GUI: Fix Tomato RAF theme for all builds. Compilation typo.
[tomato.git] / release / src-rt-6.x.4708 / linux / linux-2.6.36 / kernel / sysctl.c
blob21d9fcfd80fae7663ae6525cb3f3f5d5e9fb3a6e
1 /*
2 * sysctl.c: General linux system control interface
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
21 #include <linux/module.h>
22 #include <linux/mm.h>
23 #include <linux/swap.h>
24 #include <linux/slab.h>
25 #include <linux/sysctl.h>
26 #include <linux/signal.h>
27 #include <linux/proc_fs.h>
28 #include <linux/security.h>
29 #include <linux/ctype.h>
30 #include <linux/kmemcheck.h>
31 #include <linux/fs.h>
32 #include <linux/init.h>
33 #include <linux/kernel.h>
34 #include <linux/kobject.h>
35 #include <linux/net.h>
36 #include <linux/sysrq.h>
37 #include <linux/highuid.h>
38 #include <linux/writeback.h>
39 #include <linux/ratelimit.h>
40 #include <linux/compaction.h>
41 #include <linux/hugetlb.h>
42 #include <linux/initrd.h>
43 #include <linux/key.h>
44 #include <linux/times.h>
45 #include <linux/limits.h>
46 #include <linux/dcache.h>
47 #include <linux/dnotify.h>
48 #include <linux/syscalls.h>
49 #include <linux/vmstat.h>
50 #include <linux/nfs_fs.h>
51 #include <linux/acpi.h>
52 #include <linux/reboot.h>
53 #include <linux/ftrace.h>
54 #include <linux/perf_event.h>
55 #include <linux/kprobes.h>
56 #include <linux/pipe_fs_i.h>
57 #include <linux/oom.h>
59 #include <asm/uaccess.h>
60 #include <asm/processor.h>
62 #ifdef CONFIG_X86
63 #include <asm/nmi.h>
64 #include <asm/stacktrace.h>
65 #include <asm/io.h>
66 #endif
67 #ifdef CONFIG_BSD_PROCESS_ACCT
68 #include <linux/acct.h>
69 #endif
70 #ifdef CONFIG_RT_MUTEXES
71 #include <linux/rtmutex.h>
72 #endif
73 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
74 #include <linux/lockdep.h>
75 #endif
76 #ifdef CONFIG_CHR_DEV_SG
77 #include <scsi/sg.h>
78 #endif
80 #ifdef CONFIG_LOCKUP_DETECTOR
81 #include <linux/nmi.h>
82 #endif
85 #if defined(CONFIG_SYSCTL)
87 /* External variables not in a header file. */
88 extern int sysctl_overcommit_memory;
89 extern int sysctl_overcommit_ratio;
90 extern int max_threads;
91 extern int core_uses_pid;
92 extern int suid_dumpable;
93 extern char core_pattern[];
94 extern unsigned int core_pipe_limit;
95 extern int pid_max;
96 extern int min_free_kbytes;
97 extern int pid_max_min, pid_max_max;
98 extern int sysctl_drop_caches;
99 extern int percpu_pagelist_fraction;
100 extern int compat_log;
101 extern int latencytop_enabled;
102 extern int sysctl_nr_open_min, sysctl_nr_open_max;
103 #ifndef CONFIG_MMU
104 extern int sysctl_nr_trim_pages;
105 #endif
106 #ifdef CONFIG_BLOCK
107 extern int blk_iopoll_enabled;
108 #endif
110 /* Constants used for minimum and maximum */
111 #ifdef CONFIG_LOCKUP_DETECTOR
112 static int sixty = 60;
113 static int neg_one = -1;
114 #endif
116 static int zero;
117 static int __maybe_unused one = 1;
118 static int __maybe_unused two = 2;
119 static unsigned long one_ul = 1;
120 static int one_hundred = 100;
121 #ifdef CONFIG_PRINTK
122 static int ten_thousand = 10000;
123 #endif
125 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
126 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
128 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
129 static int maxolduid = 65535;
130 static int minolduid;
131 static int min_percpu_pagelist_fract = 8;
133 static int ngroups_max = NGROUPS_MAX;
135 #ifdef CONFIG_INOTIFY_USER
136 #include <linux/inotify.h>
137 #endif
138 #ifdef CONFIG_SPARC
139 #include <asm/system.h>
140 #endif
142 #ifdef CONFIG_SPARC64
143 extern int sysctl_tsb_ratio;
144 #endif
146 #ifdef __hppa__
147 extern int pwrsw_enabled;
148 extern int unaligned_enabled;
149 #endif
151 #ifdef CONFIG_S390
152 #ifdef CONFIG_MATHEMU
153 extern int sysctl_ieee_emulation_warnings;
154 #endif
155 extern int sysctl_userprocess_debug;
156 extern int spin_retry;
157 #endif
159 #ifdef CONFIG_IA64
160 extern int no_unaligned_warning;
161 extern int unaligned_dump_stack;
162 #endif
164 extern struct ratelimit_state printk_ratelimit_state;
166 #ifdef CONFIG_PROC_SYSCTL
167 static int proc_do_cad_pid(struct ctl_table *table, int write,
168 void __user *buffer, size_t *lenp, loff_t *ppos);
169 static int proc_taint(struct ctl_table *table, int write,
170 void __user *buffer, size_t *lenp, loff_t *ppos);
171 #endif
173 #ifdef CONFIG_MAGIC_SYSRQ
174 static int __sysrq_enabled; /* Note: sysrq code ises it's own private copy */
176 static int sysrq_sysctl_handler(ctl_table *table, int write,
177 void __user *buffer, size_t *lenp,
178 loff_t *ppos)
180 int error;
182 error = proc_dointvec(table, write, buffer, lenp, ppos);
183 if (error)
184 return error;
186 if (write)
187 sysrq_toggle_support(__sysrq_enabled);
189 return 0;
192 #endif
194 static struct ctl_table root_table[];
195 static struct ctl_table_root sysctl_table_root;
196 static struct ctl_table_header root_table_header = {
197 .count = 1,
198 .ctl_table = root_table,
199 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
200 .root = &sysctl_table_root,
201 .set = &sysctl_table_root.default_set,
203 static struct ctl_table_root sysctl_table_root = {
204 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
205 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
208 static struct ctl_table kern_table[];
209 static struct ctl_table vm_table[];
210 static struct ctl_table fs_table[];
211 static struct ctl_table debug_table[];
212 static struct ctl_table dev_table[];
213 extern struct ctl_table random_table[];
214 #ifdef CONFIG_EPOLL
215 extern struct ctl_table epoll_table[];
216 #endif
218 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
219 int sysctl_legacy_va_layout;
220 #endif
222 /* The default sysctl tables: */
224 static struct ctl_table root_table[] = {
226 .procname = "kernel",
227 .mode = 0555,
228 .child = kern_table,
231 .procname = "vm",
232 .mode = 0555,
233 .child = vm_table,
236 .procname = "fs",
237 .mode = 0555,
238 .child = fs_table,
241 .procname = "debug",
242 .mode = 0555,
243 .child = debug_table,
246 .procname = "dev",
247 .mode = 0555,
248 .child = dev_table,
251 * NOTE: do not add new entries to this table unless you have read
252 * Documentation/sysctl/ctl_unnumbered.txt
257 #ifdef CONFIG_SCHED_DEBUG
258 static int min_sched_granularity_ns = 100000; /* 100 usecs */
259 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
260 static int min_wakeup_granularity_ns; /* 0 usecs */
261 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
262 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
263 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
264 static int min_sched_shares_ratelimit = 100000; /* 100 usec */
265 static int max_sched_shares_ratelimit = NSEC_PER_SEC; /* 1 second */
266 #endif
268 #ifdef CONFIG_COMPACTION
269 static int min_extfrag_threshold;
270 static int max_extfrag_threshold = 1000;
271 #endif
273 static struct ctl_table kern_table[] = {
275 .procname = "sched_child_runs_first",
276 .data = &sysctl_sched_child_runs_first,
277 .maxlen = sizeof(unsigned int),
278 .mode = 0644,
279 .proc_handler = proc_dointvec,
281 #ifdef CONFIG_SCHED_DEBUG
283 .procname = "sched_min_granularity_ns",
284 .data = &sysctl_sched_min_granularity,
285 .maxlen = sizeof(unsigned int),
286 .mode = 0644,
287 .proc_handler = sched_proc_update_handler,
288 .extra1 = &min_sched_granularity_ns,
289 .extra2 = &max_sched_granularity_ns,
292 .procname = "sched_latency_ns",
293 .data = &sysctl_sched_latency,
294 .maxlen = sizeof(unsigned int),
295 .mode = 0644,
296 .proc_handler = sched_proc_update_handler,
297 .extra1 = &min_sched_granularity_ns,
298 .extra2 = &max_sched_granularity_ns,
301 .procname = "sched_wakeup_granularity_ns",
302 .data = &sysctl_sched_wakeup_granularity,
303 .maxlen = sizeof(unsigned int),
304 .mode = 0644,
305 .proc_handler = sched_proc_update_handler,
306 .extra1 = &min_wakeup_granularity_ns,
307 .extra2 = &max_wakeup_granularity_ns,
310 .procname = "sched_shares_ratelimit",
311 .data = &sysctl_sched_shares_ratelimit,
312 .maxlen = sizeof(unsigned int),
313 .mode = 0644,
314 .proc_handler = sched_proc_update_handler,
315 .extra1 = &min_sched_shares_ratelimit,
316 .extra2 = &max_sched_shares_ratelimit,
319 .procname = "sched_tunable_scaling",
320 .data = &sysctl_sched_tunable_scaling,
321 .maxlen = sizeof(enum sched_tunable_scaling),
322 .mode = 0644,
323 .proc_handler = sched_proc_update_handler,
324 .extra1 = &min_sched_tunable_scaling,
325 .extra2 = &max_sched_tunable_scaling,
328 .procname = "sched_shares_thresh",
329 .data = &sysctl_sched_shares_thresh,
330 .maxlen = sizeof(unsigned int),
331 .mode = 0644,
332 .proc_handler = proc_dointvec_minmax,
333 .extra1 = &zero,
336 .procname = "sched_migration_cost",
337 .data = &sysctl_sched_migration_cost,
338 .maxlen = sizeof(unsigned int),
339 .mode = 0644,
340 .proc_handler = proc_dointvec,
343 .procname = "sched_nr_migrate",
344 .data = &sysctl_sched_nr_migrate,
345 .maxlen = sizeof(unsigned int),
346 .mode = 0644,
347 .proc_handler = proc_dointvec,
350 .procname = "sched_time_avg",
351 .data = &sysctl_sched_time_avg,
352 .maxlen = sizeof(unsigned int),
353 .mode = 0644,
354 .proc_handler = proc_dointvec,
357 .procname = "timer_migration",
358 .data = &sysctl_timer_migration,
359 .maxlen = sizeof(unsigned int),
360 .mode = 0644,
361 .proc_handler = proc_dointvec_minmax,
362 .extra1 = &zero,
363 .extra2 = &one,
365 #endif
367 .procname = "sched_rt_period_us",
368 .data = &sysctl_sched_rt_period,
369 .maxlen = sizeof(unsigned int),
370 .mode = 0644,
371 .proc_handler = sched_rt_handler,
374 .procname = "sched_rt_runtime_us",
375 .data = &sysctl_sched_rt_runtime,
376 .maxlen = sizeof(int),
377 .mode = 0644,
378 .proc_handler = sched_rt_handler,
381 .procname = "sched_compat_yield",
382 .data = &sysctl_sched_compat_yield,
383 .maxlen = sizeof(unsigned int),
384 .mode = 0644,
385 .proc_handler = proc_dointvec,
387 #ifdef CONFIG_PROVE_LOCKING
389 .procname = "prove_locking",
390 .data = &prove_locking,
391 .maxlen = sizeof(int),
392 .mode = 0644,
393 .proc_handler = proc_dointvec,
395 #endif
396 #ifdef CONFIG_LOCK_STAT
398 .procname = "lock_stat",
399 .data = &lock_stat,
400 .maxlen = sizeof(int),
401 .mode = 0644,
402 .proc_handler = proc_dointvec,
404 #endif
406 .procname = "panic",
407 .data = &panic_timeout,
408 .maxlen = sizeof(int),
409 .mode = 0644,
410 .proc_handler = proc_dointvec,
413 .procname = "core_uses_pid",
414 .data = &core_uses_pid,
415 .maxlen = sizeof(int),
416 .mode = 0644,
417 .proc_handler = proc_dointvec,
420 .procname = "core_pattern",
421 .data = core_pattern,
422 .maxlen = CORENAME_MAX_SIZE,
423 .mode = 0644,
424 .proc_handler = proc_dostring,
427 .procname = "core_pipe_limit",
428 .data = &core_pipe_limit,
429 .maxlen = sizeof(unsigned int),
430 .mode = 0644,
431 .proc_handler = proc_dointvec,
433 #ifdef CONFIG_PROC_SYSCTL
435 .procname = "tainted",
436 .maxlen = sizeof(long),
437 .mode = 0644,
438 .proc_handler = proc_taint,
440 #endif
441 #ifdef CONFIG_LATENCYTOP
443 .procname = "latencytop",
444 .data = &latencytop_enabled,
445 .maxlen = sizeof(int),
446 .mode = 0644,
447 .proc_handler = proc_dointvec,
449 #endif
450 #ifdef CONFIG_BLK_DEV_INITRD
452 .procname = "real-root-dev",
453 .data = &real_root_dev,
454 .maxlen = sizeof(int),
455 .mode = 0644,
456 .proc_handler = proc_dointvec,
458 #endif
460 .procname = "print-fatal-signals",
461 .data = &print_fatal_signals,
462 .maxlen = sizeof(int),
463 .mode = 0644,
464 .proc_handler = proc_dointvec,
466 #ifdef CONFIG_SPARC
468 .procname = "reboot-cmd",
469 .data = reboot_command,
470 .maxlen = 256,
471 .mode = 0644,
472 .proc_handler = proc_dostring,
475 .procname = "stop-a",
476 .data = &stop_a_enabled,
477 .maxlen = sizeof (int),
478 .mode = 0644,
479 .proc_handler = proc_dointvec,
482 .procname = "scons-poweroff",
483 .data = &scons_pwroff,
484 .maxlen = sizeof (int),
485 .mode = 0644,
486 .proc_handler = proc_dointvec,
488 #endif
489 #ifdef CONFIG_SPARC64
491 .procname = "tsb-ratio",
492 .data = &sysctl_tsb_ratio,
493 .maxlen = sizeof (int),
494 .mode = 0644,
495 .proc_handler = proc_dointvec,
497 #endif
498 #ifdef __hppa__
500 .procname = "soft-power",
501 .data = &pwrsw_enabled,
502 .maxlen = sizeof (int),
503 .mode = 0644,
504 .proc_handler = proc_dointvec,
507 .procname = "unaligned-trap",
508 .data = &unaligned_enabled,
509 .maxlen = sizeof (int),
510 .mode = 0644,
511 .proc_handler = proc_dointvec,
513 #endif
515 .procname = "ctrl-alt-del",
516 .data = &C_A_D,
517 .maxlen = sizeof(int),
518 .mode = 0644,
519 .proc_handler = proc_dointvec,
521 #ifdef CONFIG_FUNCTION_TRACER
523 .procname = "ftrace_enabled",
524 .data = &ftrace_enabled,
525 .maxlen = sizeof(int),
526 .mode = 0644,
527 .proc_handler = ftrace_enable_sysctl,
529 #endif
530 #ifdef CONFIG_STACK_TRACER
532 .procname = "stack_tracer_enabled",
533 .data = &stack_tracer_enabled,
534 .maxlen = sizeof(int),
535 .mode = 0644,
536 .proc_handler = stack_trace_sysctl,
538 #endif
539 #ifdef CONFIG_TRACING
541 .procname = "ftrace_dump_on_oops",
542 .data = &ftrace_dump_on_oops,
543 .maxlen = sizeof(int),
544 .mode = 0644,
545 .proc_handler = proc_dointvec,
547 #endif
548 #ifdef CONFIG_MODULES
550 .procname = "modprobe",
551 .data = &modprobe_path,
552 .maxlen = KMOD_PATH_LEN,
553 .mode = 0644,
554 .proc_handler = proc_dostring,
557 .procname = "modules_disabled",
558 .data = &modules_disabled,
559 .maxlen = sizeof(int),
560 .mode = 0644,
561 /* only handle a transition from default "0" to "1" */
562 .proc_handler = proc_dointvec_minmax,
563 .extra1 = &one,
564 .extra2 = &one,
566 #endif
567 #ifdef CONFIG_HOTPLUG
569 .procname = "hotplug",
570 .data = &uevent_helper,
571 .maxlen = UEVENT_HELPER_PATH_LEN,
572 .mode = 0644,
573 .proc_handler = proc_dostring,
575 #endif
576 #ifdef CONFIG_CHR_DEV_SG
578 .procname = "sg-big-buff",
579 .data = &sg_big_buff,
580 .maxlen = sizeof (int),
581 .mode = 0444,
582 .proc_handler = proc_dointvec,
584 #endif
585 #ifdef CONFIG_BSD_PROCESS_ACCT
587 .procname = "acct",
588 .data = &acct_parm,
589 .maxlen = 3*sizeof(int),
590 .mode = 0644,
591 .proc_handler = proc_dointvec,
593 #endif
594 #ifdef CONFIG_MAGIC_SYSRQ
596 .procname = "sysrq",
597 .data = &__sysrq_enabled,
598 .maxlen = sizeof (int),
599 .mode = 0644,
600 .proc_handler = sysrq_sysctl_handler,
602 #endif
603 #ifdef CONFIG_PROC_SYSCTL
605 .procname = "cad_pid",
606 .data = NULL,
607 .maxlen = sizeof (int),
608 .mode = 0600,
609 .proc_handler = proc_do_cad_pid,
611 #endif
613 .procname = "threads-max",
614 .data = &max_threads,
615 .maxlen = sizeof(int),
616 .mode = 0644,
617 .proc_handler = proc_dointvec,
620 .procname = "random",
621 .mode = 0555,
622 .child = random_table,
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,
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,
642 #ifdef CONFIG_S390
643 #ifdef CONFIG_MATHEMU
645 .procname = "ieee_emulation_warnings",
646 .data = &sysctl_ieee_emulation_warnings,
647 .maxlen = sizeof(int),
648 .mode = 0644,
649 .proc_handler = proc_dointvec,
651 #endif
653 .procname = "userprocess_debug",
654 .data = &show_unhandled_signals,
655 .maxlen = sizeof(int),
656 .mode = 0644,
657 .proc_handler = proc_dointvec,
659 #endif
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,
670 .procname = "panic_on_oops",
671 .data = &panic_on_oops,
672 .maxlen = sizeof(int),
673 .mode = 0644,
674 .proc_handler = proc_dointvec,
676 #if defined CONFIG_PRINTK
678 .procname = "printk",
679 .data = &console_loglevel,
680 .maxlen = 4*sizeof(int),
681 .mode = 0644,
682 .proc_handler = proc_dointvec,
685 .procname = "printk_ratelimit",
686 .data = &printk_ratelimit_state.interval,
687 .maxlen = sizeof(int),
688 .mode = 0644,
689 .proc_handler = proc_dointvec_jiffies,
692 .procname = "printk_ratelimit_burst",
693 .data = &printk_ratelimit_state.burst,
694 .maxlen = sizeof(int),
695 .mode = 0644,
696 .proc_handler = proc_dointvec,
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,
707 #endif
709 .procname = "ngroups_max",
710 .data = &ngroups_max,
711 .maxlen = sizeof (int),
712 .mode = 0444,
713 .proc_handler = proc_dointvec,
715 #if defined(CONFIG_LOCKUP_DETECTOR)
717 .procname = "watchdog",
718 .data = &watchdog_enabled,
719 .maxlen = sizeof (int),
720 .mode = 0644,
721 .proc_handler = proc_dowatchdog_enabled,
724 .procname = "watchdog_thresh",
725 .data = &softlockup_thresh,
726 .maxlen = sizeof(int),
727 .mode = 0644,
728 .proc_handler = proc_dowatchdog_thresh,
729 .extra1 = &neg_one,
730 .extra2 = &sixty,
733 .procname = "softlockup_panic",
734 .data = &softlockup_panic,
735 .maxlen = sizeof(int),
736 .mode = 0644,
737 .proc_handler = proc_dointvec_minmax,
738 .extra1 = &zero,
739 .extra2 = &one,
741 #endif
742 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86) && \
743 !defined(CONFIG_LOCKUP_DETECTOR)
745 .procname = "unknown_nmi_panic",
746 .data = &unknown_nmi_panic,
747 .maxlen = sizeof (int),
748 .mode = 0644,
749 .proc_handler = proc_dointvec,
752 .procname = "nmi_watchdog",
753 .data = &nmi_watchdog_enabled,
754 .maxlen = sizeof (int),
755 .mode = 0644,
756 .proc_handler = proc_nmi_enabled,
758 #endif
759 #if defined(CONFIG_X86)
761 .procname = "panic_on_unrecovered_nmi",
762 .data = &panic_on_unrecovered_nmi,
763 .maxlen = sizeof(int),
764 .mode = 0644,
765 .proc_handler = proc_dointvec,
768 .procname = "panic_on_io_nmi",
769 .data = &panic_on_io_nmi,
770 .maxlen = sizeof(int),
771 .mode = 0644,
772 .proc_handler = proc_dointvec,
775 .procname = "bootloader_type",
776 .data = &bootloader_type,
777 .maxlen = sizeof (int),
778 .mode = 0444,
779 .proc_handler = proc_dointvec,
782 .procname = "bootloader_version",
783 .data = &bootloader_version,
784 .maxlen = sizeof (int),
785 .mode = 0444,
786 .proc_handler = proc_dointvec,
789 .procname = "kstack_depth_to_print",
790 .data = &kstack_depth_to_print,
791 .maxlen = sizeof(int),
792 .mode = 0644,
793 .proc_handler = proc_dointvec,
796 .procname = "io_delay_type",
797 .data = &io_delay_type,
798 .maxlen = sizeof(int),
799 .mode = 0644,
800 .proc_handler = proc_dointvec,
802 #endif
803 #if defined(CONFIG_MMU)
805 .procname = "randomize_va_space",
806 .data = &randomize_va_space,
807 .maxlen = sizeof(int),
808 .mode = 0644,
809 .proc_handler = proc_dointvec,
811 #endif
812 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
814 .procname = "spin_retry",
815 .data = &spin_retry,
816 .maxlen = sizeof (int),
817 .mode = 0644,
818 .proc_handler = proc_dointvec,
820 #endif
821 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
823 .procname = "acpi_video_flags",
824 .data = &acpi_realmode_flags,
825 .maxlen = sizeof (unsigned long),
826 .mode = 0644,
827 .proc_handler = proc_doulongvec_minmax,
829 #endif
830 #ifdef CONFIG_IA64
832 .procname = "ignore-unaligned-usertrap",
833 .data = &no_unaligned_warning,
834 .maxlen = sizeof (int),
835 .mode = 0644,
836 .proc_handler = proc_dointvec,
839 .procname = "unaligned-dump-stack",
840 .data = &unaligned_dump_stack,
841 .maxlen = sizeof (int),
842 .mode = 0644,
843 .proc_handler = proc_dointvec,
845 #endif
846 #ifdef CONFIG_DETECT_HUNG_TASK
848 .procname = "hung_task_panic",
849 .data = &sysctl_hung_task_panic,
850 .maxlen = sizeof(int),
851 .mode = 0644,
852 .proc_handler = proc_dointvec_minmax,
853 .extra1 = &zero,
854 .extra2 = &one,
857 .procname = "hung_task_check_count",
858 .data = &sysctl_hung_task_check_count,
859 .maxlen = sizeof(unsigned long),
860 .mode = 0644,
861 .proc_handler = proc_doulongvec_minmax,
864 .procname = "hung_task_timeout_secs",
865 .data = &sysctl_hung_task_timeout_secs,
866 .maxlen = sizeof(unsigned long),
867 .mode = 0644,
868 .proc_handler = proc_dohung_task_timeout_secs,
871 .procname = "hung_task_warnings",
872 .data = &sysctl_hung_task_warnings,
873 .maxlen = sizeof(unsigned long),
874 .mode = 0644,
875 .proc_handler = proc_doulongvec_minmax,
877 #endif
878 #ifdef CONFIG_COMPAT
880 .procname = "compat-log",
881 .data = &compat_log,
882 .maxlen = sizeof (int),
883 .mode = 0644,
884 .proc_handler = proc_dointvec,
886 #endif
887 #ifdef CONFIG_RT_MUTEXES
889 .procname = "max_lock_depth",
890 .data = &max_lock_depth,
891 .maxlen = sizeof(int),
892 .mode = 0644,
893 .proc_handler = proc_dointvec,
895 #endif
897 .procname = "poweroff_cmd",
898 .data = &poweroff_cmd,
899 .maxlen = POWEROFF_CMD_PATH_LEN,
900 .mode = 0644,
901 .proc_handler = proc_dostring,
903 #ifdef CONFIG_KEYS
905 .procname = "keys",
906 .mode = 0555,
907 .child = key_sysctls,
909 #endif
910 #ifdef CONFIG_RCU_TORTURE_TEST
912 .procname = "rcutorture_runnable",
913 .data = &rcutorture_runnable,
914 .maxlen = sizeof(int),
915 .mode = 0644,
916 .proc_handler = proc_dointvec,
918 #endif
919 #ifdef CONFIG_PERF_EVENTS
921 .procname = "perf_event_paranoid",
922 .data = &sysctl_perf_event_paranoid,
923 .maxlen = sizeof(sysctl_perf_event_paranoid),
924 .mode = 0644,
925 .proc_handler = proc_dointvec,
928 .procname = "perf_event_mlock_kb",
929 .data = &sysctl_perf_event_mlock,
930 .maxlen = sizeof(sysctl_perf_event_mlock),
931 .mode = 0644,
932 .proc_handler = proc_dointvec,
935 .procname = "perf_event_max_sample_rate",
936 .data = &sysctl_perf_event_sample_rate,
937 .maxlen = sizeof(sysctl_perf_event_sample_rate),
938 .mode = 0644,
939 .proc_handler = proc_dointvec,
941 #endif
942 #ifdef CONFIG_KMEMCHECK
944 .procname = "kmemcheck",
945 .data = &kmemcheck_enabled,
946 .maxlen = sizeof(int),
947 .mode = 0644,
948 .proc_handler = proc_dointvec,
950 #endif
951 #ifdef CONFIG_BLOCK
953 .procname = "blk_iopoll",
954 .data = &blk_iopoll_enabled,
955 .maxlen = sizeof(int),
956 .mode = 0644,
957 .proc_handler = proc_dointvec,
959 #endif
961 * NOTE: do not add new entries to this table unless you have read
962 * Documentation/sysctl/ctl_unnumbered.txt
967 static struct ctl_table vm_table[] = {
969 .procname = "overcommit_memory",
970 .data = &sysctl_overcommit_memory,
971 .maxlen = sizeof(sysctl_overcommit_memory),
972 .mode = 0644,
973 .proc_handler = proc_dointvec,
976 .procname = "panic_on_oom",
977 .data = &sysctl_panic_on_oom,
978 .maxlen = sizeof(sysctl_panic_on_oom),
979 .mode = 0644,
980 .proc_handler = proc_dointvec,
983 .procname = "oom_kill_allocating_task",
984 .data = &sysctl_oom_kill_allocating_task,
985 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
986 .mode = 0644,
987 .proc_handler = proc_dointvec,
990 .procname = "oom_dump_tasks",
991 .data = &sysctl_oom_dump_tasks,
992 .maxlen = sizeof(sysctl_oom_dump_tasks),
993 .mode = 0644,
994 .proc_handler = proc_dointvec,
997 .procname = "overcommit_ratio",
998 .data = &sysctl_overcommit_ratio,
999 .maxlen = sizeof(sysctl_overcommit_ratio),
1000 .mode = 0644,
1001 .proc_handler = proc_dointvec,
1004 .procname = "page-cluster",
1005 .data = &page_cluster,
1006 .maxlen = sizeof(int),
1007 .mode = 0644,
1008 .proc_handler = proc_dointvec,
1011 .procname = "dirty_background_ratio",
1012 .data = &dirty_background_ratio,
1013 .maxlen = sizeof(dirty_background_ratio),
1014 .mode = 0644,
1015 .proc_handler = dirty_background_ratio_handler,
1016 .extra1 = &zero,
1017 .extra2 = &one_hundred,
1020 .procname = "dirty_background_bytes",
1021 .data = &dirty_background_bytes,
1022 .maxlen = sizeof(dirty_background_bytes),
1023 .mode = 0644,
1024 .proc_handler = dirty_background_bytes_handler,
1025 .extra1 = &one_ul,
1028 .procname = "dirty_ratio",
1029 .data = &vm_dirty_ratio,
1030 .maxlen = sizeof(vm_dirty_ratio),
1031 .mode = 0644,
1032 .proc_handler = dirty_ratio_handler,
1033 .extra1 = &zero,
1034 .extra2 = &one_hundred,
1037 .procname = "dirty_bytes",
1038 .data = &vm_dirty_bytes,
1039 .maxlen = sizeof(vm_dirty_bytes),
1040 .mode = 0644,
1041 .proc_handler = dirty_bytes_handler,
1042 .extra1 = &dirty_bytes_min,
1045 .procname = "dirty_writeback_centisecs",
1046 .data = &dirty_writeback_interval,
1047 .maxlen = sizeof(dirty_writeback_interval),
1048 .mode = 0644,
1049 .proc_handler = dirty_writeback_centisecs_handler,
1052 .procname = "dirty_expire_centisecs",
1053 .data = &dirty_expire_interval,
1054 .maxlen = sizeof(dirty_expire_interval),
1055 .mode = 0644,
1056 .proc_handler = proc_dointvec,
1059 .procname = "nr_pdflush_threads",
1060 .data = &nr_pdflush_threads,
1061 .maxlen = sizeof nr_pdflush_threads,
1062 .mode = 0444 /* read-only*/,
1063 .proc_handler = proc_dointvec,
1066 .procname = "swappiness",
1067 .data = &vm_swappiness,
1068 .maxlen = sizeof(vm_swappiness),
1069 .mode = 0644,
1070 .proc_handler = proc_dointvec_minmax,
1071 .extra1 = &zero,
1072 .extra2 = &one_hundred,
1074 #ifdef CONFIG_HUGETLB_PAGE
1076 .procname = "nr_hugepages",
1077 .data = NULL,
1078 .maxlen = sizeof(unsigned long),
1079 .mode = 0644,
1080 .proc_handler = hugetlb_sysctl_handler,
1081 .extra1 = (void *)&hugetlb_zero,
1082 .extra2 = (void *)&hugetlb_infinity,
1084 #ifdef CONFIG_NUMA
1086 .procname = "nr_hugepages_mempolicy",
1087 .data = NULL,
1088 .maxlen = sizeof(unsigned long),
1089 .mode = 0644,
1090 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1091 .extra1 = (void *)&hugetlb_zero,
1092 .extra2 = (void *)&hugetlb_infinity,
1094 #endif
1096 .procname = "hugetlb_shm_group",
1097 .data = &sysctl_hugetlb_shm_group,
1098 .maxlen = sizeof(gid_t),
1099 .mode = 0644,
1100 .proc_handler = proc_dointvec,
1103 .procname = "hugepages_treat_as_movable",
1104 .data = &hugepages_treat_as_movable,
1105 .maxlen = sizeof(int),
1106 .mode = 0644,
1107 .proc_handler = hugetlb_treat_movable_handler,
1110 .procname = "nr_overcommit_hugepages",
1111 .data = NULL,
1112 .maxlen = sizeof(unsigned long),
1113 .mode = 0644,
1114 .proc_handler = hugetlb_overcommit_handler,
1115 .extra1 = (void *)&hugetlb_zero,
1116 .extra2 = (void *)&hugetlb_infinity,
1118 #endif
1120 .procname = "lowmem_reserve_ratio",
1121 .data = &sysctl_lowmem_reserve_ratio,
1122 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1123 .mode = 0644,
1124 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1127 .procname = "drop_caches",
1128 .data = &sysctl_drop_caches,
1129 .maxlen = sizeof(int),
1130 .mode = 0644,
1131 .proc_handler = drop_caches_sysctl_handler,
1133 #ifdef CONFIG_COMPACTION
1135 .procname = "compact_memory",
1136 .data = &sysctl_compact_memory,
1137 .maxlen = sizeof(int),
1138 .mode = 0200,
1139 .proc_handler = sysctl_compaction_handler,
1142 .procname = "extfrag_threshold",
1143 .data = &sysctl_extfrag_threshold,
1144 .maxlen = sizeof(int),
1145 .mode = 0644,
1146 .proc_handler = sysctl_extfrag_handler,
1147 .extra1 = &min_extfrag_threshold,
1148 .extra2 = &max_extfrag_threshold,
1151 #endif /* CONFIG_COMPACTION */
1153 .procname = "min_free_kbytes",
1154 .data = &min_free_kbytes,
1155 .maxlen = sizeof(min_free_kbytes),
1156 .mode = 0644,
1157 .proc_handler = min_free_kbytes_sysctl_handler,
1158 .extra1 = &zero,
1161 .procname = "percpu_pagelist_fraction",
1162 .data = &percpu_pagelist_fraction,
1163 .maxlen = sizeof(percpu_pagelist_fraction),
1164 .mode = 0644,
1165 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1166 .extra1 = &min_percpu_pagelist_fract,
1168 #ifdef CONFIG_MMU
1170 .procname = "max_map_count",
1171 .data = &sysctl_max_map_count,
1172 .maxlen = sizeof(sysctl_max_map_count),
1173 .mode = 0644,
1174 .proc_handler = proc_dointvec_minmax,
1175 .extra1 = &zero,
1177 #else
1179 .procname = "nr_trim_pages",
1180 .data = &sysctl_nr_trim_pages,
1181 .maxlen = sizeof(sysctl_nr_trim_pages),
1182 .mode = 0644,
1183 .proc_handler = proc_dointvec_minmax,
1184 .extra1 = &zero,
1186 #endif
1188 .procname = "laptop_mode",
1189 .data = &laptop_mode,
1190 .maxlen = sizeof(laptop_mode),
1191 .mode = 0644,
1192 .proc_handler = proc_dointvec_jiffies,
1195 .procname = "block_dump",
1196 .data = &block_dump,
1197 .maxlen = sizeof(block_dump),
1198 .mode = 0644,
1199 .proc_handler = proc_dointvec,
1200 .extra1 = &zero,
1203 .procname = "vfs_cache_pressure",
1204 .data = &sysctl_vfs_cache_pressure,
1205 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1206 .mode = 0644,
1207 .proc_handler = proc_dointvec,
1208 .extra1 = &zero,
1210 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1212 .procname = "legacy_va_layout",
1213 .data = &sysctl_legacy_va_layout,
1214 .maxlen = sizeof(sysctl_legacy_va_layout),
1215 .mode = 0644,
1216 .proc_handler = proc_dointvec,
1217 .extra1 = &zero,
1219 #endif
1220 #ifdef CONFIG_NUMA
1222 .procname = "zone_reclaim_mode",
1223 .data = &zone_reclaim_mode,
1224 .maxlen = sizeof(zone_reclaim_mode),
1225 .mode = 0644,
1226 .proc_handler = proc_dointvec,
1227 .extra1 = &zero,
1230 .procname = "min_unmapped_ratio",
1231 .data = &sysctl_min_unmapped_ratio,
1232 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1233 .mode = 0644,
1234 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1235 .extra1 = &zero,
1236 .extra2 = &one_hundred,
1239 .procname = "min_slab_ratio",
1240 .data = &sysctl_min_slab_ratio,
1241 .maxlen = sizeof(sysctl_min_slab_ratio),
1242 .mode = 0644,
1243 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1244 .extra1 = &zero,
1245 .extra2 = &one_hundred,
1247 #endif
1248 #ifdef CONFIG_SMP
1250 .procname = "stat_interval",
1251 .data = &sysctl_stat_interval,
1252 .maxlen = sizeof(sysctl_stat_interval),
1253 .mode = 0644,
1254 .proc_handler = proc_dointvec_jiffies,
1256 #endif
1257 #ifdef CONFIG_MMU
1259 .procname = "mmap_min_addr",
1260 .data = &dac_mmap_min_addr,
1261 .maxlen = sizeof(unsigned long),
1262 .mode = 0644,
1263 .proc_handler = mmap_min_addr_handler,
1265 #endif
1266 #ifdef CONFIG_NUMA
1268 .procname = "numa_zonelist_order",
1269 .data = &numa_zonelist_order,
1270 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1271 .mode = 0644,
1272 .proc_handler = numa_zonelist_order_handler,
1274 #endif
1275 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| (defined(CONFIG_SUPERH) && \
1276 defined(CONFIG_VSYSCALL))
1278 .procname = "vdso_enabled",
1279 .data = &vdso_enabled,
1280 .maxlen = sizeof(vdso_enabled),
1281 .mode = 0644,
1282 .proc_handler = proc_dointvec,
1283 .extra1 = &zero,
1285 #endif
1286 #ifdef CONFIG_HIGHMEM
1288 .procname = "highmem_is_dirtyable",
1289 .data = &vm_highmem_is_dirtyable,
1290 .maxlen = sizeof(vm_highmem_is_dirtyable),
1291 .mode = 0644,
1292 .proc_handler = proc_dointvec_minmax,
1293 .extra1 = &zero,
1294 .extra2 = &one,
1296 #endif
1298 .procname = "scan_unevictable_pages",
1299 .data = &scan_unevictable_pages,
1300 .maxlen = sizeof(scan_unevictable_pages),
1301 .mode = 0644,
1302 .proc_handler = scan_unevictable_handler,
1304 #ifdef CONFIG_MEMORY_FAILURE
1306 .procname = "memory_failure_early_kill",
1307 .data = &sysctl_memory_failure_early_kill,
1308 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1309 .mode = 0644,
1310 .proc_handler = proc_dointvec_minmax,
1311 .extra1 = &zero,
1312 .extra2 = &one,
1315 .procname = "memory_failure_recovery",
1316 .data = &sysctl_memory_failure_recovery,
1317 .maxlen = sizeof(sysctl_memory_failure_recovery),
1318 .mode = 0644,
1319 .proc_handler = proc_dointvec_minmax,
1320 .extra1 = &zero,
1321 .extra2 = &one,
1323 #endif
1326 * NOTE: do not add new entries to this table unless you have read
1327 * Documentation/sysctl/ctl_unnumbered.txt
1332 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1333 static struct ctl_table binfmt_misc_table[] = {
1336 #endif
1338 static struct ctl_table fs_table[] = {
1340 .procname = "inode-nr",
1341 .data = &inodes_stat,
1342 .maxlen = 2*sizeof(int),
1343 .mode = 0444,
1344 .proc_handler = proc_dointvec,
1347 .procname = "inode-state",
1348 .data = &inodes_stat,
1349 .maxlen = 7*sizeof(int),
1350 .mode = 0444,
1351 .proc_handler = proc_dointvec,
1354 .procname = "file-nr",
1355 .data = &files_stat,
1356 .maxlen = 3*sizeof(int),
1357 .mode = 0444,
1358 .proc_handler = proc_nr_files,
1361 .procname = "file-max",
1362 .data = &files_stat.max_files,
1363 .maxlen = sizeof(int),
1364 .mode = 0644,
1365 .proc_handler = proc_dointvec,
1368 .procname = "nr_open",
1369 .data = &sysctl_nr_open,
1370 .maxlen = sizeof(int),
1371 .mode = 0644,
1372 .proc_handler = proc_dointvec_minmax,
1373 .extra1 = &sysctl_nr_open_min,
1374 .extra2 = &sysctl_nr_open_max,
1377 .procname = "dentry-state",
1378 .data = &dentry_stat,
1379 .maxlen = 6*sizeof(int),
1380 .mode = 0444,
1381 .proc_handler = proc_dointvec,
1384 .procname = "overflowuid",
1385 .data = &fs_overflowuid,
1386 .maxlen = sizeof(int),
1387 .mode = 0644,
1388 .proc_handler = proc_dointvec_minmax,
1389 .extra1 = &minolduid,
1390 .extra2 = &maxolduid,
1393 .procname = "overflowgid",
1394 .data = &fs_overflowgid,
1395 .maxlen = sizeof(int),
1396 .mode = 0644,
1397 .proc_handler = proc_dointvec_minmax,
1398 .extra1 = &minolduid,
1399 .extra2 = &maxolduid,
1401 #ifdef CONFIG_FILE_LOCKING
1403 .procname = "leases-enable",
1404 .data = &leases_enable,
1405 .maxlen = sizeof(int),
1406 .mode = 0644,
1407 .proc_handler = proc_dointvec,
1409 #endif
1410 #ifdef CONFIG_DNOTIFY
1412 .procname = "dir-notify-enable",
1413 .data = &dir_notify_enable,
1414 .maxlen = sizeof(int),
1415 .mode = 0644,
1416 .proc_handler = proc_dointvec,
1418 #endif
1419 #ifdef CONFIG_MMU
1420 #ifdef CONFIG_FILE_LOCKING
1422 .procname = "lease-break-time",
1423 .data = &lease_break_time,
1424 .maxlen = sizeof(int),
1425 .mode = 0644,
1426 .proc_handler = proc_dointvec,
1428 #endif
1429 #ifdef CONFIG_AIO
1431 .procname = "aio-nr",
1432 .data = &aio_nr,
1433 .maxlen = sizeof(aio_nr),
1434 .mode = 0444,
1435 .proc_handler = proc_doulongvec_minmax,
1438 .procname = "aio-max-nr",
1439 .data = &aio_max_nr,
1440 .maxlen = sizeof(aio_max_nr),
1441 .mode = 0644,
1442 .proc_handler = proc_doulongvec_minmax,
1444 #endif /* CONFIG_AIO */
1445 #ifdef CONFIG_INOTIFY_USER
1447 .procname = "inotify",
1448 .mode = 0555,
1449 .child = inotify_table,
1451 #endif
1452 #ifdef CONFIG_EPOLL
1454 .procname = "epoll",
1455 .mode = 0555,
1456 .child = epoll_table,
1458 #endif
1459 #endif
1461 .procname = "suid_dumpable",
1462 .data = &suid_dumpable,
1463 .maxlen = sizeof(int),
1464 .mode = 0644,
1465 .proc_handler = proc_dointvec_minmax,
1466 .extra1 = &zero,
1467 .extra2 = &two,
1469 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1471 .procname = "binfmt_misc",
1472 .mode = 0555,
1473 .child = binfmt_misc_table,
1475 #endif
1477 .procname = "pipe-max-size",
1478 .data = &pipe_max_size,
1479 .maxlen = sizeof(int),
1480 .mode = 0644,
1481 .proc_handler = &pipe_proc_fn,
1482 .extra1 = &pipe_min_size,
1485 * NOTE: do not add new entries to this table unless you have read
1486 * Documentation/sysctl/ctl_unnumbered.txt
1491 static struct ctl_table debug_table[] = {
1492 #if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
1493 defined(CONFIG_S390)
1495 .procname = "exception-trace",
1496 .data = &show_unhandled_signals,
1497 .maxlen = sizeof(int),
1498 .mode = 0644,
1499 .proc_handler = proc_dointvec
1501 #endif
1502 #if defined(CONFIG_OPTPROBES)
1504 .procname = "kprobes-optimization",
1505 .data = &sysctl_kprobes_optimization,
1506 .maxlen = sizeof(int),
1507 .mode = 0644,
1508 .proc_handler = proc_kprobes_optimization_handler,
1509 .extra1 = &zero,
1510 .extra2 = &one,
1512 #endif
1516 static struct ctl_table dev_table[] = {
1520 static DEFINE_SPINLOCK(sysctl_lock);
1522 /* called under sysctl_lock */
1523 static int use_table(struct ctl_table_header *p)
1525 if (unlikely(p->unregistering))
1526 return 0;
1527 p->used++;
1528 return 1;
1531 /* called under sysctl_lock */
1532 static void unuse_table(struct ctl_table_header *p)
1534 if (!--p->used)
1535 if (unlikely(p->unregistering))
1536 complete(p->unregistering);
1539 /* called under sysctl_lock, will reacquire if has to wait */
1540 static void start_unregistering(struct ctl_table_header *p)
1543 * if p->used is 0, nobody will ever touch that entry again;
1544 * we'll eliminate all paths to it before dropping sysctl_lock
1546 if (unlikely(p->used)) {
1547 struct completion wait;
1548 init_completion(&wait);
1549 p->unregistering = &wait;
1550 spin_unlock(&sysctl_lock);
1551 wait_for_completion(&wait);
1552 spin_lock(&sysctl_lock);
1553 } else {
1554 /* anything non-NULL; we'll never dereference it */
1555 p->unregistering = ERR_PTR(-EINVAL);
1558 * do not remove from the list until nobody holds it; walking the
1559 * list in do_sysctl() relies on that.
1561 list_del_init(&p->ctl_entry);
1564 void sysctl_head_get(struct ctl_table_header *head)
1566 spin_lock(&sysctl_lock);
1567 head->count++;
1568 spin_unlock(&sysctl_lock);
1571 void sysctl_head_put(struct ctl_table_header *head)
1573 spin_lock(&sysctl_lock);
1574 if (!--head->count)
1575 kfree(head);
1576 spin_unlock(&sysctl_lock);
1579 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1581 if (!head)
1582 BUG();
1583 spin_lock(&sysctl_lock);
1584 if (!use_table(head))
1585 head = ERR_PTR(-ENOENT);
1586 spin_unlock(&sysctl_lock);
1587 return head;
1590 void sysctl_head_finish(struct ctl_table_header *head)
1592 if (!head)
1593 return;
1594 spin_lock(&sysctl_lock);
1595 unuse_table(head);
1596 spin_unlock(&sysctl_lock);
1599 static struct ctl_table_set *
1600 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1602 struct ctl_table_set *set = &root->default_set;
1603 if (root->lookup)
1604 set = root->lookup(root, namespaces);
1605 return set;
1608 static struct list_head *
1609 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1611 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1612 return &set->list;
1615 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1616 struct ctl_table_header *prev)
1618 struct ctl_table_root *root;
1619 struct list_head *header_list;
1620 struct ctl_table_header *head;
1621 struct list_head *tmp;
1623 spin_lock(&sysctl_lock);
1624 if (prev) {
1625 head = prev;
1626 tmp = &prev->ctl_entry;
1627 unuse_table(prev);
1628 goto next;
1630 tmp = &root_table_header.ctl_entry;
1631 for (;;) {
1632 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1634 if (!use_table(head))
1635 goto next;
1636 spin_unlock(&sysctl_lock);
1637 return head;
1638 next:
1639 root = head->root;
1640 tmp = tmp->next;
1641 header_list = lookup_header_list(root, namespaces);
1642 if (tmp != header_list)
1643 continue;
1645 do {
1646 root = list_entry(root->root_list.next,
1647 struct ctl_table_root, root_list);
1648 if (root == &sysctl_table_root)
1649 goto out;
1650 header_list = lookup_header_list(root, namespaces);
1651 } while (list_empty(header_list));
1652 tmp = header_list->next;
1654 out:
1655 spin_unlock(&sysctl_lock);
1656 return NULL;
1659 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1661 return __sysctl_head_next(current->nsproxy, prev);
1664 void register_sysctl_root(struct ctl_table_root *root)
1666 spin_lock(&sysctl_lock);
1667 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1668 spin_unlock(&sysctl_lock);
1672 * sysctl_perm does NOT grant the superuser all rights automatically, because
1673 * some sysctl variables are readonly even to root.
1676 static int test_perm(int mode, int op)
1678 if (!current_euid())
1679 mode >>= 6;
1680 else if (in_egroup_p(0))
1681 mode >>= 3;
1682 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1683 return 0;
1684 return -EACCES;
1687 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1689 int error;
1690 int mode;
1692 error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1693 if (error)
1694 return error;
1696 if (root->permissions)
1697 mode = root->permissions(root, current->nsproxy, table);
1698 else
1699 mode = table->mode;
1701 return test_perm(mode, op);
1704 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1706 for (; table->procname; table++) {
1707 table->parent = parent;
1708 if (table->child)
1709 sysctl_set_parent(table, table->child);
1713 static __init int sysctl_init(void)
1715 sysctl_set_parent(NULL, root_table);
1716 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1717 sysctl_check_table(current->nsproxy, root_table);
1718 #endif
1719 return 0;
1722 core_initcall(sysctl_init);
1724 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1725 struct ctl_table *table)
1727 struct ctl_table *p;
1728 const char *s = branch->procname;
1730 /* branch should have named subdirectory as its first element */
1731 if (!s || !branch->child)
1732 return NULL;
1734 /* ... and nothing else */
1735 if (branch[1].procname)
1736 return NULL;
1738 /* table should contain subdirectory with the same name */
1739 for (p = table; p->procname; p++) {
1740 if (!p->child)
1741 continue;
1742 if (p->procname && strcmp(p->procname, s) == 0)
1743 return p;
1745 return NULL;
1748 /* see if attaching q to p would be an improvement */
1749 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1751 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1752 struct ctl_table *next;
1753 int is_better = 0;
1754 int not_in_parent = !p->attached_by;
1756 while ((next = is_branch_in(by, to)) != NULL) {
1757 if (by == q->attached_by)
1758 is_better = 1;
1759 if (to == p->attached_by)
1760 not_in_parent = 1;
1761 by = by->child;
1762 to = next->child;
1765 if (is_better && not_in_parent) {
1766 q->attached_by = by;
1767 q->attached_to = to;
1768 q->parent = p;
1773 * __register_sysctl_paths - register a sysctl hierarchy
1774 * @root: List of sysctl headers to register on
1775 * @namespaces: Data to compute which lists of sysctl entries are visible
1776 * @path: The path to the directory the sysctl table is in.
1777 * @table: the top-level table structure
1779 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1780 * array. A completely 0 filled entry terminates the table.
1782 * The members of the &struct ctl_table structure are used as follows:
1784 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1785 * enter a sysctl file
1787 * data - a pointer to data for use by proc_handler
1789 * maxlen - the maximum size in bytes of the data
1791 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1793 * child - a pointer to the child sysctl table if this entry is a directory, or
1794 * %NULL.
1796 * proc_handler - the text handler routine (described below)
1798 * de - for internal use by the sysctl routines
1800 * extra1, extra2 - extra pointers usable by the proc handler routines
1802 * Leaf nodes in the sysctl tree will be represented by a single file
1803 * under /proc; non-leaf nodes will be represented by directories.
1805 * sysctl(2) can automatically manage read and write requests through
1806 * the sysctl table. The data and maxlen fields of the ctl_table
1807 * struct enable minimal validation of the values being written to be
1808 * performed, and the mode field allows minimal authentication.
1810 * There must be a proc_handler routine for any terminal nodes
1811 * mirrored under /proc/sys (non-terminals are handled by a built-in
1812 * directory handler). Several default handlers are available to
1813 * cover common cases -
1815 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1816 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1817 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1819 * It is the handler's job to read the input buffer from user memory
1820 * and process it. The handler should return 0 on success.
1822 * This routine returns %NULL on a failure to register, and a pointer
1823 * to the table header on success.
1825 struct ctl_table_header *__register_sysctl_paths(
1826 struct ctl_table_root *root,
1827 struct nsproxy *namespaces,
1828 const struct ctl_path *path, struct ctl_table *table)
1830 struct ctl_table_header *header;
1831 struct ctl_table *new, **prevp;
1832 unsigned int n, npath;
1833 struct ctl_table_set *set;
1835 /* Count the path components */
1836 for (npath = 0; path[npath].procname; ++npath)
1840 * For each path component, allocate a 2-element ctl_table array.
1841 * The first array element will be filled with the sysctl entry
1842 * for this, the second will be the sentinel (procname == 0).
1844 * We allocate everything in one go so that we don't have to
1845 * worry about freeing additional memory in unregister_sysctl_table.
1847 header = kzalloc(sizeof(struct ctl_table_header) +
1848 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1849 if (!header)
1850 return NULL;
1852 new = (struct ctl_table *) (header + 1);
1854 /* Now connect the dots */
1855 prevp = &header->ctl_table;
1856 for (n = 0; n < npath; ++n, ++path) {
1857 /* Copy the procname */
1858 new->procname = path->procname;
1859 new->mode = 0555;
1861 *prevp = new;
1862 prevp = &new->child;
1864 new += 2;
1866 *prevp = table;
1867 header->ctl_table_arg = table;
1869 INIT_LIST_HEAD(&header->ctl_entry);
1870 header->used = 0;
1871 header->unregistering = NULL;
1872 header->root = root;
1873 sysctl_set_parent(NULL, header->ctl_table);
1874 header->count = 1;
1875 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1876 if (sysctl_check_table(namespaces, header->ctl_table)) {
1877 kfree(header);
1878 return NULL;
1880 #endif
1881 spin_lock(&sysctl_lock);
1882 header->set = lookup_header_set(root, namespaces);
1883 header->attached_by = header->ctl_table;
1884 header->attached_to = root_table;
1885 header->parent = &root_table_header;
1886 for (set = header->set; set; set = set->parent) {
1887 struct ctl_table_header *p;
1888 list_for_each_entry(p, &set->list, ctl_entry) {
1889 if (p->unregistering)
1890 continue;
1891 try_attach(p, header);
1894 header->parent->count++;
1895 list_add_tail(&header->ctl_entry, &header->set->list);
1896 spin_unlock(&sysctl_lock);
1898 return header;
1902 * register_sysctl_table_path - register a sysctl table hierarchy
1903 * @path: The path to the directory the sysctl table is in.
1904 * @table: the top-level table structure
1906 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1907 * array. A completely 0 filled entry terminates the table.
1909 * See __register_sysctl_paths for more details.
1911 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1912 struct ctl_table *table)
1914 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1915 path, table);
1919 * register_sysctl_table - register a sysctl table hierarchy
1920 * @table: the top-level table structure
1922 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1923 * array. A completely 0 filled entry terminates the table.
1925 * See register_sysctl_paths for more details.
1927 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1929 static const struct ctl_path null_path[] = { {} };
1931 return register_sysctl_paths(null_path, table);
1935 * unregister_sysctl_table - unregister a sysctl table hierarchy
1936 * @header: the header returned from register_sysctl_table
1938 * Unregisters the sysctl table and all children. proc entries may not
1939 * actually be removed until they are no longer used by anyone.
1941 void unregister_sysctl_table(struct ctl_table_header * header)
1943 might_sleep();
1945 if (header == NULL)
1946 return;
1948 spin_lock(&sysctl_lock);
1949 start_unregistering(header);
1950 if (!--header->parent->count) {
1951 WARN_ON(1);
1952 kfree(header->parent);
1954 if (!--header->count)
1955 kfree(header);
1956 spin_unlock(&sysctl_lock);
1959 int sysctl_is_seen(struct ctl_table_header *p)
1961 struct ctl_table_set *set = p->set;
1962 int res;
1963 spin_lock(&sysctl_lock);
1964 if (p->unregistering)
1965 res = 0;
1966 else if (!set->is_seen)
1967 res = 1;
1968 else
1969 res = set->is_seen(set);
1970 spin_unlock(&sysctl_lock);
1971 return res;
1974 void setup_sysctl_set(struct ctl_table_set *p,
1975 struct ctl_table_set *parent,
1976 int (*is_seen)(struct ctl_table_set *))
1978 INIT_LIST_HEAD(&p->list);
1979 p->parent = parent ? parent : &sysctl_table_root.default_set;
1980 p->is_seen = is_seen;
1983 #else /* !CONFIG_SYSCTL */
1984 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1986 return NULL;
1989 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1990 struct ctl_table *table)
1992 return NULL;
1995 void unregister_sysctl_table(struct ctl_table_header * table)
1999 void setup_sysctl_set(struct ctl_table_set *p,
2000 struct ctl_table_set *parent,
2001 int (*is_seen)(struct ctl_table_set *))
2005 void sysctl_head_put(struct ctl_table_header *head)
2009 #endif /* CONFIG_SYSCTL */
2012 * /proc/sys support
2015 #ifdef CONFIG_PROC_SYSCTL
2017 static int _proc_do_string(void* data, int maxlen, int write,
2018 void __user *buffer,
2019 size_t *lenp, loff_t *ppos)
2021 size_t len;
2022 char __user *p;
2023 char c;
2025 if (!data || !maxlen || !*lenp) {
2026 *lenp = 0;
2027 return 0;
2030 if (write) {
2031 len = 0;
2032 p = buffer;
2033 while (len < *lenp) {
2034 if (get_user(c, p++))
2035 return -EFAULT;
2036 if (c == 0 || c == '\n')
2037 break;
2038 len++;
2040 if (len >= maxlen)
2041 len = maxlen-1;
2042 if(copy_from_user(data, buffer, len))
2043 return -EFAULT;
2044 ((char *) data)[len] = 0;
2045 *ppos += *lenp;
2046 } else {
2047 len = strlen(data);
2048 if (len > maxlen)
2049 len = maxlen;
2051 if (*ppos > len) {
2052 *lenp = 0;
2053 return 0;
2056 data += *ppos;
2057 len -= *ppos;
2059 if (len > *lenp)
2060 len = *lenp;
2061 if (len)
2062 if(copy_to_user(buffer, data, len))
2063 return -EFAULT;
2064 if (len < *lenp) {
2065 if(put_user('\n', ((char __user *) buffer) + len))
2066 return -EFAULT;
2067 len++;
2069 *lenp = len;
2070 *ppos += len;
2072 return 0;
2076 * proc_dostring - read a string sysctl
2077 * @table: the sysctl table
2078 * @write: %TRUE if this is a write to the sysctl file
2079 * @buffer: the user buffer
2080 * @lenp: the size of the user buffer
2081 * @ppos: file position
2083 * Reads/writes a string from/to the user buffer. If the kernel
2084 * buffer provided is not large enough to hold the string, the
2085 * string is truncated. The copied string is %NULL-terminated.
2086 * If the string is being read by the user process, it is copied
2087 * and a newline '\n' is added. It is truncated if the buffer is
2088 * not large enough.
2090 * Returns 0 on success.
2092 int proc_dostring(struct ctl_table *table, int write,
2093 void __user *buffer, size_t *lenp, loff_t *ppos)
2095 return _proc_do_string(table->data, table->maxlen, write,
2096 buffer, lenp, ppos);
2099 static size_t proc_skip_spaces(char **buf)
2101 size_t ret;
2102 char *tmp = skip_spaces(*buf);
2103 ret = tmp - *buf;
2104 *buf = tmp;
2105 return ret;
2108 static void proc_skip_char(char **buf, size_t *size, const char v)
2110 while (*size) {
2111 if (**buf != v)
2112 break;
2113 (*size)--;
2114 (*buf)++;
2118 #define TMPBUFLEN 22
2120 * proc_get_long - reads an ASCII formatted integer from a user buffer
2122 * @buf: a kernel buffer
2123 * @size: size of the kernel buffer
2124 * @val: this is where the number will be stored
2125 * @neg: set to %TRUE if number is negative
2126 * @perm_tr: a vector which contains the allowed trailers
2127 * @perm_tr_len: size of the perm_tr vector
2128 * @tr: pointer to store the trailer character
2130 * In case of success %0 is returned and @buf and @size are updated with
2131 * the amount of bytes read. If @tr is non-NULL and a trailing
2132 * character exists (size is non-zero after returning from this
2133 * function), @tr is updated with the trailing character.
2135 static int proc_get_long(char **buf, size_t *size,
2136 unsigned long *val, bool *neg,
2137 const char *perm_tr, unsigned perm_tr_len, char *tr)
2139 int len;
2140 char *p, tmp[TMPBUFLEN];
2142 if (!*size)
2143 return -EINVAL;
2145 len = *size;
2146 if (len > TMPBUFLEN - 1)
2147 len = TMPBUFLEN - 1;
2149 memcpy(tmp, *buf, len);
2151 tmp[len] = 0;
2152 p = tmp;
2153 if (*p == '-' && *size > 1) {
2154 *neg = true;
2155 p++;
2156 } else
2157 *neg = false;
2158 if (!isdigit(*p))
2159 return -EINVAL;
2161 *val = simple_strtoul(p, &p, 0);
2163 len = p - tmp;
2165 /* We don't know if the next char is whitespace thus we may accept
2166 * invalid integers (e.g. 1234...a) or two integers instead of one
2167 * (e.g. 123...1). So lets not allow such large numbers. */
2168 if (len == TMPBUFLEN - 1)
2169 return -EINVAL;
2171 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2172 return -EINVAL;
2174 if (tr && (len < *size))
2175 *tr = *p;
2177 *buf += len;
2178 *size -= len;
2180 return 0;
2184 * proc_put_long - converts an integer to a decimal ASCII formatted string
2186 * @buf: the user buffer
2187 * @size: the size of the user buffer
2188 * @val: the integer to be converted
2189 * @neg: sign of the number, %TRUE for negative
2191 * In case of success %0 is returned and @buf and @size are updated with
2192 * the amount of bytes written.
2194 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2195 bool neg)
2197 int len;
2198 char tmp[TMPBUFLEN], *p = tmp;
2200 sprintf(p, "%s%lu", neg ? "-" : "", val);
2201 len = strlen(tmp);
2202 if (len > *size)
2203 len = *size;
2204 if (copy_to_user(*buf, tmp, len))
2205 return -EFAULT;
2206 *size -= len;
2207 *buf += len;
2208 return 0;
2210 #undef TMPBUFLEN
2212 static int proc_put_char(void __user **buf, size_t *size, char c)
2214 if (*size) {
2215 char __user **buffer = (char __user **)buf;
2216 if (put_user(c, *buffer))
2217 return -EFAULT;
2218 (*size)--, (*buffer)++;
2219 *buf = *buffer;
2221 return 0;
2224 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2225 int *valp,
2226 int write, void *data)
2228 if (write) {
2229 *valp = *negp ? -*lvalp : *lvalp;
2230 } else {
2231 int val = *valp;
2232 if (val < 0) {
2233 *negp = true;
2234 *lvalp = (unsigned long)-val;
2235 } else {
2236 *negp = false;
2237 *lvalp = (unsigned long)val;
2240 return 0;
2243 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2245 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2246 int write, void __user *buffer,
2247 size_t *lenp, loff_t *ppos,
2248 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2249 int write, void *data),
2250 void *data)
2252 int *i, vleft, first = 1, err = 0;
2253 unsigned long page = 0;
2254 size_t left;
2255 char *kbuf;
2257 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2258 *lenp = 0;
2259 return 0;
2262 i = (int *) tbl_data;
2263 vleft = table->maxlen / sizeof(*i);
2264 left = *lenp;
2266 if (!conv)
2267 conv = do_proc_dointvec_conv;
2269 if (write) {
2270 if (left > PAGE_SIZE - 1)
2271 left = PAGE_SIZE - 1;
2272 page = __get_free_page(GFP_TEMPORARY);
2273 kbuf = (char *) page;
2274 if (!kbuf)
2275 return -ENOMEM;
2276 if (copy_from_user(kbuf, buffer, left)) {
2277 err = -EFAULT;
2278 goto free;
2280 kbuf[left] = 0;
2283 for (; left && vleft--; i++, first=0) {
2284 unsigned long lval;
2285 bool neg;
2287 if (write) {
2288 left -= proc_skip_spaces(&kbuf);
2290 if (!left)
2291 break;
2292 err = proc_get_long(&kbuf, &left, &lval, &neg,
2293 proc_wspace_sep,
2294 sizeof(proc_wspace_sep), NULL);
2295 if (err)
2296 break;
2297 if (conv(&neg, &lval, i, 1, data)) {
2298 err = -EINVAL;
2299 break;
2301 } else {
2302 if (conv(&neg, &lval, i, 0, data)) {
2303 err = -EINVAL;
2304 break;
2306 if (!first)
2307 err = proc_put_char(&buffer, &left, '\t');
2308 if (err)
2309 break;
2310 err = proc_put_long(&buffer, &left, lval, neg);
2311 if (err)
2312 break;
2316 if (!write && !first && left && !err)
2317 err = proc_put_char(&buffer, &left, '\n');
2318 if (write && !err && left)
2319 left -= proc_skip_spaces(&kbuf);
2320 free:
2321 if (write) {
2322 free_page(page);
2323 if (first)
2324 return err ? : -EINVAL;
2326 *lenp -= left;
2327 *ppos += *lenp;
2328 return err;
2331 static int do_proc_dointvec(struct ctl_table *table, int write,
2332 void __user *buffer, size_t *lenp, loff_t *ppos,
2333 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2334 int write, void *data),
2335 void *data)
2337 return __do_proc_dointvec(table->data, table, write,
2338 buffer, lenp, ppos, conv, data);
2342 * proc_dointvec - read a vector of integers
2343 * @table: the sysctl table
2344 * @write: %TRUE if this is a write to the sysctl file
2345 * @buffer: the user buffer
2346 * @lenp: the size of the user buffer
2347 * @ppos: file position
2349 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2350 * values from/to the user buffer, treated as an ASCII string.
2352 * Returns 0 on success.
2354 int proc_dointvec(struct ctl_table *table, int write,
2355 void __user *buffer, size_t *lenp, loff_t *ppos)
2357 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2358 NULL,NULL);
2362 * Taint values can only be increased
2363 * This means we can safely use a temporary.
2365 static int proc_taint(struct ctl_table *table, int write,
2366 void __user *buffer, size_t *lenp, loff_t *ppos)
2368 struct ctl_table t;
2369 unsigned long tmptaint = get_taint();
2370 int err;
2372 if (write && !capable(CAP_SYS_ADMIN))
2373 return -EPERM;
2375 t = *table;
2376 t.data = &tmptaint;
2377 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2378 if (err < 0)
2379 return err;
2381 if (write) {
2383 * Poor man's atomic or. Not worth adding a primitive
2384 * to everyone's atomic.h for this
2386 int i;
2387 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2388 if ((tmptaint >> i) & 1)
2389 add_taint(i);
2393 return err;
2396 struct do_proc_dointvec_minmax_conv_param {
2397 int *min;
2398 int *max;
2401 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2402 int *valp,
2403 int write, void *data)
2405 struct do_proc_dointvec_minmax_conv_param *param = data;
2406 if (write) {
2407 int val = *negp ? -*lvalp : *lvalp;
2408 if ((param->min && *param->min > val) ||
2409 (param->max && *param->max < val))
2410 return -EINVAL;
2411 *valp = val;
2412 } else {
2413 int val = *valp;
2414 if (val < 0) {
2415 *negp = true;
2416 *lvalp = (unsigned long)-val;
2417 } else {
2418 *negp = false;
2419 *lvalp = (unsigned long)val;
2422 return 0;
2426 * proc_dointvec_minmax - read a vector of integers with min/max values
2427 * @table: the sysctl table
2428 * @write: %TRUE if this is a write to the sysctl file
2429 * @buffer: the user buffer
2430 * @lenp: the size of the user buffer
2431 * @ppos: file position
2433 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2434 * values from/to the user buffer, treated as an ASCII string.
2436 * This routine will ensure the values are within the range specified by
2437 * table->extra1 (min) and table->extra2 (max).
2439 * Returns 0 on success.
2441 int proc_dointvec_minmax(struct ctl_table *table, int write,
2442 void __user *buffer, size_t *lenp, loff_t *ppos)
2444 struct do_proc_dointvec_minmax_conv_param param = {
2445 .min = (int *) table->extra1,
2446 .max = (int *) table->extra2,
2448 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2449 do_proc_dointvec_minmax_conv, &param);
2452 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2453 void __user *buffer,
2454 size_t *lenp, loff_t *ppos,
2455 unsigned long convmul,
2456 unsigned long convdiv)
2458 unsigned long *i, *min, *max;
2459 int vleft, first = 1, err = 0;
2460 unsigned long page = 0;
2461 size_t left;
2462 char *kbuf;
2464 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2465 *lenp = 0;
2466 return 0;
2469 i = (unsigned long *) data;
2470 min = (unsigned long *) table->extra1;
2471 max = (unsigned long *) table->extra2;
2472 vleft = table->maxlen / sizeof(unsigned long);
2473 left = *lenp;
2475 if (write) {
2476 if (left > PAGE_SIZE - 1)
2477 left = PAGE_SIZE - 1;
2478 page = __get_free_page(GFP_TEMPORARY);
2479 kbuf = (char *) page;
2480 if (!kbuf)
2481 return -ENOMEM;
2482 if (copy_from_user(kbuf, buffer, left)) {
2483 err = -EFAULT;
2484 goto free;
2486 kbuf[left] = 0;
2489 for (; left && vleft--; i++, first = 0) {
2490 unsigned long val;
2492 if (write) {
2493 bool neg;
2495 left -= proc_skip_spaces(&kbuf);
2497 err = proc_get_long(&kbuf, &left, &val, &neg,
2498 proc_wspace_sep,
2499 sizeof(proc_wspace_sep), NULL);
2500 if (err)
2501 break;
2502 if (neg)
2503 continue;
2504 if ((min && val < *min) || (max && val > *max))
2505 continue;
2506 *i = val;
2507 } else {
2508 val = convdiv * (*i) / convmul;
2509 if (!first)
2510 err = proc_put_char(&buffer, &left, '\t');
2511 err = proc_put_long(&buffer, &left, val, false);
2512 if (err)
2513 break;
2517 if (!write && !first && left && !err)
2518 err = proc_put_char(&buffer, &left, '\n');
2519 if (write && !err)
2520 left -= proc_skip_spaces(&kbuf);
2521 free:
2522 if (write) {
2523 free_page(page);
2524 if (first)
2525 return err ? : -EINVAL;
2527 *lenp -= left;
2528 *ppos += *lenp;
2529 return err;
2532 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2533 void __user *buffer,
2534 size_t *lenp, loff_t *ppos,
2535 unsigned long convmul,
2536 unsigned long convdiv)
2538 return __do_proc_doulongvec_minmax(table->data, table, write,
2539 buffer, lenp, ppos, convmul, convdiv);
2543 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2544 * @table: the sysctl table
2545 * @write: %TRUE if this is a write to the sysctl file
2546 * @buffer: the user buffer
2547 * @lenp: the size of the user buffer
2548 * @ppos: file position
2550 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2551 * values from/to the user buffer, treated as an ASCII string.
2553 * This routine will ensure the values are within the range specified by
2554 * table->extra1 (min) and table->extra2 (max).
2556 * Returns 0 on success.
2558 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2559 void __user *buffer, size_t *lenp, loff_t *ppos)
2561 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2565 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2566 * @table: the sysctl table
2567 * @write: %TRUE if this is a write to the sysctl file
2568 * @buffer: the user buffer
2569 * @lenp: the size of the user buffer
2570 * @ppos: file position
2572 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2573 * values from/to the user buffer, treated as an ASCII string. The values
2574 * are treated as milliseconds, and converted to jiffies when they are stored.
2576 * This routine will ensure the values are within the range specified by
2577 * table->extra1 (min) and table->extra2 (max).
2579 * Returns 0 on success.
2581 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2582 void __user *buffer,
2583 size_t *lenp, loff_t *ppos)
2585 return do_proc_doulongvec_minmax(table, write, buffer,
2586 lenp, ppos, HZ, 1000l);
2590 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2591 int *valp,
2592 int write, void *data)
2594 if (write) {
2595 if (*lvalp > LONG_MAX / HZ)
2596 return 1;
2597 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2598 } else {
2599 int val = *valp;
2600 unsigned long lval;
2601 if (val < 0) {
2602 *negp = true;
2603 lval = (unsigned long)-val;
2604 } else {
2605 *negp = false;
2606 lval = (unsigned long)val;
2608 *lvalp = lval / HZ;
2610 return 0;
2613 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2614 int *valp,
2615 int write, void *data)
2617 if (write) {
2618 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2619 return 1;
2620 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2621 } else {
2622 int val = *valp;
2623 unsigned long lval;
2624 if (val < 0) {
2625 *negp = true;
2626 lval = (unsigned long)-val;
2627 } else {
2628 *negp = false;
2629 lval = (unsigned long)val;
2631 *lvalp = jiffies_to_clock_t(lval);
2633 return 0;
2636 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2637 int *valp,
2638 int write, void *data)
2640 if (write) {
2641 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2642 } else {
2643 int val = *valp;
2644 unsigned long lval;
2645 if (val < 0) {
2646 *negp = true;
2647 lval = (unsigned long)-val;
2648 } else {
2649 *negp = false;
2650 lval = (unsigned long)val;
2652 *lvalp = jiffies_to_msecs(lval);
2654 return 0;
2658 * proc_dointvec_jiffies - read a vector of integers as seconds
2659 * @table: the sysctl table
2660 * @write: %TRUE if this is a write to the sysctl file
2661 * @buffer: the user buffer
2662 * @lenp: the size of the user buffer
2663 * @ppos: file position
2665 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2666 * values from/to the user buffer, treated as an ASCII string.
2667 * The values read are assumed to be in seconds, and are converted into
2668 * jiffies.
2670 * Returns 0 on success.
2672 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2673 void __user *buffer, size_t *lenp, loff_t *ppos)
2675 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2676 do_proc_dointvec_jiffies_conv,NULL);
2680 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2681 * @table: the sysctl table
2682 * @write: %TRUE if this is a write to the sysctl file
2683 * @buffer: the user buffer
2684 * @lenp: the size of the user buffer
2685 * @ppos: pointer to the file position
2687 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2688 * values from/to the user buffer, treated as an ASCII string.
2689 * The values read are assumed to be in 1/USER_HZ seconds, and
2690 * are converted into jiffies.
2692 * Returns 0 on success.
2694 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2695 void __user *buffer, size_t *lenp, loff_t *ppos)
2697 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2698 do_proc_dointvec_userhz_jiffies_conv,NULL);
2702 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2703 * @table: the sysctl table
2704 * @write: %TRUE if this is a write to the sysctl file
2705 * @buffer: the user buffer
2706 * @lenp: the size of the user buffer
2707 * @ppos: file position
2708 * @ppos: the current position in the file
2710 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2711 * values from/to the user buffer, treated as an ASCII string.
2712 * The values read are assumed to be in 1/1000 seconds, and
2713 * are converted into jiffies.
2715 * Returns 0 on success.
2717 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2718 void __user *buffer, size_t *lenp, loff_t *ppos)
2720 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2721 do_proc_dointvec_ms_jiffies_conv, NULL);
2724 static int proc_do_cad_pid(struct ctl_table *table, int write,
2725 void __user *buffer, size_t *lenp, loff_t *ppos)
2727 struct pid *new_pid;
2728 pid_t tmp;
2729 int r;
2731 tmp = pid_vnr(cad_pid);
2733 r = __do_proc_dointvec(&tmp, table, write, buffer,
2734 lenp, ppos, NULL, NULL);
2735 if (r || !write)
2736 return r;
2738 new_pid = find_get_pid(tmp);
2739 if (!new_pid)
2740 return -ESRCH;
2742 put_pid(xchg(&cad_pid, new_pid));
2743 return 0;
2747 * proc_do_large_bitmap - read/write from/to a large bitmap
2748 * @table: the sysctl table
2749 * @write: %TRUE if this is a write to the sysctl file
2750 * @buffer: the user buffer
2751 * @lenp: the size of the user buffer
2752 * @ppos: file position
2754 * The bitmap is stored at table->data and the bitmap length (in bits)
2755 * in table->maxlen.
2757 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2758 * large bitmaps may be represented in a compact manner. Writing into
2759 * the file will clear the bitmap then update it with the given input.
2761 * Returns 0 on success.
2763 int proc_do_large_bitmap(struct ctl_table *table, int write,
2764 void __user *buffer, size_t *lenp, loff_t *ppos)
2766 int err = 0;
2767 bool first = 1;
2768 size_t left = *lenp;
2769 unsigned long bitmap_len = table->maxlen;
2770 unsigned long *bitmap = (unsigned long *) table->data;
2771 unsigned long *tmp_bitmap = NULL;
2772 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2774 if (!bitmap_len || !left || (*ppos && !write)) {
2775 *lenp = 0;
2776 return 0;
2779 if (write) {
2780 unsigned long page = 0;
2781 char *kbuf;
2783 if (left > PAGE_SIZE - 1)
2784 left = PAGE_SIZE - 1;
2786 page = __get_free_page(GFP_TEMPORARY);
2787 kbuf = (char *) page;
2788 if (!kbuf)
2789 return -ENOMEM;
2790 if (copy_from_user(kbuf, buffer, left)) {
2791 free_page(page);
2792 return -EFAULT;
2794 kbuf[left] = 0;
2796 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2797 GFP_KERNEL);
2798 if (!tmp_bitmap) {
2799 free_page(page);
2800 return -ENOMEM;
2802 proc_skip_char(&kbuf, &left, '\n');
2803 while (!err && left) {
2804 unsigned long val_a, val_b;
2805 bool neg;
2807 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2808 sizeof(tr_a), &c);
2809 if (err)
2810 break;
2811 if (val_a >= bitmap_len || neg) {
2812 err = -EINVAL;
2813 break;
2816 val_b = val_a;
2817 if (left) {
2818 kbuf++;
2819 left--;
2822 if (c == '-') {
2823 err = proc_get_long(&kbuf, &left, &val_b,
2824 &neg, tr_b, sizeof(tr_b),
2825 &c);
2826 if (err)
2827 break;
2828 if (val_b >= bitmap_len || neg ||
2829 val_a > val_b) {
2830 err = -EINVAL;
2831 break;
2833 if (left) {
2834 kbuf++;
2835 left--;
2839 while (val_a <= val_b)
2840 set_bit(val_a++, tmp_bitmap);
2842 first = 0;
2843 proc_skip_char(&kbuf, &left, '\n');
2845 free_page(page);
2846 } else {
2847 unsigned long bit_a, bit_b = 0;
2849 while (left) {
2850 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2851 if (bit_a >= bitmap_len)
2852 break;
2853 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2854 bit_a + 1) - 1;
2856 if (!first) {
2857 err = proc_put_char(&buffer, &left, ',');
2858 if (err)
2859 break;
2861 err = proc_put_long(&buffer, &left, bit_a, false);
2862 if (err)
2863 break;
2864 if (bit_a != bit_b) {
2865 err = proc_put_char(&buffer, &left, '-');
2866 if (err)
2867 break;
2868 err = proc_put_long(&buffer, &left, bit_b, false);
2869 if (err)
2870 break;
2873 first = 0; bit_b++;
2875 if (!err)
2876 err = proc_put_char(&buffer, &left, '\n');
2879 if (!err) {
2880 if (write) {
2881 if (*ppos)
2882 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2883 else
2884 memcpy(bitmap, tmp_bitmap,
2885 BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long));
2887 kfree(tmp_bitmap);
2888 *lenp -= left;
2889 *ppos += *lenp;
2890 return 0;
2891 } else {
2892 kfree(tmp_bitmap);
2893 return err;
2897 #else /* CONFIG_PROC_FS */
2899 int proc_dostring(struct ctl_table *table, int write,
2900 void __user *buffer, size_t *lenp, loff_t *ppos)
2902 return -ENOSYS;
2905 int proc_dointvec(struct ctl_table *table, int write,
2906 void __user *buffer, size_t *lenp, loff_t *ppos)
2908 return -ENOSYS;
2911 int proc_dointvec_minmax(struct ctl_table *table, int write,
2912 void __user *buffer, size_t *lenp, loff_t *ppos)
2914 return -ENOSYS;
2917 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2918 void __user *buffer, size_t *lenp, loff_t *ppos)
2920 return -ENOSYS;
2923 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2924 void __user *buffer, size_t *lenp, loff_t *ppos)
2926 return -ENOSYS;
2929 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2930 void __user *buffer, size_t *lenp, loff_t *ppos)
2932 return -ENOSYS;
2935 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2936 void __user *buffer, size_t *lenp, loff_t *ppos)
2938 return -ENOSYS;
2941 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2942 void __user *buffer,
2943 size_t *lenp, loff_t *ppos)
2945 return -ENOSYS;
2949 #endif /* CONFIG_PROC_FS */
2952 * No sense putting this after each symbol definition, twice,
2953 * exception granted :-)
2955 EXPORT_SYMBOL(proc_dointvec);
2956 EXPORT_SYMBOL(proc_dointvec_jiffies);
2957 EXPORT_SYMBOL(proc_dointvec_minmax);
2958 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2959 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2960 EXPORT_SYMBOL(proc_dostring);
2961 EXPORT_SYMBOL(proc_doulongvec_minmax);
2962 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2963 EXPORT_SYMBOL(register_sysctl_table);
2964 EXPORT_SYMBOL(register_sysctl_paths);
2965 EXPORT_SYMBOL(unregister_sysctl_table);