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
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
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>
23 #include <linux/swap.h>
24 #include <linux/slab.h>
25 #include <linux/sysctl.h>
26 #include <linux/proc_fs.h>
27 #include <linux/security.h>
28 #include <linux/ctype.h>
29 #include <linux/kmemcheck.h>
31 #include <linux/init.h>
32 #include <linux/kernel.h>
33 #include <linux/kobject.h>
34 #include <linux/net.h>
35 #include <linux/sysrq.h>
36 #include <linux/highuid.h>
37 #include <linux/writeback.h>
38 #include <linux/ratelimit.h>
39 #include <linux/hugetlb.h>
40 #include <linux/initrd.h>
41 #include <linux/key.h>
42 #include <linux/times.h>
43 #include <linux/limits.h>
44 #include <linux/dcache.h>
45 #include <linux/syscalls.h>
46 #include <linux/vmstat.h>
47 #include <linux/nfs_fs.h>
48 #include <linux/acpi.h>
49 #include <linux/reboot.h>
50 #include <linux/ftrace.h>
51 #include <linux/slow-work.h>
52 #include <linux/perf_event.h>
54 #include <asm/uaccess.h>
55 #include <asm/processor.h>
59 #include <asm/stacktrace.h>
64 #if defined(CONFIG_SYSCTL)
66 /* External variables not in a header file. */
68 extern int print_fatal_signals;
69 extern int sysctl_overcommit_memory;
70 extern int sysctl_overcommit_ratio;
71 extern int sysctl_panic_on_oom;
72 extern int sysctl_oom_kill_allocating_task;
73 extern int sysctl_oom_dump_tasks;
74 extern int max_threads;
75 extern int core_uses_pid;
76 extern int suid_dumpable;
77 extern char core_pattern[];
78 extern unsigned int core_pipe_limit;
80 extern int min_free_kbytes;
81 extern int pid_max_min, pid_max_max;
82 extern int sysctl_drop_caches;
83 extern int percpu_pagelist_fraction;
84 extern int compat_log;
85 extern int latencytop_enabled;
86 extern int sysctl_nr_open_min, sysctl_nr_open_max;
88 extern int sysctl_nr_trim_pages;
90 #ifdef CONFIG_RCU_TORTURE_TEST
91 extern int rcutorture_runnable;
92 #endif /* #ifdef CONFIG_RCU_TORTURE_TEST */
94 extern int blk_iopoll_enabled;
97 /* Constants used for minimum and maximum */
98 #ifdef CONFIG_DETECT_SOFTLOCKUP
99 static int sixty = 60;
100 static int neg_one = -1;
104 static int __maybe_unused one = 1;
105 static int __maybe_unused two = 2;
106 static unsigned long one_ul = 1;
107 static int one_hundred = 100;
109 static int ten_thousand = 10000;
112 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
113 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
115 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
116 static int maxolduid = 65535;
117 static int minolduid;
118 static int min_percpu_pagelist_fract = 8;
120 static int ngroups_max = NGROUPS_MAX;
122 #ifdef CONFIG_MODULES
123 extern char modprobe_path[];
124 extern int modules_disabled;
126 #ifdef CONFIG_CHR_DEV_SG
127 extern int sg_big_buff;
131 #include <asm/system.h>
134 #ifdef CONFIG_SPARC64
135 extern int sysctl_tsb_ratio;
139 extern int pwrsw_enabled;
140 extern int unaligned_enabled;
144 #ifdef CONFIG_MATHEMU
145 extern int sysctl_ieee_emulation_warnings;
147 extern int sysctl_userprocess_debug;
148 extern int spin_retry;
151 #ifdef CONFIG_BSD_PROCESS_ACCT
152 extern int acct_parm[];
156 extern int no_unaligned_warning;
157 extern int unaligned_dump_stack;
160 extern struct ratelimit_state printk_ratelimit_state;
162 #ifdef CONFIG_RT_MUTEXES
163 extern int max_lock_depth;
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);
173 static struct ctl_table root_table[];
174 static struct ctl_table_root sysctl_table_root;
175 static struct ctl_table_header root_table_header = {
177 .ctl_table = root_table,
178 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
179 .root = &sysctl_table_root,
180 .set = &sysctl_table_root.default_set,
182 static struct ctl_table_root sysctl_table_root = {
183 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
184 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
187 static struct ctl_table kern_table[];
188 static struct ctl_table vm_table[];
189 static struct ctl_table fs_table[];
190 static struct ctl_table debug_table[];
191 static struct ctl_table dev_table[];
192 extern struct ctl_table random_table[];
193 #ifdef CONFIG_INOTIFY_USER
194 extern struct ctl_table inotify_table[];
197 extern struct ctl_table epoll_table[];
200 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
201 int sysctl_legacy_va_layout;
204 extern int prove_locking;
205 extern int lock_stat;
207 /* The default sysctl tables: */
209 static struct ctl_table root_table[] = {
211 .procname = "kernel",
228 .child = debug_table,
236 * NOTE: do not add new entries to this table unless you have read
237 * Documentation/sysctl/ctl_unnumbered.txt
242 #ifdef CONFIG_SCHED_DEBUG
243 static int min_sched_granularity_ns = 100000; /* 100 usecs */
244 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
245 static int min_wakeup_granularity_ns; /* 0 usecs */
246 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
247 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
248 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
249 static int min_sched_shares_ratelimit = 100000; /* 100 usec */
250 static int max_sched_shares_ratelimit = NSEC_PER_SEC; /* 1 second */
253 static struct ctl_table kern_table[] = {
255 .procname = "sched_child_runs_first",
256 .data = &sysctl_sched_child_runs_first,
257 .maxlen = sizeof(unsigned int),
259 .proc_handler = proc_dointvec,
261 #ifdef CONFIG_SCHED_DEBUG
263 .procname = "sched_min_granularity_ns",
264 .data = &sysctl_sched_min_granularity,
265 .maxlen = sizeof(unsigned int),
267 .proc_handler = sched_proc_update_handler,
268 .extra1 = &min_sched_granularity_ns,
269 .extra2 = &max_sched_granularity_ns,
272 .procname = "sched_latency_ns",
273 .data = &sysctl_sched_latency,
274 .maxlen = sizeof(unsigned int),
276 .proc_handler = sched_proc_update_handler,
277 .extra1 = &min_sched_granularity_ns,
278 .extra2 = &max_sched_granularity_ns,
281 .procname = "sched_wakeup_granularity_ns",
282 .data = &sysctl_sched_wakeup_granularity,
283 .maxlen = sizeof(unsigned int),
285 .proc_handler = sched_proc_update_handler,
286 .extra1 = &min_wakeup_granularity_ns,
287 .extra2 = &max_wakeup_granularity_ns,
290 .procname = "sched_shares_ratelimit",
291 .data = &sysctl_sched_shares_ratelimit,
292 .maxlen = sizeof(unsigned int),
294 .proc_handler = sched_proc_update_handler,
295 .extra1 = &min_sched_shares_ratelimit,
296 .extra2 = &max_sched_shares_ratelimit,
299 .procname = "sched_tunable_scaling",
300 .data = &sysctl_sched_tunable_scaling,
301 .maxlen = sizeof(enum sched_tunable_scaling),
303 .proc_handler = sched_proc_update_handler,
304 .extra1 = &min_sched_tunable_scaling,
305 .extra2 = &max_sched_tunable_scaling,
308 .procname = "sched_shares_thresh",
309 .data = &sysctl_sched_shares_thresh,
310 .maxlen = sizeof(unsigned int),
312 .proc_handler = proc_dointvec_minmax,
316 .procname = "sched_migration_cost",
317 .data = &sysctl_sched_migration_cost,
318 .maxlen = sizeof(unsigned int),
320 .proc_handler = proc_dointvec,
323 .procname = "sched_nr_migrate",
324 .data = &sysctl_sched_nr_migrate,
325 .maxlen = sizeof(unsigned int),
327 .proc_handler = proc_dointvec,
330 .procname = "sched_time_avg",
331 .data = &sysctl_sched_time_avg,
332 .maxlen = sizeof(unsigned int),
334 .proc_handler = proc_dointvec,
337 .procname = "timer_migration",
338 .data = &sysctl_timer_migration,
339 .maxlen = sizeof(unsigned int),
341 .proc_handler = proc_dointvec_minmax,
347 .procname = "sched_rt_period_us",
348 .data = &sysctl_sched_rt_period,
349 .maxlen = sizeof(unsigned int),
351 .proc_handler = sched_rt_handler,
354 .procname = "sched_rt_runtime_us",
355 .data = &sysctl_sched_rt_runtime,
356 .maxlen = sizeof(int),
358 .proc_handler = sched_rt_handler,
361 .procname = "sched_compat_yield",
362 .data = &sysctl_sched_compat_yield,
363 .maxlen = sizeof(unsigned int),
365 .proc_handler = proc_dointvec,
367 #ifdef CONFIG_PROVE_LOCKING
369 .procname = "prove_locking",
370 .data = &prove_locking,
371 .maxlen = sizeof(int),
373 .proc_handler = proc_dointvec,
376 #ifdef CONFIG_LOCK_STAT
378 .procname = "lock_stat",
380 .maxlen = sizeof(int),
382 .proc_handler = proc_dointvec,
387 .data = &panic_timeout,
388 .maxlen = sizeof(int),
390 .proc_handler = proc_dointvec,
393 .procname = "core_uses_pid",
394 .data = &core_uses_pid,
395 .maxlen = sizeof(int),
397 .proc_handler = proc_dointvec,
400 .procname = "core_pattern",
401 .data = core_pattern,
402 .maxlen = CORENAME_MAX_SIZE,
404 .proc_handler = proc_dostring,
407 .procname = "core_pipe_limit",
408 .data = &core_pipe_limit,
409 .maxlen = sizeof(unsigned int),
411 .proc_handler = proc_dointvec,
413 #ifdef CONFIG_PROC_SYSCTL
415 .procname = "tainted",
416 .maxlen = sizeof(long),
418 .proc_handler = proc_taint,
421 #ifdef CONFIG_LATENCYTOP
423 .procname = "latencytop",
424 .data = &latencytop_enabled,
425 .maxlen = sizeof(int),
427 .proc_handler = proc_dointvec,
430 #ifdef CONFIG_BLK_DEV_INITRD
432 .procname = "real-root-dev",
433 .data = &real_root_dev,
434 .maxlen = sizeof(int),
436 .proc_handler = proc_dointvec,
440 .procname = "print-fatal-signals",
441 .data = &print_fatal_signals,
442 .maxlen = sizeof(int),
444 .proc_handler = proc_dointvec,
448 .procname = "reboot-cmd",
449 .data = reboot_command,
452 .proc_handler = proc_dostring,
455 .procname = "stop-a",
456 .data = &stop_a_enabled,
457 .maxlen = sizeof (int),
459 .proc_handler = proc_dointvec,
462 .procname = "scons-poweroff",
463 .data = &scons_pwroff,
464 .maxlen = sizeof (int),
466 .proc_handler = proc_dointvec,
469 #ifdef CONFIG_SPARC64
471 .procname = "tsb-ratio",
472 .data = &sysctl_tsb_ratio,
473 .maxlen = sizeof (int),
475 .proc_handler = proc_dointvec,
480 .procname = "soft-power",
481 .data = &pwrsw_enabled,
482 .maxlen = sizeof (int),
484 .proc_handler = proc_dointvec,
487 .procname = "unaligned-trap",
488 .data = &unaligned_enabled,
489 .maxlen = sizeof (int),
491 .proc_handler = proc_dointvec,
495 .procname = "ctrl-alt-del",
497 .maxlen = sizeof(int),
499 .proc_handler = proc_dointvec,
501 #ifdef CONFIG_FUNCTION_TRACER
503 .procname = "ftrace_enabled",
504 .data = &ftrace_enabled,
505 .maxlen = sizeof(int),
507 .proc_handler = ftrace_enable_sysctl,
510 #ifdef CONFIG_STACK_TRACER
512 .procname = "stack_tracer_enabled",
513 .data = &stack_tracer_enabled,
514 .maxlen = sizeof(int),
516 .proc_handler = stack_trace_sysctl,
519 #ifdef CONFIG_TRACING
521 .procname = "ftrace_dump_on_oops",
522 .data = &ftrace_dump_on_oops,
523 .maxlen = sizeof(int),
525 .proc_handler = proc_dointvec,
528 #ifdef CONFIG_MODULES
530 .procname = "modprobe",
531 .data = &modprobe_path,
532 .maxlen = KMOD_PATH_LEN,
534 .proc_handler = proc_dostring,
537 .procname = "modules_disabled",
538 .data = &modules_disabled,
539 .maxlen = sizeof(int),
541 /* only handle a transition from default "0" to "1" */
542 .proc_handler = proc_dointvec_minmax,
547 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
549 .procname = "hotplug",
550 .data = &uevent_helper,
551 .maxlen = UEVENT_HELPER_PATH_LEN,
553 .proc_handler = proc_dostring,
556 #ifdef CONFIG_CHR_DEV_SG
558 .procname = "sg-big-buff",
559 .data = &sg_big_buff,
560 .maxlen = sizeof (int),
562 .proc_handler = proc_dointvec,
565 #ifdef CONFIG_BSD_PROCESS_ACCT
569 .maxlen = 3*sizeof(int),
571 .proc_handler = proc_dointvec,
574 #ifdef CONFIG_MAGIC_SYSRQ
577 .data = &__sysrq_enabled,
578 .maxlen = sizeof (int),
580 .proc_handler = proc_dointvec,
583 #ifdef CONFIG_PROC_SYSCTL
585 .procname = "cad_pid",
587 .maxlen = sizeof (int),
589 .proc_handler = proc_do_cad_pid,
593 .procname = "threads-max",
594 .data = &max_threads,
595 .maxlen = sizeof(int),
597 .proc_handler = proc_dointvec,
600 .procname = "random",
602 .child = random_table,
605 .procname = "overflowuid",
606 .data = &overflowuid,
607 .maxlen = sizeof(int),
609 .proc_handler = proc_dointvec_minmax,
610 .extra1 = &minolduid,
611 .extra2 = &maxolduid,
614 .procname = "overflowgid",
615 .data = &overflowgid,
616 .maxlen = sizeof(int),
618 .proc_handler = proc_dointvec_minmax,
619 .extra1 = &minolduid,
620 .extra2 = &maxolduid,
623 #ifdef CONFIG_MATHEMU
625 .procname = "ieee_emulation_warnings",
626 .data = &sysctl_ieee_emulation_warnings,
627 .maxlen = sizeof(int),
629 .proc_handler = proc_dointvec,
633 .procname = "userprocess_debug",
634 .data = &sysctl_userprocess_debug,
635 .maxlen = sizeof(int),
637 .proc_handler = proc_dointvec,
641 .procname = "pid_max",
643 .maxlen = sizeof (int),
645 .proc_handler = proc_dointvec_minmax,
646 .extra1 = &pid_max_min,
647 .extra2 = &pid_max_max,
649 #if defined(CONFIG_MODULES) && defined(CONFIG_ENTERPRISE_SUPPORT)
651 .ctl_name = CTL_UNNUMBERED,
652 .procname = "unsupported",
653 .data = &unsupported,
654 .maxlen = sizeof(int),
656 .proc_handler = &proc_dointvec,
660 .procname = "panic_on_oops",
661 .data = &panic_on_oops,
662 .maxlen = sizeof(int),
664 .proc_handler = proc_dointvec,
666 #if defined CONFIG_PRINTK
668 .procname = "printk",
669 .data = &console_loglevel,
670 .maxlen = 4*sizeof(int),
672 .proc_handler = proc_dointvec,
675 .procname = "printk_ratelimit",
676 .data = &printk_ratelimit_state.interval,
677 .maxlen = sizeof(int),
679 .proc_handler = proc_dointvec_jiffies,
682 .procname = "printk_ratelimit_burst",
683 .data = &printk_ratelimit_state.burst,
684 .maxlen = sizeof(int),
686 .proc_handler = proc_dointvec,
689 .procname = "printk_delay",
690 .data = &printk_delay_msec,
691 .maxlen = sizeof(int),
693 .proc_handler = proc_dointvec_minmax,
695 .extra2 = &ten_thousand,
699 .procname = "ngroups_max",
700 .data = &ngroups_max,
701 .maxlen = sizeof (int),
703 .proc_handler = proc_dointvec,
705 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
707 .procname = "unknown_nmi_panic",
708 .data = &unknown_nmi_panic,
709 .maxlen = sizeof (int),
711 .proc_handler = proc_dointvec,
714 .procname = "nmi_watchdog",
715 .data = &nmi_watchdog_enabled,
716 .maxlen = sizeof (int),
718 .proc_handler = proc_nmi_enabled,
721 #if defined(CONFIG_X86)
723 .procname = "panic_on_unrecovered_nmi",
724 .data = &panic_on_unrecovered_nmi,
725 .maxlen = sizeof(int),
727 .proc_handler = proc_dointvec,
730 .procname = "panic_on_io_nmi",
731 .data = &panic_on_io_nmi,
732 .maxlen = sizeof(int),
734 .proc_handler = proc_dointvec,
737 .procname = "bootloader_type",
738 .data = &bootloader_type,
739 .maxlen = sizeof (int),
741 .proc_handler = proc_dointvec,
744 .procname = "bootloader_version",
745 .data = &bootloader_version,
746 .maxlen = sizeof (int),
748 .proc_handler = proc_dointvec,
751 .procname = "kstack_depth_to_print",
752 .data = &kstack_depth_to_print,
753 .maxlen = sizeof(int),
755 .proc_handler = proc_dointvec,
758 .procname = "io_delay_type",
759 .data = &io_delay_type,
760 .maxlen = sizeof(int),
762 .proc_handler = proc_dointvec,
765 #if defined(CONFIG_MMU)
767 .procname = "randomize_va_space",
768 .data = &randomize_va_space,
769 .maxlen = sizeof(int),
771 .proc_handler = proc_dointvec,
775 .procname = "suid_dumpable",
776 .data = &suid_dumpable,
777 .maxlen = sizeof(int),
779 .proc_handler = proc_dointvec,
781 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
783 .procname = "spin_retry",
785 .maxlen = sizeof (int),
787 .proc_handler = proc_dointvec,
790 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
792 .procname = "acpi_video_flags",
793 .data = &acpi_realmode_flags,
794 .maxlen = sizeof (unsigned long),
796 .proc_handler = proc_doulongvec_minmax,
801 .procname = "ignore-unaligned-usertrap",
802 .data = &no_unaligned_warning,
803 .maxlen = sizeof (int),
805 .proc_handler = proc_dointvec,
808 .procname = "unaligned-dump-stack",
809 .data = &unaligned_dump_stack,
810 .maxlen = sizeof (int),
812 .proc_handler = proc_dointvec,
815 #ifdef CONFIG_DETECT_SOFTLOCKUP
817 .procname = "softlockup_panic",
818 .data = &softlockup_panic,
819 .maxlen = sizeof(int),
821 .proc_handler = proc_dointvec_minmax,
826 .procname = "softlockup_thresh",
827 .data = &softlockup_thresh,
828 .maxlen = sizeof(int),
830 .proc_handler = proc_dosoftlockup_thresh,
835 #ifdef CONFIG_DETECT_HUNG_TASK
837 .procname = "hung_task_panic",
838 .data = &sysctl_hung_task_panic,
839 .maxlen = sizeof(int),
841 .proc_handler = proc_dointvec_minmax,
846 .procname = "hung_task_check_count",
847 .data = &sysctl_hung_task_check_count,
848 .maxlen = sizeof(unsigned long),
850 .proc_handler = proc_doulongvec_minmax,
853 .procname = "hung_task_timeout_secs",
854 .data = &sysctl_hung_task_timeout_secs,
855 .maxlen = sizeof(unsigned long),
857 .proc_handler = proc_dohung_task_timeout_secs,
860 .procname = "hung_task_warnings",
861 .data = &sysctl_hung_task_warnings,
862 .maxlen = sizeof(unsigned long),
864 .proc_handler = proc_doulongvec_minmax,
869 .procname = "compat-log",
871 .maxlen = sizeof (int),
873 .proc_handler = proc_dointvec,
876 #ifdef CONFIG_RT_MUTEXES
878 .procname = "max_lock_depth",
879 .data = &max_lock_depth,
880 .maxlen = sizeof(int),
882 .proc_handler = proc_dointvec,
886 .procname = "poweroff_cmd",
887 .data = &poweroff_cmd,
888 .maxlen = POWEROFF_CMD_PATH_LEN,
890 .proc_handler = proc_dostring,
896 .child = key_sysctls,
899 #ifdef CONFIG_RCU_TORTURE_TEST
901 .procname = "rcutorture_runnable",
902 .data = &rcutorture_runnable,
903 .maxlen = sizeof(int),
905 .proc_handler = proc_dointvec,
908 #ifdef CONFIG_SLOW_WORK
910 .procname = "slow-work",
912 .child = slow_work_sysctls,
915 #ifdef CONFIG_PERF_EVENTS
917 .procname = "perf_event_paranoid",
918 .data = &sysctl_perf_event_paranoid,
919 .maxlen = sizeof(sysctl_perf_event_paranoid),
921 .proc_handler = proc_dointvec,
924 .procname = "perf_event_mlock_kb",
925 .data = &sysctl_perf_event_mlock,
926 .maxlen = sizeof(sysctl_perf_event_mlock),
928 .proc_handler = proc_dointvec,
931 .procname = "perf_event_max_sample_rate",
932 .data = &sysctl_perf_event_sample_rate,
933 .maxlen = sizeof(sysctl_perf_event_sample_rate),
935 .proc_handler = proc_dointvec,
938 #ifdef CONFIG_KMEMCHECK
940 .procname = "kmemcheck",
941 .data = &kmemcheck_enabled,
942 .maxlen = sizeof(int),
944 .proc_handler = proc_dointvec,
949 .procname = "blk_iopoll",
950 .data = &blk_iopoll_enabled,
951 .maxlen = sizeof(int),
953 .proc_handler = proc_dointvec,
957 * NOTE: do not add new entries to this table unless you have read
958 * Documentation/sysctl/ctl_unnumbered.txt
963 static struct ctl_table vm_table[] = {
965 .procname = "overcommit_memory",
966 .data = &sysctl_overcommit_memory,
967 .maxlen = sizeof(sysctl_overcommit_memory),
969 .proc_handler = proc_dointvec,
972 .procname = "panic_on_oom",
973 .data = &sysctl_panic_on_oom,
974 .maxlen = sizeof(sysctl_panic_on_oom),
976 .proc_handler = proc_dointvec,
979 .procname = "oom_kill_allocating_task",
980 .data = &sysctl_oom_kill_allocating_task,
981 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
983 .proc_handler = proc_dointvec,
986 .procname = "oom_dump_tasks",
987 .data = &sysctl_oom_dump_tasks,
988 .maxlen = sizeof(sysctl_oom_dump_tasks),
990 .proc_handler = proc_dointvec,
993 .procname = "overcommit_ratio",
994 .data = &sysctl_overcommit_ratio,
995 .maxlen = sizeof(sysctl_overcommit_ratio),
997 .proc_handler = proc_dointvec,
1000 .procname = "page-cluster",
1001 .data = &page_cluster,
1002 .maxlen = sizeof(int),
1004 .proc_handler = proc_dointvec,
1007 .procname = "dirty_background_ratio",
1008 .data = &dirty_background_ratio,
1009 .maxlen = sizeof(dirty_background_ratio),
1011 .proc_handler = dirty_background_ratio_handler,
1013 .extra2 = &one_hundred,
1016 .procname = "dirty_background_bytes",
1017 .data = &dirty_background_bytes,
1018 .maxlen = sizeof(dirty_background_bytes),
1020 .proc_handler = dirty_background_bytes_handler,
1024 .procname = "dirty_ratio",
1025 .data = &vm_dirty_ratio,
1026 .maxlen = sizeof(vm_dirty_ratio),
1028 .proc_handler = dirty_ratio_handler,
1030 .extra2 = &one_hundred,
1033 .procname = "dirty_bytes",
1034 .data = &vm_dirty_bytes,
1035 .maxlen = sizeof(vm_dirty_bytes),
1037 .proc_handler = dirty_bytes_handler,
1038 .extra1 = &dirty_bytes_min,
1041 .procname = "dirty_writeback_centisecs",
1042 .data = &dirty_writeback_interval,
1043 .maxlen = sizeof(dirty_writeback_interval),
1045 .proc_handler = dirty_writeback_centisecs_handler,
1048 .procname = "dirty_expire_centisecs",
1049 .data = &dirty_expire_interval,
1050 .maxlen = sizeof(dirty_expire_interval),
1052 .proc_handler = proc_dointvec,
1055 .procname = "nr_pdflush_threads",
1056 .data = &nr_pdflush_threads,
1057 .maxlen = sizeof nr_pdflush_threads,
1058 .mode = 0444 /* read-only*/,
1059 .proc_handler = proc_dointvec,
1062 .procname = "swappiness",
1063 .data = &vm_swappiness,
1064 .maxlen = sizeof(vm_swappiness),
1066 .proc_handler = proc_dointvec_minmax,
1068 .extra2 = &one_hundred,
1070 #ifdef CONFIG_HUGETLB_PAGE
1072 .procname = "nr_hugepages",
1074 .maxlen = sizeof(unsigned long),
1076 .proc_handler = hugetlb_sysctl_handler,
1077 .extra1 = (void *)&hugetlb_zero,
1078 .extra2 = (void *)&hugetlb_infinity,
1082 .procname = "nr_hugepages_mempolicy",
1084 .maxlen = sizeof(unsigned long),
1086 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1087 .extra1 = (void *)&hugetlb_zero,
1088 .extra2 = (void *)&hugetlb_infinity,
1092 .procname = "hugetlb_shm_group",
1093 .data = &sysctl_hugetlb_shm_group,
1094 .maxlen = sizeof(gid_t),
1096 .proc_handler = proc_dointvec,
1099 .procname = "hugepages_treat_as_movable",
1100 .data = &hugepages_treat_as_movable,
1101 .maxlen = sizeof(int),
1103 .proc_handler = hugetlb_treat_movable_handler,
1106 .procname = "nr_overcommit_hugepages",
1108 .maxlen = sizeof(unsigned long),
1110 .proc_handler = hugetlb_overcommit_handler,
1111 .extra1 = (void *)&hugetlb_zero,
1112 .extra2 = (void *)&hugetlb_infinity,
1116 .procname = "lowmem_reserve_ratio",
1117 .data = &sysctl_lowmem_reserve_ratio,
1118 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1120 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1123 .procname = "drop_caches",
1124 .data = &sysctl_drop_caches,
1125 .maxlen = sizeof(int),
1127 .proc_handler = drop_caches_sysctl_handler,
1130 .procname = "min_free_kbytes",
1131 .data = &min_free_kbytes,
1132 .maxlen = sizeof(min_free_kbytes),
1134 .proc_handler = min_free_kbytes_sysctl_handler,
1138 .procname = "percpu_pagelist_fraction",
1139 .data = &percpu_pagelist_fraction,
1140 .maxlen = sizeof(percpu_pagelist_fraction),
1142 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1143 .extra1 = &min_percpu_pagelist_fract,
1147 .procname = "max_map_count",
1148 .data = &sysctl_max_map_count,
1149 .maxlen = sizeof(sysctl_max_map_count),
1151 .proc_handler = proc_dointvec_minmax,
1156 .procname = "nr_trim_pages",
1157 .data = &sysctl_nr_trim_pages,
1158 .maxlen = sizeof(sysctl_nr_trim_pages),
1160 .proc_handler = proc_dointvec_minmax,
1165 .procname = "laptop_mode",
1166 .data = &laptop_mode,
1167 .maxlen = sizeof(laptop_mode),
1169 .proc_handler = proc_dointvec_jiffies,
1172 .procname = "block_dump",
1173 .data = &block_dump,
1174 .maxlen = sizeof(block_dump),
1176 .proc_handler = proc_dointvec,
1180 .procname = "vfs_cache_pressure",
1181 .data = &sysctl_vfs_cache_pressure,
1182 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1184 .proc_handler = proc_dointvec,
1187 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1189 .procname = "legacy_va_layout",
1190 .data = &sysctl_legacy_va_layout,
1191 .maxlen = sizeof(sysctl_legacy_va_layout),
1193 .proc_handler = proc_dointvec,
1199 .procname = "zone_reclaim_mode",
1200 .data = &zone_reclaim_mode,
1201 .maxlen = sizeof(zone_reclaim_mode),
1203 .proc_handler = proc_dointvec,
1207 .procname = "min_unmapped_ratio",
1208 .data = &sysctl_min_unmapped_ratio,
1209 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1211 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1213 .extra2 = &one_hundred,
1216 .procname = "min_slab_ratio",
1217 .data = &sysctl_min_slab_ratio,
1218 .maxlen = sizeof(sysctl_min_slab_ratio),
1220 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1222 .extra2 = &one_hundred,
1227 .procname = "stat_interval",
1228 .data = &sysctl_stat_interval,
1229 .maxlen = sizeof(sysctl_stat_interval),
1231 .proc_handler = proc_dointvec_jiffies,
1236 .procname = "mmap_min_addr",
1237 .data = &dac_mmap_min_addr,
1238 .maxlen = sizeof(unsigned long),
1240 .proc_handler = mmap_min_addr_handler,
1245 .procname = "numa_zonelist_order",
1246 .data = &numa_zonelist_order,
1247 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1249 .proc_handler = numa_zonelist_order_handler,
1252 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1253 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1255 .procname = "vdso_enabled",
1256 .data = &vdso_enabled,
1257 .maxlen = sizeof(vdso_enabled),
1259 .proc_handler = proc_dointvec,
1263 #ifdef CONFIG_HIGHMEM
1265 .procname = "highmem_is_dirtyable",
1266 .data = &vm_highmem_is_dirtyable,
1267 .maxlen = sizeof(vm_highmem_is_dirtyable),
1269 .proc_handler = proc_dointvec_minmax,
1275 .procname = "scan_unevictable_pages",
1276 .data = &scan_unevictable_pages,
1277 .maxlen = sizeof(scan_unevictable_pages),
1279 .proc_handler = scan_unevictable_handler,
1282 .procname = "heap-stack-gap",
1283 .data = &heap_stack_gap,
1284 .maxlen = sizeof(int),
1286 .proc_handler = proc_dointvec,
1288 #ifdef CONFIG_MEMORY_FAILURE
1290 .procname = "memory_failure_early_kill",
1291 .data = &sysctl_memory_failure_early_kill,
1292 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1294 .proc_handler = proc_dointvec_minmax,
1299 .procname = "memory_failure_recovery",
1300 .data = &sysctl_memory_failure_recovery,
1301 .maxlen = sizeof(sysctl_memory_failure_recovery),
1303 .proc_handler = proc_dointvec_minmax,
1310 * NOTE: do not add new entries to this table unless you have read
1311 * Documentation/sysctl/ctl_unnumbered.txt
1316 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1317 static struct ctl_table binfmt_misc_table[] = {
1322 static struct ctl_table fs_table[] = {
1324 .procname = "inode-nr",
1325 .data = &inodes_stat,
1326 .maxlen = 2*sizeof(int),
1328 .proc_handler = proc_dointvec,
1331 .procname = "inode-state",
1332 .data = &inodes_stat,
1333 .maxlen = 7*sizeof(int),
1335 .proc_handler = proc_dointvec,
1338 .procname = "file-nr",
1339 .data = &files_stat,
1340 .maxlen = 3*sizeof(int),
1342 .proc_handler = proc_nr_files,
1345 .procname = "file-max",
1346 .data = &files_stat.max_files,
1347 .maxlen = sizeof(int),
1349 .proc_handler = proc_dointvec,
1352 .procname = "nr_open",
1353 .data = &sysctl_nr_open,
1354 .maxlen = sizeof(int),
1356 .proc_handler = proc_dointvec_minmax,
1357 .extra1 = &sysctl_nr_open_min,
1358 .extra2 = &sysctl_nr_open_max,
1361 .procname = "dentry-state",
1362 .data = &dentry_stat,
1363 .maxlen = 6*sizeof(int),
1365 .proc_handler = proc_dointvec,
1368 .procname = "overflowuid",
1369 .data = &fs_overflowuid,
1370 .maxlen = sizeof(int),
1372 .proc_handler = proc_dointvec_minmax,
1373 .extra1 = &minolduid,
1374 .extra2 = &maxolduid,
1377 .procname = "overflowgid",
1378 .data = &fs_overflowgid,
1379 .maxlen = sizeof(int),
1381 .proc_handler = proc_dointvec_minmax,
1382 .extra1 = &minolduid,
1383 .extra2 = &maxolduid,
1385 #ifdef CONFIG_FILE_LOCKING
1387 .procname = "leases-enable",
1388 .data = &leases_enable,
1389 .maxlen = sizeof(int),
1391 .proc_handler = proc_dointvec,
1394 #ifdef CONFIG_DNOTIFY
1396 .procname = "dir-notify-enable",
1397 .data = &dir_notify_enable,
1398 .maxlen = sizeof(int),
1400 .proc_handler = proc_dointvec,
1404 #ifdef CONFIG_FILE_LOCKING
1406 .procname = "lease-break-time",
1407 .data = &lease_break_time,
1408 .maxlen = sizeof(int),
1410 .proc_handler = proc_dointvec,
1415 .procname = "aio-nr",
1417 .maxlen = sizeof(aio_nr),
1419 .proc_handler = proc_doulongvec_minmax,
1422 .procname = "aio-max-nr",
1423 .data = &aio_max_nr,
1424 .maxlen = sizeof(aio_max_nr),
1426 .proc_handler = proc_doulongvec_minmax,
1428 #endif /* CONFIG_AIO */
1429 #ifdef CONFIG_INOTIFY_USER
1431 .procname = "inotify",
1433 .child = inotify_table,
1438 .procname = "epoll",
1440 .child = epoll_table,
1445 .procname = "suid_dumpable",
1446 .data = &suid_dumpable,
1447 .maxlen = sizeof(int),
1449 .proc_handler = proc_dointvec_minmax,
1453 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1455 .procname = "binfmt_misc",
1457 .child = binfmt_misc_table,
1461 * NOTE: do not add new entries to this table unless you have read
1462 * Documentation/sysctl/ctl_unnumbered.txt
1467 static struct ctl_table debug_table[] = {
1468 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1470 .procname = "exception-trace",
1471 .data = &show_unhandled_signals,
1472 .maxlen = sizeof(int),
1474 .proc_handler = proc_dointvec
1480 static struct ctl_table dev_table[] = {
1484 static DEFINE_SPINLOCK(sysctl_lock);
1486 /* called under sysctl_lock */
1487 static int use_table(struct ctl_table_header *p)
1489 if (unlikely(p->unregistering))
1495 /* called under sysctl_lock */
1496 static void unuse_table(struct ctl_table_header *p)
1499 if (unlikely(p->unregistering))
1500 complete(p->unregistering);
1503 /* called under sysctl_lock, will reacquire if has to wait */
1504 static void start_unregistering(struct ctl_table_header *p)
1507 * if p->used is 0, nobody will ever touch that entry again;
1508 * we'll eliminate all paths to it before dropping sysctl_lock
1510 if (unlikely(p->used)) {
1511 struct completion wait;
1512 init_completion(&wait);
1513 p->unregistering = &wait;
1514 spin_unlock(&sysctl_lock);
1515 wait_for_completion(&wait);
1516 spin_lock(&sysctl_lock);
1518 /* anything non-NULL; we'll never dereference it */
1519 p->unregistering = ERR_PTR(-EINVAL);
1522 * do not remove from the list until nobody holds it; walking the
1523 * list in do_sysctl() relies on that.
1525 list_del_init(&p->ctl_entry);
1528 void sysctl_head_get(struct ctl_table_header *head)
1530 spin_lock(&sysctl_lock);
1532 spin_unlock(&sysctl_lock);
1535 void sysctl_head_put(struct ctl_table_header *head)
1537 spin_lock(&sysctl_lock);
1540 spin_unlock(&sysctl_lock);
1543 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1547 spin_lock(&sysctl_lock);
1548 if (!use_table(head))
1549 head = ERR_PTR(-ENOENT);
1550 spin_unlock(&sysctl_lock);
1554 void sysctl_head_finish(struct ctl_table_header *head)
1558 spin_lock(&sysctl_lock);
1560 spin_unlock(&sysctl_lock);
1563 static struct ctl_table_set *
1564 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1566 struct ctl_table_set *set = &root->default_set;
1568 set = root->lookup(root, namespaces);
1572 static struct list_head *
1573 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1575 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1579 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1580 struct ctl_table_header *prev)
1582 struct ctl_table_root *root;
1583 struct list_head *header_list;
1584 struct ctl_table_header *head;
1585 struct list_head *tmp;
1587 spin_lock(&sysctl_lock);
1590 tmp = &prev->ctl_entry;
1594 tmp = &root_table_header.ctl_entry;
1596 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1598 if (!use_table(head))
1600 spin_unlock(&sysctl_lock);
1605 header_list = lookup_header_list(root, namespaces);
1606 if (tmp != header_list)
1610 root = list_entry(root->root_list.next,
1611 struct ctl_table_root, root_list);
1612 if (root == &sysctl_table_root)
1614 header_list = lookup_header_list(root, namespaces);
1615 } while (list_empty(header_list));
1616 tmp = header_list->next;
1619 spin_unlock(&sysctl_lock);
1623 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1625 return __sysctl_head_next(current->nsproxy, prev);
1628 void register_sysctl_root(struct ctl_table_root *root)
1630 spin_lock(&sysctl_lock);
1631 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1632 spin_unlock(&sysctl_lock);
1636 * sysctl_perm does NOT grant the superuser all rights automatically, because
1637 * some sysctl variables are readonly even to root.
1640 static int test_perm(int mode, int op)
1642 if (!current_euid())
1644 else if (in_egroup_p(0))
1646 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1651 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1656 error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1660 if (root->permissions)
1661 mode = root->permissions(root, current->nsproxy, table);
1665 return test_perm(mode, op);
1668 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1670 for (; table->procname; table++) {
1671 table->parent = parent;
1673 sysctl_set_parent(table, table->child);
1677 static __init int sysctl_init(void)
1679 sysctl_set_parent(NULL, root_table);
1680 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1683 err = sysctl_check_table(current->nsproxy, root_table);
1689 core_initcall(sysctl_init);
1691 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1692 struct ctl_table *table)
1694 struct ctl_table *p;
1695 const char *s = branch->procname;
1697 /* branch should have named subdirectory as its first element */
1698 if (!s || !branch->child)
1701 /* ... and nothing else */
1702 if (branch[1].procname)
1705 /* table should contain subdirectory with the same name */
1706 for (p = table; p->procname; p++) {
1709 if (p->procname && strcmp(p->procname, s) == 0)
1715 /* see if attaching q to p would be an improvement */
1716 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1718 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1719 struct ctl_table *next;
1721 int not_in_parent = !p->attached_by;
1723 while ((next = is_branch_in(by, to)) != NULL) {
1724 if (by == q->attached_by)
1726 if (to == p->attached_by)
1732 if (is_better && not_in_parent) {
1733 q->attached_by = by;
1734 q->attached_to = to;
1740 * __register_sysctl_paths - register a sysctl hierarchy
1741 * @root: List of sysctl headers to register on
1742 * @namespaces: Data to compute which lists of sysctl entries are visible
1743 * @path: The path to the directory the sysctl table is in.
1744 * @table: the top-level table structure
1746 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1747 * array. A completely 0 filled entry terminates the table.
1749 * The members of the &struct ctl_table structure are used as follows:
1751 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1752 * enter a sysctl file
1754 * data - a pointer to data for use by proc_handler
1756 * maxlen - the maximum size in bytes of the data
1758 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1760 * child - a pointer to the child sysctl table if this entry is a directory, or
1763 * proc_handler - the text handler routine (described below)
1765 * de - for internal use by the sysctl routines
1767 * extra1, extra2 - extra pointers usable by the proc handler routines
1769 * Leaf nodes in the sysctl tree will be represented by a single file
1770 * under /proc; non-leaf nodes will be represented by directories.
1772 * sysctl(2) can automatically manage read and write requests through
1773 * the sysctl table. The data and maxlen fields of the ctl_table
1774 * struct enable minimal validation of the values being written to be
1775 * performed, and the mode field allows minimal authentication.
1777 * There must be a proc_handler routine for any terminal nodes
1778 * mirrored under /proc/sys (non-terminals are handled by a built-in
1779 * directory handler). Several default handlers are available to
1780 * cover common cases -
1782 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1783 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1784 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1786 * It is the handler's job to read the input buffer from user memory
1787 * and process it. The handler should return 0 on success.
1789 * This routine returns %NULL on a failure to register, and a pointer
1790 * to the table header on success.
1792 struct ctl_table_header *__register_sysctl_paths(
1793 struct ctl_table_root *root,
1794 struct nsproxy *namespaces,
1795 const struct ctl_path *path, struct ctl_table *table)
1797 struct ctl_table_header *header;
1798 struct ctl_table *new, **prevp;
1799 unsigned int n, npath;
1800 struct ctl_table_set *set;
1802 /* Count the path components */
1803 for (npath = 0; path[npath].procname; ++npath)
1807 * For each path component, allocate a 2-element ctl_table array.
1808 * The first array element will be filled with the sysctl entry
1809 * for this, the second will be the sentinel (procname == 0).
1811 * We allocate everything in one go so that we don't have to
1812 * worry about freeing additional memory in unregister_sysctl_table.
1814 header = kzalloc(sizeof(struct ctl_table_header) +
1815 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1819 new = (struct ctl_table *) (header + 1);
1821 /* Now connect the dots */
1822 prevp = &header->ctl_table;
1823 for (n = 0; n < npath; ++n, ++path) {
1824 /* Copy the procname */
1825 new->procname = path->procname;
1829 prevp = &new->child;
1834 header->ctl_table_arg = table;
1836 INIT_LIST_HEAD(&header->ctl_entry);
1838 header->unregistering = NULL;
1839 header->root = root;
1840 sysctl_set_parent(NULL, header->ctl_table);
1842 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1843 if (sysctl_check_table(namespaces, header->ctl_table)) {
1848 spin_lock(&sysctl_lock);
1849 header->set = lookup_header_set(root, namespaces);
1850 header->attached_by = header->ctl_table;
1851 header->attached_to = root_table;
1852 header->parent = &root_table_header;
1853 for (set = header->set; set; set = set->parent) {
1854 struct ctl_table_header *p;
1855 list_for_each_entry(p, &set->list, ctl_entry) {
1856 if (p->unregistering)
1858 try_attach(p, header);
1861 header->parent->count++;
1862 list_add_tail(&header->ctl_entry, &header->set->list);
1863 spin_unlock(&sysctl_lock);
1869 * register_sysctl_table_path - register a sysctl table hierarchy
1870 * @path: The path to the directory the sysctl table is in.
1871 * @table: the top-level table structure
1873 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1874 * array. A completely 0 filled entry terminates the table.
1876 * See __register_sysctl_paths for more details.
1878 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1879 struct ctl_table *table)
1881 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1886 * register_sysctl_table - register a sysctl table hierarchy
1887 * @table: the top-level table structure
1889 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1890 * array. A completely 0 filled entry terminates the table.
1892 * See register_sysctl_paths for more details.
1894 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1896 static const struct ctl_path null_path[] = { {} };
1898 return register_sysctl_paths(null_path, table);
1902 * unregister_sysctl_table - unregister a sysctl table hierarchy
1903 * @header: the header returned from register_sysctl_table
1905 * Unregisters the sysctl table and all children. proc entries may not
1906 * actually be removed until they are no longer used by anyone.
1908 void unregister_sysctl_table(struct ctl_table_header * header)
1915 spin_lock(&sysctl_lock);
1916 start_unregistering(header);
1917 if (!--header->parent->count) {
1919 kfree(header->parent);
1921 if (!--header->count)
1923 spin_unlock(&sysctl_lock);
1926 int sysctl_is_seen(struct ctl_table_header *p)
1928 struct ctl_table_set *set = p->set;
1930 spin_lock(&sysctl_lock);
1931 if (p->unregistering)
1933 else if (!set->is_seen)
1936 res = set->is_seen(set);
1937 spin_unlock(&sysctl_lock);
1941 void setup_sysctl_set(struct ctl_table_set *p,
1942 struct ctl_table_set *parent,
1943 int (*is_seen)(struct ctl_table_set *))
1945 INIT_LIST_HEAD(&p->list);
1946 p->parent = parent ? parent : &sysctl_table_root.default_set;
1947 p->is_seen = is_seen;
1950 #else /* !CONFIG_SYSCTL */
1951 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1956 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1957 struct ctl_table *table)
1962 void unregister_sysctl_table(struct ctl_table_header * table)
1966 void setup_sysctl_set(struct ctl_table_set *p,
1967 struct ctl_table_set *parent,
1968 int (*is_seen)(struct ctl_table_set *))
1972 void sysctl_head_put(struct ctl_table_header *head)
1976 #endif /* CONFIG_SYSCTL */
1982 #ifdef CONFIG_PROC_SYSCTL
1984 static int _proc_do_string(void* data, int maxlen, int write,
1985 void __user *buffer,
1986 size_t *lenp, loff_t *ppos)
1992 if (!data || !maxlen || !*lenp) {
2000 while (len < *lenp) {
2001 if (get_user(c, p++))
2003 if (c == 0 || c == '\n')
2009 if(copy_from_user(data, buffer, len))
2011 ((char *) data)[len] = 0;
2029 if(copy_to_user(buffer, data, len))
2032 if(put_user('\n', ((char __user *) buffer) + len))
2043 * proc_dostring - read a string sysctl
2044 * @table: the sysctl table
2045 * @write: %TRUE if this is a write to the sysctl file
2046 * @buffer: the user buffer
2047 * @lenp: the size of the user buffer
2048 * @ppos: file position
2050 * Reads/writes a string from/to the user buffer. If the kernel
2051 * buffer provided is not large enough to hold the string, the
2052 * string is truncated. The copied string is %NULL-terminated.
2053 * If the string is being read by the user process, it is copied
2054 * and a newline '\n' is added. It is truncated if the buffer is
2057 * Returns 0 on success.
2059 int proc_dostring(struct ctl_table *table, int write,
2060 void __user *buffer, size_t *lenp, loff_t *ppos)
2062 return _proc_do_string(table->data, table->maxlen, write,
2063 buffer, lenp, ppos);
2067 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2069 int write, void *data)
2072 *valp = *negp ? -*lvalp : *lvalp;
2077 *lvalp = (unsigned long)-val;
2080 *lvalp = (unsigned long)val;
2086 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2087 int write, void __user *buffer,
2088 size_t *lenp, loff_t *ppos,
2089 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2090 int write, void *data),
2093 #define TMPBUFLEN 21
2094 int *i, vleft, first = 1, neg;
2098 char buf[TMPBUFLEN], *p;
2099 char __user *s = buffer;
2101 if (!tbl_data || !table->maxlen || !*lenp ||
2102 (*ppos && !write)) {
2107 i = (int *) tbl_data;
2108 vleft = table->maxlen / sizeof(*i);
2112 conv = do_proc_dointvec_conv;
2114 for (; left && vleft--; i++, first=0) {
2129 if (len > sizeof(buf) - 1)
2130 len = sizeof(buf) - 1;
2131 if (copy_from_user(buf, s, len))
2135 if (*p == '-' && left > 1) {
2139 if (*p < '0' || *p > '9')
2142 lval = simple_strtoul(p, &p, 0);
2145 if ((len < left) && *p && !isspace(*p))
2150 if (conv(&neg, &lval, i, 1, data))
2157 if (conv(&neg, &lval, i, 0, data))
2160 sprintf(p, "%s%lu", neg ? "-" : "", lval);
2164 if(copy_to_user(s, buf, len))
2171 if (!write && !first && left) {
2172 if(put_user('\n', s))
2179 if (get_user(c, s++))
2194 static int do_proc_dointvec(struct ctl_table *table, int write,
2195 void __user *buffer, size_t *lenp, loff_t *ppos,
2196 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2197 int write, void *data),
2200 return __do_proc_dointvec(table->data, table, write,
2201 buffer, lenp, ppos, conv, data);
2205 * proc_dointvec - read a vector of integers
2206 * @table: the sysctl table
2207 * @write: %TRUE if this is a write to the sysctl file
2208 * @buffer: the user buffer
2209 * @lenp: the size of the user buffer
2210 * @ppos: file position
2212 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2213 * values from/to the user buffer, treated as an ASCII string.
2215 * Returns 0 on success.
2217 int proc_dointvec(struct ctl_table *table, int write,
2218 void __user *buffer, size_t *lenp, loff_t *ppos)
2220 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2225 * Taint values can only be increased
2226 * This means we can safely use a temporary.
2228 static int proc_taint(struct ctl_table *table, int write,
2229 void __user *buffer, size_t *lenp, loff_t *ppos)
2232 unsigned long tmptaint = get_taint();
2235 if (write && !capable(CAP_SYS_ADMIN))
2240 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2246 * Poor man's atomic or. Not worth adding a primitive
2247 * to everyone's atomic.h for this
2250 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2251 if ((tmptaint >> i) & 1)
2259 struct do_proc_dointvec_minmax_conv_param {
2264 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2266 int write, void *data)
2268 struct do_proc_dointvec_minmax_conv_param *param = data;
2270 int val = *negp ? -*lvalp : *lvalp;
2271 if ((param->min && *param->min > val) ||
2272 (param->max && *param->max < val))
2279 *lvalp = (unsigned long)-val;
2282 *lvalp = (unsigned long)val;
2289 * proc_dointvec_minmax - read a vector of integers with min/max values
2290 * @table: the sysctl table
2291 * @write: %TRUE if this is a write to the sysctl file
2292 * @buffer: the user buffer
2293 * @lenp: the size of the user buffer
2294 * @ppos: file position
2296 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2297 * values from/to the user buffer, treated as an ASCII string.
2299 * This routine will ensure the values are within the range specified by
2300 * table->extra1 (min) and table->extra2 (max).
2302 * Returns 0 on success.
2304 int proc_dointvec_minmax(struct ctl_table *table, int write,
2305 void __user *buffer, size_t *lenp, loff_t *ppos)
2307 struct do_proc_dointvec_minmax_conv_param param = {
2308 .min = (int *) table->extra1,
2309 .max = (int *) table->extra2,
2311 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2312 do_proc_dointvec_minmax_conv, ¶m);
2315 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2316 void __user *buffer,
2317 size_t *lenp, loff_t *ppos,
2318 unsigned long convmul,
2319 unsigned long convdiv)
2321 #define TMPBUFLEN 21
2322 unsigned long *i, *min, *max, val;
2323 int vleft, first=1, neg;
2325 char buf[TMPBUFLEN], *p;
2326 char __user *s = buffer;
2328 if (!data || !table->maxlen || !*lenp ||
2329 (*ppos && !write)) {
2334 i = (unsigned long *) data;
2335 min = (unsigned long *) table->extra1;
2336 max = (unsigned long *) table->extra2;
2337 vleft = table->maxlen / sizeof(unsigned long);
2340 for (; left && vleft--; i++, min++, max++, first=0) {
2355 if (len > TMPBUFLEN-1)
2357 if (copy_from_user(buf, s, len))
2361 if (*p == '-' && left > 1) {
2365 if (*p < '0' || *p > '9')
2367 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2369 if ((len < left) && *p && !isspace(*p))
2378 if ((min && val < *min) || (max && val > *max))
2385 sprintf(p, "%lu", convdiv * (*i) / convmul);
2389 if(copy_to_user(s, buf, len))
2396 if (!write && !first && left) {
2397 if(put_user('\n', s))
2404 if (get_user(c, s++))
2419 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2420 void __user *buffer,
2421 size_t *lenp, loff_t *ppos,
2422 unsigned long convmul,
2423 unsigned long convdiv)
2425 return __do_proc_doulongvec_minmax(table->data, table, write,
2426 buffer, lenp, ppos, convmul, convdiv);
2430 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2431 * @table: the sysctl table
2432 * @write: %TRUE if this is a write to the sysctl file
2433 * @buffer: the user buffer
2434 * @lenp: the size of the user buffer
2435 * @ppos: file position
2437 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2438 * values from/to the user buffer, treated as an ASCII string.
2440 * This routine will ensure the values are within the range specified by
2441 * table->extra1 (min) and table->extra2 (max).
2443 * Returns 0 on success.
2445 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2446 void __user *buffer, size_t *lenp, loff_t *ppos)
2448 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2452 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2453 * @table: the sysctl table
2454 * @write: %TRUE if this is a write to the sysctl file
2455 * @buffer: the user buffer
2456 * @lenp: the size of the user buffer
2457 * @ppos: file position
2459 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2460 * values from/to the user buffer, treated as an ASCII string. The values
2461 * are treated as milliseconds, and converted to jiffies when they are stored.
2463 * This routine will ensure the values are within the range specified by
2464 * table->extra1 (min) and table->extra2 (max).
2466 * Returns 0 on success.
2468 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2469 void __user *buffer,
2470 size_t *lenp, loff_t *ppos)
2472 return do_proc_doulongvec_minmax(table, write, buffer,
2473 lenp, ppos, HZ, 1000l);
2477 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2479 int write, void *data)
2482 if (*lvalp > LONG_MAX / HZ)
2484 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2490 lval = (unsigned long)-val;
2493 lval = (unsigned long)val;
2500 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2502 int write, void *data)
2505 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2507 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2513 lval = (unsigned long)-val;
2516 lval = (unsigned long)val;
2518 *lvalp = jiffies_to_clock_t(lval);
2523 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2525 int write, void *data)
2528 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2534 lval = (unsigned long)-val;
2537 lval = (unsigned long)val;
2539 *lvalp = jiffies_to_msecs(lval);
2545 * proc_dointvec_jiffies - read a vector of integers as seconds
2546 * @table: the sysctl table
2547 * @write: %TRUE if this is a write to the sysctl file
2548 * @buffer: the user buffer
2549 * @lenp: the size of the user buffer
2550 * @ppos: file position
2552 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2553 * values from/to the user buffer, treated as an ASCII string.
2554 * The values read are assumed to be in seconds, and are converted into
2557 * Returns 0 on success.
2559 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2560 void __user *buffer, size_t *lenp, loff_t *ppos)
2562 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2563 do_proc_dointvec_jiffies_conv,NULL);
2567 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2568 * @table: the sysctl table
2569 * @write: %TRUE if this is a write to the sysctl file
2570 * @buffer: the user buffer
2571 * @lenp: the size of the user buffer
2572 * @ppos: pointer to the file position
2574 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2575 * values from/to the user buffer, treated as an ASCII string.
2576 * The values read are assumed to be in 1/USER_HZ seconds, and
2577 * are converted into jiffies.
2579 * Returns 0 on success.
2581 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2582 void __user *buffer, size_t *lenp, loff_t *ppos)
2584 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2585 do_proc_dointvec_userhz_jiffies_conv,NULL);
2589 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2590 * @table: the sysctl table
2591 * @write: %TRUE if this is a write to the sysctl file
2592 * @buffer: the user buffer
2593 * @lenp: the size of the user buffer
2594 * @ppos: file position
2595 * @ppos: the current position in the file
2597 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2598 * values from/to the user buffer, treated as an ASCII string.
2599 * The values read are assumed to be in 1/1000 seconds, and
2600 * are converted into jiffies.
2602 * Returns 0 on success.
2604 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2605 void __user *buffer, size_t *lenp, loff_t *ppos)
2607 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2608 do_proc_dointvec_ms_jiffies_conv, NULL);
2611 static int proc_do_cad_pid(struct ctl_table *table, int write,
2612 void __user *buffer, size_t *lenp, loff_t *ppos)
2614 struct pid *new_pid;
2618 tmp = pid_vnr(cad_pid);
2620 r = __do_proc_dointvec(&tmp, table, write, buffer,
2621 lenp, ppos, NULL, NULL);
2625 new_pid = find_get_pid(tmp);
2629 put_pid(xchg(&cad_pid, new_pid));
2633 #else /* CONFIG_PROC_FS */
2635 int proc_dostring(struct ctl_table *table, int write,
2636 void __user *buffer, size_t *lenp, loff_t *ppos)
2641 int proc_dointvec(struct ctl_table *table, int write,
2642 void __user *buffer, size_t *lenp, loff_t *ppos)
2647 int proc_dointvec_minmax(struct ctl_table *table, int write,
2648 void __user *buffer, size_t *lenp, loff_t *ppos)
2653 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2654 void __user *buffer, size_t *lenp, loff_t *ppos)
2659 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2660 void __user *buffer, size_t *lenp, loff_t *ppos)
2665 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2666 void __user *buffer, size_t *lenp, loff_t *ppos)
2671 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2672 void __user *buffer, size_t *lenp, loff_t *ppos)
2677 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2678 void __user *buffer,
2679 size_t *lenp, loff_t *ppos)
2685 #endif /* CONFIG_PROC_FS */
2688 * No sense putting this after each symbol definition, twice,
2689 * exception granted :-)
2691 EXPORT_SYMBOL(proc_dointvec);
2692 EXPORT_SYMBOL(proc_dointvec_jiffies);
2693 EXPORT_SYMBOL(proc_dointvec_minmax);
2694 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2695 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2696 EXPORT_SYMBOL(proc_dostring);
2697 EXPORT_SYMBOL(proc_doulongvec_minmax);
2698 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2699 EXPORT_SYMBOL(register_sysctl_table);
2700 EXPORT_SYMBOL(register_sysctl_paths);
2701 EXPORT_SYMBOL(unregister_sysctl_table);