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>
22 #include <linux/aio.h>
24 #include <linux/swap.h>
25 #include <linux/slab.h>
26 #include <linux/sysctl.h>
27 #include <linux/bitmap.h>
28 #include <linux/signal.h>
29 #include <linux/printk.h>
30 #include <linux/proc_fs.h>
31 #include <linux/security.h>
32 #include <linux/ctype.h>
33 #include <linux/kmemcheck.h>
34 #include <linux/kmemleak.h>
36 #include <linux/init.h>
37 #include <linux/kernel.h>
38 #include <linux/kobject.h>
39 #include <linux/net.h>
40 #include <linux/sysrq.h>
41 #include <linux/highuid.h>
42 #include <linux/writeback.h>
43 #include <linux/ratelimit.h>
44 #include <linux/compaction.h>
45 #include <linux/hugetlb.h>
46 #include <linux/initrd.h>
47 #include <linux/key.h>
48 #include <linux/times.h>
49 #include <linux/limits.h>
50 #include <linux/dcache.h>
51 #include <linux/dnotify.h>
52 #include <linux/syscalls.h>
53 #include <linux/vmstat.h>
54 #include <linux/nfs_fs.h>
55 #include <linux/acpi.h>
56 #include <linux/reboot.h>
57 #include <linux/ftrace.h>
58 #include <linux/perf_event.h>
59 #include <linux/kprobes.h>
60 #include <linux/pipe_fs_i.h>
61 #include <linux/oom.h>
62 #include <linux/kmod.h>
63 #include <linux/capability.h>
64 #include <linux/binfmts.h>
65 #include <linux/sched/sysctl.h>
66 #include <linux/kexec.h>
67 #include <linux/bpf.h>
69 #include <asm/uaccess.h>
70 #include <asm/processor.h>
74 #include <asm/stacktrace.h>
78 #include <asm/setup.h>
80 #ifdef CONFIG_BSD_PROCESS_ACCT
81 #include <linux/acct.h>
83 #ifdef CONFIG_RT_MUTEXES
84 #include <linux/rtmutex.h>
86 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
87 #include <linux/lockdep.h>
89 #ifdef CONFIG_CHR_DEV_SG
93 #ifdef CONFIG_LOCKUP_DETECTOR
94 #include <linux/nmi.h>
97 #if defined(CONFIG_SYSCTL)
99 /* External variables not in a header file. */
100 extern int suid_dumpable;
101 #ifdef CONFIG_COREDUMP
102 extern int core_uses_pid;
103 extern char core_pattern[];
104 extern unsigned int core_pipe_limit;
107 extern int pid_max_min, pid_max_max;
108 extern int percpu_pagelist_fraction;
109 extern int latencytop_enabled;
110 extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
112 extern int sysctl_nr_trim_pages;
115 /* Constants used for minimum and maximum */
116 #ifdef CONFIG_LOCKUP_DETECTOR
117 static int sixty = 60;
120 static int __maybe_unused neg_one = -1;
123 static int __maybe_unused one = 1;
124 static int __maybe_unused two = 2;
125 static int __maybe_unused four = 4;
126 static unsigned long one_ul = 1;
127 static int one_hundred = 100;
128 static int one_thousand = 1000;
130 static int ten_thousand = 10000;
132 #ifdef CONFIG_PERF_EVENTS
133 static int six_hundred_forty_kb = 640 * 1024;
136 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
137 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
139 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
140 static int maxolduid = 65535;
141 static int minolduid;
143 static int ngroups_max = NGROUPS_MAX;
144 static const int cap_last_cap = CAP_LAST_CAP;
146 /*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */
147 #ifdef CONFIG_DETECT_HUNG_TASK
148 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
151 #ifdef CONFIG_INOTIFY_USER
152 #include <linux/inotify.h>
158 extern int pwrsw_enabled;
161 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
162 extern int unaligned_enabled;
166 extern int unaligned_dump_stack;
169 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
170 extern int no_unaligned_warning;
173 #ifdef CONFIG_PROC_SYSCTL
175 #define SYSCTL_WRITES_LEGACY -1
176 #define SYSCTL_WRITES_WARN 0
177 #define SYSCTL_WRITES_STRICT 1
179 static int sysctl_writes_strict = SYSCTL_WRITES_STRICT;
181 static int proc_do_cad_pid(struct ctl_table *table, int write,
182 void __user *buffer, size_t *lenp, loff_t *ppos);
183 static int proc_taint(struct ctl_table *table, int write,
184 void __user *buffer, size_t *lenp, loff_t *ppos);
188 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
189 void __user *buffer, size_t *lenp, loff_t *ppos);
192 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
193 void __user *buffer, size_t *lenp, loff_t *ppos);
194 #ifdef CONFIG_COREDUMP
195 static int proc_dostring_coredump(struct ctl_table *table, int write,
196 void __user *buffer, size_t *lenp, loff_t *ppos);
199 #ifdef CONFIG_MAGIC_SYSRQ
200 /* Note: sysrq code uses it's own private copy */
201 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
203 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
204 void __user *buffer, size_t *lenp,
209 error = proc_dointvec(table, write, buffer, lenp, ppos);
214 sysrq_toggle_support(__sysrq_enabled);
221 static struct ctl_table kern_table[];
222 static struct ctl_table vm_table[];
223 static struct ctl_table fs_table[];
224 static struct ctl_table debug_table[];
225 static struct ctl_table dev_table[];
226 extern struct ctl_table random_table[];
228 extern struct ctl_table epoll_table[];
231 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
232 int sysctl_legacy_va_layout;
235 /* The default sysctl tables: */
237 static struct ctl_table sysctl_base_table[] = {
239 .procname = "kernel",
256 .child = debug_table,
266 #ifdef CONFIG_SCHED_DEBUG
267 static int min_sched_granularity_ns = 100000; /* 100 usecs */
268 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
269 static int min_wakeup_granularity_ns; /* 0 usecs */
270 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
272 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
273 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
274 #endif /* CONFIG_SMP */
275 #endif /* CONFIG_SCHED_DEBUG */
277 #ifdef CONFIG_COMPACTION
278 static int min_extfrag_threshold;
279 static int max_extfrag_threshold = 1000;
282 static struct ctl_table kern_table[] = {
284 .procname = "sched_child_runs_first",
285 .data = &sysctl_sched_child_runs_first,
286 .maxlen = sizeof(unsigned int),
288 .proc_handler = proc_dointvec,
290 #ifdef CONFIG_SCHED_DEBUG
292 .procname = "sched_min_granularity_ns",
293 .data = &sysctl_sched_min_granularity,
294 .maxlen = sizeof(unsigned int),
296 .proc_handler = sched_proc_update_handler,
297 .extra1 = &min_sched_granularity_ns,
298 .extra2 = &max_sched_granularity_ns,
301 .procname = "sched_latency_ns",
302 .data = &sysctl_sched_latency,
303 .maxlen = sizeof(unsigned int),
305 .proc_handler = sched_proc_update_handler,
306 .extra1 = &min_sched_granularity_ns,
307 .extra2 = &max_sched_granularity_ns,
310 .procname = "sched_wakeup_granularity_ns",
311 .data = &sysctl_sched_wakeup_granularity,
312 .maxlen = sizeof(unsigned int),
314 .proc_handler = sched_proc_update_handler,
315 .extra1 = &min_wakeup_granularity_ns,
316 .extra2 = &max_wakeup_granularity_ns,
320 .procname = "sched_tunable_scaling",
321 .data = &sysctl_sched_tunable_scaling,
322 .maxlen = sizeof(enum sched_tunable_scaling),
324 .proc_handler = sched_proc_update_handler,
325 .extra1 = &min_sched_tunable_scaling,
326 .extra2 = &max_sched_tunable_scaling,
329 .procname = "sched_migration_cost_ns",
330 .data = &sysctl_sched_migration_cost,
331 .maxlen = sizeof(unsigned int),
333 .proc_handler = proc_dointvec,
336 .procname = "sched_nr_migrate",
337 .data = &sysctl_sched_nr_migrate,
338 .maxlen = sizeof(unsigned int),
340 .proc_handler = proc_dointvec,
343 .procname = "sched_time_avg_ms",
344 .data = &sysctl_sched_time_avg,
345 .maxlen = sizeof(unsigned int),
347 .proc_handler = proc_dointvec,
350 .procname = "sched_shares_window_ns",
351 .data = &sysctl_sched_shares_window,
352 .maxlen = sizeof(unsigned int),
354 .proc_handler = proc_dointvec,
356 #ifdef CONFIG_SCHEDSTATS
358 .procname = "sched_schedstats",
360 .maxlen = sizeof(unsigned int),
362 .proc_handler = sysctl_schedstats,
366 #endif /* CONFIG_SCHEDSTATS */
367 #endif /* CONFIG_SMP */
368 #ifdef CONFIG_NUMA_BALANCING
370 .procname = "numa_balancing_scan_delay_ms",
371 .data = &sysctl_numa_balancing_scan_delay,
372 .maxlen = sizeof(unsigned int),
374 .proc_handler = proc_dointvec,
377 .procname = "numa_balancing_scan_period_min_ms",
378 .data = &sysctl_numa_balancing_scan_period_min,
379 .maxlen = sizeof(unsigned int),
381 .proc_handler = proc_dointvec,
384 .procname = "numa_balancing_scan_period_max_ms",
385 .data = &sysctl_numa_balancing_scan_period_max,
386 .maxlen = sizeof(unsigned int),
388 .proc_handler = proc_dointvec,
391 .procname = "numa_balancing_scan_size_mb",
392 .data = &sysctl_numa_balancing_scan_size,
393 .maxlen = sizeof(unsigned int),
395 .proc_handler = proc_dointvec_minmax,
399 .procname = "numa_balancing",
400 .data = NULL, /* filled in by handler */
401 .maxlen = sizeof(unsigned int),
403 .proc_handler = sysctl_numa_balancing,
407 #endif /* CONFIG_NUMA_BALANCING */
408 #endif /* CONFIG_SCHED_DEBUG */
410 .procname = "sched_rt_period_us",
411 .data = &sysctl_sched_rt_period,
412 .maxlen = sizeof(unsigned int),
414 .proc_handler = sched_rt_handler,
417 .procname = "sched_rt_runtime_us",
418 .data = &sysctl_sched_rt_runtime,
419 .maxlen = sizeof(int),
421 .proc_handler = sched_rt_handler,
424 .procname = "sched_rr_timeslice_ms",
425 .data = &sched_rr_timeslice,
426 .maxlen = sizeof(int),
428 .proc_handler = sched_rr_handler,
430 #ifdef CONFIG_SCHED_AUTOGROUP
432 .procname = "sched_autogroup_enabled",
433 .data = &sysctl_sched_autogroup_enabled,
434 .maxlen = sizeof(unsigned int),
436 .proc_handler = proc_dointvec_minmax,
441 #ifdef CONFIG_CFS_BANDWIDTH
443 .procname = "sched_cfs_bandwidth_slice_us",
444 .data = &sysctl_sched_cfs_bandwidth_slice,
445 .maxlen = sizeof(unsigned int),
447 .proc_handler = proc_dointvec_minmax,
451 #ifdef CONFIG_PROVE_LOCKING
453 .procname = "prove_locking",
454 .data = &prove_locking,
455 .maxlen = sizeof(int),
457 .proc_handler = proc_dointvec,
460 #ifdef CONFIG_LOCK_STAT
462 .procname = "lock_stat",
464 .maxlen = sizeof(int),
466 .proc_handler = proc_dointvec,
471 .data = &panic_timeout,
472 .maxlen = sizeof(int),
474 .proc_handler = proc_dointvec,
476 #ifdef CONFIG_COREDUMP
478 .procname = "core_uses_pid",
479 .data = &core_uses_pid,
480 .maxlen = sizeof(int),
482 .proc_handler = proc_dointvec,
485 .procname = "core_pattern",
486 .data = core_pattern,
487 .maxlen = CORENAME_MAX_SIZE,
489 .proc_handler = proc_dostring_coredump,
492 .procname = "core_pipe_limit",
493 .data = &core_pipe_limit,
494 .maxlen = sizeof(unsigned int),
496 .proc_handler = proc_dointvec,
499 #ifdef CONFIG_PROC_SYSCTL
501 .procname = "tainted",
502 .maxlen = sizeof(long),
504 .proc_handler = proc_taint,
507 .procname = "sysctl_writes_strict",
508 .data = &sysctl_writes_strict,
509 .maxlen = sizeof(int),
511 .proc_handler = proc_dointvec_minmax,
516 #ifdef CONFIG_LATENCYTOP
518 .procname = "latencytop",
519 .data = &latencytop_enabled,
520 .maxlen = sizeof(int),
522 .proc_handler = sysctl_latencytop,
525 #ifdef CONFIG_BLK_DEV_INITRD
527 .procname = "real-root-dev",
528 .data = &real_root_dev,
529 .maxlen = sizeof(int),
531 .proc_handler = proc_dointvec,
535 .procname = "print-fatal-signals",
536 .data = &print_fatal_signals,
537 .maxlen = sizeof(int),
539 .proc_handler = proc_dointvec,
543 .procname = "reboot-cmd",
544 .data = reboot_command,
547 .proc_handler = proc_dostring,
550 .procname = "stop-a",
551 .data = &stop_a_enabled,
552 .maxlen = sizeof (int),
554 .proc_handler = proc_dointvec,
557 .procname = "scons-poweroff",
558 .data = &scons_pwroff,
559 .maxlen = sizeof (int),
561 .proc_handler = proc_dointvec,
564 #ifdef CONFIG_SPARC64
566 .procname = "tsb-ratio",
567 .data = &sysctl_tsb_ratio,
568 .maxlen = sizeof (int),
570 .proc_handler = proc_dointvec,
575 .procname = "soft-power",
576 .data = &pwrsw_enabled,
577 .maxlen = sizeof (int),
579 .proc_handler = proc_dointvec,
582 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
584 .procname = "unaligned-trap",
585 .data = &unaligned_enabled,
586 .maxlen = sizeof (int),
588 .proc_handler = proc_dointvec,
592 .procname = "ctrl-alt-del",
594 .maxlen = sizeof(int),
596 .proc_handler = proc_dointvec,
598 #ifdef CONFIG_FUNCTION_TRACER
600 .procname = "ftrace_enabled",
601 .data = &ftrace_enabled,
602 .maxlen = sizeof(int),
604 .proc_handler = ftrace_enable_sysctl,
607 #ifdef CONFIG_STACK_TRACER
609 .procname = "stack_tracer_enabled",
610 .data = &stack_tracer_enabled,
611 .maxlen = sizeof(int),
613 .proc_handler = stack_trace_sysctl,
616 #ifdef CONFIG_TRACING
618 .procname = "ftrace_dump_on_oops",
619 .data = &ftrace_dump_on_oops,
620 .maxlen = sizeof(int),
622 .proc_handler = proc_dointvec,
625 .procname = "traceoff_on_warning",
626 .data = &__disable_trace_on_warning,
627 .maxlen = sizeof(__disable_trace_on_warning),
629 .proc_handler = proc_dointvec,
632 .procname = "tracepoint_printk",
633 .data = &tracepoint_printk,
634 .maxlen = sizeof(tracepoint_printk),
636 .proc_handler = proc_dointvec,
639 #ifdef CONFIG_KEXEC_CORE
641 .procname = "kexec_load_disabled",
642 .data = &kexec_load_disabled,
643 .maxlen = sizeof(int),
645 /* only handle a transition from default "0" to "1" */
646 .proc_handler = proc_dointvec_minmax,
651 #ifdef CONFIG_MODULES
653 .procname = "modprobe",
654 .data = &modprobe_path,
655 .maxlen = KMOD_PATH_LEN,
657 .proc_handler = proc_dostring,
660 .procname = "modules_disabled",
661 .data = &modules_disabled,
662 .maxlen = sizeof(int),
664 /* only handle a transition from default "0" to "1" */
665 .proc_handler = proc_dointvec_minmax,
670 #ifdef CONFIG_UEVENT_HELPER
672 .procname = "hotplug",
673 .data = &uevent_helper,
674 .maxlen = UEVENT_HELPER_PATH_LEN,
676 .proc_handler = proc_dostring,
679 #ifdef CONFIG_CHR_DEV_SG
681 .procname = "sg-big-buff",
682 .data = &sg_big_buff,
683 .maxlen = sizeof (int),
685 .proc_handler = proc_dointvec,
688 #ifdef CONFIG_BSD_PROCESS_ACCT
692 .maxlen = 3*sizeof(int),
694 .proc_handler = proc_dointvec,
697 #ifdef CONFIG_MAGIC_SYSRQ
700 .data = &__sysrq_enabled,
701 .maxlen = sizeof (int),
703 .proc_handler = sysrq_sysctl_handler,
706 #ifdef CONFIG_PROC_SYSCTL
708 .procname = "cad_pid",
710 .maxlen = sizeof (int),
712 .proc_handler = proc_do_cad_pid,
716 .procname = "threads-max",
718 .maxlen = sizeof(int),
720 .proc_handler = sysctl_max_threads,
723 .procname = "random",
725 .child = random_table,
728 .procname = "usermodehelper",
730 .child = usermodehelper_table,
733 .procname = "overflowuid",
734 .data = &overflowuid,
735 .maxlen = sizeof(int),
737 .proc_handler = proc_dointvec_minmax,
738 .extra1 = &minolduid,
739 .extra2 = &maxolduid,
742 .procname = "overflowgid",
743 .data = &overflowgid,
744 .maxlen = sizeof(int),
746 .proc_handler = proc_dointvec_minmax,
747 .extra1 = &minolduid,
748 .extra2 = &maxolduid,
751 #ifdef CONFIG_MATHEMU
753 .procname = "ieee_emulation_warnings",
754 .data = &sysctl_ieee_emulation_warnings,
755 .maxlen = sizeof(int),
757 .proc_handler = proc_dointvec,
761 .procname = "userprocess_debug",
762 .data = &show_unhandled_signals,
763 .maxlen = sizeof(int),
765 .proc_handler = proc_dointvec,
769 .procname = "pid_max",
771 .maxlen = sizeof (int),
773 .proc_handler = proc_dointvec_minmax,
774 .extra1 = &pid_max_min,
775 .extra2 = &pid_max_max,
778 .procname = "panic_on_oops",
779 .data = &panic_on_oops,
780 .maxlen = sizeof(int),
782 .proc_handler = proc_dointvec,
784 #if defined CONFIG_PRINTK
786 .procname = "printk",
787 .data = &console_loglevel,
788 .maxlen = 4*sizeof(int),
790 .proc_handler = proc_dointvec,
793 .procname = "printk_ratelimit",
794 .data = &printk_ratelimit_state.interval,
795 .maxlen = sizeof(int),
797 .proc_handler = proc_dointvec_jiffies,
800 .procname = "printk_ratelimit_burst",
801 .data = &printk_ratelimit_state.burst,
802 .maxlen = sizeof(int),
804 .proc_handler = proc_dointvec,
807 .procname = "printk_delay",
808 .data = &printk_delay_msec,
809 .maxlen = sizeof(int),
811 .proc_handler = proc_dointvec_minmax,
813 .extra2 = &ten_thousand,
816 .procname = "printk_devkmsg",
817 .data = devkmsg_log_str,
818 .maxlen = DEVKMSG_STR_MAX_SIZE,
820 .proc_handler = devkmsg_sysctl_set_loglvl,
823 .procname = "dmesg_restrict",
824 .data = &dmesg_restrict,
825 .maxlen = sizeof(int),
827 .proc_handler = proc_dointvec_minmax_sysadmin,
832 .procname = "kptr_restrict",
833 .data = &kptr_restrict,
834 .maxlen = sizeof(int),
836 .proc_handler = proc_dointvec_minmax_sysadmin,
842 .procname = "ngroups_max",
843 .data = &ngroups_max,
844 .maxlen = sizeof (int),
846 .proc_handler = proc_dointvec,
849 .procname = "cap_last_cap",
850 .data = (void *)&cap_last_cap,
851 .maxlen = sizeof(int),
853 .proc_handler = proc_dointvec,
855 #if defined(CONFIG_LOCKUP_DETECTOR)
857 .procname = "watchdog",
858 .data = &watchdog_user_enabled,
859 .maxlen = sizeof (int),
861 .proc_handler = proc_watchdog,
866 .procname = "watchdog_thresh",
867 .data = &watchdog_thresh,
868 .maxlen = sizeof(int),
870 .proc_handler = proc_watchdog_thresh,
875 .procname = "nmi_watchdog",
876 .data = &nmi_watchdog_enabled,
877 .maxlen = sizeof (int),
879 .proc_handler = proc_nmi_watchdog,
881 #if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR)
888 .procname = "soft_watchdog",
889 .data = &soft_watchdog_enabled,
890 .maxlen = sizeof (int),
892 .proc_handler = proc_soft_watchdog,
897 .procname = "watchdog_cpumask",
898 .data = &watchdog_cpumask_bits,
901 .proc_handler = proc_watchdog_cpumask,
904 .procname = "softlockup_panic",
905 .data = &softlockup_panic,
906 .maxlen = sizeof(int),
908 .proc_handler = proc_dointvec_minmax,
912 #ifdef CONFIG_HARDLOCKUP_DETECTOR
914 .procname = "hardlockup_panic",
915 .data = &hardlockup_panic,
916 .maxlen = sizeof(int),
918 .proc_handler = proc_dointvec_minmax,
925 .procname = "softlockup_all_cpu_backtrace",
926 .data = &sysctl_softlockup_all_cpu_backtrace,
927 .maxlen = sizeof(int),
929 .proc_handler = proc_dointvec_minmax,
934 .procname = "hardlockup_all_cpu_backtrace",
935 .data = &sysctl_hardlockup_all_cpu_backtrace,
936 .maxlen = sizeof(int),
938 .proc_handler = proc_dointvec_minmax,
942 #endif /* CONFIG_SMP */
944 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
946 .procname = "unknown_nmi_panic",
947 .data = &unknown_nmi_panic,
948 .maxlen = sizeof (int),
950 .proc_handler = proc_dointvec,
953 #if defined(CONFIG_X86)
955 .procname = "panic_on_unrecovered_nmi",
956 .data = &panic_on_unrecovered_nmi,
957 .maxlen = sizeof(int),
959 .proc_handler = proc_dointvec,
962 .procname = "panic_on_io_nmi",
963 .data = &panic_on_io_nmi,
964 .maxlen = sizeof(int),
966 .proc_handler = proc_dointvec,
968 #ifdef CONFIG_DEBUG_STACKOVERFLOW
970 .procname = "panic_on_stackoverflow",
971 .data = &sysctl_panic_on_stackoverflow,
972 .maxlen = sizeof(int),
974 .proc_handler = proc_dointvec,
978 .procname = "bootloader_type",
979 .data = &bootloader_type,
980 .maxlen = sizeof (int),
982 .proc_handler = proc_dointvec,
985 .procname = "bootloader_version",
986 .data = &bootloader_version,
987 .maxlen = sizeof (int),
989 .proc_handler = proc_dointvec,
992 .procname = "kstack_depth_to_print",
993 .data = &kstack_depth_to_print,
994 .maxlen = sizeof(int),
996 .proc_handler = proc_dointvec,
999 .procname = "io_delay_type",
1000 .data = &io_delay_type,
1001 .maxlen = sizeof(int),
1003 .proc_handler = proc_dointvec,
1006 #if defined(CONFIG_MMU)
1008 .procname = "randomize_va_space",
1009 .data = &randomize_va_space,
1010 .maxlen = sizeof(int),
1012 .proc_handler = proc_dointvec,
1015 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1017 .procname = "spin_retry",
1018 .data = &spin_retry,
1019 .maxlen = sizeof (int),
1021 .proc_handler = proc_dointvec,
1024 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1026 .procname = "acpi_video_flags",
1027 .data = &acpi_realmode_flags,
1028 .maxlen = sizeof (unsigned long),
1030 .proc_handler = proc_doulongvec_minmax,
1033 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1035 .procname = "ignore-unaligned-usertrap",
1036 .data = &no_unaligned_warning,
1037 .maxlen = sizeof (int),
1039 .proc_handler = proc_dointvec,
1044 .procname = "unaligned-dump-stack",
1045 .data = &unaligned_dump_stack,
1046 .maxlen = sizeof (int),
1048 .proc_handler = proc_dointvec,
1051 #ifdef CONFIG_DETECT_HUNG_TASK
1053 .procname = "hung_task_panic",
1054 .data = &sysctl_hung_task_panic,
1055 .maxlen = sizeof(int),
1057 .proc_handler = proc_dointvec_minmax,
1062 .procname = "hung_task_check_count",
1063 .data = &sysctl_hung_task_check_count,
1064 .maxlen = sizeof(int),
1066 .proc_handler = proc_dointvec_minmax,
1070 .procname = "hung_task_timeout_secs",
1071 .data = &sysctl_hung_task_timeout_secs,
1072 .maxlen = sizeof(unsigned long),
1074 .proc_handler = proc_dohung_task_timeout_secs,
1075 .extra2 = &hung_task_timeout_max,
1078 .procname = "hung_task_warnings",
1079 .data = &sysctl_hung_task_warnings,
1080 .maxlen = sizeof(int),
1082 .proc_handler = proc_dointvec_minmax,
1086 #ifdef CONFIG_RT_MUTEXES
1088 .procname = "max_lock_depth",
1089 .data = &max_lock_depth,
1090 .maxlen = sizeof(int),
1092 .proc_handler = proc_dointvec,
1096 .procname = "poweroff_cmd",
1097 .data = &poweroff_cmd,
1098 .maxlen = POWEROFF_CMD_PATH_LEN,
1100 .proc_handler = proc_dostring,
1106 .child = key_sysctls,
1109 #ifdef CONFIG_PERF_EVENTS
1111 * User-space scripts rely on the existence of this file
1112 * as a feature check for perf_events being enabled.
1114 * So it's an ABI, do not remove!
1117 .procname = "perf_event_paranoid",
1118 .data = &sysctl_perf_event_paranoid,
1119 .maxlen = sizeof(sysctl_perf_event_paranoid),
1121 .proc_handler = proc_dointvec,
1124 .procname = "perf_event_mlock_kb",
1125 .data = &sysctl_perf_event_mlock,
1126 .maxlen = sizeof(sysctl_perf_event_mlock),
1128 .proc_handler = proc_dointvec,
1131 .procname = "perf_event_max_sample_rate",
1132 .data = &sysctl_perf_event_sample_rate,
1133 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1135 .proc_handler = perf_proc_update_handler,
1139 .procname = "perf_cpu_time_max_percent",
1140 .data = &sysctl_perf_cpu_time_max_percent,
1141 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1143 .proc_handler = perf_cpu_time_max_percent_handler,
1145 .extra2 = &one_hundred,
1148 .procname = "perf_event_max_stack",
1149 .data = &sysctl_perf_event_max_stack,
1150 .maxlen = sizeof(sysctl_perf_event_max_stack),
1152 .proc_handler = perf_event_max_stack_handler,
1154 .extra2 = &six_hundred_forty_kb,
1157 .procname = "perf_event_max_contexts_per_stack",
1158 .data = &sysctl_perf_event_max_contexts_per_stack,
1159 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
1161 .proc_handler = perf_event_max_stack_handler,
1163 .extra2 = &one_thousand,
1166 #ifdef CONFIG_KMEMCHECK
1168 .procname = "kmemcheck",
1169 .data = &kmemcheck_enabled,
1170 .maxlen = sizeof(int),
1172 .proc_handler = proc_dointvec,
1176 .procname = "panic_on_warn",
1177 .data = &panic_on_warn,
1178 .maxlen = sizeof(int),
1180 .proc_handler = proc_dointvec_minmax,
1184 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1186 .procname = "timer_migration",
1187 .data = &sysctl_timer_migration,
1188 .maxlen = sizeof(unsigned int),
1190 .proc_handler = timer_migration_handler,
1193 #ifdef CONFIG_BPF_SYSCALL
1195 .procname = "unprivileged_bpf_disabled",
1196 .data = &sysctl_unprivileged_bpf_disabled,
1197 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
1199 /* only handle a transition from default "0" to "1" */
1200 .proc_handler = proc_dointvec_minmax,
1205 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1207 .procname = "panic_on_rcu_stall",
1208 .data = &sysctl_panic_on_rcu_stall,
1209 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
1211 .proc_handler = proc_dointvec_minmax,
1219 static struct ctl_table vm_table[] = {
1221 .procname = "overcommit_memory",
1222 .data = &sysctl_overcommit_memory,
1223 .maxlen = sizeof(sysctl_overcommit_memory),
1225 .proc_handler = proc_dointvec_minmax,
1230 .procname = "panic_on_oom",
1231 .data = &sysctl_panic_on_oom,
1232 .maxlen = sizeof(sysctl_panic_on_oom),
1234 .proc_handler = proc_dointvec_minmax,
1239 .procname = "oom_kill_allocating_task",
1240 .data = &sysctl_oom_kill_allocating_task,
1241 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1243 .proc_handler = proc_dointvec,
1246 .procname = "oom_dump_tasks",
1247 .data = &sysctl_oom_dump_tasks,
1248 .maxlen = sizeof(sysctl_oom_dump_tasks),
1250 .proc_handler = proc_dointvec,
1253 .procname = "overcommit_ratio",
1254 .data = &sysctl_overcommit_ratio,
1255 .maxlen = sizeof(sysctl_overcommit_ratio),
1257 .proc_handler = overcommit_ratio_handler,
1260 .procname = "overcommit_kbytes",
1261 .data = &sysctl_overcommit_kbytes,
1262 .maxlen = sizeof(sysctl_overcommit_kbytes),
1264 .proc_handler = overcommit_kbytes_handler,
1267 .procname = "page-cluster",
1268 .data = &page_cluster,
1269 .maxlen = sizeof(int),
1271 .proc_handler = proc_dointvec_minmax,
1275 .procname = "dirty_background_ratio",
1276 .data = &dirty_background_ratio,
1277 .maxlen = sizeof(dirty_background_ratio),
1279 .proc_handler = dirty_background_ratio_handler,
1281 .extra2 = &one_hundred,
1284 .procname = "dirty_background_bytes",
1285 .data = &dirty_background_bytes,
1286 .maxlen = sizeof(dirty_background_bytes),
1288 .proc_handler = dirty_background_bytes_handler,
1292 .procname = "dirty_ratio",
1293 .data = &vm_dirty_ratio,
1294 .maxlen = sizeof(vm_dirty_ratio),
1296 .proc_handler = dirty_ratio_handler,
1298 .extra2 = &one_hundred,
1301 .procname = "dirty_bytes",
1302 .data = &vm_dirty_bytes,
1303 .maxlen = sizeof(vm_dirty_bytes),
1305 .proc_handler = dirty_bytes_handler,
1306 .extra1 = &dirty_bytes_min,
1309 .procname = "dirty_writeback_centisecs",
1310 .data = &dirty_writeback_interval,
1311 .maxlen = sizeof(dirty_writeback_interval),
1313 .proc_handler = dirty_writeback_centisecs_handler,
1316 .procname = "dirty_expire_centisecs",
1317 .data = &dirty_expire_interval,
1318 .maxlen = sizeof(dirty_expire_interval),
1320 .proc_handler = proc_dointvec_minmax,
1324 .procname = "dirtytime_expire_seconds",
1325 .data = &dirtytime_expire_interval,
1326 .maxlen = sizeof(dirty_expire_interval),
1328 .proc_handler = dirtytime_interval_handler,
1332 .procname = "nr_pdflush_threads",
1333 .mode = 0444 /* read-only */,
1334 .proc_handler = pdflush_proc_obsolete,
1337 .procname = "swappiness",
1338 .data = &vm_swappiness,
1339 .maxlen = sizeof(vm_swappiness),
1341 .proc_handler = proc_dointvec_minmax,
1343 .extra2 = &one_hundred,
1345 #ifdef CONFIG_HUGETLB_PAGE
1347 .procname = "nr_hugepages",
1349 .maxlen = sizeof(unsigned long),
1351 .proc_handler = hugetlb_sysctl_handler,
1355 .procname = "nr_hugepages_mempolicy",
1357 .maxlen = sizeof(unsigned long),
1359 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1363 .procname = "hugetlb_shm_group",
1364 .data = &sysctl_hugetlb_shm_group,
1365 .maxlen = sizeof(gid_t),
1367 .proc_handler = proc_dointvec,
1370 .procname = "hugepages_treat_as_movable",
1371 .data = &hugepages_treat_as_movable,
1372 .maxlen = sizeof(int),
1374 .proc_handler = proc_dointvec,
1377 .procname = "nr_overcommit_hugepages",
1379 .maxlen = sizeof(unsigned long),
1381 .proc_handler = hugetlb_overcommit_handler,
1385 .procname = "lowmem_reserve_ratio",
1386 .data = &sysctl_lowmem_reserve_ratio,
1387 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1389 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1392 .procname = "drop_caches",
1393 .data = &sysctl_drop_caches,
1394 .maxlen = sizeof(int),
1396 .proc_handler = drop_caches_sysctl_handler,
1400 #ifdef CONFIG_COMPACTION
1402 .procname = "compact_memory",
1403 .data = &sysctl_compact_memory,
1404 .maxlen = sizeof(int),
1406 .proc_handler = sysctl_compaction_handler,
1409 .procname = "extfrag_threshold",
1410 .data = &sysctl_extfrag_threshold,
1411 .maxlen = sizeof(int),
1413 .proc_handler = sysctl_extfrag_handler,
1414 .extra1 = &min_extfrag_threshold,
1415 .extra2 = &max_extfrag_threshold,
1418 .procname = "compact_unevictable_allowed",
1419 .data = &sysctl_compact_unevictable_allowed,
1420 .maxlen = sizeof(int),
1422 .proc_handler = proc_dointvec,
1427 #endif /* CONFIG_COMPACTION */
1429 .procname = "min_free_kbytes",
1430 .data = &min_free_kbytes,
1431 .maxlen = sizeof(min_free_kbytes),
1433 .proc_handler = min_free_kbytes_sysctl_handler,
1437 .procname = "watermark_scale_factor",
1438 .data = &watermark_scale_factor,
1439 .maxlen = sizeof(watermark_scale_factor),
1441 .proc_handler = watermark_scale_factor_sysctl_handler,
1443 .extra2 = &one_thousand,
1446 .procname = "percpu_pagelist_fraction",
1447 .data = &percpu_pagelist_fraction,
1448 .maxlen = sizeof(percpu_pagelist_fraction),
1450 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1455 .procname = "max_map_count",
1456 .data = &sysctl_max_map_count,
1457 .maxlen = sizeof(sysctl_max_map_count),
1459 .proc_handler = proc_dointvec_minmax,
1464 .procname = "nr_trim_pages",
1465 .data = &sysctl_nr_trim_pages,
1466 .maxlen = sizeof(sysctl_nr_trim_pages),
1468 .proc_handler = proc_dointvec_minmax,
1473 .procname = "laptop_mode",
1474 .data = &laptop_mode,
1475 .maxlen = sizeof(laptop_mode),
1477 .proc_handler = proc_dointvec_jiffies,
1480 .procname = "block_dump",
1481 .data = &block_dump,
1482 .maxlen = sizeof(block_dump),
1484 .proc_handler = proc_dointvec,
1488 .procname = "vfs_cache_pressure",
1489 .data = &sysctl_vfs_cache_pressure,
1490 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1492 .proc_handler = proc_dointvec,
1495 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1497 .procname = "legacy_va_layout",
1498 .data = &sysctl_legacy_va_layout,
1499 .maxlen = sizeof(sysctl_legacy_va_layout),
1501 .proc_handler = proc_dointvec,
1507 .procname = "zone_reclaim_mode",
1508 .data = &node_reclaim_mode,
1509 .maxlen = sizeof(node_reclaim_mode),
1511 .proc_handler = proc_dointvec,
1515 .procname = "min_unmapped_ratio",
1516 .data = &sysctl_min_unmapped_ratio,
1517 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1519 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1521 .extra2 = &one_hundred,
1524 .procname = "min_slab_ratio",
1525 .data = &sysctl_min_slab_ratio,
1526 .maxlen = sizeof(sysctl_min_slab_ratio),
1528 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1530 .extra2 = &one_hundred,
1535 .procname = "stat_interval",
1536 .data = &sysctl_stat_interval,
1537 .maxlen = sizeof(sysctl_stat_interval),
1539 .proc_handler = proc_dointvec_jiffies,
1542 .procname = "stat_refresh",
1546 .proc_handler = vmstat_refresh,
1551 .procname = "mmap_min_addr",
1552 .data = &dac_mmap_min_addr,
1553 .maxlen = sizeof(unsigned long),
1555 .proc_handler = mmap_min_addr_handler,
1560 .procname = "numa_zonelist_order",
1561 .data = &numa_zonelist_order,
1562 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1564 .proc_handler = numa_zonelist_order_handler,
1567 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1568 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1570 .procname = "vdso_enabled",
1571 #ifdef CONFIG_X86_32
1572 .data = &vdso32_enabled,
1573 .maxlen = sizeof(vdso32_enabled),
1575 .data = &vdso_enabled,
1576 .maxlen = sizeof(vdso_enabled),
1579 .proc_handler = proc_dointvec,
1583 #ifdef CONFIG_HIGHMEM
1585 .procname = "highmem_is_dirtyable",
1586 .data = &vm_highmem_is_dirtyable,
1587 .maxlen = sizeof(vm_highmem_is_dirtyable),
1589 .proc_handler = proc_dointvec_minmax,
1594 #ifdef CONFIG_MEMORY_FAILURE
1596 .procname = "memory_failure_early_kill",
1597 .data = &sysctl_memory_failure_early_kill,
1598 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1600 .proc_handler = proc_dointvec_minmax,
1605 .procname = "memory_failure_recovery",
1606 .data = &sysctl_memory_failure_recovery,
1607 .maxlen = sizeof(sysctl_memory_failure_recovery),
1609 .proc_handler = proc_dointvec_minmax,
1615 .procname = "user_reserve_kbytes",
1616 .data = &sysctl_user_reserve_kbytes,
1617 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1619 .proc_handler = proc_doulongvec_minmax,
1622 .procname = "admin_reserve_kbytes",
1623 .data = &sysctl_admin_reserve_kbytes,
1624 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1626 .proc_handler = proc_doulongvec_minmax,
1628 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1630 .procname = "mmap_rnd_bits",
1631 .data = &mmap_rnd_bits,
1632 .maxlen = sizeof(mmap_rnd_bits),
1634 .proc_handler = proc_dointvec_minmax,
1635 .extra1 = (void *)&mmap_rnd_bits_min,
1636 .extra2 = (void *)&mmap_rnd_bits_max,
1639 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1641 .procname = "mmap_rnd_compat_bits",
1642 .data = &mmap_rnd_compat_bits,
1643 .maxlen = sizeof(mmap_rnd_compat_bits),
1645 .proc_handler = proc_dointvec_minmax,
1646 .extra1 = (void *)&mmap_rnd_compat_bits_min,
1647 .extra2 = (void *)&mmap_rnd_compat_bits_max,
1653 static struct ctl_table fs_table[] = {
1655 .procname = "inode-nr",
1656 .data = &inodes_stat,
1657 .maxlen = 2*sizeof(long),
1659 .proc_handler = proc_nr_inodes,
1662 .procname = "inode-state",
1663 .data = &inodes_stat,
1664 .maxlen = 7*sizeof(long),
1666 .proc_handler = proc_nr_inodes,
1669 .procname = "file-nr",
1670 .data = &files_stat,
1671 .maxlen = sizeof(files_stat),
1673 .proc_handler = proc_nr_files,
1676 .procname = "file-max",
1677 .data = &files_stat.max_files,
1678 .maxlen = sizeof(files_stat.max_files),
1680 .proc_handler = proc_doulongvec_minmax,
1683 .procname = "nr_open",
1684 .data = &sysctl_nr_open,
1685 .maxlen = sizeof(unsigned int),
1687 .proc_handler = proc_dointvec_minmax,
1688 .extra1 = &sysctl_nr_open_min,
1689 .extra2 = &sysctl_nr_open_max,
1692 .procname = "dentry-state",
1693 .data = &dentry_stat,
1694 .maxlen = 6*sizeof(long),
1696 .proc_handler = proc_nr_dentry,
1699 .procname = "overflowuid",
1700 .data = &fs_overflowuid,
1701 .maxlen = sizeof(int),
1703 .proc_handler = proc_dointvec_minmax,
1704 .extra1 = &minolduid,
1705 .extra2 = &maxolduid,
1708 .procname = "overflowgid",
1709 .data = &fs_overflowgid,
1710 .maxlen = sizeof(int),
1712 .proc_handler = proc_dointvec_minmax,
1713 .extra1 = &minolduid,
1714 .extra2 = &maxolduid,
1716 #ifdef CONFIG_FILE_LOCKING
1718 .procname = "leases-enable",
1719 .data = &leases_enable,
1720 .maxlen = sizeof(int),
1722 .proc_handler = proc_dointvec,
1725 #ifdef CONFIG_DNOTIFY
1727 .procname = "dir-notify-enable",
1728 .data = &dir_notify_enable,
1729 .maxlen = sizeof(int),
1731 .proc_handler = proc_dointvec,
1735 #ifdef CONFIG_FILE_LOCKING
1737 .procname = "lease-break-time",
1738 .data = &lease_break_time,
1739 .maxlen = sizeof(int),
1741 .proc_handler = proc_dointvec,
1746 .procname = "aio-nr",
1748 .maxlen = sizeof(aio_nr),
1750 .proc_handler = proc_doulongvec_minmax,
1753 .procname = "aio-max-nr",
1754 .data = &aio_max_nr,
1755 .maxlen = sizeof(aio_max_nr),
1757 .proc_handler = proc_doulongvec_minmax,
1759 #endif /* CONFIG_AIO */
1760 #ifdef CONFIG_INOTIFY_USER
1762 .procname = "inotify",
1764 .child = inotify_table,
1769 .procname = "epoll",
1771 .child = epoll_table,
1776 .procname = "protected_symlinks",
1777 .data = &sysctl_protected_symlinks,
1778 .maxlen = sizeof(int),
1780 .proc_handler = proc_dointvec_minmax,
1785 .procname = "protected_hardlinks",
1786 .data = &sysctl_protected_hardlinks,
1787 .maxlen = sizeof(int),
1789 .proc_handler = proc_dointvec_minmax,
1794 .procname = "suid_dumpable",
1795 .data = &suid_dumpable,
1796 .maxlen = sizeof(int),
1798 .proc_handler = proc_dointvec_minmax_coredump,
1802 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1804 .procname = "binfmt_misc",
1806 .child = sysctl_mount_point,
1810 .procname = "pipe-max-size",
1811 .data = &pipe_max_size,
1812 .maxlen = sizeof(int),
1814 .proc_handler = &pipe_proc_fn,
1815 .extra1 = &pipe_min_size,
1818 .procname = "pipe-user-pages-hard",
1819 .data = &pipe_user_pages_hard,
1820 .maxlen = sizeof(pipe_user_pages_hard),
1822 .proc_handler = proc_doulongvec_minmax,
1825 .procname = "pipe-user-pages-soft",
1826 .data = &pipe_user_pages_soft,
1827 .maxlen = sizeof(pipe_user_pages_soft),
1829 .proc_handler = proc_doulongvec_minmax,
1834 static struct ctl_table debug_table[] = {
1835 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1837 .procname = "exception-trace",
1838 .data = &show_unhandled_signals,
1839 .maxlen = sizeof(int),
1841 .proc_handler = proc_dointvec
1844 #if defined(CONFIG_OPTPROBES)
1846 .procname = "kprobes-optimization",
1847 .data = &sysctl_kprobes_optimization,
1848 .maxlen = sizeof(int),
1850 .proc_handler = proc_kprobes_optimization_handler,
1858 static struct ctl_table dev_table[] = {
1862 int __init sysctl_init(void)
1864 struct ctl_table_header *hdr;
1866 hdr = register_sysctl_table(sysctl_base_table);
1867 kmemleak_not_leak(hdr);
1871 #endif /* CONFIG_SYSCTL */
1877 #ifdef CONFIG_PROC_SYSCTL
1879 static int _proc_do_string(char *data, int maxlen, int write,
1880 char __user *buffer,
1881 size_t *lenp, loff_t *ppos)
1887 if (!data || !maxlen || !*lenp) {
1893 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1894 /* Only continue writes not past the end of buffer. */
1896 if (len > maxlen - 1)
1903 /* Start writing from beginning of buffer. */
1909 while ((p - buffer) < *lenp && len < maxlen - 1) {
1910 if (get_user(c, p++))
1912 if (c == 0 || c == '\n')
1933 if (copy_to_user(buffer, data, len))
1936 if (put_user('\n', buffer + len))
1946 static void warn_sysctl_write(struct ctl_table *table)
1948 pr_warn_once("%s wrote to %s when file position was not 0!\n"
1949 "This will not be supported in the future. To silence this\n"
1950 "warning, set kernel.sysctl_writes_strict = -1\n",
1951 current->comm, table->procname);
1955 * proc_dostring - read a string sysctl
1956 * @table: the sysctl table
1957 * @write: %TRUE if this is a write to the sysctl file
1958 * @buffer: the user buffer
1959 * @lenp: the size of the user buffer
1960 * @ppos: file position
1962 * Reads/writes a string from/to the user buffer. If the kernel
1963 * buffer provided is not large enough to hold the string, the
1964 * string is truncated. The copied string is %NULL-terminated.
1965 * If the string is being read by the user process, it is copied
1966 * and a newline '\n' is added. It is truncated if the buffer is
1969 * Returns 0 on success.
1971 int proc_dostring(struct ctl_table *table, int write,
1972 void __user *buffer, size_t *lenp, loff_t *ppos)
1974 if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN)
1975 warn_sysctl_write(table);
1977 return _proc_do_string((char *)(table->data), table->maxlen, write,
1978 (char __user *)buffer, lenp, ppos);
1981 static size_t proc_skip_spaces(char **buf)
1984 char *tmp = skip_spaces(*buf);
1990 static void proc_skip_char(char **buf, size_t *size, const char v)
2000 #define TMPBUFLEN 22
2002 * proc_get_long - reads an ASCII formatted integer from a user buffer
2004 * @buf: a kernel buffer
2005 * @size: size of the kernel buffer
2006 * @val: this is where the number will be stored
2007 * @neg: set to %TRUE if number is negative
2008 * @perm_tr: a vector which contains the allowed trailers
2009 * @perm_tr_len: size of the perm_tr vector
2010 * @tr: pointer to store the trailer character
2012 * In case of success %0 is returned and @buf and @size are updated with
2013 * the amount of bytes read. If @tr is non-NULL and a trailing
2014 * character exists (size is non-zero after returning from this
2015 * function), @tr is updated with the trailing character.
2017 static int proc_get_long(char **buf, size_t *size,
2018 unsigned long *val, bool *neg,
2019 const char *perm_tr, unsigned perm_tr_len, char *tr)
2022 char *p, tmp[TMPBUFLEN];
2028 if (len > TMPBUFLEN - 1)
2029 len = TMPBUFLEN - 1;
2031 memcpy(tmp, *buf, len);
2035 if (*p == '-' && *size > 1) {
2043 *val = simple_strtoul(p, &p, 0);
2047 /* We don't know if the next char is whitespace thus we may accept
2048 * invalid integers (e.g. 1234...a) or two integers instead of one
2049 * (e.g. 123...1). So lets not allow such large numbers. */
2050 if (len == TMPBUFLEN - 1)
2053 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2056 if (tr && (len < *size))
2066 * proc_put_long - converts an integer to a decimal ASCII formatted string
2068 * @buf: the user buffer
2069 * @size: the size of the user buffer
2070 * @val: the integer to be converted
2071 * @neg: sign of the number, %TRUE for negative
2073 * In case of success %0 is returned and @buf and @size are updated with
2074 * the amount of bytes written.
2076 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2080 char tmp[TMPBUFLEN], *p = tmp;
2082 sprintf(p, "%s%lu", neg ? "-" : "", val);
2086 if (copy_to_user(*buf, tmp, len))
2094 static int proc_put_char(void __user **buf, size_t *size, char c)
2097 char __user **buffer = (char __user **)buf;
2098 if (put_user(c, *buffer))
2100 (*size)--, (*buffer)++;
2106 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2108 int write, void *data)
2112 if (*lvalp > (unsigned long) INT_MAX + 1)
2116 if (*lvalp > (unsigned long) INT_MAX)
2124 *lvalp = -(unsigned long)val;
2127 *lvalp = (unsigned long)val;
2133 static int do_proc_douintvec_conv(bool *negp, unsigned long *lvalp,
2135 int write, void *data)
2142 unsigned int val = *valp;
2143 *lvalp = (unsigned long)val;
2148 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2150 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2151 int write, void __user *buffer,
2152 size_t *lenp, loff_t *ppos,
2153 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2154 int write, void *data),
2157 int *i, vleft, first = 1, err = 0;
2159 char *kbuf = NULL, *p;
2161 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2166 i = (int *) tbl_data;
2167 vleft = table->maxlen / sizeof(*i);
2171 conv = do_proc_dointvec_conv;
2175 switch (sysctl_writes_strict) {
2176 case SYSCTL_WRITES_STRICT:
2178 case SYSCTL_WRITES_WARN:
2179 warn_sysctl_write(table);
2186 if (left > PAGE_SIZE - 1)
2187 left = PAGE_SIZE - 1;
2188 p = kbuf = memdup_user_nul(buffer, left);
2190 return PTR_ERR(kbuf);
2193 for (; left && vleft--; i++, first=0) {
2198 left -= proc_skip_spaces(&p);
2202 err = proc_get_long(&p, &left, &lval, &neg,
2204 sizeof(proc_wspace_sep), NULL);
2207 if (conv(&neg, &lval, i, 1, data)) {
2212 if (conv(&neg, &lval, i, 0, data)) {
2217 err = proc_put_char(&buffer, &left, '\t');
2220 err = proc_put_long(&buffer, &left, lval, neg);
2226 if (!write && !first && left && !err)
2227 err = proc_put_char(&buffer, &left, '\n');
2228 if (write && !err && left)
2229 left -= proc_skip_spaces(&p);
2233 return err ? : -EINVAL;
2241 static int do_proc_dointvec(struct ctl_table *table, int write,
2242 void __user *buffer, size_t *lenp, loff_t *ppos,
2243 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2244 int write, void *data),
2247 return __do_proc_dointvec(table->data, table, write,
2248 buffer, lenp, ppos, conv, data);
2252 * proc_dointvec - read a vector of integers
2253 * @table: the sysctl table
2254 * @write: %TRUE if this is a write to the sysctl file
2255 * @buffer: the user buffer
2256 * @lenp: the size of the user buffer
2257 * @ppos: file position
2259 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2260 * values from/to the user buffer, treated as an ASCII string.
2262 * Returns 0 on success.
2264 int proc_dointvec(struct ctl_table *table, int write,
2265 void __user *buffer, size_t *lenp, loff_t *ppos)
2267 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2271 * proc_douintvec - read a vector of unsigned integers
2272 * @table: the sysctl table
2273 * @write: %TRUE if this is a write to the sysctl file
2274 * @buffer: the user buffer
2275 * @lenp: the size of the user buffer
2276 * @ppos: file position
2278 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2279 * values from/to the user buffer, treated as an ASCII string.
2281 * Returns 0 on success.
2283 int proc_douintvec(struct ctl_table *table, int write,
2284 void __user *buffer, size_t *lenp, loff_t *ppos)
2286 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2287 do_proc_douintvec_conv, NULL);
2291 * Taint values can only be increased
2292 * This means we can safely use a temporary.
2294 static int proc_taint(struct ctl_table *table, int write,
2295 void __user *buffer, size_t *lenp, loff_t *ppos)
2298 unsigned long tmptaint = get_taint();
2301 if (write && !capable(CAP_SYS_ADMIN))
2306 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2312 * Poor man's atomic or. Not worth adding a primitive
2313 * to everyone's atomic.h for this
2316 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2317 if ((tmptaint >> i) & 1)
2318 add_taint(i, LOCKDEP_STILL_OK);
2325 #ifdef CONFIG_PRINTK
2326 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2327 void __user *buffer, size_t *lenp, loff_t *ppos)
2329 if (write && !capable(CAP_SYS_ADMIN))
2332 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2336 struct do_proc_dointvec_minmax_conv_param {
2341 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2343 int write, void *data)
2345 struct do_proc_dointvec_minmax_conv_param *param = data;
2347 int val = *negp ? -*lvalp : *lvalp;
2348 if ((param->min && *param->min > val) ||
2349 (param->max && *param->max < val))
2356 *lvalp = -(unsigned long)val;
2359 *lvalp = (unsigned long)val;
2366 * proc_dointvec_minmax - read a vector of integers with min/max values
2367 * @table: the sysctl table
2368 * @write: %TRUE if this is a write to the sysctl file
2369 * @buffer: the user buffer
2370 * @lenp: the size of the user buffer
2371 * @ppos: file position
2373 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2374 * values from/to the user buffer, treated as an ASCII string.
2376 * This routine will ensure the values are within the range specified by
2377 * table->extra1 (min) and table->extra2 (max).
2379 * Returns 0 on success.
2381 int proc_dointvec_minmax(struct ctl_table *table, int write,
2382 void __user *buffer, size_t *lenp, loff_t *ppos)
2384 struct do_proc_dointvec_minmax_conv_param param = {
2385 .min = (int *) table->extra1,
2386 .max = (int *) table->extra2,
2388 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2389 do_proc_dointvec_minmax_conv, ¶m);
2392 static void validate_coredump_safety(void)
2394 #ifdef CONFIG_COREDUMP
2395 if (suid_dumpable == SUID_DUMP_ROOT &&
2396 core_pattern[0] != '/' && core_pattern[0] != '|') {
2397 printk(KERN_WARNING "Unsafe core_pattern used with "\
2398 "suid_dumpable=2. Pipe handler or fully qualified "\
2399 "core dump path required.\n");
2404 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2405 void __user *buffer, size_t *lenp, loff_t *ppos)
2407 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2409 validate_coredump_safety();
2413 #ifdef CONFIG_COREDUMP
2414 static int proc_dostring_coredump(struct ctl_table *table, int write,
2415 void __user *buffer, size_t *lenp, loff_t *ppos)
2417 int error = proc_dostring(table, write, buffer, lenp, ppos);
2419 validate_coredump_safety();
2424 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2425 void __user *buffer,
2426 size_t *lenp, loff_t *ppos,
2427 unsigned long convmul,
2428 unsigned long convdiv)
2430 unsigned long *i, *min, *max;
2431 int vleft, first = 1, err = 0;
2433 char *kbuf = NULL, *p;
2435 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2440 i = (unsigned long *) data;
2441 min = (unsigned long *) table->extra1;
2442 max = (unsigned long *) table->extra2;
2443 vleft = table->maxlen / sizeof(unsigned long);
2448 switch (sysctl_writes_strict) {
2449 case SYSCTL_WRITES_STRICT:
2451 case SYSCTL_WRITES_WARN:
2452 warn_sysctl_write(table);
2459 if (left > PAGE_SIZE - 1)
2460 left = PAGE_SIZE - 1;
2461 p = kbuf = memdup_user_nul(buffer, left);
2463 return PTR_ERR(kbuf);
2466 for (; left && vleft--; i++, first = 0) {
2472 left -= proc_skip_spaces(&p);
2474 err = proc_get_long(&p, &left, &val, &neg,
2476 sizeof(proc_wspace_sep), NULL);
2481 if ((min && val < *min) || (max && val > *max))
2485 val = convdiv * (*i) / convmul;
2487 err = proc_put_char(&buffer, &left, '\t');
2491 err = proc_put_long(&buffer, &left, val, false);
2497 if (!write && !first && left && !err)
2498 err = proc_put_char(&buffer, &left, '\n');
2500 left -= proc_skip_spaces(&p);
2504 return err ? : -EINVAL;
2512 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2513 void __user *buffer,
2514 size_t *lenp, loff_t *ppos,
2515 unsigned long convmul,
2516 unsigned long convdiv)
2518 return __do_proc_doulongvec_minmax(table->data, table, write,
2519 buffer, lenp, ppos, convmul, convdiv);
2523 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2524 * @table: the sysctl table
2525 * @write: %TRUE if this is a write to the sysctl file
2526 * @buffer: the user buffer
2527 * @lenp: the size of the user buffer
2528 * @ppos: file position
2530 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2531 * values from/to the user buffer, treated as an ASCII string.
2533 * This routine will ensure the values are within the range specified by
2534 * table->extra1 (min) and table->extra2 (max).
2536 * Returns 0 on success.
2538 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2539 void __user *buffer, size_t *lenp, loff_t *ppos)
2541 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2545 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
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 long) unsigned long
2553 * values from/to the user buffer, treated as an ASCII string. The values
2554 * are treated as milliseconds, and converted to jiffies when they are stored.
2556 * This routine will ensure the values are within the range specified by
2557 * table->extra1 (min) and table->extra2 (max).
2559 * Returns 0 on success.
2561 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2562 void __user *buffer,
2563 size_t *lenp, loff_t *ppos)
2565 return do_proc_doulongvec_minmax(table, write, buffer,
2566 lenp, ppos, HZ, 1000l);
2570 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2572 int write, void *data)
2575 if (*lvalp > LONG_MAX / HZ)
2577 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2583 lval = -(unsigned long)val;
2586 lval = (unsigned long)val;
2593 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2595 int write, void *data)
2598 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2600 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2606 lval = -(unsigned long)val;
2609 lval = (unsigned long)val;
2611 *lvalp = jiffies_to_clock_t(lval);
2616 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2618 int write, void *data)
2621 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2631 lval = -(unsigned long)val;
2634 lval = (unsigned long)val;
2636 *lvalp = jiffies_to_msecs(lval);
2642 * proc_dointvec_jiffies - read a vector of integers as seconds
2643 * @table: the sysctl table
2644 * @write: %TRUE if this is a write to the sysctl file
2645 * @buffer: the user buffer
2646 * @lenp: the size of the user buffer
2647 * @ppos: file position
2649 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2650 * values from/to the user buffer, treated as an ASCII string.
2651 * The values read are assumed to be in seconds, and are converted into
2654 * Returns 0 on success.
2656 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2657 void __user *buffer, size_t *lenp, loff_t *ppos)
2659 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2660 do_proc_dointvec_jiffies_conv,NULL);
2664 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2665 * @table: the sysctl table
2666 * @write: %TRUE if this is a write to the sysctl file
2667 * @buffer: the user buffer
2668 * @lenp: the size of the user buffer
2669 * @ppos: pointer to the file position
2671 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2672 * values from/to the user buffer, treated as an ASCII string.
2673 * The values read are assumed to be in 1/USER_HZ seconds, and
2674 * are converted into jiffies.
2676 * Returns 0 on success.
2678 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2679 void __user *buffer, size_t *lenp, loff_t *ppos)
2681 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2682 do_proc_dointvec_userhz_jiffies_conv,NULL);
2686 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2687 * @table: the sysctl table
2688 * @write: %TRUE if this is a write to the sysctl file
2689 * @buffer: the user buffer
2690 * @lenp: the size of the user buffer
2691 * @ppos: file position
2692 * @ppos: the current position in the file
2694 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2695 * values from/to the user buffer, treated as an ASCII string.
2696 * The values read are assumed to be in 1/1000 seconds, and
2697 * are converted into jiffies.
2699 * Returns 0 on success.
2701 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2702 void __user *buffer, size_t *lenp, loff_t *ppos)
2704 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2705 do_proc_dointvec_ms_jiffies_conv, NULL);
2708 static int proc_do_cad_pid(struct ctl_table *table, int write,
2709 void __user *buffer, size_t *lenp, loff_t *ppos)
2711 struct pid *new_pid;
2715 tmp = pid_vnr(cad_pid);
2717 r = __do_proc_dointvec(&tmp, table, write, buffer,
2718 lenp, ppos, NULL, NULL);
2722 new_pid = find_get_pid(tmp);
2726 put_pid(xchg(&cad_pid, new_pid));
2731 * proc_do_large_bitmap - read/write from/to a large bitmap
2732 * @table: the sysctl table
2733 * @write: %TRUE if this is a write to the sysctl file
2734 * @buffer: the user buffer
2735 * @lenp: the size of the user buffer
2736 * @ppos: file position
2738 * The bitmap is stored at table->data and the bitmap length (in bits)
2741 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2742 * large bitmaps may be represented in a compact manner. Writing into
2743 * the file will clear the bitmap then update it with the given input.
2745 * Returns 0 on success.
2747 int proc_do_large_bitmap(struct ctl_table *table, int write,
2748 void __user *buffer, size_t *lenp, loff_t *ppos)
2752 size_t left = *lenp;
2753 unsigned long bitmap_len = table->maxlen;
2754 unsigned long *bitmap = *(unsigned long **) table->data;
2755 unsigned long *tmp_bitmap = NULL;
2756 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2758 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2766 if (left > PAGE_SIZE - 1)
2767 left = PAGE_SIZE - 1;
2769 p = kbuf = memdup_user_nul(buffer, left);
2771 return PTR_ERR(kbuf);
2773 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2779 proc_skip_char(&p, &left, '\n');
2780 while (!err && left) {
2781 unsigned long val_a, val_b;
2784 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
2788 if (val_a >= bitmap_len || neg) {
2800 err = proc_get_long(&p, &left, &val_b,
2801 &neg, tr_b, sizeof(tr_b),
2805 if (val_b >= bitmap_len || neg ||
2816 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2818 proc_skip_char(&p, &left, '\n');
2822 unsigned long bit_a, bit_b = 0;
2825 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2826 if (bit_a >= bitmap_len)
2828 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2832 err = proc_put_char(&buffer, &left, ',');
2836 err = proc_put_long(&buffer, &left, bit_a, false);
2839 if (bit_a != bit_b) {
2840 err = proc_put_char(&buffer, &left, '-');
2843 err = proc_put_long(&buffer, &left, bit_b, false);
2851 err = proc_put_char(&buffer, &left, '\n');
2857 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2859 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2871 #else /* CONFIG_PROC_SYSCTL */
2873 int proc_dostring(struct ctl_table *table, int write,
2874 void __user *buffer, size_t *lenp, loff_t *ppos)
2879 int proc_dointvec(struct ctl_table *table, int write,
2880 void __user *buffer, size_t *lenp, loff_t *ppos)
2885 int proc_douintvec(struct ctl_table *table, int write,
2886 void __user *buffer, size_t *lenp, loff_t *ppos)
2891 int proc_dointvec_minmax(struct ctl_table *table, int write,
2892 void __user *buffer, size_t *lenp, loff_t *ppos)
2897 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2898 void __user *buffer, size_t *lenp, loff_t *ppos)
2903 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2904 void __user *buffer, size_t *lenp, loff_t *ppos)
2909 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2910 void __user *buffer, size_t *lenp, loff_t *ppos)
2915 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2916 void __user *buffer, size_t *lenp, loff_t *ppos)
2921 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2922 void __user *buffer,
2923 size_t *lenp, loff_t *ppos)
2929 #endif /* CONFIG_PROC_SYSCTL */
2932 * No sense putting this after each symbol definition, twice,
2933 * exception granted :-)
2935 EXPORT_SYMBOL(proc_dointvec);
2936 EXPORT_SYMBOL(proc_douintvec);
2937 EXPORT_SYMBOL(proc_dointvec_jiffies);
2938 EXPORT_SYMBOL(proc_dointvec_minmax);
2939 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2940 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2941 EXPORT_SYMBOL(proc_dostring);
2942 EXPORT_SYMBOL(proc_doulongvec_minmax);
2943 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);