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/kmemleak.h>
35 #include <linux/init.h>
36 #include <linux/kernel.h>
37 #include <linux/kobject.h>
38 #include <linux/net.h>
39 #include <linux/sysrq.h>
40 #include <linux/highuid.h>
41 #include <linux/writeback.h>
42 #include <linux/ratelimit.h>
43 #include <linux/compaction.h>
44 #include <linux/hugetlb.h>
45 #include <linux/initrd.h>
46 #include <linux/key.h>
47 #include <linux/times.h>
48 #include <linux/limits.h>
49 #include <linux/dcache.h>
50 #include <linux/dnotify.h>
51 #include <linux/syscalls.h>
52 #include <linux/vmstat.h>
53 #include <linux/nfs_fs.h>
54 #include <linux/acpi.h>
55 #include <linux/reboot.h>
56 #include <linux/ftrace.h>
57 #include <linux/perf_event.h>
58 #include <linux/kprobes.h>
59 #include <linux/pipe_fs_i.h>
60 #include <linux/oom.h>
61 #include <linux/kmod.h>
62 #include <linux/capability.h>
63 #include <linux/binfmts.h>
64 #include <linux/sched/sysctl.h>
65 #include <linux/sched/coredump.h>
66 #include <linux/kexec.h>
67 #include <linux/bpf.h>
68 #include <linux/mount.h>
70 #include "../lib/kstrtox.h"
72 #include <linux/uaccess.h>
73 #include <asm/processor.h>
77 #include <asm/stacktrace.h>
81 #include <asm/setup.h>
83 #ifdef CONFIG_BSD_PROCESS_ACCT
84 #include <linux/acct.h>
86 #ifdef CONFIG_RT_MUTEXES
87 #include <linux/rtmutex.h>
89 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
90 #include <linux/lockdep.h>
92 #ifdef CONFIG_CHR_DEV_SG
95 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
96 #include <linux/stackleak.h>
98 #ifdef CONFIG_LOCKUP_DETECTOR
99 #include <linux/nmi.h>
102 #if defined(CONFIG_SYSCTL)
104 /* External variables not in a header file. */
105 extern int suid_dumpable;
106 #ifdef CONFIG_COREDUMP
107 extern int core_uses_pid;
108 extern char core_pattern[];
109 extern unsigned int core_pipe_limit;
112 extern int pid_max_min, pid_max_max;
113 extern int percpu_pagelist_fraction;
114 extern int latencytop_enabled;
115 extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
117 extern int sysctl_nr_trim_pages;
120 /* Constants used for minimum and maximum */
121 #ifdef CONFIG_LOCKUP_DETECTOR
122 static int sixty = 60;
125 static int __maybe_unused neg_one = -1;
128 static int __maybe_unused one = 1;
129 static int __maybe_unused two = 2;
130 static int __maybe_unused four = 4;
131 static unsigned long one_ul = 1;
132 static unsigned long long_max = LONG_MAX;
133 static int one_hundred = 100;
134 static int one_thousand = 1000;
136 static int ten_thousand = 10000;
138 #ifdef CONFIG_PERF_EVENTS
139 static int six_hundred_forty_kb = 640 * 1024;
142 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
143 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
145 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
146 static int maxolduid = 65535;
147 static int minolduid;
149 static int ngroups_max = NGROUPS_MAX;
150 static const int cap_last_cap = CAP_LAST_CAP;
153 * This is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs
154 * and hung_task_check_interval_secs
156 #ifdef CONFIG_DETECT_HUNG_TASK
157 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
160 #ifdef CONFIG_INOTIFY_USER
161 #include <linux/inotify.h>
167 extern int pwrsw_enabled;
170 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
171 extern int unaligned_enabled;
175 extern int unaligned_dump_stack;
178 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
179 extern int no_unaligned_warning;
182 #ifdef CONFIG_PROC_SYSCTL
185 * enum sysctl_writes_mode - supported sysctl write modes
187 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
188 * to be written, and multiple writes on the same sysctl file descriptor
189 * will rewrite the sysctl value, regardless of file position. No warning
190 * is issued when the initial position is not 0.
191 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
193 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
194 * file position 0 and the value must be fully contained in the buffer
195 * sent to the write syscall. If dealing with strings respect the file
196 * position, but restrict this to the max length of the buffer, anything
197 * passed the max lenght will be ignored. Multiple writes will append
200 * These write modes control how current file position affects the behavior of
201 * updating sysctl values through the proc interface on each write.
203 enum sysctl_writes_mode {
204 SYSCTL_WRITES_LEGACY = -1,
205 SYSCTL_WRITES_WARN = 0,
206 SYSCTL_WRITES_STRICT = 1,
209 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
211 static int proc_do_cad_pid(struct ctl_table *table, int write,
212 void __user *buffer, size_t *lenp, loff_t *ppos);
213 static int proc_taint(struct ctl_table *table, int write,
214 void __user *buffer, size_t *lenp, loff_t *ppos);
218 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
219 void __user *buffer, size_t *lenp, loff_t *ppos);
222 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
223 void __user *buffer, size_t *lenp, loff_t *ppos);
224 #ifdef CONFIG_COREDUMP
225 static int proc_dostring_coredump(struct ctl_table *table, int write,
226 void __user *buffer, size_t *lenp, loff_t *ppos);
228 static int proc_dopipe_max_size(struct ctl_table *table, int write,
229 void __user *buffer, size_t *lenp, loff_t *ppos);
230 #ifdef CONFIG_BPF_SYSCALL
231 static int proc_dointvec_minmax_bpf_stats(struct ctl_table *table, int write,
232 void __user *buffer, size_t *lenp,
236 #ifdef CONFIG_MAGIC_SYSRQ
237 /* Note: sysrq code uses its own private copy */
238 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
240 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
241 void __user *buffer, size_t *lenp,
246 error = proc_dointvec(table, write, buffer, lenp, ppos);
251 sysrq_toggle_support(__sysrq_enabled);
258 static struct ctl_table kern_table[];
259 static struct ctl_table vm_table[];
260 static struct ctl_table fs_table[];
261 static struct ctl_table debug_table[];
262 static struct ctl_table dev_table[];
263 extern struct ctl_table random_table[];
265 extern struct ctl_table epoll_table[];
268 #ifdef CONFIG_FW_LOADER_USER_HELPER
269 extern struct ctl_table firmware_config_table[];
272 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
273 int sysctl_legacy_va_layout;
276 /* The default sysctl tables: */
278 static struct ctl_table sysctl_base_table[] = {
280 .procname = "kernel",
297 .child = debug_table,
307 #ifdef CONFIG_SCHED_DEBUG
308 static int min_sched_granularity_ns = 100000; /* 100 usecs */
309 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
310 static int min_wakeup_granularity_ns; /* 0 usecs */
311 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
313 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
314 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
315 #endif /* CONFIG_SMP */
316 #endif /* CONFIG_SCHED_DEBUG */
318 #ifdef CONFIG_COMPACTION
319 static int min_extfrag_threshold;
320 static int max_extfrag_threshold = 1000;
323 static struct ctl_table kern_table[] = {
325 .procname = "sched_child_runs_first",
326 .data = &sysctl_sched_child_runs_first,
327 .maxlen = sizeof(unsigned int),
329 .proc_handler = proc_dointvec,
331 #ifdef CONFIG_SCHED_DEBUG
333 .procname = "sched_min_granularity_ns",
334 .data = &sysctl_sched_min_granularity,
335 .maxlen = sizeof(unsigned int),
337 .proc_handler = sched_proc_update_handler,
338 .extra1 = &min_sched_granularity_ns,
339 .extra2 = &max_sched_granularity_ns,
342 .procname = "sched_latency_ns",
343 .data = &sysctl_sched_latency,
344 .maxlen = sizeof(unsigned int),
346 .proc_handler = sched_proc_update_handler,
347 .extra1 = &min_sched_granularity_ns,
348 .extra2 = &max_sched_granularity_ns,
351 .procname = "sched_wakeup_granularity_ns",
352 .data = &sysctl_sched_wakeup_granularity,
353 .maxlen = sizeof(unsigned int),
355 .proc_handler = sched_proc_update_handler,
356 .extra1 = &min_wakeup_granularity_ns,
357 .extra2 = &max_wakeup_granularity_ns,
361 .procname = "sched_tunable_scaling",
362 .data = &sysctl_sched_tunable_scaling,
363 .maxlen = sizeof(enum sched_tunable_scaling),
365 .proc_handler = sched_proc_update_handler,
366 .extra1 = &min_sched_tunable_scaling,
367 .extra2 = &max_sched_tunable_scaling,
370 .procname = "sched_migration_cost_ns",
371 .data = &sysctl_sched_migration_cost,
372 .maxlen = sizeof(unsigned int),
374 .proc_handler = proc_dointvec,
377 .procname = "sched_nr_migrate",
378 .data = &sysctl_sched_nr_migrate,
379 .maxlen = sizeof(unsigned int),
381 .proc_handler = proc_dointvec,
383 #ifdef CONFIG_SCHEDSTATS
385 .procname = "sched_schedstats",
387 .maxlen = sizeof(unsigned int),
389 .proc_handler = sysctl_schedstats,
393 #endif /* CONFIG_SCHEDSTATS */
394 #endif /* CONFIG_SMP */
395 #ifdef CONFIG_NUMA_BALANCING
397 .procname = "numa_balancing_scan_delay_ms",
398 .data = &sysctl_numa_balancing_scan_delay,
399 .maxlen = sizeof(unsigned int),
401 .proc_handler = proc_dointvec,
404 .procname = "numa_balancing_scan_period_min_ms",
405 .data = &sysctl_numa_balancing_scan_period_min,
406 .maxlen = sizeof(unsigned int),
408 .proc_handler = proc_dointvec,
411 .procname = "numa_balancing_scan_period_max_ms",
412 .data = &sysctl_numa_balancing_scan_period_max,
413 .maxlen = sizeof(unsigned int),
415 .proc_handler = proc_dointvec,
418 .procname = "numa_balancing_scan_size_mb",
419 .data = &sysctl_numa_balancing_scan_size,
420 .maxlen = sizeof(unsigned int),
422 .proc_handler = proc_dointvec_minmax,
426 .procname = "numa_balancing",
427 .data = NULL, /* filled in by handler */
428 .maxlen = sizeof(unsigned int),
430 .proc_handler = sysctl_numa_balancing,
434 #endif /* CONFIG_NUMA_BALANCING */
435 #endif /* CONFIG_SCHED_DEBUG */
437 .procname = "sched_rt_period_us",
438 .data = &sysctl_sched_rt_period,
439 .maxlen = sizeof(unsigned int),
441 .proc_handler = sched_rt_handler,
444 .procname = "sched_rt_runtime_us",
445 .data = &sysctl_sched_rt_runtime,
446 .maxlen = sizeof(int),
448 .proc_handler = sched_rt_handler,
451 .procname = "sched_rr_timeslice_ms",
452 .data = &sysctl_sched_rr_timeslice,
453 .maxlen = sizeof(int),
455 .proc_handler = sched_rr_handler,
457 #ifdef CONFIG_SCHED_AUTOGROUP
459 .procname = "sched_autogroup_enabled",
460 .data = &sysctl_sched_autogroup_enabled,
461 .maxlen = sizeof(unsigned int),
463 .proc_handler = proc_dointvec_minmax,
468 #ifdef CONFIG_CFS_BANDWIDTH
470 .procname = "sched_cfs_bandwidth_slice_us",
471 .data = &sysctl_sched_cfs_bandwidth_slice,
472 .maxlen = sizeof(unsigned int),
474 .proc_handler = proc_dointvec_minmax,
478 #if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_CPU_FREQ_GOV_SCHEDUTIL)
480 .procname = "sched_energy_aware",
481 .data = &sysctl_sched_energy_aware,
482 .maxlen = sizeof(unsigned int),
484 .proc_handler = sched_energy_aware_handler,
489 #ifdef CONFIG_PROVE_LOCKING
491 .procname = "prove_locking",
492 .data = &prove_locking,
493 .maxlen = sizeof(int),
495 .proc_handler = proc_dointvec,
498 #ifdef CONFIG_LOCK_STAT
500 .procname = "lock_stat",
502 .maxlen = sizeof(int),
504 .proc_handler = proc_dointvec,
509 .data = &panic_timeout,
510 .maxlen = sizeof(int),
512 .proc_handler = proc_dointvec,
514 #ifdef CONFIG_COREDUMP
516 .procname = "core_uses_pid",
517 .data = &core_uses_pid,
518 .maxlen = sizeof(int),
520 .proc_handler = proc_dointvec,
523 .procname = "core_pattern",
524 .data = core_pattern,
525 .maxlen = CORENAME_MAX_SIZE,
527 .proc_handler = proc_dostring_coredump,
530 .procname = "core_pipe_limit",
531 .data = &core_pipe_limit,
532 .maxlen = sizeof(unsigned int),
534 .proc_handler = proc_dointvec,
537 #ifdef CONFIG_PROC_SYSCTL
539 .procname = "tainted",
540 .maxlen = sizeof(long),
542 .proc_handler = proc_taint,
545 .procname = "sysctl_writes_strict",
546 .data = &sysctl_writes_strict,
547 .maxlen = sizeof(int),
549 .proc_handler = proc_dointvec_minmax,
554 #ifdef CONFIG_LATENCYTOP
556 .procname = "latencytop",
557 .data = &latencytop_enabled,
558 .maxlen = sizeof(int),
560 .proc_handler = sysctl_latencytop,
563 #ifdef CONFIG_BLK_DEV_INITRD
565 .procname = "real-root-dev",
566 .data = &real_root_dev,
567 .maxlen = sizeof(int),
569 .proc_handler = proc_dointvec,
573 .procname = "print-fatal-signals",
574 .data = &print_fatal_signals,
575 .maxlen = sizeof(int),
577 .proc_handler = proc_dointvec,
581 .procname = "reboot-cmd",
582 .data = reboot_command,
585 .proc_handler = proc_dostring,
588 .procname = "stop-a",
589 .data = &stop_a_enabled,
590 .maxlen = sizeof (int),
592 .proc_handler = proc_dointvec,
595 .procname = "scons-poweroff",
596 .data = &scons_pwroff,
597 .maxlen = sizeof (int),
599 .proc_handler = proc_dointvec,
602 #ifdef CONFIG_SPARC64
604 .procname = "tsb-ratio",
605 .data = &sysctl_tsb_ratio,
606 .maxlen = sizeof (int),
608 .proc_handler = proc_dointvec,
613 .procname = "soft-power",
614 .data = &pwrsw_enabled,
615 .maxlen = sizeof (int),
617 .proc_handler = proc_dointvec,
620 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
622 .procname = "unaligned-trap",
623 .data = &unaligned_enabled,
624 .maxlen = sizeof (int),
626 .proc_handler = proc_dointvec,
630 .procname = "ctrl-alt-del",
632 .maxlen = sizeof(int),
634 .proc_handler = proc_dointvec,
636 #ifdef CONFIG_FUNCTION_TRACER
638 .procname = "ftrace_enabled",
639 .data = &ftrace_enabled,
640 .maxlen = sizeof(int),
642 .proc_handler = ftrace_enable_sysctl,
645 #ifdef CONFIG_STACK_TRACER
647 .procname = "stack_tracer_enabled",
648 .data = &stack_tracer_enabled,
649 .maxlen = sizeof(int),
651 .proc_handler = stack_trace_sysctl,
654 #ifdef CONFIG_TRACING
656 .procname = "ftrace_dump_on_oops",
657 .data = &ftrace_dump_on_oops,
658 .maxlen = sizeof(int),
660 .proc_handler = proc_dointvec,
663 .procname = "traceoff_on_warning",
664 .data = &__disable_trace_on_warning,
665 .maxlen = sizeof(__disable_trace_on_warning),
667 .proc_handler = proc_dointvec,
670 .procname = "tracepoint_printk",
671 .data = &tracepoint_printk,
672 .maxlen = sizeof(tracepoint_printk),
674 .proc_handler = tracepoint_printk_sysctl,
677 #ifdef CONFIG_KEXEC_CORE
679 .procname = "kexec_load_disabled",
680 .data = &kexec_load_disabled,
681 .maxlen = sizeof(int),
683 /* only handle a transition from default "0" to "1" */
684 .proc_handler = proc_dointvec_minmax,
689 #ifdef CONFIG_MODULES
691 .procname = "modprobe",
692 .data = &modprobe_path,
693 .maxlen = KMOD_PATH_LEN,
695 .proc_handler = proc_dostring,
698 .procname = "modules_disabled",
699 .data = &modules_disabled,
700 .maxlen = sizeof(int),
702 /* only handle a transition from default "0" to "1" */
703 .proc_handler = proc_dointvec_minmax,
708 #ifdef CONFIG_UEVENT_HELPER
710 .procname = "hotplug",
711 .data = &uevent_helper,
712 .maxlen = UEVENT_HELPER_PATH_LEN,
714 .proc_handler = proc_dostring,
717 #ifdef CONFIG_CHR_DEV_SG
719 .procname = "sg-big-buff",
720 .data = &sg_big_buff,
721 .maxlen = sizeof (int),
723 .proc_handler = proc_dointvec,
726 #ifdef CONFIG_BSD_PROCESS_ACCT
730 .maxlen = 3*sizeof(int),
732 .proc_handler = proc_dointvec,
735 #ifdef CONFIG_MAGIC_SYSRQ
738 .data = &__sysrq_enabled,
739 .maxlen = sizeof (int),
741 .proc_handler = sysrq_sysctl_handler,
744 #ifdef CONFIG_PROC_SYSCTL
746 .procname = "cad_pid",
748 .maxlen = sizeof (int),
750 .proc_handler = proc_do_cad_pid,
754 .procname = "threads-max",
756 .maxlen = sizeof(int),
758 .proc_handler = sysctl_max_threads,
761 .procname = "random",
763 .child = random_table,
766 .procname = "usermodehelper",
768 .child = usermodehelper_table,
770 #ifdef CONFIG_FW_LOADER_USER_HELPER
772 .procname = "firmware_config",
774 .child = firmware_config_table,
778 .procname = "overflowuid",
779 .data = &overflowuid,
780 .maxlen = sizeof(int),
782 .proc_handler = proc_dointvec_minmax,
783 .extra1 = &minolduid,
784 .extra2 = &maxolduid,
787 .procname = "overflowgid",
788 .data = &overflowgid,
789 .maxlen = sizeof(int),
791 .proc_handler = proc_dointvec_minmax,
792 .extra1 = &minolduid,
793 .extra2 = &maxolduid,
796 #ifdef CONFIG_MATHEMU
798 .procname = "ieee_emulation_warnings",
799 .data = &sysctl_ieee_emulation_warnings,
800 .maxlen = sizeof(int),
802 .proc_handler = proc_dointvec,
806 .procname = "userprocess_debug",
807 .data = &show_unhandled_signals,
808 .maxlen = sizeof(int),
810 .proc_handler = proc_dointvec,
814 .procname = "pid_max",
816 .maxlen = sizeof (int),
818 .proc_handler = proc_dointvec_minmax,
819 .extra1 = &pid_max_min,
820 .extra2 = &pid_max_max,
823 .procname = "panic_on_oops",
824 .data = &panic_on_oops,
825 .maxlen = sizeof(int),
827 .proc_handler = proc_dointvec,
830 .procname = "panic_print",
831 .data = &panic_print,
832 .maxlen = sizeof(unsigned long),
834 .proc_handler = proc_doulongvec_minmax,
836 #if defined CONFIG_PRINTK
838 .procname = "printk",
839 .data = &console_loglevel,
840 .maxlen = 4*sizeof(int),
842 .proc_handler = proc_dointvec,
845 .procname = "printk_ratelimit",
846 .data = &printk_ratelimit_state.interval,
847 .maxlen = sizeof(int),
849 .proc_handler = proc_dointvec_jiffies,
852 .procname = "printk_ratelimit_burst",
853 .data = &printk_ratelimit_state.burst,
854 .maxlen = sizeof(int),
856 .proc_handler = proc_dointvec,
859 .procname = "printk_delay",
860 .data = &printk_delay_msec,
861 .maxlen = sizeof(int),
863 .proc_handler = proc_dointvec_minmax,
865 .extra2 = &ten_thousand,
868 .procname = "printk_devkmsg",
869 .data = devkmsg_log_str,
870 .maxlen = DEVKMSG_STR_MAX_SIZE,
872 .proc_handler = devkmsg_sysctl_set_loglvl,
875 .procname = "dmesg_restrict",
876 .data = &dmesg_restrict,
877 .maxlen = sizeof(int),
879 .proc_handler = proc_dointvec_minmax_sysadmin,
884 .procname = "kptr_restrict",
885 .data = &kptr_restrict,
886 .maxlen = sizeof(int),
888 .proc_handler = proc_dointvec_minmax_sysadmin,
894 .procname = "ngroups_max",
895 .data = &ngroups_max,
896 .maxlen = sizeof (int),
898 .proc_handler = proc_dointvec,
901 .procname = "cap_last_cap",
902 .data = (void *)&cap_last_cap,
903 .maxlen = sizeof(int),
905 .proc_handler = proc_dointvec,
907 #if defined(CONFIG_LOCKUP_DETECTOR)
909 .procname = "watchdog",
910 .data = &watchdog_user_enabled,
911 .maxlen = sizeof(int),
913 .proc_handler = proc_watchdog,
918 .procname = "watchdog_thresh",
919 .data = &watchdog_thresh,
920 .maxlen = sizeof(int),
922 .proc_handler = proc_watchdog_thresh,
927 .procname = "nmi_watchdog",
928 .data = &nmi_watchdog_user_enabled,
929 .maxlen = sizeof(int),
930 .mode = NMI_WATCHDOG_SYSCTL_PERM,
931 .proc_handler = proc_nmi_watchdog,
936 .procname = "watchdog_cpumask",
937 .data = &watchdog_cpumask_bits,
940 .proc_handler = proc_watchdog_cpumask,
942 #ifdef CONFIG_SOFTLOCKUP_DETECTOR
944 .procname = "soft_watchdog",
945 .data = &soft_watchdog_user_enabled,
946 .maxlen = sizeof(int),
948 .proc_handler = proc_soft_watchdog,
953 .procname = "softlockup_panic",
954 .data = &softlockup_panic,
955 .maxlen = sizeof(int),
957 .proc_handler = proc_dointvec_minmax,
963 .procname = "softlockup_all_cpu_backtrace",
964 .data = &sysctl_softlockup_all_cpu_backtrace,
965 .maxlen = sizeof(int),
967 .proc_handler = proc_dointvec_minmax,
971 #endif /* CONFIG_SMP */
973 #ifdef CONFIG_HARDLOCKUP_DETECTOR
975 .procname = "hardlockup_panic",
976 .data = &hardlockup_panic,
977 .maxlen = sizeof(int),
979 .proc_handler = proc_dointvec_minmax,
985 .procname = "hardlockup_all_cpu_backtrace",
986 .data = &sysctl_hardlockup_all_cpu_backtrace,
987 .maxlen = sizeof(int),
989 .proc_handler = proc_dointvec_minmax,
993 #endif /* CONFIG_SMP */
997 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
999 .procname = "unknown_nmi_panic",
1000 .data = &unknown_nmi_panic,
1001 .maxlen = sizeof (int),
1003 .proc_handler = proc_dointvec,
1006 #if defined(CONFIG_X86)
1008 .procname = "panic_on_unrecovered_nmi",
1009 .data = &panic_on_unrecovered_nmi,
1010 .maxlen = sizeof(int),
1012 .proc_handler = proc_dointvec,
1015 .procname = "panic_on_io_nmi",
1016 .data = &panic_on_io_nmi,
1017 .maxlen = sizeof(int),
1019 .proc_handler = proc_dointvec,
1021 #ifdef CONFIG_DEBUG_STACKOVERFLOW
1023 .procname = "panic_on_stackoverflow",
1024 .data = &sysctl_panic_on_stackoverflow,
1025 .maxlen = sizeof(int),
1027 .proc_handler = proc_dointvec,
1031 .procname = "bootloader_type",
1032 .data = &bootloader_type,
1033 .maxlen = sizeof (int),
1035 .proc_handler = proc_dointvec,
1038 .procname = "bootloader_version",
1039 .data = &bootloader_version,
1040 .maxlen = sizeof (int),
1042 .proc_handler = proc_dointvec,
1045 .procname = "io_delay_type",
1046 .data = &io_delay_type,
1047 .maxlen = sizeof(int),
1049 .proc_handler = proc_dointvec,
1052 #if defined(CONFIG_MMU)
1054 .procname = "randomize_va_space",
1055 .data = &randomize_va_space,
1056 .maxlen = sizeof(int),
1058 .proc_handler = proc_dointvec,
1061 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1063 .procname = "spin_retry",
1064 .data = &spin_retry,
1065 .maxlen = sizeof (int),
1067 .proc_handler = proc_dointvec,
1070 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1072 .procname = "acpi_video_flags",
1073 .data = &acpi_realmode_flags,
1074 .maxlen = sizeof (unsigned long),
1076 .proc_handler = proc_doulongvec_minmax,
1079 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1081 .procname = "ignore-unaligned-usertrap",
1082 .data = &no_unaligned_warning,
1083 .maxlen = sizeof (int),
1085 .proc_handler = proc_dointvec,
1090 .procname = "unaligned-dump-stack",
1091 .data = &unaligned_dump_stack,
1092 .maxlen = sizeof (int),
1094 .proc_handler = proc_dointvec,
1097 #ifdef CONFIG_DETECT_HUNG_TASK
1099 .procname = "hung_task_panic",
1100 .data = &sysctl_hung_task_panic,
1101 .maxlen = sizeof(int),
1103 .proc_handler = proc_dointvec_minmax,
1108 .procname = "hung_task_check_count",
1109 .data = &sysctl_hung_task_check_count,
1110 .maxlen = sizeof(int),
1112 .proc_handler = proc_dointvec_minmax,
1116 .procname = "hung_task_timeout_secs",
1117 .data = &sysctl_hung_task_timeout_secs,
1118 .maxlen = sizeof(unsigned long),
1120 .proc_handler = proc_dohung_task_timeout_secs,
1121 .extra2 = &hung_task_timeout_max,
1124 .procname = "hung_task_check_interval_secs",
1125 .data = &sysctl_hung_task_check_interval_secs,
1126 .maxlen = sizeof(unsigned long),
1128 .proc_handler = proc_dohung_task_timeout_secs,
1129 .extra2 = &hung_task_timeout_max,
1132 .procname = "hung_task_warnings",
1133 .data = &sysctl_hung_task_warnings,
1134 .maxlen = sizeof(int),
1136 .proc_handler = proc_dointvec_minmax,
1140 #ifdef CONFIG_RT_MUTEXES
1142 .procname = "max_lock_depth",
1143 .data = &max_lock_depth,
1144 .maxlen = sizeof(int),
1146 .proc_handler = proc_dointvec,
1150 .procname = "poweroff_cmd",
1151 .data = &poweroff_cmd,
1152 .maxlen = POWEROFF_CMD_PATH_LEN,
1154 .proc_handler = proc_dostring,
1160 .child = key_sysctls,
1163 #ifdef CONFIG_PERF_EVENTS
1165 * User-space scripts rely on the existence of this file
1166 * as a feature check for perf_events being enabled.
1168 * So it's an ABI, do not remove!
1171 .procname = "perf_event_paranoid",
1172 .data = &sysctl_perf_event_paranoid,
1173 .maxlen = sizeof(sysctl_perf_event_paranoid),
1175 .proc_handler = proc_dointvec,
1178 .procname = "perf_event_mlock_kb",
1179 .data = &sysctl_perf_event_mlock,
1180 .maxlen = sizeof(sysctl_perf_event_mlock),
1182 .proc_handler = proc_dointvec,
1185 .procname = "perf_event_max_sample_rate",
1186 .data = &sysctl_perf_event_sample_rate,
1187 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1189 .proc_handler = perf_proc_update_handler,
1193 .procname = "perf_cpu_time_max_percent",
1194 .data = &sysctl_perf_cpu_time_max_percent,
1195 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1197 .proc_handler = perf_cpu_time_max_percent_handler,
1199 .extra2 = &one_hundred,
1202 .procname = "perf_event_max_stack",
1203 .data = &sysctl_perf_event_max_stack,
1204 .maxlen = sizeof(sysctl_perf_event_max_stack),
1206 .proc_handler = perf_event_max_stack_handler,
1208 .extra2 = &six_hundred_forty_kb,
1211 .procname = "perf_event_max_contexts_per_stack",
1212 .data = &sysctl_perf_event_max_contexts_per_stack,
1213 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
1215 .proc_handler = perf_event_max_stack_handler,
1217 .extra2 = &one_thousand,
1221 .procname = "panic_on_warn",
1222 .data = &panic_on_warn,
1223 .maxlen = sizeof(int),
1225 .proc_handler = proc_dointvec_minmax,
1229 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1231 .procname = "timer_migration",
1232 .data = &sysctl_timer_migration,
1233 .maxlen = sizeof(unsigned int),
1235 .proc_handler = timer_migration_handler,
1240 #ifdef CONFIG_BPF_SYSCALL
1242 .procname = "unprivileged_bpf_disabled",
1243 .data = &sysctl_unprivileged_bpf_disabled,
1244 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
1246 /* only handle a transition from default "0" to "1" */
1247 .proc_handler = proc_dointvec_minmax,
1252 .procname = "bpf_stats_enabled",
1253 .data = &sysctl_bpf_stats_enabled,
1254 .maxlen = sizeof(sysctl_bpf_stats_enabled),
1256 .proc_handler = proc_dointvec_minmax_bpf_stats,
1261 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1263 .procname = "panic_on_rcu_stall",
1264 .data = &sysctl_panic_on_rcu_stall,
1265 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
1267 .proc_handler = proc_dointvec_minmax,
1272 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
1274 .procname = "stack_erasing",
1276 .maxlen = sizeof(int),
1278 .proc_handler = stack_erasing_sysctl,
1286 static struct ctl_table vm_table[] = {
1288 .procname = "overcommit_memory",
1289 .data = &sysctl_overcommit_memory,
1290 .maxlen = sizeof(sysctl_overcommit_memory),
1292 .proc_handler = proc_dointvec_minmax,
1297 .procname = "panic_on_oom",
1298 .data = &sysctl_panic_on_oom,
1299 .maxlen = sizeof(sysctl_panic_on_oom),
1301 .proc_handler = proc_dointvec_minmax,
1306 .procname = "oom_kill_allocating_task",
1307 .data = &sysctl_oom_kill_allocating_task,
1308 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1310 .proc_handler = proc_dointvec,
1313 .procname = "oom_dump_tasks",
1314 .data = &sysctl_oom_dump_tasks,
1315 .maxlen = sizeof(sysctl_oom_dump_tasks),
1317 .proc_handler = proc_dointvec,
1320 .procname = "overcommit_ratio",
1321 .data = &sysctl_overcommit_ratio,
1322 .maxlen = sizeof(sysctl_overcommit_ratio),
1324 .proc_handler = overcommit_ratio_handler,
1327 .procname = "overcommit_kbytes",
1328 .data = &sysctl_overcommit_kbytes,
1329 .maxlen = sizeof(sysctl_overcommit_kbytes),
1331 .proc_handler = overcommit_kbytes_handler,
1334 .procname = "page-cluster",
1335 .data = &page_cluster,
1336 .maxlen = sizeof(int),
1338 .proc_handler = proc_dointvec_minmax,
1342 .procname = "dirty_background_ratio",
1343 .data = &dirty_background_ratio,
1344 .maxlen = sizeof(dirty_background_ratio),
1346 .proc_handler = dirty_background_ratio_handler,
1348 .extra2 = &one_hundred,
1351 .procname = "dirty_background_bytes",
1352 .data = &dirty_background_bytes,
1353 .maxlen = sizeof(dirty_background_bytes),
1355 .proc_handler = dirty_background_bytes_handler,
1359 .procname = "dirty_ratio",
1360 .data = &vm_dirty_ratio,
1361 .maxlen = sizeof(vm_dirty_ratio),
1363 .proc_handler = dirty_ratio_handler,
1365 .extra2 = &one_hundred,
1368 .procname = "dirty_bytes",
1369 .data = &vm_dirty_bytes,
1370 .maxlen = sizeof(vm_dirty_bytes),
1372 .proc_handler = dirty_bytes_handler,
1373 .extra1 = &dirty_bytes_min,
1376 .procname = "dirty_writeback_centisecs",
1377 .data = &dirty_writeback_interval,
1378 .maxlen = sizeof(dirty_writeback_interval),
1380 .proc_handler = dirty_writeback_centisecs_handler,
1383 .procname = "dirty_expire_centisecs",
1384 .data = &dirty_expire_interval,
1385 .maxlen = sizeof(dirty_expire_interval),
1387 .proc_handler = proc_dointvec_minmax,
1391 .procname = "dirtytime_expire_seconds",
1392 .data = &dirtytime_expire_interval,
1393 .maxlen = sizeof(dirtytime_expire_interval),
1395 .proc_handler = dirtytime_interval_handler,
1399 .procname = "swappiness",
1400 .data = &vm_swappiness,
1401 .maxlen = sizeof(vm_swappiness),
1403 .proc_handler = proc_dointvec_minmax,
1405 .extra2 = &one_hundred,
1407 #ifdef CONFIG_HUGETLB_PAGE
1409 .procname = "nr_hugepages",
1411 .maxlen = sizeof(unsigned long),
1413 .proc_handler = hugetlb_sysctl_handler,
1417 .procname = "nr_hugepages_mempolicy",
1419 .maxlen = sizeof(unsigned long),
1421 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1424 .procname = "numa_stat",
1425 .data = &sysctl_vm_numa_stat,
1426 .maxlen = sizeof(int),
1428 .proc_handler = sysctl_vm_numa_stat_handler,
1434 .procname = "hugetlb_shm_group",
1435 .data = &sysctl_hugetlb_shm_group,
1436 .maxlen = sizeof(gid_t),
1438 .proc_handler = proc_dointvec,
1441 .procname = "nr_overcommit_hugepages",
1443 .maxlen = sizeof(unsigned long),
1445 .proc_handler = hugetlb_overcommit_handler,
1449 .procname = "lowmem_reserve_ratio",
1450 .data = &sysctl_lowmem_reserve_ratio,
1451 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1453 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1456 .procname = "drop_caches",
1457 .data = &sysctl_drop_caches,
1458 .maxlen = sizeof(int),
1460 .proc_handler = drop_caches_sysctl_handler,
1464 #ifdef CONFIG_COMPACTION
1466 .procname = "compact_memory",
1467 .data = &sysctl_compact_memory,
1468 .maxlen = sizeof(int),
1470 .proc_handler = sysctl_compaction_handler,
1473 .procname = "extfrag_threshold",
1474 .data = &sysctl_extfrag_threshold,
1475 .maxlen = sizeof(int),
1477 .proc_handler = proc_dointvec_minmax,
1478 .extra1 = &min_extfrag_threshold,
1479 .extra2 = &max_extfrag_threshold,
1482 .procname = "compact_unevictable_allowed",
1483 .data = &sysctl_compact_unevictable_allowed,
1484 .maxlen = sizeof(int),
1486 .proc_handler = proc_dointvec,
1491 #endif /* CONFIG_COMPACTION */
1493 .procname = "min_free_kbytes",
1494 .data = &min_free_kbytes,
1495 .maxlen = sizeof(min_free_kbytes),
1497 .proc_handler = min_free_kbytes_sysctl_handler,
1501 .procname = "watermark_boost_factor",
1502 .data = &watermark_boost_factor,
1503 .maxlen = sizeof(watermark_boost_factor),
1505 .proc_handler = watermark_boost_factor_sysctl_handler,
1509 .procname = "watermark_scale_factor",
1510 .data = &watermark_scale_factor,
1511 .maxlen = sizeof(watermark_scale_factor),
1513 .proc_handler = watermark_scale_factor_sysctl_handler,
1515 .extra2 = &one_thousand,
1518 .procname = "percpu_pagelist_fraction",
1519 .data = &percpu_pagelist_fraction,
1520 .maxlen = sizeof(percpu_pagelist_fraction),
1522 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1527 .procname = "max_map_count",
1528 .data = &sysctl_max_map_count,
1529 .maxlen = sizeof(sysctl_max_map_count),
1531 .proc_handler = proc_dointvec_minmax,
1536 .procname = "nr_trim_pages",
1537 .data = &sysctl_nr_trim_pages,
1538 .maxlen = sizeof(sysctl_nr_trim_pages),
1540 .proc_handler = proc_dointvec_minmax,
1545 .procname = "laptop_mode",
1546 .data = &laptop_mode,
1547 .maxlen = sizeof(laptop_mode),
1549 .proc_handler = proc_dointvec_jiffies,
1552 .procname = "block_dump",
1553 .data = &block_dump,
1554 .maxlen = sizeof(block_dump),
1556 .proc_handler = proc_dointvec,
1560 .procname = "vfs_cache_pressure",
1561 .data = &sysctl_vfs_cache_pressure,
1562 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1564 .proc_handler = proc_dointvec,
1567 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1569 .procname = "legacy_va_layout",
1570 .data = &sysctl_legacy_va_layout,
1571 .maxlen = sizeof(sysctl_legacy_va_layout),
1573 .proc_handler = proc_dointvec,
1579 .procname = "zone_reclaim_mode",
1580 .data = &node_reclaim_mode,
1581 .maxlen = sizeof(node_reclaim_mode),
1583 .proc_handler = proc_dointvec,
1587 .procname = "min_unmapped_ratio",
1588 .data = &sysctl_min_unmapped_ratio,
1589 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1591 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1593 .extra2 = &one_hundred,
1596 .procname = "min_slab_ratio",
1597 .data = &sysctl_min_slab_ratio,
1598 .maxlen = sizeof(sysctl_min_slab_ratio),
1600 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1602 .extra2 = &one_hundred,
1607 .procname = "stat_interval",
1608 .data = &sysctl_stat_interval,
1609 .maxlen = sizeof(sysctl_stat_interval),
1611 .proc_handler = proc_dointvec_jiffies,
1614 .procname = "stat_refresh",
1618 .proc_handler = vmstat_refresh,
1623 .procname = "mmap_min_addr",
1624 .data = &dac_mmap_min_addr,
1625 .maxlen = sizeof(unsigned long),
1627 .proc_handler = mmap_min_addr_handler,
1632 .procname = "numa_zonelist_order",
1633 .data = &numa_zonelist_order,
1634 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1636 .proc_handler = numa_zonelist_order_handler,
1639 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1640 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1642 .procname = "vdso_enabled",
1643 #ifdef CONFIG_X86_32
1644 .data = &vdso32_enabled,
1645 .maxlen = sizeof(vdso32_enabled),
1647 .data = &vdso_enabled,
1648 .maxlen = sizeof(vdso_enabled),
1651 .proc_handler = proc_dointvec,
1655 #ifdef CONFIG_HIGHMEM
1657 .procname = "highmem_is_dirtyable",
1658 .data = &vm_highmem_is_dirtyable,
1659 .maxlen = sizeof(vm_highmem_is_dirtyable),
1661 .proc_handler = proc_dointvec_minmax,
1666 #ifdef CONFIG_MEMORY_FAILURE
1668 .procname = "memory_failure_early_kill",
1669 .data = &sysctl_memory_failure_early_kill,
1670 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1672 .proc_handler = proc_dointvec_minmax,
1677 .procname = "memory_failure_recovery",
1678 .data = &sysctl_memory_failure_recovery,
1679 .maxlen = sizeof(sysctl_memory_failure_recovery),
1681 .proc_handler = proc_dointvec_minmax,
1687 .procname = "user_reserve_kbytes",
1688 .data = &sysctl_user_reserve_kbytes,
1689 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1691 .proc_handler = proc_doulongvec_minmax,
1694 .procname = "admin_reserve_kbytes",
1695 .data = &sysctl_admin_reserve_kbytes,
1696 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1698 .proc_handler = proc_doulongvec_minmax,
1700 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1702 .procname = "mmap_rnd_bits",
1703 .data = &mmap_rnd_bits,
1704 .maxlen = sizeof(mmap_rnd_bits),
1706 .proc_handler = proc_dointvec_minmax,
1707 .extra1 = (void *)&mmap_rnd_bits_min,
1708 .extra2 = (void *)&mmap_rnd_bits_max,
1711 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1713 .procname = "mmap_rnd_compat_bits",
1714 .data = &mmap_rnd_compat_bits,
1715 .maxlen = sizeof(mmap_rnd_compat_bits),
1717 .proc_handler = proc_dointvec_minmax,
1718 .extra1 = (void *)&mmap_rnd_compat_bits_min,
1719 .extra2 = (void *)&mmap_rnd_compat_bits_max,
1725 static struct ctl_table fs_table[] = {
1727 .procname = "inode-nr",
1728 .data = &inodes_stat,
1729 .maxlen = 2*sizeof(long),
1731 .proc_handler = proc_nr_inodes,
1734 .procname = "inode-state",
1735 .data = &inodes_stat,
1736 .maxlen = 7*sizeof(long),
1738 .proc_handler = proc_nr_inodes,
1741 .procname = "file-nr",
1742 .data = &files_stat,
1743 .maxlen = sizeof(files_stat),
1745 .proc_handler = proc_nr_files,
1748 .procname = "file-max",
1749 .data = &files_stat.max_files,
1750 .maxlen = sizeof(files_stat.max_files),
1752 .proc_handler = proc_doulongvec_minmax,
1754 .extra2 = &long_max,
1757 .procname = "nr_open",
1758 .data = &sysctl_nr_open,
1759 .maxlen = sizeof(unsigned int),
1761 .proc_handler = proc_dointvec_minmax,
1762 .extra1 = &sysctl_nr_open_min,
1763 .extra2 = &sysctl_nr_open_max,
1766 .procname = "dentry-state",
1767 .data = &dentry_stat,
1768 .maxlen = 6*sizeof(long),
1770 .proc_handler = proc_nr_dentry,
1773 .procname = "overflowuid",
1774 .data = &fs_overflowuid,
1775 .maxlen = sizeof(int),
1777 .proc_handler = proc_dointvec_minmax,
1778 .extra1 = &minolduid,
1779 .extra2 = &maxolduid,
1782 .procname = "overflowgid",
1783 .data = &fs_overflowgid,
1784 .maxlen = sizeof(int),
1786 .proc_handler = proc_dointvec_minmax,
1787 .extra1 = &minolduid,
1788 .extra2 = &maxolduid,
1790 #ifdef CONFIG_FILE_LOCKING
1792 .procname = "leases-enable",
1793 .data = &leases_enable,
1794 .maxlen = sizeof(int),
1796 .proc_handler = proc_dointvec,
1799 #ifdef CONFIG_DNOTIFY
1801 .procname = "dir-notify-enable",
1802 .data = &dir_notify_enable,
1803 .maxlen = sizeof(int),
1805 .proc_handler = proc_dointvec,
1809 #ifdef CONFIG_FILE_LOCKING
1811 .procname = "lease-break-time",
1812 .data = &lease_break_time,
1813 .maxlen = sizeof(int),
1815 .proc_handler = proc_dointvec,
1820 .procname = "aio-nr",
1822 .maxlen = sizeof(aio_nr),
1824 .proc_handler = proc_doulongvec_minmax,
1827 .procname = "aio-max-nr",
1828 .data = &aio_max_nr,
1829 .maxlen = sizeof(aio_max_nr),
1831 .proc_handler = proc_doulongvec_minmax,
1833 #endif /* CONFIG_AIO */
1834 #ifdef CONFIG_INOTIFY_USER
1836 .procname = "inotify",
1838 .child = inotify_table,
1843 .procname = "epoll",
1845 .child = epoll_table,
1850 .procname = "protected_symlinks",
1851 .data = &sysctl_protected_symlinks,
1852 .maxlen = sizeof(int),
1854 .proc_handler = proc_dointvec_minmax,
1859 .procname = "protected_hardlinks",
1860 .data = &sysctl_protected_hardlinks,
1861 .maxlen = sizeof(int),
1863 .proc_handler = proc_dointvec_minmax,
1868 .procname = "protected_fifos",
1869 .data = &sysctl_protected_fifos,
1870 .maxlen = sizeof(int),
1872 .proc_handler = proc_dointvec_minmax,
1877 .procname = "protected_regular",
1878 .data = &sysctl_protected_regular,
1879 .maxlen = sizeof(int),
1881 .proc_handler = proc_dointvec_minmax,
1886 .procname = "suid_dumpable",
1887 .data = &suid_dumpable,
1888 .maxlen = sizeof(int),
1890 .proc_handler = proc_dointvec_minmax_coredump,
1894 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1896 .procname = "binfmt_misc",
1898 .child = sysctl_mount_point,
1902 .procname = "pipe-max-size",
1903 .data = &pipe_max_size,
1904 .maxlen = sizeof(pipe_max_size),
1906 .proc_handler = proc_dopipe_max_size,
1909 .procname = "pipe-user-pages-hard",
1910 .data = &pipe_user_pages_hard,
1911 .maxlen = sizeof(pipe_user_pages_hard),
1913 .proc_handler = proc_doulongvec_minmax,
1916 .procname = "pipe-user-pages-soft",
1917 .data = &pipe_user_pages_soft,
1918 .maxlen = sizeof(pipe_user_pages_soft),
1920 .proc_handler = proc_doulongvec_minmax,
1923 .procname = "mount-max",
1924 .data = &sysctl_mount_max,
1925 .maxlen = sizeof(unsigned int),
1927 .proc_handler = proc_dointvec_minmax,
1933 static struct ctl_table debug_table[] = {
1934 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1936 .procname = "exception-trace",
1937 .data = &show_unhandled_signals,
1938 .maxlen = sizeof(int),
1940 .proc_handler = proc_dointvec
1943 #if defined(CONFIG_OPTPROBES)
1945 .procname = "kprobes-optimization",
1946 .data = &sysctl_kprobes_optimization,
1947 .maxlen = sizeof(int),
1949 .proc_handler = proc_kprobes_optimization_handler,
1957 static struct ctl_table dev_table[] = {
1961 int __init sysctl_init(void)
1963 struct ctl_table_header *hdr;
1965 hdr = register_sysctl_table(sysctl_base_table);
1966 kmemleak_not_leak(hdr);
1970 #endif /* CONFIG_SYSCTL */
1976 #ifdef CONFIG_PROC_SYSCTL
1978 static int _proc_do_string(char *data, int maxlen, int write,
1979 char __user *buffer,
1980 size_t *lenp, loff_t *ppos)
1986 if (!data || !maxlen || !*lenp) {
1992 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1993 /* Only continue writes not past the end of buffer. */
1995 if (len > maxlen - 1)
2002 /* Start writing from beginning of buffer. */
2008 while ((p - buffer) < *lenp && len < maxlen - 1) {
2009 if (get_user(c, p++))
2011 if (c == 0 || c == '\n')
2032 if (copy_to_user(buffer, data, len))
2035 if (put_user('\n', buffer + len))
2045 static void warn_sysctl_write(struct ctl_table *table)
2047 pr_warn_once("%s wrote to %s when file position was not 0!\n"
2048 "This will not be supported in the future. To silence this\n"
2049 "warning, set kernel.sysctl_writes_strict = -1\n",
2050 current->comm, table->procname);
2054 * proc_first_pos_non_zero_ignore - check if first position is allowed
2055 * @ppos: file position
2056 * @table: the sysctl table
2058 * Returns true if the first position is non-zero and the sysctl_writes_strict
2059 * mode indicates this is not allowed for numeric input types. String proc
2060 * handlers can ignore the return value.
2062 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
2063 struct ctl_table *table)
2068 switch (sysctl_writes_strict) {
2069 case SYSCTL_WRITES_STRICT:
2071 case SYSCTL_WRITES_WARN:
2072 warn_sysctl_write(table);
2080 * proc_dostring - read a string sysctl
2081 * @table: the sysctl table
2082 * @write: %TRUE if this is a write to the sysctl file
2083 * @buffer: the user buffer
2084 * @lenp: the size of the user buffer
2085 * @ppos: file position
2087 * Reads/writes a string from/to the user buffer. If the kernel
2088 * buffer provided is not large enough to hold the string, the
2089 * string is truncated. The copied string is %NULL-terminated.
2090 * If the string is being read by the user process, it is copied
2091 * and a newline '\n' is added. It is truncated if the buffer is
2094 * Returns 0 on success.
2096 int proc_dostring(struct ctl_table *table, int write,
2097 void __user *buffer, size_t *lenp, loff_t *ppos)
2100 proc_first_pos_non_zero_ignore(ppos, table);
2102 return _proc_do_string((char *)(table->data), table->maxlen, write,
2103 (char __user *)buffer, lenp, ppos);
2106 static size_t proc_skip_spaces(char **buf)
2109 char *tmp = skip_spaces(*buf);
2115 static void proc_skip_char(char **buf, size_t *size, const char v)
2126 * strtoul_lenient - parse an ASCII formatted integer from a buffer and only
2129 * @cp: kernel buffer containing the string to parse
2130 * @endp: pointer to store the trailing characters
2131 * @base: the base to use
2132 * @res: where the parsed integer will be stored
2134 * In case of success 0 is returned and @res will contain the parsed integer,
2135 * @endp will hold any trailing characters.
2136 * This function will fail the parse on overflow. If there wasn't an overflow
2137 * the function will defer the decision what characters count as invalid to the
2140 static int strtoul_lenient(const char *cp, char **endp, unsigned int base,
2143 unsigned long long result;
2146 cp = _parse_integer_fixup_radix(cp, &base);
2147 rv = _parse_integer(cp, base, &result);
2148 if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result))
2156 *res = (unsigned long)result;
2160 #define TMPBUFLEN 22
2162 * proc_get_long - reads an ASCII formatted integer from a user buffer
2164 * @buf: a kernel buffer
2165 * @size: size of the kernel buffer
2166 * @val: this is where the number will be stored
2167 * @neg: set to %TRUE if number is negative
2168 * @perm_tr: a vector which contains the allowed trailers
2169 * @perm_tr_len: size of the perm_tr vector
2170 * @tr: pointer to store the trailer character
2172 * In case of success %0 is returned and @buf and @size are updated with
2173 * the amount of bytes read. If @tr is non-NULL and a trailing
2174 * character exists (size is non-zero after returning from this
2175 * function), @tr is updated with the trailing character.
2177 static int proc_get_long(char **buf, size_t *size,
2178 unsigned long *val, bool *neg,
2179 const char *perm_tr, unsigned perm_tr_len, char *tr)
2182 char *p, tmp[TMPBUFLEN];
2188 if (len > TMPBUFLEN - 1)
2189 len = TMPBUFLEN - 1;
2191 memcpy(tmp, *buf, len);
2195 if (*p == '-' && *size > 1) {
2203 if (strtoul_lenient(p, &p, 0, val))
2208 /* We don't know if the next char is whitespace thus we may accept
2209 * invalid integers (e.g. 1234...a) or two integers instead of one
2210 * (e.g. 123...1). So lets not allow such large numbers. */
2211 if (len == TMPBUFLEN - 1)
2214 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2217 if (tr && (len < *size))
2227 * proc_put_long - converts an integer to a decimal ASCII formatted string
2229 * @buf: the user buffer
2230 * @size: the size of the user buffer
2231 * @val: the integer to be converted
2232 * @neg: sign of the number, %TRUE for negative
2234 * In case of success %0 is returned and @buf and @size are updated with
2235 * the amount of bytes written.
2237 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2241 char tmp[TMPBUFLEN], *p = tmp;
2243 sprintf(p, "%s%lu", neg ? "-" : "", val);
2247 if (copy_to_user(*buf, tmp, len))
2255 static int proc_put_char(void __user **buf, size_t *size, char c)
2258 char __user **buffer = (char __user **)buf;
2259 if (put_user(c, *buffer))
2261 (*size)--, (*buffer)++;
2267 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2269 int write, void *data)
2273 if (*lvalp > (unsigned long) INT_MAX + 1)
2277 if (*lvalp > (unsigned long) INT_MAX)
2285 *lvalp = -(unsigned long)val;
2288 *lvalp = (unsigned long)val;
2294 static int do_proc_douintvec_conv(unsigned long *lvalp,
2296 int write, void *data)
2299 if (*lvalp > UINT_MAX)
2303 unsigned int val = *valp;
2304 *lvalp = (unsigned long)val;
2309 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2311 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2312 int write, void __user *buffer,
2313 size_t *lenp, loff_t *ppos,
2314 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2315 int write, void *data),
2318 int *i, vleft, first = 1, err = 0;
2320 char *kbuf = NULL, *p;
2322 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2327 i = (int *) tbl_data;
2328 vleft = table->maxlen / sizeof(*i);
2332 conv = do_proc_dointvec_conv;
2335 if (proc_first_pos_non_zero_ignore(ppos, table))
2338 if (left > PAGE_SIZE - 1)
2339 left = PAGE_SIZE - 1;
2340 p = kbuf = memdup_user_nul(buffer, left);
2342 return PTR_ERR(kbuf);
2345 for (; left && vleft--; i++, first=0) {
2350 left -= proc_skip_spaces(&p);
2354 err = proc_get_long(&p, &left, &lval, &neg,
2356 sizeof(proc_wspace_sep), NULL);
2359 if (conv(&neg, &lval, i, 1, data)) {
2364 if (conv(&neg, &lval, i, 0, data)) {
2369 err = proc_put_char(&buffer, &left, '\t');
2372 err = proc_put_long(&buffer, &left, lval, neg);
2378 if (!write && !first && left && !err)
2379 err = proc_put_char(&buffer, &left, '\n');
2380 if (write && !err && left)
2381 left -= proc_skip_spaces(&p);
2385 return err ? : -EINVAL;
2393 static int do_proc_dointvec(struct ctl_table *table, int write,
2394 void __user *buffer, size_t *lenp, loff_t *ppos,
2395 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2396 int write, void *data),
2399 return __do_proc_dointvec(table->data, table, write,
2400 buffer, lenp, ppos, conv, data);
2403 static int do_proc_douintvec_w(unsigned int *tbl_data,
2404 struct ctl_table *table,
2405 void __user *buffer,
2406 size_t *lenp, loff_t *ppos,
2407 int (*conv)(unsigned long *lvalp,
2409 int write, void *data),
2416 char *kbuf = NULL, *p;
2420 if (proc_first_pos_non_zero_ignore(ppos, table))
2423 if (left > PAGE_SIZE - 1)
2424 left = PAGE_SIZE - 1;
2426 p = kbuf = memdup_user_nul(buffer, left);
2430 left -= proc_skip_spaces(&p);
2436 err = proc_get_long(&p, &left, &lval, &neg,
2438 sizeof(proc_wspace_sep), NULL);
2444 if (conv(&lval, tbl_data, 1, data)) {
2450 left -= proc_skip_spaces(&p);
2459 /* This is in keeping with old __do_proc_dointvec() */
2465 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2466 size_t *lenp, loff_t *ppos,
2467 int (*conv)(unsigned long *lvalp,
2469 int write, void *data),
2478 if (conv(&lval, tbl_data, 0, data)) {
2483 err = proc_put_long(&buffer, &left, lval, false);
2487 err = proc_put_char(&buffer, &left, '\n');
2496 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2497 int write, void __user *buffer,
2498 size_t *lenp, loff_t *ppos,
2499 int (*conv)(unsigned long *lvalp,
2501 int write, void *data),
2504 unsigned int *i, vleft;
2506 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2511 i = (unsigned int *) tbl_data;
2512 vleft = table->maxlen / sizeof(*i);
2515 * Arrays are not supported, keep this simple. *Do not* add
2524 conv = do_proc_douintvec_conv;
2527 return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2529 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2532 static int do_proc_douintvec(struct ctl_table *table, int write,
2533 void __user *buffer, size_t *lenp, loff_t *ppos,
2534 int (*conv)(unsigned long *lvalp,
2536 int write, void *data),
2539 return __do_proc_douintvec(table->data, table, write,
2540 buffer, lenp, ppos, conv, data);
2544 * proc_dointvec - read a vector of integers
2545 * @table: the sysctl table
2546 * @write: %TRUE if this is a write to the sysctl file
2547 * @buffer: the user buffer
2548 * @lenp: the size of the user buffer
2549 * @ppos: file position
2551 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2552 * values from/to the user buffer, treated as an ASCII string.
2554 * Returns 0 on success.
2556 int proc_dointvec(struct ctl_table *table, int write,
2557 void __user *buffer, size_t *lenp, loff_t *ppos)
2559 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2563 * proc_douintvec - read a vector of unsigned integers
2564 * @table: the sysctl table
2565 * @write: %TRUE if this is a write to the sysctl file
2566 * @buffer: the user buffer
2567 * @lenp: the size of the user buffer
2568 * @ppos: file position
2570 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2571 * values from/to the user buffer, treated as an ASCII string.
2573 * Returns 0 on success.
2575 int proc_douintvec(struct ctl_table *table, int write,
2576 void __user *buffer, size_t *lenp, loff_t *ppos)
2578 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2579 do_proc_douintvec_conv, NULL);
2583 * Taint values can only be increased
2584 * This means we can safely use a temporary.
2586 static int proc_taint(struct ctl_table *table, int write,
2587 void __user *buffer, size_t *lenp, loff_t *ppos)
2590 unsigned long tmptaint = get_taint();
2593 if (write && !capable(CAP_SYS_ADMIN))
2598 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2604 * Poor man's atomic or. Not worth adding a primitive
2605 * to everyone's atomic.h for this
2608 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2609 if ((tmptaint >> i) & 1)
2610 add_taint(i, LOCKDEP_STILL_OK);
2617 #ifdef CONFIG_PRINTK
2618 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2619 void __user *buffer, size_t *lenp, loff_t *ppos)
2621 if (write && !capable(CAP_SYS_ADMIN))
2624 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2629 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
2630 * @min: pointer to minimum allowable value
2631 * @max: pointer to maximum allowable value
2633 * The do_proc_dointvec_minmax_conv_param structure provides the
2634 * minimum and maximum values for doing range checking for those sysctl
2635 * parameters that use the proc_dointvec_minmax() handler.
2637 struct do_proc_dointvec_minmax_conv_param {
2642 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2644 int write, void *data)
2647 struct do_proc_dointvec_minmax_conv_param *param = data;
2649 * If writing, first do so via a temporary local int so we can
2650 * bounds-check it before touching *valp.
2652 int *ip = write ? &tmp : valp;
2654 ret = do_proc_dointvec_conv(negp, lvalp, ip, write, data);
2659 if ((param->min && *param->min > tmp) ||
2660 (param->max && *param->max < tmp))
2669 * proc_dointvec_minmax - read a vector of integers with min/max values
2670 * @table: the sysctl table
2671 * @write: %TRUE if this is a write to the sysctl file
2672 * @buffer: the user buffer
2673 * @lenp: the size of the user buffer
2674 * @ppos: file position
2676 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2677 * values from/to the user buffer, treated as an ASCII string.
2679 * This routine will ensure the values are within the range specified by
2680 * table->extra1 (min) and table->extra2 (max).
2682 * Returns 0 on success or -EINVAL on write when the range check fails.
2684 int proc_dointvec_minmax(struct ctl_table *table, int write,
2685 void __user *buffer, size_t *lenp, loff_t *ppos)
2687 struct do_proc_dointvec_minmax_conv_param param = {
2688 .min = (int *) table->extra1,
2689 .max = (int *) table->extra2,
2691 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2692 do_proc_dointvec_minmax_conv, ¶m);
2696 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
2697 * @min: pointer to minimum allowable value
2698 * @max: pointer to maximum allowable value
2700 * The do_proc_douintvec_minmax_conv_param structure provides the
2701 * minimum and maximum values for doing range checking for those sysctl
2702 * parameters that use the proc_douintvec_minmax() handler.
2704 struct do_proc_douintvec_minmax_conv_param {
2709 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2711 int write, void *data)
2715 struct do_proc_douintvec_minmax_conv_param *param = data;
2716 /* write via temporary local uint for bounds-checking */
2717 unsigned int *up = write ? &tmp : valp;
2719 ret = do_proc_douintvec_conv(lvalp, up, write, data);
2724 if ((param->min && *param->min > tmp) ||
2725 (param->max && *param->max < tmp))
2735 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2736 * @table: the sysctl table
2737 * @write: %TRUE if this is a write to the sysctl file
2738 * @buffer: the user buffer
2739 * @lenp: the size of the user buffer
2740 * @ppos: file position
2742 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2743 * values from/to the user buffer, treated as an ASCII string. Negative
2744 * strings are not allowed.
2746 * This routine will ensure the values are within the range specified by
2747 * table->extra1 (min) and table->extra2 (max). There is a final sanity
2748 * check for UINT_MAX to avoid having to support wrap around uses from
2751 * Returns 0 on success or -ERANGE on write when the range check fails.
2753 int proc_douintvec_minmax(struct ctl_table *table, int write,
2754 void __user *buffer, size_t *lenp, loff_t *ppos)
2756 struct do_proc_douintvec_minmax_conv_param param = {
2757 .min = (unsigned int *) table->extra1,
2758 .max = (unsigned int *) table->extra2,
2760 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2761 do_proc_douintvec_minmax_conv, ¶m);
2764 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
2766 int write, void *data)
2771 val = round_pipe_size(*lvalp);
2777 unsigned int val = *valp;
2778 *lvalp = (unsigned long) val;
2784 static int proc_dopipe_max_size(struct ctl_table *table, int write,
2785 void __user *buffer, size_t *lenp, loff_t *ppos)
2787 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2788 do_proc_dopipe_max_size_conv, NULL);
2791 static void validate_coredump_safety(void)
2793 #ifdef CONFIG_COREDUMP
2794 if (suid_dumpable == SUID_DUMP_ROOT &&
2795 core_pattern[0] != '/' && core_pattern[0] != '|') {
2797 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2798 "Pipe handler or fully qualified core dump path required.\n"
2799 "Set kernel.core_pattern before fs.suid_dumpable.\n"
2805 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2806 void __user *buffer, size_t *lenp, loff_t *ppos)
2808 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2810 validate_coredump_safety();
2814 #ifdef CONFIG_COREDUMP
2815 static int proc_dostring_coredump(struct ctl_table *table, int write,
2816 void __user *buffer, size_t *lenp, loff_t *ppos)
2818 int error = proc_dostring(table, write, buffer, lenp, ppos);
2820 validate_coredump_safety();
2825 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2826 void __user *buffer,
2827 size_t *lenp, loff_t *ppos,
2828 unsigned long convmul,
2829 unsigned long convdiv)
2831 unsigned long *i, *min, *max;
2832 int vleft, first = 1, err = 0;
2834 char *kbuf = NULL, *p;
2836 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2841 i = (unsigned long *) data;
2842 min = (unsigned long *) table->extra1;
2843 max = (unsigned long *) table->extra2;
2844 vleft = table->maxlen / sizeof(unsigned long);
2848 if (proc_first_pos_non_zero_ignore(ppos, table))
2851 if (left > PAGE_SIZE - 1)
2852 left = PAGE_SIZE - 1;
2853 p = kbuf = memdup_user_nul(buffer, left);
2855 return PTR_ERR(kbuf);
2858 for (; left && vleft--; i++, first = 0) {
2864 left -= proc_skip_spaces(&p);
2868 err = proc_get_long(&p, &left, &val, &neg,
2870 sizeof(proc_wspace_sep), NULL);
2875 val = convmul * val / convdiv;
2876 if ((min && val < *min) || (max && val > *max))
2880 val = convdiv * (*i) / convmul;
2882 err = proc_put_char(&buffer, &left, '\t');
2886 err = proc_put_long(&buffer, &left, val, false);
2892 if (!write && !first && left && !err)
2893 err = proc_put_char(&buffer, &left, '\n');
2895 left -= proc_skip_spaces(&p);
2899 return err ? : -EINVAL;
2907 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2908 void __user *buffer,
2909 size_t *lenp, loff_t *ppos,
2910 unsigned long convmul,
2911 unsigned long convdiv)
2913 return __do_proc_doulongvec_minmax(table->data, table, write,
2914 buffer, lenp, ppos, convmul, convdiv);
2918 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2919 * @table: the sysctl table
2920 * @write: %TRUE if this is a write to the sysctl file
2921 * @buffer: the user buffer
2922 * @lenp: the size of the user buffer
2923 * @ppos: file position
2925 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2926 * values from/to the user buffer, treated as an ASCII string.
2928 * This routine will ensure the values are within the range specified by
2929 * table->extra1 (min) and table->extra2 (max).
2931 * Returns 0 on success.
2933 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2934 void __user *buffer, size_t *lenp, loff_t *ppos)
2936 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2940 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2941 * @table: the sysctl table
2942 * @write: %TRUE if this is a write to the sysctl file
2943 * @buffer: the user buffer
2944 * @lenp: the size of the user buffer
2945 * @ppos: file position
2947 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2948 * values from/to the user buffer, treated as an ASCII string. The values
2949 * are treated as milliseconds, and converted to jiffies when they are stored.
2951 * This routine will ensure the values are within the range specified by
2952 * table->extra1 (min) and table->extra2 (max).
2954 * Returns 0 on success.
2956 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2957 void __user *buffer,
2958 size_t *lenp, loff_t *ppos)
2960 return do_proc_doulongvec_minmax(table, write, buffer,
2961 lenp, ppos, HZ, 1000l);
2965 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2967 int write, void *data)
2970 if (*lvalp > INT_MAX / HZ)
2972 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2978 lval = -(unsigned long)val;
2981 lval = (unsigned long)val;
2988 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2990 int write, void *data)
2993 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2995 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
3001 lval = -(unsigned long)val;
3004 lval = (unsigned long)val;
3006 *lvalp = jiffies_to_clock_t(lval);
3011 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
3013 int write, void *data)
3016 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
3026 lval = -(unsigned long)val;
3029 lval = (unsigned long)val;
3031 *lvalp = jiffies_to_msecs(lval);
3037 * proc_dointvec_jiffies - read a vector of integers as seconds
3038 * @table: the sysctl table
3039 * @write: %TRUE if this is a write to the sysctl file
3040 * @buffer: the user buffer
3041 * @lenp: the size of the user buffer
3042 * @ppos: file position
3044 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3045 * values from/to the user buffer, treated as an ASCII string.
3046 * The values read are assumed to be in seconds, and are converted into
3049 * Returns 0 on success.
3051 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3052 void __user *buffer, size_t *lenp, loff_t *ppos)
3054 return do_proc_dointvec(table,write,buffer,lenp,ppos,
3055 do_proc_dointvec_jiffies_conv,NULL);
3059 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
3060 * @table: the sysctl table
3061 * @write: %TRUE if this is a write to the sysctl file
3062 * @buffer: the user buffer
3063 * @lenp: the size of the user buffer
3064 * @ppos: pointer to the file position
3066 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3067 * values from/to the user buffer, treated as an ASCII string.
3068 * The values read are assumed to be in 1/USER_HZ seconds, and
3069 * are converted into jiffies.
3071 * Returns 0 on success.
3073 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3074 void __user *buffer, size_t *lenp, loff_t *ppos)
3076 return do_proc_dointvec(table,write,buffer,lenp,ppos,
3077 do_proc_dointvec_userhz_jiffies_conv,NULL);
3081 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
3082 * @table: the sysctl table
3083 * @write: %TRUE if this is a write to the sysctl file
3084 * @buffer: the user buffer
3085 * @lenp: the size of the user buffer
3086 * @ppos: file position
3087 * @ppos: the current position in the file
3089 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3090 * values from/to the user buffer, treated as an ASCII string.
3091 * The values read are assumed to be in 1/1000 seconds, and
3092 * are converted into jiffies.
3094 * Returns 0 on success.
3096 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3097 void __user *buffer, size_t *lenp, loff_t *ppos)
3099 return do_proc_dointvec(table, write, buffer, lenp, ppos,
3100 do_proc_dointvec_ms_jiffies_conv, NULL);
3103 static int proc_do_cad_pid(struct ctl_table *table, int write,
3104 void __user *buffer, size_t *lenp, loff_t *ppos)
3106 struct pid *new_pid;
3110 tmp = pid_vnr(cad_pid);
3112 r = __do_proc_dointvec(&tmp, table, write, buffer,
3113 lenp, ppos, NULL, NULL);
3117 new_pid = find_get_pid(tmp);
3121 put_pid(xchg(&cad_pid, new_pid));
3126 * proc_do_large_bitmap - read/write from/to a large bitmap
3127 * @table: the sysctl table
3128 * @write: %TRUE if this is a write to the sysctl file
3129 * @buffer: the user buffer
3130 * @lenp: the size of the user buffer
3131 * @ppos: file position
3133 * The bitmap is stored at table->data and the bitmap length (in bits)
3136 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
3137 * large bitmaps may be represented in a compact manner. Writing into
3138 * the file will clear the bitmap then update it with the given input.
3140 * Returns 0 on success.
3142 int proc_do_large_bitmap(struct ctl_table *table, int write,
3143 void __user *buffer, size_t *lenp, loff_t *ppos)
3147 size_t left = *lenp;
3148 unsigned long bitmap_len = table->maxlen;
3149 unsigned long *bitmap = *(unsigned long **) table->data;
3150 unsigned long *tmp_bitmap = NULL;
3151 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
3153 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
3161 if (left > PAGE_SIZE - 1)
3162 left = PAGE_SIZE - 1;
3164 p = kbuf = memdup_user_nul(buffer, left);
3166 return PTR_ERR(kbuf);
3168 tmp_bitmap = kcalloc(BITS_TO_LONGS(bitmap_len),
3169 sizeof(unsigned long),
3175 proc_skip_char(&p, &left, '\n');
3176 while (!err && left) {
3177 unsigned long val_a, val_b;
3180 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3184 if (val_a >= bitmap_len || neg) {
3196 err = proc_get_long(&p, &left, &val_b,
3197 &neg, tr_b, sizeof(tr_b),
3201 if (val_b >= bitmap_len || neg ||
3212 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3214 proc_skip_char(&p, &left, '\n');
3218 unsigned long bit_a, bit_b = 0;
3221 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3222 if (bit_a >= bitmap_len)
3224 bit_b = find_next_zero_bit(bitmap, bitmap_len,
3228 err = proc_put_char(&buffer, &left, ',');
3232 err = proc_put_long(&buffer, &left, bit_a, false);
3235 if (bit_a != bit_b) {
3236 err = proc_put_char(&buffer, &left, '-');
3239 err = proc_put_long(&buffer, &left, bit_b, false);
3247 err = proc_put_char(&buffer, &left, '\n');
3253 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3255 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3265 #else /* CONFIG_PROC_SYSCTL */
3267 int proc_dostring(struct ctl_table *table, int write,
3268 void __user *buffer, size_t *lenp, loff_t *ppos)
3273 int proc_dointvec(struct ctl_table *table, int write,
3274 void __user *buffer, size_t *lenp, loff_t *ppos)
3279 int proc_douintvec(struct ctl_table *table, int write,
3280 void __user *buffer, size_t *lenp, loff_t *ppos)
3285 int proc_dointvec_minmax(struct ctl_table *table, int write,
3286 void __user *buffer, size_t *lenp, loff_t *ppos)
3291 int proc_douintvec_minmax(struct ctl_table *table, int write,
3292 void __user *buffer, size_t *lenp, loff_t *ppos)
3297 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3298 void __user *buffer, size_t *lenp, loff_t *ppos)
3303 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3304 void __user *buffer, size_t *lenp, loff_t *ppos)
3309 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3310 void __user *buffer, size_t *lenp, loff_t *ppos)
3315 int proc_doulongvec_minmax(struct ctl_table *table, int write,
3316 void __user *buffer, size_t *lenp, loff_t *ppos)
3321 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3322 void __user *buffer,
3323 size_t *lenp, loff_t *ppos)
3329 #endif /* CONFIG_PROC_SYSCTL */
3331 #if defined(CONFIG_BPF_SYSCALL) && defined(CONFIG_SYSCTL)
3332 static int proc_dointvec_minmax_bpf_stats(struct ctl_table *table, int write,
3333 void __user *buffer, size_t *lenp,
3336 int ret, bpf_stats = *(int *)table->data;
3337 struct ctl_table tmp = *table;
3339 if (write && !capable(CAP_SYS_ADMIN))
3342 tmp.data = &bpf_stats;
3343 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
3344 if (write && !ret) {
3345 *(int *)table->data = bpf_stats;
3347 static_branch_enable(&bpf_stats_enabled_key);
3349 static_branch_disable(&bpf_stats_enabled_key);
3355 * No sense putting this after each symbol definition, twice,
3356 * exception granted :-)
3358 EXPORT_SYMBOL(proc_dointvec);
3359 EXPORT_SYMBOL(proc_douintvec);
3360 EXPORT_SYMBOL(proc_dointvec_jiffies);
3361 EXPORT_SYMBOL(proc_dointvec_minmax);
3362 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3363 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3364 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3365 EXPORT_SYMBOL(proc_dostring);
3366 EXPORT_SYMBOL(proc_doulongvec_minmax);
3367 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);