1 // SPDX-License-Identifier: GPL-2.0-only
3 * sysctl.c: General linux system control interface
5 * Begun 24 March 1995, Stephen Tweedie
6 * Added /proc support, Dec 1995
7 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
8 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
9 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
10 * Dynamic registration fixes, Stephen Tweedie.
11 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
12 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
14 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
15 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
16 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
18 * The list_for_each() macro wasn't appropriate for the sysctl loop.
19 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
22 #include <linux/module.h>
23 #include <linux/aio.h>
25 #include <linux/swap.h>
26 #include <linux/slab.h>
27 #include <linux/sysctl.h>
28 #include <linux/bitmap.h>
29 #include <linux/signal.h>
30 #include <linux/printk.h>
31 #include <linux/proc_fs.h>
32 #include <linux/security.h>
33 #include <linux/ctype.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/sched/coredump.h>
67 #include <linux/kexec.h>
68 #include <linux/bpf.h>
69 #include <linux/mount.h>
70 #include <linux/userfaultfd_k.h>
72 #include "../lib/kstrtox.h"
74 #include <linux/uaccess.h>
75 #include <asm/processor.h>
79 #include <asm/stacktrace.h>
83 #include <asm/setup.h>
85 #ifdef CONFIG_BSD_PROCESS_ACCT
86 #include <linux/acct.h>
88 #ifdef CONFIG_RT_MUTEXES
89 #include <linux/rtmutex.h>
91 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
92 #include <linux/lockdep.h>
94 #ifdef CONFIG_CHR_DEV_SG
97 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
98 #include <linux/stackleak.h>
100 #ifdef CONFIG_LOCKUP_DETECTOR
101 #include <linux/nmi.h>
104 #if defined(CONFIG_SYSCTL)
106 /* External variables not in a header file. */
107 extern int suid_dumpable;
108 #ifdef CONFIG_COREDUMP
109 extern int core_uses_pid;
110 extern char core_pattern[];
111 extern unsigned int core_pipe_limit;
114 extern int pid_max_min, pid_max_max;
115 extern int percpu_pagelist_fraction;
116 extern int latencytop_enabled;
117 extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
119 extern int sysctl_nr_trim_pages;
122 /* Constants used for minimum and maximum */
123 #ifdef CONFIG_LOCKUP_DETECTOR
124 static int sixty = 60;
127 static int __maybe_unused neg_one = -1;
130 static int __maybe_unused one = 1;
131 static int __maybe_unused two = 2;
132 static int __maybe_unused four = 4;
133 static unsigned long zero_ul;
134 static unsigned long one_ul = 1;
135 static unsigned long long_max = LONG_MAX;
136 static int one_hundred = 100;
137 static int one_thousand = 1000;
139 static int ten_thousand = 10000;
141 #ifdef CONFIG_PERF_EVENTS
142 static int six_hundred_forty_kb = 640 * 1024;
145 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
146 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
148 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
149 static int maxolduid = 65535;
150 static int minolduid;
152 static int ngroups_max = NGROUPS_MAX;
153 static const int cap_last_cap = CAP_LAST_CAP;
156 * This is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs
157 * and hung_task_check_interval_secs
159 #ifdef CONFIG_DETECT_HUNG_TASK
160 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
163 #ifdef CONFIG_INOTIFY_USER
164 #include <linux/inotify.h>
170 extern int pwrsw_enabled;
173 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
174 extern int unaligned_enabled;
178 extern int unaligned_dump_stack;
181 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
182 extern int no_unaligned_warning;
185 #ifdef CONFIG_PROC_SYSCTL
188 * enum sysctl_writes_mode - supported sysctl write modes
190 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
191 * to be written, and multiple writes on the same sysctl file descriptor
192 * will rewrite the sysctl value, regardless of file position. No warning
193 * is issued when the initial position is not 0.
194 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
196 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
197 * file position 0 and the value must be fully contained in the buffer
198 * sent to the write syscall. If dealing with strings respect the file
199 * position, but restrict this to the max length of the buffer, anything
200 * passed the max lenght will be ignored. Multiple writes will append
203 * These write modes control how current file position affects the behavior of
204 * updating sysctl values through the proc interface on each write.
206 enum sysctl_writes_mode {
207 SYSCTL_WRITES_LEGACY = -1,
208 SYSCTL_WRITES_WARN = 0,
209 SYSCTL_WRITES_STRICT = 1,
212 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
214 static int proc_do_cad_pid(struct ctl_table *table, int write,
215 void __user *buffer, size_t *lenp, loff_t *ppos);
216 static int proc_taint(struct ctl_table *table, int write,
217 void __user *buffer, size_t *lenp, loff_t *ppos);
221 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
222 void __user *buffer, size_t *lenp, loff_t *ppos);
225 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
226 void __user *buffer, size_t *lenp, loff_t *ppos);
227 #ifdef CONFIG_COREDUMP
228 static int proc_dostring_coredump(struct ctl_table *table, int write,
229 void __user *buffer, size_t *lenp, loff_t *ppos);
231 static int proc_dopipe_max_size(struct ctl_table *table, int write,
232 void __user *buffer, size_t *lenp, loff_t *ppos);
233 #ifdef CONFIG_BPF_SYSCALL
234 static int proc_dointvec_minmax_bpf_stats(struct ctl_table *table, int write,
235 void __user *buffer, size_t *lenp,
239 #ifdef CONFIG_MAGIC_SYSRQ
240 /* Note: sysrq code uses its own private copy */
241 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
243 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
244 void __user *buffer, size_t *lenp,
249 error = proc_dointvec(table, write, buffer, lenp, ppos);
254 sysrq_toggle_support(__sysrq_enabled);
261 static struct ctl_table kern_table[];
262 static struct ctl_table vm_table[];
263 static struct ctl_table fs_table[];
264 static struct ctl_table debug_table[];
265 static struct ctl_table dev_table[];
266 extern struct ctl_table random_table[];
268 extern struct ctl_table epoll_table[];
271 #ifdef CONFIG_FW_LOADER_USER_HELPER
272 extern struct ctl_table firmware_config_table[];
275 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
276 int sysctl_legacy_va_layout;
279 /* The default sysctl tables: */
281 static struct ctl_table sysctl_base_table[] = {
283 .procname = "kernel",
300 .child = debug_table,
310 #ifdef CONFIG_SCHED_DEBUG
311 static int min_sched_granularity_ns = 100000; /* 100 usecs */
312 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
313 static int min_wakeup_granularity_ns; /* 0 usecs */
314 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
316 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
317 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
318 #endif /* CONFIG_SMP */
319 #endif /* CONFIG_SCHED_DEBUG */
321 #ifdef CONFIG_COMPACTION
322 static int min_extfrag_threshold;
323 static int max_extfrag_threshold = 1000;
326 static struct ctl_table kern_table[] = {
328 .procname = "sched_child_runs_first",
329 .data = &sysctl_sched_child_runs_first,
330 .maxlen = sizeof(unsigned int),
332 .proc_handler = proc_dointvec,
334 #ifdef CONFIG_SCHED_DEBUG
336 .procname = "sched_min_granularity_ns",
337 .data = &sysctl_sched_min_granularity,
338 .maxlen = sizeof(unsigned int),
340 .proc_handler = sched_proc_update_handler,
341 .extra1 = &min_sched_granularity_ns,
342 .extra2 = &max_sched_granularity_ns,
345 .procname = "sched_latency_ns",
346 .data = &sysctl_sched_latency,
347 .maxlen = sizeof(unsigned int),
349 .proc_handler = sched_proc_update_handler,
350 .extra1 = &min_sched_granularity_ns,
351 .extra2 = &max_sched_granularity_ns,
354 .procname = "sched_wakeup_granularity_ns",
355 .data = &sysctl_sched_wakeup_granularity,
356 .maxlen = sizeof(unsigned int),
358 .proc_handler = sched_proc_update_handler,
359 .extra1 = &min_wakeup_granularity_ns,
360 .extra2 = &max_wakeup_granularity_ns,
364 .procname = "sched_tunable_scaling",
365 .data = &sysctl_sched_tunable_scaling,
366 .maxlen = sizeof(enum sched_tunable_scaling),
368 .proc_handler = sched_proc_update_handler,
369 .extra1 = &min_sched_tunable_scaling,
370 .extra2 = &max_sched_tunable_scaling,
373 .procname = "sched_migration_cost_ns",
374 .data = &sysctl_sched_migration_cost,
375 .maxlen = sizeof(unsigned int),
377 .proc_handler = proc_dointvec,
380 .procname = "sched_nr_migrate",
381 .data = &sysctl_sched_nr_migrate,
382 .maxlen = sizeof(unsigned int),
384 .proc_handler = proc_dointvec,
386 #ifdef CONFIG_SCHEDSTATS
388 .procname = "sched_schedstats",
390 .maxlen = sizeof(unsigned int),
392 .proc_handler = sysctl_schedstats,
396 #endif /* CONFIG_SCHEDSTATS */
397 #endif /* CONFIG_SMP */
398 #ifdef CONFIG_NUMA_BALANCING
400 .procname = "numa_balancing_scan_delay_ms",
401 .data = &sysctl_numa_balancing_scan_delay,
402 .maxlen = sizeof(unsigned int),
404 .proc_handler = proc_dointvec,
407 .procname = "numa_balancing_scan_period_min_ms",
408 .data = &sysctl_numa_balancing_scan_period_min,
409 .maxlen = sizeof(unsigned int),
411 .proc_handler = proc_dointvec,
414 .procname = "numa_balancing_scan_period_max_ms",
415 .data = &sysctl_numa_balancing_scan_period_max,
416 .maxlen = sizeof(unsigned int),
418 .proc_handler = proc_dointvec,
421 .procname = "numa_balancing_scan_size_mb",
422 .data = &sysctl_numa_balancing_scan_size,
423 .maxlen = sizeof(unsigned int),
425 .proc_handler = proc_dointvec_minmax,
429 .procname = "numa_balancing",
430 .data = NULL, /* filled in by handler */
431 .maxlen = sizeof(unsigned int),
433 .proc_handler = sysctl_numa_balancing,
437 #endif /* CONFIG_NUMA_BALANCING */
438 #endif /* CONFIG_SCHED_DEBUG */
440 .procname = "sched_rt_period_us",
441 .data = &sysctl_sched_rt_period,
442 .maxlen = sizeof(unsigned int),
444 .proc_handler = sched_rt_handler,
447 .procname = "sched_rt_runtime_us",
448 .data = &sysctl_sched_rt_runtime,
449 .maxlen = sizeof(int),
451 .proc_handler = sched_rt_handler,
454 .procname = "sched_rr_timeslice_ms",
455 .data = &sysctl_sched_rr_timeslice,
456 .maxlen = sizeof(int),
458 .proc_handler = sched_rr_handler,
460 #ifdef CONFIG_SCHED_AUTOGROUP
462 .procname = "sched_autogroup_enabled",
463 .data = &sysctl_sched_autogroup_enabled,
464 .maxlen = sizeof(unsigned int),
466 .proc_handler = proc_dointvec_minmax,
471 #ifdef CONFIG_CFS_BANDWIDTH
473 .procname = "sched_cfs_bandwidth_slice_us",
474 .data = &sysctl_sched_cfs_bandwidth_slice,
475 .maxlen = sizeof(unsigned int),
477 .proc_handler = proc_dointvec_minmax,
481 #if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_CPU_FREQ_GOV_SCHEDUTIL)
483 .procname = "sched_energy_aware",
484 .data = &sysctl_sched_energy_aware,
485 .maxlen = sizeof(unsigned int),
487 .proc_handler = sched_energy_aware_handler,
492 #ifdef CONFIG_PROVE_LOCKING
494 .procname = "prove_locking",
495 .data = &prove_locking,
496 .maxlen = sizeof(int),
498 .proc_handler = proc_dointvec,
501 #ifdef CONFIG_LOCK_STAT
503 .procname = "lock_stat",
505 .maxlen = sizeof(int),
507 .proc_handler = proc_dointvec,
512 .data = &panic_timeout,
513 .maxlen = sizeof(int),
515 .proc_handler = proc_dointvec,
517 #ifdef CONFIG_COREDUMP
519 .procname = "core_uses_pid",
520 .data = &core_uses_pid,
521 .maxlen = sizeof(int),
523 .proc_handler = proc_dointvec,
526 .procname = "core_pattern",
527 .data = core_pattern,
528 .maxlen = CORENAME_MAX_SIZE,
530 .proc_handler = proc_dostring_coredump,
533 .procname = "core_pipe_limit",
534 .data = &core_pipe_limit,
535 .maxlen = sizeof(unsigned int),
537 .proc_handler = proc_dointvec,
540 #ifdef CONFIG_PROC_SYSCTL
542 .procname = "tainted",
543 .maxlen = sizeof(long),
545 .proc_handler = proc_taint,
548 .procname = "sysctl_writes_strict",
549 .data = &sysctl_writes_strict,
550 .maxlen = sizeof(int),
552 .proc_handler = proc_dointvec_minmax,
557 #ifdef CONFIG_LATENCYTOP
559 .procname = "latencytop",
560 .data = &latencytop_enabled,
561 .maxlen = sizeof(int),
563 .proc_handler = sysctl_latencytop,
566 #ifdef CONFIG_BLK_DEV_INITRD
568 .procname = "real-root-dev",
569 .data = &real_root_dev,
570 .maxlen = sizeof(int),
572 .proc_handler = proc_dointvec,
576 .procname = "print-fatal-signals",
577 .data = &print_fatal_signals,
578 .maxlen = sizeof(int),
580 .proc_handler = proc_dointvec,
584 .procname = "reboot-cmd",
585 .data = reboot_command,
588 .proc_handler = proc_dostring,
591 .procname = "stop-a",
592 .data = &stop_a_enabled,
593 .maxlen = sizeof (int),
595 .proc_handler = proc_dointvec,
598 .procname = "scons-poweroff",
599 .data = &scons_pwroff,
600 .maxlen = sizeof (int),
602 .proc_handler = proc_dointvec,
605 #ifdef CONFIG_SPARC64
607 .procname = "tsb-ratio",
608 .data = &sysctl_tsb_ratio,
609 .maxlen = sizeof (int),
611 .proc_handler = proc_dointvec,
616 .procname = "soft-power",
617 .data = &pwrsw_enabled,
618 .maxlen = sizeof (int),
620 .proc_handler = proc_dointvec,
623 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
625 .procname = "unaligned-trap",
626 .data = &unaligned_enabled,
627 .maxlen = sizeof (int),
629 .proc_handler = proc_dointvec,
633 .procname = "ctrl-alt-del",
635 .maxlen = sizeof(int),
637 .proc_handler = proc_dointvec,
639 #ifdef CONFIG_FUNCTION_TRACER
641 .procname = "ftrace_enabled",
642 .data = &ftrace_enabled,
643 .maxlen = sizeof(int),
645 .proc_handler = ftrace_enable_sysctl,
648 #ifdef CONFIG_STACK_TRACER
650 .procname = "stack_tracer_enabled",
651 .data = &stack_tracer_enabled,
652 .maxlen = sizeof(int),
654 .proc_handler = stack_trace_sysctl,
657 #ifdef CONFIG_TRACING
659 .procname = "ftrace_dump_on_oops",
660 .data = &ftrace_dump_on_oops,
661 .maxlen = sizeof(int),
663 .proc_handler = proc_dointvec,
666 .procname = "traceoff_on_warning",
667 .data = &__disable_trace_on_warning,
668 .maxlen = sizeof(__disable_trace_on_warning),
670 .proc_handler = proc_dointvec,
673 .procname = "tracepoint_printk",
674 .data = &tracepoint_printk,
675 .maxlen = sizeof(tracepoint_printk),
677 .proc_handler = tracepoint_printk_sysctl,
680 #ifdef CONFIG_KEXEC_CORE
682 .procname = "kexec_load_disabled",
683 .data = &kexec_load_disabled,
684 .maxlen = sizeof(int),
686 /* only handle a transition from default "0" to "1" */
687 .proc_handler = proc_dointvec_minmax,
692 #ifdef CONFIG_MODULES
694 .procname = "modprobe",
695 .data = &modprobe_path,
696 .maxlen = KMOD_PATH_LEN,
698 .proc_handler = proc_dostring,
701 .procname = "modules_disabled",
702 .data = &modules_disabled,
703 .maxlen = sizeof(int),
705 /* only handle a transition from default "0" to "1" */
706 .proc_handler = proc_dointvec_minmax,
711 #ifdef CONFIG_UEVENT_HELPER
713 .procname = "hotplug",
714 .data = &uevent_helper,
715 .maxlen = UEVENT_HELPER_PATH_LEN,
717 .proc_handler = proc_dostring,
720 #ifdef CONFIG_CHR_DEV_SG
722 .procname = "sg-big-buff",
723 .data = &sg_big_buff,
724 .maxlen = sizeof (int),
726 .proc_handler = proc_dointvec,
729 #ifdef CONFIG_BSD_PROCESS_ACCT
733 .maxlen = 3*sizeof(int),
735 .proc_handler = proc_dointvec,
738 #ifdef CONFIG_MAGIC_SYSRQ
741 .data = &__sysrq_enabled,
742 .maxlen = sizeof (int),
744 .proc_handler = sysrq_sysctl_handler,
747 #ifdef CONFIG_PROC_SYSCTL
749 .procname = "cad_pid",
751 .maxlen = sizeof (int),
753 .proc_handler = proc_do_cad_pid,
757 .procname = "threads-max",
759 .maxlen = sizeof(int),
761 .proc_handler = sysctl_max_threads,
764 .procname = "random",
766 .child = random_table,
769 .procname = "usermodehelper",
771 .child = usermodehelper_table,
773 #ifdef CONFIG_FW_LOADER_USER_HELPER
775 .procname = "firmware_config",
777 .child = firmware_config_table,
781 .procname = "overflowuid",
782 .data = &overflowuid,
783 .maxlen = sizeof(int),
785 .proc_handler = proc_dointvec_minmax,
786 .extra1 = &minolduid,
787 .extra2 = &maxolduid,
790 .procname = "overflowgid",
791 .data = &overflowgid,
792 .maxlen = sizeof(int),
794 .proc_handler = proc_dointvec_minmax,
795 .extra1 = &minolduid,
796 .extra2 = &maxolduid,
799 #ifdef CONFIG_MATHEMU
801 .procname = "ieee_emulation_warnings",
802 .data = &sysctl_ieee_emulation_warnings,
803 .maxlen = sizeof(int),
805 .proc_handler = proc_dointvec,
809 .procname = "userprocess_debug",
810 .data = &show_unhandled_signals,
811 .maxlen = sizeof(int),
813 .proc_handler = proc_dointvec,
817 .procname = "pid_max",
819 .maxlen = sizeof (int),
821 .proc_handler = proc_dointvec_minmax,
822 .extra1 = &pid_max_min,
823 .extra2 = &pid_max_max,
826 .procname = "panic_on_oops",
827 .data = &panic_on_oops,
828 .maxlen = sizeof(int),
830 .proc_handler = proc_dointvec,
833 .procname = "panic_print",
834 .data = &panic_print,
835 .maxlen = sizeof(unsigned long),
837 .proc_handler = proc_doulongvec_minmax,
839 #if defined CONFIG_PRINTK
841 .procname = "printk",
842 .data = &console_loglevel,
843 .maxlen = 4*sizeof(int),
845 .proc_handler = proc_dointvec,
848 .procname = "printk_ratelimit",
849 .data = &printk_ratelimit_state.interval,
850 .maxlen = sizeof(int),
852 .proc_handler = proc_dointvec_jiffies,
855 .procname = "printk_ratelimit_burst",
856 .data = &printk_ratelimit_state.burst,
857 .maxlen = sizeof(int),
859 .proc_handler = proc_dointvec,
862 .procname = "printk_delay",
863 .data = &printk_delay_msec,
864 .maxlen = sizeof(int),
866 .proc_handler = proc_dointvec_minmax,
868 .extra2 = &ten_thousand,
871 .procname = "printk_devkmsg",
872 .data = devkmsg_log_str,
873 .maxlen = DEVKMSG_STR_MAX_SIZE,
875 .proc_handler = devkmsg_sysctl_set_loglvl,
878 .procname = "dmesg_restrict",
879 .data = &dmesg_restrict,
880 .maxlen = sizeof(int),
882 .proc_handler = proc_dointvec_minmax_sysadmin,
887 .procname = "kptr_restrict",
888 .data = &kptr_restrict,
889 .maxlen = sizeof(int),
891 .proc_handler = proc_dointvec_minmax_sysadmin,
897 .procname = "ngroups_max",
898 .data = &ngroups_max,
899 .maxlen = sizeof (int),
901 .proc_handler = proc_dointvec,
904 .procname = "cap_last_cap",
905 .data = (void *)&cap_last_cap,
906 .maxlen = sizeof(int),
908 .proc_handler = proc_dointvec,
910 #if defined(CONFIG_LOCKUP_DETECTOR)
912 .procname = "watchdog",
913 .data = &watchdog_user_enabled,
914 .maxlen = sizeof(int),
916 .proc_handler = proc_watchdog,
921 .procname = "watchdog_thresh",
922 .data = &watchdog_thresh,
923 .maxlen = sizeof(int),
925 .proc_handler = proc_watchdog_thresh,
930 .procname = "nmi_watchdog",
931 .data = &nmi_watchdog_user_enabled,
932 .maxlen = sizeof(int),
933 .mode = NMI_WATCHDOG_SYSCTL_PERM,
934 .proc_handler = proc_nmi_watchdog,
939 .procname = "watchdog_cpumask",
940 .data = &watchdog_cpumask_bits,
943 .proc_handler = proc_watchdog_cpumask,
945 #ifdef CONFIG_SOFTLOCKUP_DETECTOR
947 .procname = "soft_watchdog",
948 .data = &soft_watchdog_user_enabled,
949 .maxlen = sizeof(int),
951 .proc_handler = proc_soft_watchdog,
956 .procname = "softlockup_panic",
957 .data = &softlockup_panic,
958 .maxlen = sizeof(int),
960 .proc_handler = proc_dointvec_minmax,
966 .procname = "softlockup_all_cpu_backtrace",
967 .data = &sysctl_softlockup_all_cpu_backtrace,
968 .maxlen = sizeof(int),
970 .proc_handler = proc_dointvec_minmax,
974 #endif /* CONFIG_SMP */
976 #ifdef CONFIG_HARDLOCKUP_DETECTOR
978 .procname = "hardlockup_panic",
979 .data = &hardlockup_panic,
980 .maxlen = sizeof(int),
982 .proc_handler = proc_dointvec_minmax,
988 .procname = "hardlockup_all_cpu_backtrace",
989 .data = &sysctl_hardlockup_all_cpu_backtrace,
990 .maxlen = sizeof(int),
992 .proc_handler = proc_dointvec_minmax,
996 #endif /* CONFIG_SMP */
1000 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
1002 .procname = "unknown_nmi_panic",
1003 .data = &unknown_nmi_panic,
1004 .maxlen = sizeof (int),
1006 .proc_handler = proc_dointvec,
1009 #if defined(CONFIG_X86)
1011 .procname = "panic_on_unrecovered_nmi",
1012 .data = &panic_on_unrecovered_nmi,
1013 .maxlen = sizeof(int),
1015 .proc_handler = proc_dointvec,
1018 .procname = "panic_on_io_nmi",
1019 .data = &panic_on_io_nmi,
1020 .maxlen = sizeof(int),
1022 .proc_handler = proc_dointvec,
1024 #ifdef CONFIG_DEBUG_STACKOVERFLOW
1026 .procname = "panic_on_stackoverflow",
1027 .data = &sysctl_panic_on_stackoverflow,
1028 .maxlen = sizeof(int),
1030 .proc_handler = proc_dointvec,
1034 .procname = "bootloader_type",
1035 .data = &bootloader_type,
1036 .maxlen = sizeof (int),
1038 .proc_handler = proc_dointvec,
1041 .procname = "bootloader_version",
1042 .data = &bootloader_version,
1043 .maxlen = sizeof (int),
1045 .proc_handler = proc_dointvec,
1048 .procname = "io_delay_type",
1049 .data = &io_delay_type,
1050 .maxlen = sizeof(int),
1052 .proc_handler = proc_dointvec,
1055 #if defined(CONFIG_MMU)
1057 .procname = "randomize_va_space",
1058 .data = &randomize_va_space,
1059 .maxlen = sizeof(int),
1061 .proc_handler = proc_dointvec,
1064 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1066 .procname = "spin_retry",
1067 .data = &spin_retry,
1068 .maxlen = sizeof (int),
1070 .proc_handler = proc_dointvec,
1073 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1075 .procname = "acpi_video_flags",
1076 .data = &acpi_realmode_flags,
1077 .maxlen = sizeof (unsigned long),
1079 .proc_handler = proc_doulongvec_minmax,
1082 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1084 .procname = "ignore-unaligned-usertrap",
1085 .data = &no_unaligned_warning,
1086 .maxlen = sizeof (int),
1088 .proc_handler = proc_dointvec,
1093 .procname = "unaligned-dump-stack",
1094 .data = &unaligned_dump_stack,
1095 .maxlen = sizeof (int),
1097 .proc_handler = proc_dointvec,
1100 #ifdef CONFIG_DETECT_HUNG_TASK
1102 .procname = "hung_task_panic",
1103 .data = &sysctl_hung_task_panic,
1104 .maxlen = sizeof(int),
1106 .proc_handler = proc_dointvec_minmax,
1111 .procname = "hung_task_check_count",
1112 .data = &sysctl_hung_task_check_count,
1113 .maxlen = sizeof(int),
1115 .proc_handler = proc_dointvec_minmax,
1119 .procname = "hung_task_timeout_secs",
1120 .data = &sysctl_hung_task_timeout_secs,
1121 .maxlen = sizeof(unsigned long),
1123 .proc_handler = proc_dohung_task_timeout_secs,
1124 .extra2 = &hung_task_timeout_max,
1127 .procname = "hung_task_check_interval_secs",
1128 .data = &sysctl_hung_task_check_interval_secs,
1129 .maxlen = sizeof(unsigned long),
1131 .proc_handler = proc_dohung_task_timeout_secs,
1132 .extra2 = &hung_task_timeout_max,
1135 .procname = "hung_task_warnings",
1136 .data = &sysctl_hung_task_warnings,
1137 .maxlen = sizeof(int),
1139 .proc_handler = proc_dointvec_minmax,
1143 #ifdef CONFIG_RT_MUTEXES
1145 .procname = "max_lock_depth",
1146 .data = &max_lock_depth,
1147 .maxlen = sizeof(int),
1149 .proc_handler = proc_dointvec,
1153 .procname = "poweroff_cmd",
1154 .data = &poweroff_cmd,
1155 .maxlen = POWEROFF_CMD_PATH_LEN,
1157 .proc_handler = proc_dostring,
1163 .child = key_sysctls,
1166 #ifdef CONFIG_PERF_EVENTS
1168 * User-space scripts rely on the existence of this file
1169 * as a feature check for perf_events being enabled.
1171 * So it's an ABI, do not remove!
1174 .procname = "perf_event_paranoid",
1175 .data = &sysctl_perf_event_paranoid,
1176 .maxlen = sizeof(sysctl_perf_event_paranoid),
1178 .proc_handler = proc_dointvec,
1181 .procname = "perf_event_mlock_kb",
1182 .data = &sysctl_perf_event_mlock,
1183 .maxlen = sizeof(sysctl_perf_event_mlock),
1185 .proc_handler = proc_dointvec,
1188 .procname = "perf_event_max_sample_rate",
1189 .data = &sysctl_perf_event_sample_rate,
1190 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1192 .proc_handler = perf_proc_update_handler,
1196 .procname = "perf_cpu_time_max_percent",
1197 .data = &sysctl_perf_cpu_time_max_percent,
1198 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1200 .proc_handler = perf_cpu_time_max_percent_handler,
1202 .extra2 = &one_hundred,
1205 .procname = "perf_event_max_stack",
1206 .data = &sysctl_perf_event_max_stack,
1207 .maxlen = sizeof(sysctl_perf_event_max_stack),
1209 .proc_handler = perf_event_max_stack_handler,
1211 .extra2 = &six_hundred_forty_kb,
1214 .procname = "perf_event_max_contexts_per_stack",
1215 .data = &sysctl_perf_event_max_contexts_per_stack,
1216 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
1218 .proc_handler = perf_event_max_stack_handler,
1220 .extra2 = &one_thousand,
1224 .procname = "panic_on_warn",
1225 .data = &panic_on_warn,
1226 .maxlen = sizeof(int),
1228 .proc_handler = proc_dointvec_minmax,
1232 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1234 .procname = "timer_migration",
1235 .data = &sysctl_timer_migration,
1236 .maxlen = sizeof(unsigned int),
1238 .proc_handler = timer_migration_handler,
1243 #ifdef CONFIG_BPF_SYSCALL
1245 .procname = "unprivileged_bpf_disabled",
1246 .data = &sysctl_unprivileged_bpf_disabled,
1247 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
1249 /* only handle a transition from default "0" to "1" */
1250 .proc_handler = proc_dointvec_minmax,
1255 .procname = "bpf_stats_enabled",
1256 .data = &sysctl_bpf_stats_enabled,
1257 .maxlen = sizeof(sysctl_bpf_stats_enabled),
1259 .proc_handler = proc_dointvec_minmax_bpf_stats,
1264 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1266 .procname = "panic_on_rcu_stall",
1267 .data = &sysctl_panic_on_rcu_stall,
1268 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
1270 .proc_handler = proc_dointvec_minmax,
1275 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
1277 .procname = "stack_erasing",
1279 .maxlen = sizeof(int),
1281 .proc_handler = stack_erasing_sysctl,
1289 static struct ctl_table vm_table[] = {
1291 .procname = "overcommit_memory",
1292 .data = &sysctl_overcommit_memory,
1293 .maxlen = sizeof(sysctl_overcommit_memory),
1295 .proc_handler = proc_dointvec_minmax,
1300 .procname = "panic_on_oom",
1301 .data = &sysctl_panic_on_oom,
1302 .maxlen = sizeof(sysctl_panic_on_oom),
1304 .proc_handler = proc_dointvec_minmax,
1309 .procname = "oom_kill_allocating_task",
1310 .data = &sysctl_oom_kill_allocating_task,
1311 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1313 .proc_handler = proc_dointvec,
1316 .procname = "oom_dump_tasks",
1317 .data = &sysctl_oom_dump_tasks,
1318 .maxlen = sizeof(sysctl_oom_dump_tasks),
1320 .proc_handler = proc_dointvec,
1323 .procname = "overcommit_ratio",
1324 .data = &sysctl_overcommit_ratio,
1325 .maxlen = sizeof(sysctl_overcommit_ratio),
1327 .proc_handler = overcommit_ratio_handler,
1330 .procname = "overcommit_kbytes",
1331 .data = &sysctl_overcommit_kbytes,
1332 .maxlen = sizeof(sysctl_overcommit_kbytes),
1334 .proc_handler = overcommit_kbytes_handler,
1337 .procname = "page-cluster",
1338 .data = &page_cluster,
1339 .maxlen = sizeof(int),
1341 .proc_handler = proc_dointvec_minmax,
1345 .procname = "dirty_background_ratio",
1346 .data = &dirty_background_ratio,
1347 .maxlen = sizeof(dirty_background_ratio),
1349 .proc_handler = dirty_background_ratio_handler,
1351 .extra2 = &one_hundred,
1354 .procname = "dirty_background_bytes",
1355 .data = &dirty_background_bytes,
1356 .maxlen = sizeof(dirty_background_bytes),
1358 .proc_handler = dirty_background_bytes_handler,
1362 .procname = "dirty_ratio",
1363 .data = &vm_dirty_ratio,
1364 .maxlen = sizeof(vm_dirty_ratio),
1366 .proc_handler = dirty_ratio_handler,
1368 .extra2 = &one_hundred,
1371 .procname = "dirty_bytes",
1372 .data = &vm_dirty_bytes,
1373 .maxlen = sizeof(vm_dirty_bytes),
1375 .proc_handler = dirty_bytes_handler,
1376 .extra1 = &dirty_bytes_min,
1379 .procname = "dirty_writeback_centisecs",
1380 .data = &dirty_writeback_interval,
1381 .maxlen = sizeof(dirty_writeback_interval),
1383 .proc_handler = dirty_writeback_centisecs_handler,
1386 .procname = "dirty_expire_centisecs",
1387 .data = &dirty_expire_interval,
1388 .maxlen = sizeof(dirty_expire_interval),
1390 .proc_handler = proc_dointvec_minmax,
1394 .procname = "dirtytime_expire_seconds",
1395 .data = &dirtytime_expire_interval,
1396 .maxlen = sizeof(dirtytime_expire_interval),
1398 .proc_handler = dirtytime_interval_handler,
1402 .procname = "swappiness",
1403 .data = &vm_swappiness,
1404 .maxlen = sizeof(vm_swappiness),
1406 .proc_handler = proc_dointvec_minmax,
1408 .extra2 = &one_hundred,
1410 #ifdef CONFIG_HUGETLB_PAGE
1412 .procname = "nr_hugepages",
1414 .maxlen = sizeof(unsigned long),
1416 .proc_handler = hugetlb_sysctl_handler,
1420 .procname = "nr_hugepages_mempolicy",
1422 .maxlen = sizeof(unsigned long),
1424 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1427 .procname = "numa_stat",
1428 .data = &sysctl_vm_numa_stat,
1429 .maxlen = sizeof(int),
1431 .proc_handler = sysctl_vm_numa_stat_handler,
1437 .procname = "hugetlb_shm_group",
1438 .data = &sysctl_hugetlb_shm_group,
1439 .maxlen = sizeof(gid_t),
1441 .proc_handler = proc_dointvec,
1444 .procname = "nr_overcommit_hugepages",
1446 .maxlen = sizeof(unsigned long),
1448 .proc_handler = hugetlb_overcommit_handler,
1452 .procname = "lowmem_reserve_ratio",
1453 .data = &sysctl_lowmem_reserve_ratio,
1454 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1456 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1459 .procname = "drop_caches",
1460 .data = &sysctl_drop_caches,
1461 .maxlen = sizeof(int),
1463 .proc_handler = drop_caches_sysctl_handler,
1467 #ifdef CONFIG_COMPACTION
1469 .procname = "compact_memory",
1470 .data = &sysctl_compact_memory,
1471 .maxlen = sizeof(int),
1473 .proc_handler = sysctl_compaction_handler,
1476 .procname = "extfrag_threshold",
1477 .data = &sysctl_extfrag_threshold,
1478 .maxlen = sizeof(int),
1480 .proc_handler = proc_dointvec_minmax,
1481 .extra1 = &min_extfrag_threshold,
1482 .extra2 = &max_extfrag_threshold,
1485 .procname = "compact_unevictable_allowed",
1486 .data = &sysctl_compact_unevictable_allowed,
1487 .maxlen = sizeof(int),
1489 .proc_handler = proc_dointvec,
1494 #endif /* CONFIG_COMPACTION */
1496 .procname = "min_free_kbytes",
1497 .data = &min_free_kbytes,
1498 .maxlen = sizeof(min_free_kbytes),
1500 .proc_handler = min_free_kbytes_sysctl_handler,
1504 .procname = "watermark_boost_factor",
1505 .data = &watermark_boost_factor,
1506 .maxlen = sizeof(watermark_boost_factor),
1508 .proc_handler = watermark_boost_factor_sysctl_handler,
1512 .procname = "watermark_scale_factor",
1513 .data = &watermark_scale_factor,
1514 .maxlen = sizeof(watermark_scale_factor),
1516 .proc_handler = watermark_scale_factor_sysctl_handler,
1518 .extra2 = &one_thousand,
1521 .procname = "percpu_pagelist_fraction",
1522 .data = &percpu_pagelist_fraction,
1523 .maxlen = sizeof(percpu_pagelist_fraction),
1525 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1530 .procname = "max_map_count",
1531 .data = &sysctl_max_map_count,
1532 .maxlen = sizeof(sysctl_max_map_count),
1534 .proc_handler = proc_dointvec_minmax,
1539 .procname = "nr_trim_pages",
1540 .data = &sysctl_nr_trim_pages,
1541 .maxlen = sizeof(sysctl_nr_trim_pages),
1543 .proc_handler = proc_dointvec_minmax,
1548 .procname = "laptop_mode",
1549 .data = &laptop_mode,
1550 .maxlen = sizeof(laptop_mode),
1552 .proc_handler = proc_dointvec_jiffies,
1555 .procname = "block_dump",
1556 .data = &block_dump,
1557 .maxlen = sizeof(block_dump),
1559 .proc_handler = proc_dointvec,
1563 .procname = "vfs_cache_pressure",
1564 .data = &sysctl_vfs_cache_pressure,
1565 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1567 .proc_handler = proc_dointvec,
1570 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1572 .procname = "legacy_va_layout",
1573 .data = &sysctl_legacy_va_layout,
1574 .maxlen = sizeof(sysctl_legacy_va_layout),
1576 .proc_handler = proc_dointvec,
1582 .procname = "zone_reclaim_mode",
1583 .data = &node_reclaim_mode,
1584 .maxlen = sizeof(node_reclaim_mode),
1586 .proc_handler = proc_dointvec,
1590 .procname = "min_unmapped_ratio",
1591 .data = &sysctl_min_unmapped_ratio,
1592 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1594 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1596 .extra2 = &one_hundred,
1599 .procname = "min_slab_ratio",
1600 .data = &sysctl_min_slab_ratio,
1601 .maxlen = sizeof(sysctl_min_slab_ratio),
1603 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1605 .extra2 = &one_hundred,
1610 .procname = "stat_interval",
1611 .data = &sysctl_stat_interval,
1612 .maxlen = sizeof(sysctl_stat_interval),
1614 .proc_handler = proc_dointvec_jiffies,
1617 .procname = "stat_refresh",
1621 .proc_handler = vmstat_refresh,
1626 .procname = "mmap_min_addr",
1627 .data = &dac_mmap_min_addr,
1628 .maxlen = sizeof(unsigned long),
1630 .proc_handler = mmap_min_addr_handler,
1635 .procname = "numa_zonelist_order",
1636 .data = &numa_zonelist_order,
1637 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1639 .proc_handler = numa_zonelist_order_handler,
1642 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1643 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1645 .procname = "vdso_enabled",
1646 #ifdef CONFIG_X86_32
1647 .data = &vdso32_enabled,
1648 .maxlen = sizeof(vdso32_enabled),
1650 .data = &vdso_enabled,
1651 .maxlen = sizeof(vdso_enabled),
1654 .proc_handler = proc_dointvec,
1658 #ifdef CONFIG_HIGHMEM
1660 .procname = "highmem_is_dirtyable",
1661 .data = &vm_highmem_is_dirtyable,
1662 .maxlen = sizeof(vm_highmem_is_dirtyable),
1664 .proc_handler = proc_dointvec_minmax,
1669 #ifdef CONFIG_MEMORY_FAILURE
1671 .procname = "memory_failure_early_kill",
1672 .data = &sysctl_memory_failure_early_kill,
1673 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1675 .proc_handler = proc_dointvec_minmax,
1680 .procname = "memory_failure_recovery",
1681 .data = &sysctl_memory_failure_recovery,
1682 .maxlen = sizeof(sysctl_memory_failure_recovery),
1684 .proc_handler = proc_dointvec_minmax,
1690 .procname = "user_reserve_kbytes",
1691 .data = &sysctl_user_reserve_kbytes,
1692 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1694 .proc_handler = proc_doulongvec_minmax,
1697 .procname = "admin_reserve_kbytes",
1698 .data = &sysctl_admin_reserve_kbytes,
1699 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1701 .proc_handler = proc_doulongvec_minmax,
1703 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1705 .procname = "mmap_rnd_bits",
1706 .data = &mmap_rnd_bits,
1707 .maxlen = sizeof(mmap_rnd_bits),
1709 .proc_handler = proc_dointvec_minmax,
1710 .extra1 = (void *)&mmap_rnd_bits_min,
1711 .extra2 = (void *)&mmap_rnd_bits_max,
1714 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1716 .procname = "mmap_rnd_compat_bits",
1717 .data = &mmap_rnd_compat_bits,
1718 .maxlen = sizeof(mmap_rnd_compat_bits),
1720 .proc_handler = proc_dointvec_minmax,
1721 .extra1 = (void *)&mmap_rnd_compat_bits_min,
1722 .extra2 = (void *)&mmap_rnd_compat_bits_max,
1725 #ifdef CONFIG_USERFAULTFD
1727 .procname = "unprivileged_userfaultfd",
1728 .data = &sysctl_unprivileged_userfaultfd,
1729 .maxlen = sizeof(sysctl_unprivileged_userfaultfd),
1731 .proc_handler = proc_dointvec_minmax,
1739 static struct ctl_table fs_table[] = {
1741 .procname = "inode-nr",
1742 .data = &inodes_stat,
1743 .maxlen = 2*sizeof(long),
1745 .proc_handler = proc_nr_inodes,
1748 .procname = "inode-state",
1749 .data = &inodes_stat,
1750 .maxlen = 7*sizeof(long),
1752 .proc_handler = proc_nr_inodes,
1755 .procname = "file-nr",
1756 .data = &files_stat,
1757 .maxlen = sizeof(files_stat),
1759 .proc_handler = proc_nr_files,
1762 .procname = "file-max",
1763 .data = &files_stat.max_files,
1764 .maxlen = sizeof(files_stat.max_files),
1766 .proc_handler = proc_doulongvec_minmax,
1768 .extra2 = &long_max,
1771 .procname = "nr_open",
1772 .data = &sysctl_nr_open,
1773 .maxlen = sizeof(unsigned int),
1775 .proc_handler = proc_dointvec_minmax,
1776 .extra1 = &sysctl_nr_open_min,
1777 .extra2 = &sysctl_nr_open_max,
1780 .procname = "dentry-state",
1781 .data = &dentry_stat,
1782 .maxlen = 6*sizeof(long),
1784 .proc_handler = proc_nr_dentry,
1787 .procname = "overflowuid",
1788 .data = &fs_overflowuid,
1789 .maxlen = sizeof(int),
1791 .proc_handler = proc_dointvec_minmax,
1792 .extra1 = &minolduid,
1793 .extra2 = &maxolduid,
1796 .procname = "overflowgid",
1797 .data = &fs_overflowgid,
1798 .maxlen = sizeof(int),
1800 .proc_handler = proc_dointvec_minmax,
1801 .extra1 = &minolduid,
1802 .extra2 = &maxolduid,
1804 #ifdef CONFIG_FILE_LOCKING
1806 .procname = "leases-enable",
1807 .data = &leases_enable,
1808 .maxlen = sizeof(int),
1810 .proc_handler = proc_dointvec,
1813 #ifdef CONFIG_DNOTIFY
1815 .procname = "dir-notify-enable",
1816 .data = &dir_notify_enable,
1817 .maxlen = sizeof(int),
1819 .proc_handler = proc_dointvec,
1823 #ifdef CONFIG_FILE_LOCKING
1825 .procname = "lease-break-time",
1826 .data = &lease_break_time,
1827 .maxlen = sizeof(int),
1829 .proc_handler = proc_dointvec,
1834 .procname = "aio-nr",
1836 .maxlen = sizeof(aio_nr),
1838 .proc_handler = proc_doulongvec_minmax,
1841 .procname = "aio-max-nr",
1842 .data = &aio_max_nr,
1843 .maxlen = sizeof(aio_max_nr),
1845 .proc_handler = proc_doulongvec_minmax,
1847 #endif /* CONFIG_AIO */
1848 #ifdef CONFIG_INOTIFY_USER
1850 .procname = "inotify",
1852 .child = inotify_table,
1857 .procname = "epoll",
1859 .child = epoll_table,
1864 .procname = "protected_symlinks",
1865 .data = &sysctl_protected_symlinks,
1866 .maxlen = sizeof(int),
1868 .proc_handler = proc_dointvec_minmax,
1873 .procname = "protected_hardlinks",
1874 .data = &sysctl_protected_hardlinks,
1875 .maxlen = sizeof(int),
1877 .proc_handler = proc_dointvec_minmax,
1882 .procname = "protected_fifos",
1883 .data = &sysctl_protected_fifos,
1884 .maxlen = sizeof(int),
1886 .proc_handler = proc_dointvec_minmax,
1891 .procname = "protected_regular",
1892 .data = &sysctl_protected_regular,
1893 .maxlen = sizeof(int),
1895 .proc_handler = proc_dointvec_minmax,
1900 .procname = "suid_dumpable",
1901 .data = &suid_dumpable,
1902 .maxlen = sizeof(int),
1904 .proc_handler = proc_dointvec_minmax_coredump,
1908 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1910 .procname = "binfmt_misc",
1912 .child = sysctl_mount_point,
1916 .procname = "pipe-max-size",
1917 .data = &pipe_max_size,
1918 .maxlen = sizeof(pipe_max_size),
1920 .proc_handler = proc_dopipe_max_size,
1923 .procname = "pipe-user-pages-hard",
1924 .data = &pipe_user_pages_hard,
1925 .maxlen = sizeof(pipe_user_pages_hard),
1927 .proc_handler = proc_doulongvec_minmax,
1930 .procname = "pipe-user-pages-soft",
1931 .data = &pipe_user_pages_soft,
1932 .maxlen = sizeof(pipe_user_pages_soft),
1934 .proc_handler = proc_doulongvec_minmax,
1937 .procname = "mount-max",
1938 .data = &sysctl_mount_max,
1939 .maxlen = sizeof(unsigned int),
1941 .proc_handler = proc_dointvec_minmax,
1947 static struct ctl_table debug_table[] = {
1948 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1950 .procname = "exception-trace",
1951 .data = &show_unhandled_signals,
1952 .maxlen = sizeof(int),
1954 .proc_handler = proc_dointvec
1957 #if defined(CONFIG_OPTPROBES)
1959 .procname = "kprobes-optimization",
1960 .data = &sysctl_kprobes_optimization,
1961 .maxlen = sizeof(int),
1963 .proc_handler = proc_kprobes_optimization_handler,
1971 static struct ctl_table dev_table[] = {
1975 int __init sysctl_init(void)
1977 struct ctl_table_header *hdr;
1979 hdr = register_sysctl_table(sysctl_base_table);
1980 kmemleak_not_leak(hdr);
1984 #endif /* CONFIG_SYSCTL */
1990 #ifdef CONFIG_PROC_SYSCTL
1992 static int _proc_do_string(char *data, int maxlen, int write,
1993 char __user *buffer,
1994 size_t *lenp, loff_t *ppos)
2000 if (!data || !maxlen || !*lenp) {
2006 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
2007 /* Only continue writes not past the end of buffer. */
2009 if (len > maxlen - 1)
2016 /* Start writing from beginning of buffer. */
2022 while ((p - buffer) < *lenp && len < maxlen - 1) {
2023 if (get_user(c, p++))
2025 if (c == 0 || c == '\n')
2046 if (copy_to_user(buffer, data, len))
2049 if (put_user('\n', buffer + len))
2059 static void warn_sysctl_write(struct ctl_table *table)
2061 pr_warn_once("%s wrote to %s when file position was not 0!\n"
2062 "This will not be supported in the future. To silence this\n"
2063 "warning, set kernel.sysctl_writes_strict = -1\n",
2064 current->comm, table->procname);
2068 * proc_first_pos_non_zero_ignore - check if first position is allowed
2069 * @ppos: file position
2070 * @table: the sysctl table
2072 * Returns true if the first position is non-zero and the sysctl_writes_strict
2073 * mode indicates this is not allowed for numeric input types. String proc
2074 * handlers can ignore the return value.
2076 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
2077 struct ctl_table *table)
2082 switch (sysctl_writes_strict) {
2083 case SYSCTL_WRITES_STRICT:
2085 case SYSCTL_WRITES_WARN:
2086 warn_sysctl_write(table);
2094 * proc_dostring - read a string sysctl
2095 * @table: the sysctl table
2096 * @write: %TRUE if this is a write to the sysctl file
2097 * @buffer: the user buffer
2098 * @lenp: the size of the user buffer
2099 * @ppos: file position
2101 * Reads/writes a string from/to the user buffer. If the kernel
2102 * buffer provided is not large enough to hold the string, the
2103 * string is truncated. The copied string is %NULL-terminated.
2104 * If the string is being read by the user process, it is copied
2105 * and a newline '\n' is added. It is truncated if the buffer is
2108 * Returns 0 on success.
2110 int proc_dostring(struct ctl_table *table, int write,
2111 void __user *buffer, size_t *lenp, loff_t *ppos)
2114 proc_first_pos_non_zero_ignore(ppos, table);
2116 return _proc_do_string((char *)(table->data), table->maxlen, write,
2117 (char __user *)buffer, lenp, ppos);
2120 static size_t proc_skip_spaces(char **buf)
2123 char *tmp = skip_spaces(*buf);
2129 static void proc_skip_char(char **buf, size_t *size, const char v)
2140 * strtoul_lenient - parse an ASCII formatted integer from a buffer and only
2143 * @cp: kernel buffer containing the string to parse
2144 * @endp: pointer to store the trailing characters
2145 * @base: the base to use
2146 * @res: where the parsed integer will be stored
2148 * In case of success 0 is returned and @res will contain the parsed integer,
2149 * @endp will hold any trailing characters.
2150 * This function will fail the parse on overflow. If there wasn't an overflow
2151 * the function will defer the decision what characters count as invalid to the
2154 static int strtoul_lenient(const char *cp, char **endp, unsigned int base,
2157 unsigned long long result;
2160 cp = _parse_integer_fixup_radix(cp, &base);
2161 rv = _parse_integer(cp, base, &result);
2162 if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result))
2170 *res = (unsigned long)result;
2174 #define TMPBUFLEN 22
2176 * proc_get_long - reads an ASCII formatted integer from a user buffer
2178 * @buf: a kernel buffer
2179 * @size: size of the kernel buffer
2180 * @val: this is where the number will be stored
2181 * @neg: set to %TRUE if number is negative
2182 * @perm_tr: a vector which contains the allowed trailers
2183 * @perm_tr_len: size of the perm_tr vector
2184 * @tr: pointer to store the trailer character
2186 * In case of success %0 is returned and @buf and @size are updated with
2187 * the amount of bytes read. If @tr is non-NULL and a trailing
2188 * character exists (size is non-zero after returning from this
2189 * function), @tr is updated with the trailing character.
2191 static int proc_get_long(char **buf, size_t *size,
2192 unsigned long *val, bool *neg,
2193 const char *perm_tr, unsigned perm_tr_len, char *tr)
2196 char *p, tmp[TMPBUFLEN];
2202 if (len > TMPBUFLEN - 1)
2203 len = TMPBUFLEN - 1;
2205 memcpy(tmp, *buf, len);
2209 if (*p == '-' && *size > 1) {
2217 if (strtoul_lenient(p, &p, 0, val))
2222 /* We don't know if the next char is whitespace thus we may accept
2223 * invalid integers (e.g. 1234...a) or two integers instead of one
2224 * (e.g. 123...1). So lets not allow such large numbers. */
2225 if (len == TMPBUFLEN - 1)
2228 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2231 if (tr && (len < *size))
2241 * proc_put_long - converts an integer to a decimal ASCII formatted string
2243 * @buf: the user buffer
2244 * @size: the size of the user buffer
2245 * @val: the integer to be converted
2246 * @neg: sign of the number, %TRUE for negative
2248 * In case of success %0 is returned and @buf and @size are updated with
2249 * the amount of bytes written.
2251 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2255 char tmp[TMPBUFLEN], *p = tmp;
2257 sprintf(p, "%s%lu", neg ? "-" : "", val);
2261 if (copy_to_user(*buf, tmp, len))
2269 static int proc_put_char(void __user **buf, size_t *size, char c)
2272 char __user **buffer = (char __user **)buf;
2273 if (put_user(c, *buffer))
2275 (*size)--, (*buffer)++;
2281 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2283 int write, void *data)
2287 if (*lvalp > (unsigned long) INT_MAX + 1)
2291 if (*lvalp > (unsigned long) INT_MAX)
2299 *lvalp = -(unsigned long)val;
2302 *lvalp = (unsigned long)val;
2308 static int do_proc_douintvec_conv(unsigned long *lvalp,
2310 int write, void *data)
2313 if (*lvalp > UINT_MAX)
2317 unsigned int val = *valp;
2318 *lvalp = (unsigned long)val;
2323 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2325 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2326 int write, void __user *buffer,
2327 size_t *lenp, loff_t *ppos,
2328 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2329 int write, void *data),
2332 int *i, vleft, first = 1, err = 0;
2334 char *kbuf = NULL, *p;
2336 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2341 i = (int *) tbl_data;
2342 vleft = table->maxlen / sizeof(*i);
2346 conv = do_proc_dointvec_conv;
2349 if (proc_first_pos_non_zero_ignore(ppos, table))
2352 if (left > PAGE_SIZE - 1)
2353 left = PAGE_SIZE - 1;
2354 p = kbuf = memdup_user_nul(buffer, left);
2356 return PTR_ERR(kbuf);
2359 for (; left && vleft--; i++, first=0) {
2364 left -= proc_skip_spaces(&p);
2368 err = proc_get_long(&p, &left, &lval, &neg,
2370 sizeof(proc_wspace_sep), NULL);
2373 if (conv(&neg, &lval, i, 1, data)) {
2378 if (conv(&neg, &lval, i, 0, data)) {
2383 err = proc_put_char(&buffer, &left, '\t');
2386 err = proc_put_long(&buffer, &left, lval, neg);
2392 if (!write && !first && left && !err)
2393 err = proc_put_char(&buffer, &left, '\n');
2394 if (write && !err && left)
2395 left -= proc_skip_spaces(&p);
2399 return err ? : -EINVAL;
2407 static int do_proc_dointvec(struct ctl_table *table, int write,
2408 void __user *buffer, size_t *lenp, loff_t *ppos,
2409 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2410 int write, void *data),
2413 return __do_proc_dointvec(table->data, table, write,
2414 buffer, lenp, ppos, conv, data);
2417 static int do_proc_douintvec_w(unsigned int *tbl_data,
2418 struct ctl_table *table,
2419 void __user *buffer,
2420 size_t *lenp, loff_t *ppos,
2421 int (*conv)(unsigned long *lvalp,
2423 int write, void *data),
2430 char *kbuf = NULL, *p;
2434 if (proc_first_pos_non_zero_ignore(ppos, table))
2437 if (left > PAGE_SIZE - 1)
2438 left = PAGE_SIZE - 1;
2440 p = kbuf = memdup_user_nul(buffer, left);
2444 left -= proc_skip_spaces(&p);
2450 err = proc_get_long(&p, &left, &lval, &neg,
2452 sizeof(proc_wspace_sep), NULL);
2458 if (conv(&lval, tbl_data, 1, data)) {
2464 left -= proc_skip_spaces(&p);
2473 /* This is in keeping with old __do_proc_dointvec() */
2479 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2480 size_t *lenp, loff_t *ppos,
2481 int (*conv)(unsigned long *lvalp,
2483 int write, void *data),
2492 if (conv(&lval, tbl_data, 0, data)) {
2497 err = proc_put_long(&buffer, &left, lval, false);
2501 err = proc_put_char(&buffer, &left, '\n');
2510 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2511 int write, void __user *buffer,
2512 size_t *lenp, loff_t *ppos,
2513 int (*conv)(unsigned long *lvalp,
2515 int write, void *data),
2518 unsigned int *i, vleft;
2520 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2525 i = (unsigned int *) tbl_data;
2526 vleft = table->maxlen / sizeof(*i);
2529 * Arrays are not supported, keep this simple. *Do not* add
2538 conv = do_proc_douintvec_conv;
2541 return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2543 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2546 static int do_proc_douintvec(struct ctl_table *table, int write,
2547 void __user *buffer, size_t *lenp, loff_t *ppos,
2548 int (*conv)(unsigned long *lvalp,
2550 int write, void *data),
2553 return __do_proc_douintvec(table->data, table, write,
2554 buffer, lenp, ppos, conv, data);
2558 * proc_dointvec - read a vector of integers
2559 * @table: the sysctl table
2560 * @write: %TRUE if this is a write to the sysctl file
2561 * @buffer: the user buffer
2562 * @lenp: the size of the user buffer
2563 * @ppos: file position
2565 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2566 * values from/to the user buffer, treated as an ASCII string.
2568 * Returns 0 on success.
2570 int proc_dointvec(struct ctl_table *table, int write,
2571 void __user *buffer, size_t *lenp, loff_t *ppos)
2573 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2577 * proc_douintvec - read a vector of unsigned integers
2578 * @table: the sysctl table
2579 * @write: %TRUE if this is a write to the sysctl file
2580 * @buffer: the user buffer
2581 * @lenp: the size of the user buffer
2582 * @ppos: file position
2584 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2585 * values from/to the user buffer, treated as an ASCII string.
2587 * Returns 0 on success.
2589 int proc_douintvec(struct ctl_table *table, int write,
2590 void __user *buffer, size_t *lenp, loff_t *ppos)
2592 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2593 do_proc_douintvec_conv, NULL);
2597 * Taint values can only be increased
2598 * This means we can safely use a temporary.
2600 static int proc_taint(struct ctl_table *table, int write,
2601 void __user *buffer, size_t *lenp, loff_t *ppos)
2604 unsigned long tmptaint = get_taint();
2607 if (write && !capable(CAP_SYS_ADMIN))
2612 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2618 * Poor man's atomic or. Not worth adding a primitive
2619 * to everyone's atomic.h for this
2622 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2623 if ((tmptaint >> i) & 1)
2624 add_taint(i, LOCKDEP_STILL_OK);
2631 #ifdef CONFIG_PRINTK
2632 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2633 void __user *buffer, size_t *lenp, loff_t *ppos)
2635 if (write && !capable(CAP_SYS_ADMIN))
2638 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2643 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
2644 * @min: pointer to minimum allowable value
2645 * @max: pointer to maximum allowable value
2647 * The do_proc_dointvec_minmax_conv_param structure provides the
2648 * minimum and maximum values for doing range checking for those sysctl
2649 * parameters that use the proc_dointvec_minmax() handler.
2651 struct do_proc_dointvec_minmax_conv_param {
2656 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2658 int write, void *data)
2661 struct do_proc_dointvec_minmax_conv_param *param = data;
2663 * If writing, first do so via a temporary local int so we can
2664 * bounds-check it before touching *valp.
2666 int *ip = write ? &tmp : valp;
2668 ret = do_proc_dointvec_conv(negp, lvalp, ip, write, data);
2673 if ((param->min && *param->min > tmp) ||
2674 (param->max && *param->max < tmp))
2683 * proc_dointvec_minmax - read a vector of integers with min/max values
2684 * @table: the sysctl table
2685 * @write: %TRUE if this is a write to the sysctl file
2686 * @buffer: the user buffer
2687 * @lenp: the size of the user buffer
2688 * @ppos: file position
2690 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2691 * values from/to the user buffer, treated as an ASCII string.
2693 * This routine will ensure the values are within the range specified by
2694 * table->extra1 (min) and table->extra2 (max).
2696 * Returns 0 on success or -EINVAL on write when the range check fails.
2698 int proc_dointvec_minmax(struct ctl_table *table, int write,
2699 void __user *buffer, size_t *lenp, loff_t *ppos)
2701 struct do_proc_dointvec_minmax_conv_param param = {
2702 .min = (int *) table->extra1,
2703 .max = (int *) table->extra2,
2705 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2706 do_proc_dointvec_minmax_conv, ¶m);
2710 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
2711 * @min: pointer to minimum allowable value
2712 * @max: pointer to maximum allowable value
2714 * The do_proc_douintvec_minmax_conv_param structure provides the
2715 * minimum and maximum values for doing range checking for those sysctl
2716 * parameters that use the proc_douintvec_minmax() handler.
2718 struct do_proc_douintvec_minmax_conv_param {
2723 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2725 int write, void *data)
2729 struct do_proc_douintvec_minmax_conv_param *param = data;
2730 /* write via temporary local uint for bounds-checking */
2731 unsigned int *up = write ? &tmp : valp;
2733 ret = do_proc_douintvec_conv(lvalp, up, write, data);
2738 if ((param->min && *param->min > tmp) ||
2739 (param->max && *param->max < tmp))
2749 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2750 * @table: the sysctl table
2751 * @write: %TRUE if this is a write to the sysctl file
2752 * @buffer: the user buffer
2753 * @lenp: the size of the user buffer
2754 * @ppos: file position
2756 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2757 * values from/to the user buffer, treated as an ASCII string. Negative
2758 * strings are not allowed.
2760 * This routine will ensure the values are within the range specified by
2761 * table->extra1 (min) and table->extra2 (max). There is a final sanity
2762 * check for UINT_MAX to avoid having to support wrap around uses from
2765 * Returns 0 on success or -ERANGE on write when the range check fails.
2767 int proc_douintvec_minmax(struct ctl_table *table, int write,
2768 void __user *buffer, size_t *lenp, loff_t *ppos)
2770 struct do_proc_douintvec_minmax_conv_param param = {
2771 .min = (unsigned int *) table->extra1,
2772 .max = (unsigned int *) table->extra2,
2774 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2775 do_proc_douintvec_minmax_conv, ¶m);
2778 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
2780 int write, void *data)
2785 val = round_pipe_size(*lvalp);
2791 unsigned int val = *valp;
2792 *lvalp = (unsigned long) val;
2798 static int proc_dopipe_max_size(struct ctl_table *table, int write,
2799 void __user *buffer, size_t *lenp, loff_t *ppos)
2801 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2802 do_proc_dopipe_max_size_conv, NULL);
2805 static void validate_coredump_safety(void)
2807 #ifdef CONFIG_COREDUMP
2808 if (suid_dumpable == SUID_DUMP_ROOT &&
2809 core_pattern[0] != '/' && core_pattern[0] != '|') {
2811 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2812 "Pipe handler or fully qualified core dump path required.\n"
2813 "Set kernel.core_pattern before fs.suid_dumpable.\n"
2819 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2820 void __user *buffer, size_t *lenp, loff_t *ppos)
2822 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2824 validate_coredump_safety();
2828 #ifdef CONFIG_COREDUMP
2829 static int proc_dostring_coredump(struct ctl_table *table, int write,
2830 void __user *buffer, size_t *lenp, loff_t *ppos)
2832 int error = proc_dostring(table, write, buffer, lenp, ppos);
2834 validate_coredump_safety();
2839 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2840 void __user *buffer,
2841 size_t *lenp, loff_t *ppos,
2842 unsigned long convmul,
2843 unsigned long convdiv)
2845 unsigned long *i, *min, *max;
2846 int vleft, first = 1, err = 0;
2848 char *kbuf = NULL, *p;
2850 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2855 i = (unsigned long *) data;
2856 min = (unsigned long *) table->extra1;
2857 max = (unsigned long *) table->extra2;
2858 vleft = table->maxlen / sizeof(unsigned long);
2862 if (proc_first_pos_non_zero_ignore(ppos, table))
2865 if (left > PAGE_SIZE - 1)
2866 left = PAGE_SIZE - 1;
2867 p = kbuf = memdup_user_nul(buffer, left);
2869 return PTR_ERR(kbuf);
2872 for (; left && vleft--; i++, first = 0) {
2878 left -= proc_skip_spaces(&p);
2882 err = proc_get_long(&p, &left, &val, &neg,
2884 sizeof(proc_wspace_sep), NULL);
2889 val = convmul * val / convdiv;
2890 if ((min && val < *min) || (max && val > *max)) {
2896 val = convdiv * (*i) / convmul;
2898 err = proc_put_char(&buffer, &left, '\t');
2902 err = proc_put_long(&buffer, &left, val, false);
2908 if (!write && !first && left && !err)
2909 err = proc_put_char(&buffer, &left, '\n');
2911 left -= proc_skip_spaces(&p);
2915 return err ? : -EINVAL;
2923 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2924 void __user *buffer,
2925 size_t *lenp, loff_t *ppos,
2926 unsigned long convmul,
2927 unsigned long convdiv)
2929 return __do_proc_doulongvec_minmax(table->data, table, write,
2930 buffer, lenp, ppos, convmul, convdiv);
2934 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2935 * @table: the sysctl table
2936 * @write: %TRUE if this is a write to the sysctl file
2937 * @buffer: the user buffer
2938 * @lenp: the size of the user buffer
2939 * @ppos: file position
2941 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2942 * values from/to the user buffer, treated as an ASCII string.
2944 * This routine will ensure the values are within the range specified by
2945 * table->extra1 (min) and table->extra2 (max).
2947 * Returns 0 on success.
2949 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2950 void __user *buffer, size_t *lenp, loff_t *ppos)
2952 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2956 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2957 * @table: the sysctl table
2958 * @write: %TRUE if this is a write to the sysctl file
2959 * @buffer: the user buffer
2960 * @lenp: the size of the user buffer
2961 * @ppos: file position
2963 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2964 * values from/to the user buffer, treated as an ASCII string. The values
2965 * are treated as milliseconds, and converted to jiffies when they are stored.
2967 * This routine will ensure the values are within the range specified by
2968 * table->extra1 (min) and table->extra2 (max).
2970 * Returns 0 on success.
2972 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2973 void __user *buffer,
2974 size_t *lenp, loff_t *ppos)
2976 return do_proc_doulongvec_minmax(table, write, buffer,
2977 lenp, ppos, HZ, 1000l);
2981 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2983 int write, void *data)
2986 if (*lvalp > INT_MAX / HZ)
2988 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2994 lval = -(unsigned long)val;
2997 lval = (unsigned long)val;
3004 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
3006 int write, void *data)
3009 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
3011 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
3017 lval = -(unsigned long)val;
3020 lval = (unsigned long)val;
3022 *lvalp = jiffies_to_clock_t(lval);
3027 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
3029 int write, void *data)
3032 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
3042 lval = -(unsigned long)val;
3045 lval = (unsigned long)val;
3047 *lvalp = jiffies_to_msecs(lval);
3053 * proc_dointvec_jiffies - read a vector of integers as seconds
3054 * @table: the sysctl table
3055 * @write: %TRUE if this is a write to the sysctl file
3056 * @buffer: the user buffer
3057 * @lenp: the size of the user buffer
3058 * @ppos: file position
3060 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3061 * values from/to the user buffer, treated as an ASCII string.
3062 * The values read are assumed to be in seconds, and are converted into
3065 * Returns 0 on success.
3067 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3068 void __user *buffer, size_t *lenp, loff_t *ppos)
3070 return do_proc_dointvec(table,write,buffer,lenp,ppos,
3071 do_proc_dointvec_jiffies_conv,NULL);
3075 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
3076 * @table: the sysctl table
3077 * @write: %TRUE if this is a write to the sysctl file
3078 * @buffer: the user buffer
3079 * @lenp: the size of the user buffer
3080 * @ppos: pointer to the file position
3082 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3083 * values from/to the user buffer, treated as an ASCII string.
3084 * The values read are assumed to be in 1/USER_HZ seconds, and
3085 * are converted into jiffies.
3087 * Returns 0 on success.
3089 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3090 void __user *buffer, size_t *lenp, loff_t *ppos)
3092 return do_proc_dointvec(table,write,buffer,lenp,ppos,
3093 do_proc_dointvec_userhz_jiffies_conv,NULL);
3097 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
3098 * @table: the sysctl table
3099 * @write: %TRUE if this is a write to the sysctl file
3100 * @buffer: the user buffer
3101 * @lenp: the size of the user buffer
3102 * @ppos: file position
3103 * @ppos: the current position in the file
3105 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3106 * values from/to the user buffer, treated as an ASCII string.
3107 * The values read are assumed to be in 1/1000 seconds, and
3108 * are converted into jiffies.
3110 * Returns 0 on success.
3112 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3113 void __user *buffer, size_t *lenp, loff_t *ppos)
3115 return do_proc_dointvec(table, write, buffer, lenp, ppos,
3116 do_proc_dointvec_ms_jiffies_conv, NULL);
3119 static int proc_do_cad_pid(struct ctl_table *table, int write,
3120 void __user *buffer, size_t *lenp, loff_t *ppos)
3122 struct pid *new_pid;
3126 tmp = pid_vnr(cad_pid);
3128 r = __do_proc_dointvec(&tmp, table, write, buffer,
3129 lenp, ppos, NULL, NULL);
3133 new_pid = find_get_pid(tmp);
3137 put_pid(xchg(&cad_pid, new_pid));
3142 * proc_do_large_bitmap - read/write from/to a large bitmap
3143 * @table: the sysctl table
3144 * @write: %TRUE if this is a write to the sysctl file
3145 * @buffer: the user buffer
3146 * @lenp: the size of the user buffer
3147 * @ppos: file position
3149 * The bitmap is stored at table->data and the bitmap length (in bits)
3152 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
3153 * large bitmaps may be represented in a compact manner. Writing into
3154 * the file will clear the bitmap then update it with the given input.
3156 * Returns 0 on success.
3158 int proc_do_large_bitmap(struct ctl_table *table, int write,
3159 void __user *buffer, size_t *lenp, loff_t *ppos)
3163 size_t left = *lenp;
3164 unsigned long bitmap_len = table->maxlen;
3165 unsigned long *bitmap = *(unsigned long **) table->data;
3166 unsigned long *tmp_bitmap = NULL;
3167 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
3169 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
3178 if (left > PAGE_SIZE - 1) {
3179 left = PAGE_SIZE - 1;
3180 /* How much of the buffer we'll skip this pass */
3181 skipped = *lenp - left;
3184 p = kbuf = memdup_user_nul(buffer, left);
3186 return PTR_ERR(kbuf);
3188 tmp_bitmap = bitmap_zalloc(bitmap_len, GFP_KERNEL);
3193 proc_skip_char(&p, &left, '\n');
3194 while (!err && left) {
3195 unsigned long val_a, val_b;
3199 /* In case we stop parsing mid-number, we can reset */
3201 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3204 * If we consumed the entirety of a truncated buffer or
3205 * only one char is left (may be a "-"), then stop here,
3206 * reset, & come back for more.
3208 if ((left <= 1) && skipped) {
3215 if (val_a >= bitmap_len || neg) {
3227 err = proc_get_long(&p, &left, &val_b,
3228 &neg, tr_b, sizeof(tr_b),
3231 * If we consumed all of a truncated buffer or
3232 * then stop here, reset, & come back for more.
3234 if (!left && skipped) {
3241 if (val_b >= bitmap_len || neg ||
3252 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3254 proc_skip_char(&p, &left, '\n');
3259 unsigned long bit_a, bit_b = 0;
3262 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3263 if (bit_a >= bitmap_len)
3265 bit_b = find_next_zero_bit(bitmap, bitmap_len,
3269 err = proc_put_char(&buffer, &left, ',');
3273 err = proc_put_long(&buffer, &left, bit_a, false);
3276 if (bit_a != bit_b) {
3277 err = proc_put_char(&buffer, &left, '-');
3280 err = proc_put_long(&buffer, &left, bit_b, false);
3288 err = proc_put_char(&buffer, &left, '\n');
3294 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3296 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3302 bitmap_free(tmp_bitmap);
3306 #else /* CONFIG_PROC_SYSCTL */
3308 int proc_dostring(struct ctl_table *table, int write,
3309 void __user *buffer, size_t *lenp, loff_t *ppos)
3314 int proc_dointvec(struct ctl_table *table, int write,
3315 void __user *buffer, size_t *lenp, loff_t *ppos)
3320 int proc_douintvec(struct ctl_table *table, int write,
3321 void __user *buffer, size_t *lenp, loff_t *ppos)
3326 int proc_dointvec_minmax(struct ctl_table *table, int write,
3327 void __user *buffer, size_t *lenp, loff_t *ppos)
3332 int proc_douintvec_minmax(struct ctl_table *table, int write,
3333 void __user *buffer, size_t *lenp, loff_t *ppos)
3338 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3339 void __user *buffer, size_t *lenp, loff_t *ppos)
3344 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3345 void __user *buffer, size_t *lenp, loff_t *ppos)
3350 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3351 void __user *buffer, size_t *lenp, loff_t *ppos)
3356 int proc_doulongvec_minmax(struct ctl_table *table, int write,
3357 void __user *buffer, size_t *lenp, loff_t *ppos)
3362 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3363 void __user *buffer,
3364 size_t *lenp, loff_t *ppos)
3369 int proc_do_large_bitmap(struct ctl_table *table, int write,
3370 void __user *buffer, size_t *lenp, loff_t *ppos)
3375 #endif /* CONFIG_PROC_SYSCTL */
3377 #if defined(CONFIG_BPF_SYSCALL) && defined(CONFIG_SYSCTL)
3378 static int proc_dointvec_minmax_bpf_stats(struct ctl_table *table, int write,
3379 void __user *buffer, size_t *lenp,
3382 int ret, bpf_stats = *(int *)table->data;
3383 struct ctl_table tmp = *table;
3385 if (write && !capable(CAP_SYS_ADMIN))
3388 tmp.data = &bpf_stats;
3389 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
3390 if (write && !ret) {
3391 *(int *)table->data = bpf_stats;
3393 static_branch_enable(&bpf_stats_enabled_key);
3395 static_branch_disable(&bpf_stats_enabled_key);
3401 * No sense putting this after each symbol definition, twice,
3402 * exception granted :-)
3404 EXPORT_SYMBOL(proc_dointvec);
3405 EXPORT_SYMBOL(proc_douintvec);
3406 EXPORT_SYMBOL(proc_dointvec_jiffies);
3407 EXPORT_SYMBOL(proc_dointvec_minmax);
3408 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3409 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3410 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3411 EXPORT_SYMBOL(proc_dostring);
3412 EXPORT_SYMBOL(proc_doulongvec_minmax);
3413 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3414 EXPORT_SYMBOL(proc_do_large_bitmap);