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;
128 static int __maybe_unused two = 2;
129 static int __maybe_unused four = 4;
130 static unsigned long zero_ul;
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 length 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);
231 #ifdef CONFIG_MAGIC_SYSRQ
232 /* Note: sysrq code uses its own private copy */
233 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
235 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
236 void __user *buffer, size_t *lenp,
241 error = proc_dointvec(table, write, buffer, lenp, ppos);
246 sysrq_toggle_support(__sysrq_enabled);
253 static struct ctl_table kern_table[];
254 static struct ctl_table vm_table[];
255 static struct ctl_table fs_table[];
256 static struct ctl_table debug_table[];
257 static struct ctl_table dev_table[];
258 extern struct ctl_table random_table[];
260 extern struct ctl_table epoll_table[];
263 #ifdef CONFIG_FW_LOADER_USER_HELPER
264 extern struct ctl_table firmware_config_table[];
267 #if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
268 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
269 int sysctl_legacy_va_layout;
272 /* The default sysctl tables: */
274 static struct ctl_table sysctl_base_table[] = {
276 .procname = "kernel",
293 .child = debug_table,
303 #ifdef CONFIG_SCHED_DEBUG
304 static int min_sched_granularity_ns = 100000; /* 100 usecs */
305 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
306 static int min_wakeup_granularity_ns; /* 0 usecs */
307 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
309 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
310 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
311 #endif /* CONFIG_SMP */
312 #endif /* CONFIG_SCHED_DEBUG */
314 #ifdef CONFIG_COMPACTION
315 static int min_extfrag_threshold;
316 static int max_extfrag_threshold = 1000;
319 static struct ctl_table kern_table[] = {
321 .procname = "sched_child_runs_first",
322 .data = &sysctl_sched_child_runs_first,
323 .maxlen = sizeof(unsigned int),
325 .proc_handler = proc_dointvec,
327 #ifdef CONFIG_SCHED_DEBUG
329 .procname = "sched_min_granularity_ns",
330 .data = &sysctl_sched_min_granularity,
331 .maxlen = sizeof(unsigned int),
333 .proc_handler = sched_proc_update_handler,
334 .extra1 = &min_sched_granularity_ns,
335 .extra2 = &max_sched_granularity_ns,
338 .procname = "sched_latency_ns",
339 .data = &sysctl_sched_latency,
340 .maxlen = sizeof(unsigned int),
342 .proc_handler = sched_proc_update_handler,
343 .extra1 = &min_sched_granularity_ns,
344 .extra2 = &max_sched_granularity_ns,
347 .procname = "sched_wakeup_granularity_ns",
348 .data = &sysctl_sched_wakeup_granularity,
349 .maxlen = sizeof(unsigned int),
351 .proc_handler = sched_proc_update_handler,
352 .extra1 = &min_wakeup_granularity_ns,
353 .extra2 = &max_wakeup_granularity_ns,
357 .procname = "sched_tunable_scaling",
358 .data = &sysctl_sched_tunable_scaling,
359 .maxlen = sizeof(enum sched_tunable_scaling),
361 .proc_handler = sched_proc_update_handler,
362 .extra1 = &min_sched_tunable_scaling,
363 .extra2 = &max_sched_tunable_scaling,
366 .procname = "sched_migration_cost_ns",
367 .data = &sysctl_sched_migration_cost,
368 .maxlen = sizeof(unsigned int),
370 .proc_handler = proc_dointvec,
373 .procname = "sched_nr_migrate",
374 .data = &sysctl_sched_nr_migrate,
375 .maxlen = sizeof(unsigned int),
377 .proc_handler = proc_dointvec,
379 #ifdef CONFIG_SCHEDSTATS
381 .procname = "sched_schedstats",
383 .maxlen = sizeof(unsigned int),
385 .proc_handler = sysctl_schedstats,
386 .extra1 = SYSCTL_ZERO,
387 .extra2 = SYSCTL_ONE,
389 #endif /* CONFIG_SCHEDSTATS */
390 #endif /* CONFIG_SMP */
391 #ifdef CONFIG_NUMA_BALANCING
393 .procname = "numa_balancing_scan_delay_ms",
394 .data = &sysctl_numa_balancing_scan_delay,
395 .maxlen = sizeof(unsigned int),
397 .proc_handler = proc_dointvec,
400 .procname = "numa_balancing_scan_period_min_ms",
401 .data = &sysctl_numa_balancing_scan_period_min,
402 .maxlen = sizeof(unsigned int),
404 .proc_handler = proc_dointvec,
407 .procname = "numa_balancing_scan_period_max_ms",
408 .data = &sysctl_numa_balancing_scan_period_max,
409 .maxlen = sizeof(unsigned int),
411 .proc_handler = proc_dointvec,
414 .procname = "numa_balancing_scan_size_mb",
415 .data = &sysctl_numa_balancing_scan_size,
416 .maxlen = sizeof(unsigned int),
418 .proc_handler = proc_dointvec_minmax,
419 .extra1 = SYSCTL_ONE,
422 .procname = "numa_balancing",
423 .data = NULL, /* filled in by handler */
424 .maxlen = sizeof(unsigned int),
426 .proc_handler = sysctl_numa_balancing,
427 .extra1 = SYSCTL_ZERO,
428 .extra2 = SYSCTL_ONE,
430 #endif /* CONFIG_NUMA_BALANCING */
431 #endif /* CONFIG_SCHED_DEBUG */
433 .procname = "sched_rt_period_us",
434 .data = &sysctl_sched_rt_period,
435 .maxlen = sizeof(unsigned int),
437 .proc_handler = sched_rt_handler,
440 .procname = "sched_rt_runtime_us",
441 .data = &sysctl_sched_rt_runtime,
442 .maxlen = sizeof(int),
444 .proc_handler = sched_rt_handler,
447 .procname = "sched_rr_timeslice_ms",
448 .data = &sysctl_sched_rr_timeslice,
449 .maxlen = sizeof(int),
451 .proc_handler = sched_rr_handler,
453 #ifdef CONFIG_UCLAMP_TASK
455 .procname = "sched_util_clamp_min",
456 .data = &sysctl_sched_uclamp_util_min,
457 .maxlen = sizeof(unsigned int),
459 .proc_handler = sysctl_sched_uclamp_handler,
462 .procname = "sched_util_clamp_max",
463 .data = &sysctl_sched_uclamp_util_max,
464 .maxlen = sizeof(unsigned int),
466 .proc_handler = sysctl_sched_uclamp_handler,
469 #ifdef CONFIG_SCHED_AUTOGROUP
471 .procname = "sched_autogroup_enabled",
472 .data = &sysctl_sched_autogroup_enabled,
473 .maxlen = sizeof(unsigned int),
475 .proc_handler = proc_dointvec_minmax,
476 .extra1 = SYSCTL_ZERO,
477 .extra2 = SYSCTL_ONE,
480 #ifdef CONFIG_CFS_BANDWIDTH
482 .procname = "sched_cfs_bandwidth_slice_us",
483 .data = &sysctl_sched_cfs_bandwidth_slice,
484 .maxlen = sizeof(unsigned int),
486 .proc_handler = proc_dointvec_minmax,
487 .extra1 = SYSCTL_ONE,
490 #if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_CPU_FREQ_GOV_SCHEDUTIL)
492 .procname = "sched_energy_aware",
493 .data = &sysctl_sched_energy_aware,
494 .maxlen = sizeof(unsigned int),
496 .proc_handler = sched_energy_aware_handler,
497 .extra1 = SYSCTL_ZERO,
498 .extra2 = SYSCTL_ONE,
501 #ifdef CONFIG_PROVE_LOCKING
503 .procname = "prove_locking",
504 .data = &prove_locking,
505 .maxlen = sizeof(int),
507 .proc_handler = proc_dointvec,
510 #ifdef CONFIG_LOCK_STAT
512 .procname = "lock_stat",
514 .maxlen = sizeof(int),
516 .proc_handler = proc_dointvec,
521 .data = &panic_timeout,
522 .maxlen = sizeof(int),
524 .proc_handler = proc_dointvec,
526 #ifdef CONFIG_COREDUMP
528 .procname = "core_uses_pid",
529 .data = &core_uses_pid,
530 .maxlen = sizeof(int),
532 .proc_handler = proc_dointvec,
535 .procname = "core_pattern",
536 .data = core_pattern,
537 .maxlen = CORENAME_MAX_SIZE,
539 .proc_handler = proc_dostring_coredump,
542 .procname = "core_pipe_limit",
543 .data = &core_pipe_limit,
544 .maxlen = sizeof(unsigned int),
546 .proc_handler = proc_dointvec,
549 #ifdef CONFIG_PROC_SYSCTL
551 .procname = "tainted",
552 .maxlen = sizeof(long),
554 .proc_handler = proc_taint,
557 .procname = "sysctl_writes_strict",
558 .data = &sysctl_writes_strict,
559 .maxlen = sizeof(int),
561 .proc_handler = proc_dointvec_minmax,
563 .extra2 = SYSCTL_ONE,
566 #ifdef CONFIG_LATENCYTOP
568 .procname = "latencytop",
569 .data = &latencytop_enabled,
570 .maxlen = sizeof(int),
572 .proc_handler = sysctl_latencytop,
575 #ifdef CONFIG_BLK_DEV_INITRD
577 .procname = "real-root-dev",
578 .data = &real_root_dev,
579 .maxlen = sizeof(int),
581 .proc_handler = proc_dointvec,
585 .procname = "print-fatal-signals",
586 .data = &print_fatal_signals,
587 .maxlen = sizeof(int),
589 .proc_handler = proc_dointvec,
593 .procname = "reboot-cmd",
594 .data = reboot_command,
597 .proc_handler = proc_dostring,
600 .procname = "stop-a",
601 .data = &stop_a_enabled,
602 .maxlen = sizeof (int),
604 .proc_handler = proc_dointvec,
607 .procname = "scons-poweroff",
608 .data = &scons_pwroff,
609 .maxlen = sizeof (int),
611 .proc_handler = proc_dointvec,
614 #ifdef CONFIG_SPARC64
616 .procname = "tsb-ratio",
617 .data = &sysctl_tsb_ratio,
618 .maxlen = sizeof (int),
620 .proc_handler = proc_dointvec,
625 .procname = "soft-power",
626 .data = &pwrsw_enabled,
627 .maxlen = sizeof (int),
629 .proc_handler = proc_dointvec,
632 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
634 .procname = "unaligned-trap",
635 .data = &unaligned_enabled,
636 .maxlen = sizeof (int),
638 .proc_handler = proc_dointvec,
642 .procname = "ctrl-alt-del",
644 .maxlen = sizeof(int),
646 .proc_handler = proc_dointvec,
648 #ifdef CONFIG_FUNCTION_TRACER
650 .procname = "ftrace_enabled",
651 .data = &ftrace_enabled,
652 .maxlen = sizeof(int),
654 .proc_handler = ftrace_enable_sysctl,
657 #ifdef CONFIG_STACK_TRACER
659 .procname = "stack_tracer_enabled",
660 .data = &stack_tracer_enabled,
661 .maxlen = sizeof(int),
663 .proc_handler = stack_trace_sysctl,
666 #ifdef CONFIG_TRACING
668 .procname = "ftrace_dump_on_oops",
669 .data = &ftrace_dump_on_oops,
670 .maxlen = sizeof(int),
672 .proc_handler = proc_dointvec,
675 .procname = "traceoff_on_warning",
676 .data = &__disable_trace_on_warning,
677 .maxlen = sizeof(__disable_trace_on_warning),
679 .proc_handler = proc_dointvec,
682 .procname = "tracepoint_printk",
683 .data = &tracepoint_printk,
684 .maxlen = sizeof(tracepoint_printk),
686 .proc_handler = tracepoint_printk_sysctl,
689 #ifdef CONFIG_KEXEC_CORE
691 .procname = "kexec_load_disabled",
692 .data = &kexec_load_disabled,
693 .maxlen = sizeof(int),
695 /* only handle a transition from default "0" to "1" */
696 .proc_handler = proc_dointvec_minmax,
697 .extra1 = SYSCTL_ONE,
698 .extra2 = SYSCTL_ONE,
701 #ifdef CONFIG_MODULES
703 .procname = "modprobe",
704 .data = &modprobe_path,
705 .maxlen = KMOD_PATH_LEN,
707 .proc_handler = proc_dostring,
710 .procname = "modules_disabled",
711 .data = &modules_disabled,
712 .maxlen = sizeof(int),
714 /* only handle a transition from default "0" to "1" */
715 .proc_handler = proc_dointvec_minmax,
716 .extra1 = SYSCTL_ONE,
717 .extra2 = SYSCTL_ONE,
720 #ifdef CONFIG_UEVENT_HELPER
722 .procname = "hotplug",
723 .data = &uevent_helper,
724 .maxlen = UEVENT_HELPER_PATH_LEN,
726 .proc_handler = proc_dostring,
729 #ifdef CONFIG_CHR_DEV_SG
731 .procname = "sg-big-buff",
732 .data = &sg_big_buff,
733 .maxlen = sizeof (int),
735 .proc_handler = proc_dointvec,
738 #ifdef CONFIG_BSD_PROCESS_ACCT
742 .maxlen = 3*sizeof(int),
744 .proc_handler = proc_dointvec,
747 #ifdef CONFIG_MAGIC_SYSRQ
750 .data = &__sysrq_enabled,
751 .maxlen = sizeof (int),
753 .proc_handler = sysrq_sysctl_handler,
756 #ifdef CONFIG_PROC_SYSCTL
758 .procname = "cad_pid",
760 .maxlen = sizeof (int),
762 .proc_handler = proc_do_cad_pid,
766 .procname = "threads-max",
768 .maxlen = sizeof(int),
770 .proc_handler = sysctl_max_threads,
773 .procname = "random",
775 .child = random_table,
778 .procname = "usermodehelper",
780 .child = usermodehelper_table,
782 #ifdef CONFIG_FW_LOADER_USER_HELPER
784 .procname = "firmware_config",
786 .child = firmware_config_table,
790 .procname = "overflowuid",
791 .data = &overflowuid,
792 .maxlen = sizeof(int),
794 .proc_handler = proc_dointvec_minmax,
795 .extra1 = &minolduid,
796 .extra2 = &maxolduid,
799 .procname = "overflowgid",
800 .data = &overflowgid,
801 .maxlen = sizeof(int),
803 .proc_handler = proc_dointvec_minmax,
804 .extra1 = &minolduid,
805 .extra2 = &maxolduid,
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,
867 .extra1 = SYSCTL_ZERO,
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,
883 .extra1 = SYSCTL_ZERO,
884 .extra2 = SYSCTL_ONE,
887 .procname = "kptr_restrict",
888 .data = &kptr_restrict,
889 .maxlen = sizeof(int),
891 .proc_handler = proc_dointvec_minmax_sysadmin,
892 .extra1 = SYSCTL_ZERO,
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,
917 .extra1 = SYSCTL_ZERO,
918 .extra2 = SYSCTL_ONE,
921 .procname = "watchdog_thresh",
922 .data = &watchdog_thresh,
923 .maxlen = sizeof(int),
925 .proc_handler = proc_watchdog_thresh,
926 .extra1 = SYSCTL_ZERO,
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,
935 .extra1 = SYSCTL_ZERO,
936 .extra2 = SYSCTL_ONE,
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,
952 .extra1 = SYSCTL_ZERO,
953 .extra2 = SYSCTL_ONE,
956 .procname = "softlockup_panic",
957 .data = &softlockup_panic,
958 .maxlen = sizeof(int),
960 .proc_handler = proc_dointvec_minmax,
961 .extra1 = SYSCTL_ZERO,
962 .extra2 = SYSCTL_ONE,
966 .procname = "softlockup_all_cpu_backtrace",
967 .data = &sysctl_softlockup_all_cpu_backtrace,
968 .maxlen = sizeof(int),
970 .proc_handler = proc_dointvec_minmax,
971 .extra1 = SYSCTL_ZERO,
972 .extra2 = SYSCTL_ONE,
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,
983 .extra1 = SYSCTL_ZERO,
984 .extra2 = SYSCTL_ONE,
988 .procname = "hardlockup_all_cpu_backtrace",
989 .data = &sysctl_hardlockup_all_cpu_backtrace,
990 .maxlen = sizeof(int),
992 .proc_handler = proc_dointvec_minmax,
993 .extra1 = SYSCTL_ZERO,
994 .extra2 = SYSCTL_ONE,
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,
1107 .extra1 = SYSCTL_ZERO,
1108 .extra2 = SYSCTL_ONE,
1111 .procname = "hung_task_check_count",
1112 .data = &sysctl_hung_task_check_count,
1113 .maxlen = sizeof(int),
1115 .proc_handler = proc_dointvec_minmax,
1116 .extra1 = SYSCTL_ZERO,
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,
1193 .extra1 = SYSCTL_ONE,
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,
1201 .extra1 = SYSCTL_ZERO,
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,
1210 .extra1 = SYSCTL_ZERO,
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,
1219 .extra1 = SYSCTL_ZERO,
1220 .extra2 = &one_thousand,
1224 .procname = "panic_on_warn",
1225 .data = &panic_on_warn,
1226 .maxlen = sizeof(int),
1228 .proc_handler = proc_dointvec_minmax,
1229 .extra1 = SYSCTL_ZERO,
1230 .extra2 = SYSCTL_ONE,
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,
1239 .extra1 = SYSCTL_ZERO,
1240 .extra2 = SYSCTL_ONE,
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,
1251 .extra1 = SYSCTL_ONE,
1252 .extra2 = SYSCTL_ONE,
1255 .procname = "bpf_stats_enabled",
1256 .data = &bpf_stats_enabled_key.key,
1257 .maxlen = sizeof(bpf_stats_enabled_key),
1259 .proc_handler = proc_do_static_key,
1262 #if defined(CONFIG_TREE_RCU)
1264 .procname = "panic_on_rcu_stall",
1265 .data = &sysctl_panic_on_rcu_stall,
1266 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
1268 .proc_handler = proc_dointvec_minmax,
1269 .extra1 = SYSCTL_ZERO,
1270 .extra2 = SYSCTL_ONE,
1273 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
1275 .procname = "stack_erasing",
1277 .maxlen = sizeof(int),
1279 .proc_handler = stack_erasing_sysctl,
1280 .extra1 = SYSCTL_ZERO,
1281 .extra2 = SYSCTL_ONE,
1287 static struct ctl_table vm_table[] = {
1289 .procname = "overcommit_memory",
1290 .data = &sysctl_overcommit_memory,
1291 .maxlen = sizeof(sysctl_overcommit_memory),
1293 .proc_handler = proc_dointvec_minmax,
1294 .extra1 = SYSCTL_ZERO,
1298 .procname = "panic_on_oom",
1299 .data = &sysctl_panic_on_oom,
1300 .maxlen = sizeof(sysctl_panic_on_oom),
1302 .proc_handler = proc_dointvec_minmax,
1303 .extra1 = SYSCTL_ZERO,
1307 .procname = "oom_kill_allocating_task",
1308 .data = &sysctl_oom_kill_allocating_task,
1309 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1311 .proc_handler = proc_dointvec,
1314 .procname = "oom_dump_tasks",
1315 .data = &sysctl_oom_dump_tasks,
1316 .maxlen = sizeof(sysctl_oom_dump_tasks),
1318 .proc_handler = proc_dointvec,
1321 .procname = "overcommit_ratio",
1322 .data = &sysctl_overcommit_ratio,
1323 .maxlen = sizeof(sysctl_overcommit_ratio),
1325 .proc_handler = overcommit_ratio_handler,
1328 .procname = "overcommit_kbytes",
1329 .data = &sysctl_overcommit_kbytes,
1330 .maxlen = sizeof(sysctl_overcommit_kbytes),
1332 .proc_handler = overcommit_kbytes_handler,
1335 .procname = "page-cluster",
1336 .data = &page_cluster,
1337 .maxlen = sizeof(int),
1339 .proc_handler = proc_dointvec_minmax,
1340 .extra1 = SYSCTL_ZERO,
1343 .procname = "dirty_background_ratio",
1344 .data = &dirty_background_ratio,
1345 .maxlen = sizeof(dirty_background_ratio),
1347 .proc_handler = dirty_background_ratio_handler,
1348 .extra1 = SYSCTL_ZERO,
1349 .extra2 = &one_hundred,
1352 .procname = "dirty_background_bytes",
1353 .data = &dirty_background_bytes,
1354 .maxlen = sizeof(dirty_background_bytes),
1356 .proc_handler = dirty_background_bytes_handler,
1360 .procname = "dirty_ratio",
1361 .data = &vm_dirty_ratio,
1362 .maxlen = sizeof(vm_dirty_ratio),
1364 .proc_handler = dirty_ratio_handler,
1365 .extra1 = SYSCTL_ZERO,
1366 .extra2 = &one_hundred,
1369 .procname = "dirty_bytes",
1370 .data = &vm_dirty_bytes,
1371 .maxlen = sizeof(vm_dirty_bytes),
1373 .proc_handler = dirty_bytes_handler,
1374 .extra1 = &dirty_bytes_min,
1377 .procname = "dirty_writeback_centisecs",
1378 .data = &dirty_writeback_interval,
1379 .maxlen = sizeof(dirty_writeback_interval),
1381 .proc_handler = dirty_writeback_centisecs_handler,
1384 .procname = "dirty_expire_centisecs",
1385 .data = &dirty_expire_interval,
1386 .maxlen = sizeof(dirty_expire_interval),
1388 .proc_handler = proc_dointvec_minmax,
1389 .extra1 = SYSCTL_ZERO,
1392 .procname = "dirtytime_expire_seconds",
1393 .data = &dirtytime_expire_interval,
1394 .maxlen = sizeof(dirtytime_expire_interval),
1396 .proc_handler = dirtytime_interval_handler,
1397 .extra1 = SYSCTL_ZERO,
1400 .procname = "swappiness",
1401 .data = &vm_swappiness,
1402 .maxlen = sizeof(vm_swappiness),
1404 .proc_handler = proc_dointvec_minmax,
1405 .extra1 = SYSCTL_ZERO,
1406 .extra2 = &one_hundred,
1408 #ifdef CONFIG_HUGETLB_PAGE
1410 .procname = "nr_hugepages",
1412 .maxlen = sizeof(unsigned long),
1414 .proc_handler = hugetlb_sysctl_handler,
1418 .procname = "nr_hugepages_mempolicy",
1420 .maxlen = sizeof(unsigned long),
1422 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1425 .procname = "numa_stat",
1426 .data = &sysctl_vm_numa_stat,
1427 .maxlen = sizeof(int),
1429 .proc_handler = sysctl_vm_numa_stat_handler,
1430 .extra1 = SYSCTL_ZERO,
1431 .extra2 = SYSCTL_ONE,
1435 .procname = "hugetlb_shm_group",
1436 .data = &sysctl_hugetlb_shm_group,
1437 .maxlen = sizeof(gid_t),
1439 .proc_handler = proc_dointvec,
1442 .procname = "nr_overcommit_hugepages",
1444 .maxlen = sizeof(unsigned long),
1446 .proc_handler = hugetlb_overcommit_handler,
1450 .procname = "lowmem_reserve_ratio",
1451 .data = &sysctl_lowmem_reserve_ratio,
1452 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1454 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1457 .procname = "drop_caches",
1458 .data = &sysctl_drop_caches,
1459 .maxlen = sizeof(int),
1461 .proc_handler = drop_caches_sysctl_handler,
1462 .extra1 = SYSCTL_ONE,
1465 #ifdef CONFIG_COMPACTION
1467 .procname = "compact_memory",
1468 .data = &sysctl_compact_memory,
1469 .maxlen = sizeof(int),
1471 .proc_handler = sysctl_compaction_handler,
1474 .procname = "extfrag_threshold",
1475 .data = &sysctl_extfrag_threshold,
1476 .maxlen = sizeof(int),
1478 .proc_handler = proc_dointvec_minmax,
1479 .extra1 = &min_extfrag_threshold,
1480 .extra2 = &max_extfrag_threshold,
1483 .procname = "compact_unevictable_allowed",
1484 .data = &sysctl_compact_unevictable_allowed,
1485 .maxlen = sizeof(int),
1487 .proc_handler = proc_dointvec,
1488 .extra1 = SYSCTL_ZERO,
1489 .extra2 = SYSCTL_ONE,
1492 #endif /* CONFIG_COMPACTION */
1494 .procname = "min_free_kbytes",
1495 .data = &min_free_kbytes,
1496 .maxlen = sizeof(min_free_kbytes),
1498 .proc_handler = min_free_kbytes_sysctl_handler,
1499 .extra1 = SYSCTL_ZERO,
1502 .procname = "watermark_boost_factor",
1503 .data = &watermark_boost_factor,
1504 .maxlen = sizeof(watermark_boost_factor),
1506 .proc_handler = watermark_boost_factor_sysctl_handler,
1507 .extra1 = SYSCTL_ZERO,
1510 .procname = "watermark_scale_factor",
1511 .data = &watermark_scale_factor,
1512 .maxlen = sizeof(watermark_scale_factor),
1514 .proc_handler = watermark_scale_factor_sysctl_handler,
1515 .extra1 = SYSCTL_ONE,
1516 .extra2 = &one_thousand,
1519 .procname = "percpu_pagelist_fraction",
1520 .data = &percpu_pagelist_fraction,
1521 .maxlen = sizeof(percpu_pagelist_fraction),
1523 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1524 .extra1 = SYSCTL_ZERO,
1528 .procname = "max_map_count",
1529 .data = &sysctl_max_map_count,
1530 .maxlen = sizeof(sysctl_max_map_count),
1532 .proc_handler = proc_dointvec_minmax,
1533 .extra1 = SYSCTL_ZERO,
1537 .procname = "nr_trim_pages",
1538 .data = &sysctl_nr_trim_pages,
1539 .maxlen = sizeof(sysctl_nr_trim_pages),
1541 .proc_handler = proc_dointvec_minmax,
1542 .extra1 = SYSCTL_ZERO,
1546 .procname = "laptop_mode",
1547 .data = &laptop_mode,
1548 .maxlen = sizeof(laptop_mode),
1550 .proc_handler = proc_dointvec_jiffies,
1553 .procname = "block_dump",
1554 .data = &block_dump,
1555 .maxlen = sizeof(block_dump),
1557 .proc_handler = proc_dointvec,
1558 .extra1 = SYSCTL_ZERO,
1561 .procname = "vfs_cache_pressure",
1562 .data = &sysctl_vfs_cache_pressure,
1563 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1565 .proc_handler = proc_dointvec,
1566 .extra1 = SYSCTL_ZERO,
1568 #if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
1569 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
1571 .procname = "legacy_va_layout",
1572 .data = &sysctl_legacy_va_layout,
1573 .maxlen = sizeof(sysctl_legacy_va_layout),
1575 .proc_handler = proc_dointvec,
1576 .extra1 = SYSCTL_ZERO,
1581 .procname = "zone_reclaim_mode",
1582 .data = &node_reclaim_mode,
1583 .maxlen = sizeof(node_reclaim_mode),
1585 .proc_handler = proc_dointvec,
1586 .extra1 = SYSCTL_ZERO,
1589 .procname = "min_unmapped_ratio",
1590 .data = &sysctl_min_unmapped_ratio,
1591 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1593 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1594 .extra1 = SYSCTL_ZERO,
1595 .extra2 = &one_hundred,
1598 .procname = "min_slab_ratio",
1599 .data = &sysctl_min_slab_ratio,
1600 .maxlen = sizeof(sysctl_min_slab_ratio),
1602 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1603 .extra1 = SYSCTL_ZERO,
1604 .extra2 = &one_hundred,
1609 .procname = "stat_interval",
1610 .data = &sysctl_stat_interval,
1611 .maxlen = sizeof(sysctl_stat_interval),
1613 .proc_handler = proc_dointvec_jiffies,
1616 .procname = "stat_refresh",
1620 .proc_handler = vmstat_refresh,
1625 .procname = "mmap_min_addr",
1626 .data = &dac_mmap_min_addr,
1627 .maxlen = sizeof(unsigned long),
1629 .proc_handler = mmap_min_addr_handler,
1634 .procname = "numa_zonelist_order",
1635 .data = &numa_zonelist_order,
1636 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1638 .proc_handler = numa_zonelist_order_handler,
1641 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1642 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1644 .procname = "vdso_enabled",
1645 #ifdef CONFIG_X86_32
1646 .data = &vdso32_enabled,
1647 .maxlen = sizeof(vdso32_enabled),
1649 .data = &vdso_enabled,
1650 .maxlen = sizeof(vdso_enabled),
1653 .proc_handler = proc_dointvec,
1654 .extra1 = SYSCTL_ZERO,
1657 #ifdef CONFIG_HIGHMEM
1659 .procname = "highmem_is_dirtyable",
1660 .data = &vm_highmem_is_dirtyable,
1661 .maxlen = sizeof(vm_highmem_is_dirtyable),
1663 .proc_handler = proc_dointvec_minmax,
1664 .extra1 = SYSCTL_ZERO,
1665 .extra2 = SYSCTL_ONE,
1668 #ifdef CONFIG_MEMORY_FAILURE
1670 .procname = "memory_failure_early_kill",
1671 .data = &sysctl_memory_failure_early_kill,
1672 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1674 .proc_handler = proc_dointvec_minmax,
1675 .extra1 = SYSCTL_ZERO,
1676 .extra2 = SYSCTL_ONE,
1679 .procname = "memory_failure_recovery",
1680 .data = &sysctl_memory_failure_recovery,
1681 .maxlen = sizeof(sysctl_memory_failure_recovery),
1683 .proc_handler = proc_dointvec_minmax,
1684 .extra1 = SYSCTL_ZERO,
1685 .extra2 = SYSCTL_ONE,
1689 .procname = "user_reserve_kbytes",
1690 .data = &sysctl_user_reserve_kbytes,
1691 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1693 .proc_handler = proc_doulongvec_minmax,
1696 .procname = "admin_reserve_kbytes",
1697 .data = &sysctl_admin_reserve_kbytes,
1698 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1700 .proc_handler = proc_doulongvec_minmax,
1702 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1704 .procname = "mmap_rnd_bits",
1705 .data = &mmap_rnd_bits,
1706 .maxlen = sizeof(mmap_rnd_bits),
1708 .proc_handler = proc_dointvec_minmax,
1709 .extra1 = (void *)&mmap_rnd_bits_min,
1710 .extra2 = (void *)&mmap_rnd_bits_max,
1713 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1715 .procname = "mmap_rnd_compat_bits",
1716 .data = &mmap_rnd_compat_bits,
1717 .maxlen = sizeof(mmap_rnd_compat_bits),
1719 .proc_handler = proc_dointvec_minmax,
1720 .extra1 = (void *)&mmap_rnd_compat_bits_min,
1721 .extra2 = (void *)&mmap_rnd_compat_bits_max,
1724 #ifdef CONFIG_USERFAULTFD
1726 .procname = "unprivileged_userfaultfd",
1727 .data = &sysctl_unprivileged_userfaultfd,
1728 .maxlen = sizeof(sysctl_unprivileged_userfaultfd),
1730 .proc_handler = proc_dointvec_minmax,
1731 .extra1 = SYSCTL_ZERO,
1732 .extra2 = SYSCTL_ONE,
1738 static struct ctl_table fs_table[] = {
1740 .procname = "inode-nr",
1741 .data = &inodes_stat,
1742 .maxlen = 2*sizeof(long),
1744 .proc_handler = proc_nr_inodes,
1747 .procname = "inode-state",
1748 .data = &inodes_stat,
1749 .maxlen = 7*sizeof(long),
1751 .proc_handler = proc_nr_inodes,
1754 .procname = "file-nr",
1755 .data = &files_stat,
1756 .maxlen = sizeof(files_stat),
1758 .proc_handler = proc_nr_files,
1761 .procname = "file-max",
1762 .data = &files_stat.max_files,
1763 .maxlen = sizeof(files_stat.max_files),
1765 .proc_handler = proc_doulongvec_minmax,
1767 .extra2 = &long_max,
1770 .procname = "nr_open",
1771 .data = &sysctl_nr_open,
1772 .maxlen = sizeof(unsigned int),
1774 .proc_handler = proc_dointvec_minmax,
1775 .extra1 = &sysctl_nr_open_min,
1776 .extra2 = &sysctl_nr_open_max,
1779 .procname = "dentry-state",
1780 .data = &dentry_stat,
1781 .maxlen = 6*sizeof(long),
1783 .proc_handler = proc_nr_dentry,
1786 .procname = "overflowuid",
1787 .data = &fs_overflowuid,
1788 .maxlen = sizeof(int),
1790 .proc_handler = proc_dointvec_minmax,
1791 .extra1 = &minolduid,
1792 .extra2 = &maxolduid,
1795 .procname = "overflowgid",
1796 .data = &fs_overflowgid,
1797 .maxlen = sizeof(int),
1799 .proc_handler = proc_dointvec_minmax,
1800 .extra1 = &minolduid,
1801 .extra2 = &maxolduid,
1803 #ifdef CONFIG_FILE_LOCKING
1805 .procname = "leases-enable",
1806 .data = &leases_enable,
1807 .maxlen = sizeof(int),
1809 .proc_handler = proc_dointvec,
1812 #ifdef CONFIG_DNOTIFY
1814 .procname = "dir-notify-enable",
1815 .data = &dir_notify_enable,
1816 .maxlen = sizeof(int),
1818 .proc_handler = proc_dointvec,
1822 #ifdef CONFIG_FILE_LOCKING
1824 .procname = "lease-break-time",
1825 .data = &lease_break_time,
1826 .maxlen = sizeof(int),
1828 .proc_handler = proc_dointvec,
1833 .procname = "aio-nr",
1835 .maxlen = sizeof(aio_nr),
1837 .proc_handler = proc_doulongvec_minmax,
1840 .procname = "aio-max-nr",
1841 .data = &aio_max_nr,
1842 .maxlen = sizeof(aio_max_nr),
1844 .proc_handler = proc_doulongvec_minmax,
1846 #endif /* CONFIG_AIO */
1847 #ifdef CONFIG_INOTIFY_USER
1849 .procname = "inotify",
1851 .child = inotify_table,
1856 .procname = "epoll",
1858 .child = epoll_table,
1863 .procname = "protected_symlinks",
1864 .data = &sysctl_protected_symlinks,
1865 .maxlen = sizeof(int),
1867 .proc_handler = proc_dointvec_minmax,
1868 .extra1 = SYSCTL_ZERO,
1869 .extra2 = SYSCTL_ONE,
1872 .procname = "protected_hardlinks",
1873 .data = &sysctl_protected_hardlinks,
1874 .maxlen = sizeof(int),
1876 .proc_handler = proc_dointvec_minmax,
1877 .extra1 = SYSCTL_ZERO,
1878 .extra2 = SYSCTL_ONE,
1881 .procname = "protected_fifos",
1882 .data = &sysctl_protected_fifos,
1883 .maxlen = sizeof(int),
1885 .proc_handler = proc_dointvec_minmax,
1886 .extra1 = SYSCTL_ZERO,
1890 .procname = "protected_regular",
1891 .data = &sysctl_protected_regular,
1892 .maxlen = sizeof(int),
1894 .proc_handler = proc_dointvec_minmax,
1895 .extra1 = SYSCTL_ZERO,
1899 .procname = "suid_dumpable",
1900 .data = &suid_dumpable,
1901 .maxlen = sizeof(int),
1903 .proc_handler = proc_dointvec_minmax_coredump,
1904 .extra1 = SYSCTL_ZERO,
1907 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1909 .procname = "binfmt_misc",
1911 .child = sysctl_mount_point,
1915 .procname = "pipe-max-size",
1916 .data = &pipe_max_size,
1917 .maxlen = sizeof(pipe_max_size),
1919 .proc_handler = proc_dopipe_max_size,
1922 .procname = "pipe-user-pages-hard",
1923 .data = &pipe_user_pages_hard,
1924 .maxlen = sizeof(pipe_user_pages_hard),
1926 .proc_handler = proc_doulongvec_minmax,
1929 .procname = "pipe-user-pages-soft",
1930 .data = &pipe_user_pages_soft,
1931 .maxlen = sizeof(pipe_user_pages_soft),
1933 .proc_handler = proc_doulongvec_minmax,
1936 .procname = "mount-max",
1937 .data = &sysctl_mount_max,
1938 .maxlen = sizeof(unsigned int),
1940 .proc_handler = proc_dointvec_minmax,
1941 .extra1 = SYSCTL_ONE,
1946 static struct ctl_table debug_table[] = {
1947 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1949 .procname = "exception-trace",
1950 .data = &show_unhandled_signals,
1951 .maxlen = sizeof(int),
1953 .proc_handler = proc_dointvec
1956 #if defined(CONFIG_OPTPROBES)
1958 .procname = "kprobes-optimization",
1959 .data = &sysctl_kprobes_optimization,
1960 .maxlen = sizeof(int),
1962 .proc_handler = proc_kprobes_optimization_handler,
1963 .extra1 = SYSCTL_ZERO,
1964 .extra2 = SYSCTL_ONE,
1970 static struct ctl_table dev_table[] = {
1974 int __init sysctl_init(void)
1976 struct ctl_table_header *hdr;
1978 hdr = register_sysctl_table(sysctl_base_table);
1979 kmemleak_not_leak(hdr);
1983 #endif /* CONFIG_SYSCTL */
1989 #ifdef CONFIG_PROC_SYSCTL
1991 static int _proc_do_string(char *data, int maxlen, int write,
1992 char __user *buffer,
1993 size_t *lenp, loff_t *ppos)
1999 if (!data || !maxlen || !*lenp) {
2005 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
2006 /* Only continue writes not past the end of buffer. */
2008 if (len > maxlen - 1)
2015 /* Start writing from beginning of buffer. */
2021 while ((p - buffer) < *lenp && len < maxlen - 1) {
2022 if (get_user(c, p++))
2024 if (c == 0 || c == '\n')
2045 if (copy_to_user(buffer, data, len))
2048 if (put_user('\n', buffer + len))
2058 static void warn_sysctl_write(struct ctl_table *table)
2060 pr_warn_once("%s wrote to %s when file position was not 0!\n"
2061 "This will not be supported in the future. To silence this\n"
2062 "warning, set kernel.sysctl_writes_strict = -1\n",
2063 current->comm, table->procname);
2067 * proc_first_pos_non_zero_ignore - check if first position is allowed
2068 * @ppos: file position
2069 * @table: the sysctl table
2071 * Returns true if the first position is non-zero and the sysctl_writes_strict
2072 * mode indicates this is not allowed for numeric input types. String proc
2073 * handlers can ignore the return value.
2075 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
2076 struct ctl_table *table)
2081 switch (sysctl_writes_strict) {
2082 case SYSCTL_WRITES_STRICT:
2084 case SYSCTL_WRITES_WARN:
2085 warn_sysctl_write(table);
2093 * proc_dostring - read a string sysctl
2094 * @table: the sysctl table
2095 * @write: %TRUE if this is a write to the sysctl file
2096 * @buffer: the user buffer
2097 * @lenp: the size of the user buffer
2098 * @ppos: file position
2100 * Reads/writes a string from/to the user buffer. If the kernel
2101 * buffer provided is not large enough to hold the string, the
2102 * string is truncated. The copied string is %NULL-terminated.
2103 * If the string is being read by the user process, it is copied
2104 * and a newline '\n' is added. It is truncated if the buffer is
2107 * Returns 0 on success.
2109 int proc_dostring(struct ctl_table *table, int write,
2110 void __user *buffer, size_t *lenp, loff_t *ppos)
2113 proc_first_pos_non_zero_ignore(ppos, table);
2115 return _proc_do_string((char *)(table->data), table->maxlen, write,
2116 (char __user *)buffer, lenp, ppos);
2119 static size_t proc_skip_spaces(char **buf)
2122 char *tmp = skip_spaces(*buf);
2128 static void proc_skip_char(char **buf, size_t *size, const char v)
2139 * strtoul_lenient - parse an ASCII formatted integer from a buffer and only
2142 * @cp: kernel buffer containing the string to parse
2143 * @endp: pointer to store the trailing characters
2144 * @base: the base to use
2145 * @res: where the parsed integer will be stored
2147 * In case of success 0 is returned and @res will contain the parsed integer,
2148 * @endp will hold any trailing characters.
2149 * This function will fail the parse on overflow. If there wasn't an overflow
2150 * the function will defer the decision what characters count as invalid to the
2153 static int strtoul_lenient(const char *cp, char **endp, unsigned int base,
2156 unsigned long long result;
2159 cp = _parse_integer_fixup_radix(cp, &base);
2160 rv = _parse_integer(cp, base, &result);
2161 if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result))
2169 *res = (unsigned long)result;
2173 #define TMPBUFLEN 22
2175 * proc_get_long - reads an ASCII formatted integer from a user buffer
2177 * @buf: a kernel buffer
2178 * @size: size of the kernel buffer
2179 * @val: this is where the number will be stored
2180 * @neg: set to %TRUE if number is negative
2181 * @perm_tr: a vector which contains the allowed trailers
2182 * @perm_tr_len: size of the perm_tr vector
2183 * @tr: pointer to store the trailer character
2185 * In case of success %0 is returned and @buf and @size are updated with
2186 * the amount of bytes read. If @tr is non-NULL and a trailing
2187 * character exists (size is non-zero after returning from this
2188 * function), @tr is updated with the trailing character.
2190 static int proc_get_long(char **buf, size_t *size,
2191 unsigned long *val, bool *neg,
2192 const char *perm_tr, unsigned perm_tr_len, char *tr)
2195 char *p, tmp[TMPBUFLEN];
2201 if (len > TMPBUFLEN - 1)
2202 len = TMPBUFLEN - 1;
2204 memcpy(tmp, *buf, len);
2208 if (*p == '-' && *size > 1) {
2216 if (strtoul_lenient(p, &p, 0, val))
2221 /* We don't know if the next char is whitespace thus we may accept
2222 * invalid integers (e.g. 1234...a) or two integers instead of one
2223 * (e.g. 123...1). So lets not allow such large numbers. */
2224 if (len == TMPBUFLEN - 1)
2227 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2230 if (tr && (len < *size))
2240 * proc_put_long - converts an integer to a decimal ASCII formatted string
2242 * @buf: the user buffer
2243 * @size: the size of the user buffer
2244 * @val: the integer to be converted
2245 * @neg: sign of the number, %TRUE for negative
2247 * In case of success %0 is returned and @buf and @size are updated with
2248 * the amount of bytes written.
2250 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2254 char tmp[TMPBUFLEN], *p = tmp;
2256 sprintf(p, "%s%lu", neg ? "-" : "", val);
2260 if (copy_to_user(*buf, tmp, len))
2268 static int proc_put_char(void __user **buf, size_t *size, char c)
2271 char __user **buffer = (char __user **)buf;
2272 if (put_user(c, *buffer))
2274 (*size)--, (*buffer)++;
2280 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2282 int write, void *data)
2286 if (*lvalp > (unsigned long) INT_MAX + 1)
2290 if (*lvalp > (unsigned long) INT_MAX)
2298 *lvalp = -(unsigned long)val;
2301 *lvalp = (unsigned long)val;
2307 static int do_proc_douintvec_conv(unsigned long *lvalp,
2309 int write, void *data)
2312 if (*lvalp > UINT_MAX)
2316 unsigned int val = *valp;
2317 *lvalp = (unsigned long)val;
2322 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2324 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2325 int write, void __user *buffer,
2326 size_t *lenp, loff_t *ppos,
2327 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2328 int write, void *data),
2331 int *i, vleft, first = 1, err = 0;
2333 char *kbuf = NULL, *p;
2335 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2340 i = (int *) tbl_data;
2341 vleft = table->maxlen / sizeof(*i);
2345 conv = do_proc_dointvec_conv;
2348 if (proc_first_pos_non_zero_ignore(ppos, table))
2351 if (left > PAGE_SIZE - 1)
2352 left = PAGE_SIZE - 1;
2353 p = kbuf = memdup_user_nul(buffer, left);
2355 return PTR_ERR(kbuf);
2358 for (; left && vleft--; i++, first=0) {
2363 left -= proc_skip_spaces(&p);
2367 err = proc_get_long(&p, &left, &lval, &neg,
2369 sizeof(proc_wspace_sep), NULL);
2372 if (conv(&neg, &lval, i, 1, data)) {
2377 if (conv(&neg, &lval, i, 0, data)) {
2382 err = proc_put_char(&buffer, &left, '\t');
2385 err = proc_put_long(&buffer, &left, lval, neg);
2391 if (!write && !first && left && !err)
2392 err = proc_put_char(&buffer, &left, '\n');
2393 if (write && !err && left)
2394 left -= proc_skip_spaces(&p);
2398 return err ? : -EINVAL;
2406 static int do_proc_dointvec(struct ctl_table *table, int write,
2407 void __user *buffer, size_t *lenp, loff_t *ppos,
2408 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2409 int write, void *data),
2412 return __do_proc_dointvec(table->data, table, write,
2413 buffer, lenp, ppos, conv, data);
2416 static int do_proc_douintvec_w(unsigned int *tbl_data,
2417 struct ctl_table *table,
2418 void __user *buffer,
2419 size_t *lenp, loff_t *ppos,
2420 int (*conv)(unsigned long *lvalp,
2422 int write, void *data),
2429 char *kbuf = NULL, *p;
2433 if (proc_first_pos_non_zero_ignore(ppos, table))
2436 if (left > PAGE_SIZE - 1)
2437 left = PAGE_SIZE - 1;
2439 p = kbuf = memdup_user_nul(buffer, left);
2443 left -= proc_skip_spaces(&p);
2449 err = proc_get_long(&p, &left, &lval, &neg,
2451 sizeof(proc_wspace_sep), NULL);
2457 if (conv(&lval, tbl_data, 1, data)) {
2463 left -= proc_skip_spaces(&p);
2472 /* This is in keeping with old __do_proc_dointvec() */
2478 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2479 size_t *lenp, loff_t *ppos,
2480 int (*conv)(unsigned long *lvalp,
2482 int write, void *data),
2491 if (conv(&lval, tbl_data, 0, data)) {
2496 err = proc_put_long(&buffer, &left, lval, false);
2500 err = proc_put_char(&buffer, &left, '\n');
2509 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2510 int write, void __user *buffer,
2511 size_t *lenp, loff_t *ppos,
2512 int (*conv)(unsigned long *lvalp,
2514 int write, void *data),
2517 unsigned int *i, vleft;
2519 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2524 i = (unsigned int *) tbl_data;
2525 vleft = table->maxlen / sizeof(*i);
2528 * Arrays are not supported, keep this simple. *Do not* add
2537 conv = do_proc_douintvec_conv;
2540 return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2542 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2545 static int do_proc_douintvec(struct ctl_table *table, int write,
2546 void __user *buffer, size_t *lenp, loff_t *ppos,
2547 int (*conv)(unsigned long *lvalp,
2549 int write, void *data),
2552 return __do_proc_douintvec(table->data, table, write,
2553 buffer, lenp, ppos, conv, data);
2557 * proc_dointvec - read a vector of integers
2558 * @table: the sysctl table
2559 * @write: %TRUE if this is a write to the sysctl file
2560 * @buffer: the user buffer
2561 * @lenp: the size of the user buffer
2562 * @ppos: file position
2564 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2565 * values from/to the user buffer, treated as an ASCII string.
2567 * Returns 0 on success.
2569 int proc_dointvec(struct ctl_table *table, int write,
2570 void __user *buffer, size_t *lenp, loff_t *ppos)
2572 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2576 * proc_douintvec - read a vector of unsigned integers
2577 * @table: the sysctl table
2578 * @write: %TRUE if this is a write to the sysctl file
2579 * @buffer: the user buffer
2580 * @lenp: the size of the user buffer
2581 * @ppos: file position
2583 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2584 * values from/to the user buffer, treated as an ASCII string.
2586 * Returns 0 on success.
2588 int proc_douintvec(struct ctl_table *table, int write,
2589 void __user *buffer, size_t *lenp, loff_t *ppos)
2591 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2592 do_proc_douintvec_conv, NULL);
2596 * Taint values can only be increased
2597 * This means we can safely use a temporary.
2599 static int proc_taint(struct ctl_table *table, int write,
2600 void __user *buffer, size_t *lenp, loff_t *ppos)
2603 unsigned long tmptaint = get_taint();
2606 if (write && !capable(CAP_SYS_ADMIN))
2611 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2617 * Poor man's atomic or. Not worth adding a primitive
2618 * to everyone's atomic.h for this
2621 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2622 if ((tmptaint >> i) & 1)
2623 add_taint(i, LOCKDEP_STILL_OK);
2630 #ifdef CONFIG_PRINTK
2631 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2632 void __user *buffer, size_t *lenp, loff_t *ppos)
2634 if (write && !capable(CAP_SYS_ADMIN))
2637 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2642 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
2643 * @min: pointer to minimum allowable value
2644 * @max: pointer to maximum allowable value
2646 * The do_proc_dointvec_minmax_conv_param structure provides the
2647 * minimum and maximum values for doing range checking for those sysctl
2648 * parameters that use the proc_dointvec_minmax() handler.
2650 struct do_proc_dointvec_minmax_conv_param {
2655 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2657 int write, void *data)
2660 struct do_proc_dointvec_minmax_conv_param *param = data;
2662 * If writing, first do so via a temporary local int so we can
2663 * bounds-check it before touching *valp.
2665 int *ip = write ? &tmp : valp;
2667 ret = do_proc_dointvec_conv(negp, lvalp, ip, write, data);
2672 if ((param->min && *param->min > tmp) ||
2673 (param->max && *param->max < tmp))
2682 * proc_dointvec_minmax - read a vector of integers with min/max values
2683 * @table: the sysctl table
2684 * @write: %TRUE if this is a write to the sysctl file
2685 * @buffer: the user buffer
2686 * @lenp: the size of the user buffer
2687 * @ppos: file position
2689 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2690 * values from/to the user buffer, treated as an ASCII string.
2692 * This routine will ensure the values are within the range specified by
2693 * table->extra1 (min) and table->extra2 (max).
2695 * Returns 0 on success or -EINVAL on write when the range check fails.
2697 int proc_dointvec_minmax(struct ctl_table *table, int write,
2698 void __user *buffer, size_t *lenp, loff_t *ppos)
2700 struct do_proc_dointvec_minmax_conv_param param = {
2701 .min = (int *) table->extra1,
2702 .max = (int *) table->extra2,
2704 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2705 do_proc_dointvec_minmax_conv, ¶m);
2709 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
2710 * @min: pointer to minimum allowable value
2711 * @max: pointer to maximum allowable value
2713 * The do_proc_douintvec_minmax_conv_param structure provides the
2714 * minimum and maximum values for doing range checking for those sysctl
2715 * parameters that use the proc_douintvec_minmax() handler.
2717 struct do_proc_douintvec_minmax_conv_param {
2722 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2724 int write, void *data)
2728 struct do_proc_douintvec_minmax_conv_param *param = data;
2729 /* write via temporary local uint for bounds-checking */
2730 unsigned int *up = write ? &tmp : valp;
2732 ret = do_proc_douintvec_conv(lvalp, up, write, data);
2737 if ((param->min && *param->min > tmp) ||
2738 (param->max && *param->max < tmp))
2748 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2749 * @table: the sysctl table
2750 * @write: %TRUE if this is a write to the sysctl file
2751 * @buffer: the user buffer
2752 * @lenp: the size of the user buffer
2753 * @ppos: file position
2755 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2756 * values from/to the user buffer, treated as an ASCII string. Negative
2757 * strings are not allowed.
2759 * This routine will ensure the values are within the range specified by
2760 * table->extra1 (min) and table->extra2 (max). There is a final sanity
2761 * check for UINT_MAX to avoid having to support wrap around uses from
2764 * Returns 0 on success or -ERANGE on write when the range check fails.
2766 int proc_douintvec_minmax(struct ctl_table *table, int write,
2767 void __user *buffer, size_t *lenp, loff_t *ppos)
2769 struct do_proc_douintvec_minmax_conv_param param = {
2770 .min = (unsigned int *) table->extra1,
2771 .max = (unsigned int *) table->extra2,
2773 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2774 do_proc_douintvec_minmax_conv, ¶m);
2777 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
2779 int write, void *data)
2784 val = round_pipe_size(*lvalp);
2790 unsigned int val = *valp;
2791 *lvalp = (unsigned long) val;
2797 static int proc_dopipe_max_size(struct ctl_table *table, int write,
2798 void __user *buffer, size_t *lenp, loff_t *ppos)
2800 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2801 do_proc_dopipe_max_size_conv, NULL);
2804 static void validate_coredump_safety(void)
2806 #ifdef CONFIG_COREDUMP
2807 if (suid_dumpable == SUID_DUMP_ROOT &&
2808 core_pattern[0] != '/' && core_pattern[0] != '|') {
2810 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2811 "Pipe handler or fully qualified core dump path required.\n"
2812 "Set kernel.core_pattern before fs.suid_dumpable.\n"
2818 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2819 void __user *buffer, size_t *lenp, loff_t *ppos)
2821 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2823 validate_coredump_safety();
2827 #ifdef CONFIG_COREDUMP
2828 static int proc_dostring_coredump(struct ctl_table *table, int write,
2829 void __user *buffer, size_t *lenp, loff_t *ppos)
2831 int error = proc_dostring(table, write, buffer, lenp, ppos);
2833 validate_coredump_safety();
2838 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2839 void __user *buffer,
2840 size_t *lenp, loff_t *ppos,
2841 unsigned long convmul,
2842 unsigned long convdiv)
2844 unsigned long *i, *min, *max;
2845 int vleft, first = 1, err = 0;
2847 char *kbuf = NULL, *p;
2849 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2854 i = (unsigned long *) data;
2855 min = (unsigned long *) table->extra1;
2856 max = (unsigned long *) table->extra2;
2857 vleft = table->maxlen / sizeof(unsigned long);
2861 if (proc_first_pos_non_zero_ignore(ppos, table))
2864 if (left > PAGE_SIZE - 1)
2865 left = PAGE_SIZE - 1;
2866 p = kbuf = memdup_user_nul(buffer, left);
2868 return PTR_ERR(kbuf);
2871 for (; left && vleft--; i++, first = 0) {
2877 left -= proc_skip_spaces(&p);
2881 err = proc_get_long(&p, &left, &val, &neg,
2883 sizeof(proc_wspace_sep), NULL);
2888 val = convmul * val / convdiv;
2889 if ((min && val < *min) || (max && val > *max)) {
2895 val = convdiv * (*i) / convmul;
2897 err = proc_put_char(&buffer, &left, '\t');
2901 err = proc_put_long(&buffer, &left, val, false);
2907 if (!write && !first && left && !err)
2908 err = proc_put_char(&buffer, &left, '\n');
2910 left -= proc_skip_spaces(&p);
2914 return err ? : -EINVAL;
2922 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2923 void __user *buffer,
2924 size_t *lenp, loff_t *ppos,
2925 unsigned long convmul,
2926 unsigned long convdiv)
2928 return __do_proc_doulongvec_minmax(table->data, table, write,
2929 buffer, lenp, ppos, convmul, convdiv);
2933 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2934 * @table: the sysctl table
2935 * @write: %TRUE if this is a write to the sysctl file
2936 * @buffer: the user buffer
2937 * @lenp: the size of the user buffer
2938 * @ppos: file position
2940 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2941 * values from/to the user buffer, treated as an ASCII string.
2943 * This routine will ensure the values are within the range specified by
2944 * table->extra1 (min) and table->extra2 (max).
2946 * Returns 0 on success.
2948 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2949 void __user *buffer, size_t *lenp, loff_t *ppos)
2951 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2955 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2956 * @table: the sysctl table
2957 * @write: %TRUE if this is a write to the sysctl file
2958 * @buffer: the user buffer
2959 * @lenp: the size of the user buffer
2960 * @ppos: file position
2962 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2963 * values from/to the user buffer, treated as an ASCII string. The values
2964 * are treated as milliseconds, and converted to jiffies when they are stored.
2966 * This routine will ensure the values are within the range specified by
2967 * table->extra1 (min) and table->extra2 (max).
2969 * Returns 0 on success.
2971 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2972 void __user *buffer,
2973 size_t *lenp, loff_t *ppos)
2975 return do_proc_doulongvec_minmax(table, write, buffer,
2976 lenp, ppos, HZ, 1000l);
2980 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2982 int write, void *data)
2985 if (*lvalp > INT_MAX / HZ)
2987 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2993 lval = -(unsigned long)val;
2996 lval = (unsigned long)val;
3003 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
3005 int write, void *data)
3008 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
3010 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
3016 lval = -(unsigned long)val;
3019 lval = (unsigned long)val;
3021 *lvalp = jiffies_to_clock_t(lval);
3026 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
3028 int write, void *data)
3031 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
3041 lval = -(unsigned long)val;
3044 lval = (unsigned long)val;
3046 *lvalp = jiffies_to_msecs(lval);
3052 * proc_dointvec_jiffies - read a vector of integers as seconds
3053 * @table: the sysctl table
3054 * @write: %TRUE if this is a write to the sysctl file
3055 * @buffer: the user buffer
3056 * @lenp: the size of the user buffer
3057 * @ppos: file position
3059 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3060 * values from/to the user buffer, treated as an ASCII string.
3061 * The values read are assumed to be in seconds, and are converted into
3064 * Returns 0 on success.
3066 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3067 void __user *buffer, size_t *lenp, loff_t *ppos)
3069 return do_proc_dointvec(table,write,buffer,lenp,ppos,
3070 do_proc_dointvec_jiffies_conv,NULL);
3074 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
3075 * @table: the sysctl table
3076 * @write: %TRUE if this is a write to the sysctl file
3077 * @buffer: the user buffer
3078 * @lenp: the size of the user buffer
3079 * @ppos: pointer to the file position
3081 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3082 * values from/to the user buffer, treated as an ASCII string.
3083 * The values read are assumed to be in 1/USER_HZ seconds, and
3084 * are converted into jiffies.
3086 * Returns 0 on success.
3088 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3089 void __user *buffer, size_t *lenp, loff_t *ppos)
3091 return do_proc_dointvec(table,write,buffer,lenp,ppos,
3092 do_proc_dointvec_userhz_jiffies_conv,NULL);
3096 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
3097 * @table: the sysctl table
3098 * @write: %TRUE if this is a write to the sysctl file
3099 * @buffer: the user buffer
3100 * @lenp: the size of the user buffer
3101 * @ppos: file position
3102 * @ppos: the current position in the file
3104 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3105 * values from/to the user buffer, treated as an ASCII string.
3106 * The values read are assumed to be in 1/1000 seconds, and
3107 * are converted into jiffies.
3109 * Returns 0 on success.
3111 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3112 void __user *buffer, size_t *lenp, loff_t *ppos)
3114 return do_proc_dointvec(table, write, buffer, lenp, ppos,
3115 do_proc_dointvec_ms_jiffies_conv, NULL);
3118 static int proc_do_cad_pid(struct ctl_table *table, int write,
3119 void __user *buffer, size_t *lenp, loff_t *ppos)
3121 struct pid *new_pid;
3125 tmp = pid_vnr(cad_pid);
3127 r = __do_proc_dointvec(&tmp, table, write, buffer,
3128 lenp, ppos, NULL, NULL);
3132 new_pid = find_get_pid(tmp);
3136 put_pid(xchg(&cad_pid, new_pid));
3141 * proc_do_large_bitmap - read/write from/to a large bitmap
3142 * @table: the sysctl table
3143 * @write: %TRUE if this is a write to the sysctl file
3144 * @buffer: the user buffer
3145 * @lenp: the size of the user buffer
3146 * @ppos: file position
3148 * The bitmap is stored at table->data and the bitmap length (in bits)
3151 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
3152 * large bitmaps may be represented in a compact manner. Writing into
3153 * the file will clear the bitmap then update it with the given input.
3155 * Returns 0 on success.
3157 int proc_do_large_bitmap(struct ctl_table *table, int write,
3158 void __user *buffer, size_t *lenp, loff_t *ppos)
3162 size_t left = *lenp;
3163 unsigned long bitmap_len = table->maxlen;
3164 unsigned long *bitmap = *(unsigned long **) table->data;
3165 unsigned long *tmp_bitmap = NULL;
3166 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
3168 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
3177 if (left > PAGE_SIZE - 1) {
3178 left = PAGE_SIZE - 1;
3179 /* How much of the buffer we'll skip this pass */
3180 skipped = *lenp - left;
3183 p = kbuf = memdup_user_nul(buffer, left);
3185 return PTR_ERR(kbuf);
3187 tmp_bitmap = bitmap_zalloc(bitmap_len, GFP_KERNEL);
3192 proc_skip_char(&p, &left, '\n');
3193 while (!err && left) {
3194 unsigned long val_a, val_b;
3198 /* In case we stop parsing mid-number, we can reset */
3200 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3203 * If we consumed the entirety of a truncated buffer or
3204 * only one char is left (may be a "-"), then stop here,
3205 * reset, & come back for more.
3207 if ((left <= 1) && skipped) {
3214 if (val_a >= bitmap_len || neg) {
3226 err = proc_get_long(&p, &left, &val_b,
3227 &neg, tr_b, sizeof(tr_b),
3230 * If we consumed all of a truncated buffer or
3231 * then stop here, reset, & come back for more.
3233 if (!left && skipped) {
3240 if (val_b >= bitmap_len || neg ||
3251 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3253 proc_skip_char(&p, &left, '\n');
3258 unsigned long bit_a, bit_b = 0;
3261 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3262 if (bit_a >= bitmap_len)
3264 bit_b = find_next_zero_bit(bitmap, bitmap_len,
3268 err = proc_put_char(&buffer, &left, ',');
3272 err = proc_put_long(&buffer, &left, bit_a, false);
3275 if (bit_a != bit_b) {
3276 err = proc_put_char(&buffer, &left, '-');
3279 err = proc_put_long(&buffer, &left, bit_b, false);
3287 err = proc_put_char(&buffer, &left, '\n');
3293 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3295 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3301 bitmap_free(tmp_bitmap);
3305 #else /* CONFIG_PROC_SYSCTL */
3307 int proc_dostring(struct ctl_table *table, int write,
3308 void __user *buffer, size_t *lenp, loff_t *ppos)
3313 int proc_dointvec(struct ctl_table *table, int write,
3314 void __user *buffer, size_t *lenp, loff_t *ppos)
3319 int proc_douintvec(struct ctl_table *table, int write,
3320 void __user *buffer, size_t *lenp, loff_t *ppos)
3325 int proc_dointvec_minmax(struct ctl_table *table, int write,
3326 void __user *buffer, size_t *lenp, loff_t *ppos)
3331 int proc_douintvec_minmax(struct ctl_table *table, int write,
3332 void __user *buffer, size_t *lenp, loff_t *ppos)
3337 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3338 void __user *buffer, size_t *lenp, loff_t *ppos)
3343 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3344 void __user *buffer, size_t *lenp, loff_t *ppos)
3349 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3350 void __user *buffer, size_t *lenp, loff_t *ppos)
3355 int proc_doulongvec_minmax(struct ctl_table *table, int write,
3356 void __user *buffer, size_t *lenp, loff_t *ppos)
3361 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3362 void __user *buffer,
3363 size_t *lenp, loff_t *ppos)
3368 int proc_do_large_bitmap(struct ctl_table *table, int write,
3369 void __user *buffer, size_t *lenp, loff_t *ppos)
3374 #endif /* CONFIG_PROC_SYSCTL */
3376 #if defined(CONFIG_SYSCTL)
3377 int proc_do_static_key(struct ctl_table *table, int write,
3378 void __user *buffer, size_t *lenp,
3381 struct static_key *key = (struct static_key *)table->data;
3382 static DEFINE_MUTEX(static_key_mutex);
3384 struct ctl_table tmp = {
3386 .maxlen = sizeof(val),
3387 .mode = table->mode,
3388 .extra1 = SYSCTL_ZERO,
3389 .extra2 = SYSCTL_ONE,
3392 if (write && !capable(CAP_SYS_ADMIN))
3395 mutex_lock(&static_key_mutex);
3396 val = static_key_enabled(key);
3397 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
3398 if (write && !ret) {
3400 static_key_enable(key);
3402 static_key_disable(key);
3404 mutex_unlock(&static_key_mutex);
3409 * No sense putting this after each symbol definition, twice,
3410 * exception granted :-)
3412 EXPORT_SYMBOL(proc_dointvec);
3413 EXPORT_SYMBOL(proc_douintvec);
3414 EXPORT_SYMBOL(proc_dointvec_jiffies);
3415 EXPORT_SYMBOL(proc_dointvec_minmax);
3416 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3417 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3418 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3419 EXPORT_SYMBOL(proc_dostring);
3420 EXPORT_SYMBOL(proc_doulongvec_minmax);
3421 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3422 EXPORT_SYMBOL(proc_do_large_bitmap);