2 * sysctl.c: General linux system control interface
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
21 #include <linux/module.h>
22 #include <linux/aio.h>
24 #include <linux/swap.h>
25 #include <linux/slab.h>
26 #include <linux/sysctl.h>
27 #include <linux/bitmap.h>
28 #include <linux/signal.h>
29 #include <linux/printk.h>
30 #include <linux/proc_fs.h>
31 #include <linux/security.h>
32 #include <linux/ctype.h>
33 #include <linux/kmemleak.h>
35 #include <linux/init.h>
36 #include <linux/kernel.h>
37 #include <linux/kobject.h>
38 #include <linux/net.h>
39 #include <linux/sysrq.h>
40 #include <linux/highuid.h>
41 #include <linux/writeback.h>
42 #include <linux/ratelimit.h>
43 #include <linux/compaction.h>
44 #include <linux/hugetlb.h>
45 #include <linux/initrd.h>
46 #include <linux/key.h>
47 #include <linux/times.h>
48 #include <linux/limits.h>
49 #include <linux/dcache.h>
50 #include <linux/dnotify.h>
51 #include <linux/syscalls.h>
52 #include <linux/vmstat.h>
53 #include <linux/nfs_fs.h>
54 #include <linux/acpi.h>
55 #include <linux/reboot.h>
56 #include <linux/ftrace.h>
57 #include <linux/perf_event.h>
58 #include <linux/kprobes.h>
59 #include <linux/pipe_fs_i.h>
60 #include <linux/oom.h>
61 #include <linux/kmod.h>
62 #include <linux/capability.h>
63 #include <linux/binfmts.h>
64 #include <linux/sched/sysctl.h>
65 #include <linux/sched/coredump.h>
66 #include <linux/kexec.h>
67 #include <linux/bpf.h>
68 #include <linux/mount.h>
70 #include <linux/uaccess.h>
71 #include <asm/processor.h>
75 #include <asm/stacktrace.h>
79 #include <asm/setup.h>
81 #ifdef CONFIG_BSD_PROCESS_ACCT
82 #include <linux/acct.h>
84 #ifdef CONFIG_RT_MUTEXES
85 #include <linux/rtmutex.h>
87 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
88 #include <linux/lockdep.h>
90 #ifdef CONFIG_CHR_DEV_SG
94 #ifdef CONFIG_LOCKUP_DETECTOR
95 #include <linux/nmi.h>
98 #if defined(CONFIG_SYSCTL)
100 /* External variables not in a header file. */
101 extern int suid_dumpable;
102 #ifdef CONFIG_COREDUMP
103 extern int core_uses_pid;
104 extern char core_pattern[];
105 extern unsigned int core_pipe_limit;
108 extern int pid_max_min, pid_max_max;
109 extern int percpu_pagelist_fraction;
110 extern int latencytop_enabled;
111 extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
113 extern int sysctl_nr_trim_pages;
116 /* Constants used for minimum and maximum */
117 #ifdef CONFIG_LOCKUP_DETECTOR
118 static int sixty = 60;
121 static int __maybe_unused neg_one = -1;
124 static int __maybe_unused one = 1;
125 static int __maybe_unused two = 2;
126 static int __maybe_unused four = 4;
127 static unsigned long one_ul = 1;
128 static int one_hundred = 100;
129 static int one_thousand = 1000;
131 static int ten_thousand = 10000;
133 #ifdef CONFIG_PERF_EVENTS
134 static int six_hundred_forty_kb = 640 * 1024;
137 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
138 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
140 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
141 static int maxolduid = 65535;
142 static int minolduid;
144 static int ngroups_max = NGROUPS_MAX;
145 static const int cap_last_cap = CAP_LAST_CAP;
147 /*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */
148 #ifdef CONFIG_DETECT_HUNG_TASK
149 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
152 #ifdef CONFIG_INOTIFY_USER
153 #include <linux/inotify.h>
159 extern int pwrsw_enabled;
162 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
163 extern int unaligned_enabled;
167 extern int unaligned_dump_stack;
170 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
171 extern int no_unaligned_warning;
174 #ifdef CONFIG_PROC_SYSCTL
177 * enum sysctl_writes_mode - supported sysctl write modes
179 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
180 * to be written, and multiple writes on the same sysctl file descriptor
181 * will rewrite the sysctl value, regardless of file position. No warning
182 * is issued when the initial position is not 0.
183 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
185 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
186 * file position 0 and the value must be fully contained in the buffer
187 * sent to the write syscall. If dealing with strings respect the file
188 * position, but restrict this to the max length of the buffer, anything
189 * passed the max lenght will be ignored. Multiple writes will append
192 * These write modes control how current file position affects the behavior of
193 * updating sysctl values through the proc interface on each write.
195 enum sysctl_writes_mode {
196 SYSCTL_WRITES_LEGACY = -1,
197 SYSCTL_WRITES_WARN = 0,
198 SYSCTL_WRITES_STRICT = 1,
201 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
203 static int proc_do_cad_pid(struct ctl_table *table, int write,
204 void __user *buffer, size_t *lenp, loff_t *ppos);
205 static int proc_taint(struct ctl_table *table, int write,
206 void __user *buffer, size_t *lenp, loff_t *ppos);
210 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
211 void __user *buffer, size_t *lenp, loff_t *ppos);
214 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
215 void __user *buffer, size_t *lenp, loff_t *ppos);
216 #ifdef CONFIG_COREDUMP
217 static int proc_dostring_coredump(struct ctl_table *table, int write,
218 void __user *buffer, size_t *lenp, loff_t *ppos);
221 #ifdef CONFIG_MAGIC_SYSRQ
222 /* Note: sysrq code uses it's own private copy */
223 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
225 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
226 void __user *buffer, size_t *lenp,
231 error = proc_dointvec(table, write, buffer, lenp, ppos);
236 sysrq_toggle_support(__sysrq_enabled);
243 static struct ctl_table kern_table[];
244 static struct ctl_table vm_table[];
245 static struct ctl_table fs_table[];
246 static struct ctl_table debug_table[];
247 static struct ctl_table dev_table[];
248 extern struct ctl_table random_table[];
250 extern struct ctl_table epoll_table[];
253 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
254 int sysctl_legacy_va_layout;
257 /* The default sysctl tables: */
259 static struct ctl_table sysctl_base_table[] = {
261 .procname = "kernel",
278 .child = debug_table,
288 #ifdef CONFIG_SCHED_DEBUG
289 static int min_sched_granularity_ns = 100000; /* 100 usecs */
290 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
291 static int min_wakeup_granularity_ns; /* 0 usecs */
292 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
294 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
295 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
296 #endif /* CONFIG_SMP */
297 #endif /* CONFIG_SCHED_DEBUG */
299 #ifdef CONFIG_COMPACTION
300 static int min_extfrag_threshold;
301 static int max_extfrag_threshold = 1000;
304 static struct ctl_table kern_table[] = {
306 .procname = "sched_child_runs_first",
307 .data = &sysctl_sched_child_runs_first,
308 .maxlen = sizeof(unsigned int),
310 .proc_handler = proc_dointvec,
312 #ifdef CONFIG_SCHED_DEBUG
314 .procname = "sched_min_granularity_ns",
315 .data = &sysctl_sched_min_granularity,
316 .maxlen = sizeof(unsigned int),
318 .proc_handler = sched_proc_update_handler,
319 .extra1 = &min_sched_granularity_ns,
320 .extra2 = &max_sched_granularity_ns,
323 .procname = "sched_latency_ns",
324 .data = &sysctl_sched_latency,
325 .maxlen = sizeof(unsigned int),
327 .proc_handler = sched_proc_update_handler,
328 .extra1 = &min_sched_granularity_ns,
329 .extra2 = &max_sched_granularity_ns,
332 .procname = "sched_wakeup_granularity_ns",
333 .data = &sysctl_sched_wakeup_granularity,
334 .maxlen = sizeof(unsigned int),
336 .proc_handler = sched_proc_update_handler,
337 .extra1 = &min_wakeup_granularity_ns,
338 .extra2 = &max_wakeup_granularity_ns,
342 .procname = "sched_tunable_scaling",
343 .data = &sysctl_sched_tunable_scaling,
344 .maxlen = sizeof(enum sched_tunable_scaling),
346 .proc_handler = sched_proc_update_handler,
347 .extra1 = &min_sched_tunable_scaling,
348 .extra2 = &max_sched_tunable_scaling,
351 .procname = "sched_migration_cost_ns",
352 .data = &sysctl_sched_migration_cost,
353 .maxlen = sizeof(unsigned int),
355 .proc_handler = proc_dointvec,
358 .procname = "sched_nr_migrate",
359 .data = &sysctl_sched_nr_migrate,
360 .maxlen = sizeof(unsigned int),
362 .proc_handler = proc_dointvec,
365 .procname = "sched_time_avg_ms",
366 .data = &sysctl_sched_time_avg,
367 .maxlen = sizeof(unsigned int),
369 .proc_handler = proc_dointvec_minmax,
372 #ifdef CONFIG_SCHEDSTATS
374 .procname = "sched_schedstats",
376 .maxlen = sizeof(unsigned int),
378 .proc_handler = sysctl_schedstats,
382 #endif /* CONFIG_SCHEDSTATS */
383 #endif /* CONFIG_SMP */
384 #ifdef CONFIG_NUMA_BALANCING
386 .procname = "numa_balancing_scan_delay_ms",
387 .data = &sysctl_numa_balancing_scan_delay,
388 .maxlen = sizeof(unsigned int),
390 .proc_handler = proc_dointvec,
393 .procname = "numa_balancing_scan_period_min_ms",
394 .data = &sysctl_numa_balancing_scan_period_min,
395 .maxlen = sizeof(unsigned int),
397 .proc_handler = proc_dointvec,
400 .procname = "numa_balancing_scan_period_max_ms",
401 .data = &sysctl_numa_balancing_scan_period_max,
402 .maxlen = sizeof(unsigned int),
404 .proc_handler = proc_dointvec,
407 .procname = "numa_balancing_scan_size_mb",
408 .data = &sysctl_numa_balancing_scan_size,
409 .maxlen = sizeof(unsigned int),
411 .proc_handler = proc_dointvec_minmax,
415 .procname = "numa_balancing",
416 .data = NULL, /* filled in by handler */
417 .maxlen = sizeof(unsigned int),
419 .proc_handler = sysctl_numa_balancing,
423 #endif /* CONFIG_NUMA_BALANCING */
424 #endif /* CONFIG_SCHED_DEBUG */
426 .procname = "sched_rt_period_us",
427 .data = &sysctl_sched_rt_period,
428 .maxlen = sizeof(unsigned int),
430 .proc_handler = sched_rt_handler,
433 .procname = "sched_rt_runtime_us",
434 .data = &sysctl_sched_rt_runtime,
435 .maxlen = sizeof(int),
437 .proc_handler = sched_rt_handler,
440 .procname = "sched_rr_timeslice_ms",
441 .data = &sysctl_sched_rr_timeslice,
442 .maxlen = sizeof(int),
444 .proc_handler = sched_rr_handler,
446 #ifdef CONFIG_SCHED_AUTOGROUP
448 .procname = "sched_autogroup_enabled",
449 .data = &sysctl_sched_autogroup_enabled,
450 .maxlen = sizeof(unsigned int),
452 .proc_handler = proc_dointvec_minmax,
457 #ifdef CONFIG_CFS_BANDWIDTH
459 .procname = "sched_cfs_bandwidth_slice_us",
460 .data = &sysctl_sched_cfs_bandwidth_slice,
461 .maxlen = sizeof(unsigned int),
463 .proc_handler = proc_dointvec_minmax,
467 #ifdef CONFIG_PROVE_LOCKING
469 .procname = "prove_locking",
470 .data = &prove_locking,
471 .maxlen = sizeof(int),
473 .proc_handler = proc_dointvec,
476 #ifdef CONFIG_LOCK_STAT
478 .procname = "lock_stat",
480 .maxlen = sizeof(int),
482 .proc_handler = proc_dointvec,
487 .data = &panic_timeout,
488 .maxlen = sizeof(int),
490 .proc_handler = proc_dointvec,
492 #ifdef CONFIG_COREDUMP
494 .procname = "core_uses_pid",
495 .data = &core_uses_pid,
496 .maxlen = sizeof(int),
498 .proc_handler = proc_dointvec,
501 .procname = "core_pattern",
502 .data = core_pattern,
503 .maxlen = CORENAME_MAX_SIZE,
505 .proc_handler = proc_dostring_coredump,
508 .procname = "core_pipe_limit",
509 .data = &core_pipe_limit,
510 .maxlen = sizeof(unsigned int),
512 .proc_handler = proc_dointvec,
515 #ifdef CONFIG_PROC_SYSCTL
517 .procname = "tainted",
518 .maxlen = sizeof(long),
520 .proc_handler = proc_taint,
523 .procname = "sysctl_writes_strict",
524 .data = &sysctl_writes_strict,
525 .maxlen = sizeof(int),
527 .proc_handler = proc_dointvec_minmax,
532 #ifdef CONFIG_LATENCYTOP
534 .procname = "latencytop",
535 .data = &latencytop_enabled,
536 .maxlen = sizeof(int),
538 .proc_handler = sysctl_latencytop,
541 #ifdef CONFIG_BLK_DEV_INITRD
543 .procname = "real-root-dev",
544 .data = &real_root_dev,
545 .maxlen = sizeof(int),
547 .proc_handler = proc_dointvec,
551 .procname = "print-fatal-signals",
552 .data = &print_fatal_signals,
553 .maxlen = sizeof(int),
555 .proc_handler = proc_dointvec,
559 .procname = "reboot-cmd",
560 .data = reboot_command,
563 .proc_handler = proc_dostring,
566 .procname = "stop-a",
567 .data = &stop_a_enabled,
568 .maxlen = sizeof (int),
570 .proc_handler = proc_dointvec,
573 .procname = "scons-poweroff",
574 .data = &scons_pwroff,
575 .maxlen = sizeof (int),
577 .proc_handler = proc_dointvec,
580 #ifdef CONFIG_SPARC64
582 .procname = "tsb-ratio",
583 .data = &sysctl_tsb_ratio,
584 .maxlen = sizeof (int),
586 .proc_handler = proc_dointvec,
591 .procname = "soft-power",
592 .data = &pwrsw_enabled,
593 .maxlen = sizeof (int),
595 .proc_handler = proc_dointvec,
598 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
600 .procname = "unaligned-trap",
601 .data = &unaligned_enabled,
602 .maxlen = sizeof (int),
604 .proc_handler = proc_dointvec,
608 .procname = "ctrl-alt-del",
610 .maxlen = sizeof(int),
612 .proc_handler = proc_dointvec,
614 #ifdef CONFIG_FUNCTION_TRACER
616 .procname = "ftrace_enabled",
617 .data = &ftrace_enabled,
618 .maxlen = sizeof(int),
620 .proc_handler = ftrace_enable_sysctl,
623 #ifdef CONFIG_STACK_TRACER
625 .procname = "stack_tracer_enabled",
626 .data = &stack_tracer_enabled,
627 .maxlen = sizeof(int),
629 .proc_handler = stack_trace_sysctl,
632 #ifdef CONFIG_TRACING
634 .procname = "ftrace_dump_on_oops",
635 .data = &ftrace_dump_on_oops,
636 .maxlen = sizeof(int),
638 .proc_handler = proc_dointvec,
641 .procname = "traceoff_on_warning",
642 .data = &__disable_trace_on_warning,
643 .maxlen = sizeof(__disable_trace_on_warning),
645 .proc_handler = proc_dointvec,
648 .procname = "tracepoint_printk",
649 .data = &tracepoint_printk,
650 .maxlen = sizeof(tracepoint_printk),
652 .proc_handler = tracepoint_printk_sysctl,
655 #ifdef CONFIG_KEXEC_CORE
657 .procname = "kexec_load_disabled",
658 .data = &kexec_load_disabled,
659 .maxlen = sizeof(int),
661 /* only handle a transition from default "0" to "1" */
662 .proc_handler = proc_dointvec_minmax,
667 #ifdef CONFIG_MODULES
669 .procname = "modprobe",
670 .data = &modprobe_path,
671 .maxlen = KMOD_PATH_LEN,
673 .proc_handler = proc_dostring,
676 .procname = "modules_disabled",
677 .data = &modules_disabled,
678 .maxlen = sizeof(int),
680 /* only handle a transition from default "0" to "1" */
681 .proc_handler = proc_dointvec_minmax,
686 #ifdef CONFIG_UEVENT_HELPER
688 .procname = "hotplug",
689 .data = &uevent_helper,
690 .maxlen = UEVENT_HELPER_PATH_LEN,
692 .proc_handler = proc_dostring,
695 #ifdef CONFIG_CHR_DEV_SG
697 .procname = "sg-big-buff",
698 .data = &sg_big_buff,
699 .maxlen = sizeof (int),
701 .proc_handler = proc_dointvec,
704 #ifdef CONFIG_BSD_PROCESS_ACCT
708 .maxlen = 3*sizeof(int),
710 .proc_handler = proc_dointvec,
713 #ifdef CONFIG_MAGIC_SYSRQ
716 .data = &__sysrq_enabled,
717 .maxlen = sizeof (int),
719 .proc_handler = sysrq_sysctl_handler,
722 #ifdef CONFIG_PROC_SYSCTL
724 .procname = "cad_pid",
726 .maxlen = sizeof (int),
728 .proc_handler = proc_do_cad_pid,
732 .procname = "threads-max",
734 .maxlen = sizeof(int),
736 .proc_handler = sysctl_max_threads,
739 .procname = "random",
741 .child = random_table,
744 .procname = "usermodehelper",
746 .child = usermodehelper_table,
749 .procname = "overflowuid",
750 .data = &overflowuid,
751 .maxlen = sizeof(int),
753 .proc_handler = proc_dointvec_minmax,
754 .extra1 = &minolduid,
755 .extra2 = &maxolduid,
758 .procname = "overflowgid",
759 .data = &overflowgid,
760 .maxlen = sizeof(int),
762 .proc_handler = proc_dointvec_minmax,
763 .extra1 = &minolduid,
764 .extra2 = &maxolduid,
767 #ifdef CONFIG_MATHEMU
769 .procname = "ieee_emulation_warnings",
770 .data = &sysctl_ieee_emulation_warnings,
771 .maxlen = sizeof(int),
773 .proc_handler = proc_dointvec,
777 .procname = "userprocess_debug",
778 .data = &show_unhandled_signals,
779 .maxlen = sizeof(int),
781 .proc_handler = proc_dointvec,
785 .procname = "pid_max",
787 .maxlen = sizeof (int),
789 .proc_handler = proc_dointvec_minmax,
790 .extra1 = &pid_max_min,
791 .extra2 = &pid_max_max,
794 .procname = "panic_on_oops",
795 .data = &panic_on_oops,
796 .maxlen = sizeof(int),
798 .proc_handler = proc_dointvec,
800 #if defined CONFIG_PRINTK
802 .procname = "printk",
803 .data = &console_loglevel,
804 .maxlen = 4*sizeof(int),
806 .proc_handler = proc_dointvec,
809 .procname = "printk_ratelimit",
810 .data = &printk_ratelimit_state.interval,
811 .maxlen = sizeof(int),
813 .proc_handler = proc_dointvec_jiffies,
816 .procname = "printk_ratelimit_burst",
817 .data = &printk_ratelimit_state.burst,
818 .maxlen = sizeof(int),
820 .proc_handler = proc_dointvec,
823 .procname = "printk_delay",
824 .data = &printk_delay_msec,
825 .maxlen = sizeof(int),
827 .proc_handler = proc_dointvec_minmax,
829 .extra2 = &ten_thousand,
832 .procname = "printk_devkmsg",
833 .data = devkmsg_log_str,
834 .maxlen = DEVKMSG_STR_MAX_SIZE,
836 .proc_handler = devkmsg_sysctl_set_loglvl,
839 .procname = "dmesg_restrict",
840 .data = &dmesg_restrict,
841 .maxlen = sizeof(int),
843 .proc_handler = proc_dointvec_minmax_sysadmin,
848 .procname = "kptr_restrict",
849 .data = &kptr_restrict,
850 .maxlen = sizeof(int),
852 .proc_handler = proc_dointvec_minmax_sysadmin,
858 .procname = "ngroups_max",
859 .data = &ngroups_max,
860 .maxlen = sizeof (int),
862 .proc_handler = proc_dointvec,
865 .procname = "cap_last_cap",
866 .data = (void *)&cap_last_cap,
867 .maxlen = sizeof(int),
869 .proc_handler = proc_dointvec,
871 #if defined(CONFIG_LOCKUP_DETECTOR)
873 .procname = "watchdog",
874 .data = &watchdog_user_enabled,
875 .maxlen = sizeof(int),
877 .proc_handler = proc_watchdog,
882 .procname = "watchdog_thresh",
883 .data = &watchdog_thresh,
884 .maxlen = sizeof(int),
886 .proc_handler = proc_watchdog_thresh,
891 .procname = "nmi_watchdog",
892 .data = &nmi_watchdog_user_enabled,
893 .maxlen = sizeof(int),
894 .mode = NMI_WATCHDOG_SYSCTL_PERM,
895 .proc_handler = proc_nmi_watchdog,
900 .procname = "watchdog_cpumask",
901 .data = &watchdog_cpumask_bits,
904 .proc_handler = proc_watchdog_cpumask,
906 #ifdef CONFIG_SOFTLOCKUP_DETECTOR
908 .procname = "soft_watchdog",
909 .data = &soft_watchdog_user_enabled,
910 .maxlen = sizeof(int),
912 .proc_handler = proc_soft_watchdog,
917 .procname = "softlockup_panic",
918 .data = &softlockup_panic,
919 .maxlen = sizeof(int),
921 .proc_handler = proc_dointvec_minmax,
927 .procname = "softlockup_all_cpu_backtrace",
928 .data = &sysctl_softlockup_all_cpu_backtrace,
929 .maxlen = sizeof(int),
931 .proc_handler = proc_dointvec_minmax,
935 #endif /* CONFIG_SMP */
937 #ifdef CONFIG_HARDLOCKUP_DETECTOR
939 .procname = "hardlockup_panic",
940 .data = &hardlockup_panic,
941 .maxlen = sizeof(int),
943 .proc_handler = proc_dointvec_minmax,
949 .procname = "hardlockup_all_cpu_backtrace",
950 .data = &sysctl_hardlockup_all_cpu_backtrace,
951 .maxlen = sizeof(int),
953 .proc_handler = proc_dointvec_minmax,
957 #endif /* CONFIG_SMP */
961 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
963 .procname = "unknown_nmi_panic",
964 .data = &unknown_nmi_panic,
965 .maxlen = sizeof (int),
967 .proc_handler = proc_dointvec,
970 #if defined(CONFIG_X86)
972 .procname = "panic_on_unrecovered_nmi",
973 .data = &panic_on_unrecovered_nmi,
974 .maxlen = sizeof(int),
976 .proc_handler = proc_dointvec,
979 .procname = "panic_on_io_nmi",
980 .data = &panic_on_io_nmi,
981 .maxlen = sizeof(int),
983 .proc_handler = proc_dointvec,
985 #ifdef CONFIG_DEBUG_STACKOVERFLOW
987 .procname = "panic_on_stackoverflow",
988 .data = &sysctl_panic_on_stackoverflow,
989 .maxlen = sizeof(int),
991 .proc_handler = proc_dointvec,
995 .procname = "bootloader_type",
996 .data = &bootloader_type,
997 .maxlen = sizeof (int),
999 .proc_handler = proc_dointvec,
1002 .procname = "bootloader_version",
1003 .data = &bootloader_version,
1004 .maxlen = sizeof (int),
1006 .proc_handler = proc_dointvec,
1009 .procname = "io_delay_type",
1010 .data = &io_delay_type,
1011 .maxlen = sizeof(int),
1013 .proc_handler = proc_dointvec,
1016 #if defined(CONFIG_MMU)
1018 .procname = "randomize_va_space",
1019 .data = &randomize_va_space,
1020 .maxlen = sizeof(int),
1022 .proc_handler = proc_dointvec,
1025 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1027 .procname = "spin_retry",
1028 .data = &spin_retry,
1029 .maxlen = sizeof (int),
1031 .proc_handler = proc_dointvec,
1034 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1036 .procname = "acpi_video_flags",
1037 .data = &acpi_realmode_flags,
1038 .maxlen = sizeof (unsigned long),
1040 .proc_handler = proc_doulongvec_minmax,
1043 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1045 .procname = "ignore-unaligned-usertrap",
1046 .data = &no_unaligned_warning,
1047 .maxlen = sizeof (int),
1049 .proc_handler = proc_dointvec,
1054 .procname = "unaligned-dump-stack",
1055 .data = &unaligned_dump_stack,
1056 .maxlen = sizeof (int),
1058 .proc_handler = proc_dointvec,
1061 #ifdef CONFIG_DETECT_HUNG_TASK
1063 .procname = "hung_task_panic",
1064 .data = &sysctl_hung_task_panic,
1065 .maxlen = sizeof(int),
1067 .proc_handler = proc_dointvec_minmax,
1072 .procname = "hung_task_check_count",
1073 .data = &sysctl_hung_task_check_count,
1074 .maxlen = sizeof(int),
1076 .proc_handler = proc_dointvec_minmax,
1080 .procname = "hung_task_timeout_secs",
1081 .data = &sysctl_hung_task_timeout_secs,
1082 .maxlen = sizeof(unsigned long),
1084 .proc_handler = proc_dohung_task_timeout_secs,
1085 .extra2 = &hung_task_timeout_max,
1088 .procname = "hung_task_warnings",
1089 .data = &sysctl_hung_task_warnings,
1090 .maxlen = sizeof(int),
1092 .proc_handler = proc_dointvec_minmax,
1096 #ifdef CONFIG_RT_MUTEXES
1098 .procname = "max_lock_depth",
1099 .data = &max_lock_depth,
1100 .maxlen = sizeof(int),
1102 .proc_handler = proc_dointvec,
1106 .procname = "poweroff_cmd",
1107 .data = &poweroff_cmd,
1108 .maxlen = POWEROFF_CMD_PATH_LEN,
1110 .proc_handler = proc_dostring,
1116 .child = key_sysctls,
1119 #ifdef CONFIG_PERF_EVENTS
1121 * User-space scripts rely on the existence of this file
1122 * as a feature check for perf_events being enabled.
1124 * So it's an ABI, do not remove!
1127 .procname = "perf_event_paranoid",
1128 .data = &sysctl_perf_event_paranoid,
1129 .maxlen = sizeof(sysctl_perf_event_paranoid),
1131 .proc_handler = proc_dointvec,
1134 .procname = "perf_event_mlock_kb",
1135 .data = &sysctl_perf_event_mlock,
1136 .maxlen = sizeof(sysctl_perf_event_mlock),
1138 .proc_handler = proc_dointvec,
1141 .procname = "perf_event_max_sample_rate",
1142 .data = &sysctl_perf_event_sample_rate,
1143 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1145 .proc_handler = perf_proc_update_handler,
1149 .procname = "perf_cpu_time_max_percent",
1150 .data = &sysctl_perf_cpu_time_max_percent,
1151 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1153 .proc_handler = perf_cpu_time_max_percent_handler,
1155 .extra2 = &one_hundred,
1158 .procname = "perf_event_max_stack",
1159 .data = &sysctl_perf_event_max_stack,
1160 .maxlen = sizeof(sysctl_perf_event_max_stack),
1162 .proc_handler = perf_event_max_stack_handler,
1164 .extra2 = &six_hundred_forty_kb,
1167 .procname = "perf_event_max_contexts_per_stack",
1168 .data = &sysctl_perf_event_max_contexts_per_stack,
1169 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
1171 .proc_handler = perf_event_max_stack_handler,
1173 .extra2 = &one_thousand,
1177 .procname = "panic_on_warn",
1178 .data = &panic_on_warn,
1179 .maxlen = sizeof(int),
1181 .proc_handler = proc_dointvec_minmax,
1185 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1187 .procname = "timer_migration",
1188 .data = &sysctl_timer_migration,
1189 .maxlen = sizeof(unsigned int),
1191 .proc_handler = timer_migration_handler,
1196 #ifdef CONFIG_BPF_SYSCALL
1198 .procname = "unprivileged_bpf_disabled",
1199 .data = &sysctl_unprivileged_bpf_disabled,
1200 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
1202 /* only handle a transition from default "0" to "1" */
1203 .proc_handler = proc_dointvec_minmax,
1208 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1210 .procname = "panic_on_rcu_stall",
1211 .data = &sysctl_panic_on_rcu_stall,
1212 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
1214 .proc_handler = proc_dointvec_minmax,
1222 static struct ctl_table vm_table[] = {
1224 .procname = "overcommit_memory",
1225 .data = &sysctl_overcommit_memory,
1226 .maxlen = sizeof(sysctl_overcommit_memory),
1228 .proc_handler = proc_dointvec_minmax,
1233 .procname = "panic_on_oom",
1234 .data = &sysctl_panic_on_oom,
1235 .maxlen = sizeof(sysctl_panic_on_oom),
1237 .proc_handler = proc_dointvec_minmax,
1242 .procname = "oom_kill_allocating_task",
1243 .data = &sysctl_oom_kill_allocating_task,
1244 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1246 .proc_handler = proc_dointvec,
1249 .procname = "oom_dump_tasks",
1250 .data = &sysctl_oom_dump_tasks,
1251 .maxlen = sizeof(sysctl_oom_dump_tasks),
1253 .proc_handler = proc_dointvec,
1256 .procname = "overcommit_ratio",
1257 .data = &sysctl_overcommit_ratio,
1258 .maxlen = sizeof(sysctl_overcommit_ratio),
1260 .proc_handler = overcommit_ratio_handler,
1263 .procname = "overcommit_kbytes",
1264 .data = &sysctl_overcommit_kbytes,
1265 .maxlen = sizeof(sysctl_overcommit_kbytes),
1267 .proc_handler = overcommit_kbytes_handler,
1270 .procname = "page-cluster",
1271 .data = &page_cluster,
1272 .maxlen = sizeof(int),
1274 .proc_handler = proc_dointvec_minmax,
1278 .procname = "dirty_background_ratio",
1279 .data = &dirty_background_ratio,
1280 .maxlen = sizeof(dirty_background_ratio),
1282 .proc_handler = dirty_background_ratio_handler,
1284 .extra2 = &one_hundred,
1287 .procname = "dirty_background_bytes",
1288 .data = &dirty_background_bytes,
1289 .maxlen = sizeof(dirty_background_bytes),
1291 .proc_handler = dirty_background_bytes_handler,
1295 .procname = "dirty_ratio",
1296 .data = &vm_dirty_ratio,
1297 .maxlen = sizeof(vm_dirty_ratio),
1299 .proc_handler = dirty_ratio_handler,
1301 .extra2 = &one_hundred,
1304 .procname = "dirty_bytes",
1305 .data = &vm_dirty_bytes,
1306 .maxlen = sizeof(vm_dirty_bytes),
1308 .proc_handler = dirty_bytes_handler,
1309 .extra1 = &dirty_bytes_min,
1312 .procname = "dirty_writeback_centisecs",
1313 .data = &dirty_writeback_interval,
1314 .maxlen = sizeof(dirty_writeback_interval),
1316 .proc_handler = dirty_writeback_centisecs_handler,
1319 .procname = "dirty_expire_centisecs",
1320 .data = &dirty_expire_interval,
1321 .maxlen = sizeof(dirty_expire_interval),
1323 .proc_handler = proc_dointvec_minmax,
1327 .procname = "dirtytime_expire_seconds",
1328 .data = &dirtytime_expire_interval,
1329 .maxlen = sizeof(dirty_expire_interval),
1331 .proc_handler = dirtytime_interval_handler,
1335 .procname = "swappiness",
1336 .data = &vm_swappiness,
1337 .maxlen = sizeof(vm_swappiness),
1339 .proc_handler = proc_dointvec_minmax,
1341 .extra2 = &one_hundred,
1343 #ifdef CONFIG_HUGETLB_PAGE
1345 .procname = "nr_hugepages",
1347 .maxlen = sizeof(unsigned long),
1349 .proc_handler = hugetlb_sysctl_handler,
1353 .procname = "nr_hugepages_mempolicy",
1355 .maxlen = sizeof(unsigned long),
1357 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1360 .procname = "numa_stat",
1361 .data = &sysctl_vm_numa_stat,
1362 .maxlen = sizeof(int),
1364 .proc_handler = sysctl_vm_numa_stat_handler,
1370 .procname = "hugetlb_shm_group",
1371 .data = &sysctl_hugetlb_shm_group,
1372 .maxlen = sizeof(gid_t),
1374 .proc_handler = proc_dointvec,
1377 .procname = "hugepages_treat_as_movable",
1378 .data = &hugepages_treat_as_movable,
1379 .maxlen = sizeof(int),
1381 .proc_handler = proc_dointvec,
1384 .procname = "nr_overcommit_hugepages",
1386 .maxlen = sizeof(unsigned long),
1388 .proc_handler = hugetlb_overcommit_handler,
1392 .procname = "lowmem_reserve_ratio",
1393 .data = &sysctl_lowmem_reserve_ratio,
1394 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1396 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1399 .procname = "drop_caches",
1400 .data = &sysctl_drop_caches,
1401 .maxlen = sizeof(int),
1403 .proc_handler = drop_caches_sysctl_handler,
1407 #ifdef CONFIG_COMPACTION
1409 .procname = "compact_memory",
1410 .data = &sysctl_compact_memory,
1411 .maxlen = sizeof(int),
1413 .proc_handler = sysctl_compaction_handler,
1416 .procname = "extfrag_threshold",
1417 .data = &sysctl_extfrag_threshold,
1418 .maxlen = sizeof(int),
1420 .proc_handler = sysctl_extfrag_handler,
1421 .extra1 = &min_extfrag_threshold,
1422 .extra2 = &max_extfrag_threshold,
1425 .procname = "compact_unevictable_allowed",
1426 .data = &sysctl_compact_unevictable_allowed,
1427 .maxlen = sizeof(int),
1429 .proc_handler = proc_dointvec,
1434 #endif /* CONFIG_COMPACTION */
1436 .procname = "min_free_kbytes",
1437 .data = &min_free_kbytes,
1438 .maxlen = sizeof(min_free_kbytes),
1440 .proc_handler = min_free_kbytes_sysctl_handler,
1444 .procname = "watermark_scale_factor",
1445 .data = &watermark_scale_factor,
1446 .maxlen = sizeof(watermark_scale_factor),
1448 .proc_handler = watermark_scale_factor_sysctl_handler,
1450 .extra2 = &one_thousand,
1453 .procname = "percpu_pagelist_fraction",
1454 .data = &percpu_pagelist_fraction,
1455 .maxlen = sizeof(percpu_pagelist_fraction),
1457 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1462 .procname = "max_map_count",
1463 .data = &sysctl_max_map_count,
1464 .maxlen = sizeof(sysctl_max_map_count),
1466 .proc_handler = proc_dointvec_minmax,
1471 .procname = "nr_trim_pages",
1472 .data = &sysctl_nr_trim_pages,
1473 .maxlen = sizeof(sysctl_nr_trim_pages),
1475 .proc_handler = proc_dointvec_minmax,
1480 .procname = "laptop_mode",
1481 .data = &laptop_mode,
1482 .maxlen = sizeof(laptop_mode),
1484 .proc_handler = proc_dointvec_jiffies,
1487 .procname = "block_dump",
1488 .data = &block_dump,
1489 .maxlen = sizeof(block_dump),
1491 .proc_handler = proc_dointvec,
1495 .procname = "vfs_cache_pressure",
1496 .data = &sysctl_vfs_cache_pressure,
1497 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1499 .proc_handler = proc_dointvec,
1502 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1504 .procname = "legacy_va_layout",
1505 .data = &sysctl_legacy_va_layout,
1506 .maxlen = sizeof(sysctl_legacy_va_layout),
1508 .proc_handler = proc_dointvec,
1514 .procname = "zone_reclaim_mode",
1515 .data = &node_reclaim_mode,
1516 .maxlen = sizeof(node_reclaim_mode),
1518 .proc_handler = proc_dointvec,
1522 .procname = "min_unmapped_ratio",
1523 .data = &sysctl_min_unmapped_ratio,
1524 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1526 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1528 .extra2 = &one_hundred,
1531 .procname = "min_slab_ratio",
1532 .data = &sysctl_min_slab_ratio,
1533 .maxlen = sizeof(sysctl_min_slab_ratio),
1535 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1537 .extra2 = &one_hundred,
1542 .procname = "stat_interval",
1543 .data = &sysctl_stat_interval,
1544 .maxlen = sizeof(sysctl_stat_interval),
1546 .proc_handler = proc_dointvec_jiffies,
1549 .procname = "stat_refresh",
1553 .proc_handler = vmstat_refresh,
1558 .procname = "mmap_min_addr",
1559 .data = &dac_mmap_min_addr,
1560 .maxlen = sizeof(unsigned long),
1562 .proc_handler = mmap_min_addr_handler,
1567 .procname = "numa_zonelist_order",
1568 .data = &numa_zonelist_order,
1569 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1571 .proc_handler = numa_zonelist_order_handler,
1574 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1575 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1577 .procname = "vdso_enabled",
1578 #ifdef CONFIG_X86_32
1579 .data = &vdso32_enabled,
1580 .maxlen = sizeof(vdso32_enabled),
1582 .data = &vdso_enabled,
1583 .maxlen = sizeof(vdso_enabled),
1586 .proc_handler = proc_dointvec,
1590 #ifdef CONFIG_HIGHMEM
1592 .procname = "highmem_is_dirtyable",
1593 .data = &vm_highmem_is_dirtyable,
1594 .maxlen = sizeof(vm_highmem_is_dirtyable),
1596 .proc_handler = proc_dointvec_minmax,
1601 #ifdef CONFIG_MEMORY_FAILURE
1603 .procname = "memory_failure_early_kill",
1604 .data = &sysctl_memory_failure_early_kill,
1605 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1607 .proc_handler = proc_dointvec_minmax,
1612 .procname = "memory_failure_recovery",
1613 .data = &sysctl_memory_failure_recovery,
1614 .maxlen = sizeof(sysctl_memory_failure_recovery),
1616 .proc_handler = proc_dointvec_minmax,
1622 .procname = "user_reserve_kbytes",
1623 .data = &sysctl_user_reserve_kbytes,
1624 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1626 .proc_handler = proc_doulongvec_minmax,
1629 .procname = "admin_reserve_kbytes",
1630 .data = &sysctl_admin_reserve_kbytes,
1631 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1633 .proc_handler = proc_doulongvec_minmax,
1635 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1637 .procname = "mmap_rnd_bits",
1638 .data = &mmap_rnd_bits,
1639 .maxlen = sizeof(mmap_rnd_bits),
1641 .proc_handler = proc_dointvec_minmax,
1642 .extra1 = (void *)&mmap_rnd_bits_min,
1643 .extra2 = (void *)&mmap_rnd_bits_max,
1646 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1648 .procname = "mmap_rnd_compat_bits",
1649 .data = &mmap_rnd_compat_bits,
1650 .maxlen = sizeof(mmap_rnd_compat_bits),
1652 .proc_handler = proc_dointvec_minmax,
1653 .extra1 = (void *)&mmap_rnd_compat_bits_min,
1654 .extra2 = (void *)&mmap_rnd_compat_bits_max,
1660 static struct ctl_table fs_table[] = {
1662 .procname = "inode-nr",
1663 .data = &inodes_stat,
1664 .maxlen = 2*sizeof(long),
1666 .proc_handler = proc_nr_inodes,
1669 .procname = "inode-state",
1670 .data = &inodes_stat,
1671 .maxlen = 7*sizeof(long),
1673 .proc_handler = proc_nr_inodes,
1676 .procname = "file-nr",
1677 .data = &files_stat,
1678 .maxlen = sizeof(files_stat),
1680 .proc_handler = proc_nr_files,
1683 .procname = "file-max",
1684 .data = &files_stat.max_files,
1685 .maxlen = sizeof(files_stat.max_files),
1687 .proc_handler = proc_doulongvec_minmax,
1690 .procname = "nr_open",
1691 .data = &sysctl_nr_open,
1692 .maxlen = sizeof(unsigned int),
1694 .proc_handler = proc_dointvec_minmax,
1695 .extra1 = &sysctl_nr_open_min,
1696 .extra2 = &sysctl_nr_open_max,
1699 .procname = "dentry-state",
1700 .data = &dentry_stat,
1701 .maxlen = 6*sizeof(long),
1703 .proc_handler = proc_nr_dentry,
1706 .procname = "overflowuid",
1707 .data = &fs_overflowuid,
1708 .maxlen = sizeof(int),
1710 .proc_handler = proc_dointvec_minmax,
1711 .extra1 = &minolduid,
1712 .extra2 = &maxolduid,
1715 .procname = "overflowgid",
1716 .data = &fs_overflowgid,
1717 .maxlen = sizeof(int),
1719 .proc_handler = proc_dointvec_minmax,
1720 .extra1 = &minolduid,
1721 .extra2 = &maxolduid,
1723 #ifdef CONFIG_FILE_LOCKING
1725 .procname = "leases-enable",
1726 .data = &leases_enable,
1727 .maxlen = sizeof(int),
1729 .proc_handler = proc_dointvec,
1732 #ifdef CONFIG_DNOTIFY
1734 .procname = "dir-notify-enable",
1735 .data = &dir_notify_enable,
1736 .maxlen = sizeof(int),
1738 .proc_handler = proc_dointvec,
1742 #ifdef CONFIG_FILE_LOCKING
1744 .procname = "lease-break-time",
1745 .data = &lease_break_time,
1746 .maxlen = sizeof(int),
1748 .proc_handler = proc_dointvec,
1753 .procname = "aio-nr",
1755 .maxlen = sizeof(aio_nr),
1757 .proc_handler = proc_doulongvec_minmax,
1760 .procname = "aio-max-nr",
1761 .data = &aio_max_nr,
1762 .maxlen = sizeof(aio_max_nr),
1764 .proc_handler = proc_doulongvec_minmax,
1766 #endif /* CONFIG_AIO */
1767 #ifdef CONFIG_INOTIFY_USER
1769 .procname = "inotify",
1771 .child = inotify_table,
1776 .procname = "epoll",
1778 .child = epoll_table,
1783 .procname = "protected_symlinks",
1784 .data = &sysctl_protected_symlinks,
1785 .maxlen = sizeof(int),
1787 .proc_handler = proc_dointvec_minmax,
1792 .procname = "protected_hardlinks",
1793 .data = &sysctl_protected_hardlinks,
1794 .maxlen = sizeof(int),
1796 .proc_handler = proc_dointvec_minmax,
1801 .procname = "suid_dumpable",
1802 .data = &suid_dumpable,
1803 .maxlen = sizeof(int),
1805 .proc_handler = proc_dointvec_minmax_coredump,
1809 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1811 .procname = "binfmt_misc",
1813 .child = sysctl_mount_point,
1817 .procname = "pipe-max-size",
1818 .data = &pipe_max_size,
1819 .maxlen = sizeof(int),
1821 .proc_handler = &pipe_proc_fn,
1822 .extra1 = &pipe_min_size,
1825 .procname = "pipe-user-pages-hard",
1826 .data = &pipe_user_pages_hard,
1827 .maxlen = sizeof(pipe_user_pages_hard),
1829 .proc_handler = proc_doulongvec_minmax,
1832 .procname = "pipe-user-pages-soft",
1833 .data = &pipe_user_pages_soft,
1834 .maxlen = sizeof(pipe_user_pages_soft),
1836 .proc_handler = proc_doulongvec_minmax,
1839 .procname = "mount-max",
1840 .data = &sysctl_mount_max,
1841 .maxlen = sizeof(unsigned int),
1843 .proc_handler = proc_dointvec_minmax,
1849 static struct ctl_table debug_table[] = {
1850 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1852 .procname = "exception-trace",
1853 .data = &show_unhandled_signals,
1854 .maxlen = sizeof(int),
1856 .proc_handler = proc_dointvec
1859 #if defined(CONFIG_OPTPROBES)
1861 .procname = "kprobes-optimization",
1862 .data = &sysctl_kprobes_optimization,
1863 .maxlen = sizeof(int),
1865 .proc_handler = proc_kprobes_optimization_handler,
1873 static struct ctl_table dev_table[] = {
1877 int __init sysctl_init(void)
1879 struct ctl_table_header *hdr;
1881 hdr = register_sysctl_table(sysctl_base_table);
1882 kmemleak_not_leak(hdr);
1886 #endif /* CONFIG_SYSCTL */
1892 #ifdef CONFIG_PROC_SYSCTL
1894 static int _proc_do_string(char *data, int maxlen, int write,
1895 char __user *buffer,
1896 size_t *lenp, loff_t *ppos)
1902 if (!data || !maxlen || !*lenp) {
1908 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1909 /* Only continue writes not past the end of buffer. */
1911 if (len > maxlen - 1)
1918 /* Start writing from beginning of buffer. */
1924 while ((p - buffer) < *lenp && len < maxlen - 1) {
1925 if (get_user(c, p++))
1927 if (c == 0 || c == '\n')
1948 if (copy_to_user(buffer, data, len))
1951 if (put_user('\n', buffer + len))
1961 static void warn_sysctl_write(struct ctl_table *table)
1963 pr_warn_once("%s wrote to %s when file position was not 0!\n"
1964 "This will not be supported in the future. To silence this\n"
1965 "warning, set kernel.sysctl_writes_strict = -1\n",
1966 current->comm, table->procname);
1970 * proc_first_pos_non_zero_ignore - check if firs position is allowed
1971 * @ppos: file position
1972 * @table: the sysctl table
1974 * Returns true if the first position is non-zero and the sysctl_writes_strict
1975 * mode indicates this is not allowed for numeric input types. String proc
1976 * hadlers can ignore the return value.
1978 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
1979 struct ctl_table *table)
1984 switch (sysctl_writes_strict) {
1985 case SYSCTL_WRITES_STRICT:
1987 case SYSCTL_WRITES_WARN:
1988 warn_sysctl_write(table);
1996 * proc_dostring - read a string sysctl
1997 * @table: the sysctl table
1998 * @write: %TRUE if this is a write to the sysctl file
1999 * @buffer: the user buffer
2000 * @lenp: the size of the user buffer
2001 * @ppos: file position
2003 * Reads/writes a string from/to the user buffer. If the kernel
2004 * buffer provided is not large enough to hold the string, the
2005 * string is truncated. The copied string is %NULL-terminated.
2006 * If the string is being read by the user process, it is copied
2007 * and a newline '\n' is added. It is truncated if the buffer is
2010 * Returns 0 on success.
2012 int proc_dostring(struct ctl_table *table, int write,
2013 void __user *buffer, size_t *lenp, loff_t *ppos)
2016 proc_first_pos_non_zero_ignore(ppos, table);
2018 return _proc_do_string((char *)(table->data), table->maxlen, write,
2019 (char __user *)buffer, lenp, ppos);
2022 static size_t proc_skip_spaces(char **buf)
2025 char *tmp = skip_spaces(*buf);
2031 static void proc_skip_char(char **buf, size_t *size, const char v)
2041 #define TMPBUFLEN 22
2043 * proc_get_long - reads an ASCII formatted integer from a user buffer
2045 * @buf: a kernel buffer
2046 * @size: size of the kernel buffer
2047 * @val: this is where the number will be stored
2048 * @neg: set to %TRUE if number is negative
2049 * @perm_tr: a vector which contains the allowed trailers
2050 * @perm_tr_len: size of the perm_tr vector
2051 * @tr: pointer to store the trailer character
2053 * In case of success %0 is returned and @buf and @size are updated with
2054 * the amount of bytes read. If @tr is non-NULL and a trailing
2055 * character exists (size is non-zero after returning from this
2056 * function), @tr is updated with the trailing character.
2058 static int proc_get_long(char **buf, size_t *size,
2059 unsigned long *val, bool *neg,
2060 const char *perm_tr, unsigned perm_tr_len, char *tr)
2063 char *p, tmp[TMPBUFLEN];
2069 if (len > TMPBUFLEN - 1)
2070 len = TMPBUFLEN - 1;
2072 memcpy(tmp, *buf, len);
2076 if (*p == '-' && *size > 1) {
2084 *val = simple_strtoul(p, &p, 0);
2088 /* We don't know if the next char is whitespace thus we may accept
2089 * invalid integers (e.g. 1234...a) or two integers instead of one
2090 * (e.g. 123...1). So lets not allow such large numbers. */
2091 if (len == TMPBUFLEN - 1)
2094 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2097 if (tr && (len < *size))
2107 * proc_put_long - converts an integer to a decimal ASCII formatted string
2109 * @buf: the user buffer
2110 * @size: the size of the user buffer
2111 * @val: the integer to be converted
2112 * @neg: sign of the number, %TRUE for negative
2114 * In case of success %0 is returned and @buf and @size are updated with
2115 * the amount of bytes written.
2117 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2121 char tmp[TMPBUFLEN], *p = tmp;
2123 sprintf(p, "%s%lu", neg ? "-" : "", val);
2127 if (copy_to_user(*buf, tmp, len))
2135 static int proc_put_char(void __user **buf, size_t *size, char c)
2138 char __user **buffer = (char __user **)buf;
2139 if (put_user(c, *buffer))
2141 (*size)--, (*buffer)++;
2147 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2149 int write, void *data)
2153 if (*lvalp > (unsigned long) INT_MAX + 1)
2157 if (*lvalp > (unsigned long) INT_MAX)
2165 *lvalp = -(unsigned long)val;
2168 *lvalp = (unsigned long)val;
2174 static int do_proc_douintvec_conv(unsigned long *lvalp,
2176 int write, void *data)
2179 if (*lvalp > UINT_MAX)
2183 unsigned int val = *valp;
2184 *lvalp = (unsigned long)val;
2189 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2191 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2192 int write, void __user *buffer,
2193 size_t *lenp, loff_t *ppos,
2194 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2195 int write, void *data),
2198 int *i, vleft, first = 1, err = 0;
2200 char *kbuf = NULL, *p;
2202 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2207 i = (int *) tbl_data;
2208 vleft = table->maxlen / sizeof(*i);
2212 conv = do_proc_dointvec_conv;
2215 if (proc_first_pos_non_zero_ignore(ppos, table))
2218 if (left > PAGE_SIZE - 1)
2219 left = PAGE_SIZE - 1;
2220 p = kbuf = memdup_user_nul(buffer, left);
2222 return PTR_ERR(kbuf);
2225 for (; left && vleft--; i++, first=0) {
2230 left -= proc_skip_spaces(&p);
2234 err = proc_get_long(&p, &left, &lval, &neg,
2236 sizeof(proc_wspace_sep), NULL);
2239 if (conv(&neg, &lval, i, 1, data)) {
2244 if (conv(&neg, &lval, i, 0, data)) {
2249 err = proc_put_char(&buffer, &left, '\t');
2252 err = proc_put_long(&buffer, &left, lval, neg);
2258 if (!write && !first && left && !err)
2259 err = proc_put_char(&buffer, &left, '\n');
2260 if (write && !err && left)
2261 left -= proc_skip_spaces(&p);
2265 return err ? : -EINVAL;
2273 static int do_proc_dointvec(struct ctl_table *table, int write,
2274 void __user *buffer, size_t *lenp, loff_t *ppos,
2275 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2276 int write, void *data),
2279 return __do_proc_dointvec(table->data, table, write,
2280 buffer, lenp, ppos, conv, data);
2283 static int do_proc_douintvec_w(unsigned int *tbl_data,
2284 struct ctl_table *table,
2285 void __user *buffer,
2286 size_t *lenp, loff_t *ppos,
2287 int (*conv)(unsigned long *lvalp,
2289 int write, void *data),
2296 char *kbuf = NULL, *p;
2300 if (proc_first_pos_non_zero_ignore(ppos, table))
2303 if (left > PAGE_SIZE - 1)
2304 left = PAGE_SIZE - 1;
2306 p = kbuf = memdup_user_nul(buffer, left);
2310 left -= proc_skip_spaces(&p);
2316 err = proc_get_long(&p, &left, &lval, &neg,
2318 sizeof(proc_wspace_sep), NULL);
2324 if (conv(&lval, tbl_data, 1, data)) {
2330 left -= proc_skip_spaces(&p);
2339 /* This is in keeping with old __do_proc_dointvec() */
2345 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2346 size_t *lenp, loff_t *ppos,
2347 int (*conv)(unsigned long *lvalp,
2349 int write, void *data),
2358 if (conv(&lval, tbl_data, 0, data)) {
2363 err = proc_put_long(&buffer, &left, lval, false);
2367 err = proc_put_char(&buffer, &left, '\n');
2376 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2377 int write, void __user *buffer,
2378 size_t *lenp, loff_t *ppos,
2379 int (*conv)(unsigned long *lvalp,
2381 int write, void *data),
2384 unsigned int *i, vleft;
2386 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2391 i = (unsigned int *) tbl_data;
2392 vleft = table->maxlen / sizeof(*i);
2395 * Arrays are not supported, keep this simple. *Do not* add
2404 conv = do_proc_douintvec_conv;
2407 return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2409 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2412 static int do_proc_douintvec(struct ctl_table *table, int write,
2413 void __user *buffer, size_t *lenp, loff_t *ppos,
2414 int (*conv)(unsigned long *lvalp,
2416 int write, void *data),
2419 return __do_proc_douintvec(table->data, table, write,
2420 buffer, lenp, ppos, conv, data);
2424 * proc_dointvec - read a vector of integers
2425 * @table: the sysctl table
2426 * @write: %TRUE if this is a write to the sysctl file
2427 * @buffer: the user buffer
2428 * @lenp: the size of the user buffer
2429 * @ppos: file position
2431 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2432 * values from/to the user buffer, treated as an ASCII string.
2434 * Returns 0 on success.
2436 int proc_dointvec(struct ctl_table *table, int write,
2437 void __user *buffer, size_t *lenp, loff_t *ppos)
2439 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2443 * proc_douintvec - read a vector of unsigned integers
2444 * @table: the sysctl table
2445 * @write: %TRUE if this is a write to the sysctl file
2446 * @buffer: the user buffer
2447 * @lenp: the size of the user buffer
2448 * @ppos: file position
2450 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2451 * values from/to the user buffer, treated as an ASCII string.
2453 * Returns 0 on success.
2455 int proc_douintvec(struct ctl_table *table, int write,
2456 void __user *buffer, size_t *lenp, loff_t *ppos)
2458 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2459 do_proc_douintvec_conv, NULL);
2463 * Taint values can only be increased
2464 * This means we can safely use a temporary.
2466 static int proc_taint(struct ctl_table *table, int write,
2467 void __user *buffer, size_t *lenp, loff_t *ppos)
2470 unsigned long tmptaint = get_taint();
2473 if (write && !capable(CAP_SYS_ADMIN))
2478 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2484 * Poor man's atomic or. Not worth adding a primitive
2485 * to everyone's atomic.h for this
2488 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2489 if ((tmptaint >> i) & 1)
2490 add_taint(i, LOCKDEP_STILL_OK);
2497 #ifdef CONFIG_PRINTK
2498 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2499 void __user *buffer, size_t *lenp, loff_t *ppos)
2501 if (write && !capable(CAP_SYS_ADMIN))
2504 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2508 struct do_proc_dointvec_minmax_conv_param {
2513 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2515 int write, void *data)
2517 struct do_proc_dointvec_minmax_conv_param *param = data;
2519 int val = *negp ? -*lvalp : *lvalp;
2520 if ((param->min && *param->min > val) ||
2521 (param->max && *param->max < val))
2528 *lvalp = -(unsigned long)val;
2531 *lvalp = (unsigned long)val;
2538 * proc_dointvec_minmax - read a vector of integers with min/max values
2539 * @table: the sysctl table
2540 * @write: %TRUE if this is a write to the sysctl file
2541 * @buffer: the user buffer
2542 * @lenp: the size of the user buffer
2543 * @ppos: file position
2545 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2546 * values from/to the user buffer, treated as an ASCII string.
2548 * This routine will ensure the values are within the range specified by
2549 * table->extra1 (min) and table->extra2 (max).
2551 * Returns 0 on success.
2553 int proc_dointvec_minmax(struct ctl_table *table, int write,
2554 void __user *buffer, size_t *lenp, loff_t *ppos)
2556 struct do_proc_dointvec_minmax_conv_param param = {
2557 .min = (int *) table->extra1,
2558 .max = (int *) table->extra2,
2560 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2561 do_proc_dointvec_minmax_conv, ¶m);
2564 struct do_proc_douintvec_minmax_conv_param {
2569 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2571 int write, void *data)
2573 struct do_proc_douintvec_minmax_conv_param *param = data;
2576 unsigned int val = *lvalp;
2578 if ((param->min && *param->min > val) ||
2579 (param->max && *param->max < val))
2582 if (*lvalp > UINT_MAX)
2586 unsigned int val = *valp;
2587 *lvalp = (unsigned long) val;
2594 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2595 * @table: the sysctl table
2596 * @write: %TRUE if this is a write to the sysctl file
2597 * @buffer: the user buffer
2598 * @lenp: the size of the user buffer
2599 * @ppos: file position
2601 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2602 * values from/to the user buffer, treated as an ASCII string. Negative
2603 * strings are not allowed.
2605 * This routine will ensure the values are within the range specified by
2606 * table->extra1 (min) and table->extra2 (max). There is a final sanity
2607 * check for UINT_MAX to avoid having to support wrap around uses from
2610 * Returns 0 on success.
2612 int proc_douintvec_minmax(struct ctl_table *table, int write,
2613 void __user *buffer, size_t *lenp, loff_t *ppos)
2615 struct do_proc_douintvec_minmax_conv_param param = {
2616 .min = (unsigned int *) table->extra1,
2617 .max = (unsigned int *) table->extra2,
2619 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2620 do_proc_douintvec_minmax_conv, ¶m);
2623 static void validate_coredump_safety(void)
2625 #ifdef CONFIG_COREDUMP
2626 if (suid_dumpable == SUID_DUMP_ROOT &&
2627 core_pattern[0] != '/' && core_pattern[0] != '|') {
2629 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2630 "Pipe handler or fully qualified core dump path required.\n"
2631 "Set kernel.core_pattern before fs.suid_dumpable.\n"
2637 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2638 void __user *buffer, size_t *lenp, loff_t *ppos)
2640 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2642 validate_coredump_safety();
2646 #ifdef CONFIG_COREDUMP
2647 static int proc_dostring_coredump(struct ctl_table *table, int write,
2648 void __user *buffer, size_t *lenp, loff_t *ppos)
2650 int error = proc_dostring(table, write, buffer, lenp, ppos);
2652 validate_coredump_safety();
2657 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2658 void __user *buffer,
2659 size_t *lenp, loff_t *ppos,
2660 unsigned long convmul,
2661 unsigned long convdiv)
2663 unsigned long *i, *min, *max;
2664 int vleft, first = 1, err = 0;
2666 char *kbuf = NULL, *p;
2668 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2673 i = (unsigned long *) data;
2674 min = (unsigned long *) table->extra1;
2675 max = (unsigned long *) table->extra2;
2676 vleft = table->maxlen / sizeof(unsigned long);
2680 if (proc_first_pos_non_zero_ignore(ppos, table))
2683 if (left > PAGE_SIZE - 1)
2684 left = PAGE_SIZE - 1;
2685 p = kbuf = memdup_user_nul(buffer, left);
2687 return PTR_ERR(kbuf);
2690 for (; left && vleft--; i++, first = 0) {
2696 left -= proc_skip_spaces(&p);
2698 err = proc_get_long(&p, &left, &val, &neg,
2700 sizeof(proc_wspace_sep), NULL);
2705 val = convmul * val / convdiv;
2706 if ((min && val < *min) || (max && val > *max))
2710 val = convdiv * (*i) / convmul;
2712 err = proc_put_char(&buffer, &left, '\t');
2716 err = proc_put_long(&buffer, &left, val, false);
2722 if (!write && !first && left && !err)
2723 err = proc_put_char(&buffer, &left, '\n');
2725 left -= proc_skip_spaces(&p);
2729 return err ? : -EINVAL;
2737 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2738 void __user *buffer,
2739 size_t *lenp, loff_t *ppos,
2740 unsigned long convmul,
2741 unsigned long convdiv)
2743 return __do_proc_doulongvec_minmax(table->data, table, write,
2744 buffer, lenp, ppos, convmul, convdiv);
2748 * proc_doulongvec_minmax - read a vector of long integers 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 long) unsigned long
2756 * values from/to the user buffer, treated as an ASCII string.
2758 * This routine will ensure the values are within the range specified by
2759 * table->extra1 (min) and table->extra2 (max).
2761 * Returns 0 on success.
2763 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2764 void __user *buffer, size_t *lenp, loff_t *ppos)
2766 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2770 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2771 * @table: the sysctl table
2772 * @write: %TRUE if this is a write to the sysctl file
2773 * @buffer: the user buffer
2774 * @lenp: the size of the user buffer
2775 * @ppos: file position
2777 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2778 * values from/to the user buffer, treated as an ASCII string. The values
2779 * are treated as milliseconds, and converted to jiffies when they are stored.
2781 * This routine will ensure the values are within the range specified by
2782 * table->extra1 (min) and table->extra2 (max).
2784 * Returns 0 on success.
2786 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2787 void __user *buffer,
2788 size_t *lenp, loff_t *ppos)
2790 return do_proc_doulongvec_minmax(table, write, buffer,
2791 lenp, ppos, HZ, 1000l);
2795 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2797 int write, void *data)
2800 if (*lvalp > INT_MAX / HZ)
2802 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2808 lval = -(unsigned long)val;
2811 lval = (unsigned long)val;
2818 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2820 int write, void *data)
2823 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2825 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2831 lval = -(unsigned long)val;
2834 lval = (unsigned long)val;
2836 *lvalp = jiffies_to_clock_t(lval);
2841 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2843 int write, void *data)
2846 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2856 lval = -(unsigned long)val;
2859 lval = (unsigned long)val;
2861 *lvalp = jiffies_to_msecs(lval);
2867 * proc_dointvec_jiffies - read a vector of integers as seconds
2868 * @table: the sysctl table
2869 * @write: %TRUE if this is a write to the sysctl file
2870 * @buffer: the user buffer
2871 * @lenp: the size of the user buffer
2872 * @ppos: file position
2874 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2875 * values from/to the user buffer, treated as an ASCII string.
2876 * The values read are assumed to be in seconds, and are converted into
2879 * Returns 0 on success.
2881 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2882 void __user *buffer, size_t *lenp, loff_t *ppos)
2884 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2885 do_proc_dointvec_jiffies_conv,NULL);
2889 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2890 * @table: the sysctl table
2891 * @write: %TRUE if this is a write to the sysctl file
2892 * @buffer: the user buffer
2893 * @lenp: the size of the user buffer
2894 * @ppos: pointer to the file position
2896 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2897 * values from/to the user buffer, treated as an ASCII string.
2898 * The values read are assumed to be in 1/USER_HZ seconds, and
2899 * are converted into jiffies.
2901 * Returns 0 on success.
2903 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2904 void __user *buffer, size_t *lenp, loff_t *ppos)
2906 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2907 do_proc_dointvec_userhz_jiffies_conv,NULL);
2911 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2912 * @table: the sysctl table
2913 * @write: %TRUE if this is a write to the sysctl file
2914 * @buffer: the user buffer
2915 * @lenp: the size of the user buffer
2916 * @ppos: file position
2917 * @ppos: the current position in the file
2919 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2920 * values from/to the user buffer, treated as an ASCII string.
2921 * The values read are assumed to be in 1/1000 seconds, and
2922 * are converted into jiffies.
2924 * Returns 0 on success.
2926 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2927 void __user *buffer, size_t *lenp, loff_t *ppos)
2929 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2930 do_proc_dointvec_ms_jiffies_conv, NULL);
2933 static int proc_do_cad_pid(struct ctl_table *table, int write,
2934 void __user *buffer, size_t *lenp, loff_t *ppos)
2936 struct pid *new_pid;
2940 tmp = pid_vnr(cad_pid);
2942 r = __do_proc_dointvec(&tmp, table, write, buffer,
2943 lenp, ppos, NULL, NULL);
2947 new_pid = find_get_pid(tmp);
2951 put_pid(xchg(&cad_pid, new_pid));
2956 * proc_do_large_bitmap - read/write from/to a large bitmap
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 * The bitmap is stored at table->data and the bitmap length (in bits)
2966 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2967 * large bitmaps may be represented in a compact manner. Writing into
2968 * the file will clear the bitmap then update it with the given input.
2970 * Returns 0 on success.
2972 int proc_do_large_bitmap(struct ctl_table *table, int write,
2973 void __user *buffer, size_t *lenp, loff_t *ppos)
2977 size_t left = *lenp;
2978 unsigned long bitmap_len = table->maxlen;
2979 unsigned long *bitmap = *(unsigned long **) table->data;
2980 unsigned long *tmp_bitmap = NULL;
2981 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2983 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2991 if (left > PAGE_SIZE - 1)
2992 left = PAGE_SIZE - 1;
2994 p = kbuf = memdup_user_nul(buffer, left);
2996 return PTR_ERR(kbuf);
2998 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
3004 proc_skip_char(&p, &left, '\n');
3005 while (!err && left) {
3006 unsigned long val_a, val_b;
3009 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3013 if (val_a >= bitmap_len || neg) {
3025 err = proc_get_long(&p, &left, &val_b,
3026 &neg, tr_b, sizeof(tr_b),
3030 if (val_b >= bitmap_len || neg ||
3041 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3043 proc_skip_char(&p, &left, '\n');
3047 unsigned long bit_a, bit_b = 0;
3050 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3051 if (bit_a >= bitmap_len)
3053 bit_b = find_next_zero_bit(bitmap, bitmap_len,
3057 err = proc_put_char(&buffer, &left, ',');
3061 err = proc_put_long(&buffer, &left, bit_a, false);
3064 if (bit_a != bit_b) {
3065 err = proc_put_char(&buffer, &left, '-');
3068 err = proc_put_long(&buffer, &left, bit_b, false);
3076 err = proc_put_char(&buffer, &left, '\n');
3082 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3084 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3096 #else /* CONFIG_PROC_SYSCTL */
3098 int proc_dostring(struct ctl_table *table, int write,
3099 void __user *buffer, size_t *lenp, loff_t *ppos)
3104 int proc_dointvec(struct ctl_table *table, int write,
3105 void __user *buffer, size_t *lenp, loff_t *ppos)
3110 int proc_douintvec(struct ctl_table *table, int write,
3111 void __user *buffer, size_t *lenp, loff_t *ppos)
3116 int proc_dointvec_minmax(struct ctl_table *table, int write,
3117 void __user *buffer, size_t *lenp, loff_t *ppos)
3122 int proc_douintvec_minmax(struct ctl_table *table, int write,
3123 void __user *buffer, size_t *lenp, loff_t *ppos)
3128 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3129 void __user *buffer, size_t *lenp, loff_t *ppos)
3134 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3135 void __user *buffer, size_t *lenp, loff_t *ppos)
3140 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3141 void __user *buffer, size_t *lenp, loff_t *ppos)
3146 int proc_doulongvec_minmax(struct ctl_table *table, int write,
3147 void __user *buffer, size_t *lenp, loff_t *ppos)
3152 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3153 void __user *buffer,
3154 size_t *lenp, loff_t *ppos)
3160 #endif /* CONFIG_PROC_SYSCTL */
3163 * No sense putting this after each symbol definition, twice,
3164 * exception granted :-)
3166 EXPORT_SYMBOL(proc_dointvec);
3167 EXPORT_SYMBOL(proc_douintvec);
3168 EXPORT_SYMBOL(proc_dointvec_jiffies);
3169 EXPORT_SYMBOL(proc_dointvec_minmax);
3170 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3171 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3172 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3173 EXPORT_SYMBOL(proc_dostring);
3174 EXPORT_SYMBOL(proc_doulongvec_minmax);
3175 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);