Loading...
1/*
2 * sysctl.c: General linux system control interface
3 *
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
12 * Horn.
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
16 * Wendling.
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
19 */
20
21#include <linux/module.h>
22#include <linux/aio.h>
23#include <linux/mm.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>
34#include <linux/fs.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>
69#include <linux/pipe_fs_i.h>
70
71#include <linux/uaccess.h>
72#include <asm/processor.h>
73
74#ifdef CONFIG_X86
75#include <asm/nmi.h>
76#include <asm/stacktrace.h>
77#include <asm/io.h>
78#endif
79#ifdef CONFIG_SPARC
80#include <asm/setup.h>
81#endif
82#ifdef CONFIG_BSD_PROCESS_ACCT
83#include <linux/acct.h>
84#endif
85#ifdef CONFIG_RT_MUTEXES
86#include <linux/rtmutex.h>
87#endif
88#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
89#include <linux/lockdep.h>
90#endif
91#ifdef CONFIG_CHR_DEV_SG
92#include <scsi/sg.h>
93#endif
94
95#ifdef CONFIG_LOCKUP_DETECTOR
96#include <linux/nmi.h>
97#endif
98
99#if defined(CONFIG_SYSCTL)
100
101/* External variables not in a header file. */
102extern int suid_dumpable;
103#ifdef CONFIG_COREDUMP
104extern int core_uses_pid;
105extern char core_pattern[];
106extern unsigned int core_pipe_limit;
107#endif
108extern int pid_max;
109extern int pid_max_min, pid_max_max;
110extern int percpu_pagelist_fraction;
111extern int latencytop_enabled;
112extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
113#ifndef CONFIG_MMU
114extern int sysctl_nr_trim_pages;
115#endif
116
117/* Constants used for minimum and maximum */
118#ifdef CONFIG_LOCKUP_DETECTOR
119static int sixty = 60;
120#endif
121
122static int __maybe_unused neg_one = -1;
123
124static int zero;
125static int __maybe_unused one = 1;
126static int __maybe_unused two = 2;
127static int __maybe_unused four = 4;
128static unsigned long one_ul = 1;
129static int one_hundred = 100;
130static int one_thousand = 1000;
131#ifdef CONFIG_PRINTK
132static int ten_thousand = 10000;
133#endif
134#ifdef CONFIG_PERF_EVENTS
135static int six_hundred_forty_kb = 640 * 1024;
136#endif
137
138/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
139static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
140
141/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
142static int maxolduid = 65535;
143static int minolduid;
144
145static int ngroups_max = NGROUPS_MAX;
146static const int cap_last_cap = CAP_LAST_CAP;
147
148/*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */
149#ifdef CONFIG_DETECT_HUNG_TASK
150static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
151#endif
152
153#ifdef CONFIG_INOTIFY_USER
154#include <linux/inotify.h>
155#endif
156#ifdef CONFIG_SPARC
157#endif
158
159#ifdef __hppa__
160extern int pwrsw_enabled;
161#endif
162
163#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
164extern int unaligned_enabled;
165#endif
166
167#ifdef CONFIG_IA64
168extern int unaligned_dump_stack;
169#endif
170
171#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
172extern int no_unaligned_warning;
173#endif
174
175#ifdef CONFIG_PROC_SYSCTL
176
177/**
178 * enum sysctl_writes_mode - supported sysctl write modes
179 *
180 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
181 * to be written, and multiple writes on the same sysctl file descriptor
182 * will rewrite the sysctl value, regardless of file position. No warning
183 * is issued when the initial position is not 0.
184 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
185 * not 0.
186 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
187 * file position 0 and the value must be fully contained in the buffer
188 * sent to the write syscall. If dealing with strings respect the file
189 * position, but restrict this to the max length of the buffer, anything
190 * passed the max lenght will be ignored. Multiple writes will append
191 * to the buffer.
192 *
193 * These write modes control how current file position affects the behavior of
194 * updating sysctl values through the proc interface on each write.
195 */
196enum sysctl_writes_mode {
197 SYSCTL_WRITES_LEGACY = -1,
198 SYSCTL_WRITES_WARN = 0,
199 SYSCTL_WRITES_STRICT = 1,
200};
201
202static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
203
204static int proc_do_cad_pid(struct ctl_table *table, int write,
205 void __user *buffer, size_t *lenp, loff_t *ppos);
206static int proc_taint(struct ctl_table *table, int write,
207 void __user *buffer, size_t *lenp, loff_t *ppos);
208#endif
209
210#ifdef CONFIG_PRINTK
211static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
212 void __user *buffer, size_t *lenp, loff_t *ppos);
213#endif
214
215static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
216 void __user *buffer, size_t *lenp, loff_t *ppos);
217#ifdef CONFIG_COREDUMP
218static int proc_dostring_coredump(struct ctl_table *table, int write,
219 void __user *buffer, size_t *lenp, loff_t *ppos);
220#endif
221static int proc_dopipe_max_size(struct ctl_table *table, int write,
222 void __user *buffer, size_t *lenp, loff_t *ppos);
223
224#ifdef CONFIG_MAGIC_SYSRQ
225/* Note: sysrq code uses it's own private copy */
226static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
227
228static int sysrq_sysctl_handler(struct ctl_table *table, int write,
229 void __user *buffer, size_t *lenp,
230 loff_t *ppos)
231{
232 int error;
233
234 error = proc_dointvec(table, write, buffer, lenp, ppos);
235 if (error)
236 return error;
237
238 if (write)
239 sysrq_toggle_support(__sysrq_enabled);
240
241 return 0;
242}
243
244#endif
245
246static struct ctl_table kern_table[];
247static struct ctl_table vm_table[];
248static struct ctl_table fs_table[];
249static struct ctl_table debug_table[];
250static struct ctl_table dev_table[];
251extern struct ctl_table random_table[];
252#ifdef CONFIG_EPOLL
253extern struct ctl_table epoll_table[];
254#endif
255
256#ifdef CONFIG_FW_LOADER_USER_HELPER
257extern struct ctl_table firmware_config_table[];
258#endif
259
260#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
261int sysctl_legacy_va_layout;
262#endif
263
264/* The default sysctl tables: */
265
266static struct ctl_table sysctl_base_table[] = {
267 {
268 .procname = "kernel",
269 .mode = 0555,
270 .child = kern_table,
271 },
272 {
273 .procname = "vm",
274 .mode = 0555,
275 .child = vm_table,
276 },
277 {
278 .procname = "fs",
279 .mode = 0555,
280 .child = fs_table,
281 },
282 {
283 .procname = "debug",
284 .mode = 0555,
285 .child = debug_table,
286 },
287 {
288 .procname = "dev",
289 .mode = 0555,
290 .child = dev_table,
291 },
292 { }
293};
294
295#ifdef CONFIG_SCHED_DEBUG
296static int min_sched_granularity_ns = 100000; /* 100 usecs */
297static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
298static int min_wakeup_granularity_ns; /* 0 usecs */
299static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
300#ifdef CONFIG_SMP
301static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
302static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
303#endif /* CONFIG_SMP */
304#endif /* CONFIG_SCHED_DEBUG */
305
306#ifdef CONFIG_COMPACTION
307static int min_extfrag_threshold;
308static int max_extfrag_threshold = 1000;
309#endif
310
311static struct ctl_table kern_table[] = {
312 {
313 .procname = "sched_child_runs_first",
314 .data = &sysctl_sched_child_runs_first,
315 .maxlen = sizeof(unsigned int),
316 .mode = 0644,
317 .proc_handler = proc_dointvec,
318 },
319#ifdef CONFIG_SCHED_DEBUG
320 {
321 .procname = "sched_min_granularity_ns",
322 .data = &sysctl_sched_min_granularity,
323 .maxlen = sizeof(unsigned int),
324 .mode = 0644,
325 .proc_handler = sched_proc_update_handler,
326 .extra1 = &min_sched_granularity_ns,
327 .extra2 = &max_sched_granularity_ns,
328 },
329 {
330 .procname = "sched_latency_ns",
331 .data = &sysctl_sched_latency,
332 .maxlen = sizeof(unsigned int),
333 .mode = 0644,
334 .proc_handler = sched_proc_update_handler,
335 .extra1 = &min_sched_granularity_ns,
336 .extra2 = &max_sched_granularity_ns,
337 },
338 {
339 .procname = "sched_wakeup_granularity_ns",
340 .data = &sysctl_sched_wakeup_granularity,
341 .maxlen = sizeof(unsigned int),
342 .mode = 0644,
343 .proc_handler = sched_proc_update_handler,
344 .extra1 = &min_wakeup_granularity_ns,
345 .extra2 = &max_wakeup_granularity_ns,
346 },
347#ifdef CONFIG_SMP
348 {
349 .procname = "sched_tunable_scaling",
350 .data = &sysctl_sched_tunable_scaling,
351 .maxlen = sizeof(enum sched_tunable_scaling),
352 .mode = 0644,
353 .proc_handler = sched_proc_update_handler,
354 .extra1 = &min_sched_tunable_scaling,
355 .extra2 = &max_sched_tunable_scaling,
356 },
357 {
358 .procname = "sched_migration_cost_ns",
359 .data = &sysctl_sched_migration_cost,
360 .maxlen = sizeof(unsigned int),
361 .mode = 0644,
362 .proc_handler = proc_dointvec,
363 },
364 {
365 .procname = "sched_nr_migrate",
366 .data = &sysctl_sched_nr_migrate,
367 .maxlen = sizeof(unsigned int),
368 .mode = 0644,
369 .proc_handler = proc_dointvec,
370 },
371 {
372 .procname = "sched_time_avg_ms",
373 .data = &sysctl_sched_time_avg,
374 .maxlen = sizeof(unsigned int),
375 .mode = 0644,
376 .proc_handler = proc_dointvec_minmax,
377 .extra1 = &one,
378 },
379#ifdef CONFIG_SCHEDSTATS
380 {
381 .procname = "sched_schedstats",
382 .data = NULL,
383 .maxlen = sizeof(unsigned int),
384 .mode = 0644,
385 .proc_handler = sysctl_schedstats,
386 .extra1 = &zero,
387 .extra2 = &one,
388 },
389#endif /* CONFIG_SCHEDSTATS */
390#endif /* CONFIG_SMP */
391#ifdef CONFIG_NUMA_BALANCING
392 {
393 .procname = "numa_balancing_scan_delay_ms",
394 .data = &sysctl_numa_balancing_scan_delay,
395 .maxlen = sizeof(unsigned int),
396 .mode = 0644,
397 .proc_handler = proc_dointvec,
398 },
399 {
400 .procname = "numa_balancing_scan_period_min_ms",
401 .data = &sysctl_numa_balancing_scan_period_min,
402 .maxlen = sizeof(unsigned int),
403 .mode = 0644,
404 .proc_handler = proc_dointvec,
405 },
406 {
407 .procname = "numa_balancing_scan_period_max_ms",
408 .data = &sysctl_numa_balancing_scan_period_max,
409 .maxlen = sizeof(unsigned int),
410 .mode = 0644,
411 .proc_handler = proc_dointvec,
412 },
413 {
414 .procname = "numa_balancing_scan_size_mb",
415 .data = &sysctl_numa_balancing_scan_size,
416 .maxlen = sizeof(unsigned int),
417 .mode = 0644,
418 .proc_handler = proc_dointvec_minmax,
419 .extra1 = &one,
420 },
421 {
422 .procname = "numa_balancing",
423 .data = NULL, /* filled in by handler */
424 .maxlen = sizeof(unsigned int),
425 .mode = 0644,
426 .proc_handler = sysctl_numa_balancing,
427 .extra1 = &zero,
428 .extra2 = &one,
429 },
430#endif /* CONFIG_NUMA_BALANCING */
431#endif /* CONFIG_SCHED_DEBUG */
432 {
433 .procname = "sched_rt_period_us",
434 .data = &sysctl_sched_rt_period,
435 .maxlen = sizeof(unsigned int),
436 .mode = 0644,
437 .proc_handler = sched_rt_handler,
438 },
439 {
440 .procname = "sched_rt_runtime_us",
441 .data = &sysctl_sched_rt_runtime,
442 .maxlen = sizeof(int),
443 .mode = 0644,
444 .proc_handler = sched_rt_handler,
445 },
446 {
447 .procname = "sched_rr_timeslice_ms",
448 .data = &sysctl_sched_rr_timeslice,
449 .maxlen = sizeof(int),
450 .mode = 0644,
451 .proc_handler = sched_rr_handler,
452 },
453#ifdef CONFIG_SCHED_AUTOGROUP
454 {
455 .procname = "sched_autogroup_enabled",
456 .data = &sysctl_sched_autogroup_enabled,
457 .maxlen = sizeof(unsigned int),
458 .mode = 0644,
459 .proc_handler = proc_dointvec_minmax,
460 .extra1 = &zero,
461 .extra2 = &one,
462 },
463#endif
464#ifdef CONFIG_CFS_BANDWIDTH
465 {
466 .procname = "sched_cfs_bandwidth_slice_us",
467 .data = &sysctl_sched_cfs_bandwidth_slice,
468 .maxlen = sizeof(unsigned int),
469 .mode = 0644,
470 .proc_handler = proc_dointvec_minmax,
471 .extra1 = &one,
472 },
473#endif
474#ifdef CONFIG_PROVE_LOCKING
475 {
476 .procname = "prove_locking",
477 .data = &prove_locking,
478 .maxlen = sizeof(int),
479 .mode = 0644,
480 .proc_handler = proc_dointvec,
481 },
482#endif
483#ifdef CONFIG_LOCK_STAT
484 {
485 .procname = "lock_stat",
486 .data = &lock_stat,
487 .maxlen = sizeof(int),
488 .mode = 0644,
489 .proc_handler = proc_dointvec,
490 },
491#endif
492 {
493 .procname = "panic",
494 .data = &panic_timeout,
495 .maxlen = sizeof(int),
496 .mode = 0644,
497 .proc_handler = proc_dointvec,
498 },
499#ifdef CONFIG_COREDUMP
500 {
501 .procname = "core_uses_pid",
502 .data = &core_uses_pid,
503 .maxlen = sizeof(int),
504 .mode = 0644,
505 .proc_handler = proc_dointvec,
506 },
507 {
508 .procname = "core_pattern",
509 .data = core_pattern,
510 .maxlen = CORENAME_MAX_SIZE,
511 .mode = 0644,
512 .proc_handler = proc_dostring_coredump,
513 },
514 {
515 .procname = "core_pipe_limit",
516 .data = &core_pipe_limit,
517 .maxlen = sizeof(unsigned int),
518 .mode = 0644,
519 .proc_handler = proc_dointvec,
520 },
521#endif
522#ifdef CONFIG_PROC_SYSCTL
523 {
524 .procname = "tainted",
525 .maxlen = sizeof(long),
526 .mode = 0644,
527 .proc_handler = proc_taint,
528 },
529 {
530 .procname = "sysctl_writes_strict",
531 .data = &sysctl_writes_strict,
532 .maxlen = sizeof(int),
533 .mode = 0644,
534 .proc_handler = proc_dointvec_minmax,
535 .extra1 = &neg_one,
536 .extra2 = &one,
537 },
538#endif
539#ifdef CONFIG_LATENCYTOP
540 {
541 .procname = "latencytop",
542 .data = &latencytop_enabled,
543 .maxlen = sizeof(int),
544 .mode = 0644,
545 .proc_handler = sysctl_latencytop,
546 },
547#endif
548#ifdef CONFIG_BLK_DEV_INITRD
549 {
550 .procname = "real-root-dev",
551 .data = &real_root_dev,
552 .maxlen = sizeof(int),
553 .mode = 0644,
554 .proc_handler = proc_dointvec,
555 },
556#endif
557 {
558 .procname = "print-fatal-signals",
559 .data = &print_fatal_signals,
560 .maxlen = sizeof(int),
561 .mode = 0644,
562 .proc_handler = proc_dointvec,
563 },
564#ifdef CONFIG_SPARC
565 {
566 .procname = "reboot-cmd",
567 .data = reboot_command,
568 .maxlen = 256,
569 .mode = 0644,
570 .proc_handler = proc_dostring,
571 },
572 {
573 .procname = "stop-a",
574 .data = &stop_a_enabled,
575 .maxlen = sizeof (int),
576 .mode = 0644,
577 .proc_handler = proc_dointvec,
578 },
579 {
580 .procname = "scons-poweroff",
581 .data = &scons_pwroff,
582 .maxlen = sizeof (int),
583 .mode = 0644,
584 .proc_handler = proc_dointvec,
585 },
586#endif
587#ifdef CONFIG_SPARC64
588 {
589 .procname = "tsb-ratio",
590 .data = &sysctl_tsb_ratio,
591 .maxlen = sizeof (int),
592 .mode = 0644,
593 .proc_handler = proc_dointvec,
594 },
595#endif
596#ifdef __hppa__
597 {
598 .procname = "soft-power",
599 .data = &pwrsw_enabled,
600 .maxlen = sizeof (int),
601 .mode = 0644,
602 .proc_handler = proc_dointvec,
603 },
604#endif
605#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
606 {
607 .procname = "unaligned-trap",
608 .data = &unaligned_enabled,
609 .maxlen = sizeof (int),
610 .mode = 0644,
611 .proc_handler = proc_dointvec,
612 },
613#endif
614 {
615 .procname = "ctrl-alt-del",
616 .data = &C_A_D,
617 .maxlen = sizeof(int),
618 .mode = 0644,
619 .proc_handler = proc_dointvec,
620 },
621#ifdef CONFIG_FUNCTION_TRACER
622 {
623 .procname = "ftrace_enabled",
624 .data = &ftrace_enabled,
625 .maxlen = sizeof(int),
626 .mode = 0644,
627 .proc_handler = ftrace_enable_sysctl,
628 },
629#endif
630#ifdef CONFIG_STACK_TRACER
631 {
632 .procname = "stack_tracer_enabled",
633 .data = &stack_tracer_enabled,
634 .maxlen = sizeof(int),
635 .mode = 0644,
636 .proc_handler = stack_trace_sysctl,
637 },
638#endif
639#ifdef CONFIG_TRACING
640 {
641 .procname = "ftrace_dump_on_oops",
642 .data = &ftrace_dump_on_oops,
643 .maxlen = sizeof(int),
644 .mode = 0644,
645 .proc_handler = proc_dointvec,
646 },
647 {
648 .procname = "traceoff_on_warning",
649 .data = &__disable_trace_on_warning,
650 .maxlen = sizeof(__disable_trace_on_warning),
651 .mode = 0644,
652 .proc_handler = proc_dointvec,
653 },
654 {
655 .procname = "tracepoint_printk",
656 .data = &tracepoint_printk,
657 .maxlen = sizeof(tracepoint_printk),
658 .mode = 0644,
659 .proc_handler = tracepoint_printk_sysctl,
660 },
661#endif
662#ifdef CONFIG_KEXEC_CORE
663 {
664 .procname = "kexec_load_disabled",
665 .data = &kexec_load_disabled,
666 .maxlen = sizeof(int),
667 .mode = 0644,
668 /* only handle a transition from default "0" to "1" */
669 .proc_handler = proc_dointvec_minmax,
670 .extra1 = &one,
671 .extra2 = &one,
672 },
673#endif
674#ifdef CONFIG_MODULES
675 {
676 .procname = "modprobe",
677 .data = &modprobe_path,
678 .maxlen = KMOD_PATH_LEN,
679 .mode = 0644,
680 .proc_handler = proc_dostring,
681 },
682 {
683 .procname = "modules_disabled",
684 .data = &modules_disabled,
685 .maxlen = sizeof(int),
686 .mode = 0644,
687 /* only handle a transition from default "0" to "1" */
688 .proc_handler = proc_dointvec_minmax,
689 .extra1 = &one,
690 .extra2 = &one,
691 },
692#endif
693#ifdef CONFIG_UEVENT_HELPER
694 {
695 .procname = "hotplug",
696 .data = &uevent_helper,
697 .maxlen = UEVENT_HELPER_PATH_LEN,
698 .mode = 0644,
699 .proc_handler = proc_dostring,
700 },
701#endif
702#ifdef CONFIG_CHR_DEV_SG
703 {
704 .procname = "sg-big-buff",
705 .data = &sg_big_buff,
706 .maxlen = sizeof (int),
707 .mode = 0444,
708 .proc_handler = proc_dointvec,
709 },
710#endif
711#ifdef CONFIG_BSD_PROCESS_ACCT
712 {
713 .procname = "acct",
714 .data = &acct_parm,
715 .maxlen = 3*sizeof(int),
716 .mode = 0644,
717 .proc_handler = proc_dointvec,
718 },
719#endif
720#ifdef CONFIG_MAGIC_SYSRQ
721 {
722 .procname = "sysrq",
723 .data = &__sysrq_enabled,
724 .maxlen = sizeof (int),
725 .mode = 0644,
726 .proc_handler = sysrq_sysctl_handler,
727 },
728#endif
729#ifdef CONFIG_PROC_SYSCTL
730 {
731 .procname = "cad_pid",
732 .data = NULL,
733 .maxlen = sizeof (int),
734 .mode = 0600,
735 .proc_handler = proc_do_cad_pid,
736 },
737#endif
738 {
739 .procname = "threads-max",
740 .data = NULL,
741 .maxlen = sizeof(int),
742 .mode = 0644,
743 .proc_handler = sysctl_max_threads,
744 },
745 {
746 .procname = "random",
747 .mode = 0555,
748 .child = random_table,
749 },
750 {
751 .procname = "usermodehelper",
752 .mode = 0555,
753 .child = usermodehelper_table,
754 },
755#ifdef CONFIG_FW_LOADER_USER_HELPER
756 {
757 .procname = "firmware_config",
758 .mode = 0555,
759 .child = firmware_config_table,
760 },
761#endif
762 {
763 .procname = "overflowuid",
764 .data = &overflowuid,
765 .maxlen = sizeof(int),
766 .mode = 0644,
767 .proc_handler = proc_dointvec_minmax,
768 .extra1 = &minolduid,
769 .extra2 = &maxolduid,
770 },
771 {
772 .procname = "overflowgid",
773 .data = &overflowgid,
774 .maxlen = sizeof(int),
775 .mode = 0644,
776 .proc_handler = proc_dointvec_minmax,
777 .extra1 = &minolduid,
778 .extra2 = &maxolduid,
779 },
780#ifdef CONFIG_S390
781#ifdef CONFIG_MATHEMU
782 {
783 .procname = "ieee_emulation_warnings",
784 .data = &sysctl_ieee_emulation_warnings,
785 .maxlen = sizeof(int),
786 .mode = 0644,
787 .proc_handler = proc_dointvec,
788 },
789#endif
790 {
791 .procname = "userprocess_debug",
792 .data = &show_unhandled_signals,
793 .maxlen = sizeof(int),
794 .mode = 0644,
795 .proc_handler = proc_dointvec,
796 },
797#endif
798 {
799 .procname = "pid_max",
800 .data = &pid_max,
801 .maxlen = sizeof (int),
802 .mode = 0644,
803 .proc_handler = proc_dointvec_minmax,
804 .extra1 = &pid_max_min,
805 .extra2 = &pid_max_max,
806 },
807 {
808 .procname = "panic_on_oops",
809 .data = &panic_on_oops,
810 .maxlen = sizeof(int),
811 .mode = 0644,
812 .proc_handler = proc_dointvec,
813 },
814#if defined CONFIG_PRINTK
815 {
816 .procname = "printk",
817 .data = &console_loglevel,
818 .maxlen = 4*sizeof(int),
819 .mode = 0644,
820 .proc_handler = proc_dointvec,
821 },
822 {
823 .procname = "printk_ratelimit",
824 .data = &printk_ratelimit_state.interval,
825 .maxlen = sizeof(int),
826 .mode = 0644,
827 .proc_handler = proc_dointvec_jiffies,
828 },
829 {
830 .procname = "printk_ratelimit_burst",
831 .data = &printk_ratelimit_state.burst,
832 .maxlen = sizeof(int),
833 .mode = 0644,
834 .proc_handler = proc_dointvec,
835 },
836 {
837 .procname = "printk_delay",
838 .data = &printk_delay_msec,
839 .maxlen = sizeof(int),
840 .mode = 0644,
841 .proc_handler = proc_dointvec_minmax,
842 .extra1 = &zero,
843 .extra2 = &ten_thousand,
844 },
845 {
846 .procname = "printk_devkmsg",
847 .data = devkmsg_log_str,
848 .maxlen = DEVKMSG_STR_MAX_SIZE,
849 .mode = 0644,
850 .proc_handler = devkmsg_sysctl_set_loglvl,
851 },
852 {
853 .procname = "dmesg_restrict",
854 .data = &dmesg_restrict,
855 .maxlen = sizeof(int),
856 .mode = 0644,
857 .proc_handler = proc_dointvec_minmax_sysadmin,
858 .extra1 = &zero,
859 .extra2 = &one,
860 },
861 {
862 .procname = "kptr_restrict",
863 .data = &kptr_restrict,
864 .maxlen = sizeof(int),
865 .mode = 0644,
866 .proc_handler = proc_dointvec_minmax_sysadmin,
867 .extra1 = &zero,
868 .extra2 = &two,
869 },
870#endif
871 {
872 .procname = "ngroups_max",
873 .data = &ngroups_max,
874 .maxlen = sizeof (int),
875 .mode = 0444,
876 .proc_handler = proc_dointvec,
877 },
878 {
879 .procname = "cap_last_cap",
880 .data = (void *)&cap_last_cap,
881 .maxlen = sizeof(int),
882 .mode = 0444,
883 .proc_handler = proc_dointvec,
884 },
885#if defined(CONFIG_LOCKUP_DETECTOR)
886 {
887 .procname = "watchdog",
888 .data = &watchdog_user_enabled,
889 .maxlen = sizeof(int),
890 .mode = 0644,
891 .proc_handler = proc_watchdog,
892 .extra1 = &zero,
893 .extra2 = &one,
894 },
895 {
896 .procname = "watchdog_thresh",
897 .data = &watchdog_thresh,
898 .maxlen = sizeof(int),
899 .mode = 0644,
900 .proc_handler = proc_watchdog_thresh,
901 .extra1 = &zero,
902 .extra2 = &sixty,
903 },
904 {
905 .procname = "nmi_watchdog",
906 .data = &nmi_watchdog_user_enabled,
907 .maxlen = sizeof(int),
908 .mode = NMI_WATCHDOG_SYSCTL_PERM,
909 .proc_handler = proc_nmi_watchdog,
910 .extra1 = &zero,
911 .extra2 = &one,
912 },
913 {
914 .procname = "watchdog_cpumask",
915 .data = &watchdog_cpumask_bits,
916 .maxlen = NR_CPUS,
917 .mode = 0644,
918 .proc_handler = proc_watchdog_cpumask,
919 },
920#ifdef CONFIG_SOFTLOCKUP_DETECTOR
921 {
922 .procname = "soft_watchdog",
923 .data = &soft_watchdog_user_enabled,
924 .maxlen = sizeof(int),
925 .mode = 0644,
926 .proc_handler = proc_soft_watchdog,
927 .extra1 = &zero,
928 .extra2 = &one,
929 },
930 {
931 .procname = "softlockup_panic",
932 .data = &softlockup_panic,
933 .maxlen = sizeof(int),
934 .mode = 0644,
935 .proc_handler = proc_dointvec_minmax,
936 .extra1 = &zero,
937 .extra2 = &one,
938 },
939#ifdef CONFIG_SMP
940 {
941 .procname = "softlockup_all_cpu_backtrace",
942 .data = &sysctl_softlockup_all_cpu_backtrace,
943 .maxlen = sizeof(int),
944 .mode = 0644,
945 .proc_handler = proc_dointvec_minmax,
946 .extra1 = &zero,
947 .extra2 = &one,
948 },
949#endif /* CONFIG_SMP */
950#endif
951#ifdef CONFIG_HARDLOCKUP_DETECTOR
952 {
953 .procname = "hardlockup_panic",
954 .data = &hardlockup_panic,
955 .maxlen = sizeof(int),
956 .mode = 0644,
957 .proc_handler = proc_dointvec_minmax,
958 .extra1 = &zero,
959 .extra2 = &one,
960 },
961#ifdef CONFIG_SMP
962 {
963 .procname = "hardlockup_all_cpu_backtrace",
964 .data = &sysctl_hardlockup_all_cpu_backtrace,
965 .maxlen = sizeof(int),
966 .mode = 0644,
967 .proc_handler = proc_dointvec_minmax,
968 .extra1 = &zero,
969 .extra2 = &one,
970 },
971#endif /* CONFIG_SMP */
972#endif
973#endif
974
975#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
976 {
977 .procname = "unknown_nmi_panic",
978 .data = &unknown_nmi_panic,
979 .maxlen = sizeof (int),
980 .mode = 0644,
981 .proc_handler = proc_dointvec,
982 },
983#endif
984#if defined(CONFIG_X86)
985 {
986 .procname = "panic_on_unrecovered_nmi",
987 .data = &panic_on_unrecovered_nmi,
988 .maxlen = sizeof(int),
989 .mode = 0644,
990 .proc_handler = proc_dointvec,
991 },
992 {
993 .procname = "panic_on_io_nmi",
994 .data = &panic_on_io_nmi,
995 .maxlen = sizeof(int),
996 .mode = 0644,
997 .proc_handler = proc_dointvec,
998 },
999#ifdef CONFIG_DEBUG_STACKOVERFLOW
1000 {
1001 .procname = "panic_on_stackoverflow",
1002 .data = &sysctl_panic_on_stackoverflow,
1003 .maxlen = sizeof(int),
1004 .mode = 0644,
1005 .proc_handler = proc_dointvec,
1006 },
1007#endif
1008 {
1009 .procname = "bootloader_type",
1010 .data = &bootloader_type,
1011 .maxlen = sizeof (int),
1012 .mode = 0444,
1013 .proc_handler = proc_dointvec,
1014 },
1015 {
1016 .procname = "bootloader_version",
1017 .data = &bootloader_version,
1018 .maxlen = sizeof (int),
1019 .mode = 0444,
1020 .proc_handler = proc_dointvec,
1021 },
1022 {
1023 .procname = "io_delay_type",
1024 .data = &io_delay_type,
1025 .maxlen = sizeof(int),
1026 .mode = 0644,
1027 .proc_handler = proc_dointvec,
1028 },
1029#endif
1030#if defined(CONFIG_MMU)
1031 {
1032 .procname = "randomize_va_space",
1033 .data = &randomize_va_space,
1034 .maxlen = sizeof(int),
1035 .mode = 0644,
1036 .proc_handler = proc_dointvec,
1037 },
1038#endif
1039#if defined(CONFIG_S390) && defined(CONFIG_SMP)
1040 {
1041 .procname = "spin_retry",
1042 .data = &spin_retry,
1043 .maxlen = sizeof (int),
1044 .mode = 0644,
1045 .proc_handler = proc_dointvec,
1046 },
1047#endif
1048#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1049 {
1050 .procname = "acpi_video_flags",
1051 .data = &acpi_realmode_flags,
1052 .maxlen = sizeof (unsigned long),
1053 .mode = 0644,
1054 .proc_handler = proc_doulongvec_minmax,
1055 },
1056#endif
1057#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1058 {
1059 .procname = "ignore-unaligned-usertrap",
1060 .data = &no_unaligned_warning,
1061 .maxlen = sizeof (int),
1062 .mode = 0644,
1063 .proc_handler = proc_dointvec,
1064 },
1065#endif
1066#ifdef CONFIG_IA64
1067 {
1068 .procname = "unaligned-dump-stack",
1069 .data = &unaligned_dump_stack,
1070 .maxlen = sizeof (int),
1071 .mode = 0644,
1072 .proc_handler = proc_dointvec,
1073 },
1074#endif
1075#ifdef CONFIG_DETECT_HUNG_TASK
1076 {
1077 .procname = "hung_task_panic",
1078 .data = &sysctl_hung_task_panic,
1079 .maxlen = sizeof(int),
1080 .mode = 0644,
1081 .proc_handler = proc_dointvec_minmax,
1082 .extra1 = &zero,
1083 .extra2 = &one,
1084 },
1085 {
1086 .procname = "hung_task_check_count",
1087 .data = &sysctl_hung_task_check_count,
1088 .maxlen = sizeof(int),
1089 .mode = 0644,
1090 .proc_handler = proc_dointvec_minmax,
1091 .extra1 = &zero,
1092 },
1093 {
1094 .procname = "hung_task_timeout_secs",
1095 .data = &sysctl_hung_task_timeout_secs,
1096 .maxlen = sizeof(unsigned long),
1097 .mode = 0644,
1098 .proc_handler = proc_dohung_task_timeout_secs,
1099 .extra2 = &hung_task_timeout_max,
1100 },
1101 {
1102 .procname = "hung_task_warnings",
1103 .data = &sysctl_hung_task_warnings,
1104 .maxlen = sizeof(int),
1105 .mode = 0644,
1106 .proc_handler = proc_dointvec_minmax,
1107 .extra1 = &neg_one,
1108 },
1109#endif
1110#ifdef CONFIG_RT_MUTEXES
1111 {
1112 .procname = "max_lock_depth",
1113 .data = &max_lock_depth,
1114 .maxlen = sizeof(int),
1115 .mode = 0644,
1116 .proc_handler = proc_dointvec,
1117 },
1118#endif
1119 {
1120 .procname = "poweroff_cmd",
1121 .data = &poweroff_cmd,
1122 .maxlen = POWEROFF_CMD_PATH_LEN,
1123 .mode = 0644,
1124 .proc_handler = proc_dostring,
1125 },
1126#ifdef CONFIG_KEYS
1127 {
1128 .procname = "keys",
1129 .mode = 0555,
1130 .child = key_sysctls,
1131 },
1132#endif
1133#ifdef CONFIG_PERF_EVENTS
1134 /*
1135 * User-space scripts rely on the existence of this file
1136 * as a feature check for perf_events being enabled.
1137 *
1138 * So it's an ABI, do not remove!
1139 */
1140 {
1141 .procname = "perf_event_paranoid",
1142 .data = &sysctl_perf_event_paranoid,
1143 .maxlen = sizeof(sysctl_perf_event_paranoid),
1144 .mode = 0644,
1145 .proc_handler = proc_dointvec,
1146 },
1147 {
1148 .procname = "perf_event_mlock_kb",
1149 .data = &sysctl_perf_event_mlock,
1150 .maxlen = sizeof(sysctl_perf_event_mlock),
1151 .mode = 0644,
1152 .proc_handler = proc_dointvec,
1153 },
1154 {
1155 .procname = "perf_event_max_sample_rate",
1156 .data = &sysctl_perf_event_sample_rate,
1157 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1158 .mode = 0644,
1159 .proc_handler = perf_proc_update_handler,
1160 .extra1 = &one,
1161 },
1162 {
1163 .procname = "perf_cpu_time_max_percent",
1164 .data = &sysctl_perf_cpu_time_max_percent,
1165 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1166 .mode = 0644,
1167 .proc_handler = perf_cpu_time_max_percent_handler,
1168 .extra1 = &zero,
1169 .extra2 = &one_hundred,
1170 },
1171 {
1172 .procname = "perf_event_max_stack",
1173 .data = &sysctl_perf_event_max_stack,
1174 .maxlen = sizeof(sysctl_perf_event_max_stack),
1175 .mode = 0644,
1176 .proc_handler = perf_event_max_stack_handler,
1177 .extra1 = &zero,
1178 .extra2 = &six_hundred_forty_kb,
1179 },
1180 {
1181 .procname = "perf_event_max_contexts_per_stack",
1182 .data = &sysctl_perf_event_max_contexts_per_stack,
1183 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
1184 .mode = 0644,
1185 .proc_handler = perf_event_max_stack_handler,
1186 .extra1 = &zero,
1187 .extra2 = &one_thousand,
1188 },
1189#endif
1190 {
1191 .procname = "panic_on_warn",
1192 .data = &panic_on_warn,
1193 .maxlen = sizeof(int),
1194 .mode = 0644,
1195 .proc_handler = proc_dointvec_minmax,
1196 .extra1 = &zero,
1197 .extra2 = &one,
1198 },
1199#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1200 {
1201 .procname = "timer_migration",
1202 .data = &sysctl_timer_migration,
1203 .maxlen = sizeof(unsigned int),
1204 .mode = 0644,
1205 .proc_handler = timer_migration_handler,
1206 .extra1 = &zero,
1207 .extra2 = &one,
1208 },
1209#endif
1210#ifdef CONFIG_BPF_SYSCALL
1211 {
1212 .procname = "unprivileged_bpf_disabled",
1213 .data = &sysctl_unprivileged_bpf_disabled,
1214 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
1215 .mode = 0644,
1216 /* only handle a transition from default "0" to "1" */
1217 .proc_handler = proc_dointvec_minmax,
1218 .extra1 = &one,
1219 .extra2 = &one,
1220 },
1221#endif
1222#if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1223 {
1224 .procname = "panic_on_rcu_stall",
1225 .data = &sysctl_panic_on_rcu_stall,
1226 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
1227 .mode = 0644,
1228 .proc_handler = proc_dointvec_minmax,
1229 .extra1 = &zero,
1230 .extra2 = &one,
1231 },
1232#endif
1233 { }
1234};
1235
1236static struct ctl_table vm_table[] = {
1237 {
1238 .procname = "overcommit_memory",
1239 .data = &sysctl_overcommit_memory,
1240 .maxlen = sizeof(sysctl_overcommit_memory),
1241 .mode = 0644,
1242 .proc_handler = proc_dointvec_minmax,
1243 .extra1 = &zero,
1244 .extra2 = &two,
1245 },
1246 {
1247 .procname = "panic_on_oom",
1248 .data = &sysctl_panic_on_oom,
1249 .maxlen = sizeof(sysctl_panic_on_oom),
1250 .mode = 0644,
1251 .proc_handler = proc_dointvec_minmax,
1252 .extra1 = &zero,
1253 .extra2 = &two,
1254 },
1255 {
1256 .procname = "oom_kill_allocating_task",
1257 .data = &sysctl_oom_kill_allocating_task,
1258 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1259 .mode = 0644,
1260 .proc_handler = proc_dointvec,
1261 },
1262 {
1263 .procname = "oom_dump_tasks",
1264 .data = &sysctl_oom_dump_tasks,
1265 .maxlen = sizeof(sysctl_oom_dump_tasks),
1266 .mode = 0644,
1267 .proc_handler = proc_dointvec,
1268 },
1269 {
1270 .procname = "overcommit_ratio",
1271 .data = &sysctl_overcommit_ratio,
1272 .maxlen = sizeof(sysctl_overcommit_ratio),
1273 .mode = 0644,
1274 .proc_handler = overcommit_ratio_handler,
1275 },
1276 {
1277 .procname = "overcommit_kbytes",
1278 .data = &sysctl_overcommit_kbytes,
1279 .maxlen = sizeof(sysctl_overcommit_kbytes),
1280 .mode = 0644,
1281 .proc_handler = overcommit_kbytes_handler,
1282 },
1283 {
1284 .procname = "page-cluster",
1285 .data = &page_cluster,
1286 .maxlen = sizeof(int),
1287 .mode = 0644,
1288 .proc_handler = proc_dointvec_minmax,
1289 .extra1 = &zero,
1290 },
1291 {
1292 .procname = "dirty_background_ratio",
1293 .data = &dirty_background_ratio,
1294 .maxlen = sizeof(dirty_background_ratio),
1295 .mode = 0644,
1296 .proc_handler = dirty_background_ratio_handler,
1297 .extra1 = &zero,
1298 .extra2 = &one_hundred,
1299 },
1300 {
1301 .procname = "dirty_background_bytes",
1302 .data = &dirty_background_bytes,
1303 .maxlen = sizeof(dirty_background_bytes),
1304 .mode = 0644,
1305 .proc_handler = dirty_background_bytes_handler,
1306 .extra1 = &one_ul,
1307 },
1308 {
1309 .procname = "dirty_ratio",
1310 .data = &vm_dirty_ratio,
1311 .maxlen = sizeof(vm_dirty_ratio),
1312 .mode = 0644,
1313 .proc_handler = dirty_ratio_handler,
1314 .extra1 = &zero,
1315 .extra2 = &one_hundred,
1316 },
1317 {
1318 .procname = "dirty_bytes",
1319 .data = &vm_dirty_bytes,
1320 .maxlen = sizeof(vm_dirty_bytes),
1321 .mode = 0644,
1322 .proc_handler = dirty_bytes_handler,
1323 .extra1 = &dirty_bytes_min,
1324 },
1325 {
1326 .procname = "dirty_writeback_centisecs",
1327 .data = &dirty_writeback_interval,
1328 .maxlen = sizeof(dirty_writeback_interval),
1329 .mode = 0644,
1330 .proc_handler = dirty_writeback_centisecs_handler,
1331 },
1332 {
1333 .procname = "dirty_expire_centisecs",
1334 .data = &dirty_expire_interval,
1335 .maxlen = sizeof(dirty_expire_interval),
1336 .mode = 0644,
1337 .proc_handler = proc_dointvec_minmax,
1338 .extra1 = &zero,
1339 },
1340 {
1341 .procname = "dirtytime_expire_seconds",
1342 .data = &dirtytime_expire_interval,
1343 .maxlen = sizeof(dirtytime_expire_interval),
1344 .mode = 0644,
1345 .proc_handler = dirtytime_interval_handler,
1346 .extra1 = &zero,
1347 },
1348 {
1349 .procname = "swappiness",
1350 .data = &vm_swappiness,
1351 .maxlen = sizeof(vm_swappiness),
1352 .mode = 0644,
1353 .proc_handler = proc_dointvec_minmax,
1354 .extra1 = &zero,
1355 .extra2 = &one_hundred,
1356 },
1357#ifdef CONFIG_HUGETLB_PAGE
1358 {
1359 .procname = "nr_hugepages",
1360 .data = NULL,
1361 .maxlen = sizeof(unsigned long),
1362 .mode = 0644,
1363 .proc_handler = hugetlb_sysctl_handler,
1364 },
1365#ifdef CONFIG_NUMA
1366 {
1367 .procname = "nr_hugepages_mempolicy",
1368 .data = NULL,
1369 .maxlen = sizeof(unsigned long),
1370 .mode = 0644,
1371 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1372 },
1373 {
1374 .procname = "numa_stat",
1375 .data = &sysctl_vm_numa_stat,
1376 .maxlen = sizeof(int),
1377 .mode = 0644,
1378 .proc_handler = sysctl_vm_numa_stat_handler,
1379 .extra1 = &zero,
1380 .extra2 = &one,
1381 },
1382#endif
1383 {
1384 .procname = "hugetlb_shm_group",
1385 .data = &sysctl_hugetlb_shm_group,
1386 .maxlen = sizeof(gid_t),
1387 .mode = 0644,
1388 .proc_handler = proc_dointvec,
1389 },
1390 {
1391 .procname = "nr_overcommit_hugepages",
1392 .data = NULL,
1393 .maxlen = sizeof(unsigned long),
1394 .mode = 0644,
1395 .proc_handler = hugetlb_overcommit_handler,
1396 },
1397#endif
1398 {
1399 .procname = "lowmem_reserve_ratio",
1400 .data = &sysctl_lowmem_reserve_ratio,
1401 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1402 .mode = 0644,
1403 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1404 },
1405 {
1406 .procname = "drop_caches",
1407 .data = &sysctl_drop_caches,
1408 .maxlen = sizeof(int),
1409 .mode = 0644,
1410 .proc_handler = drop_caches_sysctl_handler,
1411 .extra1 = &one,
1412 .extra2 = &four,
1413 },
1414#ifdef CONFIG_COMPACTION
1415 {
1416 .procname = "compact_memory",
1417 .data = &sysctl_compact_memory,
1418 .maxlen = sizeof(int),
1419 .mode = 0200,
1420 .proc_handler = sysctl_compaction_handler,
1421 },
1422 {
1423 .procname = "extfrag_threshold",
1424 .data = &sysctl_extfrag_threshold,
1425 .maxlen = sizeof(int),
1426 .mode = 0644,
1427 .proc_handler = sysctl_extfrag_handler,
1428 .extra1 = &min_extfrag_threshold,
1429 .extra2 = &max_extfrag_threshold,
1430 },
1431 {
1432 .procname = "compact_unevictable_allowed",
1433 .data = &sysctl_compact_unevictable_allowed,
1434 .maxlen = sizeof(int),
1435 .mode = 0644,
1436 .proc_handler = proc_dointvec,
1437 .extra1 = &zero,
1438 .extra2 = &one,
1439 },
1440
1441#endif /* CONFIG_COMPACTION */
1442 {
1443 .procname = "min_free_kbytes",
1444 .data = &min_free_kbytes,
1445 .maxlen = sizeof(min_free_kbytes),
1446 .mode = 0644,
1447 .proc_handler = min_free_kbytes_sysctl_handler,
1448 .extra1 = &zero,
1449 },
1450 {
1451 .procname = "watermark_scale_factor",
1452 .data = &watermark_scale_factor,
1453 .maxlen = sizeof(watermark_scale_factor),
1454 .mode = 0644,
1455 .proc_handler = watermark_scale_factor_sysctl_handler,
1456 .extra1 = &one,
1457 .extra2 = &one_thousand,
1458 },
1459 {
1460 .procname = "percpu_pagelist_fraction",
1461 .data = &percpu_pagelist_fraction,
1462 .maxlen = sizeof(percpu_pagelist_fraction),
1463 .mode = 0644,
1464 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1465 .extra1 = &zero,
1466 },
1467#ifdef CONFIG_MMU
1468 {
1469 .procname = "max_map_count",
1470 .data = &sysctl_max_map_count,
1471 .maxlen = sizeof(sysctl_max_map_count),
1472 .mode = 0644,
1473 .proc_handler = proc_dointvec_minmax,
1474 .extra1 = &zero,
1475 },
1476#else
1477 {
1478 .procname = "nr_trim_pages",
1479 .data = &sysctl_nr_trim_pages,
1480 .maxlen = sizeof(sysctl_nr_trim_pages),
1481 .mode = 0644,
1482 .proc_handler = proc_dointvec_minmax,
1483 .extra1 = &zero,
1484 },
1485#endif
1486 {
1487 .procname = "laptop_mode",
1488 .data = &laptop_mode,
1489 .maxlen = sizeof(laptop_mode),
1490 .mode = 0644,
1491 .proc_handler = proc_dointvec_jiffies,
1492 },
1493 {
1494 .procname = "block_dump",
1495 .data = &block_dump,
1496 .maxlen = sizeof(block_dump),
1497 .mode = 0644,
1498 .proc_handler = proc_dointvec,
1499 .extra1 = &zero,
1500 },
1501 {
1502 .procname = "vfs_cache_pressure",
1503 .data = &sysctl_vfs_cache_pressure,
1504 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1505 .mode = 0644,
1506 .proc_handler = proc_dointvec,
1507 .extra1 = &zero,
1508 },
1509#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1510 {
1511 .procname = "legacy_va_layout",
1512 .data = &sysctl_legacy_va_layout,
1513 .maxlen = sizeof(sysctl_legacy_va_layout),
1514 .mode = 0644,
1515 .proc_handler = proc_dointvec,
1516 .extra1 = &zero,
1517 },
1518#endif
1519#ifdef CONFIG_NUMA
1520 {
1521 .procname = "zone_reclaim_mode",
1522 .data = &node_reclaim_mode,
1523 .maxlen = sizeof(node_reclaim_mode),
1524 .mode = 0644,
1525 .proc_handler = proc_dointvec,
1526 .extra1 = &zero,
1527 },
1528 {
1529 .procname = "min_unmapped_ratio",
1530 .data = &sysctl_min_unmapped_ratio,
1531 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1532 .mode = 0644,
1533 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1534 .extra1 = &zero,
1535 .extra2 = &one_hundred,
1536 },
1537 {
1538 .procname = "min_slab_ratio",
1539 .data = &sysctl_min_slab_ratio,
1540 .maxlen = sizeof(sysctl_min_slab_ratio),
1541 .mode = 0644,
1542 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1543 .extra1 = &zero,
1544 .extra2 = &one_hundred,
1545 },
1546#endif
1547#ifdef CONFIG_SMP
1548 {
1549 .procname = "stat_interval",
1550 .data = &sysctl_stat_interval,
1551 .maxlen = sizeof(sysctl_stat_interval),
1552 .mode = 0644,
1553 .proc_handler = proc_dointvec_jiffies,
1554 },
1555 {
1556 .procname = "stat_refresh",
1557 .data = NULL,
1558 .maxlen = 0,
1559 .mode = 0600,
1560 .proc_handler = vmstat_refresh,
1561 },
1562#endif
1563#ifdef CONFIG_MMU
1564 {
1565 .procname = "mmap_min_addr",
1566 .data = &dac_mmap_min_addr,
1567 .maxlen = sizeof(unsigned long),
1568 .mode = 0644,
1569 .proc_handler = mmap_min_addr_handler,
1570 },
1571#endif
1572#ifdef CONFIG_NUMA
1573 {
1574 .procname = "numa_zonelist_order",
1575 .data = &numa_zonelist_order,
1576 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1577 .mode = 0644,
1578 .proc_handler = numa_zonelist_order_handler,
1579 },
1580#endif
1581#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1582 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1583 {
1584 .procname = "vdso_enabled",
1585#ifdef CONFIG_X86_32
1586 .data = &vdso32_enabled,
1587 .maxlen = sizeof(vdso32_enabled),
1588#else
1589 .data = &vdso_enabled,
1590 .maxlen = sizeof(vdso_enabled),
1591#endif
1592 .mode = 0644,
1593 .proc_handler = proc_dointvec,
1594 .extra1 = &zero,
1595 },
1596#endif
1597#ifdef CONFIG_HIGHMEM
1598 {
1599 .procname = "highmem_is_dirtyable",
1600 .data = &vm_highmem_is_dirtyable,
1601 .maxlen = sizeof(vm_highmem_is_dirtyable),
1602 .mode = 0644,
1603 .proc_handler = proc_dointvec_minmax,
1604 .extra1 = &zero,
1605 .extra2 = &one,
1606 },
1607#endif
1608#ifdef CONFIG_MEMORY_FAILURE
1609 {
1610 .procname = "memory_failure_early_kill",
1611 .data = &sysctl_memory_failure_early_kill,
1612 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1613 .mode = 0644,
1614 .proc_handler = proc_dointvec_minmax,
1615 .extra1 = &zero,
1616 .extra2 = &one,
1617 },
1618 {
1619 .procname = "memory_failure_recovery",
1620 .data = &sysctl_memory_failure_recovery,
1621 .maxlen = sizeof(sysctl_memory_failure_recovery),
1622 .mode = 0644,
1623 .proc_handler = proc_dointvec_minmax,
1624 .extra1 = &zero,
1625 .extra2 = &one,
1626 },
1627#endif
1628 {
1629 .procname = "user_reserve_kbytes",
1630 .data = &sysctl_user_reserve_kbytes,
1631 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1632 .mode = 0644,
1633 .proc_handler = proc_doulongvec_minmax,
1634 },
1635 {
1636 .procname = "admin_reserve_kbytes",
1637 .data = &sysctl_admin_reserve_kbytes,
1638 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1639 .mode = 0644,
1640 .proc_handler = proc_doulongvec_minmax,
1641 },
1642#ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1643 {
1644 .procname = "mmap_rnd_bits",
1645 .data = &mmap_rnd_bits,
1646 .maxlen = sizeof(mmap_rnd_bits),
1647 .mode = 0600,
1648 .proc_handler = proc_dointvec_minmax,
1649 .extra1 = (void *)&mmap_rnd_bits_min,
1650 .extra2 = (void *)&mmap_rnd_bits_max,
1651 },
1652#endif
1653#ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1654 {
1655 .procname = "mmap_rnd_compat_bits",
1656 .data = &mmap_rnd_compat_bits,
1657 .maxlen = sizeof(mmap_rnd_compat_bits),
1658 .mode = 0600,
1659 .proc_handler = proc_dointvec_minmax,
1660 .extra1 = (void *)&mmap_rnd_compat_bits_min,
1661 .extra2 = (void *)&mmap_rnd_compat_bits_max,
1662 },
1663#endif
1664 { }
1665};
1666
1667static struct ctl_table fs_table[] = {
1668 {
1669 .procname = "inode-nr",
1670 .data = &inodes_stat,
1671 .maxlen = 2*sizeof(long),
1672 .mode = 0444,
1673 .proc_handler = proc_nr_inodes,
1674 },
1675 {
1676 .procname = "inode-state",
1677 .data = &inodes_stat,
1678 .maxlen = 7*sizeof(long),
1679 .mode = 0444,
1680 .proc_handler = proc_nr_inodes,
1681 },
1682 {
1683 .procname = "file-nr",
1684 .data = &files_stat,
1685 .maxlen = sizeof(files_stat),
1686 .mode = 0444,
1687 .proc_handler = proc_nr_files,
1688 },
1689 {
1690 .procname = "file-max",
1691 .data = &files_stat.max_files,
1692 .maxlen = sizeof(files_stat.max_files),
1693 .mode = 0644,
1694 .proc_handler = proc_doulongvec_minmax,
1695 },
1696 {
1697 .procname = "nr_open",
1698 .data = &sysctl_nr_open,
1699 .maxlen = sizeof(unsigned int),
1700 .mode = 0644,
1701 .proc_handler = proc_dointvec_minmax,
1702 .extra1 = &sysctl_nr_open_min,
1703 .extra2 = &sysctl_nr_open_max,
1704 },
1705 {
1706 .procname = "dentry-state",
1707 .data = &dentry_stat,
1708 .maxlen = 6*sizeof(long),
1709 .mode = 0444,
1710 .proc_handler = proc_nr_dentry,
1711 },
1712 {
1713 .procname = "overflowuid",
1714 .data = &fs_overflowuid,
1715 .maxlen = sizeof(int),
1716 .mode = 0644,
1717 .proc_handler = proc_dointvec_minmax,
1718 .extra1 = &minolduid,
1719 .extra2 = &maxolduid,
1720 },
1721 {
1722 .procname = "overflowgid",
1723 .data = &fs_overflowgid,
1724 .maxlen = sizeof(int),
1725 .mode = 0644,
1726 .proc_handler = proc_dointvec_minmax,
1727 .extra1 = &minolduid,
1728 .extra2 = &maxolduid,
1729 },
1730#ifdef CONFIG_FILE_LOCKING
1731 {
1732 .procname = "leases-enable",
1733 .data = &leases_enable,
1734 .maxlen = sizeof(int),
1735 .mode = 0644,
1736 .proc_handler = proc_dointvec,
1737 },
1738#endif
1739#ifdef CONFIG_DNOTIFY
1740 {
1741 .procname = "dir-notify-enable",
1742 .data = &dir_notify_enable,
1743 .maxlen = sizeof(int),
1744 .mode = 0644,
1745 .proc_handler = proc_dointvec,
1746 },
1747#endif
1748#ifdef CONFIG_MMU
1749#ifdef CONFIG_FILE_LOCKING
1750 {
1751 .procname = "lease-break-time",
1752 .data = &lease_break_time,
1753 .maxlen = sizeof(int),
1754 .mode = 0644,
1755 .proc_handler = proc_dointvec,
1756 },
1757#endif
1758#ifdef CONFIG_AIO
1759 {
1760 .procname = "aio-nr",
1761 .data = &aio_nr,
1762 .maxlen = sizeof(aio_nr),
1763 .mode = 0444,
1764 .proc_handler = proc_doulongvec_minmax,
1765 },
1766 {
1767 .procname = "aio-max-nr",
1768 .data = &aio_max_nr,
1769 .maxlen = sizeof(aio_max_nr),
1770 .mode = 0644,
1771 .proc_handler = proc_doulongvec_minmax,
1772 },
1773#endif /* CONFIG_AIO */
1774#ifdef CONFIG_INOTIFY_USER
1775 {
1776 .procname = "inotify",
1777 .mode = 0555,
1778 .child = inotify_table,
1779 },
1780#endif
1781#ifdef CONFIG_EPOLL
1782 {
1783 .procname = "epoll",
1784 .mode = 0555,
1785 .child = epoll_table,
1786 },
1787#endif
1788#endif
1789 {
1790 .procname = "protected_symlinks",
1791 .data = &sysctl_protected_symlinks,
1792 .maxlen = sizeof(int),
1793 .mode = 0600,
1794 .proc_handler = proc_dointvec_minmax,
1795 .extra1 = &zero,
1796 .extra2 = &one,
1797 },
1798 {
1799 .procname = "protected_hardlinks",
1800 .data = &sysctl_protected_hardlinks,
1801 .maxlen = sizeof(int),
1802 .mode = 0600,
1803 .proc_handler = proc_dointvec_minmax,
1804 .extra1 = &zero,
1805 .extra2 = &one,
1806 },
1807 {
1808 .procname = "suid_dumpable",
1809 .data = &suid_dumpable,
1810 .maxlen = sizeof(int),
1811 .mode = 0644,
1812 .proc_handler = proc_dointvec_minmax_coredump,
1813 .extra1 = &zero,
1814 .extra2 = &two,
1815 },
1816#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1817 {
1818 .procname = "binfmt_misc",
1819 .mode = 0555,
1820 .child = sysctl_mount_point,
1821 },
1822#endif
1823 {
1824 .procname = "pipe-max-size",
1825 .data = &pipe_max_size,
1826 .maxlen = sizeof(pipe_max_size),
1827 .mode = 0644,
1828 .proc_handler = proc_dopipe_max_size,
1829 },
1830 {
1831 .procname = "pipe-user-pages-hard",
1832 .data = &pipe_user_pages_hard,
1833 .maxlen = sizeof(pipe_user_pages_hard),
1834 .mode = 0644,
1835 .proc_handler = proc_doulongvec_minmax,
1836 },
1837 {
1838 .procname = "pipe-user-pages-soft",
1839 .data = &pipe_user_pages_soft,
1840 .maxlen = sizeof(pipe_user_pages_soft),
1841 .mode = 0644,
1842 .proc_handler = proc_doulongvec_minmax,
1843 },
1844 {
1845 .procname = "mount-max",
1846 .data = &sysctl_mount_max,
1847 .maxlen = sizeof(unsigned int),
1848 .mode = 0644,
1849 .proc_handler = proc_dointvec_minmax,
1850 .extra1 = &one,
1851 },
1852 { }
1853};
1854
1855static struct ctl_table debug_table[] = {
1856#ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1857 {
1858 .procname = "exception-trace",
1859 .data = &show_unhandled_signals,
1860 .maxlen = sizeof(int),
1861 .mode = 0644,
1862 .proc_handler = proc_dointvec
1863 },
1864#endif
1865#if defined(CONFIG_OPTPROBES)
1866 {
1867 .procname = "kprobes-optimization",
1868 .data = &sysctl_kprobes_optimization,
1869 .maxlen = sizeof(int),
1870 .mode = 0644,
1871 .proc_handler = proc_kprobes_optimization_handler,
1872 .extra1 = &zero,
1873 .extra2 = &one,
1874 },
1875#endif
1876 { }
1877};
1878
1879static struct ctl_table dev_table[] = {
1880 { }
1881};
1882
1883int __init sysctl_init(void)
1884{
1885 struct ctl_table_header *hdr;
1886
1887 hdr = register_sysctl_table(sysctl_base_table);
1888 kmemleak_not_leak(hdr);
1889 return 0;
1890}
1891
1892#endif /* CONFIG_SYSCTL */
1893
1894/*
1895 * /proc/sys support
1896 */
1897
1898#ifdef CONFIG_PROC_SYSCTL
1899
1900static int _proc_do_string(char *data, int maxlen, int write,
1901 char __user *buffer,
1902 size_t *lenp, loff_t *ppos)
1903{
1904 size_t len;
1905 char __user *p;
1906 char c;
1907
1908 if (!data || !maxlen || !*lenp) {
1909 *lenp = 0;
1910 return 0;
1911 }
1912
1913 if (write) {
1914 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1915 /* Only continue writes not past the end of buffer. */
1916 len = strlen(data);
1917 if (len > maxlen - 1)
1918 len = maxlen - 1;
1919
1920 if (*ppos > len)
1921 return 0;
1922 len = *ppos;
1923 } else {
1924 /* Start writing from beginning of buffer. */
1925 len = 0;
1926 }
1927
1928 *ppos += *lenp;
1929 p = buffer;
1930 while ((p - buffer) < *lenp && len < maxlen - 1) {
1931 if (get_user(c, p++))
1932 return -EFAULT;
1933 if (c == 0 || c == '\n')
1934 break;
1935 data[len++] = c;
1936 }
1937 data[len] = 0;
1938 } else {
1939 len = strlen(data);
1940 if (len > maxlen)
1941 len = maxlen;
1942
1943 if (*ppos > len) {
1944 *lenp = 0;
1945 return 0;
1946 }
1947
1948 data += *ppos;
1949 len -= *ppos;
1950
1951 if (len > *lenp)
1952 len = *lenp;
1953 if (len)
1954 if (copy_to_user(buffer, data, len))
1955 return -EFAULT;
1956 if (len < *lenp) {
1957 if (put_user('\n', buffer + len))
1958 return -EFAULT;
1959 len++;
1960 }
1961 *lenp = len;
1962 *ppos += len;
1963 }
1964 return 0;
1965}
1966
1967static void warn_sysctl_write(struct ctl_table *table)
1968{
1969 pr_warn_once("%s wrote to %s when file position was not 0!\n"
1970 "This will not be supported in the future. To silence this\n"
1971 "warning, set kernel.sysctl_writes_strict = -1\n",
1972 current->comm, table->procname);
1973}
1974
1975/**
1976 * proc_first_pos_non_zero_ignore - check if firs position is allowed
1977 * @ppos: file position
1978 * @table: the sysctl table
1979 *
1980 * Returns true if the first position is non-zero and the sysctl_writes_strict
1981 * mode indicates this is not allowed for numeric input types. String proc
1982 * hadlers can ignore the return value.
1983 */
1984static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
1985 struct ctl_table *table)
1986{
1987 if (!*ppos)
1988 return false;
1989
1990 switch (sysctl_writes_strict) {
1991 case SYSCTL_WRITES_STRICT:
1992 return true;
1993 case SYSCTL_WRITES_WARN:
1994 warn_sysctl_write(table);
1995 return false;
1996 default:
1997 return false;
1998 }
1999}
2000
2001/**
2002 * proc_dostring - read a string sysctl
2003 * @table: the sysctl table
2004 * @write: %TRUE if this is a write to the sysctl file
2005 * @buffer: the user buffer
2006 * @lenp: the size of the user buffer
2007 * @ppos: file position
2008 *
2009 * Reads/writes a string from/to the user buffer. If the kernel
2010 * buffer provided is not large enough to hold the string, the
2011 * string is truncated. The copied string is %NULL-terminated.
2012 * If the string is being read by the user process, it is copied
2013 * and a newline '\n' is added. It is truncated if the buffer is
2014 * not large enough.
2015 *
2016 * Returns 0 on success.
2017 */
2018int proc_dostring(struct ctl_table *table, int write,
2019 void __user *buffer, size_t *lenp, loff_t *ppos)
2020{
2021 if (write)
2022 proc_first_pos_non_zero_ignore(ppos, table);
2023
2024 return _proc_do_string((char *)(table->data), table->maxlen, write,
2025 (char __user *)buffer, lenp, ppos);
2026}
2027
2028static size_t proc_skip_spaces(char **buf)
2029{
2030 size_t ret;
2031 char *tmp = skip_spaces(*buf);
2032 ret = tmp - *buf;
2033 *buf = tmp;
2034 return ret;
2035}
2036
2037static void proc_skip_char(char **buf, size_t *size, const char v)
2038{
2039 while (*size) {
2040 if (**buf != v)
2041 break;
2042 (*size)--;
2043 (*buf)++;
2044 }
2045}
2046
2047#define TMPBUFLEN 22
2048/**
2049 * proc_get_long - reads an ASCII formatted integer from a user buffer
2050 *
2051 * @buf: a kernel buffer
2052 * @size: size of the kernel buffer
2053 * @val: this is where the number will be stored
2054 * @neg: set to %TRUE if number is negative
2055 * @perm_tr: a vector which contains the allowed trailers
2056 * @perm_tr_len: size of the perm_tr vector
2057 * @tr: pointer to store the trailer character
2058 *
2059 * In case of success %0 is returned and @buf and @size are updated with
2060 * the amount of bytes read. If @tr is non-NULL and a trailing
2061 * character exists (size is non-zero after returning from this
2062 * function), @tr is updated with the trailing character.
2063 */
2064static int proc_get_long(char **buf, size_t *size,
2065 unsigned long *val, bool *neg,
2066 const char *perm_tr, unsigned perm_tr_len, char *tr)
2067{
2068 int len;
2069 char *p, tmp[TMPBUFLEN];
2070
2071 if (!*size)
2072 return -EINVAL;
2073
2074 len = *size;
2075 if (len > TMPBUFLEN - 1)
2076 len = TMPBUFLEN - 1;
2077
2078 memcpy(tmp, *buf, len);
2079
2080 tmp[len] = 0;
2081 p = tmp;
2082 if (*p == '-' && *size > 1) {
2083 *neg = true;
2084 p++;
2085 } else
2086 *neg = false;
2087 if (!isdigit(*p))
2088 return -EINVAL;
2089
2090 *val = simple_strtoul(p, &p, 0);
2091
2092 len = p - tmp;
2093
2094 /* We don't know if the next char is whitespace thus we may accept
2095 * invalid integers (e.g. 1234...a) or two integers instead of one
2096 * (e.g. 123...1). So lets not allow such large numbers. */
2097 if (len == TMPBUFLEN - 1)
2098 return -EINVAL;
2099
2100 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2101 return -EINVAL;
2102
2103 if (tr && (len < *size))
2104 *tr = *p;
2105
2106 *buf += len;
2107 *size -= len;
2108
2109 return 0;
2110}
2111
2112/**
2113 * proc_put_long - converts an integer to a decimal ASCII formatted string
2114 *
2115 * @buf: the user buffer
2116 * @size: the size of the user buffer
2117 * @val: the integer to be converted
2118 * @neg: sign of the number, %TRUE for negative
2119 *
2120 * In case of success %0 is returned and @buf and @size are updated with
2121 * the amount of bytes written.
2122 */
2123static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2124 bool neg)
2125{
2126 int len;
2127 char tmp[TMPBUFLEN], *p = tmp;
2128
2129 sprintf(p, "%s%lu", neg ? "-" : "", val);
2130 len = strlen(tmp);
2131 if (len > *size)
2132 len = *size;
2133 if (copy_to_user(*buf, tmp, len))
2134 return -EFAULT;
2135 *size -= len;
2136 *buf += len;
2137 return 0;
2138}
2139#undef TMPBUFLEN
2140
2141static int proc_put_char(void __user **buf, size_t *size, char c)
2142{
2143 if (*size) {
2144 char __user **buffer = (char __user **)buf;
2145 if (put_user(c, *buffer))
2146 return -EFAULT;
2147 (*size)--, (*buffer)++;
2148 *buf = *buffer;
2149 }
2150 return 0;
2151}
2152
2153static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2154 int *valp,
2155 int write, void *data)
2156{
2157 if (write) {
2158 if (*negp) {
2159 if (*lvalp > (unsigned long) INT_MAX + 1)
2160 return -EINVAL;
2161 *valp = -*lvalp;
2162 } else {
2163 if (*lvalp > (unsigned long) INT_MAX)
2164 return -EINVAL;
2165 *valp = *lvalp;
2166 }
2167 } else {
2168 int val = *valp;
2169 if (val < 0) {
2170 *negp = true;
2171 *lvalp = -(unsigned long)val;
2172 } else {
2173 *negp = false;
2174 *lvalp = (unsigned long)val;
2175 }
2176 }
2177 return 0;
2178}
2179
2180static int do_proc_douintvec_conv(unsigned long *lvalp,
2181 unsigned int *valp,
2182 int write, void *data)
2183{
2184 if (write) {
2185 if (*lvalp > UINT_MAX)
2186 return -EINVAL;
2187 *valp = *lvalp;
2188 } else {
2189 unsigned int val = *valp;
2190 *lvalp = (unsigned long)val;
2191 }
2192 return 0;
2193}
2194
2195static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2196
2197static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2198 int write, void __user *buffer,
2199 size_t *lenp, loff_t *ppos,
2200 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2201 int write, void *data),
2202 void *data)
2203{
2204 int *i, vleft, first = 1, err = 0;
2205 size_t left;
2206 char *kbuf = NULL, *p;
2207
2208 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2209 *lenp = 0;
2210 return 0;
2211 }
2212
2213 i = (int *) tbl_data;
2214 vleft = table->maxlen / sizeof(*i);
2215 left = *lenp;
2216
2217 if (!conv)
2218 conv = do_proc_dointvec_conv;
2219
2220 if (write) {
2221 if (proc_first_pos_non_zero_ignore(ppos, table))
2222 goto out;
2223
2224 if (left > PAGE_SIZE - 1)
2225 left = PAGE_SIZE - 1;
2226 p = kbuf = memdup_user_nul(buffer, left);
2227 if (IS_ERR(kbuf))
2228 return PTR_ERR(kbuf);
2229 }
2230
2231 for (; left && vleft--; i++, first=0) {
2232 unsigned long lval;
2233 bool neg;
2234
2235 if (write) {
2236 left -= proc_skip_spaces(&p);
2237
2238 if (!left)
2239 break;
2240 err = proc_get_long(&p, &left, &lval, &neg,
2241 proc_wspace_sep,
2242 sizeof(proc_wspace_sep), NULL);
2243 if (err)
2244 break;
2245 if (conv(&neg, &lval, i, 1, data)) {
2246 err = -EINVAL;
2247 break;
2248 }
2249 } else {
2250 if (conv(&neg, &lval, i, 0, data)) {
2251 err = -EINVAL;
2252 break;
2253 }
2254 if (!first)
2255 err = proc_put_char(&buffer, &left, '\t');
2256 if (err)
2257 break;
2258 err = proc_put_long(&buffer, &left, lval, neg);
2259 if (err)
2260 break;
2261 }
2262 }
2263
2264 if (!write && !first && left && !err)
2265 err = proc_put_char(&buffer, &left, '\n');
2266 if (write && !err && left)
2267 left -= proc_skip_spaces(&p);
2268 if (write) {
2269 kfree(kbuf);
2270 if (first)
2271 return err ? : -EINVAL;
2272 }
2273 *lenp -= left;
2274out:
2275 *ppos += *lenp;
2276 return err;
2277}
2278
2279static int do_proc_dointvec(struct ctl_table *table, int write,
2280 void __user *buffer, size_t *lenp, loff_t *ppos,
2281 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2282 int write, void *data),
2283 void *data)
2284{
2285 return __do_proc_dointvec(table->data, table, write,
2286 buffer, lenp, ppos, conv, data);
2287}
2288
2289static int do_proc_douintvec_w(unsigned int *tbl_data,
2290 struct ctl_table *table,
2291 void __user *buffer,
2292 size_t *lenp, loff_t *ppos,
2293 int (*conv)(unsigned long *lvalp,
2294 unsigned int *valp,
2295 int write, void *data),
2296 void *data)
2297{
2298 unsigned long lval;
2299 int err = 0;
2300 size_t left;
2301 bool neg;
2302 char *kbuf = NULL, *p;
2303
2304 left = *lenp;
2305
2306 if (proc_first_pos_non_zero_ignore(ppos, table))
2307 goto bail_early;
2308
2309 if (left > PAGE_SIZE - 1)
2310 left = PAGE_SIZE - 1;
2311
2312 p = kbuf = memdup_user_nul(buffer, left);
2313 if (IS_ERR(kbuf))
2314 return -EINVAL;
2315
2316 left -= proc_skip_spaces(&p);
2317 if (!left) {
2318 err = -EINVAL;
2319 goto out_free;
2320 }
2321
2322 err = proc_get_long(&p, &left, &lval, &neg,
2323 proc_wspace_sep,
2324 sizeof(proc_wspace_sep), NULL);
2325 if (err || neg) {
2326 err = -EINVAL;
2327 goto out_free;
2328 }
2329
2330 if (conv(&lval, tbl_data, 1, data)) {
2331 err = -EINVAL;
2332 goto out_free;
2333 }
2334
2335 if (!err && left)
2336 left -= proc_skip_spaces(&p);
2337
2338out_free:
2339 kfree(kbuf);
2340 if (err)
2341 return -EINVAL;
2342
2343 return 0;
2344
2345 /* This is in keeping with old __do_proc_dointvec() */
2346bail_early:
2347 *ppos += *lenp;
2348 return err;
2349}
2350
2351static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2352 size_t *lenp, loff_t *ppos,
2353 int (*conv)(unsigned long *lvalp,
2354 unsigned int *valp,
2355 int write, void *data),
2356 void *data)
2357{
2358 unsigned long lval;
2359 int err = 0;
2360 size_t left;
2361
2362 left = *lenp;
2363
2364 if (conv(&lval, tbl_data, 0, data)) {
2365 err = -EINVAL;
2366 goto out;
2367 }
2368
2369 err = proc_put_long(&buffer, &left, lval, false);
2370 if (err || !left)
2371 goto out;
2372
2373 err = proc_put_char(&buffer, &left, '\n');
2374
2375out:
2376 *lenp -= left;
2377 *ppos += *lenp;
2378
2379 return err;
2380}
2381
2382static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2383 int write, void __user *buffer,
2384 size_t *lenp, loff_t *ppos,
2385 int (*conv)(unsigned long *lvalp,
2386 unsigned int *valp,
2387 int write, void *data),
2388 void *data)
2389{
2390 unsigned int *i, vleft;
2391
2392 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2393 *lenp = 0;
2394 return 0;
2395 }
2396
2397 i = (unsigned int *) tbl_data;
2398 vleft = table->maxlen / sizeof(*i);
2399
2400 /*
2401 * Arrays are not supported, keep this simple. *Do not* add
2402 * support for them.
2403 */
2404 if (vleft != 1) {
2405 *lenp = 0;
2406 return -EINVAL;
2407 }
2408
2409 if (!conv)
2410 conv = do_proc_douintvec_conv;
2411
2412 if (write)
2413 return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2414 conv, data);
2415 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2416}
2417
2418static int do_proc_douintvec(struct ctl_table *table, int write,
2419 void __user *buffer, size_t *lenp, loff_t *ppos,
2420 int (*conv)(unsigned long *lvalp,
2421 unsigned int *valp,
2422 int write, void *data),
2423 void *data)
2424{
2425 return __do_proc_douintvec(table->data, table, write,
2426 buffer, lenp, ppos, conv, data);
2427}
2428
2429/**
2430 * proc_dointvec - read a vector of integers
2431 * @table: the sysctl table
2432 * @write: %TRUE if this is a write to the sysctl file
2433 * @buffer: the user buffer
2434 * @lenp: the size of the user buffer
2435 * @ppos: file position
2436 *
2437 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2438 * values from/to the user buffer, treated as an ASCII string.
2439 *
2440 * Returns 0 on success.
2441 */
2442int proc_dointvec(struct ctl_table *table, int write,
2443 void __user *buffer, size_t *lenp, loff_t *ppos)
2444{
2445 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2446}
2447
2448/**
2449 * proc_douintvec - read a vector of unsigned integers
2450 * @table: the sysctl table
2451 * @write: %TRUE if this is a write to the sysctl file
2452 * @buffer: the user buffer
2453 * @lenp: the size of the user buffer
2454 * @ppos: file position
2455 *
2456 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2457 * values from/to the user buffer, treated as an ASCII string.
2458 *
2459 * Returns 0 on success.
2460 */
2461int proc_douintvec(struct ctl_table *table, int write,
2462 void __user *buffer, size_t *lenp, loff_t *ppos)
2463{
2464 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2465 do_proc_douintvec_conv, NULL);
2466}
2467
2468/*
2469 * Taint values can only be increased
2470 * This means we can safely use a temporary.
2471 */
2472static int proc_taint(struct ctl_table *table, int write,
2473 void __user *buffer, size_t *lenp, loff_t *ppos)
2474{
2475 struct ctl_table t;
2476 unsigned long tmptaint = get_taint();
2477 int err;
2478
2479 if (write && !capable(CAP_SYS_ADMIN))
2480 return -EPERM;
2481
2482 t = *table;
2483 t.data = &tmptaint;
2484 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2485 if (err < 0)
2486 return err;
2487
2488 if (write) {
2489 /*
2490 * Poor man's atomic or. Not worth adding a primitive
2491 * to everyone's atomic.h for this
2492 */
2493 int i;
2494 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2495 if ((tmptaint >> i) & 1)
2496 add_taint(i, LOCKDEP_STILL_OK);
2497 }
2498 }
2499
2500 return err;
2501}
2502
2503#ifdef CONFIG_PRINTK
2504static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2505 void __user *buffer, size_t *lenp, loff_t *ppos)
2506{
2507 if (write && !capable(CAP_SYS_ADMIN))
2508 return -EPERM;
2509
2510 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2511}
2512#endif
2513
2514/**
2515 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
2516 * @min: pointer to minimum allowable value
2517 * @max: pointer to maximum allowable value
2518 *
2519 * The do_proc_dointvec_minmax_conv_param structure provides the
2520 * minimum and maximum values for doing range checking for those sysctl
2521 * parameters that use the proc_dointvec_minmax() handler.
2522 */
2523struct do_proc_dointvec_minmax_conv_param {
2524 int *min;
2525 int *max;
2526};
2527
2528static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2529 int *valp,
2530 int write, void *data)
2531{
2532 struct do_proc_dointvec_minmax_conv_param *param = data;
2533 if (write) {
2534 int val = *negp ? -*lvalp : *lvalp;
2535 if ((param->min && *param->min > val) ||
2536 (param->max && *param->max < val))
2537 return -EINVAL;
2538 *valp = val;
2539 } else {
2540 int val = *valp;
2541 if (val < 0) {
2542 *negp = true;
2543 *lvalp = -(unsigned long)val;
2544 } else {
2545 *negp = false;
2546 *lvalp = (unsigned long)val;
2547 }
2548 }
2549 return 0;
2550}
2551
2552/**
2553 * proc_dointvec_minmax - read a vector of integers with min/max values
2554 * @table: the sysctl table
2555 * @write: %TRUE if this is a write to the sysctl file
2556 * @buffer: the user buffer
2557 * @lenp: the size of the user buffer
2558 * @ppos: file position
2559 *
2560 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2561 * values from/to the user buffer, treated as an ASCII string.
2562 *
2563 * This routine will ensure the values are within the range specified by
2564 * table->extra1 (min) and table->extra2 (max).
2565 *
2566 * Returns 0 on success or -EINVAL on write when the range check fails.
2567 */
2568int proc_dointvec_minmax(struct ctl_table *table, int write,
2569 void __user *buffer, size_t *lenp, loff_t *ppos)
2570{
2571 struct do_proc_dointvec_minmax_conv_param param = {
2572 .min = (int *) table->extra1,
2573 .max = (int *) table->extra2,
2574 };
2575 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2576 do_proc_dointvec_minmax_conv, ¶m);
2577}
2578
2579/**
2580 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
2581 * @min: pointer to minimum allowable value
2582 * @max: pointer to maximum allowable value
2583 *
2584 * The do_proc_douintvec_minmax_conv_param structure provides the
2585 * minimum and maximum values for doing range checking for those sysctl
2586 * parameters that use the proc_douintvec_minmax() handler.
2587 */
2588struct do_proc_douintvec_minmax_conv_param {
2589 unsigned int *min;
2590 unsigned int *max;
2591};
2592
2593static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2594 unsigned int *valp,
2595 int write, void *data)
2596{
2597 struct do_proc_douintvec_minmax_conv_param *param = data;
2598
2599 if (write) {
2600 unsigned int val = *lvalp;
2601
2602 if (*lvalp > UINT_MAX)
2603 return -EINVAL;
2604
2605 if ((param->min && *param->min > val) ||
2606 (param->max && *param->max < val))
2607 return -ERANGE;
2608
2609 *valp = val;
2610 } else {
2611 unsigned int val = *valp;
2612 *lvalp = (unsigned long) val;
2613 }
2614
2615 return 0;
2616}
2617
2618/**
2619 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2620 * @table: the sysctl table
2621 * @write: %TRUE if this is a write to the sysctl file
2622 * @buffer: the user buffer
2623 * @lenp: the size of the user buffer
2624 * @ppos: file position
2625 *
2626 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2627 * values from/to the user buffer, treated as an ASCII string. Negative
2628 * strings are not allowed.
2629 *
2630 * This routine will ensure the values are within the range specified by
2631 * table->extra1 (min) and table->extra2 (max). There is a final sanity
2632 * check for UINT_MAX to avoid having to support wrap around uses from
2633 * userspace.
2634 *
2635 * Returns 0 on success or -ERANGE on write when the range check fails.
2636 */
2637int proc_douintvec_minmax(struct ctl_table *table, int write,
2638 void __user *buffer, size_t *lenp, loff_t *ppos)
2639{
2640 struct do_proc_douintvec_minmax_conv_param param = {
2641 .min = (unsigned int *) table->extra1,
2642 .max = (unsigned int *) table->extra2,
2643 };
2644 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2645 do_proc_douintvec_minmax_conv, ¶m);
2646}
2647
2648static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
2649 unsigned int *valp,
2650 int write, void *data)
2651{
2652 if (write) {
2653 unsigned int val;
2654
2655 val = round_pipe_size(*lvalp);
2656 if (val == 0)
2657 return -EINVAL;
2658
2659 *valp = val;
2660 } else {
2661 unsigned int val = *valp;
2662 *lvalp = (unsigned long) val;
2663 }
2664
2665 return 0;
2666}
2667
2668static int proc_dopipe_max_size(struct ctl_table *table, int write,
2669 void __user *buffer, size_t *lenp, loff_t *ppos)
2670{
2671 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2672 do_proc_dopipe_max_size_conv, NULL);
2673}
2674
2675static void validate_coredump_safety(void)
2676{
2677#ifdef CONFIG_COREDUMP
2678 if (suid_dumpable == SUID_DUMP_ROOT &&
2679 core_pattern[0] != '/' && core_pattern[0] != '|') {
2680 printk(KERN_WARNING
2681"Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2682"Pipe handler or fully qualified core dump path required.\n"
2683"Set kernel.core_pattern before fs.suid_dumpable.\n"
2684 );
2685 }
2686#endif
2687}
2688
2689static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2690 void __user *buffer, size_t *lenp, loff_t *ppos)
2691{
2692 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2693 if (!error)
2694 validate_coredump_safety();
2695 return error;
2696}
2697
2698#ifdef CONFIG_COREDUMP
2699static int proc_dostring_coredump(struct ctl_table *table, int write,
2700 void __user *buffer, size_t *lenp, loff_t *ppos)
2701{
2702 int error = proc_dostring(table, write, buffer, lenp, ppos);
2703 if (!error)
2704 validate_coredump_safety();
2705 return error;
2706}
2707#endif
2708
2709static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2710 void __user *buffer,
2711 size_t *lenp, loff_t *ppos,
2712 unsigned long convmul,
2713 unsigned long convdiv)
2714{
2715 unsigned long *i, *min, *max;
2716 int vleft, first = 1, err = 0;
2717 size_t left;
2718 char *kbuf = NULL, *p;
2719
2720 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2721 *lenp = 0;
2722 return 0;
2723 }
2724
2725 i = (unsigned long *) data;
2726 min = (unsigned long *) table->extra1;
2727 max = (unsigned long *) table->extra2;
2728 vleft = table->maxlen / sizeof(unsigned long);
2729 left = *lenp;
2730
2731 if (write) {
2732 if (proc_first_pos_non_zero_ignore(ppos, table))
2733 goto out;
2734
2735 if (left > PAGE_SIZE - 1)
2736 left = PAGE_SIZE - 1;
2737 p = kbuf = memdup_user_nul(buffer, left);
2738 if (IS_ERR(kbuf))
2739 return PTR_ERR(kbuf);
2740 }
2741
2742 for (; left && vleft--; i++, first = 0) {
2743 unsigned long val;
2744
2745 if (write) {
2746 bool neg;
2747
2748 left -= proc_skip_spaces(&p);
2749
2750 err = proc_get_long(&p, &left, &val, &neg,
2751 proc_wspace_sep,
2752 sizeof(proc_wspace_sep), NULL);
2753 if (err)
2754 break;
2755 if (neg)
2756 continue;
2757 val = convmul * val / convdiv;
2758 if ((min && val < *min) || (max && val > *max))
2759 continue;
2760 *i = val;
2761 } else {
2762 val = convdiv * (*i) / convmul;
2763 if (!first) {
2764 err = proc_put_char(&buffer, &left, '\t');
2765 if (err)
2766 break;
2767 }
2768 err = proc_put_long(&buffer, &left, val, false);
2769 if (err)
2770 break;
2771 }
2772 }
2773
2774 if (!write && !first && left && !err)
2775 err = proc_put_char(&buffer, &left, '\n');
2776 if (write && !err)
2777 left -= proc_skip_spaces(&p);
2778 if (write) {
2779 kfree(kbuf);
2780 if (first)
2781 return err ? : -EINVAL;
2782 }
2783 *lenp -= left;
2784out:
2785 *ppos += *lenp;
2786 return err;
2787}
2788
2789static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2790 void __user *buffer,
2791 size_t *lenp, loff_t *ppos,
2792 unsigned long convmul,
2793 unsigned long convdiv)
2794{
2795 return __do_proc_doulongvec_minmax(table->data, table, write,
2796 buffer, lenp, ppos, convmul, convdiv);
2797}
2798
2799/**
2800 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2801 * @table: the sysctl table
2802 * @write: %TRUE if this is a write to the sysctl file
2803 * @buffer: the user buffer
2804 * @lenp: the size of the user buffer
2805 * @ppos: file position
2806 *
2807 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2808 * values from/to the user buffer, treated as an ASCII string.
2809 *
2810 * This routine will ensure the values are within the range specified by
2811 * table->extra1 (min) and table->extra2 (max).
2812 *
2813 * Returns 0 on success.
2814 */
2815int proc_doulongvec_minmax(struct ctl_table *table, int write,
2816 void __user *buffer, size_t *lenp, loff_t *ppos)
2817{
2818 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2819}
2820
2821/**
2822 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2823 * @table: the sysctl table
2824 * @write: %TRUE if this is a write to the sysctl file
2825 * @buffer: the user buffer
2826 * @lenp: the size of the user buffer
2827 * @ppos: file position
2828 *
2829 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2830 * values from/to the user buffer, treated as an ASCII string. The values
2831 * are treated as milliseconds, and converted to jiffies when they are stored.
2832 *
2833 * This routine will ensure the values are within the range specified by
2834 * table->extra1 (min) and table->extra2 (max).
2835 *
2836 * Returns 0 on success.
2837 */
2838int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2839 void __user *buffer,
2840 size_t *lenp, loff_t *ppos)
2841{
2842 return do_proc_doulongvec_minmax(table, write, buffer,
2843 lenp, ppos, HZ, 1000l);
2844}
2845
2846
2847static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2848 int *valp,
2849 int write, void *data)
2850{
2851 if (write) {
2852 if (*lvalp > INT_MAX / HZ)
2853 return 1;
2854 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2855 } else {
2856 int val = *valp;
2857 unsigned long lval;
2858 if (val < 0) {
2859 *negp = true;
2860 lval = -(unsigned long)val;
2861 } else {
2862 *negp = false;
2863 lval = (unsigned long)val;
2864 }
2865 *lvalp = lval / HZ;
2866 }
2867 return 0;
2868}
2869
2870static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2871 int *valp,
2872 int write, void *data)
2873{
2874 if (write) {
2875 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2876 return 1;
2877 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2878 } else {
2879 int val = *valp;
2880 unsigned long lval;
2881 if (val < 0) {
2882 *negp = true;
2883 lval = -(unsigned long)val;
2884 } else {
2885 *negp = false;
2886 lval = (unsigned long)val;
2887 }
2888 *lvalp = jiffies_to_clock_t(lval);
2889 }
2890 return 0;
2891}
2892
2893static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2894 int *valp,
2895 int write, void *data)
2896{
2897 if (write) {
2898 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2899
2900 if (jif > INT_MAX)
2901 return 1;
2902 *valp = (int)jif;
2903 } else {
2904 int val = *valp;
2905 unsigned long lval;
2906 if (val < 0) {
2907 *negp = true;
2908 lval = -(unsigned long)val;
2909 } else {
2910 *negp = false;
2911 lval = (unsigned long)val;
2912 }
2913 *lvalp = jiffies_to_msecs(lval);
2914 }
2915 return 0;
2916}
2917
2918/**
2919 * proc_dointvec_jiffies - read a vector of integers as seconds
2920 * @table: the sysctl table
2921 * @write: %TRUE if this is a write to the sysctl file
2922 * @buffer: the user buffer
2923 * @lenp: the size of the user buffer
2924 * @ppos: file position
2925 *
2926 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2927 * values from/to the user buffer, treated as an ASCII string.
2928 * The values read are assumed to be in seconds, and are converted into
2929 * jiffies.
2930 *
2931 * Returns 0 on success.
2932 */
2933int proc_dointvec_jiffies(struct ctl_table *table, int write,
2934 void __user *buffer, size_t *lenp, loff_t *ppos)
2935{
2936 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2937 do_proc_dointvec_jiffies_conv,NULL);
2938}
2939
2940/**
2941 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2942 * @table: the sysctl table
2943 * @write: %TRUE if this is a write to the sysctl file
2944 * @buffer: the user buffer
2945 * @lenp: the size of the user buffer
2946 * @ppos: pointer to the file position
2947 *
2948 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2949 * values from/to the user buffer, treated as an ASCII string.
2950 * The values read are assumed to be in 1/USER_HZ seconds, and
2951 * are converted into jiffies.
2952 *
2953 * Returns 0 on success.
2954 */
2955int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2956 void __user *buffer, size_t *lenp, loff_t *ppos)
2957{
2958 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2959 do_proc_dointvec_userhz_jiffies_conv,NULL);
2960}
2961
2962/**
2963 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2964 * @table: the sysctl table
2965 * @write: %TRUE if this is a write to the sysctl file
2966 * @buffer: the user buffer
2967 * @lenp: the size of the user buffer
2968 * @ppos: file position
2969 * @ppos: the current position in the file
2970 *
2971 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2972 * values from/to the user buffer, treated as an ASCII string.
2973 * The values read are assumed to be in 1/1000 seconds, and
2974 * are converted into jiffies.
2975 *
2976 * Returns 0 on success.
2977 */
2978int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2979 void __user *buffer, size_t *lenp, loff_t *ppos)
2980{
2981 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2982 do_proc_dointvec_ms_jiffies_conv, NULL);
2983}
2984
2985static int proc_do_cad_pid(struct ctl_table *table, int write,
2986 void __user *buffer, size_t *lenp, loff_t *ppos)
2987{
2988 struct pid *new_pid;
2989 pid_t tmp;
2990 int r;
2991
2992 tmp = pid_vnr(cad_pid);
2993
2994 r = __do_proc_dointvec(&tmp, table, write, buffer,
2995 lenp, ppos, NULL, NULL);
2996 if (r || !write)
2997 return r;
2998
2999 new_pid = find_get_pid(tmp);
3000 if (!new_pid)
3001 return -ESRCH;
3002
3003 put_pid(xchg(&cad_pid, new_pid));
3004 return 0;
3005}
3006
3007/**
3008 * proc_do_large_bitmap - read/write from/to a large bitmap
3009 * @table: the sysctl table
3010 * @write: %TRUE if this is a write to the sysctl file
3011 * @buffer: the user buffer
3012 * @lenp: the size of the user buffer
3013 * @ppos: file position
3014 *
3015 * The bitmap is stored at table->data and the bitmap length (in bits)
3016 * in table->maxlen.
3017 *
3018 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
3019 * large bitmaps may be represented in a compact manner. Writing into
3020 * the file will clear the bitmap then update it with the given input.
3021 *
3022 * Returns 0 on success.
3023 */
3024int proc_do_large_bitmap(struct ctl_table *table, int write,
3025 void __user *buffer, size_t *lenp, loff_t *ppos)
3026{
3027 int err = 0;
3028 bool first = 1;
3029 size_t left = *lenp;
3030 unsigned long bitmap_len = table->maxlen;
3031 unsigned long *bitmap = *(unsigned long **) table->data;
3032 unsigned long *tmp_bitmap = NULL;
3033 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
3034
3035 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
3036 *lenp = 0;
3037 return 0;
3038 }
3039
3040 if (write) {
3041 char *kbuf, *p;
3042
3043 if (left > PAGE_SIZE - 1)
3044 left = PAGE_SIZE - 1;
3045
3046 p = kbuf = memdup_user_nul(buffer, left);
3047 if (IS_ERR(kbuf))
3048 return PTR_ERR(kbuf);
3049
3050 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
3051 GFP_KERNEL);
3052 if (!tmp_bitmap) {
3053 kfree(kbuf);
3054 return -ENOMEM;
3055 }
3056 proc_skip_char(&p, &left, '\n');
3057 while (!err && left) {
3058 unsigned long val_a, val_b;
3059 bool neg;
3060
3061 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3062 sizeof(tr_a), &c);
3063 if (err)
3064 break;
3065 if (val_a >= bitmap_len || neg) {
3066 err = -EINVAL;
3067 break;
3068 }
3069
3070 val_b = val_a;
3071 if (left) {
3072 p++;
3073 left--;
3074 }
3075
3076 if (c == '-') {
3077 err = proc_get_long(&p, &left, &val_b,
3078 &neg, tr_b, sizeof(tr_b),
3079 &c);
3080 if (err)
3081 break;
3082 if (val_b >= bitmap_len || neg ||
3083 val_a > val_b) {
3084 err = -EINVAL;
3085 break;
3086 }
3087 if (left) {
3088 p++;
3089 left--;
3090 }
3091 }
3092
3093 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3094 first = 0;
3095 proc_skip_char(&p, &left, '\n');
3096 }
3097 kfree(kbuf);
3098 } else {
3099 unsigned long bit_a, bit_b = 0;
3100
3101 while (left) {
3102 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3103 if (bit_a >= bitmap_len)
3104 break;
3105 bit_b = find_next_zero_bit(bitmap, bitmap_len,
3106 bit_a + 1) - 1;
3107
3108 if (!first) {
3109 err = proc_put_char(&buffer, &left, ',');
3110 if (err)
3111 break;
3112 }
3113 err = proc_put_long(&buffer, &left, bit_a, false);
3114 if (err)
3115 break;
3116 if (bit_a != bit_b) {
3117 err = proc_put_char(&buffer, &left, '-');
3118 if (err)
3119 break;
3120 err = proc_put_long(&buffer, &left, bit_b, false);
3121 if (err)
3122 break;
3123 }
3124
3125 first = 0; bit_b++;
3126 }
3127 if (!err)
3128 err = proc_put_char(&buffer, &left, '\n');
3129 }
3130
3131 if (!err) {
3132 if (write) {
3133 if (*ppos)
3134 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3135 else
3136 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3137 }
3138 *lenp -= left;
3139 *ppos += *lenp;
3140 }
3141
3142 kfree(tmp_bitmap);
3143 return err;
3144}
3145
3146#else /* CONFIG_PROC_SYSCTL */
3147
3148int proc_dostring(struct ctl_table *table, int write,
3149 void __user *buffer, size_t *lenp, loff_t *ppos)
3150{
3151 return -ENOSYS;
3152}
3153
3154int proc_dointvec(struct ctl_table *table, int write,
3155 void __user *buffer, size_t *lenp, loff_t *ppos)
3156{
3157 return -ENOSYS;
3158}
3159
3160int proc_douintvec(struct ctl_table *table, int write,
3161 void __user *buffer, size_t *lenp, loff_t *ppos)
3162{
3163 return -ENOSYS;
3164}
3165
3166int proc_dointvec_minmax(struct ctl_table *table, int write,
3167 void __user *buffer, size_t *lenp, loff_t *ppos)
3168{
3169 return -ENOSYS;
3170}
3171
3172int proc_douintvec_minmax(struct ctl_table *table, int write,
3173 void __user *buffer, size_t *lenp, loff_t *ppos)
3174{
3175 return -ENOSYS;
3176}
3177
3178int proc_dointvec_jiffies(struct ctl_table *table, int write,
3179 void __user *buffer, size_t *lenp, loff_t *ppos)
3180{
3181 return -ENOSYS;
3182}
3183
3184int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3185 void __user *buffer, size_t *lenp, loff_t *ppos)
3186{
3187 return -ENOSYS;
3188}
3189
3190int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3191 void __user *buffer, size_t *lenp, loff_t *ppos)
3192{
3193 return -ENOSYS;
3194}
3195
3196int proc_doulongvec_minmax(struct ctl_table *table, int write,
3197 void __user *buffer, size_t *lenp, loff_t *ppos)
3198{
3199 return -ENOSYS;
3200}
3201
3202int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3203 void __user *buffer,
3204 size_t *lenp, loff_t *ppos)
3205{
3206 return -ENOSYS;
3207}
3208
3209
3210#endif /* CONFIG_PROC_SYSCTL */
3211
3212/*
3213 * No sense putting this after each symbol definition, twice,
3214 * exception granted :-)
3215 */
3216EXPORT_SYMBOL(proc_dointvec);
3217EXPORT_SYMBOL(proc_douintvec);
3218EXPORT_SYMBOL(proc_dointvec_jiffies);
3219EXPORT_SYMBOL(proc_dointvec_minmax);
3220EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3221EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3222EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3223EXPORT_SYMBOL(proc_dostring);
3224EXPORT_SYMBOL(proc_doulongvec_minmax);
3225EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
1// SPDX-License-Identifier: GPL-2.0-only
2/*
3 * sysctl.c: General linux system control interface
4 *
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
13 * Horn.
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
17 * Wendling.
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
20 */
21
22#include <linux/module.h>
23#include <linux/aio.h>
24#include <linux/mm.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>
35#include <linux/fs.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>
71#include <linux/coredump.h>
72#include <linux/latencytop.h>
73#include <linux/pid.h>
74
75#include "../lib/kstrtox.h"
76
77#include <linux/uaccess.h>
78#include <asm/processor.h>
79
80#ifdef CONFIG_X86
81#include <asm/nmi.h>
82#include <asm/stacktrace.h>
83#include <asm/io.h>
84#endif
85#ifdef CONFIG_SPARC
86#include <asm/setup.h>
87#endif
88#ifdef CONFIG_BSD_PROCESS_ACCT
89#include <linux/acct.h>
90#endif
91#ifdef CONFIG_RT_MUTEXES
92#include <linux/rtmutex.h>
93#endif
94#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
95#include <linux/lockdep.h>
96#endif
97#ifdef CONFIG_CHR_DEV_SG
98#include <scsi/sg.h>
99#endif
100#ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
101#include <linux/stackleak.h>
102#endif
103#ifdef CONFIG_LOCKUP_DETECTOR
104#include <linux/nmi.h>
105#endif
106
107#if defined(CONFIG_SYSCTL)
108
109/* Constants used for minimum and maximum */
110#ifdef CONFIG_LOCKUP_DETECTOR
111static int sixty = 60;
112#endif
113
114static int __maybe_unused neg_one = -1;
115static int __maybe_unused two = 2;
116static int __maybe_unused four = 4;
117static unsigned long zero_ul;
118static unsigned long one_ul = 1;
119static unsigned long long_max = LONG_MAX;
120static int one_hundred = 100;
121static int two_hundred = 200;
122static int one_thousand = 1000;
123#ifdef CONFIG_PRINTK
124static int ten_thousand = 10000;
125#endif
126#ifdef CONFIG_PERF_EVENTS
127static int six_hundred_forty_kb = 640 * 1024;
128#endif
129
130/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
131static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
132
133/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
134static int maxolduid = 65535;
135static int minolduid;
136
137static int ngroups_max = NGROUPS_MAX;
138static const int cap_last_cap = CAP_LAST_CAP;
139
140/*
141 * This is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs
142 * and hung_task_check_interval_secs
143 */
144#ifdef CONFIG_DETECT_HUNG_TASK
145static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
146#endif
147
148#ifdef CONFIG_INOTIFY_USER
149#include <linux/inotify.h>
150#endif
151
152#ifdef CONFIG_PROC_SYSCTL
153
154/**
155 * enum sysctl_writes_mode - supported sysctl write modes
156 *
157 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
158 * to be written, and multiple writes on the same sysctl file descriptor
159 * will rewrite the sysctl value, regardless of file position. No warning
160 * is issued when the initial position is not 0.
161 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
162 * not 0.
163 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
164 * file position 0 and the value must be fully contained in the buffer
165 * sent to the write syscall. If dealing with strings respect the file
166 * position, but restrict this to the max length of the buffer, anything
167 * passed the max length will be ignored. Multiple writes will append
168 * to the buffer.
169 *
170 * These write modes control how current file position affects the behavior of
171 * updating sysctl values through the proc interface on each write.
172 */
173enum sysctl_writes_mode {
174 SYSCTL_WRITES_LEGACY = -1,
175 SYSCTL_WRITES_WARN = 0,
176 SYSCTL_WRITES_STRICT = 1,
177};
178
179static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
180#endif /* CONFIG_PROC_SYSCTL */
181
182#if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
183 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
184int sysctl_legacy_va_layout;
185#endif
186
187#ifdef CONFIG_SCHED_DEBUG
188static int min_sched_granularity_ns = 100000; /* 100 usecs */
189static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
190static int min_wakeup_granularity_ns; /* 0 usecs */
191static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
192#ifdef CONFIG_SMP
193static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
194static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
195#endif /* CONFIG_SMP */
196#endif /* CONFIG_SCHED_DEBUG */
197
198#ifdef CONFIG_COMPACTION
199static int min_extfrag_threshold;
200static int max_extfrag_threshold = 1000;
201#endif
202
203#endif /* CONFIG_SYSCTL */
204
205#if defined(CONFIG_BPF_SYSCALL) && defined(CONFIG_SYSCTL)
206static int bpf_stats_handler(struct ctl_table *table, int write,
207 void *buffer, size_t *lenp, loff_t *ppos)
208{
209 struct static_key *key = (struct static_key *)table->data;
210 static int saved_val;
211 int val, ret;
212 struct ctl_table tmp = {
213 .data = &val,
214 .maxlen = sizeof(val),
215 .mode = table->mode,
216 .extra1 = SYSCTL_ZERO,
217 .extra2 = SYSCTL_ONE,
218 };
219
220 if (write && !capable(CAP_SYS_ADMIN))
221 return -EPERM;
222
223 mutex_lock(&bpf_stats_enabled_mutex);
224 val = saved_val;
225 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
226 if (write && !ret && val != saved_val) {
227 if (val)
228 static_key_slow_inc(key);
229 else
230 static_key_slow_dec(key);
231 saved_val = val;
232 }
233 mutex_unlock(&bpf_stats_enabled_mutex);
234 return ret;
235}
236#endif
237
238/*
239 * /proc/sys support
240 */
241
242#ifdef CONFIG_PROC_SYSCTL
243
244static int _proc_do_string(char *data, int maxlen, int write,
245 char *buffer, size_t *lenp, loff_t *ppos)
246{
247 size_t len;
248 char c, *p;
249
250 if (!data || !maxlen || !*lenp) {
251 *lenp = 0;
252 return 0;
253 }
254
255 if (write) {
256 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
257 /* Only continue writes not past the end of buffer. */
258 len = strlen(data);
259 if (len > maxlen - 1)
260 len = maxlen - 1;
261
262 if (*ppos > len)
263 return 0;
264 len = *ppos;
265 } else {
266 /* Start writing from beginning of buffer. */
267 len = 0;
268 }
269
270 *ppos += *lenp;
271 p = buffer;
272 while ((p - buffer) < *lenp && len < maxlen - 1) {
273 c = *(p++);
274 if (c == 0 || c == '\n')
275 break;
276 data[len++] = c;
277 }
278 data[len] = 0;
279 } else {
280 len = strlen(data);
281 if (len > maxlen)
282 len = maxlen;
283
284 if (*ppos > len) {
285 *lenp = 0;
286 return 0;
287 }
288
289 data += *ppos;
290 len -= *ppos;
291
292 if (len > *lenp)
293 len = *lenp;
294 if (len)
295 memcpy(buffer, data, len);
296 if (len < *lenp) {
297 buffer[len] = '\n';
298 len++;
299 }
300 *lenp = len;
301 *ppos += len;
302 }
303 return 0;
304}
305
306static void warn_sysctl_write(struct ctl_table *table)
307{
308 pr_warn_once("%s wrote to %s when file position was not 0!\n"
309 "This will not be supported in the future. To silence this\n"
310 "warning, set kernel.sysctl_writes_strict = -1\n",
311 current->comm, table->procname);
312}
313
314/**
315 * proc_first_pos_non_zero_ignore - check if first position is allowed
316 * @ppos: file position
317 * @table: the sysctl table
318 *
319 * Returns true if the first position is non-zero and the sysctl_writes_strict
320 * mode indicates this is not allowed for numeric input types. String proc
321 * handlers can ignore the return value.
322 */
323static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
324 struct ctl_table *table)
325{
326 if (!*ppos)
327 return false;
328
329 switch (sysctl_writes_strict) {
330 case SYSCTL_WRITES_STRICT:
331 return true;
332 case SYSCTL_WRITES_WARN:
333 warn_sysctl_write(table);
334 return false;
335 default:
336 return false;
337 }
338}
339
340/**
341 * proc_dostring - read a string sysctl
342 * @table: the sysctl table
343 * @write: %TRUE if this is a write to the sysctl file
344 * @buffer: the user buffer
345 * @lenp: the size of the user buffer
346 * @ppos: file position
347 *
348 * Reads/writes a string from/to the user buffer. If the kernel
349 * buffer provided is not large enough to hold the string, the
350 * string is truncated. The copied string is %NULL-terminated.
351 * If the string is being read by the user process, it is copied
352 * and a newline '\n' is added. It is truncated if the buffer is
353 * not large enough.
354 *
355 * Returns 0 on success.
356 */
357int proc_dostring(struct ctl_table *table, int write,
358 void *buffer, size_t *lenp, loff_t *ppos)
359{
360 if (write)
361 proc_first_pos_non_zero_ignore(ppos, table);
362
363 return _proc_do_string(table->data, table->maxlen, write, buffer, lenp,
364 ppos);
365}
366
367static size_t proc_skip_spaces(char **buf)
368{
369 size_t ret;
370 char *tmp = skip_spaces(*buf);
371 ret = tmp - *buf;
372 *buf = tmp;
373 return ret;
374}
375
376static void proc_skip_char(char **buf, size_t *size, const char v)
377{
378 while (*size) {
379 if (**buf != v)
380 break;
381 (*size)--;
382 (*buf)++;
383 }
384}
385
386/**
387 * strtoul_lenient - parse an ASCII formatted integer from a buffer and only
388 * fail on overflow
389 *
390 * @cp: kernel buffer containing the string to parse
391 * @endp: pointer to store the trailing characters
392 * @base: the base to use
393 * @res: where the parsed integer will be stored
394 *
395 * In case of success 0 is returned and @res will contain the parsed integer,
396 * @endp will hold any trailing characters.
397 * This function will fail the parse on overflow. If there wasn't an overflow
398 * the function will defer the decision what characters count as invalid to the
399 * caller.
400 */
401static int strtoul_lenient(const char *cp, char **endp, unsigned int base,
402 unsigned long *res)
403{
404 unsigned long long result;
405 unsigned int rv;
406
407 cp = _parse_integer_fixup_radix(cp, &base);
408 rv = _parse_integer(cp, base, &result);
409 if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result))
410 return -ERANGE;
411
412 cp += rv;
413
414 if (endp)
415 *endp = (char *)cp;
416
417 *res = (unsigned long)result;
418 return 0;
419}
420
421#define TMPBUFLEN 22
422/**
423 * proc_get_long - reads an ASCII formatted integer from a user buffer
424 *
425 * @buf: a kernel buffer
426 * @size: size of the kernel buffer
427 * @val: this is where the number will be stored
428 * @neg: set to %TRUE if number is negative
429 * @perm_tr: a vector which contains the allowed trailers
430 * @perm_tr_len: size of the perm_tr vector
431 * @tr: pointer to store the trailer character
432 *
433 * In case of success %0 is returned and @buf and @size are updated with
434 * the amount of bytes read. If @tr is non-NULL and a trailing
435 * character exists (size is non-zero after returning from this
436 * function), @tr is updated with the trailing character.
437 */
438static int proc_get_long(char **buf, size_t *size,
439 unsigned long *val, bool *neg,
440 const char *perm_tr, unsigned perm_tr_len, char *tr)
441{
442 int len;
443 char *p, tmp[TMPBUFLEN];
444
445 if (!*size)
446 return -EINVAL;
447
448 len = *size;
449 if (len > TMPBUFLEN - 1)
450 len = TMPBUFLEN - 1;
451
452 memcpy(tmp, *buf, len);
453
454 tmp[len] = 0;
455 p = tmp;
456 if (*p == '-' && *size > 1) {
457 *neg = true;
458 p++;
459 } else
460 *neg = false;
461 if (!isdigit(*p))
462 return -EINVAL;
463
464 if (strtoul_lenient(p, &p, 0, val))
465 return -EINVAL;
466
467 len = p - tmp;
468
469 /* We don't know if the next char is whitespace thus we may accept
470 * invalid integers (e.g. 1234...a) or two integers instead of one
471 * (e.g. 123...1). So lets not allow such large numbers. */
472 if (len == TMPBUFLEN - 1)
473 return -EINVAL;
474
475 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
476 return -EINVAL;
477
478 if (tr && (len < *size))
479 *tr = *p;
480
481 *buf += len;
482 *size -= len;
483
484 return 0;
485}
486
487/**
488 * proc_put_long - converts an integer to a decimal ASCII formatted string
489 *
490 * @buf: the user buffer
491 * @size: the size of the user buffer
492 * @val: the integer to be converted
493 * @neg: sign of the number, %TRUE for negative
494 *
495 * In case of success @buf and @size are updated with the amount of bytes
496 * written.
497 */
498static void proc_put_long(void **buf, size_t *size, unsigned long val, bool neg)
499{
500 int len;
501 char tmp[TMPBUFLEN], *p = tmp;
502
503 sprintf(p, "%s%lu", neg ? "-" : "", val);
504 len = strlen(tmp);
505 if (len > *size)
506 len = *size;
507 memcpy(*buf, tmp, len);
508 *size -= len;
509 *buf += len;
510}
511#undef TMPBUFLEN
512
513static void proc_put_char(void **buf, size_t *size, char c)
514{
515 if (*size) {
516 char **buffer = (char **)buf;
517 **buffer = c;
518
519 (*size)--;
520 (*buffer)++;
521 *buf = *buffer;
522 }
523}
524
525static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
526 int *valp,
527 int write, void *data)
528{
529 if (write) {
530 if (*negp) {
531 if (*lvalp > (unsigned long) INT_MAX + 1)
532 return -EINVAL;
533 *valp = -*lvalp;
534 } else {
535 if (*lvalp > (unsigned long) INT_MAX)
536 return -EINVAL;
537 *valp = *lvalp;
538 }
539 } else {
540 int val = *valp;
541 if (val < 0) {
542 *negp = true;
543 *lvalp = -(unsigned long)val;
544 } else {
545 *negp = false;
546 *lvalp = (unsigned long)val;
547 }
548 }
549 return 0;
550}
551
552static int do_proc_douintvec_conv(unsigned long *lvalp,
553 unsigned int *valp,
554 int write, void *data)
555{
556 if (write) {
557 if (*lvalp > UINT_MAX)
558 return -EINVAL;
559 *valp = *lvalp;
560 } else {
561 unsigned int val = *valp;
562 *lvalp = (unsigned long)val;
563 }
564 return 0;
565}
566
567static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
568
569static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
570 int write, void *buffer,
571 size_t *lenp, loff_t *ppos,
572 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
573 int write, void *data),
574 void *data)
575{
576 int *i, vleft, first = 1, err = 0;
577 size_t left;
578 char *p;
579
580 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
581 *lenp = 0;
582 return 0;
583 }
584
585 i = (int *) tbl_data;
586 vleft = table->maxlen / sizeof(*i);
587 left = *lenp;
588
589 if (!conv)
590 conv = do_proc_dointvec_conv;
591
592 if (write) {
593 if (proc_first_pos_non_zero_ignore(ppos, table))
594 goto out;
595
596 if (left > PAGE_SIZE - 1)
597 left = PAGE_SIZE - 1;
598 p = buffer;
599 }
600
601 for (; left && vleft--; i++, first=0) {
602 unsigned long lval;
603 bool neg;
604
605 if (write) {
606 left -= proc_skip_spaces(&p);
607
608 if (!left)
609 break;
610 err = proc_get_long(&p, &left, &lval, &neg,
611 proc_wspace_sep,
612 sizeof(proc_wspace_sep), NULL);
613 if (err)
614 break;
615 if (conv(&neg, &lval, i, 1, data)) {
616 err = -EINVAL;
617 break;
618 }
619 } else {
620 if (conv(&neg, &lval, i, 0, data)) {
621 err = -EINVAL;
622 break;
623 }
624 if (!first)
625 proc_put_char(&buffer, &left, '\t');
626 proc_put_long(&buffer, &left, lval, neg);
627 }
628 }
629
630 if (!write && !first && left && !err)
631 proc_put_char(&buffer, &left, '\n');
632 if (write && !err && left)
633 left -= proc_skip_spaces(&p);
634 if (write && first)
635 return err ? : -EINVAL;
636 *lenp -= left;
637out:
638 *ppos += *lenp;
639 return err;
640}
641
642static int do_proc_dointvec(struct ctl_table *table, int write,
643 void *buffer, size_t *lenp, loff_t *ppos,
644 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
645 int write, void *data),
646 void *data)
647{
648 return __do_proc_dointvec(table->data, table, write,
649 buffer, lenp, ppos, conv, data);
650}
651
652static int do_proc_douintvec_w(unsigned int *tbl_data,
653 struct ctl_table *table,
654 void *buffer,
655 size_t *lenp, loff_t *ppos,
656 int (*conv)(unsigned long *lvalp,
657 unsigned int *valp,
658 int write, void *data),
659 void *data)
660{
661 unsigned long lval;
662 int err = 0;
663 size_t left;
664 bool neg;
665 char *p = buffer;
666
667 left = *lenp;
668
669 if (proc_first_pos_non_zero_ignore(ppos, table))
670 goto bail_early;
671
672 if (left > PAGE_SIZE - 1)
673 left = PAGE_SIZE - 1;
674
675 left -= proc_skip_spaces(&p);
676 if (!left) {
677 err = -EINVAL;
678 goto out_free;
679 }
680
681 err = proc_get_long(&p, &left, &lval, &neg,
682 proc_wspace_sep,
683 sizeof(proc_wspace_sep), NULL);
684 if (err || neg) {
685 err = -EINVAL;
686 goto out_free;
687 }
688
689 if (conv(&lval, tbl_data, 1, data)) {
690 err = -EINVAL;
691 goto out_free;
692 }
693
694 if (!err && left)
695 left -= proc_skip_spaces(&p);
696
697out_free:
698 if (err)
699 return -EINVAL;
700
701 return 0;
702
703 /* This is in keeping with old __do_proc_dointvec() */
704bail_early:
705 *ppos += *lenp;
706 return err;
707}
708
709static int do_proc_douintvec_r(unsigned int *tbl_data, void *buffer,
710 size_t *lenp, loff_t *ppos,
711 int (*conv)(unsigned long *lvalp,
712 unsigned int *valp,
713 int write, void *data),
714 void *data)
715{
716 unsigned long lval;
717 int err = 0;
718 size_t left;
719
720 left = *lenp;
721
722 if (conv(&lval, tbl_data, 0, data)) {
723 err = -EINVAL;
724 goto out;
725 }
726
727 proc_put_long(&buffer, &left, lval, false);
728 if (!left)
729 goto out;
730
731 proc_put_char(&buffer, &left, '\n');
732
733out:
734 *lenp -= left;
735 *ppos += *lenp;
736
737 return err;
738}
739
740static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
741 int write, void *buffer,
742 size_t *lenp, loff_t *ppos,
743 int (*conv)(unsigned long *lvalp,
744 unsigned int *valp,
745 int write, void *data),
746 void *data)
747{
748 unsigned int *i, vleft;
749
750 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
751 *lenp = 0;
752 return 0;
753 }
754
755 i = (unsigned int *) tbl_data;
756 vleft = table->maxlen / sizeof(*i);
757
758 /*
759 * Arrays are not supported, keep this simple. *Do not* add
760 * support for them.
761 */
762 if (vleft != 1) {
763 *lenp = 0;
764 return -EINVAL;
765 }
766
767 if (!conv)
768 conv = do_proc_douintvec_conv;
769
770 if (write)
771 return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
772 conv, data);
773 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
774}
775
776static int do_proc_douintvec(struct ctl_table *table, int write,
777 void *buffer, size_t *lenp, loff_t *ppos,
778 int (*conv)(unsigned long *lvalp,
779 unsigned int *valp,
780 int write, void *data),
781 void *data)
782{
783 return __do_proc_douintvec(table->data, table, write,
784 buffer, lenp, ppos, conv, data);
785}
786
787/**
788 * proc_dointvec - read a vector of integers
789 * @table: the sysctl table
790 * @write: %TRUE if this is a write to the sysctl file
791 * @buffer: the user buffer
792 * @lenp: the size of the user buffer
793 * @ppos: file position
794 *
795 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
796 * values from/to the user buffer, treated as an ASCII string.
797 *
798 * Returns 0 on success.
799 */
800int proc_dointvec(struct ctl_table *table, int write, void *buffer,
801 size_t *lenp, loff_t *ppos)
802{
803 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
804}
805
806#ifdef CONFIG_COMPACTION
807static int proc_dointvec_minmax_warn_RT_change(struct ctl_table *table,
808 int write, void *buffer, size_t *lenp, loff_t *ppos)
809{
810 int ret, old;
811
812 if (!IS_ENABLED(CONFIG_PREEMPT_RT) || !write)
813 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
814
815 old = *(int *)table->data;
816 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
817 if (ret)
818 return ret;
819 if (old != *(int *)table->data)
820 pr_warn_once("sysctl attribute %s changed by %s[%d]\n",
821 table->procname, current->comm,
822 task_pid_nr(current));
823 return ret;
824}
825#endif
826
827/**
828 * proc_douintvec - read a vector of unsigned integers
829 * @table: the sysctl table
830 * @write: %TRUE if this is a write to the sysctl file
831 * @buffer: the user buffer
832 * @lenp: the size of the user buffer
833 * @ppos: file position
834 *
835 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
836 * values from/to the user buffer, treated as an ASCII string.
837 *
838 * Returns 0 on success.
839 */
840int proc_douintvec(struct ctl_table *table, int write, void *buffer,
841 size_t *lenp, loff_t *ppos)
842{
843 return do_proc_douintvec(table, write, buffer, lenp, ppos,
844 do_proc_douintvec_conv, NULL);
845}
846
847/*
848 * Taint values can only be increased
849 * This means we can safely use a temporary.
850 */
851static int proc_taint(struct ctl_table *table, int write,
852 void *buffer, size_t *lenp, loff_t *ppos)
853{
854 struct ctl_table t;
855 unsigned long tmptaint = get_taint();
856 int err;
857
858 if (write && !capable(CAP_SYS_ADMIN))
859 return -EPERM;
860
861 t = *table;
862 t.data = &tmptaint;
863 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
864 if (err < 0)
865 return err;
866
867 if (write) {
868 int i;
869
870 /*
871 * If we are relying on panic_on_taint not producing
872 * false positives due to userspace input, bail out
873 * before setting the requested taint flags.
874 */
875 if (panic_on_taint_nousertaint && (tmptaint & panic_on_taint))
876 return -EINVAL;
877
878 /*
879 * Poor man's atomic or. Not worth adding a primitive
880 * to everyone's atomic.h for this
881 */
882 for (i = 0; i < TAINT_FLAGS_COUNT; i++)
883 if ((1UL << i) & tmptaint)
884 add_taint(i, LOCKDEP_STILL_OK);
885 }
886
887 return err;
888}
889
890#ifdef CONFIG_PRINTK
891static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
892 void *buffer, size_t *lenp, loff_t *ppos)
893{
894 if (write && !capable(CAP_SYS_ADMIN))
895 return -EPERM;
896
897 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
898}
899#endif
900
901/**
902 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
903 * @min: pointer to minimum allowable value
904 * @max: pointer to maximum allowable value
905 *
906 * The do_proc_dointvec_minmax_conv_param structure provides the
907 * minimum and maximum values for doing range checking for those sysctl
908 * parameters that use the proc_dointvec_minmax() handler.
909 */
910struct do_proc_dointvec_minmax_conv_param {
911 int *min;
912 int *max;
913};
914
915static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
916 int *valp,
917 int write, void *data)
918{
919 int tmp, ret;
920 struct do_proc_dointvec_minmax_conv_param *param = data;
921 /*
922 * If writing, first do so via a temporary local int so we can
923 * bounds-check it before touching *valp.
924 */
925 int *ip = write ? &tmp : valp;
926
927 ret = do_proc_dointvec_conv(negp, lvalp, ip, write, data);
928 if (ret)
929 return ret;
930
931 if (write) {
932 if ((param->min && *param->min > tmp) ||
933 (param->max && *param->max < tmp))
934 return -EINVAL;
935 *valp = tmp;
936 }
937
938 return 0;
939}
940
941/**
942 * proc_dointvec_minmax - read a vector of integers with min/max values
943 * @table: the sysctl table
944 * @write: %TRUE if this is a write to the sysctl file
945 * @buffer: the user buffer
946 * @lenp: the size of the user buffer
947 * @ppos: file position
948 *
949 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
950 * values from/to the user buffer, treated as an ASCII string.
951 *
952 * This routine will ensure the values are within the range specified by
953 * table->extra1 (min) and table->extra2 (max).
954 *
955 * Returns 0 on success or -EINVAL on write when the range check fails.
956 */
957int proc_dointvec_minmax(struct ctl_table *table, int write,
958 void *buffer, size_t *lenp, loff_t *ppos)
959{
960 struct do_proc_dointvec_minmax_conv_param param = {
961 .min = (int *) table->extra1,
962 .max = (int *) table->extra2,
963 };
964 return do_proc_dointvec(table, write, buffer, lenp, ppos,
965 do_proc_dointvec_minmax_conv, ¶m);
966}
967
968/**
969 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
970 * @min: pointer to minimum allowable value
971 * @max: pointer to maximum allowable value
972 *
973 * The do_proc_douintvec_minmax_conv_param structure provides the
974 * minimum and maximum values for doing range checking for those sysctl
975 * parameters that use the proc_douintvec_minmax() handler.
976 */
977struct do_proc_douintvec_minmax_conv_param {
978 unsigned int *min;
979 unsigned int *max;
980};
981
982static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
983 unsigned int *valp,
984 int write, void *data)
985{
986 int ret;
987 unsigned int tmp;
988 struct do_proc_douintvec_minmax_conv_param *param = data;
989 /* write via temporary local uint for bounds-checking */
990 unsigned int *up = write ? &tmp : valp;
991
992 ret = do_proc_douintvec_conv(lvalp, up, write, data);
993 if (ret)
994 return ret;
995
996 if (write) {
997 if ((param->min && *param->min > tmp) ||
998 (param->max && *param->max < tmp))
999 return -ERANGE;
1000
1001 *valp = tmp;
1002 }
1003
1004 return 0;
1005}
1006
1007/**
1008 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
1009 * @table: the sysctl table
1010 * @write: %TRUE if this is a write to the sysctl file
1011 * @buffer: the user buffer
1012 * @lenp: the size of the user buffer
1013 * @ppos: file position
1014 *
1015 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
1016 * values from/to the user buffer, treated as an ASCII string. Negative
1017 * strings are not allowed.
1018 *
1019 * This routine will ensure the values are within the range specified by
1020 * table->extra1 (min) and table->extra2 (max). There is a final sanity
1021 * check for UINT_MAX to avoid having to support wrap around uses from
1022 * userspace.
1023 *
1024 * Returns 0 on success or -ERANGE on write when the range check fails.
1025 */
1026int proc_douintvec_minmax(struct ctl_table *table, int write,
1027 void *buffer, size_t *lenp, loff_t *ppos)
1028{
1029 struct do_proc_douintvec_minmax_conv_param param = {
1030 .min = (unsigned int *) table->extra1,
1031 .max = (unsigned int *) table->extra2,
1032 };
1033 return do_proc_douintvec(table, write, buffer, lenp, ppos,
1034 do_proc_douintvec_minmax_conv, ¶m);
1035}
1036
1037static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
1038 unsigned int *valp,
1039 int write, void *data)
1040{
1041 if (write) {
1042 unsigned int val;
1043
1044 val = round_pipe_size(*lvalp);
1045 if (val == 0)
1046 return -EINVAL;
1047
1048 *valp = val;
1049 } else {
1050 unsigned int val = *valp;
1051 *lvalp = (unsigned long) val;
1052 }
1053
1054 return 0;
1055}
1056
1057static int proc_dopipe_max_size(struct ctl_table *table, int write,
1058 void *buffer, size_t *lenp, loff_t *ppos)
1059{
1060 return do_proc_douintvec(table, write, buffer, lenp, ppos,
1061 do_proc_dopipe_max_size_conv, NULL);
1062}
1063
1064static void validate_coredump_safety(void)
1065{
1066#ifdef CONFIG_COREDUMP
1067 if (suid_dumpable == SUID_DUMP_ROOT &&
1068 core_pattern[0] != '/' && core_pattern[0] != '|') {
1069 printk(KERN_WARNING
1070"Unsafe core_pattern used with fs.suid_dumpable=2.\n"
1071"Pipe handler or fully qualified core dump path required.\n"
1072"Set kernel.core_pattern before fs.suid_dumpable.\n"
1073 );
1074 }
1075#endif
1076}
1077
1078static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
1079 void *buffer, size_t *lenp, loff_t *ppos)
1080{
1081 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
1082 if (!error)
1083 validate_coredump_safety();
1084 return error;
1085}
1086
1087#ifdef CONFIG_COREDUMP
1088static int proc_dostring_coredump(struct ctl_table *table, int write,
1089 void *buffer, size_t *lenp, loff_t *ppos)
1090{
1091 int error = proc_dostring(table, write, buffer, lenp, ppos);
1092 if (!error)
1093 validate_coredump_safety();
1094 return error;
1095}
1096#endif
1097
1098#ifdef CONFIG_MAGIC_SYSRQ
1099static int sysrq_sysctl_handler(struct ctl_table *table, int write,
1100 void *buffer, size_t *lenp, loff_t *ppos)
1101{
1102 int tmp, ret;
1103
1104 tmp = sysrq_mask();
1105
1106 ret = __do_proc_dointvec(&tmp, table, write, buffer,
1107 lenp, ppos, NULL, NULL);
1108 if (ret || !write)
1109 return ret;
1110
1111 if (write)
1112 sysrq_toggle_support(tmp);
1113
1114 return 0;
1115}
1116#endif
1117
1118static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table,
1119 int write, void *buffer, size_t *lenp, loff_t *ppos,
1120 unsigned long convmul, unsigned long convdiv)
1121{
1122 unsigned long *i, *min, *max;
1123 int vleft, first = 1, err = 0;
1124 size_t left;
1125 char *p;
1126
1127 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
1128 *lenp = 0;
1129 return 0;
1130 }
1131
1132 i = (unsigned long *) data;
1133 min = (unsigned long *) table->extra1;
1134 max = (unsigned long *) table->extra2;
1135 vleft = table->maxlen / sizeof(unsigned long);
1136 left = *lenp;
1137
1138 if (write) {
1139 if (proc_first_pos_non_zero_ignore(ppos, table))
1140 goto out;
1141
1142 if (left > PAGE_SIZE - 1)
1143 left = PAGE_SIZE - 1;
1144 p = buffer;
1145 }
1146
1147 for (; left && vleft--; i++, first = 0) {
1148 unsigned long val;
1149
1150 if (write) {
1151 bool neg;
1152
1153 left -= proc_skip_spaces(&p);
1154 if (!left)
1155 break;
1156
1157 err = proc_get_long(&p, &left, &val, &neg,
1158 proc_wspace_sep,
1159 sizeof(proc_wspace_sep), NULL);
1160 if (err)
1161 break;
1162 if (neg)
1163 continue;
1164 val = convmul * val / convdiv;
1165 if ((min && val < *min) || (max && val > *max)) {
1166 err = -EINVAL;
1167 break;
1168 }
1169 *i = val;
1170 } else {
1171 val = convdiv * (*i) / convmul;
1172 if (!first)
1173 proc_put_char(&buffer, &left, '\t');
1174 proc_put_long(&buffer, &left, val, false);
1175 }
1176 }
1177
1178 if (!write && !first && left && !err)
1179 proc_put_char(&buffer, &left, '\n');
1180 if (write && !err)
1181 left -= proc_skip_spaces(&p);
1182 if (write && first)
1183 return err ? : -EINVAL;
1184 *lenp -= left;
1185out:
1186 *ppos += *lenp;
1187 return err;
1188}
1189
1190static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
1191 void *buffer, size_t *lenp, loff_t *ppos, unsigned long convmul,
1192 unsigned long convdiv)
1193{
1194 return __do_proc_doulongvec_minmax(table->data, table, write,
1195 buffer, lenp, ppos, convmul, convdiv);
1196}
1197
1198/**
1199 * proc_doulongvec_minmax - read a vector of long integers with min/max values
1200 * @table: the sysctl table
1201 * @write: %TRUE if this is a write to the sysctl file
1202 * @buffer: the user buffer
1203 * @lenp: the size of the user buffer
1204 * @ppos: file position
1205 *
1206 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1207 * values from/to the user buffer, treated as an ASCII string.
1208 *
1209 * This routine will ensure the values are within the range specified by
1210 * table->extra1 (min) and table->extra2 (max).
1211 *
1212 * Returns 0 on success.
1213 */
1214int proc_doulongvec_minmax(struct ctl_table *table, int write,
1215 void *buffer, size_t *lenp, loff_t *ppos)
1216{
1217 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
1218}
1219
1220/**
1221 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
1222 * @table: the sysctl table
1223 * @write: %TRUE if this is a write to the sysctl file
1224 * @buffer: the user buffer
1225 * @lenp: the size of the user buffer
1226 * @ppos: file position
1227 *
1228 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1229 * values from/to the user buffer, treated as an ASCII string. The values
1230 * are treated as milliseconds, and converted to jiffies when they are stored.
1231 *
1232 * This routine will ensure the values are within the range specified by
1233 * table->extra1 (min) and table->extra2 (max).
1234 *
1235 * Returns 0 on success.
1236 */
1237int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1238 void *buffer, size_t *lenp, loff_t *ppos)
1239{
1240 return do_proc_doulongvec_minmax(table, write, buffer,
1241 lenp, ppos, HZ, 1000l);
1242}
1243
1244
1245static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
1246 int *valp,
1247 int write, void *data)
1248{
1249 if (write) {
1250 if (*lvalp > INT_MAX / HZ)
1251 return 1;
1252 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
1253 } else {
1254 int val = *valp;
1255 unsigned long lval;
1256 if (val < 0) {
1257 *negp = true;
1258 lval = -(unsigned long)val;
1259 } else {
1260 *negp = false;
1261 lval = (unsigned long)val;
1262 }
1263 *lvalp = lval / HZ;
1264 }
1265 return 0;
1266}
1267
1268static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
1269 int *valp,
1270 int write, void *data)
1271{
1272 if (write) {
1273 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
1274 return 1;
1275 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
1276 } else {
1277 int val = *valp;
1278 unsigned long lval;
1279 if (val < 0) {
1280 *negp = true;
1281 lval = -(unsigned long)val;
1282 } else {
1283 *negp = false;
1284 lval = (unsigned long)val;
1285 }
1286 *lvalp = jiffies_to_clock_t(lval);
1287 }
1288 return 0;
1289}
1290
1291static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
1292 int *valp,
1293 int write, void *data)
1294{
1295 if (write) {
1296 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
1297
1298 if (jif > INT_MAX)
1299 return 1;
1300 *valp = (int)jif;
1301 } else {
1302 int val = *valp;
1303 unsigned long lval;
1304 if (val < 0) {
1305 *negp = true;
1306 lval = -(unsigned long)val;
1307 } else {
1308 *negp = false;
1309 lval = (unsigned long)val;
1310 }
1311 *lvalp = jiffies_to_msecs(lval);
1312 }
1313 return 0;
1314}
1315
1316/**
1317 * proc_dointvec_jiffies - read a vector of integers as seconds
1318 * @table: the sysctl table
1319 * @write: %TRUE if this is a write to the sysctl file
1320 * @buffer: the user buffer
1321 * @lenp: the size of the user buffer
1322 * @ppos: file position
1323 *
1324 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1325 * values from/to the user buffer, treated as an ASCII string.
1326 * The values read are assumed to be in seconds, and are converted into
1327 * jiffies.
1328 *
1329 * Returns 0 on success.
1330 */
1331int proc_dointvec_jiffies(struct ctl_table *table, int write,
1332 void *buffer, size_t *lenp, loff_t *ppos)
1333{
1334 return do_proc_dointvec(table,write,buffer,lenp,ppos,
1335 do_proc_dointvec_jiffies_conv,NULL);
1336}
1337
1338/**
1339 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
1340 * @table: the sysctl table
1341 * @write: %TRUE if this is a write to the sysctl file
1342 * @buffer: the user buffer
1343 * @lenp: the size of the user buffer
1344 * @ppos: pointer to the file position
1345 *
1346 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1347 * values from/to the user buffer, treated as an ASCII string.
1348 * The values read are assumed to be in 1/USER_HZ seconds, and
1349 * are converted into jiffies.
1350 *
1351 * Returns 0 on success.
1352 */
1353int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1354 void *buffer, size_t *lenp, loff_t *ppos)
1355{
1356 return do_proc_dointvec(table,write,buffer,lenp,ppos,
1357 do_proc_dointvec_userhz_jiffies_conv,NULL);
1358}
1359
1360/**
1361 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
1362 * @table: the sysctl table
1363 * @write: %TRUE if this is a write to the sysctl file
1364 * @buffer: the user buffer
1365 * @lenp: the size of the user buffer
1366 * @ppos: file position
1367 * @ppos: the current position in the file
1368 *
1369 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1370 * values from/to the user buffer, treated as an ASCII string.
1371 * The values read are assumed to be in 1/1000 seconds, and
1372 * are converted into jiffies.
1373 *
1374 * Returns 0 on success.
1375 */
1376int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, void *buffer,
1377 size_t *lenp, loff_t *ppos)
1378{
1379 return do_proc_dointvec(table, write, buffer, lenp, ppos,
1380 do_proc_dointvec_ms_jiffies_conv, NULL);
1381}
1382
1383static int proc_do_cad_pid(struct ctl_table *table, int write, void *buffer,
1384 size_t *lenp, loff_t *ppos)
1385{
1386 struct pid *new_pid;
1387 pid_t tmp;
1388 int r;
1389
1390 tmp = pid_vnr(cad_pid);
1391
1392 r = __do_proc_dointvec(&tmp, table, write, buffer,
1393 lenp, ppos, NULL, NULL);
1394 if (r || !write)
1395 return r;
1396
1397 new_pid = find_get_pid(tmp);
1398 if (!new_pid)
1399 return -ESRCH;
1400
1401 put_pid(xchg(&cad_pid, new_pid));
1402 return 0;
1403}
1404
1405/**
1406 * proc_do_large_bitmap - read/write from/to a large bitmap
1407 * @table: the sysctl table
1408 * @write: %TRUE if this is a write to the sysctl file
1409 * @buffer: the user buffer
1410 * @lenp: the size of the user buffer
1411 * @ppos: file position
1412 *
1413 * The bitmap is stored at table->data and the bitmap length (in bits)
1414 * in table->maxlen.
1415 *
1416 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
1417 * large bitmaps may be represented in a compact manner. Writing into
1418 * the file will clear the bitmap then update it with the given input.
1419 *
1420 * Returns 0 on success.
1421 */
1422int proc_do_large_bitmap(struct ctl_table *table, int write,
1423 void *buffer, size_t *lenp, loff_t *ppos)
1424{
1425 int err = 0;
1426 bool first = 1;
1427 size_t left = *lenp;
1428 unsigned long bitmap_len = table->maxlen;
1429 unsigned long *bitmap = *(unsigned long **) table->data;
1430 unsigned long *tmp_bitmap = NULL;
1431 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
1432
1433 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
1434 *lenp = 0;
1435 return 0;
1436 }
1437
1438 if (write) {
1439 char *p = buffer;
1440 size_t skipped = 0;
1441
1442 if (left > PAGE_SIZE - 1) {
1443 left = PAGE_SIZE - 1;
1444 /* How much of the buffer we'll skip this pass */
1445 skipped = *lenp - left;
1446 }
1447
1448 tmp_bitmap = bitmap_zalloc(bitmap_len, GFP_KERNEL);
1449 if (!tmp_bitmap)
1450 return -ENOMEM;
1451 proc_skip_char(&p, &left, '\n');
1452 while (!err && left) {
1453 unsigned long val_a, val_b;
1454 bool neg;
1455 size_t saved_left;
1456
1457 /* In case we stop parsing mid-number, we can reset */
1458 saved_left = left;
1459 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
1460 sizeof(tr_a), &c);
1461 /*
1462 * If we consumed the entirety of a truncated buffer or
1463 * only one char is left (may be a "-"), then stop here,
1464 * reset, & come back for more.
1465 */
1466 if ((left <= 1) && skipped) {
1467 left = saved_left;
1468 break;
1469 }
1470
1471 if (err)
1472 break;
1473 if (val_a >= bitmap_len || neg) {
1474 err = -EINVAL;
1475 break;
1476 }
1477
1478 val_b = val_a;
1479 if (left) {
1480 p++;
1481 left--;
1482 }
1483
1484 if (c == '-') {
1485 err = proc_get_long(&p, &left, &val_b,
1486 &neg, tr_b, sizeof(tr_b),
1487 &c);
1488 /*
1489 * If we consumed all of a truncated buffer or
1490 * then stop here, reset, & come back for more.
1491 */
1492 if (!left && skipped) {
1493 left = saved_left;
1494 break;
1495 }
1496
1497 if (err)
1498 break;
1499 if (val_b >= bitmap_len || neg ||
1500 val_a > val_b) {
1501 err = -EINVAL;
1502 break;
1503 }
1504 if (left) {
1505 p++;
1506 left--;
1507 }
1508 }
1509
1510 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
1511 first = 0;
1512 proc_skip_char(&p, &left, '\n');
1513 }
1514 left += skipped;
1515 } else {
1516 unsigned long bit_a, bit_b = 0;
1517
1518 while (left) {
1519 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
1520 if (bit_a >= bitmap_len)
1521 break;
1522 bit_b = find_next_zero_bit(bitmap, bitmap_len,
1523 bit_a + 1) - 1;
1524
1525 if (!first)
1526 proc_put_char(&buffer, &left, ',');
1527 proc_put_long(&buffer, &left, bit_a, false);
1528 if (bit_a != bit_b) {
1529 proc_put_char(&buffer, &left, '-');
1530 proc_put_long(&buffer, &left, bit_b, false);
1531 }
1532
1533 first = 0; bit_b++;
1534 }
1535 proc_put_char(&buffer, &left, '\n');
1536 }
1537
1538 if (!err) {
1539 if (write) {
1540 if (*ppos)
1541 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
1542 else
1543 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
1544 }
1545 *lenp -= left;
1546 *ppos += *lenp;
1547 }
1548
1549 bitmap_free(tmp_bitmap);
1550 return err;
1551}
1552
1553#else /* CONFIG_PROC_SYSCTL */
1554
1555int proc_dostring(struct ctl_table *table, int write,
1556 void *buffer, size_t *lenp, loff_t *ppos)
1557{
1558 return -ENOSYS;
1559}
1560
1561int proc_dointvec(struct ctl_table *table, int write,
1562 void *buffer, size_t *lenp, loff_t *ppos)
1563{
1564 return -ENOSYS;
1565}
1566
1567int proc_douintvec(struct ctl_table *table, int write,
1568 void *buffer, size_t *lenp, loff_t *ppos)
1569{
1570 return -ENOSYS;
1571}
1572
1573int proc_dointvec_minmax(struct ctl_table *table, int write,
1574 void *buffer, size_t *lenp, loff_t *ppos)
1575{
1576 return -ENOSYS;
1577}
1578
1579int proc_douintvec_minmax(struct ctl_table *table, int write,
1580 void *buffer, size_t *lenp, loff_t *ppos)
1581{
1582 return -ENOSYS;
1583}
1584
1585int proc_dointvec_jiffies(struct ctl_table *table, int write,
1586 void *buffer, size_t *lenp, loff_t *ppos)
1587{
1588 return -ENOSYS;
1589}
1590
1591int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1592 void *buffer, size_t *lenp, loff_t *ppos)
1593{
1594 return -ENOSYS;
1595}
1596
1597int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
1598 void *buffer, size_t *lenp, loff_t *ppos)
1599{
1600 return -ENOSYS;
1601}
1602
1603int proc_doulongvec_minmax(struct ctl_table *table, int write,
1604 void *buffer, size_t *lenp, loff_t *ppos)
1605{
1606 return -ENOSYS;
1607}
1608
1609int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1610 void *buffer, size_t *lenp, loff_t *ppos)
1611{
1612 return -ENOSYS;
1613}
1614
1615int proc_do_large_bitmap(struct ctl_table *table, int write,
1616 void *buffer, size_t *lenp, loff_t *ppos)
1617{
1618 return -ENOSYS;
1619}
1620
1621#endif /* CONFIG_PROC_SYSCTL */
1622
1623#if defined(CONFIG_SYSCTL)
1624int proc_do_static_key(struct ctl_table *table, int write,
1625 void *buffer, size_t *lenp, loff_t *ppos)
1626{
1627 struct static_key *key = (struct static_key *)table->data;
1628 static DEFINE_MUTEX(static_key_mutex);
1629 int val, ret;
1630 struct ctl_table tmp = {
1631 .data = &val,
1632 .maxlen = sizeof(val),
1633 .mode = table->mode,
1634 .extra1 = SYSCTL_ZERO,
1635 .extra2 = SYSCTL_ONE,
1636 };
1637
1638 if (write && !capable(CAP_SYS_ADMIN))
1639 return -EPERM;
1640
1641 mutex_lock(&static_key_mutex);
1642 val = static_key_enabled(key);
1643 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
1644 if (write && !ret) {
1645 if (val)
1646 static_key_enable(key);
1647 else
1648 static_key_disable(key);
1649 }
1650 mutex_unlock(&static_key_mutex);
1651 return ret;
1652}
1653
1654static struct ctl_table kern_table[] = {
1655 {
1656 .procname = "sched_child_runs_first",
1657 .data = &sysctl_sched_child_runs_first,
1658 .maxlen = sizeof(unsigned int),
1659 .mode = 0644,
1660 .proc_handler = proc_dointvec,
1661 },
1662#ifdef CONFIG_SCHED_DEBUG
1663 {
1664 .procname = "sched_min_granularity_ns",
1665 .data = &sysctl_sched_min_granularity,
1666 .maxlen = sizeof(unsigned int),
1667 .mode = 0644,
1668 .proc_handler = sched_proc_update_handler,
1669 .extra1 = &min_sched_granularity_ns,
1670 .extra2 = &max_sched_granularity_ns,
1671 },
1672 {
1673 .procname = "sched_latency_ns",
1674 .data = &sysctl_sched_latency,
1675 .maxlen = sizeof(unsigned int),
1676 .mode = 0644,
1677 .proc_handler = sched_proc_update_handler,
1678 .extra1 = &min_sched_granularity_ns,
1679 .extra2 = &max_sched_granularity_ns,
1680 },
1681 {
1682 .procname = "sched_wakeup_granularity_ns",
1683 .data = &sysctl_sched_wakeup_granularity,
1684 .maxlen = sizeof(unsigned int),
1685 .mode = 0644,
1686 .proc_handler = sched_proc_update_handler,
1687 .extra1 = &min_wakeup_granularity_ns,
1688 .extra2 = &max_wakeup_granularity_ns,
1689 },
1690#ifdef CONFIG_SMP
1691 {
1692 .procname = "sched_tunable_scaling",
1693 .data = &sysctl_sched_tunable_scaling,
1694 .maxlen = sizeof(enum sched_tunable_scaling),
1695 .mode = 0644,
1696 .proc_handler = sched_proc_update_handler,
1697 .extra1 = &min_sched_tunable_scaling,
1698 .extra2 = &max_sched_tunable_scaling,
1699 },
1700 {
1701 .procname = "sched_migration_cost_ns",
1702 .data = &sysctl_sched_migration_cost,
1703 .maxlen = sizeof(unsigned int),
1704 .mode = 0644,
1705 .proc_handler = proc_dointvec,
1706 },
1707 {
1708 .procname = "sched_nr_migrate",
1709 .data = &sysctl_sched_nr_migrate,
1710 .maxlen = sizeof(unsigned int),
1711 .mode = 0644,
1712 .proc_handler = proc_dointvec,
1713 },
1714#ifdef CONFIG_SCHEDSTATS
1715 {
1716 .procname = "sched_schedstats",
1717 .data = NULL,
1718 .maxlen = sizeof(unsigned int),
1719 .mode = 0644,
1720 .proc_handler = sysctl_schedstats,
1721 .extra1 = SYSCTL_ZERO,
1722 .extra2 = SYSCTL_ONE,
1723 },
1724#endif /* CONFIG_SCHEDSTATS */
1725#endif /* CONFIG_SMP */
1726#ifdef CONFIG_NUMA_BALANCING
1727 {
1728 .procname = "numa_balancing_scan_delay_ms",
1729 .data = &sysctl_numa_balancing_scan_delay,
1730 .maxlen = sizeof(unsigned int),
1731 .mode = 0644,
1732 .proc_handler = proc_dointvec,
1733 },
1734 {
1735 .procname = "numa_balancing_scan_period_min_ms",
1736 .data = &sysctl_numa_balancing_scan_period_min,
1737 .maxlen = sizeof(unsigned int),
1738 .mode = 0644,
1739 .proc_handler = proc_dointvec,
1740 },
1741 {
1742 .procname = "numa_balancing_scan_period_max_ms",
1743 .data = &sysctl_numa_balancing_scan_period_max,
1744 .maxlen = sizeof(unsigned int),
1745 .mode = 0644,
1746 .proc_handler = proc_dointvec,
1747 },
1748 {
1749 .procname = "numa_balancing_scan_size_mb",
1750 .data = &sysctl_numa_balancing_scan_size,
1751 .maxlen = sizeof(unsigned int),
1752 .mode = 0644,
1753 .proc_handler = proc_dointvec_minmax,
1754 .extra1 = SYSCTL_ONE,
1755 },
1756 {
1757 .procname = "numa_balancing",
1758 .data = NULL, /* filled in by handler */
1759 .maxlen = sizeof(unsigned int),
1760 .mode = 0644,
1761 .proc_handler = sysctl_numa_balancing,
1762 .extra1 = SYSCTL_ZERO,
1763 .extra2 = SYSCTL_ONE,
1764 },
1765#endif /* CONFIG_NUMA_BALANCING */
1766#endif /* CONFIG_SCHED_DEBUG */
1767 {
1768 .procname = "sched_rt_period_us",
1769 .data = &sysctl_sched_rt_period,
1770 .maxlen = sizeof(unsigned int),
1771 .mode = 0644,
1772 .proc_handler = sched_rt_handler,
1773 },
1774 {
1775 .procname = "sched_rt_runtime_us",
1776 .data = &sysctl_sched_rt_runtime,
1777 .maxlen = sizeof(int),
1778 .mode = 0644,
1779 .proc_handler = sched_rt_handler,
1780 },
1781 {
1782 .procname = "sched_deadline_period_max_us",
1783 .data = &sysctl_sched_dl_period_max,
1784 .maxlen = sizeof(unsigned int),
1785 .mode = 0644,
1786 .proc_handler = proc_dointvec,
1787 },
1788 {
1789 .procname = "sched_deadline_period_min_us",
1790 .data = &sysctl_sched_dl_period_min,
1791 .maxlen = sizeof(unsigned int),
1792 .mode = 0644,
1793 .proc_handler = proc_dointvec,
1794 },
1795 {
1796 .procname = "sched_rr_timeslice_ms",
1797 .data = &sysctl_sched_rr_timeslice,
1798 .maxlen = sizeof(int),
1799 .mode = 0644,
1800 .proc_handler = sched_rr_handler,
1801 },
1802#ifdef CONFIG_UCLAMP_TASK
1803 {
1804 .procname = "sched_util_clamp_min",
1805 .data = &sysctl_sched_uclamp_util_min,
1806 .maxlen = sizeof(unsigned int),
1807 .mode = 0644,
1808 .proc_handler = sysctl_sched_uclamp_handler,
1809 },
1810 {
1811 .procname = "sched_util_clamp_max",
1812 .data = &sysctl_sched_uclamp_util_max,
1813 .maxlen = sizeof(unsigned int),
1814 .mode = 0644,
1815 .proc_handler = sysctl_sched_uclamp_handler,
1816 },
1817 {
1818 .procname = "sched_util_clamp_min_rt_default",
1819 .data = &sysctl_sched_uclamp_util_min_rt_default,
1820 .maxlen = sizeof(unsigned int),
1821 .mode = 0644,
1822 .proc_handler = sysctl_sched_uclamp_handler,
1823 },
1824#endif
1825#ifdef CONFIG_SCHED_AUTOGROUP
1826 {
1827 .procname = "sched_autogroup_enabled",
1828 .data = &sysctl_sched_autogroup_enabled,
1829 .maxlen = sizeof(unsigned int),
1830 .mode = 0644,
1831 .proc_handler = proc_dointvec_minmax,
1832 .extra1 = SYSCTL_ZERO,
1833 .extra2 = SYSCTL_ONE,
1834 },
1835#endif
1836#ifdef CONFIG_CFS_BANDWIDTH
1837 {
1838 .procname = "sched_cfs_bandwidth_slice_us",
1839 .data = &sysctl_sched_cfs_bandwidth_slice,
1840 .maxlen = sizeof(unsigned int),
1841 .mode = 0644,
1842 .proc_handler = proc_dointvec_minmax,
1843 .extra1 = SYSCTL_ONE,
1844 },
1845#endif
1846#if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_CPU_FREQ_GOV_SCHEDUTIL)
1847 {
1848 .procname = "sched_energy_aware",
1849 .data = &sysctl_sched_energy_aware,
1850 .maxlen = sizeof(unsigned int),
1851 .mode = 0644,
1852 .proc_handler = sched_energy_aware_handler,
1853 .extra1 = SYSCTL_ZERO,
1854 .extra2 = SYSCTL_ONE,
1855 },
1856#endif
1857#ifdef CONFIG_PROVE_LOCKING
1858 {
1859 .procname = "prove_locking",
1860 .data = &prove_locking,
1861 .maxlen = sizeof(int),
1862 .mode = 0644,
1863 .proc_handler = proc_dointvec,
1864 },
1865#endif
1866#ifdef CONFIG_LOCK_STAT
1867 {
1868 .procname = "lock_stat",
1869 .data = &lock_stat,
1870 .maxlen = sizeof(int),
1871 .mode = 0644,
1872 .proc_handler = proc_dointvec,
1873 },
1874#endif
1875 {
1876 .procname = "panic",
1877 .data = &panic_timeout,
1878 .maxlen = sizeof(int),
1879 .mode = 0644,
1880 .proc_handler = proc_dointvec,
1881 },
1882#ifdef CONFIG_COREDUMP
1883 {
1884 .procname = "core_uses_pid",
1885 .data = &core_uses_pid,
1886 .maxlen = sizeof(int),
1887 .mode = 0644,
1888 .proc_handler = proc_dointvec,
1889 },
1890 {
1891 .procname = "core_pattern",
1892 .data = core_pattern,
1893 .maxlen = CORENAME_MAX_SIZE,
1894 .mode = 0644,
1895 .proc_handler = proc_dostring_coredump,
1896 },
1897 {
1898 .procname = "core_pipe_limit",
1899 .data = &core_pipe_limit,
1900 .maxlen = sizeof(unsigned int),
1901 .mode = 0644,
1902 .proc_handler = proc_dointvec,
1903 },
1904#endif
1905#ifdef CONFIG_PROC_SYSCTL
1906 {
1907 .procname = "tainted",
1908 .maxlen = sizeof(long),
1909 .mode = 0644,
1910 .proc_handler = proc_taint,
1911 },
1912 {
1913 .procname = "sysctl_writes_strict",
1914 .data = &sysctl_writes_strict,
1915 .maxlen = sizeof(int),
1916 .mode = 0644,
1917 .proc_handler = proc_dointvec_minmax,
1918 .extra1 = &neg_one,
1919 .extra2 = SYSCTL_ONE,
1920 },
1921#endif
1922#ifdef CONFIG_LATENCYTOP
1923 {
1924 .procname = "latencytop",
1925 .data = &latencytop_enabled,
1926 .maxlen = sizeof(int),
1927 .mode = 0644,
1928 .proc_handler = sysctl_latencytop,
1929 },
1930#endif
1931#ifdef CONFIG_BLK_DEV_INITRD
1932 {
1933 .procname = "real-root-dev",
1934 .data = &real_root_dev,
1935 .maxlen = sizeof(int),
1936 .mode = 0644,
1937 .proc_handler = proc_dointvec,
1938 },
1939#endif
1940 {
1941 .procname = "print-fatal-signals",
1942 .data = &print_fatal_signals,
1943 .maxlen = sizeof(int),
1944 .mode = 0644,
1945 .proc_handler = proc_dointvec,
1946 },
1947#ifdef CONFIG_SPARC
1948 {
1949 .procname = "reboot-cmd",
1950 .data = reboot_command,
1951 .maxlen = 256,
1952 .mode = 0644,
1953 .proc_handler = proc_dostring,
1954 },
1955 {
1956 .procname = "stop-a",
1957 .data = &stop_a_enabled,
1958 .maxlen = sizeof (int),
1959 .mode = 0644,
1960 .proc_handler = proc_dointvec,
1961 },
1962 {
1963 .procname = "scons-poweroff",
1964 .data = &scons_pwroff,
1965 .maxlen = sizeof (int),
1966 .mode = 0644,
1967 .proc_handler = proc_dointvec,
1968 },
1969#endif
1970#ifdef CONFIG_SPARC64
1971 {
1972 .procname = "tsb-ratio",
1973 .data = &sysctl_tsb_ratio,
1974 .maxlen = sizeof (int),
1975 .mode = 0644,
1976 .proc_handler = proc_dointvec,
1977 },
1978#endif
1979#ifdef CONFIG_PARISC
1980 {
1981 .procname = "soft-power",
1982 .data = &pwrsw_enabled,
1983 .maxlen = sizeof (int),
1984 .mode = 0644,
1985 .proc_handler = proc_dointvec,
1986 },
1987#endif
1988#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
1989 {
1990 .procname = "unaligned-trap",
1991 .data = &unaligned_enabled,
1992 .maxlen = sizeof (int),
1993 .mode = 0644,
1994 .proc_handler = proc_dointvec,
1995 },
1996#endif
1997 {
1998 .procname = "ctrl-alt-del",
1999 .data = &C_A_D,
2000 .maxlen = sizeof(int),
2001 .mode = 0644,
2002 .proc_handler = proc_dointvec,
2003 },
2004#ifdef CONFIG_FUNCTION_TRACER
2005 {
2006 .procname = "ftrace_enabled",
2007 .data = &ftrace_enabled,
2008 .maxlen = sizeof(int),
2009 .mode = 0644,
2010 .proc_handler = ftrace_enable_sysctl,
2011 },
2012#endif
2013#ifdef CONFIG_STACK_TRACER
2014 {
2015 .procname = "stack_tracer_enabled",
2016 .data = &stack_tracer_enabled,
2017 .maxlen = sizeof(int),
2018 .mode = 0644,
2019 .proc_handler = stack_trace_sysctl,
2020 },
2021#endif
2022#ifdef CONFIG_TRACING
2023 {
2024 .procname = "ftrace_dump_on_oops",
2025 .data = &ftrace_dump_on_oops,
2026 .maxlen = sizeof(int),
2027 .mode = 0644,
2028 .proc_handler = proc_dointvec,
2029 },
2030 {
2031 .procname = "traceoff_on_warning",
2032 .data = &__disable_trace_on_warning,
2033 .maxlen = sizeof(__disable_trace_on_warning),
2034 .mode = 0644,
2035 .proc_handler = proc_dointvec,
2036 },
2037 {
2038 .procname = "tracepoint_printk",
2039 .data = &tracepoint_printk,
2040 .maxlen = sizeof(tracepoint_printk),
2041 .mode = 0644,
2042 .proc_handler = tracepoint_printk_sysctl,
2043 },
2044#endif
2045#ifdef CONFIG_KEXEC_CORE
2046 {
2047 .procname = "kexec_load_disabled",
2048 .data = &kexec_load_disabled,
2049 .maxlen = sizeof(int),
2050 .mode = 0644,
2051 /* only handle a transition from default "0" to "1" */
2052 .proc_handler = proc_dointvec_minmax,
2053 .extra1 = SYSCTL_ONE,
2054 .extra2 = SYSCTL_ONE,
2055 },
2056#endif
2057#ifdef CONFIG_MODULES
2058 {
2059 .procname = "modprobe",
2060 .data = &modprobe_path,
2061 .maxlen = KMOD_PATH_LEN,
2062 .mode = 0644,
2063 .proc_handler = proc_dostring,
2064 },
2065 {
2066 .procname = "modules_disabled",
2067 .data = &modules_disabled,
2068 .maxlen = sizeof(int),
2069 .mode = 0644,
2070 /* only handle a transition from default "0" to "1" */
2071 .proc_handler = proc_dointvec_minmax,
2072 .extra1 = SYSCTL_ONE,
2073 .extra2 = SYSCTL_ONE,
2074 },
2075#endif
2076#ifdef CONFIG_UEVENT_HELPER
2077 {
2078 .procname = "hotplug",
2079 .data = &uevent_helper,
2080 .maxlen = UEVENT_HELPER_PATH_LEN,
2081 .mode = 0644,
2082 .proc_handler = proc_dostring,
2083 },
2084#endif
2085#ifdef CONFIG_CHR_DEV_SG
2086 {
2087 .procname = "sg-big-buff",
2088 .data = &sg_big_buff,
2089 .maxlen = sizeof (int),
2090 .mode = 0444,
2091 .proc_handler = proc_dointvec,
2092 },
2093#endif
2094#ifdef CONFIG_BSD_PROCESS_ACCT
2095 {
2096 .procname = "acct",
2097 .data = &acct_parm,
2098 .maxlen = 3*sizeof(int),
2099 .mode = 0644,
2100 .proc_handler = proc_dointvec,
2101 },
2102#endif
2103#ifdef CONFIG_MAGIC_SYSRQ
2104 {
2105 .procname = "sysrq",
2106 .data = NULL,
2107 .maxlen = sizeof (int),
2108 .mode = 0644,
2109 .proc_handler = sysrq_sysctl_handler,
2110 },
2111#endif
2112#ifdef CONFIG_PROC_SYSCTL
2113 {
2114 .procname = "cad_pid",
2115 .data = NULL,
2116 .maxlen = sizeof (int),
2117 .mode = 0600,
2118 .proc_handler = proc_do_cad_pid,
2119 },
2120#endif
2121 {
2122 .procname = "threads-max",
2123 .data = NULL,
2124 .maxlen = sizeof(int),
2125 .mode = 0644,
2126 .proc_handler = sysctl_max_threads,
2127 },
2128 {
2129 .procname = "random",
2130 .mode = 0555,
2131 .child = random_table,
2132 },
2133 {
2134 .procname = "usermodehelper",
2135 .mode = 0555,
2136 .child = usermodehelper_table,
2137 },
2138#ifdef CONFIG_FW_LOADER_USER_HELPER
2139 {
2140 .procname = "firmware_config",
2141 .mode = 0555,
2142 .child = firmware_config_table,
2143 },
2144#endif
2145 {
2146 .procname = "overflowuid",
2147 .data = &overflowuid,
2148 .maxlen = sizeof(int),
2149 .mode = 0644,
2150 .proc_handler = proc_dointvec_minmax,
2151 .extra1 = &minolduid,
2152 .extra2 = &maxolduid,
2153 },
2154 {
2155 .procname = "overflowgid",
2156 .data = &overflowgid,
2157 .maxlen = sizeof(int),
2158 .mode = 0644,
2159 .proc_handler = proc_dointvec_minmax,
2160 .extra1 = &minolduid,
2161 .extra2 = &maxolduid,
2162 },
2163#ifdef CONFIG_S390
2164 {
2165 .procname = "userprocess_debug",
2166 .data = &show_unhandled_signals,
2167 .maxlen = sizeof(int),
2168 .mode = 0644,
2169 .proc_handler = proc_dointvec,
2170 },
2171#endif
2172#ifdef CONFIG_SMP
2173 {
2174 .procname = "oops_all_cpu_backtrace",
2175 .data = &sysctl_oops_all_cpu_backtrace,
2176 .maxlen = sizeof(int),
2177 .mode = 0644,
2178 .proc_handler = proc_dointvec_minmax,
2179 .extra1 = SYSCTL_ZERO,
2180 .extra2 = SYSCTL_ONE,
2181 },
2182#endif /* CONFIG_SMP */
2183 {
2184 .procname = "pid_max",
2185 .data = &pid_max,
2186 .maxlen = sizeof (int),
2187 .mode = 0644,
2188 .proc_handler = proc_dointvec_minmax,
2189 .extra1 = &pid_max_min,
2190 .extra2 = &pid_max_max,
2191 },
2192 {
2193 .procname = "panic_on_oops",
2194 .data = &panic_on_oops,
2195 .maxlen = sizeof(int),
2196 .mode = 0644,
2197 .proc_handler = proc_dointvec,
2198 },
2199 {
2200 .procname = "panic_print",
2201 .data = &panic_print,
2202 .maxlen = sizeof(unsigned long),
2203 .mode = 0644,
2204 .proc_handler = proc_doulongvec_minmax,
2205 },
2206#if defined CONFIG_PRINTK
2207 {
2208 .procname = "printk",
2209 .data = &console_loglevel,
2210 .maxlen = 4*sizeof(int),
2211 .mode = 0644,
2212 .proc_handler = proc_dointvec,
2213 },
2214 {
2215 .procname = "printk_ratelimit",
2216 .data = &printk_ratelimit_state.interval,
2217 .maxlen = sizeof(int),
2218 .mode = 0644,
2219 .proc_handler = proc_dointvec_jiffies,
2220 },
2221 {
2222 .procname = "printk_ratelimit_burst",
2223 .data = &printk_ratelimit_state.burst,
2224 .maxlen = sizeof(int),
2225 .mode = 0644,
2226 .proc_handler = proc_dointvec,
2227 },
2228 {
2229 .procname = "printk_delay",
2230 .data = &printk_delay_msec,
2231 .maxlen = sizeof(int),
2232 .mode = 0644,
2233 .proc_handler = proc_dointvec_minmax,
2234 .extra1 = SYSCTL_ZERO,
2235 .extra2 = &ten_thousand,
2236 },
2237 {
2238 .procname = "printk_devkmsg",
2239 .data = devkmsg_log_str,
2240 .maxlen = DEVKMSG_STR_MAX_SIZE,
2241 .mode = 0644,
2242 .proc_handler = devkmsg_sysctl_set_loglvl,
2243 },
2244 {
2245 .procname = "dmesg_restrict",
2246 .data = &dmesg_restrict,
2247 .maxlen = sizeof(int),
2248 .mode = 0644,
2249 .proc_handler = proc_dointvec_minmax_sysadmin,
2250 .extra1 = SYSCTL_ZERO,
2251 .extra2 = SYSCTL_ONE,
2252 },
2253 {
2254 .procname = "kptr_restrict",
2255 .data = &kptr_restrict,
2256 .maxlen = sizeof(int),
2257 .mode = 0644,
2258 .proc_handler = proc_dointvec_minmax_sysadmin,
2259 .extra1 = SYSCTL_ZERO,
2260 .extra2 = &two,
2261 },
2262#endif
2263 {
2264 .procname = "ngroups_max",
2265 .data = &ngroups_max,
2266 .maxlen = sizeof (int),
2267 .mode = 0444,
2268 .proc_handler = proc_dointvec,
2269 },
2270 {
2271 .procname = "cap_last_cap",
2272 .data = (void *)&cap_last_cap,
2273 .maxlen = sizeof(int),
2274 .mode = 0444,
2275 .proc_handler = proc_dointvec,
2276 },
2277#if defined(CONFIG_LOCKUP_DETECTOR)
2278 {
2279 .procname = "watchdog",
2280 .data = &watchdog_user_enabled,
2281 .maxlen = sizeof(int),
2282 .mode = 0644,
2283 .proc_handler = proc_watchdog,
2284 .extra1 = SYSCTL_ZERO,
2285 .extra2 = SYSCTL_ONE,
2286 },
2287 {
2288 .procname = "watchdog_thresh",
2289 .data = &watchdog_thresh,
2290 .maxlen = sizeof(int),
2291 .mode = 0644,
2292 .proc_handler = proc_watchdog_thresh,
2293 .extra1 = SYSCTL_ZERO,
2294 .extra2 = &sixty,
2295 },
2296 {
2297 .procname = "nmi_watchdog",
2298 .data = &nmi_watchdog_user_enabled,
2299 .maxlen = sizeof(int),
2300 .mode = NMI_WATCHDOG_SYSCTL_PERM,
2301 .proc_handler = proc_nmi_watchdog,
2302 .extra1 = SYSCTL_ZERO,
2303 .extra2 = SYSCTL_ONE,
2304 },
2305 {
2306 .procname = "watchdog_cpumask",
2307 .data = &watchdog_cpumask_bits,
2308 .maxlen = NR_CPUS,
2309 .mode = 0644,
2310 .proc_handler = proc_watchdog_cpumask,
2311 },
2312#ifdef CONFIG_SOFTLOCKUP_DETECTOR
2313 {
2314 .procname = "soft_watchdog",
2315 .data = &soft_watchdog_user_enabled,
2316 .maxlen = sizeof(int),
2317 .mode = 0644,
2318 .proc_handler = proc_soft_watchdog,
2319 .extra1 = SYSCTL_ZERO,
2320 .extra2 = SYSCTL_ONE,
2321 },
2322 {
2323 .procname = "softlockup_panic",
2324 .data = &softlockup_panic,
2325 .maxlen = sizeof(int),
2326 .mode = 0644,
2327 .proc_handler = proc_dointvec_minmax,
2328 .extra1 = SYSCTL_ZERO,
2329 .extra2 = SYSCTL_ONE,
2330 },
2331#ifdef CONFIG_SMP
2332 {
2333 .procname = "softlockup_all_cpu_backtrace",
2334 .data = &sysctl_softlockup_all_cpu_backtrace,
2335 .maxlen = sizeof(int),
2336 .mode = 0644,
2337 .proc_handler = proc_dointvec_minmax,
2338 .extra1 = SYSCTL_ZERO,
2339 .extra2 = SYSCTL_ONE,
2340 },
2341#endif /* CONFIG_SMP */
2342#endif
2343#ifdef CONFIG_HARDLOCKUP_DETECTOR
2344 {
2345 .procname = "hardlockup_panic",
2346 .data = &hardlockup_panic,
2347 .maxlen = sizeof(int),
2348 .mode = 0644,
2349 .proc_handler = proc_dointvec_minmax,
2350 .extra1 = SYSCTL_ZERO,
2351 .extra2 = SYSCTL_ONE,
2352 },
2353#ifdef CONFIG_SMP
2354 {
2355 .procname = "hardlockup_all_cpu_backtrace",
2356 .data = &sysctl_hardlockup_all_cpu_backtrace,
2357 .maxlen = sizeof(int),
2358 .mode = 0644,
2359 .proc_handler = proc_dointvec_minmax,
2360 .extra1 = SYSCTL_ZERO,
2361 .extra2 = SYSCTL_ONE,
2362 },
2363#endif /* CONFIG_SMP */
2364#endif
2365#endif
2366
2367#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
2368 {
2369 .procname = "unknown_nmi_panic",
2370 .data = &unknown_nmi_panic,
2371 .maxlen = sizeof (int),
2372 .mode = 0644,
2373 .proc_handler = proc_dointvec,
2374 },
2375#endif
2376
2377#if (defined(CONFIG_X86_32) || defined(CONFIG_PARISC)) && \
2378 defined(CONFIG_DEBUG_STACKOVERFLOW)
2379 {
2380 .procname = "panic_on_stackoverflow",
2381 .data = &sysctl_panic_on_stackoverflow,
2382 .maxlen = sizeof(int),
2383 .mode = 0644,
2384 .proc_handler = proc_dointvec,
2385 },
2386#endif
2387#if defined(CONFIG_X86)
2388 {
2389 .procname = "panic_on_unrecovered_nmi",
2390 .data = &panic_on_unrecovered_nmi,
2391 .maxlen = sizeof(int),
2392 .mode = 0644,
2393 .proc_handler = proc_dointvec,
2394 },
2395 {
2396 .procname = "panic_on_io_nmi",
2397 .data = &panic_on_io_nmi,
2398 .maxlen = sizeof(int),
2399 .mode = 0644,
2400 .proc_handler = proc_dointvec,
2401 },
2402 {
2403 .procname = "bootloader_type",
2404 .data = &bootloader_type,
2405 .maxlen = sizeof (int),
2406 .mode = 0444,
2407 .proc_handler = proc_dointvec,
2408 },
2409 {
2410 .procname = "bootloader_version",
2411 .data = &bootloader_version,
2412 .maxlen = sizeof (int),
2413 .mode = 0444,
2414 .proc_handler = proc_dointvec,
2415 },
2416 {
2417 .procname = "io_delay_type",
2418 .data = &io_delay_type,
2419 .maxlen = sizeof(int),
2420 .mode = 0644,
2421 .proc_handler = proc_dointvec,
2422 },
2423#endif
2424#if defined(CONFIG_MMU)
2425 {
2426 .procname = "randomize_va_space",
2427 .data = &randomize_va_space,
2428 .maxlen = sizeof(int),
2429 .mode = 0644,
2430 .proc_handler = proc_dointvec,
2431 },
2432#endif
2433#if defined(CONFIG_S390) && defined(CONFIG_SMP)
2434 {
2435 .procname = "spin_retry",
2436 .data = &spin_retry,
2437 .maxlen = sizeof (int),
2438 .mode = 0644,
2439 .proc_handler = proc_dointvec,
2440 },
2441#endif
2442#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
2443 {
2444 .procname = "acpi_video_flags",
2445 .data = &acpi_realmode_flags,
2446 .maxlen = sizeof (unsigned long),
2447 .mode = 0644,
2448 .proc_handler = proc_doulongvec_minmax,
2449 },
2450#endif
2451#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
2452 {
2453 .procname = "ignore-unaligned-usertrap",
2454 .data = &no_unaligned_warning,
2455 .maxlen = sizeof (int),
2456 .mode = 0644,
2457 .proc_handler = proc_dointvec,
2458 },
2459#endif
2460#ifdef CONFIG_IA64
2461 {
2462 .procname = "unaligned-dump-stack",
2463 .data = &unaligned_dump_stack,
2464 .maxlen = sizeof (int),
2465 .mode = 0644,
2466 .proc_handler = proc_dointvec,
2467 },
2468#endif
2469#ifdef CONFIG_DETECT_HUNG_TASK
2470#ifdef CONFIG_SMP
2471 {
2472 .procname = "hung_task_all_cpu_backtrace",
2473 .data = &sysctl_hung_task_all_cpu_backtrace,
2474 .maxlen = sizeof(int),
2475 .mode = 0644,
2476 .proc_handler = proc_dointvec_minmax,
2477 .extra1 = SYSCTL_ZERO,
2478 .extra2 = SYSCTL_ONE,
2479 },
2480#endif /* CONFIG_SMP */
2481 {
2482 .procname = "hung_task_panic",
2483 .data = &sysctl_hung_task_panic,
2484 .maxlen = sizeof(int),
2485 .mode = 0644,
2486 .proc_handler = proc_dointvec_minmax,
2487 .extra1 = SYSCTL_ZERO,
2488 .extra2 = SYSCTL_ONE,
2489 },
2490 {
2491 .procname = "hung_task_check_count",
2492 .data = &sysctl_hung_task_check_count,
2493 .maxlen = sizeof(int),
2494 .mode = 0644,
2495 .proc_handler = proc_dointvec_minmax,
2496 .extra1 = SYSCTL_ZERO,
2497 },
2498 {
2499 .procname = "hung_task_timeout_secs",
2500 .data = &sysctl_hung_task_timeout_secs,
2501 .maxlen = sizeof(unsigned long),
2502 .mode = 0644,
2503 .proc_handler = proc_dohung_task_timeout_secs,
2504 .extra2 = &hung_task_timeout_max,
2505 },
2506 {
2507 .procname = "hung_task_check_interval_secs",
2508 .data = &sysctl_hung_task_check_interval_secs,
2509 .maxlen = sizeof(unsigned long),
2510 .mode = 0644,
2511 .proc_handler = proc_dohung_task_timeout_secs,
2512 .extra2 = &hung_task_timeout_max,
2513 },
2514 {
2515 .procname = "hung_task_warnings",
2516 .data = &sysctl_hung_task_warnings,
2517 .maxlen = sizeof(int),
2518 .mode = 0644,
2519 .proc_handler = proc_dointvec_minmax,
2520 .extra1 = &neg_one,
2521 },
2522#endif
2523#ifdef CONFIG_RT_MUTEXES
2524 {
2525 .procname = "max_lock_depth",
2526 .data = &max_lock_depth,
2527 .maxlen = sizeof(int),
2528 .mode = 0644,
2529 .proc_handler = proc_dointvec,
2530 },
2531#endif
2532 {
2533 .procname = "poweroff_cmd",
2534 .data = &poweroff_cmd,
2535 .maxlen = POWEROFF_CMD_PATH_LEN,
2536 .mode = 0644,
2537 .proc_handler = proc_dostring,
2538 },
2539#ifdef CONFIG_KEYS
2540 {
2541 .procname = "keys",
2542 .mode = 0555,
2543 .child = key_sysctls,
2544 },
2545#endif
2546#ifdef CONFIG_PERF_EVENTS
2547 /*
2548 * User-space scripts rely on the existence of this file
2549 * as a feature check for perf_events being enabled.
2550 *
2551 * So it's an ABI, do not remove!
2552 */
2553 {
2554 .procname = "perf_event_paranoid",
2555 .data = &sysctl_perf_event_paranoid,
2556 .maxlen = sizeof(sysctl_perf_event_paranoid),
2557 .mode = 0644,
2558 .proc_handler = proc_dointvec,
2559 },
2560 {
2561 .procname = "perf_event_mlock_kb",
2562 .data = &sysctl_perf_event_mlock,
2563 .maxlen = sizeof(sysctl_perf_event_mlock),
2564 .mode = 0644,
2565 .proc_handler = proc_dointvec,
2566 },
2567 {
2568 .procname = "perf_event_max_sample_rate",
2569 .data = &sysctl_perf_event_sample_rate,
2570 .maxlen = sizeof(sysctl_perf_event_sample_rate),
2571 .mode = 0644,
2572 .proc_handler = perf_proc_update_handler,
2573 .extra1 = SYSCTL_ONE,
2574 },
2575 {
2576 .procname = "perf_cpu_time_max_percent",
2577 .data = &sysctl_perf_cpu_time_max_percent,
2578 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
2579 .mode = 0644,
2580 .proc_handler = perf_cpu_time_max_percent_handler,
2581 .extra1 = SYSCTL_ZERO,
2582 .extra2 = &one_hundred,
2583 },
2584 {
2585 .procname = "perf_event_max_stack",
2586 .data = &sysctl_perf_event_max_stack,
2587 .maxlen = sizeof(sysctl_perf_event_max_stack),
2588 .mode = 0644,
2589 .proc_handler = perf_event_max_stack_handler,
2590 .extra1 = SYSCTL_ZERO,
2591 .extra2 = &six_hundred_forty_kb,
2592 },
2593 {
2594 .procname = "perf_event_max_contexts_per_stack",
2595 .data = &sysctl_perf_event_max_contexts_per_stack,
2596 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
2597 .mode = 0644,
2598 .proc_handler = perf_event_max_stack_handler,
2599 .extra1 = SYSCTL_ZERO,
2600 .extra2 = &one_thousand,
2601 },
2602#endif
2603 {
2604 .procname = "panic_on_warn",
2605 .data = &panic_on_warn,
2606 .maxlen = sizeof(int),
2607 .mode = 0644,
2608 .proc_handler = proc_dointvec_minmax,
2609 .extra1 = SYSCTL_ZERO,
2610 .extra2 = SYSCTL_ONE,
2611 },
2612#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
2613 {
2614 .procname = "timer_migration",
2615 .data = &sysctl_timer_migration,
2616 .maxlen = sizeof(unsigned int),
2617 .mode = 0644,
2618 .proc_handler = timer_migration_handler,
2619 .extra1 = SYSCTL_ZERO,
2620 .extra2 = SYSCTL_ONE,
2621 },
2622#endif
2623#ifdef CONFIG_BPF_SYSCALL
2624 {
2625 .procname = "unprivileged_bpf_disabled",
2626 .data = &sysctl_unprivileged_bpf_disabled,
2627 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
2628 .mode = 0644,
2629 /* only handle a transition from default "0" to "1" */
2630 .proc_handler = proc_dointvec_minmax,
2631 .extra1 = SYSCTL_ONE,
2632 .extra2 = SYSCTL_ONE,
2633 },
2634 {
2635 .procname = "bpf_stats_enabled",
2636 .data = &bpf_stats_enabled_key.key,
2637 .maxlen = sizeof(bpf_stats_enabled_key),
2638 .mode = 0644,
2639 .proc_handler = bpf_stats_handler,
2640 },
2641#endif
2642#if defined(CONFIG_TREE_RCU)
2643 {
2644 .procname = "panic_on_rcu_stall",
2645 .data = &sysctl_panic_on_rcu_stall,
2646 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
2647 .mode = 0644,
2648 .proc_handler = proc_dointvec_minmax,
2649 .extra1 = SYSCTL_ZERO,
2650 .extra2 = SYSCTL_ONE,
2651 },
2652#endif
2653#ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
2654 {
2655 .procname = "stack_erasing",
2656 .data = NULL,
2657 .maxlen = sizeof(int),
2658 .mode = 0600,
2659 .proc_handler = stack_erasing_sysctl,
2660 .extra1 = SYSCTL_ZERO,
2661 .extra2 = SYSCTL_ONE,
2662 },
2663#endif
2664 { }
2665};
2666
2667static struct ctl_table vm_table[] = {
2668 {
2669 .procname = "overcommit_memory",
2670 .data = &sysctl_overcommit_memory,
2671 .maxlen = sizeof(sysctl_overcommit_memory),
2672 .mode = 0644,
2673 .proc_handler = overcommit_policy_handler,
2674 .extra1 = SYSCTL_ZERO,
2675 .extra2 = &two,
2676 },
2677 {
2678 .procname = "panic_on_oom",
2679 .data = &sysctl_panic_on_oom,
2680 .maxlen = sizeof(sysctl_panic_on_oom),
2681 .mode = 0644,
2682 .proc_handler = proc_dointvec_minmax,
2683 .extra1 = SYSCTL_ZERO,
2684 .extra2 = &two,
2685 },
2686 {
2687 .procname = "oom_kill_allocating_task",
2688 .data = &sysctl_oom_kill_allocating_task,
2689 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
2690 .mode = 0644,
2691 .proc_handler = proc_dointvec,
2692 },
2693 {
2694 .procname = "oom_dump_tasks",
2695 .data = &sysctl_oom_dump_tasks,
2696 .maxlen = sizeof(sysctl_oom_dump_tasks),
2697 .mode = 0644,
2698 .proc_handler = proc_dointvec,
2699 },
2700 {
2701 .procname = "overcommit_ratio",
2702 .data = &sysctl_overcommit_ratio,
2703 .maxlen = sizeof(sysctl_overcommit_ratio),
2704 .mode = 0644,
2705 .proc_handler = overcommit_ratio_handler,
2706 },
2707 {
2708 .procname = "overcommit_kbytes",
2709 .data = &sysctl_overcommit_kbytes,
2710 .maxlen = sizeof(sysctl_overcommit_kbytes),
2711 .mode = 0644,
2712 .proc_handler = overcommit_kbytes_handler,
2713 },
2714 {
2715 .procname = "page-cluster",
2716 .data = &page_cluster,
2717 .maxlen = sizeof(int),
2718 .mode = 0644,
2719 .proc_handler = proc_dointvec_minmax,
2720 .extra1 = SYSCTL_ZERO,
2721 },
2722 {
2723 .procname = "dirty_background_ratio",
2724 .data = &dirty_background_ratio,
2725 .maxlen = sizeof(dirty_background_ratio),
2726 .mode = 0644,
2727 .proc_handler = dirty_background_ratio_handler,
2728 .extra1 = SYSCTL_ZERO,
2729 .extra2 = &one_hundred,
2730 },
2731 {
2732 .procname = "dirty_background_bytes",
2733 .data = &dirty_background_bytes,
2734 .maxlen = sizeof(dirty_background_bytes),
2735 .mode = 0644,
2736 .proc_handler = dirty_background_bytes_handler,
2737 .extra1 = &one_ul,
2738 },
2739 {
2740 .procname = "dirty_ratio",
2741 .data = &vm_dirty_ratio,
2742 .maxlen = sizeof(vm_dirty_ratio),
2743 .mode = 0644,
2744 .proc_handler = dirty_ratio_handler,
2745 .extra1 = SYSCTL_ZERO,
2746 .extra2 = &one_hundred,
2747 },
2748 {
2749 .procname = "dirty_bytes",
2750 .data = &vm_dirty_bytes,
2751 .maxlen = sizeof(vm_dirty_bytes),
2752 .mode = 0644,
2753 .proc_handler = dirty_bytes_handler,
2754 .extra1 = &dirty_bytes_min,
2755 },
2756 {
2757 .procname = "dirty_writeback_centisecs",
2758 .data = &dirty_writeback_interval,
2759 .maxlen = sizeof(dirty_writeback_interval),
2760 .mode = 0644,
2761 .proc_handler = dirty_writeback_centisecs_handler,
2762 },
2763 {
2764 .procname = "dirty_expire_centisecs",
2765 .data = &dirty_expire_interval,
2766 .maxlen = sizeof(dirty_expire_interval),
2767 .mode = 0644,
2768 .proc_handler = proc_dointvec_minmax,
2769 .extra1 = SYSCTL_ZERO,
2770 },
2771 {
2772 .procname = "dirtytime_expire_seconds",
2773 .data = &dirtytime_expire_interval,
2774 .maxlen = sizeof(dirtytime_expire_interval),
2775 .mode = 0644,
2776 .proc_handler = dirtytime_interval_handler,
2777 .extra1 = SYSCTL_ZERO,
2778 },
2779 {
2780 .procname = "swappiness",
2781 .data = &vm_swappiness,
2782 .maxlen = sizeof(vm_swappiness),
2783 .mode = 0644,
2784 .proc_handler = proc_dointvec_minmax,
2785 .extra1 = SYSCTL_ZERO,
2786 .extra2 = &two_hundred,
2787 },
2788#ifdef CONFIG_HUGETLB_PAGE
2789 {
2790 .procname = "nr_hugepages",
2791 .data = NULL,
2792 .maxlen = sizeof(unsigned long),
2793 .mode = 0644,
2794 .proc_handler = hugetlb_sysctl_handler,
2795 },
2796#ifdef CONFIG_NUMA
2797 {
2798 .procname = "nr_hugepages_mempolicy",
2799 .data = NULL,
2800 .maxlen = sizeof(unsigned long),
2801 .mode = 0644,
2802 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
2803 },
2804 {
2805 .procname = "numa_stat",
2806 .data = &sysctl_vm_numa_stat,
2807 .maxlen = sizeof(int),
2808 .mode = 0644,
2809 .proc_handler = sysctl_vm_numa_stat_handler,
2810 .extra1 = SYSCTL_ZERO,
2811 .extra2 = SYSCTL_ONE,
2812 },
2813#endif
2814 {
2815 .procname = "hugetlb_shm_group",
2816 .data = &sysctl_hugetlb_shm_group,
2817 .maxlen = sizeof(gid_t),
2818 .mode = 0644,
2819 .proc_handler = proc_dointvec,
2820 },
2821 {
2822 .procname = "nr_overcommit_hugepages",
2823 .data = NULL,
2824 .maxlen = sizeof(unsigned long),
2825 .mode = 0644,
2826 .proc_handler = hugetlb_overcommit_handler,
2827 },
2828#endif
2829 {
2830 .procname = "lowmem_reserve_ratio",
2831 .data = &sysctl_lowmem_reserve_ratio,
2832 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
2833 .mode = 0644,
2834 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
2835 },
2836 {
2837 .procname = "drop_caches",
2838 .data = &sysctl_drop_caches,
2839 .maxlen = sizeof(int),
2840 .mode = 0200,
2841 .proc_handler = drop_caches_sysctl_handler,
2842 .extra1 = SYSCTL_ONE,
2843 .extra2 = &four,
2844 },
2845#ifdef CONFIG_COMPACTION
2846 {
2847 .procname = "compact_memory",
2848 .data = &sysctl_compact_memory,
2849 .maxlen = sizeof(int),
2850 .mode = 0200,
2851 .proc_handler = sysctl_compaction_handler,
2852 },
2853 {
2854 .procname = "compaction_proactiveness",
2855 .data = &sysctl_compaction_proactiveness,
2856 .maxlen = sizeof(sysctl_compaction_proactiveness),
2857 .mode = 0644,
2858 .proc_handler = proc_dointvec_minmax,
2859 .extra1 = SYSCTL_ZERO,
2860 .extra2 = &one_hundred,
2861 },
2862 {
2863 .procname = "extfrag_threshold",
2864 .data = &sysctl_extfrag_threshold,
2865 .maxlen = sizeof(int),
2866 .mode = 0644,
2867 .proc_handler = proc_dointvec_minmax,
2868 .extra1 = &min_extfrag_threshold,
2869 .extra2 = &max_extfrag_threshold,
2870 },
2871 {
2872 .procname = "compact_unevictable_allowed",
2873 .data = &sysctl_compact_unevictable_allowed,
2874 .maxlen = sizeof(int),
2875 .mode = 0644,
2876 .proc_handler = proc_dointvec_minmax_warn_RT_change,
2877 .extra1 = SYSCTL_ZERO,
2878 .extra2 = SYSCTL_ONE,
2879 },
2880
2881#endif /* CONFIG_COMPACTION */
2882 {
2883 .procname = "min_free_kbytes",
2884 .data = &min_free_kbytes,
2885 .maxlen = sizeof(min_free_kbytes),
2886 .mode = 0644,
2887 .proc_handler = min_free_kbytes_sysctl_handler,
2888 .extra1 = SYSCTL_ZERO,
2889 },
2890 {
2891 .procname = "watermark_boost_factor",
2892 .data = &watermark_boost_factor,
2893 .maxlen = sizeof(watermark_boost_factor),
2894 .mode = 0644,
2895 .proc_handler = proc_dointvec_minmax,
2896 .extra1 = SYSCTL_ZERO,
2897 },
2898 {
2899 .procname = "watermark_scale_factor",
2900 .data = &watermark_scale_factor,
2901 .maxlen = sizeof(watermark_scale_factor),
2902 .mode = 0644,
2903 .proc_handler = watermark_scale_factor_sysctl_handler,
2904 .extra1 = SYSCTL_ONE,
2905 .extra2 = &one_thousand,
2906 },
2907 {
2908 .procname = "percpu_pagelist_fraction",
2909 .data = &percpu_pagelist_fraction,
2910 .maxlen = sizeof(percpu_pagelist_fraction),
2911 .mode = 0644,
2912 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
2913 .extra1 = SYSCTL_ZERO,
2914 },
2915 {
2916 .procname = "page_lock_unfairness",
2917 .data = &sysctl_page_lock_unfairness,
2918 .maxlen = sizeof(sysctl_page_lock_unfairness),
2919 .mode = 0644,
2920 .proc_handler = proc_dointvec_minmax,
2921 .extra1 = SYSCTL_ZERO,
2922 },
2923#ifdef CONFIG_MMU
2924 {
2925 .procname = "max_map_count",
2926 .data = &sysctl_max_map_count,
2927 .maxlen = sizeof(sysctl_max_map_count),
2928 .mode = 0644,
2929 .proc_handler = proc_dointvec_minmax,
2930 .extra1 = SYSCTL_ZERO,
2931 },
2932#else
2933 {
2934 .procname = "nr_trim_pages",
2935 .data = &sysctl_nr_trim_pages,
2936 .maxlen = sizeof(sysctl_nr_trim_pages),
2937 .mode = 0644,
2938 .proc_handler = proc_dointvec_minmax,
2939 .extra1 = SYSCTL_ZERO,
2940 },
2941#endif
2942 {
2943 .procname = "laptop_mode",
2944 .data = &laptop_mode,
2945 .maxlen = sizeof(laptop_mode),
2946 .mode = 0644,
2947 .proc_handler = proc_dointvec_jiffies,
2948 },
2949 {
2950 .procname = "block_dump",
2951 .data = &block_dump,
2952 .maxlen = sizeof(block_dump),
2953 .mode = 0644,
2954 .proc_handler = proc_dointvec,
2955 .extra1 = SYSCTL_ZERO,
2956 },
2957 {
2958 .procname = "vfs_cache_pressure",
2959 .data = &sysctl_vfs_cache_pressure,
2960 .maxlen = sizeof(sysctl_vfs_cache_pressure),
2961 .mode = 0644,
2962 .proc_handler = proc_dointvec,
2963 .extra1 = SYSCTL_ZERO,
2964 },
2965#if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
2966 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
2967 {
2968 .procname = "legacy_va_layout",
2969 .data = &sysctl_legacy_va_layout,
2970 .maxlen = sizeof(sysctl_legacy_va_layout),
2971 .mode = 0644,
2972 .proc_handler = proc_dointvec,
2973 .extra1 = SYSCTL_ZERO,
2974 },
2975#endif
2976#ifdef CONFIG_NUMA
2977 {
2978 .procname = "zone_reclaim_mode",
2979 .data = &node_reclaim_mode,
2980 .maxlen = sizeof(node_reclaim_mode),
2981 .mode = 0644,
2982 .proc_handler = proc_dointvec,
2983 .extra1 = SYSCTL_ZERO,
2984 },
2985 {
2986 .procname = "min_unmapped_ratio",
2987 .data = &sysctl_min_unmapped_ratio,
2988 .maxlen = sizeof(sysctl_min_unmapped_ratio),
2989 .mode = 0644,
2990 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
2991 .extra1 = SYSCTL_ZERO,
2992 .extra2 = &one_hundred,
2993 },
2994 {
2995 .procname = "min_slab_ratio",
2996 .data = &sysctl_min_slab_ratio,
2997 .maxlen = sizeof(sysctl_min_slab_ratio),
2998 .mode = 0644,
2999 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
3000 .extra1 = SYSCTL_ZERO,
3001 .extra2 = &one_hundred,
3002 },
3003#endif
3004#ifdef CONFIG_SMP
3005 {
3006 .procname = "stat_interval",
3007 .data = &sysctl_stat_interval,
3008 .maxlen = sizeof(sysctl_stat_interval),
3009 .mode = 0644,
3010 .proc_handler = proc_dointvec_jiffies,
3011 },
3012 {
3013 .procname = "stat_refresh",
3014 .data = NULL,
3015 .maxlen = 0,
3016 .mode = 0600,
3017 .proc_handler = vmstat_refresh,
3018 },
3019#endif
3020#ifdef CONFIG_MMU
3021 {
3022 .procname = "mmap_min_addr",
3023 .data = &dac_mmap_min_addr,
3024 .maxlen = sizeof(unsigned long),
3025 .mode = 0644,
3026 .proc_handler = mmap_min_addr_handler,
3027 },
3028#endif
3029#ifdef CONFIG_NUMA
3030 {
3031 .procname = "numa_zonelist_order",
3032 .data = &numa_zonelist_order,
3033 .maxlen = NUMA_ZONELIST_ORDER_LEN,
3034 .mode = 0644,
3035 .proc_handler = numa_zonelist_order_handler,
3036 },
3037#endif
3038#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
3039 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
3040 {
3041 .procname = "vdso_enabled",
3042#ifdef CONFIG_X86_32
3043 .data = &vdso32_enabled,
3044 .maxlen = sizeof(vdso32_enabled),
3045#else
3046 .data = &vdso_enabled,
3047 .maxlen = sizeof(vdso_enabled),
3048#endif
3049 .mode = 0644,
3050 .proc_handler = proc_dointvec,
3051 .extra1 = SYSCTL_ZERO,
3052 },
3053#endif
3054#ifdef CONFIG_HIGHMEM
3055 {
3056 .procname = "highmem_is_dirtyable",
3057 .data = &vm_highmem_is_dirtyable,
3058 .maxlen = sizeof(vm_highmem_is_dirtyable),
3059 .mode = 0644,
3060 .proc_handler = proc_dointvec_minmax,
3061 .extra1 = SYSCTL_ZERO,
3062 .extra2 = SYSCTL_ONE,
3063 },
3064#endif
3065#ifdef CONFIG_MEMORY_FAILURE
3066 {
3067 .procname = "memory_failure_early_kill",
3068 .data = &sysctl_memory_failure_early_kill,
3069 .maxlen = sizeof(sysctl_memory_failure_early_kill),
3070 .mode = 0644,
3071 .proc_handler = proc_dointvec_minmax,
3072 .extra1 = SYSCTL_ZERO,
3073 .extra2 = SYSCTL_ONE,
3074 },
3075 {
3076 .procname = "memory_failure_recovery",
3077 .data = &sysctl_memory_failure_recovery,
3078 .maxlen = sizeof(sysctl_memory_failure_recovery),
3079 .mode = 0644,
3080 .proc_handler = proc_dointvec_minmax,
3081 .extra1 = SYSCTL_ZERO,
3082 .extra2 = SYSCTL_ONE,
3083 },
3084#endif
3085 {
3086 .procname = "user_reserve_kbytes",
3087 .data = &sysctl_user_reserve_kbytes,
3088 .maxlen = sizeof(sysctl_user_reserve_kbytes),
3089 .mode = 0644,
3090 .proc_handler = proc_doulongvec_minmax,
3091 },
3092 {
3093 .procname = "admin_reserve_kbytes",
3094 .data = &sysctl_admin_reserve_kbytes,
3095 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
3096 .mode = 0644,
3097 .proc_handler = proc_doulongvec_minmax,
3098 },
3099#ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
3100 {
3101 .procname = "mmap_rnd_bits",
3102 .data = &mmap_rnd_bits,
3103 .maxlen = sizeof(mmap_rnd_bits),
3104 .mode = 0600,
3105 .proc_handler = proc_dointvec_minmax,
3106 .extra1 = (void *)&mmap_rnd_bits_min,
3107 .extra2 = (void *)&mmap_rnd_bits_max,
3108 },
3109#endif
3110#ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
3111 {
3112 .procname = "mmap_rnd_compat_bits",
3113 .data = &mmap_rnd_compat_bits,
3114 .maxlen = sizeof(mmap_rnd_compat_bits),
3115 .mode = 0600,
3116 .proc_handler = proc_dointvec_minmax,
3117 .extra1 = (void *)&mmap_rnd_compat_bits_min,
3118 .extra2 = (void *)&mmap_rnd_compat_bits_max,
3119 },
3120#endif
3121#ifdef CONFIG_USERFAULTFD
3122 {
3123 .procname = "unprivileged_userfaultfd",
3124 .data = &sysctl_unprivileged_userfaultfd,
3125 .maxlen = sizeof(sysctl_unprivileged_userfaultfd),
3126 .mode = 0644,
3127 .proc_handler = proc_dointvec_minmax,
3128 .extra1 = SYSCTL_ZERO,
3129 .extra2 = SYSCTL_ONE,
3130 },
3131#endif
3132 { }
3133};
3134
3135static struct ctl_table fs_table[] = {
3136 {
3137 .procname = "inode-nr",
3138 .data = &inodes_stat,
3139 .maxlen = 2*sizeof(long),
3140 .mode = 0444,
3141 .proc_handler = proc_nr_inodes,
3142 },
3143 {
3144 .procname = "inode-state",
3145 .data = &inodes_stat,
3146 .maxlen = 7*sizeof(long),
3147 .mode = 0444,
3148 .proc_handler = proc_nr_inodes,
3149 },
3150 {
3151 .procname = "file-nr",
3152 .data = &files_stat,
3153 .maxlen = sizeof(files_stat),
3154 .mode = 0444,
3155 .proc_handler = proc_nr_files,
3156 },
3157 {
3158 .procname = "file-max",
3159 .data = &files_stat.max_files,
3160 .maxlen = sizeof(files_stat.max_files),
3161 .mode = 0644,
3162 .proc_handler = proc_doulongvec_minmax,
3163 .extra1 = &zero_ul,
3164 .extra2 = &long_max,
3165 },
3166 {
3167 .procname = "nr_open",
3168 .data = &sysctl_nr_open,
3169 .maxlen = sizeof(unsigned int),
3170 .mode = 0644,
3171 .proc_handler = proc_dointvec_minmax,
3172 .extra1 = &sysctl_nr_open_min,
3173 .extra2 = &sysctl_nr_open_max,
3174 },
3175 {
3176 .procname = "dentry-state",
3177 .data = &dentry_stat,
3178 .maxlen = 6*sizeof(long),
3179 .mode = 0444,
3180 .proc_handler = proc_nr_dentry,
3181 },
3182 {
3183 .procname = "overflowuid",
3184 .data = &fs_overflowuid,
3185 .maxlen = sizeof(int),
3186 .mode = 0644,
3187 .proc_handler = proc_dointvec_minmax,
3188 .extra1 = &minolduid,
3189 .extra2 = &maxolduid,
3190 },
3191 {
3192 .procname = "overflowgid",
3193 .data = &fs_overflowgid,
3194 .maxlen = sizeof(int),
3195 .mode = 0644,
3196 .proc_handler = proc_dointvec_minmax,
3197 .extra1 = &minolduid,
3198 .extra2 = &maxolduid,
3199 },
3200#ifdef CONFIG_FILE_LOCKING
3201 {
3202 .procname = "leases-enable",
3203 .data = &leases_enable,
3204 .maxlen = sizeof(int),
3205 .mode = 0644,
3206 .proc_handler = proc_dointvec,
3207 },
3208#endif
3209#ifdef CONFIG_DNOTIFY
3210 {
3211 .procname = "dir-notify-enable",
3212 .data = &dir_notify_enable,
3213 .maxlen = sizeof(int),
3214 .mode = 0644,
3215 .proc_handler = proc_dointvec,
3216 },
3217#endif
3218#ifdef CONFIG_MMU
3219#ifdef CONFIG_FILE_LOCKING
3220 {
3221 .procname = "lease-break-time",
3222 .data = &lease_break_time,
3223 .maxlen = sizeof(int),
3224 .mode = 0644,
3225 .proc_handler = proc_dointvec,
3226 },
3227#endif
3228#ifdef CONFIG_AIO
3229 {
3230 .procname = "aio-nr",
3231 .data = &aio_nr,
3232 .maxlen = sizeof(aio_nr),
3233 .mode = 0444,
3234 .proc_handler = proc_doulongvec_minmax,
3235 },
3236 {
3237 .procname = "aio-max-nr",
3238 .data = &aio_max_nr,
3239 .maxlen = sizeof(aio_max_nr),
3240 .mode = 0644,
3241 .proc_handler = proc_doulongvec_minmax,
3242 },
3243#endif /* CONFIG_AIO */
3244#ifdef CONFIG_INOTIFY_USER
3245 {
3246 .procname = "inotify",
3247 .mode = 0555,
3248 .child = inotify_table,
3249 },
3250#endif
3251#ifdef CONFIG_EPOLL
3252 {
3253 .procname = "epoll",
3254 .mode = 0555,
3255 .child = epoll_table,
3256 },
3257#endif
3258#endif
3259 {
3260 .procname = "protected_symlinks",
3261 .data = &sysctl_protected_symlinks,
3262 .maxlen = sizeof(int),
3263 .mode = 0600,
3264 .proc_handler = proc_dointvec_minmax,
3265 .extra1 = SYSCTL_ZERO,
3266 .extra2 = SYSCTL_ONE,
3267 },
3268 {
3269 .procname = "protected_hardlinks",
3270 .data = &sysctl_protected_hardlinks,
3271 .maxlen = sizeof(int),
3272 .mode = 0600,
3273 .proc_handler = proc_dointvec_minmax,
3274 .extra1 = SYSCTL_ZERO,
3275 .extra2 = SYSCTL_ONE,
3276 },
3277 {
3278 .procname = "protected_fifos",
3279 .data = &sysctl_protected_fifos,
3280 .maxlen = sizeof(int),
3281 .mode = 0600,
3282 .proc_handler = proc_dointvec_minmax,
3283 .extra1 = SYSCTL_ZERO,
3284 .extra2 = &two,
3285 },
3286 {
3287 .procname = "protected_regular",
3288 .data = &sysctl_protected_regular,
3289 .maxlen = sizeof(int),
3290 .mode = 0600,
3291 .proc_handler = proc_dointvec_minmax,
3292 .extra1 = SYSCTL_ZERO,
3293 .extra2 = &two,
3294 },
3295 {
3296 .procname = "suid_dumpable",
3297 .data = &suid_dumpable,
3298 .maxlen = sizeof(int),
3299 .mode = 0644,
3300 .proc_handler = proc_dointvec_minmax_coredump,
3301 .extra1 = SYSCTL_ZERO,
3302 .extra2 = &two,
3303 },
3304#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
3305 {
3306 .procname = "binfmt_misc",
3307 .mode = 0555,
3308 .child = sysctl_mount_point,
3309 },
3310#endif
3311 {
3312 .procname = "pipe-max-size",
3313 .data = &pipe_max_size,
3314 .maxlen = sizeof(pipe_max_size),
3315 .mode = 0644,
3316 .proc_handler = proc_dopipe_max_size,
3317 },
3318 {
3319 .procname = "pipe-user-pages-hard",
3320 .data = &pipe_user_pages_hard,
3321 .maxlen = sizeof(pipe_user_pages_hard),
3322 .mode = 0644,
3323 .proc_handler = proc_doulongvec_minmax,
3324 },
3325 {
3326 .procname = "pipe-user-pages-soft",
3327 .data = &pipe_user_pages_soft,
3328 .maxlen = sizeof(pipe_user_pages_soft),
3329 .mode = 0644,
3330 .proc_handler = proc_doulongvec_minmax,
3331 },
3332 {
3333 .procname = "mount-max",
3334 .data = &sysctl_mount_max,
3335 .maxlen = sizeof(unsigned int),
3336 .mode = 0644,
3337 .proc_handler = proc_dointvec_minmax,
3338 .extra1 = SYSCTL_ONE,
3339 },
3340 { }
3341};
3342
3343static struct ctl_table debug_table[] = {
3344#ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
3345 {
3346 .procname = "exception-trace",
3347 .data = &show_unhandled_signals,
3348 .maxlen = sizeof(int),
3349 .mode = 0644,
3350 .proc_handler = proc_dointvec
3351 },
3352#endif
3353#if defined(CONFIG_OPTPROBES)
3354 {
3355 .procname = "kprobes-optimization",
3356 .data = &sysctl_kprobes_optimization,
3357 .maxlen = sizeof(int),
3358 .mode = 0644,
3359 .proc_handler = proc_kprobes_optimization_handler,
3360 .extra1 = SYSCTL_ZERO,
3361 .extra2 = SYSCTL_ONE,
3362 },
3363#endif
3364 { }
3365};
3366
3367static struct ctl_table dev_table[] = {
3368 { }
3369};
3370
3371static struct ctl_table sysctl_base_table[] = {
3372 {
3373 .procname = "kernel",
3374 .mode = 0555,
3375 .child = kern_table,
3376 },
3377 {
3378 .procname = "vm",
3379 .mode = 0555,
3380 .child = vm_table,
3381 },
3382 {
3383 .procname = "fs",
3384 .mode = 0555,
3385 .child = fs_table,
3386 },
3387 {
3388 .procname = "debug",
3389 .mode = 0555,
3390 .child = debug_table,
3391 },
3392 {
3393 .procname = "dev",
3394 .mode = 0555,
3395 .child = dev_table,
3396 },
3397 { }
3398};
3399
3400int __init sysctl_init(void)
3401{
3402 struct ctl_table_header *hdr;
3403
3404 hdr = register_sysctl_table(sysctl_base_table);
3405 kmemleak_not_leak(hdr);
3406 return 0;
3407}
3408#endif /* CONFIG_SYSCTL */
3409/*
3410 * No sense putting this after each symbol definition, twice,
3411 * exception granted :-)
3412 */
3413EXPORT_SYMBOL(proc_dointvec);
3414EXPORT_SYMBOL(proc_douintvec);
3415EXPORT_SYMBOL(proc_dointvec_jiffies);
3416EXPORT_SYMBOL(proc_dointvec_minmax);
3417EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3418EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3419EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3420EXPORT_SYMBOL(proc_dostring);
3421EXPORT_SYMBOL(proc_doulongvec_minmax);
3422EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3423EXPORT_SYMBOL(proc_do_large_bitmap);