Loading...
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
72#include "../lib/kstrtox.h"
73
74#include <linux/uaccess.h>
75#include <asm/processor.h>
76
77#ifdef CONFIG_X86
78#include <asm/nmi.h>
79#include <asm/stacktrace.h>
80#include <asm/io.h>
81#endif
82#ifdef CONFIG_SPARC
83#include <asm/setup.h>
84#endif
85#ifdef CONFIG_BSD_PROCESS_ACCT
86#include <linux/acct.h>
87#endif
88#ifdef CONFIG_RT_MUTEXES
89#include <linux/rtmutex.h>
90#endif
91#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
92#include <linux/lockdep.h>
93#endif
94#ifdef CONFIG_CHR_DEV_SG
95#include <scsi/sg.h>
96#endif
97#ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
98#include <linux/stackleak.h>
99#endif
100#ifdef CONFIG_LOCKUP_DETECTOR
101#include <linux/nmi.h>
102#endif
103
104#if defined(CONFIG_SYSCTL)
105
106/* External variables not in a header file. */
107extern int suid_dumpable;
108#ifdef CONFIG_COREDUMP
109extern int core_uses_pid;
110extern char core_pattern[];
111extern unsigned int core_pipe_limit;
112#endif
113extern int pid_max;
114extern int pid_max_min, pid_max_max;
115extern int percpu_pagelist_fraction;
116extern int latencytop_enabled;
117extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
118#ifndef CONFIG_MMU
119extern int sysctl_nr_trim_pages;
120#endif
121
122/* Constants used for minimum and maximum */
123#ifdef CONFIG_LOCKUP_DETECTOR
124static int sixty = 60;
125#endif
126
127static int __maybe_unused neg_one = -1;
128static int __maybe_unused two = 2;
129static int __maybe_unused four = 4;
130static unsigned long zero_ul;
131static unsigned long one_ul = 1;
132static unsigned long long_max = LONG_MAX;
133static int one_hundred = 100;
134static int one_thousand = 1000;
135#ifdef CONFIG_PRINTK
136static int ten_thousand = 10000;
137#endif
138#ifdef CONFIG_PERF_EVENTS
139static int six_hundred_forty_kb = 640 * 1024;
140#endif
141
142/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
143static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
144
145/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
146static int maxolduid = 65535;
147static int minolduid;
148
149static int ngroups_max = NGROUPS_MAX;
150static const int cap_last_cap = CAP_LAST_CAP;
151
152/*
153 * This is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs
154 * and hung_task_check_interval_secs
155 */
156#ifdef CONFIG_DETECT_HUNG_TASK
157static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
158#endif
159
160#ifdef CONFIG_INOTIFY_USER
161#include <linux/inotify.h>
162#endif
163#ifdef CONFIG_SPARC
164#endif
165
166#ifdef CONFIG_PARISC
167extern int pwrsw_enabled;
168#endif
169
170#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
171extern int unaligned_enabled;
172#endif
173
174#ifdef CONFIG_IA64
175extern int unaligned_dump_stack;
176#endif
177
178#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
179extern int no_unaligned_warning;
180#endif
181
182#ifdef CONFIG_PROC_SYSCTL
183
184/**
185 * enum sysctl_writes_mode - supported sysctl write modes
186 *
187 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
188 * to be written, and multiple writes on the same sysctl file descriptor
189 * will rewrite the sysctl value, regardless of file position. No warning
190 * is issued when the initial position is not 0.
191 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
192 * not 0.
193 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
194 * file position 0 and the value must be fully contained in the buffer
195 * sent to the write syscall. If dealing with strings respect the file
196 * position, but restrict this to the max length of the buffer, anything
197 * passed the max length will be ignored. Multiple writes will append
198 * to the buffer.
199 *
200 * These write modes control how current file position affects the behavior of
201 * updating sysctl values through the proc interface on each write.
202 */
203enum sysctl_writes_mode {
204 SYSCTL_WRITES_LEGACY = -1,
205 SYSCTL_WRITES_WARN = 0,
206 SYSCTL_WRITES_STRICT = 1,
207};
208
209static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
210
211static int proc_do_cad_pid(struct ctl_table *table, int write,
212 void __user *buffer, size_t *lenp, loff_t *ppos);
213static int proc_taint(struct ctl_table *table, int write,
214 void __user *buffer, size_t *lenp, loff_t *ppos);
215#endif
216
217#ifdef CONFIG_PRINTK
218static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
219 void __user *buffer, size_t *lenp, loff_t *ppos);
220#endif
221
222static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
223 void __user *buffer, size_t *lenp, loff_t *ppos);
224#ifdef CONFIG_COREDUMP
225static int proc_dostring_coredump(struct ctl_table *table, int write,
226 void __user *buffer, size_t *lenp, loff_t *ppos);
227#endif
228static int proc_dopipe_max_size(struct ctl_table *table, int write,
229 void __user *buffer, size_t *lenp, loff_t *ppos);
230
231#ifdef CONFIG_MAGIC_SYSRQ
232/* Note: sysrq code uses its own private copy */
233static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
234
235static int sysrq_sysctl_handler(struct ctl_table *table, int write,
236 void __user *buffer, size_t *lenp,
237 loff_t *ppos)
238{
239 int error;
240
241 error = proc_dointvec(table, write, buffer, lenp, ppos);
242 if (error)
243 return error;
244
245 if (write)
246 sysrq_toggle_support(__sysrq_enabled);
247
248 return 0;
249}
250
251#endif
252
253static struct ctl_table kern_table[];
254static struct ctl_table vm_table[];
255static struct ctl_table fs_table[];
256static struct ctl_table debug_table[];
257static struct ctl_table dev_table[];
258extern struct ctl_table random_table[];
259#ifdef CONFIG_EPOLL
260extern struct ctl_table epoll_table[];
261#endif
262
263#ifdef CONFIG_FW_LOADER_USER_HELPER
264extern struct ctl_table firmware_config_table[];
265#endif
266
267#if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
268 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
269int sysctl_legacy_va_layout;
270#endif
271
272/* The default sysctl tables: */
273
274static struct ctl_table sysctl_base_table[] = {
275 {
276 .procname = "kernel",
277 .mode = 0555,
278 .child = kern_table,
279 },
280 {
281 .procname = "vm",
282 .mode = 0555,
283 .child = vm_table,
284 },
285 {
286 .procname = "fs",
287 .mode = 0555,
288 .child = fs_table,
289 },
290 {
291 .procname = "debug",
292 .mode = 0555,
293 .child = debug_table,
294 },
295 {
296 .procname = "dev",
297 .mode = 0555,
298 .child = dev_table,
299 },
300 { }
301};
302
303#ifdef CONFIG_SCHED_DEBUG
304static int min_sched_granularity_ns = 100000; /* 100 usecs */
305static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
306static int min_wakeup_granularity_ns; /* 0 usecs */
307static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
308#ifdef CONFIG_SMP
309static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
310static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
311#endif /* CONFIG_SMP */
312#endif /* CONFIG_SCHED_DEBUG */
313
314#ifdef CONFIG_COMPACTION
315static int min_extfrag_threshold;
316static int max_extfrag_threshold = 1000;
317#endif
318
319static struct ctl_table kern_table[] = {
320 {
321 .procname = "sched_child_runs_first",
322 .data = &sysctl_sched_child_runs_first,
323 .maxlen = sizeof(unsigned int),
324 .mode = 0644,
325 .proc_handler = proc_dointvec,
326 },
327#ifdef CONFIG_SCHED_DEBUG
328 {
329 .procname = "sched_min_granularity_ns",
330 .data = &sysctl_sched_min_granularity,
331 .maxlen = sizeof(unsigned int),
332 .mode = 0644,
333 .proc_handler = sched_proc_update_handler,
334 .extra1 = &min_sched_granularity_ns,
335 .extra2 = &max_sched_granularity_ns,
336 },
337 {
338 .procname = "sched_latency_ns",
339 .data = &sysctl_sched_latency,
340 .maxlen = sizeof(unsigned int),
341 .mode = 0644,
342 .proc_handler = sched_proc_update_handler,
343 .extra1 = &min_sched_granularity_ns,
344 .extra2 = &max_sched_granularity_ns,
345 },
346 {
347 .procname = "sched_wakeup_granularity_ns",
348 .data = &sysctl_sched_wakeup_granularity,
349 .maxlen = sizeof(unsigned int),
350 .mode = 0644,
351 .proc_handler = sched_proc_update_handler,
352 .extra1 = &min_wakeup_granularity_ns,
353 .extra2 = &max_wakeup_granularity_ns,
354 },
355#ifdef CONFIG_SMP
356 {
357 .procname = "sched_tunable_scaling",
358 .data = &sysctl_sched_tunable_scaling,
359 .maxlen = sizeof(enum sched_tunable_scaling),
360 .mode = 0644,
361 .proc_handler = sched_proc_update_handler,
362 .extra1 = &min_sched_tunable_scaling,
363 .extra2 = &max_sched_tunable_scaling,
364 },
365 {
366 .procname = "sched_migration_cost_ns",
367 .data = &sysctl_sched_migration_cost,
368 .maxlen = sizeof(unsigned int),
369 .mode = 0644,
370 .proc_handler = proc_dointvec,
371 },
372 {
373 .procname = "sched_nr_migrate",
374 .data = &sysctl_sched_nr_migrate,
375 .maxlen = sizeof(unsigned int),
376 .mode = 0644,
377 .proc_handler = proc_dointvec,
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 = SYSCTL_ZERO,
387 .extra2 = SYSCTL_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 = SYSCTL_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 = SYSCTL_ZERO,
428 .extra2 = SYSCTL_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_UCLAMP_TASK
454 {
455 .procname = "sched_util_clamp_min",
456 .data = &sysctl_sched_uclamp_util_min,
457 .maxlen = sizeof(unsigned int),
458 .mode = 0644,
459 .proc_handler = sysctl_sched_uclamp_handler,
460 },
461 {
462 .procname = "sched_util_clamp_max",
463 .data = &sysctl_sched_uclamp_util_max,
464 .maxlen = sizeof(unsigned int),
465 .mode = 0644,
466 .proc_handler = sysctl_sched_uclamp_handler,
467 },
468#endif
469#ifdef CONFIG_SCHED_AUTOGROUP
470 {
471 .procname = "sched_autogroup_enabled",
472 .data = &sysctl_sched_autogroup_enabled,
473 .maxlen = sizeof(unsigned int),
474 .mode = 0644,
475 .proc_handler = proc_dointvec_minmax,
476 .extra1 = SYSCTL_ZERO,
477 .extra2 = SYSCTL_ONE,
478 },
479#endif
480#ifdef CONFIG_CFS_BANDWIDTH
481 {
482 .procname = "sched_cfs_bandwidth_slice_us",
483 .data = &sysctl_sched_cfs_bandwidth_slice,
484 .maxlen = sizeof(unsigned int),
485 .mode = 0644,
486 .proc_handler = proc_dointvec_minmax,
487 .extra1 = SYSCTL_ONE,
488 },
489#endif
490#if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_CPU_FREQ_GOV_SCHEDUTIL)
491 {
492 .procname = "sched_energy_aware",
493 .data = &sysctl_sched_energy_aware,
494 .maxlen = sizeof(unsigned int),
495 .mode = 0644,
496 .proc_handler = sched_energy_aware_handler,
497 .extra1 = SYSCTL_ZERO,
498 .extra2 = SYSCTL_ONE,
499 },
500#endif
501#ifdef CONFIG_PROVE_LOCKING
502 {
503 .procname = "prove_locking",
504 .data = &prove_locking,
505 .maxlen = sizeof(int),
506 .mode = 0644,
507 .proc_handler = proc_dointvec,
508 },
509#endif
510#ifdef CONFIG_LOCK_STAT
511 {
512 .procname = "lock_stat",
513 .data = &lock_stat,
514 .maxlen = sizeof(int),
515 .mode = 0644,
516 .proc_handler = proc_dointvec,
517 },
518#endif
519 {
520 .procname = "panic",
521 .data = &panic_timeout,
522 .maxlen = sizeof(int),
523 .mode = 0644,
524 .proc_handler = proc_dointvec,
525 },
526#ifdef CONFIG_COREDUMP
527 {
528 .procname = "core_uses_pid",
529 .data = &core_uses_pid,
530 .maxlen = sizeof(int),
531 .mode = 0644,
532 .proc_handler = proc_dointvec,
533 },
534 {
535 .procname = "core_pattern",
536 .data = core_pattern,
537 .maxlen = CORENAME_MAX_SIZE,
538 .mode = 0644,
539 .proc_handler = proc_dostring_coredump,
540 },
541 {
542 .procname = "core_pipe_limit",
543 .data = &core_pipe_limit,
544 .maxlen = sizeof(unsigned int),
545 .mode = 0644,
546 .proc_handler = proc_dointvec,
547 },
548#endif
549#ifdef CONFIG_PROC_SYSCTL
550 {
551 .procname = "tainted",
552 .maxlen = sizeof(long),
553 .mode = 0644,
554 .proc_handler = proc_taint,
555 },
556 {
557 .procname = "sysctl_writes_strict",
558 .data = &sysctl_writes_strict,
559 .maxlen = sizeof(int),
560 .mode = 0644,
561 .proc_handler = proc_dointvec_minmax,
562 .extra1 = &neg_one,
563 .extra2 = SYSCTL_ONE,
564 },
565#endif
566#ifdef CONFIG_LATENCYTOP
567 {
568 .procname = "latencytop",
569 .data = &latencytop_enabled,
570 .maxlen = sizeof(int),
571 .mode = 0644,
572 .proc_handler = sysctl_latencytop,
573 },
574#endif
575#ifdef CONFIG_BLK_DEV_INITRD
576 {
577 .procname = "real-root-dev",
578 .data = &real_root_dev,
579 .maxlen = sizeof(int),
580 .mode = 0644,
581 .proc_handler = proc_dointvec,
582 },
583#endif
584 {
585 .procname = "print-fatal-signals",
586 .data = &print_fatal_signals,
587 .maxlen = sizeof(int),
588 .mode = 0644,
589 .proc_handler = proc_dointvec,
590 },
591#ifdef CONFIG_SPARC
592 {
593 .procname = "reboot-cmd",
594 .data = reboot_command,
595 .maxlen = 256,
596 .mode = 0644,
597 .proc_handler = proc_dostring,
598 },
599 {
600 .procname = "stop-a",
601 .data = &stop_a_enabled,
602 .maxlen = sizeof (int),
603 .mode = 0644,
604 .proc_handler = proc_dointvec,
605 },
606 {
607 .procname = "scons-poweroff",
608 .data = &scons_pwroff,
609 .maxlen = sizeof (int),
610 .mode = 0644,
611 .proc_handler = proc_dointvec,
612 },
613#endif
614#ifdef CONFIG_SPARC64
615 {
616 .procname = "tsb-ratio",
617 .data = &sysctl_tsb_ratio,
618 .maxlen = sizeof (int),
619 .mode = 0644,
620 .proc_handler = proc_dointvec,
621 },
622#endif
623#ifdef CONFIG_PARISC
624 {
625 .procname = "soft-power",
626 .data = &pwrsw_enabled,
627 .maxlen = sizeof (int),
628 .mode = 0644,
629 .proc_handler = proc_dointvec,
630 },
631#endif
632#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
633 {
634 .procname = "unaligned-trap",
635 .data = &unaligned_enabled,
636 .maxlen = sizeof (int),
637 .mode = 0644,
638 .proc_handler = proc_dointvec,
639 },
640#endif
641 {
642 .procname = "ctrl-alt-del",
643 .data = &C_A_D,
644 .maxlen = sizeof(int),
645 .mode = 0644,
646 .proc_handler = proc_dointvec,
647 },
648#ifdef CONFIG_FUNCTION_TRACER
649 {
650 .procname = "ftrace_enabled",
651 .data = &ftrace_enabled,
652 .maxlen = sizeof(int),
653 .mode = 0644,
654 .proc_handler = ftrace_enable_sysctl,
655 },
656#endif
657#ifdef CONFIG_STACK_TRACER
658 {
659 .procname = "stack_tracer_enabled",
660 .data = &stack_tracer_enabled,
661 .maxlen = sizeof(int),
662 .mode = 0644,
663 .proc_handler = stack_trace_sysctl,
664 },
665#endif
666#ifdef CONFIG_TRACING
667 {
668 .procname = "ftrace_dump_on_oops",
669 .data = &ftrace_dump_on_oops,
670 .maxlen = sizeof(int),
671 .mode = 0644,
672 .proc_handler = proc_dointvec,
673 },
674 {
675 .procname = "traceoff_on_warning",
676 .data = &__disable_trace_on_warning,
677 .maxlen = sizeof(__disable_trace_on_warning),
678 .mode = 0644,
679 .proc_handler = proc_dointvec,
680 },
681 {
682 .procname = "tracepoint_printk",
683 .data = &tracepoint_printk,
684 .maxlen = sizeof(tracepoint_printk),
685 .mode = 0644,
686 .proc_handler = tracepoint_printk_sysctl,
687 },
688#endif
689#ifdef CONFIG_KEXEC_CORE
690 {
691 .procname = "kexec_load_disabled",
692 .data = &kexec_load_disabled,
693 .maxlen = sizeof(int),
694 .mode = 0644,
695 /* only handle a transition from default "0" to "1" */
696 .proc_handler = proc_dointvec_minmax,
697 .extra1 = SYSCTL_ONE,
698 .extra2 = SYSCTL_ONE,
699 },
700#endif
701#ifdef CONFIG_MODULES
702 {
703 .procname = "modprobe",
704 .data = &modprobe_path,
705 .maxlen = KMOD_PATH_LEN,
706 .mode = 0644,
707 .proc_handler = proc_dostring,
708 },
709 {
710 .procname = "modules_disabled",
711 .data = &modules_disabled,
712 .maxlen = sizeof(int),
713 .mode = 0644,
714 /* only handle a transition from default "0" to "1" */
715 .proc_handler = proc_dointvec_minmax,
716 .extra1 = SYSCTL_ONE,
717 .extra2 = SYSCTL_ONE,
718 },
719#endif
720#ifdef CONFIG_UEVENT_HELPER
721 {
722 .procname = "hotplug",
723 .data = &uevent_helper,
724 .maxlen = UEVENT_HELPER_PATH_LEN,
725 .mode = 0644,
726 .proc_handler = proc_dostring,
727 },
728#endif
729#ifdef CONFIG_CHR_DEV_SG
730 {
731 .procname = "sg-big-buff",
732 .data = &sg_big_buff,
733 .maxlen = sizeof (int),
734 .mode = 0444,
735 .proc_handler = proc_dointvec,
736 },
737#endif
738#ifdef CONFIG_BSD_PROCESS_ACCT
739 {
740 .procname = "acct",
741 .data = &acct_parm,
742 .maxlen = 3*sizeof(int),
743 .mode = 0644,
744 .proc_handler = proc_dointvec,
745 },
746#endif
747#ifdef CONFIG_MAGIC_SYSRQ
748 {
749 .procname = "sysrq",
750 .data = &__sysrq_enabled,
751 .maxlen = sizeof (int),
752 .mode = 0644,
753 .proc_handler = sysrq_sysctl_handler,
754 },
755#endif
756#ifdef CONFIG_PROC_SYSCTL
757 {
758 .procname = "cad_pid",
759 .data = NULL,
760 .maxlen = sizeof (int),
761 .mode = 0600,
762 .proc_handler = proc_do_cad_pid,
763 },
764#endif
765 {
766 .procname = "threads-max",
767 .data = NULL,
768 .maxlen = sizeof(int),
769 .mode = 0644,
770 .proc_handler = sysctl_max_threads,
771 },
772 {
773 .procname = "random",
774 .mode = 0555,
775 .child = random_table,
776 },
777 {
778 .procname = "usermodehelper",
779 .mode = 0555,
780 .child = usermodehelper_table,
781 },
782#ifdef CONFIG_FW_LOADER_USER_HELPER
783 {
784 .procname = "firmware_config",
785 .mode = 0555,
786 .child = firmware_config_table,
787 },
788#endif
789 {
790 .procname = "overflowuid",
791 .data = &overflowuid,
792 .maxlen = sizeof(int),
793 .mode = 0644,
794 .proc_handler = proc_dointvec_minmax,
795 .extra1 = &minolduid,
796 .extra2 = &maxolduid,
797 },
798 {
799 .procname = "overflowgid",
800 .data = &overflowgid,
801 .maxlen = sizeof(int),
802 .mode = 0644,
803 .proc_handler = proc_dointvec_minmax,
804 .extra1 = &minolduid,
805 .extra2 = &maxolduid,
806 },
807#ifdef CONFIG_S390
808#ifdef CONFIG_MATHEMU
809 {
810 .procname = "ieee_emulation_warnings",
811 .data = &sysctl_ieee_emulation_warnings,
812 .maxlen = sizeof(int),
813 .mode = 0644,
814 .proc_handler = proc_dointvec,
815 },
816#endif
817 {
818 .procname = "userprocess_debug",
819 .data = &show_unhandled_signals,
820 .maxlen = sizeof(int),
821 .mode = 0644,
822 .proc_handler = proc_dointvec,
823 },
824#endif
825 {
826 .procname = "pid_max",
827 .data = &pid_max,
828 .maxlen = sizeof (int),
829 .mode = 0644,
830 .proc_handler = proc_dointvec_minmax,
831 .extra1 = &pid_max_min,
832 .extra2 = &pid_max_max,
833 },
834 {
835 .procname = "panic_on_oops",
836 .data = &panic_on_oops,
837 .maxlen = sizeof(int),
838 .mode = 0644,
839 .proc_handler = proc_dointvec,
840 },
841 {
842 .procname = "panic_print",
843 .data = &panic_print,
844 .maxlen = sizeof(unsigned long),
845 .mode = 0644,
846 .proc_handler = proc_doulongvec_minmax,
847 },
848#if defined CONFIG_PRINTK
849 {
850 .procname = "printk",
851 .data = &console_loglevel,
852 .maxlen = 4*sizeof(int),
853 .mode = 0644,
854 .proc_handler = proc_dointvec,
855 },
856 {
857 .procname = "printk_ratelimit",
858 .data = &printk_ratelimit_state.interval,
859 .maxlen = sizeof(int),
860 .mode = 0644,
861 .proc_handler = proc_dointvec_jiffies,
862 },
863 {
864 .procname = "printk_ratelimit_burst",
865 .data = &printk_ratelimit_state.burst,
866 .maxlen = sizeof(int),
867 .mode = 0644,
868 .proc_handler = proc_dointvec,
869 },
870 {
871 .procname = "printk_delay",
872 .data = &printk_delay_msec,
873 .maxlen = sizeof(int),
874 .mode = 0644,
875 .proc_handler = proc_dointvec_minmax,
876 .extra1 = SYSCTL_ZERO,
877 .extra2 = &ten_thousand,
878 },
879 {
880 .procname = "printk_devkmsg",
881 .data = devkmsg_log_str,
882 .maxlen = DEVKMSG_STR_MAX_SIZE,
883 .mode = 0644,
884 .proc_handler = devkmsg_sysctl_set_loglvl,
885 },
886 {
887 .procname = "dmesg_restrict",
888 .data = &dmesg_restrict,
889 .maxlen = sizeof(int),
890 .mode = 0644,
891 .proc_handler = proc_dointvec_minmax_sysadmin,
892 .extra1 = SYSCTL_ZERO,
893 .extra2 = SYSCTL_ONE,
894 },
895 {
896 .procname = "kptr_restrict",
897 .data = &kptr_restrict,
898 .maxlen = sizeof(int),
899 .mode = 0644,
900 .proc_handler = proc_dointvec_minmax_sysadmin,
901 .extra1 = SYSCTL_ZERO,
902 .extra2 = &two,
903 },
904#endif
905 {
906 .procname = "ngroups_max",
907 .data = &ngroups_max,
908 .maxlen = sizeof (int),
909 .mode = 0444,
910 .proc_handler = proc_dointvec,
911 },
912 {
913 .procname = "cap_last_cap",
914 .data = (void *)&cap_last_cap,
915 .maxlen = sizeof(int),
916 .mode = 0444,
917 .proc_handler = proc_dointvec,
918 },
919#if defined(CONFIG_LOCKUP_DETECTOR)
920 {
921 .procname = "watchdog",
922 .data = &watchdog_user_enabled,
923 .maxlen = sizeof(int),
924 .mode = 0644,
925 .proc_handler = proc_watchdog,
926 .extra1 = SYSCTL_ZERO,
927 .extra2 = SYSCTL_ONE,
928 },
929 {
930 .procname = "watchdog_thresh",
931 .data = &watchdog_thresh,
932 .maxlen = sizeof(int),
933 .mode = 0644,
934 .proc_handler = proc_watchdog_thresh,
935 .extra1 = SYSCTL_ZERO,
936 .extra2 = &sixty,
937 },
938 {
939 .procname = "nmi_watchdog",
940 .data = &nmi_watchdog_user_enabled,
941 .maxlen = sizeof(int),
942 .mode = NMI_WATCHDOG_SYSCTL_PERM,
943 .proc_handler = proc_nmi_watchdog,
944 .extra1 = SYSCTL_ZERO,
945 .extra2 = SYSCTL_ONE,
946 },
947 {
948 .procname = "watchdog_cpumask",
949 .data = &watchdog_cpumask_bits,
950 .maxlen = NR_CPUS,
951 .mode = 0644,
952 .proc_handler = proc_watchdog_cpumask,
953 },
954#ifdef CONFIG_SOFTLOCKUP_DETECTOR
955 {
956 .procname = "soft_watchdog",
957 .data = &soft_watchdog_user_enabled,
958 .maxlen = sizeof(int),
959 .mode = 0644,
960 .proc_handler = proc_soft_watchdog,
961 .extra1 = SYSCTL_ZERO,
962 .extra2 = SYSCTL_ONE,
963 },
964 {
965 .procname = "softlockup_panic",
966 .data = &softlockup_panic,
967 .maxlen = sizeof(int),
968 .mode = 0644,
969 .proc_handler = proc_dointvec_minmax,
970 .extra1 = SYSCTL_ZERO,
971 .extra2 = SYSCTL_ONE,
972 },
973#ifdef CONFIG_SMP
974 {
975 .procname = "softlockup_all_cpu_backtrace",
976 .data = &sysctl_softlockup_all_cpu_backtrace,
977 .maxlen = sizeof(int),
978 .mode = 0644,
979 .proc_handler = proc_dointvec_minmax,
980 .extra1 = SYSCTL_ZERO,
981 .extra2 = SYSCTL_ONE,
982 },
983#endif /* CONFIG_SMP */
984#endif
985#ifdef CONFIG_HARDLOCKUP_DETECTOR
986 {
987 .procname = "hardlockup_panic",
988 .data = &hardlockup_panic,
989 .maxlen = sizeof(int),
990 .mode = 0644,
991 .proc_handler = proc_dointvec_minmax,
992 .extra1 = SYSCTL_ZERO,
993 .extra2 = SYSCTL_ONE,
994 },
995#ifdef CONFIG_SMP
996 {
997 .procname = "hardlockup_all_cpu_backtrace",
998 .data = &sysctl_hardlockup_all_cpu_backtrace,
999 .maxlen = sizeof(int),
1000 .mode = 0644,
1001 .proc_handler = proc_dointvec_minmax,
1002 .extra1 = SYSCTL_ZERO,
1003 .extra2 = SYSCTL_ONE,
1004 },
1005#endif /* CONFIG_SMP */
1006#endif
1007#endif
1008
1009#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
1010 {
1011 .procname = "unknown_nmi_panic",
1012 .data = &unknown_nmi_panic,
1013 .maxlen = sizeof (int),
1014 .mode = 0644,
1015 .proc_handler = proc_dointvec,
1016 },
1017#endif
1018#if defined(CONFIG_X86)
1019 {
1020 .procname = "panic_on_unrecovered_nmi",
1021 .data = &panic_on_unrecovered_nmi,
1022 .maxlen = sizeof(int),
1023 .mode = 0644,
1024 .proc_handler = proc_dointvec,
1025 },
1026 {
1027 .procname = "panic_on_io_nmi",
1028 .data = &panic_on_io_nmi,
1029 .maxlen = sizeof(int),
1030 .mode = 0644,
1031 .proc_handler = proc_dointvec,
1032 },
1033#ifdef CONFIG_DEBUG_STACKOVERFLOW
1034 {
1035 .procname = "panic_on_stackoverflow",
1036 .data = &sysctl_panic_on_stackoverflow,
1037 .maxlen = sizeof(int),
1038 .mode = 0644,
1039 .proc_handler = proc_dointvec,
1040 },
1041#endif
1042 {
1043 .procname = "bootloader_type",
1044 .data = &bootloader_type,
1045 .maxlen = sizeof (int),
1046 .mode = 0444,
1047 .proc_handler = proc_dointvec,
1048 },
1049 {
1050 .procname = "bootloader_version",
1051 .data = &bootloader_version,
1052 .maxlen = sizeof (int),
1053 .mode = 0444,
1054 .proc_handler = proc_dointvec,
1055 },
1056 {
1057 .procname = "io_delay_type",
1058 .data = &io_delay_type,
1059 .maxlen = sizeof(int),
1060 .mode = 0644,
1061 .proc_handler = proc_dointvec,
1062 },
1063#endif
1064#if defined(CONFIG_MMU)
1065 {
1066 .procname = "randomize_va_space",
1067 .data = &randomize_va_space,
1068 .maxlen = sizeof(int),
1069 .mode = 0644,
1070 .proc_handler = proc_dointvec,
1071 },
1072#endif
1073#if defined(CONFIG_S390) && defined(CONFIG_SMP)
1074 {
1075 .procname = "spin_retry",
1076 .data = &spin_retry,
1077 .maxlen = sizeof (int),
1078 .mode = 0644,
1079 .proc_handler = proc_dointvec,
1080 },
1081#endif
1082#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1083 {
1084 .procname = "acpi_video_flags",
1085 .data = &acpi_realmode_flags,
1086 .maxlen = sizeof (unsigned long),
1087 .mode = 0644,
1088 .proc_handler = proc_doulongvec_minmax,
1089 },
1090#endif
1091#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1092 {
1093 .procname = "ignore-unaligned-usertrap",
1094 .data = &no_unaligned_warning,
1095 .maxlen = sizeof (int),
1096 .mode = 0644,
1097 .proc_handler = proc_dointvec,
1098 },
1099#endif
1100#ifdef CONFIG_IA64
1101 {
1102 .procname = "unaligned-dump-stack",
1103 .data = &unaligned_dump_stack,
1104 .maxlen = sizeof (int),
1105 .mode = 0644,
1106 .proc_handler = proc_dointvec,
1107 },
1108#endif
1109#ifdef CONFIG_DETECT_HUNG_TASK
1110 {
1111 .procname = "hung_task_panic",
1112 .data = &sysctl_hung_task_panic,
1113 .maxlen = sizeof(int),
1114 .mode = 0644,
1115 .proc_handler = proc_dointvec_minmax,
1116 .extra1 = SYSCTL_ZERO,
1117 .extra2 = SYSCTL_ONE,
1118 },
1119 {
1120 .procname = "hung_task_check_count",
1121 .data = &sysctl_hung_task_check_count,
1122 .maxlen = sizeof(int),
1123 .mode = 0644,
1124 .proc_handler = proc_dointvec_minmax,
1125 .extra1 = SYSCTL_ZERO,
1126 },
1127 {
1128 .procname = "hung_task_timeout_secs",
1129 .data = &sysctl_hung_task_timeout_secs,
1130 .maxlen = sizeof(unsigned long),
1131 .mode = 0644,
1132 .proc_handler = proc_dohung_task_timeout_secs,
1133 .extra2 = &hung_task_timeout_max,
1134 },
1135 {
1136 .procname = "hung_task_check_interval_secs",
1137 .data = &sysctl_hung_task_check_interval_secs,
1138 .maxlen = sizeof(unsigned long),
1139 .mode = 0644,
1140 .proc_handler = proc_dohung_task_timeout_secs,
1141 .extra2 = &hung_task_timeout_max,
1142 },
1143 {
1144 .procname = "hung_task_warnings",
1145 .data = &sysctl_hung_task_warnings,
1146 .maxlen = sizeof(int),
1147 .mode = 0644,
1148 .proc_handler = proc_dointvec_minmax,
1149 .extra1 = &neg_one,
1150 },
1151#endif
1152#ifdef CONFIG_RT_MUTEXES
1153 {
1154 .procname = "max_lock_depth",
1155 .data = &max_lock_depth,
1156 .maxlen = sizeof(int),
1157 .mode = 0644,
1158 .proc_handler = proc_dointvec,
1159 },
1160#endif
1161 {
1162 .procname = "poweroff_cmd",
1163 .data = &poweroff_cmd,
1164 .maxlen = POWEROFF_CMD_PATH_LEN,
1165 .mode = 0644,
1166 .proc_handler = proc_dostring,
1167 },
1168#ifdef CONFIG_KEYS
1169 {
1170 .procname = "keys",
1171 .mode = 0555,
1172 .child = key_sysctls,
1173 },
1174#endif
1175#ifdef CONFIG_PERF_EVENTS
1176 /*
1177 * User-space scripts rely on the existence of this file
1178 * as a feature check for perf_events being enabled.
1179 *
1180 * So it's an ABI, do not remove!
1181 */
1182 {
1183 .procname = "perf_event_paranoid",
1184 .data = &sysctl_perf_event_paranoid,
1185 .maxlen = sizeof(sysctl_perf_event_paranoid),
1186 .mode = 0644,
1187 .proc_handler = proc_dointvec,
1188 },
1189 {
1190 .procname = "perf_event_mlock_kb",
1191 .data = &sysctl_perf_event_mlock,
1192 .maxlen = sizeof(sysctl_perf_event_mlock),
1193 .mode = 0644,
1194 .proc_handler = proc_dointvec,
1195 },
1196 {
1197 .procname = "perf_event_max_sample_rate",
1198 .data = &sysctl_perf_event_sample_rate,
1199 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1200 .mode = 0644,
1201 .proc_handler = perf_proc_update_handler,
1202 .extra1 = SYSCTL_ONE,
1203 },
1204 {
1205 .procname = "perf_cpu_time_max_percent",
1206 .data = &sysctl_perf_cpu_time_max_percent,
1207 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1208 .mode = 0644,
1209 .proc_handler = perf_cpu_time_max_percent_handler,
1210 .extra1 = SYSCTL_ZERO,
1211 .extra2 = &one_hundred,
1212 },
1213 {
1214 .procname = "perf_event_max_stack",
1215 .data = &sysctl_perf_event_max_stack,
1216 .maxlen = sizeof(sysctl_perf_event_max_stack),
1217 .mode = 0644,
1218 .proc_handler = perf_event_max_stack_handler,
1219 .extra1 = SYSCTL_ZERO,
1220 .extra2 = &six_hundred_forty_kb,
1221 },
1222 {
1223 .procname = "perf_event_max_contexts_per_stack",
1224 .data = &sysctl_perf_event_max_contexts_per_stack,
1225 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
1226 .mode = 0644,
1227 .proc_handler = perf_event_max_stack_handler,
1228 .extra1 = SYSCTL_ZERO,
1229 .extra2 = &one_thousand,
1230 },
1231#endif
1232 {
1233 .procname = "panic_on_warn",
1234 .data = &panic_on_warn,
1235 .maxlen = sizeof(int),
1236 .mode = 0644,
1237 .proc_handler = proc_dointvec_minmax,
1238 .extra1 = SYSCTL_ZERO,
1239 .extra2 = SYSCTL_ONE,
1240 },
1241#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1242 {
1243 .procname = "timer_migration",
1244 .data = &sysctl_timer_migration,
1245 .maxlen = sizeof(unsigned int),
1246 .mode = 0644,
1247 .proc_handler = timer_migration_handler,
1248 .extra1 = SYSCTL_ZERO,
1249 .extra2 = SYSCTL_ONE,
1250 },
1251#endif
1252#ifdef CONFIG_BPF_SYSCALL
1253 {
1254 .procname = "unprivileged_bpf_disabled",
1255 .data = &sysctl_unprivileged_bpf_disabled,
1256 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
1257 .mode = 0644,
1258 /* only handle a transition from default "0" to "1" */
1259 .proc_handler = proc_dointvec_minmax,
1260 .extra1 = SYSCTL_ONE,
1261 .extra2 = SYSCTL_ONE,
1262 },
1263 {
1264 .procname = "bpf_stats_enabled",
1265 .data = &bpf_stats_enabled_key.key,
1266 .maxlen = sizeof(bpf_stats_enabled_key),
1267 .mode = 0644,
1268 .proc_handler = proc_do_static_key,
1269 },
1270#endif
1271#if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1272 {
1273 .procname = "panic_on_rcu_stall",
1274 .data = &sysctl_panic_on_rcu_stall,
1275 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
1276 .mode = 0644,
1277 .proc_handler = proc_dointvec_minmax,
1278 .extra1 = SYSCTL_ZERO,
1279 .extra2 = SYSCTL_ONE,
1280 },
1281#endif
1282#ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
1283 {
1284 .procname = "stack_erasing",
1285 .data = NULL,
1286 .maxlen = sizeof(int),
1287 .mode = 0600,
1288 .proc_handler = stack_erasing_sysctl,
1289 .extra1 = SYSCTL_ZERO,
1290 .extra2 = SYSCTL_ONE,
1291 },
1292#endif
1293 { }
1294};
1295
1296static struct ctl_table vm_table[] = {
1297 {
1298 .procname = "overcommit_memory",
1299 .data = &sysctl_overcommit_memory,
1300 .maxlen = sizeof(sysctl_overcommit_memory),
1301 .mode = 0644,
1302 .proc_handler = proc_dointvec_minmax,
1303 .extra1 = SYSCTL_ZERO,
1304 .extra2 = &two,
1305 },
1306 {
1307 .procname = "panic_on_oom",
1308 .data = &sysctl_panic_on_oom,
1309 .maxlen = sizeof(sysctl_panic_on_oom),
1310 .mode = 0644,
1311 .proc_handler = proc_dointvec_minmax,
1312 .extra1 = SYSCTL_ZERO,
1313 .extra2 = &two,
1314 },
1315 {
1316 .procname = "oom_kill_allocating_task",
1317 .data = &sysctl_oom_kill_allocating_task,
1318 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1319 .mode = 0644,
1320 .proc_handler = proc_dointvec,
1321 },
1322 {
1323 .procname = "oom_dump_tasks",
1324 .data = &sysctl_oom_dump_tasks,
1325 .maxlen = sizeof(sysctl_oom_dump_tasks),
1326 .mode = 0644,
1327 .proc_handler = proc_dointvec,
1328 },
1329 {
1330 .procname = "overcommit_ratio",
1331 .data = &sysctl_overcommit_ratio,
1332 .maxlen = sizeof(sysctl_overcommit_ratio),
1333 .mode = 0644,
1334 .proc_handler = overcommit_ratio_handler,
1335 },
1336 {
1337 .procname = "overcommit_kbytes",
1338 .data = &sysctl_overcommit_kbytes,
1339 .maxlen = sizeof(sysctl_overcommit_kbytes),
1340 .mode = 0644,
1341 .proc_handler = overcommit_kbytes_handler,
1342 },
1343 {
1344 .procname = "page-cluster",
1345 .data = &page_cluster,
1346 .maxlen = sizeof(int),
1347 .mode = 0644,
1348 .proc_handler = proc_dointvec_minmax,
1349 .extra1 = SYSCTL_ZERO,
1350 },
1351 {
1352 .procname = "dirty_background_ratio",
1353 .data = &dirty_background_ratio,
1354 .maxlen = sizeof(dirty_background_ratio),
1355 .mode = 0644,
1356 .proc_handler = dirty_background_ratio_handler,
1357 .extra1 = SYSCTL_ZERO,
1358 .extra2 = &one_hundred,
1359 },
1360 {
1361 .procname = "dirty_background_bytes",
1362 .data = &dirty_background_bytes,
1363 .maxlen = sizeof(dirty_background_bytes),
1364 .mode = 0644,
1365 .proc_handler = dirty_background_bytes_handler,
1366 .extra1 = &one_ul,
1367 },
1368 {
1369 .procname = "dirty_ratio",
1370 .data = &vm_dirty_ratio,
1371 .maxlen = sizeof(vm_dirty_ratio),
1372 .mode = 0644,
1373 .proc_handler = dirty_ratio_handler,
1374 .extra1 = SYSCTL_ZERO,
1375 .extra2 = &one_hundred,
1376 },
1377 {
1378 .procname = "dirty_bytes",
1379 .data = &vm_dirty_bytes,
1380 .maxlen = sizeof(vm_dirty_bytes),
1381 .mode = 0644,
1382 .proc_handler = dirty_bytes_handler,
1383 .extra1 = &dirty_bytes_min,
1384 },
1385 {
1386 .procname = "dirty_writeback_centisecs",
1387 .data = &dirty_writeback_interval,
1388 .maxlen = sizeof(dirty_writeback_interval),
1389 .mode = 0644,
1390 .proc_handler = dirty_writeback_centisecs_handler,
1391 },
1392 {
1393 .procname = "dirty_expire_centisecs",
1394 .data = &dirty_expire_interval,
1395 .maxlen = sizeof(dirty_expire_interval),
1396 .mode = 0644,
1397 .proc_handler = proc_dointvec_minmax,
1398 .extra1 = SYSCTL_ZERO,
1399 },
1400 {
1401 .procname = "dirtytime_expire_seconds",
1402 .data = &dirtytime_expire_interval,
1403 .maxlen = sizeof(dirtytime_expire_interval),
1404 .mode = 0644,
1405 .proc_handler = dirtytime_interval_handler,
1406 .extra1 = SYSCTL_ZERO,
1407 },
1408 {
1409 .procname = "swappiness",
1410 .data = &vm_swappiness,
1411 .maxlen = sizeof(vm_swappiness),
1412 .mode = 0644,
1413 .proc_handler = proc_dointvec_minmax,
1414 .extra1 = SYSCTL_ZERO,
1415 .extra2 = &one_hundred,
1416 },
1417#ifdef CONFIG_HUGETLB_PAGE
1418 {
1419 .procname = "nr_hugepages",
1420 .data = NULL,
1421 .maxlen = sizeof(unsigned long),
1422 .mode = 0644,
1423 .proc_handler = hugetlb_sysctl_handler,
1424 },
1425#ifdef CONFIG_NUMA
1426 {
1427 .procname = "nr_hugepages_mempolicy",
1428 .data = NULL,
1429 .maxlen = sizeof(unsigned long),
1430 .mode = 0644,
1431 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1432 },
1433 {
1434 .procname = "numa_stat",
1435 .data = &sysctl_vm_numa_stat,
1436 .maxlen = sizeof(int),
1437 .mode = 0644,
1438 .proc_handler = sysctl_vm_numa_stat_handler,
1439 .extra1 = SYSCTL_ZERO,
1440 .extra2 = SYSCTL_ONE,
1441 },
1442#endif
1443 {
1444 .procname = "hugetlb_shm_group",
1445 .data = &sysctl_hugetlb_shm_group,
1446 .maxlen = sizeof(gid_t),
1447 .mode = 0644,
1448 .proc_handler = proc_dointvec,
1449 },
1450 {
1451 .procname = "nr_overcommit_hugepages",
1452 .data = NULL,
1453 .maxlen = sizeof(unsigned long),
1454 .mode = 0644,
1455 .proc_handler = hugetlb_overcommit_handler,
1456 },
1457#endif
1458 {
1459 .procname = "lowmem_reserve_ratio",
1460 .data = &sysctl_lowmem_reserve_ratio,
1461 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1462 .mode = 0644,
1463 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1464 },
1465 {
1466 .procname = "drop_caches",
1467 .data = &sysctl_drop_caches,
1468 .maxlen = sizeof(int),
1469 .mode = 0644,
1470 .proc_handler = drop_caches_sysctl_handler,
1471 .extra1 = SYSCTL_ONE,
1472 .extra2 = &four,
1473 },
1474#ifdef CONFIG_COMPACTION
1475 {
1476 .procname = "compact_memory",
1477 .data = &sysctl_compact_memory,
1478 .maxlen = sizeof(int),
1479 .mode = 0200,
1480 .proc_handler = sysctl_compaction_handler,
1481 },
1482 {
1483 .procname = "extfrag_threshold",
1484 .data = &sysctl_extfrag_threshold,
1485 .maxlen = sizeof(int),
1486 .mode = 0644,
1487 .proc_handler = proc_dointvec_minmax,
1488 .extra1 = &min_extfrag_threshold,
1489 .extra2 = &max_extfrag_threshold,
1490 },
1491 {
1492 .procname = "compact_unevictable_allowed",
1493 .data = &sysctl_compact_unevictable_allowed,
1494 .maxlen = sizeof(int),
1495 .mode = 0644,
1496 .proc_handler = proc_dointvec,
1497 .extra1 = SYSCTL_ZERO,
1498 .extra2 = SYSCTL_ONE,
1499 },
1500
1501#endif /* CONFIG_COMPACTION */
1502 {
1503 .procname = "min_free_kbytes",
1504 .data = &min_free_kbytes,
1505 .maxlen = sizeof(min_free_kbytes),
1506 .mode = 0644,
1507 .proc_handler = min_free_kbytes_sysctl_handler,
1508 .extra1 = SYSCTL_ZERO,
1509 },
1510 {
1511 .procname = "watermark_boost_factor",
1512 .data = &watermark_boost_factor,
1513 .maxlen = sizeof(watermark_boost_factor),
1514 .mode = 0644,
1515 .proc_handler = watermark_boost_factor_sysctl_handler,
1516 .extra1 = SYSCTL_ZERO,
1517 },
1518 {
1519 .procname = "watermark_scale_factor",
1520 .data = &watermark_scale_factor,
1521 .maxlen = sizeof(watermark_scale_factor),
1522 .mode = 0644,
1523 .proc_handler = watermark_scale_factor_sysctl_handler,
1524 .extra1 = SYSCTL_ONE,
1525 .extra2 = &one_thousand,
1526 },
1527 {
1528 .procname = "percpu_pagelist_fraction",
1529 .data = &percpu_pagelist_fraction,
1530 .maxlen = sizeof(percpu_pagelist_fraction),
1531 .mode = 0644,
1532 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1533 .extra1 = SYSCTL_ZERO,
1534 },
1535#ifdef CONFIG_MMU
1536 {
1537 .procname = "max_map_count",
1538 .data = &sysctl_max_map_count,
1539 .maxlen = sizeof(sysctl_max_map_count),
1540 .mode = 0644,
1541 .proc_handler = proc_dointvec_minmax,
1542 .extra1 = SYSCTL_ZERO,
1543 },
1544#else
1545 {
1546 .procname = "nr_trim_pages",
1547 .data = &sysctl_nr_trim_pages,
1548 .maxlen = sizeof(sysctl_nr_trim_pages),
1549 .mode = 0644,
1550 .proc_handler = proc_dointvec_minmax,
1551 .extra1 = SYSCTL_ZERO,
1552 },
1553#endif
1554 {
1555 .procname = "laptop_mode",
1556 .data = &laptop_mode,
1557 .maxlen = sizeof(laptop_mode),
1558 .mode = 0644,
1559 .proc_handler = proc_dointvec_jiffies,
1560 },
1561 {
1562 .procname = "block_dump",
1563 .data = &block_dump,
1564 .maxlen = sizeof(block_dump),
1565 .mode = 0644,
1566 .proc_handler = proc_dointvec,
1567 .extra1 = SYSCTL_ZERO,
1568 },
1569 {
1570 .procname = "vfs_cache_pressure",
1571 .data = &sysctl_vfs_cache_pressure,
1572 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1573 .mode = 0644,
1574 .proc_handler = proc_dointvec,
1575 .extra1 = SYSCTL_ZERO,
1576 },
1577#if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
1578 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
1579 {
1580 .procname = "legacy_va_layout",
1581 .data = &sysctl_legacy_va_layout,
1582 .maxlen = sizeof(sysctl_legacy_va_layout),
1583 .mode = 0644,
1584 .proc_handler = proc_dointvec,
1585 .extra1 = SYSCTL_ZERO,
1586 },
1587#endif
1588#ifdef CONFIG_NUMA
1589 {
1590 .procname = "zone_reclaim_mode",
1591 .data = &node_reclaim_mode,
1592 .maxlen = sizeof(node_reclaim_mode),
1593 .mode = 0644,
1594 .proc_handler = proc_dointvec,
1595 .extra1 = SYSCTL_ZERO,
1596 },
1597 {
1598 .procname = "min_unmapped_ratio",
1599 .data = &sysctl_min_unmapped_ratio,
1600 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1601 .mode = 0644,
1602 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1603 .extra1 = SYSCTL_ZERO,
1604 .extra2 = &one_hundred,
1605 },
1606 {
1607 .procname = "min_slab_ratio",
1608 .data = &sysctl_min_slab_ratio,
1609 .maxlen = sizeof(sysctl_min_slab_ratio),
1610 .mode = 0644,
1611 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1612 .extra1 = SYSCTL_ZERO,
1613 .extra2 = &one_hundred,
1614 },
1615#endif
1616#ifdef CONFIG_SMP
1617 {
1618 .procname = "stat_interval",
1619 .data = &sysctl_stat_interval,
1620 .maxlen = sizeof(sysctl_stat_interval),
1621 .mode = 0644,
1622 .proc_handler = proc_dointvec_jiffies,
1623 },
1624 {
1625 .procname = "stat_refresh",
1626 .data = NULL,
1627 .maxlen = 0,
1628 .mode = 0600,
1629 .proc_handler = vmstat_refresh,
1630 },
1631#endif
1632#ifdef CONFIG_MMU
1633 {
1634 .procname = "mmap_min_addr",
1635 .data = &dac_mmap_min_addr,
1636 .maxlen = sizeof(unsigned long),
1637 .mode = 0644,
1638 .proc_handler = mmap_min_addr_handler,
1639 },
1640#endif
1641#ifdef CONFIG_NUMA
1642 {
1643 .procname = "numa_zonelist_order",
1644 .data = &numa_zonelist_order,
1645 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1646 .mode = 0644,
1647 .proc_handler = numa_zonelist_order_handler,
1648 },
1649#endif
1650#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1651 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1652 {
1653 .procname = "vdso_enabled",
1654#ifdef CONFIG_X86_32
1655 .data = &vdso32_enabled,
1656 .maxlen = sizeof(vdso32_enabled),
1657#else
1658 .data = &vdso_enabled,
1659 .maxlen = sizeof(vdso_enabled),
1660#endif
1661 .mode = 0644,
1662 .proc_handler = proc_dointvec,
1663 .extra1 = SYSCTL_ZERO,
1664 },
1665#endif
1666#ifdef CONFIG_HIGHMEM
1667 {
1668 .procname = "highmem_is_dirtyable",
1669 .data = &vm_highmem_is_dirtyable,
1670 .maxlen = sizeof(vm_highmem_is_dirtyable),
1671 .mode = 0644,
1672 .proc_handler = proc_dointvec_minmax,
1673 .extra1 = SYSCTL_ZERO,
1674 .extra2 = SYSCTL_ONE,
1675 },
1676#endif
1677#ifdef CONFIG_MEMORY_FAILURE
1678 {
1679 .procname = "memory_failure_early_kill",
1680 .data = &sysctl_memory_failure_early_kill,
1681 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1682 .mode = 0644,
1683 .proc_handler = proc_dointvec_minmax,
1684 .extra1 = SYSCTL_ZERO,
1685 .extra2 = SYSCTL_ONE,
1686 },
1687 {
1688 .procname = "memory_failure_recovery",
1689 .data = &sysctl_memory_failure_recovery,
1690 .maxlen = sizeof(sysctl_memory_failure_recovery),
1691 .mode = 0644,
1692 .proc_handler = proc_dointvec_minmax,
1693 .extra1 = SYSCTL_ZERO,
1694 .extra2 = SYSCTL_ONE,
1695 },
1696#endif
1697 {
1698 .procname = "user_reserve_kbytes",
1699 .data = &sysctl_user_reserve_kbytes,
1700 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1701 .mode = 0644,
1702 .proc_handler = proc_doulongvec_minmax,
1703 },
1704 {
1705 .procname = "admin_reserve_kbytes",
1706 .data = &sysctl_admin_reserve_kbytes,
1707 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1708 .mode = 0644,
1709 .proc_handler = proc_doulongvec_minmax,
1710 },
1711#ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1712 {
1713 .procname = "mmap_rnd_bits",
1714 .data = &mmap_rnd_bits,
1715 .maxlen = sizeof(mmap_rnd_bits),
1716 .mode = 0600,
1717 .proc_handler = proc_dointvec_minmax,
1718 .extra1 = (void *)&mmap_rnd_bits_min,
1719 .extra2 = (void *)&mmap_rnd_bits_max,
1720 },
1721#endif
1722#ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1723 {
1724 .procname = "mmap_rnd_compat_bits",
1725 .data = &mmap_rnd_compat_bits,
1726 .maxlen = sizeof(mmap_rnd_compat_bits),
1727 .mode = 0600,
1728 .proc_handler = proc_dointvec_minmax,
1729 .extra1 = (void *)&mmap_rnd_compat_bits_min,
1730 .extra2 = (void *)&mmap_rnd_compat_bits_max,
1731 },
1732#endif
1733#ifdef CONFIG_USERFAULTFD
1734 {
1735 .procname = "unprivileged_userfaultfd",
1736 .data = &sysctl_unprivileged_userfaultfd,
1737 .maxlen = sizeof(sysctl_unprivileged_userfaultfd),
1738 .mode = 0644,
1739 .proc_handler = proc_dointvec_minmax,
1740 .extra1 = SYSCTL_ZERO,
1741 .extra2 = SYSCTL_ONE,
1742 },
1743#endif
1744 { }
1745};
1746
1747static struct ctl_table fs_table[] = {
1748 {
1749 .procname = "inode-nr",
1750 .data = &inodes_stat,
1751 .maxlen = 2*sizeof(long),
1752 .mode = 0444,
1753 .proc_handler = proc_nr_inodes,
1754 },
1755 {
1756 .procname = "inode-state",
1757 .data = &inodes_stat,
1758 .maxlen = 7*sizeof(long),
1759 .mode = 0444,
1760 .proc_handler = proc_nr_inodes,
1761 },
1762 {
1763 .procname = "file-nr",
1764 .data = &files_stat,
1765 .maxlen = sizeof(files_stat),
1766 .mode = 0444,
1767 .proc_handler = proc_nr_files,
1768 },
1769 {
1770 .procname = "file-max",
1771 .data = &files_stat.max_files,
1772 .maxlen = sizeof(files_stat.max_files),
1773 .mode = 0644,
1774 .proc_handler = proc_doulongvec_minmax,
1775 .extra1 = &zero_ul,
1776 .extra2 = &long_max,
1777 },
1778 {
1779 .procname = "nr_open",
1780 .data = &sysctl_nr_open,
1781 .maxlen = sizeof(unsigned int),
1782 .mode = 0644,
1783 .proc_handler = proc_dointvec_minmax,
1784 .extra1 = &sysctl_nr_open_min,
1785 .extra2 = &sysctl_nr_open_max,
1786 },
1787 {
1788 .procname = "dentry-state",
1789 .data = &dentry_stat,
1790 .maxlen = 6*sizeof(long),
1791 .mode = 0444,
1792 .proc_handler = proc_nr_dentry,
1793 },
1794 {
1795 .procname = "overflowuid",
1796 .data = &fs_overflowuid,
1797 .maxlen = sizeof(int),
1798 .mode = 0644,
1799 .proc_handler = proc_dointvec_minmax,
1800 .extra1 = &minolduid,
1801 .extra2 = &maxolduid,
1802 },
1803 {
1804 .procname = "overflowgid",
1805 .data = &fs_overflowgid,
1806 .maxlen = sizeof(int),
1807 .mode = 0644,
1808 .proc_handler = proc_dointvec_minmax,
1809 .extra1 = &minolduid,
1810 .extra2 = &maxolduid,
1811 },
1812#ifdef CONFIG_FILE_LOCKING
1813 {
1814 .procname = "leases-enable",
1815 .data = &leases_enable,
1816 .maxlen = sizeof(int),
1817 .mode = 0644,
1818 .proc_handler = proc_dointvec,
1819 },
1820#endif
1821#ifdef CONFIG_DNOTIFY
1822 {
1823 .procname = "dir-notify-enable",
1824 .data = &dir_notify_enable,
1825 .maxlen = sizeof(int),
1826 .mode = 0644,
1827 .proc_handler = proc_dointvec,
1828 },
1829#endif
1830#ifdef CONFIG_MMU
1831#ifdef CONFIG_FILE_LOCKING
1832 {
1833 .procname = "lease-break-time",
1834 .data = &lease_break_time,
1835 .maxlen = sizeof(int),
1836 .mode = 0644,
1837 .proc_handler = proc_dointvec,
1838 },
1839#endif
1840#ifdef CONFIG_AIO
1841 {
1842 .procname = "aio-nr",
1843 .data = &aio_nr,
1844 .maxlen = sizeof(aio_nr),
1845 .mode = 0444,
1846 .proc_handler = proc_doulongvec_minmax,
1847 },
1848 {
1849 .procname = "aio-max-nr",
1850 .data = &aio_max_nr,
1851 .maxlen = sizeof(aio_max_nr),
1852 .mode = 0644,
1853 .proc_handler = proc_doulongvec_minmax,
1854 },
1855#endif /* CONFIG_AIO */
1856#ifdef CONFIG_INOTIFY_USER
1857 {
1858 .procname = "inotify",
1859 .mode = 0555,
1860 .child = inotify_table,
1861 },
1862#endif
1863#ifdef CONFIG_EPOLL
1864 {
1865 .procname = "epoll",
1866 .mode = 0555,
1867 .child = epoll_table,
1868 },
1869#endif
1870#endif
1871 {
1872 .procname = "protected_symlinks",
1873 .data = &sysctl_protected_symlinks,
1874 .maxlen = sizeof(int),
1875 .mode = 0600,
1876 .proc_handler = proc_dointvec_minmax,
1877 .extra1 = SYSCTL_ZERO,
1878 .extra2 = SYSCTL_ONE,
1879 },
1880 {
1881 .procname = "protected_hardlinks",
1882 .data = &sysctl_protected_hardlinks,
1883 .maxlen = sizeof(int),
1884 .mode = 0600,
1885 .proc_handler = proc_dointvec_minmax,
1886 .extra1 = SYSCTL_ZERO,
1887 .extra2 = SYSCTL_ONE,
1888 },
1889 {
1890 .procname = "protected_fifos",
1891 .data = &sysctl_protected_fifos,
1892 .maxlen = sizeof(int),
1893 .mode = 0600,
1894 .proc_handler = proc_dointvec_minmax,
1895 .extra1 = SYSCTL_ZERO,
1896 .extra2 = &two,
1897 },
1898 {
1899 .procname = "protected_regular",
1900 .data = &sysctl_protected_regular,
1901 .maxlen = sizeof(int),
1902 .mode = 0600,
1903 .proc_handler = proc_dointvec_minmax,
1904 .extra1 = SYSCTL_ZERO,
1905 .extra2 = &two,
1906 },
1907 {
1908 .procname = "suid_dumpable",
1909 .data = &suid_dumpable,
1910 .maxlen = sizeof(int),
1911 .mode = 0644,
1912 .proc_handler = proc_dointvec_minmax_coredump,
1913 .extra1 = SYSCTL_ZERO,
1914 .extra2 = &two,
1915 },
1916#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1917 {
1918 .procname = "binfmt_misc",
1919 .mode = 0555,
1920 .child = sysctl_mount_point,
1921 },
1922#endif
1923 {
1924 .procname = "pipe-max-size",
1925 .data = &pipe_max_size,
1926 .maxlen = sizeof(pipe_max_size),
1927 .mode = 0644,
1928 .proc_handler = proc_dopipe_max_size,
1929 },
1930 {
1931 .procname = "pipe-user-pages-hard",
1932 .data = &pipe_user_pages_hard,
1933 .maxlen = sizeof(pipe_user_pages_hard),
1934 .mode = 0644,
1935 .proc_handler = proc_doulongvec_minmax,
1936 },
1937 {
1938 .procname = "pipe-user-pages-soft",
1939 .data = &pipe_user_pages_soft,
1940 .maxlen = sizeof(pipe_user_pages_soft),
1941 .mode = 0644,
1942 .proc_handler = proc_doulongvec_minmax,
1943 },
1944 {
1945 .procname = "mount-max",
1946 .data = &sysctl_mount_max,
1947 .maxlen = sizeof(unsigned int),
1948 .mode = 0644,
1949 .proc_handler = proc_dointvec_minmax,
1950 .extra1 = SYSCTL_ONE,
1951 },
1952 { }
1953};
1954
1955static struct ctl_table debug_table[] = {
1956#ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1957 {
1958 .procname = "exception-trace",
1959 .data = &show_unhandled_signals,
1960 .maxlen = sizeof(int),
1961 .mode = 0644,
1962 .proc_handler = proc_dointvec
1963 },
1964#endif
1965#if defined(CONFIG_OPTPROBES)
1966 {
1967 .procname = "kprobes-optimization",
1968 .data = &sysctl_kprobes_optimization,
1969 .maxlen = sizeof(int),
1970 .mode = 0644,
1971 .proc_handler = proc_kprobes_optimization_handler,
1972 .extra1 = SYSCTL_ZERO,
1973 .extra2 = SYSCTL_ONE,
1974 },
1975#endif
1976 { }
1977};
1978
1979static struct ctl_table dev_table[] = {
1980 { }
1981};
1982
1983int __init sysctl_init(void)
1984{
1985 struct ctl_table_header *hdr;
1986
1987 hdr = register_sysctl_table(sysctl_base_table);
1988 kmemleak_not_leak(hdr);
1989 return 0;
1990}
1991
1992#endif /* CONFIG_SYSCTL */
1993
1994/*
1995 * /proc/sys support
1996 */
1997
1998#ifdef CONFIG_PROC_SYSCTL
1999
2000static int _proc_do_string(char *data, int maxlen, int write,
2001 char __user *buffer,
2002 size_t *lenp, loff_t *ppos)
2003{
2004 size_t len;
2005 char __user *p;
2006 char c;
2007
2008 if (!data || !maxlen || !*lenp) {
2009 *lenp = 0;
2010 return 0;
2011 }
2012
2013 if (write) {
2014 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
2015 /* Only continue writes not past the end of buffer. */
2016 len = strlen(data);
2017 if (len > maxlen - 1)
2018 len = maxlen - 1;
2019
2020 if (*ppos > len)
2021 return 0;
2022 len = *ppos;
2023 } else {
2024 /* Start writing from beginning of buffer. */
2025 len = 0;
2026 }
2027
2028 *ppos += *lenp;
2029 p = buffer;
2030 while ((p - buffer) < *lenp && len < maxlen - 1) {
2031 if (get_user(c, p++))
2032 return -EFAULT;
2033 if (c == 0 || c == '\n')
2034 break;
2035 data[len++] = c;
2036 }
2037 data[len] = 0;
2038 } else {
2039 len = strlen(data);
2040 if (len > maxlen)
2041 len = maxlen;
2042
2043 if (*ppos > len) {
2044 *lenp = 0;
2045 return 0;
2046 }
2047
2048 data += *ppos;
2049 len -= *ppos;
2050
2051 if (len > *lenp)
2052 len = *lenp;
2053 if (len)
2054 if (copy_to_user(buffer, data, len))
2055 return -EFAULT;
2056 if (len < *lenp) {
2057 if (put_user('\n', buffer + len))
2058 return -EFAULT;
2059 len++;
2060 }
2061 *lenp = len;
2062 *ppos += len;
2063 }
2064 return 0;
2065}
2066
2067static void warn_sysctl_write(struct ctl_table *table)
2068{
2069 pr_warn_once("%s wrote to %s when file position was not 0!\n"
2070 "This will not be supported in the future. To silence this\n"
2071 "warning, set kernel.sysctl_writes_strict = -1\n",
2072 current->comm, table->procname);
2073}
2074
2075/**
2076 * proc_first_pos_non_zero_ignore - check if first position is allowed
2077 * @ppos: file position
2078 * @table: the sysctl table
2079 *
2080 * Returns true if the first position is non-zero and the sysctl_writes_strict
2081 * mode indicates this is not allowed for numeric input types. String proc
2082 * handlers can ignore the return value.
2083 */
2084static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
2085 struct ctl_table *table)
2086{
2087 if (!*ppos)
2088 return false;
2089
2090 switch (sysctl_writes_strict) {
2091 case SYSCTL_WRITES_STRICT:
2092 return true;
2093 case SYSCTL_WRITES_WARN:
2094 warn_sysctl_write(table);
2095 return false;
2096 default:
2097 return false;
2098 }
2099}
2100
2101/**
2102 * proc_dostring - read a string sysctl
2103 * @table: the sysctl table
2104 * @write: %TRUE if this is a write to the sysctl file
2105 * @buffer: the user buffer
2106 * @lenp: the size of the user buffer
2107 * @ppos: file position
2108 *
2109 * Reads/writes a string from/to the user buffer. If the kernel
2110 * buffer provided is not large enough to hold the string, the
2111 * string is truncated. The copied string is %NULL-terminated.
2112 * If the string is being read by the user process, it is copied
2113 * and a newline '\n' is added. It is truncated if the buffer is
2114 * not large enough.
2115 *
2116 * Returns 0 on success.
2117 */
2118int proc_dostring(struct ctl_table *table, int write,
2119 void __user *buffer, size_t *lenp, loff_t *ppos)
2120{
2121 if (write)
2122 proc_first_pos_non_zero_ignore(ppos, table);
2123
2124 return _proc_do_string((char *)(table->data), table->maxlen, write,
2125 (char __user *)buffer, lenp, ppos);
2126}
2127
2128static size_t proc_skip_spaces(char **buf)
2129{
2130 size_t ret;
2131 char *tmp = skip_spaces(*buf);
2132 ret = tmp - *buf;
2133 *buf = tmp;
2134 return ret;
2135}
2136
2137static void proc_skip_char(char **buf, size_t *size, const char v)
2138{
2139 while (*size) {
2140 if (**buf != v)
2141 break;
2142 (*size)--;
2143 (*buf)++;
2144 }
2145}
2146
2147/**
2148 * strtoul_lenient - parse an ASCII formatted integer from a buffer and only
2149 * fail on overflow
2150 *
2151 * @cp: kernel buffer containing the string to parse
2152 * @endp: pointer to store the trailing characters
2153 * @base: the base to use
2154 * @res: where the parsed integer will be stored
2155 *
2156 * In case of success 0 is returned and @res will contain the parsed integer,
2157 * @endp will hold any trailing characters.
2158 * This function will fail the parse on overflow. If there wasn't an overflow
2159 * the function will defer the decision what characters count as invalid to the
2160 * caller.
2161 */
2162static int strtoul_lenient(const char *cp, char **endp, unsigned int base,
2163 unsigned long *res)
2164{
2165 unsigned long long result;
2166 unsigned int rv;
2167
2168 cp = _parse_integer_fixup_radix(cp, &base);
2169 rv = _parse_integer(cp, base, &result);
2170 if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result))
2171 return -ERANGE;
2172
2173 cp += rv;
2174
2175 if (endp)
2176 *endp = (char *)cp;
2177
2178 *res = (unsigned long)result;
2179 return 0;
2180}
2181
2182#define TMPBUFLEN 22
2183/**
2184 * proc_get_long - reads an ASCII formatted integer from a user buffer
2185 *
2186 * @buf: a kernel buffer
2187 * @size: size of the kernel buffer
2188 * @val: this is where the number will be stored
2189 * @neg: set to %TRUE if number is negative
2190 * @perm_tr: a vector which contains the allowed trailers
2191 * @perm_tr_len: size of the perm_tr vector
2192 * @tr: pointer to store the trailer character
2193 *
2194 * In case of success %0 is returned and @buf and @size are updated with
2195 * the amount of bytes read. If @tr is non-NULL and a trailing
2196 * character exists (size is non-zero after returning from this
2197 * function), @tr is updated with the trailing character.
2198 */
2199static int proc_get_long(char **buf, size_t *size,
2200 unsigned long *val, bool *neg,
2201 const char *perm_tr, unsigned perm_tr_len, char *tr)
2202{
2203 int len;
2204 char *p, tmp[TMPBUFLEN];
2205
2206 if (!*size)
2207 return -EINVAL;
2208
2209 len = *size;
2210 if (len > TMPBUFLEN - 1)
2211 len = TMPBUFLEN - 1;
2212
2213 memcpy(tmp, *buf, len);
2214
2215 tmp[len] = 0;
2216 p = tmp;
2217 if (*p == '-' && *size > 1) {
2218 *neg = true;
2219 p++;
2220 } else
2221 *neg = false;
2222 if (!isdigit(*p))
2223 return -EINVAL;
2224
2225 if (strtoul_lenient(p, &p, 0, val))
2226 return -EINVAL;
2227
2228 len = p - tmp;
2229
2230 /* We don't know if the next char is whitespace thus we may accept
2231 * invalid integers (e.g. 1234...a) or two integers instead of one
2232 * (e.g. 123...1). So lets not allow such large numbers. */
2233 if (len == TMPBUFLEN - 1)
2234 return -EINVAL;
2235
2236 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2237 return -EINVAL;
2238
2239 if (tr && (len < *size))
2240 *tr = *p;
2241
2242 *buf += len;
2243 *size -= len;
2244
2245 return 0;
2246}
2247
2248/**
2249 * proc_put_long - converts an integer to a decimal ASCII formatted string
2250 *
2251 * @buf: the user buffer
2252 * @size: the size of the user buffer
2253 * @val: the integer to be converted
2254 * @neg: sign of the number, %TRUE for negative
2255 *
2256 * In case of success %0 is returned and @buf and @size are updated with
2257 * the amount of bytes written.
2258 */
2259static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2260 bool neg)
2261{
2262 int len;
2263 char tmp[TMPBUFLEN], *p = tmp;
2264
2265 sprintf(p, "%s%lu", neg ? "-" : "", val);
2266 len = strlen(tmp);
2267 if (len > *size)
2268 len = *size;
2269 if (copy_to_user(*buf, tmp, len))
2270 return -EFAULT;
2271 *size -= len;
2272 *buf += len;
2273 return 0;
2274}
2275#undef TMPBUFLEN
2276
2277static int proc_put_char(void __user **buf, size_t *size, char c)
2278{
2279 if (*size) {
2280 char __user **buffer = (char __user **)buf;
2281 if (put_user(c, *buffer))
2282 return -EFAULT;
2283 (*size)--, (*buffer)++;
2284 *buf = *buffer;
2285 }
2286 return 0;
2287}
2288
2289static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2290 int *valp,
2291 int write, void *data)
2292{
2293 if (write) {
2294 if (*negp) {
2295 if (*lvalp > (unsigned long) INT_MAX + 1)
2296 return -EINVAL;
2297 *valp = -*lvalp;
2298 } else {
2299 if (*lvalp > (unsigned long) INT_MAX)
2300 return -EINVAL;
2301 *valp = *lvalp;
2302 }
2303 } else {
2304 int val = *valp;
2305 if (val < 0) {
2306 *negp = true;
2307 *lvalp = -(unsigned long)val;
2308 } else {
2309 *negp = false;
2310 *lvalp = (unsigned long)val;
2311 }
2312 }
2313 return 0;
2314}
2315
2316static int do_proc_douintvec_conv(unsigned long *lvalp,
2317 unsigned int *valp,
2318 int write, void *data)
2319{
2320 if (write) {
2321 if (*lvalp > UINT_MAX)
2322 return -EINVAL;
2323 *valp = *lvalp;
2324 } else {
2325 unsigned int val = *valp;
2326 *lvalp = (unsigned long)val;
2327 }
2328 return 0;
2329}
2330
2331static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2332
2333static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2334 int write, void __user *buffer,
2335 size_t *lenp, loff_t *ppos,
2336 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2337 int write, void *data),
2338 void *data)
2339{
2340 int *i, vleft, first = 1, err = 0;
2341 size_t left;
2342 char *kbuf = NULL, *p;
2343
2344 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2345 *lenp = 0;
2346 return 0;
2347 }
2348
2349 i = (int *) tbl_data;
2350 vleft = table->maxlen / sizeof(*i);
2351 left = *lenp;
2352
2353 if (!conv)
2354 conv = do_proc_dointvec_conv;
2355
2356 if (write) {
2357 if (proc_first_pos_non_zero_ignore(ppos, table))
2358 goto out;
2359
2360 if (left > PAGE_SIZE - 1)
2361 left = PAGE_SIZE - 1;
2362 p = kbuf = memdup_user_nul(buffer, left);
2363 if (IS_ERR(kbuf))
2364 return PTR_ERR(kbuf);
2365 }
2366
2367 for (; left && vleft--; i++, first=0) {
2368 unsigned long lval;
2369 bool neg;
2370
2371 if (write) {
2372 left -= proc_skip_spaces(&p);
2373
2374 if (!left)
2375 break;
2376 err = proc_get_long(&p, &left, &lval, &neg,
2377 proc_wspace_sep,
2378 sizeof(proc_wspace_sep), NULL);
2379 if (err)
2380 break;
2381 if (conv(&neg, &lval, i, 1, data)) {
2382 err = -EINVAL;
2383 break;
2384 }
2385 } else {
2386 if (conv(&neg, &lval, i, 0, data)) {
2387 err = -EINVAL;
2388 break;
2389 }
2390 if (!first)
2391 err = proc_put_char(&buffer, &left, '\t');
2392 if (err)
2393 break;
2394 err = proc_put_long(&buffer, &left, lval, neg);
2395 if (err)
2396 break;
2397 }
2398 }
2399
2400 if (!write && !first && left && !err)
2401 err = proc_put_char(&buffer, &left, '\n');
2402 if (write && !err && left)
2403 left -= proc_skip_spaces(&p);
2404 if (write) {
2405 kfree(kbuf);
2406 if (first)
2407 return err ? : -EINVAL;
2408 }
2409 *lenp -= left;
2410out:
2411 *ppos += *lenp;
2412 return err;
2413}
2414
2415static int do_proc_dointvec(struct ctl_table *table, int write,
2416 void __user *buffer, size_t *lenp, loff_t *ppos,
2417 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2418 int write, void *data),
2419 void *data)
2420{
2421 return __do_proc_dointvec(table->data, table, write,
2422 buffer, lenp, ppos, conv, data);
2423}
2424
2425static int do_proc_douintvec_w(unsigned int *tbl_data,
2426 struct ctl_table *table,
2427 void __user *buffer,
2428 size_t *lenp, loff_t *ppos,
2429 int (*conv)(unsigned long *lvalp,
2430 unsigned int *valp,
2431 int write, void *data),
2432 void *data)
2433{
2434 unsigned long lval;
2435 int err = 0;
2436 size_t left;
2437 bool neg;
2438 char *kbuf = NULL, *p;
2439
2440 left = *lenp;
2441
2442 if (proc_first_pos_non_zero_ignore(ppos, table))
2443 goto bail_early;
2444
2445 if (left > PAGE_SIZE - 1)
2446 left = PAGE_SIZE - 1;
2447
2448 p = kbuf = memdup_user_nul(buffer, left);
2449 if (IS_ERR(kbuf))
2450 return -EINVAL;
2451
2452 left -= proc_skip_spaces(&p);
2453 if (!left) {
2454 err = -EINVAL;
2455 goto out_free;
2456 }
2457
2458 err = proc_get_long(&p, &left, &lval, &neg,
2459 proc_wspace_sep,
2460 sizeof(proc_wspace_sep), NULL);
2461 if (err || neg) {
2462 err = -EINVAL;
2463 goto out_free;
2464 }
2465
2466 if (conv(&lval, tbl_data, 1, data)) {
2467 err = -EINVAL;
2468 goto out_free;
2469 }
2470
2471 if (!err && left)
2472 left -= proc_skip_spaces(&p);
2473
2474out_free:
2475 kfree(kbuf);
2476 if (err)
2477 return -EINVAL;
2478
2479 return 0;
2480
2481 /* This is in keeping with old __do_proc_dointvec() */
2482bail_early:
2483 *ppos += *lenp;
2484 return err;
2485}
2486
2487static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2488 size_t *lenp, loff_t *ppos,
2489 int (*conv)(unsigned long *lvalp,
2490 unsigned int *valp,
2491 int write, void *data),
2492 void *data)
2493{
2494 unsigned long lval;
2495 int err = 0;
2496 size_t left;
2497
2498 left = *lenp;
2499
2500 if (conv(&lval, tbl_data, 0, data)) {
2501 err = -EINVAL;
2502 goto out;
2503 }
2504
2505 err = proc_put_long(&buffer, &left, lval, false);
2506 if (err || !left)
2507 goto out;
2508
2509 err = proc_put_char(&buffer, &left, '\n');
2510
2511out:
2512 *lenp -= left;
2513 *ppos += *lenp;
2514
2515 return err;
2516}
2517
2518static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2519 int write, void __user *buffer,
2520 size_t *lenp, loff_t *ppos,
2521 int (*conv)(unsigned long *lvalp,
2522 unsigned int *valp,
2523 int write, void *data),
2524 void *data)
2525{
2526 unsigned int *i, vleft;
2527
2528 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2529 *lenp = 0;
2530 return 0;
2531 }
2532
2533 i = (unsigned int *) tbl_data;
2534 vleft = table->maxlen / sizeof(*i);
2535
2536 /*
2537 * Arrays are not supported, keep this simple. *Do not* add
2538 * support for them.
2539 */
2540 if (vleft != 1) {
2541 *lenp = 0;
2542 return -EINVAL;
2543 }
2544
2545 if (!conv)
2546 conv = do_proc_douintvec_conv;
2547
2548 if (write)
2549 return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2550 conv, data);
2551 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2552}
2553
2554static int do_proc_douintvec(struct ctl_table *table, int write,
2555 void __user *buffer, size_t *lenp, loff_t *ppos,
2556 int (*conv)(unsigned long *lvalp,
2557 unsigned int *valp,
2558 int write, void *data),
2559 void *data)
2560{
2561 return __do_proc_douintvec(table->data, table, write,
2562 buffer, lenp, ppos, conv, data);
2563}
2564
2565/**
2566 * proc_dointvec - read a vector of integers
2567 * @table: the sysctl table
2568 * @write: %TRUE if this is a write to the sysctl file
2569 * @buffer: the user buffer
2570 * @lenp: the size of the user buffer
2571 * @ppos: file position
2572 *
2573 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2574 * values from/to the user buffer, treated as an ASCII string.
2575 *
2576 * Returns 0 on success.
2577 */
2578int proc_dointvec(struct ctl_table *table, int write,
2579 void __user *buffer, size_t *lenp, loff_t *ppos)
2580{
2581 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2582}
2583
2584/**
2585 * proc_douintvec - read a vector of unsigned integers
2586 * @table: the sysctl table
2587 * @write: %TRUE if this is a write to the sysctl file
2588 * @buffer: the user buffer
2589 * @lenp: the size of the user buffer
2590 * @ppos: file position
2591 *
2592 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2593 * values from/to the user buffer, treated as an ASCII string.
2594 *
2595 * Returns 0 on success.
2596 */
2597int proc_douintvec(struct ctl_table *table, int write,
2598 void __user *buffer, size_t *lenp, loff_t *ppos)
2599{
2600 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2601 do_proc_douintvec_conv, NULL);
2602}
2603
2604/*
2605 * Taint values can only be increased
2606 * This means we can safely use a temporary.
2607 */
2608static int proc_taint(struct ctl_table *table, int write,
2609 void __user *buffer, size_t *lenp, loff_t *ppos)
2610{
2611 struct ctl_table t;
2612 unsigned long tmptaint = get_taint();
2613 int err;
2614
2615 if (write && !capable(CAP_SYS_ADMIN))
2616 return -EPERM;
2617
2618 t = *table;
2619 t.data = &tmptaint;
2620 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2621 if (err < 0)
2622 return err;
2623
2624 if (write) {
2625 /*
2626 * Poor man's atomic or. Not worth adding a primitive
2627 * to everyone's atomic.h for this
2628 */
2629 int i;
2630 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2631 if ((tmptaint >> i) & 1)
2632 add_taint(i, LOCKDEP_STILL_OK);
2633 }
2634 }
2635
2636 return err;
2637}
2638
2639#ifdef CONFIG_PRINTK
2640static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2641 void __user *buffer, size_t *lenp, loff_t *ppos)
2642{
2643 if (write && !capable(CAP_SYS_ADMIN))
2644 return -EPERM;
2645
2646 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2647}
2648#endif
2649
2650/**
2651 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
2652 * @min: pointer to minimum allowable value
2653 * @max: pointer to maximum allowable value
2654 *
2655 * The do_proc_dointvec_minmax_conv_param structure provides the
2656 * minimum and maximum values for doing range checking for those sysctl
2657 * parameters that use the proc_dointvec_minmax() handler.
2658 */
2659struct do_proc_dointvec_minmax_conv_param {
2660 int *min;
2661 int *max;
2662};
2663
2664static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2665 int *valp,
2666 int write, void *data)
2667{
2668 int tmp, ret;
2669 struct do_proc_dointvec_minmax_conv_param *param = data;
2670 /*
2671 * If writing, first do so via a temporary local int so we can
2672 * bounds-check it before touching *valp.
2673 */
2674 int *ip = write ? &tmp : valp;
2675
2676 ret = do_proc_dointvec_conv(negp, lvalp, ip, write, data);
2677 if (ret)
2678 return ret;
2679
2680 if (write) {
2681 if ((param->min && *param->min > tmp) ||
2682 (param->max && *param->max < tmp))
2683 return -EINVAL;
2684 *valp = tmp;
2685 }
2686
2687 return 0;
2688}
2689
2690/**
2691 * proc_dointvec_minmax - read a vector of integers with min/max values
2692 * @table: the sysctl table
2693 * @write: %TRUE if this is a write to the sysctl file
2694 * @buffer: the user buffer
2695 * @lenp: the size of the user buffer
2696 * @ppos: file position
2697 *
2698 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2699 * values from/to the user buffer, treated as an ASCII string.
2700 *
2701 * This routine will ensure the values are within the range specified by
2702 * table->extra1 (min) and table->extra2 (max).
2703 *
2704 * Returns 0 on success or -EINVAL on write when the range check fails.
2705 */
2706int proc_dointvec_minmax(struct ctl_table *table, int write,
2707 void __user *buffer, size_t *lenp, loff_t *ppos)
2708{
2709 struct do_proc_dointvec_minmax_conv_param param = {
2710 .min = (int *) table->extra1,
2711 .max = (int *) table->extra2,
2712 };
2713 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2714 do_proc_dointvec_minmax_conv, ¶m);
2715}
2716
2717/**
2718 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
2719 * @min: pointer to minimum allowable value
2720 * @max: pointer to maximum allowable value
2721 *
2722 * The do_proc_douintvec_minmax_conv_param structure provides the
2723 * minimum and maximum values for doing range checking for those sysctl
2724 * parameters that use the proc_douintvec_minmax() handler.
2725 */
2726struct do_proc_douintvec_minmax_conv_param {
2727 unsigned int *min;
2728 unsigned int *max;
2729};
2730
2731static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2732 unsigned int *valp,
2733 int write, void *data)
2734{
2735 int ret;
2736 unsigned int tmp;
2737 struct do_proc_douintvec_minmax_conv_param *param = data;
2738 /* write via temporary local uint for bounds-checking */
2739 unsigned int *up = write ? &tmp : valp;
2740
2741 ret = do_proc_douintvec_conv(lvalp, up, write, data);
2742 if (ret)
2743 return ret;
2744
2745 if (write) {
2746 if ((param->min && *param->min > tmp) ||
2747 (param->max && *param->max < tmp))
2748 return -ERANGE;
2749
2750 *valp = tmp;
2751 }
2752
2753 return 0;
2754}
2755
2756/**
2757 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2758 * @table: the sysctl table
2759 * @write: %TRUE if this is a write to the sysctl file
2760 * @buffer: the user buffer
2761 * @lenp: the size of the user buffer
2762 * @ppos: file position
2763 *
2764 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2765 * values from/to the user buffer, treated as an ASCII string. Negative
2766 * strings are not allowed.
2767 *
2768 * This routine will ensure the values are within the range specified by
2769 * table->extra1 (min) and table->extra2 (max). There is a final sanity
2770 * check for UINT_MAX to avoid having to support wrap around uses from
2771 * userspace.
2772 *
2773 * Returns 0 on success or -ERANGE on write when the range check fails.
2774 */
2775int proc_douintvec_minmax(struct ctl_table *table, int write,
2776 void __user *buffer, size_t *lenp, loff_t *ppos)
2777{
2778 struct do_proc_douintvec_minmax_conv_param param = {
2779 .min = (unsigned int *) table->extra1,
2780 .max = (unsigned int *) table->extra2,
2781 };
2782 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2783 do_proc_douintvec_minmax_conv, ¶m);
2784}
2785
2786static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
2787 unsigned int *valp,
2788 int write, void *data)
2789{
2790 if (write) {
2791 unsigned int val;
2792
2793 val = round_pipe_size(*lvalp);
2794 if (val == 0)
2795 return -EINVAL;
2796
2797 *valp = val;
2798 } else {
2799 unsigned int val = *valp;
2800 *lvalp = (unsigned long) val;
2801 }
2802
2803 return 0;
2804}
2805
2806static int proc_dopipe_max_size(struct ctl_table *table, int write,
2807 void __user *buffer, size_t *lenp, loff_t *ppos)
2808{
2809 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2810 do_proc_dopipe_max_size_conv, NULL);
2811}
2812
2813static void validate_coredump_safety(void)
2814{
2815#ifdef CONFIG_COREDUMP
2816 if (suid_dumpable == SUID_DUMP_ROOT &&
2817 core_pattern[0] != '/' && core_pattern[0] != '|') {
2818 printk(KERN_WARNING
2819"Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2820"Pipe handler or fully qualified core dump path required.\n"
2821"Set kernel.core_pattern before fs.suid_dumpable.\n"
2822 );
2823 }
2824#endif
2825}
2826
2827static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2828 void __user *buffer, size_t *lenp, loff_t *ppos)
2829{
2830 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2831 if (!error)
2832 validate_coredump_safety();
2833 return error;
2834}
2835
2836#ifdef CONFIG_COREDUMP
2837static int proc_dostring_coredump(struct ctl_table *table, int write,
2838 void __user *buffer, size_t *lenp, loff_t *ppos)
2839{
2840 int error = proc_dostring(table, write, buffer, lenp, ppos);
2841 if (!error)
2842 validate_coredump_safety();
2843 return error;
2844}
2845#endif
2846
2847static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2848 void __user *buffer,
2849 size_t *lenp, loff_t *ppos,
2850 unsigned long convmul,
2851 unsigned long convdiv)
2852{
2853 unsigned long *i, *min, *max;
2854 int vleft, first = 1, err = 0;
2855 size_t left;
2856 char *kbuf = NULL, *p;
2857
2858 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2859 *lenp = 0;
2860 return 0;
2861 }
2862
2863 i = (unsigned long *) data;
2864 min = (unsigned long *) table->extra1;
2865 max = (unsigned long *) table->extra2;
2866 vleft = table->maxlen / sizeof(unsigned long);
2867 left = *lenp;
2868
2869 if (write) {
2870 if (proc_first_pos_non_zero_ignore(ppos, table))
2871 goto out;
2872
2873 if (left > PAGE_SIZE - 1)
2874 left = PAGE_SIZE - 1;
2875 p = kbuf = memdup_user_nul(buffer, left);
2876 if (IS_ERR(kbuf))
2877 return PTR_ERR(kbuf);
2878 }
2879
2880 for (; left && vleft--; i++, first = 0) {
2881 unsigned long val;
2882
2883 if (write) {
2884 bool neg;
2885
2886 left -= proc_skip_spaces(&p);
2887 if (!left)
2888 break;
2889
2890 err = proc_get_long(&p, &left, &val, &neg,
2891 proc_wspace_sep,
2892 sizeof(proc_wspace_sep), NULL);
2893 if (err)
2894 break;
2895 if (neg)
2896 continue;
2897 val = convmul * val / convdiv;
2898 if ((min && val < *min) || (max && val > *max)) {
2899 err = -EINVAL;
2900 break;
2901 }
2902 *i = val;
2903 } else {
2904 val = convdiv * (*i) / convmul;
2905 if (!first) {
2906 err = proc_put_char(&buffer, &left, '\t');
2907 if (err)
2908 break;
2909 }
2910 err = proc_put_long(&buffer, &left, val, false);
2911 if (err)
2912 break;
2913 }
2914 }
2915
2916 if (!write && !first && left && !err)
2917 err = proc_put_char(&buffer, &left, '\n');
2918 if (write && !err)
2919 left -= proc_skip_spaces(&p);
2920 if (write) {
2921 kfree(kbuf);
2922 if (first)
2923 return err ? : -EINVAL;
2924 }
2925 *lenp -= left;
2926out:
2927 *ppos += *lenp;
2928 return err;
2929}
2930
2931static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2932 void __user *buffer,
2933 size_t *lenp, loff_t *ppos,
2934 unsigned long convmul,
2935 unsigned long convdiv)
2936{
2937 return __do_proc_doulongvec_minmax(table->data, table, write,
2938 buffer, lenp, ppos, convmul, convdiv);
2939}
2940
2941/**
2942 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2943 * @table: the sysctl table
2944 * @write: %TRUE if this is a write to the sysctl file
2945 * @buffer: the user buffer
2946 * @lenp: the size of the user buffer
2947 * @ppos: file position
2948 *
2949 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2950 * values from/to the user buffer, treated as an ASCII string.
2951 *
2952 * This routine will ensure the values are within the range specified by
2953 * table->extra1 (min) and table->extra2 (max).
2954 *
2955 * Returns 0 on success.
2956 */
2957int proc_doulongvec_minmax(struct ctl_table *table, int write,
2958 void __user *buffer, size_t *lenp, loff_t *ppos)
2959{
2960 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2961}
2962
2963/**
2964 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2965 * @table: the sysctl table
2966 * @write: %TRUE if this is a write to the sysctl file
2967 * @buffer: the user buffer
2968 * @lenp: the size of the user buffer
2969 * @ppos: file position
2970 *
2971 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2972 * values from/to the user buffer, treated as an ASCII string. The values
2973 * are treated as milliseconds, and converted to jiffies when they are stored.
2974 *
2975 * This routine will ensure the values are within the range specified by
2976 * table->extra1 (min) and table->extra2 (max).
2977 *
2978 * Returns 0 on success.
2979 */
2980int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2981 void __user *buffer,
2982 size_t *lenp, loff_t *ppos)
2983{
2984 return do_proc_doulongvec_minmax(table, write, buffer,
2985 lenp, ppos, HZ, 1000l);
2986}
2987
2988
2989static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2990 int *valp,
2991 int write, void *data)
2992{
2993 if (write) {
2994 if (*lvalp > INT_MAX / HZ)
2995 return 1;
2996 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2997 } else {
2998 int val = *valp;
2999 unsigned long lval;
3000 if (val < 0) {
3001 *negp = true;
3002 lval = -(unsigned long)val;
3003 } else {
3004 *negp = false;
3005 lval = (unsigned long)val;
3006 }
3007 *lvalp = lval / HZ;
3008 }
3009 return 0;
3010}
3011
3012static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
3013 int *valp,
3014 int write, void *data)
3015{
3016 if (write) {
3017 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
3018 return 1;
3019 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
3020 } else {
3021 int val = *valp;
3022 unsigned long lval;
3023 if (val < 0) {
3024 *negp = true;
3025 lval = -(unsigned long)val;
3026 } else {
3027 *negp = false;
3028 lval = (unsigned long)val;
3029 }
3030 *lvalp = jiffies_to_clock_t(lval);
3031 }
3032 return 0;
3033}
3034
3035static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
3036 int *valp,
3037 int write, void *data)
3038{
3039 if (write) {
3040 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
3041
3042 if (jif > INT_MAX)
3043 return 1;
3044 *valp = (int)jif;
3045 } else {
3046 int val = *valp;
3047 unsigned long lval;
3048 if (val < 0) {
3049 *negp = true;
3050 lval = -(unsigned long)val;
3051 } else {
3052 *negp = false;
3053 lval = (unsigned long)val;
3054 }
3055 *lvalp = jiffies_to_msecs(lval);
3056 }
3057 return 0;
3058}
3059
3060/**
3061 * proc_dointvec_jiffies - read a vector of integers as seconds
3062 * @table: the sysctl table
3063 * @write: %TRUE if this is a write to the sysctl file
3064 * @buffer: the user buffer
3065 * @lenp: the size of the user buffer
3066 * @ppos: file position
3067 *
3068 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3069 * values from/to the user buffer, treated as an ASCII string.
3070 * The values read are assumed to be in seconds, and are converted into
3071 * jiffies.
3072 *
3073 * Returns 0 on success.
3074 */
3075int proc_dointvec_jiffies(struct ctl_table *table, int write,
3076 void __user *buffer, size_t *lenp, loff_t *ppos)
3077{
3078 return do_proc_dointvec(table,write,buffer,lenp,ppos,
3079 do_proc_dointvec_jiffies_conv,NULL);
3080}
3081
3082/**
3083 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
3084 * @table: the sysctl table
3085 * @write: %TRUE if this is a write to the sysctl file
3086 * @buffer: the user buffer
3087 * @lenp: the size of the user buffer
3088 * @ppos: pointer to the file position
3089 *
3090 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3091 * values from/to the user buffer, treated as an ASCII string.
3092 * The values read are assumed to be in 1/USER_HZ seconds, and
3093 * are converted into jiffies.
3094 *
3095 * Returns 0 on success.
3096 */
3097int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3098 void __user *buffer, size_t *lenp, loff_t *ppos)
3099{
3100 return do_proc_dointvec(table,write,buffer,lenp,ppos,
3101 do_proc_dointvec_userhz_jiffies_conv,NULL);
3102}
3103
3104/**
3105 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
3106 * @table: the sysctl table
3107 * @write: %TRUE if this is a write to the sysctl file
3108 * @buffer: the user buffer
3109 * @lenp: the size of the user buffer
3110 * @ppos: file position
3111 * @ppos: the current position in the file
3112 *
3113 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3114 * values from/to the user buffer, treated as an ASCII string.
3115 * The values read are assumed to be in 1/1000 seconds, and
3116 * are converted into jiffies.
3117 *
3118 * Returns 0 on success.
3119 */
3120int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3121 void __user *buffer, size_t *lenp, loff_t *ppos)
3122{
3123 return do_proc_dointvec(table, write, buffer, lenp, ppos,
3124 do_proc_dointvec_ms_jiffies_conv, NULL);
3125}
3126
3127static int proc_do_cad_pid(struct ctl_table *table, int write,
3128 void __user *buffer, size_t *lenp, loff_t *ppos)
3129{
3130 struct pid *new_pid;
3131 pid_t tmp;
3132 int r;
3133
3134 tmp = pid_vnr(cad_pid);
3135
3136 r = __do_proc_dointvec(&tmp, table, write, buffer,
3137 lenp, ppos, NULL, NULL);
3138 if (r || !write)
3139 return r;
3140
3141 new_pid = find_get_pid(tmp);
3142 if (!new_pid)
3143 return -ESRCH;
3144
3145 put_pid(xchg(&cad_pid, new_pid));
3146 return 0;
3147}
3148
3149/**
3150 * proc_do_large_bitmap - read/write from/to a large bitmap
3151 * @table: the sysctl table
3152 * @write: %TRUE if this is a write to the sysctl file
3153 * @buffer: the user buffer
3154 * @lenp: the size of the user buffer
3155 * @ppos: file position
3156 *
3157 * The bitmap is stored at table->data and the bitmap length (in bits)
3158 * in table->maxlen.
3159 *
3160 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
3161 * large bitmaps may be represented in a compact manner. Writing into
3162 * the file will clear the bitmap then update it with the given input.
3163 *
3164 * Returns 0 on success.
3165 */
3166int proc_do_large_bitmap(struct ctl_table *table, int write,
3167 void __user *buffer, size_t *lenp, loff_t *ppos)
3168{
3169 int err = 0;
3170 bool first = 1;
3171 size_t left = *lenp;
3172 unsigned long bitmap_len = table->maxlen;
3173 unsigned long *bitmap = *(unsigned long **) table->data;
3174 unsigned long *tmp_bitmap = NULL;
3175 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
3176
3177 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
3178 *lenp = 0;
3179 return 0;
3180 }
3181
3182 if (write) {
3183 char *kbuf, *p;
3184 size_t skipped = 0;
3185
3186 if (left > PAGE_SIZE - 1) {
3187 left = PAGE_SIZE - 1;
3188 /* How much of the buffer we'll skip this pass */
3189 skipped = *lenp - left;
3190 }
3191
3192 p = kbuf = memdup_user_nul(buffer, left);
3193 if (IS_ERR(kbuf))
3194 return PTR_ERR(kbuf);
3195
3196 tmp_bitmap = bitmap_zalloc(bitmap_len, GFP_KERNEL);
3197 if (!tmp_bitmap) {
3198 kfree(kbuf);
3199 return -ENOMEM;
3200 }
3201 proc_skip_char(&p, &left, '\n');
3202 while (!err && left) {
3203 unsigned long val_a, val_b;
3204 bool neg;
3205 size_t saved_left;
3206
3207 /* In case we stop parsing mid-number, we can reset */
3208 saved_left = left;
3209 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3210 sizeof(tr_a), &c);
3211 /*
3212 * If we consumed the entirety of a truncated buffer or
3213 * only one char is left (may be a "-"), then stop here,
3214 * reset, & come back for more.
3215 */
3216 if ((left <= 1) && skipped) {
3217 left = saved_left;
3218 break;
3219 }
3220
3221 if (err)
3222 break;
3223 if (val_a >= bitmap_len || neg) {
3224 err = -EINVAL;
3225 break;
3226 }
3227
3228 val_b = val_a;
3229 if (left) {
3230 p++;
3231 left--;
3232 }
3233
3234 if (c == '-') {
3235 err = proc_get_long(&p, &left, &val_b,
3236 &neg, tr_b, sizeof(tr_b),
3237 &c);
3238 /*
3239 * If we consumed all of a truncated buffer or
3240 * then stop here, reset, & come back for more.
3241 */
3242 if (!left && skipped) {
3243 left = saved_left;
3244 break;
3245 }
3246
3247 if (err)
3248 break;
3249 if (val_b >= bitmap_len || neg ||
3250 val_a > val_b) {
3251 err = -EINVAL;
3252 break;
3253 }
3254 if (left) {
3255 p++;
3256 left--;
3257 }
3258 }
3259
3260 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3261 first = 0;
3262 proc_skip_char(&p, &left, '\n');
3263 }
3264 kfree(kbuf);
3265 left += skipped;
3266 } else {
3267 unsigned long bit_a, bit_b = 0;
3268
3269 while (left) {
3270 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3271 if (bit_a >= bitmap_len)
3272 break;
3273 bit_b = find_next_zero_bit(bitmap, bitmap_len,
3274 bit_a + 1) - 1;
3275
3276 if (!first) {
3277 err = proc_put_char(&buffer, &left, ',');
3278 if (err)
3279 break;
3280 }
3281 err = proc_put_long(&buffer, &left, bit_a, false);
3282 if (err)
3283 break;
3284 if (bit_a != bit_b) {
3285 err = proc_put_char(&buffer, &left, '-');
3286 if (err)
3287 break;
3288 err = proc_put_long(&buffer, &left, bit_b, false);
3289 if (err)
3290 break;
3291 }
3292
3293 first = 0; bit_b++;
3294 }
3295 if (!err)
3296 err = proc_put_char(&buffer, &left, '\n');
3297 }
3298
3299 if (!err) {
3300 if (write) {
3301 if (*ppos)
3302 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3303 else
3304 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3305 }
3306 *lenp -= left;
3307 *ppos += *lenp;
3308 }
3309
3310 bitmap_free(tmp_bitmap);
3311 return err;
3312}
3313
3314#else /* CONFIG_PROC_SYSCTL */
3315
3316int proc_dostring(struct ctl_table *table, int write,
3317 void __user *buffer, size_t *lenp, loff_t *ppos)
3318{
3319 return -ENOSYS;
3320}
3321
3322int proc_dointvec(struct ctl_table *table, int write,
3323 void __user *buffer, size_t *lenp, loff_t *ppos)
3324{
3325 return -ENOSYS;
3326}
3327
3328int proc_douintvec(struct ctl_table *table, int write,
3329 void __user *buffer, size_t *lenp, loff_t *ppos)
3330{
3331 return -ENOSYS;
3332}
3333
3334int proc_dointvec_minmax(struct ctl_table *table, int write,
3335 void __user *buffer, size_t *lenp, loff_t *ppos)
3336{
3337 return -ENOSYS;
3338}
3339
3340int proc_douintvec_minmax(struct ctl_table *table, int write,
3341 void __user *buffer, size_t *lenp, loff_t *ppos)
3342{
3343 return -ENOSYS;
3344}
3345
3346int proc_dointvec_jiffies(struct ctl_table *table, int write,
3347 void __user *buffer, size_t *lenp, loff_t *ppos)
3348{
3349 return -ENOSYS;
3350}
3351
3352int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3353 void __user *buffer, size_t *lenp, loff_t *ppos)
3354{
3355 return -ENOSYS;
3356}
3357
3358int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3359 void __user *buffer, size_t *lenp, loff_t *ppos)
3360{
3361 return -ENOSYS;
3362}
3363
3364int proc_doulongvec_minmax(struct ctl_table *table, int write,
3365 void __user *buffer, size_t *lenp, loff_t *ppos)
3366{
3367 return -ENOSYS;
3368}
3369
3370int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3371 void __user *buffer,
3372 size_t *lenp, loff_t *ppos)
3373{
3374 return -ENOSYS;
3375}
3376
3377int proc_do_large_bitmap(struct ctl_table *table, int write,
3378 void __user *buffer, size_t *lenp, loff_t *ppos)
3379{
3380 return -ENOSYS;
3381}
3382
3383#endif /* CONFIG_PROC_SYSCTL */
3384
3385#if defined(CONFIG_SYSCTL)
3386int proc_do_static_key(struct ctl_table *table, int write,
3387 void __user *buffer, size_t *lenp,
3388 loff_t *ppos)
3389{
3390 struct static_key *key = (struct static_key *)table->data;
3391 static DEFINE_MUTEX(static_key_mutex);
3392 int val, ret;
3393 struct ctl_table tmp = {
3394 .data = &val,
3395 .maxlen = sizeof(val),
3396 .mode = table->mode,
3397 .extra1 = SYSCTL_ZERO,
3398 .extra2 = SYSCTL_ONE,
3399 };
3400
3401 if (write && !capable(CAP_SYS_ADMIN))
3402 return -EPERM;
3403
3404 mutex_lock(&static_key_mutex);
3405 val = static_key_enabled(key);
3406 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
3407 if (write && !ret) {
3408 if (val)
3409 static_key_enable(key);
3410 else
3411 static_key_disable(key);
3412 }
3413 mutex_unlock(&static_key_mutex);
3414 return ret;
3415}
3416#endif
3417/*
3418 * No sense putting this after each symbol definition, twice,
3419 * exception granted :-)
3420 */
3421EXPORT_SYMBOL(proc_dointvec);
3422EXPORT_SYMBOL(proc_douintvec);
3423EXPORT_SYMBOL(proc_dointvec_jiffies);
3424EXPORT_SYMBOL(proc_dointvec_minmax);
3425EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3426EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3427EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3428EXPORT_SYMBOL(proc_dostring);
3429EXPORT_SYMBOL(proc_doulongvec_minmax);
3430EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3431EXPORT_SYMBOL(proc_do_large_bitmap);
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/mm.h>
23#include <linux/swap.h>
24#include <linux/slab.h>
25#include <linux/sysctl.h>
26#include <linux/signal.h>
27#include <linux/printk.h>
28#include <linux/proc_fs.h>
29#include <linux/security.h>
30#include <linux/ctype.h>
31#include <linux/kmemcheck.h>
32#include <linux/fs.h>
33#include <linux/init.h>
34#include <linux/kernel.h>
35#include <linux/kobject.h>
36#include <linux/net.h>
37#include <linux/sysrq.h>
38#include <linux/highuid.h>
39#include <linux/writeback.h>
40#include <linux/ratelimit.h>
41#include <linux/compaction.h>
42#include <linux/hugetlb.h>
43#include <linux/initrd.h>
44#include <linux/key.h>
45#include <linux/times.h>
46#include <linux/limits.h>
47#include <linux/dcache.h>
48#include <linux/dnotify.h>
49#include <linux/syscalls.h>
50#include <linux/vmstat.h>
51#include <linux/nfs_fs.h>
52#include <linux/acpi.h>
53#include <linux/reboot.h>
54#include <linux/ftrace.h>
55#include <linux/perf_event.h>
56#include <linux/kprobes.h>
57#include <linux/pipe_fs_i.h>
58#include <linux/oom.h>
59#include <linux/kmod.h>
60
61#include <asm/uaccess.h>
62#include <asm/processor.h>
63
64#ifdef CONFIG_X86
65#include <asm/nmi.h>
66#include <asm/stacktrace.h>
67#include <asm/io.h>
68#endif
69#ifdef CONFIG_BSD_PROCESS_ACCT
70#include <linux/acct.h>
71#endif
72#ifdef CONFIG_RT_MUTEXES
73#include <linux/rtmutex.h>
74#endif
75#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
76#include <linux/lockdep.h>
77#endif
78#ifdef CONFIG_CHR_DEV_SG
79#include <scsi/sg.h>
80#endif
81
82#ifdef CONFIG_LOCKUP_DETECTOR
83#include <linux/nmi.h>
84#endif
85
86
87#if defined(CONFIG_SYSCTL)
88
89/* External variables not in a header file. */
90extern int sysctl_overcommit_memory;
91extern int sysctl_overcommit_ratio;
92extern int max_threads;
93extern int core_uses_pid;
94extern int suid_dumpable;
95extern char core_pattern[];
96extern unsigned int core_pipe_limit;
97extern int pid_max;
98extern int min_free_kbytes;
99extern int pid_max_min, pid_max_max;
100extern int sysctl_drop_caches;
101extern int percpu_pagelist_fraction;
102extern int compat_log;
103extern int latencytop_enabled;
104extern int sysctl_nr_open_min, sysctl_nr_open_max;
105#ifndef CONFIG_MMU
106extern int sysctl_nr_trim_pages;
107#endif
108#ifdef CONFIG_BLOCK
109extern int blk_iopoll_enabled;
110#endif
111
112/* Constants used for minimum and maximum */
113#ifdef CONFIG_LOCKUP_DETECTOR
114static int sixty = 60;
115static int neg_one = -1;
116#endif
117
118static int zero;
119static int __maybe_unused one = 1;
120static int __maybe_unused two = 2;
121static int __maybe_unused three = 3;
122static unsigned long one_ul = 1;
123static int one_hundred = 100;
124#ifdef CONFIG_PRINTK
125static int ten_thousand = 10000;
126#endif
127
128/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
129static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
130
131/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
132static int maxolduid = 65535;
133static int minolduid;
134static int min_percpu_pagelist_fract = 8;
135
136static int ngroups_max = NGROUPS_MAX;
137
138#ifdef CONFIG_INOTIFY_USER
139#include <linux/inotify.h>
140#endif
141#ifdef CONFIG_SPARC
142#include <asm/system.h>
143#endif
144
145#ifdef CONFIG_SPARC64
146extern int sysctl_tsb_ratio;
147#endif
148
149#ifdef __hppa__
150extern int pwrsw_enabled;
151extern int unaligned_enabled;
152#endif
153
154#ifdef CONFIG_S390
155#ifdef CONFIG_MATHEMU
156extern int sysctl_ieee_emulation_warnings;
157#endif
158extern int sysctl_userprocess_debug;
159extern int spin_retry;
160#endif
161
162#ifdef CONFIG_IA64
163extern int no_unaligned_warning;
164extern int unaligned_dump_stack;
165#endif
166
167#ifdef CONFIG_PROC_SYSCTL
168static int proc_do_cad_pid(struct ctl_table *table, int write,
169 void __user *buffer, size_t *lenp, loff_t *ppos);
170static int proc_taint(struct ctl_table *table, int write,
171 void __user *buffer, size_t *lenp, loff_t *ppos);
172#endif
173
174#ifdef CONFIG_PRINTK
175static int proc_dmesg_restrict(struct ctl_table *table, int write,
176 void __user *buffer, size_t *lenp, loff_t *ppos);
177#endif
178
179#ifdef CONFIG_MAGIC_SYSRQ
180/* Note: sysrq code uses it's own private copy */
181static int __sysrq_enabled = SYSRQ_DEFAULT_ENABLE;
182
183static int sysrq_sysctl_handler(ctl_table *table, int write,
184 void __user *buffer, size_t *lenp,
185 loff_t *ppos)
186{
187 int error;
188
189 error = proc_dointvec(table, write, buffer, lenp, ppos);
190 if (error)
191 return error;
192
193 if (write)
194 sysrq_toggle_support(__sysrq_enabled);
195
196 return 0;
197}
198
199#endif
200
201static struct ctl_table root_table[];
202static struct ctl_table_root sysctl_table_root;
203static struct ctl_table_header root_table_header = {
204 {{.count = 1,
205 .ctl_table = root_table,
206 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),}},
207 .root = &sysctl_table_root,
208 .set = &sysctl_table_root.default_set,
209};
210static struct ctl_table_root sysctl_table_root = {
211 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
212 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
213};
214
215static struct ctl_table kern_table[];
216static struct ctl_table vm_table[];
217static struct ctl_table fs_table[];
218static struct ctl_table debug_table[];
219static struct ctl_table dev_table[];
220extern struct ctl_table random_table[];
221#ifdef CONFIG_EPOLL
222extern struct ctl_table epoll_table[];
223#endif
224
225#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
226int sysctl_legacy_va_layout;
227#endif
228
229/* The default sysctl tables: */
230
231static struct ctl_table root_table[] = {
232 {
233 .procname = "kernel",
234 .mode = 0555,
235 .child = kern_table,
236 },
237 {
238 .procname = "vm",
239 .mode = 0555,
240 .child = vm_table,
241 },
242 {
243 .procname = "fs",
244 .mode = 0555,
245 .child = fs_table,
246 },
247 {
248 .procname = "debug",
249 .mode = 0555,
250 .child = debug_table,
251 },
252 {
253 .procname = "dev",
254 .mode = 0555,
255 .child = dev_table,
256 },
257 { }
258};
259
260#ifdef CONFIG_SCHED_DEBUG
261static int min_sched_granularity_ns = 100000; /* 100 usecs */
262static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
263static int min_wakeup_granularity_ns; /* 0 usecs */
264static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
265static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
266static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
267#endif
268
269#ifdef CONFIG_COMPACTION
270static int min_extfrag_threshold;
271static int max_extfrag_threshold = 1000;
272#endif
273
274static struct ctl_table kern_table[] = {
275 {
276 .procname = "sched_child_runs_first",
277 .data = &sysctl_sched_child_runs_first,
278 .maxlen = sizeof(unsigned int),
279 .mode = 0644,
280 .proc_handler = proc_dointvec,
281 },
282#ifdef CONFIG_SCHED_DEBUG
283 {
284 .procname = "sched_min_granularity_ns",
285 .data = &sysctl_sched_min_granularity,
286 .maxlen = sizeof(unsigned int),
287 .mode = 0644,
288 .proc_handler = sched_proc_update_handler,
289 .extra1 = &min_sched_granularity_ns,
290 .extra2 = &max_sched_granularity_ns,
291 },
292 {
293 .procname = "sched_latency_ns",
294 .data = &sysctl_sched_latency,
295 .maxlen = sizeof(unsigned int),
296 .mode = 0644,
297 .proc_handler = sched_proc_update_handler,
298 .extra1 = &min_sched_granularity_ns,
299 .extra2 = &max_sched_granularity_ns,
300 },
301 {
302 .procname = "sched_wakeup_granularity_ns",
303 .data = &sysctl_sched_wakeup_granularity,
304 .maxlen = sizeof(unsigned int),
305 .mode = 0644,
306 .proc_handler = sched_proc_update_handler,
307 .extra1 = &min_wakeup_granularity_ns,
308 .extra2 = &max_wakeup_granularity_ns,
309 },
310 {
311 .procname = "sched_tunable_scaling",
312 .data = &sysctl_sched_tunable_scaling,
313 .maxlen = sizeof(enum sched_tunable_scaling),
314 .mode = 0644,
315 .proc_handler = sched_proc_update_handler,
316 .extra1 = &min_sched_tunable_scaling,
317 .extra2 = &max_sched_tunable_scaling,
318 },
319 {
320 .procname = "sched_migration_cost",
321 .data = &sysctl_sched_migration_cost,
322 .maxlen = sizeof(unsigned int),
323 .mode = 0644,
324 .proc_handler = proc_dointvec,
325 },
326 {
327 .procname = "sched_nr_migrate",
328 .data = &sysctl_sched_nr_migrate,
329 .maxlen = sizeof(unsigned int),
330 .mode = 0644,
331 .proc_handler = proc_dointvec,
332 },
333 {
334 .procname = "sched_time_avg",
335 .data = &sysctl_sched_time_avg,
336 .maxlen = sizeof(unsigned int),
337 .mode = 0644,
338 .proc_handler = proc_dointvec,
339 },
340 {
341 .procname = "sched_shares_window",
342 .data = &sysctl_sched_shares_window,
343 .maxlen = sizeof(unsigned int),
344 .mode = 0644,
345 .proc_handler = proc_dointvec,
346 },
347 {
348 .procname = "timer_migration",
349 .data = &sysctl_timer_migration,
350 .maxlen = sizeof(unsigned int),
351 .mode = 0644,
352 .proc_handler = proc_dointvec_minmax,
353 .extra1 = &zero,
354 .extra2 = &one,
355 },
356#endif
357 {
358 .procname = "sched_rt_period_us",
359 .data = &sysctl_sched_rt_period,
360 .maxlen = sizeof(unsigned int),
361 .mode = 0644,
362 .proc_handler = sched_rt_handler,
363 },
364 {
365 .procname = "sched_rt_runtime_us",
366 .data = &sysctl_sched_rt_runtime,
367 .maxlen = sizeof(int),
368 .mode = 0644,
369 .proc_handler = sched_rt_handler,
370 },
371#ifdef CONFIG_SCHED_AUTOGROUP
372 {
373 .procname = "sched_autogroup_enabled",
374 .data = &sysctl_sched_autogroup_enabled,
375 .maxlen = sizeof(unsigned int),
376 .mode = 0644,
377 .proc_handler = proc_dointvec_minmax,
378 .extra1 = &zero,
379 .extra2 = &one,
380 },
381#endif
382#ifdef CONFIG_PROVE_LOCKING
383 {
384 .procname = "prove_locking",
385 .data = &prove_locking,
386 .maxlen = sizeof(int),
387 .mode = 0644,
388 .proc_handler = proc_dointvec,
389 },
390#endif
391#ifdef CONFIG_LOCK_STAT
392 {
393 .procname = "lock_stat",
394 .data = &lock_stat,
395 .maxlen = sizeof(int),
396 .mode = 0644,
397 .proc_handler = proc_dointvec,
398 },
399#endif
400 {
401 .procname = "panic",
402 .data = &panic_timeout,
403 .maxlen = sizeof(int),
404 .mode = 0644,
405 .proc_handler = proc_dointvec,
406 },
407 {
408 .procname = "core_uses_pid",
409 .data = &core_uses_pid,
410 .maxlen = sizeof(int),
411 .mode = 0644,
412 .proc_handler = proc_dointvec,
413 },
414 {
415 .procname = "core_pattern",
416 .data = core_pattern,
417 .maxlen = CORENAME_MAX_SIZE,
418 .mode = 0644,
419 .proc_handler = proc_dostring,
420 },
421 {
422 .procname = "core_pipe_limit",
423 .data = &core_pipe_limit,
424 .maxlen = sizeof(unsigned int),
425 .mode = 0644,
426 .proc_handler = proc_dointvec,
427 },
428#ifdef CONFIG_PROC_SYSCTL
429 {
430 .procname = "tainted",
431 .maxlen = sizeof(long),
432 .mode = 0644,
433 .proc_handler = proc_taint,
434 },
435#endif
436#ifdef CONFIG_LATENCYTOP
437 {
438 .procname = "latencytop",
439 .data = &latencytop_enabled,
440 .maxlen = sizeof(int),
441 .mode = 0644,
442 .proc_handler = proc_dointvec,
443 },
444#endif
445#ifdef CONFIG_BLK_DEV_INITRD
446 {
447 .procname = "real-root-dev",
448 .data = &real_root_dev,
449 .maxlen = sizeof(int),
450 .mode = 0644,
451 .proc_handler = proc_dointvec,
452 },
453#endif
454 {
455 .procname = "print-fatal-signals",
456 .data = &print_fatal_signals,
457 .maxlen = sizeof(int),
458 .mode = 0644,
459 .proc_handler = proc_dointvec,
460 },
461#ifdef CONFIG_SPARC
462 {
463 .procname = "reboot-cmd",
464 .data = reboot_command,
465 .maxlen = 256,
466 .mode = 0644,
467 .proc_handler = proc_dostring,
468 },
469 {
470 .procname = "stop-a",
471 .data = &stop_a_enabled,
472 .maxlen = sizeof (int),
473 .mode = 0644,
474 .proc_handler = proc_dointvec,
475 },
476 {
477 .procname = "scons-poweroff",
478 .data = &scons_pwroff,
479 .maxlen = sizeof (int),
480 .mode = 0644,
481 .proc_handler = proc_dointvec,
482 },
483#endif
484#ifdef CONFIG_SPARC64
485 {
486 .procname = "tsb-ratio",
487 .data = &sysctl_tsb_ratio,
488 .maxlen = sizeof (int),
489 .mode = 0644,
490 .proc_handler = proc_dointvec,
491 },
492#endif
493#ifdef __hppa__
494 {
495 .procname = "soft-power",
496 .data = &pwrsw_enabled,
497 .maxlen = sizeof (int),
498 .mode = 0644,
499 .proc_handler = proc_dointvec,
500 },
501 {
502 .procname = "unaligned-trap",
503 .data = &unaligned_enabled,
504 .maxlen = sizeof (int),
505 .mode = 0644,
506 .proc_handler = proc_dointvec,
507 },
508#endif
509 {
510 .procname = "ctrl-alt-del",
511 .data = &C_A_D,
512 .maxlen = sizeof(int),
513 .mode = 0644,
514 .proc_handler = proc_dointvec,
515 },
516#ifdef CONFIG_FUNCTION_TRACER
517 {
518 .procname = "ftrace_enabled",
519 .data = &ftrace_enabled,
520 .maxlen = sizeof(int),
521 .mode = 0644,
522 .proc_handler = ftrace_enable_sysctl,
523 },
524#endif
525#ifdef CONFIG_STACK_TRACER
526 {
527 .procname = "stack_tracer_enabled",
528 .data = &stack_tracer_enabled,
529 .maxlen = sizeof(int),
530 .mode = 0644,
531 .proc_handler = stack_trace_sysctl,
532 },
533#endif
534#ifdef CONFIG_TRACING
535 {
536 .procname = "ftrace_dump_on_oops",
537 .data = &ftrace_dump_on_oops,
538 .maxlen = sizeof(int),
539 .mode = 0644,
540 .proc_handler = proc_dointvec,
541 },
542#endif
543#ifdef CONFIG_MODULES
544 {
545 .procname = "modprobe",
546 .data = &modprobe_path,
547 .maxlen = KMOD_PATH_LEN,
548 .mode = 0644,
549 .proc_handler = proc_dostring,
550 },
551 {
552 .procname = "modules_disabled",
553 .data = &modules_disabled,
554 .maxlen = sizeof(int),
555 .mode = 0644,
556 /* only handle a transition from default "0" to "1" */
557 .proc_handler = proc_dointvec_minmax,
558 .extra1 = &one,
559 .extra2 = &one,
560 },
561#endif
562#ifdef CONFIG_HOTPLUG
563 {
564 .procname = "hotplug",
565 .data = &uevent_helper,
566 .maxlen = UEVENT_HELPER_PATH_LEN,
567 .mode = 0644,
568 .proc_handler = proc_dostring,
569 },
570#endif
571#ifdef CONFIG_CHR_DEV_SG
572 {
573 .procname = "sg-big-buff",
574 .data = &sg_big_buff,
575 .maxlen = sizeof (int),
576 .mode = 0444,
577 .proc_handler = proc_dointvec,
578 },
579#endif
580#ifdef CONFIG_BSD_PROCESS_ACCT
581 {
582 .procname = "acct",
583 .data = &acct_parm,
584 .maxlen = 3*sizeof(int),
585 .mode = 0644,
586 .proc_handler = proc_dointvec,
587 },
588#endif
589#ifdef CONFIG_MAGIC_SYSRQ
590 {
591 .procname = "sysrq",
592 .data = &__sysrq_enabled,
593 .maxlen = sizeof (int),
594 .mode = 0644,
595 .proc_handler = sysrq_sysctl_handler,
596 },
597#endif
598#ifdef CONFIG_PROC_SYSCTL
599 {
600 .procname = "cad_pid",
601 .data = NULL,
602 .maxlen = sizeof (int),
603 .mode = 0600,
604 .proc_handler = proc_do_cad_pid,
605 },
606#endif
607 {
608 .procname = "threads-max",
609 .data = &max_threads,
610 .maxlen = sizeof(int),
611 .mode = 0644,
612 .proc_handler = proc_dointvec,
613 },
614 {
615 .procname = "random",
616 .mode = 0555,
617 .child = random_table,
618 },
619 {
620 .procname = "usermodehelper",
621 .mode = 0555,
622 .child = usermodehelper_table,
623 },
624 {
625 .procname = "overflowuid",
626 .data = &overflowuid,
627 .maxlen = sizeof(int),
628 .mode = 0644,
629 .proc_handler = proc_dointvec_minmax,
630 .extra1 = &minolduid,
631 .extra2 = &maxolduid,
632 },
633 {
634 .procname = "overflowgid",
635 .data = &overflowgid,
636 .maxlen = sizeof(int),
637 .mode = 0644,
638 .proc_handler = proc_dointvec_minmax,
639 .extra1 = &minolduid,
640 .extra2 = &maxolduid,
641 },
642#ifdef CONFIG_S390
643#ifdef CONFIG_MATHEMU
644 {
645 .procname = "ieee_emulation_warnings",
646 .data = &sysctl_ieee_emulation_warnings,
647 .maxlen = sizeof(int),
648 .mode = 0644,
649 .proc_handler = proc_dointvec,
650 },
651#endif
652 {
653 .procname = "userprocess_debug",
654 .data = &show_unhandled_signals,
655 .maxlen = sizeof(int),
656 .mode = 0644,
657 .proc_handler = proc_dointvec,
658 },
659#endif
660 {
661 .procname = "pid_max",
662 .data = &pid_max,
663 .maxlen = sizeof (int),
664 .mode = 0644,
665 .proc_handler = proc_dointvec_minmax,
666 .extra1 = &pid_max_min,
667 .extra2 = &pid_max_max,
668 },
669 {
670 .procname = "panic_on_oops",
671 .data = &panic_on_oops,
672 .maxlen = sizeof(int),
673 .mode = 0644,
674 .proc_handler = proc_dointvec,
675 },
676#if defined CONFIG_PRINTK
677 {
678 .procname = "printk",
679 .data = &console_loglevel,
680 .maxlen = 4*sizeof(int),
681 .mode = 0644,
682 .proc_handler = proc_dointvec,
683 },
684 {
685 .procname = "printk_ratelimit",
686 .data = &printk_ratelimit_state.interval,
687 .maxlen = sizeof(int),
688 .mode = 0644,
689 .proc_handler = proc_dointvec_jiffies,
690 },
691 {
692 .procname = "printk_ratelimit_burst",
693 .data = &printk_ratelimit_state.burst,
694 .maxlen = sizeof(int),
695 .mode = 0644,
696 .proc_handler = proc_dointvec,
697 },
698 {
699 .procname = "printk_delay",
700 .data = &printk_delay_msec,
701 .maxlen = sizeof(int),
702 .mode = 0644,
703 .proc_handler = proc_dointvec_minmax,
704 .extra1 = &zero,
705 .extra2 = &ten_thousand,
706 },
707 {
708 .procname = "dmesg_restrict",
709 .data = &dmesg_restrict,
710 .maxlen = sizeof(int),
711 .mode = 0644,
712 .proc_handler = proc_dointvec_minmax,
713 .extra1 = &zero,
714 .extra2 = &one,
715 },
716 {
717 .procname = "kptr_restrict",
718 .data = &kptr_restrict,
719 .maxlen = sizeof(int),
720 .mode = 0644,
721 .proc_handler = proc_dmesg_restrict,
722 .extra1 = &zero,
723 .extra2 = &two,
724 },
725#endif
726 {
727 .procname = "ngroups_max",
728 .data = &ngroups_max,
729 .maxlen = sizeof (int),
730 .mode = 0444,
731 .proc_handler = proc_dointvec,
732 },
733#if defined(CONFIG_LOCKUP_DETECTOR)
734 {
735 .procname = "watchdog",
736 .data = &watchdog_enabled,
737 .maxlen = sizeof (int),
738 .mode = 0644,
739 .proc_handler = proc_dowatchdog,
740 .extra1 = &zero,
741 .extra2 = &one,
742 },
743 {
744 .procname = "watchdog_thresh",
745 .data = &watchdog_thresh,
746 .maxlen = sizeof(int),
747 .mode = 0644,
748 .proc_handler = proc_dowatchdog,
749 .extra1 = &neg_one,
750 .extra2 = &sixty,
751 },
752 {
753 .procname = "softlockup_panic",
754 .data = &softlockup_panic,
755 .maxlen = sizeof(int),
756 .mode = 0644,
757 .proc_handler = proc_dointvec_minmax,
758 .extra1 = &zero,
759 .extra2 = &one,
760 },
761 {
762 .procname = "nmi_watchdog",
763 .data = &watchdog_enabled,
764 .maxlen = sizeof (int),
765 .mode = 0644,
766 .proc_handler = proc_dowatchdog,
767 .extra1 = &zero,
768 .extra2 = &one,
769 },
770#endif
771#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
772 {
773 .procname = "unknown_nmi_panic",
774 .data = &unknown_nmi_panic,
775 .maxlen = sizeof (int),
776 .mode = 0644,
777 .proc_handler = proc_dointvec,
778 },
779#endif
780#if defined(CONFIG_X86)
781 {
782 .procname = "panic_on_unrecovered_nmi",
783 .data = &panic_on_unrecovered_nmi,
784 .maxlen = sizeof(int),
785 .mode = 0644,
786 .proc_handler = proc_dointvec,
787 },
788 {
789 .procname = "panic_on_io_nmi",
790 .data = &panic_on_io_nmi,
791 .maxlen = sizeof(int),
792 .mode = 0644,
793 .proc_handler = proc_dointvec,
794 },
795 {
796 .procname = "bootloader_type",
797 .data = &bootloader_type,
798 .maxlen = sizeof (int),
799 .mode = 0444,
800 .proc_handler = proc_dointvec,
801 },
802 {
803 .procname = "bootloader_version",
804 .data = &bootloader_version,
805 .maxlen = sizeof (int),
806 .mode = 0444,
807 .proc_handler = proc_dointvec,
808 },
809 {
810 .procname = "kstack_depth_to_print",
811 .data = &kstack_depth_to_print,
812 .maxlen = sizeof(int),
813 .mode = 0644,
814 .proc_handler = proc_dointvec,
815 },
816 {
817 .procname = "io_delay_type",
818 .data = &io_delay_type,
819 .maxlen = sizeof(int),
820 .mode = 0644,
821 .proc_handler = proc_dointvec,
822 },
823#endif
824#if defined(CONFIG_MMU)
825 {
826 .procname = "randomize_va_space",
827 .data = &randomize_va_space,
828 .maxlen = sizeof(int),
829 .mode = 0644,
830 .proc_handler = proc_dointvec,
831 },
832#endif
833#if defined(CONFIG_S390) && defined(CONFIG_SMP)
834 {
835 .procname = "spin_retry",
836 .data = &spin_retry,
837 .maxlen = sizeof (int),
838 .mode = 0644,
839 .proc_handler = proc_dointvec,
840 },
841#endif
842#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
843 {
844 .procname = "acpi_video_flags",
845 .data = &acpi_realmode_flags,
846 .maxlen = sizeof (unsigned long),
847 .mode = 0644,
848 .proc_handler = proc_doulongvec_minmax,
849 },
850#endif
851#ifdef CONFIG_IA64
852 {
853 .procname = "ignore-unaligned-usertrap",
854 .data = &no_unaligned_warning,
855 .maxlen = sizeof (int),
856 .mode = 0644,
857 .proc_handler = proc_dointvec,
858 },
859 {
860 .procname = "unaligned-dump-stack",
861 .data = &unaligned_dump_stack,
862 .maxlen = sizeof (int),
863 .mode = 0644,
864 .proc_handler = proc_dointvec,
865 },
866#endif
867#ifdef CONFIG_DETECT_HUNG_TASK
868 {
869 .procname = "hung_task_panic",
870 .data = &sysctl_hung_task_panic,
871 .maxlen = sizeof(int),
872 .mode = 0644,
873 .proc_handler = proc_dointvec_minmax,
874 .extra1 = &zero,
875 .extra2 = &one,
876 },
877 {
878 .procname = "hung_task_check_count",
879 .data = &sysctl_hung_task_check_count,
880 .maxlen = sizeof(unsigned long),
881 .mode = 0644,
882 .proc_handler = proc_doulongvec_minmax,
883 },
884 {
885 .procname = "hung_task_timeout_secs",
886 .data = &sysctl_hung_task_timeout_secs,
887 .maxlen = sizeof(unsigned long),
888 .mode = 0644,
889 .proc_handler = proc_dohung_task_timeout_secs,
890 },
891 {
892 .procname = "hung_task_warnings",
893 .data = &sysctl_hung_task_warnings,
894 .maxlen = sizeof(unsigned long),
895 .mode = 0644,
896 .proc_handler = proc_doulongvec_minmax,
897 },
898#endif
899#ifdef CONFIG_COMPAT
900 {
901 .procname = "compat-log",
902 .data = &compat_log,
903 .maxlen = sizeof (int),
904 .mode = 0644,
905 .proc_handler = proc_dointvec,
906 },
907#endif
908#ifdef CONFIG_RT_MUTEXES
909 {
910 .procname = "max_lock_depth",
911 .data = &max_lock_depth,
912 .maxlen = sizeof(int),
913 .mode = 0644,
914 .proc_handler = proc_dointvec,
915 },
916#endif
917 {
918 .procname = "poweroff_cmd",
919 .data = &poweroff_cmd,
920 .maxlen = POWEROFF_CMD_PATH_LEN,
921 .mode = 0644,
922 .proc_handler = proc_dostring,
923 },
924#ifdef CONFIG_KEYS
925 {
926 .procname = "keys",
927 .mode = 0555,
928 .child = key_sysctls,
929 },
930#endif
931#ifdef CONFIG_RCU_TORTURE_TEST
932 {
933 .procname = "rcutorture_runnable",
934 .data = &rcutorture_runnable,
935 .maxlen = sizeof(int),
936 .mode = 0644,
937 .proc_handler = proc_dointvec,
938 },
939#endif
940#ifdef CONFIG_PERF_EVENTS
941 /*
942 * User-space scripts rely on the existence of this file
943 * as a feature check for perf_events being enabled.
944 *
945 * So it's an ABI, do not remove!
946 */
947 {
948 .procname = "perf_event_paranoid",
949 .data = &sysctl_perf_event_paranoid,
950 .maxlen = sizeof(sysctl_perf_event_paranoid),
951 .mode = 0644,
952 .proc_handler = proc_dointvec,
953 },
954 {
955 .procname = "perf_event_mlock_kb",
956 .data = &sysctl_perf_event_mlock,
957 .maxlen = sizeof(sysctl_perf_event_mlock),
958 .mode = 0644,
959 .proc_handler = proc_dointvec,
960 },
961 {
962 .procname = "perf_event_max_sample_rate",
963 .data = &sysctl_perf_event_sample_rate,
964 .maxlen = sizeof(sysctl_perf_event_sample_rate),
965 .mode = 0644,
966 .proc_handler = perf_proc_update_handler,
967 },
968#endif
969#ifdef CONFIG_KMEMCHECK
970 {
971 .procname = "kmemcheck",
972 .data = &kmemcheck_enabled,
973 .maxlen = sizeof(int),
974 .mode = 0644,
975 .proc_handler = proc_dointvec,
976 },
977#endif
978#ifdef CONFIG_BLOCK
979 {
980 .procname = "blk_iopoll",
981 .data = &blk_iopoll_enabled,
982 .maxlen = sizeof(int),
983 .mode = 0644,
984 .proc_handler = proc_dointvec,
985 },
986#endif
987 { }
988};
989
990static struct ctl_table vm_table[] = {
991 {
992 .procname = "overcommit_memory",
993 .data = &sysctl_overcommit_memory,
994 .maxlen = sizeof(sysctl_overcommit_memory),
995 .mode = 0644,
996 .proc_handler = proc_dointvec_minmax,
997 .extra1 = &zero,
998 .extra2 = &two,
999 },
1000 {
1001 .procname = "panic_on_oom",
1002 .data = &sysctl_panic_on_oom,
1003 .maxlen = sizeof(sysctl_panic_on_oom),
1004 .mode = 0644,
1005 .proc_handler = proc_dointvec_minmax,
1006 .extra1 = &zero,
1007 .extra2 = &two,
1008 },
1009 {
1010 .procname = "oom_kill_allocating_task",
1011 .data = &sysctl_oom_kill_allocating_task,
1012 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1013 .mode = 0644,
1014 .proc_handler = proc_dointvec,
1015 },
1016 {
1017 .procname = "oom_dump_tasks",
1018 .data = &sysctl_oom_dump_tasks,
1019 .maxlen = sizeof(sysctl_oom_dump_tasks),
1020 .mode = 0644,
1021 .proc_handler = proc_dointvec,
1022 },
1023 {
1024 .procname = "overcommit_ratio",
1025 .data = &sysctl_overcommit_ratio,
1026 .maxlen = sizeof(sysctl_overcommit_ratio),
1027 .mode = 0644,
1028 .proc_handler = proc_dointvec,
1029 },
1030 {
1031 .procname = "page-cluster",
1032 .data = &page_cluster,
1033 .maxlen = sizeof(int),
1034 .mode = 0644,
1035 .proc_handler = proc_dointvec_minmax,
1036 .extra1 = &zero,
1037 },
1038 {
1039 .procname = "dirty_background_ratio",
1040 .data = &dirty_background_ratio,
1041 .maxlen = sizeof(dirty_background_ratio),
1042 .mode = 0644,
1043 .proc_handler = dirty_background_ratio_handler,
1044 .extra1 = &zero,
1045 .extra2 = &one_hundred,
1046 },
1047 {
1048 .procname = "dirty_background_bytes",
1049 .data = &dirty_background_bytes,
1050 .maxlen = sizeof(dirty_background_bytes),
1051 .mode = 0644,
1052 .proc_handler = dirty_background_bytes_handler,
1053 .extra1 = &one_ul,
1054 },
1055 {
1056 .procname = "dirty_ratio",
1057 .data = &vm_dirty_ratio,
1058 .maxlen = sizeof(vm_dirty_ratio),
1059 .mode = 0644,
1060 .proc_handler = dirty_ratio_handler,
1061 .extra1 = &zero,
1062 .extra2 = &one_hundred,
1063 },
1064 {
1065 .procname = "dirty_bytes",
1066 .data = &vm_dirty_bytes,
1067 .maxlen = sizeof(vm_dirty_bytes),
1068 .mode = 0644,
1069 .proc_handler = dirty_bytes_handler,
1070 .extra1 = &dirty_bytes_min,
1071 },
1072 {
1073 .procname = "dirty_writeback_centisecs",
1074 .data = &dirty_writeback_interval,
1075 .maxlen = sizeof(dirty_writeback_interval),
1076 .mode = 0644,
1077 .proc_handler = dirty_writeback_centisecs_handler,
1078 },
1079 {
1080 .procname = "dirty_expire_centisecs",
1081 .data = &dirty_expire_interval,
1082 .maxlen = sizeof(dirty_expire_interval),
1083 .mode = 0644,
1084 .proc_handler = proc_dointvec_minmax,
1085 .extra1 = &zero,
1086 },
1087 {
1088 .procname = "nr_pdflush_threads",
1089 .data = &nr_pdflush_threads,
1090 .maxlen = sizeof nr_pdflush_threads,
1091 .mode = 0444 /* read-only*/,
1092 .proc_handler = proc_dointvec,
1093 },
1094 {
1095 .procname = "swappiness",
1096 .data = &vm_swappiness,
1097 .maxlen = sizeof(vm_swappiness),
1098 .mode = 0644,
1099 .proc_handler = proc_dointvec_minmax,
1100 .extra1 = &zero,
1101 .extra2 = &one_hundred,
1102 },
1103#ifdef CONFIG_HUGETLB_PAGE
1104 {
1105 .procname = "nr_hugepages",
1106 .data = NULL,
1107 .maxlen = sizeof(unsigned long),
1108 .mode = 0644,
1109 .proc_handler = hugetlb_sysctl_handler,
1110 .extra1 = (void *)&hugetlb_zero,
1111 .extra2 = (void *)&hugetlb_infinity,
1112 },
1113#ifdef CONFIG_NUMA
1114 {
1115 .procname = "nr_hugepages_mempolicy",
1116 .data = NULL,
1117 .maxlen = sizeof(unsigned long),
1118 .mode = 0644,
1119 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1120 .extra1 = (void *)&hugetlb_zero,
1121 .extra2 = (void *)&hugetlb_infinity,
1122 },
1123#endif
1124 {
1125 .procname = "hugetlb_shm_group",
1126 .data = &sysctl_hugetlb_shm_group,
1127 .maxlen = sizeof(gid_t),
1128 .mode = 0644,
1129 .proc_handler = proc_dointvec,
1130 },
1131 {
1132 .procname = "hugepages_treat_as_movable",
1133 .data = &hugepages_treat_as_movable,
1134 .maxlen = sizeof(int),
1135 .mode = 0644,
1136 .proc_handler = hugetlb_treat_movable_handler,
1137 },
1138 {
1139 .procname = "nr_overcommit_hugepages",
1140 .data = NULL,
1141 .maxlen = sizeof(unsigned long),
1142 .mode = 0644,
1143 .proc_handler = hugetlb_overcommit_handler,
1144 .extra1 = (void *)&hugetlb_zero,
1145 .extra2 = (void *)&hugetlb_infinity,
1146 },
1147#endif
1148 {
1149 .procname = "lowmem_reserve_ratio",
1150 .data = &sysctl_lowmem_reserve_ratio,
1151 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1152 .mode = 0644,
1153 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1154 },
1155 {
1156 .procname = "drop_caches",
1157 .data = &sysctl_drop_caches,
1158 .maxlen = sizeof(int),
1159 .mode = 0644,
1160 .proc_handler = drop_caches_sysctl_handler,
1161 .extra1 = &one,
1162 .extra2 = &three,
1163 },
1164#ifdef CONFIG_COMPACTION
1165 {
1166 .procname = "compact_memory",
1167 .data = &sysctl_compact_memory,
1168 .maxlen = sizeof(int),
1169 .mode = 0200,
1170 .proc_handler = sysctl_compaction_handler,
1171 },
1172 {
1173 .procname = "extfrag_threshold",
1174 .data = &sysctl_extfrag_threshold,
1175 .maxlen = sizeof(int),
1176 .mode = 0644,
1177 .proc_handler = sysctl_extfrag_handler,
1178 .extra1 = &min_extfrag_threshold,
1179 .extra2 = &max_extfrag_threshold,
1180 },
1181
1182#endif /* CONFIG_COMPACTION */
1183 {
1184 .procname = "min_free_kbytes",
1185 .data = &min_free_kbytes,
1186 .maxlen = sizeof(min_free_kbytes),
1187 .mode = 0644,
1188 .proc_handler = min_free_kbytes_sysctl_handler,
1189 .extra1 = &zero,
1190 },
1191 {
1192 .procname = "percpu_pagelist_fraction",
1193 .data = &percpu_pagelist_fraction,
1194 .maxlen = sizeof(percpu_pagelist_fraction),
1195 .mode = 0644,
1196 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1197 .extra1 = &min_percpu_pagelist_fract,
1198 },
1199#ifdef CONFIG_MMU
1200 {
1201 .procname = "max_map_count",
1202 .data = &sysctl_max_map_count,
1203 .maxlen = sizeof(sysctl_max_map_count),
1204 .mode = 0644,
1205 .proc_handler = proc_dointvec_minmax,
1206 .extra1 = &zero,
1207 },
1208#else
1209 {
1210 .procname = "nr_trim_pages",
1211 .data = &sysctl_nr_trim_pages,
1212 .maxlen = sizeof(sysctl_nr_trim_pages),
1213 .mode = 0644,
1214 .proc_handler = proc_dointvec_minmax,
1215 .extra1 = &zero,
1216 },
1217#endif
1218 {
1219 .procname = "laptop_mode",
1220 .data = &laptop_mode,
1221 .maxlen = sizeof(laptop_mode),
1222 .mode = 0644,
1223 .proc_handler = proc_dointvec_jiffies,
1224 },
1225 {
1226 .procname = "block_dump",
1227 .data = &block_dump,
1228 .maxlen = sizeof(block_dump),
1229 .mode = 0644,
1230 .proc_handler = proc_dointvec,
1231 .extra1 = &zero,
1232 },
1233 {
1234 .procname = "vfs_cache_pressure",
1235 .data = &sysctl_vfs_cache_pressure,
1236 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1237 .mode = 0644,
1238 .proc_handler = proc_dointvec,
1239 .extra1 = &zero,
1240 },
1241#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1242 {
1243 .procname = "legacy_va_layout",
1244 .data = &sysctl_legacy_va_layout,
1245 .maxlen = sizeof(sysctl_legacy_va_layout),
1246 .mode = 0644,
1247 .proc_handler = proc_dointvec,
1248 .extra1 = &zero,
1249 },
1250#endif
1251#ifdef CONFIG_NUMA
1252 {
1253 .procname = "zone_reclaim_mode",
1254 .data = &zone_reclaim_mode,
1255 .maxlen = sizeof(zone_reclaim_mode),
1256 .mode = 0644,
1257 .proc_handler = proc_dointvec,
1258 .extra1 = &zero,
1259 },
1260 {
1261 .procname = "min_unmapped_ratio",
1262 .data = &sysctl_min_unmapped_ratio,
1263 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1264 .mode = 0644,
1265 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1266 .extra1 = &zero,
1267 .extra2 = &one_hundred,
1268 },
1269 {
1270 .procname = "min_slab_ratio",
1271 .data = &sysctl_min_slab_ratio,
1272 .maxlen = sizeof(sysctl_min_slab_ratio),
1273 .mode = 0644,
1274 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1275 .extra1 = &zero,
1276 .extra2 = &one_hundred,
1277 },
1278#endif
1279#ifdef CONFIG_SMP
1280 {
1281 .procname = "stat_interval",
1282 .data = &sysctl_stat_interval,
1283 .maxlen = sizeof(sysctl_stat_interval),
1284 .mode = 0644,
1285 .proc_handler = proc_dointvec_jiffies,
1286 },
1287#endif
1288#ifdef CONFIG_MMU
1289 {
1290 .procname = "mmap_min_addr",
1291 .data = &dac_mmap_min_addr,
1292 .maxlen = sizeof(unsigned long),
1293 .mode = 0644,
1294 .proc_handler = mmap_min_addr_handler,
1295 },
1296#endif
1297#ifdef CONFIG_NUMA
1298 {
1299 .procname = "numa_zonelist_order",
1300 .data = &numa_zonelist_order,
1301 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1302 .mode = 0644,
1303 .proc_handler = numa_zonelist_order_handler,
1304 },
1305#endif
1306#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1307 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1308 {
1309 .procname = "vdso_enabled",
1310 .data = &vdso_enabled,
1311 .maxlen = sizeof(vdso_enabled),
1312 .mode = 0644,
1313 .proc_handler = proc_dointvec,
1314 .extra1 = &zero,
1315 },
1316#endif
1317#ifdef CONFIG_HIGHMEM
1318 {
1319 .procname = "highmem_is_dirtyable",
1320 .data = &vm_highmem_is_dirtyable,
1321 .maxlen = sizeof(vm_highmem_is_dirtyable),
1322 .mode = 0644,
1323 .proc_handler = proc_dointvec_minmax,
1324 .extra1 = &zero,
1325 .extra2 = &one,
1326 },
1327#endif
1328 {
1329 .procname = "scan_unevictable_pages",
1330 .data = &scan_unevictable_pages,
1331 .maxlen = sizeof(scan_unevictable_pages),
1332 .mode = 0644,
1333 .proc_handler = scan_unevictable_handler,
1334 },
1335#ifdef CONFIG_MEMORY_FAILURE
1336 {
1337 .procname = "memory_failure_early_kill",
1338 .data = &sysctl_memory_failure_early_kill,
1339 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1340 .mode = 0644,
1341 .proc_handler = proc_dointvec_minmax,
1342 .extra1 = &zero,
1343 .extra2 = &one,
1344 },
1345 {
1346 .procname = "memory_failure_recovery",
1347 .data = &sysctl_memory_failure_recovery,
1348 .maxlen = sizeof(sysctl_memory_failure_recovery),
1349 .mode = 0644,
1350 .proc_handler = proc_dointvec_minmax,
1351 .extra1 = &zero,
1352 .extra2 = &one,
1353 },
1354#endif
1355 { }
1356};
1357
1358#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1359static struct ctl_table binfmt_misc_table[] = {
1360 { }
1361};
1362#endif
1363
1364static struct ctl_table fs_table[] = {
1365 {
1366 .procname = "inode-nr",
1367 .data = &inodes_stat,
1368 .maxlen = 2*sizeof(int),
1369 .mode = 0444,
1370 .proc_handler = proc_nr_inodes,
1371 },
1372 {
1373 .procname = "inode-state",
1374 .data = &inodes_stat,
1375 .maxlen = 7*sizeof(int),
1376 .mode = 0444,
1377 .proc_handler = proc_nr_inodes,
1378 },
1379 {
1380 .procname = "file-nr",
1381 .data = &files_stat,
1382 .maxlen = sizeof(files_stat),
1383 .mode = 0444,
1384 .proc_handler = proc_nr_files,
1385 },
1386 {
1387 .procname = "file-max",
1388 .data = &files_stat.max_files,
1389 .maxlen = sizeof(files_stat.max_files),
1390 .mode = 0644,
1391 .proc_handler = proc_doulongvec_minmax,
1392 },
1393 {
1394 .procname = "nr_open",
1395 .data = &sysctl_nr_open,
1396 .maxlen = sizeof(int),
1397 .mode = 0644,
1398 .proc_handler = proc_dointvec_minmax,
1399 .extra1 = &sysctl_nr_open_min,
1400 .extra2 = &sysctl_nr_open_max,
1401 },
1402 {
1403 .procname = "dentry-state",
1404 .data = &dentry_stat,
1405 .maxlen = 6*sizeof(int),
1406 .mode = 0444,
1407 .proc_handler = proc_nr_dentry,
1408 },
1409 {
1410 .procname = "overflowuid",
1411 .data = &fs_overflowuid,
1412 .maxlen = sizeof(int),
1413 .mode = 0644,
1414 .proc_handler = proc_dointvec_minmax,
1415 .extra1 = &minolduid,
1416 .extra2 = &maxolduid,
1417 },
1418 {
1419 .procname = "overflowgid",
1420 .data = &fs_overflowgid,
1421 .maxlen = sizeof(int),
1422 .mode = 0644,
1423 .proc_handler = proc_dointvec_minmax,
1424 .extra1 = &minolduid,
1425 .extra2 = &maxolduid,
1426 },
1427#ifdef CONFIG_FILE_LOCKING
1428 {
1429 .procname = "leases-enable",
1430 .data = &leases_enable,
1431 .maxlen = sizeof(int),
1432 .mode = 0644,
1433 .proc_handler = proc_dointvec,
1434 },
1435#endif
1436#ifdef CONFIG_DNOTIFY
1437 {
1438 .procname = "dir-notify-enable",
1439 .data = &dir_notify_enable,
1440 .maxlen = sizeof(int),
1441 .mode = 0644,
1442 .proc_handler = proc_dointvec,
1443 },
1444#endif
1445#ifdef CONFIG_MMU
1446#ifdef CONFIG_FILE_LOCKING
1447 {
1448 .procname = "lease-break-time",
1449 .data = &lease_break_time,
1450 .maxlen = sizeof(int),
1451 .mode = 0644,
1452 .proc_handler = proc_dointvec,
1453 },
1454#endif
1455#ifdef CONFIG_AIO
1456 {
1457 .procname = "aio-nr",
1458 .data = &aio_nr,
1459 .maxlen = sizeof(aio_nr),
1460 .mode = 0444,
1461 .proc_handler = proc_doulongvec_minmax,
1462 },
1463 {
1464 .procname = "aio-max-nr",
1465 .data = &aio_max_nr,
1466 .maxlen = sizeof(aio_max_nr),
1467 .mode = 0644,
1468 .proc_handler = proc_doulongvec_minmax,
1469 },
1470#endif /* CONFIG_AIO */
1471#ifdef CONFIG_INOTIFY_USER
1472 {
1473 .procname = "inotify",
1474 .mode = 0555,
1475 .child = inotify_table,
1476 },
1477#endif
1478#ifdef CONFIG_EPOLL
1479 {
1480 .procname = "epoll",
1481 .mode = 0555,
1482 .child = epoll_table,
1483 },
1484#endif
1485#endif
1486 {
1487 .procname = "suid_dumpable",
1488 .data = &suid_dumpable,
1489 .maxlen = sizeof(int),
1490 .mode = 0644,
1491 .proc_handler = proc_dointvec_minmax,
1492 .extra1 = &zero,
1493 .extra2 = &two,
1494 },
1495#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1496 {
1497 .procname = "binfmt_misc",
1498 .mode = 0555,
1499 .child = binfmt_misc_table,
1500 },
1501#endif
1502 {
1503 .procname = "pipe-max-size",
1504 .data = &pipe_max_size,
1505 .maxlen = sizeof(int),
1506 .mode = 0644,
1507 .proc_handler = &pipe_proc_fn,
1508 .extra1 = &pipe_min_size,
1509 },
1510 { }
1511};
1512
1513static struct ctl_table debug_table[] = {
1514#if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
1515 defined(CONFIG_S390) || defined(CONFIG_TILE)
1516 {
1517 .procname = "exception-trace",
1518 .data = &show_unhandled_signals,
1519 .maxlen = sizeof(int),
1520 .mode = 0644,
1521 .proc_handler = proc_dointvec
1522 },
1523#endif
1524#if defined(CONFIG_OPTPROBES)
1525 {
1526 .procname = "kprobes-optimization",
1527 .data = &sysctl_kprobes_optimization,
1528 .maxlen = sizeof(int),
1529 .mode = 0644,
1530 .proc_handler = proc_kprobes_optimization_handler,
1531 .extra1 = &zero,
1532 .extra2 = &one,
1533 },
1534#endif
1535 { }
1536};
1537
1538static struct ctl_table dev_table[] = {
1539 { }
1540};
1541
1542static DEFINE_SPINLOCK(sysctl_lock);
1543
1544/* called under sysctl_lock */
1545static int use_table(struct ctl_table_header *p)
1546{
1547 if (unlikely(p->unregistering))
1548 return 0;
1549 p->used++;
1550 return 1;
1551}
1552
1553/* called under sysctl_lock */
1554static void unuse_table(struct ctl_table_header *p)
1555{
1556 if (!--p->used)
1557 if (unlikely(p->unregistering))
1558 complete(p->unregistering);
1559}
1560
1561/* called under sysctl_lock, will reacquire if has to wait */
1562static void start_unregistering(struct ctl_table_header *p)
1563{
1564 /*
1565 * if p->used is 0, nobody will ever touch that entry again;
1566 * we'll eliminate all paths to it before dropping sysctl_lock
1567 */
1568 if (unlikely(p->used)) {
1569 struct completion wait;
1570 init_completion(&wait);
1571 p->unregistering = &wait;
1572 spin_unlock(&sysctl_lock);
1573 wait_for_completion(&wait);
1574 spin_lock(&sysctl_lock);
1575 } else {
1576 /* anything non-NULL; we'll never dereference it */
1577 p->unregistering = ERR_PTR(-EINVAL);
1578 }
1579 /*
1580 * do not remove from the list until nobody holds it; walking the
1581 * list in do_sysctl() relies on that.
1582 */
1583 list_del_init(&p->ctl_entry);
1584}
1585
1586void sysctl_head_get(struct ctl_table_header *head)
1587{
1588 spin_lock(&sysctl_lock);
1589 head->count++;
1590 spin_unlock(&sysctl_lock);
1591}
1592
1593void sysctl_head_put(struct ctl_table_header *head)
1594{
1595 spin_lock(&sysctl_lock);
1596 if (!--head->count)
1597 kfree_rcu(head, rcu);
1598 spin_unlock(&sysctl_lock);
1599}
1600
1601struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1602{
1603 if (!head)
1604 BUG();
1605 spin_lock(&sysctl_lock);
1606 if (!use_table(head))
1607 head = ERR_PTR(-ENOENT);
1608 spin_unlock(&sysctl_lock);
1609 return head;
1610}
1611
1612void sysctl_head_finish(struct ctl_table_header *head)
1613{
1614 if (!head)
1615 return;
1616 spin_lock(&sysctl_lock);
1617 unuse_table(head);
1618 spin_unlock(&sysctl_lock);
1619}
1620
1621static struct ctl_table_set *
1622lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1623{
1624 struct ctl_table_set *set = &root->default_set;
1625 if (root->lookup)
1626 set = root->lookup(root, namespaces);
1627 return set;
1628}
1629
1630static struct list_head *
1631lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1632{
1633 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1634 return &set->list;
1635}
1636
1637struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1638 struct ctl_table_header *prev)
1639{
1640 struct ctl_table_root *root;
1641 struct list_head *header_list;
1642 struct ctl_table_header *head;
1643 struct list_head *tmp;
1644
1645 spin_lock(&sysctl_lock);
1646 if (prev) {
1647 head = prev;
1648 tmp = &prev->ctl_entry;
1649 unuse_table(prev);
1650 goto next;
1651 }
1652 tmp = &root_table_header.ctl_entry;
1653 for (;;) {
1654 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1655
1656 if (!use_table(head))
1657 goto next;
1658 spin_unlock(&sysctl_lock);
1659 return head;
1660 next:
1661 root = head->root;
1662 tmp = tmp->next;
1663 header_list = lookup_header_list(root, namespaces);
1664 if (tmp != header_list)
1665 continue;
1666
1667 do {
1668 root = list_entry(root->root_list.next,
1669 struct ctl_table_root, root_list);
1670 if (root == &sysctl_table_root)
1671 goto out;
1672 header_list = lookup_header_list(root, namespaces);
1673 } while (list_empty(header_list));
1674 tmp = header_list->next;
1675 }
1676out:
1677 spin_unlock(&sysctl_lock);
1678 return NULL;
1679}
1680
1681struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1682{
1683 return __sysctl_head_next(current->nsproxy, prev);
1684}
1685
1686void register_sysctl_root(struct ctl_table_root *root)
1687{
1688 spin_lock(&sysctl_lock);
1689 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1690 spin_unlock(&sysctl_lock);
1691}
1692
1693/*
1694 * sysctl_perm does NOT grant the superuser all rights automatically, because
1695 * some sysctl variables are readonly even to root.
1696 */
1697
1698static int test_perm(int mode, int op)
1699{
1700 if (!current_euid())
1701 mode >>= 6;
1702 else if (in_egroup_p(0))
1703 mode >>= 3;
1704 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1705 return 0;
1706 return -EACCES;
1707}
1708
1709int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1710{
1711 int mode;
1712
1713 if (root->permissions)
1714 mode = root->permissions(root, current->nsproxy, table);
1715 else
1716 mode = table->mode;
1717
1718 return test_perm(mode, op);
1719}
1720
1721static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1722{
1723 for (; table->procname; table++) {
1724 table->parent = parent;
1725 if (table->child)
1726 sysctl_set_parent(table, table->child);
1727 }
1728}
1729
1730static __init int sysctl_init(void)
1731{
1732 sysctl_set_parent(NULL, root_table);
1733#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1734 sysctl_check_table(current->nsproxy, root_table);
1735#endif
1736 return 0;
1737}
1738
1739core_initcall(sysctl_init);
1740
1741static struct ctl_table *is_branch_in(struct ctl_table *branch,
1742 struct ctl_table *table)
1743{
1744 struct ctl_table *p;
1745 const char *s = branch->procname;
1746
1747 /* branch should have named subdirectory as its first element */
1748 if (!s || !branch->child)
1749 return NULL;
1750
1751 /* ... and nothing else */
1752 if (branch[1].procname)
1753 return NULL;
1754
1755 /* table should contain subdirectory with the same name */
1756 for (p = table; p->procname; p++) {
1757 if (!p->child)
1758 continue;
1759 if (p->procname && strcmp(p->procname, s) == 0)
1760 return p;
1761 }
1762 return NULL;
1763}
1764
1765/* see if attaching q to p would be an improvement */
1766static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1767{
1768 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1769 struct ctl_table *next;
1770 int is_better = 0;
1771 int not_in_parent = !p->attached_by;
1772
1773 while ((next = is_branch_in(by, to)) != NULL) {
1774 if (by == q->attached_by)
1775 is_better = 1;
1776 if (to == p->attached_by)
1777 not_in_parent = 1;
1778 by = by->child;
1779 to = next->child;
1780 }
1781
1782 if (is_better && not_in_parent) {
1783 q->attached_by = by;
1784 q->attached_to = to;
1785 q->parent = p;
1786 }
1787}
1788
1789/**
1790 * __register_sysctl_paths - register a sysctl hierarchy
1791 * @root: List of sysctl headers to register on
1792 * @namespaces: Data to compute which lists of sysctl entries are visible
1793 * @path: The path to the directory the sysctl table is in.
1794 * @table: the top-level table structure
1795 *
1796 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1797 * array. A completely 0 filled entry terminates the table.
1798 *
1799 * The members of the &struct ctl_table structure are used as follows:
1800 *
1801 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1802 * enter a sysctl file
1803 *
1804 * data - a pointer to data for use by proc_handler
1805 *
1806 * maxlen - the maximum size in bytes of the data
1807 *
1808 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1809 *
1810 * child - a pointer to the child sysctl table if this entry is a directory, or
1811 * %NULL.
1812 *
1813 * proc_handler - the text handler routine (described below)
1814 *
1815 * de - for internal use by the sysctl routines
1816 *
1817 * extra1, extra2 - extra pointers usable by the proc handler routines
1818 *
1819 * Leaf nodes in the sysctl tree will be represented by a single file
1820 * under /proc; non-leaf nodes will be represented by directories.
1821 *
1822 * sysctl(2) can automatically manage read and write requests through
1823 * the sysctl table. The data and maxlen fields of the ctl_table
1824 * struct enable minimal validation of the values being written to be
1825 * performed, and the mode field allows minimal authentication.
1826 *
1827 * There must be a proc_handler routine for any terminal nodes
1828 * mirrored under /proc/sys (non-terminals are handled by a built-in
1829 * directory handler). Several default handlers are available to
1830 * cover common cases -
1831 *
1832 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1833 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1834 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1835 *
1836 * It is the handler's job to read the input buffer from user memory
1837 * and process it. The handler should return 0 on success.
1838 *
1839 * This routine returns %NULL on a failure to register, and a pointer
1840 * to the table header on success.
1841 */
1842struct ctl_table_header *__register_sysctl_paths(
1843 struct ctl_table_root *root,
1844 struct nsproxy *namespaces,
1845 const struct ctl_path *path, struct ctl_table *table)
1846{
1847 struct ctl_table_header *header;
1848 struct ctl_table *new, **prevp;
1849 unsigned int n, npath;
1850 struct ctl_table_set *set;
1851
1852 /* Count the path components */
1853 for (npath = 0; path[npath].procname; ++npath)
1854 ;
1855
1856 /*
1857 * For each path component, allocate a 2-element ctl_table array.
1858 * The first array element will be filled with the sysctl entry
1859 * for this, the second will be the sentinel (procname == 0).
1860 *
1861 * We allocate everything in one go so that we don't have to
1862 * worry about freeing additional memory in unregister_sysctl_table.
1863 */
1864 header = kzalloc(sizeof(struct ctl_table_header) +
1865 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1866 if (!header)
1867 return NULL;
1868
1869 new = (struct ctl_table *) (header + 1);
1870
1871 /* Now connect the dots */
1872 prevp = &header->ctl_table;
1873 for (n = 0; n < npath; ++n, ++path) {
1874 /* Copy the procname */
1875 new->procname = path->procname;
1876 new->mode = 0555;
1877
1878 *prevp = new;
1879 prevp = &new->child;
1880
1881 new += 2;
1882 }
1883 *prevp = table;
1884 header->ctl_table_arg = table;
1885
1886 INIT_LIST_HEAD(&header->ctl_entry);
1887 header->used = 0;
1888 header->unregistering = NULL;
1889 header->root = root;
1890 sysctl_set_parent(NULL, header->ctl_table);
1891 header->count = 1;
1892#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1893 if (sysctl_check_table(namespaces, header->ctl_table)) {
1894 kfree(header);
1895 return NULL;
1896 }
1897#endif
1898 spin_lock(&sysctl_lock);
1899 header->set = lookup_header_set(root, namespaces);
1900 header->attached_by = header->ctl_table;
1901 header->attached_to = root_table;
1902 header->parent = &root_table_header;
1903 for (set = header->set; set; set = set->parent) {
1904 struct ctl_table_header *p;
1905 list_for_each_entry(p, &set->list, ctl_entry) {
1906 if (p->unregistering)
1907 continue;
1908 try_attach(p, header);
1909 }
1910 }
1911 header->parent->count++;
1912 list_add_tail(&header->ctl_entry, &header->set->list);
1913 spin_unlock(&sysctl_lock);
1914
1915 return header;
1916}
1917
1918/**
1919 * register_sysctl_table_path - register a sysctl table hierarchy
1920 * @path: The path to the directory the sysctl table is in.
1921 * @table: the top-level table structure
1922 *
1923 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1924 * array. A completely 0 filled entry terminates the table.
1925 *
1926 * See __register_sysctl_paths for more details.
1927 */
1928struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1929 struct ctl_table *table)
1930{
1931 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1932 path, table);
1933}
1934
1935/**
1936 * register_sysctl_table - register a sysctl table hierarchy
1937 * @table: the top-level table structure
1938 *
1939 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1940 * array. A completely 0 filled entry terminates the table.
1941 *
1942 * See register_sysctl_paths for more details.
1943 */
1944struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1945{
1946 static const struct ctl_path null_path[] = { {} };
1947
1948 return register_sysctl_paths(null_path, table);
1949}
1950
1951/**
1952 * unregister_sysctl_table - unregister a sysctl table hierarchy
1953 * @header: the header returned from register_sysctl_table
1954 *
1955 * Unregisters the sysctl table and all children. proc entries may not
1956 * actually be removed until they are no longer used by anyone.
1957 */
1958void unregister_sysctl_table(struct ctl_table_header * header)
1959{
1960 might_sleep();
1961
1962 if (header == NULL)
1963 return;
1964
1965 spin_lock(&sysctl_lock);
1966 start_unregistering(header);
1967 if (!--header->parent->count) {
1968 WARN_ON(1);
1969 kfree_rcu(header->parent, rcu);
1970 }
1971 if (!--header->count)
1972 kfree_rcu(header, rcu);
1973 spin_unlock(&sysctl_lock);
1974}
1975
1976int sysctl_is_seen(struct ctl_table_header *p)
1977{
1978 struct ctl_table_set *set = p->set;
1979 int res;
1980 spin_lock(&sysctl_lock);
1981 if (p->unregistering)
1982 res = 0;
1983 else if (!set->is_seen)
1984 res = 1;
1985 else
1986 res = set->is_seen(set);
1987 spin_unlock(&sysctl_lock);
1988 return res;
1989}
1990
1991void setup_sysctl_set(struct ctl_table_set *p,
1992 struct ctl_table_set *parent,
1993 int (*is_seen)(struct ctl_table_set *))
1994{
1995 INIT_LIST_HEAD(&p->list);
1996 p->parent = parent ? parent : &sysctl_table_root.default_set;
1997 p->is_seen = is_seen;
1998}
1999
2000#else /* !CONFIG_SYSCTL */
2001struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2002{
2003 return NULL;
2004}
2005
2006struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2007 struct ctl_table *table)
2008{
2009 return NULL;
2010}
2011
2012void unregister_sysctl_table(struct ctl_table_header * table)
2013{
2014}
2015
2016void setup_sysctl_set(struct ctl_table_set *p,
2017 struct ctl_table_set *parent,
2018 int (*is_seen)(struct ctl_table_set *))
2019{
2020}
2021
2022void sysctl_head_put(struct ctl_table_header *head)
2023{
2024}
2025
2026#endif /* CONFIG_SYSCTL */
2027
2028/*
2029 * /proc/sys support
2030 */
2031
2032#ifdef CONFIG_PROC_SYSCTL
2033
2034static int _proc_do_string(void* data, int maxlen, int write,
2035 void __user *buffer,
2036 size_t *lenp, loff_t *ppos)
2037{
2038 size_t len;
2039 char __user *p;
2040 char c;
2041
2042 if (!data || !maxlen || !*lenp) {
2043 *lenp = 0;
2044 return 0;
2045 }
2046
2047 if (write) {
2048 len = 0;
2049 p = buffer;
2050 while (len < *lenp) {
2051 if (get_user(c, p++))
2052 return -EFAULT;
2053 if (c == 0 || c == '\n')
2054 break;
2055 len++;
2056 }
2057 if (len >= maxlen)
2058 len = maxlen-1;
2059 if(copy_from_user(data, buffer, len))
2060 return -EFAULT;
2061 ((char *) data)[len] = 0;
2062 *ppos += *lenp;
2063 } else {
2064 len = strlen(data);
2065 if (len > maxlen)
2066 len = maxlen;
2067
2068 if (*ppos > len) {
2069 *lenp = 0;
2070 return 0;
2071 }
2072
2073 data += *ppos;
2074 len -= *ppos;
2075
2076 if (len > *lenp)
2077 len = *lenp;
2078 if (len)
2079 if(copy_to_user(buffer, data, len))
2080 return -EFAULT;
2081 if (len < *lenp) {
2082 if(put_user('\n', ((char __user *) buffer) + len))
2083 return -EFAULT;
2084 len++;
2085 }
2086 *lenp = len;
2087 *ppos += len;
2088 }
2089 return 0;
2090}
2091
2092/**
2093 * proc_dostring - read a string sysctl
2094 * @table: the sysctl table
2095 * @write: %TRUE if this is a write to the sysctl file
2096 * @buffer: the user buffer
2097 * @lenp: the size of the user buffer
2098 * @ppos: file position
2099 *
2100 * Reads/writes a string from/to the user buffer. If the kernel
2101 * buffer provided is not large enough to hold the string, the
2102 * string is truncated. The copied string is %NULL-terminated.
2103 * If the string is being read by the user process, it is copied
2104 * and a newline '\n' is added. It is truncated if the buffer is
2105 * not large enough.
2106 *
2107 * Returns 0 on success.
2108 */
2109int proc_dostring(struct ctl_table *table, int write,
2110 void __user *buffer, size_t *lenp, loff_t *ppos)
2111{
2112 return _proc_do_string(table->data, table->maxlen, write,
2113 buffer, lenp, ppos);
2114}
2115
2116static size_t proc_skip_spaces(char **buf)
2117{
2118 size_t ret;
2119 char *tmp = skip_spaces(*buf);
2120 ret = tmp - *buf;
2121 *buf = tmp;
2122 return ret;
2123}
2124
2125static void proc_skip_char(char **buf, size_t *size, const char v)
2126{
2127 while (*size) {
2128 if (**buf != v)
2129 break;
2130 (*size)--;
2131 (*buf)++;
2132 }
2133}
2134
2135#define TMPBUFLEN 22
2136/**
2137 * proc_get_long - reads an ASCII formatted integer from a user buffer
2138 *
2139 * @buf: a kernel buffer
2140 * @size: size of the kernel buffer
2141 * @val: this is where the number will be stored
2142 * @neg: set to %TRUE if number is negative
2143 * @perm_tr: a vector which contains the allowed trailers
2144 * @perm_tr_len: size of the perm_tr vector
2145 * @tr: pointer to store the trailer character
2146 *
2147 * In case of success %0 is returned and @buf and @size are updated with
2148 * the amount of bytes read. If @tr is non-NULL and a trailing
2149 * character exists (size is non-zero after returning from this
2150 * function), @tr is updated with the trailing character.
2151 */
2152static int proc_get_long(char **buf, size_t *size,
2153 unsigned long *val, bool *neg,
2154 const char *perm_tr, unsigned perm_tr_len, char *tr)
2155{
2156 int len;
2157 char *p, tmp[TMPBUFLEN];
2158
2159 if (!*size)
2160 return -EINVAL;
2161
2162 len = *size;
2163 if (len > TMPBUFLEN - 1)
2164 len = TMPBUFLEN - 1;
2165
2166 memcpy(tmp, *buf, len);
2167
2168 tmp[len] = 0;
2169 p = tmp;
2170 if (*p == '-' && *size > 1) {
2171 *neg = true;
2172 p++;
2173 } else
2174 *neg = false;
2175 if (!isdigit(*p))
2176 return -EINVAL;
2177
2178 *val = simple_strtoul(p, &p, 0);
2179
2180 len = p - tmp;
2181
2182 /* We don't know if the next char is whitespace thus we may accept
2183 * invalid integers (e.g. 1234...a) or two integers instead of one
2184 * (e.g. 123...1). So lets not allow such large numbers. */
2185 if (len == TMPBUFLEN - 1)
2186 return -EINVAL;
2187
2188 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2189 return -EINVAL;
2190
2191 if (tr && (len < *size))
2192 *tr = *p;
2193
2194 *buf += len;
2195 *size -= len;
2196
2197 return 0;
2198}
2199
2200/**
2201 * proc_put_long - converts an integer to a decimal ASCII formatted string
2202 *
2203 * @buf: the user buffer
2204 * @size: the size of the user buffer
2205 * @val: the integer to be converted
2206 * @neg: sign of the number, %TRUE for negative
2207 *
2208 * In case of success %0 is returned and @buf and @size are updated with
2209 * the amount of bytes written.
2210 */
2211static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2212 bool neg)
2213{
2214 int len;
2215 char tmp[TMPBUFLEN], *p = tmp;
2216
2217 sprintf(p, "%s%lu", neg ? "-" : "", val);
2218 len = strlen(tmp);
2219 if (len > *size)
2220 len = *size;
2221 if (copy_to_user(*buf, tmp, len))
2222 return -EFAULT;
2223 *size -= len;
2224 *buf += len;
2225 return 0;
2226}
2227#undef TMPBUFLEN
2228
2229static int proc_put_char(void __user **buf, size_t *size, char c)
2230{
2231 if (*size) {
2232 char __user **buffer = (char __user **)buf;
2233 if (put_user(c, *buffer))
2234 return -EFAULT;
2235 (*size)--, (*buffer)++;
2236 *buf = *buffer;
2237 }
2238 return 0;
2239}
2240
2241static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2242 int *valp,
2243 int write, void *data)
2244{
2245 if (write) {
2246 *valp = *negp ? -*lvalp : *lvalp;
2247 } else {
2248 int val = *valp;
2249 if (val < 0) {
2250 *negp = true;
2251 *lvalp = (unsigned long)-val;
2252 } else {
2253 *negp = false;
2254 *lvalp = (unsigned long)val;
2255 }
2256 }
2257 return 0;
2258}
2259
2260static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2261
2262static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2263 int write, void __user *buffer,
2264 size_t *lenp, loff_t *ppos,
2265 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2266 int write, void *data),
2267 void *data)
2268{
2269 int *i, vleft, first = 1, err = 0;
2270 unsigned long page = 0;
2271 size_t left;
2272 char *kbuf;
2273
2274 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2275 *lenp = 0;
2276 return 0;
2277 }
2278
2279 i = (int *) tbl_data;
2280 vleft = table->maxlen / sizeof(*i);
2281 left = *lenp;
2282
2283 if (!conv)
2284 conv = do_proc_dointvec_conv;
2285
2286 if (write) {
2287 if (left > PAGE_SIZE - 1)
2288 left = PAGE_SIZE - 1;
2289 page = __get_free_page(GFP_TEMPORARY);
2290 kbuf = (char *) page;
2291 if (!kbuf)
2292 return -ENOMEM;
2293 if (copy_from_user(kbuf, buffer, left)) {
2294 err = -EFAULT;
2295 goto free;
2296 }
2297 kbuf[left] = 0;
2298 }
2299
2300 for (; left && vleft--; i++, first=0) {
2301 unsigned long lval;
2302 bool neg;
2303
2304 if (write) {
2305 left -= proc_skip_spaces(&kbuf);
2306
2307 if (!left)
2308 break;
2309 err = proc_get_long(&kbuf, &left, &lval, &neg,
2310 proc_wspace_sep,
2311 sizeof(proc_wspace_sep), NULL);
2312 if (err)
2313 break;
2314 if (conv(&neg, &lval, i, 1, data)) {
2315 err = -EINVAL;
2316 break;
2317 }
2318 } else {
2319 if (conv(&neg, &lval, i, 0, data)) {
2320 err = -EINVAL;
2321 break;
2322 }
2323 if (!first)
2324 err = proc_put_char(&buffer, &left, '\t');
2325 if (err)
2326 break;
2327 err = proc_put_long(&buffer, &left, lval, neg);
2328 if (err)
2329 break;
2330 }
2331 }
2332
2333 if (!write && !first && left && !err)
2334 err = proc_put_char(&buffer, &left, '\n');
2335 if (write && !err && left)
2336 left -= proc_skip_spaces(&kbuf);
2337free:
2338 if (write) {
2339 free_page(page);
2340 if (first)
2341 return err ? : -EINVAL;
2342 }
2343 *lenp -= left;
2344 *ppos += *lenp;
2345 return err;
2346}
2347
2348static int do_proc_dointvec(struct ctl_table *table, int write,
2349 void __user *buffer, size_t *lenp, loff_t *ppos,
2350 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2351 int write, void *data),
2352 void *data)
2353{
2354 return __do_proc_dointvec(table->data, table, write,
2355 buffer, lenp, ppos, conv, data);
2356}
2357
2358/**
2359 * proc_dointvec - read a vector of integers
2360 * @table: the sysctl table
2361 * @write: %TRUE if this is a write to the sysctl file
2362 * @buffer: the user buffer
2363 * @lenp: the size of the user buffer
2364 * @ppos: file position
2365 *
2366 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2367 * values from/to the user buffer, treated as an ASCII string.
2368 *
2369 * Returns 0 on success.
2370 */
2371int proc_dointvec(struct ctl_table *table, int write,
2372 void __user *buffer, size_t *lenp, loff_t *ppos)
2373{
2374 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2375 NULL,NULL);
2376}
2377
2378/*
2379 * Taint values can only be increased
2380 * This means we can safely use a temporary.
2381 */
2382static int proc_taint(struct ctl_table *table, int write,
2383 void __user *buffer, size_t *lenp, loff_t *ppos)
2384{
2385 struct ctl_table t;
2386 unsigned long tmptaint = get_taint();
2387 int err;
2388
2389 if (write && !capable(CAP_SYS_ADMIN))
2390 return -EPERM;
2391
2392 t = *table;
2393 t.data = &tmptaint;
2394 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2395 if (err < 0)
2396 return err;
2397
2398 if (write) {
2399 /*
2400 * Poor man's atomic or. Not worth adding a primitive
2401 * to everyone's atomic.h for this
2402 */
2403 int i;
2404 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2405 if ((tmptaint >> i) & 1)
2406 add_taint(i);
2407 }
2408 }
2409
2410 return err;
2411}
2412
2413#ifdef CONFIG_PRINTK
2414static int proc_dmesg_restrict(struct ctl_table *table, int write,
2415 void __user *buffer, size_t *lenp, loff_t *ppos)
2416{
2417 if (write && !capable(CAP_SYS_ADMIN))
2418 return -EPERM;
2419
2420 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2421}
2422#endif
2423
2424struct do_proc_dointvec_minmax_conv_param {
2425 int *min;
2426 int *max;
2427};
2428
2429static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2430 int *valp,
2431 int write, void *data)
2432{
2433 struct do_proc_dointvec_minmax_conv_param *param = data;
2434 if (write) {
2435 int val = *negp ? -*lvalp : *lvalp;
2436 if ((param->min && *param->min > val) ||
2437 (param->max && *param->max < val))
2438 return -EINVAL;
2439 *valp = val;
2440 } else {
2441 int val = *valp;
2442 if (val < 0) {
2443 *negp = true;
2444 *lvalp = (unsigned long)-val;
2445 } else {
2446 *negp = false;
2447 *lvalp = (unsigned long)val;
2448 }
2449 }
2450 return 0;
2451}
2452
2453/**
2454 * proc_dointvec_minmax - read a vector of integers with min/max values
2455 * @table: the sysctl table
2456 * @write: %TRUE if this is a write to the sysctl file
2457 * @buffer: the user buffer
2458 * @lenp: the size of the user buffer
2459 * @ppos: file position
2460 *
2461 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2462 * values from/to the user buffer, treated as an ASCII string.
2463 *
2464 * This routine will ensure the values are within the range specified by
2465 * table->extra1 (min) and table->extra2 (max).
2466 *
2467 * Returns 0 on success.
2468 */
2469int proc_dointvec_minmax(struct ctl_table *table, int write,
2470 void __user *buffer, size_t *lenp, loff_t *ppos)
2471{
2472 struct do_proc_dointvec_minmax_conv_param param = {
2473 .min = (int *) table->extra1,
2474 .max = (int *) table->extra2,
2475 };
2476 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2477 do_proc_dointvec_minmax_conv, ¶m);
2478}
2479
2480static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2481 void __user *buffer,
2482 size_t *lenp, loff_t *ppos,
2483 unsigned long convmul,
2484 unsigned long convdiv)
2485{
2486 unsigned long *i, *min, *max;
2487 int vleft, first = 1, err = 0;
2488 unsigned long page = 0;
2489 size_t left;
2490 char *kbuf;
2491
2492 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2493 *lenp = 0;
2494 return 0;
2495 }
2496
2497 i = (unsigned long *) data;
2498 min = (unsigned long *) table->extra1;
2499 max = (unsigned long *) table->extra2;
2500 vleft = table->maxlen / sizeof(unsigned long);
2501 left = *lenp;
2502
2503 if (write) {
2504 if (left > PAGE_SIZE - 1)
2505 left = PAGE_SIZE - 1;
2506 page = __get_free_page(GFP_TEMPORARY);
2507 kbuf = (char *) page;
2508 if (!kbuf)
2509 return -ENOMEM;
2510 if (copy_from_user(kbuf, buffer, left)) {
2511 err = -EFAULT;
2512 goto free;
2513 }
2514 kbuf[left] = 0;
2515 }
2516
2517 for (; left && vleft--; i++, first = 0) {
2518 unsigned long val;
2519
2520 if (write) {
2521 bool neg;
2522
2523 left -= proc_skip_spaces(&kbuf);
2524
2525 err = proc_get_long(&kbuf, &left, &val, &neg,
2526 proc_wspace_sep,
2527 sizeof(proc_wspace_sep), NULL);
2528 if (err)
2529 break;
2530 if (neg)
2531 continue;
2532 if ((min && val < *min) || (max && val > *max))
2533 continue;
2534 *i = val;
2535 } else {
2536 val = convdiv * (*i) / convmul;
2537 if (!first)
2538 err = proc_put_char(&buffer, &left, '\t');
2539 err = proc_put_long(&buffer, &left, val, false);
2540 if (err)
2541 break;
2542 }
2543 }
2544
2545 if (!write && !first && left && !err)
2546 err = proc_put_char(&buffer, &left, '\n');
2547 if (write && !err)
2548 left -= proc_skip_spaces(&kbuf);
2549free:
2550 if (write) {
2551 free_page(page);
2552 if (first)
2553 return err ? : -EINVAL;
2554 }
2555 *lenp -= left;
2556 *ppos += *lenp;
2557 return err;
2558}
2559
2560static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2561 void __user *buffer,
2562 size_t *lenp, loff_t *ppos,
2563 unsigned long convmul,
2564 unsigned long convdiv)
2565{
2566 return __do_proc_doulongvec_minmax(table->data, table, write,
2567 buffer, lenp, ppos, convmul, convdiv);
2568}
2569
2570/**
2571 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2572 * @table: the sysctl table
2573 * @write: %TRUE if this is a write to the sysctl file
2574 * @buffer: the user buffer
2575 * @lenp: the size of the user buffer
2576 * @ppos: file position
2577 *
2578 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2579 * values from/to the user buffer, treated as an ASCII string.
2580 *
2581 * This routine will ensure the values are within the range specified by
2582 * table->extra1 (min) and table->extra2 (max).
2583 *
2584 * Returns 0 on success.
2585 */
2586int proc_doulongvec_minmax(struct ctl_table *table, int write,
2587 void __user *buffer, size_t *lenp, loff_t *ppos)
2588{
2589 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2590}
2591
2592/**
2593 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2594 * @table: the sysctl table
2595 * @write: %TRUE if this is a write to the sysctl file
2596 * @buffer: the user buffer
2597 * @lenp: the size of the user buffer
2598 * @ppos: file position
2599 *
2600 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2601 * values from/to the user buffer, treated as an ASCII string. The values
2602 * are treated as milliseconds, and converted to jiffies when they are stored.
2603 *
2604 * This routine will ensure the values are within the range specified by
2605 * table->extra1 (min) and table->extra2 (max).
2606 *
2607 * Returns 0 on success.
2608 */
2609int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2610 void __user *buffer,
2611 size_t *lenp, loff_t *ppos)
2612{
2613 return do_proc_doulongvec_minmax(table, write, buffer,
2614 lenp, ppos, HZ, 1000l);
2615}
2616
2617
2618static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2619 int *valp,
2620 int write, void *data)
2621{
2622 if (write) {
2623 if (*lvalp > LONG_MAX / HZ)
2624 return 1;
2625 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2626 } else {
2627 int val = *valp;
2628 unsigned long lval;
2629 if (val < 0) {
2630 *negp = true;
2631 lval = (unsigned long)-val;
2632 } else {
2633 *negp = false;
2634 lval = (unsigned long)val;
2635 }
2636 *lvalp = lval / HZ;
2637 }
2638 return 0;
2639}
2640
2641static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2642 int *valp,
2643 int write, void *data)
2644{
2645 if (write) {
2646 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2647 return 1;
2648 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2649 } else {
2650 int val = *valp;
2651 unsigned long lval;
2652 if (val < 0) {
2653 *negp = true;
2654 lval = (unsigned long)-val;
2655 } else {
2656 *negp = false;
2657 lval = (unsigned long)val;
2658 }
2659 *lvalp = jiffies_to_clock_t(lval);
2660 }
2661 return 0;
2662}
2663
2664static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2665 int *valp,
2666 int write, void *data)
2667{
2668 if (write) {
2669 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2670 } else {
2671 int val = *valp;
2672 unsigned long lval;
2673 if (val < 0) {
2674 *negp = true;
2675 lval = (unsigned long)-val;
2676 } else {
2677 *negp = false;
2678 lval = (unsigned long)val;
2679 }
2680 *lvalp = jiffies_to_msecs(lval);
2681 }
2682 return 0;
2683}
2684
2685/**
2686 * proc_dointvec_jiffies - read a vector of integers as seconds
2687 * @table: the sysctl table
2688 * @write: %TRUE if this is a write to the sysctl file
2689 * @buffer: the user buffer
2690 * @lenp: the size of the user buffer
2691 * @ppos: file position
2692 *
2693 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2694 * values from/to the user buffer, treated as an ASCII string.
2695 * The values read are assumed to be in seconds, and are converted into
2696 * jiffies.
2697 *
2698 * Returns 0 on success.
2699 */
2700int proc_dointvec_jiffies(struct ctl_table *table, int write,
2701 void __user *buffer, size_t *lenp, loff_t *ppos)
2702{
2703 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2704 do_proc_dointvec_jiffies_conv,NULL);
2705}
2706
2707/**
2708 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2709 * @table: the sysctl table
2710 * @write: %TRUE if this is a write to the sysctl file
2711 * @buffer: the user buffer
2712 * @lenp: the size of the user buffer
2713 * @ppos: pointer to the file position
2714 *
2715 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2716 * values from/to the user buffer, treated as an ASCII string.
2717 * The values read are assumed to be in 1/USER_HZ seconds, and
2718 * are converted into jiffies.
2719 *
2720 * Returns 0 on success.
2721 */
2722int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2723 void __user *buffer, size_t *lenp, loff_t *ppos)
2724{
2725 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2726 do_proc_dointvec_userhz_jiffies_conv,NULL);
2727}
2728
2729/**
2730 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2731 * @table: the sysctl table
2732 * @write: %TRUE if this is a write to the sysctl file
2733 * @buffer: the user buffer
2734 * @lenp: the size of the user buffer
2735 * @ppos: file position
2736 * @ppos: the current position in the file
2737 *
2738 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2739 * values from/to the user buffer, treated as an ASCII string.
2740 * The values read are assumed to be in 1/1000 seconds, and
2741 * are converted into jiffies.
2742 *
2743 * Returns 0 on success.
2744 */
2745int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2746 void __user *buffer, size_t *lenp, loff_t *ppos)
2747{
2748 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2749 do_proc_dointvec_ms_jiffies_conv, NULL);
2750}
2751
2752static int proc_do_cad_pid(struct ctl_table *table, int write,
2753 void __user *buffer, size_t *lenp, loff_t *ppos)
2754{
2755 struct pid *new_pid;
2756 pid_t tmp;
2757 int r;
2758
2759 tmp = pid_vnr(cad_pid);
2760
2761 r = __do_proc_dointvec(&tmp, table, write, buffer,
2762 lenp, ppos, NULL, NULL);
2763 if (r || !write)
2764 return r;
2765
2766 new_pid = find_get_pid(tmp);
2767 if (!new_pid)
2768 return -ESRCH;
2769
2770 put_pid(xchg(&cad_pid, new_pid));
2771 return 0;
2772}
2773
2774/**
2775 * proc_do_large_bitmap - read/write from/to a large bitmap
2776 * @table: the sysctl table
2777 * @write: %TRUE if this is a write to the sysctl file
2778 * @buffer: the user buffer
2779 * @lenp: the size of the user buffer
2780 * @ppos: file position
2781 *
2782 * The bitmap is stored at table->data and the bitmap length (in bits)
2783 * in table->maxlen.
2784 *
2785 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2786 * large bitmaps may be represented in a compact manner. Writing into
2787 * the file will clear the bitmap then update it with the given input.
2788 *
2789 * Returns 0 on success.
2790 */
2791int proc_do_large_bitmap(struct ctl_table *table, int write,
2792 void __user *buffer, size_t *lenp, loff_t *ppos)
2793{
2794 int err = 0;
2795 bool first = 1;
2796 size_t left = *lenp;
2797 unsigned long bitmap_len = table->maxlen;
2798 unsigned long *bitmap = (unsigned long *) table->data;
2799 unsigned long *tmp_bitmap = NULL;
2800 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2801
2802 if (!bitmap_len || !left || (*ppos && !write)) {
2803 *lenp = 0;
2804 return 0;
2805 }
2806
2807 if (write) {
2808 unsigned long page = 0;
2809 char *kbuf;
2810
2811 if (left > PAGE_SIZE - 1)
2812 left = PAGE_SIZE - 1;
2813
2814 page = __get_free_page(GFP_TEMPORARY);
2815 kbuf = (char *) page;
2816 if (!kbuf)
2817 return -ENOMEM;
2818 if (copy_from_user(kbuf, buffer, left)) {
2819 free_page(page);
2820 return -EFAULT;
2821 }
2822 kbuf[left] = 0;
2823
2824 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2825 GFP_KERNEL);
2826 if (!tmp_bitmap) {
2827 free_page(page);
2828 return -ENOMEM;
2829 }
2830 proc_skip_char(&kbuf, &left, '\n');
2831 while (!err && left) {
2832 unsigned long val_a, val_b;
2833 bool neg;
2834
2835 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2836 sizeof(tr_a), &c);
2837 if (err)
2838 break;
2839 if (val_a >= bitmap_len || neg) {
2840 err = -EINVAL;
2841 break;
2842 }
2843
2844 val_b = val_a;
2845 if (left) {
2846 kbuf++;
2847 left--;
2848 }
2849
2850 if (c == '-') {
2851 err = proc_get_long(&kbuf, &left, &val_b,
2852 &neg, tr_b, sizeof(tr_b),
2853 &c);
2854 if (err)
2855 break;
2856 if (val_b >= bitmap_len || neg ||
2857 val_a > val_b) {
2858 err = -EINVAL;
2859 break;
2860 }
2861 if (left) {
2862 kbuf++;
2863 left--;
2864 }
2865 }
2866
2867 while (val_a <= val_b)
2868 set_bit(val_a++, tmp_bitmap);
2869
2870 first = 0;
2871 proc_skip_char(&kbuf, &left, '\n');
2872 }
2873 free_page(page);
2874 } else {
2875 unsigned long bit_a, bit_b = 0;
2876
2877 while (left) {
2878 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2879 if (bit_a >= bitmap_len)
2880 break;
2881 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2882 bit_a + 1) - 1;
2883
2884 if (!first) {
2885 err = proc_put_char(&buffer, &left, ',');
2886 if (err)
2887 break;
2888 }
2889 err = proc_put_long(&buffer, &left, bit_a, false);
2890 if (err)
2891 break;
2892 if (bit_a != bit_b) {
2893 err = proc_put_char(&buffer, &left, '-');
2894 if (err)
2895 break;
2896 err = proc_put_long(&buffer, &left, bit_b, false);
2897 if (err)
2898 break;
2899 }
2900
2901 first = 0; bit_b++;
2902 }
2903 if (!err)
2904 err = proc_put_char(&buffer, &left, '\n');
2905 }
2906
2907 if (!err) {
2908 if (write) {
2909 if (*ppos)
2910 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2911 else
2912 memcpy(bitmap, tmp_bitmap,
2913 BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long));
2914 }
2915 kfree(tmp_bitmap);
2916 *lenp -= left;
2917 *ppos += *lenp;
2918 return 0;
2919 } else {
2920 kfree(tmp_bitmap);
2921 return err;
2922 }
2923}
2924
2925#else /* CONFIG_PROC_SYSCTL */
2926
2927int proc_dostring(struct ctl_table *table, int write,
2928 void __user *buffer, size_t *lenp, loff_t *ppos)
2929{
2930 return -ENOSYS;
2931}
2932
2933int proc_dointvec(struct ctl_table *table, int write,
2934 void __user *buffer, size_t *lenp, loff_t *ppos)
2935{
2936 return -ENOSYS;
2937}
2938
2939int proc_dointvec_minmax(struct ctl_table *table, int write,
2940 void __user *buffer, size_t *lenp, loff_t *ppos)
2941{
2942 return -ENOSYS;
2943}
2944
2945int proc_dointvec_jiffies(struct ctl_table *table, int write,
2946 void __user *buffer, size_t *lenp, loff_t *ppos)
2947{
2948 return -ENOSYS;
2949}
2950
2951int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2952 void __user *buffer, size_t *lenp, loff_t *ppos)
2953{
2954 return -ENOSYS;
2955}
2956
2957int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2958 void __user *buffer, size_t *lenp, loff_t *ppos)
2959{
2960 return -ENOSYS;
2961}
2962
2963int proc_doulongvec_minmax(struct ctl_table *table, int write,
2964 void __user *buffer, size_t *lenp, loff_t *ppos)
2965{
2966 return -ENOSYS;
2967}
2968
2969int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2970 void __user *buffer,
2971 size_t *lenp, loff_t *ppos)
2972{
2973 return -ENOSYS;
2974}
2975
2976
2977#endif /* CONFIG_PROC_SYSCTL */
2978
2979/*
2980 * No sense putting this after each symbol definition, twice,
2981 * exception granted :-)
2982 */
2983EXPORT_SYMBOL(proc_dointvec);
2984EXPORT_SYMBOL(proc_dointvec_jiffies);
2985EXPORT_SYMBOL(proc_dointvec_minmax);
2986EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2987EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2988EXPORT_SYMBOL(proc_dostring);
2989EXPORT_SYMBOL(proc_doulongvec_minmax);
2990EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2991EXPORT_SYMBOL(register_sysctl_table);
2992EXPORT_SYMBOL(register_sysctl_paths);
2993EXPORT_SYMBOL(unregister_sysctl_table);