projects / deliverable / linux.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
kernel/watchdog.c: perform all-CPU backtrace in case of hard lockup
[deliverable/linux.git] / kernel / sysctl.c
1 /*
2 * sysctl.c: General linux system control interface
3 *
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12 * Horn.
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16 * Wendling.
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19 */
20
21 #include <linux/module.h>
22 #include <linux/aio.h>
23 #include <linux/mm.h>
24 #include <linux/swap.h>
25 #include <linux/slab.h>
26 #include <linux/sysctl.h>
27 #include <linux/bitmap.h>
28 #include <linux/signal.h>
29 #include <linux/printk.h>
30 #include <linux/proc_fs.h>
31 #include <linux/security.h>
32 #include <linux/ctype.h>
33 #include <linux/kmemcheck.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/kexec.h>
67 #include <linux/bpf.h>
68
69 #include <asm/uaccess.h>
70 #include <asm/processor.h>
71
72 #ifdef CONFIG_X86
73 #include <asm/nmi.h>
74 #include <asm/stacktrace.h>
75 #include <asm/io.h>
76 #endif
77 #ifdef CONFIG_SPARC
78 #include <asm/setup.h>
79 #endif
80 #ifdef CONFIG_BSD_PROCESS_ACCT
81 #include <linux/acct.h>
82 #endif
83 #ifdef CONFIG_RT_MUTEXES
84 #include <linux/rtmutex.h>
85 #endif
86 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
87 #include <linux/lockdep.h>
88 #endif
89 #ifdef CONFIG_CHR_DEV_SG
90 #include <scsi/sg.h>
91 #endif
92
93 #ifdef CONFIG_LOCKUP_DETECTOR
94 #include <linux/nmi.h>
95 #endif
96
97 #if defined(CONFIG_SYSCTL)
98
99 /* External variables not in a header file. */
100 extern int suid_dumpable;
101 #ifdef CONFIG_COREDUMP
102 extern int core_uses_pid;
103 extern char core_pattern[];
104 extern unsigned int core_pipe_limit;
105 #endif
106 extern int pid_max;
107 extern int pid_max_min, pid_max_max;
108 extern int percpu_pagelist_fraction;
109 extern int compat_log;
110 extern int latencytop_enabled;
111 extern int sysctl_nr_open_min, sysctl_nr_open_max;
112 #ifndef CONFIG_MMU
113 extern int sysctl_nr_trim_pages;
114 #endif
115
116 /* Constants used for minimum and maximum */
117 #ifdef CONFIG_LOCKUP_DETECTOR
118 static int sixty = 60;
119 #endif
120
121 static int __maybe_unused neg_one = -1;
122
123 static int zero;
124 static int __maybe_unused one = 1;
125 static int __maybe_unused two = 2;
126 static int __maybe_unused four = 4;
127 static unsigned long one_ul = 1;
128 static int one_hundred = 100;
129 #ifdef CONFIG_PRINTK
130 static int ten_thousand = 10000;
131 #endif
132
133 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
134 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
135
136 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
137 static int maxolduid = 65535;
138 static int minolduid;
139
140 static int ngroups_max = NGROUPS_MAX;
141 static const int cap_last_cap = CAP_LAST_CAP;
142
143 /*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */
144 #ifdef CONFIG_DETECT_HUNG_TASK
145 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
146 #endif
147
148 #ifdef CONFIG_INOTIFY_USER
149 #include <linux/inotify.h>
150 #endif
151 #ifdef CONFIG_SPARC
152 #endif
153
154 #ifdef __hppa__
155 extern int pwrsw_enabled;
156 #endif
157
158 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
159 extern int unaligned_enabled;
160 #endif
161
162 #ifdef CONFIG_IA64
163 extern int unaligned_dump_stack;
164 #endif
165
166 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
167 extern int no_unaligned_warning;
168 #endif
169
170 #ifdef CONFIG_PROC_SYSCTL
171
172 #define SYSCTL_WRITES_LEGACY -1
173 #define SYSCTL_WRITES_WARN 0
174 #define SYSCTL_WRITES_STRICT 1
175
176 static int sysctl_writes_strict = SYSCTL_WRITES_WARN;
177
178 static int proc_do_cad_pid(struct ctl_table *table, int write,
179 void __user *buffer, size_t *lenp, loff_t *ppos);
180 static int proc_taint(struct ctl_table *table, int write,
181 void __user *buffer, size_t *lenp, loff_t *ppos);
182 #endif
183
184 #ifdef CONFIG_PRINTK
185 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
186 void __user *buffer, size_t *lenp, loff_t *ppos);
187 #endif
188
189 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
190 void __user *buffer, size_t *lenp, loff_t *ppos);
191 #ifdef CONFIG_COREDUMP
192 static int proc_dostring_coredump(struct ctl_table *table, int write,
193 void __user *buffer, size_t *lenp, loff_t *ppos);
194 #endif
195
196 #ifdef CONFIG_MAGIC_SYSRQ
197 /* Note: sysrq code uses it's own private copy */
198 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
199
200 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
201 void __user *buffer, size_t *lenp,
202 loff_t *ppos)
203 {
204 int error;
205
206 error = proc_dointvec(table, write, buffer, lenp, ppos);
207 if (error)
208 return error;
209
210 if (write)
211 sysrq_toggle_support(__sysrq_enabled);
212
213 return 0;
214 }
215
216 #endif
217
218 static struct ctl_table kern_table[];
219 static struct ctl_table vm_table[];
220 static struct ctl_table fs_table[];
221 static struct ctl_table debug_table[];
222 static struct ctl_table dev_table[];
223 extern struct ctl_table random_table[];
224 #ifdef CONFIG_EPOLL
225 extern struct ctl_table epoll_table[];
226 #endif
227
228 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
229 int sysctl_legacy_va_layout;
230 #endif
231
232 /* The default sysctl tables: */
233
234 static struct ctl_table sysctl_base_table[] = {
235 {
236 .procname = "kernel",
237 .mode = 0555,
238 .child = kern_table,
239 },
240 {
241 .procname = "vm",
242 .mode = 0555,
243 .child = vm_table,
244 },
245 {
246 .procname = "fs",
247 .mode = 0555,
248 .child = fs_table,
249 },
250 {
251 .procname = "debug",
252 .mode = 0555,
253 .child = debug_table,
254 },
255 {
256 .procname = "dev",
257 .mode = 0555,
258 .child = dev_table,
259 },
260 { }
261 };
262
263 #ifdef CONFIG_SCHED_DEBUG
264 static int min_sched_granularity_ns = 100000; /* 100 usecs */
265 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
266 static int min_wakeup_granularity_ns; /* 0 usecs */
267 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
268 #ifdef CONFIG_SMP
269 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
270 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
271 #endif /* CONFIG_SMP */
272 #endif /* CONFIG_SCHED_DEBUG */
273
274 #ifdef CONFIG_COMPACTION
275 static int min_extfrag_threshold;
276 static int max_extfrag_threshold = 1000;
277 #endif
278
279 static struct ctl_table kern_table[] = {
280 {
281 .procname = "sched_child_runs_first",
282 .data = &sysctl_sched_child_runs_first,
283 .maxlen = sizeof(unsigned int),
284 .mode = 0644,
285 .proc_handler = proc_dointvec,
286 },
287 #ifdef CONFIG_SCHED_DEBUG
288 {
289 .procname = "sched_min_granularity_ns",
290 .data = &sysctl_sched_min_granularity,
291 .maxlen = sizeof(unsigned int),
292 .mode = 0644,
293 .proc_handler = sched_proc_update_handler,
294 .extra1 = &min_sched_granularity_ns,
295 .extra2 = &max_sched_granularity_ns,
296 },
297 {
298 .procname = "sched_latency_ns",
299 .data = &sysctl_sched_latency,
300 .maxlen = sizeof(unsigned int),
301 .mode = 0644,
302 .proc_handler = sched_proc_update_handler,
303 .extra1 = &min_sched_granularity_ns,
304 .extra2 = &max_sched_granularity_ns,
305 },
306 {
307 .procname = "sched_wakeup_granularity_ns",
308 .data = &sysctl_sched_wakeup_granularity,
309 .maxlen = sizeof(unsigned int),
310 .mode = 0644,
311 .proc_handler = sched_proc_update_handler,
312 .extra1 = &min_wakeup_granularity_ns,
313 .extra2 = &max_wakeup_granularity_ns,
314 },
315 #ifdef CONFIG_SMP
316 {
317 .procname = "sched_tunable_scaling",
318 .data = &sysctl_sched_tunable_scaling,
319 .maxlen = sizeof(enum sched_tunable_scaling),
320 .mode = 0644,
321 .proc_handler = sched_proc_update_handler,
322 .extra1 = &min_sched_tunable_scaling,
323 .extra2 = &max_sched_tunable_scaling,
324 },
325 {
326 .procname = "sched_migration_cost_ns",
327 .data = &sysctl_sched_migration_cost,
328 .maxlen = sizeof(unsigned int),
329 .mode = 0644,
330 .proc_handler = proc_dointvec,
331 },
332 {
333 .procname = "sched_nr_migrate",
334 .data = &sysctl_sched_nr_migrate,
335 .maxlen = sizeof(unsigned int),
336 .mode = 0644,
337 .proc_handler = proc_dointvec,
338 },
339 {
340 .procname = "sched_time_avg_ms",
341 .data = &sysctl_sched_time_avg,
342 .maxlen = sizeof(unsigned int),
343 .mode = 0644,
344 .proc_handler = proc_dointvec,
345 },
346 {
347 .procname = "sched_shares_window_ns",
348 .data = &sysctl_sched_shares_window,
349 .maxlen = sizeof(unsigned int),
350 .mode = 0644,
351 .proc_handler = proc_dointvec,
352 },
353 #endif /* CONFIG_SMP */
354 #ifdef CONFIG_NUMA_BALANCING
355 {
356 .procname = "numa_balancing_scan_delay_ms",
357 .data = &sysctl_numa_balancing_scan_delay,
358 .maxlen = sizeof(unsigned int),
359 .mode = 0644,
360 .proc_handler = proc_dointvec,
361 },
362 {
363 .procname = "numa_balancing_scan_period_min_ms",
364 .data = &sysctl_numa_balancing_scan_period_min,
365 .maxlen = sizeof(unsigned int),
366 .mode = 0644,
367 .proc_handler = proc_dointvec,
368 },
369 {
370 .procname = "numa_balancing_scan_period_max_ms",
371 .data = &sysctl_numa_balancing_scan_period_max,
372 .maxlen = sizeof(unsigned int),
373 .mode = 0644,
374 .proc_handler = proc_dointvec,
375 },
376 {
377 .procname = "numa_balancing_scan_size_mb",
378 .data = &sysctl_numa_balancing_scan_size,
379 .maxlen = sizeof(unsigned int),
380 .mode = 0644,
381 .proc_handler = proc_dointvec_minmax,
382 .extra1 = &one,
383 },
384 {
385 .procname = "numa_balancing",
386 .data = NULL, /* filled in by handler */
387 .maxlen = sizeof(unsigned int),
388 .mode = 0644,
389 .proc_handler = sysctl_numa_balancing,
390 .extra1 = &zero,
391 .extra2 = &one,
392 },
393 #endif /* CONFIG_NUMA_BALANCING */
394 #endif /* CONFIG_SCHED_DEBUG */
395 {
396 .procname = "sched_rt_period_us",
397 .data = &sysctl_sched_rt_period,
398 .maxlen = sizeof(unsigned int),
399 .mode = 0644,
400 .proc_handler = sched_rt_handler,
401 },
402 {
403 .procname = "sched_rt_runtime_us",
404 .data = &sysctl_sched_rt_runtime,
405 .maxlen = sizeof(int),
406 .mode = 0644,
407 .proc_handler = sched_rt_handler,
408 },
409 {
410 .procname = "sched_rr_timeslice_ms",
411 .data = &sched_rr_timeslice,
412 .maxlen = sizeof(int),
413 .mode = 0644,
414 .proc_handler = sched_rr_handler,
415 },
416 #ifdef CONFIG_SCHED_AUTOGROUP
417 {
418 .procname = "sched_autogroup_enabled",
419 .data = &sysctl_sched_autogroup_enabled,
420 .maxlen = sizeof(unsigned int),
421 .mode = 0644,
422 .proc_handler = proc_dointvec_minmax,
423 .extra1 = &zero,
424 .extra2 = &one,
425 },
426 #endif
427 #ifdef CONFIG_CFS_BANDWIDTH
428 {
429 .procname = "sched_cfs_bandwidth_slice_us",
430 .data = &sysctl_sched_cfs_bandwidth_slice,
431 .maxlen = sizeof(unsigned int),
432 .mode = 0644,
433 .proc_handler = proc_dointvec_minmax,
434 .extra1 = &one,
435 },
436 #endif
437 #ifdef CONFIG_PROVE_LOCKING
438 {
439 .procname = "prove_locking",
440 .data = &prove_locking,
441 .maxlen = sizeof(int),
442 .mode = 0644,
443 .proc_handler = proc_dointvec,
444 },
445 #endif
446 #ifdef CONFIG_LOCK_STAT
447 {
448 .procname = "lock_stat",
449 .data = &lock_stat,
450 .maxlen = sizeof(int),
451 .mode = 0644,
452 .proc_handler = proc_dointvec,
453 },
454 #endif
455 {
456 .procname = "panic",
457 .data = &panic_timeout,
458 .maxlen = sizeof(int),
459 .mode = 0644,
460 .proc_handler = proc_dointvec,
461 },
462 #ifdef CONFIG_COREDUMP
463 {
464 .procname = "core_uses_pid",
465 .data = &core_uses_pid,
466 .maxlen = sizeof(int),
467 .mode = 0644,
468 .proc_handler = proc_dointvec,
469 },
470 {
471 .procname = "core_pattern",
472 .data = core_pattern,
473 .maxlen = CORENAME_MAX_SIZE,
474 .mode = 0644,
475 .proc_handler = proc_dostring_coredump,
476 },
477 {
478 .procname = "core_pipe_limit",
479 .data = &core_pipe_limit,
480 .maxlen = sizeof(unsigned int),
481 .mode = 0644,
482 .proc_handler = proc_dointvec,
483 },
484 #endif
485 #ifdef CONFIG_PROC_SYSCTL
486 {
487 .procname = "tainted",
488 .maxlen = sizeof(long),
489 .mode = 0644,
490 .proc_handler = proc_taint,
491 },
492 {
493 .procname = "sysctl_writes_strict",
494 .data = &sysctl_writes_strict,
495 .maxlen = sizeof(int),
496 .mode = 0644,
497 .proc_handler = proc_dointvec_minmax,
498 .extra1 = &neg_one,
499 .extra2 = &one,
500 },
501 #endif
502 #ifdef CONFIG_LATENCYTOP
503 {
504 .procname = "latencytop",
505 .data = &latencytop_enabled,
506 .maxlen = sizeof(int),
507 .mode = 0644,
508 .proc_handler = proc_dointvec,
509 },
510 #endif
511 #ifdef CONFIG_BLK_DEV_INITRD
512 {
513 .procname = "real-root-dev",
514 .data = &real_root_dev,
515 .maxlen = sizeof(int),
516 .mode = 0644,
517 .proc_handler = proc_dointvec,
518 },
519 #endif
520 {
521 .procname = "print-fatal-signals",
522 .data = &print_fatal_signals,
523 .maxlen = sizeof(int),
524 .mode = 0644,
525 .proc_handler = proc_dointvec,
526 },
527 #ifdef CONFIG_SPARC
528 {
529 .procname = "reboot-cmd",
530 .data = reboot_command,
531 .maxlen = 256,
532 .mode = 0644,
533 .proc_handler = proc_dostring,
534 },
535 {
536 .procname = "stop-a",
537 .data = &stop_a_enabled,
538 .maxlen = sizeof (int),
539 .mode = 0644,
540 .proc_handler = proc_dointvec,
541 },
542 {
543 .procname = "scons-poweroff",
544 .data = &scons_pwroff,
545 .maxlen = sizeof (int),
546 .mode = 0644,
547 .proc_handler = proc_dointvec,
548 },
549 #endif
550 #ifdef CONFIG_SPARC64
551 {
552 .procname = "tsb-ratio",
553 .data = &sysctl_tsb_ratio,
554 .maxlen = sizeof (int),
555 .mode = 0644,
556 .proc_handler = proc_dointvec,
557 },
558 #endif
559 #ifdef __hppa__
560 {
561 .procname = "soft-power",
562 .data = &pwrsw_enabled,
563 .maxlen = sizeof (int),
564 .mode = 0644,
565 .proc_handler = proc_dointvec,
566 },
567 #endif
568 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
569 {
570 .procname = "unaligned-trap",
571 .data = &unaligned_enabled,
572 .maxlen = sizeof (int),
573 .mode = 0644,
574 .proc_handler = proc_dointvec,
575 },
576 #endif
577 {
578 .procname = "ctrl-alt-del",
579 .data = &C_A_D,
580 .maxlen = sizeof(int),
581 .mode = 0644,
582 .proc_handler = proc_dointvec,
583 },
584 #ifdef CONFIG_FUNCTION_TRACER
585 {
586 .procname = "ftrace_enabled",
587 .data = &ftrace_enabled,
588 .maxlen = sizeof(int),
589 .mode = 0644,
590 .proc_handler = ftrace_enable_sysctl,
591 },
592 #endif
593 #ifdef CONFIG_STACK_TRACER
594 {
595 .procname = "stack_tracer_enabled",
596 .data = &stack_tracer_enabled,
597 .maxlen = sizeof(int),
598 .mode = 0644,
599 .proc_handler = stack_trace_sysctl,
600 },
601 #endif
602 #ifdef CONFIG_TRACING
603 {
604 .procname = "ftrace_dump_on_oops",
605 .data = &ftrace_dump_on_oops,
606 .maxlen = sizeof(int),
607 .mode = 0644,
608 .proc_handler = proc_dointvec,
609 },
610 {
611 .procname = "traceoff_on_warning",
612 .data = &__disable_trace_on_warning,
613 .maxlen = sizeof(__disable_trace_on_warning),
614 .mode = 0644,
615 .proc_handler = proc_dointvec,
616 },
617 {
618 .procname = "tracepoint_printk",
619 .data = &tracepoint_printk,
620 .maxlen = sizeof(tracepoint_printk),
621 .mode = 0644,
622 .proc_handler = proc_dointvec,
623 },
624 #endif
625 #ifdef CONFIG_KEXEC_CORE
626 {
627 .procname = "kexec_load_disabled",
628 .data = &kexec_load_disabled,
629 .maxlen = sizeof(int),
630 .mode = 0644,
631 /* only handle a transition from default "0" to "1" */
632 .proc_handler = proc_dointvec_minmax,
633 .extra1 = &one,
634 .extra2 = &one,
635 },
636 #endif
637 #ifdef CONFIG_MODULES
638 {
639 .procname = "modprobe",
640 .data = &modprobe_path,
641 .maxlen = KMOD_PATH_LEN,
642 .mode = 0644,
643 .proc_handler = proc_dostring,
644 },
645 {
646 .procname = "modules_disabled",
647 .data = &modules_disabled,
648 .maxlen = sizeof(int),
649 .mode = 0644,
650 /* only handle a transition from default "0" to "1" */
651 .proc_handler = proc_dointvec_minmax,
652 .extra1 = &one,
653 .extra2 = &one,
654 },
655 #endif
656 #ifdef CONFIG_UEVENT_HELPER
657 {
658 .procname = "hotplug",
659 .data = &uevent_helper,
660 .maxlen = UEVENT_HELPER_PATH_LEN,
661 .mode = 0644,
662 .proc_handler = proc_dostring,
663 },
664 #endif
665 #ifdef CONFIG_CHR_DEV_SG
666 {
667 .procname = "sg-big-buff",
668 .data = &sg_big_buff,
669 .maxlen = sizeof (int),
670 .mode = 0444,
671 .proc_handler = proc_dointvec,
672 },
673 #endif
674 #ifdef CONFIG_BSD_PROCESS_ACCT
675 {
676 .procname = "acct",
677 .data = &acct_parm,
678 .maxlen = 3*sizeof(int),
679 .mode = 0644,
680 .proc_handler = proc_dointvec,
681 },
682 #endif
683 #ifdef CONFIG_MAGIC_SYSRQ
684 {
685 .procname = "sysrq",
686 .data = &__sysrq_enabled,
687 .maxlen = sizeof (int),
688 .mode = 0644,
689 .proc_handler = sysrq_sysctl_handler,
690 },
691 #endif
692 #ifdef CONFIG_PROC_SYSCTL
693 {
694 .procname = "cad_pid",
695 .data = NULL,
696 .maxlen = sizeof (int),
697 .mode = 0600,
698 .proc_handler = proc_do_cad_pid,
699 },
700 #endif
701 {
702 .procname = "threads-max",
703 .data = NULL,
704 .maxlen = sizeof(int),
705 .mode = 0644,
706 .proc_handler = sysctl_max_threads,
707 },
708 {
709 .procname = "random",
710 .mode = 0555,
711 .child = random_table,
712 },
713 {
714 .procname = "usermodehelper",
715 .mode = 0555,
716 .child = usermodehelper_table,
717 },
718 {
719 .procname = "overflowuid",
720 .data = &overflowuid,
721 .maxlen = sizeof(int),
722 .mode = 0644,
723 .proc_handler = proc_dointvec_minmax,
724 .extra1 = &minolduid,
725 .extra2 = &maxolduid,
726 },
727 {
728 .procname = "overflowgid",
729 .data = &overflowgid,
730 .maxlen = sizeof(int),
731 .mode = 0644,
732 .proc_handler = proc_dointvec_minmax,
733 .extra1 = &minolduid,
734 .extra2 = &maxolduid,
735 },
736 #ifdef CONFIG_S390
737 #ifdef CONFIG_MATHEMU
738 {
739 .procname = "ieee_emulation_warnings",
740 .data = &sysctl_ieee_emulation_warnings,
741 .maxlen = sizeof(int),
742 .mode = 0644,
743 .proc_handler = proc_dointvec,
744 },
745 #endif
746 {
747 .procname = "userprocess_debug",
748 .data = &show_unhandled_signals,
749 .maxlen = sizeof(int),
750 .mode = 0644,
751 .proc_handler = proc_dointvec,
752 },
753 #endif
754 {
755 .procname = "pid_max",
756 .data = &pid_max,
757 .maxlen = sizeof (int),
758 .mode = 0644,
759 .proc_handler = proc_dointvec_minmax,
760 .extra1 = &pid_max_min,
761 .extra2 = &pid_max_max,
762 },
763 {
764 .procname = "panic_on_oops",
765 .data = &panic_on_oops,
766 .maxlen = sizeof(int),
767 .mode = 0644,
768 .proc_handler = proc_dointvec,
769 },
770 #if defined CONFIG_PRINTK
771 {
772 .procname = "printk",
773 .data = &console_loglevel,
774 .maxlen = 4*sizeof(int),
775 .mode = 0644,
776 .proc_handler = proc_dointvec,
777 },
778 {
779 .procname = "printk_ratelimit",
780 .data = &printk_ratelimit_state.interval,
781 .maxlen = sizeof(int),
782 .mode = 0644,
783 .proc_handler = proc_dointvec_jiffies,
784 },
785 {
786 .procname = "printk_ratelimit_burst",
787 .data = &printk_ratelimit_state.burst,
788 .maxlen = sizeof(int),
789 .mode = 0644,
790 .proc_handler = proc_dointvec,
791 },
792 {
793 .procname = "printk_delay",
794 .data = &printk_delay_msec,
795 .maxlen = sizeof(int),
796 .mode = 0644,
797 .proc_handler = proc_dointvec_minmax,
798 .extra1 = &zero,
799 .extra2 = &ten_thousand,
800 },
801 {
802 .procname = "dmesg_restrict",
803 .data = &dmesg_restrict,
804 .maxlen = sizeof(int),
805 .mode = 0644,
806 .proc_handler = proc_dointvec_minmax_sysadmin,
807 .extra1 = &zero,
808 .extra2 = &one,
809 },
810 {
811 .procname = "kptr_restrict",
812 .data = &kptr_restrict,
813 .maxlen = sizeof(int),
814 .mode = 0644,
815 .proc_handler = proc_dointvec_minmax_sysadmin,
816 .extra1 = &zero,
817 .extra2 = &two,
818 },
819 #endif
820 {
821 .procname = "ngroups_max",
822 .data = &ngroups_max,
823 .maxlen = sizeof (int),
824 .mode = 0444,
825 .proc_handler = proc_dointvec,
826 },
827 {
828 .procname = "cap_last_cap",
829 .data = (void *)&cap_last_cap,
830 .maxlen = sizeof(int),
831 .mode = 0444,
832 .proc_handler = proc_dointvec,
833 },
834 #if defined(CONFIG_LOCKUP_DETECTOR)
835 {
836 .procname = "watchdog",
837 .data = &watchdog_user_enabled,
838 .maxlen = sizeof (int),
839 .mode = 0644,
840 .proc_handler = proc_watchdog,
841 .extra1 = &zero,
842 .extra2 = &one,
843 },
844 {
845 .procname = "watchdog_thresh",
846 .data = &watchdog_thresh,
847 .maxlen = sizeof(int),
848 .mode = 0644,
849 .proc_handler = proc_watchdog_thresh,
850 .extra1 = &zero,
851 .extra2 = &sixty,
852 },
853 {
854 .procname = "nmi_watchdog",
855 .data = &nmi_watchdog_enabled,
856 .maxlen = sizeof (int),
857 .mode = 0644,
858 .proc_handler = proc_nmi_watchdog,
859 .extra1 = &zero,
860 #if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR)
861 .extra2 = &one,
862 #else
863 .extra2 = &zero,
864 #endif
865 },
866 {
867 .procname = "soft_watchdog",
868 .data = &soft_watchdog_enabled,
869 .maxlen = sizeof (int),
870 .mode = 0644,
871 .proc_handler = proc_soft_watchdog,
872 .extra1 = &zero,
873 .extra2 = &one,
874 },
875 {
876 .procname = "watchdog_cpumask",
877 .data = &watchdog_cpumask_bits,
878 .maxlen = NR_CPUS,
879 .mode = 0644,
880 .proc_handler = proc_watchdog_cpumask,
881 },
882 {
883 .procname = "softlockup_panic",
884 .data = &softlockup_panic,
885 .maxlen = sizeof(int),
886 .mode = 0644,
887 .proc_handler = proc_dointvec_minmax,
888 .extra1 = &zero,
889 .extra2 = &one,
890 },
891 #ifdef CONFIG_SMP
892 {
893 .procname = "softlockup_all_cpu_backtrace",
894 .data = &sysctl_softlockup_all_cpu_backtrace,
895 .maxlen = sizeof(int),
896 .mode = 0644,
897 .proc_handler = proc_dointvec_minmax,
898 .extra1 = &zero,
899 .extra2 = &one,
900 },
901 {
902 .procname = "hardlockup_all_cpu_backtrace",
903 .data = &sysctl_hardlockup_all_cpu_backtrace,
904 .maxlen = sizeof(int),
905 .mode = 0644,
906 .proc_handler = proc_dointvec_minmax,
907 .extra1 = &zero,
908 .extra2 = &one,
909 },
910 #endif /* CONFIG_SMP */
911 #endif
912 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
913 {
914 .procname = "unknown_nmi_panic",
915 .data = &unknown_nmi_panic,
916 .maxlen = sizeof (int),
917 .mode = 0644,
918 .proc_handler = proc_dointvec,
919 },
920 #endif
921 #if defined(CONFIG_X86)
922 {
923 .procname = "panic_on_unrecovered_nmi",
924 .data = &panic_on_unrecovered_nmi,
925 .maxlen = sizeof(int),
926 .mode = 0644,
927 .proc_handler = proc_dointvec,
928 },
929 {
930 .procname = "panic_on_io_nmi",
931 .data = &panic_on_io_nmi,
932 .maxlen = sizeof(int),
933 .mode = 0644,
934 .proc_handler = proc_dointvec,
935 },
936 #ifdef CONFIG_DEBUG_STACKOVERFLOW
937 {
938 .procname = "panic_on_stackoverflow",
939 .data = &sysctl_panic_on_stackoverflow,
940 .maxlen = sizeof(int),
941 .mode = 0644,
942 .proc_handler = proc_dointvec,
943 },
944 #endif
945 {
946 .procname = "bootloader_type",
947 .data = &bootloader_type,
948 .maxlen = sizeof (int),
949 .mode = 0444,
950 .proc_handler = proc_dointvec,
951 },
952 {
953 .procname = "bootloader_version",
954 .data = &bootloader_version,
955 .maxlen = sizeof (int),
956 .mode = 0444,
957 .proc_handler = proc_dointvec,
958 },
959 {
960 .procname = "kstack_depth_to_print",
961 .data = &kstack_depth_to_print,
962 .maxlen = sizeof(int),
963 .mode = 0644,
964 .proc_handler = proc_dointvec,
965 },
966 {
967 .procname = "io_delay_type",
968 .data = &io_delay_type,
969 .maxlen = sizeof(int),
970 .mode = 0644,
971 .proc_handler = proc_dointvec,
972 },
973 #endif
974 #if defined(CONFIG_MMU)
975 {
976 .procname = "randomize_va_space",
977 .data = &randomize_va_space,
978 .maxlen = sizeof(int),
979 .mode = 0644,
980 .proc_handler = proc_dointvec,
981 },
982 #endif
983 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
984 {
985 .procname = "spin_retry",
986 .data = &spin_retry,
987 .maxlen = sizeof (int),
988 .mode = 0644,
989 .proc_handler = proc_dointvec,
990 },
991 #endif
992 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
993 {
994 .procname = "acpi_video_flags",
995 .data = &acpi_realmode_flags,
996 .maxlen = sizeof (unsigned long),
997 .mode = 0644,
998 .proc_handler = proc_doulongvec_minmax,
999 },
1000 #endif
1001 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1002 {
1003 .procname = "ignore-unaligned-usertrap",
1004 .data = &no_unaligned_warning,
1005 .maxlen = sizeof (int),
1006 .mode = 0644,
1007 .proc_handler = proc_dointvec,
1008 },
1009 #endif
1010 #ifdef CONFIG_IA64
1011 {
1012 .procname = "unaligned-dump-stack",
1013 .data = &unaligned_dump_stack,
1014 .maxlen = sizeof (int),
1015 .mode = 0644,
1016 .proc_handler = proc_dointvec,
1017 },
1018 #endif
1019 #ifdef CONFIG_DETECT_HUNG_TASK
1020 {
1021 .procname = "hung_task_panic",
1022 .data = &sysctl_hung_task_panic,
1023 .maxlen = sizeof(int),
1024 .mode = 0644,
1025 .proc_handler = proc_dointvec_minmax,
1026 .extra1 = &zero,
1027 .extra2 = &one,
1028 },
1029 {
1030 .procname = "hung_task_check_count",
1031 .data = &sysctl_hung_task_check_count,
1032 .maxlen = sizeof(int),
1033 .mode = 0644,
1034 .proc_handler = proc_dointvec_minmax,
1035 .extra1 = &zero,
1036 },
1037 {
1038 .procname = "hung_task_timeout_secs",
1039 .data = &sysctl_hung_task_timeout_secs,
1040 .maxlen = sizeof(unsigned long),
1041 .mode = 0644,
1042 .proc_handler = proc_dohung_task_timeout_secs,
1043 .extra2 = &hung_task_timeout_max,
1044 },
1045 {
1046 .procname = "hung_task_warnings",
1047 .data = &sysctl_hung_task_warnings,
1048 .maxlen = sizeof(int),
1049 .mode = 0644,
1050 .proc_handler = proc_dointvec_minmax,
1051 .extra1 = &neg_one,
1052 },
1053 #endif
1054 #ifdef CONFIG_COMPAT
1055 {
1056 .procname = "compat-log",
1057 .data = &compat_log,
1058 .maxlen = sizeof (int),
1059 .mode = 0644,
1060 .proc_handler = proc_dointvec,
1061 },
1062 #endif
1063 #ifdef CONFIG_RT_MUTEXES
1064 {
1065 .procname = "max_lock_depth",
1066 .data = &max_lock_depth,
1067 .maxlen = sizeof(int),
1068 .mode = 0644,
1069 .proc_handler = proc_dointvec,
1070 },
1071 #endif
1072 {
1073 .procname = "poweroff_cmd",
1074 .data = &poweroff_cmd,
1075 .maxlen = POWEROFF_CMD_PATH_LEN,
1076 .mode = 0644,
1077 .proc_handler = proc_dostring,
1078 },
1079 #ifdef CONFIG_KEYS
1080 {
1081 .procname = "keys",
1082 .mode = 0555,
1083 .child = key_sysctls,
1084 },
1085 #endif
1086 #ifdef CONFIG_PERF_EVENTS
1087 /*
1088 * User-space scripts rely on the existence of this file
1089 * as a feature check for perf_events being enabled.
1090 *
1091 * So it's an ABI, do not remove!
1092 */
1093 {
1094 .procname = "perf_event_paranoid",
1095 .data = &sysctl_perf_event_paranoid,
1096 .maxlen = sizeof(sysctl_perf_event_paranoid),
1097 .mode = 0644,
1098 .proc_handler = proc_dointvec,
1099 },
1100 {
1101 .procname = "perf_event_mlock_kb",
1102 .data = &sysctl_perf_event_mlock,
1103 .maxlen = sizeof(sysctl_perf_event_mlock),
1104 .mode = 0644,
1105 .proc_handler = proc_dointvec,
1106 },
1107 {
1108 .procname = "perf_event_max_sample_rate",
1109 .data = &sysctl_perf_event_sample_rate,
1110 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1111 .mode = 0644,
1112 .proc_handler = perf_proc_update_handler,
1113 .extra1 = &one,
1114 },
1115 {
1116 .procname = "perf_cpu_time_max_percent",
1117 .data = &sysctl_perf_cpu_time_max_percent,
1118 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1119 .mode = 0644,
1120 .proc_handler = perf_cpu_time_max_percent_handler,
1121 .extra1 = &zero,
1122 .extra2 = &one_hundred,
1123 },
1124 #endif
1125 #ifdef CONFIG_KMEMCHECK
1126 {
1127 .procname = "kmemcheck",
1128 .data = &kmemcheck_enabled,
1129 .maxlen = sizeof(int),
1130 .mode = 0644,
1131 .proc_handler = proc_dointvec,
1132 },
1133 #endif
1134 {
1135 .procname = "panic_on_warn",
1136 .data = &panic_on_warn,
1137 .maxlen = sizeof(int),
1138 .mode = 0644,
1139 .proc_handler = proc_dointvec_minmax,
1140 .extra1 = &zero,
1141 .extra2 = &one,
1142 },
1143 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1144 {
1145 .procname = "timer_migration",
1146 .data = &sysctl_timer_migration,
1147 .maxlen = sizeof(unsigned int),
1148 .mode = 0644,
1149 .proc_handler = timer_migration_handler,
1150 },
1151 #endif
1152 #ifdef CONFIG_BPF_SYSCALL
1153 {
1154 .procname = "unprivileged_bpf_disabled",
1155 .data = &sysctl_unprivileged_bpf_disabled,
1156 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
1157 .mode = 0644,
1158 /* only handle a transition from default "0" to "1" */
1159 .proc_handler = proc_dointvec_minmax,
1160 .extra1 = &one,
1161 .extra2 = &one,
1162 },
1163 #endif
1164 { }
1165 };
1166
1167 static struct ctl_table vm_table[] = {
1168 {
1169 .procname = "overcommit_memory",
1170 .data = &sysctl_overcommit_memory,
1171 .maxlen = sizeof(sysctl_overcommit_memory),
1172 .mode = 0644,
1173 .proc_handler = proc_dointvec_minmax,
1174 .extra1 = &zero,
1175 .extra2 = &two,
1176 },
1177 {
1178 .procname = "panic_on_oom",
1179 .data = &sysctl_panic_on_oom,
1180 .maxlen = sizeof(sysctl_panic_on_oom),
1181 .mode = 0644,
1182 .proc_handler = proc_dointvec_minmax,
1183 .extra1 = &zero,
1184 .extra2 = &two,
1185 },
1186 {
1187 .procname = "oom_kill_allocating_task",
1188 .data = &sysctl_oom_kill_allocating_task,
1189 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1190 .mode = 0644,
1191 .proc_handler = proc_dointvec,
1192 },
1193 {
1194 .procname = "oom_dump_tasks",
1195 .data = &sysctl_oom_dump_tasks,
1196 .maxlen = sizeof(sysctl_oom_dump_tasks),
1197 .mode = 0644,
1198 .proc_handler = proc_dointvec,
1199 },
1200 {
1201 .procname = "overcommit_ratio",
1202 .data = &sysctl_overcommit_ratio,
1203 .maxlen = sizeof(sysctl_overcommit_ratio),
1204 .mode = 0644,
1205 .proc_handler = overcommit_ratio_handler,
1206 },
1207 {
1208 .procname = "overcommit_kbytes",
1209 .data = &sysctl_overcommit_kbytes,
1210 .maxlen = sizeof(sysctl_overcommit_kbytes),
1211 .mode = 0644,
1212 .proc_handler = overcommit_kbytes_handler,
1213 },
1214 {
1215 .procname = "page-cluster",
1216 .data = &page_cluster,
1217 .maxlen = sizeof(int),
1218 .mode = 0644,
1219 .proc_handler = proc_dointvec_minmax,
1220 .extra1 = &zero,
1221 },
1222 {
1223 .procname = "dirty_background_ratio",
1224 .data = &dirty_background_ratio,
1225 .maxlen = sizeof(dirty_background_ratio),
1226 .mode = 0644,
1227 .proc_handler = dirty_background_ratio_handler,
1228 .extra1 = &zero,
1229 .extra2 = &one_hundred,
1230 },
1231 {
1232 .procname = "dirty_background_bytes",
1233 .data = &dirty_background_bytes,
1234 .maxlen = sizeof(dirty_background_bytes),
1235 .mode = 0644,
1236 .proc_handler = dirty_background_bytes_handler,
1237 .extra1 = &one_ul,
1238 },
1239 {
1240 .procname = "dirty_ratio",
1241 .data = &vm_dirty_ratio,
1242 .maxlen = sizeof(vm_dirty_ratio),
1243 .mode = 0644,
1244 .proc_handler = dirty_ratio_handler,
1245 .extra1 = &zero,
1246 .extra2 = &one_hundred,
1247 },
1248 {
1249 .procname = "dirty_bytes",
1250 .data = &vm_dirty_bytes,
1251 .maxlen = sizeof(vm_dirty_bytes),
1252 .mode = 0644,
1253 .proc_handler = dirty_bytes_handler,
1254 .extra1 = &dirty_bytes_min,
1255 },
1256 {
1257 .procname = "dirty_writeback_centisecs",
1258 .data = &dirty_writeback_interval,
1259 .maxlen = sizeof(dirty_writeback_interval),
1260 .mode = 0644,
1261 .proc_handler = dirty_writeback_centisecs_handler,
1262 },
1263 {
1264 .procname = "dirty_expire_centisecs",
1265 .data = &dirty_expire_interval,
1266 .maxlen = sizeof(dirty_expire_interval),
1267 .mode = 0644,
1268 .proc_handler = proc_dointvec_minmax,
1269 .extra1 = &zero,
1270 },
1271 {
1272 .procname = "dirtytime_expire_seconds",
1273 .data = &dirtytime_expire_interval,
1274 .maxlen = sizeof(dirty_expire_interval),
1275 .mode = 0644,
1276 .proc_handler = dirtytime_interval_handler,
1277 .extra1 = &zero,
1278 },
1279 {
1280 .procname = "nr_pdflush_threads",
1281 .mode = 0444 /* read-only */,
1282 .proc_handler = pdflush_proc_obsolete,
1283 },
1284 {
1285 .procname = "swappiness",
1286 .data = &vm_swappiness,
1287 .maxlen = sizeof(vm_swappiness),
1288 .mode = 0644,
1289 .proc_handler = proc_dointvec_minmax,
1290 .extra1 = &zero,
1291 .extra2 = &one_hundred,
1292 },
1293 #ifdef CONFIG_HUGETLB_PAGE
1294 {
1295 .procname = "nr_hugepages",
1296 .data = NULL,
1297 .maxlen = sizeof(unsigned long),
1298 .mode = 0644,
1299 .proc_handler = hugetlb_sysctl_handler,
1300 },
1301 #ifdef CONFIG_NUMA
1302 {
1303 .procname = "nr_hugepages_mempolicy",
1304 .data = NULL,
1305 .maxlen = sizeof(unsigned long),
1306 .mode = 0644,
1307 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1308 },
1309 #endif
1310 {
1311 .procname = "hugetlb_shm_group",
1312 .data = &sysctl_hugetlb_shm_group,
1313 .maxlen = sizeof(gid_t),
1314 .mode = 0644,
1315 .proc_handler = proc_dointvec,
1316 },
1317 {
1318 .procname = "hugepages_treat_as_movable",
1319 .data = &hugepages_treat_as_movable,
1320 .maxlen = sizeof(int),
1321 .mode = 0644,
1322 .proc_handler = proc_dointvec,
1323 },
1324 {
1325 .procname = "nr_overcommit_hugepages",
1326 .data = NULL,
1327 .maxlen = sizeof(unsigned long),
1328 .mode = 0644,
1329 .proc_handler = hugetlb_overcommit_handler,
1330 },
1331 #endif
1332 {
1333 .procname = "lowmem_reserve_ratio",
1334 .data = &sysctl_lowmem_reserve_ratio,
1335 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1336 .mode = 0644,
1337 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1338 },
1339 {
1340 .procname = "drop_caches",
1341 .data = &sysctl_drop_caches,
1342 .maxlen = sizeof(int),
1343 .mode = 0644,
1344 .proc_handler = drop_caches_sysctl_handler,
1345 .extra1 = &one,
1346 .extra2 = &four,
1347 },
1348 #ifdef CONFIG_COMPACTION
1349 {
1350 .procname = "compact_memory",
1351 .data = &sysctl_compact_memory,
1352 .maxlen = sizeof(int),
1353 .mode = 0200,
1354 .proc_handler = sysctl_compaction_handler,
1355 },
1356 {
1357 .procname = "extfrag_threshold",
1358 .data = &sysctl_extfrag_threshold,
1359 .maxlen = sizeof(int),
1360 .mode = 0644,
1361 .proc_handler = sysctl_extfrag_handler,
1362 .extra1 = &min_extfrag_threshold,
1363 .extra2 = &max_extfrag_threshold,
1364 },
1365 {
1366 .procname = "compact_unevictable_allowed",
1367 .data = &sysctl_compact_unevictable_allowed,
1368 .maxlen = sizeof(int),
1369 .mode = 0644,
1370 .proc_handler = proc_dointvec,
1371 .extra1 = &zero,
1372 .extra2 = &one,
1373 },
1374
1375 #endif /* CONFIG_COMPACTION */
1376 {
1377 .procname = "min_free_kbytes",
1378 .data = &min_free_kbytes,
1379 .maxlen = sizeof(min_free_kbytes),
1380 .mode = 0644,
1381 .proc_handler = min_free_kbytes_sysctl_handler,
1382 .extra1 = &zero,
1383 },
1384 {
1385 .procname = "percpu_pagelist_fraction",
1386 .data = &percpu_pagelist_fraction,
1387 .maxlen = sizeof(percpu_pagelist_fraction),
1388 .mode = 0644,
1389 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1390 .extra1 = &zero,
1391 },
1392 #ifdef CONFIG_MMU
1393 {
1394 .procname = "max_map_count",
1395 .data = &sysctl_max_map_count,
1396 .maxlen = sizeof(sysctl_max_map_count),
1397 .mode = 0644,
1398 .proc_handler = proc_dointvec_minmax,
1399 .extra1 = &zero,
1400 },
1401 #else
1402 {
1403 .procname = "nr_trim_pages",
1404 .data = &sysctl_nr_trim_pages,
1405 .maxlen = sizeof(sysctl_nr_trim_pages),
1406 .mode = 0644,
1407 .proc_handler = proc_dointvec_minmax,
1408 .extra1 = &zero,
1409 },
1410 #endif
1411 {
1412 .procname = "laptop_mode",
1413 .data = &laptop_mode,
1414 .maxlen = sizeof(laptop_mode),
1415 .mode = 0644,
1416 .proc_handler = proc_dointvec_jiffies,
1417 },
1418 {
1419 .procname = "block_dump",
1420 .data = &block_dump,
1421 .maxlen = sizeof(block_dump),
1422 .mode = 0644,
1423 .proc_handler = proc_dointvec,
1424 .extra1 = &zero,
1425 },
1426 {
1427 .procname = "vfs_cache_pressure",
1428 .data = &sysctl_vfs_cache_pressure,
1429 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1430 .mode = 0644,
1431 .proc_handler = proc_dointvec,
1432 .extra1 = &zero,
1433 },
1434 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1435 {
1436 .procname = "legacy_va_layout",
1437 .data = &sysctl_legacy_va_layout,
1438 .maxlen = sizeof(sysctl_legacy_va_layout),
1439 .mode = 0644,
1440 .proc_handler = proc_dointvec,
1441 .extra1 = &zero,
1442 },
1443 #endif
1444 #ifdef CONFIG_NUMA
1445 {
1446 .procname = "zone_reclaim_mode",
1447 .data = &zone_reclaim_mode,
1448 .maxlen = sizeof(zone_reclaim_mode),
1449 .mode = 0644,
1450 .proc_handler = proc_dointvec,
1451 .extra1 = &zero,
1452 },
1453 {
1454 .procname = "min_unmapped_ratio",
1455 .data = &sysctl_min_unmapped_ratio,
1456 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1457 .mode = 0644,
1458 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1459 .extra1 = &zero,
1460 .extra2 = &one_hundred,
1461 },
1462 {
1463 .procname = "min_slab_ratio",
1464 .data = &sysctl_min_slab_ratio,
1465 .maxlen = sizeof(sysctl_min_slab_ratio),
1466 .mode = 0644,
1467 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1468 .extra1 = &zero,
1469 .extra2 = &one_hundred,
1470 },
1471 #endif
1472 #ifdef CONFIG_SMP
1473 {
1474 .procname = "stat_interval",
1475 .data = &sysctl_stat_interval,
1476 .maxlen = sizeof(sysctl_stat_interval),
1477 .mode = 0644,
1478 .proc_handler = proc_dointvec_jiffies,
1479 },
1480 #endif
1481 #ifdef CONFIG_MMU
1482 {
1483 .procname = "mmap_min_addr",
1484 .data = &dac_mmap_min_addr,
1485 .maxlen = sizeof(unsigned long),
1486 .mode = 0644,
1487 .proc_handler = mmap_min_addr_handler,
1488 },
1489 #endif
1490 #ifdef CONFIG_NUMA
1491 {
1492 .procname = "numa_zonelist_order",
1493 .data = &numa_zonelist_order,
1494 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1495 .mode = 0644,
1496 .proc_handler = numa_zonelist_order_handler,
1497 },
1498 #endif
1499 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1500 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1501 {
1502 .procname = "vdso_enabled",
1503 #ifdef CONFIG_X86_32
1504 .data = &vdso32_enabled,
1505 .maxlen = sizeof(vdso32_enabled),
1506 #else
1507 .data = &vdso_enabled,
1508 .maxlen = sizeof(vdso_enabled),
1509 #endif
1510 .mode = 0644,
1511 .proc_handler = proc_dointvec,
1512 .extra1 = &zero,
1513 },
1514 #endif
1515 #ifdef CONFIG_HIGHMEM
1516 {
1517 .procname = "highmem_is_dirtyable",
1518 .data = &vm_highmem_is_dirtyable,
1519 .maxlen = sizeof(vm_highmem_is_dirtyable),
1520 .mode = 0644,
1521 .proc_handler = proc_dointvec_minmax,
1522 .extra1 = &zero,
1523 .extra2 = &one,
1524 },
1525 #endif
1526 #ifdef CONFIG_MEMORY_FAILURE
1527 {
1528 .procname = "memory_failure_early_kill",
1529 .data = &sysctl_memory_failure_early_kill,
1530 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1531 .mode = 0644,
1532 .proc_handler = proc_dointvec_minmax,
1533 .extra1 = &zero,
1534 .extra2 = &one,
1535 },
1536 {
1537 .procname = "memory_failure_recovery",
1538 .data = &sysctl_memory_failure_recovery,
1539 .maxlen = sizeof(sysctl_memory_failure_recovery),
1540 .mode = 0644,
1541 .proc_handler = proc_dointvec_minmax,
1542 .extra1 = &zero,
1543 .extra2 = &one,
1544 },
1545 #endif
1546 {
1547 .procname = "user_reserve_kbytes",
1548 .data = &sysctl_user_reserve_kbytes,
1549 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1550 .mode = 0644,
1551 .proc_handler = proc_doulongvec_minmax,
1552 },
1553 {
1554 .procname = "admin_reserve_kbytes",
1555 .data = &sysctl_admin_reserve_kbytes,
1556 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1557 .mode = 0644,
1558 .proc_handler = proc_doulongvec_minmax,
1559 },
1560 { }
1561 };
1562
1563 static struct ctl_table fs_table[] = {
1564 {
1565 .procname = "inode-nr",
1566 .data = &inodes_stat,
1567 .maxlen = 2*sizeof(long),
1568 .mode = 0444,
1569 .proc_handler = proc_nr_inodes,
1570 },
1571 {
1572 .procname = "inode-state",
1573 .data = &inodes_stat,
1574 .maxlen = 7*sizeof(long),
1575 .mode = 0444,
1576 .proc_handler = proc_nr_inodes,
1577 },
1578 {
1579 .procname = "file-nr",
1580 .data = &files_stat,
1581 .maxlen = sizeof(files_stat),
1582 .mode = 0444,
1583 .proc_handler = proc_nr_files,
1584 },
1585 {
1586 .procname = "file-max",
1587 .data = &files_stat.max_files,
1588 .maxlen = sizeof(files_stat.max_files),
1589 .mode = 0644,
1590 .proc_handler = proc_doulongvec_minmax,
1591 },
1592 {
1593 .procname = "nr_open",
1594 .data = &sysctl_nr_open,
1595 .maxlen = sizeof(int),
1596 .mode = 0644,
1597 .proc_handler = proc_dointvec_minmax,
1598 .extra1 = &sysctl_nr_open_min,
1599 .extra2 = &sysctl_nr_open_max,
1600 },
1601 {
1602 .procname = "dentry-state",
1603 .data = &dentry_stat,
1604 .maxlen = 6*sizeof(long),
1605 .mode = 0444,
1606 .proc_handler = proc_nr_dentry,
1607 },
1608 {
1609 .procname = "overflowuid",
1610 .data = &fs_overflowuid,
1611 .maxlen = sizeof(int),
1612 .mode = 0644,
1613 .proc_handler = proc_dointvec_minmax,
1614 .extra1 = &minolduid,
1615 .extra2 = &maxolduid,
1616 },
1617 {
1618 .procname = "overflowgid",
1619 .data = &fs_overflowgid,
1620 .maxlen = sizeof(int),
1621 .mode = 0644,
1622 .proc_handler = proc_dointvec_minmax,
1623 .extra1 = &minolduid,
1624 .extra2 = &maxolduid,
1625 },
1626 #ifdef CONFIG_FILE_LOCKING
1627 {
1628 .procname = "leases-enable",
1629 .data = &leases_enable,
1630 .maxlen = sizeof(int),
1631 .mode = 0644,
1632 .proc_handler = proc_dointvec,
1633 },
1634 #endif
1635 #ifdef CONFIG_DNOTIFY
1636 {
1637 .procname = "dir-notify-enable",
1638 .data = &dir_notify_enable,
1639 .maxlen = sizeof(int),
1640 .mode = 0644,
1641 .proc_handler = proc_dointvec,
1642 },
1643 #endif
1644 #ifdef CONFIG_MMU
1645 #ifdef CONFIG_FILE_LOCKING
1646 {
1647 .procname = "lease-break-time",
1648 .data = &lease_break_time,
1649 .maxlen = sizeof(int),
1650 .mode = 0644,
1651 .proc_handler = proc_dointvec,
1652 },
1653 #endif
1654 #ifdef CONFIG_AIO
1655 {
1656 .procname = "aio-nr",
1657 .data = &aio_nr,
1658 .maxlen = sizeof(aio_nr),
1659 .mode = 0444,
1660 .proc_handler = proc_doulongvec_minmax,
1661 },
1662 {
1663 .procname = "aio-max-nr",
1664 .data = &aio_max_nr,
1665 .maxlen = sizeof(aio_max_nr),
1666 .mode = 0644,
1667 .proc_handler = proc_doulongvec_minmax,
1668 },
1669 #endif /* CONFIG_AIO */
1670 #ifdef CONFIG_INOTIFY_USER
1671 {
1672 .procname = "inotify",
1673 .mode = 0555,
1674 .child = inotify_table,
1675 },
1676 #endif
1677 #ifdef CONFIG_EPOLL
1678 {
1679 .procname = "epoll",
1680 .mode = 0555,
1681 .child = epoll_table,
1682 },
1683 #endif
1684 #endif
1685 {
1686 .procname = "protected_symlinks",
1687 .data = &sysctl_protected_symlinks,
1688 .maxlen = sizeof(int),
1689 .mode = 0600,
1690 .proc_handler = proc_dointvec_minmax,
1691 .extra1 = &zero,
1692 .extra2 = &one,
1693 },
1694 {
1695 .procname = "protected_hardlinks",
1696 .data = &sysctl_protected_hardlinks,
1697 .maxlen = sizeof(int),
1698 .mode = 0600,
1699 .proc_handler = proc_dointvec_minmax,
1700 .extra1 = &zero,
1701 .extra2 = &one,
1702 },
1703 {
1704 .procname = "suid_dumpable",
1705 .data = &suid_dumpable,
1706 .maxlen = sizeof(int),
1707 .mode = 0644,
1708 .proc_handler = proc_dointvec_minmax_coredump,
1709 .extra1 = &zero,
1710 .extra2 = &two,
1711 },
1712 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1713 {
1714 .procname = "binfmt_misc",
1715 .mode = 0555,
1716 .child = sysctl_mount_point,
1717 },
1718 #endif
1719 {
1720 .procname = "pipe-max-size",
1721 .data = &pipe_max_size,
1722 .maxlen = sizeof(int),
1723 .mode = 0644,
1724 .proc_handler = &pipe_proc_fn,
1725 .extra1 = &pipe_min_size,
1726 },
1727 { }
1728 };
1729
1730 static struct ctl_table debug_table[] = {
1731 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1732 {
1733 .procname = "exception-trace",
1734 .data = &show_unhandled_signals,
1735 .maxlen = sizeof(int),
1736 .mode = 0644,
1737 .proc_handler = proc_dointvec
1738 },
1739 #endif
1740 #if defined(CONFIG_OPTPROBES)
1741 {
1742 .procname = "kprobes-optimization",
1743 .data = &sysctl_kprobes_optimization,
1744 .maxlen = sizeof(int),
1745 .mode = 0644,
1746 .proc_handler = proc_kprobes_optimization_handler,
1747 .extra1 = &zero,
1748 .extra2 = &one,
1749 },
1750 #endif
1751 { }
1752 };
1753
1754 static struct ctl_table dev_table[] = {
1755 { }
1756 };
1757
1758 int __init sysctl_init(void)
1759 {
1760 struct ctl_table_header *hdr;
1761
1762 hdr = register_sysctl_table(sysctl_base_table);
1763 kmemleak_not_leak(hdr);
1764 return 0;
1765 }
1766
1767 #endif /* CONFIG_SYSCTL */
1768
1769 /*
1770 * /proc/sys support
1771 */
1772
1773 #ifdef CONFIG_PROC_SYSCTL
1774
1775 static int _proc_do_string(char *data, int maxlen, int write,
1776 char __user *buffer,
1777 size_t *lenp, loff_t *ppos)
1778 {
1779 size_t len;
1780 char __user *p;
1781 char c;
1782
1783 if (!data || !maxlen || !*lenp) {
1784 *lenp = 0;
1785 return 0;
1786 }
1787
1788 if (write) {
1789 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1790 /* Only continue writes not past the end of buffer. */
1791 len = strlen(data);
1792 if (len > maxlen - 1)
1793 len = maxlen - 1;
1794
1795 if (*ppos > len)
1796 return 0;
1797 len = *ppos;
1798 } else {
1799 /* Start writing from beginning of buffer. */
1800 len = 0;
1801 }
1802
1803 *ppos += *lenp;
1804 p = buffer;
1805 while ((p - buffer) < *lenp && len < maxlen - 1) {
1806 if (get_user(c, p++))
1807 return -EFAULT;
1808 if (c == 0 || c == '\n')
1809 break;
1810 data[len++] = c;
1811 }
1812 data[len] = 0;
1813 } else {
1814 len = strlen(data);
1815 if (len > maxlen)
1816 len = maxlen;
1817
1818 if (*ppos > len) {
1819 *lenp = 0;
1820 return 0;
1821 }
1822
1823 data += *ppos;
1824 len -= *ppos;
1825
1826 if (len > *lenp)
1827 len = *lenp;
1828 if (len)
1829 if (copy_to_user(buffer, data, len))
1830 return -EFAULT;
1831 if (len < *lenp) {
1832 if (put_user('\n', buffer + len))
1833 return -EFAULT;
1834 len++;
1835 }
1836 *lenp = len;
1837 *ppos += len;
1838 }
1839 return 0;
1840 }
1841
1842 static void warn_sysctl_write(struct ctl_table *table)
1843 {
1844 pr_warn_once("%s wrote to %s when file position was not 0!\n"
1845 "This will not be supported in the future. To silence this\n"
1846 "warning, set kernel.sysctl_writes_strict = -1\n",
1847 current->comm, table->procname);
1848 }
1849
1850 /**
1851 * proc_dostring - read a string sysctl
1852 * @table: the sysctl table
1853 * @write: %TRUE if this is a write to the sysctl file
1854 * @buffer: the user buffer
1855 * @lenp: the size of the user buffer
1856 * @ppos: file position
1857 *
1858 * Reads/writes a string from/to the user buffer. If the kernel
1859 * buffer provided is not large enough to hold the string, the
1860 * string is truncated. The copied string is %NULL-terminated.
1861 * If the string is being read by the user process, it is copied
1862 * and a newline '\n' is added. It is truncated if the buffer is
1863 * not large enough.
1864 *
1865 * Returns 0 on success.
1866 */
1867 int proc_dostring(struct ctl_table *table, int write,
1868 void __user *buffer, size_t *lenp, loff_t *ppos)
1869 {
1870 if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN)
1871 warn_sysctl_write(table);
1872
1873 return _proc_do_string((char *)(table->data), table->maxlen, write,
1874 (char __user *)buffer, lenp, ppos);
1875 }
1876
1877 static size_t proc_skip_spaces(char **buf)
1878 {
1879 size_t ret;
1880 char *tmp = skip_spaces(*buf);
1881 ret = tmp - *buf;
1882 *buf = tmp;
1883 return ret;
1884 }
1885
1886 static void proc_skip_char(char **buf, size_t *size, const char v)
1887 {
1888 while (*size) {
1889 if (**buf != v)
1890 break;
1891 (*size)--;
1892 (*buf)++;
1893 }
1894 }
1895
1896 #define TMPBUFLEN 22
1897 /**
1898 * proc_get_long - reads an ASCII formatted integer from a user buffer
1899 *
1900 * @buf: a kernel buffer
1901 * @size: size of the kernel buffer
1902 * @val: this is where the number will be stored
1903 * @neg: set to %TRUE if number is negative
1904 * @perm_tr: a vector which contains the allowed trailers
1905 * @perm_tr_len: size of the perm_tr vector
1906 * @tr: pointer to store the trailer character
1907 *
1908 * In case of success %0 is returned and @buf and @size are updated with
1909 * the amount of bytes read. If @tr is non-NULL and a trailing
1910 * character exists (size is non-zero after returning from this
1911 * function), @tr is updated with the trailing character.
1912 */
1913 static int proc_get_long(char **buf, size_t *size,
1914 unsigned long *val, bool *neg,
1915 const char *perm_tr, unsigned perm_tr_len, char *tr)
1916 {
1917 int len;
1918 char *p, tmp[TMPBUFLEN];
1919
1920 if (!*size)
1921 return -EINVAL;
1922
1923 len = *size;
1924 if (len > TMPBUFLEN - 1)
1925 len = TMPBUFLEN - 1;
1926
1927 memcpy(tmp, *buf, len);
1928
1929 tmp[len] = 0;
1930 p = tmp;
1931 if (*p == '-' && *size > 1) {
1932 *neg = true;
1933 p++;
1934 } else
1935 *neg = false;
1936 if (!isdigit(*p))
1937 return -EINVAL;
1938
1939 *val = simple_strtoul(p, &p, 0);
1940
1941 len = p - tmp;
1942
1943 /* We don't know if the next char is whitespace thus we may accept
1944 * invalid integers (e.g. 1234...a) or two integers instead of one
1945 * (e.g. 123...1). So lets not allow such large numbers. */
1946 if (len == TMPBUFLEN - 1)
1947 return -EINVAL;
1948
1949 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
1950 return -EINVAL;
1951
1952 if (tr && (len < *size))
1953 *tr = *p;
1954
1955 *buf += len;
1956 *size -= len;
1957
1958 return 0;
1959 }
1960
1961 /**
1962 * proc_put_long - converts an integer to a decimal ASCII formatted string
1963 *
1964 * @buf: the user buffer
1965 * @size: the size of the user buffer
1966 * @val: the integer to be converted
1967 * @neg: sign of the number, %TRUE for negative
1968 *
1969 * In case of success %0 is returned and @buf and @size are updated with
1970 * the amount of bytes written.
1971 */
1972 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
1973 bool neg)
1974 {
1975 int len;
1976 char tmp[TMPBUFLEN], *p = tmp;
1977
1978 sprintf(p, "%s%lu", neg ? "-" : "", val);
1979 len = strlen(tmp);
1980 if (len > *size)
1981 len = *size;
1982 if (copy_to_user(*buf, tmp, len))
1983 return -EFAULT;
1984 *size -= len;
1985 *buf += len;
1986 return 0;
1987 }
1988 #undef TMPBUFLEN
1989
1990 static int proc_put_char(void __user **buf, size_t *size, char c)
1991 {
1992 if (*size) {
1993 char __user **buffer = (char __user **)buf;
1994 if (put_user(c, *buffer))
1995 return -EFAULT;
1996 (*size)--, (*buffer)++;
1997 *buf = *buffer;
1998 }
1999 return 0;
2000 }
2001
2002 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2003 int *valp,
2004 int write, void *data)
2005 {
2006 if (write) {
2007 if (*negp) {
2008 if (*lvalp > (unsigned long) INT_MAX + 1)
2009 return -EINVAL;
2010 *valp = -*lvalp;
2011 } else {
2012 if (*lvalp > (unsigned long) INT_MAX)
2013 return -EINVAL;
2014 *valp = *lvalp;
2015 }
2016 } else {
2017 int val = *valp;
2018 if (val < 0) {
2019 *negp = true;
2020 *lvalp = -(unsigned long)val;
2021 } else {
2022 *negp = false;
2023 *lvalp = (unsigned long)val;
2024 }
2025 }
2026 return 0;
2027 }
2028
2029 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2030
2031 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2032 int write, void __user *buffer,
2033 size_t *lenp, loff_t *ppos,
2034 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2035 int write, void *data),
2036 void *data)
2037 {
2038 int *i, vleft, first = 1, err = 0;
2039 unsigned long page = 0;
2040 size_t left;
2041 char *kbuf;
2042
2043 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2044 *lenp = 0;
2045 return 0;
2046 }
2047
2048 i = (int *) tbl_data;
2049 vleft = table->maxlen / sizeof(*i);
2050 left = *lenp;
2051
2052 if (!conv)
2053 conv = do_proc_dointvec_conv;
2054
2055 if (write) {
2056 if (*ppos) {
2057 switch (sysctl_writes_strict) {
2058 case SYSCTL_WRITES_STRICT:
2059 goto out;
2060 case SYSCTL_WRITES_WARN:
2061 warn_sysctl_write(table);
2062 break;
2063 default:
2064 break;
2065 }
2066 }
2067
2068 if (left > PAGE_SIZE - 1)
2069 left = PAGE_SIZE - 1;
2070 page = __get_free_page(GFP_TEMPORARY);
2071 kbuf = (char *) page;
2072 if (!kbuf)
2073 return -ENOMEM;
2074 if (copy_from_user(kbuf, buffer, left)) {
2075 err = -EFAULT;
2076 goto free;
2077 }
2078 kbuf[left] = 0;
2079 }
2080
2081 for (; left && vleft--; i++, first=0) {
2082 unsigned long lval;
2083 bool neg;
2084
2085 if (write) {
2086 left -= proc_skip_spaces(&kbuf);
2087
2088 if (!left)
2089 break;
2090 err = proc_get_long(&kbuf, &left, &lval, &neg,
2091 proc_wspace_sep,
2092 sizeof(proc_wspace_sep), NULL);
2093 if (err)
2094 break;
2095 if (conv(&neg, &lval, i, 1, data)) {
2096 err = -EINVAL;
2097 break;
2098 }
2099 } else {
2100 if (conv(&neg, &lval, i, 0, data)) {
2101 err = -EINVAL;
2102 break;
2103 }
2104 if (!first)
2105 err = proc_put_char(&buffer, &left, '\t');
2106 if (err)
2107 break;
2108 err = proc_put_long(&buffer, &left, lval, neg);
2109 if (err)
2110 break;
2111 }
2112 }
2113
2114 if (!write && !first && left && !err)
2115 err = proc_put_char(&buffer, &left, '\n');
2116 if (write && !err && left)
2117 left -= proc_skip_spaces(&kbuf);
2118 free:
2119 if (write) {
2120 free_page(page);
2121 if (first)
2122 return err ? : -EINVAL;
2123 }
2124 *lenp -= left;
2125 out:
2126 *ppos += *lenp;
2127 return err;
2128 }
2129
2130 static int do_proc_dointvec(struct ctl_table *table, int write,
2131 void __user *buffer, size_t *lenp, loff_t *ppos,
2132 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2133 int write, void *data),
2134 void *data)
2135 {
2136 return __do_proc_dointvec(table->data, table, write,
2137 buffer, lenp, ppos, conv, data);
2138 }
2139
2140 /**
2141 * proc_dointvec - read a vector of integers
2142 * @table: the sysctl table
2143 * @write: %TRUE if this is a write to the sysctl file
2144 * @buffer: the user buffer
2145 * @lenp: the size of the user buffer
2146 * @ppos: file position
2147 *
2148 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2149 * values from/to the user buffer, treated as an ASCII string.
2150 *
2151 * Returns 0 on success.
2152 */
2153 int proc_dointvec(struct ctl_table *table, int write,
2154 void __user *buffer, size_t *lenp, loff_t *ppos)
2155 {
2156 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2157 NULL,NULL);
2158 }
2159
2160 /*
2161 * Taint values can only be increased
2162 * This means we can safely use a temporary.
2163 */
2164 static int proc_taint(struct ctl_table *table, int write,
2165 void __user *buffer, size_t *lenp, loff_t *ppos)
2166 {
2167 struct ctl_table t;
2168 unsigned long tmptaint = get_taint();
2169 int err;
2170
2171 if (write && !capable(CAP_SYS_ADMIN))
2172 return -EPERM;
2173
2174 t = *table;
2175 t.data = &tmptaint;
2176 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2177 if (err < 0)
2178 return err;
2179
2180 if (write) {
2181 /*
2182 * Poor man's atomic or. Not worth adding a primitive
2183 * to everyone's atomic.h for this
2184 */
2185 int i;
2186 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2187 if ((tmptaint >> i) & 1)
2188 add_taint(i, LOCKDEP_STILL_OK);
2189 }
2190 }
2191
2192 return err;
2193 }
2194
2195 #ifdef CONFIG_PRINTK
2196 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2197 void __user *buffer, size_t *lenp, loff_t *ppos)
2198 {
2199 if (write && !capable(CAP_SYS_ADMIN))
2200 return -EPERM;
2201
2202 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2203 }
2204 #endif
2205
2206 struct do_proc_dointvec_minmax_conv_param {
2207 int *min;
2208 int *max;
2209 };
2210
2211 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2212 int *valp,
2213 int write, void *data)
2214 {
2215 struct do_proc_dointvec_minmax_conv_param *param = data;
2216 if (write) {
2217 int val = *negp ? -*lvalp : *lvalp;
2218 if ((param->min && *param->min > val) ||
2219 (param->max && *param->max < val))
2220 return -EINVAL;
2221 *valp = val;
2222 } else {
2223 int val = *valp;
2224 if (val < 0) {
2225 *negp = true;
2226 *lvalp = -(unsigned long)val;
2227 } else {
2228 *negp = false;
2229 *lvalp = (unsigned long)val;
2230 }
2231 }
2232 return 0;
2233 }
2234
2235 /**
2236 * proc_dointvec_minmax - read a vector of integers with min/max values
2237 * @table: the sysctl table
2238 * @write: %TRUE if this is a write to the sysctl file
2239 * @buffer: the user buffer
2240 * @lenp: the size of the user buffer
2241 * @ppos: file position
2242 *
2243 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2244 * values from/to the user buffer, treated as an ASCII string.
2245 *
2246 * This routine will ensure the values are within the range specified by
2247 * table->extra1 (min) and table->extra2 (max).
2248 *
2249 * Returns 0 on success.
2250 */
2251 int proc_dointvec_minmax(struct ctl_table *table, int write,
2252 void __user *buffer, size_t *lenp, loff_t *ppos)
2253 {
2254 struct do_proc_dointvec_minmax_conv_param param = {
2255 .min = (int *) table->extra1,
2256 .max = (int *) table->extra2,
2257 };
2258 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2259 do_proc_dointvec_minmax_conv, &param);
2260 }
2261
2262 static void validate_coredump_safety(void)
2263 {
2264 #ifdef CONFIG_COREDUMP
2265 if (suid_dumpable == SUID_DUMP_ROOT &&
2266 core_pattern[0] != '/' && core_pattern[0] != '|') {
2267 printk(KERN_WARNING "Unsafe core_pattern used with "\
2268 "suid_dumpable=2. Pipe handler or fully qualified "\
2269 "core dump path required.\n");
2270 }
2271 #endif
2272 }
2273
2274 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2275 void __user *buffer, size_t *lenp, loff_t *ppos)
2276 {
2277 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2278 if (!error)
2279 validate_coredump_safety();
2280 return error;
2281 }
2282
2283 #ifdef CONFIG_COREDUMP
2284 static int proc_dostring_coredump(struct ctl_table *table, int write,
2285 void __user *buffer, size_t *lenp, loff_t *ppos)
2286 {
2287 int error = proc_dostring(table, write, buffer, lenp, ppos);
2288 if (!error)
2289 validate_coredump_safety();
2290 return error;
2291 }
2292 #endif
2293
2294 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2295 void __user *buffer,
2296 size_t *lenp, loff_t *ppos,
2297 unsigned long convmul,
2298 unsigned long convdiv)
2299 {
2300 unsigned long *i, *min, *max;
2301 int vleft, first = 1, err = 0;
2302 unsigned long page = 0;
2303 size_t left;
2304 char *kbuf;
2305
2306 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2307 *lenp = 0;
2308 return 0;
2309 }
2310
2311 i = (unsigned long *) data;
2312 min = (unsigned long *) table->extra1;
2313 max = (unsigned long *) table->extra2;
2314 vleft = table->maxlen / sizeof(unsigned long);
2315 left = *lenp;
2316
2317 if (write) {
2318 if (*ppos) {
2319 switch (sysctl_writes_strict) {
2320 case SYSCTL_WRITES_STRICT:
2321 goto out;
2322 case SYSCTL_WRITES_WARN:
2323 warn_sysctl_write(table);
2324 break;
2325 default:
2326 break;
2327 }
2328 }
2329
2330 if (left > PAGE_SIZE - 1)
2331 left = PAGE_SIZE - 1;
2332 page = __get_free_page(GFP_TEMPORARY);
2333 kbuf = (char *) page;
2334 if (!kbuf)
2335 return -ENOMEM;
2336 if (copy_from_user(kbuf, buffer, left)) {
2337 err = -EFAULT;
2338 goto free;
2339 }
2340 kbuf[left] = 0;
2341 }
2342
2343 for (; left && vleft--; i++, first = 0) {
2344 unsigned long val;
2345
2346 if (write) {
2347 bool neg;
2348
2349 left -= proc_skip_spaces(&kbuf);
2350
2351 err = proc_get_long(&kbuf, &left, &val, &neg,
2352 proc_wspace_sep,
2353 sizeof(proc_wspace_sep), NULL);
2354 if (err)
2355 break;
2356 if (neg)
2357 continue;
2358 if ((min && val < *min) || (max && val > *max))
2359 continue;
2360 *i = val;
2361 } else {
2362 val = convdiv * (*i) / convmul;
2363 if (!first) {
2364 err = proc_put_char(&buffer, &left, '\t');
2365 if (err)
2366 break;
2367 }
2368 err = proc_put_long(&buffer, &left, val, false);
2369 if (err)
2370 break;
2371 }
2372 }
2373
2374 if (!write && !first && left && !err)
2375 err = proc_put_char(&buffer, &left, '\n');
2376 if (write && !err)
2377 left -= proc_skip_spaces(&kbuf);
2378 free:
2379 if (write) {
2380 free_page(page);
2381 if (first)
2382 return err ? : -EINVAL;
2383 }
2384 *lenp -= left;
2385 out:
2386 *ppos += *lenp;
2387 return err;
2388 }
2389
2390 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2391 void __user *buffer,
2392 size_t *lenp, loff_t *ppos,
2393 unsigned long convmul,
2394 unsigned long convdiv)
2395 {
2396 return __do_proc_doulongvec_minmax(table->data, table, write,
2397 buffer, lenp, ppos, convmul, convdiv);
2398 }
2399
2400 /**
2401 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2402 * @table: the sysctl table
2403 * @write: %TRUE if this is a write to the sysctl file
2404 * @buffer: the user buffer
2405 * @lenp: the size of the user buffer
2406 * @ppos: file position
2407 *
2408 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2409 * values from/to the user buffer, treated as an ASCII string.
2410 *
2411 * This routine will ensure the values are within the range specified by
2412 * table->extra1 (min) and table->extra2 (max).
2413 *
2414 * Returns 0 on success.
2415 */
2416 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2417 void __user *buffer, size_t *lenp, loff_t *ppos)
2418 {
2419 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2420 }
2421
2422 /**
2423 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2424 * @table: the sysctl table
2425 * @write: %TRUE if this is a write to the sysctl file
2426 * @buffer: the user buffer
2427 * @lenp: the size of the user buffer
2428 * @ppos: file position
2429 *
2430 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2431 * values from/to the user buffer, treated as an ASCII string. The values
2432 * are treated as milliseconds, and converted to jiffies when they are stored.
2433 *
2434 * This routine will ensure the values are within the range specified by
2435 * table->extra1 (min) and table->extra2 (max).
2436 *
2437 * Returns 0 on success.
2438 */
2439 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2440 void __user *buffer,
2441 size_t *lenp, loff_t *ppos)
2442 {
2443 return do_proc_doulongvec_minmax(table, write, buffer,
2444 lenp, ppos, HZ, 1000l);
2445 }
2446
2447
2448 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2449 int *valp,
2450 int write, void *data)
2451 {
2452 if (write) {
2453 if (*lvalp > LONG_MAX / HZ)
2454 return 1;
2455 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2456 } else {
2457 int val = *valp;
2458 unsigned long lval;
2459 if (val < 0) {
2460 *negp = true;
2461 lval = -(unsigned long)val;
2462 } else {
2463 *negp = false;
2464 lval = (unsigned long)val;
2465 }
2466 *lvalp = lval / HZ;
2467 }
2468 return 0;
2469 }
2470
2471 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2472 int *valp,
2473 int write, void *data)
2474 {
2475 if (write) {
2476 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2477 return 1;
2478 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2479 } else {
2480 int val = *valp;
2481 unsigned long lval;
2482 if (val < 0) {
2483 *negp = true;
2484 lval = -(unsigned long)val;
2485 } else {
2486 *negp = false;
2487 lval = (unsigned long)val;
2488 }
2489 *lvalp = jiffies_to_clock_t(lval);
2490 }
2491 return 0;
2492 }
2493
2494 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2495 int *valp,
2496 int write, void *data)
2497 {
2498 if (write) {
2499 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2500
2501 if (jif > INT_MAX)
2502 return 1;
2503 *valp = (int)jif;
2504 } else {
2505 int val = *valp;
2506 unsigned long lval;
2507 if (val < 0) {
2508 *negp = true;
2509 lval = -(unsigned long)val;
2510 } else {
2511 *negp = false;
2512 lval = (unsigned long)val;
2513 }
2514 *lvalp = jiffies_to_msecs(lval);
2515 }
2516 return 0;
2517 }
2518
2519 /**
2520 * proc_dointvec_jiffies - read a vector of integers as seconds
2521 * @table: the sysctl table
2522 * @write: %TRUE if this is a write to the sysctl file
2523 * @buffer: the user buffer
2524 * @lenp: the size of the user buffer
2525 * @ppos: file position
2526 *
2527 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2528 * values from/to the user buffer, treated as an ASCII string.
2529 * The values read are assumed to be in seconds, and are converted into
2530 * jiffies.
2531 *
2532 * Returns 0 on success.
2533 */
2534 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2535 void __user *buffer, size_t *lenp, loff_t *ppos)
2536 {
2537 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2538 do_proc_dointvec_jiffies_conv,NULL);
2539 }
2540
2541 /**
2542 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2543 * @table: the sysctl table
2544 * @write: %TRUE if this is a write to the sysctl file
2545 * @buffer: the user buffer
2546 * @lenp: the size of the user buffer
2547 * @ppos: pointer to the file position
2548 *
2549 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2550 * values from/to the user buffer, treated as an ASCII string.
2551 * The values read are assumed to be in 1/USER_HZ seconds, and
2552 * are converted into jiffies.
2553 *
2554 * Returns 0 on success.
2555 */
2556 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2557 void __user *buffer, size_t *lenp, loff_t *ppos)
2558 {
2559 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2560 do_proc_dointvec_userhz_jiffies_conv,NULL);
2561 }
2562
2563 /**
2564 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2565 * @table: the sysctl table
2566 * @write: %TRUE if this is a write to the sysctl file
2567 * @buffer: the user buffer
2568 * @lenp: the size of the user buffer
2569 * @ppos: file position
2570 * @ppos: the current position in the file
2571 *
2572 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2573 * values from/to the user buffer, treated as an ASCII string.
2574 * The values read are assumed to be in 1/1000 seconds, and
2575 * are converted into jiffies.
2576 *
2577 * Returns 0 on success.
2578 */
2579 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2580 void __user *buffer, size_t *lenp, loff_t *ppos)
2581 {
2582 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2583 do_proc_dointvec_ms_jiffies_conv, NULL);
2584 }
2585
2586 static int proc_do_cad_pid(struct ctl_table *table, int write,
2587 void __user *buffer, size_t *lenp, loff_t *ppos)
2588 {
2589 struct pid *new_pid;
2590 pid_t tmp;
2591 int r;
2592
2593 tmp = pid_vnr(cad_pid);
2594
2595 r = __do_proc_dointvec(&tmp, table, write, buffer,
2596 lenp, ppos, NULL, NULL);
2597 if (r || !write)
2598 return r;
2599
2600 new_pid = find_get_pid(tmp);
2601 if (!new_pid)
2602 return -ESRCH;
2603
2604 put_pid(xchg(&cad_pid, new_pid));
2605 return 0;
2606 }
2607
2608 /**
2609 * proc_do_large_bitmap - read/write from/to a large bitmap
2610 * @table: the sysctl table
2611 * @write: %TRUE if this is a write to the sysctl file
2612 * @buffer: the user buffer
2613 * @lenp: the size of the user buffer
2614 * @ppos: file position
2615 *
2616 * The bitmap is stored at table->data and the bitmap length (in bits)
2617 * in table->maxlen.
2618 *
2619 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2620 * large bitmaps may be represented in a compact manner. Writing into
2621 * the file will clear the bitmap then update it with the given input.
2622 *
2623 * Returns 0 on success.
2624 */
2625 int proc_do_large_bitmap(struct ctl_table *table, int write,
2626 void __user *buffer, size_t *lenp, loff_t *ppos)
2627 {
2628 int err = 0;
2629 bool first = 1;
2630 size_t left = *lenp;
2631 unsigned long bitmap_len = table->maxlen;
2632 unsigned long *bitmap = *(unsigned long **) table->data;
2633 unsigned long *tmp_bitmap = NULL;
2634 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2635
2636 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2637 *lenp = 0;
2638 return 0;
2639 }
2640
2641 if (write) {
2642 unsigned long page = 0;
2643 char *kbuf;
2644
2645 if (left > PAGE_SIZE - 1)
2646 left = PAGE_SIZE - 1;
2647
2648 page = __get_free_page(GFP_TEMPORARY);
2649 kbuf = (char *) page;
2650 if (!kbuf)
2651 return -ENOMEM;
2652 if (copy_from_user(kbuf, buffer, left)) {
2653 free_page(page);
2654 return -EFAULT;
2655 }
2656 kbuf[left] = 0;
2657
2658 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2659 GFP_KERNEL);
2660 if (!tmp_bitmap) {
2661 free_page(page);
2662 return -ENOMEM;
2663 }
2664 proc_skip_char(&kbuf, &left, '\n');
2665 while (!err && left) {
2666 unsigned long val_a, val_b;
2667 bool neg;
2668
2669 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2670 sizeof(tr_a), &c);
2671 if (err)
2672 break;
2673 if (val_a >= bitmap_len || neg) {
2674 err = -EINVAL;
2675 break;
2676 }
2677
2678 val_b = val_a;
2679 if (left) {
2680 kbuf++;
2681 left--;
2682 }
2683
2684 if (c == '-') {
2685 err = proc_get_long(&kbuf, &left, &val_b,
2686 &neg, tr_b, sizeof(tr_b),
2687 &c);
2688 if (err)
2689 break;
2690 if (val_b >= bitmap_len || neg ||
2691 val_a > val_b) {
2692 err = -EINVAL;
2693 break;
2694 }
2695 if (left) {
2696 kbuf++;
2697 left--;
2698 }
2699 }
2700
2701 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2702 first = 0;
2703 proc_skip_char(&kbuf, &left, '\n');
2704 }
2705 free_page(page);
2706 } else {
2707 unsigned long bit_a, bit_b = 0;
2708
2709 while (left) {
2710 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2711 if (bit_a >= bitmap_len)
2712 break;
2713 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2714 bit_a + 1) - 1;
2715
2716 if (!first) {
2717 err = proc_put_char(&buffer, &left, ',');
2718 if (err)
2719 break;
2720 }
2721 err = proc_put_long(&buffer, &left, bit_a, false);
2722 if (err)
2723 break;
2724 if (bit_a != bit_b) {
2725 err = proc_put_char(&buffer, &left, '-');
2726 if (err)
2727 break;
2728 err = proc_put_long(&buffer, &left, bit_b, false);
2729 if (err)
2730 break;
2731 }
2732
2733 first = 0; bit_b++;
2734 }
2735 if (!err)
2736 err = proc_put_char(&buffer, &left, '\n');
2737 }
2738
2739 if (!err) {
2740 if (write) {
2741 if (*ppos)
2742 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2743 else
2744 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2745 }
2746 kfree(tmp_bitmap);
2747 *lenp -= left;
2748 *ppos += *lenp;
2749 return 0;
2750 } else {
2751 kfree(tmp_bitmap);
2752 return err;
2753 }
2754 }
2755
2756 #else /* CONFIG_PROC_SYSCTL */
2757
2758 int proc_dostring(struct ctl_table *table, int write,
2759 void __user *buffer, size_t *lenp, loff_t *ppos)
2760 {
2761 return -ENOSYS;
2762 }
2763
2764 int proc_dointvec(struct ctl_table *table, int write,
2765 void __user *buffer, size_t *lenp, loff_t *ppos)
2766 {
2767 return -ENOSYS;
2768 }
2769
2770 int proc_dointvec_minmax(struct ctl_table *table, int write,
2771 void __user *buffer, size_t *lenp, loff_t *ppos)
2772 {
2773 return -ENOSYS;
2774 }
2775
2776 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2777 void __user *buffer, size_t *lenp, loff_t *ppos)
2778 {
2779 return -ENOSYS;
2780 }
2781
2782 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2783 void __user *buffer, size_t *lenp, loff_t *ppos)
2784 {
2785 return -ENOSYS;
2786 }
2787
2788 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2789 void __user *buffer, size_t *lenp, loff_t *ppos)
2790 {
2791 return -ENOSYS;
2792 }
2793
2794 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2795 void __user *buffer, size_t *lenp, loff_t *ppos)
2796 {
2797 return -ENOSYS;
2798 }
2799
2800 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2801 void __user *buffer,
2802 size_t *lenp, loff_t *ppos)
2803 {
2804 return -ENOSYS;
2805 }
2806
2807
2808 #endif /* CONFIG_PROC_SYSCTL */
2809
2810 /*
2811 * No sense putting this after each symbol definition, twice,
2812 * exception granted :-)
2813 */
2814 EXPORT_SYMBOL(proc_dointvec);
2815 EXPORT_SYMBOL(proc_dointvec_jiffies);
2816 EXPORT_SYMBOL(proc_dointvec_minmax);
2817 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2818 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2819 EXPORT_SYMBOL(proc_dostring);
2820 EXPORT_SYMBOL(proc_doulongvec_minmax);
2821 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
Linux kernel - Deliverables
This page took 0.118125 seconds and 6 git commands to generate.