projects / deliverable / linux.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Merge tag 'ftracetest-3.18' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt...
[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/mm.h>
23 #include <linux/swap.h>
24 #include <linux/slab.h>
25 #include <linux/sysctl.h>
26 #include <linux/bitmap.h>
27 #include <linux/signal.h>
28 #include <linux/printk.h>
29 #include <linux/proc_fs.h>
30 #include <linux/security.h>
31 #include <linux/ctype.h>
32 #include <linux/kmemcheck.h>
33 #include <linux/kmemleak.h>
34 #include <linux/fs.h>
35 #include <linux/init.h>
36 #include <linux/kernel.h>
37 #include <linux/kobject.h>
38 #include <linux/net.h>
39 #include <linux/sysrq.h>
40 #include <linux/highuid.h>
41 #include <linux/writeback.h>
42 #include <linux/ratelimit.h>
43 #include <linux/compaction.h>
44 #include <linux/hugetlb.h>
45 #include <linux/initrd.h>
46 #include <linux/key.h>
47 #include <linux/times.h>
48 #include <linux/limits.h>
49 #include <linux/dcache.h>
50 #include <linux/dnotify.h>
51 #include <linux/syscalls.h>
52 #include <linux/vmstat.h>
53 #include <linux/nfs_fs.h>
54 #include <linux/acpi.h>
55 #include <linux/reboot.h>
56 #include <linux/ftrace.h>
57 #include <linux/perf_event.h>
58 #include <linux/kprobes.h>
59 #include <linux/pipe_fs_i.h>
60 #include <linux/oom.h>
61 #include <linux/kmod.h>
62 #include <linux/capability.h>
63 #include <linux/binfmts.h>
64 #include <linux/sched/sysctl.h>
65 #include <linux/kexec.h>
66
67 #include <asm/uaccess.h>
68 #include <asm/processor.h>
69
70 #ifdef CONFIG_X86
71 #include <asm/nmi.h>
72 #include <asm/stacktrace.h>
73 #include <asm/io.h>
74 #endif
75 #ifdef CONFIG_SPARC
76 #include <asm/setup.h>
77 #endif
78 #ifdef CONFIG_BSD_PROCESS_ACCT
79 #include <linux/acct.h>
80 #endif
81 #ifdef CONFIG_RT_MUTEXES
82 #include <linux/rtmutex.h>
83 #endif
84 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
85 #include <linux/lockdep.h>
86 #endif
87 #ifdef CONFIG_CHR_DEV_SG
88 #include <scsi/sg.h>
89 #endif
90
91 #ifdef CONFIG_LOCKUP_DETECTOR
92 #include <linux/nmi.h>
93 #endif
94
95
96 #if defined(CONFIG_SYSCTL)
97
98 /* External variables not in a header file. */
99 extern int max_threads;
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 {
354 .procname = "timer_migration",
355 .data = &sysctl_timer_migration,
356 .maxlen = sizeof(unsigned int),
357 .mode = 0644,
358 .proc_handler = proc_dointvec_minmax,
359 .extra1 = &zero,
360 .extra2 = &one,
361 },
362 #endif /* CONFIG_SMP */
363 #ifdef CONFIG_NUMA_BALANCING
364 {
365 .procname = "numa_balancing_scan_delay_ms",
366 .data = &sysctl_numa_balancing_scan_delay,
367 .maxlen = sizeof(unsigned int),
368 .mode = 0644,
369 .proc_handler = proc_dointvec,
370 },
371 {
372 .procname = "numa_balancing_scan_period_min_ms",
373 .data = &sysctl_numa_balancing_scan_period_min,
374 .maxlen = sizeof(unsigned int),
375 .mode = 0644,
376 .proc_handler = proc_dointvec,
377 },
378 {
379 .procname = "numa_balancing_scan_period_max_ms",
380 .data = &sysctl_numa_balancing_scan_period_max,
381 .maxlen = sizeof(unsigned int),
382 .mode = 0644,
383 .proc_handler = proc_dointvec,
384 },
385 {
386 .procname = "numa_balancing_scan_size_mb",
387 .data = &sysctl_numa_balancing_scan_size,
388 .maxlen = sizeof(unsigned int),
389 .mode = 0644,
390 .proc_handler = proc_dointvec,
391 },
392 {
393 .procname = "numa_balancing",
394 .data = NULL, /* filled in by handler */
395 .maxlen = sizeof(unsigned int),
396 .mode = 0644,
397 .proc_handler = sysctl_numa_balancing,
398 .extra1 = &zero,
399 .extra2 = &one,
400 },
401 #endif /* CONFIG_NUMA_BALANCING */
402 #endif /* CONFIG_SCHED_DEBUG */
403 {
404 .procname = "sched_rt_period_us",
405 .data = &sysctl_sched_rt_period,
406 .maxlen = sizeof(unsigned int),
407 .mode = 0644,
408 .proc_handler = sched_rt_handler,
409 },
410 {
411 .procname = "sched_rt_runtime_us",
412 .data = &sysctl_sched_rt_runtime,
413 .maxlen = sizeof(int),
414 .mode = 0644,
415 .proc_handler = sched_rt_handler,
416 },
417 {
418 .procname = "sched_rr_timeslice_ms",
419 .data = &sched_rr_timeslice,
420 .maxlen = sizeof(int),
421 .mode = 0644,
422 .proc_handler = sched_rr_handler,
423 },
424 #ifdef CONFIG_SCHED_AUTOGROUP
425 {
426 .procname = "sched_autogroup_enabled",
427 .data = &sysctl_sched_autogroup_enabled,
428 .maxlen = sizeof(unsigned int),
429 .mode = 0644,
430 .proc_handler = proc_dointvec_minmax,
431 .extra1 = &zero,
432 .extra2 = &one,
433 },
434 #endif
435 #ifdef CONFIG_CFS_BANDWIDTH
436 {
437 .procname = "sched_cfs_bandwidth_slice_us",
438 .data = &sysctl_sched_cfs_bandwidth_slice,
439 .maxlen = sizeof(unsigned int),
440 .mode = 0644,
441 .proc_handler = proc_dointvec_minmax,
442 .extra1 = &one,
443 },
444 #endif
445 #ifdef CONFIG_PROVE_LOCKING
446 {
447 .procname = "prove_locking",
448 .data = &prove_locking,
449 .maxlen = sizeof(int),
450 .mode = 0644,
451 .proc_handler = proc_dointvec,
452 },
453 #endif
454 #ifdef CONFIG_LOCK_STAT
455 {
456 .procname = "lock_stat",
457 .data = &lock_stat,
458 .maxlen = sizeof(int),
459 .mode = 0644,
460 .proc_handler = proc_dointvec,
461 },
462 #endif
463 {
464 .procname = "panic",
465 .data = &panic_timeout,
466 .maxlen = sizeof(int),
467 .mode = 0644,
468 .proc_handler = proc_dointvec,
469 },
470 #ifdef CONFIG_COREDUMP
471 {
472 .procname = "core_uses_pid",
473 .data = &core_uses_pid,
474 .maxlen = sizeof(int),
475 .mode = 0644,
476 .proc_handler = proc_dointvec,
477 },
478 {
479 .procname = "core_pattern",
480 .data = core_pattern,
481 .maxlen = CORENAME_MAX_SIZE,
482 .mode = 0644,
483 .proc_handler = proc_dostring_coredump,
484 },
485 {
486 .procname = "core_pipe_limit",
487 .data = &core_pipe_limit,
488 .maxlen = sizeof(unsigned int),
489 .mode = 0644,
490 .proc_handler = proc_dointvec,
491 },
492 #endif
493 #ifdef CONFIG_PROC_SYSCTL
494 {
495 .procname = "tainted",
496 .maxlen = sizeof(long),
497 .mode = 0644,
498 .proc_handler = proc_taint,
499 },
500 {
501 .procname = "sysctl_writes_strict",
502 .data = &sysctl_writes_strict,
503 .maxlen = sizeof(int),
504 .mode = 0644,
505 .proc_handler = proc_dointvec_minmax,
506 .extra1 = &neg_one,
507 .extra2 = &one,
508 },
509 #endif
510 #ifdef CONFIG_LATENCYTOP
511 {
512 .procname = "latencytop",
513 .data = &latencytop_enabled,
514 .maxlen = sizeof(int),
515 .mode = 0644,
516 .proc_handler = proc_dointvec,
517 },
518 #endif
519 #ifdef CONFIG_BLK_DEV_INITRD
520 {
521 .procname = "real-root-dev",
522 .data = &real_root_dev,
523 .maxlen = sizeof(int),
524 .mode = 0644,
525 .proc_handler = proc_dointvec,
526 },
527 #endif
528 {
529 .procname = "print-fatal-signals",
530 .data = &print_fatal_signals,
531 .maxlen = sizeof(int),
532 .mode = 0644,
533 .proc_handler = proc_dointvec,
534 },
535 #ifdef CONFIG_SPARC
536 {
537 .procname = "reboot-cmd",
538 .data = reboot_command,
539 .maxlen = 256,
540 .mode = 0644,
541 .proc_handler = proc_dostring,
542 },
543 {
544 .procname = "stop-a",
545 .data = &stop_a_enabled,
546 .maxlen = sizeof (int),
547 .mode = 0644,
548 .proc_handler = proc_dointvec,
549 },
550 {
551 .procname = "scons-poweroff",
552 .data = &scons_pwroff,
553 .maxlen = sizeof (int),
554 .mode = 0644,
555 .proc_handler = proc_dointvec,
556 },
557 #endif
558 #ifdef CONFIG_SPARC64
559 {
560 .procname = "tsb-ratio",
561 .data = &sysctl_tsb_ratio,
562 .maxlen = sizeof (int),
563 .mode = 0644,
564 .proc_handler = proc_dointvec,
565 },
566 #endif
567 #ifdef __hppa__
568 {
569 .procname = "soft-power",
570 .data = &pwrsw_enabled,
571 .maxlen = sizeof (int),
572 .mode = 0644,
573 .proc_handler = proc_dointvec,
574 },
575 #endif
576 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
577 {
578 .procname = "unaligned-trap",
579 .data = &unaligned_enabled,
580 .maxlen = sizeof (int),
581 .mode = 0644,
582 .proc_handler = proc_dointvec,
583 },
584 #endif
585 {
586 .procname = "ctrl-alt-del",
587 .data = &C_A_D,
588 .maxlen = sizeof(int),
589 .mode = 0644,
590 .proc_handler = proc_dointvec,
591 },
592 #ifdef CONFIG_FUNCTION_TRACER
593 {
594 .procname = "ftrace_enabled",
595 .data = &ftrace_enabled,
596 .maxlen = sizeof(int),
597 .mode = 0644,
598 .proc_handler = ftrace_enable_sysctl,
599 },
600 #endif
601 #ifdef CONFIG_STACK_TRACER
602 {
603 .procname = "stack_tracer_enabled",
604 .data = &stack_tracer_enabled,
605 .maxlen = sizeof(int),
606 .mode = 0644,
607 .proc_handler = stack_trace_sysctl,
608 },
609 #endif
610 #ifdef CONFIG_TRACING
611 {
612 .procname = "ftrace_dump_on_oops",
613 .data = &ftrace_dump_on_oops,
614 .maxlen = sizeof(int),
615 .mode = 0644,
616 .proc_handler = proc_dointvec,
617 },
618 {
619 .procname = "traceoff_on_warning",
620 .data = &__disable_trace_on_warning,
621 .maxlen = sizeof(__disable_trace_on_warning),
622 .mode = 0644,
623 .proc_handler = proc_dointvec,
624 },
625 #endif
626 #ifdef CONFIG_KEXEC
627 {
628 .procname = "kexec_load_disabled",
629 .data = &kexec_load_disabled,
630 .maxlen = sizeof(int),
631 .mode = 0644,
632 /* only handle a transition from default "0" to "1" */
633 .proc_handler = proc_dointvec_minmax,
634 .extra1 = &one,
635 .extra2 = &one,
636 },
637 #endif
638 #ifdef CONFIG_MODULES
639 {
640 .procname = "modprobe",
641 .data = &modprobe_path,
642 .maxlen = KMOD_PATH_LEN,
643 .mode = 0644,
644 .proc_handler = proc_dostring,
645 },
646 {
647 .procname = "modules_disabled",
648 .data = &modules_disabled,
649 .maxlen = sizeof(int),
650 .mode = 0644,
651 /* only handle a transition from default "0" to "1" */
652 .proc_handler = proc_dointvec_minmax,
653 .extra1 = &one,
654 .extra2 = &one,
655 },
656 #endif
657 #ifdef CONFIG_UEVENT_HELPER
658 {
659 .procname = "hotplug",
660 .data = &uevent_helper,
661 .maxlen = UEVENT_HELPER_PATH_LEN,
662 .mode = 0644,
663 .proc_handler = proc_dostring,
664 },
665 #endif
666 #ifdef CONFIG_CHR_DEV_SG
667 {
668 .procname = "sg-big-buff",
669 .data = &sg_big_buff,
670 .maxlen = sizeof (int),
671 .mode = 0444,
672 .proc_handler = proc_dointvec,
673 },
674 #endif
675 #ifdef CONFIG_BSD_PROCESS_ACCT
676 {
677 .procname = "acct",
678 .data = &acct_parm,
679 .maxlen = 3*sizeof(int),
680 .mode = 0644,
681 .proc_handler = proc_dointvec,
682 },
683 #endif
684 #ifdef CONFIG_MAGIC_SYSRQ
685 {
686 .procname = "sysrq",
687 .data = &__sysrq_enabled,
688 .maxlen = sizeof (int),
689 .mode = 0644,
690 .proc_handler = sysrq_sysctl_handler,
691 },
692 #endif
693 #ifdef CONFIG_PROC_SYSCTL
694 {
695 .procname = "cad_pid",
696 .data = NULL,
697 .maxlen = sizeof (int),
698 .mode = 0600,
699 .proc_handler = proc_do_cad_pid,
700 },
701 #endif
702 {
703 .procname = "threads-max",
704 .data = &max_threads,
705 .maxlen = sizeof(int),
706 .mode = 0644,
707 .proc_handler = proc_dointvec,
708 },
709 {
710 .procname = "random",
711 .mode = 0555,
712 .child = random_table,
713 },
714 {
715 .procname = "usermodehelper",
716 .mode = 0555,
717 .child = usermodehelper_table,
718 },
719 {
720 .procname = "overflowuid",
721 .data = &overflowuid,
722 .maxlen = sizeof(int),
723 .mode = 0644,
724 .proc_handler = proc_dointvec_minmax,
725 .extra1 = &minolduid,
726 .extra2 = &maxolduid,
727 },
728 {
729 .procname = "overflowgid",
730 .data = &overflowgid,
731 .maxlen = sizeof(int),
732 .mode = 0644,
733 .proc_handler = proc_dointvec_minmax,
734 .extra1 = &minolduid,
735 .extra2 = &maxolduid,
736 },
737 #ifdef CONFIG_S390
738 #ifdef CONFIG_MATHEMU
739 {
740 .procname = "ieee_emulation_warnings",
741 .data = &sysctl_ieee_emulation_warnings,
742 .maxlen = sizeof(int),
743 .mode = 0644,
744 .proc_handler = proc_dointvec,
745 },
746 #endif
747 {
748 .procname = "userprocess_debug",
749 .data = &show_unhandled_signals,
750 .maxlen = sizeof(int),
751 .mode = 0644,
752 .proc_handler = proc_dointvec,
753 },
754 #endif
755 {
756 .procname = "pid_max",
757 .data = &pid_max,
758 .maxlen = sizeof (int),
759 .mode = 0644,
760 .proc_handler = proc_dointvec_minmax,
761 .extra1 = &pid_max_min,
762 .extra2 = &pid_max_max,
763 },
764 {
765 .procname = "panic_on_oops",
766 .data = &panic_on_oops,
767 .maxlen = sizeof(int),
768 .mode = 0644,
769 .proc_handler = proc_dointvec,
770 },
771 #if defined CONFIG_PRINTK
772 {
773 .procname = "printk",
774 .data = &console_loglevel,
775 .maxlen = 4*sizeof(int),
776 .mode = 0644,
777 .proc_handler = proc_dointvec,
778 },
779 {
780 .procname = "printk_ratelimit",
781 .data = &printk_ratelimit_state.interval,
782 .maxlen = sizeof(int),
783 .mode = 0644,
784 .proc_handler = proc_dointvec_jiffies,
785 },
786 {
787 .procname = "printk_ratelimit_burst",
788 .data = &printk_ratelimit_state.burst,
789 .maxlen = sizeof(int),
790 .mode = 0644,
791 .proc_handler = proc_dointvec,
792 },
793 {
794 .procname = "printk_delay",
795 .data = &printk_delay_msec,
796 .maxlen = sizeof(int),
797 .mode = 0644,
798 .proc_handler = proc_dointvec_minmax,
799 .extra1 = &zero,
800 .extra2 = &ten_thousand,
801 },
802 {
803 .procname = "dmesg_restrict",
804 .data = &dmesg_restrict,
805 .maxlen = sizeof(int),
806 .mode = 0644,
807 .proc_handler = proc_dointvec_minmax_sysadmin,
808 .extra1 = &zero,
809 .extra2 = &one,
810 },
811 {
812 .procname = "kptr_restrict",
813 .data = &kptr_restrict,
814 .maxlen = sizeof(int),
815 .mode = 0644,
816 .proc_handler = proc_dointvec_minmax_sysadmin,
817 .extra1 = &zero,
818 .extra2 = &two,
819 },
820 #endif
821 {
822 .procname = "ngroups_max",
823 .data = &ngroups_max,
824 .maxlen = sizeof (int),
825 .mode = 0444,
826 .proc_handler = proc_dointvec,
827 },
828 {
829 .procname = "cap_last_cap",
830 .data = (void *)&cap_last_cap,
831 .maxlen = sizeof(int),
832 .mode = 0444,
833 .proc_handler = proc_dointvec,
834 },
835 #if defined(CONFIG_LOCKUP_DETECTOR)
836 {
837 .procname = "watchdog",
838 .data = &watchdog_user_enabled,
839 .maxlen = sizeof (int),
840 .mode = 0644,
841 .proc_handler = proc_dowatchdog,
842 .extra1 = &zero,
843 .extra2 = &one,
844 },
845 {
846 .procname = "watchdog_thresh",
847 .data = &watchdog_thresh,
848 .maxlen = sizeof(int),
849 .mode = 0644,
850 .proc_handler = proc_dowatchdog,
851 .extra1 = &zero,
852 .extra2 = &sixty,
853 },
854 {
855 .procname = "softlockup_panic",
856 .data = &softlockup_panic,
857 .maxlen = sizeof(int),
858 .mode = 0644,
859 .proc_handler = proc_dointvec_minmax,
860 .extra1 = &zero,
861 .extra2 = &one,
862 },
863 #ifdef CONFIG_SMP
864 {
865 .procname = "softlockup_all_cpu_backtrace",
866 .data = &sysctl_softlockup_all_cpu_backtrace,
867 .maxlen = sizeof(int),
868 .mode = 0644,
869 .proc_handler = proc_dointvec_minmax,
870 .extra1 = &zero,
871 .extra2 = &one,
872 },
873 #endif /* CONFIG_SMP */
874 {
875 .procname = "nmi_watchdog",
876 .data = &watchdog_user_enabled,
877 .maxlen = sizeof (int),
878 .mode = 0644,
879 .proc_handler = proc_dowatchdog,
880 .extra1 = &zero,
881 .extra2 = &one,
882 },
883 #endif
884 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
885 {
886 .procname = "unknown_nmi_panic",
887 .data = &unknown_nmi_panic,
888 .maxlen = sizeof (int),
889 .mode = 0644,
890 .proc_handler = proc_dointvec,
891 },
892 #endif
893 #if defined(CONFIG_X86)
894 {
895 .procname = "panic_on_unrecovered_nmi",
896 .data = &panic_on_unrecovered_nmi,
897 .maxlen = sizeof(int),
898 .mode = 0644,
899 .proc_handler = proc_dointvec,
900 },
901 {
902 .procname = "panic_on_io_nmi",
903 .data = &panic_on_io_nmi,
904 .maxlen = sizeof(int),
905 .mode = 0644,
906 .proc_handler = proc_dointvec,
907 },
908 #ifdef CONFIG_DEBUG_STACKOVERFLOW
909 {
910 .procname = "panic_on_stackoverflow",
911 .data = &sysctl_panic_on_stackoverflow,
912 .maxlen = sizeof(int),
913 .mode = 0644,
914 .proc_handler = proc_dointvec,
915 },
916 #endif
917 {
918 .procname = "bootloader_type",
919 .data = &bootloader_type,
920 .maxlen = sizeof (int),
921 .mode = 0444,
922 .proc_handler = proc_dointvec,
923 },
924 {
925 .procname = "bootloader_version",
926 .data = &bootloader_version,
927 .maxlen = sizeof (int),
928 .mode = 0444,
929 .proc_handler = proc_dointvec,
930 },
931 {
932 .procname = "kstack_depth_to_print",
933 .data = &kstack_depth_to_print,
934 .maxlen = sizeof(int),
935 .mode = 0644,
936 .proc_handler = proc_dointvec,
937 },
938 {
939 .procname = "io_delay_type",
940 .data = &io_delay_type,
941 .maxlen = sizeof(int),
942 .mode = 0644,
943 .proc_handler = proc_dointvec,
944 },
945 #endif
946 #if defined(CONFIG_MMU)
947 {
948 .procname = "randomize_va_space",
949 .data = &randomize_va_space,
950 .maxlen = sizeof(int),
951 .mode = 0644,
952 .proc_handler = proc_dointvec,
953 },
954 #endif
955 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
956 {
957 .procname = "spin_retry",
958 .data = &spin_retry,
959 .maxlen = sizeof (int),
960 .mode = 0644,
961 .proc_handler = proc_dointvec,
962 },
963 #endif
964 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
965 {
966 .procname = "acpi_video_flags",
967 .data = &acpi_realmode_flags,
968 .maxlen = sizeof (unsigned long),
969 .mode = 0644,
970 .proc_handler = proc_doulongvec_minmax,
971 },
972 #endif
973 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
974 {
975 .procname = "ignore-unaligned-usertrap",
976 .data = &no_unaligned_warning,
977 .maxlen = sizeof (int),
978 .mode = 0644,
979 .proc_handler = proc_dointvec,
980 },
981 #endif
982 #ifdef CONFIG_IA64
983 {
984 .procname = "unaligned-dump-stack",
985 .data = &unaligned_dump_stack,
986 .maxlen = sizeof (int),
987 .mode = 0644,
988 .proc_handler = proc_dointvec,
989 },
990 #endif
991 #ifdef CONFIG_DETECT_HUNG_TASK
992 {
993 .procname = "hung_task_panic",
994 .data = &sysctl_hung_task_panic,
995 .maxlen = sizeof(int),
996 .mode = 0644,
997 .proc_handler = proc_dointvec_minmax,
998 .extra1 = &zero,
999 .extra2 = &one,
1000 },
1001 {
1002 .procname = "hung_task_check_count",
1003 .data = &sysctl_hung_task_check_count,
1004 .maxlen = sizeof(int),
1005 .mode = 0644,
1006 .proc_handler = proc_dointvec_minmax,
1007 .extra1 = &zero,
1008 },
1009 {
1010 .procname = "hung_task_timeout_secs",
1011 .data = &sysctl_hung_task_timeout_secs,
1012 .maxlen = sizeof(unsigned long),
1013 .mode = 0644,
1014 .proc_handler = proc_dohung_task_timeout_secs,
1015 .extra2 = &hung_task_timeout_max,
1016 },
1017 {
1018 .procname = "hung_task_warnings",
1019 .data = &sysctl_hung_task_warnings,
1020 .maxlen = sizeof(int),
1021 .mode = 0644,
1022 .proc_handler = proc_dointvec_minmax,
1023 .extra1 = &neg_one,
1024 },
1025 #endif
1026 #ifdef CONFIG_COMPAT
1027 {
1028 .procname = "compat-log",
1029 .data = &compat_log,
1030 .maxlen = sizeof (int),
1031 .mode = 0644,
1032 .proc_handler = proc_dointvec,
1033 },
1034 #endif
1035 #ifdef CONFIG_RT_MUTEXES
1036 {
1037 .procname = "max_lock_depth",
1038 .data = &max_lock_depth,
1039 .maxlen = sizeof(int),
1040 .mode = 0644,
1041 .proc_handler = proc_dointvec,
1042 },
1043 #endif
1044 {
1045 .procname = "poweroff_cmd",
1046 .data = &poweroff_cmd,
1047 .maxlen = POWEROFF_CMD_PATH_LEN,
1048 .mode = 0644,
1049 .proc_handler = proc_dostring,
1050 },
1051 #ifdef CONFIG_KEYS
1052 {
1053 .procname = "keys",
1054 .mode = 0555,
1055 .child = key_sysctls,
1056 },
1057 #endif
1058 #ifdef CONFIG_RCU_TORTURE_TEST
1059 {
1060 .procname = "rcutorture_runnable",
1061 .data = &rcutorture_runnable,
1062 .maxlen = sizeof(int),
1063 .mode = 0644,
1064 .proc_handler = proc_dointvec,
1065 },
1066 #endif
1067 #ifdef CONFIG_PERF_EVENTS
1068 /*
1069 * User-space scripts rely on the existence of this file
1070 * as a feature check for perf_events being enabled.
1071 *
1072 * So it's an ABI, do not remove!
1073 */
1074 {
1075 .procname = "perf_event_paranoid",
1076 .data = &sysctl_perf_event_paranoid,
1077 .maxlen = sizeof(sysctl_perf_event_paranoid),
1078 .mode = 0644,
1079 .proc_handler = proc_dointvec,
1080 },
1081 {
1082 .procname = "perf_event_mlock_kb",
1083 .data = &sysctl_perf_event_mlock,
1084 .maxlen = sizeof(sysctl_perf_event_mlock),
1085 .mode = 0644,
1086 .proc_handler = proc_dointvec,
1087 },
1088 {
1089 .procname = "perf_event_max_sample_rate",
1090 .data = &sysctl_perf_event_sample_rate,
1091 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1092 .mode = 0644,
1093 .proc_handler = perf_proc_update_handler,
1094 .extra1 = &one,
1095 },
1096 {
1097 .procname = "perf_cpu_time_max_percent",
1098 .data = &sysctl_perf_cpu_time_max_percent,
1099 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1100 .mode = 0644,
1101 .proc_handler = perf_cpu_time_max_percent_handler,
1102 .extra1 = &zero,
1103 .extra2 = &one_hundred,
1104 },
1105 #endif
1106 #ifdef CONFIG_KMEMCHECK
1107 {
1108 .procname = "kmemcheck",
1109 .data = &kmemcheck_enabled,
1110 .maxlen = sizeof(int),
1111 .mode = 0644,
1112 .proc_handler = proc_dointvec,
1113 },
1114 #endif
1115 { }
1116 };
1117
1118 static struct ctl_table vm_table[] = {
1119 {
1120 .procname = "overcommit_memory",
1121 .data = &sysctl_overcommit_memory,
1122 .maxlen = sizeof(sysctl_overcommit_memory),
1123 .mode = 0644,
1124 .proc_handler = proc_dointvec_minmax,
1125 .extra1 = &zero,
1126 .extra2 = &two,
1127 },
1128 {
1129 .procname = "panic_on_oom",
1130 .data = &sysctl_panic_on_oom,
1131 .maxlen = sizeof(sysctl_panic_on_oom),
1132 .mode = 0644,
1133 .proc_handler = proc_dointvec_minmax,
1134 .extra1 = &zero,
1135 .extra2 = &two,
1136 },
1137 {
1138 .procname = "oom_kill_allocating_task",
1139 .data = &sysctl_oom_kill_allocating_task,
1140 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1141 .mode = 0644,
1142 .proc_handler = proc_dointvec,
1143 },
1144 {
1145 .procname = "oom_dump_tasks",
1146 .data = &sysctl_oom_dump_tasks,
1147 .maxlen = sizeof(sysctl_oom_dump_tasks),
1148 .mode = 0644,
1149 .proc_handler = proc_dointvec,
1150 },
1151 {
1152 .procname = "overcommit_ratio",
1153 .data = &sysctl_overcommit_ratio,
1154 .maxlen = sizeof(sysctl_overcommit_ratio),
1155 .mode = 0644,
1156 .proc_handler = overcommit_ratio_handler,
1157 },
1158 {
1159 .procname = "overcommit_kbytes",
1160 .data = &sysctl_overcommit_kbytes,
1161 .maxlen = sizeof(sysctl_overcommit_kbytes),
1162 .mode = 0644,
1163 .proc_handler = overcommit_kbytes_handler,
1164 },
1165 {
1166 .procname = "page-cluster",
1167 .data = &page_cluster,
1168 .maxlen = sizeof(int),
1169 .mode = 0644,
1170 .proc_handler = proc_dointvec_minmax,
1171 .extra1 = &zero,
1172 },
1173 {
1174 .procname = "dirty_background_ratio",
1175 .data = &dirty_background_ratio,
1176 .maxlen = sizeof(dirty_background_ratio),
1177 .mode = 0644,
1178 .proc_handler = dirty_background_ratio_handler,
1179 .extra1 = &zero,
1180 .extra2 = &one_hundred,
1181 },
1182 {
1183 .procname = "dirty_background_bytes",
1184 .data = &dirty_background_bytes,
1185 .maxlen = sizeof(dirty_background_bytes),
1186 .mode = 0644,
1187 .proc_handler = dirty_background_bytes_handler,
1188 .extra1 = &one_ul,
1189 },
1190 {
1191 .procname = "dirty_ratio",
1192 .data = &vm_dirty_ratio,
1193 .maxlen = sizeof(vm_dirty_ratio),
1194 .mode = 0644,
1195 .proc_handler = dirty_ratio_handler,
1196 .extra1 = &zero,
1197 .extra2 = &one_hundred,
1198 },
1199 {
1200 .procname = "dirty_bytes",
1201 .data = &vm_dirty_bytes,
1202 .maxlen = sizeof(vm_dirty_bytes),
1203 .mode = 0644,
1204 .proc_handler = dirty_bytes_handler,
1205 .extra1 = &dirty_bytes_min,
1206 },
1207 {
1208 .procname = "dirty_writeback_centisecs",
1209 .data = &dirty_writeback_interval,
1210 .maxlen = sizeof(dirty_writeback_interval),
1211 .mode = 0644,
1212 .proc_handler = dirty_writeback_centisecs_handler,
1213 },
1214 {
1215 .procname = "dirty_expire_centisecs",
1216 .data = &dirty_expire_interval,
1217 .maxlen = sizeof(dirty_expire_interval),
1218 .mode = 0644,
1219 .proc_handler = proc_dointvec_minmax,
1220 .extra1 = &zero,
1221 },
1222 {
1223 .procname = "nr_pdflush_threads",
1224 .mode = 0444 /* read-only */,
1225 .proc_handler = pdflush_proc_obsolete,
1226 },
1227 {
1228 .procname = "swappiness",
1229 .data = &vm_swappiness,
1230 .maxlen = sizeof(vm_swappiness),
1231 .mode = 0644,
1232 .proc_handler = proc_dointvec_minmax,
1233 .extra1 = &zero,
1234 .extra2 = &one_hundred,
1235 },
1236 #ifdef CONFIG_HUGETLB_PAGE
1237 {
1238 .procname = "nr_hugepages",
1239 .data = NULL,
1240 .maxlen = sizeof(unsigned long),
1241 .mode = 0644,
1242 .proc_handler = hugetlb_sysctl_handler,
1243 .extra1 = &zero,
1244 },
1245 #ifdef CONFIG_NUMA
1246 {
1247 .procname = "nr_hugepages_mempolicy",
1248 .data = NULL,
1249 .maxlen = sizeof(unsigned long),
1250 .mode = 0644,
1251 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1252 .extra1 = &zero,
1253 },
1254 #endif
1255 {
1256 .procname = "hugetlb_shm_group",
1257 .data = &sysctl_hugetlb_shm_group,
1258 .maxlen = sizeof(gid_t),
1259 .mode = 0644,
1260 .proc_handler = proc_dointvec,
1261 },
1262 {
1263 .procname = "hugepages_treat_as_movable",
1264 .data = &hugepages_treat_as_movable,
1265 .maxlen = sizeof(int),
1266 .mode = 0644,
1267 .proc_handler = proc_dointvec,
1268 },
1269 {
1270 .procname = "nr_overcommit_hugepages",
1271 .data = NULL,
1272 .maxlen = sizeof(unsigned long),
1273 .mode = 0644,
1274 .proc_handler = hugetlb_overcommit_handler,
1275 .extra1 = &zero,
1276 },
1277 #endif
1278 {
1279 .procname = "lowmem_reserve_ratio",
1280 .data = &sysctl_lowmem_reserve_ratio,
1281 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1282 .mode = 0644,
1283 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1284 },
1285 {
1286 .procname = "drop_caches",
1287 .data = &sysctl_drop_caches,
1288 .maxlen = sizeof(int),
1289 .mode = 0644,
1290 .proc_handler = drop_caches_sysctl_handler,
1291 .extra1 = &one,
1292 .extra2 = &four,
1293 },
1294 #ifdef CONFIG_COMPACTION
1295 {
1296 .procname = "compact_memory",
1297 .data = &sysctl_compact_memory,
1298 .maxlen = sizeof(int),
1299 .mode = 0200,
1300 .proc_handler = sysctl_compaction_handler,
1301 },
1302 {
1303 .procname = "extfrag_threshold",
1304 .data = &sysctl_extfrag_threshold,
1305 .maxlen = sizeof(int),
1306 .mode = 0644,
1307 .proc_handler = sysctl_extfrag_handler,
1308 .extra1 = &min_extfrag_threshold,
1309 .extra2 = &max_extfrag_threshold,
1310 },
1311
1312 #endif /* CONFIG_COMPACTION */
1313 {
1314 .procname = "min_free_kbytes",
1315 .data = &min_free_kbytes,
1316 .maxlen = sizeof(min_free_kbytes),
1317 .mode = 0644,
1318 .proc_handler = min_free_kbytes_sysctl_handler,
1319 .extra1 = &zero,
1320 },
1321 {
1322 .procname = "percpu_pagelist_fraction",
1323 .data = &percpu_pagelist_fraction,
1324 .maxlen = sizeof(percpu_pagelist_fraction),
1325 .mode = 0644,
1326 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1327 .extra1 = &zero,
1328 },
1329 #ifdef CONFIG_MMU
1330 {
1331 .procname = "max_map_count",
1332 .data = &sysctl_max_map_count,
1333 .maxlen = sizeof(sysctl_max_map_count),
1334 .mode = 0644,
1335 .proc_handler = proc_dointvec_minmax,
1336 .extra1 = &zero,
1337 },
1338 #else
1339 {
1340 .procname = "nr_trim_pages",
1341 .data = &sysctl_nr_trim_pages,
1342 .maxlen = sizeof(sysctl_nr_trim_pages),
1343 .mode = 0644,
1344 .proc_handler = proc_dointvec_minmax,
1345 .extra1 = &zero,
1346 },
1347 #endif
1348 {
1349 .procname = "laptop_mode",
1350 .data = &laptop_mode,
1351 .maxlen = sizeof(laptop_mode),
1352 .mode = 0644,
1353 .proc_handler = proc_dointvec_jiffies,
1354 },
1355 {
1356 .procname = "block_dump",
1357 .data = &block_dump,
1358 .maxlen = sizeof(block_dump),
1359 .mode = 0644,
1360 .proc_handler = proc_dointvec,
1361 .extra1 = &zero,
1362 },
1363 {
1364 .procname = "vfs_cache_pressure",
1365 .data = &sysctl_vfs_cache_pressure,
1366 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1367 .mode = 0644,
1368 .proc_handler = proc_dointvec,
1369 .extra1 = &zero,
1370 },
1371 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1372 {
1373 .procname = "legacy_va_layout",
1374 .data = &sysctl_legacy_va_layout,
1375 .maxlen = sizeof(sysctl_legacy_va_layout),
1376 .mode = 0644,
1377 .proc_handler = proc_dointvec,
1378 .extra1 = &zero,
1379 },
1380 #endif
1381 #ifdef CONFIG_NUMA
1382 {
1383 .procname = "zone_reclaim_mode",
1384 .data = &zone_reclaim_mode,
1385 .maxlen = sizeof(zone_reclaim_mode),
1386 .mode = 0644,
1387 .proc_handler = proc_dointvec,
1388 .extra1 = &zero,
1389 },
1390 {
1391 .procname = "min_unmapped_ratio",
1392 .data = &sysctl_min_unmapped_ratio,
1393 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1394 .mode = 0644,
1395 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1396 .extra1 = &zero,
1397 .extra2 = &one_hundred,
1398 },
1399 {
1400 .procname = "min_slab_ratio",
1401 .data = &sysctl_min_slab_ratio,
1402 .maxlen = sizeof(sysctl_min_slab_ratio),
1403 .mode = 0644,
1404 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1405 .extra1 = &zero,
1406 .extra2 = &one_hundred,
1407 },
1408 #endif
1409 #ifdef CONFIG_SMP
1410 {
1411 .procname = "stat_interval",
1412 .data = &sysctl_stat_interval,
1413 .maxlen = sizeof(sysctl_stat_interval),
1414 .mode = 0644,
1415 .proc_handler = proc_dointvec_jiffies,
1416 },
1417 #endif
1418 #ifdef CONFIG_MMU
1419 {
1420 .procname = "mmap_min_addr",
1421 .data = &dac_mmap_min_addr,
1422 .maxlen = sizeof(unsigned long),
1423 .mode = 0644,
1424 .proc_handler = mmap_min_addr_handler,
1425 },
1426 #endif
1427 #ifdef CONFIG_NUMA
1428 {
1429 .procname = "numa_zonelist_order",
1430 .data = &numa_zonelist_order,
1431 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1432 .mode = 0644,
1433 .proc_handler = numa_zonelist_order_handler,
1434 },
1435 #endif
1436 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1437 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1438 {
1439 .procname = "vdso_enabled",
1440 #ifdef CONFIG_X86_32
1441 .data = &vdso32_enabled,
1442 .maxlen = sizeof(vdso32_enabled),
1443 #else
1444 .data = &vdso_enabled,
1445 .maxlen = sizeof(vdso_enabled),
1446 #endif
1447 .mode = 0644,
1448 .proc_handler = proc_dointvec,
1449 .extra1 = &zero,
1450 },
1451 #endif
1452 #ifdef CONFIG_HIGHMEM
1453 {
1454 .procname = "highmem_is_dirtyable",
1455 .data = &vm_highmem_is_dirtyable,
1456 .maxlen = sizeof(vm_highmem_is_dirtyable),
1457 .mode = 0644,
1458 .proc_handler = proc_dointvec_minmax,
1459 .extra1 = &zero,
1460 .extra2 = &one,
1461 },
1462 #endif
1463 #ifdef CONFIG_MEMORY_FAILURE
1464 {
1465 .procname = "memory_failure_early_kill",
1466 .data = &sysctl_memory_failure_early_kill,
1467 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1468 .mode = 0644,
1469 .proc_handler = proc_dointvec_minmax,
1470 .extra1 = &zero,
1471 .extra2 = &one,
1472 },
1473 {
1474 .procname = "memory_failure_recovery",
1475 .data = &sysctl_memory_failure_recovery,
1476 .maxlen = sizeof(sysctl_memory_failure_recovery),
1477 .mode = 0644,
1478 .proc_handler = proc_dointvec_minmax,
1479 .extra1 = &zero,
1480 .extra2 = &one,
1481 },
1482 #endif
1483 {
1484 .procname = "user_reserve_kbytes",
1485 .data = &sysctl_user_reserve_kbytes,
1486 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1487 .mode = 0644,
1488 .proc_handler = proc_doulongvec_minmax,
1489 },
1490 {
1491 .procname = "admin_reserve_kbytes",
1492 .data = &sysctl_admin_reserve_kbytes,
1493 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1494 .mode = 0644,
1495 .proc_handler = proc_doulongvec_minmax,
1496 },
1497 { }
1498 };
1499
1500 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1501 static struct ctl_table binfmt_misc_table[] = {
1502 { }
1503 };
1504 #endif
1505
1506 static struct ctl_table fs_table[] = {
1507 {
1508 .procname = "inode-nr",
1509 .data = &inodes_stat,
1510 .maxlen = 2*sizeof(long),
1511 .mode = 0444,
1512 .proc_handler = proc_nr_inodes,
1513 },
1514 {
1515 .procname = "inode-state",
1516 .data = &inodes_stat,
1517 .maxlen = 7*sizeof(long),
1518 .mode = 0444,
1519 .proc_handler = proc_nr_inodes,
1520 },
1521 {
1522 .procname = "file-nr",
1523 .data = &files_stat,
1524 .maxlen = sizeof(files_stat),
1525 .mode = 0444,
1526 .proc_handler = proc_nr_files,
1527 },
1528 {
1529 .procname = "file-max",
1530 .data = &files_stat.max_files,
1531 .maxlen = sizeof(files_stat.max_files),
1532 .mode = 0644,
1533 .proc_handler = proc_doulongvec_minmax,
1534 },
1535 {
1536 .procname = "nr_open",
1537 .data = &sysctl_nr_open,
1538 .maxlen = sizeof(int),
1539 .mode = 0644,
1540 .proc_handler = proc_dointvec_minmax,
1541 .extra1 = &sysctl_nr_open_min,
1542 .extra2 = &sysctl_nr_open_max,
1543 },
1544 {
1545 .procname = "dentry-state",
1546 .data = &dentry_stat,
1547 .maxlen = 6*sizeof(long),
1548 .mode = 0444,
1549 .proc_handler = proc_nr_dentry,
1550 },
1551 {
1552 .procname = "overflowuid",
1553 .data = &fs_overflowuid,
1554 .maxlen = sizeof(int),
1555 .mode = 0644,
1556 .proc_handler = proc_dointvec_minmax,
1557 .extra1 = &minolduid,
1558 .extra2 = &maxolduid,
1559 },
1560 {
1561 .procname = "overflowgid",
1562 .data = &fs_overflowgid,
1563 .maxlen = sizeof(int),
1564 .mode = 0644,
1565 .proc_handler = proc_dointvec_minmax,
1566 .extra1 = &minolduid,
1567 .extra2 = &maxolduid,
1568 },
1569 #ifdef CONFIG_FILE_LOCKING
1570 {
1571 .procname = "leases-enable",
1572 .data = &leases_enable,
1573 .maxlen = sizeof(int),
1574 .mode = 0644,
1575 .proc_handler = proc_dointvec,
1576 },
1577 #endif
1578 #ifdef CONFIG_DNOTIFY
1579 {
1580 .procname = "dir-notify-enable",
1581 .data = &dir_notify_enable,
1582 .maxlen = sizeof(int),
1583 .mode = 0644,
1584 .proc_handler = proc_dointvec,
1585 },
1586 #endif
1587 #ifdef CONFIG_MMU
1588 #ifdef CONFIG_FILE_LOCKING
1589 {
1590 .procname = "lease-break-time",
1591 .data = &lease_break_time,
1592 .maxlen = sizeof(int),
1593 .mode = 0644,
1594 .proc_handler = proc_dointvec,
1595 },
1596 #endif
1597 #ifdef CONFIG_AIO
1598 {
1599 .procname = "aio-nr",
1600 .data = &aio_nr,
1601 .maxlen = sizeof(aio_nr),
1602 .mode = 0444,
1603 .proc_handler = proc_doulongvec_minmax,
1604 },
1605 {
1606 .procname = "aio-max-nr",
1607 .data = &aio_max_nr,
1608 .maxlen = sizeof(aio_max_nr),
1609 .mode = 0644,
1610 .proc_handler = proc_doulongvec_minmax,
1611 },
1612 #endif /* CONFIG_AIO */
1613 #ifdef CONFIG_INOTIFY_USER
1614 {
1615 .procname = "inotify",
1616 .mode = 0555,
1617 .child = inotify_table,
1618 },
1619 #endif
1620 #ifdef CONFIG_EPOLL
1621 {
1622 .procname = "epoll",
1623 .mode = 0555,
1624 .child = epoll_table,
1625 },
1626 #endif
1627 #endif
1628 {
1629 .procname = "protected_symlinks",
1630 .data = &sysctl_protected_symlinks,
1631 .maxlen = sizeof(int),
1632 .mode = 0600,
1633 .proc_handler = proc_dointvec_minmax,
1634 .extra1 = &zero,
1635 .extra2 = &one,
1636 },
1637 {
1638 .procname = "protected_hardlinks",
1639 .data = &sysctl_protected_hardlinks,
1640 .maxlen = sizeof(int),
1641 .mode = 0600,
1642 .proc_handler = proc_dointvec_minmax,
1643 .extra1 = &zero,
1644 .extra2 = &one,
1645 },
1646 {
1647 .procname = "suid_dumpable",
1648 .data = &suid_dumpable,
1649 .maxlen = sizeof(int),
1650 .mode = 0644,
1651 .proc_handler = proc_dointvec_minmax_coredump,
1652 .extra1 = &zero,
1653 .extra2 = &two,
1654 },
1655 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1656 {
1657 .procname = "binfmt_misc",
1658 .mode = 0555,
1659 .child = binfmt_misc_table,
1660 },
1661 #endif
1662 {
1663 .procname = "pipe-max-size",
1664 .data = &pipe_max_size,
1665 .maxlen = sizeof(int),
1666 .mode = 0644,
1667 .proc_handler = &pipe_proc_fn,
1668 .extra1 = &pipe_min_size,
1669 },
1670 { }
1671 };
1672
1673 static struct ctl_table debug_table[] = {
1674 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1675 {
1676 .procname = "exception-trace",
1677 .data = &show_unhandled_signals,
1678 .maxlen = sizeof(int),
1679 .mode = 0644,
1680 .proc_handler = proc_dointvec
1681 },
1682 #endif
1683 #if defined(CONFIG_OPTPROBES)
1684 {
1685 .procname = "kprobes-optimization",
1686 .data = &sysctl_kprobes_optimization,
1687 .maxlen = sizeof(int),
1688 .mode = 0644,
1689 .proc_handler = proc_kprobes_optimization_handler,
1690 .extra1 = &zero,
1691 .extra2 = &one,
1692 },
1693 #endif
1694 { }
1695 };
1696
1697 static struct ctl_table dev_table[] = {
1698 { }
1699 };
1700
1701 int __init sysctl_init(void)
1702 {
1703 struct ctl_table_header *hdr;
1704
1705 hdr = register_sysctl_table(sysctl_base_table);
1706 kmemleak_not_leak(hdr);
1707 return 0;
1708 }
1709
1710 #endif /* CONFIG_SYSCTL */
1711
1712 /*
1713 * /proc/sys support
1714 */
1715
1716 #ifdef CONFIG_PROC_SYSCTL
1717
1718 static int _proc_do_string(char *data, int maxlen, int write,
1719 char __user *buffer,
1720 size_t *lenp, loff_t *ppos)
1721 {
1722 size_t len;
1723 char __user *p;
1724 char c;
1725
1726 if (!data || !maxlen || !*lenp) {
1727 *lenp = 0;
1728 return 0;
1729 }
1730
1731 if (write) {
1732 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1733 /* Only continue writes not past the end of buffer. */
1734 len = strlen(data);
1735 if (len > maxlen - 1)
1736 len = maxlen - 1;
1737
1738 if (*ppos > len)
1739 return 0;
1740 len = *ppos;
1741 } else {
1742 /* Start writing from beginning of buffer. */
1743 len = 0;
1744 }
1745
1746 *ppos += *lenp;
1747 p = buffer;
1748 while ((p - buffer) < *lenp && len < maxlen - 1) {
1749 if (get_user(c, p++))
1750 return -EFAULT;
1751 if (c == 0 || c == '\n')
1752 break;
1753 data[len++] = c;
1754 }
1755 data[len] = 0;
1756 } else {
1757 len = strlen(data);
1758 if (len > maxlen)
1759 len = maxlen;
1760
1761 if (*ppos > len) {
1762 *lenp = 0;
1763 return 0;
1764 }
1765
1766 data += *ppos;
1767 len -= *ppos;
1768
1769 if (len > *lenp)
1770 len = *lenp;
1771 if (len)
1772 if (copy_to_user(buffer, data, len))
1773 return -EFAULT;
1774 if (len < *lenp) {
1775 if (put_user('\n', buffer + len))
1776 return -EFAULT;
1777 len++;
1778 }
1779 *lenp = len;
1780 *ppos += len;
1781 }
1782 return 0;
1783 }
1784
1785 static void warn_sysctl_write(struct ctl_table *table)
1786 {
1787 pr_warn_once("%s wrote to %s when file position was not 0!\n"
1788 "This will not be supported in the future. To silence this\n"
1789 "warning, set kernel.sysctl_writes_strict = -1\n",
1790 current->comm, table->procname);
1791 }
1792
1793 /**
1794 * proc_dostring - read a string sysctl
1795 * @table: the sysctl table
1796 * @write: %TRUE if this is a write to the sysctl file
1797 * @buffer: the user buffer
1798 * @lenp: the size of the user buffer
1799 * @ppos: file position
1800 *
1801 * Reads/writes a string from/to the user buffer. If the kernel
1802 * buffer provided is not large enough to hold the string, the
1803 * string is truncated. The copied string is %NULL-terminated.
1804 * If the string is being read by the user process, it is copied
1805 * and a newline '\n' is added. It is truncated if the buffer is
1806 * not large enough.
1807 *
1808 * Returns 0 on success.
1809 */
1810 int proc_dostring(struct ctl_table *table, int write,
1811 void __user *buffer, size_t *lenp, loff_t *ppos)
1812 {
1813 if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN)
1814 warn_sysctl_write(table);
1815
1816 return _proc_do_string((char *)(table->data), table->maxlen, write,
1817 (char __user *)buffer, lenp, ppos);
1818 }
1819
1820 static size_t proc_skip_spaces(char **buf)
1821 {
1822 size_t ret;
1823 char *tmp = skip_spaces(*buf);
1824 ret = tmp - *buf;
1825 *buf = tmp;
1826 return ret;
1827 }
1828
1829 static void proc_skip_char(char **buf, size_t *size, const char v)
1830 {
1831 while (*size) {
1832 if (**buf != v)
1833 break;
1834 (*size)--;
1835 (*buf)++;
1836 }
1837 }
1838
1839 #define TMPBUFLEN 22
1840 /**
1841 * proc_get_long - reads an ASCII formatted integer from a user buffer
1842 *
1843 * @buf: a kernel buffer
1844 * @size: size of the kernel buffer
1845 * @val: this is where the number will be stored
1846 * @neg: set to %TRUE if number is negative
1847 * @perm_tr: a vector which contains the allowed trailers
1848 * @perm_tr_len: size of the perm_tr vector
1849 * @tr: pointer to store the trailer character
1850 *
1851 * In case of success %0 is returned and @buf and @size are updated with
1852 * the amount of bytes read. If @tr is non-NULL and a trailing
1853 * character exists (size is non-zero after returning from this
1854 * function), @tr is updated with the trailing character.
1855 */
1856 static int proc_get_long(char **buf, size_t *size,
1857 unsigned long *val, bool *neg,
1858 const char *perm_tr, unsigned perm_tr_len, char *tr)
1859 {
1860 int len;
1861 char *p, tmp[TMPBUFLEN];
1862
1863 if (!*size)
1864 return -EINVAL;
1865
1866 len = *size;
1867 if (len > TMPBUFLEN - 1)
1868 len = TMPBUFLEN - 1;
1869
1870 memcpy(tmp, *buf, len);
1871
1872 tmp[len] = 0;
1873 p = tmp;
1874 if (*p == '-' && *size > 1) {
1875 *neg = true;
1876 p++;
1877 } else
1878 *neg = false;
1879 if (!isdigit(*p))
1880 return -EINVAL;
1881
1882 *val = simple_strtoul(p, &p, 0);
1883
1884 len = p - tmp;
1885
1886 /* We don't know if the next char is whitespace thus we may accept
1887 * invalid integers (e.g. 1234...a) or two integers instead of one
1888 * (e.g. 123...1). So lets not allow such large numbers. */
1889 if (len == TMPBUFLEN - 1)
1890 return -EINVAL;
1891
1892 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
1893 return -EINVAL;
1894
1895 if (tr && (len < *size))
1896 *tr = *p;
1897
1898 *buf += len;
1899 *size -= len;
1900
1901 return 0;
1902 }
1903
1904 /**
1905 * proc_put_long - converts an integer to a decimal ASCII formatted string
1906 *
1907 * @buf: the user buffer
1908 * @size: the size of the user buffer
1909 * @val: the integer to be converted
1910 * @neg: sign of the number, %TRUE for negative
1911 *
1912 * In case of success %0 is returned and @buf and @size are updated with
1913 * the amount of bytes written.
1914 */
1915 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
1916 bool neg)
1917 {
1918 int len;
1919 char tmp[TMPBUFLEN], *p = tmp;
1920
1921 sprintf(p, "%s%lu", neg ? "-" : "", val);
1922 len = strlen(tmp);
1923 if (len > *size)
1924 len = *size;
1925 if (copy_to_user(*buf, tmp, len))
1926 return -EFAULT;
1927 *size -= len;
1928 *buf += len;
1929 return 0;
1930 }
1931 #undef TMPBUFLEN
1932
1933 static int proc_put_char(void __user **buf, size_t *size, char c)
1934 {
1935 if (*size) {
1936 char __user **buffer = (char __user **)buf;
1937 if (put_user(c, *buffer))
1938 return -EFAULT;
1939 (*size)--, (*buffer)++;
1940 *buf = *buffer;
1941 }
1942 return 0;
1943 }
1944
1945 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
1946 int *valp,
1947 int write, void *data)
1948 {
1949 if (write) {
1950 *valp = *negp ? -*lvalp : *lvalp;
1951 } else {
1952 int val = *valp;
1953 if (val < 0) {
1954 *negp = true;
1955 *lvalp = (unsigned long)-val;
1956 } else {
1957 *negp = false;
1958 *lvalp = (unsigned long)val;
1959 }
1960 }
1961 return 0;
1962 }
1963
1964 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
1965
1966 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1967 int write, void __user *buffer,
1968 size_t *lenp, loff_t *ppos,
1969 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1970 int write, void *data),
1971 void *data)
1972 {
1973 int *i, vleft, first = 1, err = 0;
1974 unsigned long page = 0;
1975 size_t left;
1976 char *kbuf;
1977
1978 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
1979 *lenp = 0;
1980 return 0;
1981 }
1982
1983 i = (int *) tbl_data;
1984 vleft = table->maxlen / sizeof(*i);
1985 left = *lenp;
1986
1987 if (!conv)
1988 conv = do_proc_dointvec_conv;
1989
1990 if (write) {
1991 if (*ppos) {
1992 switch (sysctl_writes_strict) {
1993 case SYSCTL_WRITES_STRICT:
1994 goto out;
1995 case SYSCTL_WRITES_WARN:
1996 warn_sysctl_write(table);
1997 break;
1998 default:
1999 break;
2000 }
2001 }
2002
2003 if (left > PAGE_SIZE - 1)
2004 left = PAGE_SIZE - 1;
2005 page = __get_free_page(GFP_TEMPORARY);
2006 kbuf = (char *) page;
2007 if (!kbuf)
2008 return -ENOMEM;
2009 if (copy_from_user(kbuf, buffer, left)) {
2010 err = -EFAULT;
2011 goto free;
2012 }
2013 kbuf[left] = 0;
2014 }
2015
2016 for (; left && vleft--; i++, first=0) {
2017 unsigned long lval;
2018 bool neg;
2019
2020 if (write) {
2021 left -= proc_skip_spaces(&kbuf);
2022
2023 if (!left)
2024 break;
2025 err = proc_get_long(&kbuf, &left, &lval, &neg,
2026 proc_wspace_sep,
2027 sizeof(proc_wspace_sep), NULL);
2028 if (err)
2029 break;
2030 if (conv(&neg, &lval, i, 1, data)) {
2031 err = -EINVAL;
2032 break;
2033 }
2034 } else {
2035 if (conv(&neg, &lval, i, 0, data)) {
2036 err = -EINVAL;
2037 break;
2038 }
2039 if (!first)
2040 err = proc_put_char(&buffer, &left, '\t');
2041 if (err)
2042 break;
2043 err = proc_put_long(&buffer, &left, lval, neg);
2044 if (err)
2045 break;
2046 }
2047 }
2048
2049 if (!write && !first && left && !err)
2050 err = proc_put_char(&buffer, &left, '\n');
2051 if (write && !err && left)
2052 left -= proc_skip_spaces(&kbuf);
2053 free:
2054 if (write) {
2055 free_page(page);
2056 if (first)
2057 return err ? : -EINVAL;
2058 }
2059 *lenp -= left;
2060 out:
2061 *ppos += *lenp;
2062 return err;
2063 }
2064
2065 static int do_proc_dointvec(struct ctl_table *table, int write,
2066 void __user *buffer, size_t *lenp, loff_t *ppos,
2067 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2068 int write, void *data),
2069 void *data)
2070 {
2071 return __do_proc_dointvec(table->data, table, write,
2072 buffer, lenp, ppos, conv, data);
2073 }
2074
2075 /**
2076 * proc_dointvec - read a vector of integers
2077 * @table: the sysctl table
2078 * @write: %TRUE if this is a write to the sysctl file
2079 * @buffer: the user buffer
2080 * @lenp: the size of the user buffer
2081 * @ppos: file position
2082 *
2083 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2084 * values from/to the user buffer, treated as an ASCII string.
2085 *
2086 * Returns 0 on success.
2087 */
2088 int proc_dointvec(struct ctl_table *table, int write,
2089 void __user *buffer, size_t *lenp, loff_t *ppos)
2090 {
2091 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2092 NULL,NULL);
2093 }
2094
2095 /*
2096 * Taint values can only be increased
2097 * This means we can safely use a temporary.
2098 */
2099 static int proc_taint(struct ctl_table *table, int write,
2100 void __user *buffer, size_t *lenp, loff_t *ppos)
2101 {
2102 struct ctl_table t;
2103 unsigned long tmptaint = get_taint();
2104 int err;
2105
2106 if (write && !capable(CAP_SYS_ADMIN))
2107 return -EPERM;
2108
2109 t = *table;
2110 t.data = &tmptaint;
2111 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2112 if (err < 0)
2113 return err;
2114
2115 if (write) {
2116 /*
2117 * Poor man's atomic or. Not worth adding a primitive
2118 * to everyone's atomic.h for this
2119 */
2120 int i;
2121 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2122 if ((tmptaint >> i) & 1)
2123 add_taint(i, LOCKDEP_STILL_OK);
2124 }
2125 }
2126
2127 return err;
2128 }
2129
2130 #ifdef CONFIG_PRINTK
2131 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2132 void __user *buffer, size_t *lenp, loff_t *ppos)
2133 {
2134 if (write && !capable(CAP_SYS_ADMIN))
2135 return -EPERM;
2136
2137 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2138 }
2139 #endif
2140
2141 struct do_proc_dointvec_minmax_conv_param {
2142 int *min;
2143 int *max;
2144 };
2145
2146 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2147 int *valp,
2148 int write, void *data)
2149 {
2150 struct do_proc_dointvec_minmax_conv_param *param = data;
2151 if (write) {
2152 int val = *negp ? -*lvalp : *lvalp;
2153 if ((param->min && *param->min > val) ||
2154 (param->max && *param->max < val))
2155 return -EINVAL;
2156 *valp = val;
2157 } else {
2158 int val = *valp;
2159 if (val < 0) {
2160 *negp = true;
2161 *lvalp = (unsigned long)-val;
2162 } else {
2163 *negp = false;
2164 *lvalp = (unsigned long)val;
2165 }
2166 }
2167 return 0;
2168 }
2169
2170 /**
2171 * proc_dointvec_minmax - read a vector of integers with min/max values
2172 * @table: the sysctl table
2173 * @write: %TRUE if this is a write to the sysctl file
2174 * @buffer: the user buffer
2175 * @lenp: the size of the user buffer
2176 * @ppos: file position
2177 *
2178 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2179 * values from/to the user buffer, treated as an ASCII string.
2180 *
2181 * This routine will ensure the values are within the range specified by
2182 * table->extra1 (min) and table->extra2 (max).
2183 *
2184 * Returns 0 on success.
2185 */
2186 int proc_dointvec_minmax(struct ctl_table *table, int write,
2187 void __user *buffer, size_t *lenp, loff_t *ppos)
2188 {
2189 struct do_proc_dointvec_minmax_conv_param param = {
2190 .min = (int *) table->extra1,
2191 .max = (int *) table->extra2,
2192 };
2193 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2194 do_proc_dointvec_minmax_conv, &param);
2195 }
2196
2197 static void validate_coredump_safety(void)
2198 {
2199 #ifdef CONFIG_COREDUMP
2200 if (suid_dumpable == SUID_DUMP_ROOT &&
2201 core_pattern[0] != '/' && core_pattern[0] != '|') {
2202 printk(KERN_WARNING "Unsafe core_pattern used with "\
2203 "suid_dumpable=2. Pipe handler or fully qualified "\
2204 "core dump path required.\n");
2205 }
2206 #endif
2207 }
2208
2209 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2210 void __user *buffer, size_t *lenp, loff_t *ppos)
2211 {
2212 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2213 if (!error)
2214 validate_coredump_safety();
2215 return error;
2216 }
2217
2218 #ifdef CONFIG_COREDUMP
2219 static int proc_dostring_coredump(struct ctl_table *table, int write,
2220 void __user *buffer, size_t *lenp, loff_t *ppos)
2221 {
2222 int error = proc_dostring(table, write, buffer, lenp, ppos);
2223 if (!error)
2224 validate_coredump_safety();
2225 return error;
2226 }
2227 #endif
2228
2229 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2230 void __user *buffer,
2231 size_t *lenp, loff_t *ppos,
2232 unsigned long convmul,
2233 unsigned long convdiv)
2234 {
2235 unsigned long *i, *min, *max;
2236 int vleft, first = 1, err = 0;
2237 unsigned long page = 0;
2238 size_t left;
2239 char *kbuf;
2240
2241 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2242 *lenp = 0;
2243 return 0;
2244 }
2245
2246 i = (unsigned long *) data;
2247 min = (unsigned long *) table->extra1;
2248 max = (unsigned long *) table->extra2;
2249 vleft = table->maxlen / sizeof(unsigned long);
2250 left = *lenp;
2251
2252 if (write) {
2253 if (*ppos) {
2254 switch (sysctl_writes_strict) {
2255 case SYSCTL_WRITES_STRICT:
2256 goto out;
2257 case SYSCTL_WRITES_WARN:
2258 warn_sysctl_write(table);
2259 break;
2260 default:
2261 break;
2262 }
2263 }
2264
2265 if (left > PAGE_SIZE - 1)
2266 left = PAGE_SIZE - 1;
2267 page = __get_free_page(GFP_TEMPORARY);
2268 kbuf = (char *) page;
2269 if (!kbuf)
2270 return -ENOMEM;
2271 if (copy_from_user(kbuf, buffer, left)) {
2272 err = -EFAULT;
2273 goto free;
2274 }
2275 kbuf[left] = 0;
2276 }
2277
2278 for (; left && vleft--; i++, first = 0) {
2279 unsigned long val;
2280
2281 if (write) {
2282 bool neg;
2283
2284 left -= proc_skip_spaces(&kbuf);
2285
2286 err = proc_get_long(&kbuf, &left, &val, &neg,
2287 proc_wspace_sep,
2288 sizeof(proc_wspace_sep), NULL);
2289 if (err)
2290 break;
2291 if (neg)
2292 continue;
2293 if ((min && val < *min) || (max && val > *max))
2294 continue;
2295 *i = val;
2296 } else {
2297 val = convdiv * (*i) / convmul;
2298 if (!first) {
2299 err = proc_put_char(&buffer, &left, '\t');
2300 if (err)
2301 break;
2302 }
2303 err = proc_put_long(&buffer, &left, val, false);
2304 if (err)
2305 break;
2306 }
2307 }
2308
2309 if (!write && !first && left && !err)
2310 err = proc_put_char(&buffer, &left, '\n');
2311 if (write && !err)
2312 left -= proc_skip_spaces(&kbuf);
2313 free:
2314 if (write) {
2315 free_page(page);
2316 if (first)
2317 return err ? : -EINVAL;
2318 }
2319 *lenp -= left;
2320 out:
2321 *ppos += *lenp;
2322 return err;
2323 }
2324
2325 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2326 void __user *buffer,
2327 size_t *lenp, loff_t *ppos,
2328 unsigned long convmul,
2329 unsigned long convdiv)
2330 {
2331 return __do_proc_doulongvec_minmax(table->data, table, write,
2332 buffer, lenp, ppos, convmul, convdiv);
2333 }
2334
2335 /**
2336 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2337 * @table: the sysctl table
2338 * @write: %TRUE if this is a write to the sysctl file
2339 * @buffer: the user buffer
2340 * @lenp: the size of the user buffer
2341 * @ppos: file position
2342 *
2343 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2344 * values from/to the user buffer, treated as an ASCII string.
2345 *
2346 * This routine will ensure the values are within the range specified by
2347 * table->extra1 (min) and table->extra2 (max).
2348 *
2349 * Returns 0 on success.
2350 */
2351 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2352 void __user *buffer, size_t *lenp, loff_t *ppos)
2353 {
2354 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2355 }
2356
2357 /**
2358 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2359 * @table: the sysctl table
2360 * @write: %TRUE if this is a write to the sysctl file
2361 * @buffer: the user buffer
2362 * @lenp: the size of the user buffer
2363 * @ppos: file position
2364 *
2365 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2366 * values from/to the user buffer, treated as an ASCII string. The values
2367 * are treated as milliseconds, and converted to jiffies when they are stored.
2368 *
2369 * This routine will ensure the values are within the range specified by
2370 * table->extra1 (min) and table->extra2 (max).
2371 *
2372 * Returns 0 on success.
2373 */
2374 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2375 void __user *buffer,
2376 size_t *lenp, loff_t *ppos)
2377 {
2378 return do_proc_doulongvec_minmax(table, write, buffer,
2379 lenp, ppos, HZ, 1000l);
2380 }
2381
2382
2383 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2384 int *valp,
2385 int write, void *data)
2386 {
2387 if (write) {
2388 if (*lvalp > LONG_MAX / HZ)
2389 return 1;
2390 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2391 } else {
2392 int val = *valp;
2393 unsigned long lval;
2394 if (val < 0) {
2395 *negp = true;
2396 lval = (unsigned long)-val;
2397 } else {
2398 *negp = false;
2399 lval = (unsigned long)val;
2400 }
2401 *lvalp = lval / HZ;
2402 }
2403 return 0;
2404 }
2405
2406 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2407 int *valp,
2408 int write, void *data)
2409 {
2410 if (write) {
2411 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2412 return 1;
2413 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2414 } else {
2415 int val = *valp;
2416 unsigned long lval;
2417 if (val < 0) {
2418 *negp = true;
2419 lval = (unsigned long)-val;
2420 } else {
2421 *negp = false;
2422 lval = (unsigned long)val;
2423 }
2424 *lvalp = jiffies_to_clock_t(lval);
2425 }
2426 return 0;
2427 }
2428
2429 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2430 int *valp,
2431 int write, void *data)
2432 {
2433 if (write) {
2434 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2435
2436 if (jif > INT_MAX)
2437 return 1;
2438 *valp = (int)jif;
2439 } else {
2440 int val = *valp;
2441 unsigned long lval;
2442 if (val < 0) {
2443 *negp = true;
2444 lval = (unsigned long)-val;
2445 } else {
2446 *negp = false;
2447 lval = (unsigned long)val;
2448 }
2449 *lvalp = jiffies_to_msecs(lval);
2450 }
2451 return 0;
2452 }
2453
2454 /**
2455 * proc_dointvec_jiffies - read a vector of integers as seconds
2456 * @table: the sysctl table
2457 * @write: %TRUE if this is a write to the sysctl file
2458 * @buffer: the user buffer
2459 * @lenp: the size of the user buffer
2460 * @ppos: file position
2461 *
2462 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2463 * values from/to the user buffer, treated as an ASCII string.
2464 * The values read are assumed to be in seconds, and are converted into
2465 * jiffies.
2466 *
2467 * Returns 0 on success.
2468 */
2469 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2470 void __user *buffer, size_t *lenp, loff_t *ppos)
2471 {
2472 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2473 do_proc_dointvec_jiffies_conv,NULL);
2474 }
2475
2476 /**
2477 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2478 * @table: the sysctl table
2479 * @write: %TRUE if this is a write to the sysctl file
2480 * @buffer: the user buffer
2481 * @lenp: the size of the user buffer
2482 * @ppos: pointer to the file position
2483 *
2484 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2485 * values from/to the user buffer, treated as an ASCII string.
2486 * The values read are assumed to be in 1/USER_HZ seconds, and
2487 * are converted into jiffies.
2488 *
2489 * Returns 0 on success.
2490 */
2491 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2492 void __user *buffer, size_t *lenp, loff_t *ppos)
2493 {
2494 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2495 do_proc_dointvec_userhz_jiffies_conv,NULL);
2496 }
2497
2498 /**
2499 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2500 * @table: the sysctl table
2501 * @write: %TRUE if this is a write to the sysctl file
2502 * @buffer: the user buffer
2503 * @lenp: the size of the user buffer
2504 * @ppos: file position
2505 * @ppos: the current position in the file
2506 *
2507 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2508 * values from/to the user buffer, treated as an ASCII string.
2509 * The values read are assumed to be in 1/1000 seconds, and
2510 * are converted into jiffies.
2511 *
2512 * Returns 0 on success.
2513 */
2514 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2515 void __user *buffer, size_t *lenp, loff_t *ppos)
2516 {
2517 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2518 do_proc_dointvec_ms_jiffies_conv, NULL);
2519 }
2520
2521 static int proc_do_cad_pid(struct ctl_table *table, int write,
2522 void __user *buffer, size_t *lenp, loff_t *ppos)
2523 {
2524 struct pid *new_pid;
2525 pid_t tmp;
2526 int r;
2527
2528 tmp = pid_vnr(cad_pid);
2529
2530 r = __do_proc_dointvec(&tmp, table, write, buffer,
2531 lenp, ppos, NULL, NULL);
2532 if (r || !write)
2533 return r;
2534
2535 new_pid = find_get_pid(tmp);
2536 if (!new_pid)
2537 return -ESRCH;
2538
2539 put_pid(xchg(&cad_pid, new_pid));
2540 return 0;
2541 }
2542
2543 /**
2544 * proc_do_large_bitmap - read/write from/to a large bitmap
2545 * @table: the sysctl table
2546 * @write: %TRUE if this is a write to the sysctl file
2547 * @buffer: the user buffer
2548 * @lenp: the size of the user buffer
2549 * @ppos: file position
2550 *
2551 * The bitmap is stored at table->data and the bitmap length (in bits)
2552 * in table->maxlen.
2553 *
2554 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2555 * large bitmaps may be represented in a compact manner. Writing into
2556 * the file will clear the bitmap then update it with the given input.
2557 *
2558 * Returns 0 on success.
2559 */
2560 int proc_do_large_bitmap(struct ctl_table *table, int write,
2561 void __user *buffer, size_t *lenp, loff_t *ppos)
2562 {
2563 int err = 0;
2564 bool first = 1;
2565 size_t left = *lenp;
2566 unsigned long bitmap_len = table->maxlen;
2567 unsigned long *bitmap = *(unsigned long **) table->data;
2568 unsigned long *tmp_bitmap = NULL;
2569 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2570
2571 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2572 *lenp = 0;
2573 return 0;
2574 }
2575
2576 if (write) {
2577 unsigned long page = 0;
2578 char *kbuf;
2579
2580 if (left > PAGE_SIZE - 1)
2581 left = PAGE_SIZE - 1;
2582
2583 page = __get_free_page(GFP_TEMPORARY);
2584 kbuf = (char *) page;
2585 if (!kbuf)
2586 return -ENOMEM;
2587 if (copy_from_user(kbuf, buffer, left)) {
2588 free_page(page);
2589 return -EFAULT;
2590 }
2591 kbuf[left] = 0;
2592
2593 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2594 GFP_KERNEL);
2595 if (!tmp_bitmap) {
2596 free_page(page);
2597 return -ENOMEM;
2598 }
2599 proc_skip_char(&kbuf, &left, '\n');
2600 while (!err && left) {
2601 unsigned long val_a, val_b;
2602 bool neg;
2603
2604 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2605 sizeof(tr_a), &c);
2606 if (err)
2607 break;
2608 if (val_a >= bitmap_len || neg) {
2609 err = -EINVAL;
2610 break;
2611 }
2612
2613 val_b = val_a;
2614 if (left) {
2615 kbuf++;
2616 left--;
2617 }
2618
2619 if (c == '-') {
2620 err = proc_get_long(&kbuf, &left, &val_b,
2621 &neg, tr_b, sizeof(tr_b),
2622 &c);
2623 if (err)
2624 break;
2625 if (val_b >= bitmap_len || neg ||
2626 val_a > val_b) {
2627 err = -EINVAL;
2628 break;
2629 }
2630 if (left) {
2631 kbuf++;
2632 left--;
2633 }
2634 }
2635
2636 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2637 first = 0;
2638 proc_skip_char(&kbuf, &left, '\n');
2639 }
2640 free_page(page);
2641 } else {
2642 unsigned long bit_a, bit_b = 0;
2643
2644 while (left) {
2645 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2646 if (bit_a >= bitmap_len)
2647 break;
2648 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2649 bit_a + 1) - 1;
2650
2651 if (!first) {
2652 err = proc_put_char(&buffer, &left, ',');
2653 if (err)
2654 break;
2655 }
2656 err = proc_put_long(&buffer, &left, bit_a, false);
2657 if (err)
2658 break;
2659 if (bit_a != bit_b) {
2660 err = proc_put_char(&buffer, &left, '-');
2661 if (err)
2662 break;
2663 err = proc_put_long(&buffer, &left, bit_b, false);
2664 if (err)
2665 break;
2666 }
2667
2668 first = 0; bit_b++;
2669 }
2670 if (!err)
2671 err = proc_put_char(&buffer, &left, '\n');
2672 }
2673
2674 if (!err) {
2675 if (write) {
2676 if (*ppos)
2677 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2678 else
2679 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2680 }
2681 kfree(tmp_bitmap);
2682 *lenp -= left;
2683 *ppos += *lenp;
2684 return 0;
2685 } else {
2686 kfree(tmp_bitmap);
2687 return err;
2688 }
2689 }
2690
2691 #else /* CONFIG_PROC_SYSCTL */
2692
2693 int proc_dostring(struct ctl_table *table, int write,
2694 void __user *buffer, size_t *lenp, loff_t *ppos)
2695 {
2696 return -ENOSYS;
2697 }
2698
2699 int proc_dointvec(struct ctl_table *table, int write,
2700 void __user *buffer, size_t *lenp, loff_t *ppos)
2701 {
2702 return -ENOSYS;
2703 }
2704
2705 int proc_dointvec_minmax(struct ctl_table *table, int write,
2706 void __user *buffer, size_t *lenp, loff_t *ppos)
2707 {
2708 return -ENOSYS;
2709 }
2710
2711 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2712 void __user *buffer, size_t *lenp, loff_t *ppos)
2713 {
2714 return -ENOSYS;
2715 }
2716
2717 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2718 void __user *buffer, size_t *lenp, loff_t *ppos)
2719 {
2720 return -ENOSYS;
2721 }
2722
2723 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2724 void __user *buffer, size_t *lenp, loff_t *ppos)
2725 {
2726 return -ENOSYS;
2727 }
2728
2729 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2730 void __user *buffer, size_t *lenp, loff_t *ppos)
2731 {
2732 return -ENOSYS;
2733 }
2734
2735 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2736 void __user *buffer,
2737 size_t *lenp, loff_t *ppos)
2738 {
2739 return -ENOSYS;
2740 }
2741
2742
2743 #endif /* CONFIG_PROC_SYSCTL */
2744
2745 /*
2746 * No sense putting this after each symbol definition, twice,
2747 * exception granted :-)
2748 */
2749 EXPORT_SYMBOL(proc_dointvec);
2750 EXPORT_SYMBOL(proc_dointvec_jiffies);
2751 EXPORT_SYMBOL(proc_dointvec_minmax);
2752 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2753 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2754 EXPORT_SYMBOL(proc_dostring);
2755 EXPORT_SYMBOL(proc_doulongvec_minmax);
2756 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
Linux kernel - Deliverables
This page took 0.087153 seconds and 5 git commands to generate.