projects / deliverable / linux.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/trivial
[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 {
1464 .procname = "scan_unevictable_pages",
1465 .data = &scan_unevictable_pages,
1466 .maxlen = sizeof(scan_unevictable_pages),
1467 .mode = 0644,
1468 .proc_handler = scan_unevictable_handler,
1469 },
1470 #ifdef CONFIG_MEMORY_FAILURE
1471 {
1472 .procname = "memory_failure_early_kill",
1473 .data = &sysctl_memory_failure_early_kill,
1474 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1475 .mode = 0644,
1476 .proc_handler = proc_dointvec_minmax,
1477 .extra1 = &zero,
1478 .extra2 = &one,
1479 },
1480 {
1481 .procname = "memory_failure_recovery",
1482 .data = &sysctl_memory_failure_recovery,
1483 .maxlen = sizeof(sysctl_memory_failure_recovery),
1484 .mode = 0644,
1485 .proc_handler = proc_dointvec_minmax,
1486 .extra1 = &zero,
1487 .extra2 = &one,
1488 },
1489 #endif
1490 {
1491 .procname = "user_reserve_kbytes",
1492 .data = &sysctl_user_reserve_kbytes,
1493 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1494 .mode = 0644,
1495 .proc_handler = proc_doulongvec_minmax,
1496 },
1497 {
1498 .procname = "admin_reserve_kbytes",
1499 .data = &sysctl_admin_reserve_kbytes,
1500 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1501 .mode = 0644,
1502 .proc_handler = proc_doulongvec_minmax,
1503 },
1504 { }
1505 };
1506
1507 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1508 static struct ctl_table binfmt_misc_table[] = {
1509 { }
1510 };
1511 #endif
1512
1513 static struct ctl_table fs_table[] = {
1514 {
1515 .procname = "inode-nr",
1516 .data = &inodes_stat,
1517 .maxlen = 2*sizeof(long),
1518 .mode = 0444,
1519 .proc_handler = proc_nr_inodes,
1520 },
1521 {
1522 .procname = "inode-state",
1523 .data = &inodes_stat,
1524 .maxlen = 7*sizeof(long),
1525 .mode = 0444,
1526 .proc_handler = proc_nr_inodes,
1527 },
1528 {
1529 .procname = "file-nr",
1530 .data = &files_stat,
1531 .maxlen = sizeof(files_stat),
1532 .mode = 0444,
1533 .proc_handler = proc_nr_files,
1534 },
1535 {
1536 .procname = "file-max",
1537 .data = &files_stat.max_files,
1538 .maxlen = sizeof(files_stat.max_files),
1539 .mode = 0644,
1540 .proc_handler = proc_doulongvec_minmax,
1541 },
1542 {
1543 .procname = "nr_open",
1544 .data = &sysctl_nr_open,
1545 .maxlen = sizeof(int),
1546 .mode = 0644,
1547 .proc_handler = proc_dointvec_minmax,
1548 .extra1 = &sysctl_nr_open_min,
1549 .extra2 = &sysctl_nr_open_max,
1550 },
1551 {
1552 .procname = "dentry-state",
1553 .data = &dentry_stat,
1554 .maxlen = 6*sizeof(long),
1555 .mode = 0444,
1556 .proc_handler = proc_nr_dentry,
1557 },
1558 {
1559 .procname = "overflowuid",
1560 .data = &fs_overflowuid,
1561 .maxlen = sizeof(int),
1562 .mode = 0644,
1563 .proc_handler = proc_dointvec_minmax,
1564 .extra1 = &minolduid,
1565 .extra2 = &maxolduid,
1566 },
1567 {
1568 .procname = "overflowgid",
1569 .data = &fs_overflowgid,
1570 .maxlen = sizeof(int),
1571 .mode = 0644,
1572 .proc_handler = proc_dointvec_minmax,
1573 .extra1 = &minolduid,
1574 .extra2 = &maxolduid,
1575 },
1576 #ifdef CONFIG_FILE_LOCKING
1577 {
1578 .procname = "leases-enable",
1579 .data = &leases_enable,
1580 .maxlen = sizeof(int),
1581 .mode = 0644,
1582 .proc_handler = proc_dointvec,
1583 },
1584 #endif
1585 #ifdef CONFIG_DNOTIFY
1586 {
1587 .procname = "dir-notify-enable",
1588 .data = &dir_notify_enable,
1589 .maxlen = sizeof(int),
1590 .mode = 0644,
1591 .proc_handler = proc_dointvec,
1592 },
1593 #endif
1594 #ifdef CONFIG_MMU
1595 #ifdef CONFIG_FILE_LOCKING
1596 {
1597 .procname = "lease-break-time",
1598 .data = &lease_break_time,
1599 .maxlen = sizeof(int),
1600 .mode = 0644,
1601 .proc_handler = proc_dointvec,
1602 },
1603 #endif
1604 #ifdef CONFIG_AIO
1605 {
1606 .procname = "aio-nr",
1607 .data = &aio_nr,
1608 .maxlen = sizeof(aio_nr),
1609 .mode = 0444,
1610 .proc_handler = proc_doulongvec_minmax,
1611 },
1612 {
1613 .procname = "aio-max-nr",
1614 .data = &aio_max_nr,
1615 .maxlen = sizeof(aio_max_nr),
1616 .mode = 0644,
1617 .proc_handler = proc_doulongvec_minmax,
1618 },
1619 #endif /* CONFIG_AIO */
1620 #ifdef CONFIG_INOTIFY_USER
1621 {
1622 .procname = "inotify",
1623 .mode = 0555,
1624 .child = inotify_table,
1625 },
1626 #endif
1627 #ifdef CONFIG_EPOLL
1628 {
1629 .procname = "epoll",
1630 .mode = 0555,
1631 .child = epoll_table,
1632 },
1633 #endif
1634 #endif
1635 {
1636 .procname = "protected_symlinks",
1637 .data = &sysctl_protected_symlinks,
1638 .maxlen = sizeof(int),
1639 .mode = 0600,
1640 .proc_handler = proc_dointvec_minmax,
1641 .extra1 = &zero,
1642 .extra2 = &one,
1643 },
1644 {
1645 .procname = "protected_hardlinks",
1646 .data = &sysctl_protected_hardlinks,
1647 .maxlen = sizeof(int),
1648 .mode = 0600,
1649 .proc_handler = proc_dointvec_minmax,
1650 .extra1 = &zero,
1651 .extra2 = &one,
1652 },
1653 {
1654 .procname = "suid_dumpable",
1655 .data = &suid_dumpable,
1656 .maxlen = sizeof(int),
1657 .mode = 0644,
1658 .proc_handler = proc_dointvec_minmax_coredump,
1659 .extra1 = &zero,
1660 .extra2 = &two,
1661 },
1662 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1663 {
1664 .procname = "binfmt_misc",
1665 .mode = 0555,
1666 .child = binfmt_misc_table,
1667 },
1668 #endif
1669 {
1670 .procname = "pipe-max-size",
1671 .data = &pipe_max_size,
1672 .maxlen = sizeof(int),
1673 .mode = 0644,
1674 .proc_handler = &pipe_proc_fn,
1675 .extra1 = &pipe_min_size,
1676 },
1677 { }
1678 };
1679
1680 static struct ctl_table debug_table[] = {
1681 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1682 {
1683 .procname = "exception-trace",
1684 .data = &show_unhandled_signals,
1685 .maxlen = sizeof(int),
1686 .mode = 0644,
1687 .proc_handler = proc_dointvec
1688 },
1689 #endif
1690 #if defined(CONFIG_OPTPROBES)
1691 {
1692 .procname = "kprobes-optimization",
1693 .data = &sysctl_kprobes_optimization,
1694 .maxlen = sizeof(int),
1695 .mode = 0644,
1696 .proc_handler = proc_kprobes_optimization_handler,
1697 .extra1 = &zero,
1698 .extra2 = &one,
1699 },
1700 #endif
1701 { }
1702 };
1703
1704 static struct ctl_table dev_table[] = {
1705 { }
1706 };
1707
1708 int __init sysctl_init(void)
1709 {
1710 struct ctl_table_header *hdr;
1711
1712 hdr = register_sysctl_table(sysctl_base_table);
1713 kmemleak_not_leak(hdr);
1714 return 0;
1715 }
1716
1717 #endif /* CONFIG_SYSCTL */
1718
1719 /*
1720 * /proc/sys support
1721 */
1722
1723 #ifdef CONFIG_PROC_SYSCTL
1724
1725 static int _proc_do_string(char *data, int maxlen, int write,
1726 char __user *buffer,
1727 size_t *lenp, loff_t *ppos)
1728 {
1729 size_t len;
1730 char __user *p;
1731 char c;
1732
1733 if (!data || !maxlen || !*lenp) {
1734 *lenp = 0;
1735 return 0;
1736 }
1737
1738 if (write) {
1739 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1740 /* Only continue writes not past the end of buffer. */
1741 len = strlen(data);
1742 if (len > maxlen - 1)
1743 len = maxlen - 1;
1744
1745 if (*ppos > len)
1746 return 0;
1747 len = *ppos;
1748 } else {
1749 /* Start writing from beginning of buffer. */
1750 len = 0;
1751 }
1752
1753 *ppos += *lenp;
1754 p = buffer;
1755 while ((p - buffer) < *lenp && len < maxlen - 1) {
1756 if (get_user(c, p++))
1757 return -EFAULT;
1758 if (c == 0 || c == '\n')
1759 break;
1760 data[len++] = c;
1761 }
1762 data[len] = 0;
1763 } else {
1764 len = strlen(data);
1765 if (len > maxlen)
1766 len = maxlen;
1767
1768 if (*ppos > len) {
1769 *lenp = 0;
1770 return 0;
1771 }
1772
1773 data += *ppos;
1774 len -= *ppos;
1775
1776 if (len > *lenp)
1777 len = *lenp;
1778 if (len)
1779 if (copy_to_user(buffer, data, len))
1780 return -EFAULT;
1781 if (len < *lenp) {
1782 if (put_user('\n', buffer + len))
1783 return -EFAULT;
1784 len++;
1785 }
1786 *lenp = len;
1787 *ppos += len;
1788 }
1789 return 0;
1790 }
1791
1792 static void warn_sysctl_write(struct ctl_table *table)
1793 {
1794 pr_warn_once("%s wrote to %s when file position was not 0!\n"
1795 "This will not be supported in the future. To silence this\n"
1796 "warning, set kernel.sysctl_writes_strict = -1\n",
1797 current->comm, table->procname);
1798 }
1799
1800 /**
1801 * proc_dostring - read a string sysctl
1802 * @table: the sysctl table
1803 * @write: %TRUE if this is a write to the sysctl file
1804 * @buffer: the user buffer
1805 * @lenp: the size of the user buffer
1806 * @ppos: file position
1807 *
1808 * Reads/writes a string from/to the user buffer. If the kernel
1809 * buffer provided is not large enough to hold the string, the
1810 * string is truncated. The copied string is %NULL-terminated.
1811 * If the string is being read by the user process, it is copied
1812 * and a newline '\n' is added. It is truncated if the buffer is
1813 * not large enough.
1814 *
1815 * Returns 0 on success.
1816 */
1817 int proc_dostring(struct ctl_table *table, int write,
1818 void __user *buffer, size_t *lenp, loff_t *ppos)
1819 {
1820 if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN)
1821 warn_sysctl_write(table);
1822
1823 return _proc_do_string((char *)(table->data), table->maxlen, write,
1824 (char __user *)buffer, lenp, ppos);
1825 }
1826
1827 static size_t proc_skip_spaces(char **buf)
1828 {
1829 size_t ret;
1830 char *tmp = skip_spaces(*buf);
1831 ret = tmp - *buf;
1832 *buf = tmp;
1833 return ret;
1834 }
1835
1836 static void proc_skip_char(char **buf, size_t *size, const char v)
1837 {
1838 while (*size) {
1839 if (**buf != v)
1840 break;
1841 (*size)--;
1842 (*buf)++;
1843 }
1844 }
1845
1846 #define TMPBUFLEN 22
1847 /**
1848 * proc_get_long - reads an ASCII formatted integer from a user buffer
1849 *
1850 * @buf: a kernel buffer
1851 * @size: size of the kernel buffer
1852 * @val: this is where the number will be stored
1853 * @neg: set to %TRUE if number is negative
1854 * @perm_tr: a vector which contains the allowed trailers
1855 * @perm_tr_len: size of the perm_tr vector
1856 * @tr: pointer to store the trailer character
1857 *
1858 * In case of success %0 is returned and @buf and @size are updated with
1859 * the amount of bytes read. If @tr is non-NULL and a trailing
1860 * character exists (size is non-zero after returning from this
1861 * function), @tr is updated with the trailing character.
1862 */
1863 static int proc_get_long(char **buf, size_t *size,
1864 unsigned long *val, bool *neg,
1865 const char *perm_tr, unsigned perm_tr_len, char *tr)
1866 {
1867 int len;
1868 char *p, tmp[TMPBUFLEN];
1869
1870 if (!*size)
1871 return -EINVAL;
1872
1873 len = *size;
1874 if (len > TMPBUFLEN - 1)
1875 len = TMPBUFLEN - 1;
1876
1877 memcpy(tmp, *buf, len);
1878
1879 tmp[len] = 0;
1880 p = tmp;
1881 if (*p == '-' && *size > 1) {
1882 *neg = true;
1883 p++;
1884 } else
1885 *neg = false;
1886 if (!isdigit(*p))
1887 return -EINVAL;
1888
1889 *val = simple_strtoul(p, &p, 0);
1890
1891 len = p - tmp;
1892
1893 /* We don't know if the next char is whitespace thus we may accept
1894 * invalid integers (e.g. 1234...a) or two integers instead of one
1895 * (e.g. 123...1). So lets not allow such large numbers. */
1896 if (len == TMPBUFLEN - 1)
1897 return -EINVAL;
1898
1899 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
1900 return -EINVAL;
1901
1902 if (tr && (len < *size))
1903 *tr = *p;
1904
1905 *buf += len;
1906 *size -= len;
1907
1908 return 0;
1909 }
1910
1911 /**
1912 * proc_put_long - converts an integer to a decimal ASCII formatted string
1913 *
1914 * @buf: the user buffer
1915 * @size: the size of the user buffer
1916 * @val: the integer to be converted
1917 * @neg: sign of the number, %TRUE for negative
1918 *
1919 * In case of success %0 is returned and @buf and @size are updated with
1920 * the amount of bytes written.
1921 */
1922 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
1923 bool neg)
1924 {
1925 int len;
1926 char tmp[TMPBUFLEN], *p = tmp;
1927
1928 sprintf(p, "%s%lu", neg ? "-" : "", val);
1929 len = strlen(tmp);
1930 if (len > *size)
1931 len = *size;
1932 if (copy_to_user(*buf, tmp, len))
1933 return -EFAULT;
1934 *size -= len;
1935 *buf += len;
1936 return 0;
1937 }
1938 #undef TMPBUFLEN
1939
1940 static int proc_put_char(void __user **buf, size_t *size, char c)
1941 {
1942 if (*size) {
1943 char __user **buffer = (char __user **)buf;
1944 if (put_user(c, *buffer))
1945 return -EFAULT;
1946 (*size)--, (*buffer)++;
1947 *buf = *buffer;
1948 }
1949 return 0;
1950 }
1951
1952 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
1953 int *valp,
1954 int write, void *data)
1955 {
1956 if (write) {
1957 *valp = *negp ? -*lvalp : *lvalp;
1958 } else {
1959 int val = *valp;
1960 if (val < 0) {
1961 *negp = true;
1962 *lvalp = (unsigned long)-val;
1963 } else {
1964 *negp = false;
1965 *lvalp = (unsigned long)val;
1966 }
1967 }
1968 return 0;
1969 }
1970
1971 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
1972
1973 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1974 int write, void __user *buffer,
1975 size_t *lenp, loff_t *ppos,
1976 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1977 int write, void *data),
1978 void *data)
1979 {
1980 int *i, vleft, first = 1, err = 0;
1981 unsigned long page = 0;
1982 size_t left;
1983 char *kbuf;
1984
1985 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
1986 *lenp = 0;
1987 return 0;
1988 }
1989
1990 i = (int *) tbl_data;
1991 vleft = table->maxlen / sizeof(*i);
1992 left = *lenp;
1993
1994 if (!conv)
1995 conv = do_proc_dointvec_conv;
1996
1997 if (write) {
1998 if (*ppos) {
1999 switch (sysctl_writes_strict) {
2000 case SYSCTL_WRITES_STRICT:
2001 goto out;
2002 case SYSCTL_WRITES_WARN:
2003 warn_sysctl_write(table);
2004 break;
2005 default:
2006 break;
2007 }
2008 }
2009
2010 if (left > PAGE_SIZE - 1)
2011 left = PAGE_SIZE - 1;
2012 page = __get_free_page(GFP_TEMPORARY);
2013 kbuf = (char *) page;
2014 if (!kbuf)
2015 return -ENOMEM;
2016 if (copy_from_user(kbuf, buffer, left)) {
2017 err = -EFAULT;
2018 goto free;
2019 }
2020 kbuf[left] = 0;
2021 }
2022
2023 for (; left && vleft--; i++, first=0) {
2024 unsigned long lval;
2025 bool neg;
2026
2027 if (write) {
2028 left -= proc_skip_spaces(&kbuf);
2029
2030 if (!left)
2031 break;
2032 err = proc_get_long(&kbuf, &left, &lval, &neg,
2033 proc_wspace_sep,
2034 sizeof(proc_wspace_sep), NULL);
2035 if (err)
2036 break;
2037 if (conv(&neg, &lval, i, 1, data)) {
2038 err = -EINVAL;
2039 break;
2040 }
2041 } else {
2042 if (conv(&neg, &lval, i, 0, data)) {
2043 err = -EINVAL;
2044 break;
2045 }
2046 if (!first)
2047 err = proc_put_char(&buffer, &left, '\t');
2048 if (err)
2049 break;
2050 err = proc_put_long(&buffer, &left, lval, neg);
2051 if (err)
2052 break;
2053 }
2054 }
2055
2056 if (!write && !first && left && !err)
2057 err = proc_put_char(&buffer, &left, '\n');
2058 if (write && !err && left)
2059 left -= proc_skip_spaces(&kbuf);
2060 free:
2061 if (write) {
2062 free_page(page);
2063 if (first)
2064 return err ? : -EINVAL;
2065 }
2066 *lenp -= left;
2067 out:
2068 *ppos += *lenp;
2069 return err;
2070 }
2071
2072 static int do_proc_dointvec(struct ctl_table *table, int write,
2073 void __user *buffer, size_t *lenp, loff_t *ppos,
2074 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2075 int write, void *data),
2076 void *data)
2077 {
2078 return __do_proc_dointvec(table->data, table, write,
2079 buffer, lenp, ppos, conv, data);
2080 }
2081
2082 /**
2083 * proc_dointvec - read a vector of integers
2084 * @table: the sysctl table
2085 * @write: %TRUE if this is a write to the sysctl file
2086 * @buffer: the user buffer
2087 * @lenp: the size of the user buffer
2088 * @ppos: file position
2089 *
2090 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2091 * values from/to the user buffer, treated as an ASCII string.
2092 *
2093 * Returns 0 on success.
2094 */
2095 int proc_dointvec(struct ctl_table *table, int write,
2096 void __user *buffer, size_t *lenp, loff_t *ppos)
2097 {
2098 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2099 NULL,NULL);
2100 }
2101
2102 /*
2103 * Taint values can only be increased
2104 * This means we can safely use a temporary.
2105 */
2106 static int proc_taint(struct ctl_table *table, int write,
2107 void __user *buffer, size_t *lenp, loff_t *ppos)
2108 {
2109 struct ctl_table t;
2110 unsigned long tmptaint = get_taint();
2111 int err;
2112
2113 if (write && !capable(CAP_SYS_ADMIN))
2114 return -EPERM;
2115
2116 t = *table;
2117 t.data = &tmptaint;
2118 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2119 if (err < 0)
2120 return err;
2121
2122 if (write) {
2123 /*
2124 * Poor man's atomic or. Not worth adding a primitive
2125 * to everyone's atomic.h for this
2126 */
2127 int i;
2128 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2129 if ((tmptaint >> i) & 1)
2130 add_taint(i, LOCKDEP_STILL_OK);
2131 }
2132 }
2133
2134 return err;
2135 }
2136
2137 #ifdef CONFIG_PRINTK
2138 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2139 void __user *buffer, size_t *lenp, loff_t *ppos)
2140 {
2141 if (write && !capable(CAP_SYS_ADMIN))
2142 return -EPERM;
2143
2144 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2145 }
2146 #endif
2147
2148 struct do_proc_dointvec_minmax_conv_param {
2149 int *min;
2150 int *max;
2151 };
2152
2153 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2154 int *valp,
2155 int write, void *data)
2156 {
2157 struct do_proc_dointvec_minmax_conv_param *param = data;
2158 if (write) {
2159 int val = *negp ? -*lvalp : *lvalp;
2160 if ((param->min && *param->min > val) ||
2161 (param->max && *param->max < val))
2162 return -EINVAL;
2163 *valp = val;
2164 } else {
2165 int val = *valp;
2166 if (val < 0) {
2167 *negp = true;
2168 *lvalp = (unsigned long)-val;
2169 } else {
2170 *negp = false;
2171 *lvalp = (unsigned long)val;
2172 }
2173 }
2174 return 0;
2175 }
2176
2177 /**
2178 * proc_dointvec_minmax - read a vector of integers with min/max values
2179 * @table: the sysctl table
2180 * @write: %TRUE if this is a write to the sysctl file
2181 * @buffer: the user buffer
2182 * @lenp: the size of the user buffer
2183 * @ppos: file position
2184 *
2185 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2186 * values from/to the user buffer, treated as an ASCII string.
2187 *
2188 * This routine will ensure the values are within the range specified by
2189 * table->extra1 (min) and table->extra2 (max).
2190 *
2191 * Returns 0 on success.
2192 */
2193 int proc_dointvec_minmax(struct ctl_table *table, int write,
2194 void __user *buffer, size_t *lenp, loff_t *ppos)
2195 {
2196 struct do_proc_dointvec_minmax_conv_param param = {
2197 .min = (int *) table->extra1,
2198 .max = (int *) table->extra2,
2199 };
2200 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2201 do_proc_dointvec_minmax_conv, &param);
2202 }
2203
2204 static void validate_coredump_safety(void)
2205 {
2206 #ifdef CONFIG_COREDUMP
2207 if (suid_dumpable == SUID_DUMP_ROOT &&
2208 core_pattern[0] != '/' && core_pattern[0] != '|') {
2209 printk(KERN_WARNING "Unsafe core_pattern used with "\
2210 "suid_dumpable=2. Pipe handler or fully qualified "\
2211 "core dump path required.\n");
2212 }
2213 #endif
2214 }
2215
2216 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2217 void __user *buffer, size_t *lenp, loff_t *ppos)
2218 {
2219 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2220 if (!error)
2221 validate_coredump_safety();
2222 return error;
2223 }
2224
2225 #ifdef CONFIG_COREDUMP
2226 static int proc_dostring_coredump(struct ctl_table *table, int write,
2227 void __user *buffer, size_t *lenp, loff_t *ppos)
2228 {
2229 int error = proc_dostring(table, write, buffer, lenp, ppos);
2230 if (!error)
2231 validate_coredump_safety();
2232 return error;
2233 }
2234 #endif
2235
2236 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2237 void __user *buffer,
2238 size_t *lenp, loff_t *ppos,
2239 unsigned long convmul,
2240 unsigned long convdiv)
2241 {
2242 unsigned long *i, *min, *max;
2243 int vleft, first = 1, err = 0;
2244 unsigned long page = 0;
2245 size_t left;
2246 char *kbuf;
2247
2248 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2249 *lenp = 0;
2250 return 0;
2251 }
2252
2253 i = (unsigned long *) data;
2254 min = (unsigned long *) table->extra1;
2255 max = (unsigned long *) table->extra2;
2256 vleft = table->maxlen / sizeof(unsigned long);
2257 left = *lenp;
2258
2259 if (write) {
2260 if (*ppos) {
2261 switch (sysctl_writes_strict) {
2262 case SYSCTL_WRITES_STRICT:
2263 goto out;
2264 case SYSCTL_WRITES_WARN:
2265 warn_sysctl_write(table);
2266 break;
2267 default:
2268 break;
2269 }
2270 }
2271
2272 if (left > PAGE_SIZE - 1)
2273 left = PAGE_SIZE - 1;
2274 page = __get_free_page(GFP_TEMPORARY);
2275 kbuf = (char *) page;
2276 if (!kbuf)
2277 return -ENOMEM;
2278 if (copy_from_user(kbuf, buffer, left)) {
2279 err = -EFAULT;
2280 goto free;
2281 }
2282 kbuf[left] = 0;
2283 }
2284
2285 for (; left && vleft--; i++, first = 0) {
2286 unsigned long val;
2287
2288 if (write) {
2289 bool neg;
2290
2291 left -= proc_skip_spaces(&kbuf);
2292
2293 err = proc_get_long(&kbuf, &left, &val, &neg,
2294 proc_wspace_sep,
2295 sizeof(proc_wspace_sep), NULL);
2296 if (err)
2297 break;
2298 if (neg)
2299 continue;
2300 if ((min && val < *min) || (max && val > *max))
2301 continue;
2302 *i = val;
2303 } else {
2304 val = convdiv * (*i) / convmul;
2305 if (!first) {
2306 err = proc_put_char(&buffer, &left, '\t');
2307 if (err)
2308 break;
2309 }
2310 err = proc_put_long(&buffer, &left, val, false);
2311 if (err)
2312 break;
2313 }
2314 }
2315
2316 if (!write && !first && left && !err)
2317 err = proc_put_char(&buffer, &left, '\n');
2318 if (write && !err)
2319 left -= proc_skip_spaces(&kbuf);
2320 free:
2321 if (write) {
2322 free_page(page);
2323 if (first)
2324 return err ? : -EINVAL;
2325 }
2326 *lenp -= left;
2327 out:
2328 *ppos += *lenp;
2329 return err;
2330 }
2331
2332 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2333 void __user *buffer,
2334 size_t *lenp, loff_t *ppos,
2335 unsigned long convmul,
2336 unsigned long convdiv)
2337 {
2338 return __do_proc_doulongvec_minmax(table->data, table, write,
2339 buffer, lenp, ppos, convmul, convdiv);
2340 }
2341
2342 /**
2343 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2344 * @table: the sysctl table
2345 * @write: %TRUE if this is a write to the sysctl file
2346 * @buffer: the user buffer
2347 * @lenp: the size of the user buffer
2348 * @ppos: file position
2349 *
2350 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2351 * values from/to the user buffer, treated as an ASCII string.
2352 *
2353 * This routine will ensure the values are within the range specified by
2354 * table->extra1 (min) and table->extra2 (max).
2355 *
2356 * Returns 0 on success.
2357 */
2358 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2359 void __user *buffer, size_t *lenp, loff_t *ppos)
2360 {
2361 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2362 }
2363
2364 /**
2365 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2366 * @table: the sysctl table
2367 * @write: %TRUE if this is a write to the sysctl file
2368 * @buffer: the user buffer
2369 * @lenp: the size of the user buffer
2370 * @ppos: file position
2371 *
2372 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2373 * values from/to the user buffer, treated as an ASCII string. The values
2374 * are treated as milliseconds, and converted to jiffies when they are stored.
2375 *
2376 * This routine will ensure the values are within the range specified by
2377 * table->extra1 (min) and table->extra2 (max).
2378 *
2379 * Returns 0 on success.
2380 */
2381 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2382 void __user *buffer,
2383 size_t *lenp, loff_t *ppos)
2384 {
2385 return do_proc_doulongvec_minmax(table, write, buffer,
2386 lenp, ppos, HZ, 1000l);
2387 }
2388
2389
2390 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2391 int *valp,
2392 int write, void *data)
2393 {
2394 if (write) {
2395 if (*lvalp > LONG_MAX / HZ)
2396 return 1;
2397 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2398 } else {
2399 int val = *valp;
2400 unsigned long lval;
2401 if (val < 0) {
2402 *negp = true;
2403 lval = (unsigned long)-val;
2404 } else {
2405 *negp = false;
2406 lval = (unsigned long)val;
2407 }
2408 *lvalp = lval / HZ;
2409 }
2410 return 0;
2411 }
2412
2413 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2414 int *valp,
2415 int write, void *data)
2416 {
2417 if (write) {
2418 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2419 return 1;
2420 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2421 } else {
2422 int val = *valp;
2423 unsigned long lval;
2424 if (val < 0) {
2425 *negp = true;
2426 lval = (unsigned long)-val;
2427 } else {
2428 *negp = false;
2429 lval = (unsigned long)val;
2430 }
2431 *lvalp = jiffies_to_clock_t(lval);
2432 }
2433 return 0;
2434 }
2435
2436 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2437 int *valp,
2438 int write, void *data)
2439 {
2440 if (write) {
2441 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2442
2443 if (jif > INT_MAX)
2444 return 1;
2445 *valp = (int)jif;
2446 } else {
2447 int val = *valp;
2448 unsigned long lval;
2449 if (val < 0) {
2450 *negp = true;
2451 lval = (unsigned long)-val;
2452 } else {
2453 *negp = false;
2454 lval = (unsigned long)val;
2455 }
2456 *lvalp = jiffies_to_msecs(lval);
2457 }
2458 return 0;
2459 }
2460
2461 /**
2462 * proc_dointvec_jiffies - read a vector of integers as seconds
2463 * @table: the sysctl table
2464 * @write: %TRUE if this is a write to the sysctl file
2465 * @buffer: the user buffer
2466 * @lenp: the size of the user buffer
2467 * @ppos: file position
2468 *
2469 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2470 * values from/to the user buffer, treated as an ASCII string.
2471 * The values read are assumed to be in seconds, and are converted into
2472 * jiffies.
2473 *
2474 * Returns 0 on success.
2475 */
2476 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2477 void __user *buffer, size_t *lenp, loff_t *ppos)
2478 {
2479 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2480 do_proc_dointvec_jiffies_conv,NULL);
2481 }
2482
2483 /**
2484 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2485 * @table: the sysctl table
2486 * @write: %TRUE if this is a write to the sysctl file
2487 * @buffer: the user buffer
2488 * @lenp: the size of the user buffer
2489 * @ppos: pointer to the file position
2490 *
2491 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2492 * values from/to the user buffer, treated as an ASCII string.
2493 * The values read are assumed to be in 1/USER_HZ seconds, and
2494 * are converted into jiffies.
2495 *
2496 * Returns 0 on success.
2497 */
2498 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2499 void __user *buffer, size_t *lenp, loff_t *ppos)
2500 {
2501 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2502 do_proc_dointvec_userhz_jiffies_conv,NULL);
2503 }
2504
2505 /**
2506 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2507 * @table: the sysctl table
2508 * @write: %TRUE if this is a write to the sysctl file
2509 * @buffer: the user buffer
2510 * @lenp: the size of the user buffer
2511 * @ppos: file position
2512 * @ppos: the current position in the file
2513 *
2514 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2515 * values from/to the user buffer, treated as an ASCII string.
2516 * The values read are assumed to be in 1/1000 seconds, and
2517 * are converted into jiffies.
2518 *
2519 * Returns 0 on success.
2520 */
2521 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2522 void __user *buffer, size_t *lenp, loff_t *ppos)
2523 {
2524 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2525 do_proc_dointvec_ms_jiffies_conv, NULL);
2526 }
2527
2528 static int proc_do_cad_pid(struct ctl_table *table, int write,
2529 void __user *buffer, size_t *lenp, loff_t *ppos)
2530 {
2531 struct pid *new_pid;
2532 pid_t tmp;
2533 int r;
2534
2535 tmp = pid_vnr(cad_pid);
2536
2537 r = __do_proc_dointvec(&tmp, table, write, buffer,
2538 lenp, ppos, NULL, NULL);
2539 if (r || !write)
2540 return r;
2541
2542 new_pid = find_get_pid(tmp);
2543 if (!new_pid)
2544 return -ESRCH;
2545
2546 put_pid(xchg(&cad_pid, new_pid));
2547 return 0;
2548 }
2549
2550 /**
2551 * proc_do_large_bitmap - read/write from/to a large bitmap
2552 * @table: the sysctl table
2553 * @write: %TRUE if this is a write to the sysctl file
2554 * @buffer: the user buffer
2555 * @lenp: the size of the user buffer
2556 * @ppos: file position
2557 *
2558 * The bitmap is stored at table->data and the bitmap length (in bits)
2559 * in table->maxlen.
2560 *
2561 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2562 * large bitmaps may be represented in a compact manner. Writing into
2563 * the file will clear the bitmap then update it with the given input.
2564 *
2565 * Returns 0 on success.
2566 */
2567 int proc_do_large_bitmap(struct ctl_table *table, int write,
2568 void __user *buffer, size_t *lenp, loff_t *ppos)
2569 {
2570 int err = 0;
2571 bool first = 1;
2572 size_t left = *lenp;
2573 unsigned long bitmap_len = table->maxlen;
2574 unsigned long *bitmap = *(unsigned long **) table->data;
2575 unsigned long *tmp_bitmap = NULL;
2576 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2577
2578 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2579 *lenp = 0;
2580 return 0;
2581 }
2582
2583 if (write) {
2584 unsigned long page = 0;
2585 char *kbuf;
2586
2587 if (left > PAGE_SIZE - 1)
2588 left = PAGE_SIZE - 1;
2589
2590 page = __get_free_page(GFP_TEMPORARY);
2591 kbuf = (char *) page;
2592 if (!kbuf)
2593 return -ENOMEM;
2594 if (copy_from_user(kbuf, buffer, left)) {
2595 free_page(page);
2596 return -EFAULT;
2597 }
2598 kbuf[left] = 0;
2599
2600 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2601 GFP_KERNEL);
2602 if (!tmp_bitmap) {
2603 free_page(page);
2604 return -ENOMEM;
2605 }
2606 proc_skip_char(&kbuf, &left, '\n');
2607 while (!err && left) {
2608 unsigned long val_a, val_b;
2609 bool neg;
2610
2611 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2612 sizeof(tr_a), &c);
2613 if (err)
2614 break;
2615 if (val_a >= bitmap_len || neg) {
2616 err = -EINVAL;
2617 break;
2618 }
2619
2620 val_b = val_a;
2621 if (left) {
2622 kbuf++;
2623 left--;
2624 }
2625
2626 if (c == '-') {
2627 err = proc_get_long(&kbuf, &left, &val_b,
2628 &neg, tr_b, sizeof(tr_b),
2629 &c);
2630 if (err)
2631 break;
2632 if (val_b >= bitmap_len || neg ||
2633 val_a > val_b) {
2634 err = -EINVAL;
2635 break;
2636 }
2637 if (left) {
2638 kbuf++;
2639 left--;
2640 }
2641 }
2642
2643 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2644 first = 0;
2645 proc_skip_char(&kbuf, &left, '\n');
2646 }
2647 free_page(page);
2648 } else {
2649 unsigned long bit_a, bit_b = 0;
2650
2651 while (left) {
2652 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2653 if (bit_a >= bitmap_len)
2654 break;
2655 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2656 bit_a + 1) - 1;
2657
2658 if (!first) {
2659 err = proc_put_char(&buffer, &left, ',');
2660 if (err)
2661 break;
2662 }
2663 err = proc_put_long(&buffer, &left, bit_a, false);
2664 if (err)
2665 break;
2666 if (bit_a != bit_b) {
2667 err = proc_put_char(&buffer, &left, '-');
2668 if (err)
2669 break;
2670 err = proc_put_long(&buffer, &left, bit_b, false);
2671 if (err)
2672 break;
2673 }
2674
2675 first = 0; bit_b++;
2676 }
2677 if (!err)
2678 err = proc_put_char(&buffer, &left, '\n');
2679 }
2680
2681 if (!err) {
2682 if (write) {
2683 if (*ppos)
2684 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2685 else
2686 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2687 }
2688 kfree(tmp_bitmap);
2689 *lenp -= left;
2690 *ppos += *lenp;
2691 return 0;
2692 } else {
2693 kfree(tmp_bitmap);
2694 return err;
2695 }
2696 }
2697
2698 #else /* CONFIG_PROC_SYSCTL */
2699
2700 int proc_dostring(struct ctl_table *table, int write,
2701 void __user *buffer, size_t *lenp, loff_t *ppos)
2702 {
2703 return -ENOSYS;
2704 }
2705
2706 int proc_dointvec(struct ctl_table *table, int write,
2707 void __user *buffer, size_t *lenp, loff_t *ppos)
2708 {
2709 return -ENOSYS;
2710 }
2711
2712 int proc_dointvec_minmax(struct ctl_table *table, int write,
2713 void __user *buffer, size_t *lenp, loff_t *ppos)
2714 {
2715 return -ENOSYS;
2716 }
2717
2718 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2719 void __user *buffer, size_t *lenp, loff_t *ppos)
2720 {
2721 return -ENOSYS;
2722 }
2723
2724 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2725 void __user *buffer, size_t *lenp, loff_t *ppos)
2726 {
2727 return -ENOSYS;
2728 }
2729
2730 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2731 void __user *buffer, size_t *lenp, loff_t *ppos)
2732 {
2733 return -ENOSYS;
2734 }
2735
2736 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2737 void __user *buffer, size_t *lenp, loff_t *ppos)
2738 {
2739 return -ENOSYS;
2740 }
2741
2742 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2743 void __user *buffer,
2744 size_t *lenp, loff_t *ppos)
2745 {
2746 return -ENOSYS;
2747 }
2748
2749
2750 #endif /* CONFIG_PROC_SYSCTL */
2751
2752 /*
2753 * No sense putting this after each symbol definition, twice,
2754 * exception granted :-)
2755 */
2756 EXPORT_SYMBOL(proc_dointvec);
2757 EXPORT_SYMBOL(proc_dointvec_jiffies);
2758 EXPORT_SYMBOL(proc_dointvec_minmax);
2759 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2760 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2761 EXPORT_SYMBOL(proc_dostring);
2762 EXPORT_SYMBOL(proc_doulongvec_minmax);
2763 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
Linux kernel - Deliverables
This page took 0.184028 seconds and 6 git commands to generate.