projects / deliverable / linux.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Merge branch 'fixes' into for-linus
[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
65 #include <asm/uaccess.h>
66 #include <asm/processor.h>
67
68 #ifdef CONFIG_X86
69 #include <asm/nmi.h>
70 #include <asm/stacktrace.h>
71 #include <asm/io.h>
72 #endif
73 #ifdef CONFIG_SPARC
74 #include <asm/setup.h>
75 #endif
76 #ifdef CONFIG_BSD_PROCESS_ACCT
77 #include <linux/acct.h>
78 #endif
79 #ifdef CONFIG_RT_MUTEXES
80 #include <linux/rtmutex.h>
81 #endif
82 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
83 #include <linux/lockdep.h>
84 #endif
85 #ifdef CONFIG_CHR_DEV_SG
86 #include <scsi/sg.h>
87 #endif
88
89 #ifdef CONFIG_LOCKUP_DETECTOR
90 #include <linux/nmi.h>
91 #endif
92
93
94 #if defined(CONFIG_SYSCTL)
95
96 /* External variables not in a header file. */
97 extern int sysctl_overcommit_memory;
98 extern int sysctl_overcommit_ratio;
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 min_free_kbytes;
108 extern int pid_max_min, pid_max_max;
109 extern int sysctl_drop_caches;
110 extern int percpu_pagelist_fraction;
111 extern int compat_log;
112 extern int latencytop_enabled;
113 extern int sysctl_nr_open_min, sysctl_nr_open_max;
114 #ifndef CONFIG_MMU
115 extern int sysctl_nr_trim_pages;
116 #endif
117 #ifdef CONFIG_BLOCK
118 extern int blk_iopoll_enabled;
119 #endif
120
121 /* Constants used for minimum and maximum */
122 #ifdef CONFIG_LOCKUP_DETECTOR
123 static int sixty = 60;
124 static int neg_one = -1;
125 #endif
126
127 static int zero;
128 static int __maybe_unused one = 1;
129 static int __maybe_unused two = 2;
130 static int __maybe_unused three = 3;
131 static unsigned long one_ul = 1;
132 static int one_hundred = 100;
133 #ifdef CONFIG_PRINTK
134 static int ten_thousand = 10000;
135 #endif
136
137 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
138 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
139
140 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
141 static int maxolduid = 65535;
142 static int minolduid;
143 static int min_percpu_pagelist_fract = 8;
144
145 static int ngroups_max = NGROUPS_MAX;
146 static const int cap_last_cap = CAP_LAST_CAP;
147
148 #ifdef CONFIG_INOTIFY_USER
149 #include <linux/inotify.h>
150 #endif
151 #ifdef CONFIG_SPARC
152 #endif
153
154 #ifdef CONFIG_SPARC64
155 extern int sysctl_tsb_ratio;
156 #endif
157
158 #ifdef __hppa__
159 extern int pwrsw_enabled;
160 extern int unaligned_enabled;
161 #endif
162
163 #ifdef CONFIG_IA64
164 extern int no_unaligned_warning;
165 extern int unaligned_dump_stack;
166 #endif
167
168 #ifdef CONFIG_PROC_SYSCTL
169 static int proc_do_cad_pid(struct ctl_table *table, int write,
170 void __user *buffer, size_t *lenp, loff_t *ppos);
171 static int proc_taint(struct ctl_table *table, int write,
172 void __user *buffer, size_t *lenp, loff_t *ppos);
173 #endif
174
175 #ifdef CONFIG_PRINTK
176 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
177 void __user *buffer, size_t *lenp, loff_t *ppos);
178 #endif
179
180 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
181 void __user *buffer, size_t *lenp, loff_t *ppos);
182 #ifdef CONFIG_COREDUMP
183 static int proc_dostring_coredump(struct ctl_table *table, int write,
184 void __user *buffer, size_t *lenp, loff_t *ppos);
185 #endif
186
187 #ifdef CONFIG_MAGIC_SYSRQ
188 /* Note: sysrq code uses it's own private copy */
189 static int __sysrq_enabled = SYSRQ_DEFAULT_ENABLE;
190
191 static int sysrq_sysctl_handler(ctl_table *table, int write,
192 void __user *buffer, size_t *lenp,
193 loff_t *ppos)
194 {
195 int error;
196
197 error = proc_dointvec(table, write, buffer, lenp, ppos);
198 if (error)
199 return error;
200
201 if (write)
202 sysrq_toggle_support(__sysrq_enabled);
203
204 return 0;
205 }
206
207 #endif
208
209 static struct ctl_table kern_table[];
210 static struct ctl_table vm_table[];
211 static struct ctl_table fs_table[];
212 static struct ctl_table debug_table[];
213 static struct ctl_table dev_table[];
214 extern struct ctl_table random_table[];
215 #ifdef CONFIG_EPOLL
216 extern struct ctl_table epoll_table[];
217 #endif
218
219 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
220 int sysctl_legacy_va_layout;
221 #endif
222
223 /* The default sysctl tables: */
224
225 static struct ctl_table sysctl_base_table[] = {
226 {
227 .procname = "kernel",
228 .mode = 0555,
229 .child = kern_table,
230 },
231 {
232 .procname = "vm",
233 .mode = 0555,
234 .child = vm_table,
235 },
236 {
237 .procname = "fs",
238 .mode = 0555,
239 .child = fs_table,
240 },
241 {
242 .procname = "debug",
243 .mode = 0555,
244 .child = debug_table,
245 },
246 {
247 .procname = "dev",
248 .mode = 0555,
249 .child = dev_table,
250 },
251 { }
252 };
253
254 #ifdef CONFIG_SCHED_DEBUG
255 static int min_sched_granularity_ns = 100000; /* 100 usecs */
256 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
257 static int min_wakeup_granularity_ns; /* 0 usecs */
258 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
259 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
260 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
261 #endif
262
263 #ifdef CONFIG_COMPACTION
264 static int min_extfrag_threshold;
265 static int max_extfrag_threshold = 1000;
266 #endif
267
268 static struct ctl_table kern_table[] = {
269 {
270 .procname = "sched_child_runs_first",
271 .data = &sysctl_sched_child_runs_first,
272 .maxlen = sizeof(unsigned int),
273 .mode = 0644,
274 .proc_handler = proc_dointvec,
275 },
276 #ifdef CONFIG_SCHED_DEBUG
277 {
278 .procname = "sched_min_granularity_ns",
279 .data = &sysctl_sched_min_granularity,
280 .maxlen = sizeof(unsigned int),
281 .mode = 0644,
282 .proc_handler = sched_proc_update_handler,
283 .extra1 = &min_sched_granularity_ns,
284 .extra2 = &max_sched_granularity_ns,
285 },
286 {
287 .procname = "sched_latency_ns",
288 .data = &sysctl_sched_latency,
289 .maxlen = sizeof(unsigned int),
290 .mode = 0644,
291 .proc_handler = sched_proc_update_handler,
292 .extra1 = &min_sched_granularity_ns,
293 .extra2 = &max_sched_granularity_ns,
294 },
295 {
296 .procname = "sched_wakeup_granularity_ns",
297 .data = &sysctl_sched_wakeup_granularity,
298 .maxlen = sizeof(unsigned int),
299 .mode = 0644,
300 .proc_handler = sched_proc_update_handler,
301 .extra1 = &min_wakeup_granularity_ns,
302 .extra2 = &max_wakeup_granularity_ns,
303 },
304 {
305 .procname = "sched_tunable_scaling",
306 .data = &sysctl_sched_tunable_scaling,
307 .maxlen = sizeof(enum sched_tunable_scaling),
308 .mode = 0644,
309 .proc_handler = sched_proc_update_handler,
310 .extra1 = &min_sched_tunable_scaling,
311 .extra2 = &max_sched_tunable_scaling,
312 },
313 {
314 .procname = "sched_migration_cost_ns",
315 .data = &sysctl_sched_migration_cost,
316 .maxlen = sizeof(unsigned int),
317 .mode = 0644,
318 .proc_handler = proc_dointvec,
319 },
320 {
321 .procname = "sched_nr_migrate",
322 .data = &sysctl_sched_nr_migrate,
323 .maxlen = sizeof(unsigned int),
324 .mode = 0644,
325 .proc_handler = proc_dointvec,
326 },
327 {
328 .procname = "sched_time_avg_ms",
329 .data = &sysctl_sched_time_avg,
330 .maxlen = sizeof(unsigned int),
331 .mode = 0644,
332 .proc_handler = proc_dointvec,
333 },
334 {
335 .procname = "sched_shares_window_ns",
336 .data = &sysctl_sched_shares_window,
337 .maxlen = sizeof(unsigned int),
338 .mode = 0644,
339 .proc_handler = proc_dointvec,
340 },
341 {
342 .procname = "timer_migration",
343 .data = &sysctl_timer_migration,
344 .maxlen = sizeof(unsigned int),
345 .mode = 0644,
346 .proc_handler = proc_dointvec_minmax,
347 .extra1 = &zero,
348 .extra2 = &one,
349 },
350 #endif
351 {
352 .procname = "sched_rt_period_us",
353 .data = &sysctl_sched_rt_period,
354 .maxlen = sizeof(unsigned int),
355 .mode = 0644,
356 .proc_handler = sched_rt_handler,
357 },
358 {
359 .procname = "sched_rt_runtime_us",
360 .data = &sysctl_sched_rt_runtime,
361 .maxlen = sizeof(int),
362 .mode = 0644,
363 .proc_handler = sched_rt_handler,
364 },
365 #ifdef CONFIG_SCHED_AUTOGROUP
366 {
367 .procname = "sched_autogroup_enabled",
368 .data = &sysctl_sched_autogroup_enabled,
369 .maxlen = sizeof(unsigned int),
370 .mode = 0644,
371 .proc_handler = proc_dointvec_minmax,
372 .extra1 = &zero,
373 .extra2 = &one,
374 },
375 #endif
376 #ifdef CONFIG_CFS_BANDWIDTH
377 {
378 .procname = "sched_cfs_bandwidth_slice_us",
379 .data = &sysctl_sched_cfs_bandwidth_slice,
380 .maxlen = sizeof(unsigned int),
381 .mode = 0644,
382 .proc_handler = proc_dointvec_minmax,
383 .extra1 = &one,
384 },
385 #endif
386 #ifdef CONFIG_PROVE_LOCKING
387 {
388 .procname = "prove_locking",
389 .data = &prove_locking,
390 .maxlen = sizeof(int),
391 .mode = 0644,
392 .proc_handler = proc_dointvec,
393 },
394 #endif
395 #ifdef CONFIG_LOCK_STAT
396 {
397 .procname = "lock_stat",
398 .data = &lock_stat,
399 .maxlen = sizeof(int),
400 .mode = 0644,
401 .proc_handler = proc_dointvec,
402 },
403 #endif
404 {
405 .procname = "panic",
406 .data = &panic_timeout,
407 .maxlen = sizeof(int),
408 .mode = 0644,
409 .proc_handler = proc_dointvec,
410 },
411 #ifdef CONFIG_COREDUMP
412 {
413 .procname = "core_uses_pid",
414 .data = &core_uses_pid,
415 .maxlen = sizeof(int),
416 .mode = 0644,
417 .proc_handler = proc_dointvec,
418 },
419 {
420 .procname = "core_pattern",
421 .data = core_pattern,
422 .maxlen = CORENAME_MAX_SIZE,
423 .mode = 0644,
424 .proc_handler = proc_dostring_coredump,
425 },
426 {
427 .procname = "core_pipe_limit",
428 .data = &core_pipe_limit,
429 .maxlen = sizeof(unsigned int),
430 .mode = 0644,
431 .proc_handler = proc_dointvec,
432 },
433 #endif
434 #ifdef CONFIG_PROC_SYSCTL
435 {
436 .procname = "tainted",
437 .maxlen = sizeof(long),
438 .mode = 0644,
439 .proc_handler = proc_taint,
440 },
441 #endif
442 #ifdef CONFIG_LATENCYTOP
443 {
444 .procname = "latencytop",
445 .data = &latencytop_enabled,
446 .maxlen = sizeof(int),
447 .mode = 0644,
448 .proc_handler = proc_dointvec,
449 },
450 #endif
451 #ifdef CONFIG_BLK_DEV_INITRD
452 {
453 .procname = "real-root-dev",
454 .data = &real_root_dev,
455 .maxlen = sizeof(int),
456 .mode = 0644,
457 .proc_handler = proc_dointvec,
458 },
459 #endif
460 {
461 .procname = "print-fatal-signals",
462 .data = &print_fatal_signals,
463 .maxlen = sizeof(int),
464 .mode = 0644,
465 .proc_handler = proc_dointvec,
466 },
467 #ifdef CONFIG_SPARC
468 {
469 .procname = "reboot-cmd",
470 .data = reboot_command,
471 .maxlen = 256,
472 .mode = 0644,
473 .proc_handler = proc_dostring,
474 },
475 {
476 .procname = "stop-a",
477 .data = &stop_a_enabled,
478 .maxlen = sizeof (int),
479 .mode = 0644,
480 .proc_handler = proc_dointvec,
481 },
482 {
483 .procname = "scons-poweroff",
484 .data = &scons_pwroff,
485 .maxlen = sizeof (int),
486 .mode = 0644,
487 .proc_handler = proc_dointvec,
488 },
489 #endif
490 #ifdef CONFIG_SPARC64
491 {
492 .procname = "tsb-ratio",
493 .data = &sysctl_tsb_ratio,
494 .maxlen = sizeof (int),
495 .mode = 0644,
496 .proc_handler = proc_dointvec,
497 },
498 #endif
499 #ifdef __hppa__
500 {
501 .procname = "soft-power",
502 .data = &pwrsw_enabled,
503 .maxlen = sizeof (int),
504 .mode = 0644,
505 .proc_handler = proc_dointvec,
506 },
507 {
508 .procname = "unaligned-trap",
509 .data = &unaligned_enabled,
510 .maxlen = sizeof (int),
511 .mode = 0644,
512 .proc_handler = proc_dointvec,
513 },
514 #endif
515 {
516 .procname = "ctrl-alt-del",
517 .data = &C_A_D,
518 .maxlen = sizeof(int),
519 .mode = 0644,
520 .proc_handler = proc_dointvec,
521 },
522 #ifdef CONFIG_FUNCTION_TRACER
523 {
524 .procname = "ftrace_enabled",
525 .data = &ftrace_enabled,
526 .maxlen = sizeof(int),
527 .mode = 0644,
528 .proc_handler = ftrace_enable_sysctl,
529 },
530 #endif
531 #ifdef CONFIG_STACK_TRACER
532 {
533 .procname = "stack_tracer_enabled",
534 .data = &stack_tracer_enabled,
535 .maxlen = sizeof(int),
536 .mode = 0644,
537 .proc_handler = stack_trace_sysctl,
538 },
539 #endif
540 #ifdef CONFIG_TRACING
541 {
542 .procname = "ftrace_dump_on_oops",
543 .data = &ftrace_dump_on_oops,
544 .maxlen = sizeof(int),
545 .mode = 0644,
546 .proc_handler = proc_dointvec,
547 },
548 #endif
549 #ifdef CONFIG_MODULES
550 {
551 .procname = "modprobe",
552 .data = &modprobe_path,
553 .maxlen = KMOD_PATH_LEN,
554 .mode = 0644,
555 .proc_handler = proc_dostring,
556 },
557 {
558 .procname = "modules_disabled",
559 .data = &modules_disabled,
560 .maxlen = sizeof(int),
561 .mode = 0644,
562 /* only handle a transition from default "0" to "1" */
563 .proc_handler = proc_dointvec_minmax,
564 .extra1 = &one,
565 .extra2 = &one,
566 },
567 #endif
568 #ifdef CONFIG_HOTPLUG
569 {
570 .procname = "hotplug",
571 .data = &uevent_helper,
572 .maxlen = UEVENT_HELPER_PATH_LEN,
573 .mode = 0644,
574 .proc_handler = proc_dostring,
575 },
576 #endif
577 #ifdef CONFIG_CHR_DEV_SG
578 {
579 .procname = "sg-big-buff",
580 .data = &sg_big_buff,
581 .maxlen = sizeof (int),
582 .mode = 0444,
583 .proc_handler = proc_dointvec,
584 },
585 #endif
586 #ifdef CONFIG_BSD_PROCESS_ACCT
587 {
588 .procname = "acct",
589 .data = &acct_parm,
590 .maxlen = 3*sizeof(int),
591 .mode = 0644,
592 .proc_handler = proc_dointvec,
593 },
594 #endif
595 #ifdef CONFIG_MAGIC_SYSRQ
596 {
597 .procname = "sysrq",
598 .data = &__sysrq_enabled,
599 .maxlen = sizeof (int),
600 .mode = 0644,
601 .proc_handler = sysrq_sysctl_handler,
602 },
603 #endif
604 #ifdef CONFIG_PROC_SYSCTL
605 {
606 .procname = "cad_pid",
607 .data = NULL,
608 .maxlen = sizeof (int),
609 .mode = 0600,
610 .proc_handler = proc_do_cad_pid,
611 },
612 #endif
613 {
614 .procname = "threads-max",
615 .data = &max_threads,
616 .maxlen = sizeof(int),
617 .mode = 0644,
618 .proc_handler = proc_dointvec,
619 },
620 {
621 .procname = "random",
622 .mode = 0555,
623 .child = random_table,
624 },
625 {
626 .procname = "usermodehelper",
627 .mode = 0555,
628 .child = usermodehelper_table,
629 },
630 {
631 .procname = "overflowuid",
632 .data = &overflowuid,
633 .maxlen = sizeof(int),
634 .mode = 0644,
635 .proc_handler = proc_dointvec_minmax,
636 .extra1 = &minolduid,
637 .extra2 = &maxolduid,
638 },
639 {
640 .procname = "overflowgid",
641 .data = &overflowgid,
642 .maxlen = sizeof(int),
643 .mode = 0644,
644 .proc_handler = proc_dointvec_minmax,
645 .extra1 = &minolduid,
646 .extra2 = &maxolduid,
647 },
648 #ifdef CONFIG_S390
649 #ifdef CONFIG_MATHEMU
650 {
651 .procname = "ieee_emulation_warnings",
652 .data = &sysctl_ieee_emulation_warnings,
653 .maxlen = sizeof(int),
654 .mode = 0644,
655 .proc_handler = proc_dointvec,
656 },
657 #endif
658 {
659 .procname = "userprocess_debug",
660 .data = &show_unhandled_signals,
661 .maxlen = sizeof(int),
662 .mode = 0644,
663 .proc_handler = proc_dointvec,
664 },
665 #endif
666 {
667 .procname = "pid_max",
668 .data = &pid_max,
669 .maxlen = sizeof (int),
670 .mode = 0644,
671 .proc_handler = proc_dointvec_minmax,
672 .extra1 = &pid_max_min,
673 .extra2 = &pid_max_max,
674 },
675 {
676 .procname = "panic_on_oops",
677 .data = &panic_on_oops,
678 .maxlen = sizeof(int),
679 .mode = 0644,
680 .proc_handler = proc_dointvec,
681 },
682 #if defined CONFIG_PRINTK
683 {
684 .procname = "printk",
685 .data = &console_loglevel,
686 .maxlen = 4*sizeof(int),
687 .mode = 0644,
688 .proc_handler = proc_dointvec,
689 },
690 {
691 .procname = "printk_ratelimit",
692 .data = &printk_ratelimit_state.interval,
693 .maxlen = sizeof(int),
694 .mode = 0644,
695 .proc_handler = proc_dointvec_jiffies,
696 },
697 {
698 .procname = "printk_ratelimit_burst",
699 .data = &printk_ratelimit_state.burst,
700 .maxlen = sizeof(int),
701 .mode = 0644,
702 .proc_handler = proc_dointvec,
703 },
704 {
705 .procname = "printk_delay",
706 .data = &printk_delay_msec,
707 .maxlen = sizeof(int),
708 .mode = 0644,
709 .proc_handler = proc_dointvec_minmax,
710 .extra1 = &zero,
711 .extra2 = &ten_thousand,
712 },
713 {
714 .procname = "dmesg_restrict",
715 .data = &dmesg_restrict,
716 .maxlen = sizeof(int),
717 .mode = 0644,
718 .proc_handler = proc_dointvec_minmax_sysadmin,
719 .extra1 = &zero,
720 .extra2 = &one,
721 },
722 {
723 .procname = "kptr_restrict",
724 .data = &kptr_restrict,
725 .maxlen = sizeof(int),
726 .mode = 0644,
727 .proc_handler = proc_dointvec_minmax_sysadmin,
728 .extra1 = &zero,
729 .extra2 = &two,
730 },
731 #endif
732 {
733 .procname = "ngroups_max",
734 .data = &ngroups_max,
735 .maxlen = sizeof (int),
736 .mode = 0444,
737 .proc_handler = proc_dointvec,
738 },
739 {
740 .procname = "cap_last_cap",
741 .data = (void *)&cap_last_cap,
742 .maxlen = sizeof(int),
743 .mode = 0444,
744 .proc_handler = proc_dointvec,
745 },
746 #if defined(CONFIG_LOCKUP_DETECTOR)
747 {
748 .procname = "watchdog",
749 .data = &watchdog_enabled,
750 .maxlen = sizeof (int),
751 .mode = 0644,
752 .proc_handler = proc_dowatchdog,
753 .extra1 = &zero,
754 .extra2 = &one,
755 },
756 {
757 .procname = "watchdog_thresh",
758 .data = &watchdog_thresh,
759 .maxlen = sizeof(int),
760 .mode = 0644,
761 .proc_handler = proc_dowatchdog,
762 .extra1 = &neg_one,
763 .extra2 = &sixty,
764 },
765 {
766 .procname = "softlockup_panic",
767 .data = &softlockup_panic,
768 .maxlen = sizeof(int),
769 .mode = 0644,
770 .proc_handler = proc_dointvec_minmax,
771 .extra1 = &zero,
772 .extra2 = &one,
773 },
774 {
775 .procname = "nmi_watchdog",
776 .data = &watchdog_enabled,
777 .maxlen = sizeof (int),
778 .mode = 0644,
779 .proc_handler = proc_dowatchdog,
780 .extra1 = &zero,
781 .extra2 = &one,
782 },
783 #endif
784 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
785 {
786 .procname = "unknown_nmi_panic",
787 .data = &unknown_nmi_panic,
788 .maxlen = sizeof (int),
789 .mode = 0644,
790 .proc_handler = proc_dointvec,
791 },
792 #endif
793 #if defined(CONFIG_X86)
794 {
795 .procname = "panic_on_unrecovered_nmi",
796 .data = &panic_on_unrecovered_nmi,
797 .maxlen = sizeof(int),
798 .mode = 0644,
799 .proc_handler = proc_dointvec,
800 },
801 {
802 .procname = "panic_on_io_nmi",
803 .data = &panic_on_io_nmi,
804 .maxlen = sizeof(int),
805 .mode = 0644,
806 .proc_handler = proc_dointvec,
807 },
808 #ifdef CONFIG_DEBUG_STACKOVERFLOW
809 {
810 .procname = "panic_on_stackoverflow",
811 .data = &sysctl_panic_on_stackoverflow,
812 .maxlen = sizeof(int),
813 .mode = 0644,
814 .proc_handler = proc_dointvec,
815 },
816 #endif
817 {
818 .procname = "bootloader_type",
819 .data = &bootloader_type,
820 .maxlen = sizeof (int),
821 .mode = 0444,
822 .proc_handler = proc_dointvec,
823 },
824 {
825 .procname = "bootloader_version",
826 .data = &bootloader_version,
827 .maxlen = sizeof (int),
828 .mode = 0444,
829 .proc_handler = proc_dointvec,
830 },
831 {
832 .procname = "kstack_depth_to_print",
833 .data = &kstack_depth_to_print,
834 .maxlen = sizeof(int),
835 .mode = 0644,
836 .proc_handler = proc_dointvec,
837 },
838 {
839 .procname = "io_delay_type",
840 .data = &io_delay_type,
841 .maxlen = sizeof(int),
842 .mode = 0644,
843 .proc_handler = proc_dointvec,
844 },
845 #endif
846 #if defined(CONFIG_MMU)
847 {
848 .procname = "randomize_va_space",
849 .data = &randomize_va_space,
850 .maxlen = sizeof(int),
851 .mode = 0644,
852 .proc_handler = proc_dointvec,
853 },
854 #endif
855 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
856 {
857 .procname = "spin_retry",
858 .data = &spin_retry,
859 .maxlen = sizeof (int),
860 .mode = 0644,
861 .proc_handler = proc_dointvec,
862 },
863 #endif
864 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
865 {
866 .procname = "acpi_video_flags",
867 .data = &acpi_realmode_flags,
868 .maxlen = sizeof (unsigned long),
869 .mode = 0644,
870 .proc_handler = proc_doulongvec_minmax,
871 },
872 #endif
873 #ifdef CONFIG_IA64
874 {
875 .procname = "ignore-unaligned-usertrap",
876 .data = &no_unaligned_warning,
877 .maxlen = sizeof (int),
878 .mode = 0644,
879 .proc_handler = proc_dointvec,
880 },
881 {
882 .procname = "unaligned-dump-stack",
883 .data = &unaligned_dump_stack,
884 .maxlen = sizeof (int),
885 .mode = 0644,
886 .proc_handler = proc_dointvec,
887 },
888 #endif
889 #ifdef CONFIG_DETECT_HUNG_TASK
890 {
891 .procname = "hung_task_panic",
892 .data = &sysctl_hung_task_panic,
893 .maxlen = sizeof(int),
894 .mode = 0644,
895 .proc_handler = proc_dointvec_minmax,
896 .extra1 = &zero,
897 .extra2 = &one,
898 },
899 {
900 .procname = "hung_task_check_count",
901 .data = &sysctl_hung_task_check_count,
902 .maxlen = sizeof(unsigned long),
903 .mode = 0644,
904 .proc_handler = proc_doulongvec_minmax,
905 },
906 {
907 .procname = "hung_task_timeout_secs",
908 .data = &sysctl_hung_task_timeout_secs,
909 .maxlen = sizeof(unsigned long),
910 .mode = 0644,
911 .proc_handler = proc_dohung_task_timeout_secs,
912 },
913 {
914 .procname = "hung_task_warnings",
915 .data = &sysctl_hung_task_warnings,
916 .maxlen = sizeof(unsigned long),
917 .mode = 0644,
918 .proc_handler = proc_doulongvec_minmax,
919 },
920 #endif
921 #ifdef CONFIG_COMPAT
922 {
923 .procname = "compat-log",
924 .data = &compat_log,
925 .maxlen = sizeof (int),
926 .mode = 0644,
927 .proc_handler = proc_dointvec,
928 },
929 #endif
930 #ifdef CONFIG_RT_MUTEXES
931 {
932 .procname = "max_lock_depth",
933 .data = &max_lock_depth,
934 .maxlen = sizeof(int),
935 .mode = 0644,
936 .proc_handler = proc_dointvec,
937 },
938 #endif
939 {
940 .procname = "poweroff_cmd",
941 .data = &poweroff_cmd,
942 .maxlen = POWEROFF_CMD_PATH_LEN,
943 .mode = 0644,
944 .proc_handler = proc_dostring,
945 },
946 #ifdef CONFIG_KEYS
947 {
948 .procname = "keys",
949 .mode = 0555,
950 .child = key_sysctls,
951 },
952 #endif
953 #ifdef CONFIG_RCU_TORTURE_TEST
954 {
955 .procname = "rcutorture_runnable",
956 .data = &rcutorture_runnable,
957 .maxlen = sizeof(int),
958 .mode = 0644,
959 .proc_handler = proc_dointvec,
960 },
961 #endif
962 #ifdef CONFIG_PERF_EVENTS
963 /*
964 * User-space scripts rely on the existence of this file
965 * as a feature check for perf_events being enabled.
966 *
967 * So it's an ABI, do not remove!
968 */
969 {
970 .procname = "perf_event_paranoid",
971 .data = &sysctl_perf_event_paranoid,
972 .maxlen = sizeof(sysctl_perf_event_paranoid),
973 .mode = 0644,
974 .proc_handler = proc_dointvec,
975 },
976 {
977 .procname = "perf_event_mlock_kb",
978 .data = &sysctl_perf_event_mlock,
979 .maxlen = sizeof(sysctl_perf_event_mlock),
980 .mode = 0644,
981 .proc_handler = proc_dointvec,
982 },
983 {
984 .procname = "perf_event_max_sample_rate",
985 .data = &sysctl_perf_event_sample_rate,
986 .maxlen = sizeof(sysctl_perf_event_sample_rate),
987 .mode = 0644,
988 .proc_handler = perf_proc_update_handler,
989 },
990 #endif
991 #ifdef CONFIG_KMEMCHECK
992 {
993 .procname = "kmemcheck",
994 .data = &kmemcheck_enabled,
995 .maxlen = sizeof(int),
996 .mode = 0644,
997 .proc_handler = proc_dointvec,
998 },
999 #endif
1000 #ifdef CONFIG_BLOCK
1001 {
1002 .procname = "blk_iopoll",
1003 .data = &blk_iopoll_enabled,
1004 .maxlen = sizeof(int),
1005 .mode = 0644,
1006 .proc_handler = proc_dointvec,
1007 },
1008 #endif
1009 { }
1010 };
1011
1012 static struct ctl_table vm_table[] = {
1013 {
1014 .procname = "overcommit_memory",
1015 .data = &sysctl_overcommit_memory,
1016 .maxlen = sizeof(sysctl_overcommit_memory),
1017 .mode = 0644,
1018 .proc_handler = proc_dointvec_minmax,
1019 .extra1 = &zero,
1020 .extra2 = &two,
1021 },
1022 {
1023 .procname = "panic_on_oom",
1024 .data = &sysctl_panic_on_oom,
1025 .maxlen = sizeof(sysctl_panic_on_oom),
1026 .mode = 0644,
1027 .proc_handler = proc_dointvec_minmax,
1028 .extra1 = &zero,
1029 .extra2 = &two,
1030 },
1031 {
1032 .procname = "oom_kill_allocating_task",
1033 .data = &sysctl_oom_kill_allocating_task,
1034 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1035 .mode = 0644,
1036 .proc_handler = proc_dointvec,
1037 },
1038 {
1039 .procname = "oom_dump_tasks",
1040 .data = &sysctl_oom_dump_tasks,
1041 .maxlen = sizeof(sysctl_oom_dump_tasks),
1042 .mode = 0644,
1043 .proc_handler = proc_dointvec,
1044 },
1045 {
1046 .procname = "overcommit_ratio",
1047 .data = &sysctl_overcommit_ratio,
1048 .maxlen = sizeof(sysctl_overcommit_ratio),
1049 .mode = 0644,
1050 .proc_handler = proc_dointvec,
1051 },
1052 {
1053 .procname = "page-cluster",
1054 .data = &page_cluster,
1055 .maxlen = sizeof(int),
1056 .mode = 0644,
1057 .proc_handler = proc_dointvec_minmax,
1058 .extra1 = &zero,
1059 },
1060 {
1061 .procname = "dirty_background_ratio",
1062 .data = &dirty_background_ratio,
1063 .maxlen = sizeof(dirty_background_ratio),
1064 .mode = 0644,
1065 .proc_handler = dirty_background_ratio_handler,
1066 .extra1 = &zero,
1067 .extra2 = &one_hundred,
1068 },
1069 {
1070 .procname = "dirty_background_bytes",
1071 .data = &dirty_background_bytes,
1072 .maxlen = sizeof(dirty_background_bytes),
1073 .mode = 0644,
1074 .proc_handler = dirty_background_bytes_handler,
1075 .extra1 = &one_ul,
1076 },
1077 {
1078 .procname = "dirty_ratio",
1079 .data = &vm_dirty_ratio,
1080 .maxlen = sizeof(vm_dirty_ratio),
1081 .mode = 0644,
1082 .proc_handler = dirty_ratio_handler,
1083 .extra1 = &zero,
1084 .extra2 = &one_hundred,
1085 },
1086 {
1087 .procname = "dirty_bytes",
1088 .data = &vm_dirty_bytes,
1089 .maxlen = sizeof(vm_dirty_bytes),
1090 .mode = 0644,
1091 .proc_handler = dirty_bytes_handler,
1092 .extra1 = &dirty_bytes_min,
1093 },
1094 {
1095 .procname = "dirty_writeback_centisecs",
1096 .data = &dirty_writeback_interval,
1097 .maxlen = sizeof(dirty_writeback_interval),
1098 .mode = 0644,
1099 .proc_handler = dirty_writeback_centisecs_handler,
1100 },
1101 {
1102 .procname = "dirty_expire_centisecs",
1103 .data = &dirty_expire_interval,
1104 .maxlen = sizeof(dirty_expire_interval),
1105 .mode = 0644,
1106 .proc_handler = proc_dointvec_minmax,
1107 .extra1 = &zero,
1108 },
1109 {
1110 .procname = "nr_pdflush_threads",
1111 .mode = 0444 /* read-only */,
1112 .proc_handler = pdflush_proc_obsolete,
1113 },
1114 {
1115 .procname = "swappiness",
1116 .data = &vm_swappiness,
1117 .maxlen = sizeof(vm_swappiness),
1118 .mode = 0644,
1119 .proc_handler = proc_dointvec_minmax,
1120 .extra1 = &zero,
1121 .extra2 = &one_hundred,
1122 },
1123 #ifdef CONFIG_HUGETLB_PAGE
1124 {
1125 .procname = "nr_hugepages",
1126 .data = NULL,
1127 .maxlen = sizeof(unsigned long),
1128 .mode = 0644,
1129 .proc_handler = hugetlb_sysctl_handler,
1130 .extra1 = (void *)&hugetlb_zero,
1131 .extra2 = (void *)&hugetlb_infinity,
1132 },
1133 #ifdef CONFIG_NUMA
1134 {
1135 .procname = "nr_hugepages_mempolicy",
1136 .data = NULL,
1137 .maxlen = sizeof(unsigned long),
1138 .mode = 0644,
1139 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1140 .extra1 = (void *)&hugetlb_zero,
1141 .extra2 = (void *)&hugetlb_infinity,
1142 },
1143 #endif
1144 {
1145 .procname = "hugetlb_shm_group",
1146 .data = &sysctl_hugetlb_shm_group,
1147 .maxlen = sizeof(gid_t),
1148 .mode = 0644,
1149 .proc_handler = proc_dointvec,
1150 },
1151 {
1152 .procname = "hugepages_treat_as_movable",
1153 .data = &hugepages_treat_as_movable,
1154 .maxlen = sizeof(int),
1155 .mode = 0644,
1156 .proc_handler = hugetlb_treat_movable_handler,
1157 },
1158 {
1159 .procname = "nr_overcommit_hugepages",
1160 .data = NULL,
1161 .maxlen = sizeof(unsigned long),
1162 .mode = 0644,
1163 .proc_handler = hugetlb_overcommit_handler,
1164 .extra1 = (void *)&hugetlb_zero,
1165 .extra2 = (void *)&hugetlb_infinity,
1166 },
1167 #endif
1168 {
1169 .procname = "lowmem_reserve_ratio",
1170 .data = &sysctl_lowmem_reserve_ratio,
1171 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1172 .mode = 0644,
1173 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1174 },
1175 {
1176 .procname = "drop_caches",
1177 .data = &sysctl_drop_caches,
1178 .maxlen = sizeof(int),
1179 .mode = 0644,
1180 .proc_handler = drop_caches_sysctl_handler,
1181 .extra1 = &one,
1182 .extra2 = &three,
1183 },
1184 #ifdef CONFIG_COMPACTION
1185 {
1186 .procname = "compact_memory",
1187 .data = &sysctl_compact_memory,
1188 .maxlen = sizeof(int),
1189 .mode = 0200,
1190 .proc_handler = sysctl_compaction_handler,
1191 },
1192 {
1193 .procname = "extfrag_threshold",
1194 .data = &sysctl_extfrag_threshold,
1195 .maxlen = sizeof(int),
1196 .mode = 0644,
1197 .proc_handler = sysctl_extfrag_handler,
1198 .extra1 = &min_extfrag_threshold,
1199 .extra2 = &max_extfrag_threshold,
1200 },
1201
1202 #endif /* CONFIG_COMPACTION */
1203 {
1204 .procname = "min_free_kbytes",
1205 .data = &min_free_kbytes,
1206 .maxlen = sizeof(min_free_kbytes),
1207 .mode = 0644,
1208 .proc_handler = min_free_kbytes_sysctl_handler,
1209 .extra1 = &zero,
1210 },
1211 {
1212 .procname = "percpu_pagelist_fraction",
1213 .data = &percpu_pagelist_fraction,
1214 .maxlen = sizeof(percpu_pagelist_fraction),
1215 .mode = 0644,
1216 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1217 .extra1 = &min_percpu_pagelist_fract,
1218 },
1219 #ifdef CONFIG_MMU
1220 {
1221 .procname = "max_map_count",
1222 .data = &sysctl_max_map_count,
1223 .maxlen = sizeof(sysctl_max_map_count),
1224 .mode = 0644,
1225 .proc_handler = proc_dointvec_minmax,
1226 .extra1 = &zero,
1227 },
1228 #else
1229 {
1230 .procname = "nr_trim_pages",
1231 .data = &sysctl_nr_trim_pages,
1232 .maxlen = sizeof(sysctl_nr_trim_pages),
1233 .mode = 0644,
1234 .proc_handler = proc_dointvec_minmax,
1235 .extra1 = &zero,
1236 },
1237 #endif
1238 {
1239 .procname = "laptop_mode",
1240 .data = &laptop_mode,
1241 .maxlen = sizeof(laptop_mode),
1242 .mode = 0644,
1243 .proc_handler = proc_dointvec_jiffies,
1244 },
1245 {
1246 .procname = "block_dump",
1247 .data = &block_dump,
1248 .maxlen = sizeof(block_dump),
1249 .mode = 0644,
1250 .proc_handler = proc_dointvec,
1251 .extra1 = &zero,
1252 },
1253 {
1254 .procname = "vfs_cache_pressure",
1255 .data = &sysctl_vfs_cache_pressure,
1256 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1257 .mode = 0644,
1258 .proc_handler = proc_dointvec,
1259 .extra1 = &zero,
1260 },
1261 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1262 {
1263 .procname = "legacy_va_layout",
1264 .data = &sysctl_legacy_va_layout,
1265 .maxlen = sizeof(sysctl_legacy_va_layout),
1266 .mode = 0644,
1267 .proc_handler = proc_dointvec,
1268 .extra1 = &zero,
1269 },
1270 #endif
1271 #ifdef CONFIG_NUMA
1272 {
1273 .procname = "zone_reclaim_mode",
1274 .data = &zone_reclaim_mode,
1275 .maxlen = sizeof(zone_reclaim_mode),
1276 .mode = 0644,
1277 .proc_handler = proc_dointvec,
1278 .extra1 = &zero,
1279 },
1280 {
1281 .procname = "min_unmapped_ratio",
1282 .data = &sysctl_min_unmapped_ratio,
1283 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1284 .mode = 0644,
1285 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1286 .extra1 = &zero,
1287 .extra2 = &one_hundred,
1288 },
1289 {
1290 .procname = "min_slab_ratio",
1291 .data = &sysctl_min_slab_ratio,
1292 .maxlen = sizeof(sysctl_min_slab_ratio),
1293 .mode = 0644,
1294 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1295 .extra1 = &zero,
1296 .extra2 = &one_hundred,
1297 },
1298 #endif
1299 #ifdef CONFIG_SMP
1300 {
1301 .procname = "stat_interval",
1302 .data = &sysctl_stat_interval,
1303 .maxlen = sizeof(sysctl_stat_interval),
1304 .mode = 0644,
1305 .proc_handler = proc_dointvec_jiffies,
1306 },
1307 #endif
1308 #ifdef CONFIG_MMU
1309 {
1310 .procname = "mmap_min_addr",
1311 .data = &dac_mmap_min_addr,
1312 .maxlen = sizeof(unsigned long),
1313 .mode = 0644,
1314 .proc_handler = mmap_min_addr_handler,
1315 },
1316 #endif
1317 #ifdef CONFIG_NUMA
1318 {
1319 .procname = "numa_zonelist_order",
1320 .data = &numa_zonelist_order,
1321 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1322 .mode = 0644,
1323 .proc_handler = numa_zonelist_order_handler,
1324 },
1325 #endif
1326 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1327 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1328 {
1329 .procname = "vdso_enabled",
1330 .data = &vdso_enabled,
1331 .maxlen = sizeof(vdso_enabled),
1332 .mode = 0644,
1333 .proc_handler = proc_dointvec,
1334 .extra1 = &zero,
1335 },
1336 #endif
1337 #ifdef CONFIG_HIGHMEM
1338 {
1339 .procname = "highmem_is_dirtyable",
1340 .data = &vm_highmem_is_dirtyable,
1341 .maxlen = sizeof(vm_highmem_is_dirtyable),
1342 .mode = 0644,
1343 .proc_handler = proc_dointvec_minmax,
1344 .extra1 = &zero,
1345 .extra2 = &one,
1346 },
1347 #endif
1348 {
1349 .procname = "scan_unevictable_pages",
1350 .data = &scan_unevictable_pages,
1351 .maxlen = sizeof(scan_unevictable_pages),
1352 .mode = 0644,
1353 .proc_handler = scan_unevictable_handler,
1354 },
1355 #ifdef CONFIG_MEMORY_FAILURE
1356 {
1357 .procname = "memory_failure_early_kill",
1358 .data = &sysctl_memory_failure_early_kill,
1359 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1360 .mode = 0644,
1361 .proc_handler = proc_dointvec_minmax,
1362 .extra1 = &zero,
1363 .extra2 = &one,
1364 },
1365 {
1366 .procname = "memory_failure_recovery",
1367 .data = &sysctl_memory_failure_recovery,
1368 .maxlen = sizeof(sysctl_memory_failure_recovery),
1369 .mode = 0644,
1370 .proc_handler = proc_dointvec_minmax,
1371 .extra1 = &zero,
1372 .extra2 = &one,
1373 },
1374 #endif
1375 { }
1376 };
1377
1378 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1379 static struct ctl_table binfmt_misc_table[] = {
1380 { }
1381 };
1382 #endif
1383
1384 static struct ctl_table fs_table[] = {
1385 {
1386 .procname = "inode-nr",
1387 .data = &inodes_stat,
1388 .maxlen = 2*sizeof(int),
1389 .mode = 0444,
1390 .proc_handler = proc_nr_inodes,
1391 },
1392 {
1393 .procname = "inode-state",
1394 .data = &inodes_stat,
1395 .maxlen = 7*sizeof(int),
1396 .mode = 0444,
1397 .proc_handler = proc_nr_inodes,
1398 },
1399 {
1400 .procname = "file-nr",
1401 .data = &files_stat,
1402 .maxlen = sizeof(files_stat),
1403 .mode = 0444,
1404 .proc_handler = proc_nr_files,
1405 },
1406 {
1407 .procname = "file-max",
1408 .data = &files_stat.max_files,
1409 .maxlen = sizeof(files_stat.max_files),
1410 .mode = 0644,
1411 .proc_handler = proc_doulongvec_minmax,
1412 },
1413 {
1414 .procname = "nr_open",
1415 .data = &sysctl_nr_open,
1416 .maxlen = sizeof(int),
1417 .mode = 0644,
1418 .proc_handler = proc_dointvec_minmax,
1419 .extra1 = &sysctl_nr_open_min,
1420 .extra2 = &sysctl_nr_open_max,
1421 },
1422 {
1423 .procname = "dentry-state",
1424 .data = &dentry_stat,
1425 .maxlen = 6*sizeof(int),
1426 .mode = 0444,
1427 .proc_handler = proc_nr_dentry,
1428 },
1429 {
1430 .procname = "overflowuid",
1431 .data = &fs_overflowuid,
1432 .maxlen = sizeof(int),
1433 .mode = 0644,
1434 .proc_handler = proc_dointvec_minmax,
1435 .extra1 = &minolduid,
1436 .extra2 = &maxolduid,
1437 },
1438 {
1439 .procname = "overflowgid",
1440 .data = &fs_overflowgid,
1441 .maxlen = sizeof(int),
1442 .mode = 0644,
1443 .proc_handler = proc_dointvec_minmax,
1444 .extra1 = &minolduid,
1445 .extra2 = &maxolduid,
1446 },
1447 #ifdef CONFIG_FILE_LOCKING
1448 {
1449 .procname = "leases-enable",
1450 .data = &leases_enable,
1451 .maxlen = sizeof(int),
1452 .mode = 0644,
1453 .proc_handler = proc_dointvec,
1454 },
1455 #endif
1456 #ifdef CONFIG_DNOTIFY
1457 {
1458 .procname = "dir-notify-enable",
1459 .data = &dir_notify_enable,
1460 .maxlen = sizeof(int),
1461 .mode = 0644,
1462 .proc_handler = proc_dointvec,
1463 },
1464 #endif
1465 #ifdef CONFIG_MMU
1466 #ifdef CONFIG_FILE_LOCKING
1467 {
1468 .procname = "lease-break-time",
1469 .data = &lease_break_time,
1470 .maxlen = sizeof(int),
1471 .mode = 0644,
1472 .proc_handler = proc_dointvec,
1473 },
1474 #endif
1475 #ifdef CONFIG_AIO
1476 {
1477 .procname = "aio-nr",
1478 .data = &aio_nr,
1479 .maxlen = sizeof(aio_nr),
1480 .mode = 0444,
1481 .proc_handler = proc_doulongvec_minmax,
1482 },
1483 {
1484 .procname = "aio-max-nr",
1485 .data = &aio_max_nr,
1486 .maxlen = sizeof(aio_max_nr),
1487 .mode = 0644,
1488 .proc_handler = proc_doulongvec_minmax,
1489 },
1490 #endif /* CONFIG_AIO */
1491 #ifdef CONFIG_INOTIFY_USER
1492 {
1493 .procname = "inotify",
1494 .mode = 0555,
1495 .child = inotify_table,
1496 },
1497 #endif
1498 #ifdef CONFIG_EPOLL
1499 {
1500 .procname = "epoll",
1501 .mode = 0555,
1502 .child = epoll_table,
1503 },
1504 #endif
1505 #endif
1506 {
1507 .procname = "protected_symlinks",
1508 .data = &sysctl_protected_symlinks,
1509 .maxlen = sizeof(int),
1510 .mode = 0600,
1511 .proc_handler = proc_dointvec_minmax,
1512 .extra1 = &zero,
1513 .extra2 = &one,
1514 },
1515 {
1516 .procname = "protected_hardlinks",
1517 .data = &sysctl_protected_hardlinks,
1518 .maxlen = sizeof(int),
1519 .mode = 0600,
1520 .proc_handler = proc_dointvec_minmax,
1521 .extra1 = &zero,
1522 .extra2 = &one,
1523 },
1524 {
1525 .procname = "suid_dumpable",
1526 .data = &suid_dumpable,
1527 .maxlen = sizeof(int),
1528 .mode = 0644,
1529 .proc_handler = proc_dointvec_minmax_coredump,
1530 .extra1 = &zero,
1531 .extra2 = &two,
1532 },
1533 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1534 {
1535 .procname = "binfmt_misc",
1536 .mode = 0555,
1537 .child = binfmt_misc_table,
1538 },
1539 #endif
1540 {
1541 .procname = "pipe-max-size",
1542 .data = &pipe_max_size,
1543 .maxlen = sizeof(int),
1544 .mode = 0644,
1545 .proc_handler = &pipe_proc_fn,
1546 .extra1 = &pipe_min_size,
1547 },
1548 { }
1549 };
1550
1551 static struct ctl_table debug_table[] = {
1552 #if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
1553 defined(CONFIG_S390) || defined(CONFIG_TILE) || defined(CONFIG_ARM64)
1554 {
1555 .procname = "exception-trace",
1556 .data = &show_unhandled_signals,
1557 .maxlen = sizeof(int),
1558 .mode = 0644,
1559 .proc_handler = proc_dointvec
1560 },
1561 #endif
1562 #if defined(CONFIG_OPTPROBES)
1563 {
1564 .procname = "kprobes-optimization",
1565 .data = &sysctl_kprobes_optimization,
1566 .maxlen = sizeof(int),
1567 .mode = 0644,
1568 .proc_handler = proc_kprobes_optimization_handler,
1569 .extra1 = &zero,
1570 .extra2 = &one,
1571 },
1572 #endif
1573 { }
1574 };
1575
1576 static struct ctl_table dev_table[] = {
1577 { }
1578 };
1579
1580 int __init sysctl_init(void)
1581 {
1582 struct ctl_table_header *hdr;
1583
1584 hdr = register_sysctl_table(sysctl_base_table);
1585 kmemleak_not_leak(hdr);
1586 return 0;
1587 }
1588
1589 #endif /* CONFIG_SYSCTL */
1590
1591 /*
1592 * /proc/sys support
1593 */
1594
1595 #ifdef CONFIG_PROC_SYSCTL
1596
1597 static int _proc_do_string(void* data, int maxlen, int write,
1598 void __user *buffer,
1599 size_t *lenp, loff_t *ppos)
1600 {
1601 size_t len;
1602 char __user *p;
1603 char c;
1604
1605 if (!data || !maxlen || !*lenp) {
1606 *lenp = 0;
1607 return 0;
1608 }
1609
1610 if (write) {
1611 len = 0;
1612 p = buffer;
1613 while (len < *lenp) {
1614 if (get_user(c, p++))
1615 return -EFAULT;
1616 if (c == 0 || c == '\n')
1617 break;
1618 len++;
1619 }
1620 if (len >= maxlen)
1621 len = maxlen-1;
1622 if(copy_from_user(data, buffer, len))
1623 return -EFAULT;
1624 ((char *) data)[len] = 0;
1625 *ppos += *lenp;
1626 } else {
1627 len = strlen(data);
1628 if (len > maxlen)
1629 len = maxlen;
1630
1631 if (*ppos > len) {
1632 *lenp = 0;
1633 return 0;
1634 }
1635
1636 data += *ppos;
1637 len -= *ppos;
1638
1639 if (len > *lenp)
1640 len = *lenp;
1641 if (len)
1642 if(copy_to_user(buffer, data, len))
1643 return -EFAULT;
1644 if (len < *lenp) {
1645 if(put_user('\n', ((char __user *) buffer) + len))
1646 return -EFAULT;
1647 len++;
1648 }
1649 *lenp = len;
1650 *ppos += len;
1651 }
1652 return 0;
1653 }
1654
1655 /**
1656 * proc_dostring - read a string sysctl
1657 * @table: the sysctl table
1658 * @write: %TRUE if this is a write to the sysctl file
1659 * @buffer: the user buffer
1660 * @lenp: the size of the user buffer
1661 * @ppos: file position
1662 *
1663 * Reads/writes a string from/to the user buffer. If the kernel
1664 * buffer provided is not large enough to hold the string, the
1665 * string is truncated. The copied string is %NULL-terminated.
1666 * If the string is being read by the user process, it is copied
1667 * and a newline '\n' is added. It is truncated if the buffer is
1668 * not large enough.
1669 *
1670 * Returns 0 on success.
1671 */
1672 int proc_dostring(struct ctl_table *table, int write,
1673 void __user *buffer, size_t *lenp, loff_t *ppos)
1674 {
1675 return _proc_do_string(table->data, table->maxlen, write,
1676 buffer, lenp, ppos);
1677 }
1678
1679 static size_t proc_skip_spaces(char **buf)
1680 {
1681 size_t ret;
1682 char *tmp = skip_spaces(*buf);
1683 ret = tmp - *buf;
1684 *buf = tmp;
1685 return ret;
1686 }
1687
1688 static void proc_skip_char(char **buf, size_t *size, const char v)
1689 {
1690 while (*size) {
1691 if (**buf != v)
1692 break;
1693 (*size)--;
1694 (*buf)++;
1695 }
1696 }
1697
1698 #define TMPBUFLEN 22
1699 /**
1700 * proc_get_long - reads an ASCII formatted integer from a user buffer
1701 *
1702 * @buf: a kernel buffer
1703 * @size: size of the kernel buffer
1704 * @val: this is where the number will be stored
1705 * @neg: set to %TRUE if number is negative
1706 * @perm_tr: a vector which contains the allowed trailers
1707 * @perm_tr_len: size of the perm_tr vector
1708 * @tr: pointer to store the trailer character
1709 *
1710 * In case of success %0 is returned and @buf and @size are updated with
1711 * the amount of bytes read. If @tr is non-NULL and a trailing
1712 * character exists (size is non-zero after returning from this
1713 * function), @tr is updated with the trailing character.
1714 */
1715 static int proc_get_long(char **buf, size_t *size,
1716 unsigned long *val, bool *neg,
1717 const char *perm_tr, unsigned perm_tr_len, char *tr)
1718 {
1719 int len;
1720 char *p, tmp[TMPBUFLEN];
1721
1722 if (!*size)
1723 return -EINVAL;
1724
1725 len = *size;
1726 if (len > TMPBUFLEN - 1)
1727 len = TMPBUFLEN - 1;
1728
1729 memcpy(tmp, *buf, len);
1730
1731 tmp[len] = 0;
1732 p = tmp;
1733 if (*p == '-' && *size > 1) {
1734 *neg = true;
1735 p++;
1736 } else
1737 *neg = false;
1738 if (!isdigit(*p))
1739 return -EINVAL;
1740
1741 *val = simple_strtoul(p, &p, 0);
1742
1743 len = p - tmp;
1744
1745 /* We don't know if the next char is whitespace thus we may accept
1746 * invalid integers (e.g. 1234...a) or two integers instead of one
1747 * (e.g. 123...1). So lets not allow such large numbers. */
1748 if (len == TMPBUFLEN - 1)
1749 return -EINVAL;
1750
1751 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
1752 return -EINVAL;
1753
1754 if (tr && (len < *size))
1755 *tr = *p;
1756
1757 *buf += len;
1758 *size -= len;
1759
1760 return 0;
1761 }
1762
1763 /**
1764 * proc_put_long - converts an integer to a decimal ASCII formatted string
1765 *
1766 * @buf: the user buffer
1767 * @size: the size of the user buffer
1768 * @val: the integer to be converted
1769 * @neg: sign of the number, %TRUE for negative
1770 *
1771 * In case of success %0 is returned and @buf and @size are updated with
1772 * the amount of bytes written.
1773 */
1774 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
1775 bool neg)
1776 {
1777 int len;
1778 char tmp[TMPBUFLEN], *p = tmp;
1779
1780 sprintf(p, "%s%lu", neg ? "-" : "", val);
1781 len = strlen(tmp);
1782 if (len > *size)
1783 len = *size;
1784 if (copy_to_user(*buf, tmp, len))
1785 return -EFAULT;
1786 *size -= len;
1787 *buf += len;
1788 return 0;
1789 }
1790 #undef TMPBUFLEN
1791
1792 static int proc_put_char(void __user **buf, size_t *size, char c)
1793 {
1794 if (*size) {
1795 char __user **buffer = (char __user **)buf;
1796 if (put_user(c, *buffer))
1797 return -EFAULT;
1798 (*size)--, (*buffer)++;
1799 *buf = *buffer;
1800 }
1801 return 0;
1802 }
1803
1804 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
1805 int *valp,
1806 int write, void *data)
1807 {
1808 if (write) {
1809 *valp = *negp ? -*lvalp : *lvalp;
1810 } else {
1811 int val = *valp;
1812 if (val < 0) {
1813 *negp = true;
1814 *lvalp = (unsigned long)-val;
1815 } else {
1816 *negp = false;
1817 *lvalp = (unsigned long)val;
1818 }
1819 }
1820 return 0;
1821 }
1822
1823 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
1824
1825 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1826 int write, void __user *buffer,
1827 size_t *lenp, loff_t *ppos,
1828 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1829 int write, void *data),
1830 void *data)
1831 {
1832 int *i, vleft, first = 1, err = 0;
1833 unsigned long page = 0;
1834 size_t left;
1835 char *kbuf;
1836
1837 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
1838 *lenp = 0;
1839 return 0;
1840 }
1841
1842 i = (int *) tbl_data;
1843 vleft = table->maxlen / sizeof(*i);
1844 left = *lenp;
1845
1846 if (!conv)
1847 conv = do_proc_dointvec_conv;
1848
1849 if (write) {
1850 if (left > PAGE_SIZE - 1)
1851 left = PAGE_SIZE - 1;
1852 page = __get_free_page(GFP_TEMPORARY);
1853 kbuf = (char *) page;
1854 if (!kbuf)
1855 return -ENOMEM;
1856 if (copy_from_user(kbuf, buffer, left)) {
1857 err = -EFAULT;
1858 goto free;
1859 }
1860 kbuf[left] = 0;
1861 }
1862
1863 for (; left && vleft--; i++, first=0) {
1864 unsigned long lval;
1865 bool neg;
1866
1867 if (write) {
1868 left -= proc_skip_spaces(&kbuf);
1869
1870 if (!left)
1871 break;
1872 err = proc_get_long(&kbuf, &left, &lval, &neg,
1873 proc_wspace_sep,
1874 sizeof(proc_wspace_sep), NULL);
1875 if (err)
1876 break;
1877 if (conv(&neg, &lval, i, 1, data)) {
1878 err = -EINVAL;
1879 break;
1880 }
1881 } else {
1882 if (conv(&neg, &lval, i, 0, data)) {
1883 err = -EINVAL;
1884 break;
1885 }
1886 if (!first)
1887 err = proc_put_char(&buffer, &left, '\t');
1888 if (err)
1889 break;
1890 err = proc_put_long(&buffer, &left, lval, neg);
1891 if (err)
1892 break;
1893 }
1894 }
1895
1896 if (!write && !first && left && !err)
1897 err = proc_put_char(&buffer, &left, '\n');
1898 if (write && !err && left)
1899 left -= proc_skip_spaces(&kbuf);
1900 free:
1901 if (write) {
1902 free_page(page);
1903 if (first)
1904 return err ? : -EINVAL;
1905 }
1906 *lenp -= left;
1907 *ppos += *lenp;
1908 return err;
1909 }
1910
1911 static int do_proc_dointvec(struct ctl_table *table, int write,
1912 void __user *buffer, size_t *lenp, loff_t *ppos,
1913 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1914 int write, void *data),
1915 void *data)
1916 {
1917 return __do_proc_dointvec(table->data, table, write,
1918 buffer, lenp, ppos, conv, data);
1919 }
1920
1921 /**
1922 * proc_dointvec - read a vector of integers
1923 * @table: the sysctl table
1924 * @write: %TRUE if this is a write to the sysctl file
1925 * @buffer: the user buffer
1926 * @lenp: the size of the user buffer
1927 * @ppos: file position
1928 *
1929 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1930 * values from/to the user buffer, treated as an ASCII string.
1931 *
1932 * Returns 0 on success.
1933 */
1934 int proc_dointvec(struct ctl_table *table, int write,
1935 void __user *buffer, size_t *lenp, loff_t *ppos)
1936 {
1937 return do_proc_dointvec(table,write,buffer,lenp,ppos,
1938 NULL,NULL);
1939 }
1940
1941 /*
1942 * Taint values can only be increased
1943 * This means we can safely use a temporary.
1944 */
1945 static int proc_taint(struct ctl_table *table, int write,
1946 void __user *buffer, size_t *lenp, loff_t *ppos)
1947 {
1948 struct ctl_table t;
1949 unsigned long tmptaint = get_taint();
1950 int err;
1951
1952 if (write && !capable(CAP_SYS_ADMIN))
1953 return -EPERM;
1954
1955 t = *table;
1956 t.data = &tmptaint;
1957 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
1958 if (err < 0)
1959 return err;
1960
1961 if (write) {
1962 /*
1963 * Poor man's atomic or. Not worth adding a primitive
1964 * to everyone's atomic.h for this
1965 */
1966 int i;
1967 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
1968 if ((tmptaint >> i) & 1)
1969 add_taint(i);
1970 }
1971 }
1972
1973 return err;
1974 }
1975
1976 #ifdef CONFIG_PRINTK
1977 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
1978 void __user *buffer, size_t *lenp, loff_t *ppos)
1979 {
1980 if (write && !capable(CAP_SYS_ADMIN))
1981 return -EPERM;
1982
1983 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
1984 }
1985 #endif
1986
1987 struct do_proc_dointvec_minmax_conv_param {
1988 int *min;
1989 int *max;
1990 };
1991
1992 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
1993 int *valp,
1994 int write, void *data)
1995 {
1996 struct do_proc_dointvec_minmax_conv_param *param = data;
1997 if (write) {
1998 int val = *negp ? -*lvalp : *lvalp;
1999 if ((param->min && *param->min > val) ||
2000 (param->max && *param->max < val))
2001 return -EINVAL;
2002 *valp = val;
2003 } else {
2004 int val = *valp;
2005 if (val < 0) {
2006 *negp = true;
2007 *lvalp = (unsigned long)-val;
2008 } else {
2009 *negp = false;
2010 *lvalp = (unsigned long)val;
2011 }
2012 }
2013 return 0;
2014 }
2015
2016 /**
2017 * proc_dointvec_minmax - read a vector of integers with min/max values
2018 * @table: the sysctl table
2019 * @write: %TRUE if this is a write to the sysctl file
2020 * @buffer: the user buffer
2021 * @lenp: the size of the user buffer
2022 * @ppos: file position
2023 *
2024 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2025 * values from/to the user buffer, treated as an ASCII string.
2026 *
2027 * This routine will ensure the values are within the range specified by
2028 * table->extra1 (min) and table->extra2 (max).
2029 *
2030 * Returns 0 on success.
2031 */
2032 int proc_dointvec_minmax(struct ctl_table *table, int write,
2033 void __user *buffer, size_t *lenp, loff_t *ppos)
2034 {
2035 struct do_proc_dointvec_minmax_conv_param param = {
2036 .min = (int *) table->extra1,
2037 .max = (int *) table->extra2,
2038 };
2039 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2040 do_proc_dointvec_minmax_conv, &param);
2041 }
2042
2043 static void validate_coredump_safety(void)
2044 {
2045 #ifdef CONFIG_COREDUMP
2046 if (suid_dumpable == SUID_DUMPABLE_SAFE &&
2047 core_pattern[0] != '/' && core_pattern[0] != '|') {
2048 printk(KERN_WARNING "Unsafe core_pattern used with "\
2049 "suid_dumpable=2. Pipe handler or fully qualified "\
2050 "core dump path required.\n");
2051 }
2052 #endif
2053 }
2054
2055 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2056 void __user *buffer, size_t *lenp, loff_t *ppos)
2057 {
2058 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2059 if (!error)
2060 validate_coredump_safety();
2061 return error;
2062 }
2063
2064 #ifdef CONFIG_COREDUMP
2065 static int proc_dostring_coredump(struct ctl_table *table, int write,
2066 void __user *buffer, size_t *lenp, loff_t *ppos)
2067 {
2068 int error = proc_dostring(table, write, buffer, lenp, ppos);
2069 if (!error)
2070 validate_coredump_safety();
2071 return error;
2072 }
2073 #endif
2074
2075 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2076 void __user *buffer,
2077 size_t *lenp, loff_t *ppos,
2078 unsigned long convmul,
2079 unsigned long convdiv)
2080 {
2081 unsigned long *i, *min, *max;
2082 int vleft, first = 1, err = 0;
2083 unsigned long page = 0;
2084 size_t left;
2085 char *kbuf;
2086
2087 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2088 *lenp = 0;
2089 return 0;
2090 }
2091
2092 i = (unsigned long *) data;
2093 min = (unsigned long *) table->extra1;
2094 max = (unsigned long *) table->extra2;
2095 vleft = table->maxlen / sizeof(unsigned long);
2096 left = *lenp;
2097
2098 if (write) {
2099 if (left > PAGE_SIZE - 1)
2100 left = PAGE_SIZE - 1;
2101 page = __get_free_page(GFP_TEMPORARY);
2102 kbuf = (char *) page;
2103 if (!kbuf)
2104 return -ENOMEM;
2105 if (copy_from_user(kbuf, buffer, left)) {
2106 err = -EFAULT;
2107 goto free;
2108 }
2109 kbuf[left] = 0;
2110 }
2111
2112 for (; left && vleft--; i++, first = 0) {
2113 unsigned long val;
2114
2115 if (write) {
2116 bool neg;
2117
2118 left -= proc_skip_spaces(&kbuf);
2119
2120 err = proc_get_long(&kbuf, &left, &val, &neg,
2121 proc_wspace_sep,
2122 sizeof(proc_wspace_sep), NULL);
2123 if (err)
2124 break;
2125 if (neg)
2126 continue;
2127 if ((min && val < *min) || (max && val > *max))
2128 continue;
2129 *i = val;
2130 } else {
2131 val = convdiv * (*i) / convmul;
2132 if (!first)
2133 err = proc_put_char(&buffer, &left, '\t');
2134 err = proc_put_long(&buffer, &left, val, false);
2135 if (err)
2136 break;
2137 }
2138 }
2139
2140 if (!write && !first && left && !err)
2141 err = proc_put_char(&buffer, &left, '\n');
2142 if (write && !err)
2143 left -= proc_skip_spaces(&kbuf);
2144 free:
2145 if (write) {
2146 free_page(page);
2147 if (first)
2148 return err ? : -EINVAL;
2149 }
2150 *lenp -= left;
2151 *ppos += *lenp;
2152 return err;
2153 }
2154
2155 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2156 void __user *buffer,
2157 size_t *lenp, loff_t *ppos,
2158 unsigned long convmul,
2159 unsigned long convdiv)
2160 {
2161 return __do_proc_doulongvec_minmax(table->data, table, write,
2162 buffer, lenp, ppos, convmul, convdiv);
2163 }
2164
2165 /**
2166 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2167 * @table: the sysctl table
2168 * @write: %TRUE if this is a write to the sysctl file
2169 * @buffer: the user buffer
2170 * @lenp: the size of the user buffer
2171 * @ppos: file position
2172 *
2173 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2174 * values from/to the user buffer, treated as an ASCII string.
2175 *
2176 * This routine will ensure the values are within the range specified by
2177 * table->extra1 (min) and table->extra2 (max).
2178 *
2179 * Returns 0 on success.
2180 */
2181 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2182 void __user *buffer, size_t *lenp, loff_t *ppos)
2183 {
2184 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2185 }
2186
2187 /**
2188 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2189 * @table: the sysctl table
2190 * @write: %TRUE if this is a write to the sysctl file
2191 * @buffer: the user buffer
2192 * @lenp: the size of the user buffer
2193 * @ppos: file position
2194 *
2195 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2196 * values from/to the user buffer, treated as an ASCII string. The values
2197 * are treated as milliseconds, and converted to jiffies when they are stored.
2198 *
2199 * This routine will ensure the values are within the range specified by
2200 * table->extra1 (min) and table->extra2 (max).
2201 *
2202 * Returns 0 on success.
2203 */
2204 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2205 void __user *buffer,
2206 size_t *lenp, loff_t *ppos)
2207 {
2208 return do_proc_doulongvec_minmax(table, write, buffer,
2209 lenp, ppos, HZ, 1000l);
2210 }
2211
2212
2213 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2214 int *valp,
2215 int write, void *data)
2216 {
2217 if (write) {
2218 if (*lvalp > LONG_MAX / HZ)
2219 return 1;
2220 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2221 } else {
2222 int val = *valp;
2223 unsigned long lval;
2224 if (val < 0) {
2225 *negp = true;
2226 lval = (unsigned long)-val;
2227 } else {
2228 *negp = false;
2229 lval = (unsigned long)val;
2230 }
2231 *lvalp = lval / HZ;
2232 }
2233 return 0;
2234 }
2235
2236 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2237 int *valp,
2238 int write, void *data)
2239 {
2240 if (write) {
2241 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2242 return 1;
2243 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2244 } else {
2245 int val = *valp;
2246 unsigned long lval;
2247 if (val < 0) {
2248 *negp = true;
2249 lval = (unsigned long)-val;
2250 } else {
2251 *negp = false;
2252 lval = (unsigned long)val;
2253 }
2254 *lvalp = jiffies_to_clock_t(lval);
2255 }
2256 return 0;
2257 }
2258
2259 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2260 int *valp,
2261 int write, void *data)
2262 {
2263 if (write) {
2264 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2265 } else {
2266 int val = *valp;
2267 unsigned long lval;
2268 if (val < 0) {
2269 *negp = true;
2270 lval = (unsigned long)-val;
2271 } else {
2272 *negp = false;
2273 lval = (unsigned long)val;
2274 }
2275 *lvalp = jiffies_to_msecs(lval);
2276 }
2277 return 0;
2278 }
2279
2280 /**
2281 * proc_dointvec_jiffies - read a vector of integers as seconds
2282 * @table: the sysctl table
2283 * @write: %TRUE if this is a write to the sysctl file
2284 * @buffer: the user buffer
2285 * @lenp: the size of the user buffer
2286 * @ppos: file position
2287 *
2288 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2289 * values from/to the user buffer, treated as an ASCII string.
2290 * The values read are assumed to be in seconds, and are converted into
2291 * jiffies.
2292 *
2293 * Returns 0 on success.
2294 */
2295 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2296 void __user *buffer, size_t *lenp, loff_t *ppos)
2297 {
2298 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2299 do_proc_dointvec_jiffies_conv,NULL);
2300 }
2301
2302 /**
2303 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2304 * @table: the sysctl table
2305 * @write: %TRUE if this is a write to the sysctl file
2306 * @buffer: the user buffer
2307 * @lenp: the size of the user buffer
2308 * @ppos: pointer to the file position
2309 *
2310 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2311 * values from/to the user buffer, treated as an ASCII string.
2312 * The values read are assumed to be in 1/USER_HZ seconds, and
2313 * are converted into jiffies.
2314 *
2315 * Returns 0 on success.
2316 */
2317 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2318 void __user *buffer, size_t *lenp, loff_t *ppos)
2319 {
2320 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2321 do_proc_dointvec_userhz_jiffies_conv,NULL);
2322 }
2323
2324 /**
2325 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2326 * @table: the sysctl table
2327 * @write: %TRUE if this is a write to the sysctl file
2328 * @buffer: the user buffer
2329 * @lenp: the size of the user buffer
2330 * @ppos: file position
2331 * @ppos: the current position in the file
2332 *
2333 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2334 * values from/to the user buffer, treated as an ASCII string.
2335 * The values read are assumed to be in 1/1000 seconds, and
2336 * are converted into jiffies.
2337 *
2338 * Returns 0 on success.
2339 */
2340 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2341 void __user *buffer, size_t *lenp, loff_t *ppos)
2342 {
2343 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2344 do_proc_dointvec_ms_jiffies_conv, NULL);
2345 }
2346
2347 static int proc_do_cad_pid(struct ctl_table *table, int write,
2348 void __user *buffer, size_t *lenp, loff_t *ppos)
2349 {
2350 struct pid *new_pid;
2351 pid_t tmp;
2352 int r;
2353
2354 tmp = pid_vnr(cad_pid);
2355
2356 r = __do_proc_dointvec(&tmp, table, write, buffer,
2357 lenp, ppos, NULL, NULL);
2358 if (r || !write)
2359 return r;
2360
2361 new_pid = find_get_pid(tmp);
2362 if (!new_pid)
2363 return -ESRCH;
2364
2365 put_pid(xchg(&cad_pid, new_pid));
2366 return 0;
2367 }
2368
2369 /**
2370 * proc_do_large_bitmap - read/write from/to a large bitmap
2371 * @table: the sysctl table
2372 * @write: %TRUE if this is a write to the sysctl file
2373 * @buffer: the user buffer
2374 * @lenp: the size of the user buffer
2375 * @ppos: file position
2376 *
2377 * The bitmap is stored at table->data and the bitmap length (in bits)
2378 * in table->maxlen.
2379 *
2380 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2381 * large bitmaps may be represented in a compact manner. Writing into
2382 * the file will clear the bitmap then update it with the given input.
2383 *
2384 * Returns 0 on success.
2385 */
2386 int proc_do_large_bitmap(struct ctl_table *table, int write,
2387 void __user *buffer, size_t *lenp, loff_t *ppos)
2388 {
2389 int err = 0;
2390 bool first = 1;
2391 size_t left = *lenp;
2392 unsigned long bitmap_len = table->maxlen;
2393 unsigned long *bitmap = (unsigned long *) table->data;
2394 unsigned long *tmp_bitmap = NULL;
2395 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2396
2397 if (!bitmap_len || !left || (*ppos && !write)) {
2398 *lenp = 0;
2399 return 0;
2400 }
2401
2402 if (write) {
2403 unsigned long page = 0;
2404 char *kbuf;
2405
2406 if (left > PAGE_SIZE - 1)
2407 left = PAGE_SIZE - 1;
2408
2409 page = __get_free_page(GFP_TEMPORARY);
2410 kbuf = (char *) page;
2411 if (!kbuf)
2412 return -ENOMEM;
2413 if (copy_from_user(kbuf, buffer, left)) {
2414 free_page(page);
2415 return -EFAULT;
2416 }
2417 kbuf[left] = 0;
2418
2419 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2420 GFP_KERNEL);
2421 if (!tmp_bitmap) {
2422 free_page(page);
2423 return -ENOMEM;
2424 }
2425 proc_skip_char(&kbuf, &left, '\n');
2426 while (!err && left) {
2427 unsigned long val_a, val_b;
2428 bool neg;
2429
2430 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2431 sizeof(tr_a), &c);
2432 if (err)
2433 break;
2434 if (val_a >= bitmap_len || neg) {
2435 err = -EINVAL;
2436 break;
2437 }
2438
2439 val_b = val_a;
2440 if (left) {
2441 kbuf++;
2442 left--;
2443 }
2444
2445 if (c == '-') {
2446 err = proc_get_long(&kbuf, &left, &val_b,
2447 &neg, tr_b, sizeof(tr_b),
2448 &c);
2449 if (err)
2450 break;
2451 if (val_b >= bitmap_len || neg ||
2452 val_a > val_b) {
2453 err = -EINVAL;
2454 break;
2455 }
2456 if (left) {
2457 kbuf++;
2458 left--;
2459 }
2460 }
2461
2462 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2463 first = 0;
2464 proc_skip_char(&kbuf, &left, '\n');
2465 }
2466 free_page(page);
2467 } else {
2468 unsigned long bit_a, bit_b = 0;
2469
2470 while (left) {
2471 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2472 if (bit_a >= bitmap_len)
2473 break;
2474 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2475 bit_a + 1) - 1;
2476
2477 if (!first) {
2478 err = proc_put_char(&buffer, &left, ',');
2479 if (err)
2480 break;
2481 }
2482 err = proc_put_long(&buffer, &left, bit_a, false);
2483 if (err)
2484 break;
2485 if (bit_a != bit_b) {
2486 err = proc_put_char(&buffer, &left, '-');
2487 if (err)
2488 break;
2489 err = proc_put_long(&buffer, &left, bit_b, false);
2490 if (err)
2491 break;
2492 }
2493
2494 first = 0; bit_b++;
2495 }
2496 if (!err)
2497 err = proc_put_char(&buffer, &left, '\n');
2498 }
2499
2500 if (!err) {
2501 if (write) {
2502 if (*ppos)
2503 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2504 else
2505 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2506 }
2507 kfree(tmp_bitmap);
2508 *lenp -= left;
2509 *ppos += *lenp;
2510 return 0;
2511 } else {
2512 kfree(tmp_bitmap);
2513 return err;
2514 }
2515 }
2516
2517 #else /* CONFIG_PROC_SYSCTL */
2518
2519 int proc_dostring(struct ctl_table *table, int write,
2520 void __user *buffer, size_t *lenp, loff_t *ppos)
2521 {
2522 return -ENOSYS;
2523 }
2524
2525 int proc_dointvec(struct ctl_table *table, int write,
2526 void __user *buffer, size_t *lenp, loff_t *ppos)
2527 {
2528 return -ENOSYS;
2529 }
2530
2531 int proc_dointvec_minmax(struct ctl_table *table, int write,
2532 void __user *buffer, size_t *lenp, loff_t *ppos)
2533 {
2534 return -ENOSYS;
2535 }
2536
2537 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2538 void __user *buffer, size_t *lenp, loff_t *ppos)
2539 {
2540 return -ENOSYS;
2541 }
2542
2543 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2544 void __user *buffer, size_t *lenp, loff_t *ppos)
2545 {
2546 return -ENOSYS;
2547 }
2548
2549 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2550 void __user *buffer, size_t *lenp, loff_t *ppos)
2551 {
2552 return -ENOSYS;
2553 }
2554
2555 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2556 void __user *buffer, size_t *lenp, loff_t *ppos)
2557 {
2558 return -ENOSYS;
2559 }
2560
2561 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2562 void __user *buffer,
2563 size_t *lenp, loff_t *ppos)
2564 {
2565 return -ENOSYS;
2566 }
2567
2568
2569 #endif /* CONFIG_PROC_SYSCTL */
2570
2571 /*
2572 * No sense putting this after each symbol definition, twice,
2573 * exception granted :-)
2574 */
2575 EXPORT_SYMBOL(proc_dointvec);
2576 EXPORT_SYMBOL(proc_dointvec_jiffies);
2577 EXPORT_SYMBOL(proc_dointvec_minmax);
2578 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2579 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2580 EXPORT_SYMBOL(proc_dostring);
2581 EXPORT_SYMBOL(proc_doulongvec_minmax);
2582 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
Linux kernel - Deliverables
This page took 0.082686 seconds and 6 git commands to generate.