projects / deliverable / linux.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Merge branch 'acpi-hotplug'
[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
66 #include <asm/uaccess.h>
67 #include <asm/processor.h>
68
69 #ifdef CONFIG_X86
70 #include <asm/nmi.h>
71 #include <asm/stacktrace.h>
72 #include <asm/io.h>
73 #endif
74 #ifdef CONFIG_SPARC
75 #include <asm/setup.h>
76 #endif
77 #ifdef CONFIG_BSD_PROCESS_ACCT
78 #include <linux/acct.h>
79 #endif
80 #ifdef CONFIG_RT_MUTEXES
81 #include <linux/rtmutex.h>
82 #endif
83 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
84 #include <linux/lockdep.h>
85 #endif
86 #ifdef CONFIG_CHR_DEV_SG
87 #include <scsi/sg.h>
88 #endif
89
90 #ifdef CONFIG_LOCKUP_DETECTOR
91 #include <linux/nmi.h>
92 #endif
93
94
95 #if defined(CONFIG_SYSCTL)
96
97 /* External variables not in a header file. */
98 extern int sysctl_overcommit_memory;
99 extern int sysctl_overcommit_ratio;
100 extern int max_threads;
101 extern int suid_dumpable;
102 #ifdef CONFIG_COREDUMP
103 extern int core_uses_pid;
104 extern char core_pattern[];
105 extern unsigned int core_pipe_limit;
106 #endif
107 extern int pid_max;
108 extern int pid_max_min, pid_max_max;
109 extern int percpu_pagelist_fraction;
110 extern int compat_log;
111 extern int latencytop_enabled;
112 extern int sysctl_nr_open_min, sysctl_nr_open_max;
113 #ifndef CONFIG_MMU
114 extern int sysctl_nr_trim_pages;
115 #endif
116 #ifdef CONFIG_BLOCK
117 extern int blk_iopoll_enabled;
118 #endif
119
120 /* Constants used for minimum and maximum */
121 #ifdef CONFIG_LOCKUP_DETECTOR
122 static int sixty = 60;
123 #endif
124
125 static int zero;
126 static int __maybe_unused one = 1;
127 static int __maybe_unused two = 2;
128 static int __maybe_unused three = 3;
129 static unsigned long one_ul = 1;
130 static int one_hundred = 100;
131 #ifdef CONFIG_PRINTK
132 static int ten_thousand = 10000;
133 #endif
134
135 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
136 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
137
138 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
139 static int maxolduid = 65535;
140 static int minolduid;
141 static int min_percpu_pagelist_fract = 8;
142
143 static int ngroups_max = NGROUPS_MAX;
144 static const int cap_last_cap = CAP_LAST_CAP;
145
146 #ifdef CONFIG_INOTIFY_USER
147 #include <linux/inotify.h>
148 #endif
149 #ifdef CONFIG_SPARC
150 #endif
151
152 #ifdef CONFIG_SPARC64
153 extern int sysctl_tsb_ratio;
154 #endif
155
156 #ifdef __hppa__
157 extern int pwrsw_enabled;
158 #endif
159
160 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
161 extern int unaligned_enabled;
162 #endif
163
164 #ifdef CONFIG_IA64
165 extern int unaligned_dump_stack;
166 #endif
167
168 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
169 extern int no_unaligned_warning;
170 #endif
171
172 #ifdef CONFIG_PROC_SYSCTL
173 static int proc_do_cad_pid(struct ctl_table *table, int write,
174 void __user *buffer, size_t *lenp, loff_t *ppos);
175 static int proc_taint(struct ctl_table *table, int write,
176 void __user *buffer, size_t *lenp, loff_t *ppos);
177 #endif
178
179 #ifdef CONFIG_PRINTK
180 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
181 void __user *buffer, size_t *lenp, loff_t *ppos);
182 #endif
183
184 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
185 void __user *buffer, size_t *lenp, loff_t *ppos);
186 #ifdef CONFIG_COREDUMP
187 static int proc_dostring_coredump(struct ctl_table *table, int write,
188 void __user *buffer, size_t *lenp, loff_t *ppos);
189 #endif
190
191 #ifdef CONFIG_MAGIC_SYSRQ
192 /* Note: sysrq code uses it's own private copy */
193 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
194
195 static int sysrq_sysctl_handler(ctl_table *table, int write,
196 void __user *buffer, size_t *lenp,
197 loff_t *ppos)
198 {
199 int error;
200
201 error = proc_dointvec(table, write, buffer, lenp, ppos);
202 if (error)
203 return error;
204
205 if (write)
206 sysrq_toggle_support(__sysrq_enabled);
207
208 return 0;
209 }
210
211 #endif
212
213 static struct ctl_table kern_table[];
214 static struct ctl_table vm_table[];
215 static struct ctl_table fs_table[];
216 static struct ctl_table debug_table[];
217 static struct ctl_table dev_table[];
218 extern struct ctl_table random_table[];
219 #ifdef CONFIG_EPOLL
220 extern struct ctl_table epoll_table[];
221 #endif
222
223 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
224 int sysctl_legacy_va_layout;
225 #endif
226
227 /* The default sysctl tables: */
228
229 static struct ctl_table sysctl_base_table[] = {
230 {
231 .procname = "kernel",
232 .mode = 0555,
233 .child = kern_table,
234 },
235 {
236 .procname = "vm",
237 .mode = 0555,
238 .child = vm_table,
239 },
240 {
241 .procname = "fs",
242 .mode = 0555,
243 .child = fs_table,
244 },
245 {
246 .procname = "debug",
247 .mode = 0555,
248 .child = debug_table,
249 },
250 {
251 .procname = "dev",
252 .mode = 0555,
253 .child = dev_table,
254 },
255 { }
256 };
257
258 #ifdef CONFIG_SCHED_DEBUG
259 static int min_sched_granularity_ns = 100000; /* 100 usecs */
260 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
261 static int min_wakeup_granularity_ns; /* 0 usecs */
262 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
263 #ifdef CONFIG_SMP
264 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
265 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
266 #endif /* CONFIG_SMP */
267 #endif /* CONFIG_SCHED_DEBUG */
268
269 #ifdef CONFIG_COMPACTION
270 static int min_extfrag_threshold;
271 static int max_extfrag_threshold = 1000;
272 #endif
273
274 static struct ctl_table kern_table[] = {
275 {
276 .procname = "sched_child_runs_first",
277 .data = &sysctl_sched_child_runs_first,
278 .maxlen = sizeof(unsigned int),
279 .mode = 0644,
280 .proc_handler = proc_dointvec,
281 },
282 #ifdef CONFIG_SCHED_DEBUG
283 {
284 .procname = "sched_min_granularity_ns",
285 .data = &sysctl_sched_min_granularity,
286 .maxlen = sizeof(unsigned int),
287 .mode = 0644,
288 .proc_handler = sched_proc_update_handler,
289 .extra1 = &min_sched_granularity_ns,
290 .extra2 = &max_sched_granularity_ns,
291 },
292 {
293 .procname = "sched_latency_ns",
294 .data = &sysctl_sched_latency,
295 .maxlen = sizeof(unsigned int),
296 .mode = 0644,
297 .proc_handler = sched_proc_update_handler,
298 .extra1 = &min_sched_granularity_ns,
299 .extra2 = &max_sched_granularity_ns,
300 },
301 {
302 .procname = "sched_wakeup_granularity_ns",
303 .data = &sysctl_sched_wakeup_granularity,
304 .maxlen = sizeof(unsigned int),
305 .mode = 0644,
306 .proc_handler = sched_proc_update_handler,
307 .extra1 = &min_wakeup_granularity_ns,
308 .extra2 = &max_wakeup_granularity_ns,
309 },
310 #ifdef CONFIG_SMP
311 {
312 .procname = "sched_tunable_scaling",
313 .data = &sysctl_sched_tunable_scaling,
314 .maxlen = sizeof(enum sched_tunable_scaling),
315 .mode = 0644,
316 .proc_handler = sched_proc_update_handler,
317 .extra1 = &min_sched_tunable_scaling,
318 .extra2 = &max_sched_tunable_scaling,
319 },
320 {
321 .procname = "sched_migration_cost_ns",
322 .data = &sysctl_sched_migration_cost,
323 .maxlen = sizeof(unsigned int),
324 .mode = 0644,
325 .proc_handler = proc_dointvec,
326 },
327 {
328 .procname = "sched_nr_migrate",
329 .data = &sysctl_sched_nr_migrate,
330 .maxlen = sizeof(unsigned int),
331 .mode = 0644,
332 .proc_handler = proc_dointvec,
333 },
334 {
335 .procname = "sched_time_avg_ms",
336 .data = &sysctl_sched_time_avg,
337 .maxlen = sizeof(unsigned int),
338 .mode = 0644,
339 .proc_handler = proc_dointvec,
340 },
341 {
342 .procname = "sched_shares_window_ns",
343 .data = &sysctl_sched_shares_window,
344 .maxlen = sizeof(unsigned int),
345 .mode = 0644,
346 .proc_handler = proc_dointvec,
347 },
348 {
349 .procname = "timer_migration",
350 .data = &sysctl_timer_migration,
351 .maxlen = sizeof(unsigned int),
352 .mode = 0644,
353 .proc_handler = proc_dointvec_minmax,
354 .extra1 = &zero,
355 .extra2 = &one,
356 },
357 #endif /* CONFIG_SMP */
358 #ifdef CONFIG_NUMA_BALANCING
359 {
360 .procname = "numa_balancing_scan_delay_ms",
361 .data = &sysctl_numa_balancing_scan_delay,
362 .maxlen = sizeof(unsigned int),
363 .mode = 0644,
364 .proc_handler = proc_dointvec,
365 },
366 {
367 .procname = "numa_balancing_scan_period_min_ms",
368 .data = &sysctl_numa_balancing_scan_period_min,
369 .maxlen = sizeof(unsigned int),
370 .mode = 0644,
371 .proc_handler = proc_dointvec,
372 },
373 {
374 .procname = "numa_balancing_scan_period_max_ms",
375 .data = &sysctl_numa_balancing_scan_period_max,
376 .maxlen = sizeof(unsigned int),
377 .mode = 0644,
378 .proc_handler = proc_dointvec,
379 },
380 {
381 .procname = "numa_balancing_scan_size_mb",
382 .data = &sysctl_numa_balancing_scan_size,
383 .maxlen = sizeof(unsigned int),
384 .mode = 0644,
385 .proc_handler = proc_dointvec,
386 },
387 {
388 .procname = "numa_balancing_settle_count",
389 .data = &sysctl_numa_balancing_settle_count,
390 .maxlen = sizeof(unsigned int),
391 .mode = 0644,
392 .proc_handler = proc_dointvec,
393 },
394 {
395 .procname = "numa_balancing_migrate_deferred",
396 .data = &sysctl_numa_balancing_migrate_deferred,
397 .maxlen = sizeof(unsigned int),
398 .mode = 0644,
399 .proc_handler = proc_dointvec,
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 #endif
501 #ifdef CONFIG_LATENCYTOP
502 {
503 .procname = "latencytop",
504 .data = &latencytop_enabled,
505 .maxlen = sizeof(int),
506 .mode = 0644,
507 .proc_handler = proc_dointvec,
508 },
509 #endif
510 #ifdef CONFIG_BLK_DEV_INITRD
511 {
512 .procname = "real-root-dev",
513 .data = &real_root_dev,
514 .maxlen = sizeof(int),
515 .mode = 0644,
516 .proc_handler = proc_dointvec,
517 },
518 #endif
519 {
520 .procname = "print-fatal-signals",
521 .data = &print_fatal_signals,
522 .maxlen = sizeof(int),
523 .mode = 0644,
524 .proc_handler = proc_dointvec,
525 },
526 #ifdef CONFIG_SPARC
527 {
528 .procname = "reboot-cmd",
529 .data = reboot_command,
530 .maxlen = 256,
531 .mode = 0644,
532 .proc_handler = proc_dostring,
533 },
534 {
535 .procname = "stop-a",
536 .data = &stop_a_enabled,
537 .maxlen = sizeof (int),
538 .mode = 0644,
539 .proc_handler = proc_dointvec,
540 },
541 {
542 .procname = "scons-poweroff",
543 .data = &scons_pwroff,
544 .maxlen = sizeof (int),
545 .mode = 0644,
546 .proc_handler = proc_dointvec,
547 },
548 #endif
549 #ifdef CONFIG_SPARC64
550 {
551 .procname = "tsb-ratio",
552 .data = &sysctl_tsb_ratio,
553 .maxlen = sizeof (int),
554 .mode = 0644,
555 .proc_handler = proc_dointvec,
556 },
557 #endif
558 #ifdef __hppa__
559 {
560 .procname = "soft-power",
561 .data = &pwrsw_enabled,
562 .maxlen = sizeof (int),
563 .mode = 0644,
564 .proc_handler = proc_dointvec,
565 },
566 #endif
567 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
568 {
569 .procname = "unaligned-trap",
570 .data = &unaligned_enabled,
571 .maxlen = sizeof (int),
572 .mode = 0644,
573 .proc_handler = proc_dointvec,
574 },
575 #endif
576 {
577 .procname = "ctrl-alt-del",
578 .data = &C_A_D,
579 .maxlen = sizeof(int),
580 .mode = 0644,
581 .proc_handler = proc_dointvec,
582 },
583 #ifdef CONFIG_FUNCTION_TRACER
584 {
585 .procname = "ftrace_enabled",
586 .data = &ftrace_enabled,
587 .maxlen = sizeof(int),
588 .mode = 0644,
589 .proc_handler = ftrace_enable_sysctl,
590 },
591 #endif
592 #ifdef CONFIG_STACK_TRACER
593 {
594 .procname = "stack_tracer_enabled",
595 .data = &stack_tracer_enabled,
596 .maxlen = sizeof(int),
597 .mode = 0644,
598 .proc_handler = stack_trace_sysctl,
599 },
600 #endif
601 #ifdef CONFIG_TRACING
602 {
603 .procname = "ftrace_dump_on_oops",
604 .data = &ftrace_dump_on_oops,
605 .maxlen = sizeof(int),
606 .mode = 0644,
607 .proc_handler = proc_dointvec,
608 },
609 {
610 .procname = "traceoff_on_warning",
611 .data = &__disable_trace_on_warning,
612 .maxlen = sizeof(__disable_trace_on_warning),
613 .mode = 0644,
614 .proc_handler = proc_dointvec,
615 },
616 #endif
617 #ifdef CONFIG_MODULES
618 {
619 .procname = "modprobe",
620 .data = &modprobe_path,
621 .maxlen = KMOD_PATH_LEN,
622 .mode = 0644,
623 .proc_handler = proc_dostring,
624 },
625 {
626 .procname = "modules_disabled",
627 .data = &modules_disabled,
628 .maxlen = sizeof(int),
629 .mode = 0644,
630 /* only handle a transition from default "0" to "1" */
631 .proc_handler = proc_dointvec_minmax,
632 .extra1 = &one,
633 .extra2 = &one,
634 },
635 #endif
636
637 {
638 .procname = "hotplug",
639 .data = &uevent_helper,
640 .maxlen = UEVENT_HELPER_PATH_LEN,
641 .mode = 0644,
642 .proc_handler = proc_dostring,
643 },
644
645 #ifdef CONFIG_CHR_DEV_SG
646 {
647 .procname = "sg-big-buff",
648 .data = &sg_big_buff,
649 .maxlen = sizeof (int),
650 .mode = 0444,
651 .proc_handler = proc_dointvec,
652 },
653 #endif
654 #ifdef CONFIG_BSD_PROCESS_ACCT
655 {
656 .procname = "acct",
657 .data = &acct_parm,
658 .maxlen = 3*sizeof(int),
659 .mode = 0644,
660 .proc_handler = proc_dointvec,
661 },
662 #endif
663 #ifdef CONFIG_MAGIC_SYSRQ
664 {
665 .procname = "sysrq",
666 .data = &__sysrq_enabled,
667 .maxlen = sizeof (int),
668 .mode = 0644,
669 .proc_handler = sysrq_sysctl_handler,
670 },
671 #endif
672 #ifdef CONFIG_PROC_SYSCTL
673 {
674 .procname = "cad_pid",
675 .data = NULL,
676 .maxlen = sizeof (int),
677 .mode = 0600,
678 .proc_handler = proc_do_cad_pid,
679 },
680 #endif
681 {
682 .procname = "threads-max",
683 .data = &max_threads,
684 .maxlen = sizeof(int),
685 .mode = 0644,
686 .proc_handler = proc_dointvec,
687 },
688 {
689 .procname = "random",
690 .mode = 0555,
691 .child = random_table,
692 },
693 {
694 .procname = "usermodehelper",
695 .mode = 0555,
696 .child = usermodehelper_table,
697 },
698 {
699 .procname = "overflowuid",
700 .data = &overflowuid,
701 .maxlen = sizeof(int),
702 .mode = 0644,
703 .proc_handler = proc_dointvec_minmax,
704 .extra1 = &minolduid,
705 .extra2 = &maxolduid,
706 },
707 {
708 .procname = "overflowgid",
709 .data = &overflowgid,
710 .maxlen = sizeof(int),
711 .mode = 0644,
712 .proc_handler = proc_dointvec_minmax,
713 .extra1 = &minolduid,
714 .extra2 = &maxolduid,
715 },
716 #ifdef CONFIG_S390
717 #ifdef CONFIG_MATHEMU
718 {
719 .procname = "ieee_emulation_warnings",
720 .data = &sysctl_ieee_emulation_warnings,
721 .maxlen = sizeof(int),
722 .mode = 0644,
723 .proc_handler = proc_dointvec,
724 },
725 #endif
726 {
727 .procname = "userprocess_debug",
728 .data = &show_unhandled_signals,
729 .maxlen = sizeof(int),
730 .mode = 0644,
731 .proc_handler = proc_dointvec,
732 },
733 #endif
734 {
735 .procname = "pid_max",
736 .data = &pid_max,
737 .maxlen = sizeof (int),
738 .mode = 0644,
739 .proc_handler = proc_dointvec_minmax,
740 .extra1 = &pid_max_min,
741 .extra2 = &pid_max_max,
742 },
743 {
744 .procname = "panic_on_oops",
745 .data = &panic_on_oops,
746 .maxlen = sizeof(int),
747 .mode = 0644,
748 .proc_handler = proc_dointvec,
749 },
750 #if defined CONFIG_PRINTK
751 {
752 .procname = "printk",
753 .data = &console_loglevel,
754 .maxlen = 4*sizeof(int),
755 .mode = 0644,
756 .proc_handler = proc_dointvec,
757 },
758 {
759 .procname = "printk_ratelimit",
760 .data = &printk_ratelimit_state.interval,
761 .maxlen = sizeof(int),
762 .mode = 0644,
763 .proc_handler = proc_dointvec_jiffies,
764 },
765 {
766 .procname = "printk_ratelimit_burst",
767 .data = &printk_ratelimit_state.burst,
768 .maxlen = sizeof(int),
769 .mode = 0644,
770 .proc_handler = proc_dointvec,
771 },
772 {
773 .procname = "printk_delay",
774 .data = &printk_delay_msec,
775 .maxlen = sizeof(int),
776 .mode = 0644,
777 .proc_handler = proc_dointvec_minmax,
778 .extra1 = &zero,
779 .extra2 = &ten_thousand,
780 },
781 {
782 .procname = "dmesg_restrict",
783 .data = &dmesg_restrict,
784 .maxlen = sizeof(int),
785 .mode = 0644,
786 .proc_handler = proc_dointvec_minmax_sysadmin,
787 .extra1 = &zero,
788 .extra2 = &one,
789 },
790 {
791 .procname = "kptr_restrict",
792 .data = &kptr_restrict,
793 .maxlen = sizeof(int),
794 .mode = 0644,
795 .proc_handler = proc_dointvec_minmax_sysadmin,
796 .extra1 = &zero,
797 .extra2 = &two,
798 },
799 #endif
800 {
801 .procname = "ngroups_max",
802 .data = &ngroups_max,
803 .maxlen = sizeof (int),
804 .mode = 0444,
805 .proc_handler = proc_dointvec,
806 },
807 {
808 .procname = "cap_last_cap",
809 .data = (void *)&cap_last_cap,
810 .maxlen = sizeof(int),
811 .mode = 0444,
812 .proc_handler = proc_dointvec,
813 },
814 #if defined(CONFIG_LOCKUP_DETECTOR)
815 {
816 .procname = "watchdog",
817 .data = &watchdog_user_enabled,
818 .maxlen = sizeof (int),
819 .mode = 0644,
820 .proc_handler = proc_dowatchdog,
821 .extra1 = &zero,
822 .extra2 = &one,
823 },
824 {
825 .procname = "watchdog_thresh",
826 .data = &watchdog_thresh,
827 .maxlen = sizeof(int),
828 .mode = 0644,
829 .proc_handler = proc_dowatchdog,
830 .extra1 = &zero,
831 .extra2 = &sixty,
832 },
833 {
834 .procname = "softlockup_panic",
835 .data = &softlockup_panic,
836 .maxlen = sizeof(int),
837 .mode = 0644,
838 .proc_handler = proc_dointvec_minmax,
839 .extra1 = &zero,
840 .extra2 = &one,
841 },
842 {
843 .procname = "nmi_watchdog",
844 .data = &watchdog_user_enabled,
845 .maxlen = sizeof (int),
846 .mode = 0644,
847 .proc_handler = proc_dowatchdog,
848 .extra1 = &zero,
849 .extra2 = &one,
850 },
851 #endif
852 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
853 {
854 .procname = "unknown_nmi_panic",
855 .data = &unknown_nmi_panic,
856 .maxlen = sizeof (int),
857 .mode = 0644,
858 .proc_handler = proc_dointvec,
859 },
860 #endif
861 #if defined(CONFIG_X86)
862 {
863 .procname = "panic_on_unrecovered_nmi",
864 .data = &panic_on_unrecovered_nmi,
865 .maxlen = sizeof(int),
866 .mode = 0644,
867 .proc_handler = proc_dointvec,
868 },
869 {
870 .procname = "panic_on_io_nmi",
871 .data = &panic_on_io_nmi,
872 .maxlen = sizeof(int),
873 .mode = 0644,
874 .proc_handler = proc_dointvec,
875 },
876 #ifdef CONFIG_DEBUG_STACKOVERFLOW
877 {
878 .procname = "panic_on_stackoverflow",
879 .data = &sysctl_panic_on_stackoverflow,
880 .maxlen = sizeof(int),
881 .mode = 0644,
882 .proc_handler = proc_dointvec,
883 },
884 #endif
885 {
886 .procname = "bootloader_type",
887 .data = &bootloader_type,
888 .maxlen = sizeof (int),
889 .mode = 0444,
890 .proc_handler = proc_dointvec,
891 },
892 {
893 .procname = "bootloader_version",
894 .data = &bootloader_version,
895 .maxlen = sizeof (int),
896 .mode = 0444,
897 .proc_handler = proc_dointvec,
898 },
899 {
900 .procname = "kstack_depth_to_print",
901 .data = &kstack_depth_to_print,
902 .maxlen = sizeof(int),
903 .mode = 0644,
904 .proc_handler = proc_dointvec,
905 },
906 {
907 .procname = "io_delay_type",
908 .data = &io_delay_type,
909 .maxlen = sizeof(int),
910 .mode = 0644,
911 .proc_handler = proc_dointvec,
912 },
913 #endif
914 #if defined(CONFIG_MMU)
915 {
916 .procname = "randomize_va_space",
917 .data = &randomize_va_space,
918 .maxlen = sizeof(int),
919 .mode = 0644,
920 .proc_handler = proc_dointvec,
921 },
922 #endif
923 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
924 {
925 .procname = "spin_retry",
926 .data = &spin_retry,
927 .maxlen = sizeof (int),
928 .mode = 0644,
929 .proc_handler = proc_dointvec,
930 },
931 #endif
932 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
933 {
934 .procname = "acpi_video_flags",
935 .data = &acpi_realmode_flags,
936 .maxlen = sizeof (unsigned long),
937 .mode = 0644,
938 .proc_handler = proc_doulongvec_minmax,
939 },
940 #endif
941 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
942 {
943 .procname = "ignore-unaligned-usertrap",
944 .data = &no_unaligned_warning,
945 .maxlen = sizeof (int),
946 .mode = 0644,
947 .proc_handler = proc_dointvec,
948 },
949 #endif
950 #ifdef CONFIG_IA64
951 {
952 .procname = "unaligned-dump-stack",
953 .data = &unaligned_dump_stack,
954 .maxlen = sizeof (int),
955 .mode = 0644,
956 .proc_handler = proc_dointvec,
957 },
958 #endif
959 #ifdef CONFIG_DETECT_HUNG_TASK
960 {
961 .procname = "hung_task_panic",
962 .data = &sysctl_hung_task_panic,
963 .maxlen = sizeof(int),
964 .mode = 0644,
965 .proc_handler = proc_dointvec_minmax,
966 .extra1 = &zero,
967 .extra2 = &one,
968 },
969 {
970 .procname = "hung_task_check_count",
971 .data = &sysctl_hung_task_check_count,
972 .maxlen = sizeof(unsigned long),
973 .mode = 0644,
974 .proc_handler = proc_doulongvec_minmax,
975 },
976 {
977 .procname = "hung_task_timeout_secs",
978 .data = &sysctl_hung_task_timeout_secs,
979 .maxlen = sizeof(unsigned long),
980 .mode = 0644,
981 .proc_handler = proc_dohung_task_timeout_secs,
982 },
983 {
984 .procname = "hung_task_warnings",
985 .data = &sysctl_hung_task_warnings,
986 .maxlen = sizeof(unsigned long),
987 .mode = 0644,
988 .proc_handler = proc_doulongvec_minmax,
989 },
990 #endif
991 #ifdef CONFIG_COMPAT
992 {
993 .procname = "compat-log",
994 .data = &compat_log,
995 .maxlen = sizeof (int),
996 .mode = 0644,
997 .proc_handler = proc_dointvec,
998 },
999 #endif
1000 #ifdef CONFIG_RT_MUTEXES
1001 {
1002 .procname = "max_lock_depth",
1003 .data = &max_lock_depth,
1004 .maxlen = sizeof(int),
1005 .mode = 0644,
1006 .proc_handler = proc_dointvec,
1007 },
1008 #endif
1009 {
1010 .procname = "poweroff_cmd",
1011 .data = &poweroff_cmd,
1012 .maxlen = POWEROFF_CMD_PATH_LEN,
1013 .mode = 0644,
1014 .proc_handler = proc_dostring,
1015 },
1016 #ifdef CONFIG_KEYS
1017 {
1018 .procname = "keys",
1019 .mode = 0555,
1020 .child = key_sysctls,
1021 },
1022 #endif
1023 #ifdef CONFIG_RCU_TORTURE_TEST
1024 {
1025 .procname = "rcutorture_runnable",
1026 .data = &rcutorture_runnable,
1027 .maxlen = sizeof(int),
1028 .mode = 0644,
1029 .proc_handler = proc_dointvec,
1030 },
1031 #endif
1032 #ifdef CONFIG_PERF_EVENTS
1033 /*
1034 * User-space scripts rely on the existence of this file
1035 * as a feature check for perf_events being enabled.
1036 *
1037 * So it's an ABI, do not remove!
1038 */
1039 {
1040 .procname = "perf_event_paranoid",
1041 .data = &sysctl_perf_event_paranoid,
1042 .maxlen = sizeof(sysctl_perf_event_paranoid),
1043 .mode = 0644,
1044 .proc_handler = proc_dointvec,
1045 },
1046 {
1047 .procname = "perf_event_mlock_kb",
1048 .data = &sysctl_perf_event_mlock,
1049 .maxlen = sizeof(sysctl_perf_event_mlock),
1050 .mode = 0644,
1051 .proc_handler = proc_dointvec,
1052 },
1053 {
1054 .procname = "perf_event_max_sample_rate",
1055 .data = &sysctl_perf_event_sample_rate,
1056 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1057 .mode = 0644,
1058 .proc_handler = perf_proc_update_handler,
1059 .extra1 = &one,
1060 },
1061 {
1062 .procname = "perf_cpu_time_max_percent",
1063 .data = &sysctl_perf_cpu_time_max_percent,
1064 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1065 .mode = 0644,
1066 .proc_handler = perf_cpu_time_max_percent_handler,
1067 .extra1 = &zero,
1068 .extra2 = &one_hundred,
1069 },
1070 #endif
1071 #ifdef CONFIG_KMEMCHECK
1072 {
1073 .procname = "kmemcheck",
1074 .data = &kmemcheck_enabled,
1075 .maxlen = sizeof(int),
1076 .mode = 0644,
1077 .proc_handler = proc_dointvec,
1078 },
1079 #endif
1080 #ifdef CONFIG_BLOCK
1081 {
1082 .procname = "blk_iopoll",
1083 .data = &blk_iopoll_enabled,
1084 .maxlen = sizeof(int),
1085 .mode = 0644,
1086 .proc_handler = proc_dointvec,
1087 },
1088 #endif
1089 { }
1090 };
1091
1092 static struct ctl_table vm_table[] = {
1093 {
1094 .procname = "overcommit_memory",
1095 .data = &sysctl_overcommit_memory,
1096 .maxlen = sizeof(sysctl_overcommit_memory),
1097 .mode = 0644,
1098 .proc_handler = proc_dointvec_minmax,
1099 .extra1 = &zero,
1100 .extra2 = &two,
1101 },
1102 {
1103 .procname = "panic_on_oom",
1104 .data = &sysctl_panic_on_oom,
1105 .maxlen = sizeof(sysctl_panic_on_oom),
1106 .mode = 0644,
1107 .proc_handler = proc_dointvec_minmax,
1108 .extra1 = &zero,
1109 .extra2 = &two,
1110 },
1111 {
1112 .procname = "oom_kill_allocating_task",
1113 .data = &sysctl_oom_kill_allocating_task,
1114 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1115 .mode = 0644,
1116 .proc_handler = proc_dointvec,
1117 },
1118 {
1119 .procname = "oom_dump_tasks",
1120 .data = &sysctl_oom_dump_tasks,
1121 .maxlen = sizeof(sysctl_oom_dump_tasks),
1122 .mode = 0644,
1123 .proc_handler = proc_dointvec,
1124 },
1125 {
1126 .procname = "overcommit_ratio",
1127 .data = &sysctl_overcommit_ratio,
1128 .maxlen = sizeof(sysctl_overcommit_ratio),
1129 .mode = 0644,
1130 .proc_handler = proc_dointvec,
1131 },
1132 {
1133 .procname = "page-cluster",
1134 .data = &page_cluster,
1135 .maxlen = sizeof(int),
1136 .mode = 0644,
1137 .proc_handler = proc_dointvec_minmax,
1138 .extra1 = &zero,
1139 },
1140 {
1141 .procname = "dirty_background_ratio",
1142 .data = &dirty_background_ratio,
1143 .maxlen = sizeof(dirty_background_ratio),
1144 .mode = 0644,
1145 .proc_handler = dirty_background_ratio_handler,
1146 .extra1 = &zero,
1147 .extra2 = &one_hundred,
1148 },
1149 {
1150 .procname = "dirty_background_bytes",
1151 .data = &dirty_background_bytes,
1152 .maxlen = sizeof(dirty_background_bytes),
1153 .mode = 0644,
1154 .proc_handler = dirty_background_bytes_handler,
1155 .extra1 = &one_ul,
1156 },
1157 {
1158 .procname = "dirty_ratio",
1159 .data = &vm_dirty_ratio,
1160 .maxlen = sizeof(vm_dirty_ratio),
1161 .mode = 0644,
1162 .proc_handler = dirty_ratio_handler,
1163 .extra1 = &zero,
1164 .extra2 = &one_hundred,
1165 },
1166 {
1167 .procname = "dirty_bytes",
1168 .data = &vm_dirty_bytes,
1169 .maxlen = sizeof(vm_dirty_bytes),
1170 .mode = 0644,
1171 .proc_handler = dirty_bytes_handler,
1172 .extra1 = &dirty_bytes_min,
1173 },
1174 {
1175 .procname = "dirty_writeback_centisecs",
1176 .data = &dirty_writeback_interval,
1177 .maxlen = sizeof(dirty_writeback_interval),
1178 .mode = 0644,
1179 .proc_handler = dirty_writeback_centisecs_handler,
1180 },
1181 {
1182 .procname = "dirty_expire_centisecs",
1183 .data = &dirty_expire_interval,
1184 .maxlen = sizeof(dirty_expire_interval),
1185 .mode = 0644,
1186 .proc_handler = proc_dointvec_minmax,
1187 .extra1 = &zero,
1188 },
1189 {
1190 .procname = "nr_pdflush_threads",
1191 .mode = 0444 /* read-only */,
1192 .proc_handler = pdflush_proc_obsolete,
1193 },
1194 {
1195 .procname = "swappiness",
1196 .data = &vm_swappiness,
1197 .maxlen = sizeof(vm_swappiness),
1198 .mode = 0644,
1199 .proc_handler = proc_dointvec_minmax,
1200 .extra1 = &zero,
1201 .extra2 = &one_hundred,
1202 },
1203 #ifdef CONFIG_HUGETLB_PAGE
1204 {
1205 .procname = "nr_hugepages",
1206 .data = NULL,
1207 .maxlen = sizeof(unsigned long),
1208 .mode = 0644,
1209 .proc_handler = hugetlb_sysctl_handler,
1210 .extra1 = (void *)&hugetlb_zero,
1211 .extra2 = (void *)&hugetlb_infinity,
1212 },
1213 #ifdef CONFIG_NUMA
1214 {
1215 .procname = "nr_hugepages_mempolicy",
1216 .data = NULL,
1217 .maxlen = sizeof(unsigned long),
1218 .mode = 0644,
1219 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1220 .extra1 = (void *)&hugetlb_zero,
1221 .extra2 = (void *)&hugetlb_infinity,
1222 },
1223 #endif
1224 {
1225 .procname = "hugetlb_shm_group",
1226 .data = &sysctl_hugetlb_shm_group,
1227 .maxlen = sizeof(gid_t),
1228 .mode = 0644,
1229 .proc_handler = proc_dointvec,
1230 },
1231 {
1232 .procname = "hugepages_treat_as_movable",
1233 .data = &hugepages_treat_as_movable,
1234 .maxlen = sizeof(int),
1235 .mode = 0644,
1236 .proc_handler = proc_dointvec,
1237 },
1238 {
1239 .procname = "nr_overcommit_hugepages",
1240 .data = NULL,
1241 .maxlen = sizeof(unsigned long),
1242 .mode = 0644,
1243 .proc_handler = hugetlb_overcommit_handler,
1244 .extra1 = (void *)&hugetlb_zero,
1245 .extra2 = (void *)&hugetlb_infinity,
1246 },
1247 #endif
1248 {
1249 .procname = "lowmem_reserve_ratio",
1250 .data = &sysctl_lowmem_reserve_ratio,
1251 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1252 .mode = 0644,
1253 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1254 },
1255 {
1256 .procname = "drop_caches",
1257 .data = &sysctl_drop_caches,
1258 .maxlen = sizeof(int),
1259 .mode = 0644,
1260 .proc_handler = drop_caches_sysctl_handler,
1261 .extra1 = &one,
1262 .extra2 = &three,
1263 },
1264 #ifdef CONFIG_COMPACTION
1265 {
1266 .procname = "compact_memory",
1267 .data = &sysctl_compact_memory,
1268 .maxlen = sizeof(int),
1269 .mode = 0200,
1270 .proc_handler = sysctl_compaction_handler,
1271 },
1272 {
1273 .procname = "extfrag_threshold",
1274 .data = &sysctl_extfrag_threshold,
1275 .maxlen = sizeof(int),
1276 .mode = 0644,
1277 .proc_handler = sysctl_extfrag_handler,
1278 .extra1 = &min_extfrag_threshold,
1279 .extra2 = &max_extfrag_threshold,
1280 },
1281
1282 #endif /* CONFIG_COMPACTION */
1283 {
1284 .procname = "min_free_kbytes",
1285 .data = &min_free_kbytes,
1286 .maxlen = sizeof(min_free_kbytes),
1287 .mode = 0644,
1288 .proc_handler = min_free_kbytes_sysctl_handler,
1289 .extra1 = &zero,
1290 },
1291 {
1292 .procname = "percpu_pagelist_fraction",
1293 .data = &percpu_pagelist_fraction,
1294 .maxlen = sizeof(percpu_pagelist_fraction),
1295 .mode = 0644,
1296 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1297 .extra1 = &min_percpu_pagelist_fract,
1298 },
1299 #ifdef CONFIG_MMU
1300 {
1301 .procname = "max_map_count",
1302 .data = &sysctl_max_map_count,
1303 .maxlen = sizeof(sysctl_max_map_count),
1304 .mode = 0644,
1305 .proc_handler = proc_dointvec_minmax,
1306 .extra1 = &zero,
1307 },
1308 #else
1309 {
1310 .procname = "nr_trim_pages",
1311 .data = &sysctl_nr_trim_pages,
1312 .maxlen = sizeof(sysctl_nr_trim_pages),
1313 .mode = 0644,
1314 .proc_handler = proc_dointvec_minmax,
1315 .extra1 = &zero,
1316 },
1317 #endif
1318 {
1319 .procname = "laptop_mode",
1320 .data = &laptop_mode,
1321 .maxlen = sizeof(laptop_mode),
1322 .mode = 0644,
1323 .proc_handler = proc_dointvec_jiffies,
1324 },
1325 {
1326 .procname = "block_dump",
1327 .data = &block_dump,
1328 .maxlen = sizeof(block_dump),
1329 .mode = 0644,
1330 .proc_handler = proc_dointvec,
1331 .extra1 = &zero,
1332 },
1333 {
1334 .procname = "vfs_cache_pressure",
1335 .data = &sysctl_vfs_cache_pressure,
1336 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1337 .mode = 0644,
1338 .proc_handler = proc_dointvec,
1339 .extra1 = &zero,
1340 },
1341 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1342 {
1343 .procname = "legacy_va_layout",
1344 .data = &sysctl_legacy_va_layout,
1345 .maxlen = sizeof(sysctl_legacy_va_layout),
1346 .mode = 0644,
1347 .proc_handler = proc_dointvec,
1348 .extra1 = &zero,
1349 },
1350 #endif
1351 #ifdef CONFIG_NUMA
1352 {
1353 .procname = "zone_reclaim_mode",
1354 .data = &zone_reclaim_mode,
1355 .maxlen = sizeof(zone_reclaim_mode),
1356 .mode = 0644,
1357 .proc_handler = proc_dointvec,
1358 .extra1 = &zero,
1359 },
1360 {
1361 .procname = "min_unmapped_ratio",
1362 .data = &sysctl_min_unmapped_ratio,
1363 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1364 .mode = 0644,
1365 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1366 .extra1 = &zero,
1367 .extra2 = &one_hundred,
1368 },
1369 {
1370 .procname = "min_slab_ratio",
1371 .data = &sysctl_min_slab_ratio,
1372 .maxlen = sizeof(sysctl_min_slab_ratio),
1373 .mode = 0644,
1374 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1375 .extra1 = &zero,
1376 .extra2 = &one_hundred,
1377 },
1378 #endif
1379 #ifdef CONFIG_SMP
1380 {
1381 .procname = "stat_interval",
1382 .data = &sysctl_stat_interval,
1383 .maxlen = sizeof(sysctl_stat_interval),
1384 .mode = 0644,
1385 .proc_handler = proc_dointvec_jiffies,
1386 },
1387 #endif
1388 #ifdef CONFIG_MMU
1389 {
1390 .procname = "mmap_min_addr",
1391 .data = &dac_mmap_min_addr,
1392 .maxlen = sizeof(unsigned long),
1393 .mode = 0644,
1394 .proc_handler = mmap_min_addr_handler,
1395 },
1396 #endif
1397 #ifdef CONFIG_NUMA
1398 {
1399 .procname = "numa_zonelist_order",
1400 .data = &numa_zonelist_order,
1401 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1402 .mode = 0644,
1403 .proc_handler = numa_zonelist_order_handler,
1404 },
1405 #endif
1406 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1407 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1408 {
1409 .procname = "vdso_enabled",
1410 .data = &vdso_enabled,
1411 .maxlen = sizeof(vdso_enabled),
1412 .mode = 0644,
1413 .proc_handler = proc_dointvec,
1414 .extra1 = &zero,
1415 },
1416 #endif
1417 #ifdef CONFIG_HIGHMEM
1418 {
1419 .procname = "highmem_is_dirtyable",
1420 .data = &vm_highmem_is_dirtyable,
1421 .maxlen = sizeof(vm_highmem_is_dirtyable),
1422 .mode = 0644,
1423 .proc_handler = proc_dointvec_minmax,
1424 .extra1 = &zero,
1425 .extra2 = &one,
1426 },
1427 #endif
1428 {
1429 .procname = "scan_unevictable_pages",
1430 .data = &scan_unevictable_pages,
1431 .maxlen = sizeof(scan_unevictable_pages),
1432 .mode = 0644,
1433 .proc_handler = scan_unevictable_handler,
1434 },
1435 #ifdef CONFIG_MEMORY_FAILURE
1436 {
1437 .procname = "memory_failure_early_kill",
1438 .data = &sysctl_memory_failure_early_kill,
1439 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1440 .mode = 0644,
1441 .proc_handler = proc_dointvec_minmax,
1442 .extra1 = &zero,
1443 .extra2 = &one,
1444 },
1445 {
1446 .procname = "memory_failure_recovery",
1447 .data = &sysctl_memory_failure_recovery,
1448 .maxlen = sizeof(sysctl_memory_failure_recovery),
1449 .mode = 0644,
1450 .proc_handler = proc_dointvec_minmax,
1451 .extra1 = &zero,
1452 .extra2 = &one,
1453 },
1454 #endif
1455 {
1456 .procname = "user_reserve_kbytes",
1457 .data = &sysctl_user_reserve_kbytes,
1458 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1459 .mode = 0644,
1460 .proc_handler = proc_doulongvec_minmax,
1461 },
1462 {
1463 .procname = "admin_reserve_kbytes",
1464 .data = &sysctl_admin_reserve_kbytes,
1465 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1466 .mode = 0644,
1467 .proc_handler = proc_doulongvec_minmax,
1468 },
1469 { }
1470 };
1471
1472 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1473 static struct ctl_table binfmt_misc_table[] = {
1474 { }
1475 };
1476 #endif
1477
1478 static struct ctl_table fs_table[] = {
1479 {
1480 .procname = "inode-nr",
1481 .data = &inodes_stat,
1482 .maxlen = 2*sizeof(long),
1483 .mode = 0444,
1484 .proc_handler = proc_nr_inodes,
1485 },
1486 {
1487 .procname = "inode-state",
1488 .data = &inodes_stat,
1489 .maxlen = 7*sizeof(long),
1490 .mode = 0444,
1491 .proc_handler = proc_nr_inodes,
1492 },
1493 {
1494 .procname = "file-nr",
1495 .data = &files_stat,
1496 .maxlen = sizeof(files_stat),
1497 .mode = 0444,
1498 .proc_handler = proc_nr_files,
1499 },
1500 {
1501 .procname = "file-max",
1502 .data = &files_stat.max_files,
1503 .maxlen = sizeof(files_stat.max_files),
1504 .mode = 0644,
1505 .proc_handler = proc_doulongvec_minmax,
1506 },
1507 {
1508 .procname = "nr_open",
1509 .data = &sysctl_nr_open,
1510 .maxlen = sizeof(int),
1511 .mode = 0644,
1512 .proc_handler = proc_dointvec_minmax,
1513 .extra1 = &sysctl_nr_open_min,
1514 .extra2 = &sysctl_nr_open_max,
1515 },
1516 {
1517 .procname = "dentry-state",
1518 .data = &dentry_stat,
1519 .maxlen = 6*sizeof(long),
1520 .mode = 0444,
1521 .proc_handler = proc_nr_dentry,
1522 },
1523 {
1524 .procname = "overflowuid",
1525 .data = &fs_overflowuid,
1526 .maxlen = sizeof(int),
1527 .mode = 0644,
1528 .proc_handler = proc_dointvec_minmax,
1529 .extra1 = &minolduid,
1530 .extra2 = &maxolduid,
1531 },
1532 {
1533 .procname = "overflowgid",
1534 .data = &fs_overflowgid,
1535 .maxlen = sizeof(int),
1536 .mode = 0644,
1537 .proc_handler = proc_dointvec_minmax,
1538 .extra1 = &minolduid,
1539 .extra2 = &maxolduid,
1540 },
1541 #ifdef CONFIG_FILE_LOCKING
1542 {
1543 .procname = "leases-enable",
1544 .data = &leases_enable,
1545 .maxlen = sizeof(int),
1546 .mode = 0644,
1547 .proc_handler = proc_dointvec,
1548 },
1549 #endif
1550 #ifdef CONFIG_DNOTIFY
1551 {
1552 .procname = "dir-notify-enable",
1553 .data = &dir_notify_enable,
1554 .maxlen = sizeof(int),
1555 .mode = 0644,
1556 .proc_handler = proc_dointvec,
1557 },
1558 #endif
1559 #ifdef CONFIG_MMU
1560 #ifdef CONFIG_FILE_LOCKING
1561 {
1562 .procname = "lease-break-time",
1563 .data = &lease_break_time,
1564 .maxlen = sizeof(int),
1565 .mode = 0644,
1566 .proc_handler = proc_dointvec,
1567 },
1568 #endif
1569 #ifdef CONFIG_AIO
1570 {
1571 .procname = "aio-nr",
1572 .data = &aio_nr,
1573 .maxlen = sizeof(aio_nr),
1574 .mode = 0444,
1575 .proc_handler = proc_doulongvec_minmax,
1576 },
1577 {
1578 .procname = "aio-max-nr",
1579 .data = &aio_max_nr,
1580 .maxlen = sizeof(aio_max_nr),
1581 .mode = 0644,
1582 .proc_handler = proc_doulongvec_minmax,
1583 },
1584 #endif /* CONFIG_AIO */
1585 #ifdef CONFIG_INOTIFY_USER
1586 {
1587 .procname = "inotify",
1588 .mode = 0555,
1589 .child = inotify_table,
1590 },
1591 #endif
1592 #ifdef CONFIG_EPOLL
1593 {
1594 .procname = "epoll",
1595 .mode = 0555,
1596 .child = epoll_table,
1597 },
1598 #endif
1599 #endif
1600 {
1601 .procname = "protected_symlinks",
1602 .data = &sysctl_protected_symlinks,
1603 .maxlen = sizeof(int),
1604 .mode = 0600,
1605 .proc_handler = proc_dointvec_minmax,
1606 .extra1 = &zero,
1607 .extra2 = &one,
1608 },
1609 {
1610 .procname = "protected_hardlinks",
1611 .data = &sysctl_protected_hardlinks,
1612 .maxlen = sizeof(int),
1613 .mode = 0600,
1614 .proc_handler = proc_dointvec_minmax,
1615 .extra1 = &zero,
1616 .extra2 = &one,
1617 },
1618 {
1619 .procname = "suid_dumpable",
1620 .data = &suid_dumpable,
1621 .maxlen = sizeof(int),
1622 .mode = 0644,
1623 .proc_handler = proc_dointvec_minmax_coredump,
1624 .extra1 = &zero,
1625 .extra2 = &two,
1626 },
1627 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1628 {
1629 .procname = "binfmt_misc",
1630 .mode = 0555,
1631 .child = binfmt_misc_table,
1632 },
1633 #endif
1634 {
1635 .procname = "pipe-max-size",
1636 .data = &pipe_max_size,
1637 .maxlen = sizeof(int),
1638 .mode = 0644,
1639 .proc_handler = &pipe_proc_fn,
1640 .extra1 = &pipe_min_size,
1641 },
1642 { }
1643 };
1644
1645 static struct ctl_table debug_table[] = {
1646 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1647 {
1648 .procname = "exception-trace",
1649 .data = &show_unhandled_signals,
1650 .maxlen = sizeof(int),
1651 .mode = 0644,
1652 .proc_handler = proc_dointvec
1653 },
1654 #endif
1655 #if defined(CONFIG_OPTPROBES)
1656 {
1657 .procname = "kprobes-optimization",
1658 .data = &sysctl_kprobes_optimization,
1659 .maxlen = sizeof(int),
1660 .mode = 0644,
1661 .proc_handler = proc_kprobes_optimization_handler,
1662 .extra1 = &zero,
1663 .extra2 = &one,
1664 },
1665 #endif
1666 { }
1667 };
1668
1669 static struct ctl_table dev_table[] = {
1670 { }
1671 };
1672
1673 int __init sysctl_init(void)
1674 {
1675 struct ctl_table_header *hdr;
1676
1677 hdr = register_sysctl_table(sysctl_base_table);
1678 kmemleak_not_leak(hdr);
1679 return 0;
1680 }
1681
1682 #endif /* CONFIG_SYSCTL */
1683
1684 /*
1685 * /proc/sys support
1686 */
1687
1688 #ifdef CONFIG_PROC_SYSCTL
1689
1690 static int _proc_do_string(void* data, int maxlen, int write,
1691 void __user *buffer,
1692 size_t *lenp, loff_t *ppos)
1693 {
1694 size_t len;
1695 char __user *p;
1696 char c;
1697
1698 if (!data || !maxlen || !*lenp) {
1699 *lenp = 0;
1700 return 0;
1701 }
1702
1703 if (write) {
1704 len = 0;
1705 p = buffer;
1706 while (len < *lenp) {
1707 if (get_user(c, p++))
1708 return -EFAULT;
1709 if (c == 0 || c == '\n')
1710 break;
1711 len++;
1712 }
1713 if (len >= maxlen)
1714 len = maxlen-1;
1715 if(copy_from_user(data, buffer, len))
1716 return -EFAULT;
1717 ((char *) data)[len] = 0;
1718 *ppos += *lenp;
1719 } else {
1720 len = strlen(data);
1721 if (len > maxlen)
1722 len = maxlen;
1723
1724 if (*ppos > len) {
1725 *lenp = 0;
1726 return 0;
1727 }
1728
1729 data += *ppos;
1730 len -= *ppos;
1731
1732 if (len > *lenp)
1733 len = *lenp;
1734 if (len)
1735 if(copy_to_user(buffer, data, len))
1736 return -EFAULT;
1737 if (len < *lenp) {
1738 if(put_user('\n', ((char __user *) buffer) + len))
1739 return -EFAULT;
1740 len++;
1741 }
1742 *lenp = len;
1743 *ppos += len;
1744 }
1745 return 0;
1746 }
1747
1748 /**
1749 * proc_dostring - read a string sysctl
1750 * @table: the sysctl table
1751 * @write: %TRUE if this is a write to the sysctl file
1752 * @buffer: the user buffer
1753 * @lenp: the size of the user buffer
1754 * @ppos: file position
1755 *
1756 * Reads/writes a string from/to the user buffer. If the kernel
1757 * buffer provided is not large enough to hold the string, the
1758 * string is truncated. The copied string is %NULL-terminated.
1759 * If the string is being read by the user process, it is copied
1760 * and a newline '\n' is added. It is truncated if the buffer is
1761 * not large enough.
1762 *
1763 * Returns 0 on success.
1764 */
1765 int proc_dostring(struct ctl_table *table, int write,
1766 void __user *buffer, size_t *lenp, loff_t *ppos)
1767 {
1768 return _proc_do_string(table->data, table->maxlen, write,
1769 buffer, lenp, ppos);
1770 }
1771
1772 static size_t proc_skip_spaces(char **buf)
1773 {
1774 size_t ret;
1775 char *tmp = skip_spaces(*buf);
1776 ret = tmp - *buf;
1777 *buf = tmp;
1778 return ret;
1779 }
1780
1781 static void proc_skip_char(char **buf, size_t *size, const char v)
1782 {
1783 while (*size) {
1784 if (**buf != v)
1785 break;
1786 (*size)--;
1787 (*buf)++;
1788 }
1789 }
1790
1791 #define TMPBUFLEN 22
1792 /**
1793 * proc_get_long - reads an ASCII formatted integer from a user buffer
1794 *
1795 * @buf: a kernel buffer
1796 * @size: size of the kernel buffer
1797 * @val: this is where the number will be stored
1798 * @neg: set to %TRUE if number is negative
1799 * @perm_tr: a vector which contains the allowed trailers
1800 * @perm_tr_len: size of the perm_tr vector
1801 * @tr: pointer to store the trailer character
1802 *
1803 * In case of success %0 is returned and @buf and @size are updated with
1804 * the amount of bytes read. If @tr is non-NULL and a trailing
1805 * character exists (size is non-zero after returning from this
1806 * function), @tr is updated with the trailing character.
1807 */
1808 static int proc_get_long(char **buf, size_t *size,
1809 unsigned long *val, bool *neg,
1810 const char *perm_tr, unsigned perm_tr_len, char *tr)
1811 {
1812 int len;
1813 char *p, tmp[TMPBUFLEN];
1814
1815 if (!*size)
1816 return -EINVAL;
1817
1818 len = *size;
1819 if (len > TMPBUFLEN - 1)
1820 len = TMPBUFLEN - 1;
1821
1822 memcpy(tmp, *buf, len);
1823
1824 tmp[len] = 0;
1825 p = tmp;
1826 if (*p == '-' && *size > 1) {
1827 *neg = true;
1828 p++;
1829 } else
1830 *neg = false;
1831 if (!isdigit(*p))
1832 return -EINVAL;
1833
1834 *val = simple_strtoul(p, &p, 0);
1835
1836 len = p - tmp;
1837
1838 /* We don't know if the next char is whitespace thus we may accept
1839 * invalid integers (e.g. 1234...a) or two integers instead of one
1840 * (e.g. 123...1). So lets not allow such large numbers. */
1841 if (len == TMPBUFLEN - 1)
1842 return -EINVAL;
1843
1844 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
1845 return -EINVAL;
1846
1847 if (tr && (len < *size))
1848 *tr = *p;
1849
1850 *buf += len;
1851 *size -= len;
1852
1853 return 0;
1854 }
1855
1856 /**
1857 * proc_put_long - converts an integer to a decimal ASCII formatted string
1858 *
1859 * @buf: the user buffer
1860 * @size: the size of the user buffer
1861 * @val: the integer to be converted
1862 * @neg: sign of the number, %TRUE for negative
1863 *
1864 * In case of success %0 is returned and @buf and @size are updated with
1865 * the amount of bytes written.
1866 */
1867 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
1868 bool neg)
1869 {
1870 int len;
1871 char tmp[TMPBUFLEN], *p = tmp;
1872
1873 sprintf(p, "%s%lu", neg ? "-" : "", val);
1874 len = strlen(tmp);
1875 if (len > *size)
1876 len = *size;
1877 if (copy_to_user(*buf, tmp, len))
1878 return -EFAULT;
1879 *size -= len;
1880 *buf += len;
1881 return 0;
1882 }
1883 #undef TMPBUFLEN
1884
1885 static int proc_put_char(void __user **buf, size_t *size, char c)
1886 {
1887 if (*size) {
1888 char __user **buffer = (char __user **)buf;
1889 if (put_user(c, *buffer))
1890 return -EFAULT;
1891 (*size)--, (*buffer)++;
1892 *buf = *buffer;
1893 }
1894 return 0;
1895 }
1896
1897 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
1898 int *valp,
1899 int write, void *data)
1900 {
1901 if (write) {
1902 *valp = *negp ? -*lvalp : *lvalp;
1903 } else {
1904 int val = *valp;
1905 if (val < 0) {
1906 *negp = true;
1907 *lvalp = (unsigned long)-val;
1908 } else {
1909 *negp = false;
1910 *lvalp = (unsigned long)val;
1911 }
1912 }
1913 return 0;
1914 }
1915
1916 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
1917
1918 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1919 int write, void __user *buffer,
1920 size_t *lenp, loff_t *ppos,
1921 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1922 int write, void *data),
1923 void *data)
1924 {
1925 int *i, vleft, first = 1, err = 0;
1926 unsigned long page = 0;
1927 size_t left;
1928 char *kbuf;
1929
1930 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
1931 *lenp = 0;
1932 return 0;
1933 }
1934
1935 i = (int *) tbl_data;
1936 vleft = table->maxlen / sizeof(*i);
1937 left = *lenp;
1938
1939 if (!conv)
1940 conv = do_proc_dointvec_conv;
1941
1942 if (write) {
1943 if (left > PAGE_SIZE - 1)
1944 left = PAGE_SIZE - 1;
1945 page = __get_free_page(GFP_TEMPORARY);
1946 kbuf = (char *) page;
1947 if (!kbuf)
1948 return -ENOMEM;
1949 if (copy_from_user(kbuf, buffer, left)) {
1950 err = -EFAULT;
1951 goto free;
1952 }
1953 kbuf[left] = 0;
1954 }
1955
1956 for (; left && vleft--; i++, first=0) {
1957 unsigned long lval;
1958 bool neg;
1959
1960 if (write) {
1961 left -= proc_skip_spaces(&kbuf);
1962
1963 if (!left)
1964 break;
1965 err = proc_get_long(&kbuf, &left, &lval, &neg,
1966 proc_wspace_sep,
1967 sizeof(proc_wspace_sep), NULL);
1968 if (err)
1969 break;
1970 if (conv(&neg, &lval, i, 1, data)) {
1971 err = -EINVAL;
1972 break;
1973 }
1974 } else {
1975 if (conv(&neg, &lval, i, 0, data)) {
1976 err = -EINVAL;
1977 break;
1978 }
1979 if (!first)
1980 err = proc_put_char(&buffer, &left, '\t');
1981 if (err)
1982 break;
1983 err = proc_put_long(&buffer, &left, lval, neg);
1984 if (err)
1985 break;
1986 }
1987 }
1988
1989 if (!write && !first && left && !err)
1990 err = proc_put_char(&buffer, &left, '\n');
1991 if (write && !err && left)
1992 left -= proc_skip_spaces(&kbuf);
1993 free:
1994 if (write) {
1995 free_page(page);
1996 if (first)
1997 return err ? : -EINVAL;
1998 }
1999 *lenp -= left;
2000 *ppos += *lenp;
2001 return err;
2002 }
2003
2004 static int do_proc_dointvec(struct ctl_table *table, int write,
2005 void __user *buffer, size_t *lenp, loff_t *ppos,
2006 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2007 int write, void *data),
2008 void *data)
2009 {
2010 return __do_proc_dointvec(table->data, table, write,
2011 buffer, lenp, ppos, conv, data);
2012 }
2013
2014 /**
2015 * proc_dointvec - read a vector of integers
2016 * @table: the sysctl table
2017 * @write: %TRUE if this is a write to the sysctl file
2018 * @buffer: the user buffer
2019 * @lenp: the size of the user buffer
2020 * @ppos: file position
2021 *
2022 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2023 * values from/to the user buffer, treated as an ASCII string.
2024 *
2025 * Returns 0 on success.
2026 */
2027 int proc_dointvec(struct ctl_table *table, int write,
2028 void __user *buffer, size_t *lenp, loff_t *ppos)
2029 {
2030 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2031 NULL,NULL);
2032 }
2033
2034 /*
2035 * Taint values can only be increased
2036 * This means we can safely use a temporary.
2037 */
2038 static int proc_taint(struct ctl_table *table, int write,
2039 void __user *buffer, size_t *lenp, loff_t *ppos)
2040 {
2041 struct ctl_table t;
2042 unsigned long tmptaint = get_taint();
2043 int err;
2044
2045 if (write && !capable(CAP_SYS_ADMIN))
2046 return -EPERM;
2047
2048 t = *table;
2049 t.data = &tmptaint;
2050 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2051 if (err < 0)
2052 return err;
2053
2054 if (write) {
2055 /*
2056 * Poor man's atomic or. Not worth adding a primitive
2057 * to everyone's atomic.h for this
2058 */
2059 int i;
2060 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2061 if ((tmptaint >> i) & 1)
2062 add_taint(i, LOCKDEP_STILL_OK);
2063 }
2064 }
2065
2066 return err;
2067 }
2068
2069 #ifdef CONFIG_PRINTK
2070 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2071 void __user *buffer, size_t *lenp, loff_t *ppos)
2072 {
2073 if (write && !capable(CAP_SYS_ADMIN))
2074 return -EPERM;
2075
2076 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2077 }
2078 #endif
2079
2080 struct do_proc_dointvec_minmax_conv_param {
2081 int *min;
2082 int *max;
2083 };
2084
2085 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2086 int *valp,
2087 int write, void *data)
2088 {
2089 struct do_proc_dointvec_minmax_conv_param *param = data;
2090 if (write) {
2091 int val = *negp ? -*lvalp : *lvalp;
2092 if ((param->min && *param->min > val) ||
2093 (param->max && *param->max < val))
2094 return -EINVAL;
2095 *valp = val;
2096 } else {
2097 int val = *valp;
2098 if (val < 0) {
2099 *negp = true;
2100 *lvalp = (unsigned long)-val;
2101 } else {
2102 *negp = false;
2103 *lvalp = (unsigned long)val;
2104 }
2105 }
2106 return 0;
2107 }
2108
2109 /**
2110 * proc_dointvec_minmax - read a vector of integers with min/max values
2111 * @table: the sysctl table
2112 * @write: %TRUE if this is a write to the sysctl file
2113 * @buffer: the user buffer
2114 * @lenp: the size of the user buffer
2115 * @ppos: file position
2116 *
2117 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2118 * values from/to the user buffer, treated as an ASCII string.
2119 *
2120 * This routine will ensure the values are within the range specified by
2121 * table->extra1 (min) and table->extra2 (max).
2122 *
2123 * Returns 0 on success.
2124 */
2125 int proc_dointvec_minmax(struct ctl_table *table, int write,
2126 void __user *buffer, size_t *lenp, loff_t *ppos)
2127 {
2128 struct do_proc_dointvec_minmax_conv_param param = {
2129 .min = (int *) table->extra1,
2130 .max = (int *) table->extra2,
2131 };
2132 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2133 do_proc_dointvec_minmax_conv, &param);
2134 }
2135
2136 static void validate_coredump_safety(void)
2137 {
2138 #ifdef CONFIG_COREDUMP
2139 if (suid_dumpable == SUID_DUMP_ROOT &&
2140 core_pattern[0] != '/' && core_pattern[0] != '|') {
2141 printk(KERN_WARNING "Unsafe core_pattern used with "\
2142 "suid_dumpable=2. Pipe handler or fully qualified "\
2143 "core dump path required.\n");
2144 }
2145 #endif
2146 }
2147
2148 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2149 void __user *buffer, size_t *lenp, loff_t *ppos)
2150 {
2151 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2152 if (!error)
2153 validate_coredump_safety();
2154 return error;
2155 }
2156
2157 #ifdef CONFIG_COREDUMP
2158 static int proc_dostring_coredump(struct ctl_table *table, int write,
2159 void __user *buffer, size_t *lenp, loff_t *ppos)
2160 {
2161 int error = proc_dostring(table, write, buffer, lenp, ppos);
2162 if (!error)
2163 validate_coredump_safety();
2164 return error;
2165 }
2166 #endif
2167
2168 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2169 void __user *buffer,
2170 size_t *lenp, loff_t *ppos,
2171 unsigned long convmul,
2172 unsigned long convdiv)
2173 {
2174 unsigned long *i, *min, *max;
2175 int vleft, first = 1, err = 0;
2176 unsigned long page = 0;
2177 size_t left;
2178 char *kbuf;
2179
2180 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2181 *lenp = 0;
2182 return 0;
2183 }
2184
2185 i = (unsigned long *) data;
2186 min = (unsigned long *) table->extra1;
2187 max = (unsigned long *) table->extra2;
2188 vleft = table->maxlen / sizeof(unsigned long);
2189 left = *lenp;
2190
2191 if (write) {
2192 if (left > PAGE_SIZE - 1)
2193 left = PAGE_SIZE - 1;
2194 page = __get_free_page(GFP_TEMPORARY);
2195 kbuf = (char *) page;
2196 if (!kbuf)
2197 return -ENOMEM;
2198 if (copy_from_user(kbuf, buffer, left)) {
2199 err = -EFAULT;
2200 goto free;
2201 }
2202 kbuf[left] = 0;
2203 }
2204
2205 for (; left && vleft--; i++, first = 0) {
2206 unsigned long val;
2207
2208 if (write) {
2209 bool neg;
2210
2211 left -= proc_skip_spaces(&kbuf);
2212
2213 err = proc_get_long(&kbuf, &left, &val, &neg,
2214 proc_wspace_sep,
2215 sizeof(proc_wspace_sep), NULL);
2216 if (err)
2217 break;
2218 if (neg)
2219 continue;
2220 if ((min && val < *min) || (max && val > *max))
2221 continue;
2222 *i = val;
2223 } else {
2224 val = convdiv * (*i) / convmul;
2225 if (!first) {
2226 err = proc_put_char(&buffer, &left, '\t');
2227 if (err)
2228 break;
2229 }
2230 err = proc_put_long(&buffer, &left, val, false);
2231 if (err)
2232 break;
2233 }
2234 }
2235
2236 if (!write && !first && left && !err)
2237 err = proc_put_char(&buffer, &left, '\n');
2238 if (write && !err)
2239 left -= proc_skip_spaces(&kbuf);
2240 free:
2241 if (write) {
2242 free_page(page);
2243 if (first)
2244 return err ? : -EINVAL;
2245 }
2246 *lenp -= left;
2247 *ppos += *lenp;
2248 return err;
2249 }
2250
2251 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2252 void __user *buffer,
2253 size_t *lenp, loff_t *ppos,
2254 unsigned long convmul,
2255 unsigned long convdiv)
2256 {
2257 return __do_proc_doulongvec_minmax(table->data, table, write,
2258 buffer, lenp, ppos, convmul, convdiv);
2259 }
2260
2261 /**
2262 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2263 * @table: the sysctl table
2264 * @write: %TRUE if this is a write to the sysctl file
2265 * @buffer: the user buffer
2266 * @lenp: the size of the user buffer
2267 * @ppos: file position
2268 *
2269 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2270 * values from/to the user buffer, treated as an ASCII string.
2271 *
2272 * This routine will ensure the values are within the range specified by
2273 * table->extra1 (min) and table->extra2 (max).
2274 *
2275 * Returns 0 on success.
2276 */
2277 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2278 void __user *buffer, size_t *lenp, loff_t *ppos)
2279 {
2280 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2281 }
2282
2283 /**
2284 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2285 * @table: the sysctl table
2286 * @write: %TRUE if this is a write to the sysctl file
2287 * @buffer: the user buffer
2288 * @lenp: the size of the user buffer
2289 * @ppos: file position
2290 *
2291 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2292 * values from/to the user buffer, treated as an ASCII string. The values
2293 * are treated as milliseconds, and converted to jiffies when they are stored.
2294 *
2295 * This routine will ensure the values are within the range specified by
2296 * table->extra1 (min) and table->extra2 (max).
2297 *
2298 * Returns 0 on success.
2299 */
2300 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2301 void __user *buffer,
2302 size_t *lenp, loff_t *ppos)
2303 {
2304 return do_proc_doulongvec_minmax(table, write, buffer,
2305 lenp, ppos, HZ, 1000l);
2306 }
2307
2308
2309 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2310 int *valp,
2311 int write, void *data)
2312 {
2313 if (write) {
2314 if (*lvalp > LONG_MAX / HZ)
2315 return 1;
2316 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2317 } else {
2318 int val = *valp;
2319 unsigned long lval;
2320 if (val < 0) {
2321 *negp = true;
2322 lval = (unsigned long)-val;
2323 } else {
2324 *negp = false;
2325 lval = (unsigned long)val;
2326 }
2327 *lvalp = lval / HZ;
2328 }
2329 return 0;
2330 }
2331
2332 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2333 int *valp,
2334 int write, void *data)
2335 {
2336 if (write) {
2337 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2338 return 1;
2339 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2340 } else {
2341 int val = *valp;
2342 unsigned long lval;
2343 if (val < 0) {
2344 *negp = true;
2345 lval = (unsigned long)-val;
2346 } else {
2347 *negp = false;
2348 lval = (unsigned long)val;
2349 }
2350 *lvalp = jiffies_to_clock_t(lval);
2351 }
2352 return 0;
2353 }
2354
2355 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2356 int *valp,
2357 int write, void *data)
2358 {
2359 if (write) {
2360 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2361
2362 if (jif > INT_MAX)
2363 return 1;
2364 *valp = (int)jif;
2365 } else {
2366 int val = *valp;
2367 unsigned long lval;
2368 if (val < 0) {
2369 *negp = true;
2370 lval = (unsigned long)-val;
2371 } else {
2372 *negp = false;
2373 lval = (unsigned long)val;
2374 }
2375 *lvalp = jiffies_to_msecs(lval);
2376 }
2377 return 0;
2378 }
2379
2380 /**
2381 * proc_dointvec_jiffies - read a vector of integers as seconds
2382 * @table: the sysctl table
2383 * @write: %TRUE if this is a write to the sysctl file
2384 * @buffer: the user buffer
2385 * @lenp: the size of the user buffer
2386 * @ppos: file position
2387 *
2388 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2389 * values from/to the user buffer, treated as an ASCII string.
2390 * The values read are assumed to be in seconds, and are converted into
2391 * jiffies.
2392 *
2393 * Returns 0 on success.
2394 */
2395 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2396 void __user *buffer, size_t *lenp, loff_t *ppos)
2397 {
2398 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2399 do_proc_dointvec_jiffies_conv,NULL);
2400 }
2401
2402 /**
2403 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2404 * @table: the sysctl table
2405 * @write: %TRUE if this is a write to the sysctl file
2406 * @buffer: the user buffer
2407 * @lenp: the size of the user buffer
2408 * @ppos: pointer to the file position
2409 *
2410 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2411 * values from/to the user buffer, treated as an ASCII string.
2412 * The values read are assumed to be in 1/USER_HZ seconds, and
2413 * are converted into jiffies.
2414 *
2415 * Returns 0 on success.
2416 */
2417 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2418 void __user *buffer, size_t *lenp, loff_t *ppos)
2419 {
2420 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2421 do_proc_dointvec_userhz_jiffies_conv,NULL);
2422 }
2423
2424 /**
2425 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2426 * @table: the sysctl table
2427 * @write: %TRUE if this is a write to the sysctl file
2428 * @buffer: the user buffer
2429 * @lenp: the size of the user buffer
2430 * @ppos: file position
2431 * @ppos: the current position in the file
2432 *
2433 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2434 * values from/to the user buffer, treated as an ASCII string.
2435 * The values read are assumed to be in 1/1000 seconds, and
2436 * are converted into jiffies.
2437 *
2438 * Returns 0 on success.
2439 */
2440 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2441 void __user *buffer, size_t *lenp, loff_t *ppos)
2442 {
2443 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2444 do_proc_dointvec_ms_jiffies_conv, NULL);
2445 }
2446
2447 static int proc_do_cad_pid(struct ctl_table *table, int write,
2448 void __user *buffer, size_t *lenp, loff_t *ppos)
2449 {
2450 struct pid *new_pid;
2451 pid_t tmp;
2452 int r;
2453
2454 tmp = pid_vnr(cad_pid);
2455
2456 r = __do_proc_dointvec(&tmp, table, write, buffer,
2457 lenp, ppos, NULL, NULL);
2458 if (r || !write)
2459 return r;
2460
2461 new_pid = find_get_pid(tmp);
2462 if (!new_pid)
2463 return -ESRCH;
2464
2465 put_pid(xchg(&cad_pid, new_pid));
2466 return 0;
2467 }
2468
2469 /**
2470 * proc_do_large_bitmap - read/write from/to a large bitmap
2471 * @table: the sysctl table
2472 * @write: %TRUE if this is a write to the sysctl file
2473 * @buffer: the user buffer
2474 * @lenp: the size of the user buffer
2475 * @ppos: file position
2476 *
2477 * The bitmap is stored at table->data and the bitmap length (in bits)
2478 * in table->maxlen.
2479 *
2480 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2481 * large bitmaps may be represented in a compact manner. Writing into
2482 * the file will clear the bitmap then update it with the given input.
2483 *
2484 * Returns 0 on success.
2485 */
2486 int proc_do_large_bitmap(struct ctl_table *table, int write,
2487 void __user *buffer, size_t *lenp, loff_t *ppos)
2488 {
2489 int err = 0;
2490 bool first = 1;
2491 size_t left = *lenp;
2492 unsigned long bitmap_len = table->maxlen;
2493 unsigned long *bitmap = (unsigned long *) table->data;
2494 unsigned long *tmp_bitmap = NULL;
2495 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2496
2497 if (!bitmap_len || !left || (*ppos && !write)) {
2498 *lenp = 0;
2499 return 0;
2500 }
2501
2502 if (write) {
2503 unsigned long page = 0;
2504 char *kbuf;
2505
2506 if (left > PAGE_SIZE - 1)
2507 left = PAGE_SIZE - 1;
2508
2509 page = __get_free_page(GFP_TEMPORARY);
2510 kbuf = (char *) page;
2511 if (!kbuf)
2512 return -ENOMEM;
2513 if (copy_from_user(kbuf, buffer, left)) {
2514 free_page(page);
2515 return -EFAULT;
2516 }
2517 kbuf[left] = 0;
2518
2519 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2520 GFP_KERNEL);
2521 if (!tmp_bitmap) {
2522 free_page(page);
2523 return -ENOMEM;
2524 }
2525 proc_skip_char(&kbuf, &left, '\n');
2526 while (!err && left) {
2527 unsigned long val_a, val_b;
2528 bool neg;
2529
2530 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2531 sizeof(tr_a), &c);
2532 if (err)
2533 break;
2534 if (val_a >= bitmap_len || neg) {
2535 err = -EINVAL;
2536 break;
2537 }
2538
2539 val_b = val_a;
2540 if (left) {
2541 kbuf++;
2542 left--;
2543 }
2544
2545 if (c == '-') {
2546 err = proc_get_long(&kbuf, &left, &val_b,
2547 &neg, tr_b, sizeof(tr_b),
2548 &c);
2549 if (err)
2550 break;
2551 if (val_b >= bitmap_len || neg ||
2552 val_a > val_b) {
2553 err = -EINVAL;
2554 break;
2555 }
2556 if (left) {
2557 kbuf++;
2558 left--;
2559 }
2560 }
2561
2562 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2563 first = 0;
2564 proc_skip_char(&kbuf, &left, '\n');
2565 }
2566 free_page(page);
2567 } else {
2568 unsigned long bit_a, bit_b = 0;
2569
2570 while (left) {
2571 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2572 if (bit_a >= bitmap_len)
2573 break;
2574 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2575 bit_a + 1) - 1;
2576
2577 if (!first) {
2578 err = proc_put_char(&buffer, &left, ',');
2579 if (err)
2580 break;
2581 }
2582 err = proc_put_long(&buffer, &left, bit_a, false);
2583 if (err)
2584 break;
2585 if (bit_a != bit_b) {
2586 err = proc_put_char(&buffer, &left, '-');
2587 if (err)
2588 break;
2589 err = proc_put_long(&buffer, &left, bit_b, false);
2590 if (err)
2591 break;
2592 }
2593
2594 first = 0; bit_b++;
2595 }
2596 if (!err)
2597 err = proc_put_char(&buffer, &left, '\n');
2598 }
2599
2600 if (!err) {
2601 if (write) {
2602 if (*ppos)
2603 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2604 else
2605 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2606 }
2607 kfree(tmp_bitmap);
2608 *lenp -= left;
2609 *ppos += *lenp;
2610 return 0;
2611 } else {
2612 kfree(tmp_bitmap);
2613 return err;
2614 }
2615 }
2616
2617 #else /* CONFIG_PROC_SYSCTL */
2618
2619 int proc_dostring(struct ctl_table *table, int write,
2620 void __user *buffer, size_t *lenp, loff_t *ppos)
2621 {
2622 return -ENOSYS;
2623 }
2624
2625 int proc_dointvec(struct ctl_table *table, int write,
2626 void __user *buffer, size_t *lenp, loff_t *ppos)
2627 {
2628 return -ENOSYS;
2629 }
2630
2631 int proc_dointvec_minmax(struct ctl_table *table, int write,
2632 void __user *buffer, size_t *lenp, loff_t *ppos)
2633 {
2634 return -ENOSYS;
2635 }
2636
2637 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2638 void __user *buffer, size_t *lenp, loff_t *ppos)
2639 {
2640 return -ENOSYS;
2641 }
2642
2643 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2644 void __user *buffer, size_t *lenp, loff_t *ppos)
2645 {
2646 return -ENOSYS;
2647 }
2648
2649 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2650 void __user *buffer, size_t *lenp, loff_t *ppos)
2651 {
2652 return -ENOSYS;
2653 }
2654
2655 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2656 void __user *buffer, size_t *lenp, loff_t *ppos)
2657 {
2658 return -ENOSYS;
2659 }
2660
2661 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2662 void __user *buffer,
2663 size_t *lenp, loff_t *ppos)
2664 {
2665 return -ENOSYS;
2666 }
2667
2668
2669 #endif /* CONFIG_PROC_SYSCTL */
2670
2671 /*
2672 * No sense putting this after each symbol definition, twice,
2673 * exception granted :-)
2674 */
2675 EXPORT_SYMBOL(proc_dointvec);
2676 EXPORT_SYMBOL(proc_dointvec_jiffies);
2677 EXPORT_SYMBOL(proc_dointvec_minmax);
2678 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2679 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2680 EXPORT_SYMBOL(proc_dostring);
2681 EXPORT_SYMBOL(proc_doulongvec_minmax);
2682 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
Linux kernel - Deliverables
This page took 0.08189 seconds and 6 git commands to generate.