projects / deliverable / linux.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Merge remote-tracking branches 'asoc/fix/arizona', 'asoc/fix/fsl', 'asoc/fix/fsl...
[deliverable/linux.git] / kernel / sysctl.c
1 /*
2 * sysctl.c: General linux system control interface
3 *
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12 * Horn.
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16 * Wendling.
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19 */
20
21 #include <linux/module.h>
22 #include <linux/mm.h>
23 #include <linux/swap.h>
24 #include <linux/slab.h>
25 #include <linux/sysctl.h>
26 #include <linux/bitmap.h>
27 #include <linux/signal.h>
28 #include <linux/printk.h>
29 #include <linux/proc_fs.h>
30 #include <linux/security.h>
31 #include <linux/ctype.h>
32 #include <linux/kmemcheck.h>
33 #include <linux/kmemleak.h>
34 #include <linux/fs.h>
35 #include <linux/init.h>
36 #include <linux/kernel.h>
37 #include <linux/kobject.h>
38 #include <linux/net.h>
39 #include <linux/sysrq.h>
40 #include <linux/highuid.h>
41 #include <linux/writeback.h>
42 #include <linux/ratelimit.h>
43 #include <linux/compaction.h>
44 #include <linux/hugetlb.h>
45 #include <linux/initrd.h>
46 #include <linux/key.h>
47 #include <linux/times.h>
48 #include <linux/limits.h>
49 #include <linux/dcache.h>
50 #include <linux/dnotify.h>
51 #include <linux/syscalls.h>
52 #include <linux/vmstat.h>
53 #include <linux/nfs_fs.h>
54 #include <linux/acpi.h>
55 #include <linux/reboot.h>
56 #include <linux/ftrace.h>
57 #include <linux/perf_event.h>
58 #include <linux/kprobes.h>
59 #include <linux/pipe_fs_i.h>
60 #include <linux/oom.h>
61 #include <linux/kmod.h>
62 #include <linux/capability.h>
63 #include <linux/binfmts.h>
64 #include <linux/sched/sysctl.h>
65 #include <linux/kexec.h>
66
67 #include <asm/uaccess.h>
68 #include <asm/processor.h>
69
70 #ifdef CONFIG_X86
71 #include <asm/nmi.h>
72 #include <asm/stacktrace.h>
73 #include <asm/io.h>
74 #endif
75 #ifdef CONFIG_SPARC
76 #include <asm/setup.h>
77 #endif
78 #ifdef CONFIG_BSD_PROCESS_ACCT
79 #include <linux/acct.h>
80 #endif
81 #ifdef CONFIG_RT_MUTEXES
82 #include <linux/rtmutex.h>
83 #endif
84 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
85 #include <linux/lockdep.h>
86 #endif
87 #ifdef CONFIG_CHR_DEV_SG
88 #include <scsi/sg.h>
89 #endif
90
91 #ifdef CONFIG_LOCKUP_DETECTOR
92 #include <linux/nmi.h>
93 #endif
94
95
96 #if defined(CONFIG_SYSCTL)
97
98 /* External variables not in a header file. */
99 extern int max_threads;
100 extern int suid_dumpable;
101 #ifdef CONFIG_COREDUMP
102 extern int core_uses_pid;
103 extern char core_pattern[];
104 extern unsigned int core_pipe_limit;
105 #endif
106 extern int pid_max;
107 extern int pid_max_min, pid_max_max;
108 extern int percpu_pagelist_fraction;
109 extern int compat_log;
110 extern int latencytop_enabled;
111 extern int sysctl_nr_open_min, sysctl_nr_open_max;
112 #ifndef CONFIG_MMU
113 extern int sysctl_nr_trim_pages;
114 #endif
115
116 /* Constants used for minimum and maximum */
117 #ifdef CONFIG_LOCKUP_DETECTOR
118 static int sixty = 60;
119 #endif
120
121 static int __maybe_unused neg_one = -1;
122
123 static int zero;
124 static int __maybe_unused one = 1;
125 static int __maybe_unused two = 2;
126 static int __maybe_unused four = 4;
127 static unsigned long one_ul = 1;
128 static int one_hundred = 100;
129 #ifdef CONFIG_PRINTK
130 static int ten_thousand = 10000;
131 #endif
132
133 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
134 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
135
136 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
137 static int maxolduid = 65535;
138 static int minolduid;
139
140 static int ngroups_max = NGROUPS_MAX;
141 static const int cap_last_cap = CAP_LAST_CAP;
142
143 /*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */
144 #ifdef CONFIG_DETECT_HUNG_TASK
145 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
146 #endif
147
148 #ifdef CONFIG_INOTIFY_USER
149 #include <linux/inotify.h>
150 #endif
151 #ifdef CONFIG_SPARC
152 #endif
153
154 #ifdef __hppa__
155 extern int pwrsw_enabled;
156 #endif
157
158 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
159 extern int unaligned_enabled;
160 #endif
161
162 #ifdef CONFIG_IA64
163 extern int unaligned_dump_stack;
164 #endif
165
166 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
167 extern int no_unaligned_warning;
168 #endif
169
170 #ifdef CONFIG_PROC_SYSCTL
171
172 #define SYSCTL_WRITES_LEGACY -1
173 #define SYSCTL_WRITES_WARN 0
174 #define SYSCTL_WRITES_STRICT 1
175
176 static int sysctl_writes_strict = SYSCTL_WRITES_WARN;
177
178 static int proc_do_cad_pid(struct ctl_table *table, int write,
179 void __user *buffer, size_t *lenp, loff_t *ppos);
180 static int proc_taint(struct ctl_table *table, int write,
181 void __user *buffer, size_t *lenp, loff_t *ppos);
182 #endif
183
184 #ifdef CONFIG_PRINTK
185 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
186 void __user *buffer, size_t *lenp, loff_t *ppos);
187 #endif
188
189 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
190 void __user *buffer, size_t *lenp, loff_t *ppos);
191 #ifdef CONFIG_COREDUMP
192 static int proc_dostring_coredump(struct ctl_table *table, int write,
193 void __user *buffer, size_t *lenp, loff_t *ppos);
194 #endif
195
196 #ifdef CONFIG_MAGIC_SYSRQ
197 /* Note: sysrq code uses it's own private copy */
198 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
199
200 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
201 void __user *buffer, size_t *lenp,
202 loff_t *ppos)
203 {
204 int error;
205
206 error = proc_dointvec(table, write, buffer, lenp, ppos);
207 if (error)
208 return error;
209
210 if (write)
211 sysrq_toggle_support(__sysrq_enabled);
212
213 return 0;
214 }
215
216 #endif
217
218 static struct ctl_table kern_table[];
219 static struct ctl_table vm_table[];
220 static struct ctl_table fs_table[];
221 static struct ctl_table debug_table[];
222 static struct ctl_table dev_table[];
223 extern struct ctl_table random_table[];
224 #ifdef CONFIG_EPOLL
225 extern struct ctl_table epoll_table[];
226 #endif
227
228 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
229 int sysctl_legacy_va_layout;
230 #endif
231
232 /* The default sysctl tables: */
233
234 static struct ctl_table sysctl_base_table[] = {
235 {
236 .procname = "kernel",
237 .mode = 0555,
238 .child = kern_table,
239 },
240 {
241 .procname = "vm",
242 .mode = 0555,
243 .child = vm_table,
244 },
245 {
246 .procname = "fs",
247 .mode = 0555,
248 .child = fs_table,
249 },
250 {
251 .procname = "debug",
252 .mode = 0555,
253 .child = debug_table,
254 },
255 {
256 .procname = "dev",
257 .mode = 0555,
258 .child = dev_table,
259 },
260 { }
261 };
262
263 #ifdef CONFIG_SCHED_DEBUG
264 static int min_sched_granularity_ns = 100000; /* 100 usecs */
265 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
266 static int min_wakeup_granularity_ns; /* 0 usecs */
267 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
268 #ifdef CONFIG_SMP
269 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
270 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
271 #endif /* CONFIG_SMP */
272 #endif /* CONFIG_SCHED_DEBUG */
273
274 #ifdef CONFIG_COMPACTION
275 static int min_extfrag_threshold;
276 static int max_extfrag_threshold = 1000;
277 #endif
278
279 static struct ctl_table kern_table[] = {
280 {
281 .procname = "sched_child_runs_first",
282 .data = &sysctl_sched_child_runs_first,
283 .maxlen = sizeof(unsigned int),
284 .mode = 0644,
285 .proc_handler = proc_dointvec,
286 },
287 #ifdef CONFIG_SCHED_DEBUG
288 {
289 .procname = "sched_min_granularity_ns",
290 .data = &sysctl_sched_min_granularity,
291 .maxlen = sizeof(unsigned int),
292 .mode = 0644,
293 .proc_handler = sched_proc_update_handler,
294 .extra1 = &min_sched_granularity_ns,
295 .extra2 = &max_sched_granularity_ns,
296 },
297 {
298 .procname = "sched_latency_ns",
299 .data = &sysctl_sched_latency,
300 .maxlen = sizeof(unsigned int),
301 .mode = 0644,
302 .proc_handler = sched_proc_update_handler,
303 .extra1 = &min_sched_granularity_ns,
304 .extra2 = &max_sched_granularity_ns,
305 },
306 {
307 .procname = "sched_wakeup_granularity_ns",
308 .data = &sysctl_sched_wakeup_granularity,
309 .maxlen = sizeof(unsigned int),
310 .mode = 0644,
311 .proc_handler = sched_proc_update_handler,
312 .extra1 = &min_wakeup_granularity_ns,
313 .extra2 = &max_wakeup_granularity_ns,
314 },
315 #ifdef CONFIG_SMP
316 {
317 .procname = "sched_tunable_scaling",
318 .data = &sysctl_sched_tunable_scaling,
319 .maxlen = sizeof(enum sched_tunable_scaling),
320 .mode = 0644,
321 .proc_handler = sched_proc_update_handler,
322 .extra1 = &min_sched_tunable_scaling,
323 .extra2 = &max_sched_tunable_scaling,
324 },
325 {
326 .procname = "sched_migration_cost_ns",
327 .data = &sysctl_sched_migration_cost,
328 .maxlen = sizeof(unsigned int),
329 .mode = 0644,
330 .proc_handler = proc_dointvec,
331 },
332 {
333 .procname = "sched_nr_migrate",
334 .data = &sysctl_sched_nr_migrate,
335 .maxlen = sizeof(unsigned int),
336 .mode = 0644,
337 .proc_handler = proc_dointvec,
338 },
339 {
340 .procname = "sched_time_avg_ms",
341 .data = &sysctl_sched_time_avg,
342 .maxlen = sizeof(unsigned int),
343 .mode = 0644,
344 .proc_handler = proc_dointvec,
345 },
346 {
347 .procname = "sched_shares_window_ns",
348 .data = &sysctl_sched_shares_window,
349 .maxlen = sizeof(unsigned int),
350 .mode = 0644,
351 .proc_handler = proc_dointvec,
352 },
353 {
354 .procname = "timer_migration",
355 .data = &sysctl_timer_migration,
356 .maxlen = sizeof(unsigned int),
357 .mode = 0644,
358 .proc_handler = proc_dointvec_minmax,
359 .extra1 = &zero,
360 .extra2 = &one,
361 },
362 #endif /* CONFIG_SMP */
363 #ifdef CONFIG_NUMA_BALANCING
364 {
365 .procname = "numa_balancing_scan_delay_ms",
366 .data = &sysctl_numa_balancing_scan_delay,
367 .maxlen = sizeof(unsigned int),
368 .mode = 0644,
369 .proc_handler = proc_dointvec,
370 },
371 {
372 .procname = "numa_balancing_scan_period_min_ms",
373 .data = &sysctl_numa_balancing_scan_period_min,
374 .maxlen = sizeof(unsigned int),
375 .mode = 0644,
376 .proc_handler = proc_dointvec,
377 },
378 {
379 .procname = "numa_balancing_scan_period_max_ms",
380 .data = &sysctl_numa_balancing_scan_period_max,
381 .maxlen = sizeof(unsigned int),
382 .mode = 0644,
383 .proc_handler = proc_dointvec,
384 },
385 {
386 .procname = "numa_balancing_scan_size_mb",
387 .data = &sysctl_numa_balancing_scan_size,
388 .maxlen = sizeof(unsigned int),
389 .mode = 0644,
390 .proc_handler = proc_dointvec,
391 },
392 {
393 .procname = "numa_balancing",
394 .data = NULL, /* filled in by handler */
395 .maxlen = sizeof(unsigned int),
396 .mode = 0644,
397 .proc_handler = sysctl_numa_balancing,
398 .extra1 = &zero,
399 .extra2 = &one,
400 },
401 #endif /* CONFIG_NUMA_BALANCING */
402 #endif /* CONFIG_SCHED_DEBUG */
403 {
404 .procname = "sched_rt_period_us",
405 .data = &sysctl_sched_rt_period,
406 .maxlen = sizeof(unsigned int),
407 .mode = 0644,
408 .proc_handler = sched_rt_handler,
409 },
410 {
411 .procname = "sched_rt_runtime_us",
412 .data = &sysctl_sched_rt_runtime,
413 .maxlen = sizeof(int),
414 .mode = 0644,
415 .proc_handler = sched_rt_handler,
416 },
417 {
418 .procname = "sched_rr_timeslice_ms",
419 .data = &sched_rr_timeslice,
420 .maxlen = sizeof(int),
421 .mode = 0644,
422 .proc_handler = sched_rr_handler,
423 },
424 #ifdef CONFIG_SCHED_AUTOGROUP
425 {
426 .procname = "sched_autogroup_enabled",
427 .data = &sysctl_sched_autogroup_enabled,
428 .maxlen = sizeof(unsigned int),
429 .mode = 0644,
430 .proc_handler = proc_dointvec_minmax,
431 .extra1 = &zero,
432 .extra2 = &one,
433 },
434 #endif
435 #ifdef CONFIG_CFS_BANDWIDTH
436 {
437 .procname = "sched_cfs_bandwidth_slice_us",
438 .data = &sysctl_sched_cfs_bandwidth_slice,
439 .maxlen = sizeof(unsigned int),
440 .mode = 0644,
441 .proc_handler = proc_dointvec_minmax,
442 .extra1 = &one,
443 },
444 #endif
445 #ifdef CONFIG_PROVE_LOCKING
446 {
447 .procname = "prove_locking",
448 .data = &prove_locking,
449 .maxlen = sizeof(int),
450 .mode = 0644,
451 .proc_handler = proc_dointvec,
452 },
453 #endif
454 #ifdef CONFIG_LOCK_STAT
455 {
456 .procname = "lock_stat",
457 .data = &lock_stat,
458 .maxlen = sizeof(int),
459 .mode = 0644,
460 .proc_handler = proc_dointvec,
461 },
462 #endif
463 {
464 .procname = "panic",
465 .data = &panic_timeout,
466 .maxlen = sizeof(int),
467 .mode = 0644,
468 .proc_handler = proc_dointvec,
469 },
470 #ifdef CONFIG_COREDUMP
471 {
472 .procname = "core_uses_pid",
473 .data = &core_uses_pid,
474 .maxlen = sizeof(int),
475 .mode = 0644,
476 .proc_handler = proc_dointvec,
477 },
478 {
479 .procname = "core_pattern",
480 .data = core_pattern,
481 .maxlen = CORENAME_MAX_SIZE,
482 .mode = 0644,
483 .proc_handler = proc_dostring_coredump,
484 },
485 {
486 .procname = "core_pipe_limit",
487 .data = &core_pipe_limit,
488 .maxlen = sizeof(unsigned int),
489 .mode = 0644,
490 .proc_handler = proc_dointvec,
491 },
492 #endif
493 #ifdef CONFIG_PROC_SYSCTL
494 {
495 .procname = "tainted",
496 .maxlen = sizeof(long),
497 .mode = 0644,
498 .proc_handler = proc_taint,
499 },
500 {
501 .procname = "sysctl_writes_strict",
502 .data = &sysctl_writes_strict,
503 .maxlen = sizeof(int),
504 .mode = 0644,
505 .proc_handler = proc_dointvec_minmax,
506 .extra1 = &neg_one,
507 .extra2 = &one,
508 },
509 #endif
510 #ifdef CONFIG_LATENCYTOP
511 {
512 .procname = "latencytop",
513 .data = &latencytop_enabled,
514 .maxlen = sizeof(int),
515 .mode = 0644,
516 .proc_handler = proc_dointvec,
517 },
518 #endif
519 #ifdef CONFIG_BLK_DEV_INITRD
520 {
521 .procname = "real-root-dev",
522 .data = &real_root_dev,
523 .maxlen = sizeof(int),
524 .mode = 0644,
525 .proc_handler = proc_dointvec,
526 },
527 #endif
528 {
529 .procname = "print-fatal-signals",
530 .data = &print_fatal_signals,
531 .maxlen = sizeof(int),
532 .mode = 0644,
533 .proc_handler = proc_dointvec,
534 },
535 #ifdef CONFIG_SPARC
536 {
537 .procname = "reboot-cmd",
538 .data = reboot_command,
539 .maxlen = 256,
540 .mode = 0644,
541 .proc_handler = proc_dostring,
542 },
543 {
544 .procname = "stop-a",
545 .data = &stop_a_enabled,
546 .maxlen = sizeof (int),
547 .mode = 0644,
548 .proc_handler = proc_dointvec,
549 },
550 {
551 .procname = "scons-poweroff",
552 .data = &scons_pwroff,
553 .maxlen = sizeof (int),
554 .mode = 0644,
555 .proc_handler = proc_dointvec,
556 },
557 #endif
558 #ifdef CONFIG_SPARC64
559 {
560 .procname = "tsb-ratio",
561 .data = &sysctl_tsb_ratio,
562 .maxlen = sizeof (int),
563 .mode = 0644,
564 .proc_handler = proc_dointvec,
565 },
566 #endif
567 #ifdef __hppa__
568 {
569 .procname = "soft-power",
570 .data = &pwrsw_enabled,
571 .maxlen = sizeof (int),
572 .mode = 0644,
573 .proc_handler = proc_dointvec,
574 },
575 #endif
576 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
577 {
578 .procname = "unaligned-trap",
579 .data = &unaligned_enabled,
580 .maxlen = sizeof (int),
581 .mode = 0644,
582 .proc_handler = proc_dointvec,
583 },
584 #endif
585 {
586 .procname = "ctrl-alt-del",
587 .data = &C_A_D,
588 .maxlen = sizeof(int),
589 .mode = 0644,
590 .proc_handler = proc_dointvec,
591 },
592 #ifdef CONFIG_FUNCTION_TRACER
593 {
594 .procname = "ftrace_enabled",
595 .data = &ftrace_enabled,
596 .maxlen = sizeof(int),
597 .mode = 0644,
598 .proc_handler = ftrace_enable_sysctl,
599 },
600 #endif
601 #ifdef CONFIG_STACK_TRACER
602 {
603 .procname = "stack_tracer_enabled",
604 .data = &stack_tracer_enabled,
605 .maxlen = sizeof(int),
606 .mode = 0644,
607 .proc_handler = stack_trace_sysctl,
608 },
609 #endif
610 #ifdef CONFIG_TRACING
611 {
612 .procname = "ftrace_dump_on_oops",
613 .data = &ftrace_dump_on_oops,
614 .maxlen = sizeof(int),
615 .mode = 0644,
616 .proc_handler = proc_dointvec,
617 },
618 {
619 .procname = "traceoff_on_warning",
620 .data = &__disable_trace_on_warning,
621 .maxlen = sizeof(__disable_trace_on_warning),
622 .mode = 0644,
623 .proc_handler = proc_dointvec,
624 },
625 #endif
626 #ifdef CONFIG_KEXEC
627 {
628 .procname = "kexec_load_disabled",
629 .data = &kexec_load_disabled,
630 .maxlen = sizeof(int),
631 .mode = 0644,
632 /* only handle a transition from default "0" to "1" */
633 .proc_handler = proc_dointvec_minmax,
634 .extra1 = &one,
635 .extra2 = &one,
636 },
637 #endif
638 #ifdef CONFIG_MODULES
639 {
640 .procname = "modprobe",
641 .data = &modprobe_path,
642 .maxlen = KMOD_PATH_LEN,
643 .mode = 0644,
644 .proc_handler = proc_dostring,
645 },
646 {
647 .procname = "modules_disabled",
648 .data = &modules_disabled,
649 .maxlen = sizeof(int),
650 .mode = 0644,
651 /* only handle a transition from default "0" to "1" */
652 .proc_handler = proc_dointvec_minmax,
653 .extra1 = &one,
654 .extra2 = &one,
655 },
656 #endif
657 #ifdef CONFIG_UEVENT_HELPER
658 {
659 .procname = "hotplug",
660 .data = &uevent_helper,
661 .maxlen = UEVENT_HELPER_PATH_LEN,
662 .mode = 0644,
663 .proc_handler = proc_dostring,
664 },
665 #endif
666 #ifdef CONFIG_CHR_DEV_SG
667 {
668 .procname = "sg-big-buff",
669 .data = &sg_big_buff,
670 .maxlen = sizeof (int),
671 .mode = 0444,
672 .proc_handler = proc_dointvec,
673 },
674 #endif
675 #ifdef CONFIG_BSD_PROCESS_ACCT
676 {
677 .procname = "acct",
678 .data = &acct_parm,
679 .maxlen = 3*sizeof(int),
680 .mode = 0644,
681 .proc_handler = proc_dointvec,
682 },
683 #endif
684 #ifdef CONFIG_MAGIC_SYSRQ
685 {
686 .procname = "sysrq",
687 .data = &__sysrq_enabled,
688 .maxlen = sizeof (int),
689 .mode = 0644,
690 .proc_handler = sysrq_sysctl_handler,
691 },
692 #endif
693 #ifdef CONFIG_PROC_SYSCTL
694 {
695 .procname = "cad_pid",
696 .data = NULL,
697 .maxlen = sizeof (int),
698 .mode = 0600,
699 .proc_handler = proc_do_cad_pid,
700 },
701 #endif
702 {
703 .procname = "threads-max",
704 .data = &max_threads,
705 .maxlen = sizeof(int),
706 .mode = 0644,
707 .proc_handler = proc_dointvec,
708 },
709 {
710 .procname = "random",
711 .mode = 0555,
712 .child = random_table,
713 },
714 {
715 .procname = "usermodehelper",
716 .mode = 0555,
717 .child = usermodehelper_table,
718 },
719 {
720 .procname = "overflowuid",
721 .data = &overflowuid,
722 .maxlen = sizeof(int),
723 .mode = 0644,
724 .proc_handler = proc_dointvec_minmax,
725 .extra1 = &minolduid,
726 .extra2 = &maxolduid,
727 },
728 {
729 .procname = "overflowgid",
730 .data = &overflowgid,
731 .maxlen = sizeof(int),
732 .mode = 0644,
733 .proc_handler = proc_dointvec_minmax,
734 .extra1 = &minolduid,
735 .extra2 = &maxolduid,
736 },
737 #ifdef CONFIG_S390
738 #ifdef CONFIG_MATHEMU
739 {
740 .procname = "ieee_emulation_warnings",
741 .data = &sysctl_ieee_emulation_warnings,
742 .maxlen = sizeof(int),
743 .mode = 0644,
744 .proc_handler = proc_dointvec,
745 },
746 #endif
747 {
748 .procname = "userprocess_debug",
749 .data = &show_unhandled_signals,
750 .maxlen = sizeof(int),
751 .mode = 0644,
752 .proc_handler = proc_dointvec,
753 },
754 #endif
755 {
756 .procname = "pid_max",
757 .data = &pid_max,
758 .maxlen = sizeof (int),
759 .mode = 0644,
760 .proc_handler = proc_dointvec_minmax,
761 .extra1 = &pid_max_min,
762 .extra2 = &pid_max_max,
763 },
764 {
765 .procname = "panic_on_oops",
766 .data = &panic_on_oops,
767 .maxlen = sizeof(int),
768 .mode = 0644,
769 .proc_handler = proc_dointvec,
770 },
771 #if defined CONFIG_PRINTK
772 {
773 .procname = "printk",
774 .data = &console_loglevel,
775 .maxlen = 4*sizeof(int),
776 .mode = 0644,
777 .proc_handler = proc_dointvec,
778 },
779 {
780 .procname = "printk_ratelimit",
781 .data = &printk_ratelimit_state.interval,
782 .maxlen = sizeof(int),
783 .mode = 0644,
784 .proc_handler = proc_dointvec_jiffies,
785 },
786 {
787 .procname = "printk_ratelimit_burst",
788 .data = &printk_ratelimit_state.burst,
789 .maxlen = sizeof(int),
790 .mode = 0644,
791 .proc_handler = proc_dointvec,
792 },
793 {
794 .procname = "printk_delay",
795 .data = &printk_delay_msec,
796 .maxlen = sizeof(int),
797 .mode = 0644,
798 .proc_handler = proc_dointvec_minmax,
799 .extra1 = &zero,
800 .extra2 = &ten_thousand,
801 },
802 {
803 .procname = "dmesg_restrict",
804 .data = &dmesg_restrict,
805 .maxlen = sizeof(int),
806 .mode = 0644,
807 .proc_handler = proc_dointvec_minmax_sysadmin,
808 .extra1 = &zero,
809 .extra2 = &one,
810 },
811 {
812 .procname = "kptr_restrict",
813 .data = &kptr_restrict,
814 .maxlen = sizeof(int),
815 .mode = 0644,
816 .proc_handler = proc_dointvec_minmax_sysadmin,
817 .extra1 = &zero,
818 .extra2 = &two,
819 },
820 #endif
821 {
822 .procname = "ngroups_max",
823 .data = &ngroups_max,
824 .maxlen = sizeof (int),
825 .mode = 0444,
826 .proc_handler = proc_dointvec,
827 },
828 {
829 .procname = "cap_last_cap",
830 .data = (void *)&cap_last_cap,
831 .maxlen = sizeof(int),
832 .mode = 0444,
833 .proc_handler = proc_dointvec,
834 },
835 #if defined(CONFIG_LOCKUP_DETECTOR)
836 {
837 .procname = "watchdog",
838 .data = &watchdog_user_enabled,
839 .maxlen = sizeof (int),
840 .mode = 0644,
841 .proc_handler = proc_dowatchdog,
842 .extra1 = &zero,
843 .extra2 = &one,
844 },
845 {
846 .procname = "watchdog_thresh",
847 .data = &watchdog_thresh,
848 .maxlen = sizeof(int),
849 .mode = 0644,
850 .proc_handler = proc_dowatchdog,
851 .extra1 = &zero,
852 .extra2 = &sixty,
853 },
854 {
855 .procname = "softlockup_panic",
856 .data = &softlockup_panic,
857 .maxlen = sizeof(int),
858 .mode = 0644,
859 .proc_handler = proc_dointvec_minmax,
860 .extra1 = &zero,
861 .extra2 = &one,
862 },
863 #ifdef CONFIG_SMP
864 {
865 .procname = "softlockup_all_cpu_backtrace",
866 .data = &sysctl_softlockup_all_cpu_backtrace,
867 .maxlen = sizeof(int),
868 .mode = 0644,
869 .proc_handler = proc_dointvec_minmax,
870 .extra1 = &zero,
871 .extra2 = &one,
872 },
873 #endif /* CONFIG_SMP */
874 {
875 .procname = "nmi_watchdog",
876 .data = &watchdog_user_enabled,
877 .maxlen = sizeof (int),
878 .mode = 0644,
879 .proc_handler = proc_dowatchdog,
880 .extra1 = &zero,
881 .extra2 = &one,
882 },
883 #endif
884 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
885 {
886 .procname = "unknown_nmi_panic",
887 .data = &unknown_nmi_panic,
888 .maxlen = sizeof (int),
889 .mode = 0644,
890 .proc_handler = proc_dointvec,
891 },
892 #endif
893 #if defined(CONFIG_X86)
894 {
895 .procname = "panic_on_unrecovered_nmi",
896 .data = &panic_on_unrecovered_nmi,
897 .maxlen = sizeof(int),
898 .mode = 0644,
899 .proc_handler = proc_dointvec,
900 },
901 {
902 .procname = "panic_on_io_nmi",
903 .data = &panic_on_io_nmi,
904 .maxlen = sizeof(int),
905 .mode = 0644,
906 .proc_handler = proc_dointvec,
907 },
908 #ifdef CONFIG_DEBUG_STACKOVERFLOW
909 {
910 .procname = "panic_on_stackoverflow",
911 .data = &sysctl_panic_on_stackoverflow,
912 .maxlen = sizeof(int),
913 .mode = 0644,
914 .proc_handler = proc_dointvec,
915 },
916 #endif
917 {
918 .procname = "bootloader_type",
919 .data = &bootloader_type,
920 .maxlen = sizeof (int),
921 .mode = 0444,
922 .proc_handler = proc_dointvec,
923 },
924 {
925 .procname = "bootloader_version",
926 .data = &bootloader_version,
927 .maxlen = sizeof (int),
928 .mode = 0444,
929 .proc_handler = proc_dointvec,
930 },
931 {
932 .procname = "kstack_depth_to_print",
933 .data = &kstack_depth_to_print,
934 .maxlen = sizeof(int),
935 .mode = 0644,
936 .proc_handler = proc_dointvec,
937 },
938 {
939 .procname = "io_delay_type",
940 .data = &io_delay_type,
941 .maxlen = sizeof(int),
942 .mode = 0644,
943 .proc_handler = proc_dointvec,
944 },
945 #endif
946 #if defined(CONFIG_MMU)
947 {
948 .procname = "randomize_va_space",
949 .data = &randomize_va_space,
950 .maxlen = sizeof(int),
951 .mode = 0644,
952 .proc_handler = proc_dointvec,
953 },
954 #endif
955 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
956 {
957 .procname = "spin_retry",
958 .data = &spin_retry,
959 .maxlen = sizeof (int),
960 .mode = 0644,
961 .proc_handler = proc_dointvec,
962 },
963 #endif
964 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
965 {
966 .procname = "acpi_video_flags",
967 .data = &acpi_realmode_flags,
968 .maxlen = sizeof (unsigned long),
969 .mode = 0644,
970 .proc_handler = proc_doulongvec_minmax,
971 },
972 #endif
973 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
974 {
975 .procname = "ignore-unaligned-usertrap",
976 .data = &no_unaligned_warning,
977 .maxlen = sizeof (int),
978 .mode = 0644,
979 .proc_handler = proc_dointvec,
980 },
981 #endif
982 #ifdef CONFIG_IA64
983 {
984 .procname = "unaligned-dump-stack",
985 .data = &unaligned_dump_stack,
986 .maxlen = sizeof (int),
987 .mode = 0644,
988 .proc_handler = proc_dointvec,
989 },
990 #endif
991 #ifdef CONFIG_DETECT_HUNG_TASK
992 {
993 .procname = "hung_task_panic",
994 .data = &sysctl_hung_task_panic,
995 .maxlen = sizeof(int),
996 .mode = 0644,
997 .proc_handler = proc_dointvec_minmax,
998 .extra1 = &zero,
999 .extra2 = &one,
1000 },
1001 {
1002 .procname = "hung_task_check_count",
1003 .data = &sysctl_hung_task_check_count,
1004 .maxlen = sizeof(int),
1005 .mode = 0644,
1006 .proc_handler = proc_dointvec_minmax,
1007 .extra1 = &zero,
1008 },
1009 {
1010 .procname = "hung_task_timeout_secs",
1011 .data = &sysctl_hung_task_timeout_secs,
1012 .maxlen = sizeof(unsigned long),
1013 .mode = 0644,
1014 .proc_handler = proc_dohung_task_timeout_secs,
1015 .extra2 = &hung_task_timeout_max,
1016 },
1017 {
1018 .procname = "hung_task_warnings",
1019 .data = &sysctl_hung_task_warnings,
1020 .maxlen = sizeof(int),
1021 .mode = 0644,
1022 .proc_handler = proc_dointvec_minmax,
1023 .extra1 = &neg_one,
1024 },
1025 #endif
1026 #ifdef CONFIG_COMPAT
1027 {
1028 .procname = "compat-log",
1029 .data = &compat_log,
1030 .maxlen = sizeof (int),
1031 .mode = 0644,
1032 .proc_handler = proc_dointvec,
1033 },
1034 #endif
1035 #ifdef CONFIG_RT_MUTEXES
1036 {
1037 .procname = "max_lock_depth",
1038 .data = &max_lock_depth,
1039 .maxlen = sizeof(int),
1040 .mode = 0644,
1041 .proc_handler = proc_dointvec,
1042 },
1043 #endif
1044 {
1045 .procname = "poweroff_cmd",
1046 .data = &poweroff_cmd,
1047 .maxlen = POWEROFF_CMD_PATH_LEN,
1048 .mode = 0644,
1049 .proc_handler = proc_dostring,
1050 },
1051 #ifdef CONFIG_KEYS
1052 {
1053 .procname = "keys",
1054 .mode = 0555,
1055 .child = key_sysctls,
1056 },
1057 #endif
1058 #ifdef CONFIG_RCU_TORTURE_TEST
1059 {
1060 .procname = "rcutorture_runnable",
1061 .data = &rcutorture_runnable,
1062 .maxlen = sizeof(int),
1063 .mode = 0644,
1064 .proc_handler = proc_dointvec,
1065 },
1066 #endif
1067 #ifdef CONFIG_PERF_EVENTS
1068 /*
1069 * User-space scripts rely on the existence of this file
1070 * as a feature check for perf_events being enabled.
1071 *
1072 * So it's an ABI, do not remove!
1073 */
1074 {
1075 .procname = "perf_event_paranoid",
1076 .data = &sysctl_perf_event_paranoid,
1077 .maxlen = sizeof(sysctl_perf_event_paranoid),
1078 .mode = 0644,
1079 .proc_handler = proc_dointvec,
1080 },
1081 {
1082 .procname = "perf_event_mlock_kb",
1083 .data = &sysctl_perf_event_mlock,
1084 .maxlen = sizeof(sysctl_perf_event_mlock),
1085 .mode = 0644,
1086 .proc_handler = proc_dointvec,
1087 },
1088 {
1089 .procname = "perf_event_max_sample_rate",
1090 .data = &sysctl_perf_event_sample_rate,
1091 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1092 .mode = 0644,
1093 .proc_handler = perf_proc_update_handler,
1094 .extra1 = &one,
1095 },
1096 {
1097 .procname = "perf_cpu_time_max_percent",
1098 .data = &sysctl_perf_cpu_time_max_percent,
1099 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1100 .mode = 0644,
1101 .proc_handler = perf_cpu_time_max_percent_handler,
1102 .extra1 = &zero,
1103 .extra2 = &one_hundred,
1104 },
1105 #endif
1106 #ifdef CONFIG_KMEMCHECK
1107 {
1108 .procname = "kmemcheck",
1109 .data = &kmemcheck_enabled,
1110 .maxlen = sizeof(int),
1111 .mode = 0644,
1112 .proc_handler = proc_dointvec,
1113 },
1114 #endif
1115 { }
1116 };
1117
1118 static struct ctl_table vm_table[] = {
1119 {
1120 .procname = "overcommit_memory",
1121 .data = &sysctl_overcommit_memory,
1122 .maxlen = sizeof(sysctl_overcommit_memory),
1123 .mode = 0644,
1124 .proc_handler = proc_dointvec_minmax,
1125 .extra1 = &zero,
1126 .extra2 = &two,
1127 },
1128 {
1129 .procname = "panic_on_oom",
1130 .data = &sysctl_panic_on_oom,
1131 .maxlen = sizeof(sysctl_panic_on_oom),
1132 .mode = 0644,
1133 .proc_handler = proc_dointvec_minmax,
1134 .extra1 = &zero,
1135 .extra2 = &two,
1136 },
1137 {
1138 .procname = "oom_kill_allocating_task",
1139 .data = &sysctl_oom_kill_allocating_task,
1140 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1141 .mode = 0644,
1142 .proc_handler = proc_dointvec,
1143 },
1144 {
1145 .procname = "oom_dump_tasks",
1146 .data = &sysctl_oom_dump_tasks,
1147 .maxlen = sizeof(sysctl_oom_dump_tasks),
1148 .mode = 0644,
1149 .proc_handler = proc_dointvec,
1150 },
1151 {
1152 .procname = "overcommit_ratio",
1153 .data = &sysctl_overcommit_ratio,
1154 .maxlen = sizeof(sysctl_overcommit_ratio),
1155 .mode = 0644,
1156 .proc_handler = overcommit_ratio_handler,
1157 },
1158 {
1159 .procname = "overcommit_kbytes",
1160 .data = &sysctl_overcommit_kbytes,
1161 .maxlen = sizeof(sysctl_overcommit_kbytes),
1162 .mode = 0644,
1163 .proc_handler = overcommit_kbytes_handler,
1164 },
1165 {
1166 .procname = "page-cluster",
1167 .data = &page_cluster,
1168 .maxlen = sizeof(int),
1169 .mode = 0644,
1170 .proc_handler = proc_dointvec_minmax,
1171 .extra1 = &zero,
1172 },
1173 {
1174 .procname = "dirty_background_ratio",
1175 .data = &dirty_background_ratio,
1176 .maxlen = sizeof(dirty_background_ratio),
1177 .mode = 0644,
1178 .proc_handler = dirty_background_ratio_handler,
1179 .extra1 = &zero,
1180 .extra2 = &one_hundred,
1181 },
1182 {
1183 .procname = "dirty_background_bytes",
1184 .data = &dirty_background_bytes,
1185 .maxlen = sizeof(dirty_background_bytes),
1186 .mode = 0644,
1187 .proc_handler = dirty_background_bytes_handler,
1188 .extra1 = &one_ul,
1189 },
1190 {
1191 .procname = "dirty_ratio",
1192 .data = &vm_dirty_ratio,
1193 .maxlen = sizeof(vm_dirty_ratio),
1194 .mode = 0644,
1195 .proc_handler = dirty_ratio_handler,
1196 .extra1 = &zero,
1197 .extra2 = &one_hundred,
1198 },
1199 {
1200 .procname = "dirty_bytes",
1201 .data = &vm_dirty_bytes,
1202 .maxlen = sizeof(vm_dirty_bytes),
1203 .mode = 0644,
1204 .proc_handler = dirty_bytes_handler,
1205 .extra1 = &dirty_bytes_min,
1206 },
1207 {
1208 .procname = "dirty_writeback_centisecs",
1209 .data = &dirty_writeback_interval,
1210 .maxlen = sizeof(dirty_writeback_interval),
1211 .mode = 0644,
1212 .proc_handler = dirty_writeback_centisecs_handler,
1213 },
1214 {
1215 .procname = "dirty_expire_centisecs",
1216 .data = &dirty_expire_interval,
1217 .maxlen = sizeof(dirty_expire_interval),
1218 .mode = 0644,
1219 .proc_handler = proc_dointvec_minmax,
1220 .extra1 = &zero,
1221 },
1222 {
1223 .procname = "nr_pdflush_threads",
1224 .mode = 0444 /* read-only */,
1225 .proc_handler = pdflush_proc_obsolete,
1226 },
1227 {
1228 .procname = "swappiness",
1229 .data = &vm_swappiness,
1230 .maxlen = sizeof(vm_swappiness),
1231 .mode = 0644,
1232 .proc_handler = proc_dointvec_minmax,
1233 .extra1 = &zero,
1234 .extra2 = &one_hundred,
1235 },
1236 #ifdef CONFIG_HUGETLB_PAGE
1237 {
1238 .procname = "nr_hugepages",
1239 .data = NULL,
1240 .maxlen = sizeof(unsigned long),
1241 .mode = 0644,
1242 .proc_handler = hugetlb_sysctl_handler,
1243 .extra1 = (void *)&hugetlb_zero,
1244 .extra2 = (void *)&hugetlb_infinity,
1245 },
1246 #ifdef CONFIG_NUMA
1247 {
1248 .procname = "nr_hugepages_mempolicy",
1249 .data = NULL,
1250 .maxlen = sizeof(unsigned long),
1251 .mode = 0644,
1252 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1253 .extra1 = (void *)&hugetlb_zero,
1254 .extra2 = (void *)&hugetlb_infinity,
1255 },
1256 #endif
1257 {
1258 .procname = "hugetlb_shm_group",
1259 .data = &sysctl_hugetlb_shm_group,
1260 .maxlen = sizeof(gid_t),
1261 .mode = 0644,
1262 .proc_handler = proc_dointvec,
1263 },
1264 {
1265 .procname = "hugepages_treat_as_movable",
1266 .data = &hugepages_treat_as_movable,
1267 .maxlen = sizeof(int),
1268 .mode = 0644,
1269 .proc_handler = proc_dointvec,
1270 },
1271 {
1272 .procname = "nr_overcommit_hugepages",
1273 .data = NULL,
1274 .maxlen = sizeof(unsigned long),
1275 .mode = 0644,
1276 .proc_handler = hugetlb_overcommit_handler,
1277 .extra1 = (void *)&hugetlb_zero,
1278 .extra2 = (void *)&hugetlb_infinity,
1279 },
1280 #endif
1281 {
1282 .procname = "lowmem_reserve_ratio",
1283 .data = &sysctl_lowmem_reserve_ratio,
1284 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1285 .mode = 0644,
1286 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1287 },
1288 {
1289 .procname = "drop_caches",
1290 .data = &sysctl_drop_caches,
1291 .maxlen = sizeof(int),
1292 .mode = 0644,
1293 .proc_handler = drop_caches_sysctl_handler,
1294 .extra1 = &one,
1295 .extra2 = &four,
1296 },
1297 #ifdef CONFIG_COMPACTION
1298 {
1299 .procname = "compact_memory",
1300 .data = &sysctl_compact_memory,
1301 .maxlen = sizeof(int),
1302 .mode = 0200,
1303 .proc_handler = sysctl_compaction_handler,
1304 },
1305 {
1306 .procname = "extfrag_threshold",
1307 .data = &sysctl_extfrag_threshold,
1308 .maxlen = sizeof(int),
1309 .mode = 0644,
1310 .proc_handler = sysctl_extfrag_handler,
1311 .extra1 = &min_extfrag_threshold,
1312 .extra2 = &max_extfrag_threshold,
1313 },
1314
1315 #endif /* CONFIG_COMPACTION */
1316 {
1317 .procname = "min_free_kbytes",
1318 .data = &min_free_kbytes,
1319 .maxlen = sizeof(min_free_kbytes),
1320 .mode = 0644,
1321 .proc_handler = min_free_kbytes_sysctl_handler,
1322 .extra1 = &zero,
1323 },
1324 {
1325 .procname = "percpu_pagelist_fraction",
1326 .data = &percpu_pagelist_fraction,
1327 .maxlen = sizeof(percpu_pagelist_fraction),
1328 .mode = 0644,
1329 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1330 .extra1 = &zero,
1331 },
1332 #ifdef CONFIG_MMU
1333 {
1334 .procname = "max_map_count",
1335 .data = &sysctl_max_map_count,
1336 .maxlen = sizeof(sysctl_max_map_count),
1337 .mode = 0644,
1338 .proc_handler = proc_dointvec_minmax,
1339 .extra1 = &zero,
1340 },
1341 #else
1342 {
1343 .procname = "nr_trim_pages",
1344 .data = &sysctl_nr_trim_pages,
1345 .maxlen = sizeof(sysctl_nr_trim_pages),
1346 .mode = 0644,
1347 .proc_handler = proc_dointvec_minmax,
1348 .extra1 = &zero,
1349 },
1350 #endif
1351 {
1352 .procname = "laptop_mode",
1353 .data = &laptop_mode,
1354 .maxlen = sizeof(laptop_mode),
1355 .mode = 0644,
1356 .proc_handler = proc_dointvec_jiffies,
1357 },
1358 {
1359 .procname = "block_dump",
1360 .data = &block_dump,
1361 .maxlen = sizeof(block_dump),
1362 .mode = 0644,
1363 .proc_handler = proc_dointvec,
1364 .extra1 = &zero,
1365 },
1366 {
1367 .procname = "vfs_cache_pressure",
1368 .data = &sysctl_vfs_cache_pressure,
1369 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1370 .mode = 0644,
1371 .proc_handler = proc_dointvec,
1372 .extra1 = &zero,
1373 },
1374 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1375 {
1376 .procname = "legacy_va_layout",
1377 .data = &sysctl_legacy_va_layout,
1378 .maxlen = sizeof(sysctl_legacy_va_layout),
1379 .mode = 0644,
1380 .proc_handler = proc_dointvec,
1381 .extra1 = &zero,
1382 },
1383 #endif
1384 #ifdef CONFIG_NUMA
1385 {
1386 .procname = "zone_reclaim_mode",
1387 .data = &zone_reclaim_mode,
1388 .maxlen = sizeof(zone_reclaim_mode),
1389 .mode = 0644,
1390 .proc_handler = proc_dointvec,
1391 .extra1 = &zero,
1392 },
1393 {
1394 .procname = "min_unmapped_ratio",
1395 .data = &sysctl_min_unmapped_ratio,
1396 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1397 .mode = 0644,
1398 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1399 .extra1 = &zero,
1400 .extra2 = &one_hundred,
1401 },
1402 {
1403 .procname = "min_slab_ratio",
1404 .data = &sysctl_min_slab_ratio,
1405 .maxlen = sizeof(sysctl_min_slab_ratio),
1406 .mode = 0644,
1407 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1408 .extra1 = &zero,
1409 .extra2 = &one_hundred,
1410 },
1411 #endif
1412 #ifdef CONFIG_SMP
1413 {
1414 .procname = "stat_interval",
1415 .data = &sysctl_stat_interval,
1416 .maxlen = sizeof(sysctl_stat_interval),
1417 .mode = 0644,
1418 .proc_handler = proc_dointvec_jiffies,
1419 },
1420 #endif
1421 #ifdef CONFIG_MMU
1422 {
1423 .procname = "mmap_min_addr",
1424 .data = &dac_mmap_min_addr,
1425 .maxlen = sizeof(unsigned long),
1426 .mode = 0644,
1427 .proc_handler = mmap_min_addr_handler,
1428 },
1429 #endif
1430 #ifdef CONFIG_NUMA
1431 {
1432 .procname = "numa_zonelist_order",
1433 .data = &numa_zonelist_order,
1434 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1435 .mode = 0644,
1436 .proc_handler = numa_zonelist_order_handler,
1437 },
1438 #endif
1439 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1440 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1441 {
1442 .procname = "vdso_enabled",
1443 #ifdef CONFIG_X86_32
1444 .data = &vdso32_enabled,
1445 .maxlen = sizeof(vdso32_enabled),
1446 #else
1447 .data = &vdso_enabled,
1448 .maxlen = sizeof(vdso_enabled),
1449 #endif
1450 .mode = 0644,
1451 .proc_handler = proc_dointvec,
1452 .extra1 = &zero,
1453 },
1454 #endif
1455 #ifdef CONFIG_HIGHMEM
1456 {
1457 .procname = "highmem_is_dirtyable",
1458 .data = &vm_highmem_is_dirtyable,
1459 .maxlen = sizeof(vm_highmem_is_dirtyable),
1460 .mode = 0644,
1461 .proc_handler = proc_dointvec_minmax,
1462 .extra1 = &zero,
1463 .extra2 = &one,
1464 },
1465 #endif
1466 {
1467 .procname = "scan_unevictable_pages",
1468 .data = &scan_unevictable_pages,
1469 .maxlen = sizeof(scan_unevictable_pages),
1470 .mode = 0644,
1471 .proc_handler = scan_unevictable_handler,
1472 },
1473 #ifdef CONFIG_MEMORY_FAILURE
1474 {
1475 .procname = "memory_failure_early_kill",
1476 .data = &sysctl_memory_failure_early_kill,
1477 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1478 .mode = 0644,
1479 .proc_handler = proc_dointvec_minmax,
1480 .extra1 = &zero,
1481 .extra2 = &one,
1482 },
1483 {
1484 .procname = "memory_failure_recovery",
1485 .data = &sysctl_memory_failure_recovery,
1486 .maxlen = sizeof(sysctl_memory_failure_recovery),
1487 .mode = 0644,
1488 .proc_handler = proc_dointvec_minmax,
1489 .extra1 = &zero,
1490 .extra2 = &one,
1491 },
1492 #endif
1493 {
1494 .procname = "user_reserve_kbytes",
1495 .data = &sysctl_user_reserve_kbytes,
1496 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1497 .mode = 0644,
1498 .proc_handler = proc_doulongvec_minmax,
1499 },
1500 {
1501 .procname = "admin_reserve_kbytes",
1502 .data = &sysctl_admin_reserve_kbytes,
1503 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1504 .mode = 0644,
1505 .proc_handler = proc_doulongvec_minmax,
1506 },
1507 { }
1508 };
1509
1510 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1511 static struct ctl_table binfmt_misc_table[] = {
1512 { }
1513 };
1514 #endif
1515
1516 static struct ctl_table fs_table[] = {
1517 {
1518 .procname = "inode-nr",
1519 .data = &inodes_stat,
1520 .maxlen = 2*sizeof(long),
1521 .mode = 0444,
1522 .proc_handler = proc_nr_inodes,
1523 },
1524 {
1525 .procname = "inode-state",
1526 .data = &inodes_stat,
1527 .maxlen = 7*sizeof(long),
1528 .mode = 0444,
1529 .proc_handler = proc_nr_inodes,
1530 },
1531 {
1532 .procname = "file-nr",
1533 .data = &files_stat,
1534 .maxlen = sizeof(files_stat),
1535 .mode = 0444,
1536 .proc_handler = proc_nr_files,
1537 },
1538 {
1539 .procname = "file-max",
1540 .data = &files_stat.max_files,
1541 .maxlen = sizeof(files_stat.max_files),
1542 .mode = 0644,
1543 .proc_handler = proc_doulongvec_minmax,
1544 },
1545 {
1546 .procname = "nr_open",
1547 .data = &sysctl_nr_open,
1548 .maxlen = sizeof(int),
1549 .mode = 0644,
1550 .proc_handler = proc_dointvec_minmax,
1551 .extra1 = &sysctl_nr_open_min,
1552 .extra2 = &sysctl_nr_open_max,
1553 },
1554 {
1555 .procname = "dentry-state",
1556 .data = &dentry_stat,
1557 .maxlen = 6*sizeof(long),
1558 .mode = 0444,
1559 .proc_handler = proc_nr_dentry,
1560 },
1561 {
1562 .procname = "overflowuid",
1563 .data = &fs_overflowuid,
1564 .maxlen = sizeof(int),
1565 .mode = 0644,
1566 .proc_handler = proc_dointvec_minmax,
1567 .extra1 = &minolduid,
1568 .extra2 = &maxolduid,
1569 },
1570 {
1571 .procname = "overflowgid",
1572 .data = &fs_overflowgid,
1573 .maxlen = sizeof(int),
1574 .mode = 0644,
1575 .proc_handler = proc_dointvec_minmax,
1576 .extra1 = &minolduid,
1577 .extra2 = &maxolduid,
1578 },
1579 #ifdef CONFIG_FILE_LOCKING
1580 {
1581 .procname = "leases-enable",
1582 .data = &leases_enable,
1583 .maxlen = sizeof(int),
1584 .mode = 0644,
1585 .proc_handler = proc_dointvec,
1586 },
1587 #endif
1588 #ifdef CONFIG_DNOTIFY
1589 {
1590 .procname = "dir-notify-enable",
1591 .data = &dir_notify_enable,
1592 .maxlen = sizeof(int),
1593 .mode = 0644,
1594 .proc_handler = proc_dointvec,
1595 },
1596 #endif
1597 #ifdef CONFIG_MMU
1598 #ifdef CONFIG_FILE_LOCKING
1599 {
1600 .procname = "lease-break-time",
1601 .data = &lease_break_time,
1602 .maxlen = sizeof(int),
1603 .mode = 0644,
1604 .proc_handler = proc_dointvec,
1605 },
1606 #endif
1607 #ifdef CONFIG_AIO
1608 {
1609 .procname = "aio-nr",
1610 .data = &aio_nr,
1611 .maxlen = sizeof(aio_nr),
1612 .mode = 0444,
1613 .proc_handler = proc_doulongvec_minmax,
1614 },
1615 {
1616 .procname = "aio-max-nr",
1617 .data = &aio_max_nr,
1618 .maxlen = sizeof(aio_max_nr),
1619 .mode = 0644,
1620 .proc_handler = proc_doulongvec_minmax,
1621 },
1622 #endif /* CONFIG_AIO */
1623 #ifdef CONFIG_INOTIFY_USER
1624 {
1625 .procname = "inotify",
1626 .mode = 0555,
1627 .child = inotify_table,
1628 },
1629 #endif
1630 #ifdef CONFIG_EPOLL
1631 {
1632 .procname = "epoll",
1633 .mode = 0555,
1634 .child = epoll_table,
1635 },
1636 #endif
1637 #endif
1638 {
1639 .procname = "protected_symlinks",
1640 .data = &sysctl_protected_symlinks,
1641 .maxlen = sizeof(int),
1642 .mode = 0600,
1643 .proc_handler = proc_dointvec_minmax,
1644 .extra1 = &zero,
1645 .extra2 = &one,
1646 },
1647 {
1648 .procname = "protected_hardlinks",
1649 .data = &sysctl_protected_hardlinks,
1650 .maxlen = sizeof(int),
1651 .mode = 0600,
1652 .proc_handler = proc_dointvec_minmax,
1653 .extra1 = &zero,
1654 .extra2 = &one,
1655 },
1656 {
1657 .procname = "suid_dumpable",
1658 .data = &suid_dumpable,
1659 .maxlen = sizeof(int),
1660 .mode = 0644,
1661 .proc_handler = proc_dointvec_minmax_coredump,
1662 .extra1 = &zero,
1663 .extra2 = &two,
1664 },
1665 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1666 {
1667 .procname = "binfmt_misc",
1668 .mode = 0555,
1669 .child = binfmt_misc_table,
1670 },
1671 #endif
1672 {
1673 .procname = "pipe-max-size",
1674 .data = &pipe_max_size,
1675 .maxlen = sizeof(int),
1676 .mode = 0644,
1677 .proc_handler = &pipe_proc_fn,
1678 .extra1 = &pipe_min_size,
1679 },
1680 { }
1681 };
1682
1683 static struct ctl_table debug_table[] = {
1684 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1685 {
1686 .procname = "exception-trace",
1687 .data = &show_unhandled_signals,
1688 .maxlen = sizeof(int),
1689 .mode = 0644,
1690 .proc_handler = proc_dointvec
1691 },
1692 #endif
1693 #if defined(CONFIG_OPTPROBES)
1694 {
1695 .procname = "kprobes-optimization",
1696 .data = &sysctl_kprobes_optimization,
1697 .maxlen = sizeof(int),
1698 .mode = 0644,
1699 .proc_handler = proc_kprobes_optimization_handler,
1700 .extra1 = &zero,
1701 .extra2 = &one,
1702 },
1703 #endif
1704 { }
1705 };
1706
1707 static struct ctl_table dev_table[] = {
1708 { }
1709 };
1710
1711 int __init sysctl_init(void)
1712 {
1713 struct ctl_table_header *hdr;
1714
1715 hdr = register_sysctl_table(sysctl_base_table);
1716 kmemleak_not_leak(hdr);
1717 return 0;
1718 }
1719
1720 #endif /* CONFIG_SYSCTL */
1721
1722 /*
1723 * /proc/sys support
1724 */
1725
1726 #ifdef CONFIG_PROC_SYSCTL
1727
1728 static int _proc_do_string(char *data, int maxlen, int write,
1729 char __user *buffer,
1730 size_t *lenp, loff_t *ppos)
1731 {
1732 size_t len;
1733 char __user *p;
1734 char c;
1735
1736 if (!data || !maxlen || !*lenp) {
1737 *lenp = 0;
1738 return 0;
1739 }
1740
1741 if (write) {
1742 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1743 /* Only continue writes not past the end of buffer. */
1744 len = strlen(data);
1745 if (len > maxlen - 1)
1746 len = maxlen - 1;
1747
1748 if (*ppos > len)
1749 return 0;
1750 len = *ppos;
1751 } else {
1752 /* Start writing from beginning of buffer. */
1753 len = 0;
1754 }
1755
1756 *ppos += *lenp;
1757 p = buffer;
1758 while ((p - buffer) < *lenp && len < maxlen - 1) {
1759 if (get_user(c, p++))
1760 return -EFAULT;
1761 if (c == 0 || c == '\n')
1762 break;
1763 data[len++] = c;
1764 }
1765 data[len] = 0;
1766 } else {
1767 len = strlen(data);
1768 if (len > maxlen)
1769 len = maxlen;
1770
1771 if (*ppos > len) {
1772 *lenp = 0;
1773 return 0;
1774 }
1775
1776 data += *ppos;
1777 len -= *ppos;
1778
1779 if (len > *lenp)
1780 len = *lenp;
1781 if (len)
1782 if (copy_to_user(buffer, data, len))
1783 return -EFAULT;
1784 if (len < *lenp) {
1785 if (put_user('\n', buffer + len))
1786 return -EFAULT;
1787 len++;
1788 }
1789 *lenp = len;
1790 *ppos += len;
1791 }
1792 return 0;
1793 }
1794
1795 static void warn_sysctl_write(struct ctl_table *table)
1796 {
1797 pr_warn_once("%s wrote to %s when file position was not 0!\n"
1798 "This will not be supported in the future. To silence this\n"
1799 "warning, set kernel.sysctl_writes_strict = -1\n",
1800 current->comm, table->procname);
1801 }
1802
1803 /**
1804 * proc_dostring - read a string sysctl
1805 * @table: the sysctl table
1806 * @write: %TRUE if this is a write to the sysctl file
1807 * @buffer: the user buffer
1808 * @lenp: the size of the user buffer
1809 * @ppos: file position
1810 *
1811 * Reads/writes a string from/to the user buffer. If the kernel
1812 * buffer provided is not large enough to hold the string, the
1813 * string is truncated. The copied string is %NULL-terminated.
1814 * If the string is being read by the user process, it is copied
1815 * and a newline '\n' is added. It is truncated if the buffer is
1816 * not large enough.
1817 *
1818 * Returns 0 on success.
1819 */
1820 int proc_dostring(struct ctl_table *table, int write,
1821 void __user *buffer, size_t *lenp, loff_t *ppos)
1822 {
1823 if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN)
1824 warn_sysctl_write(table);
1825
1826 return _proc_do_string((char *)(table->data), table->maxlen, write,
1827 (char __user *)buffer, lenp, ppos);
1828 }
1829
1830 static size_t proc_skip_spaces(char **buf)
1831 {
1832 size_t ret;
1833 char *tmp = skip_spaces(*buf);
1834 ret = tmp - *buf;
1835 *buf = tmp;
1836 return ret;
1837 }
1838
1839 static void proc_skip_char(char **buf, size_t *size, const char v)
1840 {
1841 while (*size) {
1842 if (**buf != v)
1843 break;
1844 (*size)--;
1845 (*buf)++;
1846 }
1847 }
1848
1849 #define TMPBUFLEN 22
1850 /**
1851 * proc_get_long - reads an ASCII formatted integer from a user buffer
1852 *
1853 * @buf: a kernel buffer
1854 * @size: size of the kernel buffer
1855 * @val: this is where the number will be stored
1856 * @neg: set to %TRUE if number is negative
1857 * @perm_tr: a vector which contains the allowed trailers
1858 * @perm_tr_len: size of the perm_tr vector
1859 * @tr: pointer to store the trailer character
1860 *
1861 * In case of success %0 is returned and @buf and @size are updated with
1862 * the amount of bytes read. If @tr is non-NULL and a trailing
1863 * character exists (size is non-zero after returning from this
1864 * function), @tr is updated with the trailing character.
1865 */
1866 static int proc_get_long(char **buf, size_t *size,
1867 unsigned long *val, bool *neg,
1868 const char *perm_tr, unsigned perm_tr_len, char *tr)
1869 {
1870 int len;
1871 char *p, tmp[TMPBUFLEN];
1872
1873 if (!*size)
1874 return -EINVAL;
1875
1876 len = *size;
1877 if (len > TMPBUFLEN - 1)
1878 len = TMPBUFLEN - 1;
1879
1880 memcpy(tmp, *buf, len);
1881
1882 tmp[len] = 0;
1883 p = tmp;
1884 if (*p == '-' && *size > 1) {
1885 *neg = true;
1886 p++;
1887 } else
1888 *neg = false;
1889 if (!isdigit(*p))
1890 return -EINVAL;
1891
1892 *val = simple_strtoul(p, &p, 0);
1893
1894 len = p - tmp;
1895
1896 /* We don't know if the next char is whitespace thus we may accept
1897 * invalid integers (e.g. 1234...a) or two integers instead of one
1898 * (e.g. 123...1). So lets not allow such large numbers. */
1899 if (len == TMPBUFLEN - 1)
1900 return -EINVAL;
1901
1902 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
1903 return -EINVAL;
1904
1905 if (tr && (len < *size))
1906 *tr = *p;
1907
1908 *buf += len;
1909 *size -= len;
1910
1911 return 0;
1912 }
1913
1914 /**
1915 * proc_put_long - converts an integer to a decimal ASCII formatted string
1916 *
1917 * @buf: the user buffer
1918 * @size: the size of the user buffer
1919 * @val: the integer to be converted
1920 * @neg: sign of the number, %TRUE for negative
1921 *
1922 * In case of success %0 is returned and @buf and @size are updated with
1923 * the amount of bytes written.
1924 */
1925 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
1926 bool neg)
1927 {
1928 int len;
1929 char tmp[TMPBUFLEN], *p = tmp;
1930
1931 sprintf(p, "%s%lu", neg ? "-" : "", val);
1932 len = strlen(tmp);
1933 if (len > *size)
1934 len = *size;
1935 if (copy_to_user(*buf, tmp, len))
1936 return -EFAULT;
1937 *size -= len;
1938 *buf += len;
1939 return 0;
1940 }
1941 #undef TMPBUFLEN
1942
1943 static int proc_put_char(void __user **buf, size_t *size, char c)
1944 {
1945 if (*size) {
1946 char __user **buffer = (char __user **)buf;
1947 if (put_user(c, *buffer))
1948 return -EFAULT;
1949 (*size)--, (*buffer)++;
1950 *buf = *buffer;
1951 }
1952 return 0;
1953 }
1954
1955 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
1956 int *valp,
1957 int write, void *data)
1958 {
1959 if (write) {
1960 *valp = *negp ? -*lvalp : *lvalp;
1961 } else {
1962 int val = *valp;
1963 if (val < 0) {
1964 *negp = true;
1965 *lvalp = (unsigned long)-val;
1966 } else {
1967 *negp = false;
1968 *lvalp = (unsigned long)val;
1969 }
1970 }
1971 return 0;
1972 }
1973
1974 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
1975
1976 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1977 int write, void __user *buffer,
1978 size_t *lenp, loff_t *ppos,
1979 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1980 int write, void *data),
1981 void *data)
1982 {
1983 int *i, vleft, first = 1, err = 0;
1984 unsigned long page = 0;
1985 size_t left;
1986 char *kbuf;
1987
1988 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
1989 *lenp = 0;
1990 return 0;
1991 }
1992
1993 i = (int *) tbl_data;
1994 vleft = table->maxlen / sizeof(*i);
1995 left = *lenp;
1996
1997 if (!conv)
1998 conv = do_proc_dointvec_conv;
1999
2000 if (write) {
2001 if (*ppos) {
2002 switch (sysctl_writes_strict) {
2003 case SYSCTL_WRITES_STRICT:
2004 goto out;
2005 case SYSCTL_WRITES_WARN:
2006 warn_sysctl_write(table);
2007 break;
2008 default:
2009 break;
2010 }
2011 }
2012
2013 if (left > PAGE_SIZE - 1)
2014 left = PAGE_SIZE - 1;
2015 page = __get_free_page(GFP_TEMPORARY);
2016 kbuf = (char *) page;
2017 if (!kbuf)
2018 return -ENOMEM;
2019 if (copy_from_user(kbuf, buffer, left)) {
2020 err = -EFAULT;
2021 goto free;
2022 }
2023 kbuf[left] = 0;
2024 }
2025
2026 for (; left && vleft--; i++, first=0) {
2027 unsigned long lval;
2028 bool neg;
2029
2030 if (write) {
2031 left -= proc_skip_spaces(&kbuf);
2032
2033 if (!left)
2034 break;
2035 err = proc_get_long(&kbuf, &left, &lval, &neg,
2036 proc_wspace_sep,
2037 sizeof(proc_wspace_sep), NULL);
2038 if (err)
2039 break;
2040 if (conv(&neg, &lval, i, 1, data)) {
2041 err = -EINVAL;
2042 break;
2043 }
2044 } else {
2045 if (conv(&neg, &lval, i, 0, data)) {
2046 err = -EINVAL;
2047 break;
2048 }
2049 if (!first)
2050 err = proc_put_char(&buffer, &left, '\t');
2051 if (err)
2052 break;
2053 err = proc_put_long(&buffer, &left, lval, neg);
2054 if (err)
2055 break;
2056 }
2057 }
2058
2059 if (!write && !first && left && !err)
2060 err = proc_put_char(&buffer, &left, '\n');
2061 if (write && !err && left)
2062 left -= proc_skip_spaces(&kbuf);
2063 free:
2064 if (write) {
2065 free_page(page);
2066 if (first)
2067 return err ? : -EINVAL;
2068 }
2069 *lenp -= left;
2070 out:
2071 *ppos += *lenp;
2072 return err;
2073 }
2074
2075 static int do_proc_dointvec(struct ctl_table *table, int write,
2076 void __user *buffer, size_t *lenp, loff_t *ppos,
2077 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2078 int write, void *data),
2079 void *data)
2080 {
2081 return __do_proc_dointvec(table->data, table, write,
2082 buffer, lenp, ppos, conv, data);
2083 }
2084
2085 /**
2086 * proc_dointvec - read a vector of integers
2087 * @table: the sysctl table
2088 * @write: %TRUE if this is a write to the sysctl file
2089 * @buffer: the user buffer
2090 * @lenp: the size of the user buffer
2091 * @ppos: file position
2092 *
2093 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2094 * values from/to the user buffer, treated as an ASCII string.
2095 *
2096 * Returns 0 on success.
2097 */
2098 int proc_dointvec(struct ctl_table *table, int write,
2099 void __user *buffer, size_t *lenp, loff_t *ppos)
2100 {
2101 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2102 NULL,NULL);
2103 }
2104
2105 /*
2106 * Taint values can only be increased
2107 * This means we can safely use a temporary.
2108 */
2109 static int proc_taint(struct ctl_table *table, int write,
2110 void __user *buffer, size_t *lenp, loff_t *ppos)
2111 {
2112 struct ctl_table t;
2113 unsigned long tmptaint = get_taint();
2114 int err;
2115
2116 if (write && !capable(CAP_SYS_ADMIN))
2117 return -EPERM;
2118
2119 t = *table;
2120 t.data = &tmptaint;
2121 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2122 if (err < 0)
2123 return err;
2124
2125 if (write) {
2126 /*
2127 * Poor man's atomic or. Not worth adding a primitive
2128 * to everyone's atomic.h for this
2129 */
2130 int i;
2131 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2132 if ((tmptaint >> i) & 1)
2133 add_taint(i, LOCKDEP_STILL_OK);
2134 }
2135 }
2136
2137 return err;
2138 }
2139
2140 #ifdef CONFIG_PRINTK
2141 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2142 void __user *buffer, size_t *lenp, loff_t *ppos)
2143 {
2144 if (write && !capable(CAP_SYS_ADMIN))
2145 return -EPERM;
2146
2147 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2148 }
2149 #endif
2150
2151 struct do_proc_dointvec_minmax_conv_param {
2152 int *min;
2153 int *max;
2154 };
2155
2156 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2157 int *valp,
2158 int write, void *data)
2159 {
2160 struct do_proc_dointvec_minmax_conv_param *param = data;
2161 if (write) {
2162 int val = *negp ? -*lvalp : *lvalp;
2163 if ((param->min && *param->min > val) ||
2164 (param->max && *param->max < val))
2165 return -EINVAL;
2166 *valp = val;
2167 } else {
2168 int val = *valp;
2169 if (val < 0) {
2170 *negp = true;
2171 *lvalp = (unsigned long)-val;
2172 } else {
2173 *negp = false;
2174 *lvalp = (unsigned long)val;
2175 }
2176 }
2177 return 0;
2178 }
2179
2180 /**
2181 * proc_dointvec_minmax - read a vector of integers with min/max values
2182 * @table: the sysctl table
2183 * @write: %TRUE if this is a write to the sysctl file
2184 * @buffer: the user buffer
2185 * @lenp: the size of the user buffer
2186 * @ppos: file position
2187 *
2188 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2189 * values from/to the user buffer, treated as an ASCII string.
2190 *
2191 * This routine will ensure the values are within the range specified by
2192 * table->extra1 (min) and table->extra2 (max).
2193 *
2194 * Returns 0 on success.
2195 */
2196 int proc_dointvec_minmax(struct ctl_table *table, int write,
2197 void __user *buffer, size_t *lenp, loff_t *ppos)
2198 {
2199 struct do_proc_dointvec_minmax_conv_param param = {
2200 .min = (int *) table->extra1,
2201 .max = (int *) table->extra2,
2202 };
2203 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2204 do_proc_dointvec_minmax_conv, &param);
2205 }
2206
2207 static void validate_coredump_safety(void)
2208 {
2209 #ifdef CONFIG_COREDUMP
2210 if (suid_dumpable == SUID_DUMP_ROOT &&
2211 core_pattern[0] != '/' && core_pattern[0] != '|') {
2212 printk(KERN_WARNING "Unsafe core_pattern used with "\
2213 "suid_dumpable=2. Pipe handler or fully qualified "\
2214 "core dump path required.\n");
2215 }
2216 #endif
2217 }
2218
2219 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2220 void __user *buffer, size_t *lenp, loff_t *ppos)
2221 {
2222 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2223 if (!error)
2224 validate_coredump_safety();
2225 return error;
2226 }
2227
2228 #ifdef CONFIG_COREDUMP
2229 static int proc_dostring_coredump(struct ctl_table *table, int write,
2230 void __user *buffer, size_t *lenp, loff_t *ppos)
2231 {
2232 int error = proc_dostring(table, write, buffer, lenp, ppos);
2233 if (!error)
2234 validate_coredump_safety();
2235 return error;
2236 }
2237 #endif
2238
2239 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2240 void __user *buffer,
2241 size_t *lenp, loff_t *ppos,
2242 unsigned long convmul,
2243 unsigned long convdiv)
2244 {
2245 unsigned long *i, *min, *max;
2246 int vleft, first = 1, err = 0;
2247 unsigned long page = 0;
2248 size_t left;
2249 char *kbuf;
2250
2251 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2252 *lenp = 0;
2253 return 0;
2254 }
2255
2256 i = (unsigned long *) data;
2257 min = (unsigned long *) table->extra1;
2258 max = (unsigned long *) table->extra2;
2259 vleft = table->maxlen / sizeof(unsigned long);
2260 left = *lenp;
2261
2262 if (write) {
2263 if (*ppos) {
2264 switch (sysctl_writes_strict) {
2265 case SYSCTL_WRITES_STRICT:
2266 goto out;
2267 case SYSCTL_WRITES_WARN:
2268 warn_sysctl_write(table);
2269 break;
2270 default:
2271 break;
2272 }
2273 }
2274
2275 if (left > PAGE_SIZE - 1)
2276 left = PAGE_SIZE - 1;
2277 page = __get_free_page(GFP_TEMPORARY);
2278 kbuf = (char *) page;
2279 if (!kbuf)
2280 return -ENOMEM;
2281 if (copy_from_user(kbuf, buffer, left)) {
2282 err = -EFAULT;
2283 goto free;
2284 }
2285 kbuf[left] = 0;
2286 }
2287
2288 for (; left && vleft--; i++, first = 0) {
2289 unsigned long val;
2290
2291 if (write) {
2292 bool neg;
2293
2294 left -= proc_skip_spaces(&kbuf);
2295
2296 err = proc_get_long(&kbuf, &left, &val, &neg,
2297 proc_wspace_sep,
2298 sizeof(proc_wspace_sep), NULL);
2299 if (err)
2300 break;
2301 if (neg)
2302 continue;
2303 if ((min && val < *min) || (max && val > *max))
2304 continue;
2305 *i = val;
2306 } else {
2307 val = convdiv * (*i) / convmul;
2308 if (!first) {
2309 err = proc_put_char(&buffer, &left, '\t');
2310 if (err)
2311 break;
2312 }
2313 err = proc_put_long(&buffer, &left, val, false);
2314 if (err)
2315 break;
2316 }
2317 }
2318
2319 if (!write && !first && left && !err)
2320 err = proc_put_char(&buffer, &left, '\n');
2321 if (write && !err)
2322 left -= proc_skip_spaces(&kbuf);
2323 free:
2324 if (write) {
2325 free_page(page);
2326 if (first)
2327 return err ? : -EINVAL;
2328 }
2329 *lenp -= left;
2330 out:
2331 *ppos += *lenp;
2332 return err;
2333 }
2334
2335 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2336 void __user *buffer,
2337 size_t *lenp, loff_t *ppos,
2338 unsigned long convmul,
2339 unsigned long convdiv)
2340 {
2341 return __do_proc_doulongvec_minmax(table->data, table, write,
2342 buffer, lenp, ppos, convmul, convdiv);
2343 }
2344
2345 /**
2346 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2347 * @table: the sysctl table
2348 * @write: %TRUE if this is a write to the sysctl file
2349 * @buffer: the user buffer
2350 * @lenp: the size of the user buffer
2351 * @ppos: file position
2352 *
2353 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2354 * values from/to the user buffer, treated as an ASCII string.
2355 *
2356 * This routine will ensure the values are within the range specified by
2357 * table->extra1 (min) and table->extra2 (max).
2358 *
2359 * Returns 0 on success.
2360 */
2361 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2362 void __user *buffer, size_t *lenp, loff_t *ppos)
2363 {
2364 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2365 }
2366
2367 /**
2368 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2369 * @table: the sysctl table
2370 * @write: %TRUE if this is a write to the sysctl file
2371 * @buffer: the user buffer
2372 * @lenp: the size of the user buffer
2373 * @ppos: file position
2374 *
2375 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2376 * values from/to the user buffer, treated as an ASCII string. The values
2377 * are treated as milliseconds, and converted to jiffies when they are stored.
2378 *
2379 * This routine will ensure the values are within the range specified by
2380 * table->extra1 (min) and table->extra2 (max).
2381 *
2382 * Returns 0 on success.
2383 */
2384 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2385 void __user *buffer,
2386 size_t *lenp, loff_t *ppos)
2387 {
2388 return do_proc_doulongvec_minmax(table, write, buffer,
2389 lenp, ppos, HZ, 1000l);
2390 }
2391
2392
2393 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2394 int *valp,
2395 int write, void *data)
2396 {
2397 if (write) {
2398 if (*lvalp > LONG_MAX / HZ)
2399 return 1;
2400 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2401 } else {
2402 int val = *valp;
2403 unsigned long lval;
2404 if (val < 0) {
2405 *negp = true;
2406 lval = (unsigned long)-val;
2407 } else {
2408 *negp = false;
2409 lval = (unsigned long)val;
2410 }
2411 *lvalp = lval / HZ;
2412 }
2413 return 0;
2414 }
2415
2416 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2417 int *valp,
2418 int write, void *data)
2419 {
2420 if (write) {
2421 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2422 return 1;
2423 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2424 } else {
2425 int val = *valp;
2426 unsigned long lval;
2427 if (val < 0) {
2428 *negp = true;
2429 lval = (unsigned long)-val;
2430 } else {
2431 *negp = false;
2432 lval = (unsigned long)val;
2433 }
2434 *lvalp = jiffies_to_clock_t(lval);
2435 }
2436 return 0;
2437 }
2438
2439 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2440 int *valp,
2441 int write, void *data)
2442 {
2443 if (write) {
2444 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2445
2446 if (jif > INT_MAX)
2447 return 1;
2448 *valp = (int)jif;
2449 } else {
2450 int val = *valp;
2451 unsigned long lval;
2452 if (val < 0) {
2453 *negp = true;
2454 lval = (unsigned long)-val;
2455 } else {
2456 *negp = false;
2457 lval = (unsigned long)val;
2458 }
2459 *lvalp = jiffies_to_msecs(lval);
2460 }
2461 return 0;
2462 }
2463
2464 /**
2465 * proc_dointvec_jiffies - read a vector of integers as seconds
2466 * @table: the sysctl table
2467 * @write: %TRUE if this is a write to the sysctl file
2468 * @buffer: the user buffer
2469 * @lenp: the size of the user buffer
2470 * @ppos: file position
2471 *
2472 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2473 * values from/to the user buffer, treated as an ASCII string.
2474 * The values read are assumed to be in seconds, and are converted into
2475 * jiffies.
2476 *
2477 * Returns 0 on success.
2478 */
2479 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2480 void __user *buffer, size_t *lenp, loff_t *ppos)
2481 {
2482 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2483 do_proc_dointvec_jiffies_conv,NULL);
2484 }
2485
2486 /**
2487 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2488 * @table: the sysctl table
2489 * @write: %TRUE if this is a write to the sysctl file
2490 * @buffer: the user buffer
2491 * @lenp: the size of the user buffer
2492 * @ppos: pointer to the file position
2493 *
2494 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2495 * values from/to the user buffer, treated as an ASCII string.
2496 * The values read are assumed to be in 1/USER_HZ seconds, and
2497 * are converted into jiffies.
2498 *
2499 * Returns 0 on success.
2500 */
2501 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2502 void __user *buffer, size_t *lenp, loff_t *ppos)
2503 {
2504 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2505 do_proc_dointvec_userhz_jiffies_conv,NULL);
2506 }
2507
2508 /**
2509 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2510 * @table: the sysctl table
2511 * @write: %TRUE if this is a write to the sysctl file
2512 * @buffer: the user buffer
2513 * @lenp: the size of the user buffer
2514 * @ppos: file position
2515 * @ppos: the current position in the file
2516 *
2517 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2518 * values from/to the user buffer, treated as an ASCII string.
2519 * The values read are assumed to be in 1/1000 seconds, and
2520 * are converted into jiffies.
2521 *
2522 * Returns 0 on success.
2523 */
2524 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2525 void __user *buffer, size_t *lenp, loff_t *ppos)
2526 {
2527 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2528 do_proc_dointvec_ms_jiffies_conv, NULL);
2529 }
2530
2531 static int proc_do_cad_pid(struct ctl_table *table, int write,
2532 void __user *buffer, size_t *lenp, loff_t *ppos)
2533 {
2534 struct pid *new_pid;
2535 pid_t tmp;
2536 int r;
2537
2538 tmp = pid_vnr(cad_pid);
2539
2540 r = __do_proc_dointvec(&tmp, table, write, buffer,
2541 lenp, ppos, NULL, NULL);
2542 if (r || !write)
2543 return r;
2544
2545 new_pid = find_get_pid(tmp);
2546 if (!new_pid)
2547 return -ESRCH;
2548
2549 put_pid(xchg(&cad_pid, new_pid));
2550 return 0;
2551 }
2552
2553 /**
2554 * proc_do_large_bitmap - read/write from/to a large bitmap
2555 * @table: the sysctl table
2556 * @write: %TRUE if this is a write to the sysctl file
2557 * @buffer: the user buffer
2558 * @lenp: the size of the user buffer
2559 * @ppos: file position
2560 *
2561 * The bitmap is stored at table->data and the bitmap length (in bits)
2562 * in table->maxlen.
2563 *
2564 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2565 * large bitmaps may be represented in a compact manner. Writing into
2566 * the file will clear the bitmap then update it with the given input.
2567 *
2568 * Returns 0 on success.
2569 */
2570 int proc_do_large_bitmap(struct ctl_table *table, int write,
2571 void __user *buffer, size_t *lenp, loff_t *ppos)
2572 {
2573 int err = 0;
2574 bool first = 1;
2575 size_t left = *lenp;
2576 unsigned long bitmap_len = table->maxlen;
2577 unsigned long *bitmap = *(unsigned long **) table->data;
2578 unsigned long *tmp_bitmap = NULL;
2579 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2580
2581 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2582 *lenp = 0;
2583 return 0;
2584 }
2585
2586 if (write) {
2587 unsigned long page = 0;
2588 char *kbuf;
2589
2590 if (left > PAGE_SIZE - 1)
2591 left = PAGE_SIZE - 1;
2592
2593 page = __get_free_page(GFP_TEMPORARY);
2594 kbuf = (char *) page;
2595 if (!kbuf)
2596 return -ENOMEM;
2597 if (copy_from_user(kbuf, buffer, left)) {
2598 free_page(page);
2599 return -EFAULT;
2600 }
2601 kbuf[left] = 0;
2602
2603 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2604 GFP_KERNEL);
2605 if (!tmp_bitmap) {
2606 free_page(page);
2607 return -ENOMEM;
2608 }
2609 proc_skip_char(&kbuf, &left, '\n');
2610 while (!err && left) {
2611 unsigned long val_a, val_b;
2612 bool neg;
2613
2614 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2615 sizeof(tr_a), &c);
2616 if (err)
2617 break;
2618 if (val_a >= bitmap_len || neg) {
2619 err = -EINVAL;
2620 break;
2621 }
2622
2623 val_b = val_a;
2624 if (left) {
2625 kbuf++;
2626 left--;
2627 }
2628
2629 if (c == '-') {
2630 err = proc_get_long(&kbuf, &left, &val_b,
2631 &neg, tr_b, sizeof(tr_b),
2632 &c);
2633 if (err)
2634 break;
2635 if (val_b >= bitmap_len || neg ||
2636 val_a > val_b) {
2637 err = -EINVAL;
2638 break;
2639 }
2640 if (left) {
2641 kbuf++;
2642 left--;
2643 }
2644 }
2645
2646 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2647 first = 0;
2648 proc_skip_char(&kbuf, &left, '\n');
2649 }
2650 free_page(page);
2651 } else {
2652 unsigned long bit_a, bit_b = 0;
2653
2654 while (left) {
2655 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2656 if (bit_a >= bitmap_len)
2657 break;
2658 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2659 bit_a + 1) - 1;
2660
2661 if (!first) {
2662 err = proc_put_char(&buffer, &left, ',');
2663 if (err)
2664 break;
2665 }
2666 err = proc_put_long(&buffer, &left, bit_a, false);
2667 if (err)
2668 break;
2669 if (bit_a != bit_b) {
2670 err = proc_put_char(&buffer, &left, '-');
2671 if (err)
2672 break;
2673 err = proc_put_long(&buffer, &left, bit_b, false);
2674 if (err)
2675 break;
2676 }
2677
2678 first = 0; bit_b++;
2679 }
2680 if (!err)
2681 err = proc_put_char(&buffer, &left, '\n');
2682 }
2683
2684 if (!err) {
2685 if (write) {
2686 if (*ppos)
2687 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2688 else
2689 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2690 }
2691 kfree(tmp_bitmap);
2692 *lenp -= left;
2693 *ppos += *lenp;
2694 return 0;
2695 } else {
2696 kfree(tmp_bitmap);
2697 return err;
2698 }
2699 }
2700
2701 #else /* CONFIG_PROC_SYSCTL */
2702
2703 int proc_dostring(struct ctl_table *table, int write,
2704 void __user *buffer, size_t *lenp, loff_t *ppos)
2705 {
2706 return -ENOSYS;
2707 }
2708
2709 int proc_dointvec(struct ctl_table *table, int write,
2710 void __user *buffer, size_t *lenp, loff_t *ppos)
2711 {
2712 return -ENOSYS;
2713 }
2714
2715 int proc_dointvec_minmax(struct ctl_table *table, int write,
2716 void __user *buffer, size_t *lenp, loff_t *ppos)
2717 {
2718 return -ENOSYS;
2719 }
2720
2721 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2722 void __user *buffer, size_t *lenp, loff_t *ppos)
2723 {
2724 return -ENOSYS;
2725 }
2726
2727 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2728 void __user *buffer, size_t *lenp, loff_t *ppos)
2729 {
2730 return -ENOSYS;
2731 }
2732
2733 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2734 void __user *buffer, size_t *lenp, loff_t *ppos)
2735 {
2736 return -ENOSYS;
2737 }
2738
2739 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2740 void __user *buffer, size_t *lenp, loff_t *ppos)
2741 {
2742 return -ENOSYS;
2743 }
2744
2745 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2746 void __user *buffer,
2747 size_t *lenp, loff_t *ppos)
2748 {
2749 return -ENOSYS;
2750 }
2751
2752
2753 #endif /* CONFIG_PROC_SYSCTL */
2754
2755 /*
2756 * No sense putting this after each symbol definition, twice,
2757 * exception granted :-)
2758 */
2759 EXPORT_SYMBOL(proc_dointvec);
2760 EXPORT_SYMBOL(proc_dointvec_jiffies);
2761 EXPORT_SYMBOL(proc_dointvec_minmax);
2762 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2763 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2764 EXPORT_SYMBOL(proc_dostring);
2765 EXPORT_SYMBOL(proc_doulongvec_minmax);
2766 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
Linux kernel - Deliverables
This page took 0.095738 seconds and 6 git commands to generate.