projects / deliverable / linux.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Merge branch 'exec_domain_rip_v2' of git://git.kernel.org/pub/scm/linux/kernel/git...
[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/aio.h>
23 #include <linux/mm.h>
24 #include <linux/swap.h>
25 #include <linux/slab.h>
26 #include <linux/sysctl.h>
27 #include <linux/bitmap.h>
28 #include <linux/signal.h>
29 #include <linux/printk.h>
30 #include <linux/proc_fs.h>
31 #include <linux/security.h>
32 #include <linux/ctype.h>
33 #include <linux/kmemcheck.h>
34 #include <linux/kmemleak.h>
35 #include <linux/fs.h>
36 #include <linux/init.h>
37 #include <linux/kernel.h>
38 #include <linux/kobject.h>
39 #include <linux/net.h>
40 #include <linux/sysrq.h>
41 #include <linux/highuid.h>
42 #include <linux/writeback.h>
43 #include <linux/ratelimit.h>
44 #include <linux/compaction.h>
45 #include <linux/hugetlb.h>
46 #include <linux/initrd.h>
47 #include <linux/key.h>
48 #include <linux/times.h>
49 #include <linux/limits.h>
50 #include <linux/dcache.h>
51 #include <linux/dnotify.h>
52 #include <linux/syscalls.h>
53 #include <linux/vmstat.h>
54 #include <linux/nfs_fs.h>
55 #include <linux/acpi.h>
56 #include <linux/reboot.h>
57 #include <linux/ftrace.h>
58 #include <linux/perf_event.h>
59 #include <linux/kprobes.h>
60 #include <linux/pipe_fs_i.h>
61 #include <linux/oom.h>
62 #include <linux/kmod.h>
63 #include <linux/capability.h>
64 #include <linux/binfmts.h>
65 #include <linux/sched/sysctl.h>
66 #include <linux/kexec.h>
67
68 #include <asm/uaccess.h>
69 #include <asm/processor.h>
70
71 #ifdef CONFIG_X86
72 #include <asm/nmi.h>
73 #include <asm/stacktrace.h>
74 #include <asm/io.h>
75 #endif
76 #ifdef CONFIG_SPARC
77 #include <asm/setup.h>
78 #endif
79 #ifdef CONFIG_BSD_PROCESS_ACCT
80 #include <linux/acct.h>
81 #endif
82 #ifdef CONFIG_RT_MUTEXES
83 #include <linux/rtmutex.h>
84 #endif
85 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
86 #include <linux/lockdep.h>
87 #endif
88 #ifdef CONFIG_CHR_DEV_SG
89 #include <scsi/sg.h>
90 #endif
91
92 #ifdef CONFIG_LOCKUP_DETECTOR
93 #include <linux/nmi.h>
94 #endif
95
96
97 #if defined(CONFIG_SYSCTL)
98
99 /* External variables not in a header file. */
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
117 /* Constants used for minimum and maximum */
118 #ifdef CONFIG_LOCKUP_DETECTOR
119 static int sixty = 60;
120 #endif
121
122 static int __maybe_unused neg_one = -1;
123
124 static int zero;
125 static int __maybe_unused one = 1;
126 static int __maybe_unused two = 2;
127 static int __maybe_unused four = 4;
128 static unsigned long one_ul = 1;
129 static int one_hundred = 100;
130 #ifdef CONFIG_PRINTK
131 static int ten_thousand = 10000;
132 #endif
133
134 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
135 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
136
137 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
138 static int maxolduid = 65535;
139 static int minolduid;
140
141 static int ngroups_max = NGROUPS_MAX;
142 static const int cap_last_cap = CAP_LAST_CAP;
143
144 /*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */
145 #ifdef CONFIG_DETECT_HUNG_TASK
146 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
147 #endif
148
149 #ifdef CONFIG_INOTIFY_USER
150 #include <linux/inotify.h>
151 #endif
152 #ifdef CONFIG_SPARC
153 #endif
154
155 #ifdef __hppa__
156 extern int pwrsw_enabled;
157 #endif
158
159 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
160 extern int unaligned_enabled;
161 #endif
162
163 #ifdef CONFIG_IA64
164 extern int unaligned_dump_stack;
165 #endif
166
167 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
168 extern int no_unaligned_warning;
169 #endif
170
171 #ifdef CONFIG_PROC_SYSCTL
172
173 #define SYSCTL_WRITES_LEGACY -1
174 #define SYSCTL_WRITES_WARN 0
175 #define SYSCTL_WRITES_STRICT 1
176
177 static int sysctl_writes_strict = SYSCTL_WRITES_WARN;
178
179 static int proc_do_cad_pid(struct ctl_table *table, int write,
180 void __user *buffer, size_t *lenp, loff_t *ppos);
181 static int proc_taint(struct ctl_table *table, int write,
182 void __user *buffer, size_t *lenp, loff_t *ppos);
183 #endif
184
185 #ifdef CONFIG_PRINTK
186 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
187 void __user *buffer, size_t *lenp, loff_t *ppos);
188 #endif
189
190 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
191 void __user *buffer, size_t *lenp, loff_t *ppos);
192 #ifdef CONFIG_COREDUMP
193 static int proc_dostring_coredump(struct ctl_table *table, int write,
194 void __user *buffer, size_t *lenp, loff_t *ppos);
195 #endif
196
197 #ifdef CONFIG_MAGIC_SYSRQ
198 /* Note: sysrq code uses it's own private copy */
199 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
200
201 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
202 void __user *buffer, size_t *lenp,
203 loff_t *ppos)
204 {
205 int error;
206
207 error = proc_dointvec(table, write, buffer, lenp, ppos);
208 if (error)
209 return error;
210
211 if (write)
212 sysrq_toggle_support(__sysrq_enabled);
213
214 return 0;
215 }
216
217 #endif
218
219 static struct ctl_table kern_table[];
220 static struct ctl_table vm_table[];
221 static struct ctl_table fs_table[];
222 static struct ctl_table debug_table[];
223 static struct ctl_table dev_table[];
224 extern struct ctl_table random_table[];
225 #ifdef CONFIG_EPOLL
226 extern struct ctl_table epoll_table[];
227 #endif
228
229 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
230 int sysctl_legacy_va_layout;
231 #endif
232
233 /* The default sysctl tables: */
234
235 static struct ctl_table sysctl_base_table[] = {
236 {
237 .procname = "kernel",
238 .mode = 0555,
239 .child = kern_table,
240 },
241 {
242 .procname = "vm",
243 .mode = 0555,
244 .child = vm_table,
245 },
246 {
247 .procname = "fs",
248 .mode = 0555,
249 .child = fs_table,
250 },
251 {
252 .procname = "debug",
253 .mode = 0555,
254 .child = debug_table,
255 },
256 {
257 .procname = "dev",
258 .mode = 0555,
259 .child = dev_table,
260 },
261 { }
262 };
263
264 #ifdef CONFIG_SCHED_DEBUG
265 static int min_sched_granularity_ns = 100000; /* 100 usecs */
266 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
267 static int min_wakeup_granularity_ns; /* 0 usecs */
268 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
269 #ifdef CONFIG_SMP
270 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
271 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
272 #endif /* CONFIG_SMP */
273 #endif /* CONFIG_SCHED_DEBUG */
274
275 #ifdef CONFIG_COMPACTION
276 static int min_extfrag_threshold;
277 static int max_extfrag_threshold = 1000;
278 #endif
279
280 static struct ctl_table kern_table[] = {
281 {
282 .procname = "sched_child_runs_first",
283 .data = &sysctl_sched_child_runs_first,
284 .maxlen = sizeof(unsigned int),
285 .mode = 0644,
286 .proc_handler = proc_dointvec,
287 },
288 #ifdef CONFIG_SCHED_DEBUG
289 {
290 .procname = "sched_min_granularity_ns",
291 .data = &sysctl_sched_min_granularity,
292 .maxlen = sizeof(unsigned int),
293 .mode = 0644,
294 .proc_handler = sched_proc_update_handler,
295 .extra1 = &min_sched_granularity_ns,
296 .extra2 = &max_sched_granularity_ns,
297 },
298 {
299 .procname = "sched_latency_ns",
300 .data = &sysctl_sched_latency,
301 .maxlen = sizeof(unsigned int),
302 .mode = 0644,
303 .proc_handler = sched_proc_update_handler,
304 .extra1 = &min_sched_granularity_ns,
305 .extra2 = &max_sched_granularity_ns,
306 },
307 {
308 .procname = "sched_wakeup_granularity_ns",
309 .data = &sysctl_sched_wakeup_granularity,
310 .maxlen = sizeof(unsigned int),
311 .mode = 0644,
312 .proc_handler = sched_proc_update_handler,
313 .extra1 = &min_wakeup_granularity_ns,
314 .extra2 = &max_wakeup_granularity_ns,
315 },
316 #ifdef CONFIG_SMP
317 {
318 .procname = "sched_tunable_scaling",
319 .data = &sysctl_sched_tunable_scaling,
320 .maxlen = sizeof(enum sched_tunable_scaling),
321 .mode = 0644,
322 .proc_handler = sched_proc_update_handler,
323 .extra1 = &min_sched_tunable_scaling,
324 .extra2 = &max_sched_tunable_scaling,
325 },
326 {
327 .procname = "sched_migration_cost_ns",
328 .data = &sysctl_sched_migration_cost,
329 .maxlen = sizeof(unsigned int),
330 .mode = 0644,
331 .proc_handler = proc_dointvec,
332 },
333 {
334 .procname = "sched_nr_migrate",
335 .data = &sysctl_sched_nr_migrate,
336 .maxlen = sizeof(unsigned int),
337 .mode = 0644,
338 .proc_handler = proc_dointvec,
339 },
340 {
341 .procname = "sched_time_avg_ms",
342 .data = &sysctl_sched_time_avg,
343 .maxlen = sizeof(unsigned int),
344 .mode = 0644,
345 .proc_handler = proc_dointvec,
346 },
347 {
348 .procname = "sched_shares_window_ns",
349 .data = &sysctl_sched_shares_window,
350 .maxlen = sizeof(unsigned int),
351 .mode = 0644,
352 .proc_handler = proc_dointvec,
353 },
354 {
355 .procname = "timer_migration",
356 .data = &sysctl_timer_migration,
357 .maxlen = sizeof(unsigned int),
358 .mode = 0644,
359 .proc_handler = proc_dointvec_minmax,
360 .extra1 = &zero,
361 .extra2 = &one,
362 },
363 #endif /* CONFIG_SMP */
364 #ifdef CONFIG_NUMA_BALANCING
365 {
366 .procname = "numa_balancing_scan_delay_ms",
367 .data = &sysctl_numa_balancing_scan_delay,
368 .maxlen = sizeof(unsigned int),
369 .mode = 0644,
370 .proc_handler = proc_dointvec,
371 },
372 {
373 .procname = "numa_balancing_scan_period_min_ms",
374 .data = &sysctl_numa_balancing_scan_period_min,
375 .maxlen = sizeof(unsigned int),
376 .mode = 0644,
377 .proc_handler = proc_dointvec,
378 },
379 {
380 .procname = "numa_balancing_scan_period_max_ms",
381 .data = &sysctl_numa_balancing_scan_period_max,
382 .maxlen = sizeof(unsigned int),
383 .mode = 0644,
384 .proc_handler = proc_dointvec,
385 },
386 {
387 .procname = "numa_balancing_scan_size_mb",
388 .data = &sysctl_numa_balancing_scan_size,
389 .maxlen = sizeof(unsigned int),
390 .mode = 0644,
391 .proc_handler = proc_dointvec_minmax,
392 .extra1 = &one,
393 },
394 {
395 .procname = "numa_balancing",
396 .data = NULL, /* filled in by handler */
397 .maxlen = sizeof(unsigned int),
398 .mode = 0644,
399 .proc_handler = sysctl_numa_balancing,
400 .extra1 = &zero,
401 .extra2 = &one,
402 },
403 #endif /* CONFIG_NUMA_BALANCING */
404 #endif /* CONFIG_SCHED_DEBUG */
405 {
406 .procname = "sched_rt_period_us",
407 .data = &sysctl_sched_rt_period,
408 .maxlen = sizeof(unsigned int),
409 .mode = 0644,
410 .proc_handler = sched_rt_handler,
411 },
412 {
413 .procname = "sched_rt_runtime_us",
414 .data = &sysctl_sched_rt_runtime,
415 .maxlen = sizeof(int),
416 .mode = 0644,
417 .proc_handler = sched_rt_handler,
418 },
419 {
420 .procname = "sched_rr_timeslice_ms",
421 .data = &sched_rr_timeslice,
422 .maxlen = sizeof(int),
423 .mode = 0644,
424 .proc_handler = sched_rr_handler,
425 },
426 #ifdef CONFIG_SCHED_AUTOGROUP
427 {
428 .procname = "sched_autogroup_enabled",
429 .data = &sysctl_sched_autogroup_enabled,
430 .maxlen = sizeof(unsigned int),
431 .mode = 0644,
432 .proc_handler = proc_dointvec_minmax,
433 .extra1 = &zero,
434 .extra2 = &one,
435 },
436 #endif
437 #ifdef CONFIG_CFS_BANDWIDTH
438 {
439 .procname = "sched_cfs_bandwidth_slice_us",
440 .data = &sysctl_sched_cfs_bandwidth_slice,
441 .maxlen = sizeof(unsigned int),
442 .mode = 0644,
443 .proc_handler = proc_dointvec_minmax,
444 .extra1 = &one,
445 },
446 #endif
447 #ifdef CONFIG_PROVE_LOCKING
448 {
449 .procname = "prove_locking",
450 .data = &prove_locking,
451 .maxlen = sizeof(int),
452 .mode = 0644,
453 .proc_handler = proc_dointvec,
454 },
455 #endif
456 #ifdef CONFIG_LOCK_STAT
457 {
458 .procname = "lock_stat",
459 .data = &lock_stat,
460 .maxlen = sizeof(int),
461 .mode = 0644,
462 .proc_handler = proc_dointvec,
463 },
464 #endif
465 {
466 .procname = "panic",
467 .data = &panic_timeout,
468 .maxlen = sizeof(int),
469 .mode = 0644,
470 .proc_handler = proc_dointvec,
471 },
472 #ifdef CONFIG_COREDUMP
473 {
474 .procname = "core_uses_pid",
475 .data = &core_uses_pid,
476 .maxlen = sizeof(int),
477 .mode = 0644,
478 .proc_handler = proc_dointvec,
479 },
480 {
481 .procname = "core_pattern",
482 .data = core_pattern,
483 .maxlen = CORENAME_MAX_SIZE,
484 .mode = 0644,
485 .proc_handler = proc_dostring_coredump,
486 },
487 {
488 .procname = "core_pipe_limit",
489 .data = &core_pipe_limit,
490 .maxlen = sizeof(unsigned int),
491 .mode = 0644,
492 .proc_handler = proc_dointvec,
493 },
494 #endif
495 #ifdef CONFIG_PROC_SYSCTL
496 {
497 .procname = "tainted",
498 .maxlen = sizeof(long),
499 .mode = 0644,
500 .proc_handler = proc_taint,
501 },
502 {
503 .procname = "sysctl_writes_strict",
504 .data = &sysctl_writes_strict,
505 .maxlen = sizeof(int),
506 .mode = 0644,
507 .proc_handler = proc_dointvec_minmax,
508 .extra1 = &neg_one,
509 .extra2 = &one,
510 },
511 #endif
512 #ifdef CONFIG_LATENCYTOP
513 {
514 .procname = "latencytop",
515 .data = &latencytop_enabled,
516 .maxlen = sizeof(int),
517 .mode = 0644,
518 .proc_handler = proc_dointvec,
519 },
520 #endif
521 #ifdef CONFIG_BLK_DEV_INITRD
522 {
523 .procname = "real-root-dev",
524 .data = &real_root_dev,
525 .maxlen = sizeof(int),
526 .mode = 0644,
527 .proc_handler = proc_dointvec,
528 },
529 #endif
530 {
531 .procname = "print-fatal-signals",
532 .data = &print_fatal_signals,
533 .maxlen = sizeof(int),
534 .mode = 0644,
535 .proc_handler = proc_dointvec,
536 },
537 #ifdef CONFIG_SPARC
538 {
539 .procname = "reboot-cmd",
540 .data = reboot_command,
541 .maxlen = 256,
542 .mode = 0644,
543 .proc_handler = proc_dostring,
544 },
545 {
546 .procname = "stop-a",
547 .data = &stop_a_enabled,
548 .maxlen = sizeof (int),
549 .mode = 0644,
550 .proc_handler = proc_dointvec,
551 },
552 {
553 .procname = "scons-poweroff",
554 .data = &scons_pwroff,
555 .maxlen = sizeof (int),
556 .mode = 0644,
557 .proc_handler = proc_dointvec,
558 },
559 #endif
560 #ifdef CONFIG_SPARC64
561 {
562 .procname = "tsb-ratio",
563 .data = &sysctl_tsb_ratio,
564 .maxlen = sizeof (int),
565 .mode = 0644,
566 .proc_handler = proc_dointvec,
567 },
568 #endif
569 #ifdef __hppa__
570 {
571 .procname = "soft-power",
572 .data = &pwrsw_enabled,
573 .maxlen = sizeof (int),
574 .mode = 0644,
575 .proc_handler = proc_dointvec,
576 },
577 #endif
578 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
579 {
580 .procname = "unaligned-trap",
581 .data = &unaligned_enabled,
582 .maxlen = sizeof (int),
583 .mode = 0644,
584 .proc_handler = proc_dointvec,
585 },
586 #endif
587 {
588 .procname = "ctrl-alt-del",
589 .data = &C_A_D,
590 .maxlen = sizeof(int),
591 .mode = 0644,
592 .proc_handler = proc_dointvec,
593 },
594 #ifdef CONFIG_FUNCTION_TRACER
595 {
596 .procname = "ftrace_enabled",
597 .data = &ftrace_enabled,
598 .maxlen = sizeof(int),
599 .mode = 0644,
600 .proc_handler = ftrace_enable_sysctl,
601 },
602 #endif
603 #ifdef CONFIG_STACK_TRACER
604 {
605 .procname = "stack_tracer_enabled",
606 .data = &stack_tracer_enabled,
607 .maxlen = sizeof(int),
608 .mode = 0644,
609 .proc_handler = stack_trace_sysctl,
610 },
611 #endif
612 #ifdef CONFIG_TRACING
613 {
614 .procname = "ftrace_dump_on_oops",
615 .data = &ftrace_dump_on_oops,
616 .maxlen = sizeof(int),
617 .mode = 0644,
618 .proc_handler = proc_dointvec,
619 },
620 {
621 .procname = "traceoff_on_warning",
622 .data = &__disable_trace_on_warning,
623 .maxlen = sizeof(__disable_trace_on_warning),
624 .mode = 0644,
625 .proc_handler = proc_dointvec,
626 },
627 {
628 .procname = "tracepoint_printk",
629 .data = &tracepoint_printk,
630 .maxlen = sizeof(tracepoint_printk),
631 .mode = 0644,
632 .proc_handler = proc_dointvec,
633 },
634 #endif
635 #ifdef CONFIG_KEXEC
636 {
637 .procname = "kexec_load_disabled",
638 .data = &kexec_load_disabled,
639 .maxlen = sizeof(int),
640 .mode = 0644,
641 /* only handle a transition from default "0" to "1" */
642 .proc_handler = proc_dointvec_minmax,
643 .extra1 = &one,
644 .extra2 = &one,
645 },
646 #endif
647 #ifdef CONFIG_MODULES
648 {
649 .procname = "modprobe",
650 .data = &modprobe_path,
651 .maxlen = KMOD_PATH_LEN,
652 .mode = 0644,
653 .proc_handler = proc_dostring,
654 },
655 {
656 .procname = "modules_disabled",
657 .data = &modules_disabled,
658 .maxlen = sizeof(int),
659 .mode = 0644,
660 /* only handle a transition from default "0" to "1" */
661 .proc_handler = proc_dointvec_minmax,
662 .extra1 = &one,
663 .extra2 = &one,
664 },
665 #endif
666 #ifdef CONFIG_UEVENT_HELPER
667 {
668 .procname = "hotplug",
669 .data = &uevent_helper,
670 .maxlen = UEVENT_HELPER_PATH_LEN,
671 .mode = 0644,
672 .proc_handler = proc_dostring,
673 },
674 #endif
675 #ifdef CONFIG_CHR_DEV_SG
676 {
677 .procname = "sg-big-buff",
678 .data = &sg_big_buff,
679 .maxlen = sizeof (int),
680 .mode = 0444,
681 .proc_handler = proc_dointvec,
682 },
683 #endif
684 #ifdef CONFIG_BSD_PROCESS_ACCT
685 {
686 .procname = "acct",
687 .data = &acct_parm,
688 .maxlen = 3*sizeof(int),
689 .mode = 0644,
690 .proc_handler = proc_dointvec,
691 },
692 #endif
693 #ifdef CONFIG_MAGIC_SYSRQ
694 {
695 .procname = "sysrq",
696 .data = &__sysrq_enabled,
697 .maxlen = sizeof (int),
698 .mode = 0644,
699 .proc_handler = sysrq_sysctl_handler,
700 },
701 #endif
702 #ifdef CONFIG_PROC_SYSCTL
703 {
704 .procname = "cad_pid",
705 .data = NULL,
706 .maxlen = sizeof (int),
707 .mode = 0600,
708 .proc_handler = proc_do_cad_pid,
709 },
710 #endif
711 {
712 .procname = "threads-max",
713 .data = &max_threads,
714 .maxlen = sizeof(int),
715 .mode = 0644,
716 .proc_handler = proc_dointvec,
717 },
718 {
719 .procname = "random",
720 .mode = 0555,
721 .child = random_table,
722 },
723 {
724 .procname = "usermodehelper",
725 .mode = 0555,
726 .child = usermodehelper_table,
727 },
728 {
729 .procname = "overflowuid",
730 .data = &overflowuid,
731 .maxlen = sizeof(int),
732 .mode = 0644,
733 .proc_handler = proc_dointvec_minmax,
734 .extra1 = &minolduid,
735 .extra2 = &maxolduid,
736 },
737 {
738 .procname = "overflowgid",
739 .data = &overflowgid,
740 .maxlen = sizeof(int),
741 .mode = 0644,
742 .proc_handler = proc_dointvec_minmax,
743 .extra1 = &minolduid,
744 .extra2 = &maxolduid,
745 },
746 #ifdef CONFIG_S390
747 #ifdef CONFIG_MATHEMU
748 {
749 .procname = "ieee_emulation_warnings",
750 .data = &sysctl_ieee_emulation_warnings,
751 .maxlen = sizeof(int),
752 .mode = 0644,
753 .proc_handler = proc_dointvec,
754 },
755 #endif
756 {
757 .procname = "userprocess_debug",
758 .data = &show_unhandled_signals,
759 .maxlen = sizeof(int),
760 .mode = 0644,
761 .proc_handler = proc_dointvec,
762 },
763 #endif
764 {
765 .procname = "pid_max",
766 .data = &pid_max,
767 .maxlen = sizeof (int),
768 .mode = 0644,
769 .proc_handler = proc_dointvec_minmax,
770 .extra1 = &pid_max_min,
771 .extra2 = &pid_max_max,
772 },
773 {
774 .procname = "panic_on_oops",
775 .data = &panic_on_oops,
776 .maxlen = sizeof(int),
777 .mode = 0644,
778 .proc_handler = proc_dointvec,
779 },
780 #if defined CONFIG_PRINTK
781 {
782 .procname = "printk",
783 .data = &console_loglevel,
784 .maxlen = 4*sizeof(int),
785 .mode = 0644,
786 .proc_handler = proc_dointvec,
787 },
788 {
789 .procname = "printk_ratelimit",
790 .data = &printk_ratelimit_state.interval,
791 .maxlen = sizeof(int),
792 .mode = 0644,
793 .proc_handler = proc_dointvec_jiffies,
794 },
795 {
796 .procname = "printk_ratelimit_burst",
797 .data = &printk_ratelimit_state.burst,
798 .maxlen = sizeof(int),
799 .mode = 0644,
800 .proc_handler = proc_dointvec,
801 },
802 {
803 .procname = "printk_delay",
804 .data = &printk_delay_msec,
805 .maxlen = sizeof(int),
806 .mode = 0644,
807 .proc_handler = proc_dointvec_minmax,
808 .extra1 = &zero,
809 .extra2 = &ten_thousand,
810 },
811 {
812 .procname = "dmesg_restrict",
813 .data = &dmesg_restrict,
814 .maxlen = sizeof(int),
815 .mode = 0644,
816 .proc_handler = proc_dointvec_minmax_sysadmin,
817 .extra1 = &zero,
818 .extra2 = &one,
819 },
820 {
821 .procname = "kptr_restrict",
822 .data = &kptr_restrict,
823 .maxlen = sizeof(int),
824 .mode = 0644,
825 .proc_handler = proc_dointvec_minmax_sysadmin,
826 .extra1 = &zero,
827 .extra2 = &two,
828 },
829 #endif
830 {
831 .procname = "ngroups_max",
832 .data = &ngroups_max,
833 .maxlen = sizeof (int),
834 .mode = 0444,
835 .proc_handler = proc_dointvec,
836 },
837 {
838 .procname = "cap_last_cap",
839 .data = (void *)&cap_last_cap,
840 .maxlen = sizeof(int),
841 .mode = 0444,
842 .proc_handler = proc_dointvec,
843 },
844 #if defined(CONFIG_LOCKUP_DETECTOR)
845 {
846 .procname = "watchdog",
847 .data = &watchdog_user_enabled,
848 .maxlen = sizeof (int),
849 .mode = 0644,
850 .proc_handler = proc_watchdog,
851 .extra1 = &zero,
852 .extra2 = &one,
853 },
854 {
855 .procname = "watchdog_thresh",
856 .data = &watchdog_thresh,
857 .maxlen = sizeof(int),
858 .mode = 0644,
859 .proc_handler = proc_watchdog_thresh,
860 .extra1 = &zero,
861 .extra2 = &sixty,
862 },
863 {
864 .procname = "nmi_watchdog",
865 .data = &nmi_watchdog_enabled,
866 .maxlen = sizeof (int),
867 .mode = 0644,
868 .proc_handler = proc_nmi_watchdog,
869 .extra1 = &zero,
870 #if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR)
871 .extra2 = &one,
872 #else
873 .extra2 = &zero,
874 #endif
875 },
876 {
877 .procname = "soft_watchdog",
878 .data = &soft_watchdog_enabled,
879 .maxlen = sizeof (int),
880 .mode = 0644,
881 .proc_handler = proc_soft_watchdog,
882 .extra1 = &zero,
883 .extra2 = &one,
884 },
885 {
886 .procname = "softlockup_panic",
887 .data = &softlockup_panic,
888 .maxlen = sizeof(int),
889 .mode = 0644,
890 .proc_handler = proc_dointvec_minmax,
891 .extra1 = &zero,
892 .extra2 = &one,
893 },
894 #ifdef CONFIG_SMP
895 {
896 .procname = "softlockup_all_cpu_backtrace",
897 .data = &sysctl_softlockup_all_cpu_backtrace,
898 .maxlen = sizeof(int),
899 .mode = 0644,
900 .proc_handler = proc_dointvec_minmax,
901 .extra1 = &zero,
902 .extra2 = &one,
903 },
904 #endif /* CONFIG_SMP */
905 #endif
906 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
907 {
908 .procname = "unknown_nmi_panic",
909 .data = &unknown_nmi_panic,
910 .maxlen = sizeof (int),
911 .mode = 0644,
912 .proc_handler = proc_dointvec,
913 },
914 #endif
915 #if defined(CONFIG_X86)
916 {
917 .procname = "panic_on_unrecovered_nmi",
918 .data = &panic_on_unrecovered_nmi,
919 .maxlen = sizeof(int),
920 .mode = 0644,
921 .proc_handler = proc_dointvec,
922 },
923 {
924 .procname = "panic_on_io_nmi",
925 .data = &panic_on_io_nmi,
926 .maxlen = sizeof(int),
927 .mode = 0644,
928 .proc_handler = proc_dointvec,
929 },
930 #ifdef CONFIG_DEBUG_STACKOVERFLOW
931 {
932 .procname = "panic_on_stackoverflow",
933 .data = &sysctl_panic_on_stackoverflow,
934 .maxlen = sizeof(int),
935 .mode = 0644,
936 .proc_handler = proc_dointvec,
937 },
938 #endif
939 {
940 .procname = "bootloader_type",
941 .data = &bootloader_type,
942 .maxlen = sizeof (int),
943 .mode = 0444,
944 .proc_handler = proc_dointvec,
945 },
946 {
947 .procname = "bootloader_version",
948 .data = &bootloader_version,
949 .maxlen = sizeof (int),
950 .mode = 0444,
951 .proc_handler = proc_dointvec,
952 },
953 {
954 .procname = "kstack_depth_to_print",
955 .data = &kstack_depth_to_print,
956 .maxlen = sizeof(int),
957 .mode = 0644,
958 .proc_handler = proc_dointvec,
959 },
960 {
961 .procname = "io_delay_type",
962 .data = &io_delay_type,
963 .maxlen = sizeof(int),
964 .mode = 0644,
965 .proc_handler = proc_dointvec,
966 },
967 #endif
968 #if defined(CONFIG_MMU)
969 {
970 .procname = "randomize_va_space",
971 .data = &randomize_va_space,
972 .maxlen = sizeof(int),
973 .mode = 0644,
974 .proc_handler = proc_dointvec,
975 },
976 #endif
977 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
978 {
979 .procname = "spin_retry",
980 .data = &spin_retry,
981 .maxlen = sizeof (int),
982 .mode = 0644,
983 .proc_handler = proc_dointvec,
984 },
985 #endif
986 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
987 {
988 .procname = "acpi_video_flags",
989 .data = &acpi_realmode_flags,
990 .maxlen = sizeof (unsigned long),
991 .mode = 0644,
992 .proc_handler = proc_doulongvec_minmax,
993 },
994 #endif
995 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
996 {
997 .procname = "ignore-unaligned-usertrap",
998 .data = &no_unaligned_warning,
999 .maxlen = sizeof (int),
1000 .mode = 0644,
1001 .proc_handler = proc_dointvec,
1002 },
1003 #endif
1004 #ifdef CONFIG_IA64
1005 {
1006 .procname = "unaligned-dump-stack",
1007 .data = &unaligned_dump_stack,
1008 .maxlen = sizeof (int),
1009 .mode = 0644,
1010 .proc_handler = proc_dointvec,
1011 },
1012 #endif
1013 #ifdef CONFIG_DETECT_HUNG_TASK
1014 {
1015 .procname = "hung_task_panic",
1016 .data = &sysctl_hung_task_panic,
1017 .maxlen = sizeof(int),
1018 .mode = 0644,
1019 .proc_handler = proc_dointvec_minmax,
1020 .extra1 = &zero,
1021 .extra2 = &one,
1022 },
1023 {
1024 .procname = "hung_task_check_count",
1025 .data = &sysctl_hung_task_check_count,
1026 .maxlen = sizeof(int),
1027 .mode = 0644,
1028 .proc_handler = proc_dointvec_minmax,
1029 .extra1 = &zero,
1030 },
1031 {
1032 .procname = "hung_task_timeout_secs",
1033 .data = &sysctl_hung_task_timeout_secs,
1034 .maxlen = sizeof(unsigned long),
1035 .mode = 0644,
1036 .proc_handler = proc_dohung_task_timeout_secs,
1037 .extra2 = &hung_task_timeout_max,
1038 },
1039 {
1040 .procname = "hung_task_warnings",
1041 .data = &sysctl_hung_task_warnings,
1042 .maxlen = sizeof(int),
1043 .mode = 0644,
1044 .proc_handler = proc_dointvec_minmax,
1045 .extra1 = &neg_one,
1046 },
1047 #endif
1048 #ifdef CONFIG_COMPAT
1049 {
1050 .procname = "compat-log",
1051 .data = &compat_log,
1052 .maxlen = sizeof (int),
1053 .mode = 0644,
1054 .proc_handler = proc_dointvec,
1055 },
1056 #endif
1057 #ifdef CONFIG_RT_MUTEXES
1058 {
1059 .procname = "max_lock_depth",
1060 .data = &max_lock_depth,
1061 .maxlen = sizeof(int),
1062 .mode = 0644,
1063 .proc_handler = proc_dointvec,
1064 },
1065 #endif
1066 {
1067 .procname = "poweroff_cmd",
1068 .data = &poweroff_cmd,
1069 .maxlen = POWEROFF_CMD_PATH_LEN,
1070 .mode = 0644,
1071 .proc_handler = proc_dostring,
1072 },
1073 #ifdef CONFIG_KEYS
1074 {
1075 .procname = "keys",
1076 .mode = 0555,
1077 .child = key_sysctls,
1078 },
1079 #endif
1080 #ifdef CONFIG_PERF_EVENTS
1081 /*
1082 * User-space scripts rely on the existence of this file
1083 * as a feature check for perf_events being enabled.
1084 *
1085 * So it's an ABI, do not remove!
1086 */
1087 {
1088 .procname = "perf_event_paranoid",
1089 .data = &sysctl_perf_event_paranoid,
1090 .maxlen = sizeof(sysctl_perf_event_paranoid),
1091 .mode = 0644,
1092 .proc_handler = proc_dointvec,
1093 },
1094 {
1095 .procname = "perf_event_mlock_kb",
1096 .data = &sysctl_perf_event_mlock,
1097 .maxlen = sizeof(sysctl_perf_event_mlock),
1098 .mode = 0644,
1099 .proc_handler = proc_dointvec,
1100 },
1101 {
1102 .procname = "perf_event_max_sample_rate",
1103 .data = &sysctl_perf_event_sample_rate,
1104 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1105 .mode = 0644,
1106 .proc_handler = perf_proc_update_handler,
1107 .extra1 = &one,
1108 },
1109 {
1110 .procname = "perf_cpu_time_max_percent",
1111 .data = &sysctl_perf_cpu_time_max_percent,
1112 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1113 .mode = 0644,
1114 .proc_handler = perf_cpu_time_max_percent_handler,
1115 .extra1 = &zero,
1116 .extra2 = &one_hundred,
1117 },
1118 #endif
1119 #ifdef CONFIG_KMEMCHECK
1120 {
1121 .procname = "kmemcheck",
1122 .data = &kmemcheck_enabled,
1123 .maxlen = sizeof(int),
1124 .mode = 0644,
1125 .proc_handler = proc_dointvec,
1126 },
1127 #endif
1128 {
1129 .procname = "panic_on_warn",
1130 .data = &panic_on_warn,
1131 .maxlen = sizeof(int),
1132 .mode = 0644,
1133 .proc_handler = proc_dointvec_minmax,
1134 .extra1 = &zero,
1135 .extra2 = &one,
1136 },
1137 { }
1138 };
1139
1140 static struct ctl_table vm_table[] = {
1141 {
1142 .procname = "overcommit_memory",
1143 .data = &sysctl_overcommit_memory,
1144 .maxlen = sizeof(sysctl_overcommit_memory),
1145 .mode = 0644,
1146 .proc_handler = proc_dointvec_minmax,
1147 .extra1 = &zero,
1148 .extra2 = &two,
1149 },
1150 {
1151 .procname = "panic_on_oom",
1152 .data = &sysctl_panic_on_oom,
1153 .maxlen = sizeof(sysctl_panic_on_oom),
1154 .mode = 0644,
1155 .proc_handler = proc_dointvec_minmax,
1156 .extra1 = &zero,
1157 .extra2 = &two,
1158 },
1159 {
1160 .procname = "oom_kill_allocating_task",
1161 .data = &sysctl_oom_kill_allocating_task,
1162 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1163 .mode = 0644,
1164 .proc_handler = proc_dointvec,
1165 },
1166 {
1167 .procname = "oom_dump_tasks",
1168 .data = &sysctl_oom_dump_tasks,
1169 .maxlen = sizeof(sysctl_oom_dump_tasks),
1170 .mode = 0644,
1171 .proc_handler = proc_dointvec,
1172 },
1173 {
1174 .procname = "overcommit_ratio",
1175 .data = &sysctl_overcommit_ratio,
1176 .maxlen = sizeof(sysctl_overcommit_ratio),
1177 .mode = 0644,
1178 .proc_handler = overcommit_ratio_handler,
1179 },
1180 {
1181 .procname = "overcommit_kbytes",
1182 .data = &sysctl_overcommit_kbytes,
1183 .maxlen = sizeof(sysctl_overcommit_kbytes),
1184 .mode = 0644,
1185 .proc_handler = overcommit_kbytes_handler,
1186 },
1187 {
1188 .procname = "page-cluster",
1189 .data = &page_cluster,
1190 .maxlen = sizeof(int),
1191 .mode = 0644,
1192 .proc_handler = proc_dointvec_minmax,
1193 .extra1 = &zero,
1194 },
1195 {
1196 .procname = "dirty_background_ratio",
1197 .data = &dirty_background_ratio,
1198 .maxlen = sizeof(dirty_background_ratio),
1199 .mode = 0644,
1200 .proc_handler = dirty_background_ratio_handler,
1201 .extra1 = &zero,
1202 .extra2 = &one_hundred,
1203 },
1204 {
1205 .procname = "dirty_background_bytes",
1206 .data = &dirty_background_bytes,
1207 .maxlen = sizeof(dirty_background_bytes),
1208 .mode = 0644,
1209 .proc_handler = dirty_background_bytes_handler,
1210 .extra1 = &one_ul,
1211 },
1212 {
1213 .procname = "dirty_ratio",
1214 .data = &vm_dirty_ratio,
1215 .maxlen = sizeof(vm_dirty_ratio),
1216 .mode = 0644,
1217 .proc_handler = dirty_ratio_handler,
1218 .extra1 = &zero,
1219 .extra2 = &one_hundred,
1220 },
1221 {
1222 .procname = "dirty_bytes",
1223 .data = &vm_dirty_bytes,
1224 .maxlen = sizeof(vm_dirty_bytes),
1225 .mode = 0644,
1226 .proc_handler = dirty_bytes_handler,
1227 .extra1 = &dirty_bytes_min,
1228 },
1229 {
1230 .procname = "dirty_writeback_centisecs",
1231 .data = &dirty_writeback_interval,
1232 .maxlen = sizeof(dirty_writeback_interval),
1233 .mode = 0644,
1234 .proc_handler = dirty_writeback_centisecs_handler,
1235 },
1236 {
1237 .procname = "dirty_expire_centisecs",
1238 .data = &dirty_expire_interval,
1239 .maxlen = sizeof(dirty_expire_interval),
1240 .mode = 0644,
1241 .proc_handler = proc_dointvec_minmax,
1242 .extra1 = &zero,
1243 },
1244 {
1245 .procname = "dirtytime_expire_seconds",
1246 .data = &dirtytime_expire_interval,
1247 .maxlen = sizeof(dirty_expire_interval),
1248 .mode = 0644,
1249 .proc_handler = dirtytime_interval_handler,
1250 .extra1 = &zero,
1251 },
1252 {
1253 .procname = "nr_pdflush_threads",
1254 .mode = 0444 /* read-only */,
1255 .proc_handler = pdflush_proc_obsolete,
1256 },
1257 {
1258 .procname = "swappiness",
1259 .data = &vm_swappiness,
1260 .maxlen = sizeof(vm_swappiness),
1261 .mode = 0644,
1262 .proc_handler = proc_dointvec_minmax,
1263 .extra1 = &zero,
1264 .extra2 = &one_hundred,
1265 },
1266 #ifdef CONFIG_HUGETLB_PAGE
1267 {
1268 .procname = "nr_hugepages",
1269 .data = NULL,
1270 .maxlen = sizeof(unsigned long),
1271 .mode = 0644,
1272 .proc_handler = hugetlb_sysctl_handler,
1273 },
1274 #ifdef CONFIG_NUMA
1275 {
1276 .procname = "nr_hugepages_mempolicy",
1277 .data = NULL,
1278 .maxlen = sizeof(unsigned long),
1279 .mode = 0644,
1280 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1281 },
1282 #endif
1283 {
1284 .procname = "hugetlb_shm_group",
1285 .data = &sysctl_hugetlb_shm_group,
1286 .maxlen = sizeof(gid_t),
1287 .mode = 0644,
1288 .proc_handler = proc_dointvec,
1289 },
1290 {
1291 .procname = "hugepages_treat_as_movable",
1292 .data = &hugepages_treat_as_movable,
1293 .maxlen = sizeof(int),
1294 .mode = 0644,
1295 .proc_handler = proc_dointvec,
1296 },
1297 {
1298 .procname = "nr_overcommit_hugepages",
1299 .data = NULL,
1300 .maxlen = sizeof(unsigned long),
1301 .mode = 0644,
1302 .proc_handler = hugetlb_overcommit_handler,
1303 },
1304 #endif
1305 {
1306 .procname = "lowmem_reserve_ratio",
1307 .data = &sysctl_lowmem_reserve_ratio,
1308 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1309 .mode = 0644,
1310 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1311 },
1312 {
1313 .procname = "drop_caches",
1314 .data = &sysctl_drop_caches,
1315 .maxlen = sizeof(int),
1316 .mode = 0644,
1317 .proc_handler = drop_caches_sysctl_handler,
1318 .extra1 = &one,
1319 .extra2 = &four,
1320 },
1321 #ifdef CONFIG_COMPACTION
1322 {
1323 .procname = "compact_memory",
1324 .data = &sysctl_compact_memory,
1325 .maxlen = sizeof(int),
1326 .mode = 0200,
1327 .proc_handler = sysctl_compaction_handler,
1328 },
1329 {
1330 .procname = "extfrag_threshold",
1331 .data = &sysctl_extfrag_threshold,
1332 .maxlen = sizeof(int),
1333 .mode = 0644,
1334 .proc_handler = sysctl_extfrag_handler,
1335 .extra1 = &min_extfrag_threshold,
1336 .extra2 = &max_extfrag_threshold,
1337 },
1338
1339 #endif /* CONFIG_COMPACTION */
1340 {
1341 .procname = "min_free_kbytes",
1342 .data = &min_free_kbytes,
1343 .maxlen = sizeof(min_free_kbytes),
1344 .mode = 0644,
1345 .proc_handler = min_free_kbytes_sysctl_handler,
1346 .extra1 = &zero,
1347 },
1348 {
1349 .procname = "percpu_pagelist_fraction",
1350 .data = &percpu_pagelist_fraction,
1351 .maxlen = sizeof(percpu_pagelist_fraction),
1352 .mode = 0644,
1353 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1354 .extra1 = &zero,
1355 },
1356 #ifdef CONFIG_MMU
1357 {
1358 .procname = "max_map_count",
1359 .data = &sysctl_max_map_count,
1360 .maxlen = sizeof(sysctl_max_map_count),
1361 .mode = 0644,
1362 .proc_handler = proc_dointvec_minmax,
1363 .extra1 = &zero,
1364 },
1365 #else
1366 {
1367 .procname = "nr_trim_pages",
1368 .data = &sysctl_nr_trim_pages,
1369 .maxlen = sizeof(sysctl_nr_trim_pages),
1370 .mode = 0644,
1371 .proc_handler = proc_dointvec_minmax,
1372 .extra1 = &zero,
1373 },
1374 #endif
1375 {
1376 .procname = "laptop_mode",
1377 .data = &laptop_mode,
1378 .maxlen = sizeof(laptop_mode),
1379 .mode = 0644,
1380 .proc_handler = proc_dointvec_jiffies,
1381 },
1382 {
1383 .procname = "block_dump",
1384 .data = &block_dump,
1385 .maxlen = sizeof(block_dump),
1386 .mode = 0644,
1387 .proc_handler = proc_dointvec,
1388 .extra1 = &zero,
1389 },
1390 {
1391 .procname = "vfs_cache_pressure",
1392 .data = &sysctl_vfs_cache_pressure,
1393 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1394 .mode = 0644,
1395 .proc_handler = proc_dointvec,
1396 .extra1 = &zero,
1397 },
1398 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1399 {
1400 .procname = "legacy_va_layout",
1401 .data = &sysctl_legacy_va_layout,
1402 .maxlen = sizeof(sysctl_legacy_va_layout),
1403 .mode = 0644,
1404 .proc_handler = proc_dointvec,
1405 .extra1 = &zero,
1406 },
1407 #endif
1408 #ifdef CONFIG_NUMA
1409 {
1410 .procname = "zone_reclaim_mode",
1411 .data = &zone_reclaim_mode,
1412 .maxlen = sizeof(zone_reclaim_mode),
1413 .mode = 0644,
1414 .proc_handler = proc_dointvec,
1415 .extra1 = &zero,
1416 },
1417 {
1418 .procname = "min_unmapped_ratio",
1419 .data = &sysctl_min_unmapped_ratio,
1420 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1421 .mode = 0644,
1422 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1423 .extra1 = &zero,
1424 .extra2 = &one_hundred,
1425 },
1426 {
1427 .procname = "min_slab_ratio",
1428 .data = &sysctl_min_slab_ratio,
1429 .maxlen = sizeof(sysctl_min_slab_ratio),
1430 .mode = 0644,
1431 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1432 .extra1 = &zero,
1433 .extra2 = &one_hundred,
1434 },
1435 #endif
1436 #ifdef CONFIG_SMP
1437 {
1438 .procname = "stat_interval",
1439 .data = &sysctl_stat_interval,
1440 .maxlen = sizeof(sysctl_stat_interval),
1441 .mode = 0644,
1442 .proc_handler = proc_dointvec_jiffies,
1443 },
1444 #endif
1445 #ifdef CONFIG_MMU
1446 {
1447 .procname = "mmap_min_addr",
1448 .data = &dac_mmap_min_addr,
1449 .maxlen = sizeof(unsigned long),
1450 .mode = 0644,
1451 .proc_handler = mmap_min_addr_handler,
1452 },
1453 #endif
1454 #ifdef CONFIG_NUMA
1455 {
1456 .procname = "numa_zonelist_order",
1457 .data = &numa_zonelist_order,
1458 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1459 .mode = 0644,
1460 .proc_handler = numa_zonelist_order_handler,
1461 },
1462 #endif
1463 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1464 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1465 {
1466 .procname = "vdso_enabled",
1467 #ifdef CONFIG_X86_32
1468 .data = &vdso32_enabled,
1469 .maxlen = sizeof(vdso32_enabled),
1470 #else
1471 .data = &vdso_enabled,
1472 .maxlen = sizeof(vdso_enabled),
1473 #endif
1474 .mode = 0644,
1475 .proc_handler = proc_dointvec,
1476 .extra1 = &zero,
1477 },
1478 #endif
1479 #ifdef CONFIG_HIGHMEM
1480 {
1481 .procname = "highmem_is_dirtyable",
1482 .data = &vm_highmem_is_dirtyable,
1483 .maxlen = sizeof(vm_highmem_is_dirtyable),
1484 .mode = 0644,
1485 .proc_handler = proc_dointvec_minmax,
1486 .extra1 = &zero,
1487 .extra2 = &one,
1488 },
1489 #endif
1490 #ifdef CONFIG_MEMORY_FAILURE
1491 {
1492 .procname = "memory_failure_early_kill",
1493 .data = &sysctl_memory_failure_early_kill,
1494 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1495 .mode = 0644,
1496 .proc_handler = proc_dointvec_minmax,
1497 .extra1 = &zero,
1498 .extra2 = &one,
1499 },
1500 {
1501 .procname = "memory_failure_recovery",
1502 .data = &sysctl_memory_failure_recovery,
1503 .maxlen = sizeof(sysctl_memory_failure_recovery),
1504 .mode = 0644,
1505 .proc_handler = proc_dointvec_minmax,
1506 .extra1 = &zero,
1507 .extra2 = &one,
1508 },
1509 #endif
1510 {
1511 .procname = "user_reserve_kbytes",
1512 .data = &sysctl_user_reserve_kbytes,
1513 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1514 .mode = 0644,
1515 .proc_handler = proc_doulongvec_minmax,
1516 },
1517 {
1518 .procname = "admin_reserve_kbytes",
1519 .data = &sysctl_admin_reserve_kbytes,
1520 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1521 .mode = 0644,
1522 .proc_handler = proc_doulongvec_minmax,
1523 },
1524 { }
1525 };
1526
1527 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1528 static struct ctl_table binfmt_misc_table[] = {
1529 { }
1530 };
1531 #endif
1532
1533 static struct ctl_table fs_table[] = {
1534 {
1535 .procname = "inode-nr",
1536 .data = &inodes_stat,
1537 .maxlen = 2*sizeof(long),
1538 .mode = 0444,
1539 .proc_handler = proc_nr_inodes,
1540 },
1541 {
1542 .procname = "inode-state",
1543 .data = &inodes_stat,
1544 .maxlen = 7*sizeof(long),
1545 .mode = 0444,
1546 .proc_handler = proc_nr_inodes,
1547 },
1548 {
1549 .procname = "file-nr",
1550 .data = &files_stat,
1551 .maxlen = sizeof(files_stat),
1552 .mode = 0444,
1553 .proc_handler = proc_nr_files,
1554 },
1555 {
1556 .procname = "file-max",
1557 .data = &files_stat.max_files,
1558 .maxlen = sizeof(files_stat.max_files),
1559 .mode = 0644,
1560 .proc_handler = proc_doulongvec_minmax,
1561 },
1562 {
1563 .procname = "nr_open",
1564 .data = &sysctl_nr_open,
1565 .maxlen = sizeof(int),
1566 .mode = 0644,
1567 .proc_handler = proc_dointvec_minmax,
1568 .extra1 = &sysctl_nr_open_min,
1569 .extra2 = &sysctl_nr_open_max,
1570 },
1571 {
1572 .procname = "dentry-state",
1573 .data = &dentry_stat,
1574 .maxlen = 6*sizeof(long),
1575 .mode = 0444,
1576 .proc_handler = proc_nr_dentry,
1577 },
1578 {
1579 .procname = "overflowuid",
1580 .data = &fs_overflowuid,
1581 .maxlen = sizeof(int),
1582 .mode = 0644,
1583 .proc_handler = proc_dointvec_minmax,
1584 .extra1 = &minolduid,
1585 .extra2 = &maxolduid,
1586 },
1587 {
1588 .procname = "overflowgid",
1589 .data = &fs_overflowgid,
1590 .maxlen = sizeof(int),
1591 .mode = 0644,
1592 .proc_handler = proc_dointvec_minmax,
1593 .extra1 = &minolduid,
1594 .extra2 = &maxolduid,
1595 },
1596 #ifdef CONFIG_FILE_LOCKING
1597 {
1598 .procname = "leases-enable",
1599 .data = &leases_enable,
1600 .maxlen = sizeof(int),
1601 .mode = 0644,
1602 .proc_handler = proc_dointvec,
1603 },
1604 #endif
1605 #ifdef CONFIG_DNOTIFY
1606 {
1607 .procname = "dir-notify-enable",
1608 .data = &dir_notify_enable,
1609 .maxlen = sizeof(int),
1610 .mode = 0644,
1611 .proc_handler = proc_dointvec,
1612 },
1613 #endif
1614 #ifdef CONFIG_MMU
1615 #ifdef CONFIG_FILE_LOCKING
1616 {
1617 .procname = "lease-break-time",
1618 .data = &lease_break_time,
1619 .maxlen = sizeof(int),
1620 .mode = 0644,
1621 .proc_handler = proc_dointvec,
1622 },
1623 #endif
1624 #ifdef CONFIG_AIO
1625 {
1626 .procname = "aio-nr",
1627 .data = &aio_nr,
1628 .maxlen = sizeof(aio_nr),
1629 .mode = 0444,
1630 .proc_handler = proc_doulongvec_minmax,
1631 },
1632 {
1633 .procname = "aio-max-nr",
1634 .data = &aio_max_nr,
1635 .maxlen = sizeof(aio_max_nr),
1636 .mode = 0644,
1637 .proc_handler = proc_doulongvec_minmax,
1638 },
1639 #endif /* CONFIG_AIO */
1640 #ifdef CONFIG_INOTIFY_USER
1641 {
1642 .procname = "inotify",
1643 .mode = 0555,
1644 .child = inotify_table,
1645 },
1646 #endif
1647 #ifdef CONFIG_EPOLL
1648 {
1649 .procname = "epoll",
1650 .mode = 0555,
1651 .child = epoll_table,
1652 },
1653 #endif
1654 #endif
1655 {
1656 .procname = "protected_symlinks",
1657 .data = &sysctl_protected_symlinks,
1658 .maxlen = sizeof(int),
1659 .mode = 0600,
1660 .proc_handler = proc_dointvec_minmax,
1661 .extra1 = &zero,
1662 .extra2 = &one,
1663 },
1664 {
1665 .procname = "protected_hardlinks",
1666 .data = &sysctl_protected_hardlinks,
1667 .maxlen = sizeof(int),
1668 .mode = 0600,
1669 .proc_handler = proc_dointvec_minmax,
1670 .extra1 = &zero,
1671 .extra2 = &one,
1672 },
1673 {
1674 .procname = "suid_dumpable",
1675 .data = &suid_dumpable,
1676 .maxlen = sizeof(int),
1677 .mode = 0644,
1678 .proc_handler = proc_dointvec_minmax_coredump,
1679 .extra1 = &zero,
1680 .extra2 = &two,
1681 },
1682 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1683 {
1684 .procname = "binfmt_misc",
1685 .mode = 0555,
1686 .child = binfmt_misc_table,
1687 },
1688 #endif
1689 {
1690 .procname = "pipe-max-size",
1691 .data = &pipe_max_size,
1692 .maxlen = sizeof(int),
1693 .mode = 0644,
1694 .proc_handler = &pipe_proc_fn,
1695 .extra1 = &pipe_min_size,
1696 },
1697 { }
1698 };
1699
1700 static struct ctl_table debug_table[] = {
1701 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1702 {
1703 .procname = "exception-trace",
1704 .data = &show_unhandled_signals,
1705 .maxlen = sizeof(int),
1706 .mode = 0644,
1707 .proc_handler = proc_dointvec
1708 },
1709 #endif
1710 #if defined(CONFIG_OPTPROBES)
1711 {
1712 .procname = "kprobes-optimization",
1713 .data = &sysctl_kprobes_optimization,
1714 .maxlen = sizeof(int),
1715 .mode = 0644,
1716 .proc_handler = proc_kprobes_optimization_handler,
1717 .extra1 = &zero,
1718 .extra2 = &one,
1719 },
1720 #endif
1721 { }
1722 };
1723
1724 static struct ctl_table dev_table[] = {
1725 { }
1726 };
1727
1728 int __init sysctl_init(void)
1729 {
1730 struct ctl_table_header *hdr;
1731
1732 hdr = register_sysctl_table(sysctl_base_table);
1733 kmemleak_not_leak(hdr);
1734 return 0;
1735 }
1736
1737 #endif /* CONFIG_SYSCTL */
1738
1739 /*
1740 * /proc/sys support
1741 */
1742
1743 #ifdef CONFIG_PROC_SYSCTL
1744
1745 static int _proc_do_string(char *data, int maxlen, int write,
1746 char __user *buffer,
1747 size_t *lenp, loff_t *ppos)
1748 {
1749 size_t len;
1750 char __user *p;
1751 char c;
1752
1753 if (!data || !maxlen || !*lenp) {
1754 *lenp = 0;
1755 return 0;
1756 }
1757
1758 if (write) {
1759 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1760 /* Only continue writes not past the end of buffer. */
1761 len = strlen(data);
1762 if (len > maxlen - 1)
1763 len = maxlen - 1;
1764
1765 if (*ppos > len)
1766 return 0;
1767 len = *ppos;
1768 } else {
1769 /* Start writing from beginning of buffer. */
1770 len = 0;
1771 }
1772
1773 *ppos += *lenp;
1774 p = buffer;
1775 while ((p - buffer) < *lenp && len < maxlen - 1) {
1776 if (get_user(c, p++))
1777 return -EFAULT;
1778 if (c == 0 || c == '\n')
1779 break;
1780 data[len++] = c;
1781 }
1782 data[len] = 0;
1783 } else {
1784 len = strlen(data);
1785 if (len > maxlen)
1786 len = maxlen;
1787
1788 if (*ppos > len) {
1789 *lenp = 0;
1790 return 0;
1791 }
1792
1793 data += *ppos;
1794 len -= *ppos;
1795
1796 if (len > *lenp)
1797 len = *lenp;
1798 if (len)
1799 if (copy_to_user(buffer, data, len))
1800 return -EFAULT;
1801 if (len < *lenp) {
1802 if (put_user('\n', buffer + len))
1803 return -EFAULT;
1804 len++;
1805 }
1806 *lenp = len;
1807 *ppos += len;
1808 }
1809 return 0;
1810 }
1811
1812 static void warn_sysctl_write(struct ctl_table *table)
1813 {
1814 pr_warn_once("%s wrote to %s when file position was not 0!\n"
1815 "This will not be supported in the future. To silence this\n"
1816 "warning, set kernel.sysctl_writes_strict = -1\n",
1817 current->comm, table->procname);
1818 }
1819
1820 /**
1821 * proc_dostring - read a string sysctl
1822 * @table: the sysctl table
1823 * @write: %TRUE if this is a write to the sysctl file
1824 * @buffer: the user buffer
1825 * @lenp: the size of the user buffer
1826 * @ppos: file position
1827 *
1828 * Reads/writes a string from/to the user buffer. If the kernel
1829 * buffer provided is not large enough to hold the string, the
1830 * string is truncated. The copied string is %NULL-terminated.
1831 * If the string is being read by the user process, it is copied
1832 * and a newline '\n' is added. It is truncated if the buffer is
1833 * not large enough.
1834 *
1835 * Returns 0 on success.
1836 */
1837 int proc_dostring(struct ctl_table *table, int write,
1838 void __user *buffer, size_t *lenp, loff_t *ppos)
1839 {
1840 if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN)
1841 warn_sysctl_write(table);
1842
1843 return _proc_do_string((char *)(table->data), table->maxlen, write,
1844 (char __user *)buffer, lenp, ppos);
1845 }
1846
1847 static size_t proc_skip_spaces(char **buf)
1848 {
1849 size_t ret;
1850 char *tmp = skip_spaces(*buf);
1851 ret = tmp - *buf;
1852 *buf = tmp;
1853 return ret;
1854 }
1855
1856 static void proc_skip_char(char **buf, size_t *size, const char v)
1857 {
1858 while (*size) {
1859 if (**buf != v)
1860 break;
1861 (*size)--;
1862 (*buf)++;
1863 }
1864 }
1865
1866 #define TMPBUFLEN 22
1867 /**
1868 * proc_get_long - reads an ASCII formatted integer from a user buffer
1869 *
1870 * @buf: a kernel buffer
1871 * @size: size of the kernel buffer
1872 * @val: this is where the number will be stored
1873 * @neg: set to %TRUE if number is negative
1874 * @perm_tr: a vector which contains the allowed trailers
1875 * @perm_tr_len: size of the perm_tr vector
1876 * @tr: pointer to store the trailer character
1877 *
1878 * In case of success %0 is returned and @buf and @size are updated with
1879 * the amount of bytes read. If @tr is non-NULL and a trailing
1880 * character exists (size is non-zero after returning from this
1881 * function), @tr is updated with the trailing character.
1882 */
1883 static int proc_get_long(char **buf, size_t *size,
1884 unsigned long *val, bool *neg,
1885 const char *perm_tr, unsigned perm_tr_len, char *tr)
1886 {
1887 int len;
1888 char *p, tmp[TMPBUFLEN];
1889
1890 if (!*size)
1891 return -EINVAL;
1892
1893 len = *size;
1894 if (len > TMPBUFLEN - 1)
1895 len = TMPBUFLEN - 1;
1896
1897 memcpy(tmp, *buf, len);
1898
1899 tmp[len] = 0;
1900 p = tmp;
1901 if (*p == '-' && *size > 1) {
1902 *neg = true;
1903 p++;
1904 } else
1905 *neg = false;
1906 if (!isdigit(*p))
1907 return -EINVAL;
1908
1909 *val = simple_strtoul(p, &p, 0);
1910
1911 len = p - tmp;
1912
1913 /* We don't know if the next char is whitespace thus we may accept
1914 * invalid integers (e.g. 1234...a) or two integers instead of one
1915 * (e.g. 123...1). So lets not allow such large numbers. */
1916 if (len == TMPBUFLEN - 1)
1917 return -EINVAL;
1918
1919 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
1920 return -EINVAL;
1921
1922 if (tr && (len < *size))
1923 *tr = *p;
1924
1925 *buf += len;
1926 *size -= len;
1927
1928 return 0;
1929 }
1930
1931 /**
1932 * proc_put_long - converts an integer to a decimal ASCII formatted string
1933 *
1934 * @buf: the user buffer
1935 * @size: the size of the user buffer
1936 * @val: the integer to be converted
1937 * @neg: sign of the number, %TRUE for negative
1938 *
1939 * In case of success %0 is returned and @buf and @size are updated with
1940 * the amount of bytes written.
1941 */
1942 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
1943 bool neg)
1944 {
1945 int len;
1946 char tmp[TMPBUFLEN], *p = tmp;
1947
1948 sprintf(p, "%s%lu", neg ? "-" : "", val);
1949 len = strlen(tmp);
1950 if (len > *size)
1951 len = *size;
1952 if (copy_to_user(*buf, tmp, len))
1953 return -EFAULT;
1954 *size -= len;
1955 *buf += len;
1956 return 0;
1957 }
1958 #undef TMPBUFLEN
1959
1960 static int proc_put_char(void __user **buf, size_t *size, char c)
1961 {
1962 if (*size) {
1963 char __user **buffer = (char __user **)buf;
1964 if (put_user(c, *buffer))
1965 return -EFAULT;
1966 (*size)--, (*buffer)++;
1967 *buf = *buffer;
1968 }
1969 return 0;
1970 }
1971
1972 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
1973 int *valp,
1974 int write, void *data)
1975 {
1976 if (write) {
1977 *valp = *negp ? -*lvalp : *lvalp;
1978 } else {
1979 int val = *valp;
1980 if (val < 0) {
1981 *negp = true;
1982 *lvalp = (unsigned long)-val;
1983 } else {
1984 *negp = false;
1985 *lvalp = (unsigned long)val;
1986 }
1987 }
1988 return 0;
1989 }
1990
1991 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
1992
1993 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1994 int write, void __user *buffer,
1995 size_t *lenp, loff_t *ppos,
1996 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1997 int write, void *data),
1998 void *data)
1999 {
2000 int *i, vleft, first = 1, err = 0;
2001 unsigned long page = 0;
2002 size_t left;
2003 char *kbuf;
2004
2005 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2006 *lenp = 0;
2007 return 0;
2008 }
2009
2010 i = (int *) tbl_data;
2011 vleft = table->maxlen / sizeof(*i);
2012 left = *lenp;
2013
2014 if (!conv)
2015 conv = do_proc_dointvec_conv;
2016
2017 if (write) {
2018 if (*ppos) {
2019 switch (sysctl_writes_strict) {
2020 case SYSCTL_WRITES_STRICT:
2021 goto out;
2022 case SYSCTL_WRITES_WARN:
2023 warn_sysctl_write(table);
2024 break;
2025 default:
2026 break;
2027 }
2028 }
2029
2030 if (left > PAGE_SIZE - 1)
2031 left = PAGE_SIZE - 1;
2032 page = __get_free_page(GFP_TEMPORARY);
2033 kbuf = (char *) page;
2034 if (!kbuf)
2035 return -ENOMEM;
2036 if (copy_from_user(kbuf, buffer, left)) {
2037 err = -EFAULT;
2038 goto free;
2039 }
2040 kbuf[left] = 0;
2041 }
2042
2043 for (; left && vleft--; i++, first=0) {
2044 unsigned long lval;
2045 bool neg;
2046
2047 if (write) {
2048 left -= proc_skip_spaces(&kbuf);
2049
2050 if (!left)
2051 break;
2052 err = proc_get_long(&kbuf, &left, &lval, &neg,
2053 proc_wspace_sep,
2054 sizeof(proc_wspace_sep), NULL);
2055 if (err)
2056 break;
2057 if (conv(&neg, &lval, i, 1, data)) {
2058 err = -EINVAL;
2059 break;
2060 }
2061 } else {
2062 if (conv(&neg, &lval, i, 0, data)) {
2063 err = -EINVAL;
2064 break;
2065 }
2066 if (!first)
2067 err = proc_put_char(&buffer, &left, '\t');
2068 if (err)
2069 break;
2070 err = proc_put_long(&buffer, &left, lval, neg);
2071 if (err)
2072 break;
2073 }
2074 }
2075
2076 if (!write && !first && left && !err)
2077 err = proc_put_char(&buffer, &left, '\n');
2078 if (write && !err && left)
2079 left -= proc_skip_spaces(&kbuf);
2080 free:
2081 if (write) {
2082 free_page(page);
2083 if (first)
2084 return err ? : -EINVAL;
2085 }
2086 *lenp -= left;
2087 out:
2088 *ppos += *lenp;
2089 return err;
2090 }
2091
2092 static int do_proc_dointvec(struct ctl_table *table, int write,
2093 void __user *buffer, size_t *lenp, loff_t *ppos,
2094 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2095 int write, void *data),
2096 void *data)
2097 {
2098 return __do_proc_dointvec(table->data, table, write,
2099 buffer, lenp, ppos, conv, data);
2100 }
2101
2102 /**
2103 * proc_dointvec - read a vector of integers
2104 * @table: the sysctl table
2105 * @write: %TRUE if this is a write to the sysctl file
2106 * @buffer: the user buffer
2107 * @lenp: the size of the user buffer
2108 * @ppos: file position
2109 *
2110 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2111 * values from/to the user buffer, treated as an ASCII string.
2112 *
2113 * Returns 0 on success.
2114 */
2115 int proc_dointvec(struct ctl_table *table, int write,
2116 void __user *buffer, size_t *lenp, loff_t *ppos)
2117 {
2118 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2119 NULL,NULL);
2120 }
2121
2122 /*
2123 * Taint values can only be increased
2124 * This means we can safely use a temporary.
2125 */
2126 static int proc_taint(struct ctl_table *table, int write,
2127 void __user *buffer, size_t *lenp, loff_t *ppos)
2128 {
2129 struct ctl_table t;
2130 unsigned long tmptaint = get_taint();
2131 int err;
2132
2133 if (write && !capable(CAP_SYS_ADMIN))
2134 return -EPERM;
2135
2136 t = *table;
2137 t.data = &tmptaint;
2138 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2139 if (err < 0)
2140 return err;
2141
2142 if (write) {
2143 /*
2144 * Poor man's atomic or. Not worth adding a primitive
2145 * to everyone's atomic.h for this
2146 */
2147 int i;
2148 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2149 if ((tmptaint >> i) & 1)
2150 add_taint(i, LOCKDEP_STILL_OK);
2151 }
2152 }
2153
2154 return err;
2155 }
2156
2157 #ifdef CONFIG_PRINTK
2158 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2159 void __user *buffer, size_t *lenp, loff_t *ppos)
2160 {
2161 if (write && !capable(CAP_SYS_ADMIN))
2162 return -EPERM;
2163
2164 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2165 }
2166 #endif
2167
2168 struct do_proc_dointvec_minmax_conv_param {
2169 int *min;
2170 int *max;
2171 };
2172
2173 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2174 int *valp,
2175 int write, void *data)
2176 {
2177 struct do_proc_dointvec_minmax_conv_param *param = data;
2178 if (write) {
2179 int val = *negp ? -*lvalp : *lvalp;
2180 if ((param->min && *param->min > val) ||
2181 (param->max && *param->max < val))
2182 return -EINVAL;
2183 *valp = val;
2184 } else {
2185 int val = *valp;
2186 if (val < 0) {
2187 *negp = true;
2188 *lvalp = (unsigned long)-val;
2189 } else {
2190 *negp = false;
2191 *lvalp = (unsigned long)val;
2192 }
2193 }
2194 return 0;
2195 }
2196
2197 /**
2198 * proc_dointvec_minmax - read a vector of integers with min/max values
2199 * @table: the sysctl table
2200 * @write: %TRUE if this is a write to the sysctl file
2201 * @buffer: the user buffer
2202 * @lenp: the size of the user buffer
2203 * @ppos: file position
2204 *
2205 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2206 * values from/to the user buffer, treated as an ASCII string.
2207 *
2208 * This routine will ensure the values are within the range specified by
2209 * table->extra1 (min) and table->extra2 (max).
2210 *
2211 * Returns 0 on success.
2212 */
2213 int proc_dointvec_minmax(struct ctl_table *table, int write,
2214 void __user *buffer, size_t *lenp, loff_t *ppos)
2215 {
2216 struct do_proc_dointvec_minmax_conv_param param = {
2217 .min = (int *) table->extra1,
2218 .max = (int *) table->extra2,
2219 };
2220 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2221 do_proc_dointvec_minmax_conv, &param);
2222 }
2223
2224 static void validate_coredump_safety(void)
2225 {
2226 #ifdef CONFIG_COREDUMP
2227 if (suid_dumpable == SUID_DUMP_ROOT &&
2228 core_pattern[0] != '/' && core_pattern[0] != '|') {
2229 printk(KERN_WARNING "Unsafe core_pattern used with "\
2230 "suid_dumpable=2. Pipe handler or fully qualified "\
2231 "core dump path required.\n");
2232 }
2233 #endif
2234 }
2235
2236 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2237 void __user *buffer, size_t *lenp, loff_t *ppos)
2238 {
2239 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2240 if (!error)
2241 validate_coredump_safety();
2242 return error;
2243 }
2244
2245 #ifdef CONFIG_COREDUMP
2246 static int proc_dostring_coredump(struct ctl_table *table, int write,
2247 void __user *buffer, size_t *lenp, loff_t *ppos)
2248 {
2249 int error = proc_dostring(table, write, buffer, lenp, ppos);
2250 if (!error)
2251 validate_coredump_safety();
2252 return error;
2253 }
2254 #endif
2255
2256 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2257 void __user *buffer,
2258 size_t *lenp, loff_t *ppos,
2259 unsigned long convmul,
2260 unsigned long convdiv)
2261 {
2262 unsigned long *i, *min, *max;
2263 int vleft, first = 1, err = 0;
2264 unsigned long page = 0;
2265 size_t left;
2266 char *kbuf;
2267
2268 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2269 *lenp = 0;
2270 return 0;
2271 }
2272
2273 i = (unsigned long *) data;
2274 min = (unsigned long *) table->extra1;
2275 max = (unsigned long *) table->extra2;
2276 vleft = table->maxlen / sizeof(unsigned long);
2277 left = *lenp;
2278
2279 if (write) {
2280 if (*ppos) {
2281 switch (sysctl_writes_strict) {
2282 case SYSCTL_WRITES_STRICT:
2283 goto out;
2284 case SYSCTL_WRITES_WARN:
2285 warn_sysctl_write(table);
2286 break;
2287 default:
2288 break;
2289 }
2290 }
2291
2292 if (left > PAGE_SIZE - 1)
2293 left = PAGE_SIZE - 1;
2294 page = __get_free_page(GFP_TEMPORARY);
2295 kbuf = (char *) page;
2296 if (!kbuf)
2297 return -ENOMEM;
2298 if (copy_from_user(kbuf, buffer, left)) {
2299 err = -EFAULT;
2300 goto free;
2301 }
2302 kbuf[left] = 0;
2303 }
2304
2305 for (; left && vleft--; i++, first = 0) {
2306 unsigned long val;
2307
2308 if (write) {
2309 bool neg;
2310
2311 left -= proc_skip_spaces(&kbuf);
2312
2313 err = proc_get_long(&kbuf, &left, &val, &neg,
2314 proc_wspace_sep,
2315 sizeof(proc_wspace_sep), NULL);
2316 if (err)
2317 break;
2318 if (neg)
2319 continue;
2320 if ((min && val < *min) || (max && val > *max))
2321 continue;
2322 *i = val;
2323 } else {
2324 val = convdiv * (*i) / convmul;
2325 if (!first) {
2326 err = proc_put_char(&buffer, &left, '\t');
2327 if (err)
2328 break;
2329 }
2330 err = proc_put_long(&buffer, &left, val, false);
2331 if (err)
2332 break;
2333 }
2334 }
2335
2336 if (!write && !first && left && !err)
2337 err = proc_put_char(&buffer, &left, '\n');
2338 if (write && !err)
2339 left -= proc_skip_spaces(&kbuf);
2340 free:
2341 if (write) {
2342 free_page(page);
2343 if (first)
2344 return err ? : -EINVAL;
2345 }
2346 *lenp -= left;
2347 out:
2348 *ppos += *lenp;
2349 return err;
2350 }
2351
2352 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2353 void __user *buffer,
2354 size_t *lenp, loff_t *ppos,
2355 unsigned long convmul,
2356 unsigned long convdiv)
2357 {
2358 return __do_proc_doulongvec_minmax(table->data, table, write,
2359 buffer, lenp, ppos, convmul, convdiv);
2360 }
2361
2362 /**
2363 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2364 * @table: the sysctl table
2365 * @write: %TRUE if this is a write to the sysctl file
2366 * @buffer: the user buffer
2367 * @lenp: the size of the user buffer
2368 * @ppos: file position
2369 *
2370 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2371 * values from/to the user buffer, treated as an ASCII string.
2372 *
2373 * This routine will ensure the values are within the range specified by
2374 * table->extra1 (min) and table->extra2 (max).
2375 *
2376 * Returns 0 on success.
2377 */
2378 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2379 void __user *buffer, size_t *lenp, loff_t *ppos)
2380 {
2381 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2382 }
2383
2384 /**
2385 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2386 * @table: the sysctl table
2387 * @write: %TRUE if this is a write to the sysctl file
2388 * @buffer: the user buffer
2389 * @lenp: the size of the user buffer
2390 * @ppos: file position
2391 *
2392 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2393 * values from/to the user buffer, treated as an ASCII string. The values
2394 * are treated as milliseconds, and converted to jiffies when they are stored.
2395 *
2396 * This routine will ensure the values are within the range specified by
2397 * table->extra1 (min) and table->extra2 (max).
2398 *
2399 * Returns 0 on success.
2400 */
2401 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2402 void __user *buffer,
2403 size_t *lenp, loff_t *ppos)
2404 {
2405 return do_proc_doulongvec_minmax(table, write, buffer,
2406 lenp, ppos, HZ, 1000l);
2407 }
2408
2409
2410 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2411 int *valp,
2412 int write, void *data)
2413 {
2414 if (write) {
2415 if (*lvalp > LONG_MAX / HZ)
2416 return 1;
2417 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2418 } else {
2419 int val = *valp;
2420 unsigned long lval;
2421 if (val < 0) {
2422 *negp = true;
2423 lval = (unsigned long)-val;
2424 } else {
2425 *negp = false;
2426 lval = (unsigned long)val;
2427 }
2428 *lvalp = lval / HZ;
2429 }
2430 return 0;
2431 }
2432
2433 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2434 int *valp,
2435 int write, void *data)
2436 {
2437 if (write) {
2438 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2439 return 1;
2440 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2441 } else {
2442 int val = *valp;
2443 unsigned long lval;
2444 if (val < 0) {
2445 *negp = true;
2446 lval = (unsigned long)-val;
2447 } else {
2448 *negp = false;
2449 lval = (unsigned long)val;
2450 }
2451 *lvalp = jiffies_to_clock_t(lval);
2452 }
2453 return 0;
2454 }
2455
2456 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2457 int *valp,
2458 int write, void *data)
2459 {
2460 if (write) {
2461 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2462
2463 if (jif > INT_MAX)
2464 return 1;
2465 *valp = (int)jif;
2466 } else {
2467 int val = *valp;
2468 unsigned long lval;
2469 if (val < 0) {
2470 *negp = true;
2471 lval = (unsigned long)-val;
2472 } else {
2473 *negp = false;
2474 lval = (unsigned long)val;
2475 }
2476 *lvalp = jiffies_to_msecs(lval);
2477 }
2478 return 0;
2479 }
2480
2481 /**
2482 * proc_dointvec_jiffies - read a vector of integers as seconds
2483 * @table: the sysctl table
2484 * @write: %TRUE if this is a write to the sysctl file
2485 * @buffer: the user buffer
2486 * @lenp: the size of the user buffer
2487 * @ppos: file position
2488 *
2489 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2490 * values from/to the user buffer, treated as an ASCII string.
2491 * The values read are assumed to be in seconds, and are converted into
2492 * jiffies.
2493 *
2494 * Returns 0 on success.
2495 */
2496 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2497 void __user *buffer, size_t *lenp, loff_t *ppos)
2498 {
2499 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2500 do_proc_dointvec_jiffies_conv,NULL);
2501 }
2502
2503 /**
2504 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2505 * @table: the sysctl table
2506 * @write: %TRUE if this is a write to the sysctl file
2507 * @buffer: the user buffer
2508 * @lenp: the size of the user buffer
2509 * @ppos: pointer to the file position
2510 *
2511 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2512 * values from/to the user buffer, treated as an ASCII string.
2513 * The values read are assumed to be in 1/USER_HZ seconds, and
2514 * are converted into jiffies.
2515 *
2516 * Returns 0 on success.
2517 */
2518 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2519 void __user *buffer, size_t *lenp, loff_t *ppos)
2520 {
2521 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2522 do_proc_dointvec_userhz_jiffies_conv,NULL);
2523 }
2524
2525 /**
2526 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2527 * @table: the sysctl table
2528 * @write: %TRUE if this is a write to the sysctl file
2529 * @buffer: the user buffer
2530 * @lenp: the size of the user buffer
2531 * @ppos: file position
2532 * @ppos: the current position in the file
2533 *
2534 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2535 * values from/to the user buffer, treated as an ASCII string.
2536 * The values read are assumed to be in 1/1000 seconds, and
2537 * are converted into jiffies.
2538 *
2539 * Returns 0 on success.
2540 */
2541 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2542 void __user *buffer, size_t *lenp, loff_t *ppos)
2543 {
2544 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2545 do_proc_dointvec_ms_jiffies_conv, NULL);
2546 }
2547
2548 static int proc_do_cad_pid(struct ctl_table *table, int write,
2549 void __user *buffer, size_t *lenp, loff_t *ppos)
2550 {
2551 struct pid *new_pid;
2552 pid_t tmp;
2553 int r;
2554
2555 tmp = pid_vnr(cad_pid);
2556
2557 r = __do_proc_dointvec(&tmp, table, write, buffer,
2558 lenp, ppos, NULL, NULL);
2559 if (r || !write)
2560 return r;
2561
2562 new_pid = find_get_pid(tmp);
2563 if (!new_pid)
2564 return -ESRCH;
2565
2566 put_pid(xchg(&cad_pid, new_pid));
2567 return 0;
2568 }
2569
2570 /**
2571 * proc_do_large_bitmap - read/write from/to a large bitmap
2572 * @table: the sysctl table
2573 * @write: %TRUE if this is a write to the sysctl file
2574 * @buffer: the user buffer
2575 * @lenp: the size of the user buffer
2576 * @ppos: file position
2577 *
2578 * The bitmap is stored at table->data and the bitmap length (in bits)
2579 * in table->maxlen.
2580 *
2581 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2582 * large bitmaps may be represented in a compact manner. Writing into
2583 * the file will clear the bitmap then update it with the given input.
2584 *
2585 * Returns 0 on success.
2586 */
2587 int proc_do_large_bitmap(struct ctl_table *table, int write,
2588 void __user *buffer, size_t *lenp, loff_t *ppos)
2589 {
2590 int err = 0;
2591 bool first = 1;
2592 size_t left = *lenp;
2593 unsigned long bitmap_len = table->maxlen;
2594 unsigned long *bitmap = *(unsigned long **) table->data;
2595 unsigned long *tmp_bitmap = NULL;
2596 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2597
2598 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2599 *lenp = 0;
2600 return 0;
2601 }
2602
2603 if (write) {
2604 unsigned long page = 0;
2605 char *kbuf;
2606
2607 if (left > PAGE_SIZE - 1)
2608 left = PAGE_SIZE - 1;
2609
2610 page = __get_free_page(GFP_TEMPORARY);
2611 kbuf = (char *) page;
2612 if (!kbuf)
2613 return -ENOMEM;
2614 if (copy_from_user(kbuf, buffer, left)) {
2615 free_page(page);
2616 return -EFAULT;
2617 }
2618 kbuf[left] = 0;
2619
2620 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2621 GFP_KERNEL);
2622 if (!tmp_bitmap) {
2623 free_page(page);
2624 return -ENOMEM;
2625 }
2626 proc_skip_char(&kbuf, &left, '\n');
2627 while (!err && left) {
2628 unsigned long val_a, val_b;
2629 bool neg;
2630
2631 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2632 sizeof(tr_a), &c);
2633 if (err)
2634 break;
2635 if (val_a >= bitmap_len || neg) {
2636 err = -EINVAL;
2637 break;
2638 }
2639
2640 val_b = val_a;
2641 if (left) {
2642 kbuf++;
2643 left--;
2644 }
2645
2646 if (c == '-') {
2647 err = proc_get_long(&kbuf, &left, &val_b,
2648 &neg, tr_b, sizeof(tr_b),
2649 &c);
2650 if (err)
2651 break;
2652 if (val_b >= bitmap_len || neg ||
2653 val_a > val_b) {
2654 err = -EINVAL;
2655 break;
2656 }
2657 if (left) {
2658 kbuf++;
2659 left--;
2660 }
2661 }
2662
2663 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2664 first = 0;
2665 proc_skip_char(&kbuf, &left, '\n');
2666 }
2667 free_page(page);
2668 } else {
2669 unsigned long bit_a, bit_b = 0;
2670
2671 while (left) {
2672 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2673 if (bit_a >= bitmap_len)
2674 break;
2675 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2676 bit_a + 1) - 1;
2677
2678 if (!first) {
2679 err = proc_put_char(&buffer, &left, ',');
2680 if (err)
2681 break;
2682 }
2683 err = proc_put_long(&buffer, &left, bit_a, false);
2684 if (err)
2685 break;
2686 if (bit_a != bit_b) {
2687 err = proc_put_char(&buffer, &left, '-');
2688 if (err)
2689 break;
2690 err = proc_put_long(&buffer, &left, bit_b, false);
2691 if (err)
2692 break;
2693 }
2694
2695 first = 0; bit_b++;
2696 }
2697 if (!err)
2698 err = proc_put_char(&buffer, &left, '\n');
2699 }
2700
2701 if (!err) {
2702 if (write) {
2703 if (*ppos)
2704 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2705 else
2706 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2707 }
2708 kfree(tmp_bitmap);
2709 *lenp -= left;
2710 *ppos += *lenp;
2711 return 0;
2712 } else {
2713 kfree(tmp_bitmap);
2714 return err;
2715 }
2716 }
2717
2718 #else /* CONFIG_PROC_SYSCTL */
2719
2720 int proc_dostring(struct ctl_table *table, int write,
2721 void __user *buffer, size_t *lenp, loff_t *ppos)
2722 {
2723 return -ENOSYS;
2724 }
2725
2726 int proc_dointvec(struct ctl_table *table, int write,
2727 void __user *buffer, size_t *lenp, loff_t *ppos)
2728 {
2729 return -ENOSYS;
2730 }
2731
2732 int proc_dointvec_minmax(struct ctl_table *table, int write,
2733 void __user *buffer, size_t *lenp, loff_t *ppos)
2734 {
2735 return -ENOSYS;
2736 }
2737
2738 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2739 void __user *buffer, size_t *lenp, loff_t *ppos)
2740 {
2741 return -ENOSYS;
2742 }
2743
2744 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2745 void __user *buffer, size_t *lenp, loff_t *ppos)
2746 {
2747 return -ENOSYS;
2748 }
2749
2750 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2751 void __user *buffer, size_t *lenp, loff_t *ppos)
2752 {
2753 return -ENOSYS;
2754 }
2755
2756 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2757 void __user *buffer, size_t *lenp, loff_t *ppos)
2758 {
2759 return -ENOSYS;
2760 }
2761
2762 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2763 void __user *buffer,
2764 size_t *lenp, loff_t *ppos)
2765 {
2766 return -ENOSYS;
2767 }
2768
2769
2770 #endif /* CONFIG_PROC_SYSCTL */
2771
2772 /*
2773 * No sense putting this after each symbol definition, twice,
2774 * exception granted :-)
2775 */
2776 EXPORT_SYMBOL(proc_dointvec);
2777 EXPORT_SYMBOL(proc_dointvec_jiffies);
2778 EXPORT_SYMBOL(proc_dointvec_minmax);
2779 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2780 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2781 EXPORT_SYMBOL(proc_dostring);
2782 EXPORT_SYMBOL(proc_doulongvec_minmax);
2783 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
Linux kernel - Deliverables
This page took 0.091404 seconds and 6 git commands to generate.