projects / deliverable / linux.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Merge tag 'jfs-4.1' of git://github.com/kleikamp/linux-shaggy
[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_dowatchdog,
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_dowatchdog,
860 .extra1 = &zero,
861 .extra2 = &sixty,
862 },
863 {
864 .procname = "softlockup_panic",
865 .data = &softlockup_panic,
866 .maxlen = sizeof(int),
867 .mode = 0644,
868 .proc_handler = proc_dointvec_minmax,
869 .extra1 = &zero,
870 .extra2 = &one,
871 },
872 #ifdef CONFIG_SMP
873 {
874 .procname = "softlockup_all_cpu_backtrace",
875 .data = &sysctl_softlockup_all_cpu_backtrace,
876 .maxlen = sizeof(int),
877 .mode = 0644,
878 .proc_handler = proc_dointvec_minmax,
879 .extra1 = &zero,
880 .extra2 = &one,
881 },
882 #endif /* CONFIG_SMP */
883 {
884 .procname = "nmi_watchdog",
885 .data = &watchdog_user_enabled,
886 .maxlen = sizeof (int),
887 .mode = 0644,
888 .proc_handler = proc_dowatchdog,
889 .extra1 = &zero,
890 .extra2 = &one,
891 },
892 #endif
893 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
894 {
895 .procname = "unknown_nmi_panic",
896 .data = &unknown_nmi_panic,
897 .maxlen = sizeof (int),
898 .mode = 0644,
899 .proc_handler = proc_dointvec,
900 },
901 #endif
902 #if defined(CONFIG_X86)
903 {
904 .procname = "panic_on_unrecovered_nmi",
905 .data = &panic_on_unrecovered_nmi,
906 .maxlen = sizeof(int),
907 .mode = 0644,
908 .proc_handler = proc_dointvec,
909 },
910 {
911 .procname = "panic_on_io_nmi",
912 .data = &panic_on_io_nmi,
913 .maxlen = sizeof(int),
914 .mode = 0644,
915 .proc_handler = proc_dointvec,
916 },
917 #ifdef CONFIG_DEBUG_STACKOVERFLOW
918 {
919 .procname = "panic_on_stackoverflow",
920 .data = &sysctl_panic_on_stackoverflow,
921 .maxlen = sizeof(int),
922 .mode = 0644,
923 .proc_handler = proc_dointvec,
924 },
925 #endif
926 {
927 .procname = "bootloader_type",
928 .data = &bootloader_type,
929 .maxlen = sizeof (int),
930 .mode = 0444,
931 .proc_handler = proc_dointvec,
932 },
933 {
934 .procname = "bootloader_version",
935 .data = &bootloader_version,
936 .maxlen = sizeof (int),
937 .mode = 0444,
938 .proc_handler = proc_dointvec,
939 },
940 {
941 .procname = "kstack_depth_to_print",
942 .data = &kstack_depth_to_print,
943 .maxlen = sizeof(int),
944 .mode = 0644,
945 .proc_handler = proc_dointvec,
946 },
947 {
948 .procname = "io_delay_type",
949 .data = &io_delay_type,
950 .maxlen = sizeof(int),
951 .mode = 0644,
952 .proc_handler = proc_dointvec,
953 },
954 #endif
955 #if defined(CONFIG_MMU)
956 {
957 .procname = "randomize_va_space",
958 .data = &randomize_va_space,
959 .maxlen = sizeof(int),
960 .mode = 0644,
961 .proc_handler = proc_dointvec,
962 },
963 #endif
964 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
965 {
966 .procname = "spin_retry",
967 .data = &spin_retry,
968 .maxlen = sizeof (int),
969 .mode = 0644,
970 .proc_handler = proc_dointvec,
971 },
972 #endif
973 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
974 {
975 .procname = "acpi_video_flags",
976 .data = &acpi_realmode_flags,
977 .maxlen = sizeof (unsigned long),
978 .mode = 0644,
979 .proc_handler = proc_doulongvec_minmax,
980 },
981 #endif
982 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
983 {
984 .procname = "ignore-unaligned-usertrap",
985 .data = &no_unaligned_warning,
986 .maxlen = sizeof (int),
987 .mode = 0644,
988 .proc_handler = proc_dointvec,
989 },
990 #endif
991 #ifdef CONFIG_IA64
992 {
993 .procname = "unaligned-dump-stack",
994 .data = &unaligned_dump_stack,
995 .maxlen = sizeof (int),
996 .mode = 0644,
997 .proc_handler = proc_dointvec,
998 },
999 #endif
1000 #ifdef CONFIG_DETECT_HUNG_TASK
1001 {
1002 .procname = "hung_task_panic",
1003 .data = &sysctl_hung_task_panic,
1004 .maxlen = sizeof(int),
1005 .mode = 0644,
1006 .proc_handler = proc_dointvec_minmax,
1007 .extra1 = &zero,
1008 .extra2 = &one,
1009 },
1010 {
1011 .procname = "hung_task_check_count",
1012 .data = &sysctl_hung_task_check_count,
1013 .maxlen = sizeof(int),
1014 .mode = 0644,
1015 .proc_handler = proc_dointvec_minmax,
1016 .extra1 = &zero,
1017 },
1018 {
1019 .procname = "hung_task_timeout_secs",
1020 .data = &sysctl_hung_task_timeout_secs,
1021 .maxlen = sizeof(unsigned long),
1022 .mode = 0644,
1023 .proc_handler = proc_dohung_task_timeout_secs,
1024 .extra2 = &hung_task_timeout_max,
1025 },
1026 {
1027 .procname = "hung_task_warnings",
1028 .data = &sysctl_hung_task_warnings,
1029 .maxlen = sizeof(int),
1030 .mode = 0644,
1031 .proc_handler = proc_dointvec_minmax,
1032 .extra1 = &neg_one,
1033 },
1034 #endif
1035 #ifdef CONFIG_COMPAT
1036 {
1037 .procname = "compat-log",
1038 .data = &compat_log,
1039 .maxlen = sizeof (int),
1040 .mode = 0644,
1041 .proc_handler = proc_dointvec,
1042 },
1043 #endif
1044 #ifdef CONFIG_RT_MUTEXES
1045 {
1046 .procname = "max_lock_depth",
1047 .data = &max_lock_depth,
1048 .maxlen = sizeof(int),
1049 .mode = 0644,
1050 .proc_handler = proc_dointvec,
1051 },
1052 #endif
1053 {
1054 .procname = "poweroff_cmd",
1055 .data = &poweroff_cmd,
1056 .maxlen = POWEROFF_CMD_PATH_LEN,
1057 .mode = 0644,
1058 .proc_handler = proc_dostring,
1059 },
1060 #ifdef CONFIG_KEYS
1061 {
1062 .procname = "keys",
1063 .mode = 0555,
1064 .child = key_sysctls,
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 .procname = "panic_on_warn",
1117 .data = &panic_on_warn,
1118 .maxlen = sizeof(int),
1119 .mode = 0644,
1120 .proc_handler = proc_dointvec_minmax,
1121 .extra1 = &zero,
1122 .extra2 = &one,
1123 },
1124 { }
1125 };
1126
1127 static struct ctl_table vm_table[] = {
1128 {
1129 .procname = "overcommit_memory",
1130 .data = &sysctl_overcommit_memory,
1131 .maxlen = sizeof(sysctl_overcommit_memory),
1132 .mode = 0644,
1133 .proc_handler = proc_dointvec_minmax,
1134 .extra1 = &zero,
1135 .extra2 = &two,
1136 },
1137 {
1138 .procname = "panic_on_oom",
1139 .data = &sysctl_panic_on_oom,
1140 .maxlen = sizeof(sysctl_panic_on_oom),
1141 .mode = 0644,
1142 .proc_handler = proc_dointvec_minmax,
1143 .extra1 = &zero,
1144 .extra2 = &two,
1145 },
1146 {
1147 .procname = "oom_kill_allocating_task",
1148 .data = &sysctl_oom_kill_allocating_task,
1149 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1150 .mode = 0644,
1151 .proc_handler = proc_dointvec,
1152 },
1153 {
1154 .procname = "oom_dump_tasks",
1155 .data = &sysctl_oom_dump_tasks,
1156 .maxlen = sizeof(sysctl_oom_dump_tasks),
1157 .mode = 0644,
1158 .proc_handler = proc_dointvec,
1159 },
1160 {
1161 .procname = "overcommit_ratio",
1162 .data = &sysctl_overcommit_ratio,
1163 .maxlen = sizeof(sysctl_overcommit_ratio),
1164 .mode = 0644,
1165 .proc_handler = overcommit_ratio_handler,
1166 },
1167 {
1168 .procname = "overcommit_kbytes",
1169 .data = &sysctl_overcommit_kbytes,
1170 .maxlen = sizeof(sysctl_overcommit_kbytes),
1171 .mode = 0644,
1172 .proc_handler = overcommit_kbytes_handler,
1173 },
1174 {
1175 .procname = "page-cluster",
1176 .data = &page_cluster,
1177 .maxlen = sizeof(int),
1178 .mode = 0644,
1179 .proc_handler = proc_dointvec_minmax,
1180 .extra1 = &zero,
1181 },
1182 {
1183 .procname = "dirty_background_ratio",
1184 .data = &dirty_background_ratio,
1185 .maxlen = sizeof(dirty_background_ratio),
1186 .mode = 0644,
1187 .proc_handler = dirty_background_ratio_handler,
1188 .extra1 = &zero,
1189 .extra2 = &one_hundred,
1190 },
1191 {
1192 .procname = "dirty_background_bytes",
1193 .data = &dirty_background_bytes,
1194 .maxlen = sizeof(dirty_background_bytes),
1195 .mode = 0644,
1196 .proc_handler = dirty_background_bytes_handler,
1197 .extra1 = &one_ul,
1198 },
1199 {
1200 .procname = "dirty_ratio",
1201 .data = &vm_dirty_ratio,
1202 .maxlen = sizeof(vm_dirty_ratio),
1203 .mode = 0644,
1204 .proc_handler = dirty_ratio_handler,
1205 .extra1 = &zero,
1206 .extra2 = &one_hundred,
1207 },
1208 {
1209 .procname = "dirty_bytes",
1210 .data = &vm_dirty_bytes,
1211 .maxlen = sizeof(vm_dirty_bytes),
1212 .mode = 0644,
1213 .proc_handler = dirty_bytes_handler,
1214 .extra1 = &dirty_bytes_min,
1215 },
1216 {
1217 .procname = "dirty_writeback_centisecs",
1218 .data = &dirty_writeback_interval,
1219 .maxlen = sizeof(dirty_writeback_interval),
1220 .mode = 0644,
1221 .proc_handler = dirty_writeback_centisecs_handler,
1222 },
1223 {
1224 .procname = "dirty_expire_centisecs",
1225 .data = &dirty_expire_interval,
1226 .maxlen = sizeof(dirty_expire_interval),
1227 .mode = 0644,
1228 .proc_handler = proc_dointvec_minmax,
1229 .extra1 = &zero,
1230 },
1231 {
1232 .procname = "dirtytime_expire_seconds",
1233 .data = &dirtytime_expire_interval,
1234 .maxlen = sizeof(dirty_expire_interval),
1235 .mode = 0644,
1236 .proc_handler = dirtytime_interval_handler,
1237 .extra1 = &zero,
1238 },
1239 {
1240 .procname = "nr_pdflush_threads",
1241 .mode = 0444 /* read-only */,
1242 .proc_handler = pdflush_proc_obsolete,
1243 },
1244 {
1245 .procname = "swappiness",
1246 .data = &vm_swappiness,
1247 .maxlen = sizeof(vm_swappiness),
1248 .mode = 0644,
1249 .proc_handler = proc_dointvec_minmax,
1250 .extra1 = &zero,
1251 .extra2 = &one_hundred,
1252 },
1253 #ifdef CONFIG_HUGETLB_PAGE
1254 {
1255 .procname = "nr_hugepages",
1256 .data = NULL,
1257 .maxlen = sizeof(unsigned long),
1258 .mode = 0644,
1259 .proc_handler = hugetlb_sysctl_handler,
1260 },
1261 #ifdef CONFIG_NUMA
1262 {
1263 .procname = "nr_hugepages_mempolicy",
1264 .data = NULL,
1265 .maxlen = sizeof(unsigned long),
1266 .mode = 0644,
1267 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1268 },
1269 #endif
1270 {
1271 .procname = "hugetlb_shm_group",
1272 .data = &sysctl_hugetlb_shm_group,
1273 .maxlen = sizeof(gid_t),
1274 .mode = 0644,
1275 .proc_handler = proc_dointvec,
1276 },
1277 {
1278 .procname = "hugepages_treat_as_movable",
1279 .data = &hugepages_treat_as_movable,
1280 .maxlen = sizeof(int),
1281 .mode = 0644,
1282 .proc_handler = proc_dointvec,
1283 },
1284 {
1285 .procname = "nr_overcommit_hugepages",
1286 .data = NULL,
1287 .maxlen = sizeof(unsigned long),
1288 .mode = 0644,
1289 .proc_handler = hugetlb_overcommit_handler,
1290 },
1291 #endif
1292 {
1293 .procname = "lowmem_reserve_ratio",
1294 .data = &sysctl_lowmem_reserve_ratio,
1295 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1296 .mode = 0644,
1297 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1298 },
1299 {
1300 .procname = "drop_caches",
1301 .data = &sysctl_drop_caches,
1302 .maxlen = sizeof(int),
1303 .mode = 0644,
1304 .proc_handler = drop_caches_sysctl_handler,
1305 .extra1 = &one,
1306 .extra2 = &four,
1307 },
1308 #ifdef CONFIG_COMPACTION
1309 {
1310 .procname = "compact_memory",
1311 .data = &sysctl_compact_memory,
1312 .maxlen = sizeof(int),
1313 .mode = 0200,
1314 .proc_handler = sysctl_compaction_handler,
1315 },
1316 {
1317 .procname = "extfrag_threshold",
1318 .data = &sysctl_extfrag_threshold,
1319 .maxlen = sizeof(int),
1320 .mode = 0644,
1321 .proc_handler = sysctl_extfrag_handler,
1322 .extra1 = &min_extfrag_threshold,
1323 .extra2 = &max_extfrag_threshold,
1324 },
1325
1326 #endif /* CONFIG_COMPACTION */
1327 {
1328 .procname = "min_free_kbytes",
1329 .data = &min_free_kbytes,
1330 .maxlen = sizeof(min_free_kbytes),
1331 .mode = 0644,
1332 .proc_handler = min_free_kbytes_sysctl_handler,
1333 .extra1 = &zero,
1334 },
1335 {
1336 .procname = "percpu_pagelist_fraction",
1337 .data = &percpu_pagelist_fraction,
1338 .maxlen = sizeof(percpu_pagelist_fraction),
1339 .mode = 0644,
1340 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1341 .extra1 = &zero,
1342 },
1343 #ifdef CONFIG_MMU
1344 {
1345 .procname = "max_map_count",
1346 .data = &sysctl_max_map_count,
1347 .maxlen = sizeof(sysctl_max_map_count),
1348 .mode = 0644,
1349 .proc_handler = proc_dointvec_minmax,
1350 .extra1 = &zero,
1351 },
1352 #else
1353 {
1354 .procname = "nr_trim_pages",
1355 .data = &sysctl_nr_trim_pages,
1356 .maxlen = sizeof(sysctl_nr_trim_pages),
1357 .mode = 0644,
1358 .proc_handler = proc_dointvec_minmax,
1359 .extra1 = &zero,
1360 },
1361 #endif
1362 {
1363 .procname = "laptop_mode",
1364 .data = &laptop_mode,
1365 .maxlen = sizeof(laptop_mode),
1366 .mode = 0644,
1367 .proc_handler = proc_dointvec_jiffies,
1368 },
1369 {
1370 .procname = "block_dump",
1371 .data = &block_dump,
1372 .maxlen = sizeof(block_dump),
1373 .mode = 0644,
1374 .proc_handler = proc_dointvec,
1375 .extra1 = &zero,
1376 },
1377 {
1378 .procname = "vfs_cache_pressure",
1379 .data = &sysctl_vfs_cache_pressure,
1380 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1381 .mode = 0644,
1382 .proc_handler = proc_dointvec,
1383 .extra1 = &zero,
1384 },
1385 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1386 {
1387 .procname = "legacy_va_layout",
1388 .data = &sysctl_legacy_va_layout,
1389 .maxlen = sizeof(sysctl_legacy_va_layout),
1390 .mode = 0644,
1391 .proc_handler = proc_dointvec,
1392 .extra1 = &zero,
1393 },
1394 #endif
1395 #ifdef CONFIG_NUMA
1396 {
1397 .procname = "zone_reclaim_mode",
1398 .data = &zone_reclaim_mode,
1399 .maxlen = sizeof(zone_reclaim_mode),
1400 .mode = 0644,
1401 .proc_handler = proc_dointvec,
1402 .extra1 = &zero,
1403 },
1404 {
1405 .procname = "min_unmapped_ratio",
1406 .data = &sysctl_min_unmapped_ratio,
1407 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1408 .mode = 0644,
1409 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1410 .extra1 = &zero,
1411 .extra2 = &one_hundred,
1412 },
1413 {
1414 .procname = "min_slab_ratio",
1415 .data = &sysctl_min_slab_ratio,
1416 .maxlen = sizeof(sysctl_min_slab_ratio),
1417 .mode = 0644,
1418 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1419 .extra1 = &zero,
1420 .extra2 = &one_hundred,
1421 },
1422 #endif
1423 #ifdef CONFIG_SMP
1424 {
1425 .procname = "stat_interval",
1426 .data = &sysctl_stat_interval,
1427 .maxlen = sizeof(sysctl_stat_interval),
1428 .mode = 0644,
1429 .proc_handler = proc_dointvec_jiffies,
1430 },
1431 #endif
1432 #ifdef CONFIG_MMU
1433 {
1434 .procname = "mmap_min_addr",
1435 .data = &dac_mmap_min_addr,
1436 .maxlen = sizeof(unsigned long),
1437 .mode = 0644,
1438 .proc_handler = mmap_min_addr_handler,
1439 },
1440 #endif
1441 #ifdef CONFIG_NUMA
1442 {
1443 .procname = "numa_zonelist_order",
1444 .data = &numa_zonelist_order,
1445 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1446 .mode = 0644,
1447 .proc_handler = numa_zonelist_order_handler,
1448 },
1449 #endif
1450 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1451 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1452 {
1453 .procname = "vdso_enabled",
1454 #ifdef CONFIG_X86_32
1455 .data = &vdso32_enabled,
1456 .maxlen = sizeof(vdso32_enabled),
1457 #else
1458 .data = &vdso_enabled,
1459 .maxlen = sizeof(vdso_enabled),
1460 #endif
1461 .mode = 0644,
1462 .proc_handler = proc_dointvec,
1463 .extra1 = &zero,
1464 },
1465 #endif
1466 #ifdef CONFIG_HIGHMEM
1467 {
1468 .procname = "highmem_is_dirtyable",
1469 .data = &vm_highmem_is_dirtyable,
1470 .maxlen = sizeof(vm_highmem_is_dirtyable),
1471 .mode = 0644,
1472 .proc_handler = proc_dointvec_minmax,
1473 .extra1 = &zero,
1474 .extra2 = &one,
1475 },
1476 #endif
1477 #ifdef CONFIG_MEMORY_FAILURE
1478 {
1479 .procname = "memory_failure_early_kill",
1480 .data = &sysctl_memory_failure_early_kill,
1481 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1482 .mode = 0644,
1483 .proc_handler = proc_dointvec_minmax,
1484 .extra1 = &zero,
1485 .extra2 = &one,
1486 },
1487 {
1488 .procname = "memory_failure_recovery",
1489 .data = &sysctl_memory_failure_recovery,
1490 .maxlen = sizeof(sysctl_memory_failure_recovery),
1491 .mode = 0644,
1492 .proc_handler = proc_dointvec_minmax,
1493 .extra1 = &zero,
1494 .extra2 = &one,
1495 },
1496 #endif
1497 {
1498 .procname = "user_reserve_kbytes",
1499 .data = &sysctl_user_reserve_kbytes,
1500 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1501 .mode = 0644,
1502 .proc_handler = proc_doulongvec_minmax,
1503 },
1504 {
1505 .procname = "admin_reserve_kbytes",
1506 .data = &sysctl_admin_reserve_kbytes,
1507 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1508 .mode = 0644,
1509 .proc_handler = proc_doulongvec_minmax,
1510 },
1511 { }
1512 };
1513
1514 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1515 static struct ctl_table binfmt_misc_table[] = {
1516 { }
1517 };
1518 #endif
1519
1520 static struct ctl_table fs_table[] = {
1521 {
1522 .procname = "inode-nr",
1523 .data = &inodes_stat,
1524 .maxlen = 2*sizeof(long),
1525 .mode = 0444,
1526 .proc_handler = proc_nr_inodes,
1527 },
1528 {
1529 .procname = "inode-state",
1530 .data = &inodes_stat,
1531 .maxlen = 7*sizeof(long),
1532 .mode = 0444,
1533 .proc_handler = proc_nr_inodes,
1534 },
1535 {
1536 .procname = "file-nr",
1537 .data = &files_stat,
1538 .maxlen = sizeof(files_stat),
1539 .mode = 0444,
1540 .proc_handler = proc_nr_files,
1541 },
1542 {
1543 .procname = "file-max",
1544 .data = &files_stat.max_files,
1545 .maxlen = sizeof(files_stat.max_files),
1546 .mode = 0644,
1547 .proc_handler = proc_doulongvec_minmax,
1548 },
1549 {
1550 .procname = "nr_open",
1551 .data = &sysctl_nr_open,
1552 .maxlen = sizeof(int),
1553 .mode = 0644,
1554 .proc_handler = proc_dointvec_minmax,
1555 .extra1 = &sysctl_nr_open_min,
1556 .extra2 = &sysctl_nr_open_max,
1557 },
1558 {
1559 .procname = "dentry-state",
1560 .data = &dentry_stat,
1561 .maxlen = 6*sizeof(long),
1562 .mode = 0444,
1563 .proc_handler = proc_nr_dentry,
1564 },
1565 {
1566 .procname = "overflowuid",
1567 .data = &fs_overflowuid,
1568 .maxlen = sizeof(int),
1569 .mode = 0644,
1570 .proc_handler = proc_dointvec_minmax,
1571 .extra1 = &minolduid,
1572 .extra2 = &maxolduid,
1573 },
1574 {
1575 .procname = "overflowgid",
1576 .data = &fs_overflowgid,
1577 .maxlen = sizeof(int),
1578 .mode = 0644,
1579 .proc_handler = proc_dointvec_minmax,
1580 .extra1 = &minolduid,
1581 .extra2 = &maxolduid,
1582 },
1583 #ifdef CONFIG_FILE_LOCKING
1584 {
1585 .procname = "leases-enable",
1586 .data = &leases_enable,
1587 .maxlen = sizeof(int),
1588 .mode = 0644,
1589 .proc_handler = proc_dointvec,
1590 },
1591 #endif
1592 #ifdef CONFIG_DNOTIFY
1593 {
1594 .procname = "dir-notify-enable",
1595 .data = &dir_notify_enable,
1596 .maxlen = sizeof(int),
1597 .mode = 0644,
1598 .proc_handler = proc_dointvec,
1599 },
1600 #endif
1601 #ifdef CONFIG_MMU
1602 #ifdef CONFIG_FILE_LOCKING
1603 {
1604 .procname = "lease-break-time",
1605 .data = &lease_break_time,
1606 .maxlen = sizeof(int),
1607 .mode = 0644,
1608 .proc_handler = proc_dointvec,
1609 },
1610 #endif
1611 #ifdef CONFIG_AIO
1612 {
1613 .procname = "aio-nr",
1614 .data = &aio_nr,
1615 .maxlen = sizeof(aio_nr),
1616 .mode = 0444,
1617 .proc_handler = proc_doulongvec_minmax,
1618 },
1619 {
1620 .procname = "aio-max-nr",
1621 .data = &aio_max_nr,
1622 .maxlen = sizeof(aio_max_nr),
1623 .mode = 0644,
1624 .proc_handler = proc_doulongvec_minmax,
1625 },
1626 #endif /* CONFIG_AIO */
1627 #ifdef CONFIG_INOTIFY_USER
1628 {
1629 .procname = "inotify",
1630 .mode = 0555,
1631 .child = inotify_table,
1632 },
1633 #endif
1634 #ifdef CONFIG_EPOLL
1635 {
1636 .procname = "epoll",
1637 .mode = 0555,
1638 .child = epoll_table,
1639 },
1640 #endif
1641 #endif
1642 {
1643 .procname = "protected_symlinks",
1644 .data = &sysctl_protected_symlinks,
1645 .maxlen = sizeof(int),
1646 .mode = 0600,
1647 .proc_handler = proc_dointvec_minmax,
1648 .extra1 = &zero,
1649 .extra2 = &one,
1650 },
1651 {
1652 .procname = "protected_hardlinks",
1653 .data = &sysctl_protected_hardlinks,
1654 .maxlen = sizeof(int),
1655 .mode = 0600,
1656 .proc_handler = proc_dointvec_minmax,
1657 .extra1 = &zero,
1658 .extra2 = &one,
1659 },
1660 {
1661 .procname = "suid_dumpable",
1662 .data = &suid_dumpable,
1663 .maxlen = sizeof(int),
1664 .mode = 0644,
1665 .proc_handler = proc_dointvec_minmax_coredump,
1666 .extra1 = &zero,
1667 .extra2 = &two,
1668 },
1669 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1670 {
1671 .procname = "binfmt_misc",
1672 .mode = 0555,
1673 .child = binfmt_misc_table,
1674 },
1675 #endif
1676 {
1677 .procname = "pipe-max-size",
1678 .data = &pipe_max_size,
1679 .maxlen = sizeof(int),
1680 .mode = 0644,
1681 .proc_handler = &pipe_proc_fn,
1682 .extra1 = &pipe_min_size,
1683 },
1684 { }
1685 };
1686
1687 static struct ctl_table debug_table[] = {
1688 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1689 {
1690 .procname = "exception-trace",
1691 .data = &show_unhandled_signals,
1692 .maxlen = sizeof(int),
1693 .mode = 0644,
1694 .proc_handler = proc_dointvec
1695 },
1696 #endif
1697 #if defined(CONFIG_OPTPROBES)
1698 {
1699 .procname = "kprobes-optimization",
1700 .data = &sysctl_kprobes_optimization,
1701 .maxlen = sizeof(int),
1702 .mode = 0644,
1703 .proc_handler = proc_kprobes_optimization_handler,
1704 .extra1 = &zero,
1705 .extra2 = &one,
1706 },
1707 #endif
1708 { }
1709 };
1710
1711 static struct ctl_table dev_table[] = {
1712 { }
1713 };
1714
1715 int __init sysctl_init(void)
1716 {
1717 struct ctl_table_header *hdr;
1718
1719 hdr = register_sysctl_table(sysctl_base_table);
1720 kmemleak_not_leak(hdr);
1721 return 0;
1722 }
1723
1724 #endif /* CONFIG_SYSCTL */
1725
1726 /*
1727 * /proc/sys support
1728 */
1729
1730 #ifdef CONFIG_PROC_SYSCTL
1731
1732 static int _proc_do_string(char *data, int maxlen, int write,
1733 char __user *buffer,
1734 size_t *lenp, loff_t *ppos)
1735 {
1736 size_t len;
1737 char __user *p;
1738 char c;
1739
1740 if (!data || !maxlen || !*lenp) {
1741 *lenp = 0;
1742 return 0;
1743 }
1744
1745 if (write) {
1746 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1747 /* Only continue writes not past the end of buffer. */
1748 len = strlen(data);
1749 if (len > maxlen - 1)
1750 len = maxlen - 1;
1751
1752 if (*ppos > len)
1753 return 0;
1754 len = *ppos;
1755 } else {
1756 /* Start writing from beginning of buffer. */
1757 len = 0;
1758 }
1759
1760 *ppos += *lenp;
1761 p = buffer;
1762 while ((p - buffer) < *lenp && len < maxlen - 1) {
1763 if (get_user(c, p++))
1764 return -EFAULT;
1765 if (c == 0 || c == '\n')
1766 break;
1767 data[len++] = c;
1768 }
1769 data[len] = 0;
1770 } else {
1771 len = strlen(data);
1772 if (len > maxlen)
1773 len = maxlen;
1774
1775 if (*ppos > len) {
1776 *lenp = 0;
1777 return 0;
1778 }
1779
1780 data += *ppos;
1781 len -= *ppos;
1782
1783 if (len > *lenp)
1784 len = *lenp;
1785 if (len)
1786 if (copy_to_user(buffer, data, len))
1787 return -EFAULT;
1788 if (len < *lenp) {
1789 if (put_user('\n', buffer + len))
1790 return -EFAULT;
1791 len++;
1792 }
1793 *lenp = len;
1794 *ppos += len;
1795 }
1796 return 0;
1797 }
1798
1799 static void warn_sysctl_write(struct ctl_table *table)
1800 {
1801 pr_warn_once("%s wrote to %s when file position was not 0!\n"
1802 "This will not be supported in the future. To silence this\n"
1803 "warning, set kernel.sysctl_writes_strict = -1\n",
1804 current->comm, table->procname);
1805 }
1806
1807 /**
1808 * proc_dostring - read a string sysctl
1809 * @table: the sysctl table
1810 * @write: %TRUE if this is a write to the sysctl file
1811 * @buffer: the user buffer
1812 * @lenp: the size of the user buffer
1813 * @ppos: file position
1814 *
1815 * Reads/writes a string from/to the user buffer. If the kernel
1816 * buffer provided is not large enough to hold the string, the
1817 * string is truncated. The copied string is %NULL-terminated.
1818 * If the string is being read by the user process, it is copied
1819 * and a newline '\n' is added. It is truncated if the buffer is
1820 * not large enough.
1821 *
1822 * Returns 0 on success.
1823 */
1824 int proc_dostring(struct ctl_table *table, int write,
1825 void __user *buffer, size_t *lenp, loff_t *ppos)
1826 {
1827 if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN)
1828 warn_sysctl_write(table);
1829
1830 return _proc_do_string((char *)(table->data), table->maxlen, write,
1831 (char __user *)buffer, lenp, ppos);
1832 }
1833
1834 static size_t proc_skip_spaces(char **buf)
1835 {
1836 size_t ret;
1837 char *tmp = skip_spaces(*buf);
1838 ret = tmp - *buf;
1839 *buf = tmp;
1840 return ret;
1841 }
1842
1843 static void proc_skip_char(char **buf, size_t *size, const char v)
1844 {
1845 while (*size) {
1846 if (**buf != v)
1847 break;
1848 (*size)--;
1849 (*buf)++;
1850 }
1851 }
1852
1853 #define TMPBUFLEN 22
1854 /**
1855 * proc_get_long - reads an ASCII formatted integer from a user buffer
1856 *
1857 * @buf: a kernel buffer
1858 * @size: size of the kernel buffer
1859 * @val: this is where the number will be stored
1860 * @neg: set to %TRUE if number is negative
1861 * @perm_tr: a vector which contains the allowed trailers
1862 * @perm_tr_len: size of the perm_tr vector
1863 * @tr: pointer to store the trailer character
1864 *
1865 * In case of success %0 is returned and @buf and @size are updated with
1866 * the amount of bytes read. If @tr is non-NULL and a trailing
1867 * character exists (size is non-zero after returning from this
1868 * function), @tr is updated with the trailing character.
1869 */
1870 static int proc_get_long(char **buf, size_t *size,
1871 unsigned long *val, bool *neg,
1872 const char *perm_tr, unsigned perm_tr_len, char *tr)
1873 {
1874 int len;
1875 char *p, tmp[TMPBUFLEN];
1876
1877 if (!*size)
1878 return -EINVAL;
1879
1880 len = *size;
1881 if (len > TMPBUFLEN - 1)
1882 len = TMPBUFLEN - 1;
1883
1884 memcpy(tmp, *buf, len);
1885
1886 tmp[len] = 0;
1887 p = tmp;
1888 if (*p == '-' && *size > 1) {
1889 *neg = true;
1890 p++;
1891 } else
1892 *neg = false;
1893 if (!isdigit(*p))
1894 return -EINVAL;
1895
1896 *val = simple_strtoul(p, &p, 0);
1897
1898 len = p - tmp;
1899
1900 /* We don't know if the next char is whitespace thus we may accept
1901 * invalid integers (e.g. 1234...a) or two integers instead of one
1902 * (e.g. 123...1). So lets not allow such large numbers. */
1903 if (len == TMPBUFLEN - 1)
1904 return -EINVAL;
1905
1906 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
1907 return -EINVAL;
1908
1909 if (tr && (len < *size))
1910 *tr = *p;
1911
1912 *buf += len;
1913 *size -= len;
1914
1915 return 0;
1916 }
1917
1918 /**
1919 * proc_put_long - converts an integer to a decimal ASCII formatted string
1920 *
1921 * @buf: the user buffer
1922 * @size: the size of the user buffer
1923 * @val: the integer to be converted
1924 * @neg: sign of the number, %TRUE for negative
1925 *
1926 * In case of success %0 is returned and @buf and @size are updated with
1927 * the amount of bytes written.
1928 */
1929 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
1930 bool neg)
1931 {
1932 int len;
1933 char tmp[TMPBUFLEN], *p = tmp;
1934
1935 sprintf(p, "%s%lu", neg ? "-" : "", val);
1936 len = strlen(tmp);
1937 if (len > *size)
1938 len = *size;
1939 if (copy_to_user(*buf, tmp, len))
1940 return -EFAULT;
1941 *size -= len;
1942 *buf += len;
1943 return 0;
1944 }
1945 #undef TMPBUFLEN
1946
1947 static int proc_put_char(void __user **buf, size_t *size, char c)
1948 {
1949 if (*size) {
1950 char __user **buffer = (char __user **)buf;
1951 if (put_user(c, *buffer))
1952 return -EFAULT;
1953 (*size)--, (*buffer)++;
1954 *buf = *buffer;
1955 }
1956 return 0;
1957 }
1958
1959 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
1960 int *valp,
1961 int write, void *data)
1962 {
1963 if (write) {
1964 *valp = *negp ? -*lvalp : *lvalp;
1965 } else {
1966 int val = *valp;
1967 if (val < 0) {
1968 *negp = true;
1969 *lvalp = (unsigned long)-val;
1970 } else {
1971 *negp = false;
1972 *lvalp = (unsigned long)val;
1973 }
1974 }
1975 return 0;
1976 }
1977
1978 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
1979
1980 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1981 int write, void __user *buffer,
1982 size_t *lenp, loff_t *ppos,
1983 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1984 int write, void *data),
1985 void *data)
1986 {
1987 int *i, vleft, first = 1, err = 0;
1988 unsigned long page = 0;
1989 size_t left;
1990 char *kbuf;
1991
1992 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
1993 *lenp = 0;
1994 return 0;
1995 }
1996
1997 i = (int *) tbl_data;
1998 vleft = table->maxlen / sizeof(*i);
1999 left = *lenp;
2000
2001 if (!conv)
2002 conv = do_proc_dointvec_conv;
2003
2004 if (write) {
2005 if (*ppos) {
2006 switch (sysctl_writes_strict) {
2007 case SYSCTL_WRITES_STRICT:
2008 goto out;
2009 case SYSCTL_WRITES_WARN:
2010 warn_sysctl_write(table);
2011 break;
2012 default:
2013 break;
2014 }
2015 }
2016
2017 if (left > PAGE_SIZE - 1)
2018 left = PAGE_SIZE - 1;
2019 page = __get_free_page(GFP_TEMPORARY);
2020 kbuf = (char *) page;
2021 if (!kbuf)
2022 return -ENOMEM;
2023 if (copy_from_user(kbuf, buffer, left)) {
2024 err = -EFAULT;
2025 goto free;
2026 }
2027 kbuf[left] = 0;
2028 }
2029
2030 for (; left && vleft--; i++, first=0) {
2031 unsigned long lval;
2032 bool neg;
2033
2034 if (write) {
2035 left -= proc_skip_spaces(&kbuf);
2036
2037 if (!left)
2038 break;
2039 err = proc_get_long(&kbuf, &left, &lval, &neg,
2040 proc_wspace_sep,
2041 sizeof(proc_wspace_sep), NULL);
2042 if (err)
2043 break;
2044 if (conv(&neg, &lval, i, 1, data)) {
2045 err = -EINVAL;
2046 break;
2047 }
2048 } else {
2049 if (conv(&neg, &lval, i, 0, data)) {
2050 err = -EINVAL;
2051 break;
2052 }
2053 if (!first)
2054 err = proc_put_char(&buffer, &left, '\t');
2055 if (err)
2056 break;
2057 err = proc_put_long(&buffer, &left, lval, neg);
2058 if (err)
2059 break;
2060 }
2061 }
2062
2063 if (!write && !first && left && !err)
2064 err = proc_put_char(&buffer, &left, '\n');
2065 if (write && !err && left)
2066 left -= proc_skip_spaces(&kbuf);
2067 free:
2068 if (write) {
2069 free_page(page);
2070 if (first)
2071 return err ? : -EINVAL;
2072 }
2073 *lenp -= left;
2074 out:
2075 *ppos += *lenp;
2076 return err;
2077 }
2078
2079 static int do_proc_dointvec(struct ctl_table *table, int write,
2080 void __user *buffer, size_t *lenp, loff_t *ppos,
2081 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2082 int write, void *data),
2083 void *data)
2084 {
2085 return __do_proc_dointvec(table->data, table, write,
2086 buffer, lenp, ppos, conv, data);
2087 }
2088
2089 /**
2090 * proc_dointvec - read a vector of integers
2091 * @table: the sysctl table
2092 * @write: %TRUE if this is a write to the sysctl file
2093 * @buffer: the user buffer
2094 * @lenp: the size of the user buffer
2095 * @ppos: file position
2096 *
2097 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2098 * values from/to the user buffer, treated as an ASCII string.
2099 *
2100 * Returns 0 on success.
2101 */
2102 int proc_dointvec(struct ctl_table *table, int write,
2103 void __user *buffer, size_t *lenp, loff_t *ppos)
2104 {
2105 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2106 NULL,NULL);
2107 }
2108
2109 /*
2110 * Taint values can only be increased
2111 * This means we can safely use a temporary.
2112 */
2113 static int proc_taint(struct ctl_table *table, int write,
2114 void __user *buffer, size_t *lenp, loff_t *ppos)
2115 {
2116 struct ctl_table t;
2117 unsigned long tmptaint = get_taint();
2118 int err;
2119
2120 if (write && !capable(CAP_SYS_ADMIN))
2121 return -EPERM;
2122
2123 t = *table;
2124 t.data = &tmptaint;
2125 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2126 if (err < 0)
2127 return err;
2128
2129 if (write) {
2130 /*
2131 * Poor man's atomic or. Not worth adding a primitive
2132 * to everyone's atomic.h for this
2133 */
2134 int i;
2135 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2136 if ((tmptaint >> i) & 1)
2137 add_taint(i, LOCKDEP_STILL_OK);
2138 }
2139 }
2140
2141 return err;
2142 }
2143
2144 #ifdef CONFIG_PRINTK
2145 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2146 void __user *buffer, size_t *lenp, loff_t *ppos)
2147 {
2148 if (write && !capable(CAP_SYS_ADMIN))
2149 return -EPERM;
2150
2151 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2152 }
2153 #endif
2154
2155 struct do_proc_dointvec_minmax_conv_param {
2156 int *min;
2157 int *max;
2158 };
2159
2160 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2161 int *valp,
2162 int write, void *data)
2163 {
2164 struct do_proc_dointvec_minmax_conv_param *param = data;
2165 if (write) {
2166 int val = *negp ? -*lvalp : *lvalp;
2167 if ((param->min && *param->min > val) ||
2168 (param->max && *param->max < val))
2169 return -EINVAL;
2170 *valp = val;
2171 } else {
2172 int val = *valp;
2173 if (val < 0) {
2174 *negp = true;
2175 *lvalp = (unsigned long)-val;
2176 } else {
2177 *negp = false;
2178 *lvalp = (unsigned long)val;
2179 }
2180 }
2181 return 0;
2182 }
2183
2184 /**
2185 * proc_dointvec_minmax - read a vector of integers with min/max values
2186 * @table: the sysctl table
2187 * @write: %TRUE if this is a write to the sysctl file
2188 * @buffer: the user buffer
2189 * @lenp: the size of the user buffer
2190 * @ppos: file position
2191 *
2192 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2193 * values from/to the user buffer, treated as an ASCII string.
2194 *
2195 * This routine will ensure the values are within the range specified by
2196 * table->extra1 (min) and table->extra2 (max).
2197 *
2198 * Returns 0 on success.
2199 */
2200 int proc_dointvec_minmax(struct ctl_table *table, int write,
2201 void __user *buffer, size_t *lenp, loff_t *ppos)
2202 {
2203 struct do_proc_dointvec_minmax_conv_param param = {
2204 .min = (int *) table->extra1,
2205 .max = (int *) table->extra2,
2206 };
2207 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2208 do_proc_dointvec_minmax_conv, &param);
2209 }
2210
2211 static void validate_coredump_safety(void)
2212 {
2213 #ifdef CONFIG_COREDUMP
2214 if (suid_dumpable == SUID_DUMP_ROOT &&
2215 core_pattern[0] != '/' && core_pattern[0] != '|') {
2216 printk(KERN_WARNING "Unsafe core_pattern used with "\
2217 "suid_dumpable=2. Pipe handler or fully qualified "\
2218 "core dump path required.\n");
2219 }
2220 #endif
2221 }
2222
2223 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2224 void __user *buffer, size_t *lenp, loff_t *ppos)
2225 {
2226 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2227 if (!error)
2228 validate_coredump_safety();
2229 return error;
2230 }
2231
2232 #ifdef CONFIG_COREDUMP
2233 static int proc_dostring_coredump(struct ctl_table *table, int write,
2234 void __user *buffer, size_t *lenp, loff_t *ppos)
2235 {
2236 int error = proc_dostring(table, write, buffer, lenp, ppos);
2237 if (!error)
2238 validate_coredump_safety();
2239 return error;
2240 }
2241 #endif
2242
2243 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2244 void __user *buffer,
2245 size_t *lenp, loff_t *ppos,
2246 unsigned long convmul,
2247 unsigned long convdiv)
2248 {
2249 unsigned long *i, *min, *max;
2250 int vleft, first = 1, err = 0;
2251 unsigned long page = 0;
2252 size_t left;
2253 char *kbuf;
2254
2255 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2256 *lenp = 0;
2257 return 0;
2258 }
2259
2260 i = (unsigned long *) data;
2261 min = (unsigned long *) table->extra1;
2262 max = (unsigned long *) table->extra2;
2263 vleft = table->maxlen / sizeof(unsigned long);
2264 left = *lenp;
2265
2266 if (write) {
2267 if (*ppos) {
2268 switch (sysctl_writes_strict) {
2269 case SYSCTL_WRITES_STRICT:
2270 goto out;
2271 case SYSCTL_WRITES_WARN:
2272 warn_sysctl_write(table);
2273 break;
2274 default:
2275 break;
2276 }
2277 }
2278
2279 if (left > PAGE_SIZE - 1)
2280 left = PAGE_SIZE - 1;
2281 page = __get_free_page(GFP_TEMPORARY);
2282 kbuf = (char *) page;
2283 if (!kbuf)
2284 return -ENOMEM;
2285 if (copy_from_user(kbuf, buffer, left)) {
2286 err = -EFAULT;
2287 goto free;
2288 }
2289 kbuf[left] = 0;
2290 }
2291
2292 for (; left && vleft--; i++, first = 0) {
2293 unsigned long val;
2294
2295 if (write) {
2296 bool neg;
2297
2298 left -= proc_skip_spaces(&kbuf);
2299
2300 err = proc_get_long(&kbuf, &left, &val, &neg,
2301 proc_wspace_sep,
2302 sizeof(proc_wspace_sep), NULL);
2303 if (err)
2304 break;
2305 if (neg)
2306 continue;
2307 if ((min && val < *min) || (max && val > *max))
2308 continue;
2309 *i = val;
2310 } else {
2311 val = convdiv * (*i) / convmul;
2312 if (!first) {
2313 err = proc_put_char(&buffer, &left, '\t');
2314 if (err)
2315 break;
2316 }
2317 err = proc_put_long(&buffer, &left, val, false);
2318 if (err)
2319 break;
2320 }
2321 }
2322
2323 if (!write && !first && left && !err)
2324 err = proc_put_char(&buffer, &left, '\n');
2325 if (write && !err)
2326 left -= proc_skip_spaces(&kbuf);
2327 free:
2328 if (write) {
2329 free_page(page);
2330 if (first)
2331 return err ? : -EINVAL;
2332 }
2333 *lenp -= left;
2334 out:
2335 *ppos += *lenp;
2336 return err;
2337 }
2338
2339 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2340 void __user *buffer,
2341 size_t *lenp, loff_t *ppos,
2342 unsigned long convmul,
2343 unsigned long convdiv)
2344 {
2345 return __do_proc_doulongvec_minmax(table->data, table, write,
2346 buffer, lenp, ppos, convmul, convdiv);
2347 }
2348
2349 /**
2350 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2351 * @table: the sysctl table
2352 * @write: %TRUE if this is a write to the sysctl file
2353 * @buffer: the user buffer
2354 * @lenp: the size of the user buffer
2355 * @ppos: file position
2356 *
2357 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2358 * values from/to the user buffer, treated as an ASCII string.
2359 *
2360 * This routine will ensure the values are within the range specified by
2361 * table->extra1 (min) and table->extra2 (max).
2362 *
2363 * Returns 0 on success.
2364 */
2365 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2366 void __user *buffer, size_t *lenp, loff_t *ppos)
2367 {
2368 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2369 }
2370
2371 /**
2372 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2373 * @table: the sysctl table
2374 * @write: %TRUE if this is a write to the sysctl file
2375 * @buffer: the user buffer
2376 * @lenp: the size of the user buffer
2377 * @ppos: file position
2378 *
2379 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2380 * values from/to the user buffer, treated as an ASCII string. The values
2381 * are treated as milliseconds, and converted to jiffies when they are stored.
2382 *
2383 * This routine will ensure the values are within the range specified by
2384 * table->extra1 (min) and table->extra2 (max).
2385 *
2386 * Returns 0 on success.
2387 */
2388 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2389 void __user *buffer,
2390 size_t *lenp, loff_t *ppos)
2391 {
2392 return do_proc_doulongvec_minmax(table, write, buffer,
2393 lenp, ppos, HZ, 1000l);
2394 }
2395
2396
2397 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2398 int *valp,
2399 int write, void *data)
2400 {
2401 if (write) {
2402 if (*lvalp > LONG_MAX / HZ)
2403 return 1;
2404 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2405 } else {
2406 int val = *valp;
2407 unsigned long lval;
2408 if (val < 0) {
2409 *negp = true;
2410 lval = (unsigned long)-val;
2411 } else {
2412 *negp = false;
2413 lval = (unsigned long)val;
2414 }
2415 *lvalp = lval / HZ;
2416 }
2417 return 0;
2418 }
2419
2420 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2421 int *valp,
2422 int write, void *data)
2423 {
2424 if (write) {
2425 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2426 return 1;
2427 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2428 } else {
2429 int val = *valp;
2430 unsigned long lval;
2431 if (val < 0) {
2432 *negp = true;
2433 lval = (unsigned long)-val;
2434 } else {
2435 *negp = false;
2436 lval = (unsigned long)val;
2437 }
2438 *lvalp = jiffies_to_clock_t(lval);
2439 }
2440 return 0;
2441 }
2442
2443 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2444 int *valp,
2445 int write, void *data)
2446 {
2447 if (write) {
2448 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2449
2450 if (jif > INT_MAX)
2451 return 1;
2452 *valp = (int)jif;
2453 } else {
2454 int val = *valp;
2455 unsigned long lval;
2456 if (val < 0) {
2457 *negp = true;
2458 lval = (unsigned long)-val;
2459 } else {
2460 *negp = false;
2461 lval = (unsigned long)val;
2462 }
2463 *lvalp = jiffies_to_msecs(lval);
2464 }
2465 return 0;
2466 }
2467
2468 /**
2469 * proc_dointvec_jiffies - read a vector of integers as seconds
2470 * @table: the sysctl table
2471 * @write: %TRUE if this is a write to the sysctl file
2472 * @buffer: the user buffer
2473 * @lenp: the size of the user buffer
2474 * @ppos: file position
2475 *
2476 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2477 * values from/to the user buffer, treated as an ASCII string.
2478 * The values read are assumed to be in seconds, and are converted into
2479 * jiffies.
2480 *
2481 * Returns 0 on success.
2482 */
2483 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2484 void __user *buffer, size_t *lenp, loff_t *ppos)
2485 {
2486 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2487 do_proc_dointvec_jiffies_conv,NULL);
2488 }
2489
2490 /**
2491 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2492 * @table: the sysctl table
2493 * @write: %TRUE if this is a write to the sysctl file
2494 * @buffer: the user buffer
2495 * @lenp: the size of the user buffer
2496 * @ppos: pointer to the file position
2497 *
2498 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2499 * values from/to the user buffer, treated as an ASCII string.
2500 * The values read are assumed to be in 1/USER_HZ seconds, and
2501 * are converted into jiffies.
2502 *
2503 * Returns 0 on success.
2504 */
2505 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2506 void __user *buffer, size_t *lenp, loff_t *ppos)
2507 {
2508 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2509 do_proc_dointvec_userhz_jiffies_conv,NULL);
2510 }
2511
2512 /**
2513 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2514 * @table: the sysctl table
2515 * @write: %TRUE if this is a write to the sysctl file
2516 * @buffer: the user buffer
2517 * @lenp: the size of the user buffer
2518 * @ppos: file position
2519 * @ppos: the current position in the file
2520 *
2521 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2522 * values from/to the user buffer, treated as an ASCII string.
2523 * The values read are assumed to be in 1/1000 seconds, and
2524 * are converted into jiffies.
2525 *
2526 * Returns 0 on success.
2527 */
2528 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2529 void __user *buffer, size_t *lenp, loff_t *ppos)
2530 {
2531 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2532 do_proc_dointvec_ms_jiffies_conv, NULL);
2533 }
2534
2535 static int proc_do_cad_pid(struct ctl_table *table, int write,
2536 void __user *buffer, size_t *lenp, loff_t *ppos)
2537 {
2538 struct pid *new_pid;
2539 pid_t tmp;
2540 int r;
2541
2542 tmp = pid_vnr(cad_pid);
2543
2544 r = __do_proc_dointvec(&tmp, table, write, buffer,
2545 lenp, ppos, NULL, NULL);
2546 if (r || !write)
2547 return r;
2548
2549 new_pid = find_get_pid(tmp);
2550 if (!new_pid)
2551 return -ESRCH;
2552
2553 put_pid(xchg(&cad_pid, new_pid));
2554 return 0;
2555 }
2556
2557 /**
2558 * proc_do_large_bitmap - read/write from/to a large bitmap
2559 * @table: the sysctl table
2560 * @write: %TRUE if this is a write to the sysctl file
2561 * @buffer: the user buffer
2562 * @lenp: the size of the user buffer
2563 * @ppos: file position
2564 *
2565 * The bitmap is stored at table->data and the bitmap length (in bits)
2566 * in table->maxlen.
2567 *
2568 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2569 * large bitmaps may be represented in a compact manner. Writing into
2570 * the file will clear the bitmap then update it with the given input.
2571 *
2572 * Returns 0 on success.
2573 */
2574 int proc_do_large_bitmap(struct ctl_table *table, int write,
2575 void __user *buffer, size_t *lenp, loff_t *ppos)
2576 {
2577 int err = 0;
2578 bool first = 1;
2579 size_t left = *lenp;
2580 unsigned long bitmap_len = table->maxlen;
2581 unsigned long *bitmap = *(unsigned long **) table->data;
2582 unsigned long *tmp_bitmap = NULL;
2583 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2584
2585 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2586 *lenp = 0;
2587 return 0;
2588 }
2589
2590 if (write) {
2591 unsigned long page = 0;
2592 char *kbuf;
2593
2594 if (left > PAGE_SIZE - 1)
2595 left = PAGE_SIZE - 1;
2596
2597 page = __get_free_page(GFP_TEMPORARY);
2598 kbuf = (char *) page;
2599 if (!kbuf)
2600 return -ENOMEM;
2601 if (copy_from_user(kbuf, buffer, left)) {
2602 free_page(page);
2603 return -EFAULT;
2604 }
2605 kbuf[left] = 0;
2606
2607 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2608 GFP_KERNEL);
2609 if (!tmp_bitmap) {
2610 free_page(page);
2611 return -ENOMEM;
2612 }
2613 proc_skip_char(&kbuf, &left, '\n');
2614 while (!err && left) {
2615 unsigned long val_a, val_b;
2616 bool neg;
2617
2618 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2619 sizeof(tr_a), &c);
2620 if (err)
2621 break;
2622 if (val_a >= bitmap_len || neg) {
2623 err = -EINVAL;
2624 break;
2625 }
2626
2627 val_b = val_a;
2628 if (left) {
2629 kbuf++;
2630 left--;
2631 }
2632
2633 if (c == '-') {
2634 err = proc_get_long(&kbuf, &left, &val_b,
2635 &neg, tr_b, sizeof(tr_b),
2636 &c);
2637 if (err)
2638 break;
2639 if (val_b >= bitmap_len || neg ||
2640 val_a > val_b) {
2641 err = -EINVAL;
2642 break;
2643 }
2644 if (left) {
2645 kbuf++;
2646 left--;
2647 }
2648 }
2649
2650 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2651 first = 0;
2652 proc_skip_char(&kbuf, &left, '\n');
2653 }
2654 free_page(page);
2655 } else {
2656 unsigned long bit_a, bit_b = 0;
2657
2658 while (left) {
2659 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2660 if (bit_a >= bitmap_len)
2661 break;
2662 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2663 bit_a + 1) - 1;
2664
2665 if (!first) {
2666 err = proc_put_char(&buffer, &left, ',');
2667 if (err)
2668 break;
2669 }
2670 err = proc_put_long(&buffer, &left, bit_a, false);
2671 if (err)
2672 break;
2673 if (bit_a != bit_b) {
2674 err = proc_put_char(&buffer, &left, '-');
2675 if (err)
2676 break;
2677 err = proc_put_long(&buffer, &left, bit_b, false);
2678 if (err)
2679 break;
2680 }
2681
2682 first = 0; bit_b++;
2683 }
2684 if (!err)
2685 err = proc_put_char(&buffer, &left, '\n');
2686 }
2687
2688 if (!err) {
2689 if (write) {
2690 if (*ppos)
2691 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2692 else
2693 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2694 }
2695 kfree(tmp_bitmap);
2696 *lenp -= left;
2697 *ppos += *lenp;
2698 return 0;
2699 } else {
2700 kfree(tmp_bitmap);
2701 return err;
2702 }
2703 }
2704
2705 #else /* CONFIG_PROC_SYSCTL */
2706
2707 int proc_dostring(struct ctl_table *table, int write,
2708 void __user *buffer, size_t *lenp, loff_t *ppos)
2709 {
2710 return -ENOSYS;
2711 }
2712
2713 int proc_dointvec(struct ctl_table *table, int write,
2714 void __user *buffer, size_t *lenp, loff_t *ppos)
2715 {
2716 return -ENOSYS;
2717 }
2718
2719 int proc_dointvec_minmax(struct ctl_table *table, int write,
2720 void __user *buffer, size_t *lenp, loff_t *ppos)
2721 {
2722 return -ENOSYS;
2723 }
2724
2725 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2726 void __user *buffer, size_t *lenp, loff_t *ppos)
2727 {
2728 return -ENOSYS;
2729 }
2730
2731 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2732 void __user *buffer, size_t *lenp, loff_t *ppos)
2733 {
2734 return -ENOSYS;
2735 }
2736
2737 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2738 void __user *buffer, size_t *lenp, loff_t *ppos)
2739 {
2740 return -ENOSYS;
2741 }
2742
2743 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2744 void __user *buffer, size_t *lenp, loff_t *ppos)
2745 {
2746 return -ENOSYS;
2747 }
2748
2749 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2750 void __user *buffer,
2751 size_t *lenp, loff_t *ppos)
2752 {
2753 return -ENOSYS;
2754 }
2755
2756
2757 #endif /* CONFIG_PROC_SYSCTL */
2758
2759 /*
2760 * No sense putting this after each symbol definition, twice,
2761 * exception granted :-)
2762 */
2763 EXPORT_SYMBOL(proc_dointvec);
2764 EXPORT_SYMBOL(proc_dointvec_jiffies);
2765 EXPORT_SYMBOL(proc_dointvec_minmax);
2766 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2767 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2768 EXPORT_SYMBOL(proc_dostring);
2769 EXPORT_SYMBOL(proc_doulongvec_minmax);
2770 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
Linux kernel - Deliverables
This page took 0.123926 seconds and 6 git commands to generate.