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