?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
sysctl extern cleanup: module
[linux-2.6/x86.git] / kernel / sysctl.c
blob44e9492368fdd2a79a9703fbb2140673c8a3f784
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
65
66 #if defined(CONFIG_SYSCTL)
67
68 /* External variables not in a header file. */
69 extern int sysctl_overcommit_memory;
70 extern int sysctl_overcommit_ratio;
71 extern int sysctl_panic_on_oom;
72 extern int sysctl_oom_kill_allocating_task;
73 extern int sysctl_oom_dump_tasks;
74 extern int max_threads;
75 extern int core_uses_pid;
76 extern int suid_dumpable;
77 extern char core_pattern[];
78 extern unsigned int core_pipe_limit;
79 extern int pid_max;
80 extern int min_free_kbytes;
81 extern int pid_max_min, pid_max_max;
82 extern int sysctl_drop_caches;
83 extern int percpu_pagelist_fraction;
84 extern int compat_log;
85 extern int latencytop_enabled;
86 extern int sysctl_nr_open_min, sysctl_nr_open_max;
87 #ifndef CONFIG_MMU
88 extern int sysctl_nr_trim_pages;
89 #endif
90 #ifdef CONFIG_BLOCK
91 extern int blk_iopoll_enabled;
92 #endif
93
94 /* Constants used for minimum and maximum */
95 #ifdef CONFIG_DETECT_SOFTLOCKUP
96 static int sixty = 60;
97 static int neg_one = -1;
98 #endif
99
100 static int zero;
101 static int __maybe_unused one = 1;
102 static int __maybe_unused two = 2;
103 static unsigned long one_ul = 1;
104 static int one_hundred = 100;
105 #ifdef CONFIG_PRINTK
106 static int ten_thousand = 10000;
107 #endif
108
109 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
110 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
111
112 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
113 static int maxolduid = 65535;
114 static int minolduid;
115 static int min_percpu_pagelist_fract = 8;
116
117 static int ngroups_max = NGROUPS_MAX;
118
119 #ifdef CONFIG_CHR_DEV_SG
120 extern int sg_big_buff;
121 #endif
122
123 #ifdef CONFIG_SPARC
124 #include <asm/system.h>
125 #endif
126
127 #ifdef CONFIG_SPARC64
128 extern int sysctl_tsb_ratio;
129 #endif
130
131 #ifdef __hppa__
132 extern int pwrsw_enabled;
133 extern int unaligned_enabled;
134 #endif
135
136 #ifdef CONFIG_S390
137 #ifdef CONFIG_MATHEMU
138 extern int sysctl_ieee_emulation_warnings;
139 #endif
140 extern int sysctl_userprocess_debug;
141 extern int spin_retry;
142 #endif
143
144 #ifdef CONFIG_BSD_PROCESS_ACCT
145 extern int acct_parm[];
146 #endif
147
148 #ifdef CONFIG_IA64
149 extern int no_unaligned_warning;
150 extern int unaligned_dump_stack;
151 #endif
152
153 extern struct ratelimit_state printk_ratelimit_state;
154
155 #ifdef CONFIG_RT_MUTEXES
156 extern int max_lock_depth;
157 #endif
158
159 #ifdef CONFIG_PROC_SYSCTL
160 static int proc_do_cad_pid(struct ctl_table *table, int write,
161 void __user *buffer, size_t *lenp, loff_t *ppos);
162 static int proc_taint(struct ctl_table *table, int write,
163 void __user *buffer, size_t *lenp, loff_t *ppos);
164 #endif
165
166 static struct ctl_table root_table[];
167 static struct ctl_table_root sysctl_table_root;
168 static struct ctl_table_header root_table_header = {
169 .count = 1,
170 .ctl_table = root_table,
171 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
172 .root = &sysctl_table_root,
173 .set = &sysctl_table_root.default_set,
174 };
175 static struct ctl_table_root sysctl_table_root = {
176 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
177 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
178 };
179
180 static struct ctl_table kern_table[];
181 static struct ctl_table vm_table[];
182 static struct ctl_table fs_table[];
183 static struct ctl_table debug_table[];
184 static struct ctl_table dev_table[];
185 extern struct ctl_table random_table[];
186 #ifdef CONFIG_INOTIFY_USER
187 extern struct ctl_table inotify_table[];
188 #endif
189 #ifdef CONFIG_EPOLL
190 extern struct ctl_table epoll_table[];
191 #endif
192
193 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
194 int sysctl_legacy_va_layout;
195 #endif
196
197 extern int prove_locking;
198 extern int lock_stat;
199
200 /* The default sysctl tables: */
201
202 static struct ctl_table root_table[] = {
203 {
204 .procname = "kernel",
205 .mode = 0555,
206 .child = kern_table,
207 },
208 {
209 .procname = "vm",
210 .mode = 0555,
211 .child = vm_table,
212 },
213 {
214 .procname = "fs",
215 .mode = 0555,
216 .child = fs_table,
217 },
218 {
219 .procname = "debug",
220 .mode = 0555,
221 .child = debug_table,
222 },
223 {
224 .procname = "dev",
225 .mode = 0555,
226 .child = dev_table,
227 },
228 /*
229 * NOTE: do not add new entries to this table unless you have read
230 * Documentation/sysctl/ctl_unnumbered.txt
231 */
232 { }
233 };
234
235 #ifdef CONFIG_SCHED_DEBUG
236 static int min_sched_granularity_ns = 100000; /* 100 usecs */
237 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
238 static int min_wakeup_granularity_ns; /* 0 usecs */
239 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
240 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
241 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
242 static int min_sched_shares_ratelimit = 100000; /* 100 usec */
243 static int max_sched_shares_ratelimit = NSEC_PER_SEC; /* 1 second */
244 #endif
245
246 static struct ctl_table kern_table[] = {
247 {
248 .procname = "sched_child_runs_first",
249 .data = &sysctl_sched_child_runs_first,
250 .maxlen = sizeof(unsigned int),
251 .mode = 0644,
252 .proc_handler = proc_dointvec,
253 },
254 #ifdef CONFIG_SCHED_DEBUG
255 {
256 .procname = "sched_min_granularity_ns",
257 .data = &sysctl_sched_min_granularity,
258 .maxlen = sizeof(unsigned int),
259 .mode = 0644,
260 .proc_handler = sched_proc_update_handler,
261 .extra1 = &min_sched_granularity_ns,
262 .extra2 = &max_sched_granularity_ns,
263 },
264 {
265 .procname = "sched_latency_ns",
266 .data = &sysctl_sched_latency,
267 .maxlen = sizeof(unsigned int),
268 .mode = 0644,
269 .proc_handler = sched_proc_update_handler,
270 .extra1 = &min_sched_granularity_ns,
271 .extra2 = &max_sched_granularity_ns,
272 },
273 {
274 .procname = "sched_wakeup_granularity_ns",
275 .data = &sysctl_sched_wakeup_granularity,
276 .maxlen = sizeof(unsigned int),
277 .mode = 0644,
278 .proc_handler = sched_proc_update_handler,
279 .extra1 = &min_wakeup_granularity_ns,
280 .extra2 = &max_wakeup_granularity_ns,
281 },
282 {
283 .procname = "sched_shares_ratelimit",
284 .data = &sysctl_sched_shares_ratelimit,
285 .maxlen = sizeof(unsigned int),
286 .mode = 0644,
287 .proc_handler = sched_proc_update_handler,
288 .extra1 = &min_sched_shares_ratelimit,
289 .extra2 = &max_sched_shares_ratelimit,
290 },
291 {
292 .procname = "sched_tunable_scaling",
293 .data = &sysctl_sched_tunable_scaling,
294 .maxlen = sizeof(enum sched_tunable_scaling),
295 .mode = 0644,
296 .proc_handler = sched_proc_update_handler,
297 .extra1 = &min_sched_tunable_scaling,
298 .extra2 = &max_sched_tunable_scaling,
299 },
300 {
301 .procname = "sched_shares_thresh",
302 .data = &sysctl_sched_shares_thresh,
303 .maxlen = sizeof(unsigned int),
304 .mode = 0644,
305 .proc_handler = proc_dointvec_minmax,
306 .extra1 = &zero,
307 },
308 {
309 .procname = "sched_migration_cost",
310 .data = &sysctl_sched_migration_cost,
311 .maxlen = sizeof(unsigned int),
312 .mode = 0644,
313 .proc_handler = proc_dointvec,
314 },
315 {
316 .procname = "sched_nr_migrate",
317 .data = &sysctl_sched_nr_migrate,
318 .maxlen = sizeof(unsigned int),
319 .mode = 0644,
320 .proc_handler = proc_dointvec,
321 },
322 {
323 .procname = "sched_time_avg",
324 .data = &sysctl_sched_time_avg,
325 .maxlen = sizeof(unsigned int),
326 .mode = 0644,
327 .proc_handler = proc_dointvec,
328 },
329 {
330 .procname = "timer_migration",
331 .data = &sysctl_timer_migration,
332 .maxlen = sizeof(unsigned int),
333 .mode = 0644,
334 .proc_handler = proc_dointvec_minmax,
335 .extra1 = &zero,
336 .extra2 = &one,
337 },
338 #endif
339 {
340 .procname = "sched_rt_period_us",
341 .data = &sysctl_sched_rt_period,
342 .maxlen = sizeof(unsigned int),
343 .mode = 0644,
344 .proc_handler = sched_rt_handler,
345 },
346 {
347 .procname = "sched_rt_runtime_us",
348 .data = &sysctl_sched_rt_runtime,
349 .maxlen = sizeof(int),
350 .mode = 0644,
351 .proc_handler = sched_rt_handler,
352 },
353 {
354 .procname = "sched_compat_yield",
355 .data = &sysctl_sched_compat_yield,
356 .maxlen = sizeof(unsigned int),
357 .mode = 0644,
358 .proc_handler = proc_dointvec,
359 },
360 #ifdef CONFIG_PROVE_LOCKING
361 {
362 .procname = "prove_locking",
363 .data = &prove_locking,
364 .maxlen = sizeof(int),
365 .mode = 0644,
366 .proc_handler = proc_dointvec,
367 },
368 #endif
369 #ifdef CONFIG_LOCK_STAT
370 {
371 .procname = "lock_stat",
372 .data = &lock_stat,
373 .maxlen = sizeof(int),
374 .mode = 0644,
375 .proc_handler = proc_dointvec,
376 },
377 #endif
378 {
379 .procname = "panic",
380 .data = &panic_timeout,
381 .maxlen = sizeof(int),
382 .mode = 0644,
383 .proc_handler = proc_dointvec,
384 },
385 {
386 .procname = "core_uses_pid",
387 .data = &core_uses_pid,
388 .maxlen = sizeof(int),
389 .mode = 0644,
390 .proc_handler = proc_dointvec,
391 },
392 {
393 .procname = "core_pattern",
394 .data = core_pattern,
395 .maxlen = CORENAME_MAX_SIZE,
396 .mode = 0644,
397 .proc_handler = proc_dostring,
398 },
399 {
400 .procname = "core_pipe_limit",
401 .data = &core_pipe_limit,
402 .maxlen = sizeof(unsigned int),
403 .mode = 0644,
404 .proc_handler = proc_dointvec,
405 },
406 #ifdef CONFIG_PROC_SYSCTL
407 {
408 .procname = "tainted",
409 .maxlen = sizeof(long),
410 .mode = 0644,
411 .proc_handler = proc_taint,
412 },
413 #endif
414 #ifdef CONFIG_LATENCYTOP
415 {
416 .procname = "latencytop",
417 .data = &latencytop_enabled,
418 .maxlen = sizeof(int),
419 .mode = 0644,
420 .proc_handler = proc_dointvec,
421 },
422 #endif
423 #ifdef CONFIG_BLK_DEV_INITRD
424 {
425 .procname = "real-root-dev",
426 .data = &real_root_dev,
427 .maxlen = sizeof(int),
428 .mode = 0644,
429 .proc_handler = proc_dointvec,
430 },
431 #endif
432 {
433 .procname = "print-fatal-signals",
434 .data = &print_fatal_signals,
435 .maxlen = sizeof(int),
436 .mode = 0644,
437 .proc_handler = proc_dointvec,
438 },
439 #ifdef CONFIG_SPARC
440 {
441 .procname = "reboot-cmd",
442 .data = reboot_command,
443 .maxlen = 256,
444 .mode = 0644,
445 .proc_handler = proc_dostring,
446 },
447 {
448 .procname = "stop-a",
449 .data = &stop_a_enabled,
450 .maxlen = sizeof (int),
451 .mode = 0644,
452 .proc_handler = proc_dointvec,
453 },
454 {
455 .procname = "scons-poweroff",
456 .data = &scons_pwroff,
457 .maxlen = sizeof (int),
458 .mode = 0644,
459 .proc_handler = proc_dointvec,
460 },
461 #endif
462 #ifdef CONFIG_SPARC64
463 {
464 .procname = "tsb-ratio",
465 .data = &sysctl_tsb_ratio,
466 .maxlen = sizeof (int),
467 .mode = 0644,
468 .proc_handler = proc_dointvec,
469 },
470 #endif
471 #ifdef __hppa__
472 {
473 .procname = "soft-power",
474 .data = &pwrsw_enabled,
475 .maxlen = sizeof (int),
476 .mode = 0644,
477 .proc_handler = proc_dointvec,
478 },
479 {
480 .procname = "unaligned-trap",
481 .data = &unaligned_enabled,
482 .maxlen = sizeof (int),
483 .mode = 0644,
484 .proc_handler = proc_dointvec,
485 },
486 #endif
487 {
488 .procname = "ctrl-alt-del",
489 .data = &C_A_D,
490 .maxlen = sizeof(int),
491 .mode = 0644,
492 .proc_handler = proc_dointvec,
493 },
494 #ifdef CONFIG_FUNCTION_TRACER
495 {
496 .procname = "ftrace_enabled",
497 .data = &ftrace_enabled,
498 .maxlen = sizeof(int),
499 .mode = 0644,
500 .proc_handler = ftrace_enable_sysctl,
501 },
502 #endif
503 #ifdef CONFIG_STACK_TRACER
504 {
505 .procname = "stack_tracer_enabled",
506 .data = &stack_tracer_enabled,
507 .maxlen = sizeof(int),
508 .mode = 0644,
509 .proc_handler = stack_trace_sysctl,
510 },
511 #endif
512 #ifdef CONFIG_TRACING
513 {
514 .procname = "ftrace_dump_on_oops",
515 .data = &ftrace_dump_on_oops,
516 .maxlen = sizeof(int),
517 .mode = 0644,
518 .proc_handler = proc_dointvec,
519 },
520 #endif
521 #ifdef CONFIG_MODULES
522 {
523 .procname = "modprobe",
524 .data = &modprobe_path,
525 .maxlen = KMOD_PATH_LEN,
526 .mode = 0644,
527 .proc_handler = proc_dostring,
528 },
529 {
530 .procname = "modules_disabled",
531 .data = &modules_disabled,
532 .maxlen = sizeof(int),
533 .mode = 0644,
534 /* only handle a transition from default "0" to "1" */
535 .proc_handler = proc_dointvec_minmax,
536 .extra1 = &one,
537 .extra2 = &one,
538 },
539 #endif
540 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
541 {
542 .procname = "hotplug",
543 .data = &uevent_helper,
544 .maxlen = UEVENT_HELPER_PATH_LEN,
545 .mode = 0644,
546 .proc_handler = proc_dostring,
547 },
548 #endif
549 #ifdef CONFIG_CHR_DEV_SG
550 {
551 .procname = "sg-big-buff",
552 .data = &sg_big_buff,
553 .maxlen = sizeof (int),
554 .mode = 0444,
555 .proc_handler = proc_dointvec,
556 },
557 #endif
558 #ifdef CONFIG_BSD_PROCESS_ACCT
559 {
560 .procname = "acct",
561 .data = &acct_parm,
562 .maxlen = 3*sizeof(int),
563 .mode = 0644,
564 .proc_handler = proc_dointvec,
565 },
566 #endif
567 #ifdef CONFIG_MAGIC_SYSRQ
568 {
569 .procname = "sysrq",
570 .data = &__sysrq_enabled,
571 .maxlen = sizeof (int),
572 .mode = 0644,
573 .proc_handler = proc_dointvec,
574 },
575 #endif
576 #ifdef CONFIG_PROC_SYSCTL
577 {
578 .procname = "cad_pid",
579 .data = NULL,
580 .maxlen = sizeof (int),
581 .mode = 0600,
582 .proc_handler = proc_do_cad_pid,
583 },
584 #endif
585 {
586 .procname = "threads-max",
587 .data = &max_threads,
588 .maxlen = sizeof(int),
589 .mode = 0644,
590 .proc_handler = proc_dointvec,
591 },
592 {
593 .procname = "random",
594 .mode = 0555,
595 .child = random_table,
596 },
597 {
598 .procname = "overflowuid",
599 .data = &overflowuid,
600 .maxlen = sizeof(int),
601 .mode = 0644,
602 .proc_handler = proc_dointvec_minmax,
603 .extra1 = &minolduid,
604 .extra2 = &maxolduid,
605 },
606 {
607 .procname = "overflowgid",
608 .data = &overflowgid,
609 .maxlen = sizeof(int),
610 .mode = 0644,
611 .proc_handler = proc_dointvec_minmax,
612 .extra1 = &minolduid,
613 .extra2 = &maxolduid,
614 },
615 #ifdef CONFIG_S390
616 #ifdef CONFIG_MATHEMU
617 {
618 .procname = "ieee_emulation_warnings",
619 .data = &sysctl_ieee_emulation_warnings,
620 .maxlen = sizeof(int),
621 .mode = 0644,
622 .proc_handler = proc_dointvec,
623 },
624 #endif
625 {
626 .procname = "userprocess_debug",
627 .data = &sysctl_userprocess_debug,
628 .maxlen = sizeof(int),
629 .mode = 0644,
630 .proc_handler = proc_dointvec,
631 },
632 #endif
633 {
634 .procname = "pid_max",
635 .data = &pid_max,
636 .maxlen = sizeof (int),
637 .mode = 0644,
638 .proc_handler = proc_dointvec_minmax,
639 .extra1 = &pid_max_min,
640 .extra2 = &pid_max_max,
641 },
642 {
643 .procname = "panic_on_oops",
644 .data = &panic_on_oops,
645 .maxlen = sizeof(int),
646 .mode = 0644,
647 .proc_handler = proc_dointvec,
648 },
649 #if defined CONFIG_PRINTK
650 {
651 .procname = "printk",
652 .data = &console_loglevel,
653 .maxlen = 4*sizeof(int),
654 .mode = 0644,
655 .proc_handler = proc_dointvec,
656 },
657 {
658 .procname = "printk_ratelimit",
659 .data = &printk_ratelimit_state.interval,
660 .maxlen = sizeof(int),
661 .mode = 0644,
662 .proc_handler = proc_dointvec_jiffies,
663 },
664 {
665 .procname = "printk_ratelimit_burst",
666 .data = &printk_ratelimit_state.burst,
667 .maxlen = sizeof(int),
668 .mode = 0644,
669 .proc_handler = proc_dointvec,
670 },
671 {
672 .procname = "printk_delay",
673 .data = &printk_delay_msec,
674 .maxlen = sizeof(int),
675 .mode = 0644,
676 .proc_handler = proc_dointvec_minmax,
677 .extra1 = &zero,
678 .extra2 = &ten_thousand,
679 },
680 #endif
681 {
682 .procname = "ngroups_max",
683 .data = &ngroups_max,
684 .maxlen = sizeof (int),
685 .mode = 0444,
686 .proc_handler = proc_dointvec,
687 },
688 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
689 {
690 .procname = "unknown_nmi_panic",
691 .data = &unknown_nmi_panic,
692 .maxlen = sizeof (int),
693 .mode = 0644,
694 .proc_handler = proc_dointvec,
695 },
696 {
697 .procname = "nmi_watchdog",
698 .data = &nmi_watchdog_enabled,
699 .maxlen = sizeof (int),
700 .mode = 0644,
701 .proc_handler = proc_nmi_enabled,
702 },
703 #endif
704 #if defined(CONFIG_X86)
705 {
706 .procname = "panic_on_unrecovered_nmi",
707 .data = &panic_on_unrecovered_nmi,
708 .maxlen = sizeof(int),
709 .mode = 0644,
710 .proc_handler = proc_dointvec,
711 },
712 {
713 .procname = "panic_on_io_nmi",
714 .data = &panic_on_io_nmi,
715 .maxlen = sizeof(int),
716 .mode = 0644,
717 .proc_handler = proc_dointvec,
718 },
719 {
720 .procname = "bootloader_type",
721 .data = &bootloader_type,
722 .maxlen = sizeof (int),
723 .mode = 0444,
724 .proc_handler = proc_dointvec,
725 },
726 {
727 .procname = "bootloader_version",
728 .data = &bootloader_version,
729 .maxlen = sizeof (int),
730 .mode = 0444,
731 .proc_handler = proc_dointvec,
732 },
733 {
734 .procname = "kstack_depth_to_print",
735 .data = &kstack_depth_to_print,
736 .maxlen = sizeof(int),
737 .mode = 0644,
738 .proc_handler = proc_dointvec,
739 },
740 {
741 .procname = "io_delay_type",
742 .data = &io_delay_type,
743 .maxlen = sizeof(int),
744 .mode = 0644,
745 .proc_handler = proc_dointvec,
746 },
747 #endif
748 #if defined(CONFIG_MMU)
749 {
750 .procname = "randomize_va_space",
751 .data = &randomize_va_space,
752 .maxlen = sizeof(int),
753 .mode = 0644,
754 .proc_handler = proc_dointvec,
755 },
756 #endif
757 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
758 {
759 .procname = "spin_retry",
760 .data = &spin_retry,
761 .maxlen = sizeof (int),
762 .mode = 0644,
763 .proc_handler = proc_dointvec,
764 },
765 #endif
766 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
767 {
768 .procname = "acpi_video_flags",
769 .data = &acpi_realmode_flags,
770 .maxlen = sizeof (unsigned long),
771 .mode = 0644,
772 .proc_handler = proc_doulongvec_minmax,
773 },
774 #endif
775 #ifdef CONFIG_IA64
776 {
777 .procname = "ignore-unaligned-usertrap",
778 .data = &no_unaligned_warning,
779 .maxlen = sizeof (int),
780 .mode = 0644,
781 .proc_handler = proc_dointvec,
782 },
783 {
784 .procname = "unaligned-dump-stack",
785 .data = &unaligned_dump_stack,
786 .maxlen = sizeof (int),
787 .mode = 0644,
788 .proc_handler = proc_dointvec,
789 },
790 #endif
791 #ifdef CONFIG_DETECT_SOFTLOCKUP
792 {
793 .procname = "softlockup_panic",
794 .data = &softlockup_panic,
795 .maxlen = sizeof(int),
796 .mode = 0644,
797 .proc_handler = proc_dointvec_minmax,
798 .extra1 = &zero,
799 .extra2 = &one,
800 },
801 {
802 .procname = "softlockup_thresh",
803 .data = &softlockup_thresh,
804 .maxlen = sizeof(int),
805 .mode = 0644,
806 .proc_handler = proc_dosoftlockup_thresh,
807 .extra1 = &neg_one,
808 .extra2 = &sixty,
809 },
810 #endif
811 #ifdef CONFIG_DETECT_HUNG_TASK
812 {
813 .procname = "hung_task_panic",
814 .data = &sysctl_hung_task_panic,
815 .maxlen = sizeof(int),
816 .mode = 0644,
817 .proc_handler = proc_dointvec_minmax,
818 .extra1 = &zero,
819 .extra2 = &one,
820 },
821 {
822 .procname = "hung_task_check_count",
823 .data = &sysctl_hung_task_check_count,
824 .maxlen = sizeof(unsigned long),
825 .mode = 0644,
826 .proc_handler = proc_doulongvec_minmax,
827 },
828 {
829 .procname = "hung_task_timeout_secs",
830 .data = &sysctl_hung_task_timeout_secs,
831 .maxlen = sizeof(unsigned long),
832 .mode = 0644,
833 .proc_handler = proc_dohung_task_timeout_secs,
834 },
835 {
836 .procname = "hung_task_warnings",
837 .data = &sysctl_hung_task_warnings,
838 .maxlen = sizeof(unsigned long),
839 .mode = 0644,
840 .proc_handler = proc_doulongvec_minmax,
841 },
842 #endif
843 #ifdef CONFIG_COMPAT
844 {
845 .procname = "compat-log",
846 .data = &compat_log,
847 .maxlen = sizeof (int),
848 .mode = 0644,
849 .proc_handler = proc_dointvec,
850 },
851 #endif
852 #ifdef CONFIG_RT_MUTEXES
853 {
854 .procname = "max_lock_depth",
855 .data = &max_lock_depth,
856 .maxlen = sizeof(int),
857 .mode = 0644,
858 .proc_handler = proc_dointvec,
859 },
860 #endif
861 {
862 .procname = "poweroff_cmd",
863 .data = &poweroff_cmd,
864 .maxlen = POWEROFF_CMD_PATH_LEN,
865 .mode = 0644,
866 .proc_handler = proc_dostring,
867 },
868 #ifdef CONFIG_KEYS
869 {
870 .procname = "keys",
871 .mode = 0555,
872 .child = key_sysctls,
873 },
874 #endif
875 #ifdef CONFIG_RCU_TORTURE_TEST
876 {
877 .procname = "rcutorture_runnable",
878 .data = &rcutorture_runnable,
879 .maxlen = sizeof(int),
880 .mode = 0644,
881 .proc_handler = proc_dointvec,
882 },
883 #endif
884 #ifdef CONFIG_SLOW_WORK
885 {
886 .procname = "slow-work",
887 .mode = 0555,
888 .child = slow_work_sysctls,
889 },
890 #endif
891 #ifdef CONFIG_PERF_EVENTS
892 {
893 .procname = "perf_event_paranoid",
894 .data = &sysctl_perf_event_paranoid,
895 .maxlen = sizeof(sysctl_perf_event_paranoid),
896 .mode = 0644,
897 .proc_handler = proc_dointvec,
898 },
899 {
900 .procname = "perf_event_mlock_kb",
901 .data = &sysctl_perf_event_mlock,
902 .maxlen = sizeof(sysctl_perf_event_mlock),
903 .mode = 0644,
904 .proc_handler = proc_dointvec,
905 },
906 {
907 .procname = "perf_event_max_sample_rate",
908 .data = &sysctl_perf_event_sample_rate,
909 .maxlen = sizeof(sysctl_perf_event_sample_rate),
910 .mode = 0644,
911 .proc_handler = proc_dointvec,
912 },
913 #endif
914 #ifdef CONFIG_KMEMCHECK
915 {
916 .procname = "kmemcheck",
917 .data = &kmemcheck_enabled,
918 .maxlen = sizeof(int),
919 .mode = 0644,
920 .proc_handler = proc_dointvec,
921 },
922 #endif
923 #ifdef CONFIG_BLOCK
924 {
925 .procname = "blk_iopoll",
926 .data = &blk_iopoll_enabled,
927 .maxlen = sizeof(int),
928 .mode = 0644,
929 .proc_handler = proc_dointvec,
930 },
931 #endif
932 /*
933 * NOTE: do not add new entries to this table unless you have read
934 * Documentation/sysctl/ctl_unnumbered.txt
935 */
936 { }
937 };
938
939 static struct ctl_table vm_table[] = {
940 {
941 .procname = "overcommit_memory",
942 .data = &sysctl_overcommit_memory,
943 .maxlen = sizeof(sysctl_overcommit_memory),
944 .mode = 0644,
945 .proc_handler = proc_dointvec,
946 },
947 {
948 .procname = "panic_on_oom",
949 .data = &sysctl_panic_on_oom,
950 .maxlen = sizeof(sysctl_panic_on_oom),
951 .mode = 0644,
952 .proc_handler = proc_dointvec,
953 },
954 {
955 .procname = "oom_kill_allocating_task",
956 .data = &sysctl_oom_kill_allocating_task,
957 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
958 .mode = 0644,
959 .proc_handler = proc_dointvec,
960 },
961 {
962 .procname = "oom_dump_tasks",
963 .data = &sysctl_oom_dump_tasks,
964 .maxlen = sizeof(sysctl_oom_dump_tasks),
965 .mode = 0644,
966 .proc_handler = proc_dointvec,
967 },
968 {
969 .procname = "overcommit_ratio",
970 .data = &sysctl_overcommit_ratio,
971 .maxlen = sizeof(sysctl_overcommit_ratio),
972 .mode = 0644,
973 .proc_handler = proc_dointvec,
974 },
975 {
976 .procname = "page-cluster",
977 .data = &page_cluster,
978 .maxlen = sizeof(int),
979 .mode = 0644,
980 .proc_handler = proc_dointvec,
981 },
982 {
983 .procname = "dirty_background_ratio",
984 .data = &dirty_background_ratio,
985 .maxlen = sizeof(dirty_background_ratio),
986 .mode = 0644,
987 .proc_handler = dirty_background_ratio_handler,
988 .extra1 = &zero,
989 .extra2 = &one_hundred,
990 },
991 {
992 .procname = "dirty_background_bytes",
993 .data = &dirty_background_bytes,
994 .maxlen = sizeof(dirty_background_bytes),
995 .mode = 0644,
996 .proc_handler = dirty_background_bytes_handler,
997 .extra1 = &one_ul,
998 },
999 {
1000 .procname = "dirty_ratio",
1001 .data = &vm_dirty_ratio,
1002 .maxlen = sizeof(vm_dirty_ratio),
1003 .mode = 0644,
1004 .proc_handler = dirty_ratio_handler,
1005 .extra1 = &zero,
1006 .extra2 = &one_hundred,
1007 },
1008 {
1009 .procname = "dirty_bytes",
1010 .data = &vm_dirty_bytes,
1011 .maxlen = sizeof(vm_dirty_bytes),
1012 .mode = 0644,
1013 .proc_handler = dirty_bytes_handler,
1014 .extra1 = &dirty_bytes_min,
1015 },
1016 {
1017 .procname = "dirty_writeback_centisecs",
1018 .data = &dirty_writeback_interval,
1019 .maxlen = sizeof(dirty_writeback_interval),
1020 .mode = 0644,
1021 .proc_handler = dirty_writeback_centisecs_handler,
1022 },
1023 {
1024 .procname = "dirty_expire_centisecs",
1025 .data = &dirty_expire_interval,
1026 .maxlen = sizeof(dirty_expire_interval),
1027 .mode = 0644,
1028 .proc_handler = proc_dointvec,
1029 },
1030 {
1031 .procname = "nr_pdflush_threads",
1032 .data = &nr_pdflush_threads,
1033 .maxlen = sizeof nr_pdflush_threads,
1034 .mode = 0444 /* read-only*/,
1035 .proc_handler = proc_dointvec,
1036 },
1037 {
1038 .procname = "swappiness",
1039 .data = &vm_swappiness,
1040 .maxlen = sizeof(vm_swappiness),
1041 .mode = 0644,
1042 .proc_handler = proc_dointvec_minmax,
1043 .extra1 = &zero,
1044 .extra2 = &one_hundred,
1045 },
1046 #ifdef CONFIG_HUGETLB_PAGE
1047 {
1048 .procname = "nr_hugepages",
1049 .data = NULL,
1050 .maxlen = sizeof(unsigned long),
1051 .mode = 0644,
1052 .proc_handler = hugetlb_sysctl_handler,
1053 .extra1 = (void *)&hugetlb_zero,
1054 .extra2 = (void *)&hugetlb_infinity,
1055 },
1056 #ifdef CONFIG_NUMA
1057 {
1058 .procname = "nr_hugepages_mempolicy",
1059 .data = NULL,
1060 .maxlen = sizeof(unsigned long),
1061 .mode = 0644,
1062 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1063 .extra1 = (void *)&hugetlb_zero,
1064 .extra2 = (void *)&hugetlb_infinity,
1065 },
1066 #endif
1067 {
1068 .procname = "hugetlb_shm_group",
1069 .data = &sysctl_hugetlb_shm_group,
1070 .maxlen = sizeof(gid_t),
1071 .mode = 0644,
1072 .proc_handler = proc_dointvec,
1073 },
1074 {
1075 .procname = "hugepages_treat_as_movable",
1076 .data = &hugepages_treat_as_movable,
1077 .maxlen = sizeof(int),
1078 .mode = 0644,
1079 .proc_handler = hugetlb_treat_movable_handler,
1080 },
1081 {
1082 .procname = "nr_overcommit_hugepages",
1083 .data = NULL,
1084 .maxlen = sizeof(unsigned long),
1085 .mode = 0644,
1086 .proc_handler = hugetlb_overcommit_handler,
1087 .extra1 = (void *)&hugetlb_zero,
1088 .extra2 = (void *)&hugetlb_infinity,
1089 },
1090 #endif
1091 {
1092 .procname = "lowmem_reserve_ratio",
1093 .data = &sysctl_lowmem_reserve_ratio,
1094 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1095 .mode = 0644,
1096 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1097 },
1098 {
1099 .procname = "drop_caches",
1100 .data = &sysctl_drop_caches,
1101 .maxlen = sizeof(int),
1102 .mode = 0644,
1103 .proc_handler = drop_caches_sysctl_handler,
1104 },
1105 {
1106 .procname = "min_free_kbytes",
1107 .data = &min_free_kbytes,
1108 .maxlen = sizeof(min_free_kbytes),
1109 .mode = 0644,
1110 .proc_handler = min_free_kbytes_sysctl_handler,
1111 .extra1 = &zero,
1112 },
1113 {
1114 .procname = "percpu_pagelist_fraction",
1115 .data = &percpu_pagelist_fraction,
1116 .maxlen = sizeof(percpu_pagelist_fraction),
1117 .mode = 0644,
1118 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1119 .extra1 = &min_percpu_pagelist_fract,
1120 },
1121 #ifdef CONFIG_MMU
1122 {
1123 .procname = "max_map_count",
1124 .data = &sysctl_max_map_count,
1125 .maxlen = sizeof(sysctl_max_map_count),
1126 .mode = 0644,
1127 .proc_handler = proc_dointvec_minmax,
1128 .extra1 = &zero,
1129 },
1130 #else
1131 {
1132 .procname = "nr_trim_pages",
1133 .data = &sysctl_nr_trim_pages,
1134 .maxlen = sizeof(sysctl_nr_trim_pages),
1135 .mode = 0644,
1136 .proc_handler = proc_dointvec_minmax,
1137 .extra1 = &zero,
1138 },
1139 #endif
1140 {
1141 .procname = "laptop_mode",
1142 .data = &laptop_mode,
1143 .maxlen = sizeof(laptop_mode),
1144 .mode = 0644,
1145 .proc_handler = proc_dointvec_jiffies,
1146 },
1147 {
1148 .procname = "block_dump",
1149 .data = &block_dump,
1150 .maxlen = sizeof(block_dump),
1151 .mode = 0644,
1152 .proc_handler = proc_dointvec,
1153 .extra1 = &zero,
1154 },
1155 {
1156 .procname = "vfs_cache_pressure",
1157 .data = &sysctl_vfs_cache_pressure,
1158 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1159 .mode = 0644,
1160 .proc_handler = proc_dointvec,
1161 .extra1 = &zero,
1162 },
1163 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1164 {
1165 .procname = "legacy_va_layout",
1166 .data = &sysctl_legacy_va_layout,
1167 .maxlen = sizeof(sysctl_legacy_va_layout),
1168 .mode = 0644,
1169 .proc_handler = proc_dointvec,
1170 .extra1 = &zero,
1171 },
1172 #endif
1173 #ifdef CONFIG_NUMA
1174 {
1175 .procname = "zone_reclaim_mode",
1176 .data = &zone_reclaim_mode,
1177 .maxlen = sizeof(zone_reclaim_mode),
1178 .mode = 0644,
1179 .proc_handler = proc_dointvec,
1180 .extra1 = &zero,
1181 },
1182 {
1183 .procname = "min_unmapped_ratio",
1184 .data = &sysctl_min_unmapped_ratio,
1185 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1186 .mode = 0644,
1187 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1188 .extra1 = &zero,
1189 .extra2 = &one_hundred,
1190 },
1191 {
1192 .procname = "min_slab_ratio",
1193 .data = &sysctl_min_slab_ratio,
1194 .maxlen = sizeof(sysctl_min_slab_ratio),
1195 .mode = 0644,
1196 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1197 .extra1 = &zero,
1198 .extra2 = &one_hundred,
1199 },
1200 #endif
1201 #ifdef CONFIG_SMP
1202 {
1203 .procname = "stat_interval",
1204 .data = &sysctl_stat_interval,
1205 .maxlen = sizeof(sysctl_stat_interval),
1206 .mode = 0644,
1207 .proc_handler = proc_dointvec_jiffies,
1208 },
1209 #endif
1210 #ifdef CONFIG_MMU
1211 {
1212 .procname = "mmap_min_addr",
1213 .data = &dac_mmap_min_addr,
1214 .maxlen = sizeof(unsigned long),
1215 .mode = 0644,
1216 .proc_handler = mmap_min_addr_handler,
1217 },
1218 #endif
1219 #ifdef CONFIG_NUMA
1220 {
1221 .procname = "numa_zonelist_order",
1222 .data = &numa_zonelist_order,
1223 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1224 .mode = 0644,
1225 .proc_handler = numa_zonelist_order_handler,
1226 },
1227 #endif
1228 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1229 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1230 {
1231 .procname = "vdso_enabled",
1232 .data = &vdso_enabled,
1233 .maxlen = sizeof(vdso_enabled),
1234 .mode = 0644,
1235 .proc_handler = proc_dointvec,
1236 .extra1 = &zero,
1237 },
1238 #endif
1239 #ifdef CONFIG_HIGHMEM
1240 {
1241 .procname = "highmem_is_dirtyable",
1242 .data = &vm_highmem_is_dirtyable,
1243 .maxlen = sizeof(vm_highmem_is_dirtyable),
1244 .mode = 0644,
1245 .proc_handler = proc_dointvec_minmax,
1246 .extra1 = &zero,
1247 .extra2 = &one,
1248 },
1249 #endif
1250 {
1251 .procname = "scan_unevictable_pages",
1252 .data = &scan_unevictable_pages,
1253 .maxlen = sizeof(scan_unevictable_pages),
1254 .mode = 0644,
1255 .proc_handler = scan_unevictable_handler,
1256 },
1257 #ifdef CONFIG_MEMORY_FAILURE
1258 {
1259 .procname = "memory_failure_early_kill",
1260 .data = &sysctl_memory_failure_early_kill,
1261 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1262 .mode = 0644,
1263 .proc_handler = proc_dointvec_minmax,
1264 .extra1 = &zero,
1265 .extra2 = &one,
1266 },
1267 {
1268 .procname = "memory_failure_recovery",
1269 .data = &sysctl_memory_failure_recovery,
1270 .maxlen = sizeof(sysctl_memory_failure_recovery),
1271 .mode = 0644,
1272 .proc_handler = proc_dointvec_minmax,
1273 .extra1 = &zero,
1274 .extra2 = &one,
1275 },
1276 #endif
1277
1278 /*
1279 * NOTE: do not add new entries to this table unless you have read
1280 * Documentation/sysctl/ctl_unnumbered.txt
1281 */
1282 { }
1283 };
1284
1285 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1286 static struct ctl_table binfmt_misc_table[] = {
1287 { }
1288 };
1289 #endif
1290
1291 static struct ctl_table fs_table[] = {
1292 {
1293 .procname = "inode-nr",
1294 .data = &inodes_stat,
1295 .maxlen = 2*sizeof(int),
1296 .mode = 0444,
1297 .proc_handler = proc_dointvec,
1298 },
1299 {
1300 .procname = "inode-state",
1301 .data = &inodes_stat,
1302 .maxlen = 7*sizeof(int),
1303 .mode = 0444,
1304 .proc_handler = proc_dointvec,
1305 },
1306 {
1307 .procname = "file-nr",
1308 .data = &files_stat,
1309 .maxlen = 3*sizeof(int),
1310 .mode = 0444,
1311 .proc_handler = proc_nr_files,
1312 },
1313 {
1314 .procname = "file-max",
1315 .data = &files_stat.max_files,
1316 .maxlen = sizeof(int),
1317 .mode = 0644,
1318 .proc_handler = proc_dointvec,
1319 },
1320 {
1321 .procname = "nr_open",
1322 .data = &sysctl_nr_open,
1323 .maxlen = sizeof(int),
1324 .mode = 0644,
1325 .proc_handler = proc_dointvec_minmax,
1326 .extra1 = &sysctl_nr_open_min,
1327 .extra2 = &sysctl_nr_open_max,
1328 },
1329 {
1330 .procname = "dentry-state",
1331 .data = &dentry_stat,
1332 .maxlen = 6*sizeof(int),
1333 .mode = 0444,
1334 .proc_handler = proc_dointvec,
1335 },
1336 {
1337 .procname = "overflowuid",
1338 .data = &fs_overflowuid,
1339 .maxlen = sizeof(int),
1340 .mode = 0644,
1341 .proc_handler = proc_dointvec_minmax,
1342 .extra1 = &minolduid,
1343 .extra2 = &maxolduid,
1344 },
1345 {
1346 .procname = "overflowgid",
1347 .data = &fs_overflowgid,
1348 .maxlen = sizeof(int),
1349 .mode = 0644,
1350 .proc_handler = proc_dointvec_minmax,
1351 .extra1 = &minolduid,
1352 .extra2 = &maxolduid,
1353 },
1354 #ifdef CONFIG_FILE_LOCKING
1355 {
1356 .procname = "leases-enable",
1357 .data = &leases_enable,
1358 .maxlen = sizeof(int),
1359 .mode = 0644,
1360 .proc_handler = proc_dointvec,
1361 },
1362 #endif
1363 #ifdef CONFIG_DNOTIFY
1364 {
1365 .procname = "dir-notify-enable",
1366 .data = &dir_notify_enable,
1367 .maxlen = sizeof(int),
1368 .mode = 0644,
1369 .proc_handler = proc_dointvec,
1370 },
1371 #endif
1372 #ifdef CONFIG_MMU
1373 #ifdef CONFIG_FILE_LOCKING
1374 {
1375 .procname = "lease-break-time",
1376 .data = &lease_break_time,
1377 .maxlen = sizeof(int),
1378 .mode = 0644,
1379 .proc_handler = proc_dointvec,
1380 },
1381 #endif
1382 #ifdef CONFIG_AIO
1383 {
1384 .procname = "aio-nr",
1385 .data = &aio_nr,
1386 .maxlen = sizeof(aio_nr),
1387 .mode = 0444,
1388 .proc_handler = proc_doulongvec_minmax,
1389 },
1390 {
1391 .procname = "aio-max-nr",
1392 .data = &aio_max_nr,
1393 .maxlen = sizeof(aio_max_nr),
1394 .mode = 0644,
1395 .proc_handler = proc_doulongvec_minmax,
1396 },
1397 #endif /* CONFIG_AIO */
1398 #ifdef CONFIG_INOTIFY_USER
1399 {
1400 .procname = "inotify",
1401 .mode = 0555,
1402 .child = inotify_table,
1403 },
1404 #endif
1405 #ifdef CONFIG_EPOLL
1406 {
1407 .procname = "epoll",
1408 .mode = 0555,
1409 .child = epoll_table,
1410 },
1411 #endif
1412 #endif
1413 {
1414 .procname = "suid_dumpable",
1415 .data = &suid_dumpable,
1416 .maxlen = sizeof(int),
1417 .mode = 0644,
1418 .proc_handler = proc_dointvec_minmax,
1419 .extra1 = &zero,
1420 .extra2 = &two,
1421 },
1422 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1423 {
1424 .procname = "binfmt_misc",
1425 .mode = 0555,
1426 .child = binfmt_misc_table,
1427 },
1428 #endif
1429 /*
1430 * NOTE: do not add new entries to this table unless you have read
1431 * Documentation/sysctl/ctl_unnumbered.txt
1432 */
1433 { }
1434 };
1435
1436 static struct ctl_table debug_table[] = {
1437 #if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC)
1438 {
1439 .procname = "exception-trace",
1440 .data = &show_unhandled_signals,
1441 .maxlen = sizeof(int),
1442 .mode = 0644,
1443 .proc_handler = proc_dointvec
1444 },
1445 #endif
1446 #if defined(CONFIG_OPTPROBES)
1447 {
1448 .procname = "kprobes-optimization",
1449 .data = &sysctl_kprobes_optimization,
1450 .maxlen = sizeof(int),
1451 .mode = 0644,
1452 .proc_handler = proc_kprobes_optimization_handler,
1453 .extra1 = &zero,
1454 .extra2 = &one,
1455 },
1456 #endif
1457 { }
1458 };
1459
1460 static struct ctl_table dev_table[] = {
1461 { }
1462 };
1463
1464 static DEFINE_SPINLOCK(sysctl_lock);
1465
1466 /* called under sysctl_lock */
1467 static int use_table(struct ctl_table_header *p)
1468 {
1469 if (unlikely(p->unregistering))
1470 return 0;
1471 p->used++;
1472 return 1;
1473 }
1474
1475 /* called under sysctl_lock */
1476 static void unuse_table(struct ctl_table_header *p)
1477 {
1478 if (!--p->used)
1479 if (unlikely(p->unregistering))
1480 complete(p->unregistering);
1481 }
1482
1483 /* called under sysctl_lock, will reacquire if has to wait */
1484 static void start_unregistering(struct ctl_table_header *p)
1485 {
1486 /*
1487 * if p->used is 0, nobody will ever touch that entry again;
1488 * we'll eliminate all paths to it before dropping sysctl_lock
1489 */
1490 if (unlikely(p->used)) {
1491 struct completion wait;
1492 init_completion(&wait);
1493 p->unregistering = &wait;
1494 spin_unlock(&sysctl_lock);
1495 wait_for_completion(&wait);
1496 spin_lock(&sysctl_lock);
1497 } else {
1498 /* anything non-NULL; we'll never dereference it */
1499 p->unregistering = ERR_PTR(-EINVAL);
1500 }
1501 /*
1502 * do not remove from the list until nobody holds it; walking the
1503 * list in do_sysctl() relies on that.
1504 */
1505 list_del_init(&p->ctl_entry);
1506 }
1507
1508 void sysctl_head_get(struct ctl_table_header *head)
1509 {
1510 spin_lock(&sysctl_lock);
1511 head->count++;
1512 spin_unlock(&sysctl_lock);
1513 }
1514
1515 void sysctl_head_put(struct ctl_table_header *head)
1516 {
1517 spin_lock(&sysctl_lock);
1518 if (!--head->count)
1519 kfree(head);
1520 spin_unlock(&sysctl_lock);
1521 }
1522
1523 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1524 {
1525 if (!head)
1526 BUG();
1527 spin_lock(&sysctl_lock);
1528 if (!use_table(head))
1529 head = ERR_PTR(-ENOENT);
1530 spin_unlock(&sysctl_lock);
1531 return head;
1532 }
1533
1534 void sysctl_head_finish(struct ctl_table_header *head)
1535 {
1536 if (!head)
1537 return;
1538 spin_lock(&sysctl_lock);
1539 unuse_table(head);
1540 spin_unlock(&sysctl_lock);
1541 }
1542
1543 static struct ctl_table_set *
1544 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1545 {
1546 struct ctl_table_set *set = &root->default_set;
1547 if (root->lookup)
1548 set = root->lookup(root, namespaces);
1549 return set;
1550 }
1551
1552 static struct list_head *
1553 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1554 {
1555 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1556 return &set->list;
1557 }
1558
1559 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1560 struct ctl_table_header *prev)
1561 {
1562 struct ctl_table_root *root;
1563 struct list_head *header_list;
1564 struct ctl_table_header *head;
1565 struct list_head *tmp;
1566
1567 spin_lock(&sysctl_lock);
1568 if (prev) {
1569 head = prev;
1570 tmp = &prev->ctl_entry;
1571 unuse_table(prev);
1572 goto next;
1573 }
1574 tmp = &root_table_header.ctl_entry;
1575 for (;;) {
1576 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1577
1578 if (!use_table(head))
1579 goto next;
1580 spin_unlock(&sysctl_lock);
1581 return head;
1582 next:
1583 root = head->root;
1584 tmp = tmp->next;
1585 header_list = lookup_header_list(root, namespaces);
1586 if (tmp != header_list)
1587 continue;
1588
1589 do {
1590 root = list_entry(root->root_list.next,
1591 struct ctl_table_root, root_list);
1592 if (root == &sysctl_table_root)
1593 goto out;
1594 header_list = lookup_header_list(root, namespaces);
1595 } while (list_empty(header_list));
1596 tmp = header_list->next;
1597 }
1598 out:
1599 spin_unlock(&sysctl_lock);
1600 return NULL;
1601 }
1602
1603 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1604 {
1605 return __sysctl_head_next(current->nsproxy, prev);
1606 }
1607
1608 void register_sysctl_root(struct ctl_table_root *root)
1609 {
1610 spin_lock(&sysctl_lock);
1611 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1612 spin_unlock(&sysctl_lock);
1613 }
1614
1615 /*
1616 * sysctl_perm does NOT grant the superuser all rights automatically, because
1617 * some sysctl variables are readonly even to root.
1618 */
1619
1620 static int test_perm(int mode, int op)
1621 {
1622 if (!current_euid())
1623 mode >>= 6;
1624 else if (in_egroup_p(0))
1625 mode >>= 3;
1626 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1627 return 0;
1628 return -EACCES;
1629 }
1630
1631 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1632 {
1633 int error;
1634 int mode;
1635
1636 error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1637 if (error)
1638 return error;
1639
1640 if (root->permissions)
1641 mode = root->permissions(root, current->nsproxy, table);
1642 else
1643 mode = table->mode;
1644
1645 return test_perm(mode, op);
1646 }
1647
1648 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1649 {
1650 for (; table->procname; table++) {
1651 table->parent = parent;
1652 if (table->child)
1653 sysctl_set_parent(table, table->child);
1654 }
1655 }
1656
1657 static __init int sysctl_init(void)
1658 {
1659 sysctl_set_parent(NULL, root_table);
1660 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1661 {
1662 int err;
1663 err = sysctl_check_table(current->nsproxy, root_table);
1664 }
1665 #endif
1666 return 0;
1667 }
1668
1669 core_initcall(sysctl_init);
1670
1671 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1672 struct ctl_table *table)
1673 {
1674 struct ctl_table *p;
1675 const char *s = branch->procname;
1676
1677 /* branch should have named subdirectory as its first element */
1678 if (!s || !branch->child)
1679 return NULL;
1680
1681 /* ... and nothing else */
1682 if (branch[1].procname)
1683 return NULL;
1684
1685 /* table should contain subdirectory with the same name */
1686 for (p = table; p->procname; p++) {
1687 if (!p->child)
1688 continue;
1689 if (p->procname && strcmp(p->procname, s) == 0)
1690 return p;
1691 }
1692 return NULL;
1693 }
1694
1695 /* see if attaching q to p would be an improvement */
1696 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1697 {
1698 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1699 struct ctl_table *next;
1700 int is_better = 0;
1701 int not_in_parent = !p->attached_by;
1702
1703 while ((next = is_branch_in(by, to)) != NULL) {
1704 if (by == q->attached_by)
1705 is_better = 1;
1706 if (to == p->attached_by)
1707 not_in_parent = 1;
1708 by = by->child;
1709 to = next->child;
1710 }
1711
1712 if (is_better && not_in_parent) {
1713 q->attached_by = by;
1714 q->attached_to = to;
1715 q->parent = p;
1716 }
1717 }
1718
1719 /**
1720 * __register_sysctl_paths - register a sysctl hierarchy
1721 * @root: List of sysctl headers to register on
1722 * @namespaces: Data to compute which lists of sysctl entries are visible
1723 * @path: The path to the directory the sysctl table is in.
1724 * @table: the top-level table structure
1725 *
1726 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1727 * array. A completely 0 filled entry terminates the table.
1728 *
1729 * The members of the &struct ctl_table structure are used as follows:
1730 *
1731 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1732 * enter a sysctl file
1733 *
1734 * data - a pointer to data for use by proc_handler
1735 *
1736 * maxlen - the maximum size in bytes of the data
1737 *
1738 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1739 *
1740 * child - a pointer to the child sysctl table if this entry is a directory, or
1741 * %NULL.
1742 *
1743 * proc_handler - the text handler routine (described below)
1744 *
1745 * de - for internal use by the sysctl routines
1746 *
1747 * extra1, extra2 - extra pointers usable by the proc handler routines
1748 *
1749 * Leaf nodes in the sysctl tree will be represented by a single file
1750 * under /proc; non-leaf nodes will be represented by directories.
1751 *
1752 * sysctl(2) can automatically manage read and write requests through
1753 * the sysctl table. The data and maxlen fields of the ctl_table
1754 * struct enable minimal validation of the values being written to be
1755 * performed, and the mode field allows minimal authentication.
1756 *
1757 * There must be a proc_handler routine for any terminal nodes
1758 * mirrored under /proc/sys (non-terminals are handled by a built-in
1759 * directory handler). Several default handlers are available to
1760 * cover common cases -
1761 *
1762 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1763 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1764 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1765 *
1766 * It is the handler's job to read the input buffer from user memory
1767 * and process it. The handler should return 0 on success.
1768 *
1769 * This routine returns %NULL on a failure to register, and a pointer
1770 * to the table header on success.
1771 */
1772 struct ctl_table_header *__register_sysctl_paths(
1773 struct ctl_table_root *root,
1774 struct nsproxy *namespaces,
1775 const struct ctl_path *path, struct ctl_table *table)
1776 {
1777 struct ctl_table_header *header;
1778 struct ctl_table *new, **prevp;
1779 unsigned int n, npath;
1780 struct ctl_table_set *set;
1781
1782 /* Count the path components */
1783 for (npath = 0; path[npath].procname; ++npath)
1784 ;
1785
1786 /*
1787 * For each path component, allocate a 2-element ctl_table array.
1788 * The first array element will be filled with the sysctl entry
1789 * for this, the second will be the sentinel (procname == 0).
1790 *
1791 * We allocate everything in one go so that we don't have to
1792 * worry about freeing additional memory in unregister_sysctl_table.
1793 */
1794 header = kzalloc(sizeof(struct ctl_table_header) +
1795 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1796 if (!header)
1797 return NULL;
1798
1799 new = (struct ctl_table *) (header + 1);
1800
1801 /* Now connect the dots */
1802 prevp = &header->ctl_table;
1803 for (n = 0; n < npath; ++n, ++path) {
1804 /* Copy the procname */
1805 new->procname = path->procname;
1806 new->mode = 0555;
1807
1808 *prevp = new;
1809 prevp = &new->child;
1810
1811 new += 2;
1812 }
1813 *prevp = table;
1814 header->ctl_table_arg = table;
1815
1816 INIT_LIST_HEAD(&header->ctl_entry);
1817 header->used = 0;
1818 header->unregistering = NULL;
1819 header->root = root;
1820 sysctl_set_parent(NULL, header->ctl_table);
1821 header->count = 1;
1822 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1823 if (sysctl_check_table(namespaces, header->ctl_table)) {
1824 kfree(header);
1825 return NULL;
1826 }
1827 #endif
1828 spin_lock(&sysctl_lock);
1829 header->set = lookup_header_set(root, namespaces);
1830 header->attached_by = header->ctl_table;
1831 header->attached_to = root_table;
1832 header->parent = &root_table_header;
1833 for (set = header->set; set; set = set->parent) {
1834 struct ctl_table_header *p;
1835 list_for_each_entry(p, &set->list, ctl_entry) {
1836 if (p->unregistering)
1837 continue;
1838 try_attach(p, header);
1839 }
1840 }
1841 header->parent->count++;
1842 list_add_tail(&header->ctl_entry, &header->set->list);
1843 spin_unlock(&sysctl_lock);
1844
1845 return header;
1846 }
1847
1848 /**
1849 * register_sysctl_table_path - register a sysctl table hierarchy
1850 * @path: The path to the directory the sysctl table is in.
1851 * @table: the top-level table structure
1852 *
1853 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1854 * array. A completely 0 filled entry terminates the table.
1855 *
1856 * See __register_sysctl_paths for more details.
1857 */
1858 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1859 struct ctl_table *table)
1860 {
1861 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1862 path, table);
1863 }
1864
1865 /**
1866 * register_sysctl_table - register a sysctl table hierarchy
1867 * @table: the top-level table structure
1868 *
1869 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1870 * array. A completely 0 filled entry terminates the table.
1871 *
1872 * See register_sysctl_paths for more details.
1873 */
1874 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1875 {
1876 static const struct ctl_path null_path[] = { {} };
1877
1878 return register_sysctl_paths(null_path, table);
1879 }
1880
1881 /**
1882 * unregister_sysctl_table - unregister a sysctl table hierarchy
1883 * @header: the header returned from register_sysctl_table
1884 *
1885 * Unregisters the sysctl table and all children. proc entries may not
1886 * actually be removed until they are no longer used by anyone.
1887 */
1888 void unregister_sysctl_table(struct ctl_table_header * header)
1889 {
1890 might_sleep();
1891
1892 if (header == NULL)
1893 return;
1894
1895 spin_lock(&sysctl_lock);
1896 start_unregistering(header);
1897 if (!--header->parent->count) {
1898 WARN_ON(1);
1899 kfree(header->parent);
1900 }
1901 if (!--header->count)
1902 kfree(header);
1903 spin_unlock(&sysctl_lock);
1904 }
1905
1906 int sysctl_is_seen(struct ctl_table_header *p)
1907 {
1908 struct ctl_table_set *set = p->set;
1909 int res;
1910 spin_lock(&sysctl_lock);
1911 if (p->unregistering)
1912 res = 0;
1913 else if (!set->is_seen)
1914 res = 1;
1915 else
1916 res = set->is_seen(set);
1917 spin_unlock(&sysctl_lock);
1918 return res;
1919 }
1920
1921 void setup_sysctl_set(struct ctl_table_set *p,
1922 struct ctl_table_set *parent,
1923 int (*is_seen)(struct ctl_table_set *))
1924 {
1925 INIT_LIST_HEAD(&p->list);
1926 p->parent = parent ? parent : &sysctl_table_root.default_set;
1927 p->is_seen = is_seen;
1928 }
1929
1930 #else /* !CONFIG_SYSCTL */
1931 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1932 {
1933 return NULL;
1934 }
1935
1936 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1937 struct ctl_table *table)
1938 {
1939 return NULL;
1940 }
1941
1942 void unregister_sysctl_table(struct ctl_table_header * table)
1943 {
1944 }
1945
1946 void setup_sysctl_set(struct ctl_table_set *p,
1947 struct ctl_table_set *parent,
1948 int (*is_seen)(struct ctl_table_set *))
1949 {
1950 }
1951
1952 void sysctl_head_put(struct ctl_table_header *head)
1953 {
1954 }
1955
1956 #endif /* CONFIG_SYSCTL */
1957
1958 /*
1959 * /proc/sys support
1960 */
1961
1962 #ifdef CONFIG_PROC_SYSCTL
1963
1964 static int _proc_do_string(void* data, int maxlen, int write,
1965 void __user *buffer,
1966 size_t *lenp, loff_t *ppos)
1967 {
1968 size_t len;
1969 char __user *p;
1970 char c;
1971
1972 if (!data || !maxlen || !*lenp) {
1973 *lenp = 0;
1974 return 0;
1975 }
1976
1977 if (write) {
1978 len = 0;
1979 p = buffer;
1980 while (len < *lenp) {
1981 if (get_user(c, p++))
1982 return -EFAULT;
1983 if (c == 0 || c == '\n')
1984 break;
1985 len++;
1986 }
1987 if (len >= maxlen)
1988 len = maxlen-1;
1989 if(copy_from_user(data, buffer, len))
1990 return -EFAULT;
1991 ((char *) data)[len] = 0;
1992 *ppos += *lenp;
1993 } else {
1994 len = strlen(data);
1995 if (len > maxlen)
1996 len = maxlen;
1997
1998 if (*ppos > len) {
1999 *lenp = 0;
2000 return 0;
2001 }
2002
2003 data += *ppos;
2004 len -= *ppos;
2005
2006 if (len > *lenp)
2007 len = *lenp;
2008 if (len)
2009 if(copy_to_user(buffer, data, len))
2010 return -EFAULT;
2011 if (len < *lenp) {
2012 if(put_user('\n', ((char __user *) buffer) + len))
2013 return -EFAULT;
2014 len++;
2015 }
2016 *lenp = len;
2017 *ppos += len;
2018 }
2019 return 0;
2020 }
2021
2022 /**
2023 * proc_dostring - read a string sysctl
2024 * @table: the sysctl table
2025 * @write: %TRUE if this is a write to the sysctl file
2026 * @buffer: the user buffer
2027 * @lenp: the size of the user buffer
2028 * @ppos: file position
2029 *
2030 * Reads/writes a string from/to the user buffer. If the kernel
2031 * buffer provided is not large enough to hold the string, the
2032 * string is truncated. The copied string is %NULL-terminated.
2033 * If the string is being read by the user process, it is copied
2034 * and a newline '\n' is added. It is truncated if the buffer is
2035 * not large enough.
2036 *
2037 * Returns 0 on success.
2038 */
2039 int proc_dostring(struct ctl_table *table, int write,
2040 void __user *buffer, size_t *lenp, loff_t *ppos)
2041 {
2042 return _proc_do_string(table->data, table->maxlen, write,
2043 buffer, lenp, ppos);
2044 }
2045
2046
2047 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2048 int *valp,
2049 int write, void *data)
2050 {
2051 if (write) {
2052 *valp = *negp ? -*lvalp : *lvalp;
2053 } else {
2054 int val = *valp;
2055 if (val < 0) {
2056 *negp = -1;
2057 *lvalp = (unsigned long)-val;
2058 } else {
2059 *negp = 0;
2060 *lvalp = (unsigned long)val;
2061 }
2062 }
2063 return 0;
2064 }
2065
2066 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2067 int write, void __user *buffer,
2068 size_t *lenp, loff_t *ppos,
2069 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2070 int write, void *data),
2071 void *data)
2072 {
2073 #define TMPBUFLEN 21
2074 int *i, vleft, first = 1, neg;
2075 unsigned long lval;
2076 size_t left, len;
2077
2078 char buf[TMPBUFLEN], *p;
2079 char __user *s = buffer;
2080
2081 if (!tbl_data || !table->maxlen || !*lenp ||
2082 (*ppos && !write)) {
2083 *lenp = 0;
2084 return 0;
2085 }
2086
2087 i = (int *) tbl_data;
2088 vleft = table->maxlen / sizeof(*i);
2089 left = *lenp;
2090
2091 if (!conv)
2092 conv = do_proc_dointvec_conv;
2093
2094 for (; left && vleft--; i++, first=0) {
2095 if (write) {
2096 while (left) {
2097 char c;
2098 if (get_user(c, s))
2099 return -EFAULT;
2100 if (!isspace(c))
2101 break;
2102 left--;
2103 s++;
2104 }
2105 if (!left)
2106 break;
2107 neg = 0;
2108 len = left;
2109 if (len > sizeof(buf) - 1)
2110 len = sizeof(buf) - 1;
2111 if (copy_from_user(buf, s, len))
2112 return -EFAULT;
2113 buf[len] = 0;
2114 p = buf;
2115 if (*p == '-' && left > 1) {
2116 neg = 1;
2117 p++;
2118 }
2119 if (*p < '0' || *p > '9')
2120 break;
2121
2122 lval = simple_strtoul(p, &p, 0);
2123
2124 len = p-buf;
2125 if ((len < left) && *p && !isspace(*p))
2126 break;
2127 s += len;
2128 left -= len;
2129
2130 if (conv(&neg, &lval, i, 1, data))
2131 break;
2132 } else {
2133 p = buf;
2134 if (!first)
2135 *p++ = '\t';
2136
2137 if (conv(&neg, &lval, i, 0, data))
2138 break;
2139
2140 sprintf(p, "%s%lu", neg ? "-" : "", lval);
2141 len = strlen(buf);
2142 if (len > left)
2143 len = left;
2144 if(copy_to_user(s, buf, len))
2145 return -EFAULT;
2146 left -= len;
2147 s += len;
2148 }
2149 }
2150
2151 if (!write && !first && left) {
2152 if(put_user('\n', s))
2153 return -EFAULT;
2154 left--, s++;
2155 }
2156 if (write) {
2157 while (left) {
2158 char c;
2159 if (get_user(c, s++))
2160 return -EFAULT;
2161 if (!isspace(c))
2162 break;
2163 left--;
2164 }
2165 }
2166 if (write && first)
2167 return -EINVAL;
2168 *lenp -= left;
2169 *ppos += *lenp;
2170 return 0;
2171 #undef TMPBUFLEN
2172 }
2173
2174 static int do_proc_dointvec(struct ctl_table *table, int write,
2175 void __user *buffer, size_t *lenp, loff_t *ppos,
2176 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2177 int write, void *data),
2178 void *data)
2179 {
2180 return __do_proc_dointvec(table->data, table, write,
2181 buffer, lenp, ppos, conv, data);
2182 }
2183
2184 /**
2185 * proc_dointvec - read a vector of integers
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 * Returns 0 on success.
2196 */
2197 int proc_dointvec(struct ctl_table *table, int write,
2198 void __user *buffer, size_t *lenp, loff_t *ppos)
2199 {
2200 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2201 NULL,NULL);
2202 }
2203
2204 /*
2205 * Taint values can only be increased
2206 * This means we can safely use a temporary.
2207 */
2208 static int proc_taint(struct ctl_table *table, int write,
2209 void __user *buffer, size_t *lenp, loff_t *ppos)
2210 {
2211 struct ctl_table t;
2212 unsigned long tmptaint = get_taint();
2213 int err;
2214
2215 if (write && !capable(CAP_SYS_ADMIN))
2216 return -EPERM;
2217
2218 t = *table;
2219 t.data = &tmptaint;
2220 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2221 if (err < 0)
2222 return err;
2223
2224 if (write) {
2225 /*
2226 * Poor man's atomic or. Not worth adding a primitive
2227 * to everyone's atomic.h for this
2228 */
2229 int i;
2230 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2231 if ((tmptaint >> i) & 1)
2232 add_taint(i);
2233 }
2234 }
2235
2236 return err;
2237 }
2238
2239 struct do_proc_dointvec_minmax_conv_param {
2240 int *min;
2241 int *max;
2242 };
2243
2244 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2245 int *valp,
2246 int write, void *data)
2247 {
2248 struct do_proc_dointvec_minmax_conv_param *param = data;
2249 if (write) {
2250 int val = *negp ? -*lvalp : *lvalp;
2251 if ((param->min && *param->min > val) ||
2252 (param->max && *param->max < val))
2253 return -EINVAL;
2254 *valp = val;
2255 } else {
2256 int val = *valp;
2257 if (val < 0) {
2258 *negp = -1;
2259 *lvalp = (unsigned long)-val;
2260 } else {
2261 *negp = 0;
2262 *lvalp = (unsigned long)val;
2263 }
2264 }
2265 return 0;
2266 }
2267
2268 /**
2269 * proc_dointvec_minmax - read a vector of integers with min/max values
2270 * @table: the sysctl table
2271 * @write: %TRUE if this is a write to the sysctl file
2272 * @buffer: the user buffer
2273 * @lenp: the size of the user buffer
2274 * @ppos: file position
2275 *
2276 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2277 * values from/to the user buffer, treated as an ASCII string.
2278 *
2279 * This routine will ensure the values are within the range specified by
2280 * table->extra1 (min) and table->extra2 (max).
2281 *
2282 * Returns 0 on success.
2283 */
2284 int proc_dointvec_minmax(struct ctl_table *table, int write,
2285 void __user *buffer, size_t *lenp, loff_t *ppos)
2286 {
2287 struct do_proc_dointvec_minmax_conv_param param = {
2288 .min = (int *) table->extra1,
2289 .max = (int *) table->extra2,
2290 };
2291 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2292 do_proc_dointvec_minmax_conv, &param);
2293 }
2294
2295 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2296 void __user *buffer,
2297 size_t *lenp, loff_t *ppos,
2298 unsigned long convmul,
2299 unsigned long convdiv)
2300 {
2301 #define TMPBUFLEN 21
2302 unsigned long *i, *min, *max, val;
2303 int vleft, first=1, neg;
2304 size_t len, left;
2305 char buf[TMPBUFLEN], *p;
2306 char __user *s = buffer;
2307
2308 if (!data || !table->maxlen || !*lenp ||
2309 (*ppos && !write)) {
2310 *lenp = 0;
2311 return 0;
2312 }
2313
2314 i = (unsigned long *) data;
2315 min = (unsigned long *) table->extra1;
2316 max = (unsigned long *) table->extra2;
2317 vleft = table->maxlen / sizeof(unsigned long);
2318 left = *lenp;
2319
2320 for (; left && vleft--; i++, min++, max++, first=0) {
2321 if (write) {
2322 while (left) {
2323 char c;
2324 if (get_user(c, s))
2325 return -EFAULT;
2326 if (!isspace(c))
2327 break;
2328 left--;
2329 s++;
2330 }
2331 if (!left)
2332 break;
2333 neg = 0;
2334 len = left;
2335 if (len > TMPBUFLEN-1)
2336 len = TMPBUFLEN-1;
2337 if (copy_from_user(buf, s, len))
2338 return -EFAULT;
2339 buf[len] = 0;
2340 p = buf;
2341 if (*p == '-' && left > 1) {
2342 neg = 1;
2343 p++;
2344 }
2345 if (*p < '0' || *p > '9')
2346 break;
2347 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2348 len = p-buf;
2349 if ((len < left) && *p && !isspace(*p))
2350 break;
2351 if (neg)
2352 val = -val;
2353 s += len;
2354 left -= len;
2355
2356 if(neg)
2357 continue;
2358 if ((min && val < *min) || (max && val > *max))
2359 continue;
2360 *i = val;
2361 } else {
2362 p = buf;
2363 if (!first)
2364 *p++ = '\t';
2365 sprintf(p, "%lu", convdiv * (*i) / convmul);
2366 len = strlen(buf);
2367 if (len > left)
2368 len = left;
2369 if(copy_to_user(s, buf, len))
2370 return -EFAULT;
2371 left -= len;
2372 s += len;
2373 }
2374 }
2375
2376 if (!write && !first && left) {
2377 if(put_user('\n', s))
2378 return -EFAULT;
2379 left--, s++;
2380 }
2381 if (write) {
2382 while (left) {
2383 char c;
2384 if (get_user(c, s++))
2385 return -EFAULT;
2386 if (!isspace(c))
2387 break;
2388 left--;
2389 }
2390 }
2391 if (write && first)
2392 return -EINVAL;
2393 *lenp -= left;
2394 *ppos += *lenp;
2395 return 0;
2396 #undef TMPBUFLEN
2397 }
2398
2399 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2400 void __user *buffer,
2401 size_t *lenp, loff_t *ppos,
2402 unsigned long convmul,
2403 unsigned long convdiv)
2404 {
2405 return __do_proc_doulongvec_minmax(table->data, table, write,
2406 buffer, lenp, ppos, convmul, convdiv);
2407 }
2408
2409 /**
2410 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2411 * @table: the sysctl table
2412 * @write: %TRUE if this is a write to the sysctl file
2413 * @buffer: the user buffer
2414 * @lenp: the size of the user buffer
2415 * @ppos: file position
2416 *
2417 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2418 * values from/to the user buffer, treated as an ASCII string.
2419 *
2420 * This routine will ensure the values are within the range specified by
2421 * table->extra1 (min) and table->extra2 (max).
2422 *
2423 * Returns 0 on success.
2424 */
2425 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2426 void __user *buffer, size_t *lenp, loff_t *ppos)
2427 {
2428 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2429 }
2430
2431 /**
2432 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2433 * @table: the sysctl table
2434 * @write: %TRUE if this is a write to the sysctl file
2435 * @buffer: the user buffer
2436 * @lenp: the size of the user buffer
2437 * @ppos: file position
2438 *
2439 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2440 * values from/to the user buffer, treated as an ASCII string. The values
2441 * are treated as milliseconds, and converted to jiffies when they are stored.
2442 *
2443 * This routine will ensure the values are within the range specified by
2444 * table->extra1 (min) and table->extra2 (max).
2445 *
2446 * Returns 0 on success.
2447 */
2448 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2449 void __user *buffer,
2450 size_t *lenp, loff_t *ppos)
2451 {
2452 return do_proc_doulongvec_minmax(table, write, buffer,
2453 lenp, ppos, HZ, 1000l);
2454 }
2455
2456
2457 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2458 int *valp,
2459 int write, void *data)
2460 {
2461 if (write) {
2462 if (*lvalp > LONG_MAX / HZ)
2463 return 1;
2464 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2465 } else {
2466 int val = *valp;
2467 unsigned long lval;
2468 if (val < 0) {
2469 *negp = -1;
2470 lval = (unsigned long)-val;
2471 } else {
2472 *negp = 0;
2473 lval = (unsigned long)val;
2474 }
2475 *lvalp = lval / HZ;
2476 }
2477 return 0;
2478 }
2479
2480 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2481 int *valp,
2482 int write, void *data)
2483 {
2484 if (write) {
2485 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2486 return 1;
2487 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2488 } else {
2489 int val = *valp;
2490 unsigned long lval;
2491 if (val < 0) {
2492 *negp = -1;
2493 lval = (unsigned long)-val;
2494 } else {
2495 *negp = 0;
2496 lval = (unsigned long)val;
2497 }
2498 *lvalp = jiffies_to_clock_t(lval);
2499 }
2500 return 0;
2501 }
2502
2503 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2504 int *valp,
2505 int write, void *data)
2506 {
2507 if (write) {
2508 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2509 } else {
2510 int val = *valp;
2511 unsigned long lval;
2512 if (val < 0) {
2513 *negp = -1;
2514 lval = (unsigned long)-val;
2515 } else {
2516 *negp = 0;
2517 lval = (unsigned long)val;
2518 }
2519 *lvalp = jiffies_to_msecs(lval);
2520 }
2521 return 0;
2522 }
2523
2524 /**
2525 * proc_dointvec_jiffies - read a vector of integers as seconds
2526 * @table: the sysctl table
2527 * @write: %TRUE if this is a write to the sysctl file
2528 * @buffer: the user buffer
2529 * @lenp: the size of the user buffer
2530 * @ppos: file position
2531 *
2532 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2533 * values from/to the user buffer, treated as an ASCII string.
2534 * The values read are assumed to be in seconds, and are converted into
2535 * jiffies.
2536 *
2537 * Returns 0 on success.
2538 */
2539 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2540 void __user *buffer, size_t *lenp, loff_t *ppos)
2541 {
2542 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2543 do_proc_dointvec_jiffies_conv,NULL);
2544 }
2545
2546 /**
2547 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2548 * @table: the sysctl table
2549 * @write: %TRUE if this is a write to the sysctl file
2550 * @buffer: the user buffer
2551 * @lenp: the size of the user buffer
2552 * @ppos: pointer to the file position
2553 *
2554 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2555 * values from/to the user buffer, treated as an ASCII string.
2556 * The values read are assumed to be in 1/USER_HZ seconds, and
2557 * are converted into jiffies.
2558 *
2559 * Returns 0 on success.
2560 */
2561 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2562 void __user *buffer, size_t *lenp, loff_t *ppos)
2563 {
2564 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2565 do_proc_dointvec_userhz_jiffies_conv,NULL);
2566 }
2567
2568 /**
2569 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2570 * @table: the sysctl table
2571 * @write: %TRUE if this is a write to the sysctl file
2572 * @buffer: the user buffer
2573 * @lenp: the size of the user buffer
2574 * @ppos: file position
2575 * @ppos: the current position in the file
2576 *
2577 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2578 * values from/to the user buffer, treated as an ASCII string.
2579 * The values read are assumed to be in 1/1000 seconds, and
2580 * are converted into jiffies.
2581 *
2582 * Returns 0 on success.
2583 */
2584 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2585 void __user *buffer, size_t *lenp, loff_t *ppos)
2586 {
2587 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2588 do_proc_dointvec_ms_jiffies_conv, NULL);
2589 }
2590
2591 static int proc_do_cad_pid(struct ctl_table *table, int write,
2592 void __user *buffer, size_t *lenp, loff_t *ppos)
2593 {
2594 struct pid *new_pid;
2595 pid_t tmp;
2596 int r;
2597
2598 tmp = pid_vnr(cad_pid);
2599
2600 r = __do_proc_dointvec(&tmp, table, write, buffer,
2601 lenp, ppos, NULL, NULL);
2602 if (r || !write)
2603 return r;
2604
2605 new_pid = find_get_pid(tmp);
2606 if (!new_pid)
2607 return -ESRCH;
2608
2609 put_pid(xchg(&cad_pid, new_pid));
2610 return 0;
2611 }
2612
2613 #else /* CONFIG_PROC_FS */
2614
2615 int proc_dostring(struct ctl_table *table, int write,
2616 void __user *buffer, size_t *lenp, loff_t *ppos)
2617 {
2618 return -ENOSYS;
2619 }
2620
2621 int proc_dointvec(struct ctl_table *table, int write,
2622 void __user *buffer, size_t *lenp, loff_t *ppos)
2623 {
2624 return -ENOSYS;
2625 }
2626
2627 int proc_dointvec_minmax(struct ctl_table *table, int write,
2628 void __user *buffer, size_t *lenp, loff_t *ppos)
2629 {
2630 return -ENOSYS;
2631 }
2632
2633 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2634 void __user *buffer, size_t *lenp, loff_t *ppos)
2635 {
2636 return -ENOSYS;
2637 }
2638
2639 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2640 void __user *buffer, size_t *lenp, loff_t *ppos)
2641 {
2642 return -ENOSYS;
2643 }
2644
2645 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2646 void __user *buffer, size_t *lenp, loff_t *ppos)
2647 {
2648 return -ENOSYS;
2649 }
2650
2651 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2652 void __user *buffer, size_t *lenp, loff_t *ppos)
2653 {
2654 return -ENOSYS;
2655 }
2656
2657 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2658 void __user *buffer,
2659 size_t *lenp, loff_t *ppos)
2660 {
2661 return -ENOSYS;
2662 }
2663
2664
2665 #endif /* CONFIG_PROC_FS */
2666
2667 /*
2668 * No sense putting this after each symbol definition, twice,
2669 * exception granted :-)
2670 */
2671 EXPORT_SYMBOL(proc_dointvec);
2672 EXPORT_SYMBOL(proc_dointvec_jiffies);
2673 EXPORT_SYMBOL(proc_dointvec_minmax);
2674 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2675 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2676 EXPORT_SYMBOL(proc_dostring);
2677 EXPORT_SYMBOL(proc_doulongvec_minmax);
2678 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2679 EXPORT_SYMBOL(register_sysctl_table);
2680 EXPORT_SYMBOL(register_sysctl_paths);
2681 EXPORT_SYMBOL(unregister_sysctl_table);
x86 "subsystem" repository