?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
NFSv4: Release the sequence id before restarting a CLOSE rpc call
[linux-2.6.git] / kernel / sysctl.c
blob554ac4894f0f893a6f1e26a06e3076510f6dffe9
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/proc_fs.h>
27 #include <linux/security.h>
28 #include <linux/ctype.h>
29 #include <linux/kmemcheck.h>
30 #include <linux/fs.h>
31 #include <linux/init.h>
32 #include <linux/kernel.h>
33 #include <linux/kobject.h>
34 #include <linux/net.h>
35 #include <linux/sysrq.h>
36 #include <linux/highuid.h>
37 #include <linux/writeback.h>
38 #include <linux/ratelimit.h>
39 #include <linux/hugetlb.h>
40 #include <linux/initrd.h>
41 #include <linux/key.h>
42 #include <linux/times.h>
43 #include <linux/limits.h>
44 #include <linux/dcache.h>
45 #include <linux/syscalls.h>
46 #include <linux/vmstat.h>
47 #include <linux/nfs_fs.h>
48 #include <linux/acpi.h>
49 #include <linux/reboot.h>
50 #include <linux/ftrace.h>
51 #include <linux/slow-work.h>
52 #include <linux/perf_event.h>
53
54 #include <asm/uaccess.h>
55 #include <asm/processor.h>
56
57 #ifdef CONFIG_X86
58 #include <asm/nmi.h>
59 #include <asm/stacktrace.h>
60 #include <asm/io.h>
61 #endif
62
63
64 #if defined(CONFIG_SYSCTL)
65
66 /* External variables not in a header file. */
67 extern int C_A_D;
68 extern int print_fatal_signals;
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_RCU_TORTURE_TEST
91 extern int rcutorture_runnable;
92 #endif /* #ifdef CONFIG_RCU_TORTURE_TEST */
93 #ifdef CONFIG_BLOCK
94 extern int blk_iopoll_enabled;
95 #endif
96
97 /* Constants used for minimum and maximum */
98 #ifdef CONFIG_DETECT_SOFTLOCKUP
99 static int sixty = 60;
100 static int neg_one = -1;
101 #endif
102
103 static int zero;
104 static int __maybe_unused one = 1;
105 static int __maybe_unused two = 2;
106 static unsigned long one_ul = 1;
107 static int one_hundred = 100;
108 #ifdef CONFIG_PRINTK
109 static int ten_thousand = 10000;
110 #endif
111
112 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
113 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
114
115 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
116 static int maxolduid = 65535;
117 static int minolduid;
118 static int min_percpu_pagelist_fract = 8;
119
120 static int ngroups_max = NGROUPS_MAX;
121
122 #ifdef CONFIG_MODULES
123 extern char modprobe_path[];
124 extern int modules_disabled;
125 #endif
126 #ifdef CONFIG_CHR_DEV_SG
127 extern int sg_big_buff;
128 #endif
129
130 #ifdef CONFIG_SPARC
131 #include <asm/system.h>
132 #endif
133
134 #ifdef CONFIG_SPARC64
135 extern int sysctl_tsb_ratio;
136 #endif
137
138 #ifdef __hppa__
139 extern int pwrsw_enabled;
140 extern int unaligned_enabled;
141 #endif
142
143 #ifdef CONFIG_S390
144 #ifdef CONFIG_MATHEMU
145 extern int sysctl_ieee_emulation_warnings;
146 #endif
147 extern int sysctl_userprocess_debug;
148 extern int spin_retry;
149 #endif
150
151 #ifdef CONFIG_BSD_PROCESS_ACCT
152 extern int acct_parm[];
153 #endif
154
155 #ifdef CONFIG_IA64
156 extern int no_unaligned_warning;
157 extern int unaligned_dump_stack;
158 #endif
159
160 extern struct ratelimit_state printk_ratelimit_state;
161
162 #ifdef CONFIG_RT_MUTEXES
163 extern int max_lock_depth;
164 #endif
165
166 #ifdef CONFIG_PROC_SYSCTL
167 static int proc_do_cad_pid(struct ctl_table *table, int write,
168 void __user *buffer, size_t *lenp, loff_t *ppos);
169 static int proc_taint(struct ctl_table *table, int write,
170 void __user *buffer, size_t *lenp, loff_t *ppos);
171 #endif
172
173 static struct ctl_table root_table[];
174 static struct ctl_table_root sysctl_table_root;
175 static struct ctl_table_header root_table_header = {
176 .count = 1,
177 .ctl_table = root_table,
178 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
179 .root = &sysctl_table_root,
180 .set = &sysctl_table_root.default_set,
181 };
182 static struct ctl_table_root sysctl_table_root = {
183 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
184 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
185 };
186
187 static struct ctl_table kern_table[];
188 static struct ctl_table vm_table[];
189 static struct ctl_table fs_table[];
190 static struct ctl_table debug_table[];
191 static struct ctl_table dev_table[];
192 extern struct ctl_table random_table[];
193 #ifdef CONFIG_INOTIFY_USER
194 extern struct ctl_table inotify_table[];
195 #endif
196 #ifdef CONFIG_EPOLL
197 extern struct ctl_table epoll_table[];
198 #endif
199
200 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
201 int sysctl_legacy_va_layout;
202 #endif
203
204 extern int prove_locking;
205 extern int lock_stat;
206
207 /* The default sysctl tables: */
208
209 static struct ctl_table root_table[] = {
210 {
211 .procname = "kernel",
212 .mode = 0555,
213 .child = kern_table,
214 },
215 {
216 .procname = "vm",
217 .mode = 0555,
218 .child = vm_table,
219 },
220 {
221 .procname = "fs",
222 .mode = 0555,
223 .child = fs_table,
224 },
225 {
226 .procname = "debug",
227 .mode = 0555,
228 .child = debug_table,
229 },
230 {
231 .procname = "dev",
232 .mode = 0555,
233 .child = dev_table,
234 },
235 /*
236 * NOTE: do not add new entries to this table unless you have read
237 * Documentation/sysctl/ctl_unnumbered.txt
238 */
239 { }
240 };
241
242 #ifdef CONFIG_SCHED_DEBUG
243 static int min_sched_granularity_ns = 100000; /* 100 usecs */
244 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
245 static int min_wakeup_granularity_ns; /* 0 usecs */
246 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
247 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
248 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
249 static int min_sched_shares_ratelimit = 100000; /* 100 usec */
250 static int max_sched_shares_ratelimit = NSEC_PER_SEC; /* 1 second */
251 #endif
252
253 static struct ctl_table kern_table[] = {
254 {
255 .procname = "sched_child_runs_first",
256 .data = &sysctl_sched_child_runs_first,
257 .maxlen = sizeof(unsigned int),
258 .mode = 0644,
259 .proc_handler = proc_dointvec,
260 },
261 #ifdef CONFIG_SCHED_DEBUG
262 {
263 .procname = "sched_min_granularity_ns",
264 .data = &sysctl_sched_min_granularity,
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_latency_ns",
273 .data = &sysctl_sched_latency,
274 .maxlen = sizeof(unsigned int),
275 .mode = 0644,
276 .proc_handler = sched_proc_update_handler,
277 .extra1 = &min_sched_granularity_ns,
278 .extra2 = &max_sched_granularity_ns,
279 },
280 {
281 .procname = "sched_wakeup_granularity_ns",
282 .data = &sysctl_sched_wakeup_granularity,
283 .maxlen = sizeof(unsigned int),
284 .mode = 0644,
285 .proc_handler = sched_proc_update_handler,
286 .extra1 = &min_wakeup_granularity_ns,
287 .extra2 = &max_wakeup_granularity_ns,
288 },
289 {
290 .procname = "sched_shares_ratelimit",
291 .data = &sysctl_sched_shares_ratelimit,
292 .maxlen = sizeof(unsigned int),
293 .mode = 0644,
294 .proc_handler = sched_proc_update_handler,
295 .extra1 = &min_sched_shares_ratelimit,
296 .extra2 = &max_sched_shares_ratelimit,
297 },
298 {
299 .procname = "sched_tunable_scaling",
300 .data = &sysctl_sched_tunable_scaling,
301 .maxlen = sizeof(enum sched_tunable_scaling),
302 .mode = 0644,
303 .proc_handler = sched_proc_update_handler,
304 .extra1 = &min_sched_tunable_scaling,
305 .extra2 = &max_sched_tunable_scaling,
306 },
307 {
308 .procname = "sched_shares_thresh",
309 .data = &sysctl_sched_shares_thresh,
310 .maxlen = sizeof(unsigned int),
311 .mode = 0644,
312 .proc_handler = proc_dointvec_minmax,
313 .extra1 = &zero,
314 },
315 {
316 .procname = "sched_migration_cost",
317 .data = &sysctl_sched_migration_cost,
318 .maxlen = sizeof(unsigned int),
319 .mode = 0644,
320 .proc_handler = proc_dointvec,
321 },
322 {
323 .procname = "sched_nr_migrate",
324 .data = &sysctl_sched_nr_migrate,
325 .maxlen = sizeof(unsigned int),
326 .mode = 0644,
327 .proc_handler = proc_dointvec,
328 },
329 {
330 .procname = "sched_time_avg",
331 .data = &sysctl_sched_time_avg,
332 .maxlen = sizeof(unsigned int),
333 .mode = 0644,
334 .proc_handler = proc_dointvec,
335 },
336 {
337 .procname = "timer_migration",
338 .data = &sysctl_timer_migration,
339 .maxlen = sizeof(unsigned int),
340 .mode = 0644,
341 .proc_handler = proc_dointvec_minmax,
342 .extra1 = &zero,
343 .extra2 = &one,
344 },
345 #endif
346 {
347 .procname = "sched_rt_period_us",
348 .data = &sysctl_sched_rt_period,
349 .maxlen = sizeof(unsigned int),
350 .mode = 0644,
351 .proc_handler = sched_rt_handler,
352 },
353 {
354 .procname = "sched_rt_runtime_us",
355 .data = &sysctl_sched_rt_runtime,
356 .maxlen = sizeof(int),
357 .mode = 0644,
358 .proc_handler = sched_rt_handler,
359 },
360 {
361 .procname = "sched_compat_yield",
362 .data = &sysctl_sched_compat_yield,
363 .maxlen = sizeof(unsigned int),
364 .mode = 0644,
365 .proc_handler = proc_dointvec,
366 },
367 #ifdef CONFIG_PROVE_LOCKING
368 {
369 .procname = "prove_locking",
370 .data = &prove_locking,
371 .maxlen = sizeof(int),
372 .mode = 0644,
373 .proc_handler = proc_dointvec,
374 },
375 #endif
376 #ifdef CONFIG_LOCK_STAT
377 {
378 .procname = "lock_stat",
379 .data = &lock_stat,
380 .maxlen = sizeof(int),
381 .mode = 0644,
382 .proc_handler = proc_dointvec,
383 },
384 #endif
385 {
386 .procname = "panic",
387 .data = &panic_timeout,
388 .maxlen = sizeof(int),
389 .mode = 0644,
390 .proc_handler = proc_dointvec,
391 },
392 {
393 .procname = "core_uses_pid",
394 .data = &core_uses_pid,
395 .maxlen = sizeof(int),
396 .mode = 0644,
397 .proc_handler = proc_dointvec,
398 },
399 {
400 .procname = "core_pattern",
401 .data = core_pattern,
402 .maxlen = CORENAME_MAX_SIZE,
403 .mode = 0644,
404 .proc_handler = proc_dostring,
405 },
406 {
407 .procname = "core_pipe_limit",
408 .data = &core_pipe_limit,
409 .maxlen = sizeof(unsigned int),
410 .mode = 0644,
411 .proc_handler = proc_dointvec,
412 },
413 #ifdef CONFIG_PROC_SYSCTL
414 {
415 .procname = "tainted",
416 .maxlen = sizeof(long),
417 .mode = 0644,
418 .proc_handler = proc_taint,
419 },
420 #endif
421 #ifdef CONFIG_LATENCYTOP
422 {
423 .procname = "latencytop",
424 .data = &latencytop_enabled,
425 .maxlen = sizeof(int),
426 .mode = 0644,
427 .proc_handler = proc_dointvec,
428 },
429 #endif
430 #ifdef CONFIG_BLK_DEV_INITRD
431 {
432 .procname = "real-root-dev",
433 .data = &real_root_dev,
434 .maxlen = sizeof(int),
435 .mode = 0644,
436 .proc_handler = proc_dointvec,
437 },
438 #endif
439 {
440 .procname = "print-fatal-signals",
441 .data = &print_fatal_signals,
442 .maxlen = sizeof(int),
443 .mode = 0644,
444 .proc_handler = proc_dointvec,
445 },
446 #ifdef CONFIG_SPARC
447 {
448 .procname = "reboot-cmd",
449 .data = reboot_command,
450 .maxlen = 256,
451 .mode = 0644,
452 .proc_handler = proc_dostring,
453 },
454 {
455 .procname = "stop-a",
456 .data = &stop_a_enabled,
457 .maxlen = sizeof (int),
458 .mode = 0644,
459 .proc_handler = proc_dointvec,
460 },
461 {
462 .procname = "scons-poweroff",
463 .data = &scons_pwroff,
464 .maxlen = sizeof (int),
465 .mode = 0644,
466 .proc_handler = proc_dointvec,
467 },
468 #endif
469 #ifdef CONFIG_SPARC64
470 {
471 .procname = "tsb-ratio",
472 .data = &sysctl_tsb_ratio,
473 .maxlen = sizeof (int),
474 .mode = 0644,
475 .proc_handler = proc_dointvec,
476 },
477 #endif
478 #ifdef __hppa__
479 {
480 .procname = "soft-power",
481 .data = &pwrsw_enabled,
482 .maxlen = sizeof (int),
483 .mode = 0644,
484 .proc_handler = proc_dointvec,
485 },
486 {
487 .procname = "unaligned-trap",
488 .data = &unaligned_enabled,
489 .maxlen = sizeof (int),
490 .mode = 0644,
491 .proc_handler = proc_dointvec,
492 },
493 #endif
494 {
495 .procname = "ctrl-alt-del",
496 .data = &C_A_D,
497 .maxlen = sizeof(int),
498 .mode = 0644,
499 .proc_handler = proc_dointvec,
500 },
501 #ifdef CONFIG_FUNCTION_TRACER
502 {
503 .procname = "ftrace_enabled",
504 .data = &ftrace_enabled,
505 .maxlen = sizeof(int),
506 .mode = 0644,
507 .proc_handler = ftrace_enable_sysctl,
508 },
509 #endif
510 #ifdef CONFIG_STACK_TRACER
511 {
512 .procname = "stack_tracer_enabled",
513 .data = &stack_tracer_enabled,
514 .maxlen = sizeof(int),
515 .mode = 0644,
516 .proc_handler = stack_trace_sysctl,
517 },
518 #endif
519 #ifdef CONFIG_TRACING
520 {
521 .procname = "ftrace_dump_on_oops",
522 .data = &ftrace_dump_on_oops,
523 .maxlen = sizeof(int),
524 .mode = 0644,
525 .proc_handler = proc_dointvec,
526 },
527 #endif
528 #ifdef CONFIG_MODULES
529 {
530 .procname = "modprobe",
531 .data = &modprobe_path,
532 .maxlen = KMOD_PATH_LEN,
533 .mode = 0644,
534 .proc_handler = proc_dostring,
535 },
536 {
537 .procname = "modules_disabled",
538 .data = &modules_disabled,
539 .maxlen = sizeof(int),
540 .mode = 0644,
541 /* only handle a transition from default "0" to "1" */
542 .proc_handler = proc_dointvec_minmax,
543 .extra1 = &one,
544 .extra2 = &one,
545 },
546 #endif
547 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
548 {
549 .procname = "hotplug",
550 .data = &uevent_helper,
551 .maxlen = UEVENT_HELPER_PATH_LEN,
552 .mode = 0644,
553 .proc_handler = proc_dostring,
554 },
555 #endif
556 #ifdef CONFIG_CHR_DEV_SG
557 {
558 .procname = "sg-big-buff",
559 .data = &sg_big_buff,
560 .maxlen = sizeof (int),
561 .mode = 0444,
562 .proc_handler = proc_dointvec,
563 },
564 #endif
565 #ifdef CONFIG_BSD_PROCESS_ACCT
566 {
567 .procname = "acct",
568 .data = &acct_parm,
569 .maxlen = 3*sizeof(int),
570 .mode = 0644,
571 .proc_handler = proc_dointvec,
572 },
573 #endif
574 #ifdef CONFIG_MAGIC_SYSRQ
575 {
576 .procname = "sysrq",
577 .data = &__sysrq_enabled,
578 .maxlen = sizeof (int),
579 .mode = 0644,
580 .proc_handler = proc_dointvec,
581 },
582 #endif
583 #ifdef CONFIG_PROC_SYSCTL
584 {
585 .procname = "cad_pid",
586 .data = NULL,
587 .maxlen = sizeof (int),
588 .mode = 0600,
589 .proc_handler = proc_do_cad_pid,
590 },
591 #endif
592 {
593 .procname = "threads-max",
594 .data = &max_threads,
595 .maxlen = sizeof(int),
596 .mode = 0644,
597 .proc_handler = proc_dointvec,
598 },
599 {
600 .procname = "random",
601 .mode = 0555,
602 .child = random_table,
603 },
604 {
605 .procname = "overflowuid",
606 .data = &overflowuid,
607 .maxlen = sizeof(int),
608 .mode = 0644,
609 .proc_handler = proc_dointvec_minmax,
610 .extra1 = &minolduid,
611 .extra2 = &maxolduid,
612 },
613 {
614 .procname = "overflowgid",
615 .data = &overflowgid,
616 .maxlen = sizeof(int),
617 .mode = 0644,
618 .proc_handler = proc_dointvec_minmax,
619 .extra1 = &minolduid,
620 .extra2 = &maxolduid,
621 },
622 #ifdef CONFIG_S390
623 #ifdef CONFIG_MATHEMU
624 {
625 .procname = "ieee_emulation_warnings",
626 .data = &sysctl_ieee_emulation_warnings,
627 .maxlen = sizeof(int),
628 .mode = 0644,
629 .proc_handler = proc_dointvec,
630 },
631 #endif
632 {
633 .procname = "userprocess_debug",
634 .data = &sysctl_userprocess_debug,
635 .maxlen = sizeof(int),
636 .mode = 0644,
637 .proc_handler = proc_dointvec,
638 },
639 #endif
640 {
641 .procname = "pid_max",
642 .data = &pid_max,
643 .maxlen = sizeof (int),
644 .mode = 0644,
645 .proc_handler = proc_dointvec_minmax,
646 .extra1 = &pid_max_min,
647 .extra2 = &pid_max_max,
648 },
649 {
650 .procname = "panic_on_oops",
651 .data = &panic_on_oops,
652 .maxlen = sizeof(int),
653 .mode = 0644,
654 .proc_handler = proc_dointvec,
655 },
656 #if defined CONFIG_PRINTK
657 {
658 .procname = "printk",
659 .data = &console_loglevel,
660 .maxlen = 4*sizeof(int),
661 .mode = 0644,
662 .proc_handler = proc_dointvec,
663 },
664 {
665 .procname = "printk_ratelimit",
666 .data = &printk_ratelimit_state.interval,
667 .maxlen = sizeof(int),
668 .mode = 0644,
669 .proc_handler = proc_dointvec_jiffies,
670 },
671 {
672 .procname = "printk_ratelimit_burst",
673 .data = &printk_ratelimit_state.burst,
674 .maxlen = sizeof(int),
675 .mode = 0644,
676 .proc_handler = proc_dointvec,
677 },
678 {
679 .procname = "printk_delay",
680 .data = &printk_delay_msec,
681 .maxlen = sizeof(int),
682 .mode = 0644,
683 .proc_handler = proc_dointvec_minmax,
684 .extra1 = &zero,
685 .extra2 = &ten_thousand,
686 },
687 #endif
688 {
689 .procname = "ngroups_max",
690 .data = &ngroups_max,
691 .maxlen = sizeof (int),
692 .mode = 0444,
693 .proc_handler = proc_dointvec,
694 },
695 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
696 {
697 .procname = "unknown_nmi_panic",
698 .data = &unknown_nmi_panic,
699 .maxlen = sizeof (int),
700 .mode = 0644,
701 .proc_handler = proc_dointvec,
702 },
703 {
704 .procname = "nmi_watchdog",
705 .data = &nmi_watchdog_enabled,
706 .maxlen = sizeof (int),
707 .mode = 0644,
708 .proc_handler = proc_nmi_enabled,
709 },
710 #endif
711 #if defined(CONFIG_X86)
712 {
713 .procname = "panic_on_unrecovered_nmi",
714 .data = &panic_on_unrecovered_nmi,
715 .maxlen = sizeof(int),
716 .mode = 0644,
717 .proc_handler = proc_dointvec,
718 },
719 {
720 .procname = "panic_on_io_nmi",
721 .data = &panic_on_io_nmi,
722 .maxlen = sizeof(int),
723 .mode = 0644,
724 .proc_handler = proc_dointvec,
725 },
726 {
727 .procname = "bootloader_type",
728 .data = &bootloader_type,
729 .maxlen = sizeof (int),
730 .mode = 0444,
731 .proc_handler = proc_dointvec,
732 },
733 {
734 .procname = "bootloader_version",
735 .data = &bootloader_version,
736 .maxlen = sizeof (int),
737 .mode = 0444,
738 .proc_handler = proc_dointvec,
739 },
740 {
741 .procname = "kstack_depth_to_print",
742 .data = &kstack_depth_to_print,
743 .maxlen = sizeof(int),
744 .mode = 0644,
745 .proc_handler = proc_dointvec,
746 },
747 {
748 .procname = "io_delay_type",
749 .data = &io_delay_type,
750 .maxlen = sizeof(int),
751 .mode = 0644,
752 .proc_handler = proc_dointvec,
753 },
754 #endif
755 #if defined(CONFIG_MMU)
756 {
757 .procname = "randomize_va_space",
758 .data = &randomize_va_space,
759 .maxlen = sizeof(int),
760 .mode = 0644,
761 .proc_handler = proc_dointvec,
762 },
763 #endif
764 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
765 {
766 .procname = "spin_retry",
767 .data = &spin_retry,
768 .maxlen = sizeof (int),
769 .mode = 0644,
770 .proc_handler = proc_dointvec,
771 },
772 #endif
773 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
774 {
775 .procname = "acpi_video_flags",
776 .data = &acpi_realmode_flags,
777 .maxlen = sizeof (unsigned long),
778 .mode = 0644,
779 .proc_handler = proc_doulongvec_minmax,
780 },
781 #endif
782 #ifdef CONFIG_IA64
783 {
784 .procname = "ignore-unaligned-usertrap",
785 .data = &no_unaligned_warning,
786 .maxlen = sizeof (int),
787 .mode = 0644,
788 .proc_handler = proc_dointvec,
789 },
790 {
791 .procname = "unaligned-dump-stack",
792 .data = &unaligned_dump_stack,
793 .maxlen = sizeof (int),
794 .mode = 0644,
795 .proc_handler = proc_dointvec,
796 },
797 #endif
798 #ifdef CONFIG_DETECT_SOFTLOCKUP
799 {
800 .procname = "softlockup_panic",
801 .data = &softlockup_panic,
802 .maxlen = sizeof(int),
803 .mode = 0644,
804 .proc_handler = proc_dointvec_minmax,
805 .extra1 = &zero,
806 .extra2 = &one,
807 },
808 {
809 .procname = "softlockup_thresh",
810 .data = &softlockup_thresh,
811 .maxlen = sizeof(int),
812 .mode = 0644,
813 .proc_handler = proc_dosoftlockup_thresh,
814 .extra1 = &neg_one,
815 .extra2 = &sixty,
816 },
817 #endif
818 #ifdef CONFIG_DETECT_HUNG_TASK
819 {
820 .procname = "hung_task_panic",
821 .data = &sysctl_hung_task_panic,
822 .maxlen = sizeof(int),
823 .mode = 0644,
824 .proc_handler = proc_dointvec_minmax,
825 .extra1 = &zero,
826 .extra2 = &one,
827 },
828 {
829 .procname = "hung_task_check_count",
830 .data = &sysctl_hung_task_check_count,
831 .maxlen = sizeof(unsigned long),
832 .mode = 0644,
833 .proc_handler = proc_doulongvec_minmax,
834 },
835 {
836 .procname = "hung_task_timeout_secs",
837 .data = &sysctl_hung_task_timeout_secs,
838 .maxlen = sizeof(unsigned long),
839 .mode = 0644,
840 .proc_handler = proc_dohung_task_timeout_secs,
841 },
842 {
843 .procname = "hung_task_warnings",
844 .data = &sysctl_hung_task_warnings,
845 .maxlen = sizeof(unsigned long),
846 .mode = 0644,
847 .proc_handler = proc_doulongvec_minmax,
848 },
849 #endif
850 #ifdef CONFIG_COMPAT
851 {
852 .procname = "compat-log",
853 .data = &compat_log,
854 .maxlen = sizeof (int),
855 .mode = 0644,
856 .proc_handler = proc_dointvec,
857 },
858 #endif
859 #ifdef CONFIG_RT_MUTEXES
860 {
861 .procname = "max_lock_depth",
862 .data = &max_lock_depth,
863 .maxlen = sizeof(int),
864 .mode = 0644,
865 .proc_handler = proc_dointvec,
866 },
867 #endif
868 {
869 .procname = "poweroff_cmd",
870 .data = &poweroff_cmd,
871 .maxlen = POWEROFF_CMD_PATH_LEN,
872 .mode = 0644,
873 .proc_handler = proc_dostring,
874 },
875 #ifdef CONFIG_KEYS
876 {
877 .procname = "keys",
878 .mode = 0555,
879 .child = key_sysctls,
880 },
881 #endif
882 #ifdef CONFIG_RCU_TORTURE_TEST
883 {
884 .procname = "rcutorture_runnable",
885 .data = &rcutorture_runnable,
886 .maxlen = sizeof(int),
887 .mode = 0644,
888 .proc_handler = proc_dointvec,
889 },
890 #endif
891 #ifdef CONFIG_SLOW_WORK
892 {
893 .procname = "slow-work",
894 .mode = 0555,
895 .child = slow_work_sysctls,
896 },
897 #endif
898 #ifdef CONFIG_PERF_EVENTS
899 {
900 .procname = "perf_event_paranoid",
901 .data = &sysctl_perf_event_paranoid,
902 .maxlen = sizeof(sysctl_perf_event_paranoid),
903 .mode = 0644,
904 .proc_handler = proc_dointvec,
905 },
906 {
907 .procname = "perf_event_mlock_kb",
908 .data = &sysctl_perf_event_mlock,
909 .maxlen = sizeof(sysctl_perf_event_mlock),
910 .mode = 0644,
911 .proc_handler = proc_dointvec,
912 },
913 {
914 .procname = "perf_event_max_sample_rate",
915 .data = &sysctl_perf_event_sample_rate,
916 .maxlen = sizeof(sysctl_perf_event_sample_rate),
917 .mode = 0644,
918 .proc_handler = proc_dointvec,
919 },
920 #endif
921 #ifdef CONFIG_KMEMCHECK
922 {
923 .procname = "kmemcheck",
924 .data = &kmemcheck_enabled,
925 .maxlen = sizeof(int),
926 .mode = 0644,
927 .proc_handler = proc_dointvec,
928 },
929 #endif
930 #ifdef CONFIG_BLOCK
931 {
932 .procname = "blk_iopoll",
933 .data = &blk_iopoll_enabled,
934 .maxlen = sizeof(int),
935 .mode = 0644,
936 .proc_handler = proc_dointvec,
937 },
938 #endif
939 /*
940 * NOTE: do not add new entries to this table unless you have read
941 * Documentation/sysctl/ctl_unnumbered.txt
942 */
943 { }
944 };
945
946 static struct ctl_table vm_table[] = {
947 {
948 .procname = "overcommit_memory",
949 .data = &sysctl_overcommit_memory,
950 .maxlen = sizeof(sysctl_overcommit_memory),
951 .mode = 0644,
952 .proc_handler = proc_dointvec,
953 },
954 {
955 .procname = "panic_on_oom",
956 .data = &sysctl_panic_on_oom,
957 .maxlen = sizeof(sysctl_panic_on_oom),
958 .mode = 0644,
959 .proc_handler = proc_dointvec,
960 },
961 {
962 .procname = "oom_kill_allocating_task",
963 .data = &sysctl_oom_kill_allocating_task,
964 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
965 .mode = 0644,
966 .proc_handler = proc_dointvec,
967 },
968 {
969 .procname = "oom_dump_tasks",
970 .data = &sysctl_oom_dump_tasks,
971 .maxlen = sizeof(sysctl_oom_dump_tasks),
972 .mode = 0644,
973 .proc_handler = proc_dointvec,
974 },
975 {
976 .procname = "overcommit_ratio",
977 .data = &sysctl_overcommit_ratio,
978 .maxlen = sizeof(sysctl_overcommit_ratio),
979 .mode = 0644,
980 .proc_handler = proc_dointvec,
981 },
982 {
983 .procname = "page-cluster",
984 .data = &page_cluster,
985 .maxlen = sizeof(int),
986 .mode = 0644,
987 .proc_handler = proc_dointvec,
988 },
989 {
990 .procname = "dirty_background_ratio",
991 .data = &dirty_background_ratio,
992 .maxlen = sizeof(dirty_background_ratio),
993 .mode = 0644,
994 .proc_handler = dirty_background_ratio_handler,
995 .extra1 = &zero,
996 .extra2 = &one_hundred,
997 },
998 {
999 .procname = "dirty_background_bytes",
1000 .data = &dirty_background_bytes,
1001 .maxlen = sizeof(dirty_background_bytes),
1002 .mode = 0644,
1003 .proc_handler = dirty_background_bytes_handler,
1004 .extra1 = &one_ul,
1005 },
1006 {
1007 .procname = "dirty_ratio",
1008 .data = &vm_dirty_ratio,
1009 .maxlen = sizeof(vm_dirty_ratio),
1010 .mode = 0644,
1011 .proc_handler = dirty_ratio_handler,
1012 .extra1 = &zero,
1013 .extra2 = &one_hundred,
1014 },
1015 {
1016 .procname = "dirty_bytes",
1017 .data = &vm_dirty_bytes,
1018 .maxlen = sizeof(vm_dirty_bytes),
1019 .mode = 0644,
1020 .proc_handler = dirty_bytes_handler,
1021 .extra1 = &dirty_bytes_min,
1022 },
1023 {
1024 .procname = "dirty_writeback_centisecs",
1025 .data = &dirty_writeback_interval,
1026 .maxlen = sizeof(dirty_writeback_interval),
1027 .mode = 0644,
1028 .proc_handler = dirty_writeback_centisecs_handler,
1029 },
1030 {
1031 .procname = "dirty_expire_centisecs",
1032 .data = &dirty_expire_interval,
1033 .maxlen = sizeof(dirty_expire_interval),
1034 .mode = 0644,
1035 .proc_handler = proc_dointvec,
1036 },
1037 {
1038 .procname = "nr_pdflush_threads",
1039 .data = &nr_pdflush_threads,
1040 .maxlen = sizeof nr_pdflush_threads,
1041 .mode = 0444 /* read-only*/,
1042 .proc_handler = proc_dointvec,
1043 },
1044 {
1045 .procname = "swappiness",
1046 .data = &vm_swappiness,
1047 .maxlen = sizeof(vm_swappiness),
1048 .mode = 0644,
1049 .proc_handler = proc_dointvec_minmax,
1050 .extra1 = &zero,
1051 .extra2 = &one_hundred,
1052 },
1053 #ifdef CONFIG_HUGETLB_PAGE
1054 {
1055 .procname = "nr_hugepages",
1056 .data = NULL,
1057 .maxlen = sizeof(unsigned long),
1058 .mode = 0644,
1059 .proc_handler = hugetlb_sysctl_handler,
1060 .extra1 = (void *)&hugetlb_zero,
1061 .extra2 = (void *)&hugetlb_infinity,
1062 },
1063 {
1064 .procname = "hugetlb_shm_group",
1065 .data = &sysctl_hugetlb_shm_group,
1066 .maxlen = sizeof(gid_t),
1067 .mode = 0644,
1068 .proc_handler = proc_dointvec,
1069 },
1070 {
1071 .procname = "hugepages_treat_as_movable",
1072 .data = &hugepages_treat_as_movable,
1073 .maxlen = sizeof(int),
1074 .mode = 0644,
1075 .proc_handler = hugetlb_treat_movable_handler,
1076 },
1077 {
1078 .procname = "nr_overcommit_hugepages",
1079 .data = NULL,
1080 .maxlen = sizeof(unsigned long),
1081 .mode = 0644,
1082 .proc_handler = hugetlb_overcommit_handler,
1083 .extra1 = (void *)&hugetlb_zero,
1084 .extra2 = (void *)&hugetlb_infinity,
1085 },
1086 #endif
1087 {
1088 .procname = "lowmem_reserve_ratio",
1089 .data = &sysctl_lowmem_reserve_ratio,
1090 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1091 .mode = 0644,
1092 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1093 },
1094 {
1095 .procname = "drop_caches",
1096 .data = &sysctl_drop_caches,
1097 .maxlen = sizeof(int),
1098 .mode = 0644,
1099 .proc_handler = drop_caches_sysctl_handler,
1100 },
1101 {
1102 .procname = "min_free_kbytes",
1103 .data = &min_free_kbytes,
1104 .maxlen = sizeof(min_free_kbytes),
1105 .mode = 0644,
1106 .proc_handler = min_free_kbytes_sysctl_handler,
1107 .extra1 = &zero,
1108 },
1109 {
1110 .procname = "percpu_pagelist_fraction",
1111 .data = &percpu_pagelist_fraction,
1112 .maxlen = sizeof(percpu_pagelist_fraction),
1113 .mode = 0644,
1114 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1115 .extra1 = &min_percpu_pagelist_fract,
1116 },
1117 #ifdef CONFIG_MMU
1118 {
1119 .procname = "max_map_count",
1120 .data = &sysctl_max_map_count,
1121 .maxlen = sizeof(sysctl_max_map_count),
1122 .mode = 0644,
1123 .proc_handler = proc_dointvec
1124 },
1125 #else
1126 {
1127 .procname = "nr_trim_pages",
1128 .data = &sysctl_nr_trim_pages,
1129 .maxlen = sizeof(sysctl_nr_trim_pages),
1130 .mode = 0644,
1131 .proc_handler = proc_dointvec_minmax,
1132 .extra1 = &zero,
1133 },
1134 #endif
1135 {
1136 .procname = "laptop_mode",
1137 .data = &laptop_mode,
1138 .maxlen = sizeof(laptop_mode),
1139 .mode = 0644,
1140 .proc_handler = proc_dointvec_jiffies,
1141 },
1142 {
1143 .procname = "block_dump",
1144 .data = &block_dump,
1145 .maxlen = sizeof(block_dump),
1146 .mode = 0644,
1147 .proc_handler = proc_dointvec,
1148 .extra1 = &zero,
1149 },
1150 {
1151 .procname = "vfs_cache_pressure",
1152 .data = &sysctl_vfs_cache_pressure,
1153 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1154 .mode = 0644,
1155 .proc_handler = proc_dointvec,
1156 .extra1 = &zero,
1157 },
1158 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1159 {
1160 .procname = "legacy_va_layout",
1161 .data = &sysctl_legacy_va_layout,
1162 .maxlen = sizeof(sysctl_legacy_va_layout),
1163 .mode = 0644,
1164 .proc_handler = proc_dointvec,
1165 .extra1 = &zero,
1166 },
1167 #endif
1168 #ifdef CONFIG_NUMA
1169 {
1170 .procname = "zone_reclaim_mode",
1171 .data = &zone_reclaim_mode,
1172 .maxlen = sizeof(zone_reclaim_mode),
1173 .mode = 0644,
1174 .proc_handler = proc_dointvec,
1175 .extra1 = &zero,
1176 },
1177 {
1178 .procname = "min_unmapped_ratio",
1179 .data = &sysctl_min_unmapped_ratio,
1180 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1181 .mode = 0644,
1182 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1183 .extra1 = &zero,
1184 .extra2 = &one_hundred,
1185 },
1186 {
1187 .procname = "min_slab_ratio",
1188 .data = &sysctl_min_slab_ratio,
1189 .maxlen = sizeof(sysctl_min_slab_ratio),
1190 .mode = 0644,
1191 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1192 .extra1 = &zero,
1193 .extra2 = &one_hundred,
1194 },
1195 #endif
1196 #ifdef CONFIG_SMP
1197 {
1198 .procname = "stat_interval",
1199 .data = &sysctl_stat_interval,
1200 .maxlen = sizeof(sysctl_stat_interval),
1201 .mode = 0644,
1202 .proc_handler = proc_dointvec_jiffies,
1203 },
1204 #endif
1205 {
1206 .procname = "mmap_min_addr",
1207 .data = &dac_mmap_min_addr,
1208 .maxlen = sizeof(unsigned long),
1209 .mode = 0644,
1210 .proc_handler = mmap_min_addr_handler,
1211 },
1212 #ifdef CONFIG_NUMA
1213 {
1214 .procname = "numa_zonelist_order",
1215 .data = &numa_zonelist_order,
1216 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1217 .mode = 0644,
1218 .proc_handler = numa_zonelist_order_handler,
1219 },
1220 #endif
1221 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1222 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1223 {
1224 .procname = "vdso_enabled",
1225 .data = &vdso_enabled,
1226 .maxlen = sizeof(vdso_enabled),
1227 .mode = 0644,
1228 .proc_handler = proc_dointvec,
1229 .extra1 = &zero,
1230 },
1231 #endif
1232 #ifdef CONFIG_HIGHMEM
1233 {
1234 .procname = "highmem_is_dirtyable",
1235 .data = &vm_highmem_is_dirtyable,
1236 .maxlen = sizeof(vm_highmem_is_dirtyable),
1237 .mode = 0644,
1238 .proc_handler = proc_dointvec_minmax,
1239 .extra1 = &zero,
1240 .extra2 = &one,
1241 },
1242 #endif
1243 {
1244 .procname = "scan_unevictable_pages",
1245 .data = &scan_unevictable_pages,
1246 .maxlen = sizeof(scan_unevictable_pages),
1247 .mode = 0644,
1248 .proc_handler = scan_unevictable_handler,
1249 },
1250 #ifdef CONFIG_MEMORY_FAILURE
1251 {
1252 .procname = "memory_failure_early_kill",
1253 .data = &sysctl_memory_failure_early_kill,
1254 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1255 .mode = 0644,
1256 .proc_handler = proc_dointvec_minmax,
1257 .extra1 = &zero,
1258 .extra2 = &one,
1259 },
1260 {
1261 .procname = "memory_failure_recovery",
1262 .data = &sysctl_memory_failure_recovery,
1263 .maxlen = sizeof(sysctl_memory_failure_recovery),
1264 .mode = 0644,
1265 .proc_handler = proc_dointvec_minmax,
1266 .extra1 = &zero,
1267 .extra2 = &one,
1268 },
1269 #endif
1270
1271 /*
1272 * NOTE: do not add new entries to this table unless you have read
1273 * Documentation/sysctl/ctl_unnumbered.txt
1274 */
1275 { }
1276 };
1277
1278 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1279 static struct ctl_table binfmt_misc_table[] = {
1280 { }
1281 };
1282 #endif
1283
1284 static struct ctl_table fs_table[] = {
1285 {
1286 .procname = "inode-nr",
1287 .data = &inodes_stat,
1288 .maxlen = 2*sizeof(int),
1289 .mode = 0444,
1290 .proc_handler = proc_dointvec,
1291 },
1292 {
1293 .procname = "inode-state",
1294 .data = &inodes_stat,
1295 .maxlen = 7*sizeof(int),
1296 .mode = 0444,
1297 .proc_handler = proc_dointvec,
1298 },
1299 {
1300 .procname = "file-nr",
1301 .data = &files_stat,
1302 .maxlen = 3*sizeof(int),
1303 .mode = 0444,
1304 .proc_handler = proc_nr_files,
1305 },
1306 {
1307 .procname = "file-max",
1308 .data = &files_stat.max_files,
1309 .maxlen = sizeof(int),
1310 .mode = 0644,
1311 .proc_handler = proc_dointvec,
1312 },
1313 {
1314 .procname = "nr_open",
1315 .data = &sysctl_nr_open,
1316 .maxlen = sizeof(int),
1317 .mode = 0644,
1318 .proc_handler = proc_dointvec_minmax,
1319 .extra1 = &sysctl_nr_open_min,
1320 .extra2 = &sysctl_nr_open_max,
1321 },
1322 {
1323 .procname = "dentry-state",
1324 .data = &dentry_stat,
1325 .maxlen = 6*sizeof(int),
1326 .mode = 0444,
1327 .proc_handler = proc_dointvec,
1328 },
1329 {
1330 .procname = "overflowuid",
1331 .data = &fs_overflowuid,
1332 .maxlen = sizeof(int),
1333 .mode = 0644,
1334 .proc_handler = proc_dointvec_minmax,
1335 .extra1 = &minolduid,
1336 .extra2 = &maxolduid,
1337 },
1338 {
1339 .procname = "overflowgid",
1340 .data = &fs_overflowgid,
1341 .maxlen = sizeof(int),
1342 .mode = 0644,
1343 .proc_handler = proc_dointvec_minmax,
1344 .extra1 = &minolduid,
1345 .extra2 = &maxolduid,
1346 },
1347 #ifdef CONFIG_FILE_LOCKING
1348 {
1349 .procname = "leases-enable",
1350 .data = &leases_enable,
1351 .maxlen = sizeof(int),
1352 .mode = 0644,
1353 .proc_handler = proc_dointvec,
1354 },
1355 #endif
1356 #ifdef CONFIG_DNOTIFY
1357 {
1358 .procname = "dir-notify-enable",
1359 .data = &dir_notify_enable,
1360 .maxlen = sizeof(int),
1361 .mode = 0644,
1362 .proc_handler = proc_dointvec,
1363 },
1364 #endif
1365 #ifdef CONFIG_MMU
1366 #ifdef CONFIG_FILE_LOCKING
1367 {
1368 .procname = "lease-break-time",
1369 .data = &lease_break_time,
1370 .maxlen = sizeof(int),
1371 .mode = 0644,
1372 .proc_handler = proc_dointvec,
1373 },
1374 #endif
1375 #ifdef CONFIG_AIO
1376 {
1377 .procname = "aio-nr",
1378 .data = &aio_nr,
1379 .maxlen = sizeof(aio_nr),
1380 .mode = 0444,
1381 .proc_handler = proc_doulongvec_minmax,
1382 },
1383 {
1384 .procname = "aio-max-nr",
1385 .data = &aio_max_nr,
1386 .maxlen = sizeof(aio_max_nr),
1387 .mode = 0644,
1388 .proc_handler = proc_doulongvec_minmax,
1389 },
1390 #endif /* CONFIG_AIO */
1391 #ifdef CONFIG_INOTIFY_USER
1392 {
1393 .procname = "inotify",
1394 .mode = 0555,
1395 .child = inotify_table,
1396 },
1397 #endif
1398 #ifdef CONFIG_EPOLL
1399 {
1400 .procname = "epoll",
1401 .mode = 0555,
1402 .child = epoll_table,
1403 },
1404 #endif
1405 #endif
1406 {
1407 .procname = "suid_dumpable",
1408 .data = &suid_dumpable,
1409 .maxlen = sizeof(int),
1410 .mode = 0644,
1411 .proc_handler = proc_dointvec_minmax,
1412 .extra1 = &zero,
1413 .extra2 = &two,
1414 },
1415 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1416 {
1417 .procname = "binfmt_misc",
1418 .mode = 0555,
1419 .child = binfmt_misc_table,
1420 },
1421 #endif
1422 /*
1423 * NOTE: do not add new entries to this table unless you have read
1424 * Documentation/sysctl/ctl_unnumbered.txt
1425 */
1426 { }
1427 };
1428
1429 static struct ctl_table debug_table[] = {
1430 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1431 {
1432 .procname = "exception-trace",
1433 .data = &show_unhandled_signals,
1434 .maxlen = sizeof(int),
1435 .mode = 0644,
1436 .proc_handler = proc_dointvec
1437 },
1438 #endif
1439 { }
1440 };
1441
1442 static struct ctl_table dev_table[] = {
1443 { }
1444 };
1445
1446 static DEFINE_SPINLOCK(sysctl_lock);
1447
1448 /* called under sysctl_lock */
1449 static int use_table(struct ctl_table_header *p)
1450 {
1451 if (unlikely(p->unregistering))
1452 return 0;
1453 p->used++;
1454 return 1;
1455 }
1456
1457 /* called under sysctl_lock */
1458 static void unuse_table(struct ctl_table_header *p)
1459 {
1460 if (!--p->used)
1461 if (unlikely(p->unregistering))
1462 complete(p->unregistering);
1463 }
1464
1465 /* called under sysctl_lock, will reacquire if has to wait */
1466 static void start_unregistering(struct ctl_table_header *p)
1467 {
1468 /*
1469 * if p->used is 0, nobody will ever touch that entry again;
1470 * we'll eliminate all paths to it before dropping sysctl_lock
1471 */
1472 if (unlikely(p->used)) {
1473 struct completion wait;
1474 init_completion(&wait);
1475 p->unregistering = &wait;
1476 spin_unlock(&sysctl_lock);
1477 wait_for_completion(&wait);
1478 spin_lock(&sysctl_lock);
1479 } else {
1480 /* anything non-NULL; we'll never dereference it */
1481 p->unregistering = ERR_PTR(-EINVAL);
1482 }
1483 /*
1484 * do not remove from the list until nobody holds it; walking the
1485 * list in do_sysctl() relies on that.
1486 */
1487 list_del_init(&p->ctl_entry);
1488 }
1489
1490 void sysctl_head_get(struct ctl_table_header *head)
1491 {
1492 spin_lock(&sysctl_lock);
1493 head->count++;
1494 spin_unlock(&sysctl_lock);
1495 }
1496
1497 void sysctl_head_put(struct ctl_table_header *head)
1498 {
1499 spin_lock(&sysctl_lock);
1500 if (!--head->count)
1501 kfree(head);
1502 spin_unlock(&sysctl_lock);
1503 }
1504
1505 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1506 {
1507 if (!head)
1508 BUG();
1509 spin_lock(&sysctl_lock);
1510 if (!use_table(head))
1511 head = ERR_PTR(-ENOENT);
1512 spin_unlock(&sysctl_lock);
1513 return head;
1514 }
1515
1516 void sysctl_head_finish(struct ctl_table_header *head)
1517 {
1518 if (!head)
1519 return;
1520 spin_lock(&sysctl_lock);
1521 unuse_table(head);
1522 spin_unlock(&sysctl_lock);
1523 }
1524
1525 static struct ctl_table_set *
1526 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1527 {
1528 struct ctl_table_set *set = &root->default_set;
1529 if (root->lookup)
1530 set = root->lookup(root, namespaces);
1531 return set;
1532 }
1533
1534 static struct list_head *
1535 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1536 {
1537 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1538 return &set->list;
1539 }
1540
1541 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1542 struct ctl_table_header *prev)
1543 {
1544 struct ctl_table_root *root;
1545 struct list_head *header_list;
1546 struct ctl_table_header *head;
1547 struct list_head *tmp;
1548
1549 spin_lock(&sysctl_lock);
1550 if (prev) {
1551 head = prev;
1552 tmp = &prev->ctl_entry;
1553 unuse_table(prev);
1554 goto next;
1555 }
1556 tmp = &root_table_header.ctl_entry;
1557 for (;;) {
1558 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1559
1560 if (!use_table(head))
1561 goto next;
1562 spin_unlock(&sysctl_lock);
1563 return head;
1564 next:
1565 root = head->root;
1566 tmp = tmp->next;
1567 header_list = lookup_header_list(root, namespaces);
1568 if (tmp != header_list)
1569 continue;
1570
1571 do {
1572 root = list_entry(root->root_list.next,
1573 struct ctl_table_root, root_list);
1574 if (root == &sysctl_table_root)
1575 goto out;
1576 header_list = lookup_header_list(root, namespaces);
1577 } while (list_empty(header_list));
1578 tmp = header_list->next;
1579 }
1580 out:
1581 spin_unlock(&sysctl_lock);
1582 return NULL;
1583 }
1584
1585 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1586 {
1587 return __sysctl_head_next(current->nsproxy, prev);
1588 }
1589
1590 void register_sysctl_root(struct ctl_table_root *root)
1591 {
1592 spin_lock(&sysctl_lock);
1593 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1594 spin_unlock(&sysctl_lock);
1595 }
1596
1597 /*
1598 * sysctl_perm does NOT grant the superuser all rights automatically, because
1599 * some sysctl variables are readonly even to root.
1600 */
1601
1602 static int test_perm(int mode, int op)
1603 {
1604 if (!current_euid())
1605 mode >>= 6;
1606 else if (in_egroup_p(0))
1607 mode >>= 3;
1608 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1609 return 0;
1610 return -EACCES;
1611 }
1612
1613 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1614 {
1615 int error;
1616 int mode;
1617
1618 error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1619 if (error)
1620 return error;
1621
1622 if (root->permissions)
1623 mode = root->permissions(root, current->nsproxy, table);
1624 else
1625 mode = table->mode;
1626
1627 return test_perm(mode, op);
1628 }
1629
1630 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1631 {
1632 for (; table->procname; table++) {
1633 table->parent = parent;
1634 if (table->child)
1635 sysctl_set_parent(table, table->child);
1636 }
1637 }
1638
1639 static __init int sysctl_init(void)
1640 {
1641 sysctl_set_parent(NULL, root_table);
1642 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1643 {
1644 int err;
1645 err = sysctl_check_table(current->nsproxy, root_table);
1646 }
1647 #endif
1648 return 0;
1649 }
1650
1651 core_initcall(sysctl_init);
1652
1653 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1654 struct ctl_table *table)
1655 {
1656 struct ctl_table *p;
1657 const char *s = branch->procname;
1658
1659 /* branch should have named subdirectory as its first element */
1660 if (!s || !branch->child)
1661 return NULL;
1662
1663 /* ... and nothing else */
1664 if (branch[1].procname)
1665 return NULL;
1666
1667 /* table should contain subdirectory with the same name */
1668 for (p = table; p->procname; p++) {
1669 if (!p->child)
1670 continue;
1671 if (p->procname && strcmp(p->procname, s) == 0)
1672 return p;
1673 }
1674 return NULL;
1675 }
1676
1677 /* see if attaching q to p would be an improvement */
1678 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1679 {
1680 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1681 struct ctl_table *next;
1682 int is_better = 0;
1683 int not_in_parent = !p->attached_by;
1684
1685 while ((next = is_branch_in(by, to)) != NULL) {
1686 if (by == q->attached_by)
1687 is_better = 1;
1688 if (to == p->attached_by)
1689 not_in_parent = 1;
1690 by = by->child;
1691 to = next->child;
1692 }
1693
1694 if (is_better && not_in_parent) {
1695 q->attached_by = by;
1696 q->attached_to = to;
1697 q->parent = p;
1698 }
1699 }
1700
1701 /**
1702 * __register_sysctl_paths - register a sysctl hierarchy
1703 * @root: List of sysctl headers to register on
1704 * @namespaces: Data to compute which lists of sysctl entries are visible
1705 * @path: The path to the directory the sysctl table is in.
1706 * @table: the top-level table structure
1707 *
1708 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1709 * array. A completely 0 filled entry terminates the table.
1710 *
1711 * The members of the &struct ctl_table structure are used as follows:
1712 *
1713 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1714 * enter a sysctl file
1715 *
1716 * data - a pointer to data for use by proc_handler
1717 *
1718 * maxlen - the maximum size in bytes of the data
1719 *
1720 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1721 *
1722 * child - a pointer to the child sysctl table if this entry is a directory, or
1723 * %NULL.
1724 *
1725 * proc_handler - the text handler routine (described below)
1726 *
1727 * de - for internal use by the sysctl routines
1728 *
1729 * extra1, extra2 - extra pointers usable by the proc handler routines
1730 *
1731 * Leaf nodes in the sysctl tree will be represented by a single file
1732 * under /proc; non-leaf nodes will be represented by directories.
1733 *
1734 * sysctl(2) can automatically manage read and write requests through
1735 * the sysctl table. The data and maxlen fields of the ctl_table
1736 * struct enable minimal validation of the values being written to be
1737 * performed, and the mode field allows minimal authentication.
1738 *
1739 * There must be a proc_handler routine for any terminal nodes
1740 * mirrored under /proc/sys (non-terminals are handled by a built-in
1741 * directory handler). Several default handlers are available to
1742 * cover common cases -
1743 *
1744 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1745 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1746 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1747 *
1748 * It is the handler's job to read the input buffer from user memory
1749 * and process it. The handler should return 0 on success.
1750 *
1751 * This routine returns %NULL on a failure to register, and a pointer
1752 * to the table header on success.
1753 */
1754 struct ctl_table_header *__register_sysctl_paths(
1755 struct ctl_table_root *root,
1756 struct nsproxy *namespaces,
1757 const struct ctl_path *path, struct ctl_table *table)
1758 {
1759 struct ctl_table_header *header;
1760 struct ctl_table *new, **prevp;
1761 unsigned int n, npath;
1762 struct ctl_table_set *set;
1763
1764 /* Count the path components */
1765 for (npath = 0; path[npath].procname; ++npath)
1766 ;
1767
1768 /*
1769 * For each path component, allocate a 2-element ctl_table array.
1770 * The first array element will be filled with the sysctl entry
1771 * for this, the second will be the sentinel (procname == 0).
1772 *
1773 * We allocate everything in one go so that we don't have to
1774 * worry about freeing additional memory in unregister_sysctl_table.
1775 */
1776 header = kzalloc(sizeof(struct ctl_table_header) +
1777 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1778 if (!header)
1779 return NULL;
1780
1781 new = (struct ctl_table *) (header + 1);
1782
1783 /* Now connect the dots */
1784 prevp = &header->ctl_table;
1785 for (n = 0; n < npath; ++n, ++path) {
1786 /* Copy the procname */
1787 new->procname = path->procname;
1788 new->mode = 0555;
1789
1790 *prevp = new;
1791 prevp = &new->child;
1792
1793 new += 2;
1794 }
1795 *prevp = table;
1796 header->ctl_table_arg = table;
1797
1798 INIT_LIST_HEAD(&header->ctl_entry);
1799 header->used = 0;
1800 header->unregistering = NULL;
1801 header->root = root;
1802 sysctl_set_parent(NULL, header->ctl_table);
1803 header->count = 1;
1804 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1805 if (sysctl_check_table(namespaces, header->ctl_table)) {
1806 kfree(header);
1807 return NULL;
1808 }
1809 #endif
1810 spin_lock(&sysctl_lock);
1811 header->set = lookup_header_set(root, namespaces);
1812 header->attached_by = header->ctl_table;
1813 header->attached_to = root_table;
1814 header->parent = &root_table_header;
1815 for (set = header->set; set; set = set->parent) {
1816 struct ctl_table_header *p;
1817 list_for_each_entry(p, &set->list, ctl_entry) {
1818 if (p->unregistering)
1819 continue;
1820 try_attach(p, header);
1821 }
1822 }
1823 header->parent->count++;
1824 list_add_tail(&header->ctl_entry, &header->set->list);
1825 spin_unlock(&sysctl_lock);
1826
1827 return header;
1828 }
1829
1830 /**
1831 * register_sysctl_table_path - register a sysctl table hierarchy
1832 * @path: The path to the directory the sysctl table is in.
1833 * @table: the top-level table structure
1834 *
1835 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1836 * array. A completely 0 filled entry terminates the table.
1837 *
1838 * See __register_sysctl_paths for more details.
1839 */
1840 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1841 struct ctl_table *table)
1842 {
1843 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1844 path, table);
1845 }
1846
1847 /**
1848 * register_sysctl_table - register a sysctl table hierarchy
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_table(struct ctl_table *table)
1857 {
1858 static const struct ctl_path null_path[] = { {} };
1859
1860 return register_sysctl_paths(null_path, table);
1861 }
1862
1863 /**
1864 * unregister_sysctl_table - unregister a sysctl table hierarchy
1865 * @header: the header returned from register_sysctl_table
1866 *
1867 * Unregisters the sysctl table and all children. proc entries may not
1868 * actually be removed until they are no longer used by anyone.
1869 */
1870 void unregister_sysctl_table(struct ctl_table_header * header)
1871 {
1872 might_sleep();
1873
1874 if (header == NULL)
1875 return;
1876
1877 spin_lock(&sysctl_lock);
1878 start_unregistering(header);
1879 if (!--header->parent->count) {
1880 WARN_ON(1);
1881 kfree(header->parent);
1882 }
1883 if (!--header->count)
1884 kfree(header);
1885 spin_unlock(&sysctl_lock);
1886 }
1887
1888 int sysctl_is_seen(struct ctl_table_header *p)
1889 {
1890 struct ctl_table_set *set = p->set;
1891 int res;
1892 spin_lock(&sysctl_lock);
1893 if (p->unregistering)
1894 res = 0;
1895 else if (!set->is_seen)
1896 res = 1;
1897 else
1898 res = set->is_seen(set);
1899 spin_unlock(&sysctl_lock);
1900 return res;
1901 }
1902
1903 void setup_sysctl_set(struct ctl_table_set *p,
1904 struct ctl_table_set *parent,
1905 int (*is_seen)(struct ctl_table_set *))
1906 {
1907 INIT_LIST_HEAD(&p->list);
1908 p->parent = parent ? parent : &sysctl_table_root.default_set;
1909 p->is_seen = is_seen;
1910 }
1911
1912 #else /* !CONFIG_SYSCTL */
1913 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1914 {
1915 return NULL;
1916 }
1917
1918 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1919 struct ctl_table *table)
1920 {
1921 return NULL;
1922 }
1923
1924 void unregister_sysctl_table(struct ctl_table_header * table)
1925 {
1926 }
1927
1928 void setup_sysctl_set(struct ctl_table_set *p,
1929 struct ctl_table_set *parent,
1930 int (*is_seen)(struct ctl_table_set *))
1931 {
1932 }
1933
1934 void sysctl_head_put(struct ctl_table_header *head)
1935 {
1936 }
1937
1938 #endif /* CONFIG_SYSCTL */
1939
1940 /*
1941 * /proc/sys support
1942 */
1943
1944 #ifdef CONFIG_PROC_SYSCTL
1945
1946 static int _proc_do_string(void* data, int maxlen, int write,
1947 void __user *buffer,
1948 size_t *lenp, loff_t *ppos)
1949 {
1950 size_t len;
1951 char __user *p;
1952 char c;
1953
1954 if (!data || !maxlen || !*lenp) {
1955 *lenp = 0;
1956 return 0;
1957 }
1958
1959 if (write) {
1960 len = 0;
1961 p = buffer;
1962 while (len < *lenp) {
1963 if (get_user(c, p++))
1964 return -EFAULT;
1965 if (c == 0 || c == '\n')
1966 break;
1967 len++;
1968 }
1969 if (len >= maxlen)
1970 len = maxlen-1;
1971 if(copy_from_user(data, buffer, len))
1972 return -EFAULT;
1973 ((char *) data)[len] = 0;
1974 *ppos += *lenp;
1975 } else {
1976 len = strlen(data);
1977 if (len > maxlen)
1978 len = maxlen;
1979
1980 if (*ppos > len) {
1981 *lenp = 0;
1982 return 0;
1983 }
1984
1985 data += *ppos;
1986 len -= *ppos;
1987
1988 if (len > *lenp)
1989 len = *lenp;
1990 if (len)
1991 if(copy_to_user(buffer, data, len))
1992 return -EFAULT;
1993 if (len < *lenp) {
1994 if(put_user('\n', ((char __user *) buffer) + len))
1995 return -EFAULT;
1996 len++;
1997 }
1998 *lenp = len;
1999 *ppos += len;
2000 }
2001 return 0;
2002 }
2003
2004 /**
2005 * proc_dostring - read a string sysctl
2006 * @table: the sysctl table
2007 * @write: %TRUE if this is a write to the sysctl file
2008 * @buffer: the user buffer
2009 * @lenp: the size of the user buffer
2010 * @ppos: file position
2011 *
2012 * Reads/writes a string from/to the user buffer. If the kernel
2013 * buffer provided is not large enough to hold the string, the
2014 * string is truncated. The copied string is %NULL-terminated.
2015 * If the string is being read by the user process, it is copied
2016 * and a newline '\n' is added. It is truncated if the buffer is
2017 * not large enough.
2018 *
2019 * Returns 0 on success.
2020 */
2021 int proc_dostring(struct ctl_table *table, int write,
2022 void __user *buffer, size_t *lenp, loff_t *ppos)
2023 {
2024 return _proc_do_string(table->data, table->maxlen, write,
2025 buffer, lenp, ppos);
2026 }
2027
2028
2029 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2030 int *valp,
2031 int write, void *data)
2032 {
2033 if (write) {
2034 *valp = *negp ? -*lvalp : *lvalp;
2035 } else {
2036 int val = *valp;
2037 if (val < 0) {
2038 *negp = -1;
2039 *lvalp = (unsigned long)-val;
2040 } else {
2041 *negp = 0;
2042 *lvalp = (unsigned long)val;
2043 }
2044 }
2045 return 0;
2046 }
2047
2048 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2049 int write, void __user *buffer,
2050 size_t *lenp, loff_t *ppos,
2051 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2052 int write, void *data),
2053 void *data)
2054 {
2055 #define TMPBUFLEN 21
2056 int *i, vleft, first = 1, neg;
2057 unsigned long lval;
2058 size_t left, len;
2059
2060 char buf[TMPBUFLEN], *p;
2061 char __user *s = buffer;
2062
2063 if (!tbl_data || !table->maxlen || !*lenp ||
2064 (*ppos && !write)) {
2065 *lenp = 0;
2066 return 0;
2067 }
2068
2069 i = (int *) tbl_data;
2070 vleft = table->maxlen / sizeof(*i);
2071 left = *lenp;
2072
2073 if (!conv)
2074 conv = do_proc_dointvec_conv;
2075
2076 for (; left && vleft--; i++, first=0) {
2077 if (write) {
2078 while (left) {
2079 char c;
2080 if (get_user(c, s))
2081 return -EFAULT;
2082 if (!isspace(c))
2083 break;
2084 left--;
2085 s++;
2086 }
2087 if (!left)
2088 break;
2089 neg = 0;
2090 len = left;
2091 if (len > sizeof(buf) - 1)
2092 len = sizeof(buf) - 1;
2093 if (copy_from_user(buf, s, len))
2094 return -EFAULT;
2095 buf[len] = 0;
2096 p = buf;
2097 if (*p == '-' && left > 1) {
2098 neg = 1;
2099 p++;
2100 }
2101 if (*p < '0' || *p > '9')
2102 break;
2103
2104 lval = simple_strtoul(p, &p, 0);
2105
2106 len = p-buf;
2107 if ((len < left) && *p && !isspace(*p))
2108 break;
2109 s += len;
2110 left -= len;
2111
2112 if (conv(&neg, &lval, i, 1, data))
2113 break;
2114 } else {
2115 p = buf;
2116 if (!first)
2117 *p++ = '\t';
2118
2119 if (conv(&neg, &lval, i, 0, data))
2120 break;
2121
2122 sprintf(p, "%s%lu", neg ? "-" : "", lval);
2123 len = strlen(buf);
2124 if (len > left)
2125 len = left;
2126 if(copy_to_user(s, buf, len))
2127 return -EFAULT;
2128 left -= len;
2129 s += len;
2130 }
2131 }
2132
2133 if (!write && !first && left) {
2134 if(put_user('\n', s))
2135 return -EFAULT;
2136 left--, s++;
2137 }
2138 if (write) {
2139 while (left) {
2140 char c;
2141 if (get_user(c, s++))
2142 return -EFAULT;
2143 if (!isspace(c))
2144 break;
2145 left--;
2146 }
2147 }
2148 if (write && first)
2149 return -EINVAL;
2150 *lenp -= left;
2151 *ppos += *lenp;
2152 return 0;
2153 #undef TMPBUFLEN
2154 }
2155
2156 static int do_proc_dointvec(struct ctl_table *table, int write,
2157 void __user *buffer, size_t *lenp, loff_t *ppos,
2158 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2159 int write, void *data),
2160 void *data)
2161 {
2162 return __do_proc_dointvec(table->data, table, write,
2163 buffer, lenp, ppos, conv, data);
2164 }
2165
2166 /**
2167 * proc_dointvec - read a vector of integers
2168 * @table: the sysctl table
2169 * @write: %TRUE if this is a write to the sysctl file
2170 * @buffer: the user buffer
2171 * @lenp: the size of the user buffer
2172 * @ppos: file position
2173 *
2174 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2175 * values from/to the user buffer, treated as an ASCII string.
2176 *
2177 * Returns 0 on success.
2178 */
2179 int proc_dointvec(struct ctl_table *table, int write,
2180 void __user *buffer, size_t *lenp, loff_t *ppos)
2181 {
2182 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2183 NULL,NULL);
2184 }
2185
2186 /*
2187 * Taint values can only be increased
2188 * This means we can safely use a temporary.
2189 */
2190 static int proc_taint(struct ctl_table *table, int write,
2191 void __user *buffer, size_t *lenp, loff_t *ppos)
2192 {
2193 struct ctl_table t;
2194 unsigned long tmptaint = get_taint();
2195 int err;
2196
2197 if (write && !capable(CAP_SYS_ADMIN))
2198 return -EPERM;
2199
2200 t = *table;
2201 t.data = &tmptaint;
2202 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2203 if (err < 0)
2204 return err;
2205
2206 if (write) {
2207 /*
2208 * Poor man's atomic or. Not worth adding a primitive
2209 * to everyone's atomic.h for this
2210 */
2211 int i;
2212 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2213 if ((tmptaint >> i) & 1)
2214 add_taint(i);
2215 }
2216 }
2217
2218 return err;
2219 }
2220
2221 struct do_proc_dointvec_minmax_conv_param {
2222 int *min;
2223 int *max;
2224 };
2225
2226 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2227 int *valp,
2228 int write, void *data)
2229 {
2230 struct do_proc_dointvec_minmax_conv_param *param = data;
2231 if (write) {
2232 int val = *negp ? -*lvalp : *lvalp;
2233 if ((param->min && *param->min > val) ||
2234 (param->max && *param->max < val))
2235 return -EINVAL;
2236 *valp = val;
2237 } else {
2238 int val = *valp;
2239 if (val < 0) {
2240 *negp = -1;
2241 *lvalp = (unsigned long)-val;
2242 } else {
2243 *negp = 0;
2244 *lvalp = (unsigned long)val;
2245 }
2246 }
2247 return 0;
2248 }
2249
2250 /**
2251 * proc_dointvec_minmax - read a vector of integers with min/max values
2252 * @table: the sysctl table
2253 * @write: %TRUE if this is a write to the sysctl file
2254 * @buffer: the user buffer
2255 * @lenp: the size of the user buffer
2256 * @ppos: file position
2257 *
2258 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2259 * values from/to the user buffer, treated as an ASCII string.
2260 *
2261 * This routine will ensure the values are within the range specified by
2262 * table->extra1 (min) and table->extra2 (max).
2263 *
2264 * Returns 0 on success.
2265 */
2266 int proc_dointvec_minmax(struct ctl_table *table, int write,
2267 void __user *buffer, size_t *lenp, loff_t *ppos)
2268 {
2269 struct do_proc_dointvec_minmax_conv_param param = {
2270 .min = (int *) table->extra1,
2271 .max = (int *) table->extra2,
2272 };
2273 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2274 do_proc_dointvec_minmax_conv, &param);
2275 }
2276
2277 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2278 void __user *buffer,
2279 size_t *lenp, loff_t *ppos,
2280 unsigned long convmul,
2281 unsigned long convdiv)
2282 {
2283 #define TMPBUFLEN 21
2284 unsigned long *i, *min, *max, val;
2285 int vleft, first=1, neg;
2286 size_t len, left;
2287 char buf[TMPBUFLEN], *p;
2288 char __user *s = buffer;
2289
2290 if (!data || !table->maxlen || !*lenp ||
2291 (*ppos && !write)) {
2292 *lenp = 0;
2293 return 0;
2294 }
2295
2296 i = (unsigned long *) data;
2297 min = (unsigned long *) table->extra1;
2298 max = (unsigned long *) table->extra2;
2299 vleft = table->maxlen / sizeof(unsigned long);
2300 left = *lenp;
2301
2302 for (; left && vleft--; i++, min++, max++, first=0) {
2303 if (write) {
2304 while (left) {
2305 char c;
2306 if (get_user(c, s))
2307 return -EFAULT;
2308 if (!isspace(c))
2309 break;
2310 left--;
2311 s++;
2312 }
2313 if (!left)
2314 break;
2315 neg = 0;
2316 len = left;
2317 if (len > TMPBUFLEN-1)
2318 len = TMPBUFLEN-1;
2319 if (copy_from_user(buf, s, len))
2320 return -EFAULT;
2321 buf[len] = 0;
2322 p = buf;
2323 if (*p == '-' && left > 1) {
2324 neg = 1;
2325 p++;
2326 }
2327 if (*p < '0' || *p > '9')
2328 break;
2329 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2330 len = p-buf;
2331 if ((len < left) && *p && !isspace(*p))
2332 break;
2333 if (neg)
2334 val = -val;
2335 s += len;
2336 left -= len;
2337
2338 if(neg)
2339 continue;
2340 if ((min && val < *min) || (max && val > *max))
2341 continue;
2342 *i = val;
2343 } else {
2344 p = buf;
2345 if (!first)
2346 *p++ = '\t';
2347 sprintf(p, "%lu", convdiv * (*i) / convmul);
2348 len = strlen(buf);
2349 if (len > left)
2350 len = left;
2351 if(copy_to_user(s, buf, len))
2352 return -EFAULT;
2353 left -= len;
2354 s += len;
2355 }
2356 }
2357
2358 if (!write && !first && left) {
2359 if(put_user('\n', s))
2360 return -EFAULT;
2361 left--, s++;
2362 }
2363 if (write) {
2364 while (left) {
2365 char c;
2366 if (get_user(c, s++))
2367 return -EFAULT;
2368 if (!isspace(c))
2369 break;
2370 left--;
2371 }
2372 }
2373 if (write && first)
2374 return -EINVAL;
2375 *lenp -= left;
2376 *ppos += *lenp;
2377 return 0;
2378 #undef TMPBUFLEN
2379 }
2380
2381 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2382 void __user *buffer,
2383 size_t *lenp, loff_t *ppos,
2384 unsigned long convmul,
2385 unsigned long convdiv)
2386 {
2387 return __do_proc_doulongvec_minmax(table->data, table, write,
2388 buffer, lenp, ppos, convmul, convdiv);
2389 }
2390
2391 /**
2392 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2393 * @table: the sysctl table
2394 * @write: %TRUE if this is a write to the sysctl file
2395 * @buffer: the user buffer
2396 * @lenp: the size of the user buffer
2397 * @ppos: file position
2398 *
2399 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2400 * values from/to the user buffer, treated as an ASCII string.
2401 *
2402 * This routine will ensure the values are within the range specified by
2403 * table->extra1 (min) and table->extra2 (max).
2404 *
2405 * Returns 0 on success.
2406 */
2407 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2408 void __user *buffer, size_t *lenp, loff_t *ppos)
2409 {
2410 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2411 }
2412
2413 /**
2414 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2415 * @table: the sysctl table
2416 * @write: %TRUE if this is a write to the sysctl file
2417 * @buffer: the user buffer
2418 * @lenp: the size of the user buffer
2419 * @ppos: file position
2420 *
2421 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2422 * values from/to the user buffer, treated as an ASCII string. The values
2423 * are treated as milliseconds, and converted to jiffies when they are stored.
2424 *
2425 * This routine will ensure the values are within the range specified by
2426 * table->extra1 (min) and table->extra2 (max).
2427 *
2428 * Returns 0 on success.
2429 */
2430 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2431 void __user *buffer,
2432 size_t *lenp, loff_t *ppos)
2433 {
2434 return do_proc_doulongvec_minmax(table, write, buffer,
2435 lenp, ppos, HZ, 1000l);
2436 }
2437
2438
2439 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2440 int *valp,
2441 int write, void *data)
2442 {
2443 if (write) {
2444 if (*lvalp > LONG_MAX / HZ)
2445 return 1;
2446 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2447 } else {
2448 int val = *valp;
2449 unsigned long lval;
2450 if (val < 0) {
2451 *negp = -1;
2452 lval = (unsigned long)-val;
2453 } else {
2454 *negp = 0;
2455 lval = (unsigned long)val;
2456 }
2457 *lvalp = lval / HZ;
2458 }
2459 return 0;
2460 }
2461
2462 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2463 int *valp,
2464 int write, void *data)
2465 {
2466 if (write) {
2467 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2468 return 1;
2469 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2470 } else {
2471 int val = *valp;
2472 unsigned long lval;
2473 if (val < 0) {
2474 *negp = -1;
2475 lval = (unsigned long)-val;
2476 } else {
2477 *negp = 0;
2478 lval = (unsigned long)val;
2479 }
2480 *lvalp = jiffies_to_clock_t(lval);
2481 }
2482 return 0;
2483 }
2484
2485 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2486 int *valp,
2487 int write, void *data)
2488 {
2489 if (write) {
2490 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2491 } else {
2492 int val = *valp;
2493 unsigned long lval;
2494 if (val < 0) {
2495 *negp = -1;
2496 lval = (unsigned long)-val;
2497 } else {
2498 *negp = 0;
2499 lval = (unsigned long)val;
2500 }
2501 *lvalp = jiffies_to_msecs(lval);
2502 }
2503 return 0;
2504 }
2505
2506 /**
2507 * proc_dointvec_jiffies - read a vector of integers as seconds
2508 * @table: the sysctl table
2509 * @write: %TRUE if this is a write to the sysctl file
2510 * @buffer: the user buffer
2511 * @lenp: the size of the user buffer
2512 * @ppos: file position
2513 *
2514 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2515 * values from/to the user buffer, treated as an ASCII string.
2516 * The values read are assumed to be in seconds, and are converted into
2517 * jiffies.
2518 *
2519 * Returns 0 on success.
2520 */
2521 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2522 void __user *buffer, size_t *lenp, loff_t *ppos)
2523 {
2524 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2525 do_proc_dointvec_jiffies_conv,NULL);
2526 }
2527
2528 /**
2529 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2530 * @table: the sysctl table
2531 * @write: %TRUE if this is a write to the sysctl file
2532 * @buffer: the user buffer
2533 * @lenp: the size of the user buffer
2534 * @ppos: pointer to the file position
2535 *
2536 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2537 * values from/to the user buffer, treated as an ASCII string.
2538 * The values read are assumed to be in 1/USER_HZ seconds, and
2539 * are converted into jiffies.
2540 *
2541 * Returns 0 on success.
2542 */
2543 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2544 void __user *buffer, size_t *lenp, loff_t *ppos)
2545 {
2546 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2547 do_proc_dointvec_userhz_jiffies_conv,NULL);
2548 }
2549
2550 /**
2551 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2552 * @table: the sysctl table
2553 * @write: %TRUE if this is a write to the sysctl file
2554 * @buffer: the user buffer
2555 * @lenp: the size of the user buffer
2556 * @ppos: file position
2557 * @ppos: the current position in the file
2558 *
2559 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2560 * values from/to the user buffer, treated as an ASCII string.
2561 * The values read are assumed to be in 1/1000 seconds, and
2562 * are converted into jiffies.
2563 *
2564 * Returns 0 on success.
2565 */
2566 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2567 void __user *buffer, size_t *lenp, loff_t *ppos)
2568 {
2569 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2570 do_proc_dointvec_ms_jiffies_conv, NULL);
2571 }
2572
2573 static int proc_do_cad_pid(struct ctl_table *table, int write,
2574 void __user *buffer, size_t *lenp, loff_t *ppos)
2575 {
2576 struct pid *new_pid;
2577 pid_t tmp;
2578 int r;
2579
2580 tmp = pid_vnr(cad_pid);
2581
2582 r = __do_proc_dointvec(&tmp, table, write, buffer,
2583 lenp, ppos, NULL, NULL);
2584 if (r || !write)
2585 return r;
2586
2587 new_pid = find_get_pid(tmp);
2588 if (!new_pid)
2589 return -ESRCH;
2590
2591 put_pid(xchg(&cad_pid, new_pid));
2592 return 0;
2593 }
2594
2595 #else /* CONFIG_PROC_FS */
2596
2597 int proc_dostring(struct ctl_table *table, int write,
2598 void __user *buffer, size_t *lenp, loff_t *ppos)
2599 {
2600 return -ENOSYS;
2601 }
2602
2603 int proc_dointvec(struct ctl_table *table, int write,
2604 void __user *buffer, size_t *lenp, loff_t *ppos)
2605 {
2606 return -ENOSYS;
2607 }
2608
2609 int proc_dointvec_minmax(struct ctl_table *table, int write,
2610 void __user *buffer, size_t *lenp, loff_t *ppos)
2611 {
2612 return -ENOSYS;
2613 }
2614
2615 int proc_dointvec_jiffies(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_userhz_jiffies(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_ms_jiffies(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_doulongvec_minmax(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_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2640 void __user *buffer,
2641 size_t *lenp, loff_t *ppos)
2642 {
2643 return -ENOSYS;
2644 }
2645
2646
2647 #endif /* CONFIG_PROC_FS */
2648
2649 /*
2650 * No sense putting this after each symbol definition, twice,
2651 * exception granted :-)
2652 */
2653 EXPORT_SYMBOL(proc_dointvec);
2654 EXPORT_SYMBOL(proc_dointvec_jiffies);
2655 EXPORT_SYMBOL(proc_dointvec_minmax);
2656 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2657 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2658 EXPORT_SYMBOL(proc_dostring);
2659 EXPORT_SYMBOL(proc_doulongvec_minmax);
2660 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2661 EXPORT_SYMBOL(register_sysctl_table);
2662 EXPORT_SYMBOL(register_sysctl_paths);
2663 EXPORT_SYMBOL(unregister_sysctl_table);
Linus' kernel tree