?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu
[linux-2.6/linux-acpi-2.6/ibm-acpi-2.6.git] / kernel / sysctl.c
blob90f536d846439439593af5f85c92bace046c611e
1 /*
2 * sysctl.c: General linux system control interface
3 *
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12 * Horn.
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16 * Wendling.
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19 */
20
21 #include <linux/module.h>
22 #include <linux/mm.h>
23 #include <linux/swap.h>
24 #include <linux/slab.h>
25 #include <linux/sysctl.h>
26 #include <linux/signal.h>
27 #include <linux/proc_fs.h>
28 #include <linux/security.h>
29 #include <linux/ctype.h>
30 #include <linux/kmemcheck.h>
31 #include <linux/fs.h>
32 #include <linux/init.h>
33 #include <linux/kernel.h>
34 #include <linux/kobject.h>
35 #include <linux/net.h>
36 #include <linux/sysrq.h>
37 #include <linux/highuid.h>
38 #include <linux/writeback.h>
39 #include <linux/ratelimit.h>
40 #include <linux/hugetlb.h>
41 #include <linux/initrd.h>
42 #include <linux/key.h>
43 #include <linux/times.h>
44 #include <linux/limits.h>
45 #include <linux/dcache.h>
46 #include <linux/syscalls.h>
47 #include <linux/vmstat.h>
48 #include <linux/nfs_fs.h>
49 #include <linux/acpi.h>
50 #include <linux/reboot.h>
51 #include <linux/ftrace.h>
52 #include <linux/slow-work.h>
53 #include <linux/perf_event.h>
54 #include <linux/kprobes.h>
55
56 #include <asm/uaccess.h>
57 #include <asm/processor.h>
58
59 #ifdef CONFIG_X86
60 #include <asm/nmi.h>
61 #include <asm/stacktrace.h>
62 #include <asm/io.h>
63 #endif
64 #ifdef CONFIG_BSD_PROCESS_ACCT
65 #include <linux/acct.h>
66 #endif
67 #ifdef CONFIG_RT_MUTEXES
68 #include <linux/rtmutex.h>
69 #endif
70 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
71 #include <linux/lockdep.h>
72 #endif
73 #ifdef CONFIG_CHR_DEV_SG
74 #include <scsi/sg.h>
75 #endif
76
77
78 #if defined(CONFIG_SYSCTL)
79
80 /* External variables not in a header file. */
81 extern int sysctl_overcommit_memory;
82 extern int sysctl_overcommit_ratio;
83 extern int sysctl_panic_on_oom;
84 extern int sysctl_oom_kill_allocating_task;
85 extern int sysctl_oom_dump_tasks;
86 extern int max_threads;
87 extern int core_uses_pid;
88 extern int suid_dumpable;
89 extern char core_pattern[];
90 extern unsigned int core_pipe_limit;
91 extern int pid_max;
92 extern int min_free_kbytes;
93 extern int pid_max_min, pid_max_max;
94 extern int sysctl_drop_caches;
95 extern int percpu_pagelist_fraction;
96 extern int compat_log;
97 extern int latencytop_enabled;
98 extern int sysctl_nr_open_min, sysctl_nr_open_max;
99 #ifndef CONFIG_MMU
100 extern int sysctl_nr_trim_pages;
101 #endif
102 #ifdef CONFIG_BLOCK
103 extern int blk_iopoll_enabled;
104 #endif
105
106 /* Constants used for minimum and maximum */
107 #ifdef CONFIG_DETECT_SOFTLOCKUP
108 static int sixty = 60;
109 static int neg_one = -1;
110 #endif
111
112 static int zero;
113 static int __maybe_unused one = 1;
114 static int __maybe_unused two = 2;
115 static unsigned long one_ul = 1;
116 static int one_hundred = 100;
117 #ifdef CONFIG_PRINTK
118 static int ten_thousand = 10000;
119 #endif
120
121 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
122 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
123
124 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
125 static int maxolduid = 65535;
126 static int minolduid;
127 static int min_percpu_pagelist_fract = 8;
128
129 static int ngroups_max = NGROUPS_MAX;
130
131 #ifdef CONFIG_SPARC
132 #include <asm/system.h>
133 #endif
134
135 #ifdef CONFIG_SPARC64
136 extern int sysctl_tsb_ratio;
137 #endif
138
139 #ifdef __hppa__
140 extern int pwrsw_enabled;
141 extern int unaligned_enabled;
142 #endif
143
144 #ifdef CONFIG_S390
145 #ifdef CONFIG_MATHEMU
146 extern int sysctl_ieee_emulation_warnings;
147 #endif
148 extern int sysctl_userprocess_debug;
149 extern int spin_retry;
150 #endif
151
152 #ifdef CONFIG_IA64
153 extern int no_unaligned_warning;
154 extern int unaligned_dump_stack;
155 #endif
156
157 extern struct ratelimit_state printk_ratelimit_state;
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 /* The default sysctl tables: */
198
199 static struct ctl_table root_table[] = {
200 {
201 .procname = "kernel",
202 .mode = 0555,
203 .child = kern_table,
204 },
205 {
206 .procname = "vm",
207 .mode = 0555,
208 .child = vm_table,
209 },
210 {
211 .procname = "fs",
212 .mode = 0555,
213 .child = fs_table,
214 },
215 {
216 .procname = "debug",
217 .mode = 0555,
218 .child = debug_table,
219 },
220 {
221 .procname = "dev",
222 .mode = 0555,
223 .child = dev_table,
224 },
225 /*
226 * NOTE: do not add new entries to this table unless you have read
227 * Documentation/sysctl/ctl_unnumbered.txt
228 */
229 { }
230 };
231
232 #ifdef CONFIG_SCHED_DEBUG
233 static int min_sched_granularity_ns = 100000; /* 100 usecs */
234 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
235 static int min_wakeup_granularity_ns; /* 0 usecs */
236 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
237 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
238 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
239 static int min_sched_shares_ratelimit = 100000; /* 100 usec */
240 static int max_sched_shares_ratelimit = NSEC_PER_SEC; /* 1 second */
241 #endif
242
243 static struct ctl_table kern_table[] = {
244 {
245 .procname = "sched_child_runs_first",
246 .data = &sysctl_sched_child_runs_first,
247 .maxlen = sizeof(unsigned int),
248 .mode = 0644,
249 .proc_handler = proc_dointvec,
250 },
251 #ifdef CONFIG_SCHED_DEBUG
252 {
253 .procname = "sched_min_granularity_ns",
254 .data = &sysctl_sched_min_granularity,
255 .maxlen = sizeof(unsigned int),
256 .mode = 0644,
257 .proc_handler = sched_proc_update_handler,
258 .extra1 = &min_sched_granularity_ns,
259 .extra2 = &max_sched_granularity_ns,
260 },
261 {
262 .procname = "sched_latency_ns",
263 .data = &sysctl_sched_latency,
264 .maxlen = sizeof(unsigned int),
265 .mode = 0644,
266 .proc_handler = sched_proc_update_handler,
267 .extra1 = &min_sched_granularity_ns,
268 .extra2 = &max_sched_granularity_ns,
269 },
270 {
271 .procname = "sched_wakeup_granularity_ns",
272 .data = &sysctl_sched_wakeup_granularity,
273 .maxlen = sizeof(unsigned int),
274 .mode = 0644,
275 .proc_handler = sched_proc_update_handler,
276 .extra1 = &min_wakeup_granularity_ns,
277 .extra2 = &max_wakeup_granularity_ns,
278 },
279 {
280 .procname = "sched_shares_ratelimit",
281 .data = &sysctl_sched_shares_ratelimit,
282 .maxlen = sizeof(unsigned int),
283 .mode = 0644,
284 .proc_handler = sched_proc_update_handler,
285 .extra1 = &min_sched_shares_ratelimit,
286 .extra2 = &max_sched_shares_ratelimit,
287 },
288 {
289 .procname = "sched_tunable_scaling",
290 .data = &sysctl_sched_tunable_scaling,
291 .maxlen = sizeof(enum sched_tunable_scaling),
292 .mode = 0644,
293 .proc_handler = sched_proc_update_handler,
294 .extra1 = &min_sched_tunable_scaling,
295 .extra2 = &max_sched_tunable_scaling,
296 },
297 {
298 .procname = "sched_shares_thresh",
299 .data = &sysctl_sched_shares_thresh,
300 .maxlen = sizeof(unsigned int),
301 .mode = 0644,
302 .proc_handler = proc_dointvec_minmax,
303 .extra1 = &zero,
304 },
305 {
306 .procname = "sched_migration_cost",
307 .data = &sysctl_sched_migration_cost,
308 .maxlen = sizeof(unsigned int),
309 .mode = 0644,
310 .proc_handler = proc_dointvec,
311 },
312 {
313 .procname = "sched_nr_migrate",
314 .data = &sysctl_sched_nr_migrate,
315 .maxlen = sizeof(unsigned int),
316 .mode = 0644,
317 .proc_handler = proc_dointvec,
318 },
319 {
320 .procname = "sched_time_avg",
321 .data = &sysctl_sched_time_avg,
322 .maxlen = sizeof(unsigned int),
323 .mode = 0644,
324 .proc_handler = proc_dointvec,
325 },
326 {
327 .procname = "timer_migration",
328 .data = &sysctl_timer_migration,
329 .maxlen = sizeof(unsigned int),
330 .mode = 0644,
331 .proc_handler = proc_dointvec_minmax,
332 .extra1 = &zero,
333 .extra2 = &one,
334 },
335 #endif
336 {
337 .procname = "sched_rt_period_us",
338 .data = &sysctl_sched_rt_period,
339 .maxlen = sizeof(unsigned int),
340 .mode = 0644,
341 .proc_handler = sched_rt_handler,
342 },
343 {
344 .procname = "sched_rt_runtime_us",
345 .data = &sysctl_sched_rt_runtime,
346 .maxlen = sizeof(int),
347 .mode = 0644,
348 .proc_handler = sched_rt_handler,
349 },
350 {
351 .procname = "sched_compat_yield",
352 .data = &sysctl_sched_compat_yield,
353 .maxlen = sizeof(unsigned int),
354 .mode = 0644,
355 .proc_handler = proc_dointvec,
356 },
357 #ifdef CONFIG_PROVE_LOCKING
358 {
359 .procname = "prove_locking",
360 .data = &prove_locking,
361 .maxlen = sizeof(int),
362 .mode = 0644,
363 .proc_handler = proc_dointvec,
364 },
365 #endif
366 #ifdef CONFIG_LOCK_STAT
367 {
368 .procname = "lock_stat",
369 .data = &lock_stat,
370 .maxlen = sizeof(int),
371 .mode = 0644,
372 .proc_handler = proc_dointvec,
373 },
374 #endif
375 {
376 .procname = "panic",
377 .data = &panic_timeout,
378 .maxlen = sizeof(int),
379 .mode = 0644,
380 .proc_handler = proc_dointvec,
381 },
382 {
383 .procname = "core_uses_pid",
384 .data = &core_uses_pid,
385 .maxlen = sizeof(int),
386 .mode = 0644,
387 .proc_handler = proc_dointvec,
388 },
389 {
390 .procname = "core_pattern",
391 .data = core_pattern,
392 .maxlen = CORENAME_MAX_SIZE,
393 .mode = 0644,
394 .proc_handler = proc_dostring,
395 },
396 {
397 .procname = "core_pipe_limit",
398 .data = &core_pipe_limit,
399 .maxlen = sizeof(unsigned int),
400 .mode = 0644,
401 .proc_handler = proc_dointvec,
402 },
403 #ifdef CONFIG_PROC_SYSCTL
404 {
405 .procname = "tainted",
406 .maxlen = sizeof(long),
407 .mode = 0644,
408 .proc_handler = proc_taint,
409 },
410 #endif
411 #ifdef CONFIG_LATENCYTOP
412 {
413 .procname = "latencytop",
414 .data = &latencytop_enabled,
415 .maxlen = sizeof(int),
416 .mode = 0644,
417 .proc_handler = proc_dointvec,
418 },
419 #endif
420 #ifdef CONFIG_BLK_DEV_INITRD
421 {
422 .procname = "real-root-dev",
423 .data = &real_root_dev,
424 .maxlen = sizeof(int),
425 .mode = 0644,
426 .proc_handler = proc_dointvec,
427 },
428 #endif
429 {
430 .procname = "print-fatal-signals",
431 .data = &print_fatal_signals,
432 .maxlen = sizeof(int),
433 .mode = 0644,
434 .proc_handler = proc_dointvec,
435 },
436 #ifdef CONFIG_SPARC
437 {
438 .procname = "reboot-cmd",
439 .data = reboot_command,
440 .maxlen = 256,
441 .mode = 0644,
442 .proc_handler = proc_dostring,
443 },
444 {
445 .procname = "stop-a",
446 .data = &stop_a_enabled,
447 .maxlen = sizeof (int),
448 .mode = 0644,
449 .proc_handler = proc_dointvec,
450 },
451 {
452 .procname = "scons-poweroff",
453 .data = &scons_pwroff,
454 .maxlen = sizeof (int),
455 .mode = 0644,
456 .proc_handler = proc_dointvec,
457 },
458 #endif
459 #ifdef CONFIG_SPARC64
460 {
461 .procname = "tsb-ratio",
462 .data = &sysctl_tsb_ratio,
463 .maxlen = sizeof (int),
464 .mode = 0644,
465 .proc_handler = proc_dointvec,
466 },
467 #endif
468 #ifdef __hppa__
469 {
470 .procname = "soft-power",
471 .data = &pwrsw_enabled,
472 .maxlen = sizeof (int),
473 .mode = 0644,
474 .proc_handler = proc_dointvec,
475 },
476 {
477 .procname = "unaligned-trap",
478 .data = &unaligned_enabled,
479 .maxlen = sizeof (int),
480 .mode = 0644,
481 .proc_handler = proc_dointvec,
482 },
483 #endif
484 {
485 .procname = "ctrl-alt-del",
486 .data = &C_A_D,
487 .maxlen = sizeof(int),
488 .mode = 0644,
489 .proc_handler = proc_dointvec,
490 },
491 #ifdef CONFIG_FUNCTION_TRACER
492 {
493 .procname = "ftrace_enabled",
494 .data = &ftrace_enabled,
495 .maxlen = sizeof(int),
496 .mode = 0644,
497 .proc_handler = ftrace_enable_sysctl,
498 },
499 #endif
500 #ifdef CONFIG_STACK_TRACER
501 {
502 .procname = "stack_tracer_enabled",
503 .data = &stack_tracer_enabled,
504 .maxlen = sizeof(int),
505 .mode = 0644,
506 .proc_handler = stack_trace_sysctl,
507 },
508 #endif
509 #ifdef CONFIG_TRACING
510 {
511 .procname = "ftrace_dump_on_oops",
512 .data = &ftrace_dump_on_oops,
513 .maxlen = sizeof(int),
514 .mode = 0644,
515 .proc_handler = proc_dointvec,
516 },
517 #endif
518 #ifdef CONFIG_MODULES
519 {
520 .procname = "modprobe",
521 .data = &modprobe_path,
522 .maxlen = KMOD_PATH_LEN,
523 .mode = 0644,
524 .proc_handler = proc_dostring,
525 },
526 {
527 .procname = "modules_disabled",
528 .data = &modules_disabled,
529 .maxlen = sizeof(int),
530 .mode = 0644,
531 /* only handle a transition from default "0" to "1" */
532 .proc_handler = proc_dointvec_minmax,
533 .extra1 = &one,
534 .extra2 = &one,
535 },
536 #endif
537 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
538 {
539 .procname = "hotplug",
540 .data = &uevent_helper,
541 .maxlen = UEVENT_HELPER_PATH_LEN,
542 .mode = 0644,
543 .proc_handler = proc_dostring,
544 },
545 #endif
546 #ifdef CONFIG_CHR_DEV_SG
547 {
548 .procname = "sg-big-buff",
549 .data = &sg_big_buff,
550 .maxlen = sizeof (int),
551 .mode = 0444,
552 .proc_handler = proc_dointvec,
553 },
554 #endif
555 #ifdef CONFIG_BSD_PROCESS_ACCT
556 {
557 .procname = "acct",
558 .data = &acct_parm,
559 .maxlen = 3*sizeof(int),
560 .mode = 0644,
561 .proc_handler = proc_dointvec,
562 },
563 #endif
564 #ifdef CONFIG_MAGIC_SYSRQ
565 {
566 .procname = "sysrq",
567 .data = &__sysrq_enabled,
568 .maxlen = sizeof (int),
569 .mode = 0644,
570 .proc_handler = proc_dointvec,
571 },
572 #endif
573 #ifdef CONFIG_PROC_SYSCTL
574 {
575 .procname = "cad_pid",
576 .data = NULL,
577 .maxlen = sizeof (int),
578 .mode = 0600,
579 .proc_handler = proc_do_cad_pid,
580 },
581 #endif
582 {
583 .procname = "threads-max",
584 .data = &max_threads,
585 .maxlen = sizeof(int),
586 .mode = 0644,
587 .proc_handler = proc_dointvec,
588 },
589 {
590 .procname = "random",
591 .mode = 0555,
592 .child = random_table,
593 },
594 {
595 .procname = "overflowuid",
596 .data = &overflowuid,
597 .maxlen = sizeof(int),
598 .mode = 0644,
599 .proc_handler = proc_dointvec_minmax,
600 .extra1 = &minolduid,
601 .extra2 = &maxolduid,
602 },
603 {
604 .procname = "overflowgid",
605 .data = &overflowgid,
606 .maxlen = sizeof(int),
607 .mode = 0644,
608 .proc_handler = proc_dointvec_minmax,
609 .extra1 = &minolduid,
610 .extra2 = &maxolduid,
611 },
612 #ifdef CONFIG_S390
613 #ifdef CONFIG_MATHEMU
614 {
615 .procname = "ieee_emulation_warnings",
616 .data = &sysctl_ieee_emulation_warnings,
617 .maxlen = sizeof(int),
618 .mode = 0644,
619 .proc_handler = proc_dointvec,
620 },
621 #endif
622 {
623 .procname = "userprocess_debug",
624 .data = &show_unhandled_signals,
625 .maxlen = sizeof(int),
626 .mode = 0644,
627 .proc_handler = proc_dointvec,
628 },
629 #endif
630 {
631 .procname = "pid_max",
632 .data = &pid_max,
633 .maxlen = sizeof (int),
634 .mode = 0644,
635 .proc_handler = proc_dointvec_minmax,
636 .extra1 = &pid_max_min,
637 .extra2 = &pid_max_max,
638 },
639 {
640 .procname = "panic_on_oops",
641 .data = &panic_on_oops,
642 .maxlen = sizeof(int),
643 .mode = 0644,
644 .proc_handler = proc_dointvec,
645 },
646 #if defined CONFIG_PRINTK
647 {
648 .procname = "printk",
649 .data = &console_loglevel,
650 .maxlen = 4*sizeof(int),
651 .mode = 0644,
652 .proc_handler = proc_dointvec,
653 },
654 {
655 .procname = "printk_ratelimit",
656 .data = &printk_ratelimit_state.interval,
657 .maxlen = sizeof(int),
658 .mode = 0644,
659 .proc_handler = proc_dointvec_jiffies,
660 },
661 {
662 .procname = "printk_ratelimit_burst",
663 .data = &printk_ratelimit_state.burst,
664 .maxlen = sizeof(int),
665 .mode = 0644,
666 .proc_handler = proc_dointvec,
667 },
668 {
669 .procname = "printk_delay",
670 .data = &printk_delay_msec,
671 .maxlen = sizeof(int),
672 .mode = 0644,
673 .proc_handler = proc_dointvec_minmax,
674 .extra1 = &zero,
675 .extra2 = &ten_thousand,
676 },
677 #endif
678 {
679 .procname = "ngroups_max",
680 .data = &ngroups_max,
681 .maxlen = sizeof (int),
682 .mode = 0444,
683 .proc_handler = proc_dointvec,
684 },
685 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
686 {
687 .procname = "unknown_nmi_panic",
688 .data = &unknown_nmi_panic,
689 .maxlen = sizeof (int),
690 .mode = 0644,
691 .proc_handler = proc_dointvec,
692 },
693 {
694 .procname = "nmi_watchdog",
695 .data = &nmi_watchdog_enabled,
696 .maxlen = sizeof (int),
697 .mode = 0644,
698 .proc_handler = proc_nmi_enabled,
699 },
700 #endif
701 #if defined(CONFIG_X86)
702 {
703 .procname = "panic_on_unrecovered_nmi",
704 .data = &panic_on_unrecovered_nmi,
705 .maxlen = sizeof(int),
706 .mode = 0644,
707 .proc_handler = proc_dointvec,
708 },
709 {
710 .procname = "panic_on_io_nmi",
711 .data = &panic_on_io_nmi,
712 .maxlen = sizeof(int),
713 .mode = 0644,
714 .proc_handler = proc_dointvec,
715 },
716 {
717 .procname = "bootloader_type",
718 .data = &bootloader_type,
719 .maxlen = sizeof (int),
720 .mode = 0444,
721 .proc_handler = proc_dointvec,
722 },
723 {
724 .procname = "bootloader_version",
725 .data = &bootloader_version,
726 .maxlen = sizeof (int),
727 .mode = 0444,
728 .proc_handler = proc_dointvec,
729 },
730 {
731 .procname = "kstack_depth_to_print",
732 .data = &kstack_depth_to_print,
733 .maxlen = sizeof(int),
734 .mode = 0644,
735 .proc_handler = proc_dointvec,
736 },
737 {
738 .procname = "io_delay_type",
739 .data = &io_delay_type,
740 .maxlen = sizeof(int),
741 .mode = 0644,
742 .proc_handler = proc_dointvec,
743 },
744 #endif
745 #if defined(CONFIG_MMU)
746 {
747 .procname = "randomize_va_space",
748 .data = &randomize_va_space,
749 .maxlen = sizeof(int),
750 .mode = 0644,
751 .proc_handler = proc_dointvec,
752 },
753 #endif
754 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
755 {
756 .procname = "spin_retry",
757 .data = &spin_retry,
758 .maxlen = sizeof (int),
759 .mode = 0644,
760 .proc_handler = proc_dointvec,
761 },
762 #endif
763 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
764 {
765 .procname = "acpi_video_flags",
766 .data = &acpi_realmode_flags,
767 .maxlen = sizeof (unsigned long),
768 .mode = 0644,
769 .proc_handler = proc_doulongvec_minmax,
770 },
771 #endif
772 #ifdef CONFIG_IA64
773 {
774 .procname = "ignore-unaligned-usertrap",
775 .data = &no_unaligned_warning,
776 .maxlen = sizeof (int),
777 .mode = 0644,
778 .proc_handler = proc_dointvec,
779 },
780 {
781 .procname = "unaligned-dump-stack",
782 .data = &unaligned_dump_stack,
783 .maxlen = sizeof (int),
784 .mode = 0644,
785 .proc_handler = proc_dointvec,
786 },
787 #endif
788 #ifdef CONFIG_DETECT_SOFTLOCKUP
789 {
790 .procname = "softlockup_panic",
791 .data = &softlockup_panic,
792 .maxlen = sizeof(int),
793 .mode = 0644,
794 .proc_handler = proc_dointvec_minmax,
795 .extra1 = &zero,
796 .extra2 = &one,
797 },
798 {
799 .procname = "softlockup_thresh",
800 .data = &softlockup_thresh,
801 .maxlen = sizeof(int),
802 .mode = 0644,
803 .proc_handler = proc_dosoftlockup_thresh,
804 .extra1 = &neg_one,
805 .extra2 = &sixty,
806 },
807 #endif
808 #ifdef CONFIG_DETECT_HUNG_TASK
809 {
810 .procname = "hung_task_panic",
811 .data = &sysctl_hung_task_panic,
812 .maxlen = sizeof(int),
813 .mode = 0644,
814 .proc_handler = proc_dointvec_minmax,
815 .extra1 = &zero,
816 .extra2 = &one,
817 },
818 {
819 .procname = "hung_task_check_count",
820 .data = &sysctl_hung_task_check_count,
821 .maxlen = sizeof(unsigned long),
822 .mode = 0644,
823 .proc_handler = proc_doulongvec_minmax,
824 },
825 {
826 .procname = "hung_task_timeout_secs",
827 .data = &sysctl_hung_task_timeout_secs,
828 .maxlen = sizeof(unsigned long),
829 .mode = 0644,
830 .proc_handler = proc_dohung_task_timeout_secs,
831 },
832 {
833 .procname = "hung_task_warnings",
834 .data = &sysctl_hung_task_warnings,
835 .maxlen = sizeof(unsigned long),
836 .mode = 0644,
837 .proc_handler = proc_doulongvec_minmax,
838 },
839 #endif
840 #ifdef CONFIG_COMPAT
841 {
842 .procname = "compat-log",
843 .data = &compat_log,
844 .maxlen = sizeof (int),
845 .mode = 0644,
846 .proc_handler = proc_dointvec,
847 },
848 #endif
849 #ifdef CONFIG_RT_MUTEXES
850 {
851 .procname = "max_lock_depth",
852 .data = &max_lock_depth,
853 .maxlen = sizeof(int),
854 .mode = 0644,
855 .proc_handler = proc_dointvec,
856 },
857 #endif
858 {
859 .procname = "poweroff_cmd",
860 .data = &poweroff_cmd,
861 .maxlen = POWEROFF_CMD_PATH_LEN,
862 .mode = 0644,
863 .proc_handler = proc_dostring,
864 },
865 #ifdef CONFIG_KEYS
866 {
867 .procname = "keys",
868 .mode = 0555,
869 .child = key_sysctls,
870 },
871 #endif
872 #ifdef CONFIG_RCU_TORTURE_TEST
873 {
874 .procname = "rcutorture_runnable",
875 .data = &rcutorture_runnable,
876 .maxlen = sizeof(int),
877 .mode = 0644,
878 .proc_handler = proc_dointvec,
879 },
880 #endif
881 #ifdef CONFIG_SLOW_WORK
882 {
883 .procname = "slow-work",
884 .mode = 0555,
885 .child = slow_work_sysctls,
886 },
887 #endif
888 #ifdef CONFIG_PERF_EVENTS
889 {
890 .procname = "perf_event_paranoid",
891 .data = &sysctl_perf_event_paranoid,
892 .maxlen = sizeof(sysctl_perf_event_paranoid),
893 .mode = 0644,
894 .proc_handler = proc_dointvec,
895 },
896 {
897 .procname = "perf_event_mlock_kb",
898 .data = &sysctl_perf_event_mlock,
899 .maxlen = sizeof(sysctl_perf_event_mlock),
900 .mode = 0644,
901 .proc_handler = proc_dointvec,
902 },
903 {
904 .procname = "perf_event_max_sample_rate",
905 .data = &sysctl_perf_event_sample_rate,
906 .maxlen = sizeof(sysctl_perf_event_sample_rate),
907 .mode = 0644,
908 .proc_handler = proc_dointvec,
909 },
910 #endif
911 #ifdef CONFIG_KMEMCHECK
912 {
913 .procname = "kmemcheck",
914 .data = &kmemcheck_enabled,
915 .maxlen = sizeof(int),
916 .mode = 0644,
917 .proc_handler = proc_dointvec,
918 },
919 #endif
920 #ifdef CONFIG_BLOCK
921 {
922 .procname = "blk_iopoll",
923 .data = &blk_iopoll_enabled,
924 .maxlen = sizeof(int),
925 .mode = 0644,
926 .proc_handler = proc_dointvec,
927 },
928 #endif
929 /*
930 * NOTE: do not add new entries to this table unless you have read
931 * Documentation/sysctl/ctl_unnumbered.txt
932 */
933 { }
934 };
935
936 static struct ctl_table vm_table[] = {
937 {
938 .procname = "overcommit_memory",
939 .data = &sysctl_overcommit_memory,
940 .maxlen = sizeof(sysctl_overcommit_memory),
941 .mode = 0644,
942 .proc_handler = proc_dointvec,
943 },
944 {
945 .procname = "panic_on_oom",
946 .data = &sysctl_panic_on_oom,
947 .maxlen = sizeof(sysctl_panic_on_oom),
948 .mode = 0644,
949 .proc_handler = proc_dointvec,
950 },
951 {
952 .procname = "oom_kill_allocating_task",
953 .data = &sysctl_oom_kill_allocating_task,
954 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
955 .mode = 0644,
956 .proc_handler = proc_dointvec,
957 },
958 {
959 .procname = "oom_dump_tasks",
960 .data = &sysctl_oom_dump_tasks,
961 .maxlen = sizeof(sysctl_oom_dump_tasks),
962 .mode = 0644,
963 .proc_handler = proc_dointvec,
964 },
965 {
966 .procname = "overcommit_ratio",
967 .data = &sysctl_overcommit_ratio,
968 .maxlen = sizeof(sysctl_overcommit_ratio),
969 .mode = 0644,
970 .proc_handler = proc_dointvec,
971 },
972 {
973 .procname = "page-cluster",
974 .data = &page_cluster,
975 .maxlen = sizeof(int),
976 .mode = 0644,
977 .proc_handler = proc_dointvec,
978 },
979 {
980 .procname = "dirty_background_ratio",
981 .data = &dirty_background_ratio,
982 .maxlen = sizeof(dirty_background_ratio),
983 .mode = 0644,
984 .proc_handler = dirty_background_ratio_handler,
985 .extra1 = &zero,
986 .extra2 = &one_hundred,
987 },
988 {
989 .procname = "dirty_background_bytes",
990 .data = &dirty_background_bytes,
991 .maxlen = sizeof(dirty_background_bytes),
992 .mode = 0644,
993 .proc_handler = dirty_background_bytes_handler,
994 .extra1 = &one_ul,
995 },
996 {
997 .procname = "dirty_ratio",
998 .data = &vm_dirty_ratio,
999 .maxlen = sizeof(vm_dirty_ratio),
1000 .mode = 0644,
1001 .proc_handler = dirty_ratio_handler,
1002 .extra1 = &zero,
1003 .extra2 = &one_hundred,
1004 },
1005 {
1006 .procname = "dirty_bytes",
1007 .data = &vm_dirty_bytes,
1008 .maxlen = sizeof(vm_dirty_bytes),
1009 .mode = 0644,
1010 .proc_handler = dirty_bytes_handler,
1011 .extra1 = &dirty_bytes_min,
1012 },
1013 {
1014 .procname = "dirty_writeback_centisecs",
1015 .data = &dirty_writeback_interval,
1016 .maxlen = sizeof(dirty_writeback_interval),
1017 .mode = 0644,
1018 .proc_handler = dirty_writeback_centisecs_handler,
1019 },
1020 {
1021 .procname = "dirty_expire_centisecs",
1022 .data = &dirty_expire_interval,
1023 .maxlen = sizeof(dirty_expire_interval),
1024 .mode = 0644,
1025 .proc_handler = proc_dointvec,
1026 },
1027 {
1028 .procname = "nr_pdflush_threads",
1029 .data = &nr_pdflush_threads,
1030 .maxlen = sizeof nr_pdflush_threads,
1031 .mode = 0444 /* read-only*/,
1032 .proc_handler = proc_dointvec,
1033 },
1034 {
1035 .procname = "swappiness",
1036 .data = &vm_swappiness,
1037 .maxlen = sizeof(vm_swappiness),
1038 .mode = 0644,
1039 .proc_handler = proc_dointvec_minmax,
1040 .extra1 = &zero,
1041 .extra2 = &one_hundred,
1042 },
1043 #ifdef CONFIG_HUGETLB_PAGE
1044 {
1045 .procname = "nr_hugepages",
1046 .data = NULL,
1047 .maxlen = sizeof(unsigned long),
1048 .mode = 0644,
1049 .proc_handler = hugetlb_sysctl_handler,
1050 .extra1 = (void *)&hugetlb_zero,
1051 .extra2 = (void *)&hugetlb_infinity,
1052 },
1053 #ifdef CONFIG_NUMA
1054 {
1055 .procname = "nr_hugepages_mempolicy",
1056 .data = NULL,
1057 .maxlen = sizeof(unsigned long),
1058 .mode = 0644,
1059 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1060 .extra1 = (void *)&hugetlb_zero,
1061 .extra2 = (void *)&hugetlb_infinity,
1062 },
1063 #endif
1064 {
1065 .procname = "hugetlb_shm_group",
1066 .data = &sysctl_hugetlb_shm_group,
1067 .maxlen = sizeof(gid_t),
1068 .mode = 0644,
1069 .proc_handler = proc_dointvec,
1070 },
1071 {
1072 .procname = "hugepages_treat_as_movable",
1073 .data = &hugepages_treat_as_movable,
1074 .maxlen = sizeof(int),
1075 .mode = 0644,
1076 .proc_handler = hugetlb_treat_movable_handler,
1077 },
1078 {
1079 .procname = "nr_overcommit_hugepages",
1080 .data = NULL,
1081 .maxlen = sizeof(unsigned long),
1082 .mode = 0644,
1083 .proc_handler = hugetlb_overcommit_handler,
1084 .extra1 = (void *)&hugetlb_zero,
1085 .extra2 = (void *)&hugetlb_infinity,
1086 },
1087 #endif
1088 {
1089 .procname = "lowmem_reserve_ratio",
1090 .data = &sysctl_lowmem_reserve_ratio,
1091 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1092 .mode = 0644,
1093 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1094 },
1095 {
1096 .procname = "drop_caches",
1097 .data = &sysctl_drop_caches,
1098 .maxlen = sizeof(int),
1099 .mode = 0644,
1100 .proc_handler = drop_caches_sysctl_handler,
1101 },
1102 {
1103 .procname = "min_free_kbytes",
1104 .data = &min_free_kbytes,
1105 .maxlen = sizeof(min_free_kbytes),
1106 .mode = 0644,
1107 .proc_handler = min_free_kbytes_sysctl_handler,
1108 .extra1 = &zero,
1109 },
1110 {
1111 .procname = "percpu_pagelist_fraction",
1112 .data = &percpu_pagelist_fraction,
1113 .maxlen = sizeof(percpu_pagelist_fraction),
1114 .mode = 0644,
1115 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1116 .extra1 = &min_percpu_pagelist_fract,
1117 },
1118 #ifdef CONFIG_MMU
1119 {
1120 .procname = "max_map_count",
1121 .data = &sysctl_max_map_count,
1122 .maxlen = sizeof(sysctl_max_map_count),
1123 .mode = 0644,
1124 .proc_handler = proc_dointvec_minmax,
1125 .extra1 = &zero,
1126 },
1127 #else
1128 {
1129 .procname = "nr_trim_pages",
1130 .data = &sysctl_nr_trim_pages,
1131 .maxlen = sizeof(sysctl_nr_trim_pages),
1132 .mode = 0644,
1133 .proc_handler = proc_dointvec_minmax,
1134 .extra1 = &zero,
1135 },
1136 #endif
1137 {
1138 .procname = "laptop_mode",
1139 .data = &laptop_mode,
1140 .maxlen = sizeof(laptop_mode),
1141 .mode = 0644,
1142 .proc_handler = proc_dointvec_jiffies,
1143 },
1144 {
1145 .procname = "block_dump",
1146 .data = &block_dump,
1147 .maxlen = sizeof(block_dump),
1148 .mode = 0644,
1149 .proc_handler = proc_dointvec,
1150 .extra1 = &zero,
1151 },
1152 {
1153 .procname = "vfs_cache_pressure",
1154 .data = &sysctl_vfs_cache_pressure,
1155 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1156 .mode = 0644,
1157 .proc_handler = proc_dointvec,
1158 .extra1 = &zero,
1159 },
1160 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1161 {
1162 .procname = "legacy_va_layout",
1163 .data = &sysctl_legacy_va_layout,
1164 .maxlen = sizeof(sysctl_legacy_va_layout),
1165 .mode = 0644,
1166 .proc_handler = proc_dointvec,
1167 .extra1 = &zero,
1168 },
1169 #endif
1170 #ifdef CONFIG_NUMA
1171 {
1172 .procname = "zone_reclaim_mode",
1173 .data = &zone_reclaim_mode,
1174 .maxlen = sizeof(zone_reclaim_mode),
1175 .mode = 0644,
1176 .proc_handler = proc_dointvec,
1177 .extra1 = &zero,
1178 },
1179 {
1180 .procname = "min_unmapped_ratio",
1181 .data = &sysctl_min_unmapped_ratio,
1182 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1183 .mode = 0644,
1184 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1185 .extra1 = &zero,
1186 .extra2 = &one_hundred,
1187 },
1188 {
1189 .procname = "min_slab_ratio",
1190 .data = &sysctl_min_slab_ratio,
1191 .maxlen = sizeof(sysctl_min_slab_ratio),
1192 .mode = 0644,
1193 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1194 .extra1 = &zero,
1195 .extra2 = &one_hundred,
1196 },
1197 #endif
1198 #ifdef CONFIG_SMP
1199 {
1200 .procname = "stat_interval",
1201 .data = &sysctl_stat_interval,
1202 .maxlen = sizeof(sysctl_stat_interval),
1203 .mode = 0644,
1204 .proc_handler = proc_dointvec_jiffies,
1205 },
1206 #endif
1207 #ifdef CONFIG_MMU
1208 {
1209 .procname = "mmap_min_addr",
1210 .data = &dac_mmap_min_addr,
1211 .maxlen = sizeof(unsigned long),
1212 .mode = 0644,
1213 .proc_handler = mmap_min_addr_handler,
1214 },
1215 #endif
1216 #ifdef CONFIG_NUMA
1217 {
1218 .procname = "numa_zonelist_order",
1219 .data = &numa_zonelist_order,
1220 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1221 .mode = 0644,
1222 .proc_handler = numa_zonelist_order_handler,
1223 },
1224 #endif
1225 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1226 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1227 {
1228 .procname = "vdso_enabled",
1229 .data = &vdso_enabled,
1230 .maxlen = sizeof(vdso_enabled),
1231 .mode = 0644,
1232 .proc_handler = proc_dointvec,
1233 .extra1 = &zero,
1234 },
1235 #endif
1236 #ifdef CONFIG_HIGHMEM
1237 {
1238 .procname = "highmem_is_dirtyable",
1239 .data = &vm_highmem_is_dirtyable,
1240 .maxlen = sizeof(vm_highmem_is_dirtyable),
1241 .mode = 0644,
1242 .proc_handler = proc_dointvec_minmax,
1243 .extra1 = &zero,
1244 .extra2 = &one,
1245 },
1246 #endif
1247 {
1248 .procname = "scan_unevictable_pages",
1249 .data = &scan_unevictable_pages,
1250 .maxlen = sizeof(scan_unevictable_pages),
1251 .mode = 0644,
1252 .proc_handler = scan_unevictable_handler,
1253 },
1254 #ifdef CONFIG_MEMORY_FAILURE
1255 {
1256 .procname = "memory_failure_early_kill",
1257 .data = &sysctl_memory_failure_early_kill,
1258 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1259 .mode = 0644,
1260 .proc_handler = proc_dointvec_minmax,
1261 .extra1 = &zero,
1262 .extra2 = &one,
1263 },
1264 {
1265 .procname = "memory_failure_recovery",
1266 .data = &sysctl_memory_failure_recovery,
1267 .maxlen = sizeof(sysctl_memory_failure_recovery),
1268 .mode = 0644,
1269 .proc_handler = proc_dointvec_minmax,
1270 .extra1 = &zero,
1271 .extra2 = &one,
1272 },
1273 #endif
1274
1275 /*
1276 * NOTE: do not add new entries to this table unless you have read
1277 * Documentation/sysctl/ctl_unnumbered.txt
1278 */
1279 { }
1280 };
1281
1282 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1283 static struct ctl_table binfmt_misc_table[] = {
1284 { }
1285 };
1286 #endif
1287
1288 static struct ctl_table fs_table[] = {
1289 {
1290 .procname = "inode-nr",
1291 .data = &inodes_stat,
1292 .maxlen = 2*sizeof(int),
1293 .mode = 0444,
1294 .proc_handler = proc_dointvec,
1295 },
1296 {
1297 .procname = "inode-state",
1298 .data = &inodes_stat,
1299 .maxlen = 7*sizeof(int),
1300 .mode = 0444,
1301 .proc_handler = proc_dointvec,
1302 },
1303 {
1304 .procname = "file-nr",
1305 .data = &files_stat,
1306 .maxlen = 3*sizeof(int),
1307 .mode = 0444,
1308 .proc_handler = proc_nr_files,
1309 },
1310 {
1311 .procname = "file-max",
1312 .data = &files_stat.max_files,
1313 .maxlen = sizeof(int),
1314 .mode = 0644,
1315 .proc_handler = proc_dointvec,
1316 },
1317 {
1318 .procname = "nr_open",
1319 .data = &sysctl_nr_open,
1320 .maxlen = sizeof(int),
1321 .mode = 0644,
1322 .proc_handler = proc_dointvec_minmax,
1323 .extra1 = &sysctl_nr_open_min,
1324 .extra2 = &sysctl_nr_open_max,
1325 },
1326 {
1327 .procname = "dentry-state",
1328 .data = &dentry_stat,
1329 .maxlen = 6*sizeof(int),
1330 .mode = 0444,
1331 .proc_handler = proc_dointvec,
1332 },
1333 {
1334 .procname = "overflowuid",
1335 .data = &fs_overflowuid,
1336 .maxlen = sizeof(int),
1337 .mode = 0644,
1338 .proc_handler = proc_dointvec_minmax,
1339 .extra1 = &minolduid,
1340 .extra2 = &maxolduid,
1341 },
1342 {
1343 .procname = "overflowgid",
1344 .data = &fs_overflowgid,
1345 .maxlen = sizeof(int),
1346 .mode = 0644,
1347 .proc_handler = proc_dointvec_minmax,
1348 .extra1 = &minolduid,
1349 .extra2 = &maxolduid,
1350 },
1351 #ifdef CONFIG_FILE_LOCKING
1352 {
1353 .procname = "leases-enable",
1354 .data = &leases_enable,
1355 .maxlen = sizeof(int),
1356 .mode = 0644,
1357 .proc_handler = proc_dointvec,
1358 },
1359 #endif
1360 #ifdef CONFIG_DNOTIFY
1361 {
1362 .procname = "dir-notify-enable",
1363 .data = &dir_notify_enable,
1364 .maxlen = sizeof(int),
1365 .mode = 0644,
1366 .proc_handler = proc_dointvec,
1367 },
1368 #endif
1369 #ifdef CONFIG_MMU
1370 #ifdef CONFIG_FILE_LOCKING
1371 {
1372 .procname = "lease-break-time",
1373 .data = &lease_break_time,
1374 .maxlen = sizeof(int),
1375 .mode = 0644,
1376 .proc_handler = proc_dointvec,
1377 },
1378 #endif
1379 #ifdef CONFIG_AIO
1380 {
1381 .procname = "aio-nr",
1382 .data = &aio_nr,
1383 .maxlen = sizeof(aio_nr),
1384 .mode = 0444,
1385 .proc_handler = proc_doulongvec_minmax,
1386 },
1387 {
1388 .procname = "aio-max-nr",
1389 .data = &aio_max_nr,
1390 .maxlen = sizeof(aio_max_nr),
1391 .mode = 0644,
1392 .proc_handler = proc_doulongvec_minmax,
1393 },
1394 #endif /* CONFIG_AIO */
1395 #ifdef CONFIG_INOTIFY_USER
1396 {
1397 .procname = "inotify",
1398 .mode = 0555,
1399 .child = inotify_table,
1400 },
1401 #endif
1402 #ifdef CONFIG_EPOLL
1403 {
1404 .procname = "epoll",
1405 .mode = 0555,
1406 .child = epoll_table,
1407 },
1408 #endif
1409 #endif
1410 {
1411 .procname = "suid_dumpable",
1412 .data = &suid_dumpable,
1413 .maxlen = sizeof(int),
1414 .mode = 0644,
1415 .proc_handler = proc_dointvec_minmax,
1416 .extra1 = &zero,
1417 .extra2 = &two,
1418 },
1419 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1420 {
1421 .procname = "binfmt_misc",
1422 .mode = 0555,
1423 .child = binfmt_misc_table,
1424 },
1425 #endif
1426 /*
1427 * NOTE: do not add new entries to this table unless you have read
1428 * Documentation/sysctl/ctl_unnumbered.txt
1429 */
1430 { }
1431 };
1432
1433 static struct ctl_table debug_table[] = {
1434 #if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
1435 defined(CONFIG_S390)
1436 {
1437 .procname = "exception-trace",
1438 .data = &show_unhandled_signals,
1439 .maxlen = sizeof(int),
1440 .mode = 0644,
1441 .proc_handler = proc_dointvec
1442 },
1443 #endif
1444 #if defined(CONFIG_OPTPROBES)
1445 {
1446 .procname = "kprobes-optimization",
1447 .data = &sysctl_kprobes_optimization,
1448 .maxlen = sizeof(int),
1449 .mode = 0644,
1450 .proc_handler = proc_kprobes_optimization_handler,
1451 .extra1 = &zero,
1452 .extra2 = &one,
1453 },
1454 #endif
1455 { }
1456 };
1457
1458 static struct ctl_table dev_table[] = {
1459 { }
1460 };
1461
1462 static DEFINE_SPINLOCK(sysctl_lock);
1463
1464 /* called under sysctl_lock */
1465 static int use_table(struct ctl_table_header *p)
1466 {
1467 if (unlikely(p->unregistering))
1468 return 0;
1469 p->used++;
1470 return 1;
1471 }
1472
1473 /* called under sysctl_lock */
1474 static void unuse_table(struct ctl_table_header *p)
1475 {
1476 if (!--p->used)
1477 if (unlikely(p->unregistering))
1478 complete(p->unregistering);
1479 }
1480
1481 /* called under sysctl_lock, will reacquire if has to wait */
1482 static void start_unregistering(struct ctl_table_header *p)
1483 {
1484 /*
1485 * if p->used is 0, nobody will ever touch that entry again;
1486 * we'll eliminate all paths to it before dropping sysctl_lock
1487 */
1488 if (unlikely(p->used)) {
1489 struct completion wait;
1490 init_completion(&wait);
1491 p->unregistering = &wait;
1492 spin_unlock(&sysctl_lock);
1493 wait_for_completion(&wait);
1494 spin_lock(&sysctl_lock);
1495 } else {
1496 /* anything non-NULL; we'll never dereference it */
1497 p->unregistering = ERR_PTR(-EINVAL);
1498 }
1499 /*
1500 * do not remove from the list until nobody holds it; walking the
1501 * list in do_sysctl() relies on that.
1502 */
1503 list_del_init(&p->ctl_entry);
1504 }
1505
1506 void sysctl_head_get(struct ctl_table_header *head)
1507 {
1508 spin_lock(&sysctl_lock);
1509 head->count++;
1510 spin_unlock(&sysctl_lock);
1511 }
1512
1513 void sysctl_head_put(struct ctl_table_header *head)
1514 {
1515 spin_lock(&sysctl_lock);
1516 if (!--head->count)
1517 kfree(head);
1518 spin_unlock(&sysctl_lock);
1519 }
1520
1521 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1522 {
1523 if (!head)
1524 BUG();
1525 spin_lock(&sysctl_lock);
1526 if (!use_table(head))
1527 head = ERR_PTR(-ENOENT);
1528 spin_unlock(&sysctl_lock);
1529 return head;
1530 }
1531
1532 void sysctl_head_finish(struct ctl_table_header *head)
1533 {
1534 if (!head)
1535 return;
1536 spin_lock(&sysctl_lock);
1537 unuse_table(head);
1538 spin_unlock(&sysctl_lock);
1539 }
1540
1541 static struct ctl_table_set *
1542 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1543 {
1544 struct ctl_table_set *set = &root->default_set;
1545 if (root->lookup)
1546 set = root->lookup(root, namespaces);
1547 return set;
1548 }
1549
1550 static struct list_head *
1551 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1552 {
1553 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1554 return &set->list;
1555 }
1556
1557 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1558 struct ctl_table_header *prev)
1559 {
1560 struct ctl_table_root *root;
1561 struct list_head *header_list;
1562 struct ctl_table_header *head;
1563 struct list_head *tmp;
1564
1565 spin_lock(&sysctl_lock);
1566 if (prev) {
1567 head = prev;
1568 tmp = &prev->ctl_entry;
1569 unuse_table(prev);
1570 goto next;
1571 }
1572 tmp = &root_table_header.ctl_entry;
1573 for (;;) {
1574 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1575
1576 if (!use_table(head))
1577 goto next;
1578 spin_unlock(&sysctl_lock);
1579 return head;
1580 next:
1581 root = head->root;
1582 tmp = tmp->next;
1583 header_list = lookup_header_list(root, namespaces);
1584 if (tmp != header_list)
1585 continue;
1586
1587 do {
1588 root = list_entry(root->root_list.next,
1589 struct ctl_table_root, root_list);
1590 if (root == &sysctl_table_root)
1591 goto out;
1592 header_list = lookup_header_list(root, namespaces);
1593 } while (list_empty(header_list));
1594 tmp = header_list->next;
1595 }
1596 out:
1597 spin_unlock(&sysctl_lock);
1598 return NULL;
1599 }
1600
1601 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1602 {
1603 return __sysctl_head_next(current->nsproxy, prev);
1604 }
1605
1606 void register_sysctl_root(struct ctl_table_root *root)
1607 {
1608 spin_lock(&sysctl_lock);
1609 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1610 spin_unlock(&sysctl_lock);
1611 }
1612
1613 /*
1614 * sysctl_perm does NOT grant the superuser all rights automatically, because
1615 * some sysctl variables are readonly even to root.
1616 */
1617
1618 static int test_perm(int mode, int op)
1619 {
1620 if (!current_euid())
1621 mode >>= 6;
1622 else if (in_egroup_p(0))
1623 mode >>= 3;
1624 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1625 return 0;
1626 return -EACCES;
1627 }
1628
1629 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1630 {
1631 int error;
1632 int mode;
1633
1634 error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1635 if (error)
1636 return error;
1637
1638 if (root->permissions)
1639 mode = root->permissions(root, current->nsproxy, table);
1640 else
1641 mode = table->mode;
1642
1643 return test_perm(mode, op);
1644 }
1645
1646 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1647 {
1648 for (; table->procname; table++) {
1649 table->parent = parent;
1650 if (table->child)
1651 sysctl_set_parent(table, table->child);
1652 }
1653 }
1654
1655 static __init int sysctl_init(void)
1656 {
1657 sysctl_set_parent(NULL, root_table);
1658 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1659 {
1660 int err;
1661 err = sysctl_check_table(current->nsproxy, root_table);
1662 }
1663 #endif
1664 return 0;
1665 }
1666
1667 core_initcall(sysctl_init);
1668
1669 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1670 struct ctl_table *table)
1671 {
1672 struct ctl_table *p;
1673 const char *s = branch->procname;
1674
1675 /* branch should have named subdirectory as its first element */
1676 if (!s || !branch->child)
1677 return NULL;
1678
1679 /* ... and nothing else */
1680 if (branch[1].procname)
1681 return NULL;
1682
1683 /* table should contain subdirectory with the same name */
1684 for (p = table; p->procname; p++) {
1685 if (!p->child)
1686 continue;
1687 if (p->procname && strcmp(p->procname, s) == 0)
1688 return p;
1689 }
1690 return NULL;
1691 }
1692
1693 /* see if attaching q to p would be an improvement */
1694 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1695 {
1696 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1697 struct ctl_table *next;
1698 int is_better = 0;
1699 int not_in_parent = !p->attached_by;
1700
1701 while ((next = is_branch_in(by, to)) != NULL) {
1702 if (by == q->attached_by)
1703 is_better = 1;
1704 if (to == p->attached_by)
1705 not_in_parent = 1;
1706 by = by->child;
1707 to = next->child;
1708 }
1709
1710 if (is_better && not_in_parent) {
1711 q->attached_by = by;
1712 q->attached_to = to;
1713 q->parent = p;
1714 }
1715 }
1716
1717 /**
1718 * __register_sysctl_paths - register a sysctl hierarchy
1719 * @root: List of sysctl headers to register on
1720 * @namespaces: Data to compute which lists of sysctl entries are visible
1721 * @path: The path to the directory the sysctl table is in.
1722 * @table: the top-level table structure
1723 *
1724 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1725 * array. A completely 0 filled entry terminates the table.
1726 *
1727 * The members of the &struct ctl_table structure are used as follows:
1728 *
1729 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1730 * enter a sysctl file
1731 *
1732 * data - a pointer to data for use by proc_handler
1733 *
1734 * maxlen - the maximum size in bytes of the data
1735 *
1736 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1737 *
1738 * child - a pointer to the child sysctl table if this entry is a directory, or
1739 * %NULL.
1740 *
1741 * proc_handler - the text handler routine (described below)
1742 *
1743 * de - for internal use by the sysctl routines
1744 *
1745 * extra1, extra2 - extra pointers usable by the proc handler routines
1746 *
1747 * Leaf nodes in the sysctl tree will be represented by a single file
1748 * under /proc; non-leaf nodes will be represented by directories.
1749 *
1750 * sysctl(2) can automatically manage read and write requests through
1751 * the sysctl table. The data and maxlen fields of the ctl_table
1752 * struct enable minimal validation of the values being written to be
1753 * performed, and the mode field allows minimal authentication.
1754 *
1755 * There must be a proc_handler routine for any terminal nodes
1756 * mirrored under /proc/sys (non-terminals are handled by a built-in
1757 * directory handler). Several default handlers are available to
1758 * cover common cases -
1759 *
1760 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1761 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1762 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1763 *
1764 * It is the handler's job to read the input buffer from user memory
1765 * and process it. The handler should return 0 on success.
1766 *
1767 * This routine returns %NULL on a failure to register, and a pointer
1768 * to the table header on success.
1769 */
1770 struct ctl_table_header *__register_sysctl_paths(
1771 struct ctl_table_root *root,
1772 struct nsproxy *namespaces,
1773 const struct ctl_path *path, struct ctl_table *table)
1774 {
1775 struct ctl_table_header *header;
1776 struct ctl_table *new, **prevp;
1777 unsigned int n, npath;
1778 struct ctl_table_set *set;
1779
1780 /* Count the path components */
1781 for (npath = 0; path[npath].procname; ++npath)
1782 ;
1783
1784 /*
1785 * For each path component, allocate a 2-element ctl_table array.
1786 * The first array element will be filled with the sysctl entry
1787 * for this, the second will be the sentinel (procname == 0).
1788 *
1789 * We allocate everything in one go so that we don't have to
1790 * worry about freeing additional memory in unregister_sysctl_table.
1791 */
1792 header = kzalloc(sizeof(struct ctl_table_header) +
1793 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1794 if (!header)
1795 return NULL;
1796
1797 new = (struct ctl_table *) (header + 1);
1798
1799 /* Now connect the dots */
1800 prevp = &header->ctl_table;
1801 for (n = 0; n < npath; ++n, ++path) {
1802 /* Copy the procname */
1803 new->procname = path->procname;
1804 new->mode = 0555;
1805
1806 *prevp = new;
1807 prevp = &new->child;
1808
1809 new += 2;
1810 }
1811 *prevp = table;
1812 header->ctl_table_arg = table;
1813
1814 INIT_LIST_HEAD(&header->ctl_entry);
1815 header->used = 0;
1816 header->unregistering = NULL;
1817 header->root = root;
1818 sysctl_set_parent(NULL, header->ctl_table);
1819 header->count = 1;
1820 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1821 if (sysctl_check_table(namespaces, header->ctl_table)) {
1822 kfree(header);
1823 return NULL;
1824 }
1825 #endif
1826 spin_lock(&sysctl_lock);
1827 header->set = lookup_header_set(root, namespaces);
1828 header->attached_by = header->ctl_table;
1829 header->attached_to = root_table;
1830 header->parent = &root_table_header;
1831 for (set = header->set; set; set = set->parent) {
1832 struct ctl_table_header *p;
1833 list_for_each_entry(p, &set->list, ctl_entry) {
1834 if (p->unregistering)
1835 continue;
1836 try_attach(p, header);
1837 }
1838 }
1839 header->parent->count++;
1840 list_add_tail(&header->ctl_entry, &header->set->list);
1841 spin_unlock(&sysctl_lock);
1842
1843 return header;
1844 }
1845
1846 /**
1847 * register_sysctl_table_path - register a sysctl table hierarchy
1848 * @path: The path to the directory the sysctl table is in.
1849 * @table: the top-level table structure
1850 *
1851 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1852 * array. A completely 0 filled entry terminates the table.
1853 *
1854 * See __register_sysctl_paths for more details.
1855 */
1856 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1857 struct ctl_table *table)
1858 {
1859 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1860 path, table);
1861 }
1862
1863 /**
1864 * register_sysctl_table - register a sysctl table hierarchy
1865 * @table: the top-level table structure
1866 *
1867 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1868 * array. A completely 0 filled entry terminates the table.
1869 *
1870 * See register_sysctl_paths for more details.
1871 */
1872 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1873 {
1874 static const struct ctl_path null_path[] = { {} };
1875
1876 return register_sysctl_paths(null_path, table);
1877 }
1878
1879 /**
1880 * unregister_sysctl_table - unregister a sysctl table hierarchy
1881 * @header: the header returned from register_sysctl_table
1882 *
1883 * Unregisters the sysctl table and all children. proc entries may not
1884 * actually be removed until they are no longer used by anyone.
1885 */
1886 void unregister_sysctl_table(struct ctl_table_header * header)
1887 {
1888 might_sleep();
1889
1890 if (header == NULL)
1891 return;
1892
1893 spin_lock(&sysctl_lock);
1894 start_unregistering(header);
1895 if (!--header->parent->count) {
1896 WARN_ON(1);
1897 kfree(header->parent);
1898 }
1899 if (!--header->count)
1900 kfree(header);
1901 spin_unlock(&sysctl_lock);
1902 }
1903
1904 int sysctl_is_seen(struct ctl_table_header *p)
1905 {
1906 struct ctl_table_set *set = p->set;
1907 int res;
1908 spin_lock(&sysctl_lock);
1909 if (p->unregistering)
1910 res = 0;
1911 else if (!set->is_seen)
1912 res = 1;
1913 else
1914 res = set->is_seen(set);
1915 spin_unlock(&sysctl_lock);
1916 return res;
1917 }
1918
1919 void setup_sysctl_set(struct ctl_table_set *p,
1920 struct ctl_table_set *parent,
1921 int (*is_seen)(struct ctl_table_set *))
1922 {
1923 INIT_LIST_HEAD(&p->list);
1924 p->parent = parent ? parent : &sysctl_table_root.default_set;
1925 p->is_seen = is_seen;
1926 }
1927
1928 #else /* !CONFIG_SYSCTL */
1929 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1930 {
1931 return NULL;
1932 }
1933
1934 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1935 struct ctl_table *table)
1936 {
1937 return NULL;
1938 }
1939
1940 void unregister_sysctl_table(struct ctl_table_header * table)
1941 {
1942 }
1943
1944 void setup_sysctl_set(struct ctl_table_set *p,
1945 struct ctl_table_set *parent,
1946 int (*is_seen)(struct ctl_table_set *))
1947 {
1948 }
1949
1950 void sysctl_head_put(struct ctl_table_header *head)
1951 {
1952 }
1953
1954 #endif /* CONFIG_SYSCTL */
1955
1956 /*
1957 * /proc/sys support
1958 */
1959
1960 #ifdef CONFIG_PROC_SYSCTL
1961
1962 static int _proc_do_string(void* data, int maxlen, int write,
1963 void __user *buffer,
1964 size_t *lenp, loff_t *ppos)
1965 {
1966 size_t len;
1967 char __user *p;
1968 char c;
1969
1970 if (!data || !maxlen || !*lenp) {
1971 *lenp = 0;
1972 return 0;
1973 }
1974
1975 if (write) {
1976 len = 0;
1977 p = buffer;
1978 while (len < *lenp) {
1979 if (get_user(c, p++))
1980 return -EFAULT;
1981 if (c == 0 || c == '\n')
1982 break;
1983 len++;
1984 }
1985 if (len >= maxlen)
1986 len = maxlen-1;
1987 if(copy_from_user(data, buffer, len))
1988 return -EFAULT;
1989 ((char *) data)[len] = 0;
1990 *ppos += *lenp;
1991 } else {
1992 len = strlen(data);
1993 if (len > maxlen)
1994 len = maxlen;
1995
1996 if (*ppos > len) {
1997 *lenp = 0;
1998 return 0;
1999 }
2000
2001 data += *ppos;
2002 len -= *ppos;
2003
2004 if (len > *lenp)
2005 len = *lenp;
2006 if (len)
2007 if(copy_to_user(buffer, data, len))
2008 return -EFAULT;
2009 if (len < *lenp) {
2010 if(put_user('\n', ((char __user *) buffer) + len))
2011 return -EFAULT;
2012 len++;
2013 }
2014 *lenp = len;
2015 *ppos += len;
2016 }
2017 return 0;
2018 }
2019
2020 /**
2021 * proc_dostring - read a string sysctl
2022 * @table: the sysctl table
2023 * @write: %TRUE if this is a write to the sysctl file
2024 * @buffer: the user buffer
2025 * @lenp: the size of the user buffer
2026 * @ppos: file position
2027 *
2028 * Reads/writes a string from/to the user buffer. If the kernel
2029 * buffer provided is not large enough to hold the string, the
2030 * string is truncated. The copied string is %NULL-terminated.
2031 * If the string is being read by the user process, it is copied
2032 * and a newline '\n' is added. It is truncated if the buffer is
2033 * not large enough.
2034 *
2035 * Returns 0 on success.
2036 */
2037 int proc_dostring(struct ctl_table *table, int write,
2038 void __user *buffer, size_t *lenp, loff_t *ppos)
2039 {
2040 return _proc_do_string(table->data, table->maxlen, write,
2041 buffer, lenp, ppos);
2042 }
2043
2044
2045 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2046 int *valp,
2047 int write, void *data)
2048 {
2049 if (write) {
2050 *valp = *negp ? -*lvalp : *lvalp;
2051 } else {
2052 int val = *valp;
2053 if (val < 0) {
2054 *negp = -1;
2055 *lvalp = (unsigned long)-val;
2056 } else {
2057 *negp = 0;
2058 *lvalp = (unsigned long)val;
2059 }
2060 }
2061 return 0;
2062 }
2063
2064 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2065 int write, void __user *buffer,
2066 size_t *lenp, loff_t *ppos,
2067 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2068 int write, void *data),
2069 void *data)
2070 {
2071 #define TMPBUFLEN 21
2072 int *i, vleft, first = 1, neg;
2073 unsigned long lval;
2074 size_t left, len;
2075
2076 char buf[TMPBUFLEN], *p;
2077 char __user *s = buffer;
2078
2079 if (!tbl_data || !table->maxlen || !*lenp ||
2080 (*ppos && !write)) {
2081 *lenp = 0;
2082 return 0;
2083 }
2084
2085 i = (int *) tbl_data;
2086 vleft = table->maxlen / sizeof(*i);
2087 left = *lenp;
2088
2089 if (!conv)
2090 conv = do_proc_dointvec_conv;
2091
2092 for (; left && vleft--; i++, first=0) {
2093 if (write) {
2094 while (left) {
2095 char c;
2096 if (get_user(c, s))
2097 return -EFAULT;
2098 if (!isspace(c))
2099 break;
2100 left--;
2101 s++;
2102 }
2103 if (!left)
2104 break;
2105 neg = 0;
2106 len = left;
2107 if (len > sizeof(buf) - 1)
2108 len = sizeof(buf) - 1;
2109 if (copy_from_user(buf, s, len))
2110 return -EFAULT;
2111 buf[len] = 0;
2112 p = buf;
2113 if (*p == '-' && left > 1) {
2114 neg = 1;
2115 p++;
2116 }
2117 if (*p < '0' || *p > '9')
2118 break;
2119
2120 lval = simple_strtoul(p, &p, 0);
2121
2122 len = p-buf;
2123 if ((len < left) && *p && !isspace(*p))
2124 break;
2125 s += len;
2126 left -= len;
2127
2128 if (conv(&neg, &lval, i, 1, data))
2129 break;
2130 } else {
2131 p = buf;
2132 if (!first)
2133 *p++ = '\t';
2134
2135 if (conv(&neg, &lval, i, 0, data))
2136 break;
2137
2138 sprintf(p, "%s%lu", neg ? "-" : "", lval);
2139 len = strlen(buf);
2140 if (len > left)
2141 len = left;
2142 if(copy_to_user(s, buf, len))
2143 return -EFAULT;
2144 left -= len;
2145 s += len;
2146 }
2147 }
2148
2149 if (!write && !first && left) {
2150 if(put_user('\n', s))
2151 return -EFAULT;
2152 left--, s++;
2153 }
2154 if (write) {
2155 while (left) {
2156 char c;
2157 if (get_user(c, s++))
2158 return -EFAULT;
2159 if (!isspace(c))
2160 break;
2161 left--;
2162 }
2163 }
2164 if (write && first)
2165 return -EINVAL;
2166 *lenp -= left;
2167 *ppos += *lenp;
2168 return 0;
2169 #undef TMPBUFLEN
2170 }
2171
2172 static int do_proc_dointvec(struct ctl_table *table, int write,
2173 void __user *buffer, size_t *lenp, loff_t *ppos,
2174 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2175 int write, void *data),
2176 void *data)
2177 {
2178 return __do_proc_dointvec(table->data, table, write,
2179 buffer, lenp, ppos, conv, data);
2180 }
2181
2182 /**
2183 * proc_dointvec - read a vector of integers
2184 * @table: the sysctl table
2185 * @write: %TRUE if this is a write to the sysctl file
2186 * @buffer: the user buffer
2187 * @lenp: the size of the user buffer
2188 * @ppos: file position
2189 *
2190 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2191 * values from/to the user buffer, treated as an ASCII string.
2192 *
2193 * Returns 0 on success.
2194 */
2195 int proc_dointvec(struct ctl_table *table, int write,
2196 void __user *buffer, size_t *lenp, loff_t *ppos)
2197 {
2198 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2199 NULL,NULL);
2200 }
2201
2202 /*
2203 * Taint values can only be increased
2204 * This means we can safely use a temporary.
2205 */
2206 static int proc_taint(struct ctl_table *table, int write,
2207 void __user *buffer, size_t *lenp, loff_t *ppos)
2208 {
2209 struct ctl_table t;
2210 unsigned long tmptaint = get_taint();
2211 int err;
2212
2213 if (write && !capable(CAP_SYS_ADMIN))
2214 return -EPERM;
2215
2216 t = *table;
2217 t.data = &tmptaint;
2218 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2219 if (err < 0)
2220 return err;
2221
2222 if (write) {
2223 /*
2224 * Poor man's atomic or. Not worth adding a primitive
2225 * to everyone's atomic.h for this
2226 */
2227 int i;
2228 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2229 if ((tmptaint >> i) & 1)
2230 add_taint(i);
2231 }
2232 }
2233
2234 return err;
2235 }
2236
2237 struct do_proc_dointvec_minmax_conv_param {
2238 int *min;
2239 int *max;
2240 };
2241
2242 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2243 int *valp,
2244 int write, void *data)
2245 {
2246 struct do_proc_dointvec_minmax_conv_param *param = data;
2247 if (write) {
2248 int val = *negp ? -*lvalp : *lvalp;
2249 if ((param->min && *param->min > val) ||
2250 (param->max && *param->max < val))
2251 return -EINVAL;
2252 *valp = val;
2253 } else {
2254 int val = *valp;
2255 if (val < 0) {
2256 *negp = -1;
2257 *lvalp = (unsigned long)-val;
2258 } else {
2259 *negp = 0;
2260 *lvalp = (unsigned long)val;
2261 }
2262 }
2263 return 0;
2264 }
2265
2266 /**
2267 * proc_dointvec_minmax - read a vector of integers with min/max values
2268 * @table: the sysctl table
2269 * @write: %TRUE if this is a write to the sysctl file
2270 * @buffer: the user buffer
2271 * @lenp: the size of the user buffer
2272 * @ppos: file position
2273 *
2274 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2275 * values from/to the user buffer, treated as an ASCII string.
2276 *
2277 * This routine will ensure the values are within the range specified by
2278 * table->extra1 (min) and table->extra2 (max).
2279 *
2280 * Returns 0 on success.
2281 */
2282 int proc_dointvec_minmax(struct ctl_table *table, int write,
2283 void __user *buffer, size_t *lenp, loff_t *ppos)
2284 {
2285 struct do_proc_dointvec_minmax_conv_param param = {
2286 .min = (int *) table->extra1,
2287 .max = (int *) table->extra2,
2288 };
2289 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2290 do_proc_dointvec_minmax_conv, &param);
2291 }
2292
2293 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2294 void __user *buffer,
2295 size_t *lenp, loff_t *ppos,
2296 unsigned long convmul,
2297 unsigned long convdiv)
2298 {
2299 #define TMPBUFLEN 21
2300 unsigned long *i, *min, *max, val;
2301 int vleft, first=1, neg;
2302 size_t len, left;
2303 char buf[TMPBUFLEN], *p;
2304 char __user *s = buffer;
2305
2306 if (!data || !table->maxlen || !*lenp ||
2307 (*ppos && !write)) {
2308 *lenp = 0;
2309 return 0;
2310 }
2311
2312 i = (unsigned long *) data;
2313 min = (unsigned long *) table->extra1;
2314 max = (unsigned long *) table->extra2;
2315 vleft = table->maxlen / sizeof(unsigned long);
2316 left = *lenp;
2317
2318 for (; left && vleft--; i++, min++, max++, first=0) {
2319 if (write) {
2320 while (left) {
2321 char c;
2322 if (get_user(c, s))
2323 return -EFAULT;
2324 if (!isspace(c))
2325 break;
2326 left--;
2327 s++;
2328 }
2329 if (!left)
2330 break;
2331 neg = 0;
2332 len = left;
2333 if (len > TMPBUFLEN-1)
2334 len = TMPBUFLEN-1;
2335 if (copy_from_user(buf, s, len))
2336 return -EFAULT;
2337 buf[len] = 0;
2338 p = buf;
2339 if (*p == '-' && left > 1) {
2340 neg = 1;
2341 p++;
2342 }
2343 if (*p < '0' || *p > '9')
2344 break;
2345 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2346 len = p-buf;
2347 if ((len < left) && *p && !isspace(*p))
2348 break;
2349 if (neg)
2350 val = -val;
2351 s += len;
2352 left -= len;
2353
2354 if(neg)
2355 continue;
2356 if ((min && val < *min) || (max && val > *max))
2357 continue;
2358 *i = val;
2359 } else {
2360 p = buf;
2361 if (!first)
2362 *p++ = '\t';
2363 sprintf(p, "%lu", convdiv * (*i) / convmul);
2364 len = strlen(buf);
2365 if (len > left)
2366 len = left;
2367 if(copy_to_user(s, buf, len))
2368 return -EFAULT;
2369 left -= len;
2370 s += len;
2371 }
2372 }
2373
2374 if (!write && !first && left) {
2375 if(put_user('\n', s))
2376 return -EFAULT;
2377 left--, s++;
2378 }
2379 if (write) {
2380 while (left) {
2381 char c;
2382 if (get_user(c, s++))
2383 return -EFAULT;
2384 if (!isspace(c))
2385 break;
2386 left--;
2387 }
2388 }
2389 if (write && first)
2390 return -EINVAL;
2391 *lenp -= left;
2392 *ppos += *lenp;
2393 return 0;
2394 #undef TMPBUFLEN
2395 }
2396
2397 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2398 void __user *buffer,
2399 size_t *lenp, loff_t *ppos,
2400 unsigned long convmul,
2401 unsigned long convdiv)
2402 {
2403 return __do_proc_doulongvec_minmax(table->data, table, write,
2404 buffer, lenp, ppos, convmul, convdiv);
2405 }
2406
2407 /**
2408 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2409 * @table: the sysctl table
2410 * @write: %TRUE if this is a write to the sysctl file
2411 * @buffer: the user buffer
2412 * @lenp: the size of the user buffer
2413 * @ppos: file position
2414 *
2415 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2416 * values from/to the user buffer, treated as an ASCII string.
2417 *
2418 * This routine will ensure the values are within the range specified by
2419 * table->extra1 (min) and table->extra2 (max).
2420 *
2421 * Returns 0 on success.
2422 */
2423 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2424 void __user *buffer, size_t *lenp, loff_t *ppos)
2425 {
2426 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2427 }
2428
2429 /**
2430 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2431 * @table: the sysctl table
2432 * @write: %TRUE if this is a write to the sysctl file
2433 * @buffer: the user buffer
2434 * @lenp: the size of the user buffer
2435 * @ppos: file position
2436 *
2437 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2438 * values from/to the user buffer, treated as an ASCII string. The values
2439 * are treated as milliseconds, and converted to jiffies when they are stored.
2440 *
2441 * This routine will ensure the values are within the range specified by
2442 * table->extra1 (min) and table->extra2 (max).
2443 *
2444 * Returns 0 on success.
2445 */
2446 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2447 void __user *buffer,
2448 size_t *lenp, loff_t *ppos)
2449 {
2450 return do_proc_doulongvec_minmax(table, write, buffer,
2451 lenp, ppos, HZ, 1000l);
2452 }
2453
2454
2455 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2456 int *valp,
2457 int write, void *data)
2458 {
2459 if (write) {
2460 if (*lvalp > LONG_MAX / HZ)
2461 return 1;
2462 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2463 } else {
2464 int val = *valp;
2465 unsigned long lval;
2466 if (val < 0) {
2467 *negp = -1;
2468 lval = (unsigned long)-val;
2469 } else {
2470 *negp = 0;
2471 lval = (unsigned long)val;
2472 }
2473 *lvalp = lval / HZ;
2474 }
2475 return 0;
2476 }
2477
2478 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2479 int *valp,
2480 int write, void *data)
2481 {
2482 if (write) {
2483 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2484 return 1;
2485 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2486 } else {
2487 int val = *valp;
2488 unsigned long lval;
2489 if (val < 0) {
2490 *negp = -1;
2491 lval = (unsigned long)-val;
2492 } else {
2493 *negp = 0;
2494 lval = (unsigned long)val;
2495 }
2496 *lvalp = jiffies_to_clock_t(lval);
2497 }
2498 return 0;
2499 }
2500
2501 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2502 int *valp,
2503 int write, void *data)
2504 {
2505 if (write) {
2506 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2507 } else {
2508 int val = *valp;
2509 unsigned long lval;
2510 if (val < 0) {
2511 *negp = -1;
2512 lval = (unsigned long)-val;
2513 } else {
2514 *negp = 0;
2515 lval = (unsigned long)val;
2516 }
2517 *lvalp = jiffies_to_msecs(lval);
2518 }
2519 return 0;
2520 }
2521
2522 /**
2523 * proc_dointvec_jiffies - read a vector of integers as seconds
2524 * @table: the sysctl table
2525 * @write: %TRUE if this is a write to the sysctl file
2526 * @buffer: the user buffer
2527 * @lenp: the size of the user buffer
2528 * @ppos: file position
2529 *
2530 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2531 * values from/to the user buffer, treated as an ASCII string.
2532 * The values read are assumed to be in seconds, and are converted into
2533 * jiffies.
2534 *
2535 * Returns 0 on success.
2536 */
2537 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2538 void __user *buffer, size_t *lenp, loff_t *ppos)
2539 {
2540 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2541 do_proc_dointvec_jiffies_conv,NULL);
2542 }
2543
2544 /**
2545 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2546 * @table: the sysctl table
2547 * @write: %TRUE if this is a write to the sysctl file
2548 * @buffer: the user buffer
2549 * @lenp: the size of the user buffer
2550 * @ppos: pointer to the file position
2551 *
2552 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2553 * values from/to the user buffer, treated as an ASCII string.
2554 * The values read are assumed to be in 1/USER_HZ seconds, and
2555 * are converted into jiffies.
2556 *
2557 * Returns 0 on success.
2558 */
2559 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2560 void __user *buffer, size_t *lenp, loff_t *ppos)
2561 {
2562 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2563 do_proc_dointvec_userhz_jiffies_conv,NULL);
2564 }
2565
2566 /**
2567 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2568 * @table: the sysctl table
2569 * @write: %TRUE if this is a write to the sysctl file
2570 * @buffer: the user buffer
2571 * @lenp: the size of the user buffer
2572 * @ppos: file position
2573 * @ppos: the current position in the file
2574 *
2575 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2576 * values from/to the user buffer, treated as an ASCII string.
2577 * The values read are assumed to be in 1/1000 seconds, and
2578 * are converted into jiffies.
2579 *
2580 * Returns 0 on success.
2581 */
2582 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2583 void __user *buffer, size_t *lenp, loff_t *ppos)
2584 {
2585 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2586 do_proc_dointvec_ms_jiffies_conv, NULL);
2587 }
2588
2589 static int proc_do_cad_pid(struct ctl_table *table, int write,
2590 void __user *buffer, size_t *lenp, loff_t *ppos)
2591 {
2592 struct pid *new_pid;
2593 pid_t tmp;
2594 int r;
2595
2596 tmp = pid_vnr(cad_pid);
2597
2598 r = __do_proc_dointvec(&tmp, table, write, buffer,
2599 lenp, ppos, NULL, NULL);
2600 if (r || !write)
2601 return r;
2602
2603 new_pid = find_get_pid(tmp);
2604 if (!new_pid)
2605 return -ESRCH;
2606
2607 put_pid(xchg(&cad_pid, new_pid));
2608 return 0;
2609 }
2610
2611 #else /* CONFIG_PROC_FS */
2612
2613 int proc_dostring(struct ctl_table *table, int write,
2614 void __user *buffer, size_t *lenp, loff_t *ppos)
2615 {
2616 return -ENOSYS;
2617 }
2618
2619 int proc_dointvec(struct ctl_table *table, int write,
2620 void __user *buffer, size_t *lenp, loff_t *ppos)
2621 {
2622 return -ENOSYS;
2623 }
2624
2625 int proc_dointvec_minmax(struct ctl_table *table, int write,
2626 void __user *buffer, size_t *lenp, loff_t *ppos)
2627 {
2628 return -ENOSYS;
2629 }
2630
2631 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2632 void __user *buffer, size_t *lenp, loff_t *ppos)
2633 {
2634 return -ENOSYS;
2635 }
2636
2637 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2638 void __user *buffer, size_t *lenp, loff_t *ppos)
2639 {
2640 return -ENOSYS;
2641 }
2642
2643 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2644 void __user *buffer, size_t *lenp, loff_t *ppos)
2645 {
2646 return -ENOSYS;
2647 }
2648
2649 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2650 void __user *buffer, size_t *lenp, loff_t *ppos)
2651 {
2652 return -ENOSYS;
2653 }
2654
2655 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2656 void __user *buffer,
2657 size_t *lenp, loff_t *ppos)
2658 {
2659 return -ENOSYS;
2660 }
2661
2662
2663 #endif /* CONFIG_PROC_FS */
2664
2665 /*
2666 * No sense putting this after each symbol definition, twice,
2667 * exception granted :-)
2668 */
2669 EXPORT_SYMBOL(proc_dointvec);
2670 EXPORT_SYMBOL(proc_dointvec_jiffies);
2671 EXPORT_SYMBOL(proc_dointvec_minmax);
2672 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2673 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2674 EXPORT_SYMBOL(proc_dostring);
2675 EXPORT_SYMBOL(proc_doulongvec_minmax);
2676 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2677 EXPORT_SYMBOL(register_sysctl_table);
2678 EXPORT_SYMBOL(register_sysctl_paths);
2679 EXPORT_SYMBOL(unregister_sysctl_table);
Linux 2.6 ibm-acpi/thinkpad-acpi driver git tree