sysctl: Remove the last of the generic binary sysctl support
[GitHub/mt8127/android_kernel_alcatel_ttab.git] / kernel / sysctl.c
1 /*
2 * sysctl.c: General linux system control interface
3 *
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12 * Horn.
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16 * Wendling.
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19 */
20
21 #include <linux/module.h>
22 #include <linux/mm.h>
23 #include <linux/swap.h>
24 #include <linux/slab.h>
25 #include <linux/sysctl.h>
26 #include <linux/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/hugetlb.h>
39 #include <linux/initrd.h>
40 #include <linux/key.h>
41 #include <linux/times.h>
42 #include <linux/limits.h>
43 #include <linux/dcache.h>
44 #include <linux/syscalls.h>
45 #include <linux/vmstat.h>
46 #include <linux/nfs_fs.h>
47 #include <linux/acpi.h>
48 #include <linux/reboot.h>
49 #include <linux/ftrace.h>
50 #include <linux/slow-work.h>
51 #include <linux/perf_event.h>
52
53 #include <asm/uaccess.h>
54 #include <asm/processor.h>
55
56 #ifdef CONFIG_X86
57 #include <asm/nmi.h>
58 #include <asm/stacktrace.h>
59 #include <asm/io.h>
60 #endif
61
62
63 #if defined(CONFIG_SYSCTL)
64
65 /* External variables not in a header file. */
66 extern int C_A_D;
67 extern int print_fatal_signals;
68 extern int sysctl_overcommit_memory;
69 extern int sysctl_overcommit_ratio;
70 extern int sysctl_panic_on_oom;
71 extern int sysctl_oom_kill_allocating_task;
72 extern int sysctl_oom_dump_tasks;
73 extern int max_threads;
74 extern int core_uses_pid;
75 extern int suid_dumpable;
76 extern char core_pattern[];
77 extern unsigned int core_pipe_limit;
78 extern int pid_max;
79 extern int min_free_kbytes;
80 extern int pid_max_min, pid_max_max;
81 extern int sysctl_drop_caches;
82 extern int percpu_pagelist_fraction;
83 extern int compat_log;
84 extern int latencytop_enabled;
85 extern int sysctl_nr_open_min, sysctl_nr_open_max;
86 #ifndef CONFIG_MMU
87 extern int sysctl_nr_trim_pages;
88 #endif
89 #ifdef CONFIG_RCU_TORTURE_TEST
90 extern int rcutorture_runnable;
91 #endif /* #ifdef CONFIG_RCU_TORTURE_TEST */
92 #ifdef CONFIG_BLOCK
93 extern int blk_iopoll_enabled;
94 #endif
95
96 /* Constants used for minimum and maximum */
97 #ifdef CONFIG_DETECT_SOFTLOCKUP
98 static int sixty = 60;
99 static int neg_one = -1;
100 #endif
101
102 static int zero;
103 static int __maybe_unused one = 1;
104 static int __maybe_unused two = 2;
105 static unsigned long one_ul = 1;
106 static int one_hundred = 100;
107 #ifdef CONFIG_PRINTK
108 static int ten_thousand = 10000;
109 #endif
110
111 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
112 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
113
114 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
115 static int maxolduid = 65535;
116 static int minolduid;
117 static int min_percpu_pagelist_fract = 8;
118
119 static int ngroups_max = NGROUPS_MAX;
120
121 #ifdef CONFIG_MODULES
122 extern char modprobe_path[];
123 extern int modules_disabled;
124 #endif
125 #ifdef CONFIG_CHR_DEV_SG
126 extern int sg_big_buff;
127 #endif
128
129 #ifdef CONFIG_SPARC
130 #include <asm/system.h>
131 #endif
132
133 #ifdef CONFIG_SPARC64
134 extern int sysctl_tsb_ratio;
135 #endif
136
137 #ifdef __hppa__
138 extern int pwrsw_enabled;
139 extern int unaligned_enabled;
140 #endif
141
142 #ifdef CONFIG_S390
143 #ifdef CONFIG_MATHEMU
144 extern int sysctl_ieee_emulation_warnings;
145 #endif
146 extern int sysctl_userprocess_debug;
147 extern int spin_retry;
148 #endif
149
150 #ifdef CONFIG_BSD_PROCESS_ACCT
151 extern int acct_parm[];
152 #endif
153
154 #ifdef CONFIG_IA64
155 extern int no_unaligned_warning;
156 extern int unaligned_dump_stack;
157 #endif
158
159 #ifdef CONFIG_RT_MUTEXES
160 extern int max_lock_depth;
161 #endif
162
163 #ifdef CONFIG_PROC_SYSCTL
164 static int proc_do_cad_pid(struct ctl_table *table, int write,
165 void __user *buffer, size_t *lenp, loff_t *ppos);
166 static int proc_taint(struct ctl_table *table, int write,
167 void __user *buffer, size_t *lenp, loff_t *ppos);
168 #endif
169
170 static struct ctl_table root_table[];
171 static struct ctl_table_root sysctl_table_root;
172 static struct ctl_table_header root_table_header = {
173 .count = 1,
174 .ctl_table = root_table,
175 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
176 .root = &sysctl_table_root,
177 .set = &sysctl_table_root.default_set,
178 };
179 static struct ctl_table_root sysctl_table_root = {
180 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
181 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
182 };
183
184 static struct ctl_table kern_table[];
185 static struct ctl_table vm_table[];
186 static struct ctl_table fs_table[];
187 static struct ctl_table debug_table[];
188 static struct ctl_table dev_table[];
189 extern struct ctl_table random_table[];
190 #ifdef CONFIG_INOTIFY_USER
191 extern struct ctl_table inotify_table[];
192 #endif
193 #ifdef CONFIG_EPOLL
194 extern struct ctl_table epoll_table[];
195 #endif
196
197 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
198 int sysctl_legacy_va_layout;
199 #endif
200
201 extern int prove_locking;
202 extern int lock_stat;
203
204 /* The default sysctl tables: */
205
206 static struct ctl_table root_table[] = {
207 {
208 .procname = "kernel",
209 .mode = 0555,
210 .child = kern_table,
211 },
212 {
213 .procname = "vm",
214 .mode = 0555,
215 .child = vm_table,
216 },
217 {
218 .procname = "fs",
219 .mode = 0555,
220 .child = fs_table,
221 },
222 {
223 .procname = "debug",
224 .mode = 0555,
225 .child = debug_table,
226 },
227 {
228 .procname = "dev",
229 .mode = 0555,
230 .child = dev_table,
231 },
232 /*
233 * NOTE: do not add new entries to this table unless you have read
234 * Documentation/sysctl/ctl_unnumbered.txt
235 */
236 { }
237 };
238
239 #ifdef CONFIG_SCHED_DEBUG
240 static int min_sched_granularity_ns = 100000; /* 100 usecs */
241 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
242 static int min_wakeup_granularity_ns; /* 0 usecs */
243 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
244 #endif
245
246 static struct ctl_table kern_table[] = {
247 {
248 .procname = "sched_child_runs_first",
249 .data = &sysctl_sched_child_runs_first,
250 .maxlen = sizeof(unsigned int),
251 .mode = 0644,
252 .proc_handler = &proc_dointvec,
253 },
254 #ifdef CONFIG_SCHED_DEBUG
255 {
256 .procname = "sched_min_granularity_ns",
257 .data = &sysctl_sched_min_granularity,
258 .maxlen = sizeof(unsigned int),
259 .mode = 0644,
260 .proc_handler = &sched_nr_latency_handler,
261 .extra1 = &min_sched_granularity_ns,
262 .extra2 = &max_sched_granularity_ns,
263 },
264 {
265 .procname = "sched_latency_ns",
266 .data = &sysctl_sched_latency,
267 .maxlen = sizeof(unsigned int),
268 .mode = 0644,
269 .proc_handler = &sched_nr_latency_handler,
270 .extra1 = &min_sched_granularity_ns,
271 .extra2 = &max_sched_granularity_ns,
272 },
273 {
274 .procname = "sched_wakeup_granularity_ns",
275 .data = &sysctl_sched_wakeup_granularity,
276 .maxlen = sizeof(unsigned int),
277 .mode = 0644,
278 .proc_handler = &proc_dointvec_minmax,
279 .extra1 = &min_wakeup_granularity_ns,
280 .extra2 = &max_wakeup_granularity_ns,
281 },
282 {
283 .procname = "sched_shares_ratelimit",
284 .data = &sysctl_sched_shares_ratelimit,
285 .maxlen = sizeof(unsigned int),
286 .mode = 0644,
287 .proc_handler = &proc_dointvec,
288 },
289 {
290 .procname = "sched_shares_thresh",
291 .data = &sysctl_sched_shares_thresh,
292 .maxlen = sizeof(unsigned int),
293 .mode = 0644,
294 .proc_handler = &proc_dointvec_minmax,
295 .extra1 = &zero,
296 },
297 {
298 .procname = "sched_features",
299 .data = &sysctl_sched_features,
300 .maxlen = sizeof(unsigned int),
301 .mode = 0644,
302 .proc_handler = &proc_dointvec,
303 },
304 {
305 .procname = "sched_migration_cost",
306 .data = &sysctl_sched_migration_cost,
307 .maxlen = sizeof(unsigned int),
308 .mode = 0644,
309 .proc_handler = &proc_dointvec,
310 },
311 {
312 .procname = "sched_nr_migrate",
313 .data = &sysctl_sched_nr_migrate,
314 .maxlen = sizeof(unsigned int),
315 .mode = 0644,
316 .proc_handler = &proc_dointvec,
317 },
318 {
319 .procname = "sched_time_avg",
320 .data = &sysctl_sched_time_avg,
321 .maxlen = sizeof(unsigned int),
322 .mode = 0644,
323 .proc_handler = &proc_dointvec,
324 },
325 {
326 .procname = "timer_migration",
327 .data = &sysctl_timer_migration,
328 .maxlen = sizeof(unsigned int),
329 .mode = 0644,
330 .proc_handler = &proc_dointvec_minmax,
331 .extra1 = &zero,
332 .extra2 = &one,
333 },
334 #endif
335 {
336 .procname = "sched_rt_period_us",
337 .data = &sysctl_sched_rt_period,
338 .maxlen = sizeof(unsigned int),
339 .mode = 0644,
340 .proc_handler = &sched_rt_handler,
341 },
342 {
343 .procname = "sched_rt_runtime_us",
344 .data = &sysctl_sched_rt_runtime,
345 .maxlen = sizeof(int),
346 .mode = 0644,
347 .proc_handler = &sched_rt_handler,
348 },
349 {
350 .procname = "sched_compat_yield",
351 .data = &sysctl_sched_compat_yield,
352 .maxlen = sizeof(unsigned int),
353 .mode = 0644,
354 .proc_handler = &proc_dointvec,
355 },
356 #ifdef CONFIG_PROVE_LOCKING
357 {
358 .procname = "prove_locking",
359 .data = &prove_locking,
360 .maxlen = sizeof(int),
361 .mode = 0644,
362 .proc_handler = &proc_dointvec,
363 },
364 #endif
365 #ifdef CONFIG_LOCK_STAT
366 {
367 .procname = "lock_stat",
368 .data = &lock_stat,
369 .maxlen = sizeof(int),
370 .mode = 0644,
371 .proc_handler = &proc_dointvec,
372 },
373 #endif
374 {
375 .procname = "panic",
376 .data = &panic_timeout,
377 .maxlen = sizeof(int),
378 .mode = 0644,
379 .proc_handler = &proc_dointvec,
380 },
381 {
382 .procname = "core_uses_pid",
383 .data = &core_uses_pid,
384 .maxlen = sizeof(int),
385 .mode = 0644,
386 .proc_handler = &proc_dointvec,
387 },
388 {
389 .procname = "core_pattern",
390 .data = core_pattern,
391 .maxlen = CORENAME_MAX_SIZE,
392 .mode = 0644,
393 .proc_handler = &proc_dostring,
394 },
395 {
396 .procname = "core_pipe_limit",
397 .data = &core_pipe_limit,
398 .maxlen = sizeof(unsigned int),
399 .mode = 0644,
400 .proc_handler = &proc_dointvec,
401 },
402 #ifdef CONFIG_PROC_SYSCTL
403 {
404 .procname = "tainted",
405 .maxlen = sizeof(long),
406 .mode = 0644,
407 .proc_handler = &proc_taint,
408 },
409 #endif
410 #ifdef CONFIG_LATENCYTOP
411 {
412 .procname = "latencytop",
413 .data = &latencytop_enabled,
414 .maxlen = sizeof(int),
415 .mode = 0644,
416 .proc_handler = &proc_dointvec,
417 },
418 #endif
419 #ifdef CONFIG_BLK_DEV_INITRD
420 {
421 .procname = "real-root-dev",
422 .data = &real_root_dev,
423 .maxlen = sizeof(int),
424 .mode = 0644,
425 .proc_handler = &proc_dointvec,
426 },
427 #endif
428 {
429 .procname = "print-fatal-signals",
430 .data = &print_fatal_signals,
431 .maxlen = sizeof(int),
432 .mode = 0644,
433 .proc_handler = &proc_dointvec,
434 },
435 #ifdef CONFIG_SPARC
436 {
437 .procname = "reboot-cmd",
438 .data = reboot_command,
439 .maxlen = 256,
440 .mode = 0644,
441 .proc_handler = &proc_dostring,
442 },
443 {
444 .procname = "stop-a",
445 .data = &stop_a_enabled,
446 .maxlen = sizeof (int),
447 .mode = 0644,
448 .proc_handler = &proc_dointvec,
449 },
450 {
451 .procname = "scons-poweroff",
452 .data = &scons_pwroff,
453 .maxlen = sizeof (int),
454 .mode = 0644,
455 .proc_handler = &proc_dointvec,
456 },
457 #endif
458 #ifdef CONFIG_SPARC64
459 {
460 .procname = "tsb-ratio",
461 .data = &sysctl_tsb_ratio,
462 .maxlen = sizeof (int),
463 .mode = 0644,
464 .proc_handler = &proc_dointvec,
465 },
466 #endif
467 #ifdef __hppa__
468 {
469 .procname = "soft-power",
470 .data = &pwrsw_enabled,
471 .maxlen = sizeof (int),
472 .mode = 0644,
473 .proc_handler = &proc_dointvec,
474 },
475 {
476 .procname = "unaligned-trap",
477 .data = &unaligned_enabled,
478 .maxlen = sizeof (int),
479 .mode = 0644,
480 .proc_handler = &proc_dointvec,
481 },
482 #endif
483 {
484 .procname = "ctrl-alt-del",
485 .data = &C_A_D,
486 .maxlen = sizeof(int),
487 .mode = 0644,
488 .proc_handler = &proc_dointvec,
489 },
490 #ifdef CONFIG_FUNCTION_TRACER
491 {
492 .procname = "ftrace_enabled",
493 .data = &ftrace_enabled,
494 .maxlen = sizeof(int),
495 .mode = 0644,
496 .proc_handler = &ftrace_enable_sysctl,
497 },
498 #endif
499 #ifdef CONFIG_STACK_TRACER
500 {
501 .procname = "stack_tracer_enabled",
502 .data = &stack_tracer_enabled,
503 .maxlen = sizeof(int),
504 .mode = 0644,
505 .proc_handler = &stack_trace_sysctl,
506 },
507 #endif
508 #ifdef CONFIG_TRACING
509 {
510 .procname = "ftrace_dump_on_oops",
511 .data = &ftrace_dump_on_oops,
512 .maxlen = sizeof(int),
513 .mode = 0644,
514 .proc_handler = &proc_dointvec,
515 },
516 #endif
517 #ifdef CONFIG_MODULES
518 {
519 .procname = "modprobe",
520 .data = &modprobe_path,
521 .maxlen = KMOD_PATH_LEN,
522 .mode = 0644,
523 .proc_handler = &proc_dostring,
524 },
525 {
526 .procname = "modules_disabled",
527 .data = &modules_disabled,
528 .maxlen = sizeof(int),
529 .mode = 0644,
530 /* only handle a transition from default "0" to "1" */
531 .proc_handler = &proc_dointvec_minmax,
532 .extra1 = &one,
533 .extra2 = &one,
534 },
535 #endif
536 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
537 {
538 .procname = "hotplug",
539 .data = &uevent_helper,
540 .maxlen = UEVENT_HELPER_PATH_LEN,
541 .mode = 0644,
542 .proc_handler = &proc_dostring,
543 },
544 #endif
545 #ifdef CONFIG_CHR_DEV_SG
546 {
547 .procname = "sg-big-buff",
548 .data = &sg_big_buff,
549 .maxlen = sizeof (int),
550 .mode = 0444,
551 .proc_handler = &proc_dointvec,
552 },
553 #endif
554 #ifdef CONFIG_BSD_PROCESS_ACCT
555 {
556 .procname = "acct",
557 .data = &acct_parm,
558 .maxlen = 3*sizeof(int),
559 .mode = 0644,
560 .proc_handler = &proc_dointvec,
561 },
562 #endif
563 #ifdef CONFIG_MAGIC_SYSRQ
564 {
565 .procname = "sysrq",
566 .data = &__sysrq_enabled,
567 .maxlen = sizeof (int),
568 .mode = 0644,
569 .proc_handler = &proc_dointvec,
570 },
571 #endif
572 #ifdef CONFIG_PROC_SYSCTL
573 {
574 .procname = "cad_pid",
575 .data = NULL,
576 .maxlen = sizeof (int),
577 .mode = 0600,
578 .proc_handler = &proc_do_cad_pid,
579 },
580 #endif
581 {
582 .procname = "threads-max",
583 .data = &max_threads,
584 .maxlen = sizeof(int),
585 .mode = 0644,
586 .proc_handler = &proc_dointvec,
587 },
588 {
589 .procname = "random",
590 .mode = 0555,
591 .child = random_table,
592 },
593 {
594 .procname = "overflowuid",
595 .data = &overflowuid,
596 .maxlen = sizeof(int),
597 .mode = 0644,
598 .proc_handler = &proc_dointvec_minmax,
599 .extra1 = &minolduid,
600 .extra2 = &maxolduid,
601 },
602 {
603 .procname = "overflowgid",
604 .data = &overflowgid,
605 .maxlen = sizeof(int),
606 .mode = 0644,
607 .proc_handler = &proc_dointvec_minmax,
608 .extra1 = &minolduid,
609 .extra2 = &maxolduid,
610 },
611 #ifdef CONFIG_S390
612 #ifdef CONFIG_MATHEMU
613 {
614 .procname = "ieee_emulation_warnings",
615 .data = &sysctl_ieee_emulation_warnings,
616 .maxlen = sizeof(int),
617 .mode = 0644,
618 .proc_handler = &proc_dointvec,
619 },
620 #endif
621 {
622 .procname = "userprocess_debug",
623 .data = &sysctl_userprocess_debug,
624 .maxlen = sizeof(int),
625 .mode = 0644,
626 .proc_handler = &proc_dointvec,
627 },
628 #endif
629 {
630 .procname = "pid_max",
631 .data = &pid_max,
632 .maxlen = sizeof (int),
633 .mode = 0644,
634 .proc_handler = &proc_dointvec_minmax,
635 .extra1 = &pid_max_min,
636 .extra2 = &pid_max_max,
637 },
638 {
639 .procname = "panic_on_oops",
640 .data = &panic_on_oops,
641 .maxlen = sizeof(int),
642 .mode = 0644,
643 .proc_handler = &proc_dointvec,
644 },
645 #if defined CONFIG_PRINTK
646 {
647 .procname = "printk",
648 .data = &console_loglevel,
649 .maxlen = 4*sizeof(int),
650 .mode = 0644,
651 .proc_handler = &proc_dointvec,
652 },
653 {
654 .procname = "printk_ratelimit",
655 .data = &printk_ratelimit_state.interval,
656 .maxlen = sizeof(int),
657 .mode = 0644,
658 .proc_handler = &proc_dointvec_jiffies,
659 },
660 {
661 .procname = "printk_ratelimit_burst",
662 .data = &printk_ratelimit_state.burst,
663 .maxlen = sizeof(int),
664 .mode = 0644,
665 .proc_handler = &proc_dointvec,
666 },
667 {
668 .procname = "printk_delay",
669 .data = &printk_delay_msec,
670 .maxlen = sizeof(int),
671 .mode = 0644,
672 .proc_handler = &proc_dointvec_minmax,
673 .extra1 = &zero,
674 .extra2 = &ten_thousand,
675 },
676 #endif
677 {
678 .procname = "ngroups_max",
679 .data = &ngroups_max,
680 .maxlen = sizeof (int),
681 .mode = 0444,
682 .proc_handler = &proc_dointvec,
683 },
684 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
685 {
686 .procname = "unknown_nmi_panic",
687 .data = &unknown_nmi_panic,
688 .maxlen = sizeof (int),
689 .mode = 0644,
690 .proc_handler = &proc_dointvec,
691 },
692 {
693 .procname = "nmi_watchdog",
694 .data = &nmi_watchdog_enabled,
695 .maxlen = sizeof (int),
696 .mode = 0644,
697 .proc_handler = &proc_nmi_enabled,
698 },
699 #endif
700 #if defined(CONFIG_X86)
701 {
702 .procname = "panic_on_unrecovered_nmi",
703 .data = &panic_on_unrecovered_nmi,
704 .maxlen = sizeof(int),
705 .mode = 0644,
706 .proc_handler = &proc_dointvec,
707 },
708 {
709 .procname = "panic_on_io_nmi",
710 .data = &panic_on_io_nmi,
711 .maxlen = sizeof(int),
712 .mode = 0644,
713 .proc_handler = &proc_dointvec,
714 },
715 {
716 .procname = "bootloader_type",
717 .data = &bootloader_type,
718 .maxlen = sizeof (int),
719 .mode = 0444,
720 .proc_handler = &proc_dointvec,
721 },
722 {
723 .procname = "bootloader_version",
724 .data = &bootloader_version,
725 .maxlen = sizeof (int),
726 .mode = 0444,
727 .proc_handler = &proc_dointvec,
728 },
729 {
730 .procname = "kstack_depth_to_print",
731 .data = &kstack_depth_to_print,
732 .maxlen = sizeof(int),
733 .mode = 0644,
734 .proc_handler = &proc_dointvec,
735 },
736 {
737 .procname = "io_delay_type",
738 .data = &io_delay_type,
739 .maxlen = sizeof(int),
740 .mode = 0644,
741 .proc_handler = &proc_dointvec,
742 },
743 #endif
744 #if defined(CONFIG_MMU)
745 {
746 .procname = "randomize_va_space",
747 .data = &randomize_va_space,
748 .maxlen = sizeof(int),
749 .mode = 0644,
750 .proc_handler = &proc_dointvec,
751 },
752 #endif
753 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
754 {
755 .procname = "spin_retry",
756 .data = &spin_retry,
757 .maxlen = sizeof (int),
758 .mode = 0644,
759 .proc_handler = &proc_dointvec,
760 },
761 #endif
762 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
763 {
764 .procname = "acpi_video_flags",
765 .data = &acpi_realmode_flags,
766 .maxlen = sizeof (unsigned long),
767 .mode = 0644,
768 .proc_handler = &proc_doulongvec_minmax,
769 },
770 #endif
771 #ifdef CONFIG_IA64
772 {
773 .procname = "ignore-unaligned-usertrap",
774 .data = &no_unaligned_warning,
775 .maxlen = sizeof (int),
776 .mode = 0644,
777 .proc_handler = &proc_dointvec,
778 },
779 {
780 .procname = "unaligned-dump-stack",
781 .data = &unaligned_dump_stack,
782 .maxlen = sizeof (int),
783 .mode = 0644,
784 .proc_handler = &proc_dointvec,
785 },
786 #endif
787 #ifdef CONFIG_DETECT_SOFTLOCKUP
788 {
789 .procname = "softlockup_panic",
790 .data = &softlockup_panic,
791 .maxlen = sizeof(int),
792 .mode = 0644,
793 .proc_handler = &proc_dointvec_minmax,
794 .extra1 = &zero,
795 .extra2 = &one,
796 },
797 {
798 .procname = "softlockup_thresh",
799 .data = &softlockup_thresh,
800 .maxlen = sizeof(int),
801 .mode = 0644,
802 .proc_handler = &proc_dosoftlockup_thresh,
803 .extra1 = &neg_one,
804 .extra2 = &sixty,
805 },
806 #endif
807 #ifdef CONFIG_DETECT_HUNG_TASK
808 {
809 .procname = "hung_task_panic",
810 .data = &sysctl_hung_task_panic,
811 .maxlen = sizeof(int),
812 .mode = 0644,
813 .proc_handler = &proc_dointvec_minmax,
814 .extra1 = &zero,
815 .extra2 = &one,
816 },
817 {
818 .procname = "hung_task_check_count",
819 .data = &sysctl_hung_task_check_count,
820 .maxlen = sizeof(unsigned long),
821 .mode = 0644,
822 .proc_handler = &proc_doulongvec_minmax,
823 },
824 {
825 .procname = "hung_task_timeout_secs",
826 .data = &sysctl_hung_task_timeout_secs,
827 .maxlen = sizeof(unsigned long),
828 .mode = 0644,
829 .proc_handler = &proc_dohung_task_timeout_secs,
830 },
831 {
832 .procname = "hung_task_warnings",
833 .data = &sysctl_hung_task_warnings,
834 .maxlen = sizeof(unsigned long),
835 .mode = 0644,
836 .proc_handler = &proc_doulongvec_minmax,
837 },
838 #endif
839 #ifdef CONFIG_COMPAT
840 {
841 .procname = "compat-log",
842 .data = &compat_log,
843 .maxlen = sizeof (int),
844 .mode = 0644,
845 .proc_handler = &proc_dointvec,
846 },
847 #endif
848 #ifdef CONFIG_RT_MUTEXES
849 {
850 .procname = "max_lock_depth",
851 .data = &max_lock_depth,
852 .maxlen = sizeof(int),
853 .mode = 0644,
854 .proc_handler = &proc_dointvec,
855 },
856 #endif
857 {
858 .procname = "poweroff_cmd",
859 .data = &poweroff_cmd,
860 .maxlen = POWEROFF_CMD_PATH_LEN,
861 .mode = 0644,
862 .proc_handler = &proc_dostring,
863 },
864 #ifdef CONFIG_KEYS
865 {
866 .procname = "keys",
867 .mode = 0555,
868 .child = key_sysctls,
869 },
870 #endif
871 #ifdef CONFIG_RCU_TORTURE_TEST
872 {
873 .procname = "rcutorture_runnable",
874 .data = &rcutorture_runnable,
875 .maxlen = sizeof(int),
876 .mode = 0644,
877 .proc_handler = &proc_dointvec,
878 },
879 #endif
880 #ifdef CONFIG_SLOW_WORK
881 {
882 .procname = "slow-work",
883 .mode = 0555,
884 .child = slow_work_sysctls,
885 },
886 #endif
887 #ifdef CONFIG_PERF_EVENTS
888 {
889 .procname = "perf_event_paranoid",
890 .data = &sysctl_perf_event_paranoid,
891 .maxlen = sizeof(sysctl_perf_event_paranoid),
892 .mode = 0644,
893 .proc_handler = &proc_dointvec,
894 },
895 {
896 .procname = "perf_event_mlock_kb",
897 .data = &sysctl_perf_event_mlock,
898 .maxlen = sizeof(sysctl_perf_event_mlock),
899 .mode = 0644,
900 .proc_handler = &proc_dointvec,
901 },
902 {
903 .procname = "perf_event_max_sample_rate",
904 .data = &sysctl_perf_event_sample_rate,
905 .maxlen = sizeof(sysctl_perf_event_sample_rate),
906 .mode = 0644,
907 .proc_handler = &proc_dointvec,
908 },
909 #endif
910 #ifdef CONFIG_KMEMCHECK
911 {
912 .procname = "kmemcheck",
913 .data = &kmemcheck_enabled,
914 .maxlen = sizeof(int),
915 .mode = 0644,
916 .proc_handler = &proc_dointvec,
917 },
918 #endif
919 #ifdef CONFIG_BLOCK
920 {
921 .procname = "blk_iopoll",
922 .data = &blk_iopoll_enabled,
923 .maxlen = sizeof(int),
924 .mode = 0644,
925 .proc_handler = &proc_dointvec,
926 },
927 #endif
928 /*
929 * NOTE: do not add new entries to this table unless you have read
930 * Documentation/sysctl/ctl_unnumbered.txt
931 */
932 { }
933 };
934
935 static struct ctl_table vm_table[] = {
936 {
937 .procname = "overcommit_memory",
938 .data = &sysctl_overcommit_memory,
939 .maxlen = sizeof(sysctl_overcommit_memory),
940 .mode = 0644,
941 .proc_handler = &proc_dointvec,
942 },
943 {
944 .procname = "panic_on_oom",
945 .data = &sysctl_panic_on_oom,
946 .maxlen = sizeof(sysctl_panic_on_oom),
947 .mode = 0644,
948 .proc_handler = &proc_dointvec,
949 },
950 {
951 .procname = "oom_kill_allocating_task",
952 .data = &sysctl_oom_kill_allocating_task,
953 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
954 .mode = 0644,
955 .proc_handler = &proc_dointvec,
956 },
957 {
958 .procname = "oom_dump_tasks",
959 .data = &sysctl_oom_dump_tasks,
960 .maxlen = sizeof(sysctl_oom_dump_tasks),
961 .mode = 0644,
962 .proc_handler = &proc_dointvec,
963 },
964 {
965 .procname = "overcommit_ratio",
966 .data = &sysctl_overcommit_ratio,
967 .maxlen = sizeof(sysctl_overcommit_ratio),
968 .mode = 0644,
969 .proc_handler = &proc_dointvec,
970 },
971 {
972 .procname = "page-cluster",
973 .data = &page_cluster,
974 .maxlen = sizeof(int),
975 .mode = 0644,
976 .proc_handler = &proc_dointvec,
977 },
978 {
979 .procname = "dirty_background_ratio",
980 .data = &dirty_background_ratio,
981 .maxlen = sizeof(dirty_background_ratio),
982 .mode = 0644,
983 .proc_handler = &dirty_background_ratio_handler,
984 .extra1 = &zero,
985 .extra2 = &one_hundred,
986 },
987 {
988 .procname = "dirty_background_bytes",
989 .data = &dirty_background_bytes,
990 .maxlen = sizeof(dirty_background_bytes),
991 .mode = 0644,
992 .proc_handler = &dirty_background_bytes_handler,
993 .extra1 = &one_ul,
994 },
995 {
996 .procname = "dirty_ratio",
997 .data = &vm_dirty_ratio,
998 .maxlen = sizeof(vm_dirty_ratio),
999 .mode = 0644,
1000 .proc_handler = &dirty_ratio_handler,
1001 .extra1 = &zero,
1002 .extra2 = &one_hundred,
1003 },
1004 {
1005 .procname = "dirty_bytes",
1006 .data = &vm_dirty_bytes,
1007 .maxlen = sizeof(vm_dirty_bytes),
1008 .mode = 0644,
1009 .proc_handler = &dirty_bytes_handler,
1010 .extra1 = &dirty_bytes_min,
1011 },
1012 {
1013 .procname = "dirty_writeback_centisecs",
1014 .data = &dirty_writeback_interval,
1015 .maxlen = sizeof(dirty_writeback_interval),
1016 .mode = 0644,
1017 .proc_handler = &dirty_writeback_centisecs_handler,
1018 },
1019 {
1020 .procname = "dirty_expire_centisecs",
1021 .data = &dirty_expire_interval,
1022 .maxlen = sizeof(dirty_expire_interval),
1023 .mode = 0644,
1024 .proc_handler = &proc_dointvec,
1025 },
1026 {
1027 .procname = "nr_pdflush_threads",
1028 .data = &nr_pdflush_threads,
1029 .maxlen = sizeof nr_pdflush_threads,
1030 .mode = 0444 /* read-only*/,
1031 .proc_handler = &proc_dointvec,
1032 },
1033 {
1034 .procname = "swappiness",
1035 .data = &vm_swappiness,
1036 .maxlen = sizeof(vm_swappiness),
1037 .mode = 0644,
1038 .proc_handler = &proc_dointvec_minmax,
1039 .extra1 = &zero,
1040 .extra2 = &one_hundred,
1041 },
1042 #ifdef CONFIG_HUGETLB_PAGE
1043 {
1044 .procname = "nr_hugepages",
1045 .data = NULL,
1046 .maxlen = sizeof(unsigned long),
1047 .mode = 0644,
1048 .proc_handler = &hugetlb_sysctl_handler,
1049 .extra1 = (void *)&hugetlb_zero,
1050 .extra2 = (void *)&hugetlb_infinity,
1051 },
1052 {
1053 .procname = "hugetlb_shm_group",
1054 .data = &sysctl_hugetlb_shm_group,
1055 .maxlen = sizeof(gid_t),
1056 .mode = 0644,
1057 .proc_handler = &proc_dointvec,
1058 },
1059 {
1060 .procname = "hugepages_treat_as_movable",
1061 .data = &hugepages_treat_as_movable,
1062 .maxlen = sizeof(int),
1063 .mode = 0644,
1064 .proc_handler = &hugetlb_treat_movable_handler,
1065 },
1066 {
1067 .procname = "nr_overcommit_hugepages",
1068 .data = NULL,
1069 .maxlen = sizeof(unsigned long),
1070 .mode = 0644,
1071 .proc_handler = &hugetlb_overcommit_handler,
1072 .extra1 = (void *)&hugetlb_zero,
1073 .extra2 = (void *)&hugetlb_infinity,
1074 },
1075 #endif
1076 {
1077 .procname = "lowmem_reserve_ratio",
1078 .data = &sysctl_lowmem_reserve_ratio,
1079 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1080 .mode = 0644,
1081 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
1082 },
1083 {
1084 .procname = "drop_caches",
1085 .data = &sysctl_drop_caches,
1086 .maxlen = sizeof(int),
1087 .mode = 0644,
1088 .proc_handler = drop_caches_sysctl_handler,
1089 },
1090 {
1091 .procname = "min_free_kbytes",
1092 .data = &min_free_kbytes,
1093 .maxlen = sizeof(min_free_kbytes),
1094 .mode = 0644,
1095 .proc_handler = &min_free_kbytes_sysctl_handler,
1096 .extra1 = &zero,
1097 },
1098 {
1099 .procname = "percpu_pagelist_fraction",
1100 .data = &percpu_pagelist_fraction,
1101 .maxlen = sizeof(percpu_pagelist_fraction),
1102 .mode = 0644,
1103 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
1104 .extra1 = &min_percpu_pagelist_fract,
1105 },
1106 #ifdef CONFIG_MMU
1107 {
1108 .procname = "max_map_count",
1109 .data = &sysctl_max_map_count,
1110 .maxlen = sizeof(sysctl_max_map_count),
1111 .mode = 0644,
1112 .proc_handler = &proc_dointvec
1113 },
1114 #else
1115 {
1116 .procname = "nr_trim_pages",
1117 .data = &sysctl_nr_trim_pages,
1118 .maxlen = sizeof(sysctl_nr_trim_pages),
1119 .mode = 0644,
1120 .proc_handler = &proc_dointvec_minmax,
1121 .extra1 = &zero,
1122 },
1123 #endif
1124 {
1125 .procname = "laptop_mode",
1126 .data = &laptop_mode,
1127 .maxlen = sizeof(laptop_mode),
1128 .mode = 0644,
1129 .proc_handler = &proc_dointvec_jiffies,
1130 },
1131 {
1132 .procname = "block_dump",
1133 .data = &block_dump,
1134 .maxlen = sizeof(block_dump),
1135 .mode = 0644,
1136 .proc_handler = &proc_dointvec,
1137 .extra1 = &zero,
1138 },
1139 {
1140 .procname = "vfs_cache_pressure",
1141 .data = &sysctl_vfs_cache_pressure,
1142 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1143 .mode = 0644,
1144 .proc_handler = &proc_dointvec,
1145 .extra1 = &zero,
1146 },
1147 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1148 {
1149 .procname = "legacy_va_layout",
1150 .data = &sysctl_legacy_va_layout,
1151 .maxlen = sizeof(sysctl_legacy_va_layout),
1152 .mode = 0644,
1153 .proc_handler = &proc_dointvec,
1154 .extra1 = &zero,
1155 },
1156 #endif
1157 #ifdef CONFIG_NUMA
1158 {
1159 .procname = "zone_reclaim_mode",
1160 .data = &zone_reclaim_mode,
1161 .maxlen = sizeof(zone_reclaim_mode),
1162 .mode = 0644,
1163 .proc_handler = &proc_dointvec,
1164 .extra1 = &zero,
1165 },
1166 {
1167 .procname = "min_unmapped_ratio",
1168 .data = &sysctl_min_unmapped_ratio,
1169 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1170 .mode = 0644,
1171 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
1172 .extra1 = &zero,
1173 .extra2 = &one_hundred,
1174 },
1175 {
1176 .procname = "min_slab_ratio",
1177 .data = &sysctl_min_slab_ratio,
1178 .maxlen = sizeof(sysctl_min_slab_ratio),
1179 .mode = 0644,
1180 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1181 .extra1 = &zero,
1182 .extra2 = &one_hundred,
1183 },
1184 #endif
1185 #ifdef CONFIG_SMP
1186 {
1187 .procname = "stat_interval",
1188 .data = &sysctl_stat_interval,
1189 .maxlen = sizeof(sysctl_stat_interval),
1190 .mode = 0644,
1191 .proc_handler = &proc_dointvec_jiffies,
1192 },
1193 #endif
1194 {
1195 .procname = "mmap_min_addr",
1196 .data = &dac_mmap_min_addr,
1197 .maxlen = sizeof(unsigned long),
1198 .mode = 0644,
1199 .proc_handler = &mmap_min_addr_handler,
1200 },
1201 #ifdef CONFIG_NUMA
1202 {
1203 .procname = "numa_zonelist_order",
1204 .data = &numa_zonelist_order,
1205 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1206 .mode = 0644,
1207 .proc_handler = &numa_zonelist_order_handler,
1208 },
1209 #endif
1210 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1211 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1212 {
1213 .procname = "vdso_enabled",
1214 .data = &vdso_enabled,
1215 .maxlen = sizeof(vdso_enabled),
1216 .mode = 0644,
1217 .proc_handler = &proc_dointvec,
1218 .extra1 = &zero,
1219 },
1220 #endif
1221 #ifdef CONFIG_HIGHMEM
1222 {
1223 .procname = "highmem_is_dirtyable",
1224 .data = &vm_highmem_is_dirtyable,
1225 .maxlen = sizeof(vm_highmem_is_dirtyable),
1226 .mode = 0644,
1227 .proc_handler = &proc_dointvec_minmax,
1228 .extra1 = &zero,
1229 .extra2 = &one,
1230 },
1231 #endif
1232 {
1233 .procname = "scan_unevictable_pages",
1234 .data = &scan_unevictable_pages,
1235 .maxlen = sizeof(scan_unevictable_pages),
1236 .mode = 0644,
1237 .proc_handler = &scan_unevictable_handler,
1238 },
1239 #ifdef CONFIG_MEMORY_FAILURE
1240 {
1241 .procname = "memory_failure_early_kill",
1242 .data = &sysctl_memory_failure_early_kill,
1243 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1244 .mode = 0644,
1245 .proc_handler = &proc_dointvec_minmax,
1246 .extra1 = &zero,
1247 .extra2 = &one,
1248 },
1249 {
1250 .procname = "memory_failure_recovery",
1251 .data = &sysctl_memory_failure_recovery,
1252 .maxlen = sizeof(sysctl_memory_failure_recovery),
1253 .mode = 0644,
1254 .proc_handler = &proc_dointvec_minmax,
1255 .extra1 = &zero,
1256 .extra2 = &one,
1257 },
1258 #endif
1259
1260 /*
1261 * NOTE: do not add new entries to this table unless you have read
1262 * Documentation/sysctl/ctl_unnumbered.txt
1263 */
1264 { }
1265 };
1266
1267 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1268 static struct ctl_table binfmt_misc_table[] = {
1269 { }
1270 };
1271 #endif
1272
1273 static struct ctl_table fs_table[] = {
1274 {
1275 .procname = "inode-nr",
1276 .data = &inodes_stat,
1277 .maxlen = 2*sizeof(int),
1278 .mode = 0444,
1279 .proc_handler = &proc_dointvec,
1280 },
1281 {
1282 .procname = "inode-state",
1283 .data = &inodes_stat,
1284 .maxlen = 7*sizeof(int),
1285 .mode = 0444,
1286 .proc_handler = &proc_dointvec,
1287 },
1288 {
1289 .procname = "file-nr",
1290 .data = &files_stat,
1291 .maxlen = 3*sizeof(int),
1292 .mode = 0444,
1293 .proc_handler = &proc_nr_files,
1294 },
1295 {
1296 .procname = "file-max",
1297 .data = &files_stat.max_files,
1298 .maxlen = sizeof(int),
1299 .mode = 0644,
1300 .proc_handler = &proc_dointvec,
1301 },
1302 {
1303 .procname = "nr_open",
1304 .data = &sysctl_nr_open,
1305 .maxlen = sizeof(int),
1306 .mode = 0644,
1307 .proc_handler = &proc_dointvec_minmax,
1308 .extra1 = &sysctl_nr_open_min,
1309 .extra2 = &sysctl_nr_open_max,
1310 },
1311 {
1312 .procname = "dentry-state",
1313 .data = &dentry_stat,
1314 .maxlen = 6*sizeof(int),
1315 .mode = 0444,
1316 .proc_handler = &proc_dointvec,
1317 },
1318 {
1319 .procname = "overflowuid",
1320 .data = &fs_overflowuid,
1321 .maxlen = sizeof(int),
1322 .mode = 0644,
1323 .proc_handler = &proc_dointvec_minmax,
1324 .extra1 = &minolduid,
1325 .extra2 = &maxolduid,
1326 },
1327 {
1328 .procname = "overflowgid",
1329 .data = &fs_overflowgid,
1330 .maxlen = sizeof(int),
1331 .mode = 0644,
1332 .proc_handler = &proc_dointvec_minmax,
1333 .extra1 = &minolduid,
1334 .extra2 = &maxolduid,
1335 },
1336 #ifdef CONFIG_FILE_LOCKING
1337 {
1338 .procname = "leases-enable",
1339 .data = &leases_enable,
1340 .maxlen = sizeof(int),
1341 .mode = 0644,
1342 .proc_handler = &proc_dointvec,
1343 },
1344 #endif
1345 #ifdef CONFIG_DNOTIFY
1346 {
1347 .procname = "dir-notify-enable",
1348 .data = &dir_notify_enable,
1349 .maxlen = sizeof(int),
1350 .mode = 0644,
1351 .proc_handler = &proc_dointvec,
1352 },
1353 #endif
1354 #ifdef CONFIG_MMU
1355 #ifdef CONFIG_FILE_LOCKING
1356 {
1357 .procname = "lease-break-time",
1358 .data = &lease_break_time,
1359 .maxlen = sizeof(int),
1360 .mode = 0644,
1361 .proc_handler = &proc_dointvec,
1362 },
1363 #endif
1364 #ifdef CONFIG_AIO
1365 {
1366 .procname = "aio-nr",
1367 .data = &aio_nr,
1368 .maxlen = sizeof(aio_nr),
1369 .mode = 0444,
1370 .proc_handler = &proc_doulongvec_minmax,
1371 },
1372 {
1373 .procname = "aio-max-nr",
1374 .data = &aio_max_nr,
1375 .maxlen = sizeof(aio_max_nr),
1376 .mode = 0644,
1377 .proc_handler = &proc_doulongvec_minmax,
1378 },
1379 #endif /* CONFIG_AIO */
1380 #ifdef CONFIG_INOTIFY_USER
1381 {
1382 .procname = "inotify",
1383 .mode = 0555,
1384 .child = inotify_table,
1385 },
1386 #endif
1387 #ifdef CONFIG_EPOLL
1388 {
1389 .procname = "epoll",
1390 .mode = 0555,
1391 .child = epoll_table,
1392 },
1393 #endif
1394 #endif
1395 {
1396 .procname = "suid_dumpable",
1397 .data = &suid_dumpable,
1398 .maxlen = sizeof(int),
1399 .mode = 0644,
1400 .proc_handler = &proc_dointvec_minmax,
1401 .extra1 = &zero,
1402 .extra2 = &two,
1403 },
1404 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1405 {
1406 .procname = "binfmt_misc",
1407 .mode = 0555,
1408 .child = binfmt_misc_table,
1409 },
1410 #endif
1411 /*
1412 * NOTE: do not add new entries to this table unless you have read
1413 * Documentation/sysctl/ctl_unnumbered.txt
1414 */
1415 { }
1416 };
1417
1418 static struct ctl_table debug_table[] = {
1419 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1420 {
1421 .procname = "exception-trace",
1422 .data = &show_unhandled_signals,
1423 .maxlen = sizeof(int),
1424 .mode = 0644,
1425 .proc_handler = proc_dointvec
1426 },
1427 #endif
1428 { }
1429 };
1430
1431 static struct ctl_table dev_table[] = {
1432 { }
1433 };
1434
1435 static DEFINE_SPINLOCK(sysctl_lock);
1436
1437 /* called under sysctl_lock */
1438 static int use_table(struct ctl_table_header *p)
1439 {
1440 if (unlikely(p->unregistering))
1441 return 0;
1442 p->used++;
1443 return 1;
1444 }
1445
1446 /* called under sysctl_lock */
1447 static void unuse_table(struct ctl_table_header *p)
1448 {
1449 if (!--p->used)
1450 if (unlikely(p->unregistering))
1451 complete(p->unregistering);
1452 }
1453
1454 /* called under sysctl_lock, will reacquire if has to wait */
1455 static void start_unregistering(struct ctl_table_header *p)
1456 {
1457 /*
1458 * if p->used is 0, nobody will ever touch that entry again;
1459 * we'll eliminate all paths to it before dropping sysctl_lock
1460 */
1461 if (unlikely(p->used)) {
1462 struct completion wait;
1463 init_completion(&wait);
1464 p->unregistering = &wait;
1465 spin_unlock(&sysctl_lock);
1466 wait_for_completion(&wait);
1467 spin_lock(&sysctl_lock);
1468 } else {
1469 /* anything non-NULL; we'll never dereference it */
1470 p->unregistering = ERR_PTR(-EINVAL);
1471 }
1472 /*
1473 * do not remove from the list until nobody holds it; walking the
1474 * list in do_sysctl() relies on that.
1475 */
1476 list_del_init(&p->ctl_entry);
1477 }
1478
1479 void sysctl_head_get(struct ctl_table_header *head)
1480 {
1481 spin_lock(&sysctl_lock);
1482 head->count++;
1483 spin_unlock(&sysctl_lock);
1484 }
1485
1486 void sysctl_head_put(struct ctl_table_header *head)
1487 {
1488 spin_lock(&sysctl_lock);
1489 if (!--head->count)
1490 kfree(head);
1491 spin_unlock(&sysctl_lock);
1492 }
1493
1494 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1495 {
1496 if (!head)
1497 BUG();
1498 spin_lock(&sysctl_lock);
1499 if (!use_table(head))
1500 head = ERR_PTR(-ENOENT);
1501 spin_unlock(&sysctl_lock);
1502 return head;
1503 }
1504
1505 void sysctl_head_finish(struct ctl_table_header *head)
1506 {
1507 if (!head)
1508 return;
1509 spin_lock(&sysctl_lock);
1510 unuse_table(head);
1511 spin_unlock(&sysctl_lock);
1512 }
1513
1514 static struct ctl_table_set *
1515 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1516 {
1517 struct ctl_table_set *set = &root->default_set;
1518 if (root->lookup)
1519 set = root->lookup(root, namespaces);
1520 return set;
1521 }
1522
1523 static struct list_head *
1524 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1525 {
1526 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1527 return &set->list;
1528 }
1529
1530 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1531 struct ctl_table_header *prev)
1532 {
1533 struct ctl_table_root *root;
1534 struct list_head *header_list;
1535 struct ctl_table_header *head;
1536 struct list_head *tmp;
1537
1538 spin_lock(&sysctl_lock);
1539 if (prev) {
1540 head = prev;
1541 tmp = &prev->ctl_entry;
1542 unuse_table(prev);
1543 goto next;
1544 }
1545 tmp = &root_table_header.ctl_entry;
1546 for (;;) {
1547 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1548
1549 if (!use_table(head))
1550 goto next;
1551 spin_unlock(&sysctl_lock);
1552 return head;
1553 next:
1554 root = head->root;
1555 tmp = tmp->next;
1556 header_list = lookup_header_list(root, namespaces);
1557 if (tmp != header_list)
1558 continue;
1559
1560 do {
1561 root = list_entry(root->root_list.next,
1562 struct ctl_table_root, root_list);
1563 if (root == &sysctl_table_root)
1564 goto out;
1565 header_list = lookup_header_list(root, namespaces);
1566 } while (list_empty(header_list));
1567 tmp = header_list->next;
1568 }
1569 out:
1570 spin_unlock(&sysctl_lock);
1571 return NULL;
1572 }
1573
1574 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1575 {
1576 return __sysctl_head_next(current->nsproxy, prev);
1577 }
1578
1579 void register_sysctl_root(struct ctl_table_root *root)
1580 {
1581 spin_lock(&sysctl_lock);
1582 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1583 spin_unlock(&sysctl_lock);
1584 }
1585
1586 /*
1587 * sysctl_perm does NOT grant the superuser all rights automatically, because
1588 * some sysctl variables are readonly even to root.
1589 */
1590
1591 static int test_perm(int mode, int op)
1592 {
1593 if (!current_euid())
1594 mode >>= 6;
1595 else if (in_egroup_p(0))
1596 mode >>= 3;
1597 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1598 return 0;
1599 return -EACCES;
1600 }
1601
1602 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1603 {
1604 int error;
1605 int mode;
1606
1607 error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1608 if (error)
1609 return error;
1610
1611 if (root->permissions)
1612 mode = root->permissions(root, current->nsproxy, table);
1613 else
1614 mode = table->mode;
1615
1616 return test_perm(mode, op);
1617 }
1618
1619 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1620 {
1621 for (; table->procname; table++) {
1622 table->parent = parent;
1623 if (table->child)
1624 sysctl_set_parent(table, table->child);
1625 }
1626 }
1627
1628 static __init int sysctl_init(void)
1629 {
1630 sysctl_set_parent(NULL, root_table);
1631 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1632 {
1633 int err;
1634 err = sysctl_check_table(current->nsproxy, root_table);
1635 }
1636 #endif
1637 return 0;
1638 }
1639
1640 core_initcall(sysctl_init);
1641
1642 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1643 struct ctl_table *table)
1644 {
1645 struct ctl_table *p;
1646 const char *s = branch->procname;
1647
1648 /* branch should have named subdirectory as its first element */
1649 if (!s || !branch->child)
1650 return NULL;
1651
1652 /* ... and nothing else */
1653 if (branch[1].procname)
1654 return NULL;
1655
1656 /* table should contain subdirectory with the same name */
1657 for (p = table; p->procname; p++) {
1658 if (!p->child)
1659 continue;
1660 if (p->procname && strcmp(p->procname, s) == 0)
1661 return p;
1662 }
1663 return NULL;
1664 }
1665
1666 /* see if attaching q to p would be an improvement */
1667 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1668 {
1669 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1670 struct ctl_table *next;
1671 int is_better = 0;
1672 int not_in_parent = !p->attached_by;
1673
1674 while ((next = is_branch_in(by, to)) != NULL) {
1675 if (by == q->attached_by)
1676 is_better = 1;
1677 if (to == p->attached_by)
1678 not_in_parent = 1;
1679 by = by->child;
1680 to = next->child;
1681 }
1682
1683 if (is_better && not_in_parent) {
1684 q->attached_by = by;
1685 q->attached_to = to;
1686 q->parent = p;
1687 }
1688 }
1689
1690 /**
1691 * __register_sysctl_paths - register a sysctl hierarchy
1692 * @root: List of sysctl headers to register on
1693 * @namespaces: Data to compute which lists of sysctl entries are visible
1694 * @path: The path to the directory the sysctl table is in.
1695 * @table: the top-level table structure
1696 *
1697 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1698 * array. A completely 0 filled entry terminates the table.
1699 *
1700 * The members of the &struct ctl_table structure are used as follows:
1701 *
1702 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1703 * enter a sysctl file
1704 *
1705 * data - a pointer to data for use by proc_handler
1706 *
1707 * maxlen - the maximum size in bytes of the data
1708 *
1709 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1710 *
1711 * child - a pointer to the child sysctl table if this entry is a directory, or
1712 * %NULL.
1713 *
1714 * proc_handler - the text handler routine (described below)
1715 *
1716 * de - for internal use by the sysctl routines
1717 *
1718 * extra1, extra2 - extra pointers usable by the proc handler routines
1719 *
1720 * Leaf nodes in the sysctl tree will be represented by a single file
1721 * under /proc; non-leaf nodes will be represented by directories.
1722 *
1723 * sysctl(2) can automatically manage read and write requests through
1724 * the sysctl table. The data and maxlen fields of the ctl_table
1725 * struct enable minimal validation of the values being written to be
1726 * performed, and the mode field allows minimal authentication.
1727 *
1728 * There must be a proc_handler routine for any terminal nodes
1729 * mirrored under /proc/sys (non-terminals are handled by a built-in
1730 * directory handler). Several default handlers are available to
1731 * cover common cases -
1732 *
1733 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1734 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1735 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1736 *
1737 * It is the handler's job to read the input buffer from user memory
1738 * and process it. The handler should return 0 on success.
1739 *
1740 * This routine returns %NULL on a failure to register, and a pointer
1741 * to the table header on success.
1742 */
1743 struct ctl_table_header *__register_sysctl_paths(
1744 struct ctl_table_root *root,
1745 struct nsproxy *namespaces,
1746 const struct ctl_path *path, struct ctl_table *table)
1747 {
1748 struct ctl_table_header *header;
1749 struct ctl_table *new, **prevp;
1750 unsigned int n, npath;
1751 struct ctl_table_set *set;
1752
1753 /* Count the path components */
1754 for (npath = 0; path[npath].procname; ++npath)
1755 ;
1756
1757 /*
1758 * For each path component, allocate a 2-element ctl_table array.
1759 * The first array element will be filled with the sysctl entry
1760 * for this, the second will be the sentinel (procname == 0).
1761 *
1762 * We allocate everything in one go so that we don't have to
1763 * worry about freeing additional memory in unregister_sysctl_table.
1764 */
1765 header = kzalloc(sizeof(struct ctl_table_header) +
1766 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1767 if (!header)
1768 return NULL;
1769
1770 new = (struct ctl_table *) (header + 1);
1771
1772 /* Now connect the dots */
1773 prevp = &header->ctl_table;
1774 for (n = 0; n < npath; ++n, ++path) {
1775 /* Copy the procname */
1776 new->procname = path->procname;
1777 new->mode = 0555;
1778
1779 *prevp = new;
1780 prevp = &new->child;
1781
1782 new += 2;
1783 }
1784 *prevp = table;
1785 header->ctl_table_arg = table;
1786
1787 INIT_LIST_HEAD(&header->ctl_entry);
1788 header->used = 0;
1789 header->unregistering = NULL;
1790 header->root = root;
1791 sysctl_set_parent(NULL, header->ctl_table);
1792 header->count = 1;
1793 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1794 if (sysctl_check_table(namespaces, header->ctl_table)) {
1795 kfree(header);
1796 return NULL;
1797 }
1798 #endif
1799 spin_lock(&sysctl_lock);
1800 header->set = lookup_header_set(root, namespaces);
1801 header->attached_by = header->ctl_table;
1802 header->attached_to = root_table;
1803 header->parent = &root_table_header;
1804 for (set = header->set; set; set = set->parent) {
1805 struct ctl_table_header *p;
1806 list_for_each_entry(p, &set->list, ctl_entry) {
1807 if (p->unregistering)
1808 continue;
1809 try_attach(p, header);
1810 }
1811 }
1812 header->parent->count++;
1813 list_add_tail(&header->ctl_entry, &header->set->list);
1814 spin_unlock(&sysctl_lock);
1815
1816 return header;
1817 }
1818
1819 /**
1820 * register_sysctl_table_path - register a sysctl table hierarchy
1821 * @path: The path to the directory the sysctl table is in.
1822 * @table: the top-level table structure
1823 *
1824 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1825 * array. A completely 0 filled entry terminates the table.
1826 *
1827 * See __register_sysctl_paths for more details.
1828 */
1829 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1830 struct ctl_table *table)
1831 {
1832 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1833 path, table);
1834 }
1835
1836 /**
1837 * register_sysctl_table - register a sysctl table hierarchy
1838 * @table: the top-level table structure
1839 *
1840 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1841 * array. A completely 0 filled entry terminates the table.
1842 *
1843 * See register_sysctl_paths for more details.
1844 */
1845 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1846 {
1847 static const struct ctl_path null_path[] = { {} };
1848
1849 return register_sysctl_paths(null_path, table);
1850 }
1851
1852 /**
1853 * unregister_sysctl_table - unregister a sysctl table hierarchy
1854 * @header: the header returned from register_sysctl_table
1855 *
1856 * Unregisters the sysctl table and all children. proc entries may not
1857 * actually be removed until they are no longer used by anyone.
1858 */
1859 void unregister_sysctl_table(struct ctl_table_header * header)
1860 {
1861 might_sleep();
1862
1863 if (header == NULL)
1864 return;
1865
1866 spin_lock(&sysctl_lock);
1867 start_unregistering(header);
1868 if (!--header->parent->count) {
1869 WARN_ON(1);
1870 kfree(header->parent);
1871 }
1872 if (!--header->count)
1873 kfree(header);
1874 spin_unlock(&sysctl_lock);
1875 }
1876
1877 int sysctl_is_seen(struct ctl_table_header *p)
1878 {
1879 struct ctl_table_set *set = p->set;
1880 int res;
1881 spin_lock(&sysctl_lock);
1882 if (p->unregistering)
1883 res = 0;
1884 else if (!set->is_seen)
1885 res = 1;
1886 else
1887 res = set->is_seen(set);
1888 spin_unlock(&sysctl_lock);
1889 return res;
1890 }
1891
1892 void setup_sysctl_set(struct ctl_table_set *p,
1893 struct ctl_table_set *parent,
1894 int (*is_seen)(struct ctl_table_set *))
1895 {
1896 INIT_LIST_HEAD(&p->list);
1897 p->parent = parent ? parent : &sysctl_table_root.default_set;
1898 p->is_seen = is_seen;
1899 }
1900
1901 #else /* !CONFIG_SYSCTL */
1902 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1903 {
1904 return NULL;
1905 }
1906
1907 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1908 struct ctl_table *table)
1909 {
1910 return NULL;
1911 }
1912
1913 void unregister_sysctl_table(struct ctl_table_header * table)
1914 {
1915 }
1916
1917 void setup_sysctl_set(struct ctl_table_set *p,
1918 struct ctl_table_set *parent,
1919 int (*is_seen)(struct ctl_table_set *))
1920 {
1921 }
1922
1923 void sysctl_head_put(struct ctl_table_header *head)
1924 {
1925 }
1926
1927 #endif /* CONFIG_SYSCTL */
1928
1929 /*
1930 * /proc/sys support
1931 */
1932
1933 #ifdef CONFIG_PROC_SYSCTL
1934
1935 static int _proc_do_string(void* data, int maxlen, int write,
1936 void __user *buffer,
1937 size_t *lenp, loff_t *ppos)
1938 {
1939 size_t len;
1940 char __user *p;
1941 char c;
1942
1943 if (!data || !maxlen || !*lenp) {
1944 *lenp = 0;
1945 return 0;
1946 }
1947
1948 if (write) {
1949 len = 0;
1950 p = buffer;
1951 while (len < *lenp) {
1952 if (get_user(c, p++))
1953 return -EFAULT;
1954 if (c == 0 || c == '\n')
1955 break;
1956 len++;
1957 }
1958 if (len >= maxlen)
1959 len = maxlen-1;
1960 if(copy_from_user(data, buffer, len))
1961 return -EFAULT;
1962 ((char *) data)[len] = 0;
1963 *ppos += *lenp;
1964 } else {
1965 len = strlen(data);
1966 if (len > maxlen)
1967 len = maxlen;
1968
1969 if (*ppos > len) {
1970 *lenp = 0;
1971 return 0;
1972 }
1973
1974 data += *ppos;
1975 len -= *ppos;
1976
1977 if (len > *lenp)
1978 len = *lenp;
1979 if (len)
1980 if(copy_to_user(buffer, data, len))
1981 return -EFAULT;
1982 if (len < *lenp) {
1983 if(put_user('\n', ((char __user *) buffer) + len))
1984 return -EFAULT;
1985 len++;
1986 }
1987 *lenp = len;
1988 *ppos += len;
1989 }
1990 return 0;
1991 }
1992
1993 /**
1994 * proc_dostring - read a string sysctl
1995 * @table: the sysctl table
1996 * @write: %TRUE if this is a write to the sysctl file
1997 * @buffer: the user buffer
1998 * @lenp: the size of the user buffer
1999 * @ppos: file position
2000 *
2001 * Reads/writes a string from/to the user buffer. If the kernel
2002 * buffer provided is not large enough to hold the string, the
2003 * string is truncated. The copied string is %NULL-terminated.
2004 * If the string is being read by the user process, it is copied
2005 * and a newline '\n' is added. It is truncated if the buffer is
2006 * not large enough.
2007 *
2008 * Returns 0 on success.
2009 */
2010 int proc_dostring(struct ctl_table *table, int write,
2011 void __user *buffer, size_t *lenp, loff_t *ppos)
2012 {
2013 return _proc_do_string(table->data, table->maxlen, write,
2014 buffer, lenp, ppos);
2015 }
2016
2017
2018 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2019 int *valp,
2020 int write, void *data)
2021 {
2022 if (write) {
2023 *valp = *negp ? -*lvalp : *lvalp;
2024 } else {
2025 int val = *valp;
2026 if (val < 0) {
2027 *negp = -1;
2028 *lvalp = (unsigned long)-val;
2029 } else {
2030 *negp = 0;
2031 *lvalp = (unsigned long)val;
2032 }
2033 }
2034 return 0;
2035 }
2036
2037 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2038 int write, void __user *buffer,
2039 size_t *lenp, loff_t *ppos,
2040 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2041 int write, void *data),
2042 void *data)
2043 {
2044 #define TMPBUFLEN 21
2045 int *i, vleft, first = 1, neg;
2046 unsigned long lval;
2047 size_t left, len;
2048
2049 char buf[TMPBUFLEN], *p;
2050 char __user *s = buffer;
2051
2052 if (!tbl_data || !table->maxlen || !*lenp ||
2053 (*ppos && !write)) {
2054 *lenp = 0;
2055 return 0;
2056 }
2057
2058 i = (int *) tbl_data;
2059 vleft = table->maxlen / sizeof(*i);
2060 left = *lenp;
2061
2062 if (!conv)
2063 conv = do_proc_dointvec_conv;
2064
2065 for (; left && vleft--; i++, first=0) {
2066 if (write) {
2067 while (left) {
2068 char c;
2069 if (get_user(c, s))
2070 return -EFAULT;
2071 if (!isspace(c))
2072 break;
2073 left--;
2074 s++;
2075 }
2076 if (!left)
2077 break;
2078 neg = 0;
2079 len = left;
2080 if (len > sizeof(buf) - 1)
2081 len = sizeof(buf) - 1;
2082 if (copy_from_user(buf, s, len))
2083 return -EFAULT;
2084 buf[len] = 0;
2085 p = buf;
2086 if (*p == '-' && left > 1) {
2087 neg = 1;
2088 p++;
2089 }
2090 if (*p < '0' || *p > '9')
2091 break;
2092
2093 lval = simple_strtoul(p, &p, 0);
2094
2095 len = p-buf;
2096 if ((len < left) && *p && !isspace(*p))
2097 break;
2098 s += len;
2099 left -= len;
2100
2101 if (conv(&neg, &lval, i, 1, data))
2102 break;
2103 } else {
2104 p = buf;
2105 if (!first)
2106 *p++ = '\t';
2107
2108 if (conv(&neg, &lval, i, 0, data))
2109 break;
2110
2111 sprintf(p, "%s%lu", neg ? "-" : "", lval);
2112 len = strlen(buf);
2113 if (len > left)
2114 len = left;
2115 if(copy_to_user(s, buf, len))
2116 return -EFAULT;
2117 left -= len;
2118 s += len;
2119 }
2120 }
2121
2122 if (!write && !first && left) {
2123 if(put_user('\n', s))
2124 return -EFAULT;
2125 left--, s++;
2126 }
2127 if (write) {
2128 while (left) {
2129 char c;
2130 if (get_user(c, s++))
2131 return -EFAULT;
2132 if (!isspace(c))
2133 break;
2134 left--;
2135 }
2136 }
2137 if (write && first)
2138 return -EINVAL;
2139 *lenp -= left;
2140 *ppos += *lenp;
2141 return 0;
2142 #undef TMPBUFLEN
2143 }
2144
2145 static int do_proc_dointvec(struct ctl_table *table, int write,
2146 void __user *buffer, size_t *lenp, loff_t *ppos,
2147 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2148 int write, void *data),
2149 void *data)
2150 {
2151 return __do_proc_dointvec(table->data, table, write,
2152 buffer, lenp, ppos, conv, data);
2153 }
2154
2155 /**
2156 * proc_dointvec - read a vector of integers
2157 * @table: the sysctl table
2158 * @write: %TRUE if this is a write to the sysctl file
2159 * @buffer: the user buffer
2160 * @lenp: the size of the user buffer
2161 * @ppos: file position
2162 *
2163 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2164 * values from/to the user buffer, treated as an ASCII string.
2165 *
2166 * Returns 0 on success.
2167 */
2168 int proc_dointvec(struct ctl_table *table, int write,
2169 void __user *buffer, size_t *lenp, loff_t *ppos)
2170 {
2171 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2172 NULL,NULL);
2173 }
2174
2175 /*
2176 * Taint values can only be increased
2177 * This means we can safely use a temporary.
2178 */
2179 static int proc_taint(struct ctl_table *table, int write,
2180 void __user *buffer, size_t *lenp, loff_t *ppos)
2181 {
2182 struct ctl_table t;
2183 unsigned long tmptaint = get_taint();
2184 int err;
2185
2186 if (write && !capable(CAP_SYS_ADMIN))
2187 return -EPERM;
2188
2189 t = *table;
2190 t.data = &tmptaint;
2191 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2192 if (err < 0)
2193 return err;
2194
2195 if (write) {
2196 /*
2197 * Poor man's atomic or. Not worth adding a primitive
2198 * to everyone's atomic.h for this
2199 */
2200 int i;
2201 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2202 if ((tmptaint >> i) & 1)
2203 add_taint(i);
2204 }
2205 }
2206
2207 return err;
2208 }
2209
2210 struct do_proc_dointvec_minmax_conv_param {
2211 int *min;
2212 int *max;
2213 };
2214
2215 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2216 int *valp,
2217 int write, void *data)
2218 {
2219 struct do_proc_dointvec_minmax_conv_param *param = data;
2220 if (write) {
2221 int val = *negp ? -*lvalp : *lvalp;
2222 if ((param->min && *param->min > val) ||
2223 (param->max && *param->max < val))
2224 return -EINVAL;
2225 *valp = val;
2226 } else {
2227 int val = *valp;
2228 if (val < 0) {
2229 *negp = -1;
2230 *lvalp = (unsigned long)-val;
2231 } else {
2232 *negp = 0;
2233 *lvalp = (unsigned long)val;
2234 }
2235 }
2236 return 0;
2237 }
2238
2239 /**
2240 * proc_dointvec_minmax - read a vector of integers with min/max values
2241 * @table: the sysctl table
2242 * @write: %TRUE if this is a write to the sysctl file
2243 * @buffer: the user buffer
2244 * @lenp: the size of the user buffer
2245 * @ppos: file position
2246 *
2247 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2248 * values from/to the user buffer, treated as an ASCII string.
2249 *
2250 * This routine will ensure the values are within the range specified by
2251 * table->extra1 (min) and table->extra2 (max).
2252 *
2253 * Returns 0 on success.
2254 */
2255 int proc_dointvec_minmax(struct ctl_table *table, int write,
2256 void __user *buffer, size_t *lenp, loff_t *ppos)
2257 {
2258 struct do_proc_dointvec_minmax_conv_param param = {
2259 .min = (int *) table->extra1,
2260 .max = (int *) table->extra2,
2261 };
2262 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2263 do_proc_dointvec_minmax_conv, &param);
2264 }
2265
2266 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2267 void __user *buffer,
2268 size_t *lenp, loff_t *ppos,
2269 unsigned long convmul,
2270 unsigned long convdiv)
2271 {
2272 #define TMPBUFLEN 21
2273 unsigned long *i, *min, *max, val;
2274 int vleft, first=1, neg;
2275 size_t len, left;
2276 char buf[TMPBUFLEN], *p;
2277 char __user *s = buffer;
2278
2279 if (!data || !table->maxlen || !*lenp ||
2280 (*ppos && !write)) {
2281 *lenp = 0;
2282 return 0;
2283 }
2284
2285 i = (unsigned long *) data;
2286 min = (unsigned long *) table->extra1;
2287 max = (unsigned long *) table->extra2;
2288 vleft = table->maxlen / sizeof(unsigned long);
2289 left = *lenp;
2290
2291 for (; left && vleft--; i++, min++, max++, first=0) {
2292 if (write) {
2293 while (left) {
2294 char c;
2295 if (get_user(c, s))
2296 return -EFAULT;
2297 if (!isspace(c))
2298 break;
2299 left--;
2300 s++;
2301 }
2302 if (!left)
2303 break;
2304 neg = 0;
2305 len = left;
2306 if (len > TMPBUFLEN-1)
2307 len = TMPBUFLEN-1;
2308 if (copy_from_user(buf, s, len))
2309 return -EFAULT;
2310 buf[len] = 0;
2311 p = buf;
2312 if (*p == '-' && left > 1) {
2313 neg = 1;
2314 p++;
2315 }
2316 if (*p < '0' || *p > '9')
2317 break;
2318 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2319 len = p-buf;
2320 if ((len < left) && *p && !isspace(*p))
2321 break;
2322 if (neg)
2323 val = -val;
2324 s += len;
2325 left -= len;
2326
2327 if(neg)
2328 continue;
2329 if ((min && val < *min) || (max && val > *max))
2330 continue;
2331 *i = val;
2332 } else {
2333 p = buf;
2334 if (!first)
2335 *p++ = '\t';
2336 sprintf(p, "%lu", convdiv * (*i) / convmul);
2337 len = strlen(buf);
2338 if (len > left)
2339 len = left;
2340 if(copy_to_user(s, buf, len))
2341 return -EFAULT;
2342 left -= len;
2343 s += len;
2344 }
2345 }
2346
2347 if (!write && !first && left) {
2348 if(put_user('\n', s))
2349 return -EFAULT;
2350 left--, s++;
2351 }
2352 if (write) {
2353 while (left) {
2354 char c;
2355 if (get_user(c, s++))
2356 return -EFAULT;
2357 if (!isspace(c))
2358 break;
2359 left--;
2360 }
2361 }
2362 if (write && first)
2363 return -EINVAL;
2364 *lenp -= left;
2365 *ppos += *lenp;
2366 return 0;
2367 #undef TMPBUFLEN
2368 }
2369
2370 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2371 void __user *buffer,
2372 size_t *lenp, loff_t *ppos,
2373 unsigned long convmul,
2374 unsigned long convdiv)
2375 {
2376 return __do_proc_doulongvec_minmax(table->data, table, write,
2377 buffer, lenp, ppos, convmul, convdiv);
2378 }
2379
2380 /**
2381 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2382 * @table: the sysctl table
2383 * @write: %TRUE if this is a write to the sysctl file
2384 * @buffer: the user buffer
2385 * @lenp: the size of the user buffer
2386 * @ppos: file position
2387 *
2388 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2389 * values from/to the user buffer, treated as an ASCII string.
2390 *
2391 * This routine will ensure the values are within the range specified by
2392 * table->extra1 (min) and table->extra2 (max).
2393 *
2394 * Returns 0 on success.
2395 */
2396 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2397 void __user *buffer, size_t *lenp, loff_t *ppos)
2398 {
2399 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2400 }
2401
2402 /**
2403 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2404 * @table: the sysctl table
2405 * @write: %TRUE if this is a write to the sysctl file
2406 * @buffer: the user buffer
2407 * @lenp: the size of the user buffer
2408 * @ppos: file position
2409 *
2410 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2411 * values from/to the user buffer, treated as an ASCII string. The values
2412 * are treated as milliseconds, and converted to jiffies when they are stored.
2413 *
2414 * This routine will ensure the values are within the range specified by
2415 * table->extra1 (min) and table->extra2 (max).
2416 *
2417 * Returns 0 on success.
2418 */
2419 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2420 void __user *buffer,
2421 size_t *lenp, loff_t *ppos)
2422 {
2423 return do_proc_doulongvec_minmax(table, write, buffer,
2424 lenp, ppos, HZ, 1000l);
2425 }
2426
2427
2428 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2429 int *valp,
2430 int write, void *data)
2431 {
2432 if (write) {
2433 if (*lvalp > LONG_MAX / HZ)
2434 return 1;
2435 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2436 } else {
2437 int val = *valp;
2438 unsigned long lval;
2439 if (val < 0) {
2440 *negp = -1;
2441 lval = (unsigned long)-val;
2442 } else {
2443 *negp = 0;
2444 lval = (unsigned long)val;
2445 }
2446 *lvalp = lval / HZ;
2447 }
2448 return 0;
2449 }
2450
2451 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2452 int *valp,
2453 int write, void *data)
2454 {
2455 if (write) {
2456 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2457 return 1;
2458 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2459 } else {
2460 int val = *valp;
2461 unsigned long lval;
2462 if (val < 0) {
2463 *negp = -1;
2464 lval = (unsigned long)-val;
2465 } else {
2466 *negp = 0;
2467 lval = (unsigned long)val;
2468 }
2469 *lvalp = jiffies_to_clock_t(lval);
2470 }
2471 return 0;
2472 }
2473
2474 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2475 int *valp,
2476 int write, void *data)
2477 {
2478 if (write) {
2479 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2480 } else {
2481 int val = *valp;
2482 unsigned long lval;
2483 if (val < 0) {
2484 *negp = -1;
2485 lval = (unsigned long)-val;
2486 } else {
2487 *negp = 0;
2488 lval = (unsigned long)val;
2489 }
2490 *lvalp = jiffies_to_msecs(lval);
2491 }
2492 return 0;
2493 }
2494
2495 /**
2496 * proc_dointvec_jiffies - read a vector of integers as seconds
2497 * @table: the sysctl table
2498 * @write: %TRUE if this is a write to the sysctl file
2499 * @buffer: the user buffer
2500 * @lenp: the size of the user buffer
2501 * @ppos: file position
2502 *
2503 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2504 * values from/to the user buffer, treated as an ASCII string.
2505 * The values read are assumed to be in seconds, and are converted into
2506 * jiffies.
2507 *
2508 * Returns 0 on success.
2509 */
2510 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2511 void __user *buffer, size_t *lenp, loff_t *ppos)
2512 {
2513 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2514 do_proc_dointvec_jiffies_conv,NULL);
2515 }
2516
2517 /**
2518 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2519 * @table: the sysctl table
2520 * @write: %TRUE if this is a write to the sysctl file
2521 * @buffer: the user buffer
2522 * @lenp: the size of the user buffer
2523 * @ppos: pointer to the file position
2524 *
2525 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2526 * values from/to the user buffer, treated as an ASCII string.
2527 * The values read are assumed to be in 1/USER_HZ seconds, and
2528 * are converted into jiffies.
2529 *
2530 * Returns 0 on success.
2531 */
2532 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2533 void __user *buffer, size_t *lenp, loff_t *ppos)
2534 {
2535 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2536 do_proc_dointvec_userhz_jiffies_conv,NULL);
2537 }
2538
2539 /**
2540 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2541 * @table: the sysctl table
2542 * @write: %TRUE if this is a write to the sysctl file
2543 * @buffer: the user buffer
2544 * @lenp: the size of the user buffer
2545 * @ppos: file position
2546 * @ppos: the current position in the file
2547 *
2548 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2549 * values from/to the user buffer, treated as an ASCII string.
2550 * The values read are assumed to be in 1/1000 seconds, and
2551 * are converted into jiffies.
2552 *
2553 * Returns 0 on success.
2554 */
2555 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2556 void __user *buffer, size_t *lenp, loff_t *ppos)
2557 {
2558 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2559 do_proc_dointvec_ms_jiffies_conv, NULL);
2560 }
2561
2562 static int proc_do_cad_pid(struct ctl_table *table, int write,
2563 void __user *buffer, size_t *lenp, loff_t *ppos)
2564 {
2565 struct pid *new_pid;
2566 pid_t tmp;
2567 int r;
2568
2569 tmp = pid_vnr(cad_pid);
2570
2571 r = __do_proc_dointvec(&tmp, table, write, buffer,
2572 lenp, ppos, NULL, NULL);
2573 if (r || !write)
2574 return r;
2575
2576 new_pid = find_get_pid(tmp);
2577 if (!new_pid)
2578 return -ESRCH;
2579
2580 put_pid(xchg(&cad_pid, new_pid));
2581 return 0;
2582 }
2583
2584 #else /* CONFIG_PROC_FS */
2585
2586 int proc_dostring(struct ctl_table *table, int write,
2587 void __user *buffer, size_t *lenp, loff_t *ppos)
2588 {
2589 return -ENOSYS;
2590 }
2591
2592 int proc_dointvec(struct ctl_table *table, int write,
2593 void __user *buffer, size_t *lenp, loff_t *ppos)
2594 {
2595 return -ENOSYS;
2596 }
2597
2598 int proc_dointvec_minmax(struct ctl_table *table, int write,
2599 void __user *buffer, size_t *lenp, loff_t *ppos)
2600 {
2601 return -ENOSYS;
2602 }
2603
2604 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2605 void __user *buffer, size_t *lenp, loff_t *ppos)
2606 {
2607 return -ENOSYS;
2608 }
2609
2610 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2611 void __user *buffer, size_t *lenp, loff_t *ppos)
2612 {
2613 return -ENOSYS;
2614 }
2615
2616 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2617 void __user *buffer, size_t *lenp, loff_t *ppos)
2618 {
2619 return -ENOSYS;
2620 }
2621
2622 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2623 void __user *buffer, size_t *lenp, loff_t *ppos)
2624 {
2625 return -ENOSYS;
2626 }
2627
2628 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2629 void __user *buffer,
2630 size_t *lenp, loff_t *ppos)
2631 {
2632 return -ENOSYS;
2633 }
2634
2635
2636 #endif /* CONFIG_PROC_FS */
2637
2638 /*
2639 * No sense putting this after each symbol definition, twice,
2640 * exception granted :-)
2641 */
2642 EXPORT_SYMBOL(proc_dointvec);
2643 EXPORT_SYMBOL(proc_dointvec_jiffies);
2644 EXPORT_SYMBOL(proc_dointvec_minmax);
2645 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2646 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2647 EXPORT_SYMBOL(proc_dostring);
2648 EXPORT_SYMBOL(proc_doulongvec_minmax);
2649 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2650 EXPORT_SYMBOL(register_sysctl_table);
2651 EXPORT_SYMBOL(register_sysctl_paths);
2652 EXPORT_SYMBOL(unregister_sysctl_table);