?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
sched: Fix wake_affine() vs RT tasks
[linux/fpc-iii.git] / kernel / sysctl.c
blobb8bd058c5cd24b99cdbe899528e6794b81ce504a
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/smp_lock.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/hugetlb.h>
40 #include <linux/initrd.h>
41 #include <linux/key.h>
42 #include <linux/times.h>
43 #include <linux/limits.h>
44 #include <linux/dcache.h>
45 #include <linux/syscalls.h>
46 #include <linux/vmstat.h>
47 #include <linux/nfs_fs.h>
48 #include <linux/acpi.h>
49 #include <linux/reboot.h>
50 #include <linux/ftrace.h>
51 #include <linux/slow-work.h>
52 #include <linux/perf_event.h>
53
54 #include <asm/uaccess.h>
55 #include <asm/processor.h>
56
57 #ifdef CONFIG_X86
58 #include <asm/nmi.h>
59 #include <asm/stacktrace.h>
60 #include <asm/io.h>
61 #endif
62
63 static int deprecated_sysctl_warning(struct __sysctl_args *args);
64
65 #if defined(CONFIG_SYSCTL)
66
67 /* External variables not in a header file. */
68 extern int C_A_D;
69 extern int print_fatal_signals;
70 extern int sysctl_overcommit_memory;
71 extern int sysctl_overcommit_ratio;
72 extern int sysctl_panic_on_oom;
73 extern int sysctl_oom_kill_allocating_task;
74 extern int sysctl_oom_dump_tasks;
75 extern int max_threads;
76 extern int core_uses_pid;
77 extern int suid_dumpable;
78 extern char core_pattern[];
79 extern unsigned int core_pipe_limit;
80 extern int pid_max;
81 extern int min_free_kbytes;
82 extern int pid_max_min, pid_max_max;
83 extern int sysctl_drop_caches;
84 extern int percpu_pagelist_fraction;
85 extern int compat_log;
86 extern int latencytop_enabled;
87 extern int sysctl_nr_open_min, sysctl_nr_open_max;
88 #ifndef CONFIG_MMU
89 extern int sysctl_nr_trim_pages;
90 #endif
91 #ifdef CONFIG_RCU_TORTURE_TEST
92 extern int rcutorture_runnable;
93 #endif /* #ifdef CONFIG_RCU_TORTURE_TEST */
94 #ifdef CONFIG_BLOCK
95 extern int blk_iopoll_enabled;
96 #endif
97
98 /* Constants used for minimum and maximum */
99 #ifdef CONFIG_DETECT_SOFTLOCKUP
100 static int sixty = 60;
101 static int neg_one = -1;
102 #endif
103
104 static int zero;
105 static int __maybe_unused one = 1;
106 static int __maybe_unused two = 2;
107 static unsigned long one_ul = 1;
108 static int one_hundred = 100;
109 #ifdef CONFIG_PRINTK
110 static int ten_thousand = 10000;
111 #endif
112
113 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
114 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
115
116 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
117 static int maxolduid = 65535;
118 static int minolduid;
119 static int min_percpu_pagelist_fract = 8;
120
121 static int ngroups_max = NGROUPS_MAX;
122
123 #ifdef CONFIG_MODULES
124 extern char modprobe_path[];
125 extern int modules_disabled;
126 #endif
127 #ifdef CONFIG_CHR_DEV_SG
128 extern int sg_big_buff;
129 #endif
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_BSD_PROCESS_ACCT
153 extern int acct_parm[];
154 #endif
155
156 #ifdef CONFIG_IA64
157 extern int no_unaligned_warning;
158 extern int unaligned_dump_stack;
159 #endif
160
161 #ifdef CONFIG_RT_MUTEXES
162 extern int max_lock_depth;
163 #endif
164
165 #ifdef CONFIG_PROC_SYSCTL
166 static int proc_do_cad_pid(struct ctl_table *table, int write,
167 void __user *buffer, size_t *lenp, loff_t *ppos);
168 static int proc_taint(struct ctl_table *table, int write,
169 void __user *buffer, size_t *lenp, loff_t *ppos);
170 #endif
171
172 static struct ctl_table root_table[];
173 static struct ctl_table_root sysctl_table_root;
174 static struct ctl_table_header root_table_header = {
175 .count = 1,
176 .ctl_table = root_table,
177 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
178 .root = &sysctl_table_root,
179 .set = &sysctl_table_root.default_set,
180 };
181 static struct ctl_table_root sysctl_table_root = {
182 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
183 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
184 };
185
186 static struct ctl_table kern_table[];
187 static struct ctl_table vm_table[];
188 static struct ctl_table fs_table[];
189 static struct ctl_table debug_table[];
190 static struct ctl_table dev_table[];
191 extern struct ctl_table random_table[];
192 #ifdef CONFIG_INOTIFY_USER
193 extern struct ctl_table inotify_table[];
194 #endif
195 #ifdef CONFIG_EPOLL
196 extern struct ctl_table epoll_table[];
197 #endif
198
199 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
200 int sysctl_legacy_va_layout;
201 #endif
202
203 extern int prove_locking;
204 extern int lock_stat;
205
206 /* The default sysctl tables: */
207
208 static struct ctl_table root_table[] = {
209 {
210 .ctl_name = CTL_KERN,
211 .procname = "kernel",
212 .mode = 0555,
213 .child = kern_table,
214 },
215 {
216 .ctl_name = CTL_VM,
217 .procname = "vm",
218 .mode = 0555,
219 .child = vm_table,
220 },
221 {
222 .ctl_name = CTL_FS,
223 .procname = "fs",
224 .mode = 0555,
225 .child = fs_table,
226 },
227 {
228 .ctl_name = CTL_DEBUG,
229 .procname = "debug",
230 .mode = 0555,
231 .child = debug_table,
232 },
233 {
234 .ctl_name = CTL_DEV,
235 .procname = "dev",
236 .mode = 0555,
237 .child = dev_table,
238 },
239 /*
240 * NOTE: do not add new entries to this table unless you have read
241 * Documentation/sysctl/ctl_unnumbered.txt
242 */
243 { .ctl_name = 0 }
244 };
245
246 #ifdef CONFIG_SCHED_DEBUG
247 static int min_sched_granularity_ns = 100000; /* 100 usecs */
248 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
249 static int min_wakeup_granularity_ns; /* 0 usecs */
250 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
251 #endif
252
253 static struct ctl_table kern_table[] = {
254 {
255 .ctl_name = CTL_UNNUMBERED,
256 .procname = "sched_child_runs_first",
257 .data = &sysctl_sched_child_runs_first,
258 .maxlen = sizeof(unsigned int),
259 .mode = 0644,
260 .proc_handler = &proc_dointvec,
261 },
262 #ifdef CONFIG_SCHED_DEBUG
263 {
264 .ctl_name = CTL_UNNUMBERED,
265 .procname = "sched_min_granularity_ns",
266 .data = &sysctl_sched_min_granularity,
267 .maxlen = sizeof(unsigned int),
268 .mode = 0644,
269 .proc_handler = &sched_nr_latency_handler,
270 .strategy = &sysctl_intvec,
271 .extra1 = &min_sched_granularity_ns,
272 .extra2 = &max_sched_granularity_ns,
273 },
274 {
275 .ctl_name = CTL_UNNUMBERED,
276 .procname = "sched_latency_ns",
277 .data = &sysctl_sched_latency,
278 .maxlen = sizeof(unsigned int),
279 .mode = 0644,
280 .proc_handler = &sched_nr_latency_handler,
281 .strategy = &sysctl_intvec,
282 .extra1 = &min_sched_granularity_ns,
283 .extra2 = &max_sched_granularity_ns,
284 },
285 {
286 .ctl_name = CTL_UNNUMBERED,
287 .procname = "sched_wakeup_granularity_ns",
288 .data = &sysctl_sched_wakeup_granularity,
289 .maxlen = sizeof(unsigned int),
290 .mode = 0644,
291 .proc_handler = &proc_dointvec_minmax,
292 .strategy = &sysctl_intvec,
293 .extra1 = &min_wakeup_granularity_ns,
294 .extra2 = &max_wakeup_granularity_ns,
295 },
296 {
297 .ctl_name = CTL_UNNUMBERED,
298 .procname = "sched_shares_ratelimit",
299 .data = &sysctl_sched_shares_ratelimit,
300 .maxlen = sizeof(unsigned int),
301 .mode = 0644,
302 .proc_handler = &proc_dointvec,
303 },
304 {
305 .ctl_name = CTL_UNNUMBERED,
306 .procname = "sched_shares_thresh",
307 .data = &sysctl_sched_shares_thresh,
308 .maxlen = sizeof(unsigned int),
309 .mode = 0644,
310 .proc_handler = &proc_dointvec_minmax,
311 .strategy = &sysctl_intvec,
312 .extra1 = &zero,
313 },
314 {
315 .ctl_name = CTL_UNNUMBERED,
316 .procname = "sched_features",
317 .data = &sysctl_sched_features,
318 .maxlen = sizeof(unsigned int),
319 .mode = 0644,
320 .proc_handler = &proc_dointvec,
321 },
322 {
323 .ctl_name = CTL_UNNUMBERED,
324 .procname = "sched_migration_cost",
325 .data = &sysctl_sched_migration_cost,
326 .maxlen = sizeof(unsigned int),
327 .mode = 0644,
328 .proc_handler = &proc_dointvec,
329 },
330 {
331 .ctl_name = CTL_UNNUMBERED,
332 .procname = "sched_nr_migrate",
333 .data = &sysctl_sched_nr_migrate,
334 .maxlen = sizeof(unsigned int),
335 .mode = 0644,
336 .proc_handler = &proc_dointvec,
337 },
338 {
339 .ctl_name = CTL_UNNUMBERED,
340 .procname = "sched_time_avg",
341 .data = &sysctl_sched_time_avg,
342 .maxlen = sizeof(unsigned int),
343 .mode = 0644,
344 .proc_handler = &proc_dointvec,
345 },
346 {
347 .ctl_name = CTL_UNNUMBERED,
348 .procname = "timer_migration",
349 .data = &sysctl_timer_migration,
350 .maxlen = sizeof(unsigned int),
351 .mode = 0644,
352 .proc_handler = &proc_dointvec_minmax,
353 .strategy = &sysctl_intvec,
354 .extra1 = &zero,
355 .extra2 = &one,
356 },
357 #endif
358 {
359 .ctl_name = CTL_UNNUMBERED,
360 .procname = "sched_rt_period_us",
361 .data = &sysctl_sched_rt_period,
362 .maxlen = sizeof(unsigned int),
363 .mode = 0644,
364 .proc_handler = &sched_rt_handler,
365 },
366 {
367 .ctl_name = CTL_UNNUMBERED,
368 .procname = "sched_rt_runtime_us",
369 .data = &sysctl_sched_rt_runtime,
370 .maxlen = sizeof(int),
371 .mode = 0644,
372 .proc_handler = &sched_rt_handler,
373 },
374 {
375 .ctl_name = CTL_UNNUMBERED,
376 .procname = "sched_compat_yield",
377 .data = &sysctl_sched_compat_yield,
378 .maxlen = sizeof(unsigned int),
379 .mode = 0644,
380 .proc_handler = &proc_dointvec,
381 },
382 #ifdef CONFIG_PROVE_LOCKING
383 {
384 .ctl_name = CTL_UNNUMBERED,
385 .procname = "prove_locking",
386 .data = &prove_locking,
387 .maxlen = sizeof(int),
388 .mode = 0644,
389 .proc_handler = &proc_dointvec,
390 },
391 #endif
392 #ifdef CONFIG_LOCK_STAT
393 {
394 .ctl_name = CTL_UNNUMBERED,
395 .procname = "lock_stat",
396 .data = &lock_stat,
397 .maxlen = sizeof(int),
398 .mode = 0644,
399 .proc_handler = &proc_dointvec,
400 },
401 #endif
402 {
403 .ctl_name = KERN_PANIC,
404 .procname = "panic",
405 .data = &panic_timeout,
406 .maxlen = sizeof(int),
407 .mode = 0644,
408 .proc_handler = &proc_dointvec,
409 },
410 {
411 .ctl_name = KERN_CORE_USES_PID,
412 .procname = "core_uses_pid",
413 .data = &core_uses_pid,
414 .maxlen = sizeof(int),
415 .mode = 0644,
416 .proc_handler = &proc_dointvec,
417 },
418 {
419 .ctl_name = KERN_CORE_PATTERN,
420 .procname = "core_pattern",
421 .data = core_pattern,
422 .maxlen = CORENAME_MAX_SIZE,
423 .mode = 0644,
424 .proc_handler = &proc_dostring,
425 .strategy = &sysctl_string,
426 },
427 {
428 .ctl_name = CTL_UNNUMBERED,
429 .procname = "core_pipe_limit",
430 .data = &core_pipe_limit,
431 .maxlen = sizeof(unsigned int),
432 .mode = 0644,
433 .proc_handler = &proc_dointvec,
434 },
435 #ifdef CONFIG_PROC_SYSCTL
436 {
437 .procname = "tainted",
438 .maxlen = sizeof(long),
439 .mode = 0644,
440 .proc_handler = &proc_taint,
441 },
442 #endif
443 #ifdef CONFIG_LATENCYTOP
444 {
445 .procname = "latencytop",
446 .data = &latencytop_enabled,
447 .maxlen = sizeof(int),
448 .mode = 0644,
449 .proc_handler = &proc_dointvec,
450 },
451 #endif
452 #ifdef CONFIG_BLK_DEV_INITRD
453 {
454 .ctl_name = KERN_REALROOTDEV,
455 .procname = "real-root-dev",
456 .data = &real_root_dev,
457 .maxlen = sizeof(int),
458 .mode = 0644,
459 .proc_handler = &proc_dointvec,
460 },
461 #endif
462 {
463 .ctl_name = CTL_UNNUMBERED,
464 .procname = "print-fatal-signals",
465 .data = &print_fatal_signals,
466 .maxlen = sizeof(int),
467 .mode = 0644,
468 .proc_handler = &proc_dointvec,
469 },
470 #ifdef CONFIG_SPARC
471 {
472 .ctl_name = KERN_SPARC_REBOOT,
473 .procname = "reboot-cmd",
474 .data = reboot_command,
475 .maxlen = 256,
476 .mode = 0644,
477 .proc_handler = &proc_dostring,
478 .strategy = &sysctl_string,
479 },
480 {
481 .ctl_name = KERN_SPARC_STOP_A,
482 .procname = "stop-a",
483 .data = &stop_a_enabled,
484 .maxlen = sizeof (int),
485 .mode = 0644,
486 .proc_handler = &proc_dointvec,
487 },
488 {
489 .ctl_name = KERN_SPARC_SCONS_PWROFF,
490 .procname = "scons-poweroff",
491 .data = &scons_pwroff,
492 .maxlen = sizeof (int),
493 .mode = 0644,
494 .proc_handler = &proc_dointvec,
495 },
496 #endif
497 #ifdef CONFIG_SPARC64
498 {
499 .ctl_name = CTL_UNNUMBERED,
500 .procname = "tsb-ratio",
501 .data = &sysctl_tsb_ratio,
502 .maxlen = sizeof (int),
503 .mode = 0644,
504 .proc_handler = &proc_dointvec,
505 },
506 #endif
507 #ifdef __hppa__
508 {
509 .ctl_name = KERN_HPPA_PWRSW,
510 .procname = "soft-power",
511 .data = &pwrsw_enabled,
512 .maxlen = sizeof (int),
513 .mode = 0644,
514 .proc_handler = &proc_dointvec,
515 },
516 {
517 .ctl_name = KERN_HPPA_UNALIGNED,
518 .procname = "unaligned-trap",
519 .data = &unaligned_enabled,
520 .maxlen = sizeof (int),
521 .mode = 0644,
522 .proc_handler = &proc_dointvec,
523 },
524 #endif
525 {
526 .ctl_name = KERN_CTLALTDEL,
527 .procname = "ctrl-alt-del",
528 .data = &C_A_D,
529 .maxlen = sizeof(int),
530 .mode = 0644,
531 .proc_handler = &proc_dointvec,
532 },
533 #ifdef CONFIG_FUNCTION_TRACER
534 {
535 .ctl_name = CTL_UNNUMBERED,
536 .procname = "ftrace_enabled",
537 .data = &ftrace_enabled,
538 .maxlen = sizeof(int),
539 .mode = 0644,
540 .proc_handler = &ftrace_enable_sysctl,
541 },
542 #endif
543 #ifdef CONFIG_STACK_TRACER
544 {
545 .ctl_name = CTL_UNNUMBERED,
546 .procname = "stack_tracer_enabled",
547 .data = &stack_tracer_enabled,
548 .maxlen = sizeof(int),
549 .mode = 0644,
550 .proc_handler = &stack_trace_sysctl,
551 },
552 #endif
553 #ifdef CONFIG_TRACING
554 {
555 .ctl_name = CTL_UNNUMBERED,
556 .procname = "ftrace_dump_on_oops",
557 .data = &ftrace_dump_on_oops,
558 .maxlen = sizeof(int),
559 .mode = 0644,
560 .proc_handler = &proc_dointvec,
561 },
562 #endif
563 #ifdef CONFIG_MODULES
564 {
565 .ctl_name = KERN_MODPROBE,
566 .procname = "modprobe",
567 .data = &modprobe_path,
568 .maxlen = KMOD_PATH_LEN,
569 .mode = 0644,
570 .proc_handler = &proc_dostring,
571 .strategy = &sysctl_string,
572 },
573 {
574 .ctl_name = CTL_UNNUMBERED,
575 .procname = "modules_disabled",
576 .data = &modules_disabled,
577 .maxlen = sizeof(int),
578 .mode = 0644,
579 /* only handle a transition from default "0" to "1" */
580 .proc_handler = &proc_dointvec_minmax,
581 .extra1 = &one,
582 .extra2 = &one,
583 },
584 #endif
585 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
586 {
587 .ctl_name = KERN_HOTPLUG,
588 .procname = "hotplug",
589 .data = &uevent_helper,
590 .maxlen = UEVENT_HELPER_PATH_LEN,
591 .mode = 0644,
592 .proc_handler = &proc_dostring,
593 .strategy = &sysctl_string,
594 },
595 #endif
596 #ifdef CONFIG_CHR_DEV_SG
597 {
598 .ctl_name = KERN_SG_BIG_BUFF,
599 .procname = "sg-big-buff",
600 .data = &sg_big_buff,
601 .maxlen = sizeof (int),
602 .mode = 0444,
603 .proc_handler = &proc_dointvec,
604 },
605 #endif
606 #ifdef CONFIG_BSD_PROCESS_ACCT
607 {
608 .ctl_name = KERN_ACCT,
609 .procname = "acct",
610 .data = &acct_parm,
611 .maxlen = 3*sizeof(int),
612 .mode = 0644,
613 .proc_handler = &proc_dointvec,
614 },
615 #endif
616 #ifdef CONFIG_MAGIC_SYSRQ
617 {
618 .ctl_name = KERN_SYSRQ,
619 .procname = "sysrq",
620 .data = &__sysrq_enabled,
621 .maxlen = sizeof (int),
622 .mode = 0644,
623 .proc_handler = &proc_dointvec,
624 },
625 #endif
626 #ifdef CONFIG_PROC_SYSCTL
627 {
628 .procname = "cad_pid",
629 .data = NULL,
630 .maxlen = sizeof (int),
631 .mode = 0600,
632 .proc_handler = &proc_do_cad_pid,
633 },
634 #endif
635 {
636 .ctl_name = KERN_MAX_THREADS,
637 .procname = "threads-max",
638 .data = &max_threads,
639 .maxlen = sizeof(int),
640 .mode = 0644,
641 .proc_handler = &proc_dointvec,
642 },
643 {
644 .ctl_name = KERN_RANDOM,
645 .procname = "random",
646 .mode = 0555,
647 .child = random_table,
648 },
649 {
650 .ctl_name = KERN_OVERFLOWUID,
651 .procname = "overflowuid",
652 .data = &overflowuid,
653 .maxlen = sizeof(int),
654 .mode = 0644,
655 .proc_handler = &proc_dointvec_minmax,
656 .strategy = &sysctl_intvec,
657 .extra1 = &minolduid,
658 .extra2 = &maxolduid,
659 },
660 {
661 .ctl_name = KERN_OVERFLOWGID,
662 .procname = "overflowgid",
663 .data = &overflowgid,
664 .maxlen = sizeof(int),
665 .mode = 0644,
666 .proc_handler = &proc_dointvec_minmax,
667 .strategy = &sysctl_intvec,
668 .extra1 = &minolduid,
669 .extra2 = &maxolduid,
670 },
671 #ifdef CONFIG_S390
672 #ifdef CONFIG_MATHEMU
673 {
674 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
675 .procname = "ieee_emulation_warnings",
676 .data = &sysctl_ieee_emulation_warnings,
677 .maxlen = sizeof(int),
678 .mode = 0644,
679 .proc_handler = &proc_dointvec,
680 },
681 #endif
682 {
683 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
684 .procname = "userprocess_debug",
685 .data = &sysctl_userprocess_debug,
686 .maxlen = sizeof(int),
687 .mode = 0644,
688 .proc_handler = &proc_dointvec,
689 },
690 #endif
691 {
692 .ctl_name = KERN_PIDMAX,
693 .procname = "pid_max",
694 .data = &pid_max,
695 .maxlen = sizeof (int),
696 .mode = 0644,
697 .proc_handler = &proc_dointvec_minmax,
698 .strategy = sysctl_intvec,
699 .extra1 = &pid_max_min,
700 .extra2 = &pid_max_max,
701 },
702 {
703 .ctl_name = KERN_PANIC_ON_OOPS,
704 .procname = "panic_on_oops",
705 .data = &panic_on_oops,
706 .maxlen = sizeof(int),
707 .mode = 0644,
708 .proc_handler = &proc_dointvec,
709 },
710 #if defined CONFIG_PRINTK
711 {
712 .ctl_name = KERN_PRINTK,
713 .procname = "printk",
714 .data = &console_loglevel,
715 .maxlen = 4*sizeof(int),
716 .mode = 0644,
717 .proc_handler = &proc_dointvec,
718 },
719 {
720 .ctl_name = KERN_PRINTK_RATELIMIT,
721 .procname = "printk_ratelimit",
722 .data = &printk_ratelimit_state.interval,
723 .maxlen = sizeof(int),
724 .mode = 0644,
725 .proc_handler = &proc_dointvec_jiffies,
726 .strategy = &sysctl_jiffies,
727 },
728 {
729 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
730 .procname = "printk_ratelimit_burst",
731 .data = &printk_ratelimit_state.burst,
732 .maxlen = sizeof(int),
733 .mode = 0644,
734 .proc_handler = &proc_dointvec,
735 },
736 {
737 .ctl_name = CTL_UNNUMBERED,
738 .procname = "printk_delay",
739 .data = &printk_delay_msec,
740 .maxlen = sizeof(int),
741 .mode = 0644,
742 .proc_handler = &proc_dointvec_minmax,
743 .strategy = &sysctl_intvec,
744 .extra1 = &zero,
745 .extra2 = &ten_thousand,
746 },
747 #endif
748 {
749 .ctl_name = KERN_NGROUPS_MAX,
750 .procname = "ngroups_max",
751 .data = &ngroups_max,
752 .maxlen = sizeof (int),
753 .mode = 0444,
754 .proc_handler = &proc_dointvec,
755 },
756 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
757 {
758 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
759 .procname = "unknown_nmi_panic",
760 .data = &unknown_nmi_panic,
761 .maxlen = sizeof (int),
762 .mode = 0644,
763 .proc_handler = &proc_dointvec,
764 },
765 {
766 .procname = "nmi_watchdog",
767 .data = &nmi_watchdog_enabled,
768 .maxlen = sizeof (int),
769 .mode = 0644,
770 .proc_handler = &proc_nmi_enabled,
771 },
772 #endif
773 #if defined(CONFIG_X86)
774 {
775 .ctl_name = KERN_PANIC_ON_NMI,
776 .procname = "panic_on_unrecovered_nmi",
777 .data = &panic_on_unrecovered_nmi,
778 .maxlen = sizeof(int),
779 .mode = 0644,
780 .proc_handler = &proc_dointvec,
781 },
782 {
783 .ctl_name = CTL_UNNUMBERED,
784 .procname = "panic_on_io_nmi",
785 .data = &panic_on_io_nmi,
786 .maxlen = sizeof(int),
787 .mode = 0644,
788 .proc_handler = &proc_dointvec,
789 },
790 {
791 .ctl_name = KERN_BOOTLOADER_TYPE,
792 .procname = "bootloader_type",
793 .data = &bootloader_type,
794 .maxlen = sizeof (int),
795 .mode = 0444,
796 .proc_handler = &proc_dointvec,
797 },
798 {
799 .ctl_name = CTL_UNNUMBERED,
800 .procname = "bootloader_version",
801 .data = &bootloader_version,
802 .maxlen = sizeof (int),
803 .mode = 0444,
804 .proc_handler = &proc_dointvec,
805 },
806 {
807 .ctl_name = CTL_UNNUMBERED,
808 .procname = "kstack_depth_to_print",
809 .data = &kstack_depth_to_print,
810 .maxlen = sizeof(int),
811 .mode = 0644,
812 .proc_handler = &proc_dointvec,
813 },
814 {
815 .ctl_name = CTL_UNNUMBERED,
816 .procname = "io_delay_type",
817 .data = &io_delay_type,
818 .maxlen = sizeof(int),
819 .mode = 0644,
820 .proc_handler = &proc_dointvec,
821 },
822 #endif
823 #if defined(CONFIG_MMU)
824 {
825 .ctl_name = KERN_RANDOMIZE,
826 .procname = "randomize_va_space",
827 .data = &randomize_va_space,
828 .maxlen = sizeof(int),
829 .mode = 0644,
830 .proc_handler = &proc_dointvec,
831 },
832 #endif
833 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
834 {
835 .ctl_name = KERN_SPIN_RETRY,
836 .procname = "spin_retry",
837 .data = &spin_retry,
838 .maxlen = sizeof (int),
839 .mode = 0644,
840 .proc_handler = &proc_dointvec,
841 },
842 #endif
843 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
844 {
845 .procname = "acpi_video_flags",
846 .data = &acpi_realmode_flags,
847 .maxlen = sizeof (unsigned long),
848 .mode = 0644,
849 .proc_handler = &proc_doulongvec_minmax,
850 },
851 #endif
852 #ifdef CONFIG_IA64
853 {
854 .ctl_name = KERN_IA64_UNALIGNED,
855 .procname = "ignore-unaligned-usertrap",
856 .data = &no_unaligned_warning,
857 .maxlen = sizeof (int),
858 .mode = 0644,
859 .proc_handler = &proc_dointvec,
860 },
861 {
862 .ctl_name = CTL_UNNUMBERED,
863 .procname = "unaligned-dump-stack",
864 .data = &unaligned_dump_stack,
865 .maxlen = sizeof (int),
866 .mode = 0644,
867 .proc_handler = &proc_dointvec,
868 },
869 #endif
870 #ifdef CONFIG_DETECT_SOFTLOCKUP
871 {
872 .ctl_name = CTL_UNNUMBERED,
873 .procname = "softlockup_panic",
874 .data = &softlockup_panic,
875 .maxlen = sizeof(int),
876 .mode = 0644,
877 .proc_handler = &proc_dointvec_minmax,
878 .strategy = &sysctl_intvec,
879 .extra1 = &zero,
880 .extra2 = &one,
881 },
882 {
883 .ctl_name = CTL_UNNUMBERED,
884 .procname = "softlockup_thresh",
885 .data = &softlockup_thresh,
886 .maxlen = sizeof(int),
887 .mode = 0644,
888 .proc_handler = &proc_dosoftlockup_thresh,
889 .strategy = &sysctl_intvec,
890 .extra1 = &neg_one,
891 .extra2 = &sixty,
892 },
893 #endif
894 #ifdef CONFIG_DETECT_HUNG_TASK
895 {
896 .ctl_name = CTL_UNNUMBERED,
897 .procname = "hung_task_panic",
898 .data = &sysctl_hung_task_panic,
899 .maxlen = sizeof(int),
900 .mode = 0644,
901 .proc_handler = &proc_dointvec_minmax,
902 .strategy = &sysctl_intvec,
903 .extra1 = &zero,
904 .extra2 = &one,
905 },
906 {
907 .ctl_name = CTL_UNNUMBERED,
908 .procname = "hung_task_check_count",
909 .data = &sysctl_hung_task_check_count,
910 .maxlen = sizeof(unsigned long),
911 .mode = 0644,
912 .proc_handler = &proc_doulongvec_minmax,
913 .strategy = &sysctl_intvec,
914 },
915 {
916 .ctl_name = CTL_UNNUMBERED,
917 .procname = "hung_task_timeout_secs",
918 .data = &sysctl_hung_task_timeout_secs,
919 .maxlen = sizeof(unsigned long),
920 .mode = 0644,
921 .proc_handler = &proc_dohung_task_timeout_secs,
922 .strategy = &sysctl_intvec,
923 },
924 {
925 .ctl_name = CTL_UNNUMBERED,
926 .procname = "hung_task_warnings",
927 .data = &sysctl_hung_task_warnings,
928 .maxlen = sizeof(unsigned long),
929 .mode = 0644,
930 .proc_handler = &proc_doulongvec_minmax,
931 .strategy = &sysctl_intvec,
932 },
933 #endif
934 #ifdef CONFIG_COMPAT
935 {
936 .ctl_name = KERN_COMPAT_LOG,
937 .procname = "compat-log",
938 .data = &compat_log,
939 .maxlen = sizeof (int),
940 .mode = 0644,
941 .proc_handler = &proc_dointvec,
942 },
943 #endif
944 #ifdef CONFIG_RT_MUTEXES
945 {
946 .ctl_name = KERN_MAX_LOCK_DEPTH,
947 .procname = "max_lock_depth",
948 .data = &max_lock_depth,
949 .maxlen = sizeof(int),
950 .mode = 0644,
951 .proc_handler = &proc_dointvec,
952 },
953 #endif
954 {
955 .ctl_name = CTL_UNNUMBERED,
956 .procname = "poweroff_cmd",
957 .data = &poweroff_cmd,
958 .maxlen = POWEROFF_CMD_PATH_LEN,
959 .mode = 0644,
960 .proc_handler = &proc_dostring,
961 .strategy = &sysctl_string,
962 },
963 #ifdef CONFIG_KEYS
964 {
965 .ctl_name = CTL_UNNUMBERED,
966 .procname = "keys",
967 .mode = 0555,
968 .child = key_sysctls,
969 },
970 #endif
971 #ifdef CONFIG_RCU_TORTURE_TEST
972 {
973 .ctl_name = CTL_UNNUMBERED,
974 .procname = "rcutorture_runnable",
975 .data = &rcutorture_runnable,
976 .maxlen = sizeof(int),
977 .mode = 0644,
978 .proc_handler = &proc_dointvec,
979 },
980 #endif
981 #ifdef CONFIG_SLOW_WORK
982 {
983 .ctl_name = CTL_UNNUMBERED,
984 .procname = "slow-work",
985 .mode = 0555,
986 .child = slow_work_sysctls,
987 },
988 #endif
989 #ifdef CONFIG_PERF_EVENTS
990 {
991 .ctl_name = CTL_UNNUMBERED,
992 .procname = "perf_event_paranoid",
993 .data = &sysctl_perf_event_paranoid,
994 .maxlen = sizeof(sysctl_perf_event_paranoid),
995 .mode = 0644,
996 .proc_handler = &proc_dointvec,
997 },
998 {
999 .ctl_name = CTL_UNNUMBERED,
1000 .procname = "perf_event_mlock_kb",
1001 .data = &sysctl_perf_event_mlock,
1002 .maxlen = sizeof(sysctl_perf_event_mlock),
1003 .mode = 0644,
1004 .proc_handler = &proc_dointvec,
1005 },
1006 {
1007 .ctl_name = CTL_UNNUMBERED,
1008 .procname = "perf_event_max_sample_rate",
1009 .data = &sysctl_perf_event_sample_rate,
1010 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1011 .mode = 0644,
1012 .proc_handler = &proc_dointvec,
1013 },
1014 #endif
1015 #ifdef CONFIG_KMEMCHECK
1016 {
1017 .ctl_name = CTL_UNNUMBERED,
1018 .procname = "kmemcheck",
1019 .data = &kmemcheck_enabled,
1020 .maxlen = sizeof(int),
1021 .mode = 0644,
1022 .proc_handler = &proc_dointvec,
1023 },
1024 #endif
1025 #ifdef CONFIG_BLOCK
1026 {
1027 .ctl_name = CTL_UNNUMBERED,
1028 .procname = "blk_iopoll",
1029 .data = &blk_iopoll_enabled,
1030 .maxlen = sizeof(int),
1031 .mode = 0644,
1032 .proc_handler = &proc_dointvec,
1033 },
1034 #endif
1035 /*
1036 * NOTE: do not add new entries to this table unless you have read
1037 * Documentation/sysctl/ctl_unnumbered.txt
1038 */
1039 { .ctl_name = 0 }
1040 };
1041
1042 static struct ctl_table vm_table[] = {
1043 {
1044 .ctl_name = VM_OVERCOMMIT_MEMORY,
1045 .procname = "overcommit_memory",
1046 .data = &sysctl_overcommit_memory,
1047 .maxlen = sizeof(sysctl_overcommit_memory),
1048 .mode = 0644,
1049 .proc_handler = &proc_dointvec,
1050 },
1051 {
1052 .ctl_name = VM_PANIC_ON_OOM,
1053 .procname = "panic_on_oom",
1054 .data = &sysctl_panic_on_oom,
1055 .maxlen = sizeof(sysctl_panic_on_oom),
1056 .mode = 0644,
1057 .proc_handler = &proc_dointvec,
1058 },
1059 {
1060 .ctl_name = CTL_UNNUMBERED,
1061 .procname = "oom_kill_allocating_task",
1062 .data = &sysctl_oom_kill_allocating_task,
1063 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1064 .mode = 0644,
1065 .proc_handler = &proc_dointvec,
1066 },
1067 {
1068 .ctl_name = CTL_UNNUMBERED,
1069 .procname = "oom_dump_tasks",
1070 .data = &sysctl_oom_dump_tasks,
1071 .maxlen = sizeof(sysctl_oom_dump_tasks),
1072 .mode = 0644,
1073 .proc_handler = &proc_dointvec,
1074 },
1075 {
1076 .ctl_name = VM_OVERCOMMIT_RATIO,
1077 .procname = "overcommit_ratio",
1078 .data = &sysctl_overcommit_ratio,
1079 .maxlen = sizeof(sysctl_overcommit_ratio),
1080 .mode = 0644,
1081 .proc_handler = &proc_dointvec,
1082 },
1083 {
1084 .ctl_name = VM_PAGE_CLUSTER,
1085 .procname = "page-cluster",
1086 .data = &page_cluster,
1087 .maxlen = sizeof(int),
1088 .mode = 0644,
1089 .proc_handler = &proc_dointvec,
1090 },
1091 {
1092 .ctl_name = VM_DIRTY_BACKGROUND,
1093 .procname = "dirty_background_ratio",
1094 .data = &dirty_background_ratio,
1095 .maxlen = sizeof(dirty_background_ratio),
1096 .mode = 0644,
1097 .proc_handler = &dirty_background_ratio_handler,
1098 .strategy = &sysctl_intvec,
1099 .extra1 = &zero,
1100 .extra2 = &one_hundred,
1101 },
1102 {
1103 .ctl_name = CTL_UNNUMBERED,
1104 .procname = "dirty_background_bytes",
1105 .data = &dirty_background_bytes,
1106 .maxlen = sizeof(dirty_background_bytes),
1107 .mode = 0644,
1108 .proc_handler = &dirty_background_bytes_handler,
1109 .strategy = &sysctl_intvec,
1110 .extra1 = &one_ul,
1111 },
1112 {
1113 .ctl_name = VM_DIRTY_RATIO,
1114 .procname = "dirty_ratio",
1115 .data = &vm_dirty_ratio,
1116 .maxlen = sizeof(vm_dirty_ratio),
1117 .mode = 0644,
1118 .proc_handler = &dirty_ratio_handler,
1119 .strategy = &sysctl_intvec,
1120 .extra1 = &zero,
1121 .extra2 = &one_hundred,
1122 },
1123 {
1124 .ctl_name = CTL_UNNUMBERED,
1125 .procname = "dirty_bytes",
1126 .data = &vm_dirty_bytes,
1127 .maxlen = sizeof(vm_dirty_bytes),
1128 .mode = 0644,
1129 .proc_handler = &dirty_bytes_handler,
1130 .strategy = &sysctl_intvec,
1131 .extra1 = &dirty_bytes_min,
1132 },
1133 {
1134 .procname = "dirty_writeback_centisecs",
1135 .data = &dirty_writeback_interval,
1136 .maxlen = sizeof(dirty_writeback_interval),
1137 .mode = 0644,
1138 .proc_handler = &dirty_writeback_centisecs_handler,
1139 },
1140 {
1141 .procname = "dirty_expire_centisecs",
1142 .data = &dirty_expire_interval,
1143 .maxlen = sizeof(dirty_expire_interval),
1144 .mode = 0644,
1145 .proc_handler = &proc_dointvec,
1146 },
1147 {
1148 .ctl_name = VM_NR_PDFLUSH_THREADS,
1149 .procname = "nr_pdflush_threads",
1150 .data = &nr_pdflush_threads,
1151 .maxlen = sizeof nr_pdflush_threads,
1152 .mode = 0444 /* read-only*/,
1153 .proc_handler = &proc_dointvec,
1154 },
1155 {
1156 .ctl_name = VM_SWAPPINESS,
1157 .procname = "swappiness",
1158 .data = &vm_swappiness,
1159 .maxlen = sizeof(vm_swappiness),
1160 .mode = 0644,
1161 .proc_handler = &proc_dointvec_minmax,
1162 .strategy = &sysctl_intvec,
1163 .extra1 = &zero,
1164 .extra2 = &one_hundred,
1165 },
1166 #ifdef CONFIG_HUGETLB_PAGE
1167 {
1168 .procname = "nr_hugepages",
1169 .data = NULL,
1170 .maxlen = sizeof(unsigned long),
1171 .mode = 0644,
1172 .proc_handler = &hugetlb_sysctl_handler,
1173 .extra1 = (void *)&hugetlb_zero,
1174 .extra2 = (void *)&hugetlb_infinity,
1175 },
1176 {
1177 .ctl_name = VM_HUGETLB_GROUP,
1178 .procname = "hugetlb_shm_group",
1179 .data = &sysctl_hugetlb_shm_group,
1180 .maxlen = sizeof(gid_t),
1181 .mode = 0644,
1182 .proc_handler = &proc_dointvec,
1183 },
1184 {
1185 .ctl_name = CTL_UNNUMBERED,
1186 .procname = "hugepages_treat_as_movable",
1187 .data = &hugepages_treat_as_movable,
1188 .maxlen = sizeof(int),
1189 .mode = 0644,
1190 .proc_handler = &hugetlb_treat_movable_handler,
1191 },
1192 {
1193 .ctl_name = CTL_UNNUMBERED,
1194 .procname = "nr_overcommit_hugepages",
1195 .data = NULL,
1196 .maxlen = sizeof(unsigned long),
1197 .mode = 0644,
1198 .proc_handler = &hugetlb_overcommit_handler,
1199 .extra1 = (void *)&hugetlb_zero,
1200 .extra2 = (void *)&hugetlb_infinity,
1201 },
1202 #endif
1203 {
1204 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
1205 .procname = "lowmem_reserve_ratio",
1206 .data = &sysctl_lowmem_reserve_ratio,
1207 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1208 .mode = 0644,
1209 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
1210 .strategy = &sysctl_intvec,
1211 },
1212 {
1213 .ctl_name = VM_DROP_PAGECACHE,
1214 .procname = "drop_caches",
1215 .data = &sysctl_drop_caches,
1216 .maxlen = sizeof(int),
1217 .mode = 0644,
1218 .proc_handler = drop_caches_sysctl_handler,
1219 .strategy = &sysctl_intvec,
1220 },
1221 {
1222 .ctl_name = VM_MIN_FREE_KBYTES,
1223 .procname = "min_free_kbytes",
1224 .data = &min_free_kbytes,
1225 .maxlen = sizeof(min_free_kbytes),
1226 .mode = 0644,
1227 .proc_handler = &min_free_kbytes_sysctl_handler,
1228 .strategy = &sysctl_intvec,
1229 .extra1 = &zero,
1230 },
1231 {
1232 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
1233 .procname = "percpu_pagelist_fraction",
1234 .data = &percpu_pagelist_fraction,
1235 .maxlen = sizeof(percpu_pagelist_fraction),
1236 .mode = 0644,
1237 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
1238 .strategy = &sysctl_intvec,
1239 .extra1 = &min_percpu_pagelist_fract,
1240 },
1241 #ifdef CONFIG_MMU
1242 {
1243 .ctl_name = VM_MAX_MAP_COUNT,
1244 .procname = "max_map_count",
1245 .data = &sysctl_max_map_count,
1246 .maxlen = sizeof(sysctl_max_map_count),
1247 .mode = 0644,
1248 .proc_handler = &proc_dointvec
1249 },
1250 #else
1251 {
1252 .ctl_name = CTL_UNNUMBERED,
1253 .procname = "nr_trim_pages",
1254 .data = &sysctl_nr_trim_pages,
1255 .maxlen = sizeof(sysctl_nr_trim_pages),
1256 .mode = 0644,
1257 .proc_handler = &proc_dointvec_minmax,
1258 .strategy = &sysctl_intvec,
1259 .extra1 = &zero,
1260 },
1261 #endif
1262 {
1263 .ctl_name = VM_LAPTOP_MODE,
1264 .procname = "laptop_mode",
1265 .data = &laptop_mode,
1266 .maxlen = sizeof(laptop_mode),
1267 .mode = 0644,
1268 .proc_handler = &proc_dointvec_jiffies,
1269 .strategy = &sysctl_jiffies,
1270 },
1271 {
1272 .ctl_name = VM_BLOCK_DUMP,
1273 .procname = "block_dump",
1274 .data = &block_dump,
1275 .maxlen = sizeof(block_dump),
1276 .mode = 0644,
1277 .proc_handler = &proc_dointvec,
1278 .strategy = &sysctl_intvec,
1279 .extra1 = &zero,
1280 },
1281 {
1282 .ctl_name = VM_VFS_CACHE_PRESSURE,
1283 .procname = "vfs_cache_pressure",
1284 .data = &sysctl_vfs_cache_pressure,
1285 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1286 .mode = 0644,
1287 .proc_handler = &proc_dointvec,
1288 .strategy = &sysctl_intvec,
1289 .extra1 = &zero,
1290 },
1291 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1292 {
1293 .ctl_name = VM_LEGACY_VA_LAYOUT,
1294 .procname = "legacy_va_layout",
1295 .data = &sysctl_legacy_va_layout,
1296 .maxlen = sizeof(sysctl_legacy_va_layout),
1297 .mode = 0644,
1298 .proc_handler = &proc_dointvec,
1299 .strategy = &sysctl_intvec,
1300 .extra1 = &zero,
1301 },
1302 #endif
1303 #ifdef CONFIG_NUMA
1304 {
1305 .ctl_name = VM_ZONE_RECLAIM_MODE,
1306 .procname = "zone_reclaim_mode",
1307 .data = &zone_reclaim_mode,
1308 .maxlen = sizeof(zone_reclaim_mode),
1309 .mode = 0644,
1310 .proc_handler = &proc_dointvec,
1311 .strategy = &sysctl_intvec,
1312 .extra1 = &zero,
1313 },
1314 {
1315 .ctl_name = VM_MIN_UNMAPPED,
1316 .procname = "min_unmapped_ratio",
1317 .data = &sysctl_min_unmapped_ratio,
1318 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1319 .mode = 0644,
1320 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
1321 .strategy = &sysctl_intvec,
1322 .extra1 = &zero,
1323 .extra2 = &one_hundred,
1324 },
1325 {
1326 .ctl_name = VM_MIN_SLAB,
1327 .procname = "min_slab_ratio",
1328 .data = &sysctl_min_slab_ratio,
1329 .maxlen = sizeof(sysctl_min_slab_ratio),
1330 .mode = 0644,
1331 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1332 .strategy = &sysctl_intvec,
1333 .extra1 = &zero,
1334 .extra2 = &one_hundred,
1335 },
1336 #endif
1337 #ifdef CONFIG_SMP
1338 {
1339 .ctl_name = CTL_UNNUMBERED,
1340 .procname = "stat_interval",
1341 .data = &sysctl_stat_interval,
1342 .maxlen = sizeof(sysctl_stat_interval),
1343 .mode = 0644,
1344 .proc_handler = &proc_dointvec_jiffies,
1345 .strategy = &sysctl_jiffies,
1346 },
1347 #endif
1348 #ifdef CONFIG_MMU
1349 {
1350 .ctl_name = CTL_UNNUMBERED,
1351 .procname = "mmap_min_addr",
1352 .data = &dac_mmap_min_addr,
1353 .maxlen = sizeof(unsigned long),
1354 .mode = 0644,
1355 .proc_handler = &mmap_min_addr_handler,
1356 },
1357 #endif
1358 #ifdef CONFIG_NUMA
1359 {
1360 .ctl_name = CTL_UNNUMBERED,
1361 .procname = "numa_zonelist_order",
1362 .data = &numa_zonelist_order,
1363 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1364 .mode = 0644,
1365 .proc_handler = &numa_zonelist_order_handler,
1366 .strategy = &sysctl_string,
1367 },
1368 #endif
1369 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1370 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1371 {
1372 .ctl_name = VM_VDSO_ENABLED,
1373 .procname = "vdso_enabled",
1374 .data = &vdso_enabled,
1375 .maxlen = sizeof(vdso_enabled),
1376 .mode = 0644,
1377 .proc_handler = &proc_dointvec,
1378 .strategy = &sysctl_intvec,
1379 .extra1 = &zero,
1380 },
1381 #endif
1382 #ifdef CONFIG_HIGHMEM
1383 {
1384 .ctl_name = CTL_UNNUMBERED,
1385 .procname = "highmem_is_dirtyable",
1386 .data = &vm_highmem_is_dirtyable,
1387 .maxlen = sizeof(vm_highmem_is_dirtyable),
1388 .mode = 0644,
1389 .proc_handler = &proc_dointvec_minmax,
1390 .strategy = &sysctl_intvec,
1391 .extra1 = &zero,
1392 .extra2 = &one,
1393 },
1394 #endif
1395 {
1396 .ctl_name = CTL_UNNUMBERED,
1397 .procname = "scan_unevictable_pages",
1398 .data = &scan_unevictable_pages,
1399 .maxlen = sizeof(scan_unevictable_pages),
1400 .mode = 0644,
1401 .proc_handler = &scan_unevictable_handler,
1402 },
1403 #ifdef CONFIG_MEMORY_FAILURE
1404 {
1405 .ctl_name = CTL_UNNUMBERED,
1406 .procname = "memory_failure_early_kill",
1407 .data = &sysctl_memory_failure_early_kill,
1408 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1409 .mode = 0644,
1410 .proc_handler = &proc_dointvec_minmax,
1411 .strategy = &sysctl_intvec,
1412 .extra1 = &zero,
1413 .extra2 = &one,
1414 },
1415 {
1416 .ctl_name = CTL_UNNUMBERED,
1417 .procname = "memory_failure_recovery",
1418 .data = &sysctl_memory_failure_recovery,
1419 .maxlen = sizeof(sysctl_memory_failure_recovery),
1420 .mode = 0644,
1421 .proc_handler = &proc_dointvec_minmax,
1422 .strategy = &sysctl_intvec,
1423 .extra1 = &zero,
1424 .extra2 = &one,
1425 },
1426 #endif
1427
1428 /*
1429 * NOTE: do not add new entries to this table unless you have read
1430 * Documentation/sysctl/ctl_unnumbered.txt
1431 */
1432 { .ctl_name = 0 }
1433 };
1434
1435 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1436 static struct ctl_table binfmt_misc_table[] = {
1437 { .ctl_name = 0 }
1438 };
1439 #endif
1440
1441 static struct ctl_table fs_table[] = {
1442 {
1443 .ctl_name = FS_NRINODE,
1444 .procname = "inode-nr",
1445 .data = &inodes_stat,
1446 .maxlen = 2*sizeof(int),
1447 .mode = 0444,
1448 .proc_handler = &proc_dointvec,
1449 },
1450 {
1451 .ctl_name = FS_STATINODE,
1452 .procname = "inode-state",
1453 .data = &inodes_stat,
1454 .maxlen = 7*sizeof(int),
1455 .mode = 0444,
1456 .proc_handler = &proc_dointvec,
1457 },
1458 {
1459 .procname = "file-nr",
1460 .data = &files_stat,
1461 .maxlen = 3*sizeof(int),
1462 .mode = 0444,
1463 .proc_handler = &proc_nr_files,
1464 },
1465 {
1466 .ctl_name = FS_MAXFILE,
1467 .procname = "file-max",
1468 .data = &files_stat.max_files,
1469 .maxlen = sizeof(int),
1470 .mode = 0644,
1471 .proc_handler = &proc_dointvec,
1472 },
1473 {
1474 .ctl_name = CTL_UNNUMBERED,
1475 .procname = "nr_open",
1476 .data = &sysctl_nr_open,
1477 .maxlen = sizeof(int),
1478 .mode = 0644,
1479 .proc_handler = &proc_dointvec_minmax,
1480 .extra1 = &sysctl_nr_open_min,
1481 .extra2 = &sysctl_nr_open_max,
1482 },
1483 {
1484 .ctl_name = FS_DENTRY,
1485 .procname = "dentry-state",
1486 .data = &dentry_stat,
1487 .maxlen = 6*sizeof(int),
1488 .mode = 0444,
1489 .proc_handler = &proc_dointvec,
1490 },
1491 {
1492 .ctl_name = FS_OVERFLOWUID,
1493 .procname = "overflowuid",
1494 .data = &fs_overflowuid,
1495 .maxlen = sizeof(int),
1496 .mode = 0644,
1497 .proc_handler = &proc_dointvec_minmax,
1498 .strategy = &sysctl_intvec,
1499 .extra1 = &minolduid,
1500 .extra2 = &maxolduid,
1501 },
1502 {
1503 .ctl_name = FS_OVERFLOWGID,
1504 .procname = "overflowgid",
1505 .data = &fs_overflowgid,
1506 .maxlen = sizeof(int),
1507 .mode = 0644,
1508 .proc_handler = &proc_dointvec_minmax,
1509 .strategy = &sysctl_intvec,
1510 .extra1 = &minolduid,
1511 .extra2 = &maxolduid,
1512 },
1513 #ifdef CONFIG_FILE_LOCKING
1514 {
1515 .ctl_name = FS_LEASES,
1516 .procname = "leases-enable",
1517 .data = &leases_enable,
1518 .maxlen = sizeof(int),
1519 .mode = 0644,
1520 .proc_handler = &proc_dointvec,
1521 },
1522 #endif
1523 #ifdef CONFIG_DNOTIFY
1524 {
1525 .ctl_name = FS_DIR_NOTIFY,
1526 .procname = "dir-notify-enable",
1527 .data = &dir_notify_enable,
1528 .maxlen = sizeof(int),
1529 .mode = 0644,
1530 .proc_handler = &proc_dointvec,
1531 },
1532 #endif
1533 #ifdef CONFIG_MMU
1534 #ifdef CONFIG_FILE_LOCKING
1535 {
1536 .ctl_name = FS_LEASE_TIME,
1537 .procname = "lease-break-time",
1538 .data = &lease_break_time,
1539 .maxlen = sizeof(int),
1540 .mode = 0644,
1541 .proc_handler = &proc_dointvec,
1542 },
1543 #endif
1544 #ifdef CONFIG_AIO
1545 {
1546 .procname = "aio-nr",
1547 .data = &aio_nr,
1548 .maxlen = sizeof(aio_nr),
1549 .mode = 0444,
1550 .proc_handler = &proc_doulongvec_minmax,
1551 },
1552 {
1553 .procname = "aio-max-nr",
1554 .data = &aio_max_nr,
1555 .maxlen = sizeof(aio_max_nr),
1556 .mode = 0644,
1557 .proc_handler = &proc_doulongvec_minmax,
1558 },
1559 #endif /* CONFIG_AIO */
1560 #ifdef CONFIG_INOTIFY_USER
1561 {
1562 .ctl_name = FS_INOTIFY,
1563 .procname = "inotify",
1564 .mode = 0555,
1565 .child = inotify_table,
1566 },
1567 #endif
1568 #ifdef CONFIG_EPOLL
1569 {
1570 .procname = "epoll",
1571 .mode = 0555,
1572 .child = epoll_table,
1573 },
1574 #endif
1575 #endif
1576 {
1577 .ctl_name = KERN_SETUID_DUMPABLE,
1578 .procname = "suid_dumpable",
1579 .data = &suid_dumpable,
1580 .maxlen = sizeof(int),
1581 .mode = 0644,
1582 .proc_handler = &proc_dointvec_minmax,
1583 .strategy = &sysctl_intvec,
1584 .extra1 = &zero,
1585 .extra2 = &two,
1586 },
1587 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1588 {
1589 .ctl_name = CTL_UNNUMBERED,
1590 .procname = "binfmt_misc",
1591 .mode = 0555,
1592 .child = binfmt_misc_table,
1593 },
1594 #endif
1595 /*
1596 * NOTE: do not add new entries to this table unless you have read
1597 * Documentation/sysctl/ctl_unnumbered.txt
1598 */
1599 { .ctl_name = 0 }
1600 };
1601
1602 static struct ctl_table debug_table[] = {
1603 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1604 {
1605 .ctl_name = CTL_UNNUMBERED,
1606 .procname = "exception-trace",
1607 .data = &show_unhandled_signals,
1608 .maxlen = sizeof(int),
1609 .mode = 0644,
1610 .proc_handler = proc_dointvec_minmax,
1611 .extra1 = &zero,
1612 },
1613 #endif
1614 { .ctl_name = 0 }
1615 };
1616
1617 static struct ctl_table dev_table[] = {
1618 { .ctl_name = 0 }
1619 };
1620
1621 static DEFINE_SPINLOCK(sysctl_lock);
1622
1623 /* called under sysctl_lock */
1624 static int use_table(struct ctl_table_header *p)
1625 {
1626 if (unlikely(p->unregistering))
1627 return 0;
1628 p->used++;
1629 return 1;
1630 }
1631
1632 /* called under sysctl_lock */
1633 static void unuse_table(struct ctl_table_header *p)
1634 {
1635 if (!--p->used)
1636 if (unlikely(p->unregistering))
1637 complete(p->unregistering);
1638 }
1639
1640 /* called under sysctl_lock, will reacquire if has to wait */
1641 static void start_unregistering(struct ctl_table_header *p)
1642 {
1643 /*
1644 * if p->used is 0, nobody will ever touch that entry again;
1645 * we'll eliminate all paths to it before dropping sysctl_lock
1646 */
1647 if (unlikely(p->used)) {
1648 struct completion wait;
1649 init_completion(&wait);
1650 p->unregistering = &wait;
1651 spin_unlock(&sysctl_lock);
1652 wait_for_completion(&wait);
1653 spin_lock(&sysctl_lock);
1654 } else {
1655 /* anything non-NULL; we'll never dereference it */
1656 p->unregistering = ERR_PTR(-EINVAL);
1657 }
1658 /*
1659 * do not remove from the list until nobody holds it; walking the
1660 * list in do_sysctl() relies on that.
1661 */
1662 list_del_init(&p->ctl_entry);
1663 }
1664
1665 void sysctl_head_get(struct ctl_table_header *head)
1666 {
1667 spin_lock(&sysctl_lock);
1668 head->count++;
1669 spin_unlock(&sysctl_lock);
1670 }
1671
1672 void sysctl_head_put(struct ctl_table_header *head)
1673 {
1674 spin_lock(&sysctl_lock);
1675 if (!--head->count)
1676 kfree(head);
1677 spin_unlock(&sysctl_lock);
1678 }
1679
1680 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1681 {
1682 if (!head)
1683 BUG();
1684 spin_lock(&sysctl_lock);
1685 if (!use_table(head))
1686 head = ERR_PTR(-ENOENT);
1687 spin_unlock(&sysctl_lock);
1688 return head;
1689 }
1690
1691 void sysctl_head_finish(struct ctl_table_header *head)
1692 {
1693 if (!head)
1694 return;
1695 spin_lock(&sysctl_lock);
1696 unuse_table(head);
1697 spin_unlock(&sysctl_lock);
1698 }
1699
1700 static struct ctl_table_set *
1701 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1702 {
1703 struct ctl_table_set *set = &root->default_set;
1704 if (root->lookup)
1705 set = root->lookup(root, namespaces);
1706 return set;
1707 }
1708
1709 static struct list_head *
1710 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1711 {
1712 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1713 return &set->list;
1714 }
1715
1716 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1717 struct ctl_table_header *prev)
1718 {
1719 struct ctl_table_root *root;
1720 struct list_head *header_list;
1721 struct ctl_table_header *head;
1722 struct list_head *tmp;
1723
1724 spin_lock(&sysctl_lock);
1725 if (prev) {
1726 head = prev;
1727 tmp = &prev->ctl_entry;
1728 unuse_table(prev);
1729 goto next;
1730 }
1731 tmp = &root_table_header.ctl_entry;
1732 for (;;) {
1733 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1734
1735 if (!use_table(head))
1736 goto next;
1737 spin_unlock(&sysctl_lock);
1738 return head;
1739 next:
1740 root = head->root;
1741 tmp = tmp->next;
1742 header_list = lookup_header_list(root, namespaces);
1743 if (tmp != header_list)
1744 continue;
1745
1746 do {
1747 root = list_entry(root->root_list.next,
1748 struct ctl_table_root, root_list);
1749 if (root == &sysctl_table_root)
1750 goto out;
1751 header_list = lookup_header_list(root, namespaces);
1752 } while (list_empty(header_list));
1753 tmp = header_list->next;
1754 }
1755 out:
1756 spin_unlock(&sysctl_lock);
1757 return NULL;
1758 }
1759
1760 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1761 {
1762 return __sysctl_head_next(current->nsproxy, prev);
1763 }
1764
1765 void register_sysctl_root(struct ctl_table_root *root)
1766 {
1767 spin_lock(&sysctl_lock);
1768 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1769 spin_unlock(&sysctl_lock);
1770 }
1771
1772 #ifdef CONFIG_SYSCTL_SYSCALL
1773 /* Perform the actual read/write of a sysctl table entry. */
1774 static int do_sysctl_strategy(struct ctl_table_root *root,
1775 struct ctl_table *table,
1776 void __user *oldval, size_t __user *oldlenp,
1777 void __user *newval, size_t newlen)
1778 {
1779 int op = 0, rc;
1780
1781 if (oldval)
1782 op |= MAY_READ;
1783 if (newval)
1784 op |= MAY_WRITE;
1785 if (sysctl_perm(root, table, op))
1786 return -EPERM;
1787
1788 if (table->strategy) {
1789 rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1790 if (rc < 0)
1791 return rc;
1792 if (rc > 0)
1793 return 0;
1794 }
1795
1796 /* If there is no strategy routine, or if the strategy returns
1797 * zero, proceed with automatic r/w */
1798 if (table->data && table->maxlen) {
1799 rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1800 if (rc < 0)
1801 return rc;
1802 }
1803 return 0;
1804 }
1805
1806 static int parse_table(int __user *name, int nlen,
1807 void __user *oldval, size_t __user *oldlenp,
1808 void __user *newval, size_t newlen,
1809 struct ctl_table_root *root,
1810 struct ctl_table *table)
1811 {
1812 int n;
1813 repeat:
1814 if (!nlen)
1815 return -ENOTDIR;
1816 if (get_user(n, name))
1817 return -EFAULT;
1818 for ( ; table->ctl_name || table->procname; table++) {
1819 if (!table->ctl_name)
1820 continue;
1821 if (n == table->ctl_name) {
1822 int error;
1823 if (table->child) {
1824 if (sysctl_perm(root, table, MAY_EXEC))
1825 return -EPERM;
1826 name++;
1827 nlen--;
1828 table = table->child;
1829 goto repeat;
1830 }
1831 error = do_sysctl_strategy(root, table,
1832 oldval, oldlenp,
1833 newval, newlen);
1834 return error;
1835 }
1836 }
1837 return -ENOTDIR;
1838 }
1839
1840 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1841 void __user *newval, size_t newlen)
1842 {
1843 struct ctl_table_header *head;
1844 int error = -ENOTDIR;
1845
1846 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1847 return -ENOTDIR;
1848 if (oldval) {
1849 int old_len;
1850 if (!oldlenp || get_user(old_len, oldlenp))
1851 return -EFAULT;
1852 }
1853
1854 for (head = sysctl_head_next(NULL); head;
1855 head = sysctl_head_next(head)) {
1856 error = parse_table(name, nlen, oldval, oldlenp,
1857 newval, newlen,
1858 head->root, head->ctl_table);
1859 if (error != -ENOTDIR) {
1860 sysctl_head_finish(head);
1861 break;
1862 }
1863 }
1864 return error;
1865 }
1866
1867 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
1868 {
1869 struct __sysctl_args tmp;
1870 int error;
1871
1872 if (copy_from_user(&tmp, args, sizeof(tmp)))
1873 return -EFAULT;
1874
1875 error = deprecated_sysctl_warning(&tmp);
1876 if (error)
1877 goto out;
1878
1879 lock_kernel();
1880 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1881 tmp.newval, tmp.newlen);
1882 unlock_kernel();
1883 out:
1884 return error;
1885 }
1886 #endif /* CONFIG_SYSCTL_SYSCALL */
1887
1888 /*
1889 * sysctl_perm does NOT grant the superuser all rights automatically, because
1890 * some sysctl variables are readonly even to root.
1891 */
1892
1893 static int test_perm(int mode, int op)
1894 {
1895 if (!current_euid())
1896 mode >>= 6;
1897 else if (in_egroup_p(0))
1898 mode >>= 3;
1899 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1900 return 0;
1901 return -EACCES;
1902 }
1903
1904 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1905 {
1906 int error;
1907 int mode;
1908
1909 error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1910 if (error)
1911 return error;
1912
1913 if (root->permissions)
1914 mode = root->permissions(root, current->nsproxy, table);
1915 else
1916 mode = table->mode;
1917
1918 return test_perm(mode, op);
1919 }
1920
1921 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1922 {
1923 for (; table->ctl_name || table->procname; table++) {
1924 table->parent = parent;
1925 if (table->child)
1926 sysctl_set_parent(table, table->child);
1927 }
1928 }
1929
1930 static __init int sysctl_init(void)
1931 {
1932 sysctl_set_parent(NULL, root_table);
1933 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1934 {
1935 int err;
1936 err = sysctl_check_table(current->nsproxy, root_table);
1937 }
1938 #endif
1939 return 0;
1940 }
1941
1942 core_initcall(sysctl_init);
1943
1944 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1945 struct ctl_table *table)
1946 {
1947 struct ctl_table *p;
1948 const char *s = branch->procname;
1949
1950 /* branch should have named subdirectory as its first element */
1951 if (!s || !branch->child)
1952 return NULL;
1953
1954 /* ... and nothing else */
1955 if (branch[1].procname || branch[1].ctl_name)
1956 return NULL;
1957
1958 /* table should contain subdirectory with the same name */
1959 for (p = table; p->procname || p->ctl_name; p++) {
1960 if (!p->child)
1961 continue;
1962 if (p->procname && strcmp(p->procname, s) == 0)
1963 return p;
1964 }
1965 return NULL;
1966 }
1967
1968 /* see if attaching q to p would be an improvement */
1969 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1970 {
1971 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1972 struct ctl_table *next;
1973 int is_better = 0;
1974 int not_in_parent = !p->attached_by;
1975
1976 while ((next = is_branch_in(by, to)) != NULL) {
1977 if (by == q->attached_by)
1978 is_better = 1;
1979 if (to == p->attached_by)
1980 not_in_parent = 1;
1981 by = by->child;
1982 to = next->child;
1983 }
1984
1985 if (is_better && not_in_parent) {
1986 q->attached_by = by;
1987 q->attached_to = to;
1988 q->parent = p;
1989 }
1990 }
1991
1992 /**
1993 * __register_sysctl_paths - register a sysctl hierarchy
1994 * @root: List of sysctl headers to register on
1995 * @namespaces: Data to compute which lists of sysctl entries are visible
1996 * @path: The path to the directory the sysctl table is in.
1997 * @table: the top-level table structure
1998 *
1999 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2000 * array. A completely 0 filled entry terminates the table.
2001 *
2002 * The members of the &struct ctl_table structure are used as follows:
2003 *
2004 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
2005 * must be unique within that level of sysctl
2006 *
2007 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
2008 * enter a sysctl file
2009 *
2010 * data - a pointer to data for use by proc_handler
2011 *
2012 * maxlen - the maximum size in bytes of the data
2013 *
2014 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
2015 *
2016 * child - a pointer to the child sysctl table if this entry is a directory, or
2017 * %NULL.
2018 *
2019 * proc_handler - the text handler routine (described below)
2020 *
2021 * strategy - the strategy routine (described below)
2022 *
2023 * de - for internal use by the sysctl routines
2024 *
2025 * extra1, extra2 - extra pointers usable by the proc handler routines
2026 *
2027 * Leaf nodes in the sysctl tree will be represented by a single file
2028 * under /proc; non-leaf nodes will be represented by directories.
2029 *
2030 * sysctl(2) can automatically manage read and write requests through
2031 * the sysctl table. The data and maxlen fields of the ctl_table
2032 * struct enable minimal validation of the values being written to be
2033 * performed, and the mode field allows minimal authentication.
2034 *
2035 * More sophisticated management can be enabled by the provision of a
2036 * strategy routine with the table entry. This will be called before
2037 * any automatic read or write of the data is performed.
2038 *
2039 * The strategy routine may return
2040 *
2041 * < 0 - Error occurred (error is passed to user process)
2042 *
2043 * 0 - OK - proceed with automatic read or write.
2044 *
2045 * > 0 - OK - read or write has been done by the strategy routine, so
2046 * return immediately.
2047 *
2048 * There must be a proc_handler routine for any terminal nodes
2049 * mirrored under /proc/sys (non-terminals are handled by a built-in
2050 * directory handler). Several default handlers are available to
2051 * cover common cases -
2052 *
2053 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
2054 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
2055 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
2056 *
2057 * It is the handler's job to read the input buffer from user memory
2058 * and process it. The handler should return 0 on success.
2059 *
2060 * This routine returns %NULL on a failure to register, and a pointer
2061 * to the table header on success.
2062 */
2063 struct ctl_table_header *__register_sysctl_paths(
2064 struct ctl_table_root *root,
2065 struct nsproxy *namespaces,
2066 const struct ctl_path *path, struct ctl_table *table)
2067 {
2068 struct ctl_table_header *header;
2069 struct ctl_table *new, **prevp;
2070 unsigned int n, npath;
2071 struct ctl_table_set *set;
2072
2073 /* Count the path components */
2074 for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
2075 ;
2076
2077 /*
2078 * For each path component, allocate a 2-element ctl_table array.
2079 * The first array element will be filled with the sysctl entry
2080 * for this, the second will be the sentinel (ctl_name == 0).
2081 *
2082 * We allocate everything in one go so that we don't have to
2083 * worry about freeing additional memory in unregister_sysctl_table.
2084 */
2085 header = kzalloc(sizeof(struct ctl_table_header) +
2086 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
2087 if (!header)
2088 return NULL;
2089
2090 new = (struct ctl_table *) (header + 1);
2091
2092 /* Now connect the dots */
2093 prevp = &header->ctl_table;
2094 for (n = 0; n < npath; ++n, ++path) {
2095 /* Copy the procname */
2096 new->procname = path->procname;
2097 new->ctl_name = path->ctl_name;
2098 new->mode = 0555;
2099
2100 *prevp = new;
2101 prevp = &new->child;
2102
2103 new += 2;
2104 }
2105 *prevp = table;
2106 header->ctl_table_arg = table;
2107
2108 INIT_LIST_HEAD(&header->ctl_entry);
2109 header->used = 0;
2110 header->unregistering = NULL;
2111 header->root = root;
2112 sysctl_set_parent(NULL, header->ctl_table);
2113 header->count = 1;
2114 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
2115 if (sysctl_check_table(namespaces, header->ctl_table)) {
2116 kfree(header);
2117 return NULL;
2118 }
2119 #endif
2120 spin_lock(&sysctl_lock);
2121 header->set = lookup_header_set(root, namespaces);
2122 header->attached_by = header->ctl_table;
2123 header->attached_to = root_table;
2124 header->parent = &root_table_header;
2125 for (set = header->set; set; set = set->parent) {
2126 struct ctl_table_header *p;
2127 list_for_each_entry(p, &set->list, ctl_entry) {
2128 if (p->unregistering)
2129 continue;
2130 try_attach(p, header);
2131 }
2132 }
2133 header->parent->count++;
2134 list_add_tail(&header->ctl_entry, &header->set->list);
2135 spin_unlock(&sysctl_lock);
2136
2137 return header;
2138 }
2139
2140 /**
2141 * register_sysctl_table_path - register a sysctl table hierarchy
2142 * @path: The path to the directory the sysctl table is in.
2143 * @table: the top-level table structure
2144 *
2145 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2146 * array. A completely 0 filled entry terminates the table.
2147 *
2148 * See __register_sysctl_paths for more details.
2149 */
2150 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2151 struct ctl_table *table)
2152 {
2153 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
2154 path, table);
2155 }
2156
2157 /**
2158 * register_sysctl_table - register a sysctl table hierarchy
2159 * @table: the top-level table structure
2160 *
2161 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2162 * array. A completely 0 filled entry terminates the table.
2163 *
2164 * See register_sysctl_paths for more details.
2165 */
2166 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
2167 {
2168 static const struct ctl_path null_path[] = { {} };
2169
2170 return register_sysctl_paths(null_path, table);
2171 }
2172
2173 /**
2174 * unregister_sysctl_table - unregister a sysctl table hierarchy
2175 * @header: the header returned from register_sysctl_table
2176 *
2177 * Unregisters the sysctl table and all children. proc entries may not
2178 * actually be removed until they are no longer used by anyone.
2179 */
2180 void unregister_sysctl_table(struct ctl_table_header * header)
2181 {
2182 might_sleep();
2183
2184 if (header == NULL)
2185 return;
2186
2187 spin_lock(&sysctl_lock);
2188 start_unregistering(header);
2189 if (!--header->parent->count) {
2190 WARN_ON(1);
2191 kfree(header->parent);
2192 }
2193 if (!--header->count)
2194 kfree(header);
2195 spin_unlock(&sysctl_lock);
2196 }
2197
2198 int sysctl_is_seen(struct ctl_table_header *p)
2199 {
2200 struct ctl_table_set *set = p->set;
2201 int res;
2202 spin_lock(&sysctl_lock);
2203 if (p->unregistering)
2204 res = 0;
2205 else if (!set->is_seen)
2206 res = 1;
2207 else
2208 res = set->is_seen(set);
2209 spin_unlock(&sysctl_lock);
2210 return res;
2211 }
2212
2213 void setup_sysctl_set(struct ctl_table_set *p,
2214 struct ctl_table_set *parent,
2215 int (*is_seen)(struct ctl_table_set *))
2216 {
2217 INIT_LIST_HEAD(&p->list);
2218 p->parent = parent ? parent : &sysctl_table_root.default_set;
2219 p->is_seen = is_seen;
2220 }
2221
2222 #else /* !CONFIG_SYSCTL */
2223 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2224 {
2225 return NULL;
2226 }
2227
2228 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2229 struct ctl_table *table)
2230 {
2231 return NULL;
2232 }
2233
2234 void unregister_sysctl_table(struct ctl_table_header * table)
2235 {
2236 }
2237
2238 void setup_sysctl_set(struct ctl_table_set *p,
2239 struct ctl_table_set *parent,
2240 int (*is_seen)(struct ctl_table_set *))
2241 {
2242 }
2243
2244 void sysctl_head_put(struct ctl_table_header *head)
2245 {
2246 }
2247
2248 #endif /* CONFIG_SYSCTL */
2249
2250 /*
2251 * /proc/sys support
2252 */
2253
2254 #ifdef CONFIG_PROC_SYSCTL
2255
2256 static int _proc_do_string(void* data, int maxlen, int write,
2257 void __user *buffer,
2258 size_t *lenp, loff_t *ppos)
2259 {
2260 size_t len;
2261 char __user *p;
2262 char c;
2263
2264 if (!data || !maxlen || !*lenp) {
2265 *lenp = 0;
2266 return 0;
2267 }
2268
2269 if (write) {
2270 len = 0;
2271 p = buffer;
2272 while (len < *lenp) {
2273 if (get_user(c, p++))
2274 return -EFAULT;
2275 if (c == 0 || c == '\n')
2276 break;
2277 len++;
2278 }
2279 if (len >= maxlen)
2280 len = maxlen-1;
2281 if(copy_from_user(data, buffer, len))
2282 return -EFAULT;
2283 ((char *) data)[len] = 0;
2284 *ppos += *lenp;
2285 } else {
2286 len = strlen(data);
2287 if (len > maxlen)
2288 len = maxlen;
2289
2290 if (*ppos > len) {
2291 *lenp = 0;
2292 return 0;
2293 }
2294
2295 data += *ppos;
2296 len -= *ppos;
2297
2298 if (len > *lenp)
2299 len = *lenp;
2300 if (len)
2301 if(copy_to_user(buffer, data, len))
2302 return -EFAULT;
2303 if (len < *lenp) {
2304 if(put_user('\n', ((char __user *) buffer) + len))
2305 return -EFAULT;
2306 len++;
2307 }
2308 *lenp = len;
2309 *ppos += len;
2310 }
2311 return 0;
2312 }
2313
2314 /**
2315 * proc_dostring - read a string sysctl
2316 * @table: the sysctl table
2317 * @write: %TRUE if this is a write to the sysctl file
2318 * @buffer: the user buffer
2319 * @lenp: the size of the user buffer
2320 * @ppos: file position
2321 *
2322 * Reads/writes a string from/to the user buffer. If the kernel
2323 * buffer provided is not large enough to hold the string, the
2324 * string is truncated. The copied string is %NULL-terminated.
2325 * If the string is being read by the user process, it is copied
2326 * and a newline '\n' is added. It is truncated if the buffer is
2327 * not large enough.
2328 *
2329 * Returns 0 on success.
2330 */
2331 int proc_dostring(struct ctl_table *table, int write,
2332 void __user *buffer, size_t *lenp, loff_t *ppos)
2333 {
2334 return _proc_do_string(table->data, table->maxlen, write,
2335 buffer, lenp, ppos);
2336 }
2337
2338
2339 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2340 int *valp,
2341 int write, void *data)
2342 {
2343 if (write) {
2344 *valp = *negp ? -*lvalp : *lvalp;
2345 } else {
2346 int val = *valp;
2347 if (val < 0) {
2348 *negp = -1;
2349 *lvalp = (unsigned long)-val;
2350 } else {
2351 *negp = 0;
2352 *lvalp = (unsigned long)val;
2353 }
2354 }
2355 return 0;
2356 }
2357
2358 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2359 int write, void __user *buffer,
2360 size_t *lenp, loff_t *ppos,
2361 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2362 int write, void *data),
2363 void *data)
2364 {
2365 #define TMPBUFLEN 21
2366 int *i, vleft, first = 1, neg;
2367 unsigned long lval;
2368 size_t left, len;
2369
2370 char buf[TMPBUFLEN], *p;
2371 char __user *s = buffer;
2372
2373 if (!tbl_data || !table->maxlen || !*lenp ||
2374 (*ppos && !write)) {
2375 *lenp = 0;
2376 return 0;
2377 }
2378
2379 i = (int *) tbl_data;
2380 vleft = table->maxlen / sizeof(*i);
2381 left = *lenp;
2382
2383 if (!conv)
2384 conv = do_proc_dointvec_conv;
2385
2386 for (; left && vleft--; i++, first=0) {
2387 if (write) {
2388 while (left) {
2389 char c;
2390 if (get_user(c, s))
2391 return -EFAULT;
2392 if (!isspace(c))
2393 break;
2394 left--;
2395 s++;
2396 }
2397 if (!left)
2398 break;
2399 neg = 0;
2400 len = left;
2401 if (len > sizeof(buf) - 1)
2402 len = sizeof(buf) - 1;
2403 if (copy_from_user(buf, s, len))
2404 return -EFAULT;
2405 buf[len] = 0;
2406 p = buf;
2407 if (*p == '-' && left > 1) {
2408 neg = 1;
2409 p++;
2410 }
2411 if (*p < '0' || *p > '9')
2412 break;
2413
2414 lval = simple_strtoul(p, &p, 0);
2415
2416 len = p-buf;
2417 if ((len < left) && *p && !isspace(*p))
2418 break;
2419 s += len;
2420 left -= len;
2421
2422 if (conv(&neg, &lval, i, 1, data))
2423 break;
2424 } else {
2425 p = buf;
2426 if (!first)
2427 *p++ = '\t';
2428
2429 if (conv(&neg, &lval, i, 0, data))
2430 break;
2431
2432 sprintf(p, "%s%lu", neg ? "-" : "", lval);
2433 len = strlen(buf);
2434 if (len > left)
2435 len = left;
2436 if(copy_to_user(s, buf, len))
2437 return -EFAULT;
2438 left -= len;
2439 s += len;
2440 }
2441 }
2442
2443 if (!write && !first && left) {
2444 if(put_user('\n', s))
2445 return -EFAULT;
2446 left--, s++;
2447 }
2448 if (write) {
2449 while (left) {
2450 char c;
2451 if (get_user(c, s++))
2452 return -EFAULT;
2453 if (!isspace(c))
2454 break;
2455 left--;
2456 }
2457 }
2458 if (write && first)
2459 return -EINVAL;
2460 *lenp -= left;
2461 *ppos += *lenp;
2462 return 0;
2463 #undef TMPBUFLEN
2464 }
2465
2466 static int do_proc_dointvec(struct ctl_table *table, int write,
2467 void __user *buffer, size_t *lenp, loff_t *ppos,
2468 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2469 int write, void *data),
2470 void *data)
2471 {
2472 return __do_proc_dointvec(table->data, table, write,
2473 buffer, lenp, ppos, conv, data);
2474 }
2475
2476 /**
2477 * proc_dointvec - read a vector of integers
2478 * @table: the sysctl table
2479 * @write: %TRUE if this is a write to the sysctl file
2480 * @buffer: the user buffer
2481 * @lenp: the size of the user buffer
2482 * @ppos: file position
2483 *
2484 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2485 * values from/to the user buffer, treated as an ASCII string.
2486 *
2487 * Returns 0 on success.
2488 */
2489 int proc_dointvec(struct ctl_table *table, int write,
2490 void __user *buffer, size_t *lenp, loff_t *ppos)
2491 {
2492 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2493 NULL,NULL);
2494 }
2495
2496 /*
2497 * Taint values can only be increased
2498 * This means we can safely use a temporary.
2499 */
2500 static int proc_taint(struct ctl_table *table, int write,
2501 void __user *buffer, size_t *lenp, loff_t *ppos)
2502 {
2503 struct ctl_table t;
2504 unsigned long tmptaint = get_taint();
2505 int err;
2506
2507 if (write && !capable(CAP_SYS_ADMIN))
2508 return -EPERM;
2509
2510 t = *table;
2511 t.data = &tmptaint;
2512 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2513 if (err < 0)
2514 return err;
2515
2516 if (write) {
2517 /*
2518 * Poor man's atomic or. Not worth adding a primitive
2519 * to everyone's atomic.h for this
2520 */
2521 int i;
2522 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2523 if ((tmptaint >> i) & 1)
2524 add_taint(i);
2525 }
2526 }
2527
2528 return err;
2529 }
2530
2531 struct do_proc_dointvec_minmax_conv_param {
2532 int *min;
2533 int *max;
2534 };
2535
2536 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2537 int *valp,
2538 int write, void *data)
2539 {
2540 struct do_proc_dointvec_minmax_conv_param *param = data;
2541 if (write) {
2542 int val = *negp ? -*lvalp : *lvalp;
2543 if ((param->min && *param->min > val) ||
2544 (param->max && *param->max < val))
2545 return -EINVAL;
2546 *valp = val;
2547 } else {
2548 int val = *valp;
2549 if (val < 0) {
2550 *negp = -1;
2551 *lvalp = (unsigned long)-val;
2552 } else {
2553 *negp = 0;
2554 *lvalp = (unsigned long)val;
2555 }
2556 }
2557 return 0;
2558 }
2559
2560 /**
2561 * proc_dointvec_minmax - read a vector of integers with min/max values
2562 * @table: the sysctl table
2563 * @write: %TRUE if this is a write to the sysctl file
2564 * @buffer: the user buffer
2565 * @lenp: the size of the user buffer
2566 * @ppos: file position
2567 *
2568 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2569 * values from/to the user buffer, treated as an ASCII string.
2570 *
2571 * This routine will ensure the values are within the range specified by
2572 * table->extra1 (min) and table->extra2 (max).
2573 *
2574 * Returns 0 on success.
2575 */
2576 int proc_dointvec_minmax(struct ctl_table *table, int write,
2577 void __user *buffer, size_t *lenp, loff_t *ppos)
2578 {
2579 struct do_proc_dointvec_minmax_conv_param param = {
2580 .min = (int *) table->extra1,
2581 .max = (int *) table->extra2,
2582 };
2583 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2584 do_proc_dointvec_minmax_conv, &param);
2585 }
2586
2587 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2588 void __user *buffer,
2589 size_t *lenp, loff_t *ppos,
2590 unsigned long convmul,
2591 unsigned long convdiv)
2592 {
2593 #define TMPBUFLEN 21
2594 unsigned long *i, *min, *max, val;
2595 int vleft, first=1, neg;
2596 size_t len, left;
2597 char buf[TMPBUFLEN], *p;
2598 char __user *s = buffer;
2599
2600 if (!data || !table->maxlen || !*lenp ||
2601 (*ppos && !write)) {
2602 *lenp = 0;
2603 return 0;
2604 }
2605
2606 i = (unsigned long *) data;
2607 min = (unsigned long *) table->extra1;
2608 max = (unsigned long *) table->extra2;
2609 vleft = table->maxlen / sizeof(unsigned long);
2610 left = *lenp;
2611
2612 for (; left && vleft--; i++, min++, max++, first=0) {
2613 if (write) {
2614 while (left) {
2615 char c;
2616 if (get_user(c, s))
2617 return -EFAULT;
2618 if (!isspace(c))
2619 break;
2620 left--;
2621 s++;
2622 }
2623 if (!left)
2624 break;
2625 neg = 0;
2626 len = left;
2627 if (len > TMPBUFLEN-1)
2628 len = TMPBUFLEN-1;
2629 if (copy_from_user(buf, s, len))
2630 return -EFAULT;
2631 buf[len] = 0;
2632 p = buf;
2633 if (*p == '-' && left > 1) {
2634 neg = 1;
2635 p++;
2636 }
2637 if (*p < '0' || *p > '9')
2638 break;
2639 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2640 len = p-buf;
2641 if ((len < left) && *p && !isspace(*p))
2642 break;
2643 if (neg)
2644 val = -val;
2645 s += len;
2646 left -= len;
2647
2648 if(neg)
2649 continue;
2650 if ((min && val < *min) || (max && val > *max))
2651 continue;
2652 *i = val;
2653 } else {
2654 p = buf;
2655 if (!first)
2656 *p++ = '\t';
2657 sprintf(p, "%lu", convdiv * (*i) / convmul);
2658 len = strlen(buf);
2659 if (len > left)
2660 len = left;
2661 if(copy_to_user(s, buf, len))
2662 return -EFAULT;
2663 left -= len;
2664 s += len;
2665 }
2666 }
2667
2668 if (!write && !first && left) {
2669 if(put_user('\n', s))
2670 return -EFAULT;
2671 left--, s++;
2672 }
2673 if (write) {
2674 while (left) {
2675 char c;
2676 if (get_user(c, s++))
2677 return -EFAULT;
2678 if (!isspace(c))
2679 break;
2680 left--;
2681 }
2682 }
2683 if (write && first)
2684 return -EINVAL;
2685 *lenp -= left;
2686 *ppos += *lenp;
2687 return 0;
2688 #undef TMPBUFLEN
2689 }
2690
2691 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2692 void __user *buffer,
2693 size_t *lenp, loff_t *ppos,
2694 unsigned long convmul,
2695 unsigned long convdiv)
2696 {
2697 return __do_proc_doulongvec_minmax(table->data, table, write,
2698 buffer, lenp, ppos, convmul, convdiv);
2699 }
2700
2701 /**
2702 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2703 * @table: the sysctl table
2704 * @write: %TRUE if this is a write to the sysctl file
2705 * @buffer: the user buffer
2706 * @lenp: the size of the user buffer
2707 * @ppos: file position
2708 *
2709 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2710 * values from/to the user buffer, treated as an ASCII string.
2711 *
2712 * This routine will ensure the values are within the range specified by
2713 * table->extra1 (min) and table->extra2 (max).
2714 *
2715 * Returns 0 on success.
2716 */
2717 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2718 void __user *buffer, size_t *lenp, loff_t *ppos)
2719 {
2720 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2721 }
2722
2723 /**
2724 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2725 * @table: the sysctl table
2726 * @write: %TRUE if this is a write to the sysctl file
2727 * @buffer: the user buffer
2728 * @lenp: the size of the user buffer
2729 * @ppos: file position
2730 *
2731 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2732 * values from/to the user buffer, treated as an ASCII string. The values
2733 * are treated as milliseconds, and converted to jiffies when they are stored.
2734 *
2735 * This routine will ensure the values are within the range specified by
2736 * table->extra1 (min) and table->extra2 (max).
2737 *
2738 * Returns 0 on success.
2739 */
2740 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2741 void __user *buffer,
2742 size_t *lenp, loff_t *ppos)
2743 {
2744 return do_proc_doulongvec_minmax(table, write, buffer,
2745 lenp, ppos, HZ, 1000l);
2746 }
2747
2748
2749 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2750 int *valp,
2751 int write, void *data)
2752 {
2753 if (write) {
2754 if (*lvalp > LONG_MAX / HZ)
2755 return 1;
2756 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2757 } else {
2758 int val = *valp;
2759 unsigned long lval;
2760 if (val < 0) {
2761 *negp = -1;
2762 lval = (unsigned long)-val;
2763 } else {
2764 *negp = 0;
2765 lval = (unsigned long)val;
2766 }
2767 *lvalp = lval / HZ;
2768 }
2769 return 0;
2770 }
2771
2772 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2773 int *valp,
2774 int write, void *data)
2775 {
2776 if (write) {
2777 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2778 return 1;
2779 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2780 } else {
2781 int val = *valp;
2782 unsigned long lval;
2783 if (val < 0) {
2784 *negp = -1;
2785 lval = (unsigned long)-val;
2786 } else {
2787 *negp = 0;
2788 lval = (unsigned long)val;
2789 }
2790 *lvalp = jiffies_to_clock_t(lval);
2791 }
2792 return 0;
2793 }
2794
2795 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2796 int *valp,
2797 int write, void *data)
2798 {
2799 if (write) {
2800 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2801 } else {
2802 int val = *valp;
2803 unsigned long lval;
2804 if (val < 0) {
2805 *negp = -1;
2806 lval = (unsigned long)-val;
2807 } else {
2808 *negp = 0;
2809 lval = (unsigned long)val;
2810 }
2811 *lvalp = jiffies_to_msecs(lval);
2812 }
2813 return 0;
2814 }
2815
2816 /**
2817 * proc_dointvec_jiffies - read a vector of integers as seconds
2818 * @table: the sysctl table
2819 * @write: %TRUE if this is a write to the sysctl file
2820 * @buffer: the user buffer
2821 * @lenp: the size of the user buffer
2822 * @ppos: file position
2823 *
2824 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2825 * values from/to the user buffer, treated as an ASCII string.
2826 * The values read are assumed to be in seconds, and are converted into
2827 * jiffies.
2828 *
2829 * Returns 0 on success.
2830 */
2831 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2832 void __user *buffer, size_t *lenp, loff_t *ppos)
2833 {
2834 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2835 do_proc_dointvec_jiffies_conv,NULL);
2836 }
2837
2838 /**
2839 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2840 * @table: the sysctl table
2841 * @write: %TRUE if this is a write to the sysctl file
2842 * @buffer: the user buffer
2843 * @lenp: the size of the user buffer
2844 * @ppos: pointer to the file position
2845 *
2846 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2847 * values from/to the user buffer, treated as an ASCII string.
2848 * The values read are assumed to be in 1/USER_HZ seconds, and
2849 * are converted into jiffies.
2850 *
2851 * Returns 0 on success.
2852 */
2853 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2854 void __user *buffer, size_t *lenp, loff_t *ppos)
2855 {
2856 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2857 do_proc_dointvec_userhz_jiffies_conv,NULL);
2858 }
2859
2860 /**
2861 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2862 * @table: the sysctl table
2863 * @write: %TRUE if this is a write to the sysctl file
2864 * @buffer: the user buffer
2865 * @lenp: the size of the user buffer
2866 * @ppos: file position
2867 * @ppos: the current position in the file
2868 *
2869 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2870 * values from/to the user buffer, treated as an ASCII string.
2871 * The values read are assumed to be in 1/1000 seconds, and
2872 * are converted into jiffies.
2873 *
2874 * Returns 0 on success.
2875 */
2876 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2877 void __user *buffer, size_t *lenp, loff_t *ppos)
2878 {
2879 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2880 do_proc_dointvec_ms_jiffies_conv, NULL);
2881 }
2882
2883 static int proc_do_cad_pid(struct ctl_table *table, int write,
2884 void __user *buffer, size_t *lenp, loff_t *ppos)
2885 {
2886 struct pid *new_pid;
2887 pid_t tmp;
2888 int r;
2889
2890 tmp = pid_vnr(cad_pid);
2891
2892 r = __do_proc_dointvec(&tmp, table, write, buffer,
2893 lenp, ppos, NULL, NULL);
2894 if (r || !write)
2895 return r;
2896
2897 new_pid = find_get_pid(tmp);
2898 if (!new_pid)
2899 return -ESRCH;
2900
2901 put_pid(xchg(&cad_pid, new_pid));
2902 return 0;
2903 }
2904
2905 #else /* CONFIG_PROC_FS */
2906
2907 int proc_dostring(struct ctl_table *table, int write,
2908 void __user *buffer, size_t *lenp, loff_t *ppos)
2909 {
2910 return -ENOSYS;
2911 }
2912
2913 int proc_dointvec(struct ctl_table *table, int write,
2914 void __user *buffer, size_t *lenp, loff_t *ppos)
2915 {
2916 return -ENOSYS;
2917 }
2918
2919 int proc_dointvec_minmax(struct ctl_table *table, int write,
2920 void __user *buffer, size_t *lenp, loff_t *ppos)
2921 {
2922 return -ENOSYS;
2923 }
2924
2925 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2926 void __user *buffer, size_t *lenp, loff_t *ppos)
2927 {
2928 return -ENOSYS;
2929 }
2930
2931 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2932 void __user *buffer, size_t *lenp, loff_t *ppos)
2933 {
2934 return -ENOSYS;
2935 }
2936
2937 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2938 void __user *buffer, size_t *lenp, loff_t *ppos)
2939 {
2940 return -ENOSYS;
2941 }
2942
2943 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2944 void __user *buffer, size_t *lenp, loff_t *ppos)
2945 {
2946 return -ENOSYS;
2947 }
2948
2949 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2950 void __user *buffer,
2951 size_t *lenp, loff_t *ppos)
2952 {
2953 return -ENOSYS;
2954 }
2955
2956
2957 #endif /* CONFIG_PROC_FS */
2958
2959
2960 #ifdef CONFIG_SYSCTL_SYSCALL
2961 /*
2962 * General sysctl support routines
2963 */
2964
2965 /* The generic sysctl data routine (used if no strategy routine supplied) */
2966 int sysctl_data(struct ctl_table *table,
2967 void __user *oldval, size_t __user *oldlenp,
2968 void __user *newval, size_t newlen)
2969 {
2970 size_t len;
2971
2972 /* Get out of I don't have a variable */
2973 if (!table->data || !table->maxlen)
2974 return -ENOTDIR;
2975
2976 if (oldval && oldlenp) {
2977 if (get_user(len, oldlenp))
2978 return -EFAULT;
2979 if (len) {
2980 if (len > table->maxlen)
2981 len = table->maxlen;
2982 if (copy_to_user(oldval, table->data, len))
2983 return -EFAULT;
2984 if (put_user(len, oldlenp))
2985 return -EFAULT;
2986 }
2987 }
2988
2989 if (newval && newlen) {
2990 if (newlen > table->maxlen)
2991 newlen = table->maxlen;
2992
2993 if (copy_from_user(table->data, newval, newlen))
2994 return -EFAULT;
2995 }
2996 return 1;
2997 }
2998
2999 /* The generic string strategy routine: */
3000 int sysctl_string(struct ctl_table *table,
3001 void __user *oldval, size_t __user *oldlenp,
3002 void __user *newval, size_t newlen)
3003 {
3004 if (!table->data || !table->maxlen)
3005 return -ENOTDIR;
3006
3007 if (oldval && oldlenp) {
3008 size_t bufsize;
3009 if (get_user(bufsize, oldlenp))
3010 return -EFAULT;
3011 if (bufsize) {
3012 size_t len = strlen(table->data), copied;
3013
3014 /* This shouldn't trigger for a well-formed sysctl */
3015 if (len > table->maxlen)
3016 len = table->maxlen;
3017
3018 /* Copy up to a max of bufsize-1 bytes of the string */
3019 copied = (len >= bufsize) ? bufsize - 1 : len;
3020
3021 if (copy_to_user(oldval, table->data, copied) ||
3022 put_user(0, (char __user *)(oldval + copied)))
3023 return -EFAULT;
3024 if (put_user(len, oldlenp))
3025 return -EFAULT;
3026 }
3027 }
3028 if (newval && newlen) {
3029 size_t len = newlen;
3030 if (len > table->maxlen)
3031 len = table->maxlen;
3032 if(copy_from_user(table->data, newval, len))
3033 return -EFAULT;
3034 if (len == table->maxlen)
3035 len--;
3036 ((char *) table->data)[len] = 0;
3037 }
3038 return 1;
3039 }
3040
3041 /*
3042 * This function makes sure that all of the integers in the vector
3043 * are between the minimum and maximum values given in the arrays
3044 * table->extra1 and table->extra2, respectively.
3045 */
3046 int sysctl_intvec(struct ctl_table *table,
3047 void __user *oldval, size_t __user *oldlenp,
3048 void __user *newval, size_t newlen)
3049 {
3050
3051 if (newval && newlen) {
3052 int __user *vec = (int __user *) newval;
3053 int *min = (int *) table->extra1;
3054 int *max = (int *) table->extra2;
3055 size_t length;
3056 int i;
3057
3058 if (newlen % sizeof(int) != 0)
3059 return -EINVAL;
3060
3061 if (!table->extra1 && !table->extra2)
3062 return 0;
3063
3064 if (newlen > table->maxlen)
3065 newlen = table->maxlen;
3066 length = newlen / sizeof(int);
3067
3068 for (i = 0; i < length; i++) {
3069 int value;
3070 if (get_user(value, vec + i))
3071 return -EFAULT;
3072 if (min && value < min[i])
3073 return -EINVAL;
3074 if (max && value > max[i])
3075 return -EINVAL;
3076 }
3077 }
3078 return 0;
3079 }
3080
3081 /* Strategy function to convert jiffies to seconds */
3082 int sysctl_jiffies(struct ctl_table *table,
3083 void __user *oldval, size_t __user *oldlenp,
3084 void __user *newval, size_t newlen)
3085 {
3086 if (oldval && oldlenp) {
3087 size_t olen;
3088
3089 if (get_user(olen, oldlenp))
3090 return -EFAULT;
3091 if (olen) {
3092 int val;
3093
3094 if (olen < sizeof(int))
3095 return -EINVAL;
3096
3097 val = *(int *)(table->data) / HZ;
3098 if (put_user(val, (int __user *)oldval))
3099 return -EFAULT;
3100 if (put_user(sizeof(int), oldlenp))
3101 return -EFAULT;
3102 }
3103 }
3104 if (newval && newlen) {
3105 int new;
3106 if (newlen != sizeof(int))
3107 return -EINVAL;
3108 if (get_user(new, (int __user *)newval))
3109 return -EFAULT;
3110 *(int *)(table->data) = new*HZ;
3111 }
3112 return 1;
3113 }
3114
3115 /* Strategy function to convert jiffies to seconds */
3116 int sysctl_ms_jiffies(struct ctl_table *table,
3117 void __user *oldval, size_t __user *oldlenp,
3118 void __user *newval, size_t newlen)
3119 {
3120 if (oldval && oldlenp) {
3121 size_t olen;
3122
3123 if (get_user(olen, oldlenp))
3124 return -EFAULT;
3125 if (olen) {
3126 int val;
3127
3128 if (olen < sizeof(int))
3129 return -EINVAL;
3130
3131 val = jiffies_to_msecs(*(int *)(table->data));
3132 if (put_user(val, (int __user *)oldval))
3133 return -EFAULT;
3134 if (put_user(sizeof(int), oldlenp))
3135 return -EFAULT;
3136 }
3137 }
3138 if (newval && newlen) {
3139 int new;
3140 if (newlen != sizeof(int))
3141 return -EINVAL;
3142 if (get_user(new, (int __user *)newval))
3143 return -EFAULT;
3144 *(int *)(table->data) = msecs_to_jiffies(new);
3145 }
3146 return 1;
3147 }
3148
3149
3150
3151 #else /* CONFIG_SYSCTL_SYSCALL */
3152
3153
3154 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
3155 {
3156 struct __sysctl_args tmp;
3157 int error;
3158
3159 if (copy_from_user(&tmp, args, sizeof(tmp)))
3160 return -EFAULT;
3161
3162 error = deprecated_sysctl_warning(&tmp);
3163
3164 /* If no error reading the parameters then just -ENOSYS ... */
3165 if (!error)
3166 error = -ENOSYS;
3167
3168 return error;
3169 }
3170
3171 int sysctl_data(struct ctl_table *table,
3172 void __user *oldval, size_t __user *oldlenp,
3173 void __user *newval, size_t newlen)
3174 {
3175 return -ENOSYS;
3176 }
3177
3178 int sysctl_string(struct ctl_table *table,
3179 void __user *oldval, size_t __user *oldlenp,
3180 void __user *newval, size_t newlen)
3181 {
3182 return -ENOSYS;
3183 }
3184
3185 int sysctl_intvec(struct ctl_table *table,
3186 void __user *oldval, size_t __user *oldlenp,
3187 void __user *newval, size_t newlen)
3188 {
3189 return -ENOSYS;
3190 }
3191
3192 int sysctl_jiffies(struct ctl_table *table,
3193 void __user *oldval, size_t __user *oldlenp,
3194 void __user *newval, size_t newlen)
3195 {
3196 return -ENOSYS;
3197 }
3198
3199 int sysctl_ms_jiffies(struct ctl_table *table,
3200 void __user *oldval, size_t __user *oldlenp,
3201 void __user *newval, size_t newlen)
3202 {
3203 return -ENOSYS;
3204 }
3205
3206 #endif /* CONFIG_SYSCTL_SYSCALL */
3207
3208 static int deprecated_sysctl_warning(struct __sysctl_args *args)
3209 {
3210 static int msg_count;
3211 int name[CTL_MAXNAME];
3212 int i;
3213
3214 /* Check args->nlen. */
3215 if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
3216 return -ENOTDIR;
3217
3218 /* Read in the sysctl name for better debug message logging */
3219 for (i = 0; i < args->nlen; i++)
3220 if (get_user(name[i], args->name + i))
3221 return -EFAULT;
3222
3223 /* Ignore accesses to kernel.version */
3224 if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
3225 return 0;
3226
3227 if (msg_count < 5) {
3228 msg_count++;
3229 printk(KERN_INFO
3230 "warning: process `%s' used the deprecated sysctl "
3231 "system call with ", current->comm);
3232 for (i = 0; i < args->nlen; i++)
3233 printk("%d.", name[i]);
3234 printk("\n");
3235 }
3236 return 0;
3237 }
3238
3239 /*
3240 * No sense putting this after each symbol definition, twice,
3241 * exception granted :-)
3242 */
3243 EXPORT_SYMBOL(proc_dointvec);
3244 EXPORT_SYMBOL(proc_dointvec_jiffies);
3245 EXPORT_SYMBOL(proc_dointvec_minmax);
3246 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3247 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3248 EXPORT_SYMBOL(proc_dostring);
3249 EXPORT_SYMBOL(proc_doulongvec_minmax);
3250 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3251 EXPORT_SYMBOL(register_sysctl_table);
3252 EXPORT_SYMBOL(register_sysctl_paths);
3253 EXPORT_SYMBOL(sysctl_intvec);
3254 EXPORT_SYMBOL(sysctl_jiffies);
3255 EXPORT_SYMBOL(sysctl_ms_jiffies);
3256 EXPORT_SYMBOL(sysctl_string);
3257 EXPORT_SYMBOL(sysctl_data);
3258 EXPORT_SYMBOL(unregister_sysctl_table);
Linux with added FPC-III support