?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
ring-buffer-benchmark: Add parameters to set produce/consumer priorities
[linux-2.6/linux-acpi-2.6/ibm-acpi-2.6.git] / kernel / sysctl.c
blob0d949c517412ee16822a5ca7d6e7c79218543741
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 {
1349 .ctl_name = CTL_UNNUMBERED,
1350 .procname = "mmap_min_addr",
1351 .data = &dac_mmap_min_addr,
1352 .maxlen = sizeof(unsigned long),
1353 .mode = 0644,
1354 .proc_handler = &mmap_min_addr_handler,
1355 },
1356 #ifdef CONFIG_NUMA
1357 {
1358 .ctl_name = CTL_UNNUMBERED,
1359 .procname = "numa_zonelist_order",
1360 .data = &numa_zonelist_order,
1361 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1362 .mode = 0644,
1363 .proc_handler = &numa_zonelist_order_handler,
1364 .strategy = &sysctl_string,
1365 },
1366 #endif
1367 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1368 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1369 {
1370 .ctl_name = VM_VDSO_ENABLED,
1371 .procname = "vdso_enabled",
1372 .data = &vdso_enabled,
1373 .maxlen = sizeof(vdso_enabled),
1374 .mode = 0644,
1375 .proc_handler = &proc_dointvec,
1376 .strategy = &sysctl_intvec,
1377 .extra1 = &zero,
1378 },
1379 #endif
1380 #ifdef CONFIG_HIGHMEM
1381 {
1382 .ctl_name = CTL_UNNUMBERED,
1383 .procname = "highmem_is_dirtyable",
1384 .data = &vm_highmem_is_dirtyable,
1385 .maxlen = sizeof(vm_highmem_is_dirtyable),
1386 .mode = 0644,
1387 .proc_handler = &proc_dointvec_minmax,
1388 .strategy = &sysctl_intvec,
1389 .extra1 = &zero,
1390 .extra2 = &one,
1391 },
1392 #endif
1393 {
1394 .ctl_name = CTL_UNNUMBERED,
1395 .procname = "scan_unevictable_pages",
1396 .data = &scan_unevictable_pages,
1397 .maxlen = sizeof(scan_unevictable_pages),
1398 .mode = 0644,
1399 .proc_handler = &scan_unevictable_handler,
1400 },
1401 #ifdef CONFIG_MEMORY_FAILURE
1402 {
1403 .ctl_name = CTL_UNNUMBERED,
1404 .procname = "memory_failure_early_kill",
1405 .data = &sysctl_memory_failure_early_kill,
1406 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1407 .mode = 0644,
1408 .proc_handler = &proc_dointvec_minmax,
1409 .strategy = &sysctl_intvec,
1410 .extra1 = &zero,
1411 .extra2 = &one,
1412 },
1413 {
1414 .ctl_name = CTL_UNNUMBERED,
1415 .procname = "memory_failure_recovery",
1416 .data = &sysctl_memory_failure_recovery,
1417 .maxlen = sizeof(sysctl_memory_failure_recovery),
1418 .mode = 0644,
1419 .proc_handler = &proc_dointvec_minmax,
1420 .strategy = &sysctl_intvec,
1421 .extra1 = &zero,
1422 .extra2 = &one,
1423 },
1424 #endif
1425
1426 /*
1427 * NOTE: do not add new entries to this table unless you have read
1428 * Documentation/sysctl/ctl_unnumbered.txt
1429 */
1430 { .ctl_name = 0 }
1431 };
1432
1433 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1434 static struct ctl_table binfmt_misc_table[] = {
1435 { .ctl_name = 0 }
1436 };
1437 #endif
1438
1439 static struct ctl_table fs_table[] = {
1440 {
1441 .ctl_name = FS_NRINODE,
1442 .procname = "inode-nr",
1443 .data = &inodes_stat,
1444 .maxlen = 2*sizeof(int),
1445 .mode = 0444,
1446 .proc_handler = &proc_dointvec,
1447 },
1448 {
1449 .ctl_name = FS_STATINODE,
1450 .procname = "inode-state",
1451 .data = &inodes_stat,
1452 .maxlen = 7*sizeof(int),
1453 .mode = 0444,
1454 .proc_handler = &proc_dointvec,
1455 },
1456 {
1457 .procname = "file-nr",
1458 .data = &files_stat,
1459 .maxlen = 3*sizeof(int),
1460 .mode = 0444,
1461 .proc_handler = &proc_nr_files,
1462 },
1463 {
1464 .ctl_name = FS_MAXFILE,
1465 .procname = "file-max",
1466 .data = &files_stat.max_files,
1467 .maxlen = sizeof(int),
1468 .mode = 0644,
1469 .proc_handler = &proc_dointvec,
1470 },
1471 {
1472 .ctl_name = CTL_UNNUMBERED,
1473 .procname = "nr_open",
1474 .data = &sysctl_nr_open,
1475 .maxlen = sizeof(int),
1476 .mode = 0644,
1477 .proc_handler = &proc_dointvec_minmax,
1478 .extra1 = &sysctl_nr_open_min,
1479 .extra2 = &sysctl_nr_open_max,
1480 },
1481 {
1482 .ctl_name = FS_DENTRY,
1483 .procname = "dentry-state",
1484 .data = &dentry_stat,
1485 .maxlen = 6*sizeof(int),
1486 .mode = 0444,
1487 .proc_handler = &proc_dointvec,
1488 },
1489 {
1490 .ctl_name = FS_OVERFLOWUID,
1491 .procname = "overflowuid",
1492 .data = &fs_overflowuid,
1493 .maxlen = sizeof(int),
1494 .mode = 0644,
1495 .proc_handler = &proc_dointvec_minmax,
1496 .strategy = &sysctl_intvec,
1497 .extra1 = &minolduid,
1498 .extra2 = &maxolduid,
1499 },
1500 {
1501 .ctl_name = FS_OVERFLOWGID,
1502 .procname = "overflowgid",
1503 .data = &fs_overflowgid,
1504 .maxlen = sizeof(int),
1505 .mode = 0644,
1506 .proc_handler = &proc_dointvec_minmax,
1507 .strategy = &sysctl_intvec,
1508 .extra1 = &minolduid,
1509 .extra2 = &maxolduid,
1510 },
1511 #ifdef CONFIG_FILE_LOCKING
1512 {
1513 .ctl_name = FS_LEASES,
1514 .procname = "leases-enable",
1515 .data = &leases_enable,
1516 .maxlen = sizeof(int),
1517 .mode = 0644,
1518 .proc_handler = &proc_dointvec,
1519 },
1520 #endif
1521 #ifdef CONFIG_DNOTIFY
1522 {
1523 .ctl_name = FS_DIR_NOTIFY,
1524 .procname = "dir-notify-enable",
1525 .data = &dir_notify_enable,
1526 .maxlen = sizeof(int),
1527 .mode = 0644,
1528 .proc_handler = &proc_dointvec,
1529 },
1530 #endif
1531 #ifdef CONFIG_MMU
1532 #ifdef CONFIG_FILE_LOCKING
1533 {
1534 .ctl_name = FS_LEASE_TIME,
1535 .procname = "lease-break-time",
1536 .data = &lease_break_time,
1537 .maxlen = sizeof(int),
1538 .mode = 0644,
1539 .proc_handler = &proc_dointvec,
1540 },
1541 #endif
1542 #ifdef CONFIG_AIO
1543 {
1544 .procname = "aio-nr",
1545 .data = &aio_nr,
1546 .maxlen = sizeof(aio_nr),
1547 .mode = 0444,
1548 .proc_handler = &proc_doulongvec_minmax,
1549 },
1550 {
1551 .procname = "aio-max-nr",
1552 .data = &aio_max_nr,
1553 .maxlen = sizeof(aio_max_nr),
1554 .mode = 0644,
1555 .proc_handler = &proc_doulongvec_minmax,
1556 },
1557 #endif /* CONFIG_AIO */
1558 #ifdef CONFIG_INOTIFY_USER
1559 {
1560 .ctl_name = FS_INOTIFY,
1561 .procname = "inotify",
1562 .mode = 0555,
1563 .child = inotify_table,
1564 },
1565 #endif
1566 #ifdef CONFIG_EPOLL
1567 {
1568 .procname = "epoll",
1569 .mode = 0555,
1570 .child = epoll_table,
1571 },
1572 #endif
1573 #endif
1574 {
1575 .ctl_name = KERN_SETUID_DUMPABLE,
1576 .procname = "suid_dumpable",
1577 .data = &suid_dumpable,
1578 .maxlen = sizeof(int),
1579 .mode = 0644,
1580 .proc_handler = &proc_dointvec_minmax,
1581 .strategy = &sysctl_intvec,
1582 .extra1 = &zero,
1583 .extra2 = &two,
1584 },
1585 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1586 {
1587 .ctl_name = CTL_UNNUMBERED,
1588 .procname = "binfmt_misc",
1589 .mode = 0555,
1590 .child = binfmt_misc_table,
1591 },
1592 #endif
1593 /*
1594 * NOTE: do not add new entries to this table unless you have read
1595 * Documentation/sysctl/ctl_unnumbered.txt
1596 */
1597 { .ctl_name = 0 }
1598 };
1599
1600 static struct ctl_table debug_table[] = {
1601 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1602 {
1603 .ctl_name = CTL_UNNUMBERED,
1604 .procname = "exception-trace",
1605 .data = &show_unhandled_signals,
1606 .maxlen = sizeof(int),
1607 .mode = 0644,
1608 .proc_handler = proc_dointvec
1609 },
1610 #endif
1611 { .ctl_name = 0 }
1612 };
1613
1614 static struct ctl_table dev_table[] = {
1615 { .ctl_name = 0 }
1616 };
1617
1618 static DEFINE_SPINLOCK(sysctl_lock);
1619
1620 /* called under sysctl_lock */
1621 static int use_table(struct ctl_table_header *p)
1622 {
1623 if (unlikely(p->unregistering))
1624 return 0;
1625 p->used++;
1626 return 1;
1627 }
1628
1629 /* called under sysctl_lock */
1630 static void unuse_table(struct ctl_table_header *p)
1631 {
1632 if (!--p->used)
1633 if (unlikely(p->unregistering))
1634 complete(p->unregistering);
1635 }
1636
1637 /* called under sysctl_lock, will reacquire if has to wait */
1638 static void start_unregistering(struct ctl_table_header *p)
1639 {
1640 /*
1641 * if p->used is 0, nobody will ever touch that entry again;
1642 * we'll eliminate all paths to it before dropping sysctl_lock
1643 */
1644 if (unlikely(p->used)) {
1645 struct completion wait;
1646 init_completion(&wait);
1647 p->unregistering = &wait;
1648 spin_unlock(&sysctl_lock);
1649 wait_for_completion(&wait);
1650 spin_lock(&sysctl_lock);
1651 } else {
1652 /* anything non-NULL; we'll never dereference it */
1653 p->unregistering = ERR_PTR(-EINVAL);
1654 }
1655 /*
1656 * do not remove from the list until nobody holds it; walking the
1657 * list in do_sysctl() relies on that.
1658 */
1659 list_del_init(&p->ctl_entry);
1660 }
1661
1662 void sysctl_head_get(struct ctl_table_header *head)
1663 {
1664 spin_lock(&sysctl_lock);
1665 head->count++;
1666 spin_unlock(&sysctl_lock);
1667 }
1668
1669 void sysctl_head_put(struct ctl_table_header *head)
1670 {
1671 spin_lock(&sysctl_lock);
1672 if (!--head->count)
1673 kfree(head);
1674 spin_unlock(&sysctl_lock);
1675 }
1676
1677 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1678 {
1679 if (!head)
1680 BUG();
1681 spin_lock(&sysctl_lock);
1682 if (!use_table(head))
1683 head = ERR_PTR(-ENOENT);
1684 spin_unlock(&sysctl_lock);
1685 return head;
1686 }
1687
1688 void sysctl_head_finish(struct ctl_table_header *head)
1689 {
1690 if (!head)
1691 return;
1692 spin_lock(&sysctl_lock);
1693 unuse_table(head);
1694 spin_unlock(&sysctl_lock);
1695 }
1696
1697 static struct ctl_table_set *
1698 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1699 {
1700 struct ctl_table_set *set = &root->default_set;
1701 if (root->lookup)
1702 set = root->lookup(root, namespaces);
1703 return set;
1704 }
1705
1706 static struct list_head *
1707 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1708 {
1709 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1710 return &set->list;
1711 }
1712
1713 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1714 struct ctl_table_header *prev)
1715 {
1716 struct ctl_table_root *root;
1717 struct list_head *header_list;
1718 struct ctl_table_header *head;
1719 struct list_head *tmp;
1720
1721 spin_lock(&sysctl_lock);
1722 if (prev) {
1723 head = prev;
1724 tmp = &prev->ctl_entry;
1725 unuse_table(prev);
1726 goto next;
1727 }
1728 tmp = &root_table_header.ctl_entry;
1729 for (;;) {
1730 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1731
1732 if (!use_table(head))
1733 goto next;
1734 spin_unlock(&sysctl_lock);
1735 return head;
1736 next:
1737 root = head->root;
1738 tmp = tmp->next;
1739 header_list = lookup_header_list(root, namespaces);
1740 if (tmp != header_list)
1741 continue;
1742
1743 do {
1744 root = list_entry(root->root_list.next,
1745 struct ctl_table_root, root_list);
1746 if (root == &sysctl_table_root)
1747 goto out;
1748 header_list = lookup_header_list(root, namespaces);
1749 } while (list_empty(header_list));
1750 tmp = header_list->next;
1751 }
1752 out:
1753 spin_unlock(&sysctl_lock);
1754 return NULL;
1755 }
1756
1757 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1758 {
1759 return __sysctl_head_next(current->nsproxy, prev);
1760 }
1761
1762 void register_sysctl_root(struct ctl_table_root *root)
1763 {
1764 spin_lock(&sysctl_lock);
1765 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1766 spin_unlock(&sysctl_lock);
1767 }
1768
1769 #ifdef CONFIG_SYSCTL_SYSCALL
1770 /* Perform the actual read/write of a sysctl table entry. */
1771 static int do_sysctl_strategy(struct ctl_table_root *root,
1772 struct ctl_table *table,
1773 void __user *oldval, size_t __user *oldlenp,
1774 void __user *newval, size_t newlen)
1775 {
1776 int op = 0, rc;
1777
1778 if (oldval)
1779 op |= MAY_READ;
1780 if (newval)
1781 op |= MAY_WRITE;
1782 if (sysctl_perm(root, table, op))
1783 return -EPERM;
1784
1785 if (table->strategy) {
1786 rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1787 if (rc < 0)
1788 return rc;
1789 if (rc > 0)
1790 return 0;
1791 }
1792
1793 /* If there is no strategy routine, or if the strategy returns
1794 * zero, proceed with automatic r/w */
1795 if (table->data && table->maxlen) {
1796 rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1797 if (rc < 0)
1798 return rc;
1799 }
1800 return 0;
1801 }
1802
1803 static int parse_table(int __user *name, int nlen,
1804 void __user *oldval, size_t __user *oldlenp,
1805 void __user *newval, size_t newlen,
1806 struct ctl_table_root *root,
1807 struct ctl_table *table)
1808 {
1809 int n;
1810 repeat:
1811 if (!nlen)
1812 return -ENOTDIR;
1813 if (get_user(n, name))
1814 return -EFAULT;
1815 for ( ; table->ctl_name || table->procname; table++) {
1816 if (!table->ctl_name)
1817 continue;
1818 if (n == table->ctl_name) {
1819 int error;
1820 if (table->child) {
1821 if (sysctl_perm(root, table, MAY_EXEC))
1822 return -EPERM;
1823 name++;
1824 nlen--;
1825 table = table->child;
1826 goto repeat;
1827 }
1828 error = do_sysctl_strategy(root, table,
1829 oldval, oldlenp,
1830 newval, newlen);
1831 return error;
1832 }
1833 }
1834 return -ENOTDIR;
1835 }
1836
1837 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1838 void __user *newval, size_t newlen)
1839 {
1840 struct ctl_table_header *head;
1841 int error = -ENOTDIR;
1842
1843 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1844 return -ENOTDIR;
1845 if (oldval) {
1846 int old_len;
1847 if (!oldlenp || get_user(old_len, oldlenp))
1848 return -EFAULT;
1849 }
1850
1851 for (head = sysctl_head_next(NULL); head;
1852 head = sysctl_head_next(head)) {
1853 error = parse_table(name, nlen, oldval, oldlenp,
1854 newval, newlen,
1855 head->root, head->ctl_table);
1856 if (error != -ENOTDIR) {
1857 sysctl_head_finish(head);
1858 break;
1859 }
1860 }
1861 return error;
1862 }
1863
1864 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
1865 {
1866 struct __sysctl_args tmp;
1867 int error;
1868
1869 if (copy_from_user(&tmp, args, sizeof(tmp)))
1870 return -EFAULT;
1871
1872 error = deprecated_sysctl_warning(&tmp);
1873 if (error)
1874 goto out;
1875
1876 lock_kernel();
1877 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1878 tmp.newval, tmp.newlen);
1879 unlock_kernel();
1880 out:
1881 return error;
1882 }
1883 #endif /* CONFIG_SYSCTL_SYSCALL */
1884
1885 /*
1886 * sysctl_perm does NOT grant the superuser all rights automatically, because
1887 * some sysctl variables are readonly even to root.
1888 */
1889
1890 static int test_perm(int mode, int op)
1891 {
1892 if (!current_euid())
1893 mode >>= 6;
1894 else if (in_egroup_p(0))
1895 mode >>= 3;
1896 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1897 return 0;
1898 return -EACCES;
1899 }
1900
1901 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1902 {
1903 int error;
1904 int mode;
1905
1906 error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1907 if (error)
1908 return error;
1909
1910 if (root->permissions)
1911 mode = root->permissions(root, current->nsproxy, table);
1912 else
1913 mode = table->mode;
1914
1915 return test_perm(mode, op);
1916 }
1917
1918 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1919 {
1920 for (; table->ctl_name || table->procname; table++) {
1921 table->parent = parent;
1922 if (table->child)
1923 sysctl_set_parent(table, table->child);
1924 }
1925 }
1926
1927 static __init int sysctl_init(void)
1928 {
1929 sysctl_set_parent(NULL, root_table);
1930 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1931 {
1932 int err;
1933 err = sysctl_check_table(current->nsproxy, root_table);
1934 }
1935 #endif
1936 return 0;
1937 }
1938
1939 core_initcall(sysctl_init);
1940
1941 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1942 struct ctl_table *table)
1943 {
1944 struct ctl_table *p;
1945 const char *s = branch->procname;
1946
1947 /* branch should have named subdirectory as its first element */
1948 if (!s || !branch->child)
1949 return NULL;
1950
1951 /* ... and nothing else */
1952 if (branch[1].procname || branch[1].ctl_name)
1953 return NULL;
1954
1955 /* table should contain subdirectory with the same name */
1956 for (p = table; p->procname || p->ctl_name; p++) {
1957 if (!p->child)
1958 continue;
1959 if (p->procname && strcmp(p->procname, s) == 0)
1960 return p;
1961 }
1962 return NULL;
1963 }
1964
1965 /* see if attaching q to p would be an improvement */
1966 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1967 {
1968 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1969 struct ctl_table *next;
1970 int is_better = 0;
1971 int not_in_parent = !p->attached_by;
1972
1973 while ((next = is_branch_in(by, to)) != NULL) {
1974 if (by == q->attached_by)
1975 is_better = 1;
1976 if (to == p->attached_by)
1977 not_in_parent = 1;
1978 by = by->child;
1979 to = next->child;
1980 }
1981
1982 if (is_better && not_in_parent) {
1983 q->attached_by = by;
1984 q->attached_to = to;
1985 q->parent = p;
1986 }
1987 }
1988
1989 /**
1990 * __register_sysctl_paths - register a sysctl hierarchy
1991 * @root: List of sysctl headers to register on
1992 * @namespaces: Data to compute which lists of sysctl entries are visible
1993 * @path: The path to the directory the sysctl table is in.
1994 * @table: the top-level table structure
1995 *
1996 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1997 * array. A completely 0 filled entry terminates the table.
1998 *
1999 * The members of the &struct ctl_table structure are used as follows:
2000 *
2001 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
2002 * must be unique within that level of sysctl
2003 *
2004 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
2005 * enter a sysctl file
2006 *
2007 * data - a pointer to data for use by proc_handler
2008 *
2009 * maxlen - the maximum size in bytes of the data
2010 *
2011 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
2012 *
2013 * child - a pointer to the child sysctl table if this entry is a directory, or
2014 * %NULL.
2015 *
2016 * proc_handler - the text handler routine (described below)
2017 *
2018 * strategy - the strategy routine (described below)
2019 *
2020 * de - for internal use by the sysctl routines
2021 *
2022 * extra1, extra2 - extra pointers usable by the proc handler routines
2023 *
2024 * Leaf nodes in the sysctl tree will be represented by a single file
2025 * under /proc; non-leaf nodes will be represented by directories.
2026 *
2027 * sysctl(2) can automatically manage read and write requests through
2028 * the sysctl table. The data and maxlen fields of the ctl_table
2029 * struct enable minimal validation of the values being written to be
2030 * performed, and the mode field allows minimal authentication.
2031 *
2032 * More sophisticated management can be enabled by the provision of a
2033 * strategy routine with the table entry. This will be called before
2034 * any automatic read or write of the data is performed.
2035 *
2036 * The strategy routine may return
2037 *
2038 * < 0 - Error occurred (error is passed to user process)
2039 *
2040 * 0 - OK - proceed with automatic read or write.
2041 *
2042 * > 0 - OK - read or write has been done by the strategy routine, so
2043 * return immediately.
2044 *
2045 * There must be a proc_handler routine for any terminal nodes
2046 * mirrored under /proc/sys (non-terminals are handled by a built-in
2047 * directory handler). Several default handlers are available to
2048 * cover common cases -
2049 *
2050 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
2051 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
2052 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
2053 *
2054 * It is the handler's job to read the input buffer from user memory
2055 * and process it. The handler should return 0 on success.
2056 *
2057 * This routine returns %NULL on a failure to register, and a pointer
2058 * to the table header on success.
2059 */
2060 struct ctl_table_header *__register_sysctl_paths(
2061 struct ctl_table_root *root,
2062 struct nsproxy *namespaces,
2063 const struct ctl_path *path, struct ctl_table *table)
2064 {
2065 struct ctl_table_header *header;
2066 struct ctl_table *new, **prevp;
2067 unsigned int n, npath;
2068 struct ctl_table_set *set;
2069
2070 /* Count the path components */
2071 for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
2072 ;
2073
2074 /*
2075 * For each path component, allocate a 2-element ctl_table array.
2076 * The first array element will be filled with the sysctl entry
2077 * for this, the second will be the sentinel (ctl_name == 0).
2078 *
2079 * We allocate everything in one go so that we don't have to
2080 * worry about freeing additional memory in unregister_sysctl_table.
2081 */
2082 header = kzalloc(sizeof(struct ctl_table_header) +
2083 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
2084 if (!header)
2085 return NULL;
2086
2087 new = (struct ctl_table *) (header + 1);
2088
2089 /* Now connect the dots */
2090 prevp = &header->ctl_table;
2091 for (n = 0; n < npath; ++n, ++path) {
2092 /* Copy the procname */
2093 new->procname = path->procname;
2094 new->ctl_name = path->ctl_name;
2095 new->mode = 0555;
2096
2097 *prevp = new;
2098 prevp = &new->child;
2099
2100 new += 2;
2101 }
2102 *prevp = table;
2103 header->ctl_table_arg = table;
2104
2105 INIT_LIST_HEAD(&header->ctl_entry);
2106 header->used = 0;
2107 header->unregistering = NULL;
2108 header->root = root;
2109 sysctl_set_parent(NULL, header->ctl_table);
2110 header->count = 1;
2111 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
2112 if (sysctl_check_table(namespaces, header->ctl_table)) {
2113 kfree(header);
2114 return NULL;
2115 }
2116 #endif
2117 spin_lock(&sysctl_lock);
2118 header->set = lookup_header_set(root, namespaces);
2119 header->attached_by = header->ctl_table;
2120 header->attached_to = root_table;
2121 header->parent = &root_table_header;
2122 for (set = header->set; set; set = set->parent) {
2123 struct ctl_table_header *p;
2124 list_for_each_entry(p, &set->list, ctl_entry) {
2125 if (p->unregistering)
2126 continue;
2127 try_attach(p, header);
2128 }
2129 }
2130 header->parent->count++;
2131 list_add_tail(&header->ctl_entry, &header->set->list);
2132 spin_unlock(&sysctl_lock);
2133
2134 return header;
2135 }
2136
2137 /**
2138 * register_sysctl_table_path - register a sysctl table hierarchy
2139 * @path: The path to the directory the sysctl table is in.
2140 * @table: the top-level table structure
2141 *
2142 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2143 * array. A completely 0 filled entry terminates the table.
2144 *
2145 * See __register_sysctl_paths for more details.
2146 */
2147 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2148 struct ctl_table *table)
2149 {
2150 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
2151 path, table);
2152 }
2153
2154 /**
2155 * register_sysctl_table - register a sysctl table hierarchy
2156 * @table: the top-level table structure
2157 *
2158 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2159 * array. A completely 0 filled entry terminates the table.
2160 *
2161 * See register_sysctl_paths for more details.
2162 */
2163 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
2164 {
2165 static const struct ctl_path null_path[] = { {} };
2166
2167 return register_sysctl_paths(null_path, table);
2168 }
2169
2170 /**
2171 * unregister_sysctl_table - unregister a sysctl table hierarchy
2172 * @header: the header returned from register_sysctl_table
2173 *
2174 * Unregisters the sysctl table and all children. proc entries may not
2175 * actually be removed until they are no longer used by anyone.
2176 */
2177 void unregister_sysctl_table(struct ctl_table_header * header)
2178 {
2179 might_sleep();
2180
2181 if (header == NULL)
2182 return;
2183
2184 spin_lock(&sysctl_lock);
2185 start_unregistering(header);
2186 if (!--header->parent->count) {
2187 WARN_ON(1);
2188 kfree(header->parent);
2189 }
2190 if (!--header->count)
2191 kfree(header);
2192 spin_unlock(&sysctl_lock);
2193 }
2194
2195 int sysctl_is_seen(struct ctl_table_header *p)
2196 {
2197 struct ctl_table_set *set = p->set;
2198 int res;
2199 spin_lock(&sysctl_lock);
2200 if (p->unregistering)
2201 res = 0;
2202 else if (!set->is_seen)
2203 res = 1;
2204 else
2205 res = set->is_seen(set);
2206 spin_unlock(&sysctl_lock);
2207 return res;
2208 }
2209
2210 void setup_sysctl_set(struct ctl_table_set *p,
2211 struct ctl_table_set *parent,
2212 int (*is_seen)(struct ctl_table_set *))
2213 {
2214 INIT_LIST_HEAD(&p->list);
2215 p->parent = parent ? parent : &sysctl_table_root.default_set;
2216 p->is_seen = is_seen;
2217 }
2218
2219 #else /* !CONFIG_SYSCTL */
2220 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2221 {
2222 return NULL;
2223 }
2224
2225 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2226 struct ctl_table *table)
2227 {
2228 return NULL;
2229 }
2230
2231 void unregister_sysctl_table(struct ctl_table_header * table)
2232 {
2233 }
2234
2235 void setup_sysctl_set(struct ctl_table_set *p,
2236 struct ctl_table_set *parent,
2237 int (*is_seen)(struct ctl_table_set *))
2238 {
2239 }
2240
2241 void sysctl_head_put(struct ctl_table_header *head)
2242 {
2243 }
2244
2245 #endif /* CONFIG_SYSCTL */
2246
2247 /*
2248 * /proc/sys support
2249 */
2250
2251 #ifdef CONFIG_PROC_SYSCTL
2252
2253 static int _proc_do_string(void* data, int maxlen, int write,
2254 void __user *buffer,
2255 size_t *lenp, loff_t *ppos)
2256 {
2257 size_t len;
2258 char __user *p;
2259 char c;
2260
2261 if (!data || !maxlen || !*lenp) {
2262 *lenp = 0;
2263 return 0;
2264 }
2265
2266 if (write) {
2267 len = 0;
2268 p = buffer;
2269 while (len < *lenp) {
2270 if (get_user(c, p++))
2271 return -EFAULT;
2272 if (c == 0 || c == '\n')
2273 break;
2274 len++;
2275 }
2276 if (len >= maxlen)
2277 len = maxlen-1;
2278 if(copy_from_user(data, buffer, len))
2279 return -EFAULT;
2280 ((char *) data)[len] = 0;
2281 *ppos += *lenp;
2282 } else {
2283 len = strlen(data);
2284 if (len > maxlen)
2285 len = maxlen;
2286
2287 if (*ppos > len) {
2288 *lenp = 0;
2289 return 0;
2290 }
2291
2292 data += *ppos;
2293 len -= *ppos;
2294
2295 if (len > *lenp)
2296 len = *lenp;
2297 if (len)
2298 if(copy_to_user(buffer, data, len))
2299 return -EFAULT;
2300 if (len < *lenp) {
2301 if(put_user('\n', ((char __user *) buffer) + len))
2302 return -EFAULT;
2303 len++;
2304 }
2305 *lenp = len;
2306 *ppos += len;
2307 }
2308 return 0;
2309 }
2310
2311 /**
2312 * proc_dostring - read a string sysctl
2313 * @table: the sysctl table
2314 * @write: %TRUE if this is a write to the sysctl file
2315 * @buffer: the user buffer
2316 * @lenp: the size of the user buffer
2317 * @ppos: file position
2318 *
2319 * Reads/writes a string from/to the user buffer. If the kernel
2320 * buffer provided is not large enough to hold the string, the
2321 * string is truncated. The copied string is %NULL-terminated.
2322 * If the string is being read by the user process, it is copied
2323 * and a newline '\n' is added. It is truncated if the buffer is
2324 * not large enough.
2325 *
2326 * Returns 0 on success.
2327 */
2328 int proc_dostring(struct ctl_table *table, int write,
2329 void __user *buffer, size_t *lenp, loff_t *ppos)
2330 {
2331 return _proc_do_string(table->data, table->maxlen, write,
2332 buffer, lenp, ppos);
2333 }
2334
2335
2336 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2337 int *valp,
2338 int write, void *data)
2339 {
2340 if (write) {
2341 *valp = *negp ? -*lvalp : *lvalp;
2342 } else {
2343 int val = *valp;
2344 if (val < 0) {
2345 *negp = -1;
2346 *lvalp = (unsigned long)-val;
2347 } else {
2348 *negp = 0;
2349 *lvalp = (unsigned long)val;
2350 }
2351 }
2352 return 0;
2353 }
2354
2355 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2356 int write, void __user *buffer,
2357 size_t *lenp, loff_t *ppos,
2358 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2359 int write, void *data),
2360 void *data)
2361 {
2362 #define TMPBUFLEN 21
2363 int *i, vleft, first = 1, neg;
2364 unsigned long lval;
2365 size_t left, len;
2366
2367 char buf[TMPBUFLEN], *p;
2368 char __user *s = buffer;
2369
2370 if (!tbl_data || !table->maxlen || !*lenp ||
2371 (*ppos && !write)) {
2372 *lenp = 0;
2373 return 0;
2374 }
2375
2376 i = (int *) tbl_data;
2377 vleft = table->maxlen / sizeof(*i);
2378 left = *lenp;
2379
2380 if (!conv)
2381 conv = do_proc_dointvec_conv;
2382
2383 for (; left && vleft--; i++, first=0) {
2384 if (write) {
2385 while (left) {
2386 char c;
2387 if (get_user(c, s))
2388 return -EFAULT;
2389 if (!isspace(c))
2390 break;
2391 left--;
2392 s++;
2393 }
2394 if (!left)
2395 break;
2396 neg = 0;
2397 len = left;
2398 if (len > sizeof(buf) - 1)
2399 len = sizeof(buf) - 1;
2400 if (copy_from_user(buf, s, len))
2401 return -EFAULT;
2402 buf[len] = 0;
2403 p = buf;
2404 if (*p == '-' && left > 1) {
2405 neg = 1;
2406 p++;
2407 }
2408 if (*p < '0' || *p > '9')
2409 break;
2410
2411 lval = simple_strtoul(p, &p, 0);
2412
2413 len = p-buf;
2414 if ((len < left) && *p && !isspace(*p))
2415 break;
2416 s += len;
2417 left -= len;
2418
2419 if (conv(&neg, &lval, i, 1, data))
2420 break;
2421 } else {
2422 p = buf;
2423 if (!first)
2424 *p++ = '\t';
2425
2426 if (conv(&neg, &lval, i, 0, data))
2427 break;
2428
2429 sprintf(p, "%s%lu", neg ? "-" : "", lval);
2430 len = strlen(buf);
2431 if (len > left)
2432 len = left;
2433 if(copy_to_user(s, buf, len))
2434 return -EFAULT;
2435 left -= len;
2436 s += len;
2437 }
2438 }
2439
2440 if (!write && !first && left) {
2441 if(put_user('\n', s))
2442 return -EFAULT;
2443 left--, s++;
2444 }
2445 if (write) {
2446 while (left) {
2447 char c;
2448 if (get_user(c, s++))
2449 return -EFAULT;
2450 if (!isspace(c))
2451 break;
2452 left--;
2453 }
2454 }
2455 if (write && first)
2456 return -EINVAL;
2457 *lenp -= left;
2458 *ppos += *lenp;
2459 return 0;
2460 #undef TMPBUFLEN
2461 }
2462
2463 static int do_proc_dointvec(struct ctl_table *table, int write,
2464 void __user *buffer, size_t *lenp, loff_t *ppos,
2465 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2466 int write, void *data),
2467 void *data)
2468 {
2469 return __do_proc_dointvec(table->data, table, write,
2470 buffer, lenp, ppos, conv, data);
2471 }
2472
2473 /**
2474 * proc_dointvec - read a vector of integers
2475 * @table: the sysctl table
2476 * @write: %TRUE if this is a write to the sysctl file
2477 * @buffer: the user buffer
2478 * @lenp: the size of the user buffer
2479 * @ppos: file position
2480 *
2481 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2482 * values from/to the user buffer, treated as an ASCII string.
2483 *
2484 * Returns 0 on success.
2485 */
2486 int proc_dointvec(struct ctl_table *table, int write,
2487 void __user *buffer, size_t *lenp, loff_t *ppos)
2488 {
2489 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2490 NULL,NULL);
2491 }
2492
2493 /*
2494 * Taint values can only be increased
2495 * This means we can safely use a temporary.
2496 */
2497 static int proc_taint(struct ctl_table *table, int write,
2498 void __user *buffer, size_t *lenp, loff_t *ppos)
2499 {
2500 struct ctl_table t;
2501 unsigned long tmptaint = get_taint();
2502 int err;
2503
2504 if (write && !capable(CAP_SYS_ADMIN))
2505 return -EPERM;
2506
2507 t = *table;
2508 t.data = &tmptaint;
2509 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2510 if (err < 0)
2511 return err;
2512
2513 if (write) {
2514 /*
2515 * Poor man's atomic or. Not worth adding a primitive
2516 * to everyone's atomic.h for this
2517 */
2518 int i;
2519 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2520 if ((tmptaint >> i) & 1)
2521 add_taint(i);
2522 }
2523 }
2524
2525 return err;
2526 }
2527
2528 struct do_proc_dointvec_minmax_conv_param {
2529 int *min;
2530 int *max;
2531 };
2532
2533 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2534 int *valp,
2535 int write, void *data)
2536 {
2537 struct do_proc_dointvec_minmax_conv_param *param = data;
2538 if (write) {
2539 int val = *negp ? -*lvalp : *lvalp;
2540 if ((param->min && *param->min > val) ||
2541 (param->max && *param->max < val))
2542 return -EINVAL;
2543 *valp = val;
2544 } else {
2545 int val = *valp;
2546 if (val < 0) {
2547 *negp = -1;
2548 *lvalp = (unsigned long)-val;
2549 } else {
2550 *negp = 0;
2551 *lvalp = (unsigned long)val;
2552 }
2553 }
2554 return 0;
2555 }
2556
2557 /**
2558 * proc_dointvec_minmax - read a vector of integers with min/max values
2559 * @table: the sysctl table
2560 * @write: %TRUE if this is a write to the sysctl file
2561 * @buffer: the user buffer
2562 * @lenp: the size of the user buffer
2563 * @ppos: file position
2564 *
2565 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2566 * values from/to the user buffer, treated as an ASCII string.
2567 *
2568 * This routine will ensure the values are within the range specified by
2569 * table->extra1 (min) and table->extra2 (max).
2570 *
2571 * Returns 0 on success.
2572 */
2573 int proc_dointvec_minmax(struct ctl_table *table, int write,
2574 void __user *buffer, size_t *lenp, loff_t *ppos)
2575 {
2576 struct do_proc_dointvec_minmax_conv_param param = {
2577 .min = (int *) table->extra1,
2578 .max = (int *) table->extra2,
2579 };
2580 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2581 do_proc_dointvec_minmax_conv, &param);
2582 }
2583
2584 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2585 void __user *buffer,
2586 size_t *lenp, loff_t *ppos,
2587 unsigned long convmul,
2588 unsigned long convdiv)
2589 {
2590 #define TMPBUFLEN 21
2591 unsigned long *i, *min, *max, val;
2592 int vleft, first=1, neg;
2593 size_t len, left;
2594 char buf[TMPBUFLEN], *p;
2595 char __user *s = buffer;
2596
2597 if (!data || !table->maxlen || !*lenp ||
2598 (*ppos && !write)) {
2599 *lenp = 0;
2600 return 0;
2601 }
2602
2603 i = (unsigned long *) data;
2604 min = (unsigned long *) table->extra1;
2605 max = (unsigned long *) table->extra2;
2606 vleft = table->maxlen / sizeof(unsigned long);
2607 left = *lenp;
2608
2609 for (; left && vleft--; i++, min++, max++, first=0) {
2610 if (write) {
2611 while (left) {
2612 char c;
2613 if (get_user(c, s))
2614 return -EFAULT;
2615 if (!isspace(c))
2616 break;
2617 left--;
2618 s++;
2619 }
2620 if (!left)
2621 break;
2622 neg = 0;
2623 len = left;
2624 if (len > TMPBUFLEN-1)
2625 len = TMPBUFLEN-1;
2626 if (copy_from_user(buf, s, len))
2627 return -EFAULT;
2628 buf[len] = 0;
2629 p = buf;
2630 if (*p == '-' && left > 1) {
2631 neg = 1;
2632 p++;
2633 }
2634 if (*p < '0' || *p > '9')
2635 break;
2636 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2637 len = p-buf;
2638 if ((len < left) && *p && !isspace(*p))
2639 break;
2640 if (neg)
2641 val = -val;
2642 s += len;
2643 left -= len;
2644
2645 if(neg)
2646 continue;
2647 if ((min && val < *min) || (max && val > *max))
2648 continue;
2649 *i = val;
2650 } else {
2651 p = buf;
2652 if (!first)
2653 *p++ = '\t';
2654 sprintf(p, "%lu", convdiv * (*i) / convmul);
2655 len = strlen(buf);
2656 if (len > left)
2657 len = left;
2658 if(copy_to_user(s, buf, len))
2659 return -EFAULT;
2660 left -= len;
2661 s += len;
2662 }
2663 }
2664
2665 if (!write && !first && left) {
2666 if(put_user('\n', s))
2667 return -EFAULT;
2668 left--, s++;
2669 }
2670 if (write) {
2671 while (left) {
2672 char c;
2673 if (get_user(c, s++))
2674 return -EFAULT;
2675 if (!isspace(c))
2676 break;
2677 left--;
2678 }
2679 }
2680 if (write && first)
2681 return -EINVAL;
2682 *lenp -= left;
2683 *ppos += *lenp;
2684 return 0;
2685 #undef TMPBUFLEN
2686 }
2687
2688 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2689 void __user *buffer,
2690 size_t *lenp, loff_t *ppos,
2691 unsigned long convmul,
2692 unsigned long convdiv)
2693 {
2694 return __do_proc_doulongvec_minmax(table->data, table, write,
2695 buffer, lenp, ppos, convmul, convdiv);
2696 }
2697
2698 /**
2699 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2700 * @table: the sysctl table
2701 * @write: %TRUE if this is a write to the sysctl file
2702 * @buffer: the user buffer
2703 * @lenp: the size of the user buffer
2704 * @ppos: file position
2705 *
2706 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2707 * values from/to the user buffer, treated as an ASCII string.
2708 *
2709 * This routine will ensure the values are within the range specified by
2710 * table->extra1 (min) and table->extra2 (max).
2711 *
2712 * Returns 0 on success.
2713 */
2714 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2715 void __user *buffer, size_t *lenp, loff_t *ppos)
2716 {
2717 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2718 }
2719
2720 /**
2721 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2722 * @table: the sysctl table
2723 * @write: %TRUE if this is a write to the sysctl file
2724 * @buffer: the user buffer
2725 * @lenp: the size of the user buffer
2726 * @ppos: file position
2727 *
2728 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2729 * values from/to the user buffer, treated as an ASCII string. The values
2730 * are treated as milliseconds, and converted to jiffies when they are stored.
2731 *
2732 * This routine will ensure the values are within the range specified by
2733 * table->extra1 (min) and table->extra2 (max).
2734 *
2735 * Returns 0 on success.
2736 */
2737 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2738 void __user *buffer,
2739 size_t *lenp, loff_t *ppos)
2740 {
2741 return do_proc_doulongvec_minmax(table, write, buffer,
2742 lenp, ppos, HZ, 1000l);
2743 }
2744
2745
2746 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2747 int *valp,
2748 int write, void *data)
2749 {
2750 if (write) {
2751 if (*lvalp > LONG_MAX / HZ)
2752 return 1;
2753 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2754 } else {
2755 int val = *valp;
2756 unsigned long lval;
2757 if (val < 0) {
2758 *negp = -1;
2759 lval = (unsigned long)-val;
2760 } else {
2761 *negp = 0;
2762 lval = (unsigned long)val;
2763 }
2764 *lvalp = lval / HZ;
2765 }
2766 return 0;
2767 }
2768
2769 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2770 int *valp,
2771 int write, void *data)
2772 {
2773 if (write) {
2774 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2775 return 1;
2776 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2777 } else {
2778 int val = *valp;
2779 unsigned long lval;
2780 if (val < 0) {
2781 *negp = -1;
2782 lval = (unsigned long)-val;
2783 } else {
2784 *negp = 0;
2785 lval = (unsigned long)val;
2786 }
2787 *lvalp = jiffies_to_clock_t(lval);
2788 }
2789 return 0;
2790 }
2791
2792 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2793 int *valp,
2794 int write, void *data)
2795 {
2796 if (write) {
2797 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2798 } else {
2799 int val = *valp;
2800 unsigned long lval;
2801 if (val < 0) {
2802 *negp = -1;
2803 lval = (unsigned long)-val;
2804 } else {
2805 *negp = 0;
2806 lval = (unsigned long)val;
2807 }
2808 *lvalp = jiffies_to_msecs(lval);
2809 }
2810 return 0;
2811 }
2812
2813 /**
2814 * proc_dointvec_jiffies - read a vector of integers as seconds
2815 * @table: the sysctl table
2816 * @write: %TRUE if this is a write to the sysctl file
2817 * @buffer: the user buffer
2818 * @lenp: the size of the user buffer
2819 * @ppos: file position
2820 *
2821 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2822 * values from/to the user buffer, treated as an ASCII string.
2823 * The values read are assumed to be in seconds, and are converted into
2824 * jiffies.
2825 *
2826 * Returns 0 on success.
2827 */
2828 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2829 void __user *buffer, size_t *lenp, loff_t *ppos)
2830 {
2831 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2832 do_proc_dointvec_jiffies_conv,NULL);
2833 }
2834
2835 /**
2836 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2837 * @table: the sysctl table
2838 * @write: %TRUE if this is a write to the sysctl file
2839 * @buffer: the user buffer
2840 * @lenp: the size of the user buffer
2841 * @ppos: pointer to the file position
2842 *
2843 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2844 * values from/to the user buffer, treated as an ASCII string.
2845 * The values read are assumed to be in 1/USER_HZ seconds, and
2846 * are converted into jiffies.
2847 *
2848 * Returns 0 on success.
2849 */
2850 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2851 void __user *buffer, size_t *lenp, loff_t *ppos)
2852 {
2853 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2854 do_proc_dointvec_userhz_jiffies_conv,NULL);
2855 }
2856
2857 /**
2858 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2859 * @table: the sysctl table
2860 * @write: %TRUE if this is a write to the sysctl file
2861 * @buffer: the user buffer
2862 * @lenp: the size of the user buffer
2863 * @ppos: file position
2864 * @ppos: the current position in the file
2865 *
2866 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2867 * values from/to the user buffer, treated as an ASCII string.
2868 * The values read are assumed to be in 1/1000 seconds, and
2869 * are converted into jiffies.
2870 *
2871 * Returns 0 on success.
2872 */
2873 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2874 void __user *buffer, size_t *lenp, loff_t *ppos)
2875 {
2876 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2877 do_proc_dointvec_ms_jiffies_conv, NULL);
2878 }
2879
2880 static int proc_do_cad_pid(struct ctl_table *table, int write,
2881 void __user *buffer, size_t *lenp, loff_t *ppos)
2882 {
2883 struct pid *new_pid;
2884 pid_t tmp;
2885 int r;
2886
2887 tmp = pid_vnr(cad_pid);
2888
2889 r = __do_proc_dointvec(&tmp, table, write, buffer,
2890 lenp, ppos, NULL, NULL);
2891 if (r || !write)
2892 return r;
2893
2894 new_pid = find_get_pid(tmp);
2895 if (!new_pid)
2896 return -ESRCH;
2897
2898 put_pid(xchg(&cad_pid, new_pid));
2899 return 0;
2900 }
2901
2902 #else /* CONFIG_PROC_FS */
2903
2904 int proc_dostring(struct ctl_table *table, int write,
2905 void __user *buffer, size_t *lenp, loff_t *ppos)
2906 {
2907 return -ENOSYS;
2908 }
2909
2910 int proc_dointvec(struct ctl_table *table, int write,
2911 void __user *buffer, size_t *lenp, loff_t *ppos)
2912 {
2913 return -ENOSYS;
2914 }
2915
2916 int proc_dointvec_minmax(struct ctl_table *table, int write,
2917 void __user *buffer, size_t *lenp, loff_t *ppos)
2918 {
2919 return -ENOSYS;
2920 }
2921
2922 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2923 void __user *buffer, size_t *lenp, loff_t *ppos)
2924 {
2925 return -ENOSYS;
2926 }
2927
2928 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2929 void __user *buffer, size_t *lenp, loff_t *ppos)
2930 {
2931 return -ENOSYS;
2932 }
2933
2934 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2935 void __user *buffer, size_t *lenp, loff_t *ppos)
2936 {
2937 return -ENOSYS;
2938 }
2939
2940 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2941 void __user *buffer, size_t *lenp, loff_t *ppos)
2942 {
2943 return -ENOSYS;
2944 }
2945
2946 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2947 void __user *buffer,
2948 size_t *lenp, loff_t *ppos)
2949 {
2950 return -ENOSYS;
2951 }
2952
2953
2954 #endif /* CONFIG_PROC_FS */
2955
2956
2957 #ifdef CONFIG_SYSCTL_SYSCALL
2958 /*
2959 * General sysctl support routines
2960 */
2961
2962 /* The generic sysctl data routine (used if no strategy routine supplied) */
2963 int sysctl_data(struct ctl_table *table,
2964 void __user *oldval, size_t __user *oldlenp,
2965 void __user *newval, size_t newlen)
2966 {
2967 size_t len;
2968
2969 /* Get out of I don't have a variable */
2970 if (!table->data || !table->maxlen)
2971 return -ENOTDIR;
2972
2973 if (oldval && oldlenp) {
2974 if (get_user(len, oldlenp))
2975 return -EFAULT;
2976 if (len) {
2977 if (len > table->maxlen)
2978 len = table->maxlen;
2979 if (copy_to_user(oldval, table->data, len))
2980 return -EFAULT;
2981 if (put_user(len, oldlenp))
2982 return -EFAULT;
2983 }
2984 }
2985
2986 if (newval && newlen) {
2987 if (newlen > table->maxlen)
2988 newlen = table->maxlen;
2989
2990 if (copy_from_user(table->data, newval, newlen))
2991 return -EFAULT;
2992 }
2993 return 1;
2994 }
2995
2996 /* The generic string strategy routine: */
2997 int sysctl_string(struct ctl_table *table,
2998 void __user *oldval, size_t __user *oldlenp,
2999 void __user *newval, size_t newlen)
3000 {
3001 if (!table->data || !table->maxlen)
3002 return -ENOTDIR;
3003
3004 if (oldval && oldlenp) {
3005 size_t bufsize;
3006 if (get_user(bufsize, oldlenp))
3007 return -EFAULT;
3008 if (bufsize) {
3009 size_t len = strlen(table->data), copied;
3010
3011 /* This shouldn't trigger for a well-formed sysctl */
3012 if (len > table->maxlen)
3013 len = table->maxlen;
3014
3015 /* Copy up to a max of bufsize-1 bytes of the string */
3016 copied = (len >= bufsize) ? bufsize - 1 : len;
3017
3018 if (copy_to_user(oldval, table->data, copied) ||
3019 put_user(0, (char __user *)(oldval + copied)))
3020 return -EFAULT;
3021 if (put_user(len, oldlenp))
3022 return -EFAULT;
3023 }
3024 }
3025 if (newval && newlen) {
3026 size_t len = newlen;
3027 if (len > table->maxlen)
3028 len = table->maxlen;
3029 if(copy_from_user(table->data, newval, len))
3030 return -EFAULT;
3031 if (len == table->maxlen)
3032 len--;
3033 ((char *) table->data)[len] = 0;
3034 }
3035 return 1;
3036 }
3037
3038 /*
3039 * This function makes sure that all of the integers in the vector
3040 * are between the minimum and maximum values given in the arrays
3041 * table->extra1 and table->extra2, respectively.
3042 */
3043 int sysctl_intvec(struct ctl_table *table,
3044 void __user *oldval, size_t __user *oldlenp,
3045 void __user *newval, size_t newlen)
3046 {
3047
3048 if (newval && newlen) {
3049 int __user *vec = (int __user *) newval;
3050 int *min = (int *) table->extra1;
3051 int *max = (int *) table->extra2;
3052 size_t length;
3053 int i;
3054
3055 if (newlen % sizeof(int) != 0)
3056 return -EINVAL;
3057
3058 if (!table->extra1 && !table->extra2)
3059 return 0;
3060
3061 if (newlen > table->maxlen)
3062 newlen = table->maxlen;
3063 length = newlen / sizeof(int);
3064
3065 for (i = 0; i < length; i++) {
3066 int value;
3067 if (get_user(value, vec + i))
3068 return -EFAULT;
3069 if (min && value < min[i])
3070 return -EINVAL;
3071 if (max && value > max[i])
3072 return -EINVAL;
3073 }
3074 }
3075 return 0;
3076 }
3077
3078 /* Strategy function to convert jiffies to seconds */
3079 int sysctl_jiffies(struct ctl_table *table,
3080 void __user *oldval, size_t __user *oldlenp,
3081 void __user *newval, size_t newlen)
3082 {
3083 if (oldval && oldlenp) {
3084 size_t olen;
3085
3086 if (get_user(olen, oldlenp))
3087 return -EFAULT;
3088 if (olen) {
3089 int val;
3090
3091 if (olen < sizeof(int))
3092 return -EINVAL;
3093
3094 val = *(int *)(table->data) / HZ;
3095 if (put_user(val, (int __user *)oldval))
3096 return -EFAULT;
3097 if (put_user(sizeof(int), oldlenp))
3098 return -EFAULT;
3099 }
3100 }
3101 if (newval && newlen) {
3102 int new;
3103 if (newlen != sizeof(int))
3104 return -EINVAL;
3105 if (get_user(new, (int __user *)newval))
3106 return -EFAULT;
3107 *(int *)(table->data) = new*HZ;
3108 }
3109 return 1;
3110 }
3111
3112 /* Strategy function to convert jiffies to seconds */
3113 int sysctl_ms_jiffies(struct ctl_table *table,
3114 void __user *oldval, size_t __user *oldlenp,
3115 void __user *newval, size_t newlen)
3116 {
3117 if (oldval && oldlenp) {
3118 size_t olen;
3119
3120 if (get_user(olen, oldlenp))
3121 return -EFAULT;
3122 if (olen) {
3123 int val;
3124
3125 if (olen < sizeof(int))
3126 return -EINVAL;
3127
3128 val = jiffies_to_msecs(*(int *)(table->data));
3129 if (put_user(val, (int __user *)oldval))
3130 return -EFAULT;
3131 if (put_user(sizeof(int), oldlenp))
3132 return -EFAULT;
3133 }
3134 }
3135 if (newval && newlen) {
3136 int new;
3137 if (newlen != sizeof(int))
3138 return -EINVAL;
3139 if (get_user(new, (int __user *)newval))
3140 return -EFAULT;
3141 *(int *)(table->data) = msecs_to_jiffies(new);
3142 }
3143 return 1;
3144 }
3145
3146
3147
3148 #else /* CONFIG_SYSCTL_SYSCALL */
3149
3150
3151 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
3152 {
3153 struct __sysctl_args tmp;
3154 int error;
3155
3156 if (copy_from_user(&tmp, args, sizeof(tmp)))
3157 return -EFAULT;
3158
3159 error = deprecated_sysctl_warning(&tmp);
3160
3161 /* If no error reading the parameters then just -ENOSYS ... */
3162 if (!error)
3163 error = -ENOSYS;
3164
3165 return error;
3166 }
3167
3168 int sysctl_data(struct ctl_table *table,
3169 void __user *oldval, size_t __user *oldlenp,
3170 void __user *newval, size_t newlen)
3171 {
3172 return -ENOSYS;
3173 }
3174
3175 int sysctl_string(struct ctl_table *table,
3176 void __user *oldval, size_t __user *oldlenp,
3177 void __user *newval, size_t newlen)
3178 {
3179 return -ENOSYS;
3180 }
3181
3182 int sysctl_intvec(struct ctl_table *table,
3183 void __user *oldval, size_t __user *oldlenp,
3184 void __user *newval, size_t newlen)
3185 {
3186 return -ENOSYS;
3187 }
3188
3189 int sysctl_jiffies(struct ctl_table *table,
3190 void __user *oldval, size_t __user *oldlenp,
3191 void __user *newval, size_t newlen)
3192 {
3193 return -ENOSYS;
3194 }
3195
3196 int sysctl_ms_jiffies(struct ctl_table *table,
3197 void __user *oldval, size_t __user *oldlenp,
3198 void __user *newval, size_t newlen)
3199 {
3200 return -ENOSYS;
3201 }
3202
3203 #endif /* CONFIG_SYSCTL_SYSCALL */
3204
3205 static int deprecated_sysctl_warning(struct __sysctl_args *args)
3206 {
3207 static int msg_count;
3208 int name[CTL_MAXNAME];
3209 int i;
3210
3211 /* Check args->nlen. */
3212 if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
3213 return -ENOTDIR;
3214
3215 /* Read in the sysctl name for better debug message logging */
3216 for (i = 0; i < args->nlen; i++)
3217 if (get_user(name[i], args->name + i))
3218 return -EFAULT;
3219
3220 /* Ignore accesses to kernel.version */
3221 if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
3222 return 0;
3223
3224 if (msg_count < 5) {
3225 msg_count++;
3226 printk(KERN_INFO
3227 "warning: process `%s' used the deprecated sysctl "
3228 "system call with ", current->comm);
3229 for (i = 0; i < args->nlen; i++)
3230 printk("%d.", name[i]);
3231 printk("\n");
3232 }
3233 return 0;
3234 }
3235
3236 /*
3237 * No sense putting this after each symbol definition, twice,
3238 * exception granted :-)
3239 */
3240 EXPORT_SYMBOL(proc_dointvec);
3241 EXPORT_SYMBOL(proc_dointvec_jiffies);
3242 EXPORT_SYMBOL(proc_dointvec_minmax);
3243 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3244 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3245 EXPORT_SYMBOL(proc_dostring);
3246 EXPORT_SYMBOL(proc_doulongvec_minmax);
3247 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3248 EXPORT_SYMBOL(register_sysctl_table);
3249 EXPORT_SYMBOL(register_sysctl_paths);
3250 EXPORT_SYMBOL(sysctl_intvec);
3251 EXPORT_SYMBOL(sysctl_jiffies);
3252 EXPORT_SYMBOL(sysctl_ms_jiffies);
3253 EXPORT_SYMBOL(sysctl_string);
3254 EXPORT_SYMBOL(sysctl_data);
3255 EXPORT_SYMBOL(unregister_sysctl_table);
Linux 2.6 ibm-acpi/thinkpad-acpi driver git tree