?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
HWPOISON: The high level memory error handler in the VM v7
[linux-2.6/linux-2.6-openrd.git] / kernel / sysctl.c
blobeacae77ac9fc0a3c8d0660cbdfb0e5337e61fe82
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/utsname.h>
30 #include <linux/kmemcheck.h>
31 #include <linux/smp_lock.h>
32 #include <linux/fs.h>
33 #include <linux/init.h>
34 #include <linux/kernel.h>
35 #include <linux/kobject.h>
36 #include <linux/net.h>
37 #include <linux/sysrq.h>
38 #include <linux/highuid.h>
39 #include <linux/writeback.h>
40 #include <linux/hugetlb.h>
41 #include <linux/initrd.h>
42 #include <linux/key.h>
43 #include <linux/times.h>
44 #include <linux/limits.h>
45 #include <linux/dcache.h>
46 #include <linux/syscalls.h>
47 #include <linux/vmstat.h>
48 #include <linux/nfs_fs.h>
49 #include <linux/acpi.h>
50 #include <linux/reboot.h>
51 #include <linux/ftrace.h>
52 #include <linux/slow-work.h>
53 #include <linux/perf_counter.h>
54
55 #include <asm/uaccess.h>
56 #include <asm/processor.h>
57
58 #ifdef CONFIG_X86
59 #include <asm/nmi.h>
60 #include <asm/stacktrace.h>
61 #include <asm/io.h>
62 #endif
63
64 static int deprecated_sysctl_warning(struct __sysctl_args *args);
65
66 #if defined(CONFIG_SYSCTL)
67
68 /* External variables not in a header file. */
69 extern int C_A_D;
70 extern int print_fatal_signals;
71 extern int sysctl_overcommit_memory;
72 extern int sysctl_overcommit_ratio;
73 extern int sysctl_panic_on_oom;
74 extern int sysctl_oom_kill_allocating_task;
75 extern int sysctl_oom_dump_tasks;
76 extern int max_threads;
77 extern int core_uses_pid;
78 extern int suid_dumpable;
79 extern char core_pattern[];
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 extern int blk_iopoll_enabled;
95
96 /* Constants used for minimum and maximum */
97 #ifdef CONFIG_DETECT_SOFTLOCKUP
98 static int sixty = 60;
99 static int neg_one = -1;
100 #endif
101
102 static int zero;
103 static int __maybe_unused one = 1;
104 static int __maybe_unused two = 2;
105 static unsigned long one_ul = 1;
106 static int one_hundred = 100;
107
108 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
109 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
110
111 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
112 static int maxolduid = 65535;
113 static int minolduid;
114 static int min_percpu_pagelist_fract = 8;
115
116 static int ngroups_max = NGROUPS_MAX;
117
118 #ifdef CONFIG_MODULES
119 extern char modprobe_path[];
120 extern int modules_disabled;
121 #endif
122 #ifdef CONFIG_CHR_DEV_SG
123 extern int sg_big_buff;
124 #endif
125
126 #ifdef CONFIG_SPARC
127 #include <asm/system.h>
128 #endif
129
130 #ifdef CONFIG_SPARC64
131 extern int sysctl_tsb_ratio;
132 #endif
133
134 #ifdef __hppa__
135 extern int pwrsw_enabled;
136 extern int unaligned_enabled;
137 #endif
138
139 #ifdef CONFIG_S390
140 #ifdef CONFIG_MATHEMU
141 extern int sysctl_ieee_emulation_warnings;
142 #endif
143 extern int sysctl_userprocess_debug;
144 extern int spin_retry;
145 #endif
146
147 #ifdef CONFIG_BSD_PROCESS_ACCT
148 extern int acct_parm[];
149 #endif
150
151 #ifdef CONFIG_IA64
152 extern int no_unaligned_warning;
153 extern int unaligned_dump_stack;
154 #endif
155
156 #ifdef CONFIG_RT_MUTEXES
157 extern int max_lock_depth;
158 #endif
159
160 #ifdef CONFIG_PROC_SYSCTL
161 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
162 void __user *buffer, size_t *lenp, loff_t *ppos);
163 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
164 void __user *buffer, size_t *lenp, loff_t *ppos);
165 #endif
166
167 static struct ctl_table root_table[];
168 static struct ctl_table_root sysctl_table_root;
169 static struct ctl_table_header root_table_header = {
170 .count = 1,
171 .ctl_table = root_table,
172 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
173 .root = &sysctl_table_root,
174 .set = &sysctl_table_root.default_set,
175 };
176 static struct ctl_table_root sysctl_table_root = {
177 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
178 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
179 };
180
181 static struct ctl_table kern_table[];
182 static struct ctl_table vm_table[];
183 static struct ctl_table fs_table[];
184 static struct ctl_table debug_table[];
185 static struct ctl_table dev_table[];
186 extern struct ctl_table random_table[];
187 #ifdef CONFIG_INOTIFY_USER
188 extern struct ctl_table inotify_table[];
189 #endif
190 #ifdef CONFIG_EPOLL
191 extern struct ctl_table epoll_table[];
192 #endif
193
194 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
195 int sysctl_legacy_va_layout;
196 #endif
197
198 extern int prove_locking;
199 extern int lock_stat;
200
201 /* The default sysctl tables: */
202
203 static struct ctl_table root_table[] = {
204 {
205 .ctl_name = CTL_KERN,
206 .procname = "kernel",
207 .mode = 0555,
208 .child = kern_table,
209 },
210 {
211 .ctl_name = CTL_VM,
212 .procname = "vm",
213 .mode = 0555,
214 .child = vm_table,
215 },
216 {
217 .ctl_name = CTL_FS,
218 .procname = "fs",
219 .mode = 0555,
220 .child = fs_table,
221 },
222 {
223 .ctl_name = CTL_DEBUG,
224 .procname = "debug",
225 .mode = 0555,
226 .child = debug_table,
227 },
228 {
229 .ctl_name = CTL_DEV,
230 .procname = "dev",
231 .mode = 0555,
232 .child = dev_table,
233 },
234 /*
235 * NOTE: do not add new entries to this table unless you have read
236 * Documentation/sysctl/ctl_unnumbered.txt
237 */
238 { .ctl_name = 0 }
239 };
240
241 #ifdef CONFIG_SCHED_DEBUG
242 static int min_sched_granularity_ns = 100000; /* 100 usecs */
243 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
244 static int min_wakeup_granularity_ns; /* 0 usecs */
245 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
246 #endif
247
248 static struct ctl_table kern_table[] = {
249 {
250 .ctl_name = CTL_UNNUMBERED,
251 .procname = "sched_child_runs_first",
252 .data = &sysctl_sched_child_runs_first,
253 .maxlen = sizeof(unsigned int),
254 .mode = 0644,
255 .proc_handler = &proc_dointvec,
256 },
257 #ifdef CONFIG_SCHED_DEBUG
258 {
259 .ctl_name = CTL_UNNUMBERED,
260 .procname = "sched_min_granularity_ns",
261 .data = &sysctl_sched_min_granularity,
262 .maxlen = sizeof(unsigned int),
263 .mode = 0644,
264 .proc_handler = &sched_nr_latency_handler,
265 .strategy = &sysctl_intvec,
266 .extra1 = &min_sched_granularity_ns,
267 .extra2 = &max_sched_granularity_ns,
268 },
269 {
270 .ctl_name = CTL_UNNUMBERED,
271 .procname = "sched_latency_ns",
272 .data = &sysctl_sched_latency,
273 .maxlen = sizeof(unsigned int),
274 .mode = 0644,
275 .proc_handler = &sched_nr_latency_handler,
276 .strategy = &sysctl_intvec,
277 .extra1 = &min_sched_granularity_ns,
278 .extra2 = &max_sched_granularity_ns,
279 },
280 {
281 .ctl_name = CTL_UNNUMBERED,
282 .procname = "sched_wakeup_granularity_ns",
283 .data = &sysctl_sched_wakeup_granularity,
284 .maxlen = sizeof(unsigned int),
285 .mode = 0644,
286 .proc_handler = &proc_dointvec_minmax,
287 .strategy = &sysctl_intvec,
288 .extra1 = &min_wakeup_granularity_ns,
289 .extra2 = &max_wakeup_granularity_ns,
290 },
291 {
292 .ctl_name = CTL_UNNUMBERED,
293 .procname = "sched_shares_ratelimit",
294 .data = &sysctl_sched_shares_ratelimit,
295 .maxlen = sizeof(unsigned int),
296 .mode = 0644,
297 .proc_handler = &proc_dointvec,
298 },
299 {
300 .ctl_name = CTL_UNNUMBERED,
301 .procname = "sched_shares_thresh",
302 .data = &sysctl_sched_shares_thresh,
303 .maxlen = sizeof(unsigned int),
304 .mode = 0644,
305 .proc_handler = &proc_dointvec_minmax,
306 .strategy = &sysctl_intvec,
307 .extra1 = &zero,
308 },
309 {
310 .ctl_name = CTL_UNNUMBERED,
311 .procname = "sched_features",
312 .data = &sysctl_sched_features,
313 .maxlen = sizeof(unsigned int),
314 .mode = 0644,
315 .proc_handler = &proc_dointvec,
316 },
317 {
318 .ctl_name = CTL_UNNUMBERED,
319 .procname = "sched_migration_cost",
320 .data = &sysctl_sched_migration_cost,
321 .maxlen = sizeof(unsigned int),
322 .mode = 0644,
323 .proc_handler = &proc_dointvec,
324 },
325 {
326 .ctl_name = CTL_UNNUMBERED,
327 .procname = "sched_nr_migrate",
328 .data = &sysctl_sched_nr_migrate,
329 .maxlen = sizeof(unsigned int),
330 .mode = 0644,
331 .proc_handler = &proc_dointvec,
332 },
333 {
334 .ctl_name = CTL_UNNUMBERED,
335 .procname = "sched_time_avg",
336 .data = &sysctl_sched_time_avg,
337 .maxlen = sizeof(unsigned int),
338 .mode = 0644,
339 .proc_handler = &proc_dointvec,
340 },
341 {
342 .ctl_name = CTL_UNNUMBERED,
343 .procname = "timer_migration",
344 .data = &sysctl_timer_migration,
345 .maxlen = sizeof(unsigned int),
346 .mode = 0644,
347 .proc_handler = &proc_dointvec_minmax,
348 .strategy = &sysctl_intvec,
349 .extra1 = &zero,
350 .extra2 = &one,
351 },
352 #endif
353 {
354 .ctl_name = CTL_UNNUMBERED,
355 .procname = "sched_rt_period_us",
356 .data = &sysctl_sched_rt_period,
357 .maxlen = sizeof(unsigned int),
358 .mode = 0644,
359 .proc_handler = &sched_rt_handler,
360 },
361 {
362 .ctl_name = CTL_UNNUMBERED,
363 .procname = "sched_rt_runtime_us",
364 .data = &sysctl_sched_rt_runtime,
365 .maxlen = sizeof(int),
366 .mode = 0644,
367 .proc_handler = &sched_rt_handler,
368 },
369 {
370 .ctl_name = CTL_UNNUMBERED,
371 .procname = "sched_compat_yield",
372 .data = &sysctl_sched_compat_yield,
373 .maxlen = sizeof(unsigned int),
374 .mode = 0644,
375 .proc_handler = &proc_dointvec,
376 },
377 #ifdef CONFIG_PROVE_LOCKING
378 {
379 .ctl_name = CTL_UNNUMBERED,
380 .procname = "prove_locking",
381 .data = &prove_locking,
382 .maxlen = sizeof(int),
383 .mode = 0644,
384 .proc_handler = &proc_dointvec,
385 },
386 #endif
387 #ifdef CONFIG_LOCK_STAT
388 {
389 .ctl_name = CTL_UNNUMBERED,
390 .procname = "lock_stat",
391 .data = &lock_stat,
392 .maxlen = sizeof(int),
393 .mode = 0644,
394 .proc_handler = &proc_dointvec,
395 },
396 #endif
397 {
398 .ctl_name = KERN_PANIC,
399 .procname = "panic",
400 .data = &panic_timeout,
401 .maxlen = sizeof(int),
402 .mode = 0644,
403 .proc_handler = &proc_dointvec,
404 },
405 {
406 .ctl_name = KERN_CORE_USES_PID,
407 .procname = "core_uses_pid",
408 .data = &core_uses_pid,
409 .maxlen = sizeof(int),
410 .mode = 0644,
411 .proc_handler = &proc_dointvec,
412 },
413 {
414 .ctl_name = KERN_CORE_PATTERN,
415 .procname = "core_pattern",
416 .data = core_pattern,
417 .maxlen = CORENAME_MAX_SIZE,
418 .mode = 0644,
419 .proc_handler = &proc_dostring,
420 .strategy = &sysctl_string,
421 },
422 #ifdef CONFIG_PROC_SYSCTL
423 {
424 .procname = "tainted",
425 .maxlen = sizeof(long),
426 .mode = 0644,
427 .proc_handler = &proc_taint,
428 },
429 #endif
430 #ifdef CONFIG_LATENCYTOP
431 {
432 .procname = "latencytop",
433 .data = &latencytop_enabled,
434 .maxlen = sizeof(int),
435 .mode = 0644,
436 .proc_handler = &proc_dointvec,
437 },
438 #endif
439 #ifdef CONFIG_BLK_DEV_INITRD
440 {
441 .ctl_name = KERN_REALROOTDEV,
442 .procname = "real-root-dev",
443 .data = &real_root_dev,
444 .maxlen = sizeof(int),
445 .mode = 0644,
446 .proc_handler = &proc_dointvec,
447 },
448 #endif
449 {
450 .ctl_name = CTL_UNNUMBERED,
451 .procname = "print-fatal-signals",
452 .data = &print_fatal_signals,
453 .maxlen = sizeof(int),
454 .mode = 0644,
455 .proc_handler = &proc_dointvec,
456 },
457 #ifdef CONFIG_SPARC
458 {
459 .ctl_name = KERN_SPARC_REBOOT,
460 .procname = "reboot-cmd",
461 .data = reboot_command,
462 .maxlen = 256,
463 .mode = 0644,
464 .proc_handler = &proc_dostring,
465 .strategy = &sysctl_string,
466 },
467 {
468 .ctl_name = KERN_SPARC_STOP_A,
469 .procname = "stop-a",
470 .data = &stop_a_enabled,
471 .maxlen = sizeof (int),
472 .mode = 0644,
473 .proc_handler = &proc_dointvec,
474 },
475 {
476 .ctl_name = KERN_SPARC_SCONS_PWROFF,
477 .procname = "scons-poweroff",
478 .data = &scons_pwroff,
479 .maxlen = sizeof (int),
480 .mode = 0644,
481 .proc_handler = &proc_dointvec,
482 },
483 #endif
484 #ifdef CONFIG_SPARC64
485 {
486 .ctl_name = CTL_UNNUMBERED,
487 .procname = "tsb-ratio",
488 .data = &sysctl_tsb_ratio,
489 .maxlen = sizeof (int),
490 .mode = 0644,
491 .proc_handler = &proc_dointvec,
492 },
493 #endif
494 #ifdef __hppa__
495 {
496 .ctl_name = KERN_HPPA_PWRSW,
497 .procname = "soft-power",
498 .data = &pwrsw_enabled,
499 .maxlen = sizeof (int),
500 .mode = 0644,
501 .proc_handler = &proc_dointvec,
502 },
503 {
504 .ctl_name = KERN_HPPA_UNALIGNED,
505 .procname = "unaligned-trap",
506 .data = &unaligned_enabled,
507 .maxlen = sizeof (int),
508 .mode = 0644,
509 .proc_handler = &proc_dointvec,
510 },
511 #endif
512 {
513 .ctl_name = KERN_CTLALTDEL,
514 .procname = "ctrl-alt-del",
515 .data = &C_A_D,
516 .maxlen = sizeof(int),
517 .mode = 0644,
518 .proc_handler = &proc_dointvec,
519 },
520 #ifdef CONFIG_FUNCTION_TRACER
521 {
522 .ctl_name = CTL_UNNUMBERED,
523 .procname = "ftrace_enabled",
524 .data = &ftrace_enabled,
525 .maxlen = sizeof(int),
526 .mode = 0644,
527 .proc_handler = &ftrace_enable_sysctl,
528 },
529 #endif
530 #ifdef CONFIG_STACK_TRACER
531 {
532 .ctl_name = CTL_UNNUMBERED,
533 .procname = "stack_tracer_enabled",
534 .data = &stack_tracer_enabled,
535 .maxlen = sizeof(int),
536 .mode = 0644,
537 .proc_handler = &stack_trace_sysctl,
538 },
539 #endif
540 #ifdef CONFIG_TRACING
541 {
542 .ctl_name = CTL_UNNUMBERED,
543 .procname = "ftrace_dump_on_oops",
544 .data = &ftrace_dump_on_oops,
545 .maxlen = sizeof(int),
546 .mode = 0644,
547 .proc_handler = &proc_dointvec,
548 },
549 #endif
550 #ifdef CONFIG_MODULES
551 {
552 .ctl_name = KERN_MODPROBE,
553 .procname = "modprobe",
554 .data = &modprobe_path,
555 .maxlen = KMOD_PATH_LEN,
556 .mode = 0644,
557 .proc_handler = &proc_dostring,
558 .strategy = &sysctl_string,
559 },
560 {
561 .ctl_name = CTL_UNNUMBERED,
562 .procname = "modules_disabled",
563 .data = &modules_disabled,
564 .maxlen = sizeof(int),
565 .mode = 0644,
566 /* only handle a transition from default "0" to "1" */
567 .proc_handler = &proc_dointvec_minmax,
568 .extra1 = &one,
569 .extra2 = &one,
570 },
571 #endif
572 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
573 {
574 .ctl_name = KERN_HOTPLUG,
575 .procname = "hotplug",
576 .data = &uevent_helper,
577 .maxlen = UEVENT_HELPER_PATH_LEN,
578 .mode = 0644,
579 .proc_handler = &proc_dostring,
580 .strategy = &sysctl_string,
581 },
582 #endif
583 #ifdef CONFIG_CHR_DEV_SG
584 {
585 .ctl_name = KERN_SG_BIG_BUFF,
586 .procname = "sg-big-buff",
587 .data = &sg_big_buff,
588 .maxlen = sizeof (int),
589 .mode = 0444,
590 .proc_handler = &proc_dointvec,
591 },
592 #endif
593 #ifdef CONFIG_BSD_PROCESS_ACCT
594 {
595 .ctl_name = KERN_ACCT,
596 .procname = "acct",
597 .data = &acct_parm,
598 .maxlen = 3*sizeof(int),
599 .mode = 0644,
600 .proc_handler = &proc_dointvec,
601 },
602 #endif
603 #ifdef CONFIG_MAGIC_SYSRQ
604 {
605 .ctl_name = KERN_SYSRQ,
606 .procname = "sysrq",
607 .data = &__sysrq_enabled,
608 .maxlen = sizeof (int),
609 .mode = 0644,
610 .proc_handler = &proc_dointvec,
611 },
612 #endif
613 #ifdef CONFIG_PROC_SYSCTL
614 {
615 .procname = "cad_pid",
616 .data = NULL,
617 .maxlen = sizeof (int),
618 .mode = 0600,
619 .proc_handler = &proc_do_cad_pid,
620 },
621 #endif
622 {
623 .ctl_name = KERN_MAX_THREADS,
624 .procname = "threads-max",
625 .data = &max_threads,
626 .maxlen = sizeof(int),
627 .mode = 0644,
628 .proc_handler = &proc_dointvec,
629 },
630 {
631 .ctl_name = KERN_RANDOM,
632 .procname = "random",
633 .mode = 0555,
634 .child = random_table,
635 },
636 {
637 .ctl_name = KERN_OVERFLOWUID,
638 .procname = "overflowuid",
639 .data = &overflowuid,
640 .maxlen = sizeof(int),
641 .mode = 0644,
642 .proc_handler = &proc_dointvec_minmax,
643 .strategy = &sysctl_intvec,
644 .extra1 = &minolduid,
645 .extra2 = &maxolduid,
646 },
647 {
648 .ctl_name = KERN_OVERFLOWGID,
649 .procname = "overflowgid",
650 .data = &overflowgid,
651 .maxlen = sizeof(int),
652 .mode = 0644,
653 .proc_handler = &proc_dointvec_minmax,
654 .strategy = &sysctl_intvec,
655 .extra1 = &minolduid,
656 .extra2 = &maxolduid,
657 },
658 #ifdef CONFIG_S390
659 #ifdef CONFIG_MATHEMU
660 {
661 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
662 .procname = "ieee_emulation_warnings",
663 .data = &sysctl_ieee_emulation_warnings,
664 .maxlen = sizeof(int),
665 .mode = 0644,
666 .proc_handler = &proc_dointvec,
667 },
668 #endif
669 {
670 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
671 .procname = "userprocess_debug",
672 .data = &sysctl_userprocess_debug,
673 .maxlen = sizeof(int),
674 .mode = 0644,
675 .proc_handler = &proc_dointvec,
676 },
677 #endif
678 {
679 .ctl_name = KERN_PIDMAX,
680 .procname = "pid_max",
681 .data = &pid_max,
682 .maxlen = sizeof (int),
683 .mode = 0644,
684 .proc_handler = &proc_dointvec_minmax,
685 .strategy = sysctl_intvec,
686 .extra1 = &pid_max_min,
687 .extra2 = &pid_max_max,
688 },
689 {
690 .ctl_name = KERN_PANIC_ON_OOPS,
691 .procname = "panic_on_oops",
692 .data = &panic_on_oops,
693 .maxlen = sizeof(int),
694 .mode = 0644,
695 .proc_handler = &proc_dointvec,
696 },
697 #if defined CONFIG_PRINTK
698 {
699 .ctl_name = KERN_PRINTK,
700 .procname = "printk",
701 .data = &console_loglevel,
702 .maxlen = 4*sizeof(int),
703 .mode = 0644,
704 .proc_handler = &proc_dointvec,
705 },
706 {
707 .ctl_name = KERN_PRINTK_RATELIMIT,
708 .procname = "printk_ratelimit",
709 .data = &printk_ratelimit_state.interval,
710 .maxlen = sizeof(int),
711 .mode = 0644,
712 .proc_handler = &proc_dointvec_jiffies,
713 .strategy = &sysctl_jiffies,
714 },
715 {
716 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
717 .procname = "printk_ratelimit_burst",
718 .data = &printk_ratelimit_state.burst,
719 .maxlen = sizeof(int),
720 .mode = 0644,
721 .proc_handler = &proc_dointvec,
722 },
723 #endif
724 {
725 .ctl_name = KERN_NGROUPS_MAX,
726 .procname = "ngroups_max",
727 .data = &ngroups_max,
728 .maxlen = sizeof (int),
729 .mode = 0444,
730 .proc_handler = &proc_dointvec,
731 },
732 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
733 {
734 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
735 .procname = "unknown_nmi_panic",
736 .data = &unknown_nmi_panic,
737 .maxlen = sizeof (int),
738 .mode = 0644,
739 .proc_handler = &proc_dointvec,
740 },
741 {
742 .procname = "nmi_watchdog",
743 .data = &nmi_watchdog_enabled,
744 .maxlen = sizeof (int),
745 .mode = 0644,
746 .proc_handler = &proc_nmi_enabled,
747 },
748 #endif
749 #if defined(CONFIG_X86)
750 {
751 .ctl_name = KERN_PANIC_ON_NMI,
752 .procname = "panic_on_unrecovered_nmi",
753 .data = &panic_on_unrecovered_nmi,
754 .maxlen = sizeof(int),
755 .mode = 0644,
756 .proc_handler = &proc_dointvec,
757 },
758 {
759 .ctl_name = CTL_UNNUMBERED,
760 .procname = "panic_on_io_nmi",
761 .data = &panic_on_io_nmi,
762 .maxlen = sizeof(int),
763 .mode = 0644,
764 .proc_handler = &proc_dointvec,
765 },
766 {
767 .ctl_name = KERN_BOOTLOADER_TYPE,
768 .procname = "bootloader_type",
769 .data = &bootloader_type,
770 .maxlen = sizeof (int),
771 .mode = 0444,
772 .proc_handler = &proc_dointvec,
773 },
774 {
775 .ctl_name = CTL_UNNUMBERED,
776 .procname = "bootloader_version",
777 .data = &bootloader_version,
778 .maxlen = sizeof (int),
779 .mode = 0444,
780 .proc_handler = &proc_dointvec,
781 },
782 {
783 .ctl_name = CTL_UNNUMBERED,
784 .procname = "kstack_depth_to_print",
785 .data = &kstack_depth_to_print,
786 .maxlen = sizeof(int),
787 .mode = 0644,
788 .proc_handler = &proc_dointvec,
789 },
790 {
791 .ctl_name = CTL_UNNUMBERED,
792 .procname = "io_delay_type",
793 .data = &io_delay_type,
794 .maxlen = sizeof(int),
795 .mode = 0644,
796 .proc_handler = &proc_dointvec,
797 },
798 #endif
799 #if defined(CONFIG_MMU)
800 {
801 .ctl_name = KERN_RANDOMIZE,
802 .procname = "randomize_va_space",
803 .data = &randomize_va_space,
804 .maxlen = sizeof(int),
805 .mode = 0644,
806 .proc_handler = &proc_dointvec,
807 },
808 #endif
809 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
810 {
811 .ctl_name = KERN_SPIN_RETRY,
812 .procname = "spin_retry",
813 .data = &spin_retry,
814 .maxlen = sizeof (int),
815 .mode = 0644,
816 .proc_handler = &proc_dointvec,
817 },
818 #endif
819 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
820 {
821 .procname = "acpi_video_flags",
822 .data = &acpi_realmode_flags,
823 .maxlen = sizeof (unsigned long),
824 .mode = 0644,
825 .proc_handler = &proc_doulongvec_minmax,
826 },
827 #endif
828 #ifdef CONFIG_IA64
829 {
830 .ctl_name = KERN_IA64_UNALIGNED,
831 .procname = "ignore-unaligned-usertrap",
832 .data = &no_unaligned_warning,
833 .maxlen = sizeof (int),
834 .mode = 0644,
835 .proc_handler = &proc_dointvec,
836 },
837 {
838 .ctl_name = CTL_UNNUMBERED,
839 .procname = "unaligned-dump-stack",
840 .data = &unaligned_dump_stack,
841 .maxlen = sizeof (int),
842 .mode = 0644,
843 .proc_handler = &proc_dointvec,
844 },
845 #endif
846 #ifdef CONFIG_DETECT_SOFTLOCKUP
847 {
848 .ctl_name = CTL_UNNUMBERED,
849 .procname = "softlockup_panic",
850 .data = &softlockup_panic,
851 .maxlen = sizeof(int),
852 .mode = 0644,
853 .proc_handler = &proc_dointvec_minmax,
854 .strategy = &sysctl_intvec,
855 .extra1 = &zero,
856 .extra2 = &one,
857 },
858 {
859 .ctl_name = CTL_UNNUMBERED,
860 .procname = "softlockup_thresh",
861 .data = &softlockup_thresh,
862 .maxlen = sizeof(int),
863 .mode = 0644,
864 .proc_handler = &proc_dosoftlockup_thresh,
865 .strategy = &sysctl_intvec,
866 .extra1 = &neg_one,
867 .extra2 = &sixty,
868 },
869 #endif
870 #ifdef CONFIG_DETECT_HUNG_TASK
871 {
872 .ctl_name = CTL_UNNUMBERED,
873 .procname = "hung_task_panic",
874 .data = &sysctl_hung_task_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 = "hung_task_check_count",
885 .data = &sysctl_hung_task_check_count,
886 .maxlen = sizeof(unsigned long),
887 .mode = 0644,
888 .proc_handler = &proc_doulongvec_minmax,
889 .strategy = &sysctl_intvec,
890 },
891 {
892 .ctl_name = CTL_UNNUMBERED,
893 .procname = "hung_task_timeout_secs",
894 .data = &sysctl_hung_task_timeout_secs,
895 .maxlen = sizeof(unsigned long),
896 .mode = 0644,
897 .proc_handler = &proc_dohung_task_timeout_secs,
898 .strategy = &sysctl_intvec,
899 },
900 {
901 .ctl_name = CTL_UNNUMBERED,
902 .procname = "hung_task_warnings",
903 .data = &sysctl_hung_task_warnings,
904 .maxlen = sizeof(unsigned long),
905 .mode = 0644,
906 .proc_handler = &proc_doulongvec_minmax,
907 .strategy = &sysctl_intvec,
908 },
909 #endif
910 #ifdef CONFIG_COMPAT
911 {
912 .ctl_name = KERN_COMPAT_LOG,
913 .procname = "compat-log",
914 .data = &compat_log,
915 .maxlen = sizeof (int),
916 .mode = 0644,
917 .proc_handler = &proc_dointvec,
918 },
919 #endif
920 #ifdef CONFIG_RT_MUTEXES
921 {
922 .ctl_name = KERN_MAX_LOCK_DEPTH,
923 .procname = "max_lock_depth",
924 .data = &max_lock_depth,
925 .maxlen = sizeof(int),
926 .mode = 0644,
927 .proc_handler = &proc_dointvec,
928 },
929 #endif
930 {
931 .ctl_name = CTL_UNNUMBERED,
932 .procname = "poweroff_cmd",
933 .data = &poweroff_cmd,
934 .maxlen = POWEROFF_CMD_PATH_LEN,
935 .mode = 0644,
936 .proc_handler = &proc_dostring,
937 .strategy = &sysctl_string,
938 },
939 #ifdef CONFIG_KEYS
940 {
941 .ctl_name = CTL_UNNUMBERED,
942 .procname = "keys",
943 .mode = 0555,
944 .child = key_sysctls,
945 },
946 #endif
947 #ifdef CONFIG_RCU_TORTURE_TEST
948 {
949 .ctl_name = CTL_UNNUMBERED,
950 .procname = "rcutorture_runnable",
951 .data = &rcutorture_runnable,
952 .maxlen = sizeof(int),
953 .mode = 0644,
954 .proc_handler = &proc_dointvec,
955 },
956 #endif
957 #ifdef CONFIG_SLOW_WORK
958 {
959 .ctl_name = CTL_UNNUMBERED,
960 .procname = "slow-work",
961 .mode = 0555,
962 .child = slow_work_sysctls,
963 },
964 #endif
965 #ifdef CONFIG_PERF_COUNTERS
966 {
967 .ctl_name = CTL_UNNUMBERED,
968 .procname = "perf_counter_paranoid",
969 .data = &sysctl_perf_counter_paranoid,
970 .maxlen = sizeof(sysctl_perf_counter_paranoid),
971 .mode = 0644,
972 .proc_handler = &proc_dointvec,
973 },
974 {
975 .ctl_name = CTL_UNNUMBERED,
976 .procname = "perf_counter_mlock_kb",
977 .data = &sysctl_perf_counter_mlock,
978 .maxlen = sizeof(sysctl_perf_counter_mlock),
979 .mode = 0644,
980 .proc_handler = &proc_dointvec,
981 },
982 {
983 .ctl_name = CTL_UNNUMBERED,
984 .procname = "perf_counter_max_sample_rate",
985 .data = &sysctl_perf_counter_sample_rate,
986 .maxlen = sizeof(sysctl_perf_counter_sample_rate),
987 .mode = 0644,
988 .proc_handler = &proc_dointvec,
989 },
990 #endif
991 #ifdef CONFIG_KMEMCHECK
992 {
993 .ctl_name = CTL_UNNUMBERED,
994 .procname = "kmemcheck",
995 .data = &kmemcheck_enabled,
996 .maxlen = sizeof(int),
997 .mode = 0644,
998 .proc_handler = &proc_dointvec,
999 },
1000 #endif
1001 {
1002 .ctl_name = CTL_UNNUMBERED,
1003 .procname = "blk_iopoll",
1004 .data = &blk_iopoll_enabled,
1005 .maxlen = sizeof(int),
1006 .mode = 0644,
1007 .proc_handler = &proc_dointvec,
1008 },
1009 /*
1010 * NOTE: do not add new entries to this table unless you have read
1011 * Documentation/sysctl/ctl_unnumbered.txt
1012 */
1013 { .ctl_name = 0 }
1014 };
1015
1016 static struct ctl_table vm_table[] = {
1017 {
1018 .ctl_name = VM_OVERCOMMIT_MEMORY,
1019 .procname = "overcommit_memory",
1020 .data = &sysctl_overcommit_memory,
1021 .maxlen = sizeof(sysctl_overcommit_memory),
1022 .mode = 0644,
1023 .proc_handler = &proc_dointvec,
1024 },
1025 {
1026 .ctl_name = VM_PANIC_ON_OOM,
1027 .procname = "panic_on_oom",
1028 .data = &sysctl_panic_on_oom,
1029 .maxlen = sizeof(sysctl_panic_on_oom),
1030 .mode = 0644,
1031 .proc_handler = &proc_dointvec,
1032 },
1033 {
1034 .ctl_name = CTL_UNNUMBERED,
1035 .procname = "oom_kill_allocating_task",
1036 .data = &sysctl_oom_kill_allocating_task,
1037 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1038 .mode = 0644,
1039 .proc_handler = &proc_dointvec,
1040 },
1041 {
1042 .ctl_name = CTL_UNNUMBERED,
1043 .procname = "oom_dump_tasks",
1044 .data = &sysctl_oom_dump_tasks,
1045 .maxlen = sizeof(sysctl_oom_dump_tasks),
1046 .mode = 0644,
1047 .proc_handler = &proc_dointvec,
1048 },
1049 {
1050 .ctl_name = VM_OVERCOMMIT_RATIO,
1051 .procname = "overcommit_ratio",
1052 .data = &sysctl_overcommit_ratio,
1053 .maxlen = sizeof(sysctl_overcommit_ratio),
1054 .mode = 0644,
1055 .proc_handler = &proc_dointvec,
1056 },
1057 {
1058 .ctl_name = VM_PAGE_CLUSTER,
1059 .procname = "page-cluster",
1060 .data = &page_cluster,
1061 .maxlen = sizeof(int),
1062 .mode = 0644,
1063 .proc_handler = &proc_dointvec,
1064 },
1065 {
1066 .ctl_name = VM_DIRTY_BACKGROUND,
1067 .procname = "dirty_background_ratio",
1068 .data = &dirty_background_ratio,
1069 .maxlen = sizeof(dirty_background_ratio),
1070 .mode = 0644,
1071 .proc_handler = &dirty_background_ratio_handler,
1072 .strategy = &sysctl_intvec,
1073 .extra1 = &zero,
1074 .extra2 = &one_hundred,
1075 },
1076 {
1077 .ctl_name = CTL_UNNUMBERED,
1078 .procname = "dirty_background_bytes",
1079 .data = &dirty_background_bytes,
1080 .maxlen = sizeof(dirty_background_bytes),
1081 .mode = 0644,
1082 .proc_handler = &dirty_background_bytes_handler,
1083 .strategy = &sysctl_intvec,
1084 .extra1 = &one_ul,
1085 },
1086 {
1087 .ctl_name = VM_DIRTY_RATIO,
1088 .procname = "dirty_ratio",
1089 .data = &vm_dirty_ratio,
1090 .maxlen = sizeof(vm_dirty_ratio),
1091 .mode = 0644,
1092 .proc_handler = &dirty_ratio_handler,
1093 .strategy = &sysctl_intvec,
1094 .extra1 = &zero,
1095 .extra2 = &one_hundred,
1096 },
1097 {
1098 .ctl_name = CTL_UNNUMBERED,
1099 .procname = "dirty_bytes",
1100 .data = &vm_dirty_bytes,
1101 .maxlen = sizeof(vm_dirty_bytes),
1102 .mode = 0644,
1103 .proc_handler = &dirty_bytes_handler,
1104 .strategy = &sysctl_intvec,
1105 .extra1 = &dirty_bytes_min,
1106 },
1107 {
1108 .procname = "dirty_writeback_centisecs",
1109 .data = &dirty_writeback_interval,
1110 .maxlen = sizeof(dirty_writeback_interval),
1111 .mode = 0644,
1112 .proc_handler = &dirty_writeback_centisecs_handler,
1113 },
1114 {
1115 .procname = "dirty_expire_centisecs",
1116 .data = &dirty_expire_interval,
1117 .maxlen = sizeof(dirty_expire_interval),
1118 .mode = 0644,
1119 .proc_handler = &proc_dointvec,
1120 },
1121 {
1122 .ctl_name = VM_NR_PDFLUSH_THREADS,
1123 .procname = "nr_pdflush_threads",
1124 .data = &nr_pdflush_threads,
1125 .maxlen = sizeof nr_pdflush_threads,
1126 .mode = 0444 /* read-only*/,
1127 .proc_handler = &proc_dointvec,
1128 },
1129 {
1130 .ctl_name = VM_SWAPPINESS,
1131 .procname = "swappiness",
1132 .data = &vm_swappiness,
1133 .maxlen = sizeof(vm_swappiness),
1134 .mode = 0644,
1135 .proc_handler = &proc_dointvec_minmax,
1136 .strategy = &sysctl_intvec,
1137 .extra1 = &zero,
1138 .extra2 = &one_hundred,
1139 },
1140 #ifdef CONFIG_HUGETLB_PAGE
1141 {
1142 .procname = "nr_hugepages",
1143 .data = NULL,
1144 .maxlen = sizeof(unsigned long),
1145 .mode = 0644,
1146 .proc_handler = &hugetlb_sysctl_handler,
1147 .extra1 = (void *)&hugetlb_zero,
1148 .extra2 = (void *)&hugetlb_infinity,
1149 },
1150 {
1151 .ctl_name = VM_HUGETLB_GROUP,
1152 .procname = "hugetlb_shm_group",
1153 .data = &sysctl_hugetlb_shm_group,
1154 .maxlen = sizeof(gid_t),
1155 .mode = 0644,
1156 .proc_handler = &proc_dointvec,
1157 },
1158 {
1159 .ctl_name = CTL_UNNUMBERED,
1160 .procname = "hugepages_treat_as_movable",
1161 .data = &hugepages_treat_as_movable,
1162 .maxlen = sizeof(int),
1163 .mode = 0644,
1164 .proc_handler = &hugetlb_treat_movable_handler,
1165 },
1166 {
1167 .ctl_name = CTL_UNNUMBERED,
1168 .procname = "nr_overcommit_hugepages",
1169 .data = NULL,
1170 .maxlen = sizeof(unsigned long),
1171 .mode = 0644,
1172 .proc_handler = &hugetlb_overcommit_handler,
1173 .extra1 = (void *)&hugetlb_zero,
1174 .extra2 = (void *)&hugetlb_infinity,
1175 },
1176 #endif
1177 {
1178 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
1179 .procname = "lowmem_reserve_ratio",
1180 .data = &sysctl_lowmem_reserve_ratio,
1181 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1182 .mode = 0644,
1183 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
1184 .strategy = &sysctl_intvec,
1185 },
1186 {
1187 .ctl_name = VM_DROP_PAGECACHE,
1188 .procname = "drop_caches",
1189 .data = &sysctl_drop_caches,
1190 .maxlen = sizeof(int),
1191 .mode = 0644,
1192 .proc_handler = drop_caches_sysctl_handler,
1193 .strategy = &sysctl_intvec,
1194 },
1195 {
1196 .ctl_name = VM_MIN_FREE_KBYTES,
1197 .procname = "min_free_kbytes",
1198 .data = &min_free_kbytes,
1199 .maxlen = sizeof(min_free_kbytes),
1200 .mode = 0644,
1201 .proc_handler = &min_free_kbytes_sysctl_handler,
1202 .strategy = &sysctl_intvec,
1203 .extra1 = &zero,
1204 },
1205 {
1206 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
1207 .procname = "percpu_pagelist_fraction",
1208 .data = &percpu_pagelist_fraction,
1209 .maxlen = sizeof(percpu_pagelist_fraction),
1210 .mode = 0644,
1211 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
1212 .strategy = &sysctl_intvec,
1213 .extra1 = &min_percpu_pagelist_fract,
1214 },
1215 #ifdef CONFIG_MMU
1216 {
1217 .ctl_name = VM_MAX_MAP_COUNT,
1218 .procname = "max_map_count",
1219 .data = &sysctl_max_map_count,
1220 .maxlen = sizeof(sysctl_max_map_count),
1221 .mode = 0644,
1222 .proc_handler = &proc_dointvec
1223 },
1224 #else
1225 {
1226 .ctl_name = CTL_UNNUMBERED,
1227 .procname = "nr_trim_pages",
1228 .data = &sysctl_nr_trim_pages,
1229 .maxlen = sizeof(sysctl_nr_trim_pages),
1230 .mode = 0644,
1231 .proc_handler = &proc_dointvec_minmax,
1232 .strategy = &sysctl_intvec,
1233 .extra1 = &zero,
1234 },
1235 #endif
1236 {
1237 .ctl_name = VM_LAPTOP_MODE,
1238 .procname = "laptop_mode",
1239 .data = &laptop_mode,
1240 .maxlen = sizeof(laptop_mode),
1241 .mode = 0644,
1242 .proc_handler = &proc_dointvec_jiffies,
1243 .strategy = &sysctl_jiffies,
1244 },
1245 {
1246 .ctl_name = VM_BLOCK_DUMP,
1247 .procname = "block_dump",
1248 .data = &block_dump,
1249 .maxlen = sizeof(block_dump),
1250 .mode = 0644,
1251 .proc_handler = &proc_dointvec,
1252 .strategy = &sysctl_intvec,
1253 .extra1 = &zero,
1254 },
1255 {
1256 .ctl_name = VM_VFS_CACHE_PRESSURE,
1257 .procname = "vfs_cache_pressure",
1258 .data = &sysctl_vfs_cache_pressure,
1259 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1260 .mode = 0644,
1261 .proc_handler = &proc_dointvec,
1262 .strategy = &sysctl_intvec,
1263 .extra1 = &zero,
1264 },
1265 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1266 {
1267 .ctl_name = VM_LEGACY_VA_LAYOUT,
1268 .procname = "legacy_va_layout",
1269 .data = &sysctl_legacy_va_layout,
1270 .maxlen = sizeof(sysctl_legacy_va_layout),
1271 .mode = 0644,
1272 .proc_handler = &proc_dointvec,
1273 .strategy = &sysctl_intvec,
1274 .extra1 = &zero,
1275 },
1276 #endif
1277 #ifdef CONFIG_NUMA
1278 {
1279 .ctl_name = VM_ZONE_RECLAIM_MODE,
1280 .procname = "zone_reclaim_mode",
1281 .data = &zone_reclaim_mode,
1282 .maxlen = sizeof(zone_reclaim_mode),
1283 .mode = 0644,
1284 .proc_handler = &proc_dointvec,
1285 .strategy = &sysctl_intvec,
1286 .extra1 = &zero,
1287 },
1288 {
1289 .ctl_name = VM_MIN_UNMAPPED,
1290 .procname = "min_unmapped_ratio",
1291 .data = &sysctl_min_unmapped_ratio,
1292 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1293 .mode = 0644,
1294 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
1295 .strategy = &sysctl_intvec,
1296 .extra1 = &zero,
1297 .extra2 = &one_hundred,
1298 },
1299 {
1300 .ctl_name = VM_MIN_SLAB,
1301 .procname = "min_slab_ratio",
1302 .data = &sysctl_min_slab_ratio,
1303 .maxlen = sizeof(sysctl_min_slab_ratio),
1304 .mode = 0644,
1305 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1306 .strategy = &sysctl_intvec,
1307 .extra1 = &zero,
1308 .extra2 = &one_hundred,
1309 },
1310 #endif
1311 #ifdef CONFIG_SMP
1312 {
1313 .ctl_name = CTL_UNNUMBERED,
1314 .procname = "stat_interval",
1315 .data = &sysctl_stat_interval,
1316 .maxlen = sizeof(sysctl_stat_interval),
1317 .mode = 0644,
1318 .proc_handler = &proc_dointvec_jiffies,
1319 .strategy = &sysctl_jiffies,
1320 },
1321 #endif
1322 {
1323 .ctl_name = CTL_UNNUMBERED,
1324 .procname = "mmap_min_addr",
1325 .data = &dac_mmap_min_addr,
1326 .maxlen = sizeof(unsigned long),
1327 .mode = 0644,
1328 .proc_handler = &mmap_min_addr_handler,
1329 },
1330 #ifdef CONFIG_NUMA
1331 {
1332 .ctl_name = CTL_UNNUMBERED,
1333 .procname = "numa_zonelist_order",
1334 .data = &numa_zonelist_order,
1335 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1336 .mode = 0644,
1337 .proc_handler = &numa_zonelist_order_handler,
1338 .strategy = &sysctl_string,
1339 },
1340 #endif
1341 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1342 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1343 {
1344 .ctl_name = VM_VDSO_ENABLED,
1345 .procname = "vdso_enabled",
1346 .data = &vdso_enabled,
1347 .maxlen = sizeof(vdso_enabled),
1348 .mode = 0644,
1349 .proc_handler = &proc_dointvec,
1350 .strategy = &sysctl_intvec,
1351 .extra1 = &zero,
1352 },
1353 #endif
1354 #ifdef CONFIG_HIGHMEM
1355 {
1356 .ctl_name = CTL_UNNUMBERED,
1357 .procname = "highmem_is_dirtyable",
1358 .data = &vm_highmem_is_dirtyable,
1359 .maxlen = sizeof(vm_highmem_is_dirtyable),
1360 .mode = 0644,
1361 .proc_handler = &proc_dointvec_minmax,
1362 .strategy = &sysctl_intvec,
1363 .extra1 = &zero,
1364 .extra2 = &one,
1365 },
1366 #endif
1367 {
1368 .ctl_name = CTL_UNNUMBERED,
1369 .procname = "scan_unevictable_pages",
1370 .data = &scan_unevictable_pages,
1371 .maxlen = sizeof(scan_unevictable_pages),
1372 .mode = 0644,
1373 .proc_handler = &scan_unevictable_handler,
1374 },
1375 #ifdef CONFIG_MEMORY_FAILURE
1376 {
1377 .ctl_name = CTL_UNNUMBERED,
1378 .procname = "memory_failure_early_kill",
1379 .data = &sysctl_memory_failure_early_kill,
1380 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1381 .mode = 0644,
1382 .proc_handler = &proc_dointvec_minmax,
1383 .strategy = &sysctl_intvec,
1384 .extra1 = &zero,
1385 .extra2 = &one,
1386 },
1387 {
1388 .ctl_name = CTL_UNNUMBERED,
1389 .procname = "memory_failure_recovery",
1390 .data = &sysctl_memory_failure_recovery,
1391 .maxlen = sizeof(sysctl_memory_failure_recovery),
1392 .mode = 0644,
1393 .proc_handler = &proc_dointvec_minmax,
1394 .strategy = &sysctl_intvec,
1395 .extra1 = &zero,
1396 .extra2 = &one,
1397 },
1398 #endif
1399
1400 /*
1401 * NOTE: do not add new entries to this table unless you have read
1402 * Documentation/sysctl/ctl_unnumbered.txt
1403 */
1404 { .ctl_name = 0 }
1405 };
1406
1407 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1408 static struct ctl_table binfmt_misc_table[] = {
1409 { .ctl_name = 0 }
1410 };
1411 #endif
1412
1413 static struct ctl_table fs_table[] = {
1414 {
1415 .ctl_name = FS_NRINODE,
1416 .procname = "inode-nr",
1417 .data = &inodes_stat,
1418 .maxlen = 2*sizeof(int),
1419 .mode = 0444,
1420 .proc_handler = &proc_dointvec,
1421 },
1422 {
1423 .ctl_name = FS_STATINODE,
1424 .procname = "inode-state",
1425 .data = &inodes_stat,
1426 .maxlen = 7*sizeof(int),
1427 .mode = 0444,
1428 .proc_handler = &proc_dointvec,
1429 },
1430 {
1431 .procname = "file-nr",
1432 .data = &files_stat,
1433 .maxlen = 3*sizeof(int),
1434 .mode = 0444,
1435 .proc_handler = &proc_nr_files,
1436 },
1437 {
1438 .ctl_name = FS_MAXFILE,
1439 .procname = "file-max",
1440 .data = &files_stat.max_files,
1441 .maxlen = sizeof(int),
1442 .mode = 0644,
1443 .proc_handler = &proc_dointvec,
1444 },
1445 {
1446 .ctl_name = CTL_UNNUMBERED,
1447 .procname = "nr_open",
1448 .data = &sysctl_nr_open,
1449 .maxlen = sizeof(int),
1450 .mode = 0644,
1451 .proc_handler = &proc_dointvec_minmax,
1452 .extra1 = &sysctl_nr_open_min,
1453 .extra2 = &sysctl_nr_open_max,
1454 },
1455 {
1456 .ctl_name = FS_DENTRY,
1457 .procname = "dentry-state",
1458 .data = &dentry_stat,
1459 .maxlen = 6*sizeof(int),
1460 .mode = 0444,
1461 .proc_handler = &proc_dointvec,
1462 },
1463 {
1464 .ctl_name = FS_OVERFLOWUID,
1465 .procname = "overflowuid",
1466 .data = &fs_overflowuid,
1467 .maxlen = sizeof(int),
1468 .mode = 0644,
1469 .proc_handler = &proc_dointvec_minmax,
1470 .strategy = &sysctl_intvec,
1471 .extra1 = &minolduid,
1472 .extra2 = &maxolduid,
1473 },
1474 {
1475 .ctl_name = FS_OVERFLOWGID,
1476 .procname = "overflowgid",
1477 .data = &fs_overflowgid,
1478 .maxlen = sizeof(int),
1479 .mode = 0644,
1480 .proc_handler = &proc_dointvec_minmax,
1481 .strategy = &sysctl_intvec,
1482 .extra1 = &minolduid,
1483 .extra2 = &maxolduid,
1484 },
1485 #ifdef CONFIG_FILE_LOCKING
1486 {
1487 .ctl_name = FS_LEASES,
1488 .procname = "leases-enable",
1489 .data = &leases_enable,
1490 .maxlen = sizeof(int),
1491 .mode = 0644,
1492 .proc_handler = &proc_dointvec,
1493 },
1494 #endif
1495 #ifdef CONFIG_DNOTIFY
1496 {
1497 .ctl_name = FS_DIR_NOTIFY,
1498 .procname = "dir-notify-enable",
1499 .data = &dir_notify_enable,
1500 .maxlen = sizeof(int),
1501 .mode = 0644,
1502 .proc_handler = &proc_dointvec,
1503 },
1504 #endif
1505 #ifdef CONFIG_MMU
1506 #ifdef CONFIG_FILE_LOCKING
1507 {
1508 .ctl_name = FS_LEASE_TIME,
1509 .procname = "lease-break-time",
1510 .data = &lease_break_time,
1511 .maxlen = sizeof(int),
1512 .mode = 0644,
1513 .proc_handler = &proc_dointvec,
1514 },
1515 #endif
1516 #ifdef CONFIG_AIO
1517 {
1518 .procname = "aio-nr",
1519 .data = &aio_nr,
1520 .maxlen = sizeof(aio_nr),
1521 .mode = 0444,
1522 .proc_handler = &proc_doulongvec_minmax,
1523 },
1524 {
1525 .procname = "aio-max-nr",
1526 .data = &aio_max_nr,
1527 .maxlen = sizeof(aio_max_nr),
1528 .mode = 0644,
1529 .proc_handler = &proc_doulongvec_minmax,
1530 },
1531 #endif /* CONFIG_AIO */
1532 #ifdef CONFIG_INOTIFY_USER
1533 {
1534 .ctl_name = FS_INOTIFY,
1535 .procname = "inotify",
1536 .mode = 0555,
1537 .child = inotify_table,
1538 },
1539 #endif
1540 #ifdef CONFIG_EPOLL
1541 {
1542 .procname = "epoll",
1543 .mode = 0555,
1544 .child = epoll_table,
1545 },
1546 #endif
1547 #endif
1548 {
1549 .ctl_name = KERN_SETUID_DUMPABLE,
1550 .procname = "suid_dumpable",
1551 .data = &suid_dumpable,
1552 .maxlen = sizeof(int),
1553 .mode = 0644,
1554 .proc_handler = &proc_dointvec_minmax,
1555 .strategy = &sysctl_intvec,
1556 .extra1 = &zero,
1557 .extra2 = &two,
1558 },
1559 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1560 {
1561 .ctl_name = CTL_UNNUMBERED,
1562 .procname = "binfmt_misc",
1563 .mode = 0555,
1564 .child = binfmt_misc_table,
1565 },
1566 #endif
1567 /*
1568 * NOTE: do not add new entries to this table unless you have read
1569 * Documentation/sysctl/ctl_unnumbered.txt
1570 */
1571 { .ctl_name = 0 }
1572 };
1573
1574 static struct ctl_table debug_table[] = {
1575 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1576 {
1577 .ctl_name = CTL_UNNUMBERED,
1578 .procname = "exception-trace",
1579 .data = &show_unhandled_signals,
1580 .maxlen = sizeof(int),
1581 .mode = 0644,
1582 .proc_handler = proc_dointvec
1583 },
1584 #endif
1585 { .ctl_name = 0 }
1586 };
1587
1588 static struct ctl_table dev_table[] = {
1589 { .ctl_name = 0 }
1590 };
1591
1592 static DEFINE_SPINLOCK(sysctl_lock);
1593
1594 /* called under sysctl_lock */
1595 static int use_table(struct ctl_table_header *p)
1596 {
1597 if (unlikely(p->unregistering))
1598 return 0;
1599 p->used++;
1600 return 1;
1601 }
1602
1603 /* called under sysctl_lock */
1604 static void unuse_table(struct ctl_table_header *p)
1605 {
1606 if (!--p->used)
1607 if (unlikely(p->unregistering))
1608 complete(p->unregistering);
1609 }
1610
1611 /* called under sysctl_lock, will reacquire if has to wait */
1612 static void start_unregistering(struct ctl_table_header *p)
1613 {
1614 /*
1615 * if p->used is 0, nobody will ever touch that entry again;
1616 * we'll eliminate all paths to it before dropping sysctl_lock
1617 */
1618 if (unlikely(p->used)) {
1619 struct completion wait;
1620 init_completion(&wait);
1621 p->unregistering = &wait;
1622 spin_unlock(&sysctl_lock);
1623 wait_for_completion(&wait);
1624 spin_lock(&sysctl_lock);
1625 } else {
1626 /* anything non-NULL; we'll never dereference it */
1627 p->unregistering = ERR_PTR(-EINVAL);
1628 }
1629 /*
1630 * do not remove from the list until nobody holds it; walking the
1631 * list in do_sysctl() relies on that.
1632 */
1633 list_del_init(&p->ctl_entry);
1634 }
1635
1636 void sysctl_head_get(struct ctl_table_header *head)
1637 {
1638 spin_lock(&sysctl_lock);
1639 head->count++;
1640 spin_unlock(&sysctl_lock);
1641 }
1642
1643 void sysctl_head_put(struct ctl_table_header *head)
1644 {
1645 spin_lock(&sysctl_lock);
1646 if (!--head->count)
1647 kfree(head);
1648 spin_unlock(&sysctl_lock);
1649 }
1650
1651 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1652 {
1653 if (!head)
1654 BUG();
1655 spin_lock(&sysctl_lock);
1656 if (!use_table(head))
1657 head = ERR_PTR(-ENOENT);
1658 spin_unlock(&sysctl_lock);
1659 return head;
1660 }
1661
1662 void sysctl_head_finish(struct ctl_table_header *head)
1663 {
1664 if (!head)
1665 return;
1666 spin_lock(&sysctl_lock);
1667 unuse_table(head);
1668 spin_unlock(&sysctl_lock);
1669 }
1670
1671 static struct ctl_table_set *
1672 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1673 {
1674 struct ctl_table_set *set = &root->default_set;
1675 if (root->lookup)
1676 set = root->lookup(root, namespaces);
1677 return set;
1678 }
1679
1680 static struct list_head *
1681 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1682 {
1683 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1684 return &set->list;
1685 }
1686
1687 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1688 struct ctl_table_header *prev)
1689 {
1690 struct ctl_table_root *root;
1691 struct list_head *header_list;
1692 struct ctl_table_header *head;
1693 struct list_head *tmp;
1694
1695 spin_lock(&sysctl_lock);
1696 if (prev) {
1697 head = prev;
1698 tmp = &prev->ctl_entry;
1699 unuse_table(prev);
1700 goto next;
1701 }
1702 tmp = &root_table_header.ctl_entry;
1703 for (;;) {
1704 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1705
1706 if (!use_table(head))
1707 goto next;
1708 spin_unlock(&sysctl_lock);
1709 return head;
1710 next:
1711 root = head->root;
1712 tmp = tmp->next;
1713 header_list = lookup_header_list(root, namespaces);
1714 if (tmp != header_list)
1715 continue;
1716
1717 do {
1718 root = list_entry(root->root_list.next,
1719 struct ctl_table_root, root_list);
1720 if (root == &sysctl_table_root)
1721 goto out;
1722 header_list = lookup_header_list(root, namespaces);
1723 } while (list_empty(header_list));
1724 tmp = header_list->next;
1725 }
1726 out:
1727 spin_unlock(&sysctl_lock);
1728 return NULL;
1729 }
1730
1731 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1732 {
1733 return __sysctl_head_next(current->nsproxy, prev);
1734 }
1735
1736 void register_sysctl_root(struct ctl_table_root *root)
1737 {
1738 spin_lock(&sysctl_lock);
1739 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1740 spin_unlock(&sysctl_lock);
1741 }
1742
1743 #ifdef CONFIG_SYSCTL_SYSCALL
1744 /* Perform the actual read/write of a sysctl table entry. */
1745 static int do_sysctl_strategy(struct ctl_table_root *root,
1746 struct ctl_table *table,
1747 void __user *oldval, size_t __user *oldlenp,
1748 void __user *newval, size_t newlen)
1749 {
1750 int op = 0, rc;
1751
1752 if (oldval)
1753 op |= MAY_READ;
1754 if (newval)
1755 op |= MAY_WRITE;
1756 if (sysctl_perm(root, table, op))
1757 return -EPERM;
1758
1759 if (table->strategy) {
1760 rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1761 if (rc < 0)
1762 return rc;
1763 if (rc > 0)
1764 return 0;
1765 }
1766
1767 /* If there is no strategy routine, or if the strategy returns
1768 * zero, proceed with automatic r/w */
1769 if (table->data && table->maxlen) {
1770 rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1771 if (rc < 0)
1772 return rc;
1773 }
1774 return 0;
1775 }
1776
1777 static int parse_table(int __user *name, int nlen,
1778 void __user *oldval, size_t __user *oldlenp,
1779 void __user *newval, size_t newlen,
1780 struct ctl_table_root *root,
1781 struct ctl_table *table)
1782 {
1783 int n;
1784 repeat:
1785 if (!nlen)
1786 return -ENOTDIR;
1787 if (get_user(n, name))
1788 return -EFAULT;
1789 for ( ; table->ctl_name || table->procname; table++) {
1790 if (!table->ctl_name)
1791 continue;
1792 if (n == table->ctl_name) {
1793 int error;
1794 if (table->child) {
1795 if (sysctl_perm(root, table, MAY_EXEC))
1796 return -EPERM;
1797 name++;
1798 nlen--;
1799 table = table->child;
1800 goto repeat;
1801 }
1802 error = do_sysctl_strategy(root, table,
1803 oldval, oldlenp,
1804 newval, newlen);
1805 return error;
1806 }
1807 }
1808 return -ENOTDIR;
1809 }
1810
1811 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1812 void __user *newval, size_t newlen)
1813 {
1814 struct ctl_table_header *head;
1815 int error = -ENOTDIR;
1816
1817 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1818 return -ENOTDIR;
1819 if (oldval) {
1820 int old_len;
1821 if (!oldlenp || get_user(old_len, oldlenp))
1822 return -EFAULT;
1823 }
1824
1825 for (head = sysctl_head_next(NULL); head;
1826 head = sysctl_head_next(head)) {
1827 error = parse_table(name, nlen, oldval, oldlenp,
1828 newval, newlen,
1829 head->root, head->ctl_table);
1830 if (error != -ENOTDIR) {
1831 sysctl_head_finish(head);
1832 break;
1833 }
1834 }
1835 return error;
1836 }
1837
1838 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
1839 {
1840 struct __sysctl_args tmp;
1841 int error;
1842
1843 if (copy_from_user(&tmp, args, sizeof(tmp)))
1844 return -EFAULT;
1845
1846 error = deprecated_sysctl_warning(&tmp);
1847 if (error)
1848 goto out;
1849
1850 lock_kernel();
1851 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1852 tmp.newval, tmp.newlen);
1853 unlock_kernel();
1854 out:
1855 return error;
1856 }
1857 #endif /* CONFIG_SYSCTL_SYSCALL */
1858
1859 /*
1860 * sysctl_perm does NOT grant the superuser all rights automatically, because
1861 * some sysctl variables are readonly even to root.
1862 */
1863
1864 static int test_perm(int mode, int op)
1865 {
1866 if (!current_euid())
1867 mode >>= 6;
1868 else if (in_egroup_p(0))
1869 mode >>= 3;
1870 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1871 return 0;
1872 return -EACCES;
1873 }
1874
1875 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1876 {
1877 int error;
1878 int mode;
1879
1880 error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1881 if (error)
1882 return error;
1883
1884 if (root->permissions)
1885 mode = root->permissions(root, current->nsproxy, table);
1886 else
1887 mode = table->mode;
1888
1889 return test_perm(mode, op);
1890 }
1891
1892 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1893 {
1894 for (; table->ctl_name || table->procname; table++) {
1895 table->parent = parent;
1896 if (table->child)
1897 sysctl_set_parent(table, table->child);
1898 }
1899 }
1900
1901 static __init int sysctl_init(void)
1902 {
1903 sysctl_set_parent(NULL, root_table);
1904 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1905 {
1906 int err;
1907 err = sysctl_check_table(current->nsproxy, root_table);
1908 }
1909 #endif
1910 return 0;
1911 }
1912
1913 core_initcall(sysctl_init);
1914
1915 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1916 struct ctl_table *table)
1917 {
1918 struct ctl_table *p;
1919 const char *s = branch->procname;
1920
1921 /* branch should have named subdirectory as its first element */
1922 if (!s || !branch->child)
1923 return NULL;
1924
1925 /* ... and nothing else */
1926 if (branch[1].procname || branch[1].ctl_name)
1927 return NULL;
1928
1929 /* table should contain subdirectory with the same name */
1930 for (p = table; p->procname || p->ctl_name; p++) {
1931 if (!p->child)
1932 continue;
1933 if (p->procname && strcmp(p->procname, s) == 0)
1934 return p;
1935 }
1936 return NULL;
1937 }
1938
1939 /* see if attaching q to p would be an improvement */
1940 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1941 {
1942 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1943 struct ctl_table *next;
1944 int is_better = 0;
1945 int not_in_parent = !p->attached_by;
1946
1947 while ((next = is_branch_in(by, to)) != NULL) {
1948 if (by == q->attached_by)
1949 is_better = 1;
1950 if (to == p->attached_by)
1951 not_in_parent = 1;
1952 by = by->child;
1953 to = next->child;
1954 }
1955
1956 if (is_better && not_in_parent) {
1957 q->attached_by = by;
1958 q->attached_to = to;
1959 q->parent = p;
1960 }
1961 }
1962
1963 /**
1964 * __register_sysctl_paths - register a sysctl hierarchy
1965 * @root: List of sysctl headers to register on
1966 * @namespaces: Data to compute which lists of sysctl entries are visible
1967 * @path: The path to the directory the sysctl table is in.
1968 * @table: the top-level table structure
1969 *
1970 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1971 * array. A completely 0 filled entry terminates the table.
1972 *
1973 * The members of the &struct ctl_table structure are used as follows:
1974 *
1975 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1976 * must be unique within that level of sysctl
1977 *
1978 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1979 * enter a sysctl file
1980 *
1981 * data - a pointer to data for use by proc_handler
1982 *
1983 * maxlen - the maximum size in bytes of the data
1984 *
1985 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1986 *
1987 * child - a pointer to the child sysctl table if this entry is a directory, or
1988 * %NULL.
1989 *
1990 * proc_handler - the text handler routine (described below)
1991 *
1992 * strategy - the strategy routine (described below)
1993 *
1994 * de - for internal use by the sysctl routines
1995 *
1996 * extra1, extra2 - extra pointers usable by the proc handler routines
1997 *
1998 * Leaf nodes in the sysctl tree will be represented by a single file
1999 * under /proc; non-leaf nodes will be represented by directories.
2000 *
2001 * sysctl(2) can automatically manage read and write requests through
2002 * the sysctl table. The data and maxlen fields of the ctl_table
2003 * struct enable minimal validation of the values being written to be
2004 * performed, and the mode field allows minimal authentication.
2005 *
2006 * More sophisticated management can be enabled by the provision of a
2007 * strategy routine with the table entry. This will be called before
2008 * any automatic read or write of the data is performed.
2009 *
2010 * The strategy routine may return
2011 *
2012 * < 0 - Error occurred (error is passed to user process)
2013 *
2014 * 0 - OK - proceed with automatic read or write.
2015 *
2016 * > 0 - OK - read or write has been done by the strategy routine, so
2017 * return immediately.
2018 *
2019 * There must be a proc_handler routine for any terminal nodes
2020 * mirrored under /proc/sys (non-terminals are handled by a built-in
2021 * directory handler). Several default handlers are available to
2022 * cover common cases -
2023 *
2024 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
2025 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
2026 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
2027 *
2028 * It is the handler's job to read the input buffer from user memory
2029 * and process it. The handler should return 0 on success.
2030 *
2031 * This routine returns %NULL on a failure to register, and a pointer
2032 * to the table header on success.
2033 */
2034 struct ctl_table_header *__register_sysctl_paths(
2035 struct ctl_table_root *root,
2036 struct nsproxy *namespaces,
2037 const struct ctl_path *path, struct ctl_table *table)
2038 {
2039 struct ctl_table_header *header;
2040 struct ctl_table *new, **prevp;
2041 unsigned int n, npath;
2042 struct ctl_table_set *set;
2043
2044 /* Count the path components */
2045 for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
2046 ;
2047
2048 /*
2049 * For each path component, allocate a 2-element ctl_table array.
2050 * The first array element will be filled with the sysctl entry
2051 * for this, the second will be the sentinel (ctl_name == 0).
2052 *
2053 * We allocate everything in one go so that we don't have to
2054 * worry about freeing additional memory in unregister_sysctl_table.
2055 */
2056 header = kzalloc(sizeof(struct ctl_table_header) +
2057 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
2058 if (!header)
2059 return NULL;
2060
2061 new = (struct ctl_table *) (header + 1);
2062
2063 /* Now connect the dots */
2064 prevp = &header->ctl_table;
2065 for (n = 0; n < npath; ++n, ++path) {
2066 /* Copy the procname */
2067 new->procname = path->procname;
2068 new->ctl_name = path->ctl_name;
2069 new->mode = 0555;
2070
2071 *prevp = new;
2072 prevp = &new->child;
2073
2074 new += 2;
2075 }
2076 *prevp = table;
2077 header->ctl_table_arg = table;
2078
2079 INIT_LIST_HEAD(&header->ctl_entry);
2080 header->used = 0;
2081 header->unregistering = NULL;
2082 header->root = root;
2083 sysctl_set_parent(NULL, header->ctl_table);
2084 header->count = 1;
2085 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
2086 if (sysctl_check_table(namespaces, header->ctl_table)) {
2087 kfree(header);
2088 return NULL;
2089 }
2090 #endif
2091 spin_lock(&sysctl_lock);
2092 header->set = lookup_header_set(root, namespaces);
2093 header->attached_by = header->ctl_table;
2094 header->attached_to = root_table;
2095 header->parent = &root_table_header;
2096 for (set = header->set; set; set = set->parent) {
2097 struct ctl_table_header *p;
2098 list_for_each_entry(p, &set->list, ctl_entry) {
2099 if (p->unregistering)
2100 continue;
2101 try_attach(p, header);
2102 }
2103 }
2104 header->parent->count++;
2105 list_add_tail(&header->ctl_entry, &header->set->list);
2106 spin_unlock(&sysctl_lock);
2107
2108 return header;
2109 }
2110
2111 /**
2112 * register_sysctl_table_path - register a sysctl table hierarchy
2113 * @path: The path to the directory the sysctl table is in.
2114 * @table: the top-level table structure
2115 *
2116 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2117 * array. A completely 0 filled entry terminates the table.
2118 *
2119 * See __register_sysctl_paths for more details.
2120 */
2121 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2122 struct ctl_table *table)
2123 {
2124 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
2125 path, table);
2126 }
2127
2128 /**
2129 * register_sysctl_table - register a sysctl table hierarchy
2130 * @table: the top-level table structure
2131 *
2132 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2133 * array. A completely 0 filled entry terminates the table.
2134 *
2135 * See register_sysctl_paths for more details.
2136 */
2137 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
2138 {
2139 static const struct ctl_path null_path[] = { {} };
2140
2141 return register_sysctl_paths(null_path, table);
2142 }
2143
2144 /**
2145 * unregister_sysctl_table - unregister a sysctl table hierarchy
2146 * @header: the header returned from register_sysctl_table
2147 *
2148 * Unregisters the sysctl table and all children. proc entries may not
2149 * actually be removed until they are no longer used by anyone.
2150 */
2151 void unregister_sysctl_table(struct ctl_table_header * header)
2152 {
2153 might_sleep();
2154
2155 if (header == NULL)
2156 return;
2157
2158 spin_lock(&sysctl_lock);
2159 start_unregistering(header);
2160 if (!--header->parent->count) {
2161 WARN_ON(1);
2162 kfree(header->parent);
2163 }
2164 if (!--header->count)
2165 kfree(header);
2166 spin_unlock(&sysctl_lock);
2167 }
2168
2169 int sysctl_is_seen(struct ctl_table_header *p)
2170 {
2171 struct ctl_table_set *set = p->set;
2172 int res;
2173 spin_lock(&sysctl_lock);
2174 if (p->unregistering)
2175 res = 0;
2176 else if (!set->is_seen)
2177 res = 1;
2178 else
2179 res = set->is_seen(set);
2180 spin_unlock(&sysctl_lock);
2181 return res;
2182 }
2183
2184 void setup_sysctl_set(struct ctl_table_set *p,
2185 struct ctl_table_set *parent,
2186 int (*is_seen)(struct ctl_table_set *))
2187 {
2188 INIT_LIST_HEAD(&p->list);
2189 p->parent = parent ? parent : &sysctl_table_root.default_set;
2190 p->is_seen = is_seen;
2191 }
2192
2193 #else /* !CONFIG_SYSCTL */
2194 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2195 {
2196 return NULL;
2197 }
2198
2199 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2200 struct ctl_table *table)
2201 {
2202 return NULL;
2203 }
2204
2205 void unregister_sysctl_table(struct ctl_table_header * table)
2206 {
2207 }
2208
2209 void setup_sysctl_set(struct ctl_table_set *p,
2210 struct ctl_table_set *parent,
2211 int (*is_seen)(struct ctl_table_set *))
2212 {
2213 }
2214
2215 void sysctl_head_put(struct ctl_table_header *head)
2216 {
2217 }
2218
2219 #endif /* CONFIG_SYSCTL */
2220
2221 /*
2222 * /proc/sys support
2223 */
2224
2225 #ifdef CONFIG_PROC_SYSCTL
2226
2227 static int _proc_do_string(void* data, int maxlen, int write,
2228 struct file *filp, void __user *buffer,
2229 size_t *lenp, loff_t *ppos)
2230 {
2231 size_t len;
2232 char __user *p;
2233 char c;
2234
2235 if (!data || !maxlen || !*lenp) {
2236 *lenp = 0;
2237 return 0;
2238 }
2239
2240 if (write) {
2241 len = 0;
2242 p = buffer;
2243 while (len < *lenp) {
2244 if (get_user(c, p++))
2245 return -EFAULT;
2246 if (c == 0 || c == '\n')
2247 break;
2248 len++;
2249 }
2250 if (len >= maxlen)
2251 len = maxlen-1;
2252 if(copy_from_user(data, buffer, len))
2253 return -EFAULT;
2254 ((char *) data)[len] = 0;
2255 *ppos += *lenp;
2256 } else {
2257 len = strlen(data);
2258 if (len > maxlen)
2259 len = maxlen;
2260
2261 if (*ppos > len) {
2262 *lenp = 0;
2263 return 0;
2264 }
2265
2266 data += *ppos;
2267 len -= *ppos;
2268
2269 if (len > *lenp)
2270 len = *lenp;
2271 if (len)
2272 if(copy_to_user(buffer, data, len))
2273 return -EFAULT;
2274 if (len < *lenp) {
2275 if(put_user('\n', ((char __user *) buffer) + len))
2276 return -EFAULT;
2277 len++;
2278 }
2279 *lenp = len;
2280 *ppos += len;
2281 }
2282 return 0;
2283 }
2284
2285 /**
2286 * proc_dostring - read a string sysctl
2287 * @table: the sysctl table
2288 * @write: %TRUE if this is a write to the sysctl file
2289 * @filp: the file structure
2290 * @buffer: the user buffer
2291 * @lenp: the size of the user buffer
2292 * @ppos: file position
2293 *
2294 * Reads/writes a string from/to the user buffer. If the kernel
2295 * buffer provided is not large enough to hold the string, the
2296 * string is truncated. The copied string is %NULL-terminated.
2297 * If the string is being read by the user process, it is copied
2298 * and a newline '\n' is added. It is truncated if the buffer is
2299 * not large enough.
2300 *
2301 * Returns 0 on success.
2302 */
2303 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2304 void __user *buffer, size_t *lenp, loff_t *ppos)
2305 {
2306 return _proc_do_string(table->data, table->maxlen, write, filp,
2307 buffer, lenp, ppos);
2308 }
2309
2310
2311 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2312 int *valp,
2313 int write, void *data)
2314 {
2315 if (write) {
2316 *valp = *negp ? -*lvalp : *lvalp;
2317 } else {
2318 int val = *valp;
2319 if (val < 0) {
2320 *negp = -1;
2321 *lvalp = (unsigned long)-val;
2322 } else {
2323 *negp = 0;
2324 *lvalp = (unsigned long)val;
2325 }
2326 }
2327 return 0;
2328 }
2329
2330 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2331 int write, struct file *filp, void __user *buffer,
2332 size_t *lenp, loff_t *ppos,
2333 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2334 int write, void *data),
2335 void *data)
2336 {
2337 #define TMPBUFLEN 21
2338 int *i, vleft, first = 1, neg;
2339 unsigned long lval;
2340 size_t left, len;
2341
2342 char buf[TMPBUFLEN], *p;
2343 char __user *s = buffer;
2344
2345 if (!tbl_data || !table->maxlen || !*lenp ||
2346 (*ppos && !write)) {
2347 *lenp = 0;
2348 return 0;
2349 }
2350
2351 i = (int *) tbl_data;
2352 vleft = table->maxlen / sizeof(*i);
2353 left = *lenp;
2354
2355 if (!conv)
2356 conv = do_proc_dointvec_conv;
2357
2358 for (; left && vleft--; i++, first=0) {
2359 if (write) {
2360 while (left) {
2361 char c;
2362 if (get_user(c, s))
2363 return -EFAULT;
2364 if (!isspace(c))
2365 break;
2366 left--;
2367 s++;
2368 }
2369 if (!left)
2370 break;
2371 neg = 0;
2372 len = left;
2373 if (len > sizeof(buf) - 1)
2374 len = sizeof(buf) - 1;
2375 if (copy_from_user(buf, s, len))
2376 return -EFAULT;
2377 buf[len] = 0;
2378 p = buf;
2379 if (*p == '-' && left > 1) {
2380 neg = 1;
2381 p++;
2382 }
2383 if (*p < '0' || *p > '9')
2384 break;
2385
2386 lval = simple_strtoul(p, &p, 0);
2387
2388 len = p-buf;
2389 if ((len < left) && *p && !isspace(*p))
2390 break;
2391 s += len;
2392 left -= len;
2393
2394 if (conv(&neg, &lval, i, 1, data))
2395 break;
2396 } else {
2397 p = buf;
2398 if (!first)
2399 *p++ = '\t';
2400
2401 if (conv(&neg, &lval, i, 0, data))
2402 break;
2403
2404 sprintf(p, "%s%lu", neg ? "-" : "", lval);
2405 len = strlen(buf);
2406 if (len > left)
2407 len = left;
2408 if(copy_to_user(s, buf, len))
2409 return -EFAULT;
2410 left -= len;
2411 s += len;
2412 }
2413 }
2414
2415 if (!write && !first && left) {
2416 if(put_user('\n', s))
2417 return -EFAULT;
2418 left--, s++;
2419 }
2420 if (write) {
2421 while (left) {
2422 char c;
2423 if (get_user(c, s++))
2424 return -EFAULT;
2425 if (!isspace(c))
2426 break;
2427 left--;
2428 }
2429 }
2430 if (write && first)
2431 return -EINVAL;
2432 *lenp -= left;
2433 *ppos += *lenp;
2434 return 0;
2435 #undef TMPBUFLEN
2436 }
2437
2438 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2439 void __user *buffer, size_t *lenp, loff_t *ppos,
2440 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2441 int write, void *data),
2442 void *data)
2443 {
2444 return __do_proc_dointvec(table->data, table, write, filp,
2445 buffer, lenp, ppos, conv, data);
2446 }
2447
2448 /**
2449 * proc_dointvec - read a vector of integers
2450 * @table: the sysctl table
2451 * @write: %TRUE if this is a write to the sysctl file
2452 * @filp: the file structure
2453 * @buffer: the user buffer
2454 * @lenp: the size of the user buffer
2455 * @ppos: file position
2456 *
2457 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2458 * values from/to the user buffer, treated as an ASCII string.
2459 *
2460 * Returns 0 on success.
2461 */
2462 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2463 void __user *buffer, size_t *lenp, loff_t *ppos)
2464 {
2465 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2466 NULL,NULL);
2467 }
2468
2469 /*
2470 * Taint values can only be increased
2471 * This means we can safely use a temporary.
2472 */
2473 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
2474 void __user *buffer, size_t *lenp, loff_t *ppos)
2475 {
2476 struct ctl_table t;
2477 unsigned long tmptaint = get_taint();
2478 int err;
2479
2480 if (write && !capable(CAP_SYS_ADMIN))
2481 return -EPERM;
2482
2483 t = *table;
2484 t.data = &tmptaint;
2485 err = proc_doulongvec_minmax(&t, write, filp, buffer, lenp, ppos);
2486 if (err < 0)
2487 return err;
2488
2489 if (write) {
2490 /*
2491 * Poor man's atomic or. Not worth adding a primitive
2492 * to everyone's atomic.h for this
2493 */
2494 int i;
2495 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2496 if ((tmptaint >> i) & 1)
2497 add_taint(i);
2498 }
2499 }
2500
2501 return err;
2502 }
2503
2504 struct do_proc_dointvec_minmax_conv_param {
2505 int *min;
2506 int *max;
2507 };
2508
2509 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2510 int *valp,
2511 int write, void *data)
2512 {
2513 struct do_proc_dointvec_minmax_conv_param *param = data;
2514 if (write) {
2515 int val = *negp ? -*lvalp : *lvalp;
2516 if ((param->min && *param->min > val) ||
2517 (param->max && *param->max < val))
2518 return -EINVAL;
2519 *valp = val;
2520 } else {
2521 int val = *valp;
2522 if (val < 0) {
2523 *negp = -1;
2524 *lvalp = (unsigned long)-val;
2525 } else {
2526 *negp = 0;
2527 *lvalp = (unsigned long)val;
2528 }
2529 }
2530 return 0;
2531 }
2532
2533 /**
2534 * proc_dointvec_minmax - read a vector of integers with min/max values
2535 * @table: the sysctl table
2536 * @write: %TRUE if this is a write to the sysctl file
2537 * @filp: the file structure
2538 * @buffer: the user buffer
2539 * @lenp: the size of the user buffer
2540 * @ppos: file position
2541 *
2542 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2543 * values from/to the user buffer, treated as an ASCII string.
2544 *
2545 * This routine will ensure the values are within the range specified by
2546 * table->extra1 (min) and table->extra2 (max).
2547 *
2548 * Returns 0 on success.
2549 */
2550 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2551 void __user *buffer, size_t *lenp, loff_t *ppos)
2552 {
2553 struct do_proc_dointvec_minmax_conv_param param = {
2554 .min = (int *) table->extra1,
2555 .max = (int *) table->extra2,
2556 };
2557 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2558 do_proc_dointvec_minmax_conv, &param);
2559 }
2560
2561 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2562 struct file *filp,
2563 void __user *buffer,
2564 size_t *lenp, loff_t *ppos,
2565 unsigned long convmul,
2566 unsigned long convdiv)
2567 {
2568 #define TMPBUFLEN 21
2569 unsigned long *i, *min, *max, val;
2570 int vleft, first=1, neg;
2571 size_t len, left;
2572 char buf[TMPBUFLEN], *p;
2573 char __user *s = buffer;
2574
2575 if (!data || !table->maxlen || !*lenp ||
2576 (*ppos && !write)) {
2577 *lenp = 0;
2578 return 0;
2579 }
2580
2581 i = (unsigned long *) data;
2582 min = (unsigned long *) table->extra1;
2583 max = (unsigned long *) table->extra2;
2584 vleft = table->maxlen / sizeof(unsigned long);
2585 left = *lenp;
2586
2587 for (; left && vleft--; i++, min++, max++, first=0) {
2588 if (write) {
2589 while (left) {
2590 char c;
2591 if (get_user(c, s))
2592 return -EFAULT;
2593 if (!isspace(c))
2594 break;
2595 left--;
2596 s++;
2597 }
2598 if (!left)
2599 break;
2600 neg = 0;
2601 len = left;
2602 if (len > TMPBUFLEN-1)
2603 len = TMPBUFLEN-1;
2604 if (copy_from_user(buf, s, len))
2605 return -EFAULT;
2606 buf[len] = 0;
2607 p = buf;
2608 if (*p == '-' && left > 1) {
2609 neg = 1;
2610 p++;
2611 }
2612 if (*p < '0' || *p > '9')
2613 break;
2614 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2615 len = p-buf;
2616 if ((len < left) && *p && !isspace(*p))
2617 break;
2618 if (neg)
2619 val = -val;
2620 s += len;
2621 left -= len;
2622
2623 if(neg)
2624 continue;
2625 if ((min && val < *min) || (max && val > *max))
2626 continue;
2627 *i = val;
2628 } else {
2629 p = buf;
2630 if (!first)
2631 *p++ = '\t';
2632 sprintf(p, "%lu", convdiv * (*i) / convmul);
2633 len = strlen(buf);
2634 if (len > left)
2635 len = left;
2636 if(copy_to_user(s, buf, len))
2637 return -EFAULT;
2638 left -= len;
2639 s += len;
2640 }
2641 }
2642
2643 if (!write && !first && left) {
2644 if(put_user('\n', s))
2645 return -EFAULT;
2646 left--, s++;
2647 }
2648 if (write) {
2649 while (left) {
2650 char c;
2651 if (get_user(c, s++))
2652 return -EFAULT;
2653 if (!isspace(c))
2654 break;
2655 left--;
2656 }
2657 }
2658 if (write && first)
2659 return -EINVAL;
2660 *lenp -= left;
2661 *ppos += *lenp;
2662 return 0;
2663 #undef TMPBUFLEN
2664 }
2665
2666 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2667 struct file *filp,
2668 void __user *buffer,
2669 size_t *lenp, loff_t *ppos,
2670 unsigned long convmul,
2671 unsigned long convdiv)
2672 {
2673 return __do_proc_doulongvec_minmax(table->data, table, write,
2674 filp, buffer, lenp, ppos, convmul, convdiv);
2675 }
2676
2677 /**
2678 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2679 * @table: the sysctl table
2680 * @write: %TRUE if this is a write to the sysctl file
2681 * @filp: the file structure
2682 * @buffer: the user buffer
2683 * @lenp: the size of the user buffer
2684 * @ppos: file position
2685 *
2686 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2687 * values from/to the user buffer, treated as an ASCII string.
2688 *
2689 * This routine will ensure the values are within the range specified by
2690 * table->extra1 (min) and table->extra2 (max).
2691 *
2692 * Returns 0 on success.
2693 */
2694 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2695 void __user *buffer, size_t *lenp, loff_t *ppos)
2696 {
2697 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2698 }
2699
2700 /**
2701 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2702 * @table: the sysctl table
2703 * @write: %TRUE if this is a write to the sysctl file
2704 * @filp: the file structure
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. The values
2711 * are treated as milliseconds, and converted to jiffies when they are stored.
2712 *
2713 * This routine will ensure the values are within the range specified by
2714 * table->extra1 (min) and table->extra2 (max).
2715 *
2716 * Returns 0 on success.
2717 */
2718 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2719 struct file *filp,
2720 void __user *buffer,
2721 size_t *lenp, loff_t *ppos)
2722 {
2723 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2724 lenp, ppos, HZ, 1000l);
2725 }
2726
2727
2728 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2729 int *valp,
2730 int write, void *data)
2731 {
2732 if (write) {
2733 if (*lvalp > LONG_MAX / HZ)
2734 return 1;
2735 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2736 } else {
2737 int val = *valp;
2738 unsigned long lval;
2739 if (val < 0) {
2740 *negp = -1;
2741 lval = (unsigned long)-val;
2742 } else {
2743 *negp = 0;
2744 lval = (unsigned long)val;
2745 }
2746 *lvalp = lval / HZ;
2747 }
2748 return 0;
2749 }
2750
2751 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2752 int *valp,
2753 int write, void *data)
2754 {
2755 if (write) {
2756 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2757 return 1;
2758 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2759 } else {
2760 int val = *valp;
2761 unsigned long lval;
2762 if (val < 0) {
2763 *negp = -1;
2764 lval = (unsigned long)-val;
2765 } else {
2766 *negp = 0;
2767 lval = (unsigned long)val;
2768 }
2769 *lvalp = jiffies_to_clock_t(lval);
2770 }
2771 return 0;
2772 }
2773
2774 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2775 int *valp,
2776 int write, void *data)
2777 {
2778 if (write) {
2779 *valp = msecs_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_msecs(lval);
2791 }
2792 return 0;
2793 }
2794
2795 /**
2796 * proc_dointvec_jiffies - read a vector of integers as seconds
2797 * @table: the sysctl table
2798 * @write: %TRUE if this is a write to the sysctl file
2799 * @filp: the file structure
2800 * @buffer: the user buffer
2801 * @lenp: the size of the user buffer
2802 * @ppos: file position
2803 *
2804 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2805 * values from/to the user buffer, treated as an ASCII string.
2806 * The values read are assumed to be in seconds, and are converted into
2807 * jiffies.
2808 *
2809 * Returns 0 on success.
2810 */
2811 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2812 void __user *buffer, size_t *lenp, loff_t *ppos)
2813 {
2814 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2815 do_proc_dointvec_jiffies_conv,NULL);
2816 }
2817
2818 /**
2819 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2820 * @table: the sysctl table
2821 * @write: %TRUE if this is a write to the sysctl file
2822 * @filp: the file structure
2823 * @buffer: the user buffer
2824 * @lenp: the size of the user buffer
2825 * @ppos: pointer to the file position
2826 *
2827 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2828 * values from/to the user buffer, treated as an ASCII string.
2829 * The values read are assumed to be in 1/USER_HZ seconds, and
2830 * are converted into jiffies.
2831 *
2832 * Returns 0 on success.
2833 */
2834 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2835 void __user *buffer, size_t *lenp, loff_t *ppos)
2836 {
2837 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2838 do_proc_dointvec_userhz_jiffies_conv,NULL);
2839 }
2840
2841 /**
2842 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2843 * @table: the sysctl table
2844 * @write: %TRUE if this is a write to the sysctl file
2845 * @filp: the file structure
2846 * @buffer: the user buffer
2847 * @lenp: the size of the user buffer
2848 * @ppos: file position
2849 * @ppos: the current position in the file
2850 *
2851 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2852 * values from/to the user buffer, treated as an ASCII string.
2853 * The values read are assumed to be in 1/1000 seconds, and
2854 * are converted into jiffies.
2855 *
2856 * Returns 0 on success.
2857 */
2858 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2859 void __user *buffer, size_t *lenp, loff_t *ppos)
2860 {
2861 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2862 do_proc_dointvec_ms_jiffies_conv, NULL);
2863 }
2864
2865 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2866 void __user *buffer, size_t *lenp, loff_t *ppos)
2867 {
2868 struct pid *new_pid;
2869 pid_t tmp;
2870 int r;
2871
2872 tmp = pid_vnr(cad_pid);
2873
2874 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2875 lenp, ppos, NULL, NULL);
2876 if (r || !write)
2877 return r;
2878
2879 new_pid = find_get_pid(tmp);
2880 if (!new_pid)
2881 return -ESRCH;
2882
2883 put_pid(xchg(&cad_pid, new_pid));
2884 return 0;
2885 }
2886
2887 #else /* CONFIG_PROC_FS */
2888
2889 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2890 void __user *buffer, size_t *lenp, loff_t *ppos)
2891 {
2892 return -ENOSYS;
2893 }
2894
2895 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2896 void __user *buffer, size_t *lenp, loff_t *ppos)
2897 {
2898 return -ENOSYS;
2899 }
2900
2901 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2902 void __user *buffer, size_t *lenp, loff_t *ppos)
2903 {
2904 return -ENOSYS;
2905 }
2906
2907 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2908 void __user *buffer, size_t *lenp, loff_t *ppos)
2909 {
2910 return -ENOSYS;
2911 }
2912
2913 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2914 void __user *buffer, size_t *lenp, loff_t *ppos)
2915 {
2916 return -ENOSYS;
2917 }
2918
2919 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2920 void __user *buffer, size_t *lenp, loff_t *ppos)
2921 {
2922 return -ENOSYS;
2923 }
2924
2925 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2926 void __user *buffer, size_t *lenp, loff_t *ppos)
2927 {
2928 return -ENOSYS;
2929 }
2930
2931 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2932 struct file *filp,
2933 void __user *buffer,
2934 size_t *lenp, loff_t *ppos)
2935 {
2936 return -ENOSYS;
2937 }
2938
2939
2940 #endif /* CONFIG_PROC_FS */
2941
2942
2943 #ifdef CONFIG_SYSCTL_SYSCALL
2944 /*
2945 * General sysctl support routines
2946 */
2947
2948 /* The generic sysctl data routine (used if no strategy routine supplied) */
2949 int sysctl_data(struct ctl_table *table,
2950 void __user *oldval, size_t __user *oldlenp,
2951 void __user *newval, size_t newlen)
2952 {
2953 size_t len;
2954
2955 /* Get out of I don't have a variable */
2956 if (!table->data || !table->maxlen)
2957 return -ENOTDIR;
2958
2959 if (oldval && oldlenp) {
2960 if (get_user(len, oldlenp))
2961 return -EFAULT;
2962 if (len) {
2963 if (len > table->maxlen)
2964 len = table->maxlen;
2965 if (copy_to_user(oldval, table->data, len))
2966 return -EFAULT;
2967 if (put_user(len, oldlenp))
2968 return -EFAULT;
2969 }
2970 }
2971
2972 if (newval && newlen) {
2973 if (newlen > table->maxlen)
2974 newlen = table->maxlen;
2975
2976 if (copy_from_user(table->data, newval, newlen))
2977 return -EFAULT;
2978 }
2979 return 1;
2980 }
2981
2982 /* The generic string strategy routine: */
2983 int sysctl_string(struct ctl_table *table,
2984 void __user *oldval, size_t __user *oldlenp,
2985 void __user *newval, size_t newlen)
2986 {
2987 if (!table->data || !table->maxlen)
2988 return -ENOTDIR;
2989
2990 if (oldval && oldlenp) {
2991 size_t bufsize;
2992 if (get_user(bufsize, oldlenp))
2993 return -EFAULT;
2994 if (bufsize) {
2995 size_t len = strlen(table->data), copied;
2996
2997 /* This shouldn't trigger for a well-formed sysctl */
2998 if (len > table->maxlen)
2999 len = table->maxlen;
3000
3001 /* Copy up to a max of bufsize-1 bytes of the string */
3002 copied = (len >= bufsize) ? bufsize - 1 : len;
3003
3004 if (copy_to_user(oldval, table->data, copied) ||
3005 put_user(0, (char __user *)(oldval + copied)))
3006 return -EFAULT;
3007 if (put_user(len, oldlenp))
3008 return -EFAULT;
3009 }
3010 }
3011 if (newval && newlen) {
3012 size_t len = newlen;
3013 if (len > table->maxlen)
3014 len = table->maxlen;
3015 if(copy_from_user(table->data, newval, len))
3016 return -EFAULT;
3017 if (len == table->maxlen)
3018 len--;
3019 ((char *) table->data)[len] = 0;
3020 }
3021 return 1;
3022 }
3023
3024 /*
3025 * This function makes sure that all of the integers in the vector
3026 * are between the minimum and maximum values given in the arrays
3027 * table->extra1 and table->extra2, respectively.
3028 */
3029 int sysctl_intvec(struct ctl_table *table,
3030 void __user *oldval, size_t __user *oldlenp,
3031 void __user *newval, size_t newlen)
3032 {
3033
3034 if (newval && newlen) {
3035 int __user *vec = (int __user *) newval;
3036 int *min = (int *) table->extra1;
3037 int *max = (int *) table->extra2;
3038 size_t length;
3039 int i;
3040
3041 if (newlen % sizeof(int) != 0)
3042 return -EINVAL;
3043
3044 if (!table->extra1 && !table->extra2)
3045 return 0;
3046
3047 if (newlen > table->maxlen)
3048 newlen = table->maxlen;
3049 length = newlen / sizeof(int);
3050
3051 for (i = 0; i < length; i++) {
3052 int value;
3053 if (get_user(value, vec + i))
3054 return -EFAULT;
3055 if (min && value < min[i])
3056 return -EINVAL;
3057 if (max && value > max[i])
3058 return -EINVAL;
3059 }
3060 }
3061 return 0;
3062 }
3063
3064 /* Strategy function to convert jiffies to seconds */
3065 int sysctl_jiffies(struct ctl_table *table,
3066 void __user *oldval, size_t __user *oldlenp,
3067 void __user *newval, size_t newlen)
3068 {
3069 if (oldval && oldlenp) {
3070 size_t olen;
3071
3072 if (get_user(olen, oldlenp))
3073 return -EFAULT;
3074 if (olen) {
3075 int val;
3076
3077 if (olen < sizeof(int))
3078 return -EINVAL;
3079
3080 val = *(int *)(table->data) / HZ;
3081 if (put_user(val, (int __user *)oldval))
3082 return -EFAULT;
3083 if (put_user(sizeof(int), oldlenp))
3084 return -EFAULT;
3085 }
3086 }
3087 if (newval && newlen) {
3088 int new;
3089 if (newlen != sizeof(int))
3090 return -EINVAL;
3091 if (get_user(new, (int __user *)newval))
3092 return -EFAULT;
3093 *(int *)(table->data) = new*HZ;
3094 }
3095 return 1;
3096 }
3097
3098 /* Strategy function to convert jiffies to seconds */
3099 int sysctl_ms_jiffies(struct ctl_table *table,
3100 void __user *oldval, size_t __user *oldlenp,
3101 void __user *newval, size_t newlen)
3102 {
3103 if (oldval && oldlenp) {
3104 size_t olen;
3105
3106 if (get_user(olen, oldlenp))
3107 return -EFAULT;
3108 if (olen) {
3109 int val;
3110
3111 if (olen < sizeof(int))
3112 return -EINVAL;
3113
3114 val = jiffies_to_msecs(*(int *)(table->data));
3115 if (put_user(val, (int __user *)oldval))
3116 return -EFAULT;
3117 if (put_user(sizeof(int), oldlenp))
3118 return -EFAULT;
3119 }
3120 }
3121 if (newval && newlen) {
3122 int new;
3123 if (newlen != sizeof(int))
3124 return -EINVAL;
3125 if (get_user(new, (int __user *)newval))
3126 return -EFAULT;
3127 *(int *)(table->data) = msecs_to_jiffies(new);
3128 }
3129 return 1;
3130 }
3131
3132
3133
3134 #else /* CONFIG_SYSCTL_SYSCALL */
3135
3136
3137 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
3138 {
3139 struct __sysctl_args tmp;
3140 int error;
3141
3142 if (copy_from_user(&tmp, args, sizeof(tmp)))
3143 return -EFAULT;
3144
3145 error = deprecated_sysctl_warning(&tmp);
3146
3147 /* If no error reading the parameters then just -ENOSYS ... */
3148 if (!error)
3149 error = -ENOSYS;
3150
3151 return error;
3152 }
3153
3154 int sysctl_data(struct ctl_table *table,
3155 void __user *oldval, size_t __user *oldlenp,
3156 void __user *newval, size_t newlen)
3157 {
3158 return -ENOSYS;
3159 }
3160
3161 int sysctl_string(struct ctl_table *table,
3162 void __user *oldval, size_t __user *oldlenp,
3163 void __user *newval, size_t newlen)
3164 {
3165 return -ENOSYS;
3166 }
3167
3168 int sysctl_intvec(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_jiffies(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_ms_jiffies(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 #endif /* CONFIG_SYSCTL_SYSCALL */
3190
3191 static int deprecated_sysctl_warning(struct __sysctl_args *args)
3192 {
3193 static int msg_count;
3194 int name[CTL_MAXNAME];
3195 int i;
3196
3197 /* Check args->nlen. */
3198 if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
3199 return -ENOTDIR;
3200
3201 /* Read in the sysctl name for better debug message logging */
3202 for (i = 0; i < args->nlen; i++)
3203 if (get_user(name[i], args->name + i))
3204 return -EFAULT;
3205
3206 /* Ignore accesses to kernel.version */
3207 if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
3208 return 0;
3209
3210 if (msg_count < 5) {
3211 msg_count++;
3212 printk(KERN_INFO
3213 "warning: process `%s' used the deprecated sysctl "
3214 "system call with ", current->comm);
3215 for (i = 0; i < args->nlen; i++)
3216 printk("%d.", name[i]);
3217 printk("\n");
3218 }
3219 return 0;
3220 }
3221
3222 /*
3223 * No sense putting this after each symbol definition, twice,
3224 * exception granted :-)
3225 */
3226 EXPORT_SYMBOL(proc_dointvec);
3227 EXPORT_SYMBOL(proc_dointvec_jiffies);
3228 EXPORT_SYMBOL(proc_dointvec_minmax);
3229 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3230 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3231 EXPORT_SYMBOL(proc_dostring);
3232 EXPORT_SYMBOL(proc_doulongvec_minmax);
3233 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3234 EXPORT_SYMBOL(register_sysctl_table);
3235 EXPORT_SYMBOL(register_sysctl_paths);
3236 EXPORT_SYMBOL(sysctl_intvec);
3237 EXPORT_SYMBOL(sysctl_jiffies);
3238 EXPORT_SYMBOL(sysctl_ms_jiffies);
3239 EXPORT_SYMBOL(sysctl_string);
3240 EXPORT_SYMBOL(sysctl_data);
3241 EXPORT_SYMBOL(unregister_sysctl_table);
linux 2.6 git repository for openrd