projects / deliverable / linux.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
sysctl: remove "struct file *" argument of ->proc_handler
[deliverable/linux.git] / kernel / sysctl.c
1 /*
2 * sysctl.c: General linux system control interface
3 *
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12 * Horn.
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16 * Wendling.
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19 */
20
21 #include <linux/module.h>
22 #include <linux/mm.h>
23 #include <linux/swap.h>
24 #include <linux/slab.h>
25 #include <linux/sysctl.h>
26 #include <linux/proc_fs.h>
27 #include <linux/security.h>
28 #include <linux/ctype.h>
29 #include <linux/kmemcheck.h>
30 #include <linux/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 /*
1402 * NOTE: do not add new entries to this table unless you have read
1403 * Documentation/sysctl/ctl_unnumbered.txt
1404 */
1405 { .ctl_name = 0 }
1406 };
1407
1408 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1409 static struct ctl_table binfmt_misc_table[] = {
1410 { .ctl_name = 0 }
1411 };
1412 #endif
1413
1414 static struct ctl_table fs_table[] = {
1415 {
1416 .ctl_name = FS_NRINODE,
1417 .procname = "inode-nr",
1418 .data = &inodes_stat,
1419 .maxlen = 2*sizeof(int),
1420 .mode = 0444,
1421 .proc_handler = &proc_dointvec,
1422 },
1423 {
1424 .ctl_name = FS_STATINODE,
1425 .procname = "inode-state",
1426 .data = &inodes_stat,
1427 .maxlen = 7*sizeof(int),
1428 .mode = 0444,
1429 .proc_handler = &proc_dointvec,
1430 },
1431 {
1432 .procname = "file-nr",
1433 .data = &files_stat,
1434 .maxlen = 3*sizeof(int),
1435 .mode = 0444,
1436 .proc_handler = &proc_nr_files,
1437 },
1438 {
1439 .ctl_name = FS_MAXFILE,
1440 .procname = "file-max",
1441 .data = &files_stat.max_files,
1442 .maxlen = sizeof(int),
1443 .mode = 0644,
1444 .proc_handler = &proc_dointvec,
1445 },
1446 {
1447 .ctl_name = CTL_UNNUMBERED,
1448 .procname = "nr_open",
1449 .data = &sysctl_nr_open,
1450 .maxlen = sizeof(int),
1451 .mode = 0644,
1452 .proc_handler = &proc_dointvec_minmax,
1453 .extra1 = &sysctl_nr_open_min,
1454 .extra2 = &sysctl_nr_open_max,
1455 },
1456 {
1457 .ctl_name = FS_DENTRY,
1458 .procname = "dentry-state",
1459 .data = &dentry_stat,
1460 .maxlen = 6*sizeof(int),
1461 .mode = 0444,
1462 .proc_handler = &proc_dointvec,
1463 },
1464 {
1465 .ctl_name = FS_OVERFLOWUID,
1466 .procname = "overflowuid",
1467 .data = &fs_overflowuid,
1468 .maxlen = sizeof(int),
1469 .mode = 0644,
1470 .proc_handler = &proc_dointvec_minmax,
1471 .strategy = &sysctl_intvec,
1472 .extra1 = &minolduid,
1473 .extra2 = &maxolduid,
1474 },
1475 {
1476 .ctl_name = FS_OVERFLOWGID,
1477 .procname = "overflowgid",
1478 .data = &fs_overflowgid,
1479 .maxlen = sizeof(int),
1480 .mode = 0644,
1481 .proc_handler = &proc_dointvec_minmax,
1482 .strategy = &sysctl_intvec,
1483 .extra1 = &minolduid,
1484 .extra2 = &maxolduid,
1485 },
1486 #ifdef CONFIG_FILE_LOCKING
1487 {
1488 .ctl_name = FS_LEASES,
1489 .procname = "leases-enable",
1490 .data = &leases_enable,
1491 .maxlen = sizeof(int),
1492 .mode = 0644,
1493 .proc_handler = &proc_dointvec,
1494 },
1495 #endif
1496 #ifdef CONFIG_DNOTIFY
1497 {
1498 .ctl_name = FS_DIR_NOTIFY,
1499 .procname = "dir-notify-enable",
1500 .data = &dir_notify_enable,
1501 .maxlen = sizeof(int),
1502 .mode = 0644,
1503 .proc_handler = &proc_dointvec,
1504 },
1505 #endif
1506 #ifdef CONFIG_MMU
1507 #ifdef CONFIG_FILE_LOCKING
1508 {
1509 .ctl_name = FS_LEASE_TIME,
1510 .procname = "lease-break-time",
1511 .data = &lease_break_time,
1512 .maxlen = sizeof(int),
1513 .mode = 0644,
1514 .proc_handler = &proc_dointvec,
1515 },
1516 #endif
1517 #ifdef CONFIG_AIO
1518 {
1519 .procname = "aio-nr",
1520 .data = &aio_nr,
1521 .maxlen = sizeof(aio_nr),
1522 .mode = 0444,
1523 .proc_handler = &proc_doulongvec_minmax,
1524 },
1525 {
1526 .procname = "aio-max-nr",
1527 .data = &aio_max_nr,
1528 .maxlen = sizeof(aio_max_nr),
1529 .mode = 0644,
1530 .proc_handler = &proc_doulongvec_minmax,
1531 },
1532 #endif /* CONFIG_AIO */
1533 #ifdef CONFIG_INOTIFY_USER
1534 {
1535 .ctl_name = FS_INOTIFY,
1536 .procname = "inotify",
1537 .mode = 0555,
1538 .child = inotify_table,
1539 },
1540 #endif
1541 #ifdef CONFIG_EPOLL
1542 {
1543 .procname = "epoll",
1544 .mode = 0555,
1545 .child = epoll_table,
1546 },
1547 #endif
1548 #endif
1549 {
1550 .ctl_name = KERN_SETUID_DUMPABLE,
1551 .procname = "suid_dumpable",
1552 .data = &suid_dumpable,
1553 .maxlen = sizeof(int),
1554 .mode = 0644,
1555 .proc_handler = &proc_dointvec_minmax,
1556 .strategy = &sysctl_intvec,
1557 .extra1 = &zero,
1558 .extra2 = &two,
1559 },
1560 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1561 {
1562 .ctl_name = CTL_UNNUMBERED,
1563 .procname = "binfmt_misc",
1564 .mode = 0555,
1565 .child = binfmt_misc_table,
1566 },
1567 #endif
1568 /*
1569 * NOTE: do not add new entries to this table unless you have read
1570 * Documentation/sysctl/ctl_unnumbered.txt
1571 */
1572 { .ctl_name = 0 }
1573 };
1574
1575 static struct ctl_table debug_table[] = {
1576 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1577 {
1578 .ctl_name = CTL_UNNUMBERED,
1579 .procname = "exception-trace",
1580 .data = &show_unhandled_signals,
1581 .maxlen = sizeof(int),
1582 .mode = 0644,
1583 .proc_handler = proc_dointvec
1584 },
1585 #endif
1586 { .ctl_name = 0 }
1587 };
1588
1589 static struct ctl_table dev_table[] = {
1590 { .ctl_name = 0 }
1591 };
1592
1593 static DEFINE_SPINLOCK(sysctl_lock);
1594
1595 /* called under sysctl_lock */
1596 static int use_table(struct ctl_table_header *p)
1597 {
1598 if (unlikely(p->unregistering))
1599 return 0;
1600 p->used++;
1601 return 1;
1602 }
1603
1604 /* called under sysctl_lock */
1605 static void unuse_table(struct ctl_table_header *p)
1606 {
1607 if (!--p->used)
1608 if (unlikely(p->unregistering))
1609 complete(p->unregistering);
1610 }
1611
1612 /* called under sysctl_lock, will reacquire if has to wait */
1613 static void start_unregistering(struct ctl_table_header *p)
1614 {
1615 /*
1616 * if p->used is 0, nobody will ever touch that entry again;
1617 * we'll eliminate all paths to it before dropping sysctl_lock
1618 */
1619 if (unlikely(p->used)) {
1620 struct completion wait;
1621 init_completion(&wait);
1622 p->unregistering = &wait;
1623 spin_unlock(&sysctl_lock);
1624 wait_for_completion(&wait);
1625 spin_lock(&sysctl_lock);
1626 } else {
1627 /* anything non-NULL; we'll never dereference it */
1628 p->unregistering = ERR_PTR(-EINVAL);
1629 }
1630 /*
1631 * do not remove from the list until nobody holds it; walking the
1632 * list in do_sysctl() relies on that.
1633 */
1634 list_del_init(&p->ctl_entry);
1635 }
1636
1637 void sysctl_head_get(struct ctl_table_header *head)
1638 {
1639 spin_lock(&sysctl_lock);
1640 head->count++;
1641 spin_unlock(&sysctl_lock);
1642 }
1643
1644 void sysctl_head_put(struct ctl_table_header *head)
1645 {
1646 spin_lock(&sysctl_lock);
1647 if (!--head->count)
1648 kfree(head);
1649 spin_unlock(&sysctl_lock);
1650 }
1651
1652 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1653 {
1654 if (!head)
1655 BUG();
1656 spin_lock(&sysctl_lock);
1657 if (!use_table(head))
1658 head = ERR_PTR(-ENOENT);
1659 spin_unlock(&sysctl_lock);
1660 return head;
1661 }
1662
1663 void sysctl_head_finish(struct ctl_table_header *head)
1664 {
1665 if (!head)
1666 return;
1667 spin_lock(&sysctl_lock);
1668 unuse_table(head);
1669 spin_unlock(&sysctl_lock);
1670 }
1671
1672 static struct ctl_table_set *
1673 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1674 {
1675 struct ctl_table_set *set = &root->default_set;
1676 if (root->lookup)
1677 set = root->lookup(root, namespaces);
1678 return set;
1679 }
1680
1681 static struct list_head *
1682 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1683 {
1684 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1685 return &set->list;
1686 }
1687
1688 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1689 struct ctl_table_header *prev)
1690 {
1691 struct ctl_table_root *root;
1692 struct list_head *header_list;
1693 struct ctl_table_header *head;
1694 struct list_head *tmp;
1695
1696 spin_lock(&sysctl_lock);
1697 if (prev) {
1698 head = prev;
1699 tmp = &prev->ctl_entry;
1700 unuse_table(prev);
1701 goto next;
1702 }
1703 tmp = &root_table_header.ctl_entry;
1704 for (;;) {
1705 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1706
1707 if (!use_table(head))
1708 goto next;
1709 spin_unlock(&sysctl_lock);
1710 return head;
1711 next:
1712 root = head->root;
1713 tmp = tmp->next;
1714 header_list = lookup_header_list(root, namespaces);
1715 if (tmp != header_list)
1716 continue;
1717
1718 do {
1719 root = list_entry(root->root_list.next,
1720 struct ctl_table_root, root_list);
1721 if (root == &sysctl_table_root)
1722 goto out;
1723 header_list = lookup_header_list(root, namespaces);
1724 } while (list_empty(header_list));
1725 tmp = header_list->next;
1726 }
1727 out:
1728 spin_unlock(&sysctl_lock);
1729 return NULL;
1730 }
1731
1732 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1733 {
1734 return __sysctl_head_next(current->nsproxy, prev);
1735 }
1736
1737 void register_sysctl_root(struct ctl_table_root *root)
1738 {
1739 spin_lock(&sysctl_lock);
1740 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1741 spin_unlock(&sysctl_lock);
1742 }
1743
1744 #ifdef CONFIG_SYSCTL_SYSCALL
1745 /* Perform the actual read/write of a sysctl table entry. */
1746 static int do_sysctl_strategy(struct ctl_table_root *root,
1747 struct ctl_table *table,
1748 void __user *oldval, size_t __user *oldlenp,
1749 void __user *newval, size_t newlen)
1750 {
1751 int op = 0, rc;
1752
1753 if (oldval)
1754 op |= MAY_READ;
1755 if (newval)
1756 op |= MAY_WRITE;
1757 if (sysctl_perm(root, table, op))
1758 return -EPERM;
1759
1760 if (table->strategy) {
1761 rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1762 if (rc < 0)
1763 return rc;
1764 if (rc > 0)
1765 return 0;
1766 }
1767
1768 /* If there is no strategy routine, or if the strategy returns
1769 * zero, proceed with automatic r/w */
1770 if (table->data && table->maxlen) {
1771 rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1772 if (rc < 0)
1773 return rc;
1774 }
1775 return 0;
1776 }
1777
1778 static int parse_table(int __user *name, int nlen,
1779 void __user *oldval, size_t __user *oldlenp,
1780 void __user *newval, size_t newlen,
1781 struct ctl_table_root *root,
1782 struct ctl_table *table)
1783 {
1784 int n;
1785 repeat:
1786 if (!nlen)
1787 return -ENOTDIR;
1788 if (get_user(n, name))
1789 return -EFAULT;
1790 for ( ; table->ctl_name || table->procname; table++) {
1791 if (!table->ctl_name)
1792 continue;
1793 if (n == table->ctl_name) {
1794 int error;
1795 if (table->child) {
1796 if (sysctl_perm(root, table, MAY_EXEC))
1797 return -EPERM;
1798 name++;
1799 nlen--;
1800 table = table->child;
1801 goto repeat;
1802 }
1803 error = do_sysctl_strategy(root, table,
1804 oldval, oldlenp,
1805 newval, newlen);
1806 return error;
1807 }
1808 }
1809 return -ENOTDIR;
1810 }
1811
1812 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1813 void __user *newval, size_t newlen)
1814 {
1815 struct ctl_table_header *head;
1816 int error = -ENOTDIR;
1817
1818 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1819 return -ENOTDIR;
1820 if (oldval) {
1821 int old_len;
1822 if (!oldlenp || get_user(old_len, oldlenp))
1823 return -EFAULT;
1824 }
1825
1826 for (head = sysctl_head_next(NULL); head;
1827 head = sysctl_head_next(head)) {
1828 error = parse_table(name, nlen, oldval, oldlenp,
1829 newval, newlen,
1830 head->root, head->ctl_table);
1831 if (error != -ENOTDIR) {
1832 sysctl_head_finish(head);
1833 break;
1834 }
1835 }
1836 return error;
1837 }
1838
1839 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
1840 {
1841 struct __sysctl_args tmp;
1842 int error;
1843
1844 if (copy_from_user(&tmp, args, sizeof(tmp)))
1845 return -EFAULT;
1846
1847 error = deprecated_sysctl_warning(&tmp);
1848 if (error)
1849 goto out;
1850
1851 lock_kernel();
1852 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1853 tmp.newval, tmp.newlen);
1854 unlock_kernel();
1855 out:
1856 return error;
1857 }
1858 #endif /* CONFIG_SYSCTL_SYSCALL */
1859
1860 /*
1861 * sysctl_perm does NOT grant the superuser all rights automatically, because
1862 * some sysctl variables are readonly even to root.
1863 */
1864
1865 static int test_perm(int mode, int op)
1866 {
1867 if (!current_euid())
1868 mode >>= 6;
1869 else if (in_egroup_p(0))
1870 mode >>= 3;
1871 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1872 return 0;
1873 return -EACCES;
1874 }
1875
1876 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1877 {
1878 int error;
1879 int mode;
1880
1881 error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1882 if (error)
1883 return error;
1884
1885 if (root->permissions)
1886 mode = root->permissions(root, current->nsproxy, table);
1887 else
1888 mode = table->mode;
1889
1890 return test_perm(mode, op);
1891 }
1892
1893 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1894 {
1895 for (; table->ctl_name || table->procname; table++) {
1896 table->parent = parent;
1897 if (table->child)
1898 sysctl_set_parent(table, table->child);
1899 }
1900 }
1901
1902 static __init int sysctl_init(void)
1903 {
1904 sysctl_set_parent(NULL, root_table);
1905 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1906 {
1907 int err;
1908 err = sysctl_check_table(current->nsproxy, root_table);
1909 }
1910 #endif
1911 return 0;
1912 }
1913
1914 core_initcall(sysctl_init);
1915
1916 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1917 struct ctl_table *table)
1918 {
1919 struct ctl_table *p;
1920 const char *s = branch->procname;
1921
1922 /* branch should have named subdirectory as its first element */
1923 if (!s || !branch->child)
1924 return NULL;
1925
1926 /* ... and nothing else */
1927 if (branch[1].procname || branch[1].ctl_name)
1928 return NULL;
1929
1930 /* table should contain subdirectory with the same name */
1931 for (p = table; p->procname || p->ctl_name; p++) {
1932 if (!p->child)
1933 continue;
1934 if (p->procname && strcmp(p->procname, s) == 0)
1935 return p;
1936 }
1937 return NULL;
1938 }
1939
1940 /* see if attaching q to p would be an improvement */
1941 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1942 {
1943 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1944 struct ctl_table *next;
1945 int is_better = 0;
1946 int not_in_parent = !p->attached_by;
1947
1948 while ((next = is_branch_in(by, to)) != NULL) {
1949 if (by == q->attached_by)
1950 is_better = 1;
1951 if (to == p->attached_by)
1952 not_in_parent = 1;
1953 by = by->child;
1954 to = next->child;
1955 }
1956
1957 if (is_better && not_in_parent) {
1958 q->attached_by = by;
1959 q->attached_to = to;
1960 q->parent = p;
1961 }
1962 }
1963
1964 /**
1965 * __register_sysctl_paths - register a sysctl hierarchy
1966 * @root: List of sysctl headers to register on
1967 * @namespaces: Data to compute which lists of sysctl entries are visible
1968 * @path: The path to the directory the sysctl table is in.
1969 * @table: the top-level table structure
1970 *
1971 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1972 * array. A completely 0 filled entry terminates the table.
1973 *
1974 * The members of the &struct ctl_table structure are used as follows:
1975 *
1976 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1977 * must be unique within that level of sysctl
1978 *
1979 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1980 * enter a sysctl file
1981 *
1982 * data - a pointer to data for use by proc_handler
1983 *
1984 * maxlen - the maximum size in bytes of the data
1985 *
1986 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1987 *
1988 * child - a pointer to the child sysctl table if this entry is a directory, or
1989 * %NULL.
1990 *
1991 * proc_handler - the text handler routine (described below)
1992 *
1993 * strategy - the strategy routine (described below)
1994 *
1995 * de - for internal use by the sysctl routines
1996 *
1997 * extra1, extra2 - extra pointers usable by the proc handler routines
1998 *
1999 * Leaf nodes in the sysctl tree will be represented by a single file
2000 * under /proc; non-leaf nodes will be represented by directories.
2001 *
2002 * sysctl(2) can automatically manage read and write requests through
2003 * the sysctl table. The data and maxlen fields of the ctl_table
2004 * struct enable minimal validation of the values being written to be
2005 * performed, and the mode field allows minimal authentication.
2006 *
2007 * More sophisticated management can be enabled by the provision of a
2008 * strategy routine with the table entry. This will be called before
2009 * any automatic read or write of the data is performed.
2010 *
2011 * The strategy routine may return
2012 *
2013 * < 0 - Error occurred (error is passed to user process)
2014 *
2015 * 0 - OK - proceed with automatic read or write.
2016 *
2017 * > 0 - OK - read or write has been done by the strategy routine, so
2018 * return immediately.
2019 *
2020 * There must be a proc_handler routine for any terminal nodes
2021 * mirrored under /proc/sys (non-terminals are handled by a built-in
2022 * directory handler). Several default handlers are available to
2023 * cover common cases -
2024 *
2025 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
2026 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
2027 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
2028 *
2029 * It is the handler's job to read the input buffer from user memory
2030 * and process it. The handler should return 0 on success.
2031 *
2032 * This routine returns %NULL on a failure to register, and a pointer
2033 * to the table header on success.
2034 */
2035 struct ctl_table_header *__register_sysctl_paths(
2036 struct ctl_table_root *root,
2037 struct nsproxy *namespaces,
2038 const struct ctl_path *path, struct ctl_table *table)
2039 {
2040 struct ctl_table_header *header;
2041 struct ctl_table *new, **prevp;
2042 unsigned int n, npath;
2043 struct ctl_table_set *set;
2044
2045 /* Count the path components */
2046 for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
2047 ;
2048
2049 /*
2050 * For each path component, allocate a 2-element ctl_table array.
2051 * The first array element will be filled with the sysctl entry
2052 * for this, the second will be the sentinel (ctl_name == 0).
2053 *
2054 * We allocate everything in one go so that we don't have to
2055 * worry about freeing additional memory in unregister_sysctl_table.
2056 */
2057 header = kzalloc(sizeof(struct ctl_table_header) +
2058 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
2059 if (!header)
2060 return NULL;
2061
2062 new = (struct ctl_table *) (header + 1);
2063
2064 /* Now connect the dots */
2065 prevp = &header->ctl_table;
2066 for (n = 0; n < npath; ++n, ++path) {
2067 /* Copy the procname */
2068 new->procname = path->procname;
2069 new->ctl_name = path->ctl_name;
2070 new->mode = 0555;
2071
2072 *prevp = new;
2073 prevp = &new->child;
2074
2075 new += 2;
2076 }
2077 *prevp = table;
2078 header->ctl_table_arg = table;
2079
2080 INIT_LIST_HEAD(&header->ctl_entry);
2081 header->used = 0;
2082 header->unregistering = NULL;
2083 header->root = root;
2084 sysctl_set_parent(NULL, header->ctl_table);
2085 header->count = 1;
2086 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
2087 if (sysctl_check_table(namespaces, header->ctl_table)) {
2088 kfree(header);
2089 return NULL;
2090 }
2091 #endif
2092 spin_lock(&sysctl_lock);
2093 header->set = lookup_header_set(root, namespaces);
2094 header->attached_by = header->ctl_table;
2095 header->attached_to = root_table;
2096 header->parent = &root_table_header;
2097 for (set = header->set; set; set = set->parent) {
2098 struct ctl_table_header *p;
2099 list_for_each_entry(p, &set->list, ctl_entry) {
2100 if (p->unregistering)
2101 continue;
2102 try_attach(p, header);
2103 }
2104 }
2105 header->parent->count++;
2106 list_add_tail(&header->ctl_entry, &header->set->list);
2107 spin_unlock(&sysctl_lock);
2108
2109 return header;
2110 }
2111
2112 /**
2113 * register_sysctl_table_path - register a sysctl table hierarchy
2114 * @path: The path to the directory the sysctl table is in.
2115 * @table: the top-level table structure
2116 *
2117 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2118 * array. A completely 0 filled entry terminates the table.
2119 *
2120 * See __register_sysctl_paths for more details.
2121 */
2122 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2123 struct ctl_table *table)
2124 {
2125 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
2126 path, table);
2127 }
2128
2129 /**
2130 * register_sysctl_table - register a sysctl table hierarchy
2131 * @table: the top-level table structure
2132 *
2133 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2134 * array. A completely 0 filled entry terminates the table.
2135 *
2136 * See register_sysctl_paths for more details.
2137 */
2138 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
2139 {
2140 static const struct ctl_path null_path[] = { {} };
2141
2142 return register_sysctl_paths(null_path, table);
2143 }
2144
2145 /**
2146 * unregister_sysctl_table - unregister a sysctl table hierarchy
2147 * @header: the header returned from register_sysctl_table
2148 *
2149 * Unregisters the sysctl table and all children. proc entries may not
2150 * actually be removed until they are no longer used by anyone.
2151 */
2152 void unregister_sysctl_table(struct ctl_table_header * header)
2153 {
2154 might_sleep();
2155
2156 if (header == NULL)
2157 return;
2158
2159 spin_lock(&sysctl_lock);
2160 start_unregistering(header);
2161 if (!--header->parent->count) {
2162 WARN_ON(1);
2163 kfree(header->parent);
2164 }
2165 if (!--header->count)
2166 kfree(header);
2167 spin_unlock(&sysctl_lock);
2168 }
2169
2170 int sysctl_is_seen(struct ctl_table_header *p)
2171 {
2172 struct ctl_table_set *set = p->set;
2173 int res;
2174 spin_lock(&sysctl_lock);
2175 if (p->unregistering)
2176 res = 0;
2177 else if (!set->is_seen)
2178 res = 1;
2179 else
2180 res = set->is_seen(set);
2181 spin_unlock(&sysctl_lock);
2182 return res;
2183 }
2184
2185 void setup_sysctl_set(struct ctl_table_set *p,
2186 struct ctl_table_set *parent,
2187 int (*is_seen)(struct ctl_table_set *))
2188 {
2189 INIT_LIST_HEAD(&p->list);
2190 p->parent = parent ? parent : &sysctl_table_root.default_set;
2191 p->is_seen = is_seen;
2192 }
2193
2194 #else /* !CONFIG_SYSCTL */
2195 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2196 {
2197 return NULL;
2198 }
2199
2200 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2201 struct ctl_table *table)
2202 {
2203 return NULL;
2204 }
2205
2206 void unregister_sysctl_table(struct ctl_table_header * table)
2207 {
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 }
2215
2216 void sysctl_head_put(struct ctl_table_header *head)
2217 {
2218 }
2219
2220 #endif /* CONFIG_SYSCTL */
2221
2222 /*
2223 * /proc/sys support
2224 */
2225
2226 #ifdef CONFIG_PROC_SYSCTL
2227
2228 static int _proc_do_string(void* data, int maxlen, int write,
2229 void __user *buffer,
2230 size_t *lenp, loff_t *ppos)
2231 {
2232 size_t len;
2233 char __user *p;
2234 char c;
2235
2236 if (!data || !maxlen || !*lenp) {
2237 *lenp = 0;
2238 return 0;
2239 }
2240
2241 if (write) {
2242 len = 0;
2243 p = buffer;
2244 while (len < *lenp) {
2245 if (get_user(c, p++))
2246 return -EFAULT;
2247 if (c == 0 || c == '\n')
2248 break;
2249 len++;
2250 }
2251 if (len >= maxlen)
2252 len = maxlen-1;
2253 if(copy_from_user(data, buffer, len))
2254 return -EFAULT;
2255 ((char *) data)[len] = 0;
2256 *ppos += *lenp;
2257 } else {
2258 len = strlen(data);
2259 if (len > maxlen)
2260 len = maxlen;
2261
2262 if (*ppos > len) {
2263 *lenp = 0;
2264 return 0;
2265 }
2266
2267 data += *ppos;
2268 len -= *ppos;
2269
2270 if (len > *lenp)
2271 len = *lenp;
2272 if (len)
2273 if(copy_to_user(buffer, data, len))
2274 return -EFAULT;
2275 if (len < *lenp) {
2276 if(put_user('\n', ((char __user *) buffer) + len))
2277 return -EFAULT;
2278 len++;
2279 }
2280 *lenp = len;
2281 *ppos += len;
2282 }
2283 return 0;
2284 }
2285
2286 /**
2287 * proc_dostring - read a string sysctl
2288 * @table: the sysctl table
2289 * @write: %TRUE if this is a write to the sysctl file
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,
2304 void __user *buffer, size_t *lenp, loff_t *ppos)
2305 {
2306 return _proc_do_string(table->data, table->maxlen, write,
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, 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,
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,
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 * @buffer: the user buffer
2453 * @lenp: the size of the user buffer
2454 * @ppos: file position
2455 *
2456 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2457 * values from/to the user buffer, treated as an ASCII string.
2458 *
2459 * Returns 0 on success.
2460 */
2461 int proc_dointvec(struct ctl_table *table, int write,
2462 void __user *buffer, size_t *lenp, loff_t *ppos)
2463 {
2464 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2465 NULL,NULL);
2466 }
2467
2468 /*
2469 * Taint values can only be increased
2470 * This means we can safely use a temporary.
2471 */
2472 static int proc_taint(struct ctl_table *table, int write,
2473 void __user *buffer, size_t *lenp, loff_t *ppos)
2474 {
2475 struct ctl_table t;
2476 unsigned long tmptaint = get_taint();
2477 int err;
2478
2479 if (write && !capable(CAP_SYS_ADMIN))
2480 return -EPERM;
2481
2482 t = *table;
2483 t.data = &tmptaint;
2484 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2485 if (err < 0)
2486 return err;
2487
2488 if (write) {
2489 /*
2490 * Poor man's atomic or. Not worth adding a primitive
2491 * to everyone's atomic.h for this
2492 */
2493 int i;
2494 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2495 if ((tmptaint >> i) & 1)
2496 add_taint(i);
2497 }
2498 }
2499
2500 return err;
2501 }
2502
2503 struct do_proc_dointvec_minmax_conv_param {
2504 int *min;
2505 int *max;
2506 };
2507
2508 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2509 int *valp,
2510 int write, void *data)
2511 {
2512 struct do_proc_dointvec_minmax_conv_param *param = data;
2513 if (write) {
2514 int val = *negp ? -*lvalp : *lvalp;
2515 if ((param->min && *param->min > val) ||
2516 (param->max && *param->max < val))
2517 return -EINVAL;
2518 *valp = val;
2519 } else {
2520 int val = *valp;
2521 if (val < 0) {
2522 *negp = -1;
2523 *lvalp = (unsigned long)-val;
2524 } else {
2525 *negp = 0;
2526 *lvalp = (unsigned long)val;
2527 }
2528 }
2529 return 0;
2530 }
2531
2532 /**
2533 * proc_dointvec_minmax - read a vector of integers with min/max values
2534 * @table: the sysctl table
2535 * @write: %TRUE if this is a write to the sysctl file
2536 * @buffer: the user buffer
2537 * @lenp: the size of the user buffer
2538 * @ppos: file position
2539 *
2540 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2541 * values from/to the user buffer, treated as an ASCII string.
2542 *
2543 * This routine will ensure the values are within the range specified by
2544 * table->extra1 (min) and table->extra2 (max).
2545 *
2546 * Returns 0 on success.
2547 */
2548 int proc_dointvec_minmax(struct ctl_table *table, int write,
2549 void __user *buffer, size_t *lenp, loff_t *ppos)
2550 {
2551 struct do_proc_dointvec_minmax_conv_param param = {
2552 .min = (int *) table->extra1,
2553 .max = (int *) table->extra2,
2554 };
2555 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2556 do_proc_dointvec_minmax_conv, &param);
2557 }
2558
2559 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2560 void __user *buffer,
2561 size_t *lenp, loff_t *ppos,
2562 unsigned long convmul,
2563 unsigned long convdiv)
2564 {
2565 #define TMPBUFLEN 21
2566 unsigned long *i, *min, *max, val;
2567 int vleft, first=1, neg;
2568 size_t len, left;
2569 char buf[TMPBUFLEN], *p;
2570 char __user *s = buffer;
2571
2572 if (!data || !table->maxlen || !*lenp ||
2573 (*ppos && !write)) {
2574 *lenp = 0;
2575 return 0;
2576 }
2577
2578 i = (unsigned long *) data;
2579 min = (unsigned long *) table->extra1;
2580 max = (unsigned long *) table->extra2;
2581 vleft = table->maxlen / sizeof(unsigned long);
2582 left = *lenp;
2583
2584 for (; left && vleft--; i++, min++, max++, first=0) {
2585 if (write) {
2586 while (left) {
2587 char c;
2588 if (get_user(c, s))
2589 return -EFAULT;
2590 if (!isspace(c))
2591 break;
2592 left--;
2593 s++;
2594 }
2595 if (!left)
2596 break;
2597 neg = 0;
2598 len = left;
2599 if (len > TMPBUFLEN-1)
2600 len = TMPBUFLEN-1;
2601 if (copy_from_user(buf, s, len))
2602 return -EFAULT;
2603 buf[len] = 0;
2604 p = buf;
2605 if (*p == '-' && left > 1) {
2606 neg = 1;
2607 p++;
2608 }
2609 if (*p < '0' || *p > '9')
2610 break;
2611 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2612 len = p-buf;
2613 if ((len < left) && *p && !isspace(*p))
2614 break;
2615 if (neg)
2616 val = -val;
2617 s += len;
2618 left -= len;
2619
2620 if(neg)
2621 continue;
2622 if ((min && val < *min) || (max && val > *max))
2623 continue;
2624 *i = val;
2625 } else {
2626 p = buf;
2627 if (!first)
2628 *p++ = '\t';
2629 sprintf(p, "%lu", convdiv * (*i) / convmul);
2630 len = strlen(buf);
2631 if (len > left)
2632 len = left;
2633 if(copy_to_user(s, buf, len))
2634 return -EFAULT;
2635 left -= len;
2636 s += len;
2637 }
2638 }
2639
2640 if (!write && !first && left) {
2641 if(put_user('\n', s))
2642 return -EFAULT;
2643 left--, s++;
2644 }
2645 if (write) {
2646 while (left) {
2647 char c;
2648 if (get_user(c, s++))
2649 return -EFAULT;
2650 if (!isspace(c))
2651 break;
2652 left--;
2653 }
2654 }
2655 if (write && first)
2656 return -EINVAL;
2657 *lenp -= left;
2658 *ppos += *lenp;
2659 return 0;
2660 #undef TMPBUFLEN
2661 }
2662
2663 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2664 void __user *buffer,
2665 size_t *lenp, loff_t *ppos,
2666 unsigned long convmul,
2667 unsigned long convdiv)
2668 {
2669 return __do_proc_doulongvec_minmax(table->data, table, write,
2670 buffer, lenp, ppos, convmul, convdiv);
2671 }
2672
2673 /**
2674 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2675 * @table: the sysctl table
2676 * @write: %TRUE if this is a write to the sysctl file
2677 * @buffer: the user buffer
2678 * @lenp: the size of the user buffer
2679 * @ppos: file position
2680 *
2681 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2682 * values from/to the user buffer, treated as an ASCII string.
2683 *
2684 * This routine will ensure the values are within the range specified by
2685 * table->extra1 (min) and table->extra2 (max).
2686 *
2687 * Returns 0 on success.
2688 */
2689 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2690 void __user *buffer, size_t *lenp, loff_t *ppos)
2691 {
2692 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2693 }
2694
2695 /**
2696 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2697 * @table: the sysctl table
2698 * @write: %TRUE if this is a write to the sysctl file
2699 * @buffer: the user buffer
2700 * @lenp: the size of the user buffer
2701 * @ppos: file position
2702 *
2703 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2704 * values from/to the user buffer, treated as an ASCII string. The values
2705 * are treated as milliseconds, and converted to jiffies when they are stored.
2706 *
2707 * This routine will ensure the values are within the range specified by
2708 * table->extra1 (min) and table->extra2 (max).
2709 *
2710 * Returns 0 on success.
2711 */
2712 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2713 void __user *buffer,
2714 size_t *lenp, loff_t *ppos)
2715 {
2716 return do_proc_doulongvec_minmax(table, write, buffer,
2717 lenp, ppos, HZ, 1000l);
2718 }
2719
2720
2721 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2722 int *valp,
2723 int write, void *data)
2724 {
2725 if (write) {
2726 if (*lvalp > LONG_MAX / HZ)
2727 return 1;
2728 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2729 } else {
2730 int val = *valp;
2731 unsigned long lval;
2732 if (val < 0) {
2733 *negp = -1;
2734 lval = (unsigned long)-val;
2735 } else {
2736 *negp = 0;
2737 lval = (unsigned long)val;
2738 }
2739 *lvalp = lval / HZ;
2740 }
2741 return 0;
2742 }
2743
2744 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2745 int *valp,
2746 int write, void *data)
2747 {
2748 if (write) {
2749 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2750 return 1;
2751 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2752 } else {
2753 int val = *valp;
2754 unsigned long lval;
2755 if (val < 0) {
2756 *negp = -1;
2757 lval = (unsigned long)-val;
2758 } else {
2759 *negp = 0;
2760 lval = (unsigned long)val;
2761 }
2762 *lvalp = jiffies_to_clock_t(lval);
2763 }
2764 return 0;
2765 }
2766
2767 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2768 int *valp,
2769 int write, void *data)
2770 {
2771 if (write) {
2772 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2773 } else {
2774 int val = *valp;
2775 unsigned long lval;
2776 if (val < 0) {
2777 *negp = -1;
2778 lval = (unsigned long)-val;
2779 } else {
2780 *negp = 0;
2781 lval = (unsigned long)val;
2782 }
2783 *lvalp = jiffies_to_msecs(lval);
2784 }
2785 return 0;
2786 }
2787
2788 /**
2789 * proc_dointvec_jiffies - read a vector of integers as seconds
2790 * @table: the sysctl table
2791 * @write: %TRUE if this is a write to the sysctl file
2792 * @buffer: the user buffer
2793 * @lenp: the size of the user buffer
2794 * @ppos: file position
2795 *
2796 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2797 * values from/to the user buffer, treated as an ASCII string.
2798 * The values read are assumed to be in seconds, and are converted into
2799 * jiffies.
2800 *
2801 * Returns 0 on success.
2802 */
2803 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2804 void __user *buffer, size_t *lenp, loff_t *ppos)
2805 {
2806 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2807 do_proc_dointvec_jiffies_conv,NULL);
2808 }
2809
2810 /**
2811 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2812 * @table: the sysctl table
2813 * @write: %TRUE if this is a write to the sysctl file
2814 * @buffer: the user buffer
2815 * @lenp: the size of the user buffer
2816 * @ppos: pointer to the file position
2817 *
2818 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2819 * values from/to the user buffer, treated as an ASCII string.
2820 * The values read are assumed to be in 1/USER_HZ seconds, and
2821 * are converted into jiffies.
2822 *
2823 * Returns 0 on success.
2824 */
2825 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2826 void __user *buffer, size_t *lenp, loff_t *ppos)
2827 {
2828 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2829 do_proc_dointvec_userhz_jiffies_conv,NULL);
2830 }
2831
2832 /**
2833 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2834 * @table: the sysctl table
2835 * @write: %TRUE if this is a write to the sysctl file
2836 * @buffer: the user buffer
2837 * @lenp: the size of the user buffer
2838 * @ppos: file position
2839 * @ppos: the current position in the file
2840 *
2841 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2842 * values from/to the user buffer, treated as an ASCII string.
2843 * The values read are assumed to be in 1/1000 seconds, and
2844 * are converted into jiffies.
2845 *
2846 * Returns 0 on success.
2847 */
2848 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2849 void __user *buffer, size_t *lenp, loff_t *ppos)
2850 {
2851 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2852 do_proc_dointvec_ms_jiffies_conv, NULL);
2853 }
2854
2855 static int proc_do_cad_pid(struct ctl_table *table, int write,
2856 void __user *buffer, size_t *lenp, loff_t *ppos)
2857 {
2858 struct pid *new_pid;
2859 pid_t tmp;
2860 int r;
2861
2862 tmp = pid_vnr(cad_pid);
2863
2864 r = __do_proc_dointvec(&tmp, table, write, buffer,
2865 lenp, ppos, NULL, NULL);
2866 if (r || !write)
2867 return r;
2868
2869 new_pid = find_get_pid(tmp);
2870 if (!new_pid)
2871 return -ESRCH;
2872
2873 put_pid(xchg(&cad_pid, new_pid));
2874 return 0;
2875 }
2876
2877 #else /* CONFIG_PROC_FS */
2878
2879 int proc_dostring(struct ctl_table *table, int write,
2880 void __user *buffer, size_t *lenp, loff_t *ppos)
2881 {
2882 return -ENOSYS;
2883 }
2884
2885 int proc_dointvec(struct ctl_table *table, int write,
2886 void __user *buffer, size_t *lenp, loff_t *ppos)
2887 {
2888 return -ENOSYS;
2889 }
2890
2891 int proc_dointvec_minmax(struct ctl_table *table, int write,
2892 void __user *buffer, size_t *lenp, loff_t *ppos)
2893 {
2894 return -ENOSYS;
2895 }
2896
2897 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2898 void __user *buffer, size_t *lenp, loff_t *ppos)
2899 {
2900 return -ENOSYS;
2901 }
2902
2903 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2904 void __user *buffer, size_t *lenp, loff_t *ppos)
2905 {
2906 return -ENOSYS;
2907 }
2908
2909 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2910 void __user *buffer, size_t *lenp, loff_t *ppos)
2911 {
2912 return -ENOSYS;
2913 }
2914
2915 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2916 void __user *buffer, size_t *lenp, loff_t *ppos)
2917 {
2918 return -ENOSYS;
2919 }
2920
2921 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2922 void __user *buffer,
2923 size_t *lenp, loff_t *ppos)
2924 {
2925 return -ENOSYS;
2926 }
2927
2928
2929 #endif /* CONFIG_PROC_FS */
2930
2931
2932 #ifdef CONFIG_SYSCTL_SYSCALL
2933 /*
2934 * General sysctl support routines
2935 */
2936
2937 /* The generic sysctl data routine (used if no strategy routine supplied) */
2938 int sysctl_data(struct ctl_table *table,
2939 void __user *oldval, size_t __user *oldlenp,
2940 void __user *newval, size_t newlen)
2941 {
2942 size_t len;
2943
2944 /* Get out of I don't have a variable */
2945 if (!table->data || !table->maxlen)
2946 return -ENOTDIR;
2947
2948 if (oldval && oldlenp) {
2949 if (get_user(len, oldlenp))
2950 return -EFAULT;
2951 if (len) {
2952 if (len > table->maxlen)
2953 len = table->maxlen;
2954 if (copy_to_user(oldval, table->data, len))
2955 return -EFAULT;
2956 if (put_user(len, oldlenp))
2957 return -EFAULT;
2958 }
2959 }
2960
2961 if (newval && newlen) {
2962 if (newlen > table->maxlen)
2963 newlen = table->maxlen;
2964
2965 if (copy_from_user(table->data, newval, newlen))
2966 return -EFAULT;
2967 }
2968 return 1;
2969 }
2970
2971 /* The generic string strategy routine: */
2972 int sysctl_string(struct ctl_table *table,
2973 void __user *oldval, size_t __user *oldlenp,
2974 void __user *newval, size_t newlen)
2975 {
2976 if (!table->data || !table->maxlen)
2977 return -ENOTDIR;
2978
2979 if (oldval && oldlenp) {
2980 size_t bufsize;
2981 if (get_user(bufsize, oldlenp))
2982 return -EFAULT;
2983 if (bufsize) {
2984 size_t len = strlen(table->data), copied;
2985
2986 /* This shouldn't trigger for a well-formed sysctl */
2987 if (len > table->maxlen)
2988 len = table->maxlen;
2989
2990 /* Copy up to a max of bufsize-1 bytes of the string */
2991 copied = (len >= bufsize) ? bufsize - 1 : len;
2992
2993 if (copy_to_user(oldval, table->data, copied) ||
2994 put_user(0, (char __user *)(oldval + copied)))
2995 return -EFAULT;
2996 if (put_user(len, oldlenp))
2997 return -EFAULT;
2998 }
2999 }
3000 if (newval && newlen) {
3001 size_t len = newlen;
3002 if (len > table->maxlen)
3003 len = table->maxlen;
3004 if(copy_from_user(table->data, newval, len))
3005 return -EFAULT;
3006 if (len == table->maxlen)
3007 len--;
3008 ((char *) table->data)[len] = 0;
3009 }
3010 return 1;
3011 }
3012
3013 /*
3014 * This function makes sure that all of the integers in the vector
3015 * are between the minimum and maximum values given in the arrays
3016 * table->extra1 and table->extra2, respectively.
3017 */
3018 int sysctl_intvec(struct ctl_table *table,
3019 void __user *oldval, size_t __user *oldlenp,
3020 void __user *newval, size_t newlen)
3021 {
3022
3023 if (newval && newlen) {
3024 int __user *vec = (int __user *) newval;
3025 int *min = (int *) table->extra1;
3026 int *max = (int *) table->extra2;
3027 size_t length;
3028 int i;
3029
3030 if (newlen % sizeof(int) != 0)
3031 return -EINVAL;
3032
3033 if (!table->extra1 && !table->extra2)
3034 return 0;
3035
3036 if (newlen > table->maxlen)
3037 newlen = table->maxlen;
3038 length = newlen / sizeof(int);
3039
3040 for (i = 0; i < length; i++) {
3041 int value;
3042 if (get_user(value, vec + i))
3043 return -EFAULT;
3044 if (min && value < min[i])
3045 return -EINVAL;
3046 if (max && value > max[i])
3047 return -EINVAL;
3048 }
3049 }
3050 return 0;
3051 }
3052
3053 /* Strategy function to convert jiffies to seconds */
3054 int sysctl_jiffies(struct ctl_table *table,
3055 void __user *oldval, size_t __user *oldlenp,
3056 void __user *newval, size_t newlen)
3057 {
3058 if (oldval && oldlenp) {
3059 size_t olen;
3060
3061 if (get_user(olen, oldlenp))
3062 return -EFAULT;
3063 if (olen) {
3064 int val;
3065
3066 if (olen < sizeof(int))
3067 return -EINVAL;
3068
3069 val = *(int *)(table->data) / HZ;
3070 if (put_user(val, (int __user *)oldval))
3071 return -EFAULT;
3072 if (put_user(sizeof(int), oldlenp))
3073 return -EFAULT;
3074 }
3075 }
3076 if (newval && newlen) {
3077 int new;
3078 if (newlen != sizeof(int))
3079 return -EINVAL;
3080 if (get_user(new, (int __user *)newval))
3081 return -EFAULT;
3082 *(int *)(table->data) = new*HZ;
3083 }
3084 return 1;
3085 }
3086
3087 /* Strategy function to convert jiffies to seconds */
3088 int sysctl_ms_jiffies(struct ctl_table *table,
3089 void __user *oldval, size_t __user *oldlenp,
3090 void __user *newval, size_t newlen)
3091 {
3092 if (oldval && oldlenp) {
3093 size_t olen;
3094
3095 if (get_user(olen, oldlenp))
3096 return -EFAULT;
3097 if (olen) {
3098 int val;
3099
3100 if (olen < sizeof(int))
3101 return -EINVAL;
3102
3103 val = jiffies_to_msecs(*(int *)(table->data));
3104 if (put_user(val, (int __user *)oldval))
3105 return -EFAULT;
3106 if (put_user(sizeof(int), oldlenp))
3107 return -EFAULT;
3108 }
3109 }
3110 if (newval && newlen) {
3111 int new;
3112 if (newlen != sizeof(int))
3113 return -EINVAL;
3114 if (get_user(new, (int __user *)newval))
3115 return -EFAULT;
3116 *(int *)(table->data) = msecs_to_jiffies(new);
3117 }
3118 return 1;
3119 }
3120
3121
3122
3123 #else /* CONFIG_SYSCTL_SYSCALL */
3124
3125
3126 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
3127 {
3128 struct __sysctl_args tmp;
3129 int error;
3130
3131 if (copy_from_user(&tmp, args, sizeof(tmp)))
3132 return -EFAULT;
3133
3134 error = deprecated_sysctl_warning(&tmp);
3135
3136 /* If no error reading the parameters then just -ENOSYS ... */
3137 if (!error)
3138 error = -ENOSYS;
3139
3140 return error;
3141 }
3142
3143 int sysctl_data(struct ctl_table *table,
3144 void __user *oldval, size_t __user *oldlenp,
3145 void __user *newval, size_t newlen)
3146 {
3147 return -ENOSYS;
3148 }
3149
3150 int sysctl_string(struct ctl_table *table,
3151 void __user *oldval, size_t __user *oldlenp,
3152 void __user *newval, size_t newlen)
3153 {
3154 return -ENOSYS;
3155 }
3156
3157 int sysctl_intvec(struct ctl_table *table,
3158 void __user *oldval, size_t __user *oldlenp,
3159 void __user *newval, size_t newlen)
3160 {
3161 return -ENOSYS;
3162 }
3163
3164 int sysctl_jiffies(struct ctl_table *table,
3165 void __user *oldval, size_t __user *oldlenp,
3166 void __user *newval, size_t newlen)
3167 {
3168 return -ENOSYS;
3169 }
3170
3171 int sysctl_ms_jiffies(struct ctl_table *table,
3172 void __user *oldval, size_t __user *oldlenp,
3173 void __user *newval, size_t newlen)
3174 {
3175 return -ENOSYS;
3176 }
3177
3178 #endif /* CONFIG_SYSCTL_SYSCALL */
3179
3180 static int deprecated_sysctl_warning(struct __sysctl_args *args)
3181 {
3182 static int msg_count;
3183 int name[CTL_MAXNAME];
3184 int i;
3185
3186 /* Check args->nlen. */
3187 if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
3188 return -ENOTDIR;
3189
3190 /* Read in the sysctl name for better debug message logging */
3191 for (i = 0; i < args->nlen; i++)
3192 if (get_user(name[i], args->name + i))
3193 return -EFAULT;
3194
3195 /* Ignore accesses to kernel.version */
3196 if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
3197 return 0;
3198
3199 if (msg_count < 5) {
3200 msg_count++;
3201 printk(KERN_INFO
3202 "warning: process `%s' used the deprecated sysctl "
3203 "system call with ", current->comm);
3204 for (i = 0; i < args->nlen; i++)
3205 printk("%d.", name[i]);
3206 printk("\n");
3207 }
3208 return 0;
3209 }
3210
3211 /*
3212 * No sense putting this after each symbol definition, twice,
3213 * exception granted :-)
3214 */
3215 EXPORT_SYMBOL(proc_dointvec);
3216 EXPORT_SYMBOL(proc_dointvec_jiffies);
3217 EXPORT_SYMBOL(proc_dointvec_minmax);
3218 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3219 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3220 EXPORT_SYMBOL(proc_dostring);
3221 EXPORT_SYMBOL(proc_doulongvec_minmax);
3222 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3223 EXPORT_SYMBOL(register_sysctl_table);
3224 EXPORT_SYMBOL(register_sysctl_paths);
3225 EXPORT_SYMBOL(sysctl_intvec);
3226 EXPORT_SYMBOL(sysctl_jiffies);
3227 EXPORT_SYMBOL(sysctl_ms_jiffies);
3228 EXPORT_SYMBOL(sysctl_string);
3229 EXPORT_SYMBOL(sysctl_data);
3230 EXPORT_SYMBOL(unregister_sysctl_table);
Linux kernel - Deliverables
This page took 0.096796 seconds and 5 git commands to generate.