]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/blob - kernel/sysctl.c
projects / mirror_ubuntu-bionic-kernel.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
perf_counter: More paranoia settings
[mirror_ubuntu-bionic-kernel.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/utsname.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_counter.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 int pid_max;
80 extern int min_free_kbytes;
81 extern int pid_max_min, pid_max_max;
82 extern int sysctl_drop_caches;
83 extern int percpu_pagelist_fraction;
84 extern int compat_log;
85 extern int latencytop_enabled;
86 extern int sysctl_nr_open_min, sysctl_nr_open_max;
87 #ifndef CONFIG_MMU
88 extern int sysctl_nr_trim_pages;
89 #endif
90 #ifdef CONFIG_RCU_TORTURE_TEST
91 extern int rcutorture_runnable;
92 #endif /* #ifdef CONFIG_RCU_TORTURE_TEST */
93
94 /* Constants used for minimum and maximum */
95 #ifdef CONFIG_DETECT_SOFTLOCKUP
96 static int sixty = 60;
97 static int neg_one = -1;
98 #endif
99
100 static int zero;
101 static int __maybe_unused one = 1;
102 static int __maybe_unused two = 2;
103 static unsigned long one_ul = 1;
104 static int one_hundred = 100;
105
106 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
107 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
108
109 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
110 static int maxolduid = 65535;
111 static int minolduid;
112 static int min_percpu_pagelist_fract = 8;
113
114 static int ngroups_max = NGROUPS_MAX;
115
116 #ifdef CONFIG_MODULES
117 extern char modprobe_path[];
118 #endif
119 #ifdef CONFIG_CHR_DEV_SG
120 extern int sg_big_buff;
121 #endif
122
123 #ifdef CONFIG_SPARC
124 #include <asm/system.h>
125 #endif
126
127 #ifdef CONFIG_SPARC64
128 extern int sysctl_tsb_ratio;
129 #endif
130
131 #ifdef __hppa__
132 extern int pwrsw_enabled;
133 extern int unaligned_enabled;
134 #endif
135
136 #ifdef CONFIG_S390
137 #ifdef CONFIG_MATHEMU
138 extern int sysctl_ieee_emulation_warnings;
139 #endif
140 extern int sysctl_userprocess_debug;
141 extern int spin_retry;
142 #endif
143
144 #ifdef CONFIG_BSD_PROCESS_ACCT
145 extern int acct_parm[];
146 #endif
147
148 #ifdef CONFIG_IA64
149 extern int no_unaligned_warning;
150 extern int unaligned_dump_stack;
151 #endif
152
153 #ifdef CONFIG_RT_MUTEXES
154 extern int max_lock_depth;
155 #endif
156
157 #ifdef CONFIG_PROC_SYSCTL
158 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
159 void __user *buffer, size_t *lenp, loff_t *ppos);
160 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
161 void __user *buffer, size_t *lenp, loff_t *ppos);
162 #endif
163
164 static struct ctl_table root_table[];
165 static struct ctl_table_root sysctl_table_root;
166 static struct ctl_table_header root_table_header = {
167 .count = 1,
168 .ctl_table = root_table,
169 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
170 .root = &sysctl_table_root,
171 .set = &sysctl_table_root.default_set,
172 };
173 static struct ctl_table_root sysctl_table_root = {
174 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
175 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
176 };
177
178 static struct ctl_table kern_table[];
179 static struct ctl_table vm_table[];
180 static struct ctl_table fs_table[];
181 static struct ctl_table debug_table[];
182 static struct ctl_table dev_table[];
183 extern struct ctl_table random_table[];
184 #ifdef CONFIG_INOTIFY_USER
185 extern struct ctl_table inotify_table[];
186 #endif
187 #ifdef CONFIG_EPOLL
188 extern struct ctl_table epoll_table[];
189 #endif
190
191 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
192 int sysctl_legacy_va_layout;
193 #endif
194
195 extern int prove_locking;
196 extern int lock_stat;
197
198 /* The default sysctl tables: */
199
200 static struct ctl_table root_table[] = {
201 {
202 .ctl_name = CTL_KERN,
203 .procname = "kernel",
204 .mode = 0555,
205 .child = kern_table,
206 },
207 {
208 .ctl_name = CTL_VM,
209 .procname = "vm",
210 .mode = 0555,
211 .child = vm_table,
212 },
213 {
214 .ctl_name = CTL_FS,
215 .procname = "fs",
216 .mode = 0555,
217 .child = fs_table,
218 },
219 {
220 .ctl_name = CTL_DEBUG,
221 .procname = "debug",
222 .mode = 0555,
223 .child = debug_table,
224 },
225 {
226 .ctl_name = CTL_DEV,
227 .procname = "dev",
228 .mode = 0555,
229 .child = dev_table,
230 },
231 /*
232 * NOTE: do not add new entries to this table unless you have read
233 * Documentation/sysctl/ctl_unnumbered.txt
234 */
235 { .ctl_name = 0 }
236 };
237
238 #ifdef CONFIG_SCHED_DEBUG
239 static int min_sched_granularity_ns = 100000; /* 100 usecs */
240 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
241 static int min_wakeup_granularity_ns; /* 0 usecs */
242 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
243 #endif
244
245 static struct ctl_table kern_table[] = {
246 #ifdef CONFIG_SCHED_DEBUG
247 {
248 .ctl_name = CTL_UNNUMBERED,
249 .procname = "sched_min_granularity_ns",
250 .data = &sysctl_sched_min_granularity,
251 .maxlen = sizeof(unsigned int),
252 .mode = 0644,
253 .proc_handler = &sched_nr_latency_handler,
254 .strategy = &sysctl_intvec,
255 .extra1 = &min_sched_granularity_ns,
256 .extra2 = &max_sched_granularity_ns,
257 },
258 {
259 .ctl_name = CTL_UNNUMBERED,
260 .procname = "sched_latency_ns",
261 .data = &sysctl_sched_latency,
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_wakeup_granularity_ns",
272 .data = &sysctl_sched_wakeup_granularity,
273 .maxlen = sizeof(unsigned int),
274 .mode = 0644,
275 .proc_handler = &proc_dointvec_minmax,
276 .strategy = &sysctl_intvec,
277 .extra1 = &min_wakeup_granularity_ns,
278 .extra2 = &max_wakeup_granularity_ns,
279 },
280 {
281 .ctl_name = CTL_UNNUMBERED,
282 .procname = "sched_shares_ratelimit",
283 .data = &sysctl_sched_shares_ratelimit,
284 .maxlen = sizeof(unsigned int),
285 .mode = 0644,
286 .proc_handler = &proc_dointvec,
287 },
288 {
289 .ctl_name = CTL_UNNUMBERED,
290 .procname = "sched_shares_thresh",
291 .data = &sysctl_sched_shares_thresh,
292 .maxlen = sizeof(unsigned int),
293 .mode = 0644,
294 .proc_handler = &proc_dointvec_minmax,
295 .strategy = &sysctl_intvec,
296 .extra1 = &zero,
297 },
298 {
299 .ctl_name = CTL_UNNUMBERED,
300 .procname = "sched_child_runs_first",
301 .data = &sysctl_sched_child_runs_first,
302 .maxlen = sizeof(unsigned int),
303 .mode = 0644,
304 .proc_handler = &proc_dointvec,
305 },
306 {
307 .ctl_name = CTL_UNNUMBERED,
308 .procname = "sched_features",
309 .data = &sysctl_sched_features,
310 .maxlen = sizeof(unsigned int),
311 .mode = 0644,
312 .proc_handler = &proc_dointvec,
313 },
314 {
315 .ctl_name = CTL_UNNUMBERED,
316 .procname = "sched_migration_cost",
317 .data = &sysctl_sched_migration_cost,
318 .maxlen = sizeof(unsigned int),
319 .mode = 0644,
320 .proc_handler = &proc_dointvec,
321 },
322 {
323 .ctl_name = CTL_UNNUMBERED,
324 .procname = "sched_nr_migrate",
325 .data = &sysctl_sched_nr_migrate,
326 .maxlen = sizeof(unsigned int),
327 .mode = 0644,
328 .proc_handler = &proc_dointvec,
329 },
330 #endif
331 {
332 .ctl_name = CTL_UNNUMBERED,
333 .procname = "sched_rt_period_us",
334 .data = &sysctl_sched_rt_period,
335 .maxlen = sizeof(unsigned int),
336 .mode = 0644,
337 .proc_handler = &sched_rt_handler,
338 },
339 {
340 .ctl_name = CTL_UNNUMBERED,
341 .procname = "sched_rt_runtime_us",
342 .data = &sysctl_sched_rt_runtime,
343 .maxlen = sizeof(int),
344 .mode = 0644,
345 .proc_handler = &sched_rt_handler,
346 },
347 {
348 .ctl_name = CTL_UNNUMBERED,
349 .procname = "sched_compat_yield",
350 .data = &sysctl_sched_compat_yield,
351 .maxlen = sizeof(unsigned int),
352 .mode = 0644,
353 .proc_handler = &proc_dointvec,
354 },
355 #ifdef CONFIG_PROVE_LOCKING
356 {
357 .ctl_name = CTL_UNNUMBERED,
358 .procname = "prove_locking",
359 .data = &prove_locking,
360 .maxlen = sizeof(int),
361 .mode = 0644,
362 .proc_handler = &proc_dointvec,
363 },
364 #endif
365 #ifdef CONFIG_LOCK_STAT
366 {
367 .ctl_name = CTL_UNNUMBERED,
368 .procname = "lock_stat",
369 .data = &lock_stat,
370 .maxlen = sizeof(int),
371 .mode = 0644,
372 .proc_handler = &proc_dointvec,
373 },
374 #endif
375 {
376 .ctl_name = KERN_PANIC,
377 .procname = "panic",
378 .data = &panic_timeout,
379 .maxlen = sizeof(int),
380 .mode = 0644,
381 .proc_handler = &proc_dointvec,
382 },
383 {
384 .ctl_name = KERN_CORE_USES_PID,
385 .procname = "core_uses_pid",
386 .data = &core_uses_pid,
387 .maxlen = sizeof(int),
388 .mode = 0644,
389 .proc_handler = &proc_dointvec,
390 },
391 {
392 .ctl_name = KERN_CORE_PATTERN,
393 .procname = "core_pattern",
394 .data = core_pattern,
395 .maxlen = CORENAME_MAX_SIZE,
396 .mode = 0644,
397 .proc_handler = &proc_dostring,
398 .strategy = &sysctl_string,
399 },
400 #ifdef CONFIG_PROC_SYSCTL
401 {
402 .procname = "tainted",
403 .maxlen = sizeof(long),
404 .mode = 0644,
405 .proc_handler = &proc_taint,
406 },
407 #endif
408 #ifdef CONFIG_LATENCYTOP
409 {
410 .procname = "latencytop",
411 .data = &latencytop_enabled,
412 .maxlen = sizeof(int),
413 .mode = 0644,
414 .proc_handler = &proc_dointvec,
415 },
416 #endif
417 #ifdef CONFIG_BLK_DEV_INITRD
418 {
419 .ctl_name = KERN_REALROOTDEV,
420 .procname = "real-root-dev",
421 .data = &real_root_dev,
422 .maxlen = sizeof(int),
423 .mode = 0644,
424 .proc_handler = &proc_dointvec,
425 },
426 #endif
427 {
428 .ctl_name = CTL_UNNUMBERED,
429 .procname = "print-fatal-signals",
430 .data = &print_fatal_signals,
431 .maxlen = sizeof(int),
432 .mode = 0644,
433 .proc_handler = &proc_dointvec,
434 },
435 #ifdef CONFIG_SPARC
436 {
437 .ctl_name = KERN_SPARC_REBOOT,
438 .procname = "reboot-cmd",
439 .data = reboot_command,
440 .maxlen = 256,
441 .mode = 0644,
442 .proc_handler = &proc_dostring,
443 .strategy = &sysctl_string,
444 },
445 {
446 .ctl_name = KERN_SPARC_STOP_A,
447 .procname = "stop-a",
448 .data = &stop_a_enabled,
449 .maxlen = sizeof (int),
450 .mode = 0644,
451 .proc_handler = &proc_dointvec,
452 },
453 {
454 .ctl_name = KERN_SPARC_SCONS_PWROFF,
455 .procname = "scons-poweroff",
456 .data = &scons_pwroff,
457 .maxlen = sizeof (int),
458 .mode = 0644,
459 .proc_handler = &proc_dointvec,
460 },
461 #endif
462 #ifdef CONFIG_SPARC64
463 {
464 .ctl_name = CTL_UNNUMBERED,
465 .procname = "tsb-ratio",
466 .data = &sysctl_tsb_ratio,
467 .maxlen = sizeof (int),
468 .mode = 0644,
469 .proc_handler = &proc_dointvec,
470 },
471 #endif
472 #ifdef __hppa__
473 {
474 .ctl_name = KERN_HPPA_PWRSW,
475 .procname = "soft-power",
476 .data = &pwrsw_enabled,
477 .maxlen = sizeof (int),
478 .mode = 0644,
479 .proc_handler = &proc_dointvec,
480 },
481 {
482 .ctl_name = KERN_HPPA_UNALIGNED,
483 .procname = "unaligned-trap",
484 .data = &unaligned_enabled,
485 .maxlen = sizeof (int),
486 .mode = 0644,
487 .proc_handler = &proc_dointvec,
488 },
489 #endif
490 {
491 .ctl_name = KERN_CTLALTDEL,
492 .procname = "ctrl-alt-del",
493 .data = &C_A_D,
494 .maxlen = sizeof(int),
495 .mode = 0644,
496 .proc_handler = &proc_dointvec,
497 },
498 #ifdef CONFIG_FUNCTION_TRACER
499 {
500 .ctl_name = CTL_UNNUMBERED,
501 .procname = "ftrace_enabled",
502 .data = &ftrace_enabled,
503 .maxlen = sizeof(int),
504 .mode = 0644,
505 .proc_handler = &ftrace_enable_sysctl,
506 },
507 #endif
508 #ifdef CONFIG_STACK_TRACER
509 {
510 .ctl_name = CTL_UNNUMBERED,
511 .procname = "stack_tracer_enabled",
512 .data = &stack_tracer_enabled,
513 .maxlen = sizeof(int),
514 .mode = 0644,
515 .proc_handler = &stack_trace_sysctl,
516 },
517 #endif
518 #ifdef CONFIG_TRACING
519 {
520 .ctl_name = CTL_UNNUMBERED,
521 .procname = "ftrace_dump_on_oops",
522 .data = &ftrace_dump_on_oops,
523 .maxlen = sizeof(int),
524 .mode = 0644,
525 .proc_handler = &proc_dointvec,
526 },
527 #endif
528 #ifdef CONFIG_MODULES
529 {
530 .ctl_name = KERN_MODPROBE,
531 .procname = "modprobe",
532 .data = &modprobe_path,
533 .maxlen = KMOD_PATH_LEN,
534 .mode = 0644,
535 .proc_handler = &proc_dostring,
536 .strategy = &sysctl_string,
537 },
538 #endif
539 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
540 {
541 .ctl_name = KERN_HOTPLUG,
542 .procname = "hotplug",
543 .data = &uevent_helper,
544 .maxlen = UEVENT_HELPER_PATH_LEN,
545 .mode = 0644,
546 .proc_handler = &proc_dostring,
547 .strategy = &sysctl_string,
548 },
549 #endif
550 #ifdef CONFIG_CHR_DEV_SG
551 {
552 .ctl_name = KERN_SG_BIG_BUFF,
553 .procname = "sg-big-buff",
554 .data = &sg_big_buff,
555 .maxlen = sizeof (int),
556 .mode = 0444,
557 .proc_handler = &proc_dointvec,
558 },
559 #endif
560 #ifdef CONFIG_BSD_PROCESS_ACCT
561 {
562 .ctl_name = KERN_ACCT,
563 .procname = "acct",
564 .data = &acct_parm,
565 .maxlen = 3*sizeof(int),
566 .mode = 0644,
567 .proc_handler = &proc_dointvec,
568 },
569 #endif
570 #ifdef CONFIG_MAGIC_SYSRQ
571 {
572 .ctl_name = KERN_SYSRQ,
573 .procname = "sysrq",
574 .data = &__sysrq_enabled,
575 .maxlen = sizeof (int),
576 .mode = 0644,
577 .proc_handler = &proc_dointvec,
578 },
579 #endif
580 #ifdef CONFIG_PROC_SYSCTL
581 {
582 .procname = "cad_pid",
583 .data = NULL,
584 .maxlen = sizeof (int),
585 .mode = 0600,
586 .proc_handler = &proc_do_cad_pid,
587 },
588 #endif
589 {
590 .ctl_name = KERN_MAX_THREADS,
591 .procname = "threads-max",
592 .data = &max_threads,
593 .maxlen = sizeof(int),
594 .mode = 0644,
595 .proc_handler = &proc_dointvec,
596 },
597 {
598 .ctl_name = KERN_RANDOM,
599 .procname = "random",
600 .mode = 0555,
601 .child = random_table,
602 },
603 {
604 .ctl_name = KERN_OVERFLOWUID,
605 .procname = "overflowuid",
606 .data = &overflowuid,
607 .maxlen = sizeof(int),
608 .mode = 0644,
609 .proc_handler = &proc_dointvec_minmax,
610 .strategy = &sysctl_intvec,
611 .extra1 = &minolduid,
612 .extra2 = &maxolduid,
613 },
614 {
615 .ctl_name = KERN_OVERFLOWGID,
616 .procname = "overflowgid",
617 .data = &overflowgid,
618 .maxlen = sizeof(int),
619 .mode = 0644,
620 .proc_handler = &proc_dointvec_minmax,
621 .strategy = &sysctl_intvec,
622 .extra1 = &minolduid,
623 .extra2 = &maxolduid,
624 },
625 #ifdef CONFIG_S390
626 #ifdef CONFIG_MATHEMU
627 {
628 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
629 .procname = "ieee_emulation_warnings",
630 .data = &sysctl_ieee_emulation_warnings,
631 .maxlen = sizeof(int),
632 .mode = 0644,
633 .proc_handler = &proc_dointvec,
634 },
635 #endif
636 {
637 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
638 .procname = "userprocess_debug",
639 .data = &sysctl_userprocess_debug,
640 .maxlen = sizeof(int),
641 .mode = 0644,
642 .proc_handler = &proc_dointvec,
643 },
644 #endif
645 {
646 .ctl_name = KERN_PIDMAX,
647 .procname = "pid_max",
648 .data = &pid_max,
649 .maxlen = sizeof (int),
650 .mode = 0644,
651 .proc_handler = &proc_dointvec_minmax,
652 .strategy = sysctl_intvec,
653 .extra1 = &pid_max_min,
654 .extra2 = &pid_max_max,
655 },
656 {
657 .ctl_name = KERN_PANIC_ON_OOPS,
658 .procname = "panic_on_oops",
659 .data = &panic_on_oops,
660 .maxlen = sizeof(int),
661 .mode = 0644,
662 .proc_handler = &proc_dointvec,
663 },
664 #if defined CONFIG_PRINTK
665 {
666 .ctl_name = KERN_PRINTK,
667 .procname = "printk",
668 .data = &console_loglevel,
669 .maxlen = 4*sizeof(int),
670 .mode = 0644,
671 .proc_handler = &proc_dointvec,
672 },
673 {
674 .ctl_name = KERN_PRINTK_RATELIMIT,
675 .procname = "printk_ratelimit",
676 .data = &printk_ratelimit_state.interval,
677 .maxlen = sizeof(int),
678 .mode = 0644,
679 .proc_handler = &proc_dointvec_jiffies,
680 .strategy = &sysctl_jiffies,
681 },
682 {
683 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
684 .procname = "printk_ratelimit_burst",
685 .data = &printk_ratelimit_state.burst,
686 .maxlen = sizeof(int),
687 .mode = 0644,
688 .proc_handler = &proc_dointvec,
689 },
690 #endif
691 {
692 .ctl_name = KERN_NGROUPS_MAX,
693 .procname = "ngroups_max",
694 .data = &ngroups_max,
695 .maxlen = sizeof (int),
696 .mode = 0444,
697 .proc_handler = &proc_dointvec,
698 },
699 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
700 {
701 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
702 .procname = "unknown_nmi_panic",
703 .data = &unknown_nmi_panic,
704 .maxlen = sizeof (int),
705 .mode = 0644,
706 .proc_handler = &proc_dointvec,
707 },
708 {
709 .procname = "nmi_watchdog",
710 .data = &nmi_watchdog_enabled,
711 .maxlen = sizeof (int),
712 .mode = 0644,
713 .proc_handler = &proc_nmi_enabled,
714 },
715 #endif
716 #if defined(CONFIG_X86)
717 {
718 .ctl_name = KERN_PANIC_ON_NMI,
719 .procname = "panic_on_unrecovered_nmi",
720 .data = &panic_on_unrecovered_nmi,
721 .maxlen = sizeof(int),
722 .mode = 0644,
723 .proc_handler = &proc_dointvec,
724 },
725 {
726 .ctl_name = KERN_BOOTLOADER_TYPE,
727 .procname = "bootloader_type",
728 .data = &bootloader_type,
729 .maxlen = sizeof (int),
730 .mode = 0444,
731 .proc_handler = &proc_dointvec,
732 },
733 {
734 .ctl_name = CTL_UNNUMBERED,
735 .procname = "kstack_depth_to_print",
736 .data = &kstack_depth_to_print,
737 .maxlen = sizeof(int),
738 .mode = 0644,
739 .proc_handler = &proc_dointvec,
740 },
741 {
742 .ctl_name = CTL_UNNUMBERED,
743 .procname = "io_delay_type",
744 .data = &io_delay_type,
745 .maxlen = sizeof(int),
746 .mode = 0644,
747 .proc_handler = &proc_dointvec,
748 },
749 #endif
750 #if defined(CONFIG_MMU)
751 {
752 .ctl_name = KERN_RANDOMIZE,
753 .procname = "randomize_va_space",
754 .data = &randomize_va_space,
755 .maxlen = sizeof(int),
756 .mode = 0644,
757 .proc_handler = &proc_dointvec,
758 },
759 #endif
760 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
761 {
762 .ctl_name = KERN_SPIN_RETRY,
763 .procname = "spin_retry",
764 .data = &spin_retry,
765 .maxlen = sizeof (int),
766 .mode = 0644,
767 .proc_handler = &proc_dointvec,
768 },
769 #endif
770 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
771 {
772 .procname = "acpi_video_flags",
773 .data = &acpi_realmode_flags,
774 .maxlen = sizeof (unsigned long),
775 .mode = 0644,
776 .proc_handler = &proc_doulongvec_minmax,
777 },
778 #endif
779 #ifdef CONFIG_IA64
780 {
781 .ctl_name = KERN_IA64_UNALIGNED,
782 .procname = "ignore-unaligned-usertrap",
783 .data = &no_unaligned_warning,
784 .maxlen = sizeof (int),
785 .mode = 0644,
786 .proc_handler = &proc_dointvec,
787 },
788 {
789 .ctl_name = CTL_UNNUMBERED,
790 .procname = "unaligned-dump-stack",
791 .data = &unaligned_dump_stack,
792 .maxlen = sizeof (int),
793 .mode = 0644,
794 .proc_handler = &proc_dointvec,
795 },
796 #endif
797 #ifdef CONFIG_DETECT_SOFTLOCKUP
798 {
799 .ctl_name = CTL_UNNUMBERED,
800 .procname = "softlockup_panic",
801 .data = &softlockup_panic,
802 .maxlen = sizeof(int),
803 .mode = 0644,
804 .proc_handler = &proc_dointvec_minmax,
805 .strategy = &sysctl_intvec,
806 .extra1 = &zero,
807 .extra2 = &one,
808 },
809 {
810 .ctl_name = CTL_UNNUMBERED,
811 .procname = "softlockup_thresh",
812 .data = &softlockup_thresh,
813 .maxlen = sizeof(int),
814 .mode = 0644,
815 .proc_handler = &proc_dosoftlockup_thresh,
816 .strategy = &sysctl_intvec,
817 .extra1 = &neg_one,
818 .extra2 = &sixty,
819 },
820 #endif
821 #ifdef CONFIG_DETECT_HUNG_TASK
822 {
823 .ctl_name = CTL_UNNUMBERED,
824 .procname = "hung_task_panic",
825 .data = &sysctl_hung_task_panic,
826 .maxlen = sizeof(int),
827 .mode = 0644,
828 .proc_handler = &proc_dointvec_minmax,
829 .strategy = &sysctl_intvec,
830 .extra1 = &zero,
831 .extra2 = &one,
832 },
833 {
834 .ctl_name = CTL_UNNUMBERED,
835 .procname = "hung_task_check_count",
836 .data = &sysctl_hung_task_check_count,
837 .maxlen = sizeof(unsigned long),
838 .mode = 0644,
839 .proc_handler = &proc_doulongvec_minmax,
840 .strategy = &sysctl_intvec,
841 },
842 {
843 .ctl_name = CTL_UNNUMBERED,
844 .procname = "hung_task_timeout_secs",
845 .data = &sysctl_hung_task_timeout_secs,
846 .maxlen = sizeof(unsigned long),
847 .mode = 0644,
848 .proc_handler = &proc_dohung_task_timeout_secs,
849 .strategy = &sysctl_intvec,
850 },
851 {
852 .ctl_name = CTL_UNNUMBERED,
853 .procname = "hung_task_warnings",
854 .data = &sysctl_hung_task_warnings,
855 .maxlen = sizeof(unsigned long),
856 .mode = 0644,
857 .proc_handler = &proc_doulongvec_minmax,
858 .strategy = &sysctl_intvec,
859 },
860 #endif
861 #ifdef CONFIG_COMPAT
862 {
863 .ctl_name = KERN_COMPAT_LOG,
864 .procname = "compat-log",
865 .data = &compat_log,
866 .maxlen = sizeof (int),
867 .mode = 0644,
868 .proc_handler = &proc_dointvec,
869 },
870 #endif
871 #ifdef CONFIG_RT_MUTEXES
872 {
873 .ctl_name = KERN_MAX_LOCK_DEPTH,
874 .procname = "max_lock_depth",
875 .data = &max_lock_depth,
876 .maxlen = sizeof(int),
877 .mode = 0644,
878 .proc_handler = &proc_dointvec,
879 },
880 #endif
881 {
882 .ctl_name = CTL_UNNUMBERED,
883 .procname = "poweroff_cmd",
884 .data = &poweroff_cmd,
885 .maxlen = POWEROFF_CMD_PATH_LEN,
886 .mode = 0644,
887 .proc_handler = &proc_dostring,
888 .strategy = &sysctl_string,
889 },
890 #ifdef CONFIG_KEYS
891 {
892 .ctl_name = CTL_UNNUMBERED,
893 .procname = "keys",
894 .mode = 0555,
895 .child = key_sysctls,
896 },
897 #endif
898 #ifdef CONFIG_RCU_TORTURE_TEST
899 {
900 .ctl_name = CTL_UNNUMBERED,
901 .procname = "rcutorture_runnable",
902 .data = &rcutorture_runnable,
903 .maxlen = sizeof(int),
904 .mode = 0644,
905 .proc_handler = &proc_dointvec,
906 },
907 #endif
908 #ifdef CONFIG_SLOW_WORK
909 {
910 .ctl_name = CTL_UNNUMBERED,
911 .procname = "slow-work",
912 .mode = 0555,
913 .child = slow_work_sysctls,
914 },
915 #endif
916 #ifdef CONFIG_PERF_COUNTERS
917 {
918 .ctl_name = CTL_UNNUMBERED,
919 .procname = "perf_counter_paranoid",
920 .data = &sysctl_perf_counter_paranoid,
921 .maxlen = sizeof(sysctl_perf_counter_paranoid),
922 .mode = 0644,
923 .proc_handler = &proc_dointvec,
924 },
925 {
926 .ctl_name = CTL_UNNUMBERED,
927 .procname = "perf_counter_mlock_kb",
928 .data = &sysctl_perf_counter_mlock,
929 .maxlen = sizeof(sysctl_perf_counter_mlock),
930 .mode = 0644,
931 .proc_handler = &proc_dointvec,
932 },
933 {
934 .ctl_name = CTL_UNNUMBERED,
935 .procname = "perf_counter_int_limit",
936 .data = &sysctl_perf_counter_limit,
937 .maxlen = sizeof(sysctl_perf_counter_limit),
938 .mode = 0644,
939 .proc_handler = &proc_dointvec,
940 },
941 #endif
942 /*
943 * NOTE: do not add new entries to this table unless you have read
944 * Documentation/sysctl/ctl_unnumbered.txt
945 */
946 { .ctl_name = 0 }
947 };
948
949 static struct ctl_table vm_table[] = {
950 {
951 .ctl_name = VM_OVERCOMMIT_MEMORY,
952 .procname = "overcommit_memory",
953 .data = &sysctl_overcommit_memory,
954 .maxlen = sizeof(sysctl_overcommit_memory),
955 .mode = 0644,
956 .proc_handler = &proc_dointvec,
957 },
958 {
959 .ctl_name = VM_PANIC_ON_OOM,
960 .procname = "panic_on_oom",
961 .data = &sysctl_panic_on_oom,
962 .maxlen = sizeof(sysctl_panic_on_oom),
963 .mode = 0644,
964 .proc_handler = &proc_dointvec,
965 },
966 {
967 .ctl_name = CTL_UNNUMBERED,
968 .procname = "oom_kill_allocating_task",
969 .data = &sysctl_oom_kill_allocating_task,
970 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
971 .mode = 0644,
972 .proc_handler = &proc_dointvec,
973 },
974 {
975 .ctl_name = CTL_UNNUMBERED,
976 .procname = "oom_dump_tasks",
977 .data = &sysctl_oom_dump_tasks,
978 .maxlen = sizeof(sysctl_oom_dump_tasks),
979 .mode = 0644,
980 .proc_handler = &proc_dointvec,
981 },
982 {
983 .ctl_name = VM_OVERCOMMIT_RATIO,
984 .procname = "overcommit_ratio",
985 .data = &sysctl_overcommit_ratio,
986 .maxlen = sizeof(sysctl_overcommit_ratio),
987 .mode = 0644,
988 .proc_handler = &proc_dointvec,
989 },
990 {
991 .ctl_name = VM_PAGE_CLUSTER,
992 .procname = "page-cluster",
993 .data = &page_cluster,
994 .maxlen = sizeof(int),
995 .mode = 0644,
996 .proc_handler = &proc_dointvec,
997 },
998 {
999 .ctl_name = VM_DIRTY_BACKGROUND,
1000 .procname = "dirty_background_ratio",
1001 .data = &dirty_background_ratio,
1002 .maxlen = sizeof(dirty_background_ratio),
1003 .mode = 0644,
1004 .proc_handler = &dirty_background_ratio_handler,
1005 .strategy = &sysctl_intvec,
1006 .extra1 = &zero,
1007 .extra2 = &one_hundred,
1008 },
1009 {
1010 .ctl_name = CTL_UNNUMBERED,
1011 .procname = "dirty_background_bytes",
1012 .data = &dirty_background_bytes,
1013 .maxlen = sizeof(dirty_background_bytes),
1014 .mode = 0644,
1015 .proc_handler = &dirty_background_bytes_handler,
1016 .strategy = &sysctl_intvec,
1017 .extra1 = &one_ul,
1018 },
1019 {
1020 .ctl_name = VM_DIRTY_RATIO,
1021 .procname = "dirty_ratio",
1022 .data = &vm_dirty_ratio,
1023 .maxlen = sizeof(vm_dirty_ratio),
1024 .mode = 0644,
1025 .proc_handler = &dirty_ratio_handler,
1026 .strategy = &sysctl_intvec,
1027 .extra1 = &zero,
1028 .extra2 = &one_hundred,
1029 },
1030 {
1031 .ctl_name = CTL_UNNUMBERED,
1032 .procname = "dirty_bytes",
1033 .data = &vm_dirty_bytes,
1034 .maxlen = sizeof(vm_dirty_bytes),
1035 .mode = 0644,
1036 .proc_handler = &dirty_bytes_handler,
1037 .strategy = &sysctl_intvec,
1038 .extra1 = &dirty_bytes_min,
1039 },
1040 {
1041 .procname = "dirty_writeback_centisecs",
1042 .data = &dirty_writeback_interval,
1043 .maxlen = sizeof(dirty_writeback_interval),
1044 .mode = 0644,
1045 .proc_handler = &dirty_writeback_centisecs_handler,
1046 },
1047 {
1048 .procname = "dirty_expire_centisecs",
1049 .data = &dirty_expire_interval,
1050 .maxlen = sizeof(dirty_expire_interval),
1051 .mode = 0644,
1052 .proc_handler = &proc_dointvec,
1053 },
1054 {
1055 .ctl_name = VM_NR_PDFLUSH_THREADS,
1056 .procname = "nr_pdflush_threads",
1057 .data = &nr_pdflush_threads,
1058 .maxlen = sizeof nr_pdflush_threads,
1059 .mode = 0444 /* read-only*/,
1060 .proc_handler = &proc_dointvec,
1061 },
1062 {
1063 .ctl_name = VM_SWAPPINESS,
1064 .procname = "swappiness",
1065 .data = &vm_swappiness,
1066 .maxlen = sizeof(vm_swappiness),
1067 .mode = 0644,
1068 .proc_handler = &proc_dointvec_minmax,
1069 .strategy = &sysctl_intvec,
1070 .extra1 = &zero,
1071 .extra2 = &one_hundred,
1072 },
1073 #ifdef CONFIG_HUGETLB_PAGE
1074 {
1075 .procname = "nr_hugepages",
1076 .data = NULL,
1077 .maxlen = sizeof(unsigned long),
1078 .mode = 0644,
1079 .proc_handler = &hugetlb_sysctl_handler,
1080 .extra1 = (void *)&hugetlb_zero,
1081 .extra2 = (void *)&hugetlb_infinity,
1082 },
1083 {
1084 .ctl_name = VM_HUGETLB_GROUP,
1085 .procname = "hugetlb_shm_group",
1086 .data = &sysctl_hugetlb_shm_group,
1087 .maxlen = sizeof(gid_t),
1088 .mode = 0644,
1089 .proc_handler = &proc_dointvec,
1090 },
1091 {
1092 .ctl_name = CTL_UNNUMBERED,
1093 .procname = "hugepages_treat_as_movable",
1094 .data = &hugepages_treat_as_movable,
1095 .maxlen = sizeof(int),
1096 .mode = 0644,
1097 .proc_handler = &hugetlb_treat_movable_handler,
1098 },
1099 {
1100 .ctl_name = CTL_UNNUMBERED,
1101 .procname = "nr_overcommit_hugepages",
1102 .data = NULL,
1103 .maxlen = sizeof(unsigned long),
1104 .mode = 0644,
1105 .proc_handler = &hugetlb_overcommit_handler,
1106 .extra1 = (void *)&hugetlb_zero,
1107 .extra2 = (void *)&hugetlb_infinity,
1108 },
1109 #endif
1110 {
1111 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
1112 .procname = "lowmem_reserve_ratio",
1113 .data = &sysctl_lowmem_reserve_ratio,
1114 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1115 .mode = 0644,
1116 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
1117 .strategy = &sysctl_intvec,
1118 },
1119 {
1120 .ctl_name = VM_DROP_PAGECACHE,
1121 .procname = "drop_caches",
1122 .data = &sysctl_drop_caches,
1123 .maxlen = sizeof(int),
1124 .mode = 0644,
1125 .proc_handler = drop_caches_sysctl_handler,
1126 .strategy = &sysctl_intvec,
1127 },
1128 {
1129 .ctl_name = VM_MIN_FREE_KBYTES,
1130 .procname = "min_free_kbytes",
1131 .data = &min_free_kbytes,
1132 .maxlen = sizeof(min_free_kbytes),
1133 .mode = 0644,
1134 .proc_handler = &min_free_kbytes_sysctl_handler,
1135 .strategy = &sysctl_intvec,
1136 .extra1 = &zero,
1137 },
1138 {
1139 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
1140 .procname = "percpu_pagelist_fraction",
1141 .data = &percpu_pagelist_fraction,
1142 .maxlen = sizeof(percpu_pagelist_fraction),
1143 .mode = 0644,
1144 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
1145 .strategy = &sysctl_intvec,
1146 .extra1 = &min_percpu_pagelist_fract,
1147 },
1148 #ifdef CONFIG_MMU
1149 {
1150 .ctl_name = VM_MAX_MAP_COUNT,
1151 .procname = "max_map_count",
1152 .data = &sysctl_max_map_count,
1153 .maxlen = sizeof(sysctl_max_map_count),
1154 .mode = 0644,
1155 .proc_handler = &proc_dointvec
1156 },
1157 #else
1158 {
1159 .ctl_name = CTL_UNNUMBERED,
1160 .procname = "nr_trim_pages",
1161 .data = &sysctl_nr_trim_pages,
1162 .maxlen = sizeof(sysctl_nr_trim_pages),
1163 .mode = 0644,
1164 .proc_handler = &proc_dointvec_minmax,
1165 .strategy = &sysctl_intvec,
1166 .extra1 = &zero,
1167 },
1168 #endif
1169 {
1170 .ctl_name = VM_LAPTOP_MODE,
1171 .procname = "laptop_mode",
1172 .data = &laptop_mode,
1173 .maxlen = sizeof(laptop_mode),
1174 .mode = 0644,
1175 .proc_handler = &proc_dointvec_jiffies,
1176 .strategy = &sysctl_jiffies,
1177 },
1178 {
1179 .ctl_name = VM_BLOCK_DUMP,
1180 .procname = "block_dump",
1181 .data = &block_dump,
1182 .maxlen = sizeof(block_dump),
1183 .mode = 0644,
1184 .proc_handler = &proc_dointvec,
1185 .strategy = &sysctl_intvec,
1186 .extra1 = &zero,
1187 },
1188 {
1189 .ctl_name = VM_VFS_CACHE_PRESSURE,
1190 .procname = "vfs_cache_pressure",
1191 .data = &sysctl_vfs_cache_pressure,
1192 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1193 .mode = 0644,
1194 .proc_handler = &proc_dointvec,
1195 .strategy = &sysctl_intvec,
1196 .extra1 = &zero,
1197 },
1198 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1199 {
1200 .ctl_name = VM_LEGACY_VA_LAYOUT,
1201 .procname = "legacy_va_layout",
1202 .data = &sysctl_legacy_va_layout,
1203 .maxlen = sizeof(sysctl_legacy_va_layout),
1204 .mode = 0644,
1205 .proc_handler = &proc_dointvec,
1206 .strategy = &sysctl_intvec,
1207 .extra1 = &zero,
1208 },
1209 #endif
1210 #ifdef CONFIG_NUMA
1211 {
1212 .ctl_name = VM_ZONE_RECLAIM_MODE,
1213 .procname = "zone_reclaim_mode",
1214 .data = &zone_reclaim_mode,
1215 .maxlen = sizeof(zone_reclaim_mode),
1216 .mode = 0644,
1217 .proc_handler = &proc_dointvec,
1218 .strategy = &sysctl_intvec,
1219 .extra1 = &zero,
1220 },
1221 {
1222 .ctl_name = VM_MIN_UNMAPPED,
1223 .procname = "min_unmapped_ratio",
1224 .data = &sysctl_min_unmapped_ratio,
1225 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1226 .mode = 0644,
1227 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
1228 .strategy = &sysctl_intvec,
1229 .extra1 = &zero,
1230 .extra2 = &one_hundred,
1231 },
1232 {
1233 .ctl_name = VM_MIN_SLAB,
1234 .procname = "min_slab_ratio",
1235 .data = &sysctl_min_slab_ratio,
1236 .maxlen = sizeof(sysctl_min_slab_ratio),
1237 .mode = 0644,
1238 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1239 .strategy = &sysctl_intvec,
1240 .extra1 = &zero,
1241 .extra2 = &one_hundred,
1242 },
1243 #endif
1244 #ifdef CONFIG_SMP
1245 {
1246 .ctl_name = CTL_UNNUMBERED,
1247 .procname = "stat_interval",
1248 .data = &sysctl_stat_interval,
1249 .maxlen = sizeof(sysctl_stat_interval),
1250 .mode = 0644,
1251 .proc_handler = &proc_dointvec_jiffies,
1252 .strategy = &sysctl_jiffies,
1253 },
1254 #endif
1255 #ifdef CONFIG_SECURITY
1256 {
1257 .ctl_name = CTL_UNNUMBERED,
1258 .procname = "mmap_min_addr",
1259 .data = &mmap_min_addr,
1260 .maxlen = sizeof(unsigned long),
1261 .mode = 0644,
1262 .proc_handler = &proc_doulongvec_minmax,
1263 },
1264 #endif
1265 #ifdef CONFIG_NUMA
1266 {
1267 .ctl_name = CTL_UNNUMBERED,
1268 .procname = "numa_zonelist_order",
1269 .data = &numa_zonelist_order,
1270 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1271 .mode = 0644,
1272 .proc_handler = &numa_zonelist_order_handler,
1273 .strategy = &sysctl_string,
1274 },
1275 #endif
1276 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1277 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1278 {
1279 .ctl_name = VM_VDSO_ENABLED,
1280 .procname = "vdso_enabled",
1281 .data = &vdso_enabled,
1282 .maxlen = sizeof(vdso_enabled),
1283 .mode = 0644,
1284 .proc_handler = &proc_dointvec,
1285 .strategy = &sysctl_intvec,
1286 .extra1 = &zero,
1287 },
1288 #endif
1289 #ifdef CONFIG_HIGHMEM
1290 {
1291 .ctl_name = CTL_UNNUMBERED,
1292 .procname = "highmem_is_dirtyable",
1293 .data = &vm_highmem_is_dirtyable,
1294 .maxlen = sizeof(vm_highmem_is_dirtyable),
1295 .mode = 0644,
1296 .proc_handler = &proc_dointvec_minmax,
1297 .strategy = &sysctl_intvec,
1298 .extra1 = &zero,
1299 .extra2 = &one,
1300 },
1301 #endif
1302 #ifdef CONFIG_UNEVICTABLE_LRU
1303 {
1304 .ctl_name = CTL_UNNUMBERED,
1305 .procname = "scan_unevictable_pages",
1306 .data = &scan_unevictable_pages,
1307 .maxlen = sizeof(scan_unevictable_pages),
1308 .mode = 0644,
1309 .proc_handler = &scan_unevictable_handler,
1310 },
1311 #endif
1312 /*
1313 * NOTE: do not add new entries to this table unless you have read
1314 * Documentation/sysctl/ctl_unnumbered.txt
1315 */
1316 { .ctl_name = 0 }
1317 };
1318
1319 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1320 static struct ctl_table binfmt_misc_table[] = {
1321 { .ctl_name = 0 }
1322 };
1323 #endif
1324
1325 static struct ctl_table fs_table[] = {
1326 {
1327 .ctl_name = FS_NRINODE,
1328 .procname = "inode-nr",
1329 .data = &inodes_stat,
1330 .maxlen = 2*sizeof(int),
1331 .mode = 0444,
1332 .proc_handler = &proc_dointvec,
1333 },
1334 {
1335 .ctl_name = FS_STATINODE,
1336 .procname = "inode-state",
1337 .data = &inodes_stat,
1338 .maxlen = 7*sizeof(int),
1339 .mode = 0444,
1340 .proc_handler = &proc_dointvec,
1341 },
1342 {
1343 .procname = "file-nr",
1344 .data = &files_stat,
1345 .maxlen = 3*sizeof(int),
1346 .mode = 0444,
1347 .proc_handler = &proc_nr_files,
1348 },
1349 {
1350 .ctl_name = FS_MAXFILE,
1351 .procname = "file-max",
1352 .data = &files_stat.max_files,
1353 .maxlen = sizeof(int),
1354 .mode = 0644,
1355 .proc_handler = &proc_dointvec,
1356 },
1357 {
1358 .ctl_name = CTL_UNNUMBERED,
1359 .procname = "nr_open",
1360 .data = &sysctl_nr_open,
1361 .maxlen = sizeof(int),
1362 .mode = 0644,
1363 .proc_handler = &proc_dointvec_minmax,
1364 .extra1 = &sysctl_nr_open_min,
1365 .extra2 = &sysctl_nr_open_max,
1366 },
1367 {
1368 .ctl_name = FS_DENTRY,
1369 .procname = "dentry-state",
1370 .data = &dentry_stat,
1371 .maxlen = 6*sizeof(int),
1372 .mode = 0444,
1373 .proc_handler = &proc_dointvec,
1374 },
1375 {
1376 .ctl_name = FS_OVERFLOWUID,
1377 .procname = "overflowuid",
1378 .data = &fs_overflowuid,
1379 .maxlen = sizeof(int),
1380 .mode = 0644,
1381 .proc_handler = &proc_dointvec_minmax,
1382 .strategy = &sysctl_intvec,
1383 .extra1 = &minolduid,
1384 .extra2 = &maxolduid,
1385 },
1386 {
1387 .ctl_name = FS_OVERFLOWGID,
1388 .procname = "overflowgid",
1389 .data = &fs_overflowgid,
1390 .maxlen = sizeof(int),
1391 .mode = 0644,
1392 .proc_handler = &proc_dointvec_minmax,
1393 .strategy = &sysctl_intvec,
1394 .extra1 = &minolduid,
1395 .extra2 = &maxolduid,
1396 },
1397 #ifdef CONFIG_FILE_LOCKING
1398 {
1399 .ctl_name = FS_LEASES,
1400 .procname = "leases-enable",
1401 .data = &leases_enable,
1402 .maxlen = sizeof(int),
1403 .mode = 0644,
1404 .proc_handler = &proc_dointvec,
1405 },
1406 #endif
1407 #ifdef CONFIG_DNOTIFY
1408 {
1409 .ctl_name = FS_DIR_NOTIFY,
1410 .procname = "dir-notify-enable",
1411 .data = &dir_notify_enable,
1412 .maxlen = sizeof(int),
1413 .mode = 0644,
1414 .proc_handler = &proc_dointvec,
1415 },
1416 #endif
1417 #ifdef CONFIG_MMU
1418 #ifdef CONFIG_FILE_LOCKING
1419 {
1420 .ctl_name = FS_LEASE_TIME,
1421 .procname = "lease-break-time",
1422 .data = &lease_break_time,
1423 .maxlen = sizeof(int),
1424 .mode = 0644,
1425 .proc_handler = &proc_dointvec,
1426 },
1427 #endif
1428 #ifdef CONFIG_AIO
1429 {
1430 .procname = "aio-nr",
1431 .data = &aio_nr,
1432 .maxlen = sizeof(aio_nr),
1433 .mode = 0444,
1434 .proc_handler = &proc_doulongvec_minmax,
1435 },
1436 {
1437 .procname = "aio-max-nr",
1438 .data = &aio_max_nr,
1439 .maxlen = sizeof(aio_max_nr),
1440 .mode = 0644,
1441 .proc_handler = &proc_doulongvec_minmax,
1442 },
1443 #endif /* CONFIG_AIO */
1444 #ifdef CONFIG_INOTIFY_USER
1445 {
1446 .ctl_name = FS_INOTIFY,
1447 .procname = "inotify",
1448 .mode = 0555,
1449 .child = inotify_table,
1450 },
1451 #endif
1452 #ifdef CONFIG_EPOLL
1453 {
1454 .procname = "epoll",
1455 .mode = 0555,
1456 .child = epoll_table,
1457 },
1458 #endif
1459 #endif
1460 {
1461 .ctl_name = KERN_SETUID_DUMPABLE,
1462 .procname = "suid_dumpable",
1463 .data = &suid_dumpable,
1464 .maxlen = sizeof(int),
1465 .mode = 0644,
1466 .proc_handler = &proc_dointvec_minmax,
1467 .strategy = &sysctl_intvec,
1468 .extra1 = &zero,
1469 .extra2 = &two,
1470 },
1471 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1472 {
1473 .ctl_name = CTL_UNNUMBERED,
1474 .procname = "binfmt_misc",
1475 .mode = 0555,
1476 .child = binfmt_misc_table,
1477 },
1478 #endif
1479 /*
1480 * NOTE: do not add new entries to this table unless you have read
1481 * Documentation/sysctl/ctl_unnumbered.txt
1482 */
1483 { .ctl_name = 0 }
1484 };
1485
1486 static struct ctl_table debug_table[] = {
1487 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1488 {
1489 .ctl_name = CTL_UNNUMBERED,
1490 .procname = "exception-trace",
1491 .data = &show_unhandled_signals,
1492 .maxlen = sizeof(int),
1493 .mode = 0644,
1494 .proc_handler = proc_dointvec
1495 },
1496 #endif
1497 { .ctl_name = 0 }
1498 };
1499
1500 static struct ctl_table dev_table[] = {
1501 { .ctl_name = 0 }
1502 };
1503
1504 static DEFINE_SPINLOCK(sysctl_lock);
1505
1506 /* called under sysctl_lock */
1507 static int use_table(struct ctl_table_header *p)
1508 {
1509 if (unlikely(p->unregistering))
1510 return 0;
1511 p->used++;
1512 return 1;
1513 }
1514
1515 /* called under sysctl_lock */
1516 static void unuse_table(struct ctl_table_header *p)
1517 {
1518 if (!--p->used)
1519 if (unlikely(p->unregistering))
1520 complete(p->unregistering);
1521 }
1522
1523 /* called under sysctl_lock, will reacquire if has to wait */
1524 static void start_unregistering(struct ctl_table_header *p)
1525 {
1526 /*
1527 * if p->used is 0, nobody will ever touch that entry again;
1528 * we'll eliminate all paths to it before dropping sysctl_lock
1529 */
1530 if (unlikely(p->used)) {
1531 struct completion wait;
1532 init_completion(&wait);
1533 p->unregistering = &wait;
1534 spin_unlock(&sysctl_lock);
1535 wait_for_completion(&wait);
1536 spin_lock(&sysctl_lock);
1537 } else {
1538 /* anything non-NULL; we'll never dereference it */
1539 p->unregistering = ERR_PTR(-EINVAL);
1540 }
1541 /*
1542 * do not remove from the list until nobody holds it; walking the
1543 * list in do_sysctl() relies on that.
1544 */
1545 list_del_init(&p->ctl_entry);
1546 }
1547
1548 void sysctl_head_get(struct ctl_table_header *head)
1549 {
1550 spin_lock(&sysctl_lock);
1551 head->count++;
1552 spin_unlock(&sysctl_lock);
1553 }
1554
1555 void sysctl_head_put(struct ctl_table_header *head)
1556 {
1557 spin_lock(&sysctl_lock);
1558 if (!--head->count)
1559 kfree(head);
1560 spin_unlock(&sysctl_lock);
1561 }
1562
1563 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1564 {
1565 if (!head)
1566 BUG();
1567 spin_lock(&sysctl_lock);
1568 if (!use_table(head))
1569 head = ERR_PTR(-ENOENT);
1570 spin_unlock(&sysctl_lock);
1571 return head;
1572 }
1573
1574 void sysctl_head_finish(struct ctl_table_header *head)
1575 {
1576 if (!head)
1577 return;
1578 spin_lock(&sysctl_lock);
1579 unuse_table(head);
1580 spin_unlock(&sysctl_lock);
1581 }
1582
1583 static struct ctl_table_set *
1584 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1585 {
1586 struct ctl_table_set *set = &root->default_set;
1587 if (root->lookup)
1588 set = root->lookup(root, namespaces);
1589 return set;
1590 }
1591
1592 static struct list_head *
1593 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1594 {
1595 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1596 return &set->list;
1597 }
1598
1599 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1600 struct ctl_table_header *prev)
1601 {
1602 struct ctl_table_root *root;
1603 struct list_head *header_list;
1604 struct ctl_table_header *head;
1605 struct list_head *tmp;
1606
1607 spin_lock(&sysctl_lock);
1608 if (prev) {
1609 head = prev;
1610 tmp = &prev->ctl_entry;
1611 unuse_table(prev);
1612 goto next;
1613 }
1614 tmp = &root_table_header.ctl_entry;
1615 for (;;) {
1616 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1617
1618 if (!use_table(head))
1619 goto next;
1620 spin_unlock(&sysctl_lock);
1621 return head;
1622 next:
1623 root = head->root;
1624 tmp = tmp->next;
1625 header_list = lookup_header_list(root, namespaces);
1626 if (tmp != header_list)
1627 continue;
1628
1629 do {
1630 root = list_entry(root->root_list.next,
1631 struct ctl_table_root, root_list);
1632 if (root == &sysctl_table_root)
1633 goto out;
1634 header_list = lookup_header_list(root, namespaces);
1635 } while (list_empty(header_list));
1636 tmp = header_list->next;
1637 }
1638 out:
1639 spin_unlock(&sysctl_lock);
1640 return NULL;
1641 }
1642
1643 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1644 {
1645 return __sysctl_head_next(current->nsproxy, prev);
1646 }
1647
1648 void register_sysctl_root(struct ctl_table_root *root)
1649 {
1650 spin_lock(&sysctl_lock);
1651 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1652 spin_unlock(&sysctl_lock);
1653 }
1654
1655 #ifdef CONFIG_SYSCTL_SYSCALL
1656 /* Perform the actual read/write of a sysctl table entry. */
1657 static int do_sysctl_strategy(struct ctl_table_root *root,
1658 struct ctl_table *table,
1659 void __user *oldval, size_t __user *oldlenp,
1660 void __user *newval, size_t newlen)
1661 {
1662 int op = 0, rc;
1663
1664 if (oldval)
1665 op |= MAY_READ;
1666 if (newval)
1667 op |= MAY_WRITE;
1668 if (sysctl_perm(root, table, op))
1669 return -EPERM;
1670
1671 if (table->strategy) {
1672 rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1673 if (rc < 0)
1674 return rc;
1675 if (rc > 0)
1676 return 0;
1677 }
1678
1679 /* If there is no strategy routine, or if the strategy returns
1680 * zero, proceed with automatic r/w */
1681 if (table->data && table->maxlen) {
1682 rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1683 if (rc < 0)
1684 return rc;
1685 }
1686 return 0;
1687 }
1688
1689 static int parse_table(int __user *name, int nlen,
1690 void __user *oldval, size_t __user *oldlenp,
1691 void __user *newval, size_t newlen,
1692 struct ctl_table_root *root,
1693 struct ctl_table *table)
1694 {
1695 int n;
1696 repeat:
1697 if (!nlen)
1698 return -ENOTDIR;
1699 if (get_user(n, name))
1700 return -EFAULT;
1701 for ( ; table->ctl_name || table->procname; table++) {
1702 if (!table->ctl_name)
1703 continue;
1704 if (n == table->ctl_name) {
1705 int error;
1706 if (table->child) {
1707 if (sysctl_perm(root, table, MAY_EXEC))
1708 return -EPERM;
1709 name++;
1710 nlen--;
1711 table = table->child;
1712 goto repeat;
1713 }
1714 error = do_sysctl_strategy(root, table,
1715 oldval, oldlenp,
1716 newval, newlen);
1717 return error;
1718 }
1719 }
1720 return -ENOTDIR;
1721 }
1722
1723 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1724 void __user *newval, size_t newlen)
1725 {
1726 struct ctl_table_header *head;
1727 int error = -ENOTDIR;
1728
1729 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1730 return -ENOTDIR;
1731 if (oldval) {
1732 int old_len;
1733 if (!oldlenp || get_user(old_len, oldlenp))
1734 return -EFAULT;
1735 }
1736
1737 for (head = sysctl_head_next(NULL); head;
1738 head = sysctl_head_next(head)) {
1739 error = parse_table(name, nlen, oldval, oldlenp,
1740 newval, newlen,
1741 head->root, head->ctl_table);
1742 if (error != -ENOTDIR) {
1743 sysctl_head_finish(head);
1744 break;
1745 }
1746 }
1747 return error;
1748 }
1749
1750 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
1751 {
1752 struct __sysctl_args tmp;
1753 int error;
1754
1755 if (copy_from_user(&tmp, args, sizeof(tmp)))
1756 return -EFAULT;
1757
1758 error = deprecated_sysctl_warning(&tmp);
1759 if (error)
1760 goto out;
1761
1762 lock_kernel();
1763 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1764 tmp.newval, tmp.newlen);
1765 unlock_kernel();
1766 out:
1767 return error;
1768 }
1769 #endif /* CONFIG_SYSCTL_SYSCALL */
1770
1771 /*
1772 * sysctl_perm does NOT grant the superuser all rights automatically, because
1773 * some sysctl variables are readonly even to root.
1774 */
1775
1776 static int test_perm(int mode, int op)
1777 {
1778 if (!current_euid())
1779 mode >>= 6;
1780 else if (in_egroup_p(0))
1781 mode >>= 3;
1782 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1783 return 0;
1784 return -EACCES;
1785 }
1786
1787 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1788 {
1789 int error;
1790 int mode;
1791
1792 error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1793 if (error)
1794 return error;
1795
1796 if (root->permissions)
1797 mode = root->permissions(root, current->nsproxy, table);
1798 else
1799 mode = table->mode;
1800
1801 return test_perm(mode, op);
1802 }
1803
1804 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1805 {
1806 for (; table->ctl_name || table->procname; table++) {
1807 table->parent = parent;
1808 if (table->child)
1809 sysctl_set_parent(table, table->child);
1810 }
1811 }
1812
1813 static __init int sysctl_init(void)
1814 {
1815 sysctl_set_parent(NULL, root_table);
1816 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1817 {
1818 int err;
1819 err = sysctl_check_table(current->nsproxy, root_table);
1820 }
1821 #endif
1822 return 0;
1823 }
1824
1825 core_initcall(sysctl_init);
1826
1827 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1828 struct ctl_table *table)
1829 {
1830 struct ctl_table *p;
1831 const char *s = branch->procname;
1832
1833 /* branch should have named subdirectory as its first element */
1834 if (!s || !branch->child)
1835 return NULL;
1836
1837 /* ... and nothing else */
1838 if (branch[1].procname || branch[1].ctl_name)
1839 return NULL;
1840
1841 /* table should contain subdirectory with the same name */
1842 for (p = table; p->procname || p->ctl_name; p++) {
1843 if (!p->child)
1844 continue;
1845 if (p->procname && strcmp(p->procname, s) == 0)
1846 return p;
1847 }
1848 return NULL;
1849 }
1850
1851 /* see if attaching q to p would be an improvement */
1852 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1853 {
1854 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1855 struct ctl_table *next;
1856 int is_better = 0;
1857 int not_in_parent = !p->attached_by;
1858
1859 while ((next = is_branch_in(by, to)) != NULL) {
1860 if (by == q->attached_by)
1861 is_better = 1;
1862 if (to == p->attached_by)
1863 not_in_parent = 1;
1864 by = by->child;
1865 to = next->child;
1866 }
1867
1868 if (is_better && not_in_parent) {
1869 q->attached_by = by;
1870 q->attached_to = to;
1871 q->parent = p;
1872 }
1873 }
1874
1875 /**
1876 * __register_sysctl_paths - register a sysctl hierarchy
1877 * @root: List of sysctl headers to register on
1878 * @namespaces: Data to compute which lists of sysctl entries are visible
1879 * @path: The path to the directory the sysctl table is in.
1880 * @table: the top-level table structure
1881 *
1882 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1883 * array. A completely 0 filled entry terminates the table.
1884 *
1885 * The members of the &struct ctl_table structure are used as follows:
1886 *
1887 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1888 * must be unique within that level of sysctl
1889 *
1890 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1891 * enter a sysctl file
1892 *
1893 * data - a pointer to data for use by proc_handler
1894 *
1895 * maxlen - the maximum size in bytes of the data
1896 *
1897 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1898 *
1899 * child - a pointer to the child sysctl table if this entry is a directory, or
1900 * %NULL.
1901 *
1902 * proc_handler - the text handler routine (described below)
1903 *
1904 * strategy - the strategy routine (described below)
1905 *
1906 * de - for internal use by the sysctl routines
1907 *
1908 * extra1, extra2 - extra pointers usable by the proc handler routines
1909 *
1910 * Leaf nodes in the sysctl tree will be represented by a single file
1911 * under /proc; non-leaf nodes will be represented by directories.
1912 *
1913 * sysctl(2) can automatically manage read and write requests through
1914 * the sysctl table. The data and maxlen fields of the ctl_table
1915 * struct enable minimal validation of the values being written to be
1916 * performed, and the mode field allows minimal authentication.
1917 *
1918 * More sophisticated management can be enabled by the provision of a
1919 * strategy routine with the table entry. This will be called before
1920 * any automatic read or write of the data is performed.
1921 *
1922 * The strategy routine may return
1923 *
1924 * < 0 - Error occurred (error is passed to user process)
1925 *
1926 * 0 - OK - proceed with automatic read or write.
1927 *
1928 * > 0 - OK - read or write has been done by the strategy routine, so
1929 * return immediately.
1930 *
1931 * There must be a proc_handler routine for any terminal nodes
1932 * mirrored under /proc/sys (non-terminals are handled by a built-in
1933 * directory handler). Several default handlers are available to
1934 * cover common cases -
1935 *
1936 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1937 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1938 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1939 *
1940 * It is the handler's job to read the input buffer from user memory
1941 * and process it. The handler should return 0 on success.
1942 *
1943 * This routine returns %NULL on a failure to register, and a pointer
1944 * to the table header on success.
1945 */
1946 struct ctl_table_header *__register_sysctl_paths(
1947 struct ctl_table_root *root,
1948 struct nsproxy *namespaces,
1949 const struct ctl_path *path, struct ctl_table *table)
1950 {
1951 struct ctl_table_header *header;
1952 struct ctl_table *new, **prevp;
1953 unsigned int n, npath;
1954 struct ctl_table_set *set;
1955
1956 /* Count the path components */
1957 for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1958 ;
1959
1960 /*
1961 * For each path component, allocate a 2-element ctl_table array.
1962 * The first array element will be filled with the sysctl entry
1963 * for this, the second will be the sentinel (ctl_name == 0).
1964 *
1965 * We allocate everything in one go so that we don't have to
1966 * worry about freeing additional memory in unregister_sysctl_table.
1967 */
1968 header = kzalloc(sizeof(struct ctl_table_header) +
1969 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1970 if (!header)
1971 return NULL;
1972
1973 new = (struct ctl_table *) (header + 1);
1974
1975 /* Now connect the dots */
1976 prevp = &header->ctl_table;
1977 for (n = 0; n < npath; ++n, ++path) {
1978 /* Copy the procname */
1979 new->procname = path->procname;
1980 new->ctl_name = path->ctl_name;
1981 new->mode = 0555;
1982
1983 *prevp = new;
1984 prevp = &new->child;
1985
1986 new += 2;
1987 }
1988 *prevp = table;
1989 header->ctl_table_arg = table;
1990
1991 INIT_LIST_HEAD(&header->ctl_entry);
1992 header->used = 0;
1993 header->unregistering = NULL;
1994 header->root = root;
1995 sysctl_set_parent(NULL, header->ctl_table);
1996 header->count = 1;
1997 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1998 if (sysctl_check_table(namespaces, header->ctl_table)) {
1999 kfree(header);
2000 return NULL;
2001 }
2002 #endif
2003 spin_lock(&sysctl_lock);
2004 header->set = lookup_header_set(root, namespaces);
2005 header->attached_by = header->ctl_table;
2006 header->attached_to = root_table;
2007 header->parent = &root_table_header;
2008 for (set = header->set; set; set = set->parent) {
2009 struct ctl_table_header *p;
2010 list_for_each_entry(p, &set->list, ctl_entry) {
2011 if (p->unregistering)
2012 continue;
2013 try_attach(p, header);
2014 }
2015 }
2016 header->parent->count++;
2017 list_add_tail(&header->ctl_entry, &header->set->list);
2018 spin_unlock(&sysctl_lock);
2019
2020 return header;
2021 }
2022
2023 /**
2024 * register_sysctl_table_path - register a sysctl table hierarchy
2025 * @path: The path to the directory the sysctl table is in.
2026 * @table: the top-level table structure
2027 *
2028 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2029 * array. A completely 0 filled entry terminates the table.
2030 *
2031 * See __register_sysctl_paths for more details.
2032 */
2033 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2034 struct ctl_table *table)
2035 {
2036 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
2037 path, table);
2038 }
2039
2040 /**
2041 * register_sysctl_table - register a sysctl table hierarchy
2042 * @table: the top-level table structure
2043 *
2044 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2045 * array. A completely 0 filled entry terminates the table.
2046 *
2047 * See register_sysctl_paths for more details.
2048 */
2049 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
2050 {
2051 static const struct ctl_path null_path[] = { {} };
2052
2053 return register_sysctl_paths(null_path, table);
2054 }
2055
2056 /**
2057 * unregister_sysctl_table - unregister a sysctl table hierarchy
2058 * @header: the header returned from register_sysctl_table
2059 *
2060 * Unregisters the sysctl table and all children. proc entries may not
2061 * actually be removed until they are no longer used by anyone.
2062 */
2063 void unregister_sysctl_table(struct ctl_table_header * header)
2064 {
2065 might_sleep();
2066
2067 if (header == NULL)
2068 return;
2069
2070 spin_lock(&sysctl_lock);
2071 start_unregistering(header);
2072 if (!--header->parent->count) {
2073 WARN_ON(1);
2074 kfree(header->parent);
2075 }
2076 if (!--header->count)
2077 kfree(header);
2078 spin_unlock(&sysctl_lock);
2079 }
2080
2081 int sysctl_is_seen(struct ctl_table_header *p)
2082 {
2083 struct ctl_table_set *set = p->set;
2084 int res;
2085 spin_lock(&sysctl_lock);
2086 if (p->unregistering)
2087 res = 0;
2088 else if (!set->is_seen)
2089 res = 1;
2090 else
2091 res = set->is_seen(set);
2092 spin_unlock(&sysctl_lock);
2093 return res;
2094 }
2095
2096 void setup_sysctl_set(struct ctl_table_set *p,
2097 struct ctl_table_set *parent,
2098 int (*is_seen)(struct ctl_table_set *))
2099 {
2100 INIT_LIST_HEAD(&p->list);
2101 p->parent = parent ? parent : &sysctl_table_root.default_set;
2102 p->is_seen = is_seen;
2103 }
2104
2105 #else /* !CONFIG_SYSCTL */
2106 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2107 {
2108 return NULL;
2109 }
2110
2111 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2112 struct ctl_table *table)
2113 {
2114 return NULL;
2115 }
2116
2117 void unregister_sysctl_table(struct ctl_table_header * table)
2118 {
2119 }
2120
2121 void setup_sysctl_set(struct ctl_table_set *p,
2122 struct ctl_table_set *parent,
2123 int (*is_seen)(struct ctl_table_set *))
2124 {
2125 }
2126
2127 void sysctl_head_put(struct ctl_table_header *head)
2128 {
2129 }
2130
2131 #endif /* CONFIG_SYSCTL */
2132
2133 /*
2134 * /proc/sys support
2135 */
2136
2137 #ifdef CONFIG_PROC_SYSCTL
2138
2139 static int _proc_do_string(void* data, int maxlen, int write,
2140 struct file *filp, void __user *buffer,
2141 size_t *lenp, loff_t *ppos)
2142 {
2143 size_t len;
2144 char __user *p;
2145 char c;
2146
2147 if (!data || !maxlen || !*lenp) {
2148 *lenp = 0;
2149 return 0;
2150 }
2151
2152 if (write) {
2153 len = 0;
2154 p = buffer;
2155 while (len < *lenp) {
2156 if (get_user(c, p++))
2157 return -EFAULT;
2158 if (c == 0 || c == '\n')
2159 break;
2160 len++;
2161 }
2162 if (len >= maxlen)
2163 len = maxlen-1;
2164 if(copy_from_user(data, buffer, len))
2165 return -EFAULT;
2166 ((char *) data)[len] = 0;
2167 *ppos += *lenp;
2168 } else {
2169 len = strlen(data);
2170 if (len > maxlen)
2171 len = maxlen;
2172
2173 if (*ppos > len) {
2174 *lenp = 0;
2175 return 0;
2176 }
2177
2178 data += *ppos;
2179 len -= *ppos;
2180
2181 if (len > *lenp)
2182 len = *lenp;
2183 if (len)
2184 if(copy_to_user(buffer, data, len))
2185 return -EFAULT;
2186 if (len < *lenp) {
2187 if(put_user('\n', ((char __user *) buffer) + len))
2188 return -EFAULT;
2189 len++;
2190 }
2191 *lenp = len;
2192 *ppos += len;
2193 }
2194 return 0;
2195 }
2196
2197 /**
2198 * proc_dostring - read a string sysctl
2199 * @table: the sysctl table
2200 * @write: %TRUE if this is a write to the sysctl file
2201 * @filp: the file structure
2202 * @buffer: the user buffer
2203 * @lenp: the size of the user buffer
2204 * @ppos: file position
2205 *
2206 * Reads/writes a string from/to the user buffer. If the kernel
2207 * buffer provided is not large enough to hold the string, the
2208 * string is truncated. The copied string is %NULL-terminated.
2209 * If the string is being read by the user process, it is copied
2210 * and a newline '\n' is added. It is truncated if the buffer is
2211 * not large enough.
2212 *
2213 * Returns 0 on success.
2214 */
2215 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2216 void __user *buffer, size_t *lenp, loff_t *ppos)
2217 {
2218 return _proc_do_string(table->data, table->maxlen, write, filp,
2219 buffer, lenp, ppos);
2220 }
2221
2222
2223 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2224 int *valp,
2225 int write, void *data)
2226 {
2227 if (write) {
2228 *valp = *negp ? -*lvalp : *lvalp;
2229 } else {
2230 int val = *valp;
2231 if (val < 0) {
2232 *negp = -1;
2233 *lvalp = (unsigned long)-val;
2234 } else {
2235 *negp = 0;
2236 *lvalp = (unsigned long)val;
2237 }
2238 }
2239 return 0;
2240 }
2241
2242 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2243 int write, struct file *filp, void __user *buffer,
2244 size_t *lenp, loff_t *ppos,
2245 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2246 int write, void *data),
2247 void *data)
2248 {
2249 #define TMPBUFLEN 21
2250 int *i, vleft, first=1, neg, val;
2251 unsigned long lval;
2252 size_t left, len;
2253
2254 char buf[TMPBUFLEN], *p;
2255 char __user *s = buffer;
2256
2257 if (!tbl_data || !table->maxlen || !*lenp ||
2258 (*ppos && !write)) {
2259 *lenp = 0;
2260 return 0;
2261 }
2262
2263 i = (int *) tbl_data;
2264 vleft = table->maxlen / sizeof(*i);
2265 left = *lenp;
2266
2267 if (!conv)
2268 conv = do_proc_dointvec_conv;
2269
2270 for (; left && vleft--; i++, first=0) {
2271 if (write) {
2272 while (left) {
2273 char c;
2274 if (get_user(c, s))
2275 return -EFAULT;
2276 if (!isspace(c))
2277 break;
2278 left--;
2279 s++;
2280 }
2281 if (!left)
2282 break;
2283 neg = 0;
2284 len = left;
2285 if (len > sizeof(buf) - 1)
2286 len = sizeof(buf) - 1;
2287 if (copy_from_user(buf, s, len))
2288 return -EFAULT;
2289 buf[len] = 0;
2290 p = buf;
2291 if (*p == '-' && left > 1) {
2292 neg = 1;
2293 p++;
2294 }
2295 if (*p < '0' || *p > '9')
2296 break;
2297
2298 lval = simple_strtoul(p, &p, 0);
2299
2300 len = p-buf;
2301 if ((len < left) && *p && !isspace(*p))
2302 break;
2303 if (neg)
2304 val = -val;
2305 s += len;
2306 left -= len;
2307
2308 if (conv(&neg, &lval, i, 1, data))
2309 break;
2310 } else {
2311 p = buf;
2312 if (!first)
2313 *p++ = '\t';
2314
2315 if (conv(&neg, &lval, i, 0, data))
2316 break;
2317
2318 sprintf(p, "%s%lu", neg ? "-" : "", lval);
2319 len = strlen(buf);
2320 if (len > left)
2321 len = left;
2322 if(copy_to_user(s, buf, len))
2323 return -EFAULT;
2324 left -= len;
2325 s += len;
2326 }
2327 }
2328
2329 if (!write && !first && left) {
2330 if(put_user('\n', s))
2331 return -EFAULT;
2332 left--, s++;
2333 }
2334 if (write) {
2335 while (left) {
2336 char c;
2337 if (get_user(c, s++))
2338 return -EFAULT;
2339 if (!isspace(c))
2340 break;
2341 left--;
2342 }
2343 }
2344 if (write && first)
2345 return -EINVAL;
2346 *lenp -= left;
2347 *ppos += *lenp;
2348 return 0;
2349 #undef TMPBUFLEN
2350 }
2351
2352 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2353 void __user *buffer, size_t *lenp, loff_t *ppos,
2354 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2355 int write, void *data),
2356 void *data)
2357 {
2358 return __do_proc_dointvec(table->data, table, write, filp,
2359 buffer, lenp, ppos, conv, data);
2360 }
2361
2362 /**
2363 * proc_dointvec - read a vector of integers
2364 * @table: the sysctl table
2365 * @write: %TRUE if this is a write to the sysctl file
2366 * @filp: the file structure
2367 * @buffer: the user buffer
2368 * @lenp: the size of the user buffer
2369 * @ppos: file position
2370 *
2371 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2372 * values from/to the user buffer, treated as an ASCII string.
2373 *
2374 * Returns 0 on success.
2375 */
2376 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2377 void __user *buffer, size_t *lenp, loff_t *ppos)
2378 {
2379 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2380 NULL,NULL);
2381 }
2382
2383 /*
2384 * Taint values can only be increased
2385 * This means we can safely use a temporary.
2386 */
2387 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
2388 void __user *buffer, size_t *lenp, loff_t *ppos)
2389 {
2390 struct ctl_table t;
2391 unsigned long tmptaint = get_taint();
2392 int err;
2393
2394 if (write && !capable(CAP_SYS_ADMIN))
2395 return -EPERM;
2396
2397 t = *table;
2398 t.data = &tmptaint;
2399 err = proc_doulongvec_minmax(&t, write, filp, buffer, lenp, ppos);
2400 if (err < 0)
2401 return err;
2402
2403 if (write) {
2404 /*
2405 * Poor man's atomic or. Not worth adding a primitive
2406 * to everyone's atomic.h for this
2407 */
2408 int i;
2409 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2410 if ((tmptaint >> i) & 1)
2411 add_taint(i);
2412 }
2413 }
2414
2415 return err;
2416 }
2417
2418 struct do_proc_dointvec_minmax_conv_param {
2419 int *min;
2420 int *max;
2421 };
2422
2423 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2424 int *valp,
2425 int write, void *data)
2426 {
2427 struct do_proc_dointvec_minmax_conv_param *param = data;
2428 if (write) {
2429 int val = *negp ? -*lvalp : *lvalp;
2430 if ((param->min && *param->min > val) ||
2431 (param->max && *param->max < val))
2432 return -EINVAL;
2433 *valp = val;
2434 } else {
2435 int val = *valp;
2436 if (val < 0) {
2437 *negp = -1;
2438 *lvalp = (unsigned long)-val;
2439 } else {
2440 *negp = 0;
2441 *lvalp = (unsigned long)val;
2442 }
2443 }
2444 return 0;
2445 }
2446
2447 /**
2448 * proc_dointvec_minmax - read a vector of integers with min/max values
2449 * @table: the sysctl table
2450 * @write: %TRUE if this is a write to the sysctl file
2451 * @filp: the file structure
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 * This routine will ensure the values are within the range specified by
2460 * table->extra1 (min) and table->extra2 (max).
2461 *
2462 * Returns 0 on success.
2463 */
2464 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2465 void __user *buffer, size_t *lenp, loff_t *ppos)
2466 {
2467 struct do_proc_dointvec_minmax_conv_param param = {
2468 .min = (int *) table->extra1,
2469 .max = (int *) table->extra2,
2470 };
2471 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2472 do_proc_dointvec_minmax_conv, &param);
2473 }
2474
2475 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2476 struct file *filp,
2477 void __user *buffer,
2478 size_t *lenp, loff_t *ppos,
2479 unsigned long convmul,
2480 unsigned long convdiv)
2481 {
2482 #define TMPBUFLEN 21
2483 unsigned long *i, *min, *max, val;
2484 int vleft, first=1, neg;
2485 size_t len, left;
2486 char buf[TMPBUFLEN], *p;
2487 char __user *s = buffer;
2488
2489 if (!data || !table->maxlen || !*lenp ||
2490 (*ppos && !write)) {
2491 *lenp = 0;
2492 return 0;
2493 }
2494
2495 i = (unsigned long *) data;
2496 min = (unsigned long *) table->extra1;
2497 max = (unsigned long *) table->extra2;
2498 vleft = table->maxlen / sizeof(unsigned long);
2499 left = *lenp;
2500
2501 for (; left && vleft--; i++, min++, max++, first=0) {
2502 if (write) {
2503 while (left) {
2504 char c;
2505 if (get_user(c, s))
2506 return -EFAULT;
2507 if (!isspace(c))
2508 break;
2509 left--;
2510 s++;
2511 }
2512 if (!left)
2513 break;
2514 neg = 0;
2515 len = left;
2516 if (len > TMPBUFLEN-1)
2517 len = TMPBUFLEN-1;
2518 if (copy_from_user(buf, s, len))
2519 return -EFAULT;
2520 buf[len] = 0;
2521 p = buf;
2522 if (*p == '-' && left > 1) {
2523 neg = 1;
2524 p++;
2525 }
2526 if (*p < '0' || *p > '9')
2527 break;
2528 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2529 len = p-buf;
2530 if ((len < left) && *p && !isspace(*p))
2531 break;
2532 if (neg)
2533 val = -val;
2534 s += len;
2535 left -= len;
2536
2537 if(neg)
2538 continue;
2539 if ((min && val < *min) || (max && val > *max))
2540 continue;
2541 *i = val;
2542 } else {
2543 p = buf;
2544 if (!first)
2545 *p++ = '\t';
2546 sprintf(p, "%lu", convdiv * (*i) / convmul);
2547 len = strlen(buf);
2548 if (len > left)
2549 len = left;
2550 if(copy_to_user(s, buf, len))
2551 return -EFAULT;
2552 left -= len;
2553 s += len;
2554 }
2555 }
2556
2557 if (!write && !first && left) {
2558 if(put_user('\n', s))
2559 return -EFAULT;
2560 left--, s++;
2561 }
2562 if (write) {
2563 while (left) {
2564 char c;
2565 if (get_user(c, s++))
2566 return -EFAULT;
2567 if (!isspace(c))
2568 break;
2569 left--;
2570 }
2571 }
2572 if (write && first)
2573 return -EINVAL;
2574 *lenp -= left;
2575 *ppos += *lenp;
2576 return 0;
2577 #undef TMPBUFLEN
2578 }
2579
2580 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2581 struct file *filp,
2582 void __user *buffer,
2583 size_t *lenp, loff_t *ppos,
2584 unsigned long convmul,
2585 unsigned long convdiv)
2586 {
2587 return __do_proc_doulongvec_minmax(table->data, table, write,
2588 filp, buffer, lenp, ppos, convmul, convdiv);
2589 }
2590
2591 /**
2592 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2593 * @table: the sysctl table
2594 * @write: %TRUE if this is a write to the sysctl file
2595 * @filp: the file structure
2596 * @buffer: the user buffer
2597 * @lenp: the size of the user buffer
2598 * @ppos: file position
2599 *
2600 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2601 * values from/to the user buffer, treated as an ASCII string.
2602 *
2603 * This routine will ensure the values are within the range specified by
2604 * table->extra1 (min) and table->extra2 (max).
2605 *
2606 * Returns 0 on success.
2607 */
2608 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2609 void __user *buffer, size_t *lenp, loff_t *ppos)
2610 {
2611 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2612 }
2613
2614 /**
2615 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2616 * @table: the sysctl table
2617 * @write: %TRUE if this is a write to the sysctl file
2618 * @filp: the file structure
2619 * @buffer: the user buffer
2620 * @lenp: the size of the user buffer
2621 * @ppos: file position
2622 *
2623 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2624 * values from/to the user buffer, treated as an ASCII string. The values
2625 * are treated as milliseconds, and converted to jiffies when they are stored.
2626 *
2627 * This routine will ensure the values are within the range specified by
2628 * table->extra1 (min) and table->extra2 (max).
2629 *
2630 * Returns 0 on success.
2631 */
2632 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2633 struct file *filp,
2634 void __user *buffer,
2635 size_t *lenp, loff_t *ppos)
2636 {
2637 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2638 lenp, ppos, HZ, 1000l);
2639 }
2640
2641
2642 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2643 int *valp,
2644 int write, void *data)
2645 {
2646 if (write) {
2647 if (*lvalp > LONG_MAX / HZ)
2648 return 1;
2649 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2650 } else {
2651 int val = *valp;
2652 unsigned long lval;
2653 if (val < 0) {
2654 *negp = -1;
2655 lval = (unsigned long)-val;
2656 } else {
2657 *negp = 0;
2658 lval = (unsigned long)val;
2659 }
2660 *lvalp = lval / HZ;
2661 }
2662 return 0;
2663 }
2664
2665 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2666 int *valp,
2667 int write, void *data)
2668 {
2669 if (write) {
2670 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2671 return 1;
2672 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2673 } else {
2674 int val = *valp;
2675 unsigned long lval;
2676 if (val < 0) {
2677 *negp = -1;
2678 lval = (unsigned long)-val;
2679 } else {
2680 *negp = 0;
2681 lval = (unsigned long)val;
2682 }
2683 *lvalp = jiffies_to_clock_t(lval);
2684 }
2685 return 0;
2686 }
2687
2688 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2689 int *valp,
2690 int write, void *data)
2691 {
2692 if (write) {
2693 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2694 } else {
2695 int val = *valp;
2696 unsigned long lval;
2697 if (val < 0) {
2698 *negp = -1;
2699 lval = (unsigned long)-val;
2700 } else {
2701 *negp = 0;
2702 lval = (unsigned long)val;
2703 }
2704 *lvalp = jiffies_to_msecs(lval);
2705 }
2706 return 0;
2707 }
2708
2709 /**
2710 * proc_dointvec_jiffies - read a vector of integers as seconds
2711 * @table: the sysctl table
2712 * @write: %TRUE if this is a write to the sysctl file
2713 * @filp: the file structure
2714 * @buffer: the user buffer
2715 * @lenp: the size of the user buffer
2716 * @ppos: file position
2717 *
2718 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2719 * values from/to the user buffer, treated as an ASCII string.
2720 * The values read are assumed to be in seconds, and are converted into
2721 * jiffies.
2722 *
2723 * Returns 0 on success.
2724 */
2725 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2726 void __user *buffer, size_t *lenp, loff_t *ppos)
2727 {
2728 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2729 do_proc_dointvec_jiffies_conv,NULL);
2730 }
2731
2732 /**
2733 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2734 * @table: the sysctl table
2735 * @write: %TRUE if this is a write to the sysctl file
2736 * @filp: the file structure
2737 * @buffer: the user buffer
2738 * @lenp: the size of the user buffer
2739 * @ppos: pointer to the file position
2740 *
2741 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2742 * values from/to the user buffer, treated as an ASCII string.
2743 * The values read are assumed to be in 1/USER_HZ seconds, and
2744 * are converted into jiffies.
2745 *
2746 * Returns 0 on success.
2747 */
2748 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2749 void __user *buffer, size_t *lenp, loff_t *ppos)
2750 {
2751 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2752 do_proc_dointvec_userhz_jiffies_conv,NULL);
2753 }
2754
2755 /**
2756 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2757 * @table: the sysctl table
2758 * @write: %TRUE if this is a write to the sysctl file
2759 * @filp: the file structure
2760 * @buffer: the user buffer
2761 * @lenp: the size of the user buffer
2762 * @ppos: file position
2763 * @ppos: the current position in the file
2764 *
2765 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2766 * values from/to the user buffer, treated as an ASCII string.
2767 * The values read are assumed to be in 1/1000 seconds, and
2768 * are converted into jiffies.
2769 *
2770 * Returns 0 on success.
2771 */
2772 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2773 void __user *buffer, size_t *lenp, loff_t *ppos)
2774 {
2775 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2776 do_proc_dointvec_ms_jiffies_conv, NULL);
2777 }
2778
2779 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2780 void __user *buffer, size_t *lenp, loff_t *ppos)
2781 {
2782 struct pid *new_pid;
2783 pid_t tmp;
2784 int r;
2785
2786 tmp = pid_vnr(cad_pid);
2787
2788 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2789 lenp, ppos, NULL, NULL);
2790 if (r || !write)
2791 return r;
2792
2793 new_pid = find_get_pid(tmp);
2794 if (!new_pid)
2795 return -ESRCH;
2796
2797 put_pid(xchg(&cad_pid, new_pid));
2798 return 0;
2799 }
2800
2801 #else /* CONFIG_PROC_FS */
2802
2803 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2804 void __user *buffer, size_t *lenp, loff_t *ppos)
2805 {
2806 return -ENOSYS;
2807 }
2808
2809 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2810 void __user *buffer, size_t *lenp, loff_t *ppos)
2811 {
2812 return -ENOSYS;
2813 }
2814
2815 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2816 void __user *buffer, size_t *lenp, loff_t *ppos)
2817 {
2818 return -ENOSYS;
2819 }
2820
2821 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2822 void __user *buffer, size_t *lenp, loff_t *ppos)
2823 {
2824 return -ENOSYS;
2825 }
2826
2827 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2828 void __user *buffer, size_t *lenp, loff_t *ppos)
2829 {
2830 return -ENOSYS;
2831 }
2832
2833 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2834 void __user *buffer, size_t *lenp, loff_t *ppos)
2835 {
2836 return -ENOSYS;
2837 }
2838
2839 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2840 void __user *buffer, size_t *lenp, loff_t *ppos)
2841 {
2842 return -ENOSYS;
2843 }
2844
2845 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2846 struct file *filp,
2847 void __user *buffer,
2848 size_t *lenp, loff_t *ppos)
2849 {
2850 return -ENOSYS;
2851 }
2852
2853
2854 #endif /* CONFIG_PROC_FS */
2855
2856
2857 #ifdef CONFIG_SYSCTL_SYSCALL
2858 /*
2859 * General sysctl support routines
2860 */
2861
2862 /* The generic sysctl data routine (used if no strategy routine supplied) */
2863 int sysctl_data(struct ctl_table *table,
2864 void __user *oldval, size_t __user *oldlenp,
2865 void __user *newval, size_t newlen)
2866 {
2867 size_t len;
2868
2869 /* Get out of I don't have a variable */
2870 if (!table->data || !table->maxlen)
2871 return -ENOTDIR;
2872
2873 if (oldval && oldlenp) {
2874 if (get_user(len, oldlenp))
2875 return -EFAULT;
2876 if (len) {
2877 if (len > table->maxlen)
2878 len = table->maxlen;
2879 if (copy_to_user(oldval, table->data, len))
2880 return -EFAULT;
2881 if (put_user(len, oldlenp))
2882 return -EFAULT;
2883 }
2884 }
2885
2886 if (newval && newlen) {
2887 if (newlen > table->maxlen)
2888 newlen = table->maxlen;
2889
2890 if (copy_from_user(table->data, newval, newlen))
2891 return -EFAULT;
2892 }
2893 return 1;
2894 }
2895
2896 /* The generic string strategy routine: */
2897 int sysctl_string(struct ctl_table *table,
2898 void __user *oldval, size_t __user *oldlenp,
2899 void __user *newval, size_t newlen)
2900 {
2901 if (!table->data || !table->maxlen)
2902 return -ENOTDIR;
2903
2904 if (oldval && oldlenp) {
2905 size_t bufsize;
2906 if (get_user(bufsize, oldlenp))
2907 return -EFAULT;
2908 if (bufsize) {
2909 size_t len = strlen(table->data), copied;
2910
2911 /* This shouldn't trigger for a well-formed sysctl */
2912 if (len > table->maxlen)
2913 len = table->maxlen;
2914
2915 /* Copy up to a max of bufsize-1 bytes of the string */
2916 copied = (len >= bufsize) ? bufsize - 1 : len;
2917
2918 if (copy_to_user(oldval, table->data, copied) ||
2919 put_user(0, (char __user *)(oldval + copied)))
2920 return -EFAULT;
2921 if (put_user(len, oldlenp))
2922 return -EFAULT;
2923 }
2924 }
2925 if (newval && newlen) {
2926 size_t len = newlen;
2927 if (len > table->maxlen)
2928 len = table->maxlen;
2929 if(copy_from_user(table->data, newval, len))
2930 return -EFAULT;
2931 if (len == table->maxlen)
2932 len--;
2933 ((char *) table->data)[len] = 0;
2934 }
2935 return 1;
2936 }
2937
2938 /*
2939 * This function makes sure that all of the integers in the vector
2940 * are between the minimum and maximum values given in the arrays
2941 * table->extra1 and table->extra2, respectively.
2942 */
2943 int sysctl_intvec(struct ctl_table *table,
2944 void __user *oldval, size_t __user *oldlenp,
2945 void __user *newval, size_t newlen)
2946 {
2947
2948 if (newval && newlen) {
2949 int __user *vec = (int __user *) newval;
2950 int *min = (int *) table->extra1;
2951 int *max = (int *) table->extra2;
2952 size_t length;
2953 int i;
2954
2955 if (newlen % sizeof(int) != 0)
2956 return -EINVAL;
2957
2958 if (!table->extra1 && !table->extra2)
2959 return 0;
2960
2961 if (newlen > table->maxlen)
2962 newlen = table->maxlen;
2963 length = newlen / sizeof(int);
2964
2965 for (i = 0; i < length; i++) {
2966 int value;
2967 if (get_user(value, vec + i))
2968 return -EFAULT;
2969 if (min && value < min[i])
2970 return -EINVAL;
2971 if (max && value > max[i])
2972 return -EINVAL;
2973 }
2974 }
2975 return 0;
2976 }
2977
2978 /* Strategy function to convert jiffies to seconds */
2979 int sysctl_jiffies(struct ctl_table *table,
2980 void __user *oldval, size_t __user *oldlenp,
2981 void __user *newval, size_t newlen)
2982 {
2983 if (oldval && oldlenp) {
2984 size_t olen;
2985
2986 if (get_user(olen, oldlenp))
2987 return -EFAULT;
2988 if (olen) {
2989 int val;
2990
2991 if (olen < sizeof(int))
2992 return -EINVAL;
2993
2994 val = *(int *)(table->data) / HZ;
2995 if (put_user(val, (int __user *)oldval))
2996 return -EFAULT;
2997 if (put_user(sizeof(int), oldlenp))
2998 return -EFAULT;
2999 }
3000 }
3001 if (newval && newlen) {
3002 int new;
3003 if (newlen != sizeof(int))
3004 return -EINVAL;
3005 if (get_user(new, (int __user *)newval))
3006 return -EFAULT;
3007 *(int *)(table->data) = new*HZ;
3008 }
3009 return 1;
3010 }
3011
3012 /* Strategy function to convert jiffies to seconds */
3013 int sysctl_ms_jiffies(struct ctl_table *table,
3014 void __user *oldval, size_t __user *oldlenp,
3015 void __user *newval, size_t newlen)
3016 {
3017 if (oldval && oldlenp) {
3018 size_t olen;
3019
3020 if (get_user(olen, oldlenp))
3021 return -EFAULT;
3022 if (olen) {
3023 int val;
3024
3025 if (olen < sizeof(int))
3026 return -EINVAL;
3027
3028 val = jiffies_to_msecs(*(int *)(table->data));
3029 if (put_user(val, (int __user *)oldval))
3030 return -EFAULT;
3031 if (put_user(sizeof(int), oldlenp))
3032 return -EFAULT;
3033 }
3034 }
3035 if (newval && newlen) {
3036 int new;
3037 if (newlen != sizeof(int))
3038 return -EINVAL;
3039 if (get_user(new, (int __user *)newval))
3040 return -EFAULT;
3041 *(int *)(table->data) = msecs_to_jiffies(new);
3042 }
3043 return 1;
3044 }
3045
3046
3047
3048 #else /* CONFIG_SYSCTL_SYSCALL */
3049
3050
3051 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
3052 {
3053 struct __sysctl_args tmp;
3054 int error;
3055
3056 if (copy_from_user(&tmp, args, sizeof(tmp)))
3057 return -EFAULT;
3058
3059 error = deprecated_sysctl_warning(&tmp);
3060
3061 /* If no error reading the parameters then just -ENOSYS ... */
3062 if (!error)
3063 error = -ENOSYS;
3064
3065 return error;
3066 }
3067
3068 int sysctl_data(struct ctl_table *table,
3069 void __user *oldval, size_t __user *oldlenp,
3070 void __user *newval, size_t newlen)
3071 {
3072 return -ENOSYS;
3073 }
3074
3075 int sysctl_string(struct ctl_table *table,
3076 void __user *oldval, size_t __user *oldlenp,
3077 void __user *newval, size_t newlen)
3078 {
3079 return -ENOSYS;
3080 }
3081
3082 int sysctl_intvec(struct ctl_table *table,
3083 void __user *oldval, size_t __user *oldlenp,
3084 void __user *newval, size_t newlen)
3085 {
3086 return -ENOSYS;
3087 }
3088
3089 int sysctl_jiffies(struct ctl_table *table,
3090 void __user *oldval, size_t __user *oldlenp,
3091 void __user *newval, size_t newlen)
3092 {
3093 return -ENOSYS;
3094 }
3095
3096 int sysctl_ms_jiffies(struct ctl_table *table,
3097 void __user *oldval, size_t __user *oldlenp,
3098 void __user *newval, size_t newlen)
3099 {
3100 return -ENOSYS;
3101 }
3102
3103 #endif /* CONFIG_SYSCTL_SYSCALL */
3104
3105 static int deprecated_sysctl_warning(struct __sysctl_args *args)
3106 {
3107 static int msg_count;
3108 int name[CTL_MAXNAME];
3109 int i;
3110
3111 /* Check args->nlen. */
3112 if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
3113 return -ENOTDIR;
3114
3115 /* Read in the sysctl name for better debug message logging */
3116 for (i = 0; i < args->nlen; i++)
3117 if (get_user(name[i], args->name + i))
3118 return -EFAULT;
3119
3120 /* Ignore accesses to kernel.version */
3121 if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
3122 return 0;
3123
3124 if (msg_count < 5) {
3125 msg_count++;
3126 printk(KERN_INFO
3127 "warning: process `%s' used the deprecated sysctl "
3128 "system call with ", current->comm);
3129 for (i = 0; i < args->nlen; i++)
3130 printk("%d.", name[i]);
3131 printk("\n");
3132 }
3133 return 0;
3134 }
3135
3136 /*
3137 * No sense putting this after each symbol definition, twice,
3138 * exception granted :-)
3139 */
3140 EXPORT_SYMBOL(proc_dointvec);
3141 EXPORT_SYMBOL(proc_dointvec_jiffies);
3142 EXPORT_SYMBOL(proc_dointvec_minmax);
3143 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3144 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3145 EXPORT_SYMBOL(proc_dostring);
3146 EXPORT_SYMBOL(proc_doulongvec_minmax);
3147 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3148 EXPORT_SYMBOL(register_sysctl_table);
3149 EXPORT_SYMBOL(register_sysctl_paths);
3150 EXPORT_SYMBOL(sysctl_intvec);
3151 EXPORT_SYMBOL(sysctl_jiffies);
3152 EXPORT_SYMBOL(sysctl_ms_jiffies);
3153 EXPORT_SYMBOL(sysctl_string);
3154 EXPORT_SYMBOL(sysctl_data);
3155 EXPORT_SYMBOL(unregister_sysctl_table);
ubuntu-bionic-kernel mirror