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
52 #include <asm/uaccess.h>
53 #include <asm/processor.h>
54
55 #ifdef CONFIG_X86
56 #include <asm/nmi.h>
57 #include <asm/stacktrace.h>
58 #include <asm/io.h>
59 #endif
60
61 static int deprecated_sysctl_warning(struct __sysctl_args *args);
62
63 #if defined(CONFIG_SYSCTL)
64
65 /* External variables not in a header file. */
66 extern int C_A_D;
67 extern int print_fatal_signals;
68 extern int sysctl_overcommit_memory;
69 extern int sysctl_overcommit_ratio;
70 extern int sysctl_panic_on_oom;
71 extern int sysctl_oom_kill_allocating_task;
72 extern int sysctl_oom_dump_tasks;
73 extern int max_threads;
74 extern int core_uses_pid;
75 extern int suid_dumpable;
76 extern char core_pattern[];
77 extern int pid_max;
78 extern int min_free_kbytes;
79 extern int min_free_order_shift;
80 extern int pid_max_min, pid_max_max;
81 extern int sysctl_drop_caches;
82 extern int percpu_pagelist_fraction;
83 extern int compat_log;
84 extern int latencytop_enabled;
85 extern int sysctl_nr_open_min, sysctl_nr_open_max;
86 #ifndef CONFIG_MMU
87 extern int sysctl_nr_trim_pages;
88 #endif
89 #ifdef CONFIG_RCU_TORTURE_TEST
90 extern int rcutorture_runnable;
91 #endif /* #ifdef CONFIG_RCU_TORTURE_TEST */
92
93 /* Constants used for minimum and maximum */
94 #ifdef CONFIG_DETECT_SOFTLOCKUP
95 static int sixty = 60;
96 static int neg_one = -1;
97 #endif
98
99 #if defined(CONFIG_MMU) && defined(CONFIG_FILE_LOCKING)
100 static int two = 2;
101 #endif
102
103 static int zero;
104 static int one = 1;
105 static unsigned long one_ul = 1;
106 static int one_hundred = 100;
107
108 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
109 static int maxolduid = 65535;
110 static int minolduid;
111 static int min_percpu_pagelist_fract = 8;
112
113 static int ngroups_max = NGROUPS_MAX;
114
115 #ifdef CONFIG_MODULES
116 extern char modprobe_path[];
117 #endif
118 #ifdef CONFIG_CHR_DEV_SG
119 extern int sg_big_buff;
120 #endif
121
122 #ifdef CONFIG_SPARC
123 #include <asm/system.h>
124 #endif
125
126 #ifdef CONFIG_SPARC64
127 extern int sysctl_tsb_ratio;
128 #endif
129
130 #ifdef __hppa__
131 extern int pwrsw_enabled;
132 extern int unaligned_enabled;
133 #endif
134
135 #ifdef CONFIG_S390
136 #ifdef CONFIG_MATHEMU
137 extern int sysctl_ieee_emulation_warnings;
138 #endif
139 extern int sysctl_userprocess_debug;
140 extern int spin_retry;
141 #endif
142
143 #ifdef CONFIG_BSD_PROCESS_ACCT
144 extern int acct_parm[];
145 #endif
146
147 #ifdef CONFIG_IA64
148 extern int no_unaligned_warning;
149 extern int unaligned_dump_stack;
150 #endif
151
152 #ifdef CONFIG_RT_MUTEXES
153 extern int max_lock_depth;
154 #endif
155
156 #ifdef CONFIG_PROC_SYSCTL
157 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
158 void __user *buffer, size_t *lenp, loff_t *ppos);
159 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
160 void __user *buffer, size_t *lenp, loff_t *ppos);
161 #endif
162
163 static struct ctl_table root_table[];
164 static struct ctl_table_root sysctl_table_root;
165 static struct ctl_table_header root_table_header = {
166 .count = 1,
167 .ctl_table = root_table,
168 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
169 .root = &sysctl_table_root,
170 .set = &sysctl_table_root.default_set,
171 };
172 static struct ctl_table_root sysctl_table_root = {
173 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
174 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
175 };
176
177 static struct ctl_table kern_table[];
178 static struct ctl_table vm_table[];
179 static struct ctl_table fs_table[];
180 static struct ctl_table debug_table[];
181 static struct ctl_table dev_table[];
182 extern struct ctl_table random_table[];
183 #ifdef CONFIG_INOTIFY_USER
184 extern struct ctl_table inotify_table[];
185 #endif
186 #ifdef CONFIG_EPOLL
187 extern struct ctl_table epoll_table[];
188 #endif
189
190 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
191 int sysctl_legacy_va_layout;
192 #endif
193
194 extern int prove_locking;
195 extern int lock_stat;
196
197 /* The default sysctl tables: */
198
199 static struct ctl_table root_table[] = {
200 {
201 .ctl_name = CTL_KERN,
202 .procname = "kernel",
203 .mode = 0555,
204 .child = kern_table,
205 },
206 {
207 .ctl_name = CTL_VM,
208 .procname = "vm",
209 .mode = 0555,
210 .child = vm_table,
211 },
212 {
213 .ctl_name = CTL_FS,
214 .procname = "fs",
215 .mode = 0555,
216 .child = fs_table,
217 },
218 {
219 .ctl_name = CTL_DEBUG,
220 .procname = "debug",
221 .mode = 0555,
222 .child = debug_table,
223 },
224 {
225 .ctl_name = CTL_DEV,
226 .procname = "dev",
227 .mode = 0555,
228 .child = dev_table,
229 },
230 /*
231 * NOTE: do not add new entries to this table unless you have read
232 * Documentation/sysctl/ctl_unnumbered.txt
233 */
234 { .ctl_name = 0 }
235 };
236
237 #ifdef CONFIG_SCHED_DEBUG
238 static int min_sched_granularity_ns = 100000; /* 100 usecs */
239 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
240 static int min_wakeup_granularity_ns; /* 0 usecs */
241 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
242 #endif
243
244 static struct ctl_table kern_table[] = {
245 #ifdef CONFIG_SCHED_DEBUG
246 {
247 .ctl_name = CTL_UNNUMBERED,
248 .procname = "sched_min_granularity_ns",
249 .data = &sysctl_sched_min_granularity,
250 .maxlen = sizeof(unsigned int),
251 .mode = 0644,
252 .proc_handler = &sched_nr_latency_handler,
253 .strategy = &sysctl_intvec,
254 .extra1 = &min_sched_granularity_ns,
255 .extra2 = &max_sched_granularity_ns,
256 },
257 {
258 .ctl_name = CTL_UNNUMBERED,
259 .procname = "sched_latency_ns",
260 .data = &sysctl_sched_latency,
261 .maxlen = sizeof(unsigned int),
262 .mode = 0644,
263 .proc_handler = &sched_nr_latency_handler,
264 .strategy = &sysctl_intvec,
265 .extra1 = &min_sched_granularity_ns,
266 .extra2 = &max_sched_granularity_ns,
267 },
268 {
269 .ctl_name = CTL_UNNUMBERED,
270 .procname = "sched_wakeup_granularity_ns",
271 .data = &sysctl_sched_wakeup_granularity,
272 .maxlen = sizeof(unsigned int),
273 .mode = 0644,
274 .proc_handler = &proc_dointvec_minmax,
275 .strategy = &sysctl_intvec,
276 .extra1 = &min_wakeup_granularity_ns,
277 .extra2 = &max_wakeup_granularity_ns,
278 },
279 {
280 .ctl_name = CTL_UNNUMBERED,
281 .procname = "sched_shares_ratelimit",
282 .data = &sysctl_sched_shares_ratelimit,
283 .maxlen = sizeof(unsigned int),
284 .mode = 0644,
285 .proc_handler = &proc_dointvec,
286 },
287 {
288 .ctl_name = CTL_UNNUMBERED,
289 .procname = "sched_shares_thresh",
290 .data = &sysctl_sched_shares_thresh,
291 .maxlen = sizeof(unsigned int),
292 .mode = 0644,
293 .proc_handler = &proc_dointvec_minmax,
294 .strategy = &sysctl_intvec,
295 .extra1 = &zero,
296 },
297 {
298 .ctl_name = CTL_UNNUMBERED,
299 .procname = "sched_child_runs_first",
300 .data = &sysctl_sched_child_runs_first,
301 .maxlen = sizeof(unsigned int),
302 .mode = 0644,
303 .proc_handler = &proc_dointvec,
304 },
305 {
306 .ctl_name = CTL_UNNUMBERED,
307 .procname = "sched_features",
308 .data = &sysctl_sched_features,
309 .maxlen = sizeof(unsigned int),
310 .mode = 0644,
311 .proc_handler = &proc_dointvec,
312 },
313 {
314 .ctl_name = CTL_UNNUMBERED,
315 .procname = "sched_migration_cost",
316 .data = &sysctl_sched_migration_cost,
317 .maxlen = sizeof(unsigned int),
318 .mode = 0644,
319 .proc_handler = &proc_dointvec,
320 },
321 {
322 .ctl_name = CTL_UNNUMBERED,
323 .procname = "sched_nr_migrate",
324 .data = &sysctl_sched_nr_migrate,
325 .maxlen = sizeof(unsigned int),
326 .mode = 0644,
327 .proc_handler = &proc_dointvec,
328 },
329 #endif
330 {
331 .ctl_name = CTL_UNNUMBERED,
332 .procname = "sched_rt_period_us",
333 .data = &sysctl_sched_rt_period,
334 .maxlen = sizeof(unsigned int),
335 .mode = 0644,
336 .proc_handler = &sched_rt_handler,
337 },
338 {
339 .ctl_name = CTL_UNNUMBERED,
340 .procname = "sched_rt_runtime_us",
341 .data = &sysctl_sched_rt_runtime,
342 .maxlen = sizeof(int),
343 .mode = 0644,
344 .proc_handler = &sched_rt_handler,
345 },
346 {
347 .ctl_name = CTL_UNNUMBERED,
348 .procname = "sched_compat_yield",
349 .data = &sysctl_sched_compat_yield,
350 .maxlen = sizeof(unsigned int),
351 .mode = 0644,
352 .proc_handler = &proc_dointvec,
353 },
354 #ifdef CONFIG_PROVE_LOCKING
355 {
356 .ctl_name = CTL_UNNUMBERED,
357 .procname = "prove_locking",
358 .data = &prove_locking,
359 .maxlen = sizeof(int),
360 .mode = 0644,
361 .proc_handler = &proc_dointvec,
362 },
363 #endif
364 #ifdef CONFIG_LOCK_STAT
365 {
366 .ctl_name = CTL_UNNUMBERED,
367 .procname = "lock_stat",
368 .data = &lock_stat,
369 .maxlen = sizeof(int),
370 .mode = 0644,
371 .proc_handler = &proc_dointvec,
372 },
373 #endif
374 {
375 .ctl_name = KERN_PANIC,
376 .procname = "panic",
377 .data = &panic_timeout,
378 .maxlen = sizeof(int),
379 .mode = 0644,
380 .proc_handler = &proc_dointvec,
381 },
382 {
383 .ctl_name = KERN_CORE_USES_PID,
384 .procname = "core_uses_pid",
385 .data = &core_uses_pid,
386 .maxlen = sizeof(int),
387 .mode = 0644,
388 .proc_handler = &proc_dointvec,
389 },
390 {
391 .ctl_name = KERN_CORE_PATTERN,
392 .procname = "core_pattern",
393 .data = core_pattern,
394 .maxlen = CORENAME_MAX_SIZE,
395 .mode = 0644,
396 .proc_handler = &proc_dostring,
397 .strategy = &sysctl_string,
398 },
399 #ifdef CONFIG_PROC_SYSCTL
400 {
401 .procname = "tainted",
402 .maxlen = sizeof(long),
403 .mode = 0644,
404 .proc_handler = &proc_taint,
405 },
406 #endif
407 #ifdef CONFIG_LATENCYTOP
408 {
409 .procname = "latencytop",
410 .data = &latencytop_enabled,
411 .maxlen = sizeof(int),
412 .mode = 0644,
413 .proc_handler = &proc_dointvec,
414 },
415 #endif
416 #ifdef CONFIG_BLK_DEV_INITRD
417 {
418 .ctl_name = KERN_REALROOTDEV,
419 .procname = "real-root-dev",
420 .data = &real_root_dev,
421 .maxlen = sizeof(int),
422 .mode = 0644,
423 .proc_handler = &proc_dointvec,
424 },
425 #endif
426 {
427 .ctl_name = CTL_UNNUMBERED,
428 .procname = "print-fatal-signals",
429 .data = &print_fatal_signals,
430 .maxlen = sizeof(int),
431 .mode = 0644,
432 .proc_handler = &proc_dointvec,
433 },
434 #ifdef CONFIG_SPARC
435 {
436 .ctl_name = KERN_SPARC_REBOOT,
437 .procname = "reboot-cmd",
438 .data = reboot_command,
439 .maxlen = 256,
440 .mode = 0644,
441 .proc_handler = &proc_dostring,
442 .strategy = &sysctl_string,
443 },
444 {
445 .ctl_name = KERN_SPARC_STOP_A,
446 .procname = "stop-a",
447 .data = &stop_a_enabled,
448 .maxlen = sizeof (int),
449 .mode = 0644,
450 .proc_handler = &proc_dointvec,
451 },
452 {
453 .ctl_name = KERN_SPARC_SCONS_PWROFF,
454 .procname = "scons-poweroff",
455 .data = &scons_pwroff,
456 .maxlen = sizeof (int),
457 .mode = 0644,
458 .proc_handler = &proc_dointvec,
459 },
460 #endif
461 #ifdef CONFIG_SPARC64
462 {
463 .ctl_name = CTL_UNNUMBERED,
464 .procname = "tsb-ratio",
465 .data = &sysctl_tsb_ratio,
466 .maxlen = sizeof (int),
467 .mode = 0644,
468 .proc_handler = &proc_dointvec,
469 },
470 #endif
471 #ifdef __hppa__
472 {
473 .ctl_name = KERN_HPPA_PWRSW,
474 .procname = "soft-power",
475 .data = &pwrsw_enabled,
476 .maxlen = sizeof (int),
477 .mode = 0644,
478 .proc_handler = &proc_dointvec,
479 },
480 {
481 .ctl_name = KERN_HPPA_UNALIGNED,
482 .procname = "unaligned-trap",
483 .data = &unaligned_enabled,
484 .maxlen = sizeof (int),
485 .mode = 0644,
486 .proc_handler = &proc_dointvec,
487 },
488 #endif
489 {
490 .ctl_name = KERN_CTLALTDEL,
491 .procname = "ctrl-alt-del",
492 .data = &C_A_D,
493 .maxlen = sizeof(int),
494 .mode = 0644,
495 .proc_handler = &proc_dointvec,
496 },
497 #ifdef CONFIG_FUNCTION_TRACER
498 {
499 .ctl_name = CTL_UNNUMBERED,
500 .procname = "ftrace_enabled",
501 .data = &ftrace_enabled,
502 .maxlen = sizeof(int),
503 .mode = 0644,
504 .proc_handler = &ftrace_enable_sysctl,
505 },
506 #endif
507 #ifdef CONFIG_STACK_TRACER
508 {
509 .ctl_name = CTL_UNNUMBERED,
510 .procname = "stack_tracer_enabled",
511 .data = &stack_tracer_enabled,
512 .maxlen = sizeof(int),
513 .mode = 0644,
514 .proc_handler = &stack_trace_sysctl,
515 },
516 #endif
517 #ifdef CONFIG_TRACING
518 {
519 .ctl_name = CTL_UNNUMBERED,
520 .procname = "ftrace_dump_on_oops",
521 .data = &ftrace_dump_on_oops,
522 .maxlen = sizeof(int),
523 .mode = 0644,
524 .proc_handler = &proc_dointvec,
525 },
526 #endif
527 #ifdef CONFIG_MODULES
528 {
529 .ctl_name = KERN_MODPROBE,
530 .procname = "modprobe",
531 .data = &modprobe_path,
532 .maxlen = KMOD_PATH_LEN,
533 .mode = 0644,
534 .proc_handler = &proc_dostring,
535 .strategy = &sysctl_string,
536 },
537 #endif
538 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
539 {
540 .ctl_name = KERN_HOTPLUG,
541 .procname = "hotplug",
542 .data = &uevent_helper,
543 .maxlen = UEVENT_HELPER_PATH_LEN,
544 .mode = 0644,
545 .proc_handler = &proc_dostring,
546 .strategy = &sysctl_string,
547 },
548 #endif
549 #ifdef CONFIG_CHR_DEV_SG
550 {
551 .ctl_name = KERN_SG_BIG_BUFF,
552 .procname = "sg-big-buff",
553 .data = &sg_big_buff,
554 .maxlen = sizeof (int),
555 .mode = 0444,
556 .proc_handler = &proc_dointvec,
557 },
558 #endif
559 #ifdef CONFIG_BSD_PROCESS_ACCT
560 {
561 .ctl_name = KERN_ACCT,
562 .procname = "acct",
563 .data = &acct_parm,
564 .maxlen = 3*sizeof(int),
565 .mode = 0644,
566 .proc_handler = &proc_dointvec,
567 },
568 #endif
569 #ifdef CONFIG_MAGIC_SYSRQ
570 {
571 .ctl_name = KERN_SYSRQ,
572 .procname = "sysrq",
573 .data = &__sysrq_enabled,
574 .maxlen = sizeof (int),
575 .mode = 0644,
576 .proc_handler = &proc_dointvec,
577 },
578 #endif
579 #ifdef CONFIG_PROC_SYSCTL
580 {
581 .procname = "cad_pid",
582 .data = NULL,
583 .maxlen = sizeof (int),
584 .mode = 0600,
585 .proc_handler = &proc_do_cad_pid,
586 },
587 #endif
588 {
589 .ctl_name = KERN_MAX_THREADS,
590 .procname = "threads-max",
591 .data = &max_threads,
592 .maxlen = sizeof(int),
593 .mode = 0644,
594 .proc_handler = &proc_dointvec,
595 },
596 {
597 .ctl_name = KERN_RANDOM,
598 .procname = "random",
599 .mode = 0555,
600 .child = random_table,
601 },
602 {
603 .ctl_name = KERN_OVERFLOWUID,
604 .procname = "overflowuid",
605 .data = &overflowuid,
606 .maxlen = sizeof(int),
607 .mode = 0644,
608 .proc_handler = &proc_dointvec_minmax,
609 .strategy = &sysctl_intvec,
610 .extra1 = &minolduid,
611 .extra2 = &maxolduid,
612 },
613 {
614 .ctl_name = KERN_OVERFLOWGID,
615 .procname = "overflowgid",
616 .data = &overflowgid,
617 .maxlen = sizeof(int),
618 .mode = 0644,
619 .proc_handler = &proc_dointvec_minmax,
620 .strategy = &sysctl_intvec,
621 .extra1 = &minolduid,
622 .extra2 = &maxolduid,
623 },
624 #ifdef CONFIG_S390
625 #ifdef CONFIG_MATHEMU
626 {
627 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
628 .procname = "ieee_emulation_warnings",
629 .data = &sysctl_ieee_emulation_warnings,
630 .maxlen = sizeof(int),
631 .mode = 0644,
632 .proc_handler = &proc_dointvec,
633 },
634 #endif
635 {
636 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
637 .procname = "userprocess_debug",
638 .data = &sysctl_userprocess_debug,
639 .maxlen = sizeof(int),
640 .mode = 0644,
641 .proc_handler = &proc_dointvec,
642 },
643 #endif
644 {
645 .ctl_name = KERN_PIDMAX,
646 .procname = "pid_max",
647 .data = &pid_max,
648 .maxlen = sizeof (int),
649 .mode = 0644,
650 .proc_handler = &proc_dointvec_minmax,
651 .strategy = sysctl_intvec,
652 .extra1 = &pid_max_min,
653 .extra2 = &pid_max_max,
654 },
655 {
656 .ctl_name = KERN_PANIC_ON_OOPS,
657 .procname = "panic_on_oops",
658 .data = &panic_on_oops,
659 .maxlen = sizeof(int),
660 .mode = 0644,
661 .proc_handler = &proc_dointvec,
662 },
663 #if defined CONFIG_PRINTK
664 {
665 .ctl_name = KERN_PRINTK,
666 .procname = "printk",
667 .data = &console_loglevel,
668 .maxlen = 4*sizeof(int),
669 .mode = 0644,
670 .proc_handler = &proc_dointvec,
671 },
672 {
673 .ctl_name = KERN_PRINTK_RATELIMIT,
674 .procname = "printk_ratelimit",
675 .data = &printk_ratelimit_state.interval,
676 .maxlen = sizeof(int),
677 .mode = 0644,
678 .proc_handler = &proc_dointvec_jiffies,
679 .strategy = &sysctl_jiffies,
680 },
681 {
682 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
683 .procname = "printk_ratelimit_burst",
684 .data = &printk_ratelimit_state.burst,
685 .maxlen = sizeof(int),
686 .mode = 0644,
687 .proc_handler = &proc_dointvec,
688 },
689 #endif
690 {
691 .ctl_name = KERN_NGROUPS_MAX,
692 .procname = "ngroups_max",
693 .data = &ngroups_max,
694 .maxlen = sizeof (int),
695 .mode = 0444,
696 .proc_handler = &proc_dointvec,
697 },
698 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
699 {
700 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
701 .procname = "unknown_nmi_panic",
702 .data = &unknown_nmi_panic,
703 .maxlen = sizeof (int),
704 .mode = 0644,
705 .proc_handler = &proc_dointvec,
706 },
707 {
708 .procname = "nmi_watchdog",
709 .data = &nmi_watchdog_enabled,
710 .maxlen = sizeof (int),
711 .mode = 0644,
712 .proc_handler = &proc_nmi_enabled,
713 },
714 #endif
715 #if defined(CONFIG_X86)
716 {
717 .ctl_name = KERN_PANIC_ON_NMI,
718 .procname = "panic_on_unrecovered_nmi",
719 .data = &panic_on_unrecovered_nmi,
720 .maxlen = sizeof(int),
721 .mode = 0644,
722 .proc_handler = &proc_dointvec,
723 },
724 {
725 .ctl_name = KERN_BOOTLOADER_TYPE,
726 .procname = "bootloader_type",
727 .data = &bootloader_type,
728 .maxlen = sizeof (int),
729 .mode = 0444,
730 .proc_handler = &proc_dointvec,
731 },
732 {
733 .ctl_name = CTL_UNNUMBERED,
734 .procname = "kstack_depth_to_print",
735 .data = &kstack_depth_to_print,
736 .maxlen = sizeof(int),
737 .mode = 0644,
738 .proc_handler = &proc_dointvec,
739 },
740 {
741 .ctl_name = CTL_UNNUMBERED,
742 .procname = "io_delay_type",
743 .data = &io_delay_type,
744 .maxlen = sizeof(int),
745 .mode = 0644,
746 .proc_handler = &proc_dointvec,
747 },
748 #endif
749 #if defined(CONFIG_MMU)
750 {
751 .ctl_name = KERN_RANDOMIZE,
752 .procname = "randomize_va_space",
753 .data = &randomize_va_space,
754 .maxlen = sizeof(int),
755 .mode = 0644,
756 .proc_handler = &proc_dointvec,
757 },
758 #endif
759 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
760 {
761 .ctl_name = KERN_SPIN_RETRY,
762 .procname = "spin_retry",
763 .data = &spin_retry,
764 .maxlen = sizeof (int),
765 .mode = 0644,
766 .proc_handler = &proc_dointvec,
767 },
768 #endif
769 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
770 {
771 .procname = "acpi_video_flags",
772 .data = &acpi_realmode_flags,
773 .maxlen = sizeof (unsigned long),
774 .mode = 0644,
775 .proc_handler = &proc_doulongvec_minmax,
776 },
777 #endif
778 #ifdef CONFIG_IA64
779 {
780 .ctl_name = KERN_IA64_UNALIGNED,
781 .procname = "ignore-unaligned-usertrap",
782 .data = &no_unaligned_warning,
783 .maxlen = sizeof (int),
784 .mode = 0644,
785 .proc_handler = &proc_dointvec,
786 },
787 {
788 .ctl_name = CTL_UNNUMBERED,
789 .procname = "unaligned-dump-stack",
790 .data = &unaligned_dump_stack,
791 .maxlen = sizeof (int),
792 .mode = 0644,
793 .proc_handler = &proc_dointvec,
794 },
795 #endif
796 #ifdef CONFIG_DETECT_SOFTLOCKUP
797 {
798 .ctl_name = CTL_UNNUMBERED,
799 .procname = "softlockup_panic",
800 .data = &softlockup_panic,
801 .maxlen = sizeof(int),
802 .mode = 0644,
803 .proc_handler = &proc_dointvec_minmax,
804 .strategy = &sysctl_intvec,
805 .extra1 = &zero,
806 .extra2 = &one,
807 },
808 {
809 .ctl_name = CTL_UNNUMBERED,
810 .procname = "softlockup_thresh",
811 .data = &softlockup_thresh,
812 .maxlen = sizeof(int),
813 .mode = 0644,
814 .proc_handler = &proc_dosoftlockup_thresh,
815 .strategy = &sysctl_intvec,
816 .extra1 = &neg_one,
817 .extra2 = &sixty,
818 },
819 {
820 .ctl_name = CTL_UNNUMBERED,
821 .procname = "hung_task_check_count",
822 .data = &sysctl_hung_task_check_count,
823 .maxlen = sizeof(unsigned long),
824 .mode = 0644,
825 .proc_handler = &proc_doulongvec_minmax,
826 .strategy = &sysctl_intvec,
827 },
828 {
829 .ctl_name = CTL_UNNUMBERED,
830 .procname = "hung_task_timeout_secs",
831 .data = &sysctl_hung_task_timeout_secs,
832 .maxlen = sizeof(unsigned long),
833 .mode = 0644,
834 .proc_handler = &proc_doulongvec_minmax,
835 .strategy = &sysctl_intvec,
836 },
837 {
838 .ctl_name = CTL_UNNUMBERED,
839 .procname = "hung_task_warnings",
840 .data = &sysctl_hung_task_warnings,
841 .maxlen = sizeof(unsigned long),
842 .mode = 0644,
843 .proc_handler = &proc_doulongvec_minmax,
844 .strategy = &sysctl_intvec,
845 },
846 #endif
847 #ifdef CONFIG_COMPAT
848 {
849 .ctl_name = KERN_COMPAT_LOG,
850 .procname = "compat-log",
851 .data = &compat_log,
852 .maxlen = sizeof (int),
853 .mode = 0644,
854 .proc_handler = &proc_dointvec,
855 },
856 #endif
857 #ifdef CONFIG_RT_MUTEXES
858 {
859 .ctl_name = KERN_MAX_LOCK_DEPTH,
860 .procname = "max_lock_depth",
861 .data = &max_lock_depth,
862 .maxlen = sizeof(int),
863 .mode = 0644,
864 .proc_handler = &proc_dointvec,
865 },
866 #endif
867 {
868 .ctl_name = CTL_UNNUMBERED,
869 .procname = "poweroff_cmd",
870 .data = &poweroff_cmd,
871 .maxlen = POWEROFF_CMD_PATH_LEN,
872 .mode = 0644,
873 .proc_handler = &proc_dostring,
874 .strategy = &sysctl_string,
875 },
876 #ifdef CONFIG_KEYS
877 {
878 .ctl_name = CTL_UNNUMBERED,
879 .procname = "keys",
880 .mode = 0555,
881 .child = key_sysctls,
882 },
883 #endif
884 #ifdef CONFIG_RCU_TORTURE_TEST
885 {
886 .ctl_name = CTL_UNNUMBERED,
887 .procname = "rcutorture_runnable",
888 .data = &rcutorture_runnable,
889 .maxlen = sizeof(int),
890 .mode = 0644,
891 .proc_handler = &proc_dointvec,
892 },
893 #endif
894 #ifdef CONFIG_UNEVICTABLE_LRU
895 {
896 .ctl_name = CTL_UNNUMBERED,
897 .procname = "scan_unevictable_pages",
898 .data = &scan_unevictable_pages,
899 .maxlen = sizeof(scan_unevictable_pages),
900 .mode = 0644,
901 .proc_handler = &scan_unevictable_handler,
902 },
903 #endif
904 /*
905 * NOTE: do not add new entries to this table unless you have read
906 * Documentation/sysctl/ctl_unnumbered.txt
907 */
908 { .ctl_name = 0 }
909 };
910
911 static struct ctl_table vm_table[] = {
912 {
913 .ctl_name = VM_OVERCOMMIT_MEMORY,
914 .procname = "overcommit_memory",
915 .data = &sysctl_overcommit_memory,
916 .maxlen = sizeof(sysctl_overcommit_memory),
917 .mode = 0644,
918 .proc_handler = &proc_dointvec,
919 },
920 {
921 .ctl_name = VM_PANIC_ON_OOM,
922 .procname = "panic_on_oom",
923 .data = &sysctl_panic_on_oom,
924 .maxlen = sizeof(sysctl_panic_on_oom),
925 .mode = 0644,
926 .proc_handler = &proc_dointvec,
927 },
928 {
929 .ctl_name = CTL_UNNUMBERED,
930 .procname = "oom_kill_allocating_task",
931 .data = &sysctl_oom_kill_allocating_task,
932 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
933 .mode = 0644,
934 .proc_handler = &proc_dointvec,
935 },
936 {
937 .ctl_name = CTL_UNNUMBERED,
938 .procname = "oom_dump_tasks",
939 .data = &sysctl_oom_dump_tasks,
940 .maxlen = sizeof(sysctl_oom_dump_tasks),
941 .mode = 0644,
942 .proc_handler = &proc_dointvec,
943 },
944 {
945 .ctl_name = VM_OVERCOMMIT_RATIO,
946 .procname = "overcommit_ratio",
947 .data = &sysctl_overcommit_ratio,
948 .maxlen = sizeof(sysctl_overcommit_ratio),
949 .mode = 0644,
950 .proc_handler = &proc_dointvec,
951 },
952 {
953 .ctl_name = VM_PAGE_CLUSTER,
954 .procname = "page-cluster",
955 .data = &page_cluster,
956 .maxlen = sizeof(int),
957 .mode = 0644,
958 .proc_handler = &proc_dointvec,
959 },
960 {
961 .ctl_name = VM_DIRTY_BACKGROUND,
962 .procname = "dirty_background_ratio",
963 .data = &dirty_background_ratio,
964 .maxlen = sizeof(dirty_background_ratio),
965 .mode = 0644,
966 .proc_handler = &dirty_background_ratio_handler,
967 .strategy = &sysctl_intvec,
968 .extra1 = &zero,
969 .extra2 = &one_hundred,
970 },
971 {
972 .ctl_name = CTL_UNNUMBERED,
973 .procname = "dirty_background_bytes",
974 .data = &dirty_background_bytes,
975 .maxlen = sizeof(dirty_background_bytes),
976 .mode = 0644,
977 .proc_handler = &dirty_background_bytes_handler,
978 .strategy = &sysctl_intvec,
979 .extra1 = &one_ul,
980 },
981 {
982 .ctl_name = VM_DIRTY_RATIO,
983 .procname = "dirty_ratio",
984 .data = &vm_dirty_ratio,
985 .maxlen = sizeof(vm_dirty_ratio),
986 .mode = 0644,
987 .proc_handler = &dirty_ratio_handler,
988 .strategy = &sysctl_intvec,
989 .extra1 = &zero,
990 .extra2 = &one_hundred,
991 },
992 {
993 .ctl_name = CTL_UNNUMBERED,
994 .procname = "dirty_bytes",
995 .data = &vm_dirty_bytes,
996 .maxlen = sizeof(vm_dirty_bytes),
997 .mode = 0644,
998 .proc_handler = &dirty_bytes_handler,
999 .strategy = &sysctl_intvec,
1000 .extra1 = &one_ul,
1001 },
1002 {
1003 .procname = "dirty_writeback_centisecs",
1004 .data = &dirty_writeback_interval,
1005 .maxlen = sizeof(dirty_writeback_interval),
1006 .mode = 0644,
1007 .proc_handler = &dirty_writeback_centisecs_handler,
1008 },
1009 {
1010 .procname = "dirty_expire_centisecs",
1011 .data = &dirty_expire_interval,
1012 .maxlen = sizeof(dirty_expire_interval),
1013 .mode = 0644,
1014 .proc_handler = &proc_dointvec_userhz_jiffies,
1015 },
1016 {
1017 .ctl_name = VM_NR_PDFLUSH_THREADS,
1018 .procname = "nr_pdflush_threads",
1019 .data = &nr_pdflush_threads,
1020 .maxlen = sizeof nr_pdflush_threads,
1021 .mode = 0444 /* read-only*/,
1022 .proc_handler = &proc_dointvec,
1023 },
1024 {
1025 .ctl_name = VM_SWAPPINESS,
1026 .procname = "swappiness",
1027 .data = &vm_swappiness,
1028 .maxlen = sizeof(vm_swappiness),
1029 .mode = 0644,
1030 .proc_handler = &proc_dointvec_minmax,
1031 .strategy = &sysctl_intvec,
1032 .extra1 = &zero,
1033 .extra2 = &one_hundred,
1034 },
1035 #ifdef CONFIG_HUGETLB_PAGE
1036 {
1037 .procname = "nr_hugepages",
1038 .data = NULL,
1039 .maxlen = sizeof(unsigned long),
1040 .mode = 0644,
1041 .proc_handler = &hugetlb_sysctl_handler,
1042 .extra1 = (void *)&hugetlb_zero,
1043 .extra2 = (void *)&hugetlb_infinity,
1044 },
1045 {
1046 .ctl_name = VM_HUGETLB_GROUP,
1047 .procname = "hugetlb_shm_group",
1048 .data = &sysctl_hugetlb_shm_group,
1049 .maxlen = sizeof(gid_t),
1050 .mode = 0644,
1051 .proc_handler = &proc_dointvec,
1052 },
1053 {
1054 .ctl_name = CTL_UNNUMBERED,
1055 .procname = "hugepages_treat_as_movable",
1056 .data = &hugepages_treat_as_movable,
1057 .maxlen = sizeof(int),
1058 .mode = 0644,
1059 .proc_handler = &hugetlb_treat_movable_handler,
1060 },
1061 {
1062 .ctl_name = CTL_UNNUMBERED,
1063 .procname = "nr_overcommit_hugepages",
1064 .data = NULL,
1065 .maxlen = sizeof(unsigned long),
1066 .mode = 0644,
1067 .proc_handler = &hugetlb_overcommit_handler,
1068 .extra1 = (void *)&hugetlb_zero,
1069 .extra2 = (void *)&hugetlb_infinity,
1070 },
1071 #endif
1072 {
1073 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
1074 .procname = "lowmem_reserve_ratio",
1075 .data = &sysctl_lowmem_reserve_ratio,
1076 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1077 .mode = 0644,
1078 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
1079 .strategy = &sysctl_intvec,
1080 },
1081 {
1082 .ctl_name = VM_DROP_PAGECACHE,
1083 .procname = "drop_caches",
1084 .data = &sysctl_drop_caches,
1085 .maxlen = sizeof(int),
1086 .mode = 0644,
1087 .proc_handler = drop_caches_sysctl_handler,
1088 .strategy = &sysctl_intvec,
1089 },
1090 {
1091 .ctl_name = VM_MIN_FREE_KBYTES,
1092 .procname = "min_free_kbytes",
1093 .data = &min_free_kbytes,
1094 .maxlen = sizeof(min_free_kbytes),
1095 .mode = 0644,
1096 .proc_handler = &min_free_kbytes_sysctl_handler,
1097 .strategy = &sysctl_intvec,
1098 .extra1 = &zero,
1099 },
1100 {
1101 .ctl_name = CTL_UNNUMBERED,
1102 .procname = "min_free_order_shift",
1103 .data = &min_free_order_shift,
1104 .maxlen = sizeof(min_free_order_shift),
1105 .mode = 0644,
1106 .proc_handler = &proc_dointvec
1107 },
1108 {
1109 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
1110 .procname = "percpu_pagelist_fraction",
1111 .data = &percpu_pagelist_fraction,
1112 .maxlen = sizeof(percpu_pagelist_fraction),
1113 .mode = 0644,
1114 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
1115 .strategy = &sysctl_intvec,
1116 .extra1 = &min_percpu_pagelist_fract,
1117 },
1118 #ifdef CONFIG_MMU
1119 {
1120 .ctl_name = VM_MAX_MAP_COUNT,
1121 .procname = "max_map_count",
1122 .data = &sysctl_max_map_count,
1123 .maxlen = sizeof(sysctl_max_map_count),
1124 .mode = 0644,
1125 .proc_handler = &proc_dointvec
1126 },
1127 #else
1128 {
1129 .ctl_name = CTL_UNNUMBERED,
1130 .procname = "nr_trim_pages",
1131 .data = &sysctl_nr_trim_pages,
1132 .maxlen = sizeof(sysctl_nr_trim_pages),
1133 .mode = 0644,
1134 .proc_handler = &proc_dointvec_minmax,
1135 .strategy = &sysctl_intvec,
1136 .extra1 = &zero,
1137 },
1138 #endif
1139 {
1140 .ctl_name = VM_LAPTOP_MODE,
1141 .procname = "laptop_mode",
1142 .data = &laptop_mode,
1143 .maxlen = sizeof(laptop_mode),
1144 .mode = 0644,
1145 .proc_handler = &proc_dointvec_jiffies,
1146 .strategy = &sysctl_jiffies,
1147 },
1148 {
1149 .ctl_name = VM_BLOCK_DUMP,
1150 .procname = "block_dump",
1151 .data = &block_dump,
1152 .maxlen = sizeof(block_dump),
1153 .mode = 0644,
1154 .proc_handler = &proc_dointvec,
1155 .strategy = &sysctl_intvec,
1156 .extra1 = &zero,
1157 },
1158 {
1159 .ctl_name = VM_VFS_CACHE_PRESSURE,
1160 .procname = "vfs_cache_pressure",
1161 .data = &sysctl_vfs_cache_pressure,
1162 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1163 .mode = 0644,
1164 .proc_handler = &proc_dointvec,
1165 .strategy = &sysctl_intvec,
1166 .extra1 = &zero,
1167 },
1168 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1169 {
1170 .ctl_name = VM_LEGACY_VA_LAYOUT,
1171 .procname = "legacy_va_layout",
1172 .data = &sysctl_legacy_va_layout,
1173 .maxlen = sizeof(sysctl_legacy_va_layout),
1174 .mode = 0644,
1175 .proc_handler = &proc_dointvec,
1176 .strategy = &sysctl_intvec,
1177 .extra1 = &zero,
1178 },
1179 #endif
1180 #ifdef CONFIG_NUMA
1181 {
1182 .ctl_name = VM_ZONE_RECLAIM_MODE,
1183 .procname = "zone_reclaim_mode",
1184 .data = &zone_reclaim_mode,
1185 .maxlen = sizeof(zone_reclaim_mode),
1186 .mode = 0644,
1187 .proc_handler = &proc_dointvec,
1188 .strategy = &sysctl_intvec,
1189 .extra1 = &zero,
1190 },
1191 {
1192 .ctl_name = VM_MIN_UNMAPPED,
1193 .procname = "min_unmapped_ratio",
1194 .data = &sysctl_min_unmapped_ratio,
1195 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1196 .mode = 0644,
1197 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
1198 .strategy = &sysctl_intvec,
1199 .extra1 = &zero,
1200 .extra2 = &one_hundred,
1201 },
1202 {
1203 .ctl_name = VM_MIN_SLAB,
1204 .procname = "min_slab_ratio",
1205 .data = &sysctl_min_slab_ratio,
1206 .maxlen = sizeof(sysctl_min_slab_ratio),
1207 .mode = 0644,
1208 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1209 .strategy = &sysctl_intvec,
1210 .extra1 = &zero,
1211 .extra2 = &one_hundred,
1212 },
1213 #endif
1214 #ifdef CONFIG_SMP
1215 {
1216 .ctl_name = CTL_UNNUMBERED,
1217 .procname = "stat_interval",
1218 .data = &sysctl_stat_interval,
1219 .maxlen = sizeof(sysctl_stat_interval),
1220 .mode = 0644,
1221 .proc_handler = &proc_dointvec_jiffies,
1222 .strategy = &sysctl_jiffies,
1223 },
1224 #endif
1225 {
1226 .ctl_name = CTL_UNNUMBERED,
1227 .procname = "mmap_min_addr",
1228 .data = &mmap_min_addr,
1229 .maxlen = sizeof(unsigned long),
1230 .mode = 0644,
1231 .proc_handler = &proc_doulongvec_minmax,
1232 },
1233 #ifdef CONFIG_NUMA
1234 {
1235 .ctl_name = CTL_UNNUMBERED,
1236 .procname = "numa_zonelist_order",
1237 .data = &numa_zonelist_order,
1238 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1239 .mode = 0644,
1240 .proc_handler = &numa_zonelist_order_handler,
1241 .strategy = &sysctl_string,
1242 },
1243 #endif
1244 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1245 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1246 {
1247 .ctl_name = VM_VDSO_ENABLED,
1248 .procname = "vdso_enabled",
1249 .data = &vdso_enabled,
1250 .maxlen = sizeof(vdso_enabled),
1251 .mode = 0644,
1252 .proc_handler = &proc_dointvec,
1253 .strategy = &sysctl_intvec,
1254 .extra1 = &zero,
1255 },
1256 #endif
1257 #ifdef CONFIG_HIGHMEM
1258 {
1259 .ctl_name = CTL_UNNUMBERED,
1260 .procname = "highmem_is_dirtyable",
1261 .data = &vm_highmem_is_dirtyable,
1262 .maxlen = sizeof(vm_highmem_is_dirtyable),
1263 .mode = 0644,
1264 .proc_handler = &proc_dointvec_minmax,
1265 .strategy = &sysctl_intvec,
1266 .extra1 = &zero,
1267 .extra2 = &one,
1268 },
1269 #endif
1270 /*
1271 * NOTE: do not add new entries to this table unless you have read
1272 * Documentation/sysctl/ctl_unnumbered.txt
1273 */
1274 { .ctl_name = 0 }
1275 };
1276
1277 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1278 static struct ctl_table binfmt_misc_table[] = {
1279 { .ctl_name = 0 }
1280 };
1281 #endif
1282
1283 static struct ctl_table fs_table[] = {
1284 {
1285 .ctl_name = FS_NRINODE,
1286 .procname = "inode-nr",
1287 .data = &inodes_stat,
1288 .maxlen = 2*sizeof(int),
1289 .mode = 0444,
1290 .proc_handler = &proc_dointvec,
1291 },
1292 {
1293 .ctl_name = FS_STATINODE,
1294 .procname = "inode-state",
1295 .data = &inodes_stat,
1296 .maxlen = 7*sizeof(int),
1297 .mode = 0444,
1298 .proc_handler = &proc_dointvec,
1299 },
1300 {
1301 .procname = "file-nr",
1302 .data = &files_stat,
1303 .maxlen = 3*sizeof(int),
1304 .mode = 0444,
1305 .proc_handler = &proc_nr_files,
1306 },
1307 {
1308 .ctl_name = FS_MAXFILE,
1309 .procname = "file-max",
1310 .data = &files_stat.max_files,
1311 .maxlen = sizeof(int),
1312 .mode = 0644,
1313 .proc_handler = &proc_dointvec,
1314 },
1315 {
1316 .ctl_name = CTL_UNNUMBERED,
1317 .procname = "nr_open",
1318 .data = &sysctl_nr_open,
1319 .maxlen = sizeof(int),
1320 .mode = 0644,
1321 .proc_handler = &proc_dointvec_minmax,
1322 .extra1 = &sysctl_nr_open_min,
1323 .extra2 = &sysctl_nr_open_max,
1324 },
1325 {
1326 .ctl_name = FS_DENTRY,
1327 .procname = "dentry-state",
1328 .data = &dentry_stat,
1329 .maxlen = 6*sizeof(int),
1330 .mode = 0444,
1331 .proc_handler = &proc_dointvec,
1332 },
1333 {
1334 .ctl_name = FS_OVERFLOWUID,
1335 .procname = "overflowuid",
1336 .data = &fs_overflowuid,
1337 .maxlen = sizeof(int),
1338 .mode = 0644,
1339 .proc_handler = &proc_dointvec_minmax,
1340 .strategy = &sysctl_intvec,
1341 .extra1 = &minolduid,
1342 .extra2 = &maxolduid,
1343 },
1344 {
1345 .ctl_name = FS_OVERFLOWGID,
1346 .procname = "overflowgid",
1347 .data = &fs_overflowgid,
1348 .maxlen = sizeof(int),
1349 .mode = 0644,
1350 .proc_handler = &proc_dointvec_minmax,
1351 .strategy = &sysctl_intvec,
1352 .extra1 = &minolduid,
1353 .extra2 = &maxolduid,
1354 },
1355 #ifdef CONFIG_FILE_LOCKING
1356 {
1357 .ctl_name = FS_LEASES,
1358 .procname = "leases-enable",
1359 .data = &leases_enable,
1360 .maxlen = sizeof(int),
1361 .mode = 0644,
1362 .proc_handler = &proc_dointvec,
1363 },
1364 #endif
1365 #ifdef CONFIG_DNOTIFY
1366 {
1367 .ctl_name = FS_DIR_NOTIFY,
1368 .procname = "dir-notify-enable",
1369 .data = &dir_notify_enable,
1370 .maxlen = sizeof(int),
1371 .mode = 0644,
1372 .proc_handler = &proc_dointvec,
1373 },
1374 #endif
1375 #ifdef CONFIG_MMU
1376 #ifdef CONFIG_FILE_LOCKING
1377 {
1378 .ctl_name = FS_LEASE_TIME,
1379 .procname = "lease-break-time",
1380 .data = &lease_break_time,
1381 .maxlen = sizeof(int),
1382 .mode = 0644,
1383 .proc_handler = &proc_dointvec_minmax,
1384 .strategy = &sysctl_intvec,
1385 .extra1 = &zero,
1386 .extra2 = &two,
1387 },
1388 #endif
1389 #ifdef CONFIG_AIO
1390 {
1391 .procname = "aio-nr",
1392 .data = &aio_nr,
1393 .maxlen = sizeof(aio_nr),
1394 .mode = 0444,
1395 .proc_handler = &proc_doulongvec_minmax,
1396 },
1397 {
1398 .procname = "aio-max-nr",
1399 .data = &aio_max_nr,
1400 .maxlen = sizeof(aio_max_nr),
1401 .mode = 0644,
1402 .proc_handler = &proc_doulongvec_minmax,
1403 },
1404 #endif /* CONFIG_AIO */
1405 #ifdef CONFIG_INOTIFY_USER
1406 {
1407 .ctl_name = FS_INOTIFY,
1408 .procname = "inotify",
1409 .mode = 0555,
1410 .child = inotify_table,
1411 },
1412 #endif
1413 #ifdef CONFIG_EPOLL
1414 {
1415 .procname = "epoll",
1416 .mode = 0555,
1417 .child = epoll_table,
1418 },
1419 #endif
1420 #endif
1421 {
1422 .ctl_name = KERN_SETUID_DUMPABLE,
1423 .procname = "suid_dumpable",
1424 .data = &suid_dumpable,
1425 .maxlen = sizeof(int),
1426 .mode = 0644,
1427 .proc_handler = &proc_dointvec,
1428 },
1429 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1430 {
1431 .ctl_name = CTL_UNNUMBERED,
1432 .procname = "binfmt_misc",
1433 .mode = 0555,
1434 .child = binfmt_misc_table,
1435 },
1436 #endif
1437 /*
1438 * NOTE: do not add new entries to this table unless you have read
1439 * Documentation/sysctl/ctl_unnumbered.txt
1440 */
1441 { .ctl_name = 0 }
1442 };
1443
1444 static struct ctl_table debug_table[] = {
1445 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1446 {
1447 .ctl_name = CTL_UNNUMBERED,
1448 .procname = "exception-trace",
1449 .data = &show_unhandled_signals,
1450 .maxlen = sizeof(int),
1451 .mode = 0644,
1452 .proc_handler = proc_dointvec
1453 },
1454 #endif
1455 { .ctl_name = 0 }
1456 };
1457
1458 static struct ctl_table dev_table[] = {
1459 { .ctl_name = 0 }
1460 };
1461
1462 static DEFINE_SPINLOCK(sysctl_lock);
1463
1464 /* called under sysctl_lock */
use_table(struct ctl_table_header * p)1465 static int use_table(struct ctl_table_header *p)
1466 {
1467 if (unlikely(p->unregistering))
1468 return 0;
1469 p->used++;
1470 return 1;
1471 }
1472
1473 /* called under sysctl_lock */
unuse_table(struct ctl_table_header * p)1474 static void unuse_table(struct ctl_table_header *p)
1475 {
1476 if (!--p->used)
1477 if (unlikely(p->unregistering))
1478 complete(p->unregistering);
1479 }
1480
1481 /* called under sysctl_lock, will reacquire if has to wait */
start_unregistering(struct ctl_table_header * p)1482 static void start_unregistering(struct ctl_table_header *p)
1483 {
1484 /*
1485 * if p->used is 0, nobody will ever touch that entry again;
1486 * we'll eliminate all paths to it before dropping sysctl_lock
1487 */
1488 if (unlikely(p->used)) {
1489 struct completion wait;
1490 init_completion(&wait);
1491 p->unregistering = &wait;
1492 spin_unlock(&sysctl_lock);
1493 wait_for_completion(&wait);
1494 spin_lock(&sysctl_lock);
1495 } else {
1496 /* anything non-NULL; we'll never dereference it */
1497 p->unregistering = ERR_PTR(-EINVAL);
1498 }
1499 /*
1500 * do not remove from the list until nobody holds it; walking the
1501 * list in do_sysctl() relies on that.
1502 */
1503 list_del_init(&p->ctl_entry);
1504 }
1505
sysctl_head_get(struct ctl_table_header * head)1506 void sysctl_head_get(struct ctl_table_header *head)
1507 {
1508 spin_lock(&sysctl_lock);
1509 head->count++;
1510 spin_unlock(&sysctl_lock);
1511 }
1512
sysctl_head_put(struct ctl_table_header * head)1513 void sysctl_head_put(struct ctl_table_header *head)
1514 {
1515 spin_lock(&sysctl_lock);
1516 if (!--head->count)
1517 kfree(head);
1518 spin_unlock(&sysctl_lock);
1519 }
1520
sysctl_head_grab(struct ctl_table_header * head)1521 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1522 {
1523 if (!head)
1524 BUG();
1525 spin_lock(&sysctl_lock);
1526 if (!use_table(head))
1527 head = ERR_PTR(-ENOENT);
1528 spin_unlock(&sysctl_lock);
1529 return head;
1530 }
1531
sysctl_head_finish(struct ctl_table_header * head)1532 void sysctl_head_finish(struct ctl_table_header *head)
1533 {
1534 if (!head)
1535 return;
1536 spin_lock(&sysctl_lock);
1537 unuse_table(head);
1538 spin_unlock(&sysctl_lock);
1539 }
1540
1541 static struct ctl_table_set *
lookup_header_set(struct ctl_table_root * root,struct nsproxy * namespaces)1542 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1543 {
1544 struct ctl_table_set *set = &root->default_set;
1545 if (root->lookup)
1546 set = root->lookup(root, namespaces);
1547 return set;
1548 }
1549
1550 static struct list_head *
lookup_header_list(struct ctl_table_root * root,struct nsproxy * namespaces)1551 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1552 {
1553 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1554 return &set->list;
1555 }
1556
__sysctl_head_next(struct nsproxy * namespaces,struct ctl_table_header * prev)1557 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1558 struct ctl_table_header *prev)
1559 {
1560 struct ctl_table_root *root;
1561 struct list_head *header_list;
1562 struct ctl_table_header *head;
1563 struct list_head *tmp;
1564
1565 spin_lock(&sysctl_lock);
1566 if (prev) {
1567 head = prev;
1568 tmp = &prev->ctl_entry;
1569 unuse_table(prev);
1570 goto next;
1571 }
1572 tmp = &root_table_header.ctl_entry;
1573 for (;;) {
1574 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1575
1576 if (!use_table(head))
1577 goto next;
1578 spin_unlock(&sysctl_lock);
1579 return head;
1580 next:
1581 root = head->root;
1582 tmp = tmp->next;
1583 header_list = lookup_header_list(root, namespaces);
1584 if (tmp != header_list)
1585 continue;
1586
1587 do {
1588 root = list_entry(root->root_list.next,
1589 struct ctl_table_root, root_list);
1590 if (root == &sysctl_table_root)
1591 goto out;
1592 header_list = lookup_header_list(root, namespaces);
1593 } while (list_empty(header_list));
1594 tmp = header_list->next;
1595 }
1596 out:
1597 spin_unlock(&sysctl_lock);
1598 return NULL;
1599 }
1600
sysctl_head_next(struct ctl_table_header * prev)1601 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1602 {
1603 return __sysctl_head_next(current->nsproxy, prev);
1604 }
1605
register_sysctl_root(struct ctl_table_root * root)1606 void register_sysctl_root(struct ctl_table_root *root)
1607 {
1608 spin_lock(&sysctl_lock);
1609 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1610 spin_unlock(&sysctl_lock);
1611 }
1612
1613 #ifdef CONFIG_SYSCTL_SYSCALL
1614 /* Perform the actual read/write of a sysctl table entry. */
do_sysctl_strategy(struct ctl_table_root * root,struct ctl_table * table,void __user * oldval,size_t __user * oldlenp,void __user * newval,size_t newlen)1615 static int do_sysctl_strategy(struct ctl_table_root *root,
1616 struct ctl_table *table,
1617 void __user *oldval, size_t __user *oldlenp,
1618 void __user *newval, size_t newlen)
1619 {
1620 int op = 0, rc;
1621
1622 if (oldval)
1623 op |= MAY_READ;
1624 if (newval)
1625 op |= MAY_WRITE;
1626 if (sysctl_perm(root, table, op))
1627 return -EPERM;
1628
1629 if (table->strategy) {
1630 rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1631 if (rc < 0)
1632 return rc;
1633 if (rc > 0)
1634 return 0;
1635 }
1636
1637 /* If there is no strategy routine, or if the strategy returns
1638 * zero, proceed with automatic r/w */
1639 if (table->data && table->maxlen) {
1640 rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1641 if (rc < 0)
1642 return rc;
1643 }
1644 return 0;
1645 }
1646
parse_table(int __user * name,int nlen,void __user * oldval,size_t __user * oldlenp,void __user * newval,size_t newlen,struct ctl_table_root * root,struct ctl_table * table)1647 static int parse_table(int __user *name, int nlen,
1648 void __user *oldval, size_t __user *oldlenp,
1649 void __user *newval, size_t newlen,
1650 struct ctl_table_root *root,
1651 struct ctl_table *table)
1652 {
1653 int n;
1654 repeat:
1655 if (!nlen)
1656 return -ENOTDIR;
1657 if (get_user(n, name))
1658 return -EFAULT;
1659 for ( ; table->ctl_name || table->procname; table++) {
1660 if (!table->ctl_name)
1661 continue;
1662 if (n == table->ctl_name) {
1663 int error;
1664 if (table->child) {
1665 if (sysctl_perm(root, table, MAY_EXEC))
1666 return -EPERM;
1667 name++;
1668 nlen--;
1669 table = table->child;
1670 goto repeat;
1671 }
1672 error = do_sysctl_strategy(root, table,
1673 oldval, oldlenp,
1674 newval, newlen);
1675 return error;
1676 }
1677 }
1678 return -ENOTDIR;
1679 }
1680
do_sysctl(int __user * name,int nlen,void __user * oldval,size_t __user * oldlenp,void __user * newval,size_t newlen)1681 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1682 void __user *newval, size_t newlen)
1683 {
1684 struct ctl_table_header *head;
1685 int error = -ENOTDIR;
1686
1687 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1688 return -ENOTDIR;
1689 if (oldval) {
1690 int old_len;
1691 if (!oldlenp || get_user(old_len, oldlenp))
1692 return -EFAULT;
1693 }
1694
1695 for (head = sysctl_head_next(NULL); head;
1696 head = sysctl_head_next(head)) {
1697 error = parse_table(name, nlen, oldval, oldlenp,
1698 newval, newlen,
1699 head->root, head->ctl_table);
1700 if (error != -ENOTDIR) {
1701 sysctl_head_finish(head);
1702 break;
1703 }
1704 }
1705 return error;
1706 }
1707
SYSCALL_DEFINE1(sysctl,struct __sysctl_args __user *,args)1708 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
1709 {
1710 struct __sysctl_args tmp;
1711 int error;
1712
1713 if (copy_from_user(&tmp, args, sizeof(tmp)))
1714 return -EFAULT;
1715
1716 error = deprecated_sysctl_warning(&tmp);
1717 if (error)
1718 goto out;
1719
1720 lock_kernel();
1721 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1722 tmp.newval, tmp.newlen);
1723 unlock_kernel();
1724 out:
1725 return error;
1726 }
1727 #endif /* CONFIG_SYSCTL_SYSCALL */
1728
1729 /*
1730 * sysctl_perm does NOT grant the superuser all rights automatically, because
1731 * some sysctl variables are readonly even to root.
1732 */
1733
test_perm(int mode,int op)1734 static int test_perm(int mode, int op)
1735 {
1736 if (!current_euid())
1737 mode >>= 6;
1738 else if (in_egroup_p(0))
1739 mode >>= 3;
1740 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1741 return 0;
1742 return -EACCES;
1743 }
1744
sysctl_perm(struct ctl_table_root * root,struct ctl_table * table,int op)1745 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1746 {
1747 int error;
1748 int mode;
1749
1750 error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1751 if (error)
1752 return error;
1753
1754 if (root->permissions)
1755 mode = root->permissions(root, current->nsproxy, table);
1756 else
1757 mode = table->mode;
1758
1759 return test_perm(mode, op);
1760 }
1761
sysctl_set_parent(struct ctl_table * parent,struct ctl_table * table)1762 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1763 {
1764 for (; table->ctl_name || table->procname; table++) {
1765 table->parent = parent;
1766 if (table->child)
1767 sysctl_set_parent(table, table->child);
1768 }
1769 }
1770
sysctl_init(void)1771 static __init int sysctl_init(void)
1772 {
1773 sysctl_set_parent(NULL, root_table);
1774 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1775 {
1776 int err;
1777 err = sysctl_check_table(current->nsproxy, root_table);
1778 }
1779 #endif
1780 return 0;
1781 }
1782
1783 core_initcall(sysctl_init);
1784
is_branch_in(struct ctl_table * branch,struct ctl_table * table)1785 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1786 struct ctl_table *table)
1787 {
1788 struct ctl_table *p;
1789 const char *s = branch->procname;
1790
1791 /* branch should have named subdirectory as its first element */
1792 if (!s || !branch->child)
1793 return NULL;
1794
1795 /* ... and nothing else */
1796 if (branch[1].procname || branch[1].ctl_name)
1797 return NULL;
1798
1799 /* table should contain subdirectory with the same name */
1800 for (p = table; p->procname || p->ctl_name; p++) {
1801 if (!p->child)
1802 continue;
1803 if (p->procname && strcmp(p->procname, s) == 0)
1804 return p;
1805 }
1806 return NULL;
1807 }
1808
1809 /* see if attaching q to p would be an improvement */
try_attach(struct ctl_table_header * p,struct ctl_table_header * q)1810 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1811 {
1812 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1813 struct ctl_table *next;
1814 int is_better = 0;
1815 int not_in_parent = !p->attached_by;
1816
1817 while ((next = is_branch_in(by, to)) != NULL) {
1818 if (by == q->attached_by)
1819 is_better = 1;
1820 if (to == p->attached_by)
1821 not_in_parent = 1;
1822 by = by->child;
1823 to = next->child;
1824 }
1825
1826 if (is_better && not_in_parent) {
1827 q->attached_by = by;
1828 q->attached_to = to;
1829 q->parent = p;
1830 }
1831 }
1832
1833 /**
1834 * __register_sysctl_paths - register a sysctl hierarchy
1835 * @root: List of sysctl headers to register on
1836 * @namespaces: Data to compute which lists of sysctl entries are visible
1837 * @path: The path to the directory the sysctl table is in.
1838 * @table: the top-level table structure
1839 *
1840 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1841 * array. A completely 0 filled entry terminates the table.
1842 *
1843 * The members of the &struct ctl_table structure are used as follows:
1844 *
1845 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1846 * must be unique within that level of sysctl
1847 *
1848 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1849 * enter a sysctl file
1850 *
1851 * data - a pointer to data for use by proc_handler
1852 *
1853 * maxlen - the maximum size in bytes of the data
1854 *
1855 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1856 *
1857 * child - a pointer to the child sysctl table if this entry is a directory, or
1858 * %NULL.
1859 *
1860 * proc_handler - the text handler routine (described below)
1861 *
1862 * strategy - the strategy routine (described below)
1863 *
1864 * de - for internal use by the sysctl routines
1865 *
1866 * extra1, extra2 - extra pointers usable by the proc handler routines
1867 *
1868 * Leaf nodes in the sysctl tree will be represented by a single file
1869 * under /proc; non-leaf nodes will be represented by directories.
1870 *
1871 * sysctl(2) can automatically manage read and write requests through
1872 * the sysctl table. The data and maxlen fields of the ctl_table
1873 * struct enable minimal validation of the values being written to be
1874 * performed, and the mode field allows minimal authentication.
1875 *
1876 * More sophisticated management can be enabled by the provision of a
1877 * strategy routine with the table entry. This will be called before
1878 * any automatic read or write of the data is performed.
1879 *
1880 * The strategy routine may return
1881 *
1882 * < 0 - Error occurred (error is passed to user process)
1883 *
1884 * 0 - OK - proceed with automatic read or write.
1885 *
1886 * > 0 - OK - read or write has been done by the strategy routine, so
1887 * return immediately.
1888 *
1889 * There must be a proc_handler routine for any terminal nodes
1890 * mirrored under /proc/sys (non-terminals are handled by a built-in
1891 * directory handler). Several default handlers are available to
1892 * cover common cases -
1893 *
1894 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1895 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1896 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1897 *
1898 * It is the handler's job to read the input buffer from user memory
1899 * and process it. The handler should return 0 on success.
1900 *
1901 * This routine returns %NULL on a failure to register, and a pointer
1902 * to the table header on success.
1903 */
__register_sysctl_paths(struct ctl_table_root * root,struct nsproxy * namespaces,const struct ctl_path * path,struct ctl_table * table)1904 struct ctl_table_header *__register_sysctl_paths(
1905 struct ctl_table_root *root,
1906 struct nsproxy *namespaces,
1907 const struct ctl_path *path, struct ctl_table *table)
1908 {
1909 struct ctl_table_header *header;
1910 struct ctl_table *new, **prevp;
1911 unsigned int n, npath;
1912 struct ctl_table_set *set;
1913
1914 /* Count the path components */
1915 for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1916 ;
1917
1918 /*
1919 * For each path component, allocate a 2-element ctl_table array.
1920 * The first array element will be filled with the sysctl entry
1921 * for this, the second will be the sentinel (ctl_name == 0).
1922 *
1923 * We allocate everything in one go so that we don't have to
1924 * worry about freeing additional memory in unregister_sysctl_table.
1925 */
1926 header = kzalloc(sizeof(struct ctl_table_header) +
1927 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1928 if (!header)
1929 return NULL;
1930
1931 new = (struct ctl_table *) (header + 1);
1932
1933 /* Now connect the dots */
1934 prevp = &header->ctl_table;
1935 for (n = 0; n < npath; ++n, ++path) {
1936 /* Copy the procname */
1937 new->procname = path->procname;
1938 new->ctl_name = path->ctl_name;
1939 new->mode = 0555;
1940
1941 *prevp = new;
1942 prevp = &new->child;
1943
1944 new += 2;
1945 }
1946 *prevp = table;
1947 header->ctl_table_arg = table;
1948
1949 INIT_LIST_HEAD(&header->ctl_entry);
1950 header->used = 0;
1951 header->unregistering = NULL;
1952 header->root = root;
1953 sysctl_set_parent(NULL, header->ctl_table);
1954 header->count = 1;
1955 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1956 if (sysctl_check_table(namespaces, header->ctl_table)) {
1957 kfree(header);
1958 return NULL;
1959 }
1960 #endif
1961 spin_lock(&sysctl_lock);
1962 header->set = lookup_header_set(root, namespaces);
1963 header->attached_by = header->ctl_table;
1964 header->attached_to = root_table;
1965 header->parent = &root_table_header;
1966 for (set = header->set; set; set = set->parent) {
1967 struct ctl_table_header *p;
1968 list_for_each_entry(p, &set->list, ctl_entry) {
1969 if (p->unregistering)
1970 continue;
1971 try_attach(p, header);
1972 }
1973 }
1974 header->parent->count++;
1975 list_add_tail(&header->ctl_entry, &header->set->list);
1976 spin_unlock(&sysctl_lock);
1977
1978 return header;
1979 }
1980
1981 /**
1982 * register_sysctl_table_path - register a sysctl table hierarchy
1983 * @path: The path to the directory the sysctl table is in.
1984 * @table: the top-level table structure
1985 *
1986 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1987 * array. A completely 0 filled entry terminates the table.
1988 *
1989 * See __register_sysctl_paths for more details.
1990 */
register_sysctl_paths(const struct ctl_path * path,struct ctl_table * table)1991 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1992 struct ctl_table *table)
1993 {
1994 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1995 path, table);
1996 }
1997
1998 /**
1999 * register_sysctl_table - register a sysctl table hierarchy
2000 * @table: the top-level table structure
2001 *
2002 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2003 * array. A completely 0 filled entry terminates the table.
2004 *
2005 * See register_sysctl_paths for more details.
2006 */
register_sysctl_table(struct ctl_table * table)2007 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
2008 {
2009 static const struct ctl_path null_path[] = { {} };
2010
2011 return register_sysctl_paths(null_path, table);
2012 }
2013
2014 /**
2015 * unregister_sysctl_table - unregister a sysctl table hierarchy
2016 * @header: the header returned from register_sysctl_table
2017 *
2018 * Unregisters the sysctl table and all children. proc entries may not
2019 * actually be removed until they are no longer used by anyone.
2020 */
unregister_sysctl_table(struct ctl_table_header * header)2021 void unregister_sysctl_table(struct ctl_table_header * header)
2022 {
2023 might_sleep();
2024
2025 if (header == NULL)
2026 return;
2027
2028 spin_lock(&sysctl_lock);
2029 start_unregistering(header);
2030 if (!--header->parent->count) {
2031 WARN_ON(1);
2032 kfree(header->parent);
2033 }
2034 if (!--header->count)
2035 kfree(header);
2036 spin_unlock(&sysctl_lock);
2037 }
2038
sysctl_is_seen(struct ctl_table_header * p)2039 int sysctl_is_seen(struct ctl_table_header *p)
2040 {
2041 struct ctl_table_set *set = p->set;
2042 int res;
2043 spin_lock(&sysctl_lock);
2044 if (p->unregistering)
2045 res = 0;
2046 else if (!set->is_seen)
2047 res = 1;
2048 else
2049 res = set->is_seen(set);
2050 spin_unlock(&sysctl_lock);
2051 return res;
2052 }
2053
setup_sysctl_set(struct ctl_table_set * p,struct ctl_table_set * parent,int (* is_seen)(struct ctl_table_set *))2054 void setup_sysctl_set(struct ctl_table_set *p,
2055 struct ctl_table_set *parent,
2056 int (*is_seen)(struct ctl_table_set *))
2057 {
2058 INIT_LIST_HEAD(&p->list);
2059 p->parent = parent ? parent : &sysctl_table_root.default_set;
2060 p->is_seen = is_seen;
2061 }
2062
2063 #else /* !CONFIG_SYSCTL */
register_sysctl_table(struct ctl_table * table)2064 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2065 {
2066 return NULL;
2067 }
2068
register_sysctl_paths(const struct ctl_path * path,struct ctl_table * table)2069 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2070 struct ctl_table *table)
2071 {
2072 return NULL;
2073 }
2074
unregister_sysctl_table(struct ctl_table_header * table)2075 void unregister_sysctl_table(struct ctl_table_header * table)
2076 {
2077 }
2078
setup_sysctl_set(struct ctl_table_set * p,struct ctl_table_set * parent,int (* is_seen)(struct ctl_table_set *))2079 void setup_sysctl_set(struct ctl_table_set *p,
2080 struct ctl_table_set *parent,
2081 int (*is_seen)(struct ctl_table_set *))
2082 {
2083 }
2084
sysctl_head_put(struct ctl_table_header * head)2085 void sysctl_head_put(struct ctl_table_header *head)
2086 {
2087 }
2088
2089 #endif /* CONFIG_SYSCTL */
2090
2091 /*
2092 * /proc/sys support
2093 */
2094
2095 #ifdef CONFIG_PROC_SYSCTL
2096
_proc_do_string(void * data,int maxlen,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2097 static int _proc_do_string(void* data, int maxlen, int write,
2098 struct file *filp, void __user *buffer,
2099 size_t *lenp, loff_t *ppos)
2100 {
2101 size_t len;
2102 char __user *p;
2103 char c;
2104
2105 if (!data || !maxlen || !*lenp) {
2106 *lenp = 0;
2107 return 0;
2108 }
2109
2110 if (write) {
2111 len = 0;
2112 p = buffer;
2113 while (len < *lenp) {
2114 if (get_user(c, p++))
2115 return -EFAULT;
2116 if (c == 0 || c == '\n')
2117 break;
2118 len++;
2119 }
2120 if (len >= maxlen)
2121 len = maxlen-1;
2122 if(copy_from_user(data, buffer, len))
2123 return -EFAULT;
2124 ((char *) data)[len] = 0;
2125 *ppos += *lenp;
2126 } else {
2127 len = strlen(data);
2128 if (len > maxlen)
2129 len = maxlen;
2130
2131 if (*ppos > len) {
2132 *lenp = 0;
2133 return 0;
2134 }
2135
2136 data += *ppos;
2137 len -= *ppos;
2138
2139 if (len > *lenp)
2140 len = *lenp;
2141 if (len)
2142 if(copy_to_user(buffer, data, len))
2143 return -EFAULT;
2144 if (len < *lenp) {
2145 if(put_user('\n', ((char __user *) buffer) + len))
2146 return -EFAULT;
2147 len++;
2148 }
2149 *lenp = len;
2150 *ppos += len;
2151 }
2152 return 0;
2153 }
2154
2155 /**
2156 * proc_dostring - read a string sysctl
2157 * @table: the sysctl table
2158 * @write: %TRUE if this is a write to the sysctl file
2159 * @filp: the file structure
2160 * @buffer: the user buffer
2161 * @lenp: the size of the user buffer
2162 * @ppos: file position
2163 *
2164 * Reads/writes a string from/to the user buffer. If the kernel
2165 * buffer provided is not large enough to hold the string, the
2166 * string is truncated. The copied string is %NULL-terminated.
2167 * If the string is being read by the user process, it is copied
2168 * and a newline '\n' is added. It is truncated if the buffer is
2169 * not large enough.
2170 *
2171 * Returns 0 on success.
2172 */
proc_dostring(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2173 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2174 void __user *buffer, size_t *lenp, loff_t *ppos)
2175 {
2176 return _proc_do_string(table->data, table->maxlen, write, filp,
2177 buffer, lenp, ppos);
2178 }
2179
2180
do_proc_dointvec_conv(int * negp,unsigned long * lvalp,int * valp,int write,void * data)2181 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2182 int *valp,
2183 int write, void *data)
2184 {
2185 if (write) {
2186 *valp = *negp ? -*lvalp : *lvalp;
2187 } else {
2188 int val = *valp;
2189 if (val < 0) {
2190 *negp = -1;
2191 *lvalp = (unsigned long)-val;
2192 } else {
2193 *negp = 0;
2194 *lvalp = (unsigned long)val;
2195 }
2196 }
2197 return 0;
2198 }
2199
__do_proc_dointvec(void * tbl_data,struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos,int (* conv)(int * negp,unsigned long * lvalp,int * valp,int write,void * data),void * data)2200 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2201 int write, struct file *filp, void __user *buffer,
2202 size_t *lenp, loff_t *ppos,
2203 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2204 int write, void *data),
2205 void *data)
2206 {
2207 #define TMPBUFLEN 21
2208 int *i, vleft, first=1, neg, val;
2209 unsigned long lval;
2210 size_t left, len;
2211
2212 char buf[TMPBUFLEN], *p;
2213 char __user *s = buffer;
2214
2215 if (!tbl_data || !table->maxlen || !*lenp ||
2216 (*ppos && !write)) {
2217 *lenp = 0;
2218 return 0;
2219 }
2220
2221 i = (int *) tbl_data;
2222 vleft = table->maxlen / sizeof(*i);
2223 left = *lenp;
2224
2225 if (!conv)
2226 conv = do_proc_dointvec_conv;
2227
2228 for (; left && vleft--; i++, first=0) {
2229 if (write) {
2230 while (left) {
2231 char c;
2232 if (get_user(c, s))
2233 return -EFAULT;
2234 if (!isspace(c))
2235 break;
2236 left--;
2237 s++;
2238 }
2239 if (!left)
2240 break;
2241 neg = 0;
2242 len = left;
2243 if (len > sizeof(buf) - 1)
2244 len = sizeof(buf) - 1;
2245 if (copy_from_user(buf, s, len))
2246 return -EFAULT;
2247 buf[len] = 0;
2248 p = buf;
2249 if (*p == '-' && left > 1) {
2250 neg = 1;
2251 p++;
2252 }
2253 if (*p < '0' || *p > '9')
2254 break;
2255
2256 lval = simple_strtoul(p, &p, 0);
2257
2258 len = p-buf;
2259 if ((len < left) && *p && !isspace(*p))
2260 break;
2261 if (neg)
2262 val = -val;
2263 s += len;
2264 left -= len;
2265
2266 if (conv(&neg, &lval, i, 1, data))
2267 break;
2268 } else {
2269 p = buf;
2270 if (!first)
2271 *p++ = '\t';
2272
2273 if (conv(&neg, &lval, i, 0, data))
2274 break;
2275
2276 sprintf(p, "%s%lu", neg ? "-" : "", lval);
2277 len = strlen(buf);
2278 if (len > left)
2279 len = left;
2280 if(copy_to_user(s, buf, len))
2281 return -EFAULT;
2282 left -= len;
2283 s += len;
2284 }
2285 }
2286
2287 if (!write && !first && left) {
2288 if(put_user('\n', s))
2289 return -EFAULT;
2290 left--, s++;
2291 }
2292 if (write) {
2293 while (left) {
2294 char c;
2295 if (get_user(c, s++))
2296 return -EFAULT;
2297 if (!isspace(c))
2298 break;
2299 left--;
2300 }
2301 }
2302 if (write && first)
2303 return -EINVAL;
2304 *lenp -= left;
2305 *ppos += *lenp;
2306 return 0;
2307 #undef TMPBUFLEN
2308 }
2309
do_proc_dointvec(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos,int (* conv)(int * negp,unsigned long * lvalp,int * valp,int write,void * data),void * data)2310 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2311 void __user *buffer, size_t *lenp, loff_t *ppos,
2312 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2313 int write, void *data),
2314 void *data)
2315 {
2316 return __do_proc_dointvec(table->data, table, write, filp,
2317 buffer, lenp, ppos, conv, data);
2318 }
2319
2320 /**
2321 * proc_dointvec - read a vector of integers
2322 * @table: the sysctl table
2323 * @write: %TRUE if this is a write to the sysctl file
2324 * @filp: the file structure
2325 * @buffer: the user buffer
2326 * @lenp: the size of the user buffer
2327 * @ppos: file position
2328 *
2329 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2330 * values from/to the user buffer, treated as an ASCII string.
2331 *
2332 * Returns 0 on success.
2333 */
proc_dointvec(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2334 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2335 void __user *buffer, size_t *lenp, loff_t *ppos)
2336 {
2337 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2338 NULL,NULL);
2339 }
2340
2341 /*
2342 * Taint values can only be increased
2343 * This means we can safely use a temporary.
2344 */
proc_taint(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2345 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
2346 void __user *buffer, size_t *lenp, loff_t *ppos)
2347 {
2348 struct ctl_table t;
2349 unsigned long tmptaint = get_taint();
2350 int err;
2351
2352 if (write && !capable(CAP_SYS_ADMIN))
2353 return -EPERM;
2354
2355 t = *table;
2356 t.data = &tmptaint;
2357 err = proc_doulongvec_minmax(&t, write, filp, buffer, lenp, ppos);
2358 if (err < 0)
2359 return err;
2360
2361 if (write) {
2362 /*
2363 * Poor man's atomic or. Not worth adding a primitive
2364 * to everyone's atomic.h for this
2365 */
2366 int i;
2367 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2368 if ((tmptaint >> i) & 1)
2369 add_taint(i);
2370 }
2371 }
2372
2373 return err;
2374 }
2375
2376 struct do_proc_dointvec_minmax_conv_param {
2377 int *min;
2378 int *max;
2379 };
2380
do_proc_dointvec_minmax_conv(int * negp,unsigned long * lvalp,int * valp,int write,void * data)2381 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2382 int *valp,
2383 int write, void *data)
2384 {
2385 struct do_proc_dointvec_minmax_conv_param *param = data;
2386 if (write) {
2387 int val = *negp ? -*lvalp : *lvalp;
2388 if ((param->min && *param->min > val) ||
2389 (param->max && *param->max < val))
2390 return -EINVAL;
2391 *valp = val;
2392 } else {
2393 int val = *valp;
2394 if (val < 0) {
2395 *negp = -1;
2396 *lvalp = (unsigned long)-val;
2397 } else {
2398 *negp = 0;
2399 *lvalp = (unsigned long)val;
2400 }
2401 }
2402 return 0;
2403 }
2404
2405 /**
2406 * proc_dointvec_minmax - read a vector of integers with min/max values
2407 * @table: the sysctl table
2408 * @write: %TRUE if this is a write to the sysctl file
2409 * @filp: the file structure
2410 * @buffer: the user buffer
2411 * @lenp: the size of the user buffer
2412 * @ppos: file position
2413 *
2414 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2415 * values from/to the user buffer, treated as an ASCII string.
2416 *
2417 * This routine will ensure the values are within the range specified by
2418 * table->extra1 (min) and table->extra2 (max).
2419 *
2420 * Returns 0 on success.
2421 */
proc_dointvec_minmax(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2422 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2423 void __user *buffer, size_t *lenp, loff_t *ppos)
2424 {
2425 struct do_proc_dointvec_minmax_conv_param param = {
2426 .min = (int *) table->extra1,
2427 .max = (int *) table->extra2,
2428 };
2429 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2430 do_proc_dointvec_minmax_conv, ¶m);
2431 }
2432
__do_proc_doulongvec_minmax(void * data,struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos,unsigned long convmul,unsigned long convdiv)2433 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2434 struct file *filp,
2435 void __user *buffer,
2436 size_t *lenp, loff_t *ppos,
2437 unsigned long convmul,
2438 unsigned long convdiv)
2439 {
2440 #define TMPBUFLEN 21
2441 unsigned long *i, *min, *max, val;
2442 int vleft, first=1, neg;
2443 size_t len, left;
2444 char buf[TMPBUFLEN], *p;
2445 char __user *s = buffer;
2446
2447 if (!data || !table->maxlen || !*lenp ||
2448 (*ppos && !write)) {
2449 *lenp = 0;
2450 return 0;
2451 }
2452
2453 i = (unsigned long *) data;
2454 min = (unsigned long *) table->extra1;
2455 max = (unsigned long *) table->extra2;
2456 vleft = table->maxlen / sizeof(unsigned long);
2457 left = *lenp;
2458
2459 for (; left && vleft--; i++, min++, max++, first=0) {
2460 if (write) {
2461 while (left) {
2462 char c;
2463 if (get_user(c, s))
2464 return -EFAULT;
2465 if (!isspace(c))
2466 break;
2467 left--;
2468 s++;
2469 }
2470 if (!left)
2471 break;
2472 neg = 0;
2473 len = left;
2474 if (len > TMPBUFLEN-1)
2475 len = TMPBUFLEN-1;
2476 if (copy_from_user(buf, s, len))
2477 return -EFAULT;
2478 buf[len] = 0;
2479 p = buf;
2480 if (*p == '-' && left > 1) {
2481 neg = 1;
2482 p++;
2483 }
2484 if (*p < '0' || *p > '9')
2485 break;
2486 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2487 len = p-buf;
2488 if ((len < left) && *p && !isspace(*p))
2489 break;
2490 if (neg)
2491 val = -val;
2492 s += len;
2493 left -= len;
2494
2495 if(neg)
2496 continue;
2497 if ((min && val < *min) || (max && val > *max))
2498 continue;
2499 *i = val;
2500 } else {
2501 p = buf;
2502 if (!first)
2503 *p++ = '\t';
2504 sprintf(p, "%lu", convdiv * (*i) / convmul);
2505 len = strlen(buf);
2506 if (len > left)
2507 len = left;
2508 if(copy_to_user(s, buf, len))
2509 return -EFAULT;
2510 left -= len;
2511 s += len;
2512 }
2513 }
2514
2515 if (!write && !first && left) {
2516 if(put_user('\n', s))
2517 return -EFAULT;
2518 left--, s++;
2519 }
2520 if (write) {
2521 while (left) {
2522 char c;
2523 if (get_user(c, s++))
2524 return -EFAULT;
2525 if (!isspace(c))
2526 break;
2527 left--;
2528 }
2529 }
2530 if (write && first)
2531 return -EINVAL;
2532 *lenp -= left;
2533 *ppos += *lenp;
2534 return 0;
2535 #undef TMPBUFLEN
2536 }
2537
do_proc_doulongvec_minmax(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos,unsigned long convmul,unsigned long convdiv)2538 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2539 struct file *filp,
2540 void __user *buffer,
2541 size_t *lenp, loff_t *ppos,
2542 unsigned long convmul,
2543 unsigned long convdiv)
2544 {
2545 return __do_proc_doulongvec_minmax(table->data, table, write,
2546 filp, buffer, lenp, ppos, convmul, convdiv);
2547 }
2548
2549 /**
2550 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2551 * @table: the sysctl table
2552 * @write: %TRUE if this is a write to the sysctl file
2553 * @filp: the file structure
2554 * @buffer: the user buffer
2555 * @lenp: the size of the user buffer
2556 * @ppos: file position
2557 *
2558 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2559 * values from/to the user buffer, treated as an ASCII string.
2560 *
2561 * This routine will ensure the values are within the range specified by
2562 * table->extra1 (min) and table->extra2 (max).
2563 *
2564 * Returns 0 on success.
2565 */
proc_doulongvec_minmax(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2566 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2567 void __user *buffer, size_t *lenp, loff_t *ppos)
2568 {
2569 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2570 }
2571
2572 /**
2573 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2574 * @table: the sysctl table
2575 * @write: %TRUE if this is a write to the sysctl file
2576 * @filp: the file structure
2577 * @buffer: the user buffer
2578 * @lenp: the size of the user buffer
2579 * @ppos: file position
2580 *
2581 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2582 * values from/to the user buffer, treated as an ASCII string. The values
2583 * are treated as milliseconds, and converted to jiffies when they are stored.
2584 *
2585 * This routine will ensure the values are within the range specified by
2586 * table->extra1 (min) and table->extra2 (max).
2587 *
2588 * Returns 0 on success.
2589 */
proc_doulongvec_ms_jiffies_minmax(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2590 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2591 struct file *filp,
2592 void __user *buffer,
2593 size_t *lenp, loff_t *ppos)
2594 {
2595 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2596 lenp, ppos, HZ, 1000l);
2597 }
2598
2599
do_proc_dointvec_jiffies_conv(int * negp,unsigned long * lvalp,int * valp,int write,void * data)2600 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2601 int *valp,
2602 int write, void *data)
2603 {
2604 if (write) {
2605 if (*lvalp > LONG_MAX / HZ)
2606 return 1;
2607 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2608 } else {
2609 int val = *valp;
2610 unsigned long lval;
2611 if (val < 0) {
2612 *negp = -1;
2613 lval = (unsigned long)-val;
2614 } else {
2615 *negp = 0;
2616 lval = (unsigned long)val;
2617 }
2618 *lvalp = lval / HZ;
2619 }
2620 return 0;
2621 }
2622
do_proc_dointvec_userhz_jiffies_conv(int * negp,unsigned long * lvalp,int * valp,int write,void * data)2623 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2624 int *valp,
2625 int write, void *data)
2626 {
2627 if (write) {
2628 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2629 return 1;
2630 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2631 } else {
2632 int val = *valp;
2633 unsigned long lval;
2634 if (val < 0) {
2635 *negp = -1;
2636 lval = (unsigned long)-val;
2637 } else {
2638 *negp = 0;
2639 lval = (unsigned long)val;
2640 }
2641 *lvalp = jiffies_to_clock_t(lval);
2642 }
2643 return 0;
2644 }
2645
do_proc_dointvec_ms_jiffies_conv(int * negp,unsigned long * lvalp,int * valp,int write,void * data)2646 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2647 int *valp,
2648 int write, void *data)
2649 {
2650 if (write) {
2651 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2652 } else {
2653 int val = *valp;
2654 unsigned long lval;
2655 if (val < 0) {
2656 *negp = -1;
2657 lval = (unsigned long)-val;
2658 } else {
2659 *negp = 0;
2660 lval = (unsigned long)val;
2661 }
2662 *lvalp = jiffies_to_msecs(lval);
2663 }
2664 return 0;
2665 }
2666
2667 /**
2668 * proc_dointvec_jiffies - read a vector of integers as seconds
2669 * @table: the sysctl table
2670 * @write: %TRUE if this is a write to the sysctl file
2671 * @filp: the file structure
2672 * @buffer: the user buffer
2673 * @lenp: the size of the user buffer
2674 * @ppos: file position
2675 *
2676 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2677 * values from/to the user buffer, treated as an ASCII string.
2678 * The values read are assumed to be in seconds, and are converted into
2679 * jiffies.
2680 *
2681 * Returns 0 on success.
2682 */
proc_dointvec_jiffies(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2683 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2684 void __user *buffer, size_t *lenp, loff_t *ppos)
2685 {
2686 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2687 do_proc_dointvec_jiffies_conv,NULL);
2688 }
2689
2690 /**
2691 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2692 * @table: the sysctl table
2693 * @write: %TRUE if this is a write to the sysctl file
2694 * @filp: the file structure
2695 * @buffer: the user buffer
2696 * @lenp: the size of the user buffer
2697 * @ppos: pointer to the file position
2698 *
2699 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2700 * values from/to the user buffer, treated as an ASCII string.
2701 * The values read are assumed to be in 1/USER_HZ seconds, and
2702 * are converted into jiffies.
2703 *
2704 * Returns 0 on success.
2705 */
proc_dointvec_userhz_jiffies(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2706 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2707 void __user *buffer, size_t *lenp, loff_t *ppos)
2708 {
2709 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2710 do_proc_dointvec_userhz_jiffies_conv,NULL);
2711 }
2712
2713 /**
2714 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2715 * @table: the sysctl table
2716 * @write: %TRUE if this is a write to the sysctl file
2717 * @filp: the file structure
2718 * @buffer: the user buffer
2719 * @lenp: the size of the user buffer
2720 * @ppos: file position
2721 * @ppos: the current position in the file
2722 *
2723 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2724 * values from/to the user buffer, treated as an ASCII string.
2725 * The values read are assumed to be in 1/1000 seconds, and
2726 * are converted into jiffies.
2727 *
2728 * Returns 0 on success.
2729 */
proc_dointvec_ms_jiffies(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2730 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2731 void __user *buffer, size_t *lenp, loff_t *ppos)
2732 {
2733 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2734 do_proc_dointvec_ms_jiffies_conv, NULL);
2735 }
2736
proc_do_cad_pid(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2737 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2738 void __user *buffer, size_t *lenp, loff_t *ppos)
2739 {
2740 struct pid *new_pid;
2741 pid_t tmp;
2742 int r;
2743
2744 tmp = pid_vnr(cad_pid);
2745
2746 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2747 lenp, ppos, NULL, NULL);
2748 if (r || !write)
2749 return r;
2750
2751 new_pid = find_get_pid(tmp);
2752 if (!new_pid)
2753 return -ESRCH;
2754
2755 put_pid(xchg(&cad_pid, new_pid));
2756 return 0;
2757 }
2758
2759 #else /* CONFIG_PROC_FS */
2760
proc_dostring(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2761 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2762 void __user *buffer, size_t *lenp, loff_t *ppos)
2763 {
2764 return -ENOSYS;
2765 }
2766
proc_dointvec(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2767 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2768 void __user *buffer, size_t *lenp, loff_t *ppos)
2769 {
2770 return -ENOSYS;
2771 }
2772
proc_dointvec_minmax(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2773 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2774 void __user *buffer, size_t *lenp, loff_t *ppos)
2775 {
2776 return -ENOSYS;
2777 }
2778
proc_dointvec_jiffies(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2779 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2780 void __user *buffer, size_t *lenp, loff_t *ppos)
2781 {
2782 return -ENOSYS;
2783 }
2784
proc_dointvec_userhz_jiffies(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2785 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2786 void __user *buffer, size_t *lenp, loff_t *ppos)
2787 {
2788 return -ENOSYS;
2789 }
2790
proc_dointvec_ms_jiffies(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2791 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2792 void __user *buffer, size_t *lenp, loff_t *ppos)
2793 {
2794 return -ENOSYS;
2795 }
2796
proc_doulongvec_minmax(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2797 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2798 void __user *buffer, size_t *lenp, loff_t *ppos)
2799 {
2800 return -ENOSYS;
2801 }
2802
proc_doulongvec_ms_jiffies_minmax(struct ctl_table * table,int write,struct file * filp,void __user * buffer,size_t * lenp,loff_t * ppos)2803 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2804 struct file *filp,
2805 void __user *buffer,
2806 size_t *lenp, loff_t *ppos)
2807 {
2808 return -ENOSYS;
2809 }
2810
2811
2812 #endif /* CONFIG_PROC_FS */
2813
2814
2815 #ifdef CONFIG_SYSCTL_SYSCALL
2816 /*
2817 * General sysctl support routines
2818 */
2819
2820 /* The generic sysctl data routine (used if no strategy routine supplied) */
sysctl_data(struct ctl_table * table,void __user * oldval,size_t __user * oldlenp,void __user * newval,size_t newlen)2821 int sysctl_data(struct ctl_table *table,
2822 void __user *oldval, size_t __user *oldlenp,
2823 void __user *newval, size_t newlen)
2824 {
2825 size_t len;
2826
2827 /* Get out of I don't have a variable */
2828 if (!table->data || !table->maxlen)
2829 return -ENOTDIR;
2830
2831 if (oldval && oldlenp) {
2832 if (get_user(len, oldlenp))
2833 return -EFAULT;
2834 if (len) {
2835 if (len > table->maxlen)
2836 len = table->maxlen;
2837 if (copy_to_user(oldval, table->data, len))
2838 return -EFAULT;
2839 if (put_user(len, oldlenp))
2840 return -EFAULT;
2841 }
2842 }
2843
2844 if (newval && newlen) {
2845 if (newlen > table->maxlen)
2846 newlen = table->maxlen;
2847
2848 if (copy_from_user(table->data, newval, newlen))
2849 return -EFAULT;
2850 }
2851 return 1;
2852 }
2853
2854 /* The generic string strategy routine: */
sysctl_string(struct ctl_table * table,void __user * oldval,size_t __user * oldlenp,void __user * newval,size_t newlen)2855 int sysctl_string(struct ctl_table *table,
2856 void __user *oldval, size_t __user *oldlenp,
2857 void __user *newval, size_t newlen)
2858 {
2859 if (!table->data || !table->maxlen)
2860 return -ENOTDIR;
2861
2862 if (oldval && oldlenp) {
2863 size_t bufsize;
2864 if (get_user(bufsize, oldlenp))
2865 return -EFAULT;
2866 if (bufsize) {
2867 size_t len = strlen(table->data), copied;
2868
2869 /* This shouldn't trigger for a well-formed sysctl */
2870 if (len > table->maxlen)
2871 len = table->maxlen;
2872
2873 /* Copy up to a max of bufsize-1 bytes of the string */
2874 copied = (len >= bufsize) ? bufsize - 1 : len;
2875
2876 if (copy_to_user(oldval, table->data, copied) ||
2877 put_user(0, (char __user *)(oldval + copied)))
2878 return -EFAULT;
2879 if (put_user(len, oldlenp))
2880 return -EFAULT;
2881 }
2882 }
2883 if (newval && newlen) {
2884 size_t len = newlen;
2885 if (len > table->maxlen)
2886 len = table->maxlen;
2887 if(copy_from_user(table->data, newval, len))
2888 return -EFAULT;
2889 if (len == table->maxlen)
2890 len--;
2891 ((char *) table->data)[len] = 0;
2892 }
2893 return 1;
2894 }
2895
2896 /*
2897 * This function makes sure that all of the integers in the vector
2898 * are between the minimum and maximum values given in the arrays
2899 * table->extra1 and table->extra2, respectively.
2900 */
sysctl_intvec(struct ctl_table * table,void __user * oldval,size_t __user * oldlenp,void __user * newval,size_t newlen)2901 int sysctl_intvec(struct ctl_table *table,
2902 void __user *oldval, size_t __user *oldlenp,
2903 void __user *newval, size_t newlen)
2904 {
2905
2906 if (newval && newlen) {
2907 int __user *vec = (int __user *) newval;
2908 int *min = (int *) table->extra1;
2909 int *max = (int *) table->extra2;
2910 size_t length;
2911 int i;
2912
2913 if (newlen % sizeof(int) != 0)
2914 return -EINVAL;
2915
2916 if (!table->extra1 && !table->extra2)
2917 return 0;
2918
2919 if (newlen > table->maxlen)
2920 newlen = table->maxlen;
2921 length = newlen / sizeof(int);
2922
2923 for (i = 0; i < length; i++) {
2924 int value;
2925 if (get_user(value, vec + i))
2926 return -EFAULT;
2927 if (min && value < min[i])
2928 return -EINVAL;
2929 if (max && value > max[i])
2930 return -EINVAL;
2931 }
2932 }
2933 return 0;
2934 }
2935
2936 /* Strategy function to convert jiffies to seconds */
sysctl_jiffies(struct ctl_table * table,void __user * oldval,size_t __user * oldlenp,void __user * newval,size_t newlen)2937 int sysctl_jiffies(struct ctl_table *table,
2938 void __user *oldval, size_t __user *oldlenp,
2939 void __user *newval, size_t newlen)
2940 {
2941 if (oldval && oldlenp) {
2942 size_t olen;
2943
2944 if (get_user(olen, oldlenp))
2945 return -EFAULT;
2946 if (olen) {
2947 int val;
2948
2949 if (olen < sizeof(int))
2950 return -EINVAL;
2951
2952 val = *(int *)(table->data) / HZ;
2953 if (put_user(val, (int __user *)oldval))
2954 return -EFAULT;
2955 if (put_user(sizeof(int), oldlenp))
2956 return -EFAULT;
2957 }
2958 }
2959 if (newval && newlen) {
2960 int new;
2961 if (newlen != sizeof(int))
2962 return -EINVAL;
2963 if (get_user(new, (int __user *)newval))
2964 return -EFAULT;
2965 *(int *)(table->data) = new*HZ;
2966 }
2967 return 1;
2968 }
2969
2970 /* Strategy function to convert jiffies to seconds */
sysctl_ms_jiffies(struct ctl_table * table,void __user * oldval,size_t __user * oldlenp,void __user * newval,size_t newlen)2971 int sysctl_ms_jiffies(struct ctl_table *table,
2972 void __user *oldval, size_t __user *oldlenp,
2973 void __user *newval, size_t newlen)
2974 {
2975 if (oldval && oldlenp) {
2976 size_t olen;
2977
2978 if (get_user(olen, oldlenp))
2979 return -EFAULT;
2980 if (olen) {
2981 int val;
2982
2983 if (olen < sizeof(int))
2984 return -EINVAL;
2985
2986 val = jiffies_to_msecs(*(int *)(table->data));
2987 if (put_user(val, (int __user *)oldval))
2988 return -EFAULT;
2989 if (put_user(sizeof(int), oldlenp))
2990 return -EFAULT;
2991 }
2992 }
2993 if (newval && newlen) {
2994 int new;
2995 if (newlen != sizeof(int))
2996 return -EINVAL;
2997 if (get_user(new, (int __user *)newval))
2998 return -EFAULT;
2999 *(int *)(table->data) = msecs_to_jiffies(new);
3000 }
3001 return 1;
3002 }
3003
3004
3005
3006 #else /* CONFIG_SYSCTL_SYSCALL */
3007
3008
SYSCALL_DEFINE1(sysctl,struct __sysctl_args __user *,args)3009 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
3010 {
3011 struct __sysctl_args tmp;
3012 int error;
3013
3014 if (copy_from_user(&tmp, args, sizeof(tmp)))
3015 return -EFAULT;
3016
3017 error = deprecated_sysctl_warning(&tmp);
3018
3019 /* If no error reading the parameters then just -ENOSYS ... */
3020 if (!error)
3021 error = -ENOSYS;
3022
3023 return error;
3024 }
3025
sysctl_data(struct ctl_table * table,void __user * oldval,size_t __user * oldlenp,void __user * newval,size_t newlen)3026 int sysctl_data(struct ctl_table *table,
3027 void __user *oldval, size_t __user *oldlenp,
3028 void __user *newval, size_t newlen)
3029 {
3030 return -ENOSYS;
3031 }
3032
sysctl_string(struct ctl_table * table,void __user * oldval,size_t __user * oldlenp,void __user * newval,size_t newlen)3033 int sysctl_string(struct ctl_table *table,
3034 void __user *oldval, size_t __user *oldlenp,
3035 void __user *newval, size_t newlen)
3036 {
3037 return -ENOSYS;
3038 }
3039
sysctl_intvec(struct ctl_table * table,void __user * oldval,size_t __user * oldlenp,void __user * newval,size_t newlen)3040 int sysctl_intvec(struct ctl_table *table,
3041 void __user *oldval, size_t __user *oldlenp,
3042 void __user *newval, size_t newlen)
3043 {
3044 return -ENOSYS;
3045 }
3046
sysctl_jiffies(struct ctl_table * table,void __user * oldval,size_t __user * oldlenp,void __user * newval,size_t newlen)3047 int sysctl_jiffies(struct ctl_table *table,
3048 void __user *oldval, size_t __user *oldlenp,
3049 void __user *newval, size_t newlen)
3050 {
3051 return -ENOSYS;
3052 }
3053
sysctl_ms_jiffies(struct ctl_table * table,void __user * oldval,size_t __user * oldlenp,void __user * newval,size_t newlen)3054 int sysctl_ms_jiffies(struct ctl_table *table,
3055 void __user *oldval, size_t __user *oldlenp,
3056 void __user *newval, size_t newlen)
3057 {
3058 return -ENOSYS;
3059 }
3060
3061 #endif /* CONFIG_SYSCTL_SYSCALL */
3062
deprecated_sysctl_warning(struct __sysctl_args * args)3063 static int deprecated_sysctl_warning(struct __sysctl_args *args)
3064 {
3065 static int msg_count;
3066 int name[CTL_MAXNAME];
3067 int i;
3068
3069 /* Check args->nlen. */
3070 if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
3071 return -ENOTDIR;
3072
3073 /* Read in the sysctl name for better debug message logging */
3074 for (i = 0; i < args->nlen; i++)
3075 if (get_user(name[i], args->name + i))
3076 return -EFAULT;
3077
3078 /* Ignore accesses to kernel.version */
3079 if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
3080 return 0;
3081
3082 if (msg_count < 5) {
3083 msg_count++;
3084 printk(KERN_INFO
3085 "warning: process `%s' used the deprecated sysctl "
3086 "system call with ", current->comm);
3087 for (i = 0; i < args->nlen; i++)
3088 printk("%d.", name[i]);
3089 printk("\n");
3090 }
3091 return 0;
3092 }
3093
3094 /*
3095 * No sense putting this after each symbol definition, twice,
3096 * exception granted :-)
3097 */
3098 EXPORT_SYMBOL(proc_dointvec);
3099 EXPORT_SYMBOL(proc_dointvec_jiffies);
3100 EXPORT_SYMBOL(proc_dointvec_minmax);
3101 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3102 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3103 EXPORT_SYMBOL(proc_dostring);
3104 EXPORT_SYMBOL(proc_doulongvec_minmax);
3105 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3106 EXPORT_SYMBOL(register_sysctl_table);
3107 EXPORT_SYMBOL(register_sysctl_paths);
3108 EXPORT_SYMBOL(sysctl_intvec);
3109 EXPORT_SYMBOL(sysctl_jiffies);
3110 EXPORT_SYMBOL(sysctl_ms_jiffies);
3111 EXPORT_SYMBOL(sysctl_string);
3112 EXPORT_SYMBOL(sysctl_data);
3113 EXPORT_SYMBOL(unregister_sysctl_table);
3114