?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
ext2: change the default behaviour on error
[linux-2.6/lfs.git] / kernel / sysctl.c
blob5e2ad5bf88e280fdbc83b61a380909555e121a71
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/security.h>
41 #include <linux/initrd.h>
42 #include <linux/times.h>
43 #include <linux/limits.h>
44 #include <linux/dcache.h>
45 #include <linux/syscalls.h>
46 #include <linux/nfs_fs.h>
47 #include <linux/acpi.h>
48 #include <linux/reboot.h>
49
50 #include <asm/uaccess.h>
51 #include <asm/processor.h>
52
53 #ifdef CONFIG_X86
54 #include <asm/nmi.h>
55 #include <asm/stacktrace.h>
56 #include <asm/io.h>
57 #endif
58
59 static int deprecated_sysctl_warning(struct __sysctl_args *args);
60
61 #if defined(CONFIG_SYSCTL)
62
63 /* External variables not in a header file. */
64 extern int C_A_D;
65 extern int print_fatal_signals;
66 extern int sysctl_overcommit_memory;
67 extern int sysctl_overcommit_ratio;
68 extern int sysctl_panic_on_oom;
69 extern int sysctl_oom_kill_allocating_task;
70 extern int max_threads;
71 extern int core_uses_pid;
72 extern int suid_dumpable;
73 extern char core_pattern[];
74 extern int pid_max;
75 extern int min_free_kbytes;
76 extern int printk_ratelimit_jiffies;
77 extern int printk_ratelimit_burst;
78 extern int pid_max_min, pid_max_max;
79 extern int sysctl_drop_caches;
80 extern int percpu_pagelist_fraction;
81 extern int compat_log;
82 extern int maps_protect;
83 extern int sysctl_stat_interval;
84 extern int latencytop_enabled;
85
86 /* Constants used for minimum and maximum */
87 #if defined(CONFIG_DETECT_SOFTLOCKUP) || defined(CONFIG_HIGHMEM)
88 static int one = 1;
89 #endif
90
91 #ifdef CONFIG_DETECT_SOFTLOCKUP
92 static int sixty = 60;
93 #endif
94
95 #ifdef CONFIG_MMU
96 static int two = 2;
97 #endif
98
99 static int zero;
100 static int one_hundred = 100;
101
102 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
103 static int maxolduid = 65535;
104 static int minolduid;
105 static int min_percpu_pagelist_fract = 8;
106
107 static int ngroups_max = NGROUPS_MAX;
108
109 #ifdef CONFIG_KMOD
110 extern char modprobe_path[];
111 #endif
112 #ifdef CONFIG_CHR_DEV_SG
113 extern int sg_big_buff;
114 #endif
115
116 #ifdef __sparc__
117 extern char reboot_command [];
118 extern int stop_a_enabled;
119 extern int scons_pwroff;
120 #endif
121
122 #ifdef __hppa__
123 extern int pwrsw_enabled;
124 extern int unaligned_enabled;
125 #endif
126
127 #ifdef CONFIG_S390
128 #ifdef CONFIG_MATHEMU
129 extern int sysctl_ieee_emulation_warnings;
130 #endif
131 extern int sysctl_userprocess_debug;
132 extern int spin_retry;
133 #endif
134
135 extern int sysctl_hz_timer;
136
137 #ifdef CONFIG_BSD_PROCESS_ACCT
138 extern int acct_parm[];
139 #endif
140
141 #ifdef CONFIG_IA64
142 extern int no_unaligned_warning;
143 #endif
144
145 #ifdef CONFIG_RT_MUTEXES
146 extern int max_lock_depth;
147 #endif
148
149 #ifdef CONFIG_SYSCTL_SYSCALL
150 static int parse_table(int __user *, int, void __user *, size_t __user *,
151 void __user *, size_t, struct ctl_table *);
152 #endif
153
154
155 #ifdef CONFIG_PROC_SYSCTL
156 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
157 void __user *buffer, size_t *lenp, loff_t *ppos);
158 static int proc_dointvec_taint(struct ctl_table *table, int write, struct file *filp,
159 void __user *buffer, size_t *lenp, loff_t *ppos);
160 #endif
161
162 static struct ctl_table root_table[];
163 static struct ctl_table_root sysctl_table_root;
164 static struct ctl_table_header root_table_header = {
165 .ctl_table = root_table,
166 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.header_list),
167 .root = &sysctl_table_root,
168 };
169 static struct ctl_table_root sysctl_table_root = {
170 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
171 .header_list = LIST_HEAD_INIT(root_table_header.ctl_entry),
172 };
173
174 static struct ctl_table kern_table[];
175 static struct ctl_table vm_table[];
176 static struct ctl_table fs_table[];
177 static struct ctl_table debug_table[];
178 static struct ctl_table dev_table[];
179 extern struct ctl_table random_table[];
180 #ifdef CONFIG_INOTIFY_USER
181 extern struct ctl_table inotify_table[];
182 #endif
183
184 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
185 int sysctl_legacy_va_layout;
186 #endif
187
188 extern int prove_locking;
189 extern int lock_stat;
190
191 /* The default sysctl tables: */
192
193 static struct ctl_table root_table[] = {
194 {
195 .ctl_name = CTL_KERN,
196 .procname = "kernel",
197 .mode = 0555,
198 .child = kern_table,
199 },
200 {
201 .ctl_name = CTL_VM,
202 .procname = "vm",
203 .mode = 0555,
204 .child = vm_table,
205 },
206 {
207 .ctl_name = CTL_FS,
208 .procname = "fs",
209 .mode = 0555,
210 .child = fs_table,
211 },
212 {
213 .ctl_name = CTL_DEBUG,
214 .procname = "debug",
215 .mode = 0555,
216 .child = debug_table,
217 },
218 {
219 .ctl_name = CTL_DEV,
220 .procname = "dev",
221 .mode = 0555,
222 .child = dev_table,
223 },
224 /*
225 * NOTE: do not add new entries to this table unless you have read
226 * Documentation/sysctl/ctl_unnumbered.txt
227 */
228 { .ctl_name = 0 }
229 };
230
231 #ifdef CONFIG_SCHED_DEBUG
232 static int min_sched_granularity_ns = 100000; /* 100 usecs */
233 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
234 static int min_wakeup_granularity_ns; /* 0 usecs */
235 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
236 #endif
237
238 static struct ctl_table kern_table[] = {
239 #ifdef CONFIG_SCHED_DEBUG
240 {
241 .ctl_name = CTL_UNNUMBERED,
242 .procname = "sched_min_granularity_ns",
243 .data = &sysctl_sched_min_granularity,
244 .maxlen = sizeof(unsigned int),
245 .mode = 0644,
246 .proc_handler = &sched_nr_latency_handler,
247 .strategy = &sysctl_intvec,
248 .extra1 = &min_sched_granularity_ns,
249 .extra2 = &max_sched_granularity_ns,
250 },
251 {
252 .ctl_name = CTL_UNNUMBERED,
253 .procname = "sched_latency_ns",
254 .data = &sysctl_sched_latency,
255 .maxlen = sizeof(unsigned int),
256 .mode = 0644,
257 .proc_handler = &sched_nr_latency_handler,
258 .strategy = &sysctl_intvec,
259 .extra1 = &min_sched_granularity_ns,
260 .extra2 = &max_sched_granularity_ns,
261 },
262 {
263 .ctl_name = CTL_UNNUMBERED,
264 .procname = "sched_wakeup_granularity_ns",
265 .data = &sysctl_sched_wakeup_granularity,
266 .maxlen = sizeof(unsigned int),
267 .mode = 0644,
268 .proc_handler = &proc_dointvec_minmax,
269 .strategy = &sysctl_intvec,
270 .extra1 = &min_wakeup_granularity_ns,
271 .extra2 = &max_wakeup_granularity_ns,
272 },
273 {
274 .ctl_name = CTL_UNNUMBERED,
275 .procname = "sched_batch_wakeup_granularity_ns",
276 .data = &sysctl_sched_batch_wakeup_granularity,
277 .maxlen = sizeof(unsigned int),
278 .mode = 0644,
279 .proc_handler = &proc_dointvec_minmax,
280 .strategy = &sysctl_intvec,
281 .extra1 = &min_wakeup_granularity_ns,
282 .extra2 = &max_wakeup_granularity_ns,
283 },
284 {
285 .ctl_name = CTL_UNNUMBERED,
286 .procname = "sched_child_runs_first",
287 .data = &sysctl_sched_child_runs_first,
288 .maxlen = sizeof(unsigned int),
289 .mode = 0644,
290 .proc_handler = &proc_dointvec,
291 },
292 {
293 .ctl_name = CTL_UNNUMBERED,
294 .procname = "sched_features",
295 .data = &sysctl_sched_features,
296 .maxlen = sizeof(unsigned int),
297 .mode = 0644,
298 .proc_handler = &proc_dointvec,
299 },
300 {
301 .ctl_name = CTL_UNNUMBERED,
302 .procname = "sched_migration_cost",
303 .data = &sysctl_sched_migration_cost,
304 .maxlen = sizeof(unsigned int),
305 .mode = 0644,
306 .proc_handler = &proc_dointvec,
307 },
308 {
309 .ctl_name = CTL_UNNUMBERED,
310 .procname = "sched_nr_migrate",
311 .data = &sysctl_sched_nr_migrate,
312 .maxlen = sizeof(unsigned int),
313 .mode = 0644,
314 .proc_handler = &proc_dointvec,
315 },
316 {
317 .ctl_name = CTL_UNNUMBERED,
318 .procname = "sched_rt_period_ms",
319 .data = &sysctl_sched_rt_period,
320 .maxlen = sizeof(unsigned int),
321 .mode = 0644,
322 .proc_handler = &proc_dointvec,
323 },
324 {
325 .ctl_name = CTL_UNNUMBERED,
326 .procname = "sched_rt_ratio",
327 .data = &sysctl_sched_rt_ratio,
328 .maxlen = sizeof(unsigned int),
329 .mode = 0644,
330 .proc_handler = &proc_dointvec,
331 },
332 #if defined(CONFIG_FAIR_GROUP_SCHED) && defined(CONFIG_SMP)
333 {
334 .ctl_name = CTL_UNNUMBERED,
335 .procname = "sched_min_bal_int_shares",
336 .data = &sysctl_sched_min_bal_int_shares,
337 .maxlen = sizeof(unsigned int),
338 .mode = 0644,
339 .proc_handler = &proc_dointvec,
340 },
341 {
342 .ctl_name = CTL_UNNUMBERED,
343 .procname = "sched_max_bal_int_shares",
344 .data = &sysctl_sched_max_bal_int_shares,
345 .maxlen = sizeof(unsigned int),
346 .mode = 0644,
347 .proc_handler = &proc_dointvec,
348 },
349 #endif
350 #endif
351 {
352 .ctl_name = CTL_UNNUMBERED,
353 .procname = "sched_compat_yield",
354 .data = &sysctl_sched_compat_yield,
355 .maxlen = sizeof(unsigned int),
356 .mode = 0644,
357 .proc_handler = &proc_dointvec,
358 },
359 #ifdef CONFIG_PROVE_LOCKING
360 {
361 .ctl_name = CTL_UNNUMBERED,
362 .procname = "prove_locking",
363 .data = &prove_locking,
364 .maxlen = sizeof(int),
365 .mode = 0644,
366 .proc_handler = &proc_dointvec,
367 },
368 #endif
369 #ifdef CONFIG_LOCK_STAT
370 {
371 .ctl_name = CTL_UNNUMBERED,
372 .procname = "lock_stat",
373 .data = &lock_stat,
374 .maxlen = sizeof(int),
375 .mode = 0644,
376 .proc_handler = &proc_dointvec,
377 },
378 #endif
379 {
380 .ctl_name = KERN_PANIC,
381 .procname = "panic",
382 .data = &panic_timeout,
383 .maxlen = sizeof(int),
384 .mode = 0644,
385 .proc_handler = &proc_dointvec,
386 },
387 {
388 .ctl_name = KERN_CORE_USES_PID,
389 .procname = "core_uses_pid",
390 .data = &core_uses_pid,
391 .maxlen = sizeof(int),
392 .mode = 0644,
393 .proc_handler = &proc_dointvec,
394 },
395 {
396 .ctl_name = KERN_CORE_PATTERN,
397 .procname = "core_pattern",
398 .data = core_pattern,
399 .maxlen = CORENAME_MAX_SIZE,
400 .mode = 0644,
401 .proc_handler = &proc_dostring,
402 .strategy = &sysctl_string,
403 },
404 #ifdef CONFIG_PROC_SYSCTL
405 {
406 .procname = "tainted",
407 .data = &tainted,
408 .maxlen = sizeof(int),
409 .mode = 0644,
410 .proc_handler = &proc_dointvec_taint,
411 },
412 #endif
413 #ifdef CONFIG_LATENCYTOP
414 {
415 .procname = "latencytop",
416 .data = &latencytop_enabled,
417 .maxlen = sizeof(int),
418 .mode = 0644,
419 .proc_handler = &proc_dointvec,
420 },
421 #endif
422 #ifdef CONFIG_BLK_DEV_INITRD
423 {
424 .ctl_name = KERN_REALROOTDEV,
425 .procname = "real-root-dev",
426 .data = &real_root_dev,
427 .maxlen = sizeof(int),
428 .mode = 0644,
429 .proc_handler = &proc_dointvec,
430 },
431 #endif
432 {
433 .ctl_name = CTL_UNNUMBERED,
434 .procname = "print-fatal-signals",
435 .data = &print_fatal_signals,
436 .maxlen = sizeof(int),
437 .mode = 0644,
438 .proc_handler = &proc_dointvec,
439 },
440 #ifdef __sparc__
441 {
442 .ctl_name = KERN_SPARC_REBOOT,
443 .procname = "reboot-cmd",
444 .data = reboot_command,
445 .maxlen = 256,
446 .mode = 0644,
447 .proc_handler = &proc_dostring,
448 .strategy = &sysctl_string,
449 },
450 {
451 .ctl_name = KERN_SPARC_STOP_A,
452 .procname = "stop-a",
453 .data = &stop_a_enabled,
454 .maxlen = sizeof (int),
455 .mode = 0644,
456 .proc_handler = &proc_dointvec,
457 },
458 {
459 .ctl_name = KERN_SPARC_SCONS_PWROFF,
460 .procname = "scons-poweroff",
461 .data = &scons_pwroff,
462 .maxlen = sizeof (int),
463 .mode = 0644,
464 .proc_handler = &proc_dointvec,
465 },
466 #endif
467 #ifdef __hppa__
468 {
469 .ctl_name = KERN_HPPA_PWRSW,
470 .procname = "soft-power",
471 .data = &pwrsw_enabled,
472 .maxlen = sizeof (int),
473 .mode = 0644,
474 .proc_handler = &proc_dointvec,
475 },
476 {
477 .ctl_name = KERN_HPPA_UNALIGNED,
478 .procname = "unaligned-trap",
479 .data = &unaligned_enabled,
480 .maxlen = sizeof (int),
481 .mode = 0644,
482 .proc_handler = &proc_dointvec,
483 },
484 #endif
485 {
486 .ctl_name = KERN_CTLALTDEL,
487 .procname = "ctrl-alt-del",
488 .data = &C_A_D,
489 .maxlen = sizeof(int),
490 .mode = 0644,
491 .proc_handler = &proc_dointvec,
492 },
493 {
494 .ctl_name = KERN_PRINTK,
495 .procname = "printk",
496 .data = &console_loglevel,
497 .maxlen = 4*sizeof(int),
498 .mode = 0644,
499 .proc_handler = &proc_dointvec,
500 },
501 #ifdef CONFIG_KMOD
502 {
503 .ctl_name = KERN_MODPROBE,
504 .procname = "modprobe",
505 .data = &modprobe_path,
506 .maxlen = KMOD_PATH_LEN,
507 .mode = 0644,
508 .proc_handler = &proc_dostring,
509 .strategy = &sysctl_string,
510 },
511 #endif
512 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
513 {
514 .ctl_name = KERN_HOTPLUG,
515 .procname = "hotplug",
516 .data = &uevent_helper,
517 .maxlen = UEVENT_HELPER_PATH_LEN,
518 .mode = 0644,
519 .proc_handler = &proc_dostring,
520 .strategy = &sysctl_string,
521 },
522 #endif
523 #ifdef CONFIG_CHR_DEV_SG
524 {
525 .ctl_name = KERN_SG_BIG_BUFF,
526 .procname = "sg-big-buff",
527 .data = &sg_big_buff,
528 .maxlen = sizeof (int),
529 .mode = 0444,
530 .proc_handler = &proc_dointvec,
531 },
532 #endif
533 #ifdef CONFIG_BSD_PROCESS_ACCT
534 {
535 .ctl_name = KERN_ACCT,
536 .procname = "acct",
537 .data = &acct_parm,
538 .maxlen = 3*sizeof(int),
539 .mode = 0644,
540 .proc_handler = &proc_dointvec,
541 },
542 #endif
543 #ifdef CONFIG_MAGIC_SYSRQ
544 {
545 .ctl_name = KERN_SYSRQ,
546 .procname = "sysrq",
547 .data = &__sysrq_enabled,
548 .maxlen = sizeof (int),
549 .mode = 0644,
550 .proc_handler = &proc_dointvec,
551 },
552 #endif
553 #ifdef CONFIG_PROC_SYSCTL
554 {
555 .procname = "cad_pid",
556 .data = NULL,
557 .maxlen = sizeof (int),
558 .mode = 0600,
559 .proc_handler = &proc_do_cad_pid,
560 },
561 #endif
562 {
563 .ctl_name = KERN_MAX_THREADS,
564 .procname = "threads-max",
565 .data = &max_threads,
566 .maxlen = sizeof(int),
567 .mode = 0644,
568 .proc_handler = &proc_dointvec,
569 },
570 {
571 .ctl_name = KERN_RANDOM,
572 .procname = "random",
573 .mode = 0555,
574 .child = random_table,
575 },
576 {
577 .ctl_name = KERN_OVERFLOWUID,
578 .procname = "overflowuid",
579 .data = &overflowuid,
580 .maxlen = sizeof(int),
581 .mode = 0644,
582 .proc_handler = &proc_dointvec_minmax,
583 .strategy = &sysctl_intvec,
584 .extra1 = &minolduid,
585 .extra2 = &maxolduid,
586 },
587 {
588 .ctl_name = KERN_OVERFLOWGID,
589 .procname = "overflowgid",
590 .data = &overflowgid,
591 .maxlen = sizeof(int),
592 .mode = 0644,
593 .proc_handler = &proc_dointvec_minmax,
594 .strategy = &sysctl_intvec,
595 .extra1 = &minolduid,
596 .extra2 = &maxolduid,
597 },
598 #ifdef CONFIG_S390
599 #ifdef CONFIG_MATHEMU
600 {
601 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
602 .procname = "ieee_emulation_warnings",
603 .data = &sysctl_ieee_emulation_warnings,
604 .maxlen = sizeof(int),
605 .mode = 0644,
606 .proc_handler = &proc_dointvec,
607 },
608 #endif
609 #ifdef CONFIG_NO_IDLE_HZ
610 {
611 .ctl_name = KERN_HZ_TIMER,
612 .procname = "hz_timer",
613 .data = &sysctl_hz_timer,
614 .maxlen = sizeof(int),
615 .mode = 0644,
616 .proc_handler = &proc_dointvec,
617 },
618 #endif
619 {
620 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
621 .procname = "userprocess_debug",
622 .data = &sysctl_userprocess_debug,
623 .maxlen = sizeof(int),
624 .mode = 0644,
625 .proc_handler = &proc_dointvec,
626 },
627 #endif
628 {
629 .ctl_name = KERN_PIDMAX,
630 .procname = "pid_max",
631 .data = &pid_max,
632 .maxlen = sizeof (int),
633 .mode = 0644,
634 .proc_handler = &proc_dointvec_minmax,
635 .strategy = sysctl_intvec,
636 .extra1 = &pid_max_min,
637 .extra2 = &pid_max_max,
638 },
639 {
640 .ctl_name = KERN_PANIC_ON_OOPS,
641 .procname = "panic_on_oops",
642 .data = &panic_on_oops,
643 .maxlen = sizeof(int),
644 .mode = 0644,
645 .proc_handler = &proc_dointvec,
646 },
647 {
648 .ctl_name = KERN_PRINTK_RATELIMIT,
649 .procname = "printk_ratelimit",
650 .data = &printk_ratelimit_jiffies,
651 .maxlen = sizeof(int),
652 .mode = 0644,
653 .proc_handler = &proc_dointvec_jiffies,
654 .strategy = &sysctl_jiffies,
655 },
656 {
657 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
658 .procname = "printk_ratelimit_burst",
659 .data = &printk_ratelimit_burst,
660 .maxlen = sizeof(int),
661 .mode = 0644,
662 .proc_handler = &proc_dointvec,
663 },
664 {
665 .ctl_name = KERN_NGROUPS_MAX,
666 .procname = "ngroups_max",
667 .data = &ngroups_max,
668 .maxlen = sizeof (int),
669 .mode = 0444,
670 .proc_handler = &proc_dointvec,
671 },
672 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
673 {
674 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
675 .procname = "unknown_nmi_panic",
676 .data = &unknown_nmi_panic,
677 .maxlen = sizeof (int),
678 .mode = 0644,
679 .proc_handler = &proc_dointvec,
680 },
681 {
682 .procname = "nmi_watchdog",
683 .data = &nmi_watchdog_enabled,
684 .maxlen = sizeof (int),
685 .mode = 0644,
686 .proc_handler = &proc_nmi_enabled,
687 },
688 #endif
689 #if defined(CONFIG_X86)
690 {
691 .ctl_name = KERN_PANIC_ON_NMI,
692 .procname = "panic_on_unrecovered_nmi",
693 .data = &panic_on_unrecovered_nmi,
694 .maxlen = sizeof(int),
695 .mode = 0644,
696 .proc_handler = &proc_dointvec,
697 },
698 {
699 .ctl_name = KERN_BOOTLOADER_TYPE,
700 .procname = "bootloader_type",
701 .data = &bootloader_type,
702 .maxlen = sizeof (int),
703 .mode = 0444,
704 .proc_handler = &proc_dointvec,
705 },
706 {
707 .ctl_name = CTL_UNNUMBERED,
708 .procname = "kstack_depth_to_print",
709 .data = &kstack_depth_to_print,
710 .maxlen = sizeof(int),
711 .mode = 0644,
712 .proc_handler = &proc_dointvec,
713 },
714 {
715 .ctl_name = CTL_UNNUMBERED,
716 .procname = "io_delay_type",
717 .data = &io_delay_type,
718 .maxlen = sizeof(int),
719 .mode = 0644,
720 .proc_handler = &proc_dointvec,
721 },
722 #endif
723 #if defined(CONFIG_MMU)
724 {
725 .ctl_name = KERN_RANDOMIZE,
726 .procname = "randomize_va_space",
727 .data = &randomize_va_space,
728 .maxlen = sizeof(int),
729 .mode = 0644,
730 .proc_handler = &proc_dointvec,
731 },
732 #endif
733 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
734 {
735 .ctl_name = KERN_SPIN_RETRY,
736 .procname = "spin_retry",
737 .data = &spin_retry,
738 .maxlen = sizeof (int),
739 .mode = 0644,
740 .proc_handler = &proc_dointvec,
741 },
742 #endif
743 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
744 {
745 .procname = "acpi_video_flags",
746 .data = &acpi_realmode_flags,
747 .maxlen = sizeof (unsigned long),
748 .mode = 0644,
749 .proc_handler = &proc_doulongvec_minmax,
750 },
751 #endif
752 #ifdef CONFIG_IA64
753 {
754 .ctl_name = KERN_IA64_UNALIGNED,
755 .procname = "ignore-unaligned-usertrap",
756 .data = &no_unaligned_warning,
757 .maxlen = sizeof (int),
758 .mode = 0644,
759 .proc_handler = &proc_dointvec,
760 },
761 #endif
762 #ifdef CONFIG_DETECT_SOFTLOCKUP
763 {
764 .ctl_name = CTL_UNNUMBERED,
765 .procname = "softlockup_thresh",
766 .data = &softlockup_thresh,
767 .maxlen = sizeof(unsigned long),
768 .mode = 0644,
769 .proc_handler = &proc_doulongvec_minmax,
770 .strategy = &sysctl_intvec,
771 .extra1 = &one,
772 .extra2 = &sixty,
773 },
774 {
775 .ctl_name = CTL_UNNUMBERED,
776 .procname = "hung_task_check_count",
777 .data = &sysctl_hung_task_check_count,
778 .maxlen = sizeof(unsigned long),
779 .mode = 0644,
780 .proc_handler = &proc_doulongvec_minmax,
781 .strategy = &sysctl_intvec,
782 },
783 {
784 .ctl_name = CTL_UNNUMBERED,
785 .procname = "hung_task_timeout_secs",
786 .data = &sysctl_hung_task_timeout_secs,
787 .maxlen = sizeof(unsigned long),
788 .mode = 0644,
789 .proc_handler = &proc_doulongvec_minmax,
790 .strategy = &sysctl_intvec,
791 },
792 {
793 .ctl_name = CTL_UNNUMBERED,
794 .procname = "hung_task_warnings",
795 .data = &sysctl_hung_task_warnings,
796 .maxlen = sizeof(unsigned long),
797 .mode = 0644,
798 .proc_handler = &proc_doulongvec_minmax,
799 .strategy = &sysctl_intvec,
800 },
801 #endif
802 #ifdef CONFIG_COMPAT
803 {
804 .ctl_name = KERN_COMPAT_LOG,
805 .procname = "compat-log",
806 .data = &compat_log,
807 .maxlen = sizeof (int),
808 .mode = 0644,
809 .proc_handler = &proc_dointvec,
810 },
811 #endif
812 #ifdef CONFIG_RT_MUTEXES
813 {
814 .ctl_name = KERN_MAX_LOCK_DEPTH,
815 .procname = "max_lock_depth",
816 .data = &max_lock_depth,
817 .maxlen = sizeof(int),
818 .mode = 0644,
819 .proc_handler = &proc_dointvec,
820 },
821 #endif
822 #ifdef CONFIG_PROC_FS
823 {
824 .ctl_name = CTL_UNNUMBERED,
825 .procname = "maps_protect",
826 .data = &maps_protect,
827 .maxlen = sizeof(int),
828 .mode = 0644,
829 .proc_handler = &proc_dointvec,
830 },
831 #endif
832 {
833 .ctl_name = CTL_UNNUMBERED,
834 .procname = "poweroff_cmd",
835 .data = &poweroff_cmd,
836 .maxlen = POWEROFF_CMD_PATH_LEN,
837 .mode = 0644,
838 .proc_handler = &proc_dostring,
839 .strategy = &sysctl_string,
840 },
841 /*
842 * NOTE: do not add new entries to this table unless you have read
843 * Documentation/sysctl/ctl_unnumbered.txt
844 */
845 { .ctl_name = 0 }
846 };
847
848 static struct ctl_table vm_table[] = {
849 {
850 .ctl_name = VM_OVERCOMMIT_MEMORY,
851 .procname = "overcommit_memory",
852 .data = &sysctl_overcommit_memory,
853 .maxlen = sizeof(sysctl_overcommit_memory),
854 .mode = 0644,
855 .proc_handler = &proc_dointvec,
856 },
857 {
858 .ctl_name = VM_PANIC_ON_OOM,
859 .procname = "panic_on_oom",
860 .data = &sysctl_panic_on_oom,
861 .maxlen = sizeof(sysctl_panic_on_oom),
862 .mode = 0644,
863 .proc_handler = &proc_dointvec,
864 },
865 {
866 .ctl_name = CTL_UNNUMBERED,
867 .procname = "oom_kill_allocating_task",
868 .data = &sysctl_oom_kill_allocating_task,
869 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
870 .mode = 0644,
871 .proc_handler = &proc_dointvec,
872 },
873 {
874 .ctl_name = VM_OVERCOMMIT_RATIO,
875 .procname = "overcommit_ratio",
876 .data = &sysctl_overcommit_ratio,
877 .maxlen = sizeof(sysctl_overcommit_ratio),
878 .mode = 0644,
879 .proc_handler = &proc_dointvec,
880 },
881 {
882 .ctl_name = VM_PAGE_CLUSTER,
883 .procname = "page-cluster",
884 .data = &page_cluster,
885 .maxlen = sizeof(int),
886 .mode = 0644,
887 .proc_handler = &proc_dointvec,
888 },
889 {
890 .ctl_name = VM_DIRTY_BACKGROUND,
891 .procname = "dirty_background_ratio",
892 .data = &dirty_background_ratio,
893 .maxlen = sizeof(dirty_background_ratio),
894 .mode = 0644,
895 .proc_handler = &proc_dointvec_minmax,
896 .strategy = &sysctl_intvec,
897 .extra1 = &zero,
898 .extra2 = &one_hundred,
899 },
900 {
901 .ctl_name = VM_DIRTY_RATIO,
902 .procname = "dirty_ratio",
903 .data = &vm_dirty_ratio,
904 .maxlen = sizeof(vm_dirty_ratio),
905 .mode = 0644,
906 .proc_handler = &dirty_ratio_handler,
907 .strategy = &sysctl_intvec,
908 .extra1 = &zero,
909 .extra2 = &one_hundred,
910 },
911 {
912 .procname = "dirty_writeback_centisecs",
913 .data = &dirty_writeback_interval,
914 .maxlen = sizeof(dirty_writeback_interval),
915 .mode = 0644,
916 .proc_handler = &dirty_writeback_centisecs_handler,
917 },
918 {
919 .procname = "dirty_expire_centisecs",
920 .data = &dirty_expire_interval,
921 .maxlen = sizeof(dirty_expire_interval),
922 .mode = 0644,
923 .proc_handler = &proc_dointvec_userhz_jiffies,
924 },
925 {
926 .ctl_name = VM_NR_PDFLUSH_THREADS,
927 .procname = "nr_pdflush_threads",
928 .data = &nr_pdflush_threads,
929 .maxlen = sizeof nr_pdflush_threads,
930 .mode = 0444 /* read-only*/,
931 .proc_handler = &proc_dointvec,
932 },
933 {
934 .ctl_name = VM_SWAPPINESS,
935 .procname = "swappiness",
936 .data = &vm_swappiness,
937 .maxlen = sizeof(vm_swappiness),
938 .mode = 0644,
939 .proc_handler = &proc_dointvec_minmax,
940 .strategy = &sysctl_intvec,
941 .extra1 = &zero,
942 .extra2 = &one_hundred,
943 },
944 #ifdef CONFIG_HUGETLB_PAGE
945 {
946 .procname = "nr_hugepages",
947 .data = &max_huge_pages,
948 .maxlen = sizeof(unsigned long),
949 .mode = 0644,
950 .proc_handler = &hugetlb_sysctl_handler,
951 .extra1 = (void *)&hugetlb_zero,
952 .extra2 = (void *)&hugetlb_infinity,
953 },
954 {
955 .ctl_name = VM_HUGETLB_GROUP,
956 .procname = "hugetlb_shm_group",
957 .data = &sysctl_hugetlb_shm_group,
958 .maxlen = sizeof(gid_t),
959 .mode = 0644,
960 .proc_handler = &proc_dointvec,
961 },
962 {
963 .ctl_name = CTL_UNNUMBERED,
964 .procname = "hugepages_treat_as_movable",
965 .data = &hugepages_treat_as_movable,
966 .maxlen = sizeof(int),
967 .mode = 0644,
968 .proc_handler = &hugetlb_treat_movable_handler,
969 },
970 {
971 .ctl_name = CTL_UNNUMBERED,
972 .procname = "nr_overcommit_hugepages",
973 .data = &nr_overcommit_huge_pages,
974 .maxlen = sizeof(nr_overcommit_huge_pages),
975 .mode = 0644,
976 .proc_handler = &proc_doulongvec_minmax,
977 },
978 #endif
979 {
980 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
981 .procname = "lowmem_reserve_ratio",
982 .data = &sysctl_lowmem_reserve_ratio,
983 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
984 .mode = 0644,
985 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
986 .strategy = &sysctl_intvec,
987 },
988 {
989 .ctl_name = VM_DROP_PAGECACHE,
990 .procname = "drop_caches",
991 .data = &sysctl_drop_caches,
992 .maxlen = sizeof(int),
993 .mode = 0644,
994 .proc_handler = drop_caches_sysctl_handler,
995 .strategy = &sysctl_intvec,
996 },
997 {
998 .ctl_name = VM_MIN_FREE_KBYTES,
999 .procname = "min_free_kbytes",
1000 .data = &min_free_kbytes,
1001 .maxlen = sizeof(min_free_kbytes),
1002 .mode = 0644,
1003 .proc_handler = &min_free_kbytes_sysctl_handler,
1004 .strategy = &sysctl_intvec,
1005 .extra1 = &zero,
1006 },
1007 {
1008 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
1009 .procname = "percpu_pagelist_fraction",
1010 .data = &percpu_pagelist_fraction,
1011 .maxlen = sizeof(percpu_pagelist_fraction),
1012 .mode = 0644,
1013 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
1014 .strategy = &sysctl_intvec,
1015 .extra1 = &min_percpu_pagelist_fract,
1016 },
1017 #ifdef CONFIG_MMU
1018 {
1019 .ctl_name = VM_MAX_MAP_COUNT,
1020 .procname = "max_map_count",
1021 .data = &sysctl_max_map_count,
1022 .maxlen = sizeof(sysctl_max_map_count),
1023 .mode = 0644,
1024 .proc_handler = &proc_dointvec
1025 },
1026 #endif
1027 {
1028 .ctl_name = VM_LAPTOP_MODE,
1029 .procname = "laptop_mode",
1030 .data = &laptop_mode,
1031 .maxlen = sizeof(laptop_mode),
1032 .mode = 0644,
1033 .proc_handler = &proc_dointvec_jiffies,
1034 .strategy = &sysctl_jiffies,
1035 },
1036 {
1037 .ctl_name = VM_BLOCK_DUMP,
1038 .procname = "block_dump",
1039 .data = &block_dump,
1040 .maxlen = sizeof(block_dump),
1041 .mode = 0644,
1042 .proc_handler = &proc_dointvec,
1043 .strategy = &sysctl_intvec,
1044 .extra1 = &zero,
1045 },
1046 {
1047 .ctl_name = VM_VFS_CACHE_PRESSURE,
1048 .procname = "vfs_cache_pressure",
1049 .data = &sysctl_vfs_cache_pressure,
1050 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1051 .mode = 0644,
1052 .proc_handler = &proc_dointvec,
1053 .strategy = &sysctl_intvec,
1054 .extra1 = &zero,
1055 },
1056 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1057 {
1058 .ctl_name = VM_LEGACY_VA_LAYOUT,
1059 .procname = "legacy_va_layout",
1060 .data = &sysctl_legacy_va_layout,
1061 .maxlen = sizeof(sysctl_legacy_va_layout),
1062 .mode = 0644,
1063 .proc_handler = &proc_dointvec,
1064 .strategy = &sysctl_intvec,
1065 .extra1 = &zero,
1066 },
1067 #endif
1068 #ifdef CONFIG_NUMA
1069 {
1070 .ctl_name = VM_ZONE_RECLAIM_MODE,
1071 .procname = "zone_reclaim_mode",
1072 .data = &zone_reclaim_mode,
1073 .maxlen = sizeof(zone_reclaim_mode),
1074 .mode = 0644,
1075 .proc_handler = &proc_dointvec,
1076 .strategy = &sysctl_intvec,
1077 .extra1 = &zero,
1078 },
1079 {
1080 .ctl_name = VM_MIN_UNMAPPED,
1081 .procname = "min_unmapped_ratio",
1082 .data = &sysctl_min_unmapped_ratio,
1083 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1084 .mode = 0644,
1085 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
1086 .strategy = &sysctl_intvec,
1087 .extra1 = &zero,
1088 .extra2 = &one_hundred,
1089 },
1090 {
1091 .ctl_name = VM_MIN_SLAB,
1092 .procname = "min_slab_ratio",
1093 .data = &sysctl_min_slab_ratio,
1094 .maxlen = sizeof(sysctl_min_slab_ratio),
1095 .mode = 0644,
1096 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1097 .strategy = &sysctl_intvec,
1098 .extra1 = &zero,
1099 .extra2 = &one_hundred,
1100 },
1101 #endif
1102 #ifdef CONFIG_SMP
1103 {
1104 .ctl_name = CTL_UNNUMBERED,
1105 .procname = "stat_interval",
1106 .data = &sysctl_stat_interval,
1107 .maxlen = sizeof(sysctl_stat_interval),
1108 .mode = 0644,
1109 .proc_handler = &proc_dointvec_jiffies,
1110 .strategy = &sysctl_jiffies,
1111 },
1112 #endif
1113 #ifdef CONFIG_SECURITY
1114 {
1115 .ctl_name = CTL_UNNUMBERED,
1116 .procname = "mmap_min_addr",
1117 .data = &mmap_min_addr,
1118 .maxlen = sizeof(unsigned long),
1119 .mode = 0644,
1120 .proc_handler = &proc_doulongvec_minmax,
1121 },
1122 #endif
1123 #ifdef CONFIG_NUMA
1124 {
1125 .ctl_name = CTL_UNNUMBERED,
1126 .procname = "numa_zonelist_order",
1127 .data = &numa_zonelist_order,
1128 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1129 .mode = 0644,
1130 .proc_handler = &numa_zonelist_order_handler,
1131 .strategy = &sysctl_string,
1132 },
1133 #endif
1134 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1135 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1136 {
1137 .ctl_name = VM_VDSO_ENABLED,
1138 .procname = "vdso_enabled",
1139 .data = &vdso_enabled,
1140 .maxlen = sizeof(vdso_enabled),
1141 .mode = 0644,
1142 .proc_handler = &proc_dointvec,
1143 .strategy = &sysctl_intvec,
1144 .extra1 = &zero,
1145 },
1146 #endif
1147 #ifdef CONFIG_HIGHMEM
1148 {
1149 .ctl_name = CTL_UNNUMBERED,
1150 .procname = "highmem_is_dirtyable",
1151 .data = &vm_highmem_is_dirtyable,
1152 .maxlen = sizeof(vm_highmem_is_dirtyable),
1153 .mode = 0644,
1154 .proc_handler = &proc_dointvec_minmax,
1155 .strategy = &sysctl_intvec,
1156 .extra1 = &zero,
1157 .extra2 = &one,
1158 },
1159 #endif
1160 /*
1161 * NOTE: do not add new entries to this table unless you have read
1162 * Documentation/sysctl/ctl_unnumbered.txt
1163 */
1164 { .ctl_name = 0 }
1165 };
1166
1167 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1168 static struct ctl_table binfmt_misc_table[] = {
1169 { .ctl_name = 0 }
1170 };
1171 #endif
1172
1173 static struct ctl_table fs_table[] = {
1174 {
1175 .ctl_name = FS_NRINODE,
1176 .procname = "inode-nr",
1177 .data = &inodes_stat,
1178 .maxlen = 2*sizeof(int),
1179 .mode = 0444,
1180 .proc_handler = &proc_dointvec,
1181 },
1182 {
1183 .ctl_name = FS_STATINODE,
1184 .procname = "inode-state",
1185 .data = &inodes_stat,
1186 .maxlen = 7*sizeof(int),
1187 .mode = 0444,
1188 .proc_handler = &proc_dointvec,
1189 },
1190 {
1191 .procname = "file-nr",
1192 .data = &files_stat,
1193 .maxlen = 3*sizeof(int),
1194 .mode = 0444,
1195 .proc_handler = &proc_nr_files,
1196 },
1197 {
1198 .ctl_name = FS_MAXFILE,
1199 .procname = "file-max",
1200 .data = &files_stat.max_files,
1201 .maxlen = sizeof(int),
1202 .mode = 0644,
1203 .proc_handler = &proc_dointvec,
1204 },
1205 {
1206 .ctl_name = FS_DENTRY,
1207 .procname = "dentry-state",
1208 .data = &dentry_stat,
1209 .maxlen = 6*sizeof(int),
1210 .mode = 0444,
1211 .proc_handler = &proc_dointvec,
1212 },
1213 {
1214 .ctl_name = FS_OVERFLOWUID,
1215 .procname = "overflowuid",
1216 .data = &fs_overflowuid,
1217 .maxlen = sizeof(int),
1218 .mode = 0644,
1219 .proc_handler = &proc_dointvec_minmax,
1220 .strategy = &sysctl_intvec,
1221 .extra1 = &minolduid,
1222 .extra2 = &maxolduid,
1223 },
1224 {
1225 .ctl_name = FS_OVERFLOWGID,
1226 .procname = "overflowgid",
1227 .data = &fs_overflowgid,
1228 .maxlen = sizeof(int),
1229 .mode = 0644,
1230 .proc_handler = &proc_dointvec_minmax,
1231 .strategy = &sysctl_intvec,
1232 .extra1 = &minolduid,
1233 .extra2 = &maxolduid,
1234 },
1235 {
1236 .ctl_name = FS_LEASES,
1237 .procname = "leases-enable",
1238 .data = &leases_enable,
1239 .maxlen = sizeof(int),
1240 .mode = 0644,
1241 .proc_handler = &proc_dointvec,
1242 },
1243 #ifdef CONFIG_DNOTIFY
1244 {
1245 .ctl_name = FS_DIR_NOTIFY,
1246 .procname = "dir-notify-enable",
1247 .data = &dir_notify_enable,
1248 .maxlen = sizeof(int),
1249 .mode = 0644,
1250 .proc_handler = &proc_dointvec,
1251 },
1252 #endif
1253 #ifdef CONFIG_MMU
1254 {
1255 .ctl_name = FS_LEASE_TIME,
1256 .procname = "lease-break-time",
1257 .data = &lease_break_time,
1258 .maxlen = sizeof(int),
1259 .mode = 0644,
1260 .proc_handler = &proc_dointvec_minmax,
1261 .strategy = &sysctl_intvec,
1262 .extra1 = &zero,
1263 .extra2 = &two,
1264 },
1265 {
1266 .procname = "aio-nr",
1267 .data = &aio_nr,
1268 .maxlen = sizeof(aio_nr),
1269 .mode = 0444,
1270 .proc_handler = &proc_doulongvec_minmax,
1271 },
1272 {
1273 .procname = "aio-max-nr",
1274 .data = &aio_max_nr,
1275 .maxlen = sizeof(aio_max_nr),
1276 .mode = 0644,
1277 .proc_handler = &proc_doulongvec_minmax,
1278 },
1279 #ifdef CONFIG_INOTIFY_USER
1280 {
1281 .ctl_name = FS_INOTIFY,
1282 .procname = "inotify",
1283 .mode = 0555,
1284 .child = inotify_table,
1285 },
1286 #endif
1287 #endif
1288 {
1289 .ctl_name = KERN_SETUID_DUMPABLE,
1290 .procname = "suid_dumpable",
1291 .data = &suid_dumpable,
1292 .maxlen = sizeof(int),
1293 .mode = 0644,
1294 .proc_handler = &proc_dointvec,
1295 },
1296 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1297 {
1298 .ctl_name = CTL_UNNUMBERED,
1299 .procname = "binfmt_misc",
1300 .mode = 0555,
1301 .child = binfmt_misc_table,
1302 },
1303 #endif
1304 /*
1305 * NOTE: do not add new entries to this table unless you have read
1306 * Documentation/sysctl/ctl_unnumbered.txt
1307 */
1308 { .ctl_name = 0 }
1309 };
1310
1311 static struct ctl_table debug_table[] = {
1312 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1313 {
1314 .ctl_name = CTL_UNNUMBERED,
1315 .procname = "exception-trace",
1316 .data = &show_unhandled_signals,
1317 .maxlen = sizeof(int),
1318 .mode = 0644,
1319 .proc_handler = proc_dointvec
1320 },
1321 #endif
1322 { .ctl_name = 0 }
1323 };
1324
1325 static struct ctl_table dev_table[] = {
1326 { .ctl_name = 0 }
1327 };
1328
1329 static DEFINE_SPINLOCK(sysctl_lock);
1330
1331 /* called under sysctl_lock */
1332 static int use_table(struct ctl_table_header *p)
1333 {
1334 if (unlikely(p->unregistering))
1335 return 0;
1336 p->used++;
1337 return 1;
1338 }
1339
1340 /* called under sysctl_lock */
1341 static void unuse_table(struct ctl_table_header *p)
1342 {
1343 if (!--p->used)
1344 if (unlikely(p->unregistering))
1345 complete(p->unregistering);
1346 }
1347
1348 /* called under sysctl_lock, will reacquire if has to wait */
1349 static void start_unregistering(struct ctl_table_header *p)
1350 {
1351 /*
1352 * if p->used is 0, nobody will ever touch that entry again;
1353 * we'll eliminate all paths to it before dropping sysctl_lock
1354 */
1355 if (unlikely(p->used)) {
1356 struct completion wait;
1357 init_completion(&wait);
1358 p->unregistering = &wait;
1359 spin_unlock(&sysctl_lock);
1360 wait_for_completion(&wait);
1361 spin_lock(&sysctl_lock);
1362 }
1363 /*
1364 * do not remove from the list until nobody holds it; walking the
1365 * list in do_sysctl() relies on that.
1366 */
1367 list_del_init(&p->ctl_entry);
1368 }
1369
1370 void sysctl_head_finish(struct ctl_table_header *head)
1371 {
1372 if (!head)
1373 return;
1374 spin_lock(&sysctl_lock);
1375 unuse_table(head);
1376 spin_unlock(&sysctl_lock);
1377 }
1378
1379 static struct list_head *
1380 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1381 {
1382 struct list_head *header_list;
1383 header_list = &root->header_list;
1384 if (root->lookup)
1385 header_list = root->lookup(root, namespaces);
1386 return header_list;
1387 }
1388
1389 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1390 struct ctl_table_header *prev)
1391 {
1392 struct ctl_table_root *root;
1393 struct list_head *header_list;
1394 struct ctl_table_header *head;
1395 struct list_head *tmp;
1396
1397 spin_lock(&sysctl_lock);
1398 if (prev) {
1399 head = prev;
1400 tmp = &prev->ctl_entry;
1401 unuse_table(prev);
1402 goto next;
1403 }
1404 tmp = &root_table_header.ctl_entry;
1405 for (;;) {
1406 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1407
1408 if (!use_table(head))
1409 goto next;
1410 spin_unlock(&sysctl_lock);
1411 return head;
1412 next:
1413 root = head->root;
1414 tmp = tmp->next;
1415 header_list = lookup_header_list(root, namespaces);
1416 if (tmp != header_list)
1417 continue;
1418
1419 do {
1420 root = list_entry(root->root_list.next,
1421 struct ctl_table_root, root_list);
1422 if (root == &sysctl_table_root)
1423 goto out;
1424 header_list = lookup_header_list(root, namespaces);
1425 } while (list_empty(header_list));
1426 tmp = header_list->next;
1427 }
1428 out:
1429 spin_unlock(&sysctl_lock);
1430 return NULL;
1431 }
1432
1433 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1434 {
1435 return __sysctl_head_next(current->nsproxy, prev);
1436 }
1437
1438 void register_sysctl_root(struct ctl_table_root *root)
1439 {
1440 spin_lock(&sysctl_lock);
1441 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1442 spin_unlock(&sysctl_lock);
1443 }
1444
1445 #ifdef CONFIG_SYSCTL_SYSCALL
1446 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1447 void __user *newval, size_t newlen)
1448 {
1449 struct ctl_table_header *head;
1450 int error = -ENOTDIR;
1451
1452 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1453 return -ENOTDIR;
1454 if (oldval) {
1455 int old_len;
1456 if (!oldlenp || get_user(old_len, oldlenp))
1457 return -EFAULT;
1458 }
1459
1460 for (head = sysctl_head_next(NULL); head;
1461 head = sysctl_head_next(head)) {
1462 error = parse_table(name, nlen, oldval, oldlenp,
1463 newval, newlen, head->ctl_table);
1464 if (error != -ENOTDIR) {
1465 sysctl_head_finish(head);
1466 break;
1467 }
1468 }
1469 return error;
1470 }
1471
1472 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1473 {
1474 struct __sysctl_args tmp;
1475 int error;
1476
1477 if (copy_from_user(&tmp, args, sizeof(tmp)))
1478 return -EFAULT;
1479
1480 error = deprecated_sysctl_warning(&tmp);
1481 if (error)
1482 goto out;
1483
1484 lock_kernel();
1485 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1486 tmp.newval, tmp.newlen);
1487 unlock_kernel();
1488 out:
1489 return error;
1490 }
1491 #endif /* CONFIG_SYSCTL_SYSCALL */
1492
1493 /*
1494 * sysctl_perm does NOT grant the superuser all rights automatically, because
1495 * some sysctl variables are readonly even to root.
1496 */
1497
1498 static int test_perm(int mode, int op)
1499 {
1500 if (!current->euid)
1501 mode >>= 6;
1502 else if (in_egroup_p(0))
1503 mode >>= 3;
1504 if ((mode & op & 0007) == op)
1505 return 0;
1506 return -EACCES;
1507 }
1508
1509 int sysctl_perm(struct ctl_table *table, int op)
1510 {
1511 int error;
1512 error = security_sysctl(table, op);
1513 if (error)
1514 return error;
1515 return test_perm(table->mode, op);
1516 }
1517
1518 #ifdef CONFIG_SYSCTL_SYSCALL
1519 static int parse_table(int __user *name, int nlen,
1520 void __user *oldval, size_t __user *oldlenp,
1521 void __user *newval, size_t newlen,
1522 struct ctl_table *table)
1523 {
1524 int n;
1525 repeat:
1526 if (!nlen)
1527 return -ENOTDIR;
1528 if (get_user(n, name))
1529 return -EFAULT;
1530 for ( ; table->ctl_name || table->procname; table++) {
1531 if (!table->ctl_name)
1532 continue;
1533 if (n == table->ctl_name) {
1534 int error;
1535 if (table->child) {
1536 if (sysctl_perm(table, 001))
1537 return -EPERM;
1538 name++;
1539 nlen--;
1540 table = table->child;
1541 goto repeat;
1542 }
1543 error = do_sysctl_strategy(table, name, nlen,
1544 oldval, oldlenp,
1545 newval, newlen);
1546 return error;
1547 }
1548 }
1549 return -ENOTDIR;
1550 }
1551
1552 /* Perform the actual read/write of a sysctl table entry. */
1553 int do_sysctl_strategy (struct ctl_table *table,
1554 int __user *name, int nlen,
1555 void __user *oldval, size_t __user *oldlenp,
1556 void __user *newval, size_t newlen)
1557 {
1558 int op = 0, rc;
1559
1560 if (oldval)
1561 op |= 004;
1562 if (newval)
1563 op |= 002;
1564 if (sysctl_perm(table, op))
1565 return -EPERM;
1566
1567 if (table->strategy) {
1568 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1569 newval, newlen);
1570 if (rc < 0)
1571 return rc;
1572 if (rc > 0)
1573 return 0;
1574 }
1575
1576 /* If there is no strategy routine, or if the strategy returns
1577 * zero, proceed with automatic r/w */
1578 if (table->data && table->maxlen) {
1579 rc = sysctl_data(table, name, nlen, oldval, oldlenp,
1580 newval, newlen);
1581 if (rc < 0)
1582 return rc;
1583 }
1584 return 0;
1585 }
1586 #endif /* CONFIG_SYSCTL_SYSCALL */
1587
1588 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1589 {
1590 for (; table->ctl_name || table->procname; table++) {
1591 table->parent = parent;
1592 if (table->child)
1593 sysctl_set_parent(table, table->child);
1594 }
1595 }
1596
1597 static __init int sysctl_init(void)
1598 {
1599 int err;
1600 sysctl_set_parent(NULL, root_table);
1601 err = sysctl_check_table(current->nsproxy, root_table);
1602 return 0;
1603 }
1604
1605 core_initcall(sysctl_init);
1606
1607 /**
1608 * __register_sysctl_paths - register a sysctl hierarchy
1609 * @root: List of sysctl headers to register on
1610 * @namespaces: Data to compute which lists of sysctl entries are visible
1611 * @path: The path to the directory the sysctl table is in.
1612 * @table: the top-level table structure
1613 *
1614 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1615 * array. A completely 0 filled entry terminates the table.
1616 *
1617 * The members of the &struct ctl_table structure are used as follows:
1618 *
1619 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1620 * must be unique within that level of sysctl
1621 *
1622 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1623 * enter a sysctl file
1624 *
1625 * data - a pointer to data for use by proc_handler
1626 *
1627 * maxlen - the maximum size in bytes of the data
1628 *
1629 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1630 *
1631 * child - a pointer to the child sysctl table if this entry is a directory, or
1632 * %NULL.
1633 *
1634 * proc_handler - the text handler routine (described below)
1635 *
1636 * strategy - the strategy routine (described below)
1637 *
1638 * de - for internal use by the sysctl routines
1639 *
1640 * extra1, extra2 - extra pointers usable by the proc handler routines
1641 *
1642 * Leaf nodes in the sysctl tree will be represented by a single file
1643 * under /proc; non-leaf nodes will be represented by directories.
1644 *
1645 * sysctl(2) can automatically manage read and write requests through
1646 * the sysctl table. The data and maxlen fields of the ctl_table
1647 * struct enable minimal validation of the values being written to be
1648 * performed, and the mode field allows minimal authentication.
1649 *
1650 * More sophisticated management can be enabled by the provision of a
1651 * strategy routine with the table entry. This will be called before
1652 * any automatic read or write of the data is performed.
1653 *
1654 * The strategy routine may return
1655 *
1656 * < 0 - Error occurred (error is passed to user process)
1657 *
1658 * 0 - OK - proceed with automatic read or write.
1659 *
1660 * > 0 - OK - read or write has been done by the strategy routine, so
1661 * return immediately.
1662 *
1663 * There must be a proc_handler routine for any terminal nodes
1664 * mirrored under /proc/sys (non-terminals are handled by a built-in
1665 * directory handler). Several default handlers are available to
1666 * cover common cases -
1667 *
1668 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1669 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1670 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1671 *
1672 * It is the handler's job to read the input buffer from user memory
1673 * and process it. The handler should return 0 on success.
1674 *
1675 * This routine returns %NULL on a failure to register, and a pointer
1676 * to the table header on success.
1677 */
1678 struct ctl_table_header *__register_sysctl_paths(
1679 struct ctl_table_root *root,
1680 struct nsproxy *namespaces,
1681 const struct ctl_path *path, struct ctl_table *table)
1682 {
1683 struct list_head *header_list;
1684 struct ctl_table_header *header;
1685 struct ctl_table *new, **prevp;
1686 unsigned int n, npath;
1687
1688 /* Count the path components */
1689 for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1690 ;
1691
1692 /*
1693 * For each path component, allocate a 2-element ctl_table array.
1694 * The first array element will be filled with the sysctl entry
1695 * for this, the second will be the sentinel (ctl_name == 0).
1696 *
1697 * We allocate everything in one go so that we don't have to
1698 * worry about freeing additional memory in unregister_sysctl_table.
1699 */
1700 header = kzalloc(sizeof(struct ctl_table_header) +
1701 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1702 if (!header)
1703 return NULL;
1704
1705 new = (struct ctl_table *) (header + 1);
1706
1707 /* Now connect the dots */
1708 prevp = &header->ctl_table;
1709 for (n = 0; n < npath; ++n, ++path) {
1710 /* Copy the procname */
1711 new->procname = path->procname;
1712 new->ctl_name = path->ctl_name;
1713 new->mode = 0555;
1714
1715 *prevp = new;
1716 prevp = &new->child;
1717
1718 new += 2;
1719 }
1720 *prevp = table;
1721 header->ctl_table_arg = table;
1722
1723 INIT_LIST_HEAD(&header->ctl_entry);
1724 header->used = 0;
1725 header->unregistering = NULL;
1726 header->root = root;
1727 sysctl_set_parent(NULL, header->ctl_table);
1728 if (sysctl_check_table(namespaces, header->ctl_table)) {
1729 kfree(header);
1730 return NULL;
1731 }
1732 spin_lock(&sysctl_lock);
1733 header_list = lookup_header_list(root, namespaces);
1734 list_add_tail(&header->ctl_entry, header_list);
1735 spin_unlock(&sysctl_lock);
1736
1737 return header;
1738 }
1739
1740 /**
1741 * register_sysctl_table_path - register a sysctl table hierarchy
1742 * @path: The path to the directory the sysctl table is in.
1743 * @table: the top-level table structure
1744 *
1745 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1746 * array. A completely 0 filled entry terminates the table.
1747 *
1748 * See __register_sysctl_paths for more details.
1749 */
1750 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1751 struct ctl_table *table)
1752 {
1753 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1754 path, table);
1755 }
1756
1757 /**
1758 * register_sysctl_table - register a sysctl table hierarchy
1759 * @table: the top-level table structure
1760 *
1761 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1762 * array. A completely 0 filled entry terminates the table.
1763 *
1764 * See register_sysctl_paths for more details.
1765 */
1766 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1767 {
1768 static const struct ctl_path null_path[] = { {} };
1769
1770 return register_sysctl_paths(null_path, table);
1771 }
1772
1773 /**
1774 * unregister_sysctl_table - unregister a sysctl table hierarchy
1775 * @header: the header returned from register_sysctl_table
1776 *
1777 * Unregisters the sysctl table and all children. proc entries may not
1778 * actually be removed until they are no longer used by anyone.
1779 */
1780 void unregister_sysctl_table(struct ctl_table_header * header)
1781 {
1782 might_sleep();
1783
1784 if (header == NULL)
1785 return;
1786
1787 spin_lock(&sysctl_lock);
1788 start_unregistering(header);
1789 spin_unlock(&sysctl_lock);
1790 kfree(header);
1791 }
1792
1793 #else /* !CONFIG_SYSCTL */
1794 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1795 {
1796 return NULL;
1797 }
1798
1799 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1800 struct ctl_table *table)
1801 {
1802 return NULL;
1803 }
1804
1805 void unregister_sysctl_table(struct ctl_table_header * table)
1806 {
1807 }
1808
1809 #endif /* CONFIG_SYSCTL */
1810
1811 /*
1812 * /proc/sys support
1813 */
1814
1815 #ifdef CONFIG_PROC_SYSCTL
1816
1817 static int _proc_do_string(void* data, int maxlen, int write,
1818 struct file *filp, void __user *buffer,
1819 size_t *lenp, loff_t *ppos)
1820 {
1821 size_t len;
1822 char __user *p;
1823 char c;
1824
1825 if (!data || !maxlen || !*lenp) {
1826 *lenp = 0;
1827 return 0;
1828 }
1829
1830 if (write) {
1831 len = 0;
1832 p = buffer;
1833 while (len < *lenp) {
1834 if (get_user(c, p++))
1835 return -EFAULT;
1836 if (c == 0 || c == '\n')
1837 break;
1838 len++;
1839 }
1840 if (len >= maxlen)
1841 len = maxlen-1;
1842 if(copy_from_user(data, buffer, len))
1843 return -EFAULT;
1844 ((char *) data)[len] = 0;
1845 *ppos += *lenp;
1846 } else {
1847 len = strlen(data);
1848 if (len > maxlen)
1849 len = maxlen;
1850
1851 if (*ppos > len) {
1852 *lenp = 0;
1853 return 0;
1854 }
1855
1856 data += *ppos;
1857 len -= *ppos;
1858
1859 if (len > *lenp)
1860 len = *lenp;
1861 if (len)
1862 if(copy_to_user(buffer, data, len))
1863 return -EFAULT;
1864 if (len < *lenp) {
1865 if(put_user('\n', ((char __user *) buffer) + len))
1866 return -EFAULT;
1867 len++;
1868 }
1869 *lenp = len;
1870 *ppos += len;
1871 }
1872 return 0;
1873 }
1874
1875 /**
1876 * proc_dostring - read a string sysctl
1877 * @table: the sysctl table
1878 * @write: %TRUE if this is a write to the sysctl file
1879 * @filp: the file structure
1880 * @buffer: the user buffer
1881 * @lenp: the size of the user buffer
1882 * @ppos: file position
1883 *
1884 * Reads/writes a string from/to the user buffer. If the kernel
1885 * buffer provided is not large enough to hold the string, the
1886 * string is truncated. The copied string is %NULL-terminated.
1887 * If the string is being read by the user process, it is copied
1888 * and a newline '\n' is added. It is truncated if the buffer is
1889 * not large enough.
1890 *
1891 * Returns 0 on success.
1892 */
1893 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
1894 void __user *buffer, size_t *lenp, loff_t *ppos)
1895 {
1896 return _proc_do_string(table->data, table->maxlen, write, filp,
1897 buffer, lenp, ppos);
1898 }
1899
1900
1901 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1902 int *valp,
1903 int write, void *data)
1904 {
1905 if (write) {
1906 *valp = *negp ? -*lvalp : *lvalp;
1907 } else {
1908 int val = *valp;
1909 if (val < 0) {
1910 *negp = -1;
1911 *lvalp = (unsigned long)-val;
1912 } else {
1913 *negp = 0;
1914 *lvalp = (unsigned long)val;
1915 }
1916 }
1917 return 0;
1918 }
1919
1920 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1921 int write, struct file *filp, void __user *buffer,
1922 size_t *lenp, loff_t *ppos,
1923 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1924 int write, void *data),
1925 void *data)
1926 {
1927 #define TMPBUFLEN 21
1928 int *i, vleft, first=1, neg, val;
1929 unsigned long lval;
1930 size_t left, len;
1931
1932 char buf[TMPBUFLEN], *p;
1933 char __user *s = buffer;
1934
1935 if (!tbl_data || !table->maxlen || !*lenp ||
1936 (*ppos && !write)) {
1937 *lenp = 0;
1938 return 0;
1939 }
1940
1941 i = (int *) tbl_data;
1942 vleft = table->maxlen / sizeof(*i);
1943 left = *lenp;
1944
1945 if (!conv)
1946 conv = do_proc_dointvec_conv;
1947
1948 for (; left && vleft--; i++, first=0) {
1949 if (write) {
1950 while (left) {
1951 char c;
1952 if (get_user(c, s))
1953 return -EFAULT;
1954 if (!isspace(c))
1955 break;
1956 left--;
1957 s++;
1958 }
1959 if (!left)
1960 break;
1961 neg = 0;
1962 len = left;
1963 if (len > sizeof(buf) - 1)
1964 len = sizeof(buf) - 1;
1965 if (copy_from_user(buf, s, len))
1966 return -EFAULT;
1967 buf[len] = 0;
1968 p = buf;
1969 if (*p == '-' && left > 1) {
1970 neg = 1;
1971 p++;
1972 }
1973 if (*p < '0' || *p > '9')
1974 break;
1975
1976 lval = simple_strtoul(p, &p, 0);
1977
1978 len = p-buf;
1979 if ((len < left) && *p && !isspace(*p))
1980 break;
1981 if (neg)
1982 val = -val;
1983 s += len;
1984 left -= len;
1985
1986 if (conv(&neg, &lval, i, 1, data))
1987 break;
1988 } else {
1989 p = buf;
1990 if (!first)
1991 *p++ = '\t';
1992
1993 if (conv(&neg, &lval, i, 0, data))
1994 break;
1995
1996 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1997 len = strlen(buf);
1998 if (len > left)
1999 len = left;
2000 if(copy_to_user(s, buf, len))
2001 return -EFAULT;
2002 left -= len;
2003 s += len;
2004 }
2005 }
2006
2007 if (!write && !first && left) {
2008 if(put_user('\n', s))
2009 return -EFAULT;
2010 left--, s++;
2011 }
2012 if (write) {
2013 while (left) {
2014 char c;
2015 if (get_user(c, s++))
2016 return -EFAULT;
2017 if (!isspace(c))
2018 break;
2019 left--;
2020 }
2021 }
2022 if (write && first)
2023 return -EINVAL;
2024 *lenp -= left;
2025 *ppos += *lenp;
2026 return 0;
2027 #undef TMPBUFLEN
2028 }
2029
2030 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2031 void __user *buffer, size_t *lenp, loff_t *ppos,
2032 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2033 int write, void *data),
2034 void *data)
2035 {
2036 return __do_proc_dointvec(table->data, table, write, filp,
2037 buffer, lenp, ppos, conv, data);
2038 }
2039
2040 /**
2041 * proc_dointvec - read a vector of integers
2042 * @table: the sysctl table
2043 * @write: %TRUE if this is a write to the sysctl file
2044 * @filp: the file structure
2045 * @buffer: the user buffer
2046 * @lenp: the size of the user buffer
2047 * @ppos: file position
2048 *
2049 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2050 * values from/to the user buffer, treated as an ASCII string.
2051 *
2052 * Returns 0 on success.
2053 */
2054 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2055 void __user *buffer, size_t *lenp, loff_t *ppos)
2056 {
2057 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2058 NULL,NULL);
2059 }
2060
2061 #define OP_SET 0
2062 #define OP_AND 1
2063 #define OP_OR 2
2064
2065 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
2066 int *valp,
2067 int write, void *data)
2068 {
2069 int op = *(int *)data;
2070 if (write) {
2071 int val = *negp ? -*lvalp : *lvalp;
2072 switch(op) {
2073 case OP_SET: *valp = val; break;
2074 case OP_AND: *valp &= val; break;
2075 case OP_OR: *valp |= val; break;
2076 }
2077 } else {
2078 int val = *valp;
2079 if (val < 0) {
2080 *negp = -1;
2081 *lvalp = (unsigned long)-val;
2082 } else {
2083 *negp = 0;
2084 *lvalp = (unsigned long)val;
2085 }
2086 }
2087 return 0;
2088 }
2089
2090 /*
2091 * Taint values can only be increased
2092 */
2093 static int proc_dointvec_taint(struct ctl_table *table, int write, struct file *filp,
2094 void __user *buffer, size_t *lenp, loff_t *ppos)
2095 {
2096 int op;
2097
2098 if (write && !capable(CAP_SYS_ADMIN))
2099 return -EPERM;
2100
2101 op = OP_OR;
2102 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2103 do_proc_dointvec_bset_conv,&op);
2104 }
2105
2106 struct do_proc_dointvec_minmax_conv_param {
2107 int *min;
2108 int *max;
2109 };
2110
2111 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2112 int *valp,
2113 int write, void *data)
2114 {
2115 struct do_proc_dointvec_minmax_conv_param *param = data;
2116 if (write) {
2117 int val = *negp ? -*lvalp : *lvalp;
2118 if ((param->min && *param->min > val) ||
2119 (param->max && *param->max < val))
2120 return -EINVAL;
2121 *valp = val;
2122 } else {
2123 int val = *valp;
2124 if (val < 0) {
2125 *negp = -1;
2126 *lvalp = (unsigned long)-val;
2127 } else {
2128 *negp = 0;
2129 *lvalp = (unsigned long)val;
2130 }
2131 }
2132 return 0;
2133 }
2134
2135 /**
2136 * proc_dointvec_minmax - read a vector of integers with min/max values
2137 * @table: the sysctl table
2138 * @write: %TRUE if this is a write to the sysctl file
2139 * @filp: the file structure
2140 * @buffer: the user buffer
2141 * @lenp: the size of the user buffer
2142 * @ppos: file position
2143 *
2144 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2145 * values from/to the user buffer, treated as an ASCII string.
2146 *
2147 * This routine will ensure the values are within the range specified by
2148 * table->extra1 (min) and table->extra2 (max).
2149 *
2150 * Returns 0 on success.
2151 */
2152 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2153 void __user *buffer, size_t *lenp, loff_t *ppos)
2154 {
2155 struct do_proc_dointvec_minmax_conv_param param = {
2156 .min = (int *) table->extra1,
2157 .max = (int *) table->extra2,
2158 };
2159 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2160 do_proc_dointvec_minmax_conv, &param);
2161 }
2162
2163 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2164 struct file *filp,
2165 void __user *buffer,
2166 size_t *lenp, loff_t *ppos,
2167 unsigned long convmul,
2168 unsigned long convdiv)
2169 {
2170 #define TMPBUFLEN 21
2171 unsigned long *i, *min, *max, val;
2172 int vleft, first=1, neg;
2173 size_t len, left;
2174 char buf[TMPBUFLEN], *p;
2175 char __user *s = buffer;
2176
2177 if (!data || !table->maxlen || !*lenp ||
2178 (*ppos && !write)) {
2179 *lenp = 0;
2180 return 0;
2181 }
2182
2183 i = (unsigned long *) data;
2184 min = (unsigned long *) table->extra1;
2185 max = (unsigned long *) table->extra2;
2186 vleft = table->maxlen / sizeof(unsigned long);
2187 left = *lenp;
2188
2189 for (; left && vleft--; i++, min++, max++, first=0) {
2190 if (write) {
2191 while (left) {
2192 char c;
2193 if (get_user(c, s))
2194 return -EFAULT;
2195 if (!isspace(c))
2196 break;
2197 left--;
2198 s++;
2199 }
2200 if (!left)
2201 break;
2202 neg = 0;
2203 len = left;
2204 if (len > TMPBUFLEN-1)
2205 len = TMPBUFLEN-1;
2206 if (copy_from_user(buf, s, len))
2207 return -EFAULT;
2208 buf[len] = 0;
2209 p = buf;
2210 if (*p == '-' && left > 1) {
2211 neg = 1;
2212 p++;
2213 }
2214 if (*p < '0' || *p > '9')
2215 break;
2216 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2217 len = p-buf;
2218 if ((len < left) && *p && !isspace(*p))
2219 break;
2220 if (neg)
2221 val = -val;
2222 s += len;
2223 left -= len;
2224
2225 if(neg)
2226 continue;
2227 if ((min && val < *min) || (max && val > *max))
2228 continue;
2229 *i = val;
2230 } else {
2231 p = buf;
2232 if (!first)
2233 *p++ = '\t';
2234 sprintf(p, "%lu", convdiv * (*i) / convmul);
2235 len = strlen(buf);
2236 if (len > left)
2237 len = left;
2238 if(copy_to_user(s, buf, len))
2239 return -EFAULT;
2240 left -= len;
2241 s += len;
2242 }
2243 }
2244
2245 if (!write && !first && left) {
2246 if(put_user('\n', s))
2247 return -EFAULT;
2248 left--, s++;
2249 }
2250 if (write) {
2251 while (left) {
2252 char c;
2253 if (get_user(c, s++))
2254 return -EFAULT;
2255 if (!isspace(c))
2256 break;
2257 left--;
2258 }
2259 }
2260 if (write && first)
2261 return -EINVAL;
2262 *lenp -= left;
2263 *ppos += *lenp;
2264 return 0;
2265 #undef TMPBUFLEN
2266 }
2267
2268 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2269 struct file *filp,
2270 void __user *buffer,
2271 size_t *lenp, loff_t *ppos,
2272 unsigned long convmul,
2273 unsigned long convdiv)
2274 {
2275 return __do_proc_doulongvec_minmax(table->data, table, write,
2276 filp, buffer, lenp, ppos, convmul, convdiv);
2277 }
2278
2279 /**
2280 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2281 * @table: the sysctl table
2282 * @write: %TRUE if this is a write to the sysctl file
2283 * @filp: the file structure
2284 * @buffer: the user buffer
2285 * @lenp: the size of the user buffer
2286 * @ppos: file position
2287 *
2288 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2289 * values from/to the user buffer, treated as an ASCII string.
2290 *
2291 * This routine will ensure the values are within the range specified by
2292 * table->extra1 (min) and table->extra2 (max).
2293 *
2294 * Returns 0 on success.
2295 */
2296 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2297 void __user *buffer, size_t *lenp, loff_t *ppos)
2298 {
2299 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2300 }
2301
2302 /**
2303 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2304 * @table: the sysctl table
2305 * @write: %TRUE if this is a write to the sysctl file
2306 * @filp: the file structure
2307 * @buffer: the user buffer
2308 * @lenp: the size of the user buffer
2309 * @ppos: file position
2310 *
2311 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2312 * values from/to the user buffer, treated as an ASCII string. The values
2313 * are treated as milliseconds, and converted to jiffies when they are stored.
2314 *
2315 * This routine will ensure the values are within the range specified by
2316 * table->extra1 (min) and table->extra2 (max).
2317 *
2318 * Returns 0 on success.
2319 */
2320 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2321 struct file *filp,
2322 void __user *buffer,
2323 size_t *lenp, loff_t *ppos)
2324 {
2325 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2326 lenp, ppos, HZ, 1000l);
2327 }
2328
2329
2330 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2331 int *valp,
2332 int write, void *data)
2333 {
2334 if (write) {
2335 if (*lvalp > LONG_MAX / HZ)
2336 return 1;
2337 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2338 } else {
2339 int val = *valp;
2340 unsigned long lval;
2341 if (val < 0) {
2342 *negp = -1;
2343 lval = (unsigned long)-val;
2344 } else {
2345 *negp = 0;
2346 lval = (unsigned long)val;
2347 }
2348 *lvalp = lval / HZ;
2349 }
2350 return 0;
2351 }
2352
2353 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2354 int *valp,
2355 int write, void *data)
2356 {
2357 if (write) {
2358 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2359 return 1;
2360 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2361 } else {
2362 int val = *valp;
2363 unsigned long lval;
2364 if (val < 0) {
2365 *negp = -1;
2366 lval = (unsigned long)-val;
2367 } else {
2368 *negp = 0;
2369 lval = (unsigned long)val;
2370 }
2371 *lvalp = jiffies_to_clock_t(lval);
2372 }
2373 return 0;
2374 }
2375
2376 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2377 int *valp,
2378 int write, void *data)
2379 {
2380 if (write) {
2381 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2382 } else {
2383 int val = *valp;
2384 unsigned long lval;
2385 if (val < 0) {
2386 *negp = -1;
2387 lval = (unsigned long)-val;
2388 } else {
2389 *negp = 0;
2390 lval = (unsigned long)val;
2391 }
2392 *lvalp = jiffies_to_msecs(lval);
2393 }
2394 return 0;
2395 }
2396
2397 /**
2398 * proc_dointvec_jiffies - read a vector of integers as seconds
2399 * @table: the sysctl table
2400 * @write: %TRUE if this is a write to the sysctl file
2401 * @filp: the file structure
2402 * @buffer: the user buffer
2403 * @lenp: the size of the user buffer
2404 * @ppos: file position
2405 *
2406 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2407 * values from/to the user buffer, treated as an ASCII string.
2408 * The values read are assumed to be in seconds, and are converted into
2409 * jiffies.
2410 *
2411 * Returns 0 on success.
2412 */
2413 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2414 void __user *buffer, size_t *lenp, loff_t *ppos)
2415 {
2416 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2417 do_proc_dointvec_jiffies_conv,NULL);
2418 }
2419
2420 /**
2421 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2422 * @table: the sysctl table
2423 * @write: %TRUE if this is a write to the sysctl file
2424 * @filp: the file structure
2425 * @buffer: the user buffer
2426 * @lenp: the size of the user buffer
2427 * @ppos: pointer to the file position
2428 *
2429 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2430 * values from/to the user buffer, treated as an ASCII string.
2431 * The values read are assumed to be in 1/USER_HZ seconds, and
2432 * are converted into jiffies.
2433 *
2434 * Returns 0 on success.
2435 */
2436 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2437 void __user *buffer, size_t *lenp, loff_t *ppos)
2438 {
2439 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2440 do_proc_dointvec_userhz_jiffies_conv,NULL);
2441 }
2442
2443 /**
2444 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2445 * @table: the sysctl table
2446 * @write: %TRUE if this is a write to the sysctl file
2447 * @filp: the file structure
2448 * @buffer: the user buffer
2449 * @lenp: the size of the user buffer
2450 * @ppos: file position
2451 * @ppos: the current position in the file
2452 *
2453 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2454 * values from/to the user buffer, treated as an ASCII string.
2455 * The values read are assumed to be in 1/1000 seconds, and
2456 * are converted into jiffies.
2457 *
2458 * Returns 0 on success.
2459 */
2460 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2461 void __user *buffer, size_t *lenp, loff_t *ppos)
2462 {
2463 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2464 do_proc_dointvec_ms_jiffies_conv, NULL);
2465 }
2466
2467 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2468 void __user *buffer, size_t *lenp, loff_t *ppos)
2469 {
2470 struct pid *new_pid;
2471 pid_t tmp;
2472 int r;
2473
2474 tmp = pid_nr_ns(cad_pid, current->nsproxy->pid_ns);
2475
2476 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2477 lenp, ppos, NULL, NULL);
2478 if (r || !write)
2479 return r;
2480
2481 new_pid = find_get_pid(tmp);
2482 if (!new_pid)
2483 return -ESRCH;
2484
2485 put_pid(xchg(&cad_pid, new_pid));
2486 return 0;
2487 }
2488
2489 #else /* CONFIG_PROC_FS */
2490
2491 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2492 void __user *buffer, size_t *lenp, loff_t *ppos)
2493 {
2494 return -ENOSYS;
2495 }
2496
2497 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2498 void __user *buffer, size_t *lenp, loff_t *ppos)
2499 {
2500 return -ENOSYS;
2501 }
2502
2503 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2504 void __user *buffer, size_t *lenp, loff_t *ppos)
2505 {
2506 return -ENOSYS;
2507 }
2508
2509 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2510 void __user *buffer, size_t *lenp, loff_t *ppos)
2511 {
2512 return -ENOSYS;
2513 }
2514
2515 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2516 void __user *buffer, size_t *lenp, loff_t *ppos)
2517 {
2518 return -ENOSYS;
2519 }
2520
2521 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2522 void __user *buffer, size_t *lenp, loff_t *ppos)
2523 {
2524 return -ENOSYS;
2525 }
2526
2527 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2528 void __user *buffer, size_t *lenp, loff_t *ppos)
2529 {
2530 return -ENOSYS;
2531 }
2532
2533 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2534 struct file *filp,
2535 void __user *buffer,
2536 size_t *lenp, loff_t *ppos)
2537 {
2538 return -ENOSYS;
2539 }
2540
2541
2542 #endif /* CONFIG_PROC_FS */
2543
2544
2545 #ifdef CONFIG_SYSCTL_SYSCALL
2546 /*
2547 * General sysctl support routines
2548 */
2549
2550 /* The generic sysctl data routine (used if no strategy routine supplied) */
2551 int sysctl_data(struct ctl_table *table, int __user *name, int nlen,
2552 void __user *oldval, size_t __user *oldlenp,
2553 void __user *newval, size_t newlen)
2554 {
2555 size_t len;
2556
2557 /* Get out of I don't have a variable */
2558 if (!table->data || !table->maxlen)
2559 return -ENOTDIR;
2560
2561 if (oldval && oldlenp) {
2562 if (get_user(len, oldlenp))
2563 return -EFAULT;
2564 if (len) {
2565 if (len > table->maxlen)
2566 len = table->maxlen;
2567 if (copy_to_user(oldval, table->data, len))
2568 return -EFAULT;
2569 if (put_user(len, oldlenp))
2570 return -EFAULT;
2571 }
2572 }
2573
2574 if (newval && newlen) {
2575 if (newlen > table->maxlen)
2576 newlen = table->maxlen;
2577
2578 if (copy_from_user(table->data, newval, newlen))
2579 return -EFAULT;
2580 }
2581 return 1;
2582 }
2583
2584 /* The generic string strategy routine: */
2585 int sysctl_string(struct ctl_table *table, int __user *name, int nlen,
2586 void __user *oldval, size_t __user *oldlenp,
2587 void __user *newval, size_t newlen)
2588 {
2589 if (!table->data || !table->maxlen)
2590 return -ENOTDIR;
2591
2592 if (oldval && oldlenp) {
2593 size_t bufsize;
2594 if (get_user(bufsize, oldlenp))
2595 return -EFAULT;
2596 if (bufsize) {
2597 size_t len = strlen(table->data), copied;
2598
2599 /* This shouldn't trigger for a well-formed sysctl */
2600 if (len > table->maxlen)
2601 len = table->maxlen;
2602
2603 /* Copy up to a max of bufsize-1 bytes of the string */
2604 copied = (len >= bufsize) ? bufsize - 1 : len;
2605
2606 if (copy_to_user(oldval, table->data, copied) ||
2607 put_user(0, (char __user *)(oldval + copied)))
2608 return -EFAULT;
2609 if (put_user(len, oldlenp))
2610 return -EFAULT;
2611 }
2612 }
2613 if (newval && newlen) {
2614 size_t len = newlen;
2615 if (len > table->maxlen)
2616 len = table->maxlen;
2617 if(copy_from_user(table->data, newval, len))
2618 return -EFAULT;
2619 if (len == table->maxlen)
2620 len--;
2621 ((char *) table->data)[len] = 0;
2622 }
2623 return 1;
2624 }
2625
2626 /*
2627 * This function makes sure that all of the integers in the vector
2628 * are between the minimum and maximum values given in the arrays
2629 * table->extra1 and table->extra2, respectively.
2630 */
2631 int sysctl_intvec(struct ctl_table *table, int __user *name, int nlen,
2632 void __user *oldval, size_t __user *oldlenp,
2633 void __user *newval, size_t newlen)
2634 {
2635
2636 if (newval && newlen) {
2637 int __user *vec = (int __user *) newval;
2638 int *min = (int *) table->extra1;
2639 int *max = (int *) table->extra2;
2640 size_t length;
2641 int i;
2642
2643 if (newlen % sizeof(int) != 0)
2644 return -EINVAL;
2645
2646 if (!table->extra1 && !table->extra2)
2647 return 0;
2648
2649 if (newlen > table->maxlen)
2650 newlen = table->maxlen;
2651 length = newlen / sizeof(int);
2652
2653 for (i = 0; i < length; i++) {
2654 int value;
2655 if (get_user(value, vec + i))
2656 return -EFAULT;
2657 if (min && value < min[i])
2658 return -EINVAL;
2659 if (max && value > max[i])
2660 return -EINVAL;
2661 }
2662 }
2663 return 0;
2664 }
2665
2666 /* Strategy function to convert jiffies to seconds */
2667 int sysctl_jiffies(struct ctl_table *table, int __user *name, int nlen,
2668 void __user *oldval, size_t __user *oldlenp,
2669 void __user *newval, size_t newlen)
2670 {
2671 if (oldval && oldlenp) {
2672 size_t olen;
2673
2674 if (get_user(olen, oldlenp))
2675 return -EFAULT;
2676 if (olen) {
2677 int val;
2678
2679 if (olen < sizeof(int))
2680 return -EINVAL;
2681
2682 val = *(int *)(table->data) / HZ;
2683 if (put_user(val, (int __user *)oldval))
2684 return -EFAULT;
2685 if (put_user(sizeof(int), oldlenp))
2686 return -EFAULT;
2687 }
2688 }
2689 if (newval && newlen) {
2690 int new;
2691 if (newlen != sizeof(int))
2692 return -EINVAL;
2693 if (get_user(new, (int __user *)newval))
2694 return -EFAULT;
2695 *(int *)(table->data) = new*HZ;
2696 }
2697 return 1;
2698 }
2699
2700 /* Strategy function to convert jiffies to seconds */
2701 int sysctl_ms_jiffies(struct ctl_table *table, int __user *name, int nlen,
2702 void __user *oldval, size_t __user *oldlenp,
2703 void __user *newval, size_t newlen)
2704 {
2705 if (oldval && oldlenp) {
2706 size_t olen;
2707
2708 if (get_user(olen, oldlenp))
2709 return -EFAULT;
2710 if (olen) {
2711 int val;
2712
2713 if (olen < sizeof(int))
2714 return -EINVAL;
2715
2716 val = jiffies_to_msecs(*(int *)(table->data));
2717 if (put_user(val, (int __user *)oldval))
2718 return -EFAULT;
2719 if (put_user(sizeof(int), oldlenp))
2720 return -EFAULT;
2721 }
2722 }
2723 if (newval && newlen) {
2724 int new;
2725 if (newlen != sizeof(int))
2726 return -EINVAL;
2727 if (get_user(new, (int __user *)newval))
2728 return -EFAULT;
2729 *(int *)(table->data) = msecs_to_jiffies(new);
2730 }
2731 return 1;
2732 }
2733
2734
2735
2736 #else /* CONFIG_SYSCTL_SYSCALL */
2737
2738
2739 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2740 {
2741 struct __sysctl_args tmp;
2742 int error;
2743
2744 if (copy_from_user(&tmp, args, sizeof(tmp)))
2745 return -EFAULT;
2746
2747 error = deprecated_sysctl_warning(&tmp);
2748
2749 /* If no error reading the parameters then just -ENOSYS ... */
2750 if (!error)
2751 error = -ENOSYS;
2752
2753 return error;
2754 }
2755
2756 int sysctl_data(struct ctl_table *table, int __user *name, int nlen,
2757 void __user *oldval, size_t __user *oldlenp,
2758 void __user *newval, size_t newlen)
2759 {
2760 return -ENOSYS;
2761 }
2762
2763 int sysctl_string(struct ctl_table *table, int __user *name, int nlen,
2764 void __user *oldval, size_t __user *oldlenp,
2765 void __user *newval, size_t newlen)
2766 {
2767 return -ENOSYS;
2768 }
2769
2770 int sysctl_intvec(struct ctl_table *table, int __user *name, int nlen,
2771 void __user *oldval, size_t __user *oldlenp,
2772 void __user *newval, size_t newlen)
2773 {
2774 return -ENOSYS;
2775 }
2776
2777 int sysctl_jiffies(struct ctl_table *table, int __user *name, int nlen,
2778 void __user *oldval, size_t __user *oldlenp,
2779 void __user *newval, size_t newlen)
2780 {
2781 return -ENOSYS;
2782 }
2783
2784 int sysctl_ms_jiffies(struct ctl_table *table, int __user *name, int nlen,
2785 void __user *oldval, size_t __user *oldlenp,
2786 void __user *newval, size_t newlen)
2787 {
2788 return -ENOSYS;
2789 }
2790
2791 #endif /* CONFIG_SYSCTL_SYSCALL */
2792
2793 static int deprecated_sysctl_warning(struct __sysctl_args *args)
2794 {
2795 static int msg_count;
2796 int name[CTL_MAXNAME];
2797 int i;
2798
2799 /* Check args->nlen. */
2800 if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
2801 return -ENOTDIR;
2802
2803 /* Read in the sysctl name for better debug message logging */
2804 for (i = 0; i < args->nlen; i++)
2805 if (get_user(name[i], args->name + i))
2806 return -EFAULT;
2807
2808 /* Ignore accesses to kernel.version */
2809 if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2810 return 0;
2811
2812 if (msg_count < 5) {
2813 msg_count++;
2814 printk(KERN_INFO
2815 "warning: process `%s' used the deprecated sysctl "
2816 "system call with ", current->comm);
2817 for (i = 0; i < args->nlen; i++)
2818 printk("%d.", name[i]);
2819 printk("\n");
2820 }
2821 return 0;
2822 }
2823
2824 /*
2825 * No sense putting this after each symbol definition, twice,
2826 * exception granted :-)
2827 */
2828 EXPORT_SYMBOL(proc_dointvec);
2829 EXPORT_SYMBOL(proc_dointvec_jiffies);
2830 EXPORT_SYMBOL(proc_dointvec_minmax);
2831 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2832 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2833 EXPORT_SYMBOL(proc_dostring);
2834 EXPORT_SYMBOL(proc_doulongvec_minmax);
2835 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2836 EXPORT_SYMBOL(register_sysctl_table);
2837 EXPORT_SYMBOL(register_sysctl_paths);
2838 EXPORT_SYMBOL(sysctl_intvec);
2839 EXPORT_SYMBOL(sysctl_jiffies);
2840 EXPORT_SYMBOL(sysctl_ms_jiffies);
2841 EXPORT_SYMBOL(sysctl_string);
2842 EXPORT_SYMBOL(sysctl_data);
2843 EXPORT_SYMBOL(unregister_sysctl_table);
Log-structured file system for Linux