Char: sx, fix io unmapping
[linux-2.6/mini2440.git] / kernel / sysctl.c
blob1bf369bd44234e829905d9a565577537c040c846
1 /*
2 * sysctl.c: General linux system control interface
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
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>
52 #include <asm/uaccess.h>
53 #include <asm/processor.h>
55 #ifdef CONFIG_X86
56 #include <asm/nmi.h>
57 #include <asm/stacktrace.h>
58 #include <asm/io.h>
59 #endif
61 static int deprecated_sysctl_warning(struct __sysctl_args *args);
63 #if defined(CONFIG_SYSCTL)
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 pid_max_min, pid_max_max;
80 extern int sysctl_drop_caches;
81 extern int percpu_pagelist_fraction;
82 extern int compat_log;
83 extern int maps_protect;
84 extern int latencytop_enabled;
85 extern int sysctl_nr_open_min, sysctl_nr_open_max;
86 #ifdef CONFIG_RCU_TORTURE_TEST
87 extern int rcutorture_runnable;
88 #endif /* #ifdef CONFIG_RCU_TORTURE_TEST */
90 /* Constants used for minimum and maximum */
91 #if defined(CONFIG_HIGHMEM) || defined(CONFIG_DETECT_SOFTLOCKUP)
92 static int one = 1;
93 #endif
95 #ifdef CONFIG_DETECT_SOFTLOCKUP
96 static int sixty = 60;
97 static int neg_one = -1;
98 #endif
100 #ifdef CONFIG_MMU
101 static int two = 2;
102 #endif
104 static int zero;
105 static int one_hundred = 100;
107 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
108 static int maxolduid = 65535;
109 static int minolduid;
110 static int min_percpu_pagelist_fract = 8;
112 static int ngroups_max = NGROUPS_MAX;
114 #ifdef CONFIG_MODULES
115 extern char modprobe_path[];
116 #endif
117 #ifdef CONFIG_CHR_DEV_SG
118 extern int sg_big_buff;
119 #endif
121 #ifdef CONFIG_SPARC
122 #include <asm/system.h>
123 #endif
125 #ifdef __hppa__
126 extern int pwrsw_enabled;
127 extern int unaligned_enabled;
128 #endif
130 #ifdef CONFIG_S390
131 #ifdef CONFIG_MATHEMU
132 extern int sysctl_ieee_emulation_warnings;
133 #endif
134 extern int sysctl_userprocess_debug;
135 extern int spin_retry;
136 #endif
138 #ifdef CONFIG_BSD_PROCESS_ACCT
139 extern int acct_parm[];
140 #endif
142 #ifdef CONFIG_IA64
143 extern int no_unaligned_warning;
144 #endif
146 #ifdef CONFIG_RT_MUTEXES
147 extern int max_lock_depth;
148 #endif
150 #ifdef CONFIG_PROC_SYSCTL
151 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
152 void __user *buffer, size_t *lenp, loff_t *ppos);
153 static int proc_dointvec_taint(struct ctl_table *table, int write, struct file *filp,
154 void __user *buffer, size_t *lenp, loff_t *ppos);
155 #endif
157 static struct ctl_table root_table[];
158 static struct ctl_table_root sysctl_table_root;
159 static struct ctl_table_header root_table_header = {
160 .count = 1,
161 .ctl_table = root_table,
162 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
163 .root = &sysctl_table_root,
164 .set = &sysctl_table_root.default_set,
166 static struct ctl_table_root sysctl_table_root = {
167 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
168 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
171 static struct ctl_table kern_table[];
172 static struct ctl_table vm_table[];
173 static struct ctl_table fs_table[];
174 static struct ctl_table debug_table[];
175 static struct ctl_table dev_table[];
176 extern struct ctl_table random_table[];
177 #ifdef CONFIG_INOTIFY_USER
178 extern struct ctl_table inotify_table[];
179 #endif
181 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
182 int sysctl_legacy_va_layout;
183 #endif
185 extern int prove_locking;
186 extern int lock_stat;
188 /* The default sysctl tables: */
190 static struct ctl_table root_table[] = {
192 .ctl_name = CTL_KERN,
193 .procname = "kernel",
194 .mode = 0555,
195 .child = kern_table,
198 .ctl_name = CTL_VM,
199 .procname = "vm",
200 .mode = 0555,
201 .child = vm_table,
204 .ctl_name = CTL_FS,
205 .procname = "fs",
206 .mode = 0555,
207 .child = fs_table,
210 .ctl_name = CTL_DEBUG,
211 .procname = "debug",
212 .mode = 0555,
213 .child = debug_table,
216 .ctl_name = CTL_DEV,
217 .procname = "dev",
218 .mode = 0555,
219 .child = dev_table,
222 * NOTE: do not add new entries to this table unless you have read
223 * Documentation/sysctl/ctl_unnumbered.txt
225 { .ctl_name = 0 }
228 #ifdef CONFIG_SCHED_DEBUG
229 static int min_sched_granularity_ns = 100000; /* 100 usecs */
230 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
231 static int min_wakeup_granularity_ns; /* 0 usecs */
232 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
233 #endif
235 static struct ctl_table kern_table[] = {
236 #ifdef CONFIG_SCHED_DEBUG
238 .ctl_name = CTL_UNNUMBERED,
239 .procname = "sched_min_granularity_ns",
240 .data = &sysctl_sched_min_granularity,
241 .maxlen = sizeof(unsigned int),
242 .mode = 0644,
243 .proc_handler = &sched_nr_latency_handler,
244 .strategy = &sysctl_intvec,
245 .extra1 = &min_sched_granularity_ns,
246 .extra2 = &max_sched_granularity_ns,
249 .ctl_name = CTL_UNNUMBERED,
250 .procname = "sched_latency_ns",
251 .data = &sysctl_sched_latency,
252 .maxlen = sizeof(unsigned int),
253 .mode = 0644,
254 .proc_handler = &sched_nr_latency_handler,
255 .strategy = &sysctl_intvec,
256 .extra1 = &min_sched_granularity_ns,
257 .extra2 = &max_sched_granularity_ns,
260 .ctl_name = CTL_UNNUMBERED,
261 .procname = "sched_wakeup_granularity_ns",
262 .data = &sysctl_sched_wakeup_granularity,
263 .maxlen = sizeof(unsigned int),
264 .mode = 0644,
265 .proc_handler = &proc_dointvec_minmax,
266 .strategy = &sysctl_intvec,
267 .extra1 = &min_wakeup_granularity_ns,
268 .extra2 = &max_wakeup_granularity_ns,
271 .ctl_name = CTL_UNNUMBERED,
272 .procname = "sched_shares_ratelimit",
273 .data = &sysctl_sched_shares_ratelimit,
274 .maxlen = sizeof(unsigned int),
275 .mode = 0644,
276 .proc_handler = &proc_dointvec,
279 .ctl_name = CTL_UNNUMBERED,
280 .procname = "sched_child_runs_first",
281 .data = &sysctl_sched_child_runs_first,
282 .maxlen = sizeof(unsigned int),
283 .mode = 0644,
284 .proc_handler = &proc_dointvec,
287 .ctl_name = CTL_UNNUMBERED,
288 .procname = "sched_features",
289 .data = &sysctl_sched_features,
290 .maxlen = sizeof(unsigned int),
291 .mode = 0644,
292 .proc_handler = &proc_dointvec,
295 .ctl_name = CTL_UNNUMBERED,
296 .procname = "sched_migration_cost",
297 .data = &sysctl_sched_migration_cost,
298 .maxlen = sizeof(unsigned int),
299 .mode = 0644,
300 .proc_handler = &proc_dointvec,
303 .ctl_name = CTL_UNNUMBERED,
304 .procname = "sched_nr_migrate",
305 .data = &sysctl_sched_nr_migrate,
306 .maxlen = sizeof(unsigned int),
307 .mode = 0644,
308 .proc_handler = &proc_dointvec,
310 #endif
312 .ctl_name = CTL_UNNUMBERED,
313 .procname = "sched_rt_period_us",
314 .data = &sysctl_sched_rt_period,
315 .maxlen = sizeof(unsigned int),
316 .mode = 0644,
317 .proc_handler = &sched_rt_handler,
320 .ctl_name = CTL_UNNUMBERED,
321 .procname = "sched_rt_runtime_us",
322 .data = &sysctl_sched_rt_runtime,
323 .maxlen = sizeof(int),
324 .mode = 0644,
325 .proc_handler = &sched_rt_handler,
328 .ctl_name = CTL_UNNUMBERED,
329 .procname = "sched_compat_yield",
330 .data = &sysctl_sched_compat_yield,
331 .maxlen = sizeof(unsigned int),
332 .mode = 0644,
333 .proc_handler = &proc_dointvec,
335 #ifdef CONFIG_PROVE_LOCKING
337 .ctl_name = CTL_UNNUMBERED,
338 .procname = "prove_locking",
339 .data = &prove_locking,
340 .maxlen = sizeof(int),
341 .mode = 0644,
342 .proc_handler = &proc_dointvec,
344 #endif
345 #ifdef CONFIG_LOCK_STAT
347 .ctl_name = CTL_UNNUMBERED,
348 .procname = "lock_stat",
349 .data = &lock_stat,
350 .maxlen = sizeof(int),
351 .mode = 0644,
352 .proc_handler = &proc_dointvec,
354 #endif
356 .ctl_name = KERN_PANIC,
357 .procname = "panic",
358 .data = &panic_timeout,
359 .maxlen = sizeof(int),
360 .mode = 0644,
361 .proc_handler = &proc_dointvec,
364 .ctl_name = KERN_CORE_USES_PID,
365 .procname = "core_uses_pid",
366 .data = &core_uses_pid,
367 .maxlen = sizeof(int),
368 .mode = 0644,
369 .proc_handler = &proc_dointvec,
372 .ctl_name = KERN_CORE_PATTERN,
373 .procname = "core_pattern",
374 .data = core_pattern,
375 .maxlen = CORENAME_MAX_SIZE,
376 .mode = 0644,
377 .proc_handler = &proc_dostring,
378 .strategy = &sysctl_string,
380 #ifdef CONFIG_PROC_SYSCTL
382 .procname = "tainted",
383 .data = &tainted,
384 .maxlen = sizeof(int),
385 .mode = 0644,
386 .proc_handler = &proc_dointvec_taint,
388 #endif
389 #ifdef CONFIG_LATENCYTOP
391 .procname = "latencytop",
392 .data = &latencytop_enabled,
393 .maxlen = sizeof(int),
394 .mode = 0644,
395 .proc_handler = &proc_dointvec,
397 #endif
398 #ifdef CONFIG_BLK_DEV_INITRD
400 .ctl_name = KERN_REALROOTDEV,
401 .procname = "real-root-dev",
402 .data = &real_root_dev,
403 .maxlen = sizeof(int),
404 .mode = 0644,
405 .proc_handler = &proc_dointvec,
407 #endif
409 .ctl_name = CTL_UNNUMBERED,
410 .procname = "print-fatal-signals",
411 .data = &print_fatal_signals,
412 .maxlen = sizeof(int),
413 .mode = 0644,
414 .proc_handler = &proc_dointvec,
416 #ifdef CONFIG_SPARC
418 .ctl_name = KERN_SPARC_REBOOT,
419 .procname = "reboot-cmd",
420 .data = reboot_command,
421 .maxlen = 256,
422 .mode = 0644,
423 .proc_handler = &proc_dostring,
424 .strategy = &sysctl_string,
427 .ctl_name = KERN_SPARC_STOP_A,
428 .procname = "stop-a",
429 .data = &stop_a_enabled,
430 .maxlen = sizeof (int),
431 .mode = 0644,
432 .proc_handler = &proc_dointvec,
435 .ctl_name = KERN_SPARC_SCONS_PWROFF,
436 .procname = "scons-poweroff",
437 .data = &scons_pwroff,
438 .maxlen = sizeof (int),
439 .mode = 0644,
440 .proc_handler = &proc_dointvec,
442 #endif
443 #ifdef __hppa__
445 .ctl_name = KERN_HPPA_PWRSW,
446 .procname = "soft-power",
447 .data = &pwrsw_enabled,
448 .maxlen = sizeof (int),
449 .mode = 0644,
450 .proc_handler = &proc_dointvec,
453 .ctl_name = KERN_HPPA_UNALIGNED,
454 .procname = "unaligned-trap",
455 .data = &unaligned_enabled,
456 .maxlen = sizeof (int),
457 .mode = 0644,
458 .proc_handler = &proc_dointvec,
460 #endif
462 .ctl_name = KERN_CTLALTDEL,
463 .procname = "ctrl-alt-del",
464 .data = &C_A_D,
465 .maxlen = sizeof(int),
466 .mode = 0644,
467 .proc_handler = &proc_dointvec,
469 #ifdef CONFIG_FTRACE
471 .ctl_name = CTL_UNNUMBERED,
472 .procname = "ftrace_enabled",
473 .data = &ftrace_enabled,
474 .maxlen = sizeof(int),
475 .mode = 0644,
476 .proc_handler = &ftrace_enable_sysctl,
478 #endif
479 #ifdef CONFIG_MODULES
481 .ctl_name = KERN_MODPROBE,
482 .procname = "modprobe",
483 .data = &modprobe_path,
484 .maxlen = KMOD_PATH_LEN,
485 .mode = 0644,
486 .proc_handler = &proc_dostring,
487 .strategy = &sysctl_string,
489 #endif
490 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
492 .ctl_name = KERN_HOTPLUG,
493 .procname = "hotplug",
494 .data = &uevent_helper,
495 .maxlen = UEVENT_HELPER_PATH_LEN,
496 .mode = 0644,
497 .proc_handler = &proc_dostring,
498 .strategy = &sysctl_string,
500 #endif
501 #ifdef CONFIG_CHR_DEV_SG
503 .ctl_name = KERN_SG_BIG_BUFF,
504 .procname = "sg-big-buff",
505 .data = &sg_big_buff,
506 .maxlen = sizeof (int),
507 .mode = 0444,
508 .proc_handler = &proc_dointvec,
510 #endif
511 #ifdef CONFIG_BSD_PROCESS_ACCT
513 .ctl_name = KERN_ACCT,
514 .procname = "acct",
515 .data = &acct_parm,
516 .maxlen = 3*sizeof(int),
517 .mode = 0644,
518 .proc_handler = &proc_dointvec,
520 #endif
521 #ifdef CONFIG_MAGIC_SYSRQ
523 .ctl_name = KERN_SYSRQ,
524 .procname = "sysrq",
525 .data = &__sysrq_enabled,
526 .maxlen = sizeof (int),
527 .mode = 0644,
528 .proc_handler = &proc_dointvec,
530 #endif
531 #ifdef CONFIG_PROC_SYSCTL
533 .procname = "cad_pid",
534 .data = NULL,
535 .maxlen = sizeof (int),
536 .mode = 0600,
537 .proc_handler = &proc_do_cad_pid,
539 #endif
541 .ctl_name = KERN_MAX_THREADS,
542 .procname = "threads-max",
543 .data = &max_threads,
544 .maxlen = sizeof(int),
545 .mode = 0644,
546 .proc_handler = &proc_dointvec,
549 .ctl_name = KERN_RANDOM,
550 .procname = "random",
551 .mode = 0555,
552 .child = random_table,
555 .ctl_name = KERN_OVERFLOWUID,
556 .procname = "overflowuid",
557 .data = &overflowuid,
558 .maxlen = sizeof(int),
559 .mode = 0644,
560 .proc_handler = &proc_dointvec_minmax,
561 .strategy = &sysctl_intvec,
562 .extra1 = &minolduid,
563 .extra2 = &maxolduid,
566 .ctl_name = KERN_OVERFLOWGID,
567 .procname = "overflowgid",
568 .data = &overflowgid,
569 .maxlen = sizeof(int),
570 .mode = 0644,
571 .proc_handler = &proc_dointvec_minmax,
572 .strategy = &sysctl_intvec,
573 .extra1 = &minolduid,
574 .extra2 = &maxolduid,
576 #ifdef CONFIG_S390
577 #ifdef CONFIG_MATHEMU
579 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
580 .procname = "ieee_emulation_warnings",
581 .data = &sysctl_ieee_emulation_warnings,
582 .maxlen = sizeof(int),
583 .mode = 0644,
584 .proc_handler = &proc_dointvec,
586 #endif
588 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
589 .procname = "userprocess_debug",
590 .data = &sysctl_userprocess_debug,
591 .maxlen = sizeof(int),
592 .mode = 0644,
593 .proc_handler = &proc_dointvec,
595 #endif
597 .ctl_name = KERN_PIDMAX,
598 .procname = "pid_max",
599 .data = &pid_max,
600 .maxlen = sizeof (int),
601 .mode = 0644,
602 .proc_handler = &proc_dointvec_minmax,
603 .strategy = sysctl_intvec,
604 .extra1 = &pid_max_min,
605 .extra2 = &pid_max_max,
608 .ctl_name = KERN_PANIC_ON_OOPS,
609 .procname = "panic_on_oops",
610 .data = &panic_on_oops,
611 .maxlen = sizeof(int),
612 .mode = 0644,
613 .proc_handler = &proc_dointvec,
615 #if defined CONFIG_PRINTK
617 .ctl_name = KERN_PRINTK,
618 .procname = "printk",
619 .data = &console_loglevel,
620 .maxlen = 4*sizeof(int),
621 .mode = 0644,
622 .proc_handler = &proc_dointvec,
625 .ctl_name = KERN_PRINTK_RATELIMIT,
626 .procname = "printk_ratelimit",
627 .data = &printk_ratelimit_state.interval,
628 .maxlen = sizeof(int),
629 .mode = 0644,
630 .proc_handler = &proc_dointvec_jiffies,
631 .strategy = &sysctl_jiffies,
634 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
635 .procname = "printk_ratelimit_burst",
636 .data = &printk_ratelimit_state.burst,
637 .maxlen = sizeof(int),
638 .mode = 0644,
639 .proc_handler = &proc_dointvec,
641 #endif
643 .ctl_name = KERN_NGROUPS_MAX,
644 .procname = "ngroups_max",
645 .data = &ngroups_max,
646 .maxlen = sizeof (int),
647 .mode = 0444,
648 .proc_handler = &proc_dointvec,
650 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
652 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
653 .procname = "unknown_nmi_panic",
654 .data = &unknown_nmi_panic,
655 .maxlen = sizeof (int),
656 .mode = 0644,
657 .proc_handler = &proc_dointvec,
660 .procname = "nmi_watchdog",
661 .data = &nmi_watchdog_enabled,
662 .maxlen = sizeof (int),
663 .mode = 0644,
664 .proc_handler = &proc_nmi_enabled,
666 #endif
667 #if defined(CONFIG_X86)
669 .ctl_name = KERN_PANIC_ON_NMI,
670 .procname = "panic_on_unrecovered_nmi",
671 .data = &panic_on_unrecovered_nmi,
672 .maxlen = sizeof(int),
673 .mode = 0644,
674 .proc_handler = &proc_dointvec,
677 .ctl_name = KERN_BOOTLOADER_TYPE,
678 .procname = "bootloader_type",
679 .data = &bootloader_type,
680 .maxlen = sizeof (int),
681 .mode = 0444,
682 .proc_handler = &proc_dointvec,
685 .ctl_name = CTL_UNNUMBERED,
686 .procname = "kstack_depth_to_print",
687 .data = &kstack_depth_to_print,
688 .maxlen = sizeof(int),
689 .mode = 0644,
690 .proc_handler = &proc_dointvec,
693 .ctl_name = CTL_UNNUMBERED,
694 .procname = "io_delay_type",
695 .data = &io_delay_type,
696 .maxlen = sizeof(int),
697 .mode = 0644,
698 .proc_handler = &proc_dointvec,
700 #endif
701 #if defined(CONFIG_MMU)
703 .ctl_name = KERN_RANDOMIZE,
704 .procname = "randomize_va_space",
705 .data = &randomize_va_space,
706 .maxlen = sizeof(int),
707 .mode = 0644,
708 .proc_handler = &proc_dointvec,
710 #endif
711 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
713 .ctl_name = KERN_SPIN_RETRY,
714 .procname = "spin_retry",
715 .data = &spin_retry,
716 .maxlen = sizeof (int),
717 .mode = 0644,
718 .proc_handler = &proc_dointvec,
720 #endif
721 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
723 .procname = "acpi_video_flags",
724 .data = &acpi_realmode_flags,
725 .maxlen = sizeof (unsigned long),
726 .mode = 0644,
727 .proc_handler = &proc_doulongvec_minmax,
729 #endif
730 #ifdef CONFIG_IA64
732 .ctl_name = KERN_IA64_UNALIGNED,
733 .procname = "ignore-unaligned-usertrap",
734 .data = &no_unaligned_warning,
735 .maxlen = sizeof (int),
736 .mode = 0644,
737 .proc_handler = &proc_dointvec,
739 #endif
740 #ifdef CONFIG_DETECT_SOFTLOCKUP
742 .ctl_name = CTL_UNNUMBERED,
743 .procname = "softlockup_panic",
744 .data = &softlockup_panic,
745 .maxlen = sizeof(int),
746 .mode = 0644,
747 .proc_handler = &proc_dointvec_minmax,
748 .strategy = &sysctl_intvec,
749 .extra1 = &zero,
750 .extra2 = &one,
753 .ctl_name = CTL_UNNUMBERED,
754 .procname = "softlockup_thresh",
755 .data = &softlockup_thresh,
756 .maxlen = sizeof(int),
757 .mode = 0644,
758 .proc_handler = &proc_dointvec_minmax,
759 .strategy = &sysctl_intvec,
760 .extra1 = &neg_one,
761 .extra2 = &sixty,
764 .ctl_name = CTL_UNNUMBERED,
765 .procname = "hung_task_check_count",
766 .data = &sysctl_hung_task_check_count,
767 .maxlen = sizeof(unsigned long),
768 .mode = 0644,
769 .proc_handler = &proc_doulongvec_minmax,
770 .strategy = &sysctl_intvec,
773 .ctl_name = CTL_UNNUMBERED,
774 .procname = "hung_task_timeout_secs",
775 .data = &sysctl_hung_task_timeout_secs,
776 .maxlen = sizeof(unsigned long),
777 .mode = 0644,
778 .proc_handler = &proc_doulongvec_minmax,
779 .strategy = &sysctl_intvec,
782 .ctl_name = CTL_UNNUMBERED,
783 .procname = "hung_task_warnings",
784 .data = &sysctl_hung_task_warnings,
785 .maxlen = sizeof(unsigned long),
786 .mode = 0644,
787 .proc_handler = &proc_doulongvec_minmax,
788 .strategy = &sysctl_intvec,
790 #endif
791 #ifdef CONFIG_COMPAT
793 .ctl_name = KERN_COMPAT_LOG,
794 .procname = "compat-log",
795 .data = &compat_log,
796 .maxlen = sizeof (int),
797 .mode = 0644,
798 .proc_handler = &proc_dointvec,
800 #endif
801 #ifdef CONFIG_RT_MUTEXES
803 .ctl_name = KERN_MAX_LOCK_DEPTH,
804 .procname = "max_lock_depth",
805 .data = &max_lock_depth,
806 .maxlen = sizeof(int),
807 .mode = 0644,
808 .proc_handler = &proc_dointvec,
810 #endif
811 #ifdef CONFIG_PROC_FS
813 .ctl_name = CTL_UNNUMBERED,
814 .procname = "maps_protect",
815 .data = &maps_protect,
816 .maxlen = sizeof(int),
817 .mode = 0644,
818 .proc_handler = &proc_dointvec,
820 #endif
822 .ctl_name = CTL_UNNUMBERED,
823 .procname = "poweroff_cmd",
824 .data = &poweroff_cmd,
825 .maxlen = POWEROFF_CMD_PATH_LEN,
826 .mode = 0644,
827 .proc_handler = &proc_dostring,
828 .strategy = &sysctl_string,
830 #ifdef CONFIG_KEYS
832 .ctl_name = CTL_UNNUMBERED,
833 .procname = "keys",
834 .mode = 0555,
835 .child = key_sysctls,
837 #endif
838 #ifdef CONFIG_RCU_TORTURE_TEST
840 .ctl_name = CTL_UNNUMBERED,
841 .procname = "rcutorture_runnable",
842 .data = &rcutorture_runnable,
843 .maxlen = sizeof(int),
844 .mode = 0644,
845 .proc_handler = &proc_dointvec,
847 #endif
849 * NOTE: do not add new entries to this table unless you have read
850 * Documentation/sysctl/ctl_unnumbered.txt
852 { .ctl_name = 0 }
855 static struct ctl_table vm_table[] = {
857 .ctl_name = VM_OVERCOMMIT_MEMORY,
858 .procname = "overcommit_memory",
859 .data = &sysctl_overcommit_memory,
860 .maxlen = sizeof(sysctl_overcommit_memory),
861 .mode = 0644,
862 .proc_handler = &proc_dointvec,
865 .ctl_name = VM_PANIC_ON_OOM,
866 .procname = "panic_on_oom",
867 .data = &sysctl_panic_on_oom,
868 .maxlen = sizeof(sysctl_panic_on_oom),
869 .mode = 0644,
870 .proc_handler = &proc_dointvec,
873 .ctl_name = CTL_UNNUMBERED,
874 .procname = "oom_kill_allocating_task",
875 .data = &sysctl_oom_kill_allocating_task,
876 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
877 .mode = 0644,
878 .proc_handler = &proc_dointvec,
881 .ctl_name = CTL_UNNUMBERED,
882 .procname = "oom_dump_tasks",
883 .data = &sysctl_oom_dump_tasks,
884 .maxlen = sizeof(sysctl_oom_dump_tasks),
885 .mode = 0644,
886 .proc_handler = &proc_dointvec,
889 .ctl_name = VM_OVERCOMMIT_RATIO,
890 .procname = "overcommit_ratio",
891 .data = &sysctl_overcommit_ratio,
892 .maxlen = sizeof(sysctl_overcommit_ratio),
893 .mode = 0644,
894 .proc_handler = &proc_dointvec,
897 .ctl_name = VM_PAGE_CLUSTER,
898 .procname = "page-cluster",
899 .data = &page_cluster,
900 .maxlen = sizeof(int),
901 .mode = 0644,
902 .proc_handler = &proc_dointvec,
905 .ctl_name = VM_DIRTY_BACKGROUND,
906 .procname = "dirty_background_ratio",
907 .data = &dirty_background_ratio,
908 .maxlen = sizeof(dirty_background_ratio),
909 .mode = 0644,
910 .proc_handler = &proc_dointvec_minmax,
911 .strategy = &sysctl_intvec,
912 .extra1 = &zero,
913 .extra2 = &one_hundred,
916 .ctl_name = VM_DIRTY_RATIO,
917 .procname = "dirty_ratio",
918 .data = &vm_dirty_ratio,
919 .maxlen = sizeof(vm_dirty_ratio),
920 .mode = 0644,
921 .proc_handler = &dirty_ratio_handler,
922 .strategy = &sysctl_intvec,
923 .extra1 = &zero,
924 .extra2 = &one_hundred,
927 .procname = "dirty_writeback_centisecs",
928 .data = &dirty_writeback_interval,
929 .maxlen = sizeof(dirty_writeback_interval),
930 .mode = 0644,
931 .proc_handler = &dirty_writeback_centisecs_handler,
934 .procname = "dirty_expire_centisecs",
935 .data = &dirty_expire_interval,
936 .maxlen = sizeof(dirty_expire_interval),
937 .mode = 0644,
938 .proc_handler = &proc_dointvec_userhz_jiffies,
941 .ctl_name = VM_NR_PDFLUSH_THREADS,
942 .procname = "nr_pdflush_threads",
943 .data = &nr_pdflush_threads,
944 .maxlen = sizeof nr_pdflush_threads,
945 .mode = 0444 /* read-only*/,
946 .proc_handler = &proc_dointvec,
949 .ctl_name = VM_SWAPPINESS,
950 .procname = "swappiness",
951 .data = &vm_swappiness,
952 .maxlen = sizeof(vm_swappiness),
953 .mode = 0644,
954 .proc_handler = &proc_dointvec_minmax,
955 .strategy = &sysctl_intvec,
956 .extra1 = &zero,
957 .extra2 = &one_hundred,
959 #ifdef CONFIG_HUGETLB_PAGE
961 .procname = "nr_hugepages",
962 .data = NULL,
963 .maxlen = sizeof(unsigned long),
964 .mode = 0644,
965 .proc_handler = &hugetlb_sysctl_handler,
966 .extra1 = (void *)&hugetlb_zero,
967 .extra2 = (void *)&hugetlb_infinity,
970 .ctl_name = VM_HUGETLB_GROUP,
971 .procname = "hugetlb_shm_group",
972 .data = &sysctl_hugetlb_shm_group,
973 .maxlen = sizeof(gid_t),
974 .mode = 0644,
975 .proc_handler = &proc_dointvec,
978 .ctl_name = CTL_UNNUMBERED,
979 .procname = "hugepages_treat_as_movable",
980 .data = &hugepages_treat_as_movable,
981 .maxlen = sizeof(int),
982 .mode = 0644,
983 .proc_handler = &hugetlb_treat_movable_handler,
986 .ctl_name = CTL_UNNUMBERED,
987 .procname = "nr_overcommit_hugepages",
988 .data = NULL,
989 .maxlen = sizeof(unsigned long),
990 .mode = 0644,
991 .proc_handler = &hugetlb_overcommit_handler,
992 .extra1 = (void *)&hugetlb_zero,
993 .extra2 = (void *)&hugetlb_infinity,
995 #endif
997 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
998 .procname = "lowmem_reserve_ratio",
999 .data = &sysctl_lowmem_reserve_ratio,
1000 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1001 .mode = 0644,
1002 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
1003 .strategy = &sysctl_intvec,
1006 .ctl_name = VM_DROP_PAGECACHE,
1007 .procname = "drop_caches",
1008 .data = &sysctl_drop_caches,
1009 .maxlen = sizeof(int),
1010 .mode = 0644,
1011 .proc_handler = drop_caches_sysctl_handler,
1012 .strategy = &sysctl_intvec,
1015 .ctl_name = VM_MIN_FREE_KBYTES,
1016 .procname = "min_free_kbytes",
1017 .data = &min_free_kbytes,
1018 .maxlen = sizeof(min_free_kbytes),
1019 .mode = 0644,
1020 .proc_handler = &min_free_kbytes_sysctl_handler,
1021 .strategy = &sysctl_intvec,
1022 .extra1 = &zero,
1025 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
1026 .procname = "percpu_pagelist_fraction",
1027 .data = &percpu_pagelist_fraction,
1028 .maxlen = sizeof(percpu_pagelist_fraction),
1029 .mode = 0644,
1030 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
1031 .strategy = &sysctl_intvec,
1032 .extra1 = &min_percpu_pagelist_fract,
1034 #ifdef CONFIG_MMU
1036 .ctl_name = VM_MAX_MAP_COUNT,
1037 .procname = "max_map_count",
1038 .data = &sysctl_max_map_count,
1039 .maxlen = sizeof(sysctl_max_map_count),
1040 .mode = 0644,
1041 .proc_handler = &proc_dointvec
1043 #endif
1045 .ctl_name = VM_LAPTOP_MODE,
1046 .procname = "laptop_mode",
1047 .data = &laptop_mode,
1048 .maxlen = sizeof(laptop_mode),
1049 .mode = 0644,
1050 .proc_handler = &proc_dointvec_jiffies,
1051 .strategy = &sysctl_jiffies,
1054 .ctl_name = VM_BLOCK_DUMP,
1055 .procname = "block_dump",
1056 .data = &block_dump,
1057 .maxlen = sizeof(block_dump),
1058 .mode = 0644,
1059 .proc_handler = &proc_dointvec,
1060 .strategy = &sysctl_intvec,
1061 .extra1 = &zero,
1064 .ctl_name = VM_VFS_CACHE_PRESSURE,
1065 .procname = "vfs_cache_pressure",
1066 .data = &sysctl_vfs_cache_pressure,
1067 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1068 .mode = 0644,
1069 .proc_handler = &proc_dointvec,
1070 .strategy = &sysctl_intvec,
1071 .extra1 = &zero,
1073 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1075 .ctl_name = VM_LEGACY_VA_LAYOUT,
1076 .procname = "legacy_va_layout",
1077 .data = &sysctl_legacy_va_layout,
1078 .maxlen = sizeof(sysctl_legacy_va_layout),
1079 .mode = 0644,
1080 .proc_handler = &proc_dointvec,
1081 .strategy = &sysctl_intvec,
1082 .extra1 = &zero,
1084 #endif
1085 #ifdef CONFIG_NUMA
1087 .ctl_name = VM_ZONE_RECLAIM_MODE,
1088 .procname = "zone_reclaim_mode",
1089 .data = &zone_reclaim_mode,
1090 .maxlen = sizeof(zone_reclaim_mode),
1091 .mode = 0644,
1092 .proc_handler = &proc_dointvec,
1093 .strategy = &sysctl_intvec,
1094 .extra1 = &zero,
1097 .ctl_name = VM_MIN_UNMAPPED,
1098 .procname = "min_unmapped_ratio",
1099 .data = &sysctl_min_unmapped_ratio,
1100 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1101 .mode = 0644,
1102 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
1103 .strategy = &sysctl_intvec,
1104 .extra1 = &zero,
1105 .extra2 = &one_hundred,
1108 .ctl_name = VM_MIN_SLAB,
1109 .procname = "min_slab_ratio",
1110 .data = &sysctl_min_slab_ratio,
1111 .maxlen = sizeof(sysctl_min_slab_ratio),
1112 .mode = 0644,
1113 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1114 .strategy = &sysctl_intvec,
1115 .extra1 = &zero,
1116 .extra2 = &one_hundred,
1118 #endif
1119 #ifdef CONFIG_SMP
1121 .ctl_name = CTL_UNNUMBERED,
1122 .procname = "stat_interval",
1123 .data = &sysctl_stat_interval,
1124 .maxlen = sizeof(sysctl_stat_interval),
1125 .mode = 0644,
1126 .proc_handler = &proc_dointvec_jiffies,
1127 .strategy = &sysctl_jiffies,
1129 #endif
1130 #ifdef CONFIG_SECURITY
1132 .ctl_name = CTL_UNNUMBERED,
1133 .procname = "mmap_min_addr",
1134 .data = &mmap_min_addr,
1135 .maxlen = sizeof(unsigned long),
1136 .mode = 0644,
1137 .proc_handler = &proc_doulongvec_minmax,
1139 #endif
1140 #ifdef CONFIG_NUMA
1142 .ctl_name = CTL_UNNUMBERED,
1143 .procname = "numa_zonelist_order",
1144 .data = &numa_zonelist_order,
1145 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1146 .mode = 0644,
1147 .proc_handler = &numa_zonelist_order_handler,
1148 .strategy = &sysctl_string,
1150 #endif
1151 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1152 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1154 .ctl_name = VM_VDSO_ENABLED,
1155 .procname = "vdso_enabled",
1156 .data = &vdso_enabled,
1157 .maxlen = sizeof(vdso_enabled),
1158 .mode = 0644,
1159 .proc_handler = &proc_dointvec,
1160 .strategy = &sysctl_intvec,
1161 .extra1 = &zero,
1163 #endif
1164 #ifdef CONFIG_HIGHMEM
1166 .ctl_name = CTL_UNNUMBERED,
1167 .procname = "highmem_is_dirtyable",
1168 .data = &vm_highmem_is_dirtyable,
1169 .maxlen = sizeof(vm_highmem_is_dirtyable),
1170 .mode = 0644,
1171 .proc_handler = &proc_dointvec_minmax,
1172 .strategy = &sysctl_intvec,
1173 .extra1 = &zero,
1174 .extra2 = &one,
1176 #endif
1178 * NOTE: do not add new entries to this table unless you have read
1179 * Documentation/sysctl/ctl_unnumbered.txt
1181 { .ctl_name = 0 }
1184 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1185 static struct ctl_table binfmt_misc_table[] = {
1186 { .ctl_name = 0 }
1188 #endif
1190 static struct ctl_table fs_table[] = {
1192 .ctl_name = FS_NRINODE,
1193 .procname = "inode-nr",
1194 .data = &inodes_stat,
1195 .maxlen = 2*sizeof(int),
1196 .mode = 0444,
1197 .proc_handler = &proc_dointvec,
1200 .ctl_name = FS_STATINODE,
1201 .procname = "inode-state",
1202 .data = &inodes_stat,
1203 .maxlen = 7*sizeof(int),
1204 .mode = 0444,
1205 .proc_handler = &proc_dointvec,
1208 .procname = "file-nr",
1209 .data = &files_stat,
1210 .maxlen = 3*sizeof(int),
1211 .mode = 0444,
1212 .proc_handler = &proc_nr_files,
1215 .ctl_name = FS_MAXFILE,
1216 .procname = "file-max",
1217 .data = &files_stat.max_files,
1218 .maxlen = sizeof(int),
1219 .mode = 0644,
1220 .proc_handler = &proc_dointvec,
1223 .ctl_name = CTL_UNNUMBERED,
1224 .procname = "nr_open",
1225 .data = &sysctl_nr_open,
1226 .maxlen = sizeof(int),
1227 .mode = 0644,
1228 .proc_handler = &proc_dointvec_minmax,
1229 .extra1 = &sysctl_nr_open_min,
1230 .extra2 = &sysctl_nr_open_max,
1233 .ctl_name = FS_DENTRY,
1234 .procname = "dentry-state",
1235 .data = &dentry_stat,
1236 .maxlen = 6*sizeof(int),
1237 .mode = 0444,
1238 .proc_handler = &proc_dointvec,
1241 .ctl_name = FS_OVERFLOWUID,
1242 .procname = "overflowuid",
1243 .data = &fs_overflowuid,
1244 .maxlen = sizeof(int),
1245 .mode = 0644,
1246 .proc_handler = &proc_dointvec_minmax,
1247 .strategy = &sysctl_intvec,
1248 .extra1 = &minolduid,
1249 .extra2 = &maxolduid,
1252 .ctl_name = FS_OVERFLOWGID,
1253 .procname = "overflowgid",
1254 .data = &fs_overflowgid,
1255 .maxlen = sizeof(int),
1256 .mode = 0644,
1257 .proc_handler = &proc_dointvec_minmax,
1258 .strategy = &sysctl_intvec,
1259 .extra1 = &minolduid,
1260 .extra2 = &maxolduid,
1263 .ctl_name = FS_LEASES,
1264 .procname = "leases-enable",
1265 .data = &leases_enable,
1266 .maxlen = sizeof(int),
1267 .mode = 0644,
1268 .proc_handler = &proc_dointvec,
1270 #ifdef CONFIG_DNOTIFY
1272 .ctl_name = FS_DIR_NOTIFY,
1273 .procname = "dir-notify-enable",
1274 .data = &dir_notify_enable,
1275 .maxlen = sizeof(int),
1276 .mode = 0644,
1277 .proc_handler = &proc_dointvec,
1279 #endif
1280 #ifdef CONFIG_MMU
1282 .ctl_name = FS_LEASE_TIME,
1283 .procname = "lease-break-time",
1284 .data = &lease_break_time,
1285 .maxlen = sizeof(int),
1286 .mode = 0644,
1287 .proc_handler = &proc_dointvec_minmax,
1288 .strategy = &sysctl_intvec,
1289 .extra1 = &zero,
1290 .extra2 = &two,
1293 .procname = "aio-nr",
1294 .data = &aio_nr,
1295 .maxlen = sizeof(aio_nr),
1296 .mode = 0444,
1297 .proc_handler = &proc_doulongvec_minmax,
1300 .procname = "aio-max-nr",
1301 .data = &aio_max_nr,
1302 .maxlen = sizeof(aio_max_nr),
1303 .mode = 0644,
1304 .proc_handler = &proc_doulongvec_minmax,
1306 #ifdef CONFIG_INOTIFY_USER
1308 .ctl_name = FS_INOTIFY,
1309 .procname = "inotify",
1310 .mode = 0555,
1311 .child = inotify_table,
1313 #endif
1314 #endif
1316 .ctl_name = KERN_SETUID_DUMPABLE,
1317 .procname = "suid_dumpable",
1318 .data = &suid_dumpable,
1319 .maxlen = sizeof(int),
1320 .mode = 0644,
1321 .proc_handler = &proc_dointvec,
1323 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1325 .ctl_name = CTL_UNNUMBERED,
1326 .procname = "binfmt_misc",
1327 .mode = 0555,
1328 .child = binfmt_misc_table,
1330 #endif
1332 * NOTE: do not add new entries to this table unless you have read
1333 * Documentation/sysctl/ctl_unnumbered.txt
1335 { .ctl_name = 0 }
1338 static struct ctl_table debug_table[] = {
1339 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1341 .ctl_name = CTL_UNNUMBERED,
1342 .procname = "exception-trace",
1343 .data = &show_unhandled_signals,
1344 .maxlen = sizeof(int),
1345 .mode = 0644,
1346 .proc_handler = proc_dointvec
1348 #endif
1349 { .ctl_name = 0 }
1352 static struct ctl_table dev_table[] = {
1353 { .ctl_name = 0 }
1356 static DEFINE_SPINLOCK(sysctl_lock);
1358 /* called under sysctl_lock */
1359 static int use_table(struct ctl_table_header *p)
1361 if (unlikely(p->unregistering))
1362 return 0;
1363 p->used++;
1364 return 1;
1367 /* called under sysctl_lock */
1368 static void unuse_table(struct ctl_table_header *p)
1370 if (!--p->used)
1371 if (unlikely(p->unregistering))
1372 complete(p->unregistering);
1375 /* called under sysctl_lock, will reacquire if has to wait */
1376 static void start_unregistering(struct ctl_table_header *p)
1379 * if p->used is 0, nobody will ever touch that entry again;
1380 * we'll eliminate all paths to it before dropping sysctl_lock
1382 if (unlikely(p->used)) {
1383 struct completion wait;
1384 init_completion(&wait);
1385 p->unregistering = &wait;
1386 spin_unlock(&sysctl_lock);
1387 wait_for_completion(&wait);
1388 spin_lock(&sysctl_lock);
1389 } else {
1390 /* anything non-NULL; we'll never dereference it */
1391 p->unregistering = ERR_PTR(-EINVAL);
1394 * do not remove from the list until nobody holds it; walking the
1395 * list in do_sysctl() relies on that.
1397 list_del_init(&p->ctl_entry);
1400 void sysctl_head_get(struct ctl_table_header *head)
1402 spin_lock(&sysctl_lock);
1403 head->count++;
1404 spin_unlock(&sysctl_lock);
1407 void sysctl_head_put(struct ctl_table_header *head)
1409 spin_lock(&sysctl_lock);
1410 if (!--head->count)
1411 kfree(head);
1412 spin_unlock(&sysctl_lock);
1415 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1417 if (!head)
1418 BUG();
1419 spin_lock(&sysctl_lock);
1420 if (!use_table(head))
1421 head = ERR_PTR(-ENOENT);
1422 spin_unlock(&sysctl_lock);
1423 return head;
1426 void sysctl_head_finish(struct ctl_table_header *head)
1428 if (!head)
1429 return;
1430 spin_lock(&sysctl_lock);
1431 unuse_table(head);
1432 spin_unlock(&sysctl_lock);
1435 static struct ctl_table_set *
1436 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1438 struct ctl_table_set *set = &root->default_set;
1439 if (root->lookup)
1440 set = root->lookup(root, namespaces);
1441 return set;
1444 static struct list_head *
1445 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1447 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1448 return &set->list;
1451 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1452 struct ctl_table_header *prev)
1454 struct ctl_table_root *root;
1455 struct list_head *header_list;
1456 struct ctl_table_header *head;
1457 struct list_head *tmp;
1459 spin_lock(&sysctl_lock);
1460 if (prev) {
1461 head = prev;
1462 tmp = &prev->ctl_entry;
1463 unuse_table(prev);
1464 goto next;
1466 tmp = &root_table_header.ctl_entry;
1467 for (;;) {
1468 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1470 if (!use_table(head))
1471 goto next;
1472 spin_unlock(&sysctl_lock);
1473 return head;
1474 next:
1475 root = head->root;
1476 tmp = tmp->next;
1477 header_list = lookup_header_list(root, namespaces);
1478 if (tmp != header_list)
1479 continue;
1481 do {
1482 root = list_entry(root->root_list.next,
1483 struct ctl_table_root, root_list);
1484 if (root == &sysctl_table_root)
1485 goto out;
1486 header_list = lookup_header_list(root, namespaces);
1487 } while (list_empty(header_list));
1488 tmp = header_list->next;
1490 out:
1491 spin_unlock(&sysctl_lock);
1492 return NULL;
1495 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1497 return __sysctl_head_next(current->nsproxy, prev);
1500 void register_sysctl_root(struct ctl_table_root *root)
1502 spin_lock(&sysctl_lock);
1503 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1504 spin_unlock(&sysctl_lock);
1507 #ifdef CONFIG_SYSCTL_SYSCALL
1508 /* Perform the actual read/write of a sysctl table entry. */
1509 static int do_sysctl_strategy(struct ctl_table_root *root,
1510 struct ctl_table *table,
1511 int __user *name, int nlen,
1512 void __user *oldval, size_t __user *oldlenp,
1513 void __user *newval, size_t newlen)
1515 int op = 0, rc;
1517 if (oldval)
1518 op |= MAY_READ;
1519 if (newval)
1520 op |= MAY_WRITE;
1521 if (sysctl_perm(root, table, op))
1522 return -EPERM;
1524 if (table->strategy) {
1525 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1526 newval, newlen);
1527 if (rc < 0)
1528 return rc;
1529 if (rc > 0)
1530 return 0;
1533 /* If there is no strategy routine, or if the strategy returns
1534 * zero, proceed with automatic r/w */
1535 if (table->data && table->maxlen) {
1536 rc = sysctl_data(table, name, nlen, oldval, oldlenp,
1537 newval, newlen);
1538 if (rc < 0)
1539 return rc;
1541 return 0;
1544 static int parse_table(int __user *name, int nlen,
1545 void __user *oldval, size_t __user *oldlenp,
1546 void __user *newval, size_t newlen,
1547 struct ctl_table_root *root,
1548 struct ctl_table *table)
1550 int n;
1551 repeat:
1552 if (!nlen)
1553 return -ENOTDIR;
1554 if (get_user(n, name))
1555 return -EFAULT;
1556 for ( ; table->ctl_name || table->procname; table++) {
1557 if (!table->ctl_name)
1558 continue;
1559 if (n == table->ctl_name) {
1560 int error;
1561 if (table->child) {
1562 if (sysctl_perm(root, table, MAY_EXEC))
1563 return -EPERM;
1564 name++;
1565 nlen--;
1566 table = table->child;
1567 goto repeat;
1569 error = do_sysctl_strategy(root, table, name, nlen,
1570 oldval, oldlenp,
1571 newval, newlen);
1572 return error;
1575 return -ENOTDIR;
1578 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1579 void __user *newval, size_t newlen)
1581 struct ctl_table_header *head;
1582 int error = -ENOTDIR;
1584 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1585 return -ENOTDIR;
1586 if (oldval) {
1587 int old_len;
1588 if (!oldlenp || get_user(old_len, oldlenp))
1589 return -EFAULT;
1592 for (head = sysctl_head_next(NULL); head;
1593 head = sysctl_head_next(head)) {
1594 error = parse_table(name, nlen, oldval, oldlenp,
1595 newval, newlen,
1596 head->root, head->ctl_table);
1597 if (error != -ENOTDIR) {
1598 sysctl_head_finish(head);
1599 break;
1602 return error;
1605 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1607 struct __sysctl_args tmp;
1608 int error;
1610 if (copy_from_user(&tmp, args, sizeof(tmp)))
1611 return -EFAULT;
1613 error = deprecated_sysctl_warning(&tmp);
1614 if (error)
1615 goto out;
1617 lock_kernel();
1618 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1619 tmp.newval, tmp.newlen);
1620 unlock_kernel();
1621 out:
1622 return error;
1624 #endif /* CONFIG_SYSCTL_SYSCALL */
1627 * sysctl_perm does NOT grant the superuser all rights automatically, because
1628 * some sysctl variables are readonly even to root.
1631 static int test_perm(int mode, int op)
1633 if (!current->euid)
1634 mode >>= 6;
1635 else if (in_egroup_p(0))
1636 mode >>= 3;
1637 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1638 return 0;
1639 return -EACCES;
1642 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1644 int error;
1645 int mode;
1647 error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1648 if (error)
1649 return error;
1651 if (root->permissions)
1652 mode = root->permissions(root, current->nsproxy, table);
1653 else
1654 mode = table->mode;
1656 return test_perm(mode, op);
1659 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1661 for (; table->ctl_name || table->procname; table++) {
1662 table->parent = parent;
1663 if (table->child)
1664 sysctl_set_parent(table, table->child);
1668 static __init int sysctl_init(void)
1670 sysctl_set_parent(NULL, root_table);
1671 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1673 int err;
1674 err = sysctl_check_table(current->nsproxy, root_table);
1676 #endif
1677 return 0;
1680 core_initcall(sysctl_init);
1682 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1683 struct ctl_table *table)
1685 struct ctl_table *p;
1686 const char *s = branch->procname;
1688 /* branch should have named subdirectory as its first element */
1689 if (!s || !branch->child)
1690 return NULL;
1692 /* ... and nothing else */
1693 if (branch[1].procname || branch[1].ctl_name)
1694 return NULL;
1696 /* table should contain subdirectory with the same name */
1697 for (p = table; p->procname || p->ctl_name; p++) {
1698 if (!p->child)
1699 continue;
1700 if (p->procname && strcmp(p->procname, s) == 0)
1701 return p;
1703 return NULL;
1706 /* see if attaching q to p would be an improvement */
1707 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1709 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1710 struct ctl_table *next;
1711 int is_better = 0;
1712 int not_in_parent = !p->attached_by;
1714 while ((next = is_branch_in(by, to)) != NULL) {
1715 if (by == q->attached_by)
1716 is_better = 1;
1717 if (to == p->attached_by)
1718 not_in_parent = 1;
1719 by = by->child;
1720 to = next->child;
1723 if (is_better && not_in_parent) {
1724 q->attached_by = by;
1725 q->attached_to = to;
1726 q->parent = p;
1731 * __register_sysctl_paths - register a sysctl hierarchy
1732 * @root: List of sysctl headers to register on
1733 * @namespaces: Data to compute which lists of sysctl entries are visible
1734 * @path: The path to the directory the sysctl table is in.
1735 * @table: the top-level table structure
1737 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1738 * array. A completely 0 filled entry terminates the table.
1740 * The members of the &struct ctl_table structure are used as follows:
1742 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1743 * must be unique within that level of sysctl
1745 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1746 * enter a sysctl file
1748 * data - a pointer to data for use by proc_handler
1750 * maxlen - the maximum size in bytes of the data
1752 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1754 * child - a pointer to the child sysctl table if this entry is a directory, or
1755 * %NULL.
1757 * proc_handler - the text handler routine (described below)
1759 * strategy - the strategy routine (described below)
1761 * de - for internal use by the sysctl routines
1763 * extra1, extra2 - extra pointers usable by the proc handler routines
1765 * Leaf nodes in the sysctl tree will be represented by a single file
1766 * under /proc; non-leaf nodes will be represented by directories.
1768 * sysctl(2) can automatically manage read and write requests through
1769 * the sysctl table. The data and maxlen fields of the ctl_table
1770 * struct enable minimal validation of the values being written to be
1771 * performed, and the mode field allows minimal authentication.
1773 * More sophisticated management can be enabled by the provision of a
1774 * strategy routine with the table entry. This will be called before
1775 * any automatic read or write of the data is performed.
1777 * The strategy routine may return
1779 * < 0 - Error occurred (error is passed to user process)
1781 * 0 - OK - proceed with automatic read or write.
1783 * > 0 - OK - read or write has been done by the strategy routine, so
1784 * return immediately.
1786 * There must be a proc_handler routine for any terminal nodes
1787 * mirrored under /proc/sys (non-terminals are handled by a built-in
1788 * directory handler). Several default handlers are available to
1789 * cover common cases -
1791 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1792 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1793 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1795 * It is the handler's job to read the input buffer from user memory
1796 * and process it. The handler should return 0 on success.
1798 * This routine returns %NULL on a failure to register, and a pointer
1799 * to the table header on success.
1801 struct ctl_table_header *__register_sysctl_paths(
1802 struct ctl_table_root *root,
1803 struct nsproxy *namespaces,
1804 const struct ctl_path *path, struct ctl_table *table)
1806 struct ctl_table_header *header;
1807 struct ctl_table *new, **prevp;
1808 unsigned int n, npath;
1809 struct ctl_table_set *set;
1811 /* Count the path components */
1812 for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1816 * For each path component, allocate a 2-element ctl_table array.
1817 * The first array element will be filled with the sysctl entry
1818 * for this, the second will be the sentinel (ctl_name == 0).
1820 * We allocate everything in one go so that we don't have to
1821 * worry about freeing additional memory in unregister_sysctl_table.
1823 header = kzalloc(sizeof(struct ctl_table_header) +
1824 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1825 if (!header)
1826 return NULL;
1828 new = (struct ctl_table *) (header + 1);
1830 /* Now connect the dots */
1831 prevp = &header->ctl_table;
1832 for (n = 0; n < npath; ++n, ++path) {
1833 /* Copy the procname */
1834 new->procname = path->procname;
1835 new->ctl_name = path->ctl_name;
1836 new->mode = 0555;
1838 *prevp = new;
1839 prevp = &new->child;
1841 new += 2;
1843 *prevp = table;
1844 header->ctl_table_arg = table;
1846 INIT_LIST_HEAD(&header->ctl_entry);
1847 header->used = 0;
1848 header->unregistering = NULL;
1849 header->root = root;
1850 sysctl_set_parent(NULL, header->ctl_table);
1851 header->count = 1;
1852 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1853 if (sysctl_check_table(namespaces, header->ctl_table)) {
1854 kfree(header);
1855 return NULL;
1857 #endif
1858 spin_lock(&sysctl_lock);
1859 header->set = lookup_header_set(root, namespaces);
1860 header->attached_by = header->ctl_table;
1861 header->attached_to = root_table;
1862 header->parent = &root_table_header;
1863 for (set = header->set; set; set = set->parent) {
1864 struct ctl_table_header *p;
1865 list_for_each_entry(p, &set->list, ctl_entry) {
1866 if (p->unregistering)
1867 continue;
1868 try_attach(p, header);
1871 header->parent->count++;
1872 list_add_tail(&header->ctl_entry, &header->set->list);
1873 spin_unlock(&sysctl_lock);
1875 return header;
1879 * register_sysctl_table_path - register a sysctl table hierarchy
1880 * @path: The path to the directory the sysctl table is in.
1881 * @table: the top-level table structure
1883 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1884 * array. A completely 0 filled entry terminates the table.
1886 * See __register_sysctl_paths for more details.
1888 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1889 struct ctl_table *table)
1891 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1892 path, table);
1896 * register_sysctl_table - register a sysctl table hierarchy
1897 * @table: the top-level table structure
1899 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1900 * array. A completely 0 filled entry terminates the table.
1902 * See register_sysctl_paths for more details.
1904 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1906 static const struct ctl_path null_path[] = { {} };
1908 return register_sysctl_paths(null_path, table);
1912 * unregister_sysctl_table - unregister a sysctl table hierarchy
1913 * @header: the header returned from register_sysctl_table
1915 * Unregisters the sysctl table and all children. proc entries may not
1916 * actually be removed until they are no longer used by anyone.
1918 void unregister_sysctl_table(struct ctl_table_header * header)
1920 might_sleep();
1922 if (header == NULL)
1923 return;
1925 spin_lock(&sysctl_lock);
1926 start_unregistering(header);
1927 if (!--header->parent->count) {
1928 WARN_ON(1);
1929 kfree(header->parent);
1931 if (!--header->count)
1932 kfree(header);
1933 spin_unlock(&sysctl_lock);
1936 int sysctl_is_seen(struct ctl_table_header *p)
1938 struct ctl_table_set *set = p->set;
1939 int res;
1940 spin_lock(&sysctl_lock);
1941 if (p->unregistering)
1942 res = 0;
1943 else if (!set->is_seen)
1944 res = 1;
1945 else
1946 res = set->is_seen(set);
1947 spin_unlock(&sysctl_lock);
1948 return res;
1951 void setup_sysctl_set(struct ctl_table_set *p,
1952 struct ctl_table_set *parent,
1953 int (*is_seen)(struct ctl_table_set *))
1955 INIT_LIST_HEAD(&p->list);
1956 p->parent = parent ? parent : &sysctl_table_root.default_set;
1957 p->is_seen = is_seen;
1960 #else /* !CONFIG_SYSCTL */
1961 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1963 return NULL;
1966 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1967 struct ctl_table *table)
1969 return NULL;
1972 void unregister_sysctl_table(struct ctl_table_header * table)
1976 void setup_sysctl_set(struct ctl_table_set *p,
1977 struct ctl_table_set *parent,
1978 int (*is_seen)(struct ctl_table_set *))
1982 void sysctl_head_put(struct ctl_table_header *head)
1986 #endif /* CONFIG_SYSCTL */
1989 * /proc/sys support
1992 #ifdef CONFIG_PROC_SYSCTL
1994 static int _proc_do_string(void* data, int maxlen, int write,
1995 struct file *filp, void __user *buffer,
1996 size_t *lenp, loff_t *ppos)
1998 size_t len;
1999 char __user *p;
2000 char c;
2002 if (!data || !maxlen || !*lenp) {
2003 *lenp = 0;
2004 return 0;
2007 if (write) {
2008 len = 0;
2009 p = buffer;
2010 while (len < *lenp) {
2011 if (get_user(c, p++))
2012 return -EFAULT;
2013 if (c == 0 || c == '\n')
2014 break;
2015 len++;
2017 if (len >= maxlen)
2018 len = maxlen-1;
2019 if(copy_from_user(data, buffer, len))
2020 return -EFAULT;
2021 ((char *) data)[len] = 0;
2022 *ppos += *lenp;
2023 } else {
2024 len = strlen(data);
2025 if (len > maxlen)
2026 len = maxlen;
2028 if (*ppos > len) {
2029 *lenp = 0;
2030 return 0;
2033 data += *ppos;
2034 len -= *ppos;
2036 if (len > *lenp)
2037 len = *lenp;
2038 if (len)
2039 if(copy_to_user(buffer, data, len))
2040 return -EFAULT;
2041 if (len < *lenp) {
2042 if(put_user('\n', ((char __user *) buffer) + len))
2043 return -EFAULT;
2044 len++;
2046 *lenp = len;
2047 *ppos += len;
2049 return 0;
2053 * proc_dostring - read a string sysctl
2054 * @table: the sysctl table
2055 * @write: %TRUE if this is a write to the sysctl file
2056 * @filp: the file structure
2057 * @buffer: the user buffer
2058 * @lenp: the size of the user buffer
2059 * @ppos: file position
2061 * Reads/writes a string from/to the user buffer. If the kernel
2062 * buffer provided is not large enough to hold the string, the
2063 * string is truncated. The copied string is %NULL-terminated.
2064 * If the string is being read by the user process, it is copied
2065 * and a newline '\n' is added. It is truncated if the buffer is
2066 * not large enough.
2068 * Returns 0 on success.
2070 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2071 void __user *buffer, size_t *lenp, loff_t *ppos)
2073 return _proc_do_string(table->data, table->maxlen, write, filp,
2074 buffer, lenp, ppos);
2078 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2079 int *valp,
2080 int write, void *data)
2082 if (write) {
2083 *valp = *negp ? -*lvalp : *lvalp;
2084 } else {
2085 int val = *valp;
2086 if (val < 0) {
2087 *negp = -1;
2088 *lvalp = (unsigned long)-val;
2089 } else {
2090 *negp = 0;
2091 *lvalp = (unsigned long)val;
2094 return 0;
2097 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2098 int write, struct file *filp, void __user *buffer,
2099 size_t *lenp, loff_t *ppos,
2100 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2101 int write, void *data),
2102 void *data)
2104 #define TMPBUFLEN 21
2105 int *i, vleft, first=1, neg, val;
2106 unsigned long lval;
2107 size_t left, len;
2109 char buf[TMPBUFLEN], *p;
2110 char __user *s = buffer;
2112 if (!tbl_data || !table->maxlen || !*lenp ||
2113 (*ppos && !write)) {
2114 *lenp = 0;
2115 return 0;
2118 i = (int *) tbl_data;
2119 vleft = table->maxlen / sizeof(*i);
2120 left = *lenp;
2122 if (!conv)
2123 conv = do_proc_dointvec_conv;
2125 for (; left && vleft--; i++, first=0) {
2126 if (write) {
2127 while (left) {
2128 char c;
2129 if (get_user(c, s))
2130 return -EFAULT;
2131 if (!isspace(c))
2132 break;
2133 left--;
2134 s++;
2136 if (!left)
2137 break;
2138 neg = 0;
2139 len = left;
2140 if (len > sizeof(buf) - 1)
2141 len = sizeof(buf) - 1;
2142 if (copy_from_user(buf, s, len))
2143 return -EFAULT;
2144 buf[len] = 0;
2145 p = buf;
2146 if (*p == '-' && left > 1) {
2147 neg = 1;
2148 p++;
2150 if (*p < '0' || *p > '9')
2151 break;
2153 lval = simple_strtoul(p, &p, 0);
2155 len = p-buf;
2156 if ((len < left) && *p && !isspace(*p))
2157 break;
2158 if (neg)
2159 val = -val;
2160 s += len;
2161 left -= len;
2163 if (conv(&neg, &lval, i, 1, data))
2164 break;
2165 } else {
2166 p = buf;
2167 if (!first)
2168 *p++ = '\t';
2170 if (conv(&neg, &lval, i, 0, data))
2171 break;
2173 sprintf(p, "%s%lu", neg ? "-" : "", lval);
2174 len = strlen(buf);
2175 if (len > left)
2176 len = left;
2177 if(copy_to_user(s, buf, len))
2178 return -EFAULT;
2179 left -= len;
2180 s += len;
2184 if (!write && !first && left) {
2185 if(put_user('\n', s))
2186 return -EFAULT;
2187 left--, s++;
2189 if (write) {
2190 while (left) {
2191 char c;
2192 if (get_user(c, s++))
2193 return -EFAULT;
2194 if (!isspace(c))
2195 break;
2196 left--;
2199 if (write && first)
2200 return -EINVAL;
2201 *lenp -= left;
2202 *ppos += *lenp;
2203 return 0;
2204 #undef TMPBUFLEN
2207 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2208 void __user *buffer, size_t *lenp, loff_t *ppos,
2209 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2210 int write, void *data),
2211 void *data)
2213 return __do_proc_dointvec(table->data, table, write, filp,
2214 buffer, lenp, ppos, conv, data);
2218 * proc_dointvec - read a vector of integers
2219 * @table: the sysctl table
2220 * @write: %TRUE if this is a write to the sysctl file
2221 * @filp: the file structure
2222 * @buffer: the user buffer
2223 * @lenp: the size of the user buffer
2224 * @ppos: file position
2226 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2227 * values from/to the user buffer, treated as an ASCII string.
2229 * Returns 0 on success.
2231 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2232 void __user *buffer, size_t *lenp, loff_t *ppos)
2234 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2235 NULL,NULL);
2238 #define OP_SET 0
2239 #define OP_AND 1
2240 #define OP_OR 2
2242 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
2243 int *valp,
2244 int write, void *data)
2246 int op = *(int *)data;
2247 if (write) {
2248 int val = *negp ? -*lvalp : *lvalp;
2249 switch(op) {
2250 case OP_SET: *valp = val; break;
2251 case OP_AND: *valp &= val; break;
2252 case OP_OR: *valp |= val; break;
2254 } else {
2255 int val = *valp;
2256 if (val < 0) {
2257 *negp = -1;
2258 *lvalp = (unsigned long)-val;
2259 } else {
2260 *negp = 0;
2261 *lvalp = (unsigned long)val;
2264 return 0;
2268 * Taint values can only be increased
2270 static int proc_dointvec_taint(struct ctl_table *table, int write, struct file *filp,
2271 void __user *buffer, size_t *lenp, loff_t *ppos)
2273 int op;
2275 if (write && !capable(CAP_SYS_ADMIN))
2276 return -EPERM;
2278 op = OP_OR;
2279 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2280 do_proc_dointvec_bset_conv,&op);
2283 struct do_proc_dointvec_minmax_conv_param {
2284 int *min;
2285 int *max;
2288 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2289 int *valp,
2290 int write, void *data)
2292 struct do_proc_dointvec_minmax_conv_param *param = data;
2293 if (write) {
2294 int val = *negp ? -*lvalp : *lvalp;
2295 if ((param->min && *param->min > val) ||
2296 (param->max && *param->max < val))
2297 return -EINVAL;
2298 *valp = val;
2299 } else {
2300 int val = *valp;
2301 if (val < 0) {
2302 *negp = -1;
2303 *lvalp = (unsigned long)-val;
2304 } else {
2305 *negp = 0;
2306 *lvalp = (unsigned long)val;
2309 return 0;
2313 * proc_dointvec_minmax - read a vector of integers with min/max values
2314 * @table: the sysctl table
2315 * @write: %TRUE if this is a write to the sysctl file
2316 * @filp: the file structure
2317 * @buffer: the user buffer
2318 * @lenp: the size of the user buffer
2319 * @ppos: file position
2321 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2322 * values from/to the user buffer, treated as an ASCII string.
2324 * This routine will ensure the values are within the range specified by
2325 * table->extra1 (min) and table->extra2 (max).
2327 * Returns 0 on success.
2329 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2330 void __user *buffer, size_t *lenp, loff_t *ppos)
2332 struct do_proc_dointvec_minmax_conv_param param = {
2333 .min = (int *) table->extra1,
2334 .max = (int *) table->extra2,
2336 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2337 do_proc_dointvec_minmax_conv, &param);
2340 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2341 struct file *filp,
2342 void __user *buffer,
2343 size_t *lenp, loff_t *ppos,
2344 unsigned long convmul,
2345 unsigned long convdiv)
2347 #define TMPBUFLEN 21
2348 unsigned long *i, *min, *max, val;
2349 int vleft, first=1, neg;
2350 size_t len, left;
2351 char buf[TMPBUFLEN], *p;
2352 char __user *s = buffer;
2354 if (!data || !table->maxlen || !*lenp ||
2355 (*ppos && !write)) {
2356 *lenp = 0;
2357 return 0;
2360 i = (unsigned long *) data;
2361 min = (unsigned long *) table->extra1;
2362 max = (unsigned long *) table->extra2;
2363 vleft = table->maxlen / sizeof(unsigned long);
2364 left = *lenp;
2366 for (; left && vleft--; i++, min++, max++, first=0) {
2367 if (write) {
2368 while (left) {
2369 char c;
2370 if (get_user(c, s))
2371 return -EFAULT;
2372 if (!isspace(c))
2373 break;
2374 left--;
2375 s++;
2377 if (!left)
2378 break;
2379 neg = 0;
2380 len = left;
2381 if (len > TMPBUFLEN-1)
2382 len = TMPBUFLEN-1;
2383 if (copy_from_user(buf, s, len))
2384 return -EFAULT;
2385 buf[len] = 0;
2386 p = buf;
2387 if (*p == '-' && left > 1) {
2388 neg = 1;
2389 p++;
2391 if (*p < '0' || *p > '9')
2392 break;
2393 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2394 len = p-buf;
2395 if ((len < left) && *p && !isspace(*p))
2396 break;
2397 if (neg)
2398 val = -val;
2399 s += len;
2400 left -= len;
2402 if(neg)
2403 continue;
2404 if ((min && val < *min) || (max && val > *max))
2405 continue;
2406 *i = val;
2407 } else {
2408 p = buf;
2409 if (!first)
2410 *p++ = '\t';
2411 sprintf(p, "%lu", convdiv * (*i) / convmul);
2412 len = strlen(buf);
2413 if (len > left)
2414 len = left;
2415 if(copy_to_user(s, buf, len))
2416 return -EFAULT;
2417 left -= len;
2418 s += len;
2422 if (!write && !first && left) {
2423 if(put_user('\n', s))
2424 return -EFAULT;
2425 left--, s++;
2427 if (write) {
2428 while (left) {
2429 char c;
2430 if (get_user(c, s++))
2431 return -EFAULT;
2432 if (!isspace(c))
2433 break;
2434 left--;
2437 if (write && first)
2438 return -EINVAL;
2439 *lenp -= left;
2440 *ppos += *lenp;
2441 return 0;
2442 #undef TMPBUFLEN
2445 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2446 struct file *filp,
2447 void __user *buffer,
2448 size_t *lenp, loff_t *ppos,
2449 unsigned long convmul,
2450 unsigned long convdiv)
2452 return __do_proc_doulongvec_minmax(table->data, table, write,
2453 filp, buffer, lenp, ppos, convmul, convdiv);
2457 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2458 * @table: the sysctl table
2459 * @write: %TRUE if this is a write to the sysctl file
2460 * @filp: the file structure
2461 * @buffer: the user buffer
2462 * @lenp: the size of the user buffer
2463 * @ppos: file position
2465 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2466 * values from/to the user buffer, treated as an ASCII string.
2468 * This routine will ensure the values are within the range specified by
2469 * table->extra1 (min) and table->extra2 (max).
2471 * Returns 0 on success.
2473 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2474 void __user *buffer, size_t *lenp, loff_t *ppos)
2476 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2480 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2481 * @table: the sysctl table
2482 * @write: %TRUE if this is a write to the sysctl file
2483 * @filp: the file structure
2484 * @buffer: the user buffer
2485 * @lenp: the size of the user buffer
2486 * @ppos: file position
2488 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2489 * values from/to the user buffer, treated as an ASCII string. The values
2490 * are treated as milliseconds, and converted to jiffies when they are stored.
2492 * This routine will ensure the values are within the range specified by
2493 * table->extra1 (min) and table->extra2 (max).
2495 * Returns 0 on success.
2497 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2498 struct file *filp,
2499 void __user *buffer,
2500 size_t *lenp, loff_t *ppos)
2502 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2503 lenp, ppos, HZ, 1000l);
2507 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2508 int *valp,
2509 int write, void *data)
2511 if (write) {
2512 if (*lvalp > LONG_MAX / HZ)
2513 return 1;
2514 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2515 } else {
2516 int val = *valp;
2517 unsigned long lval;
2518 if (val < 0) {
2519 *negp = -1;
2520 lval = (unsigned long)-val;
2521 } else {
2522 *negp = 0;
2523 lval = (unsigned long)val;
2525 *lvalp = lval / HZ;
2527 return 0;
2530 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2531 int *valp,
2532 int write, void *data)
2534 if (write) {
2535 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2536 return 1;
2537 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2538 } else {
2539 int val = *valp;
2540 unsigned long lval;
2541 if (val < 0) {
2542 *negp = -1;
2543 lval = (unsigned long)-val;
2544 } else {
2545 *negp = 0;
2546 lval = (unsigned long)val;
2548 *lvalp = jiffies_to_clock_t(lval);
2550 return 0;
2553 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2554 int *valp,
2555 int write, void *data)
2557 if (write) {
2558 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2559 } else {
2560 int val = *valp;
2561 unsigned long lval;
2562 if (val < 0) {
2563 *negp = -1;
2564 lval = (unsigned long)-val;
2565 } else {
2566 *negp = 0;
2567 lval = (unsigned long)val;
2569 *lvalp = jiffies_to_msecs(lval);
2571 return 0;
2575 * proc_dointvec_jiffies - read a vector of integers as seconds
2576 * @table: the sysctl table
2577 * @write: %TRUE if this is a write to the sysctl file
2578 * @filp: the file structure
2579 * @buffer: the user buffer
2580 * @lenp: the size of the user buffer
2581 * @ppos: file position
2583 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2584 * values from/to the user buffer, treated as an ASCII string.
2585 * The values read are assumed to be in seconds, and are converted into
2586 * jiffies.
2588 * Returns 0 on success.
2590 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2591 void __user *buffer, size_t *lenp, loff_t *ppos)
2593 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2594 do_proc_dointvec_jiffies_conv,NULL);
2598 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2599 * @table: the sysctl table
2600 * @write: %TRUE if this is a write to the sysctl file
2601 * @filp: the file structure
2602 * @buffer: the user buffer
2603 * @lenp: the size of the user buffer
2604 * @ppos: pointer to the file position
2606 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2607 * values from/to the user buffer, treated as an ASCII string.
2608 * The values read are assumed to be in 1/USER_HZ seconds, and
2609 * are converted into jiffies.
2611 * Returns 0 on success.
2613 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2614 void __user *buffer, size_t *lenp, loff_t *ppos)
2616 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2617 do_proc_dointvec_userhz_jiffies_conv,NULL);
2621 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2622 * @table: the sysctl table
2623 * @write: %TRUE if this is a write to the sysctl file
2624 * @filp: the file structure
2625 * @buffer: the user buffer
2626 * @lenp: the size of the user buffer
2627 * @ppos: file position
2628 * @ppos: the current position in the file
2630 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2631 * values from/to the user buffer, treated as an ASCII string.
2632 * The values read are assumed to be in 1/1000 seconds, and
2633 * are converted into jiffies.
2635 * Returns 0 on success.
2637 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2638 void __user *buffer, size_t *lenp, loff_t *ppos)
2640 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2641 do_proc_dointvec_ms_jiffies_conv, NULL);
2644 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2645 void __user *buffer, size_t *lenp, loff_t *ppos)
2647 struct pid *new_pid;
2648 pid_t tmp;
2649 int r;
2651 tmp = pid_vnr(cad_pid);
2653 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2654 lenp, ppos, NULL, NULL);
2655 if (r || !write)
2656 return r;
2658 new_pid = find_get_pid(tmp);
2659 if (!new_pid)
2660 return -ESRCH;
2662 put_pid(xchg(&cad_pid, new_pid));
2663 return 0;
2666 #else /* CONFIG_PROC_FS */
2668 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2669 void __user *buffer, size_t *lenp, loff_t *ppos)
2671 return -ENOSYS;
2674 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2675 void __user *buffer, size_t *lenp, loff_t *ppos)
2677 return -ENOSYS;
2680 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2681 void __user *buffer, size_t *lenp, loff_t *ppos)
2683 return -ENOSYS;
2686 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2687 void __user *buffer, size_t *lenp, loff_t *ppos)
2689 return -ENOSYS;
2692 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2693 void __user *buffer, size_t *lenp, loff_t *ppos)
2695 return -ENOSYS;
2698 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2699 void __user *buffer, size_t *lenp, loff_t *ppos)
2701 return -ENOSYS;
2704 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2705 void __user *buffer, size_t *lenp, loff_t *ppos)
2707 return -ENOSYS;
2710 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2711 struct file *filp,
2712 void __user *buffer,
2713 size_t *lenp, loff_t *ppos)
2715 return -ENOSYS;
2719 #endif /* CONFIG_PROC_FS */
2722 #ifdef CONFIG_SYSCTL_SYSCALL
2724 * General sysctl support routines
2727 /* The generic sysctl data routine (used if no strategy routine supplied) */
2728 int sysctl_data(struct ctl_table *table, int __user *name, int nlen,
2729 void __user *oldval, size_t __user *oldlenp,
2730 void __user *newval, size_t newlen)
2732 size_t len;
2734 /* Get out of I don't have a variable */
2735 if (!table->data || !table->maxlen)
2736 return -ENOTDIR;
2738 if (oldval && oldlenp) {
2739 if (get_user(len, oldlenp))
2740 return -EFAULT;
2741 if (len) {
2742 if (len > table->maxlen)
2743 len = table->maxlen;
2744 if (copy_to_user(oldval, table->data, len))
2745 return -EFAULT;
2746 if (put_user(len, oldlenp))
2747 return -EFAULT;
2751 if (newval && newlen) {
2752 if (newlen > table->maxlen)
2753 newlen = table->maxlen;
2755 if (copy_from_user(table->data, newval, newlen))
2756 return -EFAULT;
2758 return 1;
2761 /* The generic string strategy routine: */
2762 int sysctl_string(struct ctl_table *table, int __user *name, int nlen,
2763 void __user *oldval, size_t __user *oldlenp,
2764 void __user *newval, size_t newlen)
2766 if (!table->data || !table->maxlen)
2767 return -ENOTDIR;
2769 if (oldval && oldlenp) {
2770 size_t bufsize;
2771 if (get_user(bufsize, oldlenp))
2772 return -EFAULT;
2773 if (bufsize) {
2774 size_t len = strlen(table->data), copied;
2776 /* This shouldn't trigger for a well-formed sysctl */
2777 if (len > table->maxlen)
2778 len = table->maxlen;
2780 /* Copy up to a max of bufsize-1 bytes of the string */
2781 copied = (len >= bufsize) ? bufsize - 1 : len;
2783 if (copy_to_user(oldval, table->data, copied) ||
2784 put_user(0, (char __user *)(oldval + copied)))
2785 return -EFAULT;
2786 if (put_user(len, oldlenp))
2787 return -EFAULT;
2790 if (newval && newlen) {
2791 size_t len = newlen;
2792 if (len > table->maxlen)
2793 len = table->maxlen;
2794 if(copy_from_user(table->data, newval, len))
2795 return -EFAULT;
2796 if (len == table->maxlen)
2797 len--;
2798 ((char *) table->data)[len] = 0;
2800 return 1;
2804 * This function makes sure that all of the integers in the vector
2805 * are between the minimum and maximum values given in the arrays
2806 * table->extra1 and table->extra2, respectively.
2808 int sysctl_intvec(struct ctl_table *table, int __user *name, int nlen,
2809 void __user *oldval, size_t __user *oldlenp,
2810 void __user *newval, size_t newlen)
2813 if (newval && newlen) {
2814 int __user *vec = (int __user *) newval;
2815 int *min = (int *) table->extra1;
2816 int *max = (int *) table->extra2;
2817 size_t length;
2818 int i;
2820 if (newlen % sizeof(int) != 0)
2821 return -EINVAL;
2823 if (!table->extra1 && !table->extra2)
2824 return 0;
2826 if (newlen > table->maxlen)
2827 newlen = table->maxlen;
2828 length = newlen / sizeof(int);
2830 for (i = 0; i < length; i++) {
2831 int value;
2832 if (get_user(value, vec + i))
2833 return -EFAULT;
2834 if (min && value < min[i])
2835 return -EINVAL;
2836 if (max && value > max[i])
2837 return -EINVAL;
2840 return 0;
2843 /* Strategy function to convert jiffies to seconds */
2844 int sysctl_jiffies(struct ctl_table *table, int __user *name, int nlen,
2845 void __user *oldval, size_t __user *oldlenp,
2846 void __user *newval, size_t newlen)
2848 if (oldval && oldlenp) {
2849 size_t olen;
2851 if (get_user(olen, oldlenp))
2852 return -EFAULT;
2853 if (olen) {
2854 int val;
2856 if (olen < sizeof(int))
2857 return -EINVAL;
2859 val = *(int *)(table->data) / HZ;
2860 if (put_user(val, (int __user *)oldval))
2861 return -EFAULT;
2862 if (put_user(sizeof(int), oldlenp))
2863 return -EFAULT;
2866 if (newval && newlen) {
2867 int new;
2868 if (newlen != sizeof(int))
2869 return -EINVAL;
2870 if (get_user(new, (int __user *)newval))
2871 return -EFAULT;
2872 *(int *)(table->data) = new*HZ;
2874 return 1;
2877 /* Strategy function to convert jiffies to seconds */
2878 int sysctl_ms_jiffies(struct ctl_table *table, int __user *name, int nlen,
2879 void __user *oldval, size_t __user *oldlenp,
2880 void __user *newval, size_t newlen)
2882 if (oldval && oldlenp) {
2883 size_t olen;
2885 if (get_user(olen, oldlenp))
2886 return -EFAULT;
2887 if (olen) {
2888 int val;
2890 if (olen < sizeof(int))
2891 return -EINVAL;
2893 val = jiffies_to_msecs(*(int *)(table->data));
2894 if (put_user(val, (int __user *)oldval))
2895 return -EFAULT;
2896 if (put_user(sizeof(int), oldlenp))
2897 return -EFAULT;
2900 if (newval && newlen) {
2901 int new;
2902 if (newlen != sizeof(int))
2903 return -EINVAL;
2904 if (get_user(new, (int __user *)newval))
2905 return -EFAULT;
2906 *(int *)(table->data) = msecs_to_jiffies(new);
2908 return 1;
2913 #else /* CONFIG_SYSCTL_SYSCALL */
2916 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2918 struct __sysctl_args tmp;
2919 int error;
2921 if (copy_from_user(&tmp, args, sizeof(tmp)))
2922 return -EFAULT;
2924 error = deprecated_sysctl_warning(&tmp);
2926 /* If no error reading the parameters then just -ENOSYS ... */
2927 if (!error)
2928 error = -ENOSYS;
2930 return error;
2933 int sysctl_data(struct ctl_table *table, int __user *name, int nlen,
2934 void __user *oldval, size_t __user *oldlenp,
2935 void __user *newval, size_t newlen)
2937 return -ENOSYS;
2940 int sysctl_string(struct ctl_table *table, int __user *name, int nlen,
2941 void __user *oldval, size_t __user *oldlenp,
2942 void __user *newval, size_t newlen)
2944 return -ENOSYS;
2947 int sysctl_intvec(struct ctl_table *table, int __user *name, int nlen,
2948 void __user *oldval, size_t __user *oldlenp,
2949 void __user *newval, size_t newlen)
2951 return -ENOSYS;
2954 int sysctl_jiffies(struct ctl_table *table, int __user *name, int nlen,
2955 void __user *oldval, size_t __user *oldlenp,
2956 void __user *newval, size_t newlen)
2958 return -ENOSYS;
2961 int sysctl_ms_jiffies(struct ctl_table *table, int __user *name, int nlen,
2962 void __user *oldval, size_t __user *oldlenp,
2963 void __user *newval, size_t newlen)
2965 return -ENOSYS;
2968 #endif /* CONFIG_SYSCTL_SYSCALL */
2970 static int deprecated_sysctl_warning(struct __sysctl_args *args)
2972 static int msg_count;
2973 int name[CTL_MAXNAME];
2974 int i;
2976 /* Check args->nlen. */
2977 if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
2978 return -ENOTDIR;
2980 /* Read in the sysctl name for better debug message logging */
2981 for (i = 0; i < args->nlen; i++)
2982 if (get_user(name[i], args->name + i))
2983 return -EFAULT;
2985 /* Ignore accesses to kernel.version */
2986 if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2987 return 0;
2989 if (msg_count < 5) {
2990 msg_count++;
2991 printk(KERN_INFO
2992 "warning: process `%s' used the deprecated sysctl "
2993 "system call with ", current->comm);
2994 for (i = 0; i < args->nlen; i++)
2995 printk("%d.", name[i]);
2996 printk("\n");
2998 return 0;
3002 * No sense putting this after each symbol definition, twice,
3003 * exception granted :-)
3005 EXPORT_SYMBOL(proc_dointvec);
3006 EXPORT_SYMBOL(proc_dointvec_jiffies);
3007 EXPORT_SYMBOL(proc_dointvec_minmax);
3008 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3009 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3010 EXPORT_SYMBOL(proc_dostring);
3011 EXPORT_SYMBOL(proc_doulongvec_minmax);
3012 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3013 EXPORT_SYMBOL(register_sysctl_table);
3014 EXPORT_SYMBOL(register_sysctl_paths);
3015 EXPORT_SYMBOL(sysctl_intvec);
3016 EXPORT_SYMBOL(sysctl_jiffies);
3017 EXPORT_SYMBOL(sysctl_ms_jiffies);
3018 EXPORT_SYMBOL(sysctl_string);
3019 EXPORT_SYMBOL(sysctl_data);
3020 EXPORT_SYMBOL(unregister_sysctl_table);