ftrace: add trace_function api for other tracers to use
[linux-2.6/linux-2.6-openrd.git] / kernel / sysctl.c
blobefaf7c5500e97102c2cc90a68fb8b05997dc6e56
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/nfs_fs.h>
47 #include <linux/acpi.h>
48 #include <linux/reboot.h>
49 #include <linux/ftrace.h>
51 #include <asm/uaccess.h>
52 #include <asm/processor.h>
54 #ifdef CONFIG_X86
55 #include <asm/nmi.h>
56 #include <asm/stacktrace.h>
57 #include <asm/io.h>
58 #endif
60 static int deprecated_sysctl_warning(struct __sysctl_args *args);
62 #if defined(CONFIG_SYSCTL)
64 /* External variables not in a header file. */
65 extern int C_A_D;
66 extern int print_fatal_signals;
67 extern int sysctl_overcommit_memory;
68 extern int sysctl_overcommit_ratio;
69 extern int sysctl_panic_on_oom;
70 extern int sysctl_oom_kill_allocating_task;
71 extern int sysctl_oom_dump_tasks;
72 extern int max_threads;
73 extern int core_uses_pid;
74 extern int suid_dumpable;
75 extern char core_pattern[];
76 extern int pid_max;
77 extern int min_free_kbytes;
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 extern int sysctl_nr_open_min, sysctl_nr_open_max;
87 /* Constants used for minimum and maximum */
88 #if defined(CONFIG_DETECT_SOFTLOCKUP) || defined(CONFIG_HIGHMEM)
89 static int one = 1;
90 #endif
92 #ifdef CONFIG_DETECT_SOFTLOCKUP
93 static int sixty = 60;
94 #endif
96 #ifdef CONFIG_MMU
97 static int two = 2;
98 #endif
100 static int zero;
101 static int one_hundred = 100;
103 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
104 static int maxolduid = 65535;
105 static int minolduid;
106 static int min_percpu_pagelist_fract = 8;
108 static int ngroups_max = NGROUPS_MAX;
110 #ifdef CONFIG_KMOD
111 extern char modprobe_path[];
112 #endif
113 #ifdef CONFIG_CHR_DEV_SG
114 extern int sg_big_buff;
115 #endif
117 #ifdef __sparc__
118 extern char reboot_command [];
119 extern int stop_a_enabled;
120 extern int scons_pwroff;
121 #endif
123 #ifdef __hppa__
124 extern int pwrsw_enabled;
125 extern int unaligned_enabled;
126 #endif
128 #ifdef CONFIG_S390
129 #ifdef CONFIG_MATHEMU
130 extern int sysctl_ieee_emulation_warnings;
131 #endif
132 extern int sysctl_userprocess_debug;
133 extern int spin_retry;
134 #endif
136 extern int sysctl_hz_timer;
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 .ctl_table = root_table,
161 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.header_list),
162 .root = &sysctl_table_root,
164 static struct ctl_table_root sysctl_table_root = {
165 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
166 .header_list = LIST_HEAD_INIT(root_table_header.ctl_entry),
169 static struct ctl_table kern_table[];
170 static struct ctl_table vm_table[];
171 static struct ctl_table fs_table[];
172 static struct ctl_table debug_table[];
173 static struct ctl_table dev_table[];
174 extern struct ctl_table random_table[];
175 #ifdef CONFIG_INOTIFY_USER
176 extern struct ctl_table inotify_table[];
177 #endif
179 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
180 int sysctl_legacy_va_layout;
181 #endif
183 extern int prove_locking;
184 extern int lock_stat;
186 /* The default sysctl tables: */
188 static struct ctl_table root_table[] = {
190 .ctl_name = CTL_KERN,
191 .procname = "kernel",
192 .mode = 0555,
193 .child = kern_table,
196 .ctl_name = CTL_VM,
197 .procname = "vm",
198 .mode = 0555,
199 .child = vm_table,
202 .ctl_name = CTL_FS,
203 .procname = "fs",
204 .mode = 0555,
205 .child = fs_table,
208 .ctl_name = CTL_DEBUG,
209 .procname = "debug",
210 .mode = 0555,
211 .child = debug_table,
214 .ctl_name = CTL_DEV,
215 .procname = "dev",
216 .mode = 0555,
217 .child = dev_table,
220 * NOTE: do not add new entries to this table unless you have read
221 * Documentation/sysctl/ctl_unnumbered.txt
223 { .ctl_name = 0 }
226 #ifdef CONFIG_SCHED_DEBUG
227 static int min_sched_granularity_ns = 100000; /* 100 usecs */
228 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
229 static int min_wakeup_granularity_ns; /* 0 usecs */
230 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
231 #endif
233 static struct ctl_table kern_table[] = {
234 #ifdef CONFIG_SCHED_DEBUG
236 .ctl_name = CTL_UNNUMBERED,
237 .procname = "sched_min_granularity_ns",
238 .data = &sysctl_sched_min_granularity,
239 .maxlen = sizeof(unsigned int),
240 .mode = 0644,
241 .proc_handler = &sched_nr_latency_handler,
242 .strategy = &sysctl_intvec,
243 .extra1 = &min_sched_granularity_ns,
244 .extra2 = &max_sched_granularity_ns,
247 .ctl_name = CTL_UNNUMBERED,
248 .procname = "sched_latency_ns",
249 .data = &sysctl_sched_latency,
250 .maxlen = sizeof(unsigned int),
251 .mode = 0644,
252 .proc_handler = &sched_nr_latency_handler,
253 .strategy = &sysctl_intvec,
254 .extra1 = &min_sched_granularity_ns,
255 .extra2 = &max_sched_granularity_ns,
258 .ctl_name = CTL_UNNUMBERED,
259 .procname = "sched_wakeup_granularity_ns",
260 .data = &sysctl_sched_wakeup_granularity,
261 .maxlen = sizeof(unsigned int),
262 .mode = 0644,
263 .proc_handler = &proc_dointvec_minmax,
264 .strategy = &sysctl_intvec,
265 .extra1 = &min_wakeup_granularity_ns,
266 .extra2 = &max_wakeup_granularity_ns,
269 .ctl_name = CTL_UNNUMBERED,
270 .procname = "sched_child_runs_first",
271 .data = &sysctl_sched_child_runs_first,
272 .maxlen = sizeof(unsigned int),
273 .mode = 0644,
274 .proc_handler = &proc_dointvec,
277 .ctl_name = CTL_UNNUMBERED,
278 .procname = "sched_features",
279 .data = &sysctl_sched_features,
280 .maxlen = sizeof(unsigned int),
281 .mode = 0644,
282 .proc_handler = &proc_dointvec,
285 .ctl_name = CTL_UNNUMBERED,
286 .procname = "sched_migration_cost",
287 .data = &sysctl_sched_migration_cost,
288 .maxlen = sizeof(unsigned int),
289 .mode = 0644,
290 .proc_handler = &proc_dointvec,
293 .ctl_name = CTL_UNNUMBERED,
294 .procname = "sched_nr_migrate",
295 .data = &sysctl_sched_nr_migrate,
296 .maxlen = sizeof(unsigned int),
297 .mode = 0644,
298 .proc_handler = &proc_dointvec,
300 #endif
302 .ctl_name = CTL_UNNUMBERED,
303 .procname = "sched_rt_period_us",
304 .data = &sysctl_sched_rt_period,
305 .maxlen = sizeof(unsigned int),
306 .mode = 0644,
307 .proc_handler = &sched_rt_handler,
310 .ctl_name = CTL_UNNUMBERED,
311 .procname = "sched_rt_runtime_us",
312 .data = &sysctl_sched_rt_runtime,
313 .maxlen = sizeof(int),
314 .mode = 0644,
315 .proc_handler = &sched_rt_handler,
318 .ctl_name = CTL_UNNUMBERED,
319 .procname = "sched_compat_yield",
320 .data = &sysctl_sched_compat_yield,
321 .maxlen = sizeof(unsigned int),
322 .mode = 0644,
323 .proc_handler = &proc_dointvec,
325 #ifdef CONFIG_PROVE_LOCKING
327 .ctl_name = CTL_UNNUMBERED,
328 .procname = "prove_locking",
329 .data = &prove_locking,
330 .maxlen = sizeof(int),
331 .mode = 0644,
332 .proc_handler = &proc_dointvec,
334 #endif
335 #ifdef CONFIG_LOCK_STAT
337 .ctl_name = CTL_UNNUMBERED,
338 .procname = "lock_stat",
339 .data = &lock_stat,
340 .maxlen = sizeof(int),
341 .mode = 0644,
342 .proc_handler = &proc_dointvec,
344 #endif
346 .ctl_name = KERN_PANIC,
347 .procname = "panic",
348 .data = &panic_timeout,
349 .maxlen = sizeof(int),
350 .mode = 0644,
351 .proc_handler = &proc_dointvec,
354 .ctl_name = KERN_CORE_USES_PID,
355 .procname = "core_uses_pid",
356 .data = &core_uses_pid,
357 .maxlen = sizeof(int),
358 .mode = 0644,
359 .proc_handler = &proc_dointvec,
362 .ctl_name = KERN_CORE_PATTERN,
363 .procname = "core_pattern",
364 .data = core_pattern,
365 .maxlen = CORENAME_MAX_SIZE,
366 .mode = 0644,
367 .proc_handler = &proc_dostring,
368 .strategy = &sysctl_string,
370 #ifdef CONFIG_PROC_SYSCTL
372 .procname = "tainted",
373 .data = &tainted,
374 .maxlen = sizeof(int),
375 .mode = 0644,
376 .proc_handler = &proc_dointvec_taint,
378 #endif
379 #ifdef CONFIG_LATENCYTOP
381 .procname = "latencytop",
382 .data = &latencytop_enabled,
383 .maxlen = sizeof(int),
384 .mode = 0644,
385 .proc_handler = &proc_dointvec,
387 #endif
388 #ifdef CONFIG_BLK_DEV_INITRD
390 .ctl_name = KERN_REALROOTDEV,
391 .procname = "real-root-dev",
392 .data = &real_root_dev,
393 .maxlen = sizeof(int),
394 .mode = 0644,
395 .proc_handler = &proc_dointvec,
397 #endif
399 .ctl_name = CTL_UNNUMBERED,
400 .procname = "print-fatal-signals",
401 .data = &print_fatal_signals,
402 .maxlen = sizeof(int),
403 .mode = 0644,
404 .proc_handler = &proc_dointvec,
406 #ifdef __sparc__
408 .ctl_name = KERN_SPARC_REBOOT,
409 .procname = "reboot-cmd",
410 .data = reboot_command,
411 .maxlen = 256,
412 .mode = 0644,
413 .proc_handler = &proc_dostring,
414 .strategy = &sysctl_string,
417 .ctl_name = KERN_SPARC_STOP_A,
418 .procname = "stop-a",
419 .data = &stop_a_enabled,
420 .maxlen = sizeof (int),
421 .mode = 0644,
422 .proc_handler = &proc_dointvec,
425 .ctl_name = KERN_SPARC_SCONS_PWROFF,
426 .procname = "scons-poweroff",
427 .data = &scons_pwroff,
428 .maxlen = sizeof (int),
429 .mode = 0644,
430 .proc_handler = &proc_dointvec,
432 #endif
433 #ifdef __hppa__
435 .ctl_name = KERN_HPPA_PWRSW,
436 .procname = "soft-power",
437 .data = &pwrsw_enabled,
438 .maxlen = sizeof (int),
439 .mode = 0644,
440 .proc_handler = &proc_dointvec,
443 .ctl_name = KERN_HPPA_UNALIGNED,
444 .procname = "unaligned-trap",
445 .data = &unaligned_enabled,
446 .maxlen = sizeof (int),
447 .mode = 0644,
448 .proc_handler = &proc_dointvec,
450 #endif
452 .ctl_name = KERN_CTLALTDEL,
453 .procname = "ctrl-alt-del",
454 .data = &C_A_D,
455 .maxlen = sizeof(int),
456 .mode = 0644,
457 .proc_handler = &proc_dointvec,
459 #ifdef CONFIG_FTRACE
461 .ctl_name = CTL_UNNUMBERED,
462 .procname = "ftrace_enabled",
463 .data = &ftrace_enabled,
464 .maxlen = sizeof(int),
465 .mode = 0644,
466 .proc_handler = &ftrace_enable_sysctl,
468 #endif
469 #ifdef CONFIG_KMOD
471 .ctl_name = KERN_MODPROBE,
472 .procname = "modprobe",
473 .data = &modprobe_path,
474 .maxlen = KMOD_PATH_LEN,
475 .mode = 0644,
476 .proc_handler = &proc_dostring,
477 .strategy = &sysctl_string,
479 #endif
480 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
482 .ctl_name = KERN_HOTPLUG,
483 .procname = "hotplug",
484 .data = &uevent_helper,
485 .maxlen = UEVENT_HELPER_PATH_LEN,
486 .mode = 0644,
487 .proc_handler = &proc_dostring,
488 .strategy = &sysctl_string,
490 #endif
491 #ifdef CONFIG_CHR_DEV_SG
493 .ctl_name = KERN_SG_BIG_BUFF,
494 .procname = "sg-big-buff",
495 .data = &sg_big_buff,
496 .maxlen = sizeof (int),
497 .mode = 0444,
498 .proc_handler = &proc_dointvec,
500 #endif
501 #ifdef CONFIG_BSD_PROCESS_ACCT
503 .ctl_name = KERN_ACCT,
504 .procname = "acct",
505 .data = &acct_parm,
506 .maxlen = 3*sizeof(int),
507 .mode = 0644,
508 .proc_handler = &proc_dointvec,
510 #endif
511 #ifdef CONFIG_MAGIC_SYSRQ
513 .ctl_name = KERN_SYSRQ,
514 .procname = "sysrq",
515 .data = &__sysrq_enabled,
516 .maxlen = sizeof (int),
517 .mode = 0644,
518 .proc_handler = &proc_dointvec,
520 #endif
521 #ifdef CONFIG_PROC_SYSCTL
523 .procname = "cad_pid",
524 .data = NULL,
525 .maxlen = sizeof (int),
526 .mode = 0600,
527 .proc_handler = &proc_do_cad_pid,
529 #endif
531 .ctl_name = KERN_MAX_THREADS,
532 .procname = "threads-max",
533 .data = &max_threads,
534 .maxlen = sizeof(int),
535 .mode = 0644,
536 .proc_handler = &proc_dointvec,
539 .ctl_name = KERN_RANDOM,
540 .procname = "random",
541 .mode = 0555,
542 .child = random_table,
545 .ctl_name = KERN_OVERFLOWUID,
546 .procname = "overflowuid",
547 .data = &overflowuid,
548 .maxlen = sizeof(int),
549 .mode = 0644,
550 .proc_handler = &proc_dointvec_minmax,
551 .strategy = &sysctl_intvec,
552 .extra1 = &minolduid,
553 .extra2 = &maxolduid,
556 .ctl_name = KERN_OVERFLOWGID,
557 .procname = "overflowgid",
558 .data = &overflowgid,
559 .maxlen = sizeof(int),
560 .mode = 0644,
561 .proc_handler = &proc_dointvec_minmax,
562 .strategy = &sysctl_intvec,
563 .extra1 = &minolduid,
564 .extra2 = &maxolduid,
566 #ifdef CONFIG_S390
567 #ifdef CONFIG_MATHEMU
569 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
570 .procname = "ieee_emulation_warnings",
571 .data = &sysctl_ieee_emulation_warnings,
572 .maxlen = sizeof(int),
573 .mode = 0644,
574 .proc_handler = &proc_dointvec,
576 #endif
577 #ifdef CONFIG_NO_IDLE_HZ
579 .ctl_name = KERN_HZ_TIMER,
580 .procname = "hz_timer",
581 .data = &sysctl_hz_timer,
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_jiffies,
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_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_thresh",
744 .data = &softlockup_thresh,
745 .maxlen = sizeof(unsigned long),
746 .mode = 0644,
747 .proc_handler = &proc_doulongvec_minmax,
748 .strategy = &sysctl_intvec,
749 .extra1 = &one,
750 .extra2 = &sixty,
753 .ctl_name = CTL_UNNUMBERED,
754 .procname = "hung_task_check_count",
755 .data = &sysctl_hung_task_check_count,
756 .maxlen = sizeof(unsigned long),
757 .mode = 0644,
758 .proc_handler = &proc_doulongvec_minmax,
759 .strategy = &sysctl_intvec,
762 .ctl_name = CTL_UNNUMBERED,
763 .procname = "hung_task_timeout_secs",
764 .data = &sysctl_hung_task_timeout_secs,
765 .maxlen = sizeof(unsigned long),
766 .mode = 0644,
767 .proc_handler = &proc_doulongvec_minmax,
768 .strategy = &sysctl_intvec,
771 .ctl_name = CTL_UNNUMBERED,
772 .procname = "hung_task_warnings",
773 .data = &sysctl_hung_task_warnings,
774 .maxlen = sizeof(unsigned long),
775 .mode = 0644,
776 .proc_handler = &proc_doulongvec_minmax,
777 .strategy = &sysctl_intvec,
779 #endif
780 #ifdef CONFIG_COMPAT
782 .ctl_name = KERN_COMPAT_LOG,
783 .procname = "compat-log",
784 .data = &compat_log,
785 .maxlen = sizeof (int),
786 .mode = 0644,
787 .proc_handler = &proc_dointvec,
789 #endif
790 #ifdef CONFIG_RT_MUTEXES
792 .ctl_name = KERN_MAX_LOCK_DEPTH,
793 .procname = "max_lock_depth",
794 .data = &max_lock_depth,
795 .maxlen = sizeof(int),
796 .mode = 0644,
797 .proc_handler = &proc_dointvec,
799 #endif
800 #ifdef CONFIG_PROC_FS
802 .ctl_name = CTL_UNNUMBERED,
803 .procname = "maps_protect",
804 .data = &maps_protect,
805 .maxlen = sizeof(int),
806 .mode = 0644,
807 .proc_handler = &proc_dointvec,
809 #endif
811 .ctl_name = CTL_UNNUMBERED,
812 .procname = "poweroff_cmd",
813 .data = &poweroff_cmd,
814 .maxlen = POWEROFF_CMD_PATH_LEN,
815 .mode = 0644,
816 .proc_handler = &proc_dostring,
817 .strategy = &sysctl_string,
819 #ifdef CONFIG_KEYS
821 .ctl_name = CTL_UNNUMBERED,
822 .procname = "keys",
823 .mode = 0555,
824 .child = key_sysctls,
826 #endif
828 * NOTE: do not add new entries to this table unless you have read
829 * Documentation/sysctl/ctl_unnumbered.txt
831 { .ctl_name = 0 }
834 static struct ctl_table vm_table[] = {
836 .ctl_name = VM_OVERCOMMIT_MEMORY,
837 .procname = "overcommit_memory",
838 .data = &sysctl_overcommit_memory,
839 .maxlen = sizeof(sysctl_overcommit_memory),
840 .mode = 0644,
841 .proc_handler = &proc_dointvec,
844 .ctl_name = VM_PANIC_ON_OOM,
845 .procname = "panic_on_oom",
846 .data = &sysctl_panic_on_oom,
847 .maxlen = sizeof(sysctl_panic_on_oom),
848 .mode = 0644,
849 .proc_handler = &proc_dointvec,
852 .ctl_name = CTL_UNNUMBERED,
853 .procname = "oom_kill_allocating_task",
854 .data = &sysctl_oom_kill_allocating_task,
855 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
856 .mode = 0644,
857 .proc_handler = &proc_dointvec,
860 .ctl_name = CTL_UNNUMBERED,
861 .procname = "oom_dump_tasks",
862 .data = &sysctl_oom_dump_tasks,
863 .maxlen = sizeof(sysctl_oom_dump_tasks),
864 .mode = 0644,
865 .proc_handler = &proc_dointvec,
868 .ctl_name = VM_OVERCOMMIT_RATIO,
869 .procname = "overcommit_ratio",
870 .data = &sysctl_overcommit_ratio,
871 .maxlen = sizeof(sysctl_overcommit_ratio),
872 .mode = 0644,
873 .proc_handler = &proc_dointvec,
876 .ctl_name = VM_PAGE_CLUSTER,
877 .procname = "page-cluster",
878 .data = &page_cluster,
879 .maxlen = sizeof(int),
880 .mode = 0644,
881 .proc_handler = &proc_dointvec,
884 .ctl_name = VM_DIRTY_BACKGROUND,
885 .procname = "dirty_background_ratio",
886 .data = &dirty_background_ratio,
887 .maxlen = sizeof(dirty_background_ratio),
888 .mode = 0644,
889 .proc_handler = &proc_dointvec_minmax,
890 .strategy = &sysctl_intvec,
891 .extra1 = &zero,
892 .extra2 = &one_hundred,
895 .ctl_name = VM_DIRTY_RATIO,
896 .procname = "dirty_ratio",
897 .data = &vm_dirty_ratio,
898 .maxlen = sizeof(vm_dirty_ratio),
899 .mode = 0644,
900 .proc_handler = &dirty_ratio_handler,
901 .strategy = &sysctl_intvec,
902 .extra1 = &zero,
903 .extra2 = &one_hundred,
906 .procname = "dirty_writeback_centisecs",
907 .data = &dirty_writeback_interval,
908 .maxlen = sizeof(dirty_writeback_interval),
909 .mode = 0644,
910 .proc_handler = &dirty_writeback_centisecs_handler,
913 .procname = "dirty_expire_centisecs",
914 .data = &dirty_expire_interval,
915 .maxlen = sizeof(dirty_expire_interval),
916 .mode = 0644,
917 .proc_handler = &proc_dointvec_userhz_jiffies,
920 .ctl_name = VM_NR_PDFLUSH_THREADS,
921 .procname = "nr_pdflush_threads",
922 .data = &nr_pdflush_threads,
923 .maxlen = sizeof nr_pdflush_threads,
924 .mode = 0444 /* read-only*/,
925 .proc_handler = &proc_dointvec,
928 .ctl_name = VM_SWAPPINESS,
929 .procname = "swappiness",
930 .data = &vm_swappiness,
931 .maxlen = sizeof(vm_swappiness),
932 .mode = 0644,
933 .proc_handler = &proc_dointvec_minmax,
934 .strategy = &sysctl_intvec,
935 .extra1 = &zero,
936 .extra2 = &one_hundred,
938 #ifdef CONFIG_HUGETLB_PAGE
940 .procname = "nr_hugepages",
941 .data = &max_huge_pages,
942 .maxlen = sizeof(unsigned long),
943 .mode = 0644,
944 .proc_handler = &hugetlb_sysctl_handler,
945 .extra1 = (void *)&hugetlb_zero,
946 .extra2 = (void *)&hugetlb_infinity,
949 .ctl_name = VM_HUGETLB_GROUP,
950 .procname = "hugetlb_shm_group",
951 .data = &sysctl_hugetlb_shm_group,
952 .maxlen = sizeof(gid_t),
953 .mode = 0644,
954 .proc_handler = &proc_dointvec,
957 .ctl_name = CTL_UNNUMBERED,
958 .procname = "hugepages_treat_as_movable",
959 .data = &hugepages_treat_as_movable,
960 .maxlen = sizeof(int),
961 .mode = 0644,
962 .proc_handler = &hugetlb_treat_movable_handler,
965 .ctl_name = CTL_UNNUMBERED,
966 .procname = "nr_overcommit_hugepages",
967 .data = &sysctl_overcommit_huge_pages,
968 .maxlen = sizeof(sysctl_overcommit_huge_pages),
969 .mode = 0644,
970 .proc_handler = &hugetlb_overcommit_handler,
972 #endif
974 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
975 .procname = "lowmem_reserve_ratio",
976 .data = &sysctl_lowmem_reserve_ratio,
977 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
978 .mode = 0644,
979 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
980 .strategy = &sysctl_intvec,
983 .ctl_name = VM_DROP_PAGECACHE,
984 .procname = "drop_caches",
985 .data = &sysctl_drop_caches,
986 .maxlen = sizeof(int),
987 .mode = 0644,
988 .proc_handler = drop_caches_sysctl_handler,
989 .strategy = &sysctl_intvec,
992 .ctl_name = VM_MIN_FREE_KBYTES,
993 .procname = "min_free_kbytes",
994 .data = &min_free_kbytes,
995 .maxlen = sizeof(min_free_kbytes),
996 .mode = 0644,
997 .proc_handler = &min_free_kbytes_sysctl_handler,
998 .strategy = &sysctl_intvec,
999 .extra1 = &zero,
1002 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
1003 .procname = "percpu_pagelist_fraction",
1004 .data = &percpu_pagelist_fraction,
1005 .maxlen = sizeof(percpu_pagelist_fraction),
1006 .mode = 0644,
1007 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
1008 .strategy = &sysctl_intvec,
1009 .extra1 = &min_percpu_pagelist_fract,
1011 #ifdef CONFIG_MMU
1013 .ctl_name = VM_MAX_MAP_COUNT,
1014 .procname = "max_map_count",
1015 .data = &sysctl_max_map_count,
1016 .maxlen = sizeof(sysctl_max_map_count),
1017 .mode = 0644,
1018 .proc_handler = &proc_dointvec
1020 #endif
1022 .ctl_name = VM_LAPTOP_MODE,
1023 .procname = "laptop_mode",
1024 .data = &laptop_mode,
1025 .maxlen = sizeof(laptop_mode),
1026 .mode = 0644,
1027 .proc_handler = &proc_dointvec_jiffies,
1028 .strategy = &sysctl_jiffies,
1031 .ctl_name = VM_BLOCK_DUMP,
1032 .procname = "block_dump",
1033 .data = &block_dump,
1034 .maxlen = sizeof(block_dump),
1035 .mode = 0644,
1036 .proc_handler = &proc_dointvec,
1037 .strategy = &sysctl_intvec,
1038 .extra1 = &zero,
1041 .ctl_name = VM_VFS_CACHE_PRESSURE,
1042 .procname = "vfs_cache_pressure",
1043 .data = &sysctl_vfs_cache_pressure,
1044 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1045 .mode = 0644,
1046 .proc_handler = &proc_dointvec,
1047 .strategy = &sysctl_intvec,
1048 .extra1 = &zero,
1050 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1052 .ctl_name = VM_LEGACY_VA_LAYOUT,
1053 .procname = "legacy_va_layout",
1054 .data = &sysctl_legacy_va_layout,
1055 .maxlen = sizeof(sysctl_legacy_va_layout),
1056 .mode = 0644,
1057 .proc_handler = &proc_dointvec,
1058 .strategy = &sysctl_intvec,
1059 .extra1 = &zero,
1061 #endif
1062 #ifdef CONFIG_NUMA
1064 .ctl_name = VM_ZONE_RECLAIM_MODE,
1065 .procname = "zone_reclaim_mode",
1066 .data = &zone_reclaim_mode,
1067 .maxlen = sizeof(zone_reclaim_mode),
1068 .mode = 0644,
1069 .proc_handler = &proc_dointvec,
1070 .strategy = &sysctl_intvec,
1071 .extra1 = &zero,
1074 .ctl_name = VM_MIN_UNMAPPED,
1075 .procname = "min_unmapped_ratio",
1076 .data = &sysctl_min_unmapped_ratio,
1077 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1078 .mode = 0644,
1079 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
1080 .strategy = &sysctl_intvec,
1081 .extra1 = &zero,
1082 .extra2 = &one_hundred,
1085 .ctl_name = VM_MIN_SLAB,
1086 .procname = "min_slab_ratio",
1087 .data = &sysctl_min_slab_ratio,
1088 .maxlen = sizeof(sysctl_min_slab_ratio),
1089 .mode = 0644,
1090 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1091 .strategy = &sysctl_intvec,
1092 .extra1 = &zero,
1093 .extra2 = &one_hundred,
1095 #endif
1096 #ifdef CONFIG_SMP
1098 .ctl_name = CTL_UNNUMBERED,
1099 .procname = "stat_interval",
1100 .data = &sysctl_stat_interval,
1101 .maxlen = sizeof(sysctl_stat_interval),
1102 .mode = 0644,
1103 .proc_handler = &proc_dointvec_jiffies,
1104 .strategy = &sysctl_jiffies,
1106 #endif
1107 #ifdef CONFIG_SECURITY
1109 .ctl_name = CTL_UNNUMBERED,
1110 .procname = "mmap_min_addr",
1111 .data = &mmap_min_addr,
1112 .maxlen = sizeof(unsigned long),
1113 .mode = 0644,
1114 .proc_handler = &proc_doulongvec_minmax,
1116 #endif
1117 #ifdef CONFIG_NUMA
1119 .ctl_name = CTL_UNNUMBERED,
1120 .procname = "numa_zonelist_order",
1121 .data = &numa_zonelist_order,
1122 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1123 .mode = 0644,
1124 .proc_handler = &numa_zonelist_order_handler,
1125 .strategy = &sysctl_string,
1127 #endif
1128 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1129 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1131 .ctl_name = VM_VDSO_ENABLED,
1132 .procname = "vdso_enabled",
1133 .data = &vdso_enabled,
1134 .maxlen = sizeof(vdso_enabled),
1135 .mode = 0644,
1136 .proc_handler = &proc_dointvec,
1137 .strategy = &sysctl_intvec,
1138 .extra1 = &zero,
1140 #endif
1141 #ifdef CONFIG_HIGHMEM
1143 .ctl_name = CTL_UNNUMBERED,
1144 .procname = "highmem_is_dirtyable",
1145 .data = &vm_highmem_is_dirtyable,
1146 .maxlen = sizeof(vm_highmem_is_dirtyable),
1147 .mode = 0644,
1148 .proc_handler = &proc_dointvec_minmax,
1149 .strategy = &sysctl_intvec,
1150 .extra1 = &zero,
1151 .extra2 = &one,
1153 #endif
1155 * NOTE: do not add new entries to this table unless you have read
1156 * Documentation/sysctl/ctl_unnumbered.txt
1158 { .ctl_name = 0 }
1161 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1162 static struct ctl_table binfmt_misc_table[] = {
1163 { .ctl_name = 0 }
1165 #endif
1167 static struct ctl_table fs_table[] = {
1169 .ctl_name = FS_NRINODE,
1170 .procname = "inode-nr",
1171 .data = &inodes_stat,
1172 .maxlen = 2*sizeof(int),
1173 .mode = 0444,
1174 .proc_handler = &proc_dointvec,
1177 .ctl_name = FS_STATINODE,
1178 .procname = "inode-state",
1179 .data = &inodes_stat,
1180 .maxlen = 7*sizeof(int),
1181 .mode = 0444,
1182 .proc_handler = &proc_dointvec,
1185 .procname = "file-nr",
1186 .data = &files_stat,
1187 .maxlen = 3*sizeof(int),
1188 .mode = 0444,
1189 .proc_handler = &proc_nr_files,
1192 .ctl_name = FS_MAXFILE,
1193 .procname = "file-max",
1194 .data = &files_stat.max_files,
1195 .maxlen = sizeof(int),
1196 .mode = 0644,
1197 .proc_handler = &proc_dointvec,
1200 .ctl_name = CTL_UNNUMBERED,
1201 .procname = "nr_open",
1202 .data = &sysctl_nr_open,
1203 .maxlen = sizeof(int),
1204 .mode = 0644,
1205 .proc_handler = &proc_dointvec_minmax,
1206 .extra1 = &sysctl_nr_open_min,
1207 .extra2 = &sysctl_nr_open_max,
1210 .ctl_name = FS_DENTRY,
1211 .procname = "dentry-state",
1212 .data = &dentry_stat,
1213 .maxlen = 6*sizeof(int),
1214 .mode = 0444,
1215 .proc_handler = &proc_dointvec,
1218 .ctl_name = FS_OVERFLOWUID,
1219 .procname = "overflowuid",
1220 .data = &fs_overflowuid,
1221 .maxlen = sizeof(int),
1222 .mode = 0644,
1223 .proc_handler = &proc_dointvec_minmax,
1224 .strategy = &sysctl_intvec,
1225 .extra1 = &minolduid,
1226 .extra2 = &maxolduid,
1229 .ctl_name = FS_OVERFLOWGID,
1230 .procname = "overflowgid",
1231 .data = &fs_overflowgid,
1232 .maxlen = sizeof(int),
1233 .mode = 0644,
1234 .proc_handler = &proc_dointvec_minmax,
1235 .strategy = &sysctl_intvec,
1236 .extra1 = &minolduid,
1237 .extra2 = &maxolduid,
1240 .ctl_name = FS_LEASES,
1241 .procname = "leases-enable",
1242 .data = &leases_enable,
1243 .maxlen = sizeof(int),
1244 .mode = 0644,
1245 .proc_handler = &proc_dointvec,
1247 #ifdef CONFIG_DNOTIFY
1249 .ctl_name = FS_DIR_NOTIFY,
1250 .procname = "dir-notify-enable",
1251 .data = &dir_notify_enable,
1252 .maxlen = sizeof(int),
1253 .mode = 0644,
1254 .proc_handler = &proc_dointvec,
1256 #endif
1257 #ifdef CONFIG_MMU
1259 .ctl_name = FS_LEASE_TIME,
1260 .procname = "lease-break-time",
1261 .data = &lease_break_time,
1262 .maxlen = sizeof(int),
1263 .mode = 0644,
1264 .proc_handler = &proc_dointvec_minmax,
1265 .strategy = &sysctl_intvec,
1266 .extra1 = &zero,
1267 .extra2 = &two,
1270 .procname = "aio-nr",
1271 .data = &aio_nr,
1272 .maxlen = sizeof(aio_nr),
1273 .mode = 0444,
1274 .proc_handler = &proc_doulongvec_minmax,
1277 .procname = "aio-max-nr",
1278 .data = &aio_max_nr,
1279 .maxlen = sizeof(aio_max_nr),
1280 .mode = 0644,
1281 .proc_handler = &proc_doulongvec_minmax,
1283 #ifdef CONFIG_INOTIFY_USER
1285 .ctl_name = FS_INOTIFY,
1286 .procname = "inotify",
1287 .mode = 0555,
1288 .child = inotify_table,
1290 #endif
1291 #endif
1293 .ctl_name = KERN_SETUID_DUMPABLE,
1294 .procname = "suid_dumpable",
1295 .data = &suid_dumpable,
1296 .maxlen = sizeof(int),
1297 .mode = 0644,
1298 .proc_handler = &proc_dointvec,
1300 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1302 .ctl_name = CTL_UNNUMBERED,
1303 .procname = "binfmt_misc",
1304 .mode = 0555,
1305 .child = binfmt_misc_table,
1307 #endif
1309 * NOTE: do not add new entries to this table unless you have read
1310 * Documentation/sysctl/ctl_unnumbered.txt
1312 { .ctl_name = 0 }
1315 static struct ctl_table debug_table[] = {
1316 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1318 .ctl_name = CTL_UNNUMBERED,
1319 .procname = "exception-trace",
1320 .data = &show_unhandled_signals,
1321 .maxlen = sizeof(int),
1322 .mode = 0644,
1323 .proc_handler = proc_dointvec
1325 #endif
1326 { .ctl_name = 0 }
1329 static struct ctl_table dev_table[] = {
1330 { .ctl_name = 0 }
1333 static DEFINE_SPINLOCK(sysctl_lock);
1335 /* called under sysctl_lock */
1336 static int use_table(struct ctl_table_header *p)
1338 if (unlikely(p->unregistering))
1339 return 0;
1340 p->used++;
1341 return 1;
1344 /* called under sysctl_lock */
1345 static void unuse_table(struct ctl_table_header *p)
1347 if (!--p->used)
1348 if (unlikely(p->unregistering))
1349 complete(p->unregistering);
1352 /* called under sysctl_lock, will reacquire if has to wait */
1353 static void start_unregistering(struct ctl_table_header *p)
1356 * if p->used is 0, nobody will ever touch that entry again;
1357 * we'll eliminate all paths to it before dropping sysctl_lock
1359 if (unlikely(p->used)) {
1360 struct completion wait;
1361 init_completion(&wait);
1362 p->unregistering = &wait;
1363 spin_unlock(&sysctl_lock);
1364 wait_for_completion(&wait);
1365 spin_lock(&sysctl_lock);
1368 * do not remove from the list until nobody holds it; walking the
1369 * list in do_sysctl() relies on that.
1371 list_del_init(&p->ctl_entry);
1374 void sysctl_head_finish(struct ctl_table_header *head)
1376 if (!head)
1377 return;
1378 spin_lock(&sysctl_lock);
1379 unuse_table(head);
1380 spin_unlock(&sysctl_lock);
1383 static struct list_head *
1384 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1386 struct list_head *header_list;
1387 header_list = &root->header_list;
1388 if (root->lookup)
1389 header_list = root->lookup(root, namespaces);
1390 return header_list;
1393 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1394 struct ctl_table_header *prev)
1396 struct ctl_table_root *root;
1397 struct list_head *header_list;
1398 struct ctl_table_header *head;
1399 struct list_head *tmp;
1401 spin_lock(&sysctl_lock);
1402 if (prev) {
1403 head = prev;
1404 tmp = &prev->ctl_entry;
1405 unuse_table(prev);
1406 goto next;
1408 tmp = &root_table_header.ctl_entry;
1409 for (;;) {
1410 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1412 if (!use_table(head))
1413 goto next;
1414 spin_unlock(&sysctl_lock);
1415 return head;
1416 next:
1417 root = head->root;
1418 tmp = tmp->next;
1419 header_list = lookup_header_list(root, namespaces);
1420 if (tmp != header_list)
1421 continue;
1423 do {
1424 root = list_entry(root->root_list.next,
1425 struct ctl_table_root, root_list);
1426 if (root == &sysctl_table_root)
1427 goto out;
1428 header_list = lookup_header_list(root, namespaces);
1429 } while (list_empty(header_list));
1430 tmp = header_list->next;
1432 out:
1433 spin_unlock(&sysctl_lock);
1434 return NULL;
1437 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1439 return __sysctl_head_next(current->nsproxy, prev);
1442 void register_sysctl_root(struct ctl_table_root *root)
1444 spin_lock(&sysctl_lock);
1445 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1446 spin_unlock(&sysctl_lock);
1449 #ifdef CONFIG_SYSCTL_SYSCALL
1450 /* Perform the actual read/write of a sysctl table entry. */
1451 static int do_sysctl_strategy(struct ctl_table_root *root,
1452 struct ctl_table *table,
1453 int __user *name, int nlen,
1454 void __user *oldval, size_t __user *oldlenp,
1455 void __user *newval, size_t newlen)
1457 int op = 0, rc;
1459 if (oldval)
1460 op |= 004;
1461 if (newval)
1462 op |= 002;
1463 if (sysctl_perm(root, table, op))
1464 return -EPERM;
1466 if (table->strategy) {
1467 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1468 newval, newlen);
1469 if (rc < 0)
1470 return rc;
1471 if (rc > 0)
1472 return 0;
1475 /* If there is no strategy routine, or if the strategy returns
1476 * zero, proceed with automatic r/w */
1477 if (table->data && table->maxlen) {
1478 rc = sysctl_data(table, name, nlen, oldval, oldlenp,
1479 newval, newlen);
1480 if (rc < 0)
1481 return rc;
1483 return 0;
1486 static int parse_table(int __user *name, int nlen,
1487 void __user *oldval, size_t __user *oldlenp,
1488 void __user *newval, size_t newlen,
1489 struct ctl_table_root *root,
1490 struct ctl_table *table)
1492 int n;
1493 repeat:
1494 if (!nlen)
1495 return -ENOTDIR;
1496 if (get_user(n, name))
1497 return -EFAULT;
1498 for ( ; table->ctl_name || table->procname; table++) {
1499 if (!table->ctl_name)
1500 continue;
1501 if (n == table->ctl_name) {
1502 int error;
1503 if (table->child) {
1504 if (sysctl_perm(root, table, 001))
1505 return -EPERM;
1506 name++;
1507 nlen--;
1508 table = table->child;
1509 goto repeat;
1511 error = do_sysctl_strategy(root, table, name, nlen,
1512 oldval, oldlenp,
1513 newval, newlen);
1514 return error;
1517 return -ENOTDIR;
1520 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1521 void __user *newval, size_t newlen)
1523 struct ctl_table_header *head;
1524 int error = -ENOTDIR;
1526 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1527 return -ENOTDIR;
1528 if (oldval) {
1529 int old_len;
1530 if (!oldlenp || get_user(old_len, oldlenp))
1531 return -EFAULT;
1534 for (head = sysctl_head_next(NULL); head;
1535 head = sysctl_head_next(head)) {
1536 error = parse_table(name, nlen, oldval, oldlenp,
1537 newval, newlen,
1538 head->root, head->ctl_table);
1539 if (error != -ENOTDIR) {
1540 sysctl_head_finish(head);
1541 break;
1544 return error;
1547 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1549 struct __sysctl_args tmp;
1550 int error;
1552 if (copy_from_user(&tmp, args, sizeof(tmp)))
1553 return -EFAULT;
1555 error = deprecated_sysctl_warning(&tmp);
1556 if (error)
1557 goto out;
1559 lock_kernel();
1560 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1561 tmp.newval, tmp.newlen);
1562 unlock_kernel();
1563 out:
1564 return error;
1566 #endif /* CONFIG_SYSCTL_SYSCALL */
1569 * sysctl_perm does NOT grant the superuser all rights automatically, because
1570 * some sysctl variables are readonly even to root.
1573 static int test_perm(int mode, int op)
1575 if (!current->euid)
1576 mode >>= 6;
1577 else if (in_egroup_p(0))
1578 mode >>= 3;
1579 if ((mode & op & 0007) == op)
1580 return 0;
1581 return -EACCES;
1584 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1586 int error;
1587 int mode;
1589 error = security_sysctl(table, op);
1590 if (error)
1591 return error;
1593 if (root->permissions)
1594 mode = root->permissions(root, current->nsproxy, table);
1595 else
1596 mode = table->mode;
1598 return test_perm(mode, op);
1601 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1603 for (; table->ctl_name || table->procname; table++) {
1604 table->parent = parent;
1605 if (table->child)
1606 sysctl_set_parent(table, table->child);
1610 static __init int sysctl_init(void)
1612 sysctl_set_parent(NULL, root_table);
1613 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1615 int err;
1616 err = sysctl_check_table(current->nsproxy, root_table);
1618 #endif
1619 return 0;
1622 core_initcall(sysctl_init);
1625 * __register_sysctl_paths - register a sysctl hierarchy
1626 * @root: List of sysctl headers to register on
1627 * @namespaces: Data to compute which lists of sysctl entries are visible
1628 * @path: The path to the directory the sysctl table is in.
1629 * @table: the top-level table structure
1631 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1632 * array. A completely 0 filled entry terminates the table.
1634 * The members of the &struct ctl_table structure are used as follows:
1636 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1637 * must be unique within that level of sysctl
1639 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1640 * enter a sysctl file
1642 * data - a pointer to data for use by proc_handler
1644 * maxlen - the maximum size in bytes of the data
1646 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1648 * child - a pointer to the child sysctl table if this entry is a directory, or
1649 * %NULL.
1651 * proc_handler - the text handler routine (described below)
1653 * strategy - the strategy routine (described below)
1655 * de - for internal use by the sysctl routines
1657 * extra1, extra2 - extra pointers usable by the proc handler routines
1659 * Leaf nodes in the sysctl tree will be represented by a single file
1660 * under /proc; non-leaf nodes will be represented by directories.
1662 * sysctl(2) can automatically manage read and write requests through
1663 * the sysctl table. The data and maxlen fields of the ctl_table
1664 * struct enable minimal validation of the values being written to be
1665 * performed, and the mode field allows minimal authentication.
1667 * More sophisticated management can be enabled by the provision of a
1668 * strategy routine with the table entry. This will be called before
1669 * any automatic read or write of the data is performed.
1671 * The strategy routine may return
1673 * < 0 - Error occurred (error is passed to user process)
1675 * 0 - OK - proceed with automatic read or write.
1677 * > 0 - OK - read or write has been done by the strategy routine, so
1678 * return immediately.
1680 * There must be a proc_handler routine for any terminal nodes
1681 * mirrored under /proc/sys (non-terminals are handled by a built-in
1682 * directory handler). Several default handlers are available to
1683 * cover common cases -
1685 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1686 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1687 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1689 * It is the handler's job to read the input buffer from user memory
1690 * and process it. The handler should return 0 on success.
1692 * This routine returns %NULL on a failure to register, and a pointer
1693 * to the table header on success.
1695 struct ctl_table_header *__register_sysctl_paths(
1696 struct ctl_table_root *root,
1697 struct nsproxy *namespaces,
1698 const struct ctl_path *path, struct ctl_table *table)
1700 struct list_head *header_list;
1701 struct ctl_table_header *header;
1702 struct ctl_table *new, **prevp;
1703 unsigned int n, npath;
1705 /* Count the path components */
1706 for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1710 * For each path component, allocate a 2-element ctl_table array.
1711 * The first array element will be filled with the sysctl entry
1712 * for this, the second will be the sentinel (ctl_name == 0).
1714 * We allocate everything in one go so that we don't have to
1715 * worry about freeing additional memory in unregister_sysctl_table.
1717 header = kzalloc(sizeof(struct ctl_table_header) +
1718 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1719 if (!header)
1720 return NULL;
1722 new = (struct ctl_table *) (header + 1);
1724 /* Now connect the dots */
1725 prevp = &header->ctl_table;
1726 for (n = 0; n < npath; ++n, ++path) {
1727 /* Copy the procname */
1728 new->procname = path->procname;
1729 new->ctl_name = path->ctl_name;
1730 new->mode = 0555;
1732 *prevp = new;
1733 prevp = &new->child;
1735 new += 2;
1737 *prevp = table;
1738 header->ctl_table_arg = table;
1740 INIT_LIST_HEAD(&header->ctl_entry);
1741 header->used = 0;
1742 header->unregistering = NULL;
1743 header->root = root;
1744 sysctl_set_parent(NULL, header->ctl_table);
1745 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1746 if (sysctl_check_table(namespaces, header->ctl_table)) {
1747 kfree(header);
1748 return NULL;
1750 #endif
1751 spin_lock(&sysctl_lock);
1752 header_list = lookup_header_list(root, namespaces);
1753 list_add_tail(&header->ctl_entry, header_list);
1754 spin_unlock(&sysctl_lock);
1756 return header;
1760 * register_sysctl_table_path - register a sysctl table hierarchy
1761 * @path: The path to the directory the sysctl table is in.
1762 * @table: the top-level table structure
1764 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1765 * array. A completely 0 filled entry terminates the table.
1767 * See __register_sysctl_paths for more details.
1769 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1770 struct ctl_table *table)
1772 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1773 path, table);
1777 * register_sysctl_table - register a sysctl table hierarchy
1778 * @table: the top-level table structure
1780 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1781 * array. A completely 0 filled entry terminates the table.
1783 * See register_sysctl_paths for more details.
1785 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1787 static const struct ctl_path null_path[] = { {} };
1789 return register_sysctl_paths(null_path, table);
1793 * unregister_sysctl_table - unregister a sysctl table hierarchy
1794 * @header: the header returned from register_sysctl_table
1796 * Unregisters the sysctl table and all children. proc entries may not
1797 * actually be removed until they are no longer used by anyone.
1799 void unregister_sysctl_table(struct ctl_table_header * header)
1801 might_sleep();
1803 if (header == NULL)
1804 return;
1806 spin_lock(&sysctl_lock);
1807 start_unregistering(header);
1808 spin_unlock(&sysctl_lock);
1809 kfree(header);
1812 #else /* !CONFIG_SYSCTL */
1813 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1815 return NULL;
1818 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1819 struct ctl_table *table)
1821 return NULL;
1824 void unregister_sysctl_table(struct ctl_table_header * table)
1828 #endif /* CONFIG_SYSCTL */
1831 * /proc/sys support
1834 #ifdef CONFIG_PROC_SYSCTL
1836 static int _proc_do_string(void* data, int maxlen, int write,
1837 struct file *filp, void __user *buffer,
1838 size_t *lenp, loff_t *ppos)
1840 size_t len;
1841 char __user *p;
1842 char c;
1844 if (!data || !maxlen || !*lenp) {
1845 *lenp = 0;
1846 return 0;
1849 if (write) {
1850 len = 0;
1851 p = buffer;
1852 while (len < *lenp) {
1853 if (get_user(c, p++))
1854 return -EFAULT;
1855 if (c == 0 || c == '\n')
1856 break;
1857 len++;
1859 if (len >= maxlen)
1860 len = maxlen-1;
1861 if(copy_from_user(data, buffer, len))
1862 return -EFAULT;
1863 ((char *) data)[len] = 0;
1864 *ppos += *lenp;
1865 } else {
1866 len = strlen(data);
1867 if (len > maxlen)
1868 len = maxlen;
1870 if (*ppos > len) {
1871 *lenp = 0;
1872 return 0;
1875 data += *ppos;
1876 len -= *ppos;
1878 if (len > *lenp)
1879 len = *lenp;
1880 if (len)
1881 if(copy_to_user(buffer, data, len))
1882 return -EFAULT;
1883 if (len < *lenp) {
1884 if(put_user('\n', ((char __user *) buffer) + len))
1885 return -EFAULT;
1886 len++;
1888 *lenp = len;
1889 *ppos += len;
1891 return 0;
1895 * proc_dostring - read a string sysctl
1896 * @table: the sysctl table
1897 * @write: %TRUE if this is a write to the sysctl file
1898 * @filp: the file structure
1899 * @buffer: the user buffer
1900 * @lenp: the size of the user buffer
1901 * @ppos: file position
1903 * Reads/writes a string from/to the user buffer. If the kernel
1904 * buffer provided is not large enough to hold the string, the
1905 * string is truncated. The copied string is %NULL-terminated.
1906 * If the string is being read by the user process, it is copied
1907 * and a newline '\n' is added. It is truncated if the buffer is
1908 * not large enough.
1910 * Returns 0 on success.
1912 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
1913 void __user *buffer, size_t *lenp, loff_t *ppos)
1915 return _proc_do_string(table->data, table->maxlen, write, filp,
1916 buffer, lenp, ppos);
1920 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1921 int *valp,
1922 int write, void *data)
1924 if (write) {
1925 *valp = *negp ? -*lvalp : *lvalp;
1926 } else {
1927 int val = *valp;
1928 if (val < 0) {
1929 *negp = -1;
1930 *lvalp = (unsigned long)-val;
1931 } else {
1932 *negp = 0;
1933 *lvalp = (unsigned long)val;
1936 return 0;
1939 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1940 int write, struct file *filp, void __user *buffer,
1941 size_t *lenp, loff_t *ppos,
1942 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1943 int write, void *data),
1944 void *data)
1946 #define TMPBUFLEN 21
1947 int *i, vleft, first=1, neg, val;
1948 unsigned long lval;
1949 size_t left, len;
1951 char buf[TMPBUFLEN], *p;
1952 char __user *s = buffer;
1954 if (!tbl_data || !table->maxlen || !*lenp ||
1955 (*ppos && !write)) {
1956 *lenp = 0;
1957 return 0;
1960 i = (int *) tbl_data;
1961 vleft = table->maxlen / sizeof(*i);
1962 left = *lenp;
1964 if (!conv)
1965 conv = do_proc_dointvec_conv;
1967 for (; left && vleft--; i++, first=0) {
1968 if (write) {
1969 while (left) {
1970 char c;
1971 if (get_user(c, s))
1972 return -EFAULT;
1973 if (!isspace(c))
1974 break;
1975 left--;
1976 s++;
1978 if (!left)
1979 break;
1980 neg = 0;
1981 len = left;
1982 if (len > sizeof(buf) - 1)
1983 len = sizeof(buf) - 1;
1984 if (copy_from_user(buf, s, len))
1985 return -EFAULT;
1986 buf[len] = 0;
1987 p = buf;
1988 if (*p == '-' && left > 1) {
1989 neg = 1;
1990 p++;
1992 if (*p < '0' || *p > '9')
1993 break;
1995 lval = simple_strtoul(p, &p, 0);
1997 len = p-buf;
1998 if ((len < left) && *p && !isspace(*p))
1999 break;
2000 if (neg)
2001 val = -val;
2002 s += len;
2003 left -= len;
2005 if (conv(&neg, &lval, i, 1, data))
2006 break;
2007 } else {
2008 p = buf;
2009 if (!first)
2010 *p++ = '\t';
2012 if (conv(&neg, &lval, i, 0, data))
2013 break;
2015 sprintf(p, "%s%lu", neg ? "-" : "", lval);
2016 len = strlen(buf);
2017 if (len > left)
2018 len = left;
2019 if(copy_to_user(s, buf, len))
2020 return -EFAULT;
2021 left -= len;
2022 s += len;
2026 if (!write && !first && left) {
2027 if(put_user('\n', s))
2028 return -EFAULT;
2029 left--, s++;
2031 if (write) {
2032 while (left) {
2033 char c;
2034 if (get_user(c, s++))
2035 return -EFAULT;
2036 if (!isspace(c))
2037 break;
2038 left--;
2041 if (write && first)
2042 return -EINVAL;
2043 *lenp -= left;
2044 *ppos += *lenp;
2045 return 0;
2046 #undef TMPBUFLEN
2049 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2050 void __user *buffer, size_t *lenp, loff_t *ppos,
2051 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2052 int write, void *data),
2053 void *data)
2055 return __do_proc_dointvec(table->data, table, write, filp,
2056 buffer, lenp, ppos, conv, data);
2060 * proc_dointvec - read a vector of integers
2061 * @table: the sysctl table
2062 * @write: %TRUE if this is a write to the sysctl file
2063 * @filp: the file structure
2064 * @buffer: the user buffer
2065 * @lenp: the size of the user buffer
2066 * @ppos: file position
2068 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2069 * values from/to the user buffer, treated as an ASCII string.
2071 * Returns 0 on success.
2073 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2074 void __user *buffer, size_t *lenp, loff_t *ppos)
2076 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2077 NULL,NULL);
2080 #define OP_SET 0
2081 #define OP_AND 1
2082 #define OP_OR 2
2084 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
2085 int *valp,
2086 int write, void *data)
2088 int op = *(int *)data;
2089 if (write) {
2090 int val = *negp ? -*lvalp : *lvalp;
2091 switch(op) {
2092 case OP_SET: *valp = val; break;
2093 case OP_AND: *valp &= val; break;
2094 case OP_OR: *valp |= val; break;
2096 } else {
2097 int val = *valp;
2098 if (val < 0) {
2099 *negp = -1;
2100 *lvalp = (unsigned long)-val;
2101 } else {
2102 *negp = 0;
2103 *lvalp = (unsigned long)val;
2106 return 0;
2110 * Taint values can only be increased
2112 static int proc_dointvec_taint(struct ctl_table *table, int write, struct file *filp,
2113 void __user *buffer, size_t *lenp, loff_t *ppos)
2115 int op;
2117 if (write && !capable(CAP_SYS_ADMIN))
2118 return -EPERM;
2120 op = OP_OR;
2121 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2122 do_proc_dointvec_bset_conv,&op);
2125 struct do_proc_dointvec_minmax_conv_param {
2126 int *min;
2127 int *max;
2130 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2131 int *valp,
2132 int write, void *data)
2134 struct do_proc_dointvec_minmax_conv_param *param = data;
2135 if (write) {
2136 int val = *negp ? -*lvalp : *lvalp;
2137 if ((param->min && *param->min > val) ||
2138 (param->max && *param->max < val))
2139 return -EINVAL;
2140 *valp = val;
2141 } else {
2142 int val = *valp;
2143 if (val < 0) {
2144 *negp = -1;
2145 *lvalp = (unsigned long)-val;
2146 } else {
2147 *negp = 0;
2148 *lvalp = (unsigned long)val;
2151 return 0;
2155 * proc_dointvec_minmax - read a vector of integers with min/max values
2156 * @table: the sysctl table
2157 * @write: %TRUE if this is a write to the sysctl file
2158 * @filp: the file structure
2159 * @buffer: the user buffer
2160 * @lenp: the size of the user buffer
2161 * @ppos: file position
2163 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2164 * values from/to the user buffer, treated as an ASCII string.
2166 * This routine will ensure the values are within the range specified by
2167 * table->extra1 (min) and table->extra2 (max).
2169 * Returns 0 on success.
2171 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2172 void __user *buffer, size_t *lenp, loff_t *ppos)
2174 struct do_proc_dointvec_minmax_conv_param param = {
2175 .min = (int *) table->extra1,
2176 .max = (int *) table->extra2,
2178 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2179 do_proc_dointvec_minmax_conv, &param);
2182 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2183 struct file *filp,
2184 void __user *buffer,
2185 size_t *lenp, loff_t *ppos,
2186 unsigned long convmul,
2187 unsigned long convdiv)
2189 #define TMPBUFLEN 21
2190 unsigned long *i, *min, *max, val;
2191 int vleft, first=1, neg;
2192 size_t len, left;
2193 char buf[TMPBUFLEN], *p;
2194 char __user *s = buffer;
2196 if (!data || !table->maxlen || !*lenp ||
2197 (*ppos && !write)) {
2198 *lenp = 0;
2199 return 0;
2202 i = (unsigned long *) data;
2203 min = (unsigned long *) table->extra1;
2204 max = (unsigned long *) table->extra2;
2205 vleft = table->maxlen / sizeof(unsigned long);
2206 left = *lenp;
2208 for (; left && vleft--; i++, min++, max++, first=0) {
2209 if (write) {
2210 while (left) {
2211 char c;
2212 if (get_user(c, s))
2213 return -EFAULT;
2214 if (!isspace(c))
2215 break;
2216 left--;
2217 s++;
2219 if (!left)
2220 break;
2221 neg = 0;
2222 len = left;
2223 if (len > TMPBUFLEN-1)
2224 len = TMPBUFLEN-1;
2225 if (copy_from_user(buf, s, len))
2226 return -EFAULT;
2227 buf[len] = 0;
2228 p = buf;
2229 if (*p == '-' && left > 1) {
2230 neg = 1;
2231 p++;
2233 if (*p < '0' || *p > '9')
2234 break;
2235 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2236 len = p-buf;
2237 if ((len < left) && *p && !isspace(*p))
2238 break;
2239 if (neg)
2240 val = -val;
2241 s += len;
2242 left -= len;
2244 if(neg)
2245 continue;
2246 if ((min && val < *min) || (max && val > *max))
2247 continue;
2248 *i = val;
2249 } else {
2250 p = buf;
2251 if (!first)
2252 *p++ = '\t';
2253 sprintf(p, "%lu", convdiv * (*i) / convmul);
2254 len = strlen(buf);
2255 if (len > left)
2256 len = left;
2257 if(copy_to_user(s, buf, len))
2258 return -EFAULT;
2259 left -= len;
2260 s += len;
2264 if (!write && !first && left) {
2265 if(put_user('\n', s))
2266 return -EFAULT;
2267 left--, s++;
2269 if (write) {
2270 while (left) {
2271 char c;
2272 if (get_user(c, s++))
2273 return -EFAULT;
2274 if (!isspace(c))
2275 break;
2276 left--;
2279 if (write && first)
2280 return -EINVAL;
2281 *lenp -= left;
2282 *ppos += *lenp;
2283 return 0;
2284 #undef TMPBUFLEN
2287 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2288 struct file *filp,
2289 void __user *buffer,
2290 size_t *lenp, loff_t *ppos,
2291 unsigned long convmul,
2292 unsigned long convdiv)
2294 return __do_proc_doulongvec_minmax(table->data, table, write,
2295 filp, buffer, lenp, ppos, convmul, convdiv);
2299 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2300 * @table: the sysctl table
2301 * @write: %TRUE if this is a write to the sysctl file
2302 * @filp: the file structure
2303 * @buffer: the user buffer
2304 * @lenp: the size of the user buffer
2305 * @ppos: file position
2307 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2308 * values from/to the user buffer, treated as an ASCII string.
2310 * This routine will ensure the values are within the range specified by
2311 * table->extra1 (min) and table->extra2 (max).
2313 * Returns 0 on success.
2315 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2316 void __user *buffer, size_t *lenp, loff_t *ppos)
2318 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2322 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2323 * @table: the sysctl table
2324 * @write: %TRUE if this is a write to the sysctl file
2325 * @filp: the file structure
2326 * @buffer: the user buffer
2327 * @lenp: the size of the user buffer
2328 * @ppos: file position
2330 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2331 * values from/to the user buffer, treated as an ASCII string. The values
2332 * are treated as milliseconds, and converted to jiffies when they are stored.
2334 * This routine will ensure the values are within the range specified by
2335 * table->extra1 (min) and table->extra2 (max).
2337 * Returns 0 on success.
2339 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2340 struct file *filp,
2341 void __user *buffer,
2342 size_t *lenp, loff_t *ppos)
2344 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2345 lenp, ppos, HZ, 1000l);
2349 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2350 int *valp,
2351 int write, void *data)
2353 if (write) {
2354 if (*lvalp > LONG_MAX / HZ)
2355 return 1;
2356 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2357 } else {
2358 int val = *valp;
2359 unsigned long lval;
2360 if (val < 0) {
2361 *negp = -1;
2362 lval = (unsigned long)-val;
2363 } else {
2364 *negp = 0;
2365 lval = (unsigned long)val;
2367 *lvalp = lval / HZ;
2369 return 0;
2372 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2373 int *valp,
2374 int write, void *data)
2376 if (write) {
2377 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2378 return 1;
2379 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2380 } else {
2381 int val = *valp;
2382 unsigned long lval;
2383 if (val < 0) {
2384 *negp = -1;
2385 lval = (unsigned long)-val;
2386 } else {
2387 *negp = 0;
2388 lval = (unsigned long)val;
2390 *lvalp = jiffies_to_clock_t(lval);
2392 return 0;
2395 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2396 int *valp,
2397 int write, void *data)
2399 if (write) {
2400 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2401 } else {
2402 int val = *valp;
2403 unsigned long lval;
2404 if (val < 0) {
2405 *negp = -1;
2406 lval = (unsigned long)-val;
2407 } else {
2408 *negp = 0;
2409 lval = (unsigned long)val;
2411 *lvalp = jiffies_to_msecs(lval);
2413 return 0;
2417 * proc_dointvec_jiffies - read a vector of integers as seconds
2418 * @table: the sysctl table
2419 * @write: %TRUE if this is a write to the sysctl file
2420 * @filp: the file structure
2421 * @buffer: the user buffer
2422 * @lenp: the size of the user buffer
2423 * @ppos: file position
2425 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2426 * values from/to the user buffer, treated as an ASCII string.
2427 * The values read are assumed to be in seconds, and are converted into
2428 * jiffies.
2430 * Returns 0 on success.
2432 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2433 void __user *buffer, size_t *lenp, loff_t *ppos)
2435 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2436 do_proc_dointvec_jiffies_conv,NULL);
2440 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2441 * @table: the sysctl table
2442 * @write: %TRUE if this is a write to the sysctl file
2443 * @filp: the file structure
2444 * @buffer: the user buffer
2445 * @lenp: the size of the user buffer
2446 * @ppos: pointer to the file position
2448 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2449 * values from/to the user buffer, treated as an ASCII string.
2450 * The values read are assumed to be in 1/USER_HZ seconds, and
2451 * are converted into jiffies.
2453 * Returns 0 on success.
2455 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2456 void __user *buffer, size_t *lenp, loff_t *ppos)
2458 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2459 do_proc_dointvec_userhz_jiffies_conv,NULL);
2463 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2464 * @table: the sysctl table
2465 * @write: %TRUE if this is a write to the sysctl file
2466 * @filp: the file structure
2467 * @buffer: the user buffer
2468 * @lenp: the size of the user buffer
2469 * @ppos: file position
2470 * @ppos: the current position in the file
2472 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2473 * values from/to the user buffer, treated as an ASCII string.
2474 * The values read are assumed to be in 1/1000 seconds, and
2475 * are converted into jiffies.
2477 * Returns 0 on success.
2479 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2480 void __user *buffer, size_t *lenp, loff_t *ppos)
2482 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2483 do_proc_dointvec_ms_jiffies_conv, NULL);
2486 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2487 void __user *buffer, size_t *lenp, loff_t *ppos)
2489 struct pid *new_pid;
2490 pid_t tmp;
2491 int r;
2493 tmp = pid_vnr(cad_pid);
2495 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2496 lenp, ppos, NULL, NULL);
2497 if (r || !write)
2498 return r;
2500 new_pid = find_get_pid(tmp);
2501 if (!new_pid)
2502 return -ESRCH;
2504 put_pid(xchg(&cad_pid, new_pid));
2505 return 0;
2508 #else /* CONFIG_PROC_FS */
2510 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2511 void __user *buffer, size_t *lenp, loff_t *ppos)
2513 return -ENOSYS;
2516 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2517 void __user *buffer, size_t *lenp, loff_t *ppos)
2519 return -ENOSYS;
2522 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2523 void __user *buffer, size_t *lenp, loff_t *ppos)
2525 return -ENOSYS;
2528 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2529 void __user *buffer, size_t *lenp, loff_t *ppos)
2531 return -ENOSYS;
2534 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2535 void __user *buffer, size_t *lenp, loff_t *ppos)
2537 return -ENOSYS;
2540 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2541 void __user *buffer, size_t *lenp, loff_t *ppos)
2543 return -ENOSYS;
2546 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2547 void __user *buffer, size_t *lenp, loff_t *ppos)
2549 return -ENOSYS;
2552 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2553 struct file *filp,
2554 void __user *buffer,
2555 size_t *lenp, loff_t *ppos)
2557 return -ENOSYS;
2561 #endif /* CONFIG_PROC_FS */
2564 #ifdef CONFIG_SYSCTL_SYSCALL
2566 * General sysctl support routines
2569 /* The generic sysctl data routine (used if no strategy routine supplied) */
2570 int sysctl_data(struct ctl_table *table, int __user *name, int nlen,
2571 void __user *oldval, size_t __user *oldlenp,
2572 void __user *newval, size_t newlen)
2574 size_t len;
2576 /* Get out of I don't have a variable */
2577 if (!table->data || !table->maxlen)
2578 return -ENOTDIR;
2580 if (oldval && oldlenp) {
2581 if (get_user(len, oldlenp))
2582 return -EFAULT;
2583 if (len) {
2584 if (len > table->maxlen)
2585 len = table->maxlen;
2586 if (copy_to_user(oldval, table->data, len))
2587 return -EFAULT;
2588 if (put_user(len, oldlenp))
2589 return -EFAULT;
2593 if (newval && newlen) {
2594 if (newlen > table->maxlen)
2595 newlen = table->maxlen;
2597 if (copy_from_user(table->data, newval, newlen))
2598 return -EFAULT;
2600 return 1;
2603 /* The generic string strategy routine: */
2604 int sysctl_string(struct ctl_table *table, int __user *name, int nlen,
2605 void __user *oldval, size_t __user *oldlenp,
2606 void __user *newval, size_t newlen)
2608 if (!table->data || !table->maxlen)
2609 return -ENOTDIR;
2611 if (oldval && oldlenp) {
2612 size_t bufsize;
2613 if (get_user(bufsize, oldlenp))
2614 return -EFAULT;
2615 if (bufsize) {
2616 size_t len = strlen(table->data), copied;
2618 /* This shouldn't trigger for a well-formed sysctl */
2619 if (len > table->maxlen)
2620 len = table->maxlen;
2622 /* Copy up to a max of bufsize-1 bytes of the string */
2623 copied = (len >= bufsize) ? bufsize - 1 : len;
2625 if (copy_to_user(oldval, table->data, copied) ||
2626 put_user(0, (char __user *)(oldval + copied)))
2627 return -EFAULT;
2628 if (put_user(len, oldlenp))
2629 return -EFAULT;
2632 if (newval && newlen) {
2633 size_t len = newlen;
2634 if (len > table->maxlen)
2635 len = table->maxlen;
2636 if(copy_from_user(table->data, newval, len))
2637 return -EFAULT;
2638 if (len == table->maxlen)
2639 len--;
2640 ((char *) table->data)[len] = 0;
2642 return 1;
2646 * This function makes sure that all of the integers in the vector
2647 * are between the minimum and maximum values given in the arrays
2648 * table->extra1 and table->extra2, respectively.
2650 int sysctl_intvec(struct ctl_table *table, int __user *name, int nlen,
2651 void __user *oldval, size_t __user *oldlenp,
2652 void __user *newval, size_t newlen)
2655 if (newval && newlen) {
2656 int __user *vec = (int __user *) newval;
2657 int *min = (int *) table->extra1;
2658 int *max = (int *) table->extra2;
2659 size_t length;
2660 int i;
2662 if (newlen % sizeof(int) != 0)
2663 return -EINVAL;
2665 if (!table->extra1 && !table->extra2)
2666 return 0;
2668 if (newlen > table->maxlen)
2669 newlen = table->maxlen;
2670 length = newlen / sizeof(int);
2672 for (i = 0; i < length; i++) {
2673 int value;
2674 if (get_user(value, vec + i))
2675 return -EFAULT;
2676 if (min && value < min[i])
2677 return -EINVAL;
2678 if (max && value > max[i])
2679 return -EINVAL;
2682 return 0;
2685 /* Strategy function to convert jiffies to seconds */
2686 int sysctl_jiffies(struct ctl_table *table, int __user *name, int nlen,
2687 void __user *oldval, size_t __user *oldlenp,
2688 void __user *newval, size_t newlen)
2690 if (oldval && oldlenp) {
2691 size_t olen;
2693 if (get_user(olen, oldlenp))
2694 return -EFAULT;
2695 if (olen) {
2696 int val;
2698 if (olen < sizeof(int))
2699 return -EINVAL;
2701 val = *(int *)(table->data) / HZ;
2702 if (put_user(val, (int __user *)oldval))
2703 return -EFAULT;
2704 if (put_user(sizeof(int), oldlenp))
2705 return -EFAULT;
2708 if (newval && newlen) {
2709 int new;
2710 if (newlen != sizeof(int))
2711 return -EINVAL;
2712 if (get_user(new, (int __user *)newval))
2713 return -EFAULT;
2714 *(int *)(table->data) = new*HZ;
2716 return 1;
2719 /* Strategy function to convert jiffies to seconds */
2720 int sysctl_ms_jiffies(struct ctl_table *table, int __user *name, int nlen,
2721 void __user *oldval, size_t __user *oldlenp,
2722 void __user *newval, size_t newlen)
2724 if (oldval && oldlenp) {
2725 size_t olen;
2727 if (get_user(olen, oldlenp))
2728 return -EFAULT;
2729 if (olen) {
2730 int val;
2732 if (olen < sizeof(int))
2733 return -EINVAL;
2735 val = jiffies_to_msecs(*(int *)(table->data));
2736 if (put_user(val, (int __user *)oldval))
2737 return -EFAULT;
2738 if (put_user(sizeof(int), oldlenp))
2739 return -EFAULT;
2742 if (newval && newlen) {
2743 int new;
2744 if (newlen != sizeof(int))
2745 return -EINVAL;
2746 if (get_user(new, (int __user *)newval))
2747 return -EFAULT;
2748 *(int *)(table->data) = msecs_to_jiffies(new);
2750 return 1;
2755 #else /* CONFIG_SYSCTL_SYSCALL */
2758 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2760 struct __sysctl_args tmp;
2761 int error;
2763 if (copy_from_user(&tmp, args, sizeof(tmp)))
2764 return -EFAULT;
2766 error = deprecated_sysctl_warning(&tmp);
2768 /* If no error reading the parameters then just -ENOSYS ... */
2769 if (!error)
2770 error = -ENOSYS;
2772 return error;
2775 int sysctl_data(struct ctl_table *table, int __user *name, int nlen,
2776 void __user *oldval, size_t __user *oldlenp,
2777 void __user *newval, size_t newlen)
2779 return -ENOSYS;
2782 int sysctl_string(struct ctl_table *table, int __user *name, int nlen,
2783 void __user *oldval, size_t __user *oldlenp,
2784 void __user *newval, size_t newlen)
2786 return -ENOSYS;
2789 int sysctl_intvec(struct ctl_table *table, int __user *name, int nlen,
2790 void __user *oldval, size_t __user *oldlenp,
2791 void __user *newval, size_t newlen)
2793 return -ENOSYS;
2796 int sysctl_jiffies(struct ctl_table *table, int __user *name, int nlen,
2797 void __user *oldval, size_t __user *oldlenp,
2798 void __user *newval, size_t newlen)
2800 return -ENOSYS;
2803 int sysctl_ms_jiffies(struct ctl_table *table, int __user *name, int nlen,
2804 void __user *oldval, size_t __user *oldlenp,
2805 void __user *newval, size_t newlen)
2807 return -ENOSYS;
2810 #endif /* CONFIG_SYSCTL_SYSCALL */
2812 static int deprecated_sysctl_warning(struct __sysctl_args *args)
2814 static int msg_count;
2815 int name[CTL_MAXNAME];
2816 int i;
2818 /* Check args->nlen. */
2819 if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
2820 return -ENOTDIR;
2822 /* Read in the sysctl name for better debug message logging */
2823 for (i = 0; i < args->nlen; i++)
2824 if (get_user(name[i], args->name + i))
2825 return -EFAULT;
2827 /* Ignore accesses to kernel.version */
2828 if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2829 return 0;
2831 if (msg_count < 5) {
2832 msg_count++;
2833 printk(KERN_INFO
2834 "warning: process `%s' used the deprecated sysctl "
2835 "system call with ", current->comm);
2836 for (i = 0; i < args->nlen; i++)
2837 printk("%d.", name[i]);
2838 printk("\n");
2840 return 0;
2844 * No sense putting this after each symbol definition, twice,
2845 * exception granted :-)
2847 EXPORT_SYMBOL(proc_dointvec);
2848 EXPORT_SYMBOL(proc_dointvec_jiffies);
2849 EXPORT_SYMBOL(proc_dointvec_minmax);
2850 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2851 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2852 EXPORT_SYMBOL(proc_dostring);
2853 EXPORT_SYMBOL(proc_doulongvec_minmax);
2854 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2855 EXPORT_SYMBOL(register_sysctl_table);
2856 EXPORT_SYMBOL(register_sysctl_paths);
2857 EXPORT_SYMBOL(sysctl_intvec);
2858 EXPORT_SYMBOL(sysctl_jiffies);
2859 EXPORT_SYMBOL(sysctl_ms_jiffies);
2860 EXPORT_SYMBOL(sysctl_string);
2861 EXPORT_SYMBOL(sysctl_data);
2862 EXPORT_SYMBOL(unregister_sysctl_table);