Merge branch 'proc' of git://git.kernel.org/pub/scm/linux/kernel/git/adobriyan/proc
[linux-2.6/libata-dev.git] / kernel / sysctl.c
blobc468c3c6dfc525e0c413eae9f90a8b77d77887f0
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 latencytop_enabled;
84 extern int sysctl_nr_open_min, sysctl_nr_open_max;
85 #ifdef CONFIG_RCU_TORTURE_TEST
86 extern int rcutorture_runnable;
87 #endif /* #ifdef CONFIG_RCU_TORTURE_TEST */
89 /* Constants used for minimum and maximum */
90 #if defined(CONFIG_HIGHMEM) || defined(CONFIG_DETECT_SOFTLOCKUP)
91 static int one = 1;
92 #endif
94 #ifdef CONFIG_DETECT_SOFTLOCKUP
95 static int sixty = 60;
96 static int neg_one = -1;
97 #endif
99 #ifdef CONFIG_MMU
100 static int two = 2;
101 #endif
103 static int zero;
104 static int one_hundred = 100;
106 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
107 static int maxolduid = 65535;
108 static int minolduid;
109 static int min_percpu_pagelist_fract = 8;
111 static int ngroups_max = NGROUPS_MAX;
113 #ifdef CONFIG_MODULES
114 extern char modprobe_path[];
115 #endif
116 #ifdef CONFIG_CHR_DEV_SG
117 extern int sg_big_buff;
118 #endif
120 #ifdef CONFIG_SPARC
121 #include <asm/system.h>
122 #endif
124 #ifdef __hppa__
125 extern int pwrsw_enabled;
126 extern int unaligned_enabled;
127 #endif
129 #ifdef CONFIG_S390
130 #ifdef CONFIG_MATHEMU
131 extern int sysctl_ieee_emulation_warnings;
132 #endif
133 extern int sysctl_userprocess_debug;
134 extern int spin_retry;
135 #endif
137 #ifdef CONFIG_BSD_PROCESS_ACCT
138 extern int acct_parm[];
139 #endif
141 #ifdef CONFIG_IA64
142 extern int no_unaligned_warning;
143 #endif
145 #ifdef CONFIG_RT_MUTEXES
146 extern int max_lock_depth;
147 #endif
149 #ifdef CONFIG_PROC_SYSCTL
150 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
151 void __user *buffer, size_t *lenp, loff_t *ppos);
152 static int proc_dointvec_taint(struct ctl_table *table, int write, struct file *filp,
153 void __user *buffer, size_t *lenp, loff_t *ppos);
154 #endif
156 static struct ctl_table root_table[];
157 static struct ctl_table_root sysctl_table_root;
158 static struct ctl_table_header root_table_header = {
159 .count = 1,
160 .ctl_table = root_table,
161 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
162 .root = &sysctl_table_root,
163 .set = &sysctl_table_root.default_set,
165 static struct ctl_table_root sysctl_table_root = {
166 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
167 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
170 static struct ctl_table kern_table[];
171 static struct ctl_table vm_table[];
172 static struct ctl_table fs_table[];
173 static struct ctl_table debug_table[];
174 static struct ctl_table dev_table[];
175 extern struct ctl_table random_table[];
176 #ifdef CONFIG_INOTIFY_USER
177 extern struct ctl_table inotify_table[];
178 #endif
180 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
181 int sysctl_legacy_va_layout;
182 #endif
184 extern int prove_locking;
185 extern int lock_stat;
187 /* The default sysctl tables: */
189 static struct ctl_table root_table[] = {
191 .ctl_name = CTL_KERN,
192 .procname = "kernel",
193 .mode = 0555,
194 .child = kern_table,
197 .ctl_name = CTL_VM,
198 .procname = "vm",
199 .mode = 0555,
200 .child = vm_table,
203 .ctl_name = CTL_FS,
204 .procname = "fs",
205 .mode = 0555,
206 .child = fs_table,
209 .ctl_name = CTL_DEBUG,
210 .procname = "debug",
211 .mode = 0555,
212 .child = debug_table,
215 .ctl_name = CTL_DEV,
216 .procname = "dev",
217 .mode = 0555,
218 .child = dev_table,
221 * NOTE: do not add new entries to this table unless you have read
222 * Documentation/sysctl/ctl_unnumbered.txt
224 { .ctl_name = 0 }
227 #ifdef CONFIG_SCHED_DEBUG
228 static int min_sched_granularity_ns = 100000; /* 100 usecs */
229 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
230 static int min_wakeup_granularity_ns; /* 0 usecs */
231 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
232 #endif
234 static struct ctl_table kern_table[] = {
235 #ifdef CONFIG_SCHED_DEBUG
237 .ctl_name = CTL_UNNUMBERED,
238 .procname = "sched_min_granularity_ns",
239 .data = &sysctl_sched_min_granularity,
240 .maxlen = sizeof(unsigned int),
241 .mode = 0644,
242 .proc_handler = &sched_nr_latency_handler,
243 .strategy = &sysctl_intvec,
244 .extra1 = &min_sched_granularity_ns,
245 .extra2 = &max_sched_granularity_ns,
248 .ctl_name = CTL_UNNUMBERED,
249 .procname = "sched_latency_ns",
250 .data = &sysctl_sched_latency,
251 .maxlen = sizeof(unsigned int),
252 .mode = 0644,
253 .proc_handler = &sched_nr_latency_handler,
254 .strategy = &sysctl_intvec,
255 .extra1 = &min_sched_granularity_ns,
256 .extra2 = &max_sched_granularity_ns,
259 .ctl_name = CTL_UNNUMBERED,
260 .procname = "sched_wakeup_granularity_ns",
261 .data = &sysctl_sched_wakeup_granularity,
262 .maxlen = sizeof(unsigned int),
263 .mode = 0644,
264 .proc_handler = &proc_dointvec_minmax,
265 .strategy = &sysctl_intvec,
266 .extra1 = &min_wakeup_granularity_ns,
267 .extra2 = &max_wakeup_granularity_ns,
270 .ctl_name = CTL_UNNUMBERED,
271 .procname = "sched_shares_ratelimit",
272 .data = &sysctl_sched_shares_ratelimit,
273 .maxlen = sizeof(unsigned int),
274 .mode = 0644,
275 .proc_handler = &proc_dointvec,
278 .ctl_name = CTL_UNNUMBERED,
279 .procname = "sched_child_runs_first",
280 .data = &sysctl_sched_child_runs_first,
281 .maxlen = sizeof(unsigned int),
282 .mode = 0644,
283 .proc_handler = &proc_dointvec,
286 .ctl_name = CTL_UNNUMBERED,
287 .procname = "sched_features",
288 .data = &sysctl_sched_features,
289 .maxlen = sizeof(unsigned int),
290 .mode = 0644,
291 .proc_handler = &proc_dointvec,
294 .ctl_name = CTL_UNNUMBERED,
295 .procname = "sched_migration_cost",
296 .data = &sysctl_sched_migration_cost,
297 .maxlen = sizeof(unsigned int),
298 .mode = 0644,
299 .proc_handler = &proc_dointvec,
302 .ctl_name = CTL_UNNUMBERED,
303 .procname = "sched_nr_migrate",
304 .data = &sysctl_sched_nr_migrate,
305 .maxlen = sizeof(unsigned int),
306 .mode = 0644,
307 .proc_handler = &proc_dointvec,
309 #endif
311 .ctl_name = CTL_UNNUMBERED,
312 .procname = "sched_rt_period_us",
313 .data = &sysctl_sched_rt_period,
314 .maxlen = sizeof(unsigned int),
315 .mode = 0644,
316 .proc_handler = &sched_rt_handler,
319 .ctl_name = CTL_UNNUMBERED,
320 .procname = "sched_rt_runtime_us",
321 .data = &sysctl_sched_rt_runtime,
322 .maxlen = sizeof(int),
323 .mode = 0644,
324 .proc_handler = &sched_rt_handler,
327 .ctl_name = CTL_UNNUMBERED,
328 .procname = "sched_compat_yield",
329 .data = &sysctl_sched_compat_yield,
330 .maxlen = sizeof(unsigned int),
331 .mode = 0644,
332 .proc_handler = &proc_dointvec,
334 #ifdef CONFIG_PROVE_LOCKING
336 .ctl_name = CTL_UNNUMBERED,
337 .procname = "prove_locking",
338 .data = &prove_locking,
339 .maxlen = sizeof(int),
340 .mode = 0644,
341 .proc_handler = &proc_dointvec,
343 #endif
344 #ifdef CONFIG_LOCK_STAT
346 .ctl_name = CTL_UNNUMBERED,
347 .procname = "lock_stat",
348 .data = &lock_stat,
349 .maxlen = sizeof(int),
350 .mode = 0644,
351 .proc_handler = &proc_dointvec,
353 #endif
355 .ctl_name = KERN_PANIC,
356 .procname = "panic",
357 .data = &panic_timeout,
358 .maxlen = sizeof(int),
359 .mode = 0644,
360 .proc_handler = &proc_dointvec,
363 .ctl_name = KERN_CORE_USES_PID,
364 .procname = "core_uses_pid",
365 .data = &core_uses_pid,
366 .maxlen = sizeof(int),
367 .mode = 0644,
368 .proc_handler = &proc_dointvec,
371 .ctl_name = KERN_CORE_PATTERN,
372 .procname = "core_pattern",
373 .data = core_pattern,
374 .maxlen = CORENAME_MAX_SIZE,
375 .mode = 0644,
376 .proc_handler = &proc_dostring,
377 .strategy = &sysctl_string,
379 #ifdef CONFIG_PROC_SYSCTL
381 .procname = "tainted",
382 .data = &tainted,
383 .maxlen = sizeof(int),
384 .mode = 0644,
385 .proc_handler = &proc_dointvec_taint,
387 #endif
388 #ifdef CONFIG_LATENCYTOP
390 .procname = "latencytop",
391 .data = &latencytop_enabled,
392 .maxlen = sizeof(int),
393 .mode = 0644,
394 .proc_handler = &proc_dointvec,
396 #endif
397 #ifdef CONFIG_BLK_DEV_INITRD
399 .ctl_name = KERN_REALROOTDEV,
400 .procname = "real-root-dev",
401 .data = &real_root_dev,
402 .maxlen = sizeof(int),
403 .mode = 0644,
404 .proc_handler = &proc_dointvec,
406 #endif
408 .ctl_name = CTL_UNNUMBERED,
409 .procname = "print-fatal-signals",
410 .data = &print_fatal_signals,
411 .maxlen = sizeof(int),
412 .mode = 0644,
413 .proc_handler = &proc_dointvec,
415 #ifdef CONFIG_SPARC
417 .ctl_name = KERN_SPARC_REBOOT,
418 .procname = "reboot-cmd",
419 .data = reboot_command,
420 .maxlen = 256,
421 .mode = 0644,
422 .proc_handler = &proc_dostring,
423 .strategy = &sysctl_string,
426 .ctl_name = KERN_SPARC_STOP_A,
427 .procname = "stop-a",
428 .data = &stop_a_enabled,
429 .maxlen = sizeof (int),
430 .mode = 0644,
431 .proc_handler = &proc_dointvec,
434 .ctl_name = KERN_SPARC_SCONS_PWROFF,
435 .procname = "scons-poweroff",
436 .data = &scons_pwroff,
437 .maxlen = sizeof (int),
438 .mode = 0644,
439 .proc_handler = &proc_dointvec,
441 #endif
442 #ifdef __hppa__
444 .ctl_name = KERN_HPPA_PWRSW,
445 .procname = "soft-power",
446 .data = &pwrsw_enabled,
447 .maxlen = sizeof (int),
448 .mode = 0644,
449 .proc_handler = &proc_dointvec,
452 .ctl_name = KERN_HPPA_UNALIGNED,
453 .procname = "unaligned-trap",
454 .data = &unaligned_enabled,
455 .maxlen = sizeof (int),
456 .mode = 0644,
457 .proc_handler = &proc_dointvec,
459 #endif
461 .ctl_name = KERN_CTLALTDEL,
462 .procname = "ctrl-alt-del",
463 .data = &C_A_D,
464 .maxlen = sizeof(int),
465 .mode = 0644,
466 .proc_handler = &proc_dointvec,
468 #ifdef CONFIG_FTRACE
470 .ctl_name = CTL_UNNUMBERED,
471 .procname = "ftrace_enabled",
472 .data = &ftrace_enabled,
473 .maxlen = sizeof(int),
474 .mode = 0644,
475 .proc_handler = &ftrace_enable_sysctl,
477 #endif
478 #ifdef CONFIG_MODULES
480 .ctl_name = KERN_MODPROBE,
481 .procname = "modprobe",
482 .data = &modprobe_path,
483 .maxlen = KMOD_PATH_LEN,
484 .mode = 0644,
485 .proc_handler = &proc_dostring,
486 .strategy = &sysctl_string,
488 #endif
489 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
491 .ctl_name = KERN_HOTPLUG,
492 .procname = "hotplug",
493 .data = &uevent_helper,
494 .maxlen = UEVENT_HELPER_PATH_LEN,
495 .mode = 0644,
496 .proc_handler = &proc_dostring,
497 .strategy = &sysctl_string,
499 #endif
500 #ifdef CONFIG_CHR_DEV_SG
502 .ctl_name = KERN_SG_BIG_BUFF,
503 .procname = "sg-big-buff",
504 .data = &sg_big_buff,
505 .maxlen = sizeof (int),
506 .mode = 0444,
507 .proc_handler = &proc_dointvec,
509 #endif
510 #ifdef CONFIG_BSD_PROCESS_ACCT
512 .ctl_name = KERN_ACCT,
513 .procname = "acct",
514 .data = &acct_parm,
515 .maxlen = 3*sizeof(int),
516 .mode = 0644,
517 .proc_handler = &proc_dointvec,
519 #endif
520 #ifdef CONFIG_MAGIC_SYSRQ
522 .ctl_name = KERN_SYSRQ,
523 .procname = "sysrq",
524 .data = &__sysrq_enabled,
525 .maxlen = sizeof (int),
526 .mode = 0644,
527 .proc_handler = &proc_dointvec,
529 #endif
530 #ifdef CONFIG_PROC_SYSCTL
532 .procname = "cad_pid",
533 .data = NULL,
534 .maxlen = sizeof (int),
535 .mode = 0600,
536 .proc_handler = &proc_do_cad_pid,
538 #endif
540 .ctl_name = KERN_MAX_THREADS,
541 .procname = "threads-max",
542 .data = &max_threads,
543 .maxlen = sizeof(int),
544 .mode = 0644,
545 .proc_handler = &proc_dointvec,
548 .ctl_name = KERN_RANDOM,
549 .procname = "random",
550 .mode = 0555,
551 .child = random_table,
554 .ctl_name = KERN_OVERFLOWUID,
555 .procname = "overflowuid",
556 .data = &overflowuid,
557 .maxlen = sizeof(int),
558 .mode = 0644,
559 .proc_handler = &proc_dointvec_minmax,
560 .strategy = &sysctl_intvec,
561 .extra1 = &minolduid,
562 .extra2 = &maxolduid,
565 .ctl_name = KERN_OVERFLOWGID,
566 .procname = "overflowgid",
567 .data = &overflowgid,
568 .maxlen = sizeof(int),
569 .mode = 0644,
570 .proc_handler = &proc_dointvec_minmax,
571 .strategy = &sysctl_intvec,
572 .extra1 = &minolduid,
573 .extra2 = &maxolduid,
575 #ifdef CONFIG_S390
576 #ifdef CONFIG_MATHEMU
578 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
579 .procname = "ieee_emulation_warnings",
580 .data = &sysctl_ieee_emulation_warnings,
581 .maxlen = sizeof(int),
582 .mode = 0644,
583 .proc_handler = &proc_dointvec,
585 #endif
587 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
588 .procname = "userprocess_debug",
589 .data = &sysctl_userprocess_debug,
590 .maxlen = sizeof(int),
591 .mode = 0644,
592 .proc_handler = &proc_dointvec,
594 #endif
596 .ctl_name = KERN_PIDMAX,
597 .procname = "pid_max",
598 .data = &pid_max,
599 .maxlen = sizeof (int),
600 .mode = 0644,
601 .proc_handler = &proc_dointvec_minmax,
602 .strategy = sysctl_intvec,
603 .extra1 = &pid_max_min,
604 .extra2 = &pid_max_max,
607 .ctl_name = KERN_PANIC_ON_OOPS,
608 .procname = "panic_on_oops",
609 .data = &panic_on_oops,
610 .maxlen = sizeof(int),
611 .mode = 0644,
612 .proc_handler = &proc_dointvec,
614 #if defined CONFIG_PRINTK
616 .ctl_name = KERN_PRINTK,
617 .procname = "printk",
618 .data = &console_loglevel,
619 .maxlen = 4*sizeof(int),
620 .mode = 0644,
621 .proc_handler = &proc_dointvec,
624 .ctl_name = KERN_PRINTK_RATELIMIT,
625 .procname = "printk_ratelimit",
626 .data = &printk_ratelimit_state.interval,
627 .maxlen = sizeof(int),
628 .mode = 0644,
629 .proc_handler = &proc_dointvec_jiffies,
630 .strategy = &sysctl_jiffies,
633 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
634 .procname = "printk_ratelimit_burst",
635 .data = &printk_ratelimit_state.burst,
636 .maxlen = sizeof(int),
637 .mode = 0644,
638 .proc_handler = &proc_dointvec,
640 #endif
642 .ctl_name = KERN_NGROUPS_MAX,
643 .procname = "ngroups_max",
644 .data = &ngroups_max,
645 .maxlen = sizeof (int),
646 .mode = 0444,
647 .proc_handler = &proc_dointvec,
649 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
651 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
652 .procname = "unknown_nmi_panic",
653 .data = &unknown_nmi_panic,
654 .maxlen = sizeof (int),
655 .mode = 0644,
656 .proc_handler = &proc_dointvec,
659 .procname = "nmi_watchdog",
660 .data = &nmi_watchdog_enabled,
661 .maxlen = sizeof (int),
662 .mode = 0644,
663 .proc_handler = &proc_nmi_enabled,
665 #endif
666 #if defined(CONFIG_X86)
668 .ctl_name = KERN_PANIC_ON_NMI,
669 .procname = "panic_on_unrecovered_nmi",
670 .data = &panic_on_unrecovered_nmi,
671 .maxlen = sizeof(int),
672 .mode = 0644,
673 .proc_handler = &proc_dointvec,
676 .ctl_name = KERN_BOOTLOADER_TYPE,
677 .procname = "bootloader_type",
678 .data = &bootloader_type,
679 .maxlen = sizeof (int),
680 .mode = 0444,
681 .proc_handler = &proc_dointvec,
684 .ctl_name = CTL_UNNUMBERED,
685 .procname = "kstack_depth_to_print",
686 .data = &kstack_depth_to_print,
687 .maxlen = sizeof(int),
688 .mode = 0644,
689 .proc_handler = &proc_dointvec,
692 .ctl_name = CTL_UNNUMBERED,
693 .procname = "io_delay_type",
694 .data = &io_delay_type,
695 .maxlen = sizeof(int),
696 .mode = 0644,
697 .proc_handler = &proc_dointvec,
699 #endif
700 #if defined(CONFIG_MMU)
702 .ctl_name = KERN_RANDOMIZE,
703 .procname = "randomize_va_space",
704 .data = &randomize_va_space,
705 .maxlen = sizeof(int),
706 .mode = 0644,
707 .proc_handler = &proc_dointvec,
709 #endif
710 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
712 .ctl_name = KERN_SPIN_RETRY,
713 .procname = "spin_retry",
714 .data = &spin_retry,
715 .maxlen = sizeof (int),
716 .mode = 0644,
717 .proc_handler = &proc_dointvec,
719 #endif
720 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
722 .procname = "acpi_video_flags",
723 .data = &acpi_realmode_flags,
724 .maxlen = sizeof (unsigned long),
725 .mode = 0644,
726 .proc_handler = &proc_doulongvec_minmax,
728 #endif
729 #ifdef CONFIG_IA64
731 .ctl_name = KERN_IA64_UNALIGNED,
732 .procname = "ignore-unaligned-usertrap",
733 .data = &no_unaligned_warning,
734 .maxlen = sizeof (int),
735 .mode = 0644,
736 .proc_handler = &proc_dointvec,
738 #endif
739 #ifdef CONFIG_DETECT_SOFTLOCKUP
741 .ctl_name = CTL_UNNUMBERED,
742 .procname = "softlockup_panic",
743 .data = &softlockup_panic,
744 .maxlen = sizeof(int),
745 .mode = 0644,
746 .proc_handler = &proc_dointvec_minmax,
747 .strategy = &sysctl_intvec,
748 .extra1 = &zero,
749 .extra2 = &one,
752 .ctl_name = CTL_UNNUMBERED,
753 .procname = "softlockup_thresh",
754 .data = &softlockup_thresh,
755 .maxlen = sizeof(int),
756 .mode = 0644,
757 .proc_handler = &proc_dointvec_minmax,
758 .strategy = &sysctl_intvec,
759 .extra1 = &neg_one,
760 .extra2 = &sixty,
763 .ctl_name = CTL_UNNUMBERED,
764 .procname = "hung_task_check_count",
765 .data = &sysctl_hung_task_check_count,
766 .maxlen = sizeof(unsigned long),
767 .mode = 0644,
768 .proc_handler = &proc_doulongvec_minmax,
769 .strategy = &sysctl_intvec,
772 .ctl_name = CTL_UNNUMBERED,
773 .procname = "hung_task_timeout_secs",
774 .data = &sysctl_hung_task_timeout_secs,
775 .maxlen = sizeof(unsigned long),
776 .mode = 0644,
777 .proc_handler = &proc_doulongvec_minmax,
778 .strategy = &sysctl_intvec,
781 .ctl_name = CTL_UNNUMBERED,
782 .procname = "hung_task_warnings",
783 .data = &sysctl_hung_task_warnings,
784 .maxlen = sizeof(unsigned long),
785 .mode = 0644,
786 .proc_handler = &proc_doulongvec_minmax,
787 .strategy = &sysctl_intvec,
789 #endif
790 #ifdef CONFIG_COMPAT
792 .ctl_name = KERN_COMPAT_LOG,
793 .procname = "compat-log",
794 .data = &compat_log,
795 .maxlen = sizeof (int),
796 .mode = 0644,
797 .proc_handler = &proc_dointvec,
799 #endif
800 #ifdef CONFIG_RT_MUTEXES
802 .ctl_name = KERN_MAX_LOCK_DEPTH,
803 .procname = "max_lock_depth",
804 .data = &max_lock_depth,
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
827 #ifdef CONFIG_RCU_TORTURE_TEST
829 .ctl_name = CTL_UNNUMBERED,
830 .procname = "rcutorture_runnable",
831 .data = &rcutorture_runnable,
832 .maxlen = sizeof(int),
833 .mode = 0644,
834 .proc_handler = &proc_dointvec,
836 #endif
838 * NOTE: do not add new entries to this table unless you have read
839 * Documentation/sysctl/ctl_unnumbered.txt
841 { .ctl_name = 0 }
844 static struct ctl_table vm_table[] = {
846 .ctl_name = VM_OVERCOMMIT_MEMORY,
847 .procname = "overcommit_memory",
848 .data = &sysctl_overcommit_memory,
849 .maxlen = sizeof(sysctl_overcommit_memory),
850 .mode = 0644,
851 .proc_handler = &proc_dointvec,
854 .ctl_name = VM_PANIC_ON_OOM,
855 .procname = "panic_on_oom",
856 .data = &sysctl_panic_on_oom,
857 .maxlen = sizeof(sysctl_panic_on_oom),
858 .mode = 0644,
859 .proc_handler = &proc_dointvec,
862 .ctl_name = CTL_UNNUMBERED,
863 .procname = "oom_kill_allocating_task",
864 .data = &sysctl_oom_kill_allocating_task,
865 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
866 .mode = 0644,
867 .proc_handler = &proc_dointvec,
870 .ctl_name = CTL_UNNUMBERED,
871 .procname = "oom_dump_tasks",
872 .data = &sysctl_oom_dump_tasks,
873 .maxlen = sizeof(sysctl_oom_dump_tasks),
874 .mode = 0644,
875 .proc_handler = &proc_dointvec,
878 .ctl_name = VM_OVERCOMMIT_RATIO,
879 .procname = "overcommit_ratio",
880 .data = &sysctl_overcommit_ratio,
881 .maxlen = sizeof(sysctl_overcommit_ratio),
882 .mode = 0644,
883 .proc_handler = &proc_dointvec,
886 .ctl_name = VM_PAGE_CLUSTER,
887 .procname = "page-cluster",
888 .data = &page_cluster,
889 .maxlen = sizeof(int),
890 .mode = 0644,
891 .proc_handler = &proc_dointvec,
894 .ctl_name = VM_DIRTY_BACKGROUND,
895 .procname = "dirty_background_ratio",
896 .data = &dirty_background_ratio,
897 .maxlen = sizeof(dirty_background_ratio),
898 .mode = 0644,
899 .proc_handler = &proc_dointvec_minmax,
900 .strategy = &sysctl_intvec,
901 .extra1 = &zero,
902 .extra2 = &one_hundred,
905 .ctl_name = VM_DIRTY_RATIO,
906 .procname = "dirty_ratio",
907 .data = &vm_dirty_ratio,
908 .maxlen = sizeof(vm_dirty_ratio),
909 .mode = 0644,
910 .proc_handler = &dirty_ratio_handler,
911 .strategy = &sysctl_intvec,
912 .extra1 = &zero,
913 .extra2 = &one_hundred,
916 .procname = "dirty_writeback_centisecs",
917 .data = &dirty_writeback_interval,
918 .maxlen = sizeof(dirty_writeback_interval),
919 .mode = 0644,
920 .proc_handler = &dirty_writeback_centisecs_handler,
923 .procname = "dirty_expire_centisecs",
924 .data = &dirty_expire_interval,
925 .maxlen = sizeof(dirty_expire_interval),
926 .mode = 0644,
927 .proc_handler = &proc_dointvec_userhz_jiffies,
930 .ctl_name = VM_NR_PDFLUSH_THREADS,
931 .procname = "nr_pdflush_threads",
932 .data = &nr_pdflush_threads,
933 .maxlen = sizeof nr_pdflush_threads,
934 .mode = 0444 /* read-only*/,
935 .proc_handler = &proc_dointvec,
938 .ctl_name = VM_SWAPPINESS,
939 .procname = "swappiness",
940 .data = &vm_swappiness,
941 .maxlen = sizeof(vm_swappiness),
942 .mode = 0644,
943 .proc_handler = &proc_dointvec_minmax,
944 .strategy = &sysctl_intvec,
945 .extra1 = &zero,
946 .extra2 = &one_hundred,
948 #ifdef CONFIG_HUGETLB_PAGE
950 .procname = "nr_hugepages",
951 .data = NULL,
952 .maxlen = sizeof(unsigned long),
953 .mode = 0644,
954 .proc_handler = &hugetlb_sysctl_handler,
955 .extra1 = (void *)&hugetlb_zero,
956 .extra2 = (void *)&hugetlb_infinity,
959 .ctl_name = VM_HUGETLB_GROUP,
960 .procname = "hugetlb_shm_group",
961 .data = &sysctl_hugetlb_shm_group,
962 .maxlen = sizeof(gid_t),
963 .mode = 0644,
964 .proc_handler = &proc_dointvec,
967 .ctl_name = CTL_UNNUMBERED,
968 .procname = "hugepages_treat_as_movable",
969 .data = &hugepages_treat_as_movable,
970 .maxlen = sizeof(int),
971 .mode = 0644,
972 .proc_handler = &hugetlb_treat_movable_handler,
975 .ctl_name = CTL_UNNUMBERED,
976 .procname = "nr_overcommit_hugepages",
977 .data = NULL,
978 .maxlen = sizeof(unsigned long),
979 .mode = 0644,
980 .proc_handler = &hugetlb_overcommit_handler,
981 .extra1 = (void *)&hugetlb_zero,
982 .extra2 = (void *)&hugetlb_infinity,
984 #endif
986 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
987 .procname = "lowmem_reserve_ratio",
988 .data = &sysctl_lowmem_reserve_ratio,
989 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
990 .mode = 0644,
991 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
992 .strategy = &sysctl_intvec,
995 .ctl_name = VM_DROP_PAGECACHE,
996 .procname = "drop_caches",
997 .data = &sysctl_drop_caches,
998 .maxlen = sizeof(int),
999 .mode = 0644,
1000 .proc_handler = drop_caches_sysctl_handler,
1001 .strategy = &sysctl_intvec,
1004 .ctl_name = VM_MIN_FREE_KBYTES,
1005 .procname = "min_free_kbytes",
1006 .data = &min_free_kbytes,
1007 .maxlen = sizeof(min_free_kbytes),
1008 .mode = 0644,
1009 .proc_handler = &min_free_kbytes_sysctl_handler,
1010 .strategy = &sysctl_intvec,
1011 .extra1 = &zero,
1014 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
1015 .procname = "percpu_pagelist_fraction",
1016 .data = &percpu_pagelist_fraction,
1017 .maxlen = sizeof(percpu_pagelist_fraction),
1018 .mode = 0644,
1019 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
1020 .strategy = &sysctl_intvec,
1021 .extra1 = &min_percpu_pagelist_fract,
1023 #ifdef CONFIG_MMU
1025 .ctl_name = VM_MAX_MAP_COUNT,
1026 .procname = "max_map_count",
1027 .data = &sysctl_max_map_count,
1028 .maxlen = sizeof(sysctl_max_map_count),
1029 .mode = 0644,
1030 .proc_handler = &proc_dointvec
1032 #endif
1034 .ctl_name = VM_LAPTOP_MODE,
1035 .procname = "laptop_mode",
1036 .data = &laptop_mode,
1037 .maxlen = sizeof(laptop_mode),
1038 .mode = 0644,
1039 .proc_handler = &proc_dointvec_jiffies,
1040 .strategy = &sysctl_jiffies,
1043 .ctl_name = VM_BLOCK_DUMP,
1044 .procname = "block_dump",
1045 .data = &block_dump,
1046 .maxlen = sizeof(block_dump),
1047 .mode = 0644,
1048 .proc_handler = &proc_dointvec,
1049 .strategy = &sysctl_intvec,
1050 .extra1 = &zero,
1053 .ctl_name = VM_VFS_CACHE_PRESSURE,
1054 .procname = "vfs_cache_pressure",
1055 .data = &sysctl_vfs_cache_pressure,
1056 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1057 .mode = 0644,
1058 .proc_handler = &proc_dointvec,
1059 .strategy = &sysctl_intvec,
1060 .extra1 = &zero,
1062 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1064 .ctl_name = VM_LEGACY_VA_LAYOUT,
1065 .procname = "legacy_va_layout",
1066 .data = &sysctl_legacy_va_layout,
1067 .maxlen = sizeof(sysctl_legacy_va_layout),
1068 .mode = 0644,
1069 .proc_handler = &proc_dointvec,
1070 .strategy = &sysctl_intvec,
1071 .extra1 = &zero,
1073 #endif
1074 #ifdef CONFIG_NUMA
1076 .ctl_name = VM_ZONE_RECLAIM_MODE,
1077 .procname = "zone_reclaim_mode",
1078 .data = &zone_reclaim_mode,
1079 .maxlen = sizeof(zone_reclaim_mode),
1080 .mode = 0644,
1081 .proc_handler = &proc_dointvec,
1082 .strategy = &sysctl_intvec,
1083 .extra1 = &zero,
1086 .ctl_name = VM_MIN_UNMAPPED,
1087 .procname = "min_unmapped_ratio",
1088 .data = &sysctl_min_unmapped_ratio,
1089 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1090 .mode = 0644,
1091 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
1092 .strategy = &sysctl_intvec,
1093 .extra1 = &zero,
1094 .extra2 = &one_hundred,
1097 .ctl_name = VM_MIN_SLAB,
1098 .procname = "min_slab_ratio",
1099 .data = &sysctl_min_slab_ratio,
1100 .maxlen = sizeof(sysctl_min_slab_ratio),
1101 .mode = 0644,
1102 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1103 .strategy = &sysctl_intvec,
1104 .extra1 = &zero,
1105 .extra2 = &one_hundred,
1107 #endif
1108 #ifdef CONFIG_SMP
1110 .ctl_name = CTL_UNNUMBERED,
1111 .procname = "stat_interval",
1112 .data = &sysctl_stat_interval,
1113 .maxlen = sizeof(sysctl_stat_interval),
1114 .mode = 0644,
1115 .proc_handler = &proc_dointvec_jiffies,
1116 .strategy = &sysctl_jiffies,
1118 #endif
1119 #ifdef CONFIG_SECURITY
1121 .ctl_name = CTL_UNNUMBERED,
1122 .procname = "mmap_min_addr",
1123 .data = &mmap_min_addr,
1124 .maxlen = sizeof(unsigned long),
1125 .mode = 0644,
1126 .proc_handler = &proc_doulongvec_minmax,
1128 #endif
1129 #ifdef CONFIG_NUMA
1131 .ctl_name = CTL_UNNUMBERED,
1132 .procname = "numa_zonelist_order",
1133 .data = &numa_zonelist_order,
1134 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1135 .mode = 0644,
1136 .proc_handler = &numa_zonelist_order_handler,
1137 .strategy = &sysctl_string,
1139 #endif
1140 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1141 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1143 .ctl_name = VM_VDSO_ENABLED,
1144 .procname = "vdso_enabled",
1145 .data = &vdso_enabled,
1146 .maxlen = sizeof(vdso_enabled),
1147 .mode = 0644,
1148 .proc_handler = &proc_dointvec,
1149 .strategy = &sysctl_intvec,
1150 .extra1 = &zero,
1152 #endif
1153 #ifdef CONFIG_HIGHMEM
1155 .ctl_name = CTL_UNNUMBERED,
1156 .procname = "highmem_is_dirtyable",
1157 .data = &vm_highmem_is_dirtyable,
1158 .maxlen = sizeof(vm_highmem_is_dirtyable),
1159 .mode = 0644,
1160 .proc_handler = &proc_dointvec_minmax,
1161 .strategy = &sysctl_intvec,
1162 .extra1 = &zero,
1163 .extra2 = &one,
1165 #endif
1167 * NOTE: do not add new entries to this table unless you have read
1168 * Documentation/sysctl/ctl_unnumbered.txt
1170 { .ctl_name = 0 }
1173 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1174 static struct ctl_table binfmt_misc_table[] = {
1175 { .ctl_name = 0 }
1177 #endif
1179 static struct ctl_table fs_table[] = {
1181 .ctl_name = FS_NRINODE,
1182 .procname = "inode-nr",
1183 .data = &inodes_stat,
1184 .maxlen = 2*sizeof(int),
1185 .mode = 0444,
1186 .proc_handler = &proc_dointvec,
1189 .ctl_name = FS_STATINODE,
1190 .procname = "inode-state",
1191 .data = &inodes_stat,
1192 .maxlen = 7*sizeof(int),
1193 .mode = 0444,
1194 .proc_handler = &proc_dointvec,
1197 .procname = "file-nr",
1198 .data = &files_stat,
1199 .maxlen = 3*sizeof(int),
1200 .mode = 0444,
1201 .proc_handler = &proc_nr_files,
1204 .ctl_name = FS_MAXFILE,
1205 .procname = "file-max",
1206 .data = &files_stat.max_files,
1207 .maxlen = sizeof(int),
1208 .mode = 0644,
1209 .proc_handler = &proc_dointvec,
1212 .ctl_name = CTL_UNNUMBERED,
1213 .procname = "nr_open",
1214 .data = &sysctl_nr_open,
1215 .maxlen = sizeof(int),
1216 .mode = 0644,
1217 .proc_handler = &proc_dointvec_minmax,
1218 .extra1 = &sysctl_nr_open_min,
1219 .extra2 = &sysctl_nr_open_max,
1222 .ctl_name = FS_DENTRY,
1223 .procname = "dentry-state",
1224 .data = &dentry_stat,
1225 .maxlen = 6*sizeof(int),
1226 .mode = 0444,
1227 .proc_handler = &proc_dointvec,
1230 .ctl_name = FS_OVERFLOWUID,
1231 .procname = "overflowuid",
1232 .data = &fs_overflowuid,
1233 .maxlen = sizeof(int),
1234 .mode = 0644,
1235 .proc_handler = &proc_dointvec_minmax,
1236 .strategy = &sysctl_intvec,
1237 .extra1 = &minolduid,
1238 .extra2 = &maxolduid,
1241 .ctl_name = FS_OVERFLOWGID,
1242 .procname = "overflowgid",
1243 .data = &fs_overflowgid,
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_LEASES,
1253 .procname = "leases-enable",
1254 .data = &leases_enable,
1255 .maxlen = sizeof(int),
1256 .mode = 0644,
1257 .proc_handler = &proc_dointvec,
1259 #ifdef CONFIG_DNOTIFY
1261 .ctl_name = FS_DIR_NOTIFY,
1262 .procname = "dir-notify-enable",
1263 .data = &dir_notify_enable,
1264 .maxlen = sizeof(int),
1265 .mode = 0644,
1266 .proc_handler = &proc_dointvec,
1268 #endif
1269 #ifdef CONFIG_MMU
1271 .ctl_name = FS_LEASE_TIME,
1272 .procname = "lease-break-time",
1273 .data = &lease_break_time,
1274 .maxlen = sizeof(int),
1275 .mode = 0644,
1276 .proc_handler = &proc_dointvec_minmax,
1277 .strategy = &sysctl_intvec,
1278 .extra1 = &zero,
1279 .extra2 = &two,
1282 .procname = "aio-nr",
1283 .data = &aio_nr,
1284 .maxlen = sizeof(aio_nr),
1285 .mode = 0444,
1286 .proc_handler = &proc_doulongvec_minmax,
1289 .procname = "aio-max-nr",
1290 .data = &aio_max_nr,
1291 .maxlen = sizeof(aio_max_nr),
1292 .mode = 0644,
1293 .proc_handler = &proc_doulongvec_minmax,
1295 #ifdef CONFIG_INOTIFY_USER
1297 .ctl_name = FS_INOTIFY,
1298 .procname = "inotify",
1299 .mode = 0555,
1300 .child = inotify_table,
1302 #endif
1303 #endif
1305 .ctl_name = KERN_SETUID_DUMPABLE,
1306 .procname = "suid_dumpable",
1307 .data = &suid_dumpable,
1308 .maxlen = sizeof(int),
1309 .mode = 0644,
1310 .proc_handler = &proc_dointvec,
1312 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1314 .ctl_name = CTL_UNNUMBERED,
1315 .procname = "binfmt_misc",
1316 .mode = 0555,
1317 .child = binfmt_misc_table,
1319 #endif
1321 * NOTE: do not add new entries to this table unless you have read
1322 * Documentation/sysctl/ctl_unnumbered.txt
1324 { .ctl_name = 0 }
1327 static struct ctl_table debug_table[] = {
1328 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1330 .ctl_name = CTL_UNNUMBERED,
1331 .procname = "exception-trace",
1332 .data = &show_unhandled_signals,
1333 .maxlen = sizeof(int),
1334 .mode = 0644,
1335 .proc_handler = proc_dointvec
1337 #endif
1338 { .ctl_name = 0 }
1341 static struct ctl_table dev_table[] = {
1342 { .ctl_name = 0 }
1345 static DEFINE_SPINLOCK(sysctl_lock);
1347 /* called under sysctl_lock */
1348 static int use_table(struct ctl_table_header *p)
1350 if (unlikely(p->unregistering))
1351 return 0;
1352 p->used++;
1353 return 1;
1356 /* called under sysctl_lock */
1357 static void unuse_table(struct ctl_table_header *p)
1359 if (!--p->used)
1360 if (unlikely(p->unregistering))
1361 complete(p->unregistering);
1364 /* called under sysctl_lock, will reacquire if has to wait */
1365 static void start_unregistering(struct ctl_table_header *p)
1368 * if p->used is 0, nobody will ever touch that entry again;
1369 * we'll eliminate all paths to it before dropping sysctl_lock
1371 if (unlikely(p->used)) {
1372 struct completion wait;
1373 init_completion(&wait);
1374 p->unregistering = &wait;
1375 spin_unlock(&sysctl_lock);
1376 wait_for_completion(&wait);
1377 spin_lock(&sysctl_lock);
1378 } else {
1379 /* anything non-NULL; we'll never dereference it */
1380 p->unregistering = ERR_PTR(-EINVAL);
1383 * do not remove from the list until nobody holds it; walking the
1384 * list in do_sysctl() relies on that.
1386 list_del_init(&p->ctl_entry);
1389 void sysctl_head_get(struct ctl_table_header *head)
1391 spin_lock(&sysctl_lock);
1392 head->count++;
1393 spin_unlock(&sysctl_lock);
1396 void sysctl_head_put(struct ctl_table_header *head)
1398 spin_lock(&sysctl_lock);
1399 if (!--head->count)
1400 kfree(head);
1401 spin_unlock(&sysctl_lock);
1404 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1406 if (!head)
1407 BUG();
1408 spin_lock(&sysctl_lock);
1409 if (!use_table(head))
1410 head = ERR_PTR(-ENOENT);
1411 spin_unlock(&sysctl_lock);
1412 return head;
1415 void sysctl_head_finish(struct ctl_table_header *head)
1417 if (!head)
1418 return;
1419 spin_lock(&sysctl_lock);
1420 unuse_table(head);
1421 spin_unlock(&sysctl_lock);
1424 static struct ctl_table_set *
1425 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1427 struct ctl_table_set *set = &root->default_set;
1428 if (root->lookup)
1429 set = root->lookup(root, namespaces);
1430 return set;
1433 static struct list_head *
1434 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1436 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1437 return &set->list;
1440 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1441 struct ctl_table_header *prev)
1443 struct ctl_table_root *root;
1444 struct list_head *header_list;
1445 struct ctl_table_header *head;
1446 struct list_head *tmp;
1448 spin_lock(&sysctl_lock);
1449 if (prev) {
1450 head = prev;
1451 tmp = &prev->ctl_entry;
1452 unuse_table(prev);
1453 goto next;
1455 tmp = &root_table_header.ctl_entry;
1456 for (;;) {
1457 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1459 if (!use_table(head))
1460 goto next;
1461 spin_unlock(&sysctl_lock);
1462 return head;
1463 next:
1464 root = head->root;
1465 tmp = tmp->next;
1466 header_list = lookup_header_list(root, namespaces);
1467 if (tmp != header_list)
1468 continue;
1470 do {
1471 root = list_entry(root->root_list.next,
1472 struct ctl_table_root, root_list);
1473 if (root == &sysctl_table_root)
1474 goto out;
1475 header_list = lookup_header_list(root, namespaces);
1476 } while (list_empty(header_list));
1477 tmp = header_list->next;
1479 out:
1480 spin_unlock(&sysctl_lock);
1481 return NULL;
1484 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1486 return __sysctl_head_next(current->nsproxy, prev);
1489 void register_sysctl_root(struct ctl_table_root *root)
1491 spin_lock(&sysctl_lock);
1492 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1493 spin_unlock(&sysctl_lock);
1496 #ifdef CONFIG_SYSCTL_SYSCALL
1497 /* Perform the actual read/write of a sysctl table entry. */
1498 static int do_sysctl_strategy(struct ctl_table_root *root,
1499 struct ctl_table *table,
1500 int __user *name, int nlen,
1501 void __user *oldval, size_t __user *oldlenp,
1502 void __user *newval, size_t newlen)
1504 int op = 0, rc;
1506 if (oldval)
1507 op |= MAY_READ;
1508 if (newval)
1509 op |= MAY_WRITE;
1510 if (sysctl_perm(root, table, op))
1511 return -EPERM;
1513 if (table->strategy) {
1514 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1515 newval, newlen);
1516 if (rc < 0)
1517 return rc;
1518 if (rc > 0)
1519 return 0;
1522 /* If there is no strategy routine, or if the strategy returns
1523 * zero, proceed with automatic r/w */
1524 if (table->data && table->maxlen) {
1525 rc = sysctl_data(table, name, nlen, oldval, oldlenp,
1526 newval, newlen);
1527 if (rc < 0)
1528 return rc;
1530 return 0;
1533 static int parse_table(int __user *name, int nlen,
1534 void __user *oldval, size_t __user *oldlenp,
1535 void __user *newval, size_t newlen,
1536 struct ctl_table_root *root,
1537 struct ctl_table *table)
1539 int n;
1540 repeat:
1541 if (!nlen)
1542 return -ENOTDIR;
1543 if (get_user(n, name))
1544 return -EFAULT;
1545 for ( ; table->ctl_name || table->procname; table++) {
1546 if (!table->ctl_name)
1547 continue;
1548 if (n == table->ctl_name) {
1549 int error;
1550 if (table->child) {
1551 if (sysctl_perm(root, table, MAY_EXEC))
1552 return -EPERM;
1553 name++;
1554 nlen--;
1555 table = table->child;
1556 goto repeat;
1558 error = do_sysctl_strategy(root, table, name, nlen,
1559 oldval, oldlenp,
1560 newval, newlen);
1561 return error;
1564 return -ENOTDIR;
1567 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1568 void __user *newval, size_t newlen)
1570 struct ctl_table_header *head;
1571 int error = -ENOTDIR;
1573 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1574 return -ENOTDIR;
1575 if (oldval) {
1576 int old_len;
1577 if (!oldlenp || get_user(old_len, oldlenp))
1578 return -EFAULT;
1581 for (head = sysctl_head_next(NULL); head;
1582 head = sysctl_head_next(head)) {
1583 error = parse_table(name, nlen, oldval, oldlenp,
1584 newval, newlen,
1585 head->root, head->ctl_table);
1586 if (error != -ENOTDIR) {
1587 sysctl_head_finish(head);
1588 break;
1591 return error;
1594 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1596 struct __sysctl_args tmp;
1597 int error;
1599 if (copy_from_user(&tmp, args, sizeof(tmp)))
1600 return -EFAULT;
1602 error = deprecated_sysctl_warning(&tmp);
1603 if (error)
1604 goto out;
1606 lock_kernel();
1607 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1608 tmp.newval, tmp.newlen);
1609 unlock_kernel();
1610 out:
1611 return error;
1613 #endif /* CONFIG_SYSCTL_SYSCALL */
1616 * sysctl_perm does NOT grant the superuser all rights automatically, because
1617 * some sysctl variables are readonly even to root.
1620 static int test_perm(int mode, int op)
1622 if (!current->euid)
1623 mode >>= 6;
1624 else if (in_egroup_p(0))
1625 mode >>= 3;
1626 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1627 return 0;
1628 return -EACCES;
1631 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1633 int error;
1634 int mode;
1636 error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1637 if (error)
1638 return error;
1640 if (root->permissions)
1641 mode = root->permissions(root, current->nsproxy, table);
1642 else
1643 mode = table->mode;
1645 return test_perm(mode, op);
1648 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1650 for (; table->ctl_name || table->procname; table++) {
1651 table->parent = parent;
1652 if (table->child)
1653 sysctl_set_parent(table, table->child);
1657 static __init int sysctl_init(void)
1659 sysctl_set_parent(NULL, root_table);
1660 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1662 int err;
1663 err = sysctl_check_table(current->nsproxy, root_table);
1665 #endif
1666 return 0;
1669 core_initcall(sysctl_init);
1671 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1672 struct ctl_table *table)
1674 struct ctl_table *p;
1675 const char *s = branch->procname;
1677 /* branch should have named subdirectory as its first element */
1678 if (!s || !branch->child)
1679 return NULL;
1681 /* ... and nothing else */
1682 if (branch[1].procname || branch[1].ctl_name)
1683 return NULL;
1685 /* table should contain subdirectory with the same name */
1686 for (p = table; p->procname || p->ctl_name; p++) {
1687 if (!p->child)
1688 continue;
1689 if (p->procname && strcmp(p->procname, s) == 0)
1690 return p;
1692 return NULL;
1695 /* see if attaching q to p would be an improvement */
1696 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1698 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1699 struct ctl_table *next;
1700 int is_better = 0;
1701 int not_in_parent = !p->attached_by;
1703 while ((next = is_branch_in(by, to)) != NULL) {
1704 if (by == q->attached_by)
1705 is_better = 1;
1706 if (to == p->attached_by)
1707 not_in_parent = 1;
1708 by = by->child;
1709 to = next->child;
1712 if (is_better && not_in_parent) {
1713 q->attached_by = by;
1714 q->attached_to = to;
1715 q->parent = p;
1720 * __register_sysctl_paths - register a sysctl hierarchy
1721 * @root: List of sysctl headers to register on
1722 * @namespaces: Data to compute which lists of sysctl entries are visible
1723 * @path: The path to the directory the sysctl table is in.
1724 * @table: the top-level table structure
1726 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1727 * array. A completely 0 filled entry terminates the table.
1729 * The members of the &struct ctl_table structure are used as follows:
1731 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1732 * must be unique within that level of sysctl
1734 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1735 * enter a sysctl file
1737 * data - a pointer to data for use by proc_handler
1739 * maxlen - the maximum size in bytes of the data
1741 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1743 * child - a pointer to the child sysctl table if this entry is a directory, or
1744 * %NULL.
1746 * proc_handler - the text handler routine (described below)
1748 * strategy - the strategy routine (described below)
1750 * de - for internal use by the sysctl routines
1752 * extra1, extra2 - extra pointers usable by the proc handler routines
1754 * Leaf nodes in the sysctl tree will be represented by a single file
1755 * under /proc; non-leaf nodes will be represented by directories.
1757 * sysctl(2) can automatically manage read and write requests through
1758 * the sysctl table. The data and maxlen fields of the ctl_table
1759 * struct enable minimal validation of the values being written to be
1760 * performed, and the mode field allows minimal authentication.
1762 * More sophisticated management can be enabled by the provision of a
1763 * strategy routine with the table entry. This will be called before
1764 * any automatic read or write of the data is performed.
1766 * The strategy routine may return
1768 * < 0 - Error occurred (error is passed to user process)
1770 * 0 - OK - proceed with automatic read or write.
1772 * > 0 - OK - read or write has been done by the strategy routine, so
1773 * return immediately.
1775 * There must be a proc_handler routine for any terminal nodes
1776 * mirrored under /proc/sys (non-terminals are handled by a built-in
1777 * directory handler). Several default handlers are available to
1778 * cover common cases -
1780 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1781 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1782 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1784 * It is the handler's job to read the input buffer from user memory
1785 * and process it. The handler should return 0 on success.
1787 * This routine returns %NULL on a failure to register, and a pointer
1788 * to the table header on success.
1790 struct ctl_table_header *__register_sysctl_paths(
1791 struct ctl_table_root *root,
1792 struct nsproxy *namespaces,
1793 const struct ctl_path *path, struct ctl_table *table)
1795 struct ctl_table_header *header;
1796 struct ctl_table *new, **prevp;
1797 unsigned int n, npath;
1798 struct ctl_table_set *set;
1800 /* Count the path components */
1801 for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1805 * For each path component, allocate a 2-element ctl_table array.
1806 * The first array element will be filled with the sysctl entry
1807 * for this, the second will be the sentinel (ctl_name == 0).
1809 * We allocate everything in one go so that we don't have to
1810 * worry about freeing additional memory in unregister_sysctl_table.
1812 header = kzalloc(sizeof(struct ctl_table_header) +
1813 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1814 if (!header)
1815 return NULL;
1817 new = (struct ctl_table *) (header + 1);
1819 /* Now connect the dots */
1820 prevp = &header->ctl_table;
1821 for (n = 0; n < npath; ++n, ++path) {
1822 /* Copy the procname */
1823 new->procname = path->procname;
1824 new->ctl_name = path->ctl_name;
1825 new->mode = 0555;
1827 *prevp = new;
1828 prevp = &new->child;
1830 new += 2;
1832 *prevp = table;
1833 header->ctl_table_arg = table;
1835 INIT_LIST_HEAD(&header->ctl_entry);
1836 header->used = 0;
1837 header->unregistering = NULL;
1838 header->root = root;
1839 sysctl_set_parent(NULL, header->ctl_table);
1840 header->count = 1;
1841 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1842 if (sysctl_check_table(namespaces, header->ctl_table)) {
1843 kfree(header);
1844 return NULL;
1846 #endif
1847 spin_lock(&sysctl_lock);
1848 header->set = lookup_header_set(root, namespaces);
1849 header->attached_by = header->ctl_table;
1850 header->attached_to = root_table;
1851 header->parent = &root_table_header;
1852 for (set = header->set; set; set = set->parent) {
1853 struct ctl_table_header *p;
1854 list_for_each_entry(p, &set->list, ctl_entry) {
1855 if (p->unregistering)
1856 continue;
1857 try_attach(p, header);
1860 header->parent->count++;
1861 list_add_tail(&header->ctl_entry, &header->set->list);
1862 spin_unlock(&sysctl_lock);
1864 return header;
1868 * register_sysctl_table_path - register a sysctl table hierarchy
1869 * @path: The path to the directory the sysctl table is in.
1870 * @table: the top-level table structure
1872 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1873 * array. A completely 0 filled entry terminates the table.
1875 * See __register_sysctl_paths for more details.
1877 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1878 struct ctl_table *table)
1880 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1881 path, table);
1885 * register_sysctl_table - register a sysctl table hierarchy
1886 * @table: the top-level table structure
1888 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1889 * array. A completely 0 filled entry terminates the table.
1891 * See register_sysctl_paths for more details.
1893 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1895 static const struct ctl_path null_path[] = { {} };
1897 return register_sysctl_paths(null_path, table);
1901 * unregister_sysctl_table - unregister a sysctl table hierarchy
1902 * @header: the header returned from register_sysctl_table
1904 * Unregisters the sysctl table and all children. proc entries may not
1905 * actually be removed until they are no longer used by anyone.
1907 void unregister_sysctl_table(struct ctl_table_header * header)
1909 might_sleep();
1911 if (header == NULL)
1912 return;
1914 spin_lock(&sysctl_lock);
1915 start_unregistering(header);
1916 if (!--header->parent->count) {
1917 WARN_ON(1);
1918 kfree(header->parent);
1920 if (!--header->count)
1921 kfree(header);
1922 spin_unlock(&sysctl_lock);
1925 int sysctl_is_seen(struct ctl_table_header *p)
1927 struct ctl_table_set *set = p->set;
1928 int res;
1929 spin_lock(&sysctl_lock);
1930 if (p->unregistering)
1931 res = 0;
1932 else if (!set->is_seen)
1933 res = 1;
1934 else
1935 res = set->is_seen(set);
1936 spin_unlock(&sysctl_lock);
1937 return res;
1940 void setup_sysctl_set(struct ctl_table_set *p,
1941 struct ctl_table_set *parent,
1942 int (*is_seen)(struct ctl_table_set *))
1944 INIT_LIST_HEAD(&p->list);
1945 p->parent = parent ? parent : &sysctl_table_root.default_set;
1946 p->is_seen = is_seen;
1949 #else /* !CONFIG_SYSCTL */
1950 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1952 return NULL;
1955 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1956 struct ctl_table *table)
1958 return NULL;
1961 void unregister_sysctl_table(struct ctl_table_header * table)
1965 void setup_sysctl_set(struct ctl_table_set *p,
1966 struct ctl_table_set *parent,
1967 int (*is_seen)(struct ctl_table_set *))
1971 void sysctl_head_put(struct ctl_table_header *head)
1975 #endif /* CONFIG_SYSCTL */
1978 * /proc/sys support
1981 #ifdef CONFIG_PROC_SYSCTL
1983 static int _proc_do_string(void* data, int maxlen, int write,
1984 struct file *filp, void __user *buffer,
1985 size_t *lenp, loff_t *ppos)
1987 size_t len;
1988 char __user *p;
1989 char c;
1991 if (!data || !maxlen || !*lenp) {
1992 *lenp = 0;
1993 return 0;
1996 if (write) {
1997 len = 0;
1998 p = buffer;
1999 while (len < *lenp) {
2000 if (get_user(c, p++))
2001 return -EFAULT;
2002 if (c == 0 || c == '\n')
2003 break;
2004 len++;
2006 if (len >= maxlen)
2007 len = maxlen-1;
2008 if(copy_from_user(data, buffer, len))
2009 return -EFAULT;
2010 ((char *) data)[len] = 0;
2011 *ppos += *lenp;
2012 } else {
2013 len = strlen(data);
2014 if (len > maxlen)
2015 len = maxlen;
2017 if (*ppos > len) {
2018 *lenp = 0;
2019 return 0;
2022 data += *ppos;
2023 len -= *ppos;
2025 if (len > *lenp)
2026 len = *lenp;
2027 if (len)
2028 if(copy_to_user(buffer, data, len))
2029 return -EFAULT;
2030 if (len < *lenp) {
2031 if(put_user('\n', ((char __user *) buffer) + len))
2032 return -EFAULT;
2033 len++;
2035 *lenp = len;
2036 *ppos += len;
2038 return 0;
2042 * proc_dostring - read a string sysctl
2043 * @table: the sysctl table
2044 * @write: %TRUE if this is a write to the sysctl file
2045 * @filp: the file structure
2046 * @buffer: the user buffer
2047 * @lenp: the size of the user buffer
2048 * @ppos: file position
2050 * Reads/writes a string from/to the user buffer. If the kernel
2051 * buffer provided is not large enough to hold the string, the
2052 * string is truncated. The copied string is %NULL-terminated.
2053 * If the string is being read by the user process, it is copied
2054 * and a newline '\n' is added. It is truncated if the buffer is
2055 * not large enough.
2057 * Returns 0 on success.
2059 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2060 void __user *buffer, size_t *lenp, loff_t *ppos)
2062 return _proc_do_string(table->data, table->maxlen, write, filp,
2063 buffer, lenp, ppos);
2067 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2068 int *valp,
2069 int write, void *data)
2071 if (write) {
2072 *valp = *negp ? -*lvalp : *lvalp;
2073 } else {
2074 int val = *valp;
2075 if (val < 0) {
2076 *negp = -1;
2077 *lvalp = (unsigned long)-val;
2078 } else {
2079 *negp = 0;
2080 *lvalp = (unsigned long)val;
2083 return 0;
2086 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2087 int write, struct file *filp, void __user *buffer,
2088 size_t *lenp, loff_t *ppos,
2089 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2090 int write, void *data),
2091 void *data)
2093 #define TMPBUFLEN 21
2094 int *i, vleft, first=1, neg, val;
2095 unsigned long lval;
2096 size_t left, len;
2098 char buf[TMPBUFLEN], *p;
2099 char __user *s = buffer;
2101 if (!tbl_data || !table->maxlen || !*lenp ||
2102 (*ppos && !write)) {
2103 *lenp = 0;
2104 return 0;
2107 i = (int *) tbl_data;
2108 vleft = table->maxlen / sizeof(*i);
2109 left = *lenp;
2111 if (!conv)
2112 conv = do_proc_dointvec_conv;
2114 for (; left && vleft--; i++, first=0) {
2115 if (write) {
2116 while (left) {
2117 char c;
2118 if (get_user(c, s))
2119 return -EFAULT;
2120 if (!isspace(c))
2121 break;
2122 left--;
2123 s++;
2125 if (!left)
2126 break;
2127 neg = 0;
2128 len = left;
2129 if (len > sizeof(buf) - 1)
2130 len = sizeof(buf) - 1;
2131 if (copy_from_user(buf, s, len))
2132 return -EFAULT;
2133 buf[len] = 0;
2134 p = buf;
2135 if (*p == '-' && left > 1) {
2136 neg = 1;
2137 p++;
2139 if (*p < '0' || *p > '9')
2140 break;
2142 lval = simple_strtoul(p, &p, 0);
2144 len = p-buf;
2145 if ((len < left) && *p && !isspace(*p))
2146 break;
2147 if (neg)
2148 val = -val;
2149 s += len;
2150 left -= len;
2152 if (conv(&neg, &lval, i, 1, data))
2153 break;
2154 } else {
2155 p = buf;
2156 if (!first)
2157 *p++ = '\t';
2159 if (conv(&neg, &lval, i, 0, data))
2160 break;
2162 sprintf(p, "%s%lu", neg ? "-" : "", lval);
2163 len = strlen(buf);
2164 if (len > left)
2165 len = left;
2166 if(copy_to_user(s, buf, len))
2167 return -EFAULT;
2168 left -= len;
2169 s += len;
2173 if (!write && !first && left) {
2174 if(put_user('\n', s))
2175 return -EFAULT;
2176 left--, s++;
2178 if (write) {
2179 while (left) {
2180 char c;
2181 if (get_user(c, s++))
2182 return -EFAULT;
2183 if (!isspace(c))
2184 break;
2185 left--;
2188 if (write && first)
2189 return -EINVAL;
2190 *lenp -= left;
2191 *ppos += *lenp;
2192 return 0;
2193 #undef TMPBUFLEN
2196 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2197 void __user *buffer, size_t *lenp, loff_t *ppos,
2198 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2199 int write, void *data),
2200 void *data)
2202 return __do_proc_dointvec(table->data, table, write, filp,
2203 buffer, lenp, ppos, conv, data);
2207 * proc_dointvec - read a vector of integers
2208 * @table: the sysctl table
2209 * @write: %TRUE if this is a write to the sysctl file
2210 * @filp: the file structure
2211 * @buffer: the user buffer
2212 * @lenp: the size of the user buffer
2213 * @ppos: file position
2215 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2216 * values from/to the user buffer, treated as an ASCII string.
2218 * Returns 0 on success.
2220 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2221 void __user *buffer, size_t *lenp, loff_t *ppos)
2223 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2224 NULL,NULL);
2227 #define OP_SET 0
2228 #define OP_AND 1
2229 #define OP_OR 2
2231 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
2232 int *valp,
2233 int write, void *data)
2235 int op = *(int *)data;
2236 if (write) {
2237 int val = *negp ? -*lvalp : *lvalp;
2238 switch(op) {
2239 case OP_SET: *valp = val; break;
2240 case OP_AND: *valp &= val; break;
2241 case OP_OR: *valp |= val; break;
2243 } else {
2244 int val = *valp;
2245 if (val < 0) {
2246 *negp = -1;
2247 *lvalp = (unsigned long)-val;
2248 } else {
2249 *negp = 0;
2250 *lvalp = (unsigned long)val;
2253 return 0;
2257 * Taint values can only be increased
2259 static int proc_dointvec_taint(struct ctl_table *table, int write, struct file *filp,
2260 void __user *buffer, size_t *lenp, loff_t *ppos)
2262 int op;
2264 if (write && !capable(CAP_SYS_ADMIN))
2265 return -EPERM;
2267 op = OP_OR;
2268 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2269 do_proc_dointvec_bset_conv,&op);
2272 struct do_proc_dointvec_minmax_conv_param {
2273 int *min;
2274 int *max;
2277 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2278 int *valp,
2279 int write, void *data)
2281 struct do_proc_dointvec_minmax_conv_param *param = data;
2282 if (write) {
2283 int val = *negp ? -*lvalp : *lvalp;
2284 if ((param->min && *param->min > val) ||
2285 (param->max && *param->max < val))
2286 return -EINVAL;
2287 *valp = val;
2288 } else {
2289 int val = *valp;
2290 if (val < 0) {
2291 *negp = -1;
2292 *lvalp = (unsigned long)-val;
2293 } else {
2294 *negp = 0;
2295 *lvalp = (unsigned long)val;
2298 return 0;
2302 * proc_dointvec_minmax - read a vector of integers with min/max values
2303 * @table: the sysctl table
2304 * @write: %TRUE if this is a write to the sysctl file
2305 * @filp: the file structure
2306 * @buffer: the user buffer
2307 * @lenp: the size of the user buffer
2308 * @ppos: file position
2310 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2311 * values from/to the user buffer, treated as an ASCII string.
2313 * This routine will ensure the values are within the range specified by
2314 * table->extra1 (min) and table->extra2 (max).
2316 * Returns 0 on success.
2318 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2319 void __user *buffer, size_t *lenp, loff_t *ppos)
2321 struct do_proc_dointvec_minmax_conv_param param = {
2322 .min = (int *) table->extra1,
2323 .max = (int *) table->extra2,
2325 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2326 do_proc_dointvec_minmax_conv, &param);
2329 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2330 struct file *filp,
2331 void __user *buffer,
2332 size_t *lenp, loff_t *ppos,
2333 unsigned long convmul,
2334 unsigned long convdiv)
2336 #define TMPBUFLEN 21
2337 unsigned long *i, *min, *max, val;
2338 int vleft, first=1, neg;
2339 size_t len, left;
2340 char buf[TMPBUFLEN], *p;
2341 char __user *s = buffer;
2343 if (!data || !table->maxlen || !*lenp ||
2344 (*ppos && !write)) {
2345 *lenp = 0;
2346 return 0;
2349 i = (unsigned long *) data;
2350 min = (unsigned long *) table->extra1;
2351 max = (unsigned long *) table->extra2;
2352 vleft = table->maxlen / sizeof(unsigned long);
2353 left = *lenp;
2355 for (; left && vleft--; i++, min++, max++, first=0) {
2356 if (write) {
2357 while (left) {
2358 char c;
2359 if (get_user(c, s))
2360 return -EFAULT;
2361 if (!isspace(c))
2362 break;
2363 left--;
2364 s++;
2366 if (!left)
2367 break;
2368 neg = 0;
2369 len = left;
2370 if (len > TMPBUFLEN-1)
2371 len = TMPBUFLEN-1;
2372 if (copy_from_user(buf, s, len))
2373 return -EFAULT;
2374 buf[len] = 0;
2375 p = buf;
2376 if (*p == '-' && left > 1) {
2377 neg = 1;
2378 p++;
2380 if (*p < '0' || *p > '9')
2381 break;
2382 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2383 len = p-buf;
2384 if ((len < left) && *p && !isspace(*p))
2385 break;
2386 if (neg)
2387 val = -val;
2388 s += len;
2389 left -= len;
2391 if(neg)
2392 continue;
2393 if ((min && val < *min) || (max && val > *max))
2394 continue;
2395 *i = val;
2396 } else {
2397 p = buf;
2398 if (!first)
2399 *p++ = '\t';
2400 sprintf(p, "%lu", convdiv * (*i) / convmul);
2401 len = strlen(buf);
2402 if (len > left)
2403 len = left;
2404 if(copy_to_user(s, buf, len))
2405 return -EFAULT;
2406 left -= len;
2407 s += len;
2411 if (!write && !first && left) {
2412 if(put_user('\n', s))
2413 return -EFAULT;
2414 left--, s++;
2416 if (write) {
2417 while (left) {
2418 char c;
2419 if (get_user(c, s++))
2420 return -EFAULT;
2421 if (!isspace(c))
2422 break;
2423 left--;
2426 if (write && first)
2427 return -EINVAL;
2428 *lenp -= left;
2429 *ppos += *lenp;
2430 return 0;
2431 #undef TMPBUFLEN
2434 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2435 struct file *filp,
2436 void __user *buffer,
2437 size_t *lenp, loff_t *ppos,
2438 unsigned long convmul,
2439 unsigned long convdiv)
2441 return __do_proc_doulongvec_minmax(table->data, table, write,
2442 filp, buffer, lenp, ppos, convmul, convdiv);
2446 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2447 * @table: the sysctl table
2448 * @write: %TRUE if this is a write to the sysctl file
2449 * @filp: the file structure
2450 * @buffer: the user buffer
2451 * @lenp: the size of the user buffer
2452 * @ppos: file position
2454 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2455 * values from/to the user buffer, treated as an ASCII string.
2457 * This routine will ensure the values are within the range specified by
2458 * table->extra1 (min) and table->extra2 (max).
2460 * Returns 0 on success.
2462 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2463 void __user *buffer, size_t *lenp, loff_t *ppos)
2465 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2469 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2470 * @table: the sysctl table
2471 * @write: %TRUE if this is a write to the sysctl file
2472 * @filp: the file structure
2473 * @buffer: the user buffer
2474 * @lenp: the size of the user buffer
2475 * @ppos: file position
2477 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2478 * values from/to the user buffer, treated as an ASCII string. The values
2479 * are treated as milliseconds, and converted to jiffies when they are stored.
2481 * This routine will ensure the values are within the range specified by
2482 * table->extra1 (min) and table->extra2 (max).
2484 * Returns 0 on success.
2486 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2487 struct file *filp,
2488 void __user *buffer,
2489 size_t *lenp, loff_t *ppos)
2491 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2492 lenp, ppos, HZ, 1000l);
2496 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2497 int *valp,
2498 int write, void *data)
2500 if (write) {
2501 if (*lvalp > LONG_MAX / HZ)
2502 return 1;
2503 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2504 } else {
2505 int val = *valp;
2506 unsigned long lval;
2507 if (val < 0) {
2508 *negp = -1;
2509 lval = (unsigned long)-val;
2510 } else {
2511 *negp = 0;
2512 lval = (unsigned long)val;
2514 *lvalp = lval / HZ;
2516 return 0;
2519 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2520 int *valp,
2521 int write, void *data)
2523 if (write) {
2524 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2525 return 1;
2526 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2527 } else {
2528 int val = *valp;
2529 unsigned long lval;
2530 if (val < 0) {
2531 *negp = -1;
2532 lval = (unsigned long)-val;
2533 } else {
2534 *negp = 0;
2535 lval = (unsigned long)val;
2537 *lvalp = jiffies_to_clock_t(lval);
2539 return 0;
2542 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2543 int *valp,
2544 int write, void *data)
2546 if (write) {
2547 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2548 } else {
2549 int val = *valp;
2550 unsigned long lval;
2551 if (val < 0) {
2552 *negp = -1;
2553 lval = (unsigned long)-val;
2554 } else {
2555 *negp = 0;
2556 lval = (unsigned long)val;
2558 *lvalp = jiffies_to_msecs(lval);
2560 return 0;
2564 * proc_dointvec_jiffies - read a vector of integers as seconds
2565 * @table: the sysctl table
2566 * @write: %TRUE if this is a write to the sysctl file
2567 * @filp: the file structure
2568 * @buffer: the user buffer
2569 * @lenp: the size of the user buffer
2570 * @ppos: file position
2572 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2573 * values from/to the user buffer, treated as an ASCII string.
2574 * The values read are assumed to be in seconds, and are converted into
2575 * jiffies.
2577 * Returns 0 on success.
2579 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2580 void __user *buffer, size_t *lenp, loff_t *ppos)
2582 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2583 do_proc_dointvec_jiffies_conv,NULL);
2587 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2588 * @table: the sysctl table
2589 * @write: %TRUE if this is a write to the sysctl file
2590 * @filp: the file structure
2591 * @buffer: the user buffer
2592 * @lenp: the size of the user buffer
2593 * @ppos: pointer to the file position
2595 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2596 * values from/to the user buffer, treated as an ASCII string.
2597 * The values read are assumed to be in 1/USER_HZ seconds, and
2598 * are converted into jiffies.
2600 * Returns 0 on success.
2602 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2603 void __user *buffer, size_t *lenp, loff_t *ppos)
2605 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2606 do_proc_dointvec_userhz_jiffies_conv,NULL);
2610 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2611 * @table: the sysctl table
2612 * @write: %TRUE if this is a write to the sysctl file
2613 * @filp: the file structure
2614 * @buffer: the user buffer
2615 * @lenp: the size of the user buffer
2616 * @ppos: file position
2617 * @ppos: the current position in the file
2619 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2620 * values from/to the user buffer, treated as an ASCII string.
2621 * The values read are assumed to be in 1/1000 seconds, and
2622 * are converted into jiffies.
2624 * Returns 0 on success.
2626 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2627 void __user *buffer, size_t *lenp, loff_t *ppos)
2629 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2630 do_proc_dointvec_ms_jiffies_conv, NULL);
2633 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2634 void __user *buffer, size_t *lenp, loff_t *ppos)
2636 struct pid *new_pid;
2637 pid_t tmp;
2638 int r;
2640 tmp = pid_vnr(cad_pid);
2642 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2643 lenp, ppos, NULL, NULL);
2644 if (r || !write)
2645 return r;
2647 new_pid = find_get_pid(tmp);
2648 if (!new_pid)
2649 return -ESRCH;
2651 put_pid(xchg(&cad_pid, new_pid));
2652 return 0;
2655 #else /* CONFIG_PROC_FS */
2657 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2658 void __user *buffer, size_t *lenp, loff_t *ppos)
2660 return -ENOSYS;
2663 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2664 void __user *buffer, size_t *lenp, loff_t *ppos)
2666 return -ENOSYS;
2669 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2670 void __user *buffer, size_t *lenp, loff_t *ppos)
2672 return -ENOSYS;
2675 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2676 void __user *buffer, size_t *lenp, loff_t *ppos)
2678 return -ENOSYS;
2681 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2682 void __user *buffer, size_t *lenp, loff_t *ppos)
2684 return -ENOSYS;
2687 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2688 void __user *buffer, size_t *lenp, loff_t *ppos)
2690 return -ENOSYS;
2693 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2694 void __user *buffer, size_t *lenp, loff_t *ppos)
2696 return -ENOSYS;
2699 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2700 struct file *filp,
2701 void __user *buffer,
2702 size_t *lenp, loff_t *ppos)
2704 return -ENOSYS;
2708 #endif /* CONFIG_PROC_FS */
2711 #ifdef CONFIG_SYSCTL_SYSCALL
2713 * General sysctl support routines
2716 /* The generic sysctl data routine (used if no strategy routine supplied) */
2717 int sysctl_data(struct ctl_table *table, int __user *name, int nlen,
2718 void __user *oldval, size_t __user *oldlenp,
2719 void __user *newval, size_t newlen)
2721 size_t len;
2723 /* Get out of I don't have a variable */
2724 if (!table->data || !table->maxlen)
2725 return -ENOTDIR;
2727 if (oldval && oldlenp) {
2728 if (get_user(len, oldlenp))
2729 return -EFAULT;
2730 if (len) {
2731 if (len > table->maxlen)
2732 len = table->maxlen;
2733 if (copy_to_user(oldval, table->data, len))
2734 return -EFAULT;
2735 if (put_user(len, oldlenp))
2736 return -EFAULT;
2740 if (newval && newlen) {
2741 if (newlen > table->maxlen)
2742 newlen = table->maxlen;
2744 if (copy_from_user(table->data, newval, newlen))
2745 return -EFAULT;
2747 return 1;
2750 /* The generic string strategy routine: */
2751 int sysctl_string(struct ctl_table *table, int __user *name, int nlen,
2752 void __user *oldval, size_t __user *oldlenp,
2753 void __user *newval, size_t newlen)
2755 if (!table->data || !table->maxlen)
2756 return -ENOTDIR;
2758 if (oldval && oldlenp) {
2759 size_t bufsize;
2760 if (get_user(bufsize, oldlenp))
2761 return -EFAULT;
2762 if (bufsize) {
2763 size_t len = strlen(table->data), copied;
2765 /* This shouldn't trigger for a well-formed sysctl */
2766 if (len > table->maxlen)
2767 len = table->maxlen;
2769 /* Copy up to a max of bufsize-1 bytes of the string */
2770 copied = (len >= bufsize) ? bufsize - 1 : len;
2772 if (copy_to_user(oldval, table->data, copied) ||
2773 put_user(0, (char __user *)(oldval + copied)))
2774 return -EFAULT;
2775 if (put_user(len, oldlenp))
2776 return -EFAULT;
2779 if (newval && newlen) {
2780 size_t len = newlen;
2781 if (len > table->maxlen)
2782 len = table->maxlen;
2783 if(copy_from_user(table->data, newval, len))
2784 return -EFAULT;
2785 if (len == table->maxlen)
2786 len--;
2787 ((char *) table->data)[len] = 0;
2789 return 1;
2793 * This function makes sure that all of the integers in the vector
2794 * are between the minimum and maximum values given in the arrays
2795 * table->extra1 and table->extra2, respectively.
2797 int sysctl_intvec(struct ctl_table *table, int __user *name, int nlen,
2798 void __user *oldval, size_t __user *oldlenp,
2799 void __user *newval, size_t newlen)
2802 if (newval && newlen) {
2803 int __user *vec = (int __user *) newval;
2804 int *min = (int *) table->extra1;
2805 int *max = (int *) table->extra2;
2806 size_t length;
2807 int i;
2809 if (newlen % sizeof(int) != 0)
2810 return -EINVAL;
2812 if (!table->extra1 && !table->extra2)
2813 return 0;
2815 if (newlen > table->maxlen)
2816 newlen = table->maxlen;
2817 length = newlen / sizeof(int);
2819 for (i = 0; i < length; i++) {
2820 int value;
2821 if (get_user(value, vec + i))
2822 return -EFAULT;
2823 if (min && value < min[i])
2824 return -EINVAL;
2825 if (max && value > max[i])
2826 return -EINVAL;
2829 return 0;
2832 /* Strategy function to convert jiffies to seconds */
2833 int sysctl_jiffies(struct ctl_table *table, int __user *name, int nlen,
2834 void __user *oldval, size_t __user *oldlenp,
2835 void __user *newval, size_t newlen)
2837 if (oldval && oldlenp) {
2838 size_t olen;
2840 if (get_user(olen, oldlenp))
2841 return -EFAULT;
2842 if (olen) {
2843 int val;
2845 if (olen < sizeof(int))
2846 return -EINVAL;
2848 val = *(int *)(table->data) / HZ;
2849 if (put_user(val, (int __user *)oldval))
2850 return -EFAULT;
2851 if (put_user(sizeof(int), oldlenp))
2852 return -EFAULT;
2855 if (newval && newlen) {
2856 int new;
2857 if (newlen != sizeof(int))
2858 return -EINVAL;
2859 if (get_user(new, (int __user *)newval))
2860 return -EFAULT;
2861 *(int *)(table->data) = new*HZ;
2863 return 1;
2866 /* Strategy function to convert jiffies to seconds */
2867 int sysctl_ms_jiffies(struct ctl_table *table, int __user *name, int nlen,
2868 void __user *oldval, size_t __user *oldlenp,
2869 void __user *newval, size_t newlen)
2871 if (oldval && oldlenp) {
2872 size_t olen;
2874 if (get_user(olen, oldlenp))
2875 return -EFAULT;
2876 if (olen) {
2877 int val;
2879 if (olen < sizeof(int))
2880 return -EINVAL;
2882 val = jiffies_to_msecs(*(int *)(table->data));
2883 if (put_user(val, (int __user *)oldval))
2884 return -EFAULT;
2885 if (put_user(sizeof(int), oldlenp))
2886 return -EFAULT;
2889 if (newval && newlen) {
2890 int new;
2891 if (newlen != sizeof(int))
2892 return -EINVAL;
2893 if (get_user(new, (int __user *)newval))
2894 return -EFAULT;
2895 *(int *)(table->data) = msecs_to_jiffies(new);
2897 return 1;
2902 #else /* CONFIG_SYSCTL_SYSCALL */
2905 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2907 struct __sysctl_args tmp;
2908 int error;
2910 if (copy_from_user(&tmp, args, sizeof(tmp)))
2911 return -EFAULT;
2913 error = deprecated_sysctl_warning(&tmp);
2915 /* If no error reading the parameters then just -ENOSYS ... */
2916 if (!error)
2917 error = -ENOSYS;
2919 return error;
2922 int sysctl_data(struct ctl_table *table, int __user *name, int nlen,
2923 void __user *oldval, size_t __user *oldlenp,
2924 void __user *newval, size_t newlen)
2926 return -ENOSYS;
2929 int sysctl_string(struct ctl_table *table, int __user *name, int nlen,
2930 void __user *oldval, size_t __user *oldlenp,
2931 void __user *newval, size_t newlen)
2933 return -ENOSYS;
2936 int sysctl_intvec(struct ctl_table *table, int __user *name, int nlen,
2937 void __user *oldval, size_t __user *oldlenp,
2938 void __user *newval, size_t newlen)
2940 return -ENOSYS;
2943 int sysctl_jiffies(struct ctl_table *table, int __user *name, int nlen,
2944 void __user *oldval, size_t __user *oldlenp,
2945 void __user *newval, size_t newlen)
2947 return -ENOSYS;
2950 int sysctl_ms_jiffies(struct ctl_table *table, int __user *name, int nlen,
2951 void __user *oldval, size_t __user *oldlenp,
2952 void __user *newval, size_t newlen)
2954 return -ENOSYS;
2957 #endif /* CONFIG_SYSCTL_SYSCALL */
2959 static int deprecated_sysctl_warning(struct __sysctl_args *args)
2961 static int msg_count;
2962 int name[CTL_MAXNAME];
2963 int i;
2965 /* Check args->nlen. */
2966 if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
2967 return -ENOTDIR;
2969 /* Read in the sysctl name for better debug message logging */
2970 for (i = 0; i < args->nlen; i++)
2971 if (get_user(name[i], args->name + i))
2972 return -EFAULT;
2974 /* Ignore accesses to kernel.version */
2975 if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2976 return 0;
2978 if (msg_count < 5) {
2979 msg_count++;
2980 printk(KERN_INFO
2981 "warning: process `%s' used the deprecated sysctl "
2982 "system call with ", current->comm);
2983 for (i = 0; i < args->nlen; i++)
2984 printk("%d.", name[i]);
2985 printk("\n");
2987 return 0;
2991 * No sense putting this after each symbol definition, twice,
2992 * exception granted :-)
2994 EXPORT_SYMBOL(proc_dointvec);
2995 EXPORT_SYMBOL(proc_dointvec_jiffies);
2996 EXPORT_SYMBOL(proc_dointvec_minmax);
2997 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2998 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2999 EXPORT_SYMBOL(proc_dostring);
3000 EXPORT_SYMBOL(proc_doulongvec_minmax);
3001 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3002 EXPORT_SYMBOL(register_sysctl_table);
3003 EXPORT_SYMBOL(register_sysctl_paths);
3004 EXPORT_SYMBOL(sysctl_intvec);
3005 EXPORT_SYMBOL(sysctl_jiffies);
3006 EXPORT_SYMBOL(sysctl_ms_jiffies);
3007 EXPORT_SYMBOL(sysctl_string);
3008 EXPORT_SYMBOL(sysctl_data);
3009 EXPORT_SYMBOL(unregister_sysctl_table);