powerpc/32: Allow __ioremap on RAM addresses for kdump kernel
[linux-2.6/mini2440.git] / kernel / sysctl.c
blob3d56fe7570daede300fe01f2e6571416107b081b
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 #if defined(CONFIG_MMU) && defined(CONFIG_FILE_LOCKING)
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_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
179 #ifdef CONFIG_EPOLL
180 extern struct ctl_table epoll_table[];
181 #endif
183 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
184 int sysctl_legacy_va_layout;
185 #endif
187 extern int prove_locking;
188 extern int lock_stat;
190 /* The default sysctl tables: */
192 static struct ctl_table root_table[] = {
194 .ctl_name = CTL_KERN,
195 .procname = "kernel",
196 .mode = 0555,
197 .child = kern_table,
200 .ctl_name = CTL_VM,
201 .procname = "vm",
202 .mode = 0555,
203 .child = vm_table,
206 .ctl_name = CTL_FS,
207 .procname = "fs",
208 .mode = 0555,
209 .child = fs_table,
212 .ctl_name = CTL_DEBUG,
213 .procname = "debug",
214 .mode = 0555,
215 .child = debug_table,
218 .ctl_name = CTL_DEV,
219 .procname = "dev",
220 .mode = 0555,
221 .child = dev_table,
224 * NOTE: do not add new entries to this table unless you have read
225 * Documentation/sysctl/ctl_unnumbered.txt
227 { .ctl_name = 0 }
230 #ifdef CONFIG_SCHED_DEBUG
231 static int min_sched_granularity_ns = 100000; /* 100 usecs */
232 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
233 static int min_wakeup_granularity_ns; /* 0 usecs */
234 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
235 #endif
237 static struct ctl_table kern_table[] = {
238 #ifdef CONFIG_SCHED_DEBUG
240 .ctl_name = CTL_UNNUMBERED,
241 .procname = "sched_min_granularity_ns",
242 .data = &sysctl_sched_min_granularity,
243 .maxlen = sizeof(unsigned int),
244 .mode = 0644,
245 .proc_handler = &sched_nr_latency_handler,
246 .strategy = &sysctl_intvec,
247 .extra1 = &min_sched_granularity_ns,
248 .extra2 = &max_sched_granularity_ns,
251 .ctl_name = CTL_UNNUMBERED,
252 .procname = "sched_latency_ns",
253 .data = &sysctl_sched_latency,
254 .maxlen = sizeof(unsigned int),
255 .mode = 0644,
256 .proc_handler = &sched_nr_latency_handler,
257 .strategy = &sysctl_intvec,
258 .extra1 = &min_sched_granularity_ns,
259 .extra2 = &max_sched_granularity_ns,
262 .ctl_name = CTL_UNNUMBERED,
263 .procname = "sched_wakeup_granularity_ns",
264 .data = &sysctl_sched_wakeup_granularity,
265 .maxlen = sizeof(unsigned int),
266 .mode = 0644,
267 .proc_handler = &proc_dointvec_minmax,
268 .strategy = &sysctl_intvec,
269 .extra1 = &min_wakeup_granularity_ns,
270 .extra2 = &max_wakeup_granularity_ns,
273 .ctl_name = CTL_UNNUMBERED,
274 .procname = "sched_shares_ratelimit",
275 .data = &sysctl_sched_shares_ratelimit,
276 .maxlen = sizeof(unsigned int),
277 .mode = 0644,
278 .proc_handler = &proc_dointvec,
281 .ctl_name = CTL_UNNUMBERED,
282 .procname = "sched_shares_thresh",
283 .data = &sysctl_sched_shares_thresh,
284 .maxlen = sizeof(unsigned int),
285 .mode = 0644,
286 .proc_handler = &proc_dointvec_minmax,
287 .strategy = &sysctl_intvec,
288 .extra1 = &zero,
291 .ctl_name = CTL_UNNUMBERED,
292 .procname = "sched_child_runs_first",
293 .data = &sysctl_sched_child_runs_first,
294 .maxlen = sizeof(unsigned int),
295 .mode = 0644,
296 .proc_handler = &proc_dointvec,
299 .ctl_name = CTL_UNNUMBERED,
300 .procname = "sched_features",
301 .data = &sysctl_sched_features,
302 .maxlen = sizeof(unsigned int),
303 .mode = 0644,
304 .proc_handler = &proc_dointvec,
307 .ctl_name = CTL_UNNUMBERED,
308 .procname = "sched_migration_cost",
309 .data = &sysctl_sched_migration_cost,
310 .maxlen = sizeof(unsigned int),
311 .mode = 0644,
312 .proc_handler = &proc_dointvec,
315 .ctl_name = CTL_UNNUMBERED,
316 .procname = "sched_nr_migrate",
317 .data = &sysctl_sched_nr_migrate,
318 .maxlen = sizeof(unsigned int),
319 .mode = 0644,
320 .proc_handler = &proc_dointvec,
322 #endif
324 .ctl_name = CTL_UNNUMBERED,
325 .procname = "sched_rt_period_us",
326 .data = &sysctl_sched_rt_period,
327 .maxlen = sizeof(unsigned int),
328 .mode = 0644,
329 .proc_handler = &sched_rt_handler,
332 .ctl_name = CTL_UNNUMBERED,
333 .procname = "sched_rt_runtime_us",
334 .data = &sysctl_sched_rt_runtime,
335 .maxlen = sizeof(int),
336 .mode = 0644,
337 .proc_handler = &sched_rt_handler,
340 .ctl_name = CTL_UNNUMBERED,
341 .procname = "sched_compat_yield",
342 .data = &sysctl_sched_compat_yield,
343 .maxlen = sizeof(unsigned int),
344 .mode = 0644,
345 .proc_handler = &proc_dointvec,
347 #ifdef CONFIG_PROVE_LOCKING
349 .ctl_name = CTL_UNNUMBERED,
350 .procname = "prove_locking",
351 .data = &prove_locking,
352 .maxlen = sizeof(int),
353 .mode = 0644,
354 .proc_handler = &proc_dointvec,
356 #endif
357 #ifdef CONFIG_LOCK_STAT
359 .ctl_name = CTL_UNNUMBERED,
360 .procname = "lock_stat",
361 .data = &lock_stat,
362 .maxlen = sizeof(int),
363 .mode = 0644,
364 .proc_handler = &proc_dointvec,
366 #endif
368 .ctl_name = KERN_PANIC,
369 .procname = "panic",
370 .data = &panic_timeout,
371 .maxlen = sizeof(int),
372 .mode = 0644,
373 .proc_handler = &proc_dointvec,
376 .ctl_name = KERN_CORE_USES_PID,
377 .procname = "core_uses_pid",
378 .data = &core_uses_pid,
379 .maxlen = sizeof(int),
380 .mode = 0644,
381 .proc_handler = &proc_dointvec,
384 .ctl_name = KERN_CORE_PATTERN,
385 .procname = "core_pattern",
386 .data = core_pattern,
387 .maxlen = CORENAME_MAX_SIZE,
388 .mode = 0644,
389 .proc_handler = &proc_dostring,
390 .strategy = &sysctl_string,
392 #ifdef CONFIG_PROC_SYSCTL
394 .procname = "tainted",
395 .maxlen = sizeof(long),
396 .mode = 0644,
397 .proc_handler = &proc_taint,
399 #endif
400 #ifdef CONFIG_LATENCYTOP
402 .procname = "latencytop",
403 .data = &latencytop_enabled,
404 .maxlen = sizeof(int),
405 .mode = 0644,
406 .proc_handler = &proc_dointvec,
408 #endif
409 #ifdef CONFIG_BLK_DEV_INITRD
411 .ctl_name = KERN_REALROOTDEV,
412 .procname = "real-root-dev",
413 .data = &real_root_dev,
414 .maxlen = sizeof(int),
415 .mode = 0644,
416 .proc_handler = &proc_dointvec,
418 #endif
420 .ctl_name = CTL_UNNUMBERED,
421 .procname = "print-fatal-signals",
422 .data = &print_fatal_signals,
423 .maxlen = sizeof(int),
424 .mode = 0644,
425 .proc_handler = &proc_dointvec,
427 #ifdef CONFIG_SPARC
429 .ctl_name = KERN_SPARC_REBOOT,
430 .procname = "reboot-cmd",
431 .data = reboot_command,
432 .maxlen = 256,
433 .mode = 0644,
434 .proc_handler = &proc_dostring,
435 .strategy = &sysctl_string,
438 .ctl_name = KERN_SPARC_STOP_A,
439 .procname = "stop-a",
440 .data = &stop_a_enabled,
441 .maxlen = sizeof (int),
442 .mode = 0644,
443 .proc_handler = &proc_dointvec,
446 .ctl_name = KERN_SPARC_SCONS_PWROFF,
447 .procname = "scons-poweroff",
448 .data = &scons_pwroff,
449 .maxlen = sizeof (int),
450 .mode = 0644,
451 .proc_handler = &proc_dointvec,
453 #endif
454 #ifdef __hppa__
456 .ctl_name = KERN_HPPA_PWRSW,
457 .procname = "soft-power",
458 .data = &pwrsw_enabled,
459 .maxlen = sizeof (int),
460 .mode = 0644,
461 .proc_handler = &proc_dointvec,
464 .ctl_name = KERN_HPPA_UNALIGNED,
465 .procname = "unaligned-trap",
466 .data = &unaligned_enabled,
467 .maxlen = sizeof (int),
468 .mode = 0644,
469 .proc_handler = &proc_dointvec,
471 #endif
473 .ctl_name = KERN_CTLALTDEL,
474 .procname = "ctrl-alt-del",
475 .data = &C_A_D,
476 .maxlen = sizeof(int),
477 .mode = 0644,
478 .proc_handler = &proc_dointvec,
480 #ifdef CONFIG_FUNCTION_TRACER
482 .ctl_name = CTL_UNNUMBERED,
483 .procname = "ftrace_enabled",
484 .data = &ftrace_enabled,
485 .maxlen = sizeof(int),
486 .mode = 0644,
487 .proc_handler = &ftrace_enable_sysctl,
489 #endif
490 #ifdef CONFIG_MODULES
492 .ctl_name = KERN_MODPROBE,
493 .procname = "modprobe",
494 .data = &modprobe_path,
495 .maxlen = KMOD_PATH_LEN,
496 .mode = 0644,
497 .proc_handler = &proc_dostring,
498 .strategy = &sysctl_string,
500 #endif
501 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
503 .ctl_name = KERN_HOTPLUG,
504 .procname = "hotplug",
505 .data = &uevent_helper,
506 .maxlen = UEVENT_HELPER_PATH_LEN,
507 .mode = 0644,
508 .proc_handler = &proc_dostring,
509 .strategy = &sysctl_string,
511 #endif
512 #ifdef CONFIG_CHR_DEV_SG
514 .ctl_name = KERN_SG_BIG_BUFF,
515 .procname = "sg-big-buff",
516 .data = &sg_big_buff,
517 .maxlen = sizeof (int),
518 .mode = 0444,
519 .proc_handler = &proc_dointvec,
521 #endif
522 #ifdef CONFIG_BSD_PROCESS_ACCT
524 .ctl_name = KERN_ACCT,
525 .procname = "acct",
526 .data = &acct_parm,
527 .maxlen = 3*sizeof(int),
528 .mode = 0644,
529 .proc_handler = &proc_dointvec,
531 #endif
532 #ifdef CONFIG_MAGIC_SYSRQ
534 .ctl_name = KERN_SYSRQ,
535 .procname = "sysrq",
536 .data = &__sysrq_enabled,
537 .maxlen = sizeof (int),
538 .mode = 0644,
539 .proc_handler = &proc_dointvec,
541 #endif
542 #ifdef CONFIG_PROC_SYSCTL
544 .procname = "cad_pid",
545 .data = NULL,
546 .maxlen = sizeof (int),
547 .mode = 0600,
548 .proc_handler = &proc_do_cad_pid,
550 #endif
552 .ctl_name = KERN_MAX_THREADS,
553 .procname = "threads-max",
554 .data = &max_threads,
555 .maxlen = sizeof(int),
556 .mode = 0644,
557 .proc_handler = &proc_dointvec,
560 .ctl_name = KERN_RANDOM,
561 .procname = "random",
562 .mode = 0555,
563 .child = random_table,
566 .ctl_name = KERN_OVERFLOWUID,
567 .procname = "overflowuid",
568 .data = &overflowuid,
569 .maxlen = sizeof(int),
570 .mode = 0644,
571 .proc_handler = &proc_dointvec_minmax,
572 .strategy = &sysctl_intvec,
573 .extra1 = &minolduid,
574 .extra2 = &maxolduid,
577 .ctl_name = KERN_OVERFLOWGID,
578 .procname = "overflowgid",
579 .data = &overflowgid,
580 .maxlen = sizeof(int),
581 .mode = 0644,
582 .proc_handler = &proc_dointvec_minmax,
583 .strategy = &sysctl_intvec,
584 .extra1 = &minolduid,
585 .extra2 = &maxolduid,
587 #ifdef CONFIG_S390
588 #ifdef CONFIG_MATHEMU
590 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
591 .procname = "ieee_emulation_warnings",
592 .data = &sysctl_ieee_emulation_warnings,
593 .maxlen = sizeof(int),
594 .mode = 0644,
595 .proc_handler = &proc_dointvec,
597 #endif
599 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
600 .procname = "userprocess_debug",
601 .data = &sysctl_userprocess_debug,
602 .maxlen = sizeof(int),
603 .mode = 0644,
604 .proc_handler = &proc_dointvec,
606 #endif
608 .ctl_name = KERN_PIDMAX,
609 .procname = "pid_max",
610 .data = &pid_max,
611 .maxlen = sizeof (int),
612 .mode = 0644,
613 .proc_handler = &proc_dointvec_minmax,
614 .strategy = sysctl_intvec,
615 .extra1 = &pid_max_min,
616 .extra2 = &pid_max_max,
619 .ctl_name = KERN_PANIC_ON_OOPS,
620 .procname = "panic_on_oops",
621 .data = &panic_on_oops,
622 .maxlen = sizeof(int),
623 .mode = 0644,
624 .proc_handler = &proc_dointvec,
626 #if defined CONFIG_PRINTK
628 .ctl_name = KERN_PRINTK,
629 .procname = "printk",
630 .data = &console_loglevel,
631 .maxlen = 4*sizeof(int),
632 .mode = 0644,
633 .proc_handler = &proc_dointvec,
636 .ctl_name = KERN_PRINTK_RATELIMIT,
637 .procname = "printk_ratelimit",
638 .data = &printk_ratelimit_state.interval,
639 .maxlen = sizeof(int),
640 .mode = 0644,
641 .proc_handler = &proc_dointvec_jiffies,
642 .strategy = &sysctl_jiffies,
645 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
646 .procname = "printk_ratelimit_burst",
647 .data = &printk_ratelimit_state.burst,
648 .maxlen = sizeof(int),
649 .mode = 0644,
650 .proc_handler = &proc_dointvec,
652 #endif
654 .ctl_name = KERN_NGROUPS_MAX,
655 .procname = "ngroups_max",
656 .data = &ngroups_max,
657 .maxlen = sizeof (int),
658 .mode = 0444,
659 .proc_handler = &proc_dointvec,
661 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
663 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
664 .procname = "unknown_nmi_panic",
665 .data = &unknown_nmi_panic,
666 .maxlen = sizeof (int),
667 .mode = 0644,
668 .proc_handler = &proc_dointvec,
671 .procname = "nmi_watchdog",
672 .data = &nmi_watchdog_enabled,
673 .maxlen = sizeof (int),
674 .mode = 0644,
675 .proc_handler = &proc_nmi_enabled,
677 #endif
678 #if defined(CONFIG_X86)
680 .ctl_name = KERN_PANIC_ON_NMI,
681 .procname = "panic_on_unrecovered_nmi",
682 .data = &panic_on_unrecovered_nmi,
683 .maxlen = sizeof(int),
684 .mode = 0644,
685 .proc_handler = &proc_dointvec,
688 .ctl_name = KERN_BOOTLOADER_TYPE,
689 .procname = "bootloader_type",
690 .data = &bootloader_type,
691 .maxlen = sizeof (int),
692 .mode = 0444,
693 .proc_handler = &proc_dointvec,
696 .ctl_name = CTL_UNNUMBERED,
697 .procname = "kstack_depth_to_print",
698 .data = &kstack_depth_to_print,
699 .maxlen = sizeof(int),
700 .mode = 0644,
701 .proc_handler = &proc_dointvec,
704 .ctl_name = CTL_UNNUMBERED,
705 .procname = "io_delay_type",
706 .data = &io_delay_type,
707 .maxlen = sizeof(int),
708 .mode = 0644,
709 .proc_handler = &proc_dointvec,
711 #endif
712 #if defined(CONFIG_MMU)
714 .ctl_name = KERN_RANDOMIZE,
715 .procname = "randomize_va_space",
716 .data = &randomize_va_space,
717 .maxlen = sizeof(int),
718 .mode = 0644,
719 .proc_handler = &proc_dointvec,
721 #endif
722 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
724 .ctl_name = KERN_SPIN_RETRY,
725 .procname = "spin_retry",
726 .data = &spin_retry,
727 .maxlen = sizeof (int),
728 .mode = 0644,
729 .proc_handler = &proc_dointvec,
731 #endif
732 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
734 .procname = "acpi_video_flags",
735 .data = &acpi_realmode_flags,
736 .maxlen = sizeof (unsigned long),
737 .mode = 0644,
738 .proc_handler = &proc_doulongvec_minmax,
740 #endif
741 #ifdef CONFIG_IA64
743 .ctl_name = KERN_IA64_UNALIGNED,
744 .procname = "ignore-unaligned-usertrap",
745 .data = &no_unaligned_warning,
746 .maxlen = sizeof (int),
747 .mode = 0644,
748 .proc_handler = &proc_dointvec,
750 #endif
751 #ifdef CONFIG_DETECT_SOFTLOCKUP
753 .ctl_name = CTL_UNNUMBERED,
754 .procname = "softlockup_panic",
755 .data = &softlockup_panic,
756 .maxlen = sizeof(int),
757 .mode = 0644,
758 .proc_handler = &proc_dointvec_minmax,
759 .strategy = &sysctl_intvec,
760 .extra1 = &zero,
761 .extra2 = &one,
764 .ctl_name = CTL_UNNUMBERED,
765 .procname = "softlockup_thresh",
766 .data = &softlockup_thresh,
767 .maxlen = sizeof(int),
768 .mode = 0644,
769 .proc_handler = &proc_dointvec_minmax,
770 .strategy = &sysctl_intvec,
771 .extra1 = &neg_one,
772 .extra2 = &sixty,
775 .ctl_name = CTL_UNNUMBERED,
776 .procname = "hung_task_check_count",
777 .data = &sysctl_hung_task_check_count,
778 .maxlen = sizeof(unsigned long),
779 .mode = 0644,
780 .proc_handler = &proc_doulongvec_minmax,
781 .strategy = &sysctl_intvec,
784 .ctl_name = CTL_UNNUMBERED,
785 .procname = "hung_task_timeout_secs",
786 .data = &sysctl_hung_task_timeout_secs,
787 .maxlen = sizeof(unsigned long),
788 .mode = 0644,
789 .proc_handler = &proc_doulongvec_minmax,
790 .strategy = &sysctl_intvec,
793 .ctl_name = CTL_UNNUMBERED,
794 .procname = "hung_task_warnings",
795 .data = &sysctl_hung_task_warnings,
796 .maxlen = sizeof(unsigned long),
797 .mode = 0644,
798 .proc_handler = &proc_doulongvec_minmax,
799 .strategy = &sysctl_intvec,
801 #endif
802 #ifdef CONFIG_COMPAT
804 .ctl_name = KERN_COMPAT_LOG,
805 .procname = "compat-log",
806 .data = &compat_log,
807 .maxlen = sizeof (int),
808 .mode = 0644,
809 .proc_handler = &proc_dointvec,
811 #endif
812 #ifdef CONFIG_RT_MUTEXES
814 .ctl_name = KERN_MAX_LOCK_DEPTH,
815 .procname = "max_lock_depth",
816 .data = &max_lock_depth,
817 .maxlen = sizeof(int),
818 .mode = 0644,
819 .proc_handler = &proc_dointvec,
821 #endif
823 .ctl_name = CTL_UNNUMBERED,
824 .procname = "poweroff_cmd",
825 .data = &poweroff_cmd,
826 .maxlen = POWEROFF_CMD_PATH_LEN,
827 .mode = 0644,
828 .proc_handler = &proc_dostring,
829 .strategy = &sysctl_string,
831 #ifdef CONFIG_KEYS
833 .ctl_name = CTL_UNNUMBERED,
834 .procname = "keys",
835 .mode = 0555,
836 .child = key_sysctls,
838 #endif
839 #ifdef CONFIG_RCU_TORTURE_TEST
841 .ctl_name = CTL_UNNUMBERED,
842 .procname = "rcutorture_runnable",
843 .data = &rcutorture_runnable,
844 .maxlen = sizeof(int),
845 .mode = 0644,
846 .proc_handler = &proc_dointvec,
848 #endif
849 #ifdef CONFIG_UNEVICTABLE_LRU
851 .ctl_name = CTL_UNNUMBERED,
852 .procname = "scan_unevictable_pages",
853 .data = &scan_unevictable_pages,
854 .maxlen = sizeof(scan_unevictable_pages),
855 .mode = 0644,
856 .proc_handler = &scan_unevictable_handler,
858 #endif
860 * NOTE: do not add new entries to this table unless you have read
861 * Documentation/sysctl/ctl_unnumbered.txt
863 { .ctl_name = 0 }
866 static struct ctl_table vm_table[] = {
868 .ctl_name = VM_OVERCOMMIT_MEMORY,
869 .procname = "overcommit_memory",
870 .data = &sysctl_overcommit_memory,
871 .maxlen = sizeof(sysctl_overcommit_memory),
872 .mode = 0644,
873 .proc_handler = &proc_dointvec,
876 .ctl_name = VM_PANIC_ON_OOM,
877 .procname = "panic_on_oom",
878 .data = &sysctl_panic_on_oom,
879 .maxlen = sizeof(sysctl_panic_on_oom),
880 .mode = 0644,
881 .proc_handler = &proc_dointvec,
884 .ctl_name = CTL_UNNUMBERED,
885 .procname = "oom_kill_allocating_task",
886 .data = &sysctl_oom_kill_allocating_task,
887 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
888 .mode = 0644,
889 .proc_handler = &proc_dointvec,
892 .ctl_name = CTL_UNNUMBERED,
893 .procname = "oom_dump_tasks",
894 .data = &sysctl_oom_dump_tasks,
895 .maxlen = sizeof(sysctl_oom_dump_tasks),
896 .mode = 0644,
897 .proc_handler = &proc_dointvec,
900 .ctl_name = VM_OVERCOMMIT_RATIO,
901 .procname = "overcommit_ratio",
902 .data = &sysctl_overcommit_ratio,
903 .maxlen = sizeof(sysctl_overcommit_ratio),
904 .mode = 0644,
905 .proc_handler = &proc_dointvec,
908 .ctl_name = VM_PAGE_CLUSTER,
909 .procname = "page-cluster",
910 .data = &page_cluster,
911 .maxlen = sizeof(int),
912 .mode = 0644,
913 .proc_handler = &proc_dointvec,
916 .ctl_name = VM_DIRTY_BACKGROUND,
917 .procname = "dirty_background_ratio",
918 .data = &dirty_background_ratio,
919 .maxlen = sizeof(dirty_background_ratio),
920 .mode = 0644,
921 .proc_handler = &proc_dointvec_minmax,
922 .strategy = &sysctl_intvec,
923 .extra1 = &zero,
924 .extra2 = &one_hundred,
927 .ctl_name = VM_DIRTY_RATIO,
928 .procname = "dirty_ratio",
929 .data = &vm_dirty_ratio,
930 .maxlen = sizeof(vm_dirty_ratio),
931 .mode = 0644,
932 .proc_handler = &dirty_ratio_handler,
933 .strategy = &sysctl_intvec,
934 .extra1 = &zero,
935 .extra2 = &one_hundred,
938 .procname = "dirty_writeback_centisecs",
939 .data = &dirty_writeback_interval,
940 .maxlen = sizeof(dirty_writeback_interval),
941 .mode = 0644,
942 .proc_handler = &dirty_writeback_centisecs_handler,
945 .procname = "dirty_expire_centisecs",
946 .data = &dirty_expire_interval,
947 .maxlen = sizeof(dirty_expire_interval),
948 .mode = 0644,
949 .proc_handler = &proc_dointvec_userhz_jiffies,
952 .ctl_name = VM_NR_PDFLUSH_THREADS,
953 .procname = "nr_pdflush_threads",
954 .data = &nr_pdflush_threads,
955 .maxlen = sizeof nr_pdflush_threads,
956 .mode = 0444 /* read-only*/,
957 .proc_handler = &proc_dointvec,
960 .ctl_name = VM_SWAPPINESS,
961 .procname = "swappiness",
962 .data = &vm_swappiness,
963 .maxlen = sizeof(vm_swappiness),
964 .mode = 0644,
965 .proc_handler = &proc_dointvec_minmax,
966 .strategy = &sysctl_intvec,
967 .extra1 = &zero,
968 .extra2 = &one_hundred,
970 #ifdef CONFIG_HUGETLB_PAGE
972 .procname = "nr_hugepages",
973 .data = NULL,
974 .maxlen = sizeof(unsigned long),
975 .mode = 0644,
976 .proc_handler = &hugetlb_sysctl_handler,
977 .extra1 = (void *)&hugetlb_zero,
978 .extra2 = (void *)&hugetlb_infinity,
981 .ctl_name = VM_HUGETLB_GROUP,
982 .procname = "hugetlb_shm_group",
983 .data = &sysctl_hugetlb_shm_group,
984 .maxlen = sizeof(gid_t),
985 .mode = 0644,
986 .proc_handler = &proc_dointvec,
989 .ctl_name = CTL_UNNUMBERED,
990 .procname = "hugepages_treat_as_movable",
991 .data = &hugepages_treat_as_movable,
992 .maxlen = sizeof(int),
993 .mode = 0644,
994 .proc_handler = &hugetlb_treat_movable_handler,
997 .ctl_name = CTL_UNNUMBERED,
998 .procname = "nr_overcommit_hugepages",
999 .data = NULL,
1000 .maxlen = sizeof(unsigned long),
1001 .mode = 0644,
1002 .proc_handler = &hugetlb_overcommit_handler,
1003 .extra1 = (void *)&hugetlb_zero,
1004 .extra2 = (void *)&hugetlb_infinity,
1006 #endif
1008 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
1009 .procname = "lowmem_reserve_ratio",
1010 .data = &sysctl_lowmem_reserve_ratio,
1011 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1012 .mode = 0644,
1013 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
1014 .strategy = &sysctl_intvec,
1017 .ctl_name = VM_DROP_PAGECACHE,
1018 .procname = "drop_caches",
1019 .data = &sysctl_drop_caches,
1020 .maxlen = sizeof(int),
1021 .mode = 0644,
1022 .proc_handler = drop_caches_sysctl_handler,
1023 .strategy = &sysctl_intvec,
1026 .ctl_name = VM_MIN_FREE_KBYTES,
1027 .procname = "min_free_kbytes",
1028 .data = &min_free_kbytes,
1029 .maxlen = sizeof(min_free_kbytes),
1030 .mode = 0644,
1031 .proc_handler = &min_free_kbytes_sysctl_handler,
1032 .strategy = &sysctl_intvec,
1033 .extra1 = &zero,
1036 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
1037 .procname = "percpu_pagelist_fraction",
1038 .data = &percpu_pagelist_fraction,
1039 .maxlen = sizeof(percpu_pagelist_fraction),
1040 .mode = 0644,
1041 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
1042 .strategy = &sysctl_intvec,
1043 .extra1 = &min_percpu_pagelist_fract,
1045 #ifdef CONFIG_MMU
1047 .ctl_name = VM_MAX_MAP_COUNT,
1048 .procname = "max_map_count",
1049 .data = &sysctl_max_map_count,
1050 .maxlen = sizeof(sysctl_max_map_count),
1051 .mode = 0644,
1052 .proc_handler = &proc_dointvec
1054 #endif
1056 .ctl_name = VM_LAPTOP_MODE,
1057 .procname = "laptop_mode",
1058 .data = &laptop_mode,
1059 .maxlen = sizeof(laptop_mode),
1060 .mode = 0644,
1061 .proc_handler = &proc_dointvec_jiffies,
1062 .strategy = &sysctl_jiffies,
1065 .ctl_name = VM_BLOCK_DUMP,
1066 .procname = "block_dump",
1067 .data = &block_dump,
1068 .maxlen = sizeof(block_dump),
1069 .mode = 0644,
1070 .proc_handler = &proc_dointvec,
1071 .strategy = &sysctl_intvec,
1072 .extra1 = &zero,
1075 .ctl_name = VM_VFS_CACHE_PRESSURE,
1076 .procname = "vfs_cache_pressure",
1077 .data = &sysctl_vfs_cache_pressure,
1078 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1079 .mode = 0644,
1080 .proc_handler = &proc_dointvec,
1081 .strategy = &sysctl_intvec,
1082 .extra1 = &zero,
1084 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1086 .ctl_name = VM_LEGACY_VA_LAYOUT,
1087 .procname = "legacy_va_layout",
1088 .data = &sysctl_legacy_va_layout,
1089 .maxlen = sizeof(sysctl_legacy_va_layout),
1090 .mode = 0644,
1091 .proc_handler = &proc_dointvec,
1092 .strategy = &sysctl_intvec,
1093 .extra1 = &zero,
1095 #endif
1096 #ifdef CONFIG_NUMA
1098 .ctl_name = VM_ZONE_RECLAIM_MODE,
1099 .procname = "zone_reclaim_mode",
1100 .data = &zone_reclaim_mode,
1101 .maxlen = sizeof(zone_reclaim_mode),
1102 .mode = 0644,
1103 .proc_handler = &proc_dointvec,
1104 .strategy = &sysctl_intvec,
1105 .extra1 = &zero,
1108 .ctl_name = VM_MIN_UNMAPPED,
1109 .procname = "min_unmapped_ratio",
1110 .data = &sysctl_min_unmapped_ratio,
1111 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1112 .mode = 0644,
1113 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
1114 .strategy = &sysctl_intvec,
1115 .extra1 = &zero,
1116 .extra2 = &one_hundred,
1119 .ctl_name = VM_MIN_SLAB,
1120 .procname = "min_slab_ratio",
1121 .data = &sysctl_min_slab_ratio,
1122 .maxlen = sizeof(sysctl_min_slab_ratio),
1123 .mode = 0644,
1124 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1125 .strategy = &sysctl_intvec,
1126 .extra1 = &zero,
1127 .extra2 = &one_hundred,
1129 #endif
1130 #ifdef CONFIG_SMP
1132 .ctl_name = CTL_UNNUMBERED,
1133 .procname = "stat_interval",
1134 .data = &sysctl_stat_interval,
1135 .maxlen = sizeof(sysctl_stat_interval),
1136 .mode = 0644,
1137 .proc_handler = &proc_dointvec_jiffies,
1138 .strategy = &sysctl_jiffies,
1140 #endif
1141 #ifdef CONFIG_SECURITY
1143 .ctl_name = CTL_UNNUMBERED,
1144 .procname = "mmap_min_addr",
1145 .data = &mmap_min_addr,
1146 .maxlen = sizeof(unsigned long),
1147 .mode = 0644,
1148 .proc_handler = &proc_doulongvec_minmax,
1150 #endif
1151 #ifdef CONFIG_NUMA
1153 .ctl_name = CTL_UNNUMBERED,
1154 .procname = "numa_zonelist_order",
1155 .data = &numa_zonelist_order,
1156 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1157 .mode = 0644,
1158 .proc_handler = &numa_zonelist_order_handler,
1159 .strategy = &sysctl_string,
1161 #endif
1162 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1163 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1165 .ctl_name = VM_VDSO_ENABLED,
1166 .procname = "vdso_enabled",
1167 .data = &vdso_enabled,
1168 .maxlen = sizeof(vdso_enabled),
1169 .mode = 0644,
1170 .proc_handler = &proc_dointvec,
1171 .strategy = &sysctl_intvec,
1172 .extra1 = &zero,
1174 #endif
1175 #ifdef CONFIG_HIGHMEM
1177 .ctl_name = CTL_UNNUMBERED,
1178 .procname = "highmem_is_dirtyable",
1179 .data = &vm_highmem_is_dirtyable,
1180 .maxlen = sizeof(vm_highmem_is_dirtyable),
1181 .mode = 0644,
1182 .proc_handler = &proc_dointvec_minmax,
1183 .strategy = &sysctl_intvec,
1184 .extra1 = &zero,
1185 .extra2 = &one,
1187 #endif
1189 * NOTE: do not add new entries to this table unless you have read
1190 * Documentation/sysctl/ctl_unnumbered.txt
1192 { .ctl_name = 0 }
1195 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1196 static struct ctl_table binfmt_misc_table[] = {
1197 { .ctl_name = 0 }
1199 #endif
1201 static struct ctl_table fs_table[] = {
1203 .ctl_name = FS_NRINODE,
1204 .procname = "inode-nr",
1205 .data = &inodes_stat,
1206 .maxlen = 2*sizeof(int),
1207 .mode = 0444,
1208 .proc_handler = &proc_dointvec,
1211 .ctl_name = FS_STATINODE,
1212 .procname = "inode-state",
1213 .data = &inodes_stat,
1214 .maxlen = 7*sizeof(int),
1215 .mode = 0444,
1216 .proc_handler = &proc_dointvec,
1219 .procname = "file-nr",
1220 .data = &files_stat,
1221 .maxlen = 3*sizeof(int),
1222 .mode = 0444,
1223 .proc_handler = &proc_nr_files,
1226 .ctl_name = FS_MAXFILE,
1227 .procname = "file-max",
1228 .data = &files_stat.max_files,
1229 .maxlen = sizeof(int),
1230 .mode = 0644,
1231 .proc_handler = &proc_dointvec,
1234 .ctl_name = CTL_UNNUMBERED,
1235 .procname = "nr_open",
1236 .data = &sysctl_nr_open,
1237 .maxlen = sizeof(int),
1238 .mode = 0644,
1239 .proc_handler = &proc_dointvec_minmax,
1240 .extra1 = &sysctl_nr_open_min,
1241 .extra2 = &sysctl_nr_open_max,
1244 .ctl_name = FS_DENTRY,
1245 .procname = "dentry-state",
1246 .data = &dentry_stat,
1247 .maxlen = 6*sizeof(int),
1248 .mode = 0444,
1249 .proc_handler = &proc_dointvec,
1252 .ctl_name = FS_OVERFLOWUID,
1253 .procname = "overflowuid",
1254 .data = &fs_overflowuid,
1255 .maxlen = sizeof(int),
1256 .mode = 0644,
1257 .proc_handler = &proc_dointvec_minmax,
1258 .strategy = &sysctl_intvec,
1259 .extra1 = &minolduid,
1260 .extra2 = &maxolduid,
1263 .ctl_name = FS_OVERFLOWGID,
1264 .procname = "overflowgid",
1265 .data = &fs_overflowgid,
1266 .maxlen = sizeof(int),
1267 .mode = 0644,
1268 .proc_handler = &proc_dointvec_minmax,
1269 .strategy = &sysctl_intvec,
1270 .extra1 = &minolduid,
1271 .extra2 = &maxolduid,
1273 #ifdef CONFIG_FILE_LOCKING
1275 .ctl_name = FS_LEASES,
1276 .procname = "leases-enable",
1277 .data = &leases_enable,
1278 .maxlen = sizeof(int),
1279 .mode = 0644,
1280 .proc_handler = &proc_dointvec,
1282 #endif
1283 #ifdef CONFIG_DNOTIFY
1285 .ctl_name = FS_DIR_NOTIFY,
1286 .procname = "dir-notify-enable",
1287 .data = &dir_notify_enable,
1288 .maxlen = sizeof(int),
1289 .mode = 0644,
1290 .proc_handler = &proc_dointvec,
1292 #endif
1293 #ifdef CONFIG_MMU
1294 #ifdef CONFIG_FILE_LOCKING
1296 .ctl_name = FS_LEASE_TIME,
1297 .procname = "lease-break-time",
1298 .data = &lease_break_time,
1299 .maxlen = sizeof(int),
1300 .mode = 0644,
1301 .proc_handler = &proc_dointvec_minmax,
1302 .strategy = &sysctl_intvec,
1303 .extra1 = &zero,
1304 .extra2 = &two,
1306 #endif
1307 #ifdef CONFIG_AIO
1309 .procname = "aio-nr",
1310 .data = &aio_nr,
1311 .maxlen = sizeof(aio_nr),
1312 .mode = 0444,
1313 .proc_handler = &proc_doulongvec_minmax,
1316 .procname = "aio-max-nr",
1317 .data = &aio_max_nr,
1318 .maxlen = sizeof(aio_max_nr),
1319 .mode = 0644,
1320 .proc_handler = &proc_doulongvec_minmax,
1322 #endif /* CONFIG_AIO */
1323 #ifdef CONFIG_INOTIFY_USER
1325 .ctl_name = FS_INOTIFY,
1326 .procname = "inotify",
1327 .mode = 0555,
1328 .child = inotify_table,
1330 #endif
1331 #ifdef CONFIG_EPOLL
1333 .procname = "epoll",
1334 .mode = 0555,
1335 .child = epoll_table,
1337 #endif
1338 #endif
1340 .ctl_name = KERN_SETUID_DUMPABLE,
1341 .procname = "suid_dumpable",
1342 .data = &suid_dumpable,
1343 .maxlen = sizeof(int),
1344 .mode = 0644,
1345 .proc_handler = &proc_dointvec,
1347 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1349 .ctl_name = CTL_UNNUMBERED,
1350 .procname = "binfmt_misc",
1351 .mode = 0555,
1352 .child = binfmt_misc_table,
1354 #endif
1356 * NOTE: do not add new entries to this table unless you have read
1357 * Documentation/sysctl/ctl_unnumbered.txt
1359 { .ctl_name = 0 }
1362 static struct ctl_table debug_table[] = {
1363 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1365 .ctl_name = CTL_UNNUMBERED,
1366 .procname = "exception-trace",
1367 .data = &show_unhandled_signals,
1368 .maxlen = sizeof(int),
1369 .mode = 0644,
1370 .proc_handler = proc_dointvec
1372 #endif
1373 { .ctl_name = 0 }
1376 static struct ctl_table dev_table[] = {
1377 { .ctl_name = 0 }
1380 static DEFINE_SPINLOCK(sysctl_lock);
1382 /* called under sysctl_lock */
1383 static int use_table(struct ctl_table_header *p)
1385 if (unlikely(p->unregistering))
1386 return 0;
1387 p->used++;
1388 return 1;
1391 /* called under sysctl_lock */
1392 static void unuse_table(struct ctl_table_header *p)
1394 if (!--p->used)
1395 if (unlikely(p->unregistering))
1396 complete(p->unregistering);
1399 /* called under sysctl_lock, will reacquire if has to wait */
1400 static void start_unregistering(struct ctl_table_header *p)
1403 * if p->used is 0, nobody will ever touch that entry again;
1404 * we'll eliminate all paths to it before dropping sysctl_lock
1406 if (unlikely(p->used)) {
1407 struct completion wait;
1408 init_completion(&wait);
1409 p->unregistering = &wait;
1410 spin_unlock(&sysctl_lock);
1411 wait_for_completion(&wait);
1412 spin_lock(&sysctl_lock);
1413 } else {
1414 /* anything non-NULL; we'll never dereference it */
1415 p->unregistering = ERR_PTR(-EINVAL);
1418 * do not remove from the list until nobody holds it; walking the
1419 * list in do_sysctl() relies on that.
1421 list_del_init(&p->ctl_entry);
1424 void sysctl_head_get(struct ctl_table_header *head)
1426 spin_lock(&sysctl_lock);
1427 head->count++;
1428 spin_unlock(&sysctl_lock);
1431 void sysctl_head_put(struct ctl_table_header *head)
1433 spin_lock(&sysctl_lock);
1434 if (!--head->count)
1435 kfree(head);
1436 spin_unlock(&sysctl_lock);
1439 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1441 if (!head)
1442 BUG();
1443 spin_lock(&sysctl_lock);
1444 if (!use_table(head))
1445 head = ERR_PTR(-ENOENT);
1446 spin_unlock(&sysctl_lock);
1447 return head;
1450 void sysctl_head_finish(struct ctl_table_header *head)
1452 if (!head)
1453 return;
1454 spin_lock(&sysctl_lock);
1455 unuse_table(head);
1456 spin_unlock(&sysctl_lock);
1459 static struct ctl_table_set *
1460 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1462 struct ctl_table_set *set = &root->default_set;
1463 if (root->lookup)
1464 set = root->lookup(root, namespaces);
1465 return set;
1468 static struct list_head *
1469 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1471 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1472 return &set->list;
1475 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1476 struct ctl_table_header *prev)
1478 struct ctl_table_root *root;
1479 struct list_head *header_list;
1480 struct ctl_table_header *head;
1481 struct list_head *tmp;
1483 spin_lock(&sysctl_lock);
1484 if (prev) {
1485 head = prev;
1486 tmp = &prev->ctl_entry;
1487 unuse_table(prev);
1488 goto next;
1490 tmp = &root_table_header.ctl_entry;
1491 for (;;) {
1492 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1494 if (!use_table(head))
1495 goto next;
1496 spin_unlock(&sysctl_lock);
1497 return head;
1498 next:
1499 root = head->root;
1500 tmp = tmp->next;
1501 header_list = lookup_header_list(root, namespaces);
1502 if (tmp != header_list)
1503 continue;
1505 do {
1506 root = list_entry(root->root_list.next,
1507 struct ctl_table_root, root_list);
1508 if (root == &sysctl_table_root)
1509 goto out;
1510 header_list = lookup_header_list(root, namespaces);
1511 } while (list_empty(header_list));
1512 tmp = header_list->next;
1514 out:
1515 spin_unlock(&sysctl_lock);
1516 return NULL;
1519 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1521 return __sysctl_head_next(current->nsproxy, prev);
1524 void register_sysctl_root(struct ctl_table_root *root)
1526 spin_lock(&sysctl_lock);
1527 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1528 spin_unlock(&sysctl_lock);
1531 #ifdef CONFIG_SYSCTL_SYSCALL
1532 /* Perform the actual read/write of a sysctl table entry. */
1533 static int do_sysctl_strategy(struct ctl_table_root *root,
1534 struct ctl_table *table,
1535 void __user *oldval, size_t __user *oldlenp,
1536 void __user *newval, size_t newlen)
1538 int op = 0, rc;
1540 if (oldval)
1541 op |= MAY_READ;
1542 if (newval)
1543 op |= MAY_WRITE;
1544 if (sysctl_perm(root, table, op))
1545 return -EPERM;
1547 if (table->strategy) {
1548 rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1549 if (rc < 0)
1550 return rc;
1551 if (rc > 0)
1552 return 0;
1555 /* If there is no strategy routine, or if the strategy returns
1556 * zero, proceed with automatic r/w */
1557 if (table->data && table->maxlen) {
1558 rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1559 if (rc < 0)
1560 return rc;
1562 return 0;
1565 static int parse_table(int __user *name, int nlen,
1566 void __user *oldval, size_t __user *oldlenp,
1567 void __user *newval, size_t newlen,
1568 struct ctl_table_root *root,
1569 struct ctl_table *table)
1571 int n;
1572 repeat:
1573 if (!nlen)
1574 return -ENOTDIR;
1575 if (get_user(n, name))
1576 return -EFAULT;
1577 for ( ; table->ctl_name || table->procname; table++) {
1578 if (!table->ctl_name)
1579 continue;
1580 if (n == table->ctl_name) {
1581 int error;
1582 if (table->child) {
1583 if (sysctl_perm(root, table, MAY_EXEC))
1584 return -EPERM;
1585 name++;
1586 nlen--;
1587 table = table->child;
1588 goto repeat;
1590 error = do_sysctl_strategy(root, table,
1591 oldval, oldlenp,
1592 newval, newlen);
1593 return error;
1596 return -ENOTDIR;
1599 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1600 void __user *newval, size_t newlen)
1602 struct ctl_table_header *head;
1603 int error = -ENOTDIR;
1605 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1606 return -ENOTDIR;
1607 if (oldval) {
1608 int old_len;
1609 if (!oldlenp || get_user(old_len, oldlenp))
1610 return -EFAULT;
1613 for (head = sysctl_head_next(NULL); head;
1614 head = sysctl_head_next(head)) {
1615 error = parse_table(name, nlen, oldval, oldlenp,
1616 newval, newlen,
1617 head->root, head->ctl_table);
1618 if (error != -ENOTDIR) {
1619 sysctl_head_finish(head);
1620 break;
1623 return error;
1626 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1628 struct __sysctl_args tmp;
1629 int error;
1631 if (copy_from_user(&tmp, args, sizeof(tmp)))
1632 return -EFAULT;
1634 error = deprecated_sysctl_warning(&tmp);
1635 if (error)
1636 goto out;
1638 lock_kernel();
1639 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1640 tmp.newval, tmp.newlen);
1641 unlock_kernel();
1642 out:
1643 return error;
1645 #endif /* CONFIG_SYSCTL_SYSCALL */
1648 * sysctl_perm does NOT grant the superuser all rights automatically, because
1649 * some sysctl variables are readonly even to root.
1652 static int test_perm(int mode, int op)
1654 if (!current->euid)
1655 mode >>= 6;
1656 else if (in_egroup_p(0))
1657 mode >>= 3;
1658 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1659 return 0;
1660 return -EACCES;
1663 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1665 int error;
1666 int mode;
1668 error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1669 if (error)
1670 return error;
1672 if (root->permissions)
1673 mode = root->permissions(root, current->nsproxy, table);
1674 else
1675 mode = table->mode;
1677 return test_perm(mode, op);
1680 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1682 for (; table->ctl_name || table->procname; table++) {
1683 table->parent = parent;
1684 if (table->child)
1685 sysctl_set_parent(table, table->child);
1689 static __init int sysctl_init(void)
1691 sysctl_set_parent(NULL, root_table);
1692 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1694 int err;
1695 err = sysctl_check_table(current->nsproxy, root_table);
1697 #endif
1698 return 0;
1701 core_initcall(sysctl_init);
1703 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1704 struct ctl_table *table)
1706 struct ctl_table *p;
1707 const char *s = branch->procname;
1709 /* branch should have named subdirectory as its first element */
1710 if (!s || !branch->child)
1711 return NULL;
1713 /* ... and nothing else */
1714 if (branch[1].procname || branch[1].ctl_name)
1715 return NULL;
1717 /* table should contain subdirectory with the same name */
1718 for (p = table; p->procname || p->ctl_name; p++) {
1719 if (!p->child)
1720 continue;
1721 if (p->procname && strcmp(p->procname, s) == 0)
1722 return p;
1724 return NULL;
1727 /* see if attaching q to p would be an improvement */
1728 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1730 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1731 struct ctl_table *next;
1732 int is_better = 0;
1733 int not_in_parent = !p->attached_by;
1735 while ((next = is_branch_in(by, to)) != NULL) {
1736 if (by == q->attached_by)
1737 is_better = 1;
1738 if (to == p->attached_by)
1739 not_in_parent = 1;
1740 by = by->child;
1741 to = next->child;
1744 if (is_better && not_in_parent) {
1745 q->attached_by = by;
1746 q->attached_to = to;
1747 q->parent = p;
1752 * __register_sysctl_paths - register a sysctl hierarchy
1753 * @root: List of sysctl headers to register on
1754 * @namespaces: Data to compute which lists of sysctl entries are visible
1755 * @path: The path to the directory the sysctl table is in.
1756 * @table: the top-level table structure
1758 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1759 * array. A completely 0 filled entry terminates the table.
1761 * The members of the &struct ctl_table structure are used as follows:
1763 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1764 * must be unique within that level of sysctl
1766 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1767 * enter a sysctl file
1769 * data - a pointer to data for use by proc_handler
1771 * maxlen - the maximum size in bytes of the data
1773 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1775 * child - a pointer to the child sysctl table if this entry is a directory, or
1776 * %NULL.
1778 * proc_handler - the text handler routine (described below)
1780 * strategy - the strategy routine (described below)
1782 * de - for internal use by the sysctl routines
1784 * extra1, extra2 - extra pointers usable by the proc handler routines
1786 * Leaf nodes in the sysctl tree will be represented by a single file
1787 * under /proc; non-leaf nodes will be represented by directories.
1789 * sysctl(2) can automatically manage read and write requests through
1790 * the sysctl table. The data and maxlen fields of the ctl_table
1791 * struct enable minimal validation of the values being written to be
1792 * performed, and the mode field allows minimal authentication.
1794 * More sophisticated management can be enabled by the provision of a
1795 * strategy routine with the table entry. This will be called before
1796 * any automatic read or write of the data is performed.
1798 * The strategy routine may return
1800 * < 0 - Error occurred (error is passed to user process)
1802 * 0 - OK - proceed with automatic read or write.
1804 * > 0 - OK - read or write has been done by the strategy routine, so
1805 * return immediately.
1807 * There must be a proc_handler routine for any terminal nodes
1808 * mirrored under /proc/sys (non-terminals are handled by a built-in
1809 * directory handler). Several default handlers are available to
1810 * cover common cases -
1812 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1813 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1814 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1816 * It is the handler's job to read the input buffer from user memory
1817 * and process it. The handler should return 0 on success.
1819 * This routine returns %NULL on a failure to register, and a pointer
1820 * to the table header on success.
1822 struct ctl_table_header *__register_sysctl_paths(
1823 struct ctl_table_root *root,
1824 struct nsproxy *namespaces,
1825 const struct ctl_path *path, struct ctl_table *table)
1827 struct ctl_table_header *header;
1828 struct ctl_table *new, **prevp;
1829 unsigned int n, npath;
1830 struct ctl_table_set *set;
1832 /* Count the path components */
1833 for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1837 * For each path component, allocate a 2-element ctl_table array.
1838 * The first array element will be filled with the sysctl entry
1839 * for this, the second will be the sentinel (ctl_name == 0).
1841 * We allocate everything in one go so that we don't have to
1842 * worry about freeing additional memory in unregister_sysctl_table.
1844 header = kzalloc(sizeof(struct ctl_table_header) +
1845 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1846 if (!header)
1847 return NULL;
1849 new = (struct ctl_table *) (header + 1);
1851 /* Now connect the dots */
1852 prevp = &header->ctl_table;
1853 for (n = 0; n < npath; ++n, ++path) {
1854 /* Copy the procname */
1855 new->procname = path->procname;
1856 new->ctl_name = path->ctl_name;
1857 new->mode = 0555;
1859 *prevp = new;
1860 prevp = &new->child;
1862 new += 2;
1864 *prevp = table;
1865 header->ctl_table_arg = table;
1867 INIT_LIST_HEAD(&header->ctl_entry);
1868 header->used = 0;
1869 header->unregistering = NULL;
1870 header->root = root;
1871 sysctl_set_parent(NULL, header->ctl_table);
1872 header->count = 1;
1873 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1874 if (sysctl_check_table(namespaces, header->ctl_table)) {
1875 kfree(header);
1876 return NULL;
1878 #endif
1879 spin_lock(&sysctl_lock);
1880 header->set = lookup_header_set(root, namespaces);
1881 header->attached_by = header->ctl_table;
1882 header->attached_to = root_table;
1883 header->parent = &root_table_header;
1884 for (set = header->set; set; set = set->parent) {
1885 struct ctl_table_header *p;
1886 list_for_each_entry(p, &set->list, ctl_entry) {
1887 if (p->unregistering)
1888 continue;
1889 try_attach(p, header);
1892 header->parent->count++;
1893 list_add_tail(&header->ctl_entry, &header->set->list);
1894 spin_unlock(&sysctl_lock);
1896 return header;
1900 * register_sysctl_table_path - register a sysctl table hierarchy
1901 * @path: The path to the directory the sysctl table is in.
1902 * @table: the top-level table structure
1904 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1905 * array. A completely 0 filled entry terminates the table.
1907 * See __register_sysctl_paths for more details.
1909 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1910 struct ctl_table *table)
1912 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1913 path, table);
1917 * register_sysctl_table - register a sysctl table hierarchy
1918 * @table: the top-level table structure
1920 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1921 * array. A completely 0 filled entry terminates the table.
1923 * See register_sysctl_paths for more details.
1925 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1927 static const struct ctl_path null_path[] = { {} };
1929 return register_sysctl_paths(null_path, table);
1933 * unregister_sysctl_table - unregister a sysctl table hierarchy
1934 * @header: the header returned from register_sysctl_table
1936 * Unregisters the sysctl table and all children. proc entries may not
1937 * actually be removed until they are no longer used by anyone.
1939 void unregister_sysctl_table(struct ctl_table_header * header)
1941 might_sleep();
1943 if (header == NULL)
1944 return;
1946 spin_lock(&sysctl_lock);
1947 start_unregistering(header);
1948 if (!--header->parent->count) {
1949 WARN_ON(1);
1950 kfree(header->parent);
1952 if (!--header->count)
1953 kfree(header);
1954 spin_unlock(&sysctl_lock);
1957 int sysctl_is_seen(struct ctl_table_header *p)
1959 struct ctl_table_set *set = p->set;
1960 int res;
1961 spin_lock(&sysctl_lock);
1962 if (p->unregistering)
1963 res = 0;
1964 else if (!set->is_seen)
1965 res = 1;
1966 else
1967 res = set->is_seen(set);
1968 spin_unlock(&sysctl_lock);
1969 return res;
1972 void setup_sysctl_set(struct ctl_table_set *p,
1973 struct ctl_table_set *parent,
1974 int (*is_seen)(struct ctl_table_set *))
1976 INIT_LIST_HEAD(&p->list);
1977 p->parent = parent ? parent : &sysctl_table_root.default_set;
1978 p->is_seen = is_seen;
1981 #else /* !CONFIG_SYSCTL */
1982 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1984 return NULL;
1987 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1988 struct ctl_table *table)
1990 return NULL;
1993 void unregister_sysctl_table(struct ctl_table_header * table)
1997 void setup_sysctl_set(struct ctl_table_set *p,
1998 struct ctl_table_set *parent,
1999 int (*is_seen)(struct ctl_table_set *))
2003 void sysctl_head_put(struct ctl_table_header *head)
2007 #endif /* CONFIG_SYSCTL */
2010 * /proc/sys support
2013 #ifdef CONFIG_PROC_SYSCTL
2015 static int _proc_do_string(void* data, int maxlen, int write,
2016 struct file *filp, void __user *buffer,
2017 size_t *lenp, loff_t *ppos)
2019 size_t len;
2020 char __user *p;
2021 char c;
2023 if (!data || !maxlen || !*lenp) {
2024 *lenp = 0;
2025 return 0;
2028 if (write) {
2029 len = 0;
2030 p = buffer;
2031 while (len < *lenp) {
2032 if (get_user(c, p++))
2033 return -EFAULT;
2034 if (c == 0 || c == '\n')
2035 break;
2036 len++;
2038 if (len >= maxlen)
2039 len = maxlen-1;
2040 if(copy_from_user(data, buffer, len))
2041 return -EFAULT;
2042 ((char *) data)[len] = 0;
2043 *ppos += *lenp;
2044 } else {
2045 len = strlen(data);
2046 if (len > maxlen)
2047 len = maxlen;
2049 if (*ppos > len) {
2050 *lenp = 0;
2051 return 0;
2054 data += *ppos;
2055 len -= *ppos;
2057 if (len > *lenp)
2058 len = *lenp;
2059 if (len)
2060 if(copy_to_user(buffer, data, len))
2061 return -EFAULT;
2062 if (len < *lenp) {
2063 if(put_user('\n', ((char __user *) buffer) + len))
2064 return -EFAULT;
2065 len++;
2067 *lenp = len;
2068 *ppos += len;
2070 return 0;
2074 * proc_dostring - read a string sysctl
2075 * @table: the sysctl table
2076 * @write: %TRUE if this is a write to the sysctl file
2077 * @filp: the file structure
2078 * @buffer: the user buffer
2079 * @lenp: the size of the user buffer
2080 * @ppos: file position
2082 * Reads/writes a string from/to the user buffer. If the kernel
2083 * buffer provided is not large enough to hold the string, the
2084 * string is truncated. The copied string is %NULL-terminated.
2085 * If the string is being read by the user process, it is copied
2086 * and a newline '\n' is added. It is truncated if the buffer is
2087 * not large enough.
2089 * Returns 0 on success.
2091 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2092 void __user *buffer, size_t *lenp, loff_t *ppos)
2094 return _proc_do_string(table->data, table->maxlen, write, filp,
2095 buffer, lenp, ppos);
2099 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2100 int *valp,
2101 int write, void *data)
2103 if (write) {
2104 *valp = *negp ? -*lvalp : *lvalp;
2105 } else {
2106 int val = *valp;
2107 if (val < 0) {
2108 *negp = -1;
2109 *lvalp = (unsigned long)-val;
2110 } else {
2111 *negp = 0;
2112 *lvalp = (unsigned long)val;
2115 return 0;
2118 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2119 int write, struct file *filp, void __user *buffer,
2120 size_t *lenp, loff_t *ppos,
2121 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2122 int write, void *data),
2123 void *data)
2125 #define TMPBUFLEN 21
2126 int *i, vleft, first=1, neg, val;
2127 unsigned long lval;
2128 size_t left, len;
2130 char buf[TMPBUFLEN], *p;
2131 char __user *s = buffer;
2133 if (!tbl_data || !table->maxlen || !*lenp ||
2134 (*ppos && !write)) {
2135 *lenp = 0;
2136 return 0;
2139 i = (int *) tbl_data;
2140 vleft = table->maxlen / sizeof(*i);
2141 left = *lenp;
2143 if (!conv)
2144 conv = do_proc_dointvec_conv;
2146 for (; left && vleft--; i++, first=0) {
2147 if (write) {
2148 while (left) {
2149 char c;
2150 if (get_user(c, s))
2151 return -EFAULT;
2152 if (!isspace(c))
2153 break;
2154 left--;
2155 s++;
2157 if (!left)
2158 break;
2159 neg = 0;
2160 len = left;
2161 if (len > sizeof(buf) - 1)
2162 len = sizeof(buf) - 1;
2163 if (copy_from_user(buf, s, len))
2164 return -EFAULT;
2165 buf[len] = 0;
2166 p = buf;
2167 if (*p == '-' && left > 1) {
2168 neg = 1;
2169 p++;
2171 if (*p < '0' || *p > '9')
2172 break;
2174 lval = simple_strtoul(p, &p, 0);
2176 len = p-buf;
2177 if ((len < left) && *p && !isspace(*p))
2178 break;
2179 if (neg)
2180 val = -val;
2181 s += len;
2182 left -= len;
2184 if (conv(&neg, &lval, i, 1, data))
2185 break;
2186 } else {
2187 p = buf;
2188 if (!first)
2189 *p++ = '\t';
2191 if (conv(&neg, &lval, i, 0, data))
2192 break;
2194 sprintf(p, "%s%lu", neg ? "-" : "", lval);
2195 len = strlen(buf);
2196 if (len > left)
2197 len = left;
2198 if(copy_to_user(s, buf, len))
2199 return -EFAULT;
2200 left -= len;
2201 s += len;
2205 if (!write && !first && left) {
2206 if(put_user('\n', s))
2207 return -EFAULT;
2208 left--, s++;
2210 if (write) {
2211 while (left) {
2212 char c;
2213 if (get_user(c, s++))
2214 return -EFAULT;
2215 if (!isspace(c))
2216 break;
2217 left--;
2220 if (write && first)
2221 return -EINVAL;
2222 *lenp -= left;
2223 *ppos += *lenp;
2224 return 0;
2225 #undef TMPBUFLEN
2228 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2229 void __user *buffer, size_t *lenp, loff_t *ppos,
2230 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2231 int write, void *data),
2232 void *data)
2234 return __do_proc_dointvec(table->data, table, write, filp,
2235 buffer, lenp, ppos, conv, data);
2239 * proc_dointvec - read a vector of integers
2240 * @table: the sysctl table
2241 * @write: %TRUE if this is a write to the sysctl file
2242 * @filp: the file structure
2243 * @buffer: the user buffer
2244 * @lenp: the size of the user buffer
2245 * @ppos: file position
2247 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2248 * values from/to the user buffer, treated as an ASCII string.
2250 * Returns 0 on success.
2252 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2253 void __user *buffer, size_t *lenp, loff_t *ppos)
2255 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2256 NULL,NULL);
2260 * Taint values can only be increased
2261 * This means we can safely use a temporary.
2263 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
2264 void __user *buffer, size_t *lenp, loff_t *ppos)
2266 struct ctl_table t;
2267 unsigned long tmptaint = get_taint();
2268 int err;
2270 if (write && !capable(CAP_SYS_ADMIN))
2271 return -EPERM;
2273 t = *table;
2274 t.data = &tmptaint;
2275 err = proc_doulongvec_minmax(&t, write, filp, buffer, lenp, ppos);
2276 if (err < 0)
2277 return err;
2279 if (write) {
2281 * Poor man's atomic or. Not worth adding a primitive
2282 * to everyone's atomic.h for this
2284 int i;
2285 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2286 if ((tmptaint >> i) & 1)
2287 add_taint(i);
2291 return err;
2294 struct do_proc_dointvec_minmax_conv_param {
2295 int *min;
2296 int *max;
2299 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2300 int *valp,
2301 int write, void *data)
2303 struct do_proc_dointvec_minmax_conv_param *param = data;
2304 if (write) {
2305 int val = *negp ? -*lvalp : *lvalp;
2306 if ((param->min && *param->min > val) ||
2307 (param->max && *param->max < val))
2308 return -EINVAL;
2309 *valp = val;
2310 } else {
2311 int val = *valp;
2312 if (val < 0) {
2313 *negp = -1;
2314 *lvalp = (unsigned long)-val;
2315 } else {
2316 *negp = 0;
2317 *lvalp = (unsigned long)val;
2320 return 0;
2324 * proc_dointvec_minmax - read a vector of integers with min/max values
2325 * @table: the sysctl table
2326 * @write: %TRUE if this is a write to the sysctl file
2327 * @filp: the file structure
2328 * @buffer: the user buffer
2329 * @lenp: the size of the user buffer
2330 * @ppos: file position
2332 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2333 * values from/to the user buffer, treated as an ASCII string.
2335 * This routine will ensure the values are within the range specified by
2336 * table->extra1 (min) and table->extra2 (max).
2338 * Returns 0 on success.
2340 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2341 void __user *buffer, size_t *lenp, loff_t *ppos)
2343 struct do_proc_dointvec_minmax_conv_param param = {
2344 .min = (int *) table->extra1,
2345 .max = (int *) table->extra2,
2347 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2348 do_proc_dointvec_minmax_conv, &param);
2351 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2352 struct file *filp,
2353 void __user *buffer,
2354 size_t *lenp, loff_t *ppos,
2355 unsigned long convmul,
2356 unsigned long convdiv)
2358 #define TMPBUFLEN 21
2359 unsigned long *i, *min, *max, val;
2360 int vleft, first=1, neg;
2361 size_t len, left;
2362 char buf[TMPBUFLEN], *p;
2363 char __user *s = buffer;
2365 if (!data || !table->maxlen || !*lenp ||
2366 (*ppos && !write)) {
2367 *lenp = 0;
2368 return 0;
2371 i = (unsigned long *) data;
2372 min = (unsigned long *) table->extra1;
2373 max = (unsigned long *) table->extra2;
2374 vleft = table->maxlen / sizeof(unsigned long);
2375 left = *lenp;
2377 for (; left && vleft--; i++, min++, max++, first=0) {
2378 if (write) {
2379 while (left) {
2380 char c;
2381 if (get_user(c, s))
2382 return -EFAULT;
2383 if (!isspace(c))
2384 break;
2385 left--;
2386 s++;
2388 if (!left)
2389 break;
2390 neg = 0;
2391 len = left;
2392 if (len > TMPBUFLEN-1)
2393 len = TMPBUFLEN-1;
2394 if (copy_from_user(buf, s, len))
2395 return -EFAULT;
2396 buf[len] = 0;
2397 p = buf;
2398 if (*p == '-' && left > 1) {
2399 neg = 1;
2400 p++;
2402 if (*p < '0' || *p > '9')
2403 break;
2404 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2405 len = p-buf;
2406 if ((len < left) && *p && !isspace(*p))
2407 break;
2408 if (neg)
2409 val = -val;
2410 s += len;
2411 left -= len;
2413 if(neg)
2414 continue;
2415 if ((min && val < *min) || (max && val > *max))
2416 continue;
2417 *i = val;
2418 } else {
2419 p = buf;
2420 if (!first)
2421 *p++ = '\t';
2422 sprintf(p, "%lu", convdiv * (*i) / convmul);
2423 len = strlen(buf);
2424 if (len > left)
2425 len = left;
2426 if(copy_to_user(s, buf, len))
2427 return -EFAULT;
2428 left -= len;
2429 s += len;
2433 if (!write && !first && left) {
2434 if(put_user('\n', s))
2435 return -EFAULT;
2436 left--, s++;
2438 if (write) {
2439 while (left) {
2440 char c;
2441 if (get_user(c, s++))
2442 return -EFAULT;
2443 if (!isspace(c))
2444 break;
2445 left--;
2448 if (write && first)
2449 return -EINVAL;
2450 *lenp -= left;
2451 *ppos += *lenp;
2452 return 0;
2453 #undef TMPBUFLEN
2456 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2457 struct file *filp,
2458 void __user *buffer,
2459 size_t *lenp, loff_t *ppos,
2460 unsigned long convmul,
2461 unsigned long convdiv)
2463 return __do_proc_doulongvec_minmax(table->data, table, write,
2464 filp, buffer, lenp, ppos, convmul, convdiv);
2468 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2469 * @table: the sysctl table
2470 * @write: %TRUE if this is a write to the sysctl file
2471 * @filp: the file structure
2472 * @buffer: the user buffer
2473 * @lenp: the size of the user buffer
2474 * @ppos: file position
2476 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2477 * values from/to the user buffer, treated as an ASCII string.
2479 * This routine will ensure the values are within the range specified by
2480 * table->extra1 (min) and table->extra2 (max).
2482 * Returns 0 on success.
2484 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2485 void __user *buffer, size_t *lenp, loff_t *ppos)
2487 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2491 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2492 * @table: the sysctl table
2493 * @write: %TRUE if this is a write to the sysctl file
2494 * @filp: the file structure
2495 * @buffer: the user buffer
2496 * @lenp: the size of the user buffer
2497 * @ppos: file position
2499 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2500 * values from/to the user buffer, treated as an ASCII string. The values
2501 * are treated as milliseconds, and converted to jiffies when they are stored.
2503 * This routine will ensure the values are within the range specified by
2504 * table->extra1 (min) and table->extra2 (max).
2506 * Returns 0 on success.
2508 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2509 struct file *filp,
2510 void __user *buffer,
2511 size_t *lenp, loff_t *ppos)
2513 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2514 lenp, ppos, HZ, 1000l);
2518 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2519 int *valp,
2520 int write, void *data)
2522 if (write) {
2523 if (*lvalp > LONG_MAX / HZ)
2524 return 1;
2525 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2526 } else {
2527 int val = *valp;
2528 unsigned long lval;
2529 if (val < 0) {
2530 *negp = -1;
2531 lval = (unsigned long)-val;
2532 } else {
2533 *negp = 0;
2534 lval = (unsigned long)val;
2536 *lvalp = lval / HZ;
2538 return 0;
2541 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2542 int *valp,
2543 int write, void *data)
2545 if (write) {
2546 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2547 return 1;
2548 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2549 } else {
2550 int val = *valp;
2551 unsigned long lval;
2552 if (val < 0) {
2553 *negp = -1;
2554 lval = (unsigned long)-val;
2555 } else {
2556 *negp = 0;
2557 lval = (unsigned long)val;
2559 *lvalp = jiffies_to_clock_t(lval);
2561 return 0;
2564 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2565 int *valp,
2566 int write, void *data)
2568 if (write) {
2569 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2570 } else {
2571 int val = *valp;
2572 unsigned long lval;
2573 if (val < 0) {
2574 *negp = -1;
2575 lval = (unsigned long)-val;
2576 } else {
2577 *negp = 0;
2578 lval = (unsigned long)val;
2580 *lvalp = jiffies_to_msecs(lval);
2582 return 0;
2586 * proc_dointvec_jiffies - read a vector of integers as seconds
2587 * @table: the sysctl table
2588 * @write: %TRUE if this is a write to the sysctl file
2589 * @filp: the file structure
2590 * @buffer: the user buffer
2591 * @lenp: the size of the user buffer
2592 * @ppos: file position
2594 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2595 * values from/to the user buffer, treated as an ASCII string.
2596 * The values read are assumed to be in seconds, and are converted into
2597 * jiffies.
2599 * Returns 0 on success.
2601 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2602 void __user *buffer, size_t *lenp, loff_t *ppos)
2604 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2605 do_proc_dointvec_jiffies_conv,NULL);
2609 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2610 * @table: the sysctl table
2611 * @write: %TRUE if this is a write to the sysctl file
2612 * @filp: the file structure
2613 * @buffer: the user buffer
2614 * @lenp: the size of the user buffer
2615 * @ppos: pointer to the file position
2617 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2618 * values from/to the user buffer, treated as an ASCII string.
2619 * The values read are assumed to be in 1/USER_HZ seconds, and
2620 * are converted into jiffies.
2622 * Returns 0 on success.
2624 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2625 void __user *buffer, size_t *lenp, loff_t *ppos)
2627 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2628 do_proc_dointvec_userhz_jiffies_conv,NULL);
2632 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2633 * @table: the sysctl table
2634 * @write: %TRUE if this is a write to the sysctl file
2635 * @filp: the file structure
2636 * @buffer: the user buffer
2637 * @lenp: the size of the user buffer
2638 * @ppos: file position
2639 * @ppos: the current position in the file
2641 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2642 * values from/to the user buffer, treated as an ASCII string.
2643 * The values read are assumed to be in 1/1000 seconds, and
2644 * are converted into jiffies.
2646 * Returns 0 on success.
2648 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2649 void __user *buffer, size_t *lenp, loff_t *ppos)
2651 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2652 do_proc_dointvec_ms_jiffies_conv, NULL);
2655 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2656 void __user *buffer, size_t *lenp, loff_t *ppos)
2658 struct pid *new_pid;
2659 pid_t tmp;
2660 int r;
2662 tmp = pid_vnr(cad_pid);
2664 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2665 lenp, ppos, NULL, NULL);
2666 if (r || !write)
2667 return r;
2669 new_pid = find_get_pid(tmp);
2670 if (!new_pid)
2671 return -ESRCH;
2673 put_pid(xchg(&cad_pid, new_pid));
2674 return 0;
2677 #else /* CONFIG_PROC_FS */
2679 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2680 void __user *buffer, size_t *lenp, loff_t *ppos)
2682 return -ENOSYS;
2685 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2686 void __user *buffer, size_t *lenp, loff_t *ppos)
2688 return -ENOSYS;
2691 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2692 void __user *buffer, size_t *lenp, loff_t *ppos)
2694 return -ENOSYS;
2697 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2698 void __user *buffer, size_t *lenp, loff_t *ppos)
2700 return -ENOSYS;
2703 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2704 void __user *buffer, size_t *lenp, loff_t *ppos)
2706 return -ENOSYS;
2709 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2710 void __user *buffer, size_t *lenp, loff_t *ppos)
2712 return -ENOSYS;
2715 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2716 void __user *buffer, size_t *lenp, loff_t *ppos)
2718 return -ENOSYS;
2721 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2722 struct file *filp,
2723 void __user *buffer,
2724 size_t *lenp, loff_t *ppos)
2726 return -ENOSYS;
2730 #endif /* CONFIG_PROC_FS */
2733 #ifdef CONFIG_SYSCTL_SYSCALL
2735 * General sysctl support routines
2738 /* The generic sysctl data routine (used if no strategy routine supplied) */
2739 int sysctl_data(struct ctl_table *table,
2740 void __user *oldval, size_t __user *oldlenp,
2741 void __user *newval, size_t newlen)
2743 size_t len;
2745 /* Get out of I don't have a variable */
2746 if (!table->data || !table->maxlen)
2747 return -ENOTDIR;
2749 if (oldval && oldlenp) {
2750 if (get_user(len, oldlenp))
2751 return -EFAULT;
2752 if (len) {
2753 if (len > table->maxlen)
2754 len = table->maxlen;
2755 if (copy_to_user(oldval, table->data, len))
2756 return -EFAULT;
2757 if (put_user(len, oldlenp))
2758 return -EFAULT;
2762 if (newval && newlen) {
2763 if (newlen > table->maxlen)
2764 newlen = table->maxlen;
2766 if (copy_from_user(table->data, newval, newlen))
2767 return -EFAULT;
2769 return 1;
2772 /* The generic string strategy routine: */
2773 int sysctl_string(struct ctl_table *table,
2774 void __user *oldval, size_t __user *oldlenp,
2775 void __user *newval, size_t newlen)
2777 if (!table->data || !table->maxlen)
2778 return -ENOTDIR;
2780 if (oldval && oldlenp) {
2781 size_t bufsize;
2782 if (get_user(bufsize, oldlenp))
2783 return -EFAULT;
2784 if (bufsize) {
2785 size_t len = strlen(table->data), copied;
2787 /* This shouldn't trigger for a well-formed sysctl */
2788 if (len > table->maxlen)
2789 len = table->maxlen;
2791 /* Copy up to a max of bufsize-1 bytes of the string */
2792 copied = (len >= bufsize) ? bufsize - 1 : len;
2794 if (copy_to_user(oldval, table->data, copied) ||
2795 put_user(0, (char __user *)(oldval + copied)))
2796 return -EFAULT;
2797 if (put_user(len, oldlenp))
2798 return -EFAULT;
2801 if (newval && newlen) {
2802 size_t len = newlen;
2803 if (len > table->maxlen)
2804 len = table->maxlen;
2805 if(copy_from_user(table->data, newval, len))
2806 return -EFAULT;
2807 if (len == table->maxlen)
2808 len--;
2809 ((char *) table->data)[len] = 0;
2811 return 1;
2815 * This function makes sure that all of the integers in the vector
2816 * are between the minimum and maximum values given in the arrays
2817 * table->extra1 and table->extra2, respectively.
2819 int sysctl_intvec(struct ctl_table *table,
2820 void __user *oldval, size_t __user *oldlenp,
2821 void __user *newval, size_t newlen)
2824 if (newval && newlen) {
2825 int __user *vec = (int __user *) newval;
2826 int *min = (int *) table->extra1;
2827 int *max = (int *) table->extra2;
2828 size_t length;
2829 int i;
2831 if (newlen % sizeof(int) != 0)
2832 return -EINVAL;
2834 if (!table->extra1 && !table->extra2)
2835 return 0;
2837 if (newlen > table->maxlen)
2838 newlen = table->maxlen;
2839 length = newlen / sizeof(int);
2841 for (i = 0; i < length; i++) {
2842 int value;
2843 if (get_user(value, vec + i))
2844 return -EFAULT;
2845 if (min && value < min[i])
2846 return -EINVAL;
2847 if (max && value > max[i])
2848 return -EINVAL;
2851 return 0;
2854 /* Strategy function to convert jiffies to seconds */
2855 int sysctl_jiffies(struct ctl_table *table,
2856 void __user *oldval, size_t __user *oldlenp,
2857 void __user *newval, size_t newlen)
2859 if (oldval && oldlenp) {
2860 size_t olen;
2862 if (get_user(olen, oldlenp))
2863 return -EFAULT;
2864 if (olen) {
2865 int val;
2867 if (olen < sizeof(int))
2868 return -EINVAL;
2870 val = *(int *)(table->data) / HZ;
2871 if (put_user(val, (int __user *)oldval))
2872 return -EFAULT;
2873 if (put_user(sizeof(int), oldlenp))
2874 return -EFAULT;
2877 if (newval && newlen) {
2878 int new;
2879 if (newlen != sizeof(int))
2880 return -EINVAL;
2881 if (get_user(new, (int __user *)newval))
2882 return -EFAULT;
2883 *(int *)(table->data) = new*HZ;
2885 return 1;
2888 /* Strategy function to convert jiffies to seconds */
2889 int sysctl_ms_jiffies(struct ctl_table *table,
2890 void __user *oldval, size_t __user *oldlenp,
2891 void __user *newval, size_t newlen)
2893 if (oldval && oldlenp) {
2894 size_t olen;
2896 if (get_user(olen, oldlenp))
2897 return -EFAULT;
2898 if (olen) {
2899 int val;
2901 if (olen < sizeof(int))
2902 return -EINVAL;
2904 val = jiffies_to_msecs(*(int *)(table->data));
2905 if (put_user(val, (int __user *)oldval))
2906 return -EFAULT;
2907 if (put_user(sizeof(int), oldlenp))
2908 return -EFAULT;
2911 if (newval && newlen) {
2912 int new;
2913 if (newlen != sizeof(int))
2914 return -EINVAL;
2915 if (get_user(new, (int __user *)newval))
2916 return -EFAULT;
2917 *(int *)(table->data) = msecs_to_jiffies(new);
2919 return 1;
2924 #else /* CONFIG_SYSCTL_SYSCALL */
2927 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2929 struct __sysctl_args tmp;
2930 int error;
2932 if (copy_from_user(&tmp, args, sizeof(tmp)))
2933 return -EFAULT;
2935 error = deprecated_sysctl_warning(&tmp);
2937 /* If no error reading the parameters then just -ENOSYS ... */
2938 if (!error)
2939 error = -ENOSYS;
2941 return error;
2944 int sysctl_data(struct ctl_table *table,
2945 void __user *oldval, size_t __user *oldlenp,
2946 void __user *newval, size_t newlen)
2948 return -ENOSYS;
2951 int sysctl_string(struct ctl_table *table,
2952 void __user *oldval, size_t __user *oldlenp,
2953 void __user *newval, size_t newlen)
2955 return -ENOSYS;
2958 int sysctl_intvec(struct ctl_table *table,
2959 void __user *oldval, size_t __user *oldlenp,
2960 void __user *newval, size_t newlen)
2962 return -ENOSYS;
2965 int sysctl_jiffies(struct ctl_table *table,
2966 void __user *oldval, size_t __user *oldlenp,
2967 void __user *newval, size_t newlen)
2969 return -ENOSYS;
2972 int sysctl_ms_jiffies(struct ctl_table *table,
2973 void __user *oldval, size_t __user *oldlenp,
2974 void __user *newval, size_t newlen)
2976 return -ENOSYS;
2979 #endif /* CONFIG_SYSCTL_SYSCALL */
2981 static int deprecated_sysctl_warning(struct __sysctl_args *args)
2983 static int msg_count;
2984 int name[CTL_MAXNAME];
2985 int i;
2987 /* Check args->nlen. */
2988 if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
2989 return -ENOTDIR;
2991 /* Read in the sysctl name for better debug message logging */
2992 for (i = 0; i < args->nlen; i++)
2993 if (get_user(name[i], args->name + i))
2994 return -EFAULT;
2996 /* Ignore accesses to kernel.version */
2997 if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2998 return 0;
3000 if (msg_count < 5) {
3001 msg_count++;
3002 printk(KERN_INFO
3003 "warning: process `%s' used the deprecated sysctl "
3004 "system call with ", current->comm);
3005 for (i = 0; i < args->nlen; i++)
3006 printk("%d.", name[i]);
3007 printk("\n");
3009 return 0;
3013 * No sense putting this after each symbol definition, twice,
3014 * exception granted :-)
3016 EXPORT_SYMBOL(proc_dointvec);
3017 EXPORT_SYMBOL(proc_dointvec_jiffies);
3018 EXPORT_SYMBOL(proc_dointvec_minmax);
3019 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3020 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3021 EXPORT_SYMBOL(proc_dostring);
3022 EXPORT_SYMBOL(proc_doulongvec_minmax);
3023 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3024 EXPORT_SYMBOL(register_sysctl_table);
3025 EXPORT_SYMBOL(register_sysctl_paths);
3026 EXPORT_SYMBOL(sysctl_intvec);
3027 EXPORT_SYMBOL(sysctl_jiffies);
3028 EXPORT_SYMBOL(sysctl_ms_jiffies);
3029 EXPORT_SYMBOL(sysctl_string);
3030 EXPORT_SYMBOL(sysctl_data);
3031 EXPORT_SYMBOL(unregister_sysctl_table);