projects/linux·blob v2.6.26-rc6 · refs

2851 lines · 67.1 KiB

1/*
2 * sysctl.c: General linux system control interface
3 *
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12 * Horn.
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16 * Wendling.
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19 */
20
21#include <linux/module.h>
22#include <linux/mm.h>
23#include <linux/swap.h>
24#include <linux/slab.h>
25#include <linux/sysctl.h>
26#include <linux/proc_fs.h>
27#include <linux/security.h>
28#include <linux/ctype.h>
29#include <linux/utsname.h>
30#include <linux/smp_lock.h>
31#include <linux/fs.h>
32#include <linux/init.h>
33#include <linux/kernel.h>
34#include <linux/kobject.h>
35#include <linux/net.h>
36#include <linux/sysrq.h>
37#include <linux/highuid.h>
38#include <linux/writeback.h>
39#include <linux/hugetlb.h>
40#include <linux/initrd.h>
41#include <linux/key.h>
42#include <linux/times.h>
43#include <linux/limits.h>
44#include <linux/dcache.h>
45#include <linux/syscalls.h>
46#include <linux/nfs_fs.h>
47#include <linux/acpi.h>
48#include <linux/reboot.h>
49
50#include <asm/uaccess.h>
51#include <asm/processor.h>
52
53#ifdef CONFIG_X86
54#include <asm/nmi.h>
55#include <asm/stacktrace.h>
56#include <asm/io.h>
57#endif
58
59static int deprecated_sysctl_warning(struct __sysctl_args *args);
60
61#if defined(CONFIG_SYSCTL)
62
63/* External variables not in a header file. */
64extern int C_A_D;
65extern int print_fatal_signals;
66extern int sysctl_overcommit_memory;
67extern int sysctl_overcommit_ratio;
68extern int sysctl_panic_on_oom;
69extern int sysctl_oom_kill_allocating_task;
70extern int sysctl_oom_dump_tasks;
71extern int max_threads;
72extern int core_uses_pid;
73extern int suid_dumpable;
74extern char core_pattern[];
75extern int pid_max;
76extern int min_free_kbytes;
77extern int pid_max_min, pid_max_max;
78extern int sysctl_drop_caches;
79extern int percpu_pagelist_fraction;
80extern int compat_log;
81extern int maps_protect;
82extern int sysctl_stat_interval;
83extern int latencytop_enabled;
84extern int sysctl_nr_open_min, sysctl_nr_open_max;
85
86/* Constants used for minimum and maximum */
87#if defined(CONFIG_DETECT_SOFTLOCKUP) || defined(CONFIG_HIGHMEM)
88static int one = 1;
89#endif
90
91#ifdef CONFIG_DETECT_SOFTLOCKUP
92static int sixty = 60;
93#endif
94
95#ifdef CONFIG_MMU
96static int two = 2;
97#endif
98
99static int zero;
100static int one_hundred = 100;
101
102/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
103static int maxolduid = 65535;
104static int minolduid;
105static int min_percpu_pagelist_fract = 8;
106
107static int ngroups_max = NGROUPS_MAX;
108
109#ifdef CONFIG_KMOD
110extern char modprobe_path[];
111#endif
112#ifdef CONFIG_CHR_DEV_SG
113extern int sg_big_buff;
114#endif
115
116#ifdef __sparc__
117extern char reboot_command [];
118extern int stop_a_enabled;
119extern int scons_pwroff;
120#endif
121
122#ifdef __hppa__
123extern int pwrsw_enabled;
124extern int unaligned_enabled;
125#endif
126
127#ifdef CONFIG_S390
128#ifdef CONFIG_MATHEMU
129extern int sysctl_ieee_emulation_warnings;
130#endif
131extern int sysctl_userprocess_debug;
132extern int spin_retry;
133#endif
134
135extern int sysctl_hz_timer;
136
137#ifdef CONFIG_BSD_PROCESS_ACCT
138extern int acct_parm[];
139#endif
140
141#ifdef CONFIG_IA64
142extern int no_unaligned_warning;
143#endif
144
145#ifdef CONFIG_RT_MUTEXES
146extern int max_lock_depth;
147#endif
148
149#ifdef CONFIG_PROC_SYSCTL
150static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
151 void __user *buffer, size_t *lenp, loff_t *ppos);
152static 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
155
156static struct ctl_table root_table[];
157static struct ctl_table_root sysctl_table_root;
158static struct ctl_table_header root_table_header = {
159 .ctl_table = root_table,
160 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.header_list),
161 .root = &sysctl_table_root,
162};
163static struct ctl_table_root sysctl_table_root = {
164 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
165 .header_list = LIST_HEAD_INIT(root_table_header.ctl_entry),
166};
167
168static struct ctl_table kern_table[];
169static struct ctl_table vm_table[];
170static struct ctl_table fs_table[];
171static struct ctl_table debug_table[];
172static struct ctl_table dev_table[];
173extern struct ctl_table random_table[];
174#ifdef CONFIG_INOTIFY_USER
175extern struct ctl_table inotify_table[];
176#endif
177
178#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
179int sysctl_legacy_va_layout;
180#endif
181
182extern int prove_locking;
183extern int lock_stat;
184
185/* The default sysctl tables: */
186
187static struct ctl_table root_table[] = {
188 {
189 .ctl_name = CTL_KERN,
190 .procname = "kernel",
191 .mode = 0555,
192 .child = kern_table,
193 },
194 {
195 .ctl_name = CTL_VM,
196 .procname = "vm",
197 .mode = 0555,
198 .child = vm_table,
199 },
200 {
201 .ctl_name = CTL_FS,
202 .procname = "fs",
203 .mode = 0555,
204 .child = fs_table,
205 },
206 {
207 .ctl_name = CTL_DEBUG,
208 .procname = "debug",
209 .mode = 0555,
210 .child = debug_table,
211 },
212 {
213 .ctl_name = CTL_DEV,
214 .procname = "dev",
215 .mode = 0555,
216 .child = dev_table,
217 },
218/*
219 * NOTE: do not add new entries to this table unless you have read
220 * Documentation/sysctl/ctl_unnumbered.txt
221 */
222 { .ctl_name = 0 }
223};
224
225#ifdef CONFIG_SCHED_DEBUG
226static int min_sched_granularity_ns = 100000; /* 100 usecs */
227static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
228static int min_wakeup_granularity_ns; /* 0 usecs */
229static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
230#endif
231
232static struct ctl_table kern_table[] = {
233#ifdef CONFIG_SCHED_DEBUG
234 {
235 .ctl_name = CTL_UNNUMBERED,
236 .procname = "sched_min_granularity_ns",
237 .data = &sysctl_sched_min_granularity,
238 .maxlen = sizeof(unsigned int),
239 .mode = 0644,
240 .proc_handler = &sched_nr_latency_handler,
241 .strategy = &sysctl_intvec,
242 .extra1 = &min_sched_granularity_ns,
243 .extra2 = &max_sched_granularity_ns,
244 },
245 {
246 .ctl_name = CTL_UNNUMBERED,
247 .procname = "sched_latency_ns",
248 .data = &sysctl_sched_latency,
249 .maxlen = sizeof(unsigned int),
250 .mode = 0644,
251 .proc_handler = &sched_nr_latency_handler,
252 .strategy = &sysctl_intvec,
253 .extra1 = &min_sched_granularity_ns,
254 .extra2 = &max_sched_granularity_ns,
255 },
256 {
257 .ctl_name = CTL_UNNUMBERED,
258 .procname = "sched_wakeup_granularity_ns",
259 .data = &sysctl_sched_wakeup_granularity,
260 .maxlen = sizeof(unsigned int),
261 .mode = 0644,
262 .proc_handler = &proc_dointvec_minmax,
263 .strategy = &sysctl_intvec,
264 .extra1 = &min_wakeup_granularity_ns,
265 .extra2 = &max_wakeup_granularity_ns,
266 },
267 {
268 .ctl_name = CTL_UNNUMBERED,
269 .procname = "sched_child_runs_first",
270 .data = &sysctl_sched_child_runs_first,
271 .maxlen = sizeof(unsigned int),
272 .mode = 0644,
273 .proc_handler = &proc_dointvec,
274 },
275 {
276 .ctl_name = CTL_UNNUMBERED,
277 .procname = "sched_features",
278 .data = &sysctl_sched_features,
279 .maxlen = sizeof(unsigned int),
280 .mode = 0644,
281 .proc_handler = &proc_dointvec,
282 },
283 {
284 .ctl_name = CTL_UNNUMBERED,
285 .procname = "sched_migration_cost",
286 .data = &sysctl_sched_migration_cost,
287 .maxlen = sizeof(unsigned int),
288 .mode = 0644,
289 .proc_handler = &proc_dointvec,
290 },
291 {
292 .ctl_name = CTL_UNNUMBERED,
293 .procname = "sched_nr_migrate",
294 .data = &sysctl_sched_nr_migrate,
295 .maxlen = sizeof(unsigned int),
296 .mode = 0644,
297 .proc_handler = &proc_dointvec,
298 },
299#endif
300 {
301 .ctl_name = CTL_UNNUMBERED,
302 .procname = "sched_rt_period_us",
303 .data = &sysctl_sched_rt_period,
304 .maxlen = sizeof(unsigned int),
305 .mode = 0644,
306 .proc_handler = &sched_rt_handler,
307 },
308 {
309 .ctl_name = CTL_UNNUMBERED,
310 .procname = "sched_rt_runtime_us",
311 .data = &sysctl_sched_rt_runtime,
312 .maxlen = sizeof(int),
313 .mode = 0644,
314 .proc_handler = &sched_rt_handler,
315 },
316 {
317 .ctl_name = CTL_UNNUMBERED,
318 .procname = "sched_compat_yield",
319 .data = &sysctl_sched_compat_yield,
320 .maxlen = sizeof(unsigned int),
321 .mode = 0644,
322 .proc_handler = &proc_dointvec,
323 },
324#ifdef CONFIG_PROVE_LOCKING
325 {
326 .ctl_name = CTL_UNNUMBERED,
327 .procname = "prove_locking",
328 .data = &prove_locking,
329 .maxlen = sizeof(int),
330 .mode = 0644,
331 .proc_handler = &proc_dointvec,
332 },
333#endif
334#ifdef CONFIG_LOCK_STAT
335 {
336 .ctl_name = CTL_UNNUMBERED,
337 .procname = "lock_stat",
338 .data = &lock_stat,
339 .maxlen = sizeof(int),
340 .mode = 0644,
341 .proc_handler = &proc_dointvec,
342 },
343#endif
344 {
345 .ctl_name = KERN_PANIC,
346 .procname = "panic",
347 .data = &panic_timeout,
348 .maxlen = sizeof(int),
349 .mode = 0644,
350 .proc_handler = &proc_dointvec,
351 },
352 {
353 .ctl_name = KERN_CORE_USES_PID,
354 .procname = "core_uses_pid",
355 .data = &core_uses_pid,
356 .maxlen = sizeof(int),
357 .mode = 0644,
358 .proc_handler = &proc_dointvec,
359 },
360 {
361 .ctl_name = KERN_CORE_PATTERN,
362 .procname = "core_pattern",
363 .data = core_pattern,
364 .maxlen = CORENAME_MAX_SIZE,
365 .mode = 0644,
366 .proc_handler = &proc_dostring,
367 .strategy = &sysctl_string,
368 },
369#ifdef CONFIG_PROC_SYSCTL
370 {
371 .procname = "tainted",
372 .data = &tainted,
373 .maxlen = sizeof(int),
374 .mode = 0644,
375 .proc_handler = &proc_dointvec_taint,
376 },
377#endif
378#ifdef CONFIG_LATENCYTOP
379 {
380 .procname = "latencytop",
381 .data = &latencytop_enabled,
382 .maxlen = sizeof(int),
383 .mode = 0644,
384 .proc_handler = &proc_dointvec,
385 },
386#endif
387#ifdef CONFIG_BLK_DEV_INITRD
388 {
389 .ctl_name = KERN_REALROOTDEV,
390 .procname = "real-root-dev",
391 .data = &real_root_dev,
392 .maxlen = sizeof(int),
393 .mode = 0644,
394 .proc_handler = &proc_dointvec,
395 },
396#endif
397 {
398 .ctl_name = CTL_UNNUMBERED,
399 .procname = "print-fatal-signals",
400 .data = &print_fatal_signals,
401 .maxlen = sizeof(int),
402 .mode = 0644,
403 .proc_handler = &proc_dointvec,
404 },
405#ifdef __sparc__
406 {
407 .ctl_name = KERN_SPARC_REBOOT,
408 .procname = "reboot-cmd",
409 .data = reboot_command,
410 .maxlen = 256,
411 .mode = 0644,
412 .proc_handler = &proc_dostring,
413 .strategy = &sysctl_string,
414 },
415 {
416 .ctl_name = KERN_SPARC_STOP_A,
417 .procname = "stop-a",
418 .data = &stop_a_enabled,
419 .maxlen = sizeof (int),
420 .mode = 0644,
421 .proc_handler = &proc_dointvec,
422 },
423 {
424 .ctl_name = KERN_SPARC_SCONS_PWROFF,
425 .procname = "scons-poweroff",
426 .data = &scons_pwroff,
427 .maxlen = sizeof (int),
428 .mode = 0644,
429 .proc_handler = &proc_dointvec,
430 },
431#endif
432#ifdef __hppa__
433 {
434 .ctl_name = KERN_HPPA_PWRSW,
435 .procname = "soft-power",
436 .data = &pwrsw_enabled,
437 .maxlen = sizeof (int),
438 .mode = 0644,
439 .proc_handler = &proc_dointvec,
440 },
441 {
442 .ctl_name = KERN_HPPA_UNALIGNED,
443 .procname = "unaligned-trap",
444 .data = &unaligned_enabled,
445 .maxlen = sizeof (int),
446 .mode = 0644,
447 .proc_handler = &proc_dointvec,
448 },
449#endif
450 {
451 .ctl_name = KERN_CTLALTDEL,
452 .procname = "ctrl-alt-del",
453 .data = &C_A_D,
454 .maxlen = sizeof(int),
455 .mode = 0644,
456 .proc_handler = &proc_dointvec,
457 },
458#ifdef CONFIG_KMOD
459 {
460 .ctl_name = KERN_MODPROBE,
461 .procname = "modprobe",
462 .data = &modprobe_path,
463 .maxlen = KMOD_PATH_LEN,
464 .mode = 0644,
465 .proc_handler = &proc_dostring,
466 .strategy = &sysctl_string,
467 },
468#endif
469#if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
470 {
471 .ctl_name = KERN_HOTPLUG,
472 .procname = "hotplug",
473 .data = &uevent_helper,
474 .maxlen = UEVENT_HELPER_PATH_LEN,
475 .mode = 0644,
476 .proc_handler = &proc_dostring,
477 .strategy = &sysctl_string,
478 },
479#endif
480#ifdef CONFIG_CHR_DEV_SG
481 {
482 .ctl_name = KERN_SG_BIG_BUFF,
483 .procname = "sg-big-buff",
484 .data = &sg_big_buff,
485 .maxlen = sizeof (int),
486 .mode = 0444,
487 .proc_handler = &proc_dointvec,
488 },
489#endif
490#ifdef CONFIG_BSD_PROCESS_ACCT
491 {
492 .ctl_name = KERN_ACCT,
493 .procname = "acct",
494 .data = &acct_parm,
495 .maxlen = 3*sizeof(int),
496 .mode = 0644,
497 .proc_handler = &proc_dointvec,
498 },
499#endif
500#ifdef CONFIG_MAGIC_SYSRQ
501 {
502 .ctl_name = KERN_SYSRQ,
503 .procname = "sysrq",
504 .data = &__sysrq_enabled,
505 .maxlen = sizeof (int),
506 .mode = 0644,
507 .proc_handler = &proc_dointvec,
508 },
509#endif
510#ifdef CONFIG_PROC_SYSCTL
511 {
512 .procname = "cad_pid",
513 .data = NULL,
514 .maxlen = sizeof (int),
515 .mode = 0600,
516 .proc_handler = &proc_do_cad_pid,
517 },
518#endif
519 {
520 .ctl_name = KERN_MAX_THREADS,
521 .procname = "threads-max",
522 .data = &max_threads,
523 .maxlen = sizeof(int),
524 .mode = 0644,
525 .proc_handler = &proc_dointvec,
526 },
527 {
528 .ctl_name = KERN_RANDOM,
529 .procname = "random",
530 .mode = 0555,
531 .child = random_table,
532 },
533 {
534 .ctl_name = KERN_OVERFLOWUID,
535 .procname = "overflowuid",
536 .data = &overflowuid,
537 .maxlen = sizeof(int),
538 .mode = 0644,
539 .proc_handler = &proc_dointvec_minmax,
540 .strategy = &sysctl_intvec,
541 .extra1 = &minolduid,
542 .extra2 = &maxolduid,
543 },
544 {
545 .ctl_name = KERN_OVERFLOWGID,
546 .procname = "overflowgid",
547 .data = &overflowgid,
548 .maxlen = sizeof(int),
549 .mode = 0644,
550 .proc_handler = &proc_dointvec_minmax,
551 .strategy = &sysctl_intvec,
552 .extra1 = &minolduid,
553 .extra2 = &maxolduid,
554 },
555#ifdef CONFIG_S390
556#ifdef CONFIG_MATHEMU
557 {
558 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
559 .procname = "ieee_emulation_warnings",
560 .data = &sysctl_ieee_emulation_warnings,
561 .maxlen = sizeof(int),
562 .mode = 0644,
563 .proc_handler = &proc_dointvec,
564 },
565#endif
566#ifdef CONFIG_NO_IDLE_HZ
567 {
568 .ctl_name = KERN_HZ_TIMER,
569 .procname = "hz_timer",
570 .data = &sysctl_hz_timer,
571 .maxlen = sizeof(int),
572 .mode = 0644,
573 .proc_handler = &proc_dointvec,
574 },
575#endif
576 {
577 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
578 .procname = "userprocess_debug",
579 .data = &sysctl_userprocess_debug,
580 .maxlen = sizeof(int),
581 .mode = 0644,
582 .proc_handler = &proc_dointvec,
583 },
584#endif
585 {
586 .ctl_name = KERN_PIDMAX,
587 .procname = "pid_max",
588 .data = &pid_max,
589 .maxlen = sizeof (int),
590 .mode = 0644,
591 .proc_handler = &proc_dointvec_minmax,
592 .strategy = sysctl_intvec,
593 .extra1 = &pid_max_min,
594 .extra2 = &pid_max_max,
595 },
596 {
597 .ctl_name = KERN_PANIC_ON_OOPS,
598 .procname = "panic_on_oops",
599 .data = &panic_on_oops,
600 .maxlen = sizeof(int),
601 .mode = 0644,
602 .proc_handler = &proc_dointvec,
603 },
604#if defined CONFIG_PRINTK
605 {
606 .ctl_name = KERN_PRINTK,
607 .procname = "printk",
608 .data = &console_loglevel,
609 .maxlen = 4*sizeof(int),
610 .mode = 0644,
611 .proc_handler = &proc_dointvec,
612 },
613 {
614 .ctl_name = KERN_PRINTK_RATELIMIT,
615 .procname = "printk_ratelimit",
616 .data = &printk_ratelimit_jiffies,
617 .maxlen = sizeof(int),
618 .mode = 0644,
619 .proc_handler = &proc_dointvec_jiffies,
620 .strategy = &sysctl_jiffies,
621 },
622 {
623 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
624 .procname = "printk_ratelimit_burst",
625 .data = &printk_ratelimit_burst,
626 .maxlen = sizeof(int),
627 .mode = 0644,
628 .proc_handler = &proc_dointvec,
629 },
630#endif
631 {
632 .ctl_name = KERN_NGROUPS_MAX,
633 .procname = "ngroups_max",
634 .data = &ngroups_max,
635 .maxlen = sizeof (int),
636 .mode = 0444,
637 .proc_handler = &proc_dointvec,
638 },
639#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
640 {
641 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
642 .procname = "unknown_nmi_panic",
643 .data = &unknown_nmi_panic,
644 .maxlen = sizeof (int),
645 .mode = 0644,
646 .proc_handler = &proc_dointvec,
647 },
648 {
649 .procname = "nmi_watchdog",
650 .data = &nmi_watchdog_enabled,
651 .maxlen = sizeof (int),
652 .mode = 0644,
653 .proc_handler = &proc_nmi_enabled,
654 },
655#endif
656#if defined(CONFIG_X86)
657 {
658 .ctl_name = KERN_PANIC_ON_NMI,
659 .procname = "panic_on_unrecovered_nmi",
660 .data = &panic_on_unrecovered_nmi,
661 .maxlen = sizeof(int),
662 .mode = 0644,
663 .proc_handler = &proc_dointvec,
664 },
665 {
666 .ctl_name = KERN_BOOTLOADER_TYPE,
667 .procname = "bootloader_type",
668 .data = &bootloader_type,
669 .maxlen = sizeof (int),
670 .mode = 0444,
671 .proc_handler = &proc_dointvec,
672 },
673 {
674 .ctl_name = CTL_UNNUMBERED,
675 .procname = "kstack_depth_to_print",
676 .data = &kstack_depth_to_print,
677 .maxlen = sizeof(int),
678 .mode = 0644,
679 .proc_handler = &proc_dointvec,
680 },
681 {
682 .ctl_name = CTL_UNNUMBERED,
683 .procname = "io_delay_type",
684 .data = &io_delay_type,
685 .maxlen = sizeof(int),
686 .mode = 0644,
687 .proc_handler = &proc_dointvec,
688 },
689#endif
690#if defined(CONFIG_MMU)
691 {
692 .ctl_name = KERN_RANDOMIZE,
693 .procname = "randomize_va_space",
694 .data = &randomize_va_space,
695 .maxlen = sizeof(int),
696 .mode = 0644,
697 .proc_handler = &proc_dointvec,
698 },
699#endif
700#if defined(CONFIG_S390) && defined(CONFIG_SMP)
701 {
702 .ctl_name = KERN_SPIN_RETRY,
703 .procname = "spin_retry",
704 .data = &spin_retry,
705 .maxlen = sizeof (int),
706 .mode = 0644,
707 .proc_handler = &proc_dointvec,
708 },
709#endif
710#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
711 {
712 .procname = "acpi_video_flags",
713 .data = &acpi_realmode_flags,
714 .maxlen = sizeof (unsigned long),
715 .mode = 0644,
716 .proc_handler = &proc_doulongvec_minmax,
717 },
718#endif
719#ifdef CONFIG_IA64
720 {
721 .ctl_name = KERN_IA64_UNALIGNED,
722 .procname = "ignore-unaligned-usertrap",
723 .data = &no_unaligned_warning,
724 .maxlen = sizeof (int),
725 .mode = 0644,
726 .proc_handler = &proc_dointvec,
727 },
728#endif
729#ifdef CONFIG_DETECT_SOFTLOCKUP
730 {
731 .ctl_name = CTL_UNNUMBERED,
732 .procname = "softlockup_thresh",
733 .data = &softlockup_thresh,
734 .maxlen = sizeof(unsigned long),
735 .mode = 0644,
736 .proc_handler = &proc_doulongvec_minmax,
737 .strategy = &sysctl_intvec,
738 .extra1 = &one,
739 .extra2 = &sixty,
740 },
741 {
742 .ctl_name = CTL_UNNUMBERED,
743 .procname = "hung_task_check_count",
744 .data = &sysctl_hung_task_check_count,
745 .maxlen = sizeof(unsigned long),
746 .mode = 0644,
747 .proc_handler = &proc_doulongvec_minmax,
748 .strategy = &sysctl_intvec,
749 },
750 {
751 .ctl_name = CTL_UNNUMBERED,
752 .procname = "hung_task_timeout_secs",
753 .data = &sysctl_hung_task_timeout_secs,
754 .maxlen = sizeof(unsigned long),
755 .mode = 0644,
756 .proc_handler = &proc_doulongvec_minmax,
757 .strategy = &sysctl_intvec,
758 },
759 {
760 .ctl_name = CTL_UNNUMBERED,
761 .procname = "hung_task_warnings",
762 .data = &sysctl_hung_task_warnings,
763 .maxlen = sizeof(unsigned long),
764 .mode = 0644,
765 .proc_handler = &proc_doulongvec_minmax,
766 .strategy = &sysctl_intvec,
767 },
768#endif
769#ifdef CONFIG_COMPAT
770 {
771 .ctl_name = KERN_COMPAT_LOG,
772 .procname = "compat-log",
773 .data = &compat_log,
774 .maxlen = sizeof (int),
775 .mode = 0644,
776 .proc_handler = &proc_dointvec,
777 },
778#endif
779#ifdef CONFIG_RT_MUTEXES
780 {
781 .ctl_name = KERN_MAX_LOCK_DEPTH,
782 .procname = "max_lock_depth",
783 .data = &max_lock_depth,
784 .maxlen = sizeof(int),
785 .mode = 0644,
786 .proc_handler = &proc_dointvec,
787 },
788#endif
789#ifdef CONFIG_PROC_FS
790 {
791 .ctl_name = CTL_UNNUMBERED,
792 .procname = "maps_protect",
793 .data = &maps_protect,
794 .maxlen = sizeof(int),
795 .mode = 0644,
796 .proc_handler = &proc_dointvec,
797 },
798#endif
799 {
800 .ctl_name = CTL_UNNUMBERED,
801 .procname = "poweroff_cmd",
802 .data = &poweroff_cmd,
803 .maxlen = POWEROFF_CMD_PATH_LEN,
804 .mode = 0644,
805 .proc_handler = &proc_dostring,
806 .strategy = &sysctl_string,
807 },
808#ifdef CONFIG_KEYS
809 {
810 .ctl_name = CTL_UNNUMBERED,
811 .procname = "keys",
812 .mode = 0555,
813 .child = key_sysctls,
814 },
815#endif
816/*
817 * NOTE: do not add new entries to this table unless you have read
818 * Documentation/sysctl/ctl_unnumbered.txt
819 */
820 { .ctl_name = 0 }
821};
822
823static struct ctl_table vm_table[] = {
824 {
825 .ctl_name = VM_OVERCOMMIT_MEMORY,
826 .procname = "overcommit_memory",
827 .data = &sysctl_overcommit_memory,
828 .maxlen = sizeof(sysctl_overcommit_memory),
829 .mode = 0644,
830 .proc_handler = &proc_dointvec,
831 },
832 {
833 .ctl_name = VM_PANIC_ON_OOM,
834 .procname = "panic_on_oom",
835 .data = &sysctl_panic_on_oom,
836 .maxlen = sizeof(sysctl_panic_on_oom),
837 .mode = 0644,
838 .proc_handler = &proc_dointvec,
839 },
840 {
841 .ctl_name = CTL_UNNUMBERED,
842 .procname = "oom_kill_allocating_task",
843 .data = &sysctl_oom_kill_allocating_task,
844 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
845 .mode = 0644,
846 .proc_handler = &proc_dointvec,
847 },
848 {
849 .ctl_name = CTL_UNNUMBERED,
850 .procname = "oom_dump_tasks",
851 .data = &sysctl_oom_dump_tasks,
852 .maxlen = sizeof(sysctl_oom_dump_tasks),
853 .mode = 0644,
854 .proc_handler = &proc_dointvec,
855 },
856 {
857 .ctl_name = VM_OVERCOMMIT_RATIO,
858 .procname = "overcommit_ratio",
859 .data = &sysctl_overcommit_ratio,
860 .maxlen = sizeof(sysctl_overcommit_ratio),
861 .mode = 0644,
862 .proc_handler = &proc_dointvec,
863 },
864 {
865 .ctl_name = VM_PAGE_CLUSTER,
866 .procname = "page-cluster",
867 .data = &page_cluster,
868 .maxlen = sizeof(int),
869 .mode = 0644,
870 .proc_handler = &proc_dointvec,
871 },
872 {
873 .ctl_name = VM_DIRTY_BACKGROUND,
874 .procname = "dirty_background_ratio",
875 .data = &dirty_background_ratio,
876 .maxlen = sizeof(dirty_background_ratio),
877 .mode = 0644,
878 .proc_handler = &proc_dointvec_minmax,
879 .strategy = &sysctl_intvec,
880 .extra1 = &zero,
881 .extra2 = &one_hundred,
882 },
883 {
884 .ctl_name = VM_DIRTY_RATIO,
885 .procname = "dirty_ratio",
886 .data = &vm_dirty_ratio,
887 .maxlen = sizeof(vm_dirty_ratio),
888 .mode = 0644,
889 .proc_handler = &dirty_ratio_handler,
890 .strategy = &sysctl_intvec,
891 .extra1 = &zero,
892 .extra2 = &one_hundred,
893 },
894 {
895 .procname = "dirty_writeback_centisecs",
896 .data = &dirty_writeback_interval,
897 .maxlen = sizeof(dirty_writeback_interval),
898 .mode = 0644,
899 .proc_handler = &dirty_writeback_centisecs_handler,
900 },
901 {
902 .procname = "dirty_expire_centisecs",
903 .data = &dirty_expire_interval,
904 .maxlen = sizeof(dirty_expire_interval),
905 .mode = 0644,
906 .proc_handler = &proc_dointvec_userhz_jiffies,
907 },
908 {
909 .ctl_name = VM_NR_PDFLUSH_THREADS,
910 .procname = "nr_pdflush_threads",
911 .data = &nr_pdflush_threads,
912 .maxlen = sizeof nr_pdflush_threads,
913 .mode = 0444 /* read-only*/,
914 .proc_handler = &proc_dointvec,
915 },
916 {
917 .ctl_name = VM_SWAPPINESS,
918 .procname = "swappiness",
919 .data = &vm_swappiness,
920 .maxlen = sizeof(vm_swappiness),
921 .mode = 0644,
922 .proc_handler = &proc_dointvec_minmax,
923 .strategy = &sysctl_intvec,
924 .extra1 = &zero,
925 .extra2 = &one_hundred,
926 },
927#ifdef CONFIG_HUGETLB_PAGE
928 {
929 .procname = "nr_hugepages",
930 .data = &max_huge_pages,
931 .maxlen = sizeof(unsigned long),
932 .mode = 0644,
933 .proc_handler = &hugetlb_sysctl_handler,
934 .extra1 = (void *)&hugetlb_zero,
935 .extra2 = (void *)&hugetlb_infinity,
936 },
937 {
938 .ctl_name = VM_HUGETLB_GROUP,
939 .procname = "hugetlb_shm_group",
940 .data = &sysctl_hugetlb_shm_group,
941 .maxlen = sizeof(gid_t),
942 .mode = 0644,
943 .proc_handler = &proc_dointvec,
944 },
945 {
946 .ctl_name = CTL_UNNUMBERED,
947 .procname = "hugepages_treat_as_movable",
948 .data = &hugepages_treat_as_movable,
949 .maxlen = sizeof(int),
950 .mode = 0644,
951 .proc_handler = &hugetlb_treat_movable_handler,
952 },
953 {
954 .ctl_name = CTL_UNNUMBERED,
955 .procname = "nr_overcommit_hugepages",
956 .data = &sysctl_overcommit_huge_pages,
957 .maxlen = sizeof(sysctl_overcommit_huge_pages),
958 .mode = 0644,
959 .proc_handler = &hugetlb_overcommit_handler,
960 },
961#endif
962 {
963 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
964 .procname = "lowmem_reserve_ratio",
965 .data = &sysctl_lowmem_reserve_ratio,
966 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
967 .mode = 0644,
968 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
969 .strategy = &sysctl_intvec,
970 },
971 {
972 .ctl_name = VM_DROP_PAGECACHE,
973 .procname = "drop_caches",
974 .data = &sysctl_drop_caches,
975 .maxlen = sizeof(int),
976 .mode = 0644,
977 .proc_handler = drop_caches_sysctl_handler,
978 .strategy = &sysctl_intvec,
979 },
980 {
981 .ctl_name = VM_MIN_FREE_KBYTES,
982 .procname = "min_free_kbytes",
983 .data = &min_free_kbytes,
984 .maxlen = sizeof(min_free_kbytes),
985 .mode = 0644,
986 .proc_handler = &min_free_kbytes_sysctl_handler,
987 .strategy = &sysctl_intvec,
988 .extra1 = &zero,
989 },
990 {
991 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
992 .procname = "percpu_pagelist_fraction",
993 .data = &percpu_pagelist_fraction,
994 .maxlen = sizeof(percpu_pagelist_fraction),
995 .mode = 0644,
996 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
997 .strategy = &sysctl_intvec,
998 .extra1 = &min_percpu_pagelist_fract,
999 },
1000#ifdef CONFIG_MMU
1001 {
1002 .ctl_name = VM_MAX_MAP_COUNT,
1003 .procname = "max_map_count",
1004 .data = &sysctl_max_map_count,
1005 .maxlen = sizeof(sysctl_max_map_count),
1006 .mode = 0644,
1007 .proc_handler = &proc_dointvec
1008 },
1009#endif
1010 {
1011 .ctl_name = VM_LAPTOP_MODE,
1012 .procname = "laptop_mode",
1013 .data = &laptop_mode,
1014 .maxlen = sizeof(laptop_mode),
1015 .mode = 0644,
1016 .proc_handler = &proc_dointvec_jiffies,
1017 .strategy = &sysctl_jiffies,
1018 },
1019 {
1020 .ctl_name = VM_BLOCK_DUMP,
1021 .procname = "block_dump",
1022 .data = &block_dump,
1023 .maxlen = sizeof(block_dump),
1024 .mode = 0644,
1025 .proc_handler = &proc_dointvec,
1026 .strategy = &sysctl_intvec,
1027 .extra1 = &zero,
1028 },
1029 {
1030 .ctl_name = VM_VFS_CACHE_PRESSURE,
1031 .procname = "vfs_cache_pressure",
1032 .data = &sysctl_vfs_cache_pressure,
1033 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1034 .mode = 0644,
1035 .proc_handler = &proc_dointvec,
1036 .strategy = &sysctl_intvec,
1037 .extra1 = &zero,
1038 },
1039#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1040 {
1041 .ctl_name = VM_LEGACY_VA_LAYOUT,
1042 .procname = "legacy_va_layout",
1043 .data = &sysctl_legacy_va_layout,
1044 .maxlen = sizeof(sysctl_legacy_va_layout),
1045 .mode = 0644,
1046 .proc_handler = &proc_dointvec,
1047 .strategy = &sysctl_intvec,
1048 .extra1 = &zero,
1049 },
1050#endif
1051#ifdef CONFIG_NUMA
1052 {
1053 .ctl_name = VM_ZONE_RECLAIM_MODE,
1054 .procname = "zone_reclaim_mode",
1055 .data = &zone_reclaim_mode,
1056 .maxlen = sizeof(zone_reclaim_mode),
1057 .mode = 0644,
1058 .proc_handler = &proc_dointvec,
1059 .strategy = &sysctl_intvec,
1060 .extra1 = &zero,
1061 },
1062 {
1063 .ctl_name = VM_MIN_UNMAPPED,
1064 .procname = "min_unmapped_ratio",
1065 .data = &sysctl_min_unmapped_ratio,
1066 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1067 .mode = 0644,
1068 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
1069 .strategy = &sysctl_intvec,
1070 .extra1 = &zero,
1071 .extra2 = &one_hundred,
1072 },
1073 {
1074 .ctl_name = VM_MIN_SLAB,
1075 .procname = "min_slab_ratio",
1076 .data = &sysctl_min_slab_ratio,
1077 .maxlen = sizeof(sysctl_min_slab_ratio),
1078 .mode = 0644,
1079 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1080 .strategy = &sysctl_intvec,
1081 .extra1 = &zero,
1082 .extra2 = &one_hundred,
1083 },
1084#endif
1085#ifdef CONFIG_SMP
1086 {
1087 .ctl_name = CTL_UNNUMBERED,
1088 .procname = "stat_interval",
1089 .data = &sysctl_stat_interval,
1090 .maxlen = sizeof(sysctl_stat_interval),
1091 .mode = 0644,
1092 .proc_handler = &proc_dointvec_jiffies,
1093 .strategy = &sysctl_jiffies,
1094 },
1095#endif
1096#ifdef CONFIG_SECURITY
1097 {
1098 .ctl_name = CTL_UNNUMBERED,
1099 .procname = "mmap_min_addr",
1100 .data = &mmap_min_addr,
1101 .maxlen = sizeof(unsigned long),
1102 .mode = 0644,
1103 .proc_handler = &proc_doulongvec_minmax,
1104 },
1105#endif
1106#ifdef CONFIG_NUMA
1107 {
1108 .ctl_name = CTL_UNNUMBERED,
1109 .procname = "numa_zonelist_order",
1110 .data = &numa_zonelist_order,
1111 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1112 .mode = 0644,
1113 .proc_handler = &numa_zonelist_order_handler,
1114 .strategy = &sysctl_string,
1115 },
1116#endif
1117#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1118 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1119 {
1120 .ctl_name = VM_VDSO_ENABLED,
1121 .procname = "vdso_enabled",
1122 .data = &vdso_enabled,
1123 .maxlen = sizeof(vdso_enabled),
1124 .mode = 0644,
1125 .proc_handler = &proc_dointvec,
1126 .strategy = &sysctl_intvec,
1127 .extra1 = &zero,
1128 },
1129#endif
1130#ifdef CONFIG_HIGHMEM
1131 {
1132 .ctl_name = CTL_UNNUMBERED,
1133 .procname = "highmem_is_dirtyable",
1134 .data = &vm_highmem_is_dirtyable,
1135 .maxlen = sizeof(vm_highmem_is_dirtyable),
1136 .mode = 0644,
1137 .proc_handler = &proc_dointvec_minmax,
1138 .strategy = &sysctl_intvec,
1139 .extra1 = &zero,
1140 .extra2 = &one,
1141 },
1142#endif
1143/*
1144 * NOTE: do not add new entries to this table unless you have read
1145 * Documentation/sysctl/ctl_unnumbered.txt
1146 */
1147 { .ctl_name = 0 }
1148};
1149
1150#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1151static struct ctl_table binfmt_misc_table[] = {
1152 { .ctl_name = 0 }
1153};
1154#endif
1155
1156static struct ctl_table fs_table[] = {
1157 {
1158 .ctl_name = FS_NRINODE,
1159 .procname = "inode-nr",
1160 .data = &inodes_stat,
1161 .maxlen = 2*sizeof(int),
1162 .mode = 0444,
1163 .proc_handler = &proc_dointvec,
1164 },
1165 {
1166 .ctl_name = FS_STATINODE,
1167 .procname = "inode-state",
1168 .data = &inodes_stat,
1169 .maxlen = 7*sizeof(int),
1170 .mode = 0444,
1171 .proc_handler = &proc_dointvec,
1172 },
1173 {
1174 .procname = "file-nr",
1175 .data = &files_stat,
1176 .maxlen = 3*sizeof(int),
1177 .mode = 0444,
1178 .proc_handler = &proc_nr_files,
1179 },
1180 {
1181 .ctl_name = FS_MAXFILE,
1182 .procname = "file-max",
1183 .data = &files_stat.max_files,
1184 .maxlen = sizeof(int),
1185 .mode = 0644,
1186 .proc_handler = &proc_dointvec,
1187 },
1188 {
1189 .ctl_name = CTL_UNNUMBERED,
1190 .procname = "nr_open",
1191 .data = &sysctl_nr_open,
1192 .maxlen = sizeof(int),
1193 .mode = 0644,
1194 .proc_handler = &proc_dointvec_minmax,
1195 .extra1 = &sysctl_nr_open_min,
1196 .extra2 = &sysctl_nr_open_max,
1197 },
1198 {
1199 .ctl_name = FS_DENTRY,
1200 .procname = "dentry-state",
1201 .data = &dentry_stat,
1202 .maxlen = 6*sizeof(int),
1203 .mode = 0444,
1204 .proc_handler = &proc_dointvec,
1205 },
1206 {
1207 .ctl_name = FS_OVERFLOWUID,
1208 .procname = "overflowuid",
1209 .data = &fs_overflowuid,
1210 .maxlen = sizeof(int),
1211 .mode = 0644,
1212 .proc_handler = &proc_dointvec_minmax,
1213 .strategy = &sysctl_intvec,
1214 .extra1 = &minolduid,
1215 .extra2 = &maxolduid,
1216 },
1217 {
1218 .ctl_name = FS_OVERFLOWGID,
1219 .procname = "overflowgid",
1220 .data = &fs_overflowgid,
1221 .maxlen = sizeof(int),
1222 .mode = 0644,
1223 .proc_handler = &proc_dointvec_minmax,
1224 .strategy = &sysctl_intvec,
1225 .extra1 = &minolduid,
1226 .extra2 = &maxolduid,
1227 },
1228 {
1229 .ctl_name = FS_LEASES,
1230 .procname = "leases-enable",
1231 .data = &leases_enable,
1232 .maxlen = sizeof(int),
1233 .mode = 0644,
1234 .proc_handler = &proc_dointvec,
1235 },
1236#ifdef CONFIG_DNOTIFY
1237 {
1238 .ctl_name = FS_DIR_NOTIFY,
1239 .procname = "dir-notify-enable",
1240 .data = &dir_notify_enable,
1241 .maxlen = sizeof(int),
1242 .mode = 0644,
1243 .proc_handler = &proc_dointvec,
1244 },
1245#endif
1246#ifdef CONFIG_MMU
1247 {
1248 .ctl_name = FS_LEASE_TIME,
1249 .procname = "lease-break-time",
1250 .data = &lease_break_time,
1251 .maxlen = sizeof(int),
1252 .mode = 0644,
1253 .proc_handler = &proc_dointvec_minmax,
1254 .strategy = &sysctl_intvec,
1255 .extra1 = &zero,
1256 .extra2 = &two,
1257 },
1258 {
1259 .procname = "aio-nr",
1260 .data = &aio_nr,
1261 .maxlen = sizeof(aio_nr),
1262 .mode = 0444,
1263 .proc_handler = &proc_doulongvec_minmax,
1264 },
1265 {
1266 .procname = "aio-max-nr",
1267 .data = &aio_max_nr,
1268 .maxlen = sizeof(aio_max_nr),
1269 .mode = 0644,
1270 .proc_handler = &proc_doulongvec_minmax,
1271 },
1272#ifdef CONFIG_INOTIFY_USER
1273 {
1274 .ctl_name = FS_INOTIFY,
1275 .procname = "inotify",
1276 .mode = 0555,
1277 .child = inotify_table,
1278 },
1279#endif
1280#endif
1281 {
1282 .ctl_name = KERN_SETUID_DUMPABLE,
1283 .procname = "suid_dumpable",
1284 .data = &suid_dumpable,
1285 .maxlen = sizeof(int),
1286 .mode = 0644,
1287 .proc_handler = &proc_dointvec,
1288 },
1289#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1290 {
1291 .ctl_name = CTL_UNNUMBERED,
1292 .procname = "binfmt_misc",
1293 .mode = 0555,
1294 .child = binfmt_misc_table,
1295 },
1296#endif
1297/*
1298 * NOTE: do not add new entries to this table unless you have read
1299 * Documentation/sysctl/ctl_unnumbered.txt
1300 */
1301 { .ctl_name = 0 }
1302};
1303
1304static struct ctl_table debug_table[] = {
1305#if defined(CONFIG_X86) || defined(CONFIG_PPC)
1306 {
1307 .ctl_name = CTL_UNNUMBERED,
1308 .procname = "exception-trace",
1309 .data = &show_unhandled_signals,
1310 .maxlen = sizeof(int),
1311 .mode = 0644,
1312 .proc_handler = proc_dointvec
1313 },
1314#endif
1315 { .ctl_name = 0 }
1316};
1317
1318static struct ctl_table dev_table[] = {
1319 { .ctl_name = 0 }
1320};
1321
1322static DEFINE_SPINLOCK(sysctl_lock);
1323
1324/* called under sysctl_lock */
1325static int use_table(struct ctl_table_header *p)
1326{
1327 if (unlikely(p->unregistering))
1328 return 0;
1329 p->used++;
1330 return 1;
1331}
1332
1333/* called under sysctl_lock */
1334static void unuse_table(struct ctl_table_header *p)
1335{
1336 if (!--p->used)
1337 if (unlikely(p->unregistering))
1338 complete(p->unregistering);
1339}
1340
1341/* called under sysctl_lock, will reacquire if has to wait */
1342static void start_unregistering(struct ctl_table_header *p)
1343{
1344 /*
1345 * if p->used is 0, nobody will ever touch that entry again;
1346 * we'll eliminate all paths to it before dropping sysctl_lock
1347 */
1348 if (unlikely(p->used)) {
1349 struct completion wait;
1350 init_completion(&wait);
1351 p->unregistering = &wait;
1352 spin_unlock(&sysctl_lock);
1353 wait_for_completion(&wait);
1354 spin_lock(&sysctl_lock);
1355 }
1356 /*
1357 * do not remove from the list until nobody holds it; walking the
1358 * list in do_sysctl() relies on that.
1359 */
1360 list_del_init(&p->ctl_entry);
1361}
1362
1363void sysctl_head_finish(struct ctl_table_header *head)
1364{
1365 if (!head)
1366 return;
1367 spin_lock(&sysctl_lock);
1368 unuse_table(head);
1369 spin_unlock(&sysctl_lock);
1370}
1371
1372static struct list_head *
1373lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1374{
1375 struct list_head *header_list;
1376 header_list = &root->header_list;
1377 if (root->lookup)
1378 header_list = root->lookup(root, namespaces);
1379 return header_list;
1380}
1381
1382struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1383 struct ctl_table_header *prev)
1384{
1385 struct ctl_table_root *root;
1386 struct list_head *header_list;
1387 struct ctl_table_header *head;
1388 struct list_head *tmp;
1389
1390 spin_lock(&sysctl_lock);
1391 if (prev) {
1392 head = prev;
1393 tmp = &prev->ctl_entry;
1394 unuse_table(prev);
1395 goto next;
1396 }
1397 tmp = &root_table_header.ctl_entry;
1398 for (;;) {
1399 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1400
1401 if (!use_table(head))
1402 goto next;
1403 spin_unlock(&sysctl_lock);
1404 return head;
1405 next:
1406 root = head->root;
1407 tmp = tmp->next;
1408 header_list = lookup_header_list(root, namespaces);
1409 if (tmp != header_list)
1410 continue;
1411
1412 do {
1413 root = list_entry(root->root_list.next,
1414 struct ctl_table_root, root_list);
1415 if (root == &sysctl_table_root)
1416 goto out;
1417 header_list = lookup_header_list(root, namespaces);
1418 } while (list_empty(header_list));
1419 tmp = header_list->next;
1420 }
1421out:
1422 spin_unlock(&sysctl_lock);
1423 return NULL;
1424}
1425
1426struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1427{
1428 return __sysctl_head_next(current->nsproxy, prev);
1429}
1430
1431void register_sysctl_root(struct ctl_table_root *root)
1432{
1433 spin_lock(&sysctl_lock);
1434 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1435 spin_unlock(&sysctl_lock);
1436}
1437
1438#ifdef CONFIG_SYSCTL_SYSCALL
1439/* Perform the actual read/write of a sysctl table entry. */
1440static int do_sysctl_strategy(struct ctl_table_root *root,
1441 struct ctl_table *table,
1442 int __user *name, int nlen,
1443 void __user *oldval, size_t __user *oldlenp,
1444 void __user *newval, size_t newlen)
1445{
1446 int op = 0, rc;
1447
1448 if (oldval)
1449 op |= 004;
1450 if (newval)
1451 op |= 002;
1452 if (sysctl_perm(root, table, op))
1453 return -EPERM;
1454
1455 if (table->strategy) {
1456 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1457 newval, newlen);
1458 if (rc < 0)
1459 return rc;
1460 if (rc > 0)
1461 return 0;
1462 }
1463
1464 /* If there is no strategy routine, or if the strategy returns
1465 * zero, proceed with automatic r/w */
1466 if (table->data && table->maxlen) {
1467 rc = sysctl_data(table, name, nlen, oldval, oldlenp,
1468 newval, newlen);
1469 if (rc < 0)
1470 return rc;
1471 }
1472 return 0;
1473}
1474
1475static int parse_table(int __user *name, int nlen,
1476 void __user *oldval, size_t __user *oldlenp,
1477 void __user *newval, size_t newlen,
1478 struct ctl_table_root *root,
1479 struct ctl_table *table)
1480{
1481 int n;
1482repeat:
1483 if (!nlen)
1484 return -ENOTDIR;
1485 if (get_user(n, name))
1486 return -EFAULT;
1487 for ( ; table->ctl_name || table->procname; table++) {
1488 if (!table->ctl_name)
1489 continue;
1490 if (n == table->ctl_name) {
1491 int error;
1492 if (table->child) {
1493 if (sysctl_perm(root, table, 001))
1494 return -EPERM;
1495 name++;
1496 nlen--;
1497 table = table->child;
1498 goto repeat;
1499 }
1500 error = do_sysctl_strategy(root, table, name, nlen,
1501 oldval, oldlenp,
1502 newval, newlen);
1503 return error;
1504 }
1505 }
1506 return -ENOTDIR;
1507}
1508
1509int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1510 void __user *newval, size_t newlen)
1511{
1512 struct ctl_table_header *head;
1513 int error = -ENOTDIR;
1514
1515 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1516 return -ENOTDIR;
1517 if (oldval) {
1518 int old_len;
1519 if (!oldlenp || get_user(old_len, oldlenp))
1520 return -EFAULT;
1521 }
1522
1523 for (head = sysctl_head_next(NULL); head;
1524 head = sysctl_head_next(head)) {
1525 error = parse_table(name, nlen, oldval, oldlenp,
1526 newval, newlen,
1527 head->root, head->ctl_table);
1528 if (error != -ENOTDIR) {
1529 sysctl_head_finish(head);
1530 break;
1531 }
1532 }
1533 return error;
1534}
1535
1536asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1537{
1538 struct __sysctl_args tmp;
1539 int error;
1540
1541 if (copy_from_user(&tmp, args, sizeof(tmp)))
1542 return -EFAULT;
1543
1544 error = deprecated_sysctl_warning(&tmp);
1545 if (error)
1546 goto out;
1547
1548 lock_kernel();
1549 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1550 tmp.newval, tmp.newlen);
1551 unlock_kernel();
1552out:
1553 return error;
1554}
1555#endif /* CONFIG_SYSCTL_SYSCALL */
1556
1557/*
1558 * sysctl_perm does NOT grant the superuser all rights automatically, because
1559 * some sysctl variables are readonly even to root.
1560 */
1561
1562static int test_perm(int mode, int op)
1563{
1564 if (!current->euid)
1565 mode >>= 6;
1566 else if (in_egroup_p(0))
1567 mode >>= 3;
1568 if ((mode & op & 0007) == op)
1569 return 0;
1570 return -EACCES;
1571}
1572
1573int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1574{
1575 int error;
1576 int mode;
1577
1578 error = security_sysctl(table, op);
1579 if (error)
1580 return error;
1581
1582 if (root->permissions)
1583 mode = root->permissions(root, current->nsproxy, table);
1584 else
1585 mode = table->mode;
1586
1587 return test_perm(mode, op);
1588}
1589
1590static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1591{
1592 for (; table->ctl_name || table->procname; table++) {
1593 table->parent = parent;
1594 if (table->child)
1595 sysctl_set_parent(table, table->child);
1596 }
1597}
1598
1599static __init int sysctl_init(void)
1600{
1601 sysctl_set_parent(NULL, root_table);
1602#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1603 {
1604 int err;
1605 err = sysctl_check_table(current->nsproxy, root_table);
1606 }
1607#endif
1608 return 0;
1609}
1610
1611core_initcall(sysctl_init);
1612
1613/**
1614 * __register_sysctl_paths - register a sysctl hierarchy
1615 * @root: List of sysctl headers to register on
1616 * @namespaces: Data to compute which lists of sysctl entries are visible
1617 * @path: The path to the directory the sysctl table is in.
1618 * @table: the top-level table structure
1619 *
1620 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1621 * array. A completely 0 filled entry terminates the table.
1622 *
1623 * The members of the &struct ctl_table structure are used as follows:
1624 *
1625 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1626 * must be unique within that level of sysctl
1627 *
1628 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1629 * enter a sysctl file
1630 *
1631 * data - a pointer to data for use by proc_handler
1632 *
1633 * maxlen - the maximum size in bytes of the data
1634 *
1635 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1636 *
1637 * child - a pointer to the child sysctl table if this entry is a directory, or
1638 * %NULL.
1639 *
1640 * proc_handler - the text handler routine (described below)
1641 *
1642 * strategy - the strategy routine (described below)
1643 *
1644 * de - for internal use by the sysctl routines
1645 *
1646 * extra1, extra2 - extra pointers usable by the proc handler routines
1647 *
1648 * Leaf nodes in the sysctl tree will be represented by a single file
1649 * under /proc; non-leaf nodes will be represented by directories.
1650 *
1651 * sysctl(2) can automatically manage read and write requests through
1652 * the sysctl table. The data and maxlen fields of the ctl_table
1653 * struct enable minimal validation of the values being written to be
1654 * performed, and the mode field allows minimal authentication.
1655 *
1656 * More sophisticated management can be enabled by the provision of a
1657 * strategy routine with the table entry. This will be called before
1658 * any automatic read or write of the data is performed.
1659 *
1660 * The strategy routine may return
1661 *
1662 * < 0 - Error occurred (error is passed to user process)
1663 *
1664 * 0 - OK - proceed with automatic read or write.
1665 *
1666 * > 0 - OK - read or write has been done by the strategy routine, so
1667 * return immediately.
1668 *
1669 * There must be a proc_handler routine for any terminal nodes
1670 * mirrored under /proc/sys (non-terminals are handled by a built-in
1671 * directory handler). Several default handlers are available to
1672 * cover common cases -
1673 *
1674 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1675 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1676 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1677 *
1678 * It is the handler's job to read the input buffer from user memory
1679 * and process it. The handler should return 0 on success.
1680 *
1681 * This routine returns %NULL on a failure to register, and a pointer
1682 * to the table header on success.
1683 */
1684struct ctl_table_header *__register_sysctl_paths(
1685 struct ctl_table_root *root,
1686 struct nsproxy *namespaces,
1687 const struct ctl_path *path, struct ctl_table *table)
1688{
1689 struct list_head *header_list;
1690 struct ctl_table_header *header;
1691 struct ctl_table *new, **prevp;
1692 unsigned int n, npath;
1693
1694 /* Count the path components */
1695 for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1696 ;
1697
1698 /*
1699 * For each path component, allocate a 2-element ctl_table array.
1700 * The first array element will be filled with the sysctl entry
1701 * for this, the second will be the sentinel (ctl_name == 0).
1702 *
1703 * We allocate everything in one go so that we don't have to
1704 * worry about freeing additional memory in unregister_sysctl_table.
1705 */
1706 header = kzalloc(sizeof(struct ctl_table_header) +
1707 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1708 if (!header)
1709 return NULL;
1710
1711 new = (struct ctl_table *) (header + 1);
1712
1713 /* Now connect the dots */
1714 prevp = &header->ctl_table;
1715 for (n = 0; n < npath; ++n, ++path) {
1716 /* Copy the procname */
1717 new->procname = path->procname;
1718 new->ctl_name = path->ctl_name;
1719 new->mode = 0555;
1720
1721 *prevp = new;
1722 prevp = &new->child;
1723
1724 new += 2;
1725 }
1726 *prevp = table;
1727 header->ctl_table_arg = table;
1728
1729 INIT_LIST_HEAD(&header->ctl_entry);
1730 header->used = 0;
1731 header->unregistering = NULL;
1732 header->root = root;
1733 sysctl_set_parent(NULL, header->ctl_table);
1734#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1735 if (sysctl_check_table(namespaces, header->ctl_table)) {
1736 kfree(header);
1737 return NULL;
1738 }
1739#endif
1740 spin_lock(&sysctl_lock);
1741 header_list = lookup_header_list(root, namespaces);
1742 list_add_tail(&header->ctl_entry, header_list);
1743 spin_unlock(&sysctl_lock);
1744
1745 return header;
1746}
1747
1748/**
1749 * register_sysctl_table_path - register a sysctl table hierarchy
1750 * @path: The path to the directory the sysctl table is in.
1751 * @table: the top-level table structure
1752 *
1753 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1754 * array. A completely 0 filled entry terminates the table.
1755 *
1756 * See __register_sysctl_paths for more details.
1757 */
1758struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1759 struct ctl_table *table)
1760{
1761 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1762 path, table);
1763}
1764
1765/**
1766 * register_sysctl_table - register a sysctl table hierarchy
1767 * @table: the top-level table structure
1768 *
1769 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1770 * array. A completely 0 filled entry terminates the table.
1771 *
1772 * See register_sysctl_paths for more details.
1773 */
1774struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1775{
1776 static const struct ctl_path null_path[] = { {} };
1777
1778 return register_sysctl_paths(null_path, table);
1779}
1780
1781/**
1782 * unregister_sysctl_table - unregister a sysctl table hierarchy
1783 * @header: the header returned from register_sysctl_table
1784 *
1785 * Unregisters the sysctl table and all children. proc entries may not
1786 * actually be removed until they are no longer used by anyone.
1787 */
1788void unregister_sysctl_table(struct ctl_table_header * header)
1789{
1790 might_sleep();
1791
1792 if (header == NULL)
1793 return;
1794
1795 spin_lock(&sysctl_lock);
1796 start_unregistering(header);
1797 spin_unlock(&sysctl_lock);
1798 kfree(header);
1799}
1800
1801#else /* !CONFIG_SYSCTL */
1802struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1803{
1804 return NULL;
1805}
1806
1807struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1808 struct ctl_table *table)
1809{
1810 return NULL;
1811}
1812
1813void unregister_sysctl_table(struct ctl_table_header * table)
1814{
1815}
1816
1817#endif /* CONFIG_SYSCTL */
1818
1819/*
1820 * /proc/sys support
1821 */
1822
1823#ifdef CONFIG_PROC_SYSCTL
1824
1825static int _proc_do_string(void* data, int maxlen, int write,
1826 struct file *filp, void __user *buffer,
1827 size_t *lenp, loff_t *ppos)
1828{
1829 size_t len;
1830 char __user *p;
1831 char c;
1832
1833 if (!data || !maxlen || !*lenp) {
1834 *lenp = 0;
1835 return 0;
1836 }
1837
1838 if (write) {
1839 len = 0;
1840 p = buffer;
1841 while (len < *lenp) {
1842 if (get_user(c, p++))
1843 return -EFAULT;
1844 if (c == 0 || c == '\n')
1845 break;
1846 len++;
1847 }
1848 if (len >= maxlen)
1849 len = maxlen-1;
1850 if(copy_from_user(data, buffer, len))
1851 return -EFAULT;
1852 ((char *) data)[len] = 0;
1853 *ppos += *lenp;
1854 } else {
1855 len = strlen(data);
1856 if (len > maxlen)
1857 len = maxlen;
1858
1859 if (*ppos > len) {
1860 *lenp = 0;
1861 return 0;
1862 }
1863
1864 data += *ppos;
1865 len -= *ppos;
1866
1867 if (len > *lenp)
1868 len = *lenp;
1869 if (len)
1870 if(copy_to_user(buffer, data, len))
1871 return -EFAULT;
1872 if (len < *lenp) {
1873 if(put_user('\n', ((char __user *) buffer) + len))
1874 return -EFAULT;
1875 len++;
1876 }
1877 *lenp = len;
1878 *ppos += len;
1879 }
1880 return 0;
1881}
1882
1883/**
1884 * proc_dostring - read a string sysctl
1885 * @table: the sysctl table
1886 * @write: %TRUE if this is a write to the sysctl file
1887 * @filp: the file structure
1888 * @buffer: the user buffer
1889 * @lenp: the size of the user buffer
1890 * @ppos: file position
1891 *
1892 * Reads/writes a string from/to the user buffer. If the kernel
1893 * buffer provided is not large enough to hold the string, the
1894 * string is truncated. The copied string is %NULL-terminated.
1895 * If the string is being read by the user process, it is copied
1896 * and a newline '\n' is added. It is truncated if the buffer is
1897 * not large enough.
1898 *
1899 * Returns 0 on success.
1900 */
1901int proc_dostring(struct ctl_table *table, int write, struct file *filp,
1902 void __user *buffer, size_t *lenp, loff_t *ppos)
1903{
1904 return _proc_do_string(table->data, table->maxlen, write, filp,
1905 buffer, lenp, ppos);
1906}
1907
1908
1909static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1910 int *valp,
1911 int write, void *data)
1912{
1913 if (write) {
1914 *valp = *negp ? -*lvalp : *lvalp;
1915 } else {
1916 int val = *valp;
1917 if (val < 0) {
1918 *negp = -1;
1919 *lvalp = (unsigned long)-val;
1920 } else {
1921 *negp = 0;
1922 *lvalp = (unsigned long)val;
1923 }
1924 }
1925 return 0;
1926}
1927
1928static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1929 int write, struct file *filp, void __user *buffer,
1930 size_t *lenp, loff_t *ppos,
1931 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1932 int write, void *data),
1933 void *data)
1934{
1935#define TMPBUFLEN 21
1936 int *i, vleft, first=1, neg, val;
1937 unsigned long lval;
1938 size_t left, len;
1939
1940 char buf[TMPBUFLEN], *p;
1941 char __user *s = buffer;
1942
1943 if (!tbl_data || !table->maxlen || !*lenp ||
1944 (*ppos && !write)) {
1945 *lenp = 0;
1946 return 0;
1947 }
1948
1949 i = (int *) tbl_data;
1950 vleft = table->maxlen / sizeof(*i);
1951 left = *lenp;
1952
1953 if (!conv)
1954 conv = do_proc_dointvec_conv;
1955
1956 for (; left && vleft--; i++, first=0) {
1957 if (write) {
1958 while (left) {
1959 char c;
1960 if (get_user(c, s))
1961 return -EFAULT;
1962 if (!isspace(c))
1963 break;
1964 left--;
1965 s++;
1966 }
1967 if (!left)
1968 break;
1969 neg = 0;
1970 len = left;
1971 if (len > sizeof(buf) - 1)
1972 len = sizeof(buf) - 1;
1973 if (copy_from_user(buf, s, len))
1974 return -EFAULT;
1975 buf[len] = 0;
1976 p = buf;
1977 if (*p == '-' && left > 1) {
1978 neg = 1;
1979 p++;
1980 }
1981 if (*p < '0' || *p > '9')
1982 break;
1983
1984 lval = simple_strtoul(p, &p, 0);
1985
1986 len = p-buf;
1987 if ((len < left) && *p && !isspace(*p))
1988 break;
1989 if (neg)
1990 val = -val;
1991 s += len;
1992 left -= len;
1993
1994 if (conv(&neg, &lval, i, 1, data))
1995 break;
1996 } else {
1997 p = buf;
1998 if (!first)
1999 *p++ = '\t';
2000
2001 if (conv(&neg, &lval, i, 0, data))
2002 break;
2003
2004 sprintf(p, "%s%lu", neg ? "-" : "", lval);
2005 len = strlen(buf);
2006 if (len > left)
2007 len = left;
2008 if(copy_to_user(s, buf, len))
2009 return -EFAULT;
2010 left -= len;
2011 s += len;
2012 }
2013 }
2014
2015 if (!write && !first && left) {
2016 if(put_user('\n', s))
2017 return -EFAULT;
2018 left--, s++;
2019 }
2020 if (write) {
2021 while (left) {
2022 char c;
2023 if (get_user(c, s++))
2024 return -EFAULT;
2025 if (!isspace(c))
2026 break;
2027 left--;
2028 }
2029 }
2030 if (write && first)
2031 return -EINVAL;
2032 *lenp -= left;
2033 *ppos += *lenp;
2034 return 0;
2035#undef TMPBUFLEN
2036}
2037
2038static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2039 void __user *buffer, size_t *lenp, loff_t *ppos,
2040 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2041 int write, void *data),
2042 void *data)
2043{
2044 return __do_proc_dointvec(table->data, table, write, filp,
2045 buffer, lenp, ppos, conv, data);
2046}
2047
2048/**
2049 * proc_dointvec - read a vector of integers
2050 * @table: the sysctl table
2051 * @write: %TRUE if this is a write to the sysctl file
2052 * @filp: the file structure
2053 * @buffer: the user buffer
2054 * @lenp: the size of the user buffer
2055 * @ppos: file position
2056 *
2057 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2058 * values from/to the user buffer, treated as an ASCII string.
2059 *
2060 * Returns 0 on success.
2061 */
2062int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2063 void __user *buffer, size_t *lenp, loff_t *ppos)
2064{
2065 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2066 NULL,NULL);
2067}
2068
2069#define OP_SET 0
2070#define OP_AND 1
2071#define OP_OR 2
2072
2073static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
2074 int *valp,
2075 int write, void *data)
2076{
2077 int op = *(int *)data;
2078 if (write) {
2079 int val = *negp ? -*lvalp : *lvalp;
2080 switch(op) {
2081 case OP_SET: *valp = val; break;
2082 case OP_AND: *valp &= val; break;
2083 case OP_OR: *valp |= val; break;
2084 }
2085 } else {
2086 int val = *valp;
2087 if (val < 0) {
2088 *negp = -1;
2089 *lvalp = (unsigned long)-val;
2090 } else {
2091 *negp = 0;
2092 *lvalp = (unsigned long)val;
2093 }
2094 }
2095 return 0;
2096}
2097
2098/*
2099 * Taint values can only be increased
2100 */
2101static int proc_dointvec_taint(struct ctl_table *table, int write, struct file *filp,
2102 void __user *buffer, size_t *lenp, loff_t *ppos)
2103{
2104 int op;
2105
2106 if (write && !capable(CAP_SYS_ADMIN))
2107 return -EPERM;
2108
2109 op = OP_OR;
2110 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2111 do_proc_dointvec_bset_conv,&op);
2112}
2113
2114struct do_proc_dointvec_minmax_conv_param {
2115 int *min;
2116 int *max;
2117};
2118
2119static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2120 int *valp,
2121 int write, void *data)
2122{
2123 struct do_proc_dointvec_minmax_conv_param *param = data;
2124 if (write) {
2125 int val = *negp ? -*lvalp : *lvalp;
2126 if ((param->min && *param->min > val) ||
2127 (param->max && *param->max < val))
2128 return -EINVAL;
2129 *valp = val;
2130 } else {
2131 int val = *valp;
2132 if (val < 0) {
2133 *negp = -1;
2134 *lvalp = (unsigned long)-val;
2135 } else {
2136 *negp = 0;
2137 *lvalp = (unsigned long)val;
2138 }
2139 }
2140 return 0;
2141}
2142
2143/**
2144 * proc_dointvec_minmax - read a vector of integers with min/max values
2145 * @table: the sysctl table
2146 * @write: %TRUE if this is a write to the sysctl file
2147 * @filp: the file structure
2148 * @buffer: the user buffer
2149 * @lenp: the size of the user buffer
2150 * @ppos: file position
2151 *
2152 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2153 * values from/to the user buffer, treated as an ASCII string.
2154 *
2155 * This routine will ensure the values are within the range specified by
2156 * table->extra1 (min) and table->extra2 (max).
2157 *
2158 * Returns 0 on success.
2159 */
2160int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2161 void __user *buffer, size_t *lenp, loff_t *ppos)
2162{
2163 struct do_proc_dointvec_minmax_conv_param param = {
2164 .min = (int *) table->extra1,
2165 .max = (int *) table->extra2,
2166 };
2167 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2168 do_proc_dointvec_minmax_conv, &param);
2169}
2170
2171static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2172 struct file *filp,
2173 void __user *buffer,
2174 size_t *lenp, loff_t *ppos,
2175 unsigned long convmul,
2176 unsigned long convdiv)
2177{
2178#define TMPBUFLEN 21
2179 unsigned long *i, *min, *max, val;
2180 int vleft, first=1, neg;
2181 size_t len, left;
2182 char buf[TMPBUFLEN], *p;
2183 char __user *s = buffer;
2184
2185 if (!data || !table->maxlen || !*lenp ||
2186 (*ppos && !write)) {
2187 *lenp = 0;
2188 return 0;
2189 }
2190
2191 i = (unsigned long *) data;
2192 min = (unsigned long *) table->extra1;
2193 max = (unsigned long *) table->extra2;
2194 vleft = table->maxlen / sizeof(unsigned long);
2195 left = *lenp;
2196
2197 for (; left && vleft--; i++, min++, max++, first=0) {
2198 if (write) {
2199 while (left) {
2200 char c;
2201 if (get_user(c, s))
2202 return -EFAULT;
2203 if (!isspace(c))
2204 break;
2205 left--;
2206 s++;
2207 }
2208 if (!left)
2209 break;
2210 neg = 0;
2211 len = left;
2212 if (len > TMPBUFLEN-1)
2213 len = TMPBUFLEN-1;
2214 if (copy_from_user(buf, s, len))
2215 return -EFAULT;
2216 buf[len] = 0;
2217 p = buf;
2218 if (*p == '-' && left > 1) {
2219 neg = 1;
2220 p++;
2221 }
2222 if (*p < '0' || *p > '9')
2223 break;
2224 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2225 len = p-buf;
2226 if ((len < left) && *p && !isspace(*p))
2227 break;
2228 if (neg)
2229 val = -val;
2230 s += len;
2231 left -= len;
2232
2233 if(neg)
2234 continue;
2235 if ((min && val < *min) || (max && val > *max))
2236 continue;
2237 *i = val;
2238 } else {
2239 p = buf;
2240 if (!first)
2241 *p++ = '\t';
2242 sprintf(p, "%lu", convdiv * (*i) / convmul);
2243 len = strlen(buf);
2244 if (len > left)
2245 len = left;
2246 if(copy_to_user(s, buf, len))
2247 return -EFAULT;
2248 left -= len;
2249 s += len;
2250 }
2251 }
2252
2253 if (!write && !first && left) {
2254 if(put_user('\n', s))
2255 return -EFAULT;
2256 left--, s++;
2257 }
2258 if (write) {
2259 while (left) {
2260 char c;
2261 if (get_user(c, s++))
2262 return -EFAULT;
2263 if (!isspace(c))
2264 break;
2265 left--;
2266 }
2267 }
2268 if (write && first)
2269 return -EINVAL;
2270 *lenp -= left;
2271 *ppos += *lenp;
2272 return 0;
2273#undef TMPBUFLEN
2274}
2275
2276static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2277 struct file *filp,
2278 void __user *buffer,
2279 size_t *lenp, loff_t *ppos,
2280 unsigned long convmul,
2281 unsigned long convdiv)
2282{
2283 return __do_proc_doulongvec_minmax(table->data, table, write,
2284 filp, buffer, lenp, ppos, convmul, convdiv);
2285}
2286
2287/**
2288 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2289 * @table: the sysctl table
2290 * @write: %TRUE if this is a write to the sysctl file
2291 * @filp: the file structure
2292 * @buffer: the user buffer
2293 * @lenp: the size of the user buffer
2294 * @ppos: file position
2295 *
2296 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2297 * values from/to the user buffer, treated as an ASCII string.
2298 *
2299 * This routine will ensure the values are within the range specified by
2300 * table->extra1 (min) and table->extra2 (max).
2301 *
2302 * Returns 0 on success.
2303 */
2304int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2305 void __user *buffer, size_t *lenp, loff_t *ppos)
2306{
2307 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2308}
2309
2310/**
2311 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2312 * @table: the sysctl table
2313 * @write: %TRUE if this is a write to the sysctl file
2314 * @filp: the file structure
2315 * @buffer: the user buffer
2316 * @lenp: the size of the user buffer
2317 * @ppos: file position
2318 *
2319 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2320 * values from/to the user buffer, treated as an ASCII string. The values
2321 * are treated as milliseconds, and converted to jiffies when they are stored.
2322 *
2323 * This routine will ensure the values are within the range specified by
2324 * table->extra1 (min) and table->extra2 (max).
2325 *
2326 * Returns 0 on success.
2327 */
2328int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2329 struct file *filp,
2330 void __user *buffer,
2331 size_t *lenp, loff_t *ppos)
2332{
2333 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2334 lenp, ppos, HZ, 1000l);
2335}
2336
2337
2338static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2339 int *valp,
2340 int write, void *data)
2341{
2342 if (write) {
2343 if (*lvalp > LONG_MAX / HZ)
2344 return 1;
2345 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2346 } else {
2347 int val = *valp;
2348 unsigned long lval;
2349 if (val < 0) {
2350 *negp = -1;
2351 lval = (unsigned long)-val;
2352 } else {
2353 *negp = 0;
2354 lval = (unsigned long)val;
2355 }
2356 *lvalp = lval / HZ;
2357 }
2358 return 0;
2359}
2360
2361static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2362 int *valp,
2363 int write, void *data)
2364{
2365 if (write) {
2366 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2367 return 1;
2368 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2369 } else {
2370 int val = *valp;
2371 unsigned long lval;
2372 if (val < 0) {
2373 *negp = -1;
2374 lval = (unsigned long)-val;
2375 } else {
2376 *negp = 0;
2377 lval = (unsigned long)val;
2378 }
2379 *lvalp = jiffies_to_clock_t(lval);
2380 }
2381 return 0;
2382}
2383
2384static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2385 int *valp,
2386 int write, void *data)
2387{
2388 if (write) {
2389 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2390 } else {
2391 int val = *valp;
2392 unsigned long lval;
2393 if (val < 0) {
2394 *negp = -1;
2395 lval = (unsigned long)-val;
2396 } else {
2397 *negp = 0;
2398 lval = (unsigned long)val;
2399 }
2400 *lvalp = jiffies_to_msecs(lval);
2401 }
2402 return 0;
2403}
2404
2405/**
2406 * proc_dointvec_jiffies - read a vector of integers as seconds
2407 * @table: the sysctl table
2408 * @write: %TRUE if this is a write to the sysctl file
2409 * @filp: the file structure
2410 * @buffer: the user buffer
2411 * @lenp: the size of the user buffer
2412 * @ppos: file position
2413 *
2414 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2415 * values from/to the user buffer, treated as an ASCII string.
2416 * The values read are assumed to be in seconds, and are converted into
2417 * jiffies.
2418 *
2419 * Returns 0 on success.
2420 */
2421int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2422 void __user *buffer, size_t *lenp, loff_t *ppos)
2423{
2424 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2425 do_proc_dointvec_jiffies_conv,NULL);
2426}
2427
2428/**
2429 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2430 * @table: the sysctl table
2431 * @write: %TRUE if this is a write to the sysctl file
2432 * @filp: the file structure
2433 * @buffer: the user buffer
2434 * @lenp: the size of the user buffer
2435 * @ppos: pointer to the file position
2436 *
2437 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2438 * values from/to the user buffer, treated as an ASCII string.
2439 * The values read are assumed to be in 1/USER_HZ seconds, and
2440 * are converted into jiffies.
2441 *
2442 * Returns 0 on success.
2443 */
2444int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2445 void __user *buffer, size_t *lenp, loff_t *ppos)
2446{
2447 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2448 do_proc_dointvec_userhz_jiffies_conv,NULL);
2449}
2450
2451/**
2452 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2453 * @table: the sysctl table
2454 * @write: %TRUE if this is a write to the sysctl file
2455 * @filp: the file structure
2456 * @buffer: the user buffer
2457 * @lenp: the size of the user buffer
2458 * @ppos: file position
2459 * @ppos: the current position in the file
2460 *
2461 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2462 * values from/to the user buffer, treated as an ASCII string.
2463 * The values read are assumed to be in 1/1000 seconds, and
2464 * are converted into jiffies.
2465 *
2466 * Returns 0 on success.
2467 */
2468int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2469 void __user *buffer, size_t *lenp, loff_t *ppos)
2470{
2471 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2472 do_proc_dointvec_ms_jiffies_conv, NULL);
2473}
2474
2475static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2476 void __user *buffer, size_t *lenp, loff_t *ppos)
2477{
2478 struct pid *new_pid;
2479 pid_t tmp;
2480 int r;
2481
2482 tmp = pid_vnr(cad_pid);
2483
2484 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2485 lenp, ppos, NULL, NULL);
2486 if (r || !write)
2487 return r;
2488
2489 new_pid = find_get_pid(tmp);
2490 if (!new_pid)
2491 return -ESRCH;
2492
2493 put_pid(xchg(&cad_pid, new_pid));
2494 return 0;
2495}
2496
2497#else /* CONFIG_PROC_FS */
2498
2499int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2500 void __user *buffer, size_t *lenp, loff_t *ppos)
2501{
2502 return -ENOSYS;
2503}
2504
2505int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2506 void __user *buffer, size_t *lenp, loff_t *ppos)
2507{
2508 return -ENOSYS;
2509}
2510
2511int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2512 void __user *buffer, size_t *lenp, loff_t *ppos)
2513{
2514 return -ENOSYS;
2515}
2516
2517int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2518 void __user *buffer, size_t *lenp, loff_t *ppos)
2519{
2520 return -ENOSYS;
2521}
2522
2523int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2524 void __user *buffer, size_t *lenp, loff_t *ppos)
2525{
2526 return -ENOSYS;
2527}
2528
2529int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2530 void __user *buffer, size_t *lenp, loff_t *ppos)
2531{
2532 return -ENOSYS;
2533}
2534
2535int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2536 void __user *buffer, size_t *lenp, loff_t *ppos)
2537{
2538 return -ENOSYS;
2539}
2540
2541int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2542 struct file *filp,
2543 void __user *buffer,
2544 size_t *lenp, loff_t *ppos)
2545{
2546 return -ENOSYS;
2547}
2548
2549
2550#endif /* CONFIG_PROC_FS */
2551
2552
2553#ifdef CONFIG_SYSCTL_SYSCALL
2554/*
2555 * General sysctl support routines
2556 */
2557
2558/* The generic sysctl data routine (used if no strategy routine supplied) */
2559int sysctl_data(struct ctl_table *table, int __user *name, int nlen,
2560 void __user *oldval, size_t __user *oldlenp,
2561 void __user *newval, size_t newlen)
2562{
2563 size_t len;
2564
2565 /* Get out of I don't have a variable */
2566 if (!table->data || !table->maxlen)
2567 return -ENOTDIR;
2568
2569 if (oldval && oldlenp) {
2570 if (get_user(len, oldlenp))
2571 return -EFAULT;
2572 if (len) {
2573 if (len > table->maxlen)
2574 len = table->maxlen;
2575 if (copy_to_user(oldval, table->data, len))
2576 return -EFAULT;
2577 if (put_user(len, oldlenp))
2578 return -EFAULT;
2579 }
2580 }
2581
2582 if (newval && newlen) {
2583 if (newlen > table->maxlen)
2584 newlen = table->maxlen;
2585
2586 if (copy_from_user(table->data, newval, newlen))
2587 return -EFAULT;
2588 }
2589 return 1;
2590}
2591
2592/* The generic string strategy routine: */
2593int sysctl_string(struct ctl_table *table, int __user *name, int nlen,
2594 void __user *oldval, size_t __user *oldlenp,
2595 void __user *newval, size_t newlen)
2596{
2597 if (!table->data || !table->maxlen)
2598 return -ENOTDIR;
2599
2600 if (oldval && oldlenp) {
2601 size_t bufsize;
2602 if (get_user(bufsize, oldlenp))
2603 return -EFAULT;
2604 if (bufsize) {
2605 size_t len = strlen(table->data), copied;
2606
2607 /* This shouldn't trigger for a well-formed sysctl */
2608 if (len > table->maxlen)
2609 len = table->maxlen;
2610
2611 /* Copy up to a max of bufsize-1 bytes of the string */
2612 copied = (len >= bufsize) ? bufsize - 1 : len;
2613
2614 if (copy_to_user(oldval, table->data, copied) ||
2615 put_user(0, (char __user *)(oldval + copied)))
2616 return -EFAULT;
2617 if (put_user(len, oldlenp))
2618 return -EFAULT;
2619 }
2620 }
2621 if (newval && newlen) {
2622 size_t len = newlen;
2623 if (len > table->maxlen)
2624 len = table->maxlen;
2625 if(copy_from_user(table->data, newval, len))
2626 return -EFAULT;
2627 if (len == table->maxlen)
2628 len--;
2629 ((char *) table->data)[len] = 0;
2630 }
2631 return 1;
2632}
2633
2634/*
2635 * This function makes sure that all of the integers in the vector
2636 * are between the minimum and maximum values given in the arrays
2637 * table->extra1 and table->extra2, respectively.
2638 */
2639int sysctl_intvec(struct ctl_table *table, int __user *name, int nlen,
2640 void __user *oldval, size_t __user *oldlenp,
2641 void __user *newval, size_t newlen)
2642{
2643
2644 if (newval && newlen) {
2645 int __user *vec = (int __user *) newval;
2646 int *min = (int *) table->extra1;
2647 int *max = (int *) table->extra2;
2648 size_t length;
2649 int i;
2650
2651 if (newlen % sizeof(int) != 0)
2652 return -EINVAL;
2653
2654 if (!table->extra1 && !table->extra2)
2655 return 0;
2656
2657 if (newlen > table->maxlen)
2658 newlen = table->maxlen;
2659 length = newlen / sizeof(int);
2660
2661 for (i = 0; i < length; i++) {
2662 int value;
2663 if (get_user(value, vec + i))
2664 return -EFAULT;
2665 if (min && value < min[i])
2666 return -EINVAL;
2667 if (max && value > max[i])
2668 return -EINVAL;
2669 }
2670 }
2671 return 0;
2672}
2673
2674/* Strategy function to convert jiffies to seconds */
2675int sysctl_jiffies(struct ctl_table *table, int __user *name, int nlen,
2676 void __user *oldval, size_t __user *oldlenp,
2677 void __user *newval, size_t newlen)
2678{
2679 if (oldval && oldlenp) {
2680 size_t olen;
2681
2682 if (get_user(olen, oldlenp))
2683 return -EFAULT;
2684 if (olen) {
2685 int val;
2686
2687 if (olen < sizeof(int))
2688 return -EINVAL;
2689
2690 val = *(int *)(table->data) / HZ;
2691 if (put_user(val, (int __user *)oldval))
2692 return -EFAULT;
2693 if (put_user(sizeof(int), oldlenp))
2694 return -EFAULT;
2695 }
2696 }
2697 if (newval && newlen) {
2698 int new;
2699 if (newlen != sizeof(int))
2700 return -EINVAL;
2701 if (get_user(new, (int __user *)newval))
2702 return -EFAULT;
2703 *(int *)(table->data) = new*HZ;
2704 }
2705 return 1;
2706}
2707
2708/* Strategy function to convert jiffies to seconds */
2709int sysctl_ms_jiffies(struct ctl_table *table, int __user *name, int nlen,
2710 void __user *oldval, size_t __user *oldlenp,
2711 void __user *newval, size_t newlen)
2712{
2713 if (oldval && oldlenp) {
2714 size_t olen;
2715
2716 if (get_user(olen, oldlenp))
2717 return -EFAULT;
2718 if (olen) {
2719 int val;
2720
2721 if (olen < sizeof(int))
2722 return -EINVAL;
2723
2724 val = jiffies_to_msecs(*(int *)(table->data));
2725 if (put_user(val, (int __user *)oldval))
2726 return -EFAULT;
2727 if (put_user(sizeof(int), oldlenp))
2728 return -EFAULT;
2729 }
2730 }
2731 if (newval && newlen) {
2732 int new;
2733 if (newlen != sizeof(int))
2734 return -EINVAL;
2735 if (get_user(new, (int __user *)newval))
2736 return -EFAULT;
2737 *(int *)(table->data) = msecs_to_jiffies(new);
2738 }
2739 return 1;
2740}
2741
2742
2743
2744#else /* CONFIG_SYSCTL_SYSCALL */
2745
2746
2747asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2748{
2749 struct __sysctl_args tmp;
2750 int error;
2751
2752 if (copy_from_user(&tmp, args, sizeof(tmp)))
2753 return -EFAULT;
2754
2755 error = deprecated_sysctl_warning(&tmp);
2756
2757 /* If no error reading the parameters then just -ENOSYS ... */
2758 if (!error)
2759 error = -ENOSYS;
2760
2761 return error;
2762}
2763
2764int sysctl_data(struct ctl_table *table, int __user *name, int nlen,
2765 void __user *oldval, size_t __user *oldlenp,
2766 void __user *newval, size_t newlen)
2767{
2768 return -ENOSYS;
2769}
2770
2771int sysctl_string(struct ctl_table *table, int __user *name, int nlen,
2772 void __user *oldval, size_t __user *oldlenp,
2773 void __user *newval, size_t newlen)
2774{
2775 return -ENOSYS;
2776}
2777
2778int sysctl_intvec(struct ctl_table *table, int __user *name, int nlen,
2779 void __user *oldval, size_t __user *oldlenp,
2780 void __user *newval, size_t newlen)
2781{
2782 return -ENOSYS;
2783}
2784
2785int sysctl_jiffies(struct ctl_table *table, int __user *name, int nlen,
2786 void __user *oldval, size_t __user *oldlenp,
2787 void __user *newval, size_t newlen)
2788{
2789 return -ENOSYS;
2790}
2791
2792int sysctl_ms_jiffies(struct ctl_table *table, int __user *name, int nlen,
2793 void __user *oldval, size_t __user *oldlenp,
2794 void __user *newval, size_t newlen)
2795{
2796 return -ENOSYS;
2797}
2798
2799#endif /* CONFIG_SYSCTL_SYSCALL */
2800
2801static int deprecated_sysctl_warning(struct __sysctl_args *args)
2802{
2803 static int msg_count;
2804 int name[CTL_MAXNAME];
2805 int i;
2806
2807 /* Check args->nlen. */
2808 if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
2809 return -ENOTDIR;
2810
2811 /* Read in the sysctl name for better debug message logging */
2812 for (i = 0; i < args->nlen; i++)
2813 if (get_user(name[i], args->name + i))
2814 return -EFAULT;
2815
2816 /* Ignore accesses to kernel.version */
2817 if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2818 return 0;
2819
2820 if (msg_count < 5) {
2821 msg_count++;
2822 printk(KERN_INFO
2823 "warning: process `%s' used the deprecated sysctl "
2824 "system call with ", current->comm);
2825 for (i = 0; i < args->nlen; i++)
2826 printk("%d.", name[i]);
2827 printk("\n");
2828 }
2829 return 0;
2830}
2831
2832/*
2833 * No sense putting this after each symbol definition, twice,
2834 * exception granted :-)
2835 */
2836EXPORT_SYMBOL(proc_dointvec);
2837EXPORT_SYMBOL(proc_dointvec_jiffies);
2838EXPORT_SYMBOL(proc_dointvec_minmax);
2839EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2840EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2841EXPORT_SYMBOL(proc_dostring);
2842EXPORT_SYMBOL(proc_doulongvec_minmax);
2843EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2844EXPORT_SYMBOL(register_sysctl_table);
2845EXPORT_SYMBOL(register_sysctl_paths);
2846EXPORT_SYMBOL(sysctl_intvec);
2847EXPORT_SYMBOL(sysctl_jiffies);
2848EXPORT_SYMBOL(sysctl_ms_jiffies);
2849EXPORT_SYMBOL(sysctl_string);
2850EXPORT_SYMBOL(sysctl_data);
2851EXPORT_SYMBOL(unregister_sysctl_table);