?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc
[linux-2.6/lfs.git] / kernel / sysctl.c
blobd7ffdc59816ad3f822389d508f3c5ff1b030578d
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
59 static int deprecated_sysctl_warning(struct __sysctl_args *args);
60
61 #if defined(CONFIG_SYSCTL)
62
63 /* External variables not in a header file. */
64 extern int C_A_D;
65 extern int print_fatal_signals;
66 extern int sysctl_overcommit_memory;
67 extern int sysctl_overcommit_ratio;
68 extern int sysctl_panic_on_oom;
69 extern int sysctl_oom_kill_allocating_task;
70 extern int sysctl_oom_dump_tasks;
71 extern int max_threads;
72 extern int core_uses_pid;
73 extern int suid_dumpable;
74 extern char core_pattern[];
75 extern int pid_max;
76 extern int min_free_kbytes;
77 extern int pid_max_min, pid_max_max;
78 extern int sysctl_drop_caches;
79 extern int percpu_pagelist_fraction;
80 extern int compat_log;
81 extern int maps_protect;
82 extern int sysctl_stat_interval;
83 extern int latencytop_enabled;
84
85 /* Constants used for minimum and maximum */
86 #if defined(CONFIG_DETECT_SOFTLOCKUP) || defined(CONFIG_HIGHMEM)
87 static int one = 1;
88 #endif
89
90 #ifdef CONFIG_DETECT_SOFTLOCKUP
91 static int sixty = 60;
92 #endif
93
94 #ifdef CONFIG_MMU
95 static int two = 2;
96 #endif
97
98 static int zero;
99 static int one_hundred = 100;
100
101 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
102 static int maxolduid = 65535;
103 static int minolduid;
104 static int min_percpu_pagelist_fract = 8;
105
106 static int ngroups_max = NGROUPS_MAX;
107
108 #ifdef CONFIG_KMOD
109 extern char modprobe_path[];
110 #endif
111 #ifdef CONFIG_CHR_DEV_SG
112 extern int sg_big_buff;
113 #endif
114
115 #ifdef __sparc__
116 extern char reboot_command [];
117 extern int stop_a_enabled;
118 extern int scons_pwroff;
119 #endif
120
121 #ifdef __hppa__
122 extern int pwrsw_enabled;
123 extern int unaligned_enabled;
124 #endif
125
126 #ifdef CONFIG_S390
127 #ifdef CONFIG_MATHEMU
128 extern int sysctl_ieee_emulation_warnings;
129 #endif
130 extern int sysctl_userprocess_debug;
131 extern int spin_retry;
132 #endif
133
134 extern int sysctl_hz_timer;
135
136 #ifdef CONFIG_BSD_PROCESS_ACCT
137 extern int acct_parm[];
138 #endif
139
140 #ifdef CONFIG_IA64
141 extern int no_unaligned_warning;
142 #endif
143
144 #ifdef CONFIG_RT_MUTEXES
145 extern int max_lock_depth;
146 #endif
147
148 #ifdef CONFIG_PROC_SYSCTL
149 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
150 void __user *buffer, size_t *lenp, loff_t *ppos);
151 static int proc_dointvec_taint(struct ctl_table *table, int write, struct file *filp,
152 void __user *buffer, size_t *lenp, loff_t *ppos);
153 #endif
154
155 static struct ctl_table root_table[];
156 static struct ctl_table_root sysctl_table_root;
157 static struct ctl_table_header root_table_header = {
158 .ctl_table = root_table,
159 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.header_list),
160 .root = &sysctl_table_root,
161 };
162 static struct ctl_table_root sysctl_table_root = {
163 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
164 .header_list = LIST_HEAD_INIT(root_table_header.ctl_entry),
165 };
166
167 static struct ctl_table kern_table[];
168 static struct ctl_table vm_table[];
169 static struct ctl_table fs_table[];
170 static struct ctl_table debug_table[];
171 static struct ctl_table dev_table[];
172 extern struct ctl_table random_table[];
173 #ifdef CONFIG_INOTIFY_USER
174 extern struct ctl_table inotify_table[];
175 #endif
176
177 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
178 int sysctl_legacy_va_layout;
179 #endif
180
181 extern int prove_locking;
182 extern int lock_stat;
183
184 /* The default sysctl tables: */
185
186 static struct ctl_table root_table[] = {
187 {
188 .ctl_name = CTL_KERN,
189 .procname = "kernel",
190 .mode = 0555,
191 .child = kern_table,
192 },
193 {
194 .ctl_name = CTL_VM,
195 .procname = "vm",
196 .mode = 0555,
197 .child = vm_table,
198 },
199 {
200 .ctl_name = CTL_FS,
201 .procname = "fs",
202 .mode = 0555,
203 .child = fs_table,
204 },
205 {
206 .ctl_name = CTL_DEBUG,
207 .procname = "debug",
208 .mode = 0555,
209 .child = debug_table,
210 },
211 {
212 .ctl_name = CTL_DEV,
213 .procname = "dev",
214 .mode = 0555,
215 .child = dev_table,
216 },
217 /*
218 * NOTE: do not add new entries to this table unless you have read
219 * Documentation/sysctl/ctl_unnumbered.txt
220 */
221 { .ctl_name = 0 }
222 };
223
224 #ifdef CONFIG_SCHED_DEBUG
225 static int min_sched_granularity_ns = 100000; /* 100 usecs */
226 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
227 static int min_wakeup_granularity_ns; /* 0 usecs */
228 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
229 #endif
230
231 static struct ctl_table kern_table[] = {
232 #ifdef CONFIG_SCHED_DEBUG
233 {
234 .ctl_name = CTL_UNNUMBERED,
235 .procname = "sched_min_granularity_ns",
236 .data = &sysctl_sched_min_granularity,
237 .maxlen = sizeof(unsigned int),
238 .mode = 0644,
239 .proc_handler = &sched_nr_latency_handler,
240 .strategy = &sysctl_intvec,
241 .extra1 = &min_sched_granularity_ns,
242 .extra2 = &max_sched_granularity_ns,
243 },
244 {
245 .ctl_name = CTL_UNNUMBERED,
246 .procname = "sched_latency_ns",
247 .data = &sysctl_sched_latency,
248 .maxlen = sizeof(unsigned int),
249 .mode = 0644,
250 .proc_handler = &sched_nr_latency_handler,
251 .strategy = &sysctl_intvec,
252 .extra1 = &min_sched_granularity_ns,
253 .extra2 = &max_sched_granularity_ns,
254 },
255 {
256 .ctl_name = CTL_UNNUMBERED,
257 .procname = "sched_wakeup_granularity_ns",
258 .data = &sysctl_sched_wakeup_granularity,
259 .maxlen = sizeof(unsigned int),
260 .mode = 0644,
261 .proc_handler = &proc_dointvec_minmax,
262 .strategy = &sysctl_intvec,
263 .extra1 = &min_wakeup_granularity_ns,
264 .extra2 = &max_wakeup_granularity_ns,
265 },
266 {
267 .ctl_name = CTL_UNNUMBERED,
268 .procname = "sched_child_runs_first",
269 .data = &sysctl_sched_child_runs_first,
270 .maxlen = sizeof(unsigned int),
271 .mode = 0644,
272 .proc_handler = &proc_dointvec,
273 },
274 {
275 .ctl_name = CTL_UNNUMBERED,
276 .procname = "sched_features",
277 .data = &sysctl_sched_features,
278 .maxlen = sizeof(unsigned int),
279 .mode = 0644,
280 .proc_handler = &proc_dointvec,
281 },
282 {
283 .ctl_name = CTL_UNNUMBERED,
284 .procname = "sched_migration_cost",
285 .data = &sysctl_sched_migration_cost,
286 .maxlen = sizeof(unsigned int),
287 .mode = 0644,
288 .proc_handler = &proc_dointvec,
289 },
290 {
291 .ctl_name = CTL_UNNUMBERED,
292 .procname = "sched_nr_migrate",
293 .data = &sysctl_sched_nr_migrate,
294 .maxlen = sizeof(unsigned int),
295 .mode = 0644,
296 .proc_handler = &proc_dointvec,
297 },
298 #endif
299 {
300 .ctl_name = CTL_UNNUMBERED,
301 .procname = "sched_rt_period_us",
302 .data = &sysctl_sched_rt_period,
303 .maxlen = sizeof(unsigned int),
304 .mode = 0644,
305 .proc_handler = &sched_rt_handler,
306 },
307 {
308 .ctl_name = CTL_UNNUMBERED,
309 .procname = "sched_rt_runtime_us",
310 .data = &sysctl_sched_rt_runtime,
311 .maxlen = sizeof(int),
312 .mode = 0644,
313 .proc_handler = &sched_rt_handler,
314 },
315 {
316 .ctl_name = CTL_UNNUMBERED,
317 .procname = "sched_compat_yield",
318 .data = &sysctl_sched_compat_yield,
319 .maxlen = sizeof(unsigned int),
320 .mode = 0644,
321 .proc_handler = &proc_dointvec,
322 },
323 #ifdef CONFIG_PROVE_LOCKING
324 {
325 .ctl_name = CTL_UNNUMBERED,
326 .procname = "prove_locking",
327 .data = &prove_locking,
328 .maxlen = sizeof(int),
329 .mode = 0644,
330 .proc_handler = &proc_dointvec,
331 },
332 #endif
333 #ifdef CONFIG_LOCK_STAT
334 {
335 .ctl_name = CTL_UNNUMBERED,
336 .procname = "lock_stat",
337 .data = &lock_stat,
338 .maxlen = sizeof(int),
339 .mode = 0644,
340 .proc_handler = &proc_dointvec,
341 },
342 #endif
343 {
344 .ctl_name = KERN_PANIC,
345 .procname = "panic",
346 .data = &panic_timeout,
347 .maxlen = sizeof(int),
348 .mode = 0644,
349 .proc_handler = &proc_dointvec,
350 },
351 {
352 .ctl_name = KERN_CORE_USES_PID,
353 .procname = "core_uses_pid",
354 .data = &core_uses_pid,
355 .maxlen = sizeof(int),
356 .mode = 0644,
357 .proc_handler = &proc_dointvec,
358 },
359 {
360 .ctl_name = KERN_CORE_PATTERN,
361 .procname = "core_pattern",
362 .data = core_pattern,
363 .maxlen = CORENAME_MAX_SIZE,
364 .mode = 0644,
365 .proc_handler = &proc_dostring,
366 .strategy = &sysctl_string,
367 },
368 #ifdef CONFIG_PROC_SYSCTL
369 {
370 .procname = "tainted",
371 .data = &tainted,
372 .maxlen = sizeof(int),
373 .mode = 0644,
374 .proc_handler = &proc_dointvec_taint,
375 },
376 #endif
377 #ifdef CONFIG_LATENCYTOP
378 {
379 .procname = "latencytop",
380 .data = &latencytop_enabled,
381 .maxlen = sizeof(int),
382 .mode = 0644,
383 .proc_handler = &proc_dointvec,
384 },
385 #endif
386 #ifdef CONFIG_BLK_DEV_INITRD
387 {
388 .ctl_name = KERN_REALROOTDEV,
389 .procname = "real-root-dev",
390 .data = &real_root_dev,
391 .maxlen = sizeof(int),
392 .mode = 0644,
393 .proc_handler = &proc_dointvec,
394 },
395 #endif
396 {
397 .ctl_name = CTL_UNNUMBERED,
398 .procname = "print-fatal-signals",
399 .data = &print_fatal_signals,
400 .maxlen = sizeof(int),
401 .mode = 0644,
402 .proc_handler = &proc_dointvec,
403 },
404 #ifdef __sparc__
405 {
406 .ctl_name = KERN_SPARC_REBOOT,
407 .procname = "reboot-cmd",
408 .data = reboot_command,
409 .maxlen = 256,
410 .mode = 0644,
411 .proc_handler = &proc_dostring,
412 .strategy = &sysctl_string,
413 },
414 {
415 .ctl_name = KERN_SPARC_STOP_A,
416 .procname = "stop-a",
417 .data = &stop_a_enabled,
418 .maxlen = sizeof (int),
419 .mode = 0644,
420 .proc_handler = &proc_dointvec,
421 },
422 {
423 .ctl_name = KERN_SPARC_SCONS_PWROFF,
424 .procname = "scons-poweroff",
425 .data = &scons_pwroff,
426 .maxlen = sizeof (int),
427 .mode = 0644,
428 .proc_handler = &proc_dointvec,
429 },
430 #endif
431 #ifdef __hppa__
432 {
433 .ctl_name = KERN_HPPA_PWRSW,
434 .procname = "soft-power",
435 .data = &pwrsw_enabled,
436 .maxlen = sizeof (int),
437 .mode = 0644,
438 .proc_handler = &proc_dointvec,
439 },
440 {
441 .ctl_name = KERN_HPPA_UNALIGNED,
442 .procname = "unaligned-trap",
443 .data = &unaligned_enabled,
444 .maxlen = sizeof (int),
445 .mode = 0644,
446 .proc_handler = &proc_dointvec,
447 },
448 #endif
449 {
450 .ctl_name = KERN_CTLALTDEL,
451 .procname = "ctrl-alt-del",
452 .data = &C_A_D,
453 .maxlen = sizeof(int),
454 .mode = 0644,
455 .proc_handler = &proc_dointvec,
456 },
457 #ifdef CONFIG_KMOD
458 {
459 .ctl_name = KERN_MODPROBE,
460 .procname = "modprobe",
461 .data = &modprobe_path,
462 .maxlen = KMOD_PATH_LEN,
463 .mode = 0644,
464 .proc_handler = &proc_dostring,
465 .strategy = &sysctl_string,
466 },
467 #endif
468 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
469 {
470 .ctl_name = KERN_HOTPLUG,
471 .procname = "hotplug",
472 .data = &uevent_helper,
473 .maxlen = UEVENT_HELPER_PATH_LEN,
474 .mode = 0644,
475 .proc_handler = &proc_dostring,
476 .strategy = &sysctl_string,
477 },
478 #endif
479 #ifdef CONFIG_CHR_DEV_SG
480 {
481 .ctl_name = KERN_SG_BIG_BUFF,
482 .procname = "sg-big-buff",
483 .data = &sg_big_buff,
484 .maxlen = sizeof (int),
485 .mode = 0444,
486 .proc_handler = &proc_dointvec,
487 },
488 #endif
489 #ifdef CONFIG_BSD_PROCESS_ACCT
490 {
491 .ctl_name = KERN_ACCT,
492 .procname = "acct",
493 .data = &acct_parm,
494 .maxlen = 3*sizeof(int),
495 .mode = 0644,
496 .proc_handler = &proc_dointvec,
497 },
498 #endif
499 #ifdef CONFIG_MAGIC_SYSRQ
500 {
501 .ctl_name = KERN_SYSRQ,
502 .procname = "sysrq",
503 .data = &__sysrq_enabled,
504 .maxlen = sizeof (int),
505 .mode = 0644,
506 .proc_handler = &proc_dointvec,
507 },
508 #endif
509 #ifdef CONFIG_PROC_SYSCTL
510 {
511 .procname = "cad_pid",
512 .data = NULL,
513 .maxlen = sizeof (int),
514 .mode = 0600,
515 .proc_handler = &proc_do_cad_pid,
516 },
517 #endif
518 {
519 .ctl_name = KERN_MAX_THREADS,
520 .procname = "threads-max",
521 .data = &max_threads,
522 .maxlen = sizeof(int),
523 .mode = 0644,
524 .proc_handler = &proc_dointvec,
525 },
526 {
527 .ctl_name = KERN_RANDOM,
528 .procname = "random",
529 .mode = 0555,
530 .child = random_table,
531 },
532 {
533 .ctl_name = KERN_OVERFLOWUID,
534 .procname = "overflowuid",
535 .data = &overflowuid,
536 .maxlen = sizeof(int),
537 .mode = 0644,
538 .proc_handler = &proc_dointvec_minmax,
539 .strategy = &sysctl_intvec,
540 .extra1 = &minolduid,
541 .extra2 = &maxolduid,
542 },
543 {
544 .ctl_name = KERN_OVERFLOWGID,
545 .procname = "overflowgid",
546 .data = &overflowgid,
547 .maxlen = sizeof(int),
548 .mode = 0644,
549 .proc_handler = &proc_dointvec_minmax,
550 .strategy = &sysctl_intvec,
551 .extra1 = &minolduid,
552 .extra2 = &maxolduid,
553 },
554 #ifdef CONFIG_S390
555 #ifdef CONFIG_MATHEMU
556 {
557 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
558 .procname = "ieee_emulation_warnings",
559 .data = &sysctl_ieee_emulation_warnings,
560 .maxlen = sizeof(int),
561 .mode = 0644,
562 .proc_handler = &proc_dointvec,
563 },
564 #endif
565 #ifdef CONFIG_NO_IDLE_HZ
566 {
567 .ctl_name = KERN_HZ_TIMER,
568 .procname = "hz_timer",
569 .data = &sysctl_hz_timer,
570 .maxlen = sizeof(int),
571 .mode = 0644,
572 .proc_handler = &proc_dointvec,
573 },
574 #endif
575 {
576 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
577 .procname = "userprocess_debug",
578 .data = &sysctl_userprocess_debug,
579 .maxlen = sizeof(int),
580 .mode = 0644,
581 .proc_handler = &proc_dointvec,
582 },
583 #endif
584 {
585 .ctl_name = KERN_PIDMAX,
586 .procname = "pid_max",
587 .data = &pid_max,
588 .maxlen = sizeof (int),
589 .mode = 0644,
590 .proc_handler = &proc_dointvec_minmax,
591 .strategy = sysctl_intvec,
592 .extra1 = &pid_max_min,
593 .extra2 = &pid_max_max,
594 },
595 {
596 .ctl_name = KERN_PANIC_ON_OOPS,
597 .procname = "panic_on_oops",
598 .data = &panic_on_oops,
599 .maxlen = sizeof(int),
600 .mode = 0644,
601 .proc_handler = &proc_dointvec,
602 },
603 #if defined CONFIG_PRINTK
604 {
605 .ctl_name = KERN_PRINTK,
606 .procname = "printk",
607 .data = &console_loglevel,
608 .maxlen = 4*sizeof(int),
609 .mode = 0644,
610 .proc_handler = &proc_dointvec,
611 },
612 {
613 .ctl_name = KERN_PRINTK_RATELIMIT,
614 .procname = "printk_ratelimit",
615 .data = &printk_ratelimit_jiffies,
616 .maxlen = sizeof(int),
617 .mode = 0644,
618 .proc_handler = &proc_dointvec_jiffies,
619 .strategy = &sysctl_jiffies,
620 },
621 {
622 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
623 .procname = "printk_ratelimit_burst",
624 .data = &printk_ratelimit_burst,
625 .maxlen = sizeof(int),
626 .mode = 0644,
627 .proc_handler = &proc_dointvec,
628 },
629 #endif
630 {
631 .ctl_name = KERN_NGROUPS_MAX,
632 .procname = "ngroups_max",
633 .data = &ngroups_max,
634 .maxlen = sizeof (int),
635 .mode = 0444,
636 .proc_handler = &proc_dointvec,
637 },
638 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
639 {
640 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
641 .procname = "unknown_nmi_panic",
642 .data = &unknown_nmi_panic,
643 .maxlen = sizeof (int),
644 .mode = 0644,
645 .proc_handler = &proc_dointvec,
646 },
647 {
648 .procname = "nmi_watchdog",
649 .data = &nmi_watchdog_enabled,
650 .maxlen = sizeof (int),
651 .mode = 0644,
652 .proc_handler = &proc_nmi_enabled,
653 },
654 #endif
655 #if defined(CONFIG_X86)
656 {
657 .ctl_name = KERN_PANIC_ON_NMI,
658 .procname = "panic_on_unrecovered_nmi",
659 .data = &panic_on_unrecovered_nmi,
660 .maxlen = sizeof(int),
661 .mode = 0644,
662 .proc_handler = &proc_dointvec,
663 },
664 {
665 .ctl_name = KERN_BOOTLOADER_TYPE,
666 .procname = "bootloader_type",
667 .data = &bootloader_type,
668 .maxlen = sizeof (int),
669 .mode = 0444,
670 .proc_handler = &proc_dointvec,
671 },
672 {
673 .ctl_name = CTL_UNNUMBERED,
674 .procname = "kstack_depth_to_print",
675 .data = &kstack_depth_to_print,
676 .maxlen = sizeof(int),
677 .mode = 0644,
678 .proc_handler = &proc_dointvec,
679 },
680 {
681 .ctl_name = CTL_UNNUMBERED,
682 .procname = "io_delay_type",
683 .data = &io_delay_type,
684 .maxlen = sizeof(int),
685 .mode = 0644,
686 .proc_handler = &proc_dointvec,
687 },
688 #endif
689 #if defined(CONFIG_MMU)
690 {
691 .ctl_name = KERN_RANDOMIZE,
692 .procname = "randomize_va_space",
693 .data = &randomize_va_space,
694 .maxlen = sizeof(int),
695 .mode = 0644,
696 .proc_handler = &proc_dointvec,
697 },
698 #endif
699 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
700 {
701 .ctl_name = KERN_SPIN_RETRY,
702 .procname = "spin_retry",
703 .data = &spin_retry,
704 .maxlen = sizeof (int),
705 .mode = 0644,
706 .proc_handler = &proc_dointvec,
707 },
708 #endif
709 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
710 {
711 .procname = "acpi_video_flags",
712 .data = &acpi_realmode_flags,
713 .maxlen = sizeof (unsigned long),
714 .mode = 0644,
715 .proc_handler = &proc_doulongvec_minmax,
716 },
717 #endif
718 #ifdef CONFIG_IA64
719 {
720 .ctl_name = KERN_IA64_UNALIGNED,
721 .procname = "ignore-unaligned-usertrap",
722 .data = &no_unaligned_warning,
723 .maxlen = sizeof (int),
724 .mode = 0644,
725 .proc_handler = &proc_dointvec,
726 },
727 #endif
728 #ifdef CONFIG_DETECT_SOFTLOCKUP
729 {
730 .ctl_name = CTL_UNNUMBERED,
731 .procname = "softlockup_thresh",
732 .data = &softlockup_thresh,
733 .maxlen = sizeof(unsigned long),
734 .mode = 0644,
735 .proc_handler = &proc_doulongvec_minmax,
736 .strategy = &sysctl_intvec,
737 .extra1 = &one,
738 .extra2 = &sixty,
739 },
740 {
741 .ctl_name = CTL_UNNUMBERED,
742 .procname = "hung_task_check_count",
743 .data = &sysctl_hung_task_check_count,
744 .maxlen = sizeof(unsigned long),
745 .mode = 0644,
746 .proc_handler = &proc_doulongvec_minmax,
747 .strategy = &sysctl_intvec,
748 },
749 {
750 .ctl_name = CTL_UNNUMBERED,
751 .procname = "hung_task_timeout_secs",
752 .data = &sysctl_hung_task_timeout_secs,
753 .maxlen = sizeof(unsigned long),
754 .mode = 0644,
755 .proc_handler = &proc_doulongvec_minmax,
756 .strategy = &sysctl_intvec,
757 },
758 {
759 .ctl_name = CTL_UNNUMBERED,
760 .procname = "hung_task_warnings",
761 .data = &sysctl_hung_task_warnings,
762 .maxlen = sizeof(unsigned long),
763 .mode = 0644,
764 .proc_handler = &proc_doulongvec_minmax,
765 .strategy = &sysctl_intvec,
766 },
767 #endif
768 #ifdef CONFIG_COMPAT
769 {
770 .ctl_name = KERN_COMPAT_LOG,
771 .procname = "compat-log",
772 .data = &compat_log,
773 .maxlen = sizeof (int),
774 .mode = 0644,
775 .proc_handler = &proc_dointvec,
776 },
777 #endif
778 #ifdef CONFIG_RT_MUTEXES
779 {
780 .ctl_name = KERN_MAX_LOCK_DEPTH,
781 .procname = "max_lock_depth",
782 .data = &max_lock_depth,
783 .maxlen = sizeof(int),
784 .mode = 0644,
785 .proc_handler = &proc_dointvec,
786 },
787 #endif
788 #ifdef CONFIG_PROC_FS
789 {
790 .ctl_name = CTL_UNNUMBERED,
791 .procname = "maps_protect",
792 .data = &maps_protect,
793 .maxlen = sizeof(int),
794 .mode = 0644,
795 .proc_handler = &proc_dointvec,
796 },
797 #endif
798 {
799 .ctl_name = CTL_UNNUMBERED,
800 .procname = "poweroff_cmd",
801 .data = &poweroff_cmd,
802 .maxlen = POWEROFF_CMD_PATH_LEN,
803 .mode = 0644,
804 .proc_handler = &proc_dostring,
805 .strategy = &sysctl_string,
806 },
807 #ifdef CONFIG_KEYS
808 {
809 .ctl_name = CTL_UNNUMBERED,
810 .procname = "keys",
811 .mode = 0555,
812 .child = key_sysctls,
813 },
814 #endif
815 /*
816 * NOTE: do not add new entries to this table unless you have read
817 * Documentation/sysctl/ctl_unnumbered.txt
818 */
819 { .ctl_name = 0 }
820 };
821
822 static struct ctl_table vm_table[] = {
823 {
824 .ctl_name = VM_OVERCOMMIT_MEMORY,
825 .procname = "overcommit_memory",
826 .data = &sysctl_overcommit_memory,
827 .maxlen = sizeof(sysctl_overcommit_memory),
828 .mode = 0644,
829 .proc_handler = &proc_dointvec,
830 },
831 {
832 .ctl_name = VM_PANIC_ON_OOM,
833 .procname = "panic_on_oom",
834 .data = &sysctl_panic_on_oom,
835 .maxlen = sizeof(sysctl_panic_on_oom),
836 .mode = 0644,
837 .proc_handler = &proc_dointvec,
838 },
839 {
840 .ctl_name = CTL_UNNUMBERED,
841 .procname = "oom_kill_allocating_task",
842 .data = &sysctl_oom_kill_allocating_task,
843 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
844 .mode = 0644,
845 .proc_handler = &proc_dointvec,
846 },
847 {
848 .ctl_name = CTL_UNNUMBERED,
849 .procname = "oom_dump_tasks",
850 .data = &sysctl_oom_dump_tasks,
851 .maxlen = sizeof(sysctl_oom_dump_tasks),
852 .mode = 0644,
853 .proc_handler = &proc_dointvec,
854 },
855 {
856 .ctl_name = VM_OVERCOMMIT_RATIO,
857 .procname = "overcommit_ratio",
858 .data = &sysctl_overcommit_ratio,
859 .maxlen = sizeof(sysctl_overcommit_ratio),
860 .mode = 0644,
861 .proc_handler = &proc_dointvec,
862 },
863 {
864 .ctl_name = VM_PAGE_CLUSTER,
865 .procname = "page-cluster",
866 .data = &page_cluster,
867 .maxlen = sizeof(int),
868 .mode = 0644,
869 .proc_handler = &proc_dointvec,
870 },
871 {
872 .ctl_name = VM_DIRTY_BACKGROUND,
873 .procname = "dirty_background_ratio",
874 .data = &dirty_background_ratio,
875 .maxlen = sizeof(dirty_background_ratio),
876 .mode = 0644,
877 .proc_handler = &proc_dointvec_minmax,
878 .strategy = &sysctl_intvec,
879 .extra1 = &zero,
880 .extra2 = &one_hundred,
881 },
882 {
883 .ctl_name = VM_DIRTY_RATIO,
884 .procname = "dirty_ratio",
885 .data = &vm_dirty_ratio,
886 .maxlen = sizeof(vm_dirty_ratio),
887 .mode = 0644,
888 .proc_handler = &dirty_ratio_handler,
889 .strategy = &sysctl_intvec,
890 .extra1 = &zero,
891 .extra2 = &one_hundred,
892 },
893 {
894 .procname = "dirty_writeback_centisecs",
895 .data = &dirty_writeback_interval,
896 .maxlen = sizeof(dirty_writeback_interval),
897 .mode = 0644,
898 .proc_handler = &dirty_writeback_centisecs_handler,
899 },
900 {
901 .procname = "dirty_expire_centisecs",
902 .data = &dirty_expire_interval,
903 .maxlen = sizeof(dirty_expire_interval),
904 .mode = 0644,
905 .proc_handler = &proc_dointvec_userhz_jiffies,
906 },
907 {
908 .ctl_name = VM_NR_PDFLUSH_THREADS,
909 .procname = "nr_pdflush_threads",
910 .data = &nr_pdflush_threads,
911 .maxlen = sizeof nr_pdflush_threads,
912 .mode = 0444 /* read-only*/,
913 .proc_handler = &proc_dointvec,
914 },
915 {
916 .ctl_name = VM_SWAPPINESS,
917 .procname = "swappiness",
918 .data = &vm_swappiness,
919 .maxlen = sizeof(vm_swappiness),
920 .mode = 0644,
921 .proc_handler = &proc_dointvec_minmax,
922 .strategy = &sysctl_intvec,
923 .extra1 = &zero,
924 .extra2 = &one_hundred,
925 },
926 #ifdef CONFIG_HUGETLB_PAGE
927 {
928 .procname = "nr_hugepages",
929 .data = &max_huge_pages,
930 .maxlen = sizeof(unsigned long),
931 .mode = 0644,
932 .proc_handler = &hugetlb_sysctl_handler,
933 .extra1 = (void *)&hugetlb_zero,
934 .extra2 = (void *)&hugetlb_infinity,
935 },
936 {
937 .ctl_name = VM_HUGETLB_GROUP,
938 .procname = "hugetlb_shm_group",
939 .data = &sysctl_hugetlb_shm_group,
940 .maxlen = sizeof(gid_t),
941 .mode = 0644,
942 .proc_handler = &proc_dointvec,
943 },
944 {
945 .ctl_name = CTL_UNNUMBERED,
946 .procname = "hugepages_treat_as_movable",
947 .data = &hugepages_treat_as_movable,
948 .maxlen = sizeof(int),
949 .mode = 0644,
950 .proc_handler = &hugetlb_treat_movable_handler,
951 },
952 {
953 .ctl_name = CTL_UNNUMBERED,
954 .procname = "nr_overcommit_hugepages",
955 .data = &sysctl_overcommit_huge_pages,
956 .maxlen = sizeof(sysctl_overcommit_huge_pages),
957 .mode = 0644,
958 .proc_handler = &hugetlb_overcommit_handler,
959 },
960 #endif
961 {
962 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
963 .procname = "lowmem_reserve_ratio",
964 .data = &sysctl_lowmem_reserve_ratio,
965 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
966 .mode = 0644,
967 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
968 .strategy = &sysctl_intvec,
969 },
970 {
971 .ctl_name = VM_DROP_PAGECACHE,
972 .procname = "drop_caches",
973 .data = &sysctl_drop_caches,
974 .maxlen = sizeof(int),
975 .mode = 0644,
976 .proc_handler = drop_caches_sysctl_handler,
977 .strategy = &sysctl_intvec,
978 },
979 {
980 .ctl_name = VM_MIN_FREE_KBYTES,
981 .procname = "min_free_kbytes",
982 .data = &min_free_kbytes,
983 .maxlen = sizeof(min_free_kbytes),
984 .mode = 0644,
985 .proc_handler = &min_free_kbytes_sysctl_handler,
986 .strategy = &sysctl_intvec,
987 .extra1 = &zero,
988 },
989 {
990 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
991 .procname = "percpu_pagelist_fraction",
992 .data = &percpu_pagelist_fraction,
993 .maxlen = sizeof(percpu_pagelist_fraction),
994 .mode = 0644,
995 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
996 .strategy = &sysctl_intvec,
997 .extra1 = &min_percpu_pagelist_fract,
998 },
999 #ifdef CONFIG_MMU
1000 {
1001 .ctl_name = VM_MAX_MAP_COUNT,
1002 .procname = "max_map_count",
1003 .data = &sysctl_max_map_count,
1004 .maxlen = sizeof(sysctl_max_map_count),
1005 .mode = 0644,
1006 .proc_handler = &proc_dointvec
1007 },
1008 #endif
1009 {
1010 .ctl_name = VM_LAPTOP_MODE,
1011 .procname = "laptop_mode",
1012 .data = &laptop_mode,
1013 .maxlen = sizeof(laptop_mode),
1014 .mode = 0644,
1015 .proc_handler = &proc_dointvec_jiffies,
1016 .strategy = &sysctl_jiffies,
1017 },
1018 {
1019 .ctl_name = VM_BLOCK_DUMP,
1020 .procname = "block_dump",
1021 .data = &block_dump,
1022 .maxlen = sizeof(block_dump),
1023 .mode = 0644,
1024 .proc_handler = &proc_dointvec,
1025 .strategy = &sysctl_intvec,
1026 .extra1 = &zero,
1027 },
1028 {
1029 .ctl_name = VM_VFS_CACHE_PRESSURE,
1030 .procname = "vfs_cache_pressure",
1031 .data = &sysctl_vfs_cache_pressure,
1032 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1033 .mode = 0644,
1034 .proc_handler = &proc_dointvec,
1035 .strategy = &sysctl_intvec,
1036 .extra1 = &zero,
1037 },
1038 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1039 {
1040 .ctl_name = VM_LEGACY_VA_LAYOUT,
1041 .procname = "legacy_va_layout",
1042 .data = &sysctl_legacy_va_layout,
1043 .maxlen = sizeof(sysctl_legacy_va_layout),
1044 .mode = 0644,
1045 .proc_handler = &proc_dointvec,
1046 .strategy = &sysctl_intvec,
1047 .extra1 = &zero,
1048 },
1049 #endif
1050 #ifdef CONFIG_NUMA
1051 {
1052 .ctl_name = VM_ZONE_RECLAIM_MODE,
1053 .procname = "zone_reclaim_mode",
1054 .data = &zone_reclaim_mode,
1055 .maxlen = sizeof(zone_reclaim_mode),
1056 .mode = 0644,
1057 .proc_handler = &proc_dointvec,
1058 .strategy = &sysctl_intvec,
1059 .extra1 = &zero,
1060 },
1061 {
1062 .ctl_name = VM_MIN_UNMAPPED,
1063 .procname = "min_unmapped_ratio",
1064 .data = &sysctl_min_unmapped_ratio,
1065 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1066 .mode = 0644,
1067 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
1068 .strategy = &sysctl_intvec,
1069 .extra1 = &zero,
1070 .extra2 = &one_hundred,
1071 },
1072 {
1073 .ctl_name = VM_MIN_SLAB,
1074 .procname = "min_slab_ratio",
1075 .data = &sysctl_min_slab_ratio,
1076 .maxlen = sizeof(sysctl_min_slab_ratio),
1077 .mode = 0644,
1078 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1079 .strategy = &sysctl_intvec,
1080 .extra1 = &zero,
1081 .extra2 = &one_hundred,
1082 },
1083 #endif
1084 #ifdef CONFIG_SMP
1085 {
1086 .ctl_name = CTL_UNNUMBERED,
1087 .procname = "stat_interval",
1088 .data = &sysctl_stat_interval,
1089 .maxlen = sizeof(sysctl_stat_interval),
1090 .mode = 0644,
1091 .proc_handler = &proc_dointvec_jiffies,
1092 .strategy = &sysctl_jiffies,
1093 },
1094 #endif
1095 #ifdef CONFIG_SECURITY
1096 {
1097 .ctl_name = CTL_UNNUMBERED,
1098 .procname = "mmap_min_addr",
1099 .data = &mmap_min_addr,
1100 .maxlen = sizeof(unsigned long),
1101 .mode = 0644,
1102 .proc_handler = &proc_doulongvec_minmax,
1103 },
1104 #endif
1105 #ifdef CONFIG_NUMA
1106 {
1107 .ctl_name = CTL_UNNUMBERED,
1108 .procname = "numa_zonelist_order",
1109 .data = &numa_zonelist_order,
1110 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1111 .mode = 0644,
1112 .proc_handler = &numa_zonelist_order_handler,
1113 .strategy = &sysctl_string,
1114 },
1115 #endif
1116 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1117 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1118 {
1119 .ctl_name = VM_VDSO_ENABLED,
1120 .procname = "vdso_enabled",
1121 .data = &vdso_enabled,
1122 .maxlen = sizeof(vdso_enabled),
1123 .mode = 0644,
1124 .proc_handler = &proc_dointvec,
1125 .strategy = &sysctl_intvec,
1126 .extra1 = &zero,
1127 },
1128 #endif
1129 #ifdef CONFIG_HIGHMEM
1130 {
1131 .ctl_name = CTL_UNNUMBERED,
1132 .procname = "highmem_is_dirtyable",
1133 .data = &vm_highmem_is_dirtyable,
1134 .maxlen = sizeof(vm_highmem_is_dirtyable),
1135 .mode = 0644,
1136 .proc_handler = &proc_dointvec_minmax,
1137 .strategy = &sysctl_intvec,
1138 .extra1 = &zero,
1139 .extra2 = &one,
1140 },
1141 #endif
1142 /*
1143 * NOTE: do not add new entries to this table unless you have read
1144 * Documentation/sysctl/ctl_unnumbered.txt
1145 */
1146 { .ctl_name = 0 }
1147 };
1148
1149 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1150 static struct ctl_table binfmt_misc_table[] = {
1151 { .ctl_name = 0 }
1152 };
1153 #endif
1154
1155 static struct ctl_table fs_table[] = {
1156 {
1157 .ctl_name = FS_NRINODE,
1158 .procname = "inode-nr",
1159 .data = &inodes_stat,
1160 .maxlen = 2*sizeof(int),
1161 .mode = 0444,
1162 .proc_handler = &proc_dointvec,
1163 },
1164 {
1165 .ctl_name = FS_STATINODE,
1166 .procname = "inode-state",
1167 .data = &inodes_stat,
1168 .maxlen = 7*sizeof(int),
1169 .mode = 0444,
1170 .proc_handler = &proc_dointvec,
1171 },
1172 {
1173 .procname = "file-nr",
1174 .data = &files_stat,
1175 .maxlen = 3*sizeof(int),
1176 .mode = 0444,
1177 .proc_handler = &proc_nr_files,
1178 },
1179 {
1180 .ctl_name = FS_MAXFILE,
1181 .procname = "file-max",
1182 .data = &files_stat.max_files,
1183 .maxlen = sizeof(int),
1184 .mode = 0644,
1185 .proc_handler = &proc_dointvec,
1186 },
1187 {
1188 .ctl_name = CTL_UNNUMBERED,
1189 .procname = "nr_open",
1190 .data = &sysctl_nr_open,
1191 .maxlen = sizeof(int),
1192 .mode = 0644,
1193 .proc_handler = &proc_dointvec,
1194 },
1195 {
1196 .ctl_name = FS_DENTRY,
1197 .procname = "dentry-state",
1198 .data = &dentry_stat,
1199 .maxlen = 6*sizeof(int),
1200 .mode = 0444,
1201 .proc_handler = &proc_dointvec,
1202 },
1203 {
1204 .ctl_name = FS_OVERFLOWUID,
1205 .procname = "overflowuid",
1206 .data = &fs_overflowuid,
1207 .maxlen = sizeof(int),
1208 .mode = 0644,
1209 .proc_handler = &proc_dointvec_minmax,
1210 .strategy = &sysctl_intvec,
1211 .extra1 = &minolduid,
1212 .extra2 = &maxolduid,
1213 },
1214 {
1215 .ctl_name = FS_OVERFLOWGID,
1216 .procname = "overflowgid",
1217 .data = &fs_overflowgid,
1218 .maxlen = sizeof(int),
1219 .mode = 0644,
1220 .proc_handler = &proc_dointvec_minmax,
1221 .strategy = &sysctl_intvec,
1222 .extra1 = &minolduid,
1223 .extra2 = &maxolduid,
1224 },
1225 {
1226 .ctl_name = FS_LEASES,
1227 .procname = "leases-enable",
1228 .data = &leases_enable,
1229 .maxlen = sizeof(int),
1230 .mode = 0644,
1231 .proc_handler = &proc_dointvec,
1232 },
1233 #ifdef CONFIG_DNOTIFY
1234 {
1235 .ctl_name = FS_DIR_NOTIFY,
1236 .procname = "dir-notify-enable",
1237 .data = &dir_notify_enable,
1238 .maxlen = sizeof(int),
1239 .mode = 0644,
1240 .proc_handler = &proc_dointvec,
1241 },
1242 #endif
1243 #ifdef CONFIG_MMU
1244 {
1245 .ctl_name = FS_LEASE_TIME,
1246 .procname = "lease-break-time",
1247 .data = &lease_break_time,
1248 .maxlen = sizeof(int),
1249 .mode = 0644,
1250 .proc_handler = &proc_dointvec_minmax,
1251 .strategy = &sysctl_intvec,
1252 .extra1 = &zero,
1253 .extra2 = &two,
1254 },
1255 {
1256 .procname = "aio-nr",
1257 .data = &aio_nr,
1258 .maxlen = sizeof(aio_nr),
1259 .mode = 0444,
1260 .proc_handler = &proc_doulongvec_minmax,
1261 },
1262 {
1263 .procname = "aio-max-nr",
1264 .data = &aio_max_nr,
1265 .maxlen = sizeof(aio_max_nr),
1266 .mode = 0644,
1267 .proc_handler = &proc_doulongvec_minmax,
1268 },
1269 #ifdef CONFIG_INOTIFY_USER
1270 {
1271 .ctl_name = FS_INOTIFY,
1272 .procname = "inotify",
1273 .mode = 0555,
1274 .child = inotify_table,
1275 },
1276 #endif
1277 #endif
1278 {
1279 .ctl_name = KERN_SETUID_DUMPABLE,
1280 .procname = "suid_dumpable",
1281 .data = &suid_dumpable,
1282 .maxlen = sizeof(int),
1283 .mode = 0644,
1284 .proc_handler = &proc_dointvec,
1285 },
1286 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1287 {
1288 .ctl_name = CTL_UNNUMBERED,
1289 .procname = "binfmt_misc",
1290 .mode = 0555,
1291 .child = binfmt_misc_table,
1292 },
1293 #endif
1294 /*
1295 * NOTE: do not add new entries to this table unless you have read
1296 * Documentation/sysctl/ctl_unnumbered.txt
1297 */
1298 { .ctl_name = 0 }
1299 };
1300
1301 static struct ctl_table debug_table[] = {
1302 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1303 {
1304 .ctl_name = CTL_UNNUMBERED,
1305 .procname = "exception-trace",
1306 .data = &show_unhandled_signals,
1307 .maxlen = sizeof(int),
1308 .mode = 0644,
1309 .proc_handler = proc_dointvec
1310 },
1311 #endif
1312 { .ctl_name = 0 }
1313 };
1314
1315 static struct ctl_table dev_table[] = {
1316 { .ctl_name = 0 }
1317 };
1318
1319 static DEFINE_SPINLOCK(sysctl_lock);
1320
1321 /* called under sysctl_lock */
1322 static int use_table(struct ctl_table_header *p)
1323 {
1324 if (unlikely(p->unregistering))
1325 return 0;
1326 p->used++;
1327 return 1;
1328 }
1329
1330 /* called under sysctl_lock */
1331 static void unuse_table(struct ctl_table_header *p)
1332 {
1333 if (!--p->used)
1334 if (unlikely(p->unregistering))
1335 complete(p->unregistering);
1336 }
1337
1338 /* called under sysctl_lock, will reacquire if has to wait */
1339 static void start_unregistering(struct ctl_table_header *p)
1340 {
1341 /*
1342 * if p->used is 0, nobody will ever touch that entry again;
1343 * we'll eliminate all paths to it before dropping sysctl_lock
1344 */
1345 if (unlikely(p->used)) {
1346 struct completion wait;
1347 init_completion(&wait);
1348 p->unregistering = &wait;
1349 spin_unlock(&sysctl_lock);
1350 wait_for_completion(&wait);
1351 spin_lock(&sysctl_lock);
1352 }
1353 /*
1354 * do not remove from the list until nobody holds it; walking the
1355 * list in do_sysctl() relies on that.
1356 */
1357 list_del_init(&p->ctl_entry);
1358 }
1359
1360 void sysctl_head_finish(struct ctl_table_header *head)
1361 {
1362 if (!head)
1363 return;
1364 spin_lock(&sysctl_lock);
1365 unuse_table(head);
1366 spin_unlock(&sysctl_lock);
1367 }
1368
1369 static struct list_head *
1370 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1371 {
1372 struct list_head *header_list;
1373 header_list = &root->header_list;
1374 if (root->lookup)
1375 header_list = root->lookup(root, namespaces);
1376 return header_list;
1377 }
1378
1379 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1380 struct ctl_table_header *prev)
1381 {
1382 struct ctl_table_root *root;
1383 struct list_head *header_list;
1384 struct ctl_table_header *head;
1385 struct list_head *tmp;
1386
1387 spin_lock(&sysctl_lock);
1388 if (prev) {
1389 head = prev;
1390 tmp = &prev->ctl_entry;
1391 unuse_table(prev);
1392 goto next;
1393 }
1394 tmp = &root_table_header.ctl_entry;
1395 for (;;) {
1396 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1397
1398 if (!use_table(head))
1399 goto next;
1400 spin_unlock(&sysctl_lock);
1401 return head;
1402 next:
1403 root = head->root;
1404 tmp = tmp->next;
1405 header_list = lookup_header_list(root, namespaces);
1406 if (tmp != header_list)
1407 continue;
1408
1409 do {
1410 root = list_entry(root->root_list.next,
1411 struct ctl_table_root, root_list);
1412 if (root == &sysctl_table_root)
1413 goto out;
1414 header_list = lookup_header_list(root, namespaces);
1415 } while (list_empty(header_list));
1416 tmp = header_list->next;
1417 }
1418 out:
1419 spin_unlock(&sysctl_lock);
1420 return NULL;
1421 }
1422
1423 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1424 {
1425 return __sysctl_head_next(current->nsproxy, prev);
1426 }
1427
1428 void register_sysctl_root(struct ctl_table_root *root)
1429 {
1430 spin_lock(&sysctl_lock);
1431 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1432 spin_unlock(&sysctl_lock);
1433 }
1434
1435 #ifdef CONFIG_SYSCTL_SYSCALL
1436 /* Perform the actual read/write of a sysctl table entry. */
1437 static int do_sysctl_strategy(struct ctl_table_root *root,
1438 struct ctl_table *table,
1439 int __user *name, int nlen,
1440 void __user *oldval, size_t __user *oldlenp,
1441 void __user *newval, size_t newlen)
1442 {
1443 int op = 0, rc;
1444
1445 if (oldval)
1446 op |= 004;
1447 if (newval)
1448 op |= 002;
1449 if (sysctl_perm(root, table, op))
1450 return -EPERM;
1451
1452 if (table->strategy) {
1453 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1454 newval, newlen);
1455 if (rc < 0)
1456 return rc;
1457 if (rc > 0)
1458 return 0;
1459 }
1460
1461 /* If there is no strategy routine, or if the strategy returns
1462 * zero, proceed with automatic r/w */
1463 if (table->data && table->maxlen) {
1464 rc = sysctl_data(table, name, nlen, oldval, oldlenp,
1465 newval, newlen);
1466 if (rc < 0)
1467 return rc;
1468 }
1469 return 0;
1470 }
1471
1472 static int parse_table(int __user *name, int nlen,
1473 void __user *oldval, size_t __user *oldlenp,
1474 void __user *newval, size_t newlen,
1475 struct ctl_table_root *root,
1476 struct ctl_table *table)
1477 {
1478 int n;
1479 repeat:
1480 if (!nlen)
1481 return -ENOTDIR;
1482 if (get_user(n, name))
1483 return -EFAULT;
1484 for ( ; table->ctl_name || table->procname; table++) {
1485 if (!table->ctl_name)
1486 continue;
1487 if (n == table->ctl_name) {
1488 int error;
1489 if (table->child) {
1490 if (sysctl_perm(root, table, 001))
1491 return -EPERM;
1492 name++;
1493 nlen--;
1494 table = table->child;
1495 goto repeat;
1496 }
1497 error = do_sysctl_strategy(root, table, name, nlen,
1498 oldval, oldlenp,
1499 newval, newlen);
1500 return error;
1501 }
1502 }
1503 return -ENOTDIR;
1504 }
1505
1506 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1507 void __user *newval, size_t newlen)
1508 {
1509 struct ctl_table_header *head;
1510 int error = -ENOTDIR;
1511
1512 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1513 return -ENOTDIR;
1514 if (oldval) {
1515 int old_len;
1516 if (!oldlenp || get_user(old_len, oldlenp))
1517 return -EFAULT;
1518 }
1519
1520 for (head = sysctl_head_next(NULL); head;
1521 head = sysctl_head_next(head)) {
1522 error = parse_table(name, nlen, oldval, oldlenp,
1523 newval, newlen,
1524 head->root, head->ctl_table);
1525 if (error != -ENOTDIR) {
1526 sysctl_head_finish(head);
1527 break;
1528 }
1529 }
1530 return error;
1531 }
1532
1533 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1534 {
1535 struct __sysctl_args tmp;
1536 int error;
1537
1538 if (copy_from_user(&tmp, args, sizeof(tmp)))
1539 return -EFAULT;
1540
1541 error = deprecated_sysctl_warning(&tmp);
1542 if (error)
1543 goto out;
1544
1545 lock_kernel();
1546 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1547 tmp.newval, tmp.newlen);
1548 unlock_kernel();
1549 out:
1550 return error;
1551 }
1552 #endif /* CONFIG_SYSCTL_SYSCALL */
1553
1554 /*
1555 * sysctl_perm does NOT grant the superuser all rights automatically, because
1556 * some sysctl variables are readonly even to root.
1557 */
1558
1559 static int test_perm(int mode, int op)
1560 {
1561 if (!current->euid)
1562 mode >>= 6;
1563 else if (in_egroup_p(0))
1564 mode >>= 3;
1565 if ((mode & op & 0007) == op)
1566 return 0;
1567 return -EACCES;
1568 }
1569
1570 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1571 {
1572 int error;
1573 int mode;
1574
1575 error = security_sysctl(table, op);
1576 if (error)
1577 return error;
1578
1579 if (root->permissions)
1580 mode = root->permissions(root, current->nsproxy, table);
1581 else
1582 mode = table->mode;
1583
1584 return test_perm(mode, op);
1585 }
1586
1587 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1588 {
1589 for (; table->ctl_name || table->procname; table++) {
1590 table->parent = parent;
1591 if (table->child)
1592 sysctl_set_parent(table, table->child);
1593 }
1594 }
1595
1596 static __init int sysctl_init(void)
1597 {
1598 sysctl_set_parent(NULL, root_table);
1599 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1600 {
1601 int err;
1602 err = sysctl_check_table(current->nsproxy, root_table);
1603 }
1604 #endif
1605 return 0;
1606 }
1607
1608 core_initcall(sysctl_init);
1609
1610 /**
1611 * __register_sysctl_paths - register a sysctl hierarchy
1612 * @root: List of sysctl headers to register on
1613 * @namespaces: Data to compute which lists of sysctl entries are visible
1614 * @path: The path to the directory the sysctl table is in.
1615 * @table: the top-level table structure
1616 *
1617 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1618 * array. A completely 0 filled entry terminates the table.
1619 *
1620 * The members of the &struct ctl_table structure are used as follows:
1621 *
1622 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1623 * must be unique within that level of sysctl
1624 *
1625 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1626 * enter a sysctl file
1627 *
1628 * data - a pointer to data for use by proc_handler
1629 *
1630 * maxlen - the maximum size in bytes of the data
1631 *
1632 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1633 *
1634 * child - a pointer to the child sysctl table if this entry is a directory, or
1635 * %NULL.
1636 *
1637 * proc_handler - the text handler routine (described below)
1638 *
1639 * strategy - the strategy routine (described below)
1640 *
1641 * de - for internal use by the sysctl routines
1642 *
1643 * extra1, extra2 - extra pointers usable by the proc handler routines
1644 *
1645 * Leaf nodes in the sysctl tree will be represented by a single file
1646 * under /proc; non-leaf nodes will be represented by directories.
1647 *
1648 * sysctl(2) can automatically manage read and write requests through
1649 * the sysctl table. The data and maxlen fields of the ctl_table
1650 * struct enable minimal validation of the values being written to be
1651 * performed, and the mode field allows minimal authentication.
1652 *
1653 * More sophisticated management can be enabled by the provision of a
1654 * strategy routine with the table entry. This will be called before
1655 * any automatic read or write of the data is performed.
1656 *
1657 * The strategy routine may return
1658 *
1659 * < 0 - Error occurred (error is passed to user process)
1660 *
1661 * 0 - OK - proceed with automatic read or write.
1662 *
1663 * > 0 - OK - read or write has been done by the strategy routine, so
1664 * return immediately.
1665 *
1666 * There must be a proc_handler routine for any terminal nodes
1667 * mirrored under /proc/sys (non-terminals are handled by a built-in
1668 * directory handler). Several default handlers are available to
1669 * cover common cases -
1670 *
1671 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1672 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1673 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1674 *
1675 * It is the handler's job to read the input buffer from user memory
1676 * and process it. The handler should return 0 on success.
1677 *
1678 * This routine returns %NULL on a failure to register, and a pointer
1679 * to the table header on success.
1680 */
1681 struct ctl_table_header *__register_sysctl_paths(
1682 struct ctl_table_root *root,
1683 struct nsproxy *namespaces,
1684 const struct ctl_path *path, struct ctl_table *table)
1685 {
1686 struct list_head *header_list;
1687 struct ctl_table_header *header;
1688 struct ctl_table *new, **prevp;
1689 unsigned int n, npath;
1690
1691 /* Count the path components */
1692 for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1693 ;
1694
1695 /*
1696 * For each path component, allocate a 2-element ctl_table array.
1697 * The first array element will be filled with the sysctl entry
1698 * for this, the second will be the sentinel (ctl_name == 0).
1699 *
1700 * We allocate everything in one go so that we don't have to
1701 * worry about freeing additional memory in unregister_sysctl_table.
1702 */
1703 header = kzalloc(sizeof(struct ctl_table_header) +
1704 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1705 if (!header)
1706 return NULL;
1707
1708 new = (struct ctl_table *) (header + 1);
1709
1710 /* Now connect the dots */
1711 prevp = &header->ctl_table;
1712 for (n = 0; n < npath; ++n, ++path) {
1713 /* Copy the procname */
1714 new->procname = path->procname;
1715 new->ctl_name = path->ctl_name;
1716 new->mode = 0555;
1717
1718 *prevp = new;
1719 prevp = &new->child;
1720
1721 new += 2;
1722 }
1723 *prevp = table;
1724 header->ctl_table_arg = table;
1725
1726 INIT_LIST_HEAD(&header->ctl_entry);
1727 header->used = 0;
1728 header->unregistering = NULL;
1729 header->root = root;
1730 sysctl_set_parent(NULL, header->ctl_table);
1731 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1732 if (sysctl_check_table(namespaces, header->ctl_table)) {
1733 kfree(header);
1734 return NULL;
1735 }
1736 #endif
1737 spin_lock(&sysctl_lock);
1738 header_list = lookup_header_list(root, namespaces);
1739 list_add_tail(&header->ctl_entry, header_list);
1740 spin_unlock(&sysctl_lock);
1741
1742 return header;
1743 }
1744
1745 /**
1746 * register_sysctl_table_path - register a sysctl table hierarchy
1747 * @path: The path to the directory the sysctl table is in.
1748 * @table: the top-level table structure
1749 *
1750 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1751 * array. A completely 0 filled entry terminates the table.
1752 *
1753 * See __register_sysctl_paths for more details.
1754 */
1755 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1756 struct ctl_table *table)
1757 {
1758 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1759 path, table);
1760 }
1761
1762 /**
1763 * register_sysctl_table - register a sysctl table hierarchy
1764 * @table: the top-level table structure
1765 *
1766 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1767 * array. A completely 0 filled entry terminates the table.
1768 *
1769 * See register_sysctl_paths for more details.
1770 */
1771 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1772 {
1773 static const struct ctl_path null_path[] = { {} };
1774
1775 return register_sysctl_paths(null_path, table);
1776 }
1777
1778 /**
1779 * unregister_sysctl_table - unregister a sysctl table hierarchy
1780 * @header: the header returned from register_sysctl_table
1781 *
1782 * Unregisters the sysctl table and all children. proc entries may not
1783 * actually be removed until they are no longer used by anyone.
1784 */
1785 void unregister_sysctl_table(struct ctl_table_header * header)
1786 {
1787 might_sleep();
1788
1789 if (header == NULL)
1790 return;
1791
1792 spin_lock(&sysctl_lock);
1793 start_unregistering(header);
1794 spin_unlock(&sysctl_lock);
1795 kfree(header);
1796 }
1797
1798 #else /* !CONFIG_SYSCTL */
1799 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1800 {
1801 return NULL;
1802 }
1803
1804 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1805 struct ctl_table *table)
1806 {
1807 return NULL;
1808 }
1809
1810 void unregister_sysctl_table(struct ctl_table_header * table)
1811 {
1812 }
1813
1814 #endif /* CONFIG_SYSCTL */
1815
1816 /*
1817 * /proc/sys support
1818 */
1819
1820 #ifdef CONFIG_PROC_SYSCTL
1821
1822 static int _proc_do_string(void* data, int maxlen, int write,
1823 struct file *filp, void __user *buffer,
1824 size_t *lenp, loff_t *ppos)
1825 {
1826 size_t len;
1827 char __user *p;
1828 char c;
1829
1830 if (!data || !maxlen || !*lenp) {
1831 *lenp = 0;
1832 return 0;
1833 }
1834
1835 if (write) {
1836 len = 0;
1837 p = buffer;
1838 while (len < *lenp) {
1839 if (get_user(c, p++))
1840 return -EFAULT;
1841 if (c == 0 || c == '\n')
1842 break;
1843 len++;
1844 }
1845 if (len >= maxlen)
1846 len = maxlen-1;
1847 if(copy_from_user(data, buffer, len))
1848 return -EFAULT;
1849 ((char *) data)[len] = 0;
1850 *ppos += *lenp;
1851 } else {
1852 len = strlen(data);
1853 if (len > maxlen)
1854 len = maxlen;
1855
1856 if (*ppos > len) {
1857 *lenp = 0;
1858 return 0;
1859 }
1860
1861 data += *ppos;
1862 len -= *ppos;
1863
1864 if (len > *lenp)
1865 len = *lenp;
1866 if (len)
1867 if(copy_to_user(buffer, data, len))
1868 return -EFAULT;
1869 if (len < *lenp) {
1870 if(put_user('\n', ((char __user *) buffer) + len))
1871 return -EFAULT;
1872 len++;
1873 }
1874 *lenp = len;
1875 *ppos += len;
1876 }
1877 return 0;
1878 }
1879
1880 /**
1881 * proc_dostring - read a string sysctl
1882 * @table: the sysctl table
1883 * @write: %TRUE if this is a write to the sysctl file
1884 * @filp: the file structure
1885 * @buffer: the user buffer
1886 * @lenp: the size of the user buffer
1887 * @ppos: file position
1888 *
1889 * Reads/writes a string from/to the user buffer. If the kernel
1890 * buffer provided is not large enough to hold the string, the
1891 * string is truncated. The copied string is %NULL-terminated.
1892 * If the string is being read by the user process, it is copied
1893 * and a newline '\n' is added. It is truncated if the buffer is
1894 * not large enough.
1895 *
1896 * Returns 0 on success.
1897 */
1898 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
1899 void __user *buffer, size_t *lenp, loff_t *ppos)
1900 {
1901 return _proc_do_string(table->data, table->maxlen, write, filp,
1902 buffer, lenp, ppos);
1903 }
1904
1905
1906 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1907 int *valp,
1908 int write, void *data)
1909 {
1910 if (write) {
1911 *valp = *negp ? -*lvalp : *lvalp;
1912 } else {
1913 int val = *valp;
1914 if (val < 0) {
1915 *negp = -1;
1916 *lvalp = (unsigned long)-val;
1917 } else {
1918 *negp = 0;
1919 *lvalp = (unsigned long)val;
1920 }
1921 }
1922 return 0;
1923 }
1924
1925 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1926 int write, struct file *filp, void __user *buffer,
1927 size_t *lenp, loff_t *ppos,
1928 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1929 int write, void *data),
1930 void *data)
1931 {
1932 #define TMPBUFLEN 21
1933 int *i, vleft, first=1, neg, val;
1934 unsigned long lval;
1935 size_t left, len;
1936
1937 char buf[TMPBUFLEN], *p;
1938 char __user *s = buffer;
1939
1940 if (!tbl_data || !table->maxlen || !*lenp ||
1941 (*ppos && !write)) {
1942 *lenp = 0;
1943 return 0;
1944 }
1945
1946 i = (int *) tbl_data;
1947 vleft = table->maxlen / sizeof(*i);
1948 left = *lenp;
1949
1950 if (!conv)
1951 conv = do_proc_dointvec_conv;
1952
1953 for (; left && vleft--; i++, first=0) {
1954 if (write) {
1955 while (left) {
1956 char c;
1957 if (get_user(c, s))
1958 return -EFAULT;
1959 if (!isspace(c))
1960 break;
1961 left--;
1962 s++;
1963 }
1964 if (!left)
1965 break;
1966 neg = 0;
1967 len = left;
1968 if (len > sizeof(buf) - 1)
1969 len = sizeof(buf) - 1;
1970 if (copy_from_user(buf, s, len))
1971 return -EFAULT;
1972 buf[len] = 0;
1973 p = buf;
1974 if (*p == '-' && left > 1) {
1975 neg = 1;
1976 p++;
1977 }
1978 if (*p < '0' || *p > '9')
1979 break;
1980
1981 lval = simple_strtoul(p, &p, 0);
1982
1983 len = p-buf;
1984 if ((len < left) && *p && !isspace(*p))
1985 break;
1986 if (neg)
1987 val = -val;
1988 s += len;
1989 left -= len;
1990
1991 if (conv(&neg, &lval, i, 1, data))
1992 break;
1993 } else {
1994 p = buf;
1995 if (!first)
1996 *p++ = '\t';
1997
1998 if (conv(&neg, &lval, i, 0, data))
1999 break;
2000
2001 sprintf(p, "%s%lu", neg ? "-" : "", lval);
2002 len = strlen(buf);
2003 if (len > left)
2004 len = left;
2005 if(copy_to_user(s, buf, len))
2006 return -EFAULT;
2007 left -= len;
2008 s += len;
2009 }
2010 }
2011
2012 if (!write && !first && left) {
2013 if(put_user('\n', s))
2014 return -EFAULT;
2015 left--, s++;
2016 }
2017 if (write) {
2018 while (left) {
2019 char c;
2020 if (get_user(c, s++))
2021 return -EFAULT;
2022 if (!isspace(c))
2023 break;
2024 left--;
2025 }
2026 }
2027 if (write && first)
2028 return -EINVAL;
2029 *lenp -= left;
2030 *ppos += *lenp;
2031 return 0;
2032 #undef TMPBUFLEN
2033 }
2034
2035 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2036 void __user *buffer, size_t *lenp, loff_t *ppos,
2037 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2038 int write, void *data),
2039 void *data)
2040 {
2041 return __do_proc_dointvec(table->data, table, write, filp,
2042 buffer, lenp, ppos, conv, data);
2043 }
2044
2045 /**
2046 * proc_dointvec - read a vector of integers
2047 * @table: the sysctl table
2048 * @write: %TRUE if this is a write to the sysctl file
2049 * @filp: the file structure
2050 * @buffer: the user buffer
2051 * @lenp: the size of the user buffer
2052 * @ppos: file position
2053 *
2054 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2055 * values from/to the user buffer, treated as an ASCII string.
2056 *
2057 * Returns 0 on success.
2058 */
2059 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2060 void __user *buffer, size_t *lenp, loff_t *ppos)
2061 {
2062 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2063 NULL,NULL);
2064 }
2065
2066 #define OP_SET 0
2067 #define OP_AND 1
2068 #define OP_OR 2
2069
2070 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
2071 int *valp,
2072 int write, void *data)
2073 {
2074 int op = *(int *)data;
2075 if (write) {
2076 int val = *negp ? -*lvalp : *lvalp;
2077 switch(op) {
2078 case OP_SET: *valp = val; break;
2079 case OP_AND: *valp &= val; break;
2080 case OP_OR: *valp |= val; break;
2081 }
2082 } else {
2083 int val = *valp;
2084 if (val < 0) {
2085 *negp = -1;
2086 *lvalp = (unsigned long)-val;
2087 } else {
2088 *negp = 0;
2089 *lvalp = (unsigned long)val;
2090 }
2091 }
2092 return 0;
2093 }
2094
2095 /*
2096 * Taint values can only be increased
2097 */
2098 static int proc_dointvec_taint(struct ctl_table *table, int write, struct file *filp,
2099 void __user *buffer, size_t *lenp, loff_t *ppos)
2100 {
2101 int op;
2102
2103 if (write && !capable(CAP_SYS_ADMIN))
2104 return -EPERM;
2105
2106 op = OP_OR;
2107 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2108 do_proc_dointvec_bset_conv,&op);
2109 }
2110
2111 struct do_proc_dointvec_minmax_conv_param {
2112 int *min;
2113 int *max;
2114 };
2115
2116 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2117 int *valp,
2118 int write, void *data)
2119 {
2120 struct do_proc_dointvec_minmax_conv_param *param = data;
2121 if (write) {
2122 int val = *negp ? -*lvalp : *lvalp;
2123 if ((param->min && *param->min > val) ||
2124 (param->max && *param->max < val))
2125 return -EINVAL;
2126 *valp = val;
2127 } else {
2128 int val = *valp;
2129 if (val < 0) {
2130 *negp = -1;
2131 *lvalp = (unsigned long)-val;
2132 } else {
2133 *negp = 0;
2134 *lvalp = (unsigned long)val;
2135 }
2136 }
2137 return 0;
2138 }
2139
2140 /**
2141 * proc_dointvec_minmax - read a vector of integers with min/max values
2142 * @table: the sysctl table
2143 * @write: %TRUE if this is a write to the sysctl file
2144 * @filp: the file structure
2145 * @buffer: the user buffer
2146 * @lenp: the size of the user buffer
2147 * @ppos: file position
2148 *
2149 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2150 * values from/to the user buffer, treated as an ASCII string.
2151 *
2152 * This routine will ensure the values are within the range specified by
2153 * table->extra1 (min) and table->extra2 (max).
2154 *
2155 * Returns 0 on success.
2156 */
2157 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2158 void __user *buffer, size_t *lenp, loff_t *ppos)
2159 {
2160 struct do_proc_dointvec_minmax_conv_param param = {
2161 .min = (int *) table->extra1,
2162 .max = (int *) table->extra2,
2163 };
2164 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2165 do_proc_dointvec_minmax_conv, &param);
2166 }
2167
2168 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2169 struct file *filp,
2170 void __user *buffer,
2171 size_t *lenp, loff_t *ppos,
2172 unsigned long convmul,
2173 unsigned long convdiv)
2174 {
2175 #define TMPBUFLEN 21
2176 unsigned long *i, *min, *max, val;
2177 int vleft, first=1, neg;
2178 size_t len, left;
2179 char buf[TMPBUFLEN], *p;
2180 char __user *s = buffer;
2181
2182 if (!data || !table->maxlen || !*lenp ||
2183 (*ppos && !write)) {
2184 *lenp = 0;
2185 return 0;
2186 }
2187
2188 i = (unsigned long *) data;
2189 min = (unsigned long *) table->extra1;
2190 max = (unsigned long *) table->extra2;
2191 vleft = table->maxlen / sizeof(unsigned long);
2192 left = *lenp;
2193
2194 for (; left && vleft--; i++, min++, max++, first=0) {
2195 if (write) {
2196 while (left) {
2197 char c;
2198 if (get_user(c, s))
2199 return -EFAULT;
2200 if (!isspace(c))
2201 break;
2202 left--;
2203 s++;
2204 }
2205 if (!left)
2206 break;
2207 neg = 0;
2208 len = left;
2209 if (len > TMPBUFLEN-1)
2210 len = TMPBUFLEN-1;
2211 if (copy_from_user(buf, s, len))
2212 return -EFAULT;
2213 buf[len] = 0;
2214 p = buf;
2215 if (*p == '-' && left > 1) {
2216 neg = 1;
2217 p++;
2218 }
2219 if (*p < '0' || *p > '9')
2220 break;
2221 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2222 len = p-buf;
2223 if ((len < left) && *p && !isspace(*p))
2224 break;
2225 if (neg)
2226 val = -val;
2227 s += len;
2228 left -= len;
2229
2230 if(neg)
2231 continue;
2232 if ((min && val < *min) || (max && val > *max))
2233 continue;
2234 *i = val;
2235 } else {
2236 p = buf;
2237 if (!first)
2238 *p++ = '\t';
2239 sprintf(p, "%lu", convdiv * (*i) / convmul);
2240 len = strlen(buf);
2241 if (len > left)
2242 len = left;
2243 if(copy_to_user(s, buf, len))
2244 return -EFAULT;
2245 left -= len;
2246 s += len;
2247 }
2248 }
2249
2250 if (!write && !first && left) {
2251 if(put_user('\n', s))
2252 return -EFAULT;
2253 left--, s++;
2254 }
2255 if (write) {
2256 while (left) {
2257 char c;
2258 if (get_user(c, s++))
2259 return -EFAULT;
2260 if (!isspace(c))
2261 break;
2262 left--;
2263 }
2264 }
2265 if (write && first)
2266 return -EINVAL;
2267 *lenp -= left;
2268 *ppos += *lenp;
2269 return 0;
2270 #undef TMPBUFLEN
2271 }
2272
2273 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2274 struct file *filp,
2275 void __user *buffer,
2276 size_t *lenp, loff_t *ppos,
2277 unsigned long convmul,
2278 unsigned long convdiv)
2279 {
2280 return __do_proc_doulongvec_minmax(table->data, table, write,
2281 filp, buffer, lenp, ppos, convmul, convdiv);
2282 }
2283
2284 /**
2285 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2286 * @table: the sysctl table
2287 * @write: %TRUE if this is a write to the sysctl file
2288 * @filp: the file structure
2289 * @buffer: the user buffer
2290 * @lenp: the size of the user buffer
2291 * @ppos: file position
2292 *
2293 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2294 * values from/to the user buffer, treated as an ASCII string.
2295 *
2296 * This routine will ensure the values are within the range specified by
2297 * table->extra1 (min) and table->extra2 (max).
2298 *
2299 * Returns 0 on success.
2300 */
2301 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2302 void __user *buffer, size_t *lenp, loff_t *ppos)
2303 {
2304 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2305 }
2306
2307 /**
2308 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2309 * @table: the sysctl table
2310 * @write: %TRUE if this is a write to the sysctl file
2311 * @filp: the file structure
2312 * @buffer: the user buffer
2313 * @lenp: the size of the user buffer
2314 * @ppos: file position
2315 *
2316 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2317 * values from/to the user buffer, treated as an ASCII string. The values
2318 * are treated as milliseconds, and converted to jiffies when they are stored.
2319 *
2320 * This routine will ensure the values are within the range specified by
2321 * table->extra1 (min) and table->extra2 (max).
2322 *
2323 * Returns 0 on success.
2324 */
2325 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2326 struct file *filp,
2327 void __user *buffer,
2328 size_t *lenp, loff_t *ppos)
2329 {
2330 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2331 lenp, ppos, HZ, 1000l);
2332 }
2333
2334
2335 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2336 int *valp,
2337 int write, void *data)
2338 {
2339 if (write) {
2340 if (*lvalp > LONG_MAX / HZ)
2341 return 1;
2342 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2343 } else {
2344 int val = *valp;
2345 unsigned long lval;
2346 if (val < 0) {
2347 *negp = -1;
2348 lval = (unsigned long)-val;
2349 } else {
2350 *negp = 0;
2351 lval = (unsigned long)val;
2352 }
2353 *lvalp = lval / HZ;
2354 }
2355 return 0;
2356 }
2357
2358 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2359 int *valp,
2360 int write, void *data)
2361 {
2362 if (write) {
2363 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2364 return 1;
2365 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2366 } else {
2367 int val = *valp;
2368 unsigned long lval;
2369 if (val < 0) {
2370 *negp = -1;
2371 lval = (unsigned long)-val;
2372 } else {
2373 *negp = 0;
2374 lval = (unsigned long)val;
2375 }
2376 *lvalp = jiffies_to_clock_t(lval);
2377 }
2378 return 0;
2379 }
2380
2381 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2382 int *valp,
2383 int write, void *data)
2384 {
2385 if (write) {
2386 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2387 } else {
2388 int val = *valp;
2389 unsigned long lval;
2390 if (val < 0) {
2391 *negp = -1;
2392 lval = (unsigned long)-val;
2393 } else {
2394 *negp = 0;
2395 lval = (unsigned long)val;
2396 }
2397 *lvalp = jiffies_to_msecs(lval);
2398 }
2399 return 0;
2400 }
2401
2402 /**
2403 * proc_dointvec_jiffies - read a vector of integers as seconds
2404 * @table: the sysctl table
2405 * @write: %TRUE if this is a write to the sysctl file
2406 * @filp: the file structure
2407 * @buffer: the user buffer
2408 * @lenp: the size of the user buffer
2409 * @ppos: file position
2410 *
2411 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2412 * values from/to the user buffer, treated as an ASCII string.
2413 * The values read are assumed to be in seconds, and are converted into
2414 * jiffies.
2415 *
2416 * Returns 0 on success.
2417 */
2418 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2419 void __user *buffer, size_t *lenp, loff_t *ppos)
2420 {
2421 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2422 do_proc_dointvec_jiffies_conv,NULL);
2423 }
2424
2425 /**
2426 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2427 * @table: the sysctl table
2428 * @write: %TRUE if this is a write to the sysctl file
2429 * @filp: the file structure
2430 * @buffer: the user buffer
2431 * @lenp: the size of the user buffer
2432 * @ppos: pointer to the file position
2433 *
2434 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2435 * values from/to the user buffer, treated as an ASCII string.
2436 * The values read are assumed to be in 1/USER_HZ seconds, and
2437 * are converted into jiffies.
2438 *
2439 * Returns 0 on success.
2440 */
2441 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2442 void __user *buffer, size_t *lenp, loff_t *ppos)
2443 {
2444 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2445 do_proc_dointvec_userhz_jiffies_conv,NULL);
2446 }
2447
2448 /**
2449 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2450 * @table: the sysctl table
2451 * @write: %TRUE if this is a write to the sysctl file
2452 * @filp: the file structure
2453 * @buffer: the user buffer
2454 * @lenp: the size of the user buffer
2455 * @ppos: file position
2456 * @ppos: the current position in the file
2457 *
2458 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2459 * values from/to the user buffer, treated as an ASCII string.
2460 * The values read are assumed to be in 1/1000 seconds, and
2461 * are converted into jiffies.
2462 *
2463 * Returns 0 on success.
2464 */
2465 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2466 void __user *buffer, size_t *lenp, loff_t *ppos)
2467 {
2468 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2469 do_proc_dointvec_ms_jiffies_conv, NULL);
2470 }
2471
2472 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2473 void __user *buffer, size_t *lenp, loff_t *ppos)
2474 {
2475 struct pid *new_pid;
2476 pid_t tmp;
2477 int r;
2478
2479 tmp = pid_vnr(cad_pid);
2480
2481 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2482 lenp, ppos, NULL, NULL);
2483 if (r || !write)
2484 return r;
2485
2486 new_pid = find_get_pid(tmp);
2487 if (!new_pid)
2488 return -ESRCH;
2489
2490 put_pid(xchg(&cad_pid, new_pid));
2491 return 0;
2492 }
2493
2494 #else /* CONFIG_PROC_FS */
2495
2496 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2497 void __user *buffer, size_t *lenp, loff_t *ppos)
2498 {
2499 return -ENOSYS;
2500 }
2501
2502 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2503 void __user *buffer, size_t *lenp, loff_t *ppos)
2504 {
2505 return -ENOSYS;
2506 }
2507
2508 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2509 void __user *buffer, size_t *lenp, loff_t *ppos)
2510 {
2511 return -ENOSYS;
2512 }
2513
2514 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2515 void __user *buffer, size_t *lenp, loff_t *ppos)
2516 {
2517 return -ENOSYS;
2518 }
2519
2520 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2521 void __user *buffer, size_t *lenp, loff_t *ppos)
2522 {
2523 return -ENOSYS;
2524 }
2525
2526 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2527 void __user *buffer, size_t *lenp, loff_t *ppos)
2528 {
2529 return -ENOSYS;
2530 }
2531
2532 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2533 void __user *buffer, size_t *lenp, loff_t *ppos)
2534 {
2535 return -ENOSYS;
2536 }
2537
2538 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2539 struct file *filp,
2540 void __user *buffer,
2541 size_t *lenp, loff_t *ppos)
2542 {
2543 return -ENOSYS;
2544 }
2545
2546
2547 #endif /* CONFIG_PROC_FS */
2548
2549
2550 #ifdef CONFIG_SYSCTL_SYSCALL
2551 /*
2552 * General sysctl support routines
2553 */
2554
2555 /* The generic sysctl data routine (used if no strategy routine supplied) */
2556 int sysctl_data(struct ctl_table *table, int __user *name, int nlen,
2557 void __user *oldval, size_t __user *oldlenp,
2558 void __user *newval, size_t newlen)
2559 {
2560 size_t len;
2561
2562 /* Get out of I don't have a variable */
2563 if (!table->data || !table->maxlen)
2564 return -ENOTDIR;
2565
2566 if (oldval && oldlenp) {
2567 if (get_user(len, oldlenp))
2568 return -EFAULT;
2569 if (len) {
2570 if (len > table->maxlen)
2571 len = table->maxlen;
2572 if (copy_to_user(oldval, table->data, len))
2573 return -EFAULT;
2574 if (put_user(len, oldlenp))
2575 return -EFAULT;
2576 }
2577 }
2578
2579 if (newval && newlen) {
2580 if (newlen > table->maxlen)
2581 newlen = table->maxlen;
2582
2583 if (copy_from_user(table->data, newval, newlen))
2584 return -EFAULT;
2585 }
2586 return 1;
2587 }
2588
2589 /* The generic string strategy routine: */
2590 int sysctl_string(struct ctl_table *table, int __user *name, int nlen,
2591 void __user *oldval, size_t __user *oldlenp,
2592 void __user *newval, size_t newlen)
2593 {
2594 if (!table->data || !table->maxlen)
2595 return -ENOTDIR;
2596
2597 if (oldval && oldlenp) {
2598 size_t bufsize;
2599 if (get_user(bufsize, oldlenp))
2600 return -EFAULT;
2601 if (bufsize) {
2602 size_t len = strlen(table->data), copied;
2603
2604 /* This shouldn't trigger for a well-formed sysctl */
2605 if (len > table->maxlen)
2606 len = table->maxlen;
2607
2608 /* Copy up to a max of bufsize-1 bytes of the string */
2609 copied = (len >= bufsize) ? bufsize - 1 : len;
2610
2611 if (copy_to_user(oldval, table->data, copied) ||
2612 put_user(0, (char __user *)(oldval + copied)))
2613 return -EFAULT;
2614 if (put_user(len, oldlenp))
2615 return -EFAULT;
2616 }
2617 }
2618 if (newval && newlen) {
2619 size_t len = newlen;
2620 if (len > table->maxlen)
2621 len = table->maxlen;
2622 if(copy_from_user(table->data, newval, len))
2623 return -EFAULT;
2624 if (len == table->maxlen)
2625 len--;
2626 ((char *) table->data)[len] = 0;
2627 }
2628 return 1;
2629 }
2630
2631 /*
2632 * This function makes sure that all of the integers in the vector
2633 * are between the minimum and maximum values given in the arrays
2634 * table->extra1 and table->extra2, respectively.
2635 */
2636 int sysctl_intvec(struct ctl_table *table, int __user *name, int nlen,
2637 void __user *oldval, size_t __user *oldlenp,
2638 void __user *newval, size_t newlen)
2639 {
2640
2641 if (newval && newlen) {
2642 int __user *vec = (int __user *) newval;
2643 int *min = (int *) table->extra1;
2644 int *max = (int *) table->extra2;
2645 size_t length;
2646 int i;
2647
2648 if (newlen % sizeof(int) != 0)
2649 return -EINVAL;
2650
2651 if (!table->extra1 && !table->extra2)
2652 return 0;
2653
2654 if (newlen > table->maxlen)
2655 newlen = table->maxlen;
2656 length = newlen / sizeof(int);
2657
2658 for (i = 0; i < length; i++) {
2659 int value;
2660 if (get_user(value, vec + i))
2661 return -EFAULT;
2662 if (min && value < min[i])
2663 return -EINVAL;
2664 if (max && value > max[i])
2665 return -EINVAL;
2666 }
2667 }
2668 return 0;
2669 }
2670
2671 /* Strategy function to convert jiffies to seconds */
2672 int sysctl_jiffies(struct ctl_table *table, int __user *name, int nlen,
2673 void __user *oldval, size_t __user *oldlenp,
2674 void __user *newval, size_t newlen)
2675 {
2676 if (oldval && oldlenp) {
2677 size_t olen;
2678
2679 if (get_user(olen, oldlenp))
2680 return -EFAULT;
2681 if (olen) {
2682 int val;
2683
2684 if (olen < sizeof(int))
2685 return -EINVAL;
2686
2687 val = *(int *)(table->data) / HZ;
2688 if (put_user(val, (int __user *)oldval))
2689 return -EFAULT;
2690 if (put_user(sizeof(int), oldlenp))
2691 return -EFAULT;
2692 }
2693 }
2694 if (newval && newlen) {
2695 int new;
2696 if (newlen != sizeof(int))
2697 return -EINVAL;
2698 if (get_user(new, (int __user *)newval))
2699 return -EFAULT;
2700 *(int *)(table->data) = new*HZ;
2701 }
2702 return 1;
2703 }
2704
2705 /* Strategy function to convert jiffies to seconds */
2706 int sysctl_ms_jiffies(struct ctl_table *table, int __user *name, int nlen,
2707 void __user *oldval, size_t __user *oldlenp,
2708 void __user *newval, size_t newlen)
2709 {
2710 if (oldval && oldlenp) {
2711 size_t olen;
2712
2713 if (get_user(olen, oldlenp))
2714 return -EFAULT;
2715 if (olen) {
2716 int val;
2717
2718 if (olen < sizeof(int))
2719 return -EINVAL;
2720
2721 val = jiffies_to_msecs(*(int *)(table->data));
2722 if (put_user(val, (int __user *)oldval))
2723 return -EFAULT;
2724 if (put_user(sizeof(int), oldlenp))
2725 return -EFAULT;
2726 }
2727 }
2728 if (newval && newlen) {
2729 int new;
2730 if (newlen != sizeof(int))
2731 return -EINVAL;
2732 if (get_user(new, (int __user *)newval))
2733 return -EFAULT;
2734 *(int *)(table->data) = msecs_to_jiffies(new);
2735 }
2736 return 1;
2737 }
2738
2739
2740
2741 #else /* CONFIG_SYSCTL_SYSCALL */
2742
2743
2744 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2745 {
2746 struct __sysctl_args tmp;
2747 int error;
2748
2749 if (copy_from_user(&tmp, args, sizeof(tmp)))
2750 return -EFAULT;
2751
2752 error = deprecated_sysctl_warning(&tmp);
2753
2754 /* If no error reading the parameters then just -ENOSYS ... */
2755 if (!error)
2756 error = -ENOSYS;
2757
2758 return error;
2759 }
2760
2761 int sysctl_data(struct ctl_table *table, int __user *name, int nlen,
2762 void __user *oldval, size_t __user *oldlenp,
2763 void __user *newval, size_t newlen)
2764 {
2765 return -ENOSYS;
2766 }
2767
2768 int sysctl_string(struct ctl_table *table, int __user *name, int nlen,
2769 void __user *oldval, size_t __user *oldlenp,
2770 void __user *newval, size_t newlen)
2771 {
2772 return -ENOSYS;
2773 }
2774
2775 int sysctl_intvec(struct ctl_table *table, int __user *name, int nlen,
2776 void __user *oldval, size_t __user *oldlenp,
2777 void __user *newval, size_t newlen)
2778 {
2779 return -ENOSYS;
2780 }
2781
2782 int sysctl_jiffies(struct ctl_table *table, int __user *name, int nlen,
2783 void __user *oldval, size_t __user *oldlenp,
2784 void __user *newval, size_t newlen)
2785 {
2786 return -ENOSYS;
2787 }
2788
2789 int sysctl_ms_jiffies(struct ctl_table *table, int __user *name, int nlen,
2790 void __user *oldval, size_t __user *oldlenp,
2791 void __user *newval, size_t newlen)
2792 {
2793 return -ENOSYS;
2794 }
2795
2796 #endif /* CONFIG_SYSCTL_SYSCALL */
2797
2798 static int deprecated_sysctl_warning(struct __sysctl_args *args)
2799 {
2800 static int msg_count;
2801 int name[CTL_MAXNAME];
2802 int i;
2803
2804 /* Check args->nlen. */
2805 if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
2806 return -ENOTDIR;
2807
2808 /* Read in the sysctl name for better debug message logging */
2809 for (i = 0; i < args->nlen; i++)
2810 if (get_user(name[i], args->name + i))
2811 return -EFAULT;
2812
2813 /* Ignore accesses to kernel.version */
2814 if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2815 return 0;
2816
2817 if (msg_count < 5) {
2818 msg_count++;
2819 printk(KERN_INFO
2820 "warning: process `%s' used the deprecated sysctl "
2821 "system call with ", current->comm);
2822 for (i = 0; i < args->nlen; i++)
2823 printk("%d.", name[i]);
2824 printk("\n");
2825 }
2826 return 0;
2827 }
2828
2829 /*
2830 * No sense putting this after each symbol definition, twice,
2831 * exception granted :-)
2832 */
2833 EXPORT_SYMBOL(proc_dointvec);
2834 EXPORT_SYMBOL(proc_dointvec_jiffies);
2835 EXPORT_SYMBOL(proc_dointvec_minmax);
2836 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2837 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2838 EXPORT_SYMBOL(proc_dostring);
2839 EXPORT_SYMBOL(proc_doulongvec_minmax);
2840 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2841 EXPORT_SYMBOL(register_sysctl_table);
2842 EXPORT_SYMBOL(register_sysctl_paths);
2843 EXPORT_SYMBOL(sysctl_intvec);
2844 EXPORT_SYMBOL(sysctl_jiffies);
2845 EXPORT_SYMBOL(sysctl_ms_jiffies);
2846 EXPORT_SYMBOL(sysctl_string);
2847 EXPORT_SYMBOL(sysctl_data);
2848 EXPORT_SYMBOL(unregister_sysctl_table);
Log-structured file system for Linux