]> git.proxmox.com Git - mirror_ubuntu-zesty-kernel.git/blob - kernel/sysctl.c
5418ef61e16ee32dbea099c52318093beebce682
[mirror_ubuntu-zesty-kernel.git] / kernel / sysctl.c
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/security.h>
41 #include <linux/initrd.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 #endif
57
58 static int deprecated_sysctl_warning(struct __sysctl_args *args);
59
60 #if defined(CONFIG_SYSCTL)
61
62 /* External variables not in a header file. */
63 extern int C_A_D;
64 extern int print_fatal_signals;
65 extern int sysctl_overcommit_memory;
66 extern int sysctl_overcommit_ratio;
67 extern int sysctl_panic_on_oom;
68 extern int sysctl_oom_kill_allocating_task;
69 extern int max_threads;
70 extern int core_uses_pid;
71 extern int suid_dumpable;
72 extern char core_pattern[];
73 extern int pid_max;
74 extern int min_free_kbytes;
75 extern int printk_ratelimit_jiffies;
76 extern int printk_ratelimit_burst;
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 audit_argv_kb;
84 extern int latencytop_enabled;
85
86 /* Constants used for minimum and maximum */
87 #ifdef CONFIG_DETECT_SOFTLOCKUP
88 static int one = 1;
89 static int sixty = 60;
90 #endif
91
92 #ifdef CONFIG_MMU
93 static int two = 2;
94 #endif
95
96 static int zero;
97 static int one_hundred = 100;
98
99 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
100 static int maxolduid = 65535;
101 static int minolduid;
102 static int min_percpu_pagelist_fract = 8;
103
104 static int ngroups_max = NGROUPS_MAX;
105
106 #ifdef CONFIG_KMOD
107 extern char modprobe_path[];
108 #endif
109 #ifdef CONFIG_CHR_DEV_SG
110 extern int sg_big_buff;
111 #endif
112
113 #ifdef __sparc__
114 extern char reboot_command [];
115 extern int stop_a_enabled;
116 extern int scons_pwroff;
117 #endif
118
119 #ifdef __hppa__
120 extern int pwrsw_enabled;
121 extern int unaligned_enabled;
122 #endif
123
124 #ifdef CONFIG_S390
125 #ifdef CONFIG_MATHEMU
126 extern int sysctl_ieee_emulation_warnings;
127 #endif
128 extern int sysctl_userprocess_debug;
129 extern int spin_retry;
130 #endif
131
132 extern int sysctl_hz_timer;
133
134 #ifdef CONFIG_BSD_PROCESS_ACCT
135 extern int acct_parm[];
136 #endif
137
138 #ifdef CONFIG_IA64
139 extern int no_unaligned_warning;
140 #endif
141
142 #ifdef CONFIG_RT_MUTEXES
143 extern int max_lock_depth;
144 #endif
145
146 #ifdef CONFIG_SYSCTL_SYSCALL
147 static int parse_table(int __user *, int, void __user *, size_t __user *,
148 void __user *, size_t, struct ctl_table *);
149 #endif
150
151
152 #ifdef CONFIG_PROC_SYSCTL
153 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
154 void __user *buffer, size_t *lenp, loff_t *ppos);
155 static int proc_dointvec_taint(struct ctl_table *table, int write, struct file *filp,
156 void __user *buffer, size_t *lenp, loff_t *ppos);
157 #endif
158
159 static struct ctl_table root_table[];
160 static struct ctl_table_header root_table_header =
161 { root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
162
163 static struct ctl_table kern_table[];
164 static struct ctl_table vm_table[];
165 static struct ctl_table fs_table[];
166 static struct ctl_table debug_table[];
167 static struct ctl_table dev_table[];
168 extern struct ctl_table random_table[];
169 #ifdef CONFIG_INOTIFY_USER
170 extern struct ctl_table inotify_table[];
171 #endif
172
173 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
174 int sysctl_legacy_va_layout;
175 #endif
176
177 extern int prove_locking;
178 extern int lock_stat;
179
180 /* The default sysctl tables: */
181
182 static struct ctl_table root_table[] = {
183 {
184 .ctl_name = CTL_KERN,
185 .procname = "kernel",
186 .mode = 0555,
187 .child = kern_table,
188 },
189 {
190 .ctl_name = CTL_VM,
191 .procname = "vm",
192 .mode = 0555,
193 .child = vm_table,
194 },
195 #ifdef CONFIG_NET
196 {
197 .ctl_name = CTL_NET,
198 .procname = "net",
199 .mode = 0555,
200 .child = net_table,
201 },
202 #endif
203 {
204 .ctl_name = CTL_FS,
205 .procname = "fs",
206 .mode = 0555,
207 .child = fs_table,
208 },
209 {
210 .ctl_name = CTL_DEBUG,
211 .procname = "debug",
212 .mode = 0555,
213 .child = debug_table,
214 },
215 {
216 .ctl_name = CTL_DEV,
217 .procname = "dev",
218 .mode = 0555,
219 .child = dev_table,
220 },
221 /*
222 * NOTE: do not add new entries to this table unless you have read
223 * Documentation/sysctl/ctl_unnumbered.txt
224 */
225 { .ctl_name = 0 }
226 };
227
228 #ifdef CONFIG_SCHED_DEBUG
229 static int min_sched_granularity_ns = 100000; /* 100 usecs */
230 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
231 static int min_wakeup_granularity_ns; /* 0 usecs */
232 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
233 #endif
234
235 static struct ctl_table kern_table[] = {
236 #ifdef CONFIG_SCHED_DEBUG
237 {
238 .ctl_name = CTL_UNNUMBERED,
239 .procname = "sched_min_granularity_ns",
240 .data = &sysctl_sched_min_granularity,
241 .maxlen = sizeof(unsigned int),
242 .mode = 0644,
243 .proc_handler = &sched_nr_latency_handler,
244 .strategy = &sysctl_intvec,
245 .extra1 = &min_sched_granularity_ns,
246 .extra2 = &max_sched_granularity_ns,
247 },
248 {
249 .ctl_name = CTL_UNNUMBERED,
250 .procname = "sched_latency_ns",
251 .data = &sysctl_sched_latency,
252 .maxlen = sizeof(unsigned int),
253 .mode = 0644,
254 .proc_handler = &sched_nr_latency_handler,
255 .strategy = &sysctl_intvec,
256 .extra1 = &min_sched_granularity_ns,
257 .extra2 = &max_sched_granularity_ns,
258 },
259 {
260 .ctl_name = CTL_UNNUMBERED,
261 .procname = "sched_wakeup_granularity_ns",
262 .data = &sysctl_sched_wakeup_granularity,
263 .maxlen = sizeof(unsigned int),
264 .mode = 0644,
265 .proc_handler = &proc_dointvec_minmax,
266 .strategy = &sysctl_intvec,
267 .extra1 = &min_wakeup_granularity_ns,
268 .extra2 = &max_wakeup_granularity_ns,
269 },
270 {
271 .ctl_name = CTL_UNNUMBERED,
272 .procname = "sched_batch_wakeup_granularity_ns",
273 .data = &sysctl_sched_batch_wakeup_granularity,
274 .maxlen = sizeof(unsigned int),
275 .mode = 0644,
276 .proc_handler = &proc_dointvec_minmax,
277 .strategy = &sysctl_intvec,
278 .extra1 = &min_wakeup_granularity_ns,
279 .extra2 = &max_wakeup_granularity_ns,
280 },
281 {
282 .ctl_name = CTL_UNNUMBERED,
283 .procname = "sched_child_runs_first",
284 .data = &sysctl_sched_child_runs_first,
285 .maxlen = sizeof(unsigned int),
286 .mode = 0644,
287 .proc_handler = &proc_dointvec,
288 },
289 {
290 .ctl_name = CTL_UNNUMBERED,
291 .procname = "sched_features",
292 .data = &sysctl_sched_features,
293 .maxlen = sizeof(unsigned int),
294 .mode = 0644,
295 .proc_handler = &proc_dointvec,
296 },
297 {
298 .ctl_name = CTL_UNNUMBERED,
299 .procname = "sched_migration_cost",
300 .data = &sysctl_sched_migration_cost,
301 .maxlen = sizeof(unsigned int),
302 .mode = 0644,
303 .proc_handler = &proc_dointvec,
304 },
305 {
306 .ctl_name = CTL_UNNUMBERED,
307 .procname = "sched_nr_migrate",
308 .data = &sysctl_sched_nr_migrate,
309 .maxlen = sizeof(unsigned int),
310 .mode = 0644,
311 .proc_handler = &proc_dointvec,
312 },
313 {
314 .ctl_name = CTL_UNNUMBERED,
315 .procname = "sched_rt_period_ms",
316 .data = &sysctl_sched_rt_period,
317 .maxlen = sizeof(unsigned int),
318 .mode = 0644,
319 .proc_handler = &proc_dointvec,
320 },
321 {
322 .ctl_name = CTL_UNNUMBERED,
323 .procname = "sched_rt_ratio",
324 .data = &sysctl_sched_rt_ratio,
325 .maxlen = sizeof(unsigned int),
326 .mode = 0644,
327 .proc_handler = &proc_dointvec,
328 },
329 #if defined(CONFIG_FAIR_GROUP_SCHED) && defined(CONFIG_SMP)
330 {
331 .ctl_name = CTL_UNNUMBERED,
332 .procname = "sched_min_bal_int_shares",
333 .data = &sysctl_sched_min_bal_int_shares,
334 .maxlen = sizeof(unsigned int),
335 .mode = 0644,
336 .proc_handler = &proc_dointvec,
337 },
338 {
339 .ctl_name = CTL_UNNUMBERED,
340 .procname = "sched_max_bal_int_shares",
341 .data = &sysctl_sched_max_bal_int_shares,
342 .maxlen = sizeof(unsigned int),
343 .mode = 0644,
344 .proc_handler = &proc_dointvec,
345 },
346 #endif
347 #endif
348 {
349 .ctl_name = CTL_UNNUMBERED,
350 .procname = "sched_compat_yield",
351 .data = &sysctl_sched_compat_yield,
352 .maxlen = sizeof(unsigned int),
353 .mode = 0644,
354 .proc_handler = &proc_dointvec,
355 },
356 #ifdef CONFIG_PROVE_LOCKING
357 {
358 .ctl_name = CTL_UNNUMBERED,
359 .procname = "prove_locking",
360 .data = &prove_locking,
361 .maxlen = sizeof(int),
362 .mode = 0644,
363 .proc_handler = &proc_dointvec,
364 },
365 #endif
366 #ifdef CONFIG_LOCK_STAT
367 {
368 .ctl_name = CTL_UNNUMBERED,
369 .procname = "lock_stat",
370 .data = &lock_stat,
371 .maxlen = sizeof(int),
372 .mode = 0644,
373 .proc_handler = &proc_dointvec,
374 },
375 #endif
376 {
377 .ctl_name = KERN_PANIC,
378 .procname = "panic",
379 .data = &panic_timeout,
380 .maxlen = sizeof(int),
381 .mode = 0644,
382 .proc_handler = &proc_dointvec,
383 },
384 {
385 .ctl_name = KERN_CORE_USES_PID,
386 .procname = "core_uses_pid",
387 .data = &core_uses_pid,
388 .maxlen = sizeof(int),
389 .mode = 0644,
390 .proc_handler = &proc_dointvec,
391 },
392 #ifdef CONFIG_AUDITSYSCALL
393 {
394 .ctl_name = CTL_UNNUMBERED,
395 .procname = "audit_argv_kb",
396 .data = &audit_argv_kb,
397 .maxlen = sizeof(int),
398 .mode = 0644,
399 .proc_handler = &proc_dointvec,
400 },
401 #endif
402 {
403 .ctl_name = KERN_CORE_PATTERN,
404 .procname = "core_pattern",
405 .data = core_pattern,
406 .maxlen = CORENAME_MAX_SIZE,
407 .mode = 0644,
408 .proc_handler = &proc_dostring,
409 .strategy = &sysctl_string,
410 },
411 #ifdef CONFIG_PROC_SYSCTL
412 {
413 .procname = "tainted",
414 .data = &tainted,
415 .maxlen = sizeof(int),
416 .mode = 0644,
417 .proc_handler = &proc_dointvec_taint,
418 },
419 #endif
420 #ifdef CONFIG_LATENCYTOP
421 {
422 .procname = "latencytop",
423 .data = &latencytop_enabled,
424 .maxlen = sizeof(int),
425 .mode = 0644,
426 .proc_handler = &proc_dointvec,
427 },
428 #endif
429 #ifdef CONFIG_SECURITY_CAPABILITIES
430 {
431 .procname = "cap-bound",
432 .data = &cap_bset,
433 .maxlen = sizeof(kernel_cap_t),
434 .mode = 0600,
435 .proc_handler = &proc_dointvec_bset,
436 },
437 #endif /* def CONFIG_SECURITY_CAPABILITIES */
438 #ifdef CONFIG_BLK_DEV_INITRD
439 {
440 .ctl_name = KERN_REALROOTDEV,
441 .procname = "real-root-dev",
442 .data = &real_root_dev,
443 .maxlen = sizeof(int),
444 .mode = 0644,
445 .proc_handler = &proc_dointvec,
446 },
447 #endif
448 {
449 .ctl_name = CTL_UNNUMBERED,
450 .procname = "print-fatal-signals",
451 .data = &print_fatal_signals,
452 .maxlen = sizeof(int),
453 .mode = 0644,
454 .proc_handler = &proc_dointvec,
455 },
456 #ifdef __sparc__
457 {
458 .ctl_name = KERN_SPARC_REBOOT,
459 .procname = "reboot-cmd",
460 .data = reboot_command,
461 .maxlen = 256,
462 .mode = 0644,
463 .proc_handler = &proc_dostring,
464 .strategy = &sysctl_string,
465 },
466 {
467 .ctl_name = KERN_SPARC_STOP_A,
468 .procname = "stop-a",
469 .data = &stop_a_enabled,
470 .maxlen = sizeof (int),
471 .mode = 0644,
472 .proc_handler = &proc_dointvec,
473 },
474 {
475 .ctl_name = KERN_SPARC_SCONS_PWROFF,
476 .procname = "scons-poweroff",
477 .data = &scons_pwroff,
478 .maxlen = sizeof (int),
479 .mode = 0644,
480 .proc_handler = &proc_dointvec,
481 },
482 #endif
483 #ifdef __hppa__
484 {
485 .ctl_name = KERN_HPPA_PWRSW,
486 .procname = "soft-power",
487 .data = &pwrsw_enabled,
488 .maxlen = sizeof (int),
489 .mode = 0644,
490 .proc_handler = &proc_dointvec,
491 },
492 {
493 .ctl_name = KERN_HPPA_UNALIGNED,
494 .procname = "unaligned-trap",
495 .data = &unaligned_enabled,
496 .maxlen = sizeof (int),
497 .mode = 0644,
498 .proc_handler = &proc_dointvec,
499 },
500 #endif
501 {
502 .ctl_name = KERN_CTLALTDEL,
503 .procname = "ctrl-alt-del",
504 .data = &C_A_D,
505 .maxlen = sizeof(int),
506 .mode = 0644,
507 .proc_handler = &proc_dointvec,
508 },
509 {
510 .ctl_name = KERN_PRINTK,
511 .procname = "printk",
512 .data = &console_loglevel,
513 .maxlen = 4*sizeof(int),
514 .mode = 0644,
515 .proc_handler = &proc_dointvec,
516 },
517 #ifdef CONFIG_KMOD
518 {
519 .ctl_name = KERN_MODPROBE,
520 .procname = "modprobe",
521 .data = &modprobe_path,
522 .maxlen = KMOD_PATH_LEN,
523 .mode = 0644,
524 .proc_handler = &proc_dostring,
525 .strategy = &sysctl_string,
526 },
527 #endif
528 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
529 {
530 .ctl_name = KERN_HOTPLUG,
531 .procname = "hotplug",
532 .data = &uevent_helper,
533 .maxlen = UEVENT_HELPER_PATH_LEN,
534 .mode = 0644,
535 .proc_handler = &proc_dostring,
536 .strategy = &sysctl_string,
537 },
538 #endif
539 #ifdef CONFIG_CHR_DEV_SG
540 {
541 .ctl_name = KERN_SG_BIG_BUFF,
542 .procname = "sg-big-buff",
543 .data = &sg_big_buff,
544 .maxlen = sizeof (int),
545 .mode = 0444,
546 .proc_handler = &proc_dointvec,
547 },
548 #endif
549 #ifdef CONFIG_BSD_PROCESS_ACCT
550 {
551 .ctl_name = KERN_ACCT,
552 .procname = "acct",
553 .data = &acct_parm,
554 .maxlen = 3*sizeof(int),
555 .mode = 0644,
556 .proc_handler = &proc_dointvec,
557 },
558 #endif
559 #ifdef CONFIG_MAGIC_SYSRQ
560 {
561 .ctl_name = KERN_SYSRQ,
562 .procname = "sysrq",
563 .data = &__sysrq_enabled,
564 .maxlen = sizeof (int),
565 .mode = 0644,
566 .proc_handler = &proc_dointvec,
567 },
568 #endif
569 #ifdef CONFIG_PROC_SYSCTL
570 {
571 .procname = "cad_pid",
572 .data = NULL,
573 .maxlen = sizeof (int),
574 .mode = 0600,
575 .proc_handler = &proc_do_cad_pid,
576 },
577 #endif
578 {
579 .ctl_name = KERN_MAX_THREADS,
580 .procname = "threads-max",
581 .data = &max_threads,
582 .maxlen = sizeof(int),
583 .mode = 0644,
584 .proc_handler = &proc_dointvec,
585 },
586 {
587 .ctl_name = KERN_RANDOM,
588 .procname = "random",
589 .mode = 0555,
590 .child = random_table,
591 },
592 {
593 .ctl_name = KERN_OVERFLOWUID,
594 .procname = "overflowuid",
595 .data = &overflowuid,
596 .maxlen = sizeof(int),
597 .mode = 0644,
598 .proc_handler = &proc_dointvec_minmax,
599 .strategy = &sysctl_intvec,
600 .extra1 = &minolduid,
601 .extra2 = &maxolduid,
602 },
603 {
604 .ctl_name = KERN_OVERFLOWGID,
605 .procname = "overflowgid",
606 .data = &overflowgid,
607 .maxlen = sizeof(int),
608 .mode = 0644,
609 .proc_handler = &proc_dointvec_minmax,
610 .strategy = &sysctl_intvec,
611 .extra1 = &minolduid,
612 .extra2 = &maxolduid,
613 },
614 #ifdef CONFIG_S390
615 #ifdef CONFIG_MATHEMU
616 {
617 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
618 .procname = "ieee_emulation_warnings",
619 .data = &sysctl_ieee_emulation_warnings,
620 .maxlen = sizeof(int),
621 .mode = 0644,
622 .proc_handler = &proc_dointvec,
623 },
624 #endif
625 #ifdef CONFIG_NO_IDLE_HZ
626 {
627 .ctl_name = KERN_HZ_TIMER,
628 .procname = "hz_timer",
629 .data = &sysctl_hz_timer,
630 .maxlen = sizeof(int),
631 .mode = 0644,
632 .proc_handler = &proc_dointvec,
633 },
634 #endif
635 {
636 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
637 .procname = "userprocess_debug",
638 .data = &sysctl_userprocess_debug,
639 .maxlen = sizeof(int),
640 .mode = 0644,
641 .proc_handler = &proc_dointvec,
642 },
643 #endif
644 {
645 .ctl_name = KERN_PIDMAX,
646 .procname = "pid_max",
647 .data = &pid_max,
648 .maxlen = sizeof (int),
649 .mode = 0644,
650 .proc_handler = &proc_dointvec_minmax,
651 .strategy = sysctl_intvec,
652 .extra1 = &pid_max_min,
653 .extra2 = &pid_max_max,
654 },
655 {
656 .ctl_name = KERN_PANIC_ON_OOPS,
657 .procname = "panic_on_oops",
658 .data = &panic_on_oops,
659 .maxlen = sizeof(int),
660 .mode = 0644,
661 .proc_handler = &proc_dointvec,
662 },
663 {
664 .ctl_name = KERN_PRINTK_RATELIMIT,
665 .procname = "printk_ratelimit",
666 .data = &printk_ratelimit_jiffies,
667 .maxlen = sizeof(int),
668 .mode = 0644,
669 .proc_handler = &proc_dointvec_jiffies,
670 .strategy = &sysctl_jiffies,
671 },
672 {
673 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
674 .procname = "printk_ratelimit_burst",
675 .data = &printk_ratelimit_burst,
676 .maxlen = sizeof(int),
677 .mode = 0644,
678 .proc_handler = &proc_dointvec,
679 },
680 {
681 .ctl_name = KERN_NGROUPS_MAX,
682 .procname = "ngroups_max",
683 .data = &ngroups_max,
684 .maxlen = sizeof (int),
685 .mode = 0444,
686 .proc_handler = &proc_dointvec,
687 },
688 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
689 {
690 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
691 .procname = "unknown_nmi_panic",
692 .data = &unknown_nmi_panic,
693 .maxlen = sizeof (int),
694 .mode = 0644,
695 .proc_handler = &proc_dointvec,
696 },
697 {
698 .procname = "nmi_watchdog",
699 .data = &nmi_watchdog_enabled,
700 .maxlen = sizeof (int),
701 .mode = 0644,
702 .proc_handler = &proc_nmi_enabled,
703 },
704 #endif
705 #if defined(CONFIG_X86)
706 {
707 .ctl_name = KERN_PANIC_ON_NMI,
708 .procname = "panic_on_unrecovered_nmi",
709 .data = &panic_on_unrecovered_nmi,
710 .maxlen = sizeof(int),
711 .mode = 0644,
712 .proc_handler = &proc_dointvec,
713 },
714 {
715 .ctl_name = KERN_BOOTLOADER_TYPE,
716 .procname = "bootloader_type",
717 .data = &bootloader_type,
718 .maxlen = sizeof (int),
719 .mode = 0444,
720 .proc_handler = &proc_dointvec,
721 },
722 {
723 .ctl_name = CTL_UNNUMBERED,
724 .procname = "kstack_depth_to_print",
725 .data = &kstack_depth_to_print,
726 .maxlen = sizeof(int),
727 .mode = 0644,
728 .proc_handler = &proc_dointvec,
729 },
730 #endif
731 #if defined(CONFIG_MMU)
732 {
733 .ctl_name = KERN_RANDOMIZE,
734 .procname = "randomize_va_space",
735 .data = &randomize_va_space,
736 .maxlen = sizeof(int),
737 .mode = 0644,
738 .proc_handler = &proc_dointvec,
739 },
740 #endif
741 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
742 {
743 .ctl_name = KERN_SPIN_RETRY,
744 .procname = "spin_retry",
745 .data = &spin_retry,
746 .maxlen = sizeof (int),
747 .mode = 0644,
748 .proc_handler = &proc_dointvec,
749 },
750 #endif
751 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
752 {
753 .procname = "acpi_video_flags",
754 .data = &acpi_realmode_flags,
755 .maxlen = sizeof (unsigned long),
756 .mode = 0644,
757 .proc_handler = &proc_doulongvec_minmax,
758 },
759 #endif
760 #ifdef CONFIG_IA64
761 {
762 .ctl_name = KERN_IA64_UNALIGNED,
763 .procname = "ignore-unaligned-usertrap",
764 .data = &no_unaligned_warning,
765 .maxlen = sizeof (int),
766 .mode = 0644,
767 .proc_handler = &proc_dointvec,
768 },
769 #endif
770 #ifdef CONFIG_DETECT_SOFTLOCKUP
771 {
772 .ctl_name = CTL_UNNUMBERED,
773 .procname = "softlockup_thresh",
774 .data = &softlockup_thresh,
775 .maxlen = sizeof(int),
776 .mode = 0644,
777 .proc_handler = &proc_dointvec_minmax,
778 .strategy = &sysctl_intvec,
779 .extra1 = &one,
780 .extra2 = &sixty,
781 },
782 {
783 .ctl_name = CTL_UNNUMBERED,
784 .procname = "hung_task_check_count",
785 .data = &sysctl_hung_task_check_count,
786 .maxlen = sizeof(int),
787 .mode = 0644,
788 .proc_handler = &proc_dointvec_minmax,
789 .strategy = &sysctl_intvec,
790 },
791 {
792 .ctl_name = CTL_UNNUMBERED,
793 .procname = "hung_task_timeout_secs",
794 .data = &sysctl_hung_task_timeout_secs,
795 .maxlen = sizeof(int),
796 .mode = 0644,
797 .proc_handler = &proc_dointvec_minmax,
798 .strategy = &sysctl_intvec,
799 },
800 {
801 .ctl_name = CTL_UNNUMBERED,
802 .procname = "hung_task_warnings",
803 .data = &sysctl_hung_task_warnings,
804 .maxlen = sizeof(int),
805 .mode = 0644,
806 .proc_handler = &proc_dointvec_minmax,
807 .strategy = &sysctl_intvec,
808 },
809 #endif
810 #ifdef CONFIG_COMPAT
811 {
812 .ctl_name = KERN_COMPAT_LOG,
813 .procname = "compat-log",
814 .data = &compat_log,
815 .maxlen = sizeof (int),
816 .mode = 0644,
817 .proc_handler = &proc_dointvec,
818 },
819 #endif
820 #ifdef CONFIG_RT_MUTEXES
821 {
822 .ctl_name = KERN_MAX_LOCK_DEPTH,
823 .procname = "max_lock_depth",
824 .data = &max_lock_depth,
825 .maxlen = sizeof(int),
826 .mode = 0644,
827 .proc_handler = &proc_dointvec,
828 },
829 #endif
830 #ifdef CONFIG_PROC_FS
831 {
832 .ctl_name = CTL_UNNUMBERED,
833 .procname = "maps_protect",
834 .data = &maps_protect,
835 .maxlen = sizeof(int),
836 .mode = 0644,
837 .proc_handler = &proc_dointvec,
838 },
839 #endif
840 {
841 .ctl_name = CTL_UNNUMBERED,
842 .procname = "poweroff_cmd",
843 .data = &poweroff_cmd,
844 .maxlen = POWEROFF_CMD_PATH_LEN,
845 .mode = 0644,
846 .proc_handler = &proc_dostring,
847 .strategy = &sysctl_string,
848 },
849 /*
850 * NOTE: do not add new entries to this table unless you have read
851 * Documentation/sysctl/ctl_unnumbered.txt
852 */
853 { .ctl_name = 0 }
854 };
855
856 static struct ctl_table vm_table[] = {
857 {
858 .ctl_name = VM_OVERCOMMIT_MEMORY,
859 .procname = "overcommit_memory",
860 .data = &sysctl_overcommit_memory,
861 .maxlen = sizeof(sysctl_overcommit_memory),
862 .mode = 0644,
863 .proc_handler = &proc_dointvec,
864 },
865 {
866 .ctl_name = VM_PANIC_ON_OOM,
867 .procname = "panic_on_oom",
868 .data = &sysctl_panic_on_oom,
869 .maxlen = sizeof(sysctl_panic_on_oom),
870 .mode = 0644,
871 .proc_handler = &proc_dointvec,
872 },
873 {
874 .ctl_name = CTL_UNNUMBERED,
875 .procname = "oom_kill_allocating_task",
876 .data = &sysctl_oom_kill_allocating_task,
877 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
878 .mode = 0644,
879 .proc_handler = &proc_dointvec,
880 },
881 {
882 .ctl_name = VM_OVERCOMMIT_RATIO,
883 .procname = "overcommit_ratio",
884 .data = &sysctl_overcommit_ratio,
885 .maxlen = sizeof(sysctl_overcommit_ratio),
886 .mode = 0644,
887 .proc_handler = &proc_dointvec,
888 },
889 {
890 .ctl_name = VM_PAGE_CLUSTER,
891 .procname = "page-cluster",
892 .data = &page_cluster,
893 .maxlen = sizeof(int),
894 .mode = 0644,
895 .proc_handler = &proc_dointvec,
896 },
897 {
898 .ctl_name = VM_DIRTY_BACKGROUND,
899 .procname = "dirty_background_ratio",
900 .data = &dirty_background_ratio,
901 .maxlen = sizeof(dirty_background_ratio),
902 .mode = 0644,
903 .proc_handler = &proc_dointvec_minmax,
904 .strategy = &sysctl_intvec,
905 .extra1 = &zero,
906 .extra2 = &one_hundred,
907 },
908 {
909 .ctl_name = VM_DIRTY_RATIO,
910 .procname = "dirty_ratio",
911 .data = &vm_dirty_ratio,
912 .maxlen = sizeof(vm_dirty_ratio),
913 .mode = 0644,
914 .proc_handler = &dirty_ratio_handler,
915 .strategy = &sysctl_intvec,
916 .extra1 = &zero,
917 .extra2 = &one_hundred,
918 },
919 {
920 .procname = "dirty_writeback_centisecs",
921 .data = &dirty_writeback_interval,
922 .maxlen = sizeof(dirty_writeback_interval),
923 .mode = 0644,
924 .proc_handler = &dirty_writeback_centisecs_handler,
925 },
926 {
927 .procname = "dirty_expire_centisecs",
928 .data = &dirty_expire_interval,
929 .maxlen = sizeof(dirty_expire_interval),
930 .mode = 0644,
931 .proc_handler = &proc_dointvec_userhz_jiffies,
932 },
933 {
934 .ctl_name = VM_NR_PDFLUSH_THREADS,
935 .procname = "nr_pdflush_threads",
936 .data = &nr_pdflush_threads,
937 .maxlen = sizeof nr_pdflush_threads,
938 .mode = 0444 /* read-only*/,
939 .proc_handler = &proc_dointvec,
940 },
941 {
942 .ctl_name = VM_SWAPPINESS,
943 .procname = "swappiness",
944 .data = &vm_swappiness,
945 .maxlen = sizeof(vm_swappiness),
946 .mode = 0644,
947 .proc_handler = &proc_dointvec_minmax,
948 .strategy = &sysctl_intvec,
949 .extra1 = &zero,
950 .extra2 = &one_hundred,
951 },
952 #ifdef CONFIG_HUGETLB_PAGE
953 {
954 .procname = "nr_hugepages",
955 .data = &max_huge_pages,
956 .maxlen = sizeof(unsigned long),
957 .mode = 0644,
958 .proc_handler = &hugetlb_sysctl_handler,
959 .extra1 = (void *)&hugetlb_zero,
960 .extra2 = (void *)&hugetlb_infinity,
961 },
962 {
963 .ctl_name = VM_HUGETLB_GROUP,
964 .procname = "hugetlb_shm_group",
965 .data = &sysctl_hugetlb_shm_group,
966 .maxlen = sizeof(gid_t),
967 .mode = 0644,
968 .proc_handler = &proc_dointvec,
969 },
970 {
971 .ctl_name = CTL_UNNUMBERED,
972 .procname = "hugepages_treat_as_movable",
973 .data = &hugepages_treat_as_movable,
974 .maxlen = sizeof(int),
975 .mode = 0644,
976 .proc_handler = &hugetlb_treat_movable_handler,
977 },
978 {
979 .ctl_name = CTL_UNNUMBERED,
980 .procname = "nr_overcommit_hugepages",
981 .data = &nr_overcommit_huge_pages,
982 .maxlen = sizeof(nr_overcommit_huge_pages),
983 .mode = 0644,
984 .proc_handler = &proc_doulongvec_minmax,
985 },
986 #endif
987 {
988 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
989 .procname = "lowmem_reserve_ratio",
990 .data = &sysctl_lowmem_reserve_ratio,
991 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
992 .mode = 0644,
993 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
994 .strategy = &sysctl_intvec,
995 },
996 {
997 .ctl_name = VM_DROP_PAGECACHE,
998 .procname = "drop_caches",
999 .data = &sysctl_drop_caches,
1000 .maxlen = sizeof(int),
1001 .mode = 0644,
1002 .proc_handler = drop_caches_sysctl_handler,
1003 .strategy = &sysctl_intvec,
1004 },
1005 {
1006 .ctl_name = VM_MIN_FREE_KBYTES,
1007 .procname = "min_free_kbytes",
1008 .data = &min_free_kbytes,
1009 .maxlen = sizeof(min_free_kbytes),
1010 .mode = 0644,
1011 .proc_handler = &min_free_kbytes_sysctl_handler,
1012 .strategy = &sysctl_intvec,
1013 .extra1 = &zero,
1014 },
1015 {
1016 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
1017 .procname = "percpu_pagelist_fraction",
1018 .data = &percpu_pagelist_fraction,
1019 .maxlen = sizeof(percpu_pagelist_fraction),
1020 .mode = 0644,
1021 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
1022 .strategy = &sysctl_intvec,
1023 .extra1 = &min_percpu_pagelist_fract,
1024 },
1025 #ifdef CONFIG_MMU
1026 {
1027 .ctl_name = VM_MAX_MAP_COUNT,
1028 .procname = "max_map_count",
1029 .data = &sysctl_max_map_count,
1030 .maxlen = sizeof(sysctl_max_map_count),
1031 .mode = 0644,
1032 .proc_handler = &proc_dointvec
1033 },
1034 #endif
1035 {
1036 .ctl_name = VM_LAPTOP_MODE,
1037 .procname = "laptop_mode",
1038 .data = &laptop_mode,
1039 .maxlen = sizeof(laptop_mode),
1040 .mode = 0644,
1041 .proc_handler = &proc_dointvec_jiffies,
1042 .strategy = &sysctl_jiffies,
1043 },
1044 {
1045 .ctl_name = VM_BLOCK_DUMP,
1046 .procname = "block_dump",
1047 .data = &block_dump,
1048 .maxlen = sizeof(block_dump),
1049 .mode = 0644,
1050 .proc_handler = &proc_dointvec,
1051 .strategy = &sysctl_intvec,
1052 .extra1 = &zero,
1053 },
1054 {
1055 .ctl_name = VM_VFS_CACHE_PRESSURE,
1056 .procname = "vfs_cache_pressure",
1057 .data = &sysctl_vfs_cache_pressure,
1058 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1059 .mode = 0644,
1060 .proc_handler = &proc_dointvec,
1061 .strategy = &sysctl_intvec,
1062 .extra1 = &zero,
1063 },
1064 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1065 {
1066 .ctl_name = VM_LEGACY_VA_LAYOUT,
1067 .procname = "legacy_va_layout",
1068 .data = &sysctl_legacy_va_layout,
1069 .maxlen = sizeof(sysctl_legacy_va_layout),
1070 .mode = 0644,
1071 .proc_handler = &proc_dointvec,
1072 .strategy = &sysctl_intvec,
1073 .extra1 = &zero,
1074 },
1075 #endif
1076 #ifdef CONFIG_NUMA
1077 {
1078 .ctl_name = VM_ZONE_RECLAIM_MODE,
1079 .procname = "zone_reclaim_mode",
1080 .data = &zone_reclaim_mode,
1081 .maxlen = sizeof(zone_reclaim_mode),
1082 .mode = 0644,
1083 .proc_handler = &proc_dointvec,
1084 .strategy = &sysctl_intvec,
1085 .extra1 = &zero,
1086 },
1087 {
1088 .ctl_name = VM_MIN_UNMAPPED,
1089 .procname = "min_unmapped_ratio",
1090 .data = &sysctl_min_unmapped_ratio,
1091 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1092 .mode = 0644,
1093 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
1094 .strategy = &sysctl_intvec,
1095 .extra1 = &zero,
1096 .extra2 = &one_hundred,
1097 },
1098 {
1099 .ctl_name = VM_MIN_SLAB,
1100 .procname = "min_slab_ratio",
1101 .data = &sysctl_min_slab_ratio,
1102 .maxlen = sizeof(sysctl_min_slab_ratio),
1103 .mode = 0644,
1104 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1105 .strategy = &sysctl_intvec,
1106 .extra1 = &zero,
1107 .extra2 = &one_hundred,
1108 },
1109 #endif
1110 #ifdef CONFIG_SMP
1111 {
1112 .ctl_name = CTL_UNNUMBERED,
1113 .procname = "stat_interval",
1114 .data = &sysctl_stat_interval,
1115 .maxlen = sizeof(sysctl_stat_interval),
1116 .mode = 0644,
1117 .proc_handler = &proc_dointvec_jiffies,
1118 .strategy = &sysctl_jiffies,
1119 },
1120 #endif
1121 #ifdef CONFIG_SECURITY
1122 {
1123 .ctl_name = CTL_UNNUMBERED,
1124 .procname = "mmap_min_addr",
1125 .data = &mmap_min_addr,
1126 .maxlen = sizeof(unsigned long),
1127 .mode = 0644,
1128 .proc_handler = &proc_doulongvec_minmax,
1129 },
1130 #endif
1131 #ifdef CONFIG_NUMA
1132 {
1133 .ctl_name = CTL_UNNUMBERED,
1134 .procname = "numa_zonelist_order",
1135 .data = &numa_zonelist_order,
1136 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1137 .mode = 0644,
1138 .proc_handler = &numa_zonelist_order_handler,
1139 .strategy = &sysctl_string,
1140 },
1141 #endif
1142 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1143 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1144 {
1145 .ctl_name = VM_VDSO_ENABLED,
1146 .procname = "vdso_enabled",
1147 .data = &vdso_enabled,
1148 .maxlen = sizeof(vdso_enabled),
1149 .mode = 0644,
1150 .proc_handler = &proc_dointvec,
1151 .strategy = &sysctl_intvec,
1152 .extra1 = &zero,
1153 },
1154 #endif
1155 /*
1156 * NOTE: do not add new entries to this table unless you have read
1157 * Documentation/sysctl/ctl_unnumbered.txt
1158 */
1159 { .ctl_name = 0 }
1160 };
1161
1162 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1163 static struct ctl_table binfmt_misc_table[] = {
1164 { .ctl_name = 0 }
1165 };
1166 #endif
1167
1168 static struct ctl_table fs_table[] = {
1169 {
1170 .ctl_name = FS_NRINODE,
1171 .procname = "inode-nr",
1172 .data = &inodes_stat,
1173 .maxlen = 2*sizeof(int),
1174 .mode = 0444,
1175 .proc_handler = &proc_dointvec,
1176 },
1177 {
1178 .ctl_name = FS_STATINODE,
1179 .procname = "inode-state",
1180 .data = &inodes_stat,
1181 .maxlen = 7*sizeof(int),
1182 .mode = 0444,
1183 .proc_handler = &proc_dointvec,
1184 },
1185 {
1186 .procname = "file-nr",
1187 .data = &files_stat,
1188 .maxlen = 3*sizeof(int),
1189 .mode = 0444,
1190 .proc_handler = &proc_nr_files,
1191 },
1192 {
1193 .ctl_name = FS_MAXFILE,
1194 .procname = "file-max",
1195 .data = &files_stat.max_files,
1196 .maxlen = sizeof(int),
1197 .mode = 0644,
1198 .proc_handler = &proc_dointvec,
1199 },
1200 {
1201 .ctl_name = FS_DENTRY,
1202 .procname = "dentry-state",
1203 .data = &dentry_stat,
1204 .maxlen = 6*sizeof(int),
1205 .mode = 0444,
1206 .proc_handler = &proc_dointvec,
1207 },
1208 {
1209 .ctl_name = FS_OVERFLOWUID,
1210 .procname = "overflowuid",
1211 .data = &fs_overflowuid,
1212 .maxlen = sizeof(int),
1213 .mode = 0644,
1214 .proc_handler = &proc_dointvec_minmax,
1215 .strategy = &sysctl_intvec,
1216 .extra1 = &minolduid,
1217 .extra2 = &maxolduid,
1218 },
1219 {
1220 .ctl_name = FS_OVERFLOWGID,
1221 .procname = "overflowgid",
1222 .data = &fs_overflowgid,
1223 .maxlen = sizeof(int),
1224 .mode = 0644,
1225 .proc_handler = &proc_dointvec_minmax,
1226 .strategy = &sysctl_intvec,
1227 .extra1 = &minolduid,
1228 .extra2 = &maxolduid,
1229 },
1230 {
1231 .ctl_name = FS_LEASES,
1232 .procname = "leases-enable",
1233 .data = &leases_enable,
1234 .maxlen = sizeof(int),
1235 .mode = 0644,
1236 .proc_handler = &proc_dointvec,
1237 },
1238 #ifdef CONFIG_DNOTIFY
1239 {
1240 .ctl_name = FS_DIR_NOTIFY,
1241 .procname = "dir-notify-enable",
1242 .data = &dir_notify_enable,
1243 .maxlen = sizeof(int),
1244 .mode = 0644,
1245 .proc_handler = &proc_dointvec,
1246 },
1247 #endif
1248 #ifdef CONFIG_MMU
1249 {
1250 .ctl_name = FS_LEASE_TIME,
1251 .procname = "lease-break-time",
1252 .data = &lease_break_time,
1253 .maxlen = sizeof(int),
1254 .mode = 0644,
1255 .proc_handler = &proc_dointvec_minmax,
1256 .strategy = &sysctl_intvec,
1257 .extra1 = &zero,
1258 .extra2 = &two,
1259 },
1260 {
1261 .procname = "aio-nr",
1262 .data = &aio_nr,
1263 .maxlen = sizeof(aio_nr),
1264 .mode = 0444,
1265 .proc_handler = &proc_doulongvec_minmax,
1266 },
1267 {
1268 .procname = "aio-max-nr",
1269 .data = &aio_max_nr,
1270 .maxlen = sizeof(aio_max_nr),
1271 .mode = 0644,
1272 .proc_handler = &proc_doulongvec_minmax,
1273 },
1274 #ifdef CONFIG_INOTIFY_USER
1275 {
1276 .ctl_name = FS_INOTIFY,
1277 .procname = "inotify",
1278 .mode = 0555,
1279 .child = inotify_table,
1280 },
1281 #endif
1282 #endif
1283 {
1284 .ctl_name = KERN_SETUID_DUMPABLE,
1285 .procname = "suid_dumpable",
1286 .data = &suid_dumpable,
1287 .maxlen = sizeof(int),
1288 .mode = 0644,
1289 .proc_handler = &proc_dointvec,
1290 },
1291 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1292 {
1293 .ctl_name = CTL_UNNUMBERED,
1294 .procname = "binfmt_misc",
1295 .mode = 0555,
1296 .child = binfmt_misc_table,
1297 },
1298 #endif
1299 /*
1300 * NOTE: do not add new entries to this table unless you have read
1301 * Documentation/sysctl/ctl_unnumbered.txt
1302 */
1303 { .ctl_name = 0 }
1304 };
1305
1306 static struct ctl_table debug_table[] = {
1307 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1308 {
1309 .ctl_name = CTL_UNNUMBERED,
1310 .procname = "exception-trace",
1311 .data = &show_unhandled_signals,
1312 .maxlen = sizeof(int),
1313 .mode = 0644,
1314 .proc_handler = proc_dointvec
1315 },
1316 #endif
1317 { .ctl_name = 0 }
1318 };
1319
1320 static struct ctl_table dev_table[] = {
1321 { .ctl_name = 0 }
1322 };
1323
1324 static DEFINE_SPINLOCK(sysctl_lock);
1325
1326 /* called under sysctl_lock */
1327 static int use_table(struct ctl_table_header *p)
1328 {
1329 if (unlikely(p->unregistering))
1330 return 0;
1331 p->used++;
1332 return 1;
1333 }
1334
1335 /* called under sysctl_lock */
1336 static void unuse_table(struct ctl_table_header *p)
1337 {
1338 if (!--p->used)
1339 if (unlikely(p->unregistering))
1340 complete(p->unregistering);
1341 }
1342
1343 /* called under sysctl_lock, will reacquire if has to wait */
1344 static void start_unregistering(struct ctl_table_header *p)
1345 {
1346 /*
1347 * if p->used is 0, nobody will ever touch that entry again;
1348 * we'll eliminate all paths to it before dropping sysctl_lock
1349 */
1350 if (unlikely(p->used)) {
1351 struct completion wait;
1352 init_completion(&wait);
1353 p->unregistering = &wait;
1354 spin_unlock(&sysctl_lock);
1355 wait_for_completion(&wait);
1356 spin_lock(&sysctl_lock);
1357 }
1358 /*
1359 * do not remove from the list until nobody holds it; walking the
1360 * list in do_sysctl() relies on that.
1361 */
1362 list_del_init(&p->ctl_entry);
1363 }
1364
1365 void sysctl_head_finish(struct ctl_table_header *head)
1366 {
1367 if (!head)
1368 return;
1369 spin_lock(&sysctl_lock);
1370 unuse_table(head);
1371 spin_unlock(&sysctl_lock);
1372 }
1373
1374 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1375 {
1376 struct ctl_table_header *head;
1377 struct list_head *tmp;
1378 spin_lock(&sysctl_lock);
1379 if (prev) {
1380 tmp = &prev->ctl_entry;
1381 unuse_table(prev);
1382 goto next;
1383 }
1384 tmp = &root_table_header.ctl_entry;
1385 for (;;) {
1386 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1387
1388 if (!use_table(head))
1389 goto next;
1390 spin_unlock(&sysctl_lock);
1391 return head;
1392 next:
1393 tmp = tmp->next;
1394 if (tmp == &root_table_header.ctl_entry)
1395 break;
1396 }
1397 spin_unlock(&sysctl_lock);
1398 return NULL;
1399 }
1400
1401 #ifdef CONFIG_SYSCTL_SYSCALL
1402 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1403 void __user *newval, size_t newlen)
1404 {
1405 struct ctl_table_header *head;
1406 int error = -ENOTDIR;
1407
1408 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1409 return -ENOTDIR;
1410 if (oldval) {
1411 int old_len;
1412 if (!oldlenp || get_user(old_len, oldlenp))
1413 return -EFAULT;
1414 }
1415
1416 for (head = sysctl_head_next(NULL); head;
1417 head = sysctl_head_next(head)) {
1418 error = parse_table(name, nlen, oldval, oldlenp,
1419 newval, newlen, head->ctl_table);
1420 if (error != -ENOTDIR) {
1421 sysctl_head_finish(head);
1422 break;
1423 }
1424 }
1425 return error;
1426 }
1427
1428 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1429 {
1430 struct __sysctl_args tmp;
1431 int error;
1432
1433 if (copy_from_user(&tmp, args, sizeof(tmp)))
1434 return -EFAULT;
1435
1436 error = deprecated_sysctl_warning(&tmp);
1437 if (error)
1438 goto out;
1439
1440 lock_kernel();
1441 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1442 tmp.newval, tmp.newlen);
1443 unlock_kernel();
1444 out:
1445 return error;
1446 }
1447 #endif /* CONFIG_SYSCTL_SYSCALL */
1448
1449 /*
1450 * sysctl_perm does NOT grant the superuser all rights automatically, because
1451 * some sysctl variables are readonly even to root.
1452 */
1453
1454 static int test_perm(int mode, int op)
1455 {
1456 if (!current->euid)
1457 mode >>= 6;
1458 else if (in_egroup_p(0))
1459 mode >>= 3;
1460 if ((mode & op & 0007) == op)
1461 return 0;
1462 return -EACCES;
1463 }
1464
1465 int sysctl_perm(struct ctl_table *table, int op)
1466 {
1467 int error;
1468 error = security_sysctl(table, op);
1469 if (error)
1470 return error;
1471 return test_perm(table->mode, op);
1472 }
1473
1474 #ifdef CONFIG_SYSCTL_SYSCALL
1475 static 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 *table)
1479 {
1480 int n;
1481 repeat:
1482 if (!nlen)
1483 return -ENOTDIR;
1484 if (get_user(n, name))
1485 return -EFAULT;
1486 for ( ; table->ctl_name || table->procname; table++) {
1487 if (!table->ctl_name)
1488 continue;
1489 if (n == table->ctl_name) {
1490 int error;
1491 if (table->child) {
1492 if (sysctl_perm(table, 001))
1493 return -EPERM;
1494 name++;
1495 nlen--;
1496 table = table->child;
1497 goto repeat;
1498 }
1499 error = do_sysctl_strategy(table, name, nlen,
1500 oldval, oldlenp,
1501 newval, newlen);
1502 return error;
1503 }
1504 }
1505 return -ENOTDIR;
1506 }
1507
1508 /* Perform the actual read/write of a sysctl table entry. */
1509 int do_sysctl_strategy (struct ctl_table *table,
1510 int __user *name, int nlen,
1511 void __user *oldval, size_t __user *oldlenp,
1512 void __user *newval, size_t newlen)
1513 {
1514 int op = 0, rc;
1515
1516 if (oldval)
1517 op |= 004;
1518 if (newval)
1519 op |= 002;
1520 if (sysctl_perm(table, op))
1521 return -EPERM;
1522
1523 if (table->strategy) {
1524 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1525 newval, newlen);
1526 if (rc < 0)
1527 return rc;
1528 if (rc > 0)
1529 return 0;
1530 }
1531
1532 /* If there is no strategy routine, or if the strategy returns
1533 * zero, proceed with automatic r/w */
1534 if (table->data && table->maxlen) {
1535 rc = sysctl_data(table, name, nlen, oldval, oldlenp,
1536 newval, newlen);
1537 if (rc < 0)
1538 return rc;
1539 }
1540 return 0;
1541 }
1542 #endif /* CONFIG_SYSCTL_SYSCALL */
1543
1544 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1545 {
1546 for (; table->ctl_name || table->procname; table++) {
1547 table->parent = parent;
1548 if (table->child)
1549 sysctl_set_parent(table, table->child);
1550 }
1551 }
1552
1553 static __init int sysctl_init(void)
1554 {
1555 int err;
1556 sysctl_set_parent(NULL, root_table);
1557 err = sysctl_check_table(root_table);
1558 return 0;
1559 }
1560
1561 core_initcall(sysctl_init);
1562
1563 /**
1564 * register_sysctl_table - register a sysctl hierarchy
1565 * @table: the top-level table structure
1566 *
1567 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1568 * array. An entry with a ctl_name of 0 terminates the table.
1569 *
1570 * The members of the &struct ctl_table structure are used as follows:
1571 *
1572 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1573 * must be unique within that level of sysctl
1574 *
1575 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1576 * enter a sysctl file
1577 *
1578 * data - a pointer to data for use by proc_handler
1579 *
1580 * maxlen - the maximum size in bytes of the data
1581 *
1582 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1583 *
1584 * child - a pointer to the child sysctl table if this entry is a directory, or
1585 * %NULL.
1586 *
1587 * proc_handler - the text handler routine (described below)
1588 *
1589 * strategy - the strategy routine (described below)
1590 *
1591 * de - for internal use by the sysctl routines
1592 *
1593 * extra1, extra2 - extra pointers usable by the proc handler routines
1594 *
1595 * Leaf nodes in the sysctl tree will be represented by a single file
1596 * under /proc; non-leaf nodes will be represented by directories.
1597 *
1598 * sysctl(2) can automatically manage read and write requests through
1599 * the sysctl table. The data and maxlen fields of the ctl_table
1600 * struct enable minimal validation of the values being written to be
1601 * performed, and the mode field allows minimal authentication.
1602 *
1603 * More sophisticated management can be enabled by the provision of a
1604 * strategy routine with the table entry. This will be called before
1605 * any automatic read or write of the data is performed.
1606 *
1607 * The strategy routine may return
1608 *
1609 * < 0 - Error occurred (error is passed to user process)
1610 *
1611 * 0 - OK - proceed with automatic read or write.
1612 *
1613 * > 0 - OK - read or write has been done by the strategy routine, so
1614 * return immediately.
1615 *
1616 * There must be a proc_handler routine for any terminal nodes
1617 * mirrored under /proc/sys (non-terminals are handled by a built-in
1618 * directory handler). Several default handlers are available to
1619 * cover common cases -
1620 *
1621 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1622 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1623 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1624 *
1625 * It is the handler's job to read the input buffer from user memory
1626 * and process it. The handler should return 0 on success.
1627 *
1628 * This routine returns %NULL on a failure to register, and a pointer
1629 * to the table header on success.
1630 */
1631 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1632 {
1633 struct ctl_table_header *tmp;
1634 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1635 if (!tmp)
1636 return NULL;
1637 tmp->ctl_table = table;
1638 INIT_LIST_HEAD(&tmp->ctl_entry);
1639 tmp->used = 0;
1640 tmp->unregistering = NULL;
1641 sysctl_set_parent(NULL, table);
1642 if (sysctl_check_table(tmp->ctl_table)) {
1643 kfree(tmp);
1644 return NULL;
1645 }
1646 spin_lock(&sysctl_lock);
1647 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
1648 spin_unlock(&sysctl_lock);
1649 return tmp;
1650 }
1651
1652 /**
1653 * unregister_sysctl_table - unregister a sysctl table hierarchy
1654 * @header: the header returned from register_sysctl_table
1655 *
1656 * Unregisters the sysctl table and all children. proc entries may not
1657 * actually be removed until they are no longer used by anyone.
1658 */
1659 void unregister_sysctl_table(struct ctl_table_header * header)
1660 {
1661 might_sleep();
1662
1663 if (header == NULL)
1664 return;
1665
1666 spin_lock(&sysctl_lock);
1667 start_unregistering(header);
1668 spin_unlock(&sysctl_lock);
1669 kfree(header);
1670 }
1671
1672 #else /* !CONFIG_SYSCTL */
1673 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1674 {
1675 return NULL;
1676 }
1677
1678 void unregister_sysctl_table(struct ctl_table_header * table)
1679 {
1680 }
1681
1682 #endif /* CONFIG_SYSCTL */
1683
1684 /*
1685 * /proc/sys support
1686 */
1687
1688 #ifdef CONFIG_PROC_SYSCTL
1689
1690 static int _proc_do_string(void* data, int maxlen, int write,
1691 struct file *filp, void __user *buffer,
1692 size_t *lenp, loff_t *ppos)
1693 {
1694 size_t len;
1695 char __user *p;
1696 char c;
1697
1698 if (!data || !maxlen || !*lenp) {
1699 *lenp = 0;
1700 return 0;
1701 }
1702
1703 if (write) {
1704 len = 0;
1705 p = buffer;
1706 while (len < *lenp) {
1707 if (get_user(c, p++))
1708 return -EFAULT;
1709 if (c == 0 || c == '\n')
1710 break;
1711 len++;
1712 }
1713 if (len >= maxlen)
1714 len = maxlen-1;
1715 if(copy_from_user(data, buffer, len))
1716 return -EFAULT;
1717 ((char *) data)[len] = 0;
1718 *ppos += *lenp;
1719 } else {
1720 len = strlen(data);
1721 if (len > maxlen)
1722 len = maxlen;
1723
1724 if (*ppos > len) {
1725 *lenp = 0;
1726 return 0;
1727 }
1728
1729 data += *ppos;
1730 len -= *ppos;
1731
1732 if (len > *lenp)
1733 len = *lenp;
1734 if (len)
1735 if(copy_to_user(buffer, data, len))
1736 return -EFAULT;
1737 if (len < *lenp) {
1738 if(put_user('\n', ((char __user *) buffer) + len))
1739 return -EFAULT;
1740 len++;
1741 }
1742 *lenp = len;
1743 *ppos += len;
1744 }
1745 return 0;
1746 }
1747
1748 /**
1749 * proc_dostring - read a string sysctl
1750 * @table: the sysctl table
1751 * @write: %TRUE if this is a write to the sysctl file
1752 * @filp: the file structure
1753 * @buffer: the user buffer
1754 * @lenp: the size of the user buffer
1755 * @ppos: file position
1756 *
1757 * Reads/writes a string from/to the user buffer. If the kernel
1758 * buffer provided is not large enough to hold the string, the
1759 * string is truncated. The copied string is %NULL-terminated.
1760 * If the string is being read by the user process, it is copied
1761 * and a newline '\n' is added. It is truncated if the buffer is
1762 * not large enough.
1763 *
1764 * Returns 0 on success.
1765 */
1766 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
1767 void __user *buffer, size_t *lenp, loff_t *ppos)
1768 {
1769 return _proc_do_string(table->data, table->maxlen, write, filp,
1770 buffer, lenp, ppos);
1771 }
1772
1773
1774 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1775 int *valp,
1776 int write, void *data)
1777 {
1778 if (write) {
1779 *valp = *negp ? -*lvalp : *lvalp;
1780 } else {
1781 int val = *valp;
1782 if (val < 0) {
1783 *negp = -1;
1784 *lvalp = (unsigned long)-val;
1785 } else {
1786 *negp = 0;
1787 *lvalp = (unsigned long)val;
1788 }
1789 }
1790 return 0;
1791 }
1792
1793 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1794 int write, struct file *filp, void __user *buffer,
1795 size_t *lenp, loff_t *ppos,
1796 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1797 int write, void *data),
1798 void *data)
1799 {
1800 #define TMPBUFLEN 21
1801 int *i, vleft, first=1, neg, val;
1802 unsigned long lval;
1803 size_t left, len;
1804
1805 char buf[TMPBUFLEN], *p;
1806 char __user *s = buffer;
1807
1808 if (!tbl_data || !table->maxlen || !*lenp ||
1809 (*ppos && !write)) {
1810 *lenp = 0;
1811 return 0;
1812 }
1813
1814 i = (int *) tbl_data;
1815 vleft = table->maxlen / sizeof(*i);
1816 left = *lenp;
1817
1818 if (!conv)
1819 conv = do_proc_dointvec_conv;
1820
1821 for (; left && vleft--; i++, first=0) {
1822 if (write) {
1823 while (left) {
1824 char c;
1825 if (get_user(c, s))
1826 return -EFAULT;
1827 if (!isspace(c))
1828 break;
1829 left--;
1830 s++;
1831 }
1832 if (!left)
1833 break;
1834 neg = 0;
1835 len = left;
1836 if (len > sizeof(buf) - 1)
1837 len = sizeof(buf) - 1;
1838 if (copy_from_user(buf, s, len))
1839 return -EFAULT;
1840 buf[len] = 0;
1841 p = buf;
1842 if (*p == '-' && left > 1) {
1843 neg = 1;
1844 p++;
1845 }
1846 if (*p < '0' || *p > '9')
1847 break;
1848
1849 lval = simple_strtoul(p, &p, 0);
1850
1851 len = p-buf;
1852 if ((len < left) && *p && !isspace(*p))
1853 break;
1854 if (neg)
1855 val = -val;
1856 s += len;
1857 left -= len;
1858
1859 if (conv(&neg, &lval, i, 1, data))
1860 break;
1861 } else {
1862 p = buf;
1863 if (!first)
1864 *p++ = '\t';
1865
1866 if (conv(&neg, &lval, i, 0, data))
1867 break;
1868
1869 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1870 len = strlen(buf);
1871 if (len > left)
1872 len = left;
1873 if(copy_to_user(s, buf, len))
1874 return -EFAULT;
1875 left -= len;
1876 s += len;
1877 }
1878 }
1879
1880 if (!write && !first && left) {
1881 if(put_user('\n', s))
1882 return -EFAULT;
1883 left--, s++;
1884 }
1885 if (write) {
1886 while (left) {
1887 char c;
1888 if (get_user(c, s++))
1889 return -EFAULT;
1890 if (!isspace(c))
1891 break;
1892 left--;
1893 }
1894 }
1895 if (write && first)
1896 return -EINVAL;
1897 *lenp -= left;
1898 *ppos += *lenp;
1899 return 0;
1900 #undef TMPBUFLEN
1901 }
1902
1903 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
1904 void __user *buffer, size_t *lenp, loff_t *ppos,
1905 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1906 int write, void *data),
1907 void *data)
1908 {
1909 return __do_proc_dointvec(table->data, table, write, filp,
1910 buffer, lenp, ppos, conv, data);
1911 }
1912
1913 /**
1914 * proc_dointvec - read a vector of integers
1915 * @table: the sysctl table
1916 * @write: %TRUE if this is a write to the sysctl file
1917 * @filp: the file structure
1918 * @buffer: the user buffer
1919 * @lenp: the size of the user buffer
1920 * @ppos: file position
1921 *
1922 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1923 * values from/to the user buffer, treated as an ASCII string.
1924 *
1925 * Returns 0 on success.
1926 */
1927 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
1928 void __user *buffer, size_t *lenp, loff_t *ppos)
1929 {
1930 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1931 NULL,NULL);
1932 }
1933
1934 #define OP_SET 0
1935 #define OP_AND 1
1936 #define OP_OR 2
1937
1938 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1939 int *valp,
1940 int write, void *data)
1941 {
1942 int op = *(int *)data;
1943 if (write) {
1944 int val = *negp ? -*lvalp : *lvalp;
1945 switch(op) {
1946 case OP_SET: *valp = val; break;
1947 case OP_AND: *valp &= val; break;
1948 case OP_OR: *valp |= val; break;
1949 }
1950 } else {
1951 int val = *valp;
1952 if (val < 0) {
1953 *negp = -1;
1954 *lvalp = (unsigned long)-val;
1955 } else {
1956 *negp = 0;
1957 *lvalp = (unsigned long)val;
1958 }
1959 }
1960 return 0;
1961 }
1962
1963 #ifdef CONFIG_SECURITY_CAPABILITIES
1964 /*
1965 * init may raise the set.
1966 */
1967
1968 int proc_dointvec_bset(struct ctl_table *table, int write, struct file *filp,
1969 void __user *buffer, size_t *lenp, loff_t *ppos)
1970 {
1971 int op;
1972
1973 if (write && !capable(CAP_SYS_MODULE)) {
1974 return -EPERM;
1975 }
1976
1977 op = is_global_init(current) ? OP_SET : OP_AND;
1978 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1979 do_proc_dointvec_bset_conv,&op);
1980 }
1981 #endif /* def CONFIG_SECURITY_CAPABILITIES */
1982
1983 /*
1984 * Taint values can only be increased
1985 */
1986 static int proc_dointvec_taint(struct ctl_table *table, int write, struct file *filp,
1987 void __user *buffer, size_t *lenp, loff_t *ppos)
1988 {
1989 int op;
1990
1991 if (write && !capable(CAP_SYS_ADMIN))
1992 return -EPERM;
1993
1994 op = OP_OR;
1995 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1996 do_proc_dointvec_bset_conv,&op);
1997 }
1998
1999 struct do_proc_dointvec_minmax_conv_param {
2000 int *min;
2001 int *max;
2002 };
2003
2004 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2005 int *valp,
2006 int write, void *data)
2007 {
2008 struct do_proc_dointvec_minmax_conv_param *param = data;
2009 if (write) {
2010 int val = *negp ? -*lvalp : *lvalp;
2011 if ((param->min && *param->min > val) ||
2012 (param->max && *param->max < val))
2013 return -EINVAL;
2014 *valp = val;
2015 } else {
2016 int val = *valp;
2017 if (val < 0) {
2018 *negp = -1;
2019 *lvalp = (unsigned long)-val;
2020 } else {
2021 *negp = 0;
2022 *lvalp = (unsigned long)val;
2023 }
2024 }
2025 return 0;
2026 }
2027
2028 /**
2029 * proc_dointvec_minmax - read a vector of integers with min/max values
2030 * @table: the sysctl table
2031 * @write: %TRUE if this is a write to the sysctl file
2032 * @filp: the file structure
2033 * @buffer: the user buffer
2034 * @lenp: the size of the user buffer
2035 * @ppos: file position
2036 *
2037 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2038 * values from/to the user buffer, treated as an ASCII string.
2039 *
2040 * This routine will ensure the values are within the range specified by
2041 * table->extra1 (min) and table->extra2 (max).
2042 *
2043 * Returns 0 on success.
2044 */
2045 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2046 void __user *buffer, size_t *lenp, loff_t *ppos)
2047 {
2048 struct do_proc_dointvec_minmax_conv_param param = {
2049 .min = (int *) table->extra1,
2050 .max = (int *) table->extra2,
2051 };
2052 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2053 do_proc_dointvec_minmax_conv, &param);
2054 }
2055
2056 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2057 struct file *filp,
2058 void __user *buffer,
2059 size_t *lenp, loff_t *ppos,
2060 unsigned long convmul,
2061 unsigned long convdiv)
2062 {
2063 #define TMPBUFLEN 21
2064 unsigned long *i, *min, *max, val;
2065 int vleft, first=1, neg;
2066 size_t len, left;
2067 char buf[TMPBUFLEN], *p;
2068 char __user *s = buffer;
2069
2070 if (!data || !table->maxlen || !*lenp ||
2071 (*ppos && !write)) {
2072 *lenp = 0;
2073 return 0;
2074 }
2075
2076 i = (unsigned long *) data;
2077 min = (unsigned long *) table->extra1;
2078 max = (unsigned long *) table->extra2;
2079 vleft = table->maxlen / sizeof(unsigned long);
2080 left = *lenp;
2081
2082 for (; left && vleft--; i++, min++, max++, first=0) {
2083 if (write) {
2084 while (left) {
2085 char c;
2086 if (get_user(c, s))
2087 return -EFAULT;
2088 if (!isspace(c))
2089 break;
2090 left--;
2091 s++;
2092 }
2093 if (!left)
2094 break;
2095 neg = 0;
2096 len = left;
2097 if (len > TMPBUFLEN-1)
2098 len = TMPBUFLEN-1;
2099 if (copy_from_user(buf, s, len))
2100 return -EFAULT;
2101 buf[len] = 0;
2102 p = buf;
2103 if (*p == '-' && left > 1) {
2104 neg = 1;
2105 p++;
2106 }
2107 if (*p < '0' || *p > '9')
2108 break;
2109 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2110 len = p-buf;
2111 if ((len < left) && *p && !isspace(*p))
2112 break;
2113 if (neg)
2114 val = -val;
2115 s += len;
2116 left -= len;
2117
2118 if(neg)
2119 continue;
2120 if ((min && val < *min) || (max && val > *max))
2121 continue;
2122 *i = val;
2123 } else {
2124 p = buf;
2125 if (!first)
2126 *p++ = '\t';
2127 sprintf(p, "%lu", convdiv * (*i) / convmul);
2128 len = strlen(buf);
2129 if (len > left)
2130 len = left;
2131 if(copy_to_user(s, buf, len))
2132 return -EFAULT;
2133 left -= len;
2134 s += len;
2135 }
2136 }
2137
2138 if (!write && !first && left) {
2139 if(put_user('\n', s))
2140 return -EFAULT;
2141 left--, s++;
2142 }
2143 if (write) {
2144 while (left) {
2145 char c;
2146 if (get_user(c, s++))
2147 return -EFAULT;
2148 if (!isspace(c))
2149 break;
2150 left--;
2151 }
2152 }
2153 if (write && first)
2154 return -EINVAL;
2155 *lenp -= left;
2156 *ppos += *lenp;
2157 return 0;
2158 #undef TMPBUFLEN
2159 }
2160
2161 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2162 struct file *filp,
2163 void __user *buffer,
2164 size_t *lenp, loff_t *ppos,
2165 unsigned long convmul,
2166 unsigned long convdiv)
2167 {
2168 return __do_proc_doulongvec_minmax(table->data, table, write,
2169 filp, buffer, lenp, ppos, convmul, convdiv);
2170 }
2171
2172 /**
2173 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2174 * @table: the sysctl table
2175 * @write: %TRUE if this is a write to the sysctl file
2176 * @filp: the file structure
2177 * @buffer: the user buffer
2178 * @lenp: the size of the user buffer
2179 * @ppos: file position
2180 *
2181 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2182 * values from/to the user buffer, treated as an ASCII string.
2183 *
2184 * This routine will ensure the values are within the range specified by
2185 * table->extra1 (min) and table->extra2 (max).
2186 *
2187 * Returns 0 on success.
2188 */
2189 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2190 void __user *buffer, size_t *lenp, loff_t *ppos)
2191 {
2192 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2193 }
2194
2195 /**
2196 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2197 * @table: the sysctl table
2198 * @write: %TRUE if this is a write to the sysctl file
2199 * @filp: the file structure
2200 * @buffer: the user buffer
2201 * @lenp: the size of the user buffer
2202 * @ppos: file position
2203 *
2204 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2205 * values from/to the user buffer, treated as an ASCII string. The values
2206 * are treated as milliseconds, and converted to jiffies when they are stored.
2207 *
2208 * This routine will ensure the values are within the range specified by
2209 * table->extra1 (min) and table->extra2 (max).
2210 *
2211 * Returns 0 on success.
2212 */
2213 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2214 struct file *filp,
2215 void __user *buffer,
2216 size_t *lenp, loff_t *ppos)
2217 {
2218 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2219 lenp, ppos, HZ, 1000l);
2220 }
2221
2222
2223 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2224 int *valp,
2225 int write, void *data)
2226 {
2227 if (write) {
2228 if (*lvalp > LONG_MAX / HZ)
2229 return 1;
2230 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2231 } else {
2232 int val = *valp;
2233 unsigned long lval;
2234 if (val < 0) {
2235 *negp = -1;
2236 lval = (unsigned long)-val;
2237 } else {
2238 *negp = 0;
2239 lval = (unsigned long)val;
2240 }
2241 *lvalp = lval / HZ;
2242 }
2243 return 0;
2244 }
2245
2246 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2247 int *valp,
2248 int write, void *data)
2249 {
2250 if (write) {
2251 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2252 return 1;
2253 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2254 } else {
2255 int val = *valp;
2256 unsigned long lval;
2257 if (val < 0) {
2258 *negp = -1;
2259 lval = (unsigned long)-val;
2260 } else {
2261 *negp = 0;
2262 lval = (unsigned long)val;
2263 }
2264 *lvalp = jiffies_to_clock_t(lval);
2265 }
2266 return 0;
2267 }
2268
2269 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2270 int *valp,
2271 int write, void *data)
2272 {
2273 if (write) {
2274 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2275 } else {
2276 int val = *valp;
2277 unsigned long lval;
2278 if (val < 0) {
2279 *negp = -1;
2280 lval = (unsigned long)-val;
2281 } else {
2282 *negp = 0;
2283 lval = (unsigned long)val;
2284 }
2285 *lvalp = jiffies_to_msecs(lval);
2286 }
2287 return 0;
2288 }
2289
2290 /**
2291 * proc_dointvec_jiffies - read a vector of integers as seconds
2292 * @table: the sysctl table
2293 * @write: %TRUE if this is a write to the sysctl file
2294 * @filp: the file structure
2295 * @buffer: the user buffer
2296 * @lenp: the size of the user buffer
2297 * @ppos: file position
2298 *
2299 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2300 * values from/to the user buffer, treated as an ASCII string.
2301 * The values read are assumed to be in seconds, and are converted into
2302 * jiffies.
2303 *
2304 * Returns 0 on success.
2305 */
2306 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2307 void __user *buffer, size_t *lenp, loff_t *ppos)
2308 {
2309 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2310 do_proc_dointvec_jiffies_conv,NULL);
2311 }
2312
2313 /**
2314 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2315 * @table: the sysctl table
2316 * @write: %TRUE if this is a write to the sysctl file
2317 * @filp: the file structure
2318 * @buffer: the user buffer
2319 * @lenp: the size of the user buffer
2320 * @ppos: pointer to the file position
2321 *
2322 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2323 * values from/to the user buffer, treated as an ASCII string.
2324 * The values read are assumed to be in 1/USER_HZ seconds, and
2325 * are converted into jiffies.
2326 *
2327 * Returns 0 on success.
2328 */
2329 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2330 void __user *buffer, size_t *lenp, loff_t *ppos)
2331 {
2332 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2333 do_proc_dointvec_userhz_jiffies_conv,NULL);
2334 }
2335
2336 /**
2337 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2338 * @table: the sysctl table
2339 * @write: %TRUE if this is a write to the sysctl file
2340 * @filp: the file structure
2341 * @buffer: the user buffer
2342 * @lenp: the size of the user buffer
2343 * @ppos: file position
2344 * @ppos: the current position in the file
2345 *
2346 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2347 * values from/to the user buffer, treated as an ASCII string.
2348 * The values read are assumed to be in 1/1000 seconds, and
2349 * are converted into jiffies.
2350 *
2351 * Returns 0 on success.
2352 */
2353 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2354 void __user *buffer, size_t *lenp, loff_t *ppos)
2355 {
2356 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2357 do_proc_dointvec_ms_jiffies_conv, NULL);
2358 }
2359
2360 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2361 void __user *buffer, size_t *lenp, loff_t *ppos)
2362 {
2363 struct pid *new_pid;
2364 pid_t tmp;
2365 int r;
2366
2367 tmp = pid_nr_ns(cad_pid, current->nsproxy->pid_ns);
2368
2369 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2370 lenp, ppos, NULL, NULL);
2371 if (r || !write)
2372 return r;
2373
2374 new_pid = find_get_pid(tmp);
2375 if (!new_pid)
2376 return -ESRCH;
2377
2378 put_pid(xchg(&cad_pid, new_pid));
2379 return 0;
2380 }
2381
2382 #else /* CONFIG_PROC_FS */
2383
2384 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2385 void __user *buffer, size_t *lenp, loff_t *ppos)
2386 {
2387 return -ENOSYS;
2388 }
2389
2390 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2391 void __user *buffer, size_t *lenp, loff_t *ppos)
2392 {
2393 return -ENOSYS;
2394 }
2395
2396 int proc_dointvec_bset(struct ctl_table *table, int write, struct file *filp,
2397 void __user *buffer, size_t *lenp, loff_t *ppos)
2398 {
2399 return -ENOSYS;
2400 }
2401
2402 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2403 void __user *buffer, size_t *lenp, loff_t *ppos)
2404 {
2405 return -ENOSYS;
2406 }
2407
2408 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2409 void __user *buffer, size_t *lenp, loff_t *ppos)
2410 {
2411 return -ENOSYS;
2412 }
2413
2414 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2415 void __user *buffer, size_t *lenp, loff_t *ppos)
2416 {
2417 return -ENOSYS;
2418 }
2419
2420 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2421 void __user *buffer, size_t *lenp, loff_t *ppos)
2422 {
2423 return -ENOSYS;
2424 }
2425
2426 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2427 void __user *buffer, size_t *lenp, loff_t *ppos)
2428 {
2429 return -ENOSYS;
2430 }
2431
2432 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2433 struct file *filp,
2434 void __user *buffer,
2435 size_t *lenp, loff_t *ppos)
2436 {
2437 return -ENOSYS;
2438 }
2439
2440
2441 #endif /* CONFIG_PROC_FS */
2442
2443
2444 #ifdef CONFIG_SYSCTL_SYSCALL
2445 /*
2446 * General sysctl support routines
2447 */
2448
2449 /* The generic sysctl data routine (used if no strategy routine supplied) */
2450 int sysctl_data(struct ctl_table *table, int __user *name, int nlen,
2451 void __user *oldval, size_t __user *oldlenp,
2452 void __user *newval, size_t newlen)
2453 {
2454 size_t len;
2455
2456 /* Get out of I don't have a variable */
2457 if (!table->data || !table->maxlen)
2458 return -ENOTDIR;
2459
2460 if (oldval && oldlenp) {
2461 if (get_user(len, oldlenp))
2462 return -EFAULT;
2463 if (len) {
2464 if (len > table->maxlen)
2465 len = table->maxlen;
2466 if (copy_to_user(oldval, table->data, len))
2467 return -EFAULT;
2468 if (put_user(len, oldlenp))
2469 return -EFAULT;
2470 }
2471 }
2472
2473 if (newval && newlen) {
2474 if (newlen > table->maxlen)
2475 newlen = table->maxlen;
2476
2477 if (copy_from_user(table->data, newval, newlen))
2478 return -EFAULT;
2479 }
2480 return 1;
2481 }
2482
2483 /* The generic string strategy routine: */
2484 int sysctl_string(struct ctl_table *table, int __user *name, int nlen,
2485 void __user *oldval, size_t __user *oldlenp,
2486 void __user *newval, size_t newlen)
2487 {
2488 if (!table->data || !table->maxlen)
2489 return -ENOTDIR;
2490
2491 if (oldval && oldlenp) {
2492 size_t bufsize;
2493 if (get_user(bufsize, oldlenp))
2494 return -EFAULT;
2495 if (bufsize) {
2496 size_t len = strlen(table->data), copied;
2497
2498 /* This shouldn't trigger for a well-formed sysctl */
2499 if (len > table->maxlen)
2500 len = table->maxlen;
2501
2502 /* Copy up to a max of bufsize-1 bytes of the string */
2503 copied = (len >= bufsize) ? bufsize - 1 : len;
2504
2505 if (copy_to_user(oldval, table->data, copied) ||
2506 put_user(0, (char __user *)(oldval + copied)))
2507 return -EFAULT;
2508 if (put_user(len, oldlenp))
2509 return -EFAULT;
2510 }
2511 }
2512 if (newval && newlen) {
2513 size_t len = newlen;
2514 if (len > table->maxlen)
2515 len = table->maxlen;
2516 if(copy_from_user(table->data, newval, len))
2517 return -EFAULT;
2518 if (len == table->maxlen)
2519 len--;
2520 ((char *) table->data)[len] = 0;
2521 }
2522 return 1;
2523 }
2524
2525 /*
2526 * This function makes sure that all of the integers in the vector
2527 * are between the minimum and maximum values given in the arrays
2528 * table->extra1 and table->extra2, respectively.
2529 */
2530 int sysctl_intvec(struct ctl_table *table, int __user *name, int nlen,
2531 void __user *oldval, size_t __user *oldlenp,
2532 void __user *newval, size_t newlen)
2533 {
2534
2535 if (newval && newlen) {
2536 int __user *vec = (int __user *) newval;
2537 int *min = (int *) table->extra1;
2538 int *max = (int *) table->extra2;
2539 size_t length;
2540 int i;
2541
2542 if (newlen % sizeof(int) != 0)
2543 return -EINVAL;
2544
2545 if (!table->extra1 && !table->extra2)
2546 return 0;
2547
2548 if (newlen > table->maxlen)
2549 newlen = table->maxlen;
2550 length = newlen / sizeof(int);
2551
2552 for (i = 0; i < length; i++) {
2553 int value;
2554 if (get_user(value, vec + i))
2555 return -EFAULT;
2556 if (min && value < min[i])
2557 return -EINVAL;
2558 if (max && value > max[i])
2559 return -EINVAL;
2560 }
2561 }
2562 return 0;
2563 }
2564
2565 /* Strategy function to convert jiffies to seconds */
2566 int sysctl_jiffies(struct ctl_table *table, int __user *name, int nlen,
2567 void __user *oldval, size_t __user *oldlenp,
2568 void __user *newval, size_t newlen)
2569 {
2570 if (oldval && oldlenp) {
2571 size_t olen;
2572
2573 if (get_user(olen, oldlenp))
2574 return -EFAULT;
2575 if (olen) {
2576 int val;
2577
2578 if (olen < sizeof(int))
2579 return -EINVAL;
2580
2581 val = *(int *)(table->data) / HZ;
2582 if (put_user(val, (int __user *)oldval))
2583 return -EFAULT;
2584 if (put_user(sizeof(int), oldlenp))
2585 return -EFAULT;
2586 }
2587 }
2588 if (newval && newlen) {
2589 int new;
2590 if (newlen != sizeof(int))
2591 return -EINVAL;
2592 if (get_user(new, (int __user *)newval))
2593 return -EFAULT;
2594 *(int *)(table->data) = new*HZ;
2595 }
2596 return 1;
2597 }
2598
2599 /* Strategy function to convert jiffies to seconds */
2600 int sysctl_ms_jiffies(struct ctl_table *table, int __user *name, int nlen,
2601 void __user *oldval, size_t __user *oldlenp,
2602 void __user *newval, size_t newlen)
2603 {
2604 if (oldval && oldlenp) {
2605 size_t olen;
2606
2607 if (get_user(olen, oldlenp))
2608 return -EFAULT;
2609 if (olen) {
2610 int val;
2611
2612 if (olen < sizeof(int))
2613 return -EINVAL;
2614
2615 val = jiffies_to_msecs(*(int *)(table->data));
2616 if (put_user(val, (int __user *)oldval))
2617 return -EFAULT;
2618 if (put_user(sizeof(int), oldlenp))
2619 return -EFAULT;
2620 }
2621 }
2622 if (newval && newlen) {
2623 int new;
2624 if (newlen != sizeof(int))
2625 return -EINVAL;
2626 if (get_user(new, (int __user *)newval))
2627 return -EFAULT;
2628 *(int *)(table->data) = msecs_to_jiffies(new);
2629 }
2630 return 1;
2631 }
2632
2633
2634
2635 #else /* CONFIG_SYSCTL_SYSCALL */
2636
2637
2638 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2639 {
2640 struct __sysctl_args tmp;
2641 int error;
2642
2643 if (copy_from_user(&tmp, args, sizeof(tmp)))
2644 return -EFAULT;
2645
2646 error = deprecated_sysctl_warning(&tmp);
2647
2648 /* If no error reading the parameters then just -ENOSYS ... */
2649 if (!error)
2650 error = -ENOSYS;
2651
2652 return error;
2653 }
2654
2655 int sysctl_data(struct ctl_table *table, int __user *name, int nlen,
2656 void __user *oldval, size_t __user *oldlenp,
2657 void __user *newval, size_t newlen)
2658 {
2659 return -ENOSYS;
2660 }
2661
2662 int sysctl_string(struct ctl_table *table, int __user *name, int nlen,
2663 void __user *oldval, size_t __user *oldlenp,
2664 void __user *newval, size_t newlen)
2665 {
2666 return -ENOSYS;
2667 }
2668
2669 int sysctl_intvec(struct ctl_table *table, int __user *name, int nlen,
2670 void __user *oldval, size_t __user *oldlenp,
2671 void __user *newval, size_t newlen)
2672 {
2673 return -ENOSYS;
2674 }
2675
2676 int sysctl_jiffies(struct ctl_table *table, int __user *name, int nlen,
2677 void __user *oldval, size_t __user *oldlenp,
2678 void __user *newval, size_t newlen)
2679 {
2680 return -ENOSYS;
2681 }
2682
2683 int sysctl_ms_jiffies(struct ctl_table *table, int __user *name, int nlen,
2684 void __user *oldval, size_t __user *oldlenp,
2685 void __user *newval, size_t newlen)
2686 {
2687 return -ENOSYS;
2688 }
2689
2690 #endif /* CONFIG_SYSCTL_SYSCALL */
2691
2692 static int deprecated_sysctl_warning(struct __sysctl_args *args)
2693 {
2694 static int msg_count;
2695 int name[CTL_MAXNAME];
2696 int i;
2697
2698 /* Check args->nlen. */
2699 if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
2700 return -ENOTDIR;
2701
2702 /* Read in the sysctl name for better debug message logging */
2703 for (i = 0; i < args->nlen; i++)
2704 if (get_user(name[i], args->name + i))
2705 return -EFAULT;
2706
2707 /* Ignore accesses to kernel.version */
2708 if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2709 return 0;
2710
2711 if (msg_count < 5) {
2712 msg_count++;
2713 printk(KERN_INFO
2714 "warning: process `%s' used the deprecated sysctl "
2715 "system call with ", current->comm);
2716 for (i = 0; i < args->nlen; i++)
2717 printk("%d.", name[i]);
2718 printk("\n");
2719 }
2720 return 0;
2721 }
2722
2723 /*
2724 * No sense putting this after each symbol definition, twice,
2725 * exception granted :-)
2726 */
2727 EXPORT_SYMBOL(proc_dointvec);
2728 EXPORT_SYMBOL(proc_dointvec_jiffies);
2729 EXPORT_SYMBOL(proc_dointvec_minmax);
2730 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2731 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2732 EXPORT_SYMBOL(proc_dostring);
2733 EXPORT_SYMBOL(proc_doulongvec_minmax);
2734 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2735 EXPORT_SYMBOL(register_sysctl_table);
2736 EXPORT_SYMBOL(sysctl_intvec);
2737 EXPORT_SYMBOL(sysctl_jiffies);
2738 EXPORT_SYMBOL(sysctl_ms_jiffies);
2739 EXPORT_SYMBOL(sysctl_string);
2740 EXPORT_SYMBOL(sysctl_data);
2741 EXPORT_SYMBOL(unregister_sysctl_table);