projects / deliverable / linux.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Merge branch 'pxa' into devel
[deliverable/linux.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/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 extern int sysctl_nr_open_min, sysctl_nr_open_max;
85
86 /* Constants used for minimum and maximum */
87 #if defined(CONFIG_DETECT_SOFTLOCKUP) || defined(CONFIG_HIGHMEM)
88 static int one = 1;
89 #endif
90
91 #ifdef CONFIG_DETECT_SOFTLOCKUP
92 static int sixty = 60;
93 #endif
94
95 #ifdef CONFIG_MMU
96 static int two = 2;
97 #endif
98
99 static int zero;
100 static int one_hundred = 100;
101
102 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
103 static int maxolduid = 65535;
104 static int minolduid;
105 static int min_percpu_pagelist_fract = 8;
106
107 static int ngroups_max = NGROUPS_MAX;
108
109 #ifdef CONFIG_KMOD
110 extern char modprobe_path[];
111 #endif
112 #ifdef CONFIG_CHR_DEV_SG
113 extern int sg_big_buff;
114 #endif
115
116 #ifdef __sparc__
117 extern char reboot_command [];
118 extern int stop_a_enabled;
119 extern int scons_pwroff;
120 #endif
121
122 #ifdef __hppa__
123 extern int pwrsw_enabled;
124 extern int unaligned_enabled;
125 #endif
126
127 #ifdef CONFIG_S390
128 #ifdef CONFIG_MATHEMU
129 extern int sysctl_ieee_emulation_warnings;
130 #endif
131 extern int sysctl_userprocess_debug;
132 extern int spin_retry;
133 #endif
134
135 #ifdef CONFIG_BSD_PROCESS_ACCT
136 extern int acct_parm[];
137 #endif
138
139 #ifdef CONFIG_IA64
140 extern int no_unaligned_warning;
141 #endif
142
143 #ifdef CONFIG_RT_MUTEXES
144 extern int max_lock_depth;
145 #endif
146
147 #ifdef CONFIG_PROC_SYSCTL
148 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
149 void __user *buffer, size_t *lenp, loff_t *ppos);
150 static int proc_dointvec_taint(struct ctl_table *table, int write, struct file *filp,
151 void __user *buffer, size_t *lenp, loff_t *ppos);
152 #endif
153
154 static struct ctl_table root_table[];
155 static struct ctl_table_root sysctl_table_root;
156 static struct ctl_table_header root_table_header = {
157 .ctl_table = root_table,
158 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.header_list),
159 .root = &sysctl_table_root,
160 };
161 static struct ctl_table_root sysctl_table_root = {
162 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
163 .header_list = LIST_HEAD_INIT(root_table_header.ctl_entry),
164 };
165
166 static struct ctl_table kern_table[];
167 static struct ctl_table vm_table[];
168 static struct ctl_table fs_table[];
169 static struct ctl_table debug_table[];
170 static struct ctl_table dev_table[];
171 extern struct ctl_table random_table[];
172 #ifdef CONFIG_INOTIFY_USER
173 extern struct ctl_table inotify_table[];
174 #endif
175
176 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
177 int sysctl_legacy_va_layout;
178 #endif
179
180 extern int prove_locking;
181 extern int lock_stat;
182
183 /* The default sysctl tables: */
184
185 static struct ctl_table root_table[] = {
186 {
187 .ctl_name = CTL_KERN,
188 .procname = "kernel",
189 .mode = 0555,
190 .child = kern_table,
191 },
192 {
193 .ctl_name = CTL_VM,
194 .procname = "vm",
195 .mode = 0555,
196 .child = vm_table,
197 },
198 {
199 .ctl_name = CTL_FS,
200 .procname = "fs",
201 .mode = 0555,
202 .child = fs_table,
203 },
204 {
205 .ctl_name = CTL_DEBUG,
206 .procname = "debug",
207 .mode = 0555,
208 .child = debug_table,
209 },
210 {
211 .ctl_name = CTL_DEV,
212 .procname = "dev",
213 .mode = 0555,
214 .child = dev_table,
215 },
216 /*
217 * NOTE: do not add new entries to this table unless you have read
218 * Documentation/sysctl/ctl_unnumbered.txt
219 */
220 { .ctl_name = 0 }
221 };
222
223 #ifdef CONFIG_SCHED_DEBUG
224 static int min_sched_granularity_ns = 100000; /* 100 usecs */
225 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
226 static int min_wakeup_granularity_ns; /* 0 usecs */
227 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
228 #endif
229
230 static struct ctl_table kern_table[] = {
231 #ifdef CONFIG_SCHED_DEBUG
232 {
233 .ctl_name = CTL_UNNUMBERED,
234 .procname = "sched_min_granularity_ns",
235 .data = &sysctl_sched_min_granularity,
236 .maxlen = sizeof(unsigned int),
237 .mode = 0644,
238 .proc_handler = &sched_nr_latency_handler,
239 .strategy = &sysctl_intvec,
240 .extra1 = &min_sched_granularity_ns,
241 .extra2 = &max_sched_granularity_ns,
242 },
243 {
244 .ctl_name = CTL_UNNUMBERED,
245 .procname = "sched_latency_ns",
246 .data = &sysctl_sched_latency,
247 .maxlen = sizeof(unsigned int),
248 .mode = 0644,
249 .proc_handler = &sched_nr_latency_handler,
250 .strategy = &sysctl_intvec,
251 .extra1 = &min_sched_granularity_ns,
252 .extra2 = &max_sched_granularity_ns,
253 },
254 {
255 .ctl_name = CTL_UNNUMBERED,
256 .procname = "sched_wakeup_granularity_ns",
257 .data = &sysctl_sched_wakeup_granularity,
258 .maxlen = sizeof(unsigned int),
259 .mode = 0644,
260 .proc_handler = &proc_dointvec_minmax,
261 .strategy = &sysctl_intvec,
262 .extra1 = &min_wakeup_granularity_ns,
263 .extra2 = &max_wakeup_granularity_ns,
264 },
265 {
266 .ctl_name = CTL_UNNUMBERED,
267 .procname = "sched_child_runs_first",
268 .data = &sysctl_sched_child_runs_first,
269 .maxlen = sizeof(unsigned int),
270 .mode = 0644,
271 .proc_handler = &proc_dointvec,
272 },
273 {
274 .ctl_name = CTL_UNNUMBERED,
275 .procname = "sched_features",
276 .data = &sysctl_sched_features,
277 .maxlen = sizeof(unsigned int),
278 .mode = 0644,
279 .proc_handler = &proc_dointvec,
280 },
281 {
282 .ctl_name = CTL_UNNUMBERED,
283 .procname = "sched_migration_cost",
284 .data = &sysctl_sched_migration_cost,
285 .maxlen = sizeof(unsigned int),
286 .mode = 0644,
287 .proc_handler = &proc_dointvec,
288 },
289 {
290 .ctl_name = CTL_UNNUMBERED,
291 .procname = "sched_nr_migrate",
292 .data = &sysctl_sched_nr_migrate,
293 .maxlen = sizeof(unsigned int),
294 .mode = 0644,
295 .proc_handler = &proc_dointvec,
296 },
297 #endif
298 {
299 .ctl_name = CTL_UNNUMBERED,
300 .procname = "sched_rt_period_us",
301 .data = &sysctl_sched_rt_period,
302 .maxlen = sizeof(unsigned int),
303 .mode = 0644,
304 .proc_handler = &sched_rt_handler,
305 },
306 {
307 .ctl_name = CTL_UNNUMBERED,
308 .procname = "sched_rt_runtime_us",
309 .data = &sysctl_sched_rt_runtime,
310 .maxlen = sizeof(int),
311 .mode = 0644,
312 .proc_handler = &sched_rt_handler,
313 },
314 {
315 .ctl_name = CTL_UNNUMBERED,
316 .procname = "sched_compat_yield",
317 .data = &sysctl_sched_compat_yield,
318 .maxlen = sizeof(unsigned int),
319 .mode = 0644,
320 .proc_handler = &proc_dointvec,
321 },
322 #ifdef CONFIG_PROVE_LOCKING
323 {
324 .ctl_name = CTL_UNNUMBERED,
325 .procname = "prove_locking",
326 .data = &prove_locking,
327 .maxlen = sizeof(int),
328 .mode = 0644,
329 .proc_handler = &proc_dointvec,
330 },
331 #endif
332 #ifdef CONFIG_LOCK_STAT
333 {
334 .ctl_name = CTL_UNNUMBERED,
335 .procname = "lock_stat",
336 .data = &lock_stat,
337 .maxlen = sizeof(int),
338 .mode = 0644,
339 .proc_handler = &proc_dointvec,
340 },
341 #endif
342 {
343 .ctl_name = KERN_PANIC,
344 .procname = "panic",
345 .data = &panic_timeout,
346 .maxlen = sizeof(int),
347 .mode = 0644,
348 .proc_handler = &proc_dointvec,
349 },
350 {
351 .ctl_name = KERN_CORE_USES_PID,
352 .procname = "core_uses_pid",
353 .data = &core_uses_pid,
354 .maxlen = sizeof(int),
355 .mode = 0644,
356 .proc_handler = &proc_dointvec,
357 },
358 {
359 .ctl_name = KERN_CORE_PATTERN,
360 .procname = "core_pattern",
361 .data = core_pattern,
362 .maxlen = CORENAME_MAX_SIZE,
363 .mode = 0644,
364 .proc_handler = &proc_dostring,
365 .strategy = &sysctl_string,
366 },
367 #ifdef CONFIG_PROC_SYSCTL
368 {
369 .procname = "tainted",
370 .data = &tainted,
371 .maxlen = sizeof(int),
372 .mode = 0644,
373 .proc_handler = &proc_dointvec_taint,
374 },
375 #endif
376 #ifdef CONFIG_LATENCYTOP
377 {
378 .procname = "latencytop",
379 .data = &latencytop_enabled,
380 .maxlen = sizeof(int),
381 .mode = 0644,
382 .proc_handler = &proc_dointvec,
383 },
384 #endif
385 #ifdef CONFIG_BLK_DEV_INITRD
386 {
387 .ctl_name = KERN_REALROOTDEV,
388 .procname = "real-root-dev",
389 .data = &real_root_dev,
390 .maxlen = sizeof(int),
391 .mode = 0644,
392 .proc_handler = &proc_dointvec,
393 },
394 #endif
395 {
396 .ctl_name = CTL_UNNUMBERED,
397 .procname = "print-fatal-signals",
398 .data = &print_fatal_signals,
399 .maxlen = sizeof(int),
400 .mode = 0644,
401 .proc_handler = &proc_dointvec,
402 },
403 #ifdef __sparc__
404 {
405 .ctl_name = KERN_SPARC_REBOOT,
406 .procname = "reboot-cmd",
407 .data = reboot_command,
408 .maxlen = 256,
409 .mode = 0644,
410 .proc_handler = &proc_dostring,
411 .strategy = &sysctl_string,
412 },
413 {
414 .ctl_name = KERN_SPARC_STOP_A,
415 .procname = "stop-a",
416 .data = &stop_a_enabled,
417 .maxlen = sizeof (int),
418 .mode = 0644,
419 .proc_handler = &proc_dointvec,
420 },
421 {
422 .ctl_name = KERN_SPARC_SCONS_PWROFF,
423 .procname = "scons-poweroff",
424 .data = &scons_pwroff,
425 .maxlen = sizeof (int),
426 .mode = 0644,
427 .proc_handler = &proc_dointvec,
428 },
429 #endif
430 #ifdef __hppa__
431 {
432 .ctl_name = KERN_HPPA_PWRSW,
433 .procname = "soft-power",
434 .data = &pwrsw_enabled,
435 .maxlen = sizeof (int),
436 .mode = 0644,
437 .proc_handler = &proc_dointvec,
438 },
439 {
440 .ctl_name = KERN_HPPA_UNALIGNED,
441 .procname = "unaligned-trap",
442 .data = &unaligned_enabled,
443 .maxlen = sizeof (int),
444 .mode = 0644,
445 .proc_handler = &proc_dointvec,
446 },
447 #endif
448 {
449 .ctl_name = KERN_CTLALTDEL,
450 .procname = "ctrl-alt-del",
451 .data = &C_A_D,
452 .maxlen = sizeof(int),
453 .mode = 0644,
454 .proc_handler = &proc_dointvec,
455 },
456 #ifdef CONFIG_KMOD
457 {
458 .ctl_name = KERN_MODPROBE,
459 .procname = "modprobe",
460 .data = &modprobe_path,
461 .maxlen = KMOD_PATH_LEN,
462 .mode = 0644,
463 .proc_handler = &proc_dostring,
464 .strategy = &sysctl_string,
465 },
466 #endif
467 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
468 {
469 .ctl_name = KERN_HOTPLUG,
470 .procname = "hotplug",
471 .data = &uevent_helper,
472 .maxlen = UEVENT_HELPER_PATH_LEN,
473 .mode = 0644,
474 .proc_handler = &proc_dostring,
475 .strategy = &sysctl_string,
476 },
477 #endif
478 #ifdef CONFIG_CHR_DEV_SG
479 {
480 .ctl_name = KERN_SG_BIG_BUFF,
481 .procname = "sg-big-buff",
482 .data = &sg_big_buff,
483 .maxlen = sizeof (int),
484 .mode = 0444,
485 .proc_handler = &proc_dointvec,
486 },
487 #endif
488 #ifdef CONFIG_BSD_PROCESS_ACCT
489 {
490 .ctl_name = KERN_ACCT,
491 .procname = "acct",
492 .data = &acct_parm,
493 .maxlen = 3*sizeof(int),
494 .mode = 0644,
495 .proc_handler = &proc_dointvec,
496 },
497 #endif
498 #ifdef CONFIG_MAGIC_SYSRQ
499 {
500 .ctl_name = KERN_SYSRQ,
501 .procname = "sysrq",
502 .data = &__sysrq_enabled,
503 .maxlen = sizeof (int),
504 .mode = 0644,
505 .proc_handler = &proc_dointvec,
506 },
507 #endif
508 #ifdef CONFIG_PROC_SYSCTL
509 {
510 .procname = "cad_pid",
511 .data = NULL,
512 .maxlen = sizeof (int),
513 .mode = 0600,
514 .proc_handler = &proc_do_cad_pid,
515 },
516 #endif
517 {
518 .ctl_name = KERN_MAX_THREADS,
519 .procname = "threads-max",
520 .data = &max_threads,
521 .maxlen = sizeof(int),
522 .mode = 0644,
523 .proc_handler = &proc_dointvec,
524 },
525 {
526 .ctl_name = KERN_RANDOM,
527 .procname = "random",
528 .mode = 0555,
529 .child = random_table,
530 },
531 {
532 .ctl_name = KERN_OVERFLOWUID,
533 .procname = "overflowuid",
534 .data = &overflowuid,
535 .maxlen = sizeof(int),
536 .mode = 0644,
537 .proc_handler = &proc_dointvec_minmax,
538 .strategy = &sysctl_intvec,
539 .extra1 = &minolduid,
540 .extra2 = &maxolduid,
541 },
542 {
543 .ctl_name = KERN_OVERFLOWGID,
544 .procname = "overflowgid",
545 .data = &overflowgid,
546 .maxlen = sizeof(int),
547 .mode = 0644,
548 .proc_handler = &proc_dointvec_minmax,
549 .strategy = &sysctl_intvec,
550 .extra1 = &minolduid,
551 .extra2 = &maxolduid,
552 },
553 #ifdef CONFIG_S390
554 #ifdef CONFIG_MATHEMU
555 {
556 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
557 .procname = "ieee_emulation_warnings",
558 .data = &sysctl_ieee_emulation_warnings,
559 .maxlen = sizeof(int),
560 .mode = 0644,
561 .proc_handler = &proc_dointvec,
562 },
563 #endif
564 {
565 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
566 .procname = "userprocess_debug",
567 .data = &sysctl_userprocess_debug,
568 .maxlen = sizeof(int),
569 .mode = 0644,
570 .proc_handler = &proc_dointvec,
571 },
572 #endif
573 {
574 .ctl_name = KERN_PIDMAX,
575 .procname = "pid_max",
576 .data = &pid_max,
577 .maxlen = sizeof (int),
578 .mode = 0644,
579 .proc_handler = &proc_dointvec_minmax,
580 .strategy = sysctl_intvec,
581 .extra1 = &pid_max_min,
582 .extra2 = &pid_max_max,
583 },
584 {
585 .ctl_name = KERN_PANIC_ON_OOPS,
586 .procname = "panic_on_oops",
587 .data = &panic_on_oops,
588 .maxlen = sizeof(int),
589 .mode = 0644,
590 .proc_handler = &proc_dointvec,
591 },
592 #if defined CONFIG_PRINTK
593 {
594 .ctl_name = KERN_PRINTK,
595 .procname = "printk",
596 .data = &console_loglevel,
597 .maxlen = 4*sizeof(int),
598 .mode = 0644,
599 .proc_handler = &proc_dointvec,
600 },
601 {
602 .ctl_name = KERN_PRINTK_RATELIMIT,
603 .procname = "printk_ratelimit",
604 .data = &printk_ratelimit_jiffies,
605 .maxlen = sizeof(int),
606 .mode = 0644,
607 .proc_handler = &proc_dointvec_jiffies,
608 .strategy = &sysctl_jiffies,
609 },
610 {
611 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
612 .procname = "printk_ratelimit_burst",
613 .data = &printk_ratelimit_burst,
614 .maxlen = sizeof(int),
615 .mode = 0644,
616 .proc_handler = &proc_dointvec,
617 },
618 #endif
619 {
620 .ctl_name = KERN_NGROUPS_MAX,
621 .procname = "ngroups_max",
622 .data = &ngroups_max,
623 .maxlen = sizeof (int),
624 .mode = 0444,
625 .proc_handler = &proc_dointvec,
626 },
627 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
628 {
629 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
630 .procname = "unknown_nmi_panic",
631 .data = &unknown_nmi_panic,
632 .maxlen = sizeof (int),
633 .mode = 0644,
634 .proc_handler = &proc_dointvec,
635 },
636 {
637 .procname = "nmi_watchdog",
638 .data = &nmi_watchdog_enabled,
639 .maxlen = sizeof (int),
640 .mode = 0644,
641 .proc_handler = &proc_nmi_enabled,
642 },
643 #endif
644 #if defined(CONFIG_X86)
645 {
646 .ctl_name = KERN_PANIC_ON_NMI,
647 .procname = "panic_on_unrecovered_nmi",
648 .data = &panic_on_unrecovered_nmi,
649 .maxlen = sizeof(int),
650 .mode = 0644,
651 .proc_handler = &proc_dointvec,
652 },
653 {
654 .ctl_name = KERN_BOOTLOADER_TYPE,
655 .procname = "bootloader_type",
656 .data = &bootloader_type,
657 .maxlen = sizeof (int),
658 .mode = 0444,
659 .proc_handler = &proc_dointvec,
660 },
661 {
662 .ctl_name = CTL_UNNUMBERED,
663 .procname = "kstack_depth_to_print",
664 .data = &kstack_depth_to_print,
665 .maxlen = sizeof(int),
666 .mode = 0644,
667 .proc_handler = &proc_dointvec,
668 },
669 {
670 .ctl_name = CTL_UNNUMBERED,
671 .procname = "io_delay_type",
672 .data = &io_delay_type,
673 .maxlen = sizeof(int),
674 .mode = 0644,
675 .proc_handler = &proc_dointvec,
676 },
677 #endif
678 #if defined(CONFIG_MMU)
679 {
680 .ctl_name = KERN_RANDOMIZE,
681 .procname = "randomize_va_space",
682 .data = &randomize_va_space,
683 .maxlen = sizeof(int),
684 .mode = 0644,
685 .proc_handler = &proc_dointvec,
686 },
687 #endif
688 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
689 {
690 .ctl_name = KERN_SPIN_RETRY,
691 .procname = "spin_retry",
692 .data = &spin_retry,
693 .maxlen = sizeof (int),
694 .mode = 0644,
695 .proc_handler = &proc_dointvec,
696 },
697 #endif
698 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
699 {
700 .procname = "acpi_video_flags",
701 .data = &acpi_realmode_flags,
702 .maxlen = sizeof (unsigned long),
703 .mode = 0644,
704 .proc_handler = &proc_doulongvec_minmax,
705 },
706 #endif
707 #ifdef CONFIG_IA64
708 {
709 .ctl_name = KERN_IA64_UNALIGNED,
710 .procname = "ignore-unaligned-usertrap",
711 .data = &no_unaligned_warning,
712 .maxlen = sizeof (int),
713 .mode = 0644,
714 .proc_handler = &proc_dointvec,
715 },
716 #endif
717 #ifdef CONFIG_DETECT_SOFTLOCKUP
718 {
719 .ctl_name = CTL_UNNUMBERED,
720 .procname = "softlockup_thresh",
721 .data = &softlockup_thresh,
722 .maxlen = sizeof(unsigned long),
723 .mode = 0644,
724 .proc_handler = &proc_doulongvec_minmax,
725 .strategy = &sysctl_intvec,
726 .extra1 = &one,
727 .extra2 = &sixty,
728 },
729 {
730 .ctl_name = CTL_UNNUMBERED,
731 .procname = "hung_task_check_count",
732 .data = &sysctl_hung_task_check_count,
733 .maxlen = sizeof(unsigned long),
734 .mode = 0644,
735 .proc_handler = &proc_doulongvec_minmax,
736 .strategy = &sysctl_intvec,
737 },
738 {
739 .ctl_name = CTL_UNNUMBERED,
740 .procname = "hung_task_timeout_secs",
741 .data = &sysctl_hung_task_timeout_secs,
742 .maxlen = sizeof(unsigned long),
743 .mode = 0644,
744 .proc_handler = &proc_doulongvec_minmax,
745 .strategy = &sysctl_intvec,
746 },
747 {
748 .ctl_name = CTL_UNNUMBERED,
749 .procname = "hung_task_warnings",
750 .data = &sysctl_hung_task_warnings,
751 .maxlen = sizeof(unsigned long),
752 .mode = 0644,
753 .proc_handler = &proc_doulongvec_minmax,
754 .strategy = &sysctl_intvec,
755 },
756 #endif
757 #ifdef CONFIG_COMPAT
758 {
759 .ctl_name = KERN_COMPAT_LOG,
760 .procname = "compat-log",
761 .data = &compat_log,
762 .maxlen = sizeof (int),
763 .mode = 0644,
764 .proc_handler = &proc_dointvec,
765 },
766 #endif
767 #ifdef CONFIG_RT_MUTEXES
768 {
769 .ctl_name = KERN_MAX_LOCK_DEPTH,
770 .procname = "max_lock_depth",
771 .data = &max_lock_depth,
772 .maxlen = sizeof(int),
773 .mode = 0644,
774 .proc_handler = &proc_dointvec,
775 },
776 #endif
777 #ifdef CONFIG_PROC_FS
778 {
779 .ctl_name = CTL_UNNUMBERED,
780 .procname = "maps_protect",
781 .data = &maps_protect,
782 .maxlen = sizeof(int),
783 .mode = 0644,
784 .proc_handler = &proc_dointvec,
785 },
786 #endif
787 {
788 .ctl_name = CTL_UNNUMBERED,
789 .procname = "poweroff_cmd",
790 .data = &poweroff_cmd,
791 .maxlen = POWEROFF_CMD_PATH_LEN,
792 .mode = 0644,
793 .proc_handler = &proc_dostring,
794 .strategy = &sysctl_string,
795 },
796 #ifdef CONFIG_KEYS
797 {
798 .ctl_name = CTL_UNNUMBERED,
799 .procname = "keys",
800 .mode = 0555,
801 .child = key_sysctls,
802 },
803 #endif
804 /*
805 * NOTE: do not add new entries to this table unless you have read
806 * Documentation/sysctl/ctl_unnumbered.txt
807 */
808 { .ctl_name = 0 }
809 };
810
811 static struct ctl_table vm_table[] = {
812 {
813 .ctl_name = VM_OVERCOMMIT_MEMORY,
814 .procname = "overcommit_memory",
815 .data = &sysctl_overcommit_memory,
816 .maxlen = sizeof(sysctl_overcommit_memory),
817 .mode = 0644,
818 .proc_handler = &proc_dointvec,
819 },
820 {
821 .ctl_name = VM_PANIC_ON_OOM,
822 .procname = "panic_on_oom",
823 .data = &sysctl_panic_on_oom,
824 .maxlen = sizeof(sysctl_panic_on_oom),
825 .mode = 0644,
826 .proc_handler = &proc_dointvec,
827 },
828 {
829 .ctl_name = CTL_UNNUMBERED,
830 .procname = "oom_kill_allocating_task",
831 .data = &sysctl_oom_kill_allocating_task,
832 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
833 .mode = 0644,
834 .proc_handler = &proc_dointvec,
835 },
836 {
837 .ctl_name = CTL_UNNUMBERED,
838 .procname = "oom_dump_tasks",
839 .data = &sysctl_oom_dump_tasks,
840 .maxlen = sizeof(sysctl_oom_dump_tasks),
841 .mode = 0644,
842 .proc_handler = &proc_dointvec,
843 },
844 {
845 .ctl_name = VM_OVERCOMMIT_RATIO,
846 .procname = "overcommit_ratio",
847 .data = &sysctl_overcommit_ratio,
848 .maxlen = sizeof(sysctl_overcommit_ratio),
849 .mode = 0644,
850 .proc_handler = &proc_dointvec,
851 },
852 {
853 .ctl_name = VM_PAGE_CLUSTER,
854 .procname = "page-cluster",
855 .data = &page_cluster,
856 .maxlen = sizeof(int),
857 .mode = 0644,
858 .proc_handler = &proc_dointvec,
859 },
860 {
861 .ctl_name = VM_DIRTY_BACKGROUND,
862 .procname = "dirty_background_ratio",
863 .data = &dirty_background_ratio,
864 .maxlen = sizeof(dirty_background_ratio),
865 .mode = 0644,
866 .proc_handler = &proc_dointvec_minmax,
867 .strategy = &sysctl_intvec,
868 .extra1 = &zero,
869 .extra2 = &one_hundred,
870 },
871 {
872 .ctl_name = VM_DIRTY_RATIO,
873 .procname = "dirty_ratio",
874 .data = &vm_dirty_ratio,
875 .maxlen = sizeof(vm_dirty_ratio),
876 .mode = 0644,
877 .proc_handler = &dirty_ratio_handler,
878 .strategy = &sysctl_intvec,
879 .extra1 = &zero,
880 .extra2 = &one_hundred,
881 },
882 {
883 .procname = "dirty_writeback_centisecs",
884 .data = &dirty_writeback_interval,
885 .maxlen = sizeof(dirty_writeback_interval),
886 .mode = 0644,
887 .proc_handler = &dirty_writeback_centisecs_handler,
888 },
889 {
890 .procname = "dirty_expire_centisecs",
891 .data = &dirty_expire_interval,
892 .maxlen = sizeof(dirty_expire_interval),
893 .mode = 0644,
894 .proc_handler = &proc_dointvec_userhz_jiffies,
895 },
896 {
897 .ctl_name = VM_NR_PDFLUSH_THREADS,
898 .procname = "nr_pdflush_threads",
899 .data = &nr_pdflush_threads,
900 .maxlen = sizeof nr_pdflush_threads,
901 .mode = 0444 /* read-only*/,
902 .proc_handler = &proc_dointvec,
903 },
904 {
905 .ctl_name = VM_SWAPPINESS,
906 .procname = "swappiness",
907 .data = &vm_swappiness,
908 .maxlen = sizeof(vm_swappiness),
909 .mode = 0644,
910 .proc_handler = &proc_dointvec_minmax,
911 .strategy = &sysctl_intvec,
912 .extra1 = &zero,
913 .extra2 = &one_hundred,
914 },
915 #ifdef CONFIG_HUGETLB_PAGE
916 {
917 .procname = "nr_hugepages",
918 .data = &max_huge_pages,
919 .maxlen = sizeof(unsigned long),
920 .mode = 0644,
921 .proc_handler = &hugetlb_sysctl_handler,
922 .extra1 = (void *)&hugetlb_zero,
923 .extra2 = (void *)&hugetlb_infinity,
924 },
925 {
926 .ctl_name = VM_HUGETLB_GROUP,
927 .procname = "hugetlb_shm_group",
928 .data = &sysctl_hugetlb_shm_group,
929 .maxlen = sizeof(gid_t),
930 .mode = 0644,
931 .proc_handler = &proc_dointvec,
932 },
933 {
934 .ctl_name = CTL_UNNUMBERED,
935 .procname = "hugepages_treat_as_movable",
936 .data = &hugepages_treat_as_movable,
937 .maxlen = sizeof(int),
938 .mode = 0644,
939 .proc_handler = &hugetlb_treat_movable_handler,
940 },
941 {
942 .ctl_name = CTL_UNNUMBERED,
943 .procname = "nr_overcommit_hugepages",
944 .data = &sysctl_overcommit_huge_pages,
945 .maxlen = sizeof(sysctl_overcommit_huge_pages),
946 .mode = 0644,
947 .proc_handler = &hugetlb_overcommit_handler,
948 },
949 #endif
950 {
951 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
952 .procname = "lowmem_reserve_ratio",
953 .data = &sysctl_lowmem_reserve_ratio,
954 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
955 .mode = 0644,
956 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
957 .strategy = &sysctl_intvec,
958 },
959 {
960 .ctl_name = VM_DROP_PAGECACHE,
961 .procname = "drop_caches",
962 .data = &sysctl_drop_caches,
963 .maxlen = sizeof(int),
964 .mode = 0644,
965 .proc_handler = drop_caches_sysctl_handler,
966 .strategy = &sysctl_intvec,
967 },
968 {
969 .ctl_name = VM_MIN_FREE_KBYTES,
970 .procname = "min_free_kbytes",
971 .data = &min_free_kbytes,
972 .maxlen = sizeof(min_free_kbytes),
973 .mode = 0644,
974 .proc_handler = &min_free_kbytes_sysctl_handler,
975 .strategy = &sysctl_intvec,
976 .extra1 = &zero,
977 },
978 {
979 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
980 .procname = "percpu_pagelist_fraction",
981 .data = &percpu_pagelist_fraction,
982 .maxlen = sizeof(percpu_pagelist_fraction),
983 .mode = 0644,
984 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
985 .strategy = &sysctl_intvec,
986 .extra1 = &min_percpu_pagelist_fract,
987 },
988 #ifdef CONFIG_MMU
989 {
990 .ctl_name = VM_MAX_MAP_COUNT,
991 .procname = "max_map_count",
992 .data = &sysctl_max_map_count,
993 .maxlen = sizeof(sysctl_max_map_count),
994 .mode = 0644,
995 .proc_handler = &proc_dointvec
996 },
997 #endif
998 {
999 .ctl_name = VM_LAPTOP_MODE,
1000 .procname = "laptop_mode",
1001 .data = &laptop_mode,
1002 .maxlen = sizeof(laptop_mode),
1003 .mode = 0644,
1004 .proc_handler = &proc_dointvec_jiffies,
1005 .strategy = &sysctl_jiffies,
1006 },
1007 {
1008 .ctl_name = VM_BLOCK_DUMP,
1009 .procname = "block_dump",
1010 .data = &block_dump,
1011 .maxlen = sizeof(block_dump),
1012 .mode = 0644,
1013 .proc_handler = &proc_dointvec,
1014 .strategy = &sysctl_intvec,
1015 .extra1 = &zero,
1016 },
1017 {
1018 .ctl_name = VM_VFS_CACHE_PRESSURE,
1019 .procname = "vfs_cache_pressure",
1020 .data = &sysctl_vfs_cache_pressure,
1021 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1022 .mode = 0644,
1023 .proc_handler = &proc_dointvec,
1024 .strategy = &sysctl_intvec,
1025 .extra1 = &zero,
1026 },
1027 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1028 {
1029 .ctl_name = VM_LEGACY_VA_LAYOUT,
1030 .procname = "legacy_va_layout",
1031 .data = &sysctl_legacy_va_layout,
1032 .maxlen = sizeof(sysctl_legacy_va_layout),
1033 .mode = 0644,
1034 .proc_handler = &proc_dointvec,
1035 .strategy = &sysctl_intvec,
1036 .extra1 = &zero,
1037 },
1038 #endif
1039 #ifdef CONFIG_NUMA
1040 {
1041 .ctl_name = VM_ZONE_RECLAIM_MODE,
1042 .procname = "zone_reclaim_mode",
1043 .data = &zone_reclaim_mode,
1044 .maxlen = sizeof(zone_reclaim_mode),
1045 .mode = 0644,
1046 .proc_handler = &proc_dointvec,
1047 .strategy = &sysctl_intvec,
1048 .extra1 = &zero,
1049 },
1050 {
1051 .ctl_name = VM_MIN_UNMAPPED,
1052 .procname = "min_unmapped_ratio",
1053 .data = &sysctl_min_unmapped_ratio,
1054 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1055 .mode = 0644,
1056 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
1057 .strategy = &sysctl_intvec,
1058 .extra1 = &zero,
1059 .extra2 = &one_hundred,
1060 },
1061 {
1062 .ctl_name = VM_MIN_SLAB,
1063 .procname = "min_slab_ratio",
1064 .data = &sysctl_min_slab_ratio,
1065 .maxlen = sizeof(sysctl_min_slab_ratio),
1066 .mode = 0644,
1067 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1068 .strategy = &sysctl_intvec,
1069 .extra1 = &zero,
1070 .extra2 = &one_hundred,
1071 },
1072 #endif
1073 #ifdef CONFIG_SMP
1074 {
1075 .ctl_name = CTL_UNNUMBERED,
1076 .procname = "stat_interval",
1077 .data = &sysctl_stat_interval,
1078 .maxlen = sizeof(sysctl_stat_interval),
1079 .mode = 0644,
1080 .proc_handler = &proc_dointvec_jiffies,
1081 .strategy = &sysctl_jiffies,
1082 },
1083 #endif
1084 #ifdef CONFIG_SECURITY
1085 {
1086 .ctl_name = CTL_UNNUMBERED,
1087 .procname = "mmap_min_addr",
1088 .data = &mmap_min_addr,
1089 .maxlen = sizeof(unsigned long),
1090 .mode = 0644,
1091 .proc_handler = &proc_doulongvec_minmax,
1092 },
1093 #endif
1094 #ifdef CONFIG_NUMA
1095 {
1096 .ctl_name = CTL_UNNUMBERED,
1097 .procname = "numa_zonelist_order",
1098 .data = &numa_zonelist_order,
1099 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1100 .mode = 0644,
1101 .proc_handler = &numa_zonelist_order_handler,
1102 .strategy = &sysctl_string,
1103 },
1104 #endif
1105 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1106 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1107 {
1108 .ctl_name = VM_VDSO_ENABLED,
1109 .procname = "vdso_enabled",
1110 .data = &vdso_enabled,
1111 .maxlen = sizeof(vdso_enabled),
1112 .mode = 0644,
1113 .proc_handler = &proc_dointvec,
1114 .strategy = &sysctl_intvec,
1115 .extra1 = &zero,
1116 },
1117 #endif
1118 #ifdef CONFIG_HIGHMEM
1119 {
1120 .ctl_name = CTL_UNNUMBERED,
1121 .procname = "highmem_is_dirtyable",
1122 .data = &vm_highmem_is_dirtyable,
1123 .maxlen = sizeof(vm_highmem_is_dirtyable),
1124 .mode = 0644,
1125 .proc_handler = &proc_dointvec_minmax,
1126 .strategy = &sysctl_intvec,
1127 .extra1 = &zero,
1128 .extra2 = &one,
1129 },
1130 #endif
1131 /*
1132 * NOTE: do not add new entries to this table unless you have read
1133 * Documentation/sysctl/ctl_unnumbered.txt
1134 */
1135 { .ctl_name = 0 }
1136 };
1137
1138 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1139 static struct ctl_table binfmt_misc_table[] = {
1140 { .ctl_name = 0 }
1141 };
1142 #endif
1143
1144 static struct ctl_table fs_table[] = {
1145 {
1146 .ctl_name = FS_NRINODE,
1147 .procname = "inode-nr",
1148 .data = &inodes_stat,
1149 .maxlen = 2*sizeof(int),
1150 .mode = 0444,
1151 .proc_handler = &proc_dointvec,
1152 },
1153 {
1154 .ctl_name = FS_STATINODE,
1155 .procname = "inode-state",
1156 .data = &inodes_stat,
1157 .maxlen = 7*sizeof(int),
1158 .mode = 0444,
1159 .proc_handler = &proc_dointvec,
1160 },
1161 {
1162 .procname = "file-nr",
1163 .data = &files_stat,
1164 .maxlen = 3*sizeof(int),
1165 .mode = 0444,
1166 .proc_handler = &proc_nr_files,
1167 },
1168 {
1169 .ctl_name = FS_MAXFILE,
1170 .procname = "file-max",
1171 .data = &files_stat.max_files,
1172 .maxlen = sizeof(int),
1173 .mode = 0644,
1174 .proc_handler = &proc_dointvec,
1175 },
1176 {
1177 .ctl_name = CTL_UNNUMBERED,
1178 .procname = "nr_open",
1179 .data = &sysctl_nr_open,
1180 .maxlen = sizeof(int),
1181 .mode = 0644,
1182 .proc_handler = &proc_dointvec_minmax,
1183 .extra1 = &sysctl_nr_open_min,
1184 .extra2 = &sysctl_nr_open_max,
1185 },
1186 {
1187 .ctl_name = FS_DENTRY,
1188 .procname = "dentry-state",
1189 .data = &dentry_stat,
1190 .maxlen = 6*sizeof(int),
1191 .mode = 0444,
1192 .proc_handler = &proc_dointvec,
1193 },
1194 {
1195 .ctl_name = FS_OVERFLOWUID,
1196 .procname = "overflowuid",
1197 .data = &fs_overflowuid,
1198 .maxlen = sizeof(int),
1199 .mode = 0644,
1200 .proc_handler = &proc_dointvec_minmax,
1201 .strategy = &sysctl_intvec,
1202 .extra1 = &minolduid,
1203 .extra2 = &maxolduid,
1204 },
1205 {
1206 .ctl_name = FS_OVERFLOWGID,
1207 .procname = "overflowgid",
1208 .data = &fs_overflowgid,
1209 .maxlen = sizeof(int),
1210 .mode = 0644,
1211 .proc_handler = &proc_dointvec_minmax,
1212 .strategy = &sysctl_intvec,
1213 .extra1 = &minolduid,
1214 .extra2 = &maxolduid,
1215 },
1216 {
1217 .ctl_name = FS_LEASES,
1218 .procname = "leases-enable",
1219 .data = &leases_enable,
1220 .maxlen = sizeof(int),
1221 .mode = 0644,
1222 .proc_handler = &proc_dointvec,
1223 },
1224 #ifdef CONFIG_DNOTIFY
1225 {
1226 .ctl_name = FS_DIR_NOTIFY,
1227 .procname = "dir-notify-enable",
1228 .data = &dir_notify_enable,
1229 .maxlen = sizeof(int),
1230 .mode = 0644,
1231 .proc_handler = &proc_dointvec,
1232 },
1233 #endif
1234 #ifdef CONFIG_MMU
1235 {
1236 .ctl_name = FS_LEASE_TIME,
1237 .procname = "lease-break-time",
1238 .data = &lease_break_time,
1239 .maxlen = sizeof(int),
1240 .mode = 0644,
1241 .proc_handler = &proc_dointvec_minmax,
1242 .strategy = &sysctl_intvec,
1243 .extra1 = &zero,
1244 .extra2 = &two,
1245 },
1246 {
1247 .procname = "aio-nr",
1248 .data = &aio_nr,
1249 .maxlen = sizeof(aio_nr),
1250 .mode = 0444,
1251 .proc_handler = &proc_doulongvec_minmax,
1252 },
1253 {
1254 .procname = "aio-max-nr",
1255 .data = &aio_max_nr,
1256 .maxlen = sizeof(aio_max_nr),
1257 .mode = 0644,
1258 .proc_handler = &proc_doulongvec_minmax,
1259 },
1260 #ifdef CONFIG_INOTIFY_USER
1261 {
1262 .ctl_name = FS_INOTIFY,
1263 .procname = "inotify",
1264 .mode = 0555,
1265 .child = inotify_table,
1266 },
1267 #endif
1268 #endif
1269 {
1270 .ctl_name = KERN_SETUID_DUMPABLE,
1271 .procname = "suid_dumpable",
1272 .data = &suid_dumpable,
1273 .maxlen = sizeof(int),
1274 .mode = 0644,
1275 .proc_handler = &proc_dointvec,
1276 },
1277 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1278 {
1279 .ctl_name = CTL_UNNUMBERED,
1280 .procname = "binfmt_misc",
1281 .mode = 0555,
1282 .child = binfmt_misc_table,
1283 },
1284 #endif
1285 /*
1286 * NOTE: do not add new entries to this table unless you have read
1287 * Documentation/sysctl/ctl_unnumbered.txt
1288 */
1289 { .ctl_name = 0 }
1290 };
1291
1292 static struct ctl_table debug_table[] = {
1293 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1294 {
1295 .ctl_name = CTL_UNNUMBERED,
1296 .procname = "exception-trace",
1297 .data = &show_unhandled_signals,
1298 .maxlen = sizeof(int),
1299 .mode = 0644,
1300 .proc_handler = proc_dointvec
1301 },
1302 #endif
1303 { .ctl_name = 0 }
1304 };
1305
1306 static struct ctl_table dev_table[] = {
1307 { .ctl_name = 0 }
1308 };
1309
1310 static DEFINE_SPINLOCK(sysctl_lock);
1311
1312 /* called under sysctl_lock */
1313 static int use_table(struct ctl_table_header *p)
1314 {
1315 if (unlikely(p->unregistering))
1316 return 0;
1317 p->used++;
1318 return 1;
1319 }
1320
1321 /* called under sysctl_lock */
1322 static void unuse_table(struct ctl_table_header *p)
1323 {
1324 if (!--p->used)
1325 if (unlikely(p->unregistering))
1326 complete(p->unregistering);
1327 }
1328
1329 /* called under sysctl_lock, will reacquire if has to wait */
1330 static void start_unregistering(struct ctl_table_header *p)
1331 {
1332 /*
1333 * if p->used is 0, nobody will ever touch that entry again;
1334 * we'll eliminate all paths to it before dropping sysctl_lock
1335 */
1336 if (unlikely(p->used)) {
1337 struct completion wait;
1338 init_completion(&wait);
1339 p->unregistering = &wait;
1340 spin_unlock(&sysctl_lock);
1341 wait_for_completion(&wait);
1342 spin_lock(&sysctl_lock);
1343 }
1344 /*
1345 * do not remove from the list until nobody holds it; walking the
1346 * list in do_sysctl() relies on that.
1347 */
1348 list_del_init(&p->ctl_entry);
1349 }
1350
1351 void sysctl_head_finish(struct ctl_table_header *head)
1352 {
1353 if (!head)
1354 return;
1355 spin_lock(&sysctl_lock);
1356 unuse_table(head);
1357 spin_unlock(&sysctl_lock);
1358 }
1359
1360 static struct list_head *
1361 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1362 {
1363 struct list_head *header_list;
1364 header_list = &root->header_list;
1365 if (root->lookup)
1366 header_list = root->lookup(root, namespaces);
1367 return header_list;
1368 }
1369
1370 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1371 struct ctl_table_header *prev)
1372 {
1373 struct ctl_table_root *root;
1374 struct list_head *header_list;
1375 struct ctl_table_header *head;
1376 struct list_head *tmp;
1377
1378 spin_lock(&sysctl_lock);
1379 if (prev) {
1380 head = prev;
1381 tmp = &prev->ctl_entry;
1382 unuse_table(prev);
1383 goto next;
1384 }
1385 tmp = &root_table_header.ctl_entry;
1386 for (;;) {
1387 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1388
1389 if (!use_table(head))
1390 goto next;
1391 spin_unlock(&sysctl_lock);
1392 return head;
1393 next:
1394 root = head->root;
1395 tmp = tmp->next;
1396 header_list = lookup_header_list(root, namespaces);
1397 if (tmp != header_list)
1398 continue;
1399
1400 do {
1401 root = list_entry(root->root_list.next,
1402 struct ctl_table_root, root_list);
1403 if (root == &sysctl_table_root)
1404 goto out;
1405 header_list = lookup_header_list(root, namespaces);
1406 } while (list_empty(header_list));
1407 tmp = header_list->next;
1408 }
1409 out:
1410 spin_unlock(&sysctl_lock);
1411 return NULL;
1412 }
1413
1414 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1415 {
1416 return __sysctl_head_next(current->nsproxy, prev);
1417 }
1418
1419 void register_sysctl_root(struct ctl_table_root *root)
1420 {
1421 spin_lock(&sysctl_lock);
1422 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1423 spin_unlock(&sysctl_lock);
1424 }
1425
1426 #ifdef CONFIG_SYSCTL_SYSCALL
1427 /* Perform the actual read/write of a sysctl table entry. */
1428 static int do_sysctl_strategy(struct ctl_table_root *root,
1429 struct ctl_table *table,
1430 int __user *name, int nlen,
1431 void __user *oldval, size_t __user *oldlenp,
1432 void __user *newval, size_t newlen)
1433 {
1434 int op = 0, rc;
1435
1436 if (oldval)
1437 op |= 004;
1438 if (newval)
1439 op |= 002;
1440 if (sysctl_perm(root, table, op))
1441 return -EPERM;
1442
1443 if (table->strategy) {
1444 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1445 newval, newlen);
1446 if (rc < 0)
1447 return rc;
1448 if (rc > 0)
1449 return 0;
1450 }
1451
1452 /* If there is no strategy routine, or if the strategy returns
1453 * zero, proceed with automatic r/w */
1454 if (table->data && table->maxlen) {
1455 rc = sysctl_data(table, name, nlen, oldval, oldlenp,
1456 newval, newlen);
1457 if (rc < 0)
1458 return rc;
1459 }
1460 return 0;
1461 }
1462
1463 static int parse_table(int __user *name, int nlen,
1464 void __user *oldval, size_t __user *oldlenp,
1465 void __user *newval, size_t newlen,
1466 struct ctl_table_root *root,
1467 struct ctl_table *table)
1468 {
1469 int n;
1470 repeat:
1471 if (!nlen)
1472 return -ENOTDIR;
1473 if (get_user(n, name))
1474 return -EFAULT;
1475 for ( ; table->ctl_name || table->procname; table++) {
1476 if (!table->ctl_name)
1477 continue;
1478 if (n == table->ctl_name) {
1479 int error;
1480 if (table->child) {
1481 if (sysctl_perm(root, table, 001))
1482 return -EPERM;
1483 name++;
1484 nlen--;
1485 table = table->child;
1486 goto repeat;
1487 }
1488 error = do_sysctl_strategy(root, table, name, nlen,
1489 oldval, oldlenp,
1490 newval, newlen);
1491 return error;
1492 }
1493 }
1494 return -ENOTDIR;
1495 }
1496
1497 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1498 void __user *newval, size_t newlen)
1499 {
1500 struct ctl_table_header *head;
1501 int error = -ENOTDIR;
1502
1503 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1504 return -ENOTDIR;
1505 if (oldval) {
1506 int old_len;
1507 if (!oldlenp || get_user(old_len, oldlenp))
1508 return -EFAULT;
1509 }
1510
1511 for (head = sysctl_head_next(NULL); head;
1512 head = sysctl_head_next(head)) {
1513 error = parse_table(name, nlen, oldval, oldlenp,
1514 newval, newlen,
1515 head->root, head->ctl_table);
1516 if (error != -ENOTDIR) {
1517 sysctl_head_finish(head);
1518 break;
1519 }
1520 }
1521 return error;
1522 }
1523
1524 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1525 {
1526 struct __sysctl_args tmp;
1527 int error;
1528
1529 if (copy_from_user(&tmp, args, sizeof(tmp)))
1530 return -EFAULT;
1531
1532 error = deprecated_sysctl_warning(&tmp);
1533 if (error)
1534 goto out;
1535
1536 lock_kernel();
1537 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1538 tmp.newval, tmp.newlen);
1539 unlock_kernel();
1540 out:
1541 return error;
1542 }
1543 #endif /* CONFIG_SYSCTL_SYSCALL */
1544
1545 /*
1546 * sysctl_perm does NOT grant the superuser all rights automatically, because
1547 * some sysctl variables are readonly even to root.
1548 */
1549
1550 static int test_perm(int mode, int op)
1551 {
1552 if (!current->euid)
1553 mode >>= 6;
1554 else if (in_egroup_p(0))
1555 mode >>= 3;
1556 if ((mode & op & 0007) == op)
1557 return 0;
1558 return -EACCES;
1559 }
1560
1561 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1562 {
1563 int error;
1564 int mode;
1565
1566 error = security_sysctl(table, op);
1567 if (error)
1568 return error;
1569
1570 if (root->permissions)
1571 mode = root->permissions(root, current->nsproxy, table);
1572 else
1573 mode = table->mode;
1574
1575 return test_perm(mode, op);
1576 }
1577
1578 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1579 {
1580 for (; table->ctl_name || table->procname; table++) {
1581 table->parent = parent;
1582 if (table->child)
1583 sysctl_set_parent(table, table->child);
1584 }
1585 }
1586
1587 static __init int sysctl_init(void)
1588 {
1589 sysctl_set_parent(NULL, root_table);
1590 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1591 {
1592 int err;
1593 err = sysctl_check_table(current->nsproxy, root_table);
1594 }
1595 #endif
1596 return 0;
1597 }
1598
1599 core_initcall(sysctl_init);
1600
1601 /**
1602 * __register_sysctl_paths - register a sysctl hierarchy
1603 * @root: List of sysctl headers to register on
1604 * @namespaces: Data to compute which lists of sysctl entries are visible
1605 * @path: The path to the directory the sysctl table is in.
1606 * @table: the top-level table structure
1607 *
1608 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1609 * array. A completely 0 filled entry terminates the table.
1610 *
1611 * The members of the &struct ctl_table structure are used as follows:
1612 *
1613 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1614 * must be unique within that level of sysctl
1615 *
1616 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1617 * enter a sysctl file
1618 *
1619 * data - a pointer to data for use by proc_handler
1620 *
1621 * maxlen - the maximum size in bytes of the data
1622 *
1623 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1624 *
1625 * child - a pointer to the child sysctl table if this entry is a directory, or
1626 * %NULL.
1627 *
1628 * proc_handler - the text handler routine (described below)
1629 *
1630 * strategy - the strategy routine (described below)
1631 *
1632 * de - for internal use by the sysctl routines
1633 *
1634 * extra1, extra2 - extra pointers usable by the proc handler routines
1635 *
1636 * Leaf nodes in the sysctl tree will be represented by a single file
1637 * under /proc; non-leaf nodes will be represented by directories.
1638 *
1639 * sysctl(2) can automatically manage read and write requests through
1640 * the sysctl table. The data and maxlen fields of the ctl_table
1641 * struct enable minimal validation of the values being written to be
1642 * performed, and the mode field allows minimal authentication.
1643 *
1644 * More sophisticated management can be enabled by the provision of a
1645 * strategy routine with the table entry. This will be called before
1646 * any automatic read or write of the data is performed.
1647 *
1648 * The strategy routine may return
1649 *
1650 * < 0 - Error occurred (error is passed to user process)
1651 *
1652 * 0 - OK - proceed with automatic read or write.
1653 *
1654 * > 0 - OK - read or write has been done by the strategy routine, so
1655 * return immediately.
1656 *
1657 * There must be a proc_handler routine for any terminal nodes
1658 * mirrored under /proc/sys (non-terminals are handled by a built-in
1659 * directory handler). Several default handlers are available to
1660 * cover common cases -
1661 *
1662 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1663 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1664 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1665 *
1666 * It is the handler's job to read the input buffer from user memory
1667 * and process it. The handler should return 0 on success.
1668 *
1669 * This routine returns %NULL on a failure to register, and a pointer
1670 * to the table header on success.
1671 */
1672 struct ctl_table_header *__register_sysctl_paths(
1673 struct ctl_table_root *root,
1674 struct nsproxy *namespaces,
1675 const struct ctl_path *path, struct ctl_table *table)
1676 {
1677 struct list_head *header_list;
1678 struct ctl_table_header *header;
1679 struct ctl_table *new, **prevp;
1680 unsigned int n, npath;
1681
1682 /* Count the path components */
1683 for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1684 ;
1685
1686 /*
1687 * For each path component, allocate a 2-element ctl_table array.
1688 * The first array element will be filled with the sysctl entry
1689 * for this, the second will be the sentinel (ctl_name == 0).
1690 *
1691 * We allocate everything in one go so that we don't have to
1692 * worry about freeing additional memory in unregister_sysctl_table.
1693 */
1694 header = kzalloc(sizeof(struct ctl_table_header) +
1695 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1696 if (!header)
1697 return NULL;
1698
1699 new = (struct ctl_table *) (header + 1);
1700
1701 /* Now connect the dots */
1702 prevp = &header->ctl_table;
1703 for (n = 0; n < npath; ++n, ++path) {
1704 /* Copy the procname */
1705 new->procname = path->procname;
1706 new->ctl_name = path->ctl_name;
1707 new->mode = 0555;
1708
1709 *prevp = new;
1710 prevp = &new->child;
1711
1712 new += 2;
1713 }
1714 *prevp = table;
1715 header->ctl_table_arg = table;
1716
1717 INIT_LIST_HEAD(&header->ctl_entry);
1718 header->used = 0;
1719 header->unregistering = NULL;
1720 header->root = root;
1721 sysctl_set_parent(NULL, header->ctl_table);
1722 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1723 if (sysctl_check_table(namespaces, header->ctl_table)) {
1724 kfree(header);
1725 return NULL;
1726 }
1727 #endif
1728 spin_lock(&sysctl_lock);
1729 header_list = lookup_header_list(root, namespaces);
1730 list_add_tail(&header->ctl_entry, header_list);
1731 spin_unlock(&sysctl_lock);
1732
1733 return header;
1734 }
1735
1736 /**
1737 * register_sysctl_table_path - register a sysctl table hierarchy
1738 * @path: The path to the directory the sysctl table is in.
1739 * @table: the top-level table structure
1740 *
1741 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1742 * array. A completely 0 filled entry terminates the table.
1743 *
1744 * See __register_sysctl_paths for more details.
1745 */
1746 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1747 struct ctl_table *table)
1748 {
1749 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1750 path, table);
1751 }
1752
1753 /**
1754 * register_sysctl_table - register a sysctl table hierarchy
1755 * @table: the top-level table structure
1756 *
1757 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1758 * array. A completely 0 filled entry terminates the table.
1759 *
1760 * See register_sysctl_paths for more details.
1761 */
1762 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1763 {
1764 static const struct ctl_path null_path[] = { {} };
1765
1766 return register_sysctl_paths(null_path, table);
1767 }
1768
1769 /**
1770 * unregister_sysctl_table - unregister a sysctl table hierarchy
1771 * @header: the header returned from register_sysctl_table
1772 *
1773 * Unregisters the sysctl table and all children. proc entries may not
1774 * actually be removed until they are no longer used by anyone.
1775 */
1776 void unregister_sysctl_table(struct ctl_table_header * header)
1777 {
1778 might_sleep();
1779
1780 if (header == NULL)
1781 return;
1782
1783 spin_lock(&sysctl_lock);
1784 start_unregistering(header);
1785 spin_unlock(&sysctl_lock);
1786 kfree(header);
1787 }
1788
1789 #else /* !CONFIG_SYSCTL */
1790 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1791 {
1792 return NULL;
1793 }
1794
1795 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1796 struct ctl_table *table)
1797 {
1798 return NULL;
1799 }
1800
1801 void unregister_sysctl_table(struct ctl_table_header * table)
1802 {
1803 }
1804
1805 #endif /* CONFIG_SYSCTL */
1806
1807 /*
1808 * /proc/sys support
1809 */
1810
1811 #ifdef CONFIG_PROC_SYSCTL
1812
1813 static int _proc_do_string(void* data, int maxlen, int write,
1814 struct file *filp, void __user *buffer,
1815 size_t *lenp, loff_t *ppos)
1816 {
1817 size_t len;
1818 char __user *p;
1819 char c;
1820
1821 if (!data || !maxlen || !*lenp) {
1822 *lenp = 0;
1823 return 0;
1824 }
1825
1826 if (write) {
1827 len = 0;
1828 p = buffer;
1829 while (len < *lenp) {
1830 if (get_user(c, p++))
1831 return -EFAULT;
1832 if (c == 0 || c == '\n')
1833 break;
1834 len++;
1835 }
1836 if (len >= maxlen)
1837 len = maxlen-1;
1838 if(copy_from_user(data, buffer, len))
1839 return -EFAULT;
1840 ((char *) data)[len] = 0;
1841 *ppos += *lenp;
1842 } else {
1843 len = strlen(data);
1844 if (len > maxlen)
1845 len = maxlen;
1846
1847 if (*ppos > len) {
1848 *lenp = 0;
1849 return 0;
1850 }
1851
1852 data += *ppos;
1853 len -= *ppos;
1854
1855 if (len > *lenp)
1856 len = *lenp;
1857 if (len)
1858 if(copy_to_user(buffer, data, len))
1859 return -EFAULT;
1860 if (len < *lenp) {
1861 if(put_user('\n', ((char __user *) buffer) + len))
1862 return -EFAULT;
1863 len++;
1864 }
1865 *lenp = len;
1866 *ppos += len;
1867 }
1868 return 0;
1869 }
1870
1871 /**
1872 * proc_dostring - read a string sysctl
1873 * @table: the sysctl table
1874 * @write: %TRUE if this is a write to the sysctl file
1875 * @filp: the file structure
1876 * @buffer: the user buffer
1877 * @lenp: the size of the user buffer
1878 * @ppos: file position
1879 *
1880 * Reads/writes a string from/to the user buffer. If the kernel
1881 * buffer provided is not large enough to hold the string, the
1882 * string is truncated. The copied string is %NULL-terminated.
1883 * If the string is being read by the user process, it is copied
1884 * and a newline '\n' is added. It is truncated if the buffer is
1885 * not large enough.
1886 *
1887 * Returns 0 on success.
1888 */
1889 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
1890 void __user *buffer, size_t *lenp, loff_t *ppos)
1891 {
1892 return _proc_do_string(table->data, table->maxlen, write, filp,
1893 buffer, lenp, ppos);
1894 }
1895
1896
1897 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1898 int *valp,
1899 int write, void *data)
1900 {
1901 if (write) {
1902 *valp = *negp ? -*lvalp : *lvalp;
1903 } else {
1904 int val = *valp;
1905 if (val < 0) {
1906 *negp = -1;
1907 *lvalp = (unsigned long)-val;
1908 } else {
1909 *negp = 0;
1910 *lvalp = (unsigned long)val;
1911 }
1912 }
1913 return 0;
1914 }
1915
1916 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1917 int write, struct file *filp, void __user *buffer,
1918 size_t *lenp, loff_t *ppos,
1919 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1920 int write, void *data),
1921 void *data)
1922 {
1923 #define TMPBUFLEN 21
1924 int *i, vleft, first=1, neg, val;
1925 unsigned long lval;
1926 size_t left, len;
1927
1928 char buf[TMPBUFLEN], *p;
1929 char __user *s = buffer;
1930
1931 if (!tbl_data || !table->maxlen || !*lenp ||
1932 (*ppos && !write)) {
1933 *lenp = 0;
1934 return 0;
1935 }
1936
1937 i = (int *) tbl_data;
1938 vleft = table->maxlen / sizeof(*i);
1939 left = *lenp;
1940
1941 if (!conv)
1942 conv = do_proc_dointvec_conv;
1943
1944 for (; left && vleft--; i++, first=0) {
1945 if (write) {
1946 while (left) {
1947 char c;
1948 if (get_user(c, s))
1949 return -EFAULT;
1950 if (!isspace(c))
1951 break;
1952 left--;
1953 s++;
1954 }
1955 if (!left)
1956 break;
1957 neg = 0;
1958 len = left;
1959 if (len > sizeof(buf) - 1)
1960 len = sizeof(buf) - 1;
1961 if (copy_from_user(buf, s, len))
1962 return -EFAULT;
1963 buf[len] = 0;
1964 p = buf;
1965 if (*p == '-' && left > 1) {
1966 neg = 1;
1967 p++;
1968 }
1969 if (*p < '0' || *p > '9')
1970 break;
1971
1972 lval = simple_strtoul(p, &p, 0);
1973
1974 len = p-buf;
1975 if ((len < left) && *p && !isspace(*p))
1976 break;
1977 if (neg)
1978 val = -val;
1979 s += len;
1980 left -= len;
1981
1982 if (conv(&neg, &lval, i, 1, data))
1983 break;
1984 } else {
1985 p = buf;
1986 if (!first)
1987 *p++ = '\t';
1988
1989 if (conv(&neg, &lval, i, 0, data))
1990 break;
1991
1992 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1993 len = strlen(buf);
1994 if (len > left)
1995 len = left;
1996 if(copy_to_user(s, buf, len))
1997 return -EFAULT;
1998 left -= len;
1999 s += len;
2000 }
2001 }
2002
2003 if (!write && !first && left) {
2004 if(put_user('\n', s))
2005 return -EFAULT;
2006 left--, s++;
2007 }
2008 if (write) {
2009 while (left) {
2010 char c;
2011 if (get_user(c, s++))
2012 return -EFAULT;
2013 if (!isspace(c))
2014 break;
2015 left--;
2016 }
2017 }
2018 if (write && first)
2019 return -EINVAL;
2020 *lenp -= left;
2021 *ppos += *lenp;
2022 return 0;
2023 #undef TMPBUFLEN
2024 }
2025
2026 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2027 void __user *buffer, size_t *lenp, loff_t *ppos,
2028 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2029 int write, void *data),
2030 void *data)
2031 {
2032 return __do_proc_dointvec(table->data, table, write, filp,
2033 buffer, lenp, ppos, conv, data);
2034 }
2035
2036 /**
2037 * proc_dointvec - read a vector of integers
2038 * @table: the sysctl table
2039 * @write: %TRUE if this is a write to the sysctl file
2040 * @filp: the file structure
2041 * @buffer: the user buffer
2042 * @lenp: the size of the user buffer
2043 * @ppos: file position
2044 *
2045 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2046 * values from/to the user buffer, treated as an ASCII string.
2047 *
2048 * Returns 0 on success.
2049 */
2050 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2051 void __user *buffer, size_t *lenp, loff_t *ppos)
2052 {
2053 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2054 NULL,NULL);
2055 }
2056
2057 #define OP_SET 0
2058 #define OP_AND 1
2059 #define OP_OR 2
2060
2061 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
2062 int *valp,
2063 int write, void *data)
2064 {
2065 int op = *(int *)data;
2066 if (write) {
2067 int val = *negp ? -*lvalp : *lvalp;
2068 switch(op) {
2069 case OP_SET: *valp = val; break;
2070 case OP_AND: *valp &= val; break;
2071 case OP_OR: *valp |= val; break;
2072 }
2073 } else {
2074 int val = *valp;
2075 if (val < 0) {
2076 *negp = -1;
2077 *lvalp = (unsigned long)-val;
2078 } else {
2079 *negp = 0;
2080 *lvalp = (unsigned long)val;
2081 }
2082 }
2083 return 0;
2084 }
2085
2086 /*
2087 * Taint values can only be increased
2088 */
2089 static int proc_dointvec_taint(struct ctl_table *table, int write, struct file *filp,
2090 void __user *buffer, size_t *lenp, loff_t *ppos)
2091 {
2092 int op;
2093
2094 if (write && !capable(CAP_SYS_ADMIN))
2095 return -EPERM;
2096
2097 op = OP_OR;
2098 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2099 do_proc_dointvec_bset_conv,&op);
2100 }
2101
2102 struct do_proc_dointvec_minmax_conv_param {
2103 int *min;
2104 int *max;
2105 };
2106
2107 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2108 int *valp,
2109 int write, void *data)
2110 {
2111 struct do_proc_dointvec_minmax_conv_param *param = data;
2112 if (write) {
2113 int val = *negp ? -*lvalp : *lvalp;
2114 if ((param->min && *param->min > val) ||
2115 (param->max && *param->max < val))
2116 return -EINVAL;
2117 *valp = val;
2118 } else {
2119 int val = *valp;
2120 if (val < 0) {
2121 *negp = -1;
2122 *lvalp = (unsigned long)-val;
2123 } else {
2124 *negp = 0;
2125 *lvalp = (unsigned long)val;
2126 }
2127 }
2128 return 0;
2129 }
2130
2131 /**
2132 * proc_dointvec_minmax - read a vector of integers with min/max values
2133 * @table: the sysctl table
2134 * @write: %TRUE if this is a write to the sysctl file
2135 * @filp: the file structure
2136 * @buffer: the user buffer
2137 * @lenp: the size of the user buffer
2138 * @ppos: file position
2139 *
2140 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2141 * values from/to the user buffer, treated as an ASCII string.
2142 *
2143 * This routine will ensure the values are within the range specified by
2144 * table->extra1 (min) and table->extra2 (max).
2145 *
2146 * Returns 0 on success.
2147 */
2148 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2149 void __user *buffer, size_t *lenp, loff_t *ppos)
2150 {
2151 struct do_proc_dointvec_minmax_conv_param param = {
2152 .min = (int *) table->extra1,
2153 .max = (int *) table->extra2,
2154 };
2155 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2156 do_proc_dointvec_minmax_conv, &param);
2157 }
2158
2159 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2160 struct file *filp,
2161 void __user *buffer,
2162 size_t *lenp, loff_t *ppos,
2163 unsigned long convmul,
2164 unsigned long convdiv)
2165 {
2166 #define TMPBUFLEN 21
2167 unsigned long *i, *min, *max, val;
2168 int vleft, first=1, neg;
2169 size_t len, left;
2170 char buf[TMPBUFLEN], *p;
2171 char __user *s = buffer;
2172
2173 if (!data || !table->maxlen || !*lenp ||
2174 (*ppos && !write)) {
2175 *lenp = 0;
2176 return 0;
2177 }
2178
2179 i = (unsigned long *) data;
2180 min = (unsigned long *) table->extra1;
2181 max = (unsigned long *) table->extra2;
2182 vleft = table->maxlen / sizeof(unsigned long);
2183 left = *lenp;
2184
2185 for (; left && vleft--; i++, min++, max++, first=0) {
2186 if (write) {
2187 while (left) {
2188 char c;
2189 if (get_user(c, s))
2190 return -EFAULT;
2191 if (!isspace(c))
2192 break;
2193 left--;
2194 s++;
2195 }
2196 if (!left)
2197 break;
2198 neg = 0;
2199 len = left;
2200 if (len > TMPBUFLEN-1)
2201 len = TMPBUFLEN-1;
2202 if (copy_from_user(buf, s, len))
2203 return -EFAULT;
2204 buf[len] = 0;
2205 p = buf;
2206 if (*p == '-' && left > 1) {
2207 neg = 1;
2208 p++;
2209 }
2210 if (*p < '0' || *p > '9')
2211 break;
2212 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2213 len = p-buf;
2214 if ((len < left) && *p && !isspace(*p))
2215 break;
2216 if (neg)
2217 val = -val;
2218 s += len;
2219 left -= len;
2220
2221 if(neg)
2222 continue;
2223 if ((min && val < *min) || (max && val > *max))
2224 continue;
2225 *i = val;
2226 } else {
2227 p = buf;
2228 if (!first)
2229 *p++ = '\t';
2230 sprintf(p, "%lu", convdiv * (*i) / convmul);
2231 len = strlen(buf);
2232 if (len > left)
2233 len = left;
2234 if(copy_to_user(s, buf, len))
2235 return -EFAULT;
2236 left -= len;
2237 s += len;
2238 }
2239 }
2240
2241 if (!write && !first && left) {
2242 if(put_user('\n', s))
2243 return -EFAULT;
2244 left--, s++;
2245 }
2246 if (write) {
2247 while (left) {
2248 char c;
2249 if (get_user(c, s++))
2250 return -EFAULT;
2251 if (!isspace(c))
2252 break;
2253 left--;
2254 }
2255 }
2256 if (write && first)
2257 return -EINVAL;
2258 *lenp -= left;
2259 *ppos += *lenp;
2260 return 0;
2261 #undef TMPBUFLEN
2262 }
2263
2264 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2265 struct file *filp,
2266 void __user *buffer,
2267 size_t *lenp, loff_t *ppos,
2268 unsigned long convmul,
2269 unsigned long convdiv)
2270 {
2271 return __do_proc_doulongvec_minmax(table->data, table, write,
2272 filp, buffer, lenp, ppos, convmul, convdiv);
2273 }
2274
2275 /**
2276 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2277 * @table: the sysctl table
2278 * @write: %TRUE if this is a write to the sysctl file
2279 * @filp: the file structure
2280 * @buffer: the user buffer
2281 * @lenp: the size of the user buffer
2282 * @ppos: file position
2283 *
2284 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2285 * values from/to the user buffer, treated as an ASCII string.
2286 *
2287 * This routine will ensure the values are within the range specified by
2288 * table->extra1 (min) and table->extra2 (max).
2289 *
2290 * Returns 0 on success.
2291 */
2292 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2293 void __user *buffer, size_t *lenp, loff_t *ppos)
2294 {
2295 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2296 }
2297
2298 /**
2299 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2300 * @table: the sysctl table
2301 * @write: %TRUE if this is a write to the sysctl file
2302 * @filp: the file structure
2303 * @buffer: the user buffer
2304 * @lenp: the size of the user buffer
2305 * @ppos: file position
2306 *
2307 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2308 * values from/to the user buffer, treated as an ASCII string. The values
2309 * are treated as milliseconds, and converted to jiffies when they are stored.
2310 *
2311 * This routine will ensure the values are within the range specified by
2312 * table->extra1 (min) and table->extra2 (max).
2313 *
2314 * Returns 0 on success.
2315 */
2316 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2317 struct file *filp,
2318 void __user *buffer,
2319 size_t *lenp, loff_t *ppos)
2320 {
2321 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2322 lenp, ppos, HZ, 1000l);
2323 }
2324
2325
2326 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2327 int *valp,
2328 int write, void *data)
2329 {
2330 if (write) {
2331 if (*lvalp > LONG_MAX / HZ)
2332 return 1;
2333 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2334 } else {
2335 int val = *valp;
2336 unsigned long lval;
2337 if (val < 0) {
2338 *negp = -1;
2339 lval = (unsigned long)-val;
2340 } else {
2341 *negp = 0;
2342 lval = (unsigned long)val;
2343 }
2344 *lvalp = lval / HZ;
2345 }
2346 return 0;
2347 }
2348
2349 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2350 int *valp,
2351 int write, void *data)
2352 {
2353 if (write) {
2354 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2355 return 1;
2356 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2357 } else {
2358 int val = *valp;
2359 unsigned long lval;
2360 if (val < 0) {
2361 *negp = -1;
2362 lval = (unsigned long)-val;
2363 } else {
2364 *negp = 0;
2365 lval = (unsigned long)val;
2366 }
2367 *lvalp = jiffies_to_clock_t(lval);
2368 }
2369 return 0;
2370 }
2371
2372 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2373 int *valp,
2374 int write, void *data)
2375 {
2376 if (write) {
2377 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2378 } else {
2379 int val = *valp;
2380 unsigned long lval;
2381 if (val < 0) {
2382 *negp = -1;
2383 lval = (unsigned long)-val;
2384 } else {
2385 *negp = 0;
2386 lval = (unsigned long)val;
2387 }
2388 *lvalp = jiffies_to_msecs(lval);
2389 }
2390 return 0;
2391 }
2392
2393 /**
2394 * proc_dointvec_jiffies - read a vector of integers as seconds
2395 * @table: the sysctl table
2396 * @write: %TRUE if this is a write to the sysctl file
2397 * @filp: the file structure
2398 * @buffer: the user buffer
2399 * @lenp: the size of the user buffer
2400 * @ppos: file position
2401 *
2402 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2403 * values from/to the user buffer, treated as an ASCII string.
2404 * The values read are assumed to be in seconds, and are converted into
2405 * jiffies.
2406 *
2407 * Returns 0 on success.
2408 */
2409 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2410 void __user *buffer, size_t *lenp, loff_t *ppos)
2411 {
2412 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2413 do_proc_dointvec_jiffies_conv,NULL);
2414 }
2415
2416 /**
2417 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2418 * @table: the sysctl table
2419 * @write: %TRUE if this is a write to the sysctl file
2420 * @filp: the file structure
2421 * @buffer: the user buffer
2422 * @lenp: the size of the user buffer
2423 * @ppos: pointer to the file position
2424 *
2425 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2426 * values from/to the user buffer, treated as an ASCII string.
2427 * The values read are assumed to be in 1/USER_HZ seconds, and
2428 * are converted into jiffies.
2429 *
2430 * Returns 0 on success.
2431 */
2432 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2433 void __user *buffer, size_t *lenp, loff_t *ppos)
2434 {
2435 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2436 do_proc_dointvec_userhz_jiffies_conv,NULL);
2437 }
2438
2439 /**
2440 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2441 * @table: the sysctl table
2442 * @write: %TRUE if this is a write to the sysctl file
2443 * @filp: the file structure
2444 * @buffer: the user buffer
2445 * @lenp: the size of the user buffer
2446 * @ppos: file position
2447 * @ppos: the current position in the file
2448 *
2449 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2450 * values from/to the user buffer, treated as an ASCII string.
2451 * The values read are assumed to be in 1/1000 seconds, and
2452 * are converted into jiffies.
2453 *
2454 * Returns 0 on success.
2455 */
2456 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2457 void __user *buffer, size_t *lenp, loff_t *ppos)
2458 {
2459 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2460 do_proc_dointvec_ms_jiffies_conv, NULL);
2461 }
2462
2463 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2464 void __user *buffer, size_t *lenp, loff_t *ppos)
2465 {
2466 struct pid *new_pid;
2467 pid_t tmp;
2468 int r;
2469
2470 tmp = pid_vnr(cad_pid);
2471
2472 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2473 lenp, ppos, NULL, NULL);
2474 if (r || !write)
2475 return r;
2476
2477 new_pid = find_get_pid(tmp);
2478 if (!new_pid)
2479 return -ESRCH;
2480
2481 put_pid(xchg(&cad_pid, new_pid));
2482 return 0;
2483 }
2484
2485 #else /* CONFIG_PROC_FS */
2486
2487 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2488 void __user *buffer, size_t *lenp, loff_t *ppos)
2489 {
2490 return -ENOSYS;
2491 }
2492
2493 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2494 void __user *buffer, size_t *lenp, loff_t *ppos)
2495 {
2496 return -ENOSYS;
2497 }
2498
2499 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2500 void __user *buffer, size_t *lenp, loff_t *ppos)
2501 {
2502 return -ENOSYS;
2503 }
2504
2505 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2506 void __user *buffer, size_t *lenp, loff_t *ppos)
2507 {
2508 return -ENOSYS;
2509 }
2510
2511 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2512 void __user *buffer, size_t *lenp, loff_t *ppos)
2513 {
2514 return -ENOSYS;
2515 }
2516
2517 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2518 void __user *buffer, size_t *lenp, loff_t *ppos)
2519 {
2520 return -ENOSYS;
2521 }
2522
2523 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2524 void __user *buffer, size_t *lenp, loff_t *ppos)
2525 {
2526 return -ENOSYS;
2527 }
2528
2529 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2530 struct file *filp,
2531 void __user *buffer,
2532 size_t *lenp, loff_t *ppos)
2533 {
2534 return -ENOSYS;
2535 }
2536
2537
2538 #endif /* CONFIG_PROC_FS */
2539
2540
2541 #ifdef CONFIG_SYSCTL_SYSCALL
2542 /*
2543 * General sysctl support routines
2544 */
2545
2546 /* The generic sysctl data routine (used if no strategy routine supplied) */
2547 int sysctl_data(struct ctl_table *table, int __user *name, int nlen,
2548 void __user *oldval, size_t __user *oldlenp,
2549 void __user *newval, size_t newlen)
2550 {
2551 size_t len;
2552
2553 /* Get out of I don't have a variable */
2554 if (!table->data || !table->maxlen)
2555 return -ENOTDIR;
2556
2557 if (oldval && oldlenp) {
2558 if (get_user(len, oldlenp))
2559 return -EFAULT;
2560 if (len) {
2561 if (len > table->maxlen)
2562 len = table->maxlen;
2563 if (copy_to_user(oldval, table->data, len))
2564 return -EFAULT;
2565 if (put_user(len, oldlenp))
2566 return -EFAULT;
2567 }
2568 }
2569
2570 if (newval && newlen) {
2571 if (newlen > table->maxlen)
2572 newlen = table->maxlen;
2573
2574 if (copy_from_user(table->data, newval, newlen))
2575 return -EFAULT;
2576 }
2577 return 1;
2578 }
2579
2580 /* The generic string strategy routine: */
2581 int sysctl_string(struct ctl_table *table, int __user *name, int nlen,
2582 void __user *oldval, size_t __user *oldlenp,
2583 void __user *newval, size_t newlen)
2584 {
2585 if (!table->data || !table->maxlen)
2586 return -ENOTDIR;
2587
2588 if (oldval && oldlenp) {
2589 size_t bufsize;
2590 if (get_user(bufsize, oldlenp))
2591 return -EFAULT;
2592 if (bufsize) {
2593 size_t len = strlen(table->data), copied;
2594
2595 /* This shouldn't trigger for a well-formed sysctl */
2596 if (len > table->maxlen)
2597 len = table->maxlen;
2598
2599 /* Copy up to a max of bufsize-1 bytes of the string */
2600 copied = (len >= bufsize) ? bufsize - 1 : len;
2601
2602 if (copy_to_user(oldval, table->data, copied) ||
2603 put_user(0, (char __user *)(oldval + copied)))
2604 return -EFAULT;
2605 if (put_user(len, oldlenp))
2606 return -EFAULT;
2607 }
2608 }
2609 if (newval && newlen) {
2610 size_t len = newlen;
2611 if (len > table->maxlen)
2612 len = table->maxlen;
2613 if(copy_from_user(table->data, newval, len))
2614 return -EFAULT;
2615 if (len == table->maxlen)
2616 len--;
2617 ((char *) table->data)[len] = 0;
2618 }
2619 return 1;
2620 }
2621
2622 /*
2623 * This function makes sure that all of the integers in the vector
2624 * are between the minimum and maximum values given in the arrays
2625 * table->extra1 and table->extra2, respectively.
2626 */
2627 int sysctl_intvec(struct ctl_table *table, int __user *name, int nlen,
2628 void __user *oldval, size_t __user *oldlenp,
2629 void __user *newval, size_t newlen)
2630 {
2631
2632 if (newval && newlen) {
2633 int __user *vec = (int __user *) newval;
2634 int *min = (int *) table->extra1;
2635 int *max = (int *) table->extra2;
2636 size_t length;
2637 int i;
2638
2639 if (newlen % sizeof(int) != 0)
2640 return -EINVAL;
2641
2642 if (!table->extra1 && !table->extra2)
2643 return 0;
2644
2645 if (newlen > table->maxlen)
2646 newlen = table->maxlen;
2647 length = newlen / sizeof(int);
2648
2649 for (i = 0; i < length; i++) {
2650 int value;
2651 if (get_user(value, vec + i))
2652 return -EFAULT;
2653 if (min && value < min[i])
2654 return -EINVAL;
2655 if (max && value > max[i])
2656 return -EINVAL;
2657 }
2658 }
2659 return 0;
2660 }
2661
2662 /* Strategy function to convert jiffies to seconds */
2663 int sysctl_jiffies(struct ctl_table *table, int __user *name, int nlen,
2664 void __user *oldval, size_t __user *oldlenp,
2665 void __user *newval, size_t newlen)
2666 {
2667 if (oldval && oldlenp) {
2668 size_t olen;
2669
2670 if (get_user(olen, oldlenp))
2671 return -EFAULT;
2672 if (olen) {
2673 int val;
2674
2675 if (olen < sizeof(int))
2676 return -EINVAL;
2677
2678 val = *(int *)(table->data) / HZ;
2679 if (put_user(val, (int __user *)oldval))
2680 return -EFAULT;
2681 if (put_user(sizeof(int), oldlenp))
2682 return -EFAULT;
2683 }
2684 }
2685 if (newval && newlen) {
2686 int new;
2687 if (newlen != sizeof(int))
2688 return -EINVAL;
2689 if (get_user(new, (int __user *)newval))
2690 return -EFAULT;
2691 *(int *)(table->data) = new*HZ;
2692 }
2693 return 1;
2694 }
2695
2696 /* Strategy function to convert jiffies to seconds */
2697 int sysctl_ms_jiffies(struct ctl_table *table, int __user *name, int nlen,
2698 void __user *oldval, size_t __user *oldlenp,
2699 void __user *newval, size_t newlen)
2700 {
2701 if (oldval && oldlenp) {
2702 size_t olen;
2703
2704 if (get_user(olen, oldlenp))
2705 return -EFAULT;
2706 if (olen) {
2707 int val;
2708
2709 if (olen < sizeof(int))
2710 return -EINVAL;
2711
2712 val = jiffies_to_msecs(*(int *)(table->data));
2713 if (put_user(val, (int __user *)oldval))
2714 return -EFAULT;
2715 if (put_user(sizeof(int), oldlenp))
2716 return -EFAULT;
2717 }
2718 }
2719 if (newval && newlen) {
2720 int new;
2721 if (newlen != sizeof(int))
2722 return -EINVAL;
2723 if (get_user(new, (int __user *)newval))
2724 return -EFAULT;
2725 *(int *)(table->data) = msecs_to_jiffies(new);
2726 }
2727 return 1;
2728 }
2729
2730
2731
2732 #else /* CONFIG_SYSCTL_SYSCALL */
2733
2734
2735 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2736 {
2737 struct __sysctl_args tmp;
2738 int error;
2739
2740 if (copy_from_user(&tmp, args, sizeof(tmp)))
2741 return -EFAULT;
2742
2743 error = deprecated_sysctl_warning(&tmp);
2744
2745 /* If no error reading the parameters then just -ENOSYS ... */
2746 if (!error)
2747 error = -ENOSYS;
2748
2749 return error;
2750 }
2751
2752 int sysctl_data(struct ctl_table *table, int __user *name, int nlen,
2753 void __user *oldval, size_t __user *oldlenp,
2754 void __user *newval, size_t newlen)
2755 {
2756 return -ENOSYS;
2757 }
2758
2759 int sysctl_string(struct ctl_table *table, int __user *name, int nlen,
2760 void __user *oldval, size_t __user *oldlenp,
2761 void __user *newval, size_t newlen)
2762 {
2763 return -ENOSYS;
2764 }
2765
2766 int sysctl_intvec(struct ctl_table *table, int __user *name, int nlen,
2767 void __user *oldval, size_t __user *oldlenp,
2768 void __user *newval, size_t newlen)
2769 {
2770 return -ENOSYS;
2771 }
2772
2773 int sysctl_jiffies(struct ctl_table *table, int __user *name, int nlen,
2774 void __user *oldval, size_t __user *oldlenp,
2775 void __user *newval, size_t newlen)
2776 {
2777 return -ENOSYS;
2778 }
2779
2780 int sysctl_ms_jiffies(struct ctl_table *table, int __user *name, int nlen,
2781 void __user *oldval, size_t __user *oldlenp,
2782 void __user *newval, size_t newlen)
2783 {
2784 return -ENOSYS;
2785 }
2786
2787 #endif /* CONFIG_SYSCTL_SYSCALL */
2788
2789 static int deprecated_sysctl_warning(struct __sysctl_args *args)
2790 {
2791 static int msg_count;
2792 int name[CTL_MAXNAME];
2793 int i;
2794
2795 /* Check args->nlen. */
2796 if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
2797 return -ENOTDIR;
2798
2799 /* Read in the sysctl name for better debug message logging */
2800 for (i = 0; i < args->nlen; i++)
2801 if (get_user(name[i], args->name + i))
2802 return -EFAULT;
2803
2804 /* Ignore accesses to kernel.version */
2805 if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2806 return 0;
2807
2808 if (msg_count < 5) {
2809 msg_count++;
2810 printk(KERN_INFO
2811 "warning: process `%s' used the deprecated sysctl "
2812 "system call with ", current->comm);
2813 for (i = 0; i < args->nlen; i++)
2814 printk("%d.", name[i]);
2815 printk("\n");
2816 }
2817 return 0;
2818 }
2819
2820 /*
2821 * No sense putting this after each symbol definition, twice,
2822 * exception granted :-)
2823 */
2824 EXPORT_SYMBOL(proc_dointvec);
2825 EXPORT_SYMBOL(proc_dointvec_jiffies);
2826 EXPORT_SYMBOL(proc_dointvec_minmax);
2827 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2828 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2829 EXPORT_SYMBOL(proc_dostring);
2830 EXPORT_SYMBOL(proc_doulongvec_minmax);
2831 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2832 EXPORT_SYMBOL(register_sysctl_table);
2833 EXPORT_SYMBOL(register_sysctl_paths);
2834 EXPORT_SYMBOL(sysctl_intvec);
2835 EXPORT_SYMBOL(sysctl_jiffies);
2836 EXPORT_SYMBOL(sysctl_ms_jiffies);
2837 EXPORT_SYMBOL(sysctl_string);
2838 EXPORT_SYMBOL(sysctl_data);
2839 EXPORT_SYMBOL(unregister_sysctl_table);
Linux kernel - Deliverables
This page took 0.087201 seconds and 6 git commands to generate.