?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
sysctl: Error on bad sysctl tables
[linux-2.6/btrfs-unstable.git] / kernel / sysctl.c
blob62e53a0de4a3055843b545749d6e0cc570fde0f4
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/capability.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 #if defined(CONFIG_SYSCTL)
59
60 /* External variables not in a header file. */
61 extern int C_A_D;
62 extern int print_fatal_signals;
63 extern int sysctl_overcommit_memory;
64 extern int sysctl_overcommit_ratio;
65 extern int sysctl_panic_on_oom;
66 extern int sysctl_oom_kill_allocating_task;
67 extern int max_threads;
68 extern int core_uses_pid;
69 extern int suid_dumpable;
70 extern char core_pattern[];
71 extern int pid_max;
72 extern int min_free_kbytes;
73 extern int printk_ratelimit_jiffies;
74 extern int printk_ratelimit_burst;
75 extern int pid_max_min, pid_max_max;
76 extern int sysctl_drop_caches;
77 extern int percpu_pagelist_fraction;
78 extern int compat_log;
79 extern int maps_protect;
80 extern int sysctl_stat_interval;
81 extern int audit_argv_kb;
82
83 /* Constants used for minimum and maximum */
84 #ifdef CONFIG_DETECT_SOFTLOCKUP
85 static int one = 1;
86 static int sixty = 60;
87 #endif
88
89 #ifdef CONFIG_MMU
90 static int two = 2;
91 #endif
92
93 static int zero;
94 static int one_hundred = 100;
95
96 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
97 static int maxolduid = 65535;
98 static int minolduid;
99 static int min_percpu_pagelist_fract = 8;
100
101 static int ngroups_max = NGROUPS_MAX;
102
103 #ifdef CONFIG_KMOD
104 extern char modprobe_path[];
105 #endif
106 #ifdef CONFIG_CHR_DEV_SG
107 extern int sg_big_buff;
108 #endif
109
110 #ifdef __sparc__
111 extern char reboot_command [];
112 extern int stop_a_enabled;
113 extern int scons_pwroff;
114 #endif
115
116 #ifdef __hppa__
117 extern int pwrsw_enabled;
118 extern int unaligned_enabled;
119 #endif
120
121 #ifdef CONFIG_S390
122 #ifdef CONFIG_MATHEMU
123 extern int sysctl_ieee_emulation_warnings;
124 #endif
125 extern int sysctl_userprocess_debug;
126 extern int spin_retry;
127 #endif
128
129 extern int sysctl_hz_timer;
130
131 #ifdef CONFIG_BSD_PROCESS_ACCT
132 extern int acct_parm[];
133 #endif
134
135 #ifdef CONFIG_IA64
136 extern int no_unaligned_warning;
137 #endif
138
139 #ifdef CONFIG_RT_MUTEXES
140 extern int max_lock_depth;
141 #endif
142
143 #ifdef CONFIG_SYSCTL_SYSCALL
144 static int parse_table(int __user *, int, void __user *, size_t __user *,
145 void __user *, size_t, struct ctl_table *);
146 #endif
147
148
149 #ifdef CONFIG_PROC_SYSCTL
150 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
151 void __user *buffer, size_t *lenp, loff_t *ppos);
152 static int proc_dointvec_taint(struct ctl_table *table, int write, struct file *filp,
153 void __user *buffer, size_t *lenp, loff_t *ppos);
154 #endif
155
156 static struct ctl_table root_table[];
157 static struct ctl_table_header root_table_header =
158 { root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
159
160 static struct ctl_table kern_table[];
161 static struct ctl_table vm_table[];
162 static struct ctl_table fs_table[];
163 static struct ctl_table debug_table[];
164 static struct ctl_table dev_table[];
165 extern struct ctl_table random_table[];
166 #ifdef CONFIG_INOTIFY_USER
167 extern struct ctl_table inotify_table[];
168 #endif
169
170 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
171 int sysctl_legacy_va_layout;
172 #endif
173
174 extern int prove_locking;
175 extern int lock_stat;
176
177 /* The default sysctl tables: */
178
179 static struct ctl_table root_table[] = {
180 {
181 .ctl_name = CTL_KERN,
182 .procname = "kernel",
183 .mode = 0555,
184 .child = kern_table,
185 },
186 {
187 .ctl_name = CTL_VM,
188 .procname = "vm",
189 .mode = 0555,
190 .child = vm_table,
191 },
192 #ifdef CONFIG_NET
193 {
194 .ctl_name = CTL_NET,
195 .procname = "net",
196 .mode = 0555,
197 .child = net_table,
198 },
199 #endif
200 {
201 .ctl_name = CTL_FS,
202 .procname = "fs",
203 .mode = 0555,
204 .child = fs_table,
205 },
206 {
207 .ctl_name = CTL_DEBUG,
208 .procname = "debug",
209 .mode = 0555,
210 .child = debug_table,
211 },
212 {
213 .ctl_name = CTL_DEV,
214 .procname = "dev",
215 .mode = 0555,
216 .child = dev_table,
217 },
218 /*
219 * NOTE: do not add new entries to this table unless you have read
220 * Documentation/sysctl/ctl_unnumbered.txt
221 */
222 { .ctl_name = 0 }
223 };
224
225 #ifdef CONFIG_SCHED_DEBUG
226 static unsigned long min_sched_granularity_ns = 100000; /* 100 usecs */
227 static unsigned long max_sched_granularity_ns = 1000000000; /* 1 second */
228 static unsigned long min_wakeup_granularity_ns; /* 0 usecs */
229 static unsigned long max_wakeup_granularity_ns = 1000000000; /* 1 second */
230 #endif
231
232 static struct ctl_table kern_table[] = {
233 #ifdef CONFIG_SCHED_DEBUG
234 {
235 .ctl_name = CTL_UNNUMBERED,
236 .procname = "sched_nr_latency",
237 .data = &sysctl_sched_nr_latency,
238 .maxlen = sizeof(unsigned int),
239 .mode = 0644,
240 .proc_handler = &proc_dointvec,
241 },
242 {
243 .ctl_name = CTL_UNNUMBERED,
244 .procname = "sched_latency_ns",
245 .data = &sysctl_sched_latency,
246 .maxlen = sizeof(unsigned int),
247 .mode = 0644,
248 .proc_handler = &proc_dointvec_minmax,
249 .strategy = &sysctl_intvec,
250 .extra1 = &min_sched_granularity_ns,
251 .extra2 = &max_sched_granularity_ns,
252 },
253 {
254 .ctl_name = CTL_UNNUMBERED,
255 .procname = "sched_wakeup_granularity_ns",
256 .data = &sysctl_sched_wakeup_granularity,
257 .maxlen = sizeof(unsigned int),
258 .mode = 0644,
259 .proc_handler = &proc_dointvec_minmax,
260 .strategy = &sysctl_intvec,
261 .extra1 = &min_wakeup_granularity_ns,
262 .extra2 = &max_wakeup_granularity_ns,
263 },
264 {
265 .ctl_name = CTL_UNNUMBERED,
266 .procname = "sched_batch_wakeup_granularity_ns",
267 .data = &sysctl_sched_batch_wakeup_granularity,
268 .maxlen = sizeof(unsigned int),
269 .mode = 0644,
270 .proc_handler = &proc_dointvec_minmax,
271 .strategy = &sysctl_intvec,
272 .extra1 = &min_wakeup_granularity_ns,
273 .extra2 = &max_wakeup_granularity_ns,
274 },
275 {
276 .ctl_name = CTL_UNNUMBERED,
277 .procname = "sched_child_runs_first",
278 .data = &sysctl_sched_child_runs_first,
279 .maxlen = sizeof(unsigned int),
280 .mode = 0644,
281 .proc_handler = &proc_dointvec,
282 },
283 {
284 .ctl_name = CTL_UNNUMBERED,
285 .procname = "sched_features",
286 .data = &sysctl_sched_features,
287 .maxlen = sizeof(unsigned int),
288 .mode = 0644,
289 .proc_handler = &proc_dointvec,
290 },
291 {
292 .ctl_name = CTL_UNNUMBERED,
293 .procname = "sched_migration_cost",
294 .data = &sysctl_sched_migration_cost,
295 .maxlen = sizeof(unsigned int),
296 .mode = 0644,
297 .proc_handler = &proc_dointvec,
298 },
299 #endif
300 {
301 .ctl_name = CTL_UNNUMBERED,
302 .procname = "sched_compat_yield",
303 .data = &sysctl_sched_compat_yield,
304 .maxlen = sizeof(unsigned int),
305 .mode = 0644,
306 .proc_handler = &proc_dointvec,
307 },
308 #ifdef CONFIG_PROVE_LOCKING
309 {
310 .ctl_name = CTL_UNNUMBERED,
311 .procname = "prove_locking",
312 .data = &prove_locking,
313 .maxlen = sizeof(int),
314 .mode = 0644,
315 .proc_handler = &proc_dointvec,
316 },
317 #endif
318 #ifdef CONFIG_LOCK_STAT
319 {
320 .ctl_name = CTL_UNNUMBERED,
321 .procname = "lock_stat",
322 .data = &lock_stat,
323 .maxlen = sizeof(int),
324 .mode = 0644,
325 .proc_handler = &proc_dointvec,
326 },
327 #endif
328 {
329 .ctl_name = KERN_PANIC,
330 .procname = "panic",
331 .data = &panic_timeout,
332 .maxlen = sizeof(int),
333 .mode = 0644,
334 .proc_handler = &proc_dointvec,
335 },
336 {
337 .ctl_name = KERN_CORE_USES_PID,
338 .procname = "core_uses_pid",
339 .data = &core_uses_pid,
340 .maxlen = sizeof(int),
341 .mode = 0644,
342 .proc_handler = &proc_dointvec,
343 },
344 #ifdef CONFIG_AUDITSYSCALL
345 {
346 .ctl_name = CTL_UNNUMBERED,
347 .procname = "audit_argv_kb",
348 .data = &audit_argv_kb,
349 .maxlen = sizeof(int),
350 .mode = 0644,
351 .proc_handler = &proc_dointvec,
352 },
353 #endif
354 {
355 .ctl_name = KERN_CORE_PATTERN,
356 .procname = "core_pattern",
357 .data = core_pattern,
358 .maxlen = CORENAME_MAX_SIZE,
359 .mode = 0644,
360 .proc_handler = &proc_dostring,
361 .strategy = &sysctl_string,
362 },
363 #ifdef CONFIG_PROC_SYSCTL
364 {
365 .procname = "tainted",
366 .data = &tainted,
367 .maxlen = sizeof(int),
368 .mode = 0644,
369 .proc_handler = &proc_dointvec_taint,
370 },
371 #endif
372 {
373 .procname = "cap-bound",
374 .data = &cap_bset,
375 .maxlen = sizeof(kernel_cap_t),
376 .mode = 0600,
377 .proc_handler = &proc_dointvec_bset,
378 },
379 #ifdef CONFIG_BLK_DEV_INITRD
380 {
381 .ctl_name = KERN_REALROOTDEV,
382 .procname = "real-root-dev",
383 .data = &real_root_dev,
384 .maxlen = sizeof(int),
385 .mode = 0644,
386 .proc_handler = &proc_dointvec,
387 },
388 #endif
389 {
390 .ctl_name = CTL_UNNUMBERED,
391 .procname = "print-fatal-signals",
392 .data = &print_fatal_signals,
393 .maxlen = sizeof(int),
394 .mode = 0644,
395 .proc_handler = &proc_dointvec,
396 },
397 #ifdef __sparc__
398 {
399 .ctl_name = KERN_SPARC_REBOOT,
400 .procname = "reboot-cmd",
401 .data = reboot_command,
402 .maxlen = 256,
403 .mode = 0644,
404 .proc_handler = &proc_dostring,
405 .strategy = &sysctl_string,
406 },
407 {
408 .ctl_name = KERN_SPARC_STOP_A,
409 .procname = "stop-a",
410 .data = &stop_a_enabled,
411 .maxlen = sizeof (int),
412 .mode = 0644,
413 .proc_handler = &proc_dointvec,
414 },
415 {
416 .ctl_name = KERN_SPARC_SCONS_PWROFF,
417 .procname = "scons-poweroff",
418 .data = &scons_pwroff,
419 .maxlen = sizeof (int),
420 .mode = 0644,
421 .proc_handler = &proc_dointvec,
422 },
423 #endif
424 #ifdef __hppa__
425 {
426 .ctl_name = KERN_HPPA_PWRSW,
427 .procname = "soft-power",
428 .data = &pwrsw_enabled,
429 .maxlen = sizeof (int),
430 .mode = 0644,
431 .proc_handler = &proc_dointvec,
432 },
433 {
434 .ctl_name = KERN_HPPA_UNALIGNED,
435 .procname = "unaligned-trap",
436 .data = &unaligned_enabled,
437 .maxlen = sizeof (int),
438 .mode = 0644,
439 .proc_handler = &proc_dointvec,
440 },
441 #endif
442 {
443 .ctl_name = KERN_CTLALTDEL,
444 .procname = "ctrl-alt-del",
445 .data = &C_A_D,
446 .maxlen = sizeof(int),
447 .mode = 0644,
448 .proc_handler = &proc_dointvec,
449 },
450 {
451 .ctl_name = KERN_PRINTK,
452 .procname = "printk",
453 .data = &console_loglevel,
454 .maxlen = 4*sizeof(int),
455 .mode = 0644,
456 .proc_handler = &proc_dointvec,
457 },
458 #ifdef CONFIG_KMOD
459 {
460 .ctl_name = KERN_MODPROBE,
461 .procname = "modprobe",
462 .data = &modprobe_path,
463 .maxlen = KMOD_PATH_LEN,
464 .mode = 0644,
465 .proc_handler = &proc_dostring,
466 .strategy = &sysctl_string,
467 },
468 #endif
469 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
470 {
471 .ctl_name = KERN_HOTPLUG,
472 .procname = "hotplug",
473 .data = &uevent_helper,
474 .maxlen = UEVENT_HELPER_PATH_LEN,
475 .mode = 0644,
476 .proc_handler = &proc_dostring,
477 .strategy = &sysctl_string,
478 },
479 #endif
480 #ifdef CONFIG_CHR_DEV_SG
481 {
482 .ctl_name = KERN_SG_BIG_BUFF,
483 .procname = "sg-big-buff",
484 .data = &sg_big_buff,
485 .maxlen = sizeof (int),
486 .mode = 0444,
487 .proc_handler = &proc_dointvec,
488 },
489 #endif
490 #ifdef CONFIG_BSD_PROCESS_ACCT
491 {
492 .ctl_name = KERN_ACCT,
493 .procname = "acct",
494 .data = &acct_parm,
495 .maxlen = 3*sizeof(int),
496 .mode = 0644,
497 .proc_handler = &proc_dointvec,
498 },
499 #endif
500 #ifdef CONFIG_MAGIC_SYSRQ
501 {
502 .ctl_name = KERN_SYSRQ,
503 .procname = "sysrq",
504 .data = &__sysrq_enabled,
505 .maxlen = sizeof (int),
506 .mode = 0644,
507 .proc_handler = &proc_dointvec,
508 },
509 #endif
510 #ifdef CONFIG_PROC_SYSCTL
511 {
512 .procname = "cad_pid",
513 .data = NULL,
514 .maxlen = sizeof (int),
515 .mode = 0600,
516 .proc_handler = &proc_do_cad_pid,
517 },
518 #endif
519 {
520 .ctl_name = KERN_MAX_THREADS,
521 .procname = "threads-max",
522 .data = &max_threads,
523 .maxlen = sizeof(int),
524 .mode = 0644,
525 .proc_handler = &proc_dointvec,
526 },
527 {
528 .ctl_name = KERN_RANDOM,
529 .procname = "random",
530 .mode = 0555,
531 .child = random_table,
532 },
533 {
534 .ctl_name = KERN_OVERFLOWUID,
535 .procname = "overflowuid",
536 .data = &overflowuid,
537 .maxlen = sizeof(int),
538 .mode = 0644,
539 .proc_handler = &proc_dointvec_minmax,
540 .strategy = &sysctl_intvec,
541 .extra1 = &minolduid,
542 .extra2 = &maxolduid,
543 },
544 {
545 .ctl_name = KERN_OVERFLOWGID,
546 .procname = "overflowgid",
547 .data = &overflowgid,
548 .maxlen = sizeof(int),
549 .mode = 0644,
550 .proc_handler = &proc_dointvec_minmax,
551 .strategy = &sysctl_intvec,
552 .extra1 = &minolduid,
553 .extra2 = &maxolduid,
554 },
555 #ifdef CONFIG_S390
556 #ifdef CONFIG_MATHEMU
557 {
558 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
559 .procname = "ieee_emulation_warnings",
560 .data = &sysctl_ieee_emulation_warnings,
561 .maxlen = sizeof(int),
562 .mode = 0644,
563 .proc_handler = &proc_dointvec,
564 },
565 #endif
566 #ifdef CONFIG_NO_IDLE_HZ
567 {
568 .ctl_name = KERN_HZ_TIMER,
569 .procname = "hz_timer",
570 .data = &sysctl_hz_timer,
571 .maxlen = sizeof(int),
572 .mode = 0644,
573 .proc_handler = &proc_dointvec,
574 },
575 #endif
576 {
577 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
578 .procname = "userprocess_debug",
579 .data = &sysctl_userprocess_debug,
580 .maxlen = sizeof(int),
581 .mode = 0644,
582 .proc_handler = &proc_dointvec,
583 },
584 #endif
585 {
586 .ctl_name = KERN_PIDMAX,
587 .procname = "pid_max",
588 .data = &pid_max,
589 .maxlen = sizeof (int),
590 .mode = 0644,
591 .proc_handler = &proc_dointvec_minmax,
592 .strategy = sysctl_intvec,
593 .extra1 = &pid_max_min,
594 .extra2 = &pid_max_max,
595 },
596 {
597 .ctl_name = KERN_PANIC_ON_OOPS,
598 .procname = "panic_on_oops",
599 .data = &panic_on_oops,
600 .maxlen = sizeof(int),
601 .mode = 0644,
602 .proc_handler = &proc_dointvec,
603 },
604 {
605 .ctl_name = KERN_PRINTK_RATELIMIT,
606 .procname = "printk_ratelimit",
607 .data = &printk_ratelimit_jiffies,
608 .maxlen = sizeof(int),
609 .mode = 0644,
610 .proc_handler = &proc_dointvec_jiffies,
611 .strategy = &sysctl_jiffies,
612 },
613 {
614 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
615 .procname = "printk_ratelimit_burst",
616 .data = &printk_ratelimit_burst,
617 .maxlen = sizeof(int),
618 .mode = 0644,
619 .proc_handler = &proc_dointvec,
620 },
621 {
622 .ctl_name = KERN_NGROUPS_MAX,
623 .procname = "ngroups_max",
624 .data = &ngroups_max,
625 .maxlen = sizeof (int),
626 .mode = 0444,
627 .proc_handler = &proc_dointvec,
628 },
629 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
630 {
631 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
632 .procname = "unknown_nmi_panic",
633 .data = &unknown_nmi_panic,
634 .maxlen = sizeof (int),
635 .mode = 0644,
636 .proc_handler = &proc_dointvec,
637 },
638 {
639 .procname = "nmi_watchdog",
640 .data = &nmi_watchdog_enabled,
641 .maxlen = sizeof (int),
642 .mode = 0644,
643 .proc_handler = &proc_nmi_enabled,
644 },
645 #endif
646 #if defined(CONFIG_X86)
647 {
648 .ctl_name = KERN_PANIC_ON_NMI,
649 .procname = "panic_on_unrecovered_nmi",
650 .data = &panic_on_unrecovered_nmi,
651 .maxlen = sizeof(int),
652 .mode = 0644,
653 .proc_handler = &proc_dointvec,
654 },
655 {
656 .ctl_name = KERN_BOOTLOADER_TYPE,
657 .procname = "bootloader_type",
658 .data = &bootloader_type,
659 .maxlen = sizeof (int),
660 .mode = 0444,
661 .proc_handler = &proc_dointvec,
662 },
663 {
664 .ctl_name = CTL_UNNUMBERED,
665 .procname = "kstack_depth_to_print",
666 .data = &kstack_depth_to_print,
667 .maxlen = sizeof(int),
668 .mode = 0644,
669 .proc_handler = &proc_dointvec,
670 },
671 #endif
672 #if defined(CONFIG_MMU)
673 {
674 .ctl_name = KERN_RANDOMIZE,
675 .procname = "randomize_va_space",
676 .data = &randomize_va_space,
677 .maxlen = sizeof(int),
678 .mode = 0644,
679 .proc_handler = &proc_dointvec,
680 },
681 #endif
682 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
683 {
684 .ctl_name = KERN_SPIN_RETRY,
685 .procname = "spin_retry",
686 .data = &spin_retry,
687 .maxlen = sizeof (int),
688 .mode = 0644,
689 .proc_handler = &proc_dointvec,
690 },
691 #endif
692 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
693 {
694 .procname = "acpi_video_flags",
695 .data = &acpi_realmode_flags,
696 .maxlen = sizeof (unsigned long),
697 .mode = 0644,
698 .proc_handler = &proc_doulongvec_minmax,
699 },
700 #endif
701 #ifdef CONFIG_IA64
702 {
703 .ctl_name = KERN_IA64_UNALIGNED,
704 .procname = "ignore-unaligned-usertrap",
705 .data = &no_unaligned_warning,
706 .maxlen = sizeof (int),
707 .mode = 0644,
708 .proc_handler = &proc_dointvec,
709 },
710 #endif
711 #ifdef CONFIG_DETECT_SOFTLOCKUP
712 {
713 .ctl_name = CTL_UNNUMBERED,
714 .procname = "softlockup_thresh",
715 .data = &softlockup_thresh,
716 .maxlen = sizeof(int),
717 .mode = 0644,
718 .proc_handler = &proc_dointvec_minmax,
719 .strategy = &sysctl_intvec,
720 .extra1 = &one,
721 .extra2 = &sixty,
722 },
723 #endif
724 #ifdef CONFIG_COMPAT
725 {
726 .ctl_name = KERN_COMPAT_LOG,
727 .procname = "compat-log",
728 .data = &compat_log,
729 .maxlen = sizeof (int),
730 .mode = 0644,
731 .proc_handler = &proc_dointvec,
732 },
733 #endif
734 #ifdef CONFIG_RT_MUTEXES
735 {
736 .ctl_name = KERN_MAX_LOCK_DEPTH,
737 .procname = "max_lock_depth",
738 .data = &max_lock_depth,
739 .maxlen = sizeof(int),
740 .mode = 0644,
741 .proc_handler = &proc_dointvec,
742 },
743 #endif
744 #ifdef CONFIG_PROC_FS
745 {
746 .ctl_name = CTL_UNNUMBERED,
747 .procname = "maps_protect",
748 .data = &maps_protect,
749 .maxlen = sizeof(int),
750 .mode = 0644,
751 .proc_handler = &proc_dointvec,
752 },
753 #endif
754 {
755 .ctl_name = CTL_UNNUMBERED,
756 .procname = "poweroff_cmd",
757 .data = &poweroff_cmd,
758 .maxlen = POWEROFF_CMD_PATH_LEN,
759 .mode = 0644,
760 .proc_handler = &proc_dostring,
761 .strategy = &sysctl_string,
762 },
763 /*
764 * NOTE: do not add new entries to this table unless you have read
765 * Documentation/sysctl/ctl_unnumbered.txt
766 */
767 { .ctl_name = 0 }
768 };
769
770 static struct ctl_table vm_table[] = {
771 {
772 .ctl_name = VM_OVERCOMMIT_MEMORY,
773 .procname = "overcommit_memory",
774 .data = &sysctl_overcommit_memory,
775 .maxlen = sizeof(sysctl_overcommit_memory),
776 .mode = 0644,
777 .proc_handler = &proc_dointvec,
778 },
779 {
780 .ctl_name = VM_PANIC_ON_OOM,
781 .procname = "panic_on_oom",
782 .data = &sysctl_panic_on_oom,
783 .maxlen = sizeof(sysctl_panic_on_oom),
784 .mode = 0644,
785 .proc_handler = &proc_dointvec,
786 },
787 {
788 .ctl_name = CTL_UNNUMBERED,
789 .procname = "oom_kill_allocating_task",
790 .data = &sysctl_oom_kill_allocating_task,
791 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
792 .mode = 0644,
793 .proc_handler = &proc_dointvec,
794 },
795 {
796 .ctl_name = VM_OVERCOMMIT_RATIO,
797 .procname = "overcommit_ratio",
798 .data = &sysctl_overcommit_ratio,
799 .maxlen = sizeof(sysctl_overcommit_ratio),
800 .mode = 0644,
801 .proc_handler = &proc_dointvec,
802 },
803 {
804 .ctl_name = VM_PAGE_CLUSTER,
805 .procname = "page-cluster",
806 .data = &page_cluster,
807 .maxlen = sizeof(int),
808 .mode = 0644,
809 .proc_handler = &proc_dointvec,
810 },
811 {
812 .ctl_name = VM_DIRTY_BACKGROUND,
813 .procname = "dirty_background_ratio",
814 .data = &dirty_background_ratio,
815 .maxlen = sizeof(dirty_background_ratio),
816 .mode = 0644,
817 .proc_handler = &proc_dointvec_minmax,
818 .strategy = &sysctl_intvec,
819 .extra1 = &zero,
820 .extra2 = &one_hundred,
821 },
822 {
823 .ctl_name = VM_DIRTY_RATIO,
824 .procname = "dirty_ratio",
825 .data = &vm_dirty_ratio,
826 .maxlen = sizeof(vm_dirty_ratio),
827 .mode = 0644,
828 .proc_handler = &dirty_ratio_handler,
829 .strategy = &sysctl_intvec,
830 .extra1 = &zero,
831 .extra2 = &one_hundred,
832 },
833 {
834 .procname = "dirty_writeback_centisecs",
835 .data = &dirty_writeback_interval,
836 .maxlen = sizeof(dirty_writeback_interval),
837 .mode = 0644,
838 .proc_handler = &dirty_writeback_centisecs_handler,
839 },
840 {
841 .procname = "dirty_expire_centisecs",
842 .data = &dirty_expire_interval,
843 .maxlen = sizeof(dirty_expire_interval),
844 .mode = 0644,
845 .proc_handler = &proc_dointvec_userhz_jiffies,
846 },
847 {
848 .ctl_name = VM_NR_PDFLUSH_THREADS,
849 .procname = "nr_pdflush_threads",
850 .data = &nr_pdflush_threads,
851 .maxlen = sizeof nr_pdflush_threads,
852 .mode = 0444 /* read-only*/,
853 .proc_handler = &proc_dointvec,
854 },
855 {
856 .ctl_name = VM_SWAPPINESS,
857 .procname = "swappiness",
858 .data = &vm_swappiness,
859 .maxlen = sizeof(vm_swappiness),
860 .mode = 0644,
861 .proc_handler = &proc_dointvec_minmax,
862 .strategy = &sysctl_intvec,
863 .extra1 = &zero,
864 .extra2 = &one_hundred,
865 },
866 #ifdef CONFIG_HUGETLB_PAGE
867 {
868 .procname = "nr_hugepages",
869 .data = &max_huge_pages,
870 .maxlen = sizeof(unsigned long),
871 .mode = 0644,
872 .proc_handler = &hugetlb_sysctl_handler,
873 .extra1 = (void *)&hugetlb_zero,
874 .extra2 = (void *)&hugetlb_infinity,
875 },
876 {
877 .ctl_name = VM_HUGETLB_GROUP,
878 .procname = "hugetlb_shm_group",
879 .data = &sysctl_hugetlb_shm_group,
880 .maxlen = sizeof(gid_t),
881 .mode = 0644,
882 .proc_handler = &proc_dointvec,
883 },
884 {
885 .ctl_name = CTL_UNNUMBERED,
886 .procname = "hugepages_treat_as_movable",
887 .data = &hugepages_treat_as_movable,
888 .maxlen = sizeof(int),
889 .mode = 0644,
890 .proc_handler = &hugetlb_treat_movable_handler,
891 },
892 {
893 .ctl_name = CTL_UNNUMBERED,
894 .procname = "hugetlb_dynamic_pool",
895 .data = &hugetlb_dynamic_pool,
896 .maxlen = sizeof(hugetlb_dynamic_pool),
897 .mode = 0644,
898 .proc_handler = &proc_dointvec,
899 },
900 #endif
901 {
902 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
903 .procname = "lowmem_reserve_ratio",
904 .data = &sysctl_lowmem_reserve_ratio,
905 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
906 .mode = 0644,
907 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
908 .strategy = &sysctl_intvec,
909 },
910 {
911 .ctl_name = VM_DROP_PAGECACHE,
912 .procname = "drop_caches",
913 .data = &sysctl_drop_caches,
914 .maxlen = sizeof(int),
915 .mode = 0644,
916 .proc_handler = drop_caches_sysctl_handler,
917 .strategy = &sysctl_intvec,
918 },
919 {
920 .ctl_name = VM_MIN_FREE_KBYTES,
921 .procname = "min_free_kbytes",
922 .data = &min_free_kbytes,
923 .maxlen = sizeof(min_free_kbytes),
924 .mode = 0644,
925 .proc_handler = &min_free_kbytes_sysctl_handler,
926 .strategy = &sysctl_intvec,
927 .extra1 = &zero,
928 },
929 {
930 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
931 .procname = "percpu_pagelist_fraction",
932 .data = &percpu_pagelist_fraction,
933 .maxlen = sizeof(percpu_pagelist_fraction),
934 .mode = 0644,
935 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
936 .strategy = &sysctl_intvec,
937 .extra1 = &min_percpu_pagelist_fract,
938 },
939 #ifdef CONFIG_MMU
940 {
941 .ctl_name = VM_MAX_MAP_COUNT,
942 .procname = "max_map_count",
943 .data = &sysctl_max_map_count,
944 .maxlen = sizeof(sysctl_max_map_count),
945 .mode = 0644,
946 .proc_handler = &proc_dointvec
947 },
948 #endif
949 {
950 .ctl_name = VM_LAPTOP_MODE,
951 .procname = "laptop_mode",
952 .data = &laptop_mode,
953 .maxlen = sizeof(laptop_mode),
954 .mode = 0644,
955 .proc_handler = &proc_dointvec_jiffies,
956 .strategy = &sysctl_jiffies,
957 },
958 {
959 .ctl_name = VM_BLOCK_DUMP,
960 .procname = "block_dump",
961 .data = &block_dump,
962 .maxlen = sizeof(block_dump),
963 .mode = 0644,
964 .proc_handler = &proc_dointvec,
965 .strategy = &sysctl_intvec,
966 .extra1 = &zero,
967 },
968 {
969 .ctl_name = VM_VFS_CACHE_PRESSURE,
970 .procname = "vfs_cache_pressure",
971 .data = &sysctl_vfs_cache_pressure,
972 .maxlen = sizeof(sysctl_vfs_cache_pressure),
973 .mode = 0644,
974 .proc_handler = &proc_dointvec,
975 .strategy = &sysctl_intvec,
976 .extra1 = &zero,
977 },
978 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
979 {
980 .ctl_name = VM_LEGACY_VA_LAYOUT,
981 .procname = "legacy_va_layout",
982 .data = &sysctl_legacy_va_layout,
983 .maxlen = sizeof(sysctl_legacy_va_layout),
984 .mode = 0644,
985 .proc_handler = &proc_dointvec,
986 .strategy = &sysctl_intvec,
987 .extra1 = &zero,
988 },
989 #endif
990 #ifdef CONFIG_NUMA
991 {
992 .ctl_name = VM_ZONE_RECLAIM_MODE,
993 .procname = "zone_reclaim_mode",
994 .data = &zone_reclaim_mode,
995 .maxlen = sizeof(zone_reclaim_mode),
996 .mode = 0644,
997 .proc_handler = &proc_dointvec,
998 .strategy = &sysctl_intvec,
999 .extra1 = &zero,
1000 },
1001 {
1002 .ctl_name = VM_MIN_UNMAPPED,
1003 .procname = "min_unmapped_ratio",
1004 .data = &sysctl_min_unmapped_ratio,
1005 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1006 .mode = 0644,
1007 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
1008 .strategy = &sysctl_intvec,
1009 .extra1 = &zero,
1010 .extra2 = &one_hundred,
1011 },
1012 {
1013 .ctl_name = VM_MIN_SLAB,
1014 .procname = "min_slab_ratio",
1015 .data = &sysctl_min_slab_ratio,
1016 .maxlen = sizeof(sysctl_min_slab_ratio),
1017 .mode = 0644,
1018 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1019 .strategy = &sysctl_intvec,
1020 .extra1 = &zero,
1021 .extra2 = &one_hundred,
1022 },
1023 #endif
1024 #ifdef CONFIG_SMP
1025 {
1026 .ctl_name = CTL_UNNUMBERED,
1027 .procname = "stat_interval",
1028 .data = &sysctl_stat_interval,
1029 .maxlen = sizeof(sysctl_stat_interval),
1030 .mode = 0644,
1031 .proc_handler = &proc_dointvec_jiffies,
1032 .strategy = &sysctl_jiffies,
1033 },
1034 #endif
1035 #ifdef CONFIG_SECURITY
1036 {
1037 .ctl_name = CTL_UNNUMBERED,
1038 .procname = "mmap_min_addr",
1039 .data = &mmap_min_addr,
1040 .maxlen = sizeof(unsigned long),
1041 .mode = 0644,
1042 .proc_handler = &proc_doulongvec_minmax,
1043 },
1044 #endif
1045 #ifdef CONFIG_NUMA
1046 {
1047 .ctl_name = CTL_UNNUMBERED,
1048 .procname = "numa_zonelist_order",
1049 .data = &numa_zonelist_order,
1050 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1051 .mode = 0644,
1052 .proc_handler = &numa_zonelist_order_handler,
1053 .strategy = &sysctl_string,
1054 },
1055 #endif
1056 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1057 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1058 {
1059 .ctl_name = VM_VDSO_ENABLED,
1060 .procname = "vdso_enabled",
1061 .data = &vdso_enabled,
1062 .maxlen = sizeof(vdso_enabled),
1063 .mode = 0644,
1064 .proc_handler = &proc_dointvec,
1065 .strategy = &sysctl_intvec,
1066 .extra1 = &zero,
1067 },
1068 #endif
1069 /*
1070 * NOTE: do not add new entries to this table unless you have read
1071 * Documentation/sysctl/ctl_unnumbered.txt
1072 */
1073 { .ctl_name = 0 }
1074 };
1075
1076 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1077 static struct ctl_table binfmt_misc_table[] = {
1078 { .ctl_name = 0 }
1079 };
1080 #endif
1081
1082 static struct ctl_table fs_table[] = {
1083 {
1084 .ctl_name = FS_NRINODE,
1085 .procname = "inode-nr",
1086 .data = &inodes_stat,
1087 .maxlen = 2*sizeof(int),
1088 .mode = 0444,
1089 .proc_handler = &proc_dointvec,
1090 },
1091 {
1092 .ctl_name = FS_STATINODE,
1093 .procname = "inode-state",
1094 .data = &inodes_stat,
1095 .maxlen = 7*sizeof(int),
1096 .mode = 0444,
1097 .proc_handler = &proc_dointvec,
1098 },
1099 {
1100 .procname = "file-nr",
1101 .data = &files_stat,
1102 .maxlen = 3*sizeof(int),
1103 .mode = 0444,
1104 .proc_handler = &proc_nr_files,
1105 },
1106 {
1107 .ctl_name = FS_MAXFILE,
1108 .procname = "file-max",
1109 .data = &files_stat.max_files,
1110 .maxlen = sizeof(int),
1111 .mode = 0644,
1112 .proc_handler = &proc_dointvec,
1113 },
1114 {
1115 .ctl_name = FS_DENTRY,
1116 .procname = "dentry-state",
1117 .data = &dentry_stat,
1118 .maxlen = 6*sizeof(int),
1119 .mode = 0444,
1120 .proc_handler = &proc_dointvec,
1121 },
1122 {
1123 .ctl_name = FS_OVERFLOWUID,
1124 .procname = "overflowuid",
1125 .data = &fs_overflowuid,
1126 .maxlen = sizeof(int),
1127 .mode = 0644,
1128 .proc_handler = &proc_dointvec_minmax,
1129 .strategy = &sysctl_intvec,
1130 .extra1 = &minolduid,
1131 .extra2 = &maxolduid,
1132 },
1133 {
1134 .ctl_name = FS_OVERFLOWGID,
1135 .procname = "overflowgid",
1136 .data = &fs_overflowgid,
1137 .maxlen = sizeof(int),
1138 .mode = 0644,
1139 .proc_handler = &proc_dointvec_minmax,
1140 .strategy = &sysctl_intvec,
1141 .extra1 = &minolduid,
1142 .extra2 = &maxolduid,
1143 },
1144 {
1145 .ctl_name = FS_LEASES,
1146 .procname = "leases-enable",
1147 .data = &leases_enable,
1148 .maxlen = sizeof(int),
1149 .mode = 0644,
1150 .proc_handler = &proc_dointvec,
1151 },
1152 #ifdef CONFIG_DNOTIFY
1153 {
1154 .ctl_name = FS_DIR_NOTIFY,
1155 .procname = "dir-notify-enable",
1156 .data = &dir_notify_enable,
1157 .maxlen = sizeof(int),
1158 .mode = 0644,
1159 .proc_handler = &proc_dointvec,
1160 },
1161 #endif
1162 #ifdef CONFIG_MMU
1163 {
1164 .ctl_name = FS_LEASE_TIME,
1165 .procname = "lease-break-time",
1166 .data = &lease_break_time,
1167 .maxlen = sizeof(int),
1168 .mode = 0644,
1169 .proc_handler = &proc_dointvec_minmax,
1170 .strategy = &sysctl_intvec,
1171 .extra1 = &zero,
1172 .extra2 = &two,
1173 },
1174 {
1175 .procname = "aio-nr",
1176 .data = &aio_nr,
1177 .maxlen = sizeof(aio_nr),
1178 .mode = 0444,
1179 .proc_handler = &proc_doulongvec_minmax,
1180 },
1181 {
1182 .procname = "aio-max-nr",
1183 .data = &aio_max_nr,
1184 .maxlen = sizeof(aio_max_nr),
1185 .mode = 0644,
1186 .proc_handler = &proc_doulongvec_minmax,
1187 },
1188 #ifdef CONFIG_INOTIFY_USER
1189 {
1190 .ctl_name = FS_INOTIFY,
1191 .procname = "inotify",
1192 .mode = 0555,
1193 .child = inotify_table,
1194 },
1195 #endif
1196 #endif
1197 {
1198 .ctl_name = KERN_SETUID_DUMPABLE,
1199 .procname = "suid_dumpable",
1200 .data = &suid_dumpable,
1201 .maxlen = sizeof(int),
1202 .mode = 0644,
1203 .proc_handler = &proc_dointvec,
1204 },
1205 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1206 {
1207 .ctl_name = CTL_UNNUMBERED,
1208 .procname = "binfmt_misc",
1209 .mode = 0555,
1210 .child = binfmt_misc_table,
1211 },
1212 #endif
1213 /*
1214 * NOTE: do not add new entries to this table unless you have read
1215 * Documentation/sysctl/ctl_unnumbered.txt
1216 */
1217 { .ctl_name = 0 }
1218 };
1219
1220 static struct ctl_table debug_table[] = {
1221 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1222 {
1223 .ctl_name = CTL_UNNUMBERED,
1224 .procname = "exception-trace",
1225 .data = &show_unhandled_signals,
1226 .maxlen = sizeof(int),
1227 .mode = 0644,
1228 .proc_handler = proc_dointvec
1229 },
1230 #endif
1231 { .ctl_name = 0 }
1232 };
1233
1234 static struct ctl_table dev_table[] = {
1235 { .ctl_name = 0 }
1236 };
1237
1238 static DEFINE_SPINLOCK(sysctl_lock);
1239
1240 /* called under sysctl_lock */
1241 static int use_table(struct ctl_table_header *p)
1242 {
1243 if (unlikely(p->unregistering))
1244 return 0;
1245 p->used++;
1246 return 1;
1247 }
1248
1249 /* called under sysctl_lock */
1250 static void unuse_table(struct ctl_table_header *p)
1251 {
1252 if (!--p->used)
1253 if (unlikely(p->unregistering))
1254 complete(p->unregistering);
1255 }
1256
1257 /* called under sysctl_lock, will reacquire if has to wait */
1258 static void start_unregistering(struct ctl_table_header *p)
1259 {
1260 /*
1261 * if p->used is 0, nobody will ever touch that entry again;
1262 * we'll eliminate all paths to it before dropping sysctl_lock
1263 */
1264 if (unlikely(p->used)) {
1265 struct completion wait;
1266 init_completion(&wait);
1267 p->unregistering = &wait;
1268 spin_unlock(&sysctl_lock);
1269 wait_for_completion(&wait);
1270 spin_lock(&sysctl_lock);
1271 }
1272 /*
1273 * do not remove from the list until nobody holds it; walking the
1274 * list in do_sysctl() relies on that.
1275 */
1276 list_del_init(&p->ctl_entry);
1277 }
1278
1279 void sysctl_head_finish(struct ctl_table_header *head)
1280 {
1281 if (!head)
1282 return;
1283 spin_lock(&sysctl_lock);
1284 unuse_table(head);
1285 spin_unlock(&sysctl_lock);
1286 }
1287
1288 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1289 {
1290 struct ctl_table_header *head;
1291 struct list_head *tmp;
1292 spin_lock(&sysctl_lock);
1293 if (prev) {
1294 tmp = &prev->ctl_entry;
1295 unuse_table(prev);
1296 goto next;
1297 }
1298 tmp = &root_table_header.ctl_entry;
1299 for (;;) {
1300 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1301
1302 if (!use_table(head))
1303 goto next;
1304 spin_unlock(&sysctl_lock);
1305 return head;
1306 next:
1307 tmp = tmp->next;
1308 if (tmp == &root_table_header.ctl_entry)
1309 break;
1310 }
1311 spin_unlock(&sysctl_lock);
1312 return NULL;
1313 }
1314
1315 #ifdef CONFIG_SYSCTL_SYSCALL
1316 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1317 void __user *newval, size_t newlen)
1318 {
1319 struct ctl_table_header *head;
1320 int error = -ENOTDIR;
1321
1322 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1323 return -ENOTDIR;
1324 if (oldval) {
1325 int old_len;
1326 if (!oldlenp || get_user(old_len, oldlenp))
1327 return -EFAULT;
1328 }
1329
1330 for (head = sysctl_head_next(NULL); head;
1331 head = sysctl_head_next(head)) {
1332 error = parse_table(name, nlen, oldval, oldlenp,
1333 newval, newlen, head->ctl_table);
1334 if (error != -ENOTDIR) {
1335 sysctl_head_finish(head);
1336 break;
1337 }
1338 }
1339 return error;
1340 }
1341
1342 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1343 {
1344 struct __sysctl_args tmp;
1345 int error;
1346
1347 if (copy_from_user(&tmp, args, sizeof(tmp)))
1348 return -EFAULT;
1349
1350 lock_kernel();
1351 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1352 tmp.newval, tmp.newlen);
1353 unlock_kernel();
1354 return error;
1355 }
1356 #endif /* CONFIG_SYSCTL_SYSCALL */
1357
1358 /*
1359 * sysctl_perm does NOT grant the superuser all rights automatically, because
1360 * some sysctl variables are readonly even to root.
1361 */
1362
1363 static int test_perm(int mode, int op)
1364 {
1365 if (!current->euid)
1366 mode >>= 6;
1367 else if (in_egroup_p(0))
1368 mode >>= 3;
1369 if ((mode & op & 0007) == op)
1370 return 0;
1371 return -EACCES;
1372 }
1373
1374 int sysctl_perm(struct ctl_table *table, int op)
1375 {
1376 int error;
1377 error = security_sysctl(table, op);
1378 if (error)
1379 return error;
1380 return test_perm(table->mode, op);
1381 }
1382
1383 #ifdef CONFIG_SYSCTL_SYSCALL
1384 static int parse_table(int __user *name, int nlen,
1385 void __user *oldval, size_t __user *oldlenp,
1386 void __user *newval, size_t newlen,
1387 struct ctl_table *table)
1388 {
1389 int n;
1390 repeat:
1391 if (!nlen)
1392 return -ENOTDIR;
1393 if (get_user(n, name))
1394 return -EFAULT;
1395 for ( ; table->ctl_name || table->procname; table++) {
1396 if (!table->ctl_name)
1397 continue;
1398 if (n == table->ctl_name) {
1399 int error;
1400 if (table->child) {
1401 if (sysctl_perm(table, 001))
1402 return -EPERM;
1403 name++;
1404 nlen--;
1405 table = table->child;
1406 goto repeat;
1407 }
1408 error = do_sysctl_strategy(table, name, nlen,
1409 oldval, oldlenp,
1410 newval, newlen);
1411 return error;
1412 }
1413 }
1414 return -ENOTDIR;
1415 }
1416
1417 /* Perform the actual read/write of a sysctl table entry. */
1418 int do_sysctl_strategy (struct ctl_table *table,
1419 int __user *name, int nlen,
1420 void __user *oldval, size_t __user *oldlenp,
1421 void __user *newval, size_t newlen)
1422 {
1423 int op = 0, rc;
1424
1425 if (oldval)
1426 op |= 004;
1427 if (newval)
1428 op |= 002;
1429 if (sysctl_perm(table, op))
1430 return -EPERM;
1431
1432 if (table->strategy) {
1433 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1434 newval, newlen);
1435 if (rc < 0)
1436 return rc;
1437 if (rc > 0)
1438 return 0;
1439 }
1440
1441 /* If there is no strategy routine, or if the strategy returns
1442 * zero, proceed with automatic r/w */
1443 if (table->data && table->maxlen) {
1444 rc = sysctl_data(table, name, nlen, oldval, oldlenp,
1445 newval, newlen);
1446 if (rc < 0)
1447 return rc;
1448 }
1449 return 0;
1450 }
1451 #endif /* CONFIG_SYSCTL_SYSCALL */
1452
1453 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1454 {
1455 for (; table->ctl_name || table->procname; table++) {
1456 table->parent = parent;
1457 if (table->child)
1458 sysctl_set_parent(table, table->child);
1459 }
1460 }
1461
1462 static __init int sysctl_init(void)
1463 {
1464 int err;
1465 sysctl_set_parent(NULL, root_table);
1466 err = sysctl_check_table(root_table);
1467 return 0;
1468 }
1469
1470 core_initcall(sysctl_init);
1471
1472 /**
1473 * register_sysctl_table - register a sysctl hierarchy
1474 * @table: the top-level table structure
1475 *
1476 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1477 * array. An entry with a ctl_name of 0 terminates the table.
1478 *
1479 * The members of the &struct ctl_table structure are used as follows:
1480 *
1481 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1482 * must be unique within that level of sysctl
1483 *
1484 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1485 * enter a sysctl file
1486 *
1487 * data - a pointer to data for use by proc_handler
1488 *
1489 * maxlen - the maximum size in bytes of the data
1490 *
1491 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1492 *
1493 * child - a pointer to the child sysctl table if this entry is a directory, or
1494 * %NULL.
1495 *
1496 * proc_handler - the text handler routine (described below)
1497 *
1498 * strategy - the strategy routine (described below)
1499 *
1500 * de - for internal use by the sysctl routines
1501 *
1502 * extra1, extra2 - extra pointers usable by the proc handler routines
1503 *
1504 * Leaf nodes in the sysctl tree will be represented by a single file
1505 * under /proc; non-leaf nodes will be represented by directories.
1506 *
1507 * sysctl(2) can automatically manage read and write requests through
1508 * the sysctl table. The data and maxlen fields of the ctl_table
1509 * struct enable minimal validation of the values being written to be
1510 * performed, and the mode field allows minimal authentication.
1511 *
1512 * More sophisticated management can be enabled by the provision of a
1513 * strategy routine with the table entry. This will be called before
1514 * any automatic read or write of the data is performed.
1515 *
1516 * The strategy routine may return
1517 *
1518 * < 0 - Error occurred (error is passed to user process)
1519 *
1520 * 0 - OK - proceed with automatic read or write.
1521 *
1522 * > 0 - OK - read or write has been done by the strategy routine, so
1523 * return immediately.
1524 *
1525 * There must be a proc_handler routine for any terminal nodes
1526 * mirrored under /proc/sys (non-terminals are handled by a built-in
1527 * directory handler). Several default handlers are available to
1528 * cover common cases -
1529 *
1530 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1531 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1532 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1533 *
1534 * It is the handler's job to read the input buffer from user memory
1535 * and process it. The handler should return 0 on success.
1536 *
1537 * This routine returns %NULL on a failure to register, and a pointer
1538 * to the table header on success.
1539 */
1540 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1541 {
1542 struct ctl_table_header *tmp;
1543 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1544 if (!tmp)
1545 return NULL;
1546 tmp->ctl_table = table;
1547 INIT_LIST_HEAD(&tmp->ctl_entry);
1548 tmp->used = 0;
1549 tmp->unregistering = NULL;
1550 sysctl_set_parent(NULL, table);
1551 if (sysctl_check_table(tmp->ctl_table)) {
1552 kfree(tmp);
1553 return NULL;
1554 }
1555 spin_lock(&sysctl_lock);
1556 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
1557 spin_unlock(&sysctl_lock);
1558 return tmp;
1559 }
1560
1561 /**
1562 * unregister_sysctl_table - unregister a sysctl table hierarchy
1563 * @header: the header returned from register_sysctl_table
1564 *
1565 * Unregisters the sysctl table and all children. proc entries may not
1566 * actually be removed until they are no longer used by anyone.
1567 */
1568 void unregister_sysctl_table(struct ctl_table_header * header)
1569 {
1570 might_sleep();
1571 spin_lock(&sysctl_lock);
1572 start_unregistering(header);
1573 spin_unlock(&sysctl_lock);
1574 kfree(header);
1575 }
1576
1577 #else /* !CONFIG_SYSCTL */
1578 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1579 {
1580 return NULL;
1581 }
1582
1583 void unregister_sysctl_table(struct ctl_table_header * table)
1584 {
1585 }
1586
1587 #endif /* CONFIG_SYSCTL */
1588
1589 /*
1590 * /proc/sys support
1591 */
1592
1593 #ifdef CONFIG_PROC_SYSCTL
1594
1595 static int _proc_do_string(void* data, int maxlen, int write,
1596 struct file *filp, void __user *buffer,
1597 size_t *lenp, loff_t *ppos)
1598 {
1599 size_t len;
1600 char __user *p;
1601 char c;
1602
1603 if (!data || !maxlen || !*lenp) {
1604 *lenp = 0;
1605 return 0;
1606 }
1607
1608 if (write) {
1609 len = 0;
1610 p = buffer;
1611 while (len < *lenp) {
1612 if (get_user(c, p++))
1613 return -EFAULT;
1614 if (c == 0 || c == '\n')
1615 break;
1616 len++;
1617 }
1618 if (len >= maxlen)
1619 len = maxlen-1;
1620 if(copy_from_user(data, buffer, len))
1621 return -EFAULT;
1622 ((char *) data)[len] = 0;
1623 *ppos += *lenp;
1624 } else {
1625 len = strlen(data);
1626 if (len > maxlen)
1627 len = maxlen;
1628
1629 if (*ppos > len) {
1630 *lenp = 0;
1631 return 0;
1632 }
1633
1634 data += *ppos;
1635 len -= *ppos;
1636
1637 if (len > *lenp)
1638 len = *lenp;
1639 if (len)
1640 if(copy_to_user(buffer, data, len))
1641 return -EFAULT;
1642 if (len < *lenp) {
1643 if(put_user('\n', ((char __user *) buffer) + len))
1644 return -EFAULT;
1645 len++;
1646 }
1647 *lenp = len;
1648 *ppos += len;
1649 }
1650 return 0;
1651 }
1652
1653 /**
1654 * proc_dostring - read a string sysctl
1655 * @table: the sysctl table
1656 * @write: %TRUE if this is a write to the sysctl file
1657 * @filp: the file structure
1658 * @buffer: the user buffer
1659 * @lenp: the size of the user buffer
1660 * @ppos: file position
1661 *
1662 * Reads/writes a string from/to the user buffer. If the kernel
1663 * buffer provided is not large enough to hold the string, the
1664 * string is truncated. The copied string is %NULL-terminated.
1665 * If the string is being read by the user process, it is copied
1666 * and a newline '\n' is added. It is truncated if the buffer is
1667 * not large enough.
1668 *
1669 * Returns 0 on success.
1670 */
1671 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
1672 void __user *buffer, size_t *lenp, loff_t *ppos)
1673 {
1674 return _proc_do_string(table->data, table->maxlen, write, filp,
1675 buffer, lenp, ppos);
1676 }
1677
1678
1679 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1680 int *valp,
1681 int write, void *data)
1682 {
1683 if (write) {
1684 *valp = *negp ? -*lvalp : *lvalp;
1685 } else {
1686 int val = *valp;
1687 if (val < 0) {
1688 *negp = -1;
1689 *lvalp = (unsigned long)-val;
1690 } else {
1691 *negp = 0;
1692 *lvalp = (unsigned long)val;
1693 }
1694 }
1695 return 0;
1696 }
1697
1698 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1699 int write, struct file *filp, void __user *buffer,
1700 size_t *lenp, loff_t *ppos,
1701 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1702 int write, void *data),
1703 void *data)
1704 {
1705 #define TMPBUFLEN 21
1706 int *i, vleft, first=1, neg, val;
1707 unsigned long lval;
1708 size_t left, len;
1709
1710 char buf[TMPBUFLEN], *p;
1711 char __user *s = buffer;
1712
1713 if (!tbl_data || !table->maxlen || !*lenp ||
1714 (*ppos && !write)) {
1715 *lenp = 0;
1716 return 0;
1717 }
1718
1719 i = (int *) tbl_data;
1720 vleft = table->maxlen / sizeof(*i);
1721 left = *lenp;
1722
1723 if (!conv)
1724 conv = do_proc_dointvec_conv;
1725
1726 for (; left && vleft--; i++, first=0) {
1727 if (write) {
1728 while (left) {
1729 char c;
1730 if (get_user(c, s))
1731 return -EFAULT;
1732 if (!isspace(c))
1733 break;
1734 left--;
1735 s++;
1736 }
1737 if (!left)
1738 break;
1739 neg = 0;
1740 len = left;
1741 if (len > sizeof(buf) - 1)
1742 len = sizeof(buf) - 1;
1743 if (copy_from_user(buf, s, len))
1744 return -EFAULT;
1745 buf[len] = 0;
1746 p = buf;
1747 if (*p == '-' && left > 1) {
1748 neg = 1;
1749 p++;
1750 }
1751 if (*p < '0' || *p > '9')
1752 break;
1753
1754 lval = simple_strtoul(p, &p, 0);
1755
1756 len = p-buf;
1757 if ((len < left) && *p && !isspace(*p))
1758 break;
1759 if (neg)
1760 val = -val;
1761 s += len;
1762 left -= len;
1763
1764 if (conv(&neg, &lval, i, 1, data))
1765 break;
1766 } else {
1767 p = buf;
1768 if (!first)
1769 *p++ = '\t';
1770
1771 if (conv(&neg, &lval, i, 0, data))
1772 break;
1773
1774 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1775 len = strlen(buf);
1776 if (len > left)
1777 len = left;
1778 if(copy_to_user(s, buf, len))
1779 return -EFAULT;
1780 left -= len;
1781 s += len;
1782 }
1783 }
1784
1785 if (!write && !first && left) {
1786 if(put_user('\n', s))
1787 return -EFAULT;
1788 left--, s++;
1789 }
1790 if (write) {
1791 while (left) {
1792 char c;
1793 if (get_user(c, s++))
1794 return -EFAULT;
1795 if (!isspace(c))
1796 break;
1797 left--;
1798 }
1799 }
1800 if (write && first)
1801 return -EINVAL;
1802 *lenp -= left;
1803 *ppos += *lenp;
1804 return 0;
1805 #undef TMPBUFLEN
1806 }
1807
1808 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
1809 void __user *buffer, size_t *lenp, loff_t *ppos,
1810 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1811 int write, void *data),
1812 void *data)
1813 {
1814 return __do_proc_dointvec(table->data, table, write, filp,
1815 buffer, lenp, ppos, conv, data);
1816 }
1817
1818 /**
1819 * proc_dointvec - read a vector of integers
1820 * @table: the sysctl table
1821 * @write: %TRUE if this is a write to the sysctl file
1822 * @filp: the file structure
1823 * @buffer: the user buffer
1824 * @lenp: the size of the user buffer
1825 * @ppos: file position
1826 *
1827 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1828 * values from/to the user buffer, treated as an ASCII string.
1829 *
1830 * Returns 0 on success.
1831 */
1832 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
1833 void __user *buffer, size_t *lenp, loff_t *ppos)
1834 {
1835 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1836 NULL,NULL);
1837 }
1838
1839 #define OP_SET 0
1840 #define OP_AND 1
1841 #define OP_OR 2
1842
1843 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1844 int *valp,
1845 int write, void *data)
1846 {
1847 int op = *(int *)data;
1848 if (write) {
1849 int val = *negp ? -*lvalp : *lvalp;
1850 switch(op) {
1851 case OP_SET: *valp = val; break;
1852 case OP_AND: *valp &= val; break;
1853 case OP_OR: *valp |= val; break;
1854 }
1855 } else {
1856 int val = *valp;
1857 if (val < 0) {
1858 *negp = -1;
1859 *lvalp = (unsigned long)-val;
1860 } else {
1861 *negp = 0;
1862 *lvalp = (unsigned long)val;
1863 }
1864 }
1865 return 0;
1866 }
1867
1868 /*
1869 * init may raise the set.
1870 */
1871
1872 int proc_dointvec_bset(struct ctl_table *table, int write, struct file *filp,
1873 void __user *buffer, size_t *lenp, loff_t *ppos)
1874 {
1875 int op;
1876
1877 if (write && !capable(CAP_SYS_MODULE)) {
1878 return -EPERM;
1879 }
1880
1881 op = is_init(current) ? OP_SET : OP_AND;
1882 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1883 do_proc_dointvec_bset_conv,&op);
1884 }
1885
1886 /*
1887 * Taint values can only be increased
1888 */
1889 static int proc_dointvec_taint(struct ctl_table *table, int write, struct file *filp,
1890 void __user *buffer, size_t *lenp, loff_t *ppos)
1891 {
1892 int op;
1893
1894 if (write && !capable(CAP_SYS_ADMIN))
1895 return -EPERM;
1896
1897 op = OP_OR;
1898 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1899 do_proc_dointvec_bset_conv,&op);
1900 }
1901
1902 struct do_proc_dointvec_minmax_conv_param {
1903 int *min;
1904 int *max;
1905 };
1906
1907 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1908 int *valp,
1909 int write, void *data)
1910 {
1911 struct do_proc_dointvec_minmax_conv_param *param = data;
1912 if (write) {
1913 int val = *negp ? -*lvalp : *lvalp;
1914 if ((param->min && *param->min > val) ||
1915 (param->max && *param->max < val))
1916 return -EINVAL;
1917 *valp = val;
1918 } else {
1919 int val = *valp;
1920 if (val < 0) {
1921 *negp = -1;
1922 *lvalp = (unsigned long)-val;
1923 } else {
1924 *negp = 0;
1925 *lvalp = (unsigned long)val;
1926 }
1927 }
1928 return 0;
1929 }
1930
1931 /**
1932 * proc_dointvec_minmax - read a vector of integers with min/max values
1933 * @table: the sysctl table
1934 * @write: %TRUE if this is a write to the sysctl file
1935 * @filp: the file structure
1936 * @buffer: the user buffer
1937 * @lenp: the size of the user buffer
1938 * @ppos: file position
1939 *
1940 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1941 * values from/to the user buffer, treated as an ASCII string.
1942 *
1943 * This routine will ensure the values are within the range specified by
1944 * table->extra1 (min) and table->extra2 (max).
1945 *
1946 * Returns 0 on success.
1947 */
1948 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
1949 void __user *buffer, size_t *lenp, loff_t *ppos)
1950 {
1951 struct do_proc_dointvec_minmax_conv_param param = {
1952 .min = (int *) table->extra1,
1953 .max = (int *) table->extra2,
1954 };
1955 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1956 do_proc_dointvec_minmax_conv, &param);
1957 }
1958
1959 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
1960 struct file *filp,
1961 void __user *buffer,
1962 size_t *lenp, loff_t *ppos,
1963 unsigned long convmul,
1964 unsigned long convdiv)
1965 {
1966 #define TMPBUFLEN 21
1967 unsigned long *i, *min, *max, val;
1968 int vleft, first=1, neg;
1969 size_t len, left;
1970 char buf[TMPBUFLEN], *p;
1971 char __user *s = buffer;
1972
1973 if (!data || !table->maxlen || !*lenp ||
1974 (*ppos && !write)) {
1975 *lenp = 0;
1976 return 0;
1977 }
1978
1979 i = (unsigned long *) data;
1980 min = (unsigned long *) table->extra1;
1981 max = (unsigned long *) table->extra2;
1982 vleft = table->maxlen / sizeof(unsigned long);
1983 left = *lenp;
1984
1985 for (; left && vleft--; i++, min++, max++, first=0) {
1986 if (write) {
1987 while (left) {
1988 char c;
1989 if (get_user(c, s))
1990 return -EFAULT;
1991 if (!isspace(c))
1992 break;
1993 left--;
1994 s++;
1995 }
1996 if (!left)
1997 break;
1998 neg = 0;
1999 len = left;
2000 if (len > TMPBUFLEN-1)
2001 len = TMPBUFLEN-1;
2002 if (copy_from_user(buf, s, len))
2003 return -EFAULT;
2004 buf[len] = 0;
2005 p = buf;
2006 if (*p == '-' && left > 1) {
2007 neg = 1;
2008 p++;
2009 }
2010 if (*p < '0' || *p > '9')
2011 break;
2012 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2013 len = p-buf;
2014 if ((len < left) && *p && !isspace(*p))
2015 break;
2016 if (neg)
2017 val = -val;
2018 s += len;
2019 left -= len;
2020
2021 if(neg)
2022 continue;
2023 if ((min && val < *min) || (max && val > *max))
2024 continue;
2025 *i = val;
2026 } else {
2027 p = buf;
2028 if (!first)
2029 *p++ = '\t';
2030 sprintf(p, "%lu", convdiv * (*i) / convmul);
2031 len = strlen(buf);
2032 if (len > left)
2033 len = left;
2034 if(copy_to_user(s, buf, len))
2035 return -EFAULT;
2036 left -= len;
2037 s += len;
2038 }
2039 }
2040
2041 if (!write && !first && left) {
2042 if(put_user('\n', s))
2043 return -EFAULT;
2044 left--, s++;
2045 }
2046 if (write) {
2047 while (left) {
2048 char c;
2049 if (get_user(c, s++))
2050 return -EFAULT;
2051 if (!isspace(c))
2052 break;
2053 left--;
2054 }
2055 }
2056 if (write && first)
2057 return -EINVAL;
2058 *lenp -= left;
2059 *ppos += *lenp;
2060 return 0;
2061 #undef TMPBUFLEN
2062 }
2063
2064 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2065 struct file *filp,
2066 void __user *buffer,
2067 size_t *lenp, loff_t *ppos,
2068 unsigned long convmul,
2069 unsigned long convdiv)
2070 {
2071 return __do_proc_doulongvec_minmax(table->data, table, write,
2072 filp, buffer, lenp, ppos, convmul, convdiv);
2073 }
2074
2075 /**
2076 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2077 * @table: the sysctl table
2078 * @write: %TRUE if this is a write to the sysctl file
2079 * @filp: the file structure
2080 * @buffer: the user buffer
2081 * @lenp: the size of the user buffer
2082 * @ppos: file position
2083 *
2084 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2085 * values from/to the user buffer, treated as an ASCII string.
2086 *
2087 * This routine will ensure the values are within the range specified by
2088 * table->extra1 (min) and table->extra2 (max).
2089 *
2090 * Returns 0 on success.
2091 */
2092 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2093 void __user *buffer, size_t *lenp, loff_t *ppos)
2094 {
2095 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2096 }
2097
2098 /**
2099 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2100 * @table: the sysctl table
2101 * @write: %TRUE if this is a write to the sysctl file
2102 * @filp: the file structure
2103 * @buffer: the user buffer
2104 * @lenp: the size of the user buffer
2105 * @ppos: file position
2106 *
2107 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2108 * values from/to the user buffer, treated as an ASCII string. The values
2109 * are treated as milliseconds, and converted to jiffies when they are stored.
2110 *
2111 * This routine will ensure the values are within the range specified by
2112 * table->extra1 (min) and table->extra2 (max).
2113 *
2114 * Returns 0 on success.
2115 */
2116 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2117 struct file *filp,
2118 void __user *buffer,
2119 size_t *lenp, loff_t *ppos)
2120 {
2121 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2122 lenp, ppos, HZ, 1000l);
2123 }
2124
2125
2126 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2127 int *valp,
2128 int write, void *data)
2129 {
2130 if (write) {
2131 if (*lvalp > LONG_MAX / HZ)
2132 return 1;
2133 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2134 } else {
2135 int val = *valp;
2136 unsigned long lval;
2137 if (val < 0) {
2138 *negp = -1;
2139 lval = (unsigned long)-val;
2140 } else {
2141 *negp = 0;
2142 lval = (unsigned long)val;
2143 }
2144 *lvalp = lval / HZ;
2145 }
2146 return 0;
2147 }
2148
2149 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2150 int *valp,
2151 int write, void *data)
2152 {
2153 if (write) {
2154 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2155 return 1;
2156 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2157 } else {
2158 int val = *valp;
2159 unsigned long lval;
2160 if (val < 0) {
2161 *negp = -1;
2162 lval = (unsigned long)-val;
2163 } else {
2164 *negp = 0;
2165 lval = (unsigned long)val;
2166 }
2167 *lvalp = jiffies_to_clock_t(lval);
2168 }
2169 return 0;
2170 }
2171
2172 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2173 int *valp,
2174 int write, void *data)
2175 {
2176 if (write) {
2177 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2178 } else {
2179 int val = *valp;
2180 unsigned long lval;
2181 if (val < 0) {
2182 *negp = -1;
2183 lval = (unsigned long)-val;
2184 } else {
2185 *negp = 0;
2186 lval = (unsigned long)val;
2187 }
2188 *lvalp = jiffies_to_msecs(lval);
2189 }
2190 return 0;
2191 }
2192
2193 /**
2194 * proc_dointvec_jiffies - read a vector of integers as seconds
2195 * @table: the sysctl table
2196 * @write: %TRUE if this is a write to the sysctl file
2197 * @filp: the file structure
2198 * @buffer: the user buffer
2199 * @lenp: the size of the user buffer
2200 * @ppos: file position
2201 *
2202 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2203 * values from/to the user buffer, treated as an ASCII string.
2204 * The values read are assumed to be in seconds, and are converted into
2205 * jiffies.
2206 *
2207 * Returns 0 on success.
2208 */
2209 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2210 void __user *buffer, size_t *lenp, loff_t *ppos)
2211 {
2212 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2213 do_proc_dointvec_jiffies_conv,NULL);
2214 }
2215
2216 /**
2217 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2218 * @table: the sysctl table
2219 * @write: %TRUE if this is a write to the sysctl file
2220 * @filp: the file structure
2221 * @buffer: the user buffer
2222 * @lenp: the size of the user buffer
2223 * @ppos: pointer to the file position
2224 *
2225 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2226 * values from/to the user buffer, treated as an ASCII string.
2227 * The values read are assumed to be in 1/USER_HZ seconds, and
2228 * are converted into jiffies.
2229 *
2230 * Returns 0 on success.
2231 */
2232 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2233 void __user *buffer, size_t *lenp, loff_t *ppos)
2234 {
2235 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2236 do_proc_dointvec_userhz_jiffies_conv,NULL);
2237 }
2238
2239 /**
2240 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2241 * @table: the sysctl table
2242 * @write: %TRUE if this is a write to the sysctl file
2243 * @filp: the file structure
2244 * @buffer: the user buffer
2245 * @lenp: the size of the user buffer
2246 * @ppos: file position
2247 * @ppos: the current position in the file
2248 *
2249 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2250 * values from/to the user buffer, treated as an ASCII string.
2251 * The values read are assumed to be in 1/1000 seconds, and
2252 * are converted into jiffies.
2253 *
2254 * Returns 0 on success.
2255 */
2256 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2257 void __user *buffer, size_t *lenp, loff_t *ppos)
2258 {
2259 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2260 do_proc_dointvec_ms_jiffies_conv, NULL);
2261 }
2262
2263 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2264 void __user *buffer, size_t *lenp, loff_t *ppos)
2265 {
2266 struct pid *new_pid;
2267 pid_t tmp;
2268 int r;
2269
2270 tmp = pid_nr(cad_pid);
2271
2272 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2273 lenp, ppos, NULL, NULL);
2274 if (r || !write)
2275 return r;
2276
2277 new_pid = find_get_pid(tmp);
2278 if (!new_pid)
2279 return -ESRCH;
2280
2281 put_pid(xchg(&cad_pid, new_pid));
2282 return 0;
2283 }
2284
2285 #else /* CONFIG_PROC_FS */
2286
2287 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2288 void __user *buffer, size_t *lenp, loff_t *ppos)
2289 {
2290 return -ENOSYS;
2291 }
2292
2293 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2294 void __user *buffer, size_t *lenp, loff_t *ppos)
2295 {
2296 return -ENOSYS;
2297 }
2298
2299 int proc_dointvec_bset(struct ctl_table *table, int write, struct file *filp,
2300 void __user *buffer, size_t *lenp, loff_t *ppos)
2301 {
2302 return -ENOSYS;
2303 }
2304
2305 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2306 void __user *buffer, size_t *lenp, loff_t *ppos)
2307 {
2308 return -ENOSYS;
2309 }
2310
2311 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2312 void __user *buffer, size_t *lenp, loff_t *ppos)
2313 {
2314 return -ENOSYS;
2315 }
2316
2317 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2318 void __user *buffer, size_t *lenp, loff_t *ppos)
2319 {
2320 return -ENOSYS;
2321 }
2322
2323 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2324 void __user *buffer, size_t *lenp, loff_t *ppos)
2325 {
2326 return -ENOSYS;
2327 }
2328
2329 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2330 void __user *buffer, size_t *lenp, loff_t *ppos)
2331 {
2332 return -ENOSYS;
2333 }
2334
2335 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2336 struct file *filp,
2337 void __user *buffer,
2338 size_t *lenp, loff_t *ppos)
2339 {
2340 return -ENOSYS;
2341 }
2342
2343
2344 #endif /* CONFIG_PROC_FS */
2345
2346
2347 #ifdef CONFIG_SYSCTL_SYSCALL
2348 /*
2349 * General sysctl support routines
2350 */
2351
2352 /* The generic sysctl data routine (used if no strategy routine supplied) */
2353 int sysctl_data(struct ctl_table *table, int __user *name, int nlen,
2354 void __user *oldval, size_t __user *oldlenp,
2355 void __user *newval, size_t newlen)
2356 {
2357 size_t len;
2358
2359 /* Get out of I don't have a variable */
2360 if (!table->data || !table->maxlen)
2361 return -ENOTDIR;
2362
2363 if (oldval && oldlenp) {
2364 if (get_user(len, oldlenp))
2365 return -EFAULT;
2366 if (len) {
2367 if (len > table->maxlen)
2368 len = table->maxlen;
2369 if (copy_to_user(oldval, table->data, len))
2370 return -EFAULT;
2371 if (put_user(len, oldlenp))
2372 return -EFAULT;
2373 }
2374 }
2375
2376 if (newval && newlen) {
2377 if (newlen > table->maxlen)
2378 newlen = table->maxlen;
2379
2380 if (copy_from_user(table->data, newval, newlen))
2381 return -EFAULT;
2382 }
2383 return 1;
2384 }
2385
2386 /* The generic string strategy routine: */
2387 int sysctl_string(struct ctl_table *table, int __user *name, int nlen,
2388 void __user *oldval, size_t __user *oldlenp,
2389 void __user *newval, size_t newlen)
2390 {
2391 if (!table->data || !table->maxlen)
2392 return -ENOTDIR;
2393
2394 if (oldval && oldlenp) {
2395 size_t bufsize;
2396 if (get_user(bufsize, oldlenp))
2397 return -EFAULT;
2398 if (bufsize) {
2399 size_t len = strlen(table->data), copied;
2400
2401 /* This shouldn't trigger for a well-formed sysctl */
2402 if (len > table->maxlen)
2403 len = table->maxlen;
2404
2405 /* Copy up to a max of bufsize-1 bytes of the string */
2406 copied = (len >= bufsize) ? bufsize - 1 : len;
2407
2408 if (copy_to_user(oldval, table->data, copied) ||
2409 put_user(0, (char __user *)(oldval + copied)))
2410 return -EFAULT;
2411 if (put_user(len, oldlenp))
2412 return -EFAULT;
2413 }
2414 }
2415 if (newval && newlen) {
2416 size_t len = newlen;
2417 if (len > table->maxlen)
2418 len = table->maxlen;
2419 if(copy_from_user(table->data, newval, len))
2420 return -EFAULT;
2421 if (len == table->maxlen)
2422 len--;
2423 ((char *) table->data)[len] = 0;
2424 }
2425 return 1;
2426 }
2427
2428 /*
2429 * This function makes sure that all of the integers in the vector
2430 * are between the minimum and maximum values given in the arrays
2431 * table->extra1 and table->extra2, respectively.
2432 */
2433 int sysctl_intvec(struct ctl_table *table, int __user *name, int nlen,
2434 void __user *oldval, size_t __user *oldlenp,
2435 void __user *newval, size_t newlen)
2436 {
2437
2438 if (newval && newlen) {
2439 int __user *vec = (int __user *) newval;
2440 int *min = (int *) table->extra1;
2441 int *max = (int *) table->extra2;
2442 size_t length;
2443 int i;
2444
2445 if (newlen % sizeof(int) != 0)
2446 return -EINVAL;
2447
2448 if (!table->extra1 && !table->extra2)
2449 return 0;
2450
2451 if (newlen > table->maxlen)
2452 newlen = table->maxlen;
2453 length = newlen / sizeof(int);
2454
2455 for (i = 0; i < length; i++) {
2456 int value;
2457 if (get_user(value, vec + i))
2458 return -EFAULT;
2459 if (min && value < min[i])
2460 return -EINVAL;
2461 if (max && value > max[i])
2462 return -EINVAL;
2463 }
2464 }
2465 return 0;
2466 }
2467
2468 /* Strategy function to convert jiffies to seconds */
2469 int sysctl_jiffies(struct ctl_table *table, int __user *name, int nlen,
2470 void __user *oldval, size_t __user *oldlenp,
2471 void __user *newval, size_t newlen)
2472 {
2473 if (oldval && oldlenp) {
2474 size_t olen;
2475
2476 if (get_user(olen, oldlenp))
2477 return -EFAULT;
2478 if (olen) {
2479 int val;
2480
2481 if (olen < sizeof(int))
2482 return -EINVAL;
2483
2484 val = *(int *)(table->data) / HZ;
2485 if (put_user(val, (int __user *)oldval))
2486 return -EFAULT;
2487 if (put_user(sizeof(int), oldlenp))
2488 return -EFAULT;
2489 }
2490 }
2491 if (newval && newlen) {
2492 int new;
2493 if (newlen != sizeof(int))
2494 return -EINVAL;
2495 if (get_user(new, (int __user *)newval))
2496 return -EFAULT;
2497 *(int *)(table->data) = new*HZ;
2498 }
2499 return 1;
2500 }
2501
2502 /* Strategy function to convert jiffies to seconds */
2503 int sysctl_ms_jiffies(struct ctl_table *table, int __user *name, int nlen,
2504 void __user *oldval, size_t __user *oldlenp,
2505 void __user *newval, size_t newlen)
2506 {
2507 if (oldval && oldlenp) {
2508 size_t olen;
2509
2510 if (get_user(olen, oldlenp))
2511 return -EFAULT;
2512 if (olen) {
2513 int val;
2514
2515 if (olen < sizeof(int))
2516 return -EINVAL;
2517
2518 val = jiffies_to_msecs(*(int *)(table->data));
2519 if (put_user(val, (int __user *)oldval))
2520 return -EFAULT;
2521 if (put_user(sizeof(int), oldlenp))
2522 return -EFAULT;
2523 }
2524 }
2525 if (newval && newlen) {
2526 int new;
2527 if (newlen != sizeof(int))
2528 return -EINVAL;
2529 if (get_user(new, (int __user *)newval))
2530 return -EFAULT;
2531 *(int *)(table->data) = msecs_to_jiffies(new);
2532 }
2533 return 1;
2534 }
2535
2536
2537
2538 #else /* CONFIG_SYSCTL_SYSCALL */
2539
2540
2541 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2542 {
2543 static int msg_count;
2544 struct __sysctl_args tmp;
2545 int name[CTL_MAXNAME];
2546 int i;
2547
2548 /* Read in the sysctl name for better debug message logging */
2549 if (copy_from_user(&tmp, args, sizeof(tmp)))
2550 return -EFAULT;
2551 if (tmp.nlen <= 0 || tmp.nlen >= CTL_MAXNAME)
2552 return -ENOTDIR;
2553 for (i = 0; i < tmp.nlen; i++)
2554 if (get_user(name[i], tmp.name + i))
2555 return -EFAULT;
2556
2557 /* Ignore accesses to kernel.version */
2558 if ((tmp.nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2559 goto out;
2560
2561 if (msg_count < 5) {
2562 msg_count++;
2563 printk(KERN_INFO
2564 "warning: process `%s' used the removed sysctl "
2565 "system call with ", current->comm);
2566 for (i = 0; i < tmp.nlen; i++)
2567 printk("%d.", name[i]);
2568 printk("\n");
2569 }
2570 out:
2571 return -ENOSYS;
2572 }
2573
2574 int sysctl_data(struct ctl_table *table, int __user *name, int nlen,
2575 void __user *oldval, size_t __user *oldlenp,
2576 void __user *newval, size_t newlen)
2577 {
2578 return -ENOSYS;
2579 }
2580
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 return -ENOSYS;
2586 }
2587
2588 int sysctl_intvec(struct ctl_table *table, int __user *name, int nlen,
2589 void __user *oldval, size_t __user *oldlenp,
2590 void __user *newval, size_t newlen)
2591 {
2592 return -ENOSYS;
2593 }
2594
2595 int sysctl_jiffies(struct ctl_table *table, int __user *name, int nlen,
2596 void __user *oldval, size_t __user *oldlenp,
2597 void __user *newval, size_t newlen)
2598 {
2599 return -ENOSYS;
2600 }
2601
2602 int sysctl_ms_jiffies(struct ctl_table *table, int __user *name, int nlen,
2603 void __user *oldval, size_t __user *oldlenp,
2604 void __user *newval, size_t newlen)
2605 {
2606 return -ENOSYS;
2607 }
2608
2609 #endif /* CONFIG_SYSCTL_SYSCALL */
2610
2611 /*
2612 * No sense putting this after each symbol definition, twice,
2613 * exception granted :-)
2614 */
2615 EXPORT_SYMBOL(proc_dointvec);
2616 EXPORT_SYMBOL(proc_dointvec_jiffies);
2617 EXPORT_SYMBOL(proc_dointvec_minmax);
2618 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2619 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2620 EXPORT_SYMBOL(proc_dostring);
2621 EXPORT_SYMBOL(proc_doulongvec_minmax);
2622 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2623 EXPORT_SYMBOL(register_sysctl_table);
2624 EXPORT_SYMBOL(sysctl_intvec);
2625 EXPORT_SYMBOL(sysctl_jiffies);
2626 EXPORT_SYMBOL(sysctl_ms_jiffies);
2627 EXPORT_SYMBOL(sysctl_string);
2628 EXPORT_SYMBOL(sysctl_data);
2629 EXPORT_SYMBOL(unregister_sysctl_table);
(deprecated) Btrfs devel tree