]> bbs.cooldavid.org Git - net-next-2.6.git/blame - kernel/sysctl.c
[PATCH] i386/x86-64: Remove un/set_nmi_callback and reserve/release_lapic_nmi functions
[net-next-2.6.git] / kernel / sysctl.c
CommitLineData
1da177e4
LT
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
1da177e4
LT
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>
c59ede7b 27#include <linux/capability.h>
1da177e4
LT
28#include <linux/ctype.h>
29#include <linux/utsname.h>
30#include <linux/capability.h>
31#include <linux/smp_lock.h>
32#include <linux/init.h>
33#include <linux/kernel.h>
0296b228 34#include <linux/kobject.h>
20380731 35#include <linux/net.h>
1da177e4
LT
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>
c255d844
PM
46#include <linux/nfs_fs.h>
47#include <linux/acpi.h>
1da177e4
LT
48
49#include <asm/uaccess.h>
50#include <asm/processor.h>
51
529bf6be
DS
52extern int proc_nr_files(ctl_table *table, int write, struct file *filp,
53 void __user *buffer, size_t *lenp, loff_t *ppos);
54
1da177e4
LT
55#if defined(CONFIG_SYSCTL)
56
57/* External variables not in a header file. */
58extern int C_A_D;
59extern int sysctl_overcommit_memory;
60extern int sysctl_overcommit_ratio;
fadd8fbd 61extern int sysctl_panic_on_oom;
1da177e4
LT
62extern int max_threads;
63extern int sysrq_enabled;
64extern int core_uses_pid;
d6e71144 65extern int suid_dumpable;
1da177e4
LT
66extern char core_pattern[];
67extern int cad_pid;
68extern int pid_max;
69extern int min_free_kbytes;
70extern int printk_ratelimit_jiffies;
71extern int printk_ratelimit_burst;
72extern int pid_max_min, pid_max_max;
9d0243bc 73extern int sysctl_drop_caches;
8ad4b1fb 74extern int percpu_pagelist_fraction;
bebfa101 75extern int compat_log;
1da177e4
LT
76
77#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
78int unknown_nmi_panic;
1da177e4
LT
79#endif
80
81/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
82static int maxolduid = 65535;
83static int minolduid;
8ad4b1fb 84static int min_percpu_pagelist_fract = 8;
1da177e4
LT
85
86static int ngroups_max = NGROUPS_MAX;
87
88#ifdef CONFIG_KMOD
89extern char modprobe_path[];
90#endif
1da177e4
LT
91#ifdef CONFIG_CHR_DEV_SG
92extern int sg_big_buff;
93#endif
94#ifdef CONFIG_SYSVIPC
95extern size_t shm_ctlmax;
96extern size_t shm_ctlall;
97extern int shm_ctlmni;
98extern int msg_ctlmax;
99extern int msg_ctlmnb;
100extern int msg_ctlmni;
101extern int sem_ctls[];
102#endif
103
104#ifdef __sparc__
105extern char reboot_command [];
106extern int stop_a_enabled;
107extern int scons_pwroff;
108#endif
109
110#ifdef __hppa__
111extern int pwrsw_enabled;
112extern int unaligned_enabled;
113#endif
114
347a8dc3 115#ifdef CONFIG_S390
1da177e4
LT
116#ifdef CONFIG_MATHEMU
117extern int sysctl_ieee_emulation_warnings;
118#endif
119extern int sysctl_userprocess_debug;
951f22d5 120extern int spin_retry;
1da177e4
LT
121#endif
122
123extern int sysctl_hz_timer;
124
125#ifdef CONFIG_BSD_PROCESS_ACCT
126extern int acct_parm[];
127#endif
128
d2b176ed
JS
129#ifdef CONFIG_IA64
130extern int no_unaligned_warning;
131#endif
132
23f78d4a
IM
133#ifdef CONFIG_RT_MUTEXES
134extern int max_lock_depth;
135#endif
136
1da177e4
LT
137static int parse_table(int __user *, int, void __user *, size_t __user *, void __user *, size_t,
138 ctl_table *, void **);
139static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
140 void __user *buffer, size_t *lenp, loff_t *ppos);
141
142static ctl_table root_table[];
143static struct ctl_table_header root_table_header =
144 { root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
145
146static ctl_table kern_table[];
147static ctl_table vm_table[];
1da177e4
LT
148static ctl_table fs_table[];
149static ctl_table debug_table[];
150static ctl_table dev_table[];
151extern ctl_table random_table[];
152#ifdef CONFIG_UNIX98_PTYS
153extern ctl_table pty_table[];
154#endif
2d9048e2 155#ifdef CONFIG_INOTIFY_USER
0399cb08
RL
156extern ctl_table inotify_table[];
157#endif
1da177e4
LT
158
159#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
160int sysctl_legacy_va_layout;
161#endif
162
163/* /proc declarations: */
164
165#ifdef CONFIG_PROC_FS
166
167static ssize_t proc_readsys(struct file *, char __user *, size_t, loff_t *);
168static ssize_t proc_writesys(struct file *, const char __user *, size_t, loff_t *);
169static int proc_opensys(struct inode *, struct file *);
170
171struct file_operations proc_sys_file_operations = {
172 .open = proc_opensys,
173 .read = proc_readsys,
174 .write = proc_writesys,
175};
176
177extern struct proc_dir_entry *proc_sys_root;
178
330d57fb 179static void register_proc_table(ctl_table *, struct proc_dir_entry *, void *);
1da177e4
LT
180static void unregister_proc_table(ctl_table *, struct proc_dir_entry *);
181#endif
182
183/* The default sysctl tables: */
184
185static 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#ifdef CONFIG_NET
199 {
200 .ctl_name = CTL_NET,
201 .procname = "net",
202 .mode = 0555,
203 .child = net_table,
204 },
205#endif
1da177e4
LT
206 {
207 .ctl_name = CTL_FS,
208 .procname = "fs",
209 .mode = 0555,
210 .child = fs_table,
211 },
212 {
213 .ctl_name = CTL_DEBUG,
214 .procname = "debug",
215 .mode = 0555,
216 .child = debug_table,
217 },
218 {
219 .ctl_name = CTL_DEV,
220 .procname = "dev",
221 .mode = 0555,
222 .child = dev_table,
223 },
0eeca283 224
1da177e4
LT
225 { .ctl_name = 0 }
226};
227
228static ctl_table kern_table[] = {
229 {
230 .ctl_name = KERN_OSTYPE,
231 .procname = "ostype",
232 .data = system_utsname.sysname,
233 .maxlen = sizeof(system_utsname.sysname),
234 .mode = 0444,
235 .proc_handler = &proc_doutsstring,
236 .strategy = &sysctl_string,
237 },
238 {
239 .ctl_name = KERN_OSRELEASE,
240 .procname = "osrelease",
241 .data = system_utsname.release,
242 .maxlen = sizeof(system_utsname.release),
243 .mode = 0444,
244 .proc_handler = &proc_doutsstring,
245 .strategy = &sysctl_string,
246 },
247 {
248 .ctl_name = KERN_VERSION,
249 .procname = "version",
250 .data = system_utsname.version,
251 .maxlen = sizeof(system_utsname.version),
252 .mode = 0444,
253 .proc_handler = &proc_doutsstring,
254 .strategy = &sysctl_string,
255 },
256 {
257 .ctl_name = KERN_NODENAME,
258 .procname = "hostname",
259 .data = system_utsname.nodename,
260 .maxlen = sizeof(system_utsname.nodename),
261 .mode = 0644,
262 .proc_handler = &proc_doutsstring,
263 .strategy = &sysctl_string,
264 },
265 {
266 .ctl_name = KERN_DOMAINNAME,
267 .procname = "domainname",
268 .data = system_utsname.domainname,
269 .maxlen = sizeof(system_utsname.domainname),
270 .mode = 0644,
271 .proc_handler = &proc_doutsstring,
272 .strategy = &sysctl_string,
273 },
274 {
275 .ctl_name = KERN_PANIC,
276 .procname = "panic",
277 .data = &panic_timeout,
278 .maxlen = sizeof(int),
279 .mode = 0644,
280 .proc_handler = &proc_dointvec,
281 },
282 {
283 .ctl_name = KERN_CORE_USES_PID,
284 .procname = "core_uses_pid",
285 .data = &core_uses_pid,
286 .maxlen = sizeof(int),
287 .mode = 0644,
288 .proc_handler = &proc_dointvec,
289 },
290 {
291 .ctl_name = KERN_CORE_PATTERN,
292 .procname = "core_pattern",
293 .data = core_pattern,
294 .maxlen = 64,
295 .mode = 0644,
296 .proc_handler = &proc_dostring,
297 .strategy = &sysctl_string,
298 },
299 {
300 .ctl_name = KERN_TAINTED,
301 .procname = "tainted",
302 .data = &tainted,
303 .maxlen = sizeof(int),
304 .mode = 0444,
305 .proc_handler = &proc_dointvec,
306 },
307 {
308 .ctl_name = KERN_CAP_BSET,
309 .procname = "cap-bound",
310 .data = &cap_bset,
311 .maxlen = sizeof(kernel_cap_t),
312 .mode = 0600,
313 .proc_handler = &proc_dointvec_bset,
314 },
315#ifdef CONFIG_BLK_DEV_INITRD
316 {
317 .ctl_name = KERN_REALROOTDEV,
318 .procname = "real-root-dev",
319 .data = &real_root_dev,
320 .maxlen = sizeof(int),
321 .mode = 0644,
322 .proc_handler = &proc_dointvec,
323 },
324#endif
325#ifdef __sparc__
326 {
327 .ctl_name = KERN_SPARC_REBOOT,
328 .procname = "reboot-cmd",
329 .data = reboot_command,
330 .maxlen = 256,
331 .mode = 0644,
332 .proc_handler = &proc_dostring,
333 .strategy = &sysctl_string,
334 },
335 {
336 .ctl_name = KERN_SPARC_STOP_A,
337 .procname = "stop-a",
338 .data = &stop_a_enabled,
339 .maxlen = sizeof (int),
340 .mode = 0644,
341 .proc_handler = &proc_dointvec,
342 },
343 {
344 .ctl_name = KERN_SPARC_SCONS_PWROFF,
345 .procname = "scons-poweroff",
346 .data = &scons_pwroff,
347 .maxlen = sizeof (int),
348 .mode = 0644,
349 .proc_handler = &proc_dointvec,
350 },
351#endif
352#ifdef __hppa__
353 {
354 .ctl_name = KERN_HPPA_PWRSW,
355 .procname = "soft-power",
356 .data = &pwrsw_enabled,
357 .maxlen = sizeof (int),
358 .mode = 0644,
359 .proc_handler = &proc_dointvec,
360 },
361 {
362 .ctl_name = KERN_HPPA_UNALIGNED,
363 .procname = "unaligned-trap",
364 .data = &unaligned_enabled,
365 .maxlen = sizeof (int),
366 .mode = 0644,
367 .proc_handler = &proc_dointvec,
368 },
369#endif
370 {
371 .ctl_name = KERN_CTLALTDEL,
372 .procname = "ctrl-alt-del",
373 .data = &C_A_D,
374 .maxlen = sizeof(int),
375 .mode = 0644,
376 .proc_handler = &proc_dointvec,
377 },
378 {
379 .ctl_name = KERN_PRINTK,
380 .procname = "printk",
381 .data = &console_loglevel,
382 .maxlen = 4*sizeof(int),
383 .mode = 0644,
384 .proc_handler = &proc_dointvec,
385 },
386#ifdef CONFIG_KMOD
387 {
388 .ctl_name = KERN_MODPROBE,
389 .procname = "modprobe",
390 .data = &modprobe_path,
391 .maxlen = KMOD_PATH_LEN,
392 .mode = 0644,
393 .proc_handler = &proc_dostring,
394 .strategy = &sysctl_string,
395 },
396#endif
57ae2508 397#if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
1da177e4
LT
398 {
399 .ctl_name = KERN_HOTPLUG,
400 .procname = "hotplug",
312c004d
KS
401 .data = &uevent_helper,
402 .maxlen = UEVENT_HELPER_PATH_LEN,
1da177e4
LT
403 .mode = 0644,
404 .proc_handler = &proc_dostring,
405 .strategy = &sysctl_string,
406 },
407#endif
408#ifdef CONFIG_CHR_DEV_SG
409 {
410 .ctl_name = KERN_SG_BIG_BUFF,
411 .procname = "sg-big-buff",
412 .data = &sg_big_buff,
413 .maxlen = sizeof (int),
414 .mode = 0444,
415 .proc_handler = &proc_dointvec,
416 },
417#endif
418#ifdef CONFIG_BSD_PROCESS_ACCT
419 {
420 .ctl_name = KERN_ACCT,
421 .procname = "acct",
422 .data = &acct_parm,
423 .maxlen = 3*sizeof(int),
424 .mode = 0644,
425 .proc_handler = &proc_dointvec,
426 },
427#endif
428#ifdef CONFIG_SYSVIPC
429 {
430 .ctl_name = KERN_SHMMAX,
431 .procname = "shmmax",
432 .data = &shm_ctlmax,
433 .maxlen = sizeof (size_t),
434 .mode = 0644,
435 .proc_handler = &proc_doulongvec_minmax,
436 },
437 {
438 .ctl_name = KERN_SHMALL,
439 .procname = "shmall",
440 .data = &shm_ctlall,
441 .maxlen = sizeof (size_t),
442 .mode = 0644,
443 .proc_handler = &proc_doulongvec_minmax,
444 },
445 {
446 .ctl_name = KERN_SHMMNI,
447 .procname = "shmmni",
448 .data = &shm_ctlmni,
449 .maxlen = sizeof (int),
450 .mode = 0644,
451 .proc_handler = &proc_dointvec,
452 },
453 {
454 .ctl_name = KERN_MSGMAX,
455 .procname = "msgmax",
456 .data = &msg_ctlmax,
457 .maxlen = sizeof (int),
458 .mode = 0644,
459 .proc_handler = &proc_dointvec,
460 },
461 {
462 .ctl_name = KERN_MSGMNI,
463 .procname = "msgmni",
464 .data = &msg_ctlmni,
465 .maxlen = sizeof (int),
466 .mode = 0644,
467 .proc_handler = &proc_dointvec,
468 },
469 {
470 .ctl_name = KERN_MSGMNB,
471 .procname = "msgmnb",
472 .data = &msg_ctlmnb,
473 .maxlen = sizeof (int),
474 .mode = 0644,
475 .proc_handler = &proc_dointvec,
476 },
477 {
478 .ctl_name = KERN_SEM,
479 .procname = "sem",
480 .data = &sem_ctls,
481 .maxlen = 4*sizeof (int),
482 .mode = 0644,
483 .proc_handler = &proc_dointvec,
484 },
485#endif
486#ifdef CONFIG_MAGIC_SYSRQ
487 {
488 .ctl_name = KERN_SYSRQ,
489 .procname = "sysrq",
490 .data = &sysrq_enabled,
491 .maxlen = sizeof (int),
492 .mode = 0644,
493 .proc_handler = &proc_dointvec,
494 },
495#endif
496 {
497 .ctl_name = KERN_CADPID,
498 .procname = "cad_pid",
499 .data = &cad_pid,
500 .maxlen = sizeof (int),
501 .mode = 0600,
502 .proc_handler = &proc_dointvec,
503 },
504 {
505 .ctl_name = KERN_MAX_THREADS,
506 .procname = "threads-max",
507 .data = &max_threads,
508 .maxlen = sizeof(int),
509 .mode = 0644,
510 .proc_handler = &proc_dointvec,
511 },
512 {
513 .ctl_name = KERN_RANDOM,
514 .procname = "random",
515 .mode = 0555,
516 .child = random_table,
517 },
518#ifdef CONFIG_UNIX98_PTYS
519 {
520 .ctl_name = KERN_PTY,
521 .procname = "pty",
522 .mode = 0555,
523 .child = pty_table,
524 },
525#endif
526 {
527 .ctl_name = KERN_OVERFLOWUID,
528 .procname = "overflowuid",
529 .data = &overflowuid,
530 .maxlen = sizeof(int),
531 .mode = 0644,
532 .proc_handler = &proc_dointvec_minmax,
533 .strategy = &sysctl_intvec,
534 .extra1 = &minolduid,
535 .extra2 = &maxolduid,
536 },
537 {
538 .ctl_name = KERN_OVERFLOWGID,
539 .procname = "overflowgid",
540 .data = &overflowgid,
541 .maxlen = sizeof(int),
542 .mode = 0644,
543 .proc_handler = &proc_dointvec_minmax,
544 .strategy = &sysctl_intvec,
545 .extra1 = &minolduid,
546 .extra2 = &maxolduid,
547 },
347a8dc3 548#ifdef CONFIG_S390
1da177e4
LT
549#ifdef CONFIG_MATHEMU
550 {
551 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
552 .procname = "ieee_emulation_warnings",
553 .data = &sysctl_ieee_emulation_warnings,
554 .maxlen = sizeof(int),
555 .mode = 0644,
556 .proc_handler = &proc_dointvec,
557 },
558#endif
559#ifdef CONFIG_NO_IDLE_HZ
560 {
561 .ctl_name = KERN_HZ_TIMER,
562 .procname = "hz_timer",
563 .data = &sysctl_hz_timer,
564 .maxlen = sizeof(int),
565 .mode = 0644,
566 .proc_handler = &proc_dointvec,
567 },
568#endif
569 {
570 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
571 .procname = "userprocess_debug",
572 .data = &sysctl_userprocess_debug,
573 .maxlen = sizeof(int),
574 .mode = 0644,
575 .proc_handler = &proc_dointvec,
576 },
577#endif
578 {
579 .ctl_name = KERN_PIDMAX,
580 .procname = "pid_max",
581 .data = &pid_max,
582 .maxlen = sizeof (int),
583 .mode = 0644,
584 .proc_handler = &proc_dointvec_minmax,
585 .strategy = sysctl_intvec,
586 .extra1 = &pid_max_min,
587 .extra2 = &pid_max_max,
588 },
589 {
590 .ctl_name = KERN_PANIC_ON_OOPS,
591 .procname = "panic_on_oops",
592 .data = &panic_on_oops,
593 .maxlen = sizeof(int),
594 .mode = 0644,
595 .proc_handler = &proc_dointvec,
596 },
597 {
598 .ctl_name = KERN_PRINTK_RATELIMIT,
599 .procname = "printk_ratelimit",
600 .data = &printk_ratelimit_jiffies,
601 .maxlen = sizeof(int),
602 .mode = 0644,
603 .proc_handler = &proc_dointvec_jiffies,
604 .strategy = &sysctl_jiffies,
605 },
606 {
607 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
608 .procname = "printk_ratelimit_burst",
609 .data = &printk_ratelimit_burst,
610 .maxlen = sizeof(int),
611 .mode = 0644,
612 .proc_handler = &proc_dointvec,
613 },
614 {
615 .ctl_name = KERN_NGROUPS_MAX,
616 .procname = "ngroups_max",
617 .data = &ngroups_max,
618 .maxlen = sizeof (int),
619 .mode = 0444,
620 .proc_handler = &proc_dointvec,
621 },
622#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
623 {
624 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
625 .procname = "unknown_nmi_panic",
626 .data = &unknown_nmi_panic,
627 .maxlen = sizeof (int),
628 .mode = 0644,
2fbe7b25 629 .proc_handler = &proc_dointvec,
1da177e4
LT
630 },
631#endif
632#if defined(CONFIG_X86)
633 {
634 .ctl_name = KERN_BOOTLOADER_TYPE,
635 .procname = "bootloader_type",
636 .data = &bootloader_type,
637 .maxlen = sizeof (int),
638 .mode = 0444,
639 .proc_handler = &proc_dointvec,
640 },
641#endif
7a9166e3 642#if defined(CONFIG_MMU)
1da177e4
LT
643 {
644 .ctl_name = KERN_RANDOMIZE,
645 .procname = "randomize_va_space",
646 .data = &randomize_va_space,
647 .maxlen = sizeof(int),
648 .mode = 0644,
649 .proc_handler = &proc_dointvec,
650 },
7a9166e3 651#endif
0152fb37 652#if defined(CONFIG_S390) && defined(CONFIG_SMP)
951f22d5
MS
653 {
654 .ctl_name = KERN_SPIN_RETRY,
655 .procname = "spin_retry",
656 .data = &spin_retry,
657 .maxlen = sizeof (int),
658 .mode = 0644,
659 .proc_handler = &proc_dointvec,
660 },
c255d844
PM
661#endif
662#ifdef CONFIG_ACPI_SLEEP
663 {
664 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
665 .procname = "acpi_video_flags",
666 .data = &acpi_video_flags,
667 .maxlen = sizeof (unsigned long),
668 .mode = 0644,
7f99f06f 669 .proc_handler = &proc_doulongvec_minmax,
c255d844 670 },
d2b176ed
JS
671#endif
672#ifdef CONFIG_IA64
673 {
674 .ctl_name = KERN_IA64_UNALIGNED,
675 .procname = "ignore-unaligned-usertrap",
676 .data = &no_unaligned_warning,
677 .maxlen = sizeof (int),
678 .mode = 0644,
679 .proc_handler = &proc_dointvec,
680 },
bebfa101
AK
681#endif
682#ifdef CONFIG_COMPAT
683 {
684 .ctl_name = KERN_COMPAT_LOG,
685 .procname = "compat-log",
686 .data = &compat_log,
687 .maxlen = sizeof (int),
688 .mode = 0644,
689 .proc_handler = &proc_dointvec,
690 },
951f22d5 691#endif
23f78d4a
IM
692#ifdef CONFIG_RT_MUTEXES
693 {
694 .ctl_name = KERN_MAX_LOCK_DEPTH,
695 .procname = "max_lock_depth",
696 .data = &max_lock_depth,
697 .maxlen = sizeof(int),
698 .mode = 0644,
699 .proc_handler = &proc_dointvec,
700 },
701#endif
702
1da177e4
LT
703 { .ctl_name = 0 }
704};
705
706/* Constants for minimum and maximum testing in vm_table.
707 We use these as one-element integer vectors. */
708static int zero;
709static int one_hundred = 100;
710
711
712static ctl_table vm_table[] = {
713 {
714 .ctl_name = VM_OVERCOMMIT_MEMORY,
715 .procname = "overcommit_memory",
716 .data = &sysctl_overcommit_memory,
717 .maxlen = sizeof(sysctl_overcommit_memory),
718 .mode = 0644,
719 .proc_handler = &proc_dointvec,
720 },
fadd8fbd
KH
721 {
722 .ctl_name = VM_PANIC_ON_OOM,
723 .procname = "panic_on_oom",
724 .data = &sysctl_panic_on_oom,
725 .maxlen = sizeof(sysctl_panic_on_oom),
726 .mode = 0644,
727 .proc_handler = &proc_dointvec,
728 },
1da177e4
LT
729 {
730 .ctl_name = VM_OVERCOMMIT_RATIO,
731 .procname = "overcommit_ratio",
732 .data = &sysctl_overcommit_ratio,
733 .maxlen = sizeof(sysctl_overcommit_ratio),
734 .mode = 0644,
735 .proc_handler = &proc_dointvec,
736 },
737 {
738 .ctl_name = VM_PAGE_CLUSTER,
739 .procname = "page-cluster",
740 .data = &page_cluster,
741 .maxlen = sizeof(int),
742 .mode = 0644,
743 .proc_handler = &proc_dointvec,
744 },
745 {
746 .ctl_name = VM_DIRTY_BACKGROUND,
747 .procname = "dirty_background_ratio",
748 .data = &dirty_background_ratio,
749 .maxlen = sizeof(dirty_background_ratio),
750 .mode = 0644,
751 .proc_handler = &proc_dointvec_minmax,
752 .strategy = &sysctl_intvec,
753 .extra1 = &zero,
754 .extra2 = &one_hundred,
755 },
756 {
757 .ctl_name = VM_DIRTY_RATIO,
758 .procname = "dirty_ratio",
759 .data = &vm_dirty_ratio,
760 .maxlen = sizeof(vm_dirty_ratio),
761 .mode = 0644,
762 .proc_handler = &proc_dointvec_minmax,
763 .strategy = &sysctl_intvec,
764 .extra1 = &zero,
765 .extra2 = &one_hundred,
766 },
767 {
768 .ctl_name = VM_DIRTY_WB_CS,
769 .procname = "dirty_writeback_centisecs",
f6ef9438
BS
770 .data = &dirty_writeback_interval,
771 .maxlen = sizeof(dirty_writeback_interval),
1da177e4
LT
772 .mode = 0644,
773 .proc_handler = &dirty_writeback_centisecs_handler,
774 },
775 {
776 .ctl_name = VM_DIRTY_EXPIRE_CS,
777 .procname = "dirty_expire_centisecs",
f6ef9438
BS
778 .data = &dirty_expire_interval,
779 .maxlen = sizeof(dirty_expire_interval),
1da177e4 780 .mode = 0644,
f6ef9438 781 .proc_handler = &proc_dointvec_userhz_jiffies,
1da177e4
LT
782 },
783 {
784 .ctl_name = VM_NR_PDFLUSH_THREADS,
785 .procname = "nr_pdflush_threads",
786 .data = &nr_pdflush_threads,
787 .maxlen = sizeof nr_pdflush_threads,
788 .mode = 0444 /* read-only*/,
789 .proc_handler = &proc_dointvec,
790 },
791 {
792 .ctl_name = VM_SWAPPINESS,
793 .procname = "swappiness",
794 .data = &vm_swappiness,
795 .maxlen = sizeof(vm_swappiness),
796 .mode = 0644,
797 .proc_handler = &proc_dointvec_minmax,
798 .strategy = &sysctl_intvec,
799 .extra1 = &zero,
800 .extra2 = &one_hundred,
801 },
802#ifdef CONFIG_HUGETLB_PAGE
803 {
804 .ctl_name = VM_HUGETLB_PAGES,
805 .procname = "nr_hugepages",
806 .data = &max_huge_pages,
807 .maxlen = sizeof(unsigned long),
808 .mode = 0644,
809 .proc_handler = &hugetlb_sysctl_handler,
810 .extra1 = (void *)&hugetlb_zero,
811 .extra2 = (void *)&hugetlb_infinity,
812 },
813 {
814 .ctl_name = VM_HUGETLB_GROUP,
815 .procname = "hugetlb_shm_group",
816 .data = &sysctl_hugetlb_shm_group,
817 .maxlen = sizeof(gid_t),
818 .mode = 0644,
819 .proc_handler = &proc_dointvec,
820 },
821#endif
822 {
823 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
824 .procname = "lowmem_reserve_ratio",
825 .data = &sysctl_lowmem_reserve_ratio,
826 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
827 .mode = 0644,
828 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
829 .strategy = &sysctl_intvec,
830 },
9d0243bc
AM
831 {
832 .ctl_name = VM_DROP_PAGECACHE,
833 .procname = "drop_caches",
834 .data = &sysctl_drop_caches,
835 .maxlen = sizeof(int),
836 .mode = 0644,
837 .proc_handler = drop_caches_sysctl_handler,
838 .strategy = &sysctl_intvec,
839 },
1da177e4
LT
840 {
841 .ctl_name = VM_MIN_FREE_KBYTES,
842 .procname = "min_free_kbytes",
843 .data = &min_free_kbytes,
844 .maxlen = sizeof(min_free_kbytes),
845 .mode = 0644,
846 .proc_handler = &min_free_kbytes_sysctl_handler,
847 .strategy = &sysctl_intvec,
848 .extra1 = &zero,
849 },
8ad4b1fb
RS
850 {
851 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
852 .procname = "percpu_pagelist_fraction",
853 .data = &percpu_pagelist_fraction,
854 .maxlen = sizeof(percpu_pagelist_fraction),
855 .mode = 0644,
856 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
857 .strategy = &sysctl_intvec,
858 .extra1 = &min_percpu_pagelist_fract,
859 },
1da177e4
LT
860#ifdef CONFIG_MMU
861 {
862 .ctl_name = VM_MAX_MAP_COUNT,
863 .procname = "max_map_count",
864 .data = &sysctl_max_map_count,
865 .maxlen = sizeof(sysctl_max_map_count),
866 .mode = 0644,
867 .proc_handler = &proc_dointvec
868 },
869#endif
870 {
871 .ctl_name = VM_LAPTOP_MODE,
872 .procname = "laptop_mode",
873 .data = &laptop_mode,
874 .maxlen = sizeof(laptop_mode),
875 .mode = 0644,
ed5b43f1
BS
876 .proc_handler = &proc_dointvec_jiffies,
877 .strategy = &sysctl_jiffies,
1da177e4
LT
878 },
879 {
880 .ctl_name = VM_BLOCK_DUMP,
881 .procname = "block_dump",
882 .data = &block_dump,
883 .maxlen = sizeof(block_dump),
884 .mode = 0644,
885 .proc_handler = &proc_dointvec,
886 .strategy = &sysctl_intvec,
887 .extra1 = &zero,
888 },
889 {
890 .ctl_name = VM_VFS_CACHE_PRESSURE,
891 .procname = "vfs_cache_pressure",
892 .data = &sysctl_vfs_cache_pressure,
893 .maxlen = sizeof(sysctl_vfs_cache_pressure),
894 .mode = 0644,
895 .proc_handler = &proc_dointvec,
896 .strategy = &sysctl_intvec,
897 .extra1 = &zero,
898 },
899#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
900 {
901 .ctl_name = VM_LEGACY_VA_LAYOUT,
902 .procname = "legacy_va_layout",
903 .data = &sysctl_legacy_va_layout,
904 .maxlen = sizeof(sysctl_legacy_va_layout),
905 .mode = 0644,
906 .proc_handler = &proc_dointvec,
907 .strategy = &sysctl_intvec,
908 .extra1 = &zero,
909 },
910#endif
911#ifdef CONFIG_SWAP
912 {
913 .ctl_name = VM_SWAP_TOKEN_TIMEOUT,
914 .procname = "swap_token_timeout",
915 .data = &swap_token_default_timeout,
916 .maxlen = sizeof(swap_token_default_timeout),
917 .mode = 0644,
918 .proc_handler = &proc_dointvec_jiffies,
919 .strategy = &sysctl_jiffies,
920 },
1743660b
CL
921#endif
922#ifdef CONFIG_NUMA
923 {
924 .ctl_name = VM_ZONE_RECLAIM_MODE,
925 .procname = "zone_reclaim_mode",
926 .data = &zone_reclaim_mode,
927 .maxlen = sizeof(zone_reclaim_mode),
928 .mode = 0644,
929 .proc_handler = &proc_dointvec,
c84db23c
CL
930 .strategy = &sysctl_intvec,
931 .extra1 = &zero,
1743660b 932 },
9614634f
CL
933 {
934 .ctl_name = VM_MIN_UNMAPPED,
935 .procname = "min_unmapped_ratio",
936 .data = &sysctl_min_unmapped_ratio,
937 .maxlen = sizeof(sysctl_min_unmapped_ratio),
938 .mode = 0644,
939 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
940 .strategy = &sysctl_intvec,
941 .extra1 = &zero,
942 .extra2 = &one_hundred,
943 },
e6e5494c
IM
944#endif
945#ifdef CONFIG_X86_32
946 {
947 .ctl_name = VM_VDSO_ENABLED,
948 .procname = "vdso_enabled",
949 .data = &vdso_enabled,
950 .maxlen = sizeof(vdso_enabled),
951 .mode = 0644,
952 .proc_handler = &proc_dointvec,
953 .strategy = &sysctl_intvec,
954 .extra1 = &zero,
955 },
1da177e4
LT
956#endif
957 { .ctl_name = 0 }
958};
959
1da177e4
LT
960static ctl_table fs_table[] = {
961 {
962 .ctl_name = FS_NRINODE,
963 .procname = "inode-nr",
964 .data = &inodes_stat,
965 .maxlen = 2*sizeof(int),
966 .mode = 0444,
967 .proc_handler = &proc_dointvec,
968 },
969 {
970 .ctl_name = FS_STATINODE,
971 .procname = "inode-state",
972 .data = &inodes_stat,
973 .maxlen = 7*sizeof(int),
974 .mode = 0444,
975 .proc_handler = &proc_dointvec,
976 },
977 {
978 .ctl_name = FS_NRFILE,
979 .procname = "file-nr",
980 .data = &files_stat,
981 .maxlen = 3*sizeof(int),
982 .mode = 0444,
529bf6be 983 .proc_handler = &proc_nr_files,
1da177e4
LT
984 },
985 {
986 .ctl_name = FS_MAXFILE,
987 .procname = "file-max",
988 .data = &files_stat.max_files,
989 .maxlen = sizeof(int),
990 .mode = 0644,
991 .proc_handler = &proc_dointvec,
992 },
993 {
994 .ctl_name = FS_DENTRY,
995 .procname = "dentry-state",
996 .data = &dentry_stat,
997 .maxlen = 6*sizeof(int),
998 .mode = 0444,
999 .proc_handler = &proc_dointvec,
1000 },
1001 {
1002 .ctl_name = FS_OVERFLOWUID,
1003 .procname = "overflowuid",
1004 .data = &fs_overflowuid,
1005 .maxlen = sizeof(int),
1006 .mode = 0644,
1007 .proc_handler = &proc_dointvec_minmax,
1008 .strategy = &sysctl_intvec,
1009 .extra1 = &minolduid,
1010 .extra2 = &maxolduid,
1011 },
1012 {
1013 .ctl_name = FS_OVERFLOWGID,
1014 .procname = "overflowgid",
1015 .data = &fs_overflowgid,
1016 .maxlen = sizeof(int),
1017 .mode = 0644,
1018 .proc_handler = &proc_dointvec_minmax,
1019 .strategy = &sysctl_intvec,
1020 .extra1 = &minolduid,
1021 .extra2 = &maxolduid,
1022 },
1023 {
1024 .ctl_name = FS_LEASES,
1025 .procname = "leases-enable",
1026 .data = &leases_enable,
1027 .maxlen = sizeof(int),
1028 .mode = 0644,
1029 .proc_handler = &proc_dointvec,
1030 },
1031#ifdef CONFIG_DNOTIFY
1032 {
1033 .ctl_name = FS_DIR_NOTIFY,
1034 .procname = "dir-notify-enable",
1035 .data = &dir_notify_enable,
1036 .maxlen = sizeof(int),
1037 .mode = 0644,
1038 .proc_handler = &proc_dointvec,
1039 },
1040#endif
1041#ifdef CONFIG_MMU
1042 {
1043 .ctl_name = FS_LEASE_TIME,
1044 .procname = "lease-break-time",
1045 .data = &lease_break_time,
1046 .maxlen = sizeof(int),
1047 .mode = 0644,
1048 .proc_handler = &proc_dointvec,
1049 },
1050 {
1051 .ctl_name = FS_AIO_NR,
1052 .procname = "aio-nr",
1053 .data = &aio_nr,
1054 .maxlen = sizeof(aio_nr),
1055 .mode = 0444,
d55b5fda 1056 .proc_handler = &proc_doulongvec_minmax,
1da177e4
LT
1057 },
1058 {
1059 .ctl_name = FS_AIO_MAX_NR,
1060 .procname = "aio-max-nr",
1061 .data = &aio_max_nr,
1062 .maxlen = sizeof(aio_max_nr),
1063 .mode = 0644,
d55b5fda 1064 .proc_handler = &proc_doulongvec_minmax,
1da177e4 1065 },
2d9048e2 1066#ifdef CONFIG_INOTIFY_USER
0399cb08
RL
1067 {
1068 .ctl_name = FS_INOTIFY,
1069 .procname = "inotify",
1070 .mode = 0555,
1071 .child = inotify_table,
1072 },
1073#endif
1da177e4 1074#endif
d6e71144
AC
1075 {
1076 .ctl_name = KERN_SETUID_DUMPABLE,
1077 .procname = "suid_dumpable",
1078 .data = &suid_dumpable,
1079 .maxlen = sizeof(int),
1080 .mode = 0644,
1081 .proc_handler = &proc_dointvec,
1082 },
1da177e4
LT
1083 { .ctl_name = 0 }
1084};
1085
1086static ctl_table debug_table[] = {
1087 { .ctl_name = 0 }
1088};
1089
1090static ctl_table dev_table[] = {
1091 { .ctl_name = 0 }
0eeca283 1092};
1da177e4
LT
1093
1094extern void init_irq_proc (void);
1095
330d57fb
AV
1096static DEFINE_SPINLOCK(sysctl_lock);
1097
1098/* called under sysctl_lock */
1099static int use_table(struct ctl_table_header *p)
1100{
1101 if (unlikely(p->unregistering))
1102 return 0;
1103 p->used++;
1104 return 1;
1105}
1106
1107/* called under sysctl_lock */
1108static void unuse_table(struct ctl_table_header *p)
1109{
1110 if (!--p->used)
1111 if (unlikely(p->unregistering))
1112 complete(p->unregistering);
1113}
1114
1115/* called under sysctl_lock, will reacquire if has to wait */
1116static void start_unregistering(struct ctl_table_header *p)
1117{
1118 /*
1119 * if p->used is 0, nobody will ever touch that entry again;
1120 * we'll eliminate all paths to it before dropping sysctl_lock
1121 */
1122 if (unlikely(p->used)) {
1123 struct completion wait;
1124 init_completion(&wait);
1125 p->unregistering = &wait;
1126 spin_unlock(&sysctl_lock);
1127 wait_for_completion(&wait);
1128 spin_lock(&sysctl_lock);
1129 }
1130 /*
1131 * do not remove from the list until nobody holds it; walking the
1132 * list in do_sysctl() relies on that.
1133 */
1134 list_del_init(&p->ctl_entry);
1135}
1136
1da177e4
LT
1137void __init sysctl_init(void)
1138{
1139#ifdef CONFIG_PROC_FS
330d57fb 1140 register_proc_table(root_table, proc_sys_root, &root_table_header);
1da177e4
LT
1141 init_irq_proc();
1142#endif
1143}
1144
1145int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1146 void __user *newval, size_t newlen)
1147{
1148 struct list_head *tmp;
330d57fb 1149 int error = -ENOTDIR;
1da177e4
LT
1150
1151 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1152 return -ENOTDIR;
1153 if (oldval) {
1154 int old_len;
1155 if (!oldlenp || get_user(old_len, oldlenp))
1156 return -EFAULT;
1157 }
330d57fb 1158 spin_lock(&sysctl_lock);
1da177e4
LT
1159 tmp = &root_table_header.ctl_entry;
1160 do {
1161 struct ctl_table_header *head =
1162 list_entry(tmp, struct ctl_table_header, ctl_entry);
1163 void *context = NULL;
330d57fb
AV
1164
1165 if (!use_table(head))
1166 continue;
1167
1168 spin_unlock(&sysctl_lock);
1169
1170 error = parse_table(name, nlen, oldval, oldlenp,
1da177e4
LT
1171 newval, newlen, head->ctl_table,
1172 &context);
5a6b454f 1173 kfree(context);
330d57fb
AV
1174
1175 spin_lock(&sysctl_lock);
1176 unuse_table(head);
1da177e4 1177 if (error != -ENOTDIR)
330d57fb
AV
1178 break;
1179 } while ((tmp = tmp->next) != &root_table_header.ctl_entry);
1180 spin_unlock(&sysctl_lock);
1181 return error;
1da177e4
LT
1182}
1183
1184asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1185{
1186 struct __sysctl_args tmp;
1187 int error;
1188
1189 if (copy_from_user(&tmp, args, sizeof(tmp)))
1190 return -EFAULT;
1191
1192 lock_kernel();
1193 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1194 tmp.newval, tmp.newlen);
1195 unlock_kernel();
1196 return error;
1197}
1198
1199/*
1200 * ctl_perm does NOT grant the superuser all rights automatically, because
1201 * some sysctl variables are readonly even to root.
1202 */
1203
1204static int test_perm(int mode, int op)
1205{
1206 if (!current->euid)
1207 mode >>= 6;
1208 else if (in_egroup_p(0))
1209 mode >>= 3;
1210 if ((mode & op & 0007) == op)
1211 return 0;
1212 return -EACCES;
1213}
1214
1215static inline int ctl_perm(ctl_table *table, int op)
1216{
1217 int error;
1218 error = security_sysctl(table, op);
1219 if (error)
1220 return error;
1221 return test_perm(table->mode, op);
1222}
1223
1224static int parse_table(int __user *name, int nlen,
1225 void __user *oldval, size_t __user *oldlenp,
1226 void __user *newval, size_t newlen,
1227 ctl_table *table, void **context)
1228{
1229 int n;
1230repeat:
1231 if (!nlen)
1232 return -ENOTDIR;
1233 if (get_user(n, name))
1234 return -EFAULT;
1235 for ( ; table->ctl_name; table++) {
1236 if (n == table->ctl_name || table->ctl_name == CTL_ANY) {
1237 int error;
1238 if (table->child) {
1239 if (ctl_perm(table, 001))
1240 return -EPERM;
1241 if (table->strategy) {
1242 error = table->strategy(
1243 table, name, nlen,
1244 oldval, oldlenp,
1245 newval, newlen, context);
1246 if (error)
1247 return error;
1248 }
1249 name++;
1250 nlen--;
1251 table = table->child;
1252 goto repeat;
1253 }
1254 error = do_sysctl_strategy(table, name, nlen,
1255 oldval, oldlenp,
1256 newval, newlen, context);
1257 return error;
1258 }
1259 }
1260 return -ENOTDIR;
1261}
1262
1263/* Perform the actual read/write of a sysctl table entry. */
1264int do_sysctl_strategy (ctl_table *table,
1265 int __user *name, int nlen,
1266 void __user *oldval, size_t __user *oldlenp,
1267 void __user *newval, size_t newlen, void **context)
1268{
1269 int op = 0, rc;
1270 size_t len;
1271
1272 if (oldval)
1273 op |= 004;
1274 if (newval)
1275 op |= 002;
1276 if (ctl_perm(table, op))
1277 return -EPERM;
1278
1279 if (table->strategy) {
1280 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1281 newval, newlen, context);
1282 if (rc < 0)
1283 return rc;
1284 if (rc > 0)
1285 return 0;
1286 }
1287
1288 /* If there is no strategy routine, or if the strategy returns
1289 * zero, proceed with automatic r/w */
1290 if (table->data && table->maxlen) {
1291 if (oldval && oldlenp) {
1292 if (get_user(len, oldlenp))
1293 return -EFAULT;
1294 if (len) {
1295 if (len > table->maxlen)
1296 len = table->maxlen;
1297 if(copy_to_user(oldval, table->data, len))
1298 return -EFAULT;
1299 if(put_user(len, oldlenp))
1300 return -EFAULT;
1301 }
1302 }
1303 if (newval && newlen) {
1304 len = newlen;
1305 if (len > table->maxlen)
1306 len = table->maxlen;
1307 if(copy_from_user(table->data, newval, len))
1308 return -EFAULT;
1309 }
1310 }
1311 return 0;
1312}
1313
1314/**
1315 * register_sysctl_table - register a sysctl hierarchy
1316 * @table: the top-level table structure
1317 * @insert_at_head: whether the entry should be inserted in front or at the end
1318 *
1319 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1320 * array. An entry with a ctl_name of 0 terminates the table.
1321 *
1322 * The members of the &ctl_table structure are used as follows:
1323 *
1324 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1325 * must be unique within that level of sysctl
1326 *
1327 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1328 * enter a sysctl file
1329 *
1330 * data - a pointer to data for use by proc_handler
1331 *
1332 * maxlen - the maximum size in bytes of the data
1333 *
1334 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1335 *
1336 * child - a pointer to the child sysctl table if this entry is a directory, or
1337 * %NULL.
1338 *
1339 * proc_handler - the text handler routine (described below)
1340 *
1341 * strategy - the strategy routine (described below)
1342 *
1343 * de - for internal use by the sysctl routines
1344 *
1345 * extra1, extra2 - extra pointers usable by the proc handler routines
1346 *
1347 * Leaf nodes in the sysctl tree will be represented by a single file
1348 * under /proc; non-leaf nodes will be represented by directories.
1349 *
1350 * sysctl(2) can automatically manage read and write requests through
1351 * the sysctl table. The data and maxlen fields of the ctl_table
1352 * struct enable minimal validation of the values being written to be
1353 * performed, and the mode field allows minimal authentication.
1354 *
1355 * More sophisticated management can be enabled by the provision of a
1356 * strategy routine with the table entry. This will be called before
1357 * any automatic read or write of the data is performed.
1358 *
1359 * The strategy routine may return
1360 *
1361 * < 0 - Error occurred (error is passed to user process)
1362 *
1363 * 0 - OK - proceed with automatic read or write.
1364 *
1365 * > 0 - OK - read or write has been done by the strategy routine, so
1366 * return immediately.
1367 *
1368 * There must be a proc_handler routine for any terminal nodes
1369 * mirrored under /proc/sys (non-terminals are handled by a built-in
1370 * directory handler). Several default handlers are available to
1371 * cover common cases -
1372 *
1373 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1374 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1375 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1376 *
1377 * It is the handler's job to read the input buffer from user memory
1378 * and process it. The handler should return 0 on success.
1379 *
1380 * This routine returns %NULL on a failure to register, and a pointer
1381 * to the table header on success.
1382 */
1383struct ctl_table_header *register_sysctl_table(ctl_table * table,
1384 int insert_at_head)
1385{
1386 struct ctl_table_header *tmp;
1387 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1388 if (!tmp)
1389 return NULL;
1390 tmp->ctl_table = table;
1391 INIT_LIST_HEAD(&tmp->ctl_entry);
330d57fb
AV
1392 tmp->used = 0;
1393 tmp->unregistering = NULL;
1394 spin_lock(&sysctl_lock);
1da177e4
LT
1395 if (insert_at_head)
1396 list_add(&tmp->ctl_entry, &root_table_header.ctl_entry);
1397 else
1398 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
330d57fb 1399 spin_unlock(&sysctl_lock);
1da177e4 1400#ifdef CONFIG_PROC_FS
330d57fb 1401 register_proc_table(table, proc_sys_root, tmp);
1da177e4
LT
1402#endif
1403 return tmp;
1404}
1405
1406/**
1407 * unregister_sysctl_table - unregister a sysctl table hierarchy
1408 * @header: the header returned from register_sysctl_table
1409 *
1410 * Unregisters the sysctl table and all children. proc entries may not
1411 * actually be removed until they are no longer used by anyone.
1412 */
1413void unregister_sysctl_table(struct ctl_table_header * header)
1414{
330d57fb
AV
1415 might_sleep();
1416 spin_lock(&sysctl_lock);
1417 start_unregistering(header);
1da177e4
LT
1418#ifdef CONFIG_PROC_FS
1419 unregister_proc_table(header->ctl_table, proc_sys_root);
1420#endif
330d57fb 1421 spin_unlock(&sysctl_lock);
1da177e4
LT
1422 kfree(header);
1423}
1424
1425/*
1426 * /proc/sys support
1427 */
1428
1429#ifdef CONFIG_PROC_FS
1430
1431/* Scan the sysctl entries in table and add them all into /proc */
330d57fb 1432static void register_proc_table(ctl_table * table, struct proc_dir_entry *root, void *set)
1da177e4
LT
1433{
1434 struct proc_dir_entry *de;
1435 int len;
1436 mode_t mode;
1437
1438 for (; table->ctl_name; table++) {
1439 /* Can't do anything without a proc name. */
1440 if (!table->procname)
1441 continue;
1442 /* Maybe we can't do anything with it... */
1443 if (!table->proc_handler && !table->child) {
1444 printk(KERN_WARNING "SYSCTL: Can't register %s\n",
1445 table->procname);
1446 continue;
1447 }
1448
1449 len = strlen(table->procname);
1450 mode = table->mode;
1451
1452 de = NULL;
1453 if (table->proc_handler)
1454 mode |= S_IFREG;
1455 else {
1456 mode |= S_IFDIR;
1457 for (de = root->subdir; de; de = de->next) {
1458 if (proc_match(len, table->procname, de))
1459 break;
1460 }
1461 /* If the subdir exists already, de is non-NULL */
1462 }
1463
1464 if (!de) {
1465 de = create_proc_entry(table->procname, mode, root);
1466 if (!de)
1467 continue;
330d57fb 1468 de->set = set;
1da177e4
LT
1469 de->data = (void *) table;
1470 if (table->proc_handler)
1471 de->proc_fops = &proc_sys_file_operations;
1472 }
1473 table->de = de;
1474 if (de->mode & S_IFDIR)
330d57fb 1475 register_proc_table(table->child, de, set);
1da177e4
LT
1476 }
1477}
1478
1479/*
1480 * Unregister a /proc sysctl table and any subdirectories.
1481 */
1482static void unregister_proc_table(ctl_table * table, struct proc_dir_entry *root)
1483{
1484 struct proc_dir_entry *de;
1485 for (; table->ctl_name; table++) {
1486 if (!(de = table->de))
1487 continue;
1488 if (de->mode & S_IFDIR) {
1489 if (!table->child) {
1490 printk (KERN_ALERT "Help - malformed sysctl tree on free\n");
1491 continue;
1492 }
1493 unregister_proc_table(table->child, de);
1494
1495 /* Don't unregister directories which still have entries.. */
1496 if (de->subdir)
1497 continue;
1498 }
1499
330d57fb
AV
1500 /*
1501 * In any case, mark the entry as goner; we'll keep it
1502 * around if it's busy, but we'll know to do nothing with
1503 * its fields. We are under sysctl_lock here.
1504 */
1505 de->data = NULL;
1506
1da177e4
LT
1507 /* Don't unregister proc entries that are still being used.. */
1508 if (atomic_read(&de->count))
1509 continue;
1510
1511 table->de = NULL;
1512 remove_proc_entry(table->procname, root);
1513 }
1514}
1515
1516static ssize_t do_rw_proc(int write, struct file * file, char __user * buf,
1517 size_t count, loff_t *ppos)
1518{
1519 int op;
330d57fb 1520 struct proc_dir_entry *de = PDE(file->f_dentry->d_inode);
1da177e4
LT
1521 struct ctl_table *table;
1522 size_t res;
330d57fb 1523 ssize_t error = -ENOTDIR;
1da177e4 1524
330d57fb
AV
1525 spin_lock(&sysctl_lock);
1526 if (de && de->data && use_table(de->set)) {
1527 /*
1528 * at that point we know that sysctl was not unregistered
1529 * and won't be until we finish
1530 */
1531 spin_unlock(&sysctl_lock);
1532 table = (struct ctl_table *) de->data;
1533 if (!table || !table->proc_handler)
1534 goto out;
1535 error = -EPERM;
1536 op = (write ? 002 : 004);
1537 if (ctl_perm(table, op))
1538 goto out;
1539
1540 /* careful: calling conventions are nasty here */
1541 res = count;
1542 error = (*table->proc_handler)(table, write, file,
1543 buf, &res, ppos);
1544 if (!error)
1545 error = res;
1546 out:
1547 spin_lock(&sysctl_lock);
1548 unuse_table(de->set);
1549 }
1550 spin_unlock(&sysctl_lock);
1551 return error;
1da177e4
LT
1552}
1553
1554static int proc_opensys(struct inode *inode, struct file *file)
1555{
1556 if (file->f_mode & FMODE_WRITE) {
1557 /*
1558 * sysctl entries that are not writable,
1559 * are _NOT_ writable, capabilities or not.
1560 */
1561 if (!(inode->i_mode & S_IWUSR))
1562 return -EPERM;
1563 }
1564
1565 return 0;
1566}
1567
1568static ssize_t proc_readsys(struct file * file, char __user * buf,
1569 size_t count, loff_t *ppos)
1570{
1571 return do_rw_proc(0, file, buf, count, ppos);
1572}
1573
1574static ssize_t proc_writesys(struct file * file, const char __user * buf,
1575 size_t count, loff_t *ppos)
1576{
1577 return do_rw_proc(1, file, (char __user *) buf, count, ppos);
1578}
1579
1580/**
1581 * proc_dostring - read a string sysctl
1582 * @table: the sysctl table
1583 * @write: %TRUE if this is a write to the sysctl file
1584 * @filp: the file structure
1585 * @buffer: the user buffer
1586 * @lenp: the size of the user buffer
1587 * @ppos: file position
1588 *
1589 * Reads/writes a string from/to the user buffer. If the kernel
1590 * buffer provided is not large enough to hold the string, the
1591 * string is truncated. The copied string is %NULL-terminated.
1592 * If the string is being read by the user process, it is copied
1593 * and a newline '\n' is added. It is truncated if the buffer is
1594 * not large enough.
1595 *
1596 * Returns 0 on success.
1597 */
1598int proc_dostring(ctl_table *table, int write, struct file *filp,
1599 void __user *buffer, size_t *lenp, loff_t *ppos)
1600{
1601 size_t len;
1602 char __user *p;
1603 char c;
1604
1605 if (!table->data || !table->maxlen || !*lenp ||
1606 (*ppos && !write)) {
1607 *lenp = 0;
1608 return 0;
1609 }
1610
1611 if (write) {
1612 len = 0;
1613 p = buffer;
1614 while (len < *lenp) {
1615 if (get_user(c, p++))
1616 return -EFAULT;
1617 if (c == 0 || c == '\n')
1618 break;
1619 len++;
1620 }
1621 if (len >= table->maxlen)
1622 len = table->maxlen-1;
1623 if(copy_from_user(table->data, buffer, len))
1624 return -EFAULT;
1625 ((char *) table->data)[len] = 0;
1626 *ppos += *lenp;
1627 } else {
1628 len = strlen(table->data);
1629 if (len > table->maxlen)
1630 len = table->maxlen;
1631 if (len > *lenp)
1632 len = *lenp;
1633 if (len)
1634 if(copy_to_user(buffer, table->data, len))
1635 return -EFAULT;
1636 if (len < *lenp) {
1637 if(put_user('\n', ((char __user *) buffer) + len))
1638 return -EFAULT;
1639 len++;
1640 }
1641 *lenp = len;
1642 *ppos += len;
1643 }
1644 return 0;
1645}
1646
1647/*
1648 * Special case of dostring for the UTS structure. This has locks
1649 * to observe. Should this be in kernel/sys.c ????
1650 */
1651
1652static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
1653 void __user *buffer, size_t *lenp, loff_t *ppos)
1654{
1655 int r;
1656
1657 if (!write) {
1658 down_read(&uts_sem);
1659 r=proc_dostring(table,0,filp,buffer,lenp, ppos);
1660 up_read(&uts_sem);
1661 } else {
1662 down_write(&uts_sem);
1663 r=proc_dostring(table,1,filp,buffer,lenp, ppos);
1664 up_write(&uts_sem);
1665 }
1666 return r;
1667}
1668
1669static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1670 int *valp,
1671 int write, void *data)
1672{
1673 if (write) {
1674 *valp = *negp ? -*lvalp : *lvalp;
1675 } else {
1676 int val = *valp;
1677 if (val < 0) {
1678 *negp = -1;
1679 *lvalp = (unsigned long)-val;
1680 } else {
1681 *negp = 0;
1682 *lvalp = (unsigned long)val;
1683 }
1684 }
1685 return 0;
1686}
1687
1688static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1689 void __user *buffer, size_t *lenp, loff_t *ppos,
1690 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1691 int write, void *data),
1692 void *data)
1693{
1694#define TMPBUFLEN 21
1695 int *i, vleft, first=1, neg, val;
1696 unsigned long lval;
1697 size_t left, len;
1698
1699 char buf[TMPBUFLEN], *p;
1700 char __user *s = buffer;
1701
1702 if (!table->data || !table->maxlen || !*lenp ||
1703 (*ppos && !write)) {
1704 *lenp = 0;
1705 return 0;
1706 }
1707
1708 i = (int *) table->data;
1709 vleft = table->maxlen / sizeof(*i);
1710 left = *lenp;
1711
1712 if (!conv)
1713 conv = do_proc_dointvec_conv;
1714
1715 for (; left && vleft--; i++, first=0) {
1716 if (write) {
1717 while (left) {
1718 char c;
1719 if (get_user(c, s))
1720 return -EFAULT;
1721 if (!isspace(c))
1722 break;
1723 left--;
1724 s++;
1725 }
1726 if (!left)
1727 break;
1728 neg = 0;
1729 len = left;
1730 if (len > sizeof(buf) - 1)
1731 len = sizeof(buf) - 1;
1732 if (copy_from_user(buf, s, len))
1733 return -EFAULT;
1734 buf[len] = 0;
1735 p = buf;
1736 if (*p == '-' && left > 1) {
1737 neg = 1;
1738 left--, p++;
1739 }
1740 if (*p < '0' || *p > '9')
1741 break;
1742
1743 lval = simple_strtoul(p, &p, 0);
1744
1745 len = p-buf;
1746 if ((len < left) && *p && !isspace(*p))
1747 break;
1748 if (neg)
1749 val = -val;
1750 s += len;
1751 left -= len;
1752
1753 if (conv(&neg, &lval, i, 1, data))
1754 break;
1755 } else {
1756 p = buf;
1757 if (!first)
1758 *p++ = '\t';
1759
1760 if (conv(&neg, &lval, i, 0, data))
1761 break;
1762
1763 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1764 len = strlen(buf);
1765 if (len > left)
1766 len = left;
1767 if(copy_to_user(s, buf, len))
1768 return -EFAULT;
1769 left -= len;
1770 s += len;
1771 }
1772 }
1773
1774 if (!write && !first && left) {
1775 if(put_user('\n', s))
1776 return -EFAULT;
1777 left--, s++;
1778 }
1779 if (write) {
1780 while (left) {
1781 char c;
1782 if (get_user(c, s++))
1783 return -EFAULT;
1784 if (!isspace(c))
1785 break;
1786 left--;
1787 }
1788 }
1789 if (write && first)
1790 return -EINVAL;
1791 *lenp -= left;
1792 *ppos += *lenp;
1793 return 0;
1794#undef TMPBUFLEN
1795}
1796
1797/**
1798 * proc_dointvec - read a vector of integers
1799 * @table: the sysctl table
1800 * @write: %TRUE if this is a write to the sysctl file
1801 * @filp: the file structure
1802 * @buffer: the user buffer
1803 * @lenp: the size of the user buffer
1804 * @ppos: file position
1805 *
1806 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1807 * values from/to the user buffer, treated as an ASCII string.
1808 *
1809 * Returns 0 on success.
1810 */
1811int proc_dointvec(ctl_table *table, int write, struct file *filp,
1812 void __user *buffer, size_t *lenp, loff_t *ppos)
1813{
1814 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1815 NULL,NULL);
1816}
1817
1818#define OP_SET 0
1819#define OP_AND 1
1820#define OP_OR 2
1821#define OP_MAX 3
1822#define OP_MIN 4
1823
1824static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1825 int *valp,
1826 int write, void *data)
1827{
1828 int op = *(int *)data;
1829 if (write) {
1830 int val = *negp ? -*lvalp : *lvalp;
1831 switch(op) {
1832 case OP_SET: *valp = val; break;
1833 case OP_AND: *valp &= val; break;
1834 case OP_OR: *valp |= val; break;
1835 case OP_MAX: if(*valp < val)
1836 *valp = val;
1837 break;
1838 case OP_MIN: if(*valp > val)
1839 *valp = val;
1840 break;
1841 }
1842 } else {
1843 int val = *valp;
1844 if (val < 0) {
1845 *negp = -1;
1846 *lvalp = (unsigned long)-val;
1847 } else {
1848 *negp = 0;
1849 *lvalp = (unsigned long)val;
1850 }
1851 }
1852 return 0;
1853}
1854
1855/*
1856 * init may raise the set.
1857 */
1858
1859int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1860 void __user *buffer, size_t *lenp, loff_t *ppos)
1861{
1862 int op;
1863
1864 if (!capable(CAP_SYS_MODULE)) {
1865 return -EPERM;
1866 }
1867
1868 op = (current->pid == 1) ? OP_SET : OP_AND;
1869 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1870 do_proc_dointvec_bset_conv,&op);
1871}
1872
1873struct do_proc_dointvec_minmax_conv_param {
1874 int *min;
1875 int *max;
1876};
1877
1878static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1879 int *valp,
1880 int write, void *data)
1881{
1882 struct do_proc_dointvec_minmax_conv_param *param = data;
1883 if (write) {
1884 int val = *negp ? -*lvalp : *lvalp;
1885 if ((param->min && *param->min > val) ||
1886 (param->max && *param->max < val))
1887 return -EINVAL;
1888 *valp = val;
1889 } else {
1890 int val = *valp;
1891 if (val < 0) {
1892 *negp = -1;
1893 *lvalp = (unsigned long)-val;
1894 } else {
1895 *negp = 0;
1896 *lvalp = (unsigned long)val;
1897 }
1898 }
1899 return 0;
1900}
1901
1902/**
1903 * proc_dointvec_minmax - read a vector of integers with min/max values
1904 * @table: the sysctl table
1905 * @write: %TRUE if this is a write to the sysctl file
1906 * @filp: the file structure
1907 * @buffer: the user buffer
1908 * @lenp: the size of the user buffer
1909 * @ppos: file position
1910 *
1911 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1912 * values from/to the user buffer, treated as an ASCII string.
1913 *
1914 * This routine will ensure the values are within the range specified by
1915 * table->extra1 (min) and table->extra2 (max).
1916 *
1917 * Returns 0 on success.
1918 */
1919int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1920 void __user *buffer, size_t *lenp, loff_t *ppos)
1921{
1922 struct do_proc_dointvec_minmax_conv_param param = {
1923 .min = (int *) table->extra1,
1924 .max = (int *) table->extra2,
1925 };
1926 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1927 do_proc_dointvec_minmax_conv, &param);
1928}
1929
1930static int do_proc_doulongvec_minmax(ctl_table *table, int write,
1931 struct file *filp,
1932 void __user *buffer,
1933 size_t *lenp, loff_t *ppos,
1934 unsigned long convmul,
1935 unsigned long convdiv)
1936{
1937#define TMPBUFLEN 21
1938 unsigned long *i, *min, *max, val;
1939 int vleft, first=1, neg;
1940 size_t len, left;
1941 char buf[TMPBUFLEN], *p;
1942 char __user *s = buffer;
1943
1944 if (!table->data || !table->maxlen || !*lenp ||
1945 (*ppos && !write)) {
1946 *lenp = 0;
1947 return 0;
1948 }
1949
1950 i = (unsigned long *) table->data;
1951 min = (unsigned long *) table->extra1;
1952 max = (unsigned long *) table->extra2;
1953 vleft = table->maxlen / sizeof(unsigned long);
1954 left = *lenp;
1955
1956 for (; left && vleft--; i++, min++, max++, first=0) {
1957 if (write) {
1958 while (left) {
1959 char c;
1960 if (get_user(c, s))
1961 return -EFAULT;
1962 if (!isspace(c))
1963 break;
1964 left--;
1965 s++;
1966 }
1967 if (!left)
1968 break;
1969 neg = 0;
1970 len = left;
1971 if (len > TMPBUFLEN-1)
1972 len = TMPBUFLEN-1;
1973 if (copy_from_user(buf, s, len))
1974 return -EFAULT;
1975 buf[len] = 0;
1976 p = buf;
1977 if (*p == '-' && left > 1) {
1978 neg = 1;
1979 left--, p++;
1980 }
1981 if (*p < '0' || *p > '9')
1982 break;
1983 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
1984 len = p-buf;
1985 if ((len < left) && *p && !isspace(*p))
1986 break;
1987 if (neg)
1988 val = -val;
1989 s += len;
1990 left -= len;
1991
1992 if(neg)
1993 continue;
1994 if ((min && val < *min) || (max && val > *max))
1995 continue;
1996 *i = val;
1997 } else {
1998 p = buf;
1999 if (!first)
2000 *p++ = '\t';
2001 sprintf(p, "%lu", convdiv * (*i) / convmul);
2002 len = strlen(buf);
2003 if (len > left)
2004 len = left;
2005 if(copy_to_user(s, buf, len))
2006 return -EFAULT;
2007 left -= len;
2008 s += len;
2009 }
2010 }
2011
2012 if (!write && !first && left) {
2013 if(put_user('\n', s))
2014 return -EFAULT;
2015 left--, s++;
2016 }
2017 if (write) {
2018 while (left) {
2019 char c;
2020 if (get_user(c, s++))
2021 return -EFAULT;
2022 if (!isspace(c))
2023 break;
2024 left--;
2025 }
2026 }
2027 if (write && first)
2028 return -EINVAL;
2029 *lenp -= left;
2030 *ppos += *lenp;
2031 return 0;
2032#undef TMPBUFLEN
2033}
2034
2035/**
2036 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2037 * @table: the sysctl table
2038 * @write: %TRUE if this is a write to the sysctl file
2039 * @filp: the file structure
2040 * @buffer: the user buffer
2041 * @lenp: the size of the user buffer
2042 * @ppos: file position
2043 *
2044 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2045 * values from/to the user buffer, treated as an ASCII string.
2046 *
2047 * This routine will ensure the values are within the range specified by
2048 * table->extra1 (min) and table->extra2 (max).
2049 *
2050 * Returns 0 on success.
2051 */
2052int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2053 void __user *buffer, size_t *lenp, loff_t *ppos)
2054{
2055 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2056}
2057
2058/**
2059 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2060 * @table: the sysctl table
2061 * @write: %TRUE if this is a write to the sysctl file
2062 * @filp: the file structure
2063 * @buffer: the user buffer
2064 * @lenp: the size of the user buffer
2065 * @ppos: file position
2066 *
2067 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2068 * values from/to the user buffer, treated as an ASCII string. The values
2069 * are treated as milliseconds, and converted to jiffies when they are stored.
2070 *
2071 * This routine will ensure the values are within the range specified by
2072 * table->extra1 (min) and table->extra2 (max).
2073 *
2074 * Returns 0 on success.
2075 */
2076int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2077 struct file *filp,
2078 void __user *buffer,
2079 size_t *lenp, loff_t *ppos)
2080{
2081 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2082 lenp, ppos, HZ, 1000l);
2083}
2084
2085
2086static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2087 int *valp,
2088 int write, void *data)
2089{
2090 if (write) {
cba9f33d
BS
2091 if (*lvalp > LONG_MAX / HZ)
2092 return 1;
1da177e4
LT
2093 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2094 } else {
2095 int val = *valp;
2096 unsigned long lval;
2097 if (val < 0) {
2098 *negp = -1;
2099 lval = (unsigned long)-val;
2100 } else {
2101 *negp = 0;
2102 lval = (unsigned long)val;
2103 }
2104 *lvalp = lval / HZ;
2105 }
2106 return 0;
2107}
2108
2109static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2110 int *valp,
2111 int write, void *data)
2112{
2113 if (write) {
cba9f33d
BS
2114 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2115 return 1;
1da177e4
LT
2116 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2117 } else {
2118 int val = *valp;
2119 unsigned long lval;
2120 if (val < 0) {
2121 *negp = -1;
2122 lval = (unsigned long)-val;
2123 } else {
2124 *negp = 0;
2125 lval = (unsigned long)val;
2126 }
2127 *lvalp = jiffies_to_clock_t(lval);
2128 }
2129 return 0;
2130}
2131
2132static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2133 int *valp,
2134 int write, void *data)
2135{
2136 if (write) {
2137 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2138 } else {
2139 int val = *valp;
2140 unsigned long lval;
2141 if (val < 0) {
2142 *negp = -1;
2143 lval = (unsigned long)-val;
2144 } else {
2145 *negp = 0;
2146 lval = (unsigned long)val;
2147 }
2148 *lvalp = jiffies_to_msecs(lval);
2149 }
2150 return 0;
2151}
2152
2153/**
2154 * proc_dointvec_jiffies - read a vector of integers as seconds
2155 * @table: the sysctl table
2156 * @write: %TRUE if this is a write to the sysctl file
2157 * @filp: the file structure
2158 * @buffer: the user buffer
2159 * @lenp: the size of the user buffer
2160 * @ppos: file position
2161 *
2162 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2163 * values from/to the user buffer, treated as an ASCII string.
2164 * The values read are assumed to be in seconds, and are converted into
2165 * jiffies.
2166 *
2167 * Returns 0 on success.
2168 */
2169int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2170 void __user *buffer, size_t *lenp, loff_t *ppos)
2171{
2172 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2173 do_proc_dointvec_jiffies_conv,NULL);
2174}
2175
2176/**
2177 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2178 * @table: the sysctl table
2179 * @write: %TRUE if this is a write to the sysctl file
2180 * @filp: the file structure
2181 * @buffer: the user buffer
2182 * @lenp: the size of the user buffer
1e5d5331 2183 * @ppos: pointer to the file position
1da177e4
LT
2184 *
2185 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2186 * values from/to the user buffer, treated as an ASCII string.
2187 * The values read are assumed to be in 1/USER_HZ seconds, and
2188 * are converted into jiffies.
2189 *
2190 * Returns 0 on success.
2191 */
2192int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2193 void __user *buffer, size_t *lenp, loff_t *ppos)
2194{
2195 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2196 do_proc_dointvec_userhz_jiffies_conv,NULL);
2197}
2198
2199/**
2200 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2201 * @table: the sysctl table
2202 * @write: %TRUE if this is a write to the sysctl file
2203 * @filp: the file structure
2204 * @buffer: the user buffer
2205 * @lenp: the size of the user buffer
67be2dd1
MW
2206 * @ppos: file position
2207 * @ppos: the current position in the file
1da177e4
LT
2208 *
2209 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2210 * values from/to the user buffer, treated as an ASCII string.
2211 * The values read are assumed to be in 1/1000 seconds, and
2212 * are converted into jiffies.
2213 *
2214 * Returns 0 on success.
2215 */
2216int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2217 void __user *buffer, size_t *lenp, loff_t *ppos)
2218{
2219 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2220 do_proc_dointvec_ms_jiffies_conv, NULL);
2221}
2222
2223#else /* CONFIG_PROC_FS */
2224
2225int proc_dostring(ctl_table *table, int write, struct file *filp,
2226 void __user *buffer, size_t *lenp, loff_t *ppos)
2227{
2228 return -ENOSYS;
2229}
2230
2231static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
2232 void __user *buffer, size_t *lenp, loff_t *ppos)
2233{
2234 return -ENOSYS;
2235}
2236
2237int proc_dointvec(ctl_table *table, int write, struct file *filp,
2238 void __user *buffer, size_t *lenp, loff_t *ppos)
2239{
2240 return -ENOSYS;
2241}
2242
2243int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2244 void __user *buffer, size_t *lenp, loff_t *ppos)
2245{
2246 return -ENOSYS;
2247}
2248
2249int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2250 void __user *buffer, size_t *lenp, loff_t *ppos)
2251{
2252 return -ENOSYS;
2253}
2254
2255int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2256 void __user *buffer, size_t *lenp, loff_t *ppos)
2257{
2258 return -ENOSYS;
2259}
2260
2261int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2262 void __user *buffer, size_t *lenp, loff_t *ppos)
2263{
2264 return -ENOSYS;
2265}
2266
2267int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2268 void __user *buffer, size_t *lenp, loff_t *ppos)
2269{
2270 return -ENOSYS;
2271}
2272
2273int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2274 void __user *buffer, size_t *lenp, loff_t *ppos)
2275{
2276 return -ENOSYS;
2277}
2278
2279int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2280 struct file *filp,
2281 void __user *buffer,
2282 size_t *lenp, loff_t *ppos)
2283{
2284 return -ENOSYS;
2285}
2286
2287
2288#endif /* CONFIG_PROC_FS */
2289
2290
2291/*
2292 * General sysctl support routines
2293 */
2294
2295/* The generic string strategy routine: */
2296int sysctl_string(ctl_table *table, int __user *name, int nlen,
2297 void __user *oldval, size_t __user *oldlenp,
2298 void __user *newval, size_t newlen, void **context)
2299{
1da177e4
LT
2300 if (!table->data || !table->maxlen)
2301 return -ENOTDIR;
2302
2303 if (oldval && oldlenp) {
de9e007d
LT
2304 size_t bufsize;
2305 if (get_user(bufsize, oldlenp))
1da177e4 2306 return -EFAULT;
de9e007d
LT
2307 if (bufsize) {
2308 size_t len = strlen(table->data), copied;
2309
2310 /* This shouldn't trigger for a well-formed sysctl */
2311 if (len > table->maxlen)
1da177e4 2312 len = table->maxlen;
de9e007d
LT
2313
2314 /* Copy up to a max of bufsize-1 bytes of the string */
2315 copied = (len >= bufsize) ? bufsize - 1 : len;
2316
2317 if (copy_to_user(oldval, table->data, copied) ||
2318 put_user(0, (char __user *)(oldval + copied)))
1da177e4 2319 return -EFAULT;
de9e007d 2320 if (put_user(len, oldlenp))
1da177e4
LT
2321 return -EFAULT;
2322 }
2323 }
2324 if (newval && newlen) {
de9e007d 2325 size_t len = newlen;
1da177e4
LT
2326 if (len > table->maxlen)
2327 len = table->maxlen;
2328 if(copy_from_user(table->data, newval, len))
2329 return -EFAULT;
2330 if (len == table->maxlen)
2331 len--;
2332 ((char *) table->data)[len] = 0;
2333 }
82c9df82 2334 return 1;
1da177e4
LT
2335}
2336
2337/*
2338 * This function makes sure that all of the integers in the vector
2339 * are between the minimum and maximum values given in the arrays
2340 * table->extra1 and table->extra2, respectively.
2341 */
2342int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2343 void __user *oldval, size_t __user *oldlenp,
2344 void __user *newval, size_t newlen, void **context)
2345{
2346
2347 if (newval && newlen) {
2348 int __user *vec = (int __user *) newval;
2349 int *min = (int *) table->extra1;
2350 int *max = (int *) table->extra2;
2351 size_t length;
2352 int i;
2353
2354 if (newlen % sizeof(int) != 0)
2355 return -EINVAL;
2356
2357 if (!table->extra1 && !table->extra2)
2358 return 0;
2359
2360 if (newlen > table->maxlen)
2361 newlen = table->maxlen;
2362 length = newlen / sizeof(int);
2363
2364 for (i = 0; i < length; i++) {
2365 int value;
2366 if (get_user(value, vec + i))
2367 return -EFAULT;
2368 if (min && value < min[i])
2369 return -EINVAL;
2370 if (max && value > max[i])
2371 return -EINVAL;
2372 }
2373 }
2374 return 0;
2375}
2376
2377/* Strategy function to convert jiffies to seconds */
2378int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2379 void __user *oldval, size_t __user *oldlenp,
2380 void __user *newval, size_t newlen, void **context)
2381{
2382 if (oldval) {
2383 size_t olen;
2384 if (oldlenp) {
2385 if (get_user(olen, oldlenp))
2386 return -EFAULT;
2387 if (olen!=sizeof(int))
2388 return -EINVAL;
2389 }
2390 if (put_user(*(int *)(table->data)/HZ, (int __user *)oldval) ||
2391 (oldlenp && put_user(sizeof(int),oldlenp)))
2392 return -EFAULT;
2393 }
2394 if (newval && newlen) {
2395 int new;
2396 if (newlen != sizeof(int))
2397 return -EINVAL;
2398 if (get_user(new, (int __user *)newval))
2399 return -EFAULT;
2400 *(int *)(table->data) = new*HZ;
2401 }
2402 return 1;
2403}
2404
2405/* Strategy function to convert jiffies to seconds */
2406int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2407 void __user *oldval, size_t __user *oldlenp,
2408 void __user *newval, size_t newlen, void **context)
2409{
2410 if (oldval) {
2411 size_t olen;
2412 if (oldlenp) {
2413 if (get_user(olen, oldlenp))
2414 return -EFAULT;
2415 if (olen!=sizeof(int))
2416 return -EINVAL;
2417 }
2418 if (put_user(jiffies_to_msecs(*(int *)(table->data)), (int __user *)oldval) ||
2419 (oldlenp && put_user(sizeof(int),oldlenp)))
2420 return -EFAULT;
2421 }
2422 if (newval && newlen) {
2423 int new;
2424 if (newlen != sizeof(int))
2425 return -EINVAL;
2426 if (get_user(new, (int __user *)newval))
2427 return -EFAULT;
2428 *(int *)(table->data) = msecs_to_jiffies(new);
2429 }
2430 return 1;
2431}
2432
2433#else /* CONFIG_SYSCTL */
2434
2435
2436asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2437{
2438 return -ENOSYS;
2439}
2440
2441int sysctl_string(ctl_table *table, int __user *name, int nlen,
2442 void __user *oldval, size_t __user *oldlenp,
2443 void __user *newval, size_t newlen, void **context)
2444{
2445 return -ENOSYS;
2446}
2447
2448int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2449 void __user *oldval, size_t __user *oldlenp,
2450 void __user *newval, size_t newlen, void **context)
2451{
2452 return -ENOSYS;
2453}
2454
2455int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2456 void __user *oldval, size_t __user *oldlenp,
2457 void __user *newval, size_t newlen, void **context)
2458{
2459 return -ENOSYS;
2460}
2461
2462int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2463 void __user *oldval, size_t __user *oldlenp,
2464 void __user *newval, size_t newlen, void **context)
2465{
2466 return -ENOSYS;
2467}
2468
2469int proc_dostring(ctl_table *table, int write, struct file *filp,
2470 void __user *buffer, size_t *lenp, loff_t *ppos)
2471{
2472 return -ENOSYS;
2473}
2474
2475int proc_dointvec(ctl_table *table, int write, struct file *filp,
2476 void __user *buffer, size_t *lenp, loff_t *ppos)
2477{
2478 return -ENOSYS;
2479}
2480
2481int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2482 void __user *buffer, size_t *lenp, loff_t *ppos)
2483{
2484 return -ENOSYS;
2485}
2486
2487int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2488 void __user *buffer, size_t *lenp, loff_t *ppos)
2489{
2490 return -ENOSYS;
2491}
2492
2493int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2494 void __user *buffer, size_t *lenp, loff_t *ppos)
2495{
2496 return -ENOSYS;
2497}
2498
2499int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2500 void __user *buffer, size_t *lenp, loff_t *ppos)
2501{
2502 return -ENOSYS;
2503}
2504
2505int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2506 void __user *buffer, size_t *lenp, loff_t *ppos)
2507{
2508 return -ENOSYS;
2509}
2510
2511int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2512 void __user *buffer, size_t *lenp, loff_t *ppos)
2513{
2514 return -ENOSYS;
2515}
2516
2517int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2518 struct file *filp,
2519 void __user *buffer,
2520 size_t *lenp, loff_t *ppos)
2521{
2522 return -ENOSYS;
2523}
2524
2525struct ctl_table_header * register_sysctl_table(ctl_table * table,
2526 int insert_at_head)
2527{
2528 return NULL;
2529}
2530
2531void unregister_sysctl_table(struct ctl_table_header * table)
2532{
2533}
2534
2535#endif /* CONFIG_SYSCTL */
2536
2537/*
2538 * No sense putting this after each symbol definition, twice,
2539 * exception granted :-)
2540 */
2541EXPORT_SYMBOL(proc_dointvec);
2542EXPORT_SYMBOL(proc_dointvec_jiffies);
2543EXPORT_SYMBOL(proc_dointvec_minmax);
2544EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2545EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2546EXPORT_SYMBOL(proc_dostring);
2547EXPORT_SYMBOL(proc_doulongvec_minmax);
2548EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2549EXPORT_SYMBOL(register_sysctl_table);
2550EXPORT_SYMBOL(sysctl_intvec);
2551EXPORT_SYMBOL(sysctl_jiffies);
2552EXPORT_SYMBOL(sysctl_ms_jiffies);
2553EXPORT_SYMBOL(sysctl_string);
2554EXPORT_SYMBOL(unregister_sysctl_table);