]> bbs.cooldavid.org Git - net-next-2.6.git/blame - kernel/sysctl.c
[PATCH] Fix warnings for WARN_ON if CONFIG_BUG is disabled
[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
29cbc78b
AK
55#ifdef CONFIG_X86
56#include <asm/nmi.h>
57#endif
58
1da177e4
LT
59#if defined(CONFIG_SYSCTL)
60
61/* External variables not in a header file. */
62extern int C_A_D;
63extern int sysctl_overcommit_memory;
64extern int sysctl_overcommit_ratio;
fadd8fbd 65extern int sysctl_panic_on_oom;
1da177e4
LT
66extern int max_threads;
67extern int sysrq_enabled;
68extern int core_uses_pid;
d6e71144 69extern int suid_dumpable;
1da177e4 70extern char core_pattern[];
1da177e4
LT
71extern int pid_max;
72extern int min_free_kbytes;
73extern int printk_ratelimit_jiffies;
74extern int printk_ratelimit_burst;
75extern int pid_max_min, pid_max_max;
9d0243bc 76extern int sysctl_drop_caches;
8ad4b1fb 77extern int percpu_pagelist_fraction;
bebfa101 78extern int compat_log;
1da177e4 79
1da177e4
LT
80/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
81static int maxolduid = 65535;
82static int minolduid;
8ad4b1fb 83static int min_percpu_pagelist_fract = 8;
1da177e4
LT
84
85static int ngroups_max = NGROUPS_MAX;
86
87#ifdef CONFIG_KMOD
88extern char modprobe_path[];
89#endif
1da177e4
LT
90#ifdef CONFIG_CHR_DEV_SG
91extern int sg_big_buff;
92#endif
93#ifdef CONFIG_SYSVIPC
fcfbd547
KK
94static int proc_do_ipc_string(ctl_table *table, int write, struct file *filp,
95 void __user *buffer, size_t *lenp, loff_t *ppos);
1da177e4
LT
96#endif
97
98#ifdef __sparc__
99extern char reboot_command [];
100extern int stop_a_enabled;
101extern int scons_pwroff;
102#endif
103
104#ifdef __hppa__
105extern int pwrsw_enabled;
106extern int unaligned_enabled;
107#endif
108
347a8dc3 109#ifdef CONFIG_S390
1da177e4
LT
110#ifdef CONFIG_MATHEMU
111extern int sysctl_ieee_emulation_warnings;
112#endif
113extern int sysctl_userprocess_debug;
951f22d5 114extern int spin_retry;
1da177e4
LT
115#endif
116
117extern int sysctl_hz_timer;
118
119#ifdef CONFIG_BSD_PROCESS_ACCT
120extern int acct_parm[];
121#endif
122
d2b176ed
JS
123#ifdef CONFIG_IA64
124extern int no_unaligned_warning;
125#endif
126
23f78d4a
IM
127#ifdef CONFIG_RT_MUTEXES
128extern int max_lock_depth;
129#endif
130
b89a8171
EB
131#ifdef CONFIG_SYSCTL_SYSCALL
132static int parse_table(int __user *, int, void __user *, size_t __user *,
133 void __user *, size_t, ctl_table *, void **);
134#endif
135
8218c74c 136static int proc_do_uts_string(ctl_table *table, int write, struct file *filp,
1da177e4
LT
137 void __user *buffer, size_t *lenp, loff_t *ppos);
138
9ec52099
CLG
139static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
140 void __user *buffer, size_t *lenp, loff_t *ppos);
141
1da177e4
LT
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
b89a8171 165#ifdef CONFIG_PROC_SYSCTL
1da177e4
LT
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[] = {
8218c74c 229#ifndef CONFIG_UTS_NS
1da177e4
LT
230 {
231 .ctl_name = KERN_OSTYPE,
232 .procname = "ostype",
8218c74c
SH
233 .data = init_uts_ns.name.sysname,
234 .maxlen = sizeof(init_uts_ns.name.sysname),
1da177e4 235 .mode = 0444,
8218c74c 236 .proc_handler = &proc_do_uts_string,
1da177e4
LT
237 .strategy = &sysctl_string,
238 },
239 {
240 .ctl_name = KERN_OSRELEASE,
241 .procname = "osrelease",
8218c74c
SH
242 .data = init_uts_ns.name.release,
243 .maxlen = sizeof(init_uts_ns.name.release),
1da177e4 244 .mode = 0444,
8218c74c 245 .proc_handler = &proc_do_uts_string,
1da177e4
LT
246 .strategy = &sysctl_string,
247 },
248 {
249 .ctl_name = KERN_VERSION,
250 .procname = "version",
8218c74c
SH
251 .data = init_uts_ns.name.version,
252 .maxlen = sizeof(init_uts_ns.name.version),
1da177e4 253 .mode = 0444,
8218c74c 254 .proc_handler = &proc_do_uts_string,
1da177e4
LT
255 .strategy = &sysctl_string,
256 },
257 {
258 .ctl_name = KERN_NODENAME,
259 .procname = "hostname",
8218c74c
SH
260 .data = init_uts_ns.name.nodename,
261 .maxlen = sizeof(init_uts_ns.name.nodename),
1da177e4 262 .mode = 0644,
8218c74c 263 .proc_handler = &proc_do_uts_string,
1da177e4
LT
264 .strategy = &sysctl_string,
265 },
266 {
267 .ctl_name = KERN_DOMAINNAME,
268 .procname = "domainname",
8218c74c
SH
269 .data = init_uts_ns.name.domainname,
270 .maxlen = sizeof(init_uts_ns.name.domainname),
1da177e4 271 .mode = 0644,
8218c74c 272 .proc_handler = &proc_do_uts_string,
1da177e4
LT
273 .strategy = &sysctl_string,
274 },
8218c74c
SH
275#else /* !CONFIG_UTS_NS */
276 {
277 .ctl_name = KERN_OSTYPE,
278 .procname = "ostype",
279 .data = NULL,
280 /* could maybe use __NEW_UTS_LEN here? */
281 .maxlen = FIELD_SIZEOF(struct new_utsname, sysname),
282 .mode = 0444,
283 .proc_handler = &proc_do_uts_string,
284 .strategy = &sysctl_string,
285 },
286 {
287 .ctl_name = KERN_OSRELEASE,
288 .procname = "osrelease",
289 .data = NULL,
290 .maxlen = FIELD_SIZEOF(struct new_utsname, release),
291 .mode = 0444,
292 .proc_handler = &proc_do_uts_string,
293 .strategy = &sysctl_string,
294 },
295 {
296 .ctl_name = KERN_VERSION,
297 .procname = "version",
298 .data = NULL,
299 .maxlen = FIELD_SIZEOF(struct new_utsname, version),
300 .mode = 0444,
301 .proc_handler = &proc_do_uts_string,
302 .strategy = &sysctl_string,
303 },
304 {
305 .ctl_name = KERN_NODENAME,
306 .procname = "hostname",
307 .data = NULL,
308 .maxlen = FIELD_SIZEOF(struct new_utsname, nodename),
309 .mode = 0644,
310 .proc_handler = &proc_do_uts_string,
311 .strategy = &sysctl_string,
312 },
313 {
314 .ctl_name = KERN_DOMAINNAME,
315 .procname = "domainname",
316 .data = NULL,
317 .maxlen = FIELD_SIZEOF(struct new_utsname, domainname),
318 .mode = 0644,
319 .proc_handler = &proc_do_uts_string,
320 .strategy = &sysctl_string,
321 },
322#endif /* !CONFIG_UTS_NS */
1da177e4
LT
323 {
324 .ctl_name = KERN_PANIC,
325 .procname = "panic",
326 .data = &panic_timeout,
327 .maxlen = sizeof(int),
328 .mode = 0644,
329 .proc_handler = &proc_dointvec,
330 },
331 {
332 .ctl_name = KERN_CORE_USES_PID,
333 .procname = "core_uses_pid",
334 .data = &core_uses_pid,
335 .maxlen = sizeof(int),
336 .mode = 0644,
337 .proc_handler = &proc_dointvec,
338 },
339 {
340 .ctl_name = KERN_CORE_PATTERN,
341 .procname = "core_pattern",
342 .data = core_pattern,
d025c9db 343 .maxlen = 128,
1da177e4
LT
344 .mode = 0644,
345 .proc_handler = &proc_dostring,
346 .strategy = &sysctl_string,
347 },
348 {
349 .ctl_name = KERN_TAINTED,
350 .procname = "tainted",
351 .data = &tainted,
352 .maxlen = sizeof(int),
353 .mode = 0444,
354 .proc_handler = &proc_dointvec,
355 },
356 {
357 .ctl_name = KERN_CAP_BSET,
358 .procname = "cap-bound",
359 .data = &cap_bset,
360 .maxlen = sizeof(kernel_cap_t),
361 .mode = 0600,
362 .proc_handler = &proc_dointvec_bset,
363 },
364#ifdef CONFIG_BLK_DEV_INITRD
365 {
366 .ctl_name = KERN_REALROOTDEV,
367 .procname = "real-root-dev",
368 .data = &real_root_dev,
369 .maxlen = sizeof(int),
370 .mode = 0644,
371 .proc_handler = &proc_dointvec,
372 },
373#endif
374#ifdef __sparc__
375 {
376 .ctl_name = KERN_SPARC_REBOOT,
377 .procname = "reboot-cmd",
378 .data = reboot_command,
379 .maxlen = 256,
380 .mode = 0644,
381 .proc_handler = &proc_dostring,
382 .strategy = &sysctl_string,
383 },
384 {
385 .ctl_name = KERN_SPARC_STOP_A,
386 .procname = "stop-a",
387 .data = &stop_a_enabled,
388 .maxlen = sizeof (int),
389 .mode = 0644,
390 .proc_handler = &proc_dointvec,
391 },
392 {
393 .ctl_name = KERN_SPARC_SCONS_PWROFF,
394 .procname = "scons-poweroff",
395 .data = &scons_pwroff,
396 .maxlen = sizeof (int),
397 .mode = 0644,
398 .proc_handler = &proc_dointvec,
399 },
400#endif
401#ifdef __hppa__
402 {
403 .ctl_name = KERN_HPPA_PWRSW,
404 .procname = "soft-power",
405 .data = &pwrsw_enabled,
406 .maxlen = sizeof (int),
407 .mode = 0644,
408 .proc_handler = &proc_dointvec,
409 },
410 {
411 .ctl_name = KERN_HPPA_UNALIGNED,
412 .procname = "unaligned-trap",
413 .data = &unaligned_enabled,
414 .maxlen = sizeof (int),
415 .mode = 0644,
416 .proc_handler = &proc_dointvec,
417 },
418#endif
419 {
420 .ctl_name = KERN_CTLALTDEL,
421 .procname = "ctrl-alt-del",
422 .data = &C_A_D,
423 .maxlen = sizeof(int),
424 .mode = 0644,
425 .proc_handler = &proc_dointvec,
426 },
427 {
428 .ctl_name = KERN_PRINTK,
429 .procname = "printk",
430 .data = &console_loglevel,
431 .maxlen = 4*sizeof(int),
432 .mode = 0644,
433 .proc_handler = &proc_dointvec,
434 },
435#ifdef CONFIG_KMOD
436 {
437 .ctl_name = KERN_MODPROBE,
438 .procname = "modprobe",
439 .data = &modprobe_path,
440 .maxlen = KMOD_PATH_LEN,
441 .mode = 0644,
442 .proc_handler = &proc_dostring,
443 .strategy = &sysctl_string,
444 },
445#endif
57ae2508 446#if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
1da177e4
LT
447 {
448 .ctl_name = KERN_HOTPLUG,
449 .procname = "hotplug",
312c004d
KS
450 .data = &uevent_helper,
451 .maxlen = UEVENT_HELPER_PATH_LEN,
1da177e4
LT
452 .mode = 0644,
453 .proc_handler = &proc_dostring,
454 .strategy = &sysctl_string,
455 },
456#endif
457#ifdef CONFIG_CHR_DEV_SG
458 {
459 .ctl_name = KERN_SG_BIG_BUFF,
460 .procname = "sg-big-buff",
461 .data = &sg_big_buff,
462 .maxlen = sizeof (int),
463 .mode = 0444,
464 .proc_handler = &proc_dointvec,
465 },
466#endif
467#ifdef CONFIG_BSD_PROCESS_ACCT
468 {
469 .ctl_name = KERN_ACCT,
470 .procname = "acct",
471 .data = &acct_parm,
472 .maxlen = 3*sizeof(int),
473 .mode = 0644,
474 .proc_handler = &proc_dointvec,
475 },
476#endif
477#ifdef CONFIG_SYSVIPC
478 {
479 .ctl_name = KERN_SHMMAX,
480 .procname = "shmmax",
fcfbd547 481 .data = NULL,
1da177e4
LT
482 .maxlen = sizeof (size_t),
483 .mode = 0644,
fcfbd547 484 .proc_handler = &proc_do_ipc_string,
1da177e4
LT
485 },
486 {
487 .ctl_name = KERN_SHMALL,
488 .procname = "shmall",
fcfbd547 489 .data = NULL,
1da177e4
LT
490 .maxlen = sizeof (size_t),
491 .mode = 0644,
fcfbd547 492 .proc_handler = &proc_do_ipc_string,
1da177e4
LT
493 },
494 {
495 .ctl_name = KERN_SHMMNI,
496 .procname = "shmmni",
fcfbd547 497 .data = NULL,
1da177e4
LT
498 .maxlen = sizeof (int),
499 .mode = 0644,
fcfbd547 500 .proc_handler = &proc_do_ipc_string,
1da177e4
LT
501 },
502 {
503 .ctl_name = KERN_MSGMAX,
504 .procname = "msgmax",
fcfbd547 505 .data = NULL,
1da177e4
LT
506 .maxlen = sizeof (int),
507 .mode = 0644,
fcfbd547 508 .proc_handler = &proc_do_ipc_string,
1da177e4
LT
509 },
510 {
511 .ctl_name = KERN_MSGMNI,
512 .procname = "msgmni",
fcfbd547 513 .data = NULL,
1da177e4
LT
514 .maxlen = sizeof (int),
515 .mode = 0644,
fcfbd547 516 .proc_handler = &proc_do_ipc_string,
1da177e4
LT
517 },
518 {
519 .ctl_name = KERN_MSGMNB,
520 .procname = "msgmnb",
fcfbd547 521 .data = NULL,
1da177e4
LT
522 .maxlen = sizeof (int),
523 .mode = 0644,
fcfbd547 524 .proc_handler = &proc_do_ipc_string,
1da177e4
LT
525 },
526 {
527 .ctl_name = KERN_SEM,
528 .procname = "sem",
fcfbd547 529 .data = NULL,
1da177e4
LT
530 .maxlen = 4*sizeof (int),
531 .mode = 0644,
fcfbd547 532 .proc_handler = &proc_do_ipc_string,
1da177e4
LT
533 },
534#endif
535#ifdef CONFIG_MAGIC_SYSRQ
536 {
537 .ctl_name = KERN_SYSRQ,
538 .procname = "sysrq",
539 .data = &sysrq_enabled,
540 .maxlen = sizeof (int),
541 .mode = 0644,
542 .proc_handler = &proc_dointvec,
543 },
544#endif
545 {
546 .ctl_name = KERN_CADPID,
547 .procname = "cad_pid",
9ec52099 548 .data = NULL,
1da177e4
LT
549 .maxlen = sizeof (int),
550 .mode = 0600,
9ec52099 551 .proc_handler = &proc_do_cad_pid,
1da177e4
LT
552 },
553 {
554 .ctl_name = KERN_MAX_THREADS,
555 .procname = "threads-max",
556 .data = &max_threads,
557 .maxlen = sizeof(int),
558 .mode = 0644,
559 .proc_handler = &proc_dointvec,
560 },
561 {
562 .ctl_name = KERN_RANDOM,
563 .procname = "random",
564 .mode = 0555,
565 .child = random_table,
566 },
567#ifdef CONFIG_UNIX98_PTYS
568 {
569 .ctl_name = KERN_PTY,
570 .procname = "pty",
571 .mode = 0555,
572 .child = pty_table,
573 },
574#endif
575 {
576 .ctl_name = KERN_OVERFLOWUID,
577 .procname = "overflowuid",
578 .data = &overflowuid,
579 .maxlen = sizeof(int),
580 .mode = 0644,
581 .proc_handler = &proc_dointvec_minmax,
582 .strategy = &sysctl_intvec,
583 .extra1 = &minolduid,
584 .extra2 = &maxolduid,
585 },
586 {
587 .ctl_name = KERN_OVERFLOWGID,
588 .procname = "overflowgid",
589 .data = &overflowgid,
590 .maxlen = sizeof(int),
591 .mode = 0644,
592 .proc_handler = &proc_dointvec_minmax,
593 .strategy = &sysctl_intvec,
594 .extra1 = &minolduid,
595 .extra2 = &maxolduid,
596 },
347a8dc3 597#ifdef CONFIG_S390
1da177e4
LT
598#ifdef CONFIG_MATHEMU
599 {
600 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
601 .procname = "ieee_emulation_warnings",
602 .data = &sysctl_ieee_emulation_warnings,
603 .maxlen = sizeof(int),
604 .mode = 0644,
605 .proc_handler = &proc_dointvec,
606 },
607#endif
608#ifdef CONFIG_NO_IDLE_HZ
609 {
610 .ctl_name = KERN_HZ_TIMER,
611 .procname = "hz_timer",
612 .data = &sysctl_hz_timer,
613 .maxlen = sizeof(int),
614 .mode = 0644,
615 .proc_handler = &proc_dointvec,
616 },
617#endif
618 {
619 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
620 .procname = "userprocess_debug",
621 .data = &sysctl_userprocess_debug,
622 .maxlen = sizeof(int),
623 .mode = 0644,
624 .proc_handler = &proc_dointvec,
625 },
626#endif
627 {
628 .ctl_name = KERN_PIDMAX,
629 .procname = "pid_max",
630 .data = &pid_max,
631 .maxlen = sizeof (int),
632 .mode = 0644,
633 .proc_handler = &proc_dointvec_minmax,
634 .strategy = sysctl_intvec,
635 .extra1 = &pid_max_min,
636 .extra2 = &pid_max_max,
637 },
638 {
639 .ctl_name = KERN_PANIC_ON_OOPS,
640 .procname = "panic_on_oops",
641 .data = &panic_on_oops,
642 .maxlen = sizeof(int),
643 .mode = 0644,
644 .proc_handler = &proc_dointvec,
645 },
646 {
647 .ctl_name = KERN_PRINTK_RATELIMIT,
648 .procname = "printk_ratelimit",
649 .data = &printk_ratelimit_jiffies,
650 .maxlen = sizeof(int),
651 .mode = 0644,
652 .proc_handler = &proc_dointvec_jiffies,
653 .strategy = &sysctl_jiffies,
654 },
655 {
656 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
657 .procname = "printk_ratelimit_burst",
658 .data = &printk_ratelimit_burst,
659 .maxlen = sizeof(int),
660 .mode = 0644,
661 .proc_handler = &proc_dointvec,
662 },
663 {
664 .ctl_name = KERN_NGROUPS_MAX,
665 .procname = "ngroups_max",
666 .data = &ngroups_max,
667 .maxlen = sizeof (int),
668 .mode = 0444,
669 .proc_handler = &proc_dointvec,
670 },
671#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
672 {
673 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
674 .procname = "unknown_nmi_panic",
675 .data = &unknown_nmi_panic,
676 .maxlen = sizeof (int),
677 .mode = 0644,
2fbe7b25 678 .proc_handler = &proc_dointvec,
1da177e4 679 },
407984f1
DZ
680 {
681 .ctl_name = KERN_NMI_WATCHDOG,
682 .procname = "nmi_watchdog",
683 .data = &nmi_watchdog_enabled,
684 .maxlen = sizeof (int),
685 .mode = 0644,
686 .proc_handler = &proc_nmi_enabled,
1da177e4
LT
687 },
688#endif
689#if defined(CONFIG_X86)
8da5adda
DZ
690 {
691 .ctl_name = KERN_PANIC_ON_NMI,
692 .procname = "panic_on_unrecovered_nmi",
693 .data = &panic_on_unrecovered_nmi,
694 .maxlen = sizeof(int),
695 .mode = 0644,
696 .proc_handler = &proc_dointvec,
697 },
1da177e4
LT
698 {
699 .ctl_name = KERN_BOOTLOADER_TYPE,
700 .procname = "bootloader_type",
701 .data = &bootloader_type,
702 .maxlen = sizeof (int),
703 .mode = 0444,
704 .proc_handler = &proc_dointvec,
705 },
706#endif
7a9166e3 707#if defined(CONFIG_MMU)
1da177e4
LT
708 {
709 .ctl_name = KERN_RANDOMIZE,
710 .procname = "randomize_va_space",
711 .data = &randomize_va_space,
712 .maxlen = sizeof(int),
713 .mode = 0644,
714 .proc_handler = &proc_dointvec,
715 },
7a9166e3 716#endif
0152fb37 717#if defined(CONFIG_S390) && defined(CONFIG_SMP)
951f22d5
MS
718 {
719 .ctl_name = KERN_SPIN_RETRY,
720 .procname = "spin_retry",
721 .data = &spin_retry,
722 .maxlen = sizeof (int),
723 .mode = 0644,
724 .proc_handler = &proc_dointvec,
725 },
c255d844
PM
726#endif
727#ifdef CONFIG_ACPI_SLEEP
728 {
729 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
730 .procname = "acpi_video_flags",
731 .data = &acpi_video_flags,
732 .maxlen = sizeof (unsigned long),
733 .mode = 0644,
7f99f06f 734 .proc_handler = &proc_doulongvec_minmax,
c255d844 735 },
d2b176ed
JS
736#endif
737#ifdef CONFIG_IA64
738 {
739 .ctl_name = KERN_IA64_UNALIGNED,
740 .procname = "ignore-unaligned-usertrap",
741 .data = &no_unaligned_warning,
742 .maxlen = sizeof (int),
743 .mode = 0644,
744 .proc_handler = &proc_dointvec,
745 },
bebfa101
AK
746#endif
747#ifdef CONFIG_COMPAT
748 {
749 .ctl_name = KERN_COMPAT_LOG,
750 .procname = "compat-log",
751 .data = &compat_log,
752 .maxlen = sizeof (int),
753 .mode = 0644,
754 .proc_handler = &proc_dointvec,
755 },
951f22d5 756#endif
23f78d4a
IM
757#ifdef CONFIG_RT_MUTEXES
758 {
759 .ctl_name = KERN_MAX_LOCK_DEPTH,
760 .procname = "max_lock_depth",
761 .data = &max_lock_depth,
762 .maxlen = sizeof(int),
763 .mode = 0644,
764 .proc_handler = &proc_dointvec,
765 },
766#endif
767
1da177e4
LT
768 { .ctl_name = 0 }
769};
770
771/* Constants for minimum and maximum testing in vm_table.
772 We use these as one-element integer vectors. */
773static int zero;
774static int one_hundred = 100;
775
776
777static ctl_table vm_table[] = {
778 {
779 .ctl_name = VM_OVERCOMMIT_MEMORY,
780 .procname = "overcommit_memory",
781 .data = &sysctl_overcommit_memory,
782 .maxlen = sizeof(sysctl_overcommit_memory),
783 .mode = 0644,
784 .proc_handler = &proc_dointvec,
785 },
fadd8fbd
KH
786 {
787 .ctl_name = VM_PANIC_ON_OOM,
788 .procname = "panic_on_oom",
789 .data = &sysctl_panic_on_oom,
790 .maxlen = sizeof(sysctl_panic_on_oom),
791 .mode = 0644,
792 .proc_handler = &proc_dointvec,
793 },
1da177e4
LT
794 {
795 .ctl_name = VM_OVERCOMMIT_RATIO,
796 .procname = "overcommit_ratio",
797 .data = &sysctl_overcommit_ratio,
798 .maxlen = sizeof(sysctl_overcommit_ratio),
799 .mode = 0644,
800 .proc_handler = &proc_dointvec,
801 },
802 {
803 .ctl_name = VM_PAGE_CLUSTER,
804 .procname = "page-cluster",
805 .data = &page_cluster,
806 .maxlen = sizeof(int),
807 .mode = 0644,
808 .proc_handler = &proc_dointvec,
809 },
810 {
811 .ctl_name = VM_DIRTY_BACKGROUND,
812 .procname = "dirty_background_ratio",
813 .data = &dirty_background_ratio,
814 .maxlen = sizeof(dirty_background_ratio),
815 .mode = 0644,
816 .proc_handler = &proc_dointvec_minmax,
817 .strategy = &sysctl_intvec,
818 .extra1 = &zero,
819 .extra2 = &one_hundred,
820 },
821 {
822 .ctl_name = VM_DIRTY_RATIO,
823 .procname = "dirty_ratio",
824 .data = &vm_dirty_ratio,
825 .maxlen = sizeof(vm_dirty_ratio),
826 .mode = 0644,
827 .proc_handler = &proc_dointvec_minmax,
828 .strategy = &sysctl_intvec,
829 .extra1 = &zero,
830 .extra2 = &one_hundred,
831 },
832 {
833 .ctl_name = VM_DIRTY_WB_CS,
834 .procname = "dirty_writeback_centisecs",
f6ef9438
BS
835 .data = &dirty_writeback_interval,
836 .maxlen = sizeof(dirty_writeback_interval),
1da177e4
LT
837 .mode = 0644,
838 .proc_handler = &dirty_writeback_centisecs_handler,
839 },
840 {
841 .ctl_name = VM_DIRTY_EXPIRE_CS,
842 .procname = "dirty_expire_centisecs",
f6ef9438
BS
843 .data = &dirty_expire_interval,
844 .maxlen = sizeof(dirty_expire_interval),
1da177e4 845 .mode = 0644,
f6ef9438 846 .proc_handler = &proc_dointvec_userhz_jiffies,
1da177e4
LT
847 },
848 {
849 .ctl_name = VM_NR_PDFLUSH_THREADS,
850 .procname = "nr_pdflush_threads",
851 .data = &nr_pdflush_threads,
852 .maxlen = sizeof nr_pdflush_threads,
853 .mode = 0444 /* read-only*/,
854 .proc_handler = &proc_dointvec,
855 },
856 {
857 .ctl_name = VM_SWAPPINESS,
858 .procname = "swappiness",
859 .data = &vm_swappiness,
860 .maxlen = sizeof(vm_swappiness),
861 .mode = 0644,
862 .proc_handler = &proc_dointvec_minmax,
863 .strategy = &sysctl_intvec,
864 .extra1 = &zero,
865 .extra2 = &one_hundred,
866 },
867#ifdef CONFIG_HUGETLB_PAGE
868 {
869 .ctl_name = VM_HUGETLB_PAGES,
870 .procname = "nr_hugepages",
871 .data = &max_huge_pages,
872 .maxlen = sizeof(unsigned long),
873 .mode = 0644,
874 .proc_handler = &hugetlb_sysctl_handler,
875 .extra1 = (void *)&hugetlb_zero,
876 .extra2 = (void *)&hugetlb_infinity,
877 },
878 {
879 .ctl_name = VM_HUGETLB_GROUP,
880 .procname = "hugetlb_shm_group",
881 .data = &sysctl_hugetlb_shm_group,
882 .maxlen = sizeof(gid_t),
883 .mode = 0644,
884 .proc_handler = &proc_dointvec,
885 },
886#endif
887 {
888 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
889 .procname = "lowmem_reserve_ratio",
890 .data = &sysctl_lowmem_reserve_ratio,
891 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
892 .mode = 0644,
893 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
894 .strategy = &sysctl_intvec,
895 },
9d0243bc
AM
896 {
897 .ctl_name = VM_DROP_PAGECACHE,
898 .procname = "drop_caches",
899 .data = &sysctl_drop_caches,
900 .maxlen = sizeof(int),
901 .mode = 0644,
902 .proc_handler = drop_caches_sysctl_handler,
903 .strategy = &sysctl_intvec,
904 },
1da177e4
LT
905 {
906 .ctl_name = VM_MIN_FREE_KBYTES,
907 .procname = "min_free_kbytes",
908 .data = &min_free_kbytes,
909 .maxlen = sizeof(min_free_kbytes),
910 .mode = 0644,
911 .proc_handler = &min_free_kbytes_sysctl_handler,
912 .strategy = &sysctl_intvec,
913 .extra1 = &zero,
914 },
8ad4b1fb
RS
915 {
916 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
917 .procname = "percpu_pagelist_fraction",
918 .data = &percpu_pagelist_fraction,
919 .maxlen = sizeof(percpu_pagelist_fraction),
920 .mode = 0644,
921 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
922 .strategy = &sysctl_intvec,
923 .extra1 = &min_percpu_pagelist_fract,
924 },
1da177e4
LT
925#ifdef CONFIG_MMU
926 {
927 .ctl_name = VM_MAX_MAP_COUNT,
928 .procname = "max_map_count",
929 .data = &sysctl_max_map_count,
930 .maxlen = sizeof(sysctl_max_map_count),
931 .mode = 0644,
932 .proc_handler = &proc_dointvec
933 },
934#endif
935 {
936 .ctl_name = VM_LAPTOP_MODE,
937 .procname = "laptop_mode",
938 .data = &laptop_mode,
939 .maxlen = sizeof(laptop_mode),
940 .mode = 0644,
ed5b43f1
BS
941 .proc_handler = &proc_dointvec_jiffies,
942 .strategy = &sysctl_jiffies,
1da177e4
LT
943 },
944 {
945 .ctl_name = VM_BLOCK_DUMP,
946 .procname = "block_dump",
947 .data = &block_dump,
948 .maxlen = sizeof(block_dump),
949 .mode = 0644,
950 .proc_handler = &proc_dointvec,
951 .strategy = &sysctl_intvec,
952 .extra1 = &zero,
953 },
954 {
955 .ctl_name = VM_VFS_CACHE_PRESSURE,
956 .procname = "vfs_cache_pressure",
957 .data = &sysctl_vfs_cache_pressure,
958 .maxlen = sizeof(sysctl_vfs_cache_pressure),
959 .mode = 0644,
960 .proc_handler = &proc_dointvec,
961 .strategy = &sysctl_intvec,
962 .extra1 = &zero,
963 },
964#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
965 {
966 .ctl_name = VM_LEGACY_VA_LAYOUT,
967 .procname = "legacy_va_layout",
968 .data = &sysctl_legacy_va_layout,
969 .maxlen = sizeof(sysctl_legacy_va_layout),
970 .mode = 0644,
971 .proc_handler = &proc_dointvec,
972 .strategy = &sysctl_intvec,
973 .extra1 = &zero,
974 },
975#endif
976#ifdef CONFIG_SWAP
977 {
978 .ctl_name = VM_SWAP_TOKEN_TIMEOUT,
979 .procname = "swap_token_timeout",
980 .data = &swap_token_default_timeout,
981 .maxlen = sizeof(swap_token_default_timeout),
982 .mode = 0644,
983 .proc_handler = &proc_dointvec_jiffies,
984 .strategy = &sysctl_jiffies,
985 },
1743660b
CL
986#endif
987#ifdef CONFIG_NUMA
988 {
989 .ctl_name = VM_ZONE_RECLAIM_MODE,
990 .procname = "zone_reclaim_mode",
991 .data = &zone_reclaim_mode,
992 .maxlen = sizeof(zone_reclaim_mode),
993 .mode = 0644,
994 .proc_handler = &proc_dointvec,
c84db23c
CL
995 .strategy = &sysctl_intvec,
996 .extra1 = &zero,
1743660b 997 },
9614634f
CL
998 {
999 .ctl_name = VM_MIN_UNMAPPED,
1000 .procname = "min_unmapped_ratio",
1001 .data = &sysctl_min_unmapped_ratio,
1002 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1003 .mode = 0644,
1004 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
1005 .strategy = &sysctl_intvec,
1006 .extra1 = &zero,
1007 .extra2 = &one_hundred,
1008 },
0ff38490
CL
1009 {
1010 .ctl_name = VM_MIN_SLAB,
1011 .procname = "min_slab_ratio",
1012 .data = &sysctl_min_slab_ratio,
1013 .maxlen = sizeof(sysctl_min_slab_ratio),
1014 .mode = 0644,
1015 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1016 .strategy = &sysctl_intvec,
1017 .extra1 = &zero,
1018 .extra2 = &one_hundred,
1019 },
e6e5494c
IM
1020#endif
1021#ifdef CONFIG_X86_32
1022 {
1023 .ctl_name = VM_VDSO_ENABLED,
1024 .procname = "vdso_enabled",
1025 .data = &vdso_enabled,
1026 .maxlen = sizeof(vdso_enabled),
1027 .mode = 0644,
1028 .proc_handler = &proc_dointvec,
1029 .strategy = &sysctl_intvec,
1030 .extra1 = &zero,
1031 },
1da177e4
LT
1032#endif
1033 { .ctl_name = 0 }
1034};
1035
1da177e4
LT
1036static ctl_table fs_table[] = {
1037 {
1038 .ctl_name = FS_NRINODE,
1039 .procname = "inode-nr",
1040 .data = &inodes_stat,
1041 .maxlen = 2*sizeof(int),
1042 .mode = 0444,
1043 .proc_handler = &proc_dointvec,
1044 },
1045 {
1046 .ctl_name = FS_STATINODE,
1047 .procname = "inode-state",
1048 .data = &inodes_stat,
1049 .maxlen = 7*sizeof(int),
1050 .mode = 0444,
1051 .proc_handler = &proc_dointvec,
1052 },
1053 {
1054 .ctl_name = FS_NRFILE,
1055 .procname = "file-nr",
1056 .data = &files_stat,
1057 .maxlen = 3*sizeof(int),
1058 .mode = 0444,
529bf6be 1059 .proc_handler = &proc_nr_files,
1da177e4
LT
1060 },
1061 {
1062 .ctl_name = FS_MAXFILE,
1063 .procname = "file-max",
1064 .data = &files_stat.max_files,
1065 .maxlen = sizeof(int),
1066 .mode = 0644,
1067 .proc_handler = &proc_dointvec,
1068 },
1069 {
1070 .ctl_name = FS_DENTRY,
1071 .procname = "dentry-state",
1072 .data = &dentry_stat,
1073 .maxlen = 6*sizeof(int),
1074 .mode = 0444,
1075 .proc_handler = &proc_dointvec,
1076 },
1077 {
1078 .ctl_name = FS_OVERFLOWUID,
1079 .procname = "overflowuid",
1080 .data = &fs_overflowuid,
1081 .maxlen = sizeof(int),
1082 .mode = 0644,
1083 .proc_handler = &proc_dointvec_minmax,
1084 .strategy = &sysctl_intvec,
1085 .extra1 = &minolduid,
1086 .extra2 = &maxolduid,
1087 },
1088 {
1089 .ctl_name = FS_OVERFLOWGID,
1090 .procname = "overflowgid",
1091 .data = &fs_overflowgid,
1092 .maxlen = sizeof(int),
1093 .mode = 0644,
1094 .proc_handler = &proc_dointvec_minmax,
1095 .strategy = &sysctl_intvec,
1096 .extra1 = &minolduid,
1097 .extra2 = &maxolduid,
1098 },
1099 {
1100 .ctl_name = FS_LEASES,
1101 .procname = "leases-enable",
1102 .data = &leases_enable,
1103 .maxlen = sizeof(int),
1104 .mode = 0644,
1105 .proc_handler = &proc_dointvec,
1106 },
1107#ifdef CONFIG_DNOTIFY
1108 {
1109 .ctl_name = FS_DIR_NOTIFY,
1110 .procname = "dir-notify-enable",
1111 .data = &dir_notify_enable,
1112 .maxlen = sizeof(int),
1113 .mode = 0644,
1114 .proc_handler = &proc_dointvec,
1115 },
1116#endif
1117#ifdef CONFIG_MMU
1118 {
1119 .ctl_name = FS_LEASE_TIME,
1120 .procname = "lease-break-time",
1121 .data = &lease_break_time,
1122 .maxlen = sizeof(int),
1123 .mode = 0644,
1124 .proc_handler = &proc_dointvec,
1125 },
1126 {
1127 .ctl_name = FS_AIO_NR,
1128 .procname = "aio-nr",
1129 .data = &aio_nr,
1130 .maxlen = sizeof(aio_nr),
1131 .mode = 0444,
d55b5fda 1132 .proc_handler = &proc_doulongvec_minmax,
1da177e4
LT
1133 },
1134 {
1135 .ctl_name = FS_AIO_MAX_NR,
1136 .procname = "aio-max-nr",
1137 .data = &aio_max_nr,
1138 .maxlen = sizeof(aio_max_nr),
1139 .mode = 0644,
d55b5fda 1140 .proc_handler = &proc_doulongvec_minmax,
1da177e4 1141 },
2d9048e2 1142#ifdef CONFIG_INOTIFY_USER
0399cb08
RL
1143 {
1144 .ctl_name = FS_INOTIFY,
1145 .procname = "inotify",
1146 .mode = 0555,
1147 .child = inotify_table,
1148 },
1149#endif
1da177e4 1150#endif
d6e71144
AC
1151 {
1152 .ctl_name = KERN_SETUID_DUMPABLE,
1153 .procname = "suid_dumpable",
1154 .data = &suid_dumpable,
1155 .maxlen = sizeof(int),
1156 .mode = 0644,
1157 .proc_handler = &proc_dointvec,
1158 },
1da177e4
LT
1159 { .ctl_name = 0 }
1160};
1161
1162static ctl_table debug_table[] = {
1163 { .ctl_name = 0 }
1164};
1165
1166static ctl_table dev_table[] = {
1167 { .ctl_name = 0 }
0eeca283 1168};
1da177e4
LT
1169
1170extern void init_irq_proc (void);
1171
330d57fb
AV
1172static DEFINE_SPINLOCK(sysctl_lock);
1173
1174/* called under sysctl_lock */
1175static int use_table(struct ctl_table_header *p)
1176{
1177 if (unlikely(p->unregistering))
1178 return 0;
1179 p->used++;
1180 return 1;
1181}
1182
1183/* called under sysctl_lock */
1184static void unuse_table(struct ctl_table_header *p)
1185{
1186 if (!--p->used)
1187 if (unlikely(p->unregistering))
1188 complete(p->unregistering);
1189}
1190
1191/* called under sysctl_lock, will reacquire if has to wait */
1192static void start_unregistering(struct ctl_table_header *p)
1193{
1194 /*
1195 * if p->used is 0, nobody will ever touch that entry again;
1196 * we'll eliminate all paths to it before dropping sysctl_lock
1197 */
1198 if (unlikely(p->used)) {
1199 struct completion wait;
1200 init_completion(&wait);
1201 p->unregistering = &wait;
1202 spin_unlock(&sysctl_lock);
1203 wait_for_completion(&wait);
1204 spin_lock(&sysctl_lock);
1205 }
1206 /*
1207 * do not remove from the list until nobody holds it; walking the
1208 * list in do_sysctl() relies on that.
1209 */
1210 list_del_init(&p->ctl_entry);
1211}
1212
1da177e4
LT
1213void __init sysctl_init(void)
1214{
b89a8171 1215#ifdef CONFIG_PROC_SYSCTL
330d57fb 1216 register_proc_table(root_table, proc_sys_root, &root_table_header);
1da177e4
LT
1217 init_irq_proc();
1218#endif
1219}
1220
b89a8171 1221#ifdef CONFIG_SYSCTL_SYSCALL
1da177e4
LT
1222int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1223 void __user *newval, size_t newlen)
1224{
1225 struct list_head *tmp;
330d57fb 1226 int error = -ENOTDIR;
1da177e4
LT
1227
1228 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1229 return -ENOTDIR;
1230 if (oldval) {
1231 int old_len;
1232 if (!oldlenp || get_user(old_len, oldlenp))
1233 return -EFAULT;
1234 }
330d57fb 1235 spin_lock(&sysctl_lock);
1da177e4
LT
1236 tmp = &root_table_header.ctl_entry;
1237 do {
1238 struct ctl_table_header *head =
1239 list_entry(tmp, struct ctl_table_header, ctl_entry);
1240 void *context = NULL;
330d57fb
AV
1241
1242 if (!use_table(head))
1243 continue;
1244
1245 spin_unlock(&sysctl_lock);
1246
1247 error = parse_table(name, nlen, oldval, oldlenp,
1da177e4
LT
1248 newval, newlen, head->ctl_table,
1249 &context);
5a6b454f 1250 kfree(context);
330d57fb
AV
1251
1252 spin_lock(&sysctl_lock);
1253 unuse_table(head);
1da177e4 1254 if (error != -ENOTDIR)
330d57fb
AV
1255 break;
1256 } while ((tmp = tmp->next) != &root_table_header.ctl_entry);
1257 spin_unlock(&sysctl_lock);
1258 return error;
1da177e4
LT
1259}
1260
1261asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1262{
1263 struct __sysctl_args tmp;
1264 int error;
1265
1266 if (copy_from_user(&tmp, args, sizeof(tmp)))
1267 return -EFAULT;
1268
1269 lock_kernel();
1270 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1271 tmp.newval, tmp.newlen);
1272 unlock_kernel();
1273 return error;
1274}
b89a8171 1275#endif /* CONFIG_SYSCTL_SYSCALL */
1da177e4
LT
1276
1277/*
1278 * ctl_perm does NOT grant the superuser all rights automatically, because
1279 * some sysctl variables are readonly even to root.
1280 */
1281
1282static int test_perm(int mode, int op)
1283{
1284 if (!current->euid)
1285 mode >>= 6;
1286 else if (in_egroup_p(0))
1287 mode >>= 3;
1288 if ((mode & op & 0007) == op)
1289 return 0;
1290 return -EACCES;
1291}
1292
1293static inline int ctl_perm(ctl_table *table, int op)
1294{
1295 int error;
1296 error = security_sysctl(table, op);
1297 if (error)
1298 return error;
1299 return test_perm(table->mode, op);
1300}
1301
b89a8171 1302#ifdef CONFIG_SYSCTL_SYSCALL
1da177e4
LT
1303static int parse_table(int __user *name, int nlen,
1304 void __user *oldval, size_t __user *oldlenp,
1305 void __user *newval, size_t newlen,
1306 ctl_table *table, void **context)
1307{
1308 int n;
1309repeat:
1310 if (!nlen)
1311 return -ENOTDIR;
1312 if (get_user(n, name))
1313 return -EFAULT;
1314 for ( ; table->ctl_name; table++) {
1315 if (n == table->ctl_name || table->ctl_name == CTL_ANY) {
1316 int error;
1317 if (table->child) {
1318 if (ctl_perm(table, 001))
1319 return -EPERM;
1320 if (table->strategy) {
1321 error = table->strategy(
1322 table, name, nlen,
1323 oldval, oldlenp,
1324 newval, newlen, context);
1325 if (error)
1326 return error;
1327 }
1328 name++;
1329 nlen--;
1330 table = table->child;
1331 goto repeat;
1332 }
1333 error = do_sysctl_strategy(table, name, nlen,
1334 oldval, oldlenp,
1335 newval, newlen, context);
1336 return error;
1337 }
1338 }
1339 return -ENOTDIR;
1340}
1341
1342/* Perform the actual read/write of a sysctl table entry. */
1343int do_sysctl_strategy (ctl_table *table,
1344 int __user *name, int nlen,
1345 void __user *oldval, size_t __user *oldlenp,
1346 void __user *newval, size_t newlen, void **context)
1347{
1348 int op = 0, rc;
1349 size_t len;
1350
1351 if (oldval)
1352 op |= 004;
1353 if (newval)
1354 op |= 002;
1355 if (ctl_perm(table, op))
1356 return -EPERM;
1357
1358 if (table->strategy) {
1359 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1360 newval, newlen, context);
1361 if (rc < 0)
1362 return rc;
1363 if (rc > 0)
1364 return 0;
1365 }
1366
1367 /* If there is no strategy routine, or if the strategy returns
1368 * zero, proceed with automatic r/w */
1369 if (table->data && table->maxlen) {
1370 if (oldval && oldlenp) {
1371 if (get_user(len, oldlenp))
1372 return -EFAULT;
1373 if (len) {
1374 if (len > table->maxlen)
1375 len = table->maxlen;
1376 if(copy_to_user(oldval, table->data, len))
1377 return -EFAULT;
1378 if(put_user(len, oldlenp))
1379 return -EFAULT;
1380 }
1381 }
1382 if (newval && newlen) {
1383 len = newlen;
1384 if (len > table->maxlen)
1385 len = table->maxlen;
1386 if(copy_from_user(table->data, newval, len))
1387 return -EFAULT;
1388 }
1389 }
1390 return 0;
1391}
b89a8171 1392#endif /* CONFIG_SYSCTL_SYSCALL */
1da177e4
LT
1393
1394/**
1395 * register_sysctl_table - register a sysctl hierarchy
1396 * @table: the top-level table structure
1397 * @insert_at_head: whether the entry should be inserted in front or at the end
1398 *
1399 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1400 * array. An entry with a ctl_name of 0 terminates the table.
1401 *
1402 * The members of the &ctl_table structure are used as follows:
1403 *
1404 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1405 * must be unique within that level of sysctl
1406 *
1407 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1408 * enter a sysctl file
1409 *
1410 * data - a pointer to data for use by proc_handler
1411 *
1412 * maxlen - the maximum size in bytes of the data
1413 *
1414 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1415 *
1416 * child - a pointer to the child sysctl table if this entry is a directory, or
1417 * %NULL.
1418 *
1419 * proc_handler - the text handler routine (described below)
1420 *
1421 * strategy - the strategy routine (described below)
1422 *
1423 * de - for internal use by the sysctl routines
1424 *
1425 * extra1, extra2 - extra pointers usable by the proc handler routines
1426 *
1427 * Leaf nodes in the sysctl tree will be represented by a single file
1428 * under /proc; non-leaf nodes will be represented by directories.
1429 *
1430 * sysctl(2) can automatically manage read and write requests through
1431 * the sysctl table. The data and maxlen fields of the ctl_table
1432 * struct enable minimal validation of the values being written to be
1433 * performed, and the mode field allows minimal authentication.
1434 *
1435 * More sophisticated management can be enabled by the provision of a
1436 * strategy routine with the table entry. This will be called before
1437 * any automatic read or write of the data is performed.
1438 *
1439 * The strategy routine may return
1440 *
1441 * < 0 - Error occurred (error is passed to user process)
1442 *
1443 * 0 - OK - proceed with automatic read or write.
1444 *
1445 * > 0 - OK - read or write has been done by the strategy routine, so
1446 * return immediately.
1447 *
1448 * There must be a proc_handler routine for any terminal nodes
1449 * mirrored under /proc/sys (non-terminals are handled by a built-in
1450 * directory handler). Several default handlers are available to
1451 * cover common cases -
1452 *
1453 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1454 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1455 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1456 *
1457 * It is the handler's job to read the input buffer from user memory
1458 * and process it. The handler should return 0 on success.
1459 *
1460 * This routine returns %NULL on a failure to register, and a pointer
1461 * to the table header on success.
1462 */
1463struct ctl_table_header *register_sysctl_table(ctl_table * table,
1464 int insert_at_head)
1465{
1466 struct ctl_table_header *tmp;
1467 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1468 if (!tmp)
1469 return NULL;
1470 tmp->ctl_table = table;
1471 INIT_LIST_HEAD(&tmp->ctl_entry);
330d57fb
AV
1472 tmp->used = 0;
1473 tmp->unregistering = NULL;
1474 spin_lock(&sysctl_lock);
1da177e4
LT
1475 if (insert_at_head)
1476 list_add(&tmp->ctl_entry, &root_table_header.ctl_entry);
1477 else
1478 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
330d57fb 1479 spin_unlock(&sysctl_lock);
b89a8171 1480#ifdef CONFIG_PROC_SYSCTL
330d57fb 1481 register_proc_table(table, proc_sys_root, tmp);
1da177e4
LT
1482#endif
1483 return tmp;
1484}
1485
1486/**
1487 * unregister_sysctl_table - unregister a sysctl table hierarchy
1488 * @header: the header returned from register_sysctl_table
1489 *
1490 * Unregisters the sysctl table and all children. proc entries may not
1491 * actually be removed until they are no longer used by anyone.
1492 */
1493void unregister_sysctl_table(struct ctl_table_header * header)
1494{
330d57fb
AV
1495 might_sleep();
1496 spin_lock(&sysctl_lock);
1497 start_unregistering(header);
b89a8171 1498#ifdef CONFIG_PROC_SYSCTL
1da177e4
LT
1499 unregister_proc_table(header->ctl_table, proc_sys_root);
1500#endif
330d57fb 1501 spin_unlock(&sysctl_lock);
1da177e4
LT
1502 kfree(header);
1503}
1504
b89a8171
EB
1505#else /* !CONFIG_SYSCTL */
1506struct ctl_table_header * register_sysctl_table(ctl_table * table,
1507 int insert_at_head)
1508{
1509 return NULL;
1510}
1511
1512void unregister_sysctl_table(struct ctl_table_header * table)
1513{
1514}
1515
1516#endif /* CONFIG_SYSCTL */
1517
1da177e4
LT
1518/*
1519 * /proc/sys support
1520 */
1521
b89a8171 1522#ifdef CONFIG_PROC_SYSCTL
1da177e4
LT
1523
1524/* Scan the sysctl entries in table and add them all into /proc */
330d57fb 1525static void register_proc_table(ctl_table * table, struct proc_dir_entry *root, void *set)
1da177e4
LT
1526{
1527 struct proc_dir_entry *de;
1528 int len;
1529 mode_t mode;
1530
1531 for (; table->ctl_name; table++) {
1532 /* Can't do anything without a proc name. */
1533 if (!table->procname)
1534 continue;
1535 /* Maybe we can't do anything with it... */
1536 if (!table->proc_handler && !table->child) {
1537 printk(KERN_WARNING "SYSCTL: Can't register %s\n",
1538 table->procname);
1539 continue;
1540 }
1541
1542 len = strlen(table->procname);
1543 mode = table->mode;
1544
1545 de = NULL;
1546 if (table->proc_handler)
1547 mode |= S_IFREG;
1548 else {
1549 mode |= S_IFDIR;
1550 for (de = root->subdir; de; de = de->next) {
1551 if (proc_match(len, table->procname, de))
1552 break;
1553 }
1554 /* If the subdir exists already, de is non-NULL */
1555 }
1556
1557 if (!de) {
1558 de = create_proc_entry(table->procname, mode, root);
1559 if (!de)
1560 continue;
330d57fb 1561 de->set = set;
1da177e4
LT
1562 de->data = (void *) table;
1563 if (table->proc_handler)
1564 de->proc_fops = &proc_sys_file_operations;
1565 }
1566 table->de = de;
1567 if (de->mode & S_IFDIR)
330d57fb 1568 register_proc_table(table->child, de, set);
1da177e4
LT
1569 }
1570}
1571
1572/*
1573 * Unregister a /proc sysctl table and any subdirectories.
1574 */
1575static void unregister_proc_table(ctl_table * table, struct proc_dir_entry *root)
1576{
1577 struct proc_dir_entry *de;
1578 for (; table->ctl_name; table++) {
1579 if (!(de = table->de))
1580 continue;
1581 if (de->mode & S_IFDIR) {
1582 if (!table->child) {
1583 printk (KERN_ALERT "Help - malformed sysctl tree on free\n");
1584 continue;
1585 }
1586 unregister_proc_table(table->child, de);
1587
1588 /* Don't unregister directories which still have entries.. */
1589 if (de->subdir)
1590 continue;
1591 }
1592
330d57fb
AV
1593 /*
1594 * In any case, mark the entry as goner; we'll keep it
1595 * around if it's busy, but we'll know to do nothing with
1596 * its fields. We are under sysctl_lock here.
1597 */
1598 de->data = NULL;
1599
1da177e4
LT
1600 /* Don't unregister proc entries that are still being used.. */
1601 if (atomic_read(&de->count))
1602 continue;
1603
1604 table->de = NULL;
1605 remove_proc_entry(table->procname, root);
1606 }
1607}
1608
1609static ssize_t do_rw_proc(int write, struct file * file, char __user * buf,
1610 size_t count, loff_t *ppos)
1611{
1612 int op;
330d57fb 1613 struct proc_dir_entry *de = PDE(file->f_dentry->d_inode);
1da177e4
LT
1614 struct ctl_table *table;
1615 size_t res;
330d57fb 1616 ssize_t error = -ENOTDIR;
1da177e4 1617
330d57fb
AV
1618 spin_lock(&sysctl_lock);
1619 if (de && de->data && use_table(de->set)) {
1620 /*
1621 * at that point we know that sysctl was not unregistered
1622 * and won't be until we finish
1623 */
1624 spin_unlock(&sysctl_lock);
1625 table = (struct ctl_table *) de->data;
1626 if (!table || !table->proc_handler)
1627 goto out;
1628 error = -EPERM;
1629 op = (write ? 002 : 004);
1630 if (ctl_perm(table, op))
1631 goto out;
1632
1633 /* careful: calling conventions are nasty here */
1634 res = count;
1635 error = (*table->proc_handler)(table, write, file,
1636 buf, &res, ppos);
1637 if (!error)
1638 error = res;
1639 out:
1640 spin_lock(&sysctl_lock);
1641 unuse_table(de->set);
1642 }
1643 spin_unlock(&sysctl_lock);
1644 return error;
1da177e4
LT
1645}
1646
1647static int proc_opensys(struct inode *inode, struct file *file)
1648{
1649 if (file->f_mode & FMODE_WRITE) {
1650 /*
1651 * sysctl entries that are not writable,
1652 * are _NOT_ writable, capabilities or not.
1653 */
1654 if (!(inode->i_mode & S_IWUSR))
1655 return -EPERM;
1656 }
1657
1658 return 0;
1659}
1660
1661static ssize_t proc_readsys(struct file * file, char __user * buf,
1662 size_t count, loff_t *ppos)
1663{
1664 return do_rw_proc(0, file, buf, count, ppos);
1665}
1666
1667static ssize_t proc_writesys(struct file * file, const char __user * buf,
1668 size_t count, loff_t *ppos)
1669{
1670 return do_rw_proc(1, file, (char __user *) buf, count, ppos);
1671}
1672
b1ba4ddd
AB
1673static int _proc_do_string(void* data, int maxlen, int write,
1674 struct file *filp, void __user *buffer,
1675 size_t *lenp, loff_t *ppos)
1da177e4
LT
1676{
1677 size_t len;
1678 char __user *p;
1679 char c;
1680
f5dd3d6f 1681 if (!data || !maxlen || !*lenp ||
1da177e4
LT
1682 (*ppos && !write)) {
1683 *lenp = 0;
1684 return 0;
1685 }
1686
1687 if (write) {
1688 len = 0;
1689 p = buffer;
1690 while (len < *lenp) {
1691 if (get_user(c, p++))
1692 return -EFAULT;
1693 if (c == 0 || c == '\n')
1694 break;
1695 len++;
1696 }
f5dd3d6f
SV
1697 if (len >= maxlen)
1698 len = maxlen-1;
1699 if(copy_from_user(data, buffer, len))
1da177e4 1700 return -EFAULT;
f5dd3d6f 1701 ((char *) data)[len] = 0;
1da177e4
LT
1702 *ppos += *lenp;
1703 } else {
f5dd3d6f
SV
1704 len = strlen(data);
1705 if (len > maxlen)
1706 len = maxlen;
1da177e4
LT
1707 if (len > *lenp)
1708 len = *lenp;
1709 if (len)
f5dd3d6f 1710 if(copy_to_user(buffer, data, len))
1da177e4
LT
1711 return -EFAULT;
1712 if (len < *lenp) {
1713 if(put_user('\n', ((char __user *) buffer) + len))
1714 return -EFAULT;
1715 len++;
1716 }
1717 *lenp = len;
1718 *ppos += len;
1719 }
1720 return 0;
1721}
1722
f5dd3d6f
SV
1723/**
1724 * proc_dostring - read a string sysctl
1725 * @table: the sysctl table
1726 * @write: %TRUE if this is a write to the sysctl file
1727 * @filp: the file structure
1728 * @buffer: the user buffer
1729 * @lenp: the size of the user buffer
1730 * @ppos: file position
1731 *
1732 * Reads/writes a string from/to the user buffer. If the kernel
1733 * buffer provided is not large enough to hold the string, the
1734 * string is truncated. The copied string is %NULL-terminated.
1735 * If the string is being read by the user process, it is copied
1736 * and a newline '\n' is added. It is truncated if the buffer is
1737 * not large enough.
1738 *
1739 * Returns 0 on success.
1740 */
1741int proc_dostring(ctl_table *table, int write, struct file *filp,
1742 void __user *buffer, size_t *lenp, loff_t *ppos)
1743{
1744 return _proc_do_string(table->data, table->maxlen, write, filp,
1745 buffer, lenp, ppos);
1746}
1747
1da177e4
LT
1748/*
1749 * Special case of dostring for the UTS structure. This has locks
1750 * to observe. Should this be in kernel/sys.c ????
1751 */
1752
8218c74c
SH
1753#ifndef CONFIG_UTS_NS
1754static int proc_do_uts_string(ctl_table *table, int write, struct file *filp,
1da177e4
LT
1755 void __user *buffer, size_t *lenp, loff_t *ppos)
1756{
1757 int r;
1758
1759 if (!write) {
1760 down_read(&uts_sem);
1761 r=proc_dostring(table,0,filp,buffer,lenp, ppos);
1762 up_read(&uts_sem);
1763 } else {
1764 down_write(&uts_sem);
1765 r=proc_dostring(table,1,filp,buffer,lenp, ppos);
1766 up_write(&uts_sem);
1767 }
1768 return r;
1769}
8218c74c
SH
1770#else /* !CONFIG_UTS_NS */
1771static int proc_do_uts_string(ctl_table *table, int write, struct file *filp,
1772 void __user *buffer, size_t *lenp, loff_t *ppos)
1773{
1774 int r;
1775 struct uts_namespace* uts_ns = current->nsproxy->uts_ns;
1776 char* which;
1777
1778 switch (table->ctl_name) {
1779 case KERN_OSTYPE:
1780 which = uts_ns->name.sysname;
1781 break;
1782 case KERN_NODENAME:
1783 which = uts_ns->name.nodename;
1784 break;
1785 case KERN_OSRELEASE:
1786 which = uts_ns->name.release;
1787 break;
1788 case KERN_VERSION:
1789 which = uts_ns->name.version;
1790 break;
1791 case KERN_DOMAINNAME:
1792 which = uts_ns->name.domainname;
1793 break;
1794 default:
1795 r = -EINVAL;
1796 goto out;
1797 }
1798
1799 if (!write) {
1800 down_read(&uts_sem);
1801 r=_proc_do_string(which,table->maxlen,0,filp,buffer,lenp, ppos);
1802 up_read(&uts_sem);
1803 } else {
1804 down_write(&uts_sem);
1805 r=_proc_do_string(which,table->maxlen,1,filp,buffer,lenp, ppos);
1806 up_write(&uts_sem);
1807 }
1808 out:
1809 return r;
1810}
1811#endif /* !CONFIG_UTS_NS */
1da177e4
LT
1812
1813static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1814 int *valp,
1815 int write, void *data)
1816{
1817 if (write) {
1818 *valp = *negp ? -*lvalp : *lvalp;
1819 } else {
1820 int val = *valp;
1821 if (val < 0) {
1822 *negp = -1;
1823 *lvalp = (unsigned long)-val;
1824 } else {
1825 *negp = 0;
1826 *lvalp = (unsigned long)val;
1827 }
1828 }
1829 return 0;
1830}
1831
fcfbd547
KK
1832static int __do_proc_dointvec(void *tbl_data, ctl_table *table,
1833 int write, struct file *filp, void __user *buffer,
1834 size_t *lenp, loff_t *ppos,
1da177e4
LT
1835 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1836 int write, void *data),
1837 void *data)
1838{
1839#define TMPBUFLEN 21
1840 int *i, vleft, first=1, neg, val;
1841 unsigned long lval;
1842 size_t left, len;
1843
1844 char buf[TMPBUFLEN], *p;
1845 char __user *s = buffer;
1846
fcfbd547 1847 if (!tbl_data || !table->maxlen || !*lenp ||
1da177e4
LT
1848 (*ppos && !write)) {
1849 *lenp = 0;
1850 return 0;
1851 }
1852
fcfbd547 1853 i = (int *) tbl_data;
1da177e4
LT
1854 vleft = table->maxlen / sizeof(*i);
1855 left = *lenp;
1856
1857 if (!conv)
1858 conv = do_proc_dointvec_conv;
1859
1860 for (; left && vleft--; i++, first=0) {
1861 if (write) {
1862 while (left) {
1863 char c;
1864 if (get_user(c, s))
1865 return -EFAULT;
1866 if (!isspace(c))
1867 break;
1868 left--;
1869 s++;
1870 }
1871 if (!left)
1872 break;
1873 neg = 0;
1874 len = left;
1875 if (len > sizeof(buf) - 1)
1876 len = sizeof(buf) - 1;
1877 if (copy_from_user(buf, s, len))
1878 return -EFAULT;
1879 buf[len] = 0;
1880 p = buf;
1881 if (*p == '-' && left > 1) {
1882 neg = 1;
1883 left--, p++;
1884 }
1885 if (*p < '0' || *p > '9')
1886 break;
1887
1888 lval = simple_strtoul(p, &p, 0);
1889
1890 len = p-buf;
1891 if ((len < left) && *p && !isspace(*p))
1892 break;
1893 if (neg)
1894 val = -val;
1895 s += len;
1896 left -= len;
1897
1898 if (conv(&neg, &lval, i, 1, data))
1899 break;
1900 } else {
1901 p = buf;
1902 if (!first)
1903 *p++ = '\t';
1904
1905 if (conv(&neg, &lval, i, 0, data))
1906 break;
1907
1908 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1909 len = strlen(buf);
1910 if (len > left)
1911 len = left;
1912 if(copy_to_user(s, buf, len))
1913 return -EFAULT;
1914 left -= len;
1915 s += len;
1916 }
1917 }
1918
1919 if (!write && !first && left) {
1920 if(put_user('\n', s))
1921 return -EFAULT;
1922 left--, s++;
1923 }
1924 if (write) {
1925 while (left) {
1926 char c;
1927 if (get_user(c, s++))
1928 return -EFAULT;
1929 if (!isspace(c))
1930 break;
1931 left--;
1932 }
1933 }
1934 if (write && first)
1935 return -EINVAL;
1936 *lenp -= left;
1937 *ppos += *lenp;
1938 return 0;
1939#undef TMPBUFLEN
1940}
1941
fcfbd547
KK
1942static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1943 void __user *buffer, size_t *lenp, loff_t *ppos,
1944 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1945 int write, void *data),
1946 void *data)
1947{
1948 return __do_proc_dointvec(table->data, table, write, filp,
1949 buffer, lenp, ppos, conv, data);
1950}
1951
1da177e4
LT
1952/**
1953 * proc_dointvec - read a vector of integers
1954 * @table: the sysctl table
1955 * @write: %TRUE if this is a write to the sysctl file
1956 * @filp: the file structure
1957 * @buffer: the user buffer
1958 * @lenp: the size of the user buffer
1959 * @ppos: file position
1960 *
1961 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1962 * values from/to the user buffer, treated as an ASCII string.
1963 *
1964 * Returns 0 on success.
1965 */
1966int proc_dointvec(ctl_table *table, int write, struct file *filp,
1967 void __user *buffer, size_t *lenp, loff_t *ppos)
1968{
1969 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1970 NULL,NULL);
1971}
1972
1973#define OP_SET 0
1974#define OP_AND 1
1975#define OP_OR 2
1976#define OP_MAX 3
1977#define OP_MIN 4
1978
1979static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1980 int *valp,
1981 int write, void *data)
1982{
1983 int op = *(int *)data;
1984 if (write) {
1985 int val = *negp ? -*lvalp : *lvalp;
1986 switch(op) {
1987 case OP_SET: *valp = val; break;
1988 case OP_AND: *valp &= val; break;
1989 case OP_OR: *valp |= val; break;
1990 case OP_MAX: if(*valp < val)
1991 *valp = val;
1992 break;
1993 case OP_MIN: if(*valp > val)
1994 *valp = val;
1995 break;
1996 }
1997 } else {
1998 int val = *valp;
1999 if (val < 0) {
2000 *negp = -1;
2001 *lvalp = (unsigned long)-val;
2002 } else {
2003 *negp = 0;
2004 *lvalp = (unsigned long)val;
2005 }
2006 }
2007 return 0;
2008}
2009
2010/*
2011 * init may raise the set.
2012 */
2013
2014int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2015 void __user *buffer, size_t *lenp, loff_t *ppos)
2016{
2017 int op;
2018
2019 if (!capable(CAP_SYS_MODULE)) {
2020 return -EPERM;
2021 }
2022
f400e198 2023 op = is_init(current) ? OP_SET : OP_AND;
1da177e4
LT
2024 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2025 do_proc_dointvec_bset_conv,&op);
2026}
2027
2028struct do_proc_dointvec_minmax_conv_param {
2029 int *min;
2030 int *max;
2031};
2032
2033static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2034 int *valp,
2035 int write, void *data)
2036{
2037 struct do_proc_dointvec_minmax_conv_param *param = data;
2038 if (write) {
2039 int val = *negp ? -*lvalp : *lvalp;
2040 if ((param->min && *param->min > val) ||
2041 (param->max && *param->max < val))
2042 return -EINVAL;
2043 *valp = val;
2044 } else {
2045 int val = *valp;
2046 if (val < 0) {
2047 *negp = -1;
2048 *lvalp = (unsigned long)-val;
2049 } else {
2050 *negp = 0;
2051 *lvalp = (unsigned long)val;
2052 }
2053 }
2054 return 0;
2055}
2056
2057/**
2058 * proc_dointvec_minmax - read a vector of integers with min/max values
2059 * @table: the sysctl table
2060 * @write: %TRUE if this is a write to the sysctl file
2061 * @filp: the file structure
2062 * @buffer: the user buffer
2063 * @lenp: the size of the user buffer
2064 * @ppos: file position
2065 *
2066 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2067 * values from/to the user buffer, treated as an ASCII string.
2068 *
2069 * This routine will ensure the values are within the range specified by
2070 * table->extra1 (min) and table->extra2 (max).
2071 *
2072 * Returns 0 on success.
2073 */
2074int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2075 void __user *buffer, size_t *lenp, loff_t *ppos)
2076{
2077 struct do_proc_dointvec_minmax_conv_param param = {
2078 .min = (int *) table->extra1,
2079 .max = (int *) table->extra2,
2080 };
2081 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2082 do_proc_dointvec_minmax_conv, &param);
2083}
2084
fcfbd547 2085static int __do_proc_doulongvec_minmax(void *data, ctl_table *table, int write,
1da177e4
LT
2086 struct file *filp,
2087 void __user *buffer,
2088 size_t *lenp, loff_t *ppos,
2089 unsigned long convmul,
2090 unsigned long convdiv)
2091{
2092#define TMPBUFLEN 21
2093 unsigned long *i, *min, *max, val;
2094 int vleft, first=1, neg;
2095 size_t len, left;
2096 char buf[TMPBUFLEN], *p;
2097 char __user *s = buffer;
2098
fcfbd547 2099 if (!data || !table->maxlen || !*lenp ||
1da177e4
LT
2100 (*ppos && !write)) {
2101 *lenp = 0;
2102 return 0;
2103 }
2104
fcfbd547 2105 i = (unsigned long *) data;
1da177e4
LT
2106 min = (unsigned long *) table->extra1;
2107 max = (unsigned long *) table->extra2;
2108 vleft = table->maxlen / sizeof(unsigned long);
2109 left = *lenp;
2110
2111 for (; left && vleft--; i++, min++, max++, first=0) {
2112 if (write) {
2113 while (left) {
2114 char c;
2115 if (get_user(c, s))
2116 return -EFAULT;
2117 if (!isspace(c))
2118 break;
2119 left--;
2120 s++;
2121 }
2122 if (!left)
2123 break;
2124 neg = 0;
2125 len = left;
2126 if (len > TMPBUFLEN-1)
2127 len = TMPBUFLEN-1;
2128 if (copy_from_user(buf, s, len))
2129 return -EFAULT;
2130 buf[len] = 0;
2131 p = buf;
2132 if (*p == '-' && left > 1) {
2133 neg = 1;
2134 left--, p++;
2135 }
2136 if (*p < '0' || *p > '9')
2137 break;
2138 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2139 len = p-buf;
2140 if ((len < left) && *p && !isspace(*p))
2141 break;
2142 if (neg)
2143 val = -val;
2144 s += len;
2145 left -= len;
2146
2147 if(neg)
2148 continue;
2149 if ((min && val < *min) || (max && val > *max))
2150 continue;
2151 *i = val;
2152 } else {
2153 p = buf;
2154 if (!first)
2155 *p++ = '\t';
2156 sprintf(p, "%lu", convdiv * (*i) / convmul);
2157 len = strlen(buf);
2158 if (len > left)
2159 len = left;
2160 if(copy_to_user(s, buf, len))
2161 return -EFAULT;
2162 left -= len;
2163 s += len;
2164 }
2165 }
2166
2167 if (!write && !first && left) {
2168 if(put_user('\n', s))
2169 return -EFAULT;
2170 left--, s++;
2171 }
2172 if (write) {
2173 while (left) {
2174 char c;
2175 if (get_user(c, s++))
2176 return -EFAULT;
2177 if (!isspace(c))
2178 break;
2179 left--;
2180 }
2181 }
2182 if (write && first)
2183 return -EINVAL;
2184 *lenp -= left;
2185 *ppos += *lenp;
2186 return 0;
2187#undef TMPBUFLEN
2188}
2189
fcfbd547
KK
2190static int do_proc_doulongvec_minmax(ctl_table *table, int write,
2191 struct file *filp,
2192 void __user *buffer,
2193 size_t *lenp, loff_t *ppos,
2194 unsigned long convmul,
2195 unsigned long convdiv)
2196{
2197 return __do_proc_doulongvec_minmax(table->data, table, write,
2198 filp, buffer, lenp, ppos, convmul, convdiv);
2199}
2200
1da177e4
LT
2201/**
2202 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2203 * @table: the sysctl table
2204 * @write: %TRUE if this is a write to the sysctl file
2205 * @filp: the file structure
2206 * @buffer: the user buffer
2207 * @lenp: the size of the user buffer
2208 * @ppos: file position
2209 *
2210 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2211 * values from/to the user buffer, treated as an ASCII string.
2212 *
2213 * This routine will ensure the values are within the range specified by
2214 * table->extra1 (min) and table->extra2 (max).
2215 *
2216 * Returns 0 on success.
2217 */
2218int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2219 void __user *buffer, size_t *lenp, loff_t *ppos)
2220{
2221 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2222}
2223
2224/**
2225 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2226 * @table: the sysctl table
2227 * @write: %TRUE if this is a write to the sysctl file
2228 * @filp: the file structure
2229 * @buffer: the user buffer
2230 * @lenp: the size of the user buffer
2231 * @ppos: file position
2232 *
2233 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2234 * values from/to the user buffer, treated as an ASCII string. The values
2235 * are treated as milliseconds, and converted to jiffies when they are stored.
2236 *
2237 * This routine will ensure the values are within the range specified by
2238 * table->extra1 (min) and table->extra2 (max).
2239 *
2240 * Returns 0 on success.
2241 */
2242int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2243 struct file *filp,
2244 void __user *buffer,
2245 size_t *lenp, loff_t *ppos)
2246{
2247 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2248 lenp, ppos, HZ, 1000l);
2249}
2250
2251
2252static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2253 int *valp,
2254 int write, void *data)
2255{
2256 if (write) {
cba9f33d
BS
2257 if (*lvalp > LONG_MAX / HZ)
2258 return 1;
1da177e4
LT
2259 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2260 } else {
2261 int val = *valp;
2262 unsigned long lval;
2263 if (val < 0) {
2264 *negp = -1;
2265 lval = (unsigned long)-val;
2266 } else {
2267 *negp = 0;
2268 lval = (unsigned long)val;
2269 }
2270 *lvalp = lval / HZ;
2271 }
2272 return 0;
2273}
2274
2275static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2276 int *valp,
2277 int write, void *data)
2278{
2279 if (write) {
cba9f33d
BS
2280 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2281 return 1;
1da177e4
LT
2282 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2283 } else {
2284 int val = *valp;
2285 unsigned long lval;
2286 if (val < 0) {
2287 *negp = -1;
2288 lval = (unsigned long)-val;
2289 } else {
2290 *negp = 0;
2291 lval = (unsigned long)val;
2292 }
2293 *lvalp = jiffies_to_clock_t(lval);
2294 }
2295 return 0;
2296}
2297
2298static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2299 int *valp,
2300 int write, void *data)
2301{
2302 if (write) {
2303 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2304 } else {
2305 int val = *valp;
2306 unsigned long lval;
2307 if (val < 0) {
2308 *negp = -1;
2309 lval = (unsigned long)-val;
2310 } else {
2311 *negp = 0;
2312 lval = (unsigned long)val;
2313 }
2314 *lvalp = jiffies_to_msecs(lval);
2315 }
2316 return 0;
2317}
2318
2319/**
2320 * proc_dointvec_jiffies - read a vector of integers as seconds
2321 * @table: the sysctl table
2322 * @write: %TRUE if this is a write to the sysctl file
2323 * @filp: the file structure
2324 * @buffer: the user buffer
2325 * @lenp: the size of the user buffer
2326 * @ppos: file position
2327 *
2328 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2329 * values from/to the user buffer, treated as an ASCII string.
2330 * The values read are assumed to be in seconds, and are converted into
2331 * jiffies.
2332 *
2333 * Returns 0 on success.
2334 */
2335int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2336 void __user *buffer, size_t *lenp, loff_t *ppos)
2337{
2338 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2339 do_proc_dointvec_jiffies_conv,NULL);
2340}
2341
2342/**
2343 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2344 * @table: the sysctl table
2345 * @write: %TRUE if this is a write to the sysctl file
2346 * @filp: the file structure
2347 * @buffer: the user buffer
2348 * @lenp: the size of the user buffer
1e5d5331 2349 * @ppos: pointer to the file position
1da177e4
LT
2350 *
2351 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2352 * values from/to the user buffer, treated as an ASCII string.
2353 * The values read are assumed to be in 1/USER_HZ seconds, and
2354 * are converted into jiffies.
2355 *
2356 * Returns 0 on success.
2357 */
2358int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2359 void __user *buffer, size_t *lenp, loff_t *ppos)
2360{
2361 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2362 do_proc_dointvec_userhz_jiffies_conv,NULL);
2363}
2364
2365/**
2366 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2367 * @table: the sysctl table
2368 * @write: %TRUE if this is a write to the sysctl file
2369 * @filp: the file structure
2370 * @buffer: the user buffer
2371 * @lenp: the size of the user buffer
67be2dd1
MW
2372 * @ppos: file position
2373 * @ppos: the current position in the file
1da177e4
LT
2374 *
2375 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2376 * values from/to the user buffer, treated as an ASCII string.
2377 * The values read are assumed to be in 1/1000 seconds, and
2378 * are converted into jiffies.
2379 *
2380 * Returns 0 on success.
2381 */
2382int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2383 void __user *buffer, size_t *lenp, loff_t *ppos)
2384{
2385 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2386 do_proc_dointvec_ms_jiffies_conv, NULL);
2387}
2388
fcfbd547
KK
2389#ifdef CONFIG_SYSVIPC
2390static int proc_do_ipc_string(ctl_table *table, int write, struct file *filp,
2391 void __user *buffer, size_t *lenp, loff_t *ppos)
2392{
2393 void *data;
2394 struct ipc_namespace *ns;
2395
2396 ns = current->nsproxy->ipc_ns;
2397
2398 switch (table->ctl_name) {
2399 case KERN_SHMMAX:
2400 data = &ns->shm_ctlmax;
2401 goto proc_minmax;
2402 case KERN_SHMALL:
2403 data = &ns->shm_ctlall;
2404 goto proc_minmax;
2405 case KERN_SHMMNI:
2406 data = &ns->shm_ctlmni;
2407 break;
2408 case KERN_MSGMAX:
2409 data = &ns->msg_ctlmax;
2410 break;
2411 case KERN_MSGMNI:
2412 data = &ns->msg_ctlmni;
2413 break;
2414 case KERN_MSGMNB:
2415 data = &ns->msg_ctlmnb;
2416 break;
2417 case KERN_SEM:
2418 data = &ns->sem_ctls;
2419 break;
2420 default:
2421 return -EINVAL;
2422 }
2423
2424 return __do_proc_dointvec(data, table, write, filp, buffer,
2425 lenp, ppos, NULL, NULL);
2426proc_minmax:
2427 return __do_proc_doulongvec_minmax(data, table, write, filp, buffer,
2428 lenp, ppos, 1l, 1l);
2429}
2430#endif
2431
9ec52099
CLG
2432static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
2433 void __user *buffer, size_t *lenp, loff_t *ppos)
2434{
2435 struct pid *new_pid;
2436 pid_t tmp;
2437 int r;
2438
2439 tmp = pid_nr(cad_pid);
2440
2441 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2442 lenp, ppos, NULL, NULL);
2443 if (r || !write)
2444 return r;
2445
2446 new_pid = find_get_pid(tmp);
2447 if (!new_pid)
2448 return -ESRCH;
2449
2450 put_pid(xchg(&cad_pid, new_pid));
2451 return 0;
2452}
2453
1da177e4
LT
2454#else /* CONFIG_PROC_FS */
2455
2456int proc_dostring(ctl_table *table, int write, struct file *filp,
2457 void __user *buffer, size_t *lenp, loff_t *ppos)
2458{
2459 return -ENOSYS;
2460}
2461
8218c74c
SH
2462static int proc_do_uts_string(ctl_table *table, int write, struct file *filp,
2463 void __user *buffer, size_t *lenp, loff_t *ppos)
1da177e4
LT
2464{
2465 return -ENOSYS;
2466}
2467
8218c74c
SH
2468#ifdef CONFIG_SYSVIPC
2469static int proc_do_ipc_string(ctl_table *table, int write, struct file *filp,
2470 void __user *buffer, size_t *lenp, loff_t *ppos)
2471{
2472 return -ENOSYS;
2473}
2474#endif
2475
1da177e4
LT
2476int proc_dointvec(ctl_table *table, int write, struct file *filp,
2477 void __user *buffer, size_t *lenp, loff_t *ppos)
2478{
2479 return -ENOSYS;
2480}
2481
2482int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2483 void __user *buffer, size_t *lenp, loff_t *ppos)
2484{
2485 return -ENOSYS;
2486}
2487
2488int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2489 void __user *buffer, size_t *lenp, loff_t *ppos)
2490{
2491 return -ENOSYS;
2492}
2493
2494int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2495 void __user *buffer, size_t *lenp, loff_t *ppos)
2496{
2497 return -ENOSYS;
2498}
2499
2500int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2501 void __user *buffer, size_t *lenp, loff_t *ppos)
2502{
2503 return -ENOSYS;
2504}
2505
2506int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2507 void __user *buffer, size_t *lenp, loff_t *ppos)
2508{
2509 return -ENOSYS;
2510}
2511
2512int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2513 void __user *buffer, size_t *lenp, loff_t *ppos)
2514{
2515 return -ENOSYS;
2516}
2517
2518int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2519 struct file *filp,
2520 void __user *buffer,
2521 size_t *lenp, loff_t *ppos)
2522{
2523 return -ENOSYS;
2524}
2525
2526
2527#endif /* CONFIG_PROC_FS */
2528
2529
b89a8171 2530#ifdef CONFIG_SYSCTL_SYSCALL
1da177e4
LT
2531/*
2532 * General sysctl support routines
2533 */
2534
2535/* The generic string strategy routine: */
2536int sysctl_string(ctl_table *table, int __user *name, int nlen,
2537 void __user *oldval, size_t __user *oldlenp,
2538 void __user *newval, size_t newlen, void **context)
2539{
1da177e4
LT
2540 if (!table->data || !table->maxlen)
2541 return -ENOTDIR;
2542
2543 if (oldval && oldlenp) {
de9e007d
LT
2544 size_t bufsize;
2545 if (get_user(bufsize, oldlenp))
1da177e4 2546 return -EFAULT;
de9e007d
LT
2547 if (bufsize) {
2548 size_t len = strlen(table->data), copied;
2549
2550 /* This shouldn't trigger for a well-formed sysctl */
2551 if (len > table->maxlen)
1da177e4 2552 len = table->maxlen;
de9e007d
LT
2553
2554 /* Copy up to a max of bufsize-1 bytes of the string */
2555 copied = (len >= bufsize) ? bufsize - 1 : len;
2556
2557 if (copy_to_user(oldval, table->data, copied) ||
2558 put_user(0, (char __user *)(oldval + copied)))
1da177e4 2559 return -EFAULT;
de9e007d 2560 if (put_user(len, oldlenp))
1da177e4
LT
2561 return -EFAULT;
2562 }
2563 }
2564 if (newval && newlen) {
de9e007d 2565 size_t len = newlen;
1da177e4
LT
2566 if (len > table->maxlen)
2567 len = table->maxlen;
2568 if(copy_from_user(table->data, newval, len))
2569 return -EFAULT;
2570 if (len == table->maxlen)
2571 len--;
2572 ((char *) table->data)[len] = 0;
2573 }
82c9df82 2574 return 1;
1da177e4
LT
2575}
2576
2577/*
2578 * This function makes sure that all of the integers in the vector
2579 * are between the minimum and maximum values given in the arrays
2580 * table->extra1 and table->extra2, respectively.
2581 */
2582int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2583 void __user *oldval, size_t __user *oldlenp,
2584 void __user *newval, size_t newlen, void **context)
2585{
2586
2587 if (newval && newlen) {
2588 int __user *vec = (int __user *) newval;
2589 int *min = (int *) table->extra1;
2590 int *max = (int *) table->extra2;
2591 size_t length;
2592 int i;
2593
2594 if (newlen % sizeof(int) != 0)
2595 return -EINVAL;
2596
2597 if (!table->extra1 && !table->extra2)
2598 return 0;
2599
2600 if (newlen > table->maxlen)
2601 newlen = table->maxlen;
2602 length = newlen / sizeof(int);
2603
2604 for (i = 0; i < length; i++) {
2605 int value;
2606 if (get_user(value, vec + i))
2607 return -EFAULT;
2608 if (min && value < min[i])
2609 return -EINVAL;
2610 if (max && value > max[i])
2611 return -EINVAL;
2612 }
2613 }
2614 return 0;
2615}
2616
2617/* Strategy function to convert jiffies to seconds */
2618int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2619 void __user *oldval, size_t __user *oldlenp,
2620 void __user *newval, size_t newlen, void **context)
2621{
2622 if (oldval) {
2623 size_t olen;
2624 if (oldlenp) {
2625 if (get_user(olen, oldlenp))
2626 return -EFAULT;
2627 if (olen!=sizeof(int))
2628 return -EINVAL;
2629 }
2630 if (put_user(*(int *)(table->data)/HZ, (int __user *)oldval) ||
2631 (oldlenp && put_user(sizeof(int),oldlenp)))
2632 return -EFAULT;
2633 }
2634 if (newval && newlen) {
2635 int new;
2636 if (newlen != sizeof(int))
2637 return -EINVAL;
2638 if (get_user(new, (int __user *)newval))
2639 return -EFAULT;
2640 *(int *)(table->data) = new*HZ;
2641 }
2642 return 1;
2643}
2644
2645/* Strategy function to convert jiffies to seconds */
2646int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2647 void __user *oldval, size_t __user *oldlenp,
2648 void __user *newval, size_t newlen, void **context)
2649{
2650 if (oldval) {
2651 size_t olen;
2652 if (oldlenp) {
2653 if (get_user(olen, oldlenp))
2654 return -EFAULT;
2655 if (olen!=sizeof(int))
2656 return -EINVAL;
2657 }
2658 if (put_user(jiffies_to_msecs(*(int *)(table->data)), (int __user *)oldval) ||
2659 (oldlenp && put_user(sizeof(int),oldlenp)))
2660 return -EFAULT;
2661 }
2662 if (newval && newlen) {
2663 int new;
2664 if (newlen != sizeof(int))
2665 return -EINVAL;
2666 if (get_user(new, (int __user *)newval))
2667 return -EFAULT;
2668 *(int *)(table->data) = msecs_to_jiffies(new);
2669 }
2670 return 1;
2671}
2672
b89a8171 2673#else /* CONFIG_SYSCTL_SYSCALL */
1da177e4
LT
2674
2675
2676asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2677{
b89a8171
EB
2678 static int msg_count;
2679
2680 if (msg_count < 5) {
2681 msg_count++;
2682 printk(KERN_INFO
2683 "warning: process `%s' used the removed sysctl "
2684 "system call\n", current->comm);
2685 }
1da177e4
LT
2686 return -ENOSYS;
2687}
2688
2689int sysctl_string(ctl_table *table, int __user *name, int nlen,
2690 void __user *oldval, size_t __user *oldlenp,
2691 void __user *newval, size_t newlen, void **context)
2692{
2693 return -ENOSYS;
2694}
2695
2696int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2697 void __user *oldval, size_t __user *oldlenp,
2698 void __user *newval, size_t newlen, void **context)
2699{
2700 return -ENOSYS;
2701}
2702
2703int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2704 void __user *oldval, size_t __user *oldlenp,
2705 void __user *newval, size_t newlen, void **context)
2706{
2707 return -ENOSYS;
2708}
2709
2710int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2711 void __user *oldval, size_t __user *oldlenp,
2712 void __user *newval, size_t newlen, void **context)
2713{
2714 return -ENOSYS;
2715}
2716
b89a8171 2717#endif /* CONFIG_SYSCTL_SYSCALL */
1da177e4
LT
2718
2719/*
2720 * No sense putting this after each symbol definition, twice,
2721 * exception granted :-)
2722 */
2723EXPORT_SYMBOL(proc_dointvec);
2724EXPORT_SYMBOL(proc_dointvec_jiffies);
2725EXPORT_SYMBOL(proc_dointvec_minmax);
2726EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2727EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2728EXPORT_SYMBOL(proc_dostring);
2729EXPORT_SYMBOL(proc_doulongvec_minmax);
2730EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2731EXPORT_SYMBOL(register_sysctl_table);
2732EXPORT_SYMBOL(sysctl_intvec);
2733EXPORT_SYMBOL(sysctl_jiffies);
2734EXPORT_SYMBOL(sysctl_ms_jiffies);
2735EXPORT_SYMBOL(sysctl_string);
2736EXPORT_SYMBOL(unregister_sysctl_table);