]> bbs.cooldavid.org Git - net-next-2.6.git/blame - kernel/sysctl.c
git://bbs.cooldavid.org / net-next-2.6.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
acpi-thermal: fix mod_timer() interval
[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>
0741f4d2 57#include <asm/stacktrace.h>
29cbc78b
AK		 58#endif
59
1da177e4
LT		 60#if defined(CONFIG_SYSCTL)
61
62/* External variables not in a header file. */
63extern int C_A_D;
64extern int sysctl_overcommit_memory;
65extern int sysctl_overcommit_ratio;
fadd8fbd 66extern int sysctl_panic_on_oom;
1da177e4 67extern int max_threads;
1da177e4 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
1da177e4
LT		 93
94#ifdef __sparc__
95extern char reboot_command [];
96extern int stop_a_enabled;
97extern int scons_pwroff;
98#endif
99
100#ifdef __hppa__
101extern int pwrsw_enabled;
102extern int unaligned_enabled;
103#endif
104
347a8dc3 105#ifdef CONFIG_S390
1da177e4
LT		 106#ifdef CONFIG_MATHEMU
107extern int sysctl_ieee_emulation_warnings;
108#endif
109extern int sysctl_userprocess_debug;
951f22d5 110extern int spin_retry;
1da177e4
LT		 111#endif
112
113extern int sysctl_hz_timer;
114
115#ifdef CONFIG_BSD_PROCESS_ACCT
116extern int acct_parm[];
117#endif
118
d2b176ed
JS		 119#ifdef CONFIG_IA64
120extern int no_unaligned_warning;
121#endif
122
23f78d4a
IM		 123#ifdef CONFIG_RT_MUTEXES
124extern int max_lock_depth;
125#endif
126
b89a8171
EB		 127#ifdef CONFIG_SYSCTL_SYSCALL
128static int parse_table(int __user *, int, void __user *, size_t __user *,
1f29bcd7 129 void __user *, size_t, ctl_table *);
b89a8171
EB		 130#endif
131
6b49a257 132
d6f8ff73 133#ifdef CONFIG_PROC_SYSCTL
9ec52099
CLG		 134static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
135 void __user *buffer, size_t *lenp, loff_t *ppos);
34f5a398
TT		 136static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
137 void __user *buffer, size_t *lenp, loff_t *ppos);
d6f8ff73 138#endif
9ec52099 139
1da177e4
LT		 140static ctl_table root_table[];
141static struct ctl_table_header root_table_header =
142 { root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
143
144static ctl_table kern_table[];
145static ctl_table vm_table[];
1da177e4
LT		 146static ctl_table fs_table[];
147static ctl_table debug_table[];
148static ctl_table dev_table[];
149extern ctl_table random_table[];
150#ifdef CONFIG_UNIX98_PTYS
151extern ctl_table pty_table[];
152#endif
2d9048e2 153#ifdef CONFIG_INOTIFY_USER
0399cb08
RL		 154extern ctl_table inotify_table[];
155#endif
1da177e4
LT		 156
157#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
158int sysctl_legacy_va_layout;
159#endif
160
9bc9a6bd 161
1da177e4
LT		 162/* The default sysctl tables: */
163
164static ctl_table root_table[] = {
165 {
166 .ctl_name = CTL_KERN,
167 .procname = "kernel",
168 .mode = 0555,
169 .child = kern_table,
170 },
171 {
172 .ctl_name = CTL_VM,
173 .procname = "vm",
174 .mode = 0555,
175 .child = vm_table,
176 },
177#ifdef CONFIG_NET
178 {
179 .ctl_name = CTL_NET,
180 .procname = "net",
181 .mode = 0555,
182 .child = net_table,
183 },
184#endif
1da177e4
LT		 185 {
186 .ctl_name = CTL_FS,
187 .procname = "fs",
188 .mode = 0555,
189 .child = fs_table,
190 },
191 {
192 .ctl_name = CTL_DEBUG,
193 .procname = "debug",
194 .mode = 0555,
195 .child = debug_table,
196 },
197 {
198 .ctl_name = CTL_DEV,
199 .procname = "dev",
200 .mode = 0555,
201 .child = dev_table,
202 },
0eeca283 203
1da177e4
LT		 204 { .ctl_name = 0 }
205};
206
207static ctl_table kern_table[] = {
1da177e4
LT		 208 {
209 .ctl_name = KERN_PANIC,
210 .procname = "panic",
211 .data = &panic_timeout,
212 .maxlen = sizeof(int),
213 .mode = 0644,
214 .proc_handler = &proc_dointvec,
215 },
216 {
217 .ctl_name = KERN_CORE_USES_PID,
218 .procname = "core_uses_pid",
219 .data = &core_uses_pid,
220 .maxlen = sizeof(int),
221 .mode = 0644,
222 .proc_handler = &proc_dointvec,
223 },
224 {
225 .ctl_name = KERN_CORE_PATTERN,
226 .procname = "core_pattern",
227 .data = core_pattern,
d025c9db 228 .maxlen = 128,
1da177e4
LT		 229 .mode = 0644,
230 .proc_handler = &proc_dostring,
231 .strategy = &sysctl_string,
232 },
34f5a398 233#ifdef CONFIG_PROC_SYSCTL
1da177e4
LT		 234 {
235 .ctl_name = KERN_TAINTED,
236 .procname = "tainted",
237 .data = &tainted,
238 .maxlen = sizeof(int),
34f5a398
TT		 239 .mode = 0644,
240 .proc_handler = &proc_dointvec_taint,
1da177e4 241 },
34f5a398 242#endif
1da177e4
LT		 243 {
244 .ctl_name = KERN_CAP_BSET,
245 .procname = "cap-bound",
246 .data = &cap_bset,
247 .maxlen = sizeof(kernel_cap_t),
248 .mode = 0600,
249 .proc_handler = &proc_dointvec_bset,
250 },
251#ifdef CONFIG_BLK_DEV_INITRD
252 {
253 .ctl_name = KERN_REALROOTDEV,
254 .procname = "real-root-dev",
255 .data = &real_root_dev,
256 .maxlen = sizeof(int),
257 .mode = 0644,
258 .proc_handler = &proc_dointvec,
259 },
260#endif
261#ifdef __sparc__
262 {
263 .ctl_name = KERN_SPARC_REBOOT,
264 .procname = "reboot-cmd",
265 .data = reboot_command,
266 .maxlen = 256,
267 .mode = 0644,
268 .proc_handler = &proc_dostring,
269 .strategy = &sysctl_string,
270 },
271 {
272 .ctl_name = KERN_SPARC_STOP_A,
273 .procname = "stop-a",
274 .data = &stop_a_enabled,
275 .maxlen = sizeof (int),
276 .mode = 0644,
277 .proc_handler = &proc_dointvec,
278 },
279 {
280 .ctl_name = KERN_SPARC_SCONS_PWROFF,
281 .procname = "scons-poweroff",
282 .data = &scons_pwroff,
283 .maxlen = sizeof (int),
284 .mode = 0644,
285 .proc_handler = &proc_dointvec,
286 },
287#endif
288#ifdef __hppa__
289 {
290 .ctl_name = KERN_HPPA_PWRSW,
291 .procname = "soft-power",
292 .data = &pwrsw_enabled,
293 .maxlen = sizeof (int),
294 .mode = 0644,
295 .proc_handler = &proc_dointvec,
296 },
297 {
298 .ctl_name = KERN_HPPA_UNALIGNED,
299 .procname = "unaligned-trap",
300 .data = &unaligned_enabled,
301 .maxlen = sizeof (int),
302 .mode = 0644,
303 .proc_handler = &proc_dointvec,
304 },
305#endif
306 {
307 .ctl_name = KERN_CTLALTDEL,
308 .procname = "ctrl-alt-del",
309 .data = &C_A_D,
310 .maxlen = sizeof(int),
311 .mode = 0644,
312 .proc_handler = &proc_dointvec,
313 },
314 {
315 .ctl_name = KERN_PRINTK,
316 .procname = "printk",
317 .data = &console_loglevel,
318 .maxlen = 4*sizeof(int),
319 .mode = 0644,
320 .proc_handler = &proc_dointvec,
321 },
322#ifdef CONFIG_KMOD
323 {
324 .ctl_name = KERN_MODPROBE,
325 .procname = "modprobe",
326 .data = &modprobe_path,
327 .maxlen = KMOD_PATH_LEN,
328 .mode = 0644,
329 .proc_handler = &proc_dostring,
330 .strategy = &sysctl_string,
331 },
332#endif
57ae2508 333#if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
1da177e4
LT		 334 {
335 .ctl_name = KERN_HOTPLUG,
336 .procname = "hotplug",
312c004d
KS		 337 .data = &uevent_helper,
338 .maxlen = UEVENT_HELPER_PATH_LEN,
1da177e4
LT		 339 .mode = 0644,
340 .proc_handler = &proc_dostring,
341 .strategy = &sysctl_string,
342 },
343#endif
344#ifdef CONFIG_CHR_DEV_SG
345 {
346 .ctl_name = KERN_SG_BIG_BUFF,
347 .procname = "sg-big-buff",
348 .data = &sg_big_buff,
349 .maxlen = sizeof (int),
350 .mode = 0444,
351 .proc_handler = &proc_dointvec,
352 },
353#endif
354#ifdef CONFIG_BSD_PROCESS_ACCT
355 {
356 .ctl_name = KERN_ACCT,
357 .procname = "acct",
358 .data = &acct_parm,
359 .maxlen = 3*sizeof(int),
360 .mode = 0644,
361 .proc_handler = &proc_dointvec,
362 },
363#endif
1da177e4
LT		 364#ifdef CONFIG_MAGIC_SYSRQ
365 {
366 .ctl_name = KERN_SYSRQ,
367 .procname = "sysrq",
5d6f647f 368 .data = &__sysrq_enabled,
1da177e4
LT		 369 .maxlen = sizeof (int),
370 .mode = 0644,
371 .proc_handler = &proc_dointvec,
372 },
373#endif
d6f8ff73 374#ifdef CONFIG_PROC_SYSCTL
1da177e4
LT		 375 {
376 .ctl_name = KERN_CADPID,
377 .procname = "cad_pid",
9ec52099 378 .data = NULL,
1da177e4
LT		 379 .maxlen = sizeof (int),
380 .mode = 0600,
9ec52099 381 .proc_handler = &proc_do_cad_pid,
1da177e4 382 },
d6f8ff73 383#endif
1da177e4
LT		 384 {
385 .ctl_name = KERN_MAX_THREADS,
386 .procname = "threads-max",
387 .data = &max_threads,
388 .maxlen = sizeof(int),
389 .mode = 0644,
390 .proc_handler = &proc_dointvec,
391 },
392 {
393 .ctl_name = KERN_RANDOM,
394 .procname = "random",
395 .mode = 0555,
396 .child = random_table,
397 },
398#ifdef CONFIG_UNIX98_PTYS
399 {
400 .ctl_name = KERN_PTY,
401 .procname = "pty",
402 .mode = 0555,
403 .child = pty_table,
404 },
405#endif
406 {
407 .ctl_name = KERN_OVERFLOWUID,
408 .procname = "overflowuid",
409 .data = &overflowuid,
410 .maxlen = sizeof(int),
411 .mode = 0644,
412 .proc_handler = &proc_dointvec_minmax,
413 .strategy = &sysctl_intvec,
414 .extra1 = &minolduid,
415 .extra2 = &maxolduid,
416 },
417 {
418 .ctl_name = KERN_OVERFLOWGID,
419 .procname = "overflowgid",
420 .data = &overflowgid,
421 .maxlen = sizeof(int),
422 .mode = 0644,
423 .proc_handler = &proc_dointvec_minmax,
424 .strategy = &sysctl_intvec,
425 .extra1 = &minolduid,
426 .extra2 = &maxolduid,
427 },
347a8dc3 428#ifdef CONFIG_S390
1da177e4
LT		 429#ifdef CONFIG_MATHEMU
430 {
431 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
432 .procname = "ieee_emulation_warnings",
433 .data = &sysctl_ieee_emulation_warnings,
434 .maxlen = sizeof(int),
435 .mode = 0644,
436 .proc_handler = &proc_dointvec,
437 },
438#endif
439#ifdef CONFIG_NO_IDLE_HZ
440 {
441 .ctl_name = KERN_HZ_TIMER,
442 .procname = "hz_timer",
443 .data = &sysctl_hz_timer,
444 .maxlen = sizeof(int),
445 .mode = 0644,
446 .proc_handler = &proc_dointvec,
447 },
448#endif
449 {
450 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
451 .procname = "userprocess_debug",
452 .data = &sysctl_userprocess_debug,
453 .maxlen = sizeof(int),
454 .mode = 0644,
455 .proc_handler = &proc_dointvec,
456 },
457#endif
458 {
459 .ctl_name = KERN_PIDMAX,
460 .procname = "pid_max",
461 .data = &pid_max,
462 .maxlen = sizeof (int),
463 .mode = 0644,
464 .proc_handler = &proc_dointvec_minmax,
465 .strategy = sysctl_intvec,
466 .extra1 = &pid_max_min,
467 .extra2 = &pid_max_max,
468 },
469 {
470 .ctl_name = KERN_PANIC_ON_OOPS,
471 .procname = "panic_on_oops",
472 .data = &panic_on_oops,
473 .maxlen = sizeof(int),
474 .mode = 0644,
475 .proc_handler = &proc_dointvec,
476 },
477 {
478 .ctl_name = KERN_PRINTK_RATELIMIT,
479 .procname = "printk_ratelimit",
480 .data = &printk_ratelimit_jiffies,
481 .maxlen = sizeof(int),
482 .mode = 0644,
483 .proc_handler = &proc_dointvec_jiffies,
484 .strategy = &sysctl_jiffies,
485 },
486 {
487 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
488 .procname = "printk_ratelimit_burst",
489 .data = &printk_ratelimit_burst,
490 .maxlen = sizeof(int),
491 .mode = 0644,
492 .proc_handler = &proc_dointvec,
493 },
494 {
495 .ctl_name = KERN_NGROUPS_MAX,
496 .procname = "ngroups_max",
497 .data = &ngroups_max,
498 .maxlen = sizeof (int),
499 .mode = 0444,
500 .proc_handler = &proc_dointvec,
501 },
502#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
503 {
504 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
505 .procname = "unknown_nmi_panic",
506 .data = &unknown_nmi_panic,
507 .maxlen = sizeof (int),
508 .mode = 0644,
2fbe7b25 509 .proc_handler = &proc_dointvec,
1da177e4 510 },
407984f1
DZ		 511 {
512 .ctl_name = KERN_NMI_WATCHDOG,
513 .procname = "nmi_watchdog",
514 .data = &nmi_watchdog_enabled,
515 .maxlen = sizeof (int),
516 .mode = 0644,
517 .proc_handler = &proc_nmi_enabled,
1da177e4
LT		 518 },
519#endif
520#if defined(CONFIG_X86)
8da5adda
DZ		 521 {
522 .ctl_name = KERN_PANIC_ON_NMI,
523 .procname = "panic_on_unrecovered_nmi",
524 .data = &panic_on_unrecovered_nmi,
525 .maxlen = sizeof(int),
526 .mode = 0644,
527 .proc_handler = &proc_dointvec,
528 },
1da177e4
LT		 529 {
530 .ctl_name = KERN_BOOTLOADER_TYPE,
531 .procname = "bootloader_type",
532 .data = &bootloader_type,
533 .maxlen = sizeof (int),
534 .mode = 0444,
535 .proc_handler = &proc_dointvec,
536 },
0741f4d2
CE		 537 {
538 .ctl_name = CTL_UNNUMBERED,
539 .procname = "kstack_depth_to_print",
540 .data = &kstack_depth_to_print,
541 .maxlen = sizeof(int),
542 .mode = 0644,
543 .proc_handler = &proc_dointvec,
544 },
1da177e4 545#endif
7a9166e3 546#if defined(CONFIG_MMU)
1da177e4
LT		 547 {
548 .ctl_name = KERN_RANDOMIZE,
549 .procname = "randomize_va_space",
550 .data = &randomize_va_space,
551 .maxlen = sizeof(int),
552 .mode = 0644,
553 .proc_handler = &proc_dointvec,
554 },
7a9166e3 555#endif
0152fb37 556#if defined(CONFIG_S390) && defined(CONFIG_SMP)
951f22d5
MS		 557 {
558 .ctl_name = KERN_SPIN_RETRY,
559 .procname = "spin_retry",
560 .data = &spin_retry,
561 .maxlen = sizeof (int),
562 .mode = 0644,
563 .proc_handler = &proc_dointvec,
564 },
c255d844
PM		 565#endif
566#ifdef CONFIG_ACPI_SLEEP
567 {
568 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
569 .procname = "acpi_video_flags",
570 .data = &acpi_video_flags,
571 .maxlen = sizeof (unsigned long),
572 .mode = 0644,
7f99f06f 573 .proc_handler = &proc_doulongvec_minmax,
c255d844 574 },
d2b176ed
JS		 575#endif
576#ifdef CONFIG_IA64
577 {
578 .ctl_name = KERN_IA64_UNALIGNED,
579 .procname = "ignore-unaligned-usertrap",
580 .data = &no_unaligned_warning,
581 .maxlen = sizeof (int),
582 .mode = 0644,
583 .proc_handler = &proc_dointvec,
584 },
bebfa101
AK		 585#endif
586#ifdef CONFIG_COMPAT
587 {
588 .ctl_name = KERN_COMPAT_LOG,
589 .procname = "compat-log",
590 .data = &compat_log,
591 .maxlen = sizeof (int),
592 .mode = 0644,
593 .proc_handler = &proc_dointvec,
594 },
951f22d5 595#endif
23f78d4a
IM		 596#ifdef CONFIG_RT_MUTEXES
597 {
598 .ctl_name = KERN_MAX_LOCK_DEPTH,
599 .procname = "max_lock_depth",
600 .data = &max_lock_depth,
601 .maxlen = sizeof(int),
602 .mode = 0644,
603 .proc_handler = &proc_dointvec,
604 },
605#endif
606
1da177e4
LT		 607 { .ctl_name = 0 }
608};
609
610/* Constants for minimum and maximum testing in vm_table.
611 We use these as one-element integer vectors. */
612static int zero;
613static int one_hundred = 100;
614
615
616static ctl_table vm_table[] = {
617 {
618 .ctl_name = VM_OVERCOMMIT_MEMORY,
619 .procname = "overcommit_memory",
620 .data = &sysctl_overcommit_memory,
621 .maxlen = sizeof(sysctl_overcommit_memory),
622 .mode = 0644,
623 .proc_handler = &proc_dointvec,
624 },
fadd8fbd
KH		 625 {
626 .ctl_name = VM_PANIC_ON_OOM,
627 .procname = "panic_on_oom",
628 .data = &sysctl_panic_on_oom,
629 .maxlen = sizeof(sysctl_panic_on_oom),
630 .mode = 0644,
631 .proc_handler = &proc_dointvec,
632 },
1da177e4
LT		 633 {
634 .ctl_name = VM_OVERCOMMIT_RATIO,
635 .procname = "overcommit_ratio",
636 .data = &sysctl_overcommit_ratio,
637 .maxlen = sizeof(sysctl_overcommit_ratio),
638 .mode = 0644,
639 .proc_handler = &proc_dointvec,
640 },
641 {
642 .ctl_name = VM_PAGE_CLUSTER,
643 .procname = "page-cluster",
644 .data = &page_cluster,
645 .maxlen = sizeof(int),
646 .mode = 0644,
647 .proc_handler = &proc_dointvec,
648 },
649 {
650 .ctl_name = VM_DIRTY_BACKGROUND,
651 .procname = "dirty_background_ratio",
652 .data = &dirty_background_ratio,
653 .maxlen = sizeof(dirty_background_ratio),
654 .mode = 0644,
655 .proc_handler = &proc_dointvec_minmax,
656 .strategy = &sysctl_intvec,
657 .extra1 = &zero,
658 .extra2 = &one_hundred,
659 },
660 {
661 .ctl_name = VM_DIRTY_RATIO,
662 .procname = "dirty_ratio",
663 .data = &vm_dirty_ratio,
664 .maxlen = sizeof(vm_dirty_ratio),
665 .mode = 0644,
666 .proc_handler = &proc_dointvec_minmax,
667 .strategy = &sysctl_intvec,
668 .extra1 = &zero,
669 .extra2 = &one_hundred,
670 },
671 {
672 .ctl_name = VM_DIRTY_WB_CS,
673 .procname = "dirty_writeback_centisecs",
f6ef9438
BS		 674 .data = &dirty_writeback_interval,
675 .maxlen = sizeof(dirty_writeback_interval),
1da177e4
LT		 676 .mode = 0644,
677 .proc_handler = &dirty_writeback_centisecs_handler,
678 },
679 {
680 .ctl_name = VM_DIRTY_EXPIRE_CS,
681 .procname = "dirty_expire_centisecs",
f6ef9438
BS		 682 .data = &dirty_expire_interval,
683 .maxlen = sizeof(dirty_expire_interval),
1da177e4 684 .mode = 0644,
f6ef9438 685 .proc_handler = &proc_dointvec_userhz_jiffies,
1da177e4
LT		 686 },
687 {
688 .ctl_name = VM_NR_PDFLUSH_THREADS,
689 .procname = "nr_pdflush_threads",
690 .data = &nr_pdflush_threads,
691 .maxlen = sizeof nr_pdflush_threads,
692 .mode = 0444 /* read-only*/,
693 .proc_handler = &proc_dointvec,
694 },
695 {
696 .ctl_name = VM_SWAPPINESS,
697 .procname = "swappiness",
698 .data = &vm_swappiness,
699 .maxlen = sizeof(vm_swappiness),
700 .mode = 0644,
701 .proc_handler = &proc_dointvec_minmax,
702 .strategy = &sysctl_intvec,
703 .extra1 = &zero,
704 .extra2 = &one_hundred,
705 },
706#ifdef CONFIG_HUGETLB_PAGE
707 {
708 .ctl_name = VM_HUGETLB_PAGES,
709 .procname = "nr_hugepages",
710 .data = &max_huge_pages,
711 .maxlen = sizeof(unsigned long),
712 .mode = 0644,
713 .proc_handler = &hugetlb_sysctl_handler,
714 .extra1 = (void *)&hugetlb_zero,
715 .extra2 = (void *)&hugetlb_infinity,
716 },
717 {
718 .ctl_name = VM_HUGETLB_GROUP,
719 .procname = "hugetlb_shm_group",
720 .data = &sysctl_hugetlb_shm_group,
721 .maxlen = sizeof(gid_t),
722 .mode = 0644,
723 .proc_handler = &proc_dointvec,
724 },
725#endif
726 {
727 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
728 .procname = "lowmem_reserve_ratio",
729 .data = &sysctl_lowmem_reserve_ratio,
730 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
731 .mode = 0644,
732 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
733 .strategy = &sysctl_intvec,
734 },
9d0243bc
AM		 735 {
736 .ctl_name = VM_DROP_PAGECACHE,
737 .procname = "drop_caches",
738 .data = &sysctl_drop_caches,
739 .maxlen = sizeof(int),
740 .mode = 0644,
741 .proc_handler = drop_caches_sysctl_handler,
742 .strategy = &sysctl_intvec,
743 },
1da177e4
LT		 744 {
745 .ctl_name = VM_MIN_FREE_KBYTES,
746 .procname = "min_free_kbytes",
747 .data = &min_free_kbytes,
748 .maxlen = sizeof(min_free_kbytes),
749 .mode = 0644,
750 .proc_handler = &min_free_kbytes_sysctl_handler,
751 .strategy = &sysctl_intvec,
752 .extra1 = &zero,
753 },
8ad4b1fb
RS		 754 {
755 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
756 .procname = "percpu_pagelist_fraction",
757 .data = &percpu_pagelist_fraction,
758 .maxlen = sizeof(percpu_pagelist_fraction),
759 .mode = 0644,
760 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
761 .strategy = &sysctl_intvec,
762 .extra1 = &min_percpu_pagelist_fract,
763 },
1da177e4
LT		 764#ifdef CONFIG_MMU
765 {
766 .ctl_name = VM_MAX_MAP_COUNT,
767 .procname = "max_map_count",
768 .data = &sysctl_max_map_count,
769 .maxlen = sizeof(sysctl_max_map_count),
770 .mode = 0644,
771 .proc_handler = &proc_dointvec
772 },
773#endif
774 {
775 .ctl_name = VM_LAPTOP_MODE,
776 .procname = "laptop_mode",
777 .data = &laptop_mode,
778 .maxlen = sizeof(laptop_mode),
779 .mode = 0644,
ed5b43f1
BS		 780 .proc_handler = &proc_dointvec_jiffies,
781 .strategy = &sysctl_jiffies,
1da177e4
LT		 782 },
783 {
784 .ctl_name = VM_BLOCK_DUMP,
785 .procname = "block_dump",
786 .data = &block_dump,
787 .maxlen = sizeof(block_dump),
788 .mode = 0644,
789 .proc_handler = &proc_dointvec,
790 .strategy = &sysctl_intvec,
791 .extra1 = &zero,
792 },
793 {
794 .ctl_name = VM_VFS_CACHE_PRESSURE,
795 .procname = "vfs_cache_pressure",
796 .data = &sysctl_vfs_cache_pressure,
797 .maxlen = sizeof(sysctl_vfs_cache_pressure),
798 .mode = 0644,
799 .proc_handler = &proc_dointvec,
800 .strategy = &sysctl_intvec,
801 .extra1 = &zero,
802 },
803#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
804 {
805 .ctl_name = VM_LEGACY_VA_LAYOUT,
806 .procname = "legacy_va_layout",
807 .data = &sysctl_legacy_va_layout,
808 .maxlen = sizeof(sysctl_legacy_va_layout),
809 .mode = 0644,
810 .proc_handler = &proc_dointvec,
811 .strategy = &sysctl_intvec,
812 .extra1 = &zero,
813 },
814#endif
1743660b
CL		 815#ifdef CONFIG_NUMA
816 {
817 .ctl_name = VM_ZONE_RECLAIM_MODE,
818 .procname = "zone_reclaim_mode",
819 .data = &zone_reclaim_mode,
820 .maxlen = sizeof(zone_reclaim_mode),
821 .mode = 0644,
822 .proc_handler = &proc_dointvec,
c84db23c
CL		 823 .strategy = &sysctl_intvec,
824 .extra1 = &zero,
1743660b 825 },
9614634f
CL		 826 {
827 .ctl_name = VM_MIN_UNMAPPED,
828 .procname = "min_unmapped_ratio",
829 .data = &sysctl_min_unmapped_ratio,
830 .maxlen = sizeof(sysctl_min_unmapped_ratio),
831 .mode = 0644,
832 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
833 .strategy = &sysctl_intvec,
834 .extra1 = &zero,
835 .extra2 = &one_hundred,
836 },
0ff38490
CL		 837 {
838 .ctl_name = VM_MIN_SLAB,
839 .procname = "min_slab_ratio",
840 .data = &sysctl_min_slab_ratio,
841 .maxlen = sizeof(sysctl_min_slab_ratio),
842 .mode = 0644,
843 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
844 .strategy = &sysctl_intvec,
845 .extra1 = &zero,
846 .extra2 = &one_hundred,
847 },
e6e5494c 848#endif
5c36e657
PM		 849#if defined(CONFIG_X86_32) || \
850 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
e6e5494c
IM		 851 {
852 .ctl_name = VM_VDSO_ENABLED,
853 .procname = "vdso_enabled",
854 .data = &vdso_enabled,
855 .maxlen = sizeof(vdso_enabled),
856 .mode = 0644,
857 .proc_handler = &proc_dointvec,
858 .strategy = &sysctl_intvec,
859 .extra1 = &zero,
860 },
1da177e4
LT		 861#endif
862 { .ctl_name = 0 }
863};
864
2abc26fc
EB		 865#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
866static ctl_table binfmt_misc_table[] = {
867 { .ctl_name = 0 }
868};
869#endif
870
1da177e4
LT		 871static ctl_table fs_table[] = {
872 {
873 .ctl_name = FS_NRINODE,
874 .procname = "inode-nr",
875 .data = &inodes_stat,
876 .maxlen = 2*sizeof(int),
877 .mode = 0444,
878 .proc_handler = &proc_dointvec,
879 },
880 {
881 .ctl_name = FS_STATINODE,
882 .procname = "inode-state",
883 .data = &inodes_stat,
884 .maxlen = 7*sizeof(int),
885 .mode = 0444,
886 .proc_handler = &proc_dointvec,
887 },
888 {
889 .ctl_name = FS_NRFILE,
890 .procname = "file-nr",
891 .data = &files_stat,
892 .maxlen = 3*sizeof(int),
893 .mode = 0444,
529bf6be 894 .proc_handler = &proc_nr_files,
1da177e4
LT		 895 },
896 {
897 .ctl_name = FS_MAXFILE,
898 .procname = "file-max",
899 .data = &files_stat.max_files,
900 .maxlen = sizeof(int),
901 .mode = 0644,
902 .proc_handler = &proc_dointvec,
903 },
904 {
905 .ctl_name = FS_DENTRY,
906 .procname = "dentry-state",
907 .data = &dentry_stat,
908 .maxlen = 6*sizeof(int),
909 .mode = 0444,
910 .proc_handler = &proc_dointvec,
911 },
912 {
913 .ctl_name = FS_OVERFLOWUID,
914 .procname = "overflowuid",
915 .data = &fs_overflowuid,
916 .maxlen = sizeof(int),
917 .mode = 0644,
918 .proc_handler = &proc_dointvec_minmax,
919 .strategy = &sysctl_intvec,
920 .extra1 = &minolduid,
921 .extra2 = &maxolduid,
922 },
923 {
924 .ctl_name = FS_OVERFLOWGID,
925 .procname = "overflowgid",
926 .data = &fs_overflowgid,
927 .maxlen = sizeof(int),
928 .mode = 0644,
929 .proc_handler = &proc_dointvec_minmax,
930 .strategy = &sysctl_intvec,
931 .extra1 = &minolduid,
932 .extra2 = &maxolduid,
933 },
934 {
935 .ctl_name = FS_LEASES,
936 .procname = "leases-enable",
937 .data = &leases_enable,
938 .maxlen = sizeof(int),
939 .mode = 0644,
940 .proc_handler = &proc_dointvec,
941 },
942#ifdef CONFIG_DNOTIFY
943 {
944 .ctl_name = FS_DIR_NOTIFY,
945 .procname = "dir-notify-enable",
946 .data = &dir_notify_enable,
947 .maxlen = sizeof(int),
948 .mode = 0644,
949 .proc_handler = &proc_dointvec,
950 },
951#endif
952#ifdef CONFIG_MMU
953 {
954 .ctl_name = FS_LEASE_TIME,
955 .procname = "lease-break-time",
956 .data = &lease_break_time,
957 .maxlen = sizeof(int),
958 .mode = 0644,
959 .proc_handler = &proc_dointvec,
960 },
961 {
962 .ctl_name = FS_AIO_NR,
963 .procname = "aio-nr",
964 .data = &aio_nr,
965 .maxlen = sizeof(aio_nr),
966 .mode = 0444,
d55b5fda 967 .proc_handler = &proc_doulongvec_minmax,
1da177e4
LT		 968 },
969 {
970 .ctl_name = FS_AIO_MAX_NR,
971 .procname = "aio-max-nr",
972 .data = &aio_max_nr,
973 .maxlen = sizeof(aio_max_nr),
974 .mode = 0644,
d55b5fda 975 .proc_handler = &proc_doulongvec_minmax,
1da177e4 976 },
2d9048e2 977#ifdef CONFIG_INOTIFY_USER
0399cb08
RL		 978 {
979 .ctl_name = FS_INOTIFY,
980 .procname = "inotify",
981 .mode = 0555,
982 .child = inotify_table,
983 },
984#endif
1da177e4 985#endif
d6e71144
AC		 986 {
987 .ctl_name = KERN_SETUID_DUMPABLE,
988 .procname = "suid_dumpable",
989 .data = &suid_dumpable,
990 .maxlen = sizeof(int),
991 .mode = 0644,
992 .proc_handler = &proc_dointvec,
993 },
2abc26fc
EB		 994#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
995 {
996 .ctl_name = CTL_UNNUMBERED,
997 .procname = "binfmt_misc",
998 .mode = 0555,
999 .child = binfmt_misc_table,
1000 },
1001#endif
1da177e4
LT		 1002 { .ctl_name = 0 }
1003};
1004
1005static ctl_table debug_table[] = {
1006 { .ctl_name = 0 }
1007};
1008
1009static ctl_table dev_table[] = {
1010 { .ctl_name = 0 }
0eeca283 1011};
1da177e4 1012
330d57fb
AV		 1013static DEFINE_SPINLOCK(sysctl_lock);
1014
1015/* called under sysctl_lock */
1016static int use_table(struct ctl_table_header *p)
1017{
1018 if (unlikely(p->unregistering))
1019 return 0;
1020 p->used++;
1021 return 1;
1022}
1023
1024/* called under sysctl_lock */
1025static void unuse_table(struct ctl_table_header *p)
1026{
1027 if (!--p->used)
1028 if (unlikely(p->unregistering))
1029 complete(p->unregistering);
1030}
1031
1032/* called under sysctl_lock, will reacquire if has to wait */
1033static void start_unregistering(struct ctl_table_header *p)
1034{
1035 /*
1036 * if p->used is 0, nobody will ever touch that entry again;
1037 * we'll eliminate all paths to it before dropping sysctl_lock
1038 */
1039 if (unlikely(p->used)) {
1040 struct completion wait;
1041 init_completion(&wait);
1042 p->unregistering = &wait;
1043 spin_unlock(&sysctl_lock);
1044 wait_for_completion(&wait);
1045 spin_lock(&sysctl_lock);
1046 }
1047 /*
1048 * do not remove from the list until nobody holds it; walking the
1049 * list in do_sysctl() relies on that.
1050 */
1051 list_del_init(&p->ctl_entry);
1052}
1053
805b5d5e
EB		 1054void sysctl_head_finish(struct ctl_table_header *head)
1055{
1056 if (!head)
1057 return;
1058 spin_lock(&sysctl_lock);
1059 unuse_table(head);
1060 spin_unlock(&sysctl_lock);
1061}
1062
1063struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1064{
1065 struct ctl_table_header *head;
1066 struct list_head *tmp;
1067 spin_lock(&sysctl_lock);
1068 if (prev) {
1069 tmp = &prev->ctl_entry;
1070 unuse_table(prev);
1071 goto next;
1072 }
1073 tmp = &root_table_header.ctl_entry;
1074 for (;;) {
1075 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1076
1077 if (!use_table(head))
1078 goto next;
1079 spin_unlock(&sysctl_lock);
1080 return head;
1081 next:
1082 tmp = tmp->next;
1083 if (tmp == &root_table_header.ctl_entry)
1084 break;
1085 }
1086 spin_unlock(&sysctl_lock);
1087 return NULL;
1088}
1089
b89a8171 1090#ifdef CONFIG_SYSCTL_SYSCALL
1da177e4
LT		 1091int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1092 void __user *newval, size_t newlen)
1093{
805b5d5e 1094 struct ctl_table_header *head;
330d57fb 1095 int error = -ENOTDIR;
1da177e4
LT		 1096
1097 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1098 return -ENOTDIR;
1099 if (oldval) {
1100 int old_len;
1101 if (!oldlenp || get_user(old_len, oldlenp))
1102 return -EFAULT;
1103 }
330d57fb 1104
805b5d5e
EB		 1105 for (head = sysctl_head_next(NULL); head;
1106 head = sysctl_head_next(head)) {
330d57fb 1107 error = parse_table(name, nlen, oldval, oldlenp,
1f29bcd7 1108 newval, newlen, head->ctl_table);
805b5d5e
EB		 1109 if (error != -ENOTDIR) {
1110 sysctl_head_finish(head);
330d57fb 1111 break;
805b5d5e
EB		 1112 }
1113 }
330d57fb 1114 return error;
1da177e4
LT		 1115}
1116
1117asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1118{
1119 struct __sysctl_args tmp;
1120 int error;
1121
1122 if (copy_from_user(&tmp, args, sizeof(tmp)))
1123 return -EFAULT;
1124
1125 lock_kernel();
1126 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1127 tmp.newval, tmp.newlen);
1128 unlock_kernel();
1129 return error;
1130}
b89a8171 1131#endif /* CONFIG_SYSCTL_SYSCALL */
1da177e4
LT		 1132
1133/*
1ff007eb 1134 * sysctl_perm does NOT grant the superuser all rights automatically, because
1da177e4
LT		 1135 * some sysctl variables are readonly even to root.
1136 */
1137
1138static int test_perm(int mode, int op)
1139{
1140 if (!current->euid)
1141 mode >>= 6;
1142 else if (in_egroup_p(0))
1143 mode >>= 3;
1144 if ((mode & op & 0007) == op)
1145 return 0;
1146 return -EACCES;
1147}
1148
1ff007eb 1149int sysctl_perm(ctl_table *table, int op)
1da177e4
LT		 1150{
1151 int error;
1152 error = security_sysctl(table, op);
1153 if (error)
1154 return error;
1155 return test_perm(table->mode, op);
1156}
1157
b89a8171 1158#ifdef CONFIG_SYSCTL_SYSCALL
1da177e4
LT		 1159static int parse_table(int __user *name, int nlen,
1160 void __user *oldval, size_t __user *oldlenp,
1161 void __user *newval, size_t newlen,
1f29bcd7 1162 ctl_table *table)
1da177e4
LT		 1163{
1164 int n;
1165repeat:
1166 if (!nlen)
1167 return -ENOTDIR;
1168 if (get_user(n, name))
1169 return -EFAULT;
d99f160a
EB		 1170 for ( ; table->ctl_name || table->procname; table++) {
1171 if (!table->ctl_name)
1172 continue;
6703ddfc 1173 if (n == table->ctl_name) {
1da177e4
LT		 1174 int error;
1175 if (table->child) {
1ff007eb 1176 if (sysctl_perm(table, 001))
1da177e4 1177 return -EPERM;
1da177e4
LT		 1178 name++;
1179 nlen--;
1180 table = table->child;
1181 goto repeat;
1182 }
1183 error = do_sysctl_strategy(table, name, nlen,
1184 oldval, oldlenp,
1f29bcd7 1185 newval, newlen);
1da177e4
LT		 1186 return error;
1187 }
1188 }
1189 return -ENOTDIR;
1190}
1191
1192/* Perform the actual read/write of a sysctl table entry. */
1193int do_sysctl_strategy (ctl_table *table,
1194 int __user *name, int nlen,
1195 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 1196 void __user *newval, size_t newlen)
1da177e4
LT		 1197{
1198 int op = 0, rc;
1199 size_t len;
1200
1201 if (oldval)
1202 op |= 004;
1203 if (newval)
1204 op |= 002;
1ff007eb 1205 if (sysctl_perm(table, op))
1da177e4
LT		 1206 return -EPERM;
1207
1208 if (table->strategy) {
1209 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1f29bcd7 1210 newval, newlen);
1da177e4
LT		 1211 if (rc < 0)
1212 return rc;
1213 if (rc > 0)
1214 return 0;
1215 }
1216
1217 /* If there is no strategy routine, or if the strategy returns
1218 * zero, proceed with automatic r/w */
1219 if (table->data && table->maxlen) {
1220 if (oldval && oldlenp) {
1221 if (get_user(len, oldlenp))
1222 return -EFAULT;
1223 if (len) {
1224 if (len > table->maxlen)
1225 len = table->maxlen;
1226 if(copy_to_user(oldval, table->data, len))
1227 return -EFAULT;
1228 if(put_user(len, oldlenp))
1229 return -EFAULT;
1230 }
1231 }
1232 if (newval && newlen) {
1233 len = newlen;
1234 if (len > table->maxlen)
1235 len = table->maxlen;
1236 if(copy_from_user(table->data, newval, len))
1237 return -EFAULT;
1238 }
1239 }
1240 return 0;
1241}
b89a8171 1242#endif /* CONFIG_SYSCTL_SYSCALL */
1da177e4 1243
d912b0cc
EB		 1244static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1245{
1246 for (; table->ctl_name || table->procname; table++) {
1247 table->parent = parent;
1248 if (table->child)
1249 sysctl_set_parent(table, table->child);
1250 }
1251}
1252
1253static __init int sysctl_init(void)
1254{
1255 sysctl_set_parent(NULL, root_table);
1256 return 0;
1257}
1258
1259core_initcall(sysctl_init);
1260
1da177e4
LT		 1261/**
1262 * register_sysctl_table - register a sysctl hierarchy
1263 * @table: the top-level table structure
1da177e4
LT		 1264 *
1265 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1266 * array. An entry with a ctl_name of 0 terminates the table.
1267 *
1268 * The members of the &ctl_table structure are used as follows:
1269 *
1270 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1271 * must be unique within that level of sysctl
1272 *
1273 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1274 * enter a sysctl file
1275 *
1276 * data - a pointer to data for use by proc_handler
1277 *
1278 * maxlen - the maximum size in bytes of the data
1279 *
1280 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1281 *
1282 * child - a pointer to the child sysctl table if this entry is a directory, or
1283 * %NULL.
1284 *
1285 * proc_handler - the text handler routine (described below)
1286 *
1287 * strategy - the strategy routine (described below)
1288 *
1289 * de - for internal use by the sysctl routines
1290 *
1291 * extra1, extra2 - extra pointers usable by the proc handler routines
1292 *
1293 * Leaf nodes in the sysctl tree will be represented by a single file
1294 * under /proc; non-leaf nodes will be represented by directories.
1295 *
1296 * sysctl(2) can automatically manage read and write requests through
1297 * the sysctl table. The data and maxlen fields of the ctl_table
1298 * struct enable minimal validation of the values being written to be
1299 * performed, and the mode field allows minimal authentication.
1300 *
1301 * More sophisticated management can be enabled by the provision of a
1302 * strategy routine with the table entry. This will be called before
1303 * any automatic read or write of the data is performed.
1304 *
1305 * The strategy routine may return
1306 *
1307 * < 0 - Error occurred (error is passed to user process)
1308 *
1309 * 0 - OK - proceed with automatic read or write.
1310 *
1311 * > 0 - OK - read or write has been done by the strategy routine, so
1312 * return immediately.
1313 *
1314 * There must be a proc_handler routine for any terminal nodes
1315 * mirrored under /proc/sys (non-terminals are handled by a built-in
1316 * directory handler). Several default handlers are available to
1317 * cover common cases -
1318 *
1319 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1320 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1321 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1322 *
1323 * It is the handler's job to read the input buffer from user memory
1324 * and process it. The handler should return 0 on success.
1325 *
1326 * This routine returns %NULL on a failure to register, and a pointer
1327 * to the table header on success.
1328 */
0b4d4147 1329struct ctl_table_header *register_sysctl_table(ctl_table * table)
1da177e4
LT		 1330{
1331 struct ctl_table_header *tmp;
1332 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1333 if (!tmp)
1334 return NULL;
1335 tmp->ctl_table = table;
1336 INIT_LIST_HEAD(&tmp->ctl_entry);
330d57fb
AV		 1337 tmp->used = 0;
1338 tmp->unregistering = NULL;
d912b0cc 1339 sysctl_set_parent(NULL, table);
330d57fb 1340 spin_lock(&sysctl_lock);
0b4d4147 1341 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
330d57fb 1342 spin_unlock(&sysctl_lock);
1da177e4
LT		 1343 return tmp;
1344}
1345
1346/**
1347 * unregister_sysctl_table - unregister a sysctl table hierarchy
1348 * @header: the header returned from register_sysctl_table
1349 *
1350 * Unregisters the sysctl table and all children. proc entries may not
1351 * actually be removed until they are no longer used by anyone.
1352 */
1353void unregister_sysctl_table(struct ctl_table_header * header)
1354{
330d57fb
AV		 1355 might_sleep();
1356 spin_lock(&sysctl_lock);
1357 start_unregistering(header);
330d57fb 1358 spin_unlock(&sysctl_lock);
1da177e4
LT		 1359 kfree(header);
1360}
1361
b89a8171 1362#else /* !CONFIG_SYSCTL */
93a6fefe 1363struct ctl_table_header *register_sysctl_table(ctl_table * table)
b89a8171
EB		 1364{
1365 return NULL;
1366}
1367
1368void unregister_sysctl_table(struct ctl_table_header * table)
1369{
1370}
1371
1372#endif /* CONFIG_SYSCTL */
1373
1da177e4
LT		 1374/*
1375 * /proc/sys support
1376 */
1377
b89a8171 1378#ifdef CONFIG_PROC_SYSCTL
1da177e4 1379
b1ba4ddd
AB		 1380static int _proc_do_string(void* data, int maxlen, int write,
1381 struct file *filp, void __user *buffer,
1382 size_t *lenp, loff_t *ppos)
1da177e4
LT		 1383{
1384 size_t len;
1385 char __user *p;
1386 char c;
8d060877
ON		 1387
1388 if (!data || !maxlen || !*lenp) {
1da177e4
LT		 1389 *lenp = 0;
1390 return 0;
1391 }
8d060877 1392
1da177e4
LT		 1393 if (write) {
1394 len = 0;
1395 p = buffer;
1396 while (len < *lenp) {
1397 if (get_user(c, p++))
1398 return -EFAULT;
1399 if (c == 0 || c == '\n')
1400 break;
1401 len++;
1402 }
f5dd3d6f
SV		 1403 if (len >= maxlen)
1404 len = maxlen-1;
1405 if(copy_from_user(data, buffer, len))
1da177e4 1406 return -EFAULT;
f5dd3d6f 1407 ((char *) data)[len] = 0;
1da177e4
LT		 1408 *ppos += *lenp;
1409 } else {
f5dd3d6f
SV		 1410 len = strlen(data);
1411 if (len > maxlen)
1412 len = maxlen;
8d060877
ON		 1413
1414 if (*ppos > len) {
1415 *lenp = 0;
1416 return 0;
1417 }
1418
1419 data += *ppos;
1420 len -= *ppos;
1421
1da177e4
LT		 1422 if (len > *lenp)
1423 len = *lenp;
1424 if (len)
f5dd3d6f 1425 if(copy_to_user(buffer, data, len))
1da177e4
LT		 1426 return -EFAULT;
1427 if (len < *lenp) {
1428 if(put_user('\n', ((char __user *) buffer) + len))
1429 return -EFAULT;
1430 len++;
1431 }
1432 *lenp = len;
1433 *ppos += len;
1434 }
1435 return 0;
1436}
1437
f5dd3d6f
SV		 1438/**
1439 * proc_dostring - read a string sysctl
1440 * @table: the sysctl table
1441 * @write: %TRUE if this is a write to the sysctl file
1442 * @filp: the file structure
1443 * @buffer: the user buffer
1444 * @lenp: the size of the user buffer
1445 * @ppos: file position
1446 *
1447 * Reads/writes a string from/to the user buffer. If the kernel
1448 * buffer provided is not large enough to hold the string, the
1449 * string is truncated. The copied string is %NULL-terminated.
1450 * If the string is being read by the user process, it is copied
1451 * and a newline '\n' is added. It is truncated if the buffer is
1452 * not large enough.
1453 *
1454 * Returns 0 on success.
1455 */
1456int proc_dostring(ctl_table *table, int write, struct file *filp,
1457 void __user *buffer, size_t *lenp, loff_t *ppos)
1458{
1459 return _proc_do_string(table->data, table->maxlen, write, filp,
1460 buffer, lenp, ppos);
1461}
1462
1da177e4
LT		 1463
1464static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1465 int *valp,
1466 int write, void *data)
1467{
1468 if (write) {
1469 *valp = *negp ? -*lvalp : *lvalp;
1470 } else {
1471 int val = *valp;
1472 if (val < 0) {
1473 *negp = -1;
1474 *lvalp = (unsigned long)-val;
1475 } else {
1476 *negp = 0;
1477 *lvalp = (unsigned long)val;
1478 }
1479 }
1480 return 0;
1481}
1482
fcfbd547
KK		 1483static int __do_proc_dointvec(void *tbl_data, ctl_table *table,
1484 int write, struct file *filp, void __user *buffer,
1485 size_t *lenp, loff_t *ppos,
1da177e4
LT		 1486 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1487 int write, void *data),
1488 void *data)
1489{
1490#define TMPBUFLEN 21
1491 int *i, vleft, first=1, neg, val;
1492 unsigned long lval;
1493 size_t left, len;
1494
1495 char buf[TMPBUFLEN], *p;
1496 char __user *s = buffer;
1497
fcfbd547 1498 if (!tbl_data || !table->maxlen || !*lenp ||
1da177e4
LT		 1499 (*ppos && !write)) {
1500 *lenp = 0;
1501 return 0;
1502 }
1503
fcfbd547 1504 i = (int *) tbl_data;
1da177e4
LT		 1505 vleft = table->maxlen / sizeof(*i);
1506 left = *lenp;
1507
1508 if (!conv)
1509 conv = do_proc_dointvec_conv;
1510
1511 for (; left && vleft--; i++, first=0) {
1512 if (write) {
1513 while (left) {
1514 char c;
1515 if (get_user(c, s))
1516 return -EFAULT;
1517 if (!isspace(c))
1518 break;
1519 left--;
1520 s++;
1521 }
1522 if (!left)
1523 break;
1524 neg = 0;
1525 len = left;
1526 if (len > sizeof(buf) - 1)
1527 len = sizeof(buf) - 1;
1528 if (copy_from_user(buf, s, len))
1529 return -EFAULT;
1530 buf[len] = 0;
1531 p = buf;
1532 if (*p == '-' && left > 1) {
1533 neg = 1;
bd9b0bac 1534 p++;
1da177e4
LT		 1535 }
1536 if (*p < '0' || *p > '9')
1537 break;
1538
1539 lval = simple_strtoul(p, &p, 0);
1540
1541 len = p-buf;
1542 if ((len < left) && *p && !isspace(*p))
1543 break;
1544 if (neg)
1545 val = -val;
1546 s += len;
1547 left -= len;
1548
1549 if (conv(&neg, &lval, i, 1, data))
1550 break;
1551 } else {
1552 p = buf;
1553 if (!first)
1554 *p++ = '\t';
1555
1556 if (conv(&neg, &lval, i, 0, data))
1557 break;
1558
1559 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1560 len = strlen(buf);
1561 if (len > left)
1562 len = left;
1563 if(copy_to_user(s, buf, len))
1564 return -EFAULT;
1565 left -= len;
1566 s += len;
1567 }
1568 }
1569
1570 if (!write && !first && left) {
1571 if(put_user('\n', s))
1572 return -EFAULT;
1573 left--, s++;
1574 }
1575 if (write) {
1576 while (left) {
1577 char c;
1578 if (get_user(c, s++))
1579 return -EFAULT;
1580 if (!isspace(c))
1581 break;
1582 left--;
1583 }
1584 }
1585 if (write && first)
1586 return -EINVAL;
1587 *lenp -= left;
1588 *ppos += *lenp;
1589 return 0;
1590#undef TMPBUFLEN
1591}
1592
fcfbd547
KK		 1593static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1594 void __user *buffer, size_t *lenp, loff_t *ppos,
1595 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1596 int write, void *data),
1597 void *data)
1598{
1599 return __do_proc_dointvec(table->data, table, write, filp,
1600 buffer, lenp, ppos, conv, data);
1601}
1602
1da177e4
LT		 1603/**
1604 * proc_dointvec - read a vector of integers
1605 * @table: the sysctl table
1606 * @write: %TRUE if this is a write to the sysctl file
1607 * @filp: the file structure
1608 * @buffer: the user buffer
1609 * @lenp: the size of the user buffer
1610 * @ppos: file position
1611 *
1612 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1613 * values from/to the user buffer, treated as an ASCII string.
1614 *
1615 * Returns 0 on success.
1616 */
1617int proc_dointvec(ctl_table *table, int write, struct file *filp,
1618 void __user *buffer, size_t *lenp, loff_t *ppos)
1619{
1620 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1621 NULL,NULL);
1622}
1623
1624#define OP_SET 0
1625#define OP_AND 1
34f5a398 1626#define OP_OR 2
1da177e4
LT		 1627
1628static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1629 int *valp,
1630 int write, void *data)
1631{
1632 int op = *(int *)data;
1633 if (write) {
1634 int val = *negp ? -*lvalp : *lvalp;
1635 switch(op) {
1636 case OP_SET: *valp = val; break;
1637 case OP_AND: *valp &= val; break;
34f5a398 1638 case OP_OR: *valp |= val; break;
1da177e4
LT		 1639 }
1640 } else {
1641 int val = *valp;
1642 if (val < 0) {
1643 *negp = -1;
1644 *lvalp = (unsigned long)-val;
1645 } else {
1646 *negp = 0;
1647 *lvalp = (unsigned long)val;
1648 }
1649 }
1650 return 0;
1651}
1652
1653/*
1654 * init may raise the set.
1655 */
1656
1657int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1658 void __user *buffer, size_t *lenp, loff_t *ppos)
1659{
1660 int op;
1661
6ff1b442 1662 if (write && !capable(CAP_SYS_MODULE)) {
1da177e4
LT		 1663 return -EPERM;
1664 }
1665
f400e198 1666 op = is_init(current) ? OP_SET : OP_AND;
1da177e4
LT		 1667 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1668 do_proc_dointvec_bset_conv,&op);
1669}
1670
34f5a398
TT		 1671/*
1672 * Taint values can only be increased
1673 */
1674static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
1675 void __user *buffer, size_t *lenp, loff_t *ppos)
1676{
1677 int op;
1678
1679 if (!capable(CAP_SYS_ADMIN))
1680 return -EPERM;
1681
1682 op = OP_OR;
1683 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1684 do_proc_dointvec_bset_conv,&op);
1685}
1686
1da177e4
LT		 1687struct do_proc_dointvec_minmax_conv_param {
1688 int *min;
1689 int *max;
1690};
1691
1692static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1693 int *valp,
1694 int write, void *data)
1695{
1696 struct do_proc_dointvec_minmax_conv_param *param = data;
1697 if (write) {
1698 int val = *negp ? -*lvalp : *lvalp;
1699 if ((param->min && *param->min > val) ||
1700 (param->max && *param->max < val))
1701 return -EINVAL;
1702 *valp = val;
1703 } else {
1704 int val = *valp;
1705 if (val < 0) {
1706 *negp = -1;
1707 *lvalp = (unsigned long)-val;
1708 } else {
1709 *negp = 0;
1710 *lvalp = (unsigned long)val;
1711 }
1712 }
1713 return 0;
1714}
1715
1716/**
1717 * proc_dointvec_minmax - read a vector of integers with min/max values
1718 * @table: the sysctl table
1719 * @write: %TRUE if this is a write to the sysctl file
1720 * @filp: the file structure
1721 * @buffer: the user buffer
1722 * @lenp: the size of the user buffer
1723 * @ppos: file position
1724 *
1725 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1726 * values from/to the user buffer, treated as an ASCII string.
1727 *
1728 * This routine will ensure the values are within the range specified by
1729 * table->extra1 (min) and table->extra2 (max).
1730 *
1731 * Returns 0 on success.
1732 */
1733int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1734 void __user *buffer, size_t *lenp, loff_t *ppos)
1735{
1736 struct do_proc_dointvec_minmax_conv_param param = {
1737 .min = (int *) table->extra1,
1738 .max = (int *) table->extra2,
1739 };
1740 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1741 do_proc_dointvec_minmax_conv, &param);
1742}
1743
fcfbd547 1744static int __do_proc_doulongvec_minmax(void *data, ctl_table *table, int write,
1da177e4
LT		 1745 struct file *filp,
1746 void __user *buffer,
1747 size_t *lenp, loff_t *ppos,
1748 unsigned long convmul,
1749 unsigned long convdiv)
1750{
1751#define TMPBUFLEN 21
1752 unsigned long *i, *min, *max, val;
1753 int vleft, first=1, neg;
1754 size_t len, left;
1755 char buf[TMPBUFLEN], *p;
1756 char __user *s = buffer;
1757
fcfbd547 1758 if (!data || !table->maxlen || !*lenp ||
1da177e4
LT		 1759 (*ppos && !write)) {
1760 *lenp = 0;
1761 return 0;
1762 }
1763
fcfbd547 1764 i = (unsigned long *) data;
1da177e4
LT		 1765 min = (unsigned long *) table->extra1;
1766 max = (unsigned long *) table->extra2;
1767 vleft = table->maxlen / sizeof(unsigned long);
1768 left = *lenp;
1769
1770 for (; left && vleft--; i++, min++, max++, first=0) {
1771 if (write) {
1772 while (left) {
1773 char c;
1774 if (get_user(c, s))
1775 return -EFAULT;
1776 if (!isspace(c))
1777 break;
1778 left--;
1779 s++;
1780 }
1781 if (!left)
1782 break;
1783 neg = 0;
1784 len = left;
1785 if (len > TMPBUFLEN-1)
1786 len = TMPBUFLEN-1;
1787 if (copy_from_user(buf, s, len))
1788 return -EFAULT;
1789 buf[len] = 0;
1790 p = buf;
1791 if (*p == '-' && left > 1) {
1792 neg = 1;
bd9b0bac 1793 p++;
1da177e4
LT		 1794 }
1795 if (*p < '0' || *p > '9')
1796 break;
1797 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
1798 len = p-buf;
1799 if ((len < left) && *p && !isspace(*p))
1800 break;
1801 if (neg)
1802 val = -val;
1803 s += len;
1804 left -= len;
1805
1806 if(neg)
1807 continue;
1808 if ((min && val < *min) || (max && val > *max))
1809 continue;
1810 *i = val;
1811 } else {
1812 p = buf;
1813 if (!first)
1814 *p++ = '\t';
1815 sprintf(p, "%lu", convdiv * (*i) / convmul);
1816 len = strlen(buf);
1817 if (len > left)
1818 len = left;
1819 if(copy_to_user(s, buf, len))
1820 return -EFAULT;
1821 left -= len;
1822 s += len;
1823 }
1824 }
1825
1826 if (!write && !first && left) {
1827 if(put_user('\n', s))
1828 return -EFAULT;
1829 left--, s++;
1830 }
1831 if (write) {
1832 while (left) {
1833 char c;
1834 if (get_user(c, s++))
1835 return -EFAULT;
1836 if (!isspace(c))
1837 break;
1838 left--;
1839 }
1840 }
1841 if (write && first)
1842 return -EINVAL;
1843 *lenp -= left;
1844 *ppos += *lenp;
1845 return 0;
1846#undef TMPBUFLEN
1847}
1848
fcfbd547
KK		 1849static int do_proc_doulongvec_minmax(ctl_table *table, int write,
1850 struct file *filp,
1851 void __user *buffer,
1852 size_t *lenp, loff_t *ppos,
1853 unsigned long convmul,
1854 unsigned long convdiv)
1855{
1856 return __do_proc_doulongvec_minmax(table->data, table, write,
1857 filp, buffer, lenp, ppos, convmul, convdiv);
1858}
1859
1da177e4
LT		 1860/**
1861 * proc_doulongvec_minmax - read a vector of long integers with min/max values
1862 * @table: the sysctl table
1863 * @write: %TRUE if this is a write to the sysctl file
1864 * @filp: the file structure
1865 * @buffer: the user buffer
1866 * @lenp: the size of the user buffer
1867 * @ppos: file position
1868 *
1869 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1870 * values from/to the user buffer, treated as an ASCII string.
1871 *
1872 * This routine will ensure the values are within the range specified by
1873 * table->extra1 (min) and table->extra2 (max).
1874 *
1875 * Returns 0 on success.
1876 */
1877int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
1878 void __user *buffer, size_t *lenp, loff_t *ppos)
1879{
1880 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
1881}
1882
1883/**
1884 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
1885 * @table: the sysctl table
1886 * @write: %TRUE if this is a write to the sysctl file
1887 * @filp: the file structure
1888 * @buffer: the user buffer
1889 * @lenp: the size of the user buffer
1890 * @ppos: file position
1891 *
1892 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1893 * values from/to the user buffer, treated as an ASCII string. The values
1894 * are treated as milliseconds, and converted to jiffies when they are stored.
1895 *
1896 * This routine will ensure the values are within the range specified by
1897 * table->extra1 (min) and table->extra2 (max).
1898 *
1899 * Returns 0 on success.
1900 */
1901int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
1902 struct file *filp,
1903 void __user *buffer,
1904 size_t *lenp, loff_t *ppos)
1905{
1906 return do_proc_doulongvec_minmax(table, write, filp, buffer,
1907 lenp, ppos, HZ, 1000l);
1908}
1909
1910
1911static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
1912 int *valp,
1913 int write, void *data)
1914{
1915 if (write) {
cba9f33d
BS		 1916 if (*lvalp > LONG_MAX / HZ)
1917 return 1;
1da177e4
LT		 1918 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
1919 } else {
1920 int val = *valp;
1921 unsigned long lval;
1922 if (val < 0) {
1923 *negp = -1;
1924 lval = (unsigned long)-val;
1925 } else {
1926 *negp = 0;
1927 lval = (unsigned long)val;
1928 }
1929 *lvalp = lval / HZ;
1930 }
1931 return 0;
1932}
1933
1934static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
1935 int *valp,
1936 int write, void *data)
1937{
1938 if (write) {
cba9f33d
BS		 1939 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
1940 return 1;
1da177e4
LT		 1941 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
1942 } else {
1943 int val = *valp;
1944 unsigned long lval;
1945 if (val < 0) {
1946 *negp = -1;
1947 lval = (unsigned long)-val;
1948 } else {
1949 *negp = 0;
1950 lval = (unsigned long)val;
1951 }
1952 *lvalp = jiffies_to_clock_t(lval);
1953 }
1954 return 0;
1955}
1956
1957static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
1958 int *valp,
1959 int write, void *data)
1960{
1961 if (write) {
1962 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
1963 } else {
1964 int val = *valp;
1965 unsigned long lval;
1966 if (val < 0) {
1967 *negp = -1;
1968 lval = (unsigned long)-val;
1969 } else {
1970 *negp = 0;
1971 lval = (unsigned long)val;
1972 }
1973 *lvalp = jiffies_to_msecs(lval);
1974 }
1975 return 0;
1976}
1977
1978/**
1979 * proc_dointvec_jiffies - read a vector of integers as seconds
1980 * @table: the sysctl table
1981 * @write: %TRUE if this is a write to the sysctl file
1982 * @filp: the file structure
1983 * @buffer: the user buffer
1984 * @lenp: the size of the user buffer
1985 * @ppos: file position
1986 *
1987 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1988 * values from/to the user buffer, treated as an ASCII string.
1989 * The values read are assumed to be in seconds, and are converted into
1990 * jiffies.
1991 *
1992 * Returns 0 on success.
1993 */
1994int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
1995 void __user *buffer, size_t *lenp, loff_t *ppos)
1996{
1997 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1998 do_proc_dointvec_jiffies_conv,NULL);
1999}
2000
2001/**
2002 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2003 * @table: the sysctl table
2004 * @write: %TRUE if this is a write to the sysctl file
2005 * @filp: the file structure
2006 * @buffer: the user buffer
2007 * @lenp: the size of the user buffer
1e5d5331 2008 * @ppos: pointer to the file position
1da177e4
LT		 2009 *
2010 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2011 * values from/to the user buffer, treated as an ASCII string.
2012 * The values read are assumed to be in 1/USER_HZ seconds, and
2013 * are converted into jiffies.
2014 *
2015 * Returns 0 on success.
2016 */
2017int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2018 void __user *buffer, size_t *lenp, loff_t *ppos)
2019{
2020 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2021 do_proc_dointvec_userhz_jiffies_conv,NULL);
2022}
2023
2024/**
2025 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2026 * @table: the sysctl table
2027 * @write: %TRUE if this is a write to the sysctl file
2028 * @filp: the file structure
2029 * @buffer: the user buffer
2030 * @lenp: the size of the user buffer
67be2dd1
MW		 2031 * @ppos: file position
2032 * @ppos: the current position in the file
1da177e4
LT		 2033 *
2034 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2035 * values from/to the user buffer, treated as an ASCII string.
2036 * The values read are assumed to be in 1/1000 seconds, and
2037 * are converted into jiffies.
2038 *
2039 * Returns 0 on success.
2040 */
2041int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2042 void __user *buffer, size_t *lenp, loff_t *ppos)
2043{
2044 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2045 do_proc_dointvec_ms_jiffies_conv, NULL);
2046}
2047
9ec52099
CLG		 2048static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
2049 void __user *buffer, size_t *lenp, loff_t *ppos)
2050{
2051 struct pid *new_pid;
2052 pid_t tmp;
2053 int r;
2054
2055 tmp = pid_nr(cad_pid);
2056
2057 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2058 lenp, ppos, NULL, NULL);
2059 if (r || !write)
2060 return r;
2061
2062 new_pid = find_get_pid(tmp);
2063 if (!new_pid)
2064 return -ESRCH;
2065
2066 put_pid(xchg(&cad_pid, new_pid));
2067 return 0;
2068}
2069
1da177e4
LT		 2070#else /* CONFIG_PROC_FS */
2071
2072int proc_dostring(ctl_table *table, int write, struct file *filp,
2073 void __user *buffer, size_t *lenp, loff_t *ppos)
2074{
2075 return -ENOSYS;
2076}
2077
1da177e4
LT		 2078int proc_dointvec(ctl_table *table, int write, struct file *filp,
2079 void __user *buffer, size_t *lenp, loff_t *ppos)
2080{
2081 return -ENOSYS;
2082}
2083
2084int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2085 void __user *buffer, size_t *lenp, loff_t *ppos)
2086{
2087 return -ENOSYS;
2088}
2089
2090int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2091 void __user *buffer, size_t *lenp, loff_t *ppos)
2092{
2093 return -ENOSYS;
2094}
2095
2096int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2097 void __user *buffer, size_t *lenp, loff_t *ppos)
2098{
2099 return -ENOSYS;
2100}
2101
2102int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2103 void __user *buffer, size_t *lenp, loff_t *ppos)
2104{
2105 return -ENOSYS;
2106}
2107
2108int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2109 void __user *buffer, size_t *lenp, loff_t *ppos)
2110{
2111 return -ENOSYS;
2112}
2113
2114int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2115 void __user *buffer, size_t *lenp, loff_t *ppos)
2116{
2117 return -ENOSYS;
2118}
2119
2120int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2121 struct file *filp,
2122 void __user *buffer,
2123 size_t *lenp, loff_t *ppos)
2124{
2125 return -ENOSYS;
2126}
2127
2128
2129#endif /* CONFIG_PROC_FS */
2130
2131
b89a8171 2132#ifdef CONFIG_SYSCTL_SYSCALL
1da177e4
LT		 2133/*
2134 * General sysctl support routines
2135 */
2136
2137/* The generic string strategy routine: */
2138int sysctl_string(ctl_table *table, int __user *name, int nlen,
2139 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2140 void __user *newval, size_t newlen)
1da177e4 2141{
1da177e4
LT		 2142 if (!table->data || !table->maxlen)
2143 return -ENOTDIR;
2144
2145 if (oldval && oldlenp) {
de9e007d
LT		 2146 size_t bufsize;
2147 if (get_user(bufsize, oldlenp))
1da177e4 2148 return -EFAULT;
de9e007d
LT		 2149 if (bufsize) {
2150 size_t len = strlen(table->data), copied;
2151
2152 /* This shouldn't trigger for a well-formed sysctl */
2153 if (len > table->maxlen)
1da177e4 2154 len = table->maxlen;
de9e007d
LT		 2155
2156 /* Copy up to a max of bufsize-1 bytes of the string */
2157 copied = (len >= bufsize) ? bufsize - 1 : len;
2158
2159 if (copy_to_user(oldval, table->data, copied) ||
2160 put_user(0, (char __user *)(oldval + copied)))
1da177e4 2161 return -EFAULT;
de9e007d 2162 if (put_user(len, oldlenp))
1da177e4
LT		 2163 return -EFAULT;
2164 }
2165 }
2166 if (newval && newlen) {
de9e007d 2167 size_t len = newlen;
1da177e4
LT		 2168 if (len > table->maxlen)
2169 len = table->maxlen;
2170 if(copy_from_user(table->data, newval, len))
2171 return -EFAULT;
2172 if (len == table->maxlen)
2173 len--;
2174 ((char *) table->data)[len] = 0;
2175 }
82c9df82 2176 return 1;
1da177e4
LT		 2177}
2178
2179/*
2180 * This function makes sure that all of the integers in the vector
2181 * are between the minimum and maximum values given in the arrays
2182 * table->extra1 and table->extra2, respectively.
2183 */
2184int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2185 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2186 void __user *newval, size_t newlen)
1da177e4
LT		 2187{
2188
2189 if (newval && newlen) {
2190 int __user *vec = (int __user *) newval;
2191 int *min = (int *) table->extra1;
2192 int *max = (int *) table->extra2;
2193 size_t length;
2194 int i;
2195
2196 if (newlen % sizeof(int) != 0)
2197 return -EINVAL;
2198
2199 if (!table->extra1 && !table->extra2)
2200 return 0;
2201
2202 if (newlen > table->maxlen)
2203 newlen = table->maxlen;
2204 length = newlen / sizeof(int);
2205
2206 for (i = 0; i < length; i++) {
2207 int value;
2208 if (get_user(value, vec + i))
2209 return -EFAULT;
2210 if (min && value < min[i])
2211 return -EINVAL;
2212 if (max && value > max[i])
2213 return -EINVAL;
2214 }
2215 }
2216 return 0;
2217}
2218
2219/* Strategy function to convert jiffies to seconds */
2220int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2221 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2222 void __user *newval, size_t newlen)
1da177e4 2223{
3ee75ac3 2224 if (oldval && oldlenp) {
1da177e4 2225 size_t olen;
3ee75ac3
AD		 2226
2227 if (get_user(olen, oldlenp))
2228 return -EFAULT;
2229 if (olen) {
2230 int val;
2231
2232 if (olen < sizeof(int))
2233 return -EINVAL;
2234
2235 val = *(int *)(table->data) / HZ;
2236 if (put_user(val, (int __user *)oldval))
2237 return -EFAULT;
2238 if (put_user(sizeof(int), oldlenp))
1da177e4 2239 return -EFAULT;
1da177e4 2240 }
1da177e4
LT		 2241 }
2242 if (newval && newlen) {
2243 int new;
2244 if (newlen != sizeof(int))
2245 return -EINVAL;
2246 if (get_user(new, (int __user *)newval))
2247 return -EFAULT;
2248 *(int *)(table->data) = new*HZ;
2249 }
2250 return 1;
2251}
2252
2253/* Strategy function to convert jiffies to seconds */
2254int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2255 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2256 void __user *newval, size_t newlen)
1da177e4 2257{
3ee75ac3 2258 if (oldval && oldlenp) {
1da177e4 2259 size_t olen;
3ee75ac3
AD		 2260
2261 if (get_user(olen, oldlenp))
2262 return -EFAULT;
2263 if (olen) {
2264 int val;
2265
2266 if (olen < sizeof(int))
2267 return -EINVAL;
2268
2269 val = jiffies_to_msecs(*(int *)(table->data));
2270 if (put_user(val, (int __user *)oldval))
2271 return -EFAULT;
2272 if (put_user(sizeof(int), oldlenp))
1da177e4 2273 return -EFAULT;
1da177e4 2274 }
1da177e4
LT		 2275 }
2276 if (newval && newlen) {
2277 int new;
2278 if (newlen != sizeof(int))
2279 return -EINVAL;
2280 if (get_user(new, (int __user *)newval))
2281 return -EFAULT;
2282 *(int *)(table->data) = msecs_to_jiffies(new);
2283 }
2284 return 1;
2285}
2286
c4b8b769 2287
c4b8b769 2288
b89a8171 2289#else /* CONFIG_SYSCTL_SYSCALL */
1da177e4
LT		 2290
2291
2292asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2293{
b89a8171 2294 static int msg_count;
0e009be8
EB		 2295 struct __sysctl_args tmp;
2296 int name[CTL_MAXNAME];
2297 int i;
2298
2299 /* Read in the sysctl name for better debug message logging */
2300 if (copy_from_user(&tmp, args, sizeof(tmp)))
2301 return -EFAULT;
2302 if (tmp.nlen <= 0 || tmp.nlen >= CTL_MAXNAME)
2303 return -ENOTDIR;
2304 for (i = 0; i < tmp.nlen; i++)
2305 if (get_user(name[i], tmp.name + i))
2306 return -EFAULT;
2307
2308 /* Ignore accesses to kernel.version */
2309 if ((tmp.nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2310 goto out;
b89a8171
EB		 2311
2312 if (msg_count < 5) {
2313 msg_count++;
2314 printk(KERN_INFO
2315 "warning: process `%s' used the removed sysctl "
0e009be8
EB		 2316 "system call with ", current->comm);
2317 for (i = 0; i < tmp.nlen; i++)
2318 printk("%d.", name[i]);
2319 printk("\n");
b89a8171 2320 }
0e009be8 2321out:
1da177e4
LT		 2322 return -ENOSYS;
2323}
2324
2325int sysctl_string(ctl_table *table, int __user *name, int nlen,
2326 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2327 void __user *newval, size_t newlen)
1da177e4
LT		 2328{
2329 return -ENOSYS;
2330}
2331
2332int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2333 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2334 void __user *newval, size_t newlen)
1da177e4
LT		 2335{
2336 return -ENOSYS;
2337}
2338
2339int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2340 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2341 void __user *newval, size_t newlen)
1da177e4
LT		 2342{
2343 return -ENOSYS;
2344}
2345
2346int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2347 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2348 void __user *newval, size_t newlen)
1da177e4
LT		 2349{
2350 return -ENOSYS;
2351}
2352
b89a8171 2353#endif /* CONFIG_SYSCTL_SYSCALL */
1da177e4
LT		 2354
2355/*
2356 * No sense putting this after each symbol definition, twice,
2357 * exception granted :-)
2358 */
2359EXPORT_SYMBOL(proc_dointvec);
2360EXPORT_SYMBOL(proc_dointvec_jiffies);
2361EXPORT_SYMBOL(proc_dointvec_minmax);
2362EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2363EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2364EXPORT_SYMBOL(proc_dostring);
2365EXPORT_SYMBOL(proc_doulongvec_minmax);
2366EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2367EXPORT_SYMBOL(register_sysctl_table);
2368EXPORT_SYMBOL(sysctl_intvec);
2369EXPORT_SYMBOL(sysctl_jiffies);
2370EXPORT_SYMBOL(sysctl_ms_jiffies);
2371EXPORT_SYMBOL(sysctl_string);
2372EXPORT_SYMBOL(unregister_sysctl_table);
Clone of davem's net-next-2.6 tree