serial: convert early_uart to earlycon for 8250
[linux-2.6-block.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;
5096add8 79extern int maps_protect;
77461ab3 80extern int sysctl_stat_interval;
1da177e4 81
1da177e4
LT
82/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
83static int maxolduid = 65535;
84static int minolduid;
8ad4b1fb 85static int min_percpu_pagelist_fract = 8;
1da177e4
LT
86
87static int ngroups_max = NGROUPS_MAX;
88
89#ifdef CONFIG_KMOD
90extern char modprobe_path[];
91#endif
1da177e4
LT
92#ifdef CONFIG_CHR_DEV_SG
93extern int sg_big_buff;
94#endif
1da177e4
LT
95
96#ifdef __sparc__
97extern char reboot_command [];
98extern int stop_a_enabled;
99extern int scons_pwroff;
100#endif
101
102#ifdef __hppa__
103extern int pwrsw_enabled;
104extern int unaligned_enabled;
105#endif
106
347a8dc3 107#ifdef CONFIG_S390
1da177e4
LT
108#ifdef CONFIG_MATHEMU
109extern int sysctl_ieee_emulation_warnings;
110#endif
111extern int sysctl_userprocess_debug;
951f22d5 112extern int spin_retry;
1da177e4
LT
113#endif
114
115extern int sysctl_hz_timer;
116
117#ifdef CONFIG_BSD_PROCESS_ACCT
118extern int acct_parm[];
119#endif
120
d2b176ed
JS
121#ifdef CONFIG_IA64
122extern int no_unaligned_warning;
123#endif
124
23f78d4a
IM
125#ifdef CONFIG_RT_MUTEXES
126extern int max_lock_depth;
127#endif
128
b89a8171
EB
129#ifdef CONFIG_SYSCTL_SYSCALL
130static int parse_table(int __user *, int, void __user *, size_t __user *,
1f29bcd7 131 void __user *, size_t, ctl_table *);
b89a8171
EB
132#endif
133
6b49a257 134
d6f8ff73 135#ifdef CONFIG_PROC_SYSCTL
9ec52099
CLG
136static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
137 void __user *buffer, size_t *lenp, loff_t *ppos);
34f5a398
TT
138static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
139 void __user *buffer, size_t *lenp, loff_t *ppos);
d6f8ff73 140#endif
9ec52099 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
9bc9a6bd 163
1da177e4
LT
164/* The default sysctl tables: */
165
166static ctl_table root_table[] = {
167 {
168 .ctl_name = CTL_KERN,
169 .procname = "kernel",
170 .mode = 0555,
171 .child = kern_table,
172 },
173 {
174 .ctl_name = CTL_VM,
175 .procname = "vm",
176 .mode = 0555,
177 .child = vm_table,
178 },
179#ifdef CONFIG_NET
180 {
181 .ctl_name = CTL_NET,
182 .procname = "net",
183 .mode = 0555,
184 .child = net_table,
185 },
186#endif
1da177e4
LT
187 {
188 .ctl_name = CTL_FS,
189 .procname = "fs",
190 .mode = 0555,
191 .child = fs_table,
192 },
193 {
194 .ctl_name = CTL_DEBUG,
195 .procname = "debug",
196 .mode = 0555,
197 .child = debug_table,
198 },
199 {
200 .ctl_name = CTL_DEV,
201 .procname = "dev",
202 .mode = 0555,
203 .child = dev_table,
204 },
0eeca283 205
1da177e4
LT
206 { .ctl_name = 0 }
207};
208
77e54a1f
IM
209#ifdef CONFIG_SCHED_DEBUG
210static unsigned long min_sched_granularity_ns = 100000; /* 100 usecs */
211static unsigned long max_sched_granularity_ns = 1000000000; /* 1 second */
212static unsigned long min_wakeup_granularity_ns; /* 0 usecs */
213static unsigned long max_wakeup_granularity_ns = 1000000000; /* 1 second */
214#endif
215
1da177e4 216static ctl_table kern_table[] = {
77e54a1f
IM
217#ifdef CONFIG_SCHED_DEBUG
218 {
219 .ctl_name = CTL_UNNUMBERED,
220 .procname = "sched_granularity_ns",
221 .data = &sysctl_sched_granularity,
222 .maxlen = sizeof(unsigned int),
223 .mode = 0644,
224 .proc_handler = &proc_dointvec_minmax,
225 .strategy = &sysctl_intvec,
226 .extra1 = &min_sched_granularity_ns,
227 .extra2 = &max_sched_granularity_ns,
228 },
229 {
230 .ctl_name = CTL_UNNUMBERED,
231 .procname = "sched_wakeup_granularity_ns",
232 .data = &sysctl_sched_wakeup_granularity,
233 .maxlen = sizeof(unsigned int),
234 .mode = 0644,
235 .proc_handler = &proc_dointvec_minmax,
236 .strategy = &sysctl_intvec,
237 .extra1 = &min_wakeup_granularity_ns,
238 .extra2 = &max_wakeup_granularity_ns,
239 },
240 {
241 .ctl_name = CTL_UNNUMBERED,
242 .procname = "sched_batch_wakeup_granularity_ns",
243 .data = &sysctl_sched_batch_wakeup_granularity,
244 .maxlen = sizeof(unsigned int),
245 .mode = 0644,
246 .proc_handler = &proc_dointvec_minmax,
247 .strategy = &sysctl_intvec,
248 .extra1 = &min_wakeup_granularity_ns,
249 .extra2 = &max_wakeup_granularity_ns,
250 },
251 {
252 .ctl_name = CTL_UNNUMBERED,
253 .procname = "sched_stat_granularity_ns",
254 .data = &sysctl_sched_stat_granularity,
255 .maxlen = sizeof(unsigned int),
256 .mode = 0644,
257 .proc_handler = &proc_dointvec_minmax,
258 .strategy = &sysctl_intvec,
259 .extra1 = &min_wakeup_granularity_ns,
260 .extra2 = &max_wakeup_granularity_ns,
261 },
262 {
263 .ctl_name = CTL_UNNUMBERED,
264 .procname = "sched_runtime_limit_ns",
265 .data = &sysctl_sched_runtime_limit,
266 .maxlen = sizeof(unsigned int),
267 .mode = 0644,
268 .proc_handler = &proc_dointvec_minmax,
269 .strategy = &sysctl_intvec,
270 .extra1 = &min_sched_granularity_ns,
271 .extra2 = &max_sched_granularity_ns,
272 },
273 {
274 .ctl_name = CTL_UNNUMBERED,
275 .procname = "sched_child_runs_first",
276 .data = &sysctl_sched_child_runs_first,
277 .maxlen = sizeof(unsigned int),
278 .mode = 0644,
279 .proc_handler = &proc_dointvec,
280 },
281 {
282 .ctl_name = CTL_UNNUMBERED,
283 .procname = "sched_features",
284 .data = &sysctl_sched_features,
285 .maxlen = sizeof(unsigned int),
286 .mode = 0644,
287 .proc_handler = &proc_dointvec,
288 },
289#endif
1da177e4
LT
290 {
291 .ctl_name = KERN_PANIC,
292 .procname = "panic",
293 .data = &panic_timeout,
294 .maxlen = sizeof(int),
295 .mode = 0644,
296 .proc_handler = &proc_dointvec,
297 },
298 {
299 .ctl_name = KERN_CORE_USES_PID,
300 .procname = "core_uses_pid",
301 .data = &core_uses_pid,
302 .maxlen = sizeof(int),
303 .mode = 0644,
304 .proc_handler = &proc_dointvec,
305 },
306 {
307 .ctl_name = KERN_CORE_PATTERN,
308 .procname = "core_pattern",
309 .data = core_pattern,
71ce92f3 310 .maxlen = CORENAME_MAX_SIZE,
1da177e4
LT
311 .mode = 0644,
312 .proc_handler = &proc_dostring,
313 .strategy = &sysctl_string,
314 },
34f5a398 315#ifdef CONFIG_PROC_SYSCTL
1da177e4
LT
316 {
317 .ctl_name = KERN_TAINTED,
318 .procname = "tainted",
319 .data = &tainted,
320 .maxlen = sizeof(int),
34f5a398
TT
321 .mode = 0644,
322 .proc_handler = &proc_dointvec_taint,
1da177e4 323 },
34f5a398 324#endif
1da177e4
LT
325 {
326 .ctl_name = KERN_CAP_BSET,
327 .procname = "cap-bound",
328 .data = &cap_bset,
329 .maxlen = sizeof(kernel_cap_t),
330 .mode = 0600,
331 .proc_handler = &proc_dointvec_bset,
332 },
333#ifdef CONFIG_BLK_DEV_INITRD
334 {
335 .ctl_name = KERN_REALROOTDEV,
336 .procname = "real-root-dev",
337 .data = &real_root_dev,
338 .maxlen = sizeof(int),
339 .mode = 0644,
340 .proc_handler = &proc_dointvec,
341 },
342#endif
343#ifdef __sparc__
344 {
345 .ctl_name = KERN_SPARC_REBOOT,
346 .procname = "reboot-cmd",
347 .data = reboot_command,
348 .maxlen = 256,
349 .mode = 0644,
350 .proc_handler = &proc_dostring,
351 .strategy = &sysctl_string,
352 },
353 {
354 .ctl_name = KERN_SPARC_STOP_A,
355 .procname = "stop-a",
356 .data = &stop_a_enabled,
357 .maxlen = sizeof (int),
358 .mode = 0644,
359 .proc_handler = &proc_dointvec,
360 },
361 {
362 .ctl_name = KERN_SPARC_SCONS_PWROFF,
363 .procname = "scons-poweroff",
364 .data = &scons_pwroff,
365 .maxlen = sizeof (int),
366 .mode = 0644,
367 .proc_handler = &proc_dointvec,
368 },
369#endif
370#ifdef __hppa__
371 {
372 .ctl_name = KERN_HPPA_PWRSW,
373 .procname = "soft-power",
374 .data = &pwrsw_enabled,
375 .maxlen = sizeof (int),
376 .mode = 0644,
377 .proc_handler = &proc_dointvec,
378 },
379 {
380 .ctl_name = KERN_HPPA_UNALIGNED,
381 .procname = "unaligned-trap",
382 .data = &unaligned_enabled,
383 .maxlen = sizeof (int),
384 .mode = 0644,
385 .proc_handler = &proc_dointvec,
386 },
387#endif
388 {
389 .ctl_name = KERN_CTLALTDEL,
390 .procname = "ctrl-alt-del",
391 .data = &C_A_D,
392 .maxlen = sizeof(int),
393 .mode = 0644,
394 .proc_handler = &proc_dointvec,
395 },
396 {
397 .ctl_name = KERN_PRINTK,
398 .procname = "printk",
399 .data = &console_loglevel,
400 .maxlen = 4*sizeof(int),
401 .mode = 0644,
402 .proc_handler = &proc_dointvec,
403 },
404#ifdef CONFIG_KMOD
405 {
406 .ctl_name = KERN_MODPROBE,
407 .procname = "modprobe",
408 .data = &modprobe_path,
409 .maxlen = KMOD_PATH_LEN,
410 .mode = 0644,
411 .proc_handler = &proc_dostring,
412 .strategy = &sysctl_string,
413 },
414#endif
57ae2508 415#if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
1da177e4
LT
416 {
417 .ctl_name = KERN_HOTPLUG,
418 .procname = "hotplug",
312c004d
KS
419 .data = &uevent_helper,
420 .maxlen = UEVENT_HELPER_PATH_LEN,
1da177e4
LT
421 .mode = 0644,
422 .proc_handler = &proc_dostring,
423 .strategy = &sysctl_string,
424 },
425#endif
426#ifdef CONFIG_CHR_DEV_SG
427 {
428 .ctl_name = KERN_SG_BIG_BUFF,
429 .procname = "sg-big-buff",
430 .data = &sg_big_buff,
431 .maxlen = sizeof (int),
432 .mode = 0444,
433 .proc_handler = &proc_dointvec,
434 },
435#endif
436#ifdef CONFIG_BSD_PROCESS_ACCT
437 {
438 .ctl_name = KERN_ACCT,
439 .procname = "acct",
440 .data = &acct_parm,
441 .maxlen = 3*sizeof(int),
442 .mode = 0644,
443 .proc_handler = &proc_dointvec,
444 },
445#endif
1da177e4
LT
446#ifdef CONFIG_MAGIC_SYSRQ
447 {
448 .ctl_name = KERN_SYSRQ,
449 .procname = "sysrq",
5d6f647f 450 .data = &__sysrq_enabled,
1da177e4
LT
451 .maxlen = sizeof (int),
452 .mode = 0644,
453 .proc_handler = &proc_dointvec,
454 },
455#endif
d6f8ff73 456#ifdef CONFIG_PROC_SYSCTL
1da177e4
LT
457 {
458 .ctl_name = KERN_CADPID,
459 .procname = "cad_pid",
9ec52099 460 .data = NULL,
1da177e4
LT
461 .maxlen = sizeof (int),
462 .mode = 0600,
9ec52099 463 .proc_handler = &proc_do_cad_pid,
1da177e4 464 },
d6f8ff73 465#endif
1da177e4
LT
466 {
467 .ctl_name = KERN_MAX_THREADS,
468 .procname = "threads-max",
469 .data = &max_threads,
470 .maxlen = sizeof(int),
471 .mode = 0644,
472 .proc_handler = &proc_dointvec,
473 },
474 {
475 .ctl_name = KERN_RANDOM,
476 .procname = "random",
477 .mode = 0555,
478 .child = random_table,
479 },
480#ifdef CONFIG_UNIX98_PTYS
481 {
482 .ctl_name = KERN_PTY,
483 .procname = "pty",
484 .mode = 0555,
485 .child = pty_table,
486 },
487#endif
488 {
489 .ctl_name = KERN_OVERFLOWUID,
490 .procname = "overflowuid",
491 .data = &overflowuid,
492 .maxlen = sizeof(int),
493 .mode = 0644,
494 .proc_handler = &proc_dointvec_minmax,
495 .strategy = &sysctl_intvec,
496 .extra1 = &minolduid,
497 .extra2 = &maxolduid,
498 },
499 {
500 .ctl_name = KERN_OVERFLOWGID,
501 .procname = "overflowgid",
502 .data = &overflowgid,
503 .maxlen = sizeof(int),
504 .mode = 0644,
505 .proc_handler = &proc_dointvec_minmax,
506 .strategy = &sysctl_intvec,
507 .extra1 = &minolduid,
508 .extra2 = &maxolduid,
509 },
347a8dc3 510#ifdef CONFIG_S390
1da177e4
LT
511#ifdef CONFIG_MATHEMU
512 {
513 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
514 .procname = "ieee_emulation_warnings",
515 .data = &sysctl_ieee_emulation_warnings,
516 .maxlen = sizeof(int),
517 .mode = 0644,
518 .proc_handler = &proc_dointvec,
519 },
520#endif
521#ifdef CONFIG_NO_IDLE_HZ
522 {
523 .ctl_name = KERN_HZ_TIMER,
524 .procname = "hz_timer",
525 .data = &sysctl_hz_timer,
526 .maxlen = sizeof(int),
527 .mode = 0644,
528 .proc_handler = &proc_dointvec,
529 },
530#endif
531 {
532 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
533 .procname = "userprocess_debug",
534 .data = &sysctl_userprocess_debug,
535 .maxlen = sizeof(int),
536 .mode = 0644,
537 .proc_handler = &proc_dointvec,
538 },
539#endif
540 {
541 .ctl_name = KERN_PIDMAX,
542 .procname = "pid_max",
543 .data = &pid_max,
544 .maxlen = sizeof (int),
545 .mode = 0644,
546 .proc_handler = &proc_dointvec_minmax,
547 .strategy = sysctl_intvec,
548 .extra1 = &pid_max_min,
549 .extra2 = &pid_max_max,
550 },
551 {
552 .ctl_name = KERN_PANIC_ON_OOPS,
553 .procname = "panic_on_oops",
554 .data = &panic_on_oops,
555 .maxlen = sizeof(int),
556 .mode = 0644,
557 .proc_handler = &proc_dointvec,
558 },
559 {
560 .ctl_name = KERN_PRINTK_RATELIMIT,
561 .procname = "printk_ratelimit",
562 .data = &printk_ratelimit_jiffies,
563 .maxlen = sizeof(int),
564 .mode = 0644,
565 .proc_handler = &proc_dointvec_jiffies,
566 .strategy = &sysctl_jiffies,
567 },
568 {
569 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
570 .procname = "printk_ratelimit_burst",
571 .data = &printk_ratelimit_burst,
572 .maxlen = sizeof(int),
573 .mode = 0644,
574 .proc_handler = &proc_dointvec,
575 },
576 {
577 .ctl_name = KERN_NGROUPS_MAX,
578 .procname = "ngroups_max",
579 .data = &ngroups_max,
580 .maxlen = sizeof (int),
581 .mode = 0444,
582 .proc_handler = &proc_dointvec,
583 },
584#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
585 {
586 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
587 .procname = "unknown_nmi_panic",
588 .data = &unknown_nmi_panic,
589 .maxlen = sizeof (int),
590 .mode = 0644,
2fbe7b25 591 .proc_handler = &proc_dointvec,
1da177e4 592 },
407984f1
DZ
593 {
594 .ctl_name = KERN_NMI_WATCHDOG,
595 .procname = "nmi_watchdog",
596 .data = &nmi_watchdog_enabled,
597 .maxlen = sizeof (int),
598 .mode = 0644,
599 .proc_handler = &proc_nmi_enabled,
1da177e4
LT
600 },
601#endif
602#if defined(CONFIG_X86)
8da5adda
DZ
603 {
604 .ctl_name = KERN_PANIC_ON_NMI,
605 .procname = "panic_on_unrecovered_nmi",
606 .data = &panic_on_unrecovered_nmi,
607 .maxlen = sizeof(int),
608 .mode = 0644,
609 .proc_handler = &proc_dointvec,
610 },
1da177e4
LT
611 {
612 .ctl_name = KERN_BOOTLOADER_TYPE,
613 .procname = "bootloader_type",
614 .data = &bootloader_type,
615 .maxlen = sizeof (int),
616 .mode = 0444,
617 .proc_handler = &proc_dointvec,
618 },
0741f4d2
CE
619 {
620 .ctl_name = CTL_UNNUMBERED,
621 .procname = "kstack_depth_to_print",
622 .data = &kstack_depth_to_print,
623 .maxlen = sizeof(int),
624 .mode = 0644,
625 .proc_handler = &proc_dointvec,
626 },
1da177e4 627#endif
7a9166e3 628#if defined(CONFIG_MMU)
1da177e4
LT
629 {
630 .ctl_name = KERN_RANDOMIZE,
631 .procname = "randomize_va_space",
632 .data = &randomize_va_space,
633 .maxlen = sizeof(int),
634 .mode = 0644,
635 .proc_handler = &proc_dointvec,
636 },
7a9166e3 637#endif
0152fb37 638#if defined(CONFIG_S390) && defined(CONFIG_SMP)
951f22d5
MS
639 {
640 .ctl_name = KERN_SPIN_RETRY,
641 .procname = "spin_retry",
642 .data = &spin_retry,
643 .maxlen = sizeof (int),
644 .mode = 0644,
645 .proc_handler = &proc_dointvec,
646 },
c255d844
PM
647#endif
648#ifdef CONFIG_ACPI_SLEEP
649 {
650 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
651 .procname = "acpi_video_flags",
652 .data = &acpi_video_flags,
653 .maxlen = sizeof (unsigned long),
654 .mode = 0644,
7f99f06f 655 .proc_handler = &proc_doulongvec_minmax,
c255d844 656 },
d2b176ed
JS
657#endif
658#ifdef CONFIG_IA64
659 {
660 .ctl_name = KERN_IA64_UNALIGNED,
661 .procname = "ignore-unaligned-usertrap",
662 .data = &no_unaligned_warning,
663 .maxlen = sizeof (int),
664 .mode = 0644,
665 .proc_handler = &proc_dointvec,
666 },
bebfa101
AK
667#endif
668#ifdef CONFIG_COMPAT
669 {
670 .ctl_name = KERN_COMPAT_LOG,
671 .procname = "compat-log",
672 .data = &compat_log,
673 .maxlen = sizeof (int),
674 .mode = 0644,
675 .proc_handler = &proc_dointvec,
676 },
951f22d5 677#endif
23f78d4a
IM
678#ifdef CONFIG_RT_MUTEXES
679 {
680 .ctl_name = KERN_MAX_LOCK_DEPTH,
681 .procname = "max_lock_depth",
682 .data = &max_lock_depth,
683 .maxlen = sizeof(int),
684 .mode = 0644,
685 .proc_handler = &proc_dointvec,
686 },
687#endif
5096add8
KC
688#ifdef CONFIG_PROC_FS
689 {
690 .ctl_name = CTL_UNNUMBERED,
691 .procname = "maps_protect",
692 .data = &maps_protect,
693 .maxlen = sizeof(int),
694 .mode = 0644,
695 .proc_handler = &proc_dointvec,
696 },
697#endif
23f78d4a 698
1da177e4
LT
699 { .ctl_name = 0 }
700};
701
702/* Constants for minimum and maximum testing in vm_table.
703 We use these as one-element integer vectors. */
704static int zero;
705static int one_hundred = 100;
706
707
708static ctl_table vm_table[] = {
709 {
710 .ctl_name = VM_OVERCOMMIT_MEMORY,
711 .procname = "overcommit_memory",
712 .data = &sysctl_overcommit_memory,
713 .maxlen = sizeof(sysctl_overcommit_memory),
714 .mode = 0644,
715 .proc_handler = &proc_dointvec,
716 },
fadd8fbd
KH
717 {
718 .ctl_name = VM_PANIC_ON_OOM,
719 .procname = "panic_on_oom",
720 .data = &sysctl_panic_on_oom,
721 .maxlen = sizeof(sysctl_panic_on_oom),
722 .mode = 0644,
723 .proc_handler = &proc_dointvec,
724 },
1da177e4
LT
725 {
726 .ctl_name = VM_OVERCOMMIT_RATIO,
727 .procname = "overcommit_ratio",
728 .data = &sysctl_overcommit_ratio,
729 .maxlen = sizeof(sysctl_overcommit_ratio),
730 .mode = 0644,
731 .proc_handler = &proc_dointvec,
732 },
733 {
734 .ctl_name = VM_PAGE_CLUSTER,
735 .procname = "page-cluster",
736 .data = &page_cluster,
737 .maxlen = sizeof(int),
738 .mode = 0644,
739 .proc_handler = &proc_dointvec,
740 },
741 {
742 .ctl_name = VM_DIRTY_BACKGROUND,
743 .procname = "dirty_background_ratio",
744 .data = &dirty_background_ratio,
745 .maxlen = sizeof(dirty_background_ratio),
746 .mode = 0644,
747 .proc_handler = &proc_dointvec_minmax,
748 .strategy = &sysctl_intvec,
749 .extra1 = &zero,
750 .extra2 = &one_hundred,
751 },
752 {
753 .ctl_name = VM_DIRTY_RATIO,
754 .procname = "dirty_ratio",
755 .data = &vm_dirty_ratio,
756 .maxlen = sizeof(vm_dirty_ratio),
757 .mode = 0644,
758 .proc_handler = &proc_dointvec_minmax,
759 .strategy = &sysctl_intvec,
760 .extra1 = &zero,
761 .extra2 = &one_hundred,
762 },
763 {
764 .ctl_name = VM_DIRTY_WB_CS,
765 .procname = "dirty_writeback_centisecs",
f6ef9438
BS
766 .data = &dirty_writeback_interval,
767 .maxlen = sizeof(dirty_writeback_interval),
1da177e4
LT
768 .mode = 0644,
769 .proc_handler = &dirty_writeback_centisecs_handler,
770 },
771 {
772 .ctl_name = VM_DIRTY_EXPIRE_CS,
773 .procname = "dirty_expire_centisecs",
f6ef9438
BS
774 .data = &dirty_expire_interval,
775 .maxlen = sizeof(dirty_expire_interval),
1da177e4 776 .mode = 0644,
f6ef9438 777 .proc_handler = &proc_dointvec_userhz_jiffies,
1da177e4
LT
778 },
779 {
780 .ctl_name = VM_NR_PDFLUSH_THREADS,
781 .procname = "nr_pdflush_threads",
782 .data = &nr_pdflush_threads,
783 .maxlen = sizeof nr_pdflush_threads,
784 .mode = 0444 /* read-only*/,
785 .proc_handler = &proc_dointvec,
786 },
787 {
788 .ctl_name = VM_SWAPPINESS,
789 .procname = "swappiness",
790 .data = &vm_swappiness,
791 .maxlen = sizeof(vm_swappiness),
792 .mode = 0644,
793 .proc_handler = &proc_dointvec_minmax,
794 .strategy = &sysctl_intvec,
795 .extra1 = &zero,
796 .extra2 = &one_hundred,
797 },
798#ifdef CONFIG_HUGETLB_PAGE
799 {
800 .ctl_name = VM_HUGETLB_PAGES,
801 .procname = "nr_hugepages",
802 .data = &max_huge_pages,
803 .maxlen = sizeof(unsigned long),
804 .mode = 0644,
805 .proc_handler = &hugetlb_sysctl_handler,
806 .extra1 = (void *)&hugetlb_zero,
807 .extra2 = (void *)&hugetlb_infinity,
808 },
809 {
810 .ctl_name = VM_HUGETLB_GROUP,
811 .procname = "hugetlb_shm_group",
812 .data = &sysctl_hugetlb_shm_group,
813 .maxlen = sizeof(gid_t),
814 .mode = 0644,
815 .proc_handler = &proc_dointvec,
816 },
817#endif
818 {
819 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
820 .procname = "lowmem_reserve_ratio",
821 .data = &sysctl_lowmem_reserve_ratio,
822 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
823 .mode = 0644,
824 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
825 .strategy = &sysctl_intvec,
826 },
9d0243bc
AM
827 {
828 .ctl_name = VM_DROP_PAGECACHE,
829 .procname = "drop_caches",
830 .data = &sysctl_drop_caches,
831 .maxlen = sizeof(int),
832 .mode = 0644,
833 .proc_handler = drop_caches_sysctl_handler,
834 .strategy = &sysctl_intvec,
835 },
1da177e4
LT
836 {
837 .ctl_name = VM_MIN_FREE_KBYTES,
838 .procname = "min_free_kbytes",
839 .data = &min_free_kbytes,
840 .maxlen = sizeof(min_free_kbytes),
841 .mode = 0644,
842 .proc_handler = &min_free_kbytes_sysctl_handler,
843 .strategy = &sysctl_intvec,
844 .extra1 = &zero,
845 },
8ad4b1fb
RS
846 {
847 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
848 .procname = "percpu_pagelist_fraction",
849 .data = &percpu_pagelist_fraction,
850 .maxlen = sizeof(percpu_pagelist_fraction),
851 .mode = 0644,
852 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
853 .strategy = &sysctl_intvec,
854 .extra1 = &min_percpu_pagelist_fract,
855 },
1da177e4
LT
856#ifdef CONFIG_MMU
857 {
858 .ctl_name = VM_MAX_MAP_COUNT,
859 .procname = "max_map_count",
860 .data = &sysctl_max_map_count,
861 .maxlen = sizeof(sysctl_max_map_count),
862 .mode = 0644,
863 .proc_handler = &proc_dointvec
864 },
865#endif
866 {
867 .ctl_name = VM_LAPTOP_MODE,
868 .procname = "laptop_mode",
869 .data = &laptop_mode,
870 .maxlen = sizeof(laptop_mode),
871 .mode = 0644,
ed5b43f1
BS
872 .proc_handler = &proc_dointvec_jiffies,
873 .strategy = &sysctl_jiffies,
1da177e4
LT
874 },
875 {
876 .ctl_name = VM_BLOCK_DUMP,
877 .procname = "block_dump",
878 .data = &block_dump,
879 .maxlen = sizeof(block_dump),
880 .mode = 0644,
881 .proc_handler = &proc_dointvec,
882 .strategy = &sysctl_intvec,
883 .extra1 = &zero,
884 },
885 {
886 .ctl_name = VM_VFS_CACHE_PRESSURE,
887 .procname = "vfs_cache_pressure",
888 .data = &sysctl_vfs_cache_pressure,
889 .maxlen = sizeof(sysctl_vfs_cache_pressure),
890 .mode = 0644,
891 .proc_handler = &proc_dointvec,
892 .strategy = &sysctl_intvec,
893 .extra1 = &zero,
894 },
895#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
896 {
897 .ctl_name = VM_LEGACY_VA_LAYOUT,
898 .procname = "legacy_va_layout",
899 .data = &sysctl_legacy_va_layout,
900 .maxlen = sizeof(sysctl_legacy_va_layout),
901 .mode = 0644,
902 .proc_handler = &proc_dointvec,
903 .strategy = &sysctl_intvec,
904 .extra1 = &zero,
905 },
906#endif
1743660b
CL
907#ifdef CONFIG_NUMA
908 {
909 .ctl_name = VM_ZONE_RECLAIM_MODE,
910 .procname = "zone_reclaim_mode",
911 .data = &zone_reclaim_mode,
912 .maxlen = sizeof(zone_reclaim_mode),
913 .mode = 0644,
914 .proc_handler = &proc_dointvec,
c84db23c
CL
915 .strategy = &sysctl_intvec,
916 .extra1 = &zero,
1743660b 917 },
9614634f
CL
918 {
919 .ctl_name = VM_MIN_UNMAPPED,
920 .procname = "min_unmapped_ratio",
921 .data = &sysctl_min_unmapped_ratio,
922 .maxlen = sizeof(sysctl_min_unmapped_ratio),
923 .mode = 0644,
924 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
925 .strategy = &sysctl_intvec,
926 .extra1 = &zero,
927 .extra2 = &one_hundred,
928 },
0ff38490
CL
929 {
930 .ctl_name = VM_MIN_SLAB,
931 .procname = "min_slab_ratio",
932 .data = &sysctl_min_slab_ratio,
933 .maxlen = sizeof(sysctl_min_slab_ratio),
934 .mode = 0644,
935 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
936 .strategy = &sysctl_intvec,
937 .extra1 = &zero,
938 .extra2 = &one_hundred,
939 },
e6e5494c 940#endif
77461ab3
CL
941#ifdef CONFIG_SMP
942 {
943 .ctl_name = CTL_UNNUMBERED,
944 .procname = "stat_interval",
945 .data = &sysctl_stat_interval,
946 .maxlen = sizeof(sysctl_stat_interval),
947 .mode = 0644,
948 .proc_handler = &proc_dointvec_jiffies,
949 .strategy = &sysctl_jiffies,
950 },
951#endif
ed032189
EP
952#ifdef CONFIG_SECURITY
953 {
954 .ctl_name = CTL_UNNUMBERED,
955 .procname = "mmap_min_addr",
956 .data = &mmap_min_addr,
957 .maxlen = sizeof(unsigned long),
958 .mode = 0644,
959 .proc_handler = &proc_doulongvec_minmax,
960 },
961#endif
5c36e657
PM
962#if defined(CONFIG_X86_32) || \
963 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
e6e5494c
IM
964 {
965 .ctl_name = VM_VDSO_ENABLED,
966 .procname = "vdso_enabled",
967 .data = &vdso_enabled,
968 .maxlen = sizeof(vdso_enabled),
969 .mode = 0644,
970 .proc_handler = &proc_dointvec,
971 .strategy = &sysctl_intvec,
972 .extra1 = &zero,
973 },
1da177e4
LT
974#endif
975 { .ctl_name = 0 }
976};
977
2abc26fc
EB
978#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
979static ctl_table binfmt_misc_table[] = {
980 { .ctl_name = 0 }
981};
982#endif
983
1da177e4
LT
984static ctl_table fs_table[] = {
985 {
986 .ctl_name = FS_NRINODE,
987 .procname = "inode-nr",
988 .data = &inodes_stat,
989 .maxlen = 2*sizeof(int),
990 .mode = 0444,
991 .proc_handler = &proc_dointvec,
992 },
993 {
994 .ctl_name = FS_STATINODE,
995 .procname = "inode-state",
996 .data = &inodes_stat,
997 .maxlen = 7*sizeof(int),
998 .mode = 0444,
999 .proc_handler = &proc_dointvec,
1000 },
1001 {
1002 .ctl_name = FS_NRFILE,
1003 .procname = "file-nr",
1004 .data = &files_stat,
1005 .maxlen = 3*sizeof(int),
1006 .mode = 0444,
529bf6be 1007 .proc_handler = &proc_nr_files,
1da177e4
LT
1008 },
1009 {
1010 .ctl_name = FS_MAXFILE,
1011 .procname = "file-max",
1012 .data = &files_stat.max_files,
1013 .maxlen = sizeof(int),
1014 .mode = 0644,
1015 .proc_handler = &proc_dointvec,
1016 },
1017 {
1018 .ctl_name = FS_DENTRY,
1019 .procname = "dentry-state",
1020 .data = &dentry_stat,
1021 .maxlen = 6*sizeof(int),
1022 .mode = 0444,
1023 .proc_handler = &proc_dointvec,
1024 },
1025 {
1026 .ctl_name = FS_OVERFLOWUID,
1027 .procname = "overflowuid",
1028 .data = &fs_overflowuid,
1029 .maxlen = sizeof(int),
1030 .mode = 0644,
1031 .proc_handler = &proc_dointvec_minmax,
1032 .strategy = &sysctl_intvec,
1033 .extra1 = &minolduid,
1034 .extra2 = &maxolduid,
1035 },
1036 {
1037 .ctl_name = FS_OVERFLOWGID,
1038 .procname = "overflowgid",
1039 .data = &fs_overflowgid,
1040 .maxlen = sizeof(int),
1041 .mode = 0644,
1042 .proc_handler = &proc_dointvec_minmax,
1043 .strategy = &sysctl_intvec,
1044 .extra1 = &minolduid,
1045 .extra2 = &maxolduid,
1046 },
1047 {
1048 .ctl_name = FS_LEASES,
1049 .procname = "leases-enable",
1050 .data = &leases_enable,
1051 .maxlen = sizeof(int),
1052 .mode = 0644,
1053 .proc_handler = &proc_dointvec,
1054 },
1055#ifdef CONFIG_DNOTIFY
1056 {
1057 .ctl_name = FS_DIR_NOTIFY,
1058 .procname = "dir-notify-enable",
1059 .data = &dir_notify_enable,
1060 .maxlen = sizeof(int),
1061 .mode = 0644,
1062 .proc_handler = &proc_dointvec,
1063 },
1064#endif
1065#ifdef CONFIG_MMU
1066 {
1067 .ctl_name = FS_LEASE_TIME,
1068 .procname = "lease-break-time",
1069 .data = &lease_break_time,
1070 .maxlen = sizeof(int),
1071 .mode = 0644,
1072 .proc_handler = &proc_dointvec,
1073 },
1074 {
1075 .ctl_name = FS_AIO_NR,
1076 .procname = "aio-nr",
1077 .data = &aio_nr,
1078 .maxlen = sizeof(aio_nr),
1079 .mode = 0444,
d55b5fda 1080 .proc_handler = &proc_doulongvec_minmax,
1da177e4
LT
1081 },
1082 {
1083 .ctl_name = FS_AIO_MAX_NR,
1084 .procname = "aio-max-nr",
1085 .data = &aio_max_nr,
1086 .maxlen = sizeof(aio_max_nr),
1087 .mode = 0644,
d55b5fda 1088 .proc_handler = &proc_doulongvec_minmax,
1da177e4 1089 },
2d9048e2 1090#ifdef CONFIG_INOTIFY_USER
0399cb08
RL
1091 {
1092 .ctl_name = FS_INOTIFY,
1093 .procname = "inotify",
1094 .mode = 0555,
1095 .child = inotify_table,
1096 },
1097#endif
1da177e4 1098#endif
d6e71144
AC
1099 {
1100 .ctl_name = KERN_SETUID_DUMPABLE,
1101 .procname = "suid_dumpable",
1102 .data = &suid_dumpable,
1103 .maxlen = sizeof(int),
1104 .mode = 0644,
1105 .proc_handler = &proc_dointvec,
1106 },
2abc26fc
EB
1107#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1108 {
1109 .ctl_name = CTL_UNNUMBERED,
1110 .procname = "binfmt_misc",
1111 .mode = 0555,
1112 .child = binfmt_misc_table,
1113 },
1114#endif
1da177e4
LT
1115 { .ctl_name = 0 }
1116};
1117
1118static ctl_table debug_table[] = {
1119 { .ctl_name = 0 }
1120};
1121
1122static ctl_table dev_table[] = {
1123 { .ctl_name = 0 }
0eeca283 1124};
1da177e4 1125
330d57fb
AV
1126static DEFINE_SPINLOCK(sysctl_lock);
1127
1128/* called under sysctl_lock */
1129static int use_table(struct ctl_table_header *p)
1130{
1131 if (unlikely(p->unregistering))
1132 return 0;
1133 p->used++;
1134 return 1;
1135}
1136
1137/* called under sysctl_lock */
1138static void unuse_table(struct ctl_table_header *p)
1139{
1140 if (!--p->used)
1141 if (unlikely(p->unregistering))
1142 complete(p->unregistering);
1143}
1144
1145/* called under sysctl_lock, will reacquire if has to wait */
1146static void start_unregistering(struct ctl_table_header *p)
1147{
1148 /*
1149 * if p->used is 0, nobody will ever touch that entry again;
1150 * we'll eliminate all paths to it before dropping sysctl_lock
1151 */
1152 if (unlikely(p->used)) {
1153 struct completion wait;
1154 init_completion(&wait);
1155 p->unregistering = &wait;
1156 spin_unlock(&sysctl_lock);
1157 wait_for_completion(&wait);
1158 spin_lock(&sysctl_lock);
1159 }
1160 /*
1161 * do not remove from the list until nobody holds it; walking the
1162 * list in do_sysctl() relies on that.
1163 */
1164 list_del_init(&p->ctl_entry);
1165}
1166
805b5d5e
EB
1167void sysctl_head_finish(struct ctl_table_header *head)
1168{
1169 if (!head)
1170 return;
1171 spin_lock(&sysctl_lock);
1172 unuse_table(head);
1173 spin_unlock(&sysctl_lock);
1174}
1175
1176struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1177{
1178 struct ctl_table_header *head;
1179 struct list_head *tmp;
1180 spin_lock(&sysctl_lock);
1181 if (prev) {
1182 tmp = &prev->ctl_entry;
1183 unuse_table(prev);
1184 goto next;
1185 }
1186 tmp = &root_table_header.ctl_entry;
1187 for (;;) {
1188 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1189
1190 if (!use_table(head))
1191 goto next;
1192 spin_unlock(&sysctl_lock);
1193 return head;
1194 next:
1195 tmp = tmp->next;
1196 if (tmp == &root_table_header.ctl_entry)
1197 break;
1198 }
1199 spin_unlock(&sysctl_lock);
1200 return NULL;
1201}
1202
b89a8171 1203#ifdef CONFIG_SYSCTL_SYSCALL
1da177e4
LT
1204int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1205 void __user *newval, size_t newlen)
1206{
805b5d5e 1207 struct ctl_table_header *head;
330d57fb 1208 int error = -ENOTDIR;
1da177e4
LT
1209
1210 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1211 return -ENOTDIR;
1212 if (oldval) {
1213 int old_len;
1214 if (!oldlenp || get_user(old_len, oldlenp))
1215 return -EFAULT;
1216 }
330d57fb 1217
805b5d5e
EB
1218 for (head = sysctl_head_next(NULL); head;
1219 head = sysctl_head_next(head)) {
330d57fb 1220 error = parse_table(name, nlen, oldval, oldlenp,
1f29bcd7 1221 newval, newlen, head->ctl_table);
805b5d5e
EB
1222 if (error != -ENOTDIR) {
1223 sysctl_head_finish(head);
330d57fb 1224 break;
805b5d5e
EB
1225 }
1226 }
330d57fb 1227 return error;
1da177e4
LT
1228}
1229
1230asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1231{
1232 struct __sysctl_args tmp;
1233 int error;
1234
1235 if (copy_from_user(&tmp, args, sizeof(tmp)))
1236 return -EFAULT;
1237
1238 lock_kernel();
1239 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1240 tmp.newval, tmp.newlen);
1241 unlock_kernel();
1242 return error;
1243}
b89a8171 1244#endif /* CONFIG_SYSCTL_SYSCALL */
1da177e4
LT
1245
1246/*
1ff007eb 1247 * sysctl_perm does NOT grant the superuser all rights automatically, because
1da177e4
LT
1248 * some sysctl variables are readonly even to root.
1249 */
1250
1251static int test_perm(int mode, int op)
1252{
1253 if (!current->euid)
1254 mode >>= 6;
1255 else if (in_egroup_p(0))
1256 mode >>= 3;
1257 if ((mode & op & 0007) == op)
1258 return 0;
1259 return -EACCES;
1260}
1261
1ff007eb 1262int sysctl_perm(ctl_table *table, int op)
1da177e4
LT
1263{
1264 int error;
1265 error = security_sysctl(table, op);
1266 if (error)
1267 return error;
1268 return test_perm(table->mode, op);
1269}
1270
b89a8171 1271#ifdef CONFIG_SYSCTL_SYSCALL
1da177e4
LT
1272static int parse_table(int __user *name, int nlen,
1273 void __user *oldval, size_t __user *oldlenp,
1274 void __user *newval, size_t newlen,
1f29bcd7 1275 ctl_table *table)
1da177e4
LT
1276{
1277 int n;
1278repeat:
1279 if (!nlen)
1280 return -ENOTDIR;
1281 if (get_user(n, name))
1282 return -EFAULT;
d99f160a
EB
1283 for ( ; table->ctl_name || table->procname; table++) {
1284 if (!table->ctl_name)
1285 continue;
6703ddfc 1286 if (n == table->ctl_name) {
1da177e4
LT
1287 int error;
1288 if (table->child) {
1ff007eb 1289 if (sysctl_perm(table, 001))
1da177e4 1290 return -EPERM;
1da177e4
LT
1291 name++;
1292 nlen--;
1293 table = table->child;
1294 goto repeat;
1295 }
1296 error = do_sysctl_strategy(table, name, nlen,
1297 oldval, oldlenp,
1f29bcd7 1298 newval, newlen);
1da177e4
LT
1299 return error;
1300 }
1301 }
1302 return -ENOTDIR;
1303}
1304
1305/* Perform the actual read/write of a sysctl table entry. */
1306int do_sysctl_strategy (ctl_table *table,
1307 int __user *name, int nlen,
1308 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 1309 void __user *newval, size_t newlen)
1da177e4
LT
1310{
1311 int op = 0, rc;
1312 size_t len;
1313
1314 if (oldval)
1315 op |= 004;
1316 if (newval)
1317 op |= 002;
1ff007eb 1318 if (sysctl_perm(table, op))
1da177e4
LT
1319 return -EPERM;
1320
1321 if (table->strategy) {
1322 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1f29bcd7 1323 newval, newlen);
1da177e4
LT
1324 if (rc < 0)
1325 return rc;
1326 if (rc > 0)
1327 return 0;
1328 }
1329
1330 /* If there is no strategy routine, or if the strategy returns
1331 * zero, proceed with automatic r/w */
1332 if (table->data && table->maxlen) {
1333 if (oldval && oldlenp) {
1334 if (get_user(len, oldlenp))
1335 return -EFAULT;
1336 if (len) {
1337 if (len > table->maxlen)
1338 len = table->maxlen;
1339 if(copy_to_user(oldval, table->data, len))
1340 return -EFAULT;
1341 if(put_user(len, oldlenp))
1342 return -EFAULT;
1343 }
1344 }
1345 if (newval && newlen) {
1346 len = newlen;
1347 if (len > table->maxlen)
1348 len = table->maxlen;
1349 if(copy_from_user(table->data, newval, len))
1350 return -EFAULT;
1351 }
1352 }
1353 return 0;
1354}
b89a8171 1355#endif /* CONFIG_SYSCTL_SYSCALL */
1da177e4 1356
d912b0cc
EB
1357static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1358{
1359 for (; table->ctl_name || table->procname; table++) {
1360 table->parent = parent;
1361 if (table->child)
1362 sysctl_set_parent(table, table->child);
1363 }
1364}
1365
1366static __init int sysctl_init(void)
1367{
1368 sysctl_set_parent(NULL, root_table);
1369 return 0;
1370}
1371
1372core_initcall(sysctl_init);
1373
1da177e4
LT
1374/**
1375 * register_sysctl_table - register a sysctl hierarchy
1376 * @table: the top-level table structure
1da177e4
LT
1377 *
1378 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1379 * array. An entry with a ctl_name of 0 terminates the table.
1380 *
1381 * The members of the &ctl_table structure are used as follows:
1382 *
1383 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1384 * must be unique within that level of sysctl
1385 *
1386 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1387 * enter a sysctl file
1388 *
1389 * data - a pointer to data for use by proc_handler
1390 *
1391 * maxlen - the maximum size in bytes of the data
1392 *
1393 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1394 *
1395 * child - a pointer to the child sysctl table if this entry is a directory, or
1396 * %NULL.
1397 *
1398 * proc_handler - the text handler routine (described below)
1399 *
1400 * strategy - the strategy routine (described below)
1401 *
1402 * de - for internal use by the sysctl routines
1403 *
1404 * extra1, extra2 - extra pointers usable by the proc handler routines
1405 *
1406 * Leaf nodes in the sysctl tree will be represented by a single file
1407 * under /proc; non-leaf nodes will be represented by directories.
1408 *
1409 * sysctl(2) can automatically manage read and write requests through
1410 * the sysctl table. The data and maxlen fields of the ctl_table
1411 * struct enable minimal validation of the values being written to be
1412 * performed, and the mode field allows minimal authentication.
1413 *
1414 * More sophisticated management can be enabled by the provision of a
1415 * strategy routine with the table entry. This will be called before
1416 * any automatic read or write of the data is performed.
1417 *
1418 * The strategy routine may return
1419 *
1420 * < 0 - Error occurred (error is passed to user process)
1421 *
1422 * 0 - OK - proceed with automatic read or write.
1423 *
1424 * > 0 - OK - read or write has been done by the strategy routine, so
1425 * return immediately.
1426 *
1427 * There must be a proc_handler routine for any terminal nodes
1428 * mirrored under /proc/sys (non-terminals are handled by a built-in
1429 * directory handler). Several default handlers are available to
1430 * cover common cases -
1431 *
1432 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1433 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1434 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1435 *
1436 * It is the handler's job to read the input buffer from user memory
1437 * and process it. The handler should return 0 on success.
1438 *
1439 * This routine returns %NULL on a failure to register, and a pointer
1440 * to the table header on success.
1441 */
0b4d4147 1442struct ctl_table_header *register_sysctl_table(ctl_table * table)
1da177e4
LT
1443{
1444 struct ctl_table_header *tmp;
1445 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1446 if (!tmp)
1447 return NULL;
1448 tmp->ctl_table = table;
1449 INIT_LIST_HEAD(&tmp->ctl_entry);
330d57fb
AV
1450 tmp->used = 0;
1451 tmp->unregistering = NULL;
d912b0cc 1452 sysctl_set_parent(NULL, table);
330d57fb 1453 spin_lock(&sysctl_lock);
0b4d4147 1454 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
330d57fb 1455 spin_unlock(&sysctl_lock);
1da177e4
LT
1456 return tmp;
1457}
1458
1459/**
1460 * unregister_sysctl_table - unregister a sysctl table hierarchy
1461 * @header: the header returned from register_sysctl_table
1462 *
1463 * Unregisters the sysctl table and all children. proc entries may not
1464 * actually be removed until they are no longer used by anyone.
1465 */
1466void unregister_sysctl_table(struct ctl_table_header * header)
1467{
330d57fb
AV
1468 might_sleep();
1469 spin_lock(&sysctl_lock);
1470 start_unregistering(header);
330d57fb 1471 spin_unlock(&sysctl_lock);
1da177e4
LT
1472 kfree(header);
1473}
1474
b89a8171 1475#else /* !CONFIG_SYSCTL */
93a6fefe 1476struct ctl_table_header *register_sysctl_table(ctl_table * table)
b89a8171
EB
1477{
1478 return NULL;
1479}
1480
1481void unregister_sysctl_table(struct ctl_table_header * table)
1482{
1483}
1484
1485#endif /* CONFIG_SYSCTL */
1486
1da177e4
LT
1487/*
1488 * /proc/sys support
1489 */
1490
b89a8171 1491#ifdef CONFIG_PROC_SYSCTL
1da177e4 1492
b1ba4ddd
AB
1493static int _proc_do_string(void* data, int maxlen, int write,
1494 struct file *filp, void __user *buffer,
1495 size_t *lenp, loff_t *ppos)
1da177e4
LT
1496{
1497 size_t len;
1498 char __user *p;
1499 char c;
8d060877
ON
1500
1501 if (!data || !maxlen || !*lenp) {
1da177e4
LT
1502 *lenp = 0;
1503 return 0;
1504 }
8d060877 1505
1da177e4
LT
1506 if (write) {
1507 len = 0;
1508 p = buffer;
1509 while (len < *lenp) {
1510 if (get_user(c, p++))
1511 return -EFAULT;
1512 if (c == 0 || c == '\n')
1513 break;
1514 len++;
1515 }
f5dd3d6f
SV
1516 if (len >= maxlen)
1517 len = maxlen-1;
1518 if(copy_from_user(data, buffer, len))
1da177e4 1519 return -EFAULT;
f5dd3d6f 1520 ((char *) data)[len] = 0;
1da177e4
LT
1521 *ppos += *lenp;
1522 } else {
f5dd3d6f
SV
1523 len = strlen(data);
1524 if (len > maxlen)
1525 len = maxlen;
8d060877
ON
1526
1527 if (*ppos > len) {
1528 *lenp = 0;
1529 return 0;
1530 }
1531
1532 data += *ppos;
1533 len -= *ppos;
1534
1da177e4
LT
1535 if (len > *lenp)
1536 len = *lenp;
1537 if (len)
f5dd3d6f 1538 if(copy_to_user(buffer, data, len))
1da177e4
LT
1539 return -EFAULT;
1540 if (len < *lenp) {
1541 if(put_user('\n', ((char __user *) buffer) + len))
1542 return -EFAULT;
1543 len++;
1544 }
1545 *lenp = len;
1546 *ppos += len;
1547 }
1548 return 0;
1549}
1550
f5dd3d6f
SV
1551/**
1552 * proc_dostring - read a string sysctl
1553 * @table: the sysctl table
1554 * @write: %TRUE if this is a write to the sysctl file
1555 * @filp: the file structure
1556 * @buffer: the user buffer
1557 * @lenp: the size of the user buffer
1558 * @ppos: file position
1559 *
1560 * Reads/writes a string from/to the user buffer. If the kernel
1561 * buffer provided is not large enough to hold the string, the
1562 * string is truncated. The copied string is %NULL-terminated.
1563 * If the string is being read by the user process, it is copied
1564 * and a newline '\n' is added. It is truncated if the buffer is
1565 * not large enough.
1566 *
1567 * Returns 0 on success.
1568 */
1569int proc_dostring(ctl_table *table, int write, struct file *filp,
1570 void __user *buffer, size_t *lenp, loff_t *ppos)
1571{
1572 return _proc_do_string(table->data, table->maxlen, write, filp,
1573 buffer, lenp, ppos);
1574}
1575
1da177e4
LT
1576
1577static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1578 int *valp,
1579 int write, void *data)
1580{
1581 if (write) {
1582 *valp = *negp ? -*lvalp : *lvalp;
1583 } else {
1584 int val = *valp;
1585 if (val < 0) {
1586 *negp = -1;
1587 *lvalp = (unsigned long)-val;
1588 } else {
1589 *negp = 0;
1590 *lvalp = (unsigned long)val;
1591 }
1592 }
1593 return 0;
1594}
1595
fcfbd547
KK
1596static int __do_proc_dointvec(void *tbl_data, ctl_table *table,
1597 int write, struct file *filp, void __user *buffer,
1598 size_t *lenp, loff_t *ppos,
1da177e4
LT
1599 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1600 int write, void *data),
1601 void *data)
1602{
1603#define TMPBUFLEN 21
1604 int *i, vleft, first=1, neg, val;
1605 unsigned long lval;
1606 size_t left, len;
1607
1608 char buf[TMPBUFLEN], *p;
1609 char __user *s = buffer;
1610
fcfbd547 1611 if (!tbl_data || !table->maxlen || !*lenp ||
1da177e4
LT
1612 (*ppos && !write)) {
1613 *lenp = 0;
1614 return 0;
1615 }
1616
fcfbd547 1617 i = (int *) tbl_data;
1da177e4
LT
1618 vleft = table->maxlen / sizeof(*i);
1619 left = *lenp;
1620
1621 if (!conv)
1622 conv = do_proc_dointvec_conv;
1623
1624 for (; left && vleft--; i++, first=0) {
1625 if (write) {
1626 while (left) {
1627 char c;
1628 if (get_user(c, s))
1629 return -EFAULT;
1630 if (!isspace(c))
1631 break;
1632 left--;
1633 s++;
1634 }
1635 if (!left)
1636 break;
1637 neg = 0;
1638 len = left;
1639 if (len > sizeof(buf) - 1)
1640 len = sizeof(buf) - 1;
1641 if (copy_from_user(buf, s, len))
1642 return -EFAULT;
1643 buf[len] = 0;
1644 p = buf;
1645 if (*p == '-' && left > 1) {
1646 neg = 1;
bd9b0bac 1647 p++;
1da177e4
LT
1648 }
1649 if (*p < '0' || *p > '9')
1650 break;
1651
1652 lval = simple_strtoul(p, &p, 0);
1653
1654 len = p-buf;
1655 if ((len < left) && *p && !isspace(*p))
1656 break;
1657 if (neg)
1658 val = -val;
1659 s += len;
1660 left -= len;
1661
1662 if (conv(&neg, &lval, i, 1, data))
1663 break;
1664 } else {
1665 p = buf;
1666 if (!first)
1667 *p++ = '\t';
1668
1669 if (conv(&neg, &lval, i, 0, data))
1670 break;
1671
1672 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1673 len = strlen(buf);
1674 if (len > left)
1675 len = left;
1676 if(copy_to_user(s, buf, len))
1677 return -EFAULT;
1678 left -= len;
1679 s += len;
1680 }
1681 }
1682
1683 if (!write && !first && left) {
1684 if(put_user('\n', s))
1685 return -EFAULT;
1686 left--, s++;
1687 }
1688 if (write) {
1689 while (left) {
1690 char c;
1691 if (get_user(c, s++))
1692 return -EFAULT;
1693 if (!isspace(c))
1694 break;
1695 left--;
1696 }
1697 }
1698 if (write && first)
1699 return -EINVAL;
1700 *lenp -= left;
1701 *ppos += *lenp;
1702 return 0;
1703#undef TMPBUFLEN
1704}
1705
fcfbd547
KK
1706static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1707 void __user *buffer, size_t *lenp, loff_t *ppos,
1708 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1709 int write, void *data),
1710 void *data)
1711{
1712 return __do_proc_dointvec(table->data, table, write, filp,
1713 buffer, lenp, ppos, conv, data);
1714}
1715
1da177e4
LT
1716/**
1717 * proc_dointvec - read a vector of integers
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 * Returns 0 on success.
1729 */
1730int proc_dointvec(ctl_table *table, int write, struct file *filp,
1731 void __user *buffer, size_t *lenp, loff_t *ppos)
1732{
1733 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1734 NULL,NULL);
1735}
1736
1737#define OP_SET 0
1738#define OP_AND 1
34f5a398 1739#define OP_OR 2
1da177e4
LT
1740
1741static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1742 int *valp,
1743 int write, void *data)
1744{
1745 int op = *(int *)data;
1746 if (write) {
1747 int val = *negp ? -*lvalp : *lvalp;
1748 switch(op) {
1749 case OP_SET: *valp = val; break;
1750 case OP_AND: *valp &= val; break;
34f5a398 1751 case OP_OR: *valp |= val; break;
1da177e4
LT
1752 }
1753 } else {
1754 int val = *valp;
1755 if (val < 0) {
1756 *negp = -1;
1757 *lvalp = (unsigned long)-val;
1758 } else {
1759 *negp = 0;
1760 *lvalp = (unsigned long)val;
1761 }
1762 }
1763 return 0;
1764}
1765
1766/*
1767 * init may raise the set.
1768 */
1769
1770int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1771 void __user *buffer, size_t *lenp, loff_t *ppos)
1772{
1773 int op;
1774
6ff1b442 1775 if (write && !capable(CAP_SYS_MODULE)) {
1da177e4
LT
1776 return -EPERM;
1777 }
1778
f400e198 1779 op = is_init(current) ? OP_SET : OP_AND;
1da177e4
LT
1780 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1781 do_proc_dointvec_bset_conv,&op);
1782}
1783
34f5a398
TT
1784/*
1785 * Taint values can only be increased
1786 */
1787static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
1788 void __user *buffer, size_t *lenp, loff_t *ppos)
1789{
1790 int op;
1791
91fcd412 1792 if (write && !capable(CAP_SYS_ADMIN))
34f5a398
TT
1793 return -EPERM;
1794
1795 op = OP_OR;
1796 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1797 do_proc_dointvec_bset_conv,&op);
1798}
1799
1da177e4
LT
1800struct do_proc_dointvec_minmax_conv_param {
1801 int *min;
1802 int *max;
1803};
1804
1805static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1806 int *valp,
1807 int write, void *data)
1808{
1809 struct do_proc_dointvec_minmax_conv_param *param = data;
1810 if (write) {
1811 int val = *negp ? -*lvalp : *lvalp;
1812 if ((param->min && *param->min > val) ||
1813 (param->max && *param->max < val))
1814 return -EINVAL;
1815 *valp = val;
1816 } else {
1817 int val = *valp;
1818 if (val < 0) {
1819 *negp = -1;
1820 *lvalp = (unsigned long)-val;
1821 } else {
1822 *negp = 0;
1823 *lvalp = (unsigned long)val;
1824 }
1825 }
1826 return 0;
1827}
1828
1829/**
1830 * proc_dointvec_minmax - read a vector of integers with min/max values
1831 * @table: the sysctl table
1832 * @write: %TRUE if this is a write to the sysctl file
1833 * @filp: the file structure
1834 * @buffer: the user buffer
1835 * @lenp: the size of the user buffer
1836 * @ppos: file position
1837 *
1838 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1839 * values from/to the user buffer, treated as an ASCII string.
1840 *
1841 * This routine will ensure the values are within the range specified by
1842 * table->extra1 (min) and table->extra2 (max).
1843 *
1844 * Returns 0 on success.
1845 */
1846int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1847 void __user *buffer, size_t *lenp, loff_t *ppos)
1848{
1849 struct do_proc_dointvec_minmax_conv_param param = {
1850 .min = (int *) table->extra1,
1851 .max = (int *) table->extra2,
1852 };
1853 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1854 do_proc_dointvec_minmax_conv, &param);
1855}
1856
fcfbd547 1857static int __do_proc_doulongvec_minmax(void *data, ctl_table *table, int write,
1da177e4
LT
1858 struct file *filp,
1859 void __user *buffer,
1860 size_t *lenp, loff_t *ppos,
1861 unsigned long convmul,
1862 unsigned long convdiv)
1863{
1864#define TMPBUFLEN 21
1865 unsigned long *i, *min, *max, val;
1866 int vleft, first=1, neg;
1867 size_t len, left;
1868 char buf[TMPBUFLEN], *p;
1869 char __user *s = buffer;
1870
fcfbd547 1871 if (!data || !table->maxlen || !*lenp ||
1da177e4
LT
1872 (*ppos && !write)) {
1873 *lenp = 0;
1874 return 0;
1875 }
1876
fcfbd547 1877 i = (unsigned long *) data;
1da177e4
LT
1878 min = (unsigned long *) table->extra1;
1879 max = (unsigned long *) table->extra2;
1880 vleft = table->maxlen / sizeof(unsigned long);
1881 left = *lenp;
1882
1883 for (; left && vleft--; i++, min++, max++, first=0) {
1884 if (write) {
1885 while (left) {
1886 char c;
1887 if (get_user(c, s))
1888 return -EFAULT;
1889 if (!isspace(c))
1890 break;
1891 left--;
1892 s++;
1893 }
1894 if (!left)
1895 break;
1896 neg = 0;
1897 len = left;
1898 if (len > TMPBUFLEN-1)
1899 len = TMPBUFLEN-1;
1900 if (copy_from_user(buf, s, len))
1901 return -EFAULT;
1902 buf[len] = 0;
1903 p = buf;
1904 if (*p == '-' && left > 1) {
1905 neg = 1;
bd9b0bac 1906 p++;
1da177e4
LT
1907 }
1908 if (*p < '0' || *p > '9')
1909 break;
1910 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
1911 len = p-buf;
1912 if ((len < left) && *p && !isspace(*p))
1913 break;
1914 if (neg)
1915 val = -val;
1916 s += len;
1917 left -= len;
1918
1919 if(neg)
1920 continue;
1921 if ((min && val < *min) || (max && val > *max))
1922 continue;
1923 *i = val;
1924 } else {
1925 p = buf;
1926 if (!first)
1927 *p++ = '\t';
1928 sprintf(p, "%lu", convdiv * (*i) / convmul);
1929 len = strlen(buf);
1930 if (len > left)
1931 len = left;
1932 if(copy_to_user(s, buf, len))
1933 return -EFAULT;
1934 left -= len;
1935 s += len;
1936 }
1937 }
1938
1939 if (!write && !first && left) {
1940 if(put_user('\n', s))
1941 return -EFAULT;
1942 left--, s++;
1943 }
1944 if (write) {
1945 while (left) {
1946 char c;
1947 if (get_user(c, s++))
1948 return -EFAULT;
1949 if (!isspace(c))
1950 break;
1951 left--;
1952 }
1953 }
1954 if (write && first)
1955 return -EINVAL;
1956 *lenp -= left;
1957 *ppos += *lenp;
1958 return 0;
1959#undef TMPBUFLEN
1960}
1961
fcfbd547
KK
1962static int do_proc_doulongvec_minmax(ctl_table *table, int write,
1963 struct file *filp,
1964 void __user *buffer,
1965 size_t *lenp, loff_t *ppos,
1966 unsigned long convmul,
1967 unsigned long convdiv)
1968{
1969 return __do_proc_doulongvec_minmax(table->data, table, write,
1970 filp, buffer, lenp, ppos, convmul, convdiv);
1971}
1972
1da177e4
LT
1973/**
1974 * proc_doulongvec_minmax - read a vector of long integers with min/max values
1975 * @table: the sysctl table
1976 * @write: %TRUE if this is a write to the sysctl file
1977 * @filp: the file structure
1978 * @buffer: the user buffer
1979 * @lenp: the size of the user buffer
1980 * @ppos: file position
1981 *
1982 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1983 * values from/to the user buffer, treated as an ASCII string.
1984 *
1985 * This routine will ensure the values are within the range specified by
1986 * table->extra1 (min) and table->extra2 (max).
1987 *
1988 * Returns 0 on success.
1989 */
1990int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
1991 void __user *buffer, size_t *lenp, loff_t *ppos)
1992{
1993 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
1994}
1995
1996/**
1997 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
1998 * @table: the sysctl table
1999 * @write: %TRUE if this is a write to the sysctl file
2000 * @filp: the file structure
2001 * @buffer: the user buffer
2002 * @lenp: the size of the user buffer
2003 * @ppos: file position
2004 *
2005 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2006 * values from/to the user buffer, treated as an ASCII string. The values
2007 * are treated as milliseconds, and converted to jiffies when they are stored.
2008 *
2009 * This routine will ensure the values are within the range specified by
2010 * table->extra1 (min) and table->extra2 (max).
2011 *
2012 * Returns 0 on success.
2013 */
2014int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2015 struct file *filp,
2016 void __user *buffer,
2017 size_t *lenp, loff_t *ppos)
2018{
2019 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2020 lenp, ppos, HZ, 1000l);
2021}
2022
2023
2024static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2025 int *valp,
2026 int write, void *data)
2027{
2028 if (write) {
cba9f33d
BS
2029 if (*lvalp > LONG_MAX / HZ)
2030 return 1;
1da177e4
LT
2031 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2032 } else {
2033 int val = *valp;
2034 unsigned long lval;
2035 if (val < 0) {
2036 *negp = -1;
2037 lval = (unsigned long)-val;
2038 } else {
2039 *negp = 0;
2040 lval = (unsigned long)val;
2041 }
2042 *lvalp = lval / HZ;
2043 }
2044 return 0;
2045}
2046
2047static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2048 int *valp,
2049 int write, void *data)
2050{
2051 if (write) {
cba9f33d
BS
2052 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2053 return 1;
1da177e4
LT
2054 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2055 } else {
2056 int val = *valp;
2057 unsigned long lval;
2058 if (val < 0) {
2059 *negp = -1;
2060 lval = (unsigned long)-val;
2061 } else {
2062 *negp = 0;
2063 lval = (unsigned long)val;
2064 }
2065 *lvalp = jiffies_to_clock_t(lval);
2066 }
2067 return 0;
2068}
2069
2070static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2071 int *valp,
2072 int write, void *data)
2073{
2074 if (write) {
2075 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2076 } else {
2077 int val = *valp;
2078 unsigned long lval;
2079 if (val < 0) {
2080 *negp = -1;
2081 lval = (unsigned long)-val;
2082 } else {
2083 *negp = 0;
2084 lval = (unsigned long)val;
2085 }
2086 *lvalp = jiffies_to_msecs(lval);
2087 }
2088 return 0;
2089}
2090
2091/**
2092 * proc_dointvec_jiffies - read a vector of integers as seconds
2093 * @table: the sysctl table
2094 * @write: %TRUE if this is a write to the sysctl file
2095 * @filp: the file structure
2096 * @buffer: the user buffer
2097 * @lenp: the size of the user buffer
2098 * @ppos: file position
2099 *
2100 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2101 * values from/to the user buffer, treated as an ASCII string.
2102 * The values read are assumed to be in seconds, and are converted into
2103 * jiffies.
2104 *
2105 * Returns 0 on success.
2106 */
2107int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2108 void __user *buffer, size_t *lenp, loff_t *ppos)
2109{
2110 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2111 do_proc_dointvec_jiffies_conv,NULL);
2112}
2113
2114/**
2115 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2116 * @table: the sysctl table
2117 * @write: %TRUE if this is a write to the sysctl file
2118 * @filp: the file structure
2119 * @buffer: the user buffer
2120 * @lenp: the size of the user buffer
1e5d5331 2121 * @ppos: pointer to the file position
1da177e4
LT
2122 *
2123 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2124 * values from/to the user buffer, treated as an ASCII string.
2125 * The values read are assumed to be in 1/USER_HZ seconds, and
2126 * are converted into jiffies.
2127 *
2128 * Returns 0 on success.
2129 */
2130int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2131 void __user *buffer, size_t *lenp, loff_t *ppos)
2132{
2133 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2134 do_proc_dointvec_userhz_jiffies_conv,NULL);
2135}
2136
2137/**
2138 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2139 * @table: the sysctl table
2140 * @write: %TRUE if this is a write to the sysctl file
2141 * @filp: the file structure
2142 * @buffer: the user buffer
2143 * @lenp: the size of the user buffer
67be2dd1
MW
2144 * @ppos: file position
2145 * @ppos: the current position in the file
1da177e4
LT
2146 *
2147 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2148 * values from/to the user buffer, treated as an ASCII string.
2149 * The values read are assumed to be in 1/1000 seconds, and
2150 * are converted into jiffies.
2151 *
2152 * Returns 0 on success.
2153 */
2154int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2155 void __user *buffer, size_t *lenp, loff_t *ppos)
2156{
2157 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2158 do_proc_dointvec_ms_jiffies_conv, NULL);
2159}
2160
9ec52099
CLG
2161static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
2162 void __user *buffer, size_t *lenp, loff_t *ppos)
2163{
2164 struct pid *new_pid;
2165 pid_t tmp;
2166 int r;
2167
2168 tmp = pid_nr(cad_pid);
2169
2170 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2171 lenp, ppos, NULL, NULL);
2172 if (r || !write)
2173 return r;
2174
2175 new_pid = find_get_pid(tmp);
2176 if (!new_pid)
2177 return -ESRCH;
2178
2179 put_pid(xchg(&cad_pid, new_pid));
2180 return 0;
2181}
2182
1da177e4
LT
2183#else /* CONFIG_PROC_FS */
2184
2185int proc_dostring(ctl_table *table, int write, struct file *filp,
2186 void __user *buffer, size_t *lenp, loff_t *ppos)
2187{
2188 return -ENOSYS;
2189}
2190
1da177e4
LT
2191int proc_dointvec(ctl_table *table, int write, struct file *filp,
2192 void __user *buffer, size_t *lenp, loff_t *ppos)
2193{
2194 return -ENOSYS;
2195}
2196
2197int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2198 void __user *buffer, size_t *lenp, loff_t *ppos)
2199{
2200 return -ENOSYS;
2201}
2202
2203int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2204 void __user *buffer, size_t *lenp, loff_t *ppos)
2205{
2206 return -ENOSYS;
2207}
2208
2209int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2210 void __user *buffer, size_t *lenp, loff_t *ppos)
2211{
2212 return -ENOSYS;
2213}
2214
2215int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2216 void __user *buffer, size_t *lenp, loff_t *ppos)
2217{
2218 return -ENOSYS;
2219}
2220
2221int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2222 void __user *buffer, size_t *lenp, loff_t *ppos)
2223{
2224 return -ENOSYS;
2225}
2226
2227int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2228 void __user *buffer, size_t *lenp, loff_t *ppos)
2229{
2230 return -ENOSYS;
2231}
2232
2233int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2234 struct file *filp,
2235 void __user *buffer,
2236 size_t *lenp, loff_t *ppos)
2237{
2238 return -ENOSYS;
2239}
2240
2241
2242#endif /* CONFIG_PROC_FS */
2243
2244
b89a8171 2245#ifdef CONFIG_SYSCTL_SYSCALL
1da177e4
LT
2246/*
2247 * General sysctl support routines
2248 */
2249
2250/* The generic string strategy routine: */
2251int sysctl_string(ctl_table *table, int __user *name, int nlen,
2252 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2253 void __user *newval, size_t newlen)
1da177e4 2254{
1da177e4
LT
2255 if (!table->data || !table->maxlen)
2256 return -ENOTDIR;
2257
2258 if (oldval && oldlenp) {
de9e007d
LT
2259 size_t bufsize;
2260 if (get_user(bufsize, oldlenp))
1da177e4 2261 return -EFAULT;
de9e007d
LT
2262 if (bufsize) {
2263 size_t len = strlen(table->data), copied;
2264
2265 /* This shouldn't trigger for a well-formed sysctl */
2266 if (len > table->maxlen)
1da177e4 2267 len = table->maxlen;
de9e007d
LT
2268
2269 /* Copy up to a max of bufsize-1 bytes of the string */
2270 copied = (len >= bufsize) ? bufsize - 1 : len;
2271
2272 if (copy_to_user(oldval, table->data, copied) ||
2273 put_user(0, (char __user *)(oldval + copied)))
1da177e4 2274 return -EFAULT;
de9e007d 2275 if (put_user(len, oldlenp))
1da177e4
LT
2276 return -EFAULT;
2277 }
2278 }
2279 if (newval && newlen) {
de9e007d 2280 size_t len = newlen;
1da177e4
LT
2281 if (len > table->maxlen)
2282 len = table->maxlen;
2283 if(copy_from_user(table->data, newval, len))
2284 return -EFAULT;
2285 if (len == table->maxlen)
2286 len--;
2287 ((char *) table->data)[len] = 0;
2288 }
82c9df82 2289 return 1;
1da177e4
LT
2290}
2291
2292/*
2293 * This function makes sure that all of the integers in the vector
2294 * are between the minimum and maximum values given in the arrays
2295 * table->extra1 and table->extra2, respectively.
2296 */
2297int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2298 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2299 void __user *newval, size_t newlen)
1da177e4
LT
2300{
2301
2302 if (newval && newlen) {
2303 int __user *vec = (int __user *) newval;
2304 int *min = (int *) table->extra1;
2305 int *max = (int *) table->extra2;
2306 size_t length;
2307 int i;
2308
2309 if (newlen % sizeof(int) != 0)
2310 return -EINVAL;
2311
2312 if (!table->extra1 && !table->extra2)
2313 return 0;
2314
2315 if (newlen > table->maxlen)
2316 newlen = table->maxlen;
2317 length = newlen / sizeof(int);
2318
2319 for (i = 0; i < length; i++) {
2320 int value;
2321 if (get_user(value, vec + i))
2322 return -EFAULT;
2323 if (min && value < min[i])
2324 return -EINVAL;
2325 if (max && value > max[i])
2326 return -EINVAL;
2327 }
2328 }
2329 return 0;
2330}
2331
2332/* Strategy function to convert jiffies to seconds */
2333int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2334 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2335 void __user *newval, size_t newlen)
1da177e4 2336{
3ee75ac3 2337 if (oldval && oldlenp) {
1da177e4 2338 size_t olen;
3ee75ac3
AD
2339
2340 if (get_user(olen, oldlenp))
2341 return -EFAULT;
2342 if (olen) {
2343 int val;
2344
2345 if (olen < sizeof(int))
2346 return -EINVAL;
2347
2348 val = *(int *)(table->data) / HZ;
2349 if (put_user(val, (int __user *)oldval))
2350 return -EFAULT;
2351 if (put_user(sizeof(int), oldlenp))
1da177e4 2352 return -EFAULT;
1da177e4 2353 }
1da177e4
LT
2354 }
2355 if (newval && newlen) {
2356 int new;
2357 if (newlen != sizeof(int))
2358 return -EINVAL;
2359 if (get_user(new, (int __user *)newval))
2360 return -EFAULT;
2361 *(int *)(table->data) = new*HZ;
2362 }
2363 return 1;
2364}
2365
2366/* Strategy function to convert jiffies to seconds */
2367int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2368 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2369 void __user *newval, size_t newlen)
1da177e4 2370{
3ee75ac3 2371 if (oldval && oldlenp) {
1da177e4 2372 size_t olen;
3ee75ac3
AD
2373
2374 if (get_user(olen, oldlenp))
2375 return -EFAULT;
2376 if (olen) {
2377 int val;
2378
2379 if (olen < sizeof(int))
2380 return -EINVAL;
2381
2382 val = jiffies_to_msecs(*(int *)(table->data));
2383 if (put_user(val, (int __user *)oldval))
2384 return -EFAULT;
2385 if (put_user(sizeof(int), oldlenp))
1da177e4 2386 return -EFAULT;
1da177e4 2387 }
1da177e4
LT
2388 }
2389 if (newval && newlen) {
2390 int new;
2391 if (newlen != sizeof(int))
2392 return -EINVAL;
2393 if (get_user(new, (int __user *)newval))
2394 return -EFAULT;
2395 *(int *)(table->data) = msecs_to_jiffies(new);
2396 }
2397 return 1;
2398}
2399
c4b8b769 2400
c4b8b769 2401
b89a8171 2402#else /* CONFIG_SYSCTL_SYSCALL */
1da177e4
LT
2403
2404
2405asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2406{
b89a8171 2407 static int msg_count;
0e009be8
EB
2408 struct __sysctl_args tmp;
2409 int name[CTL_MAXNAME];
2410 int i;
2411
2412 /* Read in the sysctl name for better debug message logging */
2413 if (copy_from_user(&tmp, args, sizeof(tmp)))
2414 return -EFAULT;
2415 if (tmp.nlen <= 0 || tmp.nlen >= CTL_MAXNAME)
2416 return -ENOTDIR;
2417 for (i = 0; i < tmp.nlen; i++)
2418 if (get_user(name[i], tmp.name + i))
2419 return -EFAULT;
2420
2421 /* Ignore accesses to kernel.version */
2422 if ((tmp.nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2423 goto out;
b89a8171
EB
2424
2425 if (msg_count < 5) {
2426 msg_count++;
2427 printk(KERN_INFO
2428 "warning: process `%s' used the removed sysctl "
0e009be8
EB
2429 "system call with ", current->comm);
2430 for (i = 0; i < tmp.nlen; i++)
2431 printk("%d.", name[i]);
2432 printk("\n");
b89a8171 2433 }
0e009be8 2434out:
1da177e4
LT
2435 return -ENOSYS;
2436}
2437
2438int sysctl_string(ctl_table *table, int __user *name, int nlen,
2439 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2440 void __user *newval, size_t newlen)
1da177e4
LT
2441{
2442 return -ENOSYS;
2443}
2444
2445int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2446 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2447 void __user *newval, size_t newlen)
1da177e4
LT
2448{
2449 return -ENOSYS;
2450}
2451
2452int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2453 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2454 void __user *newval, size_t newlen)
1da177e4
LT
2455{
2456 return -ENOSYS;
2457}
2458
2459int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2460 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2461 void __user *newval, size_t newlen)
1da177e4
LT
2462{
2463 return -ENOSYS;
2464}
2465
b89a8171 2466#endif /* CONFIG_SYSCTL_SYSCALL */
1da177e4
LT
2467
2468/*
2469 * No sense putting this after each symbol definition, twice,
2470 * exception granted :-)
2471 */
2472EXPORT_SYMBOL(proc_dointvec);
2473EXPORT_SYMBOL(proc_dointvec_jiffies);
2474EXPORT_SYMBOL(proc_dointvec_minmax);
2475EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2476EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2477EXPORT_SYMBOL(proc_dostring);
2478EXPORT_SYMBOL(proc_doulongvec_minmax);
2479EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2480EXPORT_SYMBOL(register_sysctl_table);
2481EXPORT_SYMBOL(sysctl_intvec);
2482EXPORT_SYMBOL(sysctl_jiffies);
2483EXPORT_SYMBOL(sysctl_ms_jiffies);
2484EXPORT_SYMBOL(sysctl_string);
2485EXPORT_SYMBOL(unregister_sysctl_table);