]> 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
[PATCH] do_no_pfn()
[net-next-2.6.git] / kernel / sysctl.c
CommitLineData
1da177e4
LT		 1/*
2 * sysctl.c: General linux system control interface
3 *
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12 * Horn.
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16 * Wendling.
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19 */
20
1da177e4
LT		 21#include <linux/module.h>
22#include <linux/mm.h>
23#include <linux/swap.h>
24#include <linux/slab.h>
25#include <linux/sysctl.h>
26#include <linux/proc_fs.h>
c59ede7b 27#include <linux/capability.h>
1da177e4
LT		 28#include <linux/ctype.h>
29#include <linux/utsname.h>
30#include <linux/capability.h>
31#include <linux/smp_lock.h>
32#include <linux/init.h>
33#include <linux/kernel.h>
0296b228 34#include <linux/kobject.h>
20380731 35#include <linux/net.h>
1da177e4
LT		 36#include <linux/sysrq.h>
37#include <linux/highuid.h>
38#include <linux/writeback.h>
39#include <linux/hugetlb.h>
40#include <linux/security.h>
41#include <linux/initrd.h>
42#include <linux/times.h>
43#include <linux/limits.h>
44#include <linux/dcache.h>
45#include <linux/syscalls.h>
c255d844
PM		 46#include <linux/nfs_fs.h>
47#include <linux/acpi.h>
1da177e4
LT		 48
49#include <asm/uaccess.h>
50#include <asm/processor.h>
51
529bf6be
DS		 52extern int proc_nr_files(ctl_table *table, int write, struct file *filp,
53 void __user *buffer, size_t *lenp, loff_t *ppos);
54
1da177e4
LT		 55#if defined(CONFIG_SYSCTL)
56
57/* External variables not in a header file. */
58extern int C_A_D;
59extern int sysctl_overcommit_memory;
60extern int sysctl_overcommit_ratio;
fadd8fbd 61extern int sysctl_panic_on_oom;
1da177e4
LT		 62extern int max_threads;
63extern int sysrq_enabled;
64extern int core_uses_pid;
d6e71144 65extern int suid_dumpable;
1da177e4
LT		 66extern char core_pattern[];
67extern int cad_pid;
68extern int pid_max;
69extern int min_free_kbytes;
70extern int printk_ratelimit_jiffies;
71extern int printk_ratelimit_burst;
72extern int pid_max_min, pid_max_max;
9d0243bc 73extern int sysctl_drop_caches;
8ad4b1fb 74extern int percpu_pagelist_fraction;
bebfa101 75extern int compat_log;
1da177e4
LT		 76
77#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
78int unknown_nmi_panic;
407984f1
DZ		 79int nmi_watchdog_enabled;
80extern int proc_nmi_enabled(struct ctl_table *, int , struct file *,
81 void __user *, size_t *, loff_t *);
1da177e4
LT		 82#endif
83
84/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
85static int maxolduid = 65535;
86static int minolduid;
8ad4b1fb 87static int min_percpu_pagelist_fract = 8;
1da177e4
LT		 88
89static int ngroups_max = NGROUPS_MAX;
90
91#ifdef CONFIG_KMOD
92extern char modprobe_path[];
93#endif
1da177e4
LT		 94#ifdef CONFIG_CHR_DEV_SG
95extern int sg_big_buff;
96#endif
97#ifdef CONFIG_SYSVIPC
98extern size_t shm_ctlmax;
99extern size_t shm_ctlall;
100extern int shm_ctlmni;
101extern int msg_ctlmax;
102extern int msg_ctlmnb;
103extern int msg_ctlmni;
104extern int sem_ctls[];
105#endif
106
107#ifdef __sparc__
108extern char reboot_command [];
109extern int stop_a_enabled;
110extern int scons_pwroff;
111#endif
112
113#ifdef __hppa__
114extern int pwrsw_enabled;
115extern int unaligned_enabled;
116#endif
117
347a8dc3 118#ifdef CONFIG_S390
1da177e4
LT		 119#ifdef CONFIG_MATHEMU
120extern int sysctl_ieee_emulation_warnings;
121#endif
122extern int sysctl_userprocess_debug;
951f22d5 123extern int spin_retry;
1da177e4
LT		 124#endif
125
126extern int sysctl_hz_timer;
127
128#ifdef CONFIG_BSD_PROCESS_ACCT
129extern int acct_parm[];
130#endif
131
d2b176ed
JS		 132#ifdef CONFIG_IA64
133extern int no_unaligned_warning;
134#endif
135
23f78d4a
IM		 136#ifdef CONFIG_RT_MUTEXES
137extern int max_lock_depth;
138#endif
139
1da177e4
LT		 140static int parse_table(int __user *, int, void __user *, size_t __user *, void __user *, size_t,
141 ctl_table *, void **);
142static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
143 void __user *buffer, size_t *lenp, loff_t *ppos);
144
145static ctl_table root_table[];
146static struct ctl_table_header root_table_header =
147 { root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
148
149static ctl_table kern_table[];
150static ctl_table vm_table[];
1da177e4
LT		 151static ctl_table fs_table[];
152static ctl_table debug_table[];
153static ctl_table dev_table[];
154extern ctl_table random_table[];
155#ifdef CONFIG_UNIX98_PTYS
156extern ctl_table pty_table[];
157#endif
2d9048e2 158#ifdef CONFIG_INOTIFY_USER
0399cb08
RL		 159extern ctl_table inotify_table[];
160#endif
1da177e4
LT		 161
162#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
163int sysctl_legacy_va_layout;
164#endif
165
166/* /proc declarations: */
167
168#ifdef CONFIG_PROC_FS
169
170static ssize_t proc_readsys(struct file *, char __user *, size_t, loff_t *);
171static ssize_t proc_writesys(struct file *, const char __user *, size_t, loff_t *);
172static int proc_opensys(struct inode *, struct file *);
173
174struct file_operations proc_sys_file_operations = {
175 .open = proc_opensys,
176 .read = proc_readsys,
177 .write = proc_writesys,
178};
179
180extern struct proc_dir_entry *proc_sys_root;
181
330d57fb 182static void register_proc_table(ctl_table *, struct proc_dir_entry *, void *);
1da177e4
LT		 183static void unregister_proc_table(ctl_table *, struct proc_dir_entry *);
184#endif
185
186/* The default sysctl tables: */
187
188static ctl_table root_table[] = {
189 {
190 .ctl_name = CTL_KERN,
191 .procname = "kernel",
192 .mode = 0555,
193 .child = kern_table,
194 },
195 {
196 .ctl_name = CTL_VM,
197 .procname = "vm",
198 .mode = 0555,
199 .child = vm_table,
200 },
201#ifdef CONFIG_NET
202 {
203 .ctl_name = CTL_NET,
204 .procname = "net",
205 .mode = 0555,
206 .child = net_table,
207 },
208#endif
1da177e4
LT		 209 {
210 .ctl_name = CTL_FS,
211 .procname = "fs",
212 .mode = 0555,
213 .child = fs_table,
214 },
215 {
216 .ctl_name = CTL_DEBUG,
217 .procname = "debug",
218 .mode = 0555,
219 .child = debug_table,
220 },
221 {
222 .ctl_name = CTL_DEV,
223 .procname = "dev",
224 .mode = 0555,
225 .child = dev_table,
226 },
0eeca283 227
1da177e4
LT		 228 { .ctl_name = 0 }
229};
230
231static ctl_table kern_table[] = {
232 {
233 .ctl_name = KERN_OSTYPE,
234 .procname = "ostype",
235 .data = system_utsname.sysname,
236 .maxlen = sizeof(system_utsname.sysname),
237 .mode = 0444,
238 .proc_handler = &proc_doutsstring,
239 .strategy = &sysctl_string,
240 },
241 {
242 .ctl_name = KERN_OSRELEASE,
243 .procname = "osrelease",
244 .data = system_utsname.release,
245 .maxlen = sizeof(system_utsname.release),
246 .mode = 0444,
247 .proc_handler = &proc_doutsstring,
248 .strategy = &sysctl_string,
249 },
250 {
251 .ctl_name = KERN_VERSION,
252 .procname = "version",
253 .data = system_utsname.version,
254 .maxlen = sizeof(system_utsname.version),
255 .mode = 0444,
256 .proc_handler = &proc_doutsstring,
257 .strategy = &sysctl_string,
258 },
259 {
260 .ctl_name = KERN_NODENAME,
261 .procname = "hostname",
262 .data = system_utsname.nodename,
263 .maxlen = sizeof(system_utsname.nodename),
264 .mode = 0644,
265 .proc_handler = &proc_doutsstring,
266 .strategy = &sysctl_string,
267 },
268 {
269 .ctl_name = KERN_DOMAINNAME,
270 .procname = "domainname",
271 .data = system_utsname.domainname,
272 .maxlen = sizeof(system_utsname.domainname),
273 .mode = 0644,
274 .proc_handler = &proc_doutsstring,
275 .strategy = &sysctl_string,
276 },
277 {
278 .ctl_name = KERN_PANIC,
279 .procname = "panic",
280 .data = &panic_timeout,
281 .maxlen = sizeof(int),
282 .mode = 0644,
283 .proc_handler = &proc_dointvec,
284 },
285 {
286 .ctl_name = KERN_CORE_USES_PID,
287 .procname = "core_uses_pid",
288 .data = &core_uses_pid,
289 .maxlen = sizeof(int),
290 .mode = 0644,
291 .proc_handler = &proc_dointvec,
292 },
293 {
294 .ctl_name = KERN_CORE_PATTERN,
295 .procname = "core_pattern",
296 .data = core_pattern,
297 .maxlen = 64,
298 .mode = 0644,
299 .proc_handler = &proc_dostring,
300 .strategy = &sysctl_string,
301 },
302 {
303 .ctl_name = KERN_TAINTED,
304 .procname = "tainted",
305 .data = &tainted,
306 .maxlen = sizeof(int),
307 .mode = 0444,
308 .proc_handler = &proc_dointvec,
309 },
310 {
311 .ctl_name = KERN_CAP_BSET,
312 .procname = "cap-bound",
313 .data = &cap_bset,
314 .maxlen = sizeof(kernel_cap_t),
315 .mode = 0600,
316 .proc_handler = &proc_dointvec_bset,
317 },
318#ifdef CONFIG_BLK_DEV_INITRD
319 {
320 .ctl_name = KERN_REALROOTDEV,
321 .procname = "real-root-dev",
322 .data = &real_root_dev,
323 .maxlen = sizeof(int),
324 .mode = 0644,
325 .proc_handler = &proc_dointvec,
326 },
327#endif
328#ifdef __sparc__
329 {
330 .ctl_name = KERN_SPARC_REBOOT,
331 .procname = "reboot-cmd",
332 .data = reboot_command,
333 .maxlen = 256,
334 .mode = 0644,
335 .proc_handler = &proc_dostring,
336 .strategy = &sysctl_string,
337 },
338 {
339 .ctl_name = KERN_SPARC_STOP_A,
340 .procname = "stop-a",
341 .data = &stop_a_enabled,
342 .maxlen = sizeof (int),
343 .mode = 0644,
344 .proc_handler = &proc_dointvec,
345 },
346 {
347 .ctl_name = KERN_SPARC_SCONS_PWROFF,
348 .procname = "scons-poweroff",
349 .data = &scons_pwroff,
350 .maxlen = sizeof (int),
351 .mode = 0644,
352 .proc_handler = &proc_dointvec,
353 },
354#endif
355#ifdef __hppa__
356 {
357 .ctl_name = KERN_HPPA_PWRSW,
358 .procname = "soft-power",
359 .data = &pwrsw_enabled,
360 .maxlen = sizeof (int),
361 .mode = 0644,
362 .proc_handler = &proc_dointvec,
363 },
364 {
365 .ctl_name = KERN_HPPA_UNALIGNED,
366 .procname = "unaligned-trap",
367 .data = &unaligned_enabled,
368 .maxlen = sizeof (int),
369 .mode = 0644,
370 .proc_handler = &proc_dointvec,
371 },
372#endif
373 {
374 .ctl_name = KERN_CTLALTDEL,
375 .procname = "ctrl-alt-del",
376 .data = &C_A_D,
377 .maxlen = sizeof(int),
378 .mode = 0644,
379 .proc_handler = &proc_dointvec,
380 },
381 {
382 .ctl_name = KERN_PRINTK,
383 .procname = "printk",
384 .data = &console_loglevel,
385 .maxlen = 4*sizeof(int),
386 .mode = 0644,
387 .proc_handler = &proc_dointvec,
388 },
389#ifdef CONFIG_KMOD
390 {
391 .ctl_name = KERN_MODPROBE,
392 .procname = "modprobe",
393 .data = &modprobe_path,
394 .maxlen = KMOD_PATH_LEN,
395 .mode = 0644,
396 .proc_handler = &proc_dostring,
397 .strategy = &sysctl_string,
398 },
399#endif
57ae2508 400#if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
1da177e4
LT		 401 {
402 .ctl_name = KERN_HOTPLUG,
403 .procname = "hotplug",
312c004d
KS		 404 .data = &uevent_helper,
405 .maxlen = UEVENT_HELPER_PATH_LEN,
1da177e4
LT		 406 .mode = 0644,
407 .proc_handler = &proc_dostring,
408 .strategy = &sysctl_string,
409 },
410#endif
411#ifdef CONFIG_CHR_DEV_SG
412 {
413 .ctl_name = KERN_SG_BIG_BUFF,
414 .procname = "sg-big-buff",
415 .data = &sg_big_buff,
416 .maxlen = sizeof (int),
417 .mode = 0444,
418 .proc_handler = &proc_dointvec,
419 },
420#endif
421#ifdef CONFIG_BSD_PROCESS_ACCT
422 {
423 .ctl_name = KERN_ACCT,
424 .procname = "acct",
425 .data = &acct_parm,
426 .maxlen = 3*sizeof(int),
427 .mode = 0644,
428 .proc_handler = &proc_dointvec,
429 },
430#endif
431#ifdef CONFIG_SYSVIPC
432 {
433 .ctl_name = KERN_SHMMAX,
434 .procname = "shmmax",
435 .data = &shm_ctlmax,
436 .maxlen = sizeof (size_t),
437 .mode = 0644,
438 .proc_handler = &proc_doulongvec_minmax,
439 },
440 {
441 .ctl_name = KERN_SHMALL,
442 .procname = "shmall",
443 .data = &shm_ctlall,
444 .maxlen = sizeof (size_t),
445 .mode = 0644,
446 .proc_handler = &proc_doulongvec_minmax,
447 },
448 {
449 .ctl_name = KERN_SHMMNI,
450 .procname = "shmmni",
451 .data = &shm_ctlmni,
452 .maxlen = sizeof (int),
453 .mode = 0644,
454 .proc_handler = &proc_dointvec,
455 },
456 {
457 .ctl_name = KERN_MSGMAX,
458 .procname = "msgmax",
459 .data = &msg_ctlmax,
460 .maxlen = sizeof (int),
461 .mode = 0644,
462 .proc_handler = &proc_dointvec,
463 },
464 {
465 .ctl_name = KERN_MSGMNI,
466 .procname = "msgmni",
467 .data = &msg_ctlmni,
468 .maxlen = sizeof (int),
469 .mode = 0644,
470 .proc_handler = &proc_dointvec,
471 },
472 {
473 .ctl_name = KERN_MSGMNB,
474 .procname = "msgmnb",
475 .data = &msg_ctlmnb,
476 .maxlen = sizeof (int),
477 .mode = 0644,
478 .proc_handler = &proc_dointvec,
479 },
480 {
481 .ctl_name = KERN_SEM,
482 .procname = "sem",
483 .data = &sem_ctls,
484 .maxlen = 4*sizeof (int),
485 .mode = 0644,
486 .proc_handler = &proc_dointvec,
487 },
488#endif
489#ifdef CONFIG_MAGIC_SYSRQ
490 {
491 .ctl_name = KERN_SYSRQ,
492 .procname = "sysrq",
493 .data = &sysrq_enabled,
494 .maxlen = sizeof (int),
495 .mode = 0644,
496 .proc_handler = &proc_dointvec,
497 },
498#endif
499 {
500 .ctl_name = KERN_CADPID,
501 .procname = "cad_pid",
502 .data = &cad_pid,
503 .maxlen = sizeof (int),
504 .mode = 0600,
505 .proc_handler = &proc_dointvec,
506 },
507 {
508 .ctl_name = KERN_MAX_THREADS,
509 .procname = "threads-max",
510 .data = &max_threads,
511 .maxlen = sizeof(int),
512 .mode = 0644,
513 .proc_handler = &proc_dointvec,
514 },
515 {
516 .ctl_name = KERN_RANDOM,
517 .procname = "random",
518 .mode = 0555,
519 .child = random_table,
520 },
521#ifdef CONFIG_UNIX98_PTYS
522 {
523 .ctl_name = KERN_PTY,
524 .procname = "pty",
525 .mode = 0555,
526 .child = pty_table,
527 },
528#endif
529 {
530 .ctl_name = KERN_OVERFLOWUID,
531 .procname = "overflowuid",
532 .data = &overflowuid,
533 .maxlen = sizeof(int),
534 .mode = 0644,
535 .proc_handler = &proc_dointvec_minmax,
536 .strategy = &sysctl_intvec,
537 .extra1 = &minolduid,
538 .extra2 = &maxolduid,
539 },
540 {
541 .ctl_name = KERN_OVERFLOWGID,
542 .procname = "overflowgid",
543 .data = &overflowgid,
544 .maxlen = sizeof(int),
545 .mode = 0644,
546 .proc_handler = &proc_dointvec_minmax,
547 .strategy = &sysctl_intvec,
548 .extra1 = &minolduid,
549 .extra2 = &maxolduid,
550 },
347a8dc3 551#ifdef CONFIG_S390
1da177e4
LT		 552#ifdef CONFIG_MATHEMU
553 {
554 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
555 .procname = "ieee_emulation_warnings",
556 .data = &sysctl_ieee_emulation_warnings,
557 .maxlen = sizeof(int),
558 .mode = 0644,
559 .proc_handler = &proc_dointvec,
560 },
561#endif
562#ifdef CONFIG_NO_IDLE_HZ
563 {
564 .ctl_name = KERN_HZ_TIMER,
565 .procname = "hz_timer",
566 .data = &sysctl_hz_timer,
567 .maxlen = sizeof(int),
568 .mode = 0644,
569 .proc_handler = &proc_dointvec,
570 },
571#endif
572 {
573 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
574 .procname = "userprocess_debug",
575 .data = &sysctl_userprocess_debug,
576 .maxlen = sizeof(int),
577 .mode = 0644,
578 .proc_handler = &proc_dointvec,
579 },
580#endif
581 {
582 .ctl_name = KERN_PIDMAX,
583 .procname = "pid_max",
584 .data = &pid_max,
585 .maxlen = sizeof (int),
586 .mode = 0644,
587 .proc_handler = &proc_dointvec_minmax,
588 .strategy = sysctl_intvec,
589 .extra1 = &pid_max_min,
590 .extra2 = &pid_max_max,
591 },
592 {
593 .ctl_name = KERN_PANIC_ON_OOPS,
594 .procname = "panic_on_oops",
595 .data = &panic_on_oops,
596 .maxlen = sizeof(int),
597 .mode = 0644,
598 .proc_handler = &proc_dointvec,
599 },
600 {
601 .ctl_name = KERN_PRINTK_RATELIMIT,
602 .procname = "printk_ratelimit",
603 .data = &printk_ratelimit_jiffies,
604 .maxlen = sizeof(int),
605 .mode = 0644,
606 .proc_handler = &proc_dointvec_jiffies,
607 .strategy = &sysctl_jiffies,
608 },
609 {
610 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
611 .procname = "printk_ratelimit_burst",
612 .data = &printk_ratelimit_burst,
613 .maxlen = sizeof(int),
614 .mode = 0644,
615 .proc_handler = &proc_dointvec,
616 },
617 {
618 .ctl_name = KERN_NGROUPS_MAX,
619 .procname = "ngroups_max",
620 .data = &ngroups_max,
621 .maxlen = sizeof (int),
622 .mode = 0444,
623 .proc_handler = &proc_dointvec,
624 },
625#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
626 {
627 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
628 .procname = "unknown_nmi_panic",
629 .data = &unknown_nmi_panic,
630 .maxlen = sizeof (int),
631 .mode = 0644,
2fbe7b25 632 .proc_handler = &proc_dointvec,
1da177e4 633 },
407984f1
DZ		 634 {
635 .ctl_name = KERN_NMI_WATCHDOG,
636 .procname = "nmi_watchdog",
637 .data = &nmi_watchdog_enabled,
638 .maxlen = sizeof (int),
639 .mode = 0644,
640 .proc_handler = &proc_nmi_enabled,
1da177e4
LT		 641 },
642#endif
643#if defined(CONFIG_X86)
8da5adda
DZ		 644 {
645 .ctl_name = KERN_PANIC_ON_NMI,
646 .procname = "panic_on_unrecovered_nmi",
647 .data = &panic_on_unrecovered_nmi,
648 .maxlen = sizeof(int),
649 .mode = 0644,
650 .proc_handler = &proc_dointvec,
651 },
1da177e4
LT		 652 {
653 .ctl_name = KERN_BOOTLOADER_TYPE,
654 .procname = "bootloader_type",
655 .data = &bootloader_type,
656 .maxlen = sizeof (int),
657 .mode = 0444,
658 .proc_handler = &proc_dointvec,
659 },
660#endif
7a9166e3 661#if defined(CONFIG_MMU)
1da177e4
LT		 662 {
663 .ctl_name = KERN_RANDOMIZE,
664 .procname = "randomize_va_space",
665 .data = &randomize_va_space,
666 .maxlen = sizeof(int),
667 .mode = 0644,
668 .proc_handler = &proc_dointvec,
669 },
7a9166e3 670#endif
0152fb37 671#if defined(CONFIG_S390) && defined(CONFIG_SMP)
951f22d5
MS		 672 {
673 .ctl_name = KERN_SPIN_RETRY,
674 .procname = "spin_retry",
675 .data = &spin_retry,
676 .maxlen = sizeof (int),
677 .mode = 0644,
678 .proc_handler = &proc_dointvec,
679 },
c255d844
PM		 680#endif
681#ifdef CONFIG_ACPI_SLEEP
682 {
683 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
684 .procname = "acpi_video_flags",
685 .data = &acpi_video_flags,
686 .maxlen = sizeof (unsigned long),
687 .mode = 0644,
7f99f06f 688 .proc_handler = &proc_doulongvec_minmax,
c255d844 689 },
d2b176ed
JS		 690#endif
691#ifdef CONFIG_IA64
692 {
693 .ctl_name = KERN_IA64_UNALIGNED,
694 .procname = "ignore-unaligned-usertrap",
695 .data = &no_unaligned_warning,
696 .maxlen = sizeof (int),
697 .mode = 0644,
698 .proc_handler = &proc_dointvec,
699 },
bebfa101
AK		 700#endif
701#ifdef CONFIG_COMPAT
702 {
703 .ctl_name = KERN_COMPAT_LOG,
704 .procname = "compat-log",
705 .data = &compat_log,
706 .maxlen = sizeof (int),
707 .mode = 0644,
708 .proc_handler = &proc_dointvec,
709 },
951f22d5 710#endif
23f78d4a
IM		 711#ifdef CONFIG_RT_MUTEXES
712 {
713 .ctl_name = KERN_MAX_LOCK_DEPTH,
714 .procname = "max_lock_depth",
715 .data = &max_lock_depth,
716 .maxlen = sizeof(int),
717 .mode = 0644,
718 .proc_handler = &proc_dointvec,
719 },
720#endif
721
1da177e4
LT		 722 { .ctl_name = 0 }
723};
724
725/* Constants for minimum and maximum testing in vm_table.
726 We use these as one-element integer vectors. */
727static int zero;
728static int one_hundred = 100;
729
730
731static ctl_table vm_table[] = {
732 {
733 .ctl_name = VM_OVERCOMMIT_MEMORY,
734 .procname = "overcommit_memory",
735 .data = &sysctl_overcommit_memory,
736 .maxlen = sizeof(sysctl_overcommit_memory),
737 .mode = 0644,
738 .proc_handler = &proc_dointvec,
739 },
fadd8fbd
KH		 740 {
741 .ctl_name = VM_PANIC_ON_OOM,
742 .procname = "panic_on_oom",
743 .data = &sysctl_panic_on_oom,
744 .maxlen = sizeof(sysctl_panic_on_oom),
745 .mode = 0644,
746 .proc_handler = &proc_dointvec,
747 },
1da177e4
LT		 748 {
749 .ctl_name = VM_OVERCOMMIT_RATIO,
750 .procname = "overcommit_ratio",
751 .data = &sysctl_overcommit_ratio,
752 .maxlen = sizeof(sysctl_overcommit_ratio),
753 .mode = 0644,
754 .proc_handler = &proc_dointvec,
755 },
756 {
757 .ctl_name = VM_PAGE_CLUSTER,
758 .procname = "page-cluster",
759 .data = &page_cluster,
760 .maxlen = sizeof(int),
761 .mode = 0644,
762 .proc_handler = &proc_dointvec,
763 },
764 {
765 .ctl_name = VM_DIRTY_BACKGROUND,
766 .procname = "dirty_background_ratio",
767 .data = &dirty_background_ratio,
768 .maxlen = sizeof(dirty_background_ratio),
769 .mode = 0644,
770 .proc_handler = &proc_dointvec_minmax,
771 .strategy = &sysctl_intvec,
772 .extra1 = &zero,
773 .extra2 = &one_hundred,
774 },
775 {
776 .ctl_name = VM_DIRTY_RATIO,
777 .procname = "dirty_ratio",
778 .data = &vm_dirty_ratio,
779 .maxlen = sizeof(vm_dirty_ratio),
780 .mode = 0644,
781 .proc_handler = &proc_dointvec_minmax,
782 .strategy = &sysctl_intvec,
783 .extra1 = &zero,
784 .extra2 = &one_hundred,
785 },
786 {
787 .ctl_name = VM_DIRTY_WB_CS,
788 .procname = "dirty_writeback_centisecs",
f6ef9438
BS		 789 .data = &dirty_writeback_interval,
790 .maxlen = sizeof(dirty_writeback_interval),
1da177e4
LT		 791 .mode = 0644,
792 .proc_handler = &dirty_writeback_centisecs_handler,
793 },
794 {
795 .ctl_name = VM_DIRTY_EXPIRE_CS,
796 .procname = "dirty_expire_centisecs",
f6ef9438
BS		 797 .data = &dirty_expire_interval,
798 .maxlen = sizeof(dirty_expire_interval),
1da177e4 799 .mode = 0644,
f6ef9438 800 .proc_handler = &proc_dointvec_userhz_jiffies,
1da177e4
LT		 801 },
802 {
803 .ctl_name = VM_NR_PDFLUSH_THREADS,
804 .procname = "nr_pdflush_threads",
805 .data = &nr_pdflush_threads,
806 .maxlen = sizeof nr_pdflush_threads,
807 .mode = 0444 /* read-only*/,
808 .proc_handler = &proc_dointvec,
809 },
810 {
811 .ctl_name = VM_SWAPPINESS,
812 .procname = "swappiness",
813 .data = &vm_swappiness,
814 .maxlen = sizeof(vm_swappiness),
815 .mode = 0644,
816 .proc_handler = &proc_dointvec_minmax,
817 .strategy = &sysctl_intvec,
818 .extra1 = &zero,
819 .extra2 = &one_hundred,
820 },
821#ifdef CONFIG_HUGETLB_PAGE
822 {
823 .ctl_name = VM_HUGETLB_PAGES,
824 .procname = "nr_hugepages",
825 .data = &max_huge_pages,
826 .maxlen = sizeof(unsigned long),
827 .mode = 0644,
828 .proc_handler = &hugetlb_sysctl_handler,
829 .extra1 = (void *)&hugetlb_zero,
830 .extra2 = (void *)&hugetlb_infinity,
831 },
832 {
833 .ctl_name = VM_HUGETLB_GROUP,
834 .procname = "hugetlb_shm_group",
835 .data = &sysctl_hugetlb_shm_group,
836 .maxlen = sizeof(gid_t),
837 .mode = 0644,
838 .proc_handler = &proc_dointvec,
839 },
840#endif
841 {
842 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
843 .procname = "lowmem_reserve_ratio",
844 .data = &sysctl_lowmem_reserve_ratio,
845 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
846 .mode = 0644,
847 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
848 .strategy = &sysctl_intvec,
849 },
9d0243bc
AM		 850 {
851 .ctl_name = VM_DROP_PAGECACHE,
852 .procname = "drop_caches",
853 .data = &sysctl_drop_caches,
854 .maxlen = sizeof(int),
855 .mode = 0644,
856 .proc_handler = drop_caches_sysctl_handler,
857 .strategy = &sysctl_intvec,
858 },
1da177e4
LT		 859 {
860 .ctl_name = VM_MIN_FREE_KBYTES,
861 .procname = "min_free_kbytes",
862 .data = &min_free_kbytes,
863 .maxlen = sizeof(min_free_kbytes),
864 .mode = 0644,
865 .proc_handler = &min_free_kbytes_sysctl_handler,
866 .strategy = &sysctl_intvec,
867 .extra1 = &zero,
868 },
8ad4b1fb
RS		 869 {
870 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
871 .procname = "percpu_pagelist_fraction",
872 .data = &percpu_pagelist_fraction,
873 .maxlen = sizeof(percpu_pagelist_fraction),
874 .mode = 0644,
875 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
876 .strategy = &sysctl_intvec,
877 .extra1 = &min_percpu_pagelist_fract,
878 },
1da177e4
LT		 879#ifdef CONFIG_MMU
880 {
881 .ctl_name = VM_MAX_MAP_COUNT,
882 .procname = "max_map_count",
883 .data = &sysctl_max_map_count,
884 .maxlen = sizeof(sysctl_max_map_count),
885 .mode = 0644,
886 .proc_handler = &proc_dointvec
887 },
888#endif
889 {
890 .ctl_name = VM_LAPTOP_MODE,
891 .procname = "laptop_mode",
892 .data = &laptop_mode,
893 .maxlen = sizeof(laptop_mode),
894 .mode = 0644,
ed5b43f1
BS		 895 .proc_handler = &proc_dointvec_jiffies,
896 .strategy = &sysctl_jiffies,
1da177e4
LT		 897 },
898 {
899 .ctl_name = VM_BLOCK_DUMP,
900 .procname = "block_dump",
901 .data = &block_dump,
902 .maxlen = sizeof(block_dump),
903 .mode = 0644,
904 .proc_handler = &proc_dointvec,
905 .strategy = &sysctl_intvec,
906 .extra1 = &zero,
907 },
908 {
909 .ctl_name = VM_VFS_CACHE_PRESSURE,
910 .procname = "vfs_cache_pressure",
911 .data = &sysctl_vfs_cache_pressure,
912 .maxlen = sizeof(sysctl_vfs_cache_pressure),
913 .mode = 0644,
914 .proc_handler = &proc_dointvec,
915 .strategy = &sysctl_intvec,
916 .extra1 = &zero,
917 },
918#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
919 {
920 .ctl_name = VM_LEGACY_VA_LAYOUT,
921 .procname = "legacy_va_layout",
922 .data = &sysctl_legacy_va_layout,
923 .maxlen = sizeof(sysctl_legacy_va_layout),
924 .mode = 0644,
925 .proc_handler = &proc_dointvec,
926 .strategy = &sysctl_intvec,
927 .extra1 = &zero,
928 },
929#endif
930#ifdef CONFIG_SWAP
931 {
932 .ctl_name = VM_SWAP_TOKEN_TIMEOUT,
933 .procname = "swap_token_timeout",
934 .data = &swap_token_default_timeout,
935 .maxlen = sizeof(swap_token_default_timeout),
936 .mode = 0644,
937 .proc_handler = &proc_dointvec_jiffies,
938 .strategy = &sysctl_jiffies,
939 },
1743660b
CL		 940#endif
941#ifdef CONFIG_NUMA
942 {
943 .ctl_name = VM_ZONE_RECLAIM_MODE,
944 .procname = "zone_reclaim_mode",
945 .data = &zone_reclaim_mode,
946 .maxlen = sizeof(zone_reclaim_mode),
947 .mode = 0644,
948 .proc_handler = &proc_dointvec,
c84db23c
CL		 949 .strategy = &sysctl_intvec,
950 .extra1 = &zero,
1743660b 951 },
9614634f
CL		 952 {
953 .ctl_name = VM_MIN_UNMAPPED,
954 .procname = "min_unmapped_ratio",
955 .data = &sysctl_min_unmapped_ratio,
956 .maxlen = sizeof(sysctl_min_unmapped_ratio),
957 .mode = 0644,
958 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
959 .strategy = &sysctl_intvec,
960 .extra1 = &zero,
961 .extra2 = &one_hundred,
962 },
0ff38490
CL		 963 {
964 .ctl_name = VM_MIN_SLAB,
965 .procname = "min_slab_ratio",
966 .data = &sysctl_min_slab_ratio,
967 .maxlen = sizeof(sysctl_min_slab_ratio),
968 .mode = 0644,
969 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
970 .strategy = &sysctl_intvec,
971 .extra1 = &zero,
972 .extra2 = &one_hundred,
973 },
e6e5494c
IM		 974#endif
975#ifdef CONFIG_X86_32
976 {
977 .ctl_name = VM_VDSO_ENABLED,
978 .procname = "vdso_enabled",
979 .data = &vdso_enabled,
980 .maxlen = sizeof(vdso_enabled),
981 .mode = 0644,
982 .proc_handler = &proc_dointvec,
983 .strategy = &sysctl_intvec,
984 .extra1 = &zero,
985 },
1da177e4
LT		 986#endif
987 { .ctl_name = 0 }
988};
989
1da177e4
LT		 990static ctl_table fs_table[] = {
991 {
992 .ctl_name = FS_NRINODE,
993 .procname = "inode-nr",
994 .data = &inodes_stat,
995 .maxlen = 2*sizeof(int),
996 .mode = 0444,
997 .proc_handler = &proc_dointvec,
998 },
999 {
1000 .ctl_name = FS_STATINODE,
1001 .procname = "inode-state",
1002 .data = &inodes_stat,
1003 .maxlen = 7*sizeof(int),
1004 .mode = 0444,
1005 .proc_handler = &proc_dointvec,
1006 },
1007 {
1008 .ctl_name = FS_NRFILE,
1009 .procname = "file-nr",
1010 .data = &files_stat,
1011 .maxlen = 3*sizeof(int),
1012 .mode = 0444,
529bf6be 1013 .proc_handler = &proc_nr_files,
1da177e4
LT		 1014 },
1015 {
1016 .ctl_name = FS_MAXFILE,
1017 .procname = "file-max",
1018 .data = &files_stat.max_files,
1019 .maxlen = sizeof(int),
1020 .mode = 0644,
1021 .proc_handler = &proc_dointvec,
1022 },
1023 {
1024 .ctl_name = FS_DENTRY,
1025 .procname = "dentry-state",
1026 .data = &dentry_stat,
1027 .maxlen = 6*sizeof(int),
1028 .mode = 0444,
1029 .proc_handler = &proc_dointvec,
1030 },
1031 {
1032 .ctl_name = FS_OVERFLOWUID,
1033 .procname = "overflowuid",
1034 .data = &fs_overflowuid,
1035 .maxlen = sizeof(int),
1036 .mode = 0644,
1037 .proc_handler = &proc_dointvec_minmax,
1038 .strategy = &sysctl_intvec,
1039 .extra1 = &minolduid,
1040 .extra2 = &maxolduid,
1041 },
1042 {
1043 .ctl_name = FS_OVERFLOWGID,
1044 .procname = "overflowgid",
1045 .data = &fs_overflowgid,
1046 .maxlen = sizeof(int),
1047 .mode = 0644,
1048 .proc_handler = &proc_dointvec_minmax,
1049 .strategy = &sysctl_intvec,
1050 .extra1 = &minolduid,
1051 .extra2 = &maxolduid,
1052 },
1053 {
1054 .ctl_name = FS_LEASES,
1055 .procname = "leases-enable",
1056 .data = &leases_enable,
1057 .maxlen = sizeof(int),
1058 .mode = 0644,
1059 .proc_handler = &proc_dointvec,
1060 },
1061#ifdef CONFIG_DNOTIFY
1062 {
1063 .ctl_name = FS_DIR_NOTIFY,
1064 .procname = "dir-notify-enable",
1065 .data = &dir_notify_enable,
1066 .maxlen = sizeof(int),
1067 .mode = 0644,
1068 .proc_handler = &proc_dointvec,
1069 },
1070#endif
1071#ifdef CONFIG_MMU
1072 {
1073 .ctl_name = FS_LEASE_TIME,
1074 .procname = "lease-break-time",
1075 .data = &lease_break_time,
1076 .maxlen = sizeof(int),
1077 .mode = 0644,
1078 .proc_handler = &proc_dointvec,
1079 },
1080 {
1081 .ctl_name = FS_AIO_NR,
1082 .procname = "aio-nr",
1083 .data = &aio_nr,
1084 .maxlen = sizeof(aio_nr),
1085 .mode = 0444,
d55b5fda 1086 .proc_handler = &proc_doulongvec_minmax,
1da177e4
LT		 1087 },
1088 {
1089 .ctl_name = FS_AIO_MAX_NR,
1090 .procname = "aio-max-nr",
1091 .data = &aio_max_nr,
1092 .maxlen = sizeof(aio_max_nr),
1093 .mode = 0644,
d55b5fda 1094 .proc_handler = &proc_doulongvec_minmax,
1da177e4 1095 },
2d9048e2 1096#ifdef CONFIG_INOTIFY_USER
0399cb08
RL		 1097 {
1098 .ctl_name = FS_INOTIFY,
1099 .procname = "inotify",
1100 .mode = 0555,
1101 .child = inotify_table,
1102 },
1103#endif
1da177e4 1104#endif
d6e71144
AC		 1105 {
1106 .ctl_name = KERN_SETUID_DUMPABLE,
1107 .procname = "suid_dumpable",
1108 .data = &suid_dumpable,
1109 .maxlen = sizeof(int),
1110 .mode = 0644,
1111 .proc_handler = &proc_dointvec,
1112 },
1da177e4
LT		 1113 { .ctl_name = 0 }
1114};
1115
1116static ctl_table debug_table[] = {
1117 { .ctl_name = 0 }
1118};
1119
1120static ctl_table dev_table[] = {
1121 { .ctl_name = 0 }
0eeca283 1122};
1da177e4
LT		 1123
1124extern void init_irq_proc (void);
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
1da177e4
LT		 1167void __init sysctl_init(void)
1168{
1169#ifdef CONFIG_PROC_FS
330d57fb 1170 register_proc_table(root_table, proc_sys_root, &root_table_header);
1da177e4
LT		 1171 init_irq_proc();
1172#endif
1173}
1174
1175int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1176 void __user *newval, size_t newlen)
1177{
1178 struct list_head *tmp;
330d57fb 1179 int error = -ENOTDIR;
1da177e4
LT		 1180
1181 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1182 return -ENOTDIR;
1183 if (oldval) {
1184 int old_len;
1185 if (!oldlenp || get_user(old_len, oldlenp))
1186 return -EFAULT;
1187 }
330d57fb 1188 spin_lock(&sysctl_lock);
1da177e4
LT		 1189 tmp = &root_table_header.ctl_entry;
1190 do {
1191 struct ctl_table_header *head =
1192 list_entry(tmp, struct ctl_table_header, ctl_entry);
1193 void *context = NULL;
330d57fb
AV		 1194
1195 if (!use_table(head))
1196 continue;
1197
1198 spin_unlock(&sysctl_lock);
1199
1200 error = parse_table(name, nlen, oldval, oldlenp,
1da177e4
LT		 1201 newval, newlen, head->ctl_table,
1202 &context);
5a6b454f 1203 kfree(context);
330d57fb
AV		 1204
1205 spin_lock(&sysctl_lock);
1206 unuse_table(head);
1da177e4 1207 if (error != -ENOTDIR)
330d57fb
AV		 1208 break;
1209 } while ((tmp = tmp->next) != &root_table_header.ctl_entry);
1210 spin_unlock(&sysctl_lock);
1211 return error;
1da177e4
LT		 1212}
1213
1214asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1215{
1216 struct __sysctl_args tmp;
1217 int error;
1218
1219 if (copy_from_user(&tmp, args, sizeof(tmp)))
1220 return -EFAULT;
1221
1222 lock_kernel();
1223 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1224 tmp.newval, tmp.newlen);
1225 unlock_kernel();
1226 return error;
1227}
1228
1229/*
1230 * ctl_perm does NOT grant the superuser all rights automatically, because
1231 * some sysctl variables are readonly even to root.
1232 */
1233
1234static int test_perm(int mode, int op)
1235{
1236 if (!current->euid)
1237 mode >>= 6;
1238 else if (in_egroup_p(0))
1239 mode >>= 3;
1240 if ((mode & op & 0007) == op)
1241 return 0;
1242 return -EACCES;
1243}
1244
1245static inline int ctl_perm(ctl_table *table, int op)
1246{
1247 int error;
1248 error = security_sysctl(table, op);
1249 if (error)
1250 return error;
1251 return test_perm(table->mode, op);
1252}
1253
1254static int parse_table(int __user *name, int nlen,
1255 void __user *oldval, size_t __user *oldlenp,
1256 void __user *newval, size_t newlen,
1257 ctl_table *table, void **context)
1258{
1259 int n;
1260repeat:
1261 if (!nlen)
1262 return -ENOTDIR;
1263 if (get_user(n, name))
1264 return -EFAULT;
1265 for ( ; table->ctl_name; table++) {
1266 if (n == table->ctl_name || table->ctl_name == CTL_ANY) {
1267 int error;
1268 if (table->child) {
1269 if (ctl_perm(table, 001))
1270 return -EPERM;
1271 if (table->strategy) {
1272 error = table->strategy(
1273 table, name, nlen,
1274 oldval, oldlenp,
1275 newval, newlen, context);
1276 if (error)
1277 return error;
1278 }
1279 name++;
1280 nlen--;
1281 table = table->child;
1282 goto repeat;
1283 }
1284 error = do_sysctl_strategy(table, name, nlen,
1285 oldval, oldlenp,
1286 newval, newlen, context);
1287 return error;
1288 }
1289 }
1290 return -ENOTDIR;
1291}
1292
1293/* Perform the actual read/write of a sysctl table entry. */
1294int do_sysctl_strategy (ctl_table *table,
1295 int __user *name, int nlen,
1296 void __user *oldval, size_t __user *oldlenp,
1297 void __user *newval, size_t newlen, void **context)
1298{
1299 int op = 0, rc;
1300 size_t len;
1301
1302 if (oldval)
1303 op |= 004;
1304 if (newval)
1305 op |= 002;
1306 if (ctl_perm(table, op))
1307 return -EPERM;
1308
1309 if (table->strategy) {
1310 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1311 newval, newlen, context);
1312 if (rc < 0)
1313 return rc;
1314 if (rc > 0)
1315 return 0;
1316 }
1317
1318 /* If there is no strategy routine, or if the strategy returns
1319 * zero, proceed with automatic r/w */
1320 if (table->data && table->maxlen) {
1321 if (oldval && oldlenp) {
1322 if (get_user(len, oldlenp))
1323 return -EFAULT;
1324 if (len) {
1325 if (len > table->maxlen)
1326 len = table->maxlen;
1327 if(copy_to_user(oldval, table->data, len))
1328 return -EFAULT;
1329 if(put_user(len, oldlenp))
1330 return -EFAULT;
1331 }
1332 }
1333 if (newval && newlen) {
1334 len = newlen;
1335 if (len > table->maxlen)
1336 len = table->maxlen;
1337 if(copy_from_user(table->data, newval, len))
1338 return -EFAULT;
1339 }
1340 }
1341 return 0;
1342}
1343
1344/**
1345 * register_sysctl_table - register a sysctl hierarchy
1346 * @table: the top-level table structure
1347 * @insert_at_head: whether the entry should be inserted in front or at the end
1348 *
1349 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1350 * array. An entry with a ctl_name of 0 terminates the table.
1351 *
1352 * The members of the &ctl_table structure are used as follows:
1353 *
1354 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1355 * must be unique within that level of sysctl
1356 *
1357 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1358 * enter a sysctl file
1359 *
1360 * data - a pointer to data for use by proc_handler
1361 *
1362 * maxlen - the maximum size in bytes of the data
1363 *
1364 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1365 *
1366 * child - a pointer to the child sysctl table if this entry is a directory, or
1367 * %NULL.
1368 *
1369 * proc_handler - the text handler routine (described below)
1370 *
1371 * strategy - the strategy routine (described below)
1372 *
1373 * de - for internal use by the sysctl routines
1374 *
1375 * extra1, extra2 - extra pointers usable by the proc handler routines
1376 *
1377 * Leaf nodes in the sysctl tree will be represented by a single file
1378 * under /proc; non-leaf nodes will be represented by directories.
1379 *
1380 * sysctl(2) can automatically manage read and write requests through
1381 * the sysctl table. The data and maxlen fields of the ctl_table
1382 * struct enable minimal validation of the values being written to be
1383 * performed, and the mode field allows minimal authentication.
1384 *
1385 * More sophisticated management can be enabled by the provision of a
1386 * strategy routine with the table entry. This will be called before
1387 * any automatic read or write of the data is performed.
1388 *
1389 * The strategy routine may return
1390 *
1391 * < 0 - Error occurred (error is passed to user process)
1392 *
1393 * 0 - OK - proceed with automatic read or write.
1394 *
1395 * > 0 - OK - read or write has been done by the strategy routine, so
1396 * return immediately.
1397 *
1398 * There must be a proc_handler routine for any terminal nodes
1399 * mirrored under /proc/sys (non-terminals are handled by a built-in
1400 * directory handler). Several default handlers are available to
1401 * cover common cases -
1402 *
1403 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1404 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1405 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1406 *
1407 * It is the handler's job to read the input buffer from user memory
1408 * and process it. The handler should return 0 on success.
1409 *
1410 * This routine returns %NULL on a failure to register, and a pointer
1411 * to the table header on success.
1412 */
1413struct ctl_table_header *register_sysctl_table(ctl_table * table,
1414 int insert_at_head)
1415{
1416 struct ctl_table_header *tmp;
1417 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1418 if (!tmp)
1419 return NULL;
1420 tmp->ctl_table = table;
1421 INIT_LIST_HEAD(&tmp->ctl_entry);
330d57fb
AV		 1422 tmp->used = 0;
1423 tmp->unregistering = NULL;
1424 spin_lock(&sysctl_lock);
1da177e4
LT		 1425 if (insert_at_head)
1426 list_add(&tmp->ctl_entry, &root_table_header.ctl_entry);
1427 else
1428 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
330d57fb 1429 spin_unlock(&sysctl_lock);
1da177e4 1430#ifdef CONFIG_PROC_FS
330d57fb 1431 register_proc_table(table, proc_sys_root, tmp);
1da177e4
LT		 1432#endif
1433 return tmp;
1434}
1435
1436/**
1437 * unregister_sysctl_table - unregister a sysctl table hierarchy
1438 * @header: the header returned from register_sysctl_table
1439 *
1440 * Unregisters the sysctl table and all children. proc entries may not
1441 * actually be removed until they are no longer used by anyone.
1442 */
1443void unregister_sysctl_table(struct ctl_table_header * header)
1444{
330d57fb
AV		 1445 might_sleep();
1446 spin_lock(&sysctl_lock);
1447 start_unregistering(header);
1da177e4
LT		 1448#ifdef CONFIG_PROC_FS
1449 unregister_proc_table(header->ctl_table, proc_sys_root);
1450#endif
330d57fb 1451 spin_unlock(&sysctl_lock);
1da177e4
LT		 1452 kfree(header);
1453}
1454
1455/*
1456 * /proc/sys support
1457 */
1458
1459#ifdef CONFIG_PROC_FS
1460
1461/* Scan the sysctl entries in table and add them all into /proc */
330d57fb 1462static void register_proc_table(ctl_table * table, struct proc_dir_entry *root, void *set)
1da177e4
LT		 1463{
1464 struct proc_dir_entry *de;
1465 int len;
1466 mode_t mode;
1467
1468 for (; table->ctl_name; table++) {
1469 /* Can't do anything without a proc name. */
1470 if (!table->procname)
1471 continue;
1472 /* Maybe we can't do anything with it... */
1473 if (!table->proc_handler && !table->child) {
1474 printk(KERN_WARNING "SYSCTL: Can't register %s\n",
1475 table->procname);
1476 continue;
1477 }
1478
1479 len = strlen(table->procname);
1480 mode = table->mode;
1481
1482 de = NULL;
1483 if (table->proc_handler)
1484 mode |= S_IFREG;
1485 else {
1486 mode |= S_IFDIR;
1487 for (de = root->subdir; de; de = de->next) {
1488 if (proc_match(len, table->procname, de))
1489 break;
1490 }
1491 /* If the subdir exists already, de is non-NULL */
1492 }
1493
1494 if (!de) {
1495 de = create_proc_entry(table->procname, mode, root);
1496 if (!de)
1497 continue;
330d57fb 1498 de->set = set;
1da177e4
LT		 1499 de->data = (void *) table;
1500 if (table->proc_handler)
1501 de->proc_fops = &proc_sys_file_operations;
1502 }
1503 table->de = de;
1504 if (de->mode & S_IFDIR)
330d57fb 1505 register_proc_table(table->child, de, set);
1da177e4
LT		 1506 }
1507}
1508
1509/*
1510 * Unregister a /proc sysctl table and any subdirectories.
1511 */
1512static void unregister_proc_table(ctl_table * table, struct proc_dir_entry *root)
1513{
1514 struct proc_dir_entry *de;
1515 for (; table->ctl_name; table++) {
1516 if (!(de = table->de))
1517 continue;
1518 if (de->mode & S_IFDIR) {
1519 if (!table->child) {
1520 printk (KERN_ALERT "Help - malformed sysctl tree on free\n");
1521 continue;
1522 }
1523 unregister_proc_table(table->child, de);
1524
1525 /* Don't unregister directories which still have entries.. */
1526 if (de->subdir)
1527 continue;
1528 }
1529
330d57fb
AV		 1530 /*
1531 * In any case, mark the entry as goner; we'll keep it
1532 * around if it's busy, but we'll know to do nothing with
1533 * its fields. We are under sysctl_lock here.
1534 */
1535 de->data = NULL;
1536
1da177e4
LT		 1537 /* Don't unregister proc entries that are still being used.. */
1538 if (atomic_read(&de->count))
1539 continue;
1540
1541 table->de = NULL;
1542 remove_proc_entry(table->procname, root);
1543 }
1544}
1545
1546static ssize_t do_rw_proc(int write, struct file * file, char __user * buf,
1547 size_t count, loff_t *ppos)
1548{
1549 int op;
330d57fb 1550 struct proc_dir_entry *de = PDE(file->f_dentry->d_inode);
1da177e4
LT		 1551 struct ctl_table *table;
1552 size_t res;
330d57fb 1553 ssize_t error = -ENOTDIR;
1da177e4 1554
330d57fb
AV		 1555 spin_lock(&sysctl_lock);
1556 if (de && de->data && use_table(de->set)) {
1557 /*
1558 * at that point we know that sysctl was not unregistered
1559 * and won't be until we finish
1560 */
1561 spin_unlock(&sysctl_lock);
1562 table = (struct ctl_table *) de->data;
1563 if (!table || !table->proc_handler)
1564 goto out;
1565 error = -EPERM;
1566 op = (write ? 002 : 004);
1567 if (ctl_perm(table, op))
1568 goto out;
1569
1570 /* careful: calling conventions are nasty here */
1571 res = count;
1572 error = (*table->proc_handler)(table, write, file,
1573 buf, &res, ppos);
1574 if (!error)
1575 error = res;
1576 out:
1577 spin_lock(&sysctl_lock);
1578 unuse_table(de->set);
1579 }
1580 spin_unlock(&sysctl_lock);
1581 return error;
1da177e4
LT		 1582}
1583
1584static int proc_opensys(struct inode *inode, struct file *file)
1585{
1586 if (file->f_mode & FMODE_WRITE) {
1587 /*
1588 * sysctl entries that are not writable,
1589 * are _NOT_ writable, capabilities or not.
1590 */
1591 if (!(inode->i_mode & S_IWUSR))
1592 return -EPERM;
1593 }
1594
1595 return 0;
1596}
1597
1598static ssize_t proc_readsys(struct file * file, char __user * buf,
1599 size_t count, loff_t *ppos)
1600{
1601 return do_rw_proc(0, file, buf, count, ppos);
1602}
1603
1604static ssize_t proc_writesys(struct file * file, const char __user * buf,
1605 size_t count, loff_t *ppos)
1606{
1607 return do_rw_proc(1, file, (char __user *) buf, count, ppos);
1608}
1609
1610/**
1611 * proc_dostring - read a string sysctl
1612 * @table: the sysctl table
1613 * @write: %TRUE if this is a write to the sysctl file
1614 * @filp: the file structure
1615 * @buffer: the user buffer
1616 * @lenp: the size of the user buffer
1617 * @ppos: file position
1618 *
1619 * Reads/writes a string from/to the user buffer. If the kernel
1620 * buffer provided is not large enough to hold the string, the
1621 * string is truncated. The copied string is %NULL-terminated.
1622 * If the string is being read by the user process, it is copied
1623 * and a newline '\n' is added. It is truncated if the buffer is
1624 * not large enough.
1625 *
1626 * Returns 0 on success.
1627 */
1628int proc_dostring(ctl_table *table, int write, struct file *filp,
1629 void __user *buffer, size_t *lenp, loff_t *ppos)
1630{
1631 size_t len;
1632 char __user *p;
1633 char c;
1634
1635 if (!table->data || !table->maxlen || !*lenp ||
1636 (*ppos && !write)) {
1637 *lenp = 0;
1638 return 0;
1639 }
1640
1641 if (write) {
1642 len = 0;
1643 p = buffer;
1644 while (len < *lenp) {
1645 if (get_user(c, p++))
1646 return -EFAULT;
1647 if (c == 0 || c == '\n')
1648 break;
1649 len++;
1650 }
1651 if (len >= table->maxlen)
1652 len = table->maxlen-1;
1653 if(copy_from_user(table->data, buffer, len))
1654 return -EFAULT;
1655 ((char *) table->data)[len] = 0;
1656 *ppos += *lenp;
1657 } else {
1658 len = strlen(table->data);
1659 if (len > table->maxlen)
1660 len = table->maxlen;
1661 if (len > *lenp)
1662 len = *lenp;
1663 if (len)
1664 if(copy_to_user(buffer, table->data, len))
1665 return -EFAULT;
1666 if (len < *lenp) {
1667 if(put_user('\n', ((char __user *) buffer) + len))
1668 return -EFAULT;
1669 len++;
1670 }
1671 *lenp = len;
1672 *ppos += len;
1673 }
1674 return 0;
1675}
1676
1677/*
1678 * Special case of dostring for the UTS structure. This has locks
1679 * to observe. Should this be in kernel/sys.c ????
1680 */
1681
1682static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
1683 void __user *buffer, size_t *lenp, loff_t *ppos)
1684{
1685 int r;
1686
1687 if (!write) {
1688 down_read(&uts_sem);
1689 r=proc_dostring(table,0,filp,buffer,lenp, ppos);
1690 up_read(&uts_sem);
1691 } else {
1692 down_write(&uts_sem);
1693 r=proc_dostring(table,1,filp,buffer,lenp, ppos);
1694 up_write(&uts_sem);
1695 }
1696 return r;
1697}
1698
1699static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1700 int *valp,
1701 int write, void *data)
1702{
1703 if (write) {
1704 *valp = *negp ? -*lvalp : *lvalp;
1705 } else {
1706 int val = *valp;
1707 if (val < 0) {
1708 *negp = -1;
1709 *lvalp = (unsigned long)-val;
1710 } else {
1711 *negp = 0;
1712 *lvalp = (unsigned long)val;
1713 }
1714 }
1715 return 0;
1716}
1717
1718static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1719 void __user *buffer, size_t *lenp, loff_t *ppos,
1720 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1721 int write, void *data),
1722 void *data)
1723{
1724#define TMPBUFLEN 21
1725 int *i, vleft, first=1, neg, val;
1726 unsigned long lval;
1727 size_t left, len;
1728
1729 char buf[TMPBUFLEN], *p;
1730 char __user *s = buffer;
1731
1732 if (!table->data || !table->maxlen || !*lenp ||
1733 (*ppos && !write)) {
1734 *lenp = 0;
1735 return 0;
1736 }
1737
1738 i = (int *) table->data;
1739 vleft = table->maxlen / sizeof(*i);
1740 left = *lenp;
1741
1742 if (!conv)
1743 conv = do_proc_dointvec_conv;
1744
1745 for (; left && vleft--; i++, first=0) {
1746 if (write) {
1747 while (left) {
1748 char c;
1749 if (get_user(c, s))
1750 return -EFAULT;
1751 if (!isspace(c))
1752 break;
1753 left--;
1754 s++;
1755 }
1756 if (!left)
1757 break;
1758 neg = 0;
1759 len = left;
1760 if (len > sizeof(buf) - 1)
1761 len = sizeof(buf) - 1;
1762 if (copy_from_user(buf, s, len))
1763 return -EFAULT;
1764 buf[len] = 0;
1765 p = buf;
1766 if (*p == '-' && left > 1) {
1767 neg = 1;
1768 left--, p++;
1769 }
1770 if (*p < '0' || *p > '9')
1771 break;
1772
1773 lval = simple_strtoul(p, &p, 0);
1774
1775 len = p-buf;
1776 if ((len < left) && *p && !isspace(*p))
1777 break;
1778 if (neg)
1779 val = -val;
1780 s += len;
1781 left -= len;
1782
1783 if (conv(&neg, &lval, i, 1, data))
1784 break;
1785 } else {
1786 p = buf;
1787 if (!first)
1788 *p++ = '\t';
1789
1790 if (conv(&neg, &lval, i, 0, data))
1791 break;
1792
1793 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1794 len = strlen(buf);
1795 if (len > left)
1796 len = left;
1797 if(copy_to_user(s, buf, len))
1798 return -EFAULT;
1799 left -= len;
1800 s += len;
1801 }
1802 }
1803
1804 if (!write && !first && left) {
1805 if(put_user('\n', s))
1806 return -EFAULT;
1807 left--, s++;
1808 }
1809 if (write) {
1810 while (left) {
1811 char c;
1812 if (get_user(c, s++))
1813 return -EFAULT;
1814 if (!isspace(c))
1815 break;
1816 left--;
1817 }
1818 }
1819 if (write && first)
1820 return -EINVAL;
1821 *lenp -= left;
1822 *ppos += *lenp;
1823 return 0;
1824#undef TMPBUFLEN
1825}
1826
1827/**
1828 * proc_dointvec - read a vector of integers
1829 * @table: the sysctl table
1830 * @write: %TRUE if this is a write to the sysctl file
1831 * @filp: the file structure
1832 * @buffer: the user buffer
1833 * @lenp: the size of the user buffer
1834 * @ppos: file position
1835 *
1836 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1837 * values from/to the user buffer, treated as an ASCII string.
1838 *
1839 * Returns 0 on success.
1840 */
1841int proc_dointvec(ctl_table *table, int write, struct file *filp,
1842 void __user *buffer, size_t *lenp, loff_t *ppos)
1843{
1844 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1845 NULL,NULL);
1846}
1847
1848#define OP_SET 0
1849#define OP_AND 1
1850#define OP_OR 2
1851#define OP_MAX 3
1852#define OP_MIN 4
1853
1854static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1855 int *valp,
1856 int write, void *data)
1857{
1858 int op = *(int *)data;
1859 if (write) {
1860 int val = *negp ? -*lvalp : *lvalp;
1861 switch(op) {
1862 case OP_SET: *valp = val; break;
1863 case OP_AND: *valp &= val; break;
1864 case OP_OR: *valp |= val; break;
1865 case OP_MAX: if(*valp < val)
1866 *valp = val;
1867 break;
1868 case OP_MIN: if(*valp > val)
1869 *valp = val;
1870 break;
1871 }
1872 } else {
1873 int val = *valp;
1874 if (val < 0) {
1875 *negp = -1;
1876 *lvalp = (unsigned long)-val;
1877 } else {
1878 *negp = 0;
1879 *lvalp = (unsigned long)val;
1880 }
1881 }
1882 return 0;
1883}
1884
1885/*
1886 * init may raise the set.
1887 */
1888
1889int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1890 void __user *buffer, size_t *lenp, loff_t *ppos)
1891{
1892 int op;
1893
1894 if (!capable(CAP_SYS_MODULE)) {
1895 return -EPERM;
1896 }
1897
1898 op = (current->pid == 1) ? OP_SET : OP_AND;
1899 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1900 do_proc_dointvec_bset_conv,&op);
1901}
1902
1903struct do_proc_dointvec_minmax_conv_param {
1904 int *min;
1905 int *max;
1906};
1907
1908static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1909 int *valp,
1910 int write, void *data)
1911{
1912 struct do_proc_dointvec_minmax_conv_param *param = data;
1913 if (write) {
1914 int val = *negp ? -*lvalp : *lvalp;
1915 if ((param->min && *param->min > val) ||
1916 (param->max && *param->max < val))
1917 return -EINVAL;
1918 *valp = val;
1919 } else {
1920 int val = *valp;
1921 if (val < 0) {
1922 *negp = -1;
1923 *lvalp = (unsigned long)-val;
1924 } else {
1925 *negp = 0;
1926 *lvalp = (unsigned long)val;
1927 }
1928 }
1929 return 0;
1930}
1931
1932/**
1933 * proc_dointvec_minmax - read a vector of integers with min/max values
1934 * @table: the sysctl table
1935 * @write: %TRUE if this is a write to the sysctl file
1936 * @filp: the file structure
1937 * @buffer: the user buffer
1938 * @lenp: the size of the user buffer
1939 * @ppos: file position
1940 *
1941 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1942 * values from/to the user buffer, treated as an ASCII string.
1943 *
1944 * This routine will ensure the values are within the range specified by
1945 * table->extra1 (min) and table->extra2 (max).
1946 *
1947 * Returns 0 on success.
1948 */
1949int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1950 void __user *buffer, size_t *lenp, loff_t *ppos)
1951{
1952 struct do_proc_dointvec_minmax_conv_param param = {
1953 .min = (int *) table->extra1,
1954 .max = (int *) table->extra2,
1955 };
1956 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1957 do_proc_dointvec_minmax_conv, &param);
1958}
1959
1960static int do_proc_doulongvec_minmax(ctl_table *table, int write,
1961 struct file *filp,
1962 void __user *buffer,
1963 size_t *lenp, loff_t *ppos,
1964 unsigned long convmul,
1965 unsigned long convdiv)
1966{
1967#define TMPBUFLEN 21
1968 unsigned long *i, *min, *max, val;
1969 int vleft, first=1, neg;
1970 size_t len, left;
1971 char buf[TMPBUFLEN], *p;
1972 char __user *s = buffer;
1973
1974 if (!table->data || !table->maxlen || !*lenp ||
1975 (*ppos && !write)) {
1976 *lenp = 0;
1977 return 0;
1978 }
1979
1980 i = (unsigned long *) table->data;
1981 min = (unsigned long *) table->extra1;
1982 max = (unsigned long *) table->extra2;
1983 vleft = table->maxlen / sizeof(unsigned long);
1984 left = *lenp;
1985
1986 for (; left && vleft--; i++, min++, max++, first=0) {
1987 if (write) {
1988 while (left) {
1989 char c;
1990 if (get_user(c, s))
1991 return -EFAULT;
1992 if (!isspace(c))
1993 break;
1994 left--;
1995 s++;
1996 }
1997 if (!left)
1998 break;
1999 neg = 0;
2000 len = left;
2001 if (len > TMPBUFLEN-1)
2002 len = TMPBUFLEN-1;
2003 if (copy_from_user(buf, s, len))
2004 return -EFAULT;
2005 buf[len] = 0;
2006 p = buf;
2007 if (*p == '-' && left > 1) {
2008 neg = 1;
2009 left--, p++;
2010 }
2011 if (*p < '0' || *p > '9')
2012 break;
2013 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2014 len = p-buf;
2015 if ((len < left) && *p && !isspace(*p))
2016 break;
2017 if (neg)
2018 val = -val;
2019 s += len;
2020 left -= len;
2021
2022 if(neg)
2023 continue;
2024 if ((min && val < *min) || (max && val > *max))
2025 continue;
2026 *i = val;
2027 } else {
2028 p = buf;
2029 if (!first)
2030 *p++ = '\t';
2031 sprintf(p, "%lu", convdiv * (*i) / convmul);
2032 len = strlen(buf);
2033 if (len > left)
2034 len = left;
2035 if(copy_to_user(s, buf, len))
2036 return -EFAULT;
2037 left -= len;
2038 s += len;
2039 }
2040 }
2041
2042 if (!write && !first && left) {
2043 if(put_user('\n', s))
2044 return -EFAULT;
2045 left--, s++;
2046 }
2047 if (write) {
2048 while (left) {
2049 char c;
2050 if (get_user(c, s++))
2051 return -EFAULT;
2052 if (!isspace(c))
2053 break;
2054 left--;
2055 }
2056 }
2057 if (write && first)
2058 return -EINVAL;
2059 *lenp -= left;
2060 *ppos += *lenp;
2061 return 0;
2062#undef TMPBUFLEN
2063}
2064
2065/**
2066 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2067 * @table: the sysctl table
2068 * @write: %TRUE if this is a write to the sysctl file
2069 * @filp: the file structure
2070 * @buffer: the user buffer
2071 * @lenp: the size of the user buffer
2072 * @ppos: file position
2073 *
2074 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2075 * values from/to the user buffer, treated as an ASCII string.
2076 *
2077 * This routine will ensure the values are within the range specified by
2078 * table->extra1 (min) and table->extra2 (max).
2079 *
2080 * Returns 0 on success.
2081 */
2082int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2083 void __user *buffer, size_t *lenp, loff_t *ppos)
2084{
2085 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2086}
2087
2088/**
2089 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2090 * @table: the sysctl table
2091 * @write: %TRUE if this is a write to the sysctl file
2092 * @filp: the file structure
2093 * @buffer: the user buffer
2094 * @lenp: the size of the user buffer
2095 * @ppos: file position
2096 *
2097 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2098 * values from/to the user buffer, treated as an ASCII string. The values
2099 * are treated as milliseconds, and converted to jiffies when they are stored.
2100 *
2101 * This routine will ensure the values are within the range specified by
2102 * table->extra1 (min) and table->extra2 (max).
2103 *
2104 * Returns 0 on success.
2105 */
2106int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2107 struct file *filp,
2108 void __user *buffer,
2109 size_t *lenp, loff_t *ppos)
2110{
2111 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2112 lenp, ppos, HZ, 1000l);
2113}
2114
2115
2116static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2117 int *valp,
2118 int write, void *data)
2119{
2120 if (write) {
cba9f33d
BS		 2121 if (*lvalp > LONG_MAX / HZ)
2122 return 1;
1da177e4
LT		 2123 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2124 } else {
2125 int val = *valp;
2126 unsigned long lval;
2127 if (val < 0) {
2128 *negp = -1;
2129 lval = (unsigned long)-val;
2130 } else {
2131 *negp = 0;
2132 lval = (unsigned long)val;
2133 }
2134 *lvalp = lval / HZ;
2135 }
2136 return 0;
2137}
2138
2139static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2140 int *valp,
2141 int write, void *data)
2142{
2143 if (write) {
cba9f33d
BS		 2144 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2145 return 1;
1da177e4
LT		 2146 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2147 } else {
2148 int val = *valp;
2149 unsigned long lval;
2150 if (val < 0) {
2151 *negp = -1;
2152 lval = (unsigned long)-val;
2153 } else {
2154 *negp = 0;
2155 lval = (unsigned long)val;
2156 }
2157 *lvalp = jiffies_to_clock_t(lval);
2158 }
2159 return 0;
2160}
2161
2162static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2163 int *valp,
2164 int write, void *data)
2165{
2166 if (write) {
2167 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2168 } else {
2169 int val = *valp;
2170 unsigned long lval;
2171 if (val < 0) {
2172 *negp = -1;
2173 lval = (unsigned long)-val;
2174 } else {
2175 *negp = 0;
2176 lval = (unsigned long)val;
2177 }
2178 *lvalp = jiffies_to_msecs(lval);
2179 }
2180 return 0;
2181}
2182
2183/**
2184 * proc_dointvec_jiffies - read a vector of integers as seconds
2185 * @table: the sysctl table
2186 * @write: %TRUE if this is a write to the sysctl file
2187 * @filp: the file structure
2188 * @buffer: the user buffer
2189 * @lenp: the size of the user buffer
2190 * @ppos: file position
2191 *
2192 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2193 * values from/to the user buffer, treated as an ASCII string.
2194 * The values read are assumed to be in seconds, and are converted into
2195 * jiffies.
2196 *
2197 * Returns 0 on success.
2198 */
2199int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2200 void __user *buffer, size_t *lenp, loff_t *ppos)
2201{
2202 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2203 do_proc_dointvec_jiffies_conv,NULL);
2204}
2205
2206/**
2207 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2208 * @table: the sysctl table
2209 * @write: %TRUE if this is a write to the sysctl file
2210 * @filp: the file structure
2211 * @buffer: the user buffer
2212 * @lenp: the size of the user buffer
1e5d5331 2213 * @ppos: pointer to the file position
1da177e4
LT		 2214 *
2215 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2216 * values from/to the user buffer, treated as an ASCII string.
2217 * The values read are assumed to be in 1/USER_HZ seconds, and
2218 * are converted into jiffies.
2219 *
2220 * Returns 0 on success.
2221 */
2222int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2223 void __user *buffer, size_t *lenp, loff_t *ppos)
2224{
2225 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2226 do_proc_dointvec_userhz_jiffies_conv,NULL);
2227}
2228
2229/**
2230 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2231 * @table: the sysctl table
2232 * @write: %TRUE if this is a write to the sysctl file
2233 * @filp: the file structure
2234 * @buffer: the user buffer
2235 * @lenp: the size of the user buffer
67be2dd1
MW		 2236 * @ppos: file position
2237 * @ppos: the current position in the file
1da177e4
LT		 2238 *
2239 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2240 * values from/to the user buffer, treated as an ASCII string.
2241 * The values read are assumed to be in 1/1000 seconds, and
2242 * are converted into jiffies.
2243 *
2244 * Returns 0 on success.
2245 */
2246int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2247 void __user *buffer, size_t *lenp, loff_t *ppos)
2248{
2249 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2250 do_proc_dointvec_ms_jiffies_conv, NULL);
2251}
2252
2253#else /* CONFIG_PROC_FS */
2254
2255int proc_dostring(ctl_table *table, int write, struct file *filp,
2256 void __user *buffer, size_t *lenp, loff_t *ppos)
2257{
2258 return -ENOSYS;
2259}
2260
2261static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
2262 void __user *buffer, size_t *lenp, loff_t *ppos)
2263{
2264 return -ENOSYS;
2265}
2266
2267int proc_dointvec(ctl_table *table, int write, struct file *filp,
2268 void __user *buffer, size_t *lenp, loff_t *ppos)
2269{
2270 return -ENOSYS;
2271}
2272
2273int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2274 void __user *buffer, size_t *lenp, loff_t *ppos)
2275{
2276 return -ENOSYS;
2277}
2278
2279int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2280 void __user *buffer, size_t *lenp, loff_t *ppos)
2281{
2282 return -ENOSYS;
2283}
2284
2285int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2286 void __user *buffer, size_t *lenp, loff_t *ppos)
2287{
2288 return -ENOSYS;
2289}
2290
2291int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2292 void __user *buffer, size_t *lenp, loff_t *ppos)
2293{
2294 return -ENOSYS;
2295}
2296
2297int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2298 void __user *buffer, size_t *lenp, loff_t *ppos)
2299{
2300 return -ENOSYS;
2301}
2302
2303int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2304 void __user *buffer, size_t *lenp, loff_t *ppos)
2305{
2306 return -ENOSYS;
2307}
2308
2309int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2310 struct file *filp,
2311 void __user *buffer,
2312 size_t *lenp, loff_t *ppos)
2313{
2314 return -ENOSYS;
2315}
2316
2317
2318#endif /* CONFIG_PROC_FS */
2319
2320
2321/*
2322 * General sysctl support routines
2323 */
2324
2325/* The generic string strategy routine: */
2326int sysctl_string(ctl_table *table, int __user *name, int nlen,
2327 void __user *oldval, size_t __user *oldlenp,
2328 void __user *newval, size_t newlen, void **context)
2329{
1da177e4
LT		 2330 if (!table->data || !table->maxlen)
2331 return -ENOTDIR;
2332
2333 if (oldval && oldlenp) {
de9e007d
LT		 2334 size_t bufsize;
2335 if (get_user(bufsize, oldlenp))
1da177e4 2336 return -EFAULT;
de9e007d
LT		 2337 if (bufsize) {
2338 size_t len = strlen(table->data), copied;
2339
2340 /* This shouldn't trigger for a well-formed sysctl */
2341 if (len > table->maxlen)
1da177e4 2342 len = table->maxlen;
de9e007d
LT		 2343
2344 /* Copy up to a max of bufsize-1 bytes of the string */
2345 copied = (len >= bufsize) ? bufsize - 1 : len;
2346
2347 if (copy_to_user(oldval, table->data, copied) ||
2348 put_user(0, (char __user *)(oldval + copied)))
1da177e4 2349 return -EFAULT;
de9e007d 2350 if (put_user(len, oldlenp))
1da177e4
LT		 2351 return -EFAULT;
2352 }
2353 }
2354 if (newval && newlen) {
de9e007d 2355 size_t len = newlen;
1da177e4
LT		 2356 if (len > table->maxlen)
2357 len = table->maxlen;
2358 if(copy_from_user(table->data, newval, len))
2359 return -EFAULT;
2360 if (len == table->maxlen)
2361 len--;
2362 ((char *) table->data)[len] = 0;
2363 }
82c9df82 2364 return 1;
1da177e4
LT		 2365}
2366
2367/*
2368 * This function makes sure that all of the integers in the vector
2369 * are between the minimum and maximum values given in the arrays
2370 * table->extra1 and table->extra2, respectively.
2371 */
2372int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2373 void __user *oldval, size_t __user *oldlenp,
2374 void __user *newval, size_t newlen, void **context)
2375{
2376
2377 if (newval && newlen) {
2378 int __user *vec = (int __user *) newval;
2379 int *min = (int *) table->extra1;
2380 int *max = (int *) table->extra2;
2381 size_t length;
2382 int i;
2383
2384 if (newlen % sizeof(int) != 0)
2385 return -EINVAL;
2386
2387 if (!table->extra1 && !table->extra2)
2388 return 0;
2389
2390 if (newlen > table->maxlen)
2391 newlen = table->maxlen;
2392 length = newlen / sizeof(int);
2393
2394 for (i = 0; i < length; i++) {
2395 int value;
2396 if (get_user(value, vec + i))
2397 return -EFAULT;
2398 if (min && value < min[i])
2399 return -EINVAL;
2400 if (max && value > max[i])
2401 return -EINVAL;
2402 }
2403 }
2404 return 0;
2405}
2406
2407/* Strategy function to convert jiffies to seconds */
2408int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2409 void __user *oldval, size_t __user *oldlenp,
2410 void __user *newval, size_t newlen, void **context)
2411{
2412 if (oldval) {
2413 size_t olen;
2414 if (oldlenp) {
2415 if (get_user(olen, oldlenp))
2416 return -EFAULT;
2417 if (olen!=sizeof(int))
2418 return -EINVAL;
2419 }
2420 if (put_user(*(int *)(table->data)/HZ, (int __user *)oldval) ||
2421 (oldlenp && put_user(sizeof(int),oldlenp)))
2422 return -EFAULT;
2423 }
2424 if (newval && newlen) {
2425 int new;
2426 if (newlen != sizeof(int))
2427 return -EINVAL;
2428 if (get_user(new, (int __user *)newval))
2429 return -EFAULT;
2430 *(int *)(table->data) = new*HZ;
2431 }
2432 return 1;
2433}
2434
2435/* Strategy function to convert jiffies to seconds */
2436int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2437 void __user *oldval, size_t __user *oldlenp,
2438 void __user *newval, size_t newlen, void **context)
2439{
2440 if (oldval) {
2441 size_t olen;
2442 if (oldlenp) {
2443 if (get_user(olen, oldlenp))
2444 return -EFAULT;
2445 if (olen!=sizeof(int))
2446 return -EINVAL;
2447 }
2448 if (put_user(jiffies_to_msecs(*(int *)(table->data)), (int __user *)oldval) ||
2449 (oldlenp && put_user(sizeof(int),oldlenp)))
2450 return -EFAULT;
2451 }
2452 if (newval && newlen) {
2453 int new;
2454 if (newlen != sizeof(int))
2455 return -EINVAL;
2456 if (get_user(new, (int __user *)newval))
2457 return -EFAULT;
2458 *(int *)(table->data) = msecs_to_jiffies(new);
2459 }
2460 return 1;
2461}
2462
2463#else /* CONFIG_SYSCTL */
2464
2465
2466asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2467{
2468 return -ENOSYS;
2469}
2470
2471int sysctl_string(ctl_table *table, int __user *name, int nlen,
2472 void __user *oldval, size_t __user *oldlenp,
2473 void __user *newval, size_t newlen, void **context)
2474{
2475 return -ENOSYS;
2476}
2477
2478int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2479 void __user *oldval, size_t __user *oldlenp,
2480 void __user *newval, size_t newlen, void **context)
2481{
2482 return -ENOSYS;
2483}
2484
2485int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2486 void __user *oldval, size_t __user *oldlenp,
2487 void __user *newval, size_t newlen, void **context)
2488{
2489 return -ENOSYS;
2490}
2491
2492int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2493 void __user *oldval, size_t __user *oldlenp,
2494 void __user *newval, size_t newlen, void **context)
2495{
2496 return -ENOSYS;
2497}
2498
2499int proc_dostring(ctl_table *table, int write, struct file *filp,
2500 void __user *buffer, size_t *lenp, loff_t *ppos)
2501{
2502 return -ENOSYS;
2503}
2504
2505int proc_dointvec(ctl_table *table, int write, struct file *filp,
2506 void __user *buffer, size_t *lenp, loff_t *ppos)
2507{
2508 return -ENOSYS;
2509}
2510
2511int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2512 void __user *buffer, size_t *lenp, loff_t *ppos)
2513{
2514 return -ENOSYS;
2515}
2516
2517int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2518 void __user *buffer, size_t *lenp, loff_t *ppos)
2519{
2520 return -ENOSYS;
2521}
2522
2523int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2524 void __user *buffer, size_t *lenp, loff_t *ppos)
2525{
2526 return -ENOSYS;
2527}
2528
2529int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2530 void __user *buffer, size_t *lenp, loff_t *ppos)
2531{
2532 return -ENOSYS;
2533}
2534
2535int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2536 void __user *buffer, size_t *lenp, loff_t *ppos)
2537{
2538 return -ENOSYS;
2539}
2540
2541int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2542 void __user *buffer, size_t *lenp, loff_t *ppos)
2543{
2544 return -ENOSYS;
2545}
2546
2547int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2548 struct file *filp,
2549 void __user *buffer,
2550 size_t *lenp, loff_t *ppos)
2551{
2552 return -ENOSYS;
2553}
2554
2555struct ctl_table_header * register_sysctl_table(ctl_table * table,
2556 int insert_at_head)
2557{
2558 return NULL;
2559}
2560
2561void unregister_sysctl_table(struct ctl_table_header * table)
2562{
2563}
2564
2565#endif /* CONFIG_SYSCTL */
2566
2567/*
2568 * No sense putting this after each symbol definition, twice,
2569 * exception granted :-)
2570 */
2571EXPORT_SYMBOL(proc_dointvec);
2572EXPORT_SYMBOL(proc_dointvec_jiffies);
2573EXPORT_SYMBOL(proc_dointvec_minmax);
2574EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2575EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2576EXPORT_SYMBOL(proc_dostring);
2577EXPORT_SYMBOL(proc_doulongvec_minmax);
2578EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2579EXPORT_SYMBOL(register_sysctl_table);
2580EXPORT_SYMBOL(sysctl_intvec);
2581EXPORT_SYMBOL(sysctl_jiffies);
2582EXPORT_SYMBOL(sysctl_ms_jiffies);
2583EXPORT_SYMBOL(sysctl_string);
2584EXPORT_SYMBOL(unregister_sysctl_table);
Clone of davem's net-next-2.6 tree