]> 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] free_uid() locking improvement
[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
21#include <linux/config.h>
22#include <linux/module.h>
23#include <linux/mm.h>
24#include <linux/swap.h>
25#include <linux/slab.h>
26#include <linux/sysctl.h>
27#include <linux/proc_fs.h>
c59ede7b 28#include <linux/capability.h>
1da177e4
LT		 29#include <linux/ctype.h>
30#include <linux/utsname.h>
31#include <linux/capability.h>
32#include <linux/smp_lock.h>
33#include <linux/init.h>
34#include <linux/kernel.h>
0296b228 35#include <linux/kobject.h>
20380731 36#include <linux/net.h>
1da177e4
LT		 37#include <linux/sysrq.h>
38#include <linux/highuid.h>
39#include <linux/writeback.h>
40#include <linux/hugetlb.h>
41#include <linux/security.h>
42#include <linux/initrd.h>
43#include <linux/times.h>
44#include <linux/limits.h>
45#include <linux/dcache.h>
46#include <linux/syscalls.h>
c255d844
PM		 47#include <linux/nfs_fs.h>
48#include <linux/acpi.h>
1da177e4
LT		 49
50#include <asm/uaccess.h>
51#include <asm/processor.h>
52
529bf6be
DS		 53extern int proc_nr_files(ctl_table *table, int write, struct file *filp,
54 void __user *buffer, size_t *lenp, loff_t *ppos);
55
1da177e4
LT		 56#if defined(CONFIG_SYSCTL)
57
58/* External variables not in a header file. */
59extern int C_A_D;
60extern int sysctl_overcommit_memory;
61extern int sysctl_overcommit_ratio;
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;
1da177e4
LT		 75
76#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
77int unknown_nmi_panic;
78extern int proc_unknown_nmi_panic(ctl_table *, int, struct file *,
79 void __user *, size_t *, loff_t *);
80#endif
81
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
95#ifdef CONFIG_SYSVIPC
96extern size_t shm_ctlmax;
97extern size_t shm_ctlall;
98extern int shm_ctlmni;
99extern int msg_ctlmax;
100extern int msg_ctlmnb;
101extern int msg_ctlmni;
102extern int sem_ctls[];
103#endif
104
105#ifdef __sparc__
106extern char reboot_command [];
107extern int stop_a_enabled;
108extern int scons_pwroff;
109#endif
110
111#ifdef __hppa__
112extern int pwrsw_enabled;
113extern int unaligned_enabled;
114#endif
115
347a8dc3 116#ifdef CONFIG_S390
1da177e4
LT		 117#ifdef CONFIG_MATHEMU
118extern int sysctl_ieee_emulation_warnings;
119#endif
120extern int sysctl_userprocess_debug;
951f22d5 121extern int spin_retry;
1da177e4
LT		 122#endif
123
124extern int sysctl_hz_timer;
125
126#ifdef CONFIG_BSD_PROCESS_ACCT
127extern int acct_parm[];
128#endif
129
d2b176ed
JS		 130#ifdef CONFIG_IA64
131extern int no_unaligned_warning;
132#endif
133
1da177e4
LT		 134static int parse_table(int __user *, int, void __user *, size_t __user *, void __user *, size_t,
135 ctl_table *, void **);
136static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
137 void __user *buffer, size_t *lenp, loff_t *ppos);
138
139static ctl_table root_table[];
140static struct ctl_table_header root_table_header =
141 { root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
142
143static ctl_table kern_table[];
144static ctl_table vm_table[];
1da177e4
LT		 145static ctl_table proc_table[];
146static ctl_table fs_table[];
147static ctl_table debug_table[];
148static ctl_table dev_table[];
149extern ctl_table random_table[];
150#ifdef CONFIG_UNIX98_PTYS
151extern ctl_table pty_table[];
152#endif
0399cb08
RL		 153#ifdef CONFIG_INOTIFY
154extern ctl_table inotify_table[];
155#endif
1da177e4
LT		 156
157#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
158int sysctl_legacy_va_layout;
159#endif
160
161/* /proc declarations: */
162
163#ifdef CONFIG_PROC_FS
164
165static ssize_t proc_readsys(struct file *, char __user *, size_t, loff_t *);
166static ssize_t proc_writesys(struct file *, const char __user *, size_t, loff_t *);
167static int proc_opensys(struct inode *, struct file *);
168
169struct file_operations proc_sys_file_operations = {
170 .open = proc_opensys,
171 .read = proc_readsys,
172 .write = proc_writesys,
173};
174
175extern struct proc_dir_entry *proc_sys_root;
176
330d57fb 177static void register_proc_table(ctl_table *, struct proc_dir_entry *, void *);
1da177e4
LT		 178static void unregister_proc_table(ctl_table *, struct proc_dir_entry *);
179#endif
180
181/* The default sysctl tables: */
182
183static ctl_table root_table[] = {
184 {
185 .ctl_name = CTL_KERN,
186 .procname = "kernel",
187 .mode = 0555,
188 .child = kern_table,
189 },
190 {
191 .ctl_name = CTL_VM,
192 .procname = "vm",
193 .mode = 0555,
194 .child = vm_table,
195 },
196#ifdef CONFIG_NET
197 {
198 .ctl_name = CTL_NET,
199 .procname = "net",
200 .mode = 0555,
201 .child = net_table,
202 },
203#endif
204 {
205 .ctl_name = CTL_PROC,
206 .procname = "proc",
207 .mode = 0555,
208 .child = proc_table,
209 },
210 {
211 .ctl_name = CTL_FS,
212 .procname = "fs",
213 .mode = 0555,
214 .child = fs_table,
215 },
216 {
217 .ctl_name = CTL_DEBUG,
218 .procname = "debug",
219 .mode = 0555,
220 .child = debug_table,
221 },
222 {
223 .ctl_name = CTL_DEV,
224 .procname = "dev",
225 .mode = 0555,
226 .child = dev_table,
227 },
0eeca283 228
1da177e4
LT		 229 { .ctl_name = 0 }
230};
231
232static ctl_table kern_table[] = {
233 {
234 .ctl_name = KERN_OSTYPE,
235 .procname = "ostype",
236 .data = system_utsname.sysname,
237 .maxlen = sizeof(system_utsname.sysname),
238 .mode = 0444,
239 .proc_handler = &proc_doutsstring,
240 .strategy = &sysctl_string,
241 },
242 {
243 .ctl_name = KERN_OSRELEASE,
244 .procname = "osrelease",
245 .data = system_utsname.release,
246 .maxlen = sizeof(system_utsname.release),
247 .mode = 0444,
248 .proc_handler = &proc_doutsstring,
249 .strategy = &sysctl_string,
250 },
251 {
252 .ctl_name = KERN_VERSION,
253 .procname = "version",
254 .data = system_utsname.version,
255 .maxlen = sizeof(system_utsname.version),
256 .mode = 0444,
257 .proc_handler = &proc_doutsstring,
258 .strategy = &sysctl_string,
259 },
260 {
261 .ctl_name = KERN_NODENAME,
262 .procname = "hostname",
263 .data = system_utsname.nodename,
264 .maxlen = sizeof(system_utsname.nodename),
265 .mode = 0644,
266 .proc_handler = &proc_doutsstring,
267 .strategy = &sysctl_string,
268 },
269 {
270 .ctl_name = KERN_DOMAINNAME,
271 .procname = "domainname",
272 .data = system_utsname.domainname,
273 .maxlen = sizeof(system_utsname.domainname),
274 .mode = 0644,
275 .proc_handler = &proc_doutsstring,
276 .strategy = &sysctl_string,
277 },
278 {
279 .ctl_name = KERN_PANIC,
280 .procname = "panic",
281 .data = &panic_timeout,
282 .maxlen = sizeof(int),
283 .mode = 0644,
284 .proc_handler = &proc_dointvec,
285 },
286 {
287 .ctl_name = KERN_CORE_USES_PID,
288 .procname = "core_uses_pid",
289 .data = &core_uses_pid,
290 .maxlen = sizeof(int),
291 .mode = 0644,
292 .proc_handler = &proc_dointvec,
293 },
294 {
295 .ctl_name = KERN_CORE_PATTERN,
296 .procname = "core_pattern",
297 .data = core_pattern,
298 .maxlen = 64,
299 .mode = 0644,
300 .proc_handler = &proc_dostring,
301 .strategy = &sysctl_string,
302 },
303 {
304 .ctl_name = KERN_TAINTED,
305 .procname = "tainted",
306 .data = &tainted,
307 .maxlen = sizeof(int),
308 .mode = 0444,
309 .proc_handler = &proc_dointvec,
310 },
311 {
312 .ctl_name = KERN_CAP_BSET,
313 .procname = "cap-bound",
314 .data = &cap_bset,
315 .maxlen = sizeof(kernel_cap_t),
316 .mode = 0600,
317 .proc_handler = &proc_dointvec_bset,
318 },
319#ifdef CONFIG_BLK_DEV_INITRD
320 {
321 .ctl_name = KERN_REALROOTDEV,
322 .procname = "real-root-dev",
323 .data = &real_root_dev,
324 .maxlen = sizeof(int),
325 .mode = 0644,
326 .proc_handler = &proc_dointvec,
327 },
328#endif
329#ifdef __sparc__
330 {
331 .ctl_name = KERN_SPARC_REBOOT,
332 .procname = "reboot-cmd",
333 .data = reboot_command,
334 .maxlen = 256,
335 .mode = 0644,
336 .proc_handler = &proc_dostring,
337 .strategy = &sysctl_string,
338 },
339 {
340 .ctl_name = KERN_SPARC_STOP_A,
341 .procname = "stop-a",
342 .data = &stop_a_enabled,
343 .maxlen = sizeof (int),
344 .mode = 0644,
345 .proc_handler = &proc_dointvec,
346 },
347 {
348 .ctl_name = KERN_SPARC_SCONS_PWROFF,
349 .procname = "scons-poweroff",
350 .data = &scons_pwroff,
351 .maxlen = sizeof (int),
352 .mode = 0644,
353 .proc_handler = &proc_dointvec,
354 },
355#endif
356#ifdef __hppa__
357 {
358 .ctl_name = KERN_HPPA_PWRSW,
359 .procname = "soft-power",
360 .data = &pwrsw_enabled,
361 .maxlen = sizeof (int),
362 .mode = 0644,
363 .proc_handler = &proc_dointvec,
364 },
365 {
366 .ctl_name = KERN_HPPA_UNALIGNED,
367 .procname = "unaligned-trap",
368 .data = &unaligned_enabled,
369 .maxlen = sizeof (int),
370 .mode = 0644,
371 .proc_handler = &proc_dointvec,
372 },
373#endif
374 {
375 .ctl_name = KERN_CTLALTDEL,
376 .procname = "ctrl-alt-del",
377 .data = &C_A_D,
378 .maxlen = sizeof(int),
379 .mode = 0644,
380 .proc_handler = &proc_dointvec,
381 },
382 {
383 .ctl_name = KERN_PRINTK,
384 .procname = "printk",
385 .data = &console_loglevel,
386 .maxlen = 4*sizeof(int),
387 .mode = 0644,
388 .proc_handler = &proc_dointvec,
389 },
390#ifdef CONFIG_KMOD
391 {
392 .ctl_name = KERN_MODPROBE,
393 .procname = "modprobe",
394 .data = &modprobe_path,
395 .maxlen = KMOD_PATH_LEN,
396 .mode = 0644,
397 .proc_handler = &proc_dostring,
398 .strategy = &sysctl_string,
399 },
400#endif
401#ifdef CONFIG_HOTPLUG
402 {
403 .ctl_name = KERN_HOTPLUG,
404 .procname = "hotplug",
312c004d
KS		 405 .data = &uevent_helper,
406 .maxlen = UEVENT_HELPER_PATH_LEN,
1da177e4
LT		 407 .mode = 0644,
408 .proc_handler = &proc_dostring,
409 .strategy = &sysctl_string,
410 },
411#endif
412#ifdef CONFIG_CHR_DEV_SG
413 {
414 .ctl_name = KERN_SG_BIG_BUFF,
415 .procname = "sg-big-buff",
416 .data = &sg_big_buff,
417 .maxlen = sizeof (int),
418 .mode = 0444,
419 .proc_handler = &proc_dointvec,
420 },
421#endif
422#ifdef CONFIG_BSD_PROCESS_ACCT
423 {
424 .ctl_name = KERN_ACCT,
425 .procname = "acct",
426 .data = &acct_parm,
427 .maxlen = 3*sizeof(int),
428 .mode = 0644,
429 .proc_handler = &proc_dointvec,
430 },
431#endif
432#ifdef CONFIG_SYSVIPC
433 {
434 .ctl_name = KERN_SHMMAX,
435 .procname = "shmmax",
436 .data = &shm_ctlmax,
437 .maxlen = sizeof (size_t),
438 .mode = 0644,
439 .proc_handler = &proc_doulongvec_minmax,
440 },
441 {
442 .ctl_name = KERN_SHMALL,
443 .procname = "shmall",
444 .data = &shm_ctlall,
445 .maxlen = sizeof (size_t),
446 .mode = 0644,
447 .proc_handler = &proc_doulongvec_minmax,
448 },
449 {
450 .ctl_name = KERN_SHMMNI,
451 .procname = "shmmni",
452 .data = &shm_ctlmni,
453 .maxlen = sizeof (int),
454 .mode = 0644,
455 .proc_handler = &proc_dointvec,
456 },
457 {
458 .ctl_name = KERN_MSGMAX,
459 .procname = "msgmax",
460 .data = &msg_ctlmax,
461 .maxlen = sizeof (int),
462 .mode = 0644,
463 .proc_handler = &proc_dointvec,
464 },
465 {
466 .ctl_name = KERN_MSGMNI,
467 .procname = "msgmni",
468 .data = &msg_ctlmni,
469 .maxlen = sizeof (int),
470 .mode = 0644,
471 .proc_handler = &proc_dointvec,
472 },
473 {
474 .ctl_name = KERN_MSGMNB,
475 .procname = "msgmnb",
476 .data = &msg_ctlmnb,
477 .maxlen = sizeof (int),
478 .mode = 0644,
479 .proc_handler = &proc_dointvec,
480 },
481 {
482 .ctl_name = KERN_SEM,
483 .procname = "sem",
484 .data = &sem_ctls,
485 .maxlen = 4*sizeof (int),
486 .mode = 0644,
487 .proc_handler = &proc_dointvec,
488 },
489#endif
490#ifdef CONFIG_MAGIC_SYSRQ
491 {
492 .ctl_name = KERN_SYSRQ,
493 .procname = "sysrq",
494 .data = &sysrq_enabled,
495 .maxlen = sizeof (int),
496 .mode = 0644,
497 .proc_handler = &proc_dointvec,
498 },
499#endif
500 {
501 .ctl_name = KERN_CADPID,
502 .procname = "cad_pid",
503 .data = &cad_pid,
504 .maxlen = sizeof (int),
505 .mode = 0600,
506 .proc_handler = &proc_dointvec,
507 },
508 {
509 .ctl_name = KERN_MAX_THREADS,
510 .procname = "threads-max",
511 .data = &max_threads,
512 .maxlen = sizeof(int),
513 .mode = 0644,
514 .proc_handler = &proc_dointvec,
515 },
516 {
517 .ctl_name = KERN_RANDOM,
518 .procname = "random",
519 .mode = 0555,
520 .child = random_table,
521 },
522#ifdef CONFIG_UNIX98_PTYS
523 {
524 .ctl_name = KERN_PTY,
525 .procname = "pty",
526 .mode = 0555,
527 .child = pty_table,
528 },
529#endif
530 {
531 .ctl_name = KERN_OVERFLOWUID,
532 .procname = "overflowuid",
533 .data = &overflowuid,
534 .maxlen = sizeof(int),
535 .mode = 0644,
536 .proc_handler = &proc_dointvec_minmax,
537 .strategy = &sysctl_intvec,
538 .extra1 = &minolduid,
539 .extra2 = &maxolduid,
540 },
541 {
542 .ctl_name = KERN_OVERFLOWGID,
543 .procname = "overflowgid",
544 .data = &overflowgid,
545 .maxlen = sizeof(int),
546 .mode = 0644,
547 .proc_handler = &proc_dointvec_minmax,
548 .strategy = &sysctl_intvec,
549 .extra1 = &minolduid,
550 .extra2 = &maxolduid,
551 },
347a8dc3 552#ifdef CONFIG_S390
1da177e4
LT		 553#ifdef CONFIG_MATHEMU
554 {
555 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
556 .procname = "ieee_emulation_warnings",
557 .data = &sysctl_ieee_emulation_warnings,
558 .maxlen = sizeof(int),
559 .mode = 0644,
560 .proc_handler = &proc_dointvec,
561 },
562#endif
563#ifdef CONFIG_NO_IDLE_HZ
564 {
565 .ctl_name = KERN_HZ_TIMER,
566 .procname = "hz_timer",
567 .data = &sysctl_hz_timer,
568 .maxlen = sizeof(int),
569 .mode = 0644,
570 .proc_handler = &proc_dointvec,
571 },
572#endif
573 {
574 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
575 .procname = "userprocess_debug",
576 .data = &sysctl_userprocess_debug,
577 .maxlen = sizeof(int),
578 .mode = 0644,
579 .proc_handler = &proc_dointvec,
580 },
581#endif
582 {
583 .ctl_name = KERN_PIDMAX,
584 .procname = "pid_max",
585 .data = &pid_max,
586 .maxlen = sizeof (int),
587 .mode = 0644,
588 .proc_handler = &proc_dointvec_minmax,
589 .strategy = sysctl_intvec,
590 .extra1 = &pid_max_min,
591 .extra2 = &pid_max_max,
592 },
593 {
594 .ctl_name = KERN_PANIC_ON_OOPS,
595 .procname = "panic_on_oops",
596 .data = &panic_on_oops,
597 .maxlen = sizeof(int),
598 .mode = 0644,
599 .proc_handler = &proc_dointvec,
600 },
601 {
602 .ctl_name = KERN_PRINTK_RATELIMIT,
603 .procname = "printk_ratelimit",
604 .data = &printk_ratelimit_jiffies,
605 .maxlen = sizeof(int),
606 .mode = 0644,
607 .proc_handler = &proc_dointvec_jiffies,
608 .strategy = &sysctl_jiffies,
609 },
610 {
611 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
612 .procname = "printk_ratelimit_burst",
613 .data = &printk_ratelimit_burst,
614 .maxlen = sizeof(int),
615 .mode = 0644,
616 .proc_handler = &proc_dointvec,
617 },
618 {
619 .ctl_name = KERN_NGROUPS_MAX,
620 .procname = "ngroups_max",
621 .data = &ngroups_max,
622 .maxlen = sizeof (int),
623 .mode = 0444,
624 .proc_handler = &proc_dointvec,
625 },
626#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
627 {
628 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
629 .procname = "unknown_nmi_panic",
630 .data = &unknown_nmi_panic,
631 .maxlen = sizeof (int),
632 .mode = 0644,
633 .proc_handler = &proc_unknown_nmi_panic,
634 },
635#endif
636#if defined(CONFIG_X86)
637 {
638 .ctl_name = KERN_BOOTLOADER_TYPE,
639 .procname = "bootloader_type",
640 .data = &bootloader_type,
641 .maxlen = sizeof (int),
642 .mode = 0444,
643 .proc_handler = &proc_dointvec,
644 },
645#endif
7a9166e3 646#if defined(CONFIG_MMU)
1da177e4
LT		 647 {
648 .ctl_name = KERN_RANDOMIZE,
649 .procname = "randomize_va_space",
650 .data = &randomize_va_space,
651 .maxlen = sizeof(int),
652 .mode = 0644,
653 .proc_handler = &proc_dointvec,
654 },
7a9166e3 655#endif
0152fb37 656#if defined(CONFIG_S390) && defined(CONFIG_SMP)
951f22d5
MS		 657 {
658 .ctl_name = KERN_SPIN_RETRY,
659 .procname = "spin_retry",
660 .data = &spin_retry,
661 .maxlen = sizeof (int),
662 .mode = 0644,
663 .proc_handler = &proc_dointvec,
664 },
c255d844
PM		 665#endif
666#ifdef CONFIG_ACPI_SLEEP
667 {
668 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
669 .procname = "acpi_video_flags",
670 .data = &acpi_video_flags,
671 .maxlen = sizeof (unsigned long),
672 .mode = 0644,
7f99f06f 673 .proc_handler = &proc_doulongvec_minmax,
c255d844 674 },
d2b176ed
JS		 675#endif
676#ifdef CONFIG_IA64
677 {
678 .ctl_name = KERN_IA64_UNALIGNED,
679 .procname = "ignore-unaligned-usertrap",
680 .data = &no_unaligned_warning,
681 .maxlen = sizeof (int),
682 .mode = 0644,
683 .proc_handler = &proc_dointvec,
684 },
951f22d5 685#endif
1da177e4
LT		 686 { .ctl_name = 0 }
687};
688
689/* Constants for minimum and maximum testing in vm_table.
690 We use these as one-element integer vectors. */
691static int zero;
692static int one_hundred = 100;
693
694
695static ctl_table vm_table[] = {
696 {
697 .ctl_name = VM_OVERCOMMIT_MEMORY,
698 .procname = "overcommit_memory",
699 .data = &sysctl_overcommit_memory,
700 .maxlen = sizeof(sysctl_overcommit_memory),
701 .mode = 0644,
702 .proc_handler = &proc_dointvec,
703 },
704 {
705 .ctl_name = VM_OVERCOMMIT_RATIO,
706 .procname = "overcommit_ratio",
707 .data = &sysctl_overcommit_ratio,
708 .maxlen = sizeof(sysctl_overcommit_ratio),
709 .mode = 0644,
710 .proc_handler = &proc_dointvec,
711 },
712 {
713 .ctl_name = VM_PAGE_CLUSTER,
714 .procname = "page-cluster",
715 .data = &page_cluster,
716 .maxlen = sizeof(int),
717 .mode = 0644,
718 .proc_handler = &proc_dointvec,
719 },
720 {
721 .ctl_name = VM_DIRTY_BACKGROUND,
722 .procname = "dirty_background_ratio",
723 .data = &dirty_background_ratio,
724 .maxlen = sizeof(dirty_background_ratio),
725 .mode = 0644,
726 .proc_handler = &proc_dointvec_minmax,
727 .strategy = &sysctl_intvec,
728 .extra1 = &zero,
729 .extra2 = &one_hundred,
730 },
731 {
732 .ctl_name = VM_DIRTY_RATIO,
733 .procname = "dirty_ratio",
734 .data = &vm_dirty_ratio,
735 .maxlen = sizeof(vm_dirty_ratio),
736 .mode = 0644,
737 .proc_handler = &proc_dointvec_minmax,
738 .strategy = &sysctl_intvec,
739 .extra1 = &zero,
740 .extra2 = &one_hundred,
741 },
742 {
743 .ctl_name = VM_DIRTY_WB_CS,
744 .procname = "dirty_writeback_centisecs",
745 .data = &dirty_writeback_centisecs,
746 .maxlen = sizeof(dirty_writeback_centisecs),
747 .mode = 0644,
748 .proc_handler = &dirty_writeback_centisecs_handler,
749 },
750 {
751 .ctl_name = VM_DIRTY_EXPIRE_CS,
752 .procname = "dirty_expire_centisecs",
753 .data = &dirty_expire_centisecs,
754 .maxlen = sizeof(dirty_expire_centisecs),
755 .mode = 0644,
756 .proc_handler = &proc_dointvec,
757 },
758 {
759 .ctl_name = VM_NR_PDFLUSH_THREADS,
760 .procname = "nr_pdflush_threads",
761 .data = &nr_pdflush_threads,
762 .maxlen = sizeof nr_pdflush_threads,
763 .mode = 0444 /* read-only*/,
764 .proc_handler = &proc_dointvec,
765 },
766 {
767 .ctl_name = VM_SWAPPINESS,
768 .procname = "swappiness",
769 .data = &vm_swappiness,
770 .maxlen = sizeof(vm_swappiness),
771 .mode = 0644,
772 .proc_handler = &proc_dointvec_minmax,
773 .strategy = &sysctl_intvec,
774 .extra1 = &zero,
775 .extra2 = &one_hundred,
776 },
777#ifdef CONFIG_HUGETLB_PAGE
778 {
779 .ctl_name = VM_HUGETLB_PAGES,
780 .procname = "nr_hugepages",
781 .data = &max_huge_pages,
782 .maxlen = sizeof(unsigned long),
783 .mode = 0644,
784 .proc_handler = &hugetlb_sysctl_handler,
785 .extra1 = (void *)&hugetlb_zero,
786 .extra2 = (void *)&hugetlb_infinity,
787 },
788 {
789 .ctl_name = VM_HUGETLB_GROUP,
790 .procname = "hugetlb_shm_group",
791 .data = &sysctl_hugetlb_shm_group,
792 .maxlen = sizeof(gid_t),
793 .mode = 0644,
794 .proc_handler = &proc_dointvec,
795 },
796#endif
797 {
798 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
799 .procname = "lowmem_reserve_ratio",
800 .data = &sysctl_lowmem_reserve_ratio,
801 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
802 .mode = 0644,
803 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
804 .strategy = &sysctl_intvec,
805 },
9d0243bc
AM		 806 {
807 .ctl_name = VM_DROP_PAGECACHE,
808 .procname = "drop_caches",
809 .data = &sysctl_drop_caches,
810 .maxlen = sizeof(int),
811 .mode = 0644,
812 .proc_handler = drop_caches_sysctl_handler,
813 .strategy = &sysctl_intvec,
814 },
1da177e4
LT		 815 {
816 .ctl_name = VM_MIN_FREE_KBYTES,
817 .procname = "min_free_kbytes",
818 .data = &min_free_kbytes,
819 .maxlen = sizeof(min_free_kbytes),
820 .mode = 0644,
821 .proc_handler = &min_free_kbytes_sysctl_handler,
822 .strategy = &sysctl_intvec,
823 .extra1 = &zero,
824 },
8ad4b1fb
RS		 825 {
826 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
827 .procname = "percpu_pagelist_fraction",
828 .data = &percpu_pagelist_fraction,
829 .maxlen = sizeof(percpu_pagelist_fraction),
830 .mode = 0644,
831 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
832 .strategy = &sysctl_intvec,
833 .extra1 = &min_percpu_pagelist_fract,
834 },
1da177e4
LT		 835#ifdef CONFIG_MMU
836 {
837 .ctl_name = VM_MAX_MAP_COUNT,
838 .procname = "max_map_count",
839 .data = &sysctl_max_map_count,
840 .maxlen = sizeof(sysctl_max_map_count),
841 .mode = 0644,
842 .proc_handler = &proc_dointvec
843 },
844#endif
845 {
846 .ctl_name = VM_LAPTOP_MODE,
847 .procname = "laptop_mode",
848 .data = &laptop_mode,
849 .maxlen = sizeof(laptop_mode),
850 .mode = 0644,
851 .proc_handler = &proc_dointvec,
852 .strategy = &sysctl_intvec,
853 .extra1 = &zero,
854 },
855 {
856 .ctl_name = VM_BLOCK_DUMP,
857 .procname = "block_dump",
858 .data = &block_dump,
859 .maxlen = sizeof(block_dump),
860 .mode = 0644,
861 .proc_handler = &proc_dointvec,
862 .strategy = &sysctl_intvec,
863 .extra1 = &zero,
864 },
865 {
866 .ctl_name = VM_VFS_CACHE_PRESSURE,
867 .procname = "vfs_cache_pressure",
868 .data = &sysctl_vfs_cache_pressure,
869 .maxlen = sizeof(sysctl_vfs_cache_pressure),
870 .mode = 0644,
871 .proc_handler = &proc_dointvec,
872 .strategy = &sysctl_intvec,
873 .extra1 = &zero,
874 },
875#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
876 {
877 .ctl_name = VM_LEGACY_VA_LAYOUT,
878 .procname = "legacy_va_layout",
879 .data = &sysctl_legacy_va_layout,
880 .maxlen = sizeof(sysctl_legacy_va_layout),
881 .mode = 0644,
882 .proc_handler = &proc_dointvec,
883 .strategy = &sysctl_intvec,
884 .extra1 = &zero,
885 },
886#endif
887#ifdef CONFIG_SWAP
888 {
889 .ctl_name = VM_SWAP_TOKEN_TIMEOUT,
890 .procname = "swap_token_timeout",
891 .data = &swap_token_default_timeout,
892 .maxlen = sizeof(swap_token_default_timeout),
893 .mode = 0644,
894 .proc_handler = &proc_dointvec_jiffies,
895 .strategy = &sysctl_jiffies,
896 },
1743660b
CL		 897#endif
898#ifdef CONFIG_NUMA
899 {
900 .ctl_name = VM_ZONE_RECLAIM_MODE,
901 .procname = "zone_reclaim_mode",
902 .data = &zone_reclaim_mode,
903 .maxlen = sizeof(zone_reclaim_mode),
904 .mode = 0644,
905 .proc_handler = &proc_dointvec,
c84db23c
CL		 906 .strategy = &sysctl_intvec,
907 .extra1 = &zero,
1743660b 908 },
2a11ff06
CL		 909 {
910 .ctl_name = VM_ZONE_RECLAIM_INTERVAL,
911 .procname = "zone_reclaim_interval",
912 .data = &zone_reclaim_interval,
913 .maxlen = sizeof(zone_reclaim_interval),
914 .mode = 0644,
915 .proc_handler = &proc_dointvec_jiffies,
916 .strategy = &sysctl_jiffies,
917 },
1da177e4
LT		 918#endif
919 { .ctl_name = 0 }
920};
921
922static ctl_table proc_table[] = {
923 { .ctl_name = 0 }
924};
925
926static ctl_table fs_table[] = {
927 {
928 .ctl_name = FS_NRINODE,
929 .procname = "inode-nr",
930 .data = &inodes_stat,
931 .maxlen = 2*sizeof(int),
932 .mode = 0444,
933 .proc_handler = &proc_dointvec,
934 },
935 {
936 .ctl_name = FS_STATINODE,
937 .procname = "inode-state",
938 .data = &inodes_stat,
939 .maxlen = 7*sizeof(int),
940 .mode = 0444,
941 .proc_handler = &proc_dointvec,
942 },
943 {
944 .ctl_name = FS_NRFILE,
945 .procname = "file-nr",
946 .data = &files_stat,
947 .maxlen = 3*sizeof(int),
948 .mode = 0444,
529bf6be 949 .proc_handler = &proc_nr_files,
1da177e4
LT		 950 },
951 {
952 .ctl_name = FS_MAXFILE,
953 .procname = "file-max",
954 .data = &files_stat.max_files,
955 .maxlen = sizeof(int),
956 .mode = 0644,
957 .proc_handler = &proc_dointvec,
958 },
959 {
960 .ctl_name = FS_DENTRY,
961 .procname = "dentry-state",
962 .data = &dentry_stat,
963 .maxlen = 6*sizeof(int),
964 .mode = 0444,
965 .proc_handler = &proc_dointvec,
966 },
967 {
968 .ctl_name = FS_OVERFLOWUID,
969 .procname = "overflowuid",
970 .data = &fs_overflowuid,
971 .maxlen = sizeof(int),
972 .mode = 0644,
973 .proc_handler = &proc_dointvec_minmax,
974 .strategy = &sysctl_intvec,
975 .extra1 = &minolduid,
976 .extra2 = &maxolduid,
977 },
978 {
979 .ctl_name = FS_OVERFLOWGID,
980 .procname = "overflowgid",
981 .data = &fs_overflowgid,
982 .maxlen = sizeof(int),
983 .mode = 0644,
984 .proc_handler = &proc_dointvec_minmax,
985 .strategy = &sysctl_intvec,
986 .extra1 = &minolduid,
987 .extra2 = &maxolduid,
988 },
989 {
990 .ctl_name = FS_LEASES,
991 .procname = "leases-enable",
992 .data = &leases_enable,
993 .maxlen = sizeof(int),
994 .mode = 0644,
995 .proc_handler = &proc_dointvec,
996 },
997#ifdef CONFIG_DNOTIFY
998 {
999 .ctl_name = FS_DIR_NOTIFY,
1000 .procname = "dir-notify-enable",
1001 .data = &dir_notify_enable,
1002 .maxlen = sizeof(int),
1003 .mode = 0644,
1004 .proc_handler = &proc_dointvec,
1005 },
1006#endif
1007#ifdef CONFIG_MMU
1008 {
1009 .ctl_name = FS_LEASE_TIME,
1010 .procname = "lease-break-time",
1011 .data = &lease_break_time,
1012 .maxlen = sizeof(int),
1013 .mode = 0644,
1014 .proc_handler = &proc_dointvec,
1015 },
1016 {
1017 .ctl_name = FS_AIO_NR,
1018 .procname = "aio-nr",
1019 .data = &aio_nr,
1020 .maxlen = sizeof(aio_nr),
1021 .mode = 0444,
d55b5fda 1022 .proc_handler = &proc_doulongvec_minmax,
1da177e4
LT		 1023 },
1024 {
1025 .ctl_name = FS_AIO_MAX_NR,
1026 .procname = "aio-max-nr",
1027 .data = &aio_max_nr,
1028 .maxlen = sizeof(aio_max_nr),
1029 .mode = 0644,
d55b5fda 1030 .proc_handler = &proc_doulongvec_minmax,
1da177e4 1031 },
0399cb08
RL		 1032#ifdef CONFIG_INOTIFY
1033 {
1034 .ctl_name = FS_INOTIFY,
1035 .procname = "inotify",
1036 .mode = 0555,
1037 .child = inotify_table,
1038 },
1039#endif
1da177e4 1040#endif
d6e71144
AC		 1041 {
1042 .ctl_name = KERN_SETUID_DUMPABLE,
1043 .procname = "suid_dumpable",
1044 .data = &suid_dumpable,
1045 .maxlen = sizeof(int),
1046 .mode = 0644,
1047 .proc_handler = &proc_dointvec,
1048 },
1da177e4
LT		 1049 { .ctl_name = 0 }
1050};
1051
1052static ctl_table debug_table[] = {
1053 { .ctl_name = 0 }
1054};
1055
1056static ctl_table dev_table[] = {
1057 { .ctl_name = 0 }
0eeca283 1058};
1da177e4
LT		 1059
1060extern void init_irq_proc (void);
1061
330d57fb
AV		 1062static DEFINE_SPINLOCK(sysctl_lock);
1063
1064/* called under sysctl_lock */
1065static int use_table(struct ctl_table_header *p)
1066{
1067 if (unlikely(p->unregistering))
1068 return 0;
1069 p->used++;
1070 return 1;
1071}
1072
1073/* called under sysctl_lock */
1074static void unuse_table(struct ctl_table_header *p)
1075{
1076 if (!--p->used)
1077 if (unlikely(p->unregistering))
1078 complete(p->unregistering);
1079}
1080
1081/* called under sysctl_lock, will reacquire if has to wait */
1082static void start_unregistering(struct ctl_table_header *p)
1083{
1084 /*
1085 * if p->used is 0, nobody will ever touch that entry again;
1086 * we'll eliminate all paths to it before dropping sysctl_lock
1087 */
1088 if (unlikely(p->used)) {
1089 struct completion wait;
1090 init_completion(&wait);
1091 p->unregistering = &wait;
1092 spin_unlock(&sysctl_lock);
1093 wait_for_completion(&wait);
1094 spin_lock(&sysctl_lock);
1095 }
1096 /*
1097 * do not remove from the list until nobody holds it; walking the
1098 * list in do_sysctl() relies on that.
1099 */
1100 list_del_init(&p->ctl_entry);
1101}
1102
1da177e4
LT		 1103void __init sysctl_init(void)
1104{
1105#ifdef CONFIG_PROC_FS
330d57fb 1106 register_proc_table(root_table, proc_sys_root, &root_table_header);
1da177e4
LT		 1107 init_irq_proc();
1108#endif
1109}
1110
1111int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1112 void __user *newval, size_t newlen)
1113{
1114 struct list_head *tmp;
330d57fb 1115 int error = -ENOTDIR;
1da177e4
LT		 1116
1117 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1118 return -ENOTDIR;
1119 if (oldval) {
1120 int old_len;
1121 if (!oldlenp || get_user(old_len, oldlenp))
1122 return -EFAULT;
1123 }
330d57fb 1124 spin_lock(&sysctl_lock);
1da177e4
LT		 1125 tmp = &root_table_header.ctl_entry;
1126 do {
1127 struct ctl_table_header *head =
1128 list_entry(tmp, struct ctl_table_header, ctl_entry);
1129 void *context = NULL;
330d57fb
AV		 1130
1131 if (!use_table(head))
1132 continue;
1133
1134 spin_unlock(&sysctl_lock);
1135
1136 error = parse_table(name, nlen, oldval, oldlenp,
1da177e4
LT		 1137 newval, newlen, head->ctl_table,
1138 &context);
5a6b454f 1139 kfree(context);
330d57fb
AV		 1140
1141 spin_lock(&sysctl_lock);
1142 unuse_table(head);
1da177e4 1143 if (error != -ENOTDIR)
330d57fb
AV		 1144 break;
1145 } while ((tmp = tmp->next) != &root_table_header.ctl_entry);
1146 spin_unlock(&sysctl_lock);
1147 return error;
1da177e4
LT		 1148}
1149
1150asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1151{
1152 struct __sysctl_args tmp;
1153 int error;
1154
1155 if (copy_from_user(&tmp, args, sizeof(tmp)))
1156 return -EFAULT;
1157
1158 lock_kernel();
1159 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1160 tmp.newval, tmp.newlen);
1161 unlock_kernel();
1162 return error;
1163}
1164
1165/*
1166 * ctl_perm does NOT grant the superuser all rights automatically, because
1167 * some sysctl variables are readonly even to root.
1168 */
1169
1170static int test_perm(int mode, int op)
1171{
1172 if (!current->euid)
1173 mode >>= 6;
1174 else if (in_egroup_p(0))
1175 mode >>= 3;
1176 if ((mode & op & 0007) == op)
1177 return 0;
1178 return -EACCES;
1179}
1180
1181static inline int ctl_perm(ctl_table *table, int op)
1182{
1183 int error;
1184 error = security_sysctl(table, op);
1185 if (error)
1186 return error;
1187 return test_perm(table->mode, op);
1188}
1189
1190static int parse_table(int __user *name, int nlen,
1191 void __user *oldval, size_t __user *oldlenp,
1192 void __user *newval, size_t newlen,
1193 ctl_table *table, void **context)
1194{
1195 int n;
1196repeat:
1197 if (!nlen)
1198 return -ENOTDIR;
1199 if (get_user(n, name))
1200 return -EFAULT;
1201 for ( ; table->ctl_name; table++) {
1202 if (n == table->ctl_name || table->ctl_name == CTL_ANY) {
1203 int error;
1204 if (table->child) {
1205 if (ctl_perm(table, 001))
1206 return -EPERM;
1207 if (table->strategy) {
1208 error = table->strategy(
1209 table, name, nlen,
1210 oldval, oldlenp,
1211 newval, newlen, context);
1212 if (error)
1213 return error;
1214 }
1215 name++;
1216 nlen--;
1217 table = table->child;
1218 goto repeat;
1219 }
1220 error = do_sysctl_strategy(table, name, nlen,
1221 oldval, oldlenp,
1222 newval, newlen, context);
1223 return error;
1224 }
1225 }
1226 return -ENOTDIR;
1227}
1228
1229/* Perform the actual read/write of a sysctl table entry. */
1230int do_sysctl_strategy (ctl_table *table,
1231 int __user *name, int nlen,
1232 void __user *oldval, size_t __user *oldlenp,
1233 void __user *newval, size_t newlen, void **context)
1234{
1235 int op = 0, rc;
1236 size_t len;
1237
1238 if (oldval)
1239 op |= 004;
1240 if (newval)
1241 op |= 002;
1242 if (ctl_perm(table, op))
1243 return -EPERM;
1244
1245 if (table->strategy) {
1246 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1247 newval, newlen, context);
1248 if (rc < 0)
1249 return rc;
1250 if (rc > 0)
1251 return 0;
1252 }
1253
1254 /* If there is no strategy routine, or if the strategy returns
1255 * zero, proceed with automatic r/w */
1256 if (table->data && table->maxlen) {
1257 if (oldval && oldlenp) {
1258 if (get_user(len, oldlenp))
1259 return -EFAULT;
1260 if (len) {
1261 if (len > table->maxlen)
1262 len = table->maxlen;
1263 if(copy_to_user(oldval, table->data, len))
1264 return -EFAULT;
1265 if(put_user(len, oldlenp))
1266 return -EFAULT;
1267 }
1268 }
1269 if (newval && newlen) {
1270 len = newlen;
1271 if (len > table->maxlen)
1272 len = table->maxlen;
1273 if(copy_from_user(table->data, newval, len))
1274 return -EFAULT;
1275 }
1276 }
1277 return 0;
1278}
1279
1280/**
1281 * register_sysctl_table - register a sysctl hierarchy
1282 * @table: the top-level table structure
1283 * @insert_at_head: whether the entry should be inserted in front or at the end
1284 *
1285 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1286 * array. An entry with a ctl_name of 0 terminates the table.
1287 *
1288 * The members of the &ctl_table structure are used as follows:
1289 *
1290 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1291 * must be unique within that level of sysctl
1292 *
1293 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1294 * enter a sysctl file
1295 *
1296 * data - a pointer to data for use by proc_handler
1297 *
1298 * maxlen - the maximum size in bytes of the data
1299 *
1300 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1301 *
1302 * child - a pointer to the child sysctl table if this entry is a directory, or
1303 * %NULL.
1304 *
1305 * proc_handler - the text handler routine (described below)
1306 *
1307 * strategy - the strategy routine (described below)
1308 *
1309 * de - for internal use by the sysctl routines
1310 *
1311 * extra1, extra2 - extra pointers usable by the proc handler routines
1312 *
1313 * Leaf nodes in the sysctl tree will be represented by a single file
1314 * under /proc; non-leaf nodes will be represented by directories.
1315 *
1316 * sysctl(2) can automatically manage read and write requests through
1317 * the sysctl table. The data and maxlen fields of the ctl_table
1318 * struct enable minimal validation of the values being written to be
1319 * performed, and the mode field allows minimal authentication.
1320 *
1321 * More sophisticated management can be enabled by the provision of a
1322 * strategy routine with the table entry. This will be called before
1323 * any automatic read or write of the data is performed.
1324 *
1325 * The strategy routine may return
1326 *
1327 * < 0 - Error occurred (error is passed to user process)
1328 *
1329 * 0 - OK - proceed with automatic read or write.
1330 *
1331 * > 0 - OK - read or write has been done by the strategy routine, so
1332 * return immediately.
1333 *
1334 * There must be a proc_handler routine for any terminal nodes
1335 * mirrored under /proc/sys (non-terminals are handled by a built-in
1336 * directory handler). Several default handlers are available to
1337 * cover common cases -
1338 *
1339 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1340 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1341 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1342 *
1343 * It is the handler's job to read the input buffer from user memory
1344 * and process it. The handler should return 0 on success.
1345 *
1346 * This routine returns %NULL on a failure to register, and a pointer
1347 * to the table header on success.
1348 */
1349struct ctl_table_header *register_sysctl_table(ctl_table * table,
1350 int insert_at_head)
1351{
1352 struct ctl_table_header *tmp;
1353 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1354 if (!tmp)
1355 return NULL;
1356 tmp->ctl_table = table;
1357 INIT_LIST_HEAD(&tmp->ctl_entry);
330d57fb
AV		 1358 tmp->used = 0;
1359 tmp->unregistering = NULL;
1360 spin_lock(&sysctl_lock);
1da177e4
LT		 1361 if (insert_at_head)
1362 list_add(&tmp->ctl_entry, &root_table_header.ctl_entry);
1363 else
1364 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
330d57fb 1365 spin_unlock(&sysctl_lock);
1da177e4 1366#ifdef CONFIG_PROC_FS
330d57fb 1367 register_proc_table(table, proc_sys_root, tmp);
1da177e4
LT		 1368#endif
1369 return tmp;
1370}
1371
1372/**
1373 * unregister_sysctl_table - unregister a sysctl table hierarchy
1374 * @header: the header returned from register_sysctl_table
1375 *
1376 * Unregisters the sysctl table and all children. proc entries may not
1377 * actually be removed until they are no longer used by anyone.
1378 */
1379void unregister_sysctl_table(struct ctl_table_header * header)
1380{
330d57fb
AV		 1381 might_sleep();
1382 spin_lock(&sysctl_lock);
1383 start_unregistering(header);
1da177e4
LT		 1384#ifdef CONFIG_PROC_FS
1385 unregister_proc_table(header->ctl_table, proc_sys_root);
1386#endif
330d57fb 1387 spin_unlock(&sysctl_lock);
1da177e4
LT		 1388 kfree(header);
1389}
1390
1391/*
1392 * /proc/sys support
1393 */
1394
1395#ifdef CONFIG_PROC_FS
1396
1397/* Scan the sysctl entries in table and add them all into /proc */
330d57fb 1398static void register_proc_table(ctl_table * table, struct proc_dir_entry *root, void *set)
1da177e4
LT		 1399{
1400 struct proc_dir_entry *de;
1401 int len;
1402 mode_t mode;
1403
1404 for (; table->ctl_name; table++) {
1405 /* Can't do anything without a proc name. */
1406 if (!table->procname)
1407 continue;
1408 /* Maybe we can't do anything with it... */
1409 if (!table->proc_handler && !table->child) {
1410 printk(KERN_WARNING "SYSCTL: Can't register %s\n",
1411 table->procname);
1412 continue;
1413 }
1414
1415 len = strlen(table->procname);
1416 mode = table->mode;
1417
1418 de = NULL;
1419 if (table->proc_handler)
1420 mode |= S_IFREG;
1421 else {
1422 mode |= S_IFDIR;
1423 for (de = root->subdir; de; de = de->next) {
1424 if (proc_match(len, table->procname, de))
1425 break;
1426 }
1427 /* If the subdir exists already, de is non-NULL */
1428 }
1429
1430 if (!de) {
1431 de = create_proc_entry(table->procname, mode, root);
1432 if (!de)
1433 continue;
330d57fb 1434 de->set = set;
1da177e4
LT		 1435 de->data = (void *) table;
1436 if (table->proc_handler)
1437 de->proc_fops = &proc_sys_file_operations;
1438 }
1439 table->de = de;
1440 if (de->mode & S_IFDIR)
330d57fb 1441 register_proc_table(table->child, de, set);
1da177e4
LT		 1442 }
1443}
1444
1445/*
1446 * Unregister a /proc sysctl table and any subdirectories.
1447 */
1448static void unregister_proc_table(ctl_table * table, struct proc_dir_entry *root)
1449{
1450 struct proc_dir_entry *de;
1451 for (; table->ctl_name; table++) {
1452 if (!(de = table->de))
1453 continue;
1454 if (de->mode & S_IFDIR) {
1455 if (!table->child) {
1456 printk (KERN_ALERT "Help - malformed sysctl tree on free\n");
1457 continue;
1458 }
1459 unregister_proc_table(table->child, de);
1460
1461 /* Don't unregister directories which still have entries.. */
1462 if (de->subdir)
1463 continue;
1464 }
1465
330d57fb
AV		 1466 /*
1467 * In any case, mark the entry as goner; we'll keep it
1468 * around if it's busy, but we'll know to do nothing with
1469 * its fields. We are under sysctl_lock here.
1470 */
1471 de->data = NULL;
1472
1da177e4
LT		 1473 /* Don't unregister proc entries that are still being used.. */
1474 if (atomic_read(&de->count))
1475 continue;
1476
1477 table->de = NULL;
1478 remove_proc_entry(table->procname, root);
1479 }
1480}
1481
1482static ssize_t do_rw_proc(int write, struct file * file, char __user * buf,
1483 size_t count, loff_t *ppos)
1484{
1485 int op;
330d57fb 1486 struct proc_dir_entry *de = PDE(file->f_dentry->d_inode);
1da177e4
LT		 1487 struct ctl_table *table;
1488 size_t res;
330d57fb 1489 ssize_t error = -ENOTDIR;
1da177e4 1490
330d57fb
AV		 1491 spin_lock(&sysctl_lock);
1492 if (de && de->data && use_table(de->set)) {
1493 /*
1494 * at that point we know that sysctl was not unregistered
1495 * and won't be until we finish
1496 */
1497 spin_unlock(&sysctl_lock);
1498 table = (struct ctl_table *) de->data;
1499 if (!table || !table->proc_handler)
1500 goto out;
1501 error = -EPERM;
1502 op = (write ? 002 : 004);
1503 if (ctl_perm(table, op))
1504 goto out;
1505
1506 /* careful: calling conventions are nasty here */
1507 res = count;
1508 error = (*table->proc_handler)(table, write, file,
1509 buf, &res, ppos);
1510 if (!error)
1511 error = res;
1512 out:
1513 spin_lock(&sysctl_lock);
1514 unuse_table(de->set);
1515 }
1516 spin_unlock(&sysctl_lock);
1517 return error;
1da177e4
LT		 1518}
1519
1520static int proc_opensys(struct inode *inode, struct file *file)
1521{
1522 if (file->f_mode & FMODE_WRITE) {
1523 /*
1524 * sysctl entries that are not writable,
1525 * are _NOT_ writable, capabilities or not.
1526 */
1527 if (!(inode->i_mode & S_IWUSR))
1528 return -EPERM;
1529 }
1530
1531 return 0;
1532}
1533
1534static ssize_t proc_readsys(struct file * file, char __user * buf,
1535 size_t count, loff_t *ppos)
1536{
1537 return do_rw_proc(0, file, buf, count, ppos);
1538}
1539
1540static ssize_t proc_writesys(struct file * file, const char __user * buf,
1541 size_t count, loff_t *ppos)
1542{
1543 return do_rw_proc(1, file, (char __user *) buf, count, ppos);
1544}
1545
1546/**
1547 * proc_dostring - read a string sysctl
1548 * @table: the sysctl table
1549 * @write: %TRUE if this is a write to the sysctl file
1550 * @filp: the file structure
1551 * @buffer: the user buffer
1552 * @lenp: the size of the user buffer
1553 * @ppos: file position
1554 *
1555 * Reads/writes a string from/to the user buffer. If the kernel
1556 * buffer provided is not large enough to hold the string, the
1557 * string is truncated. The copied string is %NULL-terminated.
1558 * If the string is being read by the user process, it is copied
1559 * and a newline '\n' is added. It is truncated if the buffer is
1560 * not large enough.
1561 *
1562 * Returns 0 on success.
1563 */
1564int proc_dostring(ctl_table *table, int write, struct file *filp,
1565 void __user *buffer, size_t *lenp, loff_t *ppos)
1566{
1567 size_t len;
1568 char __user *p;
1569 char c;
1570
1571 if (!table->data || !table->maxlen || !*lenp ||
1572 (*ppos && !write)) {
1573 *lenp = 0;
1574 return 0;
1575 }
1576
1577 if (write) {
1578 len = 0;
1579 p = buffer;
1580 while (len < *lenp) {
1581 if (get_user(c, p++))
1582 return -EFAULT;
1583 if (c == 0 || c == '\n')
1584 break;
1585 len++;
1586 }
1587 if (len >= table->maxlen)
1588 len = table->maxlen-1;
1589 if(copy_from_user(table->data, buffer, len))
1590 return -EFAULT;
1591 ((char *) table->data)[len] = 0;
1592 *ppos += *lenp;
1593 } else {
1594 len = strlen(table->data);
1595 if (len > table->maxlen)
1596 len = table->maxlen;
1597 if (len > *lenp)
1598 len = *lenp;
1599 if (len)
1600 if(copy_to_user(buffer, table->data, len))
1601 return -EFAULT;
1602 if (len < *lenp) {
1603 if(put_user('\n', ((char __user *) buffer) + len))
1604 return -EFAULT;
1605 len++;
1606 }
1607 *lenp = len;
1608 *ppos += len;
1609 }
1610 return 0;
1611}
1612
1613/*
1614 * Special case of dostring for the UTS structure. This has locks
1615 * to observe. Should this be in kernel/sys.c ????
1616 */
1617
1618static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
1619 void __user *buffer, size_t *lenp, loff_t *ppos)
1620{
1621 int r;
1622
1623 if (!write) {
1624 down_read(&uts_sem);
1625 r=proc_dostring(table,0,filp,buffer,lenp, ppos);
1626 up_read(&uts_sem);
1627 } else {
1628 down_write(&uts_sem);
1629 r=proc_dostring(table,1,filp,buffer,lenp, ppos);
1630 up_write(&uts_sem);
1631 }
1632 return r;
1633}
1634
1635static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1636 int *valp,
1637 int write, void *data)
1638{
1639 if (write) {
1640 *valp = *negp ? -*lvalp : *lvalp;
1641 } else {
1642 int val = *valp;
1643 if (val < 0) {
1644 *negp = -1;
1645 *lvalp = (unsigned long)-val;
1646 } else {
1647 *negp = 0;
1648 *lvalp = (unsigned long)val;
1649 }
1650 }
1651 return 0;
1652}
1653
1654static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1655 void __user *buffer, size_t *lenp, loff_t *ppos,
1656 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1657 int write, void *data),
1658 void *data)
1659{
1660#define TMPBUFLEN 21
1661 int *i, vleft, first=1, neg, val;
1662 unsigned long lval;
1663 size_t left, len;
1664
1665 char buf[TMPBUFLEN], *p;
1666 char __user *s = buffer;
1667
1668 if (!table->data || !table->maxlen || !*lenp ||
1669 (*ppos && !write)) {
1670 *lenp = 0;
1671 return 0;
1672 }
1673
1674 i = (int *) table->data;
1675 vleft = table->maxlen / sizeof(*i);
1676 left = *lenp;
1677
1678 if (!conv)
1679 conv = do_proc_dointvec_conv;
1680
1681 for (; left && vleft--; i++, first=0) {
1682 if (write) {
1683 while (left) {
1684 char c;
1685 if (get_user(c, s))
1686 return -EFAULT;
1687 if (!isspace(c))
1688 break;
1689 left--;
1690 s++;
1691 }
1692 if (!left)
1693 break;
1694 neg = 0;
1695 len = left;
1696 if (len > sizeof(buf) - 1)
1697 len = sizeof(buf) - 1;
1698 if (copy_from_user(buf, s, len))
1699 return -EFAULT;
1700 buf[len] = 0;
1701 p = buf;
1702 if (*p == '-' && left > 1) {
1703 neg = 1;
1704 left--, p++;
1705 }
1706 if (*p < '0' || *p > '9')
1707 break;
1708
1709 lval = simple_strtoul(p, &p, 0);
1710
1711 len = p-buf;
1712 if ((len < left) && *p && !isspace(*p))
1713 break;
1714 if (neg)
1715 val = -val;
1716 s += len;
1717 left -= len;
1718
1719 if (conv(&neg, &lval, i, 1, data))
1720 break;
1721 } else {
1722 p = buf;
1723 if (!first)
1724 *p++ = '\t';
1725
1726 if (conv(&neg, &lval, i, 0, data))
1727 break;
1728
1729 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1730 len = strlen(buf);
1731 if (len > left)
1732 len = left;
1733 if(copy_to_user(s, buf, len))
1734 return -EFAULT;
1735 left -= len;
1736 s += len;
1737 }
1738 }
1739
1740 if (!write && !first && left) {
1741 if(put_user('\n', s))
1742 return -EFAULT;
1743 left--, s++;
1744 }
1745 if (write) {
1746 while (left) {
1747 char c;
1748 if (get_user(c, s++))
1749 return -EFAULT;
1750 if (!isspace(c))
1751 break;
1752 left--;
1753 }
1754 }
1755 if (write && first)
1756 return -EINVAL;
1757 *lenp -= left;
1758 *ppos += *lenp;
1759 return 0;
1760#undef TMPBUFLEN
1761}
1762
1763/**
1764 * proc_dointvec - read a vector of integers
1765 * @table: the sysctl table
1766 * @write: %TRUE if this is a write to the sysctl file
1767 * @filp: the file structure
1768 * @buffer: the user buffer
1769 * @lenp: the size of the user buffer
1770 * @ppos: file position
1771 *
1772 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1773 * values from/to the user buffer, treated as an ASCII string.
1774 *
1775 * Returns 0 on success.
1776 */
1777int proc_dointvec(ctl_table *table, int write, struct file *filp,
1778 void __user *buffer, size_t *lenp, loff_t *ppos)
1779{
1780 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1781 NULL,NULL);
1782}
1783
1784#define OP_SET 0
1785#define OP_AND 1
1786#define OP_OR 2
1787#define OP_MAX 3
1788#define OP_MIN 4
1789
1790static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1791 int *valp,
1792 int write, void *data)
1793{
1794 int op = *(int *)data;
1795 if (write) {
1796 int val = *negp ? -*lvalp : *lvalp;
1797 switch(op) {
1798 case OP_SET: *valp = val; break;
1799 case OP_AND: *valp &= val; break;
1800 case OP_OR: *valp |= val; break;
1801 case OP_MAX: if(*valp < val)
1802 *valp = val;
1803 break;
1804 case OP_MIN: if(*valp > val)
1805 *valp = val;
1806 break;
1807 }
1808 } else {
1809 int val = *valp;
1810 if (val < 0) {
1811 *negp = -1;
1812 *lvalp = (unsigned long)-val;
1813 } else {
1814 *negp = 0;
1815 *lvalp = (unsigned long)val;
1816 }
1817 }
1818 return 0;
1819}
1820
1821/*
1822 * init may raise the set.
1823 */
1824
1825int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1826 void __user *buffer, size_t *lenp, loff_t *ppos)
1827{
1828 int op;
1829
1830 if (!capable(CAP_SYS_MODULE)) {
1831 return -EPERM;
1832 }
1833
1834 op = (current->pid == 1) ? OP_SET : OP_AND;
1835 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1836 do_proc_dointvec_bset_conv,&op);
1837}
1838
1839struct do_proc_dointvec_minmax_conv_param {
1840 int *min;
1841 int *max;
1842};
1843
1844static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1845 int *valp,
1846 int write, void *data)
1847{
1848 struct do_proc_dointvec_minmax_conv_param *param = data;
1849 if (write) {
1850 int val = *negp ? -*lvalp : *lvalp;
1851 if ((param->min && *param->min > val) ||
1852 (param->max && *param->max < val))
1853 return -EINVAL;
1854 *valp = val;
1855 } else {
1856 int val = *valp;
1857 if (val < 0) {
1858 *negp = -1;
1859 *lvalp = (unsigned long)-val;
1860 } else {
1861 *negp = 0;
1862 *lvalp = (unsigned long)val;
1863 }
1864 }
1865 return 0;
1866}
1867
1868/**
1869 * proc_dointvec_minmax - read a vector of integers with min/max values
1870 * @table: the sysctl table
1871 * @write: %TRUE if this is a write to the sysctl file
1872 * @filp: the file structure
1873 * @buffer: the user buffer
1874 * @lenp: the size of the user buffer
1875 * @ppos: file position
1876 *
1877 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1878 * values from/to the user buffer, treated as an ASCII string.
1879 *
1880 * This routine will ensure the values are within the range specified by
1881 * table->extra1 (min) and table->extra2 (max).
1882 *
1883 * Returns 0 on success.
1884 */
1885int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1886 void __user *buffer, size_t *lenp, loff_t *ppos)
1887{
1888 struct do_proc_dointvec_minmax_conv_param param = {
1889 .min = (int *) table->extra1,
1890 .max = (int *) table->extra2,
1891 };
1892 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1893 do_proc_dointvec_minmax_conv, &param);
1894}
1895
1896static int do_proc_doulongvec_minmax(ctl_table *table, int write,
1897 struct file *filp,
1898 void __user *buffer,
1899 size_t *lenp, loff_t *ppos,
1900 unsigned long convmul,
1901 unsigned long convdiv)
1902{
1903#define TMPBUFLEN 21
1904 unsigned long *i, *min, *max, val;
1905 int vleft, first=1, neg;
1906 size_t len, left;
1907 char buf[TMPBUFLEN], *p;
1908 char __user *s = buffer;
1909
1910 if (!table->data || !table->maxlen || !*lenp ||
1911 (*ppos && !write)) {
1912 *lenp = 0;
1913 return 0;
1914 }
1915
1916 i = (unsigned long *) table->data;
1917 min = (unsigned long *) table->extra1;
1918 max = (unsigned long *) table->extra2;
1919 vleft = table->maxlen / sizeof(unsigned long);
1920 left = *lenp;
1921
1922 for (; left && vleft--; i++, min++, max++, first=0) {
1923 if (write) {
1924 while (left) {
1925 char c;
1926 if (get_user(c, s))
1927 return -EFAULT;
1928 if (!isspace(c))
1929 break;
1930 left--;
1931 s++;
1932 }
1933 if (!left)
1934 break;
1935 neg = 0;
1936 len = left;
1937 if (len > TMPBUFLEN-1)
1938 len = TMPBUFLEN-1;
1939 if (copy_from_user(buf, s, len))
1940 return -EFAULT;
1941 buf[len] = 0;
1942 p = buf;
1943 if (*p == '-' && left > 1) {
1944 neg = 1;
1945 left--, p++;
1946 }
1947 if (*p < '0' || *p > '9')
1948 break;
1949 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
1950 len = p-buf;
1951 if ((len < left) && *p && !isspace(*p))
1952 break;
1953 if (neg)
1954 val = -val;
1955 s += len;
1956 left -= len;
1957
1958 if(neg)
1959 continue;
1960 if ((min && val < *min) || (max && val > *max))
1961 continue;
1962 *i = val;
1963 } else {
1964 p = buf;
1965 if (!first)
1966 *p++ = '\t';
1967 sprintf(p, "%lu", convdiv * (*i) / convmul);
1968 len = strlen(buf);
1969 if (len > left)
1970 len = left;
1971 if(copy_to_user(s, buf, len))
1972 return -EFAULT;
1973 left -= len;
1974 s += len;
1975 }
1976 }
1977
1978 if (!write && !first && left) {
1979 if(put_user('\n', s))
1980 return -EFAULT;
1981 left--, s++;
1982 }
1983 if (write) {
1984 while (left) {
1985 char c;
1986 if (get_user(c, s++))
1987 return -EFAULT;
1988 if (!isspace(c))
1989 break;
1990 left--;
1991 }
1992 }
1993 if (write && first)
1994 return -EINVAL;
1995 *lenp -= left;
1996 *ppos += *lenp;
1997 return 0;
1998#undef TMPBUFLEN
1999}
2000
2001/**
2002 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2003 * @table: the sysctl table
2004 * @write: %TRUE if this is a write to the sysctl file
2005 * @filp: the file structure
2006 * @buffer: the user buffer
2007 * @lenp: the size of the user buffer
2008 * @ppos: file position
2009 *
2010 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2011 * values from/to the user buffer, treated as an ASCII string.
2012 *
2013 * This routine will ensure the values are within the range specified by
2014 * table->extra1 (min) and table->extra2 (max).
2015 *
2016 * Returns 0 on success.
2017 */
2018int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2019 void __user *buffer, size_t *lenp, loff_t *ppos)
2020{
2021 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2022}
2023
2024/**
2025 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2026 * @table: the sysctl table
2027 * @write: %TRUE if this is a write to the sysctl file
2028 * @filp: the file structure
2029 * @buffer: the user buffer
2030 * @lenp: the size of the user buffer
2031 * @ppos: file position
2032 *
2033 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2034 * values from/to the user buffer, treated as an ASCII string. The values
2035 * are treated as milliseconds, and converted to jiffies when they are stored.
2036 *
2037 * This routine will ensure the values are within the range specified by
2038 * table->extra1 (min) and table->extra2 (max).
2039 *
2040 * Returns 0 on success.
2041 */
2042int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2043 struct file *filp,
2044 void __user *buffer,
2045 size_t *lenp, loff_t *ppos)
2046{
2047 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2048 lenp, ppos, HZ, 1000l);
2049}
2050
2051
2052static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2053 int *valp,
2054 int write, void *data)
2055{
2056 if (write) {
2057 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2058 } else {
2059 int val = *valp;
2060 unsigned long lval;
2061 if (val < 0) {
2062 *negp = -1;
2063 lval = (unsigned long)-val;
2064 } else {
2065 *negp = 0;
2066 lval = (unsigned long)val;
2067 }
2068 *lvalp = lval / HZ;
2069 }
2070 return 0;
2071}
2072
2073static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2074 int *valp,
2075 int write, void *data)
2076{
2077 if (write) {
2078 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2079 } else {
2080 int val = *valp;
2081 unsigned long lval;
2082 if (val < 0) {
2083 *negp = -1;
2084 lval = (unsigned long)-val;
2085 } else {
2086 *negp = 0;
2087 lval = (unsigned long)val;
2088 }
2089 *lvalp = jiffies_to_clock_t(lval);
2090 }
2091 return 0;
2092}
2093
2094static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2095 int *valp,
2096 int write, void *data)
2097{
2098 if (write) {
2099 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2100 } else {
2101 int val = *valp;
2102 unsigned long lval;
2103 if (val < 0) {
2104 *negp = -1;
2105 lval = (unsigned long)-val;
2106 } else {
2107 *negp = 0;
2108 lval = (unsigned long)val;
2109 }
2110 *lvalp = jiffies_to_msecs(lval);
2111 }
2112 return 0;
2113}
2114
2115/**
2116 * proc_dointvec_jiffies - read a vector of integers as seconds
2117 * @table: the sysctl table
2118 * @write: %TRUE if this is a write to the sysctl file
2119 * @filp: the file structure
2120 * @buffer: the user buffer
2121 * @lenp: the size of the user buffer
2122 * @ppos: file position
2123 *
2124 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2125 * values from/to the user buffer, treated as an ASCII string.
2126 * The values read are assumed to be in seconds, and are converted into
2127 * jiffies.
2128 *
2129 * Returns 0 on success.
2130 */
2131int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2132 void __user *buffer, size_t *lenp, loff_t *ppos)
2133{
2134 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2135 do_proc_dointvec_jiffies_conv,NULL);
2136}
2137
2138/**
2139 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2140 * @table: the sysctl table
2141 * @write: %TRUE if this is a write to the sysctl file
2142 * @filp: the file structure
2143 * @buffer: the user buffer
2144 * @lenp: the size of the user buffer
1e5d5331 2145 * @ppos: pointer to the file position
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/USER_HZ seconds, and
2150 * are converted into jiffies.
2151 *
2152 * Returns 0 on success.
2153 */
2154int proc_dointvec_userhz_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_userhz_jiffies_conv,NULL);
2159}
2160
2161/**
2162 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2163 * @table: the sysctl table
2164 * @write: %TRUE if this is a write to the sysctl file
2165 * @filp: the file structure
2166 * @buffer: the user buffer
2167 * @lenp: the size of the user buffer
67be2dd1
MW		 2168 * @ppos: file position
2169 * @ppos: the current position in the file
1da177e4
LT		 2170 *
2171 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2172 * values from/to the user buffer, treated as an ASCII string.
2173 * The values read are assumed to be in 1/1000 seconds, and
2174 * are converted into jiffies.
2175 *
2176 * Returns 0 on success.
2177 */
2178int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2179 void __user *buffer, size_t *lenp, loff_t *ppos)
2180{
2181 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2182 do_proc_dointvec_ms_jiffies_conv, NULL);
2183}
2184
2185#else /* CONFIG_PROC_FS */
2186
2187int proc_dostring(ctl_table *table, int write, struct file *filp,
2188 void __user *buffer, size_t *lenp, loff_t *ppos)
2189{
2190 return -ENOSYS;
2191}
2192
2193static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
2194 void __user *buffer, size_t *lenp, loff_t *ppos)
2195{
2196 return -ENOSYS;
2197}
2198
2199int proc_dointvec(ctl_table *table, int write, struct file *filp,
2200 void __user *buffer, size_t *lenp, loff_t *ppos)
2201{
2202 return -ENOSYS;
2203}
2204
2205int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2206 void __user *buffer, size_t *lenp, loff_t *ppos)
2207{
2208 return -ENOSYS;
2209}
2210
2211int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2212 void __user *buffer, size_t *lenp, loff_t *ppos)
2213{
2214 return -ENOSYS;
2215}
2216
2217int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2218 void __user *buffer, size_t *lenp, loff_t *ppos)
2219{
2220 return -ENOSYS;
2221}
2222
2223int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2224 void __user *buffer, size_t *lenp, loff_t *ppos)
2225{
2226 return -ENOSYS;
2227}
2228
2229int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2230 void __user *buffer, size_t *lenp, loff_t *ppos)
2231{
2232 return -ENOSYS;
2233}
2234
2235int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2236 void __user *buffer, size_t *lenp, loff_t *ppos)
2237{
2238 return -ENOSYS;
2239}
2240
2241int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2242 struct file *filp,
2243 void __user *buffer,
2244 size_t *lenp, loff_t *ppos)
2245{
2246 return -ENOSYS;
2247}
2248
2249
2250#endif /* CONFIG_PROC_FS */
2251
2252
2253/*
2254 * General sysctl support routines
2255 */
2256
2257/* The generic string strategy routine: */
2258int sysctl_string(ctl_table *table, int __user *name, int nlen,
2259 void __user *oldval, size_t __user *oldlenp,
2260 void __user *newval, size_t newlen, void **context)
2261{
1da177e4
LT		 2262 if (!table->data || !table->maxlen)
2263 return -ENOTDIR;
2264
2265 if (oldval && oldlenp) {
de9e007d
LT		 2266 size_t bufsize;
2267 if (get_user(bufsize, oldlenp))
1da177e4 2268 return -EFAULT;
de9e007d
LT		 2269 if (bufsize) {
2270 size_t len = strlen(table->data), copied;
2271
2272 /* This shouldn't trigger for a well-formed sysctl */
2273 if (len > table->maxlen)
1da177e4 2274 len = table->maxlen;
de9e007d
LT		 2275
2276 /* Copy up to a max of bufsize-1 bytes of the string */
2277 copied = (len >= bufsize) ? bufsize - 1 : len;
2278
2279 if (copy_to_user(oldval, table->data, copied) ||
2280 put_user(0, (char __user *)(oldval + copied)))
1da177e4 2281 return -EFAULT;
de9e007d 2282 if (put_user(len, oldlenp))
1da177e4
LT		 2283 return -EFAULT;
2284 }
2285 }
2286 if (newval && newlen) {
de9e007d 2287 size_t len = newlen;
1da177e4
LT		 2288 if (len > table->maxlen)
2289 len = table->maxlen;
2290 if(copy_from_user(table->data, newval, len))
2291 return -EFAULT;
2292 if (len == table->maxlen)
2293 len--;
2294 ((char *) table->data)[len] = 0;
2295 }
82c9df82 2296 return 1;
1da177e4
LT		 2297}
2298
2299/*
2300 * This function makes sure that all of the integers in the vector
2301 * are between the minimum and maximum values given in the arrays
2302 * table->extra1 and table->extra2, respectively.
2303 */
2304int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2305 void __user *oldval, size_t __user *oldlenp,
2306 void __user *newval, size_t newlen, void **context)
2307{
2308
2309 if (newval && newlen) {
2310 int __user *vec = (int __user *) newval;
2311 int *min = (int *) table->extra1;
2312 int *max = (int *) table->extra2;
2313 size_t length;
2314 int i;
2315
2316 if (newlen % sizeof(int) != 0)
2317 return -EINVAL;
2318
2319 if (!table->extra1 && !table->extra2)
2320 return 0;
2321
2322 if (newlen > table->maxlen)
2323 newlen = table->maxlen;
2324 length = newlen / sizeof(int);
2325
2326 for (i = 0; i < length; i++) {
2327 int value;
2328 if (get_user(value, vec + i))
2329 return -EFAULT;
2330 if (min && value < min[i])
2331 return -EINVAL;
2332 if (max && value > max[i])
2333 return -EINVAL;
2334 }
2335 }
2336 return 0;
2337}
2338
2339/* Strategy function to convert jiffies to seconds */
2340int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2341 void __user *oldval, size_t __user *oldlenp,
2342 void __user *newval, size_t newlen, void **context)
2343{
2344 if (oldval) {
2345 size_t olen;
2346 if (oldlenp) {
2347 if (get_user(olen, oldlenp))
2348 return -EFAULT;
2349 if (olen!=sizeof(int))
2350 return -EINVAL;
2351 }
2352 if (put_user(*(int *)(table->data)/HZ, (int __user *)oldval) ||
2353 (oldlenp && put_user(sizeof(int),oldlenp)))
2354 return -EFAULT;
2355 }
2356 if (newval && newlen) {
2357 int new;
2358 if (newlen != sizeof(int))
2359 return -EINVAL;
2360 if (get_user(new, (int __user *)newval))
2361 return -EFAULT;
2362 *(int *)(table->data) = new*HZ;
2363 }
2364 return 1;
2365}
2366
2367/* Strategy function to convert jiffies to seconds */
2368int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2369 void __user *oldval, size_t __user *oldlenp,
2370 void __user *newval, size_t newlen, void **context)
2371{
2372 if (oldval) {
2373 size_t olen;
2374 if (oldlenp) {
2375 if (get_user(olen, oldlenp))
2376 return -EFAULT;
2377 if (olen!=sizeof(int))
2378 return -EINVAL;
2379 }
2380 if (put_user(jiffies_to_msecs(*(int *)(table->data)), (int __user *)oldval) ||
2381 (oldlenp && put_user(sizeof(int),oldlenp)))
2382 return -EFAULT;
2383 }
2384 if (newval && newlen) {
2385 int new;
2386 if (newlen != sizeof(int))
2387 return -EINVAL;
2388 if (get_user(new, (int __user *)newval))
2389 return -EFAULT;
2390 *(int *)(table->data) = msecs_to_jiffies(new);
2391 }
2392 return 1;
2393}
2394
2395#else /* CONFIG_SYSCTL */
2396
2397
2398asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2399{
2400 return -ENOSYS;
2401}
2402
2403int sysctl_string(ctl_table *table, int __user *name, int nlen,
2404 void __user *oldval, size_t __user *oldlenp,
2405 void __user *newval, size_t newlen, void **context)
2406{
2407 return -ENOSYS;
2408}
2409
2410int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2411 void __user *oldval, size_t __user *oldlenp,
2412 void __user *newval, size_t newlen, void **context)
2413{
2414 return -ENOSYS;
2415}
2416
2417int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2418 void __user *oldval, size_t __user *oldlenp,
2419 void __user *newval, size_t newlen, void **context)
2420{
2421 return -ENOSYS;
2422}
2423
2424int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2425 void __user *oldval, size_t __user *oldlenp,
2426 void __user *newval, size_t newlen, void **context)
2427{
2428 return -ENOSYS;
2429}
2430
2431int proc_dostring(ctl_table *table, int write, struct file *filp,
2432 void __user *buffer, size_t *lenp, loff_t *ppos)
2433{
2434 return -ENOSYS;
2435}
2436
2437int proc_dointvec(ctl_table *table, int write, struct file *filp,
2438 void __user *buffer, size_t *lenp, loff_t *ppos)
2439{
2440 return -ENOSYS;
2441}
2442
2443int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2444 void __user *buffer, size_t *lenp, loff_t *ppos)
2445{
2446 return -ENOSYS;
2447}
2448
2449int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2450 void __user *buffer, size_t *lenp, loff_t *ppos)
2451{
2452 return -ENOSYS;
2453}
2454
2455int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2456 void __user *buffer, size_t *lenp, loff_t *ppos)
2457{
2458 return -ENOSYS;
2459}
2460
2461int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2462 void __user *buffer, size_t *lenp, loff_t *ppos)
2463{
2464 return -ENOSYS;
2465}
2466
2467int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2468 void __user *buffer, size_t *lenp, loff_t *ppos)
2469{
2470 return -ENOSYS;
2471}
2472
2473int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2474 void __user *buffer, size_t *lenp, loff_t *ppos)
2475{
2476 return -ENOSYS;
2477}
2478
2479int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2480 struct file *filp,
2481 void __user *buffer,
2482 size_t *lenp, loff_t *ppos)
2483{
2484 return -ENOSYS;
2485}
2486
2487struct ctl_table_header * register_sysctl_table(ctl_table * table,
2488 int insert_at_head)
2489{
2490 return NULL;
2491}
2492
2493void unregister_sysctl_table(struct ctl_table_header * table)
2494{
2495}
2496
2497#endif /* CONFIG_SYSCTL */
2498
2499/*
2500 * No sense putting this after each symbol definition, twice,
2501 * exception granted :-)
2502 */
2503EXPORT_SYMBOL(proc_dointvec);
2504EXPORT_SYMBOL(proc_dointvec_jiffies);
2505EXPORT_SYMBOL(proc_dointvec_minmax);
2506EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2507EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2508EXPORT_SYMBOL(proc_dostring);
2509EXPORT_SYMBOL(proc_doulongvec_minmax);
2510EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2511EXPORT_SYMBOL(register_sysctl_table);
2512EXPORT_SYMBOL(sysctl_intvec);
2513EXPORT_SYMBOL(sysctl_jiffies);
2514EXPORT_SYMBOL(sysctl_ms_jiffies);
2515EXPORT_SYMBOL(sysctl_string);
2516EXPORT_SYMBOL(unregister_sysctl_table);
Clone of davem's net-next-2.6 tree