projects / linux-2.6-block.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
[PATCH] voyager: fix compile after setup rework
[linux-2.6-block.git] / kernel / sysctl.c
CommitLineData
1da177e4
LT		 1/*
2 * sysctl.c: General linux system control interface
3 *
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12 * Horn.
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16 * Wendling.
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19 */
20
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;
fadd8fbd 62extern int sysctl_panic_on_oom;
1da177e4
LT		 63extern int max_threads;
64extern int sysrq_enabled;
65extern int core_uses_pid;
d6e71144 66extern int suid_dumpable;
1da177e4
LT		 67extern char core_pattern[];
68extern int cad_pid;
69extern int pid_max;
70extern int min_free_kbytes;
71extern int printk_ratelimit_jiffies;
72extern int printk_ratelimit_burst;
73extern int pid_max_min, pid_max_max;
9d0243bc 74extern int sysctl_drop_caches;
8ad4b1fb 75extern int percpu_pagelist_fraction;
bebfa101 76extern int compat_log;
1da177e4
LT		 77
78#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
79int unknown_nmi_panic;
80extern int proc_unknown_nmi_panic(ctl_table *, int, struct file *,
81 void __user *, size_t *, loff_t *);
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
1da177e4
LT		 136static int parse_table(int __user *, int, void __user *, size_t __user *, void __user *, size_t,
137 ctl_table *, void **);
138static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
139 void __user *buffer, size_t *lenp, loff_t *ppos);
140
141static ctl_table root_table[];
142static struct ctl_table_header root_table_header =
143 { root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
144
145static ctl_table kern_table[];
146static ctl_table vm_table[];
1da177e4
LT		 147static ctl_table fs_table[];
148static ctl_table debug_table[];
149static ctl_table dev_table[];
150extern ctl_table random_table[];
151#ifdef CONFIG_UNIX98_PTYS
152extern ctl_table pty_table[];
153#endif
2d9048e2 154#ifdef CONFIG_INOTIFY_USER
0399cb08
RL		 155extern ctl_table inotify_table[];
156#endif
1da177e4
LT		 157
158#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
159int sysctl_legacy_va_layout;
160#endif
161
162/* /proc declarations: */
163
164#ifdef CONFIG_PROC_FS
165
166static ssize_t proc_readsys(struct file *, char __user *, size_t, loff_t *);
167static ssize_t proc_writesys(struct file *, const char __user *, size_t, loff_t *);
168static int proc_opensys(struct inode *, struct file *);
169
170struct file_operations proc_sys_file_operations = {
171 .open = proc_opensys,
172 .read = proc_readsys,
173 .write = proc_writesys,
174};
175
176extern struct proc_dir_entry *proc_sys_root;
177
330d57fb 178static void register_proc_table(ctl_table *, struct proc_dir_entry *, void *);
1da177e4
LT		 179static void unregister_proc_table(ctl_table *, struct proc_dir_entry *);
180#endif
181
182/* The default sysctl tables: */
183
184static ctl_table root_table[] = {
185 {
186 .ctl_name = CTL_KERN,
187 .procname = "kernel",
188 .mode = 0555,
189 .child = kern_table,
190 },
191 {
192 .ctl_name = CTL_VM,
193 .procname = "vm",
194 .mode = 0555,
195 .child = vm_table,
196 },
197#ifdef CONFIG_NET
198 {
199 .ctl_name = CTL_NET,
200 .procname = "net",
201 .mode = 0555,
202 .child = net_table,
203 },
204#endif
1da177e4
LT		 205 {
206 .ctl_name = CTL_FS,
207 .procname = "fs",
208 .mode = 0555,
209 .child = fs_table,
210 },
211 {
212 .ctl_name = CTL_DEBUG,
213 .procname = "debug",
214 .mode = 0555,
215 .child = debug_table,
216 },
217 {
218 .ctl_name = CTL_DEV,
219 .procname = "dev",
220 .mode = 0555,
221 .child = dev_table,
222 },
0eeca283 223
1da177e4
LT		 224 { .ctl_name = 0 }
225};
226
227static ctl_table kern_table[] = {
228 {
229 .ctl_name = KERN_OSTYPE,
230 .procname = "ostype",
231 .data = system_utsname.sysname,
232 .maxlen = sizeof(system_utsname.sysname),
233 .mode = 0444,
234 .proc_handler = &proc_doutsstring,
235 .strategy = &sysctl_string,
236 },
237 {
238 .ctl_name = KERN_OSRELEASE,
239 .procname = "osrelease",
240 .data = system_utsname.release,
241 .maxlen = sizeof(system_utsname.release),
242 .mode = 0444,
243 .proc_handler = &proc_doutsstring,
244 .strategy = &sysctl_string,
245 },
246 {
247 .ctl_name = KERN_VERSION,
248 .procname = "version",
249 .data = system_utsname.version,
250 .maxlen = sizeof(system_utsname.version),
251 .mode = 0444,
252 .proc_handler = &proc_doutsstring,
253 .strategy = &sysctl_string,
254 },
255 {
256 .ctl_name = KERN_NODENAME,
257 .procname = "hostname",
258 .data = system_utsname.nodename,
259 .maxlen = sizeof(system_utsname.nodename),
260 .mode = 0644,
261 .proc_handler = &proc_doutsstring,
262 .strategy = &sysctl_string,
263 },
264 {
265 .ctl_name = KERN_DOMAINNAME,
266 .procname = "domainname",
267 .data = system_utsname.domainname,
268 .maxlen = sizeof(system_utsname.domainname),
269 .mode = 0644,
270 .proc_handler = &proc_doutsstring,
271 .strategy = &sysctl_string,
272 },
273 {
274 .ctl_name = KERN_PANIC,
275 .procname = "panic",
276 .data = &panic_timeout,
277 .maxlen = sizeof(int),
278 .mode = 0644,
279 .proc_handler = &proc_dointvec,
280 },
281 {
282 .ctl_name = KERN_CORE_USES_PID,
283 .procname = "core_uses_pid",
284 .data = &core_uses_pid,
285 .maxlen = sizeof(int),
286 .mode = 0644,
287 .proc_handler = &proc_dointvec,
288 },
289 {
290 .ctl_name = KERN_CORE_PATTERN,
291 .procname = "core_pattern",
292 .data = core_pattern,
293 .maxlen = 64,
294 .mode = 0644,
295 .proc_handler = &proc_dostring,
296 .strategy = &sysctl_string,
297 },
298 {
299 .ctl_name = KERN_TAINTED,
300 .procname = "tainted",
301 .data = &tainted,
302 .maxlen = sizeof(int),
303 .mode = 0444,
304 .proc_handler = &proc_dointvec,
305 },
306 {
307 .ctl_name = KERN_CAP_BSET,
308 .procname = "cap-bound",
309 .data = &cap_bset,
310 .maxlen = sizeof(kernel_cap_t),
311 .mode = 0600,
312 .proc_handler = &proc_dointvec_bset,
313 },
314#ifdef CONFIG_BLK_DEV_INITRD
315 {
316 .ctl_name = KERN_REALROOTDEV,
317 .procname = "real-root-dev",
318 .data = &real_root_dev,
319 .maxlen = sizeof(int),
320 .mode = 0644,
321 .proc_handler = &proc_dointvec,
322 },
323#endif
324#ifdef __sparc__
325 {
326 .ctl_name = KERN_SPARC_REBOOT,
327 .procname = "reboot-cmd",
328 .data = reboot_command,
329 .maxlen = 256,
330 .mode = 0644,
331 .proc_handler = &proc_dostring,
332 .strategy = &sysctl_string,
333 },
334 {
335 .ctl_name = KERN_SPARC_STOP_A,
336 .procname = "stop-a",
337 .data = &stop_a_enabled,
338 .maxlen = sizeof (int),
339 .mode = 0644,
340 .proc_handler = &proc_dointvec,
341 },
342 {
343 .ctl_name = KERN_SPARC_SCONS_PWROFF,
344 .procname = "scons-poweroff",
345 .data = &scons_pwroff,
346 .maxlen = sizeof (int),
347 .mode = 0644,
348 .proc_handler = &proc_dointvec,
349 },
350#endif
351#ifdef __hppa__
352 {
353 .ctl_name = KERN_HPPA_PWRSW,
354 .procname = "soft-power",
355 .data = &pwrsw_enabled,
356 .maxlen = sizeof (int),
357 .mode = 0644,
358 .proc_handler = &proc_dointvec,
359 },
360 {
361 .ctl_name = KERN_HPPA_UNALIGNED,
362 .procname = "unaligned-trap",
363 .data = &unaligned_enabled,
364 .maxlen = sizeof (int),
365 .mode = 0644,
366 .proc_handler = &proc_dointvec,
367 },
368#endif
369 {
370 .ctl_name = KERN_CTLALTDEL,
371 .procname = "ctrl-alt-del",
372 .data = &C_A_D,
373 .maxlen = sizeof(int),
374 .mode = 0644,
375 .proc_handler = &proc_dointvec,
376 },
377 {
378 .ctl_name = KERN_PRINTK,
379 .procname = "printk",
380 .data = &console_loglevel,
381 .maxlen = 4*sizeof(int),
382 .mode = 0644,
383 .proc_handler = &proc_dointvec,
384 },
385#ifdef CONFIG_KMOD
386 {
387 .ctl_name = KERN_MODPROBE,
388 .procname = "modprobe",
389 .data = &modprobe_path,
390 .maxlen = KMOD_PATH_LEN,
391 .mode = 0644,
392 .proc_handler = &proc_dostring,
393 .strategy = &sysctl_string,
394 },
395#endif
57ae2508 396#if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
1da177e4
LT		 397 {
398 .ctl_name = KERN_HOTPLUG,
399 .procname = "hotplug",
312c004d
KS		 400 .data = &uevent_helper,
401 .maxlen = UEVENT_HELPER_PATH_LEN,
1da177e4
LT		 402 .mode = 0644,
403 .proc_handler = &proc_dostring,
404 .strategy = &sysctl_string,
405 },
406#endif
407#ifdef CONFIG_CHR_DEV_SG
408 {
409 .ctl_name = KERN_SG_BIG_BUFF,
410 .procname = "sg-big-buff",
411 .data = &sg_big_buff,
412 .maxlen = sizeof (int),
413 .mode = 0444,
414 .proc_handler = &proc_dointvec,
415 },
416#endif
417#ifdef CONFIG_BSD_PROCESS_ACCT
418 {
419 .ctl_name = KERN_ACCT,
420 .procname = "acct",
421 .data = &acct_parm,
422 .maxlen = 3*sizeof(int),
423 .mode = 0644,
424 .proc_handler = &proc_dointvec,
425 },
426#endif
427#ifdef CONFIG_SYSVIPC
428 {
429 .ctl_name = KERN_SHMMAX,
430 .procname = "shmmax",
431 .data = &shm_ctlmax,
432 .maxlen = sizeof (size_t),
433 .mode = 0644,
434 .proc_handler = &proc_doulongvec_minmax,
435 },
436 {
437 .ctl_name = KERN_SHMALL,
438 .procname = "shmall",
439 .data = &shm_ctlall,
440 .maxlen = sizeof (size_t),
441 .mode = 0644,
442 .proc_handler = &proc_doulongvec_minmax,
443 },
444 {
445 .ctl_name = KERN_SHMMNI,
446 .procname = "shmmni",
447 .data = &shm_ctlmni,
448 .maxlen = sizeof (int),
449 .mode = 0644,
450 .proc_handler = &proc_dointvec,
451 },
452 {
453 .ctl_name = KERN_MSGMAX,
454 .procname = "msgmax",
455 .data = &msg_ctlmax,
456 .maxlen = sizeof (int),
457 .mode = 0644,
458 .proc_handler = &proc_dointvec,
459 },
460 {
461 .ctl_name = KERN_MSGMNI,
462 .procname = "msgmni",
463 .data = &msg_ctlmni,
464 .maxlen = sizeof (int),
465 .mode = 0644,
466 .proc_handler = &proc_dointvec,
467 },
468 {
469 .ctl_name = KERN_MSGMNB,
470 .procname = "msgmnb",
471 .data = &msg_ctlmnb,
472 .maxlen = sizeof (int),
473 .mode = 0644,
474 .proc_handler = &proc_dointvec,
475 },
476 {
477 .ctl_name = KERN_SEM,
478 .procname = "sem",
479 .data = &sem_ctls,
480 .maxlen = 4*sizeof (int),
481 .mode = 0644,
482 .proc_handler = &proc_dointvec,
483 },
484#endif
485#ifdef CONFIG_MAGIC_SYSRQ
486 {
487 .ctl_name = KERN_SYSRQ,
488 .procname = "sysrq",
489 .data = &sysrq_enabled,
490 .maxlen = sizeof (int),
491 .mode = 0644,
492 .proc_handler = &proc_dointvec,
493 },
494#endif
495 {
496 .ctl_name = KERN_CADPID,
497 .procname = "cad_pid",
498 .data = &cad_pid,
499 .maxlen = sizeof (int),
500 .mode = 0600,
501 .proc_handler = &proc_dointvec,
502 },
503 {
504 .ctl_name = KERN_MAX_THREADS,
505 .procname = "threads-max",
506 .data = &max_threads,
507 .maxlen = sizeof(int),
508 .mode = 0644,
509 .proc_handler = &proc_dointvec,
510 },
511 {
512 .ctl_name = KERN_RANDOM,
513 .procname = "random",
514 .mode = 0555,
515 .child = random_table,
516 },
517#ifdef CONFIG_UNIX98_PTYS
518 {
519 .ctl_name = KERN_PTY,
520 .procname = "pty",
521 .mode = 0555,
522 .child = pty_table,
523 },
524#endif
525 {
526 .ctl_name = KERN_OVERFLOWUID,
527 .procname = "overflowuid",
528 .data = &overflowuid,
529 .maxlen = sizeof(int),
530 .mode = 0644,
531 .proc_handler = &proc_dointvec_minmax,
532 .strategy = &sysctl_intvec,
533 .extra1 = &minolduid,
534 .extra2 = &maxolduid,
535 },
536 {
537 .ctl_name = KERN_OVERFLOWGID,
538 .procname = "overflowgid",
539 .data = &overflowgid,
540 .maxlen = sizeof(int),
541 .mode = 0644,
542 .proc_handler = &proc_dointvec_minmax,
543 .strategy = &sysctl_intvec,
544 .extra1 = &minolduid,
545 .extra2 = &maxolduid,
546 },
347a8dc3 547#ifdef CONFIG_S390
1da177e4
LT		 548#ifdef CONFIG_MATHEMU
549 {
550 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
551 .procname = "ieee_emulation_warnings",
552 .data = &sysctl_ieee_emulation_warnings,
553 .maxlen = sizeof(int),
554 .mode = 0644,
555 .proc_handler = &proc_dointvec,
556 },
557#endif
558#ifdef CONFIG_NO_IDLE_HZ
559 {
560 .ctl_name = KERN_HZ_TIMER,
561 .procname = "hz_timer",
562 .data = &sysctl_hz_timer,
563 .maxlen = sizeof(int),
564 .mode = 0644,
565 .proc_handler = &proc_dointvec,
566 },
567#endif
568 {
569 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
570 .procname = "userprocess_debug",
571 .data = &sysctl_userprocess_debug,
572 .maxlen = sizeof(int),
573 .mode = 0644,
574 .proc_handler = &proc_dointvec,
575 },
576#endif
577 {
578 .ctl_name = KERN_PIDMAX,
579 .procname = "pid_max",
580 .data = &pid_max,
581 .maxlen = sizeof (int),
582 .mode = 0644,
583 .proc_handler = &proc_dointvec_minmax,
584 .strategy = sysctl_intvec,
585 .extra1 = &pid_max_min,
586 .extra2 = &pid_max_max,
587 },
588 {
589 .ctl_name = KERN_PANIC_ON_OOPS,
590 .procname = "panic_on_oops",
591 .data = &panic_on_oops,
592 .maxlen = sizeof(int),
593 .mode = 0644,
594 .proc_handler = &proc_dointvec,
595 },
596 {
597 .ctl_name = KERN_PRINTK_RATELIMIT,
598 .procname = "printk_ratelimit",
599 .data = &printk_ratelimit_jiffies,
600 .maxlen = sizeof(int),
601 .mode = 0644,
602 .proc_handler = &proc_dointvec_jiffies,
603 .strategy = &sysctl_jiffies,
604 },
605 {
606 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
607 .procname = "printk_ratelimit_burst",
608 .data = &printk_ratelimit_burst,
609 .maxlen = sizeof(int),
610 .mode = 0644,
611 .proc_handler = &proc_dointvec,
612 },
613 {
614 .ctl_name = KERN_NGROUPS_MAX,
615 .procname = "ngroups_max",
616 .data = &ngroups_max,
617 .maxlen = sizeof (int),
618 .mode = 0444,
619 .proc_handler = &proc_dointvec,
620 },
621#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
622 {
623 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
624 .procname = "unknown_nmi_panic",
625 .data = &unknown_nmi_panic,
626 .maxlen = sizeof (int),
627 .mode = 0644,
628 .proc_handler = &proc_unknown_nmi_panic,
629 },
630#endif
631#if defined(CONFIG_X86)
632 {
633 .ctl_name = KERN_BOOTLOADER_TYPE,
634 .procname = "bootloader_type",
635 .data = &bootloader_type,
636 .maxlen = sizeof (int),
637 .mode = 0444,
638 .proc_handler = &proc_dointvec,
639 },
640#endif
7a9166e3 641#if defined(CONFIG_MMU)
1da177e4
LT		 642 {
643 .ctl_name = KERN_RANDOMIZE,
644 .procname = "randomize_va_space",
645 .data = &randomize_va_space,
646 .maxlen = sizeof(int),
647 .mode = 0644,
648 .proc_handler = &proc_dointvec,
649 },
7a9166e3 650#endif
0152fb37 651#if defined(CONFIG_S390) && defined(CONFIG_SMP)
951f22d5
MS		 652 {
653 .ctl_name = KERN_SPIN_RETRY,
654 .procname = "spin_retry",
655 .data = &spin_retry,
656 .maxlen = sizeof (int),
657 .mode = 0644,
658 .proc_handler = &proc_dointvec,
659 },
c255d844
PM		 660#endif
661#ifdef CONFIG_ACPI_SLEEP
662 {
663 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
664 .procname = "acpi_video_flags",
665 .data = &acpi_video_flags,
666 .maxlen = sizeof (unsigned long),
667 .mode = 0644,
7f99f06f 668 .proc_handler = &proc_doulongvec_minmax,
c255d844 669 },
d2b176ed
JS		 670#endif
671#ifdef CONFIG_IA64
672 {
673 .ctl_name = KERN_IA64_UNALIGNED,
674 .procname = "ignore-unaligned-usertrap",
675 .data = &no_unaligned_warning,
676 .maxlen = sizeof (int),
677 .mode = 0644,
678 .proc_handler = &proc_dointvec,
679 },
bebfa101
AK		 680#endif
681#ifdef CONFIG_COMPAT
682 {
683 .ctl_name = KERN_COMPAT_LOG,
684 .procname = "compat-log",
685 .data = &compat_log,
686 .maxlen = sizeof (int),
687 .mode = 0644,
688 .proc_handler = &proc_dointvec,
689 },
951f22d5 690#endif
1da177e4
LT		 691 { .ctl_name = 0 }
692};
693
694/* Constants for minimum and maximum testing in vm_table.
695 We use these as one-element integer vectors. */
696static int zero;
697static int one_hundred = 100;
698
699
700static ctl_table vm_table[] = {
701 {
702 .ctl_name = VM_OVERCOMMIT_MEMORY,
703 .procname = "overcommit_memory",
704 .data = &sysctl_overcommit_memory,
705 .maxlen = sizeof(sysctl_overcommit_memory),
706 .mode = 0644,
707 .proc_handler = &proc_dointvec,
708 },
fadd8fbd
KH		 709 {
710 .ctl_name = VM_PANIC_ON_OOM,
711 .procname = "panic_on_oom",
712 .data = &sysctl_panic_on_oom,
713 .maxlen = sizeof(sysctl_panic_on_oom),
714 .mode = 0644,
715 .proc_handler = &proc_dointvec,
716 },
1da177e4
LT		 717 {
718 .ctl_name = VM_OVERCOMMIT_RATIO,
719 .procname = "overcommit_ratio",
720 .data = &sysctl_overcommit_ratio,
721 .maxlen = sizeof(sysctl_overcommit_ratio),
722 .mode = 0644,
723 .proc_handler = &proc_dointvec,
724 },
725 {
726 .ctl_name = VM_PAGE_CLUSTER,
727 .procname = "page-cluster",
728 .data = &page_cluster,
729 .maxlen = sizeof(int),
730 .mode = 0644,
731 .proc_handler = &proc_dointvec,
732 },
733 {
734 .ctl_name = VM_DIRTY_BACKGROUND,
735 .procname = "dirty_background_ratio",
736 .data = &dirty_background_ratio,
737 .maxlen = sizeof(dirty_background_ratio),
738 .mode = 0644,
739 .proc_handler = &proc_dointvec_minmax,
740 .strategy = &sysctl_intvec,
741 .extra1 = &zero,
742 .extra2 = &one_hundred,
743 },
744 {
745 .ctl_name = VM_DIRTY_RATIO,
746 .procname = "dirty_ratio",
747 .data = &vm_dirty_ratio,
748 .maxlen = sizeof(vm_dirty_ratio),
749 .mode = 0644,
750 .proc_handler = &proc_dointvec_minmax,
751 .strategy = &sysctl_intvec,
752 .extra1 = &zero,
753 .extra2 = &one_hundred,
754 },
755 {
756 .ctl_name = VM_DIRTY_WB_CS,
757 .procname = "dirty_writeback_centisecs",
f6ef9438
BS		 758 .data = &dirty_writeback_interval,
759 .maxlen = sizeof(dirty_writeback_interval),
1da177e4
LT		 760 .mode = 0644,
761 .proc_handler = &dirty_writeback_centisecs_handler,
762 },
763 {
764 .ctl_name = VM_DIRTY_EXPIRE_CS,
765 .procname = "dirty_expire_centisecs",
f6ef9438
BS		 766 .data = &dirty_expire_interval,
767 .maxlen = sizeof(dirty_expire_interval),
1da177e4 768 .mode = 0644,
f6ef9438 769 .proc_handler = &proc_dointvec_userhz_jiffies,
1da177e4
LT		 770 },
771 {
772 .ctl_name = VM_NR_PDFLUSH_THREADS,
773 .procname = "nr_pdflush_threads",
774 .data = &nr_pdflush_threads,
775 .maxlen = sizeof nr_pdflush_threads,
776 .mode = 0444 /* read-only*/,
777 .proc_handler = &proc_dointvec,
778 },
779 {
780 .ctl_name = VM_SWAPPINESS,
781 .procname = "swappiness",
782 .data = &vm_swappiness,
783 .maxlen = sizeof(vm_swappiness),
784 .mode = 0644,
785 .proc_handler = &proc_dointvec_minmax,
786 .strategy = &sysctl_intvec,
787 .extra1 = &zero,
788 .extra2 = &one_hundred,
789 },
790#ifdef CONFIG_HUGETLB_PAGE
791 {
792 .ctl_name = VM_HUGETLB_PAGES,
793 .procname = "nr_hugepages",
794 .data = &max_huge_pages,
795 .maxlen = sizeof(unsigned long),
796 .mode = 0644,
797 .proc_handler = &hugetlb_sysctl_handler,
798 .extra1 = (void *)&hugetlb_zero,
799 .extra2 = (void *)&hugetlb_infinity,
800 },
801 {
802 .ctl_name = VM_HUGETLB_GROUP,
803 .procname = "hugetlb_shm_group",
804 .data = &sysctl_hugetlb_shm_group,
805 .maxlen = sizeof(gid_t),
806 .mode = 0644,
807 .proc_handler = &proc_dointvec,
808 },
809#endif
810 {
811 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
812 .procname = "lowmem_reserve_ratio",
813 .data = &sysctl_lowmem_reserve_ratio,
814 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
815 .mode = 0644,
816 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
817 .strategy = &sysctl_intvec,
818 },
9d0243bc
AM		 819 {
820 .ctl_name = VM_DROP_PAGECACHE,
821 .procname = "drop_caches",
822 .data = &sysctl_drop_caches,
823 .maxlen = sizeof(int),
824 .mode = 0644,
825 .proc_handler = drop_caches_sysctl_handler,
826 .strategy = &sysctl_intvec,
827 },
1da177e4
LT		 828 {
829 .ctl_name = VM_MIN_FREE_KBYTES,
830 .procname = "min_free_kbytes",
831 .data = &min_free_kbytes,
832 .maxlen = sizeof(min_free_kbytes),
833 .mode = 0644,
834 .proc_handler = &min_free_kbytes_sysctl_handler,
835 .strategy = &sysctl_intvec,
836 .extra1 = &zero,
837 },
8ad4b1fb
RS		 838 {
839 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
840 .procname = "percpu_pagelist_fraction",
841 .data = &percpu_pagelist_fraction,
842 .maxlen = sizeof(percpu_pagelist_fraction),
843 .mode = 0644,
844 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
845 .strategy = &sysctl_intvec,
846 .extra1 = &min_percpu_pagelist_fract,
847 },
1da177e4
LT		 848#ifdef CONFIG_MMU
849 {
850 .ctl_name = VM_MAX_MAP_COUNT,
851 .procname = "max_map_count",
852 .data = &sysctl_max_map_count,
853 .maxlen = sizeof(sysctl_max_map_count),
854 .mode = 0644,
855 .proc_handler = &proc_dointvec
856 },
857#endif
858 {
859 .ctl_name = VM_LAPTOP_MODE,
860 .procname = "laptop_mode",
861 .data = &laptop_mode,
862 .maxlen = sizeof(laptop_mode),
863 .mode = 0644,
ed5b43f1
BS		 864 .proc_handler = &proc_dointvec_jiffies,
865 .strategy = &sysctl_jiffies,
1da177e4
LT		 866 },
867 {
868 .ctl_name = VM_BLOCK_DUMP,
869 .procname = "block_dump",
870 .data = &block_dump,
871 .maxlen = sizeof(block_dump),
872 .mode = 0644,
873 .proc_handler = &proc_dointvec,
874 .strategy = &sysctl_intvec,
875 .extra1 = &zero,
876 },
877 {
878 .ctl_name = VM_VFS_CACHE_PRESSURE,
879 .procname = "vfs_cache_pressure",
880 .data = &sysctl_vfs_cache_pressure,
881 .maxlen = sizeof(sysctl_vfs_cache_pressure),
882 .mode = 0644,
883 .proc_handler = &proc_dointvec,
884 .strategy = &sysctl_intvec,
885 .extra1 = &zero,
886 },
887#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
888 {
889 .ctl_name = VM_LEGACY_VA_LAYOUT,
890 .procname = "legacy_va_layout",
891 .data = &sysctl_legacy_va_layout,
892 .maxlen = sizeof(sysctl_legacy_va_layout),
893 .mode = 0644,
894 .proc_handler = &proc_dointvec,
895 .strategy = &sysctl_intvec,
896 .extra1 = &zero,
897 },
898#endif
899#ifdef CONFIG_SWAP
900 {
901 .ctl_name = VM_SWAP_TOKEN_TIMEOUT,
902 .procname = "swap_token_timeout",
903 .data = &swap_token_default_timeout,
904 .maxlen = sizeof(swap_token_default_timeout),
905 .mode = 0644,
906 .proc_handler = &proc_dointvec_jiffies,
907 .strategy = &sysctl_jiffies,
908 },
1743660b
CL		 909#endif
910#ifdef CONFIG_NUMA
911 {
912 .ctl_name = VM_ZONE_RECLAIM_MODE,
913 .procname = "zone_reclaim_mode",
914 .data = &zone_reclaim_mode,
915 .maxlen = sizeof(zone_reclaim_mode),
916 .mode = 0644,
917 .proc_handler = &proc_dointvec,
c84db23c
CL		 918 .strategy = &sysctl_intvec,
919 .extra1 = &zero,
1743660b 920 },
2a11ff06
CL		 921 {
922 .ctl_name = VM_ZONE_RECLAIM_INTERVAL,
923 .procname = "zone_reclaim_interval",
924 .data = &zone_reclaim_interval,
925 .maxlen = sizeof(zone_reclaim_interval),
926 .mode = 0644,
927 .proc_handler = &proc_dointvec_jiffies,
928 .strategy = &sysctl_jiffies,
929 },
1da177e4
LT		 930#endif
931 { .ctl_name = 0 }
932};
933
1da177e4
LT		 934static ctl_table fs_table[] = {
935 {
936 .ctl_name = FS_NRINODE,
937 .procname = "inode-nr",
938 .data = &inodes_stat,
939 .maxlen = 2*sizeof(int),
940 .mode = 0444,
941 .proc_handler = &proc_dointvec,
942 },
943 {
944 .ctl_name = FS_STATINODE,
945 .procname = "inode-state",
946 .data = &inodes_stat,
947 .maxlen = 7*sizeof(int),
948 .mode = 0444,
949 .proc_handler = &proc_dointvec,
950 },
951 {
952 .ctl_name = FS_NRFILE,
953 .procname = "file-nr",
954 .data = &files_stat,
955 .maxlen = 3*sizeof(int),
956 .mode = 0444,
529bf6be 957 .proc_handler = &proc_nr_files,
1da177e4
LT		 958 },
959 {
960 .ctl_name = FS_MAXFILE,
961 .procname = "file-max",
962 .data = &files_stat.max_files,
963 .maxlen = sizeof(int),
964 .mode = 0644,
965 .proc_handler = &proc_dointvec,
966 },
967 {
968 .ctl_name = FS_DENTRY,
969 .procname = "dentry-state",
970 .data = &dentry_stat,
971 .maxlen = 6*sizeof(int),
972 .mode = 0444,
973 .proc_handler = &proc_dointvec,
974 },
975 {
976 .ctl_name = FS_OVERFLOWUID,
977 .procname = "overflowuid",
978 .data = &fs_overflowuid,
979 .maxlen = sizeof(int),
980 .mode = 0644,
981 .proc_handler = &proc_dointvec_minmax,
982 .strategy = &sysctl_intvec,
983 .extra1 = &minolduid,
984 .extra2 = &maxolduid,
985 },
986 {
987 .ctl_name = FS_OVERFLOWGID,
988 .procname = "overflowgid",
989 .data = &fs_overflowgid,
990 .maxlen = sizeof(int),
991 .mode = 0644,
992 .proc_handler = &proc_dointvec_minmax,
993 .strategy = &sysctl_intvec,
994 .extra1 = &minolduid,
995 .extra2 = &maxolduid,
996 },
997 {
998 .ctl_name = FS_LEASES,
999 .procname = "leases-enable",
1000 .data = &leases_enable,
1001 .maxlen = sizeof(int),
1002 .mode = 0644,
1003 .proc_handler = &proc_dointvec,
1004 },
1005#ifdef CONFIG_DNOTIFY
1006 {
1007 .ctl_name = FS_DIR_NOTIFY,
1008 .procname = "dir-notify-enable",
1009 .data = &dir_notify_enable,
1010 .maxlen = sizeof(int),
1011 .mode = 0644,
1012 .proc_handler = &proc_dointvec,
1013 },
1014#endif
1015#ifdef CONFIG_MMU
1016 {
1017 .ctl_name = FS_LEASE_TIME,
1018 .procname = "lease-break-time",
1019 .data = &lease_break_time,
1020 .maxlen = sizeof(int),
1021 .mode = 0644,
1022 .proc_handler = &proc_dointvec,
1023 },
1024 {
1025 .ctl_name = FS_AIO_NR,
1026 .procname = "aio-nr",
1027 .data = &aio_nr,
1028 .maxlen = sizeof(aio_nr),
1029 .mode = 0444,
d55b5fda 1030 .proc_handler = &proc_doulongvec_minmax,
1da177e4
LT		 1031 },
1032 {
1033 .ctl_name = FS_AIO_MAX_NR,
1034 .procname = "aio-max-nr",
1035 .data = &aio_max_nr,
1036 .maxlen = sizeof(aio_max_nr),
1037 .mode = 0644,
d55b5fda 1038 .proc_handler = &proc_doulongvec_minmax,
1da177e4 1039 },
2d9048e2 1040#ifdef CONFIG_INOTIFY_USER
0399cb08
RL		 1041 {
1042 .ctl_name = FS_INOTIFY,
1043 .procname = "inotify",
1044 .mode = 0555,
1045 .child = inotify_table,
1046 },
1047#endif
1da177e4 1048#endif
d6e71144
AC		 1049 {
1050 .ctl_name = KERN_SETUID_DUMPABLE,
1051 .procname = "suid_dumpable",
1052 .data = &suid_dumpable,
1053 .maxlen = sizeof(int),
1054 .mode = 0644,
1055 .proc_handler = &proc_dointvec,
1056 },
1da177e4
LT		 1057 { .ctl_name = 0 }
1058};
1059
1060static ctl_table debug_table[] = {
1061 { .ctl_name = 0 }
1062};
1063
1064static ctl_table dev_table[] = {
1065 { .ctl_name = 0 }
0eeca283 1066};
1da177e4
LT		 1067
1068extern void init_irq_proc (void);
1069
330d57fb
AV		 1070static DEFINE_SPINLOCK(sysctl_lock);
1071
1072/* called under sysctl_lock */
1073static int use_table(struct ctl_table_header *p)
1074{
1075 if (unlikely(p->unregistering))
1076 return 0;
1077 p->used++;
1078 return 1;
1079}
1080
1081/* called under sysctl_lock */
1082static void unuse_table(struct ctl_table_header *p)
1083{
1084 if (!--p->used)
1085 if (unlikely(p->unregistering))
1086 complete(p->unregistering);
1087}
1088
1089/* called under sysctl_lock, will reacquire if has to wait */
1090static void start_unregistering(struct ctl_table_header *p)
1091{
1092 /*
1093 * if p->used is 0, nobody will ever touch that entry again;
1094 * we'll eliminate all paths to it before dropping sysctl_lock
1095 */
1096 if (unlikely(p->used)) {
1097 struct completion wait;
1098 init_completion(&wait);
1099 p->unregistering = &wait;
1100 spin_unlock(&sysctl_lock);
1101 wait_for_completion(&wait);
1102 spin_lock(&sysctl_lock);
1103 }
1104 /*
1105 * do not remove from the list until nobody holds it; walking the
1106 * list in do_sysctl() relies on that.
1107 */
1108 list_del_init(&p->ctl_entry);
1109}
1110
1da177e4
LT		 1111void __init sysctl_init(void)
1112{
1113#ifdef CONFIG_PROC_FS
330d57fb 1114 register_proc_table(root_table, proc_sys_root, &root_table_header);
1da177e4
LT		 1115 init_irq_proc();
1116#endif
1117}
1118
1119int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1120 void __user *newval, size_t newlen)
1121{
1122 struct list_head *tmp;
330d57fb 1123 int error = -ENOTDIR;
1da177e4
LT		 1124
1125 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1126 return -ENOTDIR;
1127 if (oldval) {
1128 int old_len;
1129 if (!oldlenp || get_user(old_len, oldlenp))
1130 return -EFAULT;
1131 }
330d57fb 1132 spin_lock(&sysctl_lock);
1da177e4
LT		 1133 tmp = &root_table_header.ctl_entry;
1134 do {
1135 struct ctl_table_header *head =
1136 list_entry(tmp, struct ctl_table_header, ctl_entry);
1137 void *context = NULL;
330d57fb
AV		 1138
1139 if (!use_table(head))
1140 continue;
1141
1142 spin_unlock(&sysctl_lock);
1143
1144 error = parse_table(name, nlen, oldval, oldlenp,
1da177e4
LT		 1145 newval, newlen, head->ctl_table,
1146 &context);
5a6b454f 1147 kfree(context);
330d57fb
AV		 1148
1149 spin_lock(&sysctl_lock);
1150 unuse_table(head);
1da177e4 1151 if (error != -ENOTDIR)
330d57fb
AV		 1152 break;
1153 } while ((tmp = tmp->next) != &root_table_header.ctl_entry);
1154 spin_unlock(&sysctl_lock);
1155 return error;
1da177e4
LT		 1156}
1157
1158asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1159{
1160 struct __sysctl_args tmp;
1161 int error;
1162
1163 if (copy_from_user(&tmp, args, sizeof(tmp)))
1164 return -EFAULT;
1165
1166 lock_kernel();
1167 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1168 tmp.newval, tmp.newlen);
1169 unlock_kernel();
1170 return error;
1171}
1172
1173/*
1174 * ctl_perm does NOT grant the superuser all rights automatically, because
1175 * some sysctl variables are readonly even to root.
1176 */
1177
1178static int test_perm(int mode, int op)
1179{
1180 if (!current->euid)
1181 mode >>= 6;
1182 else if (in_egroup_p(0))
1183 mode >>= 3;
1184 if ((mode & op & 0007) == op)
1185 return 0;
1186 return -EACCES;
1187}
1188
1189static inline int ctl_perm(ctl_table *table, int op)
1190{
1191 int error;
1192 error = security_sysctl(table, op);
1193 if (error)
1194 return error;
1195 return test_perm(table->mode, op);
1196}
1197
1198static int parse_table(int __user *name, int nlen,
1199 void __user *oldval, size_t __user *oldlenp,
1200 void __user *newval, size_t newlen,
1201 ctl_table *table, void **context)
1202{
1203 int n;
1204repeat:
1205 if (!nlen)
1206 return -ENOTDIR;
1207 if (get_user(n, name))
1208 return -EFAULT;
1209 for ( ; table->ctl_name; table++) {
1210 if (n == table->ctl_name || table->ctl_name == CTL_ANY) {
1211 int error;
1212 if (table->child) {
1213 if (ctl_perm(table, 001))
1214 return -EPERM;
1215 if (table->strategy) {
1216 error = table->strategy(
1217 table, name, nlen,
1218 oldval, oldlenp,
1219 newval, newlen, context);
1220 if (error)
1221 return error;
1222 }
1223 name++;
1224 nlen--;
1225 table = table->child;
1226 goto repeat;
1227 }
1228 error = do_sysctl_strategy(table, name, nlen,
1229 oldval, oldlenp,
1230 newval, newlen, context);
1231 return error;
1232 }
1233 }
1234 return -ENOTDIR;
1235}
1236
1237/* Perform the actual read/write of a sysctl table entry. */
1238int do_sysctl_strategy (ctl_table *table,
1239 int __user *name, int nlen,
1240 void __user *oldval, size_t __user *oldlenp,
1241 void __user *newval, size_t newlen, void **context)
1242{
1243 int op = 0, rc;
1244 size_t len;
1245
1246 if (oldval)
1247 op |= 004;
1248 if (newval)
1249 op |= 002;
1250 if (ctl_perm(table, op))
1251 return -EPERM;
1252
1253 if (table->strategy) {
1254 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1255 newval, newlen, context);
1256 if (rc < 0)
1257 return rc;
1258 if (rc > 0)
1259 return 0;
1260 }
1261
1262 /* If there is no strategy routine, or if the strategy returns
1263 * zero, proceed with automatic r/w */
1264 if (table->data && table->maxlen) {
1265 if (oldval && oldlenp) {
1266 if (get_user(len, oldlenp))
1267 return -EFAULT;
1268 if (len) {
1269 if (len > table->maxlen)
1270 len = table->maxlen;
1271 if(copy_to_user(oldval, table->data, len))
1272 return -EFAULT;
1273 if(put_user(len, oldlenp))
1274 return -EFAULT;
1275 }
1276 }
1277 if (newval && newlen) {
1278 len = newlen;
1279 if (len > table->maxlen)
1280 len = table->maxlen;
1281 if(copy_from_user(table->data, newval, len))
1282 return -EFAULT;
1283 }
1284 }
1285 return 0;
1286}
1287
1288/**
1289 * register_sysctl_table - register a sysctl hierarchy
1290 * @table: the top-level table structure
1291 * @insert_at_head: whether the entry should be inserted in front or at the end
1292 *
1293 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1294 * array. An entry with a ctl_name of 0 terminates the table.
1295 *
1296 * The members of the &ctl_table structure are used as follows:
1297 *
1298 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1299 * must be unique within that level of sysctl
1300 *
1301 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1302 * enter a sysctl file
1303 *
1304 * data - a pointer to data for use by proc_handler
1305 *
1306 * maxlen - the maximum size in bytes of the data
1307 *
1308 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1309 *
1310 * child - a pointer to the child sysctl table if this entry is a directory, or
1311 * %NULL.
1312 *
1313 * proc_handler - the text handler routine (described below)
1314 *
1315 * strategy - the strategy routine (described below)
1316 *
1317 * de - for internal use by the sysctl routines
1318 *
1319 * extra1, extra2 - extra pointers usable by the proc handler routines
1320 *
1321 * Leaf nodes in the sysctl tree will be represented by a single file
1322 * under /proc; non-leaf nodes will be represented by directories.
1323 *
1324 * sysctl(2) can automatically manage read and write requests through
1325 * the sysctl table. The data and maxlen fields of the ctl_table
1326 * struct enable minimal validation of the values being written to be
1327 * performed, and the mode field allows minimal authentication.
1328 *
1329 * More sophisticated management can be enabled by the provision of a
1330 * strategy routine with the table entry. This will be called before
1331 * any automatic read or write of the data is performed.
1332 *
1333 * The strategy routine may return
1334 *
1335 * < 0 - Error occurred (error is passed to user process)
1336 *
1337 * 0 - OK - proceed with automatic read or write.
1338 *
1339 * > 0 - OK - read or write has been done by the strategy routine, so
1340 * return immediately.
1341 *
1342 * There must be a proc_handler routine for any terminal nodes
1343 * mirrored under /proc/sys (non-terminals are handled by a built-in
1344 * directory handler). Several default handlers are available to
1345 * cover common cases -
1346 *
1347 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1348 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1349 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1350 *
1351 * It is the handler's job to read the input buffer from user memory
1352 * and process it. The handler should return 0 on success.
1353 *
1354 * This routine returns %NULL on a failure to register, and a pointer
1355 * to the table header on success.
1356 */
1357struct ctl_table_header *register_sysctl_table(ctl_table * table,
1358 int insert_at_head)
1359{
1360 struct ctl_table_header *tmp;
1361 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1362 if (!tmp)
1363 return NULL;
1364 tmp->ctl_table = table;
1365 INIT_LIST_HEAD(&tmp->ctl_entry);
330d57fb
AV		 1366 tmp->used = 0;
1367 tmp->unregistering = NULL;
1368 spin_lock(&sysctl_lock);
1da177e4
LT		 1369 if (insert_at_head)
1370 list_add(&tmp->ctl_entry, &root_table_header.ctl_entry);
1371 else
1372 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
330d57fb 1373 spin_unlock(&sysctl_lock);
1da177e4 1374#ifdef CONFIG_PROC_FS
330d57fb 1375 register_proc_table(table, proc_sys_root, tmp);
1da177e4
LT		 1376#endif
1377 return tmp;
1378}
1379
1380/**
1381 * unregister_sysctl_table - unregister a sysctl table hierarchy
1382 * @header: the header returned from register_sysctl_table
1383 *
1384 * Unregisters the sysctl table and all children. proc entries may not
1385 * actually be removed until they are no longer used by anyone.
1386 */
1387void unregister_sysctl_table(struct ctl_table_header * header)
1388{
330d57fb
AV		 1389 might_sleep();
1390 spin_lock(&sysctl_lock);
1391 start_unregistering(header);
1da177e4
LT		 1392#ifdef CONFIG_PROC_FS
1393 unregister_proc_table(header->ctl_table, proc_sys_root);
1394#endif
330d57fb 1395 spin_unlock(&sysctl_lock);
1da177e4
LT		 1396 kfree(header);
1397}
1398
1399/*
1400 * /proc/sys support
1401 */
1402
1403#ifdef CONFIG_PROC_FS
1404
1405/* Scan the sysctl entries in table and add them all into /proc */
330d57fb 1406static void register_proc_table(ctl_table * table, struct proc_dir_entry *root, void *set)
1da177e4
LT		 1407{
1408 struct proc_dir_entry *de;
1409 int len;
1410 mode_t mode;
1411
1412 for (; table->ctl_name; table++) {
1413 /* Can't do anything without a proc name. */
1414 if (!table->procname)
1415 continue;
1416 /* Maybe we can't do anything with it... */
1417 if (!table->proc_handler && !table->child) {
1418 printk(KERN_WARNING "SYSCTL: Can't register %s\n",
1419 table->procname);
1420 continue;
1421 }
1422
1423 len = strlen(table->procname);
1424 mode = table->mode;
1425
1426 de = NULL;
1427 if (table->proc_handler)
1428 mode |= S_IFREG;
1429 else {
1430 mode |= S_IFDIR;
1431 for (de = root->subdir; de; de = de->next) {
1432 if (proc_match(len, table->procname, de))
1433 break;
1434 }
1435 /* If the subdir exists already, de is non-NULL */
1436 }
1437
1438 if (!de) {
1439 de = create_proc_entry(table->procname, mode, root);
1440 if (!de)
1441 continue;
330d57fb 1442 de->set = set;
1da177e4
LT		 1443 de->data = (void *) table;
1444 if (table->proc_handler)
1445 de->proc_fops = &proc_sys_file_operations;
1446 }
1447 table->de = de;
1448 if (de->mode & S_IFDIR)
330d57fb 1449 register_proc_table(table->child, de, set);
1da177e4
LT		 1450 }
1451}
1452
1453/*
1454 * Unregister a /proc sysctl table and any subdirectories.
1455 */
1456static void unregister_proc_table(ctl_table * table, struct proc_dir_entry *root)
1457{
1458 struct proc_dir_entry *de;
1459 for (; table->ctl_name; table++) {
1460 if (!(de = table->de))
1461 continue;
1462 if (de->mode & S_IFDIR) {
1463 if (!table->child) {
1464 printk (KERN_ALERT "Help - malformed sysctl tree on free\n");
1465 continue;
1466 }
1467 unregister_proc_table(table->child, de);
1468
1469 /* Don't unregister directories which still have entries.. */
1470 if (de->subdir)
1471 continue;
1472 }
1473
330d57fb
AV		 1474 /*
1475 * In any case, mark the entry as goner; we'll keep it
1476 * around if it's busy, but we'll know to do nothing with
1477 * its fields. We are under sysctl_lock here.
1478 */
1479 de->data = NULL;
1480
1da177e4
LT		 1481 /* Don't unregister proc entries that are still being used.. */
1482 if (atomic_read(&de->count))
1483 continue;
1484
1485 table->de = NULL;
1486 remove_proc_entry(table->procname, root);
1487 }
1488}
1489
1490static ssize_t do_rw_proc(int write, struct file * file, char __user * buf,
1491 size_t count, loff_t *ppos)
1492{
1493 int op;
330d57fb 1494 struct proc_dir_entry *de = PDE(file->f_dentry->d_inode);
1da177e4
LT		 1495 struct ctl_table *table;
1496 size_t res;
330d57fb 1497 ssize_t error = -ENOTDIR;
1da177e4 1498
330d57fb
AV		 1499 spin_lock(&sysctl_lock);
1500 if (de && de->data && use_table(de->set)) {
1501 /*
1502 * at that point we know that sysctl was not unregistered
1503 * and won't be until we finish
1504 */
1505 spin_unlock(&sysctl_lock);
1506 table = (struct ctl_table *) de->data;
1507 if (!table || !table->proc_handler)
1508 goto out;
1509 error = -EPERM;
1510 op = (write ? 002 : 004);
1511 if (ctl_perm(table, op))
1512 goto out;
1513
1514 /* careful: calling conventions are nasty here */
1515 res = count;
1516 error = (*table->proc_handler)(table, write, file,
1517 buf, &res, ppos);
1518 if (!error)
1519 error = res;
1520 out:
1521 spin_lock(&sysctl_lock);
1522 unuse_table(de->set);
1523 }
1524 spin_unlock(&sysctl_lock);
1525 return error;
1da177e4
LT		 1526}
1527
1528static int proc_opensys(struct inode *inode, struct file *file)
1529{
1530 if (file->f_mode & FMODE_WRITE) {
1531 /*
1532 * sysctl entries that are not writable,
1533 * are _NOT_ writable, capabilities or not.
1534 */
1535 if (!(inode->i_mode & S_IWUSR))
1536 return -EPERM;
1537 }
1538
1539 return 0;
1540}
1541
1542static ssize_t proc_readsys(struct file * file, char __user * buf,
1543 size_t count, loff_t *ppos)
1544{
1545 return do_rw_proc(0, file, buf, count, ppos);
1546}
1547
1548static ssize_t proc_writesys(struct file * file, const char __user * buf,
1549 size_t count, loff_t *ppos)
1550{
1551 return do_rw_proc(1, file, (char __user *) buf, count, ppos);
1552}
1553
1554/**
1555 * proc_dostring - read a string sysctl
1556 * @table: the sysctl table
1557 * @write: %TRUE if this is a write to the sysctl file
1558 * @filp: the file structure
1559 * @buffer: the user buffer
1560 * @lenp: the size of the user buffer
1561 * @ppos: file position
1562 *
1563 * Reads/writes a string from/to the user buffer. If the kernel
1564 * buffer provided is not large enough to hold the string, the
1565 * string is truncated. The copied string is %NULL-terminated.
1566 * If the string is being read by the user process, it is copied
1567 * and a newline '\n' is added. It is truncated if the buffer is
1568 * not large enough.
1569 *
1570 * Returns 0 on success.
1571 */
1572int proc_dostring(ctl_table *table, int write, struct file *filp,
1573 void __user *buffer, size_t *lenp, loff_t *ppos)
1574{
1575 size_t len;
1576 char __user *p;
1577 char c;
1578
1579 if (!table->data || !table->maxlen || !*lenp ||
1580 (*ppos && !write)) {
1581 *lenp = 0;
1582 return 0;
1583 }
1584
1585 if (write) {
1586 len = 0;
1587 p = buffer;
1588 while (len < *lenp) {
1589 if (get_user(c, p++))
1590 return -EFAULT;
1591 if (c == 0 || c == '\n')
1592 break;
1593 len++;
1594 }
1595 if (len >= table->maxlen)
1596 len = table->maxlen-1;
1597 if(copy_from_user(table->data, buffer, len))
1598 return -EFAULT;
1599 ((char *) table->data)[len] = 0;
1600 *ppos += *lenp;
1601 } else {
1602 len = strlen(table->data);
1603 if (len > table->maxlen)
1604 len = table->maxlen;
1605 if (len > *lenp)
1606 len = *lenp;
1607 if (len)
1608 if(copy_to_user(buffer, table->data, len))
1609 return -EFAULT;
1610 if (len < *lenp) {
1611 if(put_user('\n', ((char __user *) buffer) + len))
1612 return -EFAULT;
1613 len++;
1614 }
1615 *lenp = len;
1616 *ppos += len;
1617 }
1618 return 0;
1619}
1620
1621/*
1622 * Special case of dostring for the UTS structure. This has locks
1623 * to observe. Should this be in kernel/sys.c ????
1624 */
1625
1626static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
1627 void __user *buffer, size_t *lenp, loff_t *ppos)
1628{
1629 int r;
1630
1631 if (!write) {
1632 down_read(&uts_sem);
1633 r=proc_dostring(table,0,filp,buffer,lenp, ppos);
1634 up_read(&uts_sem);
1635 } else {
1636 down_write(&uts_sem);
1637 r=proc_dostring(table,1,filp,buffer,lenp, ppos);
1638 up_write(&uts_sem);
1639 }
1640 return r;
1641}
1642
1643static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1644 int *valp,
1645 int write, void *data)
1646{
1647 if (write) {
1648 *valp = *negp ? -*lvalp : *lvalp;
1649 } else {
1650 int val = *valp;
1651 if (val < 0) {
1652 *negp = -1;
1653 *lvalp = (unsigned long)-val;
1654 } else {
1655 *negp = 0;
1656 *lvalp = (unsigned long)val;
1657 }
1658 }
1659 return 0;
1660}
1661
1662static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1663 void __user *buffer, size_t *lenp, loff_t *ppos,
1664 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1665 int write, void *data),
1666 void *data)
1667{
1668#define TMPBUFLEN 21
1669 int *i, vleft, first=1, neg, val;
1670 unsigned long lval;
1671 size_t left, len;
1672
1673 char buf[TMPBUFLEN], *p;
1674 char __user *s = buffer;
1675
1676 if (!table->data || !table->maxlen || !*lenp ||
1677 (*ppos && !write)) {
1678 *lenp = 0;
1679 return 0;
1680 }
1681
1682 i = (int *) table->data;
1683 vleft = table->maxlen / sizeof(*i);
1684 left = *lenp;
1685
1686 if (!conv)
1687 conv = do_proc_dointvec_conv;
1688
1689 for (; left && vleft--; i++, first=0) {
1690 if (write) {
1691 while (left) {
1692 char c;
1693 if (get_user(c, s))
1694 return -EFAULT;
1695 if (!isspace(c))
1696 break;
1697 left--;
1698 s++;
1699 }
1700 if (!left)
1701 break;
1702 neg = 0;
1703 len = left;
1704 if (len > sizeof(buf) - 1)
1705 len = sizeof(buf) - 1;
1706 if (copy_from_user(buf, s, len))
1707 return -EFAULT;
1708 buf[len] = 0;
1709 p = buf;
1710 if (*p == '-' && left > 1) {
1711 neg = 1;
1712 left--, p++;
1713 }
1714 if (*p < '0' || *p > '9')
1715 break;
1716
1717 lval = simple_strtoul(p, &p, 0);
1718
1719 len = p-buf;
1720 if ((len < left) && *p && !isspace(*p))
1721 break;
1722 if (neg)
1723 val = -val;
1724 s += len;
1725 left -= len;
1726
1727 if (conv(&neg, &lval, i, 1, data))
1728 break;
1729 } else {
1730 p = buf;
1731 if (!first)
1732 *p++ = '\t';
1733
1734 if (conv(&neg, &lval, i, 0, data))
1735 break;
1736
1737 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1738 len = strlen(buf);
1739 if (len > left)
1740 len = left;
1741 if(copy_to_user(s, buf, len))
1742 return -EFAULT;
1743 left -= len;
1744 s += len;
1745 }
1746 }
1747
1748 if (!write && !first && left) {
1749 if(put_user('\n', s))
1750 return -EFAULT;
1751 left--, s++;
1752 }
1753 if (write) {
1754 while (left) {
1755 char c;
1756 if (get_user(c, s++))
1757 return -EFAULT;
1758 if (!isspace(c))
1759 break;
1760 left--;
1761 }
1762 }
1763 if (write && first)
1764 return -EINVAL;
1765 *lenp -= left;
1766 *ppos += *lenp;
1767 return 0;
1768#undef TMPBUFLEN
1769}
1770
1771/**
1772 * proc_dointvec - read a vector of integers
1773 * @table: the sysctl table
1774 * @write: %TRUE if this is a write to the sysctl file
1775 * @filp: the file structure
1776 * @buffer: the user buffer
1777 * @lenp: the size of the user buffer
1778 * @ppos: file position
1779 *
1780 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1781 * values from/to the user buffer, treated as an ASCII string.
1782 *
1783 * Returns 0 on success.
1784 */
1785int proc_dointvec(ctl_table *table, int write, struct file *filp,
1786 void __user *buffer, size_t *lenp, loff_t *ppos)
1787{
1788 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1789 NULL,NULL);
1790}
1791
1792#define OP_SET 0
1793#define OP_AND 1
1794#define OP_OR 2
1795#define OP_MAX 3
1796#define OP_MIN 4
1797
1798static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1799 int *valp,
1800 int write, void *data)
1801{
1802 int op = *(int *)data;
1803 if (write) {
1804 int val = *negp ? -*lvalp : *lvalp;
1805 switch(op) {
1806 case OP_SET: *valp = val; break;
1807 case OP_AND: *valp &= val; break;
1808 case OP_OR: *valp |= val; break;
1809 case OP_MAX: if(*valp < val)
1810 *valp = val;
1811 break;
1812 case OP_MIN: if(*valp > val)
1813 *valp = val;
1814 break;
1815 }
1816 } else {
1817 int val = *valp;
1818 if (val < 0) {
1819 *negp = -1;
1820 *lvalp = (unsigned long)-val;
1821 } else {
1822 *negp = 0;
1823 *lvalp = (unsigned long)val;
1824 }
1825 }
1826 return 0;
1827}
1828
1829/*
1830 * init may raise the set.
1831 */
1832
1833int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1834 void __user *buffer, size_t *lenp, loff_t *ppos)
1835{
1836 int op;
1837
1838 if (!capable(CAP_SYS_MODULE)) {
1839 return -EPERM;
1840 }
1841
1842 op = (current->pid == 1) ? OP_SET : OP_AND;
1843 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1844 do_proc_dointvec_bset_conv,&op);
1845}
1846
1847struct do_proc_dointvec_minmax_conv_param {
1848 int *min;
1849 int *max;
1850};
1851
1852static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1853 int *valp,
1854 int write, void *data)
1855{
1856 struct do_proc_dointvec_minmax_conv_param *param = data;
1857 if (write) {
1858 int val = *negp ? -*lvalp : *lvalp;
1859 if ((param->min && *param->min > val) ||
1860 (param->max && *param->max < val))
1861 return -EINVAL;
1862 *valp = val;
1863 } else {
1864 int val = *valp;
1865 if (val < 0) {
1866 *negp = -1;
1867 *lvalp = (unsigned long)-val;
1868 } else {
1869 *negp = 0;
1870 *lvalp = (unsigned long)val;
1871 }
1872 }
1873 return 0;
1874}
1875
1876/**
1877 * proc_dointvec_minmax - read a vector of integers with min/max values
1878 * @table: the sysctl table
1879 * @write: %TRUE if this is a write to the sysctl file
1880 * @filp: the file structure
1881 * @buffer: the user buffer
1882 * @lenp: the size of the user buffer
1883 * @ppos: file position
1884 *
1885 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1886 * values from/to the user buffer, treated as an ASCII string.
1887 *
1888 * This routine will ensure the values are within the range specified by
1889 * table->extra1 (min) and table->extra2 (max).
1890 *
1891 * Returns 0 on success.
1892 */
1893int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1894 void __user *buffer, size_t *lenp, loff_t *ppos)
1895{
1896 struct do_proc_dointvec_minmax_conv_param param = {
1897 .min = (int *) table->extra1,
1898 .max = (int *) table->extra2,
1899 };
1900 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1901 do_proc_dointvec_minmax_conv, &param);
1902}
1903
1904static int do_proc_doulongvec_minmax(ctl_table *table, int write,
1905 struct file *filp,
1906 void __user *buffer,
1907 size_t *lenp, loff_t *ppos,
1908 unsigned long convmul,
1909 unsigned long convdiv)
1910{
1911#define TMPBUFLEN 21
1912 unsigned long *i, *min, *max, val;
1913 int vleft, first=1, neg;
1914 size_t len, left;
1915 char buf[TMPBUFLEN], *p;
1916 char __user *s = buffer;
1917
1918 if (!table->data || !table->maxlen || !*lenp ||
1919 (*ppos && !write)) {
1920 *lenp = 0;
1921 return 0;
1922 }
1923
1924 i = (unsigned long *) table->data;
1925 min = (unsigned long *) table->extra1;
1926 max = (unsigned long *) table->extra2;
1927 vleft = table->maxlen / sizeof(unsigned long);
1928 left = *lenp;
1929
1930 for (; left && vleft--; i++, min++, max++, first=0) {
1931 if (write) {
1932 while (left) {
1933 char c;
1934 if (get_user(c, s))
1935 return -EFAULT;
1936 if (!isspace(c))
1937 break;
1938 left--;
1939 s++;
1940 }
1941 if (!left)
1942 break;
1943 neg = 0;
1944 len = left;
1945 if (len > TMPBUFLEN-1)
1946 len = TMPBUFLEN-1;
1947 if (copy_from_user(buf, s, len))
1948 return -EFAULT;
1949 buf[len] = 0;
1950 p = buf;
1951 if (*p == '-' && left > 1) {
1952 neg = 1;
1953 left--, p++;
1954 }
1955 if (*p < '0' || *p > '9')
1956 break;
1957 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
1958 len = p-buf;
1959 if ((len < left) && *p && !isspace(*p))
1960 break;
1961 if (neg)
1962 val = -val;
1963 s += len;
1964 left -= len;
1965
1966 if(neg)
1967 continue;
1968 if ((min && val < *min) || (max && val > *max))
1969 continue;
1970 *i = val;
1971 } else {
1972 p = buf;
1973 if (!first)
1974 *p++ = '\t';
1975 sprintf(p, "%lu", convdiv * (*i) / convmul);
1976 len = strlen(buf);
1977 if (len > left)
1978 len = left;
1979 if(copy_to_user(s, buf, len))
1980 return -EFAULT;
1981 left -= len;
1982 s += len;
1983 }
1984 }
1985
1986 if (!write && !first && left) {
1987 if(put_user('\n', s))
1988 return -EFAULT;
1989 left--, s++;
1990 }
1991 if (write) {
1992 while (left) {
1993 char c;
1994 if (get_user(c, s++))
1995 return -EFAULT;
1996 if (!isspace(c))
1997 break;
1998 left--;
1999 }
2000 }
2001 if (write && first)
2002 return -EINVAL;
2003 *lenp -= left;
2004 *ppos += *lenp;
2005 return 0;
2006#undef TMPBUFLEN
2007}
2008
2009/**
2010 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2011 * @table: the sysctl table
2012 * @write: %TRUE if this is a write to the sysctl file
2013 * @filp: the file structure
2014 * @buffer: the user buffer
2015 * @lenp: the size of the user buffer
2016 * @ppos: file position
2017 *
2018 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2019 * values from/to the user buffer, treated as an ASCII string.
2020 *
2021 * This routine will ensure the values are within the range specified by
2022 * table->extra1 (min) and table->extra2 (max).
2023 *
2024 * Returns 0 on success.
2025 */
2026int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2027 void __user *buffer, size_t *lenp, loff_t *ppos)
2028{
2029 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2030}
2031
2032/**
2033 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2034 * @table: the sysctl table
2035 * @write: %TRUE if this is a write to the sysctl file
2036 * @filp: the file structure
2037 * @buffer: the user buffer
2038 * @lenp: the size of the user buffer
2039 * @ppos: file position
2040 *
2041 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2042 * values from/to the user buffer, treated as an ASCII string. The values
2043 * are treated as milliseconds, and converted to jiffies when they are stored.
2044 *
2045 * This routine will ensure the values are within the range specified by
2046 * table->extra1 (min) and table->extra2 (max).
2047 *
2048 * Returns 0 on success.
2049 */
2050int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2051 struct file *filp,
2052 void __user *buffer,
2053 size_t *lenp, loff_t *ppos)
2054{
2055 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2056 lenp, ppos, HZ, 1000l);
2057}
2058
2059
2060static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2061 int *valp,
2062 int write, void *data)
2063{
2064 if (write) {
cba9f33d
BS		 2065 if (*lvalp > LONG_MAX / HZ)
2066 return 1;
1da177e4
LT		 2067 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2068 } else {
2069 int val = *valp;
2070 unsigned long lval;
2071 if (val < 0) {
2072 *negp = -1;
2073 lval = (unsigned long)-val;
2074 } else {
2075 *negp = 0;
2076 lval = (unsigned long)val;
2077 }
2078 *lvalp = lval / HZ;
2079 }
2080 return 0;
2081}
2082
2083static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2084 int *valp,
2085 int write, void *data)
2086{
2087 if (write) {
cba9f33d
BS		 2088 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2089 return 1;
1da177e4
LT		 2090 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2091 } else {
2092 int val = *valp;
2093 unsigned long lval;
2094 if (val < 0) {
2095 *negp = -1;
2096 lval = (unsigned long)-val;
2097 } else {
2098 *negp = 0;
2099 lval = (unsigned long)val;
2100 }
2101 *lvalp = jiffies_to_clock_t(lval);
2102 }
2103 return 0;
2104}
2105
2106static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2107 int *valp,
2108 int write, void *data)
2109{
2110 if (write) {
2111 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2112 } else {
2113 int val = *valp;
2114 unsigned long lval;
2115 if (val < 0) {
2116 *negp = -1;
2117 lval = (unsigned long)-val;
2118 } else {
2119 *negp = 0;
2120 lval = (unsigned long)val;
2121 }
2122 *lvalp = jiffies_to_msecs(lval);
2123 }
2124 return 0;
2125}
2126
2127/**
2128 * proc_dointvec_jiffies - read a vector of integers as seconds
2129 * @table: the sysctl table
2130 * @write: %TRUE if this is a write to the sysctl file
2131 * @filp: the file structure
2132 * @buffer: the user buffer
2133 * @lenp: the size of the user buffer
2134 * @ppos: file position
2135 *
2136 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2137 * values from/to the user buffer, treated as an ASCII string.
2138 * The values read are assumed to be in seconds, and are converted into
2139 * jiffies.
2140 *
2141 * Returns 0 on success.
2142 */
2143int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2144 void __user *buffer, size_t *lenp, loff_t *ppos)
2145{
2146 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2147 do_proc_dointvec_jiffies_conv,NULL);
2148}
2149
2150/**
2151 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2152 * @table: the sysctl table
2153 * @write: %TRUE if this is a write to the sysctl file
2154 * @filp: the file structure
2155 * @buffer: the user buffer
2156 * @lenp: the size of the user buffer
1e5d5331 2157 * @ppos: pointer to the file position
1da177e4
LT		 2158 *
2159 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2160 * values from/to the user buffer, treated as an ASCII string.
2161 * The values read are assumed to be in 1/USER_HZ seconds, and
2162 * are converted into jiffies.
2163 *
2164 * Returns 0 on success.
2165 */
2166int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2167 void __user *buffer, size_t *lenp, loff_t *ppos)
2168{
2169 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2170 do_proc_dointvec_userhz_jiffies_conv,NULL);
2171}
2172
2173/**
2174 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2175 * @table: the sysctl table
2176 * @write: %TRUE if this is a write to the sysctl file
2177 * @filp: the file structure
2178 * @buffer: the user buffer
2179 * @lenp: the size of the user buffer
67be2dd1
MW		 2180 * @ppos: file position
2181 * @ppos: the current position in the file
1da177e4
LT		 2182 *
2183 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2184 * values from/to the user buffer, treated as an ASCII string.
2185 * The values read are assumed to be in 1/1000 seconds, and
2186 * are converted into jiffies.
2187 *
2188 * Returns 0 on success.
2189 */
2190int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2191 void __user *buffer, size_t *lenp, loff_t *ppos)
2192{
2193 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2194 do_proc_dointvec_ms_jiffies_conv, NULL);
2195}
2196
2197#else /* CONFIG_PROC_FS */
2198
2199int proc_dostring(ctl_table *table, int write, struct file *filp,
2200 void __user *buffer, size_t *lenp, loff_t *ppos)
2201{
2202 return -ENOSYS;
2203}
2204
2205static int proc_doutsstring(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(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_bset(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_minmax(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_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_dointvec_userhz_jiffies(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_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2242 void __user *buffer, size_t *lenp, loff_t *ppos)
2243{
2244 return -ENOSYS;
2245}
2246
2247int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2248 void __user *buffer, size_t *lenp, loff_t *ppos)
2249{
2250 return -ENOSYS;
2251}
2252
2253int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2254 struct file *filp,
2255 void __user *buffer,
2256 size_t *lenp, loff_t *ppos)
2257{
2258 return -ENOSYS;
2259}
2260
2261
2262#endif /* CONFIG_PROC_FS */
2263
2264
2265/*
2266 * General sysctl support routines
2267 */
2268
2269/* The generic string strategy routine: */
2270int sysctl_string(ctl_table *table, int __user *name, int nlen,
2271 void __user *oldval, size_t __user *oldlenp,
2272 void __user *newval, size_t newlen, void **context)
2273{
1da177e4
LT		 2274 if (!table->data || !table->maxlen)
2275 return -ENOTDIR;
2276
2277 if (oldval && oldlenp) {
de9e007d
LT		 2278 size_t bufsize;
2279 if (get_user(bufsize, oldlenp))
1da177e4 2280 return -EFAULT;
de9e007d
LT		 2281 if (bufsize) {
2282 size_t len = strlen(table->data), copied;
2283
2284 /* This shouldn't trigger for a well-formed sysctl */
2285 if (len > table->maxlen)
1da177e4 2286 len = table->maxlen;
de9e007d
LT		 2287
2288 /* Copy up to a max of bufsize-1 bytes of the string */
2289 copied = (len >= bufsize) ? bufsize - 1 : len;
2290
2291 if (copy_to_user(oldval, table->data, copied) ||
2292 put_user(0, (char __user *)(oldval + copied)))
1da177e4 2293 return -EFAULT;
de9e007d 2294 if (put_user(len, oldlenp))
1da177e4
LT		 2295 return -EFAULT;
2296 }
2297 }
2298 if (newval && newlen) {
de9e007d 2299 size_t len = newlen;
1da177e4
LT		 2300 if (len > table->maxlen)
2301 len = table->maxlen;
2302 if(copy_from_user(table->data, newval, len))
2303 return -EFAULT;
2304 if (len == table->maxlen)
2305 len--;
2306 ((char *) table->data)[len] = 0;
2307 }
82c9df82 2308 return 1;
1da177e4
LT		 2309}
2310
2311/*
2312 * This function makes sure that all of the integers in the vector
2313 * are between the minimum and maximum values given in the arrays
2314 * table->extra1 and table->extra2, respectively.
2315 */
2316int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2317 void __user *oldval, size_t __user *oldlenp,
2318 void __user *newval, size_t newlen, void **context)
2319{
2320
2321 if (newval && newlen) {
2322 int __user *vec = (int __user *) newval;
2323 int *min = (int *) table->extra1;
2324 int *max = (int *) table->extra2;
2325 size_t length;
2326 int i;
2327
2328 if (newlen % sizeof(int) != 0)
2329 return -EINVAL;
2330
2331 if (!table->extra1 && !table->extra2)
2332 return 0;
2333
2334 if (newlen > table->maxlen)
2335 newlen = table->maxlen;
2336 length = newlen / sizeof(int);
2337
2338 for (i = 0; i < length; i++) {
2339 int value;
2340 if (get_user(value, vec + i))
2341 return -EFAULT;
2342 if (min && value < min[i])
2343 return -EINVAL;
2344 if (max && value > max[i])
2345 return -EINVAL;
2346 }
2347 }
2348 return 0;
2349}
2350
2351/* Strategy function to convert jiffies to seconds */
2352int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2353 void __user *oldval, size_t __user *oldlenp,
2354 void __user *newval, size_t newlen, void **context)
2355{
2356 if (oldval) {
2357 size_t olen;
2358 if (oldlenp) {
2359 if (get_user(olen, oldlenp))
2360 return -EFAULT;
2361 if (olen!=sizeof(int))
2362 return -EINVAL;
2363 }
2364 if (put_user(*(int *)(table->data)/HZ, (int __user *)oldval) ||
2365 (oldlenp && put_user(sizeof(int),oldlenp)))
2366 return -EFAULT;
2367 }
2368 if (newval && newlen) {
2369 int new;
2370 if (newlen != sizeof(int))
2371 return -EINVAL;
2372 if (get_user(new, (int __user *)newval))
2373 return -EFAULT;
2374 *(int *)(table->data) = new*HZ;
2375 }
2376 return 1;
2377}
2378
2379/* Strategy function to convert jiffies to seconds */
2380int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2381 void __user *oldval, size_t __user *oldlenp,
2382 void __user *newval, size_t newlen, void **context)
2383{
2384 if (oldval) {
2385 size_t olen;
2386 if (oldlenp) {
2387 if (get_user(olen, oldlenp))
2388 return -EFAULT;
2389 if (olen!=sizeof(int))
2390 return -EINVAL;
2391 }
2392 if (put_user(jiffies_to_msecs(*(int *)(table->data)), (int __user *)oldval) ||
2393 (oldlenp && put_user(sizeof(int),oldlenp)))
2394 return -EFAULT;
2395 }
2396 if (newval && newlen) {
2397 int new;
2398 if (newlen != sizeof(int))
2399 return -EINVAL;
2400 if (get_user(new, (int __user *)newval))
2401 return -EFAULT;
2402 *(int *)(table->data) = msecs_to_jiffies(new);
2403 }
2404 return 1;
2405}
2406
2407#else /* CONFIG_SYSCTL */
2408
2409
2410asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2411{
2412 return -ENOSYS;
2413}
2414
2415int sysctl_string(ctl_table *table, int __user *name, int nlen,
2416 void __user *oldval, size_t __user *oldlenp,
2417 void __user *newval, size_t newlen, void **context)
2418{
2419 return -ENOSYS;
2420}
2421
2422int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2423 void __user *oldval, size_t __user *oldlenp,
2424 void __user *newval, size_t newlen, void **context)
2425{
2426 return -ENOSYS;
2427}
2428
2429int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2430 void __user *oldval, size_t __user *oldlenp,
2431 void __user *newval, size_t newlen, void **context)
2432{
2433 return -ENOSYS;
2434}
2435
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 return -ENOSYS;
2441}
2442
2443int proc_dostring(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(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_bset(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_minmax(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_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_dointvec_userhz_jiffies(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_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2480 void __user *buffer, size_t *lenp, loff_t *ppos)
2481{
2482 return -ENOSYS;
2483}
2484
2485int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2486 void __user *buffer, size_t *lenp, loff_t *ppos)
2487{
2488 return -ENOSYS;
2489}
2490
2491int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2492 struct file *filp,
2493 void __user *buffer,
2494 size_t *lenp, loff_t *ppos)
2495{
2496 return -ENOSYS;
2497}
2498
2499struct ctl_table_header * register_sysctl_table(ctl_table * table,
2500 int insert_at_head)
2501{
2502 return NULL;
2503}
2504
2505void unregister_sysctl_table(struct ctl_table_header * table)
2506{
2507}
2508
2509#endif /* CONFIG_SYSCTL */
2510
2511/*
2512 * No sense putting this after each symbol definition, twice,
2513 * exception granted :-)
2514 */
2515EXPORT_SYMBOL(proc_dointvec);
2516EXPORT_SYMBOL(proc_dointvec_jiffies);
2517EXPORT_SYMBOL(proc_dointvec_minmax);
2518EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2519EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2520EXPORT_SYMBOL(proc_dostring);
2521EXPORT_SYMBOL(proc_doulongvec_minmax);
2522EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2523EXPORT_SYMBOL(register_sysctl_table);
2524EXPORT_SYMBOL(sysctl_intvec);
2525EXPORT_SYMBOL(sysctl_jiffies);
2526EXPORT_SYMBOL(sysctl_ms_jiffies);
2527EXPORT_SYMBOL(sysctl_string);
2528EXPORT_SYMBOL(unregister_sysctl_table);
Linux 6.x block layer and io_uring tree(s)