projects / linux-2.6-block.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
[PATCH] powermac: Fix loss of ethernet PHY on sleep
[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
1da177e4
LT		 53#if defined(CONFIG_SYSCTL)
54
55/* External variables not in a header file. */
56extern int C_A_D;
57extern int sysctl_overcommit_memory;
58extern int sysctl_overcommit_ratio;
59extern int max_threads;
60extern int sysrq_enabled;
61extern int core_uses_pid;
d6e71144 62extern int suid_dumpable;
1da177e4
LT		 63extern char core_pattern[];
64extern int cad_pid;
65extern int pid_max;
66extern int min_free_kbytes;
67extern int printk_ratelimit_jiffies;
68extern int printk_ratelimit_burst;
69extern int pid_max_min, pid_max_max;
9d0243bc 70extern int sysctl_drop_caches;
8ad4b1fb 71extern int percpu_pagelist_fraction;
1da177e4
LT		 72
73#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
74int unknown_nmi_panic;
75extern int proc_unknown_nmi_panic(ctl_table *, int, struct file *,
76 void __user *, size_t *, loff_t *);
77#endif
78
79/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
80static int maxolduid = 65535;
81static int minolduid;
8ad4b1fb 82static int min_percpu_pagelist_fract = 8;
1da177e4
LT		 83
84static int ngroups_max = NGROUPS_MAX;
85
86#ifdef CONFIG_KMOD
87extern char modprobe_path[];
88#endif
1da177e4
LT		 89#ifdef CONFIG_CHR_DEV_SG
90extern int sg_big_buff;
91#endif
92#ifdef CONFIG_SYSVIPC
93extern size_t shm_ctlmax;
94extern size_t shm_ctlall;
95extern int shm_ctlmni;
96extern int msg_ctlmax;
97extern int msg_ctlmnb;
98extern int msg_ctlmni;
99extern int sem_ctls[];
100#endif
101
102#ifdef __sparc__
103extern char reboot_command [];
104extern int stop_a_enabled;
105extern int scons_pwroff;
106#endif
107
108#ifdef __hppa__
109extern int pwrsw_enabled;
110extern int unaligned_enabled;
111#endif
112
347a8dc3 113#ifdef CONFIG_S390
1da177e4
LT		 114#ifdef CONFIG_MATHEMU
115extern int sysctl_ieee_emulation_warnings;
116#endif
117extern int sysctl_userprocess_debug;
951f22d5 118extern int spin_retry;
1da177e4
LT		 119#endif
120
121extern int sysctl_hz_timer;
122
123#ifdef CONFIG_BSD_PROCESS_ACCT
124extern int acct_parm[];
125#endif
126
1da177e4
LT		 127static int parse_table(int __user *, int, void __user *, size_t __user *, void __user *, size_t,
128 ctl_table *, void **);
129static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
130 void __user *buffer, size_t *lenp, loff_t *ppos);
131
132static ctl_table root_table[];
133static struct ctl_table_header root_table_header =
134 { root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
135
136static ctl_table kern_table[];
137static ctl_table vm_table[];
1da177e4
LT		 138static ctl_table proc_table[];
139static ctl_table fs_table[];
140static ctl_table debug_table[];
141static ctl_table dev_table[];
142extern ctl_table random_table[];
143#ifdef CONFIG_UNIX98_PTYS
144extern ctl_table pty_table[];
145#endif
0399cb08
RL		 146#ifdef CONFIG_INOTIFY
147extern ctl_table inotify_table[];
148#endif
1da177e4
LT		 149
150#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
151int sysctl_legacy_va_layout;
152#endif
153
154/* /proc declarations: */
155
156#ifdef CONFIG_PROC_FS
157
158static ssize_t proc_readsys(struct file *, char __user *, size_t, loff_t *);
159static ssize_t proc_writesys(struct file *, const char __user *, size_t, loff_t *);
160static int proc_opensys(struct inode *, struct file *);
161
162struct file_operations proc_sys_file_operations = {
163 .open = proc_opensys,
164 .read = proc_readsys,
165 .write = proc_writesys,
166};
167
168extern struct proc_dir_entry *proc_sys_root;
169
330d57fb 170static void register_proc_table(ctl_table *, struct proc_dir_entry *, void *);
1da177e4
LT		 171static void unregister_proc_table(ctl_table *, struct proc_dir_entry *);
172#endif
173
174/* The default sysctl tables: */
175
176static ctl_table root_table[] = {
177 {
178 .ctl_name = CTL_KERN,
179 .procname = "kernel",
180 .mode = 0555,
181 .child = kern_table,
182 },
183 {
184 .ctl_name = CTL_VM,
185 .procname = "vm",
186 .mode = 0555,
187 .child = vm_table,
188 },
189#ifdef CONFIG_NET
190 {
191 .ctl_name = CTL_NET,
192 .procname = "net",
193 .mode = 0555,
194 .child = net_table,
195 },
196#endif
197 {
198 .ctl_name = CTL_PROC,
199 .procname = "proc",
200 .mode = 0555,
201 .child = proc_table,
202 },
203 {
204 .ctl_name = CTL_FS,
205 .procname = "fs",
206 .mode = 0555,
207 .child = fs_table,
208 },
209 {
210 .ctl_name = CTL_DEBUG,
211 .procname = "debug",
212 .mode = 0555,
213 .child = debug_table,
214 },
215 {
216 .ctl_name = CTL_DEV,
217 .procname = "dev",
218 .mode = 0555,
219 .child = dev_table,
220 },
0eeca283 221
1da177e4
LT		 222 { .ctl_name = 0 }
223};
224
225static ctl_table kern_table[] = {
226 {
227 .ctl_name = KERN_OSTYPE,
228 .procname = "ostype",
229 .data = system_utsname.sysname,
230 .maxlen = sizeof(system_utsname.sysname),
231 .mode = 0444,
232 .proc_handler = &proc_doutsstring,
233 .strategy = &sysctl_string,
234 },
235 {
236 .ctl_name = KERN_OSRELEASE,
237 .procname = "osrelease",
238 .data = system_utsname.release,
239 .maxlen = sizeof(system_utsname.release),
240 .mode = 0444,
241 .proc_handler = &proc_doutsstring,
242 .strategy = &sysctl_string,
243 },
244 {
245 .ctl_name = KERN_VERSION,
246 .procname = "version",
247 .data = system_utsname.version,
248 .maxlen = sizeof(system_utsname.version),
249 .mode = 0444,
250 .proc_handler = &proc_doutsstring,
251 .strategy = &sysctl_string,
252 },
253 {
254 .ctl_name = KERN_NODENAME,
255 .procname = "hostname",
256 .data = system_utsname.nodename,
257 .maxlen = sizeof(system_utsname.nodename),
258 .mode = 0644,
259 .proc_handler = &proc_doutsstring,
260 .strategy = &sysctl_string,
261 },
262 {
263 .ctl_name = KERN_DOMAINNAME,
264 .procname = "domainname",
265 .data = system_utsname.domainname,
266 .maxlen = sizeof(system_utsname.domainname),
267 .mode = 0644,
268 .proc_handler = &proc_doutsstring,
269 .strategy = &sysctl_string,
270 },
271 {
272 .ctl_name = KERN_PANIC,
273 .procname = "panic",
274 .data = &panic_timeout,
275 .maxlen = sizeof(int),
276 .mode = 0644,
277 .proc_handler = &proc_dointvec,
278 },
279 {
280 .ctl_name = KERN_CORE_USES_PID,
281 .procname = "core_uses_pid",
282 .data = &core_uses_pid,
283 .maxlen = sizeof(int),
284 .mode = 0644,
285 .proc_handler = &proc_dointvec,
286 },
287 {
288 .ctl_name = KERN_CORE_PATTERN,
289 .procname = "core_pattern",
290 .data = core_pattern,
291 .maxlen = 64,
292 .mode = 0644,
293 .proc_handler = &proc_dostring,
294 .strategy = &sysctl_string,
295 },
296 {
297 .ctl_name = KERN_TAINTED,
298 .procname = "tainted",
299 .data = &tainted,
300 .maxlen = sizeof(int),
301 .mode = 0444,
302 .proc_handler = &proc_dointvec,
303 },
304 {
305 .ctl_name = KERN_CAP_BSET,
306 .procname = "cap-bound",
307 .data = &cap_bset,
308 .maxlen = sizeof(kernel_cap_t),
309 .mode = 0600,
310 .proc_handler = &proc_dointvec_bset,
311 },
312#ifdef CONFIG_BLK_DEV_INITRD
313 {
314 .ctl_name = KERN_REALROOTDEV,
315 .procname = "real-root-dev",
316 .data = &real_root_dev,
317 .maxlen = sizeof(int),
318 .mode = 0644,
319 .proc_handler = &proc_dointvec,
320 },
321#endif
322#ifdef __sparc__
323 {
324 .ctl_name = KERN_SPARC_REBOOT,
325 .procname = "reboot-cmd",
326 .data = reboot_command,
327 .maxlen = 256,
328 .mode = 0644,
329 .proc_handler = &proc_dostring,
330 .strategy = &sysctl_string,
331 },
332 {
333 .ctl_name = KERN_SPARC_STOP_A,
334 .procname = "stop-a",
335 .data = &stop_a_enabled,
336 .maxlen = sizeof (int),
337 .mode = 0644,
338 .proc_handler = &proc_dointvec,
339 },
340 {
341 .ctl_name = KERN_SPARC_SCONS_PWROFF,
342 .procname = "scons-poweroff",
343 .data = &scons_pwroff,
344 .maxlen = sizeof (int),
345 .mode = 0644,
346 .proc_handler = &proc_dointvec,
347 },
348#endif
349#ifdef __hppa__
350 {
351 .ctl_name = KERN_HPPA_PWRSW,
352 .procname = "soft-power",
353 .data = &pwrsw_enabled,
354 .maxlen = sizeof (int),
355 .mode = 0644,
356 .proc_handler = &proc_dointvec,
357 },
358 {
359 .ctl_name = KERN_HPPA_UNALIGNED,
360 .procname = "unaligned-trap",
361 .data = &unaligned_enabled,
362 .maxlen = sizeof (int),
363 .mode = 0644,
364 .proc_handler = &proc_dointvec,
365 },
366#endif
367 {
368 .ctl_name = KERN_CTLALTDEL,
369 .procname = "ctrl-alt-del",
370 .data = &C_A_D,
371 .maxlen = sizeof(int),
372 .mode = 0644,
373 .proc_handler = &proc_dointvec,
374 },
375 {
376 .ctl_name = KERN_PRINTK,
377 .procname = "printk",
378 .data = &console_loglevel,
379 .maxlen = 4*sizeof(int),
380 .mode = 0644,
381 .proc_handler = &proc_dointvec,
382 },
383#ifdef CONFIG_KMOD
384 {
385 .ctl_name = KERN_MODPROBE,
386 .procname = "modprobe",
387 .data = &modprobe_path,
388 .maxlen = KMOD_PATH_LEN,
389 .mode = 0644,
390 .proc_handler = &proc_dostring,
391 .strategy = &sysctl_string,
392 },
393#endif
394#ifdef CONFIG_HOTPLUG
395 {
396 .ctl_name = KERN_HOTPLUG,
397 .procname = "hotplug",
312c004d
KS		 398 .data = &uevent_helper,
399 .maxlen = UEVENT_HELPER_PATH_LEN,
1da177e4
LT		 400 .mode = 0644,
401 .proc_handler = &proc_dostring,
402 .strategy = &sysctl_string,
403 },
404#endif
405#ifdef CONFIG_CHR_DEV_SG
406 {
407 .ctl_name = KERN_SG_BIG_BUFF,
408 .procname = "sg-big-buff",
409 .data = &sg_big_buff,
410 .maxlen = sizeof (int),
411 .mode = 0444,
412 .proc_handler = &proc_dointvec,
413 },
414#endif
415#ifdef CONFIG_BSD_PROCESS_ACCT
416 {
417 .ctl_name = KERN_ACCT,
418 .procname = "acct",
419 .data = &acct_parm,
420 .maxlen = 3*sizeof(int),
421 .mode = 0644,
422 .proc_handler = &proc_dointvec,
423 },
424#endif
425#ifdef CONFIG_SYSVIPC
426 {
427 .ctl_name = KERN_SHMMAX,
428 .procname = "shmmax",
429 .data = &shm_ctlmax,
430 .maxlen = sizeof (size_t),
431 .mode = 0644,
432 .proc_handler = &proc_doulongvec_minmax,
433 },
434 {
435 .ctl_name = KERN_SHMALL,
436 .procname = "shmall",
437 .data = &shm_ctlall,
438 .maxlen = sizeof (size_t),
439 .mode = 0644,
440 .proc_handler = &proc_doulongvec_minmax,
441 },
442 {
443 .ctl_name = KERN_SHMMNI,
444 .procname = "shmmni",
445 .data = &shm_ctlmni,
446 .maxlen = sizeof (int),
447 .mode = 0644,
448 .proc_handler = &proc_dointvec,
449 },
450 {
451 .ctl_name = KERN_MSGMAX,
452 .procname = "msgmax",
453 .data = &msg_ctlmax,
454 .maxlen = sizeof (int),
455 .mode = 0644,
456 .proc_handler = &proc_dointvec,
457 },
458 {
459 .ctl_name = KERN_MSGMNI,
460 .procname = "msgmni",
461 .data = &msg_ctlmni,
462 .maxlen = sizeof (int),
463 .mode = 0644,
464 .proc_handler = &proc_dointvec,
465 },
466 {
467 .ctl_name = KERN_MSGMNB,
468 .procname = "msgmnb",
469 .data = &msg_ctlmnb,
470 .maxlen = sizeof (int),
471 .mode = 0644,
472 .proc_handler = &proc_dointvec,
473 },
474 {
475 .ctl_name = KERN_SEM,
476 .procname = "sem",
477 .data = &sem_ctls,
478 .maxlen = 4*sizeof (int),
479 .mode = 0644,
480 .proc_handler = &proc_dointvec,
481 },
482#endif
483#ifdef CONFIG_MAGIC_SYSRQ
484 {
485 .ctl_name = KERN_SYSRQ,
486 .procname = "sysrq",
487 .data = &sysrq_enabled,
488 .maxlen = sizeof (int),
489 .mode = 0644,
490 .proc_handler = &proc_dointvec,
491 },
492#endif
493 {
494 .ctl_name = KERN_CADPID,
495 .procname = "cad_pid",
496 .data = &cad_pid,
497 .maxlen = sizeof (int),
498 .mode = 0600,
499 .proc_handler = &proc_dointvec,
500 },
501 {
502 .ctl_name = KERN_MAX_THREADS,
503 .procname = "threads-max",
504 .data = &max_threads,
505 .maxlen = sizeof(int),
506 .mode = 0644,
507 .proc_handler = &proc_dointvec,
508 },
509 {
510 .ctl_name = KERN_RANDOM,
511 .procname = "random",
512 .mode = 0555,
513 .child = random_table,
514 },
515#ifdef CONFIG_UNIX98_PTYS
516 {
517 .ctl_name = KERN_PTY,
518 .procname = "pty",
519 .mode = 0555,
520 .child = pty_table,
521 },
522#endif
523 {
524 .ctl_name = KERN_OVERFLOWUID,
525 .procname = "overflowuid",
526 .data = &overflowuid,
527 .maxlen = sizeof(int),
528 .mode = 0644,
529 .proc_handler = &proc_dointvec_minmax,
530 .strategy = &sysctl_intvec,
531 .extra1 = &minolduid,
532 .extra2 = &maxolduid,
533 },
534 {
535 .ctl_name = KERN_OVERFLOWGID,
536 .procname = "overflowgid",
537 .data = &overflowgid,
538 .maxlen = sizeof(int),
539 .mode = 0644,
540 .proc_handler = &proc_dointvec_minmax,
541 .strategy = &sysctl_intvec,
542 .extra1 = &minolduid,
543 .extra2 = &maxolduid,
544 },
347a8dc3 545#ifdef CONFIG_S390
1da177e4
LT		 546#ifdef CONFIG_MATHEMU
547 {
548 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
549 .procname = "ieee_emulation_warnings",
550 .data = &sysctl_ieee_emulation_warnings,
551 .maxlen = sizeof(int),
552 .mode = 0644,
553 .proc_handler = &proc_dointvec,
554 },
555#endif
556#ifdef CONFIG_NO_IDLE_HZ
557 {
558 .ctl_name = KERN_HZ_TIMER,
559 .procname = "hz_timer",
560 .data = &sysctl_hz_timer,
561 .maxlen = sizeof(int),
562 .mode = 0644,
563 .proc_handler = &proc_dointvec,
564 },
565#endif
566 {
567 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
568 .procname = "userprocess_debug",
569 .data = &sysctl_userprocess_debug,
570 .maxlen = sizeof(int),
571 .mode = 0644,
572 .proc_handler = &proc_dointvec,
573 },
574#endif
575 {
576 .ctl_name = KERN_PIDMAX,
577 .procname = "pid_max",
578 .data = &pid_max,
579 .maxlen = sizeof (int),
580 .mode = 0644,
581 .proc_handler = &proc_dointvec_minmax,
582 .strategy = sysctl_intvec,
583 .extra1 = &pid_max_min,
584 .extra2 = &pid_max_max,
585 },
586 {
587 .ctl_name = KERN_PANIC_ON_OOPS,
588 .procname = "panic_on_oops",
589 .data = &panic_on_oops,
590 .maxlen = sizeof(int),
591 .mode = 0644,
592 .proc_handler = &proc_dointvec,
593 },
594 {
595 .ctl_name = KERN_PRINTK_RATELIMIT,
596 .procname = "printk_ratelimit",
597 .data = &printk_ratelimit_jiffies,
598 .maxlen = sizeof(int),
599 .mode = 0644,
600 .proc_handler = &proc_dointvec_jiffies,
601 .strategy = &sysctl_jiffies,
602 },
603 {
604 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
605 .procname = "printk_ratelimit_burst",
606 .data = &printk_ratelimit_burst,
607 .maxlen = sizeof(int),
608 .mode = 0644,
609 .proc_handler = &proc_dointvec,
610 },
611 {
612 .ctl_name = KERN_NGROUPS_MAX,
613 .procname = "ngroups_max",
614 .data = &ngroups_max,
615 .maxlen = sizeof (int),
616 .mode = 0444,
617 .proc_handler = &proc_dointvec,
618 },
619#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
620 {
621 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
622 .procname = "unknown_nmi_panic",
623 .data = &unknown_nmi_panic,
624 .maxlen = sizeof (int),
625 .mode = 0644,
626 .proc_handler = &proc_unknown_nmi_panic,
627 },
628#endif
629#if defined(CONFIG_X86)
630 {
631 .ctl_name = KERN_BOOTLOADER_TYPE,
632 .procname = "bootloader_type",
633 .data = &bootloader_type,
634 .maxlen = sizeof (int),
635 .mode = 0444,
636 .proc_handler = &proc_dointvec,
637 },
638#endif
639 {
640 .ctl_name = KERN_RANDOMIZE,
641 .procname = "randomize_va_space",
642 .data = &randomize_va_space,
643 .maxlen = sizeof(int),
644 .mode = 0644,
645 .proc_handler = &proc_dointvec,
646 },
0152fb37 647#if defined(CONFIG_S390) && defined(CONFIG_SMP)
951f22d5
MS		 648 {
649 .ctl_name = KERN_SPIN_RETRY,
650 .procname = "spin_retry",
651 .data = &spin_retry,
652 .maxlen = sizeof (int),
653 .mode = 0644,
654 .proc_handler = &proc_dointvec,
655 },
c255d844
PM		 656#endif
657#ifdef CONFIG_ACPI_SLEEP
658 {
659 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
660 .procname = "acpi_video_flags",
661 .data = &acpi_video_flags,
662 .maxlen = sizeof (unsigned long),
663 .mode = 0644,
664 .proc_handler = &proc_dointvec,
665 },
951f22d5 666#endif
1da177e4
LT		 667 { .ctl_name = 0 }
668};
669
670/* Constants for minimum and maximum testing in vm_table.
671 We use these as one-element integer vectors. */
672static int zero;
673static int one_hundred = 100;
674
675
676static ctl_table vm_table[] = {
677 {
678 .ctl_name = VM_OVERCOMMIT_MEMORY,
679 .procname = "overcommit_memory",
680 .data = &sysctl_overcommit_memory,
681 .maxlen = sizeof(sysctl_overcommit_memory),
682 .mode = 0644,
683 .proc_handler = &proc_dointvec,
684 },
685 {
686 .ctl_name = VM_OVERCOMMIT_RATIO,
687 .procname = "overcommit_ratio",
688 .data = &sysctl_overcommit_ratio,
689 .maxlen = sizeof(sysctl_overcommit_ratio),
690 .mode = 0644,
691 .proc_handler = &proc_dointvec,
692 },
693 {
694 .ctl_name = VM_PAGE_CLUSTER,
695 .procname = "page-cluster",
696 .data = &page_cluster,
697 .maxlen = sizeof(int),
698 .mode = 0644,
699 .proc_handler = &proc_dointvec,
700 },
701 {
702 .ctl_name = VM_DIRTY_BACKGROUND,
703 .procname = "dirty_background_ratio",
704 .data = &dirty_background_ratio,
705 .maxlen = sizeof(dirty_background_ratio),
706 .mode = 0644,
707 .proc_handler = &proc_dointvec_minmax,
708 .strategy = &sysctl_intvec,
709 .extra1 = &zero,
710 .extra2 = &one_hundred,
711 },
712 {
713 .ctl_name = VM_DIRTY_RATIO,
714 .procname = "dirty_ratio",
715 .data = &vm_dirty_ratio,
716 .maxlen = sizeof(vm_dirty_ratio),
717 .mode = 0644,
718 .proc_handler = &proc_dointvec_minmax,
719 .strategy = &sysctl_intvec,
720 .extra1 = &zero,
721 .extra2 = &one_hundred,
722 },
723 {
724 .ctl_name = VM_DIRTY_WB_CS,
725 .procname = "dirty_writeback_centisecs",
726 .data = &dirty_writeback_centisecs,
727 .maxlen = sizeof(dirty_writeback_centisecs),
728 .mode = 0644,
729 .proc_handler = &dirty_writeback_centisecs_handler,
730 },
731 {
732 .ctl_name = VM_DIRTY_EXPIRE_CS,
733 .procname = "dirty_expire_centisecs",
734 .data = &dirty_expire_centisecs,
735 .maxlen = sizeof(dirty_expire_centisecs),
736 .mode = 0644,
737 .proc_handler = &proc_dointvec,
738 },
739 {
740 .ctl_name = VM_NR_PDFLUSH_THREADS,
741 .procname = "nr_pdflush_threads",
742 .data = &nr_pdflush_threads,
743 .maxlen = sizeof nr_pdflush_threads,
744 .mode = 0444 /* read-only*/,
745 .proc_handler = &proc_dointvec,
746 },
747 {
748 .ctl_name = VM_SWAPPINESS,
749 .procname = "swappiness",
750 .data = &vm_swappiness,
751 .maxlen = sizeof(vm_swappiness),
752 .mode = 0644,
753 .proc_handler = &proc_dointvec_minmax,
754 .strategy = &sysctl_intvec,
755 .extra1 = &zero,
756 .extra2 = &one_hundred,
757 },
758#ifdef CONFIG_HUGETLB_PAGE
759 {
760 .ctl_name = VM_HUGETLB_PAGES,
761 .procname = "nr_hugepages",
762 .data = &max_huge_pages,
763 .maxlen = sizeof(unsigned long),
764 .mode = 0644,
765 .proc_handler = &hugetlb_sysctl_handler,
766 .extra1 = (void *)&hugetlb_zero,
767 .extra2 = (void *)&hugetlb_infinity,
768 },
769 {
770 .ctl_name = VM_HUGETLB_GROUP,
771 .procname = "hugetlb_shm_group",
772 .data = &sysctl_hugetlb_shm_group,
773 .maxlen = sizeof(gid_t),
774 .mode = 0644,
775 .proc_handler = &proc_dointvec,
776 },
777#endif
778 {
779 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
780 .procname = "lowmem_reserve_ratio",
781 .data = &sysctl_lowmem_reserve_ratio,
782 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
783 .mode = 0644,
784 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
785 .strategy = &sysctl_intvec,
786 },
9d0243bc
AM		 787 {
788 .ctl_name = VM_DROP_PAGECACHE,
789 .procname = "drop_caches",
790 .data = &sysctl_drop_caches,
791 .maxlen = sizeof(int),
792 .mode = 0644,
793 .proc_handler = drop_caches_sysctl_handler,
794 .strategy = &sysctl_intvec,
795 },
1da177e4
LT		 796 {
797 .ctl_name = VM_MIN_FREE_KBYTES,
798 .procname = "min_free_kbytes",
799 .data = &min_free_kbytes,
800 .maxlen = sizeof(min_free_kbytes),
801 .mode = 0644,
802 .proc_handler = &min_free_kbytes_sysctl_handler,
803 .strategy = &sysctl_intvec,
804 .extra1 = &zero,
805 },
8ad4b1fb
RS		 806 {
807 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
808 .procname = "percpu_pagelist_fraction",
809 .data = &percpu_pagelist_fraction,
810 .maxlen = sizeof(percpu_pagelist_fraction),
811 .mode = 0644,
812 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
813 .strategy = &sysctl_intvec,
814 .extra1 = &min_percpu_pagelist_fract,
815 },
1da177e4
LT		 816#ifdef CONFIG_MMU
817 {
818 .ctl_name = VM_MAX_MAP_COUNT,
819 .procname = "max_map_count",
820 .data = &sysctl_max_map_count,
821 .maxlen = sizeof(sysctl_max_map_count),
822 .mode = 0644,
823 .proc_handler = &proc_dointvec
824 },
825#endif
826 {
827 .ctl_name = VM_LAPTOP_MODE,
828 .procname = "laptop_mode",
829 .data = &laptop_mode,
830 .maxlen = sizeof(laptop_mode),
831 .mode = 0644,
832 .proc_handler = &proc_dointvec,
833 .strategy = &sysctl_intvec,
834 .extra1 = &zero,
835 },
836 {
837 .ctl_name = VM_BLOCK_DUMP,
838 .procname = "block_dump",
839 .data = &block_dump,
840 .maxlen = sizeof(block_dump),
841 .mode = 0644,
842 .proc_handler = &proc_dointvec,
843 .strategy = &sysctl_intvec,
844 .extra1 = &zero,
845 },
846 {
847 .ctl_name = VM_VFS_CACHE_PRESSURE,
848 .procname = "vfs_cache_pressure",
849 .data = &sysctl_vfs_cache_pressure,
850 .maxlen = sizeof(sysctl_vfs_cache_pressure),
851 .mode = 0644,
852 .proc_handler = &proc_dointvec,
853 .strategy = &sysctl_intvec,
854 .extra1 = &zero,
855 },
856#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
857 {
858 .ctl_name = VM_LEGACY_VA_LAYOUT,
859 .procname = "legacy_va_layout",
860 .data = &sysctl_legacy_va_layout,
861 .maxlen = sizeof(sysctl_legacy_va_layout),
862 .mode = 0644,
863 .proc_handler = &proc_dointvec,
864 .strategy = &sysctl_intvec,
865 .extra1 = &zero,
866 },
867#endif
868#ifdef CONFIG_SWAP
869 {
870 .ctl_name = VM_SWAP_TOKEN_TIMEOUT,
871 .procname = "swap_token_timeout",
872 .data = &swap_token_default_timeout,
873 .maxlen = sizeof(swap_token_default_timeout),
874 .mode = 0644,
875 .proc_handler = &proc_dointvec_jiffies,
876 .strategy = &sysctl_jiffies,
877 },
1743660b
CL		 878#endif
879#ifdef CONFIG_NUMA
880 {
881 .ctl_name = VM_ZONE_RECLAIM_MODE,
882 .procname = "zone_reclaim_mode",
883 .data = &zone_reclaim_mode,
884 .maxlen = sizeof(zone_reclaim_mode),
885 .mode = 0644,
886 .proc_handler = &proc_dointvec,
c84db23c
CL		 887 .strategy = &sysctl_intvec,
888 .extra1 = &zero,
1743660b 889 },
2a11ff06
CL		 890 {
891 .ctl_name = VM_ZONE_RECLAIM_INTERVAL,
892 .procname = "zone_reclaim_interval",
893 .data = &zone_reclaim_interval,
894 .maxlen = sizeof(zone_reclaim_interval),
895 .mode = 0644,
896 .proc_handler = &proc_dointvec_jiffies,
897 .strategy = &sysctl_jiffies,
898 },
1da177e4
LT		 899#endif
900 { .ctl_name = 0 }
901};
902
903static ctl_table proc_table[] = {
904 { .ctl_name = 0 }
905};
906
907static ctl_table fs_table[] = {
908 {
909 .ctl_name = FS_NRINODE,
910 .procname = "inode-nr",
911 .data = &inodes_stat,
912 .maxlen = 2*sizeof(int),
913 .mode = 0444,
914 .proc_handler = &proc_dointvec,
915 },
916 {
917 .ctl_name = FS_STATINODE,
918 .procname = "inode-state",
919 .data = &inodes_stat,
920 .maxlen = 7*sizeof(int),
921 .mode = 0444,
922 .proc_handler = &proc_dointvec,
923 },
924 {
925 .ctl_name = FS_NRFILE,
926 .procname = "file-nr",
927 .data = &files_stat,
928 .maxlen = 3*sizeof(int),
929 .mode = 0444,
930 .proc_handler = &proc_dointvec,
931 },
932 {
933 .ctl_name = FS_MAXFILE,
934 .procname = "file-max",
935 .data = &files_stat.max_files,
936 .maxlen = sizeof(int),
937 .mode = 0644,
938 .proc_handler = &proc_dointvec,
939 },
940 {
941 .ctl_name = FS_DENTRY,
942 .procname = "dentry-state",
943 .data = &dentry_stat,
944 .maxlen = 6*sizeof(int),
945 .mode = 0444,
946 .proc_handler = &proc_dointvec,
947 },
948 {
949 .ctl_name = FS_OVERFLOWUID,
950 .procname = "overflowuid",
951 .data = &fs_overflowuid,
952 .maxlen = sizeof(int),
953 .mode = 0644,
954 .proc_handler = &proc_dointvec_minmax,
955 .strategy = &sysctl_intvec,
956 .extra1 = &minolduid,
957 .extra2 = &maxolduid,
958 },
959 {
960 .ctl_name = FS_OVERFLOWGID,
961 .procname = "overflowgid",
962 .data = &fs_overflowgid,
963 .maxlen = sizeof(int),
964 .mode = 0644,
965 .proc_handler = &proc_dointvec_minmax,
966 .strategy = &sysctl_intvec,
967 .extra1 = &minolduid,
968 .extra2 = &maxolduid,
969 },
970 {
971 .ctl_name = FS_LEASES,
972 .procname = "leases-enable",
973 .data = &leases_enable,
974 .maxlen = sizeof(int),
975 .mode = 0644,
976 .proc_handler = &proc_dointvec,
977 },
978#ifdef CONFIG_DNOTIFY
979 {
980 .ctl_name = FS_DIR_NOTIFY,
981 .procname = "dir-notify-enable",
982 .data = &dir_notify_enable,
983 .maxlen = sizeof(int),
984 .mode = 0644,
985 .proc_handler = &proc_dointvec,
986 },
987#endif
988#ifdef CONFIG_MMU
989 {
990 .ctl_name = FS_LEASE_TIME,
991 .procname = "lease-break-time",
992 .data = &lease_break_time,
993 .maxlen = sizeof(int),
994 .mode = 0644,
995 .proc_handler = &proc_dointvec,
996 },
997 {
998 .ctl_name = FS_AIO_NR,
999 .procname = "aio-nr",
1000 .data = &aio_nr,
1001 .maxlen = sizeof(aio_nr),
1002 .mode = 0444,
d55b5fda 1003 .proc_handler = &proc_doulongvec_minmax,
1da177e4
LT		 1004 },
1005 {
1006 .ctl_name = FS_AIO_MAX_NR,
1007 .procname = "aio-max-nr",
1008 .data = &aio_max_nr,
1009 .maxlen = sizeof(aio_max_nr),
1010 .mode = 0644,
d55b5fda 1011 .proc_handler = &proc_doulongvec_minmax,
1da177e4 1012 },
0399cb08
RL		 1013#ifdef CONFIG_INOTIFY
1014 {
1015 .ctl_name = FS_INOTIFY,
1016 .procname = "inotify",
1017 .mode = 0555,
1018 .child = inotify_table,
1019 },
1020#endif
1da177e4 1021#endif
d6e71144
AC		 1022 {
1023 .ctl_name = KERN_SETUID_DUMPABLE,
1024 .procname = "suid_dumpable",
1025 .data = &suid_dumpable,
1026 .maxlen = sizeof(int),
1027 .mode = 0644,
1028 .proc_handler = &proc_dointvec,
1029 },
1da177e4
LT		 1030 { .ctl_name = 0 }
1031};
1032
1033static ctl_table debug_table[] = {
1034 { .ctl_name = 0 }
1035};
1036
1037static ctl_table dev_table[] = {
1038 { .ctl_name = 0 }
0eeca283 1039};
1da177e4
LT		 1040
1041extern void init_irq_proc (void);
1042
330d57fb
AV		 1043static DEFINE_SPINLOCK(sysctl_lock);
1044
1045/* called under sysctl_lock */
1046static int use_table(struct ctl_table_header *p)
1047{
1048 if (unlikely(p->unregistering))
1049 return 0;
1050 p->used++;
1051 return 1;
1052}
1053
1054/* called under sysctl_lock */
1055static void unuse_table(struct ctl_table_header *p)
1056{
1057 if (!--p->used)
1058 if (unlikely(p->unregistering))
1059 complete(p->unregistering);
1060}
1061
1062/* called under sysctl_lock, will reacquire if has to wait */
1063static void start_unregistering(struct ctl_table_header *p)
1064{
1065 /*
1066 * if p->used is 0, nobody will ever touch that entry again;
1067 * we'll eliminate all paths to it before dropping sysctl_lock
1068 */
1069 if (unlikely(p->used)) {
1070 struct completion wait;
1071 init_completion(&wait);
1072 p->unregistering = &wait;
1073 spin_unlock(&sysctl_lock);
1074 wait_for_completion(&wait);
1075 spin_lock(&sysctl_lock);
1076 }
1077 /*
1078 * do not remove from the list until nobody holds it; walking the
1079 * list in do_sysctl() relies on that.
1080 */
1081 list_del_init(&p->ctl_entry);
1082}
1083
1da177e4
LT		 1084void __init sysctl_init(void)
1085{
1086#ifdef CONFIG_PROC_FS
330d57fb 1087 register_proc_table(root_table, proc_sys_root, &root_table_header);
1da177e4
LT		 1088 init_irq_proc();
1089#endif
1090}
1091
1092int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1093 void __user *newval, size_t newlen)
1094{
1095 struct list_head *tmp;
330d57fb 1096 int error = -ENOTDIR;
1da177e4
LT		 1097
1098 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1099 return -ENOTDIR;
1100 if (oldval) {
1101 int old_len;
1102 if (!oldlenp || get_user(old_len, oldlenp))
1103 return -EFAULT;
1104 }
330d57fb 1105 spin_lock(&sysctl_lock);
1da177e4
LT		 1106 tmp = &root_table_header.ctl_entry;
1107 do {
1108 struct ctl_table_header *head =
1109 list_entry(tmp, struct ctl_table_header, ctl_entry);
1110 void *context = NULL;
330d57fb
AV		 1111
1112 if (!use_table(head))
1113 continue;
1114
1115 spin_unlock(&sysctl_lock);
1116
1117 error = parse_table(name, nlen, oldval, oldlenp,
1da177e4
LT		 1118 newval, newlen, head->ctl_table,
1119 &context);
5a6b454f 1120 kfree(context);
330d57fb
AV		 1121
1122 spin_lock(&sysctl_lock);
1123 unuse_table(head);
1da177e4 1124 if (error != -ENOTDIR)
330d57fb
AV		 1125 break;
1126 } while ((tmp = tmp->next) != &root_table_header.ctl_entry);
1127 spin_unlock(&sysctl_lock);
1128 return error;
1da177e4
LT		 1129}
1130
1131asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1132{
1133 struct __sysctl_args tmp;
1134 int error;
1135
1136 if (copy_from_user(&tmp, args, sizeof(tmp)))
1137 return -EFAULT;
1138
1139 lock_kernel();
1140 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1141 tmp.newval, tmp.newlen);
1142 unlock_kernel();
1143 return error;
1144}
1145
1146/*
1147 * ctl_perm does NOT grant the superuser all rights automatically, because
1148 * some sysctl variables are readonly even to root.
1149 */
1150
1151static int test_perm(int mode, int op)
1152{
1153 if (!current->euid)
1154 mode >>= 6;
1155 else if (in_egroup_p(0))
1156 mode >>= 3;
1157 if ((mode & op & 0007) == op)
1158 return 0;
1159 return -EACCES;
1160}
1161
1162static inline int ctl_perm(ctl_table *table, int op)
1163{
1164 int error;
1165 error = security_sysctl(table, op);
1166 if (error)
1167 return error;
1168 return test_perm(table->mode, op);
1169}
1170
1171static int parse_table(int __user *name, int nlen,
1172 void __user *oldval, size_t __user *oldlenp,
1173 void __user *newval, size_t newlen,
1174 ctl_table *table, void **context)
1175{
1176 int n;
1177repeat:
1178 if (!nlen)
1179 return -ENOTDIR;
1180 if (get_user(n, name))
1181 return -EFAULT;
1182 for ( ; table->ctl_name; table++) {
1183 if (n == table->ctl_name || table->ctl_name == CTL_ANY) {
1184 int error;
1185 if (table->child) {
1186 if (ctl_perm(table, 001))
1187 return -EPERM;
1188 if (table->strategy) {
1189 error = table->strategy(
1190 table, name, nlen,
1191 oldval, oldlenp,
1192 newval, newlen, context);
1193 if (error)
1194 return error;
1195 }
1196 name++;
1197 nlen--;
1198 table = table->child;
1199 goto repeat;
1200 }
1201 error = do_sysctl_strategy(table, name, nlen,
1202 oldval, oldlenp,
1203 newval, newlen, context);
1204 return error;
1205 }
1206 }
1207 return -ENOTDIR;
1208}
1209
1210/* Perform the actual read/write of a sysctl table entry. */
1211int do_sysctl_strategy (ctl_table *table,
1212 int __user *name, int nlen,
1213 void __user *oldval, size_t __user *oldlenp,
1214 void __user *newval, size_t newlen, void **context)
1215{
1216 int op = 0, rc;
1217 size_t len;
1218
1219 if (oldval)
1220 op |= 004;
1221 if (newval)
1222 op |= 002;
1223 if (ctl_perm(table, op))
1224 return -EPERM;
1225
1226 if (table->strategy) {
1227 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1228 newval, newlen, context);
1229 if (rc < 0)
1230 return rc;
1231 if (rc > 0)
1232 return 0;
1233 }
1234
1235 /* If there is no strategy routine, or if the strategy returns
1236 * zero, proceed with automatic r/w */
1237 if (table->data && table->maxlen) {
1238 if (oldval && oldlenp) {
1239 if (get_user(len, oldlenp))
1240 return -EFAULT;
1241 if (len) {
1242 if (len > table->maxlen)
1243 len = table->maxlen;
1244 if(copy_to_user(oldval, table->data, len))
1245 return -EFAULT;
1246 if(put_user(len, oldlenp))
1247 return -EFAULT;
1248 }
1249 }
1250 if (newval && newlen) {
1251 len = newlen;
1252 if (len > table->maxlen)
1253 len = table->maxlen;
1254 if(copy_from_user(table->data, newval, len))
1255 return -EFAULT;
1256 }
1257 }
1258 return 0;
1259}
1260
1261/**
1262 * register_sysctl_table - register a sysctl hierarchy
1263 * @table: the top-level table structure
1264 * @insert_at_head: whether the entry should be inserted in front or at the end
1265 *
1266 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1267 * array. An entry with a ctl_name of 0 terminates the table.
1268 *
1269 * The members of the &ctl_table structure are used as follows:
1270 *
1271 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1272 * must be unique within that level of sysctl
1273 *
1274 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1275 * enter a sysctl file
1276 *
1277 * data - a pointer to data for use by proc_handler
1278 *
1279 * maxlen - the maximum size in bytes of the data
1280 *
1281 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1282 *
1283 * child - a pointer to the child sysctl table if this entry is a directory, or
1284 * %NULL.
1285 *
1286 * proc_handler - the text handler routine (described below)
1287 *
1288 * strategy - the strategy routine (described below)
1289 *
1290 * de - for internal use by the sysctl routines
1291 *
1292 * extra1, extra2 - extra pointers usable by the proc handler routines
1293 *
1294 * Leaf nodes in the sysctl tree will be represented by a single file
1295 * under /proc; non-leaf nodes will be represented by directories.
1296 *
1297 * sysctl(2) can automatically manage read and write requests through
1298 * the sysctl table. The data and maxlen fields of the ctl_table
1299 * struct enable minimal validation of the values being written to be
1300 * performed, and the mode field allows minimal authentication.
1301 *
1302 * More sophisticated management can be enabled by the provision of a
1303 * strategy routine with the table entry. This will be called before
1304 * any automatic read or write of the data is performed.
1305 *
1306 * The strategy routine may return
1307 *
1308 * < 0 - Error occurred (error is passed to user process)
1309 *
1310 * 0 - OK - proceed with automatic read or write.
1311 *
1312 * > 0 - OK - read or write has been done by the strategy routine, so
1313 * return immediately.
1314 *
1315 * There must be a proc_handler routine for any terminal nodes
1316 * mirrored under /proc/sys (non-terminals are handled by a built-in
1317 * directory handler). Several default handlers are available to
1318 * cover common cases -
1319 *
1320 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1321 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1322 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1323 *
1324 * It is the handler's job to read the input buffer from user memory
1325 * and process it. The handler should return 0 on success.
1326 *
1327 * This routine returns %NULL on a failure to register, and a pointer
1328 * to the table header on success.
1329 */
1330struct ctl_table_header *register_sysctl_table(ctl_table * table,
1331 int insert_at_head)
1332{
1333 struct ctl_table_header *tmp;
1334 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1335 if (!tmp)
1336 return NULL;
1337 tmp->ctl_table = table;
1338 INIT_LIST_HEAD(&tmp->ctl_entry);
330d57fb
AV		 1339 tmp->used = 0;
1340 tmp->unregistering = NULL;
1341 spin_lock(&sysctl_lock);
1da177e4
LT		 1342 if (insert_at_head)
1343 list_add(&tmp->ctl_entry, &root_table_header.ctl_entry);
1344 else
1345 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
330d57fb 1346 spin_unlock(&sysctl_lock);
1da177e4 1347#ifdef CONFIG_PROC_FS
330d57fb 1348 register_proc_table(table, proc_sys_root, tmp);
1da177e4
LT		 1349#endif
1350 return tmp;
1351}
1352
1353/**
1354 * unregister_sysctl_table - unregister a sysctl table hierarchy
1355 * @header: the header returned from register_sysctl_table
1356 *
1357 * Unregisters the sysctl table and all children. proc entries may not
1358 * actually be removed until they are no longer used by anyone.
1359 */
1360void unregister_sysctl_table(struct ctl_table_header * header)
1361{
330d57fb
AV		 1362 might_sleep();
1363 spin_lock(&sysctl_lock);
1364 start_unregistering(header);
1da177e4
LT		 1365#ifdef CONFIG_PROC_FS
1366 unregister_proc_table(header->ctl_table, proc_sys_root);
1367#endif
330d57fb 1368 spin_unlock(&sysctl_lock);
1da177e4
LT		 1369 kfree(header);
1370}
1371
1372/*
1373 * /proc/sys support
1374 */
1375
1376#ifdef CONFIG_PROC_FS
1377
1378/* Scan the sysctl entries in table and add them all into /proc */
330d57fb 1379static void register_proc_table(ctl_table * table, struct proc_dir_entry *root, void *set)
1da177e4
LT		 1380{
1381 struct proc_dir_entry *de;
1382 int len;
1383 mode_t mode;
1384
1385 for (; table->ctl_name; table++) {
1386 /* Can't do anything without a proc name. */
1387 if (!table->procname)
1388 continue;
1389 /* Maybe we can't do anything with it... */
1390 if (!table->proc_handler && !table->child) {
1391 printk(KERN_WARNING "SYSCTL: Can't register %s\n",
1392 table->procname);
1393 continue;
1394 }
1395
1396 len = strlen(table->procname);
1397 mode = table->mode;
1398
1399 de = NULL;
1400 if (table->proc_handler)
1401 mode |= S_IFREG;
1402 else {
1403 mode |= S_IFDIR;
1404 for (de = root->subdir; de; de = de->next) {
1405 if (proc_match(len, table->procname, de))
1406 break;
1407 }
1408 /* If the subdir exists already, de is non-NULL */
1409 }
1410
1411 if (!de) {
1412 de = create_proc_entry(table->procname, mode, root);
1413 if (!de)
1414 continue;
330d57fb 1415 de->set = set;
1da177e4
LT		 1416 de->data = (void *) table;
1417 if (table->proc_handler)
1418 de->proc_fops = &proc_sys_file_operations;
1419 }
1420 table->de = de;
1421 if (de->mode & S_IFDIR)
330d57fb 1422 register_proc_table(table->child, de, set);
1da177e4
LT		 1423 }
1424}
1425
1426/*
1427 * Unregister a /proc sysctl table and any subdirectories.
1428 */
1429static void unregister_proc_table(ctl_table * table, struct proc_dir_entry *root)
1430{
1431 struct proc_dir_entry *de;
1432 for (; table->ctl_name; table++) {
1433 if (!(de = table->de))
1434 continue;
1435 if (de->mode & S_IFDIR) {
1436 if (!table->child) {
1437 printk (KERN_ALERT "Help - malformed sysctl tree on free\n");
1438 continue;
1439 }
1440 unregister_proc_table(table->child, de);
1441
1442 /* Don't unregister directories which still have entries.. */
1443 if (de->subdir)
1444 continue;
1445 }
1446
330d57fb
AV		 1447 /*
1448 * In any case, mark the entry as goner; we'll keep it
1449 * around if it's busy, but we'll know to do nothing with
1450 * its fields. We are under sysctl_lock here.
1451 */
1452 de->data = NULL;
1453
1da177e4
LT		 1454 /* Don't unregister proc entries that are still being used.. */
1455 if (atomic_read(&de->count))
1456 continue;
1457
1458 table->de = NULL;
1459 remove_proc_entry(table->procname, root);
1460 }
1461}
1462
1463static ssize_t do_rw_proc(int write, struct file * file, char __user * buf,
1464 size_t count, loff_t *ppos)
1465{
1466 int op;
330d57fb 1467 struct proc_dir_entry *de = PDE(file->f_dentry->d_inode);
1da177e4
LT		 1468 struct ctl_table *table;
1469 size_t res;
330d57fb 1470 ssize_t error = -ENOTDIR;
1da177e4 1471
330d57fb
AV		 1472 spin_lock(&sysctl_lock);
1473 if (de && de->data && use_table(de->set)) {
1474 /*
1475 * at that point we know that sysctl was not unregistered
1476 * and won't be until we finish
1477 */
1478 spin_unlock(&sysctl_lock);
1479 table = (struct ctl_table *) de->data;
1480 if (!table || !table->proc_handler)
1481 goto out;
1482 error = -EPERM;
1483 op = (write ? 002 : 004);
1484 if (ctl_perm(table, op))
1485 goto out;
1486
1487 /* careful: calling conventions are nasty here */
1488 res = count;
1489 error = (*table->proc_handler)(table, write, file,
1490 buf, &res, ppos);
1491 if (!error)
1492 error = res;
1493 out:
1494 spin_lock(&sysctl_lock);
1495 unuse_table(de->set);
1496 }
1497 spin_unlock(&sysctl_lock);
1498 return error;
1da177e4
LT		 1499}
1500
1501static int proc_opensys(struct inode *inode, struct file *file)
1502{
1503 if (file->f_mode & FMODE_WRITE) {
1504 /*
1505 * sysctl entries that are not writable,
1506 * are _NOT_ writable, capabilities or not.
1507 */
1508 if (!(inode->i_mode & S_IWUSR))
1509 return -EPERM;
1510 }
1511
1512 return 0;
1513}
1514
1515static ssize_t proc_readsys(struct file * file, char __user * buf,
1516 size_t count, loff_t *ppos)
1517{
1518 return do_rw_proc(0, file, buf, count, ppos);
1519}
1520
1521static ssize_t proc_writesys(struct file * file, const char __user * buf,
1522 size_t count, loff_t *ppos)
1523{
1524 return do_rw_proc(1, file, (char __user *) buf, count, ppos);
1525}
1526
1527/**
1528 * proc_dostring - read a string sysctl
1529 * @table: the sysctl table
1530 * @write: %TRUE if this is a write to the sysctl file
1531 * @filp: the file structure
1532 * @buffer: the user buffer
1533 * @lenp: the size of the user buffer
1534 * @ppos: file position
1535 *
1536 * Reads/writes a string from/to the user buffer. If the kernel
1537 * buffer provided is not large enough to hold the string, the
1538 * string is truncated. The copied string is %NULL-terminated.
1539 * If the string is being read by the user process, it is copied
1540 * and a newline '\n' is added. It is truncated if the buffer is
1541 * not large enough.
1542 *
1543 * Returns 0 on success.
1544 */
1545int proc_dostring(ctl_table *table, int write, struct file *filp,
1546 void __user *buffer, size_t *lenp, loff_t *ppos)
1547{
1548 size_t len;
1549 char __user *p;
1550 char c;
1551
1552 if (!table->data || !table->maxlen || !*lenp ||
1553 (*ppos && !write)) {
1554 *lenp = 0;
1555 return 0;
1556 }
1557
1558 if (write) {
1559 len = 0;
1560 p = buffer;
1561 while (len < *lenp) {
1562 if (get_user(c, p++))
1563 return -EFAULT;
1564 if (c == 0 || c == '\n')
1565 break;
1566 len++;
1567 }
1568 if (len >= table->maxlen)
1569 len = table->maxlen-1;
1570 if(copy_from_user(table->data, buffer, len))
1571 return -EFAULT;
1572 ((char *) table->data)[len] = 0;
1573 *ppos += *lenp;
1574 } else {
1575 len = strlen(table->data);
1576 if (len > table->maxlen)
1577 len = table->maxlen;
1578 if (len > *lenp)
1579 len = *lenp;
1580 if (len)
1581 if(copy_to_user(buffer, table->data, len))
1582 return -EFAULT;
1583 if (len < *lenp) {
1584 if(put_user('\n', ((char __user *) buffer) + len))
1585 return -EFAULT;
1586 len++;
1587 }
1588 *lenp = len;
1589 *ppos += len;
1590 }
1591 return 0;
1592}
1593
1594/*
1595 * Special case of dostring for the UTS structure. This has locks
1596 * to observe. Should this be in kernel/sys.c ????
1597 */
1598
1599static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
1600 void __user *buffer, size_t *lenp, loff_t *ppos)
1601{
1602 int r;
1603
1604 if (!write) {
1605 down_read(&uts_sem);
1606 r=proc_dostring(table,0,filp,buffer,lenp, ppos);
1607 up_read(&uts_sem);
1608 } else {
1609 down_write(&uts_sem);
1610 r=proc_dostring(table,1,filp,buffer,lenp, ppos);
1611 up_write(&uts_sem);
1612 }
1613 return r;
1614}
1615
1616static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1617 int *valp,
1618 int write, void *data)
1619{
1620 if (write) {
1621 *valp = *negp ? -*lvalp : *lvalp;
1622 } else {
1623 int val = *valp;
1624 if (val < 0) {
1625 *negp = -1;
1626 *lvalp = (unsigned long)-val;
1627 } else {
1628 *negp = 0;
1629 *lvalp = (unsigned long)val;
1630 }
1631 }
1632 return 0;
1633}
1634
1635static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1636 void __user *buffer, size_t *lenp, loff_t *ppos,
1637 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1638 int write, void *data),
1639 void *data)
1640{
1641#define TMPBUFLEN 21
1642 int *i, vleft, first=1, neg, val;
1643 unsigned long lval;
1644 size_t left, len;
1645
1646 char buf[TMPBUFLEN], *p;
1647 char __user *s = buffer;
1648
1649 if (!table->data || !table->maxlen || !*lenp ||
1650 (*ppos && !write)) {
1651 *lenp = 0;
1652 return 0;
1653 }
1654
1655 i = (int *) table->data;
1656 vleft = table->maxlen / sizeof(*i);
1657 left = *lenp;
1658
1659 if (!conv)
1660 conv = do_proc_dointvec_conv;
1661
1662 for (; left && vleft--; i++, first=0) {
1663 if (write) {
1664 while (left) {
1665 char c;
1666 if (get_user(c, s))
1667 return -EFAULT;
1668 if (!isspace(c))
1669 break;
1670 left--;
1671 s++;
1672 }
1673 if (!left)
1674 break;
1675 neg = 0;
1676 len = left;
1677 if (len > sizeof(buf) - 1)
1678 len = sizeof(buf) - 1;
1679 if (copy_from_user(buf, s, len))
1680 return -EFAULT;
1681 buf[len] = 0;
1682 p = buf;
1683 if (*p == '-' && left > 1) {
1684 neg = 1;
1685 left--, p++;
1686 }
1687 if (*p < '0' || *p > '9')
1688 break;
1689
1690 lval = simple_strtoul(p, &p, 0);
1691
1692 len = p-buf;
1693 if ((len < left) && *p && !isspace(*p))
1694 break;
1695 if (neg)
1696 val = -val;
1697 s += len;
1698 left -= len;
1699
1700 if (conv(&neg, &lval, i, 1, data))
1701 break;
1702 } else {
1703 p = buf;
1704 if (!first)
1705 *p++ = '\t';
1706
1707 if (conv(&neg, &lval, i, 0, data))
1708 break;
1709
1710 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1711 len = strlen(buf);
1712 if (len > left)
1713 len = left;
1714 if(copy_to_user(s, buf, len))
1715 return -EFAULT;
1716 left -= len;
1717 s += len;
1718 }
1719 }
1720
1721 if (!write && !first && left) {
1722 if(put_user('\n', s))
1723 return -EFAULT;
1724 left--, s++;
1725 }
1726 if (write) {
1727 while (left) {
1728 char c;
1729 if (get_user(c, s++))
1730 return -EFAULT;
1731 if (!isspace(c))
1732 break;
1733 left--;
1734 }
1735 }
1736 if (write && first)
1737 return -EINVAL;
1738 *lenp -= left;
1739 *ppos += *lenp;
1740 return 0;
1741#undef TMPBUFLEN
1742}
1743
1744/**
1745 * proc_dointvec - read a vector of integers
1746 * @table: the sysctl table
1747 * @write: %TRUE if this is a write to the sysctl file
1748 * @filp: the file structure
1749 * @buffer: the user buffer
1750 * @lenp: the size of the user buffer
1751 * @ppos: file position
1752 *
1753 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1754 * values from/to the user buffer, treated as an ASCII string.
1755 *
1756 * Returns 0 on success.
1757 */
1758int proc_dointvec(ctl_table *table, int write, struct file *filp,
1759 void __user *buffer, size_t *lenp, loff_t *ppos)
1760{
1761 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1762 NULL,NULL);
1763}
1764
1765#define OP_SET 0
1766#define OP_AND 1
1767#define OP_OR 2
1768#define OP_MAX 3
1769#define OP_MIN 4
1770
1771static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1772 int *valp,
1773 int write, void *data)
1774{
1775 int op = *(int *)data;
1776 if (write) {
1777 int val = *negp ? -*lvalp : *lvalp;
1778 switch(op) {
1779 case OP_SET: *valp = val; break;
1780 case OP_AND: *valp &= val; break;
1781 case OP_OR: *valp |= val; break;
1782 case OP_MAX: if(*valp < val)
1783 *valp = val;
1784 break;
1785 case OP_MIN: if(*valp > val)
1786 *valp = val;
1787 break;
1788 }
1789 } else {
1790 int val = *valp;
1791 if (val < 0) {
1792 *negp = -1;
1793 *lvalp = (unsigned long)-val;
1794 } else {
1795 *negp = 0;
1796 *lvalp = (unsigned long)val;
1797 }
1798 }
1799 return 0;
1800}
1801
1802/*
1803 * init may raise the set.
1804 */
1805
1806int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1807 void __user *buffer, size_t *lenp, loff_t *ppos)
1808{
1809 int op;
1810
1811 if (!capable(CAP_SYS_MODULE)) {
1812 return -EPERM;
1813 }
1814
1815 op = (current->pid == 1) ? OP_SET : OP_AND;
1816 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1817 do_proc_dointvec_bset_conv,&op);
1818}
1819
1820struct do_proc_dointvec_minmax_conv_param {
1821 int *min;
1822 int *max;
1823};
1824
1825static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1826 int *valp,
1827 int write, void *data)
1828{
1829 struct do_proc_dointvec_minmax_conv_param *param = data;
1830 if (write) {
1831 int val = *negp ? -*lvalp : *lvalp;
1832 if ((param->min && *param->min > val) ||
1833 (param->max && *param->max < val))
1834 return -EINVAL;
1835 *valp = val;
1836 } else {
1837 int val = *valp;
1838 if (val < 0) {
1839 *negp = -1;
1840 *lvalp = (unsigned long)-val;
1841 } else {
1842 *negp = 0;
1843 *lvalp = (unsigned long)val;
1844 }
1845 }
1846 return 0;
1847}
1848
1849/**
1850 * proc_dointvec_minmax - read a vector of integers with min/max values
1851 * @table: the sysctl table
1852 * @write: %TRUE if this is a write to the sysctl file
1853 * @filp: the file structure
1854 * @buffer: the user buffer
1855 * @lenp: the size of the user buffer
1856 * @ppos: file position
1857 *
1858 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1859 * values from/to the user buffer, treated as an ASCII string.
1860 *
1861 * This routine will ensure the values are within the range specified by
1862 * table->extra1 (min) and table->extra2 (max).
1863 *
1864 * Returns 0 on success.
1865 */
1866int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1867 void __user *buffer, size_t *lenp, loff_t *ppos)
1868{
1869 struct do_proc_dointvec_minmax_conv_param param = {
1870 .min = (int *) table->extra1,
1871 .max = (int *) table->extra2,
1872 };
1873 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1874 do_proc_dointvec_minmax_conv, &param);
1875}
1876
1877static int do_proc_doulongvec_minmax(ctl_table *table, int write,
1878 struct file *filp,
1879 void __user *buffer,
1880 size_t *lenp, loff_t *ppos,
1881 unsigned long convmul,
1882 unsigned long convdiv)
1883{
1884#define TMPBUFLEN 21
1885 unsigned long *i, *min, *max, val;
1886 int vleft, first=1, neg;
1887 size_t len, left;
1888 char buf[TMPBUFLEN], *p;
1889 char __user *s = buffer;
1890
1891 if (!table->data || !table->maxlen || !*lenp ||
1892 (*ppos && !write)) {
1893 *lenp = 0;
1894 return 0;
1895 }
1896
1897 i = (unsigned long *) table->data;
1898 min = (unsigned long *) table->extra1;
1899 max = (unsigned long *) table->extra2;
1900 vleft = table->maxlen / sizeof(unsigned long);
1901 left = *lenp;
1902
1903 for (; left && vleft--; i++, min++, max++, first=0) {
1904 if (write) {
1905 while (left) {
1906 char c;
1907 if (get_user(c, s))
1908 return -EFAULT;
1909 if (!isspace(c))
1910 break;
1911 left--;
1912 s++;
1913 }
1914 if (!left)
1915 break;
1916 neg = 0;
1917 len = left;
1918 if (len > TMPBUFLEN-1)
1919 len = TMPBUFLEN-1;
1920 if (copy_from_user(buf, s, len))
1921 return -EFAULT;
1922 buf[len] = 0;
1923 p = buf;
1924 if (*p == '-' && left > 1) {
1925 neg = 1;
1926 left--, p++;
1927 }
1928 if (*p < '0' || *p > '9')
1929 break;
1930 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
1931 len = p-buf;
1932 if ((len < left) && *p && !isspace(*p))
1933 break;
1934 if (neg)
1935 val = -val;
1936 s += len;
1937 left -= len;
1938
1939 if(neg)
1940 continue;
1941 if ((min && val < *min) || (max && val > *max))
1942 continue;
1943 *i = val;
1944 } else {
1945 p = buf;
1946 if (!first)
1947 *p++ = '\t';
1948 sprintf(p, "%lu", convdiv * (*i) / convmul);
1949 len = strlen(buf);
1950 if (len > left)
1951 len = left;
1952 if(copy_to_user(s, buf, len))
1953 return -EFAULT;
1954 left -= len;
1955 s += len;
1956 }
1957 }
1958
1959 if (!write && !first && left) {
1960 if(put_user('\n', s))
1961 return -EFAULT;
1962 left--, s++;
1963 }
1964 if (write) {
1965 while (left) {
1966 char c;
1967 if (get_user(c, s++))
1968 return -EFAULT;
1969 if (!isspace(c))
1970 break;
1971 left--;
1972 }
1973 }
1974 if (write && first)
1975 return -EINVAL;
1976 *lenp -= left;
1977 *ppos += *lenp;
1978 return 0;
1979#undef TMPBUFLEN
1980}
1981
1982/**
1983 * proc_doulongvec_minmax - read a vector of long integers with min/max values
1984 * @table: the sysctl table
1985 * @write: %TRUE if this is a write to the sysctl file
1986 * @filp: the file structure
1987 * @buffer: the user buffer
1988 * @lenp: the size of the user buffer
1989 * @ppos: file position
1990 *
1991 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1992 * values from/to the user buffer, treated as an ASCII string.
1993 *
1994 * This routine will ensure the values are within the range specified by
1995 * table->extra1 (min) and table->extra2 (max).
1996 *
1997 * Returns 0 on success.
1998 */
1999int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2000 void __user *buffer, size_t *lenp, loff_t *ppos)
2001{
2002 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2003}
2004
2005/**
2006 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2007 * @table: the sysctl table
2008 * @write: %TRUE if this is a write to the sysctl file
2009 * @filp: the file structure
2010 * @buffer: the user buffer
2011 * @lenp: the size of the user buffer
2012 * @ppos: file position
2013 *
2014 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2015 * values from/to the user buffer, treated as an ASCII string. The values
2016 * are treated as milliseconds, and converted to jiffies when they are stored.
2017 *
2018 * This routine will ensure the values are within the range specified by
2019 * table->extra1 (min) and table->extra2 (max).
2020 *
2021 * Returns 0 on success.
2022 */
2023int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2024 struct file *filp,
2025 void __user *buffer,
2026 size_t *lenp, loff_t *ppos)
2027{
2028 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2029 lenp, ppos, HZ, 1000l);
2030}
2031
2032
2033static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2034 int *valp,
2035 int write, void *data)
2036{
2037 if (write) {
2038 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2039 } else {
2040 int val = *valp;
2041 unsigned long lval;
2042 if (val < 0) {
2043 *negp = -1;
2044 lval = (unsigned long)-val;
2045 } else {
2046 *negp = 0;
2047 lval = (unsigned long)val;
2048 }
2049 *lvalp = lval / HZ;
2050 }
2051 return 0;
2052}
2053
2054static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2055 int *valp,
2056 int write, void *data)
2057{
2058 if (write) {
2059 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2060 } else {
2061 int val = *valp;
2062 unsigned long lval;
2063 if (val < 0) {
2064 *negp = -1;
2065 lval = (unsigned long)-val;
2066 } else {
2067 *negp = 0;
2068 lval = (unsigned long)val;
2069 }
2070 *lvalp = jiffies_to_clock_t(lval);
2071 }
2072 return 0;
2073}
2074
2075static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2076 int *valp,
2077 int write, void *data)
2078{
2079 if (write) {
2080 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2081 } else {
2082 int val = *valp;
2083 unsigned long lval;
2084 if (val < 0) {
2085 *negp = -1;
2086 lval = (unsigned long)-val;
2087 } else {
2088 *negp = 0;
2089 lval = (unsigned long)val;
2090 }
2091 *lvalp = jiffies_to_msecs(lval);
2092 }
2093 return 0;
2094}
2095
2096/**
2097 * proc_dointvec_jiffies - read a vector of integers as seconds
2098 * @table: the sysctl table
2099 * @write: %TRUE if this is a write to the sysctl file
2100 * @filp: the file structure
2101 * @buffer: the user buffer
2102 * @lenp: the size of the user buffer
2103 * @ppos: file position
2104 *
2105 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2106 * values from/to the user buffer, treated as an ASCII string.
2107 * The values read are assumed to be in seconds, and are converted into
2108 * jiffies.
2109 *
2110 * Returns 0 on success.
2111 */
2112int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2113 void __user *buffer, size_t *lenp, loff_t *ppos)
2114{
2115 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2116 do_proc_dointvec_jiffies_conv,NULL);
2117}
2118
2119/**
2120 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2121 * @table: the sysctl table
2122 * @write: %TRUE if this is a write to the sysctl file
2123 * @filp: the file structure
2124 * @buffer: the user buffer
2125 * @lenp: the size of the user buffer
1e5d5331 2126 * @ppos: pointer to the file position
1da177e4
LT		 2127 *
2128 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2129 * values from/to the user buffer, treated as an ASCII string.
2130 * The values read are assumed to be in 1/USER_HZ seconds, and
2131 * are converted into jiffies.
2132 *
2133 * Returns 0 on success.
2134 */
2135int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2136 void __user *buffer, size_t *lenp, loff_t *ppos)
2137{
2138 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2139 do_proc_dointvec_userhz_jiffies_conv,NULL);
2140}
2141
2142/**
2143 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2144 * @table: the sysctl table
2145 * @write: %TRUE if this is a write to the sysctl file
2146 * @filp: the file structure
2147 * @buffer: the user buffer
2148 * @lenp: the size of the user buffer
67be2dd1
MW		 2149 * @ppos: file position
2150 * @ppos: the current position in the file
1da177e4
LT		 2151 *
2152 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2153 * values from/to the user buffer, treated as an ASCII string.
2154 * The values read are assumed to be in 1/1000 seconds, and
2155 * are converted into jiffies.
2156 *
2157 * Returns 0 on success.
2158 */
2159int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2160 void __user *buffer, size_t *lenp, loff_t *ppos)
2161{
2162 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2163 do_proc_dointvec_ms_jiffies_conv, NULL);
2164}
2165
2166#else /* CONFIG_PROC_FS */
2167
2168int proc_dostring(ctl_table *table, int write, struct file *filp,
2169 void __user *buffer, size_t *lenp, loff_t *ppos)
2170{
2171 return -ENOSYS;
2172}
2173
2174static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
2175 void __user *buffer, size_t *lenp, loff_t *ppos)
2176{
2177 return -ENOSYS;
2178}
2179
2180int proc_dointvec(ctl_table *table, int write, struct file *filp,
2181 void __user *buffer, size_t *lenp, loff_t *ppos)
2182{
2183 return -ENOSYS;
2184}
2185
2186int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2187 void __user *buffer, size_t *lenp, loff_t *ppos)
2188{
2189 return -ENOSYS;
2190}
2191
2192int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2193 void __user *buffer, size_t *lenp, loff_t *ppos)
2194{
2195 return -ENOSYS;
2196}
2197
2198int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2199 void __user *buffer, size_t *lenp, loff_t *ppos)
2200{
2201 return -ENOSYS;
2202}
2203
2204int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2205 void __user *buffer, size_t *lenp, loff_t *ppos)
2206{
2207 return -ENOSYS;
2208}
2209
2210int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2211 void __user *buffer, size_t *lenp, loff_t *ppos)
2212{
2213 return -ENOSYS;
2214}
2215
2216int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2217 void __user *buffer, size_t *lenp, loff_t *ppos)
2218{
2219 return -ENOSYS;
2220}
2221
2222int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2223 struct file *filp,
2224 void __user *buffer,
2225 size_t *lenp, loff_t *ppos)
2226{
2227 return -ENOSYS;
2228}
2229
2230
2231#endif /* CONFIG_PROC_FS */
2232
2233
2234/*
2235 * General sysctl support routines
2236 */
2237
2238/* The generic string strategy routine: */
2239int sysctl_string(ctl_table *table, int __user *name, int nlen,
2240 void __user *oldval, size_t __user *oldlenp,
2241 void __user *newval, size_t newlen, void **context)
2242{
1da177e4
LT		 2243 if (!table->data || !table->maxlen)
2244 return -ENOTDIR;
2245
2246 if (oldval && oldlenp) {
de9e007d
LT		 2247 size_t bufsize;
2248 if (get_user(bufsize, oldlenp))
1da177e4 2249 return -EFAULT;
de9e007d
LT		 2250 if (bufsize) {
2251 size_t len = strlen(table->data), copied;
2252
2253 /* This shouldn't trigger for a well-formed sysctl */
2254 if (len > table->maxlen)
1da177e4 2255 len = table->maxlen;
de9e007d
LT		 2256
2257 /* Copy up to a max of bufsize-1 bytes of the string */
2258 copied = (len >= bufsize) ? bufsize - 1 : len;
2259
2260 if (copy_to_user(oldval, table->data, copied) ||
2261 put_user(0, (char __user *)(oldval + copied)))
1da177e4 2262 return -EFAULT;
de9e007d 2263 if (put_user(len, oldlenp))
1da177e4
LT		 2264 return -EFAULT;
2265 }
2266 }
2267 if (newval && newlen) {
de9e007d 2268 size_t len = newlen;
1da177e4
LT		 2269 if (len > table->maxlen)
2270 len = table->maxlen;
2271 if(copy_from_user(table->data, newval, len))
2272 return -EFAULT;
2273 if (len == table->maxlen)
2274 len--;
2275 ((char *) table->data)[len] = 0;
2276 }
82c9df82 2277 return 1;
1da177e4
LT		 2278}
2279
2280/*
2281 * This function makes sure that all of the integers in the vector
2282 * are between the minimum and maximum values given in the arrays
2283 * table->extra1 and table->extra2, respectively.
2284 */
2285int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2286 void __user *oldval, size_t __user *oldlenp,
2287 void __user *newval, size_t newlen, void **context)
2288{
2289
2290 if (newval && newlen) {
2291 int __user *vec = (int __user *) newval;
2292 int *min = (int *) table->extra1;
2293 int *max = (int *) table->extra2;
2294 size_t length;
2295 int i;
2296
2297 if (newlen % sizeof(int) != 0)
2298 return -EINVAL;
2299
2300 if (!table->extra1 && !table->extra2)
2301 return 0;
2302
2303 if (newlen > table->maxlen)
2304 newlen = table->maxlen;
2305 length = newlen / sizeof(int);
2306
2307 for (i = 0; i < length; i++) {
2308 int value;
2309 if (get_user(value, vec + i))
2310 return -EFAULT;
2311 if (min && value < min[i])
2312 return -EINVAL;
2313 if (max && value > max[i])
2314 return -EINVAL;
2315 }
2316 }
2317 return 0;
2318}
2319
2320/* Strategy function to convert jiffies to seconds */
2321int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2322 void __user *oldval, size_t __user *oldlenp,
2323 void __user *newval, size_t newlen, void **context)
2324{
2325 if (oldval) {
2326 size_t olen;
2327 if (oldlenp) {
2328 if (get_user(olen, oldlenp))
2329 return -EFAULT;
2330 if (olen!=sizeof(int))
2331 return -EINVAL;
2332 }
2333 if (put_user(*(int *)(table->data)/HZ, (int __user *)oldval) ||
2334 (oldlenp && put_user(sizeof(int),oldlenp)))
2335 return -EFAULT;
2336 }
2337 if (newval && newlen) {
2338 int new;
2339 if (newlen != sizeof(int))
2340 return -EINVAL;
2341 if (get_user(new, (int __user *)newval))
2342 return -EFAULT;
2343 *(int *)(table->data) = new*HZ;
2344 }
2345 return 1;
2346}
2347
2348/* Strategy function to convert jiffies to seconds */
2349int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2350 void __user *oldval, size_t __user *oldlenp,
2351 void __user *newval, size_t newlen, void **context)
2352{
2353 if (oldval) {
2354 size_t olen;
2355 if (oldlenp) {
2356 if (get_user(olen, oldlenp))
2357 return -EFAULT;
2358 if (olen!=sizeof(int))
2359 return -EINVAL;
2360 }
2361 if (put_user(jiffies_to_msecs(*(int *)(table->data)), (int __user *)oldval) ||
2362 (oldlenp && put_user(sizeof(int),oldlenp)))
2363 return -EFAULT;
2364 }
2365 if (newval && newlen) {
2366 int new;
2367 if (newlen != sizeof(int))
2368 return -EINVAL;
2369 if (get_user(new, (int __user *)newval))
2370 return -EFAULT;
2371 *(int *)(table->data) = msecs_to_jiffies(new);
2372 }
2373 return 1;
2374}
2375
2376#else /* CONFIG_SYSCTL */
2377
2378
2379asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2380{
2381 return -ENOSYS;
2382}
2383
2384int sysctl_string(ctl_table *table, int __user *name, int nlen,
2385 void __user *oldval, size_t __user *oldlenp,
2386 void __user *newval, size_t newlen, void **context)
2387{
2388 return -ENOSYS;
2389}
2390
2391int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2392 void __user *oldval, size_t __user *oldlenp,
2393 void __user *newval, size_t newlen, void **context)
2394{
2395 return -ENOSYS;
2396}
2397
2398int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2399 void __user *oldval, size_t __user *oldlenp,
2400 void __user *newval, size_t newlen, void **context)
2401{
2402 return -ENOSYS;
2403}
2404
2405int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2406 void __user *oldval, size_t __user *oldlenp,
2407 void __user *newval, size_t newlen, void **context)
2408{
2409 return -ENOSYS;
2410}
2411
2412int proc_dostring(ctl_table *table, int write, struct file *filp,
2413 void __user *buffer, size_t *lenp, loff_t *ppos)
2414{
2415 return -ENOSYS;
2416}
2417
2418int proc_dointvec(ctl_table *table, int write, struct file *filp,
2419 void __user *buffer, size_t *lenp, loff_t *ppos)
2420{
2421 return -ENOSYS;
2422}
2423
2424int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2425 void __user *buffer, size_t *lenp, loff_t *ppos)
2426{
2427 return -ENOSYS;
2428}
2429
2430int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2431 void __user *buffer, size_t *lenp, loff_t *ppos)
2432{
2433 return -ENOSYS;
2434}
2435
2436int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2437 void __user *buffer, size_t *lenp, loff_t *ppos)
2438{
2439 return -ENOSYS;
2440}
2441
2442int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2443 void __user *buffer, size_t *lenp, loff_t *ppos)
2444{
2445 return -ENOSYS;
2446}
2447
2448int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2449 void __user *buffer, size_t *lenp, loff_t *ppos)
2450{
2451 return -ENOSYS;
2452}
2453
2454int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2455 void __user *buffer, size_t *lenp, loff_t *ppos)
2456{
2457 return -ENOSYS;
2458}
2459
2460int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2461 struct file *filp,
2462 void __user *buffer,
2463 size_t *lenp, loff_t *ppos)
2464{
2465 return -ENOSYS;
2466}
2467
2468struct ctl_table_header * register_sysctl_table(ctl_table * table,
2469 int insert_at_head)
2470{
2471 return NULL;
2472}
2473
2474void unregister_sysctl_table(struct ctl_table_header * table)
2475{
2476}
2477
2478#endif /* CONFIG_SYSCTL */
2479
2480/*
2481 * No sense putting this after each symbol definition, twice,
2482 * exception granted :-)
2483 */
2484EXPORT_SYMBOL(proc_dointvec);
2485EXPORT_SYMBOL(proc_dointvec_jiffies);
2486EXPORT_SYMBOL(proc_dointvec_minmax);
2487EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2488EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2489EXPORT_SYMBOL(proc_dostring);
2490EXPORT_SYMBOL(proc_doulongvec_minmax);
2491EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2492EXPORT_SYMBOL(register_sysctl_table);
2493EXPORT_SYMBOL(sysctl_intvec);
2494EXPORT_SYMBOL(sysctl_jiffies);
2495EXPORT_SYMBOL(sysctl_ms_jiffies);
2496EXPORT_SYMBOL(sysctl_string);
2497EXPORT_SYMBOL(unregister_sysctl_table);
Linux 6.x block layer and io_uring tree(s)