?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
ACPI: handle battery notify event on broken BIOS
[linux-2.6/linux-acpi-2.6/ibm-acpi-2.6.git] / kernel / sysctl.c
blob99a58f2790777481f646c8f2af0eced8a9b5e0f2
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/module.h>
22 #include <linux/mm.h>
23 #include <linux/swap.h>
24 #include <linux/slab.h>
25 #include <linux/sysctl.h>
26 #include <linux/proc_fs.h>
27 #include <linux/capability.h>
28 #include <linux/ctype.h>
29 #include <linux/utsname.h>
30 #include <linux/capability.h>
31 #include <linux/smp_lock.h>
32 #include <linux/init.h>
33 #include <linux/kernel.h>
34 #include <linux/kobject.h>
35 #include <linux/net.h>
36 #include <linux/sysrq.h>
37 #include <linux/highuid.h>
38 #include <linux/writeback.h>
39 #include <linux/hugetlb.h>
40 #include <linux/security.h>
41 #include <linux/initrd.h>
42 #include <linux/times.h>
43 #include <linux/limits.h>
44 #include <linux/dcache.h>
45 #include <linux/syscalls.h>
46 #include <linux/nfs_fs.h>
47 #include <linux/acpi.h>
48
49 #include <asm/uaccess.h>
50 #include <asm/processor.h>
51
52 extern int proc_nr_files(ctl_table *table, int write, struct file *filp,
53 void __user *buffer, size_t *lenp, loff_t *ppos);
54
55 #if defined(CONFIG_SYSCTL)
56
57 /* External variables not in a header file. */
58 extern int C_A_D;
59 extern int sysctl_overcommit_memory;
60 extern int sysctl_overcommit_ratio;
61 extern int sysctl_panic_on_oom;
62 extern int max_threads;
63 extern int sysrq_enabled;
64 extern int core_uses_pid;
65 extern int suid_dumpable;
66 extern char core_pattern[];
67 extern int cad_pid;
68 extern int pid_max;
69 extern int min_free_kbytes;
70 extern int printk_ratelimit_jiffies;
71 extern int printk_ratelimit_burst;
72 extern int pid_max_min, pid_max_max;
73 extern int sysctl_drop_caches;
74 extern int percpu_pagelist_fraction;
75 extern int compat_log;
76
77 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
78 int unknown_nmi_panic;
79 extern int proc_unknown_nmi_panic(ctl_table *, int, struct file *,
80 void __user *, size_t *, loff_t *);
81 #endif
82
83 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
84 static int maxolduid = 65535;
85 static int minolduid;
86 static int min_percpu_pagelist_fract = 8;
87
88 static int ngroups_max = NGROUPS_MAX;
89
90 #ifdef CONFIG_KMOD
91 extern char modprobe_path[];
92 #endif
93 #ifdef CONFIG_CHR_DEV_SG
94 extern int sg_big_buff;
95 #endif
96 #ifdef CONFIG_SYSVIPC
97 extern size_t shm_ctlmax;
98 extern size_t shm_ctlall;
99 extern int shm_ctlmni;
100 extern int msg_ctlmax;
101 extern int msg_ctlmnb;
102 extern int msg_ctlmni;
103 extern int sem_ctls[];
104 #endif
105
106 #ifdef __sparc__
107 extern char reboot_command [];
108 extern int stop_a_enabled;
109 extern int scons_pwroff;
110 #endif
111
112 #ifdef __hppa__
113 extern int pwrsw_enabled;
114 extern int unaligned_enabled;
115 #endif
116
117 #ifdef CONFIG_S390
118 #ifdef CONFIG_MATHEMU
119 extern int sysctl_ieee_emulation_warnings;
120 #endif
121 extern int sysctl_userprocess_debug;
122 extern int spin_retry;
123 #endif
124
125 extern int sysctl_hz_timer;
126
127 #ifdef CONFIG_BSD_PROCESS_ACCT
128 extern int acct_parm[];
129 #endif
130
131 #ifdef CONFIG_IA64
132 extern int no_unaligned_warning;
133 #endif
134
135 #ifdef CONFIG_RT_MUTEXES
136 extern int max_lock_depth;
137 #endif
138
139 static int parse_table(int __user *, int, void __user *, size_t __user *, void __user *, size_t,
140 ctl_table *, void **);
141 static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
142 void __user *buffer, size_t *lenp, loff_t *ppos);
143
144 static ctl_table root_table[];
145 static struct ctl_table_header root_table_header =
146 { root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
147
148 static ctl_table kern_table[];
149 static ctl_table vm_table[];
150 static ctl_table fs_table[];
151 static ctl_table debug_table[];
152 static ctl_table dev_table[];
153 extern ctl_table random_table[];
154 #ifdef CONFIG_UNIX98_PTYS
155 extern ctl_table pty_table[];
156 #endif
157 #ifdef CONFIG_INOTIFY_USER
158 extern ctl_table inotify_table[];
159 #endif
160
161 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
162 int sysctl_legacy_va_layout;
163 #endif
164
165 /* /proc declarations: */
166
167 #ifdef CONFIG_PROC_FS
168
169 static ssize_t proc_readsys(struct file *, char __user *, size_t, loff_t *);
170 static ssize_t proc_writesys(struct file *, const char __user *, size_t, loff_t *);
171 static int proc_opensys(struct inode *, struct file *);
172
173 struct file_operations proc_sys_file_operations = {
174 .open = proc_opensys,
175 .read = proc_readsys,
176 .write = proc_writesys,
177 };
178
179 extern struct proc_dir_entry *proc_sys_root;
180
181 static void register_proc_table(ctl_table *, struct proc_dir_entry *, void *);
182 static void unregister_proc_table(ctl_table *, struct proc_dir_entry *);
183 #endif
184
185 /* The default sysctl tables: */
186
187 static ctl_table root_table[] = {
188 {
189 .ctl_name = CTL_KERN,
190 .procname = "kernel",
191 .mode = 0555,
192 .child = kern_table,
193 },
194 {
195 .ctl_name = CTL_VM,
196 .procname = "vm",
197 .mode = 0555,
198 .child = vm_table,
199 },
200 #ifdef CONFIG_NET
201 {
202 .ctl_name = CTL_NET,
203 .procname = "net",
204 .mode = 0555,
205 .child = net_table,
206 },
207 #endif
208 {
209 .ctl_name = CTL_FS,
210 .procname = "fs",
211 .mode = 0555,
212 .child = fs_table,
213 },
214 {
215 .ctl_name = CTL_DEBUG,
216 .procname = "debug",
217 .mode = 0555,
218 .child = debug_table,
219 },
220 {
221 .ctl_name = CTL_DEV,
222 .procname = "dev",
223 .mode = 0555,
224 .child = dev_table,
225 },
226
227 { .ctl_name = 0 }
228 };
229
230 static ctl_table kern_table[] = {
231 {
232 .ctl_name = KERN_OSTYPE,
233 .procname = "ostype",
234 .data = system_utsname.sysname,
235 .maxlen = sizeof(system_utsname.sysname),
236 .mode = 0444,
237 .proc_handler = &proc_doutsstring,
238 .strategy = &sysctl_string,
239 },
240 {
241 .ctl_name = KERN_OSRELEASE,
242 .procname = "osrelease",
243 .data = system_utsname.release,
244 .maxlen = sizeof(system_utsname.release),
245 .mode = 0444,
246 .proc_handler = &proc_doutsstring,
247 .strategy = &sysctl_string,
248 },
249 {
250 .ctl_name = KERN_VERSION,
251 .procname = "version",
252 .data = system_utsname.version,
253 .maxlen = sizeof(system_utsname.version),
254 .mode = 0444,
255 .proc_handler = &proc_doutsstring,
256 .strategy = &sysctl_string,
257 },
258 {
259 .ctl_name = KERN_NODENAME,
260 .procname = "hostname",
261 .data = system_utsname.nodename,
262 .maxlen = sizeof(system_utsname.nodename),
263 .mode = 0644,
264 .proc_handler = &proc_doutsstring,
265 .strategy = &sysctl_string,
266 },
267 {
268 .ctl_name = KERN_DOMAINNAME,
269 .procname = "domainname",
270 .data = system_utsname.domainname,
271 .maxlen = sizeof(system_utsname.domainname),
272 .mode = 0644,
273 .proc_handler = &proc_doutsstring,
274 .strategy = &sysctl_string,
275 },
276 {
277 .ctl_name = KERN_PANIC,
278 .procname = "panic",
279 .data = &panic_timeout,
280 .maxlen = sizeof(int),
281 .mode = 0644,
282 .proc_handler = &proc_dointvec,
283 },
284 {
285 .ctl_name = KERN_CORE_USES_PID,
286 .procname = "core_uses_pid",
287 .data = &core_uses_pid,
288 .maxlen = sizeof(int),
289 .mode = 0644,
290 .proc_handler = &proc_dointvec,
291 },
292 {
293 .ctl_name = KERN_CORE_PATTERN,
294 .procname = "core_pattern",
295 .data = core_pattern,
296 .maxlen = 64,
297 .mode = 0644,
298 .proc_handler = &proc_dostring,
299 .strategy = &sysctl_string,
300 },
301 {
302 .ctl_name = KERN_TAINTED,
303 .procname = "tainted",
304 .data = &tainted,
305 .maxlen = sizeof(int),
306 .mode = 0444,
307 .proc_handler = &proc_dointvec,
308 },
309 {
310 .ctl_name = KERN_CAP_BSET,
311 .procname = "cap-bound",
312 .data = &cap_bset,
313 .maxlen = sizeof(kernel_cap_t),
314 .mode = 0600,
315 .proc_handler = &proc_dointvec_bset,
316 },
317 #ifdef CONFIG_BLK_DEV_INITRD
318 {
319 .ctl_name = KERN_REALROOTDEV,
320 .procname = "real-root-dev",
321 .data = &real_root_dev,
322 .maxlen = sizeof(int),
323 .mode = 0644,
324 .proc_handler = &proc_dointvec,
325 },
326 #endif
327 #ifdef __sparc__
328 {
329 .ctl_name = KERN_SPARC_REBOOT,
330 .procname = "reboot-cmd",
331 .data = reboot_command,
332 .maxlen = 256,
333 .mode = 0644,
334 .proc_handler = &proc_dostring,
335 .strategy = &sysctl_string,
336 },
337 {
338 .ctl_name = KERN_SPARC_STOP_A,
339 .procname = "stop-a",
340 .data = &stop_a_enabled,
341 .maxlen = sizeof (int),
342 .mode = 0644,
343 .proc_handler = &proc_dointvec,
344 },
345 {
346 .ctl_name = KERN_SPARC_SCONS_PWROFF,
347 .procname = "scons-poweroff",
348 .data = &scons_pwroff,
349 .maxlen = sizeof (int),
350 .mode = 0644,
351 .proc_handler = &proc_dointvec,
352 },
353 #endif
354 #ifdef __hppa__
355 {
356 .ctl_name = KERN_HPPA_PWRSW,
357 .procname = "soft-power",
358 .data = &pwrsw_enabled,
359 .maxlen = sizeof (int),
360 .mode = 0644,
361 .proc_handler = &proc_dointvec,
362 },
363 {
364 .ctl_name = KERN_HPPA_UNALIGNED,
365 .procname = "unaligned-trap",
366 .data = &unaligned_enabled,
367 .maxlen = sizeof (int),
368 .mode = 0644,
369 .proc_handler = &proc_dointvec,
370 },
371 #endif
372 {
373 .ctl_name = KERN_CTLALTDEL,
374 .procname = "ctrl-alt-del",
375 .data = &C_A_D,
376 .maxlen = sizeof(int),
377 .mode = 0644,
378 .proc_handler = &proc_dointvec,
379 },
380 {
381 .ctl_name = KERN_PRINTK,
382 .procname = "printk",
383 .data = &console_loglevel,
384 .maxlen = 4*sizeof(int),
385 .mode = 0644,
386 .proc_handler = &proc_dointvec,
387 },
388 #ifdef CONFIG_KMOD
389 {
390 .ctl_name = KERN_MODPROBE,
391 .procname = "modprobe",
392 .data = &modprobe_path,
393 .maxlen = KMOD_PATH_LEN,
394 .mode = 0644,
395 .proc_handler = &proc_dostring,
396 .strategy = &sysctl_string,
397 },
398 #endif
399 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
400 {
401 .ctl_name = KERN_HOTPLUG,
402 .procname = "hotplug",
403 .data = &uevent_helper,
404 .maxlen = UEVENT_HELPER_PATH_LEN,
405 .mode = 0644,
406 .proc_handler = &proc_dostring,
407 .strategy = &sysctl_string,
408 },
409 #endif
410 #ifdef CONFIG_CHR_DEV_SG
411 {
412 .ctl_name = KERN_SG_BIG_BUFF,
413 .procname = "sg-big-buff",
414 .data = &sg_big_buff,
415 .maxlen = sizeof (int),
416 .mode = 0444,
417 .proc_handler = &proc_dointvec,
418 },
419 #endif
420 #ifdef CONFIG_BSD_PROCESS_ACCT
421 {
422 .ctl_name = KERN_ACCT,
423 .procname = "acct",
424 .data = &acct_parm,
425 .maxlen = 3*sizeof(int),
426 .mode = 0644,
427 .proc_handler = &proc_dointvec,
428 },
429 #endif
430 #ifdef CONFIG_SYSVIPC
431 {
432 .ctl_name = KERN_SHMMAX,
433 .procname = "shmmax",
434 .data = &shm_ctlmax,
435 .maxlen = sizeof (size_t),
436 .mode = 0644,
437 .proc_handler = &proc_doulongvec_minmax,
438 },
439 {
440 .ctl_name = KERN_SHMALL,
441 .procname = "shmall",
442 .data = &shm_ctlall,
443 .maxlen = sizeof (size_t),
444 .mode = 0644,
445 .proc_handler = &proc_doulongvec_minmax,
446 },
447 {
448 .ctl_name = KERN_SHMMNI,
449 .procname = "shmmni",
450 .data = &shm_ctlmni,
451 .maxlen = sizeof (int),
452 .mode = 0644,
453 .proc_handler = &proc_dointvec,
454 },
455 {
456 .ctl_name = KERN_MSGMAX,
457 .procname = "msgmax",
458 .data = &msg_ctlmax,
459 .maxlen = sizeof (int),
460 .mode = 0644,
461 .proc_handler = &proc_dointvec,
462 },
463 {
464 .ctl_name = KERN_MSGMNI,
465 .procname = "msgmni",
466 .data = &msg_ctlmni,
467 .maxlen = sizeof (int),
468 .mode = 0644,
469 .proc_handler = &proc_dointvec,
470 },
471 {
472 .ctl_name = KERN_MSGMNB,
473 .procname = "msgmnb",
474 .data = &msg_ctlmnb,
475 .maxlen = sizeof (int),
476 .mode = 0644,
477 .proc_handler = &proc_dointvec,
478 },
479 {
480 .ctl_name = KERN_SEM,
481 .procname = "sem",
482 .data = &sem_ctls,
483 .maxlen = 4*sizeof (int),
484 .mode = 0644,
485 .proc_handler = &proc_dointvec,
486 },
487 #endif
488 #ifdef CONFIG_MAGIC_SYSRQ
489 {
490 .ctl_name = KERN_SYSRQ,
491 .procname = "sysrq",
492 .data = &sysrq_enabled,
493 .maxlen = sizeof (int),
494 .mode = 0644,
495 .proc_handler = &proc_dointvec,
496 },
497 #endif
498 {
499 .ctl_name = KERN_CADPID,
500 .procname = "cad_pid",
501 .data = &cad_pid,
502 .maxlen = sizeof (int),
503 .mode = 0600,
504 .proc_handler = &proc_dointvec,
505 },
506 {
507 .ctl_name = KERN_MAX_THREADS,
508 .procname = "threads-max",
509 .data = &max_threads,
510 .maxlen = sizeof(int),
511 .mode = 0644,
512 .proc_handler = &proc_dointvec,
513 },
514 {
515 .ctl_name = KERN_RANDOM,
516 .procname = "random",
517 .mode = 0555,
518 .child = random_table,
519 },
520 #ifdef CONFIG_UNIX98_PTYS
521 {
522 .ctl_name = KERN_PTY,
523 .procname = "pty",
524 .mode = 0555,
525 .child = pty_table,
526 },
527 #endif
528 {
529 .ctl_name = KERN_OVERFLOWUID,
530 .procname = "overflowuid",
531 .data = &overflowuid,
532 .maxlen = sizeof(int),
533 .mode = 0644,
534 .proc_handler = &proc_dointvec_minmax,
535 .strategy = &sysctl_intvec,
536 .extra1 = &minolduid,
537 .extra2 = &maxolduid,
538 },
539 {
540 .ctl_name = KERN_OVERFLOWGID,
541 .procname = "overflowgid",
542 .data = &overflowgid,
543 .maxlen = sizeof(int),
544 .mode = 0644,
545 .proc_handler = &proc_dointvec_minmax,
546 .strategy = &sysctl_intvec,
547 .extra1 = &minolduid,
548 .extra2 = &maxolduid,
549 },
550 #ifdef CONFIG_S390
551 #ifdef CONFIG_MATHEMU
552 {
553 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
554 .procname = "ieee_emulation_warnings",
555 .data = &sysctl_ieee_emulation_warnings,
556 .maxlen = sizeof(int),
557 .mode = 0644,
558 .proc_handler = &proc_dointvec,
559 },
560 #endif
561 #ifdef CONFIG_NO_IDLE_HZ
562 {
563 .ctl_name = KERN_HZ_TIMER,
564 .procname = "hz_timer",
565 .data = &sysctl_hz_timer,
566 .maxlen = sizeof(int),
567 .mode = 0644,
568 .proc_handler = &proc_dointvec,
569 },
570 #endif
571 {
572 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
573 .procname = "userprocess_debug",
574 .data = &sysctl_userprocess_debug,
575 .maxlen = sizeof(int),
576 .mode = 0644,
577 .proc_handler = &proc_dointvec,
578 },
579 #endif
580 {
581 .ctl_name = KERN_PIDMAX,
582 .procname = "pid_max",
583 .data = &pid_max,
584 .maxlen = sizeof (int),
585 .mode = 0644,
586 .proc_handler = &proc_dointvec_minmax,
587 .strategy = sysctl_intvec,
588 .extra1 = &pid_max_min,
589 .extra2 = &pid_max_max,
590 },
591 {
592 .ctl_name = KERN_PANIC_ON_OOPS,
593 .procname = "panic_on_oops",
594 .data = &panic_on_oops,
595 .maxlen = sizeof(int),
596 .mode = 0644,
597 .proc_handler = &proc_dointvec,
598 },
599 {
600 .ctl_name = KERN_PRINTK_RATELIMIT,
601 .procname = "printk_ratelimit",
602 .data = &printk_ratelimit_jiffies,
603 .maxlen = sizeof(int),
604 .mode = 0644,
605 .proc_handler = &proc_dointvec_jiffies,
606 .strategy = &sysctl_jiffies,
607 },
608 {
609 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
610 .procname = "printk_ratelimit_burst",
611 .data = &printk_ratelimit_burst,
612 .maxlen = sizeof(int),
613 .mode = 0644,
614 .proc_handler = &proc_dointvec,
615 },
616 {
617 .ctl_name = KERN_NGROUPS_MAX,
618 .procname = "ngroups_max",
619 .data = &ngroups_max,
620 .maxlen = sizeof (int),
621 .mode = 0444,
622 .proc_handler = &proc_dointvec,
623 },
624 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
625 {
626 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
627 .procname = "unknown_nmi_panic",
628 .data = &unknown_nmi_panic,
629 .maxlen = sizeof (int),
630 .mode = 0644,
631 .proc_handler = &proc_unknown_nmi_panic,
632 },
633 #endif
634 #if defined(CONFIG_X86)
635 {
636 .ctl_name = KERN_BOOTLOADER_TYPE,
637 .procname = "bootloader_type",
638 .data = &bootloader_type,
639 .maxlen = sizeof (int),
640 .mode = 0444,
641 .proc_handler = &proc_dointvec,
642 },
643 #endif
644 #if defined(CONFIG_MMU)
645 {
646 .ctl_name = KERN_RANDOMIZE,
647 .procname = "randomize_va_space",
648 .data = &randomize_va_space,
649 .maxlen = sizeof(int),
650 .mode = 0644,
651 .proc_handler = &proc_dointvec,
652 },
653 #endif
654 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
655 {
656 .ctl_name = KERN_SPIN_RETRY,
657 .procname = "spin_retry",
658 .data = &spin_retry,
659 .maxlen = sizeof (int),
660 .mode = 0644,
661 .proc_handler = &proc_dointvec,
662 },
663 #endif
664 #ifdef CONFIG_ACPI_SLEEP
665 {
666 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
667 .procname = "acpi_video_flags",
668 .data = &acpi_video_flags,
669 .maxlen = sizeof (unsigned long),
670 .mode = 0644,
671 .proc_handler = &proc_doulongvec_minmax,
672 },
673 #endif
674 #ifdef CONFIG_IA64
675 {
676 .ctl_name = KERN_IA64_UNALIGNED,
677 .procname = "ignore-unaligned-usertrap",
678 .data = &no_unaligned_warning,
679 .maxlen = sizeof (int),
680 .mode = 0644,
681 .proc_handler = &proc_dointvec,
682 },
683 #endif
684 #ifdef CONFIG_COMPAT
685 {
686 .ctl_name = KERN_COMPAT_LOG,
687 .procname = "compat-log",
688 .data = &compat_log,
689 .maxlen = sizeof (int),
690 .mode = 0644,
691 .proc_handler = &proc_dointvec,
692 },
693 #endif
694 #ifdef CONFIG_RT_MUTEXES
695 {
696 .ctl_name = KERN_MAX_LOCK_DEPTH,
697 .procname = "max_lock_depth",
698 .data = &max_lock_depth,
699 .maxlen = sizeof(int),
700 .mode = 0644,
701 .proc_handler = &proc_dointvec,
702 },
703 #endif
704
705 { .ctl_name = 0 }
706 };
707
708 /* Constants for minimum and maximum testing in vm_table.
709 We use these as one-element integer vectors. */
710 static int zero;
711 static int one_hundred = 100;
712
713
714 static ctl_table vm_table[] = {
715 {
716 .ctl_name = VM_OVERCOMMIT_MEMORY,
717 .procname = "overcommit_memory",
718 .data = &sysctl_overcommit_memory,
719 .maxlen = sizeof(sysctl_overcommit_memory),
720 .mode = 0644,
721 .proc_handler = &proc_dointvec,
722 },
723 {
724 .ctl_name = VM_PANIC_ON_OOM,
725 .procname = "panic_on_oom",
726 .data = &sysctl_panic_on_oom,
727 .maxlen = sizeof(sysctl_panic_on_oom),
728 .mode = 0644,
729 .proc_handler = &proc_dointvec,
730 },
731 {
732 .ctl_name = VM_OVERCOMMIT_RATIO,
733 .procname = "overcommit_ratio",
734 .data = &sysctl_overcommit_ratio,
735 .maxlen = sizeof(sysctl_overcommit_ratio),
736 .mode = 0644,
737 .proc_handler = &proc_dointvec,
738 },
739 {
740 .ctl_name = VM_PAGE_CLUSTER,
741 .procname = "page-cluster",
742 .data = &page_cluster,
743 .maxlen = sizeof(int),
744 .mode = 0644,
745 .proc_handler = &proc_dointvec,
746 },
747 {
748 .ctl_name = VM_DIRTY_BACKGROUND,
749 .procname = "dirty_background_ratio",
750 .data = &dirty_background_ratio,
751 .maxlen = sizeof(dirty_background_ratio),
752 .mode = 0644,
753 .proc_handler = &proc_dointvec_minmax,
754 .strategy = &sysctl_intvec,
755 .extra1 = &zero,
756 .extra2 = &one_hundred,
757 },
758 {
759 .ctl_name = VM_DIRTY_RATIO,
760 .procname = "dirty_ratio",
761 .data = &vm_dirty_ratio,
762 .maxlen = sizeof(vm_dirty_ratio),
763 .mode = 0644,
764 .proc_handler = &proc_dointvec_minmax,
765 .strategy = &sysctl_intvec,
766 .extra1 = &zero,
767 .extra2 = &one_hundred,
768 },
769 {
770 .ctl_name = VM_DIRTY_WB_CS,
771 .procname = "dirty_writeback_centisecs",
772 .data = &dirty_writeback_interval,
773 .maxlen = sizeof(dirty_writeback_interval),
774 .mode = 0644,
775 .proc_handler = &dirty_writeback_centisecs_handler,
776 },
777 {
778 .ctl_name = VM_DIRTY_EXPIRE_CS,
779 .procname = "dirty_expire_centisecs",
780 .data = &dirty_expire_interval,
781 .maxlen = sizeof(dirty_expire_interval),
782 .mode = 0644,
783 .proc_handler = &proc_dointvec_userhz_jiffies,
784 },
785 {
786 .ctl_name = VM_NR_PDFLUSH_THREADS,
787 .procname = "nr_pdflush_threads",
788 .data = &nr_pdflush_threads,
789 .maxlen = sizeof nr_pdflush_threads,
790 .mode = 0444 /* read-only*/,
791 .proc_handler = &proc_dointvec,
792 },
793 {
794 .ctl_name = VM_SWAPPINESS,
795 .procname = "swappiness",
796 .data = &vm_swappiness,
797 .maxlen = sizeof(vm_swappiness),
798 .mode = 0644,
799 .proc_handler = &proc_dointvec_minmax,
800 .strategy = &sysctl_intvec,
801 .extra1 = &zero,
802 .extra2 = &one_hundred,
803 },
804 #ifdef CONFIG_HUGETLB_PAGE
805 {
806 .ctl_name = VM_HUGETLB_PAGES,
807 .procname = "nr_hugepages",
808 .data = &max_huge_pages,
809 .maxlen = sizeof(unsigned long),
810 .mode = 0644,
811 .proc_handler = &hugetlb_sysctl_handler,
812 .extra1 = (void *)&hugetlb_zero,
813 .extra2 = (void *)&hugetlb_infinity,
814 },
815 {
816 .ctl_name = VM_HUGETLB_GROUP,
817 .procname = "hugetlb_shm_group",
818 .data = &sysctl_hugetlb_shm_group,
819 .maxlen = sizeof(gid_t),
820 .mode = 0644,
821 .proc_handler = &proc_dointvec,
822 },
823 #endif
824 {
825 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
826 .procname = "lowmem_reserve_ratio",
827 .data = &sysctl_lowmem_reserve_ratio,
828 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
829 .mode = 0644,
830 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
831 .strategy = &sysctl_intvec,
832 },
833 {
834 .ctl_name = VM_DROP_PAGECACHE,
835 .procname = "drop_caches",
836 .data = &sysctl_drop_caches,
837 .maxlen = sizeof(int),
838 .mode = 0644,
839 .proc_handler = drop_caches_sysctl_handler,
840 .strategy = &sysctl_intvec,
841 },
842 {
843 .ctl_name = VM_MIN_FREE_KBYTES,
844 .procname = "min_free_kbytes",
845 .data = &min_free_kbytes,
846 .maxlen = sizeof(min_free_kbytes),
847 .mode = 0644,
848 .proc_handler = &min_free_kbytes_sysctl_handler,
849 .strategy = &sysctl_intvec,
850 .extra1 = &zero,
851 },
852 {
853 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
854 .procname = "percpu_pagelist_fraction",
855 .data = &percpu_pagelist_fraction,
856 .maxlen = sizeof(percpu_pagelist_fraction),
857 .mode = 0644,
858 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
859 .strategy = &sysctl_intvec,
860 .extra1 = &min_percpu_pagelist_fract,
861 },
862 #ifdef CONFIG_MMU
863 {
864 .ctl_name = VM_MAX_MAP_COUNT,
865 .procname = "max_map_count",
866 .data = &sysctl_max_map_count,
867 .maxlen = sizeof(sysctl_max_map_count),
868 .mode = 0644,
869 .proc_handler = &proc_dointvec
870 },
871 #endif
872 {
873 .ctl_name = VM_LAPTOP_MODE,
874 .procname = "laptop_mode",
875 .data = &laptop_mode,
876 .maxlen = sizeof(laptop_mode),
877 .mode = 0644,
878 .proc_handler = &proc_dointvec_jiffies,
879 .strategy = &sysctl_jiffies,
880 },
881 {
882 .ctl_name = VM_BLOCK_DUMP,
883 .procname = "block_dump",
884 .data = &block_dump,
885 .maxlen = sizeof(block_dump),
886 .mode = 0644,
887 .proc_handler = &proc_dointvec,
888 .strategy = &sysctl_intvec,
889 .extra1 = &zero,
890 },
891 {
892 .ctl_name = VM_VFS_CACHE_PRESSURE,
893 .procname = "vfs_cache_pressure",
894 .data = &sysctl_vfs_cache_pressure,
895 .maxlen = sizeof(sysctl_vfs_cache_pressure),
896 .mode = 0644,
897 .proc_handler = &proc_dointvec,
898 .strategy = &sysctl_intvec,
899 .extra1 = &zero,
900 },
901 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
902 {
903 .ctl_name = VM_LEGACY_VA_LAYOUT,
904 .procname = "legacy_va_layout",
905 .data = &sysctl_legacy_va_layout,
906 .maxlen = sizeof(sysctl_legacy_va_layout),
907 .mode = 0644,
908 .proc_handler = &proc_dointvec,
909 .strategy = &sysctl_intvec,
910 .extra1 = &zero,
911 },
912 #endif
913 #ifdef CONFIG_SWAP
914 {
915 .ctl_name = VM_SWAP_TOKEN_TIMEOUT,
916 .procname = "swap_token_timeout",
917 .data = &swap_token_default_timeout,
918 .maxlen = sizeof(swap_token_default_timeout),
919 .mode = 0644,
920 .proc_handler = &proc_dointvec_jiffies,
921 .strategy = &sysctl_jiffies,
922 },
923 #endif
924 #ifdef CONFIG_NUMA
925 {
926 .ctl_name = VM_ZONE_RECLAIM_MODE,
927 .procname = "zone_reclaim_mode",
928 .data = &zone_reclaim_mode,
929 .maxlen = sizeof(zone_reclaim_mode),
930 .mode = 0644,
931 .proc_handler = &proc_dointvec,
932 .strategy = &sysctl_intvec,
933 .extra1 = &zero,
934 },
935 #endif
936 #ifdef CONFIG_X86_32
937 {
938 .ctl_name = VM_VDSO_ENABLED,
939 .procname = "vdso_enabled",
940 .data = &vdso_enabled,
941 .maxlen = sizeof(vdso_enabled),
942 .mode = 0644,
943 .proc_handler = &proc_dointvec,
944 .strategy = &sysctl_intvec,
945 .extra1 = &zero,
946 },
947 #endif
948 { .ctl_name = 0 }
949 };
950
951 static ctl_table fs_table[] = {
952 {
953 .ctl_name = FS_NRINODE,
954 .procname = "inode-nr",
955 .data = &inodes_stat,
956 .maxlen = 2*sizeof(int),
957 .mode = 0444,
958 .proc_handler = &proc_dointvec,
959 },
960 {
961 .ctl_name = FS_STATINODE,
962 .procname = "inode-state",
963 .data = &inodes_stat,
964 .maxlen = 7*sizeof(int),
965 .mode = 0444,
966 .proc_handler = &proc_dointvec,
967 },
968 {
969 .ctl_name = FS_NRFILE,
970 .procname = "file-nr",
971 .data = &files_stat,
972 .maxlen = 3*sizeof(int),
973 .mode = 0444,
974 .proc_handler = &proc_nr_files,
975 },
976 {
977 .ctl_name = FS_MAXFILE,
978 .procname = "file-max",
979 .data = &files_stat.max_files,
980 .maxlen = sizeof(int),
981 .mode = 0644,
982 .proc_handler = &proc_dointvec,
983 },
984 {
985 .ctl_name = FS_DENTRY,
986 .procname = "dentry-state",
987 .data = &dentry_stat,
988 .maxlen = 6*sizeof(int),
989 .mode = 0444,
990 .proc_handler = &proc_dointvec,
991 },
992 {
993 .ctl_name = FS_OVERFLOWUID,
994 .procname = "overflowuid",
995 .data = &fs_overflowuid,
996 .maxlen = sizeof(int),
997 .mode = 0644,
998 .proc_handler = &proc_dointvec_minmax,
999 .strategy = &sysctl_intvec,
1000 .extra1 = &minolduid,
1001 .extra2 = &maxolduid,
1002 },
1003 {
1004 .ctl_name = FS_OVERFLOWGID,
1005 .procname = "overflowgid",
1006 .data = &fs_overflowgid,
1007 .maxlen = sizeof(int),
1008 .mode = 0644,
1009 .proc_handler = &proc_dointvec_minmax,
1010 .strategy = &sysctl_intvec,
1011 .extra1 = &minolduid,
1012 .extra2 = &maxolduid,
1013 },
1014 {
1015 .ctl_name = FS_LEASES,
1016 .procname = "leases-enable",
1017 .data = &leases_enable,
1018 .maxlen = sizeof(int),
1019 .mode = 0644,
1020 .proc_handler = &proc_dointvec,
1021 },
1022 #ifdef CONFIG_DNOTIFY
1023 {
1024 .ctl_name = FS_DIR_NOTIFY,
1025 .procname = "dir-notify-enable",
1026 .data = &dir_notify_enable,
1027 .maxlen = sizeof(int),
1028 .mode = 0644,
1029 .proc_handler = &proc_dointvec,
1030 },
1031 #endif
1032 #ifdef CONFIG_MMU
1033 {
1034 .ctl_name = FS_LEASE_TIME,
1035 .procname = "lease-break-time",
1036 .data = &lease_break_time,
1037 .maxlen = sizeof(int),
1038 .mode = 0644,
1039 .proc_handler = &proc_dointvec,
1040 },
1041 {
1042 .ctl_name = FS_AIO_NR,
1043 .procname = "aio-nr",
1044 .data = &aio_nr,
1045 .maxlen = sizeof(aio_nr),
1046 .mode = 0444,
1047 .proc_handler = &proc_doulongvec_minmax,
1048 },
1049 {
1050 .ctl_name = FS_AIO_MAX_NR,
1051 .procname = "aio-max-nr",
1052 .data = &aio_max_nr,
1053 .maxlen = sizeof(aio_max_nr),
1054 .mode = 0644,
1055 .proc_handler = &proc_doulongvec_minmax,
1056 },
1057 #ifdef CONFIG_INOTIFY_USER
1058 {
1059 .ctl_name = FS_INOTIFY,
1060 .procname = "inotify",
1061 .mode = 0555,
1062 .child = inotify_table,
1063 },
1064 #endif
1065 #endif
1066 {
1067 .ctl_name = KERN_SETUID_DUMPABLE,
1068 .procname = "suid_dumpable",
1069 .data = &suid_dumpable,
1070 .maxlen = sizeof(int),
1071 .mode = 0644,
1072 .proc_handler = &proc_dointvec,
1073 },
1074 { .ctl_name = 0 }
1075 };
1076
1077 static ctl_table debug_table[] = {
1078 { .ctl_name = 0 }
1079 };
1080
1081 static ctl_table dev_table[] = {
1082 { .ctl_name = 0 }
1083 };
1084
1085 extern void init_irq_proc (void);
1086
1087 static DEFINE_SPINLOCK(sysctl_lock);
1088
1089 /* called under sysctl_lock */
1090 static int use_table(struct ctl_table_header *p)
1091 {
1092 if (unlikely(p->unregistering))
1093 return 0;
1094 p->used++;
1095 return 1;
1096 }
1097
1098 /* called under sysctl_lock */
1099 static void unuse_table(struct ctl_table_header *p)
1100 {
1101 if (!--p->used)
1102 if (unlikely(p->unregistering))
1103 complete(p->unregistering);
1104 }
1105
1106 /* called under sysctl_lock, will reacquire if has to wait */
1107 static void start_unregistering(struct ctl_table_header *p)
1108 {
1109 /*
1110 * if p->used is 0, nobody will ever touch that entry again;
1111 * we'll eliminate all paths to it before dropping sysctl_lock
1112 */
1113 if (unlikely(p->used)) {
1114 struct completion wait;
1115 init_completion(&wait);
1116 p->unregistering = &wait;
1117 spin_unlock(&sysctl_lock);
1118 wait_for_completion(&wait);
1119 spin_lock(&sysctl_lock);
1120 }
1121 /*
1122 * do not remove from the list until nobody holds it; walking the
1123 * list in do_sysctl() relies on that.
1124 */
1125 list_del_init(&p->ctl_entry);
1126 }
1127
1128 void __init sysctl_init(void)
1129 {
1130 #ifdef CONFIG_PROC_FS
1131 register_proc_table(root_table, proc_sys_root, &root_table_header);
1132 init_irq_proc();
1133 #endif
1134 }
1135
1136 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1137 void __user *newval, size_t newlen)
1138 {
1139 struct list_head *tmp;
1140 int error = -ENOTDIR;
1141
1142 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1143 return -ENOTDIR;
1144 if (oldval) {
1145 int old_len;
1146 if (!oldlenp || get_user(old_len, oldlenp))
1147 return -EFAULT;
1148 }
1149 spin_lock(&sysctl_lock);
1150 tmp = &root_table_header.ctl_entry;
1151 do {
1152 struct ctl_table_header *head =
1153 list_entry(tmp, struct ctl_table_header, ctl_entry);
1154 void *context = NULL;
1155
1156 if (!use_table(head))
1157 continue;
1158
1159 spin_unlock(&sysctl_lock);
1160
1161 error = parse_table(name, nlen, oldval, oldlenp,
1162 newval, newlen, head->ctl_table,
1163 &context);
1164 kfree(context);
1165
1166 spin_lock(&sysctl_lock);
1167 unuse_table(head);
1168 if (error != -ENOTDIR)
1169 break;
1170 } while ((tmp = tmp->next) != &root_table_header.ctl_entry);
1171 spin_unlock(&sysctl_lock);
1172 return error;
1173 }
1174
1175 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1176 {
1177 struct __sysctl_args tmp;
1178 int error;
1179
1180 if (copy_from_user(&tmp, args, sizeof(tmp)))
1181 return -EFAULT;
1182
1183 lock_kernel();
1184 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1185 tmp.newval, tmp.newlen);
1186 unlock_kernel();
1187 return error;
1188 }
1189
1190 /*
1191 * ctl_perm does NOT grant the superuser all rights automatically, because
1192 * some sysctl variables are readonly even to root.
1193 */
1194
1195 static int test_perm(int mode, int op)
1196 {
1197 if (!current->euid)
1198 mode >>= 6;
1199 else if (in_egroup_p(0))
1200 mode >>= 3;
1201 if ((mode & op & 0007) == op)
1202 return 0;
1203 return -EACCES;
1204 }
1205
1206 static inline int ctl_perm(ctl_table *table, int op)
1207 {
1208 int error;
1209 error = security_sysctl(table, op);
1210 if (error)
1211 return error;
1212 return test_perm(table->mode, op);
1213 }
1214
1215 static int parse_table(int __user *name, int nlen,
1216 void __user *oldval, size_t __user *oldlenp,
1217 void __user *newval, size_t newlen,
1218 ctl_table *table, void **context)
1219 {
1220 int n;
1221 repeat:
1222 if (!nlen)
1223 return -ENOTDIR;
1224 if (get_user(n, name))
1225 return -EFAULT;
1226 for ( ; table->ctl_name; table++) {
1227 if (n == table->ctl_name || table->ctl_name == CTL_ANY) {
1228 int error;
1229 if (table->child) {
1230 if (ctl_perm(table, 001))
1231 return -EPERM;
1232 if (table->strategy) {
1233 error = table->strategy(
1234 table, name, nlen,
1235 oldval, oldlenp,
1236 newval, newlen, context);
1237 if (error)
1238 return error;
1239 }
1240 name++;
1241 nlen--;
1242 table = table->child;
1243 goto repeat;
1244 }
1245 error = do_sysctl_strategy(table, name, nlen,
1246 oldval, oldlenp,
1247 newval, newlen, context);
1248 return error;
1249 }
1250 }
1251 return -ENOTDIR;
1252 }
1253
1254 /* Perform the actual read/write of a sysctl table entry. */
1255 int do_sysctl_strategy (ctl_table *table,
1256 int __user *name, int nlen,
1257 void __user *oldval, size_t __user *oldlenp,
1258 void __user *newval, size_t newlen, void **context)
1259 {
1260 int op = 0, rc;
1261 size_t len;
1262
1263 if (oldval)
1264 op |= 004;
1265 if (newval)
1266 op |= 002;
1267 if (ctl_perm(table, op))
1268 return -EPERM;
1269
1270 if (table->strategy) {
1271 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1272 newval, newlen, context);
1273 if (rc < 0)
1274 return rc;
1275 if (rc > 0)
1276 return 0;
1277 }
1278
1279 /* If there is no strategy routine, or if the strategy returns
1280 * zero, proceed with automatic r/w */
1281 if (table->data && table->maxlen) {
1282 if (oldval && oldlenp) {
1283 if (get_user(len, oldlenp))
1284 return -EFAULT;
1285 if (len) {
1286 if (len > table->maxlen)
1287 len = table->maxlen;
1288 if(copy_to_user(oldval, table->data, len))
1289 return -EFAULT;
1290 if(put_user(len, oldlenp))
1291 return -EFAULT;
1292 }
1293 }
1294 if (newval && newlen) {
1295 len = newlen;
1296 if (len > table->maxlen)
1297 len = table->maxlen;
1298 if(copy_from_user(table->data, newval, len))
1299 return -EFAULT;
1300 }
1301 }
1302 return 0;
1303 }
1304
1305 /**
1306 * register_sysctl_table - register a sysctl hierarchy
1307 * @table: the top-level table structure
1308 * @insert_at_head: whether the entry should be inserted in front or at the end
1309 *
1310 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1311 * array. An entry with a ctl_name of 0 terminates the table.
1312 *
1313 * The members of the &ctl_table structure are used as follows:
1314 *
1315 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1316 * must be unique within that level of sysctl
1317 *
1318 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1319 * enter a sysctl file
1320 *
1321 * data - a pointer to data for use by proc_handler
1322 *
1323 * maxlen - the maximum size in bytes of the data
1324 *
1325 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1326 *
1327 * child - a pointer to the child sysctl table if this entry is a directory, or
1328 * %NULL.
1329 *
1330 * proc_handler - the text handler routine (described below)
1331 *
1332 * strategy - the strategy routine (described below)
1333 *
1334 * de - for internal use by the sysctl routines
1335 *
1336 * extra1, extra2 - extra pointers usable by the proc handler routines
1337 *
1338 * Leaf nodes in the sysctl tree will be represented by a single file
1339 * under /proc; non-leaf nodes will be represented by directories.
1340 *
1341 * sysctl(2) can automatically manage read and write requests through
1342 * the sysctl table. The data and maxlen fields of the ctl_table
1343 * struct enable minimal validation of the values being written to be
1344 * performed, and the mode field allows minimal authentication.
1345 *
1346 * More sophisticated management can be enabled by the provision of a
1347 * strategy routine with the table entry. This will be called before
1348 * any automatic read or write of the data is performed.
1349 *
1350 * The strategy routine may return
1351 *
1352 * < 0 - Error occurred (error is passed to user process)
1353 *
1354 * 0 - OK - proceed with automatic read or write.
1355 *
1356 * > 0 - OK - read or write has been done by the strategy routine, so
1357 * return immediately.
1358 *
1359 * There must be a proc_handler routine for any terminal nodes
1360 * mirrored under /proc/sys (non-terminals are handled by a built-in
1361 * directory handler). Several default handlers are available to
1362 * cover common cases -
1363 *
1364 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1365 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1366 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1367 *
1368 * It is the handler's job to read the input buffer from user memory
1369 * and process it. The handler should return 0 on success.
1370 *
1371 * This routine returns %NULL on a failure to register, and a pointer
1372 * to the table header on success.
1373 */
1374 struct ctl_table_header *register_sysctl_table(ctl_table * table,
1375 int insert_at_head)
1376 {
1377 struct ctl_table_header *tmp;
1378 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1379 if (!tmp)
1380 return NULL;
1381 tmp->ctl_table = table;
1382 INIT_LIST_HEAD(&tmp->ctl_entry);
1383 tmp->used = 0;
1384 tmp->unregistering = NULL;
1385 spin_lock(&sysctl_lock);
1386 if (insert_at_head)
1387 list_add(&tmp->ctl_entry, &root_table_header.ctl_entry);
1388 else
1389 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
1390 spin_unlock(&sysctl_lock);
1391 #ifdef CONFIG_PROC_FS
1392 register_proc_table(table, proc_sys_root, tmp);
1393 #endif
1394 return tmp;
1395 }
1396
1397 /**
1398 * unregister_sysctl_table - unregister a sysctl table hierarchy
1399 * @header: the header returned from register_sysctl_table
1400 *
1401 * Unregisters the sysctl table and all children. proc entries may not
1402 * actually be removed until they are no longer used by anyone.
1403 */
1404 void unregister_sysctl_table(struct ctl_table_header * header)
1405 {
1406 might_sleep();
1407 spin_lock(&sysctl_lock);
1408 start_unregistering(header);
1409 #ifdef CONFIG_PROC_FS
1410 unregister_proc_table(header->ctl_table, proc_sys_root);
1411 #endif
1412 spin_unlock(&sysctl_lock);
1413 kfree(header);
1414 }
1415
1416 /*
1417 * /proc/sys support
1418 */
1419
1420 #ifdef CONFIG_PROC_FS
1421
1422 /* Scan the sysctl entries in table and add them all into /proc */
1423 static void register_proc_table(ctl_table * table, struct proc_dir_entry *root, void *set)
1424 {
1425 struct proc_dir_entry *de;
1426 int len;
1427 mode_t mode;
1428
1429 for (; table->ctl_name; table++) {
1430 /* Can't do anything without a proc name. */
1431 if (!table->procname)
1432 continue;
1433 /* Maybe we can't do anything with it... */
1434 if (!table->proc_handler && !table->child) {
1435 printk(KERN_WARNING "SYSCTL: Can't register %s\n",
1436 table->procname);
1437 continue;
1438 }
1439
1440 len = strlen(table->procname);
1441 mode = table->mode;
1442
1443 de = NULL;
1444 if (table->proc_handler)
1445 mode |= S_IFREG;
1446 else {
1447 mode |= S_IFDIR;
1448 for (de = root->subdir; de; de = de->next) {
1449 if (proc_match(len, table->procname, de))
1450 break;
1451 }
1452 /* If the subdir exists already, de is non-NULL */
1453 }
1454
1455 if (!de) {
1456 de = create_proc_entry(table->procname, mode, root);
1457 if (!de)
1458 continue;
1459 de->set = set;
1460 de->data = (void *) table;
1461 if (table->proc_handler)
1462 de->proc_fops = &proc_sys_file_operations;
1463 }
1464 table->de = de;
1465 if (de->mode & S_IFDIR)
1466 register_proc_table(table->child, de, set);
1467 }
1468 }
1469
1470 /*
1471 * Unregister a /proc sysctl table and any subdirectories.
1472 */
1473 static void unregister_proc_table(ctl_table * table, struct proc_dir_entry *root)
1474 {
1475 struct proc_dir_entry *de;
1476 for (; table->ctl_name; table++) {
1477 if (!(de = table->de))
1478 continue;
1479 if (de->mode & S_IFDIR) {
1480 if (!table->child) {
1481 printk (KERN_ALERT "Help - malformed sysctl tree on free\n");
1482 continue;
1483 }
1484 unregister_proc_table(table->child, de);
1485
1486 /* Don't unregister directories which still have entries.. */
1487 if (de->subdir)
1488 continue;
1489 }
1490
1491 /*
1492 * In any case, mark the entry as goner; we'll keep it
1493 * around if it's busy, but we'll know to do nothing with
1494 * its fields. We are under sysctl_lock here.
1495 */
1496 de->data = NULL;
1497
1498 /* Don't unregister proc entries that are still being used.. */
1499 if (atomic_read(&de->count))
1500 continue;
1501
1502 table->de = NULL;
1503 remove_proc_entry(table->procname, root);
1504 }
1505 }
1506
1507 static ssize_t do_rw_proc(int write, struct file * file, char __user * buf,
1508 size_t count, loff_t *ppos)
1509 {
1510 int op;
1511 struct proc_dir_entry *de = PDE(file->f_dentry->d_inode);
1512 struct ctl_table *table;
1513 size_t res;
1514 ssize_t error = -ENOTDIR;
1515
1516 spin_lock(&sysctl_lock);
1517 if (de && de->data && use_table(de->set)) {
1518 /*
1519 * at that point we know that sysctl was not unregistered
1520 * and won't be until we finish
1521 */
1522 spin_unlock(&sysctl_lock);
1523 table = (struct ctl_table *) de->data;
1524 if (!table || !table->proc_handler)
1525 goto out;
1526 error = -EPERM;
1527 op = (write ? 002 : 004);
1528 if (ctl_perm(table, op))
1529 goto out;
1530
1531 /* careful: calling conventions are nasty here */
1532 res = count;
1533 error = (*table->proc_handler)(table, write, file,
1534 buf, &res, ppos);
1535 if (!error)
1536 error = res;
1537 out:
1538 spin_lock(&sysctl_lock);
1539 unuse_table(de->set);
1540 }
1541 spin_unlock(&sysctl_lock);
1542 return error;
1543 }
1544
1545 static int proc_opensys(struct inode *inode, struct file *file)
1546 {
1547 if (file->f_mode & FMODE_WRITE) {
1548 /*
1549 * sysctl entries that are not writable,
1550 * are _NOT_ writable, capabilities or not.
1551 */
1552 if (!(inode->i_mode & S_IWUSR))
1553 return -EPERM;
1554 }
1555
1556 return 0;
1557 }
1558
1559 static ssize_t proc_readsys(struct file * file, char __user * buf,
1560 size_t count, loff_t *ppos)
1561 {
1562 return do_rw_proc(0, file, buf, count, ppos);
1563 }
1564
1565 static ssize_t proc_writesys(struct file * file, const char __user * buf,
1566 size_t count, loff_t *ppos)
1567 {
1568 return do_rw_proc(1, file, (char __user *) buf, count, ppos);
1569 }
1570
1571 /**
1572 * proc_dostring - read a string sysctl
1573 * @table: the sysctl table
1574 * @write: %TRUE if this is a write to the sysctl file
1575 * @filp: the file structure
1576 * @buffer: the user buffer
1577 * @lenp: the size of the user buffer
1578 * @ppos: file position
1579 *
1580 * Reads/writes a string from/to the user buffer. If the kernel
1581 * buffer provided is not large enough to hold the string, the
1582 * string is truncated. The copied string is %NULL-terminated.
1583 * If the string is being read by the user process, it is copied
1584 * and a newline '\n' is added. It is truncated if the buffer is
1585 * not large enough.
1586 *
1587 * Returns 0 on success.
1588 */
1589 int proc_dostring(ctl_table *table, int write, struct file *filp,
1590 void __user *buffer, size_t *lenp, loff_t *ppos)
1591 {
1592 size_t len;
1593 char __user *p;
1594 char c;
1595
1596 if (!table->data || !table->maxlen || !*lenp ||
1597 (*ppos && !write)) {
1598 *lenp = 0;
1599 return 0;
1600 }
1601
1602 if (write) {
1603 len = 0;
1604 p = buffer;
1605 while (len < *lenp) {
1606 if (get_user(c, p++))
1607 return -EFAULT;
1608 if (c == 0 || c == '\n')
1609 break;
1610 len++;
1611 }
1612 if (len >= table->maxlen)
1613 len = table->maxlen-1;
1614 if(copy_from_user(table->data, buffer, len))
1615 return -EFAULT;
1616 ((char *) table->data)[len] = 0;
1617 *ppos += *lenp;
1618 } else {
1619 len = strlen(table->data);
1620 if (len > table->maxlen)
1621 len = table->maxlen;
1622 if (len > *lenp)
1623 len = *lenp;
1624 if (len)
1625 if(copy_to_user(buffer, table->data, len))
1626 return -EFAULT;
1627 if (len < *lenp) {
1628 if(put_user('\n', ((char __user *) buffer) + len))
1629 return -EFAULT;
1630 len++;
1631 }
1632 *lenp = len;
1633 *ppos += len;
1634 }
1635 return 0;
1636 }
1637
1638 /*
1639 * Special case of dostring for the UTS structure. This has locks
1640 * to observe. Should this be in kernel/sys.c ????
1641 */
1642
1643 static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
1644 void __user *buffer, size_t *lenp, loff_t *ppos)
1645 {
1646 int r;
1647
1648 if (!write) {
1649 down_read(&uts_sem);
1650 r=proc_dostring(table,0,filp,buffer,lenp, ppos);
1651 up_read(&uts_sem);
1652 } else {
1653 down_write(&uts_sem);
1654 r=proc_dostring(table,1,filp,buffer,lenp, ppos);
1655 up_write(&uts_sem);
1656 }
1657 return r;
1658 }
1659
1660 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1661 int *valp,
1662 int write, void *data)
1663 {
1664 if (write) {
1665 *valp = *negp ? -*lvalp : *lvalp;
1666 } else {
1667 int val = *valp;
1668 if (val < 0) {
1669 *negp = -1;
1670 *lvalp = (unsigned long)-val;
1671 } else {
1672 *negp = 0;
1673 *lvalp = (unsigned long)val;
1674 }
1675 }
1676 return 0;
1677 }
1678
1679 static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1680 void __user *buffer, size_t *lenp, loff_t *ppos,
1681 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1682 int write, void *data),
1683 void *data)
1684 {
1685 #define TMPBUFLEN 21
1686 int *i, vleft, first=1, neg, val;
1687 unsigned long lval;
1688 size_t left, len;
1689
1690 char buf[TMPBUFLEN], *p;
1691 char __user *s = buffer;
1692
1693 if (!table->data || !table->maxlen || !*lenp ||
1694 (*ppos && !write)) {
1695 *lenp = 0;
1696 return 0;
1697 }
1698
1699 i = (int *) table->data;
1700 vleft = table->maxlen / sizeof(*i);
1701 left = *lenp;
1702
1703 if (!conv)
1704 conv = do_proc_dointvec_conv;
1705
1706 for (; left && vleft--; i++, first=0) {
1707 if (write) {
1708 while (left) {
1709 char c;
1710 if (get_user(c, s))
1711 return -EFAULT;
1712 if (!isspace(c))
1713 break;
1714 left--;
1715 s++;
1716 }
1717 if (!left)
1718 break;
1719 neg = 0;
1720 len = left;
1721 if (len > sizeof(buf) - 1)
1722 len = sizeof(buf) - 1;
1723 if (copy_from_user(buf, s, len))
1724 return -EFAULT;
1725 buf[len] = 0;
1726 p = buf;
1727 if (*p == '-' && left > 1) {
1728 neg = 1;
1729 left--, p++;
1730 }
1731 if (*p < '0' || *p > '9')
1732 break;
1733
1734 lval = simple_strtoul(p, &p, 0);
1735
1736 len = p-buf;
1737 if ((len < left) && *p && !isspace(*p))
1738 break;
1739 if (neg)
1740 val = -val;
1741 s += len;
1742 left -= len;
1743
1744 if (conv(&neg, &lval, i, 1, data))
1745 break;
1746 } else {
1747 p = buf;
1748 if (!first)
1749 *p++ = '\t';
1750
1751 if (conv(&neg, &lval, i, 0, data))
1752 break;
1753
1754 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1755 len = strlen(buf);
1756 if (len > left)
1757 len = left;
1758 if(copy_to_user(s, buf, len))
1759 return -EFAULT;
1760 left -= len;
1761 s += len;
1762 }
1763 }
1764
1765 if (!write && !first && left) {
1766 if(put_user('\n', s))
1767 return -EFAULT;
1768 left--, s++;
1769 }
1770 if (write) {
1771 while (left) {
1772 char c;
1773 if (get_user(c, s++))
1774 return -EFAULT;
1775 if (!isspace(c))
1776 break;
1777 left--;
1778 }
1779 }
1780 if (write && first)
1781 return -EINVAL;
1782 *lenp -= left;
1783 *ppos += *lenp;
1784 return 0;
1785 #undef TMPBUFLEN
1786 }
1787
1788 /**
1789 * proc_dointvec - read a vector of integers
1790 * @table: the sysctl table
1791 * @write: %TRUE if this is a write to the sysctl file
1792 * @filp: the file structure
1793 * @buffer: the user buffer
1794 * @lenp: the size of the user buffer
1795 * @ppos: file position
1796 *
1797 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1798 * values from/to the user buffer, treated as an ASCII string.
1799 *
1800 * Returns 0 on success.
1801 */
1802 int proc_dointvec(ctl_table *table, int write, struct file *filp,
1803 void __user *buffer, size_t *lenp, loff_t *ppos)
1804 {
1805 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1806 NULL,NULL);
1807 }
1808
1809 #define OP_SET 0
1810 #define OP_AND 1
1811 #define OP_OR 2
1812 #define OP_MAX 3
1813 #define OP_MIN 4
1814
1815 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1816 int *valp,
1817 int write, void *data)
1818 {
1819 int op = *(int *)data;
1820 if (write) {
1821 int val = *negp ? -*lvalp : *lvalp;
1822 switch(op) {
1823 case OP_SET: *valp = val; break;
1824 case OP_AND: *valp &= val; break;
1825 case OP_OR: *valp |= val; break;
1826 case OP_MAX: if(*valp < val)
1827 *valp = val;
1828 break;
1829 case OP_MIN: if(*valp > val)
1830 *valp = val;
1831 break;
1832 }
1833 } else {
1834 int val = *valp;
1835 if (val < 0) {
1836 *negp = -1;
1837 *lvalp = (unsigned long)-val;
1838 } else {
1839 *negp = 0;
1840 *lvalp = (unsigned long)val;
1841 }
1842 }
1843 return 0;
1844 }
1845
1846 /*
1847 * init may raise the set.
1848 */
1849
1850 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1851 void __user *buffer, size_t *lenp, loff_t *ppos)
1852 {
1853 int op;
1854
1855 if (!capable(CAP_SYS_MODULE)) {
1856 return -EPERM;
1857 }
1858
1859 op = (current->pid == 1) ? OP_SET : OP_AND;
1860 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1861 do_proc_dointvec_bset_conv,&op);
1862 }
1863
1864 struct do_proc_dointvec_minmax_conv_param {
1865 int *min;
1866 int *max;
1867 };
1868
1869 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1870 int *valp,
1871 int write, void *data)
1872 {
1873 struct do_proc_dointvec_minmax_conv_param *param = data;
1874 if (write) {
1875 int val = *negp ? -*lvalp : *lvalp;
1876 if ((param->min && *param->min > val) ||
1877 (param->max && *param->max < val))
1878 return -EINVAL;
1879 *valp = val;
1880 } else {
1881 int val = *valp;
1882 if (val < 0) {
1883 *negp = -1;
1884 *lvalp = (unsigned long)-val;
1885 } else {
1886 *negp = 0;
1887 *lvalp = (unsigned long)val;
1888 }
1889 }
1890 return 0;
1891 }
1892
1893 /**
1894 * proc_dointvec_minmax - read a vector of integers with min/max values
1895 * @table: the sysctl table
1896 * @write: %TRUE if this is a write to the sysctl file
1897 * @filp: the file structure
1898 * @buffer: the user buffer
1899 * @lenp: the size of the user buffer
1900 * @ppos: file position
1901 *
1902 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1903 * values from/to the user buffer, treated as an ASCII string.
1904 *
1905 * This routine will ensure the values are within the range specified by
1906 * table->extra1 (min) and table->extra2 (max).
1907 *
1908 * Returns 0 on success.
1909 */
1910 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1911 void __user *buffer, size_t *lenp, loff_t *ppos)
1912 {
1913 struct do_proc_dointvec_minmax_conv_param param = {
1914 .min = (int *) table->extra1,
1915 .max = (int *) table->extra2,
1916 };
1917 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1918 do_proc_dointvec_minmax_conv, &param);
1919 }
1920
1921 static int do_proc_doulongvec_minmax(ctl_table *table, int write,
1922 struct file *filp,
1923 void __user *buffer,
1924 size_t *lenp, loff_t *ppos,
1925 unsigned long convmul,
1926 unsigned long convdiv)
1927 {
1928 #define TMPBUFLEN 21
1929 unsigned long *i, *min, *max, val;
1930 int vleft, first=1, neg;
1931 size_t len, left;
1932 char buf[TMPBUFLEN], *p;
1933 char __user *s = buffer;
1934
1935 if (!table->data || !table->maxlen || !*lenp ||
1936 (*ppos && !write)) {
1937 *lenp = 0;
1938 return 0;
1939 }
1940
1941 i = (unsigned long *) table->data;
1942 min = (unsigned long *) table->extra1;
1943 max = (unsigned long *) table->extra2;
1944 vleft = table->maxlen / sizeof(unsigned long);
1945 left = *lenp;
1946
1947 for (; left && vleft--; i++, min++, max++, first=0) {
1948 if (write) {
1949 while (left) {
1950 char c;
1951 if (get_user(c, s))
1952 return -EFAULT;
1953 if (!isspace(c))
1954 break;
1955 left--;
1956 s++;
1957 }
1958 if (!left)
1959 break;
1960 neg = 0;
1961 len = left;
1962 if (len > TMPBUFLEN-1)
1963 len = TMPBUFLEN-1;
1964 if (copy_from_user(buf, s, len))
1965 return -EFAULT;
1966 buf[len] = 0;
1967 p = buf;
1968 if (*p == '-' && left > 1) {
1969 neg = 1;
1970 left--, p++;
1971 }
1972 if (*p < '0' || *p > '9')
1973 break;
1974 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
1975 len = p-buf;
1976 if ((len < left) && *p && !isspace(*p))
1977 break;
1978 if (neg)
1979 val = -val;
1980 s += len;
1981 left -= len;
1982
1983 if(neg)
1984 continue;
1985 if ((min && val < *min) || (max && val > *max))
1986 continue;
1987 *i = val;
1988 } else {
1989 p = buf;
1990 if (!first)
1991 *p++ = '\t';
1992 sprintf(p, "%lu", convdiv * (*i) / convmul);
1993 len = strlen(buf);
1994 if (len > left)
1995 len = left;
1996 if(copy_to_user(s, buf, len))
1997 return -EFAULT;
1998 left -= len;
1999 s += len;
2000 }
2001 }
2002
2003 if (!write && !first && left) {
2004 if(put_user('\n', s))
2005 return -EFAULT;
2006 left--, s++;
2007 }
2008 if (write) {
2009 while (left) {
2010 char c;
2011 if (get_user(c, s++))
2012 return -EFAULT;
2013 if (!isspace(c))
2014 break;
2015 left--;
2016 }
2017 }
2018 if (write && first)
2019 return -EINVAL;
2020 *lenp -= left;
2021 *ppos += *lenp;
2022 return 0;
2023 #undef TMPBUFLEN
2024 }
2025
2026 /**
2027 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2028 * @table: the sysctl table
2029 * @write: %TRUE if this is a write to the sysctl file
2030 * @filp: the file structure
2031 * @buffer: the user buffer
2032 * @lenp: the size of the user buffer
2033 * @ppos: file position
2034 *
2035 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2036 * values from/to the user buffer, treated as an ASCII string.
2037 *
2038 * This routine will ensure the values are within the range specified by
2039 * table->extra1 (min) and table->extra2 (max).
2040 *
2041 * Returns 0 on success.
2042 */
2043 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2044 void __user *buffer, size_t *lenp, loff_t *ppos)
2045 {
2046 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2047 }
2048
2049 /**
2050 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2051 * @table: the sysctl table
2052 * @write: %TRUE if this is a write to the sysctl file
2053 * @filp: the file structure
2054 * @buffer: the user buffer
2055 * @lenp: the size of the user buffer
2056 * @ppos: file position
2057 *
2058 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2059 * values from/to the user buffer, treated as an ASCII string. The values
2060 * are treated as milliseconds, and converted to jiffies when they are stored.
2061 *
2062 * This routine will ensure the values are within the range specified by
2063 * table->extra1 (min) and table->extra2 (max).
2064 *
2065 * Returns 0 on success.
2066 */
2067 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2068 struct file *filp,
2069 void __user *buffer,
2070 size_t *lenp, loff_t *ppos)
2071 {
2072 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2073 lenp, ppos, HZ, 1000l);
2074 }
2075
2076
2077 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2078 int *valp,
2079 int write, void *data)
2080 {
2081 if (write) {
2082 if (*lvalp > LONG_MAX / HZ)
2083 return 1;
2084 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2085 } else {
2086 int val = *valp;
2087 unsigned long lval;
2088 if (val < 0) {
2089 *negp = -1;
2090 lval = (unsigned long)-val;
2091 } else {
2092 *negp = 0;
2093 lval = (unsigned long)val;
2094 }
2095 *lvalp = lval / HZ;
2096 }
2097 return 0;
2098 }
2099
2100 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2101 int *valp,
2102 int write, void *data)
2103 {
2104 if (write) {
2105 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2106 return 1;
2107 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2108 } else {
2109 int val = *valp;
2110 unsigned long lval;
2111 if (val < 0) {
2112 *negp = -1;
2113 lval = (unsigned long)-val;
2114 } else {
2115 *negp = 0;
2116 lval = (unsigned long)val;
2117 }
2118 *lvalp = jiffies_to_clock_t(lval);
2119 }
2120 return 0;
2121 }
2122
2123 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2124 int *valp,
2125 int write, void *data)
2126 {
2127 if (write) {
2128 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2129 } else {
2130 int val = *valp;
2131 unsigned long lval;
2132 if (val < 0) {
2133 *negp = -1;
2134 lval = (unsigned long)-val;
2135 } else {
2136 *negp = 0;
2137 lval = (unsigned long)val;
2138 }
2139 *lvalp = jiffies_to_msecs(lval);
2140 }
2141 return 0;
2142 }
2143
2144 /**
2145 * proc_dointvec_jiffies - read a vector of integers as seconds
2146 * @table: the sysctl table
2147 * @write: %TRUE if this is a write to the sysctl file
2148 * @filp: the file structure
2149 * @buffer: the user buffer
2150 * @lenp: the size of the user buffer
2151 * @ppos: file position
2152 *
2153 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2154 * values from/to the user buffer, treated as an ASCII string.
2155 * The values read are assumed to be in seconds, and are converted into
2156 * jiffies.
2157 *
2158 * Returns 0 on success.
2159 */
2160 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2161 void __user *buffer, size_t *lenp, loff_t *ppos)
2162 {
2163 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2164 do_proc_dointvec_jiffies_conv,NULL);
2165 }
2166
2167 /**
2168 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2169 * @table: the sysctl table
2170 * @write: %TRUE if this is a write to the sysctl file
2171 * @filp: the file structure
2172 * @buffer: the user buffer
2173 * @lenp: the size of the user buffer
2174 * @ppos: pointer to the file position
2175 *
2176 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2177 * values from/to the user buffer, treated as an ASCII string.
2178 * The values read are assumed to be in 1/USER_HZ seconds, and
2179 * are converted into jiffies.
2180 *
2181 * Returns 0 on success.
2182 */
2183 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2184 void __user *buffer, size_t *lenp, loff_t *ppos)
2185 {
2186 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2187 do_proc_dointvec_userhz_jiffies_conv,NULL);
2188 }
2189
2190 /**
2191 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2192 * @table: the sysctl table
2193 * @write: %TRUE if this is a write to the sysctl file
2194 * @filp: the file structure
2195 * @buffer: the user buffer
2196 * @lenp: the size of the user buffer
2197 * @ppos: file position
2198 * @ppos: the current position in the file
2199 *
2200 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2201 * values from/to the user buffer, treated as an ASCII string.
2202 * The values read are assumed to be in 1/1000 seconds, and
2203 * are converted into jiffies.
2204 *
2205 * Returns 0 on success.
2206 */
2207 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2208 void __user *buffer, size_t *lenp, loff_t *ppos)
2209 {
2210 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2211 do_proc_dointvec_ms_jiffies_conv, NULL);
2212 }
2213
2214 #else /* CONFIG_PROC_FS */
2215
2216 int proc_dostring(ctl_table *table, int write, struct file *filp,
2217 void __user *buffer, size_t *lenp, loff_t *ppos)
2218 {
2219 return -ENOSYS;
2220 }
2221
2222 static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
2223 void __user *buffer, size_t *lenp, loff_t *ppos)
2224 {
2225 return -ENOSYS;
2226 }
2227
2228 int proc_dointvec(ctl_table *table, int write, struct file *filp,
2229 void __user *buffer, size_t *lenp, loff_t *ppos)
2230 {
2231 return -ENOSYS;
2232 }
2233
2234 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2235 void __user *buffer, size_t *lenp, loff_t *ppos)
2236 {
2237 return -ENOSYS;
2238 }
2239
2240 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2241 void __user *buffer, size_t *lenp, loff_t *ppos)
2242 {
2243 return -ENOSYS;
2244 }
2245
2246 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2247 void __user *buffer, size_t *lenp, loff_t *ppos)
2248 {
2249 return -ENOSYS;
2250 }
2251
2252 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2253 void __user *buffer, size_t *lenp, loff_t *ppos)
2254 {
2255 return -ENOSYS;
2256 }
2257
2258 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2259 void __user *buffer, size_t *lenp, loff_t *ppos)
2260 {
2261 return -ENOSYS;
2262 }
2263
2264 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2265 void __user *buffer, size_t *lenp, loff_t *ppos)
2266 {
2267 return -ENOSYS;
2268 }
2269
2270 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2271 struct file *filp,
2272 void __user *buffer,
2273 size_t *lenp, loff_t *ppos)
2274 {
2275 return -ENOSYS;
2276 }
2277
2278
2279 #endif /* CONFIG_PROC_FS */
2280
2281
2282 /*
2283 * General sysctl support routines
2284 */
2285
2286 /* The generic string strategy routine: */
2287 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2288 void __user *oldval, size_t __user *oldlenp,
2289 void __user *newval, size_t newlen, void **context)
2290 {
2291 if (!table->data || !table->maxlen)
2292 return -ENOTDIR;
2293
2294 if (oldval && oldlenp) {
2295 size_t bufsize;
2296 if (get_user(bufsize, oldlenp))
2297 return -EFAULT;
2298 if (bufsize) {
2299 size_t len = strlen(table->data), copied;
2300
2301 /* This shouldn't trigger for a well-formed sysctl */
2302 if (len > table->maxlen)
2303 len = table->maxlen;
2304
2305 /* Copy up to a max of bufsize-1 bytes of the string */
2306 copied = (len >= bufsize) ? bufsize - 1 : len;
2307
2308 if (copy_to_user(oldval, table->data, copied) ||
2309 put_user(0, (char __user *)(oldval + copied)))
2310 return -EFAULT;
2311 if (put_user(len, oldlenp))
2312 return -EFAULT;
2313 }
2314 }
2315 if (newval && newlen) {
2316 size_t len = newlen;
2317 if (len > table->maxlen)
2318 len = table->maxlen;
2319 if(copy_from_user(table->data, newval, len))
2320 return -EFAULT;
2321 if (len == table->maxlen)
2322 len--;
2323 ((char *) table->data)[len] = 0;
2324 }
2325 return 1;
2326 }
2327
2328 /*
2329 * This function makes sure that all of the integers in the vector
2330 * are between the minimum and maximum values given in the arrays
2331 * table->extra1 and table->extra2, respectively.
2332 */
2333 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2334 void __user *oldval, size_t __user *oldlenp,
2335 void __user *newval, size_t newlen, void **context)
2336 {
2337
2338 if (newval && newlen) {
2339 int __user *vec = (int __user *) newval;
2340 int *min = (int *) table->extra1;
2341 int *max = (int *) table->extra2;
2342 size_t length;
2343 int i;
2344
2345 if (newlen % sizeof(int) != 0)
2346 return -EINVAL;
2347
2348 if (!table->extra1 && !table->extra2)
2349 return 0;
2350
2351 if (newlen > table->maxlen)
2352 newlen = table->maxlen;
2353 length = newlen / sizeof(int);
2354
2355 for (i = 0; i < length; i++) {
2356 int value;
2357 if (get_user(value, vec + i))
2358 return -EFAULT;
2359 if (min && value < min[i])
2360 return -EINVAL;
2361 if (max && value > max[i])
2362 return -EINVAL;
2363 }
2364 }
2365 return 0;
2366 }
2367
2368 /* Strategy function to convert jiffies to seconds */
2369 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2370 void __user *oldval, size_t __user *oldlenp,
2371 void __user *newval, size_t newlen, void **context)
2372 {
2373 if (oldval) {
2374 size_t olen;
2375 if (oldlenp) {
2376 if (get_user(olen, oldlenp))
2377 return -EFAULT;
2378 if (olen!=sizeof(int))
2379 return -EINVAL;
2380 }
2381 if (put_user(*(int *)(table->data)/HZ, (int __user *)oldval) ||
2382 (oldlenp && put_user(sizeof(int),oldlenp)))
2383 return -EFAULT;
2384 }
2385 if (newval && newlen) {
2386 int new;
2387 if (newlen != sizeof(int))
2388 return -EINVAL;
2389 if (get_user(new, (int __user *)newval))
2390 return -EFAULT;
2391 *(int *)(table->data) = new*HZ;
2392 }
2393 return 1;
2394 }
2395
2396 /* Strategy function to convert jiffies to seconds */
2397 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2398 void __user *oldval, size_t __user *oldlenp,
2399 void __user *newval, size_t newlen, void **context)
2400 {
2401 if (oldval) {
2402 size_t olen;
2403 if (oldlenp) {
2404 if (get_user(olen, oldlenp))
2405 return -EFAULT;
2406 if (olen!=sizeof(int))
2407 return -EINVAL;
2408 }
2409 if (put_user(jiffies_to_msecs(*(int *)(table->data)), (int __user *)oldval) ||
2410 (oldlenp && put_user(sizeof(int),oldlenp)))
2411 return -EFAULT;
2412 }
2413 if (newval && newlen) {
2414 int new;
2415 if (newlen != sizeof(int))
2416 return -EINVAL;
2417 if (get_user(new, (int __user *)newval))
2418 return -EFAULT;
2419 *(int *)(table->data) = msecs_to_jiffies(new);
2420 }
2421 return 1;
2422 }
2423
2424 #else /* CONFIG_SYSCTL */
2425
2426
2427 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2428 {
2429 return -ENOSYS;
2430 }
2431
2432 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2433 void __user *oldval, size_t __user *oldlenp,
2434 void __user *newval, size_t newlen, void **context)
2435 {
2436 return -ENOSYS;
2437 }
2438
2439 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2440 void __user *oldval, size_t __user *oldlenp,
2441 void __user *newval, size_t newlen, void **context)
2442 {
2443 return -ENOSYS;
2444 }
2445
2446 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2447 void __user *oldval, size_t __user *oldlenp,
2448 void __user *newval, size_t newlen, void **context)
2449 {
2450 return -ENOSYS;
2451 }
2452
2453 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2454 void __user *oldval, size_t __user *oldlenp,
2455 void __user *newval, size_t newlen, void **context)
2456 {
2457 return -ENOSYS;
2458 }
2459
2460 int proc_dostring(ctl_table *table, int write, struct file *filp,
2461 void __user *buffer, size_t *lenp, loff_t *ppos)
2462 {
2463 return -ENOSYS;
2464 }
2465
2466 int proc_dointvec(ctl_table *table, int write, struct file *filp,
2467 void __user *buffer, size_t *lenp, loff_t *ppos)
2468 {
2469 return -ENOSYS;
2470 }
2471
2472 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2473 void __user *buffer, size_t *lenp, loff_t *ppos)
2474 {
2475 return -ENOSYS;
2476 }
2477
2478 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2479 void __user *buffer, size_t *lenp, loff_t *ppos)
2480 {
2481 return -ENOSYS;
2482 }
2483
2484 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2485 void __user *buffer, size_t *lenp, loff_t *ppos)
2486 {
2487 return -ENOSYS;
2488 }
2489
2490 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2491 void __user *buffer, size_t *lenp, loff_t *ppos)
2492 {
2493 return -ENOSYS;
2494 }
2495
2496 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2497 void __user *buffer, size_t *lenp, loff_t *ppos)
2498 {
2499 return -ENOSYS;
2500 }
2501
2502 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2503 void __user *buffer, size_t *lenp, loff_t *ppos)
2504 {
2505 return -ENOSYS;
2506 }
2507
2508 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2509 struct file *filp,
2510 void __user *buffer,
2511 size_t *lenp, loff_t *ppos)
2512 {
2513 return -ENOSYS;
2514 }
2515
2516 struct ctl_table_header * register_sysctl_table(ctl_table * table,
2517 int insert_at_head)
2518 {
2519 return NULL;
2520 }
2521
2522 void unregister_sysctl_table(struct ctl_table_header * table)
2523 {
2524 }
2525
2526 #endif /* CONFIG_SYSCTL */
2527
2528 /*
2529 * No sense putting this after each symbol definition, twice,
2530 * exception granted :-)
2531 */
2532 EXPORT_SYMBOL(proc_dointvec);
2533 EXPORT_SYMBOL(proc_dointvec_jiffies);
2534 EXPORT_SYMBOL(proc_dointvec_minmax);
2535 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2536 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2537 EXPORT_SYMBOL(proc_dostring);
2538 EXPORT_SYMBOL(proc_doulongvec_minmax);
2539 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2540 EXPORT_SYMBOL(register_sysctl_table);
2541 EXPORT_SYMBOL(sysctl_intvec);
2542 EXPORT_SYMBOL(sysctl_jiffies);
2543 EXPORT_SYMBOL(sysctl_ms_jiffies);
2544 EXPORT_SYMBOL(sysctl_string);
2545 EXPORT_SYMBOL(unregister_sysctl_table);
Linux 2.6 ibm-acpi/thinkpad-acpi driver git tree