?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
[PATCH] sysctl: move init_irq_proc into init/main where it belongs
[linux-2.6/linux-mips.git] / kernel / sysctl.c
blob7ba4b0c16d6a324091e4bcf103dd072d062f086c
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 #ifdef CONFIG_X86
56 #include <asm/nmi.h>
57 #include <asm/stacktrace.h>
58 #endif
59
60 #if defined(CONFIG_SYSCTL)
61
62 /* External variables not in a header file. */
63 extern int C_A_D;
64 extern int sysctl_overcommit_memory;
65 extern int sysctl_overcommit_ratio;
66 extern int sysctl_panic_on_oom;
67 extern int max_threads;
68 extern int core_uses_pid;
69 extern int suid_dumpable;
70 extern char core_pattern[];
71 extern int pid_max;
72 extern int min_free_kbytes;
73 extern int printk_ratelimit_jiffies;
74 extern int printk_ratelimit_burst;
75 extern int pid_max_min, pid_max_max;
76 extern int sysctl_drop_caches;
77 extern int percpu_pagelist_fraction;
78 extern int compat_log;
79
80 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
81 static int maxolduid = 65535;
82 static int minolduid;
83 static int min_percpu_pagelist_fract = 8;
84
85 static int ngroups_max = NGROUPS_MAX;
86
87 #ifdef CONFIG_KMOD
88 extern char modprobe_path[];
89 #endif
90 #ifdef CONFIG_CHR_DEV_SG
91 extern int sg_big_buff;
92 #endif
93 #ifdef CONFIG_SYSVIPC
94 static int proc_ipc_dointvec(ctl_table *table, int write, struct file *filp,
95 void __user *buffer, size_t *lenp, loff_t *ppos);
96 static int proc_ipc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
97 void __user *buffer, size_t *lenp, loff_t *ppos);
98 #endif
99
100 #ifdef __sparc__
101 extern char reboot_command [];
102 extern int stop_a_enabled;
103 extern int scons_pwroff;
104 #endif
105
106 #ifdef __hppa__
107 extern int pwrsw_enabled;
108 extern int unaligned_enabled;
109 #endif
110
111 #ifdef CONFIG_S390
112 #ifdef CONFIG_MATHEMU
113 extern int sysctl_ieee_emulation_warnings;
114 #endif
115 extern int sysctl_userprocess_debug;
116 extern int spin_retry;
117 #endif
118
119 extern int sysctl_hz_timer;
120
121 #ifdef CONFIG_BSD_PROCESS_ACCT
122 extern int acct_parm[];
123 #endif
124
125 #ifdef CONFIG_IA64
126 extern int no_unaligned_warning;
127 #endif
128
129 #ifdef CONFIG_RT_MUTEXES
130 extern int max_lock_depth;
131 #endif
132
133 #ifdef CONFIG_SYSCTL_SYSCALL
134 static int parse_table(int __user *, int, void __user *, size_t __user *,
135 void __user *, size_t, ctl_table *);
136 #endif
137
138 static int proc_do_uts_string(ctl_table *table, int write, struct file *filp,
139 void __user *buffer, size_t *lenp, loff_t *ppos);
140
141 static int sysctl_uts_string(ctl_table *table, int __user *name, int nlen,
142 void __user *oldval, size_t __user *oldlenp,
143 void __user *newval, size_t newlen);
144
145 #ifdef CONFIG_SYSVIPC
146 static int sysctl_ipc_data(ctl_table *table, int __user *name, int nlen,
147 void __user *oldval, size_t __user *oldlenp,
148 void __user *newval, size_t newlen);
149 #endif
150
151 #ifdef CONFIG_PROC_SYSCTL
152 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
153 void __user *buffer, size_t *lenp, loff_t *ppos);
154 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
155 void __user *buffer, size_t *lenp, loff_t *ppos);
156 #endif
157
158 static ctl_table root_table[];
159 static struct ctl_table_header root_table_header =
160 { root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
161
162 static ctl_table kern_table[];
163 static ctl_table vm_table[];
164 static ctl_table fs_table[];
165 static ctl_table debug_table[];
166 static ctl_table dev_table[];
167 extern ctl_table random_table[];
168 #ifdef CONFIG_UNIX98_PTYS
169 extern ctl_table pty_table[];
170 #endif
171 #ifdef CONFIG_INOTIFY_USER
172 extern ctl_table inotify_table[];
173 #endif
174
175 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
176 int sysctl_legacy_va_layout;
177 #endif
178
179
180 static void *get_uts(ctl_table *table, int write)
181 {
182 char *which = table->data;
183 #ifdef CONFIG_UTS_NS
184 struct uts_namespace *uts_ns = current->nsproxy->uts_ns;
185 which = (which - (char *)&init_uts_ns) + (char *)uts_ns;
186 #endif
187 if (!write)
188 down_read(&uts_sem);
189 else
190 down_write(&uts_sem);
191 return which;
192 }
193
194 static void put_uts(ctl_table *table, int write, void *which)
195 {
196 if (!write)
197 up_read(&uts_sem);
198 else
199 up_write(&uts_sem);
200 }
201
202 #ifdef CONFIG_SYSVIPC
203 static void *get_ipc(ctl_table *table, int write)
204 {
205 char *which = table->data;
206 struct ipc_namespace *ipc_ns = current->nsproxy->ipc_ns;
207 which = (which - (char *)&init_ipc_ns) + (char *)ipc_ns;
208 return which;
209 }
210 #else
211 #define get_ipc(T,W) ((T)->data)
212 #endif
213
214 /* /proc declarations: */
215
216 #ifdef CONFIG_PROC_SYSCTL
217
218 static ssize_t proc_readsys(struct file *, char __user *, size_t, loff_t *);
219 static ssize_t proc_writesys(struct file *, const char __user *, size_t, loff_t *);
220 static int proc_opensys(struct inode *, struct file *);
221
222 const struct file_operations proc_sys_file_operations = {
223 .open = proc_opensys,
224 .read = proc_readsys,
225 .write = proc_writesys,
226 };
227
228 extern struct proc_dir_entry *proc_sys_root;
229
230 static void register_proc_table(ctl_table *, struct proc_dir_entry *, void *);
231 static void unregister_proc_table(ctl_table *, struct proc_dir_entry *);
232 #endif
233
234 /* The default sysctl tables: */
235
236 static ctl_table root_table[] = {
237 {
238 .ctl_name = CTL_KERN,
239 .procname = "kernel",
240 .mode = 0555,
241 .child = kern_table,
242 },
243 {
244 .ctl_name = CTL_VM,
245 .procname = "vm",
246 .mode = 0555,
247 .child = vm_table,
248 },
249 #ifdef CONFIG_NET
250 {
251 .ctl_name = CTL_NET,
252 .procname = "net",
253 .mode = 0555,
254 .child = net_table,
255 },
256 #endif
257 {
258 .ctl_name = CTL_FS,
259 .procname = "fs",
260 .mode = 0555,
261 .child = fs_table,
262 },
263 {
264 .ctl_name = CTL_DEBUG,
265 .procname = "debug",
266 .mode = 0555,
267 .child = debug_table,
268 },
269 {
270 .ctl_name = CTL_DEV,
271 .procname = "dev",
272 .mode = 0555,
273 .child = dev_table,
274 },
275
276 { .ctl_name = 0 }
277 };
278
279 static ctl_table kern_table[] = {
280 {
281 .ctl_name = KERN_OSTYPE,
282 .procname = "ostype",
283 .data = init_uts_ns.name.sysname,
284 .maxlen = sizeof(init_uts_ns.name.sysname),
285 .mode = 0444,
286 .proc_handler = &proc_do_uts_string,
287 .strategy = &sysctl_uts_string,
288 },
289 {
290 .ctl_name = KERN_OSRELEASE,
291 .procname = "osrelease",
292 .data = init_uts_ns.name.release,
293 .maxlen = sizeof(init_uts_ns.name.release),
294 .mode = 0444,
295 .proc_handler = &proc_do_uts_string,
296 .strategy = &sysctl_uts_string,
297 },
298 {
299 .ctl_name = KERN_VERSION,
300 .procname = "version",
301 .data = init_uts_ns.name.version,
302 .maxlen = sizeof(init_uts_ns.name.version),
303 .mode = 0444,
304 .proc_handler = &proc_do_uts_string,
305 .strategy = &sysctl_uts_string,
306 },
307 {
308 .ctl_name = KERN_NODENAME,
309 .procname = "hostname",
310 .data = init_uts_ns.name.nodename,
311 .maxlen = sizeof(init_uts_ns.name.nodename),
312 .mode = 0644,
313 .proc_handler = &proc_do_uts_string,
314 .strategy = &sysctl_uts_string,
315 },
316 {
317 .ctl_name = KERN_DOMAINNAME,
318 .procname = "domainname",
319 .data = init_uts_ns.name.domainname,
320 .maxlen = sizeof(init_uts_ns.name.domainname),
321 .mode = 0644,
322 .proc_handler = &proc_do_uts_string,
323 .strategy = &sysctl_uts_string,
324 },
325 {
326 .ctl_name = KERN_PANIC,
327 .procname = "panic",
328 .data = &panic_timeout,
329 .maxlen = sizeof(int),
330 .mode = 0644,
331 .proc_handler = &proc_dointvec,
332 },
333 {
334 .ctl_name = KERN_CORE_USES_PID,
335 .procname = "core_uses_pid",
336 .data = &core_uses_pid,
337 .maxlen = sizeof(int),
338 .mode = 0644,
339 .proc_handler = &proc_dointvec,
340 },
341 {
342 .ctl_name = KERN_CORE_PATTERN,
343 .procname = "core_pattern",
344 .data = core_pattern,
345 .maxlen = 128,
346 .mode = 0644,
347 .proc_handler = &proc_dostring,
348 .strategy = &sysctl_string,
349 },
350 #ifdef CONFIG_PROC_SYSCTL
351 {
352 .ctl_name = KERN_TAINTED,
353 .procname = "tainted",
354 .data = &tainted,
355 .maxlen = sizeof(int),
356 .mode = 0644,
357 .proc_handler = &proc_dointvec_taint,
358 },
359 #endif
360 {
361 .ctl_name = KERN_CAP_BSET,
362 .procname = "cap-bound",
363 .data = &cap_bset,
364 .maxlen = sizeof(kernel_cap_t),
365 .mode = 0600,
366 .proc_handler = &proc_dointvec_bset,
367 },
368 #ifdef CONFIG_BLK_DEV_INITRD
369 {
370 .ctl_name = KERN_REALROOTDEV,
371 .procname = "real-root-dev",
372 .data = &real_root_dev,
373 .maxlen = sizeof(int),
374 .mode = 0644,
375 .proc_handler = &proc_dointvec,
376 },
377 #endif
378 #ifdef __sparc__
379 {
380 .ctl_name = KERN_SPARC_REBOOT,
381 .procname = "reboot-cmd",
382 .data = reboot_command,
383 .maxlen = 256,
384 .mode = 0644,
385 .proc_handler = &proc_dostring,
386 .strategy = &sysctl_string,
387 },
388 {
389 .ctl_name = KERN_SPARC_STOP_A,
390 .procname = "stop-a",
391 .data = &stop_a_enabled,
392 .maxlen = sizeof (int),
393 .mode = 0644,
394 .proc_handler = &proc_dointvec,
395 },
396 {
397 .ctl_name = KERN_SPARC_SCONS_PWROFF,
398 .procname = "scons-poweroff",
399 .data = &scons_pwroff,
400 .maxlen = sizeof (int),
401 .mode = 0644,
402 .proc_handler = &proc_dointvec,
403 },
404 #endif
405 #ifdef __hppa__
406 {
407 .ctl_name = KERN_HPPA_PWRSW,
408 .procname = "soft-power",
409 .data = &pwrsw_enabled,
410 .maxlen = sizeof (int),
411 .mode = 0644,
412 .proc_handler = &proc_dointvec,
413 },
414 {
415 .ctl_name = KERN_HPPA_UNALIGNED,
416 .procname = "unaligned-trap",
417 .data = &unaligned_enabled,
418 .maxlen = sizeof (int),
419 .mode = 0644,
420 .proc_handler = &proc_dointvec,
421 },
422 #endif
423 {
424 .ctl_name = KERN_CTLALTDEL,
425 .procname = "ctrl-alt-del",
426 .data = &C_A_D,
427 .maxlen = sizeof(int),
428 .mode = 0644,
429 .proc_handler = &proc_dointvec,
430 },
431 {
432 .ctl_name = KERN_PRINTK,
433 .procname = "printk",
434 .data = &console_loglevel,
435 .maxlen = 4*sizeof(int),
436 .mode = 0644,
437 .proc_handler = &proc_dointvec,
438 },
439 #ifdef CONFIG_KMOD
440 {
441 .ctl_name = KERN_MODPROBE,
442 .procname = "modprobe",
443 .data = &modprobe_path,
444 .maxlen = KMOD_PATH_LEN,
445 .mode = 0644,
446 .proc_handler = &proc_dostring,
447 .strategy = &sysctl_string,
448 },
449 #endif
450 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
451 {
452 .ctl_name = KERN_HOTPLUG,
453 .procname = "hotplug",
454 .data = &uevent_helper,
455 .maxlen = UEVENT_HELPER_PATH_LEN,
456 .mode = 0644,
457 .proc_handler = &proc_dostring,
458 .strategy = &sysctl_string,
459 },
460 #endif
461 #ifdef CONFIG_CHR_DEV_SG
462 {
463 .ctl_name = KERN_SG_BIG_BUFF,
464 .procname = "sg-big-buff",
465 .data = &sg_big_buff,
466 .maxlen = sizeof (int),
467 .mode = 0444,
468 .proc_handler = &proc_dointvec,
469 },
470 #endif
471 #ifdef CONFIG_BSD_PROCESS_ACCT
472 {
473 .ctl_name = KERN_ACCT,
474 .procname = "acct",
475 .data = &acct_parm,
476 .maxlen = 3*sizeof(int),
477 .mode = 0644,
478 .proc_handler = &proc_dointvec,
479 },
480 #endif
481 #ifdef CONFIG_SYSVIPC
482 {
483 .ctl_name = KERN_SHMMAX,
484 .procname = "shmmax",
485 .data = &init_ipc_ns.shm_ctlmax,
486 .maxlen = sizeof (init_ipc_ns.shm_ctlmax),
487 .mode = 0644,
488 .proc_handler = &proc_ipc_doulongvec_minmax,
489 .strategy = sysctl_ipc_data,
490 },
491 {
492 .ctl_name = KERN_SHMALL,
493 .procname = "shmall",
494 .data = &init_ipc_ns.shm_ctlall,
495 .maxlen = sizeof (init_ipc_ns.shm_ctlall),
496 .mode = 0644,
497 .proc_handler = &proc_ipc_doulongvec_minmax,
498 .strategy = sysctl_ipc_data,
499 },
500 {
501 .ctl_name = KERN_SHMMNI,
502 .procname = "shmmni",
503 .data = &init_ipc_ns.shm_ctlmni,
504 .maxlen = sizeof (init_ipc_ns.shm_ctlmni),
505 .mode = 0644,
506 .proc_handler = &proc_ipc_dointvec,
507 .strategy = sysctl_ipc_data,
508 },
509 {
510 .ctl_name = KERN_MSGMAX,
511 .procname = "msgmax",
512 .data = &init_ipc_ns.msg_ctlmax,
513 .maxlen = sizeof (init_ipc_ns.msg_ctlmax),
514 .mode = 0644,
515 .proc_handler = &proc_ipc_dointvec,
516 .strategy = sysctl_ipc_data,
517 },
518 {
519 .ctl_name = KERN_MSGMNI,
520 .procname = "msgmni",
521 .data = &init_ipc_ns.msg_ctlmni,
522 .maxlen = sizeof (init_ipc_ns.msg_ctlmni),
523 .mode = 0644,
524 .proc_handler = &proc_ipc_dointvec,
525 .strategy = sysctl_ipc_data,
526 },
527 {
528 .ctl_name = KERN_MSGMNB,
529 .procname = "msgmnb",
530 .data = &init_ipc_ns.msg_ctlmnb,
531 .maxlen = sizeof (init_ipc_ns.msg_ctlmnb),
532 .mode = 0644,
533 .proc_handler = &proc_ipc_dointvec,
534 .strategy = sysctl_ipc_data,
535 },
536 {
537 .ctl_name = KERN_SEM,
538 .procname = "sem",
539 .data = &init_ipc_ns.sem_ctls,
540 .maxlen = 4*sizeof (int),
541 .mode = 0644,
542 .proc_handler = &proc_ipc_dointvec,
543 .strategy = sysctl_ipc_data,
544 },
545 #endif
546 #ifdef CONFIG_MAGIC_SYSRQ
547 {
548 .ctl_name = KERN_SYSRQ,
549 .procname = "sysrq",
550 .data = &__sysrq_enabled,
551 .maxlen = sizeof (int),
552 .mode = 0644,
553 .proc_handler = &proc_dointvec,
554 },
555 #endif
556 #ifdef CONFIG_PROC_SYSCTL
557 {
558 .ctl_name = KERN_CADPID,
559 .procname = "cad_pid",
560 .data = NULL,
561 .maxlen = sizeof (int),
562 .mode = 0600,
563 .proc_handler = &proc_do_cad_pid,
564 },
565 #endif
566 {
567 .ctl_name = KERN_MAX_THREADS,
568 .procname = "threads-max",
569 .data = &max_threads,
570 .maxlen = sizeof(int),
571 .mode = 0644,
572 .proc_handler = &proc_dointvec,
573 },
574 {
575 .ctl_name = KERN_RANDOM,
576 .procname = "random",
577 .mode = 0555,
578 .child = random_table,
579 },
580 #ifdef CONFIG_UNIX98_PTYS
581 {
582 .ctl_name = KERN_PTY,
583 .procname = "pty",
584 .mode = 0555,
585 .child = pty_table,
586 },
587 #endif
588 {
589 .ctl_name = KERN_OVERFLOWUID,
590 .procname = "overflowuid",
591 .data = &overflowuid,
592 .maxlen = sizeof(int),
593 .mode = 0644,
594 .proc_handler = &proc_dointvec_minmax,
595 .strategy = &sysctl_intvec,
596 .extra1 = &minolduid,
597 .extra2 = &maxolduid,
598 },
599 {
600 .ctl_name = KERN_OVERFLOWGID,
601 .procname = "overflowgid",
602 .data = &overflowgid,
603 .maxlen = sizeof(int),
604 .mode = 0644,
605 .proc_handler = &proc_dointvec_minmax,
606 .strategy = &sysctl_intvec,
607 .extra1 = &minolduid,
608 .extra2 = &maxolduid,
609 },
610 #ifdef CONFIG_S390
611 #ifdef CONFIG_MATHEMU
612 {
613 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
614 .procname = "ieee_emulation_warnings",
615 .data = &sysctl_ieee_emulation_warnings,
616 .maxlen = sizeof(int),
617 .mode = 0644,
618 .proc_handler = &proc_dointvec,
619 },
620 #endif
621 #ifdef CONFIG_NO_IDLE_HZ
622 {
623 .ctl_name = KERN_HZ_TIMER,
624 .procname = "hz_timer",
625 .data = &sysctl_hz_timer,
626 .maxlen = sizeof(int),
627 .mode = 0644,
628 .proc_handler = &proc_dointvec,
629 },
630 #endif
631 {
632 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
633 .procname = "userprocess_debug",
634 .data = &sysctl_userprocess_debug,
635 .maxlen = sizeof(int),
636 .mode = 0644,
637 .proc_handler = &proc_dointvec,
638 },
639 #endif
640 {
641 .ctl_name = KERN_PIDMAX,
642 .procname = "pid_max",
643 .data = &pid_max,
644 .maxlen = sizeof (int),
645 .mode = 0644,
646 .proc_handler = &proc_dointvec_minmax,
647 .strategy = sysctl_intvec,
648 .extra1 = &pid_max_min,
649 .extra2 = &pid_max_max,
650 },
651 {
652 .ctl_name = KERN_PANIC_ON_OOPS,
653 .procname = "panic_on_oops",
654 .data = &panic_on_oops,
655 .maxlen = sizeof(int),
656 .mode = 0644,
657 .proc_handler = &proc_dointvec,
658 },
659 {
660 .ctl_name = KERN_PRINTK_RATELIMIT,
661 .procname = "printk_ratelimit",
662 .data = &printk_ratelimit_jiffies,
663 .maxlen = sizeof(int),
664 .mode = 0644,
665 .proc_handler = &proc_dointvec_jiffies,
666 .strategy = &sysctl_jiffies,
667 },
668 {
669 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
670 .procname = "printk_ratelimit_burst",
671 .data = &printk_ratelimit_burst,
672 .maxlen = sizeof(int),
673 .mode = 0644,
674 .proc_handler = &proc_dointvec,
675 },
676 {
677 .ctl_name = KERN_NGROUPS_MAX,
678 .procname = "ngroups_max",
679 .data = &ngroups_max,
680 .maxlen = sizeof (int),
681 .mode = 0444,
682 .proc_handler = &proc_dointvec,
683 },
684 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
685 {
686 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
687 .procname = "unknown_nmi_panic",
688 .data = &unknown_nmi_panic,
689 .maxlen = sizeof (int),
690 .mode = 0644,
691 .proc_handler = &proc_dointvec,
692 },
693 {
694 .ctl_name = KERN_NMI_WATCHDOG,
695 .procname = "nmi_watchdog",
696 .data = &nmi_watchdog_enabled,
697 .maxlen = sizeof (int),
698 .mode = 0644,
699 .proc_handler = &proc_nmi_enabled,
700 },
701 #endif
702 #if defined(CONFIG_X86)
703 {
704 .ctl_name = KERN_PANIC_ON_NMI,
705 .procname = "panic_on_unrecovered_nmi",
706 .data = &panic_on_unrecovered_nmi,
707 .maxlen = sizeof(int),
708 .mode = 0644,
709 .proc_handler = &proc_dointvec,
710 },
711 {
712 .ctl_name = KERN_BOOTLOADER_TYPE,
713 .procname = "bootloader_type",
714 .data = &bootloader_type,
715 .maxlen = sizeof (int),
716 .mode = 0444,
717 .proc_handler = &proc_dointvec,
718 },
719 {
720 .ctl_name = CTL_UNNUMBERED,
721 .procname = "kstack_depth_to_print",
722 .data = &kstack_depth_to_print,
723 .maxlen = sizeof(int),
724 .mode = 0644,
725 .proc_handler = &proc_dointvec,
726 },
727 #endif
728 #if defined(CONFIG_MMU)
729 {
730 .ctl_name = KERN_RANDOMIZE,
731 .procname = "randomize_va_space",
732 .data = &randomize_va_space,
733 .maxlen = sizeof(int),
734 .mode = 0644,
735 .proc_handler = &proc_dointvec,
736 },
737 #endif
738 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
739 {
740 .ctl_name = KERN_SPIN_RETRY,
741 .procname = "spin_retry",
742 .data = &spin_retry,
743 .maxlen = sizeof (int),
744 .mode = 0644,
745 .proc_handler = &proc_dointvec,
746 },
747 #endif
748 #ifdef CONFIG_ACPI_SLEEP
749 {
750 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
751 .procname = "acpi_video_flags",
752 .data = &acpi_video_flags,
753 .maxlen = sizeof (unsigned long),
754 .mode = 0644,
755 .proc_handler = &proc_doulongvec_minmax,
756 },
757 #endif
758 #ifdef CONFIG_IA64
759 {
760 .ctl_name = KERN_IA64_UNALIGNED,
761 .procname = "ignore-unaligned-usertrap",
762 .data = &no_unaligned_warning,
763 .maxlen = sizeof (int),
764 .mode = 0644,
765 .proc_handler = &proc_dointvec,
766 },
767 #endif
768 #ifdef CONFIG_COMPAT
769 {
770 .ctl_name = KERN_COMPAT_LOG,
771 .procname = "compat-log",
772 .data = &compat_log,
773 .maxlen = sizeof (int),
774 .mode = 0644,
775 .proc_handler = &proc_dointvec,
776 },
777 #endif
778 #ifdef CONFIG_RT_MUTEXES
779 {
780 .ctl_name = KERN_MAX_LOCK_DEPTH,
781 .procname = "max_lock_depth",
782 .data = &max_lock_depth,
783 .maxlen = sizeof(int),
784 .mode = 0644,
785 .proc_handler = &proc_dointvec,
786 },
787 #endif
788
789 { .ctl_name = 0 }
790 };
791
792 /* Constants for minimum and maximum testing in vm_table.
793 We use these as one-element integer vectors. */
794 static int zero;
795 static int one_hundred = 100;
796
797
798 static ctl_table vm_table[] = {
799 {
800 .ctl_name = VM_OVERCOMMIT_MEMORY,
801 .procname = "overcommit_memory",
802 .data = &sysctl_overcommit_memory,
803 .maxlen = sizeof(sysctl_overcommit_memory),
804 .mode = 0644,
805 .proc_handler = &proc_dointvec,
806 },
807 {
808 .ctl_name = VM_PANIC_ON_OOM,
809 .procname = "panic_on_oom",
810 .data = &sysctl_panic_on_oom,
811 .maxlen = sizeof(sysctl_panic_on_oom),
812 .mode = 0644,
813 .proc_handler = &proc_dointvec,
814 },
815 {
816 .ctl_name = VM_OVERCOMMIT_RATIO,
817 .procname = "overcommit_ratio",
818 .data = &sysctl_overcommit_ratio,
819 .maxlen = sizeof(sysctl_overcommit_ratio),
820 .mode = 0644,
821 .proc_handler = &proc_dointvec,
822 },
823 {
824 .ctl_name = VM_PAGE_CLUSTER,
825 .procname = "page-cluster",
826 .data = &page_cluster,
827 .maxlen = sizeof(int),
828 .mode = 0644,
829 .proc_handler = &proc_dointvec,
830 },
831 {
832 .ctl_name = VM_DIRTY_BACKGROUND,
833 .procname = "dirty_background_ratio",
834 .data = &dirty_background_ratio,
835 .maxlen = sizeof(dirty_background_ratio),
836 .mode = 0644,
837 .proc_handler = &proc_dointvec_minmax,
838 .strategy = &sysctl_intvec,
839 .extra1 = &zero,
840 .extra2 = &one_hundred,
841 },
842 {
843 .ctl_name = VM_DIRTY_RATIO,
844 .procname = "dirty_ratio",
845 .data = &vm_dirty_ratio,
846 .maxlen = sizeof(vm_dirty_ratio),
847 .mode = 0644,
848 .proc_handler = &proc_dointvec_minmax,
849 .strategy = &sysctl_intvec,
850 .extra1 = &zero,
851 .extra2 = &one_hundred,
852 },
853 {
854 .ctl_name = VM_DIRTY_WB_CS,
855 .procname = "dirty_writeback_centisecs",
856 .data = &dirty_writeback_interval,
857 .maxlen = sizeof(dirty_writeback_interval),
858 .mode = 0644,
859 .proc_handler = &dirty_writeback_centisecs_handler,
860 },
861 {
862 .ctl_name = VM_DIRTY_EXPIRE_CS,
863 .procname = "dirty_expire_centisecs",
864 .data = &dirty_expire_interval,
865 .maxlen = sizeof(dirty_expire_interval),
866 .mode = 0644,
867 .proc_handler = &proc_dointvec_userhz_jiffies,
868 },
869 {
870 .ctl_name = VM_NR_PDFLUSH_THREADS,
871 .procname = "nr_pdflush_threads",
872 .data = &nr_pdflush_threads,
873 .maxlen = sizeof nr_pdflush_threads,
874 .mode = 0444 /* read-only*/,
875 .proc_handler = &proc_dointvec,
876 },
877 {
878 .ctl_name = VM_SWAPPINESS,
879 .procname = "swappiness",
880 .data = &vm_swappiness,
881 .maxlen = sizeof(vm_swappiness),
882 .mode = 0644,
883 .proc_handler = &proc_dointvec_minmax,
884 .strategy = &sysctl_intvec,
885 .extra1 = &zero,
886 .extra2 = &one_hundred,
887 },
888 #ifdef CONFIG_HUGETLB_PAGE
889 {
890 .ctl_name = VM_HUGETLB_PAGES,
891 .procname = "nr_hugepages",
892 .data = &max_huge_pages,
893 .maxlen = sizeof(unsigned long),
894 .mode = 0644,
895 .proc_handler = &hugetlb_sysctl_handler,
896 .extra1 = (void *)&hugetlb_zero,
897 .extra2 = (void *)&hugetlb_infinity,
898 },
899 {
900 .ctl_name = VM_HUGETLB_GROUP,
901 .procname = "hugetlb_shm_group",
902 .data = &sysctl_hugetlb_shm_group,
903 .maxlen = sizeof(gid_t),
904 .mode = 0644,
905 .proc_handler = &proc_dointvec,
906 },
907 #endif
908 {
909 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
910 .procname = "lowmem_reserve_ratio",
911 .data = &sysctl_lowmem_reserve_ratio,
912 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
913 .mode = 0644,
914 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
915 .strategy = &sysctl_intvec,
916 },
917 {
918 .ctl_name = VM_DROP_PAGECACHE,
919 .procname = "drop_caches",
920 .data = &sysctl_drop_caches,
921 .maxlen = sizeof(int),
922 .mode = 0644,
923 .proc_handler = drop_caches_sysctl_handler,
924 .strategy = &sysctl_intvec,
925 },
926 {
927 .ctl_name = VM_MIN_FREE_KBYTES,
928 .procname = "min_free_kbytes",
929 .data = &min_free_kbytes,
930 .maxlen = sizeof(min_free_kbytes),
931 .mode = 0644,
932 .proc_handler = &min_free_kbytes_sysctl_handler,
933 .strategy = &sysctl_intvec,
934 .extra1 = &zero,
935 },
936 {
937 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
938 .procname = "percpu_pagelist_fraction",
939 .data = &percpu_pagelist_fraction,
940 .maxlen = sizeof(percpu_pagelist_fraction),
941 .mode = 0644,
942 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
943 .strategy = &sysctl_intvec,
944 .extra1 = &min_percpu_pagelist_fract,
945 },
946 #ifdef CONFIG_MMU
947 {
948 .ctl_name = VM_MAX_MAP_COUNT,
949 .procname = "max_map_count",
950 .data = &sysctl_max_map_count,
951 .maxlen = sizeof(sysctl_max_map_count),
952 .mode = 0644,
953 .proc_handler = &proc_dointvec
954 },
955 #endif
956 {
957 .ctl_name = VM_LAPTOP_MODE,
958 .procname = "laptop_mode",
959 .data = &laptop_mode,
960 .maxlen = sizeof(laptop_mode),
961 .mode = 0644,
962 .proc_handler = &proc_dointvec_jiffies,
963 .strategy = &sysctl_jiffies,
964 },
965 {
966 .ctl_name = VM_BLOCK_DUMP,
967 .procname = "block_dump",
968 .data = &block_dump,
969 .maxlen = sizeof(block_dump),
970 .mode = 0644,
971 .proc_handler = &proc_dointvec,
972 .strategy = &sysctl_intvec,
973 .extra1 = &zero,
974 },
975 {
976 .ctl_name = VM_VFS_CACHE_PRESSURE,
977 .procname = "vfs_cache_pressure",
978 .data = &sysctl_vfs_cache_pressure,
979 .maxlen = sizeof(sysctl_vfs_cache_pressure),
980 .mode = 0644,
981 .proc_handler = &proc_dointvec,
982 .strategy = &sysctl_intvec,
983 .extra1 = &zero,
984 },
985 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
986 {
987 .ctl_name = VM_LEGACY_VA_LAYOUT,
988 .procname = "legacy_va_layout",
989 .data = &sysctl_legacy_va_layout,
990 .maxlen = sizeof(sysctl_legacy_va_layout),
991 .mode = 0644,
992 .proc_handler = &proc_dointvec,
993 .strategy = &sysctl_intvec,
994 .extra1 = &zero,
995 },
996 #endif
997 #ifdef CONFIG_NUMA
998 {
999 .ctl_name = VM_ZONE_RECLAIM_MODE,
1000 .procname = "zone_reclaim_mode",
1001 .data = &zone_reclaim_mode,
1002 .maxlen = sizeof(zone_reclaim_mode),
1003 .mode = 0644,
1004 .proc_handler = &proc_dointvec,
1005 .strategy = &sysctl_intvec,
1006 .extra1 = &zero,
1007 },
1008 {
1009 .ctl_name = VM_MIN_UNMAPPED,
1010 .procname = "min_unmapped_ratio",
1011 .data = &sysctl_min_unmapped_ratio,
1012 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1013 .mode = 0644,
1014 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
1015 .strategy = &sysctl_intvec,
1016 .extra1 = &zero,
1017 .extra2 = &one_hundred,
1018 },
1019 {
1020 .ctl_name = VM_MIN_SLAB,
1021 .procname = "min_slab_ratio",
1022 .data = &sysctl_min_slab_ratio,
1023 .maxlen = sizeof(sysctl_min_slab_ratio),
1024 .mode = 0644,
1025 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1026 .strategy = &sysctl_intvec,
1027 .extra1 = &zero,
1028 .extra2 = &one_hundred,
1029 },
1030 #endif
1031 #ifdef CONFIG_X86_32
1032 {
1033 .ctl_name = VM_VDSO_ENABLED,
1034 .procname = "vdso_enabled",
1035 .data = &vdso_enabled,
1036 .maxlen = sizeof(vdso_enabled),
1037 .mode = 0644,
1038 .proc_handler = &proc_dointvec,
1039 .strategy = &sysctl_intvec,
1040 .extra1 = &zero,
1041 },
1042 #endif
1043 { .ctl_name = 0 }
1044 };
1045
1046 static ctl_table fs_table[] = {
1047 {
1048 .ctl_name = FS_NRINODE,
1049 .procname = "inode-nr",
1050 .data = &inodes_stat,
1051 .maxlen = 2*sizeof(int),
1052 .mode = 0444,
1053 .proc_handler = &proc_dointvec,
1054 },
1055 {
1056 .ctl_name = FS_STATINODE,
1057 .procname = "inode-state",
1058 .data = &inodes_stat,
1059 .maxlen = 7*sizeof(int),
1060 .mode = 0444,
1061 .proc_handler = &proc_dointvec,
1062 },
1063 {
1064 .ctl_name = FS_NRFILE,
1065 .procname = "file-nr",
1066 .data = &files_stat,
1067 .maxlen = 3*sizeof(int),
1068 .mode = 0444,
1069 .proc_handler = &proc_nr_files,
1070 },
1071 {
1072 .ctl_name = FS_MAXFILE,
1073 .procname = "file-max",
1074 .data = &files_stat.max_files,
1075 .maxlen = sizeof(int),
1076 .mode = 0644,
1077 .proc_handler = &proc_dointvec,
1078 },
1079 {
1080 .ctl_name = FS_DENTRY,
1081 .procname = "dentry-state",
1082 .data = &dentry_stat,
1083 .maxlen = 6*sizeof(int),
1084 .mode = 0444,
1085 .proc_handler = &proc_dointvec,
1086 },
1087 {
1088 .ctl_name = FS_OVERFLOWUID,
1089 .procname = "overflowuid",
1090 .data = &fs_overflowuid,
1091 .maxlen = sizeof(int),
1092 .mode = 0644,
1093 .proc_handler = &proc_dointvec_minmax,
1094 .strategy = &sysctl_intvec,
1095 .extra1 = &minolduid,
1096 .extra2 = &maxolduid,
1097 },
1098 {
1099 .ctl_name = FS_OVERFLOWGID,
1100 .procname = "overflowgid",
1101 .data = &fs_overflowgid,
1102 .maxlen = sizeof(int),
1103 .mode = 0644,
1104 .proc_handler = &proc_dointvec_minmax,
1105 .strategy = &sysctl_intvec,
1106 .extra1 = &minolduid,
1107 .extra2 = &maxolduid,
1108 },
1109 {
1110 .ctl_name = FS_LEASES,
1111 .procname = "leases-enable",
1112 .data = &leases_enable,
1113 .maxlen = sizeof(int),
1114 .mode = 0644,
1115 .proc_handler = &proc_dointvec,
1116 },
1117 #ifdef CONFIG_DNOTIFY
1118 {
1119 .ctl_name = FS_DIR_NOTIFY,
1120 .procname = "dir-notify-enable",
1121 .data = &dir_notify_enable,
1122 .maxlen = sizeof(int),
1123 .mode = 0644,
1124 .proc_handler = &proc_dointvec,
1125 },
1126 #endif
1127 #ifdef CONFIG_MMU
1128 {
1129 .ctl_name = FS_LEASE_TIME,
1130 .procname = "lease-break-time",
1131 .data = &lease_break_time,
1132 .maxlen = sizeof(int),
1133 .mode = 0644,
1134 .proc_handler = &proc_dointvec,
1135 },
1136 {
1137 .ctl_name = FS_AIO_NR,
1138 .procname = "aio-nr",
1139 .data = &aio_nr,
1140 .maxlen = sizeof(aio_nr),
1141 .mode = 0444,
1142 .proc_handler = &proc_doulongvec_minmax,
1143 },
1144 {
1145 .ctl_name = FS_AIO_MAX_NR,
1146 .procname = "aio-max-nr",
1147 .data = &aio_max_nr,
1148 .maxlen = sizeof(aio_max_nr),
1149 .mode = 0644,
1150 .proc_handler = &proc_doulongvec_minmax,
1151 },
1152 #ifdef CONFIG_INOTIFY_USER
1153 {
1154 .ctl_name = FS_INOTIFY,
1155 .procname = "inotify",
1156 .mode = 0555,
1157 .child = inotify_table,
1158 },
1159 #endif
1160 #endif
1161 {
1162 .ctl_name = KERN_SETUID_DUMPABLE,
1163 .procname = "suid_dumpable",
1164 .data = &suid_dumpable,
1165 .maxlen = sizeof(int),
1166 .mode = 0644,
1167 .proc_handler = &proc_dointvec,
1168 },
1169 { .ctl_name = 0 }
1170 };
1171
1172 static ctl_table debug_table[] = {
1173 { .ctl_name = 0 }
1174 };
1175
1176 static ctl_table dev_table[] = {
1177 { .ctl_name = 0 }
1178 };
1179
1180 static DEFINE_SPINLOCK(sysctl_lock);
1181
1182 /* called under sysctl_lock */
1183 static int use_table(struct ctl_table_header *p)
1184 {
1185 if (unlikely(p->unregistering))
1186 return 0;
1187 p->used++;
1188 return 1;
1189 }
1190
1191 /* called under sysctl_lock */
1192 static void unuse_table(struct ctl_table_header *p)
1193 {
1194 if (!--p->used)
1195 if (unlikely(p->unregistering))
1196 complete(p->unregistering);
1197 }
1198
1199 /* called under sysctl_lock, will reacquire if has to wait */
1200 static void start_unregistering(struct ctl_table_header *p)
1201 {
1202 /*
1203 * if p->used is 0, nobody will ever touch that entry again;
1204 * we'll eliminate all paths to it before dropping sysctl_lock
1205 */
1206 if (unlikely(p->used)) {
1207 struct completion wait;
1208 init_completion(&wait);
1209 p->unregistering = &wait;
1210 spin_unlock(&sysctl_lock);
1211 wait_for_completion(&wait);
1212 spin_lock(&sysctl_lock);
1213 }
1214 /*
1215 * do not remove from the list until nobody holds it; walking the
1216 * list in do_sysctl() relies on that.
1217 */
1218 list_del_init(&p->ctl_entry);
1219 }
1220
1221 void __init sysctl_init(void)
1222 {
1223 #ifdef CONFIG_PROC_SYSCTL
1224 register_proc_table(root_table, proc_sys_root, &root_table_header);
1225 #endif
1226 }
1227
1228 #ifdef CONFIG_SYSCTL_SYSCALL
1229 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1230 void __user *newval, size_t newlen)
1231 {
1232 struct list_head *tmp;
1233 int error = -ENOTDIR;
1234
1235 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1236 return -ENOTDIR;
1237 if (oldval) {
1238 int old_len;
1239 if (!oldlenp || get_user(old_len, oldlenp))
1240 return -EFAULT;
1241 }
1242 spin_lock(&sysctl_lock);
1243 tmp = &root_table_header.ctl_entry;
1244 do {
1245 struct ctl_table_header *head =
1246 list_entry(tmp, struct ctl_table_header, ctl_entry);
1247
1248 if (!use_table(head))
1249 continue;
1250
1251 spin_unlock(&sysctl_lock);
1252
1253 error = parse_table(name, nlen, oldval, oldlenp,
1254 newval, newlen, head->ctl_table);
1255
1256 spin_lock(&sysctl_lock);
1257 unuse_table(head);
1258 if (error != -ENOTDIR)
1259 break;
1260 } while ((tmp = tmp->next) != &root_table_header.ctl_entry);
1261 spin_unlock(&sysctl_lock);
1262 return error;
1263 }
1264
1265 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1266 {
1267 struct __sysctl_args tmp;
1268 int error;
1269
1270 if (copy_from_user(&tmp, args, sizeof(tmp)))
1271 return -EFAULT;
1272
1273 lock_kernel();
1274 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1275 tmp.newval, tmp.newlen);
1276 unlock_kernel();
1277 return error;
1278 }
1279 #endif /* CONFIG_SYSCTL_SYSCALL */
1280
1281 /*
1282 * ctl_perm does NOT grant the superuser all rights automatically, because
1283 * some sysctl variables are readonly even to root.
1284 */
1285
1286 static int test_perm(int mode, int op)
1287 {
1288 if (!current->euid)
1289 mode >>= 6;
1290 else if (in_egroup_p(0))
1291 mode >>= 3;
1292 if ((mode & op & 0007) == op)
1293 return 0;
1294 return -EACCES;
1295 }
1296
1297 static inline int ctl_perm(ctl_table *table, int op)
1298 {
1299 int error;
1300 error = security_sysctl(table, op);
1301 if (error)
1302 return error;
1303 return test_perm(table->mode, op);
1304 }
1305
1306 #ifdef CONFIG_SYSCTL_SYSCALL
1307 static int parse_table(int __user *name, int nlen,
1308 void __user *oldval, size_t __user *oldlenp,
1309 void __user *newval, size_t newlen,
1310 ctl_table *table)
1311 {
1312 int n;
1313 repeat:
1314 if (!nlen)
1315 return -ENOTDIR;
1316 if (get_user(n, name))
1317 return -EFAULT;
1318 for ( ; table->ctl_name || table->procname; table++) {
1319 if (!table->ctl_name)
1320 continue;
1321 if (n == table->ctl_name || table->ctl_name == CTL_ANY) {
1322 int error;
1323 if (table->child) {
1324 if (ctl_perm(table, 001))
1325 return -EPERM;
1326 if (table->strategy) {
1327 error = table->strategy(
1328 table, name, nlen,
1329 oldval, oldlenp,
1330 newval, newlen);
1331 if (error)
1332 return error;
1333 }
1334 name++;
1335 nlen--;
1336 table = table->child;
1337 goto repeat;
1338 }
1339 error = do_sysctl_strategy(table, name, nlen,
1340 oldval, oldlenp,
1341 newval, newlen);
1342 return error;
1343 }
1344 }
1345 return -ENOTDIR;
1346 }
1347
1348 /* Perform the actual read/write of a sysctl table entry. */
1349 int do_sysctl_strategy (ctl_table *table,
1350 int __user *name, int nlen,
1351 void __user *oldval, size_t __user *oldlenp,
1352 void __user *newval, size_t newlen)
1353 {
1354 int op = 0, rc;
1355 size_t len;
1356
1357 if (oldval)
1358 op |= 004;
1359 if (newval)
1360 op |= 002;
1361 if (ctl_perm(table, op))
1362 return -EPERM;
1363
1364 if (table->strategy) {
1365 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1366 newval, newlen);
1367 if (rc < 0)
1368 return rc;
1369 if (rc > 0)
1370 return 0;
1371 }
1372
1373 /* If there is no strategy routine, or if the strategy returns
1374 * zero, proceed with automatic r/w */
1375 if (table->data && table->maxlen) {
1376 if (oldval && oldlenp) {
1377 if (get_user(len, oldlenp))
1378 return -EFAULT;
1379 if (len) {
1380 if (len > table->maxlen)
1381 len = table->maxlen;
1382 if(copy_to_user(oldval, table->data, len))
1383 return -EFAULT;
1384 if(put_user(len, oldlenp))
1385 return -EFAULT;
1386 }
1387 }
1388 if (newval && newlen) {
1389 len = newlen;
1390 if (len > table->maxlen)
1391 len = table->maxlen;
1392 if(copy_from_user(table->data, newval, len))
1393 return -EFAULT;
1394 }
1395 }
1396 return 0;
1397 }
1398 #endif /* CONFIG_SYSCTL_SYSCALL */
1399
1400 /**
1401 * register_sysctl_table - register a sysctl hierarchy
1402 * @table: the top-level table structure
1403 * @insert_at_head: whether the entry should be inserted in front or at the end
1404 *
1405 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1406 * array. An entry with a ctl_name of 0 terminates the table.
1407 *
1408 * The members of the &ctl_table structure are used as follows:
1409 *
1410 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1411 * must be unique within that level of sysctl
1412 *
1413 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1414 * enter a sysctl file
1415 *
1416 * data - a pointer to data for use by proc_handler
1417 *
1418 * maxlen - the maximum size in bytes of the data
1419 *
1420 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1421 *
1422 * child - a pointer to the child sysctl table if this entry is a directory, or
1423 * %NULL.
1424 *
1425 * proc_handler - the text handler routine (described below)
1426 *
1427 * strategy - the strategy routine (described below)
1428 *
1429 * de - for internal use by the sysctl routines
1430 *
1431 * extra1, extra2 - extra pointers usable by the proc handler routines
1432 *
1433 * Leaf nodes in the sysctl tree will be represented by a single file
1434 * under /proc; non-leaf nodes will be represented by directories.
1435 *
1436 * sysctl(2) can automatically manage read and write requests through
1437 * the sysctl table. The data and maxlen fields of the ctl_table
1438 * struct enable minimal validation of the values being written to be
1439 * performed, and the mode field allows minimal authentication.
1440 *
1441 * More sophisticated management can be enabled by the provision of a
1442 * strategy routine with the table entry. This will be called before
1443 * any automatic read or write of the data is performed.
1444 *
1445 * The strategy routine may return
1446 *
1447 * < 0 - Error occurred (error is passed to user process)
1448 *
1449 * 0 - OK - proceed with automatic read or write.
1450 *
1451 * > 0 - OK - read or write has been done by the strategy routine, so
1452 * return immediately.
1453 *
1454 * There must be a proc_handler routine for any terminal nodes
1455 * mirrored under /proc/sys (non-terminals are handled by a built-in
1456 * directory handler). Several default handlers are available to
1457 * cover common cases -
1458 *
1459 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1460 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1461 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1462 *
1463 * It is the handler's job to read the input buffer from user memory
1464 * and process it. The handler should return 0 on success.
1465 *
1466 * This routine returns %NULL on a failure to register, and a pointer
1467 * to the table header on success.
1468 */
1469 struct ctl_table_header *register_sysctl_table(ctl_table * table,
1470 int insert_at_head)
1471 {
1472 struct ctl_table_header *tmp;
1473 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1474 if (!tmp)
1475 return NULL;
1476 tmp->ctl_table = table;
1477 INIT_LIST_HEAD(&tmp->ctl_entry);
1478 tmp->used = 0;
1479 tmp->unregistering = NULL;
1480 spin_lock(&sysctl_lock);
1481 if (insert_at_head)
1482 list_add(&tmp->ctl_entry, &root_table_header.ctl_entry);
1483 else
1484 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
1485 spin_unlock(&sysctl_lock);
1486 #ifdef CONFIG_PROC_SYSCTL
1487 register_proc_table(table, proc_sys_root, tmp);
1488 #endif
1489 return tmp;
1490 }
1491
1492 /**
1493 * unregister_sysctl_table - unregister a sysctl table hierarchy
1494 * @header: the header returned from register_sysctl_table
1495 *
1496 * Unregisters the sysctl table and all children. proc entries may not
1497 * actually be removed until they are no longer used by anyone.
1498 */
1499 void unregister_sysctl_table(struct ctl_table_header * header)
1500 {
1501 might_sleep();
1502 spin_lock(&sysctl_lock);
1503 start_unregistering(header);
1504 #ifdef CONFIG_PROC_SYSCTL
1505 unregister_proc_table(header->ctl_table, proc_sys_root);
1506 #endif
1507 spin_unlock(&sysctl_lock);
1508 kfree(header);
1509 }
1510
1511 #else /* !CONFIG_SYSCTL */
1512 struct ctl_table_header * register_sysctl_table(ctl_table * table,
1513 int insert_at_head)
1514 {
1515 return NULL;
1516 }
1517
1518 void unregister_sysctl_table(struct ctl_table_header * table)
1519 {
1520 }
1521
1522 #endif /* CONFIG_SYSCTL */
1523
1524 /*
1525 * /proc/sys support
1526 */
1527
1528 #ifdef CONFIG_PROC_SYSCTL
1529
1530 /* Scan the sysctl entries in table and add them all into /proc */
1531 static void register_proc_table(ctl_table * table, struct proc_dir_entry *root, void *set)
1532 {
1533 struct proc_dir_entry *de;
1534 int len;
1535 mode_t mode;
1536
1537 for (; table->ctl_name || table->procname; table++) {
1538 /* Can't do anything without a proc name. */
1539 if (!table->procname)
1540 continue;
1541 /* Maybe we can't do anything with it... */
1542 if (!table->proc_handler && !table->child) {
1543 printk(KERN_WARNING "SYSCTL: Can't register %s\n",
1544 table->procname);
1545 continue;
1546 }
1547
1548 len = strlen(table->procname);
1549 mode = table->mode;
1550
1551 de = NULL;
1552 if (table->proc_handler)
1553 mode |= S_IFREG;
1554 else {
1555 mode |= S_IFDIR;
1556 for (de = root->subdir; de; de = de->next) {
1557 if (proc_match(len, table->procname, de))
1558 break;
1559 }
1560 /* If the subdir exists already, de is non-NULL */
1561 }
1562
1563 if (!de) {
1564 de = create_proc_entry(table->procname, mode, root);
1565 if (!de)
1566 continue;
1567 de->set = set;
1568 de->data = (void *) table;
1569 if (table->proc_handler)
1570 de->proc_fops = &proc_sys_file_operations;
1571 }
1572 table->de = de;
1573 if (de->mode & S_IFDIR)
1574 register_proc_table(table->child, de, set);
1575 }
1576 }
1577
1578 /*
1579 * Unregister a /proc sysctl table and any subdirectories.
1580 */
1581 static void unregister_proc_table(ctl_table * table, struct proc_dir_entry *root)
1582 {
1583 struct proc_dir_entry *de;
1584 for (; table->ctl_name || table->procname; table++) {
1585 if (!(de = table->de))
1586 continue;
1587 if (de->mode & S_IFDIR) {
1588 if (!table->child) {
1589 printk (KERN_ALERT "Help - malformed sysctl tree on free\n");
1590 continue;
1591 }
1592 unregister_proc_table(table->child, de);
1593
1594 /* Don't unregister directories which still have entries.. */
1595 if (de->subdir)
1596 continue;
1597 }
1598
1599 /*
1600 * In any case, mark the entry as goner; we'll keep it
1601 * around if it's busy, but we'll know to do nothing with
1602 * its fields. We are under sysctl_lock here.
1603 */
1604 de->data = NULL;
1605
1606 /* Don't unregister proc entries that are still being used.. */
1607 if (atomic_read(&de->count))
1608 continue;
1609
1610 table->de = NULL;
1611 remove_proc_entry(table->procname, root);
1612 }
1613 }
1614
1615 static ssize_t do_rw_proc(int write, struct file * file, char __user * buf,
1616 size_t count, loff_t *ppos)
1617 {
1618 int op;
1619 struct proc_dir_entry *de = PDE(file->f_path.dentry->d_inode);
1620 struct ctl_table *table;
1621 size_t res;
1622 ssize_t error = -ENOTDIR;
1623
1624 spin_lock(&sysctl_lock);
1625 if (de && de->data && use_table(de->set)) {
1626 /*
1627 * at that point we know that sysctl was not unregistered
1628 * and won't be until we finish
1629 */
1630 spin_unlock(&sysctl_lock);
1631 table = (struct ctl_table *) de->data;
1632 if (!table || !table->proc_handler)
1633 goto out;
1634 error = -EPERM;
1635 op = (write ? 002 : 004);
1636 if (ctl_perm(table, op))
1637 goto out;
1638
1639 /* careful: calling conventions are nasty here */
1640 res = count;
1641 error = (*table->proc_handler)(table, write, file,
1642 buf, &res, ppos);
1643 if (!error)
1644 error = res;
1645 out:
1646 spin_lock(&sysctl_lock);
1647 unuse_table(de->set);
1648 }
1649 spin_unlock(&sysctl_lock);
1650 return error;
1651 }
1652
1653 static int proc_opensys(struct inode *inode, struct file *file)
1654 {
1655 if (file->f_mode & FMODE_WRITE) {
1656 /*
1657 * sysctl entries that are not writable,
1658 * are _NOT_ writable, capabilities or not.
1659 */
1660 if (!(inode->i_mode & S_IWUSR))
1661 return -EPERM;
1662 }
1663
1664 return 0;
1665 }
1666
1667 static ssize_t proc_readsys(struct file * file, char __user * buf,
1668 size_t count, loff_t *ppos)
1669 {
1670 return do_rw_proc(0, file, buf, count, ppos);
1671 }
1672
1673 static ssize_t proc_writesys(struct file * file, const char __user * buf,
1674 size_t count, loff_t *ppos)
1675 {
1676 return do_rw_proc(1, file, (char __user *) buf, count, ppos);
1677 }
1678
1679 static int _proc_do_string(void* data, int maxlen, int write,
1680 struct file *filp, void __user *buffer,
1681 size_t *lenp, loff_t *ppos)
1682 {
1683 size_t len;
1684 char __user *p;
1685 char c;
1686
1687 if (!data || !maxlen || !*lenp) {
1688 *lenp = 0;
1689 return 0;
1690 }
1691
1692 if (write) {
1693 len = 0;
1694 p = buffer;
1695 while (len < *lenp) {
1696 if (get_user(c, p++))
1697 return -EFAULT;
1698 if (c == 0 || c == '\n')
1699 break;
1700 len++;
1701 }
1702 if (len >= maxlen)
1703 len = maxlen-1;
1704 if(copy_from_user(data, buffer, len))
1705 return -EFAULT;
1706 ((char *) data)[len] = 0;
1707 *ppos += *lenp;
1708 } else {
1709 len = strlen(data);
1710 if (len > maxlen)
1711 len = maxlen;
1712
1713 if (*ppos > len) {
1714 *lenp = 0;
1715 return 0;
1716 }
1717
1718 data += *ppos;
1719 len -= *ppos;
1720
1721 if (len > *lenp)
1722 len = *lenp;
1723 if (len)
1724 if(copy_to_user(buffer, data, len))
1725 return -EFAULT;
1726 if (len < *lenp) {
1727 if(put_user('\n', ((char __user *) buffer) + len))
1728 return -EFAULT;
1729 len++;
1730 }
1731 *lenp = len;
1732 *ppos += len;
1733 }
1734 return 0;
1735 }
1736
1737 /**
1738 * proc_dostring - read a string sysctl
1739 * @table: the sysctl table
1740 * @write: %TRUE if this is a write to the sysctl file
1741 * @filp: the file structure
1742 * @buffer: the user buffer
1743 * @lenp: the size of the user buffer
1744 * @ppos: file position
1745 *
1746 * Reads/writes a string from/to the user buffer. If the kernel
1747 * buffer provided is not large enough to hold the string, the
1748 * string is truncated. The copied string is %NULL-terminated.
1749 * If the string is being read by the user process, it is copied
1750 * and a newline '\n' is added. It is truncated if the buffer is
1751 * not large enough.
1752 *
1753 * Returns 0 on success.
1754 */
1755 int proc_dostring(ctl_table *table, int write, struct file *filp,
1756 void __user *buffer, size_t *lenp, loff_t *ppos)
1757 {
1758 return _proc_do_string(table->data, table->maxlen, write, filp,
1759 buffer, lenp, ppos);
1760 }
1761
1762 /*
1763 * Special case of dostring for the UTS structure. This has locks
1764 * to observe. Should this be in kernel/sys.c ????
1765 */
1766
1767 static int proc_do_uts_string(ctl_table *table, int write, struct file *filp,
1768 void __user *buffer, size_t *lenp, loff_t *ppos)
1769 {
1770 int r;
1771 void *which;
1772 which = get_uts(table, write);
1773 r = _proc_do_string(which, table->maxlen,write,filp,buffer,lenp, ppos);
1774 put_uts(table, write, which);
1775 return r;
1776 }
1777
1778 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1779 int *valp,
1780 int write, void *data)
1781 {
1782 if (write) {
1783 *valp = *negp ? -*lvalp : *lvalp;
1784 } else {
1785 int val = *valp;
1786 if (val < 0) {
1787 *negp = -1;
1788 *lvalp = (unsigned long)-val;
1789 } else {
1790 *negp = 0;
1791 *lvalp = (unsigned long)val;
1792 }
1793 }
1794 return 0;
1795 }
1796
1797 static int __do_proc_dointvec(void *tbl_data, ctl_table *table,
1798 int write, struct file *filp, void __user *buffer,
1799 size_t *lenp, loff_t *ppos,
1800 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1801 int write, void *data),
1802 void *data)
1803 {
1804 #define TMPBUFLEN 21
1805 int *i, vleft, first=1, neg, val;
1806 unsigned long lval;
1807 size_t left, len;
1808
1809 char buf[TMPBUFLEN], *p;
1810 char __user *s = buffer;
1811
1812 if (!tbl_data || !table->maxlen || !*lenp ||
1813 (*ppos && !write)) {
1814 *lenp = 0;
1815 return 0;
1816 }
1817
1818 i = (int *) tbl_data;
1819 vleft = table->maxlen / sizeof(*i);
1820 left = *lenp;
1821
1822 if (!conv)
1823 conv = do_proc_dointvec_conv;
1824
1825 for (; left && vleft--; i++, first=0) {
1826 if (write) {
1827 while (left) {
1828 char c;
1829 if (get_user(c, s))
1830 return -EFAULT;
1831 if (!isspace(c))
1832 break;
1833 left--;
1834 s++;
1835 }
1836 if (!left)
1837 break;
1838 neg = 0;
1839 len = left;
1840 if (len > sizeof(buf) - 1)
1841 len = sizeof(buf) - 1;
1842 if (copy_from_user(buf, s, len))
1843 return -EFAULT;
1844 buf[len] = 0;
1845 p = buf;
1846 if (*p == '-' && left > 1) {
1847 neg = 1;
1848 p++;
1849 }
1850 if (*p < '0' || *p > '9')
1851 break;
1852
1853 lval = simple_strtoul(p, &p, 0);
1854
1855 len = p-buf;
1856 if ((len < left) && *p && !isspace(*p))
1857 break;
1858 if (neg)
1859 val = -val;
1860 s += len;
1861 left -= len;
1862
1863 if (conv(&neg, &lval, i, 1, data))
1864 break;
1865 } else {
1866 p = buf;
1867 if (!first)
1868 *p++ = '\t';
1869
1870 if (conv(&neg, &lval, i, 0, data))
1871 break;
1872
1873 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1874 len = strlen(buf);
1875 if (len > left)
1876 len = left;
1877 if(copy_to_user(s, buf, len))
1878 return -EFAULT;
1879 left -= len;
1880 s += len;
1881 }
1882 }
1883
1884 if (!write && !first && left) {
1885 if(put_user('\n', s))
1886 return -EFAULT;
1887 left--, s++;
1888 }
1889 if (write) {
1890 while (left) {
1891 char c;
1892 if (get_user(c, s++))
1893 return -EFAULT;
1894 if (!isspace(c))
1895 break;
1896 left--;
1897 }
1898 }
1899 if (write && first)
1900 return -EINVAL;
1901 *lenp -= left;
1902 *ppos += *lenp;
1903 return 0;
1904 #undef TMPBUFLEN
1905 }
1906
1907 static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1908 void __user *buffer, size_t *lenp, loff_t *ppos,
1909 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1910 int write, void *data),
1911 void *data)
1912 {
1913 return __do_proc_dointvec(table->data, table, write, filp,
1914 buffer, lenp, ppos, conv, data);
1915 }
1916
1917 /**
1918 * proc_dointvec - read a vector of integers
1919 * @table: the sysctl table
1920 * @write: %TRUE if this is a write to the sysctl file
1921 * @filp: the file structure
1922 * @buffer: the user buffer
1923 * @lenp: the size of the user buffer
1924 * @ppos: file position
1925 *
1926 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1927 * values from/to the user buffer, treated as an ASCII string.
1928 *
1929 * Returns 0 on success.
1930 */
1931 int proc_dointvec(ctl_table *table, int write, struct file *filp,
1932 void __user *buffer, size_t *lenp, loff_t *ppos)
1933 {
1934 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1935 NULL,NULL);
1936 }
1937
1938 #define OP_SET 0
1939 #define OP_AND 1
1940 #define OP_OR 2
1941
1942 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1943 int *valp,
1944 int write, void *data)
1945 {
1946 int op = *(int *)data;
1947 if (write) {
1948 int val = *negp ? -*lvalp : *lvalp;
1949 switch(op) {
1950 case OP_SET: *valp = val; break;
1951 case OP_AND: *valp &= val; break;
1952 case OP_OR: *valp |= val; break;
1953 }
1954 } else {
1955 int val = *valp;
1956 if (val < 0) {
1957 *negp = -1;
1958 *lvalp = (unsigned long)-val;
1959 } else {
1960 *negp = 0;
1961 *lvalp = (unsigned long)val;
1962 }
1963 }
1964 return 0;
1965 }
1966
1967 /*
1968 * init may raise the set.
1969 */
1970
1971 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1972 void __user *buffer, size_t *lenp, loff_t *ppos)
1973 {
1974 int op;
1975
1976 if (write && !capable(CAP_SYS_MODULE)) {
1977 return -EPERM;
1978 }
1979
1980 op = is_init(current) ? OP_SET : OP_AND;
1981 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1982 do_proc_dointvec_bset_conv,&op);
1983 }
1984
1985 /*
1986 * Taint values can only be increased
1987 */
1988 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
1989 void __user *buffer, size_t *lenp, loff_t *ppos)
1990 {
1991 int op;
1992
1993 if (!capable(CAP_SYS_ADMIN))
1994 return -EPERM;
1995
1996 op = OP_OR;
1997 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1998 do_proc_dointvec_bset_conv,&op);
1999 }
2000
2001 struct do_proc_dointvec_minmax_conv_param {
2002 int *min;
2003 int *max;
2004 };
2005
2006 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2007 int *valp,
2008 int write, void *data)
2009 {
2010 struct do_proc_dointvec_minmax_conv_param *param = data;
2011 if (write) {
2012 int val = *negp ? -*lvalp : *lvalp;
2013 if ((param->min && *param->min > val) ||
2014 (param->max && *param->max < val))
2015 return -EINVAL;
2016 *valp = val;
2017 } else {
2018 int val = *valp;
2019 if (val < 0) {
2020 *negp = -1;
2021 *lvalp = (unsigned long)-val;
2022 } else {
2023 *negp = 0;
2024 *lvalp = (unsigned long)val;
2025 }
2026 }
2027 return 0;
2028 }
2029
2030 /**
2031 * proc_dointvec_minmax - read a vector of integers with min/max values
2032 * @table: the sysctl table
2033 * @write: %TRUE if this is a write to the sysctl file
2034 * @filp: the file structure
2035 * @buffer: the user buffer
2036 * @lenp: the size of the user buffer
2037 * @ppos: file position
2038 *
2039 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2040 * values from/to the user buffer, treated as an ASCII string.
2041 *
2042 * This routine will ensure the values are within the range specified by
2043 * table->extra1 (min) and table->extra2 (max).
2044 *
2045 * Returns 0 on success.
2046 */
2047 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2048 void __user *buffer, size_t *lenp, loff_t *ppos)
2049 {
2050 struct do_proc_dointvec_minmax_conv_param param = {
2051 .min = (int *) table->extra1,
2052 .max = (int *) table->extra2,
2053 };
2054 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2055 do_proc_dointvec_minmax_conv, &param);
2056 }
2057
2058 static int __do_proc_doulongvec_minmax(void *data, ctl_table *table, int write,
2059 struct file *filp,
2060 void __user *buffer,
2061 size_t *lenp, loff_t *ppos,
2062 unsigned long convmul,
2063 unsigned long convdiv)
2064 {
2065 #define TMPBUFLEN 21
2066 unsigned long *i, *min, *max, val;
2067 int vleft, first=1, neg;
2068 size_t len, left;
2069 char buf[TMPBUFLEN], *p;
2070 char __user *s = buffer;
2071
2072 if (!data || !table->maxlen || !*lenp ||
2073 (*ppos && !write)) {
2074 *lenp = 0;
2075 return 0;
2076 }
2077
2078 i = (unsigned long *) data;
2079 min = (unsigned long *) table->extra1;
2080 max = (unsigned long *) table->extra2;
2081 vleft = table->maxlen / sizeof(unsigned long);
2082 left = *lenp;
2083
2084 for (; left && vleft--; i++, min++, max++, first=0) {
2085 if (write) {
2086 while (left) {
2087 char c;
2088 if (get_user(c, s))
2089 return -EFAULT;
2090 if (!isspace(c))
2091 break;
2092 left--;
2093 s++;
2094 }
2095 if (!left)
2096 break;
2097 neg = 0;
2098 len = left;
2099 if (len > TMPBUFLEN-1)
2100 len = TMPBUFLEN-1;
2101 if (copy_from_user(buf, s, len))
2102 return -EFAULT;
2103 buf[len] = 0;
2104 p = buf;
2105 if (*p == '-' && left > 1) {
2106 neg = 1;
2107 p++;
2108 }
2109 if (*p < '0' || *p > '9')
2110 break;
2111 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2112 len = p-buf;
2113 if ((len < left) && *p && !isspace(*p))
2114 break;
2115 if (neg)
2116 val = -val;
2117 s += len;
2118 left -= len;
2119
2120 if(neg)
2121 continue;
2122 if ((min && val < *min) || (max && val > *max))
2123 continue;
2124 *i = val;
2125 } else {
2126 p = buf;
2127 if (!first)
2128 *p++ = '\t';
2129 sprintf(p, "%lu", convdiv * (*i) / convmul);
2130 len = strlen(buf);
2131 if (len > left)
2132 len = left;
2133 if(copy_to_user(s, buf, len))
2134 return -EFAULT;
2135 left -= len;
2136 s += len;
2137 }
2138 }
2139
2140 if (!write && !first && left) {
2141 if(put_user('\n', s))
2142 return -EFAULT;
2143 left--, s++;
2144 }
2145 if (write) {
2146 while (left) {
2147 char c;
2148 if (get_user(c, s++))
2149 return -EFAULT;
2150 if (!isspace(c))
2151 break;
2152 left--;
2153 }
2154 }
2155 if (write && first)
2156 return -EINVAL;
2157 *lenp -= left;
2158 *ppos += *lenp;
2159 return 0;
2160 #undef TMPBUFLEN
2161 }
2162
2163 static int do_proc_doulongvec_minmax(ctl_table *table, int write,
2164 struct file *filp,
2165 void __user *buffer,
2166 size_t *lenp, loff_t *ppos,
2167 unsigned long convmul,
2168 unsigned long convdiv)
2169 {
2170 return __do_proc_doulongvec_minmax(table->data, table, write,
2171 filp, buffer, lenp, ppos, convmul, convdiv);
2172 }
2173
2174 /**
2175 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2176 * @table: the sysctl table
2177 * @write: %TRUE if this is a write to the sysctl file
2178 * @filp: the file structure
2179 * @buffer: the user buffer
2180 * @lenp: the size of the user buffer
2181 * @ppos: file position
2182 *
2183 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2184 * values from/to the user buffer, treated as an ASCII string.
2185 *
2186 * This routine will ensure the values are within the range specified by
2187 * table->extra1 (min) and table->extra2 (max).
2188 *
2189 * Returns 0 on success.
2190 */
2191 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2192 void __user *buffer, size_t *lenp, loff_t *ppos)
2193 {
2194 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2195 }
2196
2197 /**
2198 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2199 * @table: the sysctl table
2200 * @write: %TRUE if this is a write to the sysctl file
2201 * @filp: the file structure
2202 * @buffer: the user buffer
2203 * @lenp: the size of the user buffer
2204 * @ppos: file position
2205 *
2206 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2207 * values from/to the user buffer, treated as an ASCII string. The values
2208 * are treated as milliseconds, and converted to jiffies when they are stored.
2209 *
2210 * This routine will ensure the values are within the range specified by
2211 * table->extra1 (min) and table->extra2 (max).
2212 *
2213 * Returns 0 on success.
2214 */
2215 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2216 struct file *filp,
2217 void __user *buffer,
2218 size_t *lenp, loff_t *ppos)
2219 {
2220 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2221 lenp, ppos, HZ, 1000l);
2222 }
2223
2224
2225 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2226 int *valp,
2227 int write, void *data)
2228 {
2229 if (write) {
2230 if (*lvalp > LONG_MAX / HZ)
2231 return 1;
2232 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2233 } else {
2234 int val = *valp;
2235 unsigned long lval;
2236 if (val < 0) {
2237 *negp = -1;
2238 lval = (unsigned long)-val;
2239 } else {
2240 *negp = 0;
2241 lval = (unsigned long)val;
2242 }
2243 *lvalp = lval / HZ;
2244 }
2245 return 0;
2246 }
2247
2248 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2249 int *valp,
2250 int write, void *data)
2251 {
2252 if (write) {
2253 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2254 return 1;
2255 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2256 } else {
2257 int val = *valp;
2258 unsigned long lval;
2259 if (val < 0) {
2260 *negp = -1;
2261 lval = (unsigned long)-val;
2262 } else {
2263 *negp = 0;
2264 lval = (unsigned long)val;
2265 }
2266 *lvalp = jiffies_to_clock_t(lval);
2267 }
2268 return 0;
2269 }
2270
2271 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2272 int *valp,
2273 int write, void *data)
2274 {
2275 if (write) {
2276 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2277 } else {
2278 int val = *valp;
2279 unsigned long lval;
2280 if (val < 0) {
2281 *negp = -1;
2282 lval = (unsigned long)-val;
2283 } else {
2284 *negp = 0;
2285 lval = (unsigned long)val;
2286 }
2287 *lvalp = jiffies_to_msecs(lval);
2288 }
2289 return 0;
2290 }
2291
2292 /**
2293 * proc_dointvec_jiffies - read a vector of integers as seconds
2294 * @table: the sysctl table
2295 * @write: %TRUE if this is a write to the sysctl file
2296 * @filp: the file structure
2297 * @buffer: the user buffer
2298 * @lenp: the size of the user buffer
2299 * @ppos: file position
2300 *
2301 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2302 * values from/to the user buffer, treated as an ASCII string.
2303 * The values read are assumed to be in seconds, and are converted into
2304 * jiffies.
2305 *
2306 * Returns 0 on success.
2307 */
2308 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2309 void __user *buffer, size_t *lenp, loff_t *ppos)
2310 {
2311 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2312 do_proc_dointvec_jiffies_conv,NULL);
2313 }
2314
2315 /**
2316 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2317 * @table: the sysctl table
2318 * @write: %TRUE if this is a write to the sysctl file
2319 * @filp: the file structure
2320 * @buffer: the user buffer
2321 * @lenp: the size of the user buffer
2322 * @ppos: pointer to the file position
2323 *
2324 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2325 * values from/to the user buffer, treated as an ASCII string.
2326 * The values read are assumed to be in 1/USER_HZ seconds, and
2327 * are converted into jiffies.
2328 *
2329 * Returns 0 on success.
2330 */
2331 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2332 void __user *buffer, size_t *lenp, loff_t *ppos)
2333 {
2334 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2335 do_proc_dointvec_userhz_jiffies_conv,NULL);
2336 }
2337
2338 /**
2339 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2340 * @table: the sysctl table
2341 * @write: %TRUE if this is a write to the sysctl file
2342 * @filp: the file structure
2343 * @buffer: the user buffer
2344 * @lenp: the size of the user buffer
2345 * @ppos: file position
2346 * @ppos: the current position in the file
2347 *
2348 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2349 * values from/to the user buffer, treated as an ASCII string.
2350 * The values read are assumed to be in 1/1000 seconds, and
2351 * are converted into jiffies.
2352 *
2353 * Returns 0 on success.
2354 */
2355 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2356 void __user *buffer, size_t *lenp, loff_t *ppos)
2357 {
2358 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2359 do_proc_dointvec_ms_jiffies_conv, NULL);
2360 }
2361
2362 #ifdef CONFIG_SYSVIPC
2363 static int proc_ipc_dointvec(ctl_table *table, int write, struct file *filp,
2364 void __user *buffer, size_t *lenp, loff_t *ppos)
2365 {
2366 void *which;
2367 which = get_ipc(table, write);
2368 return __do_proc_dointvec(which, table, write, filp, buffer,
2369 lenp, ppos, NULL, NULL);
2370 }
2371
2372 static int proc_ipc_doulongvec_minmax(ctl_table *table, int write,
2373 struct file *filp, void __user *buffer, size_t *lenp, loff_t *ppos)
2374 {
2375 void *which;
2376 which = get_ipc(table, write);
2377 return __do_proc_doulongvec_minmax(which, table, write, filp, buffer,
2378 lenp, ppos, 1l, 1l);
2379 }
2380
2381 #endif
2382
2383 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
2384 void __user *buffer, size_t *lenp, loff_t *ppos)
2385 {
2386 struct pid *new_pid;
2387 pid_t tmp;
2388 int r;
2389
2390 tmp = pid_nr(cad_pid);
2391
2392 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2393 lenp, ppos, NULL, NULL);
2394 if (r || !write)
2395 return r;
2396
2397 new_pid = find_get_pid(tmp);
2398 if (!new_pid)
2399 return -ESRCH;
2400
2401 put_pid(xchg(&cad_pid, new_pid));
2402 return 0;
2403 }
2404
2405 #else /* CONFIG_PROC_FS */
2406
2407 int proc_dostring(ctl_table *table, int write, struct file *filp,
2408 void __user *buffer, size_t *lenp, loff_t *ppos)
2409 {
2410 return -ENOSYS;
2411 }
2412
2413 static int proc_do_uts_string(ctl_table *table, int write, struct file *filp,
2414 void __user *buffer, size_t *lenp, loff_t *ppos)
2415 {
2416 return -ENOSYS;
2417 }
2418
2419 #ifdef CONFIG_SYSVIPC
2420 static int proc_do_ipc_string(ctl_table *table, int write, struct file *filp,
2421 void __user *buffer, size_t *lenp, loff_t *ppos)
2422 {
2423 return -ENOSYS;
2424 }
2425 static int proc_ipc_dointvec(ctl_table *table, int write, struct file *filp,
2426 void __user *buffer, size_t *lenp, loff_t *ppos)
2427 {
2428 return -ENOSYS;
2429 }
2430 static int proc_ipc_doulongvec_minmax(ctl_table *table, int write,
2431 struct file *filp, void __user *buffer,
2432 size_t *lenp, loff_t *ppos)
2433 {
2434 return -ENOSYS;
2435 }
2436 #endif
2437
2438 int proc_dointvec(ctl_table *table, int write, struct file *filp,
2439 void __user *buffer, size_t *lenp, loff_t *ppos)
2440 {
2441 return -ENOSYS;
2442 }
2443
2444 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2445 void __user *buffer, size_t *lenp, loff_t *ppos)
2446 {
2447 return -ENOSYS;
2448 }
2449
2450 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2451 void __user *buffer, size_t *lenp, loff_t *ppos)
2452 {
2453 return -ENOSYS;
2454 }
2455
2456 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2457 void __user *buffer, size_t *lenp, loff_t *ppos)
2458 {
2459 return -ENOSYS;
2460 }
2461
2462 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2463 void __user *buffer, size_t *lenp, loff_t *ppos)
2464 {
2465 return -ENOSYS;
2466 }
2467
2468 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2469 void __user *buffer, size_t *lenp, loff_t *ppos)
2470 {
2471 return -ENOSYS;
2472 }
2473
2474 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2475 void __user *buffer, size_t *lenp, loff_t *ppos)
2476 {
2477 return -ENOSYS;
2478 }
2479
2480 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2481 struct file *filp,
2482 void __user *buffer,
2483 size_t *lenp, loff_t *ppos)
2484 {
2485 return -ENOSYS;
2486 }
2487
2488
2489 #endif /* CONFIG_PROC_FS */
2490
2491
2492 #ifdef CONFIG_SYSCTL_SYSCALL
2493 /*
2494 * General sysctl support routines
2495 */
2496
2497 /* The generic string strategy routine: */
2498 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2499 void __user *oldval, size_t __user *oldlenp,
2500 void __user *newval, size_t newlen)
2501 {
2502 if (!table->data || !table->maxlen)
2503 return -ENOTDIR;
2504
2505 if (oldval && oldlenp) {
2506 size_t bufsize;
2507 if (get_user(bufsize, oldlenp))
2508 return -EFAULT;
2509 if (bufsize) {
2510 size_t len = strlen(table->data), copied;
2511
2512 /* This shouldn't trigger for a well-formed sysctl */
2513 if (len > table->maxlen)
2514 len = table->maxlen;
2515
2516 /* Copy up to a max of bufsize-1 bytes of the string */
2517 copied = (len >= bufsize) ? bufsize - 1 : len;
2518
2519 if (copy_to_user(oldval, table->data, copied) ||
2520 put_user(0, (char __user *)(oldval + copied)))
2521 return -EFAULT;
2522 if (put_user(len, oldlenp))
2523 return -EFAULT;
2524 }
2525 }
2526 if (newval && newlen) {
2527 size_t len = newlen;
2528 if (len > table->maxlen)
2529 len = table->maxlen;
2530 if(copy_from_user(table->data, newval, len))
2531 return -EFAULT;
2532 if (len == table->maxlen)
2533 len--;
2534 ((char *) table->data)[len] = 0;
2535 }
2536 return 1;
2537 }
2538
2539 /*
2540 * This function makes sure that all of the integers in the vector
2541 * are between the minimum and maximum values given in the arrays
2542 * table->extra1 and table->extra2, respectively.
2543 */
2544 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2545 void __user *oldval, size_t __user *oldlenp,
2546 void __user *newval, size_t newlen)
2547 {
2548
2549 if (newval && newlen) {
2550 int __user *vec = (int __user *) newval;
2551 int *min = (int *) table->extra1;
2552 int *max = (int *) table->extra2;
2553 size_t length;
2554 int i;
2555
2556 if (newlen % sizeof(int) != 0)
2557 return -EINVAL;
2558
2559 if (!table->extra1 && !table->extra2)
2560 return 0;
2561
2562 if (newlen > table->maxlen)
2563 newlen = table->maxlen;
2564 length = newlen / sizeof(int);
2565
2566 for (i = 0; i < length; i++) {
2567 int value;
2568 if (get_user(value, vec + i))
2569 return -EFAULT;
2570 if (min && value < min[i])
2571 return -EINVAL;
2572 if (max && value > max[i])
2573 return -EINVAL;
2574 }
2575 }
2576 return 0;
2577 }
2578
2579 /* Strategy function to convert jiffies to seconds */
2580 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2581 void __user *oldval, size_t __user *oldlenp,
2582 void __user *newval, size_t newlen)
2583 {
2584 if (oldval && oldlenp) {
2585 size_t olen;
2586
2587 if (get_user(olen, oldlenp))
2588 return -EFAULT;
2589 if (olen) {
2590 int val;
2591
2592 if (olen < sizeof(int))
2593 return -EINVAL;
2594
2595 val = *(int *)(table->data) / HZ;
2596 if (put_user(val, (int __user *)oldval))
2597 return -EFAULT;
2598 if (put_user(sizeof(int), oldlenp))
2599 return -EFAULT;
2600 }
2601 }
2602 if (newval && newlen) {
2603 int new;
2604 if (newlen != sizeof(int))
2605 return -EINVAL;
2606 if (get_user(new, (int __user *)newval))
2607 return -EFAULT;
2608 *(int *)(table->data) = new*HZ;
2609 }
2610 return 1;
2611 }
2612
2613 /* Strategy function to convert jiffies to seconds */
2614 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2615 void __user *oldval, size_t __user *oldlenp,
2616 void __user *newval, size_t newlen)
2617 {
2618 if (oldval && oldlenp) {
2619 size_t olen;
2620
2621 if (get_user(olen, oldlenp))
2622 return -EFAULT;
2623 if (olen) {
2624 int val;
2625
2626 if (olen < sizeof(int))
2627 return -EINVAL;
2628
2629 val = jiffies_to_msecs(*(int *)(table->data));
2630 if (put_user(val, (int __user *)oldval))
2631 return -EFAULT;
2632 if (put_user(sizeof(int), oldlenp))
2633 return -EFAULT;
2634 }
2635 }
2636 if (newval && newlen) {
2637 int new;
2638 if (newlen != sizeof(int))
2639 return -EINVAL;
2640 if (get_user(new, (int __user *)newval))
2641 return -EFAULT;
2642 *(int *)(table->data) = msecs_to_jiffies(new);
2643 }
2644 return 1;
2645 }
2646
2647
2648 /* The generic string strategy routine: */
2649 static int sysctl_uts_string(ctl_table *table, int __user *name, int nlen,
2650 void __user *oldval, size_t __user *oldlenp,
2651 void __user *newval, size_t newlen)
2652 {
2653 struct ctl_table uts_table;
2654 int r, write;
2655 write = newval && newlen;
2656 memcpy(&uts_table, table, sizeof(uts_table));
2657 uts_table.data = get_uts(table, write);
2658 r = sysctl_string(&uts_table, name, nlen,
2659 oldval, oldlenp, newval, newlen);
2660 put_uts(table, write, uts_table.data);
2661 return r;
2662 }
2663
2664 #ifdef CONFIG_SYSVIPC
2665 /* The generic sysctl ipc data routine. */
2666 static int sysctl_ipc_data(ctl_table *table, int __user *name, int nlen,
2667 void __user *oldval, size_t __user *oldlenp,
2668 void __user *newval, size_t newlen)
2669 {
2670 size_t len;
2671 void *data;
2672
2673 /* Get out of I don't have a variable */
2674 if (!table->data || !table->maxlen)
2675 return -ENOTDIR;
2676
2677 data = get_ipc(table, 1);
2678 if (!data)
2679 return -ENOTDIR;
2680
2681 if (oldval && oldlenp) {
2682 if (get_user(len, oldlenp))
2683 return -EFAULT;
2684 if (len) {
2685 if (len > table->maxlen)
2686 len = table->maxlen;
2687 if (copy_to_user(oldval, data, len))
2688 return -EFAULT;
2689 if (put_user(len, oldlenp))
2690 return -EFAULT;
2691 }
2692 }
2693
2694 if (newval && newlen) {
2695 if (newlen > table->maxlen)
2696 newlen = table->maxlen;
2697
2698 if (copy_from_user(data, newval, newlen))
2699 return -EFAULT;
2700 }
2701 return 1;
2702 }
2703 #endif
2704
2705 #else /* CONFIG_SYSCTL_SYSCALL */
2706
2707
2708 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2709 {
2710 static int msg_count;
2711 struct __sysctl_args tmp;
2712 int name[CTL_MAXNAME];
2713 int i;
2714
2715 /* Read in the sysctl name for better debug message logging */
2716 if (copy_from_user(&tmp, args, sizeof(tmp)))
2717 return -EFAULT;
2718 if (tmp.nlen <= 0 || tmp.nlen >= CTL_MAXNAME)
2719 return -ENOTDIR;
2720 for (i = 0; i < tmp.nlen; i++)
2721 if (get_user(name[i], tmp.name + i))
2722 return -EFAULT;
2723
2724 /* Ignore accesses to kernel.version */
2725 if ((tmp.nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2726 goto out;
2727
2728 if (msg_count < 5) {
2729 msg_count++;
2730 printk(KERN_INFO
2731 "warning: process `%s' used the removed sysctl "
2732 "system call with ", current->comm);
2733 for (i = 0; i < tmp.nlen; i++)
2734 printk("%d.", name[i]);
2735 printk("\n");
2736 }
2737 out:
2738 return -ENOSYS;
2739 }
2740
2741 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2742 void __user *oldval, size_t __user *oldlenp,
2743 void __user *newval, size_t newlen)
2744 {
2745 return -ENOSYS;
2746 }
2747
2748 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2749 void __user *oldval, size_t __user *oldlenp,
2750 void __user *newval, size_t newlen)
2751 {
2752 return -ENOSYS;
2753 }
2754
2755 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2756 void __user *oldval, size_t __user *oldlenp,
2757 void __user *newval, size_t newlen)
2758 {
2759 return -ENOSYS;
2760 }
2761
2762 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2763 void __user *oldval, size_t __user *oldlenp,
2764 void __user *newval, size_t newlen)
2765 {
2766 return -ENOSYS;
2767 }
2768
2769 static int sysctl_uts_string(ctl_table *table, int __user *name, int nlen,
2770 void __user *oldval, size_t __user *oldlenp,
2771 void __user *newval, size_t newlen)
2772 {
2773 return -ENOSYS;
2774 }
2775 #ifdef CONFIG_SYSVIPC
2776 static int sysctl_ipc_data(ctl_table *table, int __user *name, int nlen,
2777 void __user *oldval, size_t __user *oldlenp,
2778 void __user *newval, size_t newlen)
2779 {
2780 return -ENOSYS;
2781 }
2782 #endif
2783 #endif /* CONFIG_SYSCTL_SYSCALL */
2784
2785 /*
2786 * No sense putting this after each symbol definition, twice,
2787 * exception granted :-)
2788 */
2789 EXPORT_SYMBOL(proc_dointvec);
2790 EXPORT_SYMBOL(proc_dointvec_jiffies);
2791 EXPORT_SYMBOL(proc_dointvec_minmax);
2792 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2793 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2794 EXPORT_SYMBOL(proc_dostring);
2795 EXPORT_SYMBOL(proc_doulongvec_minmax);
2796 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2797 EXPORT_SYMBOL(register_sysctl_table);
2798 EXPORT_SYMBOL(sysctl_intvec);
2799 EXPORT_SYMBOL(sysctl_jiffies);
2800 EXPORT_SYMBOL(sysctl_ms_jiffies);
2801 EXPORT_SYMBOL(sysctl_string);
2802 EXPORT_SYMBOL(unregister_sysctl_table);
Mirror of the linux-mips.org kernel tree