?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
[NETFILTER]: nf_conntrack: change nf_conntrack_l[34]proto_unregister to void
[linux-2.6.22.y-op.git] / kernel / sysctl.c
blobe0ac6cd79fcf6955c6105815b4c8ab4d9d53e97f
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 extern void init_irq_proc (void);
1181
1182 static DEFINE_SPINLOCK(sysctl_lock);
1183
1184 /* called under sysctl_lock */
1185 static int use_table(struct ctl_table_header *p)
1186 {
1187 if (unlikely(p->unregistering))
1188 return 0;
1189 p->used++;
1190 return 1;
1191 }
1192
1193 /* called under sysctl_lock */
1194 static void unuse_table(struct ctl_table_header *p)
1195 {
1196 if (!--p->used)
1197 if (unlikely(p->unregistering))
1198 complete(p->unregistering);
1199 }
1200
1201 /* called under sysctl_lock, will reacquire if has to wait */
1202 static void start_unregistering(struct ctl_table_header *p)
1203 {
1204 /*
1205 * if p->used is 0, nobody will ever touch that entry again;
1206 * we'll eliminate all paths to it before dropping sysctl_lock
1207 */
1208 if (unlikely(p->used)) {
1209 struct completion wait;
1210 init_completion(&wait);
1211 p->unregistering = &wait;
1212 spin_unlock(&sysctl_lock);
1213 wait_for_completion(&wait);
1214 spin_lock(&sysctl_lock);
1215 }
1216 /*
1217 * do not remove from the list until nobody holds it; walking the
1218 * list in do_sysctl() relies on that.
1219 */
1220 list_del_init(&p->ctl_entry);
1221 }
1222
1223 void __init sysctl_init(void)
1224 {
1225 #ifdef CONFIG_PROC_SYSCTL
1226 register_proc_table(root_table, proc_sys_root, &root_table_header);
1227 init_irq_proc();
1228 #endif
1229 }
1230
1231 #ifdef CONFIG_SYSCTL_SYSCALL
1232 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1233 void __user *newval, size_t newlen)
1234 {
1235 struct list_head *tmp;
1236 int error = -ENOTDIR;
1237
1238 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1239 return -ENOTDIR;
1240 if (oldval) {
1241 int old_len;
1242 if (!oldlenp || get_user(old_len, oldlenp))
1243 return -EFAULT;
1244 }
1245 spin_lock(&sysctl_lock);
1246 tmp = &root_table_header.ctl_entry;
1247 do {
1248 struct ctl_table_header *head =
1249 list_entry(tmp, struct ctl_table_header, ctl_entry);
1250
1251 if (!use_table(head))
1252 continue;
1253
1254 spin_unlock(&sysctl_lock);
1255
1256 error = parse_table(name, nlen, oldval, oldlenp,
1257 newval, newlen, head->ctl_table);
1258
1259 spin_lock(&sysctl_lock);
1260 unuse_table(head);
1261 if (error != -ENOTDIR)
1262 break;
1263 } while ((tmp = tmp->next) != &root_table_header.ctl_entry);
1264 spin_unlock(&sysctl_lock);
1265 return error;
1266 }
1267
1268 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1269 {
1270 struct __sysctl_args tmp;
1271 int error;
1272
1273 if (copy_from_user(&tmp, args, sizeof(tmp)))
1274 return -EFAULT;
1275
1276 lock_kernel();
1277 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1278 tmp.newval, tmp.newlen);
1279 unlock_kernel();
1280 return error;
1281 }
1282 #endif /* CONFIG_SYSCTL_SYSCALL */
1283
1284 /*
1285 * ctl_perm does NOT grant the superuser all rights automatically, because
1286 * some sysctl variables are readonly even to root.
1287 */
1288
1289 static int test_perm(int mode, int op)
1290 {
1291 if (!current->euid)
1292 mode >>= 6;
1293 else if (in_egroup_p(0))
1294 mode >>= 3;
1295 if ((mode & op & 0007) == op)
1296 return 0;
1297 return -EACCES;
1298 }
1299
1300 static inline int ctl_perm(ctl_table *table, int op)
1301 {
1302 int error;
1303 error = security_sysctl(table, op);
1304 if (error)
1305 return error;
1306 return test_perm(table->mode, op);
1307 }
1308
1309 #ifdef CONFIG_SYSCTL_SYSCALL
1310 static int parse_table(int __user *name, int nlen,
1311 void __user *oldval, size_t __user *oldlenp,
1312 void __user *newval, size_t newlen,
1313 ctl_table *table)
1314 {
1315 int n;
1316 repeat:
1317 if (!nlen)
1318 return -ENOTDIR;
1319 if (get_user(n, name))
1320 return -EFAULT;
1321 for ( ; table->ctl_name || table->procname; table++) {
1322 if (!table->ctl_name)
1323 continue;
1324 if (n == table->ctl_name || table->ctl_name == CTL_ANY) {
1325 int error;
1326 if (table->child) {
1327 if (ctl_perm(table, 001))
1328 return -EPERM;
1329 if (table->strategy) {
1330 error = table->strategy(
1331 table, name, nlen,
1332 oldval, oldlenp,
1333 newval, newlen);
1334 if (error)
1335 return error;
1336 }
1337 name++;
1338 nlen--;
1339 table = table->child;
1340 goto repeat;
1341 }
1342 error = do_sysctl_strategy(table, name, nlen,
1343 oldval, oldlenp,
1344 newval, newlen);
1345 return error;
1346 }
1347 }
1348 return -ENOTDIR;
1349 }
1350
1351 /* Perform the actual read/write of a sysctl table entry. */
1352 int do_sysctl_strategy (ctl_table *table,
1353 int __user *name, int nlen,
1354 void __user *oldval, size_t __user *oldlenp,
1355 void __user *newval, size_t newlen)
1356 {
1357 int op = 0, rc;
1358 size_t len;
1359
1360 if (oldval)
1361 op |= 004;
1362 if (newval)
1363 op |= 002;
1364 if (ctl_perm(table, op))
1365 return -EPERM;
1366
1367 if (table->strategy) {
1368 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1369 newval, newlen);
1370 if (rc < 0)
1371 return rc;
1372 if (rc > 0)
1373 return 0;
1374 }
1375
1376 /* If there is no strategy routine, or if the strategy returns
1377 * zero, proceed with automatic r/w */
1378 if (table->data && table->maxlen) {
1379 if (oldval && oldlenp) {
1380 if (get_user(len, oldlenp))
1381 return -EFAULT;
1382 if (len) {
1383 if (len > table->maxlen)
1384 len = table->maxlen;
1385 if(copy_to_user(oldval, table->data, len))
1386 return -EFAULT;
1387 if(put_user(len, oldlenp))
1388 return -EFAULT;
1389 }
1390 }
1391 if (newval && newlen) {
1392 len = newlen;
1393 if (len > table->maxlen)
1394 len = table->maxlen;
1395 if(copy_from_user(table->data, newval, len))
1396 return -EFAULT;
1397 }
1398 }
1399 return 0;
1400 }
1401 #endif /* CONFIG_SYSCTL_SYSCALL */
1402
1403 /**
1404 * register_sysctl_table - register a sysctl hierarchy
1405 * @table: the top-level table structure
1406 * @insert_at_head: whether the entry should be inserted in front or at the end
1407 *
1408 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1409 * array. An entry with a ctl_name of 0 terminates the table.
1410 *
1411 * The members of the &ctl_table structure are used as follows:
1412 *
1413 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1414 * must be unique within that level of sysctl
1415 *
1416 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1417 * enter a sysctl file
1418 *
1419 * data - a pointer to data for use by proc_handler
1420 *
1421 * maxlen - the maximum size in bytes of the data
1422 *
1423 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1424 *
1425 * child - a pointer to the child sysctl table if this entry is a directory, or
1426 * %NULL.
1427 *
1428 * proc_handler - the text handler routine (described below)
1429 *
1430 * strategy - the strategy routine (described below)
1431 *
1432 * de - for internal use by the sysctl routines
1433 *
1434 * extra1, extra2 - extra pointers usable by the proc handler routines
1435 *
1436 * Leaf nodes in the sysctl tree will be represented by a single file
1437 * under /proc; non-leaf nodes will be represented by directories.
1438 *
1439 * sysctl(2) can automatically manage read and write requests through
1440 * the sysctl table. The data and maxlen fields of the ctl_table
1441 * struct enable minimal validation of the values being written to be
1442 * performed, and the mode field allows minimal authentication.
1443 *
1444 * More sophisticated management can be enabled by the provision of a
1445 * strategy routine with the table entry. This will be called before
1446 * any automatic read or write of the data is performed.
1447 *
1448 * The strategy routine may return
1449 *
1450 * < 0 - Error occurred (error is passed to user process)
1451 *
1452 * 0 - OK - proceed with automatic read or write.
1453 *
1454 * > 0 - OK - read or write has been done by the strategy routine, so
1455 * return immediately.
1456 *
1457 * There must be a proc_handler routine for any terminal nodes
1458 * mirrored under /proc/sys (non-terminals are handled by a built-in
1459 * directory handler). Several default handlers are available to
1460 * cover common cases -
1461 *
1462 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1463 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1464 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1465 *
1466 * It is the handler's job to read the input buffer from user memory
1467 * and process it. The handler should return 0 on success.
1468 *
1469 * This routine returns %NULL on a failure to register, and a pointer
1470 * to the table header on success.
1471 */
1472 struct ctl_table_header *register_sysctl_table(ctl_table * table,
1473 int insert_at_head)
1474 {
1475 struct ctl_table_header *tmp;
1476 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1477 if (!tmp)
1478 return NULL;
1479 tmp->ctl_table = table;
1480 INIT_LIST_HEAD(&tmp->ctl_entry);
1481 tmp->used = 0;
1482 tmp->unregistering = NULL;
1483 spin_lock(&sysctl_lock);
1484 if (insert_at_head)
1485 list_add(&tmp->ctl_entry, &root_table_header.ctl_entry);
1486 else
1487 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
1488 spin_unlock(&sysctl_lock);
1489 #ifdef CONFIG_PROC_SYSCTL
1490 register_proc_table(table, proc_sys_root, tmp);
1491 #endif
1492 return tmp;
1493 }
1494
1495 /**
1496 * unregister_sysctl_table - unregister a sysctl table hierarchy
1497 * @header: the header returned from register_sysctl_table
1498 *
1499 * Unregisters the sysctl table and all children. proc entries may not
1500 * actually be removed until they are no longer used by anyone.
1501 */
1502 void unregister_sysctl_table(struct ctl_table_header * header)
1503 {
1504 might_sleep();
1505 spin_lock(&sysctl_lock);
1506 start_unregistering(header);
1507 #ifdef CONFIG_PROC_SYSCTL
1508 unregister_proc_table(header->ctl_table, proc_sys_root);
1509 #endif
1510 spin_unlock(&sysctl_lock);
1511 kfree(header);
1512 }
1513
1514 #else /* !CONFIG_SYSCTL */
1515 struct ctl_table_header * register_sysctl_table(ctl_table * table,
1516 int insert_at_head)
1517 {
1518 return NULL;
1519 }
1520
1521 void unregister_sysctl_table(struct ctl_table_header * table)
1522 {
1523 }
1524
1525 #endif /* CONFIG_SYSCTL */
1526
1527 /*
1528 * /proc/sys support
1529 */
1530
1531 #ifdef CONFIG_PROC_SYSCTL
1532
1533 /* Scan the sysctl entries in table and add them all into /proc */
1534 static void register_proc_table(ctl_table * table, struct proc_dir_entry *root, void *set)
1535 {
1536 struct proc_dir_entry *de;
1537 int len;
1538 mode_t mode;
1539
1540 for (; table->ctl_name || table->procname; table++) {
1541 /* Can't do anything without a proc name. */
1542 if (!table->procname)
1543 continue;
1544 /* Maybe we can't do anything with it... */
1545 if (!table->proc_handler && !table->child) {
1546 printk(KERN_WARNING "SYSCTL: Can't register %s\n",
1547 table->procname);
1548 continue;
1549 }
1550
1551 len = strlen(table->procname);
1552 mode = table->mode;
1553
1554 de = NULL;
1555 if (table->proc_handler)
1556 mode |= S_IFREG;
1557 else {
1558 mode |= S_IFDIR;
1559 for (de = root->subdir; de; de = de->next) {
1560 if (proc_match(len, table->procname, de))
1561 break;
1562 }
1563 /* If the subdir exists already, de is non-NULL */
1564 }
1565
1566 if (!de) {
1567 de = create_proc_entry(table->procname, mode, root);
1568 if (!de)
1569 continue;
1570 de->set = set;
1571 de->data = (void *) table;
1572 if (table->proc_handler)
1573 de->proc_fops = &proc_sys_file_operations;
1574 }
1575 table->de = de;
1576 if (de->mode & S_IFDIR)
1577 register_proc_table(table->child, de, set);
1578 }
1579 }
1580
1581 /*
1582 * Unregister a /proc sysctl table and any subdirectories.
1583 */
1584 static void unregister_proc_table(ctl_table * table, struct proc_dir_entry *root)
1585 {
1586 struct proc_dir_entry *de;
1587 for (; table->ctl_name || table->procname; table++) {
1588 if (!(de = table->de))
1589 continue;
1590 if (de->mode & S_IFDIR) {
1591 if (!table->child) {
1592 printk (KERN_ALERT "Help - malformed sysctl tree on free\n");
1593 continue;
1594 }
1595 unregister_proc_table(table->child, de);
1596
1597 /* Don't unregister directories which still have entries.. */
1598 if (de->subdir)
1599 continue;
1600 }
1601
1602 /*
1603 * In any case, mark the entry as goner; we'll keep it
1604 * around if it's busy, but we'll know to do nothing with
1605 * its fields. We are under sysctl_lock here.
1606 */
1607 de->data = NULL;
1608
1609 /* Don't unregister proc entries that are still being used.. */
1610 if (atomic_read(&de->count))
1611 continue;
1612
1613 table->de = NULL;
1614 remove_proc_entry(table->procname, root);
1615 }
1616 }
1617
1618 static ssize_t do_rw_proc(int write, struct file * file, char __user * buf,
1619 size_t count, loff_t *ppos)
1620 {
1621 int op;
1622 struct proc_dir_entry *de = PDE(file->f_path.dentry->d_inode);
1623 struct ctl_table *table;
1624 size_t res;
1625 ssize_t error = -ENOTDIR;
1626
1627 spin_lock(&sysctl_lock);
1628 if (de && de->data && use_table(de->set)) {
1629 /*
1630 * at that point we know that sysctl was not unregistered
1631 * and won't be until we finish
1632 */
1633 spin_unlock(&sysctl_lock);
1634 table = (struct ctl_table *) de->data;
1635 if (!table || !table->proc_handler)
1636 goto out;
1637 error = -EPERM;
1638 op = (write ? 002 : 004);
1639 if (ctl_perm(table, op))
1640 goto out;
1641
1642 /* careful: calling conventions are nasty here */
1643 res = count;
1644 error = (*table->proc_handler)(table, write, file,
1645 buf, &res, ppos);
1646 if (!error)
1647 error = res;
1648 out:
1649 spin_lock(&sysctl_lock);
1650 unuse_table(de->set);
1651 }
1652 spin_unlock(&sysctl_lock);
1653 return error;
1654 }
1655
1656 static int proc_opensys(struct inode *inode, struct file *file)
1657 {
1658 if (file->f_mode & FMODE_WRITE) {
1659 /*
1660 * sysctl entries that are not writable,
1661 * are _NOT_ writable, capabilities or not.
1662 */
1663 if (!(inode->i_mode & S_IWUSR))
1664 return -EPERM;
1665 }
1666
1667 return 0;
1668 }
1669
1670 static ssize_t proc_readsys(struct file * file, char __user * buf,
1671 size_t count, loff_t *ppos)
1672 {
1673 return do_rw_proc(0, file, buf, count, ppos);
1674 }
1675
1676 static ssize_t proc_writesys(struct file * file, const char __user * buf,
1677 size_t count, loff_t *ppos)
1678 {
1679 return do_rw_proc(1, file, (char __user *) buf, count, ppos);
1680 }
1681
1682 static int _proc_do_string(void* data, int maxlen, int write,
1683 struct file *filp, void __user *buffer,
1684 size_t *lenp, loff_t *ppos)
1685 {
1686 size_t len;
1687 char __user *p;
1688 char c;
1689
1690 if (!data || !maxlen || !*lenp) {
1691 *lenp = 0;
1692 return 0;
1693 }
1694
1695 if (write) {
1696 len = 0;
1697 p = buffer;
1698 while (len < *lenp) {
1699 if (get_user(c, p++))
1700 return -EFAULT;
1701 if (c == 0 || c == '\n')
1702 break;
1703 len++;
1704 }
1705 if (len >= maxlen)
1706 len = maxlen-1;
1707 if(copy_from_user(data, buffer, len))
1708 return -EFAULT;
1709 ((char *) data)[len] = 0;
1710 *ppos += *lenp;
1711 } else {
1712 len = strlen(data);
1713 if (len > maxlen)
1714 len = maxlen;
1715
1716 if (*ppos > len) {
1717 *lenp = 0;
1718 return 0;
1719 }
1720
1721 data += *ppos;
1722 len -= *ppos;
1723
1724 if (len > *lenp)
1725 len = *lenp;
1726 if (len)
1727 if(copy_to_user(buffer, data, len))
1728 return -EFAULT;
1729 if (len < *lenp) {
1730 if(put_user('\n', ((char __user *) buffer) + len))
1731 return -EFAULT;
1732 len++;
1733 }
1734 *lenp = len;
1735 *ppos += len;
1736 }
1737 return 0;
1738 }
1739
1740 /**
1741 * proc_dostring - read a string sysctl
1742 * @table: the sysctl table
1743 * @write: %TRUE if this is a write to the sysctl file
1744 * @filp: the file structure
1745 * @buffer: the user buffer
1746 * @lenp: the size of the user buffer
1747 * @ppos: file position
1748 *
1749 * Reads/writes a string from/to the user buffer. If the kernel
1750 * buffer provided is not large enough to hold the string, the
1751 * string is truncated. The copied string is %NULL-terminated.
1752 * If the string is being read by the user process, it is copied
1753 * and a newline '\n' is added. It is truncated if the buffer is
1754 * not large enough.
1755 *
1756 * Returns 0 on success.
1757 */
1758 int proc_dostring(ctl_table *table, int write, struct file *filp,
1759 void __user *buffer, size_t *lenp, loff_t *ppos)
1760 {
1761 return _proc_do_string(table->data, table->maxlen, write, filp,
1762 buffer, lenp, ppos);
1763 }
1764
1765 /*
1766 * Special case of dostring for the UTS structure. This has locks
1767 * to observe. Should this be in kernel/sys.c ????
1768 */
1769
1770 static int proc_do_uts_string(ctl_table *table, int write, struct file *filp,
1771 void __user *buffer, size_t *lenp, loff_t *ppos)
1772 {
1773 int r;
1774 void *which;
1775 which = get_uts(table, write);
1776 r = _proc_do_string(which, table->maxlen,write,filp,buffer,lenp, ppos);
1777 put_uts(table, write, which);
1778 return r;
1779 }
1780
1781 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1782 int *valp,
1783 int write, void *data)
1784 {
1785 if (write) {
1786 *valp = *negp ? -*lvalp : *lvalp;
1787 } else {
1788 int val = *valp;
1789 if (val < 0) {
1790 *negp = -1;
1791 *lvalp = (unsigned long)-val;
1792 } else {
1793 *negp = 0;
1794 *lvalp = (unsigned long)val;
1795 }
1796 }
1797 return 0;
1798 }
1799
1800 static int __do_proc_dointvec(void *tbl_data, ctl_table *table,
1801 int write, struct file *filp, void __user *buffer,
1802 size_t *lenp, loff_t *ppos,
1803 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1804 int write, void *data),
1805 void *data)
1806 {
1807 #define TMPBUFLEN 21
1808 int *i, vleft, first=1, neg, val;
1809 unsigned long lval;
1810 size_t left, len;
1811
1812 char buf[TMPBUFLEN], *p;
1813 char __user *s = buffer;
1814
1815 if (!tbl_data || !table->maxlen || !*lenp ||
1816 (*ppos && !write)) {
1817 *lenp = 0;
1818 return 0;
1819 }
1820
1821 i = (int *) tbl_data;
1822 vleft = table->maxlen / sizeof(*i);
1823 left = *lenp;
1824
1825 if (!conv)
1826 conv = do_proc_dointvec_conv;
1827
1828 for (; left && vleft--; i++, first=0) {
1829 if (write) {
1830 while (left) {
1831 char c;
1832 if (get_user(c, s))
1833 return -EFAULT;
1834 if (!isspace(c))
1835 break;
1836 left--;
1837 s++;
1838 }
1839 if (!left)
1840 break;
1841 neg = 0;
1842 len = left;
1843 if (len > sizeof(buf) - 1)
1844 len = sizeof(buf) - 1;
1845 if (copy_from_user(buf, s, len))
1846 return -EFAULT;
1847 buf[len] = 0;
1848 p = buf;
1849 if (*p == '-' && left > 1) {
1850 neg = 1;
1851 p++;
1852 }
1853 if (*p < '0' || *p > '9')
1854 break;
1855
1856 lval = simple_strtoul(p, &p, 0);
1857
1858 len = p-buf;
1859 if ((len < left) && *p && !isspace(*p))
1860 break;
1861 if (neg)
1862 val = -val;
1863 s += len;
1864 left -= len;
1865
1866 if (conv(&neg, &lval, i, 1, data))
1867 break;
1868 } else {
1869 p = buf;
1870 if (!first)
1871 *p++ = '\t';
1872
1873 if (conv(&neg, &lval, i, 0, data))
1874 break;
1875
1876 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1877 len = strlen(buf);
1878 if (len > left)
1879 len = left;
1880 if(copy_to_user(s, buf, len))
1881 return -EFAULT;
1882 left -= len;
1883 s += len;
1884 }
1885 }
1886
1887 if (!write && !first && left) {
1888 if(put_user('\n', s))
1889 return -EFAULT;
1890 left--, s++;
1891 }
1892 if (write) {
1893 while (left) {
1894 char c;
1895 if (get_user(c, s++))
1896 return -EFAULT;
1897 if (!isspace(c))
1898 break;
1899 left--;
1900 }
1901 }
1902 if (write && first)
1903 return -EINVAL;
1904 *lenp -= left;
1905 *ppos += *lenp;
1906 return 0;
1907 #undef TMPBUFLEN
1908 }
1909
1910 static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1911 void __user *buffer, size_t *lenp, loff_t *ppos,
1912 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1913 int write, void *data),
1914 void *data)
1915 {
1916 return __do_proc_dointvec(table->data, table, write, filp,
1917 buffer, lenp, ppos, conv, data);
1918 }
1919
1920 /**
1921 * proc_dointvec - read a vector of integers
1922 * @table: the sysctl table
1923 * @write: %TRUE if this is a write to the sysctl file
1924 * @filp: the file structure
1925 * @buffer: the user buffer
1926 * @lenp: the size of the user buffer
1927 * @ppos: file position
1928 *
1929 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1930 * values from/to the user buffer, treated as an ASCII string.
1931 *
1932 * Returns 0 on success.
1933 */
1934 int proc_dointvec(ctl_table *table, int write, struct file *filp,
1935 void __user *buffer, size_t *lenp, loff_t *ppos)
1936 {
1937 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1938 NULL,NULL);
1939 }
1940
1941 #define OP_SET 0
1942 #define OP_AND 1
1943 #define OP_OR 2
1944
1945 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1946 int *valp,
1947 int write, void *data)
1948 {
1949 int op = *(int *)data;
1950 if (write) {
1951 int val = *negp ? -*lvalp : *lvalp;
1952 switch(op) {
1953 case OP_SET: *valp = val; break;
1954 case OP_AND: *valp &= val; break;
1955 case OP_OR: *valp |= val; break;
1956 }
1957 } else {
1958 int val = *valp;
1959 if (val < 0) {
1960 *negp = -1;
1961 *lvalp = (unsigned long)-val;
1962 } else {
1963 *negp = 0;
1964 *lvalp = (unsigned long)val;
1965 }
1966 }
1967 return 0;
1968 }
1969
1970 /*
1971 * init may raise the set.
1972 */
1973
1974 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1975 void __user *buffer, size_t *lenp, loff_t *ppos)
1976 {
1977 int op;
1978
1979 if (write && !capable(CAP_SYS_MODULE)) {
1980 return -EPERM;
1981 }
1982
1983 op = is_init(current) ? OP_SET : OP_AND;
1984 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1985 do_proc_dointvec_bset_conv,&op);
1986 }
1987
1988 /*
1989 * Taint values can only be increased
1990 */
1991 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
1992 void __user *buffer, size_t *lenp, loff_t *ppos)
1993 {
1994 int op;
1995
1996 if (!capable(CAP_SYS_ADMIN))
1997 return -EPERM;
1998
1999 op = OP_OR;
2000 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2001 do_proc_dointvec_bset_conv,&op);
2002 }
2003
2004 struct do_proc_dointvec_minmax_conv_param {
2005 int *min;
2006 int *max;
2007 };
2008
2009 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2010 int *valp,
2011 int write, void *data)
2012 {
2013 struct do_proc_dointvec_minmax_conv_param *param = data;
2014 if (write) {
2015 int val = *negp ? -*lvalp : *lvalp;
2016 if ((param->min && *param->min > val) ||
2017 (param->max && *param->max < val))
2018 return -EINVAL;
2019 *valp = val;
2020 } else {
2021 int val = *valp;
2022 if (val < 0) {
2023 *negp = -1;
2024 *lvalp = (unsigned long)-val;
2025 } else {
2026 *negp = 0;
2027 *lvalp = (unsigned long)val;
2028 }
2029 }
2030 return 0;
2031 }
2032
2033 /**
2034 * proc_dointvec_minmax - read a vector of integers with min/max values
2035 * @table: the sysctl table
2036 * @write: %TRUE if this is a write to the sysctl file
2037 * @filp: the file structure
2038 * @buffer: the user buffer
2039 * @lenp: the size of the user buffer
2040 * @ppos: file position
2041 *
2042 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2043 * values from/to the user buffer, treated as an ASCII string.
2044 *
2045 * This routine will ensure the values are within the range specified by
2046 * table->extra1 (min) and table->extra2 (max).
2047 *
2048 * Returns 0 on success.
2049 */
2050 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2051 void __user *buffer, size_t *lenp, loff_t *ppos)
2052 {
2053 struct do_proc_dointvec_minmax_conv_param param = {
2054 .min = (int *) table->extra1,
2055 .max = (int *) table->extra2,
2056 };
2057 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2058 do_proc_dointvec_minmax_conv, &param);
2059 }
2060
2061 static int __do_proc_doulongvec_minmax(void *data, ctl_table *table, int write,
2062 struct file *filp,
2063 void __user *buffer,
2064 size_t *lenp, loff_t *ppos,
2065 unsigned long convmul,
2066 unsigned long convdiv)
2067 {
2068 #define TMPBUFLEN 21
2069 unsigned long *i, *min, *max, val;
2070 int vleft, first=1, neg;
2071 size_t len, left;
2072 char buf[TMPBUFLEN], *p;
2073 char __user *s = buffer;
2074
2075 if (!data || !table->maxlen || !*lenp ||
2076 (*ppos && !write)) {
2077 *lenp = 0;
2078 return 0;
2079 }
2080
2081 i = (unsigned long *) data;
2082 min = (unsigned long *) table->extra1;
2083 max = (unsigned long *) table->extra2;
2084 vleft = table->maxlen / sizeof(unsigned long);
2085 left = *lenp;
2086
2087 for (; left && vleft--; i++, min++, max++, first=0) {
2088 if (write) {
2089 while (left) {
2090 char c;
2091 if (get_user(c, s))
2092 return -EFAULT;
2093 if (!isspace(c))
2094 break;
2095 left--;
2096 s++;
2097 }
2098 if (!left)
2099 break;
2100 neg = 0;
2101 len = left;
2102 if (len > TMPBUFLEN-1)
2103 len = TMPBUFLEN-1;
2104 if (copy_from_user(buf, s, len))
2105 return -EFAULT;
2106 buf[len] = 0;
2107 p = buf;
2108 if (*p == '-' && left > 1) {
2109 neg = 1;
2110 p++;
2111 }
2112 if (*p < '0' || *p > '9')
2113 break;
2114 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2115 len = p-buf;
2116 if ((len < left) && *p && !isspace(*p))
2117 break;
2118 if (neg)
2119 val = -val;
2120 s += len;
2121 left -= len;
2122
2123 if(neg)
2124 continue;
2125 if ((min && val < *min) || (max && val > *max))
2126 continue;
2127 *i = val;
2128 } else {
2129 p = buf;
2130 if (!first)
2131 *p++ = '\t';
2132 sprintf(p, "%lu", convdiv * (*i) / convmul);
2133 len = strlen(buf);
2134 if (len > left)
2135 len = left;
2136 if(copy_to_user(s, buf, len))
2137 return -EFAULT;
2138 left -= len;
2139 s += len;
2140 }
2141 }
2142
2143 if (!write && !first && left) {
2144 if(put_user('\n', s))
2145 return -EFAULT;
2146 left--, s++;
2147 }
2148 if (write) {
2149 while (left) {
2150 char c;
2151 if (get_user(c, s++))
2152 return -EFAULT;
2153 if (!isspace(c))
2154 break;
2155 left--;
2156 }
2157 }
2158 if (write && first)
2159 return -EINVAL;
2160 *lenp -= left;
2161 *ppos += *lenp;
2162 return 0;
2163 #undef TMPBUFLEN
2164 }
2165
2166 static int do_proc_doulongvec_minmax(ctl_table *table, int write,
2167 struct file *filp,
2168 void __user *buffer,
2169 size_t *lenp, loff_t *ppos,
2170 unsigned long convmul,
2171 unsigned long convdiv)
2172 {
2173 return __do_proc_doulongvec_minmax(table->data, table, write,
2174 filp, buffer, lenp, ppos, convmul, convdiv);
2175 }
2176
2177 /**
2178 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2179 * @table: the sysctl table
2180 * @write: %TRUE if this is a write to the sysctl file
2181 * @filp: the file structure
2182 * @buffer: the user buffer
2183 * @lenp: the size of the user buffer
2184 * @ppos: file position
2185 *
2186 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2187 * values from/to the user buffer, treated as an ASCII string.
2188 *
2189 * This routine will ensure the values are within the range specified by
2190 * table->extra1 (min) and table->extra2 (max).
2191 *
2192 * Returns 0 on success.
2193 */
2194 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2195 void __user *buffer, size_t *lenp, loff_t *ppos)
2196 {
2197 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2198 }
2199
2200 /**
2201 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2202 * @table: the sysctl table
2203 * @write: %TRUE if this is a write to the sysctl file
2204 * @filp: the file structure
2205 * @buffer: the user buffer
2206 * @lenp: the size of the user buffer
2207 * @ppos: file position
2208 *
2209 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2210 * values from/to the user buffer, treated as an ASCII string. The values
2211 * are treated as milliseconds, and converted to jiffies when they are stored.
2212 *
2213 * This routine will ensure the values are within the range specified by
2214 * table->extra1 (min) and table->extra2 (max).
2215 *
2216 * Returns 0 on success.
2217 */
2218 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2219 struct file *filp,
2220 void __user *buffer,
2221 size_t *lenp, loff_t *ppos)
2222 {
2223 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2224 lenp, ppos, HZ, 1000l);
2225 }
2226
2227
2228 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2229 int *valp,
2230 int write, void *data)
2231 {
2232 if (write) {
2233 if (*lvalp > LONG_MAX / HZ)
2234 return 1;
2235 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2236 } else {
2237 int val = *valp;
2238 unsigned long lval;
2239 if (val < 0) {
2240 *negp = -1;
2241 lval = (unsigned long)-val;
2242 } else {
2243 *negp = 0;
2244 lval = (unsigned long)val;
2245 }
2246 *lvalp = lval / HZ;
2247 }
2248 return 0;
2249 }
2250
2251 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2252 int *valp,
2253 int write, void *data)
2254 {
2255 if (write) {
2256 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2257 return 1;
2258 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2259 } else {
2260 int val = *valp;
2261 unsigned long lval;
2262 if (val < 0) {
2263 *negp = -1;
2264 lval = (unsigned long)-val;
2265 } else {
2266 *negp = 0;
2267 lval = (unsigned long)val;
2268 }
2269 *lvalp = jiffies_to_clock_t(lval);
2270 }
2271 return 0;
2272 }
2273
2274 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2275 int *valp,
2276 int write, void *data)
2277 {
2278 if (write) {
2279 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2280 } else {
2281 int val = *valp;
2282 unsigned long lval;
2283 if (val < 0) {
2284 *negp = -1;
2285 lval = (unsigned long)-val;
2286 } else {
2287 *negp = 0;
2288 lval = (unsigned long)val;
2289 }
2290 *lvalp = jiffies_to_msecs(lval);
2291 }
2292 return 0;
2293 }
2294
2295 /**
2296 * proc_dointvec_jiffies - read a vector of integers as seconds
2297 * @table: the sysctl table
2298 * @write: %TRUE if this is a write to the sysctl file
2299 * @filp: the file structure
2300 * @buffer: the user buffer
2301 * @lenp: the size of the user buffer
2302 * @ppos: file position
2303 *
2304 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2305 * values from/to the user buffer, treated as an ASCII string.
2306 * The values read are assumed to be in seconds, and are converted into
2307 * jiffies.
2308 *
2309 * Returns 0 on success.
2310 */
2311 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2312 void __user *buffer, size_t *lenp, loff_t *ppos)
2313 {
2314 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2315 do_proc_dointvec_jiffies_conv,NULL);
2316 }
2317
2318 /**
2319 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2320 * @table: the sysctl table
2321 * @write: %TRUE if this is a write to the sysctl file
2322 * @filp: the file structure
2323 * @buffer: the user buffer
2324 * @lenp: the size of the user buffer
2325 * @ppos: pointer to the file position
2326 *
2327 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2328 * values from/to the user buffer, treated as an ASCII string.
2329 * The values read are assumed to be in 1/USER_HZ seconds, and
2330 * are converted into jiffies.
2331 *
2332 * Returns 0 on success.
2333 */
2334 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2335 void __user *buffer, size_t *lenp, loff_t *ppos)
2336 {
2337 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2338 do_proc_dointvec_userhz_jiffies_conv,NULL);
2339 }
2340
2341 /**
2342 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2343 * @table: the sysctl table
2344 * @write: %TRUE if this is a write to the sysctl file
2345 * @filp: the file structure
2346 * @buffer: the user buffer
2347 * @lenp: the size of the user buffer
2348 * @ppos: file position
2349 * @ppos: the current position in the file
2350 *
2351 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2352 * values from/to the user buffer, treated as an ASCII string.
2353 * The values read are assumed to be in 1/1000 seconds, and
2354 * are converted into jiffies.
2355 *
2356 * Returns 0 on success.
2357 */
2358 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2359 void __user *buffer, size_t *lenp, loff_t *ppos)
2360 {
2361 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2362 do_proc_dointvec_ms_jiffies_conv, NULL);
2363 }
2364
2365 #ifdef CONFIG_SYSVIPC
2366 static int proc_ipc_dointvec(ctl_table *table, int write, struct file *filp,
2367 void __user *buffer, size_t *lenp, loff_t *ppos)
2368 {
2369 void *which;
2370 which = get_ipc(table, write);
2371 return __do_proc_dointvec(which, table, write, filp, buffer,
2372 lenp, ppos, NULL, NULL);
2373 }
2374
2375 static int proc_ipc_doulongvec_minmax(ctl_table *table, int write,
2376 struct file *filp, void __user *buffer, size_t *lenp, loff_t *ppos)
2377 {
2378 void *which;
2379 which = get_ipc(table, write);
2380 return __do_proc_doulongvec_minmax(which, table, write, filp, buffer,
2381 lenp, ppos, 1l, 1l);
2382 }
2383
2384 #endif
2385
2386 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
2387 void __user *buffer, size_t *lenp, loff_t *ppos)
2388 {
2389 struct pid *new_pid;
2390 pid_t tmp;
2391 int r;
2392
2393 tmp = pid_nr(cad_pid);
2394
2395 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2396 lenp, ppos, NULL, NULL);
2397 if (r || !write)
2398 return r;
2399
2400 new_pid = find_get_pid(tmp);
2401 if (!new_pid)
2402 return -ESRCH;
2403
2404 put_pid(xchg(&cad_pid, new_pid));
2405 return 0;
2406 }
2407
2408 #else /* CONFIG_PROC_FS */
2409
2410 int proc_dostring(ctl_table *table, int write, struct file *filp,
2411 void __user *buffer, size_t *lenp, loff_t *ppos)
2412 {
2413 return -ENOSYS;
2414 }
2415
2416 static int proc_do_uts_string(ctl_table *table, int write, struct file *filp,
2417 void __user *buffer, size_t *lenp, loff_t *ppos)
2418 {
2419 return -ENOSYS;
2420 }
2421
2422 #ifdef CONFIG_SYSVIPC
2423 static int proc_do_ipc_string(ctl_table *table, int write, struct file *filp,
2424 void __user *buffer, size_t *lenp, loff_t *ppos)
2425 {
2426 return -ENOSYS;
2427 }
2428 static int proc_ipc_dointvec(ctl_table *table, int write, struct file *filp,
2429 void __user *buffer, size_t *lenp, loff_t *ppos)
2430 {
2431 return -ENOSYS;
2432 }
2433 static int proc_ipc_doulongvec_minmax(ctl_table *table, int write,
2434 struct file *filp, void __user *buffer,
2435 size_t *lenp, loff_t *ppos)
2436 {
2437 return -ENOSYS;
2438 }
2439 #endif
2440
2441 int proc_dointvec(ctl_table *table, int write, struct file *filp,
2442 void __user *buffer, size_t *lenp, loff_t *ppos)
2443 {
2444 return -ENOSYS;
2445 }
2446
2447 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2448 void __user *buffer, size_t *lenp, loff_t *ppos)
2449 {
2450 return -ENOSYS;
2451 }
2452
2453 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2454 void __user *buffer, size_t *lenp, loff_t *ppos)
2455 {
2456 return -ENOSYS;
2457 }
2458
2459 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2460 void __user *buffer, size_t *lenp, loff_t *ppos)
2461 {
2462 return -ENOSYS;
2463 }
2464
2465 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2466 void __user *buffer, size_t *lenp, loff_t *ppos)
2467 {
2468 return -ENOSYS;
2469 }
2470
2471 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2472 void __user *buffer, size_t *lenp, loff_t *ppos)
2473 {
2474 return -ENOSYS;
2475 }
2476
2477 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2478 void __user *buffer, size_t *lenp, loff_t *ppos)
2479 {
2480 return -ENOSYS;
2481 }
2482
2483 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2484 struct file *filp,
2485 void __user *buffer,
2486 size_t *lenp, loff_t *ppos)
2487 {
2488 return -ENOSYS;
2489 }
2490
2491
2492 #endif /* CONFIG_PROC_FS */
2493
2494
2495 #ifdef CONFIG_SYSCTL_SYSCALL
2496 /*
2497 * General sysctl support routines
2498 */
2499
2500 /* The generic string strategy routine: */
2501 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2502 void __user *oldval, size_t __user *oldlenp,
2503 void __user *newval, size_t newlen)
2504 {
2505 if (!table->data || !table->maxlen)
2506 return -ENOTDIR;
2507
2508 if (oldval && oldlenp) {
2509 size_t bufsize;
2510 if (get_user(bufsize, oldlenp))
2511 return -EFAULT;
2512 if (bufsize) {
2513 size_t len = strlen(table->data), copied;
2514
2515 /* This shouldn't trigger for a well-formed sysctl */
2516 if (len > table->maxlen)
2517 len = table->maxlen;
2518
2519 /* Copy up to a max of bufsize-1 bytes of the string */
2520 copied = (len >= bufsize) ? bufsize - 1 : len;
2521
2522 if (copy_to_user(oldval, table->data, copied) ||
2523 put_user(0, (char __user *)(oldval + copied)))
2524 return -EFAULT;
2525 if (put_user(len, oldlenp))
2526 return -EFAULT;
2527 }
2528 }
2529 if (newval && newlen) {
2530 size_t len = newlen;
2531 if (len > table->maxlen)
2532 len = table->maxlen;
2533 if(copy_from_user(table->data, newval, len))
2534 return -EFAULT;
2535 if (len == table->maxlen)
2536 len--;
2537 ((char *) table->data)[len] = 0;
2538 }
2539 return 1;
2540 }
2541
2542 /*
2543 * This function makes sure that all of the integers in the vector
2544 * are between the minimum and maximum values given in the arrays
2545 * table->extra1 and table->extra2, respectively.
2546 */
2547 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2548 void __user *oldval, size_t __user *oldlenp,
2549 void __user *newval, size_t newlen)
2550 {
2551
2552 if (newval && newlen) {
2553 int __user *vec = (int __user *) newval;
2554 int *min = (int *) table->extra1;
2555 int *max = (int *) table->extra2;
2556 size_t length;
2557 int i;
2558
2559 if (newlen % sizeof(int) != 0)
2560 return -EINVAL;
2561
2562 if (!table->extra1 && !table->extra2)
2563 return 0;
2564
2565 if (newlen > table->maxlen)
2566 newlen = table->maxlen;
2567 length = newlen / sizeof(int);
2568
2569 for (i = 0; i < length; i++) {
2570 int value;
2571 if (get_user(value, vec + i))
2572 return -EFAULT;
2573 if (min && value < min[i])
2574 return -EINVAL;
2575 if (max && value > max[i])
2576 return -EINVAL;
2577 }
2578 }
2579 return 0;
2580 }
2581
2582 /* Strategy function to convert jiffies to seconds */
2583 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2584 void __user *oldval, size_t __user *oldlenp,
2585 void __user *newval, size_t newlen)
2586 {
2587 if (oldval && oldlenp) {
2588 size_t olen;
2589
2590 if (get_user(olen, oldlenp))
2591 return -EFAULT;
2592 if (olen) {
2593 int val;
2594
2595 if (olen < sizeof(int))
2596 return -EINVAL;
2597
2598 val = *(int *)(table->data) / HZ;
2599 if (put_user(val, (int __user *)oldval))
2600 return -EFAULT;
2601 if (put_user(sizeof(int), oldlenp))
2602 return -EFAULT;
2603 }
2604 }
2605 if (newval && newlen) {
2606 int new;
2607 if (newlen != sizeof(int))
2608 return -EINVAL;
2609 if (get_user(new, (int __user *)newval))
2610 return -EFAULT;
2611 *(int *)(table->data) = new*HZ;
2612 }
2613 return 1;
2614 }
2615
2616 /* Strategy function to convert jiffies to seconds */
2617 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2618 void __user *oldval, size_t __user *oldlenp,
2619 void __user *newval, size_t newlen)
2620 {
2621 if (oldval && oldlenp) {
2622 size_t olen;
2623
2624 if (get_user(olen, oldlenp))
2625 return -EFAULT;
2626 if (olen) {
2627 int val;
2628
2629 if (olen < sizeof(int))
2630 return -EINVAL;
2631
2632 val = jiffies_to_msecs(*(int *)(table->data));
2633 if (put_user(val, (int __user *)oldval))
2634 return -EFAULT;
2635 if (put_user(sizeof(int), oldlenp))
2636 return -EFAULT;
2637 }
2638 }
2639 if (newval && newlen) {
2640 int new;
2641 if (newlen != sizeof(int))
2642 return -EINVAL;
2643 if (get_user(new, (int __user *)newval))
2644 return -EFAULT;
2645 *(int *)(table->data) = msecs_to_jiffies(new);
2646 }
2647 return 1;
2648 }
2649
2650
2651 /* The generic string strategy routine: */
2652 static int sysctl_uts_string(ctl_table *table, int __user *name, int nlen,
2653 void __user *oldval, size_t __user *oldlenp,
2654 void __user *newval, size_t newlen)
2655 {
2656 struct ctl_table uts_table;
2657 int r, write;
2658 write = newval && newlen;
2659 memcpy(&uts_table, table, sizeof(uts_table));
2660 uts_table.data = get_uts(table, write);
2661 r = sysctl_string(&uts_table, name, nlen,
2662 oldval, oldlenp, newval, newlen);
2663 put_uts(table, write, uts_table.data);
2664 return r;
2665 }
2666
2667 #ifdef CONFIG_SYSVIPC
2668 /* The generic sysctl ipc data routine. */
2669 static int sysctl_ipc_data(ctl_table *table, int __user *name, int nlen,
2670 void __user *oldval, size_t __user *oldlenp,
2671 void __user *newval, size_t newlen)
2672 {
2673 size_t len;
2674 void *data;
2675
2676 /* Get out of I don't have a variable */
2677 if (!table->data || !table->maxlen)
2678 return -ENOTDIR;
2679
2680 data = get_ipc(table, 1);
2681 if (!data)
2682 return -ENOTDIR;
2683
2684 if (oldval && oldlenp) {
2685 if (get_user(len, oldlenp))
2686 return -EFAULT;
2687 if (len) {
2688 if (len > table->maxlen)
2689 len = table->maxlen;
2690 if (copy_to_user(oldval, data, len))
2691 return -EFAULT;
2692 if (put_user(len, oldlenp))
2693 return -EFAULT;
2694 }
2695 }
2696
2697 if (newval && newlen) {
2698 if (newlen > table->maxlen)
2699 newlen = table->maxlen;
2700
2701 if (copy_from_user(data, newval, newlen))
2702 return -EFAULT;
2703 }
2704 return 1;
2705 }
2706 #endif
2707
2708 #else /* CONFIG_SYSCTL_SYSCALL */
2709
2710
2711 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2712 {
2713 static int msg_count;
2714 struct __sysctl_args tmp;
2715 int name[CTL_MAXNAME];
2716 int i;
2717
2718 /* Read in the sysctl name for better debug message logging */
2719 if (copy_from_user(&tmp, args, sizeof(tmp)))
2720 return -EFAULT;
2721 if (tmp.nlen <= 0 || tmp.nlen >= CTL_MAXNAME)
2722 return -ENOTDIR;
2723 for (i = 0; i < tmp.nlen; i++)
2724 if (get_user(name[i], tmp.name + i))
2725 return -EFAULT;
2726
2727 /* Ignore accesses to kernel.version */
2728 if ((tmp.nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2729 goto out;
2730
2731 if (msg_count < 5) {
2732 msg_count++;
2733 printk(KERN_INFO
2734 "warning: process `%s' used the removed sysctl "
2735 "system call with ", current->comm);
2736 for (i = 0; i < tmp.nlen; i++)
2737 printk("%d.", name[i]);
2738 printk("\n");
2739 }
2740 out:
2741 return -ENOSYS;
2742 }
2743
2744 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2745 void __user *oldval, size_t __user *oldlenp,
2746 void __user *newval, size_t newlen)
2747 {
2748 return -ENOSYS;
2749 }
2750
2751 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2752 void __user *oldval, size_t __user *oldlenp,
2753 void __user *newval, size_t newlen)
2754 {
2755 return -ENOSYS;
2756 }
2757
2758 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2759 void __user *oldval, size_t __user *oldlenp,
2760 void __user *newval, size_t newlen)
2761 {
2762 return -ENOSYS;
2763 }
2764
2765 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2766 void __user *oldval, size_t __user *oldlenp,
2767 void __user *newval, size_t newlen)
2768 {
2769 return -ENOSYS;
2770 }
2771
2772 static int sysctl_uts_string(ctl_table *table, int __user *name, int nlen,
2773 void __user *oldval, size_t __user *oldlenp,
2774 void __user *newval, size_t newlen)
2775 {
2776 return -ENOSYS;
2777 }
2778 #ifdef CONFIG_SYSVIPC
2779 static int sysctl_ipc_data(ctl_table *table, int __user *name, int nlen,
2780 void __user *oldval, size_t __user *oldlenp,
2781 void __user *newval, size_t newlen)
2782 {
2783 return -ENOSYS;
2784 }
2785 #endif
2786 #endif /* CONFIG_SYSCTL_SYSCALL */
2787
2788 /*
2789 * No sense putting this after each symbol definition, twice,
2790 * exception granted :-)
2791 */
2792 EXPORT_SYMBOL(proc_dointvec);
2793 EXPORT_SYMBOL(proc_dointvec_jiffies);
2794 EXPORT_SYMBOL(proc_dointvec_minmax);
2795 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2796 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2797 EXPORT_SYMBOL(proc_dostring);
2798 EXPORT_SYMBOL(proc_doulongvec_minmax);
2799 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2800 EXPORT_SYMBOL(register_sysctl_table);
2801 EXPORT_SYMBOL(sysctl_intvec);
2802 EXPORT_SYMBOL(sysctl_jiffies);
2803 EXPORT_SYMBOL(sysctl_ms_jiffies);
2804 EXPORT_SYMBOL(sysctl_string);
2805 EXPORT_SYMBOL(unregister_sysctl_table);
linux v2.6.22.y-op