?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
[PATCH] sysctl: factor out sysctl_head_next from do_sysctl
[linux-2.6/kvm.git] / kernel / sysctl.c
blobc3e2ac9cb5fb19e9158dc833ce6309d129c89d2d
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
94 #ifdef __sparc__
95 extern char reboot_command [];
96 extern int stop_a_enabled;
97 extern int scons_pwroff;
98 #endif
99
100 #ifdef __hppa__
101 extern int pwrsw_enabled;
102 extern int unaligned_enabled;
103 #endif
104
105 #ifdef CONFIG_S390
106 #ifdef CONFIG_MATHEMU
107 extern int sysctl_ieee_emulation_warnings;
108 #endif
109 extern int sysctl_userprocess_debug;
110 extern int spin_retry;
111 #endif
112
113 extern int sysctl_hz_timer;
114
115 #ifdef CONFIG_BSD_PROCESS_ACCT
116 extern int acct_parm[];
117 #endif
118
119 #ifdef CONFIG_IA64
120 extern int no_unaligned_warning;
121 #endif
122
123 #ifdef CONFIG_RT_MUTEXES
124 extern int max_lock_depth;
125 #endif
126
127 #ifdef CONFIG_SYSCTL_SYSCALL
128 static int parse_table(int __user *, int, void __user *, size_t __user *,
129 void __user *, size_t, ctl_table *);
130 #endif
131
132
133 #ifdef CONFIG_PROC_SYSCTL
134 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
135 void __user *buffer, size_t *lenp, loff_t *ppos);
136 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
137 void __user *buffer, size_t *lenp, loff_t *ppos);
138 #endif
139
140 static ctl_table root_table[];
141 static struct ctl_table_header root_table_header =
142 { root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
143
144 static ctl_table kern_table[];
145 static ctl_table vm_table[];
146 static ctl_table fs_table[];
147 static ctl_table debug_table[];
148 static ctl_table dev_table[];
149 extern ctl_table random_table[];
150 #ifdef CONFIG_UNIX98_PTYS
151 extern ctl_table pty_table[];
152 #endif
153 #ifdef CONFIG_INOTIFY_USER
154 extern ctl_table inotify_table[];
155 #endif
156
157 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
158 int sysctl_legacy_va_layout;
159 #endif
160
161
162 /* /proc declarations: */
163
164 #ifdef CONFIG_PROC_SYSCTL
165
166 static ssize_t proc_readsys(struct file *, char __user *, size_t, loff_t *);
167 static ssize_t proc_writesys(struct file *, const char __user *, size_t, loff_t *);
168 static int proc_opensys(struct inode *, struct file *);
169
170 const struct file_operations proc_sys_file_operations = {
171 .open = proc_opensys,
172 .read = proc_readsys,
173 .write = proc_writesys,
174 };
175
176 extern struct proc_dir_entry *proc_sys_root;
177
178 static void register_proc_table(ctl_table *, struct proc_dir_entry *, void *);
179 static void unregister_proc_table(ctl_table *, struct proc_dir_entry *);
180 #endif
181
182 /* The default sysctl tables: */
183
184 static ctl_table root_table[] = {
185 {
186 .ctl_name = CTL_KERN,
187 .procname = "kernel",
188 .mode = 0555,
189 .child = kern_table,
190 },
191 {
192 .ctl_name = CTL_VM,
193 .procname = "vm",
194 .mode = 0555,
195 .child = vm_table,
196 },
197 #ifdef CONFIG_NET
198 {
199 .ctl_name = CTL_NET,
200 .procname = "net",
201 .mode = 0555,
202 .child = net_table,
203 },
204 #endif
205 {
206 .ctl_name = CTL_FS,
207 .procname = "fs",
208 .mode = 0555,
209 .child = fs_table,
210 },
211 {
212 .ctl_name = CTL_DEBUG,
213 .procname = "debug",
214 .mode = 0555,
215 .child = debug_table,
216 },
217 {
218 .ctl_name = CTL_DEV,
219 .procname = "dev",
220 .mode = 0555,
221 .child = dev_table,
222 },
223
224 { .ctl_name = 0 }
225 };
226
227 static ctl_table kern_table[] = {
228 {
229 .ctl_name = KERN_PANIC,
230 .procname = "panic",
231 .data = &panic_timeout,
232 .maxlen = sizeof(int),
233 .mode = 0644,
234 .proc_handler = &proc_dointvec,
235 },
236 {
237 .ctl_name = KERN_CORE_USES_PID,
238 .procname = "core_uses_pid",
239 .data = &core_uses_pid,
240 .maxlen = sizeof(int),
241 .mode = 0644,
242 .proc_handler = &proc_dointvec,
243 },
244 {
245 .ctl_name = KERN_CORE_PATTERN,
246 .procname = "core_pattern",
247 .data = core_pattern,
248 .maxlen = 128,
249 .mode = 0644,
250 .proc_handler = &proc_dostring,
251 .strategy = &sysctl_string,
252 },
253 #ifdef CONFIG_PROC_SYSCTL
254 {
255 .ctl_name = KERN_TAINTED,
256 .procname = "tainted",
257 .data = &tainted,
258 .maxlen = sizeof(int),
259 .mode = 0644,
260 .proc_handler = &proc_dointvec_taint,
261 },
262 #endif
263 {
264 .ctl_name = KERN_CAP_BSET,
265 .procname = "cap-bound",
266 .data = &cap_bset,
267 .maxlen = sizeof(kernel_cap_t),
268 .mode = 0600,
269 .proc_handler = &proc_dointvec_bset,
270 },
271 #ifdef CONFIG_BLK_DEV_INITRD
272 {
273 .ctl_name = KERN_REALROOTDEV,
274 .procname = "real-root-dev",
275 .data = &real_root_dev,
276 .maxlen = sizeof(int),
277 .mode = 0644,
278 .proc_handler = &proc_dointvec,
279 },
280 #endif
281 #ifdef __sparc__
282 {
283 .ctl_name = KERN_SPARC_REBOOT,
284 .procname = "reboot-cmd",
285 .data = reboot_command,
286 .maxlen = 256,
287 .mode = 0644,
288 .proc_handler = &proc_dostring,
289 .strategy = &sysctl_string,
290 },
291 {
292 .ctl_name = KERN_SPARC_STOP_A,
293 .procname = "stop-a",
294 .data = &stop_a_enabled,
295 .maxlen = sizeof (int),
296 .mode = 0644,
297 .proc_handler = &proc_dointvec,
298 },
299 {
300 .ctl_name = KERN_SPARC_SCONS_PWROFF,
301 .procname = "scons-poweroff",
302 .data = &scons_pwroff,
303 .maxlen = sizeof (int),
304 .mode = 0644,
305 .proc_handler = &proc_dointvec,
306 },
307 #endif
308 #ifdef __hppa__
309 {
310 .ctl_name = KERN_HPPA_PWRSW,
311 .procname = "soft-power",
312 .data = &pwrsw_enabled,
313 .maxlen = sizeof (int),
314 .mode = 0644,
315 .proc_handler = &proc_dointvec,
316 },
317 {
318 .ctl_name = KERN_HPPA_UNALIGNED,
319 .procname = "unaligned-trap",
320 .data = &unaligned_enabled,
321 .maxlen = sizeof (int),
322 .mode = 0644,
323 .proc_handler = &proc_dointvec,
324 },
325 #endif
326 {
327 .ctl_name = KERN_CTLALTDEL,
328 .procname = "ctrl-alt-del",
329 .data = &C_A_D,
330 .maxlen = sizeof(int),
331 .mode = 0644,
332 .proc_handler = &proc_dointvec,
333 },
334 {
335 .ctl_name = KERN_PRINTK,
336 .procname = "printk",
337 .data = &console_loglevel,
338 .maxlen = 4*sizeof(int),
339 .mode = 0644,
340 .proc_handler = &proc_dointvec,
341 },
342 #ifdef CONFIG_KMOD
343 {
344 .ctl_name = KERN_MODPROBE,
345 .procname = "modprobe",
346 .data = &modprobe_path,
347 .maxlen = KMOD_PATH_LEN,
348 .mode = 0644,
349 .proc_handler = &proc_dostring,
350 .strategy = &sysctl_string,
351 },
352 #endif
353 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
354 {
355 .ctl_name = KERN_HOTPLUG,
356 .procname = "hotplug",
357 .data = &uevent_helper,
358 .maxlen = UEVENT_HELPER_PATH_LEN,
359 .mode = 0644,
360 .proc_handler = &proc_dostring,
361 .strategy = &sysctl_string,
362 },
363 #endif
364 #ifdef CONFIG_CHR_DEV_SG
365 {
366 .ctl_name = KERN_SG_BIG_BUFF,
367 .procname = "sg-big-buff",
368 .data = &sg_big_buff,
369 .maxlen = sizeof (int),
370 .mode = 0444,
371 .proc_handler = &proc_dointvec,
372 },
373 #endif
374 #ifdef CONFIG_BSD_PROCESS_ACCT
375 {
376 .ctl_name = KERN_ACCT,
377 .procname = "acct",
378 .data = &acct_parm,
379 .maxlen = 3*sizeof(int),
380 .mode = 0644,
381 .proc_handler = &proc_dointvec,
382 },
383 #endif
384 #ifdef CONFIG_MAGIC_SYSRQ
385 {
386 .ctl_name = KERN_SYSRQ,
387 .procname = "sysrq",
388 .data = &__sysrq_enabled,
389 .maxlen = sizeof (int),
390 .mode = 0644,
391 .proc_handler = &proc_dointvec,
392 },
393 #endif
394 #ifdef CONFIG_PROC_SYSCTL
395 {
396 .ctl_name = KERN_CADPID,
397 .procname = "cad_pid",
398 .data = NULL,
399 .maxlen = sizeof (int),
400 .mode = 0600,
401 .proc_handler = &proc_do_cad_pid,
402 },
403 #endif
404 {
405 .ctl_name = KERN_MAX_THREADS,
406 .procname = "threads-max",
407 .data = &max_threads,
408 .maxlen = sizeof(int),
409 .mode = 0644,
410 .proc_handler = &proc_dointvec,
411 },
412 {
413 .ctl_name = KERN_RANDOM,
414 .procname = "random",
415 .mode = 0555,
416 .child = random_table,
417 },
418 #ifdef CONFIG_UNIX98_PTYS
419 {
420 .ctl_name = KERN_PTY,
421 .procname = "pty",
422 .mode = 0555,
423 .child = pty_table,
424 },
425 #endif
426 {
427 .ctl_name = KERN_OVERFLOWUID,
428 .procname = "overflowuid",
429 .data = &overflowuid,
430 .maxlen = sizeof(int),
431 .mode = 0644,
432 .proc_handler = &proc_dointvec_minmax,
433 .strategy = &sysctl_intvec,
434 .extra1 = &minolduid,
435 .extra2 = &maxolduid,
436 },
437 {
438 .ctl_name = KERN_OVERFLOWGID,
439 .procname = "overflowgid",
440 .data = &overflowgid,
441 .maxlen = sizeof(int),
442 .mode = 0644,
443 .proc_handler = &proc_dointvec_minmax,
444 .strategy = &sysctl_intvec,
445 .extra1 = &minolduid,
446 .extra2 = &maxolduid,
447 },
448 #ifdef CONFIG_S390
449 #ifdef CONFIG_MATHEMU
450 {
451 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
452 .procname = "ieee_emulation_warnings",
453 .data = &sysctl_ieee_emulation_warnings,
454 .maxlen = sizeof(int),
455 .mode = 0644,
456 .proc_handler = &proc_dointvec,
457 },
458 #endif
459 #ifdef CONFIG_NO_IDLE_HZ
460 {
461 .ctl_name = KERN_HZ_TIMER,
462 .procname = "hz_timer",
463 .data = &sysctl_hz_timer,
464 .maxlen = sizeof(int),
465 .mode = 0644,
466 .proc_handler = &proc_dointvec,
467 },
468 #endif
469 {
470 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
471 .procname = "userprocess_debug",
472 .data = &sysctl_userprocess_debug,
473 .maxlen = sizeof(int),
474 .mode = 0644,
475 .proc_handler = &proc_dointvec,
476 },
477 #endif
478 {
479 .ctl_name = KERN_PIDMAX,
480 .procname = "pid_max",
481 .data = &pid_max,
482 .maxlen = sizeof (int),
483 .mode = 0644,
484 .proc_handler = &proc_dointvec_minmax,
485 .strategy = sysctl_intvec,
486 .extra1 = &pid_max_min,
487 .extra2 = &pid_max_max,
488 },
489 {
490 .ctl_name = KERN_PANIC_ON_OOPS,
491 .procname = "panic_on_oops",
492 .data = &panic_on_oops,
493 .maxlen = sizeof(int),
494 .mode = 0644,
495 .proc_handler = &proc_dointvec,
496 },
497 {
498 .ctl_name = KERN_PRINTK_RATELIMIT,
499 .procname = "printk_ratelimit",
500 .data = &printk_ratelimit_jiffies,
501 .maxlen = sizeof(int),
502 .mode = 0644,
503 .proc_handler = &proc_dointvec_jiffies,
504 .strategy = &sysctl_jiffies,
505 },
506 {
507 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
508 .procname = "printk_ratelimit_burst",
509 .data = &printk_ratelimit_burst,
510 .maxlen = sizeof(int),
511 .mode = 0644,
512 .proc_handler = &proc_dointvec,
513 },
514 {
515 .ctl_name = KERN_NGROUPS_MAX,
516 .procname = "ngroups_max",
517 .data = &ngroups_max,
518 .maxlen = sizeof (int),
519 .mode = 0444,
520 .proc_handler = &proc_dointvec,
521 },
522 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
523 {
524 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
525 .procname = "unknown_nmi_panic",
526 .data = &unknown_nmi_panic,
527 .maxlen = sizeof (int),
528 .mode = 0644,
529 .proc_handler = &proc_dointvec,
530 },
531 {
532 .ctl_name = KERN_NMI_WATCHDOG,
533 .procname = "nmi_watchdog",
534 .data = &nmi_watchdog_enabled,
535 .maxlen = sizeof (int),
536 .mode = 0644,
537 .proc_handler = &proc_nmi_enabled,
538 },
539 #endif
540 #if defined(CONFIG_X86)
541 {
542 .ctl_name = KERN_PANIC_ON_NMI,
543 .procname = "panic_on_unrecovered_nmi",
544 .data = &panic_on_unrecovered_nmi,
545 .maxlen = sizeof(int),
546 .mode = 0644,
547 .proc_handler = &proc_dointvec,
548 },
549 {
550 .ctl_name = KERN_BOOTLOADER_TYPE,
551 .procname = "bootloader_type",
552 .data = &bootloader_type,
553 .maxlen = sizeof (int),
554 .mode = 0444,
555 .proc_handler = &proc_dointvec,
556 },
557 {
558 .ctl_name = CTL_UNNUMBERED,
559 .procname = "kstack_depth_to_print",
560 .data = &kstack_depth_to_print,
561 .maxlen = sizeof(int),
562 .mode = 0644,
563 .proc_handler = &proc_dointvec,
564 },
565 #endif
566 #if defined(CONFIG_MMU)
567 {
568 .ctl_name = KERN_RANDOMIZE,
569 .procname = "randomize_va_space",
570 .data = &randomize_va_space,
571 .maxlen = sizeof(int),
572 .mode = 0644,
573 .proc_handler = &proc_dointvec,
574 },
575 #endif
576 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
577 {
578 .ctl_name = KERN_SPIN_RETRY,
579 .procname = "spin_retry",
580 .data = &spin_retry,
581 .maxlen = sizeof (int),
582 .mode = 0644,
583 .proc_handler = &proc_dointvec,
584 },
585 #endif
586 #ifdef CONFIG_ACPI_SLEEP
587 {
588 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
589 .procname = "acpi_video_flags",
590 .data = &acpi_video_flags,
591 .maxlen = sizeof (unsigned long),
592 .mode = 0644,
593 .proc_handler = &proc_doulongvec_minmax,
594 },
595 #endif
596 #ifdef CONFIG_IA64
597 {
598 .ctl_name = KERN_IA64_UNALIGNED,
599 .procname = "ignore-unaligned-usertrap",
600 .data = &no_unaligned_warning,
601 .maxlen = sizeof (int),
602 .mode = 0644,
603 .proc_handler = &proc_dointvec,
604 },
605 #endif
606 #ifdef CONFIG_COMPAT
607 {
608 .ctl_name = KERN_COMPAT_LOG,
609 .procname = "compat-log",
610 .data = &compat_log,
611 .maxlen = sizeof (int),
612 .mode = 0644,
613 .proc_handler = &proc_dointvec,
614 },
615 #endif
616 #ifdef CONFIG_RT_MUTEXES
617 {
618 .ctl_name = KERN_MAX_LOCK_DEPTH,
619 .procname = "max_lock_depth",
620 .data = &max_lock_depth,
621 .maxlen = sizeof(int),
622 .mode = 0644,
623 .proc_handler = &proc_dointvec,
624 },
625 #endif
626
627 { .ctl_name = 0 }
628 };
629
630 /* Constants for minimum and maximum testing in vm_table.
631 We use these as one-element integer vectors. */
632 static int zero;
633 static int one_hundred = 100;
634
635
636 static ctl_table vm_table[] = {
637 {
638 .ctl_name = VM_OVERCOMMIT_MEMORY,
639 .procname = "overcommit_memory",
640 .data = &sysctl_overcommit_memory,
641 .maxlen = sizeof(sysctl_overcommit_memory),
642 .mode = 0644,
643 .proc_handler = &proc_dointvec,
644 },
645 {
646 .ctl_name = VM_PANIC_ON_OOM,
647 .procname = "panic_on_oom",
648 .data = &sysctl_panic_on_oom,
649 .maxlen = sizeof(sysctl_panic_on_oom),
650 .mode = 0644,
651 .proc_handler = &proc_dointvec,
652 },
653 {
654 .ctl_name = VM_OVERCOMMIT_RATIO,
655 .procname = "overcommit_ratio",
656 .data = &sysctl_overcommit_ratio,
657 .maxlen = sizeof(sysctl_overcommit_ratio),
658 .mode = 0644,
659 .proc_handler = &proc_dointvec,
660 },
661 {
662 .ctl_name = VM_PAGE_CLUSTER,
663 .procname = "page-cluster",
664 .data = &page_cluster,
665 .maxlen = sizeof(int),
666 .mode = 0644,
667 .proc_handler = &proc_dointvec,
668 },
669 {
670 .ctl_name = VM_DIRTY_BACKGROUND,
671 .procname = "dirty_background_ratio",
672 .data = &dirty_background_ratio,
673 .maxlen = sizeof(dirty_background_ratio),
674 .mode = 0644,
675 .proc_handler = &proc_dointvec_minmax,
676 .strategy = &sysctl_intvec,
677 .extra1 = &zero,
678 .extra2 = &one_hundred,
679 },
680 {
681 .ctl_name = VM_DIRTY_RATIO,
682 .procname = "dirty_ratio",
683 .data = &vm_dirty_ratio,
684 .maxlen = sizeof(vm_dirty_ratio),
685 .mode = 0644,
686 .proc_handler = &proc_dointvec_minmax,
687 .strategy = &sysctl_intvec,
688 .extra1 = &zero,
689 .extra2 = &one_hundred,
690 },
691 {
692 .ctl_name = VM_DIRTY_WB_CS,
693 .procname = "dirty_writeback_centisecs",
694 .data = &dirty_writeback_interval,
695 .maxlen = sizeof(dirty_writeback_interval),
696 .mode = 0644,
697 .proc_handler = &dirty_writeback_centisecs_handler,
698 },
699 {
700 .ctl_name = VM_DIRTY_EXPIRE_CS,
701 .procname = "dirty_expire_centisecs",
702 .data = &dirty_expire_interval,
703 .maxlen = sizeof(dirty_expire_interval),
704 .mode = 0644,
705 .proc_handler = &proc_dointvec_userhz_jiffies,
706 },
707 {
708 .ctl_name = VM_NR_PDFLUSH_THREADS,
709 .procname = "nr_pdflush_threads",
710 .data = &nr_pdflush_threads,
711 .maxlen = sizeof nr_pdflush_threads,
712 .mode = 0444 /* read-only*/,
713 .proc_handler = &proc_dointvec,
714 },
715 {
716 .ctl_name = VM_SWAPPINESS,
717 .procname = "swappiness",
718 .data = &vm_swappiness,
719 .maxlen = sizeof(vm_swappiness),
720 .mode = 0644,
721 .proc_handler = &proc_dointvec_minmax,
722 .strategy = &sysctl_intvec,
723 .extra1 = &zero,
724 .extra2 = &one_hundred,
725 },
726 #ifdef CONFIG_HUGETLB_PAGE
727 {
728 .ctl_name = VM_HUGETLB_PAGES,
729 .procname = "nr_hugepages",
730 .data = &max_huge_pages,
731 .maxlen = sizeof(unsigned long),
732 .mode = 0644,
733 .proc_handler = &hugetlb_sysctl_handler,
734 .extra1 = (void *)&hugetlb_zero,
735 .extra2 = (void *)&hugetlb_infinity,
736 },
737 {
738 .ctl_name = VM_HUGETLB_GROUP,
739 .procname = "hugetlb_shm_group",
740 .data = &sysctl_hugetlb_shm_group,
741 .maxlen = sizeof(gid_t),
742 .mode = 0644,
743 .proc_handler = &proc_dointvec,
744 },
745 #endif
746 {
747 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
748 .procname = "lowmem_reserve_ratio",
749 .data = &sysctl_lowmem_reserve_ratio,
750 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
751 .mode = 0644,
752 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
753 .strategy = &sysctl_intvec,
754 },
755 {
756 .ctl_name = VM_DROP_PAGECACHE,
757 .procname = "drop_caches",
758 .data = &sysctl_drop_caches,
759 .maxlen = sizeof(int),
760 .mode = 0644,
761 .proc_handler = drop_caches_sysctl_handler,
762 .strategy = &sysctl_intvec,
763 },
764 {
765 .ctl_name = VM_MIN_FREE_KBYTES,
766 .procname = "min_free_kbytes",
767 .data = &min_free_kbytes,
768 .maxlen = sizeof(min_free_kbytes),
769 .mode = 0644,
770 .proc_handler = &min_free_kbytes_sysctl_handler,
771 .strategy = &sysctl_intvec,
772 .extra1 = &zero,
773 },
774 {
775 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
776 .procname = "percpu_pagelist_fraction",
777 .data = &percpu_pagelist_fraction,
778 .maxlen = sizeof(percpu_pagelist_fraction),
779 .mode = 0644,
780 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
781 .strategy = &sysctl_intvec,
782 .extra1 = &min_percpu_pagelist_fract,
783 },
784 #ifdef CONFIG_MMU
785 {
786 .ctl_name = VM_MAX_MAP_COUNT,
787 .procname = "max_map_count",
788 .data = &sysctl_max_map_count,
789 .maxlen = sizeof(sysctl_max_map_count),
790 .mode = 0644,
791 .proc_handler = &proc_dointvec
792 },
793 #endif
794 {
795 .ctl_name = VM_LAPTOP_MODE,
796 .procname = "laptop_mode",
797 .data = &laptop_mode,
798 .maxlen = sizeof(laptop_mode),
799 .mode = 0644,
800 .proc_handler = &proc_dointvec_jiffies,
801 .strategy = &sysctl_jiffies,
802 },
803 {
804 .ctl_name = VM_BLOCK_DUMP,
805 .procname = "block_dump",
806 .data = &block_dump,
807 .maxlen = sizeof(block_dump),
808 .mode = 0644,
809 .proc_handler = &proc_dointvec,
810 .strategy = &sysctl_intvec,
811 .extra1 = &zero,
812 },
813 {
814 .ctl_name = VM_VFS_CACHE_PRESSURE,
815 .procname = "vfs_cache_pressure",
816 .data = &sysctl_vfs_cache_pressure,
817 .maxlen = sizeof(sysctl_vfs_cache_pressure),
818 .mode = 0644,
819 .proc_handler = &proc_dointvec,
820 .strategy = &sysctl_intvec,
821 .extra1 = &zero,
822 },
823 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
824 {
825 .ctl_name = VM_LEGACY_VA_LAYOUT,
826 .procname = "legacy_va_layout",
827 .data = &sysctl_legacy_va_layout,
828 .maxlen = sizeof(sysctl_legacy_va_layout),
829 .mode = 0644,
830 .proc_handler = &proc_dointvec,
831 .strategy = &sysctl_intvec,
832 .extra1 = &zero,
833 },
834 #endif
835 #ifdef CONFIG_NUMA
836 {
837 .ctl_name = VM_ZONE_RECLAIM_MODE,
838 .procname = "zone_reclaim_mode",
839 .data = &zone_reclaim_mode,
840 .maxlen = sizeof(zone_reclaim_mode),
841 .mode = 0644,
842 .proc_handler = &proc_dointvec,
843 .strategy = &sysctl_intvec,
844 .extra1 = &zero,
845 },
846 {
847 .ctl_name = VM_MIN_UNMAPPED,
848 .procname = "min_unmapped_ratio",
849 .data = &sysctl_min_unmapped_ratio,
850 .maxlen = sizeof(sysctl_min_unmapped_ratio),
851 .mode = 0644,
852 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
853 .strategy = &sysctl_intvec,
854 .extra1 = &zero,
855 .extra2 = &one_hundred,
856 },
857 {
858 .ctl_name = VM_MIN_SLAB,
859 .procname = "min_slab_ratio",
860 .data = &sysctl_min_slab_ratio,
861 .maxlen = sizeof(sysctl_min_slab_ratio),
862 .mode = 0644,
863 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
864 .strategy = &sysctl_intvec,
865 .extra1 = &zero,
866 .extra2 = &one_hundred,
867 },
868 #endif
869 #ifdef CONFIG_X86_32
870 {
871 .ctl_name = VM_VDSO_ENABLED,
872 .procname = "vdso_enabled",
873 .data = &vdso_enabled,
874 .maxlen = sizeof(vdso_enabled),
875 .mode = 0644,
876 .proc_handler = &proc_dointvec,
877 .strategy = &sysctl_intvec,
878 .extra1 = &zero,
879 },
880 #endif
881 { .ctl_name = 0 }
882 };
883
884 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
885 static ctl_table binfmt_misc_table[] = {
886 { .ctl_name = 0 }
887 };
888 #endif
889
890 static ctl_table fs_table[] = {
891 {
892 .ctl_name = FS_NRINODE,
893 .procname = "inode-nr",
894 .data = &inodes_stat,
895 .maxlen = 2*sizeof(int),
896 .mode = 0444,
897 .proc_handler = &proc_dointvec,
898 },
899 {
900 .ctl_name = FS_STATINODE,
901 .procname = "inode-state",
902 .data = &inodes_stat,
903 .maxlen = 7*sizeof(int),
904 .mode = 0444,
905 .proc_handler = &proc_dointvec,
906 },
907 {
908 .ctl_name = FS_NRFILE,
909 .procname = "file-nr",
910 .data = &files_stat,
911 .maxlen = 3*sizeof(int),
912 .mode = 0444,
913 .proc_handler = &proc_nr_files,
914 },
915 {
916 .ctl_name = FS_MAXFILE,
917 .procname = "file-max",
918 .data = &files_stat.max_files,
919 .maxlen = sizeof(int),
920 .mode = 0644,
921 .proc_handler = &proc_dointvec,
922 },
923 {
924 .ctl_name = FS_DENTRY,
925 .procname = "dentry-state",
926 .data = &dentry_stat,
927 .maxlen = 6*sizeof(int),
928 .mode = 0444,
929 .proc_handler = &proc_dointvec,
930 },
931 {
932 .ctl_name = FS_OVERFLOWUID,
933 .procname = "overflowuid",
934 .data = &fs_overflowuid,
935 .maxlen = sizeof(int),
936 .mode = 0644,
937 .proc_handler = &proc_dointvec_minmax,
938 .strategy = &sysctl_intvec,
939 .extra1 = &minolduid,
940 .extra2 = &maxolduid,
941 },
942 {
943 .ctl_name = FS_OVERFLOWGID,
944 .procname = "overflowgid",
945 .data = &fs_overflowgid,
946 .maxlen = sizeof(int),
947 .mode = 0644,
948 .proc_handler = &proc_dointvec_minmax,
949 .strategy = &sysctl_intvec,
950 .extra1 = &minolduid,
951 .extra2 = &maxolduid,
952 },
953 {
954 .ctl_name = FS_LEASES,
955 .procname = "leases-enable",
956 .data = &leases_enable,
957 .maxlen = sizeof(int),
958 .mode = 0644,
959 .proc_handler = &proc_dointvec,
960 },
961 #ifdef CONFIG_DNOTIFY
962 {
963 .ctl_name = FS_DIR_NOTIFY,
964 .procname = "dir-notify-enable",
965 .data = &dir_notify_enable,
966 .maxlen = sizeof(int),
967 .mode = 0644,
968 .proc_handler = &proc_dointvec,
969 },
970 #endif
971 #ifdef CONFIG_MMU
972 {
973 .ctl_name = FS_LEASE_TIME,
974 .procname = "lease-break-time",
975 .data = &lease_break_time,
976 .maxlen = sizeof(int),
977 .mode = 0644,
978 .proc_handler = &proc_dointvec,
979 },
980 {
981 .ctl_name = FS_AIO_NR,
982 .procname = "aio-nr",
983 .data = &aio_nr,
984 .maxlen = sizeof(aio_nr),
985 .mode = 0444,
986 .proc_handler = &proc_doulongvec_minmax,
987 },
988 {
989 .ctl_name = FS_AIO_MAX_NR,
990 .procname = "aio-max-nr",
991 .data = &aio_max_nr,
992 .maxlen = sizeof(aio_max_nr),
993 .mode = 0644,
994 .proc_handler = &proc_doulongvec_minmax,
995 },
996 #ifdef CONFIG_INOTIFY_USER
997 {
998 .ctl_name = FS_INOTIFY,
999 .procname = "inotify",
1000 .mode = 0555,
1001 .child = inotify_table,
1002 },
1003 #endif
1004 #endif
1005 {
1006 .ctl_name = KERN_SETUID_DUMPABLE,
1007 .procname = "suid_dumpable",
1008 .data = &suid_dumpable,
1009 .maxlen = sizeof(int),
1010 .mode = 0644,
1011 .proc_handler = &proc_dointvec,
1012 },
1013 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1014 {
1015 .ctl_name = CTL_UNNUMBERED,
1016 .procname = "binfmt_misc",
1017 .mode = 0555,
1018 .child = binfmt_misc_table,
1019 },
1020 #endif
1021 { .ctl_name = 0 }
1022 };
1023
1024 static ctl_table debug_table[] = {
1025 { .ctl_name = 0 }
1026 };
1027
1028 static ctl_table dev_table[] = {
1029 { .ctl_name = 0 }
1030 };
1031
1032 static DEFINE_SPINLOCK(sysctl_lock);
1033
1034 /* called under sysctl_lock */
1035 static int use_table(struct ctl_table_header *p)
1036 {
1037 if (unlikely(p->unregistering))
1038 return 0;
1039 p->used++;
1040 return 1;
1041 }
1042
1043 /* called under sysctl_lock */
1044 static void unuse_table(struct ctl_table_header *p)
1045 {
1046 if (!--p->used)
1047 if (unlikely(p->unregistering))
1048 complete(p->unregistering);
1049 }
1050
1051 /* called under sysctl_lock, will reacquire if has to wait */
1052 static void start_unregistering(struct ctl_table_header *p)
1053 {
1054 /*
1055 * if p->used is 0, nobody will ever touch that entry again;
1056 * we'll eliminate all paths to it before dropping sysctl_lock
1057 */
1058 if (unlikely(p->used)) {
1059 struct completion wait;
1060 init_completion(&wait);
1061 p->unregistering = &wait;
1062 spin_unlock(&sysctl_lock);
1063 wait_for_completion(&wait);
1064 spin_lock(&sysctl_lock);
1065 }
1066 /*
1067 * do not remove from the list until nobody holds it; walking the
1068 * list in do_sysctl() relies on that.
1069 */
1070 list_del_init(&p->ctl_entry);
1071 }
1072
1073 void sysctl_head_finish(struct ctl_table_header *head)
1074 {
1075 if (!head)
1076 return;
1077 spin_lock(&sysctl_lock);
1078 unuse_table(head);
1079 spin_unlock(&sysctl_lock);
1080 }
1081
1082 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1083 {
1084 struct ctl_table_header *head;
1085 struct list_head *tmp;
1086 spin_lock(&sysctl_lock);
1087 if (prev) {
1088 tmp = &prev->ctl_entry;
1089 unuse_table(prev);
1090 goto next;
1091 }
1092 tmp = &root_table_header.ctl_entry;
1093 for (;;) {
1094 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1095
1096 if (!use_table(head))
1097 goto next;
1098 spin_unlock(&sysctl_lock);
1099 return head;
1100 next:
1101 tmp = tmp->next;
1102 if (tmp == &root_table_header.ctl_entry)
1103 break;
1104 }
1105 spin_unlock(&sysctl_lock);
1106 return NULL;
1107 }
1108
1109 void __init sysctl_init(void)
1110 {
1111 #ifdef CONFIG_PROC_SYSCTL
1112 register_proc_table(root_table, proc_sys_root, &root_table_header);
1113 #endif
1114 }
1115
1116 #ifdef CONFIG_SYSCTL_SYSCALL
1117 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1118 void __user *newval, size_t newlen)
1119 {
1120 struct ctl_table_header *head;
1121 int error = -ENOTDIR;
1122
1123 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1124 return -ENOTDIR;
1125 if (oldval) {
1126 int old_len;
1127 if (!oldlenp || get_user(old_len, oldlenp))
1128 return -EFAULT;
1129 }
1130
1131 for (head = sysctl_head_next(NULL); head;
1132 head = sysctl_head_next(head)) {
1133 error = parse_table(name, nlen, oldval, oldlenp,
1134 newval, newlen, head->ctl_table);
1135 if (error != -ENOTDIR) {
1136 sysctl_head_finish(head);
1137 break;
1138 }
1139 }
1140 return error;
1141 }
1142
1143 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1144 {
1145 struct __sysctl_args tmp;
1146 int error;
1147
1148 if (copy_from_user(&tmp, args, sizeof(tmp)))
1149 return -EFAULT;
1150
1151 lock_kernel();
1152 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1153 tmp.newval, tmp.newlen);
1154 unlock_kernel();
1155 return error;
1156 }
1157 #endif /* CONFIG_SYSCTL_SYSCALL */
1158
1159 /*
1160 * ctl_perm does NOT grant the superuser all rights automatically, because
1161 * some sysctl variables are readonly even to root.
1162 */
1163
1164 static int test_perm(int mode, int op)
1165 {
1166 if (!current->euid)
1167 mode >>= 6;
1168 else if (in_egroup_p(0))
1169 mode >>= 3;
1170 if ((mode & op & 0007) == op)
1171 return 0;
1172 return -EACCES;
1173 }
1174
1175 static inline int ctl_perm(ctl_table *table, int op)
1176 {
1177 int error;
1178 error = security_sysctl(table, op);
1179 if (error)
1180 return error;
1181 return test_perm(table->mode, op);
1182 }
1183
1184 #ifdef CONFIG_SYSCTL_SYSCALL
1185 static int parse_table(int __user *name, int nlen,
1186 void __user *oldval, size_t __user *oldlenp,
1187 void __user *newval, size_t newlen,
1188 ctl_table *table)
1189 {
1190 int n;
1191 repeat:
1192 if (!nlen)
1193 return -ENOTDIR;
1194 if (get_user(n, name))
1195 return -EFAULT;
1196 for ( ; table->ctl_name || table->procname; table++) {
1197 if (!table->ctl_name)
1198 continue;
1199 if (n == table->ctl_name) {
1200 int error;
1201 if (table->child) {
1202 if (ctl_perm(table, 001))
1203 return -EPERM;
1204 name++;
1205 nlen--;
1206 table = table->child;
1207 goto repeat;
1208 }
1209 error = do_sysctl_strategy(table, name, nlen,
1210 oldval, oldlenp,
1211 newval, newlen);
1212 return error;
1213 }
1214 }
1215 return -ENOTDIR;
1216 }
1217
1218 /* Perform the actual read/write of a sysctl table entry. */
1219 int do_sysctl_strategy (ctl_table *table,
1220 int __user *name, int nlen,
1221 void __user *oldval, size_t __user *oldlenp,
1222 void __user *newval, size_t newlen)
1223 {
1224 int op = 0, rc;
1225 size_t len;
1226
1227 if (oldval)
1228 op |= 004;
1229 if (newval)
1230 op |= 002;
1231 if (ctl_perm(table, op))
1232 return -EPERM;
1233
1234 if (table->strategy) {
1235 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1236 newval, newlen);
1237 if (rc < 0)
1238 return rc;
1239 if (rc > 0)
1240 return 0;
1241 }
1242
1243 /* If there is no strategy routine, or if the strategy returns
1244 * zero, proceed with automatic r/w */
1245 if (table->data && table->maxlen) {
1246 if (oldval && oldlenp) {
1247 if (get_user(len, oldlenp))
1248 return -EFAULT;
1249 if (len) {
1250 if (len > table->maxlen)
1251 len = table->maxlen;
1252 if(copy_to_user(oldval, table->data, len))
1253 return -EFAULT;
1254 if(put_user(len, oldlenp))
1255 return -EFAULT;
1256 }
1257 }
1258 if (newval && newlen) {
1259 len = newlen;
1260 if (len > table->maxlen)
1261 len = table->maxlen;
1262 if(copy_from_user(table->data, newval, len))
1263 return -EFAULT;
1264 }
1265 }
1266 return 0;
1267 }
1268 #endif /* CONFIG_SYSCTL_SYSCALL */
1269
1270 /**
1271 * register_sysctl_table - register a sysctl hierarchy
1272 * @table: the top-level table structure
1273 *
1274 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1275 * array. An entry with a ctl_name of 0 terminates the table.
1276 *
1277 * The members of the &ctl_table structure are used as follows:
1278 *
1279 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1280 * must be unique within that level of sysctl
1281 *
1282 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1283 * enter a sysctl file
1284 *
1285 * data - a pointer to data for use by proc_handler
1286 *
1287 * maxlen - the maximum size in bytes of the data
1288 *
1289 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1290 *
1291 * child - a pointer to the child sysctl table if this entry is a directory, or
1292 * %NULL.
1293 *
1294 * proc_handler - the text handler routine (described below)
1295 *
1296 * strategy - the strategy routine (described below)
1297 *
1298 * de - for internal use by the sysctl routines
1299 *
1300 * extra1, extra2 - extra pointers usable by the proc handler routines
1301 *
1302 * Leaf nodes in the sysctl tree will be represented by a single file
1303 * under /proc; non-leaf nodes will be represented by directories.
1304 *
1305 * sysctl(2) can automatically manage read and write requests through
1306 * the sysctl table. The data and maxlen fields of the ctl_table
1307 * struct enable minimal validation of the values being written to be
1308 * performed, and the mode field allows minimal authentication.
1309 *
1310 * More sophisticated management can be enabled by the provision of a
1311 * strategy routine with the table entry. This will be called before
1312 * any automatic read or write of the data is performed.
1313 *
1314 * The strategy routine may return
1315 *
1316 * < 0 - Error occurred (error is passed to user process)
1317 *
1318 * 0 - OK - proceed with automatic read or write.
1319 *
1320 * > 0 - OK - read or write has been done by the strategy routine, so
1321 * return immediately.
1322 *
1323 * There must be a proc_handler routine for any terminal nodes
1324 * mirrored under /proc/sys (non-terminals are handled by a built-in
1325 * directory handler). Several default handlers are available to
1326 * cover common cases -
1327 *
1328 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1329 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1330 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1331 *
1332 * It is the handler's job to read the input buffer from user memory
1333 * and process it. The handler should return 0 on success.
1334 *
1335 * This routine returns %NULL on a failure to register, and a pointer
1336 * to the table header on success.
1337 */
1338 struct ctl_table_header *register_sysctl_table(ctl_table * table)
1339 {
1340 struct ctl_table_header *tmp;
1341 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1342 if (!tmp)
1343 return NULL;
1344 tmp->ctl_table = table;
1345 INIT_LIST_HEAD(&tmp->ctl_entry);
1346 tmp->used = 0;
1347 tmp->unregistering = NULL;
1348 spin_lock(&sysctl_lock);
1349 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
1350 spin_unlock(&sysctl_lock);
1351 #ifdef CONFIG_PROC_SYSCTL
1352 register_proc_table(table, proc_sys_root, tmp);
1353 #endif
1354 return tmp;
1355 }
1356
1357 /**
1358 * unregister_sysctl_table - unregister a sysctl table hierarchy
1359 * @header: the header returned from register_sysctl_table
1360 *
1361 * Unregisters the sysctl table and all children. proc entries may not
1362 * actually be removed until they are no longer used by anyone.
1363 */
1364 void unregister_sysctl_table(struct ctl_table_header * header)
1365 {
1366 might_sleep();
1367 spin_lock(&sysctl_lock);
1368 start_unregistering(header);
1369 #ifdef CONFIG_PROC_SYSCTL
1370 unregister_proc_table(header->ctl_table, proc_sys_root);
1371 #endif
1372 spin_unlock(&sysctl_lock);
1373 kfree(header);
1374 }
1375
1376 #else /* !CONFIG_SYSCTL */
1377 struct ctl_table_header * register_sysctl_table(ctl_table * table,
1378 int insert_at_head)
1379 {
1380 return NULL;
1381 }
1382
1383 void unregister_sysctl_table(struct ctl_table_header * table)
1384 {
1385 }
1386
1387 #endif /* CONFIG_SYSCTL */
1388
1389 /*
1390 * /proc/sys support
1391 */
1392
1393 #ifdef CONFIG_PROC_SYSCTL
1394
1395 /* Scan the sysctl entries in table and add them all into /proc */
1396 static void register_proc_table(ctl_table * table, struct proc_dir_entry *root, void *set)
1397 {
1398 struct proc_dir_entry *de;
1399 int len;
1400 mode_t mode;
1401
1402 for (; table->ctl_name || table->procname; table++) {
1403 /* Can't do anything without a proc name. */
1404 if (!table->procname)
1405 continue;
1406 /* Maybe we can't do anything with it... */
1407 if (!table->proc_handler && !table->child) {
1408 printk(KERN_WARNING "SYSCTL: Can't register %s\n",
1409 table->procname);
1410 continue;
1411 }
1412
1413 len = strlen(table->procname);
1414 mode = table->mode;
1415
1416 de = NULL;
1417 if (table->proc_handler)
1418 mode |= S_IFREG;
1419 else {
1420 mode |= S_IFDIR;
1421 for (de = root->subdir; de; de = de->next) {
1422 if (proc_match(len, table->procname, de))
1423 break;
1424 }
1425 /* If the subdir exists already, de is non-NULL */
1426 }
1427
1428 if (!de) {
1429 de = create_proc_entry(table->procname, mode, root);
1430 if (!de)
1431 continue;
1432 de->set = set;
1433 de->data = (void *) table;
1434 if (table->proc_handler)
1435 de->proc_fops = &proc_sys_file_operations;
1436 }
1437 table->de = de;
1438 if (de->mode & S_IFDIR)
1439 register_proc_table(table->child, de, set);
1440 }
1441 }
1442
1443 /*
1444 * Unregister a /proc sysctl table and any subdirectories.
1445 */
1446 static void unregister_proc_table(ctl_table * table, struct proc_dir_entry *root)
1447 {
1448 struct proc_dir_entry *de;
1449 for (; table->ctl_name || table->procname; table++) {
1450 if (!(de = table->de))
1451 continue;
1452 if (de->mode & S_IFDIR) {
1453 if (!table->child) {
1454 printk (KERN_ALERT "Help - malformed sysctl tree on free\n");
1455 continue;
1456 }
1457 unregister_proc_table(table->child, de);
1458
1459 /* Don't unregister directories which still have entries.. */
1460 if (de->subdir)
1461 continue;
1462 }
1463
1464 /*
1465 * In any case, mark the entry as goner; we'll keep it
1466 * around if it's busy, but we'll know to do nothing with
1467 * its fields. We are under sysctl_lock here.
1468 */
1469 de->data = NULL;
1470
1471 /* Don't unregister proc entries that are still being used.. */
1472 if (atomic_read(&de->count))
1473 continue;
1474
1475 table->de = NULL;
1476 remove_proc_entry(table->procname, root);
1477 }
1478 }
1479
1480 static ssize_t do_rw_proc(int write, struct file * file, char __user * buf,
1481 size_t count, loff_t *ppos)
1482 {
1483 int op;
1484 struct proc_dir_entry *de = PDE(file->f_path.dentry->d_inode);
1485 struct ctl_table *table;
1486 size_t res;
1487 ssize_t error = -ENOTDIR;
1488
1489 spin_lock(&sysctl_lock);
1490 if (de && de->data && use_table(de->set)) {
1491 /*
1492 * at that point we know that sysctl was not unregistered
1493 * and won't be until we finish
1494 */
1495 spin_unlock(&sysctl_lock);
1496 table = (struct ctl_table *) de->data;
1497 if (!table || !table->proc_handler)
1498 goto out;
1499 error = -EPERM;
1500 op = (write ? 002 : 004);
1501 if (ctl_perm(table, op))
1502 goto out;
1503
1504 /* careful: calling conventions are nasty here */
1505 res = count;
1506 error = (*table->proc_handler)(table, write, file,
1507 buf, &res, ppos);
1508 if (!error)
1509 error = res;
1510 out:
1511 spin_lock(&sysctl_lock);
1512 unuse_table(de->set);
1513 }
1514 spin_unlock(&sysctl_lock);
1515 return error;
1516 }
1517
1518 static int proc_opensys(struct inode *inode, struct file *file)
1519 {
1520 if (file->f_mode & FMODE_WRITE) {
1521 /*
1522 * sysctl entries that are not writable,
1523 * are _NOT_ writable, capabilities or not.
1524 */
1525 if (!(inode->i_mode & S_IWUSR))
1526 return -EPERM;
1527 }
1528
1529 return 0;
1530 }
1531
1532 static ssize_t proc_readsys(struct file * file, char __user * buf,
1533 size_t count, loff_t *ppos)
1534 {
1535 return do_rw_proc(0, file, buf, count, ppos);
1536 }
1537
1538 static ssize_t proc_writesys(struct file * file, const char __user * buf,
1539 size_t count, loff_t *ppos)
1540 {
1541 return do_rw_proc(1, file, (char __user *) buf, count, ppos);
1542 }
1543
1544 static int _proc_do_string(void* data, int maxlen, int write,
1545 struct file *filp, void __user *buffer,
1546 size_t *lenp, loff_t *ppos)
1547 {
1548 size_t len;
1549 char __user *p;
1550 char c;
1551
1552 if (!data || !maxlen || !*lenp) {
1553 *lenp = 0;
1554 return 0;
1555 }
1556
1557 if (write) {
1558 len = 0;
1559 p = buffer;
1560 while (len < *lenp) {
1561 if (get_user(c, p++))
1562 return -EFAULT;
1563 if (c == 0 || c == '\n')
1564 break;
1565 len++;
1566 }
1567 if (len >= maxlen)
1568 len = maxlen-1;
1569 if(copy_from_user(data, buffer, len))
1570 return -EFAULT;
1571 ((char *) data)[len] = 0;
1572 *ppos += *lenp;
1573 } else {
1574 len = strlen(data);
1575 if (len > maxlen)
1576 len = maxlen;
1577
1578 if (*ppos > len) {
1579 *lenp = 0;
1580 return 0;
1581 }
1582
1583 data += *ppos;
1584 len -= *ppos;
1585
1586 if (len > *lenp)
1587 len = *lenp;
1588 if (len)
1589 if(copy_to_user(buffer, data, len))
1590 return -EFAULT;
1591 if (len < *lenp) {
1592 if(put_user('\n', ((char __user *) buffer) + len))
1593 return -EFAULT;
1594 len++;
1595 }
1596 *lenp = len;
1597 *ppos += len;
1598 }
1599 return 0;
1600 }
1601
1602 /**
1603 * proc_dostring - read a string sysctl
1604 * @table: the sysctl table
1605 * @write: %TRUE if this is a write to the sysctl file
1606 * @filp: the file structure
1607 * @buffer: the user buffer
1608 * @lenp: the size of the user buffer
1609 * @ppos: file position
1610 *
1611 * Reads/writes a string from/to the user buffer. If the kernel
1612 * buffer provided is not large enough to hold the string, the
1613 * string is truncated. The copied string is %NULL-terminated.
1614 * If the string is being read by the user process, it is copied
1615 * and a newline '\n' is added. It is truncated if the buffer is
1616 * not large enough.
1617 *
1618 * Returns 0 on success.
1619 */
1620 int proc_dostring(ctl_table *table, int write, struct file *filp,
1621 void __user *buffer, size_t *lenp, loff_t *ppos)
1622 {
1623 return _proc_do_string(table->data, table->maxlen, write, filp,
1624 buffer, lenp, ppos);
1625 }
1626
1627
1628 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1629 int *valp,
1630 int write, void *data)
1631 {
1632 if (write) {
1633 *valp = *negp ? -*lvalp : *lvalp;
1634 } else {
1635 int val = *valp;
1636 if (val < 0) {
1637 *negp = -1;
1638 *lvalp = (unsigned long)-val;
1639 } else {
1640 *negp = 0;
1641 *lvalp = (unsigned long)val;
1642 }
1643 }
1644 return 0;
1645 }
1646
1647 static int __do_proc_dointvec(void *tbl_data, ctl_table *table,
1648 int write, struct file *filp, void __user *buffer,
1649 size_t *lenp, loff_t *ppos,
1650 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1651 int write, void *data),
1652 void *data)
1653 {
1654 #define TMPBUFLEN 21
1655 int *i, vleft, first=1, neg, val;
1656 unsigned long lval;
1657 size_t left, len;
1658
1659 char buf[TMPBUFLEN], *p;
1660 char __user *s = buffer;
1661
1662 if (!tbl_data || !table->maxlen || !*lenp ||
1663 (*ppos && !write)) {
1664 *lenp = 0;
1665 return 0;
1666 }
1667
1668 i = (int *) tbl_data;
1669 vleft = table->maxlen / sizeof(*i);
1670 left = *lenp;
1671
1672 if (!conv)
1673 conv = do_proc_dointvec_conv;
1674
1675 for (; left && vleft--; i++, first=0) {
1676 if (write) {
1677 while (left) {
1678 char c;
1679 if (get_user(c, s))
1680 return -EFAULT;
1681 if (!isspace(c))
1682 break;
1683 left--;
1684 s++;
1685 }
1686 if (!left)
1687 break;
1688 neg = 0;
1689 len = left;
1690 if (len > sizeof(buf) - 1)
1691 len = sizeof(buf) - 1;
1692 if (copy_from_user(buf, s, len))
1693 return -EFAULT;
1694 buf[len] = 0;
1695 p = buf;
1696 if (*p == '-' && left > 1) {
1697 neg = 1;
1698 p++;
1699 }
1700 if (*p < '0' || *p > '9')
1701 break;
1702
1703 lval = simple_strtoul(p, &p, 0);
1704
1705 len = p-buf;
1706 if ((len < left) && *p && !isspace(*p))
1707 break;
1708 if (neg)
1709 val = -val;
1710 s += len;
1711 left -= len;
1712
1713 if (conv(&neg, &lval, i, 1, data))
1714 break;
1715 } else {
1716 p = buf;
1717 if (!first)
1718 *p++ = '\t';
1719
1720 if (conv(&neg, &lval, i, 0, data))
1721 break;
1722
1723 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1724 len = strlen(buf);
1725 if (len > left)
1726 len = left;
1727 if(copy_to_user(s, buf, len))
1728 return -EFAULT;
1729 left -= len;
1730 s += len;
1731 }
1732 }
1733
1734 if (!write && !first && left) {
1735 if(put_user('\n', s))
1736 return -EFAULT;
1737 left--, s++;
1738 }
1739 if (write) {
1740 while (left) {
1741 char c;
1742 if (get_user(c, s++))
1743 return -EFAULT;
1744 if (!isspace(c))
1745 break;
1746 left--;
1747 }
1748 }
1749 if (write && first)
1750 return -EINVAL;
1751 *lenp -= left;
1752 *ppos += *lenp;
1753 return 0;
1754 #undef TMPBUFLEN
1755 }
1756
1757 static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1758 void __user *buffer, size_t *lenp, loff_t *ppos,
1759 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1760 int write, void *data),
1761 void *data)
1762 {
1763 return __do_proc_dointvec(table->data, table, write, filp,
1764 buffer, lenp, ppos, conv, data);
1765 }
1766
1767 /**
1768 * proc_dointvec - read a vector of integers
1769 * @table: the sysctl table
1770 * @write: %TRUE if this is a write to the sysctl file
1771 * @filp: the file structure
1772 * @buffer: the user buffer
1773 * @lenp: the size of the user buffer
1774 * @ppos: file position
1775 *
1776 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1777 * values from/to the user buffer, treated as an ASCII string.
1778 *
1779 * Returns 0 on success.
1780 */
1781 int proc_dointvec(ctl_table *table, int write, struct file *filp,
1782 void __user *buffer, size_t *lenp, loff_t *ppos)
1783 {
1784 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1785 NULL,NULL);
1786 }
1787
1788 #define OP_SET 0
1789 #define OP_AND 1
1790 #define OP_OR 2
1791
1792 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1793 int *valp,
1794 int write, void *data)
1795 {
1796 int op = *(int *)data;
1797 if (write) {
1798 int val = *negp ? -*lvalp : *lvalp;
1799 switch(op) {
1800 case OP_SET: *valp = val; break;
1801 case OP_AND: *valp &= val; break;
1802 case OP_OR: *valp |= val; break;
1803 }
1804 } else {
1805 int val = *valp;
1806 if (val < 0) {
1807 *negp = -1;
1808 *lvalp = (unsigned long)-val;
1809 } else {
1810 *negp = 0;
1811 *lvalp = (unsigned long)val;
1812 }
1813 }
1814 return 0;
1815 }
1816
1817 /*
1818 * init may raise the set.
1819 */
1820
1821 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1822 void __user *buffer, size_t *lenp, loff_t *ppos)
1823 {
1824 int op;
1825
1826 if (write && !capable(CAP_SYS_MODULE)) {
1827 return -EPERM;
1828 }
1829
1830 op = is_init(current) ? OP_SET : OP_AND;
1831 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1832 do_proc_dointvec_bset_conv,&op);
1833 }
1834
1835 /*
1836 * Taint values can only be increased
1837 */
1838 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
1839 void __user *buffer, size_t *lenp, loff_t *ppos)
1840 {
1841 int op;
1842
1843 if (!capable(CAP_SYS_ADMIN))
1844 return -EPERM;
1845
1846 op = OP_OR;
1847 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1848 do_proc_dointvec_bset_conv,&op);
1849 }
1850
1851 struct do_proc_dointvec_minmax_conv_param {
1852 int *min;
1853 int *max;
1854 };
1855
1856 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1857 int *valp,
1858 int write, void *data)
1859 {
1860 struct do_proc_dointvec_minmax_conv_param *param = data;
1861 if (write) {
1862 int val = *negp ? -*lvalp : *lvalp;
1863 if ((param->min && *param->min > val) ||
1864 (param->max && *param->max < val))
1865 return -EINVAL;
1866 *valp = val;
1867 } else {
1868 int val = *valp;
1869 if (val < 0) {
1870 *negp = -1;
1871 *lvalp = (unsigned long)-val;
1872 } else {
1873 *negp = 0;
1874 *lvalp = (unsigned long)val;
1875 }
1876 }
1877 return 0;
1878 }
1879
1880 /**
1881 * proc_dointvec_minmax - read a vector of integers with min/max values
1882 * @table: the sysctl table
1883 * @write: %TRUE if this is a write to the sysctl file
1884 * @filp: the file structure
1885 * @buffer: the user buffer
1886 * @lenp: the size of the user buffer
1887 * @ppos: file position
1888 *
1889 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1890 * values from/to the user buffer, treated as an ASCII string.
1891 *
1892 * This routine will ensure the values are within the range specified by
1893 * table->extra1 (min) and table->extra2 (max).
1894 *
1895 * Returns 0 on success.
1896 */
1897 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1898 void __user *buffer, size_t *lenp, loff_t *ppos)
1899 {
1900 struct do_proc_dointvec_minmax_conv_param param = {
1901 .min = (int *) table->extra1,
1902 .max = (int *) table->extra2,
1903 };
1904 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1905 do_proc_dointvec_minmax_conv, &param);
1906 }
1907
1908 static int __do_proc_doulongvec_minmax(void *data, ctl_table *table, int write,
1909 struct file *filp,
1910 void __user *buffer,
1911 size_t *lenp, loff_t *ppos,
1912 unsigned long convmul,
1913 unsigned long convdiv)
1914 {
1915 #define TMPBUFLEN 21
1916 unsigned long *i, *min, *max, val;
1917 int vleft, first=1, neg;
1918 size_t len, left;
1919 char buf[TMPBUFLEN], *p;
1920 char __user *s = buffer;
1921
1922 if (!data || !table->maxlen || !*lenp ||
1923 (*ppos && !write)) {
1924 *lenp = 0;
1925 return 0;
1926 }
1927
1928 i = (unsigned long *) data;
1929 min = (unsigned long *) table->extra1;
1930 max = (unsigned long *) table->extra2;
1931 vleft = table->maxlen / sizeof(unsigned long);
1932 left = *lenp;
1933
1934 for (; left && vleft--; i++, min++, max++, first=0) {
1935 if (write) {
1936 while (left) {
1937 char c;
1938 if (get_user(c, s))
1939 return -EFAULT;
1940 if (!isspace(c))
1941 break;
1942 left--;
1943 s++;
1944 }
1945 if (!left)
1946 break;
1947 neg = 0;
1948 len = left;
1949 if (len > TMPBUFLEN-1)
1950 len = TMPBUFLEN-1;
1951 if (copy_from_user(buf, s, len))
1952 return -EFAULT;
1953 buf[len] = 0;
1954 p = buf;
1955 if (*p == '-' && left > 1) {
1956 neg = 1;
1957 p++;
1958 }
1959 if (*p < '0' || *p > '9')
1960 break;
1961 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
1962 len = p-buf;
1963 if ((len < left) && *p && !isspace(*p))
1964 break;
1965 if (neg)
1966 val = -val;
1967 s += len;
1968 left -= len;
1969
1970 if(neg)
1971 continue;
1972 if ((min && val < *min) || (max && val > *max))
1973 continue;
1974 *i = val;
1975 } else {
1976 p = buf;
1977 if (!first)
1978 *p++ = '\t';
1979 sprintf(p, "%lu", convdiv * (*i) / convmul);
1980 len = strlen(buf);
1981 if (len > left)
1982 len = left;
1983 if(copy_to_user(s, buf, len))
1984 return -EFAULT;
1985 left -= len;
1986 s += len;
1987 }
1988 }
1989
1990 if (!write && !first && left) {
1991 if(put_user('\n', s))
1992 return -EFAULT;
1993 left--, s++;
1994 }
1995 if (write) {
1996 while (left) {
1997 char c;
1998 if (get_user(c, s++))
1999 return -EFAULT;
2000 if (!isspace(c))
2001 break;
2002 left--;
2003 }
2004 }
2005 if (write && first)
2006 return -EINVAL;
2007 *lenp -= left;
2008 *ppos += *lenp;
2009 return 0;
2010 #undef TMPBUFLEN
2011 }
2012
2013 static int do_proc_doulongvec_minmax(ctl_table *table, int write,
2014 struct file *filp,
2015 void __user *buffer,
2016 size_t *lenp, loff_t *ppos,
2017 unsigned long convmul,
2018 unsigned long convdiv)
2019 {
2020 return __do_proc_doulongvec_minmax(table->data, table, write,
2021 filp, buffer, lenp, ppos, convmul, convdiv);
2022 }
2023
2024 /**
2025 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2026 * @table: the sysctl table
2027 * @write: %TRUE if this is a write to the sysctl file
2028 * @filp: the file structure
2029 * @buffer: the user buffer
2030 * @lenp: the size of the user buffer
2031 * @ppos: file position
2032 *
2033 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2034 * values from/to the user buffer, treated as an ASCII string.
2035 *
2036 * This routine will ensure the values are within the range specified by
2037 * table->extra1 (min) and table->extra2 (max).
2038 *
2039 * Returns 0 on success.
2040 */
2041 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2042 void __user *buffer, size_t *lenp, loff_t *ppos)
2043 {
2044 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2045 }
2046
2047 /**
2048 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2049 * @table: the sysctl table
2050 * @write: %TRUE if this is a write to the sysctl file
2051 * @filp: the file structure
2052 * @buffer: the user buffer
2053 * @lenp: the size of the user buffer
2054 * @ppos: file position
2055 *
2056 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2057 * values from/to the user buffer, treated as an ASCII string. The values
2058 * are treated as milliseconds, and converted to jiffies when they are stored.
2059 *
2060 * This routine will ensure the values are within the range specified by
2061 * table->extra1 (min) and table->extra2 (max).
2062 *
2063 * Returns 0 on success.
2064 */
2065 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2066 struct file *filp,
2067 void __user *buffer,
2068 size_t *lenp, loff_t *ppos)
2069 {
2070 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2071 lenp, ppos, HZ, 1000l);
2072 }
2073
2074
2075 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2076 int *valp,
2077 int write, void *data)
2078 {
2079 if (write) {
2080 if (*lvalp > LONG_MAX / HZ)
2081 return 1;
2082 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2083 } else {
2084 int val = *valp;
2085 unsigned long lval;
2086 if (val < 0) {
2087 *negp = -1;
2088 lval = (unsigned long)-val;
2089 } else {
2090 *negp = 0;
2091 lval = (unsigned long)val;
2092 }
2093 *lvalp = lval / HZ;
2094 }
2095 return 0;
2096 }
2097
2098 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2099 int *valp,
2100 int write, void *data)
2101 {
2102 if (write) {
2103 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2104 return 1;
2105 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2106 } else {
2107 int val = *valp;
2108 unsigned long lval;
2109 if (val < 0) {
2110 *negp = -1;
2111 lval = (unsigned long)-val;
2112 } else {
2113 *negp = 0;
2114 lval = (unsigned long)val;
2115 }
2116 *lvalp = jiffies_to_clock_t(lval);
2117 }
2118 return 0;
2119 }
2120
2121 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2122 int *valp,
2123 int write, void *data)
2124 {
2125 if (write) {
2126 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2127 } else {
2128 int val = *valp;
2129 unsigned long lval;
2130 if (val < 0) {
2131 *negp = -1;
2132 lval = (unsigned long)-val;
2133 } else {
2134 *negp = 0;
2135 lval = (unsigned long)val;
2136 }
2137 *lvalp = jiffies_to_msecs(lval);
2138 }
2139 return 0;
2140 }
2141
2142 /**
2143 * proc_dointvec_jiffies - read a vector of integers as seconds
2144 * @table: the sysctl table
2145 * @write: %TRUE if this is a write to the sysctl file
2146 * @filp: the file structure
2147 * @buffer: the user buffer
2148 * @lenp: the size of the user buffer
2149 * @ppos: file position
2150 *
2151 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2152 * values from/to the user buffer, treated as an ASCII string.
2153 * The values read are assumed to be in seconds, and are converted into
2154 * jiffies.
2155 *
2156 * Returns 0 on success.
2157 */
2158 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2159 void __user *buffer, size_t *lenp, loff_t *ppos)
2160 {
2161 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2162 do_proc_dointvec_jiffies_conv,NULL);
2163 }
2164
2165 /**
2166 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2167 * @table: the sysctl table
2168 * @write: %TRUE if this is a write to the sysctl file
2169 * @filp: the file structure
2170 * @buffer: the user buffer
2171 * @lenp: the size of the user buffer
2172 * @ppos: pointer to the file position
2173 *
2174 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2175 * values from/to the user buffer, treated as an ASCII string.
2176 * The values read are assumed to be in 1/USER_HZ seconds, and
2177 * are converted into jiffies.
2178 *
2179 * Returns 0 on success.
2180 */
2181 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2182 void __user *buffer, size_t *lenp, loff_t *ppos)
2183 {
2184 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2185 do_proc_dointvec_userhz_jiffies_conv,NULL);
2186 }
2187
2188 /**
2189 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2190 * @table: the sysctl table
2191 * @write: %TRUE if this is a write to the sysctl file
2192 * @filp: the file structure
2193 * @buffer: the user buffer
2194 * @lenp: the size of the user buffer
2195 * @ppos: file position
2196 * @ppos: the current position in the file
2197 *
2198 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2199 * values from/to the user buffer, treated as an ASCII string.
2200 * The values read are assumed to be in 1/1000 seconds, and
2201 * are converted into jiffies.
2202 *
2203 * Returns 0 on success.
2204 */
2205 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2206 void __user *buffer, size_t *lenp, loff_t *ppos)
2207 {
2208 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2209 do_proc_dointvec_ms_jiffies_conv, NULL);
2210 }
2211
2212 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
2213 void __user *buffer, size_t *lenp, loff_t *ppos)
2214 {
2215 struct pid *new_pid;
2216 pid_t tmp;
2217 int r;
2218
2219 tmp = pid_nr(cad_pid);
2220
2221 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2222 lenp, ppos, NULL, NULL);
2223 if (r || !write)
2224 return r;
2225
2226 new_pid = find_get_pid(tmp);
2227 if (!new_pid)
2228 return -ESRCH;
2229
2230 put_pid(xchg(&cad_pid, new_pid));
2231 return 0;
2232 }
2233
2234 #else /* CONFIG_PROC_FS */
2235
2236 int proc_dostring(ctl_table *table, int write, struct file *filp,
2237 void __user *buffer, size_t *lenp, loff_t *ppos)
2238 {
2239 return -ENOSYS;
2240 }
2241
2242 int proc_dointvec(ctl_table *table, int write, struct file *filp,
2243 void __user *buffer, size_t *lenp, loff_t *ppos)
2244 {
2245 return -ENOSYS;
2246 }
2247
2248 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2249 void __user *buffer, size_t *lenp, loff_t *ppos)
2250 {
2251 return -ENOSYS;
2252 }
2253
2254 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2255 void __user *buffer, size_t *lenp, loff_t *ppos)
2256 {
2257 return -ENOSYS;
2258 }
2259
2260 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2261 void __user *buffer, size_t *lenp, loff_t *ppos)
2262 {
2263 return -ENOSYS;
2264 }
2265
2266 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2267 void __user *buffer, size_t *lenp, loff_t *ppos)
2268 {
2269 return -ENOSYS;
2270 }
2271
2272 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2273 void __user *buffer, size_t *lenp, loff_t *ppos)
2274 {
2275 return -ENOSYS;
2276 }
2277
2278 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2279 void __user *buffer, size_t *lenp, loff_t *ppos)
2280 {
2281 return -ENOSYS;
2282 }
2283
2284 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2285 struct file *filp,
2286 void __user *buffer,
2287 size_t *lenp, loff_t *ppos)
2288 {
2289 return -ENOSYS;
2290 }
2291
2292
2293 #endif /* CONFIG_PROC_FS */
2294
2295
2296 #ifdef CONFIG_SYSCTL_SYSCALL
2297 /*
2298 * General sysctl support routines
2299 */
2300
2301 /* The generic string strategy routine: */
2302 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2303 void __user *oldval, size_t __user *oldlenp,
2304 void __user *newval, size_t newlen)
2305 {
2306 if (!table->data || !table->maxlen)
2307 return -ENOTDIR;
2308
2309 if (oldval && oldlenp) {
2310 size_t bufsize;
2311 if (get_user(bufsize, oldlenp))
2312 return -EFAULT;
2313 if (bufsize) {
2314 size_t len = strlen(table->data), copied;
2315
2316 /* This shouldn't trigger for a well-formed sysctl */
2317 if (len > table->maxlen)
2318 len = table->maxlen;
2319
2320 /* Copy up to a max of bufsize-1 bytes of the string */
2321 copied = (len >= bufsize) ? bufsize - 1 : len;
2322
2323 if (copy_to_user(oldval, table->data, copied) ||
2324 put_user(0, (char __user *)(oldval + copied)))
2325 return -EFAULT;
2326 if (put_user(len, oldlenp))
2327 return -EFAULT;
2328 }
2329 }
2330 if (newval && newlen) {
2331 size_t len = newlen;
2332 if (len > table->maxlen)
2333 len = table->maxlen;
2334 if(copy_from_user(table->data, newval, len))
2335 return -EFAULT;
2336 if (len == table->maxlen)
2337 len--;
2338 ((char *) table->data)[len] = 0;
2339 }
2340 return 1;
2341 }
2342
2343 /*
2344 * This function makes sure that all of the integers in the vector
2345 * are between the minimum and maximum values given in the arrays
2346 * table->extra1 and table->extra2, respectively.
2347 */
2348 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2349 void __user *oldval, size_t __user *oldlenp,
2350 void __user *newval, size_t newlen)
2351 {
2352
2353 if (newval && newlen) {
2354 int __user *vec = (int __user *) newval;
2355 int *min = (int *) table->extra1;
2356 int *max = (int *) table->extra2;
2357 size_t length;
2358 int i;
2359
2360 if (newlen % sizeof(int) != 0)
2361 return -EINVAL;
2362
2363 if (!table->extra1 && !table->extra2)
2364 return 0;
2365
2366 if (newlen > table->maxlen)
2367 newlen = table->maxlen;
2368 length = newlen / sizeof(int);
2369
2370 for (i = 0; i < length; i++) {
2371 int value;
2372 if (get_user(value, vec + i))
2373 return -EFAULT;
2374 if (min && value < min[i])
2375 return -EINVAL;
2376 if (max && value > max[i])
2377 return -EINVAL;
2378 }
2379 }
2380 return 0;
2381 }
2382
2383 /* Strategy function to convert jiffies to seconds */
2384 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2385 void __user *oldval, size_t __user *oldlenp,
2386 void __user *newval, size_t newlen)
2387 {
2388 if (oldval && oldlenp) {
2389 size_t olen;
2390
2391 if (get_user(olen, oldlenp))
2392 return -EFAULT;
2393 if (olen) {
2394 int val;
2395
2396 if (olen < sizeof(int))
2397 return -EINVAL;
2398
2399 val = *(int *)(table->data) / HZ;
2400 if (put_user(val, (int __user *)oldval))
2401 return -EFAULT;
2402 if (put_user(sizeof(int), oldlenp))
2403 return -EFAULT;
2404 }
2405 }
2406 if (newval && newlen) {
2407 int new;
2408 if (newlen != sizeof(int))
2409 return -EINVAL;
2410 if (get_user(new, (int __user *)newval))
2411 return -EFAULT;
2412 *(int *)(table->data) = new*HZ;
2413 }
2414 return 1;
2415 }
2416
2417 /* Strategy function to convert jiffies to seconds */
2418 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2419 void __user *oldval, size_t __user *oldlenp,
2420 void __user *newval, size_t newlen)
2421 {
2422 if (oldval && oldlenp) {
2423 size_t olen;
2424
2425 if (get_user(olen, oldlenp))
2426 return -EFAULT;
2427 if (olen) {
2428 int val;
2429
2430 if (olen < sizeof(int))
2431 return -EINVAL;
2432
2433 val = jiffies_to_msecs(*(int *)(table->data));
2434 if (put_user(val, (int __user *)oldval))
2435 return -EFAULT;
2436 if (put_user(sizeof(int), oldlenp))
2437 return -EFAULT;
2438 }
2439 }
2440 if (newval && newlen) {
2441 int new;
2442 if (newlen != sizeof(int))
2443 return -EINVAL;
2444 if (get_user(new, (int __user *)newval))
2445 return -EFAULT;
2446 *(int *)(table->data) = msecs_to_jiffies(new);
2447 }
2448 return 1;
2449 }
2450
2451
2452
2453 #else /* CONFIG_SYSCTL_SYSCALL */
2454
2455
2456 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2457 {
2458 static int msg_count;
2459 struct __sysctl_args tmp;
2460 int name[CTL_MAXNAME];
2461 int i;
2462
2463 /* Read in the sysctl name for better debug message logging */
2464 if (copy_from_user(&tmp, args, sizeof(tmp)))
2465 return -EFAULT;
2466 if (tmp.nlen <= 0 || tmp.nlen >= CTL_MAXNAME)
2467 return -ENOTDIR;
2468 for (i = 0; i < tmp.nlen; i++)
2469 if (get_user(name[i], tmp.name + i))
2470 return -EFAULT;
2471
2472 /* Ignore accesses to kernel.version */
2473 if ((tmp.nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2474 goto out;
2475
2476 if (msg_count < 5) {
2477 msg_count++;
2478 printk(KERN_INFO
2479 "warning: process `%s' used the removed sysctl "
2480 "system call with ", current->comm);
2481 for (i = 0; i < tmp.nlen; i++)
2482 printk("%d.", name[i]);
2483 printk("\n");
2484 }
2485 out:
2486 return -ENOSYS;
2487 }
2488
2489 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2490 void __user *oldval, size_t __user *oldlenp,
2491 void __user *newval, size_t newlen)
2492 {
2493 return -ENOSYS;
2494 }
2495
2496 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2497 void __user *oldval, size_t __user *oldlenp,
2498 void __user *newval, size_t newlen)
2499 {
2500 return -ENOSYS;
2501 }
2502
2503 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2504 void __user *oldval, size_t __user *oldlenp,
2505 void __user *newval, size_t newlen)
2506 {
2507 return -ENOSYS;
2508 }
2509
2510 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2511 void __user *oldval, size_t __user *oldlenp,
2512 void __user *newval, size_t newlen)
2513 {
2514 return -ENOSYS;
2515 }
2516
2517 #endif /* CONFIG_SYSCTL_SYSCALL */
2518
2519 /*
2520 * No sense putting this after each symbol definition, twice,
2521 * exception granted :-)
2522 */
2523 EXPORT_SYMBOL(proc_dointvec);
2524 EXPORT_SYMBOL(proc_dointvec_jiffies);
2525 EXPORT_SYMBOL(proc_dointvec_minmax);
2526 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2527 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2528 EXPORT_SYMBOL(proc_dostring);
2529 EXPORT_SYMBOL(proc_doulongvec_minmax);
2530 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2531 EXPORT_SYMBOL(register_sysctl_table);
2532 EXPORT_SYMBOL(sysctl_intvec);
2533 EXPORT_SYMBOL(sysctl_jiffies);
2534 EXPORT_SYMBOL(sysctl_ms_jiffies);
2535 EXPORT_SYMBOL(sysctl_string);
2536 EXPORT_SYMBOL(unregister_sysctl_table);
kernel-based virtual machine