projects / deliverable / linux.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
[PATCH] sysctl: create sys/fs/binfmt_misc as an ordinary sysctl entry
[deliverable/linux.git] / kernel / sysctl.c
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 __init sysctl_init(void)
1074 {
1075 #ifdef CONFIG_PROC_SYSCTL
1076 register_proc_table(root_table, proc_sys_root, &root_table_header);
1077 #endif
1078 }
1079
1080 #ifdef CONFIG_SYSCTL_SYSCALL
1081 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1082 void __user *newval, size_t newlen)
1083 {
1084 struct list_head *tmp;
1085 int error = -ENOTDIR;
1086
1087 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1088 return -ENOTDIR;
1089 if (oldval) {
1090 int old_len;
1091 if (!oldlenp || get_user(old_len, oldlenp))
1092 return -EFAULT;
1093 }
1094 spin_lock(&sysctl_lock);
1095 tmp = &root_table_header.ctl_entry;
1096 do {
1097 struct ctl_table_header *head =
1098 list_entry(tmp, struct ctl_table_header, ctl_entry);
1099
1100 if (!use_table(head))
1101 continue;
1102
1103 spin_unlock(&sysctl_lock);
1104
1105 error = parse_table(name, nlen, oldval, oldlenp,
1106 newval, newlen, head->ctl_table);
1107
1108 spin_lock(&sysctl_lock);
1109 unuse_table(head);
1110 if (error != -ENOTDIR)
1111 break;
1112 } while ((tmp = tmp->next) != &root_table_header.ctl_entry);
1113 spin_unlock(&sysctl_lock);
1114 return error;
1115 }
1116
1117 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1118 {
1119 struct __sysctl_args tmp;
1120 int error;
1121
1122 if (copy_from_user(&tmp, args, sizeof(tmp)))
1123 return -EFAULT;
1124
1125 lock_kernel();
1126 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1127 tmp.newval, tmp.newlen);
1128 unlock_kernel();
1129 return error;
1130 }
1131 #endif /* CONFIG_SYSCTL_SYSCALL */
1132
1133 /*
1134 * ctl_perm does NOT grant the superuser all rights automatically, because
1135 * some sysctl variables are readonly even to root.
1136 */
1137
1138 static int test_perm(int mode, int op)
1139 {
1140 if (!current->euid)
1141 mode >>= 6;
1142 else if (in_egroup_p(0))
1143 mode >>= 3;
1144 if ((mode & op & 0007) == op)
1145 return 0;
1146 return -EACCES;
1147 }
1148
1149 static inline int ctl_perm(ctl_table *table, int op)
1150 {
1151 int error;
1152 error = security_sysctl(table, op);
1153 if (error)
1154 return error;
1155 return test_perm(table->mode, op);
1156 }
1157
1158 #ifdef CONFIG_SYSCTL_SYSCALL
1159 static int parse_table(int __user *name, int nlen,
1160 void __user *oldval, size_t __user *oldlenp,
1161 void __user *newval, size_t newlen,
1162 ctl_table *table)
1163 {
1164 int n;
1165 repeat:
1166 if (!nlen)
1167 return -ENOTDIR;
1168 if (get_user(n, name))
1169 return -EFAULT;
1170 for ( ; table->ctl_name || table->procname; table++) {
1171 if (!table->ctl_name)
1172 continue;
1173 if (n == table->ctl_name || table->ctl_name == CTL_ANY) {
1174 int error;
1175 if (table->child) {
1176 if (ctl_perm(table, 001))
1177 return -EPERM;
1178 if (table->strategy) {
1179 error = table->strategy(
1180 table, name, nlen,
1181 oldval, oldlenp,
1182 newval, newlen);
1183 if (error)
1184 return error;
1185 }
1186 name++;
1187 nlen--;
1188 table = table->child;
1189 goto repeat;
1190 }
1191 error = do_sysctl_strategy(table, name, nlen,
1192 oldval, oldlenp,
1193 newval, newlen);
1194 return error;
1195 }
1196 }
1197 return -ENOTDIR;
1198 }
1199
1200 /* Perform the actual read/write of a sysctl table entry. */
1201 int do_sysctl_strategy (ctl_table *table,
1202 int __user *name, int nlen,
1203 void __user *oldval, size_t __user *oldlenp,
1204 void __user *newval, size_t newlen)
1205 {
1206 int op = 0, rc;
1207 size_t len;
1208
1209 if (oldval)
1210 op |= 004;
1211 if (newval)
1212 op |= 002;
1213 if (ctl_perm(table, op))
1214 return -EPERM;
1215
1216 if (table->strategy) {
1217 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1218 newval, newlen);
1219 if (rc < 0)
1220 return rc;
1221 if (rc > 0)
1222 return 0;
1223 }
1224
1225 /* If there is no strategy routine, or if the strategy returns
1226 * zero, proceed with automatic r/w */
1227 if (table->data && table->maxlen) {
1228 if (oldval && oldlenp) {
1229 if (get_user(len, oldlenp))
1230 return -EFAULT;
1231 if (len) {
1232 if (len > table->maxlen)
1233 len = table->maxlen;
1234 if(copy_to_user(oldval, table->data, len))
1235 return -EFAULT;
1236 if(put_user(len, oldlenp))
1237 return -EFAULT;
1238 }
1239 }
1240 if (newval && newlen) {
1241 len = newlen;
1242 if (len > table->maxlen)
1243 len = table->maxlen;
1244 if(copy_from_user(table->data, newval, len))
1245 return -EFAULT;
1246 }
1247 }
1248 return 0;
1249 }
1250 #endif /* CONFIG_SYSCTL_SYSCALL */
1251
1252 /**
1253 * register_sysctl_table - register a sysctl hierarchy
1254 * @table: the top-level table structure
1255 * @insert_at_head: whether the entry should be inserted in front or at the end
1256 *
1257 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1258 * array. An entry with a ctl_name of 0 terminates the table.
1259 *
1260 * The members of the &ctl_table structure are used as follows:
1261 *
1262 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1263 * must be unique within that level of sysctl
1264 *
1265 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1266 * enter a sysctl file
1267 *
1268 * data - a pointer to data for use by proc_handler
1269 *
1270 * maxlen - the maximum size in bytes of the data
1271 *
1272 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1273 *
1274 * child - a pointer to the child sysctl table if this entry is a directory, or
1275 * %NULL.
1276 *
1277 * proc_handler - the text handler routine (described below)
1278 *
1279 * strategy - the strategy routine (described below)
1280 *
1281 * de - for internal use by the sysctl routines
1282 *
1283 * extra1, extra2 - extra pointers usable by the proc handler routines
1284 *
1285 * Leaf nodes in the sysctl tree will be represented by a single file
1286 * under /proc; non-leaf nodes will be represented by directories.
1287 *
1288 * sysctl(2) can automatically manage read and write requests through
1289 * the sysctl table. The data and maxlen fields of the ctl_table
1290 * struct enable minimal validation of the values being written to be
1291 * performed, and the mode field allows minimal authentication.
1292 *
1293 * More sophisticated management can be enabled by the provision of a
1294 * strategy routine with the table entry. This will be called before
1295 * any automatic read or write of the data is performed.
1296 *
1297 * The strategy routine may return
1298 *
1299 * < 0 - Error occurred (error is passed to user process)
1300 *
1301 * 0 - OK - proceed with automatic read or write.
1302 *
1303 * > 0 - OK - read or write has been done by the strategy routine, so
1304 * return immediately.
1305 *
1306 * There must be a proc_handler routine for any terminal nodes
1307 * mirrored under /proc/sys (non-terminals are handled by a built-in
1308 * directory handler). Several default handlers are available to
1309 * cover common cases -
1310 *
1311 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1312 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1313 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1314 *
1315 * It is the handler's job to read the input buffer from user memory
1316 * and process it. The handler should return 0 on success.
1317 *
1318 * This routine returns %NULL on a failure to register, and a pointer
1319 * to the table header on success.
1320 */
1321 struct ctl_table_header *register_sysctl_table(ctl_table * table,
1322 int insert_at_head)
1323 {
1324 struct ctl_table_header *tmp;
1325 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1326 if (!tmp)
1327 return NULL;
1328 tmp->ctl_table = table;
1329 INIT_LIST_HEAD(&tmp->ctl_entry);
1330 tmp->used = 0;
1331 tmp->unregistering = NULL;
1332 spin_lock(&sysctl_lock);
1333 if (insert_at_head)
1334 list_add(&tmp->ctl_entry, &root_table_header.ctl_entry);
1335 else
1336 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
1337 spin_unlock(&sysctl_lock);
1338 #ifdef CONFIG_PROC_SYSCTL
1339 register_proc_table(table, proc_sys_root, tmp);
1340 #endif
1341 return tmp;
1342 }
1343
1344 /**
1345 * unregister_sysctl_table - unregister a sysctl table hierarchy
1346 * @header: the header returned from register_sysctl_table
1347 *
1348 * Unregisters the sysctl table and all children. proc entries may not
1349 * actually be removed until they are no longer used by anyone.
1350 */
1351 void unregister_sysctl_table(struct ctl_table_header * header)
1352 {
1353 might_sleep();
1354 spin_lock(&sysctl_lock);
1355 start_unregistering(header);
1356 #ifdef CONFIG_PROC_SYSCTL
1357 unregister_proc_table(header->ctl_table, proc_sys_root);
1358 #endif
1359 spin_unlock(&sysctl_lock);
1360 kfree(header);
1361 }
1362
1363 #else /* !CONFIG_SYSCTL */
1364 struct ctl_table_header * register_sysctl_table(ctl_table * table,
1365 int insert_at_head)
1366 {
1367 return NULL;
1368 }
1369
1370 void unregister_sysctl_table(struct ctl_table_header * table)
1371 {
1372 }
1373
1374 #endif /* CONFIG_SYSCTL */
1375
1376 /*
1377 * /proc/sys support
1378 */
1379
1380 #ifdef CONFIG_PROC_SYSCTL
1381
1382 /* Scan the sysctl entries in table and add them all into /proc */
1383 static void register_proc_table(ctl_table * table, struct proc_dir_entry *root, void *set)
1384 {
1385 struct proc_dir_entry *de;
1386 int len;
1387 mode_t mode;
1388
1389 for (; table->ctl_name || table->procname; table++) {
1390 /* Can't do anything without a proc name. */
1391 if (!table->procname)
1392 continue;
1393 /* Maybe we can't do anything with it... */
1394 if (!table->proc_handler && !table->child) {
1395 printk(KERN_WARNING "SYSCTL: Can't register %s\n",
1396 table->procname);
1397 continue;
1398 }
1399
1400 len = strlen(table->procname);
1401 mode = table->mode;
1402
1403 de = NULL;
1404 if (table->proc_handler)
1405 mode |= S_IFREG;
1406 else {
1407 mode |= S_IFDIR;
1408 for (de = root->subdir; de; de = de->next) {
1409 if (proc_match(len, table->procname, de))
1410 break;
1411 }
1412 /* If the subdir exists already, de is non-NULL */
1413 }
1414
1415 if (!de) {
1416 de = create_proc_entry(table->procname, mode, root);
1417 if (!de)
1418 continue;
1419 de->set = set;
1420 de->data = (void *) table;
1421 if (table->proc_handler)
1422 de->proc_fops = &proc_sys_file_operations;
1423 }
1424 table->de = de;
1425 if (de->mode & S_IFDIR)
1426 register_proc_table(table->child, de, set);
1427 }
1428 }
1429
1430 /*
1431 * Unregister a /proc sysctl table and any subdirectories.
1432 */
1433 static void unregister_proc_table(ctl_table * table, struct proc_dir_entry *root)
1434 {
1435 struct proc_dir_entry *de;
1436 for (; table->ctl_name || table->procname; table++) {
1437 if (!(de = table->de))
1438 continue;
1439 if (de->mode & S_IFDIR) {
1440 if (!table->child) {
1441 printk (KERN_ALERT "Help - malformed sysctl tree on free\n");
1442 continue;
1443 }
1444 unregister_proc_table(table->child, de);
1445
1446 /* Don't unregister directories which still have entries.. */
1447 if (de->subdir)
1448 continue;
1449 }
1450
1451 /*
1452 * In any case, mark the entry as goner; we'll keep it
1453 * around if it's busy, but we'll know to do nothing with
1454 * its fields. We are under sysctl_lock here.
1455 */
1456 de->data = NULL;
1457
1458 /* Don't unregister proc entries that are still being used.. */
1459 if (atomic_read(&de->count))
1460 continue;
1461
1462 table->de = NULL;
1463 remove_proc_entry(table->procname, root);
1464 }
1465 }
1466
1467 static ssize_t do_rw_proc(int write, struct file * file, char __user * buf,
1468 size_t count, loff_t *ppos)
1469 {
1470 int op;
1471 struct proc_dir_entry *de = PDE(file->f_path.dentry->d_inode);
1472 struct ctl_table *table;
1473 size_t res;
1474 ssize_t error = -ENOTDIR;
1475
1476 spin_lock(&sysctl_lock);
1477 if (de && de->data && use_table(de->set)) {
1478 /*
1479 * at that point we know that sysctl was not unregistered
1480 * and won't be until we finish
1481 */
1482 spin_unlock(&sysctl_lock);
1483 table = (struct ctl_table *) de->data;
1484 if (!table || !table->proc_handler)
1485 goto out;
1486 error = -EPERM;
1487 op = (write ? 002 : 004);
1488 if (ctl_perm(table, op))
1489 goto out;
1490
1491 /* careful: calling conventions are nasty here */
1492 res = count;
1493 error = (*table->proc_handler)(table, write, file,
1494 buf, &res, ppos);
1495 if (!error)
1496 error = res;
1497 out:
1498 spin_lock(&sysctl_lock);
1499 unuse_table(de->set);
1500 }
1501 spin_unlock(&sysctl_lock);
1502 return error;
1503 }
1504
1505 static int proc_opensys(struct inode *inode, struct file *file)
1506 {
1507 if (file->f_mode & FMODE_WRITE) {
1508 /*
1509 * sysctl entries that are not writable,
1510 * are _NOT_ writable, capabilities or not.
1511 */
1512 if (!(inode->i_mode & S_IWUSR))
1513 return -EPERM;
1514 }
1515
1516 return 0;
1517 }
1518
1519 static ssize_t proc_readsys(struct file * file, char __user * buf,
1520 size_t count, loff_t *ppos)
1521 {
1522 return do_rw_proc(0, file, buf, count, ppos);
1523 }
1524
1525 static ssize_t proc_writesys(struct file * file, const char __user * buf,
1526 size_t count, loff_t *ppos)
1527 {
1528 return do_rw_proc(1, file, (char __user *) buf, count, ppos);
1529 }
1530
1531 static int _proc_do_string(void* data, int maxlen, int write,
1532 struct file *filp, void __user *buffer,
1533 size_t *lenp, loff_t *ppos)
1534 {
1535 size_t len;
1536 char __user *p;
1537 char c;
1538
1539 if (!data || !maxlen || !*lenp) {
1540 *lenp = 0;
1541 return 0;
1542 }
1543
1544 if (write) {
1545 len = 0;
1546 p = buffer;
1547 while (len < *lenp) {
1548 if (get_user(c, p++))
1549 return -EFAULT;
1550 if (c == 0 || c == '\n')
1551 break;
1552 len++;
1553 }
1554 if (len >= maxlen)
1555 len = maxlen-1;
1556 if(copy_from_user(data, buffer, len))
1557 return -EFAULT;
1558 ((char *) data)[len] = 0;
1559 *ppos += *lenp;
1560 } else {
1561 len = strlen(data);
1562 if (len > maxlen)
1563 len = maxlen;
1564
1565 if (*ppos > len) {
1566 *lenp = 0;
1567 return 0;
1568 }
1569
1570 data += *ppos;
1571 len -= *ppos;
1572
1573 if (len > *lenp)
1574 len = *lenp;
1575 if (len)
1576 if(copy_to_user(buffer, data, len))
1577 return -EFAULT;
1578 if (len < *lenp) {
1579 if(put_user('\n', ((char __user *) buffer) + len))
1580 return -EFAULT;
1581 len++;
1582 }
1583 *lenp = len;
1584 *ppos += len;
1585 }
1586 return 0;
1587 }
1588
1589 /**
1590 * proc_dostring - read a string sysctl
1591 * @table: the sysctl table
1592 * @write: %TRUE if this is a write to the sysctl file
1593 * @filp: the file structure
1594 * @buffer: the user buffer
1595 * @lenp: the size of the user buffer
1596 * @ppos: file position
1597 *
1598 * Reads/writes a string from/to the user buffer. If the kernel
1599 * buffer provided is not large enough to hold the string, the
1600 * string is truncated. The copied string is %NULL-terminated.
1601 * If the string is being read by the user process, it is copied
1602 * and a newline '\n' is added. It is truncated if the buffer is
1603 * not large enough.
1604 *
1605 * Returns 0 on success.
1606 */
1607 int proc_dostring(ctl_table *table, int write, struct file *filp,
1608 void __user *buffer, size_t *lenp, loff_t *ppos)
1609 {
1610 return _proc_do_string(table->data, table->maxlen, write, filp,
1611 buffer, lenp, ppos);
1612 }
1613
1614
1615 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1616 int *valp,
1617 int write, void *data)
1618 {
1619 if (write) {
1620 *valp = *negp ? -*lvalp : *lvalp;
1621 } else {
1622 int val = *valp;
1623 if (val < 0) {
1624 *negp = -1;
1625 *lvalp = (unsigned long)-val;
1626 } else {
1627 *negp = 0;
1628 *lvalp = (unsigned long)val;
1629 }
1630 }
1631 return 0;
1632 }
1633
1634 static int __do_proc_dointvec(void *tbl_data, ctl_table *table,
1635 int write, struct file *filp, void __user *buffer,
1636 size_t *lenp, loff_t *ppos,
1637 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1638 int write, void *data),
1639 void *data)
1640 {
1641 #define TMPBUFLEN 21
1642 int *i, vleft, first=1, neg, val;
1643 unsigned long lval;
1644 size_t left, len;
1645
1646 char buf[TMPBUFLEN], *p;
1647 char __user *s = buffer;
1648
1649 if (!tbl_data || !table->maxlen || !*lenp ||
1650 (*ppos && !write)) {
1651 *lenp = 0;
1652 return 0;
1653 }
1654
1655 i = (int *) tbl_data;
1656 vleft = table->maxlen / sizeof(*i);
1657 left = *lenp;
1658
1659 if (!conv)
1660 conv = do_proc_dointvec_conv;
1661
1662 for (; left && vleft--; i++, first=0) {
1663 if (write) {
1664 while (left) {
1665 char c;
1666 if (get_user(c, s))
1667 return -EFAULT;
1668 if (!isspace(c))
1669 break;
1670 left--;
1671 s++;
1672 }
1673 if (!left)
1674 break;
1675 neg = 0;
1676 len = left;
1677 if (len > sizeof(buf) - 1)
1678 len = sizeof(buf) - 1;
1679 if (copy_from_user(buf, s, len))
1680 return -EFAULT;
1681 buf[len] = 0;
1682 p = buf;
1683 if (*p == '-' && left > 1) {
1684 neg = 1;
1685 p++;
1686 }
1687 if (*p < '0' || *p > '9')
1688 break;
1689
1690 lval = simple_strtoul(p, &p, 0);
1691
1692 len = p-buf;
1693 if ((len < left) && *p && !isspace(*p))
1694 break;
1695 if (neg)
1696 val = -val;
1697 s += len;
1698 left -= len;
1699
1700 if (conv(&neg, &lval, i, 1, data))
1701 break;
1702 } else {
1703 p = buf;
1704 if (!first)
1705 *p++ = '\t';
1706
1707 if (conv(&neg, &lval, i, 0, data))
1708 break;
1709
1710 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1711 len = strlen(buf);
1712 if (len > left)
1713 len = left;
1714 if(copy_to_user(s, buf, len))
1715 return -EFAULT;
1716 left -= len;
1717 s += len;
1718 }
1719 }
1720
1721 if (!write && !first && left) {
1722 if(put_user('\n', s))
1723 return -EFAULT;
1724 left--, s++;
1725 }
1726 if (write) {
1727 while (left) {
1728 char c;
1729 if (get_user(c, s++))
1730 return -EFAULT;
1731 if (!isspace(c))
1732 break;
1733 left--;
1734 }
1735 }
1736 if (write && first)
1737 return -EINVAL;
1738 *lenp -= left;
1739 *ppos += *lenp;
1740 return 0;
1741 #undef TMPBUFLEN
1742 }
1743
1744 static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1745 void __user *buffer, size_t *lenp, loff_t *ppos,
1746 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1747 int write, void *data),
1748 void *data)
1749 {
1750 return __do_proc_dointvec(table->data, table, write, filp,
1751 buffer, lenp, ppos, conv, data);
1752 }
1753
1754 /**
1755 * proc_dointvec - read a vector of integers
1756 * @table: the sysctl table
1757 * @write: %TRUE if this is a write to the sysctl file
1758 * @filp: the file structure
1759 * @buffer: the user buffer
1760 * @lenp: the size of the user buffer
1761 * @ppos: file position
1762 *
1763 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1764 * values from/to the user buffer, treated as an ASCII string.
1765 *
1766 * Returns 0 on success.
1767 */
1768 int proc_dointvec(ctl_table *table, int write, struct file *filp,
1769 void __user *buffer, size_t *lenp, loff_t *ppos)
1770 {
1771 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1772 NULL,NULL);
1773 }
1774
1775 #define OP_SET 0
1776 #define OP_AND 1
1777 #define OP_OR 2
1778
1779 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1780 int *valp,
1781 int write, void *data)
1782 {
1783 int op = *(int *)data;
1784 if (write) {
1785 int val = *negp ? -*lvalp : *lvalp;
1786 switch(op) {
1787 case OP_SET: *valp = val; break;
1788 case OP_AND: *valp &= val; break;
1789 case OP_OR: *valp |= val; break;
1790 }
1791 } else {
1792 int val = *valp;
1793 if (val < 0) {
1794 *negp = -1;
1795 *lvalp = (unsigned long)-val;
1796 } else {
1797 *negp = 0;
1798 *lvalp = (unsigned long)val;
1799 }
1800 }
1801 return 0;
1802 }
1803
1804 /*
1805 * init may raise the set.
1806 */
1807
1808 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1809 void __user *buffer, size_t *lenp, loff_t *ppos)
1810 {
1811 int op;
1812
1813 if (write && !capable(CAP_SYS_MODULE)) {
1814 return -EPERM;
1815 }
1816
1817 op = is_init(current) ? OP_SET : OP_AND;
1818 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1819 do_proc_dointvec_bset_conv,&op);
1820 }
1821
1822 /*
1823 * Taint values can only be increased
1824 */
1825 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
1826 void __user *buffer, size_t *lenp, loff_t *ppos)
1827 {
1828 int op;
1829
1830 if (!capable(CAP_SYS_ADMIN))
1831 return -EPERM;
1832
1833 op = OP_OR;
1834 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1835 do_proc_dointvec_bset_conv,&op);
1836 }
1837
1838 struct do_proc_dointvec_minmax_conv_param {
1839 int *min;
1840 int *max;
1841 };
1842
1843 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1844 int *valp,
1845 int write, void *data)
1846 {
1847 struct do_proc_dointvec_minmax_conv_param *param = data;
1848 if (write) {
1849 int val = *negp ? -*lvalp : *lvalp;
1850 if ((param->min && *param->min > val) ||
1851 (param->max && *param->max < val))
1852 return -EINVAL;
1853 *valp = val;
1854 } else {
1855 int val = *valp;
1856 if (val < 0) {
1857 *negp = -1;
1858 *lvalp = (unsigned long)-val;
1859 } else {
1860 *negp = 0;
1861 *lvalp = (unsigned long)val;
1862 }
1863 }
1864 return 0;
1865 }
1866
1867 /**
1868 * proc_dointvec_minmax - read a vector of integers with min/max values
1869 * @table: the sysctl table
1870 * @write: %TRUE if this is a write to the sysctl file
1871 * @filp: the file structure
1872 * @buffer: the user buffer
1873 * @lenp: the size of the user buffer
1874 * @ppos: file position
1875 *
1876 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1877 * values from/to the user buffer, treated as an ASCII string.
1878 *
1879 * This routine will ensure the values are within the range specified by
1880 * table->extra1 (min) and table->extra2 (max).
1881 *
1882 * Returns 0 on success.
1883 */
1884 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1885 void __user *buffer, size_t *lenp, loff_t *ppos)
1886 {
1887 struct do_proc_dointvec_minmax_conv_param param = {
1888 .min = (int *) table->extra1,
1889 .max = (int *) table->extra2,
1890 };
1891 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1892 do_proc_dointvec_minmax_conv, &param);
1893 }
1894
1895 static int __do_proc_doulongvec_minmax(void *data, ctl_table *table, int write,
1896 struct file *filp,
1897 void __user *buffer,
1898 size_t *lenp, loff_t *ppos,
1899 unsigned long convmul,
1900 unsigned long convdiv)
1901 {
1902 #define TMPBUFLEN 21
1903 unsigned long *i, *min, *max, val;
1904 int vleft, first=1, neg;
1905 size_t len, left;
1906 char buf[TMPBUFLEN], *p;
1907 char __user *s = buffer;
1908
1909 if (!data || !table->maxlen || !*lenp ||
1910 (*ppos && !write)) {
1911 *lenp = 0;
1912 return 0;
1913 }
1914
1915 i = (unsigned long *) data;
1916 min = (unsigned long *) table->extra1;
1917 max = (unsigned long *) table->extra2;
1918 vleft = table->maxlen / sizeof(unsigned long);
1919 left = *lenp;
1920
1921 for (; left && vleft--; i++, min++, max++, first=0) {
1922 if (write) {
1923 while (left) {
1924 char c;
1925 if (get_user(c, s))
1926 return -EFAULT;
1927 if (!isspace(c))
1928 break;
1929 left--;
1930 s++;
1931 }
1932 if (!left)
1933 break;
1934 neg = 0;
1935 len = left;
1936 if (len > TMPBUFLEN-1)
1937 len = TMPBUFLEN-1;
1938 if (copy_from_user(buf, s, len))
1939 return -EFAULT;
1940 buf[len] = 0;
1941 p = buf;
1942 if (*p == '-' && left > 1) {
1943 neg = 1;
1944 p++;
1945 }
1946 if (*p < '0' || *p > '9')
1947 break;
1948 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
1949 len = p-buf;
1950 if ((len < left) && *p && !isspace(*p))
1951 break;
1952 if (neg)
1953 val = -val;
1954 s += len;
1955 left -= len;
1956
1957 if(neg)
1958 continue;
1959 if ((min && val < *min) || (max && val > *max))
1960 continue;
1961 *i = val;
1962 } else {
1963 p = buf;
1964 if (!first)
1965 *p++ = '\t';
1966 sprintf(p, "%lu", convdiv * (*i) / convmul);
1967 len = strlen(buf);
1968 if (len > left)
1969 len = left;
1970 if(copy_to_user(s, buf, len))
1971 return -EFAULT;
1972 left -= len;
1973 s += len;
1974 }
1975 }
1976
1977 if (!write && !first && left) {
1978 if(put_user('\n', s))
1979 return -EFAULT;
1980 left--, s++;
1981 }
1982 if (write) {
1983 while (left) {
1984 char c;
1985 if (get_user(c, s++))
1986 return -EFAULT;
1987 if (!isspace(c))
1988 break;
1989 left--;
1990 }
1991 }
1992 if (write && first)
1993 return -EINVAL;
1994 *lenp -= left;
1995 *ppos += *lenp;
1996 return 0;
1997 #undef TMPBUFLEN
1998 }
1999
2000 static int do_proc_doulongvec_minmax(ctl_table *table, int write,
2001 struct file *filp,
2002 void __user *buffer,
2003 size_t *lenp, loff_t *ppos,
2004 unsigned long convmul,
2005 unsigned long convdiv)
2006 {
2007 return __do_proc_doulongvec_minmax(table->data, table, write,
2008 filp, buffer, lenp, ppos, convmul, convdiv);
2009 }
2010
2011 /**
2012 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2013 * @table: the sysctl table
2014 * @write: %TRUE if this is a write to the sysctl file
2015 * @filp: the file structure
2016 * @buffer: the user buffer
2017 * @lenp: the size of the user buffer
2018 * @ppos: file position
2019 *
2020 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2021 * values from/to the user buffer, treated as an ASCII string.
2022 *
2023 * This routine will ensure the values are within the range specified by
2024 * table->extra1 (min) and table->extra2 (max).
2025 *
2026 * Returns 0 on success.
2027 */
2028 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2029 void __user *buffer, size_t *lenp, loff_t *ppos)
2030 {
2031 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2032 }
2033
2034 /**
2035 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2036 * @table: the sysctl table
2037 * @write: %TRUE if this is a write to the sysctl file
2038 * @filp: the file structure
2039 * @buffer: the user buffer
2040 * @lenp: the size of the user buffer
2041 * @ppos: file position
2042 *
2043 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2044 * values from/to the user buffer, treated as an ASCII string. The values
2045 * are treated as milliseconds, and converted to jiffies when they are stored.
2046 *
2047 * This routine will ensure the values are within the range specified by
2048 * table->extra1 (min) and table->extra2 (max).
2049 *
2050 * Returns 0 on success.
2051 */
2052 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2053 struct file *filp,
2054 void __user *buffer,
2055 size_t *lenp, loff_t *ppos)
2056 {
2057 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2058 lenp, ppos, HZ, 1000l);
2059 }
2060
2061
2062 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2063 int *valp,
2064 int write, void *data)
2065 {
2066 if (write) {
2067 if (*lvalp > LONG_MAX / HZ)
2068 return 1;
2069 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2070 } else {
2071 int val = *valp;
2072 unsigned long lval;
2073 if (val < 0) {
2074 *negp = -1;
2075 lval = (unsigned long)-val;
2076 } else {
2077 *negp = 0;
2078 lval = (unsigned long)val;
2079 }
2080 *lvalp = lval / HZ;
2081 }
2082 return 0;
2083 }
2084
2085 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2086 int *valp,
2087 int write, void *data)
2088 {
2089 if (write) {
2090 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2091 return 1;
2092 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2093 } else {
2094 int val = *valp;
2095 unsigned long lval;
2096 if (val < 0) {
2097 *negp = -1;
2098 lval = (unsigned long)-val;
2099 } else {
2100 *negp = 0;
2101 lval = (unsigned long)val;
2102 }
2103 *lvalp = jiffies_to_clock_t(lval);
2104 }
2105 return 0;
2106 }
2107
2108 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2109 int *valp,
2110 int write, void *data)
2111 {
2112 if (write) {
2113 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2114 } else {
2115 int val = *valp;
2116 unsigned long lval;
2117 if (val < 0) {
2118 *negp = -1;
2119 lval = (unsigned long)-val;
2120 } else {
2121 *negp = 0;
2122 lval = (unsigned long)val;
2123 }
2124 *lvalp = jiffies_to_msecs(lval);
2125 }
2126 return 0;
2127 }
2128
2129 /**
2130 * proc_dointvec_jiffies - read a vector of integers as seconds
2131 * @table: the sysctl table
2132 * @write: %TRUE if this is a write to the sysctl file
2133 * @filp: the file structure
2134 * @buffer: the user buffer
2135 * @lenp: the size of the user buffer
2136 * @ppos: file position
2137 *
2138 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2139 * values from/to the user buffer, treated as an ASCII string.
2140 * The values read are assumed to be in seconds, and are converted into
2141 * jiffies.
2142 *
2143 * Returns 0 on success.
2144 */
2145 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2146 void __user *buffer, size_t *lenp, loff_t *ppos)
2147 {
2148 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2149 do_proc_dointvec_jiffies_conv,NULL);
2150 }
2151
2152 /**
2153 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2154 * @table: the sysctl table
2155 * @write: %TRUE if this is a write to the sysctl file
2156 * @filp: the file structure
2157 * @buffer: the user buffer
2158 * @lenp: the size of the user buffer
2159 * @ppos: pointer to the file position
2160 *
2161 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2162 * values from/to the user buffer, treated as an ASCII string.
2163 * The values read are assumed to be in 1/USER_HZ seconds, and
2164 * are converted into jiffies.
2165 *
2166 * Returns 0 on success.
2167 */
2168 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2169 void __user *buffer, size_t *lenp, loff_t *ppos)
2170 {
2171 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2172 do_proc_dointvec_userhz_jiffies_conv,NULL);
2173 }
2174
2175 /**
2176 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2177 * @table: the sysctl table
2178 * @write: %TRUE if this is a write to the sysctl file
2179 * @filp: the file structure
2180 * @buffer: the user buffer
2181 * @lenp: the size of the user buffer
2182 * @ppos: file position
2183 * @ppos: the current position in the file
2184 *
2185 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2186 * values from/to the user buffer, treated as an ASCII string.
2187 * The values read are assumed to be in 1/1000 seconds, and
2188 * are converted into jiffies.
2189 *
2190 * Returns 0 on success.
2191 */
2192 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2193 void __user *buffer, size_t *lenp, loff_t *ppos)
2194 {
2195 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2196 do_proc_dointvec_ms_jiffies_conv, NULL);
2197 }
2198
2199 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
2200 void __user *buffer, size_t *lenp, loff_t *ppos)
2201 {
2202 struct pid *new_pid;
2203 pid_t tmp;
2204 int r;
2205
2206 tmp = pid_nr(cad_pid);
2207
2208 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2209 lenp, ppos, NULL, NULL);
2210 if (r || !write)
2211 return r;
2212
2213 new_pid = find_get_pid(tmp);
2214 if (!new_pid)
2215 return -ESRCH;
2216
2217 put_pid(xchg(&cad_pid, new_pid));
2218 return 0;
2219 }
2220
2221 #else /* CONFIG_PROC_FS */
2222
2223 int proc_dostring(ctl_table *table, int write, struct file *filp,
2224 void __user *buffer, size_t *lenp, loff_t *ppos)
2225 {
2226 return -ENOSYS;
2227 }
2228
2229 int proc_dointvec(ctl_table *table, int write, struct file *filp,
2230 void __user *buffer, size_t *lenp, loff_t *ppos)
2231 {
2232 return -ENOSYS;
2233 }
2234
2235 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2236 void __user *buffer, size_t *lenp, loff_t *ppos)
2237 {
2238 return -ENOSYS;
2239 }
2240
2241 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2242 void __user *buffer, size_t *lenp, loff_t *ppos)
2243 {
2244 return -ENOSYS;
2245 }
2246
2247 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2248 void __user *buffer, size_t *lenp, loff_t *ppos)
2249 {
2250 return -ENOSYS;
2251 }
2252
2253 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2254 void __user *buffer, size_t *lenp, loff_t *ppos)
2255 {
2256 return -ENOSYS;
2257 }
2258
2259 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2260 void __user *buffer, size_t *lenp, loff_t *ppos)
2261 {
2262 return -ENOSYS;
2263 }
2264
2265 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2266 void __user *buffer, size_t *lenp, loff_t *ppos)
2267 {
2268 return -ENOSYS;
2269 }
2270
2271 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2272 struct file *filp,
2273 void __user *buffer,
2274 size_t *lenp, loff_t *ppos)
2275 {
2276 return -ENOSYS;
2277 }
2278
2279
2280 #endif /* CONFIG_PROC_FS */
2281
2282
2283 #ifdef CONFIG_SYSCTL_SYSCALL
2284 /*
2285 * General sysctl support routines
2286 */
2287
2288 /* The generic string strategy routine: */
2289 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2290 void __user *oldval, size_t __user *oldlenp,
2291 void __user *newval, size_t newlen)
2292 {
2293 if (!table->data || !table->maxlen)
2294 return -ENOTDIR;
2295
2296 if (oldval && oldlenp) {
2297 size_t bufsize;
2298 if (get_user(bufsize, oldlenp))
2299 return -EFAULT;
2300 if (bufsize) {
2301 size_t len = strlen(table->data), copied;
2302
2303 /* This shouldn't trigger for a well-formed sysctl */
2304 if (len > table->maxlen)
2305 len = table->maxlen;
2306
2307 /* Copy up to a max of bufsize-1 bytes of the string */
2308 copied = (len >= bufsize) ? bufsize - 1 : len;
2309
2310 if (copy_to_user(oldval, table->data, copied) ||
2311 put_user(0, (char __user *)(oldval + copied)))
2312 return -EFAULT;
2313 if (put_user(len, oldlenp))
2314 return -EFAULT;
2315 }
2316 }
2317 if (newval && newlen) {
2318 size_t len = newlen;
2319 if (len > table->maxlen)
2320 len = table->maxlen;
2321 if(copy_from_user(table->data, newval, len))
2322 return -EFAULT;
2323 if (len == table->maxlen)
2324 len--;
2325 ((char *) table->data)[len] = 0;
2326 }
2327 return 1;
2328 }
2329
2330 /*
2331 * This function makes sure that all of the integers in the vector
2332 * are between the minimum and maximum values given in the arrays
2333 * table->extra1 and table->extra2, respectively.
2334 */
2335 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2336 void __user *oldval, size_t __user *oldlenp,
2337 void __user *newval, size_t newlen)
2338 {
2339
2340 if (newval && newlen) {
2341 int __user *vec = (int __user *) newval;
2342 int *min = (int *) table->extra1;
2343 int *max = (int *) table->extra2;
2344 size_t length;
2345 int i;
2346
2347 if (newlen % sizeof(int) != 0)
2348 return -EINVAL;
2349
2350 if (!table->extra1 && !table->extra2)
2351 return 0;
2352
2353 if (newlen > table->maxlen)
2354 newlen = table->maxlen;
2355 length = newlen / sizeof(int);
2356
2357 for (i = 0; i < length; i++) {
2358 int value;
2359 if (get_user(value, vec + i))
2360 return -EFAULT;
2361 if (min && value < min[i])
2362 return -EINVAL;
2363 if (max && value > max[i])
2364 return -EINVAL;
2365 }
2366 }
2367 return 0;
2368 }
2369
2370 /* Strategy function to convert jiffies to seconds */
2371 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2372 void __user *oldval, size_t __user *oldlenp,
2373 void __user *newval, size_t newlen)
2374 {
2375 if (oldval && oldlenp) {
2376 size_t olen;
2377
2378 if (get_user(olen, oldlenp))
2379 return -EFAULT;
2380 if (olen) {
2381 int val;
2382
2383 if (olen < sizeof(int))
2384 return -EINVAL;
2385
2386 val = *(int *)(table->data) / HZ;
2387 if (put_user(val, (int __user *)oldval))
2388 return -EFAULT;
2389 if (put_user(sizeof(int), oldlenp))
2390 return -EFAULT;
2391 }
2392 }
2393 if (newval && newlen) {
2394 int new;
2395 if (newlen != sizeof(int))
2396 return -EINVAL;
2397 if (get_user(new, (int __user *)newval))
2398 return -EFAULT;
2399 *(int *)(table->data) = new*HZ;
2400 }
2401 return 1;
2402 }
2403
2404 /* Strategy function to convert jiffies to seconds */
2405 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2406 void __user *oldval, size_t __user *oldlenp,
2407 void __user *newval, size_t newlen)
2408 {
2409 if (oldval && oldlenp) {
2410 size_t olen;
2411
2412 if (get_user(olen, oldlenp))
2413 return -EFAULT;
2414 if (olen) {
2415 int val;
2416
2417 if (olen < sizeof(int))
2418 return -EINVAL;
2419
2420 val = jiffies_to_msecs(*(int *)(table->data));
2421 if (put_user(val, (int __user *)oldval))
2422 return -EFAULT;
2423 if (put_user(sizeof(int), oldlenp))
2424 return -EFAULT;
2425 }
2426 }
2427 if (newval && newlen) {
2428 int new;
2429 if (newlen != sizeof(int))
2430 return -EINVAL;
2431 if (get_user(new, (int __user *)newval))
2432 return -EFAULT;
2433 *(int *)(table->data) = msecs_to_jiffies(new);
2434 }
2435 return 1;
2436 }
2437
2438
2439
2440 #else /* CONFIG_SYSCTL_SYSCALL */
2441
2442
2443 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2444 {
2445 static int msg_count;
2446 struct __sysctl_args tmp;
2447 int name[CTL_MAXNAME];
2448 int i;
2449
2450 /* Read in the sysctl name for better debug message logging */
2451 if (copy_from_user(&tmp, args, sizeof(tmp)))
2452 return -EFAULT;
2453 if (tmp.nlen <= 0 || tmp.nlen >= CTL_MAXNAME)
2454 return -ENOTDIR;
2455 for (i = 0; i < tmp.nlen; i++)
2456 if (get_user(name[i], tmp.name + i))
2457 return -EFAULT;
2458
2459 /* Ignore accesses to kernel.version */
2460 if ((tmp.nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2461 goto out;
2462
2463 if (msg_count < 5) {
2464 msg_count++;
2465 printk(KERN_INFO
2466 "warning: process `%s' used the removed sysctl "
2467 "system call with ", current->comm);
2468 for (i = 0; i < tmp.nlen; i++)
2469 printk("%d.", name[i]);
2470 printk("\n");
2471 }
2472 out:
2473 return -ENOSYS;
2474 }
2475
2476 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2477 void __user *oldval, size_t __user *oldlenp,
2478 void __user *newval, size_t newlen)
2479 {
2480 return -ENOSYS;
2481 }
2482
2483 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2484 void __user *oldval, size_t __user *oldlenp,
2485 void __user *newval, size_t newlen)
2486 {
2487 return -ENOSYS;
2488 }
2489
2490 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2491 void __user *oldval, size_t __user *oldlenp,
2492 void __user *newval, size_t newlen)
2493 {
2494 return -ENOSYS;
2495 }
2496
2497 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2498 void __user *oldval, size_t __user *oldlenp,
2499 void __user *newval, size_t newlen)
2500 {
2501 return -ENOSYS;
2502 }
2503
2504 #endif /* CONFIG_SYSCTL_SYSCALL */
2505
2506 /*
2507 * No sense putting this after each symbol definition, twice,
2508 * exception granted :-)
2509 */
2510 EXPORT_SYMBOL(proc_dointvec);
2511 EXPORT_SYMBOL(proc_dointvec_jiffies);
2512 EXPORT_SYMBOL(proc_dointvec_minmax);
2513 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2514 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2515 EXPORT_SYMBOL(proc_dostring);
2516 EXPORT_SYMBOL(proc_doulongvec_minmax);
2517 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2518 EXPORT_SYMBOL(register_sysctl_table);
2519 EXPORT_SYMBOL(sysctl_intvec);
2520 EXPORT_SYMBOL(sysctl_jiffies);
2521 EXPORT_SYMBOL(sysctl_ms_jiffies);
2522 EXPORT_SYMBOL(sysctl_string);
2523 EXPORT_SYMBOL(unregister_sysctl_table);
Linux kernel - Deliverables
This page took 0.107814 seconds and 6 git commands to generate.