?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
[PATCH] shmem and simple const super_operations
[linux-2.6.22.y-op.git] / kernel / sysctl.c
blob34b2301276cee8996fc6a1771053e588ec82417b
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 /* The default sysctl tables: */
163
164 static ctl_table root_table[] = {
165 {
166 .ctl_name = CTL_KERN,
167 .procname = "kernel",
168 .mode = 0555,
169 .child = kern_table,
170 },
171 {
172 .ctl_name = CTL_VM,
173 .procname = "vm",
174 .mode = 0555,
175 .child = vm_table,
176 },
177 #ifdef CONFIG_NET
178 {
179 .ctl_name = CTL_NET,
180 .procname = "net",
181 .mode = 0555,
182 .child = net_table,
183 },
184 #endif
185 {
186 .ctl_name = CTL_FS,
187 .procname = "fs",
188 .mode = 0555,
189 .child = fs_table,
190 },
191 {
192 .ctl_name = CTL_DEBUG,
193 .procname = "debug",
194 .mode = 0555,
195 .child = debug_table,
196 },
197 {
198 .ctl_name = CTL_DEV,
199 .procname = "dev",
200 .mode = 0555,
201 .child = dev_table,
202 },
203
204 { .ctl_name = 0 }
205 };
206
207 static ctl_table kern_table[] = {
208 {
209 .ctl_name = KERN_PANIC,
210 .procname = "panic",
211 .data = &panic_timeout,
212 .maxlen = sizeof(int),
213 .mode = 0644,
214 .proc_handler = &proc_dointvec,
215 },
216 {
217 .ctl_name = KERN_CORE_USES_PID,
218 .procname = "core_uses_pid",
219 .data = &core_uses_pid,
220 .maxlen = sizeof(int),
221 .mode = 0644,
222 .proc_handler = &proc_dointvec,
223 },
224 {
225 .ctl_name = KERN_CORE_PATTERN,
226 .procname = "core_pattern",
227 .data = core_pattern,
228 .maxlen = 128,
229 .mode = 0644,
230 .proc_handler = &proc_dostring,
231 .strategy = &sysctl_string,
232 },
233 #ifdef CONFIG_PROC_SYSCTL
234 {
235 .ctl_name = KERN_TAINTED,
236 .procname = "tainted",
237 .data = &tainted,
238 .maxlen = sizeof(int),
239 .mode = 0644,
240 .proc_handler = &proc_dointvec_taint,
241 },
242 #endif
243 {
244 .ctl_name = KERN_CAP_BSET,
245 .procname = "cap-bound",
246 .data = &cap_bset,
247 .maxlen = sizeof(kernel_cap_t),
248 .mode = 0600,
249 .proc_handler = &proc_dointvec_bset,
250 },
251 #ifdef CONFIG_BLK_DEV_INITRD
252 {
253 .ctl_name = KERN_REALROOTDEV,
254 .procname = "real-root-dev",
255 .data = &real_root_dev,
256 .maxlen = sizeof(int),
257 .mode = 0644,
258 .proc_handler = &proc_dointvec,
259 },
260 #endif
261 #ifdef __sparc__
262 {
263 .ctl_name = KERN_SPARC_REBOOT,
264 .procname = "reboot-cmd",
265 .data = reboot_command,
266 .maxlen = 256,
267 .mode = 0644,
268 .proc_handler = &proc_dostring,
269 .strategy = &sysctl_string,
270 },
271 {
272 .ctl_name = KERN_SPARC_STOP_A,
273 .procname = "stop-a",
274 .data = &stop_a_enabled,
275 .maxlen = sizeof (int),
276 .mode = 0644,
277 .proc_handler = &proc_dointvec,
278 },
279 {
280 .ctl_name = KERN_SPARC_SCONS_PWROFF,
281 .procname = "scons-poweroff",
282 .data = &scons_pwroff,
283 .maxlen = sizeof (int),
284 .mode = 0644,
285 .proc_handler = &proc_dointvec,
286 },
287 #endif
288 #ifdef __hppa__
289 {
290 .ctl_name = KERN_HPPA_PWRSW,
291 .procname = "soft-power",
292 .data = &pwrsw_enabled,
293 .maxlen = sizeof (int),
294 .mode = 0644,
295 .proc_handler = &proc_dointvec,
296 },
297 {
298 .ctl_name = KERN_HPPA_UNALIGNED,
299 .procname = "unaligned-trap",
300 .data = &unaligned_enabled,
301 .maxlen = sizeof (int),
302 .mode = 0644,
303 .proc_handler = &proc_dointvec,
304 },
305 #endif
306 {
307 .ctl_name = KERN_CTLALTDEL,
308 .procname = "ctrl-alt-del",
309 .data = &C_A_D,
310 .maxlen = sizeof(int),
311 .mode = 0644,
312 .proc_handler = &proc_dointvec,
313 },
314 {
315 .ctl_name = KERN_PRINTK,
316 .procname = "printk",
317 .data = &console_loglevel,
318 .maxlen = 4*sizeof(int),
319 .mode = 0644,
320 .proc_handler = &proc_dointvec,
321 },
322 #ifdef CONFIG_KMOD
323 {
324 .ctl_name = KERN_MODPROBE,
325 .procname = "modprobe",
326 .data = &modprobe_path,
327 .maxlen = KMOD_PATH_LEN,
328 .mode = 0644,
329 .proc_handler = &proc_dostring,
330 .strategy = &sysctl_string,
331 },
332 #endif
333 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
334 {
335 .ctl_name = KERN_HOTPLUG,
336 .procname = "hotplug",
337 .data = &uevent_helper,
338 .maxlen = UEVENT_HELPER_PATH_LEN,
339 .mode = 0644,
340 .proc_handler = &proc_dostring,
341 .strategy = &sysctl_string,
342 },
343 #endif
344 #ifdef CONFIG_CHR_DEV_SG
345 {
346 .ctl_name = KERN_SG_BIG_BUFF,
347 .procname = "sg-big-buff",
348 .data = &sg_big_buff,
349 .maxlen = sizeof (int),
350 .mode = 0444,
351 .proc_handler = &proc_dointvec,
352 },
353 #endif
354 #ifdef CONFIG_BSD_PROCESS_ACCT
355 {
356 .ctl_name = KERN_ACCT,
357 .procname = "acct",
358 .data = &acct_parm,
359 .maxlen = 3*sizeof(int),
360 .mode = 0644,
361 .proc_handler = &proc_dointvec,
362 },
363 #endif
364 #ifdef CONFIG_MAGIC_SYSRQ
365 {
366 .ctl_name = KERN_SYSRQ,
367 .procname = "sysrq",
368 .data = &__sysrq_enabled,
369 .maxlen = sizeof (int),
370 .mode = 0644,
371 .proc_handler = &proc_dointvec,
372 },
373 #endif
374 #ifdef CONFIG_PROC_SYSCTL
375 {
376 .ctl_name = KERN_CADPID,
377 .procname = "cad_pid",
378 .data = NULL,
379 .maxlen = sizeof (int),
380 .mode = 0600,
381 .proc_handler = &proc_do_cad_pid,
382 },
383 #endif
384 {
385 .ctl_name = KERN_MAX_THREADS,
386 .procname = "threads-max",
387 .data = &max_threads,
388 .maxlen = sizeof(int),
389 .mode = 0644,
390 .proc_handler = &proc_dointvec,
391 },
392 {
393 .ctl_name = KERN_RANDOM,
394 .procname = "random",
395 .mode = 0555,
396 .child = random_table,
397 },
398 #ifdef CONFIG_UNIX98_PTYS
399 {
400 .ctl_name = KERN_PTY,
401 .procname = "pty",
402 .mode = 0555,
403 .child = pty_table,
404 },
405 #endif
406 {
407 .ctl_name = KERN_OVERFLOWUID,
408 .procname = "overflowuid",
409 .data = &overflowuid,
410 .maxlen = sizeof(int),
411 .mode = 0644,
412 .proc_handler = &proc_dointvec_minmax,
413 .strategy = &sysctl_intvec,
414 .extra1 = &minolduid,
415 .extra2 = &maxolduid,
416 },
417 {
418 .ctl_name = KERN_OVERFLOWGID,
419 .procname = "overflowgid",
420 .data = &overflowgid,
421 .maxlen = sizeof(int),
422 .mode = 0644,
423 .proc_handler = &proc_dointvec_minmax,
424 .strategy = &sysctl_intvec,
425 .extra1 = &minolduid,
426 .extra2 = &maxolduid,
427 },
428 #ifdef CONFIG_S390
429 #ifdef CONFIG_MATHEMU
430 {
431 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
432 .procname = "ieee_emulation_warnings",
433 .data = &sysctl_ieee_emulation_warnings,
434 .maxlen = sizeof(int),
435 .mode = 0644,
436 .proc_handler = &proc_dointvec,
437 },
438 #endif
439 #ifdef CONFIG_NO_IDLE_HZ
440 {
441 .ctl_name = KERN_HZ_TIMER,
442 .procname = "hz_timer",
443 .data = &sysctl_hz_timer,
444 .maxlen = sizeof(int),
445 .mode = 0644,
446 .proc_handler = &proc_dointvec,
447 },
448 #endif
449 {
450 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
451 .procname = "userprocess_debug",
452 .data = &sysctl_userprocess_debug,
453 .maxlen = sizeof(int),
454 .mode = 0644,
455 .proc_handler = &proc_dointvec,
456 },
457 #endif
458 {
459 .ctl_name = KERN_PIDMAX,
460 .procname = "pid_max",
461 .data = &pid_max,
462 .maxlen = sizeof (int),
463 .mode = 0644,
464 .proc_handler = &proc_dointvec_minmax,
465 .strategy = sysctl_intvec,
466 .extra1 = &pid_max_min,
467 .extra2 = &pid_max_max,
468 },
469 {
470 .ctl_name = KERN_PANIC_ON_OOPS,
471 .procname = "panic_on_oops",
472 .data = &panic_on_oops,
473 .maxlen = sizeof(int),
474 .mode = 0644,
475 .proc_handler = &proc_dointvec,
476 },
477 {
478 .ctl_name = KERN_PRINTK_RATELIMIT,
479 .procname = "printk_ratelimit",
480 .data = &printk_ratelimit_jiffies,
481 .maxlen = sizeof(int),
482 .mode = 0644,
483 .proc_handler = &proc_dointvec_jiffies,
484 .strategy = &sysctl_jiffies,
485 },
486 {
487 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
488 .procname = "printk_ratelimit_burst",
489 .data = &printk_ratelimit_burst,
490 .maxlen = sizeof(int),
491 .mode = 0644,
492 .proc_handler = &proc_dointvec,
493 },
494 {
495 .ctl_name = KERN_NGROUPS_MAX,
496 .procname = "ngroups_max",
497 .data = &ngroups_max,
498 .maxlen = sizeof (int),
499 .mode = 0444,
500 .proc_handler = &proc_dointvec,
501 },
502 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
503 {
504 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
505 .procname = "unknown_nmi_panic",
506 .data = &unknown_nmi_panic,
507 .maxlen = sizeof (int),
508 .mode = 0644,
509 .proc_handler = &proc_dointvec,
510 },
511 {
512 .ctl_name = KERN_NMI_WATCHDOG,
513 .procname = "nmi_watchdog",
514 .data = &nmi_watchdog_enabled,
515 .maxlen = sizeof (int),
516 .mode = 0644,
517 .proc_handler = &proc_nmi_enabled,
518 },
519 #endif
520 #if defined(CONFIG_X86)
521 {
522 .ctl_name = KERN_PANIC_ON_NMI,
523 .procname = "panic_on_unrecovered_nmi",
524 .data = &panic_on_unrecovered_nmi,
525 .maxlen = sizeof(int),
526 .mode = 0644,
527 .proc_handler = &proc_dointvec,
528 },
529 {
530 .ctl_name = KERN_BOOTLOADER_TYPE,
531 .procname = "bootloader_type",
532 .data = &bootloader_type,
533 .maxlen = sizeof (int),
534 .mode = 0444,
535 .proc_handler = &proc_dointvec,
536 },
537 {
538 .ctl_name = CTL_UNNUMBERED,
539 .procname = "kstack_depth_to_print",
540 .data = &kstack_depth_to_print,
541 .maxlen = sizeof(int),
542 .mode = 0644,
543 .proc_handler = &proc_dointvec,
544 },
545 #endif
546 #if defined(CONFIG_MMU)
547 {
548 .ctl_name = KERN_RANDOMIZE,
549 .procname = "randomize_va_space",
550 .data = &randomize_va_space,
551 .maxlen = sizeof(int),
552 .mode = 0644,
553 .proc_handler = &proc_dointvec,
554 },
555 #endif
556 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
557 {
558 .ctl_name = KERN_SPIN_RETRY,
559 .procname = "spin_retry",
560 .data = &spin_retry,
561 .maxlen = sizeof (int),
562 .mode = 0644,
563 .proc_handler = &proc_dointvec,
564 },
565 #endif
566 #ifdef CONFIG_ACPI_SLEEP
567 {
568 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
569 .procname = "acpi_video_flags",
570 .data = &acpi_video_flags,
571 .maxlen = sizeof (unsigned long),
572 .mode = 0644,
573 .proc_handler = &proc_doulongvec_minmax,
574 },
575 #endif
576 #ifdef CONFIG_IA64
577 {
578 .ctl_name = KERN_IA64_UNALIGNED,
579 .procname = "ignore-unaligned-usertrap",
580 .data = &no_unaligned_warning,
581 .maxlen = sizeof (int),
582 .mode = 0644,
583 .proc_handler = &proc_dointvec,
584 },
585 #endif
586 #ifdef CONFIG_COMPAT
587 {
588 .ctl_name = KERN_COMPAT_LOG,
589 .procname = "compat-log",
590 .data = &compat_log,
591 .maxlen = sizeof (int),
592 .mode = 0644,
593 .proc_handler = &proc_dointvec,
594 },
595 #endif
596 #ifdef CONFIG_RT_MUTEXES
597 {
598 .ctl_name = KERN_MAX_LOCK_DEPTH,
599 .procname = "max_lock_depth",
600 .data = &max_lock_depth,
601 .maxlen = sizeof(int),
602 .mode = 0644,
603 .proc_handler = &proc_dointvec,
604 },
605 #endif
606
607 { .ctl_name = 0 }
608 };
609
610 /* Constants for minimum and maximum testing in vm_table.
611 We use these as one-element integer vectors. */
612 static int zero;
613 static int one_hundred = 100;
614
615
616 static ctl_table vm_table[] = {
617 {
618 .ctl_name = VM_OVERCOMMIT_MEMORY,
619 .procname = "overcommit_memory",
620 .data = &sysctl_overcommit_memory,
621 .maxlen = sizeof(sysctl_overcommit_memory),
622 .mode = 0644,
623 .proc_handler = &proc_dointvec,
624 },
625 {
626 .ctl_name = VM_PANIC_ON_OOM,
627 .procname = "panic_on_oom",
628 .data = &sysctl_panic_on_oom,
629 .maxlen = sizeof(sysctl_panic_on_oom),
630 .mode = 0644,
631 .proc_handler = &proc_dointvec,
632 },
633 {
634 .ctl_name = VM_OVERCOMMIT_RATIO,
635 .procname = "overcommit_ratio",
636 .data = &sysctl_overcommit_ratio,
637 .maxlen = sizeof(sysctl_overcommit_ratio),
638 .mode = 0644,
639 .proc_handler = &proc_dointvec,
640 },
641 {
642 .ctl_name = VM_PAGE_CLUSTER,
643 .procname = "page-cluster",
644 .data = &page_cluster,
645 .maxlen = sizeof(int),
646 .mode = 0644,
647 .proc_handler = &proc_dointvec,
648 },
649 {
650 .ctl_name = VM_DIRTY_BACKGROUND,
651 .procname = "dirty_background_ratio",
652 .data = &dirty_background_ratio,
653 .maxlen = sizeof(dirty_background_ratio),
654 .mode = 0644,
655 .proc_handler = &proc_dointvec_minmax,
656 .strategy = &sysctl_intvec,
657 .extra1 = &zero,
658 .extra2 = &one_hundred,
659 },
660 {
661 .ctl_name = VM_DIRTY_RATIO,
662 .procname = "dirty_ratio",
663 .data = &vm_dirty_ratio,
664 .maxlen = sizeof(vm_dirty_ratio),
665 .mode = 0644,
666 .proc_handler = &proc_dointvec_minmax,
667 .strategy = &sysctl_intvec,
668 .extra1 = &zero,
669 .extra2 = &one_hundred,
670 },
671 {
672 .ctl_name = VM_DIRTY_WB_CS,
673 .procname = "dirty_writeback_centisecs",
674 .data = &dirty_writeback_interval,
675 .maxlen = sizeof(dirty_writeback_interval),
676 .mode = 0644,
677 .proc_handler = &dirty_writeback_centisecs_handler,
678 },
679 {
680 .ctl_name = VM_DIRTY_EXPIRE_CS,
681 .procname = "dirty_expire_centisecs",
682 .data = &dirty_expire_interval,
683 .maxlen = sizeof(dirty_expire_interval),
684 .mode = 0644,
685 .proc_handler = &proc_dointvec_userhz_jiffies,
686 },
687 {
688 .ctl_name = VM_NR_PDFLUSH_THREADS,
689 .procname = "nr_pdflush_threads",
690 .data = &nr_pdflush_threads,
691 .maxlen = sizeof nr_pdflush_threads,
692 .mode = 0444 /* read-only*/,
693 .proc_handler = &proc_dointvec,
694 },
695 {
696 .ctl_name = VM_SWAPPINESS,
697 .procname = "swappiness",
698 .data = &vm_swappiness,
699 .maxlen = sizeof(vm_swappiness),
700 .mode = 0644,
701 .proc_handler = &proc_dointvec_minmax,
702 .strategy = &sysctl_intvec,
703 .extra1 = &zero,
704 .extra2 = &one_hundred,
705 },
706 #ifdef CONFIG_HUGETLB_PAGE
707 {
708 .ctl_name = VM_HUGETLB_PAGES,
709 .procname = "nr_hugepages",
710 .data = &max_huge_pages,
711 .maxlen = sizeof(unsigned long),
712 .mode = 0644,
713 .proc_handler = &hugetlb_sysctl_handler,
714 .extra1 = (void *)&hugetlb_zero,
715 .extra2 = (void *)&hugetlb_infinity,
716 },
717 {
718 .ctl_name = VM_HUGETLB_GROUP,
719 .procname = "hugetlb_shm_group",
720 .data = &sysctl_hugetlb_shm_group,
721 .maxlen = sizeof(gid_t),
722 .mode = 0644,
723 .proc_handler = &proc_dointvec,
724 },
725 #endif
726 {
727 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
728 .procname = "lowmem_reserve_ratio",
729 .data = &sysctl_lowmem_reserve_ratio,
730 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
731 .mode = 0644,
732 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
733 .strategy = &sysctl_intvec,
734 },
735 {
736 .ctl_name = VM_DROP_PAGECACHE,
737 .procname = "drop_caches",
738 .data = &sysctl_drop_caches,
739 .maxlen = sizeof(int),
740 .mode = 0644,
741 .proc_handler = drop_caches_sysctl_handler,
742 .strategy = &sysctl_intvec,
743 },
744 {
745 .ctl_name = VM_MIN_FREE_KBYTES,
746 .procname = "min_free_kbytes",
747 .data = &min_free_kbytes,
748 .maxlen = sizeof(min_free_kbytes),
749 .mode = 0644,
750 .proc_handler = &min_free_kbytes_sysctl_handler,
751 .strategy = &sysctl_intvec,
752 .extra1 = &zero,
753 },
754 {
755 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
756 .procname = "percpu_pagelist_fraction",
757 .data = &percpu_pagelist_fraction,
758 .maxlen = sizeof(percpu_pagelist_fraction),
759 .mode = 0644,
760 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
761 .strategy = &sysctl_intvec,
762 .extra1 = &min_percpu_pagelist_fract,
763 },
764 #ifdef CONFIG_MMU
765 {
766 .ctl_name = VM_MAX_MAP_COUNT,
767 .procname = "max_map_count",
768 .data = &sysctl_max_map_count,
769 .maxlen = sizeof(sysctl_max_map_count),
770 .mode = 0644,
771 .proc_handler = &proc_dointvec
772 },
773 #endif
774 {
775 .ctl_name = VM_LAPTOP_MODE,
776 .procname = "laptop_mode",
777 .data = &laptop_mode,
778 .maxlen = sizeof(laptop_mode),
779 .mode = 0644,
780 .proc_handler = &proc_dointvec_jiffies,
781 .strategy = &sysctl_jiffies,
782 },
783 {
784 .ctl_name = VM_BLOCK_DUMP,
785 .procname = "block_dump",
786 .data = &block_dump,
787 .maxlen = sizeof(block_dump),
788 .mode = 0644,
789 .proc_handler = &proc_dointvec,
790 .strategy = &sysctl_intvec,
791 .extra1 = &zero,
792 },
793 {
794 .ctl_name = VM_VFS_CACHE_PRESSURE,
795 .procname = "vfs_cache_pressure",
796 .data = &sysctl_vfs_cache_pressure,
797 .maxlen = sizeof(sysctl_vfs_cache_pressure),
798 .mode = 0644,
799 .proc_handler = &proc_dointvec,
800 .strategy = &sysctl_intvec,
801 .extra1 = &zero,
802 },
803 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
804 {
805 .ctl_name = VM_LEGACY_VA_LAYOUT,
806 .procname = "legacy_va_layout",
807 .data = &sysctl_legacy_va_layout,
808 .maxlen = sizeof(sysctl_legacy_va_layout),
809 .mode = 0644,
810 .proc_handler = &proc_dointvec,
811 .strategy = &sysctl_intvec,
812 .extra1 = &zero,
813 },
814 #endif
815 #ifdef CONFIG_NUMA
816 {
817 .ctl_name = VM_ZONE_RECLAIM_MODE,
818 .procname = "zone_reclaim_mode",
819 .data = &zone_reclaim_mode,
820 .maxlen = sizeof(zone_reclaim_mode),
821 .mode = 0644,
822 .proc_handler = &proc_dointvec,
823 .strategy = &sysctl_intvec,
824 .extra1 = &zero,
825 },
826 {
827 .ctl_name = VM_MIN_UNMAPPED,
828 .procname = "min_unmapped_ratio",
829 .data = &sysctl_min_unmapped_ratio,
830 .maxlen = sizeof(sysctl_min_unmapped_ratio),
831 .mode = 0644,
832 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
833 .strategy = &sysctl_intvec,
834 .extra1 = &zero,
835 .extra2 = &one_hundred,
836 },
837 {
838 .ctl_name = VM_MIN_SLAB,
839 .procname = "min_slab_ratio",
840 .data = &sysctl_min_slab_ratio,
841 .maxlen = sizeof(sysctl_min_slab_ratio),
842 .mode = 0644,
843 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
844 .strategy = &sysctl_intvec,
845 .extra1 = &zero,
846 .extra2 = &one_hundred,
847 },
848 #endif
849 #ifdef CONFIG_X86_32
850 {
851 .ctl_name = VM_VDSO_ENABLED,
852 .procname = "vdso_enabled",
853 .data = &vdso_enabled,
854 .maxlen = sizeof(vdso_enabled),
855 .mode = 0644,
856 .proc_handler = &proc_dointvec,
857 .strategy = &sysctl_intvec,
858 .extra1 = &zero,
859 },
860 #endif
861 { .ctl_name = 0 }
862 };
863
864 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
865 static ctl_table binfmt_misc_table[] = {
866 { .ctl_name = 0 }
867 };
868 #endif
869
870 static ctl_table fs_table[] = {
871 {
872 .ctl_name = FS_NRINODE,
873 .procname = "inode-nr",
874 .data = &inodes_stat,
875 .maxlen = 2*sizeof(int),
876 .mode = 0444,
877 .proc_handler = &proc_dointvec,
878 },
879 {
880 .ctl_name = FS_STATINODE,
881 .procname = "inode-state",
882 .data = &inodes_stat,
883 .maxlen = 7*sizeof(int),
884 .mode = 0444,
885 .proc_handler = &proc_dointvec,
886 },
887 {
888 .ctl_name = FS_NRFILE,
889 .procname = "file-nr",
890 .data = &files_stat,
891 .maxlen = 3*sizeof(int),
892 .mode = 0444,
893 .proc_handler = &proc_nr_files,
894 },
895 {
896 .ctl_name = FS_MAXFILE,
897 .procname = "file-max",
898 .data = &files_stat.max_files,
899 .maxlen = sizeof(int),
900 .mode = 0644,
901 .proc_handler = &proc_dointvec,
902 },
903 {
904 .ctl_name = FS_DENTRY,
905 .procname = "dentry-state",
906 .data = &dentry_stat,
907 .maxlen = 6*sizeof(int),
908 .mode = 0444,
909 .proc_handler = &proc_dointvec,
910 },
911 {
912 .ctl_name = FS_OVERFLOWUID,
913 .procname = "overflowuid",
914 .data = &fs_overflowuid,
915 .maxlen = sizeof(int),
916 .mode = 0644,
917 .proc_handler = &proc_dointvec_minmax,
918 .strategy = &sysctl_intvec,
919 .extra1 = &minolduid,
920 .extra2 = &maxolduid,
921 },
922 {
923 .ctl_name = FS_OVERFLOWGID,
924 .procname = "overflowgid",
925 .data = &fs_overflowgid,
926 .maxlen = sizeof(int),
927 .mode = 0644,
928 .proc_handler = &proc_dointvec_minmax,
929 .strategy = &sysctl_intvec,
930 .extra1 = &minolduid,
931 .extra2 = &maxolduid,
932 },
933 {
934 .ctl_name = FS_LEASES,
935 .procname = "leases-enable",
936 .data = &leases_enable,
937 .maxlen = sizeof(int),
938 .mode = 0644,
939 .proc_handler = &proc_dointvec,
940 },
941 #ifdef CONFIG_DNOTIFY
942 {
943 .ctl_name = FS_DIR_NOTIFY,
944 .procname = "dir-notify-enable",
945 .data = &dir_notify_enable,
946 .maxlen = sizeof(int),
947 .mode = 0644,
948 .proc_handler = &proc_dointvec,
949 },
950 #endif
951 #ifdef CONFIG_MMU
952 {
953 .ctl_name = FS_LEASE_TIME,
954 .procname = "lease-break-time",
955 .data = &lease_break_time,
956 .maxlen = sizeof(int),
957 .mode = 0644,
958 .proc_handler = &proc_dointvec,
959 },
960 {
961 .ctl_name = FS_AIO_NR,
962 .procname = "aio-nr",
963 .data = &aio_nr,
964 .maxlen = sizeof(aio_nr),
965 .mode = 0444,
966 .proc_handler = &proc_doulongvec_minmax,
967 },
968 {
969 .ctl_name = FS_AIO_MAX_NR,
970 .procname = "aio-max-nr",
971 .data = &aio_max_nr,
972 .maxlen = sizeof(aio_max_nr),
973 .mode = 0644,
974 .proc_handler = &proc_doulongvec_minmax,
975 },
976 #ifdef CONFIG_INOTIFY_USER
977 {
978 .ctl_name = FS_INOTIFY,
979 .procname = "inotify",
980 .mode = 0555,
981 .child = inotify_table,
982 },
983 #endif
984 #endif
985 {
986 .ctl_name = KERN_SETUID_DUMPABLE,
987 .procname = "suid_dumpable",
988 .data = &suid_dumpable,
989 .maxlen = sizeof(int),
990 .mode = 0644,
991 .proc_handler = &proc_dointvec,
992 },
993 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
994 {
995 .ctl_name = CTL_UNNUMBERED,
996 .procname = "binfmt_misc",
997 .mode = 0555,
998 .child = binfmt_misc_table,
999 },
1000 #endif
1001 { .ctl_name = 0 }
1002 };
1003
1004 static ctl_table debug_table[] = {
1005 { .ctl_name = 0 }
1006 };
1007
1008 static ctl_table dev_table[] = {
1009 { .ctl_name = 0 }
1010 };
1011
1012 static DEFINE_SPINLOCK(sysctl_lock);
1013
1014 /* called under sysctl_lock */
1015 static int use_table(struct ctl_table_header *p)
1016 {
1017 if (unlikely(p->unregistering))
1018 return 0;
1019 p->used++;
1020 return 1;
1021 }
1022
1023 /* called under sysctl_lock */
1024 static void unuse_table(struct ctl_table_header *p)
1025 {
1026 if (!--p->used)
1027 if (unlikely(p->unregistering))
1028 complete(p->unregistering);
1029 }
1030
1031 /* called under sysctl_lock, will reacquire if has to wait */
1032 static void start_unregistering(struct ctl_table_header *p)
1033 {
1034 /*
1035 * if p->used is 0, nobody will ever touch that entry again;
1036 * we'll eliminate all paths to it before dropping sysctl_lock
1037 */
1038 if (unlikely(p->used)) {
1039 struct completion wait;
1040 init_completion(&wait);
1041 p->unregistering = &wait;
1042 spin_unlock(&sysctl_lock);
1043 wait_for_completion(&wait);
1044 spin_lock(&sysctl_lock);
1045 }
1046 /*
1047 * do not remove from the list until nobody holds it; walking the
1048 * list in do_sysctl() relies on that.
1049 */
1050 list_del_init(&p->ctl_entry);
1051 }
1052
1053 void sysctl_head_finish(struct ctl_table_header *head)
1054 {
1055 if (!head)
1056 return;
1057 spin_lock(&sysctl_lock);
1058 unuse_table(head);
1059 spin_unlock(&sysctl_lock);
1060 }
1061
1062 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1063 {
1064 struct ctl_table_header *head;
1065 struct list_head *tmp;
1066 spin_lock(&sysctl_lock);
1067 if (prev) {
1068 tmp = &prev->ctl_entry;
1069 unuse_table(prev);
1070 goto next;
1071 }
1072 tmp = &root_table_header.ctl_entry;
1073 for (;;) {
1074 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1075
1076 if (!use_table(head))
1077 goto next;
1078 spin_unlock(&sysctl_lock);
1079 return head;
1080 next:
1081 tmp = tmp->next;
1082 if (tmp == &root_table_header.ctl_entry)
1083 break;
1084 }
1085 spin_unlock(&sysctl_lock);
1086 return NULL;
1087 }
1088
1089 #ifdef CONFIG_SYSCTL_SYSCALL
1090 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1091 void __user *newval, size_t newlen)
1092 {
1093 struct ctl_table_header *head;
1094 int error = -ENOTDIR;
1095
1096 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1097 return -ENOTDIR;
1098 if (oldval) {
1099 int old_len;
1100 if (!oldlenp || get_user(old_len, oldlenp))
1101 return -EFAULT;
1102 }
1103
1104 for (head = sysctl_head_next(NULL); head;
1105 head = sysctl_head_next(head)) {
1106 error = parse_table(name, nlen, oldval, oldlenp,
1107 newval, newlen, head->ctl_table);
1108 if (error != -ENOTDIR) {
1109 sysctl_head_finish(head);
1110 break;
1111 }
1112 }
1113 return error;
1114 }
1115
1116 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1117 {
1118 struct __sysctl_args tmp;
1119 int error;
1120
1121 if (copy_from_user(&tmp, args, sizeof(tmp)))
1122 return -EFAULT;
1123
1124 lock_kernel();
1125 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1126 tmp.newval, tmp.newlen);
1127 unlock_kernel();
1128 return error;
1129 }
1130 #endif /* CONFIG_SYSCTL_SYSCALL */
1131
1132 /*
1133 * sysctl_perm does NOT grant the superuser all rights automatically, because
1134 * some sysctl variables are readonly even to root.
1135 */
1136
1137 static int test_perm(int mode, int op)
1138 {
1139 if (!current->euid)
1140 mode >>= 6;
1141 else if (in_egroup_p(0))
1142 mode >>= 3;
1143 if ((mode & op & 0007) == op)
1144 return 0;
1145 return -EACCES;
1146 }
1147
1148 int sysctl_perm(ctl_table *table, int op)
1149 {
1150 int error;
1151 error = security_sysctl(table, op);
1152 if (error)
1153 return error;
1154 return test_perm(table->mode, op);
1155 }
1156
1157 #ifdef CONFIG_SYSCTL_SYSCALL
1158 static int parse_table(int __user *name, int nlen,
1159 void __user *oldval, size_t __user *oldlenp,
1160 void __user *newval, size_t newlen,
1161 ctl_table *table)
1162 {
1163 int n;
1164 repeat:
1165 if (!nlen)
1166 return -ENOTDIR;
1167 if (get_user(n, name))
1168 return -EFAULT;
1169 for ( ; table->ctl_name || table->procname; table++) {
1170 if (!table->ctl_name)
1171 continue;
1172 if (n == table->ctl_name) {
1173 int error;
1174 if (table->child) {
1175 if (sysctl_perm(table, 001))
1176 return -EPERM;
1177 name++;
1178 nlen--;
1179 table = table->child;
1180 goto repeat;
1181 }
1182 error = do_sysctl_strategy(table, name, nlen,
1183 oldval, oldlenp,
1184 newval, newlen);
1185 return error;
1186 }
1187 }
1188 return -ENOTDIR;
1189 }
1190
1191 /* Perform the actual read/write of a sysctl table entry. */
1192 int do_sysctl_strategy (ctl_table *table,
1193 int __user *name, int nlen,
1194 void __user *oldval, size_t __user *oldlenp,
1195 void __user *newval, size_t newlen)
1196 {
1197 int op = 0, rc;
1198 size_t len;
1199
1200 if (oldval)
1201 op |= 004;
1202 if (newval)
1203 op |= 002;
1204 if (sysctl_perm(table, op))
1205 return -EPERM;
1206
1207 if (table->strategy) {
1208 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1209 newval, newlen);
1210 if (rc < 0)
1211 return rc;
1212 if (rc > 0)
1213 return 0;
1214 }
1215
1216 /* If there is no strategy routine, or if the strategy returns
1217 * zero, proceed with automatic r/w */
1218 if (table->data && table->maxlen) {
1219 if (oldval && oldlenp) {
1220 if (get_user(len, oldlenp))
1221 return -EFAULT;
1222 if (len) {
1223 if (len > table->maxlen)
1224 len = table->maxlen;
1225 if(copy_to_user(oldval, table->data, len))
1226 return -EFAULT;
1227 if(put_user(len, oldlenp))
1228 return -EFAULT;
1229 }
1230 }
1231 if (newval && newlen) {
1232 len = newlen;
1233 if (len > table->maxlen)
1234 len = table->maxlen;
1235 if(copy_from_user(table->data, newval, len))
1236 return -EFAULT;
1237 }
1238 }
1239 return 0;
1240 }
1241 #endif /* CONFIG_SYSCTL_SYSCALL */
1242
1243 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1244 {
1245 for (; table->ctl_name || table->procname; table++) {
1246 table->parent = parent;
1247 if (table->child)
1248 sysctl_set_parent(table, table->child);
1249 }
1250 }
1251
1252 static __init int sysctl_init(void)
1253 {
1254 sysctl_set_parent(NULL, root_table);
1255 return 0;
1256 }
1257
1258 core_initcall(sysctl_init);
1259
1260 /**
1261 * register_sysctl_table - register a sysctl hierarchy
1262 * @table: the top-level table structure
1263 *
1264 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1265 * array. An entry with a ctl_name of 0 terminates the table.
1266 *
1267 * The members of the &ctl_table structure are used as follows:
1268 *
1269 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1270 * must be unique within that level of sysctl
1271 *
1272 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1273 * enter a sysctl file
1274 *
1275 * data - a pointer to data for use by proc_handler
1276 *
1277 * maxlen - the maximum size in bytes of the data
1278 *
1279 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1280 *
1281 * child - a pointer to the child sysctl table if this entry is a directory, or
1282 * %NULL.
1283 *
1284 * proc_handler - the text handler routine (described below)
1285 *
1286 * strategy - the strategy routine (described below)
1287 *
1288 * de - for internal use by the sysctl routines
1289 *
1290 * extra1, extra2 - extra pointers usable by the proc handler routines
1291 *
1292 * Leaf nodes in the sysctl tree will be represented by a single file
1293 * under /proc; non-leaf nodes will be represented by directories.
1294 *
1295 * sysctl(2) can automatically manage read and write requests through
1296 * the sysctl table. The data and maxlen fields of the ctl_table
1297 * struct enable minimal validation of the values being written to be
1298 * performed, and the mode field allows minimal authentication.
1299 *
1300 * More sophisticated management can be enabled by the provision of a
1301 * strategy routine with the table entry. This will be called before
1302 * any automatic read or write of the data is performed.
1303 *
1304 * The strategy routine may return
1305 *
1306 * < 0 - Error occurred (error is passed to user process)
1307 *
1308 * 0 - OK - proceed with automatic read or write.
1309 *
1310 * > 0 - OK - read or write has been done by the strategy routine, so
1311 * return immediately.
1312 *
1313 * There must be a proc_handler routine for any terminal nodes
1314 * mirrored under /proc/sys (non-terminals are handled by a built-in
1315 * directory handler). Several default handlers are available to
1316 * cover common cases -
1317 *
1318 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1319 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1320 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1321 *
1322 * It is the handler's job to read the input buffer from user memory
1323 * and process it. The handler should return 0 on success.
1324 *
1325 * This routine returns %NULL on a failure to register, and a pointer
1326 * to the table header on success.
1327 */
1328 struct ctl_table_header *register_sysctl_table(ctl_table * table)
1329 {
1330 struct ctl_table_header *tmp;
1331 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1332 if (!tmp)
1333 return NULL;
1334 tmp->ctl_table = table;
1335 INIT_LIST_HEAD(&tmp->ctl_entry);
1336 tmp->used = 0;
1337 tmp->unregistering = NULL;
1338 sysctl_set_parent(NULL, table);
1339 spin_lock(&sysctl_lock);
1340 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
1341 spin_unlock(&sysctl_lock);
1342 return tmp;
1343 }
1344
1345 /**
1346 * unregister_sysctl_table - unregister a sysctl table hierarchy
1347 * @header: the header returned from register_sysctl_table
1348 *
1349 * Unregisters the sysctl table and all children. proc entries may not
1350 * actually be removed until they are no longer used by anyone.
1351 */
1352 void unregister_sysctl_table(struct ctl_table_header * header)
1353 {
1354 might_sleep();
1355 spin_lock(&sysctl_lock);
1356 start_unregistering(header);
1357 spin_unlock(&sysctl_lock);
1358 kfree(header);
1359 }
1360
1361 #else /* !CONFIG_SYSCTL */
1362 struct ctl_table_header *register_sysctl_table(ctl_table * table)
1363 {
1364 return NULL;
1365 }
1366
1367 void unregister_sysctl_table(struct ctl_table_header * table)
1368 {
1369 }
1370
1371 #endif /* CONFIG_SYSCTL */
1372
1373 /*
1374 * /proc/sys support
1375 */
1376
1377 #ifdef CONFIG_PROC_SYSCTL
1378
1379 static int _proc_do_string(void* data, int maxlen, int write,
1380 struct file *filp, void __user *buffer,
1381 size_t *lenp, loff_t *ppos)
1382 {
1383 size_t len;
1384 char __user *p;
1385 char c;
1386
1387 if (!data || !maxlen || !*lenp) {
1388 *lenp = 0;
1389 return 0;
1390 }
1391
1392 if (write) {
1393 len = 0;
1394 p = buffer;
1395 while (len < *lenp) {
1396 if (get_user(c, p++))
1397 return -EFAULT;
1398 if (c == 0 || c == '\n')
1399 break;
1400 len++;
1401 }
1402 if (len >= maxlen)
1403 len = maxlen-1;
1404 if(copy_from_user(data, buffer, len))
1405 return -EFAULT;
1406 ((char *) data)[len] = 0;
1407 *ppos += *lenp;
1408 } else {
1409 len = strlen(data);
1410 if (len > maxlen)
1411 len = maxlen;
1412
1413 if (*ppos > len) {
1414 *lenp = 0;
1415 return 0;
1416 }
1417
1418 data += *ppos;
1419 len -= *ppos;
1420
1421 if (len > *lenp)
1422 len = *lenp;
1423 if (len)
1424 if(copy_to_user(buffer, data, len))
1425 return -EFAULT;
1426 if (len < *lenp) {
1427 if(put_user('\n', ((char __user *) buffer) + len))
1428 return -EFAULT;
1429 len++;
1430 }
1431 *lenp = len;
1432 *ppos += len;
1433 }
1434 return 0;
1435 }
1436
1437 /**
1438 * proc_dostring - read a string sysctl
1439 * @table: the sysctl table
1440 * @write: %TRUE if this is a write to the sysctl file
1441 * @filp: the file structure
1442 * @buffer: the user buffer
1443 * @lenp: the size of the user buffer
1444 * @ppos: file position
1445 *
1446 * Reads/writes a string from/to the user buffer. If the kernel
1447 * buffer provided is not large enough to hold the string, the
1448 * string is truncated. The copied string is %NULL-terminated.
1449 * If the string is being read by the user process, it is copied
1450 * and a newline '\n' is added. It is truncated if the buffer is
1451 * not large enough.
1452 *
1453 * Returns 0 on success.
1454 */
1455 int proc_dostring(ctl_table *table, int write, struct file *filp,
1456 void __user *buffer, size_t *lenp, loff_t *ppos)
1457 {
1458 return _proc_do_string(table->data, table->maxlen, write, filp,
1459 buffer, lenp, ppos);
1460 }
1461
1462
1463 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1464 int *valp,
1465 int write, void *data)
1466 {
1467 if (write) {
1468 *valp = *negp ? -*lvalp : *lvalp;
1469 } else {
1470 int val = *valp;
1471 if (val < 0) {
1472 *negp = -1;
1473 *lvalp = (unsigned long)-val;
1474 } else {
1475 *negp = 0;
1476 *lvalp = (unsigned long)val;
1477 }
1478 }
1479 return 0;
1480 }
1481
1482 static int __do_proc_dointvec(void *tbl_data, ctl_table *table,
1483 int write, struct file *filp, void __user *buffer,
1484 size_t *lenp, loff_t *ppos,
1485 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1486 int write, void *data),
1487 void *data)
1488 {
1489 #define TMPBUFLEN 21
1490 int *i, vleft, first=1, neg, val;
1491 unsigned long lval;
1492 size_t left, len;
1493
1494 char buf[TMPBUFLEN], *p;
1495 char __user *s = buffer;
1496
1497 if (!tbl_data || !table->maxlen || !*lenp ||
1498 (*ppos && !write)) {
1499 *lenp = 0;
1500 return 0;
1501 }
1502
1503 i = (int *) tbl_data;
1504 vleft = table->maxlen / sizeof(*i);
1505 left = *lenp;
1506
1507 if (!conv)
1508 conv = do_proc_dointvec_conv;
1509
1510 for (; left && vleft--; i++, first=0) {
1511 if (write) {
1512 while (left) {
1513 char c;
1514 if (get_user(c, s))
1515 return -EFAULT;
1516 if (!isspace(c))
1517 break;
1518 left--;
1519 s++;
1520 }
1521 if (!left)
1522 break;
1523 neg = 0;
1524 len = left;
1525 if (len > sizeof(buf) - 1)
1526 len = sizeof(buf) - 1;
1527 if (copy_from_user(buf, s, len))
1528 return -EFAULT;
1529 buf[len] = 0;
1530 p = buf;
1531 if (*p == '-' && left > 1) {
1532 neg = 1;
1533 p++;
1534 }
1535 if (*p < '0' || *p > '9')
1536 break;
1537
1538 lval = simple_strtoul(p, &p, 0);
1539
1540 len = p-buf;
1541 if ((len < left) && *p && !isspace(*p))
1542 break;
1543 if (neg)
1544 val = -val;
1545 s += len;
1546 left -= len;
1547
1548 if (conv(&neg, &lval, i, 1, data))
1549 break;
1550 } else {
1551 p = buf;
1552 if (!first)
1553 *p++ = '\t';
1554
1555 if (conv(&neg, &lval, i, 0, data))
1556 break;
1557
1558 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1559 len = strlen(buf);
1560 if (len > left)
1561 len = left;
1562 if(copy_to_user(s, buf, len))
1563 return -EFAULT;
1564 left -= len;
1565 s += len;
1566 }
1567 }
1568
1569 if (!write && !first && left) {
1570 if(put_user('\n', s))
1571 return -EFAULT;
1572 left--, s++;
1573 }
1574 if (write) {
1575 while (left) {
1576 char c;
1577 if (get_user(c, s++))
1578 return -EFAULT;
1579 if (!isspace(c))
1580 break;
1581 left--;
1582 }
1583 }
1584 if (write && first)
1585 return -EINVAL;
1586 *lenp -= left;
1587 *ppos += *lenp;
1588 return 0;
1589 #undef TMPBUFLEN
1590 }
1591
1592 static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1593 void __user *buffer, size_t *lenp, loff_t *ppos,
1594 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1595 int write, void *data),
1596 void *data)
1597 {
1598 return __do_proc_dointvec(table->data, table, write, filp,
1599 buffer, lenp, ppos, conv, data);
1600 }
1601
1602 /**
1603 * proc_dointvec - read a vector of integers
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 up to table->maxlen/sizeof(unsigned int) integer
1612 * values from/to the user buffer, treated as an ASCII string.
1613 *
1614 * Returns 0 on success.
1615 */
1616 int proc_dointvec(ctl_table *table, int write, struct file *filp,
1617 void __user *buffer, size_t *lenp, loff_t *ppos)
1618 {
1619 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1620 NULL,NULL);
1621 }
1622
1623 #define OP_SET 0
1624 #define OP_AND 1
1625 #define OP_OR 2
1626
1627 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1628 int *valp,
1629 int write, void *data)
1630 {
1631 int op = *(int *)data;
1632 if (write) {
1633 int val = *negp ? -*lvalp : *lvalp;
1634 switch(op) {
1635 case OP_SET: *valp = val; break;
1636 case OP_AND: *valp &= val; break;
1637 case OP_OR: *valp |= val; break;
1638 }
1639 } else {
1640 int val = *valp;
1641 if (val < 0) {
1642 *negp = -1;
1643 *lvalp = (unsigned long)-val;
1644 } else {
1645 *negp = 0;
1646 *lvalp = (unsigned long)val;
1647 }
1648 }
1649 return 0;
1650 }
1651
1652 /*
1653 * init may raise the set.
1654 */
1655
1656 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1657 void __user *buffer, size_t *lenp, loff_t *ppos)
1658 {
1659 int op;
1660
1661 if (write && !capable(CAP_SYS_MODULE)) {
1662 return -EPERM;
1663 }
1664
1665 op = is_init(current) ? OP_SET : OP_AND;
1666 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1667 do_proc_dointvec_bset_conv,&op);
1668 }
1669
1670 /*
1671 * Taint values can only be increased
1672 */
1673 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
1674 void __user *buffer, size_t *lenp, loff_t *ppos)
1675 {
1676 int op;
1677
1678 if (!capable(CAP_SYS_ADMIN))
1679 return -EPERM;
1680
1681 op = OP_OR;
1682 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1683 do_proc_dointvec_bset_conv,&op);
1684 }
1685
1686 struct do_proc_dointvec_minmax_conv_param {
1687 int *min;
1688 int *max;
1689 };
1690
1691 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1692 int *valp,
1693 int write, void *data)
1694 {
1695 struct do_proc_dointvec_minmax_conv_param *param = data;
1696 if (write) {
1697 int val = *negp ? -*lvalp : *lvalp;
1698 if ((param->min && *param->min > val) ||
1699 (param->max && *param->max < val))
1700 return -EINVAL;
1701 *valp = val;
1702 } else {
1703 int val = *valp;
1704 if (val < 0) {
1705 *negp = -1;
1706 *lvalp = (unsigned long)-val;
1707 } else {
1708 *negp = 0;
1709 *lvalp = (unsigned long)val;
1710 }
1711 }
1712 return 0;
1713 }
1714
1715 /**
1716 * proc_dointvec_minmax - read a vector of integers with min/max values
1717 * @table: the sysctl table
1718 * @write: %TRUE if this is a write to the sysctl file
1719 * @filp: the file structure
1720 * @buffer: the user buffer
1721 * @lenp: the size of the user buffer
1722 * @ppos: file position
1723 *
1724 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1725 * values from/to the user buffer, treated as an ASCII string.
1726 *
1727 * This routine will ensure the values are within the range specified by
1728 * table->extra1 (min) and table->extra2 (max).
1729 *
1730 * Returns 0 on success.
1731 */
1732 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1733 void __user *buffer, size_t *lenp, loff_t *ppos)
1734 {
1735 struct do_proc_dointvec_minmax_conv_param param = {
1736 .min = (int *) table->extra1,
1737 .max = (int *) table->extra2,
1738 };
1739 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1740 do_proc_dointvec_minmax_conv, &param);
1741 }
1742
1743 static int __do_proc_doulongvec_minmax(void *data, ctl_table *table, int write,
1744 struct file *filp,
1745 void __user *buffer,
1746 size_t *lenp, loff_t *ppos,
1747 unsigned long convmul,
1748 unsigned long convdiv)
1749 {
1750 #define TMPBUFLEN 21
1751 unsigned long *i, *min, *max, val;
1752 int vleft, first=1, neg;
1753 size_t len, left;
1754 char buf[TMPBUFLEN], *p;
1755 char __user *s = buffer;
1756
1757 if (!data || !table->maxlen || !*lenp ||
1758 (*ppos && !write)) {
1759 *lenp = 0;
1760 return 0;
1761 }
1762
1763 i = (unsigned long *) data;
1764 min = (unsigned long *) table->extra1;
1765 max = (unsigned long *) table->extra2;
1766 vleft = table->maxlen / sizeof(unsigned long);
1767 left = *lenp;
1768
1769 for (; left && vleft--; i++, min++, max++, first=0) {
1770 if (write) {
1771 while (left) {
1772 char c;
1773 if (get_user(c, s))
1774 return -EFAULT;
1775 if (!isspace(c))
1776 break;
1777 left--;
1778 s++;
1779 }
1780 if (!left)
1781 break;
1782 neg = 0;
1783 len = left;
1784 if (len > TMPBUFLEN-1)
1785 len = TMPBUFLEN-1;
1786 if (copy_from_user(buf, s, len))
1787 return -EFAULT;
1788 buf[len] = 0;
1789 p = buf;
1790 if (*p == '-' && left > 1) {
1791 neg = 1;
1792 p++;
1793 }
1794 if (*p < '0' || *p > '9')
1795 break;
1796 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
1797 len = p-buf;
1798 if ((len < left) && *p && !isspace(*p))
1799 break;
1800 if (neg)
1801 val = -val;
1802 s += len;
1803 left -= len;
1804
1805 if(neg)
1806 continue;
1807 if ((min && val < *min) || (max && val > *max))
1808 continue;
1809 *i = val;
1810 } else {
1811 p = buf;
1812 if (!first)
1813 *p++ = '\t';
1814 sprintf(p, "%lu", convdiv * (*i) / convmul);
1815 len = strlen(buf);
1816 if (len > left)
1817 len = left;
1818 if(copy_to_user(s, buf, len))
1819 return -EFAULT;
1820 left -= len;
1821 s += len;
1822 }
1823 }
1824
1825 if (!write && !first && left) {
1826 if(put_user('\n', s))
1827 return -EFAULT;
1828 left--, s++;
1829 }
1830 if (write) {
1831 while (left) {
1832 char c;
1833 if (get_user(c, s++))
1834 return -EFAULT;
1835 if (!isspace(c))
1836 break;
1837 left--;
1838 }
1839 }
1840 if (write && first)
1841 return -EINVAL;
1842 *lenp -= left;
1843 *ppos += *lenp;
1844 return 0;
1845 #undef TMPBUFLEN
1846 }
1847
1848 static int do_proc_doulongvec_minmax(ctl_table *table, int write,
1849 struct file *filp,
1850 void __user *buffer,
1851 size_t *lenp, loff_t *ppos,
1852 unsigned long convmul,
1853 unsigned long convdiv)
1854 {
1855 return __do_proc_doulongvec_minmax(table->data, table, write,
1856 filp, buffer, lenp, ppos, convmul, convdiv);
1857 }
1858
1859 /**
1860 * proc_doulongvec_minmax - read a vector of long integers with min/max values
1861 * @table: the sysctl table
1862 * @write: %TRUE if this is a write to the sysctl file
1863 * @filp: the file structure
1864 * @buffer: the user buffer
1865 * @lenp: the size of the user buffer
1866 * @ppos: file position
1867 *
1868 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1869 * values from/to the user buffer, treated as an ASCII string.
1870 *
1871 * This routine will ensure the values are within the range specified by
1872 * table->extra1 (min) and table->extra2 (max).
1873 *
1874 * Returns 0 on success.
1875 */
1876 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
1877 void __user *buffer, size_t *lenp, loff_t *ppos)
1878 {
1879 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
1880 }
1881
1882 /**
1883 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
1884 * @table: the sysctl table
1885 * @write: %TRUE if this is a write to the sysctl file
1886 * @filp: the file structure
1887 * @buffer: the user buffer
1888 * @lenp: the size of the user buffer
1889 * @ppos: file position
1890 *
1891 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1892 * values from/to the user buffer, treated as an ASCII string. The values
1893 * are treated as milliseconds, and converted to jiffies when they are stored.
1894 *
1895 * This routine will ensure the values are within the range specified by
1896 * table->extra1 (min) and table->extra2 (max).
1897 *
1898 * Returns 0 on success.
1899 */
1900 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
1901 struct file *filp,
1902 void __user *buffer,
1903 size_t *lenp, loff_t *ppos)
1904 {
1905 return do_proc_doulongvec_minmax(table, write, filp, buffer,
1906 lenp, ppos, HZ, 1000l);
1907 }
1908
1909
1910 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
1911 int *valp,
1912 int write, void *data)
1913 {
1914 if (write) {
1915 if (*lvalp > LONG_MAX / HZ)
1916 return 1;
1917 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
1918 } else {
1919 int val = *valp;
1920 unsigned long lval;
1921 if (val < 0) {
1922 *negp = -1;
1923 lval = (unsigned long)-val;
1924 } else {
1925 *negp = 0;
1926 lval = (unsigned long)val;
1927 }
1928 *lvalp = lval / HZ;
1929 }
1930 return 0;
1931 }
1932
1933 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
1934 int *valp,
1935 int write, void *data)
1936 {
1937 if (write) {
1938 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
1939 return 1;
1940 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
1941 } else {
1942 int val = *valp;
1943 unsigned long lval;
1944 if (val < 0) {
1945 *negp = -1;
1946 lval = (unsigned long)-val;
1947 } else {
1948 *negp = 0;
1949 lval = (unsigned long)val;
1950 }
1951 *lvalp = jiffies_to_clock_t(lval);
1952 }
1953 return 0;
1954 }
1955
1956 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
1957 int *valp,
1958 int write, void *data)
1959 {
1960 if (write) {
1961 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
1962 } else {
1963 int val = *valp;
1964 unsigned long lval;
1965 if (val < 0) {
1966 *negp = -1;
1967 lval = (unsigned long)-val;
1968 } else {
1969 *negp = 0;
1970 lval = (unsigned long)val;
1971 }
1972 *lvalp = jiffies_to_msecs(lval);
1973 }
1974 return 0;
1975 }
1976
1977 /**
1978 * proc_dointvec_jiffies - read a vector of integers as seconds
1979 * @table: the sysctl table
1980 * @write: %TRUE if this is a write to the sysctl file
1981 * @filp: the file structure
1982 * @buffer: the user buffer
1983 * @lenp: the size of the user buffer
1984 * @ppos: file position
1985 *
1986 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1987 * values from/to the user buffer, treated as an ASCII string.
1988 * The values read are assumed to be in seconds, and are converted into
1989 * jiffies.
1990 *
1991 * Returns 0 on success.
1992 */
1993 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
1994 void __user *buffer, size_t *lenp, loff_t *ppos)
1995 {
1996 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1997 do_proc_dointvec_jiffies_conv,NULL);
1998 }
1999
2000 /**
2001 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2002 * @table: the sysctl table
2003 * @write: %TRUE if this is a write to the sysctl file
2004 * @filp: the file structure
2005 * @buffer: the user buffer
2006 * @lenp: the size of the user buffer
2007 * @ppos: pointer to the file position
2008 *
2009 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2010 * values from/to the user buffer, treated as an ASCII string.
2011 * The values read are assumed to be in 1/USER_HZ seconds, and
2012 * are converted into jiffies.
2013 *
2014 * Returns 0 on success.
2015 */
2016 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2017 void __user *buffer, size_t *lenp, loff_t *ppos)
2018 {
2019 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2020 do_proc_dointvec_userhz_jiffies_conv,NULL);
2021 }
2022
2023 /**
2024 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2025 * @table: the sysctl table
2026 * @write: %TRUE if this is a write to the sysctl file
2027 * @filp: the file structure
2028 * @buffer: the user buffer
2029 * @lenp: the size of the user buffer
2030 * @ppos: file position
2031 * @ppos: the current position in the file
2032 *
2033 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2034 * values from/to the user buffer, treated as an ASCII string.
2035 * The values read are assumed to be in 1/1000 seconds, and
2036 * are converted into jiffies.
2037 *
2038 * Returns 0 on success.
2039 */
2040 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2041 void __user *buffer, size_t *lenp, loff_t *ppos)
2042 {
2043 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2044 do_proc_dointvec_ms_jiffies_conv, NULL);
2045 }
2046
2047 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
2048 void __user *buffer, size_t *lenp, loff_t *ppos)
2049 {
2050 struct pid *new_pid;
2051 pid_t tmp;
2052 int r;
2053
2054 tmp = pid_nr(cad_pid);
2055
2056 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2057 lenp, ppos, NULL, NULL);
2058 if (r || !write)
2059 return r;
2060
2061 new_pid = find_get_pid(tmp);
2062 if (!new_pid)
2063 return -ESRCH;
2064
2065 put_pid(xchg(&cad_pid, new_pid));
2066 return 0;
2067 }
2068
2069 #else /* CONFIG_PROC_FS */
2070
2071 int proc_dostring(ctl_table *table, int write, struct file *filp,
2072 void __user *buffer, size_t *lenp, loff_t *ppos)
2073 {
2074 return -ENOSYS;
2075 }
2076
2077 int proc_dointvec(ctl_table *table, int write, struct file *filp,
2078 void __user *buffer, size_t *lenp, loff_t *ppos)
2079 {
2080 return -ENOSYS;
2081 }
2082
2083 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2084 void __user *buffer, size_t *lenp, loff_t *ppos)
2085 {
2086 return -ENOSYS;
2087 }
2088
2089 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2090 void __user *buffer, size_t *lenp, loff_t *ppos)
2091 {
2092 return -ENOSYS;
2093 }
2094
2095 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2096 void __user *buffer, size_t *lenp, loff_t *ppos)
2097 {
2098 return -ENOSYS;
2099 }
2100
2101 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2102 void __user *buffer, size_t *lenp, loff_t *ppos)
2103 {
2104 return -ENOSYS;
2105 }
2106
2107 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2108 void __user *buffer, size_t *lenp, loff_t *ppos)
2109 {
2110 return -ENOSYS;
2111 }
2112
2113 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2114 void __user *buffer, size_t *lenp, loff_t *ppos)
2115 {
2116 return -ENOSYS;
2117 }
2118
2119 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2120 struct file *filp,
2121 void __user *buffer,
2122 size_t *lenp, loff_t *ppos)
2123 {
2124 return -ENOSYS;
2125 }
2126
2127
2128 #endif /* CONFIG_PROC_FS */
2129
2130
2131 #ifdef CONFIG_SYSCTL_SYSCALL
2132 /*
2133 * General sysctl support routines
2134 */
2135
2136 /* The generic string strategy routine: */
2137 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2138 void __user *oldval, size_t __user *oldlenp,
2139 void __user *newval, size_t newlen)
2140 {
2141 if (!table->data || !table->maxlen)
2142 return -ENOTDIR;
2143
2144 if (oldval && oldlenp) {
2145 size_t bufsize;
2146 if (get_user(bufsize, oldlenp))
2147 return -EFAULT;
2148 if (bufsize) {
2149 size_t len = strlen(table->data), copied;
2150
2151 /* This shouldn't trigger for a well-formed sysctl */
2152 if (len > table->maxlen)
2153 len = table->maxlen;
2154
2155 /* Copy up to a max of bufsize-1 bytes of the string */
2156 copied = (len >= bufsize) ? bufsize - 1 : len;
2157
2158 if (copy_to_user(oldval, table->data, copied) ||
2159 put_user(0, (char __user *)(oldval + copied)))
2160 return -EFAULT;
2161 if (put_user(len, oldlenp))
2162 return -EFAULT;
2163 }
2164 }
2165 if (newval && newlen) {
2166 size_t len = newlen;
2167 if (len > table->maxlen)
2168 len = table->maxlen;
2169 if(copy_from_user(table->data, newval, len))
2170 return -EFAULT;
2171 if (len == table->maxlen)
2172 len--;
2173 ((char *) table->data)[len] = 0;
2174 }
2175 return 1;
2176 }
2177
2178 /*
2179 * This function makes sure that all of the integers in the vector
2180 * are between the minimum and maximum values given in the arrays
2181 * table->extra1 and table->extra2, respectively.
2182 */
2183 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2184 void __user *oldval, size_t __user *oldlenp,
2185 void __user *newval, size_t newlen)
2186 {
2187
2188 if (newval && newlen) {
2189 int __user *vec = (int __user *) newval;
2190 int *min = (int *) table->extra1;
2191 int *max = (int *) table->extra2;
2192 size_t length;
2193 int i;
2194
2195 if (newlen % sizeof(int) != 0)
2196 return -EINVAL;
2197
2198 if (!table->extra1 && !table->extra2)
2199 return 0;
2200
2201 if (newlen > table->maxlen)
2202 newlen = table->maxlen;
2203 length = newlen / sizeof(int);
2204
2205 for (i = 0; i < length; i++) {
2206 int value;
2207 if (get_user(value, vec + i))
2208 return -EFAULT;
2209 if (min && value < min[i])
2210 return -EINVAL;
2211 if (max && value > max[i])
2212 return -EINVAL;
2213 }
2214 }
2215 return 0;
2216 }
2217
2218 /* Strategy function to convert jiffies to seconds */
2219 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2220 void __user *oldval, size_t __user *oldlenp,
2221 void __user *newval, size_t newlen)
2222 {
2223 if (oldval && oldlenp) {
2224 size_t olen;
2225
2226 if (get_user(olen, oldlenp))
2227 return -EFAULT;
2228 if (olen) {
2229 int val;
2230
2231 if (olen < sizeof(int))
2232 return -EINVAL;
2233
2234 val = *(int *)(table->data) / HZ;
2235 if (put_user(val, (int __user *)oldval))
2236 return -EFAULT;
2237 if (put_user(sizeof(int), oldlenp))
2238 return -EFAULT;
2239 }
2240 }
2241 if (newval && newlen) {
2242 int new;
2243 if (newlen != sizeof(int))
2244 return -EINVAL;
2245 if (get_user(new, (int __user *)newval))
2246 return -EFAULT;
2247 *(int *)(table->data) = new*HZ;
2248 }
2249 return 1;
2250 }
2251
2252 /* Strategy function to convert jiffies to seconds */
2253 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2254 void __user *oldval, size_t __user *oldlenp,
2255 void __user *newval, size_t newlen)
2256 {
2257 if (oldval && oldlenp) {
2258 size_t olen;
2259
2260 if (get_user(olen, oldlenp))
2261 return -EFAULT;
2262 if (olen) {
2263 int val;
2264
2265 if (olen < sizeof(int))
2266 return -EINVAL;
2267
2268 val = jiffies_to_msecs(*(int *)(table->data));
2269 if (put_user(val, (int __user *)oldval))
2270 return -EFAULT;
2271 if (put_user(sizeof(int), oldlenp))
2272 return -EFAULT;
2273 }
2274 }
2275 if (newval && newlen) {
2276 int new;
2277 if (newlen != sizeof(int))
2278 return -EINVAL;
2279 if (get_user(new, (int __user *)newval))
2280 return -EFAULT;
2281 *(int *)(table->data) = msecs_to_jiffies(new);
2282 }
2283 return 1;
2284 }
2285
2286
2287
2288 #else /* CONFIG_SYSCTL_SYSCALL */
2289
2290
2291 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2292 {
2293 static int msg_count;
2294 struct __sysctl_args tmp;
2295 int name[CTL_MAXNAME];
2296 int i;
2297
2298 /* Read in the sysctl name for better debug message logging */
2299 if (copy_from_user(&tmp, args, sizeof(tmp)))
2300 return -EFAULT;
2301 if (tmp.nlen <= 0 || tmp.nlen >= CTL_MAXNAME)
2302 return -ENOTDIR;
2303 for (i = 0; i < tmp.nlen; i++)
2304 if (get_user(name[i], tmp.name + i))
2305 return -EFAULT;
2306
2307 /* Ignore accesses to kernel.version */
2308 if ((tmp.nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2309 goto out;
2310
2311 if (msg_count < 5) {
2312 msg_count++;
2313 printk(KERN_INFO
2314 "warning: process `%s' used the removed sysctl "
2315 "system call with ", current->comm);
2316 for (i = 0; i < tmp.nlen; i++)
2317 printk("%d.", name[i]);
2318 printk("\n");
2319 }
2320 out:
2321 return -ENOSYS;
2322 }
2323
2324 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2325 void __user *oldval, size_t __user *oldlenp,
2326 void __user *newval, size_t newlen)
2327 {
2328 return -ENOSYS;
2329 }
2330
2331 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2332 void __user *oldval, size_t __user *oldlenp,
2333 void __user *newval, size_t newlen)
2334 {
2335 return -ENOSYS;
2336 }
2337
2338 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2339 void __user *oldval, size_t __user *oldlenp,
2340 void __user *newval, size_t newlen)
2341 {
2342 return -ENOSYS;
2343 }
2344
2345 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2346 void __user *oldval, size_t __user *oldlenp,
2347 void __user *newval, size_t newlen)
2348 {
2349 return -ENOSYS;
2350 }
2351
2352 #endif /* CONFIG_SYSCTL_SYSCALL */
2353
2354 /*
2355 * No sense putting this after each symbol definition, twice,
2356 * exception granted :-)
2357 */
2358 EXPORT_SYMBOL(proc_dointvec);
2359 EXPORT_SYMBOL(proc_dointvec_jiffies);
2360 EXPORT_SYMBOL(proc_dointvec_minmax);
2361 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2362 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2363 EXPORT_SYMBOL(proc_dostring);
2364 EXPORT_SYMBOL(proc_doulongvec_minmax);
2365 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2366 EXPORT_SYMBOL(register_sysctl_table);
2367 EXPORT_SYMBOL(sysctl_intvec);
2368 EXPORT_SYMBOL(sysctl_jiffies);
2369 EXPORT_SYMBOL(sysctl_ms_jiffies);
2370 EXPORT_SYMBOL(sysctl_string);
2371 EXPORT_SYMBOL(unregister_sysctl_table);
linux v2.6.22.y-op