projects / deliverable / linux.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Use write_trylock_irqsave in ptrace_attach
[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 extern int maps_protect;
80 extern int sysctl_stat_interval;
81
82 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
83 static int maxolduid = 65535;
84 static int minolduid;
85 static int min_percpu_pagelist_fract = 8;
86
87 static int ngroups_max = NGROUPS_MAX;
88
89 #ifdef CONFIG_KMOD
90 extern char modprobe_path[];
91 #endif
92 #ifdef CONFIG_CHR_DEV_SG
93 extern int sg_big_buff;
94 #endif
95
96 #ifdef __sparc__
97 extern char reboot_command [];
98 extern int stop_a_enabled;
99 extern int scons_pwroff;
100 #endif
101
102 #ifdef __hppa__
103 extern int pwrsw_enabled;
104 extern int unaligned_enabled;
105 #endif
106
107 #ifdef CONFIG_S390
108 #ifdef CONFIG_MATHEMU
109 extern int sysctl_ieee_emulation_warnings;
110 #endif
111 extern int sysctl_userprocess_debug;
112 extern int spin_retry;
113 #endif
114
115 extern int sysctl_hz_timer;
116
117 #ifdef CONFIG_BSD_PROCESS_ACCT
118 extern int acct_parm[];
119 #endif
120
121 #ifdef CONFIG_IA64
122 extern int no_unaligned_warning;
123 #endif
124
125 #ifdef CONFIG_RT_MUTEXES
126 extern int max_lock_depth;
127 #endif
128
129 #ifdef CONFIG_SYSCTL_SYSCALL
130 static int parse_table(int __user *, int, void __user *, size_t __user *,
131 void __user *, size_t, ctl_table *);
132 #endif
133
134
135 #ifdef CONFIG_PROC_SYSCTL
136 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
137 void __user *buffer, size_t *lenp, loff_t *ppos);
138 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
139 void __user *buffer, size_t *lenp, loff_t *ppos);
140 #endif
141
142 static ctl_table root_table[];
143 static struct ctl_table_header root_table_header =
144 { root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
145
146 static ctl_table kern_table[];
147 static ctl_table vm_table[];
148 static ctl_table fs_table[];
149 static ctl_table debug_table[];
150 static ctl_table dev_table[];
151 extern ctl_table random_table[];
152 #ifdef CONFIG_UNIX98_PTYS
153 extern ctl_table pty_table[];
154 #endif
155 #ifdef CONFIG_INOTIFY_USER
156 extern ctl_table inotify_table[];
157 #endif
158
159 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
160 int sysctl_legacy_va_layout;
161 #endif
162
163
164 /* The default sysctl tables: */
165
166 static ctl_table root_table[] = {
167 {
168 .ctl_name = CTL_KERN,
169 .procname = "kernel",
170 .mode = 0555,
171 .child = kern_table,
172 },
173 {
174 .ctl_name = CTL_VM,
175 .procname = "vm",
176 .mode = 0555,
177 .child = vm_table,
178 },
179 #ifdef CONFIG_NET
180 {
181 .ctl_name = CTL_NET,
182 .procname = "net",
183 .mode = 0555,
184 .child = net_table,
185 },
186 #endif
187 {
188 .ctl_name = CTL_FS,
189 .procname = "fs",
190 .mode = 0555,
191 .child = fs_table,
192 },
193 {
194 .ctl_name = CTL_DEBUG,
195 .procname = "debug",
196 .mode = 0555,
197 .child = debug_table,
198 },
199 {
200 .ctl_name = CTL_DEV,
201 .procname = "dev",
202 .mode = 0555,
203 .child = dev_table,
204 },
205
206 { .ctl_name = 0 }
207 };
208
209 #ifdef CONFIG_SCHED_DEBUG
210 static unsigned long min_sched_granularity_ns = 100000; /* 100 usecs */
211 static unsigned long max_sched_granularity_ns = 1000000000; /* 1 second */
212 static unsigned long min_wakeup_granularity_ns; /* 0 usecs */
213 static unsigned long max_wakeup_granularity_ns = 1000000000; /* 1 second */
214 #endif
215
216 static ctl_table kern_table[] = {
217 #ifdef CONFIG_SCHED_DEBUG
218 {
219 .ctl_name = CTL_UNNUMBERED,
220 .procname = "sched_granularity_ns",
221 .data = &sysctl_sched_granularity,
222 .maxlen = sizeof(unsigned int),
223 .mode = 0644,
224 .proc_handler = &proc_dointvec_minmax,
225 .strategy = &sysctl_intvec,
226 .extra1 = &min_sched_granularity_ns,
227 .extra2 = &max_sched_granularity_ns,
228 },
229 {
230 .ctl_name = CTL_UNNUMBERED,
231 .procname = "sched_wakeup_granularity_ns",
232 .data = &sysctl_sched_wakeup_granularity,
233 .maxlen = sizeof(unsigned int),
234 .mode = 0644,
235 .proc_handler = &proc_dointvec_minmax,
236 .strategy = &sysctl_intvec,
237 .extra1 = &min_wakeup_granularity_ns,
238 .extra2 = &max_wakeup_granularity_ns,
239 },
240 {
241 .ctl_name = CTL_UNNUMBERED,
242 .procname = "sched_batch_wakeup_granularity_ns",
243 .data = &sysctl_sched_batch_wakeup_granularity,
244 .maxlen = sizeof(unsigned int),
245 .mode = 0644,
246 .proc_handler = &proc_dointvec_minmax,
247 .strategy = &sysctl_intvec,
248 .extra1 = &min_wakeup_granularity_ns,
249 .extra2 = &max_wakeup_granularity_ns,
250 },
251 {
252 .ctl_name = CTL_UNNUMBERED,
253 .procname = "sched_stat_granularity_ns",
254 .data = &sysctl_sched_stat_granularity,
255 .maxlen = sizeof(unsigned int),
256 .mode = 0644,
257 .proc_handler = &proc_dointvec_minmax,
258 .strategy = &sysctl_intvec,
259 .extra1 = &min_wakeup_granularity_ns,
260 .extra2 = &max_wakeup_granularity_ns,
261 },
262 {
263 .ctl_name = CTL_UNNUMBERED,
264 .procname = "sched_runtime_limit_ns",
265 .data = &sysctl_sched_runtime_limit,
266 .maxlen = sizeof(unsigned int),
267 .mode = 0644,
268 .proc_handler = &proc_dointvec_minmax,
269 .strategy = &sysctl_intvec,
270 .extra1 = &min_sched_granularity_ns,
271 .extra2 = &max_sched_granularity_ns,
272 },
273 {
274 .ctl_name = CTL_UNNUMBERED,
275 .procname = "sched_child_runs_first",
276 .data = &sysctl_sched_child_runs_first,
277 .maxlen = sizeof(unsigned int),
278 .mode = 0644,
279 .proc_handler = &proc_dointvec,
280 },
281 {
282 .ctl_name = CTL_UNNUMBERED,
283 .procname = "sched_features",
284 .data = &sysctl_sched_features,
285 .maxlen = sizeof(unsigned int),
286 .mode = 0644,
287 .proc_handler = &proc_dointvec,
288 },
289 #endif
290 {
291 .ctl_name = KERN_PANIC,
292 .procname = "panic",
293 .data = &panic_timeout,
294 .maxlen = sizeof(int),
295 .mode = 0644,
296 .proc_handler = &proc_dointvec,
297 },
298 {
299 .ctl_name = KERN_CORE_USES_PID,
300 .procname = "core_uses_pid",
301 .data = &core_uses_pid,
302 .maxlen = sizeof(int),
303 .mode = 0644,
304 .proc_handler = &proc_dointvec,
305 },
306 {
307 .ctl_name = KERN_CORE_PATTERN,
308 .procname = "core_pattern",
309 .data = core_pattern,
310 .maxlen = CORENAME_MAX_SIZE,
311 .mode = 0644,
312 .proc_handler = &proc_dostring,
313 .strategy = &sysctl_string,
314 },
315 #ifdef CONFIG_PROC_SYSCTL
316 {
317 .ctl_name = KERN_TAINTED,
318 .procname = "tainted",
319 .data = &tainted,
320 .maxlen = sizeof(int),
321 .mode = 0644,
322 .proc_handler = &proc_dointvec_taint,
323 },
324 #endif
325 {
326 .ctl_name = KERN_CAP_BSET,
327 .procname = "cap-bound",
328 .data = &cap_bset,
329 .maxlen = sizeof(kernel_cap_t),
330 .mode = 0600,
331 .proc_handler = &proc_dointvec_bset,
332 },
333 #ifdef CONFIG_BLK_DEV_INITRD
334 {
335 .ctl_name = KERN_REALROOTDEV,
336 .procname = "real-root-dev",
337 .data = &real_root_dev,
338 .maxlen = sizeof(int),
339 .mode = 0644,
340 .proc_handler = &proc_dointvec,
341 },
342 #endif
343 #ifdef __sparc__
344 {
345 .ctl_name = KERN_SPARC_REBOOT,
346 .procname = "reboot-cmd",
347 .data = reboot_command,
348 .maxlen = 256,
349 .mode = 0644,
350 .proc_handler = &proc_dostring,
351 .strategy = &sysctl_string,
352 },
353 {
354 .ctl_name = KERN_SPARC_STOP_A,
355 .procname = "stop-a",
356 .data = &stop_a_enabled,
357 .maxlen = sizeof (int),
358 .mode = 0644,
359 .proc_handler = &proc_dointvec,
360 },
361 {
362 .ctl_name = KERN_SPARC_SCONS_PWROFF,
363 .procname = "scons-poweroff",
364 .data = &scons_pwroff,
365 .maxlen = sizeof (int),
366 .mode = 0644,
367 .proc_handler = &proc_dointvec,
368 },
369 #endif
370 #ifdef __hppa__
371 {
372 .ctl_name = KERN_HPPA_PWRSW,
373 .procname = "soft-power",
374 .data = &pwrsw_enabled,
375 .maxlen = sizeof (int),
376 .mode = 0644,
377 .proc_handler = &proc_dointvec,
378 },
379 {
380 .ctl_name = KERN_HPPA_UNALIGNED,
381 .procname = "unaligned-trap",
382 .data = &unaligned_enabled,
383 .maxlen = sizeof (int),
384 .mode = 0644,
385 .proc_handler = &proc_dointvec,
386 },
387 #endif
388 {
389 .ctl_name = KERN_CTLALTDEL,
390 .procname = "ctrl-alt-del",
391 .data = &C_A_D,
392 .maxlen = sizeof(int),
393 .mode = 0644,
394 .proc_handler = &proc_dointvec,
395 },
396 {
397 .ctl_name = KERN_PRINTK,
398 .procname = "printk",
399 .data = &console_loglevel,
400 .maxlen = 4*sizeof(int),
401 .mode = 0644,
402 .proc_handler = &proc_dointvec,
403 },
404 #ifdef CONFIG_KMOD
405 {
406 .ctl_name = KERN_MODPROBE,
407 .procname = "modprobe",
408 .data = &modprobe_path,
409 .maxlen = KMOD_PATH_LEN,
410 .mode = 0644,
411 .proc_handler = &proc_dostring,
412 .strategy = &sysctl_string,
413 },
414 #endif
415 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
416 {
417 .ctl_name = KERN_HOTPLUG,
418 .procname = "hotplug",
419 .data = &uevent_helper,
420 .maxlen = UEVENT_HELPER_PATH_LEN,
421 .mode = 0644,
422 .proc_handler = &proc_dostring,
423 .strategy = &sysctl_string,
424 },
425 #endif
426 #ifdef CONFIG_CHR_DEV_SG
427 {
428 .ctl_name = KERN_SG_BIG_BUFF,
429 .procname = "sg-big-buff",
430 .data = &sg_big_buff,
431 .maxlen = sizeof (int),
432 .mode = 0444,
433 .proc_handler = &proc_dointvec,
434 },
435 #endif
436 #ifdef CONFIG_BSD_PROCESS_ACCT
437 {
438 .ctl_name = KERN_ACCT,
439 .procname = "acct",
440 .data = &acct_parm,
441 .maxlen = 3*sizeof(int),
442 .mode = 0644,
443 .proc_handler = &proc_dointvec,
444 },
445 #endif
446 #ifdef CONFIG_MAGIC_SYSRQ
447 {
448 .ctl_name = KERN_SYSRQ,
449 .procname = "sysrq",
450 .data = &__sysrq_enabled,
451 .maxlen = sizeof (int),
452 .mode = 0644,
453 .proc_handler = &proc_dointvec,
454 },
455 #endif
456 #ifdef CONFIG_PROC_SYSCTL
457 {
458 .ctl_name = KERN_CADPID,
459 .procname = "cad_pid",
460 .data = NULL,
461 .maxlen = sizeof (int),
462 .mode = 0600,
463 .proc_handler = &proc_do_cad_pid,
464 },
465 #endif
466 {
467 .ctl_name = KERN_MAX_THREADS,
468 .procname = "threads-max",
469 .data = &max_threads,
470 .maxlen = sizeof(int),
471 .mode = 0644,
472 .proc_handler = &proc_dointvec,
473 },
474 {
475 .ctl_name = KERN_RANDOM,
476 .procname = "random",
477 .mode = 0555,
478 .child = random_table,
479 },
480 #ifdef CONFIG_UNIX98_PTYS
481 {
482 .ctl_name = KERN_PTY,
483 .procname = "pty",
484 .mode = 0555,
485 .child = pty_table,
486 },
487 #endif
488 {
489 .ctl_name = KERN_OVERFLOWUID,
490 .procname = "overflowuid",
491 .data = &overflowuid,
492 .maxlen = sizeof(int),
493 .mode = 0644,
494 .proc_handler = &proc_dointvec_minmax,
495 .strategy = &sysctl_intvec,
496 .extra1 = &minolduid,
497 .extra2 = &maxolduid,
498 },
499 {
500 .ctl_name = KERN_OVERFLOWGID,
501 .procname = "overflowgid",
502 .data = &overflowgid,
503 .maxlen = sizeof(int),
504 .mode = 0644,
505 .proc_handler = &proc_dointvec_minmax,
506 .strategy = &sysctl_intvec,
507 .extra1 = &minolduid,
508 .extra2 = &maxolduid,
509 },
510 #ifdef CONFIG_S390
511 #ifdef CONFIG_MATHEMU
512 {
513 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
514 .procname = "ieee_emulation_warnings",
515 .data = &sysctl_ieee_emulation_warnings,
516 .maxlen = sizeof(int),
517 .mode = 0644,
518 .proc_handler = &proc_dointvec,
519 },
520 #endif
521 #ifdef CONFIG_NO_IDLE_HZ
522 {
523 .ctl_name = KERN_HZ_TIMER,
524 .procname = "hz_timer",
525 .data = &sysctl_hz_timer,
526 .maxlen = sizeof(int),
527 .mode = 0644,
528 .proc_handler = &proc_dointvec,
529 },
530 #endif
531 {
532 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
533 .procname = "userprocess_debug",
534 .data = &sysctl_userprocess_debug,
535 .maxlen = sizeof(int),
536 .mode = 0644,
537 .proc_handler = &proc_dointvec,
538 },
539 #endif
540 {
541 .ctl_name = KERN_PIDMAX,
542 .procname = "pid_max",
543 .data = &pid_max,
544 .maxlen = sizeof (int),
545 .mode = 0644,
546 .proc_handler = &proc_dointvec_minmax,
547 .strategy = sysctl_intvec,
548 .extra1 = &pid_max_min,
549 .extra2 = &pid_max_max,
550 },
551 {
552 .ctl_name = KERN_PANIC_ON_OOPS,
553 .procname = "panic_on_oops",
554 .data = &panic_on_oops,
555 .maxlen = sizeof(int),
556 .mode = 0644,
557 .proc_handler = &proc_dointvec,
558 },
559 {
560 .ctl_name = KERN_PRINTK_RATELIMIT,
561 .procname = "printk_ratelimit",
562 .data = &printk_ratelimit_jiffies,
563 .maxlen = sizeof(int),
564 .mode = 0644,
565 .proc_handler = &proc_dointvec_jiffies,
566 .strategy = &sysctl_jiffies,
567 },
568 {
569 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
570 .procname = "printk_ratelimit_burst",
571 .data = &printk_ratelimit_burst,
572 .maxlen = sizeof(int),
573 .mode = 0644,
574 .proc_handler = &proc_dointvec,
575 },
576 {
577 .ctl_name = KERN_NGROUPS_MAX,
578 .procname = "ngroups_max",
579 .data = &ngroups_max,
580 .maxlen = sizeof (int),
581 .mode = 0444,
582 .proc_handler = &proc_dointvec,
583 },
584 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
585 {
586 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
587 .procname = "unknown_nmi_panic",
588 .data = &unknown_nmi_panic,
589 .maxlen = sizeof (int),
590 .mode = 0644,
591 .proc_handler = &proc_dointvec,
592 },
593 {
594 .ctl_name = KERN_NMI_WATCHDOG,
595 .procname = "nmi_watchdog",
596 .data = &nmi_watchdog_enabled,
597 .maxlen = sizeof (int),
598 .mode = 0644,
599 .proc_handler = &proc_nmi_enabled,
600 },
601 #endif
602 #if defined(CONFIG_X86)
603 {
604 .ctl_name = KERN_PANIC_ON_NMI,
605 .procname = "panic_on_unrecovered_nmi",
606 .data = &panic_on_unrecovered_nmi,
607 .maxlen = sizeof(int),
608 .mode = 0644,
609 .proc_handler = &proc_dointvec,
610 },
611 {
612 .ctl_name = KERN_BOOTLOADER_TYPE,
613 .procname = "bootloader_type",
614 .data = &bootloader_type,
615 .maxlen = sizeof (int),
616 .mode = 0444,
617 .proc_handler = &proc_dointvec,
618 },
619 {
620 .ctl_name = CTL_UNNUMBERED,
621 .procname = "kstack_depth_to_print",
622 .data = &kstack_depth_to_print,
623 .maxlen = sizeof(int),
624 .mode = 0644,
625 .proc_handler = &proc_dointvec,
626 },
627 #endif
628 #if defined(CONFIG_MMU)
629 {
630 .ctl_name = KERN_RANDOMIZE,
631 .procname = "randomize_va_space",
632 .data = &randomize_va_space,
633 .maxlen = sizeof(int),
634 .mode = 0644,
635 .proc_handler = &proc_dointvec,
636 },
637 #endif
638 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
639 {
640 .ctl_name = KERN_SPIN_RETRY,
641 .procname = "spin_retry",
642 .data = &spin_retry,
643 .maxlen = sizeof (int),
644 .mode = 0644,
645 .proc_handler = &proc_dointvec,
646 },
647 #endif
648 #ifdef CONFIG_ACPI_SLEEP
649 {
650 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
651 .procname = "acpi_video_flags",
652 .data = &acpi_video_flags,
653 .maxlen = sizeof (unsigned long),
654 .mode = 0644,
655 .proc_handler = &proc_doulongvec_minmax,
656 },
657 #endif
658 #ifdef CONFIG_IA64
659 {
660 .ctl_name = KERN_IA64_UNALIGNED,
661 .procname = "ignore-unaligned-usertrap",
662 .data = &no_unaligned_warning,
663 .maxlen = sizeof (int),
664 .mode = 0644,
665 .proc_handler = &proc_dointvec,
666 },
667 #endif
668 #ifdef CONFIG_COMPAT
669 {
670 .ctl_name = KERN_COMPAT_LOG,
671 .procname = "compat-log",
672 .data = &compat_log,
673 .maxlen = sizeof (int),
674 .mode = 0644,
675 .proc_handler = &proc_dointvec,
676 },
677 #endif
678 #ifdef CONFIG_RT_MUTEXES
679 {
680 .ctl_name = KERN_MAX_LOCK_DEPTH,
681 .procname = "max_lock_depth",
682 .data = &max_lock_depth,
683 .maxlen = sizeof(int),
684 .mode = 0644,
685 .proc_handler = &proc_dointvec,
686 },
687 #endif
688 #ifdef CONFIG_PROC_FS
689 {
690 .ctl_name = CTL_UNNUMBERED,
691 .procname = "maps_protect",
692 .data = &maps_protect,
693 .maxlen = sizeof(int),
694 .mode = 0644,
695 .proc_handler = &proc_dointvec,
696 },
697 #endif
698
699 { .ctl_name = 0 }
700 };
701
702 /* Constants for minimum and maximum testing in vm_table.
703 We use these as one-element integer vectors. */
704 static int zero;
705 static int one_hundred = 100;
706
707
708 static ctl_table vm_table[] = {
709 {
710 .ctl_name = VM_OVERCOMMIT_MEMORY,
711 .procname = "overcommit_memory",
712 .data = &sysctl_overcommit_memory,
713 .maxlen = sizeof(sysctl_overcommit_memory),
714 .mode = 0644,
715 .proc_handler = &proc_dointvec,
716 },
717 {
718 .ctl_name = VM_PANIC_ON_OOM,
719 .procname = "panic_on_oom",
720 .data = &sysctl_panic_on_oom,
721 .maxlen = sizeof(sysctl_panic_on_oom),
722 .mode = 0644,
723 .proc_handler = &proc_dointvec,
724 },
725 {
726 .ctl_name = VM_OVERCOMMIT_RATIO,
727 .procname = "overcommit_ratio",
728 .data = &sysctl_overcommit_ratio,
729 .maxlen = sizeof(sysctl_overcommit_ratio),
730 .mode = 0644,
731 .proc_handler = &proc_dointvec,
732 },
733 {
734 .ctl_name = VM_PAGE_CLUSTER,
735 .procname = "page-cluster",
736 .data = &page_cluster,
737 .maxlen = sizeof(int),
738 .mode = 0644,
739 .proc_handler = &proc_dointvec,
740 },
741 {
742 .ctl_name = VM_DIRTY_BACKGROUND,
743 .procname = "dirty_background_ratio",
744 .data = &dirty_background_ratio,
745 .maxlen = sizeof(dirty_background_ratio),
746 .mode = 0644,
747 .proc_handler = &proc_dointvec_minmax,
748 .strategy = &sysctl_intvec,
749 .extra1 = &zero,
750 .extra2 = &one_hundred,
751 },
752 {
753 .ctl_name = VM_DIRTY_RATIO,
754 .procname = "dirty_ratio",
755 .data = &vm_dirty_ratio,
756 .maxlen = sizeof(vm_dirty_ratio),
757 .mode = 0644,
758 .proc_handler = &proc_dointvec_minmax,
759 .strategy = &sysctl_intvec,
760 .extra1 = &zero,
761 .extra2 = &one_hundred,
762 },
763 {
764 .ctl_name = VM_DIRTY_WB_CS,
765 .procname = "dirty_writeback_centisecs",
766 .data = &dirty_writeback_interval,
767 .maxlen = sizeof(dirty_writeback_interval),
768 .mode = 0644,
769 .proc_handler = &dirty_writeback_centisecs_handler,
770 },
771 {
772 .ctl_name = VM_DIRTY_EXPIRE_CS,
773 .procname = "dirty_expire_centisecs",
774 .data = &dirty_expire_interval,
775 .maxlen = sizeof(dirty_expire_interval),
776 .mode = 0644,
777 .proc_handler = &proc_dointvec_userhz_jiffies,
778 },
779 {
780 .ctl_name = VM_NR_PDFLUSH_THREADS,
781 .procname = "nr_pdflush_threads",
782 .data = &nr_pdflush_threads,
783 .maxlen = sizeof nr_pdflush_threads,
784 .mode = 0444 /* read-only*/,
785 .proc_handler = &proc_dointvec,
786 },
787 {
788 .ctl_name = VM_SWAPPINESS,
789 .procname = "swappiness",
790 .data = &vm_swappiness,
791 .maxlen = sizeof(vm_swappiness),
792 .mode = 0644,
793 .proc_handler = &proc_dointvec_minmax,
794 .strategy = &sysctl_intvec,
795 .extra1 = &zero,
796 .extra2 = &one_hundred,
797 },
798 #ifdef CONFIG_HUGETLB_PAGE
799 {
800 .ctl_name = VM_HUGETLB_PAGES,
801 .procname = "nr_hugepages",
802 .data = &max_huge_pages,
803 .maxlen = sizeof(unsigned long),
804 .mode = 0644,
805 .proc_handler = &hugetlb_sysctl_handler,
806 .extra1 = (void *)&hugetlb_zero,
807 .extra2 = (void *)&hugetlb_infinity,
808 },
809 {
810 .ctl_name = VM_HUGETLB_GROUP,
811 .procname = "hugetlb_shm_group",
812 .data = &sysctl_hugetlb_shm_group,
813 .maxlen = sizeof(gid_t),
814 .mode = 0644,
815 .proc_handler = &proc_dointvec,
816 },
817 #endif
818 {
819 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
820 .procname = "lowmem_reserve_ratio",
821 .data = &sysctl_lowmem_reserve_ratio,
822 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
823 .mode = 0644,
824 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
825 .strategy = &sysctl_intvec,
826 },
827 {
828 .ctl_name = VM_DROP_PAGECACHE,
829 .procname = "drop_caches",
830 .data = &sysctl_drop_caches,
831 .maxlen = sizeof(int),
832 .mode = 0644,
833 .proc_handler = drop_caches_sysctl_handler,
834 .strategy = &sysctl_intvec,
835 },
836 {
837 .ctl_name = VM_MIN_FREE_KBYTES,
838 .procname = "min_free_kbytes",
839 .data = &min_free_kbytes,
840 .maxlen = sizeof(min_free_kbytes),
841 .mode = 0644,
842 .proc_handler = &min_free_kbytes_sysctl_handler,
843 .strategy = &sysctl_intvec,
844 .extra1 = &zero,
845 },
846 {
847 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
848 .procname = "percpu_pagelist_fraction",
849 .data = &percpu_pagelist_fraction,
850 .maxlen = sizeof(percpu_pagelist_fraction),
851 .mode = 0644,
852 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
853 .strategy = &sysctl_intvec,
854 .extra1 = &min_percpu_pagelist_fract,
855 },
856 #ifdef CONFIG_MMU
857 {
858 .ctl_name = VM_MAX_MAP_COUNT,
859 .procname = "max_map_count",
860 .data = &sysctl_max_map_count,
861 .maxlen = sizeof(sysctl_max_map_count),
862 .mode = 0644,
863 .proc_handler = &proc_dointvec
864 },
865 #endif
866 {
867 .ctl_name = VM_LAPTOP_MODE,
868 .procname = "laptop_mode",
869 .data = &laptop_mode,
870 .maxlen = sizeof(laptop_mode),
871 .mode = 0644,
872 .proc_handler = &proc_dointvec_jiffies,
873 .strategy = &sysctl_jiffies,
874 },
875 {
876 .ctl_name = VM_BLOCK_DUMP,
877 .procname = "block_dump",
878 .data = &block_dump,
879 .maxlen = sizeof(block_dump),
880 .mode = 0644,
881 .proc_handler = &proc_dointvec,
882 .strategy = &sysctl_intvec,
883 .extra1 = &zero,
884 },
885 {
886 .ctl_name = VM_VFS_CACHE_PRESSURE,
887 .procname = "vfs_cache_pressure",
888 .data = &sysctl_vfs_cache_pressure,
889 .maxlen = sizeof(sysctl_vfs_cache_pressure),
890 .mode = 0644,
891 .proc_handler = &proc_dointvec,
892 .strategy = &sysctl_intvec,
893 .extra1 = &zero,
894 },
895 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
896 {
897 .ctl_name = VM_LEGACY_VA_LAYOUT,
898 .procname = "legacy_va_layout",
899 .data = &sysctl_legacy_va_layout,
900 .maxlen = sizeof(sysctl_legacy_va_layout),
901 .mode = 0644,
902 .proc_handler = &proc_dointvec,
903 .strategy = &sysctl_intvec,
904 .extra1 = &zero,
905 },
906 #endif
907 #ifdef CONFIG_NUMA
908 {
909 .ctl_name = VM_ZONE_RECLAIM_MODE,
910 .procname = "zone_reclaim_mode",
911 .data = &zone_reclaim_mode,
912 .maxlen = sizeof(zone_reclaim_mode),
913 .mode = 0644,
914 .proc_handler = &proc_dointvec,
915 .strategy = &sysctl_intvec,
916 .extra1 = &zero,
917 },
918 {
919 .ctl_name = VM_MIN_UNMAPPED,
920 .procname = "min_unmapped_ratio",
921 .data = &sysctl_min_unmapped_ratio,
922 .maxlen = sizeof(sysctl_min_unmapped_ratio),
923 .mode = 0644,
924 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
925 .strategy = &sysctl_intvec,
926 .extra1 = &zero,
927 .extra2 = &one_hundred,
928 },
929 {
930 .ctl_name = VM_MIN_SLAB,
931 .procname = "min_slab_ratio",
932 .data = &sysctl_min_slab_ratio,
933 .maxlen = sizeof(sysctl_min_slab_ratio),
934 .mode = 0644,
935 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
936 .strategy = &sysctl_intvec,
937 .extra1 = &zero,
938 .extra2 = &one_hundred,
939 },
940 #endif
941 #ifdef CONFIG_SMP
942 {
943 .ctl_name = CTL_UNNUMBERED,
944 .procname = "stat_interval",
945 .data = &sysctl_stat_interval,
946 .maxlen = sizeof(sysctl_stat_interval),
947 .mode = 0644,
948 .proc_handler = &proc_dointvec_jiffies,
949 .strategy = &sysctl_jiffies,
950 },
951 #endif
952 #ifdef CONFIG_SECURITY
953 {
954 .ctl_name = CTL_UNNUMBERED,
955 .procname = "mmap_min_addr",
956 .data = &mmap_min_addr,
957 .maxlen = sizeof(unsigned long),
958 .mode = 0644,
959 .proc_handler = &proc_doulongvec_minmax,
960 },
961 #ifdef CONFIG_NUMA
962 {
963 .ctl_name = CTL_UNNUMBERED,
964 .procname = "numa_zonelist_order",
965 .data = &numa_zonelist_order,
966 .maxlen = NUMA_ZONELIST_ORDER_LEN,
967 .mode = 0644,
968 .proc_handler = &numa_zonelist_order_handler,
969 .strategy = &sysctl_string,
970 },
971 #endif
972 #endif
973 #if defined(CONFIG_X86_32) || \
974 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
975 {
976 .ctl_name = VM_VDSO_ENABLED,
977 .procname = "vdso_enabled",
978 .data = &vdso_enabled,
979 .maxlen = sizeof(vdso_enabled),
980 .mode = 0644,
981 .proc_handler = &proc_dointvec,
982 .strategy = &sysctl_intvec,
983 .extra1 = &zero,
984 },
985 #endif
986 { .ctl_name = 0 }
987 };
988
989 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
990 static ctl_table binfmt_misc_table[] = {
991 { .ctl_name = 0 }
992 };
993 #endif
994
995 static ctl_table fs_table[] = {
996 {
997 .ctl_name = FS_NRINODE,
998 .procname = "inode-nr",
999 .data = &inodes_stat,
1000 .maxlen = 2*sizeof(int),
1001 .mode = 0444,
1002 .proc_handler = &proc_dointvec,
1003 },
1004 {
1005 .ctl_name = FS_STATINODE,
1006 .procname = "inode-state",
1007 .data = &inodes_stat,
1008 .maxlen = 7*sizeof(int),
1009 .mode = 0444,
1010 .proc_handler = &proc_dointvec,
1011 },
1012 {
1013 .ctl_name = FS_NRFILE,
1014 .procname = "file-nr",
1015 .data = &files_stat,
1016 .maxlen = 3*sizeof(int),
1017 .mode = 0444,
1018 .proc_handler = &proc_nr_files,
1019 },
1020 {
1021 .ctl_name = FS_MAXFILE,
1022 .procname = "file-max",
1023 .data = &files_stat.max_files,
1024 .maxlen = sizeof(int),
1025 .mode = 0644,
1026 .proc_handler = &proc_dointvec,
1027 },
1028 {
1029 .ctl_name = FS_DENTRY,
1030 .procname = "dentry-state",
1031 .data = &dentry_stat,
1032 .maxlen = 6*sizeof(int),
1033 .mode = 0444,
1034 .proc_handler = &proc_dointvec,
1035 },
1036 {
1037 .ctl_name = FS_OVERFLOWUID,
1038 .procname = "overflowuid",
1039 .data = &fs_overflowuid,
1040 .maxlen = sizeof(int),
1041 .mode = 0644,
1042 .proc_handler = &proc_dointvec_minmax,
1043 .strategy = &sysctl_intvec,
1044 .extra1 = &minolduid,
1045 .extra2 = &maxolduid,
1046 },
1047 {
1048 .ctl_name = FS_OVERFLOWGID,
1049 .procname = "overflowgid",
1050 .data = &fs_overflowgid,
1051 .maxlen = sizeof(int),
1052 .mode = 0644,
1053 .proc_handler = &proc_dointvec_minmax,
1054 .strategy = &sysctl_intvec,
1055 .extra1 = &minolduid,
1056 .extra2 = &maxolduid,
1057 },
1058 {
1059 .ctl_name = FS_LEASES,
1060 .procname = "leases-enable",
1061 .data = &leases_enable,
1062 .maxlen = sizeof(int),
1063 .mode = 0644,
1064 .proc_handler = &proc_dointvec,
1065 },
1066 #ifdef CONFIG_DNOTIFY
1067 {
1068 .ctl_name = FS_DIR_NOTIFY,
1069 .procname = "dir-notify-enable",
1070 .data = &dir_notify_enable,
1071 .maxlen = sizeof(int),
1072 .mode = 0644,
1073 .proc_handler = &proc_dointvec,
1074 },
1075 #endif
1076 #ifdef CONFIG_MMU
1077 {
1078 .ctl_name = FS_LEASE_TIME,
1079 .procname = "lease-break-time",
1080 .data = &lease_break_time,
1081 .maxlen = sizeof(int),
1082 .mode = 0644,
1083 .proc_handler = &proc_dointvec,
1084 },
1085 {
1086 .ctl_name = FS_AIO_NR,
1087 .procname = "aio-nr",
1088 .data = &aio_nr,
1089 .maxlen = sizeof(aio_nr),
1090 .mode = 0444,
1091 .proc_handler = &proc_doulongvec_minmax,
1092 },
1093 {
1094 .ctl_name = FS_AIO_MAX_NR,
1095 .procname = "aio-max-nr",
1096 .data = &aio_max_nr,
1097 .maxlen = sizeof(aio_max_nr),
1098 .mode = 0644,
1099 .proc_handler = &proc_doulongvec_minmax,
1100 },
1101 #ifdef CONFIG_INOTIFY_USER
1102 {
1103 .ctl_name = FS_INOTIFY,
1104 .procname = "inotify",
1105 .mode = 0555,
1106 .child = inotify_table,
1107 },
1108 #endif
1109 #endif
1110 {
1111 .ctl_name = KERN_SETUID_DUMPABLE,
1112 .procname = "suid_dumpable",
1113 .data = &suid_dumpable,
1114 .maxlen = sizeof(int),
1115 .mode = 0644,
1116 .proc_handler = &proc_dointvec,
1117 },
1118 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1119 {
1120 .ctl_name = CTL_UNNUMBERED,
1121 .procname = "binfmt_misc",
1122 .mode = 0555,
1123 .child = binfmt_misc_table,
1124 },
1125 #endif
1126 { .ctl_name = 0 }
1127 };
1128
1129 static ctl_table debug_table[] = {
1130 { .ctl_name = 0 }
1131 };
1132
1133 static ctl_table dev_table[] = {
1134 { .ctl_name = 0 }
1135 };
1136
1137 static DEFINE_SPINLOCK(sysctl_lock);
1138
1139 /* called under sysctl_lock */
1140 static int use_table(struct ctl_table_header *p)
1141 {
1142 if (unlikely(p->unregistering))
1143 return 0;
1144 p->used++;
1145 return 1;
1146 }
1147
1148 /* called under sysctl_lock */
1149 static void unuse_table(struct ctl_table_header *p)
1150 {
1151 if (!--p->used)
1152 if (unlikely(p->unregistering))
1153 complete(p->unregistering);
1154 }
1155
1156 /* called under sysctl_lock, will reacquire if has to wait */
1157 static void start_unregistering(struct ctl_table_header *p)
1158 {
1159 /*
1160 * if p->used is 0, nobody will ever touch that entry again;
1161 * we'll eliminate all paths to it before dropping sysctl_lock
1162 */
1163 if (unlikely(p->used)) {
1164 struct completion wait;
1165 init_completion(&wait);
1166 p->unregistering = &wait;
1167 spin_unlock(&sysctl_lock);
1168 wait_for_completion(&wait);
1169 spin_lock(&sysctl_lock);
1170 }
1171 /*
1172 * do not remove from the list until nobody holds it; walking the
1173 * list in do_sysctl() relies on that.
1174 */
1175 list_del_init(&p->ctl_entry);
1176 }
1177
1178 void sysctl_head_finish(struct ctl_table_header *head)
1179 {
1180 if (!head)
1181 return;
1182 spin_lock(&sysctl_lock);
1183 unuse_table(head);
1184 spin_unlock(&sysctl_lock);
1185 }
1186
1187 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1188 {
1189 struct ctl_table_header *head;
1190 struct list_head *tmp;
1191 spin_lock(&sysctl_lock);
1192 if (prev) {
1193 tmp = &prev->ctl_entry;
1194 unuse_table(prev);
1195 goto next;
1196 }
1197 tmp = &root_table_header.ctl_entry;
1198 for (;;) {
1199 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1200
1201 if (!use_table(head))
1202 goto next;
1203 spin_unlock(&sysctl_lock);
1204 return head;
1205 next:
1206 tmp = tmp->next;
1207 if (tmp == &root_table_header.ctl_entry)
1208 break;
1209 }
1210 spin_unlock(&sysctl_lock);
1211 return NULL;
1212 }
1213
1214 #ifdef CONFIG_SYSCTL_SYSCALL
1215 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1216 void __user *newval, size_t newlen)
1217 {
1218 struct ctl_table_header *head;
1219 int error = -ENOTDIR;
1220
1221 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1222 return -ENOTDIR;
1223 if (oldval) {
1224 int old_len;
1225 if (!oldlenp || get_user(old_len, oldlenp))
1226 return -EFAULT;
1227 }
1228
1229 for (head = sysctl_head_next(NULL); head;
1230 head = sysctl_head_next(head)) {
1231 error = parse_table(name, nlen, oldval, oldlenp,
1232 newval, newlen, head->ctl_table);
1233 if (error != -ENOTDIR) {
1234 sysctl_head_finish(head);
1235 break;
1236 }
1237 }
1238 return error;
1239 }
1240
1241 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1242 {
1243 struct __sysctl_args tmp;
1244 int error;
1245
1246 if (copy_from_user(&tmp, args, sizeof(tmp)))
1247 return -EFAULT;
1248
1249 lock_kernel();
1250 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1251 tmp.newval, tmp.newlen);
1252 unlock_kernel();
1253 return error;
1254 }
1255 #endif /* CONFIG_SYSCTL_SYSCALL */
1256
1257 /*
1258 * sysctl_perm does NOT grant the superuser all rights automatically, because
1259 * some sysctl variables are readonly even to root.
1260 */
1261
1262 static int test_perm(int mode, int op)
1263 {
1264 if (!current->euid)
1265 mode >>= 6;
1266 else if (in_egroup_p(0))
1267 mode >>= 3;
1268 if ((mode & op & 0007) == op)
1269 return 0;
1270 return -EACCES;
1271 }
1272
1273 int sysctl_perm(ctl_table *table, int op)
1274 {
1275 int error;
1276 error = security_sysctl(table, op);
1277 if (error)
1278 return error;
1279 return test_perm(table->mode, op);
1280 }
1281
1282 #ifdef CONFIG_SYSCTL_SYSCALL
1283 static int parse_table(int __user *name, int nlen,
1284 void __user *oldval, size_t __user *oldlenp,
1285 void __user *newval, size_t newlen,
1286 ctl_table *table)
1287 {
1288 int n;
1289 repeat:
1290 if (!nlen)
1291 return -ENOTDIR;
1292 if (get_user(n, name))
1293 return -EFAULT;
1294 for ( ; table->ctl_name || table->procname; table++) {
1295 if (!table->ctl_name)
1296 continue;
1297 if (n == table->ctl_name) {
1298 int error;
1299 if (table->child) {
1300 if (sysctl_perm(table, 001))
1301 return -EPERM;
1302 name++;
1303 nlen--;
1304 table = table->child;
1305 goto repeat;
1306 }
1307 error = do_sysctl_strategy(table, name, nlen,
1308 oldval, oldlenp,
1309 newval, newlen);
1310 return error;
1311 }
1312 }
1313 return -ENOTDIR;
1314 }
1315
1316 /* Perform the actual read/write of a sysctl table entry. */
1317 int do_sysctl_strategy (ctl_table *table,
1318 int __user *name, int nlen,
1319 void __user *oldval, size_t __user *oldlenp,
1320 void __user *newval, size_t newlen)
1321 {
1322 int op = 0, rc;
1323 size_t len;
1324
1325 if (oldval)
1326 op |= 004;
1327 if (newval)
1328 op |= 002;
1329 if (sysctl_perm(table, op))
1330 return -EPERM;
1331
1332 if (table->strategy) {
1333 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1334 newval, newlen);
1335 if (rc < 0)
1336 return rc;
1337 if (rc > 0)
1338 return 0;
1339 }
1340
1341 /* If there is no strategy routine, or if the strategy returns
1342 * zero, proceed with automatic r/w */
1343 if (table->data && table->maxlen) {
1344 if (oldval && oldlenp) {
1345 if (get_user(len, oldlenp))
1346 return -EFAULT;
1347 if (len) {
1348 if (len > table->maxlen)
1349 len = table->maxlen;
1350 if(copy_to_user(oldval, table->data, len))
1351 return -EFAULT;
1352 if(put_user(len, oldlenp))
1353 return -EFAULT;
1354 }
1355 }
1356 if (newval && newlen) {
1357 len = newlen;
1358 if (len > table->maxlen)
1359 len = table->maxlen;
1360 if(copy_from_user(table->data, newval, len))
1361 return -EFAULT;
1362 }
1363 }
1364 return 0;
1365 }
1366 #endif /* CONFIG_SYSCTL_SYSCALL */
1367
1368 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1369 {
1370 for (; table->ctl_name || table->procname; table++) {
1371 table->parent = parent;
1372 if (table->child)
1373 sysctl_set_parent(table, table->child);
1374 }
1375 }
1376
1377 static __init int sysctl_init(void)
1378 {
1379 sysctl_set_parent(NULL, root_table);
1380 return 0;
1381 }
1382
1383 core_initcall(sysctl_init);
1384
1385 /**
1386 * register_sysctl_table - register a sysctl hierarchy
1387 * @table: the top-level table structure
1388 *
1389 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1390 * array. An entry with a ctl_name of 0 terminates the table.
1391 *
1392 * The members of the &ctl_table structure are used as follows:
1393 *
1394 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1395 * must be unique within that level of sysctl
1396 *
1397 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1398 * enter a sysctl file
1399 *
1400 * data - a pointer to data for use by proc_handler
1401 *
1402 * maxlen - the maximum size in bytes of the data
1403 *
1404 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1405 *
1406 * child - a pointer to the child sysctl table if this entry is a directory, or
1407 * %NULL.
1408 *
1409 * proc_handler - the text handler routine (described below)
1410 *
1411 * strategy - the strategy routine (described below)
1412 *
1413 * de - for internal use by the sysctl routines
1414 *
1415 * extra1, extra2 - extra pointers usable by the proc handler routines
1416 *
1417 * Leaf nodes in the sysctl tree will be represented by a single file
1418 * under /proc; non-leaf nodes will be represented by directories.
1419 *
1420 * sysctl(2) can automatically manage read and write requests through
1421 * the sysctl table. The data and maxlen fields of the ctl_table
1422 * struct enable minimal validation of the values being written to be
1423 * performed, and the mode field allows minimal authentication.
1424 *
1425 * More sophisticated management can be enabled by the provision of a
1426 * strategy routine with the table entry. This will be called before
1427 * any automatic read or write of the data is performed.
1428 *
1429 * The strategy routine may return
1430 *
1431 * < 0 - Error occurred (error is passed to user process)
1432 *
1433 * 0 - OK - proceed with automatic read or write.
1434 *
1435 * > 0 - OK - read or write has been done by the strategy routine, so
1436 * return immediately.
1437 *
1438 * There must be a proc_handler routine for any terminal nodes
1439 * mirrored under /proc/sys (non-terminals are handled by a built-in
1440 * directory handler). Several default handlers are available to
1441 * cover common cases -
1442 *
1443 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1444 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1445 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1446 *
1447 * It is the handler's job to read the input buffer from user memory
1448 * and process it. The handler should return 0 on success.
1449 *
1450 * This routine returns %NULL on a failure to register, and a pointer
1451 * to the table header on success.
1452 */
1453 struct ctl_table_header *register_sysctl_table(ctl_table * table)
1454 {
1455 struct ctl_table_header *tmp;
1456 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1457 if (!tmp)
1458 return NULL;
1459 tmp->ctl_table = table;
1460 INIT_LIST_HEAD(&tmp->ctl_entry);
1461 tmp->used = 0;
1462 tmp->unregistering = NULL;
1463 sysctl_set_parent(NULL, table);
1464 spin_lock(&sysctl_lock);
1465 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
1466 spin_unlock(&sysctl_lock);
1467 return tmp;
1468 }
1469
1470 /**
1471 * unregister_sysctl_table - unregister a sysctl table hierarchy
1472 * @header: the header returned from register_sysctl_table
1473 *
1474 * Unregisters the sysctl table and all children. proc entries may not
1475 * actually be removed until they are no longer used by anyone.
1476 */
1477 void unregister_sysctl_table(struct ctl_table_header * header)
1478 {
1479 might_sleep();
1480 spin_lock(&sysctl_lock);
1481 start_unregistering(header);
1482 spin_unlock(&sysctl_lock);
1483 kfree(header);
1484 }
1485
1486 #else /* !CONFIG_SYSCTL */
1487 struct ctl_table_header *register_sysctl_table(ctl_table * table)
1488 {
1489 return NULL;
1490 }
1491
1492 void unregister_sysctl_table(struct ctl_table_header * table)
1493 {
1494 }
1495
1496 #endif /* CONFIG_SYSCTL */
1497
1498 /*
1499 * /proc/sys support
1500 */
1501
1502 #ifdef CONFIG_PROC_SYSCTL
1503
1504 static int _proc_do_string(void* data, int maxlen, int write,
1505 struct file *filp, void __user *buffer,
1506 size_t *lenp, loff_t *ppos)
1507 {
1508 size_t len;
1509 char __user *p;
1510 char c;
1511
1512 if (!data || !maxlen || !*lenp) {
1513 *lenp = 0;
1514 return 0;
1515 }
1516
1517 if (write) {
1518 len = 0;
1519 p = buffer;
1520 while (len < *lenp) {
1521 if (get_user(c, p++))
1522 return -EFAULT;
1523 if (c == 0 || c == '\n')
1524 break;
1525 len++;
1526 }
1527 if (len >= maxlen)
1528 len = maxlen-1;
1529 if(copy_from_user(data, buffer, len))
1530 return -EFAULT;
1531 ((char *) data)[len] = 0;
1532 *ppos += *lenp;
1533 } else {
1534 len = strlen(data);
1535 if (len > maxlen)
1536 len = maxlen;
1537
1538 if (*ppos > len) {
1539 *lenp = 0;
1540 return 0;
1541 }
1542
1543 data += *ppos;
1544 len -= *ppos;
1545
1546 if (len > *lenp)
1547 len = *lenp;
1548 if (len)
1549 if(copy_to_user(buffer, data, len))
1550 return -EFAULT;
1551 if (len < *lenp) {
1552 if(put_user('\n', ((char __user *) buffer) + len))
1553 return -EFAULT;
1554 len++;
1555 }
1556 *lenp = len;
1557 *ppos += len;
1558 }
1559 return 0;
1560 }
1561
1562 /**
1563 * proc_dostring - read a string sysctl
1564 * @table: the sysctl table
1565 * @write: %TRUE if this is a write to the sysctl file
1566 * @filp: the file structure
1567 * @buffer: the user buffer
1568 * @lenp: the size of the user buffer
1569 * @ppos: file position
1570 *
1571 * Reads/writes a string from/to the user buffer. If the kernel
1572 * buffer provided is not large enough to hold the string, the
1573 * string is truncated. The copied string is %NULL-terminated.
1574 * If the string is being read by the user process, it is copied
1575 * and a newline '\n' is added. It is truncated if the buffer is
1576 * not large enough.
1577 *
1578 * Returns 0 on success.
1579 */
1580 int proc_dostring(ctl_table *table, int write, struct file *filp,
1581 void __user *buffer, size_t *lenp, loff_t *ppos)
1582 {
1583 return _proc_do_string(table->data, table->maxlen, write, filp,
1584 buffer, lenp, ppos);
1585 }
1586
1587
1588 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1589 int *valp,
1590 int write, void *data)
1591 {
1592 if (write) {
1593 *valp = *negp ? -*lvalp : *lvalp;
1594 } else {
1595 int val = *valp;
1596 if (val < 0) {
1597 *negp = -1;
1598 *lvalp = (unsigned long)-val;
1599 } else {
1600 *negp = 0;
1601 *lvalp = (unsigned long)val;
1602 }
1603 }
1604 return 0;
1605 }
1606
1607 static int __do_proc_dointvec(void *tbl_data, ctl_table *table,
1608 int write, struct file *filp, void __user *buffer,
1609 size_t *lenp, loff_t *ppos,
1610 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1611 int write, void *data),
1612 void *data)
1613 {
1614 #define TMPBUFLEN 21
1615 int *i, vleft, first=1, neg, val;
1616 unsigned long lval;
1617 size_t left, len;
1618
1619 char buf[TMPBUFLEN], *p;
1620 char __user *s = buffer;
1621
1622 if (!tbl_data || !table->maxlen || !*lenp ||
1623 (*ppos && !write)) {
1624 *lenp = 0;
1625 return 0;
1626 }
1627
1628 i = (int *) tbl_data;
1629 vleft = table->maxlen / sizeof(*i);
1630 left = *lenp;
1631
1632 if (!conv)
1633 conv = do_proc_dointvec_conv;
1634
1635 for (; left && vleft--; i++, first=0) {
1636 if (write) {
1637 while (left) {
1638 char c;
1639 if (get_user(c, s))
1640 return -EFAULT;
1641 if (!isspace(c))
1642 break;
1643 left--;
1644 s++;
1645 }
1646 if (!left)
1647 break;
1648 neg = 0;
1649 len = left;
1650 if (len > sizeof(buf) - 1)
1651 len = sizeof(buf) - 1;
1652 if (copy_from_user(buf, s, len))
1653 return -EFAULT;
1654 buf[len] = 0;
1655 p = buf;
1656 if (*p == '-' && left > 1) {
1657 neg = 1;
1658 p++;
1659 }
1660 if (*p < '0' || *p > '9')
1661 break;
1662
1663 lval = simple_strtoul(p, &p, 0);
1664
1665 len = p-buf;
1666 if ((len < left) && *p && !isspace(*p))
1667 break;
1668 if (neg)
1669 val = -val;
1670 s += len;
1671 left -= len;
1672
1673 if (conv(&neg, &lval, i, 1, data))
1674 break;
1675 } else {
1676 p = buf;
1677 if (!first)
1678 *p++ = '\t';
1679
1680 if (conv(&neg, &lval, i, 0, data))
1681 break;
1682
1683 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1684 len = strlen(buf);
1685 if (len > left)
1686 len = left;
1687 if(copy_to_user(s, buf, len))
1688 return -EFAULT;
1689 left -= len;
1690 s += len;
1691 }
1692 }
1693
1694 if (!write && !first && left) {
1695 if(put_user('\n', s))
1696 return -EFAULT;
1697 left--, s++;
1698 }
1699 if (write) {
1700 while (left) {
1701 char c;
1702 if (get_user(c, s++))
1703 return -EFAULT;
1704 if (!isspace(c))
1705 break;
1706 left--;
1707 }
1708 }
1709 if (write && first)
1710 return -EINVAL;
1711 *lenp -= left;
1712 *ppos += *lenp;
1713 return 0;
1714 #undef TMPBUFLEN
1715 }
1716
1717 static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1718 void __user *buffer, size_t *lenp, loff_t *ppos,
1719 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1720 int write, void *data),
1721 void *data)
1722 {
1723 return __do_proc_dointvec(table->data, table, write, filp,
1724 buffer, lenp, ppos, conv, data);
1725 }
1726
1727 /**
1728 * proc_dointvec - read a vector of integers
1729 * @table: the sysctl table
1730 * @write: %TRUE if this is a write to the sysctl file
1731 * @filp: the file structure
1732 * @buffer: the user buffer
1733 * @lenp: the size of the user buffer
1734 * @ppos: file position
1735 *
1736 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1737 * values from/to the user buffer, treated as an ASCII string.
1738 *
1739 * Returns 0 on success.
1740 */
1741 int proc_dointvec(ctl_table *table, int write, struct file *filp,
1742 void __user *buffer, size_t *lenp, loff_t *ppos)
1743 {
1744 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1745 NULL,NULL);
1746 }
1747
1748 #define OP_SET 0
1749 #define OP_AND 1
1750 #define OP_OR 2
1751
1752 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1753 int *valp,
1754 int write, void *data)
1755 {
1756 int op = *(int *)data;
1757 if (write) {
1758 int val = *negp ? -*lvalp : *lvalp;
1759 switch(op) {
1760 case OP_SET: *valp = val; break;
1761 case OP_AND: *valp &= val; break;
1762 case OP_OR: *valp |= val; break;
1763 }
1764 } else {
1765 int val = *valp;
1766 if (val < 0) {
1767 *negp = -1;
1768 *lvalp = (unsigned long)-val;
1769 } else {
1770 *negp = 0;
1771 *lvalp = (unsigned long)val;
1772 }
1773 }
1774 return 0;
1775 }
1776
1777 /*
1778 * init may raise the set.
1779 */
1780
1781 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1782 void __user *buffer, size_t *lenp, loff_t *ppos)
1783 {
1784 int op;
1785
1786 if (write && !capable(CAP_SYS_MODULE)) {
1787 return -EPERM;
1788 }
1789
1790 op = is_init(current) ? OP_SET : OP_AND;
1791 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1792 do_proc_dointvec_bset_conv,&op);
1793 }
1794
1795 /*
1796 * Taint values can only be increased
1797 */
1798 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
1799 void __user *buffer, size_t *lenp, loff_t *ppos)
1800 {
1801 int op;
1802
1803 if (write && !capable(CAP_SYS_ADMIN))
1804 return -EPERM;
1805
1806 op = OP_OR;
1807 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1808 do_proc_dointvec_bset_conv,&op);
1809 }
1810
1811 struct do_proc_dointvec_minmax_conv_param {
1812 int *min;
1813 int *max;
1814 };
1815
1816 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1817 int *valp,
1818 int write, void *data)
1819 {
1820 struct do_proc_dointvec_minmax_conv_param *param = data;
1821 if (write) {
1822 int val = *negp ? -*lvalp : *lvalp;
1823 if ((param->min && *param->min > val) ||
1824 (param->max && *param->max < val))
1825 return -EINVAL;
1826 *valp = val;
1827 } else {
1828 int val = *valp;
1829 if (val < 0) {
1830 *negp = -1;
1831 *lvalp = (unsigned long)-val;
1832 } else {
1833 *negp = 0;
1834 *lvalp = (unsigned long)val;
1835 }
1836 }
1837 return 0;
1838 }
1839
1840 /**
1841 * proc_dointvec_minmax - read a vector of integers with min/max values
1842 * @table: the sysctl table
1843 * @write: %TRUE if this is a write to the sysctl file
1844 * @filp: the file structure
1845 * @buffer: the user buffer
1846 * @lenp: the size of the user buffer
1847 * @ppos: file position
1848 *
1849 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1850 * values from/to the user buffer, treated as an ASCII string.
1851 *
1852 * This routine will ensure the values are within the range specified by
1853 * table->extra1 (min) and table->extra2 (max).
1854 *
1855 * Returns 0 on success.
1856 */
1857 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1858 void __user *buffer, size_t *lenp, loff_t *ppos)
1859 {
1860 struct do_proc_dointvec_minmax_conv_param param = {
1861 .min = (int *) table->extra1,
1862 .max = (int *) table->extra2,
1863 };
1864 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1865 do_proc_dointvec_minmax_conv, &param);
1866 }
1867
1868 static int __do_proc_doulongvec_minmax(void *data, ctl_table *table, int write,
1869 struct file *filp,
1870 void __user *buffer,
1871 size_t *lenp, loff_t *ppos,
1872 unsigned long convmul,
1873 unsigned long convdiv)
1874 {
1875 #define TMPBUFLEN 21
1876 unsigned long *i, *min, *max, val;
1877 int vleft, first=1, neg;
1878 size_t len, left;
1879 char buf[TMPBUFLEN], *p;
1880 char __user *s = buffer;
1881
1882 if (!data || !table->maxlen || !*lenp ||
1883 (*ppos && !write)) {
1884 *lenp = 0;
1885 return 0;
1886 }
1887
1888 i = (unsigned long *) data;
1889 min = (unsigned long *) table->extra1;
1890 max = (unsigned long *) table->extra2;
1891 vleft = table->maxlen / sizeof(unsigned long);
1892 left = *lenp;
1893
1894 for (; left && vleft--; i++, min++, max++, first=0) {
1895 if (write) {
1896 while (left) {
1897 char c;
1898 if (get_user(c, s))
1899 return -EFAULT;
1900 if (!isspace(c))
1901 break;
1902 left--;
1903 s++;
1904 }
1905 if (!left)
1906 break;
1907 neg = 0;
1908 len = left;
1909 if (len > TMPBUFLEN-1)
1910 len = TMPBUFLEN-1;
1911 if (copy_from_user(buf, s, len))
1912 return -EFAULT;
1913 buf[len] = 0;
1914 p = buf;
1915 if (*p == '-' && left > 1) {
1916 neg = 1;
1917 p++;
1918 }
1919 if (*p < '0' || *p > '9')
1920 break;
1921 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
1922 len = p-buf;
1923 if ((len < left) && *p && !isspace(*p))
1924 break;
1925 if (neg)
1926 val = -val;
1927 s += len;
1928 left -= len;
1929
1930 if(neg)
1931 continue;
1932 if ((min && val < *min) || (max && val > *max))
1933 continue;
1934 *i = val;
1935 } else {
1936 p = buf;
1937 if (!first)
1938 *p++ = '\t';
1939 sprintf(p, "%lu", convdiv * (*i) / convmul);
1940 len = strlen(buf);
1941 if (len > left)
1942 len = left;
1943 if(copy_to_user(s, buf, len))
1944 return -EFAULT;
1945 left -= len;
1946 s += len;
1947 }
1948 }
1949
1950 if (!write && !first && left) {
1951 if(put_user('\n', s))
1952 return -EFAULT;
1953 left--, s++;
1954 }
1955 if (write) {
1956 while (left) {
1957 char c;
1958 if (get_user(c, s++))
1959 return -EFAULT;
1960 if (!isspace(c))
1961 break;
1962 left--;
1963 }
1964 }
1965 if (write && first)
1966 return -EINVAL;
1967 *lenp -= left;
1968 *ppos += *lenp;
1969 return 0;
1970 #undef TMPBUFLEN
1971 }
1972
1973 static int do_proc_doulongvec_minmax(ctl_table *table, int write,
1974 struct file *filp,
1975 void __user *buffer,
1976 size_t *lenp, loff_t *ppos,
1977 unsigned long convmul,
1978 unsigned long convdiv)
1979 {
1980 return __do_proc_doulongvec_minmax(table->data, table, write,
1981 filp, buffer, lenp, ppos, convmul, convdiv);
1982 }
1983
1984 /**
1985 * proc_doulongvec_minmax - read a vector of long integers with min/max values
1986 * @table: the sysctl table
1987 * @write: %TRUE if this is a write to the sysctl file
1988 * @filp: the file structure
1989 * @buffer: the user buffer
1990 * @lenp: the size of the user buffer
1991 * @ppos: file position
1992 *
1993 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1994 * values from/to the user buffer, treated as an ASCII string.
1995 *
1996 * This routine will ensure the values are within the range specified by
1997 * table->extra1 (min) and table->extra2 (max).
1998 *
1999 * Returns 0 on success.
2000 */
2001 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2002 void __user *buffer, size_t *lenp, loff_t *ppos)
2003 {
2004 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2005 }
2006
2007 /**
2008 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2009 * @table: the sysctl table
2010 * @write: %TRUE if this is a write to the sysctl file
2011 * @filp: the file structure
2012 * @buffer: the user buffer
2013 * @lenp: the size of the user buffer
2014 * @ppos: file position
2015 *
2016 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2017 * values from/to the user buffer, treated as an ASCII string. The values
2018 * are treated as milliseconds, and converted to jiffies when they are stored.
2019 *
2020 * This routine will ensure the values are within the range specified by
2021 * table->extra1 (min) and table->extra2 (max).
2022 *
2023 * Returns 0 on success.
2024 */
2025 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2026 struct file *filp,
2027 void __user *buffer,
2028 size_t *lenp, loff_t *ppos)
2029 {
2030 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2031 lenp, ppos, HZ, 1000l);
2032 }
2033
2034
2035 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2036 int *valp,
2037 int write, void *data)
2038 {
2039 if (write) {
2040 if (*lvalp > LONG_MAX / HZ)
2041 return 1;
2042 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2043 } else {
2044 int val = *valp;
2045 unsigned long lval;
2046 if (val < 0) {
2047 *negp = -1;
2048 lval = (unsigned long)-val;
2049 } else {
2050 *negp = 0;
2051 lval = (unsigned long)val;
2052 }
2053 *lvalp = lval / HZ;
2054 }
2055 return 0;
2056 }
2057
2058 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2059 int *valp,
2060 int write, void *data)
2061 {
2062 if (write) {
2063 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2064 return 1;
2065 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2066 } else {
2067 int val = *valp;
2068 unsigned long lval;
2069 if (val < 0) {
2070 *negp = -1;
2071 lval = (unsigned long)-val;
2072 } else {
2073 *negp = 0;
2074 lval = (unsigned long)val;
2075 }
2076 *lvalp = jiffies_to_clock_t(lval);
2077 }
2078 return 0;
2079 }
2080
2081 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2082 int *valp,
2083 int write, void *data)
2084 {
2085 if (write) {
2086 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2087 } else {
2088 int val = *valp;
2089 unsigned long lval;
2090 if (val < 0) {
2091 *negp = -1;
2092 lval = (unsigned long)-val;
2093 } else {
2094 *negp = 0;
2095 lval = (unsigned long)val;
2096 }
2097 *lvalp = jiffies_to_msecs(lval);
2098 }
2099 return 0;
2100 }
2101
2102 /**
2103 * proc_dointvec_jiffies - read a vector of integers as seconds
2104 * @table: the sysctl table
2105 * @write: %TRUE if this is a write to the sysctl file
2106 * @filp: the file structure
2107 * @buffer: the user buffer
2108 * @lenp: the size of the user buffer
2109 * @ppos: file position
2110 *
2111 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2112 * values from/to the user buffer, treated as an ASCII string.
2113 * The values read are assumed to be in seconds, and are converted into
2114 * jiffies.
2115 *
2116 * Returns 0 on success.
2117 */
2118 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2119 void __user *buffer, size_t *lenp, loff_t *ppos)
2120 {
2121 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2122 do_proc_dointvec_jiffies_conv,NULL);
2123 }
2124
2125 /**
2126 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2127 * @table: the sysctl table
2128 * @write: %TRUE if this is a write to the sysctl file
2129 * @filp: the file structure
2130 * @buffer: the user buffer
2131 * @lenp: the size of the user buffer
2132 * @ppos: pointer to the file position
2133 *
2134 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2135 * values from/to the user buffer, treated as an ASCII string.
2136 * The values read are assumed to be in 1/USER_HZ seconds, and
2137 * are converted into jiffies.
2138 *
2139 * Returns 0 on success.
2140 */
2141 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2142 void __user *buffer, size_t *lenp, loff_t *ppos)
2143 {
2144 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2145 do_proc_dointvec_userhz_jiffies_conv,NULL);
2146 }
2147
2148 /**
2149 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2150 * @table: the sysctl table
2151 * @write: %TRUE if this is a write to the sysctl file
2152 * @filp: the file structure
2153 * @buffer: the user buffer
2154 * @lenp: the size of the user buffer
2155 * @ppos: file position
2156 * @ppos: the current position in the file
2157 *
2158 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2159 * values from/to the user buffer, treated as an ASCII string.
2160 * The values read are assumed to be in 1/1000 seconds, and
2161 * are converted into jiffies.
2162 *
2163 * Returns 0 on success.
2164 */
2165 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2166 void __user *buffer, size_t *lenp, loff_t *ppos)
2167 {
2168 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2169 do_proc_dointvec_ms_jiffies_conv, NULL);
2170 }
2171
2172 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
2173 void __user *buffer, size_t *lenp, loff_t *ppos)
2174 {
2175 struct pid *new_pid;
2176 pid_t tmp;
2177 int r;
2178
2179 tmp = pid_nr(cad_pid);
2180
2181 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2182 lenp, ppos, NULL, NULL);
2183 if (r || !write)
2184 return r;
2185
2186 new_pid = find_get_pid(tmp);
2187 if (!new_pid)
2188 return -ESRCH;
2189
2190 put_pid(xchg(&cad_pid, new_pid));
2191 return 0;
2192 }
2193
2194 #else /* CONFIG_PROC_FS */
2195
2196 int proc_dostring(ctl_table *table, int write, struct file *filp,
2197 void __user *buffer, size_t *lenp, loff_t *ppos)
2198 {
2199 return -ENOSYS;
2200 }
2201
2202 int proc_dointvec(ctl_table *table, int write, struct file *filp,
2203 void __user *buffer, size_t *lenp, loff_t *ppos)
2204 {
2205 return -ENOSYS;
2206 }
2207
2208 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2209 void __user *buffer, size_t *lenp, loff_t *ppos)
2210 {
2211 return -ENOSYS;
2212 }
2213
2214 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2215 void __user *buffer, size_t *lenp, loff_t *ppos)
2216 {
2217 return -ENOSYS;
2218 }
2219
2220 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2221 void __user *buffer, size_t *lenp, loff_t *ppos)
2222 {
2223 return -ENOSYS;
2224 }
2225
2226 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2227 void __user *buffer, size_t *lenp, loff_t *ppos)
2228 {
2229 return -ENOSYS;
2230 }
2231
2232 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2233 void __user *buffer, size_t *lenp, loff_t *ppos)
2234 {
2235 return -ENOSYS;
2236 }
2237
2238 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2239 void __user *buffer, size_t *lenp, loff_t *ppos)
2240 {
2241 return -ENOSYS;
2242 }
2243
2244 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2245 struct file *filp,
2246 void __user *buffer,
2247 size_t *lenp, loff_t *ppos)
2248 {
2249 return -ENOSYS;
2250 }
2251
2252
2253 #endif /* CONFIG_PROC_FS */
2254
2255
2256 #ifdef CONFIG_SYSCTL_SYSCALL
2257 /*
2258 * General sysctl support routines
2259 */
2260
2261 /* The generic string strategy routine: */
2262 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2263 void __user *oldval, size_t __user *oldlenp,
2264 void __user *newval, size_t newlen)
2265 {
2266 if (!table->data || !table->maxlen)
2267 return -ENOTDIR;
2268
2269 if (oldval && oldlenp) {
2270 size_t bufsize;
2271 if (get_user(bufsize, oldlenp))
2272 return -EFAULT;
2273 if (bufsize) {
2274 size_t len = strlen(table->data), copied;
2275
2276 /* This shouldn't trigger for a well-formed sysctl */
2277 if (len > table->maxlen)
2278 len = table->maxlen;
2279
2280 /* Copy up to a max of bufsize-1 bytes of the string */
2281 copied = (len >= bufsize) ? bufsize - 1 : len;
2282
2283 if (copy_to_user(oldval, table->data, copied) ||
2284 put_user(0, (char __user *)(oldval + copied)))
2285 return -EFAULT;
2286 if (put_user(len, oldlenp))
2287 return -EFAULT;
2288 }
2289 }
2290 if (newval && newlen) {
2291 size_t len = newlen;
2292 if (len > table->maxlen)
2293 len = table->maxlen;
2294 if(copy_from_user(table->data, newval, len))
2295 return -EFAULT;
2296 if (len == table->maxlen)
2297 len--;
2298 ((char *) table->data)[len] = 0;
2299 }
2300 return 1;
2301 }
2302
2303 /*
2304 * This function makes sure that all of the integers in the vector
2305 * are between the minimum and maximum values given in the arrays
2306 * table->extra1 and table->extra2, respectively.
2307 */
2308 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2309 void __user *oldval, size_t __user *oldlenp,
2310 void __user *newval, size_t newlen)
2311 {
2312
2313 if (newval && newlen) {
2314 int __user *vec = (int __user *) newval;
2315 int *min = (int *) table->extra1;
2316 int *max = (int *) table->extra2;
2317 size_t length;
2318 int i;
2319
2320 if (newlen % sizeof(int) != 0)
2321 return -EINVAL;
2322
2323 if (!table->extra1 && !table->extra2)
2324 return 0;
2325
2326 if (newlen > table->maxlen)
2327 newlen = table->maxlen;
2328 length = newlen / sizeof(int);
2329
2330 for (i = 0; i < length; i++) {
2331 int value;
2332 if (get_user(value, vec + i))
2333 return -EFAULT;
2334 if (min && value < min[i])
2335 return -EINVAL;
2336 if (max && value > max[i])
2337 return -EINVAL;
2338 }
2339 }
2340 return 0;
2341 }
2342
2343 /* Strategy function to convert jiffies to seconds */
2344 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2345 void __user *oldval, size_t __user *oldlenp,
2346 void __user *newval, size_t newlen)
2347 {
2348 if (oldval && oldlenp) {
2349 size_t olen;
2350
2351 if (get_user(olen, oldlenp))
2352 return -EFAULT;
2353 if (olen) {
2354 int val;
2355
2356 if (olen < sizeof(int))
2357 return -EINVAL;
2358
2359 val = *(int *)(table->data) / HZ;
2360 if (put_user(val, (int __user *)oldval))
2361 return -EFAULT;
2362 if (put_user(sizeof(int), oldlenp))
2363 return -EFAULT;
2364 }
2365 }
2366 if (newval && newlen) {
2367 int new;
2368 if (newlen != sizeof(int))
2369 return -EINVAL;
2370 if (get_user(new, (int __user *)newval))
2371 return -EFAULT;
2372 *(int *)(table->data) = new*HZ;
2373 }
2374 return 1;
2375 }
2376
2377 /* Strategy function to convert jiffies to seconds */
2378 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2379 void __user *oldval, size_t __user *oldlenp,
2380 void __user *newval, size_t newlen)
2381 {
2382 if (oldval && oldlenp) {
2383 size_t olen;
2384
2385 if (get_user(olen, oldlenp))
2386 return -EFAULT;
2387 if (olen) {
2388 int val;
2389
2390 if (olen < sizeof(int))
2391 return -EINVAL;
2392
2393 val = jiffies_to_msecs(*(int *)(table->data));
2394 if (put_user(val, (int __user *)oldval))
2395 return -EFAULT;
2396 if (put_user(sizeof(int), oldlenp))
2397 return -EFAULT;
2398 }
2399 }
2400 if (newval && newlen) {
2401 int new;
2402 if (newlen != sizeof(int))
2403 return -EINVAL;
2404 if (get_user(new, (int __user *)newval))
2405 return -EFAULT;
2406 *(int *)(table->data) = msecs_to_jiffies(new);
2407 }
2408 return 1;
2409 }
2410
2411
2412
2413 #else /* CONFIG_SYSCTL_SYSCALL */
2414
2415
2416 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2417 {
2418 static int msg_count;
2419 struct __sysctl_args tmp;
2420 int name[CTL_MAXNAME];
2421 int i;
2422
2423 /* Read in the sysctl name for better debug message logging */
2424 if (copy_from_user(&tmp, args, sizeof(tmp)))
2425 return -EFAULT;
2426 if (tmp.nlen <= 0 || tmp.nlen >= CTL_MAXNAME)
2427 return -ENOTDIR;
2428 for (i = 0; i < tmp.nlen; i++)
2429 if (get_user(name[i], tmp.name + i))
2430 return -EFAULT;
2431
2432 /* Ignore accesses to kernel.version */
2433 if ((tmp.nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2434 goto out;
2435
2436 if (msg_count < 5) {
2437 msg_count++;
2438 printk(KERN_INFO
2439 "warning: process `%s' used the removed sysctl "
2440 "system call with ", current->comm);
2441 for (i = 0; i < tmp.nlen; i++)
2442 printk("%d.", name[i]);
2443 printk("\n");
2444 }
2445 out:
2446 return -ENOSYS;
2447 }
2448
2449 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2450 void __user *oldval, size_t __user *oldlenp,
2451 void __user *newval, size_t newlen)
2452 {
2453 return -ENOSYS;
2454 }
2455
2456 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2457 void __user *oldval, size_t __user *oldlenp,
2458 void __user *newval, size_t newlen)
2459 {
2460 return -ENOSYS;
2461 }
2462
2463 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2464 void __user *oldval, size_t __user *oldlenp,
2465 void __user *newval, size_t newlen)
2466 {
2467 return -ENOSYS;
2468 }
2469
2470 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2471 void __user *oldval, size_t __user *oldlenp,
2472 void __user *newval, size_t newlen)
2473 {
2474 return -ENOSYS;
2475 }
2476
2477 #endif /* CONFIG_SYSCTL_SYSCALL */
2478
2479 /*
2480 * No sense putting this after each symbol definition, twice,
2481 * exception granted :-)
2482 */
2483 EXPORT_SYMBOL(proc_dointvec);
2484 EXPORT_SYMBOL(proc_dointvec_jiffies);
2485 EXPORT_SYMBOL(proc_dointvec_minmax);
2486 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2487 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2488 EXPORT_SYMBOL(proc_dostring);
2489 EXPORT_SYMBOL(proc_doulongvec_minmax);
2490 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2491 EXPORT_SYMBOL(register_sysctl_table);
2492 EXPORT_SYMBOL(sysctl_intvec);
2493 EXPORT_SYMBOL(sysctl_jiffies);
2494 EXPORT_SYMBOL(sysctl_ms_jiffies);
2495 EXPORT_SYMBOL(sysctl_string);
2496 EXPORT_SYMBOL(unregister_sysctl_table);
Linux kernel - Deliverables
This page took 0.099764 seconds and 5 git commands to generate.