?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Improve behaviour of spurious IRQ detect
[linux-2.6/linux-loongson.git] / kernel / sysctl.c
blob2cce2286bdcda987a8a95e5830e5830ff1911e0f
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 print_fatal_signals;
65 extern int sysctl_overcommit_memory;
66 extern int sysctl_overcommit_ratio;
67 extern int sysctl_panic_on_oom;
68 extern int max_threads;
69 extern int core_uses_pid;
70 extern int suid_dumpable;
71 extern char core_pattern[];
72 extern int pid_max;
73 extern int min_free_kbytes;
74 extern int printk_ratelimit_jiffies;
75 extern int printk_ratelimit_burst;
76 extern int pid_max_min, pid_max_max;
77 extern int sysctl_drop_caches;
78 extern int percpu_pagelist_fraction;
79 extern int compat_log;
80 extern int maps_protect;
81 extern int sysctl_stat_interval;
82
83 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
84 static int maxolduid = 65535;
85 static int minolduid;
86 static int min_percpu_pagelist_fract = 8;
87
88 static int ngroups_max = NGROUPS_MAX;
89
90 #ifdef CONFIG_KMOD
91 extern char modprobe_path[];
92 #endif
93 #ifdef CONFIG_CHR_DEV_SG
94 extern int sg_big_buff;
95 #endif
96
97 #ifdef __sparc__
98 extern char reboot_command [];
99 extern int stop_a_enabled;
100 extern int scons_pwroff;
101 #endif
102
103 #ifdef __hppa__
104 extern int pwrsw_enabled;
105 extern int unaligned_enabled;
106 #endif
107
108 #ifdef CONFIG_S390
109 #ifdef CONFIG_MATHEMU
110 extern int sysctl_ieee_emulation_warnings;
111 #endif
112 extern int sysctl_userprocess_debug;
113 extern int spin_retry;
114 #endif
115
116 extern int sysctl_hz_timer;
117
118 #ifdef CONFIG_BSD_PROCESS_ACCT
119 extern int acct_parm[];
120 #endif
121
122 #ifdef CONFIG_IA64
123 extern int no_unaligned_warning;
124 #endif
125
126 #ifdef CONFIG_RT_MUTEXES
127 extern int max_lock_depth;
128 #endif
129
130 #ifdef CONFIG_SYSCTL_SYSCALL
131 static int parse_table(int __user *, int, void __user *, size_t __user *,
132 void __user *, size_t, ctl_table *);
133 #endif
134
135
136 #ifdef CONFIG_PROC_SYSCTL
137 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
138 void __user *buffer, size_t *lenp, loff_t *ppos);
139 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
140 void __user *buffer, size_t *lenp, loff_t *ppos);
141 #endif
142
143 static ctl_table root_table[];
144 static struct ctl_table_header root_table_header =
145 { root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
146
147 static ctl_table kern_table[];
148 static ctl_table vm_table[];
149 static ctl_table fs_table[];
150 static ctl_table debug_table[];
151 static ctl_table dev_table[];
152 extern ctl_table random_table[];
153 #ifdef CONFIG_UNIX98_PTYS
154 extern ctl_table pty_table[];
155 #endif
156 #ifdef CONFIG_INOTIFY_USER
157 extern ctl_table inotify_table[];
158 #endif
159
160 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
161 int sysctl_legacy_va_layout;
162 #endif
163
164
165 /* The default sysctl tables: */
166
167 static ctl_table root_table[] = {
168 {
169 .ctl_name = CTL_KERN,
170 .procname = "kernel",
171 .mode = 0555,
172 .child = kern_table,
173 },
174 {
175 .ctl_name = CTL_VM,
176 .procname = "vm",
177 .mode = 0555,
178 .child = vm_table,
179 },
180 #ifdef CONFIG_NET
181 {
182 .ctl_name = CTL_NET,
183 .procname = "net",
184 .mode = 0555,
185 .child = net_table,
186 },
187 #endif
188 {
189 .ctl_name = CTL_FS,
190 .procname = "fs",
191 .mode = 0555,
192 .child = fs_table,
193 },
194 {
195 .ctl_name = CTL_DEBUG,
196 .procname = "debug",
197 .mode = 0555,
198 .child = debug_table,
199 },
200 {
201 .ctl_name = CTL_DEV,
202 .procname = "dev",
203 .mode = 0555,
204 .child = dev_table,
205 },
206
207 { .ctl_name = 0 }
208 };
209
210 #ifdef CONFIG_SCHED_DEBUG
211 static unsigned long min_sched_granularity_ns = 100000; /* 100 usecs */
212 static unsigned long max_sched_granularity_ns = 1000000000; /* 1 second */
213 static unsigned long min_wakeup_granularity_ns; /* 0 usecs */
214 static unsigned long max_wakeup_granularity_ns = 1000000000; /* 1 second */
215 #endif
216
217 static ctl_table kern_table[] = {
218 #ifdef CONFIG_SCHED_DEBUG
219 {
220 .ctl_name = CTL_UNNUMBERED,
221 .procname = "sched_granularity_ns",
222 .data = &sysctl_sched_granularity,
223 .maxlen = sizeof(unsigned int),
224 .mode = 0644,
225 .proc_handler = &proc_dointvec_minmax,
226 .strategy = &sysctl_intvec,
227 .extra1 = &min_sched_granularity_ns,
228 .extra2 = &max_sched_granularity_ns,
229 },
230 {
231 .ctl_name = CTL_UNNUMBERED,
232 .procname = "sched_wakeup_granularity_ns",
233 .data = &sysctl_sched_wakeup_granularity,
234 .maxlen = sizeof(unsigned int),
235 .mode = 0644,
236 .proc_handler = &proc_dointvec_minmax,
237 .strategy = &sysctl_intvec,
238 .extra1 = &min_wakeup_granularity_ns,
239 .extra2 = &max_wakeup_granularity_ns,
240 },
241 {
242 .ctl_name = CTL_UNNUMBERED,
243 .procname = "sched_batch_wakeup_granularity_ns",
244 .data = &sysctl_sched_batch_wakeup_granularity,
245 .maxlen = sizeof(unsigned int),
246 .mode = 0644,
247 .proc_handler = &proc_dointvec_minmax,
248 .strategy = &sysctl_intvec,
249 .extra1 = &min_wakeup_granularity_ns,
250 .extra2 = &max_wakeup_granularity_ns,
251 },
252 {
253 .ctl_name = CTL_UNNUMBERED,
254 .procname = "sched_stat_granularity_ns",
255 .data = &sysctl_sched_stat_granularity,
256 .maxlen = sizeof(unsigned int),
257 .mode = 0644,
258 .proc_handler = &proc_dointvec_minmax,
259 .strategy = &sysctl_intvec,
260 .extra1 = &min_wakeup_granularity_ns,
261 .extra2 = &max_wakeup_granularity_ns,
262 },
263 {
264 .ctl_name = CTL_UNNUMBERED,
265 .procname = "sched_runtime_limit_ns",
266 .data = &sysctl_sched_runtime_limit,
267 .maxlen = sizeof(unsigned int),
268 .mode = 0644,
269 .proc_handler = &proc_dointvec_minmax,
270 .strategy = &sysctl_intvec,
271 .extra1 = &min_sched_granularity_ns,
272 .extra2 = &max_sched_granularity_ns,
273 },
274 {
275 .ctl_name = CTL_UNNUMBERED,
276 .procname = "sched_child_runs_first",
277 .data = &sysctl_sched_child_runs_first,
278 .maxlen = sizeof(unsigned int),
279 .mode = 0644,
280 .proc_handler = &proc_dointvec,
281 },
282 {
283 .ctl_name = CTL_UNNUMBERED,
284 .procname = "sched_features",
285 .data = &sysctl_sched_features,
286 .maxlen = sizeof(unsigned int),
287 .mode = 0644,
288 .proc_handler = &proc_dointvec,
289 },
290 #endif
291 {
292 .ctl_name = KERN_PANIC,
293 .procname = "panic",
294 .data = &panic_timeout,
295 .maxlen = sizeof(int),
296 .mode = 0644,
297 .proc_handler = &proc_dointvec,
298 },
299 {
300 .ctl_name = KERN_CORE_USES_PID,
301 .procname = "core_uses_pid",
302 .data = &core_uses_pid,
303 .maxlen = sizeof(int),
304 .mode = 0644,
305 .proc_handler = &proc_dointvec,
306 },
307 {
308 .ctl_name = KERN_CORE_PATTERN,
309 .procname = "core_pattern",
310 .data = core_pattern,
311 .maxlen = CORENAME_MAX_SIZE,
312 .mode = 0644,
313 .proc_handler = &proc_dostring,
314 .strategy = &sysctl_string,
315 },
316 #ifdef CONFIG_PROC_SYSCTL
317 {
318 .ctl_name = KERN_TAINTED,
319 .procname = "tainted",
320 .data = &tainted,
321 .maxlen = sizeof(int),
322 .mode = 0644,
323 .proc_handler = &proc_dointvec_taint,
324 },
325 #endif
326 {
327 .ctl_name = KERN_CAP_BSET,
328 .procname = "cap-bound",
329 .data = &cap_bset,
330 .maxlen = sizeof(kernel_cap_t),
331 .mode = 0600,
332 .proc_handler = &proc_dointvec_bset,
333 },
334 #ifdef CONFIG_BLK_DEV_INITRD
335 {
336 .ctl_name = KERN_REALROOTDEV,
337 .procname = "real-root-dev",
338 .data = &real_root_dev,
339 .maxlen = sizeof(int),
340 .mode = 0644,
341 .proc_handler = &proc_dointvec,
342 },
343 #endif
344 {
345 .ctl_name = CTL_UNNUMBERED,
346 .procname = "print-fatal-signals",
347 .data = &print_fatal_signals,
348 .maxlen = sizeof(int),
349 .mode = 0644,
350 .proc_handler = &proc_dointvec,
351 },
352 #ifdef __sparc__
353 {
354 .ctl_name = KERN_SPARC_REBOOT,
355 .procname = "reboot-cmd",
356 .data = reboot_command,
357 .maxlen = 256,
358 .mode = 0644,
359 .proc_handler = &proc_dostring,
360 .strategy = &sysctl_string,
361 },
362 {
363 .ctl_name = KERN_SPARC_STOP_A,
364 .procname = "stop-a",
365 .data = &stop_a_enabled,
366 .maxlen = sizeof (int),
367 .mode = 0644,
368 .proc_handler = &proc_dointvec,
369 },
370 {
371 .ctl_name = KERN_SPARC_SCONS_PWROFF,
372 .procname = "scons-poweroff",
373 .data = &scons_pwroff,
374 .maxlen = sizeof (int),
375 .mode = 0644,
376 .proc_handler = &proc_dointvec,
377 },
378 #endif
379 #ifdef __hppa__
380 {
381 .ctl_name = KERN_HPPA_PWRSW,
382 .procname = "soft-power",
383 .data = &pwrsw_enabled,
384 .maxlen = sizeof (int),
385 .mode = 0644,
386 .proc_handler = &proc_dointvec,
387 },
388 {
389 .ctl_name = KERN_HPPA_UNALIGNED,
390 .procname = "unaligned-trap",
391 .data = &unaligned_enabled,
392 .maxlen = sizeof (int),
393 .mode = 0644,
394 .proc_handler = &proc_dointvec,
395 },
396 #endif
397 {
398 .ctl_name = KERN_CTLALTDEL,
399 .procname = "ctrl-alt-del",
400 .data = &C_A_D,
401 .maxlen = sizeof(int),
402 .mode = 0644,
403 .proc_handler = &proc_dointvec,
404 },
405 {
406 .ctl_name = KERN_PRINTK,
407 .procname = "printk",
408 .data = &console_loglevel,
409 .maxlen = 4*sizeof(int),
410 .mode = 0644,
411 .proc_handler = &proc_dointvec,
412 },
413 #ifdef CONFIG_KMOD
414 {
415 .ctl_name = KERN_MODPROBE,
416 .procname = "modprobe",
417 .data = &modprobe_path,
418 .maxlen = KMOD_PATH_LEN,
419 .mode = 0644,
420 .proc_handler = &proc_dostring,
421 .strategy = &sysctl_string,
422 },
423 #endif
424 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
425 {
426 .ctl_name = KERN_HOTPLUG,
427 .procname = "hotplug",
428 .data = &uevent_helper,
429 .maxlen = UEVENT_HELPER_PATH_LEN,
430 .mode = 0644,
431 .proc_handler = &proc_dostring,
432 .strategy = &sysctl_string,
433 },
434 #endif
435 #ifdef CONFIG_CHR_DEV_SG
436 {
437 .ctl_name = KERN_SG_BIG_BUFF,
438 .procname = "sg-big-buff",
439 .data = &sg_big_buff,
440 .maxlen = sizeof (int),
441 .mode = 0444,
442 .proc_handler = &proc_dointvec,
443 },
444 #endif
445 #ifdef CONFIG_BSD_PROCESS_ACCT
446 {
447 .ctl_name = KERN_ACCT,
448 .procname = "acct",
449 .data = &acct_parm,
450 .maxlen = 3*sizeof(int),
451 .mode = 0644,
452 .proc_handler = &proc_dointvec,
453 },
454 #endif
455 #ifdef CONFIG_MAGIC_SYSRQ
456 {
457 .ctl_name = KERN_SYSRQ,
458 .procname = "sysrq",
459 .data = &__sysrq_enabled,
460 .maxlen = sizeof (int),
461 .mode = 0644,
462 .proc_handler = &proc_dointvec,
463 },
464 #endif
465 #ifdef CONFIG_PROC_SYSCTL
466 {
467 .ctl_name = KERN_CADPID,
468 .procname = "cad_pid",
469 .data = NULL,
470 .maxlen = sizeof (int),
471 .mode = 0600,
472 .proc_handler = &proc_do_cad_pid,
473 },
474 #endif
475 {
476 .ctl_name = KERN_MAX_THREADS,
477 .procname = "threads-max",
478 .data = &max_threads,
479 .maxlen = sizeof(int),
480 .mode = 0644,
481 .proc_handler = &proc_dointvec,
482 },
483 {
484 .ctl_name = KERN_RANDOM,
485 .procname = "random",
486 .mode = 0555,
487 .child = random_table,
488 },
489 #ifdef CONFIG_UNIX98_PTYS
490 {
491 .ctl_name = KERN_PTY,
492 .procname = "pty",
493 .mode = 0555,
494 .child = pty_table,
495 },
496 #endif
497 {
498 .ctl_name = KERN_OVERFLOWUID,
499 .procname = "overflowuid",
500 .data = &overflowuid,
501 .maxlen = sizeof(int),
502 .mode = 0644,
503 .proc_handler = &proc_dointvec_minmax,
504 .strategy = &sysctl_intvec,
505 .extra1 = &minolduid,
506 .extra2 = &maxolduid,
507 },
508 {
509 .ctl_name = KERN_OVERFLOWGID,
510 .procname = "overflowgid",
511 .data = &overflowgid,
512 .maxlen = sizeof(int),
513 .mode = 0644,
514 .proc_handler = &proc_dointvec_minmax,
515 .strategy = &sysctl_intvec,
516 .extra1 = &minolduid,
517 .extra2 = &maxolduid,
518 },
519 #ifdef CONFIG_S390
520 #ifdef CONFIG_MATHEMU
521 {
522 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
523 .procname = "ieee_emulation_warnings",
524 .data = &sysctl_ieee_emulation_warnings,
525 .maxlen = sizeof(int),
526 .mode = 0644,
527 .proc_handler = &proc_dointvec,
528 },
529 #endif
530 #ifdef CONFIG_NO_IDLE_HZ
531 {
532 .ctl_name = KERN_HZ_TIMER,
533 .procname = "hz_timer",
534 .data = &sysctl_hz_timer,
535 .maxlen = sizeof(int),
536 .mode = 0644,
537 .proc_handler = &proc_dointvec,
538 },
539 #endif
540 {
541 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
542 .procname = "userprocess_debug",
543 .data = &sysctl_userprocess_debug,
544 .maxlen = sizeof(int),
545 .mode = 0644,
546 .proc_handler = &proc_dointvec,
547 },
548 #endif
549 {
550 .ctl_name = KERN_PIDMAX,
551 .procname = "pid_max",
552 .data = &pid_max,
553 .maxlen = sizeof (int),
554 .mode = 0644,
555 .proc_handler = &proc_dointvec_minmax,
556 .strategy = sysctl_intvec,
557 .extra1 = &pid_max_min,
558 .extra2 = &pid_max_max,
559 },
560 {
561 .ctl_name = KERN_PANIC_ON_OOPS,
562 .procname = "panic_on_oops",
563 .data = &panic_on_oops,
564 .maxlen = sizeof(int),
565 .mode = 0644,
566 .proc_handler = &proc_dointvec,
567 },
568 {
569 .ctl_name = KERN_PRINTK_RATELIMIT,
570 .procname = "printk_ratelimit",
571 .data = &printk_ratelimit_jiffies,
572 .maxlen = sizeof(int),
573 .mode = 0644,
574 .proc_handler = &proc_dointvec_jiffies,
575 .strategy = &sysctl_jiffies,
576 },
577 {
578 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
579 .procname = "printk_ratelimit_burst",
580 .data = &printk_ratelimit_burst,
581 .maxlen = sizeof(int),
582 .mode = 0644,
583 .proc_handler = &proc_dointvec,
584 },
585 {
586 .ctl_name = KERN_NGROUPS_MAX,
587 .procname = "ngroups_max",
588 .data = &ngroups_max,
589 .maxlen = sizeof (int),
590 .mode = 0444,
591 .proc_handler = &proc_dointvec,
592 },
593 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
594 {
595 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
596 .procname = "unknown_nmi_panic",
597 .data = &unknown_nmi_panic,
598 .maxlen = sizeof (int),
599 .mode = 0644,
600 .proc_handler = &proc_dointvec,
601 },
602 {
603 .ctl_name = KERN_NMI_WATCHDOG,
604 .procname = "nmi_watchdog",
605 .data = &nmi_watchdog_enabled,
606 .maxlen = sizeof (int),
607 .mode = 0644,
608 .proc_handler = &proc_nmi_enabled,
609 },
610 #endif
611 #if defined(CONFIG_X86)
612 {
613 .ctl_name = KERN_PANIC_ON_NMI,
614 .procname = "panic_on_unrecovered_nmi",
615 .data = &panic_on_unrecovered_nmi,
616 .maxlen = sizeof(int),
617 .mode = 0644,
618 .proc_handler = &proc_dointvec,
619 },
620 {
621 .ctl_name = KERN_BOOTLOADER_TYPE,
622 .procname = "bootloader_type",
623 .data = &bootloader_type,
624 .maxlen = sizeof (int),
625 .mode = 0444,
626 .proc_handler = &proc_dointvec,
627 },
628 {
629 .ctl_name = CTL_UNNUMBERED,
630 .procname = "kstack_depth_to_print",
631 .data = &kstack_depth_to_print,
632 .maxlen = sizeof(int),
633 .mode = 0644,
634 .proc_handler = &proc_dointvec,
635 },
636 #endif
637 #if defined(CONFIG_MMU)
638 {
639 .ctl_name = KERN_RANDOMIZE,
640 .procname = "randomize_va_space",
641 .data = &randomize_va_space,
642 .maxlen = sizeof(int),
643 .mode = 0644,
644 .proc_handler = &proc_dointvec,
645 },
646 #endif
647 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
648 {
649 .ctl_name = KERN_SPIN_RETRY,
650 .procname = "spin_retry",
651 .data = &spin_retry,
652 .maxlen = sizeof (int),
653 .mode = 0644,
654 .proc_handler = &proc_dointvec,
655 },
656 #endif
657 #ifdef CONFIG_ACPI_SLEEP
658 {
659 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
660 .procname = "acpi_video_flags",
661 .data = &acpi_video_flags,
662 .maxlen = sizeof (unsigned long),
663 .mode = 0644,
664 .proc_handler = &proc_doulongvec_minmax,
665 },
666 #endif
667 #ifdef CONFIG_IA64
668 {
669 .ctl_name = KERN_IA64_UNALIGNED,
670 .procname = "ignore-unaligned-usertrap",
671 .data = &no_unaligned_warning,
672 .maxlen = sizeof (int),
673 .mode = 0644,
674 .proc_handler = &proc_dointvec,
675 },
676 #endif
677 #ifdef CONFIG_COMPAT
678 {
679 .ctl_name = KERN_COMPAT_LOG,
680 .procname = "compat-log",
681 .data = &compat_log,
682 .maxlen = sizeof (int),
683 .mode = 0644,
684 .proc_handler = &proc_dointvec,
685 },
686 #endif
687 #ifdef CONFIG_RT_MUTEXES
688 {
689 .ctl_name = KERN_MAX_LOCK_DEPTH,
690 .procname = "max_lock_depth",
691 .data = &max_lock_depth,
692 .maxlen = sizeof(int),
693 .mode = 0644,
694 .proc_handler = &proc_dointvec,
695 },
696 #endif
697 #ifdef CONFIG_PROC_FS
698 {
699 .ctl_name = CTL_UNNUMBERED,
700 .procname = "maps_protect",
701 .data = &maps_protect,
702 .maxlen = sizeof(int),
703 .mode = 0644,
704 .proc_handler = &proc_dointvec,
705 },
706 #endif
707
708 { .ctl_name = 0 }
709 };
710
711 /* Constants for minimum and maximum testing in vm_table.
712 We use these as one-element integer vectors. */
713 static int zero;
714 static int one_hundred = 100;
715
716
717 static ctl_table vm_table[] = {
718 {
719 .ctl_name = VM_OVERCOMMIT_MEMORY,
720 .procname = "overcommit_memory",
721 .data = &sysctl_overcommit_memory,
722 .maxlen = sizeof(sysctl_overcommit_memory),
723 .mode = 0644,
724 .proc_handler = &proc_dointvec,
725 },
726 {
727 .ctl_name = VM_PANIC_ON_OOM,
728 .procname = "panic_on_oom",
729 .data = &sysctl_panic_on_oom,
730 .maxlen = sizeof(sysctl_panic_on_oom),
731 .mode = 0644,
732 .proc_handler = &proc_dointvec,
733 },
734 {
735 .ctl_name = VM_OVERCOMMIT_RATIO,
736 .procname = "overcommit_ratio",
737 .data = &sysctl_overcommit_ratio,
738 .maxlen = sizeof(sysctl_overcommit_ratio),
739 .mode = 0644,
740 .proc_handler = &proc_dointvec,
741 },
742 {
743 .ctl_name = VM_PAGE_CLUSTER,
744 .procname = "page-cluster",
745 .data = &page_cluster,
746 .maxlen = sizeof(int),
747 .mode = 0644,
748 .proc_handler = &proc_dointvec,
749 },
750 {
751 .ctl_name = VM_DIRTY_BACKGROUND,
752 .procname = "dirty_background_ratio",
753 .data = &dirty_background_ratio,
754 .maxlen = sizeof(dirty_background_ratio),
755 .mode = 0644,
756 .proc_handler = &proc_dointvec_minmax,
757 .strategy = &sysctl_intvec,
758 .extra1 = &zero,
759 .extra2 = &one_hundred,
760 },
761 {
762 .ctl_name = VM_DIRTY_RATIO,
763 .procname = "dirty_ratio",
764 .data = &vm_dirty_ratio,
765 .maxlen = sizeof(vm_dirty_ratio),
766 .mode = 0644,
767 .proc_handler = &proc_dointvec_minmax,
768 .strategy = &sysctl_intvec,
769 .extra1 = &zero,
770 .extra2 = &one_hundred,
771 },
772 {
773 .ctl_name = VM_DIRTY_WB_CS,
774 .procname = "dirty_writeback_centisecs",
775 .data = &dirty_writeback_interval,
776 .maxlen = sizeof(dirty_writeback_interval),
777 .mode = 0644,
778 .proc_handler = &dirty_writeback_centisecs_handler,
779 },
780 {
781 .ctl_name = VM_DIRTY_EXPIRE_CS,
782 .procname = "dirty_expire_centisecs",
783 .data = &dirty_expire_interval,
784 .maxlen = sizeof(dirty_expire_interval),
785 .mode = 0644,
786 .proc_handler = &proc_dointvec_userhz_jiffies,
787 },
788 {
789 .ctl_name = VM_NR_PDFLUSH_THREADS,
790 .procname = "nr_pdflush_threads",
791 .data = &nr_pdflush_threads,
792 .maxlen = sizeof nr_pdflush_threads,
793 .mode = 0444 /* read-only*/,
794 .proc_handler = &proc_dointvec,
795 },
796 {
797 .ctl_name = VM_SWAPPINESS,
798 .procname = "swappiness",
799 .data = &vm_swappiness,
800 .maxlen = sizeof(vm_swappiness),
801 .mode = 0644,
802 .proc_handler = &proc_dointvec_minmax,
803 .strategy = &sysctl_intvec,
804 .extra1 = &zero,
805 .extra2 = &one_hundred,
806 },
807 #ifdef CONFIG_HUGETLB_PAGE
808 {
809 .ctl_name = VM_HUGETLB_PAGES,
810 .procname = "nr_hugepages",
811 .data = &max_huge_pages,
812 .maxlen = sizeof(unsigned long),
813 .mode = 0644,
814 .proc_handler = &hugetlb_sysctl_handler,
815 .extra1 = (void *)&hugetlb_zero,
816 .extra2 = (void *)&hugetlb_infinity,
817 },
818 {
819 .ctl_name = VM_HUGETLB_GROUP,
820 .procname = "hugetlb_shm_group",
821 .data = &sysctl_hugetlb_shm_group,
822 .maxlen = sizeof(gid_t),
823 .mode = 0644,
824 .proc_handler = &proc_dointvec,
825 },
826 #endif
827 {
828 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
829 .procname = "lowmem_reserve_ratio",
830 .data = &sysctl_lowmem_reserve_ratio,
831 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
832 .mode = 0644,
833 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
834 .strategy = &sysctl_intvec,
835 },
836 {
837 .ctl_name = VM_DROP_PAGECACHE,
838 .procname = "drop_caches",
839 .data = &sysctl_drop_caches,
840 .maxlen = sizeof(int),
841 .mode = 0644,
842 .proc_handler = drop_caches_sysctl_handler,
843 .strategy = &sysctl_intvec,
844 },
845 {
846 .ctl_name = VM_MIN_FREE_KBYTES,
847 .procname = "min_free_kbytes",
848 .data = &min_free_kbytes,
849 .maxlen = sizeof(min_free_kbytes),
850 .mode = 0644,
851 .proc_handler = &min_free_kbytes_sysctl_handler,
852 .strategy = &sysctl_intvec,
853 .extra1 = &zero,
854 },
855 {
856 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
857 .procname = "percpu_pagelist_fraction",
858 .data = &percpu_pagelist_fraction,
859 .maxlen = sizeof(percpu_pagelist_fraction),
860 .mode = 0644,
861 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
862 .strategy = &sysctl_intvec,
863 .extra1 = &min_percpu_pagelist_fract,
864 },
865 #ifdef CONFIG_MMU
866 {
867 .ctl_name = VM_MAX_MAP_COUNT,
868 .procname = "max_map_count",
869 .data = &sysctl_max_map_count,
870 .maxlen = sizeof(sysctl_max_map_count),
871 .mode = 0644,
872 .proc_handler = &proc_dointvec
873 },
874 #endif
875 {
876 .ctl_name = VM_LAPTOP_MODE,
877 .procname = "laptop_mode",
878 .data = &laptop_mode,
879 .maxlen = sizeof(laptop_mode),
880 .mode = 0644,
881 .proc_handler = &proc_dointvec_jiffies,
882 .strategy = &sysctl_jiffies,
883 },
884 {
885 .ctl_name = VM_BLOCK_DUMP,
886 .procname = "block_dump",
887 .data = &block_dump,
888 .maxlen = sizeof(block_dump),
889 .mode = 0644,
890 .proc_handler = &proc_dointvec,
891 .strategy = &sysctl_intvec,
892 .extra1 = &zero,
893 },
894 {
895 .ctl_name = VM_VFS_CACHE_PRESSURE,
896 .procname = "vfs_cache_pressure",
897 .data = &sysctl_vfs_cache_pressure,
898 .maxlen = sizeof(sysctl_vfs_cache_pressure),
899 .mode = 0644,
900 .proc_handler = &proc_dointvec,
901 .strategy = &sysctl_intvec,
902 .extra1 = &zero,
903 },
904 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
905 {
906 .ctl_name = VM_LEGACY_VA_LAYOUT,
907 .procname = "legacy_va_layout",
908 .data = &sysctl_legacy_va_layout,
909 .maxlen = sizeof(sysctl_legacy_va_layout),
910 .mode = 0644,
911 .proc_handler = &proc_dointvec,
912 .strategy = &sysctl_intvec,
913 .extra1 = &zero,
914 },
915 #endif
916 #ifdef CONFIG_NUMA
917 {
918 .ctl_name = VM_ZONE_RECLAIM_MODE,
919 .procname = "zone_reclaim_mode",
920 .data = &zone_reclaim_mode,
921 .maxlen = sizeof(zone_reclaim_mode),
922 .mode = 0644,
923 .proc_handler = &proc_dointvec,
924 .strategy = &sysctl_intvec,
925 .extra1 = &zero,
926 },
927 {
928 .ctl_name = VM_MIN_UNMAPPED,
929 .procname = "min_unmapped_ratio",
930 .data = &sysctl_min_unmapped_ratio,
931 .maxlen = sizeof(sysctl_min_unmapped_ratio),
932 .mode = 0644,
933 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
934 .strategy = &sysctl_intvec,
935 .extra1 = &zero,
936 .extra2 = &one_hundred,
937 },
938 {
939 .ctl_name = VM_MIN_SLAB,
940 .procname = "min_slab_ratio",
941 .data = &sysctl_min_slab_ratio,
942 .maxlen = sizeof(sysctl_min_slab_ratio),
943 .mode = 0644,
944 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
945 .strategy = &sysctl_intvec,
946 .extra1 = &zero,
947 .extra2 = &one_hundred,
948 },
949 #endif
950 #ifdef CONFIG_SMP
951 {
952 .ctl_name = CTL_UNNUMBERED,
953 .procname = "stat_interval",
954 .data = &sysctl_stat_interval,
955 .maxlen = sizeof(sysctl_stat_interval),
956 .mode = 0644,
957 .proc_handler = &proc_dointvec_jiffies,
958 .strategy = &sysctl_jiffies,
959 },
960 #endif
961 #ifdef CONFIG_SECURITY
962 {
963 .ctl_name = CTL_UNNUMBERED,
964 .procname = "mmap_min_addr",
965 .data = &mmap_min_addr,
966 .maxlen = sizeof(unsigned long),
967 .mode = 0644,
968 .proc_handler = &proc_doulongvec_minmax,
969 },
970 #ifdef CONFIG_NUMA
971 {
972 .ctl_name = CTL_UNNUMBERED,
973 .procname = "numa_zonelist_order",
974 .data = &numa_zonelist_order,
975 .maxlen = NUMA_ZONELIST_ORDER_LEN,
976 .mode = 0644,
977 .proc_handler = &numa_zonelist_order_handler,
978 .strategy = &sysctl_string,
979 },
980 #endif
981 #endif
982 #if defined(CONFIG_X86_32) || \
983 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
984 {
985 .ctl_name = VM_VDSO_ENABLED,
986 .procname = "vdso_enabled",
987 .data = &vdso_enabled,
988 .maxlen = sizeof(vdso_enabled),
989 .mode = 0644,
990 .proc_handler = &proc_dointvec,
991 .strategy = &sysctl_intvec,
992 .extra1 = &zero,
993 },
994 #endif
995 { .ctl_name = 0 }
996 };
997
998 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
999 static ctl_table binfmt_misc_table[] = {
1000 { .ctl_name = 0 }
1001 };
1002 #endif
1003
1004 static ctl_table fs_table[] = {
1005 {
1006 .ctl_name = FS_NRINODE,
1007 .procname = "inode-nr",
1008 .data = &inodes_stat,
1009 .maxlen = 2*sizeof(int),
1010 .mode = 0444,
1011 .proc_handler = &proc_dointvec,
1012 },
1013 {
1014 .ctl_name = FS_STATINODE,
1015 .procname = "inode-state",
1016 .data = &inodes_stat,
1017 .maxlen = 7*sizeof(int),
1018 .mode = 0444,
1019 .proc_handler = &proc_dointvec,
1020 },
1021 {
1022 .ctl_name = FS_NRFILE,
1023 .procname = "file-nr",
1024 .data = &files_stat,
1025 .maxlen = 3*sizeof(int),
1026 .mode = 0444,
1027 .proc_handler = &proc_nr_files,
1028 },
1029 {
1030 .ctl_name = FS_MAXFILE,
1031 .procname = "file-max",
1032 .data = &files_stat.max_files,
1033 .maxlen = sizeof(int),
1034 .mode = 0644,
1035 .proc_handler = &proc_dointvec,
1036 },
1037 {
1038 .ctl_name = FS_DENTRY,
1039 .procname = "dentry-state",
1040 .data = &dentry_stat,
1041 .maxlen = 6*sizeof(int),
1042 .mode = 0444,
1043 .proc_handler = &proc_dointvec,
1044 },
1045 {
1046 .ctl_name = FS_OVERFLOWUID,
1047 .procname = "overflowuid",
1048 .data = &fs_overflowuid,
1049 .maxlen = sizeof(int),
1050 .mode = 0644,
1051 .proc_handler = &proc_dointvec_minmax,
1052 .strategy = &sysctl_intvec,
1053 .extra1 = &minolduid,
1054 .extra2 = &maxolduid,
1055 },
1056 {
1057 .ctl_name = FS_OVERFLOWGID,
1058 .procname = "overflowgid",
1059 .data = &fs_overflowgid,
1060 .maxlen = sizeof(int),
1061 .mode = 0644,
1062 .proc_handler = &proc_dointvec_minmax,
1063 .strategy = &sysctl_intvec,
1064 .extra1 = &minolduid,
1065 .extra2 = &maxolduid,
1066 },
1067 {
1068 .ctl_name = FS_LEASES,
1069 .procname = "leases-enable",
1070 .data = &leases_enable,
1071 .maxlen = sizeof(int),
1072 .mode = 0644,
1073 .proc_handler = &proc_dointvec,
1074 },
1075 #ifdef CONFIG_DNOTIFY
1076 {
1077 .ctl_name = FS_DIR_NOTIFY,
1078 .procname = "dir-notify-enable",
1079 .data = &dir_notify_enable,
1080 .maxlen = sizeof(int),
1081 .mode = 0644,
1082 .proc_handler = &proc_dointvec,
1083 },
1084 #endif
1085 #ifdef CONFIG_MMU
1086 {
1087 .ctl_name = FS_LEASE_TIME,
1088 .procname = "lease-break-time",
1089 .data = &lease_break_time,
1090 .maxlen = sizeof(int),
1091 .mode = 0644,
1092 .proc_handler = &proc_dointvec,
1093 },
1094 {
1095 .ctl_name = FS_AIO_NR,
1096 .procname = "aio-nr",
1097 .data = &aio_nr,
1098 .maxlen = sizeof(aio_nr),
1099 .mode = 0444,
1100 .proc_handler = &proc_doulongvec_minmax,
1101 },
1102 {
1103 .ctl_name = FS_AIO_MAX_NR,
1104 .procname = "aio-max-nr",
1105 .data = &aio_max_nr,
1106 .maxlen = sizeof(aio_max_nr),
1107 .mode = 0644,
1108 .proc_handler = &proc_doulongvec_minmax,
1109 },
1110 #ifdef CONFIG_INOTIFY_USER
1111 {
1112 .ctl_name = FS_INOTIFY,
1113 .procname = "inotify",
1114 .mode = 0555,
1115 .child = inotify_table,
1116 },
1117 #endif
1118 #endif
1119 {
1120 .ctl_name = KERN_SETUID_DUMPABLE,
1121 .procname = "suid_dumpable",
1122 .data = &suid_dumpable,
1123 .maxlen = sizeof(int),
1124 .mode = 0644,
1125 .proc_handler = &proc_dointvec,
1126 },
1127 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1128 {
1129 .ctl_name = CTL_UNNUMBERED,
1130 .procname = "binfmt_misc",
1131 .mode = 0555,
1132 .child = binfmt_misc_table,
1133 },
1134 #endif
1135 { .ctl_name = 0 }
1136 };
1137
1138 static ctl_table debug_table[] = {
1139 { .ctl_name = 0 }
1140 };
1141
1142 static ctl_table dev_table[] = {
1143 { .ctl_name = 0 }
1144 };
1145
1146 static DEFINE_SPINLOCK(sysctl_lock);
1147
1148 /* called under sysctl_lock */
1149 static int use_table(struct ctl_table_header *p)
1150 {
1151 if (unlikely(p->unregistering))
1152 return 0;
1153 p->used++;
1154 return 1;
1155 }
1156
1157 /* called under sysctl_lock */
1158 static void unuse_table(struct ctl_table_header *p)
1159 {
1160 if (!--p->used)
1161 if (unlikely(p->unregistering))
1162 complete(p->unregistering);
1163 }
1164
1165 /* called under sysctl_lock, will reacquire if has to wait */
1166 static void start_unregistering(struct ctl_table_header *p)
1167 {
1168 /*
1169 * if p->used is 0, nobody will ever touch that entry again;
1170 * we'll eliminate all paths to it before dropping sysctl_lock
1171 */
1172 if (unlikely(p->used)) {
1173 struct completion wait;
1174 init_completion(&wait);
1175 p->unregistering = &wait;
1176 spin_unlock(&sysctl_lock);
1177 wait_for_completion(&wait);
1178 spin_lock(&sysctl_lock);
1179 }
1180 /*
1181 * do not remove from the list until nobody holds it; walking the
1182 * list in do_sysctl() relies on that.
1183 */
1184 list_del_init(&p->ctl_entry);
1185 }
1186
1187 void sysctl_head_finish(struct ctl_table_header *head)
1188 {
1189 if (!head)
1190 return;
1191 spin_lock(&sysctl_lock);
1192 unuse_table(head);
1193 spin_unlock(&sysctl_lock);
1194 }
1195
1196 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1197 {
1198 struct ctl_table_header *head;
1199 struct list_head *tmp;
1200 spin_lock(&sysctl_lock);
1201 if (prev) {
1202 tmp = &prev->ctl_entry;
1203 unuse_table(prev);
1204 goto next;
1205 }
1206 tmp = &root_table_header.ctl_entry;
1207 for (;;) {
1208 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1209
1210 if (!use_table(head))
1211 goto next;
1212 spin_unlock(&sysctl_lock);
1213 return head;
1214 next:
1215 tmp = tmp->next;
1216 if (tmp == &root_table_header.ctl_entry)
1217 break;
1218 }
1219 spin_unlock(&sysctl_lock);
1220 return NULL;
1221 }
1222
1223 #ifdef CONFIG_SYSCTL_SYSCALL
1224 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1225 void __user *newval, size_t newlen)
1226 {
1227 struct ctl_table_header *head;
1228 int error = -ENOTDIR;
1229
1230 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1231 return -ENOTDIR;
1232 if (oldval) {
1233 int old_len;
1234 if (!oldlenp || get_user(old_len, oldlenp))
1235 return -EFAULT;
1236 }
1237
1238 for (head = sysctl_head_next(NULL); head;
1239 head = sysctl_head_next(head)) {
1240 error = parse_table(name, nlen, oldval, oldlenp,
1241 newval, newlen, head->ctl_table);
1242 if (error != -ENOTDIR) {
1243 sysctl_head_finish(head);
1244 break;
1245 }
1246 }
1247 return error;
1248 }
1249
1250 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1251 {
1252 struct __sysctl_args tmp;
1253 int error;
1254
1255 if (copy_from_user(&tmp, args, sizeof(tmp)))
1256 return -EFAULT;
1257
1258 lock_kernel();
1259 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1260 tmp.newval, tmp.newlen);
1261 unlock_kernel();
1262 return error;
1263 }
1264 #endif /* CONFIG_SYSCTL_SYSCALL */
1265
1266 /*
1267 * sysctl_perm does NOT grant the superuser all rights automatically, because
1268 * some sysctl variables are readonly even to root.
1269 */
1270
1271 static int test_perm(int mode, int op)
1272 {
1273 if (!current->euid)
1274 mode >>= 6;
1275 else if (in_egroup_p(0))
1276 mode >>= 3;
1277 if ((mode & op & 0007) == op)
1278 return 0;
1279 return -EACCES;
1280 }
1281
1282 int sysctl_perm(ctl_table *table, int op)
1283 {
1284 int error;
1285 error = security_sysctl(table, op);
1286 if (error)
1287 return error;
1288 return test_perm(table->mode, op);
1289 }
1290
1291 #ifdef CONFIG_SYSCTL_SYSCALL
1292 static int parse_table(int __user *name, int nlen,
1293 void __user *oldval, size_t __user *oldlenp,
1294 void __user *newval, size_t newlen,
1295 ctl_table *table)
1296 {
1297 int n;
1298 repeat:
1299 if (!nlen)
1300 return -ENOTDIR;
1301 if (get_user(n, name))
1302 return -EFAULT;
1303 for ( ; table->ctl_name || table->procname; table++) {
1304 if (!table->ctl_name)
1305 continue;
1306 if (n == table->ctl_name) {
1307 int error;
1308 if (table->child) {
1309 if (sysctl_perm(table, 001))
1310 return -EPERM;
1311 name++;
1312 nlen--;
1313 table = table->child;
1314 goto repeat;
1315 }
1316 error = do_sysctl_strategy(table, name, nlen,
1317 oldval, oldlenp,
1318 newval, newlen);
1319 return error;
1320 }
1321 }
1322 return -ENOTDIR;
1323 }
1324
1325 /* Perform the actual read/write of a sysctl table entry. */
1326 int do_sysctl_strategy (ctl_table *table,
1327 int __user *name, int nlen,
1328 void __user *oldval, size_t __user *oldlenp,
1329 void __user *newval, size_t newlen)
1330 {
1331 int op = 0, rc;
1332 size_t len;
1333
1334 if (oldval)
1335 op |= 004;
1336 if (newval)
1337 op |= 002;
1338 if (sysctl_perm(table, op))
1339 return -EPERM;
1340
1341 if (table->strategy) {
1342 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1343 newval, newlen);
1344 if (rc < 0)
1345 return rc;
1346 if (rc > 0)
1347 return 0;
1348 }
1349
1350 /* If there is no strategy routine, or if the strategy returns
1351 * zero, proceed with automatic r/w */
1352 if (table->data && table->maxlen) {
1353 if (oldval && oldlenp) {
1354 if (get_user(len, oldlenp))
1355 return -EFAULT;
1356 if (len) {
1357 if (len > table->maxlen)
1358 len = table->maxlen;
1359 if(copy_to_user(oldval, table->data, len))
1360 return -EFAULT;
1361 if(put_user(len, oldlenp))
1362 return -EFAULT;
1363 }
1364 }
1365 if (newval && newlen) {
1366 len = newlen;
1367 if (len > table->maxlen)
1368 len = table->maxlen;
1369 if(copy_from_user(table->data, newval, len))
1370 return -EFAULT;
1371 }
1372 }
1373 return 0;
1374 }
1375 #endif /* CONFIG_SYSCTL_SYSCALL */
1376
1377 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1378 {
1379 for (; table->ctl_name || table->procname; table++) {
1380 table->parent = parent;
1381 if (table->child)
1382 sysctl_set_parent(table, table->child);
1383 }
1384 }
1385
1386 static __init int sysctl_init(void)
1387 {
1388 sysctl_set_parent(NULL, root_table);
1389 return 0;
1390 }
1391
1392 core_initcall(sysctl_init);
1393
1394 /**
1395 * register_sysctl_table - register a sysctl hierarchy
1396 * @table: the top-level table structure
1397 *
1398 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1399 * array. An entry with a ctl_name of 0 terminates the table.
1400 *
1401 * The members of the &ctl_table structure are used as follows:
1402 *
1403 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1404 * must be unique within that level of sysctl
1405 *
1406 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1407 * enter a sysctl file
1408 *
1409 * data - a pointer to data for use by proc_handler
1410 *
1411 * maxlen - the maximum size in bytes of the data
1412 *
1413 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1414 *
1415 * child - a pointer to the child sysctl table if this entry is a directory, or
1416 * %NULL.
1417 *
1418 * proc_handler - the text handler routine (described below)
1419 *
1420 * strategy - the strategy routine (described below)
1421 *
1422 * de - for internal use by the sysctl routines
1423 *
1424 * extra1, extra2 - extra pointers usable by the proc handler routines
1425 *
1426 * Leaf nodes in the sysctl tree will be represented by a single file
1427 * under /proc; non-leaf nodes will be represented by directories.
1428 *
1429 * sysctl(2) can automatically manage read and write requests through
1430 * the sysctl table. The data and maxlen fields of the ctl_table
1431 * struct enable minimal validation of the values being written to be
1432 * performed, and the mode field allows minimal authentication.
1433 *
1434 * More sophisticated management can be enabled by the provision of a
1435 * strategy routine with the table entry. This will be called before
1436 * any automatic read or write of the data is performed.
1437 *
1438 * The strategy routine may return
1439 *
1440 * < 0 - Error occurred (error is passed to user process)
1441 *
1442 * 0 - OK - proceed with automatic read or write.
1443 *
1444 * > 0 - OK - read or write has been done by the strategy routine, so
1445 * return immediately.
1446 *
1447 * There must be a proc_handler routine for any terminal nodes
1448 * mirrored under /proc/sys (non-terminals are handled by a built-in
1449 * directory handler). Several default handlers are available to
1450 * cover common cases -
1451 *
1452 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1453 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1454 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1455 *
1456 * It is the handler's job to read the input buffer from user memory
1457 * and process it. The handler should return 0 on success.
1458 *
1459 * This routine returns %NULL on a failure to register, and a pointer
1460 * to the table header on success.
1461 */
1462 struct ctl_table_header *register_sysctl_table(ctl_table * table)
1463 {
1464 struct ctl_table_header *tmp;
1465 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1466 if (!tmp)
1467 return NULL;
1468 tmp->ctl_table = table;
1469 INIT_LIST_HEAD(&tmp->ctl_entry);
1470 tmp->used = 0;
1471 tmp->unregistering = NULL;
1472 sysctl_set_parent(NULL, table);
1473 spin_lock(&sysctl_lock);
1474 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
1475 spin_unlock(&sysctl_lock);
1476 return tmp;
1477 }
1478
1479 /**
1480 * unregister_sysctl_table - unregister a sysctl table hierarchy
1481 * @header: the header returned from register_sysctl_table
1482 *
1483 * Unregisters the sysctl table and all children. proc entries may not
1484 * actually be removed until they are no longer used by anyone.
1485 */
1486 void unregister_sysctl_table(struct ctl_table_header * header)
1487 {
1488 might_sleep();
1489 spin_lock(&sysctl_lock);
1490 start_unregistering(header);
1491 spin_unlock(&sysctl_lock);
1492 kfree(header);
1493 }
1494
1495 #else /* !CONFIG_SYSCTL */
1496 struct ctl_table_header *register_sysctl_table(ctl_table * table)
1497 {
1498 return NULL;
1499 }
1500
1501 void unregister_sysctl_table(struct ctl_table_header * table)
1502 {
1503 }
1504
1505 #endif /* CONFIG_SYSCTL */
1506
1507 /*
1508 * /proc/sys support
1509 */
1510
1511 #ifdef CONFIG_PROC_SYSCTL
1512
1513 static int _proc_do_string(void* data, int maxlen, int write,
1514 struct file *filp, void __user *buffer,
1515 size_t *lenp, loff_t *ppos)
1516 {
1517 size_t len;
1518 char __user *p;
1519 char c;
1520
1521 if (!data || !maxlen || !*lenp) {
1522 *lenp = 0;
1523 return 0;
1524 }
1525
1526 if (write) {
1527 len = 0;
1528 p = buffer;
1529 while (len < *lenp) {
1530 if (get_user(c, p++))
1531 return -EFAULT;
1532 if (c == 0 || c == '\n')
1533 break;
1534 len++;
1535 }
1536 if (len >= maxlen)
1537 len = maxlen-1;
1538 if(copy_from_user(data, buffer, len))
1539 return -EFAULT;
1540 ((char *) data)[len] = 0;
1541 *ppos += *lenp;
1542 } else {
1543 len = strlen(data);
1544 if (len > maxlen)
1545 len = maxlen;
1546
1547 if (*ppos > len) {
1548 *lenp = 0;
1549 return 0;
1550 }
1551
1552 data += *ppos;
1553 len -= *ppos;
1554
1555 if (len > *lenp)
1556 len = *lenp;
1557 if (len)
1558 if(copy_to_user(buffer, data, len))
1559 return -EFAULT;
1560 if (len < *lenp) {
1561 if(put_user('\n', ((char __user *) buffer) + len))
1562 return -EFAULT;
1563 len++;
1564 }
1565 *lenp = len;
1566 *ppos += len;
1567 }
1568 return 0;
1569 }
1570
1571 /**
1572 * proc_dostring - read a string sysctl
1573 * @table: the sysctl table
1574 * @write: %TRUE if this is a write to the sysctl file
1575 * @filp: the file structure
1576 * @buffer: the user buffer
1577 * @lenp: the size of the user buffer
1578 * @ppos: file position
1579 *
1580 * Reads/writes a string from/to the user buffer. If the kernel
1581 * buffer provided is not large enough to hold the string, the
1582 * string is truncated. The copied string is %NULL-terminated.
1583 * If the string is being read by the user process, it is copied
1584 * and a newline '\n' is added. It is truncated if the buffer is
1585 * not large enough.
1586 *
1587 * Returns 0 on success.
1588 */
1589 int proc_dostring(ctl_table *table, int write, struct file *filp,
1590 void __user *buffer, size_t *lenp, loff_t *ppos)
1591 {
1592 return _proc_do_string(table->data, table->maxlen, write, filp,
1593 buffer, lenp, ppos);
1594 }
1595
1596
1597 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1598 int *valp,
1599 int write, void *data)
1600 {
1601 if (write) {
1602 *valp = *negp ? -*lvalp : *lvalp;
1603 } else {
1604 int val = *valp;
1605 if (val < 0) {
1606 *negp = -1;
1607 *lvalp = (unsigned long)-val;
1608 } else {
1609 *negp = 0;
1610 *lvalp = (unsigned long)val;
1611 }
1612 }
1613 return 0;
1614 }
1615
1616 static int __do_proc_dointvec(void *tbl_data, ctl_table *table,
1617 int write, struct file *filp, void __user *buffer,
1618 size_t *lenp, loff_t *ppos,
1619 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1620 int write, void *data),
1621 void *data)
1622 {
1623 #define TMPBUFLEN 21
1624 int *i, vleft, first=1, neg, val;
1625 unsigned long lval;
1626 size_t left, len;
1627
1628 char buf[TMPBUFLEN], *p;
1629 char __user *s = buffer;
1630
1631 if (!tbl_data || !table->maxlen || !*lenp ||
1632 (*ppos && !write)) {
1633 *lenp = 0;
1634 return 0;
1635 }
1636
1637 i = (int *) tbl_data;
1638 vleft = table->maxlen / sizeof(*i);
1639 left = *lenp;
1640
1641 if (!conv)
1642 conv = do_proc_dointvec_conv;
1643
1644 for (; left && vleft--; i++, first=0) {
1645 if (write) {
1646 while (left) {
1647 char c;
1648 if (get_user(c, s))
1649 return -EFAULT;
1650 if (!isspace(c))
1651 break;
1652 left--;
1653 s++;
1654 }
1655 if (!left)
1656 break;
1657 neg = 0;
1658 len = left;
1659 if (len > sizeof(buf) - 1)
1660 len = sizeof(buf) - 1;
1661 if (copy_from_user(buf, s, len))
1662 return -EFAULT;
1663 buf[len] = 0;
1664 p = buf;
1665 if (*p == '-' && left > 1) {
1666 neg = 1;
1667 p++;
1668 }
1669 if (*p < '0' || *p > '9')
1670 break;
1671
1672 lval = simple_strtoul(p, &p, 0);
1673
1674 len = p-buf;
1675 if ((len < left) && *p && !isspace(*p))
1676 break;
1677 if (neg)
1678 val = -val;
1679 s += len;
1680 left -= len;
1681
1682 if (conv(&neg, &lval, i, 1, data))
1683 break;
1684 } else {
1685 p = buf;
1686 if (!first)
1687 *p++ = '\t';
1688
1689 if (conv(&neg, &lval, i, 0, data))
1690 break;
1691
1692 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1693 len = strlen(buf);
1694 if (len > left)
1695 len = left;
1696 if(copy_to_user(s, buf, len))
1697 return -EFAULT;
1698 left -= len;
1699 s += len;
1700 }
1701 }
1702
1703 if (!write && !first && left) {
1704 if(put_user('\n', s))
1705 return -EFAULT;
1706 left--, s++;
1707 }
1708 if (write) {
1709 while (left) {
1710 char c;
1711 if (get_user(c, s++))
1712 return -EFAULT;
1713 if (!isspace(c))
1714 break;
1715 left--;
1716 }
1717 }
1718 if (write && first)
1719 return -EINVAL;
1720 *lenp -= left;
1721 *ppos += *lenp;
1722 return 0;
1723 #undef TMPBUFLEN
1724 }
1725
1726 static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1727 void __user *buffer, size_t *lenp, loff_t *ppos,
1728 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1729 int write, void *data),
1730 void *data)
1731 {
1732 return __do_proc_dointvec(table->data, table, write, filp,
1733 buffer, lenp, ppos, conv, data);
1734 }
1735
1736 /**
1737 * proc_dointvec - read a vector of integers
1738 * @table: the sysctl table
1739 * @write: %TRUE if this is a write to the sysctl file
1740 * @filp: the file structure
1741 * @buffer: the user buffer
1742 * @lenp: the size of the user buffer
1743 * @ppos: file position
1744 *
1745 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1746 * values from/to the user buffer, treated as an ASCII string.
1747 *
1748 * Returns 0 on success.
1749 */
1750 int proc_dointvec(ctl_table *table, int write, struct file *filp,
1751 void __user *buffer, size_t *lenp, loff_t *ppos)
1752 {
1753 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1754 NULL,NULL);
1755 }
1756
1757 #define OP_SET 0
1758 #define OP_AND 1
1759 #define OP_OR 2
1760
1761 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1762 int *valp,
1763 int write, void *data)
1764 {
1765 int op = *(int *)data;
1766 if (write) {
1767 int val = *negp ? -*lvalp : *lvalp;
1768 switch(op) {
1769 case OP_SET: *valp = val; break;
1770 case OP_AND: *valp &= val; break;
1771 case OP_OR: *valp |= val; break;
1772 }
1773 } else {
1774 int val = *valp;
1775 if (val < 0) {
1776 *negp = -1;
1777 *lvalp = (unsigned long)-val;
1778 } else {
1779 *negp = 0;
1780 *lvalp = (unsigned long)val;
1781 }
1782 }
1783 return 0;
1784 }
1785
1786 /*
1787 * init may raise the set.
1788 */
1789
1790 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1791 void __user *buffer, size_t *lenp, loff_t *ppos)
1792 {
1793 int op;
1794
1795 if (write && !capable(CAP_SYS_MODULE)) {
1796 return -EPERM;
1797 }
1798
1799 op = is_init(current) ? OP_SET : OP_AND;
1800 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1801 do_proc_dointvec_bset_conv,&op);
1802 }
1803
1804 /*
1805 * Taint values can only be increased
1806 */
1807 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
1808 void __user *buffer, size_t *lenp, loff_t *ppos)
1809 {
1810 int op;
1811
1812 if (write && !capable(CAP_SYS_ADMIN))
1813 return -EPERM;
1814
1815 op = OP_OR;
1816 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1817 do_proc_dointvec_bset_conv,&op);
1818 }
1819
1820 struct do_proc_dointvec_minmax_conv_param {
1821 int *min;
1822 int *max;
1823 };
1824
1825 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1826 int *valp,
1827 int write, void *data)
1828 {
1829 struct do_proc_dointvec_minmax_conv_param *param = data;
1830 if (write) {
1831 int val = *negp ? -*lvalp : *lvalp;
1832 if ((param->min && *param->min > val) ||
1833 (param->max && *param->max < val))
1834 return -EINVAL;
1835 *valp = val;
1836 } else {
1837 int val = *valp;
1838 if (val < 0) {
1839 *negp = -1;
1840 *lvalp = (unsigned long)-val;
1841 } else {
1842 *negp = 0;
1843 *lvalp = (unsigned long)val;
1844 }
1845 }
1846 return 0;
1847 }
1848
1849 /**
1850 * proc_dointvec_minmax - read a vector of integers with min/max values
1851 * @table: the sysctl table
1852 * @write: %TRUE if this is a write to the sysctl file
1853 * @filp: the file structure
1854 * @buffer: the user buffer
1855 * @lenp: the size of the user buffer
1856 * @ppos: file position
1857 *
1858 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1859 * values from/to the user buffer, treated as an ASCII string.
1860 *
1861 * This routine will ensure the values are within the range specified by
1862 * table->extra1 (min) and table->extra2 (max).
1863 *
1864 * Returns 0 on success.
1865 */
1866 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1867 void __user *buffer, size_t *lenp, loff_t *ppos)
1868 {
1869 struct do_proc_dointvec_minmax_conv_param param = {
1870 .min = (int *) table->extra1,
1871 .max = (int *) table->extra2,
1872 };
1873 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1874 do_proc_dointvec_minmax_conv, &param);
1875 }
1876
1877 static int __do_proc_doulongvec_minmax(void *data, ctl_table *table, int write,
1878 struct file *filp,
1879 void __user *buffer,
1880 size_t *lenp, loff_t *ppos,
1881 unsigned long convmul,
1882 unsigned long convdiv)
1883 {
1884 #define TMPBUFLEN 21
1885 unsigned long *i, *min, *max, val;
1886 int vleft, first=1, neg;
1887 size_t len, left;
1888 char buf[TMPBUFLEN], *p;
1889 char __user *s = buffer;
1890
1891 if (!data || !table->maxlen || !*lenp ||
1892 (*ppos && !write)) {
1893 *lenp = 0;
1894 return 0;
1895 }
1896
1897 i = (unsigned long *) data;
1898 min = (unsigned long *) table->extra1;
1899 max = (unsigned long *) table->extra2;
1900 vleft = table->maxlen / sizeof(unsigned long);
1901 left = *lenp;
1902
1903 for (; left && vleft--; i++, min++, max++, first=0) {
1904 if (write) {
1905 while (left) {
1906 char c;
1907 if (get_user(c, s))
1908 return -EFAULT;
1909 if (!isspace(c))
1910 break;
1911 left--;
1912 s++;
1913 }
1914 if (!left)
1915 break;
1916 neg = 0;
1917 len = left;
1918 if (len > TMPBUFLEN-1)
1919 len = TMPBUFLEN-1;
1920 if (copy_from_user(buf, s, len))
1921 return -EFAULT;
1922 buf[len] = 0;
1923 p = buf;
1924 if (*p == '-' && left > 1) {
1925 neg = 1;
1926 p++;
1927 }
1928 if (*p < '0' || *p > '9')
1929 break;
1930 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
1931 len = p-buf;
1932 if ((len < left) && *p && !isspace(*p))
1933 break;
1934 if (neg)
1935 val = -val;
1936 s += len;
1937 left -= len;
1938
1939 if(neg)
1940 continue;
1941 if ((min && val < *min) || (max && val > *max))
1942 continue;
1943 *i = val;
1944 } else {
1945 p = buf;
1946 if (!first)
1947 *p++ = '\t';
1948 sprintf(p, "%lu", convdiv * (*i) / convmul);
1949 len = strlen(buf);
1950 if (len > left)
1951 len = left;
1952 if(copy_to_user(s, buf, len))
1953 return -EFAULT;
1954 left -= len;
1955 s += len;
1956 }
1957 }
1958
1959 if (!write && !first && left) {
1960 if(put_user('\n', s))
1961 return -EFAULT;
1962 left--, s++;
1963 }
1964 if (write) {
1965 while (left) {
1966 char c;
1967 if (get_user(c, s++))
1968 return -EFAULT;
1969 if (!isspace(c))
1970 break;
1971 left--;
1972 }
1973 }
1974 if (write && first)
1975 return -EINVAL;
1976 *lenp -= left;
1977 *ppos += *lenp;
1978 return 0;
1979 #undef TMPBUFLEN
1980 }
1981
1982 static int do_proc_doulongvec_minmax(ctl_table *table, int write,
1983 struct file *filp,
1984 void __user *buffer,
1985 size_t *lenp, loff_t *ppos,
1986 unsigned long convmul,
1987 unsigned long convdiv)
1988 {
1989 return __do_proc_doulongvec_minmax(table->data, table, write,
1990 filp, buffer, lenp, ppos, convmul, convdiv);
1991 }
1992
1993 /**
1994 * proc_doulongvec_minmax - read a vector of long integers with min/max values
1995 * @table: the sysctl table
1996 * @write: %TRUE if this is a write to the sysctl file
1997 * @filp: the file structure
1998 * @buffer: the user buffer
1999 * @lenp: the size of the user buffer
2000 * @ppos: file position
2001 *
2002 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2003 * values from/to the user buffer, treated as an ASCII string.
2004 *
2005 * This routine will ensure the values are within the range specified by
2006 * table->extra1 (min) and table->extra2 (max).
2007 *
2008 * Returns 0 on success.
2009 */
2010 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2011 void __user *buffer, size_t *lenp, loff_t *ppos)
2012 {
2013 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2014 }
2015
2016 /**
2017 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2018 * @table: the sysctl table
2019 * @write: %TRUE if this is a write to the sysctl file
2020 * @filp: the file structure
2021 * @buffer: the user buffer
2022 * @lenp: the size of the user buffer
2023 * @ppos: file position
2024 *
2025 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2026 * values from/to the user buffer, treated as an ASCII string. The values
2027 * are treated as milliseconds, and converted to jiffies when they are stored.
2028 *
2029 * This routine will ensure the values are within the range specified by
2030 * table->extra1 (min) and table->extra2 (max).
2031 *
2032 * Returns 0 on success.
2033 */
2034 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2035 struct file *filp,
2036 void __user *buffer,
2037 size_t *lenp, loff_t *ppos)
2038 {
2039 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2040 lenp, ppos, HZ, 1000l);
2041 }
2042
2043
2044 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2045 int *valp,
2046 int write, void *data)
2047 {
2048 if (write) {
2049 if (*lvalp > LONG_MAX / HZ)
2050 return 1;
2051 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2052 } else {
2053 int val = *valp;
2054 unsigned long lval;
2055 if (val < 0) {
2056 *negp = -1;
2057 lval = (unsigned long)-val;
2058 } else {
2059 *negp = 0;
2060 lval = (unsigned long)val;
2061 }
2062 *lvalp = lval / HZ;
2063 }
2064 return 0;
2065 }
2066
2067 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2068 int *valp,
2069 int write, void *data)
2070 {
2071 if (write) {
2072 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2073 return 1;
2074 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2075 } else {
2076 int val = *valp;
2077 unsigned long lval;
2078 if (val < 0) {
2079 *negp = -1;
2080 lval = (unsigned long)-val;
2081 } else {
2082 *negp = 0;
2083 lval = (unsigned long)val;
2084 }
2085 *lvalp = jiffies_to_clock_t(lval);
2086 }
2087 return 0;
2088 }
2089
2090 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2091 int *valp,
2092 int write, void *data)
2093 {
2094 if (write) {
2095 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2096 } else {
2097 int val = *valp;
2098 unsigned long lval;
2099 if (val < 0) {
2100 *negp = -1;
2101 lval = (unsigned long)-val;
2102 } else {
2103 *negp = 0;
2104 lval = (unsigned long)val;
2105 }
2106 *lvalp = jiffies_to_msecs(lval);
2107 }
2108 return 0;
2109 }
2110
2111 /**
2112 * proc_dointvec_jiffies - read a vector of integers as seconds
2113 * @table: the sysctl table
2114 * @write: %TRUE if this is a write to the sysctl file
2115 * @filp: the file structure
2116 * @buffer: the user buffer
2117 * @lenp: the size of the user buffer
2118 * @ppos: file position
2119 *
2120 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2121 * values from/to the user buffer, treated as an ASCII string.
2122 * The values read are assumed to be in seconds, and are converted into
2123 * jiffies.
2124 *
2125 * Returns 0 on success.
2126 */
2127 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2128 void __user *buffer, size_t *lenp, loff_t *ppos)
2129 {
2130 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2131 do_proc_dointvec_jiffies_conv,NULL);
2132 }
2133
2134 /**
2135 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2136 * @table: the sysctl table
2137 * @write: %TRUE if this is a write to the sysctl file
2138 * @filp: the file structure
2139 * @buffer: the user buffer
2140 * @lenp: the size of the user buffer
2141 * @ppos: pointer to the file position
2142 *
2143 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2144 * values from/to the user buffer, treated as an ASCII string.
2145 * The values read are assumed to be in 1/USER_HZ seconds, and
2146 * are converted into jiffies.
2147 *
2148 * Returns 0 on success.
2149 */
2150 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2151 void __user *buffer, size_t *lenp, loff_t *ppos)
2152 {
2153 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2154 do_proc_dointvec_userhz_jiffies_conv,NULL);
2155 }
2156
2157 /**
2158 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2159 * @table: the sysctl table
2160 * @write: %TRUE if this is a write to the sysctl file
2161 * @filp: the file structure
2162 * @buffer: the user buffer
2163 * @lenp: the size of the user buffer
2164 * @ppos: file position
2165 * @ppos: the current position in the file
2166 *
2167 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2168 * values from/to the user buffer, treated as an ASCII string.
2169 * The values read are assumed to be in 1/1000 seconds, and
2170 * are converted into jiffies.
2171 *
2172 * Returns 0 on success.
2173 */
2174 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2175 void __user *buffer, size_t *lenp, loff_t *ppos)
2176 {
2177 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2178 do_proc_dointvec_ms_jiffies_conv, NULL);
2179 }
2180
2181 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
2182 void __user *buffer, size_t *lenp, loff_t *ppos)
2183 {
2184 struct pid *new_pid;
2185 pid_t tmp;
2186 int r;
2187
2188 tmp = pid_nr(cad_pid);
2189
2190 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2191 lenp, ppos, NULL, NULL);
2192 if (r || !write)
2193 return r;
2194
2195 new_pid = find_get_pid(tmp);
2196 if (!new_pid)
2197 return -ESRCH;
2198
2199 put_pid(xchg(&cad_pid, new_pid));
2200 return 0;
2201 }
2202
2203 #else /* CONFIG_PROC_FS */
2204
2205 int proc_dostring(ctl_table *table, int write, struct file *filp,
2206 void __user *buffer, size_t *lenp, loff_t *ppos)
2207 {
2208 return -ENOSYS;
2209 }
2210
2211 int proc_dointvec(ctl_table *table, int write, struct file *filp,
2212 void __user *buffer, size_t *lenp, loff_t *ppos)
2213 {
2214 return -ENOSYS;
2215 }
2216
2217 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2218 void __user *buffer, size_t *lenp, loff_t *ppos)
2219 {
2220 return -ENOSYS;
2221 }
2222
2223 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2224 void __user *buffer, size_t *lenp, loff_t *ppos)
2225 {
2226 return -ENOSYS;
2227 }
2228
2229 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2230 void __user *buffer, size_t *lenp, loff_t *ppos)
2231 {
2232 return -ENOSYS;
2233 }
2234
2235 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2236 void __user *buffer, size_t *lenp, loff_t *ppos)
2237 {
2238 return -ENOSYS;
2239 }
2240
2241 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2242 void __user *buffer, size_t *lenp, loff_t *ppos)
2243 {
2244 return -ENOSYS;
2245 }
2246
2247 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2248 void __user *buffer, size_t *lenp, loff_t *ppos)
2249 {
2250 return -ENOSYS;
2251 }
2252
2253 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2254 struct file *filp,
2255 void __user *buffer,
2256 size_t *lenp, loff_t *ppos)
2257 {
2258 return -ENOSYS;
2259 }
2260
2261
2262 #endif /* CONFIG_PROC_FS */
2263
2264
2265 #ifdef CONFIG_SYSCTL_SYSCALL
2266 /*
2267 * General sysctl support routines
2268 */
2269
2270 /* The generic string strategy routine: */
2271 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2272 void __user *oldval, size_t __user *oldlenp,
2273 void __user *newval, size_t newlen)
2274 {
2275 if (!table->data || !table->maxlen)
2276 return -ENOTDIR;
2277
2278 if (oldval && oldlenp) {
2279 size_t bufsize;
2280 if (get_user(bufsize, oldlenp))
2281 return -EFAULT;
2282 if (bufsize) {
2283 size_t len = strlen(table->data), copied;
2284
2285 /* This shouldn't trigger for a well-formed sysctl */
2286 if (len > table->maxlen)
2287 len = table->maxlen;
2288
2289 /* Copy up to a max of bufsize-1 bytes of the string */
2290 copied = (len >= bufsize) ? bufsize - 1 : len;
2291
2292 if (copy_to_user(oldval, table->data, copied) ||
2293 put_user(0, (char __user *)(oldval + copied)))
2294 return -EFAULT;
2295 if (put_user(len, oldlenp))
2296 return -EFAULT;
2297 }
2298 }
2299 if (newval && newlen) {
2300 size_t len = newlen;
2301 if (len > table->maxlen)
2302 len = table->maxlen;
2303 if(copy_from_user(table->data, newval, len))
2304 return -EFAULT;
2305 if (len == table->maxlen)
2306 len--;
2307 ((char *) table->data)[len] = 0;
2308 }
2309 return 1;
2310 }
2311
2312 /*
2313 * This function makes sure that all of the integers in the vector
2314 * are between the minimum and maximum values given in the arrays
2315 * table->extra1 and table->extra2, respectively.
2316 */
2317 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2318 void __user *oldval, size_t __user *oldlenp,
2319 void __user *newval, size_t newlen)
2320 {
2321
2322 if (newval && newlen) {
2323 int __user *vec = (int __user *) newval;
2324 int *min = (int *) table->extra1;
2325 int *max = (int *) table->extra2;
2326 size_t length;
2327 int i;
2328
2329 if (newlen % sizeof(int) != 0)
2330 return -EINVAL;
2331
2332 if (!table->extra1 && !table->extra2)
2333 return 0;
2334
2335 if (newlen > table->maxlen)
2336 newlen = table->maxlen;
2337 length = newlen / sizeof(int);
2338
2339 for (i = 0; i < length; i++) {
2340 int value;
2341 if (get_user(value, vec + i))
2342 return -EFAULT;
2343 if (min && value < min[i])
2344 return -EINVAL;
2345 if (max && value > max[i])
2346 return -EINVAL;
2347 }
2348 }
2349 return 0;
2350 }
2351
2352 /* Strategy function to convert jiffies to seconds */
2353 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2354 void __user *oldval, size_t __user *oldlenp,
2355 void __user *newval, size_t newlen)
2356 {
2357 if (oldval && oldlenp) {
2358 size_t olen;
2359
2360 if (get_user(olen, oldlenp))
2361 return -EFAULT;
2362 if (olen) {
2363 int val;
2364
2365 if (olen < sizeof(int))
2366 return -EINVAL;
2367
2368 val = *(int *)(table->data) / HZ;
2369 if (put_user(val, (int __user *)oldval))
2370 return -EFAULT;
2371 if (put_user(sizeof(int), oldlenp))
2372 return -EFAULT;
2373 }
2374 }
2375 if (newval && newlen) {
2376 int new;
2377 if (newlen != sizeof(int))
2378 return -EINVAL;
2379 if (get_user(new, (int __user *)newval))
2380 return -EFAULT;
2381 *(int *)(table->data) = new*HZ;
2382 }
2383 return 1;
2384 }
2385
2386 /* Strategy function to convert jiffies to seconds */
2387 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2388 void __user *oldval, size_t __user *oldlenp,
2389 void __user *newval, size_t newlen)
2390 {
2391 if (oldval && oldlenp) {
2392 size_t olen;
2393
2394 if (get_user(olen, oldlenp))
2395 return -EFAULT;
2396 if (olen) {
2397 int val;
2398
2399 if (olen < sizeof(int))
2400 return -EINVAL;
2401
2402 val = jiffies_to_msecs(*(int *)(table->data));
2403 if (put_user(val, (int __user *)oldval))
2404 return -EFAULT;
2405 if (put_user(sizeof(int), oldlenp))
2406 return -EFAULT;
2407 }
2408 }
2409 if (newval && newlen) {
2410 int new;
2411 if (newlen != sizeof(int))
2412 return -EINVAL;
2413 if (get_user(new, (int __user *)newval))
2414 return -EFAULT;
2415 *(int *)(table->data) = msecs_to_jiffies(new);
2416 }
2417 return 1;
2418 }
2419
2420
2421
2422 #else /* CONFIG_SYSCTL_SYSCALL */
2423
2424
2425 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2426 {
2427 static int msg_count;
2428 struct __sysctl_args tmp;
2429 int name[CTL_MAXNAME];
2430 int i;
2431
2432 /* Read in the sysctl name for better debug message logging */
2433 if (copy_from_user(&tmp, args, sizeof(tmp)))
2434 return -EFAULT;
2435 if (tmp.nlen <= 0 || tmp.nlen >= CTL_MAXNAME)
2436 return -ENOTDIR;
2437 for (i = 0; i < tmp.nlen; i++)
2438 if (get_user(name[i], tmp.name + i))
2439 return -EFAULT;
2440
2441 /* Ignore accesses to kernel.version */
2442 if ((tmp.nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2443 goto out;
2444
2445 if (msg_count < 5) {
2446 msg_count++;
2447 printk(KERN_INFO
2448 "warning: process `%s' used the removed sysctl "
2449 "system call with ", current->comm);
2450 for (i = 0; i < tmp.nlen; i++)
2451 printk("%d.", name[i]);
2452 printk("\n");
2453 }
2454 out:
2455 return -ENOSYS;
2456 }
2457
2458 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2459 void __user *oldval, size_t __user *oldlenp,
2460 void __user *newval, size_t newlen)
2461 {
2462 return -ENOSYS;
2463 }
2464
2465 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2466 void __user *oldval, size_t __user *oldlenp,
2467 void __user *newval, size_t newlen)
2468 {
2469 return -ENOSYS;
2470 }
2471
2472 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2473 void __user *oldval, size_t __user *oldlenp,
2474 void __user *newval, size_t newlen)
2475 {
2476 return -ENOSYS;
2477 }
2478
2479 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2480 void __user *oldval, size_t __user *oldlenp,
2481 void __user *newval, size_t newlen)
2482 {
2483 return -ENOSYS;
2484 }
2485
2486 #endif /* CONFIG_SYSCTL_SYSCALL */
2487
2488 /*
2489 * No sense putting this after each symbol definition, twice,
2490 * exception granted :-)
2491 */
2492 EXPORT_SYMBOL(proc_dointvec);
2493 EXPORT_SYMBOL(proc_dointvec_jiffies);
2494 EXPORT_SYMBOL(proc_dointvec_minmax);
2495 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2496 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2497 EXPORT_SYMBOL(proc_dostring);
2498 EXPORT_SYMBOL(proc_doulongvec_minmax);
2499 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2500 EXPORT_SYMBOL(register_sysctl_table);
2501 EXPORT_SYMBOL(sysctl_intvec);
2502 EXPORT_SYMBOL(sysctl_jiffies);
2503 EXPORT_SYMBOL(sysctl_ms_jiffies);
2504 EXPORT_SYMBOL(sysctl_string);
2505 EXPORT_SYMBOL(unregister_sysctl_table);
linux kernel for loongson