?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
[NET]: Fix csum_start update in pskb_expand_head
[linux-2.6/kmemtrace.git] / kernel / sysctl.c
blobec14aa8ac51fc83b820330af559077e1cc2cf797
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/smp_lock.h>
31 #include <linux/fs.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 #include <linux/reboot.h>
49
50 #include <asm/uaccess.h>
51 #include <asm/processor.h>
52
53 #ifdef CONFIG_X86
54 #include <asm/nmi.h>
55 #include <asm/stacktrace.h>
56 #endif
57
58 #if defined(CONFIG_SYSCTL)
59
60 /* External variables not in a header file. */
61 extern int C_A_D;
62 extern int print_fatal_signals;
63 extern int sysctl_overcommit_memory;
64 extern int sysctl_overcommit_ratio;
65 extern int sysctl_panic_on_oom;
66 extern int max_threads;
67 extern int core_uses_pid;
68 extern int suid_dumpable;
69 extern char core_pattern[];
70 extern int pid_max;
71 extern int min_free_kbytes;
72 extern int printk_ratelimit_jiffies;
73 extern int printk_ratelimit_burst;
74 extern int pid_max_min, pid_max_max;
75 extern int sysctl_drop_caches;
76 extern int percpu_pagelist_fraction;
77 extern int compat_log;
78 extern int maps_protect;
79 extern int sysctl_stat_interval;
80 extern int audit_argv_kb;
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 extern int prove_locking;
164 extern int lock_stat;
165
166 /* The default sysctl tables: */
167
168 static ctl_table root_table[] = {
169 {
170 .ctl_name = CTL_KERN,
171 .procname = "kernel",
172 .mode = 0555,
173 .child = kern_table,
174 },
175 {
176 .ctl_name = CTL_VM,
177 .procname = "vm",
178 .mode = 0555,
179 .child = vm_table,
180 },
181 #ifdef CONFIG_NET
182 {
183 .ctl_name = CTL_NET,
184 .procname = "net",
185 .mode = 0555,
186 .child = net_table,
187 },
188 #endif
189 {
190 .ctl_name = CTL_FS,
191 .procname = "fs",
192 .mode = 0555,
193 .child = fs_table,
194 },
195 {
196 .ctl_name = CTL_DEBUG,
197 .procname = "debug",
198 .mode = 0555,
199 .child = debug_table,
200 },
201 {
202 .ctl_name = CTL_DEV,
203 .procname = "dev",
204 .mode = 0555,
205 .child = dev_table,
206 },
207 /*
208 * NOTE: do not add new entries to this table unless you have read
209 * Documentation/sysctl/ctl_unnumbered.txt
210 */
211 { .ctl_name = 0 }
212 };
213
214 #ifdef CONFIG_SCHED_DEBUG
215 static unsigned long min_sched_granularity_ns = 100000; /* 100 usecs */
216 static unsigned long max_sched_granularity_ns = 1000000000; /* 1 second */
217 static unsigned long min_wakeup_granularity_ns; /* 0 usecs */
218 static unsigned long max_wakeup_granularity_ns = 1000000000; /* 1 second */
219 #endif
220
221 static ctl_table kern_table[] = {
222 #ifdef CONFIG_SCHED_DEBUG
223 {
224 .ctl_name = CTL_UNNUMBERED,
225 .procname = "sched_nr_latency",
226 .data = &sysctl_sched_nr_latency,
227 .maxlen = sizeof(unsigned int),
228 .mode = 0644,
229 .proc_handler = &proc_dointvec,
230 },
231 {
232 .ctl_name = CTL_UNNUMBERED,
233 .procname = "sched_latency_ns",
234 .data = &sysctl_sched_latency,
235 .maxlen = sizeof(unsigned int),
236 .mode = 0644,
237 .proc_handler = &proc_dointvec_minmax,
238 .strategy = &sysctl_intvec,
239 .extra1 = &min_sched_granularity_ns,
240 .extra2 = &max_sched_granularity_ns,
241 },
242 {
243 .ctl_name = CTL_UNNUMBERED,
244 .procname = "sched_wakeup_granularity_ns",
245 .data = &sysctl_sched_wakeup_granularity,
246 .maxlen = sizeof(unsigned int),
247 .mode = 0644,
248 .proc_handler = &proc_dointvec_minmax,
249 .strategy = &sysctl_intvec,
250 .extra1 = &min_wakeup_granularity_ns,
251 .extra2 = &max_wakeup_granularity_ns,
252 },
253 {
254 .ctl_name = CTL_UNNUMBERED,
255 .procname = "sched_batch_wakeup_granularity_ns",
256 .data = &sysctl_sched_batch_wakeup_granularity,
257 .maxlen = sizeof(unsigned int),
258 .mode = 0644,
259 .proc_handler = &proc_dointvec_minmax,
260 .strategy = &sysctl_intvec,
261 .extra1 = &min_wakeup_granularity_ns,
262 .extra2 = &max_wakeup_granularity_ns,
263 },
264 {
265 .ctl_name = CTL_UNNUMBERED,
266 .procname = "sched_child_runs_first",
267 .data = &sysctl_sched_child_runs_first,
268 .maxlen = sizeof(unsigned int),
269 .mode = 0644,
270 .proc_handler = &proc_dointvec,
271 },
272 {
273 .ctl_name = CTL_UNNUMBERED,
274 .procname = "sched_features",
275 .data = &sysctl_sched_features,
276 .maxlen = sizeof(unsigned int),
277 .mode = 0644,
278 .proc_handler = &proc_dointvec,
279 },
280 {
281 .ctl_name = CTL_UNNUMBERED,
282 .procname = "sched_migration_cost",
283 .data = &sysctl_sched_migration_cost,
284 .maxlen = sizeof(unsigned int),
285 .mode = 0644,
286 .proc_handler = &proc_dointvec,
287 },
288 #endif
289 {
290 .ctl_name = CTL_UNNUMBERED,
291 .procname = "sched_compat_yield",
292 .data = &sysctl_sched_compat_yield,
293 .maxlen = sizeof(unsigned int),
294 .mode = 0644,
295 .proc_handler = &proc_dointvec,
296 },
297 #ifdef CONFIG_PROVE_LOCKING
298 {
299 .ctl_name = CTL_UNNUMBERED,
300 .procname = "prove_locking",
301 .data = &prove_locking,
302 .maxlen = sizeof(int),
303 .mode = 0644,
304 .proc_handler = &proc_dointvec,
305 },
306 #endif
307 #ifdef CONFIG_LOCK_STAT
308 {
309 .ctl_name = CTL_UNNUMBERED,
310 .procname = "lock_stat",
311 .data = &lock_stat,
312 .maxlen = sizeof(int),
313 .mode = 0644,
314 .proc_handler = &proc_dointvec,
315 },
316 #endif
317 {
318 .ctl_name = KERN_PANIC,
319 .procname = "panic",
320 .data = &panic_timeout,
321 .maxlen = sizeof(int),
322 .mode = 0644,
323 .proc_handler = &proc_dointvec,
324 },
325 {
326 .ctl_name = KERN_CORE_USES_PID,
327 .procname = "core_uses_pid",
328 .data = &core_uses_pid,
329 .maxlen = sizeof(int),
330 .mode = 0644,
331 .proc_handler = &proc_dointvec,
332 },
333 #ifdef CONFIG_AUDITSYSCALL
334 {
335 .ctl_name = CTL_UNNUMBERED,
336 .procname = "audit_argv_kb",
337 .data = &audit_argv_kb,
338 .maxlen = sizeof(int),
339 .mode = 0644,
340 .proc_handler = &proc_dointvec,
341 },
342 #endif
343 {
344 .ctl_name = KERN_CORE_PATTERN,
345 .procname = "core_pattern",
346 .data = core_pattern,
347 .maxlen = CORENAME_MAX_SIZE,
348 .mode = 0644,
349 .proc_handler = &proc_dostring,
350 .strategy = &sysctl_string,
351 },
352 #ifdef CONFIG_PROC_SYSCTL
353 {
354 .ctl_name = KERN_TAINTED,
355 .procname = "tainted",
356 .data = &tainted,
357 .maxlen = sizeof(int),
358 .mode = 0644,
359 .proc_handler = &proc_dointvec_taint,
360 },
361 #endif
362 {
363 .ctl_name = KERN_CAP_BSET,
364 .procname = "cap-bound",
365 .data = &cap_bset,
366 .maxlen = sizeof(kernel_cap_t),
367 .mode = 0600,
368 .proc_handler = &proc_dointvec_bset,
369 },
370 #ifdef CONFIG_BLK_DEV_INITRD
371 {
372 .ctl_name = KERN_REALROOTDEV,
373 .procname = "real-root-dev",
374 .data = &real_root_dev,
375 .maxlen = sizeof(int),
376 .mode = 0644,
377 .proc_handler = &proc_dointvec,
378 },
379 #endif
380 {
381 .ctl_name = CTL_UNNUMBERED,
382 .procname = "print-fatal-signals",
383 .data = &print_fatal_signals,
384 .maxlen = sizeof(int),
385 .mode = 0644,
386 .proc_handler = &proc_dointvec,
387 },
388 #ifdef __sparc__
389 {
390 .ctl_name = KERN_SPARC_REBOOT,
391 .procname = "reboot-cmd",
392 .data = reboot_command,
393 .maxlen = 256,
394 .mode = 0644,
395 .proc_handler = &proc_dostring,
396 .strategy = &sysctl_string,
397 },
398 {
399 .ctl_name = KERN_SPARC_STOP_A,
400 .procname = "stop-a",
401 .data = &stop_a_enabled,
402 .maxlen = sizeof (int),
403 .mode = 0644,
404 .proc_handler = &proc_dointvec,
405 },
406 {
407 .ctl_name = KERN_SPARC_SCONS_PWROFF,
408 .procname = "scons-poweroff",
409 .data = &scons_pwroff,
410 .maxlen = sizeof (int),
411 .mode = 0644,
412 .proc_handler = &proc_dointvec,
413 },
414 #endif
415 #ifdef __hppa__
416 {
417 .ctl_name = KERN_HPPA_PWRSW,
418 .procname = "soft-power",
419 .data = &pwrsw_enabled,
420 .maxlen = sizeof (int),
421 .mode = 0644,
422 .proc_handler = &proc_dointvec,
423 },
424 {
425 .ctl_name = KERN_HPPA_UNALIGNED,
426 .procname = "unaligned-trap",
427 .data = &unaligned_enabled,
428 .maxlen = sizeof (int),
429 .mode = 0644,
430 .proc_handler = &proc_dointvec,
431 },
432 #endif
433 {
434 .ctl_name = KERN_CTLALTDEL,
435 .procname = "ctrl-alt-del",
436 .data = &C_A_D,
437 .maxlen = sizeof(int),
438 .mode = 0644,
439 .proc_handler = &proc_dointvec,
440 },
441 {
442 .ctl_name = KERN_PRINTK,
443 .procname = "printk",
444 .data = &console_loglevel,
445 .maxlen = 4*sizeof(int),
446 .mode = 0644,
447 .proc_handler = &proc_dointvec,
448 },
449 #ifdef CONFIG_KMOD
450 {
451 .ctl_name = KERN_MODPROBE,
452 .procname = "modprobe",
453 .data = &modprobe_path,
454 .maxlen = KMOD_PATH_LEN,
455 .mode = 0644,
456 .proc_handler = &proc_dostring,
457 .strategy = &sysctl_string,
458 },
459 #endif
460 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
461 {
462 .ctl_name = KERN_HOTPLUG,
463 .procname = "hotplug",
464 .data = &uevent_helper,
465 .maxlen = UEVENT_HELPER_PATH_LEN,
466 .mode = 0644,
467 .proc_handler = &proc_dostring,
468 .strategy = &sysctl_string,
469 },
470 #endif
471 #ifdef CONFIG_CHR_DEV_SG
472 {
473 .ctl_name = KERN_SG_BIG_BUFF,
474 .procname = "sg-big-buff",
475 .data = &sg_big_buff,
476 .maxlen = sizeof (int),
477 .mode = 0444,
478 .proc_handler = &proc_dointvec,
479 },
480 #endif
481 #ifdef CONFIG_BSD_PROCESS_ACCT
482 {
483 .ctl_name = KERN_ACCT,
484 .procname = "acct",
485 .data = &acct_parm,
486 .maxlen = 3*sizeof(int),
487 .mode = 0644,
488 .proc_handler = &proc_dointvec,
489 },
490 #endif
491 #ifdef CONFIG_MAGIC_SYSRQ
492 {
493 .ctl_name = KERN_SYSRQ,
494 .procname = "sysrq",
495 .data = &__sysrq_enabled,
496 .maxlen = sizeof (int),
497 .mode = 0644,
498 .proc_handler = &proc_dointvec,
499 },
500 #endif
501 #ifdef CONFIG_PROC_SYSCTL
502 {
503 .ctl_name = KERN_CADPID,
504 .procname = "cad_pid",
505 .data = NULL,
506 .maxlen = sizeof (int),
507 .mode = 0600,
508 .proc_handler = &proc_do_cad_pid,
509 },
510 #endif
511 {
512 .ctl_name = KERN_MAX_THREADS,
513 .procname = "threads-max",
514 .data = &max_threads,
515 .maxlen = sizeof(int),
516 .mode = 0644,
517 .proc_handler = &proc_dointvec,
518 },
519 {
520 .ctl_name = KERN_RANDOM,
521 .procname = "random",
522 .mode = 0555,
523 .child = random_table,
524 },
525 #ifdef CONFIG_UNIX98_PTYS
526 {
527 .ctl_name = KERN_PTY,
528 .procname = "pty",
529 .mode = 0555,
530 .child = pty_table,
531 },
532 #endif
533 {
534 .ctl_name = KERN_OVERFLOWUID,
535 .procname = "overflowuid",
536 .data = &overflowuid,
537 .maxlen = sizeof(int),
538 .mode = 0644,
539 .proc_handler = &proc_dointvec_minmax,
540 .strategy = &sysctl_intvec,
541 .extra1 = &minolduid,
542 .extra2 = &maxolduid,
543 },
544 {
545 .ctl_name = KERN_OVERFLOWGID,
546 .procname = "overflowgid",
547 .data = &overflowgid,
548 .maxlen = sizeof(int),
549 .mode = 0644,
550 .proc_handler = &proc_dointvec_minmax,
551 .strategy = &sysctl_intvec,
552 .extra1 = &minolduid,
553 .extra2 = &maxolduid,
554 },
555 #ifdef CONFIG_S390
556 #ifdef CONFIG_MATHEMU
557 {
558 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
559 .procname = "ieee_emulation_warnings",
560 .data = &sysctl_ieee_emulation_warnings,
561 .maxlen = sizeof(int),
562 .mode = 0644,
563 .proc_handler = &proc_dointvec,
564 },
565 #endif
566 #ifdef CONFIG_NO_IDLE_HZ
567 {
568 .ctl_name = KERN_HZ_TIMER,
569 .procname = "hz_timer",
570 .data = &sysctl_hz_timer,
571 .maxlen = sizeof(int),
572 .mode = 0644,
573 .proc_handler = &proc_dointvec,
574 },
575 #endif
576 {
577 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
578 .procname = "userprocess_debug",
579 .data = &sysctl_userprocess_debug,
580 .maxlen = sizeof(int),
581 .mode = 0644,
582 .proc_handler = &proc_dointvec,
583 },
584 #endif
585 {
586 .ctl_name = KERN_PIDMAX,
587 .procname = "pid_max",
588 .data = &pid_max,
589 .maxlen = sizeof (int),
590 .mode = 0644,
591 .proc_handler = &proc_dointvec_minmax,
592 .strategy = sysctl_intvec,
593 .extra1 = &pid_max_min,
594 .extra2 = &pid_max_max,
595 },
596 {
597 .ctl_name = KERN_PANIC_ON_OOPS,
598 .procname = "panic_on_oops",
599 .data = &panic_on_oops,
600 .maxlen = sizeof(int),
601 .mode = 0644,
602 .proc_handler = &proc_dointvec,
603 },
604 {
605 .ctl_name = KERN_PRINTK_RATELIMIT,
606 .procname = "printk_ratelimit",
607 .data = &printk_ratelimit_jiffies,
608 .maxlen = sizeof(int),
609 .mode = 0644,
610 .proc_handler = &proc_dointvec_jiffies,
611 .strategy = &sysctl_jiffies,
612 },
613 {
614 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
615 .procname = "printk_ratelimit_burst",
616 .data = &printk_ratelimit_burst,
617 .maxlen = sizeof(int),
618 .mode = 0644,
619 .proc_handler = &proc_dointvec,
620 },
621 {
622 .ctl_name = KERN_NGROUPS_MAX,
623 .procname = "ngroups_max",
624 .data = &ngroups_max,
625 .maxlen = sizeof (int),
626 .mode = 0444,
627 .proc_handler = &proc_dointvec,
628 },
629 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
630 {
631 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
632 .procname = "unknown_nmi_panic",
633 .data = &unknown_nmi_panic,
634 .maxlen = sizeof (int),
635 .mode = 0644,
636 .proc_handler = &proc_dointvec,
637 },
638 {
639 .ctl_name = KERN_NMI_WATCHDOG,
640 .procname = "nmi_watchdog",
641 .data = &nmi_watchdog_enabled,
642 .maxlen = sizeof (int),
643 .mode = 0644,
644 .proc_handler = &proc_nmi_enabled,
645 },
646 #endif
647 #if defined(CONFIG_X86)
648 {
649 .ctl_name = KERN_PANIC_ON_NMI,
650 .procname = "panic_on_unrecovered_nmi",
651 .data = &panic_on_unrecovered_nmi,
652 .maxlen = sizeof(int),
653 .mode = 0644,
654 .proc_handler = &proc_dointvec,
655 },
656 {
657 .ctl_name = KERN_BOOTLOADER_TYPE,
658 .procname = "bootloader_type",
659 .data = &bootloader_type,
660 .maxlen = sizeof (int),
661 .mode = 0444,
662 .proc_handler = &proc_dointvec,
663 },
664 {
665 .ctl_name = CTL_UNNUMBERED,
666 .procname = "kstack_depth_to_print",
667 .data = &kstack_depth_to_print,
668 .maxlen = sizeof(int),
669 .mode = 0644,
670 .proc_handler = &proc_dointvec,
671 },
672 #endif
673 #if defined(CONFIG_MMU)
674 {
675 .ctl_name = KERN_RANDOMIZE,
676 .procname = "randomize_va_space",
677 .data = &randomize_va_space,
678 .maxlen = sizeof(int),
679 .mode = 0644,
680 .proc_handler = &proc_dointvec,
681 },
682 #endif
683 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
684 {
685 .ctl_name = KERN_SPIN_RETRY,
686 .procname = "spin_retry",
687 .data = &spin_retry,
688 .maxlen = sizeof (int),
689 .mode = 0644,
690 .proc_handler = &proc_dointvec,
691 },
692 #endif
693 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
694 {
695 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
696 .procname = "acpi_video_flags",
697 .data = &acpi_realmode_flags,
698 .maxlen = sizeof (unsigned long),
699 .mode = 0644,
700 .proc_handler = &proc_doulongvec_minmax,
701 },
702 #endif
703 #ifdef CONFIG_IA64
704 {
705 .ctl_name = KERN_IA64_UNALIGNED,
706 .procname = "ignore-unaligned-usertrap",
707 .data = &no_unaligned_warning,
708 .maxlen = sizeof (int),
709 .mode = 0644,
710 .proc_handler = &proc_dointvec,
711 },
712 #endif
713 #ifdef CONFIG_COMPAT
714 {
715 .ctl_name = KERN_COMPAT_LOG,
716 .procname = "compat-log",
717 .data = &compat_log,
718 .maxlen = sizeof (int),
719 .mode = 0644,
720 .proc_handler = &proc_dointvec,
721 },
722 #endif
723 #ifdef CONFIG_RT_MUTEXES
724 {
725 .ctl_name = KERN_MAX_LOCK_DEPTH,
726 .procname = "max_lock_depth",
727 .data = &max_lock_depth,
728 .maxlen = sizeof(int),
729 .mode = 0644,
730 .proc_handler = &proc_dointvec,
731 },
732 #endif
733 #ifdef CONFIG_PROC_FS
734 {
735 .ctl_name = CTL_UNNUMBERED,
736 .procname = "maps_protect",
737 .data = &maps_protect,
738 .maxlen = sizeof(int),
739 .mode = 0644,
740 .proc_handler = &proc_dointvec,
741 },
742 #endif
743 {
744 .ctl_name = CTL_UNNUMBERED,
745 .procname = "poweroff_cmd",
746 .data = &poweroff_cmd,
747 .maxlen = POWEROFF_CMD_PATH_LEN,
748 .mode = 0644,
749 .proc_handler = &proc_dostring,
750 .strategy = &sysctl_string,
751 },
752 /*
753 * NOTE: do not add new entries to this table unless you have read
754 * Documentation/sysctl/ctl_unnumbered.txt
755 */
756 { .ctl_name = 0 }
757 };
758
759 /* Constants for minimum and maximum testing in vm_table.
760 We use these as one-element integer vectors. */
761 static int zero;
762 static int two = 2;
763 static int one_hundred = 100;
764
765
766 static ctl_table vm_table[] = {
767 {
768 .ctl_name = VM_OVERCOMMIT_MEMORY,
769 .procname = "overcommit_memory",
770 .data = &sysctl_overcommit_memory,
771 .maxlen = sizeof(sysctl_overcommit_memory),
772 .mode = 0644,
773 .proc_handler = &proc_dointvec,
774 },
775 {
776 .ctl_name = VM_PANIC_ON_OOM,
777 .procname = "panic_on_oom",
778 .data = &sysctl_panic_on_oom,
779 .maxlen = sizeof(sysctl_panic_on_oom),
780 .mode = 0644,
781 .proc_handler = &proc_dointvec,
782 },
783 {
784 .ctl_name = VM_OVERCOMMIT_RATIO,
785 .procname = "overcommit_ratio",
786 .data = &sysctl_overcommit_ratio,
787 .maxlen = sizeof(sysctl_overcommit_ratio),
788 .mode = 0644,
789 .proc_handler = &proc_dointvec,
790 },
791 {
792 .ctl_name = VM_PAGE_CLUSTER,
793 .procname = "page-cluster",
794 .data = &page_cluster,
795 .maxlen = sizeof(int),
796 .mode = 0644,
797 .proc_handler = &proc_dointvec,
798 },
799 {
800 .ctl_name = VM_DIRTY_BACKGROUND,
801 .procname = "dirty_background_ratio",
802 .data = &dirty_background_ratio,
803 .maxlen = sizeof(dirty_background_ratio),
804 .mode = 0644,
805 .proc_handler = &proc_dointvec_minmax,
806 .strategy = &sysctl_intvec,
807 .extra1 = &zero,
808 .extra2 = &one_hundred,
809 },
810 {
811 .ctl_name = VM_DIRTY_RATIO,
812 .procname = "dirty_ratio",
813 .data = &vm_dirty_ratio,
814 .maxlen = sizeof(vm_dirty_ratio),
815 .mode = 0644,
816 .proc_handler = &proc_dointvec_minmax,
817 .strategy = &sysctl_intvec,
818 .extra1 = &zero,
819 .extra2 = &one_hundred,
820 },
821 {
822 .ctl_name = VM_DIRTY_WB_CS,
823 .procname = "dirty_writeback_centisecs",
824 .data = &dirty_writeback_interval,
825 .maxlen = sizeof(dirty_writeback_interval),
826 .mode = 0644,
827 .proc_handler = &dirty_writeback_centisecs_handler,
828 },
829 {
830 .ctl_name = VM_DIRTY_EXPIRE_CS,
831 .procname = "dirty_expire_centisecs",
832 .data = &dirty_expire_interval,
833 .maxlen = sizeof(dirty_expire_interval),
834 .mode = 0644,
835 .proc_handler = &proc_dointvec_userhz_jiffies,
836 },
837 {
838 .ctl_name = VM_NR_PDFLUSH_THREADS,
839 .procname = "nr_pdflush_threads",
840 .data = &nr_pdflush_threads,
841 .maxlen = sizeof nr_pdflush_threads,
842 .mode = 0444 /* read-only*/,
843 .proc_handler = &proc_dointvec,
844 },
845 {
846 .ctl_name = VM_SWAPPINESS,
847 .procname = "swappiness",
848 .data = &vm_swappiness,
849 .maxlen = sizeof(vm_swappiness),
850 .mode = 0644,
851 .proc_handler = &proc_dointvec_minmax,
852 .strategy = &sysctl_intvec,
853 .extra1 = &zero,
854 .extra2 = &one_hundred,
855 },
856 #ifdef CONFIG_HUGETLB_PAGE
857 {
858 .ctl_name = VM_HUGETLB_PAGES,
859 .procname = "nr_hugepages",
860 .data = &max_huge_pages,
861 .maxlen = sizeof(unsigned long),
862 .mode = 0644,
863 .proc_handler = &hugetlb_sysctl_handler,
864 .extra1 = (void *)&hugetlb_zero,
865 .extra2 = (void *)&hugetlb_infinity,
866 },
867 {
868 .ctl_name = VM_HUGETLB_GROUP,
869 .procname = "hugetlb_shm_group",
870 .data = &sysctl_hugetlb_shm_group,
871 .maxlen = sizeof(gid_t),
872 .mode = 0644,
873 .proc_handler = &proc_dointvec,
874 },
875 {
876 .ctl_name = CTL_UNNUMBERED,
877 .procname = "hugepages_treat_as_movable",
878 .data = &hugepages_treat_as_movable,
879 .maxlen = sizeof(int),
880 .mode = 0644,
881 .proc_handler = &hugetlb_treat_movable_handler,
882 },
883 #endif
884 {
885 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
886 .procname = "lowmem_reserve_ratio",
887 .data = &sysctl_lowmem_reserve_ratio,
888 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
889 .mode = 0644,
890 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
891 .strategy = &sysctl_intvec,
892 },
893 {
894 .ctl_name = VM_DROP_PAGECACHE,
895 .procname = "drop_caches",
896 .data = &sysctl_drop_caches,
897 .maxlen = sizeof(int),
898 .mode = 0644,
899 .proc_handler = drop_caches_sysctl_handler,
900 .strategy = &sysctl_intvec,
901 },
902 {
903 .ctl_name = VM_MIN_FREE_KBYTES,
904 .procname = "min_free_kbytes",
905 .data = &min_free_kbytes,
906 .maxlen = sizeof(min_free_kbytes),
907 .mode = 0644,
908 .proc_handler = &min_free_kbytes_sysctl_handler,
909 .strategy = &sysctl_intvec,
910 .extra1 = &zero,
911 },
912 {
913 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
914 .procname = "percpu_pagelist_fraction",
915 .data = &percpu_pagelist_fraction,
916 .maxlen = sizeof(percpu_pagelist_fraction),
917 .mode = 0644,
918 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
919 .strategy = &sysctl_intvec,
920 .extra1 = &min_percpu_pagelist_fract,
921 },
922 #ifdef CONFIG_MMU
923 {
924 .ctl_name = VM_MAX_MAP_COUNT,
925 .procname = "max_map_count",
926 .data = &sysctl_max_map_count,
927 .maxlen = sizeof(sysctl_max_map_count),
928 .mode = 0644,
929 .proc_handler = &proc_dointvec
930 },
931 #endif
932 {
933 .ctl_name = VM_LAPTOP_MODE,
934 .procname = "laptop_mode",
935 .data = &laptop_mode,
936 .maxlen = sizeof(laptop_mode),
937 .mode = 0644,
938 .proc_handler = &proc_dointvec_jiffies,
939 .strategy = &sysctl_jiffies,
940 },
941 {
942 .ctl_name = VM_BLOCK_DUMP,
943 .procname = "block_dump",
944 .data = &block_dump,
945 .maxlen = sizeof(block_dump),
946 .mode = 0644,
947 .proc_handler = &proc_dointvec,
948 .strategy = &sysctl_intvec,
949 .extra1 = &zero,
950 },
951 {
952 .ctl_name = VM_VFS_CACHE_PRESSURE,
953 .procname = "vfs_cache_pressure",
954 .data = &sysctl_vfs_cache_pressure,
955 .maxlen = sizeof(sysctl_vfs_cache_pressure),
956 .mode = 0644,
957 .proc_handler = &proc_dointvec,
958 .strategy = &sysctl_intvec,
959 .extra1 = &zero,
960 },
961 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
962 {
963 .ctl_name = VM_LEGACY_VA_LAYOUT,
964 .procname = "legacy_va_layout",
965 .data = &sysctl_legacy_va_layout,
966 .maxlen = sizeof(sysctl_legacy_va_layout),
967 .mode = 0644,
968 .proc_handler = &proc_dointvec,
969 .strategy = &sysctl_intvec,
970 .extra1 = &zero,
971 },
972 #endif
973 #ifdef CONFIG_NUMA
974 {
975 .ctl_name = VM_ZONE_RECLAIM_MODE,
976 .procname = "zone_reclaim_mode",
977 .data = &zone_reclaim_mode,
978 .maxlen = sizeof(zone_reclaim_mode),
979 .mode = 0644,
980 .proc_handler = &proc_dointvec,
981 .strategy = &sysctl_intvec,
982 .extra1 = &zero,
983 },
984 {
985 .ctl_name = VM_MIN_UNMAPPED,
986 .procname = "min_unmapped_ratio",
987 .data = &sysctl_min_unmapped_ratio,
988 .maxlen = sizeof(sysctl_min_unmapped_ratio),
989 .mode = 0644,
990 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
991 .strategy = &sysctl_intvec,
992 .extra1 = &zero,
993 .extra2 = &one_hundred,
994 },
995 {
996 .ctl_name = VM_MIN_SLAB,
997 .procname = "min_slab_ratio",
998 .data = &sysctl_min_slab_ratio,
999 .maxlen = sizeof(sysctl_min_slab_ratio),
1000 .mode = 0644,
1001 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1002 .strategy = &sysctl_intvec,
1003 .extra1 = &zero,
1004 .extra2 = &one_hundred,
1005 },
1006 #endif
1007 #ifdef CONFIG_SMP
1008 {
1009 .ctl_name = CTL_UNNUMBERED,
1010 .procname = "stat_interval",
1011 .data = &sysctl_stat_interval,
1012 .maxlen = sizeof(sysctl_stat_interval),
1013 .mode = 0644,
1014 .proc_handler = &proc_dointvec_jiffies,
1015 .strategy = &sysctl_jiffies,
1016 },
1017 #endif
1018 #ifdef CONFIG_SECURITY
1019 {
1020 .ctl_name = CTL_UNNUMBERED,
1021 .procname = "mmap_min_addr",
1022 .data = &mmap_min_addr,
1023 .maxlen = sizeof(unsigned long),
1024 .mode = 0644,
1025 .proc_handler = &proc_doulongvec_minmax,
1026 },
1027 #endif
1028 #ifdef CONFIG_NUMA
1029 {
1030 .ctl_name = CTL_UNNUMBERED,
1031 .procname = "numa_zonelist_order",
1032 .data = &numa_zonelist_order,
1033 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1034 .mode = 0644,
1035 .proc_handler = &numa_zonelist_order_handler,
1036 .strategy = &sysctl_string,
1037 },
1038 #endif
1039 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1040 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1041 {
1042 .ctl_name = VM_VDSO_ENABLED,
1043 .procname = "vdso_enabled",
1044 .data = &vdso_enabled,
1045 .maxlen = sizeof(vdso_enabled),
1046 .mode = 0644,
1047 .proc_handler = &proc_dointvec,
1048 .strategy = &sysctl_intvec,
1049 .extra1 = &zero,
1050 },
1051 #endif
1052 /*
1053 * NOTE: do not add new entries to this table unless you have read
1054 * Documentation/sysctl/ctl_unnumbered.txt
1055 */
1056 { .ctl_name = 0 }
1057 };
1058
1059 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1060 static ctl_table binfmt_misc_table[] = {
1061 { .ctl_name = 0 }
1062 };
1063 #endif
1064
1065 static ctl_table fs_table[] = {
1066 {
1067 .ctl_name = FS_NRINODE,
1068 .procname = "inode-nr",
1069 .data = &inodes_stat,
1070 .maxlen = 2*sizeof(int),
1071 .mode = 0444,
1072 .proc_handler = &proc_dointvec,
1073 },
1074 {
1075 .ctl_name = FS_STATINODE,
1076 .procname = "inode-state",
1077 .data = &inodes_stat,
1078 .maxlen = 7*sizeof(int),
1079 .mode = 0444,
1080 .proc_handler = &proc_dointvec,
1081 },
1082 {
1083 .ctl_name = FS_NRFILE,
1084 .procname = "file-nr",
1085 .data = &files_stat,
1086 .maxlen = 3*sizeof(int),
1087 .mode = 0444,
1088 .proc_handler = &proc_nr_files,
1089 },
1090 {
1091 .ctl_name = FS_MAXFILE,
1092 .procname = "file-max",
1093 .data = &files_stat.max_files,
1094 .maxlen = sizeof(int),
1095 .mode = 0644,
1096 .proc_handler = &proc_dointvec,
1097 },
1098 {
1099 .ctl_name = FS_DENTRY,
1100 .procname = "dentry-state",
1101 .data = &dentry_stat,
1102 .maxlen = 6*sizeof(int),
1103 .mode = 0444,
1104 .proc_handler = &proc_dointvec,
1105 },
1106 {
1107 .ctl_name = FS_OVERFLOWUID,
1108 .procname = "overflowuid",
1109 .data = &fs_overflowuid,
1110 .maxlen = sizeof(int),
1111 .mode = 0644,
1112 .proc_handler = &proc_dointvec_minmax,
1113 .strategy = &sysctl_intvec,
1114 .extra1 = &minolduid,
1115 .extra2 = &maxolduid,
1116 },
1117 {
1118 .ctl_name = FS_OVERFLOWGID,
1119 .procname = "overflowgid",
1120 .data = &fs_overflowgid,
1121 .maxlen = sizeof(int),
1122 .mode = 0644,
1123 .proc_handler = &proc_dointvec_minmax,
1124 .strategy = &sysctl_intvec,
1125 .extra1 = &minolduid,
1126 .extra2 = &maxolduid,
1127 },
1128 {
1129 .ctl_name = FS_LEASES,
1130 .procname = "leases-enable",
1131 .data = &leases_enable,
1132 .maxlen = sizeof(int),
1133 .mode = 0644,
1134 .proc_handler = &proc_dointvec,
1135 },
1136 #ifdef CONFIG_DNOTIFY
1137 {
1138 .ctl_name = FS_DIR_NOTIFY,
1139 .procname = "dir-notify-enable",
1140 .data = &dir_notify_enable,
1141 .maxlen = sizeof(int),
1142 .mode = 0644,
1143 .proc_handler = &proc_dointvec,
1144 },
1145 #endif
1146 #ifdef CONFIG_MMU
1147 {
1148 .ctl_name = FS_LEASE_TIME,
1149 .procname = "lease-break-time",
1150 .data = &lease_break_time,
1151 .maxlen = sizeof(int),
1152 .mode = 0644,
1153 .proc_handler = &proc_dointvec_minmax,
1154 .strategy = &sysctl_intvec,
1155 .extra1 = &zero,
1156 .extra2 = &two,
1157 },
1158 {
1159 .ctl_name = FS_AIO_NR,
1160 .procname = "aio-nr",
1161 .data = &aio_nr,
1162 .maxlen = sizeof(aio_nr),
1163 .mode = 0444,
1164 .proc_handler = &proc_doulongvec_minmax,
1165 },
1166 {
1167 .ctl_name = FS_AIO_MAX_NR,
1168 .procname = "aio-max-nr",
1169 .data = &aio_max_nr,
1170 .maxlen = sizeof(aio_max_nr),
1171 .mode = 0644,
1172 .proc_handler = &proc_doulongvec_minmax,
1173 },
1174 #ifdef CONFIG_INOTIFY_USER
1175 {
1176 .ctl_name = FS_INOTIFY,
1177 .procname = "inotify",
1178 .mode = 0555,
1179 .child = inotify_table,
1180 },
1181 #endif
1182 #endif
1183 {
1184 .ctl_name = KERN_SETUID_DUMPABLE,
1185 .procname = "suid_dumpable",
1186 .data = &suid_dumpable,
1187 .maxlen = sizeof(int),
1188 .mode = 0644,
1189 .proc_handler = &proc_dointvec,
1190 },
1191 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1192 {
1193 .ctl_name = CTL_UNNUMBERED,
1194 .procname = "binfmt_misc",
1195 .mode = 0555,
1196 .child = binfmt_misc_table,
1197 },
1198 #endif
1199 /*
1200 * NOTE: do not add new entries to this table unless you have read
1201 * Documentation/sysctl/ctl_unnumbered.txt
1202 */
1203 { .ctl_name = 0 }
1204 };
1205
1206 static ctl_table debug_table[] = {
1207 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1208 {
1209 .ctl_name = CTL_UNNUMBERED,
1210 .procname = "exception-trace",
1211 .data = &show_unhandled_signals,
1212 .maxlen = sizeof(int),
1213 .mode = 0644,
1214 .proc_handler = proc_dointvec
1215 },
1216 #endif
1217 { .ctl_name = 0 }
1218 };
1219
1220 static ctl_table dev_table[] = {
1221 { .ctl_name = 0 }
1222 };
1223
1224 static DEFINE_SPINLOCK(sysctl_lock);
1225
1226 /* called under sysctl_lock */
1227 static int use_table(struct ctl_table_header *p)
1228 {
1229 if (unlikely(p->unregistering))
1230 return 0;
1231 p->used++;
1232 return 1;
1233 }
1234
1235 /* called under sysctl_lock */
1236 static void unuse_table(struct ctl_table_header *p)
1237 {
1238 if (!--p->used)
1239 if (unlikely(p->unregistering))
1240 complete(p->unregistering);
1241 }
1242
1243 /* called under sysctl_lock, will reacquire if has to wait */
1244 static void start_unregistering(struct ctl_table_header *p)
1245 {
1246 /*
1247 * if p->used is 0, nobody will ever touch that entry again;
1248 * we'll eliminate all paths to it before dropping sysctl_lock
1249 */
1250 if (unlikely(p->used)) {
1251 struct completion wait;
1252 init_completion(&wait);
1253 p->unregistering = &wait;
1254 spin_unlock(&sysctl_lock);
1255 wait_for_completion(&wait);
1256 spin_lock(&sysctl_lock);
1257 }
1258 /*
1259 * do not remove from the list until nobody holds it; walking the
1260 * list in do_sysctl() relies on that.
1261 */
1262 list_del_init(&p->ctl_entry);
1263 }
1264
1265 void sysctl_head_finish(struct ctl_table_header *head)
1266 {
1267 if (!head)
1268 return;
1269 spin_lock(&sysctl_lock);
1270 unuse_table(head);
1271 spin_unlock(&sysctl_lock);
1272 }
1273
1274 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1275 {
1276 struct ctl_table_header *head;
1277 struct list_head *tmp;
1278 spin_lock(&sysctl_lock);
1279 if (prev) {
1280 tmp = &prev->ctl_entry;
1281 unuse_table(prev);
1282 goto next;
1283 }
1284 tmp = &root_table_header.ctl_entry;
1285 for (;;) {
1286 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1287
1288 if (!use_table(head))
1289 goto next;
1290 spin_unlock(&sysctl_lock);
1291 return head;
1292 next:
1293 tmp = tmp->next;
1294 if (tmp == &root_table_header.ctl_entry)
1295 break;
1296 }
1297 spin_unlock(&sysctl_lock);
1298 return NULL;
1299 }
1300
1301 #ifdef CONFIG_SYSCTL_SYSCALL
1302 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1303 void __user *newval, size_t newlen)
1304 {
1305 struct ctl_table_header *head;
1306 int error = -ENOTDIR;
1307
1308 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1309 return -ENOTDIR;
1310 if (oldval) {
1311 int old_len;
1312 if (!oldlenp || get_user(old_len, oldlenp))
1313 return -EFAULT;
1314 }
1315
1316 for (head = sysctl_head_next(NULL); head;
1317 head = sysctl_head_next(head)) {
1318 error = parse_table(name, nlen, oldval, oldlenp,
1319 newval, newlen, head->ctl_table);
1320 if (error != -ENOTDIR) {
1321 sysctl_head_finish(head);
1322 break;
1323 }
1324 }
1325 return error;
1326 }
1327
1328 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1329 {
1330 struct __sysctl_args tmp;
1331 int error;
1332
1333 if (copy_from_user(&tmp, args, sizeof(tmp)))
1334 return -EFAULT;
1335
1336 lock_kernel();
1337 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1338 tmp.newval, tmp.newlen);
1339 unlock_kernel();
1340 return error;
1341 }
1342 #endif /* CONFIG_SYSCTL_SYSCALL */
1343
1344 /*
1345 * sysctl_perm does NOT grant the superuser all rights automatically, because
1346 * some sysctl variables are readonly even to root.
1347 */
1348
1349 static int test_perm(int mode, int op)
1350 {
1351 if (!current->euid)
1352 mode >>= 6;
1353 else if (in_egroup_p(0))
1354 mode >>= 3;
1355 if ((mode & op & 0007) == op)
1356 return 0;
1357 return -EACCES;
1358 }
1359
1360 int sysctl_perm(ctl_table *table, int op)
1361 {
1362 int error;
1363 error = security_sysctl(table, op);
1364 if (error)
1365 return error;
1366 return test_perm(table->mode, op);
1367 }
1368
1369 #ifdef CONFIG_SYSCTL_SYSCALL
1370 static int parse_table(int __user *name, int nlen,
1371 void __user *oldval, size_t __user *oldlenp,
1372 void __user *newval, size_t newlen,
1373 ctl_table *table)
1374 {
1375 int n;
1376 repeat:
1377 if (!nlen)
1378 return -ENOTDIR;
1379 if (get_user(n, name))
1380 return -EFAULT;
1381 for ( ; table->ctl_name || table->procname; table++) {
1382 if (!table->ctl_name)
1383 continue;
1384 if (n == table->ctl_name) {
1385 int error;
1386 if (table->child) {
1387 if (sysctl_perm(table, 001))
1388 return -EPERM;
1389 name++;
1390 nlen--;
1391 table = table->child;
1392 goto repeat;
1393 }
1394 error = do_sysctl_strategy(table, name, nlen,
1395 oldval, oldlenp,
1396 newval, newlen);
1397 return error;
1398 }
1399 }
1400 return -ENOTDIR;
1401 }
1402
1403 /* Perform the actual read/write of a sysctl table entry. */
1404 int do_sysctl_strategy (ctl_table *table,
1405 int __user *name, int nlen,
1406 void __user *oldval, size_t __user *oldlenp,
1407 void __user *newval, size_t newlen)
1408 {
1409 int op = 0, rc;
1410 size_t len;
1411
1412 if (oldval)
1413 op |= 004;
1414 if (newval)
1415 op |= 002;
1416 if (sysctl_perm(table, op))
1417 return -EPERM;
1418
1419 if (table->strategy) {
1420 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1421 newval, newlen);
1422 if (rc < 0)
1423 return rc;
1424 if (rc > 0)
1425 return 0;
1426 }
1427
1428 /* If there is no strategy routine, or if the strategy returns
1429 * zero, proceed with automatic r/w */
1430 if (table->data && table->maxlen) {
1431 if (oldval && oldlenp) {
1432 if (get_user(len, oldlenp))
1433 return -EFAULT;
1434 if (len) {
1435 if (len > table->maxlen)
1436 len = table->maxlen;
1437 if(copy_to_user(oldval, table->data, len))
1438 return -EFAULT;
1439 if(put_user(len, oldlenp))
1440 return -EFAULT;
1441 }
1442 }
1443 if (newval && newlen) {
1444 len = newlen;
1445 if (len > table->maxlen)
1446 len = table->maxlen;
1447 if(copy_from_user(table->data, newval, len))
1448 return -EFAULT;
1449 }
1450 }
1451 return 0;
1452 }
1453 #endif /* CONFIG_SYSCTL_SYSCALL */
1454
1455 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1456 {
1457 for (; table->ctl_name || table->procname; table++) {
1458 table->parent = parent;
1459 if (table->child)
1460 sysctl_set_parent(table, table->child);
1461 }
1462 }
1463
1464 static __init int sysctl_init(void)
1465 {
1466 sysctl_set_parent(NULL, root_table);
1467 return 0;
1468 }
1469
1470 core_initcall(sysctl_init);
1471
1472 /**
1473 * register_sysctl_table - register a sysctl hierarchy
1474 * @table: the top-level table structure
1475 *
1476 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1477 * array. An entry with a ctl_name of 0 terminates the table.
1478 *
1479 * The members of the &ctl_table structure are used as follows:
1480 *
1481 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1482 * must be unique within that level of sysctl
1483 *
1484 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1485 * enter a sysctl file
1486 *
1487 * data - a pointer to data for use by proc_handler
1488 *
1489 * maxlen - the maximum size in bytes of the data
1490 *
1491 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1492 *
1493 * child - a pointer to the child sysctl table if this entry is a directory, or
1494 * %NULL.
1495 *
1496 * proc_handler - the text handler routine (described below)
1497 *
1498 * strategy - the strategy routine (described below)
1499 *
1500 * de - for internal use by the sysctl routines
1501 *
1502 * extra1, extra2 - extra pointers usable by the proc handler routines
1503 *
1504 * Leaf nodes in the sysctl tree will be represented by a single file
1505 * under /proc; non-leaf nodes will be represented by directories.
1506 *
1507 * sysctl(2) can automatically manage read and write requests through
1508 * the sysctl table. The data and maxlen fields of the ctl_table
1509 * struct enable minimal validation of the values being written to be
1510 * performed, and the mode field allows minimal authentication.
1511 *
1512 * More sophisticated management can be enabled by the provision of a
1513 * strategy routine with the table entry. This will be called before
1514 * any automatic read or write of the data is performed.
1515 *
1516 * The strategy routine may return
1517 *
1518 * < 0 - Error occurred (error is passed to user process)
1519 *
1520 * 0 - OK - proceed with automatic read or write.
1521 *
1522 * > 0 - OK - read or write has been done by the strategy routine, so
1523 * return immediately.
1524 *
1525 * There must be a proc_handler routine for any terminal nodes
1526 * mirrored under /proc/sys (non-terminals are handled by a built-in
1527 * directory handler). Several default handlers are available to
1528 * cover common cases -
1529 *
1530 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1531 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1532 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1533 *
1534 * It is the handler's job to read the input buffer from user memory
1535 * and process it. The handler should return 0 on success.
1536 *
1537 * This routine returns %NULL on a failure to register, and a pointer
1538 * to the table header on success.
1539 */
1540 struct ctl_table_header *register_sysctl_table(ctl_table * table)
1541 {
1542 struct ctl_table_header *tmp;
1543 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1544 if (!tmp)
1545 return NULL;
1546 tmp->ctl_table = table;
1547 INIT_LIST_HEAD(&tmp->ctl_entry);
1548 tmp->used = 0;
1549 tmp->unregistering = NULL;
1550 sysctl_set_parent(NULL, table);
1551 spin_lock(&sysctl_lock);
1552 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
1553 spin_unlock(&sysctl_lock);
1554 return tmp;
1555 }
1556
1557 /**
1558 * unregister_sysctl_table - unregister a sysctl table hierarchy
1559 * @header: the header returned from register_sysctl_table
1560 *
1561 * Unregisters the sysctl table and all children. proc entries may not
1562 * actually be removed until they are no longer used by anyone.
1563 */
1564 void unregister_sysctl_table(struct ctl_table_header * header)
1565 {
1566 might_sleep();
1567 spin_lock(&sysctl_lock);
1568 start_unregistering(header);
1569 spin_unlock(&sysctl_lock);
1570 kfree(header);
1571 }
1572
1573 #else /* !CONFIG_SYSCTL */
1574 struct ctl_table_header *register_sysctl_table(ctl_table * table)
1575 {
1576 return NULL;
1577 }
1578
1579 void unregister_sysctl_table(struct ctl_table_header * table)
1580 {
1581 }
1582
1583 #endif /* CONFIG_SYSCTL */
1584
1585 /*
1586 * /proc/sys support
1587 */
1588
1589 #ifdef CONFIG_PROC_SYSCTL
1590
1591 static int _proc_do_string(void* data, int maxlen, int write,
1592 struct file *filp, void __user *buffer,
1593 size_t *lenp, loff_t *ppos)
1594 {
1595 size_t len;
1596 char __user *p;
1597 char c;
1598
1599 if (!data || !maxlen || !*lenp) {
1600 *lenp = 0;
1601 return 0;
1602 }
1603
1604 if (write) {
1605 len = 0;
1606 p = buffer;
1607 while (len < *lenp) {
1608 if (get_user(c, p++))
1609 return -EFAULT;
1610 if (c == 0 || c == '\n')
1611 break;
1612 len++;
1613 }
1614 if (len >= maxlen)
1615 len = maxlen-1;
1616 if(copy_from_user(data, buffer, len))
1617 return -EFAULT;
1618 ((char *) data)[len] = 0;
1619 *ppos += *lenp;
1620 } else {
1621 len = strlen(data);
1622 if (len > maxlen)
1623 len = maxlen;
1624
1625 if (*ppos > len) {
1626 *lenp = 0;
1627 return 0;
1628 }
1629
1630 data += *ppos;
1631 len -= *ppos;
1632
1633 if (len > *lenp)
1634 len = *lenp;
1635 if (len)
1636 if(copy_to_user(buffer, data, len))
1637 return -EFAULT;
1638 if (len < *lenp) {
1639 if(put_user('\n', ((char __user *) buffer) + len))
1640 return -EFAULT;
1641 len++;
1642 }
1643 *lenp = len;
1644 *ppos += len;
1645 }
1646 return 0;
1647 }
1648
1649 /**
1650 * proc_dostring - read a string sysctl
1651 * @table: the sysctl table
1652 * @write: %TRUE if this is a write to the sysctl file
1653 * @filp: the file structure
1654 * @buffer: the user buffer
1655 * @lenp: the size of the user buffer
1656 * @ppos: file position
1657 *
1658 * Reads/writes a string from/to the user buffer. If the kernel
1659 * buffer provided is not large enough to hold the string, the
1660 * string is truncated. The copied string is %NULL-terminated.
1661 * If the string is being read by the user process, it is copied
1662 * and a newline '\n' is added. It is truncated if the buffer is
1663 * not large enough.
1664 *
1665 * Returns 0 on success.
1666 */
1667 int proc_dostring(ctl_table *table, int write, struct file *filp,
1668 void __user *buffer, size_t *lenp, loff_t *ppos)
1669 {
1670 return _proc_do_string(table->data, table->maxlen, write, filp,
1671 buffer, lenp, ppos);
1672 }
1673
1674
1675 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1676 int *valp,
1677 int write, void *data)
1678 {
1679 if (write) {
1680 *valp = *negp ? -*lvalp : *lvalp;
1681 } else {
1682 int val = *valp;
1683 if (val < 0) {
1684 *negp = -1;
1685 *lvalp = (unsigned long)-val;
1686 } else {
1687 *negp = 0;
1688 *lvalp = (unsigned long)val;
1689 }
1690 }
1691 return 0;
1692 }
1693
1694 static int __do_proc_dointvec(void *tbl_data, ctl_table *table,
1695 int write, struct file *filp, void __user *buffer,
1696 size_t *lenp, loff_t *ppos,
1697 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1698 int write, void *data),
1699 void *data)
1700 {
1701 #define TMPBUFLEN 21
1702 int *i, vleft, first=1, neg, val;
1703 unsigned long lval;
1704 size_t left, len;
1705
1706 char buf[TMPBUFLEN], *p;
1707 char __user *s = buffer;
1708
1709 if (!tbl_data || !table->maxlen || !*lenp ||
1710 (*ppos && !write)) {
1711 *lenp = 0;
1712 return 0;
1713 }
1714
1715 i = (int *) tbl_data;
1716 vleft = table->maxlen / sizeof(*i);
1717 left = *lenp;
1718
1719 if (!conv)
1720 conv = do_proc_dointvec_conv;
1721
1722 for (; left && vleft--; i++, first=0) {
1723 if (write) {
1724 while (left) {
1725 char c;
1726 if (get_user(c, s))
1727 return -EFAULT;
1728 if (!isspace(c))
1729 break;
1730 left--;
1731 s++;
1732 }
1733 if (!left)
1734 break;
1735 neg = 0;
1736 len = left;
1737 if (len > sizeof(buf) - 1)
1738 len = sizeof(buf) - 1;
1739 if (copy_from_user(buf, s, len))
1740 return -EFAULT;
1741 buf[len] = 0;
1742 p = buf;
1743 if (*p == '-' && left > 1) {
1744 neg = 1;
1745 p++;
1746 }
1747 if (*p < '0' || *p > '9')
1748 break;
1749
1750 lval = simple_strtoul(p, &p, 0);
1751
1752 len = p-buf;
1753 if ((len < left) && *p && !isspace(*p))
1754 break;
1755 if (neg)
1756 val = -val;
1757 s += len;
1758 left -= len;
1759
1760 if (conv(&neg, &lval, i, 1, data))
1761 break;
1762 } else {
1763 p = buf;
1764 if (!first)
1765 *p++ = '\t';
1766
1767 if (conv(&neg, &lval, i, 0, data))
1768 break;
1769
1770 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1771 len = strlen(buf);
1772 if (len > left)
1773 len = left;
1774 if(copy_to_user(s, buf, len))
1775 return -EFAULT;
1776 left -= len;
1777 s += len;
1778 }
1779 }
1780
1781 if (!write && !first && left) {
1782 if(put_user('\n', s))
1783 return -EFAULT;
1784 left--, s++;
1785 }
1786 if (write) {
1787 while (left) {
1788 char c;
1789 if (get_user(c, s++))
1790 return -EFAULT;
1791 if (!isspace(c))
1792 break;
1793 left--;
1794 }
1795 }
1796 if (write && first)
1797 return -EINVAL;
1798 *lenp -= left;
1799 *ppos += *lenp;
1800 return 0;
1801 #undef TMPBUFLEN
1802 }
1803
1804 static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1805 void __user *buffer, size_t *lenp, loff_t *ppos,
1806 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1807 int write, void *data),
1808 void *data)
1809 {
1810 return __do_proc_dointvec(table->data, table, write, filp,
1811 buffer, lenp, ppos, conv, data);
1812 }
1813
1814 /**
1815 * proc_dointvec - read a vector of integers
1816 * @table: the sysctl table
1817 * @write: %TRUE if this is a write to the sysctl file
1818 * @filp: the file structure
1819 * @buffer: the user buffer
1820 * @lenp: the size of the user buffer
1821 * @ppos: file position
1822 *
1823 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1824 * values from/to the user buffer, treated as an ASCII string.
1825 *
1826 * Returns 0 on success.
1827 */
1828 int proc_dointvec(ctl_table *table, int write, struct file *filp,
1829 void __user *buffer, size_t *lenp, loff_t *ppos)
1830 {
1831 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1832 NULL,NULL);
1833 }
1834
1835 #define OP_SET 0
1836 #define OP_AND 1
1837 #define OP_OR 2
1838
1839 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1840 int *valp,
1841 int write, void *data)
1842 {
1843 int op = *(int *)data;
1844 if (write) {
1845 int val = *negp ? -*lvalp : *lvalp;
1846 switch(op) {
1847 case OP_SET: *valp = val; break;
1848 case OP_AND: *valp &= val; break;
1849 case OP_OR: *valp |= val; break;
1850 }
1851 } else {
1852 int val = *valp;
1853 if (val < 0) {
1854 *negp = -1;
1855 *lvalp = (unsigned long)-val;
1856 } else {
1857 *negp = 0;
1858 *lvalp = (unsigned long)val;
1859 }
1860 }
1861 return 0;
1862 }
1863
1864 /*
1865 * init may raise the set.
1866 */
1867
1868 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1869 void __user *buffer, size_t *lenp, loff_t *ppos)
1870 {
1871 int op;
1872
1873 if (write && !capable(CAP_SYS_MODULE)) {
1874 return -EPERM;
1875 }
1876
1877 op = is_init(current) ? OP_SET : OP_AND;
1878 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1879 do_proc_dointvec_bset_conv,&op);
1880 }
1881
1882 /*
1883 * Taint values can only be increased
1884 */
1885 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
1886 void __user *buffer, size_t *lenp, loff_t *ppos)
1887 {
1888 int op;
1889
1890 if (write && !capable(CAP_SYS_ADMIN))
1891 return -EPERM;
1892
1893 op = OP_OR;
1894 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1895 do_proc_dointvec_bset_conv,&op);
1896 }
1897
1898 struct do_proc_dointvec_minmax_conv_param {
1899 int *min;
1900 int *max;
1901 };
1902
1903 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1904 int *valp,
1905 int write, void *data)
1906 {
1907 struct do_proc_dointvec_minmax_conv_param *param = data;
1908 if (write) {
1909 int val = *negp ? -*lvalp : *lvalp;
1910 if ((param->min && *param->min > val) ||
1911 (param->max && *param->max < val))
1912 return -EINVAL;
1913 *valp = val;
1914 } else {
1915 int val = *valp;
1916 if (val < 0) {
1917 *negp = -1;
1918 *lvalp = (unsigned long)-val;
1919 } else {
1920 *negp = 0;
1921 *lvalp = (unsigned long)val;
1922 }
1923 }
1924 return 0;
1925 }
1926
1927 /**
1928 * proc_dointvec_minmax - read a vector of integers with min/max values
1929 * @table: the sysctl table
1930 * @write: %TRUE if this is a write to the sysctl file
1931 * @filp: the file structure
1932 * @buffer: the user buffer
1933 * @lenp: the size of the user buffer
1934 * @ppos: file position
1935 *
1936 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1937 * values from/to the user buffer, treated as an ASCII string.
1938 *
1939 * This routine will ensure the values are within the range specified by
1940 * table->extra1 (min) and table->extra2 (max).
1941 *
1942 * Returns 0 on success.
1943 */
1944 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1945 void __user *buffer, size_t *lenp, loff_t *ppos)
1946 {
1947 struct do_proc_dointvec_minmax_conv_param param = {
1948 .min = (int *) table->extra1,
1949 .max = (int *) table->extra2,
1950 };
1951 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1952 do_proc_dointvec_minmax_conv, &param);
1953 }
1954
1955 static int __do_proc_doulongvec_minmax(void *data, ctl_table *table, int write,
1956 struct file *filp,
1957 void __user *buffer,
1958 size_t *lenp, loff_t *ppos,
1959 unsigned long convmul,
1960 unsigned long convdiv)
1961 {
1962 #define TMPBUFLEN 21
1963 unsigned long *i, *min, *max, val;
1964 int vleft, first=1, neg;
1965 size_t len, left;
1966 char buf[TMPBUFLEN], *p;
1967 char __user *s = buffer;
1968
1969 if (!data || !table->maxlen || !*lenp ||
1970 (*ppos && !write)) {
1971 *lenp = 0;
1972 return 0;
1973 }
1974
1975 i = (unsigned long *) data;
1976 min = (unsigned long *) table->extra1;
1977 max = (unsigned long *) table->extra2;
1978 vleft = table->maxlen / sizeof(unsigned long);
1979 left = *lenp;
1980
1981 for (; left && vleft--; i++, min++, max++, first=0) {
1982 if (write) {
1983 while (left) {
1984 char c;
1985 if (get_user(c, s))
1986 return -EFAULT;
1987 if (!isspace(c))
1988 break;
1989 left--;
1990 s++;
1991 }
1992 if (!left)
1993 break;
1994 neg = 0;
1995 len = left;
1996 if (len > TMPBUFLEN-1)
1997 len = TMPBUFLEN-1;
1998 if (copy_from_user(buf, s, len))
1999 return -EFAULT;
2000 buf[len] = 0;
2001 p = buf;
2002 if (*p == '-' && left > 1) {
2003 neg = 1;
2004 p++;
2005 }
2006 if (*p < '0' || *p > '9')
2007 break;
2008 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2009 len = p-buf;
2010 if ((len < left) && *p && !isspace(*p))
2011 break;
2012 if (neg)
2013 val = -val;
2014 s += len;
2015 left -= len;
2016
2017 if(neg)
2018 continue;
2019 if ((min && val < *min) || (max && val > *max))
2020 continue;
2021 *i = val;
2022 } else {
2023 p = buf;
2024 if (!first)
2025 *p++ = '\t';
2026 sprintf(p, "%lu", convdiv * (*i) / convmul);
2027 len = strlen(buf);
2028 if (len > left)
2029 len = left;
2030 if(copy_to_user(s, buf, len))
2031 return -EFAULT;
2032 left -= len;
2033 s += len;
2034 }
2035 }
2036
2037 if (!write && !first && left) {
2038 if(put_user('\n', s))
2039 return -EFAULT;
2040 left--, s++;
2041 }
2042 if (write) {
2043 while (left) {
2044 char c;
2045 if (get_user(c, s++))
2046 return -EFAULT;
2047 if (!isspace(c))
2048 break;
2049 left--;
2050 }
2051 }
2052 if (write && first)
2053 return -EINVAL;
2054 *lenp -= left;
2055 *ppos += *lenp;
2056 return 0;
2057 #undef TMPBUFLEN
2058 }
2059
2060 static int do_proc_doulongvec_minmax(ctl_table *table, int write,
2061 struct file *filp,
2062 void __user *buffer,
2063 size_t *lenp, loff_t *ppos,
2064 unsigned long convmul,
2065 unsigned long convdiv)
2066 {
2067 return __do_proc_doulongvec_minmax(table->data, table, write,
2068 filp, buffer, lenp, ppos, convmul, convdiv);
2069 }
2070
2071 /**
2072 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2073 * @table: the sysctl table
2074 * @write: %TRUE if this is a write to the sysctl file
2075 * @filp: the file structure
2076 * @buffer: the user buffer
2077 * @lenp: the size of the user buffer
2078 * @ppos: file position
2079 *
2080 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2081 * values from/to the user buffer, treated as an ASCII string.
2082 *
2083 * This routine will ensure the values are within the range specified by
2084 * table->extra1 (min) and table->extra2 (max).
2085 *
2086 * Returns 0 on success.
2087 */
2088 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2089 void __user *buffer, size_t *lenp, loff_t *ppos)
2090 {
2091 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2092 }
2093
2094 /**
2095 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2096 * @table: the sysctl table
2097 * @write: %TRUE if this is a write to the sysctl file
2098 * @filp: the file structure
2099 * @buffer: the user buffer
2100 * @lenp: the size of the user buffer
2101 * @ppos: file position
2102 *
2103 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2104 * values from/to the user buffer, treated as an ASCII string. The values
2105 * are treated as milliseconds, and converted to jiffies when they are stored.
2106 *
2107 * This routine will ensure the values are within the range specified by
2108 * table->extra1 (min) and table->extra2 (max).
2109 *
2110 * Returns 0 on success.
2111 */
2112 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2113 struct file *filp,
2114 void __user *buffer,
2115 size_t *lenp, loff_t *ppos)
2116 {
2117 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2118 lenp, ppos, HZ, 1000l);
2119 }
2120
2121
2122 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2123 int *valp,
2124 int write, void *data)
2125 {
2126 if (write) {
2127 if (*lvalp > LONG_MAX / HZ)
2128 return 1;
2129 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2130 } else {
2131 int val = *valp;
2132 unsigned long lval;
2133 if (val < 0) {
2134 *negp = -1;
2135 lval = (unsigned long)-val;
2136 } else {
2137 *negp = 0;
2138 lval = (unsigned long)val;
2139 }
2140 *lvalp = lval / HZ;
2141 }
2142 return 0;
2143 }
2144
2145 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2146 int *valp,
2147 int write, void *data)
2148 {
2149 if (write) {
2150 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2151 return 1;
2152 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2153 } else {
2154 int val = *valp;
2155 unsigned long lval;
2156 if (val < 0) {
2157 *negp = -1;
2158 lval = (unsigned long)-val;
2159 } else {
2160 *negp = 0;
2161 lval = (unsigned long)val;
2162 }
2163 *lvalp = jiffies_to_clock_t(lval);
2164 }
2165 return 0;
2166 }
2167
2168 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2169 int *valp,
2170 int write, void *data)
2171 {
2172 if (write) {
2173 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2174 } else {
2175 int val = *valp;
2176 unsigned long lval;
2177 if (val < 0) {
2178 *negp = -1;
2179 lval = (unsigned long)-val;
2180 } else {
2181 *negp = 0;
2182 lval = (unsigned long)val;
2183 }
2184 *lvalp = jiffies_to_msecs(lval);
2185 }
2186 return 0;
2187 }
2188
2189 /**
2190 * proc_dointvec_jiffies - read a vector of integers as seconds
2191 * @table: the sysctl table
2192 * @write: %TRUE if this is a write to the sysctl file
2193 * @filp: the file structure
2194 * @buffer: the user buffer
2195 * @lenp: the size of the user buffer
2196 * @ppos: file position
2197 *
2198 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2199 * values from/to the user buffer, treated as an ASCII string.
2200 * The values read are assumed to be in seconds, and are converted into
2201 * jiffies.
2202 *
2203 * Returns 0 on success.
2204 */
2205 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2206 void __user *buffer, size_t *lenp, loff_t *ppos)
2207 {
2208 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2209 do_proc_dointvec_jiffies_conv,NULL);
2210 }
2211
2212 /**
2213 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2214 * @table: the sysctl table
2215 * @write: %TRUE if this is a write to the sysctl file
2216 * @filp: the file structure
2217 * @buffer: the user buffer
2218 * @lenp: the size of the user buffer
2219 * @ppos: pointer to the file position
2220 *
2221 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2222 * values from/to the user buffer, treated as an ASCII string.
2223 * The values read are assumed to be in 1/USER_HZ seconds, and
2224 * are converted into jiffies.
2225 *
2226 * Returns 0 on success.
2227 */
2228 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2229 void __user *buffer, size_t *lenp, loff_t *ppos)
2230 {
2231 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2232 do_proc_dointvec_userhz_jiffies_conv,NULL);
2233 }
2234
2235 /**
2236 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2237 * @table: the sysctl table
2238 * @write: %TRUE if this is a write to the sysctl file
2239 * @filp: the file structure
2240 * @buffer: the user buffer
2241 * @lenp: the size of the user buffer
2242 * @ppos: file position
2243 * @ppos: the current position in the file
2244 *
2245 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2246 * values from/to the user buffer, treated as an ASCII string.
2247 * The values read are assumed to be in 1/1000 seconds, and
2248 * are converted into jiffies.
2249 *
2250 * Returns 0 on success.
2251 */
2252 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2253 void __user *buffer, size_t *lenp, loff_t *ppos)
2254 {
2255 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2256 do_proc_dointvec_ms_jiffies_conv, NULL);
2257 }
2258
2259 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
2260 void __user *buffer, size_t *lenp, loff_t *ppos)
2261 {
2262 struct pid *new_pid;
2263 pid_t tmp;
2264 int r;
2265
2266 tmp = pid_nr(cad_pid);
2267
2268 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2269 lenp, ppos, NULL, NULL);
2270 if (r || !write)
2271 return r;
2272
2273 new_pid = find_get_pid(tmp);
2274 if (!new_pid)
2275 return -ESRCH;
2276
2277 put_pid(xchg(&cad_pid, new_pid));
2278 return 0;
2279 }
2280
2281 #else /* CONFIG_PROC_FS */
2282
2283 int proc_dostring(ctl_table *table, int write, struct file *filp,
2284 void __user *buffer, size_t *lenp, loff_t *ppos)
2285 {
2286 return -ENOSYS;
2287 }
2288
2289 int proc_dointvec(ctl_table *table, int write, struct file *filp,
2290 void __user *buffer, size_t *lenp, loff_t *ppos)
2291 {
2292 return -ENOSYS;
2293 }
2294
2295 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2296 void __user *buffer, size_t *lenp, loff_t *ppos)
2297 {
2298 return -ENOSYS;
2299 }
2300
2301 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2302 void __user *buffer, size_t *lenp, loff_t *ppos)
2303 {
2304 return -ENOSYS;
2305 }
2306
2307 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2308 void __user *buffer, size_t *lenp, loff_t *ppos)
2309 {
2310 return -ENOSYS;
2311 }
2312
2313 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2314 void __user *buffer, size_t *lenp, loff_t *ppos)
2315 {
2316 return -ENOSYS;
2317 }
2318
2319 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2320 void __user *buffer, size_t *lenp, loff_t *ppos)
2321 {
2322 return -ENOSYS;
2323 }
2324
2325 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2326 void __user *buffer, size_t *lenp, loff_t *ppos)
2327 {
2328 return -ENOSYS;
2329 }
2330
2331 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2332 struct file *filp,
2333 void __user *buffer,
2334 size_t *lenp, loff_t *ppos)
2335 {
2336 return -ENOSYS;
2337 }
2338
2339
2340 #endif /* CONFIG_PROC_FS */
2341
2342
2343 #ifdef CONFIG_SYSCTL_SYSCALL
2344 /*
2345 * General sysctl support routines
2346 */
2347
2348 /* The generic string strategy routine: */
2349 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2350 void __user *oldval, size_t __user *oldlenp,
2351 void __user *newval, size_t newlen)
2352 {
2353 if (!table->data || !table->maxlen)
2354 return -ENOTDIR;
2355
2356 if (oldval && oldlenp) {
2357 size_t bufsize;
2358 if (get_user(bufsize, oldlenp))
2359 return -EFAULT;
2360 if (bufsize) {
2361 size_t len = strlen(table->data), copied;
2362
2363 /* This shouldn't trigger for a well-formed sysctl */
2364 if (len > table->maxlen)
2365 len = table->maxlen;
2366
2367 /* Copy up to a max of bufsize-1 bytes of the string */
2368 copied = (len >= bufsize) ? bufsize - 1 : len;
2369
2370 if (copy_to_user(oldval, table->data, copied) ||
2371 put_user(0, (char __user *)(oldval + copied)))
2372 return -EFAULT;
2373 if (put_user(len, oldlenp))
2374 return -EFAULT;
2375 }
2376 }
2377 if (newval && newlen) {
2378 size_t len = newlen;
2379 if (len > table->maxlen)
2380 len = table->maxlen;
2381 if(copy_from_user(table->data, newval, len))
2382 return -EFAULT;
2383 if (len == table->maxlen)
2384 len--;
2385 ((char *) table->data)[len] = 0;
2386 }
2387 return 1;
2388 }
2389
2390 /*
2391 * This function makes sure that all of the integers in the vector
2392 * are between the minimum and maximum values given in the arrays
2393 * table->extra1 and table->extra2, respectively.
2394 */
2395 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2396 void __user *oldval, size_t __user *oldlenp,
2397 void __user *newval, size_t newlen)
2398 {
2399
2400 if (newval && newlen) {
2401 int __user *vec = (int __user *) newval;
2402 int *min = (int *) table->extra1;
2403 int *max = (int *) table->extra2;
2404 size_t length;
2405 int i;
2406
2407 if (newlen % sizeof(int) != 0)
2408 return -EINVAL;
2409
2410 if (!table->extra1 && !table->extra2)
2411 return 0;
2412
2413 if (newlen > table->maxlen)
2414 newlen = table->maxlen;
2415 length = newlen / sizeof(int);
2416
2417 for (i = 0; i < length; i++) {
2418 int value;
2419 if (get_user(value, vec + i))
2420 return -EFAULT;
2421 if (min && value < min[i])
2422 return -EINVAL;
2423 if (max && value > max[i])
2424 return -EINVAL;
2425 }
2426 }
2427 return 0;
2428 }
2429
2430 /* Strategy function to convert jiffies to seconds */
2431 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2432 void __user *oldval, size_t __user *oldlenp,
2433 void __user *newval, size_t newlen)
2434 {
2435 if (oldval && oldlenp) {
2436 size_t olen;
2437
2438 if (get_user(olen, oldlenp))
2439 return -EFAULT;
2440 if (olen) {
2441 int val;
2442
2443 if (olen < sizeof(int))
2444 return -EINVAL;
2445
2446 val = *(int *)(table->data) / HZ;
2447 if (put_user(val, (int __user *)oldval))
2448 return -EFAULT;
2449 if (put_user(sizeof(int), oldlenp))
2450 return -EFAULT;
2451 }
2452 }
2453 if (newval && newlen) {
2454 int new;
2455 if (newlen != sizeof(int))
2456 return -EINVAL;
2457 if (get_user(new, (int __user *)newval))
2458 return -EFAULT;
2459 *(int *)(table->data) = new*HZ;
2460 }
2461 return 1;
2462 }
2463
2464 /* Strategy function to convert jiffies to seconds */
2465 int sysctl_ms_jiffies(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 if (oldval && oldlenp) {
2470 size_t olen;
2471
2472 if (get_user(olen, oldlenp))
2473 return -EFAULT;
2474 if (olen) {
2475 int val;
2476
2477 if (olen < sizeof(int))
2478 return -EINVAL;
2479
2480 val = jiffies_to_msecs(*(int *)(table->data));
2481 if (put_user(val, (int __user *)oldval))
2482 return -EFAULT;
2483 if (put_user(sizeof(int), oldlenp))
2484 return -EFAULT;
2485 }
2486 }
2487 if (newval && newlen) {
2488 int new;
2489 if (newlen != sizeof(int))
2490 return -EINVAL;
2491 if (get_user(new, (int __user *)newval))
2492 return -EFAULT;
2493 *(int *)(table->data) = msecs_to_jiffies(new);
2494 }
2495 return 1;
2496 }
2497
2498
2499
2500 #else /* CONFIG_SYSCTL_SYSCALL */
2501
2502
2503 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2504 {
2505 static int msg_count;
2506 struct __sysctl_args tmp;
2507 int name[CTL_MAXNAME];
2508 int i;
2509
2510 /* Read in the sysctl name for better debug message logging */
2511 if (copy_from_user(&tmp, args, sizeof(tmp)))
2512 return -EFAULT;
2513 if (tmp.nlen <= 0 || tmp.nlen >= CTL_MAXNAME)
2514 return -ENOTDIR;
2515 for (i = 0; i < tmp.nlen; i++)
2516 if (get_user(name[i], tmp.name + i))
2517 return -EFAULT;
2518
2519 /* Ignore accesses to kernel.version */
2520 if ((tmp.nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2521 goto out;
2522
2523 if (msg_count < 5) {
2524 msg_count++;
2525 printk(KERN_INFO
2526 "warning: process `%s' used the removed sysctl "
2527 "system call with ", current->comm);
2528 for (i = 0; i < tmp.nlen; i++)
2529 printk("%d.", name[i]);
2530 printk("\n");
2531 }
2532 out:
2533 return -ENOSYS;
2534 }
2535
2536 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2537 void __user *oldval, size_t __user *oldlenp,
2538 void __user *newval, size_t newlen)
2539 {
2540 return -ENOSYS;
2541 }
2542
2543 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2544 void __user *oldval, size_t __user *oldlenp,
2545 void __user *newval, size_t newlen)
2546 {
2547 return -ENOSYS;
2548 }
2549
2550 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2551 void __user *oldval, size_t __user *oldlenp,
2552 void __user *newval, size_t newlen)
2553 {
2554 return -ENOSYS;
2555 }
2556
2557 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2558 void __user *oldval, size_t __user *oldlenp,
2559 void __user *newval, size_t newlen)
2560 {
2561 return -ENOSYS;
2562 }
2563
2564 #endif /* CONFIG_SYSCTL_SYSCALL */
2565
2566 /*
2567 * No sense putting this after each symbol definition, twice,
2568 * exception granted :-)
2569 */
2570 EXPORT_SYMBOL(proc_dointvec);
2571 EXPORT_SYMBOL(proc_dointvec_jiffies);
2572 EXPORT_SYMBOL(proc_dointvec_minmax);
2573 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2574 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2575 EXPORT_SYMBOL(proc_dostring);
2576 EXPORT_SYMBOL(proc_doulongvec_minmax);
2577 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2578 EXPORT_SYMBOL(register_sysctl_table);
2579 EXPORT_SYMBOL(sysctl_intvec);
2580 EXPORT_SYMBOL(sysctl_jiffies);
2581 EXPORT_SYMBOL(sysctl_ms_jiffies);
2582 EXPORT_SYMBOL(sysctl_string);
2583 EXPORT_SYMBOL(unregister_sysctl_table);
kmemtrace - Kernel memory tracer