sched: fix sign check error in place_entity()
[GitHub/mt8127/android_kernel_alcatel_ttab.git] / kernel / sysctl.c
1 /*
2 * sysctl.c: General linux system control interface
3 *
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12 * Horn.
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16 * Wendling.
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19 */
20
21 #include <linux/module.h>
22 #include <linux/mm.h>
23 #include <linux/swap.h>
24 #include <linux/slab.h>
25 #include <linux/sysctl.h>
26 #include <linux/proc_fs.h>
27 #include <linux/capability.h>
28 #include <linux/ctype.h>
29 #include <linux/utsname.h>
30 #include <linux/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_min_granularity_ns",
226 .data = &sysctl_sched_min_granularity,
227 .maxlen = sizeof(unsigned int),
228 .mode = 0644,
229 .proc_handler = &proc_dointvec_minmax,
230 .strategy = &sysctl_intvec,
231 .extra1 = &min_sched_granularity_ns,
232 .extra2 = &max_sched_granularity_ns,
233 },
234 {
235 .ctl_name = CTL_UNNUMBERED,
236 .procname = "sched_latency_ns",
237 .data = &sysctl_sched_latency,
238 .maxlen = sizeof(unsigned int),
239 .mode = 0644,
240 .proc_handler = &proc_dointvec_minmax,
241 .strategy = &sysctl_intvec,
242 .extra1 = &min_sched_granularity_ns,
243 .extra2 = &max_sched_granularity_ns,
244 },
245 {
246 .ctl_name = CTL_UNNUMBERED,
247 .procname = "sched_wakeup_granularity_ns",
248 .data = &sysctl_sched_wakeup_granularity,
249 .maxlen = sizeof(unsigned int),
250 .mode = 0644,
251 .proc_handler = &proc_dointvec_minmax,
252 .strategy = &sysctl_intvec,
253 .extra1 = &min_wakeup_granularity_ns,
254 .extra2 = &max_wakeup_granularity_ns,
255 },
256 {
257 .ctl_name = CTL_UNNUMBERED,
258 .procname = "sched_batch_wakeup_granularity_ns",
259 .data = &sysctl_sched_batch_wakeup_granularity,
260 .maxlen = sizeof(unsigned int),
261 .mode = 0644,
262 .proc_handler = &proc_dointvec_minmax,
263 .strategy = &sysctl_intvec,
264 .extra1 = &min_wakeup_granularity_ns,
265 .extra2 = &max_wakeup_granularity_ns,
266 },
267 {
268 .ctl_name = CTL_UNNUMBERED,
269 .procname = "sched_child_runs_first",
270 .data = &sysctl_sched_child_runs_first,
271 .maxlen = sizeof(unsigned int),
272 .mode = 0644,
273 .proc_handler = &proc_dointvec,
274 },
275 {
276 .ctl_name = CTL_UNNUMBERED,
277 .procname = "sched_features",
278 .data = &sysctl_sched_features,
279 .maxlen = sizeof(unsigned int),
280 .mode = 0644,
281 .proc_handler = &proc_dointvec,
282 },
283 #endif
284 {
285 .ctl_name = CTL_UNNUMBERED,
286 .procname = "sched_compat_yield",
287 .data = &sysctl_sched_compat_yield,
288 .maxlen = sizeof(unsigned int),
289 .mode = 0644,
290 .proc_handler = &proc_dointvec,
291 },
292 #ifdef CONFIG_PROVE_LOCKING
293 {
294 .ctl_name = CTL_UNNUMBERED,
295 .procname = "prove_locking",
296 .data = &prove_locking,
297 .maxlen = sizeof(int),
298 .mode = 0644,
299 .proc_handler = &proc_dointvec,
300 },
301 #endif
302 #ifdef CONFIG_LOCK_STAT
303 {
304 .ctl_name = CTL_UNNUMBERED,
305 .procname = "lock_stat",
306 .data = &lock_stat,
307 .maxlen = sizeof(int),
308 .mode = 0644,
309 .proc_handler = &proc_dointvec,
310 },
311 #endif
312 {
313 .ctl_name = KERN_PANIC,
314 .procname = "panic",
315 .data = &panic_timeout,
316 .maxlen = sizeof(int),
317 .mode = 0644,
318 .proc_handler = &proc_dointvec,
319 },
320 {
321 .ctl_name = KERN_CORE_USES_PID,
322 .procname = "core_uses_pid",
323 .data = &core_uses_pid,
324 .maxlen = sizeof(int),
325 .mode = 0644,
326 .proc_handler = &proc_dointvec,
327 },
328 #ifdef CONFIG_AUDITSYSCALL
329 {
330 .ctl_name = CTL_UNNUMBERED,
331 .procname = "audit_argv_kb",
332 .data = &audit_argv_kb,
333 .maxlen = sizeof(int),
334 .mode = 0644,
335 .proc_handler = &proc_dointvec,
336 },
337 #endif
338 {
339 .ctl_name = KERN_CORE_PATTERN,
340 .procname = "core_pattern",
341 .data = core_pattern,
342 .maxlen = CORENAME_MAX_SIZE,
343 .mode = 0644,
344 .proc_handler = &proc_dostring,
345 .strategy = &sysctl_string,
346 },
347 #ifdef CONFIG_PROC_SYSCTL
348 {
349 .ctl_name = KERN_TAINTED,
350 .procname = "tainted",
351 .data = &tainted,
352 .maxlen = sizeof(int),
353 .mode = 0644,
354 .proc_handler = &proc_dointvec_taint,
355 },
356 #endif
357 {
358 .ctl_name = KERN_CAP_BSET,
359 .procname = "cap-bound",
360 .data = &cap_bset,
361 .maxlen = sizeof(kernel_cap_t),
362 .mode = 0600,
363 .proc_handler = &proc_dointvec_bset,
364 },
365 #ifdef CONFIG_BLK_DEV_INITRD
366 {
367 .ctl_name = KERN_REALROOTDEV,
368 .procname = "real-root-dev",
369 .data = &real_root_dev,
370 .maxlen = sizeof(int),
371 .mode = 0644,
372 .proc_handler = &proc_dointvec,
373 },
374 #endif
375 {
376 .ctl_name = CTL_UNNUMBERED,
377 .procname = "print-fatal-signals",
378 .data = &print_fatal_signals,
379 .maxlen = sizeof(int),
380 .mode = 0644,
381 .proc_handler = &proc_dointvec,
382 },
383 #ifdef __sparc__
384 {
385 .ctl_name = KERN_SPARC_REBOOT,
386 .procname = "reboot-cmd",
387 .data = reboot_command,
388 .maxlen = 256,
389 .mode = 0644,
390 .proc_handler = &proc_dostring,
391 .strategy = &sysctl_string,
392 },
393 {
394 .ctl_name = KERN_SPARC_STOP_A,
395 .procname = "stop-a",
396 .data = &stop_a_enabled,
397 .maxlen = sizeof (int),
398 .mode = 0644,
399 .proc_handler = &proc_dointvec,
400 },
401 {
402 .ctl_name = KERN_SPARC_SCONS_PWROFF,
403 .procname = "scons-poweroff",
404 .data = &scons_pwroff,
405 .maxlen = sizeof (int),
406 .mode = 0644,
407 .proc_handler = &proc_dointvec,
408 },
409 #endif
410 #ifdef __hppa__
411 {
412 .ctl_name = KERN_HPPA_PWRSW,
413 .procname = "soft-power",
414 .data = &pwrsw_enabled,
415 .maxlen = sizeof (int),
416 .mode = 0644,
417 .proc_handler = &proc_dointvec,
418 },
419 {
420 .ctl_name = KERN_HPPA_UNALIGNED,
421 .procname = "unaligned-trap",
422 .data = &unaligned_enabled,
423 .maxlen = sizeof (int),
424 .mode = 0644,
425 .proc_handler = &proc_dointvec,
426 },
427 #endif
428 {
429 .ctl_name = KERN_CTLALTDEL,
430 .procname = "ctrl-alt-del",
431 .data = &C_A_D,
432 .maxlen = sizeof(int),
433 .mode = 0644,
434 .proc_handler = &proc_dointvec,
435 },
436 {
437 .ctl_name = KERN_PRINTK,
438 .procname = "printk",
439 .data = &console_loglevel,
440 .maxlen = 4*sizeof(int),
441 .mode = 0644,
442 .proc_handler = &proc_dointvec,
443 },
444 #ifdef CONFIG_KMOD
445 {
446 .ctl_name = KERN_MODPROBE,
447 .procname = "modprobe",
448 .data = &modprobe_path,
449 .maxlen = KMOD_PATH_LEN,
450 .mode = 0644,
451 .proc_handler = &proc_dostring,
452 .strategy = &sysctl_string,
453 },
454 #endif
455 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
456 {
457 .ctl_name = KERN_HOTPLUG,
458 .procname = "hotplug",
459 .data = &uevent_helper,
460 .maxlen = UEVENT_HELPER_PATH_LEN,
461 .mode = 0644,
462 .proc_handler = &proc_dostring,
463 .strategy = &sysctl_string,
464 },
465 #endif
466 #ifdef CONFIG_CHR_DEV_SG
467 {
468 .ctl_name = KERN_SG_BIG_BUFF,
469 .procname = "sg-big-buff",
470 .data = &sg_big_buff,
471 .maxlen = sizeof (int),
472 .mode = 0444,
473 .proc_handler = &proc_dointvec,
474 },
475 #endif
476 #ifdef CONFIG_BSD_PROCESS_ACCT
477 {
478 .ctl_name = KERN_ACCT,
479 .procname = "acct",
480 .data = &acct_parm,
481 .maxlen = 3*sizeof(int),
482 .mode = 0644,
483 .proc_handler = &proc_dointvec,
484 },
485 #endif
486 #ifdef CONFIG_MAGIC_SYSRQ
487 {
488 .ctl_name = KERN_SYSRQ,
489 .procname = "sysrq",
490 .data = &__sysrq_enabled,
491 .maxlen = sizeof (int),
492 .mode = 0644,
493 .proc_handler = &proc_dointvec,
494 },
495 #endif
496 #ifdef CONFIG_PROC_SYSCTL
497 {
498 .ctl_name = KERN_CADPID,
499 .procname = "cad_pid",
500 .data = NULL,
501 .maxlen = sizeof (int),
502 .mode = 0600,
503 .proc_handler = &proc_do_cad_pid,
504 },
505 #endif
506 {
507 .ctl_name = KERN_MAX_THREADS,
508 .procname = "threads-max",
509 .data = &max_threads,
510 .maxlen = sizeof(int),
511 .mode = 0644,
512 .proc_handler = &proc_dointvec,
513 },
514 {
515 .ctl_name = KERN_RANDOM,
516 .procname = "random",
517 .mode = 0555,
518 .child = random_table,
519 },
520 #ifdef CONFIG_UNIX98_PTYS
521 {
522 .ctl_name = KERN_PTY,
523 .procname = "pty",
524 .mode = 0555,
525 .child = pty_table,
526 },
527 #endif
528 {
529 .ctl_name = KERN_OVERFLOWUID,
530 .procname = "overflowuid",
531 .data = &overflowuid,
532 .maxlen = sizeof(int),
533 .mode = 0644,
534 .proc_handler = &proc_dointvec_minmax,
535 .strategy = &sysctl_intvec,
536 .extra1 = &minolduid,
537 .extra2 = &maxolduid,
538 },
539 {
540 .ctl_name = KERN_OVERFLOWGID,
541 .procname = "overflowgid",
542 .data = &overflowgid,
543 .maxlen = sizeof(int),
544 .mode = 0644,
545 .proc_handler = &proc_dointvec_minmax,
546 .strategy = &sysctl_intvec,
547 .extra1 = &minolduid,
548 .extra2 = &maxolduid,
549 },
550 #ifdef CONFIG_S390
551 #ifdef CONFIG_MATHEMU
552 {
553 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
554 .procname = "ieee_emulation_warnings",
555 .data = &sysctl_ieee_emulation_warnings,
556 .maxlen = sizeof(int),
557 .mode = 0644,
558 .proc_handler = &proc_dointvec,
559 },
560 #endif
561 #ifdef CONFIG_NO_IDLE_HZ
562 {
563 .ctl_name = KERN_HZ_TIMER,
564 .procname = "hz_timer",
565 .data = &sysctl_hz_timer,
566 .maxlen = sizeof(int),
567 .mode = 0644,
568 .proc_handler = &proc_dointvec,
569 },
570 #endif
571 {
572 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
573 .procname = "userprocess_debug",
574 .data = &sysctl_userprocess_debug,
575 .maxlen = sizeof(int),
576 .mode = 0644,
577 .proc_handler = &proc_dointvec,
578 },
579 #endif
580 {
581 .ctl_name = KERN_PIDMAX,
582 .procname = "pid_max",
583 .data = &pid_max,
584 .maxlen = sizeof (int),
585 .mode = 0644,
586 .proc_handler = &proc_dointvec_minmax,
587 .strategy = sysctl_intvec,
588 .extra1 = &pid_max_min,
589 .extra2 = &pid_max_max,
590 },
591 {
592 .ctl_name = KERN_PANIC_ON_OOPS,
593 .procname = "panic_on_oops",
594 .data = &panic_on_oops,
595 .maxlen = sizeof(int),
596 .mode = 0644,
597 .proc_handler = &proc_dointvec,
598 },
599 {
600 .ctl_name = KERN_PRINTK_RATELIMIT,
601 .procname = "printk_ratelimit",
602 .data = &printk_ratelimit_jiffies,
603 .maxlen = sizeof(int),
604 .mode = 0644,
605 .proc_handler = &proc_dointvec_jiffies,
606 .strategy = &sysctl_jiffies,
607 },
608 {
609 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
610 .procname = "printk_ratelimit_burst",
611 .data = &printk_ratelimit_burst,
612 .maxlen = sizeof(int),
613 .mode = 0644,
614 .proc_handler = &proc_dointvec,
615 },
616 {
617 .ctl_name = KERN_NGROUPS_MAX,
618 .procname = "ngroups_max",
619 .data = &ngroups_max,
620 .maxlen = sizeof (int),
621 .mode = 0444,
622 .proc_handler = &proc_dointvec,
623 },
624 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
625 {
626 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
627 .procname = "unknown_nmi_panic",
628 .data = &unknown_nmi_panic,
629 .maxlen = sizeof (int),
630 .mode = 0644,
631 .proc_handler = &proc_dointvec,
632 },
633 {
634 .ctl_name = KERN_NMI_WATCHDOG,
635 .procname = "nmi_watchdog",
636 .data = &nmi_watchdog_enabled,
637 .maxlen = sizeof (int),
638 .mode = 0644,
639 .proc_handler = &proc_nmi_enabled,
640 },
641 #endif
642 #if defined(CONFIG_X86)
643 {
644 .ctl_name = KERN_PANIC_ON_NMI,
645 .procname = "panic_on_unrecovered_nmi",
646 .data = &panic_on_unrecovered_nmi,
647 .maxlen = sizeof(int),
648 .mode = 0644,
649 .proc_handler = &proc_dointvec,
650 },
651 {
652 .ctl_name = KERN_BOOTLOADER_TYPE,
653 .procname = "bootloader_type",
654 .data = &bootloader_type,
655 .maxlen = sizeof (int),
656 .mode = 0444,
657 .proc_handler = &proc_dointvec,
658 },
659 {
660 .ctl_name = CTL_UNNUMBERED,
661 .procname = "kstack_depth_to_print",
662 .data = &kstack_depth_to_print,
663 .maxlen = sizeof(int),
664 .mode = 0644,
665 .proc_handler = &proc_dointvec,
666 },
667 #endif
668 #if defined(CONFIG_MMU)
669 {
670 .ctl_name = KERN_RANDOMIZE,
671 .procname = "randomize_va_space",
672 .data = &randomize_va_space,
673 .maxlen = sizeof(int),
674 .mode = 0644,
675 .proc_handler = &proc_dointvec,
676 },
677 #endif
678 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
679 {
680 .ctl_name = KERN_SPIN_RETRY,
681 .procname = "spin_retry",
682 .data = &spin_retry,
683 .maxlen = sizeof (int),
684 .mode = 0644,
685 .proc_handler = &proc_dointvec,
686 },
687 #endif
688 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
689 {
690 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
691 .procname = "acpi_video_flags",
692 .data = &acpi_realmode_flags,
693 .maxlen = sizeof (unsigned long),
694 .mode = 0644,
695 .proc_handler = &proc_doulongvec_minmax,
696 },
697 #endif
698 #ifdef CONFIG_IA64
699 {
700 .ctl_name = KERN_IA64_UNALIGNED,
701 .procname = "ignore-unaligned-usertrap",
702 .data = &no_unaligned_warning,
703 .maxlen = sizeof (int),
704 .mode = 0644,
705 .proc_handler = &proc_dointvec,
706 },
707 #endif
708 #ifdef CONFIG_COMPAT
709 {
710 .ctl_name = KERN_COMPAT_LOG,
711 .procname = "compat-log",
712 .data = &compat_log,
713 .maxlen = sizeof (int),
714 .mode = 0644,
715 .proc_handler = &proc_dointvec,
716 },
717 #endif
718 #ifdef CONFIG_RT_MUTEXES
719 {
720 .ctl_name = KERN_MAX_LOCK_DEPTH,
721 .procname = "max_lock_depth",
722 .data = &max_lock_depth,
723 .maxlen = sizeof(int),
724 .mode = 0644,
725 .proc_handler = &proc_dointvec,
726 },
727 #endif
728 #ifdef CONFIG_PROC_FS
729 {
730 .ctl_name = CTL_UNNUMBERED,
731 .procname = "maps_protect",
732 .data = &maps_protect,
733 .maxlen = sizeof(int),
734 .mode = 0644,
735 .proc_handler = &proc_dointvec,
736 },
737 #endif
738 {
739 .ctl_name = CTL_UNNUMBERED,
740 .procname = "poweroff_cmd",
741 .data = &poweroff_cmd,
742 .maxlen = POWEROFF_CMD_PATH_LEN,
743 .mode = 0644,
744 .proc_handler = &proc_dostring,
745 .strategy = &sysctl_string,
746 },
747 /*
748 * NOTE: do not add new entries to this table unless you have read
749 * Documentation/sysctl/ctl_unnumbered.txt
750 */
751 { .ctl_name = 0 }
752 };
753
754 /* Constants for minimum and maximum testing in vm_table.
755 We use these as one-element integer vectors. */
756 static int zero;
757 static int two = 2;
758 static int one_hundred = 100;
759
760
761 static ctl_table vm_table[] = {
762 {
763 .ctl_name = VM_OVERCOMMIT_MEMORY,
764 .procname = "overcommit_memory",
765 .data = &sysctl_overcommit_memory,
766 .maxlen = sizeof(sysctl_overcommit_memory),
767 .mode = 0644,
768 .proc_handler = &proc_dointvec,
769 },
770 {
771 .ctl_name = VM_PANIC_ON_OOM,
772 .procname = "panic_on_oom",
773 .data = &sysctl_panic_on_oom,
774 .maxlen = sizeof(sysctl_panic_on_oom),
775 .mode = 0644,
776 .proc_handler = &proc_dointvec,
777 },
778 {
779 .ctl_name = VM_OVERCOMMIT_RATIO,
780 .procname = "overcommit_ratio",
781 .data = &sysctl_overcommit_ratio,
782 .maxlen = sizeof(sysctl_overcommit_ratio),
783 .mode = 0644,
784 .proc_handler = &proc_dointvec,
785 },
786 {
787 .ctl_name = VM_PAGE_CLUSTER,
788 .procname = "page-cluster",
789 .data = &page_cluster,
790 .maxlen = sizeof(int),
791 .mode = 0644,
792 .proc_handler = &proc_dointvec,
793 },
794 {
795 .ctl_name = VM_DIRTY_BACKGROUND,
796 .procname = "dirty_background_ratio",
797 .data = &dirty_background_ratio,
798 .maxlen = sizeof(dirty_background_ratio),
799 .mode = 0644,
800 .proc_handler = &proc_dointvec_minmax,
801 .strategy = &sysctl_intvec,
802 .extra1 = &zero,
803 .extra2 = &one_hundred,
804 },
805 {
806 .ctl_name = VM_DIRTY_RATIO,
807 .procname = "dirty_ratio",
808 .data = &vm_dirty_ratio,
809 .maxlen = sizeof(vm_dirty_ratio),
810 .mode = 0644,
811 .proc_handler = &proc_dointvec_minmax,
812 .strategy = &sysctl_intvec,
813 .extra1 = &zero,
814 .extra2 = &one_hundred,
815 },
816 {
817 .ctl_name = VM_DIRTY_WB_CS,
818 .procname = "dirty_writeback_centisecs",
819 .data = &dirty_writeback_interval,
820 .maxlen = sizeof(dirty_writeback_interval),
821 .mode = 0644,
822 .proc_handler = &dirty_writeback_centisecs_handler,
823 },
824 {
825 .ctl_name = VM_DIRTY_EXPIRE_CS,
826 .procname = "dirty_expire_centisecs",
827 .data = &dirty_expire_interval,
828 .maxlen = sizeof(dirty_expire_interval),
829 .mode = 0644,
830 .proc_handler = &proc_dointvec_userhz_jiffies,
831 },
832 {
833 .ctl_name = VM_NR_PDFLUSH_THREADS,
834 .procname = "nr_pdflush_threads",
835 .data = &nr_pdflush_threads,
836 .maxlen = sizeof nr_pdflush_threads,
837 .mode = 0444 /* read-only*/,
838 .proc_handler = &proc_dointvec,
839 },
840 {
841 .ctl_name = VM_SWAPPINESS,
842 .procname = "swappiness",
843 .data = &vm_swappiness,
844 .maxlen = sizeof(vm_swappiness),
845 .mode = 0644,
846 .proc_handler = &proc_dointvec_minmax,
847 .strategy = &sysctl_intvec,
848 .extra1 = &zero,
849 .extra2 = &one_hundred,
850 },
851 #ifdef CONFIG_HUGETLB_PAGE
852 {
853 .ctl_name = VM_HUGETLB_PAGES,
854 .procname = "nr_hugepages",
855 .data = &max_huge_pages,
856 .maxlen = sizeof(unsigned long),
857 .mode = 0644,
858 .proc_handler = &hugetlb_sysctl_handler,
859 .extra1 = (void *)&hugetlb_zero,
860 .extra2 = (void *)&hugetlb_infinity,
861 },
862 {
863 .ctl_name = VM_HUGETLB_GROUP,
864 .procname = "hugetlb_shm_group",
865 .data = &sysctl_hugetlb_shm_group,
866 .maxlen = sizeof(gid_t),
867 .mode = 0644,
868 .proc_handler = &proc_dointvec,
869 },
870 {
871 .ctl_name = CTL_UNNUMBERED,
872 .procname = "hugepages_treat_as_movable",
873 .data = &hugepages_treat_as_movable,
874 .maxlen = sizeof(int),
875 .mode = 0644,
876 .proc_handler = &hugetlb_treat_movable_handler,
877 },
878 #endif
879 {
880 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
881 .procname = "lowmem_reserve_ratio",
882 .data = &sysctl_lowmem_reserve_ratio,
883 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
884 .mode = 0644,
885 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
886 .strategy = &sysctl_intvec,
887 },
888 {
889 .ctl_name = VM_DROP_PAGECACHE,
890 .procname = "drop_caches",
891 .data = &sysctl_drop_caches,
892 .maxlen = sizeof(int),
893 .mode = 0644,
894 .proc_handler = drop_caches_sysctl_handler,
895 .strategy = &sysctl_intvec,
896 },
897 {
898 .ctl_name = VM_MIN_FREE_KBYTES,
899 .procname = "min_free_kbytes",
900 .data = &min_free_kbytes,
901 .maxlen = sizeof(min_free_kbytes),
902 .mode = 0644,
903 .proc_handler = &min_free_kbytes_sysctl_handler,
904 .strategy = &sysctl_intvec,
905 .extra1 = &zero,
906 },
907 {
908 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
909 .procname = "percpu_pagelist_fraction",
910 .data = &percpu_pagelist_fraction,
911 .maxlen = sizeof(percpu_pagelist_fraction),
912 .mode = 0644,
913 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
914 .strategy = &sysctl_intvec,
915 .extra1 = &min_percpu_pagelist_fract,
916 },
917 #ifdef CONFIG_MMU
918 {
919 .ctl_name = VM_MAX_MAP_COUNT,
920 .procname = "max_map_count",
921 .data = &sysctl_max_map_count,
922 .maxlen = sizeof(sysctl_max_map_count),
923 .mode = 0644,
924 .proc_handler = &proc_dointvec
925 },
926 #endif
927 {
928 .ctl_name = VM_LAPTOP_MODE,
929 .procname = "laptop_mode",
930 .data = &laptop_mode,
931 .maxlen = sizeof(laptop_mode),
932 .mode = 0644,
933 .proc_handler = &proc_dointvec_jiffies,
934 .strategy = &sysctl_jiffies,
935 },
936 {
937 .ctl_name = VM_BLOCK_DUMP,
938 .procname = "block_dump",
939 .data = &block_dump,
940 .maxlen = sizeof(block_dump),
941 .mode = 0644,
942 .proc_handler = &proc_dointvec,
943 .strategy = &sysctl_intvec,
944 .extra1 = &zero,
945 },
946 {
947 .ctl_name = VM_VFS_CACHE_PRESSURE,
948 .procname = "vfs_cache_pressure",
949 .data = &sysctl_vfs_cache_pressure,
950 .maxlen = sizeof(sysctl_vfs_cache_pressure),
951 .mode = 0644,
952 .proc_handler = &proc_dointvec,
953 .strategy = &sysctl_intvec,
954 .extra1 = &zero,
955 },
956 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
957 {
958 .ctl_name = VM_LEGACY_VA_LAYOUT,
959 .procname = "legacy_va_layout",
960 .data = &sysctl_legacy_va_layout,
961 .maxlen = sizeof(sysctl_legacy_va_layout),
962 .mode = 0644,
963 .proc_handler = &proc_dointvec,
964 .strategy = &sysctl_intvec,
965 .extra1 = &zero,
966 },
967 #endif
968 #ifdef CONFIG_NUMA
969 {
970 .ctl_name = VM_ZONE_RECLAIM_MODE,
971 .procname = "zone_reclaim_mode",
972 .data = &zone_reclaim_mode,
973 .maxlen = sizeof(zone_reclaim_mode),
974 .mode = 0644,
975 .proc_handler = &proc_dointvec,
976 .strategy = &sysctl_intvec,
977 .extra1 = &zero,
978 },
979 {
980 .ctl_name = VM_MIN_UNMAPPED,
981 .procname = "min_unmapped_ratio",
982 .data = &sysctl_min_unmapped_ratio,
983 .maxlen = sizeof(sysctl_min_unmapped_ratio),
984 .mode = 0644,
985 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
986 .strategy = &sysctl_intvec,
987 .extra1 = &zero,
988 .extra2 = &one_hundred,
989 },
990 {
991 .ctl_name = VM_MIN_SLAB,
992 .procname = "min_slab_ratio",
993 .data = &sysctl_min_slab_ratio,
994 .maxlen = sizeof(sysctl_min_slab_ratio),
995 .mode = 0644,
996 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
997 .strategy = &sysctl_intvec,
998 .extra1 = &zero,
999 .extra2 = &one_hundred,
1000 },
1001 #endif
1002 #ifdef CONFIG_SMP
1003 {
1004 .ctl_name = CTL_UNNUMBERED,
1005 .procname = "stat_interval",
1006 .data = &sysctl_stat_interval,
1007 .maxlen = sizeof(sysctl_stat_interval),
1008 .mode = 0644,
1009 .proc_handler = &proc_dointvec_jiffies,
1010 .strategy = &sysctl_jiffies,
1011 },
1012 #endif
1013 #ifdef CONFIG_SECURITY
1014 {
1015 .ctl_name = CTL_UNNUMBERED,
1016 .procname = "mmap_min_addr",
1017 .data = &mmap_min_addr,
1018 .maxlen = sizeof(unsigned long),
1019 .mode = 0644,
1020 .proc_handler = &proc_doulongvec_minmax,
1021 },
1022 #endif
1023 #ifdef CONFIG_NUMA
1024 {
1025 .ctl_name = CTL_UNNUMBERED,
1026 .procname = "numa_zonelist_order",
1027 .data = &numa_zonelist_order,
1028 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1029 .mode = 0644,
1030 .proc_handler = &numa_zonelist_order_handler,
1031 .strategy = &sysctl_string,
1032 },
1033 #endif
1034 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1035 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1036 {
1037 .ctl_name = VM_VDSO_ENABLED,
1038 .procname = "vdso_enabled",
1039 .data = &vdso_enabled,
1040 .maxlen = sizeof(vdso_enabled),
1041 .mode = 0644,
1042 .proc_handler = &proc_dointvec,
1043 .strategy = &sysctl_intvec,
1044 .extra1 = &zero,
1045 },
1046 #endif
1047 /*
1048 * NOTE: do not add new entries to this table unless you have read
1049 * Documentation/sysctl/ctl_unnumbered.txt
1050 */
1051 { .ctl_name = 0 }
1052 };
1053
1054 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1055 static ctl_table binfmt_misc_table[] = {
1056 { .ctl_name = 0 }
1057 };
1058 #endif
1059
1060 static ctl_table fs_table[] = {
1061 {
1062 .ctl_name = FS_NRINODE,
1063 .procname = "inode-nr",
1064 .data = &inodes_stat,
1065 .maxlen = 2*sizeof(int),
1066 .mode = 0444,
1067 .proc_handler = &proc_dointvec,
1068 },
1069 {
1070 .ctl_name = FS_STATINODE,
1071 .procname = "inode-state",
1072 .data = &inodes_stat,
1073 .maxlen = 7*sizeof(int),
1074 .mode = 0444,
1075 .proc_handler = &proc_dointvec,
1076 },
1077 {
1078 .ctl_name = FS_NRFILE,
1079 .procname = "file-nr",
1080 .data = &files_stat,
1081 .maxlen = 3*sizeof(int),
1082 .mode = 0444,
1083 .proc_handler = &proc_nr_files,
1084 },
1085 {
1086 .ctl_name = FS_MAXFILE,
1087 .procname = "file-max",
1088 .data = &files_stat.max_files,
1089 .maxlen = sizeof(int),
1090 .mode = 0644,
1091 .proc_handler = &proc_dointvec,
1092 },
1093 {
1094 .ctl_name = FS_DENTRY,
1095 .procname = "dentry-state",
1096 .data = &dentry_stat,
1097 .maxlen = 6*sizeof(int),
1098 .mode = 0444,
1099 .proc_handler = &proc_dointvec,
1100 },
1101 {
1102 .ctl_name = FS_OVERFLOWUID,
1103 .procname = "overflowuid",
1104 .data = &fs_overflowuid,
1105 .maxlen = sizeof(int),
1106 .mode = 0644,
1107 .proc_handler = &proc_dointvec_minmax,
1108 .strategy = &sysctl_intvec,
1109 .extra1 = &minolduid,
1110 .extra2 = &maxolduid,
1111 },
1112 {
1113 .ctl_name = FS_OVERFLOWGID,
1114 .procname = "overflowgid",
1115 .data = &fs_overflowgid,
1116 .maxlen = sizeof(int),
1117 .mode = 0644,
1118 .proc_handler = &proc_dointvec_minmax,
1119 .strategy = &sysctl_intvec,
1120 .extra1 = &minolduid,
1121 .extra2 = &maxolduid,
1122 },
1123 {
1124 .ctl_name = FS_LEASES,
1125 .procname = "leases-enable",
1126 .data = &leases_enable,
1127 .maxlen = sizeof(int),
1128 .mode = 0644,
1129 .proc_handler = &proc_dointvec,
1130 },
1131 #ifdef CONFIG_DNOTIFY
1132 {
1133 .ctl_name = FS_DIR_NOTIFY,
1134 .procname = "dir-notify-enable",
1135 .data = &dir_notify_enable,
1136 .maxlen = sizeof(int),
1137 .mode = 0644,
1138 .proc_handler = &proc_dointvec,
1139 },
1140 #endif
1141 #ifdef CONFIG_MMU
1142 {
1143 .ctl_name = FS_LEASE_TIME,
1144 .procname = "lease-break-time",
1145 .data = &lease_break_time,
1146 .maxlen = sizeof(int),
1147 .mode = 0644,
1148 .proc_handler = &proc_dointvec_minmax,
1149 .strategy = &sysctl_intvec,
1150 .extra1 = &zero,
1151 .extra2 = &two,
1152 },
1153 {
1154 .ctl_name = FS_AIO_NR,
1155 .procname = "aio-nr",
1156 .data = &aio_nr,
1157 .maxlen = sizeof(aio_nr),
1158 .mode = 0444,
1159 .proc_handler = &proc_doulongvec_minmax,
1160 },
1161 {
1162 .ctl_name = FS_AIO_MAX_NR,
1163 .procname = "aio-max-nr",
1164 .data = &aio_max_nr,
1165 .maxlen = sizeof(aio_max_nr),
1166 .mode = 0644,
1167 .proc_handler = &proc_doulongvec_minmax,
1168 },
1169 #ifdef CONFIG_INOTIFY_USER
1170 {
1171 .ctl_name = FS_INOTIFY,
1172 .procname = "inotify",
1173 .mode = 0555,
1174 .child = inotify_table,
1175 },
1176 #endif
1177 #endif
1178 {
1179 .ctl_name = KERN_SETUID_DUMPABLE,
1180 .procname = "suid_dumpable",
1181 .data = &suid_dumpable,
1182 .maxlen = sizeof(int),
1183 .mode = 0644,
1184 .proc_handler = &proc_dointvec,
1185 },
1186 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1187 {
1188 .ctl_name = CTL_UNNUMBERED,
1189 .procname = "binfmt_misc",
1190 .mode = 0555,
1191 .child = binfmt_misc_table,
1192 },
1193 #endif
1194 /*
1195 * NOTE: do not add new entries to this table unless you have read
1196 * Documentation/sysctl/ctl_unnumbered.txt
1197 */
1198 { .ctl_name = 0 }
1199 };
1200
1201 static ctl_table debug_table[] = {
1202 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1203 {
1204 .ctl_name = CTL_UNNUMBERED,
1205 .procname = "exception-trace",
1206 .data = &show_unhandled_signals,
1207 .maxlen = sizeof(int),
1208 .mode = 0644,
1209 .proc_handler = proc_dointvec
1210 },
1211 #endif
1212 { .ctl_name = 0 }
1213 };
1214
1215 static ctl_table dev_table[] = {
1216 { .ctl_name = 0 }
1217 };
1218
1219 static DEFINE_SPINLOCK(sysctl_lock);
1220
1221 /* called under sysctl_lock */
1222 static int use_table(struct ctl_table_header *p)
1223 {
1224 if (unlikely(p->unregistering))
1225 return 0;
1226 p->used++;
1227 return 1;
1228 }
1229
1230 /* called under sysctl_lock */
1231 static void unuse_table(struct ctl_table_header *p)
1232 {
1233 if (!--p->used)
1234 if (unlikely(p->unregistering))
1235 complete(p->unregistering);
1236 }
1237
1238 /* called under sysctl_lock, will reacquire if has to wait */
1239 static void start_unregistering(struct ctl_table_header *p)
1240 {
1241 /*
1242 * if p->used is 0, nobody will ever touch that entry again;
1243 * we'll eliminate all paths to it before dropping sysctl_lock
1244 */
1245 if (unlikely(p->used)) {
1246 struct completion wait;
1247 init_completion(&wait);
1248 p->unregistering = &wait;
1249 spin_unlock(&sysctl_lock);
1250 wait_for_completion(&wait);
1251 spin_lock(&sysctl_lock);
1252 }
1253 /*
1254 * do not remove from the list until nobody holds it; walking the
1255 * list in do_sysctl() relies on that.
1256 */
1257 list_del_init(&p->ctl_entry);
1258 }
1259
1260 void sysctl_head_finish(struct ctl_table_header *head)
1261 {
1262 if (!head)
1263 return;
1264 spin_lock(&sysctl_lock);
1265 unuse_table(head);
1266 spin_unlock(&sysctl_lock);
1267 }
1268
1269 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1270 {
1271 struct ctl_table_header *head;
1272 struct list_head *tmp;
1273 spin_lock(&sysctl_lock);
1274 if (prev) {
1275 tmp = &prev->ctl_entry;
1276 unuse_table(prev);
1277 goto next;
1278 }
1279 tmp = &root_table_header.ctl_entry;
1280 for (;;) {
1281 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1282
1283 if (!use_table(head))
1284 goto next;
1285 spin_unlock(&sysctl_lock);
1286 return head;
1287 next:
1288 tmp = tmp->next;
1289 if (tmp == &root_table_header.ctl_entry)
1290 break;
1291 }
1292 spin_unlock(&sysctl_lock);
1293 return NULL;
1294 }
1295
1296 #ifdef CONFIG_SYSCTL_SYSCALL
1297 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1298 void __user *newval, size_t newlen)
1299 {
1300 struct ctl_table_header *head;
1301 int error = -ENOTDIR;
1302
1303 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1304 return -ENOTDIR;
1305 if (oldval) {
1306 int old_len;
1307 if (!oldlenp || get_user(old_len, oldlenp))
1308 return -EFAULT;
1309 }
1310
1311 for (head = sysctl_head_next(NULL); head;
1312 head = sysctl_head_next(head)) {
1313 error = parse_table(name, nlen, oldval, oldlenp,
1314 newval, newlen, head->ctl_table);
1315 if (error != -ENOTDIR) {
1316 sysctl_head_finish(head);
1317 break;
1318 }
1319 }
1320 return error;
1321 }
1322
1323 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1324 {
1325 struct __sysctl_args tmp;
1326 int error;
1327
1328 if (copy_from_user(&tmp, args, sizeof(tmp)))
1329 return -EFAULT;
1330
1331 lock_kernel();
1332 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1333 tmp.newval, tmp.newlen);
1334 unlock_kernel();
1335 return error;
1336 }
1337 #endif /* CONFIG_SYSCTL_SYSCALL */
1338
1339 /*
1340 * sysctl_perm does NOT grant the superuser all rights automatically, because
1341 * some sysctl variables are readonly even to root.
1342 */
1343
1344 static int test_perm(int mode, int op)
1345 {
1346 if (!current->euid)
1347 mode >>= 6;
1348 else if (in_egroup_p(0))
1349 mode >>= 3;
1350 if ((mode & op & 0007) == op)
1351 return 0;
1352 return -EACCES;
1353 }
1354
1355 int sysctl_perm(ctl_table *table, int op)
1356 {
1357 int error;
1358 error = security_sysctl(table, op);
1359 if (error)
1360 return error;
1361 return test_perm(table->mode, op);
1362 }
1363
1364 #ifdef CONFIG_SYSCTL_SYSCALL
1365 static int parse_table(int __user *name, int nlen,
1366 void __user *oldval, size_t __user *oldlenp,
1367 void __user *newval, size_t newlen,
1368 ctl_table *table)
1369 {
1370 int n;
1371 repeat:
1372 if (!nlen)
1373 return -ENOTDIR;
1374 if (get_user(n, name))
1375 return -EFAULT;
1376 for ( ; table->ctl_name || table->procname; table++) {
1377 if (!table->ctl_name)
1378 continue;
1379 if (n == table->ctl_name) {
1380 int error;
1381 if (table->child) {
1382 if (sysctl_perm(table, 001))
1383 return -EPERM;
1384 name++;
1385 nlen--;
1386 table = table->child;
1387 goto repeat;
1388 }
1389 error = do_sysctl_strategy(table, name, nlen,
1390 oldval, oldlenp,
1391 newval, newlen);
1392 return error;
1393 }
1394 }
1395 return -ENOTDIR;
1396 }
1397
1398 /* Perform the actual read/write of a sysctl table entry. */
1399 int do_sysctl_strategy (ctl_table *table,
1400 int __user *name, int nlen,
1401 void __user *oldval, size_t __user *oldlenp,
1402 void __user *newval, size_t newlen)
1403 {
1404 int op = 0, rc;
1405 size_t len;
1406
1407 if (oldval)
1408 op |= 004;
1409 if (newval)
1410 op |= 002;
1411 if (sysctl_perm(table, op))
1412 return -EPERM;
1413
1414 if (table->strategy) {
1415 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1416 newval, newlen);
1417 if (rc < 0)
1418 return rc;
1419 if (rc > 0)
1420 return 0;
1421 }
1422
1423 /* If there is no strategy routine, or if the strategy returns
1424 * zero, proceed with automatic r/w */
1425 if (table->data && table->maxlen) {
1426 if (oldval && oldlenp) {
1427 if (get_user(len, oldlenp))
1428 return -EFAULT;
1429 if (len) {
1430 if (len > table->maxlen)
1431 len = table->maxlen;
1432 if(copy_to_user(oldval, table->data, len))
1433 return -EFAULT;
1434 if(put_user(len, oldlenp))
1435 return -EFAULT;
1436 }
1437 }
1438 if (newval && newlen) {
1439 len = newlen;
1440 if (len > table->maxlen)
1441 len = table->maxlen;
1442 if(copy_from_user(table->data, newval, len))
1443 return -EFAULT;
1444 }
1445 }
1446 return 0;
1447 }
1448 #endif /* CONFIG_SYSCTL_SYSCALL */
1449
1450 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1451 {
1452 for (; table->ctl_name || table->procname; table++) {
1453 table->parent = parent;
1454 if (table->child)
1455 sysctl_set_parent(table, table->child);
1456 }
1457 }
1458
1459 static __init int sysctl_init(void)
1460 {
1461 sysctl_set_parent(NULL, root_table);
1462 return 0;
1463 }
1464
1465 core_initcall(sysctl_init);
1466
1467 /**
1468 * register_sysctl_table - register a sysctl hierarchy
1469 * @table: the top-level table structure
1470 *
1471 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1472 * array. An entry with a ctl_name of 0 terminates the table.
1473 *
1474 * The members of the &ctl_table structure are used as follows:
1475 *
1476 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1477 * must be unique within that level of sysctl
1478 *
1479 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1480 * enter a sysctl file
1481 *
1482 * data - a pointer to data for use by proc_handler
1483 *
1484 * maxlen - the maximum size in bytes of the data
1485 *
1486 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1487 *
1488 * child - a pointer to the child sysctl table if this entry is a directory, or
1489 * %NULL.
1490 *
1491 * proc_handler - the text handler routine (described below)
1492 *
1493 * strategy - the strategy routine (described below)
1494 *
1495 * de - for internal use by the sysctl routines
1496 *
1497 * extra1, extra2 - extra pointers usable by the proc handler routines
1498 *
1499 * Leaf nodes in the sysctl tree will be represented by a single file
1500 * under /proc; non-leaf nodes will be represented by directories.
1501 *
1502 * sysctl(2) can automatically manage read and write requests through
1503 * the sysctl table. The data and maxlen fields of the ctl_table
1504 * struct enable minimal validation of the values being written to be
1505 * performed, and the mode field allows minimal authentication.
1506 *
1507 * More sophisticated management can be enabled by the provision of a
1508 * strategy routine with the table entry. This will be called before
1509 * any automatic read or write of the data is performed.
1510 *
1511 * The strategy routine may return
1512 *
1513 * < 0 - Error occurred (error is passed to user process)
1514 *
1515 * 0 - OK - proceed with automatic read or write.
1516 *
1517 * > 0 - OK - read or write has been done by the strategy routine, so
1518 * return immediately.
1519 *
1520 * There must be a proc_handler routine for any terminal nodes
1521 * mirrored under /proc/sys (non-terminals are handled by a built-in
1522 * directory handler). Several default handlers are available to
1523 * cover common cases -
1524 *
1525 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1526 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1527 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1528 *
1529 * It is the handler's job to read the input buffer from user memory
1530 * and process it. The handler should return 0 on success.
1531 *
1532 * This routine returns %NULL on a failure to register, and a pointer
1533 * to the table header on success.
1534 */
1535 struct ctl_table_header *register_sysctl_table(ctl_table * table)
1536 {
1537 struct ctl_table_header *tmp;
1538 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1539 if (!tmp)
1540 return NULL;
1541 tmp->ctl_table = table;
1542 INIT_LIST_HEAD(&tmp->ctl_entry);
1543 tmp->used = 0;
1544 tmp->unregistering = NULL;
1545 sysctl_set_parent(NULL, table);
1546 spin_lock(&sysctl_lock);
1547 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
1548 spin_unlock(&sysctl_lock);
1549 return tmp;
1550 }
1551
1552 /**
1553 * unregister_sysctl_table - unregister a sysctl table hierarchy
1554 * @header: the header returned from register_sysctl_table
1555 *
1556 * Unregisters the sysctl table and all children. proc entries may not
1557 * actually be removed until they are no longer used by anyone.
1558 */
1559 void unregister_sysctl_table(struct ctl_table_header * header)
1560 {
1561 might_sleep();
1562 spin_lock(&sysctl_lock);
1563 start_unregistering(header);
1564 spin_unlock(&sysctl_lock);
1565 kfree(header);
1566 }
1567
1568 #else /* !CONFIG_SYSCTL */
1569 struct ctl_table_header *register_sysctl_table(ctl_table * table)
1570 {
1571 return NULL;
1572 }
1573
1574 void unregister_sysctl_table(struct ctl_table_header * table)
1575 {
1576 }
1577
1578 #endif /* CONFIG_SYSCTL */
1579
1580 /*
1581 * /proc/sys support
1582 */
1583
1584 #ifdef CONFIG_PROC_SYSCTL
1585
1586 static int _proc_do_string(void* data, int maxlen, int write,
1587 struct file *filp, void __user *buffer,
1588 size_t *lenp, loff_t *ppos)
1589 {
1590 size_t len;
1591 char __user *p;
1592 char c;
1593
1594 if (!data || !maxlen || !*lenp) {
1595 *lenp = 0;
1596 return 0;
1597 }
1598
1599 if (write) {
1600 len = 0;
1601 p = buffer;
1602 while (len < *lenp) {
1603 if (get_user(c, p++))
1604 return -EFAULT;
1605 if (c == 0 || c == '\n')
1606 break;
1607 len++;
1608 }
1609 if (len >= maxlen)
1610 len = maxlen-1;
1611 if(copy_from_user(data, buffer, len))
1612 return -EFAULT;
1613 ((char *) data)[len] = 0;
1614 *ppos += *lenp;
1615 } else {
1616 len = strlen(data);
1617 if (len > maxlen)
1618 len = maxlen;
1619
1620 if (*ppos > len) {
1621 *lenp = 0;
1622 return 0;
1623 }
1624
1625 data += *ppos;
1626 len -= *ppos;
1627
1628 if (len > *lenp)
1629 len = *lenp;
1630 if (len)
1631 if(copy_to_user(buffer, data, len))
1632 return -EFAULT;
1633 if (len < *lenp) {
1634 if(put_user('\n', ((char __user *) buffer) + len))
1635 return -EFAULT;
1636 len++;
1637 }
1638 *lenp = len;
1639 *ppos += len;
1640 }
1641 return 0;
1642 }
1643
1644 /**
1645 * proc_dostring - read a string sysctl
1646 * @table: the sysctl table
1647 * @write: %TRUE if this is a write to the sysctl file
1648 * @filp: the file structure
1649 * @buffer: the user buffer
1650 * @lenp: the size of the user buffer
1651 * @ppos: file position
1652 *
1653 * Reads/writes a string from/to the user buffer. If the kernel
1654 * buffer provided is not large enough to hold the string, the
1655 * string is truncated. The copied string is %NULL-terminated.
1656 * If the string is being read by the user process, it is copied
1657 * and a newline '\n' is added. It is truncated if the buffer is
1658 * not large enough.
1659 *
1660 * Returns 0 on success.
1661 */
1662 int proc_dostring(ctl_table *table, int write, struct file *filp,
1663 void __user *buffer, size_t *lenp, loff_t *ppos)
1664 {
1665 return _proc_do_string(table->data, table->maxlen, write, filp,
1666 buffer, lenp, ppos);
1667 }
1668
1669
1670 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1671 int *valp,
1672 int write, void *data)
1673 {
1674 if (write) {
1675 *valp = *negp ? -*lvalp : *lvalp;
1676 } else {
1677 int val = *valp;
1678 if (val < 0) {
1679 *negp = -1;
1680 *lvalp = (unsigned long)-val;
1681 } else {
1682 *negp = 0;
1683 *lvalp = (unsigned long)val;
1684 }
1685 }
1686 return 0;
1687 }
1688
1689 static int __do_proc_dointvec(void *tbl_data, ctl_table *table,
1690 int write, struct file *filp, void __user *buffer,
1691 size_t *lenp, loff_t *ppos,
1692 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1693 int write, void *data),
1694 void *data)
1695 {
1696 #define TMPBUFLEN 21
1697 int *i, vleft, first=1, neg, val;
1698 unsigned long lval;
1699 size_t left, len;
1700
1701 char buf[TMPBUFLEN], *p;
1702 char __user *s = buffer;
1703
1704 if (!tbl_data || !table->maxlen || !*lenp ||
1705 (*ppos && !write)) {
1706 *lenp = 0;
1707 return 0;
1708 }
1709
1710 i = (int *) tbl_data;
1711 vleft = table->maxlen / sizeof(*i);
1712 left = *lenp;
1713
1714 if (!conv)
1715 conv = do_proc_dointvec_conv;
1716
1717 for (; left && vleft--; i++, first=0) {
1718 if (write) {
1719 while (left) {
1720 char c;
1721 if (get_user(c, s))
1722 return -EFAULT;
1723 if (!isspace(c))
1724 break;
1725 left--;
1726 s++;
1727 }
1728 if (!left)
1729 break;
1730 neg = 0;
1731 len = left;
1732 if (len > sizeof(buf) - 1)
1733 len = sizeof(buf) - 1;
1734 if (copy_from_user(buf, s, len))
1735 return -EFAULT;
1736 buf[len] = 0;
1737 p = buf;
1738 if (*p == '-' && left > 1) {
1739 neg = 1;
1740 p++;
1741 }
1742 if (*p < '0' || *p > '9')
1743 break;
1744
1745 lval = simple_strtoul(p, &p, 0);
1746
1747 len = p-buf;
1748 if ((len < left) && *p && !isspace(*p))
1749 break;
1750 if (neg)
1751 val = -val;
1752 s += len;
1753 left -= len;
1754
1755 if (conv(&neg, &lval, i, 1, data))
1756 break;
1757 } else {
1758 p = buf;
1759 if (!first)
1760 *p++ = '\t';
1761
1762 if (conv(&neg, &lval, i, 0, data))
1763 break;
1764
1765 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1766 len = strlen(buf);
1767 if (len > left)
1768 len = left;
1769 if(copy_to_user(s, buf, len))
1770 return -EFAULT;
1771 left -= len;
1772 s += len;
1773 }
1774 }
1775
1776 if (!write && !first && left) {
1777 if(put_user('\n', s))
1778 return -EFAULT;
1779 left--, s++;
1780 }
1781 if (write) {
1782 while (left) {
1783 char c;
1784 if (get_user(c, s++))
1785 return -EFAULT;
1786 if (!isspace(c))
1787 break;
1788 left--;
1789 }
1790 }
1791 if (write && first)
1792 return -EINVAL;
1793 *lenp -= left;
1794 *ppos += *lenp;
1795 return 0;
1796 #undef TMPBUFLEN
1797 }
1798
1799 static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1800 void __user *buffer, size_t *lenp, loff_t *ppos,
1801 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1802 int write, void *data),
1803 void *data)
1804 {
1805 return __do_proc_dointvec(table->data, table, write, filp,
1806 buffer, lenp, ppos, conv, data);
1807 }
1808
1809 /**
1810 * proc_dointvec - read a vector of integers
1811 * @table: the sysctl table
1812 * @write: %TRUE if this is a write to the sysctl file
1813 * @filp: the file structure
1814 * @buffer: the user buffer
1815 * @lenp: the size of the user buffer
1816 * @ppos: file position
1817 *
1818 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1819 * values from/to the user buffer, treated as an ASCII string.
1820 *
1821 * Returns 0 on success.
1822 */
1823 int proc_dointvec(ctl_table *table, int write, struct file *filp,
1824 void __user *buffer, size_t *lenp, loff_t *ppos)
1825 {
1826 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1827 NULL,NULL);
1828 }
1829
1830 #define OP_SET 0
1831 #define OP_AND 1
1832 #define OP_OR 2
1833
1834 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1835 int *valp,
1836 int write, void *data)
1837 {
1838 int op = *(int *)data;
1839 if (write) {
1840 int val = *negp ? -*lvalp : *lvalp;
1841 switch(op) {
1842 case OP_SET: *valp = val; break;
1843 case OP_AND: *valp &= val; break;
1844 case OP_OR: *valp |= val; break;
1845 }
1846 } else {
1847 int val = *valp;
1848 if (val < 0) {
1849 *negp = -1;
1850 *lvalp = (unsigned long)-val;
1851 } else {
1852 *negp = 0;
1853 *lvalp = (unsigned long)val;
1854 }
1855 }
1856 return 0;
1857 }
1858
1859 /*
1860 * init may raise the set.
1861 */
1862
1863 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1864 void __user *buffer, size_t *lenp, loff_t *ppos)
1865 {
1866 int op;
1867
1868 if (write && !capable(CAP_SYS_MODULE)) {
1869 return -EPERM;
1870 }
1871
1872 op = is_init(current) ? OP_SET : OP_AND;
1873 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1874 do_proc_dointvec_bset_conv,&op);
1875 }
1876
1877 /*
1878 * Taint values can only be increased
1879 */
1880 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
1881 void __user *buffer, size_t *lenp, loff_t *ppos)
1882 {
1883 int op;
1884
1885 if (write && !capable(CAP_SYS_ADMIN))
1886 return -EPERM;
1887
1888 op = OP_OR;
1889 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1890 do_proc_dointvec_bset_conv,&op);
1891 }
1892
1893 struct do_proc_dointvec_minmax_conv_param {
1894 int *min;
1895 int *max;
1896 };
1897
1898 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1899 int *valp,
1900 int write, void *data)
1901 {
1902 struct do_proc_dointvec_minmax_conv_param *param = data;
1903 if (write) {
1904 int val = *negp ? -*lvalp : *lvalp;
1905 if ((param->min && *param->min > val) ||
1906 (param->max && *param->max < val))
1907 return -EINVAL;
1908 *valp = val;
1909 } else {
1910 int val = *valp;
1911 if (val < 0) {
1912 *negp = -1;
1913 *lvalp = (unsigned long)-val;
1914 } else {
1915 *negp = 0;
1916 *lvalp = (unsigned long)val;
1917 }
1918 }
1919 return 0;
1920 }
1921
1922 /**
1923 * proc_dointvec_minmax - read a vector of integers with min/max values
1924 * @table: the sysctl table
1925 * @write: %TRUE if this is a write to the sysctl file
1926 * @filp: the file structure
1927 * @buffer: the user buffer
1928 * @lenp: the size of the user buffer
1929 * @ppos: file position
1930 *
1931 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1932 * values from/to the user buffer, treated as an ASCII string.
1933 *
1934 * This routine will ensure the values are within the range specified by
1935 * table->extra1 (min) and table->extra2 (max).
1936 *
1937 * Returns 0 on success.
1938 */
1939 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1940 void __user *buffer, size_t *lenp, loff_t *ppos)
1941 {
1942 struct do_proc_dointvec_minmax_conv_param param = {
1943 .min = (int *) table->extra1,
1944 .max = (int *) table->extra2,
1945 };
1946 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1947 do_proc_dointvec_minmax_conv, &param);
1948 }
1949
1950 static int __do_proc_doulongvec_minmax(void *data, ctl_table *table, int write,
1951 struct file *filp,
1952 void __user *buffer,
1953 size_t *lenp, loff_t *ppos,
1954 unsigned long convmul,
1955 unsigned long convdiv)
1956 {
1957 #define TMPBUFLEN 21
1958 unsigned long *i, *min, *max, val;
1959 int vleft, first=1, neg;
1960 size_t len, left;
1961 char buf[TMPBUFLEN], *p;
1962 char __user *s = buffer;
1963
1964 if (!data || !table->maxlen || !*lenp ||
1965 (*ppos && !write)) {
1966 *lenp = 0;
1967 return 0;
1968 }
1969
1970 i = (unsigned long *) data;
1971 min = (unsigned long *) table->extra1;
1972 max = (unsigned long *) table->extra2;
1973 vleft = table->maxlen / sizeof(unsigned long);
1974 left = *lenp;
1975
1976 for (; left && vleft--; i++, min++, max++, first=0) {
1977 if (write) {
1978 while (left) {
1979 char c;
1980 if (get_user(c, s))
1981 return -EFAULT;
1982 if (!isspace(c))
1983 break;
1984 left--;
1985 s++;
1986 }
1987 if (!left)
1988 break;
1989 neg = 0;
1990 len = left;
1991 if (len > TMPBUFLEN-1)
1992 len = TMPBUFLEN-1;
1993 if (copy_from_user(buf, s, len))
1994 return -EFAULT;
1995 buf[len] = 0;
1996 p = buf;
1997 if (*p == '-' && left > 1) {
1998 neg = 1;
1999 p++;
2000 }
2001 if (*p < '0' || *p > '9')
2002 break;
2003 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2004 len = p-buf;
2005 if ((len < left) && *p && !isspace(*p))
2006 break;
2007 if (neg)
2008 val = -val;
2009 s += len;
2010 left -= len;
2011
2012 if(neg)
2013 continue;
2014 if ((min && val < *min) || (max && val > *max))
2015 continue;
2016 *i = val;
2017 } else {
2018 p = buf;
2019 if (!first)
2020 *p++ = '\t';
2021 sprintf(p, "%lu", convdiv * (*i) / convmul);
2022 len = strlen(buf);
2023 if (len > left)
2024 len = left;
2025 if(copy_to_user(s, buf, len))
2026 return -EFAULT;
2027 left -= len;
2028 s += len;
2029 }
2030 }
2031
2032 if (!write && !first && left) {
2033 if(put_user('\n', s))
2034 return -EFAULT;
2035 left--, s++;
2036 }
2037 if (write) {
2038 while (left) {
2039 char c;
2040 if (get_user(c, s++))
2041 return -EFAULT;
2042 if (!isspace(c))
2043 break;
2044 left--;
2045 }
2046 }
2047 if (write && first)
2048 return -EINVAL;
2049 *lenp -= left;
2050 *ppos += *lenp;
2051 return 0;
2052 #undef TMPBUFLEN
2053 }
2054
2055 static int do_proc_doulongvec_minmax(ctl_table *table, int write,
2056 struct file *filp,
2057 void __user *buffer,
2058 size_t *lenp, loff_t *ppos,
2059 unsigned long convmul,
2060 unsigned long convdiv)
2061 {
2062 return __do_proc_doulongvec_minmax(table->data, table, write,
2063 filp, buffer, lenp, ppos, convmul, convdiv);
2064 }
2065
2066 /**
2067 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2068 * @table: the sysctl table
2069 * @write: %TRUE if this is a write to the sysctl file
2070 * @filp: the file structure
2071 * @buffer: the user buffer
2072 * @lenp: the size of the user buffer
2073 * @ppos: file position
2074 *
2075 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2076 * values from/to the user buffer, treated as an ASCII string.
2077 *
2078 * This routine will ensure the values are within the range specified by
2079 * table->extra1 (min) and table->extra2 (max).
2080 *
2081 * Returns 0 on success.
2082 */
2083 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2084 void __user *buffer, size_t *lenp, loff_t *ppos)
2085 {
2086 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2087 }
2088
2089 /**
2090 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2091 * @table: the sysctl table
2092 * @write: %TRUE if this is a write to the sysctl file
2093 * @filp: the file structure
2094 * @buffer: the user buffer
2095 * @lenp: the size of the user buffer
2096 * @ppos: file position
2097 *
2098 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2099 * values from/to the user buffer, treated as an ASCII string. The values
2100 * are treated as milliseconds, and converted to jiffies when they are stored.
2101 *
2102 * This routine will ensure the values are within the range specified by
2103 * table->extra1 (min) and table->extra2 (max).
2104 *
2105 * Returns 0 on success.
2106 */
2107 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2108 struct file *filp,
2109 void __user *buffer,
2110 size_t *lenp, loff_t *ppos)
2111 {
2112 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2113 lenp, ppos, HZ, 1000l);
2114 }
2115
2116
2117 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2118 int *valp,
2119 int write, void *data)
2120 {
2121 if (write) {
2122 if (*lvalp > LONG_MAX / HZ)
2123 return 1;
2124 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2125 } else {
2126 int val = *valp;
2127 unsigned long lval;
2128 if (val < 0) {
2129 *negp = -1;
2130 lval = (unsigned long)-val;
2131 } else {
2132 *negp = 0;
2133 lval = (unsigned long)val;
2134 }
2135 *lvalp = lval / HZ;
2136 }
2137 return 0;
2138 }
2139
2140 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2141 int *valp,
2142 int write, void *data)
2143 {
2144 if (write) {
2145 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2146 return 1;
2147 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2148 } else {
2149 int val = *valp;
2150 unsigned long lval;
2151 if (val < 0) {
2152 *negp = -1;
2153 lval = (unsigned long)-val;
2154 } else {
2155 *negp = 0;
2156 lval = (unsigned long)val;
2157 }
2158 *lvalp = jiffies_to_clock_t(lval);
2159 }
2160 return 0;
2161 }
2162
2163 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2164 int *valp,
2165 int write, void *data)
2166 {
2167 if (write) {
2168 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2169 } else {
2170 int val = *valp;
2171 unsigned long lval;
2172 if (val < 0) {
2173 *negp = -1;
2174 lval = (unsigned long)-val;
2175 } else {
2176 *negp = 0;
2177 lval = (unsigned long)val;
2178 }
2179 *lvalp = jiffies_to_msecs(lval);
2180 }
2181 return 0;
2182 }
2183
2184 /**
2185 * proc_dointvec_jiffies - read a vector of integers as seconds
2186 * @table: the sysctl table
2187 * @write: %TRUE if this is a write to the sysctl file
2188 * @filp: the file structure
2189 * @buffer: the user buffer
2190 * @lenp: the size of the user buffer
2191 * @ppos: file position
2192 *
2193 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2194 * values from/to the user buffer, treated as an ASCII string.
2195 * The values read are assumed to be in seconds, and are converted into
2196 * jiffies.
2197 *
2198 * Returns 0 on success.
2199 */
2200 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2201 void __user *buffer, size_t *lenp, loff_t *ppos)
2202 {
2203 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2204 do_proc_dointvec_jiffies_conv,NULL);
2205 }
2206
2207 /**
2208 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2209 * @table: the sysctl table
2210 * @write: %TRUE if this is a write to the sysctl file
2211 * @filp: the file structure
2212 * @buffer: the user buffer
2213 * @lenp: the size of the user buffer
2214 * @ppos: pointer to the file position
2215 *
2216 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2217 * values from/to the user buffer, treated as an ASCII string.
2218 * The values read are assumed to be in 1/USER_HZ seconds, and
2219 * are converted into jiffies.
2220 *
2221 * Returns 0 on success.
2222 */
2223 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2224 void __user *buffer, size_t *lenp, loff_t *ppos)
2225 {
2226 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2227 do_proc_dointvec_userhz_jiffies_conv,NULL);
2228 }
2229
2230 /**
2231 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2232 * @table: the sysctl table
2233 * @write: %TRUE if this is a write to the sysctl file
2234 * @filp: the file structure
2235 * @buffer: the user buffer
2236 * @lenp: the size of the user buffer
2237 * @ppos: file position
2238 * @ppos: the current position in the file
2239 *
2240 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2241 * values from/to the user buffer, treated as an ASCII string.
2242 * The values read are assumed to be in 1/1000 seconds, and
2243 * are converted into jiffies.
2244 *
2245 * Returns 0 on success.
2246 */
2247 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2248 void __user *buffer, size_t *lenp, loff_t *ppos)
2249 {
2250 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2251 do_proc_dointvec_ms_jiffies_conv, NULL);
2252 }
2253
2254 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
2255 void __user *buffer, size_t *lenp, loff_t *ppos)
2256 {
2257 struct pid *new_pid;
2258 pid_t tmp;
2259 int r;
2260
2261 tmp = pid_nr(cad_pid);
2262
2263 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2264 lenp, ppos, NULL, NULL);
2265 if (r || !write)
2266 return r;
2267
2268 new_pid = find_get_pid(tmp);
2269 if (!new_pid)
2270 return -ESRCH;
2271
2272 put_pid(xchg(&cad_pid, new_pid));
2273 return 0;
2274 }
2275
2276 #else /* CONFIG_PROC_FS */
2277
2278 int proc_dostring(ctl_table *table, int write, struct file *filp,
2279 void __user *buffer, size_t *lenp, loff_t *ppos)
2280 {
2281 return -ENOSYS;
2282 }
2283
2284 int proc_dointvec(ctl_table *table, int write, struct file *filp,
2285 void __user *buffer, size_t *lenp, loff_t *ppos)
2286 {
2287 return -ENOSYS;
2288 }
2289
2290 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2291 void __user *buffer, size_t *lenp, loff_t *ppos)
2292 {
2293 return -ENOSYS;
2294 }
2295
2296 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2297 void __user *buffer, size_t *lenp, loff_t *ppos)
2298 {
2299 return -ENOSYS;
2300 }
2301
2302 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2303 void __user *buffer, size_t *lenp, loff_t *ppos)
2304 {
2305 return -ENOSYS;
2306 }
2307
2308 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2309 void __user *buffer, size_t *lenp, loff_t *ppos)
2310 {
2311 return -ENOSYS;
2312 }
2313
2314 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2315 void __user *buffer, size_t *lenp, loff_t *ppos)
2316 {
2317 return -ENOSYS;
2318 }
2319
2320 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2321 void __user *buffer, size_t *lenp, loff_t *ppos)
2322 {
2323 return -ENOSYS;
2324 }
2325
2326 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2327 struct file *filp,
2328 void __user *buffer,
2329 size_t *lenp, loff_t *ppos)
2330 {
2331 return -ENOSYS;
2332 }
2333
2334
2335 #endif /* CONFIG_PROC_FS */
2336
2337
2338 #ifdef CONFIG_SYSCTL_SYSCALL
2339 /*
2340 * General sysctl support routines
2341 */
2342
2343 /* The generic string strategy routine: */
2344 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2345 void __user *oldval, size_t __user *oldlenp,
2346 void __user *newval, size_t newlen)
2347 {
2348 if (!table->data || !table->maxlen)
2349 return -ENOTDIR;
2350
2351 if (oldval && oldlenp) {
2352 size_t bufsize;
2353 if (get_user(bufsize, oldlenp))
2354 return -EFAULT;
2355 if (bufsize) {
2356 size_t len = strlen(table->data), copied;
2357
2358 /* This shouldn't trigger for a well-formed sysctl */
2359 if (len > table->maxlen)
2360 len = table->maxlen;
2361
2362 /* Copy up to a max of bufsize-1 bytes of the string */
2363 copied = (len >= bufsize) ? bufsize - 1 : len;
2364
2365 if (copy_to_user(oldval, table->data, copied) ||
2366 put_user(0, (char __user *)(oldval + copied)))
2367 return -EFAULT;
2368 if (put_user(len, oldlenp))
2369 return -EFAULT;
2370 }
2371 }
2372 if (newval && newlen) {
2373 size_t len = newlen;
2374 if (len > table->maxlen)
2375 len = table->maxlen;
2376 if(copy_from_user(table->data, newval, len))
2377 return -EFAULT;
2378 if (len == table->maxlen)
2379 len--;
2380 ((char *) table->data)[len] = 0;
2381 }
2382 return 1;
2383 }
2384
2385 /*
2386 * This function makes sure that all of the integers in the vector
2387 * are between the minimum and maximum values given in the arrays
2388 * table->extra1 and table->extra2, respectively.
2389 */
2390 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2391 void __user *oldval, size_t __user *oldlenp,
2392 void __user *newval, size_t newlen)
2393 {
2394
2395 if (newval && newlen) {
2396 int __user *vec = (int __user *) newval;
2397 int *min = (int *) table->extra1;
2398 int *max = (int *) table->extra2;
2399 size_t length;
2400 int i;
2401
2402 if (newlen % sizeof(int) != 0)
2403 return -EINVAL;
2404
2405 if (!table->extra1 && !table->extra2)
2406 return 0;
2407
2408 if (newlen > table->maxlen)
2409 newlen = table->maxlen;
2410 length = newlen / sizeof(int);
2411
2412 for (i = 0; i < length; i++) {
2413 int value;
2414 if (get_user(value, vec + i))
2415 return -EFAULT;
2416 if (min && value < min[i])
2417 return -EINVAL;
2418 if (max && value > max[i])
2419 return -EINVAL;
2420 }
2421 }
2422 return 0;
2423 }
2424
2425 /* Strategy function to convert jiffies to seconds */
2426 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2427 void __user *oldval, size_t __user *oldlenp,
2428 void __user *newval, size_t newlen)
2429 {
2430 if (oldval && oldlenp) {
2431 size_t olen;
2432
2433 if (get_user(olen, oldlenp))
2434 return -EFAULT;
2435 if (olen) {
2436 int val;
2437
2438 if (olen < sizeof(int))
2439 return -EINVAL;
2440
2441 val = *(int *)(table->data) / HZ;
2442 if (put_user(val, (int __user *)oldval))
2443 return -EFAULT;
2444 if (put_user(sizeof(int), oldlenp))
2445 return -EFAULT;
2446 }
2447 }
2448 if (newval && newlen) {
2449 int new;
2450 if (newlen != sizeof(int))
2451 return -EINVAL;
2452 if (get_user(new, (int __user *)newval))
2453 return -EFAULT;
2454 *(int *)(table->data) = new*HZ;
2455 }
2456 return 1;
2457 }
2458
2459 /* Strategy function to convert jiffies to seconds */
2460 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2461 void __user *oldval, size_t __user *oldlenp,
2462 void __user *newval, size_t newlen)
2463 {
2464 if (oldval && oldlenp) {
2465 size_t olen;
2466
2467 if (get_user(olen, oldlenp))
2468 return -EFAULT;
2469 if (olen) {
2470 int val;
2471
2472 if (olen < sizeof(int))
2473 return -EINVAL;
2474
2475 val = jiffies_to_msecs(*(int *)(table->data));
2476 if (put_user(val, (int __user *)oldval))
2477 return -EFAULT;
2478 if (put_user(sizeof(int), oldlenp))
2479 return -EFAULT;
2480 }
2481 }
2482 if (newval && newlen) {
2483 int new;
2484 if (newlen != sizeof(int))
2485 return -EINVAL;
2486 if (get_user(new, (int __user *)newval))
2487 return -EFAULT;
2488 *(int *)(table->data) = msecs_to_jiffies(new);
2489 }
2490 return 1;
2491 }
2492
2493
2494
2495 #else /* CONFIG_SYSCTL_SYSCALL */
2496
2497
2498 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2499 {
2500 static int msg_count;
2501 struct __sysctl_args tmp;
2502 int name[CTL_MAXNAME];
2503 int i;
2504
2505 /* Read in the sysctl name for better debug message logging */
2506 if (copy_from_user(&tmp, args, sizeof(tmp)))
2507 return -EFAULT;
2508 if (tmp.nlen <= 0 || tmp.nlen >= CTL_MAXNAME)
2509 return -ENOTDIR;
2510 for (i = 0; i < tmp.nlen; i++)
2511 if (get_user(name[i], tmp.name + i))
2512 return -EFAULT;
2513
2514 /* Ignore accesses to kernel.version */
2515 if ((tmp.nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2516 goto out;
2517
2518 if (msg_count < 5) {
2519 msg_count++;
2520 printk(KERN_INFO
2521 "warning: process `%s' used the removed sysctl "
2522 "system call with ", current->comm);
2523 for (i = 0; i < tmp.nlen; i++)
2524 printk("%d.", name[i]);
2525 printk("\n");
2526 }
2527 out:
2528 return -ENOSYS;
2529 }
2530
2531 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2532 void __user *oldval, size_t __user *oldlenp,
2533 void __user *newval, size_t newlen)
2534 {
2535 return -ENOSYS;
2536 }
2537
2538 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2539 void __user *oldval, size_t __user *oldlenp,
2540 void __user *newval, size_t newlen)
2541 {
2542 return -ENOSYS;
2543 }
2544
2545 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2546 void __user *oldval, size_t __user *oldlenp,
2547 void __user *newval, size_t newlen)
2548 {
2549 return -ENOSYS;
2550 }
2551
2552 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2553 void __user *oldval, size_t __user *oldlenp,
2554 void __user *newval, size_t newlen)
2555 {
2556 return -ENOSYS;
2557 }
2558
2559 #endif /* CONFIG_SYSCTL_SYSCALL */
2560
2561 /*
2562 * No sense putting this after each symbol definition, twice,
2563 * exception granted :-)
2564 */
2565 EXPORT_SYMBOL(proc_dointvec);
2566 EXPORT_SYMBOL(proc_dointvec_jiffies);
2567 EXPORT_SYMBOL(proc_dointvec_minmax);
2568 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2569 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2570 EXPORT_SYMBOL(proc_dostring);
2571 EXPORT_SYMBOL(proc_doulongvec_minmax);
2572 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2573 EXPORT_SYMBOL(register_sysctl_table);
2574 EXPORT_SYMBOL(sysctl_intvec);
2575 EXPORT_SYMBOL(sysctl_jiffies);
2576 EXPORT_SYMBOL(sysctl_ms_jiffies);
2577 EXPORT_SYMBOL(sysctl_string);
2578 EXPORT_SYMBOL(unregister_sysctl_table);