xref: /openbmc/linux/kernel/sysctl.c (revision a09d2831)
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/security.h>
28 #include <linux/ctype.h>
29 #include <linux/kmemcheck.h>
30 #include <linux/fs.h>
31 #include <linux/init.h>
32 #include <linux/kernel.h>
33 #include <linux/kobject.h>
34 #include <linux/net.h>
35 #include <linux/sysrq.h>
36 #include <linux/highuid.h>
37 #include <linux/writeback.h>
38 #include <linux/ratelimit.h>
39 #include <linux/hugetlb.h>
40 #include <linux/initrd.h>
41 #include <linux/key.h>
42 #include <linux/times.h>
43 #include <linux/limits.h>
44 #include <linux/dcache.h>
45 #include <linux/syscalls.h>
46 #include <linux/vmstat.h>
47 #include <linux/nfs_fs.h>
48 #include <linux/acpi.h>
49 #include <linux/reboot.h>
50 #include <linux/ftrace.h>
51 #include <linux/slow-work.h>
52 #include <linux/perf_event.h>
53 
54 #include <asm/uaccess.h>
55 #include <asm/processor.h>
56 
57 #ifdef CONFIG_X86
58 #include <asm/nmi.h>
59 #include <asm/stacktrace.h>
60 #include <asm/io.h>
61 #endif
62 
63 
64 #if defined(CONFIG_SYSCTL)
65 
66 /* External variables not in a header file. */
67 extern int C_A_D;
68 extern int print_fatal_signals;
69 extern int sysctl_overcommit_memory;
70 extern int sysctl_overcommit_ratio;
71 extern int sysctl_panic_on_oom;
72 extern int sysctl_oom_kill_allocating_task;
73 extern int sysctl_oom_dump_tasks;
74 extern int max_threads;
75 extern int core_uses_pid;
76 extern int suid_dumpable;
77 extern char core_pattern[];
78 extern unsigned int core_pipe_limit;
79 extern int pid_max;
80 extern int min_free_kbytes;
81 extern int pid_max_min, pid_max_max;
82 extern int sysctl_drop_caches;
83 extern int percpu_pagelist_fraction;
84 extern int compat_log;
85 extern int latencytop_enabled;
86 extern int sysctl_nr_open_min, sysctl_nr_open_max;
87 #ifndef CONFIG_MMU
88 extern int sysctl_nr_trim_pages;
89 #endif
90 #ifdef CONFIG_RCU_TORTURE_TEST
91 extern int rcutorture_runnable;
92 #endif /* #ifdef CONFIG_RCU_TORTURE_TEST */
93 #ifdef CONFIG_BLOCK
94 extern int blk_iopoll_enabled;
95 #endif
96 
97 /* Constants used for minimum and  maximum */
98 #ifdef CONFIG_DETECT_SOFTLOCKUP
99 static int sixty = 60;
100 static int neg_one = -1;
101 #endif
102 
103 static int zero;
104 static int __maybe_unused one = 1;
105 static int __maybe_unused two = 2;
106 static unsigned long one_ul = 1;
107 static int one_hundred = 100;
108 #ifdef CONFIG_PRINTK
109 static int ten_thousand = 10000;
110 #endif
111 
112 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
113 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
114 
115 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
116 static int maxolduid = 65535;
117 static int minolduid;
118 static int min_percpu_pagelist_fract = 8;
119 
120 static int ngroups_max = NGROUPS_MAX;
121 
122 #ifdef CONFIG_MODULES
123 extern char modprobe_path[];
124 extern int modules_disabled;
125 #endif
126 #ifdef CONFIG_CHR_DEV_SG
127 extern int sg_big_buff;
128 #endif
129 
130 #ifdef CONFIG_SPARC
131 #include <asm/system.h>
132 #endif
133 
134 #ifdef CONFIG_SPARC64
135 extern int sysctl_tsb_ratio;
136 #endif
137 
138 #ifdef __hppa__
139 extern int pwrsw_enabled;
140 extern int unaligned_enabled;
141 #endif
142 
143 #ifdef CONFIG_S390
144 #ifdef CONFIG_MATHEMU
145 extern int sysctl_ieee_emulation_warnings;
146 #endif
147 extern int sysctl_userprocess_debug;
148 extern int spin_retry;
149 #endif
150 
151 #ifdef CONFIG_BSD_PROCESS_ACCT
152 extern int acct_parm[];
153 #endif
154 
155 #ifdef CONFIG_IA64
156 extern int no_unaligned_warning;
157 extern int unaligned_dump_stack;
158 #endif
159 
160 extern struct ratelimit_state printk_ratelimit_state;
161 
162 #ifdef CONFIG_RT_MUTEXES
163 extern int max_lock_depth;
164 #endif
165 
166 #ifdef CONFIG_PROC_SYSCTL
167 static int proc_do_cad_pid(struct ctl_table *table, int write,
168 		  void __user *buffer, size_t *lenp, loff_t *ppos);
169 static int proc_taint(struct ctl_table *table, int write,
170 			       void __user *buffer, size_t *lenp, loff_t *ppos);
171 #endif
172 
173 static struct ctl_table root_table[];
174 static struct ctl_table_root sysctl_table_root;
175 static struct ctl_table_header root_table_header = {
176 	.count = 1,
177 	.ctl_table = root_table,
178 	.ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
179 	.root = &sysctl_table_root,
180 	.set = &sysctl_table_root.default_set,
181 };
182 static struct ctl_table_root sysctl_table_root = {
183 	.root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
184 	.default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
185 };
186 
187 static struct ctl_table kern_table[];
188 static struct ctl_table vm_table[];
189 static struct ctl_table fs_table[];
190 static struct ctl_table debug_table[];
191 static struct ctl_table dev_table[];
192 extern struct ctl_table random_table[];
193 #ifdef CONFIG_INOTIFY_USER
194 extern struct ctl_table inotify_table[];
195 #endif
196 #ifdef CONFIG_EPOLL
197 extern struct ctl_table epoll_table[];
198 #endif
199 
200 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
201 int sysctl_legacy_va_layout;
202 #endif
203 
204 extern int prove_locking;
205 extern int lock_stat;
206 
207 /* The default sysctl tables: */
208 
209 static struct ctl_table root_table[] = {
210 	{
211 		.procname	= "kernel",
212 		.mode		= 0555,
213 		.child		= kern_table,
214 	},
215 	{
216 		.procname	= "vm",
217 		.mode		= 0555,
218 		.child		= vm_table,
219 	},
220 	{
221 		.procname	= "fs",
222 		.mode		= 0555,
223 		.child		= fs_table,
224 	},
225 	{
226 		.procname	= "debug",
227 		.mode		= 0555,
228 		.child		= debug_table,
229 	},
230 	{
231 		.procname	= "dev",
232 		.mode		= 0555,
233 		.child		= dev_table,
234 	},
235 /*
236  * NOTE: do not add new entries to this table unless you have read
237  * Documentation/sysctl/ctl_unnumbered.txt
238  */
239 	{ }
240 };
241 
242 #ifdef CONFIG_SCHED_DEBUG
243 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
244 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
245 static int min_wakeup_granularity_ns;			/* 0 usecs */
246 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
247 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
248 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
249 static int min_sched_shares_ratelimit = 100000; /* 100 usec */
250 static int max_sched_shares_ratelimit = NSEC_PER_SEC; /* 1 second */
251 #endif
252 
253 static struct ctl_table kern_table[] = {
254 	{
255 		.procname	= "sched_child_runs_first",
256 		.data		= &sysctl_sched_child_runs_first,
257 		.maxlen		= sizeof(unsigned int),
258 		.mode		= 0644,
259 		.proc_handler	= proc_dointvec,
260 	},
261 #ifdef CONFIG_SCHED_DEBUG
262 	{
263 		.procname	= "sched_min_granularity_ns",
264 		.data		= &sysctl_sched_min_granularity,
265 		.maxlen		= sizeof(unsigned int),
266 		.mode		= 0644,
267 		.proc_handler	= sched_proc_update_handler,
268 		.extra1		= &min_sched_granularity_ns,
269 		.extra2		= &max_sched_granularity_ns,
270 	},
271 	{
272 		.procname	= "sched_latency_ns",
273 		.data		= &sysctl_sched_latency,
274 		.maxlen		= sizeof(unsigned int),
275 		.mode		= 0644,
276 		.proc_handler	= sched_proc_update_handler,
277 		.extra1		= &min_sched_granularity_ns,
278 		.extra2		= &max_sched_granularity_ns,
279 	},
280 	{
281 		.procname	= "sched_wakeup_granularity_ns",
282 		.data		= &sysctl_sched_wakeup_granularity,
283 		.maxlen		= sizeof(unsigned int),
284 		.mode		= 0644,
285 		.proc_handler	= sched_proc_update_handler,
286 		.extra1		= &min_wakeup_granularity_ns,
287 		.extra2		= &max_wakeup_granularity_ns,
288 	},
289 	{
290 		.procname	= "sched_shares_ratelimit",
291 		.data		= &sysctl_sched_shares_ratelimit,
292 		.maxlen		= sizeof(unsigned int),
293 		.mode		= 0644,
294 		.proc_handler	= sched_proc_update_handler,
295 		.extra1		= &min_sched_shares_ratelimit,
296 		.extra2		= &max_sched_shares_ratelimit,
297 	},
298 	{
299 		.procname	= "sched_tunable_scaling",
300 		.data		= &sysctl_sched_tunable_scaling,
301 		.maxlen		= sizeof(enum sched_tunable_scaling),
302 		.mode		= 0644,
303 		.proc_handler	= sched_proc_update_handler,
304 		.extra1		= &min_sched_tunable_scaling,
305 		.extra2		= &max_sched_tunable_scaling,
306 	},
307 	{
308 		.procname	= "sched_shares_thresh",
309 		.data		= &sysctl_sched_shares_thresh,
310 		.maxlen		= sizeof(unsigned int),
311 		.mode		= 0644,
312 		.proc_handler	= proc_dointvec_minmax,
313 		.extra1		= &zero,
314 	},
315 	{
316 		.procname	= "sched_migration_cost",
317 		.data		= &sysctl_sched_migration_cost,
318 		.maxlen		= sizeof(unsigned int),
319 		.mode		= 0644,
320 		.proc_handler	= proc_dointvec,
321 	},
322 	{
323 		.procname	= "sched_nr_migrate",
324 		.data		= &sysctl_sched_nr_migrate,
325 		.maxlen		= sizeof(unsigned int),
326 		.mode		= 0644,
327 		.proc_handler	= proc_dointvec,
328 	},
329 	{
330 		.procname	= "sched_time_avg",
331 		.data		= &sysctl_sched_time_avg,
332 		.maxlen		= sizeof(unsigned int),
333 		.mode		= 0644,
334 		.proc_handler	= proc_dointvec,
335 	},
336 	{
337 		.procname	= "timer_migration",
338 		.data		= &sysctl_timer_migration,
339 		.maxlen		= sizeof(unsigned int),
340 		.mode		= 0644,
341 		.proc_handler	= proc_dointvec_minmax,
342 		.extra1		= &zero,
343 		.extra2		= &one,
344 	},
345 #endif
346 	{
347 		.procname	= "sched_rt_period_us",
348 		.data		= &sysctl_sched_rt_period,
349 		.maxlen		= sizeof(unsigned int),
350 		.mode		= 0644,
351 		.proc_handler	= sched_rt_handler,
352 	},
353 	{
354 		.procname	= "sched_rt_runtime_us",
355 		.data		= &sysctl_sched_rt_runtime,
356 		.maxlen		= sizeof(int),
357 		.mode		= 0644,
358 		.proc_handler	= sched_rt_handler,
359 	},
360 	{
361 		.procname	= "sched_compat_yield",
362 		.data		= &sysctl_sched_compat_yield,
363 		.maxlen		= sizeof(unsigned int),
364 		.mode		= 0644,
365 		.proc_handler	= proc_dointvec,
366 	},
367 #ifdef CONFIG_PROVE_LOCKING
368 	{
369 		.procname	= "prove_locking",
370 		.data		= &prove_locking,
371 		.maxlen		= sizeof(int),
372 		.mode		= 0644,
373 		.proc_handler	= proc_dointvec,
374 	},
375 #endif
376 #ifdef CONFIG_LOCK_STAT
377 	{
378 		.procname	= "lock_stat",
379 		.data		= &lock_stat,
380 		.maxlen		= sizeof(int),
381 		.mode		= 0644,
382 		.proc_handler	= proc_dointvec,
383 	},
384 #endif
385 	{
386 		.procname	= "panic",
387 		.data		= &panic_timeout,
388 		.maxlen		= sizeof(int),
389 		.mode		= 0644,
390 		.proc_handler	= proc_dointvec,
391 	},
392 	{
393 		.procname	= "core_uses_pid",
394 		.data		= &core_uses_pid,
395 		.maxlen		= sizeof(int),
396 		.mode		= 0644,
397 		.proc_handler	= proc_dointvec,
398 	},
399 	{
400 		.procname	= "core_pattern",
401 		.data		= core_pattern,
402 		.maxlen		= CORENAME_MAX_SIZE,
403 		.mode		= 0644,
404 		.proc_handler	= proc_dostring,
405 	},
406 	{
407 		.procname	= "core_pipe_limit",
408 		.data		= &core_pipe_limit,
409 		.maxlen		= sizeof(unsigned int),
410 		.mode		= 0644,
411 		.proc_handler	= proc_dointvec,
412 	},
413 #ifdef CONFIG_PROC_SYSCTL
414 	{
415 		.procname	= "tainted",
416 		.maxlen 	= sizeof(long),
417 		.mode		= 0644,
418 		.proc_handler	= proc_taint,
419 	},
420 #endif
421 #ifdef CONFIG_LATENCYTOP
422 	{
423 		.procname	= "latencytop",
424 		.data		= &latencytop_enabled,
425 		.maxlen		= sizeof(int),
426 		.mode		= 0644,
427 		.proc_handler	= proc_dointvec,
428 	},
429 #endif
430 #ifdef CONFIG_BLK_DEV_INITRD
431 	{
432 		.procname	= "real-root-dev",
433 		.data		= &real_root_dev,
434 		.maxlen		= sizeof(int),
435 		.mode		= 0644,
436 		.proc_handler	= proc_dointvec,
437 	},
438 #endif
439 	{
440 		.procname	= "print-fatal-signals",
441 		.data		= &print_fatal_signals,
442 		.maxlen		= sizeof(int),
443 		.mode		= 0644,
444 		.proc_handler	= proc_dointvec,
445 	},
446 #ifdef CONFIG_SPARC
447 	{
448 		.procname	= "reboot-cmd",
449 		.data		= reboot_command,
450 		.maxlen		= 256,
451 		.mode		= 0644,
452 		.proc_handler	= proc_dostring,
453 	},
454 	{
455 		.procname	= "stop-a",
456 		.data		= &stop_a_enabled,
457 		.maxlen		= sizeof (int),
458 		.mode		= 0644,
459 		.proc_handler	= proc_dointvec,
460 	},
461 	{
462 		.procname	= "scons-poweroff",
463 		.data		= &scons_pwroff,
464 		.maxlen		= sizeof (int),
465 		.mode		= 0644,
466 		.proc_handler	= proc_dointvec,
467 	},
468 #endif
469 #ifdef CONFIG_SPARC64
470 	{
471 		.procname	= "tsb-ratio",
472 		.data		= &sysctl_tsb_ratio,
473 		.maxlen		= sizeof (int),
474 		.mode		= 0644,
475 		.proc_handler	= proc_dointvec,
476 	},
477 #endif
478 #ifdef __hppa__
479 	{
480 		.procname	= "soft-power",
481 		.data		= &pwrsw_enabled,
482 		.maxlen		= sizeof (int),
483 	 	.mode		= 0644,
484 		.proc_handler	= proc_dointvec,
485 	},
486 	{
487 		.procname	= "unaligned-trap",
488 		.data		= &unaligned_enabled,
489 		.maxlen		= sizeof (int),
490 		.mode		= 0644,
491 		.proc_handler	= proc_dointvec,
492 	},
493 #endif
494 	{
495 		.procname	= "ctrl-alt-del",
496 		.data		= &C_A_D,
497 		.maxlen		= sizeof(int),
498 		.mode		= 0644,
499 		.proc_handler	= proc_dointvec,
500 	},
501 #ifdef CONFIG_FUNCTION_TRACER
502 	{
503 		.procname	= "ftrace_enabled",
504 		.data		= &ftrace_enabled,
505 		.maxlen		= sizeof(int),
506 		.mode		= 0644,
507 		.proc_handler	= ftrace_enable_sysctl,
508 	},
509 #endif
510 #ifdef CONFIG_STACK_TRACER
511 	{
512 		.procname	= "stack_tracer_enabled",
513 		.data		= &stack_tracer_enabled,
514 		.maxlen		= sizeof(int),
515 		.mode		= 0644,
516 		.proc_handler	= stack_trace_sysctl,
517 	},
518 #endif
519 #ifdef CONFIG_TRACING
520 	{
521 		.procname	= "ftrace_dump_on_oops",
522 		.data		= &ftrace_dump_on_oops,
523 		.maxlen		= sizeof(int),
524 		.mode		= 0644,
525 		.proc_handler	= proc_dointvec,
526 	},
527 #endif
528 #ifdef CONFIG_MODULES
529 	{
530 		.procname	= "modprobe",
531 		.data		= &modprobe_path,
532 		.maxlen		= KMOD_PATH_LEN,
533 		.mode		= 0644,
534 		.proc_handler	= proc_dostring,
535 	},
536 	{
537 		.procname	= "modules_disabled",
538 		.data		= &modules_disabled,
539 		.maxlen		= sizeof(int),
540 		.mode		= 0644,
541 		/* only handle a transition from default "0" to "1" */
542 		.proc_handler	= proc_dointvec_minmax,
543 		.extra1		= &one,
544 		.extra2		= &one,
545 	},
546 #endif
547 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
548 	{
549 		.procname	= "hotplug",
550 		.data		= &uevent_helper,
551 		.maxlen		= UEVENT_HELPER_PATH_LEN,
552 		.mode		= 0644,
553 		.proc_handler	= proc_dostring,
554 	},
555 #endif
556 #ifdef CONFIG_CHR_DEV_SG
557 	{
558 		.procname	= "sg-big-buff",
559 		.data		= &sg_big_buff,
560 		.maxlen		= sizeof (int),
561 		.mode		= 0444,
562 		.proc_handler	= proc_dointvec,
563 	},
564 #endif
565 #ifdef CONFIG_BSD_PROCESS_ACCT
566 	{
567 		.procname	= "acct",
568 		.data		= &acct_parm,
569 		.maxlen		= 3*sizeof(int),
570 		.mode		= 0644,
571 		.proc_handler	= proc_dointvec,
572 	},
573 #endif
574 #ifdef CONFIG_MAGIC_SYSRQ
575 	{
576 		.procname	= "sysrq",
577 		.data		= &__sysrq_enabled,
578 		.maxlen		= sizeof (int),
579 		.mode		= 0644,
580 		.proc_handler	= proc_dointvec,
581 	},
582 #endif
583 #ifdef CONFIG_PROC_SYSCTL
584 	{
585 		.procname	= "cad_pid",
586 		.data		= NULL,
587 		.maxlen		= sizeof (int),
588 		.mode		= 0600,
589 		.proc_handler	= proc_do_cad_pid,
590 	},
591 #endif
592 	{
593 		.procname	= "threads-max",
594 		.data		= &max_threads,
595 		.maxlen		= sizeof(int),
596 		.mode		= 0644,
597 		.proc_handler	= proc_dointvec,
598 	},
599 	{
600 		.procname	= "random",
601 		.mode		= 0555,
602 		.child		= random_table,
603 	},
604 	{
605 		.procname	= "overflowuid",
606 		.data		= &overflowuid,
607 		.maxlen		= sizeof(int),
608 		.mode		= 0644,
609 		.proc_handler	= proc_dointvec_minmax,
610 		.extra1		= &minolduid,
611 		.extra2		= &maxolduid,
612 	},
613 	{
614 		.procname	= "overflowgid",
615 		.data		= &overflowgid,
616 		.maxlen		= sizeof(int),
617 		.mode		= 0644,
618 		.proc_handler	= proc_dointvec_minmax,
619 		.extra1		= &minolduid,
620 		.extra2		= &maxolduid,
621 	},
622 #ifdef CONFIG_S390
623 #ifdef CONFIG_MATHEMU
624 	{
625 		.procname	= "ieee_emulation_warnings",
626 		.data		= &sysctl_ieee_emulation_warnings,
627 		.maxlen		= sizeof(int),
628 		.mode		= 0644,
629 		.proc_handler	= proc_dointvec,
630 	},
631 #endif
632 	{
633 		.procname	= "userprocess_debug",
634 		.data		= &sysctl_userprocess_debug,
635 		.maxlen		= sizeof(int),
636 		.mode		= 0644,
637 		.proc_handler	= proc_dointvec,
638 	},
639 #endif
640 	{
641 		.procname	= "pid_max",
642 		.data		= &pid_max,
643 		.maxlen		= sizeof (int),
644 		.mode		= 0644,
645 		.proc_handler	= proc_dointvec_minmax,
646 		.extra1		= &pid_max_min,
647 		.extra2		= &pid_max_max,
648 	},
649 	{
650 		.procname	= "panic_on_oops",
651 		.data		= &panic_on_oops,
652 		.maxlen		= sizeof(int),
653 		.mode		= 0644,
654 		.proc_handler	= proc_dointvec,
655 	},
656 #if defined CONFIG_PRINTK
657 	{
658 		.procname	= "printk",
659 		.data		= &console_loglevel,
660 		.maxlen		= 4*sizeof(int),
661 		.mode		= 0644,
662 		.proc_handler	= proc_dointvec,
663 	},
664 	{
665 		.procname	= "printk_ratelimit",
666 		.data		= &printk_ratelimit_state.interval,
667 		.maxlen		= sizeof(int),
668 		.mode		= 0644,
669 		.proc_handler	= proc_dointvec_jiffies,
670 	},
671 	{
672 		.procname	= "printk_ratelimit_burst",
673 		.data		= &printk_ratelimit_state.burst,
674 		.maxlen		= sizeof(int),
675 		.mode		= 0644,
676 		.proc_handler	= proc_dointvec,
677 	},
678 	{
679 		.procname	= "printk_delay",
680 		.data		= &printk_delay_msec,
681 		.maxlen		= sizeof(int),
682 		.mode		= 0644,
683 		.proc_handler	= proc_dointvec_minmax,
684 		.extra1		= &zero,
685 		.extra2		= &ten_thousand,
686 	},
687 #endif
688 	{
689 		.procname	= "ngroups_max",
690 		.data		= &ngroups_max,
691 		.maxlen		= sizeof (int),
692 		.mode		= 0444,
693 		.proc_handler	= proc_dointvec,
694 	},
695 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
696 	{
697 		.procname       = "unknown_nmi_panic",
698 		.data           = &unknown_nmi_panic,
699 		.maxlen         = sizeof (int),
700 		.mode           = 0644,
701 		.proc_handler   = proc_dointvec,
702 	},
703 	{
704 		.procname       = "nmi_watchdog",
705 		.data           = &nmi_watchdog_enabled,
706 		.maxlen         = sizeof (int),
707 		.mode           = 0644,
708 		.proc_handler   = proc_nmi_enabled,
709 	},
710 #endif
711 #if defined(CONFIG_X86)
712 	{
713 		.procname	= "panic_on_unrecovered_nmi",
714 		.data		= &panic_on_unrecovered_nmi,
715 		.maxlen		= sizeof(int),
716 		.mode		= 0644,
717 		.proc_handler	= proc_dointvec,
718 	},
719 	{
720 		.procname	= "panic_on_io_nmi",
721 		.data		= &panic_on_io_nmi,
722 		.maxlen		= sizeof(int),
723 		.mode		= 0644,
724 		.proc_handler	= proc_dointvec,
725 	},
726 	{
727 		.procname	= "bootloader_type",
728 		.data		= &bootloader_type,
729 		.maxlen		= sizeof (int),
730 		.mode		= 0444,
731 		.proc_handler	= proc_dointvec,
732 	},
733 	{
734 		.procname	= "bootloader_version",
735 		.data		= &bootloader_version,
736 		.maxlen		= sizeof (int),
737 		.mode		= 0444,
738 		.proc_handler	= proc_dointvec,
739 	},
740 	{
741 		.procname	= "kstack_depth_to_print",
742 		.data		= &kstack_depth_to_print,
743 		.maxlen		= sizeof(int),
744 		.mode		= 0644,
745 		.proc_handler	= proc_dointvec,
746 	},
747 	{
748 		.procname	= "io_delay_type",
749 		.data		= &io_delay_type,
750 		.maxlen		= sizeof(int),
751 		.mode		= 0644,
752 		.proc_handler	= proc_dointvec,
753 	},
754 #endif
755 #if defined(CONFIG_MMU)
756 	{
757 		.procname	= "randomize_va_space",
758 		.data		= &randomize_va_space,
759 		.maxlen		= sizeof(int),
760 		.mode		= 0644,
761 		.proc_handler	= proc_dointvec,
762 	},
763 #endif
764 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
765 	{
766 		.procname	= "spin_retry",
767 		.data		= &spin_retry,
768 		.maxlen		= sizeof (int),
769 		.mode		= 0644,
770 		.proc_handler	= proc_dointvec,
771 	},
772 #endif
773 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
774 	{
775 		.procname	= "acpi_video_flags",
776 		.data		= &acpi_realmode_flags,
777 		.maxlen		= sizeof (unsigned long),
778 		.mode		= 0644,
779 		.proc_handler	= proc_doulongvec_minmax,
780 	},
781 #endif
782 #ifdef CONFIG_IA64
783 	{
784 		.procname	= "ignore-unaligned-usertrap",
785 		.data		= &no_unaligned_warning,
786 		.maxlen		= sizeof (int),
787 	 	.mode		= 0644,
788 		.proc_handler	= proc_dointvec,
789 	},
790 	{
791 		.procname	= "unaligned-dump-stack",
792 		.data		= &unaligned_dump_stack,
793 		.maxlen		= sizeof (int),
794 		.mode		= 0644,
795 		.proc_handler	= proc_dointvec,
796 	},
797 #endif
798 #ifdef CONFIG_DETECT_SOFTLOCKUP
799 	{
800 		.procname	= "softlockup_panic",
801 		.data		= &softlockup_panic,
802 		.maxlen		= sizeof(int),
803 		.mode		= 0644,
804 		.proc_handler	= proc_dointvec_minmax,
805 		.extra1		= &zero,
806 		.extra2		= &one,
807 	},
808 	{
809 		.procname	= "softlockup_thresh",
810 		.data		= &softlockup_thresh,
811 		.maxlen		= sizeof(int),
812 		.mode		= 0644,
813 		.proc_handler	= proc_dosoftlockup_thresh,
814 		.extra1		= &neg_one,
815 		.extra2		= &sixty,
816 	},
817 #endif
818 #ifdef CONFIG_DETECT_HUNG_TASK
819 	{
820 		.procname	= "hung_task_panic",
821 		.data		= &sysctl_hung_task_panic,
822 		.maxlen		= sizeof(int),
823 		.mode		= 0644,
824 		.proc_handler	= proc_dointvec_minmax,
825 		.extra1		= &zero,
826 		.extra2		= &one,
827 	},
828 	{
829 		.procname	= "hung_task_check_count",
830 		.data		= &sysctl_hung_task_check_count,
831 		.maxlen		= sizeof(unsigned long),
832 		.mode		= 0644,
833 		.proc_handler	= proc_doulongvec_minmax,
834 	},
835 	{
836 		.procname	= "hung_task_timeout_secs",
837 		.data		= &sysctl_hung_task_timeout_secs,
838 		.maxlen		= sizeof(unsigned long),
839 		.mode		= 0644,
840 		.proc_handler	= proc_dohung_task_timeout_secs,
841 	},
842 	{
843 		.procname	= "hung_task_warnings",
844 		.data		= &sysctl_hung_task_warnings,
845 		.maxlen		= sizeof(unsigned long),
846 		.mode		= 0644,
847 		.proc_handler	= proc_doulongvec_minmax,
848 	},
849 #endif
850 #ifdef CONFIG_COMPAT
851 	{
852 		.procname	= "compat-log",
853 		.data		= &compat_log,
854 		.maxlen		= sizeof (int),
855 	 	.mode		= 0644,
856 		.proc_handler	= proc_dointvec,
857 	},
858 #endif
859 #ifdef CONFIG_RT_MUTEXES
860 	{
861 		.procname	= "max_lock_depth",
862 		.data		= &max_lock_depth,
863 		.maxlen		= sizeof(int),
864 		.mode		= 0644,
865 		.proc_handler	= proc_dointvec,
866 	},
867 #endif
868 	{
869 		.procname	= "poweroff_cmd",
870 		.data		= &poweroff_cmd,
871 		.maxlen		= POWEROFF_CMD_PATH_LEN,
872 		.mode		= 0644,
873 		.proc_handler	= proc_dostring,
874 	},
875 #ifdef CONFIG_KEYS
876 	{
877 		.procname	= "keys",
878 		.mode		= 0555,
879 		.child		= key_sysctls,
880 	},
881 #endif
882 #ifdef CONFIG_RCU_TORTURE_TEST
883 	{
884 		.procname       = "rcutorture_runnable",
885 		.data           = &rcutorture_runnable,
886 		.maxlen         = sizeof(int),
887 		.mode           = 0644,
888 		.proc_handler	= proc_dointvec,
889 	},
890 #endif
891 #ifdef CONFIG_SLOW_WORK
892 	{
893 		.procname	= "slow-work",
894 		.mode		= 0555,
895 		.child		= slow_work_sysctls,
896 	},
897 #endif
898 #ifdef CONFIG_PERF_EVENTS
899 	{
900 		.procname	= "perf_event_paranoid",
901 		.data		= &sysctl_perf_event_paranoid,
902 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
903 		.mode		= 0644,
904 		.proc_handler	= proc_dointvec,
905 	},
906 	{
907 		.procname	= "perf_event_mlock_kb",
908 		.data		= &sysctl_perf_event_mlock,
909 		.maxlen		= sizeof(sysctl_perf_event_mlock),
910 		.mode		= 0644,
911 		.proc_handler	= proc_dointvec,
912 	},
913 	{
914 		.procname	= "perf_event_max_sample_rate",
915 		.data		= &sysctl_perf_event_sample_rate,
916 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
917 		.mode		= 0644,
918 		.proc_handler	= proc_dointvec,
919 	},
920 #endif
921 #ifdef CONFIG_KMEMCHECK
922 	{
923 		.procname	= "kmemcheck",
924 		.data		= &kmemcheck_enabled,
925 		.maxlen		= sizeof(int),
926 		.mode		= 0644,
927 		.proc_handler	= proc_dointvec,
928 	},
929 #endif
930 #ifdef CONFIG_BLOCK
931 	{
932 		.procname	= "blk_iopoll",
933 		.data		= &blk_iopoll_enabled,
934 		.maxlen		= sizeof(int),
935 		.mode		= 0644,
936 		.proc_handler	= proc_dointvec,
937 	},
938 #endif
939 /*
940  * NOTE: do not add new entries to this table unless you have read
941  * Documentation/sysctl/ctl_unnumbered.txt
942  */
943 	{ }
944 };
945 
946 static struct ctl_table vm_table[] = {
947 	{
948 		.procname	= "overcommit_memory",
949 		.data		= &sysctl_overcommit_memory,
950 		.maxlen		= sizeof(sysctl_overcommit_memory),
951 		.mode		= 0644,
952 		.proc_handler	= proc_dointvec,
953 	},
954 	{
955 		.procname	= "panic_on_oom",
956 		.data		= &sysctl_panic_on_oom,
957 		.maxlen		= sizeof(sysctl_panic_on_oom),
958 		.mode		= 0644,
959 		.proc_handler	= proc_dointvec,
960 	},
961 	{
962 		.procname	= "oom_kill_allocating_task",
963 		.data		= &sysctl_oom_kill_allocating_task,
964 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
965 		.mode		= 0644,
966 		.proc_handler	= proc_dointvec,
967 	},
968 	{
969 		.procname	= "oom_dump_tasks",
970 		.data		= &sysctl_oom_dump_tasks,
971 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
972 		.mode		= 0644,
973 		.proc_handler	= proc_dointvec,
974 	},
975 	{
976 		.procname	= "overcommit_ratio",
977 		.data		= &sysctl_overcommit_ratio,
978 		.maxlen		= sizeof(sysctl_overcommit_ratio),
979 		.mode		= 0644,
980 		.proc_handler	= proc_dointvec,
981 	},
982 	{
983 		.procname	= "page-cluster",
984 		.data		= &page_cluster,
985 		.maxlen		= sizeof(int),
986 		.mode		= 0644,
987 		.proc_handler	= proc_dointvec,
988 	},
989 	{
990 		.procname	= "dirty_background_ratio",
991 		.data		= &dirty_background_ratio,
992 		.maxlen		= sizeof(dirty_background_ratio),
993 		.mode		= 0644,
994 		.proc_handler	= dirty_background_ratio_handler,
995 		.extra1		= &zero,
996 		.extra2		= &one_hundred,
997 	},
998 	{
999 		.procname	= "dirty_background_bytes",
1000 		.data		= &dirty_background_bytes,
1001 		.maxlen		= sizeof(dirty_background_bytes),
1002 		.mode		= 0644,
1003 		.proc_handler	= dirty_background_bytes_handler,
1004 		.extra1		= &one_ul,
1005 	},
1006 	{
1007 		.procname	= "dirty_ratio",
1008 		.data		= &vm_dirty_ratio,
1009 		.maxlen		= sizeof(vm_dirty_ratio),
1010 		.mode		= 0644,
1011 		.proc_handler	= dirty_ratio_handler,
1012 		.extra1		= &zero,
1013 		.extra2		= &one_hundred,
1014 	},
1015 	{
1016 		.procname	= "dirty_bytes",
1017 		.data		= &vm_dirty_bytes,
1018 		.maxlen		= sizeof(vm_dirty_bytes),
1019 		.mode		= 0644,
1020 		.proc_handler	= dirty_bytes_handler,
1021 		.extra1		= &dirty_bytes_min,
1022 	},
1023 	{
1024 		.procname	= "dirty_writeback_centisecs",
1025 		.data		= &dirty_writeback_interval,
1026 		.maxlen		= sizeof(dirty_writeback_interval),
1027 		.mode		= 0644,
1028 		.proc_handler	= dirty_writeback_centisecs_handler,
1029 	},
1030 	{
1031 		.procname	= "dirty_expire_centisecs",
1032 		.data		= &dirty_expire_interval,
1033 		.maxlen		= sizeof(dirty_expire_interval),
1034 		.mode		= 0644,
1035 		.proc_handler	= proc_dointvec,
1036 	},
1037 	{
1038 		.procname	= "nr_pdflush_threads",
1039 		.data		= &nr_pdflush_threads,
1040 		.maxlen		= sizeof nr_pdflush_threads,
1041 		.mode		= 0444 /* read-only*/,
1042 		.proc_handler	= proc_dointvec,
1043 	},
1044 	{
1045 		.procname	= "swappiness",
1046 		.data		= &vm_swappiness,
1047 		.maxlen		= sizeof(vm_swappiness),
1048 		.mode		= 0644,
1049 		.proc_handler	= proc_dointvec_minmax,
1050 		.extra1		= &zero,
1051 		.extra2		= &one_hundred,
1052 	},
1053 #ifdef CONFIG_HUGETLB_PAGE
1054 	{
1055 		.procname	= "nr_hugepages",
1056 		.data		= NULL,
1057 		.maxlen		= sizeof(unsigned long),
1058 		.mode		= 0644,
1059 		.proc_handler	= hugetlb_sysctl_handler,
1060 		.extra1		= (void *)&hugetlb_zero,
1061 		.extra2		= (void *)&hugetlb_infinity,
1062 	},
1063 #ifdef CONFIG_NUMA
1064 	{
1065 		.procname       = "nr_hugepages_mempolicy",
1066 		.data           = NULL,
1067 		.maxlen         = sizeof(unsigned long),
1068 		.mode           = 0644,
1069 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1070 		.extra1		= (void *)&hugetlb_zero,
1071 		.extra2		= (void *)&hugetlb_infinity,
1072 	},
1073 #endif
1074 	 {
1075 		.procname	= "hugetlb_shm_group",
1076 		.data		= &sysctl_hugetlb_shm_group,
1077 		.maxlen		= sizeof(gid_t),
1078 		.mode		= 0644,
1079 		.proc_handler	= proc_dointvec,
1080 	 },
1081 	 {
1082 		.procname	= "hugepages_treat_as_movable",
1083 		.data		= &hugepages_treat_as_movable,
1084 		.maxlen		= sizeof(int),
1085 		.mode		= 0644,
1086 		.proc_handler	= hugetlb_treat_movable_handler,
1087 	},
1088 	{
1089 		.procname	= "nr_overcommit_hugepages",
1090 		.data		= NULL,
1091 		.maxlen		= sizeof(unsigned long),
1092 		.mode		= 0644,
1093 		.proc_handler	= hugetlb_overcommit_handler,
1094 		.extra1		= (void *)&hugetlb_zero,
1095 		.extra2		= (void *)&hugetlb_infinity,
1096 	},
1097 #endif
1098 	{
1099 		.procname	= "lowmem_reserve_ratio",
1100 		.data		= &sysctl_lowmem_reserve_ratio,
1101 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1102 		.mode		= 0644,
1103 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1104 	},
1105 	{
1106 		.procname	= "drop_caches",
1107 		.data		= &sysctl_drop_caches,
1108 		.maxlen		= sizeof(int),
1109 		.mode		= 0644,
1110 		.proc_handler	= drop_caches_sysctl_handler,
1111 	},
1112 	{
1113 		.procname	= "min_free_kbytes",
1114 		.data		= &min_free_kbytes,
1115 		.maxlen		= sizeof(min_free_kbytes),
1116 		.mode		= 0644,
1117 		.proc_handler	= min_free_kbytes_sysctl_handler,
1118 		.extra1		= &zero,
1119 	},
1120 	{
1121 		.procname	= "percpu_pagelist_fraction",
1122 		.data		= &percpu_pagelist_fraction,
1123 		.maxlen		= sizeof(percpu_pagelist_fraction),
1124 		.mode		= 0644,
1125 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1126 		.extra1		= &min_percpu_pagelist_fract,
1127 	},
1128 #ifdef CONFIG_MMU
1129 	{
1130 		.procname	= "max_map_count",
1131 		.data		= &sysctl_max_map_count,
1132 		.maxlen		= sizeof(sysctl_max_map_count),
1133 		.mode		= 0644,
1134 		.proc_handler	= proc_dointvec_minmax,
1135 		.extra1		= &zero,
1136 	},
1137 #else
1138 	{
1139 		.procname	= "nr_trim_pages",
1140 		.data		= &sysctl_nr_trim_pages,
1141 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1142 		.mode		= 0644,
1143 		.proc_handler	= proc_dointvec_minmax,
1144 		.extra1		= &zero,
1145 	},
1146 #endif
1147 	{
1148 		.procname	= "laptop_mode",
1149 		.data		= &laptop_mode,
1150 		.maxlen		= sizeof(laptop_mode),
1151 		.mode		= 0644,
1152 		.proc_handler	= proc_dointvec_jiffies,
1153 	},
1154 	{
1155 		.procname	= "block_dump",
1156 		.data		= &block_dump,
1157 		.maxlen		= sizeof(block_dump),
1158 		.mode		= 0644,
1159 		.proc_handler	= proc_dointvec,
1160 		.extra1		= &zero,
1161 	},
1162 	{
1163 		.procname	= "vfs_cache_pressure",
1164 		.data		= &sysctl_vfs_cache_pressure,
1165 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1166 		.mode		= 0644,
1167 		.proc_handler	= proc_dointvec,
1168 		.extra1		= &zero,
1169 	},
1170 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1171 	{
1172 		.procname	= "legacy_va_layout",
1173 		.data		= &sysctl_legacy_va_layout,
1174 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1175 		.mode		= 0644,
1176 		.proc_handler	= proc_dointvec,
1177 		.extra1		= &zero,
1178 	},
1179 #endif
1180 #ifdef CONFIG_NUMA
1181 	{
1182 		.procname	= "zone_reclaim_mode",
1183 		.data		= &zone_reclaim_mode,
1184 		.maxlen		= sizeof(zone_reclaim_mode),
1185 		.mode		= 0644,
1186 		.proc_handler	= proc_dointvec,
1187 		.extra1		= &zero,
1188 	},
1189 	{
1190 		.procname	= "min_unmapped_ratio",
1191 		.data		= &sysctl_min_unmapped_ratio,
1192 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1193 		.mode		= 0644,
1194 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1195 		.extra1		= &zero,
1196 		.extra2		= &one_hundred,
1197 	},
1198 	{
1199 		.procname	= "min_slab_ratio",
1200 		.data		= &sysctl_min_slab_ratio,
1201 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1202 		.mode		= 0644,
1203 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1204 		.extra1		= &zero,
1205 		.extra2		= &one_hundred,
1206 	},
1207 #endif
1208 #ifdef CONFIG_SMP
1209 	{
1210 		.procname	= "stat_interval",
1211 		.data		= &sysctl_stat_interval,
1212 		.maxlen		= sizeof(sysctl_stat_interval),
1213 		.mode		= 0644,
1214 		.proc_handler	= proc_dointvec_jiffies,
1215 	},
1216 #endif
1217 #ifdef CONFIG_MMU
1218 	{
1219 		.procname	= "mmap_min_addr",
1220 		.data		= &dac_mmap_min_addr,
1221 		.maxlen		= sizeof(unsigned long),
1222 		.mode		= 0644,
1223 		.proc_handler	= mmap_min_addr_handler,
1224 	},
1225 #endif
1226 #ifdef CONFIG_NUMA
1227 	{
1228 		.procname	= "numa_zonelist_order",
1229 		.data		= &numa_zonelist_order,
1230 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1231 		.mode		= 0644,
1232 		.proc_handler	= numa_zonelist_order_handler,
1233 	},
1234 #endif
1235 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1236    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1237 	{
1238 		.procname	= "vdso_enabled",
1239 		.data		= &vdso_enabled,
1240 		.maxlen		= sizeof(vdso_enabled),
1241 		.mode		= 0644,
1242 		.proc_handler	= proc_dointvec,
1243 		.extra1		= &zero,
1244 	},
1245 #endif
1246 #ifdef CONFIG_HIGHMEM
1247 	{
1248 		.procname	= "highmem_is_dirtyable",
1249 		.data		= &vm_highmem_is_dirtyable,
1250 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1251 		.mode		= 0644,
1252 		.proc_handler	= proc_dointvec_minmax,
1253 		.extra1		= &zero,
1254 		.extra2		= &one,
1255 	},
1256 #endif
1257 	{
1258 		.procname	= "scan_unevictable_pages",
1259 		.data		= &scan_unevictable_pages,
1260 		.maxlen		= sizeof(scan_unevictable_pages),
1261 		.mode		= 0644,
1262 		.proc_handler	= scan_unevictable_handler,
1263 	},
1264 #ifdef CONFIG_MEMORY_FAILURE
1265 	{
1266 		.procname	= "memory_failure_early_kill",
1267 		.data		= &sysctl_memory_failure_early_kill,
1268 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1269 		.mode		= 0644,
1270 		.proc_handler	= proc_dointvec_minmax,
1271 		.extra1		= &zero,
1272 		.extra2		= &one,
1273 	},
1274 	{
1275 		.procname	= "memory_failure_recovery",
1276 		.data		= &sysctl_memory_failure_recovery,
1277 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1278 		.mode		= 0644,
1279 		.proc_handler	= proc_dointvec_minmax,
1280 		.extra1		= &zero,
1281 		.extra2		= &one,
1282 	},
1283 #endif
1284 
1285 /*
1286  * NOTE: do not add new entries to this table unless you have read
1287  * Documentation/sysctl/ctl_unnumbered.txt
1288  */
1289 	{ }
1290 };
1291 
1292 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1293 static struct ctl_table binfmt_misc_table[] = {
1294 	{ }
1295 };
1296 #endif
1297 
1298 static struct ctl_table fs_table[] = {
1299 	{
1300 		.procname	= "inode-nr",
1301 		.data		= &inodes_stat,
1302 		.maxlen		= 2*sizeof(int),
1303 		.mode		= 0444,
1304 		.proc_handler	= proc_dointvec,
1305 	},
1306 	{
1307 		.procname	= "inode-state",
1308 		.data		= &inodes_stat,
1309 		.maxlen		= 7*sizeof(int),
1310 		.mode		= 0444,
1311 		.proc_handler	= proc_dointvec,
1312 	},
1313 	{
1314 		.procname	= "file-nr",
1315 		.data		= &files_stat,
1316 		.maxlen		= 3*sizeof(int),
1317 		.mode		= 0444,
1318 		.proc_handler	= proc_nr_files,
1319 	},
1320 	{
1321 		.procname	= "file-max",
1322 		.data		= &files_stat.max_files,
1323 		.maxlen		= sizeof(int),
1324 		.mode		= 0644,
1325 		.proc_handler	= proc_dointvec,
1326 	},
1327 	{
1328 		.procname	= "nr_open",
1329 		.data		= &sysctl_nr_open,
1330 		.maxlen		= sizeof(int),
1331 		.mode		= 0644,
1332 		.proc_handler	= proc_dointvec_minmax,
1333 		.extra1		= &sysctl_nr_open_min,
1334 		.extra2		= &sysctl_nr_open_max,
1335 	},
1336 	{
1337 		.procname	= "dentry-state",
1338 		.data		= &dentry_stat,
1339 		.maxlen		= 6*sizeof(int),
1340 		.mode		= 0444,
1341 		.proc_handler	= proc_dointvec,
1342 	},
1343 	{
1344 		.procname	= "overflowuid",
1345 		.data		= &fs_overflowuid,
1346 		.maxlen		= sizeof(int),
1347 		.mode		= 0644,
1348 		.proc_handler	= proc_dointvec_minmax,
1349 		.extra1		= &minolduid,
1350 		.extra2		= &maxolduid,
1351 	},
1352 	{
1353 		.procname	= "overflowgid",
1354 		.data		= &fs_overflowgid,
1355 		.maxlen		= sizeof(int),
1356 		.mode		= 0644,
1357 		.proc_handler	= proc_dointvec_minmax,
1358 		.extra1		= &minolduid,
1359 		.extra2		= &maxolduid,
1360 	},
1361 #ifdef CONFIG_FILE_LOCKING
1362 	{
1363 		.procname	= "leases-enable",
1364 		.data		= &leases_enable,
1365 		.maxlen		= sizeof(int),
1366 		.mode		= 0644,
1367 		.proc_handler	= proc_dointvec,
1368 	},
1369 #endif
1370 #ifdef CONFIG_DNOTIFY
1371 	{
1372 		.procname	= "dir-notify-enable",
1373 		.data		= &dir_notify_enable,
1374 		.maxlen		= sizeof(int),
1375 		.mode		= 0644,
1376 		.proc_handler	= proc_dointvec,
1377 	},
1378 #endif
1379 #ifdef CONFIG_MMU
1380 #ifdef CONFIG_FILE_LOCKING
1381 	{
1382 		.procname	= "lease-break-time",
1383 		.data		= &lease_break_time,
1384 		.maxlen		= sizeof(int),
1385 		.mode		= 0644,
1386 		.proc_handler	= proc_dointvec,
1387 	},
1388 #endif
1389 #ifdef CONFIG_AIO
1390 	{
1391 		.procname	= "aio-nr",
1392 		.data		= &aio_nr,
1393 		.maxlen		= sizeof(aio_nr),
1394 		.mode		= 0444,
1395 		.proc_handler	= proc_doulongvec_minmax,
1396 	},
1397 	{
1398 		.procname	= "aio-max-nr",
1399 		.data		= &aio_max_nr,
1400 		.maxlen		= sizeof(aio_max_nr),
1401 		.mode		= 0644,
1402 		.proc_handler	= proc_doulongvec_minmax,
1403 	},
1404 #endif /* CONFIG_AIO */
1405 #ifdef CONFIG_INOTIFY_USER
1406 	{
1407 		.procname	= "inotify",
1408 		.mode		= 0555,
1409 		.child		= inotify_table,
1410 	},
1411 #endif
1412 #ifdef CONFIG_EPOLL
1413 	{
1414 		.procname	= "epoll",
1415 		.mode		= 0555,
1416 		.child		= epoll_table,
1417 	},
1418 #endif
1419 #endif
1420 	{
1421 		.procname	= "suid_dumpable",
1422 		.data		= &suid_dumpable,
1423 		.maxlen		= sizeof(int),
1424 		.mode		= 0644,
1425 		.proc_handler	= proc_dointvec_minmax,
1426 		.extra1		= &zero,
1427 		.extra2		= &two,
1428 	},
1429 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1430 	{
1431 		.procname	= "binfmt_misc",
1432 		.mode		= 0555,
1433 		.child		= binfmt_misc_table,
1434 	},
1435 #endif
1436 /*
1437  * NOTE: do not add new entries to this table unless you have read
1438  * Documentation/sysctl/ctl_unnumbered.txt
1439  */
1440 	{ }
1441 };
1442 
1443 static struct ctl_table debug_table[] = {
1444 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1445 	{
1446 		.procname	= "exception-trace",
1447 		.data		= &show_unhandled_signals,
1448 		.maxlen		= sizeof(int),
1449 		.mode		= 0644,
1450 		.proc_handler	= proc_dointvec
1451 	},
1452 #endif
1453 	{ }
1454 };
1455 
1456 static struct ctl_table dev_table[] = {
1457 	{ }
1458 };
1459 
1460 static DEFINE_SPINLOCK(sysctl_lock);
1461 
1462 /* called under sysctl_lock */
1463 static int use_table(struct ctl_table_header *p)
1464 {
1465 	if (unlikely(p->unregistering))
1466 		return 0;
1467 	p->used++;
1468 	return 1;
1469 }
1470 
1471 /* called under sysctl_lock */
1472 static void unuse_table(struct ctl_table_header *p)
1473 {
1474 	if (!--p->used)
1475 		if (unlikely(p->unregistering))
1476 			complete(p->unregistering);
1477 }
1478 
1479 /* called under sysctl_lock, will reacquire if has to wait */
1480 static void start_unregistering(struct ctl_table_header *p)
1481 {
1482 	/*
1483 	 * if p->used is 0, nobody will ever touch that entry again;
1484 	 * we'll eliminate all paths to it before dropping sysctl_lock
1485 	 */
1486 	if (unlikely(p->used)) {
1487 		struct completion wait;
1488 		init_completion(&wait);
1489 		p->unregistering = &wait;
1490 		spin_unlock(&sysctl_lock);
1491 		wait_for_completion(&wait);
1492 		spin_lock(&sysctl_lock);
1493 	} else {
1494 		/* anything non-NULL; we'll never dereference it */
1495 		p->unregistering = ERR_PTR(-EINVAL);
1496 	}
1497 	/*
1498 	 * do not remove from the list until nobody holds it; walking the
1499 	 * list in do_sysctl() relies on that.
1500 	 */
1501 	list_del_init(&p->ctl_entry);
1502 }
1503 
1504 void sysctl_head_get(struct ctl_table_header *head)
1505 {
1506 	spin_lock(&sysctl_lock);
1507 	head->count++;
1508 	spin_unlock(&sysctl_lock);
1509 }
1510 
1511 void sysctl_head_put(struct ctl_table_header *head)
1512 {
1513 	spin_lock(&sysctl_lock);
1514 	if (!--head->count)
1515 		kfree(head);
1516 	spin_unlock(&sysctl_lock);
1517 }
1518 
1519 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1520 {
1521 	if (!head)
1522 		BUG();
1523 	spin_lock(&sysctl_lock);
1524 	if (!use_table(head))
1525 		head = ERR_PTR(-ENOENT);
1526 	spin_unlock(&sysctl_lock);
1527 	return head;
1528 }
1529 
1530 void sysctl_head_finish(struct ctl_table_header *head)
1531 {
1532 	if (!head)
1533 		return;
1534 	spin_lock(&sysctl_lock);
1535 	unuse_table(head);
1536 	spin_unlock(&sysctl_lock);
1537 }
1538 
1539 static struct ctl_table_set *
1540 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1541 {
1542 	struct ctl_table_set *set = &root->default_set;
1543 	if (root->lookup)
1544 		set = root->lookup(root, namespaces);
1545 	return set;
1546 }
1547 
1548 static struct list_head *
1549 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1550 {
1551 	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1552 	return &set->list;
1553 }
1554 
1555 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1556 					    struct ctl_table_header *prev)
1557 {
1558 	struct ctl_table_root *root;
1559 	struct list_head *header_list;
1560 	struct ctl_table_header *head;
1561 	struct list_head *tmp;
1562 
1563 	spin_lock(&sysctl_lock);
1564 	if (prev) {
1565 		head = prev;
1566 		tmp = &prev->ctl_entry;
1567 		unuse_table(prev);
1568 		goto next;
1569 	}
1570 	tmp = &root_table_header.ctl_entry;
1571 	for (;;) {
1572 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1573 
1574 		if (!use_table(head))
1575 			goto next;
1576 		spin_unlock(&sysctl_lock);
1577 		return head;
1578 	next:
1579 		root = head->root;
1580 		tmp = tmp->next;
1581 		header_list = lookup_header_list(root, namespaces);
1582 		if (tmp != header_list)
1583 			continue;
1584 
1585 		do {
1586 			root = list_entry(root->root_list.next,
1587 					struct ctl_table_root, root_list);
1588 			if (root == &sysctl_table_root)
1589 				goto out;
1590 			header_list = lookup_header_list(root, namespaces);
1591 		} while (list_empty(header_list));
1592 		tmp = header_list->next;
1593 	}
1594 out:
1595 	spin_unlock(&sysctl_lock);
1596 	return NULL;
1597 }
1598 
1599 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1600 {
1601 	return __sysctl_head_next(current->nsproxy, prev);
1602 }
1603 
1604 void register_sysctl_root(struct ctl_table_root *root)
1605 {
1606 	spin_lock(&sysctl_lock);
1607 	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1608 	spin_unlock(&sysctl_lock);
1609 }
1610 
1611 /*
1612  * sysctl_perm does NOT grant the superuser all rights automatically, because
1613  * some sysctl variables are readonly even to root.
1614  */
1615 
1616 static int test_perm(int mode, int op)
1617 {
1618 	if (!current_euid())
1619 		mode >>= 6;
1620 	else if (in_egroup_p(0))
1621 		mode >>= 3;
1622 	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1623 		return 0;
1624 	return -EACCES;
1625 }
1626 
1627 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1628 {
1629 	int error;
1630 	int mode;
1631 
1632 	error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1633 	if (error)
1634 		return error;
1635 
1636 	if (root->permissions)
1637 		mode = root->permissions(root, current->nsproxy, table);
1638 	else
1639 		mode = table->mode;
1640 
1641 	return test_perm(mode, op);
1642 }
1643 
1644 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1645 {
1646 	for (; table->procname; table++) {
1647 		table->parent = parent;
1648 		if (table->child)
1649 			sysctl_set_parent(table, table->child);
1650 	}
1651 }
1652 
1653 static __init int sysctl_init(void)
1654 {
1655 	sysctl_set_parent(NULL, root_table);
1656 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1657 	{
1658 		int err;
1659 		err = sysctl_check_table(current->nsproxy, root_table);
1660 	}
1661 #endif
1662 	return 0;
1663 }
1664 
1665 core_initcall(sysctl_init);
1666 
1667 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1668 				      struct ctl_table *table)
1669 {
1670 	struct ctl_table *p;
1671 	const char *s = branch->procname;
1672 
1673 	/* branch should have named subdirectory as its first element */
1674 	if (!s || !branch->child)
1675 		return NULL;
1676 
1677 	/* ... and nothing else */
1678 	if (branch[1].procname)
1679 		return NULL;
1680 
1681 	/* table should contain subdirectory with the same name */
1682 	for (p = table; p->procname; p++) {
1683 		if (!p->child)
1684 			continue;
1685 		if (p->procname && strcmp(p->procname, s) == 0)
1686 			return p;
1687 	}
1688 	return NULL;
1689 }
1690 
1691 /* see if attaching q to p would be an improvement */
1692 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1693 {
1694 	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1695 	struct ctl_table *next;
1696 	int is_better = 0;
1697 	int not_in_parent = !p->attached_by;
1698 
1699 	while ((next = is_branch_in(by, to)) != NULL) {
1700 		if (by == q->attached_by)
1701 			is_better = 1;
1702 		if (to == p->attached_by)
1703 			not_in_parent = 1;
1704 		by = by->child;
1705 		to = next->child;
1706 	}
1707 
1708 	if (is_better && not_in_parent) {
1709 		q->attached_by = by;
1710 		q->attached_to = to;
1711 		q->parent = p;
1712 	}
1713 }
1714 
1715 /**
1716  * __register_sysctl_paths - register a sysctl hierarchy
1717  * @root: List of sysctl headers to register on
1718  * @namespaces: Data to compute which lists of sysctl entries are visible
1719  * @path: The path to the directory the sysctl table is in.
1720  * @table: the top-level table structure
1721  *
1722  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1723  * array. A completely 0 filled entry terminates the table.
1724  *
1725  * The members of the &struct ctl_table structure are used as follows:
1726  *
1727  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1728  *            enter a sysctl file
1729  *
1730  * data - a pointer to data for use by proc_handler
1731  *
1732  * maxlen - the maximum size in bytes of the data
1733  *
1734  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1735  *
1736  * child - a pointer to the child sysctl table if this entry is a directory, or
1737  *         %NULL.
1738  *
1739  * proc_handler - the text handler routine (described below)
1740  *
1741  * de - for internal use by the sysctl routines
1742  *
1743  * extra1, extra2 - extra pointers usable by the proc handler routines
1744  *
1745  * Leaf nodes in the sysctl tree will be represented by a single file
1746  * under /proc; non-leaf nodes will be represented by directories.
1747  *
1748  * sysctl(2) can automatically manage read and write requests through
1749  * the sysctl table.  The data and maxlen fields of the ctl_table
1750  * struct enable minimal validation of the values being written to be
1751  * performed, and the mode field allows minimal authentication.
1752  *
1753  * There must be a proc_handler routine for any terminal nodes
1754  * mirrored under /proc/sys (non-terminals are handled by a built-in
1755  * directory handler).  Several default handlers are available to
1756  * cover common cases -
1757  *
1758  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1759  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1760  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1761  *
1762  * It is the handler's job to read the input buffer from user memory
1763  * and process it. The handler should return 0 on success.
1764  *
1765  * This routine returns %NULL on a failure to register, and a pointer
1766  * to the table header on success.
1767  */
1768 struct ctl_table_header *__register_sysctl_paths(
1769 	struct ctl_table_root *root,
1770 	struct nsproxy *namespaces,
1771 	const struct ctl_path *path, struct ctl_table *table)
1772 {
1773 	struct ctl_table_header *header;
1774 	struct ctl_table *new, **prevp;
1775 	unsigned int n, npath;
1776 	struct ctl_table_set *set;
1777 
1778 	/* Count the path components */
1779 	for (npath = 0; path[npath].procname; ++npath)
1780 		;
1781 
1782 	/*
1783 	 * For each path component, allocate a 2-element ctl_table array.
1784 	 * The first array element will be filled with the sysctl entry
1785 	 * for this, the second will be the sentinel (procname == 0).
1786 	 *
1787 	 * We allocate everything in one go so that we don't have to
1788 	 * worry about freeing additional memory in unregister_sysctl_table.
1789 	 */
1790 	header = kzalloc(sizeof(struct ctl_table_header) +
1791 			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1792 	if (!header)
1793 		return NULL;
1794 
1795 	new = (struct ctl_table *) (header + 1);
1796 
1797 	/* Now connect the dots */
1798 	prevp = &header->ctl_table;
1799 	for (n = 0; n < npath; ++n, ++path) {
1800 		/* Copy the procname */
1801 		new->procname = path->procname;
1802 		new->mode     = 0555;
1803 
1804 		*prevp = new;
1805 		prevp = &new->child;
1806 
1807 		new += 2;
1808 	}
1809 	*prevp = table;
1810 	header->ctl_table_arg = table;
1811 
1812 	INIT_LIST_HEAD(&header->ctl_entry);
1813 	header->used = 0;
1814 	header->unregistering = NULL;
1815 	header->root = root;
1816 	sysctl_set_parent(NULL, header->ctl_table);
1817 	header->count = 1;
1818 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1819 	if (sysctl_check_table(namespaces, header->ctl_table)) {
1820 		kfree(header);
1821 		return NULL;
1822 	}
1823 #endif
1824 	spin_lock(&sysctl_lock);
1825 	header->set = lookup_header_set(root, namespaces);
1826 	header->attached_by = header->ctl_table;
1827 	header->attached_to = root_table;
1828 	header->parent = &root_table_header;
1829 	for (set = header->set; set; set = set->parent) {
1830 		struct ctl_table_header *p;
1831 		list_for_each_entry(p, &set->list, ctl_entry) {
1832 			if (p->unregistering)
1833 				continue;
1834 			try_attach(p, header);
1835 		}
1836 	}
1837 	header->parent->count++;
1838 	list_add_tail(&header->ctl_entry, &header->set->list);
1839 	spin_unlock(&sysctl_lock);
1840 
1841 	return header;
1842 }
1843 
1844 /**
1845  * register_sysctl_table_path - register a sysctl table hierarchy
1846  * @path: The path to the directory the sysctl table is in.
1847  * @table: the top-level table structure
1848  *
1849  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1850  * array. A completely 0 filled entry terminates the table.
1851  *
1852  * See __register_sysctl_paths for more details.
1853  */
1854 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1855 						struct ctl_table *table)
1856 {
1857 	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1858 					path, table);
1859 }
1860 
1861 /**
1862  * register_sysctl_table - register a sysctl table hierarchy
1863  * @table: the top-level table structure
1864  *
1865  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1866  * array. A completely 0 filled entry terminates the table.
1867  *
1868  * See register_sysctl_paths for more details.
1869  */
1870 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1871 {
1872 	static const struct ctl_path null_path[] = { {} };
1873 
1874 	return register_sysctl_paths(null_path, table);
1875 }
1876 
1877 /**
1878  * unregister_sysctl_table - unregister a sysctl table hierarchy
1879  * @header: the header returned from register_sysctl_table
1880  *
1881  * Unregisters the sysctl table and all children. proc entries may not
1882  * actually be removed until they are no longer used by anyone.
1883  */
1884 void unregister_sysctl_table(struct ctl_table_header * header)
1885 {
1886 	might_sleep();
1887 
1888 	if (header == NULL)
1889 		return;
1890 
1891 	spin_lock(&sysctl_lock);
1892 	start_unregistering(header);
1893 	if (!--header->parent->count) {
1894 		WARN_ON(1);
1895 		kfree(header->parent);
1896 	}
1897 	if (!--header->count)
1898 		kfree(header);
1899 	spin_unlock(&sysctl_lock);
1900 }
1901 
1902 int sysctl_is_seen(struct ctl_table_header *p)
1903 {
1904 	struct ctl_table_set *set = p->set;
1905 	int res;
1906 	spin_lock(&sysctl_lock);
1907 	if (p->unregistering)
1908 		res = 0;
1909 	else if (!set->is_seen)
1910 		res = 1;
1911 	else
1912 		res = set->is_seen(set);
1913 	spin_unlock(&sysctl_lock);
1914 	return res;
1915 }
1916 
1917 void setup_sysctl_set(struct ctl_table_set *p,
1918 	struct ctl_table_set *parent,
1919 	int (*is_seen)(struct ctl_table_set *))
1920 {
1921 	INIT_LIST_HEAD(&p->list);
1922 	p->parent = parent ? parent : &sysctl_table_root.default_set;
1923 	p->is_seen = is_seen;
1924 }
1925 
1926 #else /* !CONFIG_SYSCTL */
1927 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1928 {
1929 	return NULL;
1930 }
1931 
1932 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1933 						    struct ctl_table *table)
1934 {
1935 	return NULL;
1936 }
1937 
1938 void unregister_sysctl_table(struct ctl_table_header * table)
1939 {
1940 }
1941 
1942 void setup_sysctl_set(struct ctl_table_set *p,
1943 	struct ctl_table_set *parent,
1944 	int (*is_seen)(struct ctl_table_set *))
1945 {
1946 }
1947 
1948 void sysctl_head_put(struct ctl_table_header *head)
1949 {
1950 }
1951 
1952 #endif /* CONFIG_SYSCTL */
1953 
1954 /*
1955  * /proc/sys support
1956  */
1957 
1958 #ifdef CONFIG_PROC_SYSCTL
1959 
1960 static int _proc_do_string(void* data, int maxlen, int write,
1961 			   void __user *buffer,
1962 			   size_t *lenp, loff_t *ppos)
1963 {
1964 	size_t len;
1965 	char __user *p;
1966 	char c;
1967 
1968 	if (!data || !maxlen || !*lenp) {
1969 		*lenp = 0;
1970 		return 0;
1971 	}
1972 
1973 	if (write) {
1974 		len = 0;
1975 		p = buffer;
1976 		while (len < *lenp) {
1977 			if (get_user(c, p++))
1978 				return -EFAULT;
1979 			if (c == 0 || c == '\n')
1980 				break;
1981 			len++;
1982 		}
1983 		if (len >= maxlen)
1984 			len = maxlen-1;
1985 		if(copy_from_user(data, buffer, len))
1986 			return -EFAULT;
1987 		((char *) data)[len] = 0;
1988 		*ppos += *lenp;
1989 	} else {
1990 		len = strlen(data);
1991 		if (len > maxlen)
1992 			len = maxlen;
1993 
1994 		if (*ppos > len) {
1995 			*lenp = 0;
1996 			return 0;
1997 		}
1998 
1999 		data += *ppos;
2000 		len  -= *ppos;
2001 
2002 		if (len > *lenp)
2003 			len = *lenp;
2004 		if (len)
2005 			if(copy_to_user(buffer, data, len))
2006 				return -EFAULT;
2007 		if (len < *lenp) {
2008 			if(put_user('\n', ((char __user *) buffer) + len))
2009 				return -EFAULT;
2010 			len++;
2011 		}
2012 		*lenp = len;
2013 		*ppos += len;
2014 	}
2015 	return 0;
2016 }
2017 
2018 /**
2019  * proc_dostring - read a string sysctl
2020  * @table: the sysctl table
2021  * @write: %TRUE if this is a write to the sysctl file
2022  * @buffer: the user buffer
2023  * @lenp: the size of the user buffer
2024  * @ppos: file position
2025  *
2026  * Reads/writes a string from/to the user buffer. If the kernel
2027  * buffer provided is not large enough to hold the string, the
2028  * string is truncated. The copied string is %NULL-terminated.
2029  * If the string is being read by the user process, it is copied
2030  * and a newline '\n' is added. It is truncated if the buffer is
2031  * not large enough.
2032  *
2033  * Returns 0 on success.
2034  */
2035 int proc_dostring(struct ctl_table *table, int write,
2036 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2037 {
2038 	return _proc_do_string(table->data, table->maxlen, write,
2039 			       buffer, lenp, ppos);
2040 }
2041 
2042 
2043 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2044 				 int *valp,
2045 				 int write, void *data)
2046 {
2047 	if (write) {
2048 		*valp = *negp ? -*lvalp : *lvalp;
2049 	} else {
2050 		int val = *valp;
2051 		if (val < 0) {
2052 			*negp = -1;
2053 			*lvalp = (unsigned long)-val;
2054 		} else {
2055 			*negp = 0;
2056 			*lvalp = (unsigned long)val;
2057 		}
2058 	}
2059 	return 0;
2060 }
2061 
2062 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2063 		  int write, void __user *buffer,
2064 		  size_t *lenp, loff_t *ppos,
2065 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2066 			      int write, void *data),
2067 		  void *data)
2068 {
2069 #define TMPBUFLEN 21
2070 	int *i, vleft, first = 1, neg;
2071 	unsigned long lval;
2072 	size_t left, len;
2073 
2074 	char buf[TMPBUFLEN], *p;
2075 	char __user *s = buffer;
2076 
2077 	if (!tbl_data || !table->maxlen || !*lenp ||
2078 	    (*ppos && !write)) {
2079 		*lenp = 0;
2080 		return 0;
2081 	}
2082 
2083 	i = (int *) tbl_data;
2084 	vleft = table->maxlen / sizeof(*i);
2085 	left = *lenp;
2086 
2087 	if (!conv)
2088 		conv = do_proc_dointvec_conv;
2089 
2090 	for (; left && vleft--; i++, first=0) {
2091 		if (write) {
2092 			while (left) {
2093 				char c;
2094 				if (get_user(c, s))
2095 					return -EFAULT;
2096 				if (!isspace(c))
2097 					break;
2098 				left--;
2099 				s++;
2100 			}
2101 			if (!left)
2102 				break;
2103 			neg = 0;
2104 			len = left;
2105 			if (len > sizeof(buf) - 1)
2106 				len = sizeof(buf) - 1;
2107 			if (copy_from_user(buf, s, len))
2108 				return -EFAULT;
2109 			buf[len] = 0;
2110 			p = buf;
2111 			if (*p == '-' && left > 1) {
2112 				neg = 1;
2113 				p++;
2114 			}
2115 			if (*p < '0' || *p > '9')
2116 				break;
2117 
2118 			lval = simple_strtoul(p, &p, 0);
2119 
2120 			len = p-buf;
2121 			if ((len < left) && *p && !isspace(*p))
2122 				break;
2123 			s += len;
2124 			left -= len;
2125 
2126 			if (conv(&neg, &lval, i, 1, data))
2127 				break;
2128 		} else {
2129 			p = buf;
2130 			if (!first)
2131 				*p++ = '\t';
2132 
2133 			if (conv(&neg, &lval, i, 0, data))
2134 				break;
2135 
2136 			sprintf(p, "%s%lu", neg ? "-" : "", lval);
2137 			len = strlen(buf);
2138 			if (len > left)
2139 				len = left;
2140 			if(copy_to_user(s, buf, len))
2141 				return -EFAULT;
2142 			left -= len;
2143 			s += len;
2144 		}
2145 	}
2146 
2147 	if (!write && !first && left) {
2148 		if(put_user('\n', s))
2149 			return -EFAULT;
2150 		left--, s++;
2151 	}
2152 	if (write) {
2153 		while (left) {
2154 			char c;
2155 			if (get_user(c, s++))
2156 				return -EFAULT;
2157 			if (!isspace(c))
2158 				break;
2159 			left--;
2160 		}
2161 	}
2162 	if (write && first)
2163 		return -EINVAL;
2164 	*lenp -= left;
2165 	*ppos += *lenp;
2166 	return 0;
2167 #undef TMPBUFLEN
2168 }
2169 
2170 static int do_proc_dointvec(struct ctl_table *table, int write,
2171 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2172 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2173 			      int write, void *data),
2174 		  void *data)
2175 {
2176 	return __do_proc_dointvec(table->data, table, write,
2177 			buffer, lenp, ppos, conv, data);
2178 }
2179 
2180 /**
2181  * proc_dointvec - read a vector of integers
2182  * @table: the sysctl table
2183  * @write: %TRUE if this is a write to the sysctl file
2184  * @buffer: the user buffer
2185  * @lenp: the size of the user buffer
2186  * @ppos: file position
2187  *
2188  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2189  * values from/to the user buffer, treated as an ASCII string.
2190  *
2191  * Returns 0 on success.
2192  */
2193 int proc_dointvec(struct ctl_table *table, int write,
2194 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2195 {
2196     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2197 		    	    NULL,NULL);
2198 }
2199 
2200 /*
2201  * Taint values can only be increased
2202  * This means we can safely use a temporary.
2203  */
2204 static int proc_taint(struct ctl_table *table, int write,
2205 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2206 {
2207 	struct ctl_table t;
2208 	unsigned long tmptaint = get_taint();
2209 	int err;
2210 
2211 	if (write && !capable(CAP_SYS_ADMIN))
2212 		return -EPERM;
2213 
2214 	t = *table;
2215 	t.data = &tmptaint;
2216 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2217 	if (err < 0)
2218 		return err;
2219 
2220 	if (write) {
2221 		/*
2222 		 * Poor man's atomic or. Not worth adding a primitive
2223 		 * to everyone's atomic.h for this
2224 		 */
2225 		int i;
2226 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2227 			if ((tmptaint >> i) & 1)
2228 				add_taint(i);
2229 		}
2230 	}
2231 
2232 	return err;
2233 }
2234 
2235 struct do_proc_dointvec_minmax_conv_param {
2236 	int *min;
2237 	int *max;
2238 };
2239 
2240 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2241 					int *valp,
2242 					int write, void *data)
2243 {
2244 	struct do_proc_dointvec_minmax_conv_param *param = data;
2245 	if (write) {
2246 		int val = *negp ? -*lvalp : *lvalp;
2247 		if ((param->min && *param->min > val) ||
2248 		    (param->max && *param->max < val))
2249 			return -EINVAL;
2250 		*valp = val;
2251 	} else {
2252 		int val = *valp;
2253 		if (val < 0) {
2254 			*negp = -1;
2255 			*lvalp = (unsigned long)-val;
2256 		} else {
2257 			*negp = 0;
2258 			*lvalp = (unsigned long)val;
2259 		}
2260 	}
2261 	return 0;
2262 }
2263 
2264 /**
2265  * proc_dointvec_minmax - read a vector of integers with min/max values
2266  * @table: the sysctl table
2267  * @write: %TRUE if this is a write to the sysctl file
2268  * @buffer: the user buffer
2269  * @lenp: the size of the user buffer
2270  * @ppos: file position
2271  *
2272  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2273  * values from/to the user buffer, treated as an ASCII string.
2274  *
2275  * This routine will ensure the values are within the range specified by
2276  * table->extra1 (min) and table->extra2 (max).
2277  *
2278  * Returns 0 on success.
2279  */
2280 int proc_dointvec_minmax(struct ctl_table *table, int write,
2281 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2282 {
2283 	struct do_proc_dointvec_minmax_conv_param param = {
2284 		.min = (int *) table->extra1,
2285 		.max = (int *) table->extra2,
2286 	};
2287 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2288 				do_proc_dointvec_minmax_conv, &param);
2289 }
2290 
2291 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2292 				     void __user *buffer,
2293 				     size_t *lenp, loff_t *ppos,
2294 				     unsigned long convmul,
2295 				     unsigned long convdiv)
2296 {
2297 #define TMPBUFLEN 21
2298 	unsigned long *i, *min, *max, val;
2299 	int vleft, first=1, neg;
2300 	size_t len, left;
2301 	char buf[TMPBUFLEN], *p;
2302 	char __user *s = buffer;
2303 
2304 	if (!data || !table->maxlen || !*lenp ||
2305 	    (*ppos && !write)) {
2306 		*lenp = 0;
2307 		return 0;
2308 	}
2309 
2310 	i = (unsigned long *) data;
2311 	min = (unsigned long *) table->extra1;
2312 	max = (unsigned long *) table->extra2;
2313 	vleft = table->maxlen / sizeof(unsigned long);
2314 	left = *lenp;
2315 
2316 	for (; left && vleft--; i++, min++, max++, first=0) {
2317 		if (write) {
2318 			while (left) {
2319 				char c;
2320 				if (get_user(c, s))
2321 					return -EFAULT;
2322 				if (!isspace(c))
2323 					break;
2324 				left--;
2325 				s++;
2326 			}
2327 			if (!left)
2328 				break;
2329 			neg = 0;
2330 			len = left;
2331 			if (len > TMPBUFLEN-1)
2332 				len = TMPBUFLEN-1;
2333 			if (copy_from_user(buf, s, len))
2334 				return -EFAULT;
2335 			buf[len] = 0;
2336 			p = buf;
2337 			if (*p == '-' && left > 1) {
2338 				neg = 1;
2339 				p++;
2340 			}
2341 			if (*p < '0' || *p > '9')
2342 				break;
2343 			val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2344 			len = p-buf;
2345 			if ((len < left) && *p && !isspace(*p))
2346 				break;
2347 			if (neg)
2348 				val = -val;
2349 			s += len;
2350 			left -= len;
2351 
2352 			if(neg)
2353 				continue;
2354 			if ((min && val < *min) || (max && val > *max))
2355 				continue;
2356 			*i = val;
2357 		} else {
2358 			p = buf;
2359 			if (!first)
2360 				*p++ = '\t';
2361 			sprintf(p, "%lu", convdiv * (*i) / convmul);
2362 			len = strlen(buf);
2363 			if (len > left)
2364 				len = left;
2365 			if(copy_to_user(s, buf, len))
2366 				return -EFAULT;
2367 			left -= len;
2368 			s += len;
2369 		}
2370 	}
2371 
2372 	if (!write && !first && left) {
2373 		if(put_user('\n', s))
2374 			return -EFAULT;
2375 		left--, s++;
2376 	}
2377 	if (write) {
2378 		while (left) {
2379 			char c;
2380 			if (get_user(c, s++))
2381 				return -EFAULT;
2382 			if (!isspace(c))
2383 				break;
2384 			left--;
2385 		}
2386 	}
2387 	if (write && first)
2388 		return -EINVAL;
2389 	*lenp -= left;
2390 	*ppos += *lenp;
2391 	return 0;
2392 #undef TMPBUFLEN
2393 }
2394 
2395 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2396 				     void __user *buffer,
2397 				     size_t *lenp, loff_t *ppos,
2398 				     unsigned long convmul,
2399 				     unsigned long convdiv)
2400 {
2401 	return __do_proc_doulongvec_minmax(table->data, table, write,
2402 			buffer, lenp, ppos, convmul, convdiv);
2403 }
2404 
2405 /**
2406  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2407  * @table: the sysctl table
2408  * @write: %TRUE if this is a write to the sysctl file
2409  * @buffer: the user buffer
2410  * @lenp: the size of the user buffer
2411  * @ppos: file position
2412  *
2413  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2414  * values from/to the user buffer, treated as an ASCII string.
2415  *
2416  * This routine will ensure the values are within the range specified by
2417  * table->extra1 (min) and table->extra2 (max).
2418  *
2419  * Returns 0 on success.
2420  */
2421 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2422 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2423 {
2424     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2425 }
2426 
2427 /**
2428  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2429  * @table: the sysctl table
2430  * @write: %TRUE if this is a write to the sysctl file
2431  * @buffer: the user buffer
2432  * @lenp: the size of the user buffer
2433  * @ppos: file position
2434  *
2435  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2436  * values from/to the user buffer, treated as an ASCII string. The values
2437  * are treated as milliseconds, and converted to jiffies when they are stored.
2438  *
2439  * This routine will ensure the values are within the range specified by
2440  * table->extra1 (min) and table->extra2 (max).
2441  *
2442  * Returns 0 on success.
2443  */
2444 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2445 				      void __user *buffer,
2446 				      size_t *lenp, loff_t *ppos)
2447 {
2448     return do_proc_doulongvec_minmax(table, write, buffer,
2449 				     lenp, ppos, HZ, 1000l);
2450 }
2451 
2452 
2453 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2454 					 int *valp,
2455 					 int write, void *data)
2456 {
2457 	if (write) {
2458 		if (*lvalp > LONG_MAX / HZ)
2459 			return 1;
2460 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2461 	} else {
2462 		int val = *valp;
2463 		unsigned long lval;
2464 		if (val < 0) {
2465 			*negp = -1;
2466 			lval = (unsigned long)-val;
2467 		} else {
2468 			*negp = 0;
2469 			lval = (unsigned long)val;
2470 		}
2471 		*lvalp = lval / HZ;
2472 	}
2473 	return 0;
2474 }
2475 
2476 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2477 						int *valp,
2478 						int write, void *data)
2479 {
2480 	if (write) {
2481 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2482 			return 1;
2483 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2484 	} else {
2485 		int val = *valp;
2486 		unsigned long lval;
2487 		if (val < 0) {
2488 			*negp = -1;
2489 			lval = (unsigned long)-val;
2490 		} else {
2491 			*negp = 0;
2492 			lval = (unsigned long)val;
2493 		}
2494 		*lvalp = jiffies_to_clock_t(lval);
2495 	}
2496 	return 0;
2497 }
2498 
2499 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2500 					    int *valp,
2501 					    int write, void *data)
2502 {
2503 	if (write) {
2504 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2505 	} else {
2506 		int val = *valp;
2507 		unsigned long lval;
2508 		if (val < 0) {
2509 			*negp = -1;
2510 			lval = (unsigned long)-val;
2511 		} else {
2512 			*negp = 0;
2513 			lval = (unsigned long)val;
2514 		}
2515 		*lvalp = jiffies_to_msecs(lval);
2516 	}
2517 	return 0;
2518 }
2519 
2520 /**
2521  * proc_dointvec_jiffies - read a vector of integers as seconds
2522  * @table: the sysctl table
2523  * @write: %TRUE if this is a write to the sysctl file
2524  * @buffer: the user buffer
2525  * @lenp: the size of the user buffer
2526  * @ppos: file position
2527  *
2528  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2529  * values from/to the user buffer, treated as an ASCII string.
2530  * The values read are assumed to be in seconds, and are converted into
2531  * jiffies.
2532  *
2533  * Returns 0 on success.
2534  */
2535 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2536 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2537 {
2538     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2539 		    	    do_proc_dointvec_jiffies_conv,NULL);
2540 }
2541 
2542 /**
2543  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2544  * @table: the sysctl table
2545  * @write: %TRUE if this is a write to the sysctl file
2546  * @buffer: the user buffer
2547  * @lenp: the size of the user buffer
2548  * @ppos: pointer to the file position
2549  *
2550  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2551  * values from/to the user buffer, treated as an ASCII string.
2552  * The values read are assumed to be in 1/USER_HZ seconds, and
2553  * are converted into jiffies.
2554  *
2555  * Returns 0 on success.
2556  */
2557 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2558 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2559 {
2560     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2561 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2562 }
2563 
2564 /**
2565  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2566  * @table: the sysctl table
2567  * @write: %TRUE if this is a write to the sysctl file
2568  * @buffer: the user buffer
2569  * @lenp: the size of the user buffer
2570  * @ppos: file position
2571  * @ppos: the current position in the file
2572  *
2573  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2574  * values from/to the user buffer, treated as an ASCII string.
2575  * The values read are assumed to be in 1/1000 seconds, and
2576  * are converted into jiffies.
2577  *
2578  * Returns 0 on success.
2579  */
2580 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2581 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2582 {
2583 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2584 				do_proc_dointvec_ms_jiffies_conv, NULL);
2585 }
2586 
2587 static int proc_do_cad_pid(struct ctl_table *table, int write,
2588 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2589 {
2590 	struct pid *new_pid;
2591 	pid_t tmp;
2592 	int r;
2593 
2594 	tmp = pid_vnr(cad_pid);
2595 
2596 	r = __do_proc_dointvec(&tmp, table, write, buffer,
2597 			       lenp, ppos, NULL, NULL);
2598 	if (r || !write)
2599 		return r;
2600 
2601 	new_pid = find_get_pid(tmp);
2602 	if (!new_pid)
2603 		return -ESRCH;
2604 
2605 	put_pid(xchg(&cad_pid, new_pid));
2606 	return 0;
2607 }
2608 
2609 #else /* CONFIG_PROC_FS */
2610 
2611 int proc_dostring(struct ctl_table *table, int write,
2612 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2613 {
2614 	return -ENOSYS;
2615 }
2616 
2617 int proc_dointvec(struct ctl_table *table, int write,
2618 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2619 {
2620 	return -ENOSYS;
2621 }
2622 
2623 int proc_dointvec_minmax(struct ctl_table *table, int write,
2624 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2625 {
2626 	return -ENOSYS;
2627 }
2628 
2629 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2630 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2631 {
2632 	return -ENOSYS;
2633 }
2634 
2635 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2636 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2637 {
2638 	return -ENOSYS;
2639 }
2640 
2641 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2642 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2643 {
2644 	return -ENOSYS;
2645 }
2646 
2647 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2648 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2649 {
2650 	return -ENOSYS;
2651 }
2652 
2653 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2654 				      void __user *buffer,
2655 				      size_t *lenp, loff_t *ppos)
2656 {
2657     return -ENOSYS;
2658 }
2659 
2660 
2661 #endif /* CONFIG_PROC_FS */
2662 
2663 /*
2664  * No sense putting this after each symbol definition, twice,
2665  * exception granted :-)
2666  */
2667 EXPORT_SYMBOL(proc_dointvec);
2668 EXPORT_SYMBOL(proc_dointvec_jiffies);
2669 EXPORT_SYMBOL(proc_dointvec_minmax);
2670 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2671 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2672 EXPORT_SYMBOL(proc_dostring);
2673 EXPORT_SYMBOL(proc_doulongvec_minmax);
2674 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2675 EXPORT_SYMBOL(register_sysctl_table);
2676 EXPORT_SYMBOL(register_sysctl_paths);
2677 EXPORT_SYMBOL(unregister_sysctl_table);
2678