xref: /openbmc/linux/kernel/sysctl.c (revision 3e26120cc7c819c97bc07281ca1fb9017cfe9a39)
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 	{
1218 		.procname	= "mmap_min_addr",
1219 		.data		= &dac_mmap_min_addr,
1220 		.maxlen		= sizeof(unsigned long),
1221 		.mode		= 0644,
1222 		.proc_handler	= mmap_min_addr_handler,
1223 	},
1224 #ifdef CONFIG_NUMA
1225 	{
1226 		.procname	= "numa_zonelist_order",
1227 		.data		= &numa_zonelist_order,
1228 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1229 		.mode		= 0644,
1230 		.proc_handler	= numa_zonelist_order_handler,
1231 	},
1232 #endif
1233 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1234    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1235 	{
1236 		.procname	= "vdso_enabled",
1237 		.data		= &vdso_enabled,
1238 		.maxlen		= sizeof(vdso_enabled),
1239 		.mode		= 0644,
1240 		.proc_handler	= proc_dointvec,
1241 		.extra1		= &zero,
1242 	},
1243 #endif
1244 #ifdef CONFIG_HIGHMEM
1245 	{
1246 		.procname	= "highmem_is_dirtyable",
1247 		.data		= &vm_highmem_is_dirtyable,
1248 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1249 		.mode		= 0644,
1250 		.proc_handler	= proc_dointvec_minmax,
1251 		.extra1		= &zero,
1252 		.extra2		= &one,
1253 	},
1254 #endif
1255 	{
1256 		.procname	= "scan_unevictable_pages",
1257 		.data		= &scan_unevictable_pages,
1258 		.maxlen		= sizeof(scan_unevictable_pages),
1259 		.mode		= 0644,
1260 		.proc_handler	= scan_unevictable_handler,
1261 	},
1262 #ifdef CONFIG_MEMORY_FAILURE
1263 	{
1264 		.procname	= "memory_failure_early_kill",
1265 		.data		= &sysctl_memory_failure_early_kill,
1266 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1267 		.mode		= 0644,
1268 		.proc_handler	= proc_dointvec_minmax,
1269 		.extra1		= &zero,
1270 		.extra2		= &one,
1271 	},
1272 	{
1273 		.procname	= "memory_failure_recovery",
1274 		.data		= &sysctl_memory_failure_recovery,
1275 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1276 		.mode		= 0644,
1277 		.proc_handler	= proc_dointvec_minmax,
1278 		.extra1		= &zero,
1279 		.extra2		= &one,
1280 	},
1281 #endif
1282 
1283 /*
1284  * NOTE: do not add new entries to this table unless you have read
1285  * Documentation/sysctl/ctl_unnumbered.txt
1286  */
1287 	{ }
1288 };
1289 
1290 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1291 static struct ctl_table binfmt_misc_table[] = {
1292 	{ }
1293 };
1294 #endif
1295 
1296 static struct ctl_table fs_table[] = {
1297 	{
1298 		.procname	= "inode-nr",
1299 		.data		= &inodes_stat,
1300 		.maxlen		= 2*sizeof(int),
1301 		.mode		= 0444,
1302 		.proc_handler	= proc_dointvec,
1303 	},
1304 	{
1305 		.procname	= "inode-state",
1306 		.data		= &inodes_stat,
1307 		.maxlen		= 7*sizeof(int),
1308 		.mode		= 0444,
1309 		.proc_handler	= proc_dointvec,
1310 	},
1311 	{
1312 		.procname	= "file-nr",
1313 		.data		= &files_stat,
1314 		.maxlen		= 3*sizeof(int),
1315 		.mode		= 0444,
1316 		.proc_handler	= proc_nr_files,
1317 	},
1318 	{
1319 		.procname	= "file-max",
1320 		.data		= &files_stat.max_files,
1321 		.maxlen		= sizeof(int),
1322 		.mode		= 0644,
1323 		.proc_handler	= proc_dointvec,
1324 	},
1325 	{
1326 		.procname	= "nr_open",
1327 		.data		= &sysctl_nr_open,
1328 		.maxlen		= sizeof(int),
1329 		.mode		= 0644,
1330 		.proc_handler	= proc_dointvec_minmax,
1331 		.extra1		= &sysctl_nr_open_min,
1332 		.extra2		= &sysctl_nr_open_max,
1333 	},
1334 	{
1335 		.procname	= "dentry-state",
1336 		.data		= &dentry_stat,
1337 		.maxlen		= 6*sizeof(int),
1338 		.mode		= 0444,
1339 		.proc_handler	= proc_dointvec,
1340 	},
1341 	{
1342 		.procname	= "overflowuid",
1343 		.data		= &fs_overflowuid,
1344 		.maxlen		= sizeof(int),
1345 		.mode		= 0644,
1346 		.proc_handler	= proc_dointvec_minmax,
1347 		.extra1		= &minolduid,
1348 		.extra2		= &maxolduid,
1349 	},
1350 	{
1351 		.procname	= "overflowgid",
1352 		.data		= &fs_overflowgid,
1353 		.maxlen		= sizeof(int),
1354 		.mode		= 0644,
1355 		.proc_handler	= proc_dointvec_minmax,
1356 		.extra1		= &minolduid,
1357 		.extra2		= &maxolduid,
1358 	},
1359 #ifdef CONFIG_FILE_LOCKING
1360 	{
1361 		.procname	= "leases-enable",
1362 		.data		= &leases_enable,
1363 		.maxlen		= sizeof(int),
1364 		.mode		= 0644,
1365 		.proc_handler	= proc_dointvec,
1366 	},
1367 #endif
1368 #ifdef CONFIG_DNOTIFY
1369 	{
1370 		.procname	= "dir-notify-enable",
1371 		.data		= &dir_notify_enable,
1372 		.maxlen		= sizeof(int),
1373 		.mode		= 0644,
1374 		.proc_handler	= proc_dointvec,
1375 	},
1376 #endif
1377 #ifdef CONFIG_MMU
1378 #ifdef CONFIG_FILE_LOCKING
1379 	{
1380 		.procname	= "lease-break-time",
1381 		.data		= &lease_break_time,
1382 		.maxlen		= sizeof(int),
1383 		.mode		= 0644,
1384 		.proc_handler	= proc_dointvec,
1385 	},
1386 #endif
1387 #ifdef CONFIG_AIO
1388 	{
1389 		.procname	= "aio-nr",
1390 		.data		= &aio_nr,
1391 		.maxlen		= sizeof(aio_nr),
1392 		.mode		= 0444,
1393 		.proc_handler	= proc_doulongvec_minmax,
1394 	},
1395 	{
1396 		.procname	= "aio-max-nr",
1397 		.data		= &aio_max_nr,
1398 		.maxlen		= sizeof(aio_max_nr),
1399 		.mode		= 0644,
1400 		.proc_handler	= proc_doulongvec_minmax,
1401 	},
1402 #endif /* CONFIG_AIO */
1403 #ifdef CONFIG_INOTIFY_USER
1404 	{
1405 		.procname	= "inotify",
1406 		.mode		= 0555,
1407 		.child		= inotify_table,
1408 	},
1409 #endif
1410 #ifdef CONFIG_EPOLL
1411 	{
1412 		.procname	= "epoll",
1413 		.mode		= 0555,
1414 		.child		= epoll_table,
1415 	},
1416 #endif
1417 #endif
1418 	{
1419 		.procname	= "suid_dumpable",
1420 		.data		= &suid_dumpable,
1421 		.maxlen		= sizeof(int),
1422 		.mode		= 0644,
1423 		.proc_handler	= proc_dointvec_minmax,
1424 		.extra1		= &zero,
1425 		.extra2		= &two,
1426 	},
1427 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1428 	{
1429 		.procname	= "binfmt_misc",
1430 		.mode		= 0555,
1431 		.child		= binfmt_misc_table,
1432 	},
1433 #endif
1434 /*
1435  * NOTE: do not add new entries to this table unless you have read
1436  * Documentation/sysctl/ctl_unnumbered.txt
1437  */
1438 	{ }
1439 };
1440 
1441 static struct ctl_table debug_table[] = {
1442 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1443 	{
1444 		.procname	= "exception-trace",
1445 		.data		= &show_unhandled_signals,
1446 		.maxlen		= sizeof(int),
1447 		.mode		= 0644,
1448 		.proc_handler	= proc_dointvec
1449 	},
1450 #endif
1451 	{ }
1452 };
1453 
1454 static struct ctl_table dev_table[] = {
1455 	{ }
1456 };
1457 
1458 static DEFINE_SPINLOCK(sysctl_lock);
1459 
1460 /* called under sysctl_lock */
1461 static int use_table(struct ctl_table_header *p)
1462 {
1463 	if (unlikely(p->unregistering))
1464 		return 0;
1465 	p->used++;
1466 	return 1;
1467 }
1468 
1469 /* called under sysctl_lock */
1470 static void unuse_table(struct ctl_table_header *p)
1471 {
1472 	if (!--p->used)
1473 		if (unlikely(p->unregistering))
1474 			complete(p->unregistering);
1475 }
1476 
1477 /* called under sysctl_lock, will reacquire if has to wait */
1478 static void start_unregistering(struct ctl_table_header *p)
1479 {
1480 	/*
1481 	 * if p->used is 0, nobody will ever touch that entry again;
1482 	 * we'll eliminate all paths to it before dropping sysctl_lock
1483 	 */
1484 	if (unlikely(p->used)) {
1485 		struct completion wait;
1486 		init_completion(&wait);
1487 		p->unregistering = &wait;
1488 		spin_unlock(&sysctl_lock);
1489 		wait_for_completion(&wait);
1490 		spin_lock(&sysctl_lock);
1491 	} else {
1492 		/* anything non-NULL; we'll never dereference it */
1493 		p->unregistering = ERR_PTR(-EINVAL);
1494 	}
1495 	/*
1496 	 * do not remove from the list until nobody holds it; walking the
1497 	 * list in do_sysctl() relies on that.
1498 	 */
1499 	list_del_init(&p->ctl_entry);
1500 }
1501 
1502 void sysctl_head_get(struct ctl_table_header *head)
1503 {
1504 	spin_lock(&sysctl_lock);
1505 	head->count++;
1506 	spin_unlock(&sysctl_lock);
1507 }
1508 
1509 void sysctl_head_put(struct ctl_table_header *head)
1510 {
1511 	spin_lock(&sysctl_lock);
1512 	if (!--head->count)
1513 		kfree(head);
1514 	spin_unlock(&sysctl_lock);
1515 }
1516 
1517 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1518 {
1519 	if (!head)
1520 		BUG();
1521 	spin_lock(&sysctl_lock);
1522 	if (!use_table(head))
1523 		head = ERR_PTR(-ENOENT);
1524 	spin_unlock(&sysctl_lock);
1525 	return head;
1526 }
1527 
1528 void sysctl_head_finish(struct ctl_table_header *head)
1529 {
1530 	if (!head)
1531 		return;
1532 	spin_lock(&sysctl_lock);
1533 	unuse_table(head);
1534 	spin_unlock(&sysctl_lock);
1535 }
1536 
1537 static struct ctl_table_set *
1538 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1539 {
1540 	struct ctl_table_set *set = &root->default_set;
1541 	if (root->lookup)
1542 		set = root->lookup(root, namespaces);
1543 	return set;
1544 }
1545 
1546 static struct list_head *
1547 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1548 {
1549 	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1550 	return &set->list;
1551 }
1552 
1553 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1554 					    struct ctl_table_header *prev)
1555 {
1556 	struct ctl_table_root *root;
1557 	struct list_head *header_list;
1558 	struct ctl_table_header *head;
1559 	struct list_head *tmp;
1560 
1561 	spin_lock(&sysctl_lock);
1562 	if (prev) {
1563 		head = prev;
1564 		tmp = &prev->ctl_entry;
1565 		unuse_table(prev);
1566 		goto next;
1567 	}
1568 	tmp = &root_table_header.ctl_entry;
1569 	for (;;) {
1570 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1571 
1572 		if (!use_table(head))
1573 			goto next;
1574 		spin_unlock(&sysctl_lock);
1575 		return head;
1576 	next:
1577 		root = head->root;
1578 		tmp = tmp->next;
1579 		header_list = lookup_header_list(root, namespaces);
1580 		if (tmp != header_list)
1581 			continue;
1582 
1583 		do {
1584 			root = list_entry(root->root_list.next,
1585 					struct ctl_table_root, root_list);
1586 			if (root == &sysctl_table_root)
1587 				goto out;
1588 			header_list = lookup_header_list(root, namespaces);
1589 		} while (list_empty(header_list));
1590 		tmp = header_list->next;
1591 	}
1592 out:
1593 	spin_unlock(&sysctl_lock);
1594 	return NULL;
1595 }
1596 
1597 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1598 {
1599 	return __sysctl_head_next(current->nsproxy, prev);
1600 }
1601 
1602 void register_sysctl_root(struct ctl_table_root *root)
1603 {
1604 	spin_lock(&sysctl_lock);
1605 	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1606 	spin_unlock(&sysctl_lock);
1607 }
1608 
1609 /*
1610  * sysctl_perm does NOT grant the superuser all rights automatically, because
1611  * some sysctl variables are readonly even to root.
1612  */
1613 
1614 static int test_perm(int mode, int op)
1615 {
1616 	if (!current_euid())
1617 		mode >>= 6;
1618 	else if (in_egroup_p(0))
1619 		mode >>= 3;
1620 	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1621 		return 0;
1622 	return -EACCES;
1623 }
1624 
1625 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1626 {
1627 	int error;
1628 	int mode;
1629 
1630 	error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1631 	if (error)
1632 		return error;
1633 
1634 	if (root->permissions)
1635 		mode = root->permissions(root, current->nsproxy, table);
1636 	else
1637 		mode = table->mode;
1638 
1639 	return test_perm(mode, op);
1640 }
1641 
1642 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1643 {
1644 	for (; table->procname; table++) {
1645 		table->parent = parent;
1646 		if (table->child)
1647 			sysctl_set_parent(table, table->child);
1648 	}
1649 }
1650 
1651 static __init int sysctl_init(void)
1652 {
1653 	sysctl_set_parent(NULL, root_table);
1654 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1655 	{
1656 		int err;
1657 		err = sysctl_check_table(current->nsproxy, root_table);
1658 	}
1659 #endif
1660 	return 0;
1661 }
1662 
1663 core_initcall(sysctl_init);
1664 
1665 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1666 				      struct ctl_table *table)
1667 {
1668 	struct ctl_table *p;
1669 	const char *s = branch->procname;
1670 
1671 	/* branch should have named subdirectory as its first element */
1672 	if (!s || !branch->child)
1673 		return NULL;
1674 
1675 	/* ... and nothing else */
1676 	if (branch[1].procname)
1677 		return NULL;
1678 
1679 	/* table should contain subdirectory with the same name */
1680 	for (p = table; p->procname; p++) {
1681 		if (!p->child)
1682 			continue;
1683 		if (p->procname && strcmp(p->procname, s) == 0)
1684 			return p;
1685 	}
1686 	return NULL;
1687 }
1688 
1689 /* see if attaching q to p would be an improvement */
1690 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1691 {
1692 	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1693 	struct ctl_table *next;
1694 	int is_better = 0;
1695 	int not_in_parent = !p->attached_by;
1696 
1697 	while ((next = is_branch_in(by, to)) != NULL) {
1698 		if (by == q->attached_by)
1699 			is_better = 1;
1700 		if (to == p->attached_by)
1701 			not_in_parent = 1;
1702 		by = by->child;
1703 		to = next->child;
1704 	}
1705 
1706 	if (is_better && not_in_parent) {
1707 		q->attached_by = by;
1708 		q->attached_to = to;
1709 		q->parent = p;
1710 	}
1711 }
1712 
1713 /**
1714  * __register_sysctl_paths - register a sysctl hierarchy
1715  * @root: List of sysctl headers to register on
1716  * @namespaces: Data to compute which lists of sysctl entries are visible
1717  * @path: The path to the directory the sysctl table is in.
1718  * @table: the top-level table structure
1719  *
1720  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1721  * array. A completely 0 filled entry terminates the table.
1722  *
1723  * The members of the &struct ctl_table structure are used as follows:
1724  *
1725  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1726  *            enter a sysctl file
1727  *
1728  * data - a pointer to data for use by proc_handler
1729  *
1730  * maxlen - the maximum size in bytes of the data
1731  *
1732  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1733  *
1734  * child - a pointer to the child sysctl table if this entry is a directory, or
1735  *         %NULL.
1736  *
1737  * proc_handler - the text handler routine (described below)
1738  *
1739  * de - for internal use by the sysctl routines
1740  *
1741  * extra1, extra2 - extra pointers usable by the proc handler routines
1742  *
1743  * Leaf nodes in the sysctl tree will be represented by a single file
1744  * under /proc; non-leaf nodes will be represented by directories.
1745  *
1746  * sysctl(2) can automatically manage read and write requests through
1747  * the sysctl table.  The data and maxlen fields of the ctl_table
1748  * struct enable minimal validation of the values being written to be
1749  * performed, and the mode field allows minimal authentication.
1750  *
1751  * There must be a proc_handler routine for any terminal nodes
1752  * mirrored under /proc/sys (non-terminals are handled by a built-in
1753  * directory handler).  Several default handlers are available to
1754  * cover common cases -
1755  *
1756  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1757  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1758  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1759  *
1760  * It is the handler's job to read the input buffer from user memory
1761  * and process it. The handler should return 0 on success.
1762  *
1763  * This routine returns %NULL on a failure to register, and a pointer
1764  * to the table header on success.
1765  */
1766 struct ctl_table_header *__register_sysctl_paths(
1767 	struct ctl_table_root *root,
1768 	struct nsproxy *namespaces,
1769 	const struct ctl_path *path, struct ctl_table *table)
1770 {
1771 	struct ctl_table_header *header;
1772 	struct ctl_table *new, **prevp;
1773 	unsigned int n, npath;
1774 	struct ctl_table_set *set;
1775 
1776 	/* Count the path components */
1777 	for (npath = 0; path[npath].procname; ++npath)
1778 		;
1779 
1780 	/*
1781 	 * For each path component, allocate a 2-element ctl_table array.
1782 	 * The first array element will be filled with the sysctl entry
1783 	 * for this, the second will be the sentinel (procname == 0).
1784 	 *
1785 	 * We allocate everything in one go so that we don't have to
1786 	 * worry about freeing additional memory in unregister_sysctl_table.
1787 	 */
1788 	header = kzalloc(sizeof(struct ctl_table_header) +
1789 			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1790 	if (!header)
1791 		return NULL;
1792 
1793 	new = (struct ctl_table *) (header + 1);
1794 
1795 	/* Now connect the dots */
1796 	prevp = &header->ctl_table;
1797 	for (n = 0; n < npath; ++n, ++path) {
1798 		/* Copy the procname */
1799 		new->procname = path->procname;
1800 		new->mode     = 0555;
1801 
1802 		*prevp = new;
1803 		prevp = &new->child;
1804 
1805 		new += 2;
1806 	}
1807 	*prevp = table;
1808 	header->ctl_table_arg = table;
1809 
1810 	INIT_LIST_HEAD(&header->ctl_entry);
1811 	header->used = 0;
1812 	header->unregistering = NULL;
1813 	header->root = root;
1814 	sysctl_set_parent(NULL, header->ctl_table);
1815 	header->count = 1;
1816 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1817 	if (sysctl_check_table(namespaces, header->ctl_table)) {
1818 		kfree(header);
1819 		return NULL;
1820 	}
1821 #endif
1822 	spin_lock(&sysctl_lock);
1823 	header->set = lookup_header_set(root, namespaces);
1824 	header->attached_by = header->ctl_table;
1825 	header->attached_to = root_table;
1826 	header->parent = &root_table_header;
1827 	for (set = header->set; set; set = set->parent) {
1828 		struct ctl_table_header *p;
1829 		list_for_each_entry(p, &set->list, ctl_entry) {
1830 			if (p->unregistering)
1831 				continue;
1832 			try_attach(p, header);
1833 		}
1834 	}
1835 	header->parent->count++;
1836 	list_add_tail(&header->ctl_entry, &header->set->list);
1837 	spin_unlock(&sysctl_lock);
1838 
1839 	return header;
1840 }
1841 
1842 /**
1843  * register_sysctl_table_path - register a sysctl table hierarchy
1844  * @path: The path to the directory the sysctl table is in.
1845  * @table: the top-level table structure
1846  *
1847  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1848  * array. A completely 0 filled entry terminates the table.
1849  *
1850  * See __register_sysctl_paths for more details.
1851  */
1852 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1853 						struct ctl_table *table)
1854 {
1855 	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1856 					path, table);
1857 }
1858 
1859 /**
1860  * register_sysctl_table - register a sysctl table hierarchy
1861  * @table: the top-level table structure
1862  *
1863  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1864  * array. A completely 0 filled entry terminates the table.
1865  *
1866  * See register_sysctl_paths for more details.
1867  */
1868 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1869 {
1870 	static const struct ctl_path null_path[] = { {} };
1871 
1872 	return register_sysctl_paths(null_path, table);
1873 }
1874 
1875 /**
1876  * unregister_sysctl_table - unregister a sysctl table hierarchy
1877  * @header: the header returned from register_sysctl_table
1878  *
1879  * Unregisters the sysctl table and all children. proc entries may not
1880  * actually be removed until they are no longer used by anyone.
1881  */
1882 void unregister_sysctl_table(struct ctl_table_header * header)
1883 {
1884 	might_sleep();
1885 
1886 	if (header == NULL)
1887 		return;
1888 
1889 	spin_lock(&sysctl_lock);
1890 	start_unregistering(header);
1891 	if (!--header->parent->count) {
1892 		WARN_ON(1);
1893 		kfree(header->parent);
1894 	}
1895 	if (!--header->count)
1896 		kfree(header);
1897 	spin_unlock(&sysctl_lock);
1898 }
1899 
1900 int sysctl_is_seen(struct ctl_table_header *p)
1901 {
1902 	struct ctl_table_set *set = p->set;
1903 	int res;
1904 	spin_lock(&sysctl_lock);
1905 	if (p->unregistering)
1906 		res = 0;
1907 	else if (!set->is_seen)
1908 		res = 1;
1909 	else
1910 		res = set->is_seen(set);
1911 	spin_unlock(&sysctl_lock);
1912 	return res;
1913 }
1914 
1915 void setup_sysctl_set(struct ctl_table_set *p,
1916 	struct ctl_table_set *parent,
1917 	int (*is_seen)(struct ctl_table_set *))
1918 {
1919 	INIT_LIST_HEAD(&p->list);
1920 	p->parent = parent ? parent : &sysctl_table_root.default_set;
1921 	p->is_seen = is_seen;
1922 }
1923 
1924 #else /* !CONFIG_SYSCTL */
1925 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1926 {
1927 	return NULL;
1928 }
1929 
1930 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1931 						    struct ctl_table *table)
1932 {
1933 	return NULL;
1934 }
1935 
1936 void unregister_sysctl_table(struct ctl_table_header * table)
1937 {
1938 }
1939 
1940 void setup_sysctl_set(struct ctl_table_set *p,
1941 	struct ctl_table_set *parent,
1942 	int (*is_seen)(struct ctl_table_set *))
1943 {
1944 }
1945 
1946 void sysctl_head_put(struct ctl_table_header *head)
1947 {
1948 }
1949 
1950 #endif /* CONFIG_SYSCTL */
1951 
1952 /*
1953  * /proc/sys support
1954  */
1955 
1956 #ifdef CONFIG_PROC_SYSCTL
1957 
1958 static int _proc_do_string(void* data, int maxlen, int write,
1959 			   void __user *buffer,
1960 			   size_t *lenp, loff_t *ppos)
1961 {
1962 	size_t len;
1963 	char __user *p;
1964 	char c;
1965 
1966 	if (!data || !maxlen || !*lenp) {
1967 		*lenp = 0;
1968 		return 0;
1969 	}
1970 
1971 	if (write) {
1972 		len = 0;
1973 		p = buffer;
1974 		while (len < *lenp) {
1975 			if (get_user(c, p++))
1976 				return -EFAULT;
1977 			if (c == 0 || c == '\n')
1978 				break;
1979 			len++;
1980 		}
1981 		if (len >= maxlen)
1982 			len = maxlen-1;
1983 		if(copy_from_user(data, buffer, len))
1984 			return -EFAULT;
1985 		((char *) data)[len] = 0;
1986 		*ppos += *lenp;
1987 	} else {
1988 		len = strlen(data);
1989 		if (len > maxlen)
1990 			len = maxlen;
1991 
1992 		if (*ppos > len) {
1993 			*lenp = 0;
1994 			return 0;
1995 		}
1996 
1997 		data += *ppos;
1998 		len  -= *ppos;
1999 
2000 		if (len > *lenp)
2001 			len = *lenp;
2002 		if (len)
2003 			if(copy_to_user(buffer, data, len))
2004 				return -EFAULT;
2005 		if (len < *lenp) {
2006 			if(put_user('\n', ((char __user *) buffer) + len))
2007 				return -EFAULT;
2008 			len++;
2009 		}
2010 		*lenp = len;
2011 		*ppos += len;
2012 	}
2013 	return 0;
2014 }
2015 
2016 /**
2017  * proc_dostring - read a string sysctl
2018  * @table: the sysctl table
2019  * @write: %TRUE if this is a write to the sysctl file
2020  * @buffer: the user buffer
2021  * @lenp: the size of the user buffer
2022  * @ppos: file position
2023  *
2024  * Reads/writes a string from/to the user buffer. If the kernel
2025  * buffer provided is not large enough to hold the string, the
2026  * string is truncated. The copied string is %NULL-terminated.
2027  * If the string is being read by the user process, it is copied
2028  * and a newline '\n' is added. It is truncated if the buffer is
2029  * not large enough.
2030  *
2031  * Returns 0 on success.
2032  */
2033 int proc_dostring(struct ctl_table *table, int write,
2034 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2035 {
2036 	return _proc_do_string(table->data, table->maxlen, write,
2037 			       buffer, lenp, ppos);
2038 }
2039 
2040 
2041 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2042 				 int *valp,
2043 				 int write, void *data)
2044 {
2045 	if (write) {
2046 		*valp = *negp ? -*lvalp : *lvalp;
2047 	} else {
2048 		int val = *valp;
2049 		if (val < 0) {
2050 			*negp = -1;
2051 			*lvalp = (unsigned long)-val;
2052 		} else {
2053 			*negp = 0;
2054 			*lvalp = (unsigned long)val;
2055 		}
2056 	}
2057 	return 0;
2058 }
2059 
2060 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2061 		  int write, void __user *buffer,
2062 		  size_t *lenp, loff_t *ppos,
2063 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2064 			      int write, void *data),
2065 		  void *data)
2066 {
2067 #define TMPBUFLEN 21
2068 	int *i, vleft, first = 1, neg;
2069 	unsigned long lval;
2070 	size_t left, len;
2071 
2072 	char buf[TMPBUFLEN], *p;
2073 	char __user *s = buffer;
2074 
2075 	if (!tbl_data || !table->maxlen || !*lenp ||
2076 	    (*ppos && !write)) {
2077 		*lenp = 0;
2078 		return 0;
2079 	}
2080 
2081 	i = (int *) tbl_data;
2082 	vleft = table->maxlen / sizeof(*i);
2083 	left = *lenp;
2084 
2085 	if (!conv)
2086 		conv = do_proc_dointvec_conv;
2087 
2088 	for (; left && vleft--; i++, first=0) {
2089 		if (write) {
2090 			while (left) {
2091 				char c;
2092 				if (get_user(c, s))
2093 					return -EFAULT;
2094 				if (!isspace(c))
2095 					break;
2096 				left--;
2097 				s++;
2098 			}
2099 			if (!left)
2100 				break;
2101 			neg = 0;
2102 			len = left;
2103 			if (len > sizeof(buf) - 1)
2104 				len = sizeof(buf) - 1;
2105 			if (copy_from_user(buf, s, len))
2106 				return -EFAULT;
2107 			buf[len] = 0;
2108 			p = buf;
2109 			if (*p == '-' && left > 1) {
2110 				neg = 1;
2111 				p++;
2112 			}
2113 			if (*p < '0' || *p > '9')
2114 				break;
2115 
2116 			lval = simple_strtoul(p, &p, 0);
2117 
2118 			len = p-buf;
2119 			if ((len < left) && *p && !isspace(*p))
2120 				break;
2121 			s += len;
2122 			left -= len;
2123 
2124 			if (conv(&neg, &lval, i, 1, data))
2125 				break;
2126 		} else {
2127 			p = buf;
2128 			if (!first)
2129 				*p++ = '\t';
2130 
2131 			if (conv(&neg, &lval, i, 0, data))
2132 				break;
2133 
2134 			sprintf(p, "%s%lu", neg ? "-" : "", lval);
2135 			len = strlen(buf);
2136 			if (len > left)
2137 				len = left;
2138 			if(copy_to_user(s, buf, len))
2139 				return -EFAULT;
2140 			left -= len;
2141 			s += len;
2142 		}
2143 	}
2144 
2145 	if (!write && !first && left) {
2146 		if(put_user('\n', s))
2147 			return -EFAULT;
2148 		left--, s++;
2149 	}
2150 	if (write) {
2151 		while (left) {
2152 			char c;
2153 			if (get_user(c, s++))
2154 				return -EFAULT;
2155 			if (!isspace(c))
2156 				break;
2157 			left--;
2158 		}
2159 	}
2160 	if (write && first)
2161 		return -EINVAL;
2162 	*lenp -= left;
2163 	*ppos += *lenp;
2164 	return 0;
2165 #undef TMPBUFLEN
2166 }
2167 
2168 static int do_proc_dointvec(struct ctl_table *table, int write,
2169 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2170 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2171 			      int write, void *data),
2172 		  void *data)
2173 {
2174 	return __do_proc_dointvec(table->data, table, write,
2175 			buffer, lenp, ppos, conv, data);
2176 }
2177 
2178 /**
2179  * proc_dointvec - read a vector of integers
2180  * @table: the sysctl table
2181  * @write: %TRUE if this is a write to the sysctl file
2182  * @buffer: the user buffer
2183  * @lenp: the size of the user buffer
2184  * @ppos: file position
2185  *
2186  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2187  * values from/to the user buffer, treated as an ASCII string.
2188  *
2189  * Returns 0 on success.
2190  */
2191 int proc_dointvec(struct ctl_table *table, int write,
2192 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2193 {
2194     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2195 		    	    NULL,NULL);
2196 }
2197 
2198 /*
2199  * Taint values can only be increased
2200  * This means we can safely use a temporary.
2201  */
2202 static int proc_taint(struct ctl_table *table, int write,
2203 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2204 {
2205 	struct ctl_table t;
2206 	unsigned long tmptaint = get_taint();
2207 	int err;
2208 
2209 	if (write && !capable(CAP_SYS_ADMIN))
2210 		return -EPERM;
2211 
2212 	t = *table;
2213 	t.data = &tmptaint;
2214 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2215 	if (err < 0)
2216 		return err;
2217 
2218 	if (write) {
2219 		/*
2220 		 * Poor man's atomic or. Not worth adding a primitive
2221 		 * to everyone's atomic.h for this
2222 		 */
2223 		int i;
2224 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2225 			if ((tmptaint >> i) & 1)
2226 				add_taint(i);
2227 		}
2228 	}
2229 
2230 	return err;
2231 }
2232 
2233 struct do_proc_dointvec_minmax_conv_param {
2234 	int *min;
2235 	int *max;
2236 };
2237 
2238 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2239 					int *valp,
2240 					int write, void *data)
2241 {
2242 	struct do_proc_dointvec_minmax_conv_param *param = data;
2243 	if (write) {
2244 		int val = *negp ? -*lvalp : *lvalp;
2245 		if ((param->min && *param->min > val) ||
2246 		    (param->max && *param->max < val))
2247 			return -EINVAL;
2248 		*valp = val;
2249 	} else {
2250 		int val = *valp;
2251 		if (val < 0) {
2252 			*negp = -1;
2253 			*lvalp = (unsigned long)-val;
2254 		} else {
2255 			*negp = 0;
2256 			*lvalp = (unsigned long)val;
2257 		}
2258 	}
2259 	return 0;
2260 }
2261 
2262 /**
2263  * proc_dointvec_minmax - read a vector of integers with min/max values
2264  * @table: the sysctl table
2265  * @write: %TRUE if this is a write to the sysctl file
2266  * @buffer: the user buffer
2267  * @lenp: the size of the user buffer
2268  * @ppos: file position
2269  *
2270  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2271  * values from/to the user buffer, treated as an ASCII string.
2272  *
2273  * This routine will ensure the values are within the range specified by
2274  * table->extra1 (min) and table->extra2 (max).
2275  *
2276  * Returns 0 on success.
2277  */
2278 int proc_dointvec_minmax(struct ctl_table *table, int write,
2279 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2280 {
2281 	struct do_proc_dointvec_minmax_conv_param param = {
2282 		.min = (int *) table->extra1,
2283 		.max = (int *) table->extra2,
2284 	};
2285 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2286 				do_proc_dointvec_minmax_conv, &param);
2287 }
2288 
2289 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2290 				     void __user *buffer,
2291 				     size_t *lenp, loff_t *ppos,
2292 				     unsigned long convmul,
2293 				     unsigned long convdiv)
2294 {
2295 #define TMPBUFLEN 21
2296 	unsigned long *i, *min, *max, val;
2297 	int vleft, first=1, neg;
2298 	size_t len, left;
2299 	char buf[TMPBUFLEN], *p;
2300 	char __user *s = buffer;
2301 
2302 	if (!data || !table->maxlen || !*lenp ||
2303 	    (*ppos && !write)) {
2304 		*lenp = 0;
2305 		return 0;
2306 	}
2307 
2308 	i = (unsigned long *) data;
2309 	min = (unsigned long *) table->extra1;
2310 	max = (unsigned long *) table->extra2;
2311 	vleft = table->maxlen / sizeof(unsigned long);
2312 	left = *lenp;
2313 
2314 	for (; left && vleft--; i++, min++, max++, first=0) {
2315 		if (write) {
2316 			while (left) {
2317 				char c;
2318 				if (get_user(c, s))
2319 					return -EFAULT;
2320 				if (!isspace(c))
2321 					break;
2322 				left--;
2323 				s++;
2324 			}
2325 			if (!left)
2326 				break;
2327 			neg = 0;
2328 			len = left;
2329 			if (len > TMPBUFLEN-1)
2330 				len = TMPBUFLEN-1;
2331 			if (copy_from_user(buf, s, len))
2332 				return -EFAULT;
2333 			buf[len] = 0;
2334 			p = buf;
2335 			if (*p == '-' && left > 1) {
2336 				neg = 1;
2337 				p++;
2338 			}
2339 			if (*p < '0' || *p > '9')
2340 				break;
2341 			val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2342 			len = p-buf;
2343 			if ((len < left) && *p && !isspace(*p))
2344 				break;
2345 			if (neg)
2346 				val = -val;
2347 			s += len;
2348 			left -= len;
2349 
2350 			if(neg)
2351 				continue;
2352 			if ((min && val < *min) || (max && val > *max))
2353 				continue;
2354 			*i = val;
2355 		} else {
2356 			p = buf;
2357 			if (!first)
2358 				*p++ = '\t';
2359 			sprintf(p, "%lu", convdiv * (*i) / convmul);
2360 			len = strlen(buf);
2361 			if (len > left)
2362 				len = left;
2363 			if(copy_to_user(s, buf, len))
2364 				return -EFAULT;
2365 			left -= len;
2366 			s += len;
2367 		}
2368 	}
2369 
2370 	if (!write && !first && left) {
2371 		if(put_user('\n', s))
2372 			return -EFAULT;
2373 		left--, s++;
2374 	}
2375 	if (write) {
2376 		while (left) {
2377 			char c;
2378 			if (get_user(c, s++))
2379 				return -EFAULT;
2380 			if (!isspace(c))
2381 				break;
2382 			left--;
2383 		}
2384 	}
2385 	if (write && first)
2386 		return -EINVAL;
2387 	*lenp -= left;
2388 	*ppos += *lenp;
2389 	return 0;
2390 #undef TMPBUFLEN
2391 }
2392 
2393 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2394 				     void __user *buffer,
2395 				     size_t *lenp, loff_t *ppos,
2396 				     unsigned long convmul,
2397 				     unsigned long convdiv)
2398 {
2399 	return __do_proc_doulongvec_minmax(table->data, table, write,
2400 			buffer, lenp, ppos, convmul, convdiv);
2401 }
2402 
2403 /**
2404  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2405  * @table: the sysctl table
2406  * @write: %TRUE if this is a write to the sysctl file
2407  * @buffer: the user buffer
2408  * @lenp: the size of the user buffer
2409  * @ppos: file position
2410  *
2411  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2412  * values from/to the user buffer, treated as an ASCII string.
2413  *
2414  * This routine will ensure the values are within the range specified by
2415  * table->extra1 (min) and table->extra2 (max).
2416  *
2417  * Returns 0 on success.
2418  */
2419 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2420 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2421 {
2422     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2423 }
2424 
2425 /**
2426  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2427  * @table: the sysctl table
2428  * @write: %TRUE if this is a write to the sysctl file
2429  * @buffer: the user buffer
2430  * @lenp: the size of the user buffer
2431  * @ppos: file position
2432  *
2433  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2434  * values from/to the user buffer, treated as an ASCII string. The values
2435  * are treated as milliseconds, and converted to jiffies when they are stored.
2436  *
2437  * This routine will ensure the values are within the range specified by
2438  * table->extra1 (min) and table->extra2 (max).
2439  *
2440  * Returns 0 on success.
2441  */
2442 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2443 				      void __user *buffer,
2444 				      size_t *lenp, loff_t *ppos)
2445 {
2446     return do_proc_doulongvec_minmax(table, write, buffer,
2447 				     lenp, ppos, HZ, 1000l);
2448 }
2449 
2450 
2451 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2452 					 int *valp,
2453 					 int write, void *data)
2454 {
2455 	if (write) {
2456 		if (*lvalp > LONG_MAX / HZ)
2457 			return 1;
2458 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2459 	} else {
2460 		int val = *valp;
2461 		unsigned long lval;
2462 		if (val < 0) {
2463 			*negp = -1;
2464 			lval = (unsigned long)-val;
2465 		} else {
2466 			*negp = 0;
2467 			lval = (unsigned long)val;
2468 		}
2469 		*lvalp = lval / HZ;
2470 	}
2471 	return 0;
2472 }
2473 
2474 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2475 						int *valp,
2476 						int write, void *data)
2477 {
2478 	if (write) {
2479 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2480 			return 1;
2481 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2482 	} else {
2483 		int val = *valp;
2484 		unsigned long lval;
2485 		if (val < 0) {
2486 			*negp = -1;
2487 			lval = (unsigned long)-val;
2488 		} else {
2489 			*negp = 0;
2490 			lval = (unsigned long)val;
2491 		}
2492 		*lvalp = jiffies_to_clock_t(lval);
2493 	}
2494 	return 0;
2495 }
2496 
2497 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2498 					    int *valp,
2499 					    int write, void *data)
2500 {
2501 	if (write) {
2502 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2503 	} else {
2504 		int val = *valp;
2505 		unsigned long lval;
2506 		if (val < 0) {
2507 			*negp = -1;
2508 			lval = (unsigned long)-val;
2509 		} else {
2510 			*negp = 0;
2511 			lval = (unsigned long)val;
2512 		}
2513 		*lvalp = jiffies_to_msecs(lval);
2514 	}
2515 	return 0;
2516 }
2517 
2518 /**
2519  * proc_dointvec_jiffies - read a vector of integers as seconds
2520  * @table: the sysctl table
2521  * @write: %TRUE if this is a write to the sysctl file
2522  * @buffer: the user buffer
2523  * @lenp: the size of the user buffer
2524  * @ppos: file position
2525  *
2526  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2527  * values from/to the user buffer, treated as an ASCII string.
2528  * The values read are assumed to be in seconds, and are converted into
2529  * jiffies.
2530  *
2531  * Returns 0 on success.
2532  */
2533 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2534 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2535 {
2536     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2537 		    	    do_proc_dointvec_jiffies_conv,NULL);
2538 }
2539 
2540 /**
2541  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2542  * @table: the sysctl table
2543  * @write: %TRUE if this is a write to the sysctl file
2544  * @buffer: the user buffer
2545  * @lenp: the size of the user buffer
2546  * @ppos: pointer to the file position
2547  *
2548  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2549  * values from/to the user buffer, treated as an ASCII string.
2550  * The values read are assumed to be in 1/USER_HZ seconds, and
2551  * are converted into jiffies.
2552  *
2553  * Returns 0 on success.
2554  */
2555 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2556 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2557 {
2558     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2559 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2560 }
2561 
2562 /**
2563  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2564  * @table: the sysctl table
2565  * @write: %TRUE if this is a write to the sysctl file
2566  * @buffer: the user buffer
2567  * @lenp: the size of the user buffer
2568  * @ppos: file position
2569  * @ppos: the current position in the file
2570  *
2571  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2572  * values from/to the user buffer, treated as an ASCII string.
2573  * The values read are assumed to be in 1/1000 seconds, and
2574  * are converted into jiffies.
2575  *
2576  * Returns 0 on success.
2577  */
2578 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2579 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2580 {
2581 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2582 				do_proc_dointvec_ms_jiffies_conv, NULL);
2583 }
2584 
2585 static int proc_do_cad_pid(struct ctl_table *table, int write,
2586 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2587 {
2588 	struct pid *new_pid;
2589 	pid_t tmp;
2590 	int r;
2591 
2592 	tmp = pid_vnr(cad_pid);
2593 
2594 	r = __do_proc_dointvec(&tmp, table, write, buffer,
2595 			       lenp, ppos, NULL, NULL);
2596 	if (r || !write)
2597 		return r;
2598 
2599 	new_pid = find_get_pid(tmp);
2600 	if (!new_pid)
2601 		return -ESRCH;
2602 
2603 	put_pid(xchg(&cad_pid, new_pid));
2604 	return 0;
2605 }
2606 
2607 #else /* CONFIG_PROC_FS */
2608 
2609 int proc_dostring(struct ctl_table *table, int write,
2610 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2611 {
2612 	return -ENOSYS;
2613 }
2614 
2615 int proc_dointvec(struct ctl_table *table, int write,
2616 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2617 {
2618 	return -ENOSYS;
2619 }
2620 
2621 int proc_dointvec_minmax(struct ctl_table *table, int write,
2622 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2623 {
2624 	return -ENOSYS;
2625 }
2626 
2627 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2628 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2629 {
2630 	return -ENOSYS;
2631 }
2632 
2633 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2634 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2635 {
2636 	return -ENOSYS;
2637 }
2638 
2639 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2640 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2641 {
2642 	return -ENOSYS;
2643 }
2644 
2645 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2646 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2647 {
2648 	return -ENOSYS;
2649 }
2650 
2651 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2652 				      void __user *buffer,
2653 				      size_t *lenp, loff_t *ppos)
2654 {
2655     return -ENOSYS;
2656 }
2657 
2658 
2659 #endif /* CONFIG_PROC_FS */
2660 
2661 /*
2662  * No sense putting this after each symbol definition, twice,
2663  * exception granted :-)
2664  */
2665 EXPORT_SYMBOL(proc_dointvec);
2666 EXPORT_SYMBOL(proc_dointvec_jiffies);
2667 EXPORT_SYMBOL(proc_dointvec_minmax);
2668 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2669 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2670 EXPORT_SYMBOL(proc_dostring);
2671 EXPORT_SYMBOL(proc_doulongvec_minmax);
2672 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2673 EXPORT_SYMBOL(register_sysctl_table);
2674 EXPORT_SYMBOL(register_sysctl_paths);
2675 EXPORT_SYMBOL(unregister_sysctl_table);
2676