xref: /linux/kernel/sysctl.c (revision d89dffa976bcd13fd87eb76e02e3b71c3a7868e3)
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/bitmap.h>
27 #include <linux/signal.h>
28 #include <linux/printk.h>
29 #include <linux/proc_fs.h>
30 #include <linux/security.h>
31 #include <linux/ctype.h>
32 #include <linux/kmemcheck.h>
33 #include <linux/kmemleak.h>
34 #include <linux/fs.h>
35 #include <linux/init.h>
36 #include <linux/kernel.h>
37 #include <linux/kobject.h>
38 #include <linux/net.h>
39 #include <linux/sysrq.h>
40 #include <linux/highuid.h>
41 #include <linux/writeback.h>
42 #include <linux/ratelimit.h>
43 #include <linux/compaction.h>
44 #include <linux/hugetlb.h>
45 #include <linux/initrd.h>
46 #include <linux/key.h>
47 #include <linux/times.h>
48 #include <linux/limits.h>
49 #include <linux/dcache.h>
50 #include <linux/dnotify.h>
51 #include <linux/syscalls.h>
52 #include <linux/vmstat.h>
53 #include <linux/nfs_fs.h>
54 #include <linux/acpi.h>
55 #include <linux/reboot.h>
56 #include <linux/ftrace.h>
57 #include <linux/perf_event.h>
58 #include <linux/kprobes.h>
59 #include <linux/pipe_fs_i.h>
60 #include <linux/oom.h>
61 #include <linux/kmod.h>
62 #include <linux/capability.h>
63 #include <linux/binfmts.h>
64 
65 #include <asm/uaccess.h>
66 #include <asm/processor.h>
67 
68 #ifdef CONFIG_X86
69 #include <asm/nmi.h>
70 #include <asm/stacktrace.h>
71 #include <asm/io.h>
72 #endif
73 #ifdef CONFIG_SPARC
74 #include <asm/setup.h>
75 #endif
76 #ifdef CONFIG_BSD_PROCESS_ACCT
77 #include <linux/acct.h>
78 #endif
79 #ifdef CONFIG_RT_MUTEXES
80 #include <linux/rtmutex.h>
81 #endif
82 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
83 #include <linux/lockdep.h>
84 #endif
85 #ifdef CONFIG_CHR_DEV_SG
86 #include <scsi/sg.h>
87 #endif
88 
89 #ifdef CONFIG_LOCKUP_DETECTOR
90 #include <linux/nmi.h>
91 #endif
92 
93 
94 #if defined(CONFIG_SYSCTL)
95 
96 /* External variables not in a header file. */
97 extern int sysctl_overcommit_memory;
98 extern int sysctl_overcommit_ratio;
99 extern int max_threads;
100 extern int core_uses_pid;
101 extern int suid_dumpable;
102 extern char core_pattern[];
103 extern unsigned int core_pipe_limit;
104 extern int pid_max;
105 extern int min_free_kbytes;
106 extern int pid_max_min, pid_max_max;
107 extern int sysctl_drop_caches;
108 extern int percpu_pagelist_fraction;
109 extern int compat_log;
110 extern int latencytop_enabled;
111 extern int sysctl_nr_open_min, sysctl_nr_open_max;
112 #ifndef CONFIG_MMU
113 extern int sysctl_nr_trim_pages;
114 #endif
115 #ifdef CONFIG_BLOCK
116 extern int blk_iopoll_enabled;
117 #endif
118 
119 /* Constants used for minimum and  maximum */
120 #ifdef CONFIG_LOCKUP_DETECTOR
121 static int sixty = 60;
122 static int neg_one = -1;
123 #endif
124 
125 static int zero;
126 static int __maybe_unused one = 1;
127 static int __maybe_unused two = 2;
128 static int __maybe_unused three = 3;
129 static unsigned long one_ul = 1;
130 static int one_hundred = 100;
131 #ifdef CONFIG_PRINTK
132 static int ten_thousand = 10000;
133 #endif
134 
135 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
136 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
137 
138 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
139 static int maxolduid = 65535;
140 static int minolduid;
141 static int min_percpu_pagelist_fract = 8;
142 
143 static int ngroups_max = NGROUPS_MAX;
144 static const int cap_last_cap = CAP_LAST_CAP;
145 
146 #ifdef CONFIG_INOTIFY_USER
147 #include <linux/inotify.h>
148 #endif
149 #ifdef CONFIG_SPARC
150 #endif
151 
152 #ifdef CONFIG_SPARC64
153 extern int sysctl_tsb_ratio;
154 #endif
155 
156 #ifdef __hppa__
157 extern int pwrsw_enabled;
158 extern int unaligned_enabled;
159 #endif
160 
161 #ifdef CONFIG_IA64
162 extern int no_unaligned_warning;
163 extern int unaligned_dump_stack;
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 #ifdef CONFIG_PRINTK
174 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
175 				void __user *buffer, size_t *lenp, loff_t *ppos);
176 #endif
177 
178 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
179 		void __user *buffer, size_t *lenp, loff_t *ppos);
180 static int proc_dostring_coredump(struct ctl_table *table, int write,
181 		void __user *buffer, size_t *lenp, loff_t *ppos);
182 
183 #ifdef CONFIG_MAGIC_SYSRQ
184 /* Note: sysrq code uses it's own private copy */
185 static int __sysrq_enabled = SYSRQ_DEFAULT_ENABLE;
186 
187 static int sysrq_sysctl_handler(ctl_table *table, int write,
188 				void __user *buffer, size_t *lenp,
189 				loff_t *ppos)
190 {
191 	int error;
192 
193 	error = proc_dointvec(table, write, buffer, lenp, ppos);
194 	if (error)
195 		return error;
196 
197 	if (write)
198 		sysrq_toggle_support(__sysrq_enabled);
199 
200 	return 0;
201 }
202 
203 #endif
204 
205 static struct ctl_table kern_table[];
206 static struct ctl_table vm_table[];
207 static struct ctl_table fs_table[];
208 static struct ctl_table debug_table[];
209 static struct ctl_table dev_table[];
210 extern struct ctl_table random_table[];
211 #ifdef CONFIG_EPOLL
212 extern struct ctl_table epoll_table[];
213 #endif
214 
215 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
216 int sysctl_legacy_va_layout;
217 #endif
218 
219 /* The default sysctl tables: */
220 
221 static struct ctl_table sysctl_base_table[] = {
222 	{
223 		.procname	= "kernel",
224 		.mode		= 0555,
225 		.child		= kern_table,
226 	},
227 	{
228 		.procname	= "vm",
229 		.mode		= 0555,
230 		.child		= vm_table,
231 	},
232 	{
233 		.procname	= "fs",
234 		.mode		= 0555,
235 		.child		= fs_table,
236 	},
237 	{
238 		.procname	= "debug",
239 		.mode		= 0555,
240 		.child		= debug_table,
241 	},
242 	{
243 		.procname	= "dev",
244 		.mode		= 0555,
245 		.child		= dev_table,
246 	},
247 	{ }
248 };
249 
250 #ifdef CONFIG_SCHED_DEBUG
251 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
252 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
253 static int min_wakeup_granularity_ns;			/* 0 usecs */
254 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
255 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
256 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
257 #endif
258 
259 #ifdef CONFIG_COMPACTION
260 static int min_extfrag_threshold;
261 static int max_extfrag_threshold = 1000;
262 #endif
263 
264 static struct ctl_table kern_table[] = {
265 	{
266 		.procname	= "sched_child_runs_first",
267 		.data		= &sysctl_sched_child_runs_first,
268 		.maxlen		= sizeof(unsigned int),
269 		.mode		= 0644,
270 		.proc_handler	= proc_dointvec,
271 	},
272 #ifdef CONFIG_SCHED_DEBUG
273 	{
274 		.procname	= "sched_min_granularity_ns",
275 		.data		= &sysctl_sched_min_granularity,
276 		.maxlen		= sizeof(unsigned int),
277 		.mode		= 0644,
278 		.proc_handler	= sched_proc_update_handler,
279 		.extra1		= &min_sched_granularity_ns,
280 		.extra2		= &max_sched_granularity_ns,
281 	},
282 	{
283 		.procname	= "sched_latency_ns",
284 		.data		= &sysctl_sched_latency,
285 		.maxlen		= sizeof(unsigned int),
286 		.mode		= 0644,
287 		.proc_handler	= sched_proc_update_handler,
288 		.extra1		= &min_sched_granularity_ns,
289 		.extra2		= &max_sched_granularity_ns,
290 	},
291 	{
292 		.procname	= "sched_wakeup_granularity_ns",
293 		.data		= &sysctl_sched_wakeup_granularity,
294 		.maxlen		= sizeof(unsigned int),
295 		.mode		= 0644,
296 		.proc_handler	= sched_proc_update_handler,
297 		.extra1		= &min_wakeup_granularity_ns,
298 		.extra2		= &max_wakeup_granularity_ns,
299 	},
300 	{
301 		.procname	= "sched_tunable_scaling",
302 		.data		= &sysctl_sched_tunable_scaling,
303 		.maxlen		= sizeof(enum sched_tunable_scaling),
304 		.mode		= 0644,
305 		.proc_handler	= sched_proc_update_handler,
306 		.extra1		= &min_sched_tunable_scaling,
307 		.extra2		= &max_sched_tunable_scaling,
308 	},
309 	{
310 		.procname	= "sched_migration_cost",
311 		.data		= &sysctl_sched_migration_cost,
312 		.maxlen		= sizeof(unsigned int),
313 		.mode		= 0644,
314 		.proc_handler	= proc_dointvec,
315 	},
316 	{
317 		.procname	= "sched_nr_migrate",
318 		.data		= &sysctl_sched_nr_migrate,
319 		.maxlen		= sizeof(unsigned int),
320 		.mode		= 0644,
321 		.proc_handler	= proc_dointvec,
322 	},
323 	{
324 		.procname	= "sched_time_avg",
325 		.data		= &sysctl_sched_time_avg,
326 		.maxlen		= sizeof(unsigned int),
327 		.mode		= 0644,
328 		.proc_handler	= proc_dointvec,
329 	},
330 	{
331 		.procname	= "sched_shares_window",
332 		.data		= &sysctl_sched_shares_window,
333 		.maxlen		= sizeof(unsigned int),
334 		.mode		= 0644,
335 		.proc_handler	= proc_dointvec,
336 	},
337 	{
338 		.procname	= "timer_migration",
339 		.data		= &sysctl_timer_migration,
340 		.maxlen		= sizeof(unsigned int),
341 		.mode		= 0644,
342 		.proc_handler	= proc_dointvec_minmax,
343 		.extra1		= &zero,
344 		.extra2		= &one,
345 	},
346 #endif
347 	{
348 		.procname	= "sched_rt_period_us",
349 		.data		= &sysctl_sched_rt_period,
350 		.maxlen		= sizeof(unsigned int),
351 		.mode		= 0644,
352 		.proc_handler	= sched_rt_handler,
353 	},
354 	{
355 		.procname	= "sched_rt_runtime_us",
356 		.data		= &sysctl_sched_rt_runtime,
357 		.maxlen		= sizeof(int),
358 		.mode		= 0644,
359 		.proc_handler	= sched_rt_handler,
360 	},
361 #ifdef CONFIG_SCHED_AUTOGROUP
362 	{
363 		.procname	= "sched_autogroup_enabled",
364 		.data		= &sysctl_sched_autogroup_enabled,
365 		.maxlen		= sizeof(unsigned int),
366 		.mode		= 0644,
367 		.proc_handler	= proc_dointvec_minmax,
368 		.extra1		= &zero,
369 		.extra2		= &one,
370 	},
371 #endif
372 #ifdef CONFIG_CFS_BANDWIDTH
373 	{
374 		.procname	= "sched_cfs_bandwidth_slice_us",
375 		.data		= &sysctl_sched_cfs_bandwidth_slice,
376 		.maxlen		= sizeof(unsigned int),
377 		.mode		= 0644,
378 		.proc_handler	= proc_dointvec_minmax,
379 		.extra1		= &one,
380 	},
381 #endif
382 #ifdef CONFIG_PROVE_LOCKING
383 	{
384 		.procname	= "prove_locking",
385 		.data		= &prove_locking,
386 		.maxlen		= sizeof(int),
387 		.mode		= 0644,
388 		.proc_handler	= proc_dointvec,
389 	},
390 #endif
391 #ifdef CONFIG_LOCK_STAT
392 	{
393 		.procname	= "lock_stat",
394 		.data		= &lock_stat,
395 		.maxlen		= sizeof(int),
396 		.mode		= 0644,
397 		.proc_handler	= proc_dointvec,
398 	},
399 #endif
400 	{
401 		.procname	= "panic",
402 		.data		= &panic_timeout,
403 		.maxlen		= sizeof(int),
404 		.mode		= 0644,
405 		.proc_handler	= proc_dointvec,
406 	},
407 	{
408 		.procname	= "core_uses_pid",
409 		.data		= &core_uses_pid,
410 		.maxlen		= sizeof(int),
411 		.mode		= 0644,
412 		.proc_handler	= proc_dointvec,
413 	},
414 	{
415 		.procname	= "core_pattern",
416 		.data		= core_pattern,
417 		.maxlen		= CORENAME_MAX_SIZE,
418 		.mode		= 0644,
419 		.proc_handler	= proc_dostring_coredump,
420 	},
421 	{
422 		.procname	= "core_pipe_limit",
423 		.data		= &core_pipe_limit,
424 		.maxlen		= sizeof(unsigned int),
425 		.mode		= 0644,
426 		.proc_handler	= proc_dointvec,
427 	},
428 #ifdef CONFIG_PROC_SYSCTL
429 	{
430 		.procname	= "tainted",
431 		.maxlen 	= sizeof(long),
432 		.mode		= 0644,
433 		.proc_handler	= proc_taint,
434 	},
435 #endif
436 #ifdef CONFIG_LATENCYTOP
437 	{
438 		.procname	= "latencytop",
439 		.data		= &latencytop_enabled,
440 		.maxlen		= sizeof(int),
441 		.mode		= 0644,
442 		.proc_handler	= proc_dointvec,
443 	},
444 #endif
445 #ifdef CONFIG_BLK_DEV_INITRD
446 	{
447 		.procname	= "real-root-dev",
448 		.data		= &real_root_dev,
449 		.maxlen		= sizeof(int),
450 		.mode		= 0644,
451 		.proc_handler	= proc_dointvec,
452 	},
453 #endif
454 	{
455 		.procname	= "print-fatal-signals",
456 		.data		= &print_fatal_signals,
457 		.maxlen		= sizeof(int),
458 		.mode		= 0644,
459 		.proc_handler	= proc_dointvec,
460 	},
461 #ifdef CONFIG_SPARC
462 	{
463 		.procname	= "reboot-cmd",
464 		.data		= reboot_command,
465 		.maxlen		= 256,
466 		.mode		= 0644,
467 		.proc_handler	= proc_dostring,
468 	},
469 	{
470 		.procname	= "stop-a",
471 		.data		= &stop_a_enabled,
472 		.maxlen		= sizeof (int),
473 		.mode		= 0644,
474 		.proc_handler	= proc_dointvec,
475 	},
476 	{
477 		.procname	= "scons-poweroff",
478 		.data		= &scons_pwroff,
479 		.maxlen		= sizeof (int),
480 		.mode		= 0644,
481 		.proc_handler	= proc_dointvec,
482 	},
483 #endif
484 #ifdef CONFIG_SPARC64
485 	{
486 		.procname	= "tsb-ratio",
487 		.data		= &sysctl_tsb_ratio,
488 		.maxlen		= sizeof (int),
489 		.mode		= 0644,
490 		.proc_handler	= proc_dointvec,
491 	},
492 #endif
493 #ifdef __hppa__
494 	{
495 		.procname	= "soft-power",
496 		.data		= &pwrsw_enabled,
497 		.maxlen		= sizeof (int),
498 	 	.mode		= 0644,
499 		.proc_handler	= proc_dointvec,
500 	},
501 	{
502 		.procname	= "unaligned-trap",
503 		.data		= &unaligned_enabled,
504 		.maxlen		= sizeof (int),
505 		.mode		= 0644,
506 		.proc_handler	= proc_dointvec,
507 	},
508 #endif
509 	{
510 		.procname	= "ctrl-alt-del",
511 		.data		= &C_A_D,
512 		.maxlen		= sizeof(int),
513 		.mode		= 0644,
514 		.proc_handler	= proc_dointvec,
515 	},
516 #ifdef CONFIG_FUNCTION_TRACER
517 	{
518 		.procname	= "ftrace_enabled",
519 		.data		= &ftrace_enabled,
520 		.maxlen		= sizeof(int),
521 		.mode		= 0644,
522 		.proc_handler	= ftrace_enable_sysctl,
523 	},
524 #endif
525 #ifdef CONFIG_STACK_TRACER
526 	{
527 		.procname	= "stack_tracer_enabled",
528 		.data		= &stack_tracer_enabled,
529 		.maxlen		= sizeof(int),
530 		.mode		= 0644,
531 		.proc_handler	= stack_trace_sysctl,
532 	},
533 #endif
534 #ifdef CONFIG_TRACING
535 	{
536 		.procname	= "ftrace_dump_on_oops",
537 		.data		= &ftrace_dump_on_oops,
538 		.maxlen		= sizeof(int),
539 		.mode		= 0644,
540 		.proc_handler	= proc_dointvec,
541 	},
542 #endif
543 #ifdef CONFIG_MODULES
544 	{
545 		.procname	= "modprobe",
546 		.data		= &modprobe_path,
547 		.maxlen		= KMOD_PATH_LEN,
548 		.mode		= 0644,
549 		.proc_handler	= proc_dostring,
550 	},
551 	{
552 		.procname	= "modules_disabled",
553 		.data		= &modules_disabled,
554 		.maxlen		= sizeof(int),
555 		.mode		= 0644,
556 		/* only handle a transition from default "0" to "1" */
557 		.proc_handler	= proc_dointvec_minmax,
558 		.extra1		= &one,
559 		.extra2		= &one,
560 	},
561 #endif
562 #ifdef CONFIG_HOTPLUG
563 	{
564 		.procname	= "hotplug",
565 		.data		= &uevent_helper,
566 		.maxlen		= UEVENT_HELPER_PATH_LEN,
567 		.mode		= 0644,
568 		.proc_handler	= proc_dostring,
569 	},
570 #endif
571 #ifdef CONFIG_CHR_DEV_SG
572 	{
573 		.procname	= "sg-big-buff",
574 		.data		= &sg_big_buff,
575 		.maxlen		= sizeof (int),
576 		.mode		= 0444,
577 		.proc_handler	= proc_dointvec,
578 	},
579 #endif
580 #ifdef CONFIG_BSD_PROCESS_ACCT
581 	{
582 		.procname	= "acct",
583 		.data		= &acct_parm,
584 		.maxlen		= 3*sizeof(int),
585 		.mode		= 0644,
586 		.proc_handler	= proc_dointvec,
587 	},
588 #endif
589 #ifdef CONFIG_MAGIC_SYSRQ
590 	{
591 		.procname	= "sysrq",
592 		.data		= &__sysrq_enabled,
593 		.maxlen		= sizeof (int),
594 		.mode		= 0644,
595 		.proc_handler	= sysrq_sysctl_handler,
596 	},
597 #endif
598 #ifdef CONFIG_PROC_SYSCTL
599 	{
600 		.procname	= "cad_pid",
601 		.data		= NULL,
602 		.maxlen		= sizeof (int),
603 		.mode		= 0600,
604 		.proc_handler	= proc_do_cad_pid,
605 	},
606 #endif
607 	{
608 		.procname	= "threads-max",
609 		.data		= &max_threads,
610 		.maxlen		= sizeof(int),
611 		.mode		= 0644,
612 		.proc_handler	= proc_dointvec,
613 	},
614 	{
615 		.procname	= "random",
616 		.mode		= 0555,
617 		.child		= random_table,
618 	},
619 	{
620 		.procname	= "usermodehelper",
621 		.mode		= 0555,
622 		.child		= usermodehelper_table,
623 	},
624 	{
625 		.procname	= "overflowuid",
626 		.data		= &overflowuid,
627 		.maxlen		= sizeof(int),
628 		.mode		= 0644,
629 		.proc_handler	= proc_dointvec_minmax,
630 		.extra1		= &minolduid,
631 		.extra2		= &maxolduid,
632 	},
633 	{
634 		.procname	= "overflowgid",
635 		.data		= &overflowgid,
636 		.maxlen		= sizeof(int),
637 		.mode		= 0644,
638 		.proc_handler	= proc_dointvec_minmax,
639 		.extra1		= &minolduid,
640 		.extra2		= &maxolduid,
641 	},
642 #ifdef CONFIG_S390
643 #ifdef CONFIG_MATHEMU
644 	{
645 		.procname	= "ieee_emulation_warnings",
646 		.data		= &sysctl_ieee_emulation_warnings,
647 		.maxlen		= sizeof(int),
648 		.mode		= 0644,
649 		.proc_handler	= proc_dointvec,
650 	},
651 #endif
652 	{
653 		.procname	= "userprocess_debug",
654 		.data		= &show_unhandled_signals,
655 		.maxlen		= sizeof(int),
656 		.mode		= 0644,
657 		.proc_handler	= proc_dointvec,
658 	},
659 #endif
660 	{
661 		.procname	= "pid_max",
662 		.data		= &pid_max,
663 		.maxlen		= sizeof (int),
664 		.mode		= 0644,
665 		.proc_handler	= proc_dointvec_minmax,
666 		.extra1		= &pid_max_min,
667 		.extra2		= &pid_max_max,
668 	},
669 	{
670 		.procname	= "panic_on_oops",
671 		.data		= &panic_on_oops,
672 		.maxlen		= sizeof(int),
673 		.mode		= 0644,
674 		.proc_handler	= proc_dointvec,
675 	},
676 #if defined CONFIG_PRINTK
677 	{
678 		.procname	= "printk",
679 		.data		= &console_loglevel,
680 		.maxlen		= 4*sizeof(int),
681 		.mode		= 0644,
682 		.proc_handler	= proc_dointvec,
683 	},
684 	{
685 		.procname	= "printk_ratelimit",
686 		.data		= &printk_ratelimit_state.interval,
687 		.maxlen		= sizeof(int),
688 		.mode		= 0644,
689 		.proc_handler	= proc_dointvec_jiffies,
690 	},
691 	{
692 		.procname	= "printk_ratelimit_burst",
693 		.data		= &printk_ratelimit_state.burst,
694 		.maxlen		= sizeof(int),
695 		.mode		= 0644,
696 		.proc_handler	= proc_dointvec,
697 	},
698 	{
699 		.procname	= "printk_delay",
700 		.data		= &printk_delay_msec,
701 		.maxlen		= sizeof(int),
702 		.mode		= 0644,
703 		.proc_handler	= proc_dointvec_minmax,
704 		.extra1		= &zero,
705 		.extra2		= &ten_thousand,
706 	},
707 	{
708 		.procname	= "dmesg_restrict",
709 		.data		= &dmesg_restrict,
710 		.maxlen		= sizeof(int),
711 		.mode		= 0644,
712 		.proc_handler	= proc_dointvec_minmax_sysadmin,
713 		.extra1		= &zero,
714 		.extra2		= &one,
715 	},
716 	{
717 		.procname	= "kptr_restrict",
718 		.data		= &kptr_restrict,
719 		.maxlen		= sizeof(int),
720 		.mode		= 0644,
721 		.proc_handler	= proc_dointvec_minmax_sysadmin,
722 		.extra1		= &zero,
723 		.extra2		= &two,
724 	},
725 #endif
726 	{
727 		.procname	= "ngroups_max",
728 		.data		= &ngroups_max,
729 		.maxlen		= sizeof (int),
730 		.mode		= 0444,
731 		.proc_handler	= proc_dointvec,
732 	},
733 	{
734 		.procname	= "cap_last_cap",
735 		.data		= (void *)&cap_last_cap,
736 		.maxlen		= sizeof(int),
737 		.mode		= 0444,
738 		.proc_handler	= proc_dointvec,
739 	},
740 #if defined(CONFIG_LOCKUP_DETECTOR)
741 	{
742 		.procname       = "watchdog",
743 		.data           = &watchdog_enabled,
744 		.maxlen         = sizeof (int),
745 		.mode           = 0644,
746 		.proc_handler   = proc_dowatchdog,
747 		.extra1		= &zero,
748 		.extra2		= &one,
749 	},
750 	{
751 		.procname	= "watchdog_thresh",
752 		.data		= &watchdog_thresh,
753 		.maxlen		= sizeof(int),
754 		.mode		= 0644,
755 		.proc_handler	= proc_dowatchdog,
756 		.extra1		= &neg_one,
757 		.extra2		= &sixty,
758 	},
759 	{
760 		.procname	= "softlockup_panic",
761 		.data		= &softlockup_panic,
762 		.maxlen		= sizeof(int),
763 		.mode		= 0644,
764 		.proc_handler	= proc_dointvec_minmax,
765 		.extra1		= &zero,
766 		.extra2		= &one,
767 	},
768 	{
769 		.procname       = "nmi_watchdog",
770 		.data           = &watchdog_enabled,
771 		.maxlen         = sizeof (int),
772 		.mode           = 0644,
773 		.proc_handler   = proc_dowatchdog,
774 		.extra1		= &zero,
775 		.extra2		= &one,
776 	},
777 #endif
778 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
779 	{
780 		.procname       = "unknown_nmi_panic",
781 		.data           = &unknown_nmi_panic,
782 		.maxlen         = sizeof (int),
783 		.mode           = 0644,
784 		.proc_handler   = proc_dointvec,
785 	},
786 #endif
787 #if defined(CONFIG_X86)
788 	{
789 		.procname	= "panic_on_unrecovered_nmi",
790 		.data		= &panic_on_unrecovered_nmi,
791 		.maxlen		= sizeof(int),
792 		.mode		= 0644,
793 		.proc_handler	= proc_dointvec,
794 	},
795 	{
796 		.procname	= "panic_on_io_nmi",
797 		.data		= &panic_on_io_nmi,
798 		.maxlen		= sizeof(int),
799 		.mode		= 0644,
800 		.proc_handler	= proc_dointvec,
801 	},
802 #ifdef CONFIG_DEBUG_STACKOVERFLOW
803 	{
804 		.procname	= "panic_on_stackoverflow",
805 		.data		= &sysctl_panic_on_stackoverflow,
806 		.maxlen		= sizeof(int),
807 		.mode		= 0644,
808 		.proc_handler	= proc_dointvec,
809 	},
810 #endif
811 	{
812 		.procname	= "bootloader_type",
813 		.data		= &bootloader_type,
814 		.maxlen		= sizeof (int),
815 		.mode		= 0444,
816 		.proc_handler	= proc_dointvec,
817 	},
818 	{
819 		.procname	= "bootloader_version",
820 		.data		= &bootloader_version,
821 		.maxlen		= sizeof (int),
822 		.mode		= 0444,
823 		.proc_handler	= proc_dointvec,
824 	},
825 	{
826 		.procname	= "kstack_depth_to_print",
827 		.data		= &kstack_depth_to_print,
828 		.maxlen		= sizeof(int),
829 		.mode		= 0644,
830 		.proc_handler	= proc_dointvec,
831 	},
832 	{
833 		.procname	= "io_delay_type",
834 		.data		= &io_delay_type,
835 		.maxlen		= sizeof(int),
836 		.mode		= 0644,
837 		.proc_handler	= proc_dointvec,
838 	},
839 #endif
840 #if defined(CONFIG_MMU)
841 	{
842 		.procname	= "randomize_va_space",
843 		.data		= &randomize_va_space,
844 		.maxlen		= sizeof(int),
845 		.mode		= 0644,
846 		.proc_handler	= proc_dointvec,
847 	},
848 #endif
849 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
850 	{
851 		.procname	= "spin_retry",
852 		.data		= &spin_retry,
853 		.maxlen		= sizeof (int),
854 		.mode		= 0644,
855 		.proc_handler	= proc_dointvec,
856 	},
857 #endif
858 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
859 	{
860 		.procname	= "acpi_video_flags",
861 		.data		= &acpi_realmode_flags,
862 		.maxlen		= sizeof (unsigned long),
863 		.mode		= 0644,
864 		.proc_handler	= proc_doulongvec_minmax,
865 	},
866 #endif
867 #ifdef CONFIG_IA64
868 	{
869 		.procname	= "ignore-unaligned-usertrap",
870 		.data		= &no_unaligned_warning,
871 		.maxlen		= sizeof (int),
872 	 	.mode		= 0644,
873 		.proc_handler	= proc_dointvec,
874 	},
875 	{
876 		.procname	= "unaligned-dump-stack",
877 		.data		= &unaligned_dump_stack,
878 		.maxlen		= sizeof (int),
879 		.mode		= 0644,
880 		.proc_handler	= proc_dointvec,
881 	},
882 #endif
883 #ifdef CONFIG_DETECT_HUNG_TASK
884 	{
885 		.procname	= "hung_task_panic",
886 		.data		= &sysctl_hung_task_panic,
887 		.maxlen		= sizeof(int),
888 		.mode		= 0644,
889 		.proc_handler	= proc_dointvec_minmax,
890 		.extra1		= &zero,
891 		.extra2		= &one,
892 	},
893 	{
894 		.procname	= "hung_task_check_count",
895 		.data		= &sysctl_hung_task_check_count,
896 		.maxlen		= sizeof(unsigned long),
897 		.mode		= 0644,
898 		.proc_handler	= proc_doulongvec_minmax,
899 	},
900 	{
901 		.procname	= "hung_task_timeout_secs",
902 		.data		= &sysctl_hung_task_timeout_secs,
903 		.maxlen		= sizeof(unsigned long),
904 		.mode		= 0644,
905 		.proc_handler	= proc_dohung_task_timeout_secs,
906 	},
907 	{
908 		.procname	= "hung_task_warnings",
909 		.data		= &sysctl_hung_task_warnings,
910 		.maxlen		= sizeof(unsigned long),
911 		.mode		= 0644,
912 		.proc_handler	= proc_doulongvec_minmax,
913 	},
914 #endif
915 #ifdef CONFIG_COMPAT
916 	{
917 		.procname	= "compat-log",
918 		.data		= &compat_log,
919 		.maxlen		= sizeof (int),
920 	 	.mode		= 0644,
921 		.proc_handler	= proc_dointvec,
922 	},
923 #endif
924 #ifdef CONFIG_RT_MUTEXES
925 	{
926 		.procname	= "max_lock_depth",
927 		.data		= &max_lock_depth,
928 		.maxlen		= sizeof(int),
929 		.mode		= 0644,
930 		.proc_handler	= proc_dointvec,
931 	},
932 #endif
933 	{
934 		.procname	= "poweroff_cmd",
935 		.data		= &poweroff_cmd,
936 		.maxlen		= POWEROFF_CMD_PATH_LEN,
937 		.mode		= 0644,
938 		.proc_handler	= proc_dostring,
939 	},
940 #ifdef CONFIG_KEYS
941 	{
942 		.procname	= "keys",
943 		.mode		= 0555,
944 		.child		= key_sysctls,
945 	},
946 #endif
947 #ifdef CONFIG_RCU_TORTURE_TEST
948 	{
949 		.procname       = "rcutorture_runnable",
950 		.data           = &rcutorture_runnable,
951 		.maxlen         = sizeof(int),
952 		.mode           = 0644,
953 		.proc_handler	= proc_dointvec,
954 	},
955 #endif
956 #ifdef CONFIG_PERF_EVENTS
957 	/*
958 	 * User-space scripts rely on the existence of this file
959 	 * as a feature check for perf_events being enabled.
960 	 *
961 	 * So it's an ABI, do not remove!
962 	 */
963 	{
964 		.procname	= "perf_event_paranoid",
965 		.data		= &sysctl_perf_event_paranoid,
966 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
967 		.mode		= 0644,
968 		.proc_handler	= proc_dointvec,
969 	},
970 	{
971 		.procname	= "perf_event_mlock_kb",
972 		.data		= &sysctl_perf_event_mlock,
973 		.maxlen		= sizeof(sysctl_perf_event_mlock),
974 		.mode		= 0644,
975 		.proc_handler	= proc_dointvec,
976 	},
977 	{
978 		.procname	= "perf_event_max_sample_rate",
979 		.data		= &sysctl_perf_event_sample_rate,
980 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
981 		.mode		= 0644,
982 		.proc_handler	= perf_proc_update_handler,
983 	},
984 #endif
985 #ifdef CONFIG_KMEMCHECK
986 	{
987 		.procname	= "kmemcheck",
988 		.data		= &kmemcheck_enabled,
989 		.maxlen		= sizeof(int),
990 		.mode		= 0644,
991 		.proc_handler	= proc_dointvec,
992 	},
993 #endif
994 #ifdef CONFIG_BLOCK
995 	{
996 		.procname	= "blk_iopoll",
997 		.data		= &blk_iopoll_enabled,
998 		.maxlen		= sizeof(int),
999 		.mode		= 0644,
1000 		.proc_handler	= proc_dointvec,
1001 	},
1002 #endif
1003 	{ }
1004 };
1005 
1006 static struct ctl_table vm_table[] = {
1007 	{
1008 		.procname	= "overcommit_memory",
1009 		.data		= &sysctl_overcommit_memory,
1010 		.maxlen		= sizeof(sysctl_overcommit_memory),
1011 		.mode		= 0644,
1012 		.proc_handler	= proc_dointvec_minmax,
1013 		.extra1		= &zero,
1014 		.extra2		= &two,
1015 	},
1016 	{
1017 		.procname	= "panic_on_oom",
1018 		.data		= &sysctl_panic_on_oom,
1019 		.maxlen		= sizeof(sysctl_panic_on_oom),
1020 		.mode		= 0644,
1021 		.proc_handler	= proc_dointvec_minmax,
1022 		.extra1		= &zero,
1023 		.extra2		= &two,
1024 	},
1025 	{
1026 		.procname	= "oom_kill_allocating_task",
1027 		.data		= &sysctl_oom_kill_allocating_task,
1028 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1029 		.mode		= 0644,
1030 		.proc_handler	= proc_dointvec,
1031 	},
1032 	{
1033 		.procname	= "oom_dump_tasks",
1034 		.data		= &sysctl_oom_dump_tasks,
1035 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1036 		.mode		= 0644,
1037 		.proc_handler	= proc_dointvec,
1038 	},
1039 	{
1040 		.procname	= "overcommit_ratio",
1041 		.data		= &sysctl_overcommit_ratio,
1042 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1043 		.mode		= 0644,
1044 		.proc_handler	= proc_dointvec,
1045 	},
1046 	{
1047 		.procname	= "page-cluster",
1048 		.data		= &page_cluster,
1049 		.maxlen		= sizeof(int),
1050 		.mode		= 0644,
1051 		.proc_handler	= proc_dointvec_minmax,
1052 		.extra1		= &zero,
1053 	},
1054 	{
1055 		.procname	= "dirty_background_ratio",
1056 		.data		= &dirty_background_ratio,
1057 		.maxlen		= sizeof(dirty_background_ratio),
1058 		.mode		= 0644,
1059 		.proc_handler	= dirty_background_ratio_handler,
1060 		.extra1		= &zero,
1061 		.extra2		= &one_hundred,
1062 	},
1063 	{
1064 		.procname	= "dirty_background_bytes",
1065 		.data		= &dirty_background_bytes,
1066 		.maxlen		= sizeof(dirty_background_bytes),
1067 		.mode		= 0644,
1068 		.proc_handler	= dirty_background_bytes_handler,
1069 		.extra1		= &one_ul,
1070 	},
1071 	{
1072 		.procname	= "dirty_ratio",
1073 		.data		= &vm_dirty_ratio,
1074 		.maxlen		= sizeof(vm_dirty_ratio),
1075 		.mode		= 0644,
1076 		.proc_handler	= dirty_ratio_handler,
1077 		.extra1		= &zero,
1078 		.extra2		= &one_hundred,
1079 	},
1080 	{
1081 		.procname	= "dirty_bytes",
1082 		.data		= &vm_dirty_bytes,
1083 		.maxlen		= sizeof(vm_dirty_bytes),
1084 		.mode		= 0644,
1085 		.proc_handler	= dirty_bytes_handler,
1086 		.extra1		= &dirty_bytes_min,
1087 	},
1088 	{
1089 		.procname	= "dirty_writeback_centisecs",
1090 		.data		= &dirty_writeback_interval,
1091 		.maxlen		= sizeof(dirty_writeback_interval),
1092 		.mode		= 0644,
1093 		.proc_handler	= dirty_writeback_centisecs_handler,
1094 	},
1095 	{
1096 		.procname	= "dirty_expire_centisecs",
1097 		.data		= &dirty_expire_interval,
1098 		.maxlen		= sizeof(dirty_expire_interval),
1099 		.mode		= 0644,
1100 		.proc_handler	= proc_dointvec_minmax,
1101 		.extra1		= &zero,
1102 	},
1103 	{
1104 		.procname	= "nr_pdflush_threads",
1105 		.data		= &nr_pdflush_threads,
1106 		.maxlen		= sizeof nr_pdflush_threads,
1107 		.mode		= 0444 /* read-only*/,
1108 		.proc_handler	= proc_dointvec,
1109 	},
1110 	{
1111 		.procname	= "swappiness",
1112 		.data		= &vm_swappiness,
1113 		.maxlen		= sizeof(vm_swappiness),
1114 		.mode		= 0644,
1115 		.proc_handler	= proc_dointvec_minmax,
1116 		.extra1		= &zero,
1117 		.extra2		= &one_hundred,
1118 	},
1119 #ifdef CONFIG_HUGETLB_PAGE
1120 	{
1121 		.procname	= "nr_hugepages",
1122 		.data		= NULL,
1123 		.maxlen		= sizeof(unsigned long),
1124 		.mode		= 0644,
1125 		.proc_handler	= hugetlb_sysctl_handler,
1126 		.extra1		= (void *)&hugetlb_zero,
1127 		.extra2		= (void *)&hugetlb_infinity,
1128 	},
1129 #ifdef CONFIG_NUMA
1130 	{
1131 		.procname       = "nr_hugepages_mempolicy",
1132 		.data           = NULL,
1133 		.maxlen         = sizeof(unsigned long),
1134 		.mode           = 0644,
1135 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1136 		.extra1		= (void *)&hugetlb_zero,
1137 		.extra2		= (void *)&hugetlb_infinity,
1138 	},
1139 #endif
1140 	 {
1141 		.procname	= "hugetlb_shm_group",
1142 		.data		= &sysctl_hugetlb_shm_group,
1143 		.maxlen		= sizeof(gid_t),
1144 		.mode		= 0644,
1145 		.proc_handler	= proc_dointvec,
1146 	 },
1147 	 {
1148 		.procname	= "hugepages_treat_as_movable",
1149 		.data		= &hugepages_treat_as_movable,
1150 		.maxlen		= sizeof(int),
1151 		.mode		= 0644,
1152 		.proc_handler	= hugetlb_treat_movable_handler,
1153 	},
1154 	{
1155 		.procname	= "nr_overcommit_hugepages",
1156 		.data		= NULL,
1157 		.maxlen		= sizeof(unsigned long),
1158 		.mode		= 0644,
1159 		.proc_handler	= hugetlb_overcommit_handler,
1160 		.extra1		= (void *)&hugetlb_zero,
1161 		.extra2		= (void *)&hugetlb_infinity,
1162 	},
1163 #endif
1164 	{
1165 		.procname	= "lowmem_reserve_ratio",
1166 		.data		= &sysctl_lowmem_reserve_ratio,
1167 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1168 		.mode		= 0644,
1169 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1170 	},
1171 	{
1172 		.procname	= "drop_caches",
1173 		.data		= &sysctl_drop_caches,
1174 		.maxlen		= sizeof(int),
1175 		.mode		= 0644,
1176 		.proc_handler	= drop_caches_sysctl_handler,
1177 		.extra1		= &one,
1178 		.extra2		= &three,
1179 	},
1180 #ifdef CONFIG_COMPACTION
1181 	{
1182 		.procname	= "compact_memory",
1183 		.data		= &sysctl_compact_memory,
1184 		.maxlen		= sizeof(int),
1185 		.mode		= 0200,
1186 		.proc_handler	= sysctl_compaction_handler,
1187 	},
1188 	{
1189 		.procname	= "extfrag_threshold",
1190 		.data		= &sysctl_extfrag_threshold,
1191 		.maxlen		= sizeof(int),
1192 		.mode		= 0644,
1193 		.proc_handler	= sysctl_extfrag_handler,
1194 		.extra1		= &min_extfrag_threshold,
1195 		.extra2		= &max_extfrag_threshold,
1196 	},
1197 
1198 #endif /* CONFIG_COMPACTION */
1199 	{
1200 		.procname	= "min_free_kbytes",
1201 		.data		= &min_free_kbytes,
1202 		.maxlen		= sizeof(min_free_kbytes),
1203 		.mode		= 0644,
1204 		.proc_handler	= min_free_kbytes_sysctl_handler,
1205 		.extra1		= &zero,
1206 	},
1207 	{
1208 		.procname	= "percpu_pagelist_fraction",
1209 		.data		= &percpu_pagelist_fraction,
1210 		.maxlen		= sizeof(percpu_pagelist_fraction),
1211 		.mode		= 0644,
1212 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1213 		.extra1		= &min_percpu_pagelist_fract,
1214 	},
1215 #ifdef CONFIG_MMU
1216 	{
1217 		.procname	= "max_map_count",
1218 		.data		= &sysctl_max_map_count,
1219 		.maxlen		= sizeof(sysctl_max_map_count),
1220 		.mode		= 0644,
1221 		.proc_handler	= proc_dointvec_minmax,
1222 		.extra1		= &zero,
1223 	},
1224 #else
1225 	{
1226 		.procname	= "nr_trim_pages",
1227 		.data		= &sysctl_nr_trim_pages,
1228 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1229 		.mode		= 0644,
1230 		.proc_handler	= proc_dointvec_minmax,
1231 		.extra1		= &zero,
1232 	},
1233 #endif
1234 	{
1235 		.procname	= "laptop_mode",
1236 		.data		= &laptop_mode,
1237 		.maxlen		= sizeof(laptop_mode),
1238 		.mode		= 0644,
1239 		.proc_handler	= proc_dointvec_jiffies,
1240 	},
1241 	{
1242 		.procname	= "block_dump",
1243 		.data		= &block_dump,
1244 		.maxlen		= sizeof(block_dump),
1245 		.mode		= 0644,
1246 		.proc_handler	= proc_dointvec,
1247 		.extra1		= &zero,
1248 	},
1249 	{
1250 		.procname	= "vfs_cache_pressure",
1251 		.data		= &sysctl_vfs_cache_pressure,
1252 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1253 		.mode		= 0644,
1254 		.proc_handler	= proc_dointvec,
1255 		.extra1		= &zero,
1256 	},
1257 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1258 	{
1259 		.procname	= "legacy_va_layout",
1260 		.data		= &sysctl_legacy_va_layout,
1261 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1262 		.mode		= 0644,
1263 		.proc_handler	= proc_dointvec,
1264 		.extra1		= &zero,
1265 	},
1266 #endif
1267 #ifdef CONFIG_NUMA
1268 	{
1269 		.procname	= "zone_reclaim_mode",
1270 		.data		= &zone_reclaim_mode,
1271 		.maxlen		= sizeof(zone_reclaim_mode),
1272 		.mode		= 0644,
1273 		.proc_handler	= proc_dointvec,
1274 		.extra1		= &zero,
1275 	},
1276 	{
1277 		.procname	= "min_unmapped_ratio",
1278 		.data		= &sysctl_min_unmapped_ratio,
1279 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1280 		.mode		= 0644,
1281 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1282 		.extra1		= &zero,
1283 		.extra2		= &one_hundred,
1284 	},
1285 	{
1286 		.procname	= "min_slab_ratio",
1287 		.data		= &sysctl_min_slab_ratio,
1288 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1289 		.mode		= 0644,
1290 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1291 		.extra1		= &zero,
1292 		.extra2		= &one_hundred,
1293 	},
1294 #endif
1295 #ifdef CONFIG_SMP
1296 	{
1297 		.procname	= "stat_interval",
1298 		.data		= &sysctl_stat_interval,
1299 		.maxlen		= sizeof(sysctl_stat_interval),
1300 		.mode		= 0644,
1301 		.proc_handler	= proc_dointvec_jiffies,
1302 	},
1303 #endif
1304 #ifdef CONFIG_MMU
1305 	{
1306 		.procname	= "mmap_min_addr",
1307 		.data		= &dac_mmap_min_addr,
1308 		.maxlen		= sizeof(unsigned long),
1309 		.mode		= 0644,
1310 		.proc_handler	= mmap_min_addr_handler,
1311 	},
1312 #endif
1313 #ifdef CONFIG_NUMA
1314 	{
1315 		.procname	= "numa_zonelist_order",
1316 		.data		= &numa_zonelist_order,
1317 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1318 		.mode		= 0644,
1319 		.proc_handler	= numa_zonelist_order_handler,
1320 	},
1321 #endif
1322 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1323    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1324 	{
1325 		.procname	= "vdso_enabled",
1326 		.data		= &vdso_enabled,
1327 		.maxlen		= sizeof(vdso_enabled),
1328 		.mode		= 0644,
1329 		.proc_handler	= proc_dointvec,
1330 		.extra1		= &zero,
1331 	},
1332 #endif
1333 #ifdef CONFIG_HIGHMEM
1334 	{
1335 		.procname	= "highmem_is_dirtyable",
1336 		.data		= &vm_highmem_is_dirtyable,
1337 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1338 		.mode		= 0644,
1339 		.proc_handler	= proc_dointvec_minmax,
1340 		.extra1		= &zero,
1341 		.extra2		= &one,
1342 	},
1343 #endif
1344 	{
1345 		.procname	= "scan_unevictable_pages",
1346 		.data		= &scan_unevictable_pages,
1347 		.maxlen		= sizeof(scan_unevictable_pages),
1348 		.mode		= 0644,
1349 		.proc_handler	= scan_unevictable_handler,
1350 	},
1351 #ifdef CONFIG_MEMORY_FAILURE
1352 	{
1353 		.procname	= "memory_failure_early_kill",
1354 		.data		= &sysctl_memory_failure_early_kill,
1355 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1356 		.mode		= 0644,
1357 		.proc_handler	= proc_dointvec_minmax,
1358 		.extra1		= &zero,
1359 		.extra2		= &one,
1360 	},
1361 	{
1362 		.procname	= "memory_failure_recovery",
1363 		.data		= &sysctl_memory_failure_recovery,
1364 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1365 		.mode		= 0644,
1366 		.proc_handler	= proc_dointvec_minmax,
1367 		.extra1		= &zero,
1368 		.extra2		= &one,
1369 	},
1370 #endif
1371 	{ }
1372 };
1373 
1374 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1375 static struct ctl_table binfmt_misc_table[] = {
1376 	{ }
1377 };
1378 #endif
1379 
1380 static struct ctl_table fs_table[] = {
1381 	{
1382 		.procname	= "inode-nr",
1383 		.data		= &inodes_stat,
1384 		.maxlen		= 2*sizeof(int),
1385 		.mode		= 0444,
1386 		.proc_handler	= proc_nr_inodes,
1387 	},
1388 	{
1389 		.procname	= "inode-state",
1390 		.data		= &inodes_stat,
1391 		.maxlen		= 7*sizeof(int),
1392 		.mode		= 0444,
1393 		.proc_handler	= proc_nr_inodes,
1394 	},
1395 	{
1396 		.procname	= "file-nr",
1397 		.data		= &files_stat,
1398 		.maxlen		= sizeof(files_stat),
1399 		.mode		= 0444,
1400 		.proc_handler	= proc_nr_files,
1401 	},
1402 	{
1403 		.procname	= "file-max",
1404 		.data		= &files_stat.max_files,
1405 		.maxlen		= sizeof(files_stat.max_files),
1406 		.mode		= 0644,
1407 		.proc_handler	= proc_doulongvec_minmax,
1408 	},
1409 	{
1410 		.procname	= "nr_open",
1411 		.data		= &sysctl_nr_open,
1412 		.maxlen		= sizeof(int),
1413 		.mode		= 0644,
1414 		.proc_handler	= proc_dointvec_minmax,
1415 		.extra1		= &sysctl_nr_open_min,
1416 		.extra2		= &sysctl_nr_open_max,
1417 	},
1418 	{
1419 		.procname	= "dentry-state",
1420 		.data		= &dentry_stat,
1421 		.maxlen		= 6*sizeof(int),
1422 		.mode		= 0444,
1423 		.proc_handler	= proc_nr_dentry,
1424 	},
1425 	{
1426 		.procname	= "overflowuid",
1427 		.data		= &fs_overflowuid,
1428 		.maxlen		= sizeof(int),
1429 		.mode		= 0644,
1430 		.proc_handler	= proc_dointvec_minmax,
1431 		.extra1		= &minolduid,
1432 		.extra2		= &maxolduid,
1433 	},
1434 	{
1435 		.procname	= "overflowgid",
1436 		.data		= &fs_overflowgid,
1437 		.maxlen		= sizeof(int),
1438 		.mode		= 0644,
1439 		.proc_handler	= proc_dointvec_minmax,
1440 		.extra1		= &minolduid,
1441 		.extra2		= &maxolduid,
1442 	},
1443 #ifdef CONFIG_FILE_LOCKING
1444 	{
1445 		.procname	= "leases-enable",
1446 		.data		= &leases_enable,
1447 		.maxlen		= sizeof(int),
1448 		.mode		= 0644,
1449 		.proc_handler	= proc_dointvec,
1450 	},
1451 #endif
1452 #ifdef CONFIG_DNOTIFY
1453 	{
1454 		.procname	= "dir-notify-enable",
1455 		.data		= &dir_notify_enable,
1456 		.maxlen		= sizeof(int),
1457 		.mode		= 0644,
1458 		.proc_handler	= proc_dointvec,
1459 	},
1460 #endif
1461 #ifdef CONFIG_MMU
1462 #ifdef CONFIG_FILE_LOCKING
1463 	{
1464 		.procname	= "lease-break-time",
1465 		.data		= &lease_break_time,
1466 		.maxlen		= sizeof(int),
1467 		.mode		= 0644,
1468 		.proc_handler	= proc_dointvec,
1469 	},
1470 #endif
1471 #ifdef CONFIG_AIO
1472 	{
1473 		.procname	= "aio-nr",
1474 		.data		= &aio_nr,
1475 		.maxlen		= sizeof(aio_nr),
1476 		.mode		= 0444,
1477 		.proc_handler	= proc_doulongvec_minmax,
1478 	},
1479 	{
1480 		.procname	= "aio-max-nr",
1481 		.data		= &aio_max_nr,
1482 		.maxlen		= sizeof(aio_max_nr),
1483 		.mode		= 0644,
1484 		.proc_handler	= proc_doulongvec_minmax,
1485 	},
1486 #endif /* CONFIG_AIO */
1487 #ifdef CONFIG_INOTIFY_USER
1488 	{
1489 		.procname	= "inotify",
1490 		.mode		= 0555,
1491 		.child		= inotify_table,
1492 	},
1493 #endif
1494 #ifdef CONFIG_EPOLL
1495 	{
1496 		.procname	= "epoll",
1497 		.mode		= 0555,
1498 		.child		= epoll_table,
1499 	},
1500 #endif
1501 #endif
1502 	{
1503 		.procname	= "suid_dumpable",
1504 		.data		= &suid_dumpable,
1505 		.maxlen		= sizeof(int),
1506 		.mode		= 0644,
1507 		.proc_handler	= proc_dointvec_minmax_coredump,
1508 		.extra1		= &zero,
1509 		.extra2		= &two,
1510 	},
1511 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1512 	{
1513 		.procname	= "binfmt_misc",
1514 		.mode		= 0555,
1515 		.child		= binfmt_misc_table,
1516 	},
1517 #endif
1518 	{
1519 		.procname	= "pipe-max-size",
1520 		.data		= &pipe_max_size,
1521 		.maxlen		= sizeof(int),
1522 		.mode		= 0644,
1523 		.proc_handler	= &pipe_proc_fn,
1524 		.extra1		= &pipe_min_size,
1525 	},
1526 	{ }
1527 };
1528 
1529 static struct ctl_table debug_table[] = {
1530 #if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
1531     defined(CONFIG_S390) || defined(CONFIG_TILE)
1532 	{
1533 		.procname	= "exception-trace",
1534 		.data		= &show_unhandled_signals,
1535 		.maxlen		= sizeof(int),
1536 		.mode		= 0644,
1537 		.proc_handler	= proc_dointvec
1538 	},
1539 #endif
1540 #if defined(CONFIG_OPTPROBES)
1541 	{
1542 		.procname	= "kprobes-optimization",
1543 		.data		= &sysctl_kprobes_optimization,
1544 		.maxlen		= sizeof(int),
1545 		.mode		= 0644,
1546 		.proc_handler	= proc_kprobes_optimization_handler,
1547 		.extra1		= &zero,
1548 		.extra2		= &one,
1549 	},
1550 #endif
1551 	{ }
1552 };
1553 
1554 static struct ctl_table dev_table[] = {
1555 	{ }
1556 };
1557 
1558 int __init sysctl_init(void)
1559 {
1560 	struct ctl_table_header *hdr;
1561 
1562 	hdr = register_sysctl_table(sysctl_base_table);
1563 	kmemleak_not_leak(hdr);
1564 	return 0;
1565 }
1566 
1567 #endif /* CONFIG_SYSCTL */
1568 
1569 /*
1570  * /proc/sys support
1571  */
1572 
1573 #ifdef CONFIG_PROC_SYSCTL
1574 
1575 static int _proc_do_string(void* data, int maxlen, int write,
1576 			   void __user *buffer,
1577 			   size_t *lenp, loff_t *ppos)
1578 {
1579 	size_t len;
1580 	char __user *p;
1581 	char c;
1582 
1583 	if (!data || !maxlen || !*lenp) {
1584 		*lenp = 0;
1585 		return 0;
1586 	}
1587 
1588 	if (write) {
1589 		len = 0;
1590 		p = buffer;
1591 		while (len < *lenp) {
1592 			if (get_user(c, p++))
1593 				return -EFAULT;
1594 			if (c == 0 || c == '\n')
1595 				break;
1596 			len++;
1597 		}
1598 		if (len >= maxlen)
1599 			len = maxlen-1;
1600 		if(copy_from_user(data, buffer, len))
1601 			return -EFAULT;
1602 		((char *) data)[len] = 0;
1603 		*ppos += *lenp;
1604 	} else {
1605 		len = strlen(data);
1606 		if (len > maxlen)
1607 			len = maxlen;
1608 
1609 		if (*ppos > len) {
1610 			*lenp = 0;
1611 			return 0;
1612 		}
1613 
1614 		data += *ppos;
1615 		len  -= *ppos;
1616 
1617 		if (len > *lenp)
1618 			len = *lenp;
1619 		if (len)
1620 			if(copy_to_user(buffer, data, len))
1621 				return -EFAULT;
1622 		if (len < *lenp) {
1623 			if(put_user('\n', ((char __user *) buffer) + len))
1624 				return -EFAULT;
1625 			len++;
1626 		}
1627 		*lenp = len;
1628 		*ppos += len;
1629 	}
1630 	return 0;
1631 }
1632 
1633 /**
1634  * proc_dostring - read a string sysctl
1635  * @table: the sysctl table
1636  * @write: %TRUE if this is a write to the sysctl file
1637  * @buffer: the user buffer
1638  * @lenp: the size of the user buffer
1639  * @ppos: file position
1640  *
1641  * Reads/writes a string from/to the user buffer. If the kernel
1642  * buffer provided is not large enough to hold the string, the
1643  * string is truncated. The copied string is %NULL-terminated.
1644  * If the string is being read by the user process, it is copied
1645  * and a newline '\n' is added. It is truncated if the buffer is
1646  * not large enough.
1647  *
1648  * Returns 0 on success.
1649  */
1650 int proc_dostring(struct ctl_table *table, int write,
1651 		  void __user *buffer, size_t *lenp, loff_t *ppos)
1652 {
1653 	return _proc_do_string(table->data, table->maxlen, write,
1654 			       buffer, lenp, ppos);
1655 }
1656 
1657 static size_t proc_skip_spaces(char **buf)
1658 {
1659 	size_t ret;
1660 	char *tmp = skip_spaces(*buf);
1661 	ret = tmp - *buf;
1662 	*buf = tmp;
1663 	return ret;
1664 }
1665 
1666 static void proc_skip_char(char **buf, size_t *size, const char v)
1667 {
1668 	while (*size) {
1669 		if (**buf != v)
1670 			break;
1671 		(*size)--;
1672 		(*buf)++;
1673 	}
1674 }
1675 
1676 #define TMPBUFLEN 22
1677 /**
1678  * proc_get_long - reads an ASCII formatted integer from a user buffer
1679  *
1680  * @buf: a kernel buffer
1681  * @size: size of the kernel buffer
1682  * @val: this is where the number will be stored
1683  * @neg: set to %TRUE if number is negative
1684  * @perm_tr: a vector which contains the allowed trailers
1685  * @perm_tr_len: size of the perm_tr vector
1686  * @tr: pointer to store the trailer character
1687  *
1688  * In case of success %0 is returned and @buf and @size are updated with
1689  * the amount of bytes read. If @tr is non-NULL and a trailing
1690  * character exists (size is non-zero after returning from this
1691  * function), @tr is updated with the trailing character.
1692  */
1693 static int proc_get_long(char **buf, size_t *size,
1694 			  unsigned long *val, bool *neg,
1695 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
1696 {
1697 	int len;
1698 	char *p, tmp[TMPBUFLEN];
1699 
1700 	if (!*size)
1701 		return -EINVAL;
1702 
1703 	len = *size;
1704 	if (len > TMPBUFLEN - 1)
1705 		len = TMPBUFLEN - 1;
1706 
1707 	memcpy(tmp, *buf, len);
1708 
1709 	tmp[len] = 0;
1710 	p = tmp;
1711 	if (*p == '-' && *size > 1) {
1712 		*neg = true;
1713 		p++;
1714 	} else
1715 		*neg = false;
1716 	if (!isdigit(*p))
1717 		return -EINVAL;
1718 
1719 	*val = simple_strtoul(p, &p, 0);
1720 
1721 	len = p - tmp;
1722 
1723 	/* We don't know if the next char is whitespace thus we may accept
1724 	 * invalid integers (e.g. 1234...a) or two integers instead of one
1725 	 * (e.g. 123...1). So lets not allow such large numbers. */
1726 	if (len == TMPBUFLEN - 1)
1727 		return -EINVAL;
1728 
1729 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
1730 		return -EINVAL;
1731 
1732 	if (tr && (len < *size))
1733 		*tr = *p;
1734 
1735 	*buf += len;
1736 	*size -= len;
1737 
1738 	return 0;
1739 }
1740 
1741 /**
1742  * proc_put_long - converts an integer to a decimal ASCII formatted string
1743  *
1744  * @buf: the user buffer
1745  * @size: the size of the user buffer
1746  * @val: the integer to be converted
1747  * @neg: sign of the number, %TRUE for negative
1748  *
1749  * In case of success %0 is returned and @buf and @size are updated with
1750  * the amount of bytes written.
1751  */
1752 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
1753 			  bool neg)
1754 {
1755 	int len;
1756 	char tmp[TMPBUFLEN], *p = tmp;
1757 
1758 	sprintf(p, "%s%lu", neg ? "-" : "", val);
1759 	len = strlen(tmp);
1760 	if (len > *size)
1761 		len = *size;
1762 	if (copy_to_user(*buf, tmp, len))
1763 		return -EFAULT;
1764 	*size -= len;
1765 	*buf += len;
1766 	return 0;
1767 }
1768 #undef TMPBUFLEN
1769 
1770 static int proc_put_char(void __user **buf, size_t *size, char c)
1771 {
1772 	if (*size) {
1773 		char __user **buffer = (char __user **)buf;
1774 		if (put_user(c, *buffer))
1775 			return -EFAULT;
1776 		(*size)--, (*buffer)++;
1777 		*buf = *buffer;
1778 	}
1779 	return 0;
1780 }
1781 
1782 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
1783 				 int *valp,
1784 				 int write, void *data)
1785 {
1786 	if (write) {
1787 		*valp = *negp ? -*lvalp : *lvalp;
1788 	} else {
1789 		int val = *valp;
1790 		if (val < 0) {
1791 			*negp = true;
1792 			*lvalp = (unsigned long)-val;
1793 		} else {
1794 			*negp = false;
1795 			*lvalp = (unsigned long)val;
1796 		}
1797 	}
1798 	return 0;
1799 }
1800 
1801 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
1802 
1803 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1804 		  int write, void __user *buffer,
1805 		  size_t *lenp, loff_t *ppos,
1806 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1807 			      int write, void *data),
1808 		  void *data)
1809 {
1810 	int *i, vleft, first = 1, err = 0;
1811 	unsigned long page = 0;
1812 	size_t left;
1813 	char *kbuf;
1814 
1815 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
1816 		*lenp = 0;
1817 		return 0;
1818 	}
1819 
1820 	i = (int *) tbl_data;
1821 	vleft = table->maxlen / sizeof(*i);
1822 	left = *lenp;
1823 
1824 	if (!conv)
1825 		conv = do_proc_dointvec_conv;
1826 
1827 	if (write) {
1828 		if (left > PAGE_SIZE - 1)
1829 			left = PAGE_SIZE - 1;
1830 		page = __get_free_page(GFP_TEMPORARY);
1831 		kbuf = (char *) page;
1832 		if (!kbuf)
1833 			return -ENOMEM;
1834 		if (copy_from_user(kbuf, buffer, left)) {
1835 			err = -EFAULT;
1836 			goto free;
1837 		}
1838 		kbuf[left] = 0;
1839 	}
1840 
1841 	for (; left && vleft--; i++, first=0) {
1842 		unsigned long lval;
1843 		bool neg;
1844 
1845 		if (write) {
1846 			left -= proc_skip_spaces(&kbuf);
1847 
1848 			if (!left)
1849 				break;
1850 			err = proc_get_long(&kbuf, &left, &lval, &neg,
1851 					     proc_wspace_sep,
1852 					     sizeof(proc_wspace_sep), NULL);
1853 			if (err)
1854 				break;
1855 			if (conv(&neg, &lval, i, 1, data)) {
1856 				err = -EINVAL;
1857 				break;
1858 			}
1859 		} else {
1860 			if (conv(&neg, &lval, i, 0, data)) {
1861 				err = -EINVAL;
1862 				break;
1863 			}
1864 			if (!first)
1865 				err = proc_put_char(&buffer, &left, '\t');
1866 			if (err)
1867 				break;
1868 			err = proc_put_long(&buffer, &left, lval, neg);
1869 			if (err)
1870 				break;
1871 		}
1872 	}
1873 
1874 	if (!write && !first && left && !err)
1875 		err = proc_put_char(&buffer, &left, '\n');
1876 	if (write && !err && left)
1877 		left -= proc_skip_spaces(&kbuf);
1878 free:
1879 	if (write) {
1880 		free_page(page);
1881 		if (first)
1882 			return err ? : -EINVAL;
1883 	}
1884 	*lenp -= left;
1885 	*ppos += *lenp;
1886 	return err;
1887 }
1888 
1889 static int do_proc_dointvec(struct ctl_table *table, int write,
1890 		  void __user *buffer, size_t *lenp, loff_t *ppos,
1891 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1892 			      int write, void *data),
1893 		  void *data)
1894 {
1895 	return __do_proc_dointvec(table->data, table, write,
1896 			buffer, lenp, ppos, conv, data);
1897 }
1898 
1899 /**
1900  * proc_dointvec - read a vector of integers
1901  * @table: the sysctl table
1902  * @write: %TRUE if this is a write to the sysctl file
1903  * @buffer: the user buffer
1904  * @lenp: the size of the user buffer
1905  * @ppos: file position
1906  *
1907  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1908  * values from/to the user buffer, treated as an ASCII string.
1909  *
1910  * Returns 0 on success.
1911  */
1912 int proc_dointvec(struct ctl_table *table, int write,
1913 		     void __user *buffer, size_t *lenp, loff_t *ppos)
1914 {
1915     return do_proc_dointvec(table,write,buffer,lenp,ppos,
1916 		    	    NULL,NULL);
1917 }
1918 
1919 /*
1920  * Taint values can only be increased
1921  * This means we can safely use a temporary.
1922  */
1923 static int proc_taint(struct ctl_table *table, int write,
1924 			       void __user *buffer, size_t *lenp, loff_t *ppos)
1925 {
1926 	struct ctl_table t;
1927 	unsigned long tmptaint = get_taint();
1928 	int err;
1929 
1930 	if (write && !capable(CAP_SYS_ADMIN))
1931 		return -EPERM;
1932 
1933 	t = *table;
1934 	t.data = &tmptaint;
1935 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
1936 	if (err < 0)
1937 		return err;
1938 
1939 	if (write) {
1940 		/*
1941 		 * Poor man's atomic or. Not worth adding a primitive
1942 		 * to everyone's atomic.h for this
1943 		 */
1944 		int i;
1945 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
1946 			if ((tmptaint >> i) & 1)
1947 				add_taint(i);
1948 		}
1949 	}
1950 
1951 	return err;
1952 }
1953 
1954 #ifdef CONFIG_PRINTK
1955 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
1956 				void __user *buffer, size_t *lenp, loff_t *ppos)
1957 {
1958 	if (write && !capable(CAP_SYS_ADMIN))
1959 		return -EPERM;
1960 
1961 	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
1962 }
1963 #endif
1964 
1965 struct do_proc_dointvec_minmax_conv_param {
1966 	int *min;
1967 	int *max;
1968 };
1969 
1970 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
1971 					int *valp,
1972 					int write, void *data)
1973 {
1974 	struct do_proc_dointvec_minmax_conv_param *param = data;
1975 	if (write) {
1976 		int val = *negp ? -*lvalp : *lvalp;
1977 		if ((param->min && *param->min > val) ||
1978 		    (param->max && *param->max < val))
1979 			return -EINVAL;
1980 		*valp = val;
1981 	} else {
1982 		int val = *valp;
1983 		if (val < 0) {
1984 			*negp = true;
1985 			*lvalp = (unsigned long)-val;
1986 		} else {
1987 			*negp = false;
1988 			*lvalp = (unsigned long)val;
1989 		}
1990 	}
1991 	return 0;
1992 }
1993 
1994 /**
1995  * proc_dointvec_minmax - read a vector of integers with min/max values
1996  * @table: the sysctl table
1997  * @write: %TRUE if this is a write to the sysctl file
1998  * @buffer: the user buffer
1999  * @lenp: the size of the user buffer
2000  * @ppos: file position
2001  *
2002  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2003  * values from/to the user buffer, treated as an ASCII string.
2004  *
2005  * This routine will ensure the values are within the range specified by
2006  * table->extra1 (min) and table->extra2 (max).
2007  *
2008  * Returns 0 on success.
2009  */
2010 int proc_dointvec_minmax(struct ctl_table *table, int write,
2011 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2012 {
2013 	struct do_proc_dointvec_minmax_conv_param param = {
2014 		.min = (int *) table->extra1,
2015 		.max = (int *) table->extra2,
2016 	};
2017 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2018 				do_proc_dointvec_minmax_conv, &param);
2019 }
2020 
2021 static void validate_coredump_safety(void)
2022 {
2023 	if (suid_dumpable == SUID_DUMPABLE_SAFE &&
2024 	    core_pattern[0] != '/' && core_pattern[0] != '|') {
2025 		printk(KERN_WARNING "Unsafe core_pattern used with "\
2026 			"suid_dumpable=2. Pipe handler or fully qualified "\
2027 			"core dump path required.\n");
2028 	}
2029 }
2030 
2031 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2032 		void __user *buffer, size_t *lenp, loff_t *ppos)
2033 {
2034 	int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2035 	if (!error)
2036 		validate_coredump_safety();
2037 	return error;
2038 }
2039 
2040 static int proc_dostring_coredump(struct ctl_table *table, int write,
2041 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2042 {
2043 	int error = proc_dostring(table, write, buffer, lenp, ppos);
2044 	if (!error)
2045 		validate_coredump_safety();
2046 	return error;
2047 }
2048 
2049 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2050 				     void __user *buffer,
2051 				     size_t *lenp, loff_t *ppos,
2052 				     unsigned long convmul,
2053 				     unsigned long convdiv)
2054 {
2055 	unsigned long *i, *min, *max;
2056 	int vleft, first = 1, err = 0;
2057 	unsigned long page = 0;
2058 	size_t left;
2059 	char *kbuf;
2060 
2061 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2062 		*lenp = 0;
2063 		return 0;
2064 	}
2065 
2066 	i = (unsigned long *) data;
2067 	min = (unsigned long *) table->extra1;
2068 	max = (unsigned long *) table->extra2;
2069 	vleft = table->maxlen / sizeof(unsigned long);
2070 	left = *lenp;
2071 
2072 	if (write) {
2073 		if (left > PAGE_SIZE - 1)
2074 			left = PAGE_SIZE - 1;
2075 		page = __get_free_page(GFP_TEMPORARY);
2076 		kbuf = (char *) page;
2077 		if (!kbuf)
2078 			return -ENOMEM;
2079 		if (copy_from_user(kbuf, buffer, left)) {
2080 			err = -EFAULT;
2081 			goto free;
2082 		}
2083 		kbuf[left] = 0;
2084 	}
2085 
2086 	for (; left && vleft--; i++, first = 0) {
2087 		unsigned long val;
2088 
2089 		if (write) {
2090 			bool neg;
2091 
2092 			left -= proc_skip_spaces(&kbuf);
2093 
2094 			err = proc_get_long(&kbuf, &left, &val, &neg,
2095 					     proc_wspace_sep,
2096 					     sizeof(proc_wspace_sep), NULL);
2097 			if (err)
2098 				break;
2099 			if (neg)
2100 				continue;
2101 			if ((min && val < *min) || (max && val > *max))
2102 				continue;
2103 			*i = val;
2104 		} else {
2105 			val = convdiv * (*i) / convmul;
2106 			if (!first)
2107 				err = proc_put_char(&buffer, &left, '\t');
2108 			err = proc_put_long(&buffer, &left, val, false);
2109 			if (err)
2110 				break;
2111 		}
2112 	}
2113 
2114 	if (!write && !first && left && !err)
2115 		err = proc_put_char(&buffer, &left, '\n');
2116 	if (write && !err)
2117 		left -= proc_skip_spaces(&kbuf);
2118 free:
2119 	if (write) {
2120 		free_page(page);
2121 		if (first)
2122 			return err ? : -EINVAL;
2123 	}
2124 	*lenp -= left;
2125 	*ppos += *lenp;
2126 	return err;
2127 }
2128 
2129 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2130 				     void __user *buffer,
2131 				     size_t *lenp, loff_t *ppos,
2132 				     unsigned long convmul,
2133 				     unsigned long convdiv)
2134 {
2135 	return __do_proc_doulongvec_minmax(table->data, table, write,
2136 			buffer, lenp, ppos, convmul, convdiv);
2137 }
2138 
2139 /**
2140  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2141  * @table: the sysctl table
2142  * @write: %TRUE if this is a write to the sysctl file
2143  * @buffer: the user buffer
2144  * @lenp: the size of the user buffer
2145  * @ppos: file position
2146  *
2147  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2148  * values from/to the user buffer, treated as an ASCII string.
2149  *
2150  * This routine will ensure the values are within the range specified by
2151  * table->extra1 (min) and table->extra2 (max).
2152  *
2153  * Returns 0 on success.
2154  */
2155 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2156 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2157 {
2158     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2159 }
2160 
2161 /**
2162  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2163  * @table: the sysctl table
2164  * @write: %TRUE if this is a write to the sysctl file
2165  * @buffer: the user buffer
2166  * @lenp: the size of the user buffer
2167  * @ppos: file position
2168  *
2169  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2170  * values from/to the user buffer, treated as an ASCII string. The values
2171  * are treated as milliseconds, and converted to jiffies when they are stored.
2172  *
2173  * This routine will ensure the values are within the range specified by
2174  * table->extra1 (min) and table->extra2 (max).
2175  *
2176  * Returns 0 on success.
2177  */
2178 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2179 				      void __user *buffer,
2180 				      size_t *lenp, loff_t *ppos)
2181 {
2182     return do_proc_doulongvec_minmax(table, write, buffer,
2183 				     lenp, ppos, HZ, 1000l);
2184 }
2185 
2186 
2187 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2188 					 int *valp,
2189 					 int write, void *data)
2190 {
2191 	if (write) {
2192 		if (*lvalp > LONG_MAX / HZ)
2193 			return 1;
2194 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2195 	} else {
2196 		int val = *valp;
2197 		unsigned long lval;
2198 		if (val < 0) {
2199 			*negp = true;
2200 			lval = (unsigned long)-val;
2201 		} else {
2202 			*negp = false;
2203 			lval = (unsigned long)val;
2204 		}
2205 		*lvalp = lval / HZ;
2206 	}
2207 	return 0;
2208 }
2209 
2210 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2211 						int *valp,
2212 						int write, void *data)
2213 {
2214 	if (write) {
2215 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2216 			return 1;
2217 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2218 	} else {
2219 		int val = *valp;
2220 		unsigned long lval;
2221 		if (val < 0) {
2222 			*negp = true;
2223 			lval = (unsigned long)-val;
2224 		} else {
2225 			*negp = false;
2226 			lval = (unsigned long)val;
2227 		}
2228 		*lvalp = jiffies_to_clock_t(lval);
2229 	}
2230 	return 0;
2231 }
2232 
2233 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2234 					    int *valp,
2235 					    int write, void *data)
2236 {
2237 	if (write) {
2238 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2239 	} else {
2240 		int val = *valp;
2241 		unsigned long lval;
2242 		if (val < 0) {
2243 			*negp = true;
2244 			lval = (unsigned long)-val;
2245 		} else {
2246 			*negp = false;
2247 			lval = (unsigned long)val;
2248 		}
2249 		*lvalp = jiffies_to_msecs(lval);
2250 	}
2251 	return 0;
2252 }
2253 
2254 /**
2255  * proc_dointvec_jiffies - read a vector of integers as seconds
2256  * @table: the sysctl table
2257  * @write: %TRUE if this is a write to the sysctl file
2258  * @buffer: the user buffer
2259  * @lenp: the size of the user buffer
2260  * @ppos: file position
2261  *
2262  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2263  * values from/to the user buffer, treated as an ASCII string.
2264  * The values read are assumed to be in seconds, and are converted into
2265  * jiffies.
2266  *
2267  * Returns 0 on success.
2268  */
2269 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2270 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2271 {
2272     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2273 		    	    do_proc_dointvec_jiffies_conv,NULL);
2274 }
2275 
2276 /**
2277  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2278  * @table: the sysctl table
2279  * @write: %TRUE if this is a write to the sysctl file
2280  * @buffer: the user buffer
2281  * @lenp: the size of the user buffer
2282  * @ppos: pointer to the file position
2283  *
2284  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2285  * values from/to the user buffer, treated as an ASCII string.
2286  * The values read are assumed to be in 1/USER_HZ seconds, and
2287  * are converted into jiffies.
2288  *
2289  * Returns 0 on success.
2290  */
2291 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2292 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2293 {
2294     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2295 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2296 }
2297 
2298 /**
2299  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2300  * @table: the sysctl table
2301  * @write: %TRUE if this is a write to the sysctl file
2302  * @buffer: the user buffer
2303  * @lenp: the size of the user buffer
2304  * @ppos: file position
2305  * @ppos: the current position in the file
2306  *
2307  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2308  * values from/to the user buffer, treated as an ASCII string.
2309  * The values read are assumed to be in 1/1000 seconds, and
2310  * are converted into jiffies.
2311  *
2312  * Returns 0 on success.
2313  */
2314 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2315 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2316 {
2317 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2318 				do_proc_dointvec_ms_jiffies_conv, NULL);
2319 }
2320 
2321 static int proc_do_cad_pid(struct ctl_table *table, int write,
2322 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2323 {
2324 	struct pid *new_pid;
2325 	pid_t tmp;
2326 	int r;
2327 
2328 	tmp = pid_vnr(cad_pid);
2329 
2330 	r = __do_proc_dointvec(&tmp, table, write, buffer,
2331 			       lenp, ppos, NULL, NULL);
2332 	if (r || !write)
2333 		return r;
2334 
2335 	new_pid = find_get_pid(tmp);
2336 	if (!new_pid)
2337 		return -ESRCH;
2338 
2339 	put_pid(xchg(&cad_pid, new_pid));
2340 	return 0;
2341 }
2342 
2343 /**
2344  * proc_do_large_bitmap - read/write from/to a large bitmap
2345  * @table: the sysctl table
2346  * @write: %TRUE if this is a write to the sysctl file
2347  * @buffer: the user buffer
2348  * @lenp: the size of the user buffer
2349  * @ppos: file position
2350  *
2351  * The bitmap is stored at table->data and the bitmap length (in bits)
2352  * in table->maxlen.
2353  *
2354  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2355  * large bitmaps may be represented in a compact manner. Writing into
2356  * the file will clear the bitmap then update it with the given input.
2357  *
2358  * Returns 0 on success.
2359  */
2360 int proc_do_large_bitmap(struct ctl_table *table, int write,
2361 			 void __user *buffer, size_t *lenp, loff_t *ppos)
2362 {
2363 	int err = 0;
2364 	bool first = 1;
2365 	size_t left = *lenp;
2366 	unsigned long bitmap_len = table->maxlen;
2367 	unsigned long *bitmap = (unsigned long *) table->data;
2368 	unsigned long *tmp_bitmap = NULL;
2369 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2370 
2371 	if (!bitmap_len || !left || (*ppos && !write)) {
2372 		*lenp = 0;
2373 		return 0;
2374 	}
2375 
2376 	if (write) {
2377 		unsigned long page = 0;
2378 		char *kbuf;
2379 
2380 		if (left > PAGE_SIZE - 1)
2381 			left = PAGE_SIZE - 1;
2382 
2383 		page = __get_free_page(GFP_TEMPORARY);
2384 		kbuf = (char *) page;
2385 		if (!kbuf)
2386 			return -ENOMEM;
2387 		if (copy_from_user(kbuf, buffer, left)) {
2388 			free_page(page);
2389 			return -EFAULT;
2390                 }
2391 		kbuf[left] = 0;
2392 
2393 		tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2394 				     GFP_KERNEL);
2395 		if (!tmp_bitmap) {
2396 			free_page(page);
2397 			return -ENOMEM;
2398 		}
2399 		proc_skip_char(&kbuf, &left, '\n');
2400 		while (!err && left) {
2401 			unsigned long val_a, val_b;
2402 			bool neg;
2403 
2404 			err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2405 					     sizeof(tr_a), &c);
2406 			if (err)
2407 				break;
2408 			if (val_a >= bitmap_len || neg) {
2409 				err = -EINVAL;
2410 				break;
2411 			}
2412 
2413 			val_b = val_a;
2414 			if (left) {
2415 				kbuf++;
2416 				left--;
2417 			}
2418 
2419 			if (c == '-') {
2420 				err = proc_get_long(&kbuf, &left, &val_b,
2421 						     &neg, tr_b, sizeof(tr_b),
2422 						     &c);
2423 				if (err)
2424 					break;
2425 				if (val_b >= bitmap_len || neg ||
2426 				    val_a > val_b) {
2427 					err = -EINVAL;
2428 					break;
2429 				}
2430 				if (left) {
2431 					kbuf++;
2432 					left--;
2433 				}
2434 			}
2435 
2436 			bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2437 			first = 0;
2438 			proc_skip_char(&kbuf, &left, '\n');
2439 		}
2440 		free_page(page);
2441 	} else {
2442 		unsigned long bit_a, bit_b = 0;
2443 
2444 		while (left) {
2445 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2446 			if (bit_a >= bitmap_len)
2447 				break;
2448 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
2449 						   bit_a + 1) - 1;
2450 
2451 			if (!first) {
2452 				err = proc_put_char(&buffer, &left, ',');
2453 				if (err)
2454 					break;
2455 			}
2456 			err = proc_put_long(&buffer, &left, bit_a, false);
2457 			if (err)
2458 				break;
2459 			if (bit_a != bit_b) {
2460 				err = proc_put_char(&buffer, &left, '-');
2461 				if (err)
2462 					break;
2463 				err = proc_put_long(&buffer, &left, bit_b, false);
2464 				if (err)
2465 					break;
2466 			}
2467 
2468 			first = 0; bit_b++;
2469 		}
2470 		if (!err)
2471 			err = proc_put_char(&buffer, &left, '\n');
2472 	}
2473 
2474 	if (!err) {
2475 		if (write) {
2476 			if (*ppos)
2477 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2478 			else
2479 				bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2480 		}
2481 		kfree(tmp_bitmap);
2482 		*lenp -= left;
2483 		*ppos += *lenp;
2484 		return 0;
2485 	} else {
2486 		kfree(tmp_bitmap);
2487 		return err;
2488 	}
2489 }
2490 
2491 #else /* CONFIG_PROC_SYSCTL */
2492 
2493 int proc_dostring(struct ctl_table *table, int write,
2494 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2495 {
2496 	return -ENOSYS;
2497 }
2498 
2499 int proc_dointvec(struct ctl_table *table, int write,
2500 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2501 {
2502 	return -ENOSYS;
2503 }
2504 
2505 int proc_dointvec_minmax(struct ctl_table *table, int write,
2506 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2507 {
2508 	return -ENOSYS;
2509 }
2510 
2511 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2512 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2513 {
2514 	return -ENOSYS;
2515 }
2516 
2517 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2518 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2519 {
2520 	return -ENOSYS;
2521 }
2522 
2523 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2524 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2525 {
2526 	return -ENOSYS;
2527 }
2528 
2529 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2530 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2531 {
2532 	return -ENOSYS;
2533 }
2534 
2535 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2536 				      void __user *buffer,
2537 				      size_t *lenp, loff_t *ppos)
2538 {
2539     return -ENOSYS;
2540 }
2541 
2542 
2543 #endif /* CONFIG_PROC_SYSCTL */
2544 
2545 /*
2546  * No sense putting this after each symbol definition, twice,
2547  * exception granted :-)
2548  */
2549 EXPORT_SYMBOL(proc_dointvec);
2550 EXPORT_SYMBOL(proc_dointvec_jiffies);
2551 EXPORT_SYMBOL(proc_dointvec_minmax);
2552 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2553 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2554 EXPORT_SYMBOL(proc_dostring);
2555 EXPORT_SYMBOL(proc_doulongvec_minmax);
2556 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2557