xref: /linux/kernel/sysctl.c (revision 12871a0bd67dd4db4418e1daafcd46e9d329ef10)
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/signal.h>
27 #include <linux/printk.h>
28 #include <linux/proc_fs.h>
29 #include <linux/security.h>
30 #include <linux/ctype.h>
31 #include <linux/kmemcheck.h>
32 #include <linux/fs.h>
33 #include <linux/init.h>
34 #include <linux/kernel.h>
35 #include <linux/kobject.h>
36 #include <linux/net.h>
37 #include <linux/sysrq.h>
38 #include <linux/highuid.h>
39 #include <linux/writeback.h>
40 #include <linux/ratelimit.h>
41 #include <linux/compaction.h>
42 #include <linux/hugetlb.h>
43 #include <linux/initrd.h>
44 #include <linux/key.h>
45 #include <linux/times.h>
46 #include <linux/limits.h>
47 #include <linux/dcache.h>
48 #include <linux/dnotify.h>
49 #include <linux/syscalls.h>
50 #include <linux/vmstat.h>
51 #include <linux/nfs_fs.h>
52 #include <linux/acpi.h>
53 #include <linux/reboot.h>
54 #include <linux/ftrace.h>
55 #include <linux/perf_event.h>
56 #include <linux/kprobes.h>
57 #include <linux/pipe_fs_i.h>
58 #include <linux/oom.h>
59 #include <linux/kmod.h>
60 
61 #include <asm/uaccess.h>
62 #include <asm/processor.h>
63 
64 #ifdef CONFIG_X86
65 #include <asm/nmi.h>
66 #include <asm/stacktrace.h>
67 #include <asm/io.h>
68 #endif
69 #ifdef CONFIG_BSD_PROCESS_ACCT
70 #include <linux/acct.h>
71 #endif
72 #ifdef CONFIG_RT_MUTEXES
73 #include <linux/rtmutex.h>
74 #endif
75 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
76 #include <linux/lockdep.h>
77 #endif
78 #ifdef CONFIG_CHR_DEV_SG
79 #include <scsi/sg.h>
80 #endif
81 
82 #ifdef CONFIG_LOCKUP_DETECTOR
83 #include <linux/nmi.h>
84 #endif
85 
86 
87 #if defined(CONFIG_SYSCTL)
88 
89 /* External variables not in a header file. */
90 extern int sysctl_overcommit_memory;
91 extern int sysctl_overcommit_ratio;
92 extern int max_threads;
93 extern int core_uses_pid;
94 extern int suid_dumpable;
95 extern char core_pattern[];
96 extern unsigned int core_pipe_limit;
97 extern int pid_max;
98 extern int min_free_kbytes;
99 extern int pid_max_min, pid_max_max;
100 extern int sysctl_drop_caches;
101 extern int percpu_pagelist_fraction;
102 extern int compat_log;
103 extern int latencytop_enabled;
104 extern int sysctl_nr_open_min, sysctl_nr_open_max;
105 #ifndef CONFIG_MMU
106 extern int sysctl_nr_trim_pages;
107 #endif
108 #ifdef CONFIG_BLOCK
109 extern int blk_iopoll_enabled;
110 #endif
111 
112 /* Constants used for minimum and  maximum */
113 #ifdef CONFIG_LOCKUP_DETECTOR
114 static int sixty = 60;
115 static int neg_one = -1;
116 #endif
117 
118 static int zero;
119 static int __maybe_unused one = 1;
120 static int __maybe_unused two = 2;
121 static int __maybe_unused three = 3;
122 static unsigned long one_ul = 1;
123 static int one_hundred = 100;
124 #ifdef CONFIG_PRINTK
125 static int ten_thousand = 10000;
126 #endif
127 
128 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
129 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
130 
131 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
132 static int maxolduid = 65535;
133 static int minolduid;
134 static int min_percpu_pagelist_fract = 8;
135 
136 static int ngroups_max = NGROUPS_MAX;
137 
138 #ifdef CONFIG_INOTIFY_USER
139 #include <linux/inotify.h>
140 #endif
141 #ifdef CONFIG_SPARC
142 #include <asm/system.h>
143 #endif
144 
145 #ifdef CONFIG_SPARC64
146 extern int sysctl_tsb_ratio;
147 #endif
148 
149 #ifdef __hppa__
150 extern int pwrsw_enabled;
151 extern int unaligned_enabled;
152 #endif
153 
154 #ifdef CONFIG_S390
155 #ifdef CONFIG_MATHEMU
156 extern int sysctl_ieee_emulation_warnings;
157 #endif
158 extern int sysctl_userprocess_debug;
159 extern int spin_retry;
160 #endif
161 
162 #ifdef CONFIG_IA64
163 extern int no_unaligned_warning;
164 extern int unaligned_dump_stack;
165 #endif
166 
167 #ifdef CONFIG_PROC_SYSCTL
168 static int proc_do_cad_pid(struct ctl_table *table, int write,
169 		  void __user *buffer, size_t *lenp, loff_t *ppos);
170 static int proc_taint(struct ctl_table *table, int write,
171 			       void __user *buffer, size_t *lenp, loff_t *ppos);
172 #endif
173 
174 #ifdef CONFIG_PRINTK
175 static int proc_dmesg_restrict(struct ctl_table *table, int write,
176 				void __user *buffer, size_t *lenp, loff_t *ppos);
177 #endif
178 
179 #ifdef CONFIG_MAGIC_SYSRQ
180 /* Note: sysrq code uses it's own private copy */
181 static int __sysrq_enabled = SYSRQ_DEFAULT_ENABLE;
182 
183 static int sysrq_sysctl_handler(ctl_table *table, int write,
184 				void __user *buffer, size_t *lenp,
185 				loff_t *ppos)
186 {
187 	int error;
188 
189 	error = proc_dointvec(table, write, buffer, lenp, ppos);
190 	if (error)
191 		return error;
192 
193 	if (write)
194 		sysrq_toggle_support(__sysrq_enabled);
195 
196 	return 0;
197 }
198 
199 #endif
200 
201 static struct ctl_table root_table[];
202 static struct ctl_table_root sysctl_table_root;
203 static struct ctl_table_header root_table_header = {
204 	{{.count = 1,
205 	.ctl_table = root_table,
206 	.ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),}},
207 	.root = &sysctl_table_root,
208 	.set = &sysctl_table_root.default_set,
209 };
210 static struct ctl_table_root sysctl_table_root = {
211 	.root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
212 	.default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
213 };
214 
215 static struct ctl_table kern_table[];
216 static struct ctl_table vm_table[];
217 static struct ctl_table fs_table[];
218 static struct ctl_table debug_table[];
219 static struct ctl_table dev_table[];
220 extern struct ctl_table random_table[];
221 #ifdef CONFIG_EPOLL
222 extern struct ctl_table epoll_table[];
223 #endif
224 
225 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
226 int sysctl_legacy_va_layout;
227 #endif
228 
229 /* The default sysctl tables: */
230 
231 static struct ctl_table root_table[] = {
232 	{
233 		.procname	= "kernel",
234 		.mode		= 0555,
235 		.child		= kern_table,
236 	},
237 	{
238 		.procname	= "vm",
239 		.mode		= 0555,
240 		.child		= vm_table,
241 	},
242 	{
243 		.procname	= "fs",
244 		.mode		= 0555,
245 		.child		= fs_table,
246 	},
247 	{
248 		.procname	= "debug",
249 		.mode		= 0555,
250 		.child		= debug_table,
251 	},
252 	{
253 		.procname	= "dev",
254 		.mode		= 0555,
255 		.child		= dev_table,
256 	},
257 	{ }
258 };
259 
260 #ifdef CONFIG_SCHED_DEBUG
261 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
262 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
263 static int min_wakeup_granularity_ns;			/* 0 usecs */
264 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
265 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
266 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
267 #endif
268 
269 #ifdef CONFIG_COMPACTION
270 static int min_extfrag_threshold;
271 static int max_extfrag_threshold = 1000;
272 #endif
273 
274 static struct ctl_table kern_table[] = {
275 	{
276 		.procname	= "sched_child_runs_first",
277 		.data		= &sysctl_sched_child_runs_first,
278 		.maxlen		= sizeof(unsigned int),
279 		.mode		= 0644,
280 		.proc_handler	= proc_dointvec,
281 	},
282 #ifdef CONFIG_SCHED_DEBUG
283 	{
284 		.procname	= "sched_min_granularity_ns",
285 		.data		= &sysctl_sched_min_granularity,
286 		.maxlen		= sizeof(unsigned int),
287 		.mode		= 0644,
288 		.proc_handler	= sched_proc_update_handler,
289 		.extra1		= &min_sched_granularity_ns,
290 		.extra2		= &max_sched_granularity_ns,
291 	},
292 	{
293 		.procname	= "sched_latency_ns",
294 		.data		= &sysctl_sched_latency,
295 		.maxlen		= sizeof(unsigned int),
296 		.mode		= 0644,
297 		.proc_handler	= sched_proc_update_handler,
298 		.extra1		= &min_sched_granularity_ns,
299 		.extra2		= &max_sched_granularity_ns,
300 	},
301 	{
302 		.procname	= "sched_wakeup_granularity_ns",
303 		.data		= &sysctl_sched_wakeup_granularity,
304 		.maxlen		= sizeof(unsigned int),
305 		.mode		= 0644,
306 		.proc_handler	= sched_proc_update_handler,
307 		.extra1		= &min_wakeup_granularity_ns,
308 		.extra2		= &max_wakeup_granularity_ns,
309 	},
310 	{
311 		.procname	= "sched_tunable_scaling",
312 		.data		= &sysctl_sched_tunable_scaling,
313 		.maxlen		= sizeof(enum sched_tunable_scaling),
314 		.mode		= 0644,
315 		.proc_handler	= sched_proc_update_handler,
316 		.extra1		= &min_sched_tunable_scaling,
317 		.extra2		= &max_sched_tunable_scaling,
318 	},
319 	{
320 		.procname	= "sched_migration_cost",
321 		.data		= &sysctl_sched_migration_cost,
322 		.maxlen		= sizeof(unsigned int),
323 		.mode		= 0644,
324 		.proc_handler	= proc_dointvec,
325 	},
326 	{
327 		.procname	= "sched_nr_migrate",
328 		.data		= &sysctl_sched_nr_migrate,
329 		.maxlen		= sizeof(unsigned int),
330 		.mode		= 0644,
331 		.proc_handler	= proc_dointvec,
332 	},
333 	{
334 		.procname	= "sched_time_avg",
335 		.data		= &sysctl_sched_time_avg,
336 		.maxlen		= sizeof(unsigned int),
337 		.mode		= 0644,
338 		.proc_handler	= proc_dointvec,
339 	},
340 	{
341 		.procname	= "sched_shares_window",
342 		.data		= &sysctl_sched_shares_window,
343 		.maxlen		= sizeof(unsigned int),
344 		.mode		= 0644,
345 		.proc_handler	= proc_dointvec,
346 	},
347 	{
348 		.procname	= "timer_migration",
349 		.data		= &sysctl_timer_migration,
350 		.maxlen		= sizeof(unsigned int),
351 		.mode		= 0644,
352 		.proc_handler	= proc_dointvec_minmax,
353 		.extra1		= &zero,
354 		.extra2		= &one,
355 	},
356 #endif
357 	{
358 		.procname	= "sched_rt_period_us",
359 		.data		= &sysctl_sched_rt_period,
360 		.maxlen		= sizeof(unsigned int),
361 		.mode		= 0644,
362 		.proc_handler	= sched_rt_handler,
363 	},
364 	{
365 		.procname	= "sched_rt_runtime_us",
366 		.data		= &sysctl_sched_rt_runtime,
367 		.maxlen		= sizeof(int),
368 		.mode		= 0644,
369 		.proc_handler	= sched_rt_handler,
370 	},
371 #ifdef CONFIG_SCHED_AUTOGROUP
372 	{
373 		.procname	= "sched_autogroup_enabled",
374 		.data		= &sysctl_sched_autogroup_enabled,
375 		.maxlen		= sizeof(unsigned int),
376 		.mode		= 0644,
377 		.proc_handler	= proc_dointvec_minmax,
378 		.extra1		= &zero,
379 		.extra2		= &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,
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,
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_dmesg_restrict,
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 #if defined(CONFIG_LOCKUP_DETECTOR)
734 	{
735 		.procname       = "watchdog",
736 		.data           = &watchdog_enabled,
737 		.maxlen         = sizeof (int),
738 		.mode           = 0644,
739 		.proc_handler   = proc_dowatchdog,
740 		.extra1		= &zero,
741 		.extra2		= &one,
742 	},
743 	{
744 		.procname	= "watchdog_thresh",
745 		.data		= &watchdog_thresh,
746 		.maxlen		= sizeof(int),
747 		.mode		= 0644,
748 		.proc_handler	= proc_dowatchdog,
749 		.extra1		= &neg_one,
750 		.extra2		= &sixty,
751 	},
752 	{
753 		.procname	= "softlockup_panic",
754 		.data		= &softlockup_panic,
755 		.maxlen		= sizeof(int),
756 		.mode		= 0644,
757 		.proc_handler	= proc_dointvec_minmax,
758 		.extra1		= &zero,
759 		.extra2		= &one,
760 	},
761 	{
762 		.procname       = "nmi_watchdog",
763 		.data           = &watchdog_enabled,
764 		.maxlen         = sizeof (int),
765 		.mode           = 0644,
766 		.proc_handler   = proc_dowatchdog,
767 		.extra1		= &zero,
768 		.extra2		= &one,
769 	},
770 #endif
771 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
772 	{
773 		.procname       = "unknown_nmi_panic",
774 		.data           = &unknown_nmi_panic,
775 		.maxlen         = sizeof (int),
776 		.mode           = 0644,
777 		.proc_handler   = proc_dointvec,
778 	},
779 #endif
780 #if defined(CONFIG_X86)
781 	{
782 		.procname	= "panic_on_unrecovered_nmi",
783 		.data		= &panic_on_unrecovered_nmi,
784 		.maxlen		= sizeof(int),
785 		.mode		= 0644,
786 		.proc_handler	= proc_dointvec,
787 	},
788 	{
789 		.procname	= "panic_on_io_nmi",
790 		.data		= &panic_on_io_nmi,
791 		.maxlen		= sizeof(int),
792 		.mode		= 0644,
793 		.proc_handler	= proc_dointvec,
794 	},
795 	{
796 		.procname	= "bootloader_type",
797 		.data		= &bootloader_type,
798 		.maxlen		= sizeof (int),
799 		.mode		= 0444,
800 		.proc_handler	= proc_dointvec,
801 	},
802 	{
803 		.procname	= "bootloader_version",
804 		.data		= &bootloader_version,
805 		.maxlen		= sizeof (int),
806 		.mode		= 0444,
807 		.proc_handler	= proc_dointvec,
808 	},
809 	{
810 		.procname	= "kstack_depth_to_print",
811 		.data		= &kstack_depth_to_print,
812 		.maxlen		= sizeof(int),
813 		.mode		= 0644,
814 		.proc_handler	= proc_dointvec,
815 	},
816 	{
817 		.procname	= "io_delay_type",
818 		.data		= &io_delay_type,
819 		.maxlen		= sizeof(int),
820 		.mode		= 0644,
821 		.proc_handler	= proc_dointvec,
822 	},
823 #endif
824 #if defined(CONFIG_MMU)
825 	{
826 		.procname	= "randomize_va_space",
827 		.data		= &randomize_va_space,
828 		.maxlen		= sizeof(int),
829 		.mode		= 0644,
830 		.proc_handler	= proc_dointvec,
831 	},
832 #endif
833 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
834 	{
835 		.procname	= "spin_retry",
836 		.data		= &spin_retry,
837 		.maxlen		= sizeof (int),
838 		.mode		= 0644,
839 		.proc_handler	= proc_dointvec,
840 	},
841 #endif
842 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
843 	{
844 		.procname	= "acpi_video_flags",
845 		.data		= &acpi_realmode_flags,
846 		.maxlen		= sizeof (unsigned long),
847 		.mode		= 0644,
848 		.proc_handler	= proc_doulongvec_minmax,
849 	},
850 #endif
851 #ifdef CONFIG_IA64
852 	{
853 		.procname	= "ignore-unaligned-usertrap",
854 		.data		= &no_unaligned_warning,
855 		.maxlen		= sizeof (int),
856 	 	.mode		= 0644,
857 		.proc_handler	= proc_dointvec,
858 	},
859 	{
860 		.procname	= "unaligned-dump-stack",
861 		.data		= &unaligned_dump_stack,
862 		.maxlen		= sizeof (int),
863 		.mode		= 0644,
864 		.proc_handler	= proc_dointvec,
865 	},
866 #endif
867 #ifdef CONFIG_DETECT_HUNG_TASK
868 	{
869 		.procname	= "hung_task_panic",
870 		.data		= &sysctl_hung_task_panic,
871 		.maxlen		= sizeof(int),
872 		.mode		= 0644,
873 		.proc_handler	= proc_dointvec_minmax,
874 		.extra1		= &zero,
875 		.extra2		= &one,
876 	},
877 	{
878 		.procname	= "hung_task_check_count",
879 		.data		= &sysctl_hung_task_check_count,
880 		.maxlen		= sizeof(unsigned long),
881 		.mode		= 0644,
882 		.proc_handler	= proc_doulongvec_minmax,
883 	},
884 	{
885 		.procname	= "hung_task_timeout_secs",
886 		.data		= &sysctl_hung_task_timeout_secs,
887 		.maxlen		= sizeof(unsigned long),
888 		.mode		= 0644,
889 		.proc_handler	= proc_dohung_task_timeout_secs,
890 	},
891 	{
892 		.procname	= "hung_task_warnings",
893 		.data		= &sysctl_hung_task_warnings,
894 		.maxlen		= sizeof(unsigned long),
895 		.mode		= 0644,
896 		.proc_handler	= proc_doulongvec_minmax,
897 	},
898 #endif
899 #ifdef CONFIG_COMPAT
900 	{
901 		.procname	= "compat-log",
902 		.data		= &compat_log,
903 		.maxlen		= sizeof (int),
904 	 	.mode		= 0644,
905 		.proc_handler	= proc_dointvec,
906 	},
907 #endif
908 #ifdef CONFIG_RT_MUTEXES
909 	{
910 		.procname	= "max_lock_depth",
911 		.data		= &max_lock_depth,
912 		.maxlen		= sizeof(int),
913 		.mode		= 0644,
914 		.proc_handler	= proc_dointvec,
915 	},
916 #endif
917 	{
918 		.procname	= "poweroff_cmd",
919 		.data		= &poweroff_cmd,
920 		.maxlen		= POWEROFF_CMD_PATH_LEN,
921 		.mode		= 0644,
922 		.proc_handler	= proc_dostring,
923 	},
924 #ifdef CONFIG_KEYS
925 	{
926 		.procname	= "keys",
927 		.mode		= 0555,
928 		.child		= key_sysctls,
929 	},
930 #endif
931 #ifdef CONFIG_RCU_TORTURE_TEST
932 	{
933 		.procname       = "rcutorture_runnable",
934 		.data           = &rcutorture_runnable,
935 		.maxlen         = sizeof(int),
936 		.mode           = 0644,
937 		.proc_handler	= proc_dointvec,
938 	},
939 #endif
940 #ifdef CONFIG_PERF_EVENTS
941 	{
942 		.procname	= "perf_event_paranoid",
943 		.data		= &sysctl_perf_event_paranoid,
944 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
945 		.mode		= 0644,
946 		.proc_handler	= proc_dointvec,
947 	},
948 	{
949 		.procname	= "perf_event_mlock_kb",
950 		.data		= &sysctl_perf_event_mlock,
951 		.maxlen		= sizeof(sysctl_perf_event_mlock),
952 		.mode		= 0644,
953 		.proc_handler	= proc_dointvec,
954 	},
955 	{
956 		.procname	= "perf_event_max_sample_rate",
957 		.data		= &sysctl_perf_event_sample_rate,
958 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
959 		.mode		= 0644,
960 		.proc_handler	= perf_proc_update_handler,
961 	},
962 #endif
963 #ifdef CONFIG_KMEMCHECK
964 	{
965 		.procname	= "kmemcheck",
966 		.data		= &kmemcheck_enabled,
967 		.maxlen		= sizeof(int),
968 		.mode		= 0644,
969 		.proc_handler	= proc_dointvec,
970 	},
971 #endif
972 #ifdef CONFIG_BLOCK
973 	{
974 		.procname	= "blk_iopoll",
975 		.data		= &blk_iopoll_enabled,
976 		.maxlen		= sizeof(int),
977 		.mode		= 0644,
978 		.proc_handler	= proc_dointvec,
979 	},
980 #endif
981 	{ }
982 };
983 
984 static struct ctl_table vm_table[] = {
985 	{
986 		.procname	= "overcommit_memory",
987 		.data		= &sysctl_overcommit_memory,
988 		.maxlen		= sizeof(sysctl_overcommit_memory),
989 		.mode		= 0644,
990 		.proc_handler	= proc_dointvec_minmax,
991 		.extra1		= &zero,
992 		.extra2		= &two,
993 	},
994 	{
995 		.procname	= "panic_on_oom",
996 		.data		= &sysctl_panic_on_oom,
997 		.maxlen		= sizeof(sysctl_panic_on_oom),
998 		.mode		= 0644,
999 		.proc_handler	= proc_dointvec_minmax,
1000 		.extra1		= &zero,
1001 		.extra2		= &two,
1002 	},
1003 	{
1004 		.procname	= "oom_kill_allocating_task",
1005 		.data		= &sysctl_oom_kill_allocating_task,
1006 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1007 		.mode		= 0644,
1008 		.proc_handler	= proc_dointvec,
1009 	},
1010 	{
1011 		.procname	= "oom_dump_tasks",
1012 		.data		= &sysctl_oom_dump_tasks,
1013 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1014 		.mode		= 0644,
1015 		.proc_handler	= proc_dointvec,
1016 	},
1017 	{
1018 		.procname	= "overcommit_ratio",
1019 		.data		= &sysctl_overcommit_ratio,
1020 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1021 		.mode		= 0644,
1022 		.proc_handler	= proc_dointvec,
1023 	},
1024 	{
1025 		.procname	= "page-cluster",
1026 		.data		= &page_cluster,
1027 		.maxlen		= sizeof(int),
1028 		.mode		= 0644,
1029 		.proc_handler	= proc_dointvec_minmax,
1030 		.extra1		= &zero,
1031 	},
1032 	{
1033 		.procname	= "dirty_background_ratio",
1034 		.data		= &dirty_background_ratio,
1035 		.maxlen		= sizeof(dirty_background_ratio),
1036 		.mode		= 0644,
1037 		.proc_handler	= dirty_background_ratio_handler,
1038 		.extra1		= &zero,
1039 		.extra2		= &one_hundred,
1040 	},
1041 	{
1042 		.procname	= "dirty_background_bytes",
1043 		.data		= &dirty_background_bytes,
1044 		.maxlen		= sizeof(dirty_background_bytes),
1045 		.mode		= 0644,
1046 		.proc_handler	= dirty_background_bytes_handler,
1047 		.extra1		= &one_ul,
1048 	},
1049 	{
1050 		.procname	= "dirty_ratio",
1051 		.data		= &vm_dirty_ratio,
1052 		.maxlen		= sizeof(vm_dirty_ratio),
1053 		.mode		= 0644,
1054 		.proc_handler	= dirty_ratio_handler,
1055 		.extra1		= &zero,
1056 		.extra2		= &one_hundred,
1057 	},
1058 	{
1059 		.procname	= "dirty_bytes",
1060 		.data		= &vm_dirty_bytes,
1061 		.maxlen		= sizeof(vm_dirty_bytes),
1062 		.mode		= 0644,
1063 		.proc_handler	= dirty_bytes_handler,
1064 		.extra1		= &dirty_bytes_min,
1065 	},
1066 	{
1067 		.procname	= "dirty_writeback_centisecs",
1068 		.data		= &dirty_writeback_interval,
1069 		.maxlen		= sizeof(dirty_writeback_interval),
1070 		.mode		= 0644,
1071 		.proc_handler	= dirty_writeback_centisecs_handler,
1072 	},
1073 	{
1074 		.procname	= "dirty_expire_centisecs",
1075 		.data		= &dirty_expire_interval,
1076 		.maxlen		= sizeof(dirty_expire_interval),
1077 		.mode		= 0644,
1078 		.proc_handler	= proc_dointvec_minmax,
1079 		.extra1		= &zero,
1080 	},
1081 	{
1082 		.procname	= "nr_pdflush_threads",
1083 		.data		= &nr_pdflush_threads,
1084 		.maxlen		= sizeof nr_pdflush_threads,
1085 		.mode		= 0444 /* read-only*/,
1086 		.proc_handler	= proc_dointvec,
1087 	},
1088 	{
1089 		.procname	= "swappiness",
1090 		.data		= &vm_swappiness,
1091 		.maxlen		= sizeof(vm_swappiness),
1092 		.mode		= 0644,
1093 		.proc_handler	= proc_dointvec_minmax,
1094 		.extra1		= &zero,
1095 		.extra2		= &one_hundred,
1096 	},
1097 #ifdef CONFIG_HUGETLB_PAGE
1098 	{
1099 		.procname	= "nr_hugepages",
1100 		.data		= NULL,
1101 		.maxlen		= sizeof(unsigned long),
1102 		.mode		= 0644,
1103 		.proc_handler	= hugetlb_sysctl_handler,
1104 		.extra1		= (void *)&hugetlb_zero,
1105 		.extra2		= (void *)&hugetlb_infinity,
1106 	},
1107 #ifdef CONFIG_NUMA
1108 	{
1109 		.procname       = "nr_hugepages_mempolicy",
1110 		.data           = NULL,
1111 		.maxlen         = sizeof(unsigned long),
1112 		.mode           = 0644,
1113 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1114 		.extra1		= (void *)&hugetlb_zero,
1115 		.extra2		= (void *)&hugetlb_infinity,
1116 	},
1117 #endif
1118 	 {
1119 		.procname	= "hugetlb_shm_group",
1120 		.data		= &sysctl_hugetlb_shm_group,
1121 		.maxlen		= sizeof(gid_t),
1122 		.mode		= 0644,
1123 		.proc_handler	= proc_dointvec,
1124 	 },
1125 	 {
1126 		.procname	= "hugepages_treat_as_movable",
1127 		.data		= &hugepages_treat_as_movable,
1128 		.maxlen		= sizeof(int),
1129 		.mode		= 0644,
1130 		.proc_handler	= hugetlb_treat_movable_handler,
1131 	},
1132 	{
1133 		.procname	= "nr_overcommit_hugepages",
1134 		.data		= NULL,
1135 		.maxlen		= sizeof(unsigned long),
1136 		.mode		= 0644,
1137 		.proc_handler	= hugetlb_overcommit_handler,
1138 		.extra1		= (void *)&hugetlb_zero,
1139 		.extra2		= (void *)&hugetlb_infinity,
1140 	},
1141 #endif
1142 	{
1143 		.procname	= "lowmem_reserve_ratio",
1144 		.data		= &sysctl_lowmem_reserve_ratio,
1145 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1146 		.mode		= 0644,
1147 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1148 	},
1149 	{
1150 		.procname	= "drop_caches",
1151 		.data		= &sysctl_drop_caches,
1152 		.maxlen		= sizeof(int),
1153 		.mode		= 0644,
1154 		.proc_handler	= drop_caches_sysctl_handler,
1155 		.extra1		= &one,
1156 		.extra2		= &three,
1157 	},
1158 #ifdef CONFIG_COMPACTION
1159 	{
1160 		.procname	= "compact_memory",
1161 		.data		= &sysctl_compact_memory,
1162 		.maxlen		= sizeof(int),
1163 		.mode		= 0200,
1164 		.proc_handler	= sysctl_compaction_handler,
1165 	},
1166 	{
1167 		.procname	= "extfrag_threshold",
1168 		.data		= &sysctl_extfrag_threshold,
1169 		.maxlen		= sizeof(int),
1170 		.mode		= 0644,
1171 		.proc_handler	= sysctl_extfrag_handler,
1172 		.extra1		= &min_extfrag_threshold,
1173 		.extra2		= &max_extfrag_threshold,
1174 	},
1175 
1176 #endif /* CONFIG_COMPACTION */
1177 	{
1178 		.procname	= "min_free_kbytes",
1179 		.data		= &min_free_kbytes,
1180 		.maxlen		= sizeof(min_free_kbytes),
1181 		.mode		= 0644,
1182 		.proc_handler	= min_free_kbytes_sysctl_handler,
1183 		.extra1		= &zero,
1184 	},
1185 	{
1186 		.procname	= "percpu_pagelist_fraction",
1187 		.data		= &percpu_pagelist_fraction,
1188 		.maxlen		= sizeof(percpu_pagelist_fraction),
1189 		.mode		= 0644,
1190 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1191 		.extra1		= &min_percpu_pagelist_fract,
1192 	},
1193 #ifdef CONFIG_MMU
1194 	{
1195 		.procname	= "max_map_count",
1196 		.data		= &sysctl_max_map_count,
1197 		.maxlen		= sizeof(sysctl_max_map_count),
1198 		.mode		= 0644,
1199 		.proc_handler	= proc_dointvec_minmax,
1200 		.extra1		= &zero,
1201 	},
1202 #else
1203 	{
1204 		.procname	= "nr_trim_pages",
1205 		.data		= &sysctl_nr_trim_pages,
1206 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1207 		.mode		= 0644,
1208 		.proc_handler	= proc_dointvec_minmax,
1209 		.extra1		= &zero,
1210 	},
1211 #endif
1212 	{
1213 		.procname	= "laptop_mode",
1214 		.data		= &laptop_mode,
1215 		.maxlen		= sizeof(laptop_mode),
1216 		.mode		= 0644,
1217 		.proc_handler	= proc_dointvec_jiffies,
1218 	},
1219 	{
1220 		.procname	= "block_dump",
1221 		.data		= &block_dump,
1222 		.maxlen		= sizeof(block_dump),
1223 		.mode		= 0644,
1224 		.proc_handler	= proc_dointvec,
1225 		.extra1		= &zero,
1226 	},
1227 	{
1228 		.procname	= "vfs_cache_pressure",
1229 		.data		= &sysctl_vfs_cache_pressure,
1230 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1231 		.mode		= 0644,
1232 		.proc_handler	= proc_dointvec,
1233 		.extra1		= &zero,
1234 	},
1235 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1236 	{
1237 		.procname	= "legacy_va_layout",
1238 		.data		= &sysctl_legacy_va_layout,
1239 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1240 		.mode		= 0644,
1241 		.proc_handler	= proc_dointvec,
1242 		.extra1		= &zero,
1243 	},
1244 #endif
1245 #ifdef CONFIG_NUMA
1246 	{
1247 		.procname	= "zone_reclaim_mode",
1248 		.data		= &zone_reclaim_mode,
1249 		.maxlen		= sizeof(zone_reclaim_mode),
1250 		.mode		= 0644,
1251 		.proc_handler	= proc_dointvec,
1252 		.extra1		= &zero,
1253 	},
1254 	{
1255 		.procname	= "min_unmapped_ratio",
1256 		.data		= &sysctl_min_unmapped_ratio,
1257 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1258 		.mode		= 0644,
1259 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1260 		.extra1		= &zero,
1261 		.extra2		= &one_hundred,
1262 	},
1263 	{
1264 		.procname	= "min_slab_ratio",
1265 		.data		= &sysctl_min_slab_ratio,
1266 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1267 		.mode		= 0644,
1268 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1269 		.extra1		= &zero,
1270 		.extra2		= &one_hundred,
1271 	},
1272 #endif
1273 #ifdef CONFIG_SMP
1274 	{
1275 		.procname	= "stat_interval",
1276 		.data		= &sysctl_stat_interval,
1277 		.maxlen		= sizeof(sysctl_stat_interval),
1278 		.mode		= 0644,
1279 		.proc_handler	= proc_dointvec_jiffies,
1280 	},
1281 #endif
1282 #ifdef CONFIG_MMU
1283 	{
1284 		.procname	= "mmap_min_addr",
1285 		.data		= &dac_mmap_min_addr,
1286 		.maxlen		= sizeof(unsigned long),
1287 		.mode		= 0644,
1288 		.proc_handler	= mmap_min_addr_handler,
1289 	},
1290 #endif
1291 #ifdef CONFIG_NUMA
1292 	{
1293 		.procname	= "numa_zonelist_order",
1294 		.data		= &numa_zonelist_order,
1295 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1296 		.mode		= 0644,
1297 		.proc_handler	= numa_zonelist_order_handler,
1298 	},
1299 #endif
1300 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1301    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1302 	{
1303 		.procname	= "vdso_enabled",
1304 		.data		= &vdso_enabled,
1305 		.maxlen		= sizeof(vdso_enabled),
1306 		.mode		= 0644,
1307 		.proc_handler	= proc_dointvec,
1308 		.extra1		= &zero,
1309 	},
1310 #endif
1311 #ifdef CONFIG_HIGHMEM
1312 	{
1313 		.procname	= "highmem_is_dirtyable",
1314 		.data		= &vm_highmem_is_dirtyable,
1315 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1316 		.mode		= 0644,
1317 		.proc_handler	= proc_dointvec_minmax,
1318 		.extra1		= &zero,
1319 		.extra2		= &one,
1320 	},
1321 #endif
1322 	{
1323 		.procname	= "scan_unevictable_pages",
1324 		.data		= &scan_unevictable_pages,
1325 		.maxlen		= sizeof(scan_unevictable_pages),
1326 		.mode		= 0644,
1327 		.proc_handler	= scan_unevictable_handler,
1328 	},
1329 #ifdef CONFIG_MEMORY_FAILURE
1330 	{
1331 		.procname	= "memory_failure_early_kill",
1332 		.data		= &sysctl_memory_failure_early_kill,
1333 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1334 		.mode		= 0644,
1335 		.proc_handler	= proc_dointvec_minmax,
1336 		.extra1		= &zero,
1337 		.extra2		= &one,
1338 	},
1339 	{
1340 		.procname	= "memory_failure_recovery",
1341 		.data		= &sysctl_memory_failure_recovery,
1342 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1343 		.mode		= 0644,
1344 		.proc_handler	= proc_dointvec_minmax,
1345 		.extra1		= &zero,
1346 		.extra2		= &one,
1347 	},
1348 #endif
1349 	{ }
1350 };
1351 
1352 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1353 static struct ctl_table binfmt_misc_table[] = {
1354 	{ }
1355 };
1356 #endif
1357 
1358 static struct ctl_table fs_table[] = {
1359 	{
1360 		.procname	= "inode-nr",
1361 		.data		= &inodes_stat,
1362 		.maxlen		= 2*sizeof(int),
1363 		.mode		= 0444,
1364 		.proc_handler	= proc_nr_inodes,
1365 	},
1366 	{
1367 		.procname	= "inode-state",
1368 		.data		= &inodes_stat,
1369 		.maxlen		= 7*sizeof(int),
1370 		.mode		= 0444,
1371 		.proc_handler	= proc_nr_inodes,
1372 	},
1373 	{
1374 		.procname	= "file-nr",
1375 		.data		= &files_stat,
1376 		.maxlen		= sizeof(files_stat),
1377 		.mode		= 0444,
1378 		.proc_handler	= proc_nr_files,
1379 	},
1380 	{
1381 		.procname	= "file-max",
1382 		.data		= &files_stat.max_files,
1383 		.maxlen		= sizeof(files_stat.max_files),
1384 		.mode		= 0644,
1385 		.proc_handler	= proc_doulongvec_minmax,
1386 	},
1387 	{
1388 		.procname	= "nr_open",
1389 		.data		= &sysctl_nr_open,
1390 		.maxlen		= sizeof(int),
1391 		.mode		= 0644,
1392 		.proc_handler	= proc_dointvec_minmax,
1393 		.extra1		= &sysctl_nr_open_min,
1394 		.extra2		= &sysctl_nr_open_max,
1395 	},
1396 	{
1397 		.procname	= "dentry-state",
1398 		.data		= &dentry_stat,
1399 		.maxlen		= 6*sizeof(int),
1400 		.mode		= 0444,
1401 		.proc_handler	= proc_nr_dentry,
1402 	},
1403 	{
1404 		.procname	= "overflowuid",
1405 		.data		= &fs_overflowuid,
1406 		.maxlen		= sizeof(int),
1407 		.mode		= 0644,
1408 		.proc_handler	= proc_dointvec_minmax,
1409 		.extra1		= &minolduid,
1410 		.extra2		= &maxolduid,
1411 	},
1412 	{
1413 		.procname	= "overflowgid",
1414 		.data		= &fs_overflowgid,
1415 		.maxlen		= sizeof(int),
1416 		.mode		= 0644,
1417 		.proc_handler	= proc_dointvec_minmax,
1418 		.extra1		= &minolduid,
1419 		.extra2		= &maxolduid,
1420 	},
1421 #ifdef CONFIG_FILE_LOCKING
1422 	{
1423 		.procname	= "leases-enable",
1424 		.data		= &leases_enable,
1425 		.maxlen		= sizeof(int),
1426 		.mode		= 0644,
1427 		.proc_handler	= proc_dointvec,
1428 	},
1429 #endif
1430 #ifdef CONFIG_DNOTIFY
1431 	{
1432 		.procname	= "dir-notify-enable",
1433 		.data		= &dir_notify_enable,
1434 		.maxlen		= sizeof(int),
1435 		.mode		= 0644,
1436 		.proc_handler	= proc_dointvec,
1437 	},
1438 #endif
1439 #ifdef CONFIG_MMU
1440 #ifdef CONFIG_FILE_LOCKING
1441 	{
1442 		.procname	= "lease-break-time",
1443 		.data		= &lease_break_time,
1444 		.maxlen		= sizeof(int),
1445 		.mode		= 0644,
1446 		.proc_handler	= proc_dointvec,
1447 	},
1448 #endif
1449 #ifdef CONFIG_AIO
1450 	{
1451 		.procname	= "aio-nr",
1452 		.data		= &aio_nr,
1453 		.maxlen		= sizeof(aio_nr),
1454 		.mode		= 0444,
1455 		.proc_handler	= proc_doulongvec_minmax,
1456 	},
1457 	{
1458 		.procname	= "aio-max-nr",
1459 		.data		= &aio_max_nr,
1460 		.maxlen		= sizeof(aio_max_nr),
1461 		.mode		= 0644,
1462 		.proc_handler	= proc_doulongvec_minmax,
1463 	},
1464 #endif /* CONFIG_AIO */
1465 #ifdef CONFIG_INOTIFY_USER
1466 	{
1467 		.procname	= "inotify",
1468 		.mode		= 0555,
1469 		.child		= inotify_table,
1470 	},
1471 #endif
1472 #ifdef CONFIG_EPOLL
1473 	{
1474 		.procname	= "epoll",
1475 		.mode		= 0555,
1476 		.child		= epoll_table,
1477 	},
1478 #endif
1479 #endif
1480 	{
1481 		.procname	= "suid_dumpable",
1482 		.data		= &suid_dumpable,
1483 		.maxlen		= sizeof(int),
1484 		.mode		= 0644,
1485 		.proc_handler	= proc_dointvec_minmax,
1486 		.extra1		= &zero,
1487 		.extra2		= &two,
1488 	},
1489 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1490 	{
1491 		.procname	= "binfmt_misc",
1492 		.mode		= 0555,
1493 		.child		= binfmt_misc_table,
1494 	},
1495 #endif
1496 	{
1497 		.procname	= "pipe-max-size",
1498 		.data		= &pipe_max_size,
1499 		.maxlen		= sizeof(int),
1500 		.mode		= 0644,
1501 		.proc_handler	= &pipe_proc_fn,
1502 		.extra1		= &pipe_min_size,
1503 	},
1504 	{ }
1505 };
1506 
1507 static struct ctl_table debug_table[] = {
1508 #if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
1509     defined(CONFIG_S390) || defined(CONFIG_TILE)
1510 	{
1511 		.procname	= "exception-trace",
1512 		.data		= &show_unhandled_signals,
1513 		.maxlen		= sizeof(int),
1514 		.mode		= 0644,
1515 		.proc_handler	= proc_dointvec
1516 	},
1517 #endif
1518 #if defined(CONFIG_OPTPROBES)
1519 	{
1520 		.procname	= "kprobes-optimization",
1521 		.data		= &sysctl_kprobes_optimization,
1522 		.maxlen		= sizeof(int),
1523 		.mode		= 0644,
1524 		.proc_handler	= proc_kprobes_optimization_handler,
1525 		.extra1		= &zero,
1526 		.extra2		= &one,
1527 	},
1528 #endif
1529 	{ }
1530 };
1531 
1532 static struct ctl_table dev_table[] = {
1533 	{ }
1534 };
1535 
1536 static DEFINE_SPINLOCK(sysctl_lock);
1537 
1538 /* called under sysctl_lock */
1539 static int use_table(struct ctl_table_header *p)
1540 {
1541 	if (unlikely(p->unregistering))
1542 		return 0;
1543 	p->used++;
1544 	return 1;
1545 }
1546 
1547 /* called under sysctl_lock */
1548 static void unuse_table(struct ctl_table_header *p)
1549 {
1550 	if (!--p->used)
1551 		if (unlikely(p->unregistering))
1552 			complete(p->unregistering);
1553 }
1554 
1555 /* called under sysctl_lock, will reacquire if has to wait */
1556 static void start_unregistering(struct ctl_table_header *p)
1557 {
1558 	/*
1559 	 * if p->used is 0, nobody will ever touch that entry again;
1560 	 * we'll eliminate all paths to it before dropping sysctl_lock
1561 	 */
1562 	if (unlikely(p->used)) {
1563 		struct completion wait;
1564 		init_completion(&wait);
1565 		p->unregistering = &wait;
1566 		spin_unlock(&sysctl_lock);
1567 		wait_for_completion(&wait);
1568 		spin_lock(&sysctl_lock);
1569 	} else {
1570 		/* anything non-NULL; we'll never dereference it */
1571 		p->unregistering = ERR_PTR(-EINVAL);
1572 	}
1573 	/*
1574 	 * do not remove from the list until nobody holds it; walking the
1575 	 * list in do_sysctl() relies on that.
1576 	 */
1577 	list_del_init(&p->ctl_entry);
1578 }
1579 
1580 void sysctl_head_get(struct ctl_table_header *head)
1581 {
1582 	spin_lock(&sysctl_lock);
1583 	head->count++;
1584 	spin_unlock(&sysctl_lock);
1585 }
1586 
1587 static void free_head(struct rcu_head *rcu)
1588 {
1589 	kfree(container_of(rcu, struct ctl_table_header, rcu));
1590 }
1591 
1592 void sysctl_head_put(struct ctl_table_header *head)
1593 {
1594 	spin_lock(&sysctl_lock);
1595 	if (!--head->count)
1596 		call_rcu(&head->rcu, free_head);
1597 	spin_unlock(&sysctl_lock);
1598 }
1599 
1600 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1601 {
1602 	if (!head)
1603 		BUG();
1604 	spin_lock(&sysctl_lock);
1605 	if (!use_table(head))
1606 		head = ERR_PTR(-ENOENT);
1607 	spin_unlock(&sysctl_lock);
1608 	return head;
1609 }
1610 
1611 void sysctl_head_finish(struct ctl_table_header *head)
1612 {
1613 	if (!head)
1614 		return;
1615 	spin_lock(&sysctl_lock);
1616 	unuse_table(head);
1617 	spin_unlock(&sysctl_lock);
1618 }
1619 
1620 static struct ctl_table_set *
1621 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1622 {
1623 	struct ctl_table_set *set = &root->default_set;
1624 	if (root->lookup)
1625 		set = root->lookup(root, namespaces);
1626 	return set;
1627 }
1628 
1629 static struct list_head *
1630 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1631 {
1632 	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1633 	return &set->list;
1634 }
1635 
1636 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1637 					    struct ctl_table_header *prev)
1638 {
1639 	struct ctl_table_root *root;
1640 	struct list_head *header_list;
1641 	struct ctl_table_header *head;
1642 	struct list_head *tmp;
1643 
1644 	spin_lock(&sysctl_lock);
1645 	if (prev) {
1646 		head = prev;
1647 		tmp = &prev->ctl_entry;
1648 		unuse_table(prev);
1649 		goto next;
1650 	}
1651 	tmp = &root_table_header.ctl_entry;
1652 	for (;;) {
1653 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1654 
1655 		if (!use_table(head))
1656 			goto next;
1657 		spin_unlock(&sysctl_lock);
1658 		return head;
1659 	next:
1660 		root = head->root;
1661 		tmp = tmp->next;
1662 		header_list = lookup_header_list(root, namespaces);
1663 		if (tmp != header_list)
1664 			continue;
1665 
1666 		do {
1667 			root = list_entry(root->root_list.next,
1668 					struct ctl_table_root, root_list);
1669 			if (root == &sysctl_table_root)
1670 				goto out;
1671 			header_list = lookup_header_list(root, namespaces);
1672 		} while (list_empty(header_list));
1673 		tmp = header_list->next;
1674 	}
1675 out:
1676 	spin_unlock(&sysctl_lock);
1677 	return NULL;
1678 }
1679 
1680 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1681 {
1682 	return __sysctl_head_next(current->nsproxy, prev);
1683 }
1684 
1685 void register_sysctl_root(struct ctl_table_root *root)
1686 {
1687 	spin_lock(&sysctl_lock);
1688 	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1689 	spin_unlock(&sysctl_lock);
1690 }
1691 
1692 /*
1693  * sysctl_perm does NOT grant the superuser all rights automatically, because
1694  * some sysctl variables are readonly even to root.
1695  */
1696 
1697 static int test_perm(int mode, int op)
1698 {
1699 	if (!current_euid())
1700 		mode >>= 6;
1701 	else if (in_egroup_p(0))
1702 		mode >>= 3;
1703 	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1704 		return 0;
1705 	return -EACCES;
1706 }
1707 
1708 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1709 {
1710 	int mode;
1711 
1712 	if (root->permissions)
1713 		mode = root->permissions(root, current->nsproxy, table);
1714 	else
1715 		mode = table->mode;
1716 
1717 	return test_perm(mode, op);
1718 }
1719 
1720 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1721 {
1722 	for (; table->procname; table++) {
1723 		table->parent = parent;
1724 		if (table->child)
1725 			sysctl_set_parent(table, table->child);
1726 	}
1727 }
1728 
1729 static __init int sysctl_init(void)
1730 {
1731 	sysctl_set_parent(NULL, root_table);
1732 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1733 	sysctl_check_table(current->nsproxy, root_table);
1734 #endif
1735 	return 0;
1736 }
1737 
1738 core_initcall(sysctl_init);
1739 
1740 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1741 				      struct ctl_table *table)
1742 {
1743 	struct ctl_table *p;
1744 	const char *s = branch->procname;
1745 
1746 	/* branch should have named subdirectory as its first element */
1747 	if (!s || !branch->child)
1748 		return NULL;
1749 
1750 	/* ... and nothing else */
1751 	if (branch[1].procname)
1752 		return NULL;
1753 
1754 	/* table should contain subdirectory with the same name */
1755 	for (p = table; p->procname; p++) {
1756 		if (!p->child)
1757 			continue;
1758 		if (p->procname && strcmp(p->procname, s) == 0)
1759 			return p;
1760 	}
1761 	return NULL;
1762 }
1763 
1764 /* see if attaching q to p would be an improvement */
1765 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1766 {
1767 	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1768 	struct ctl_table *next;
1769 	int is_better = 0;
1770 	int not_in_parent = !p->attached_by;
1771 
1772 	while ((next = is_branch_in(by, to)) != NULL) {
1773 		if (by == q->attached_by)
1774 			is_better = 1;
1775 		if (to == p->attached_by)
1776 			not_in_parent = 1;
1777 		by = by->child;
1778 		to = next->child;
1779 	}
1780 
1781 	if (is_better && not_in_parent) {
1782 		q->attached_by = by;
1783 		q->attached_to = to;
1784 		q->parent = p;
1785 	}
1786 }
1787 
1788 /**
1789  * __register_sysctl_paths - register a sysctl hierarchy
1790  * @root: List of sysctl headers to register on
1791  * @namespaces: Data to compute which lists of sysctl entries are visible
1792  * @path: The path to the directory the sysctl table is in.
1793  * @table: the top-level table structure
1794  *
1795  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1796  * array. A completely 0 filled entry terminates the table.
1797  *
1798  * The members of the &struct ctl_table structure are used as follows:
1799  *
1800  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1801  *            enter a sysctl file
1802  *
1803  * data - a pointer to data for use by proc_handler
1804  *
1805  * maxlen - the maximum size in bytes of the data
1806  *
1807  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1808  *
1809  * child - a pointer to the child sysctl table if this entry is a directory, or
1810  *         %NULL.
1811  *
1812  * proc_handler - the text handler routine (described below)
1813  *
1814  * de - for internal use by the sysctl routines
1815  *
1816  * extra1, extra2 - extra pointers usable by the proc handler routines
1817  *
1818  * Leaf nodes in the sysctl tree will be represented by a single file
1819  * under /proc; non-leaf nodes will be represented by directories.
1820  *
1821  * sysctl(2) can automatically manage read and write requests through
1822  * the sysctl table.  The data and maxlen fields of the ctl_table
1823  * struct enable minimal validation of the values being written to be
1824  * performed, and the mode field allows minimal authentication.
1825  *
1826  * There must be a proc_handler routine for any terminal nodes
1827  * mirrored under /proc/sys (non-terminals are handled by a built-in
1828  * directory handler).  Several default handlers are available to
1829  * cover common cases -
1830  *
1831  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1832  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1833  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1834  *
1835  * It is the handler's job to read the input buffer from user memory
1836  * and process it. The handler should return 0 on success.
1837  *
1838  * This routine returns %NULL on a failure to register, and a pointer
1839  * to the table header on success.
1840  */
1841 struct ctl_table_header *__register_sysctl_paths(
1842 	struct ctl_table_root *root,
1843 	struct nsproxy *namespaces,
1844 	const struct ctl_path *path, struct ctl_table *table)
1845 {
1846 	struct ctl_table_header *header;
1847 	struct ctl_table *new, **prevp;
1848 	unsigned int n, npath;
1849 	struct ctl_table_set *set;
1850 
1851 	/* Count the path components */
1852 	for (npath = 0; path[npath].procname; ++npath)
1853 		;
1854 
1855 	/*
1856 	 * For each path component, allocate a 2-element ctl_table array.
1857 	 * The first array element will be filled with the sysctl entry
1858 	 * for this, the second will be the sentinel (procname == 0).
1859 	 *
1860 	 * We allocate everything in one go so that we don't have to
1861 	 * worry about freeing additional memory in unregister_sysctl_table.
1862 	 */
1863 	header = kzalloc(sizeof(struct ctl_table_header) +
1864 			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1865 	if (!header)
1866 		return NULL;
1867 
1868 	new = (struct ctl_table *) (header + 1);
1869 
1870 	/* Now connect the dots */
1871 	prevp = &header->ctl_table;
1872 	for (n = 0; n < npath; ++n, ++path) {
1873 		/* Copy the procname */
1874 		new->procname = path->procname;
1875 		new->mode     = 0555;
1876 
1877 		*prevp = new;
1878 		prevp = &new->child;
1879 
1880 		new += 2;
1881 	}
1882 	*prevp = table;
1883 	header->ctl_table_arg = table;
1884 
1885 	INIT_LIST_HEAD(&header->ctl_entry);
1886 	header->used = 0;
1887 	header->unregistering = NULL;
1888 	header->root = root;
1889 	sysctl_set_parent(NULL, header->ctl_table);
1890 	header->count = 1;
1891 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1892 	if (sysctl_check_table(namespaces, header->ctl_table)) {
1893 		kfree(header);
1894 		return NULL;
1895 	}
1896 #endif
1897 	spin_lock(&sysctl_lock);
1898 	header->set = lookup_header_set(root, namespaces);
1899 	header->attached_by = header->ctl_table;
1900 	header->attached_to = root_table;
1901 	header->parent = &root_table_header;
1902 	for (set = header->set; set; set = set->parent) {
1903 		struct ctl_table_header *p;
1904 		list_for_each_entry(p, &set->list, ctl_entry) {
1905 			if (p->unregistering)
1906 				continue;
1907 			try_attach(p, header);
1908 		}
1909 	}
1910 	header->parent->count++;
1911 	list_add_tail(&header->ctl_entry, &header->set->list);
1912 	spin_unlock(&sysctl_lock);
1913 
1914 	return header;
1915 }
1916 
1917 /**
1918  * register_sysctl_table_path - register a sysctl table hierarchy
1919  * @path: The path to the directory the sysctl table is in.
1920  * @table: the top-level table structure
1921  *
1922  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1923  * array. A completely 0 filled entry terminates the table.
1924  *
1925  * See __register_sysctl_paths for more details.
1926  */
1927 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1928 						struct ctl_table *table)
1929 {
1930 	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1931 					path, table);
1932 }
1933 
1934 /**
1935  * register_sysctl_table - register a sysctl table hierarchy
1936  * @table: the top-level table structure
1937  *
1938  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1939  * array. A completely 0 filled entry terminates the table.
1940  *
1941  * See register_sysctl_paths for more details.
1942  */
1943 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1944 {
1945 	static const struct ctl_path null_path[] = { {} };
1946 
1947 	return register_sysctl_paths(null_path, table);
1948 }
1949 
1950 /**
1951  * unregister_sysctl_table - unregister a sysctl table hierarchy
1952  * @header: the header returned from register_sysctl_table
1953  *
1954  * Unregisters the sysctl table and all children. proc entries may not
1955  * actually be removed until they are no longer used by anyone.
1956  */
1957 void unregister_sysctl_table(struct ctl_table_header * header)
1958 {
1959 	might_sleep();
1960 
1961 	if (header == NULL)
1962 		return;
1963 
1964 	spin_lock(&sysctl_lock);
1965 	start_unregistering(header);
1966 	if (!--header->parent->count) {
1967 		WARN_ON(1);
1968 		call_rcu(&header->parent->rcu, free_head);
1969 	}
1970 	if (!--header->count)
1971 		call_rcu(&header->rcu, free_head);
1972 	spin_unlock(&sysctl_lock);
1973 }
1974 
1975 int sysctl_is_seen(struct ctl_table_header *p)
1976 {
1977 	struct ctl_table_set *set = p->set;
1978 	int res;
1979 	spin_lock(&sysctl_lock);
1980 	if (p->unregistering)
1981 		res = 0;
1982 	else if (!set->is_seen)
1983 		res = 1;
1984 	else
1985 		res = set->is_seen(set);
1986 	spin_unlock(&sysctl_lock);
1987 	return res;
1988 }
1989 
1990 void setup_sysctl_set(struct ctl_table_set *p,
1991 	struct ctl_table_set *parent,
1992 	int (*is_seen)(struct ctl_table_set *))
1993 {
1994 	INIT_LIST_HEAD(&p->list);
1995 	p->parent = parent ? parent : &sysctl_table_root.default_set;
1996 	p->is_seen = is_seen;
1997 }
1998 
1999 #else /* !CONFIG_SYSCTL */
2000 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2001 {
2002 	return NULL;
2003 }
2004 
2005 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2006 						    struct ctl_table *table)
2007 {
2008 	return NULL;
2009 }
2010 
2011 void unregister_sysctl_table(struct ctl_table_header * table)
2012 {
2013 }
2014 
2015 void setup_sysctl_set(struct ctl_table_set *p,
2016 	struct ctl_table_set *parent,
2017 	int (*is_seen)(struct ctl_table_set *))
2018 {
2019 }
2020 
2021 void sysctl_head_put(struct ctl_table_header *head)
2022 {
2023 }
2024 
2025 #endif /* CONFIG_SYSCTL */
2026 
2027 /*
2028  * /proc/sys support
2029  */
2030 
2031 #ifdef CONFIG_PROC_SYSCTL
2032 
2033 static int _proc_do_string(void* data, int maxlen, int write,
2034 			   void __user *buffer,
2035 			   size_t *lenp, loff_t *ppos)
2036 {
2037 	size_t len;
2038 	char __user *p;
2039 	char c;
2040 
2041 	if (!data || !maxlen || !*lenp) {
2042 		*lenp = 0;
2043 		return 0;
2044 	}
2045 
2046 	if (write) {
2047 		len = 0;
2048 		p = buffer;
2049 		while (len < *lenp) {
2050 			if (get_user(c, p++))
2051 				return -EFAULT;
2052 			if (c == 0 || c == '\n')
2053 				break;
2054 			len++;
2055 		}
2056 		if (len >= maxlen)
2057 			len = maxlen-1;
2058 		if(copy_from_user(data, buffer, len))
2059 			return -EFAULT;
2060 		((char *) data)[len] = 0;
2061 		*ppos += *lenp;
2062 	} else {
2063 		len = strlen(data);
2064 		if (len > maxlen)
2065 			len = maxlen;
2066 
2067 		if (*ppos > len) {
2068 			*lenp = 0;
2069 			return 0;
2070 		}
2071 
2072 		data += *ppos;
2073 		len  -= *ppos;
2074 
2075 		if (len > *lenp)
2076 			len = *lenp;
2077 		if (len)
2078 			if(copy_to_user(buffer, data, len))
2079 				return -EFAULT;
2080 		if (len < *lenp) {
2081 			if(put_user('\n', ((char __user *) buffer) + len))
2082 				return -EFAULT;
2083 			len++;
2084 		}
2085 		*lenp = len;
2086 		*ppos += len;
2087 	}
2088 	return 0;
2089 }
2090 
2091 /**
2092  * proc_dostring - read a string sysctl
2093  * @table: the sysctl table
2094  * @write: %TRUE if this is a write to the sysctl file
2095  * @buffer: the user buffer
2096  * @lenp: the size of the user buffer
2097  * @ppos: file position
2098  *
2099  * Reads/writes a string from/to the user buffer. If the kernel
2100  * buffer provided is not large enough to hold the string, the
2101  * string is truncated. The copied string is %NULL-terminated.
2102  * If the string is being read by the user process, it is copied
2103  * and a newline '\n' is added. It is truncated if the buffer is
2104  * not large enough.
2105  *
2106  * Returns 0 on success.
2107  */
2108 int proc_dostring(struct ctl_table *table, int write,
2109 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2110 {
2111 	return _proc_do_string(table->data, table->maxlen, write,
2112 			       buffer, lenp, ppos);
2113 }
2114 
2115 static size_t proc_skip_spaces(char **buf)
2116 {
2117 	size_t ret;
2118 	char *tmp = skip_spaces(*buf);
2119 	ret = tmp - *buf;
2120 	*buf = tmp;
2121 	return ret;
2122 }
2123 
2124 static void proc_skip_char(char **buf, size_t *size, const char v)
2125 {
2126 	while (*size) {
2127 		if (**buf != v)
2128 			break;
2129 		(*size)--;
2130 		(*buf)++;
2131 	}
2132 }
2133 
2134 #define TMPBUFLEN 22
2135 /**
2136  * proc_get_long - reads an ASCII formatted integer from a user buffer
2137  *
2138  * @buf: a kernel buffer
2139  * @size: size of the kernel buffer
2140  * @val: this is where the number will be stored
2141  * @neg: set to %TRUE if number is negative
2142  * @perm_tr: a vector which contains the allowed trailers
2143  * @perm_tr_len: size of the perm_tr vector
2144  * @tr: pointer to store the trailer character
2145  *
2146  * In case of success %0 is returned and @buf and @size are updated with
2147  * the amount of bytes read. If @tr is non-NULL and a trailing
2148  * character exists (size is non-zero after returning from this
2149  * function), @tr is updated with the trailing character.
2150  */
2151 static int proc_get_long(char **buf, size_t *size,
2152 			  unsigned long *val, bool *neg,
2153 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
2154 {
2155 	int len;
2156 	char *p, tmp[TMPBUFLEN];
2157 
2158 	if (!*size)
2159 		return -EINVAL;
2160 
2161 	len = *size;
2162 	if (len > TMPBUFLEN - 1)
2163 		len = TMPBUFLEN - 1;
2164 
2165 	memcpy(tmp, *buf, len);
2166 
2167 	tmp[len] = 0;
2168 	p = tmp;
2169 	if (*p == '-' && *size > 1) {
2170 		*neg = true;
2171 		p++;
2172 	} else
2173 		*neg = false;
2174 	if (!isdigit(*p))
2175 		return -EINVAL;
2176 
2177 	*val = simple_strtoul(p, &p, 0);
2178 
2179 	len = p - tmp;
2180 
2181 	/* We don't know if the next char is whitespace thus we may accept
2182 	 * invalid integers (e.g. 1234...a) or two integers instead of one
2183 	 * (e.g. 123...1). So lets not allow such large numbers. */
2184 	if (len == TMPBUFLEN - 1)
2185 		return -EINVAL;
2186 
2187 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2188 		return -EINVAL;
2189 
2190 	if (tr && (len < *size))
2191 		*tr = *p;
2192 
2193 	*buf += len;
2194 	*size -= len;
2195 
2196 	return 0;
2197 }
2198 
2199 /**
2200  * proc_put_long - converts an integer to a decimal ASCII formatted string
2201  *
2202  * @buf: the user buffer
2203  * @size: the size of the user buffer
2204  * @val: the integer to be converted
2205  * @neg: sign of the number, %TRUE for negative
2206  *
2207  * In case of success %0 is returned and @buf and @size are updated with
2208  * the amount of bytes written.
2209  */
2210 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2211 			  bool neg)
2212 {
2213 	int len;
2214 	char tmp[TMPBUFLEN], *p = tmp;
2215 
2216 	sprintf(p, "%s%lu", neg ? "-" : "", val);
2217 	len = strlen(tmp);
2218 	if (len > *size)
2219 		len = *size;
2220 	if (copy_to_user(*buf, tmp, len))
2221 		return -EFAULT;
2222 	*size -= len;
2223 	*buf += len;
2224 	return 0;
2225 }
2226 #undef TMPBUFLEN
2227 
2228 static int proc_put_char(void __user **buf, size_t *size, char c)
2229 {
2230 	if (*size) {
2231 		char __user **buffer = (char __user **)buf;
2232 		if (put_user(c, *buffer))
2233 			return -EFAULT;
2234 		(*size)--, (*buffer)++;
2235 		*buf = *buffer;
2236 	}
2237 	return 0;
2238 }
2239 
2240 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2241 				 int *valp,
2242 				 int write, void *data)
2243 {
2244 	if (write) {
2245 		*valp = *negp ? -*lvalp : *lvalp;
2246 	} else {
2247 		int val = *valp;
2248 		if (val < 0) {
2249 			*negp = true;
2250 			*lvalp = (unsigned long)-val;
2251 		} else {
2252 			*negp = false;
2253 			*lvalp = (unsigned long)val;
2254 		}
2255 	}
2256 	return 0;
2257 }
2258 
2259 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2260 
2261 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2262 		  int write, void __user *buffer,
2263 		  size_t *lenp, loff_t *ppos,
2264 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2265 			      int write, void *data),
2266 		  void *data)
2267 {
2268 	int *i, vleft, first = 1, err = 0;
2269 	unsigned long page = 0;
2270 	size_t left;
2271 	char *kbuf;
2272 
2273 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2274 		*lenp = 0;
2275 		return 0;
2276 	}
2277 
2278 	i = (int *) tbl_data;
2279 	vleft = table->maxlen / sizeof(*i);
2280 	left = *lenp;
2281 
2282 	if (!conv)
2283 		conv = do_proc_dointvec_conv;
2284 
2285 	if (write) {
2286 		if (left > PAGE_SIZE - 1)
2287 			left = PAGE_SIZE - 1;
2288 		page = __get_free_page(GFP_TEMPORARY);
2289 		kbuf = (char *) page;
2290 		if (!kbuf)
2291 			return -ENOMEM;
2292 		if (copy_from_user(kbuf, buffer, left)) {
2293 			err = -EFAULT;
2294 			goto free;
2295 		}
2296 		kbuf[left] = 0;
2297 	}
2298 
2299 	for (; left && vleft--; i++, first=0) {
2300 		unsigned long lval;
2301 		bool neg;
2302 
2303 		if (write) {
2304 			left -= proc_skip_spaces(&kbuf);
2305 
2306 			if (!left)
2307 				break;
2308 			err = proc_get_long(&kbuf, &left, &lval, &neg,
2309 					     proc_wspace_sep,
2310 					     sizeof(proc_wspace_sep), NULL);
2311 			if (err)
2312 				break;
2313 			if (conv(&neg, &lval, i, 1, data)) {
2314 				err = -EINVAL;
2315 				break;
2316 			}
2317 		} else {
2318 			if (conv(&neg, &lval, i, 0, data)) {
2319 				err = -EINVAL;
2320 				break;
2321 			}
2322 			if (!first)
2323 				err = proc_put_char(&buffer, &left, '\t');
2324 			if (err)
2325 				break;
2326 			err = proc_put_long(&buffer, &left, lval, neg);
2327 			if (err)
2328 				break;
2329 		}
2330 	}
2331 
2332 	if (!write && !first && left && !err)
2333 		err = proc_put_char(&buffer, &left, '\n');
2334 	if (write && !err && left)
2335 		left -= proc_skip_spaces(&kbuf);
2336 free:
2337 	if (write) {
2338 		free_page(page);
2339 		if (first)
2340 			return err ? : -EINVAL;
2341 	}
2342 	*lenp -= left;
2343 	*ppos += *lenp;
2344 	return err;
2345 }
2346 
2347 static int do_proc_dointvec(struct ctl_table *table, int write,
2348 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2349 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2350 			      int write, void *data),
2351 		  void *data)
2352 {
2353 	return __do_proc_dointvec(table->data, table, write,
2354 			buffer, lenp, ppos, conv, data);
2355 }
2356 
2357 /**
2358  * proc_dointvec - read a vector of integers
2359  * @table: the sysctl table
2360  * @write: %TRUE if this is a write to the sysctl file
2361  * @buffer: the user buffer
2362  * @lenp: the size of the user buffer
2363  * @ppos: file position
2364  *
2365  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2366  * values from/to the user buffer, treated as an ASCII string.
2367  *
2368  * Returns 0 on success.
2369  */
2370 int proc_dointvec(struct ctl_table *table, int write,
2371 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2372 {
2373     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2374 		    	    NULL,NULL);
2375 }
2376 
2377 /*
2378  * Taint values can only be increased
2379  * This means we can safely use a temporary.
2380  */
2381 static int proc_taint(struct ctl_table *table, int write,
2382 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2383 {
2384 	struct ctl_table t;
2385 	unsigned long tmptaint = get_taint();
2386 	int err;
2387 
2388 	if (write && !capable(CAP_SYS_ADMIN))
2389 		return -EPERM;
2390 
2391 	t = *table;
2392 	t.data = &tmptaint;
2393 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2394 	if (err < 0)
2395 		return err;
2396 
2397 	if (write) {
2398 		/*
2399 		 * Poor man's atomic or. Not worth adding a primitive
2400 		 * to everyone's atomic.h for this
2401 		 */
2402 		int i;
2403 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2404 			if ((tmptaint >> i) & 1)
2405 				add_taint(i);
2406 		}
2407 	}
2408 
2409 	return err;
2410 }
2411 
2412 #ifdef CONFIG_PRINTK
2413 static int proc_dmesg_restrict(struct ctl_table *table, int write,
2414 				void __user *buffer, size_t *lenp, loff_t *ppos)
2415 {
2416 	if (write && !capable(CAP_SYS_ADMIN))
2417 		return -EPERM;
2418 
2419 	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2420 }
2421 #endif
2422 
2423 struct do_proc_dointvec_minmax_conv_param {
2424 	int *min;
2425 	int *max;
2426 };
2427 
2428 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2429 					int *valp,
2430 					int write, void *data)
2431 {
2432 	struct do_proc_dointvec_minmax_conv_param *param = data;
2433 	if (write) {
2434 		int val = *negp ? -*lvalp : *lvalp;
2435 		if ((param->min && *param->min > val) ||
2436 		    (param->max && *param->max < val))
2437 			return -EINVAL;
2438 		*valp = val;
2439 	} else {
2440 		int val = *valp;
2441 		if (val < 0) {
2442 			*negp = true;
2443 			*lvalp = (unsigned long)-val;
2444 		} else {
2445 			*negp = false;
2446 			*lvalp = (unsigned long)val;
2447 		}
2448 	}
2449 	return 0;
2450 }
2451 
2452 /**
2453  * proc_dointvec_minmax - read a vector of integers with min/max values
2454  * @table: the sysctl table
2455  * @write: %TRUE if this is a write to the sysctl file
2456  * @buffer: the user buffer
2457  * @lenp: the size of the user buffer
2458  * @ppos: file position
2459  *
2460  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2461  * values from/to the user buffer, treated as an ASCII string.
2462  *
2463  * This routine will ensure the values are within the range specified by
2464  * table->extra1 (min) and table->extra2 (max).
2465  *
2466  * Returns 0 on success.
2467  */
2468 int proc_dointvec_minmax(struct ctl_table *table, int write,
2469 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2470 {
2471 	struct do_proc_dointvec_minmax_conv_param param = {
2472 		.min = (int *) table->extra1,
2473 		.max = (int *) table->extra2,
2474 	};
2475 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2476 				do_proc_dointvec_minmax_conv, &param);
2477 }
2478 
2479 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2480 				     void __user *buffer,
2481 				     size_t *lenp, loff_t *ppos,
2482 				     unsigned long convmul,
2483 				     unsigned long convdiv)
2484 {
2485 	unsigned long *i, *min, *max;
2486 	int vleft, first = 1, err = 0;
2487 	unsigned long page = 0;
2488 	size_t left;
2489 	char *kbuf;
2490 
2491 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2492 		*lenp = 0;
2493 		return 0;
2494 	}
2495 
2496 	i = (unsigned long *) data;
2497 	min = (unsigned long *) table->extra1;
2498 	max = (unsigned long *) table->extra2;
2499 	vleft = table->maxlen / sizeof(unsigned long);
2500 	left = *lenp;
2501 
2502 	if (write) {
2503 		if (left > PAGE_SIZE - 1)
2504 			left = PAGE_SIZE - 1;
2505 		page = __get_free_page(GFP_TEMPORARY);
2506 		kbuf = (char *) page;
2507 		if (!kbuf)
2508 			return -ENOMEM;
2509 		if (copy_from_user(kbuf, buffer, left)) {
2510 			err = -EFAULT;
2511 			goto free;
2512 		}
2513 		kbuf[left] = 0;
2514 	}
2515 
2516 	for (; left && vleft--; i++, first = 0) {
2517 		unsigned long val;
2518 
2519 		if (write) {
2520 			bool neg;
2521 
2522 			left -= proc_skip_spaces(&kbuf);
2523 
2524 			err = proc_get_long(&kbuf, &left, &val, &neg,
2525 					     proc_wspace_sep,
2526 					     sizeof(proc_wspace_sep), NULL);
2527 			if (err)
2528 				break;
2529 			if (neg)
2530 				continue;
2531 			if ((min && val < *min) || (max && val > *max))
2532 				continue;
2533 			*i = val;
2534 		} else {
2535 			val = convdiv * (*i) / convmul;
2536 			if (!first)
2537 				err = proc_put_char(&buffer, &left, '\t');
2538 			err = proc_put_long(&buffer, &left, val, false);
2539 			if (err)
2540 				break;
2541 		}
2542 	}
2543 
2544 	if (!write && !first && left && !err)
2545 		err = proc_put_char(&buffer, &left, '\n');
2546 	if (write && !err)
2547 		left -= proc_skip_spaces(&kbuf);
2548 free:
2549 	if (write) {
2550 		free_page(page);
2551 		if (first)
2552 			return err ? : -EINVAL;
2553 	}
2554 	*lenp -= left;
2555 	*ppos += *lenp;
2556 	return err;
2557 }
2558 
2559 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2560 				     void __user *buffer,
2561 				     size_t *lenp, loff_t *ppos,
2562 				     unsigned long convmul,
2563 				     unsigned long convdiv)
2564 {
2565 	return __do_proc_doulongvec_minmax(table->data, table, write,
2566 			buffer, lenp, ppos, convmul, convdiv);
2567 }
2568 
2569 /**
2570  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2571  * @table: the sysctl table
2572  * @write: %TRUE if this is a write to the sysctl file
2573  * @buffer: the user buffer
2574  * @lenp: the size of the user buffer
2575  * @ppos: file position
2576  *
2577  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2578  * values from/to the user buffer, treated as an ASCII string.
2579  *
2580  * This routine will ensure the values are within the range specified by
2581  * table->extra1 (min) and table->extra2 (max).
2582  *
2583  * Returns 0 on success.
2584  */
2585 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2586 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2587 {
2588     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2589 }
2590 
2591 /**
2592  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2593  * @table: the sysctl table
2594  * @write: %TRUE if this is a write to the sysctl file
2595  * @buffer: the user buffer
2596  * @lenp: the size of the user buffer
2597  * @ppos: file position
2598  *
2599  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2600  * values from/to the user buffer, treated as an ASCII string. The values
2601  * are treated as milliseconds, and converted to jiffies when they are stored.
2602  *
2603  * This routine will ensure the values are within the range specified by
2604  * table->extra1 (min) and table->extra2 (max).
2605  *
2606  * Returns 0 on success.
2607  */
2608 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2609 				      void __user *buffer,
2610 				      size_t *lenp, loff_t *ppos)
2611 {
2612     return do_proc_doulongvec_minmax(table, write, buffer,
2613 				     lenp, ppos, HZ, 1000l);
2614 }
2615 
2616 
2617 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2618 					 int *valp,
2619 					 int write, void *data)
2620 {
2621 	if (write) {
2622 		if (*lvalp > LONG_MAX / HZ)
2623 			return 1;
2624 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2625 	} else {
2626 		int val = *valp;
2627 		unsigned long lval;
2628 		if (val < 0) {
2629 			*negp = true;
2630 			lval = (unsigned long)-val;
2631 		} else {
2632 			*negp = false;
2633 			lval = (unsigned long)val;
2634 		}
2635 		*lvalp = lval / HZ;
2636 	}
2637 	return 0;
2638 }
2639 
2640 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2641 						int *valp,
2642 						int write, void *data)
2643 {
2644 	if (write) {
2645 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2646 			return 1;
2647 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2648 	} else {
2649 		int val = *valp;
2650 		unsigned long lval;
2651 		if (val < 0) {
2652 			*negp = true;
2653 			lval = (unsigned long)-val;
2654 		} else {
2655 			*negp = false;
2656 			lval = (unsigned long)val;
2657 		}
2658 		*lvalp = jiffies_to_clock_t(lval);
2659 	}
2660 	return 0;
2661 }
2662 
2663 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2664 					    int *valp,
2665 					    int write, void *data)
2666 {
2667 	if (write) {
2668 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2669 	} else {
2670 		int val = *valp;
2671 		unsigned long lval;
2672 		if (val < 0) {
2673 			*negp = true;
2674 			lval = (unsigned long)-val;
2675 		} else {
2676 			*negp = false;
2677 			lval = (unsigned long)val;
2678 		}
2679 		*lvalp = jiffies_to_msecs(lval);
2680 	}
2681 	return 0;
2682 }
2683 
2684 /**
2685  * proc_dointvec_jiffies - read a vector of integers as seconds
2686  * @table: the sysctl table
2687  * @write: %TRUE if this is a write to the sysctl file
2688  * @buffer: the user buffer
2689  * @lenp: the size of the user buffer
2690  * @ppos: file position
2691  *
2692  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2693  * values from/to the user buffer, treated as an ASCII string.
2694  * The values read are assumed to be in seconds, and are converted into
2695  * jiffies.
2696  *
2697  * Returns 0 on success.
2698  */
2699 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2700 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2701 {
2702     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2703 		    	    do_proc_dointvec_jiffies_conv,NULL);
2704 }
2705 
2706 /**
2707  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2708  * @table: the sysctl table
2709  * @write: %TRUE if this is a write to the sysctl file
2710  * @buffer: the user buffer
2711  * @lenp: the size of the user buffer
2712  * @ppos: pointer to the file position
2713  *
2714  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2715  * values from/to the user buffer, treated as an ASCII string.
2716  * The values read are assumed to be in 1/USER_HZ seconds, and
2717  * are converted into jiffies.
2718  *
2719  * Returns 0 on success.
2720  */
2721 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2722 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2723 {
2724     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2725 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2726 }
2727 
2728 /**
2729  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2730  * @table: the sysctl table
2731  * @write: %TRUE if this is a write to the sysctl file
2732  * @buffer: the user buffer
2733  * @lenp: the size of the user buffer
2734  * @ppos: file position
2735  * @ppos: the current position in the file
2736  *
2737  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2738  * values from/to the user buffer, treated as an ASCII string.
2739  * The values read are assumed to be in 1/1000 seconds, and
2740  * are converted into jiffies.
2741  *
2742  * Returns 0 on success.
2743  */
2744 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2745 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2746 {
2747 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2748 				do_proc_dointvec_ms_jiffies_conv, NULL);
2749 }
2750 
2751 static int proc_do_cad_pid(struct ctl_table *table, int write,
2752 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2753 {
2754 	struct pid *new_pid;
2755 	pid_t tmp;
2756 	int r;
2757 
2758 	tmp = pid_vnr(cad_pid);
2759 
2760 	r = __do_proc_dointvec(&tmp, table, write, buffer,
2761 			       lenp, ppos, NULL, NULL);
2762 	if (r || !write)
2763 		return r;
2764 
2765 	new_pid = find_get_pid(tmp);
2766 	if (!new_pid)
2767 		return -ESRCH;
2768 
2769 	put_pid(xchg(&cad_pid, new_pid));
2770 	return 0;
2771 }
2772 
2773 /**
2774  * proc_do_large_bitmap - read/write from/to a large bitmap
2775  * @table: the sysctl table
2776  * @write: %TRUE if this is a write to the sysctl file
2777  * @buffer: the user buffer
2778  * @lenp: the size of the user buffer
2779  * @ppos: file position
2780  *
2781  * The bitmap is stored at table->data and the bitmap length (in bits)
2782  * in table->maxlen.
2783  *
2784  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2785  * large bitmaps may be represented in a compact manner. Writing into
2786  * the file will clear the bitmap then update it with the given input.
2787  *
2788  * Returns 0 on success.
2789  */
2790 int proc_do_large_bitmap(struct ctl_table *table, int write,
2791 			 void __user *buffer, size_t *lenp, loff_t *ppos)
2792 {
2793 	int err = 0;
2794 	bool first = 1;
2795 	size_t left = *lenp;
2796 	unsigned long bitmap_len = table->maxlen;
2797 	unsigned long *bitmap = (unsigned long *) table->data;
2798 	unsigned long *tmp_bitmap = NULL;
2799 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2800 
2801 	if (!bitmap_len || !left || (*ppos && !write)) {
2802 		*lenp = 0;
2803 		return 0;
2804 	}
2805 
2806 	if (write) {
2807 		unsigned long page = 0;
2808 		char *kbuf;
2809 
2810 		if (left > PAGE_SIZE - 1)
2811 			left = PAGE_SIZE - 1;
2812 
2813 		page = __get_free_page(GFP_TEMPORARY);
2814 		kbuf = (char *) page;
2815 		if (!kbuf)
2816 			return -ENOMEM;
2817 		if (copy_from_user(kbuf, buffer, left)) {
2818 			free_page(page);
2819 			return -EFAULT;
2820                 }
2821 		kbuf[left] = 0;
2822 
2823 		tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2824 				     GFP_KERNEL);
2825 		if (!tmp_bitmap) {
2826 			free_page(page);
2827 			return -ENOMEM;
2828 		}
2829 		proc_skip_char(&kbuf, &left, '\n');
2830 		while (!err && left) {
2831 			unsigned long val_a, val_b;
2832 			bool neg;
2833 
2834 			err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2835 					     sizeof(tr_a), &c);
2836 			if (err)
2837 				break;
2838 			if (val_a >= bitmap_len || neg) {
2839 				err = -EINVAL;
2840 				break;
2841 			}
2842 
2843 			val_b = val_a;
2844 			if (left) {
2845 				kbuf++;
2846 				left--;
2847 			}
2848 
2849 			if (c == '-') {
2850 				err = proc_get_long(&kbuf, &left, &val_b,
2851 						     &neg, tr_b, sizeof(tr_b),
2852 						     &c);
2853 				if (err)
2854 					break;
2855 				if (val_b >= bitmap_len || neg ||
2856 				    val_a > val_b) {
2857 					err = -EINVAL;
2858 					break;
2859 				}
2860 				if (left) {
2861 					kbuf++;
2862 					left--;
2863 				}
2864 			}
2865 
2866 			while (val_a <= val_b)
2867 				set_bit(val_a++, tmp_bitmap);
2868 
2869 			first = 0;
2870 			proc_skip_char(&kbuf, &left, '\n');
2871 		}
2872 		free_page(page);
2873 	} else {
2874 		unsigned long bit_a, bit_b = 0;
2875 
2876 		while (left) {
2877 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2878 			if (bit_a >= bitmap_len)
2879 				break;
2880 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
2881 						   bit_a + 1) - 1;
2882 
2883 			if (!first) {
2884 				err = proc_put_char(&buffer, &left, ',');
2885 				if (err)
2886 					break;
2887 			}
2888 			err = proc_put_long(&buffer, &left, bit_a, false);
2889 			if (err)
2890 				break;
2891 			if (bit_a != bit_b) {
2892 				err = proc_put_char(&buffer, &left, '-');
2893 				if (err)
2894 					break;
2895 				err = proc_put_long(&buffer, &left, bit_b, false);
2896 				if (err)
2897 					break;
2898 			}
2899 
2900 			first = 0; bit_b++;
2901 		}
2902 		if (!err)
2903 			err = proc_put_char(&buffer, &left, '\n');
2904 	}
2905 
2906 	if (!err) {
2907 		if (write) {
2908 			if (*ppos)
2909 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2910 			else
2911 				memcpy(bitmap, tmp_bitmap,
2912 					BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long));
2913 		}
2914 		kfree(tmp_bitmap);
2915 		*lenp -= left;
2916 		*ppos += *lenp;
2917 		return 0;
2918 	} else {
2919 		kfree(tmp_bitmap);
2920 		return err;
2921 	}
2922 }
2923 
2924 #else /* CONFIG_PROC_SYSCTL */
2925 
2926 int proc_dostring(struct ctl_table *table, int write,
2927 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2928 {
2929 	return -ENOSYS;
2930 }
2931 
2932 int proc_dointvec(struct ctl_table *table, int write,
2933 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2934 {
2935 	return -ENOSYS;
2936 }
2937 
2938 int proc_dointvec_minmax(struct ctl_table *table, int write,
2939 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2940 {
2941 	return -ENOSYS;
2942 }
2943 
2944 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2945 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2946 {
2947 	return -ENOSYS;
2948 }
2949 
2950 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2951 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2952 {
2953 	return -ENOSYS;
2954 }
2955 
2956 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2957 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2958 {
2959 	return -ENOSYS;
2960 }
2961 
2962 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2963 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2964 {
2965 	return -ENOSYS;
2966 }
2967 
2968 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2969 				      void __user *buffer,
2970 				      size_t *lenp, loff_t *ppos)
2971 {
2972     return -ENOSYS;
2973 }
2974 
2975 
2976 #endif /* CONFIG_PROC_SYSCTL */
2977 
2978 /*
2979  * No sense putting this after each symbol definition, twice,
2980  * exception granted :-)
2981  */
2982 EXPORT_SYMBOL(proc_dointvec);
2983 EXPORT_SYMBOL(proc_dointvec_jiffies);
2984 EXPORT_SYMBOL(proc_dointvec_minmax);
2985 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2986 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2987 EXPORT_SYMBOL(proc_dostring);
2988 EXPORT_SYMBOL(proc_doulongvec_minmax);
2989 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2990 EXPORT_SYMBOL(register_sysctl_table);
2991 EXPORT_SYMBOL(register_sysctl_paths);
2992 EXPORT_SYMBOL(unregister_sysctl_table);
2993