xref: /linux/kernel/sysctl.c (revision c54ea4918c2b7722d7242ea53271356501988a9b)
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 
60 #include <asm/uaccess.h>
61 #include <asm/processor.h>
62 
63 #ifdef CONFIG_X86
64 #include <asm/nmi.h>
65 #include <asm/stacktrace.h>
66 #include <asm/io.h>
67 #endif
68 #ifdef CONFIG_BSD_PROCESS_ACCT
69 #include <linux/acct.h>
70 #endif
71 #ifdef CONFIG_RT_MUTEXES
72 #include <linux/rtmutex.h>
73 #endif
74 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
75 #include <linux/lockdep.h>
76 #endif
77 #ifdef CONFIG_CHR_DEV_SG
78 #include <scsi/sg.h>
79 #endif
80 
81 #ifdef CONFIG_LOCKUP_DETECTOR
82 #include <linux/nmi.h>
83 #endif
84 
85 
86 #if defined(CONFIG_SYSCTL)
87 
88 /* External variables not in a header file. */
89 extern int sysctl_overcommit_memory;
90 extern int sysctl_overcommit_ratio;
91 extern int max_threads;
92 extern int core_uses_pid;
93 extern int suid_dumpable;
94 extern char core_pattern[];
95 extern unsigned int core_pipe_limit;
96 extern int pid_max;
97 extern int min_free_kbytes;
98 extern int pid_max_min, pid_max_max;
99 extern int sysctl_drop_caches;
100 extern int percpu_pagelist_fraction;
101 extern int compat_log;
102 extern int latencytop_enabled;
103 extern int sysctl_nr_open_min, sysctl_nr_open_max;
104 #ifndef CONFIG_MMU
105 extern int sysctl_nr_trim_pages;
106 #endif
107 #ifdef CONFIG_BLOCK
108 extern int blk_iopoll_enabled;
109 #endif
110 
111 /* Constants used for minimum and  maximum */
112 #ifdef CONFIG_LOCKUP_DETECTOR
113 static int sixty = 60;
114 static int neg_one = -1;
115 #endif
116 
117 static int zero;
118 static int __maybe_unused one = 1;
119 static int __maybe_unused two = 2;
120 static unsigned long one_ul = 1;
121 static int one_hundred = 100;
122 #ifdef CONFIG_PRINTK
123 static int ten_thousand = 10000;
124 #endif
125 
126 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
127 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
128 
129 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
130 static int maxolduid = 65535;
131 static int minolduid;
132 static int min_percpu_pagelist_fract = 8;
133 
134 static int ngroups_max = NGROUPS_MAX;
135 
136 #ifdef CONFIG_INOTIFY_USER
137 #include <linux/inotify.h>
138 #endif
139 #ifdef CONFIG_SPARC
140 #include <asm/system.h>
141 #endif
142 
143 #ifdef CONFIG_SPARC64
144 extern int sysctl_tsb_ratio;
145 #endif
146 
147 #ifdef __hppa__
148 extern int pwrsw_enabled;
149 extern int unaligned_enabled;
150 #endif
151 
152 #ifdef CONFIG_S390
153 #ifdef CONFIG_MATHEMU
154 extern int sysctl_ieee_emulation_warnings;
155 #endif
156 extern int sysctl_userprocess_debug;
157 extern int spin_retry;
158 #endif
159 
160 #ifdef CONFIG_IA64
161 extern int no_unaligned_warning;
162 extern int unaligned_dump_stack;
163 #endif
164 
165 #ifdef CONFIG_PROC_SYSCTL
166 static int proc_do_cad_pid(struct ctl_table *table, int write,
167 		  void __user *buffer, size_t *lenp, loff_t *ppos);
168 static int proc_taint(struct ctl_table *table, int write,
169 			       void __user *buffer, size_t *lenp, loff_t *ppos);
170 #endif
171 
172 #ifdef CONFIG_MAGIC_SYSRQ
173 /* Note: sysrq code uses it's own private copy */
174 static int __sysrq_enabled = SYSRQ_DEFAULT_ENABLE;
175 
176 static int sysrq_sysctl_handler(ctl_table *table, int write,
177 				void __user *buffer, size_t *lenp,
178 				loff_t *ppos)
179 {
180 	int error;
181 
182 	error = proc_dointvec(table, write, buffer, lenp, ppos);
183 	if (error)
184 		return error;
185 
186 	if (write)
187 		sysrq_toggle_support(__sysrq_enabled);
188 
189 	return 0;
190 }
191 
192 #endif
193 
194 static struct ctl_table root_table[];
195 static struct ctl_table_root sysctl_table_root;
196 static struct ctl_table_header root_table_header = {
197 	{{.count = 1,
198 	.ctl_table = root_table,
199 	.ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),}},
200 	.root = &sysctl_table_root,
201 	.set = &sysctl_table_root.default_set,
202 };
203 static struct ctl_table_root sysctl_table_root = {
204 	.root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
205 	.default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
206 };
207 
208 static struct ctl_table kern_table[];
209 static struct ctl_table vm_table[];
210 static struct ctl_table fs_table[];
211 static struct ctl_table debug_table[];
212 static struct ctl_table dev_table[];
213 extern struct ctl_table random_table[];
214 #ifdef CONFIG_EPOLL
215 extern struct ctl_table epoll_table[];
216 #endif
217 
218 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
219 int sysctl_legacy_va_layout;
220 #endif
221 
222 /* The default sysctl tables: */
223 
224 static struct ctl_table root_table[] = {
225 	{
226 		.procname	= "kernel",
227 		.mode		= 0555,
228 		.child		= kern_table,
229 	},
230 	{
231 		.procname	= "vm",
232 		.mode		= 0555,
233 		.child		= vm_table,
234 	},
235 	{
236 		.procname	= "fs",
237 		.mode		= 0555,
238 		.child		= fs_table,
239 	},
240 	{
241 		.procname	= "debug",
242 		.mode		= 0555,
243 		.child		= debug_table,
244 	},
245 	{
246 		.procname	= "dev",
247 		.mode		= 0555,
248 		.child		= dev_table,
249 	},
250 	{ }
251 };
252 
253 #ifdef CONFIG_SCHED_DEBUG
254 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
255 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
256 static int min_wakeup_granularity_ns;			/* 0 usecs */
257 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
258 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
259 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
260 #endif
261 
262 #ifdef CONFIG_COMPACTION
263 static int min_extfrag_threshold;
264 static int max_extfrag_threshold = 1000;
265 #endif
266 
267 static struct ctl_table kern_table[] = {
268 	{
269 		.procname	= "sched_child_runs_first",
270 		.data		= &sysctl_sched_child_runs_first,
271 		.maxlen		= sizeof(unsigned int),
272 		.mode		= 0644,
273 		.proc_handler	= proc_dointvec,
274 	},
275 #ifdef CONFIG_SCHED_DEBUG
276 	{
277 		.procname	= "sched_min_granularity_ns",
278 		.data		= &sysctl_sched_min_granularity,
279 		.maxlen		= sizeof(unsigned int),
280 		.mode		= 0644,
281 		.proc_handler	= sched_proc_update_handler,
282 		.extra1		= &min_sched_granularity_ns,
283 		.extra2		= &max_sched_granularity_ns,
284 	},
285 	{
286 		.procname	= "sched_latency_ns",
287 		.data		= &sysctl_sched_latency,
288 		.maxlen		= sizeof(unsigned int),
289 		.mode		= 0644,
290 		.proc_handler	= sched_proc_update_handler,
291 		.extra1		= &min_sched_granularity_ns,
292 		.extra2		= &max_sched_granularity_ns,
293 	},
294 	{
295 		.procname	= "sched_wakeup_granularity_ns",
296 		.data		= &sysctl_sched_wakeup_granularity,
297 		.maxlen		= sizeof(unsigned int),
298 		.mode		= 0644,
299 		.proc_handler	= sched_proc_update_handler,
300 		.extra1		= &min_wakeup_granularity_ns,
301 		.extra2		= &max_wakeup_granularity_ns,
302 	},
303 	{
304 		.procname	= "sched_tunable_scaling",
305 		.data		= &sysctl_sched_tunable_scaling,
306 		.maxlen		= sizeof(enum sched_tunable_scaling),
307 		.mode		= 0644,
308 		.proc_handler	= sched_proc_update_handler,
309 		.extra1		= &min_sched_tunable_scaling,
310 		.extra2		= &max_sched_tunable_scaling,
311 	},
312 	{
313 		.procname	= "sched_migration_cost",
314 		.data		= &sysctl_sched_migration_cost,
315 		.maxlen		= sizeof(unsigned int),
316 		.mode		= 0644,
317 		.proc_handler	= proc_dointvec,
318 	},
319 	{
320 		.procname	= "sched_nr_migrate",
321 		.data		= &sysctl_sched_nr_migrate,
322 		.maxlen		= sizeof(unsigned int),
323 		.mode		= 0644,
324 		.proc_handler	= proc_dointvec,
325 	},
326 	{
327 		.procname	= "sched_time_avg",
328 		.data		= &sysctl_sched_time_avg,
329 		.maxlen		= sizeof(unsigned int),
330 		.mode		= 0644,
331 		.proc_handler	= proc_dointvec,
332 	},
333 	{
334 		.procname	= "sched_shares_window",
335 		.data		= &sysctl_sched_shares_window,
336 		.maxlen		= sizeof(unsigned int),
337 		.mode		= 0644,
338 		.proc_handler	= proc_dointvec,
339 	},
340 	{
341 		.procname	= "timer_migration",
342 		.data		= &sysctl_timer_migration,
343 		.maxlen		= sizeof(unsigned int),
344 		.mode		= 0644,
345 		.proc_handler	= proc_dointvec_minmax,
346 		.extra1		= &zero,
347 		.extra2		= &one,
348 	},
349 #endif
350 	{
351 		.procname	= "sched_rt_period_us",
352 		.data		= &sysctl_sched_rt_period,
353 		.maxlen		= sizeof(unsigned int),
354 		.mode		= 0644,
355 		.proc_handler	= sched_rt_handler,
356 	},
357 	{
358 		.procname	= "sched_rt_runtime_us",
359 		.data		= &sysctl_sched_rt_runtime,
360 		.maxlen		= sizeof(int),
361 		.mode		= 0644,
362 		.proc_handler	= sched_rt_handler,
363 	},
364 #ifdef CONFIG_SCHED_AUTOGROUP
365 	{
366 		.procname	= "sched_autogroup_enabled",
367 		.data		= &sysctl_sched_autogroup_enabled,
368 		.maxlen		= sizeof(unsigned int),
369 		.mode		= 0644,
370 		.proc_handler	= proc_dointvec_minmax,
371 		.extra1		= &zero,
372 		.extra2		= &one,
373 	},
374 #endif
375 #ifdef CONFIG_PROVE_LOCKING
376 	{
377 		.procname	= "prove_locking",
378 		.data		= &prove_locking,
379 		.maxlen		= sizeof(int),
380 		.mode		= 0644,
381 		.proc_handler	= proc_dointvec,
382 	},
383 #endif
384 #ifdef CONFIG_LOCK_STAT
385 	{
386 		.procname	= "lock_stat",
387 		.data		= &lock_stat,
388 		.maxlen		= sizeof(int),
389 		.mode		= 0644,
390 		.proc_handler	= proc_dointvec,
391 	},
392 #endif
393 	{
394 		.procname	= "panic",
395 		.data		= &panic_timeout,
396 		.maxlen		= sizeof(int),
397 		.mode		= 0644,
398 		.proc_handler	= proc_dointvec,
399 	},
400 	{
401 		.procname	= "core_uses_pid",
402 		.data		= &core_uses_pid,
403 		.maxlen		= sizeof(int),
404 		.mode		= 0644,
405 		.proc_handler	= proc_dointvec,
406 	},
407 	{
408 		.procname	= "core_pattern",
409 		.data		= core_pattern,
410 		.maxlen		= CORENAME_MAX_SIZE,
411 		.mode		= 0644,
412 		.proc_handler	= proc_dostring,
413 	},
414 	{
415 		.procname	= "core_pipe_limit",
416 		.data		= &core_pipe_limit,
417 		.maxlen		= sizeof(unsigned int),
418 		.mode		= 0644,
419 		.proc_handler	= proc_dointvec,
420 	},
421 #ifdef CONFIG_PROC_SYSCTL
422 	{
423 		.procname	= "tainted",
424 		.maxlen 	= sizeof(long),
425 		.mode		= 0644,
426 		.proc_handler	= proc_taint,
427 	},
428 #endif
429 #ifdef CONFIG_LATENCYTOP
430 	{
431 		.procname	= "latencytop",
432 		.data		= &latencytop_enabled,
433 		.maxlen		= sizeof(int),
434 		.mode		= 0644,
435 		.proc_handler	= proc_dointvec,
436 	},
437 #endif
438 #ifdef CONFIG_BLK_DEV_INITRD
439 	{
440 		.procname	= "real-root-dev",
441 		.data		= &real_root_dev,
442 		.maxlen		= sizeof(int),
443 		.mode		= 0644,
444 		.proc_handler	= proc_dointvec,
445 	},
446 #endif
447 	{
448 		.procname	= "print-fatal-signals",
449 		.data		= &print_fatal_signals,
450 		.maxlen		= sizeof(int),
451 		.mode		= 0644,
452 		.proc_handler	= proc_dointvec,
453 	},
454 #ifdef CONFIG_SPARC
455 	{
456 		.procname	= "reboot-cmd",
457 		.data		= reboot_command,
458 		.maxlen		= 256,
459 		.mode		= 0644,
460 		.proc_handler	= proc_dostring,
461 	},
462 	{
463 		.procname	= "stop-a",
464 		.data		= &stop_a_enabled,
465 		.maxlen		= sizeof (int),
466 		.mode		= 0644,
467 		.proc_handler	= proc_dointvec,
468 	},
469 	{
470 		.procname	= "scons-poweroff",
471 		.data		= &scons_pwroff,
472 		.maxlen		= sizeof (int),
473 		.mode		= 0644,
474 		.proc_handler	= proc_dointvec,
475 	},
476 #endif
477 #ifdef CONFIG_SPARC64
478 	{
479 		.procname	= "tsb-ratio",
480 		.data		= &sysctl_tsb_ratio,
481 		.maxlen		= sizeof (int),
482 		.mode		= 0644,
483 		.proc_handler	= proc_dointvec,
484 	},
485 #endif
486 #ifdef __hppa__
487 	{
488 		.procname	= "soft-power",
489 		.data		= &pwrsw_enabled,
490 		.maxlen		= sizeof (int),
491 	 	.mode		= 0644,
492 		.proc_handler	= proc_dointvec,
493 	},
494 	{
495 		.procname	= "unaligned-trap",
496 		.data		= &unaligned_enabled,
497 		.maxlen		= sizeof (int),
498 		.mode		= 0644,
499 		.proc_handler	= proc_dointvec,
500 	},
501 #endif
502 	{
503 		.procname	= "ctrl-alt-del",
504 		.data		= &C_A_D,
505 		.maxlen		= sizeof(int),
506 		.mode		= 0644,
507 		.proc_handler	= proc_dointvec,
508 	},
509 #ifdef CONFIG_FUNCTION_TRACER
510 	{
511 		.procname	= "ftrace_enabled",
512 		.data		= &ftrace_enabled,
513 		.maxlen		= sizeof(int),
514 		.mode		= 0644,
515 		.proc_handler	= ftrace_enable_sysctl,
516 	},
517 #endif
518 #ifdef CONFIG_STACK_TRACER
519 	{
520 		.procname	= "stack_tracer_enabled",
521 		.data		= &stack_tracer_enabled,
522 		.maxlen		= sizeof(int),
523 		.mode		= 0644,
524 		.proc_handler	= stack_trace_sysctl,
525 	},
526 #endif
527 #ifdef CONFIG_TRACING
528 	{
529 		.procname	= "ftrace_dump_on_oops",
530 		.data		= &ftrace_dump_on_oops,
531 		.maxlen		= sizeof(int),
532 		.mode		= 0644,
533 		.proc_handler	= proc_dointvec,
534 	},
535 #endif
536 #ifdef CONFIG_MODULES
537 	{
538 		.procname	= "modprobe",
539 		.data		= &modprobe_path,
540 		.maxlen		= KMOD_PATH_LEN,
541 		.mode		= 0644,
542 		.proc_handler	= proc_dostring,
543 	},
544 	{
545 		.procname	= "modules_disabled",
546 		.data		= &modules_disabled,
547 		.maxlen		= sizeof(int),
548 		.mode		= 0644,
549 		/* only handle a transition from default "0" to "1" */
550 		.proc_handler	= proc_dointvec_minmax,
551 		.extra1		= &one,
552 		.extra2		= &one,
553 	},
554 #endif
555 #ifdef CONFIG_HOTPLUG
556 	{
557 		.procname	= "hotplug",
558 		.data		= &uevent_helper,
559 		.maxlen		= UEVENT_HELPER_PATH_LEN,
560 		.mode		= 0644,
561 		.proc_handler	= proc_dostring,
562 	},
563 #endif
564 #ifdef CONFIG_CHR_DEV_SG
565 	{
566 		.procname	= "sg-big-buff",
567 		.data		= &sg_big_buff,
568 		.maxlen		= sizeof (int),
569 		.mode		= 0444,
570 		.proc_handler	= proc_dointvec,
571 	},
572 #endif
573 #ifdef CONFIG_BSD_PROCESS_ACCT
574 	{
575 		.procname	= "acct",
576 		.data		= &acct_parm,
577 		.maxlen		= 3*sizeof(int),
578 		.mode		= 0644,
579 		.proc_handler	= proc_dointvec,
580 	},
581 #endif
582 #ifdef CONFIG_MAGIC_SYSRQ
583 	{
584 		.procname	= "sysrq",
585 		.data		= &__sysrq_enabled,
586 		.maxlen		= sizeof (int),
587 		.mode		= 0644,
588 		.proc_handler	= sysrq_sysctl_handler,
589 	},
590 #endif
591 #ifdef CONFIG_PROC_SYSCTL
592 	{
593 		.procname	= "cad_pid",
594 		.data		= NULL,
595 		.maxlen		= sizeof (int),
596 		.mode		= 0600,
597 		.proc_handler	= proc_do_cad_pid,
598 	},
599 #endif
600 	{
601 		.procname	= "threads-max",
602 		.data		= &max_threads,
603 		.maxlen		= sizeof(int),
604 		.mode		= 0644,
605 		.proc_handler	= proc_dointvec,
606 	},
607 	{
608 		.procname	= "random",
609 		.mode		= 0555,
610 		.child		= random_table,
611 	},
612 	{
613 		.procname	= "overflowuid",
614 		.data		= &overflowuid,
615 		.maxlen		= sizeof(int),
616 		.mode		= 0644,
617 		.proc_handler	= proc_dointvec_minmax,
618 		.extra1		= &minolduid,
619 		.extra2		= &maxolduid,
620 	},
621 	{
622 		.procname	= "overflowgid",
623 		.data		= &overflowgid,
624 		.maxlen		= sizeof(int),
625 		.mode		= 0644,
626 		.proc_handler	= proc_dointvec_minmax,
627 		.extra1		= &minolduid,
628 		.extra2		= &maxolduid,
629 	},
630 #ifdef CONFIG_S390
631 #ifdef CONFIG_MATHEMU
632 	{
633 		.procname	= "ieee_emulation_warnings",
634 		.data		= &sysctl_ieee_emulation_warnings,
635 		.maxlen		= sizeof(int),
636 		.mode		= 0644,
637 		.proc_handler	= proc_dointvec,
638 	},
639 #endif
640 	{
641 		.procname	= "userprocess_debug",
642 		.data		= &show_unhandled_signals,
643 		.maxlen		= sizeof(int),
644 		.mode		= 0644,
645 		.proc_handler	= proc_dointvec,
646 	},
647 #endif
648 	{
649 		.procname	= "pid_max",
650 		.data		= &pid_max,
651 		.maxlen		= sizeof (int),
652 		.mode		= 0644,
653 		.proc_handler	= proc_dointvec_minmax,
654 		.extra1		= &pid_max_min,
655 		.extra2		= &pid_max_max,
656 	},
657 	{
658 		.procname	= "panic_on_oops",
659 		.data		= &panic_on_oops,
660 		.maxlen		= sizeof(int),
661 		.mode		= 0644,
662 		.proc_handler	= proc_dointvec,
663 	},
664 #if defined CONFIG_PRINTK
665 	{
666 		.procname	= "printk",
667 		.data		= &console_loglevel,
668 		.maxlen		= 4*sizeof(int),
669 		.mode		= 0644,
670 		.proc_handler	= proc_dointvec,
671 	},
672 	{
673 		.procname	= "printk_ratelimit",
674 		.data		= &printk_ratelimit_state.interval,
675 		.maxlen		= sizeof(int),
676 		.mode		= 0644,
677 		.proc_handler	= proc_dointvec_jiffies,
678 	},
679 	{
680 		.procname	= "printk_ratelimit_burst",
681 		.data		= &printk_ratelimit_state.burst,
682 		.maxlen		= sizeof(int),
683 		.mode		= 0644,
684 		.proc_handler	= proc_dointvec,
685 	},
686 	{
687 		.procname	= "printk_delay",
688 		.data		= &printk_delay_msec,
689 		.maxlen		= sizeof(int),
690 		.mode		= 0644,
691 		.proc_handler	= proc_dointvec_minmax,
692 		.extra1		= &zero,
693 		.extra2		= &ten_thousand,
694 	},
695 	{
696 		.procname	= "dmesg_restrict",
697 		.data		= &dmesg_restrict,
698 		.maxlen		= sizeof(int),
699 		.mode		= 0644,
700 		.proc_handler	= proc_dointvec_minmax,
701 		.extra1		= &zero,
702 		.extra2		= &one,
703 	},
704 	{
705 		.procname	= "kptr_restrict",
706 		.data		= &kptr_restrict,
707 		.maxlen		= sizeof(int),
708 		.mode		= 0644,
709 		.proc_handler	= proc_dointvec_minmax,
710 		.extra1		= &zero,
711 		.extra2		= &two,
712 	},
713 #endif
714 	{
715 		.procname	= "ngroups_max",
716 		.data		= &ngroups_max,
717 		.maxlen		= sizeof (int),
718 		.mode		= 0444,
719 		.proc_handler	= proc_dointvec,
720 	},
721 #if defined(CONFIG_LOCKUP_DETECTOR)
722 	{
723 		.procname       = "watchdog",
724 		.data           = &watchdog_enabled,
725 		.maxlen         = sizeof (int),
726 		.mode           = 0644,
727 		.proc_handler   = proc_dowatchdog_enabled,
728 	},
729 	{
730 		.procname	= "watchdog_thresh",
731 		.data		= &softlockup_thresh,
732 		.maxlen		= sizeof(int),
733 		.mode		= 0644,
734 		.proc_handler	= proc_dowatchdog_thresh,
735 		.extra1		= &neg_one,
736 		.extra2		= &sixty,
737 	},
738 	{
739 		.procname	= "softlockup_panic",
740 		.data		= &softlockup_panic,
741 		.maxlen		= sizeof(int),
742 		.mode		= 0644,
743 		.proc_handler	= proc_dointvec_minmax,
744 		.extra1		= &zero,
745 		.extra2		= &one,
746 	},
747 	{
748 		.procname       = "nmi_watchdog",
749 		.data           = &watchdog_enabled,
750 		.maxlen         = sizeof (int),
751 		.mode           = 0644,
752 		.proc_handler   = proc_dowatchdog_enabled,
753 	},
754 #endif
755 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
756 	{
757 		.procname       = "unknown_nmi_panic",
758 		.data           = &unknown_nmi_panic,
759 		.maxlen         = sizeof (int),
760 		.mode           = 0644,
761 		.proc_handler   = proc_dointvec,
762 	},
763 #endif
764 #if defined(CONFIG_X86)
765 	{
766 		.procname	= "panic_on_unrecovered_nmi",
767 		.data		= &panic_on_unrecovered_nmi,
768 		.maxlen		= sizeof(int),
769 		.mode		= 0644,
770 		.proc_handler	= proc_dointvec,
771 	},
772 	{
773 		.procname	= "panic_on_io_nmi",
774 		.data		= &panic_on_io_nmi,
775 		.maxlen		= sizeof(int),
776 		.mode		= 0644,
777 		.proc_handler	= proc_dointvec,
778 	},
779 	{
780 		.procname	= "bootloader_type",
781 		.data		= &bootloader_type,
782 		.maxlen		= sizeof (int),
783 		.mode		= 0444,
784 		.proc_handler	= proc_dointvec,
785 	},
786 	{
787 		.procname	= "bootloader_version",
788 		.data		= &bootloader_version,
789 		.maxlen		= sizeof (int),
790 		.mode		= 0444,
791 		.proc_handler	= proc_dointvec,
792 	},
793 	{
794 		.procname	= "kstack_depth_to_print",
795 		.data		= &kstack_depth_to_print,
796 		.maxlen		= sizeof(int),
797 		.mode		= 0644,
798 		.proc_handler	= proc_dointvec,
799 	},
800 	{
801 		.procname	= "io_delay_type",
802 		.data		= &io_delay_type,
803 		.maxlen		= sizeof(int),
804 		.mode		= 0644,
805 		.proc_handler	= proc_dointvec,
806 	},
807 #endif
808 #if defined(CONFIG_MMU)
809 	{
810 		.procname	= "randomize_va_space",
811 		.data		= &randomize_va_space,
812 		.maxlen		= sizeof(int),
813 		.mode		= 0644,
814 		.proc_handler	= proc_dointvec,
815 	},
816 #endif
817 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
818 	{
819 		.procname	= "spin_retry",
820 		.data		= &spin_retry,
821 		.maxlen		= sizeof (int),
822 		.mode		= 0644,
823 		.proc_handler	= proc_dointvec,
824 	},
825 #endif
826 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
827 	{
828 		.procname	= "acpi_video_flags",
829 		.data		= &acpi_realmode_flags,
830 		.maxlen		= sizeof (unsigned long),
831 		.mode		= 0644,
832 		.proc_handler	= proc_doulongvec_minmax,
833 	},
834 #endif
835 #ifdef CONFIG_IA64
836 	{
837 		.procname	= "ignore-unaligned-usertrap",
838 		.data		= &no_unaligned_warning,
839 		.maxlen		= sizeof (int),
840 	 	.mode		= 0644,
841 		.proc_handler	= proc_dointvec,
842 	},
843 	{
844 		.procname	= "unaligned-dump-stack",
845 		.data		= &unaligned_dump_stack,
846 		.maxlen		= sizeof (int),
847 		.mode		= 0644,
848 		.proc_handler	= proc_dointvec,
849 	},
850 #endif
851 #ifdef CONFIG_DETECT_HUNG_TASK
852 	{
853 		.procname	= "hung_task_panic",
854 		.data		= &sysctl_hung_task_panic,
855 		.maxlen		= sizeof(int),
856 		.mode		= 0644,
857 		.proc_handler	= proc_dointvec_minmax,
858 		.extra1		= &zero,
859 		.extra2		= &one,
860 	},
861 	{
862 		.procname	= "hung_task_check_count",
863 		.data		= &sysctl_hung_task_check_count,
864 		.maxlen		= sizeof(unsigned long),
865 		.mode		= 0644,
866 		.proc_handler	= proc_doulongvec_minmax,
867 	},
868 	{
869 		.procname	= "hung_task_timeout_secs",
870 		.data		= &sysctl_hung_task_timeout_secs,
871 		.maxlen		= sizeof(unsigned long),
872 		.mode		= 0644,
873 		.proc_handler	= proc_dohung_task_timeout_secs,
874 	},
875 	{
876 		.procname	= "hung_task_warnings",
877 		.data		= &sysctl_hung_task_warnings,
878 		.maxlen		= sizeof(unsigned long),
879 		.mode		= 0644,
880 		.proc_handler	= proc_doulongvec_minmax,
881 	},
882 #endif
883 #ifdef CONFIG_COMPAT
884 	{
885 		.procname	= "compat-log",
886 		.data		= &compat_log,
887 		.maxlen		= sizeof (int),
888 	 	.mode		= 0644,
889 		.proc_handler	= proc_dointvec,
890 	},
891 #endif
892 #ifdef CONFIG_RT_MUTEXES
893 	{
894 		.procname	= "max_lock_depth",
895 		.data		= &max_lock_depth,
896 		.maxlen		= sizeof(int),
897 		.mode		= 0644,
898 		.proc_handler	= proc_dointvec,
899 	},
900 #endif
901 	{
902 		.procname	= "poweroff_cmd",
903 		.data		= &poweroff_cmd,
904 		.maxlen		= POWEROFF_CMD_PATH_LEN,
905 		.mode		= 0644,
906 		.proc_handler	= proc_dostring,
907 	},
908 #ifdef CONFIG_KEYS
909 	{
910 		.procname	= "keys",
911 		.mode		= 0555,
912 		.child		= key_sysctls,
913 	},
914 #endif
915 #ifdef CONFIG_RCU_TORTURE_TEST
916 	{
917 		.procname       = "rcutorture_runnable",
918 		.data           = &rcutorture_runnable,
919 		.maxlen         = sizeof(int),
920 		.mode           = 0644,
921 		.proc_handler	= proc_dointvec,
922 	},
923 #endif
924 #ifdef CONFIG_PERF_EVENTS
925 	{
926 		.procname	= "perf_event_paranoid",
927 		.data		= &sysctl_perf_event_paranoid,
928 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
929 		.mode		= 0644,
930 		.proc_handler	= proc_dointvec,
931 	},
932 	{
933 		.procname	= "perf_event_mlock_kb",
934 		.data		= &sysctl_perf_event_mlock,
935 		.maxlen		= sizeof(sysctl_perf_event_mlock),
936 		.mode		= 0644,
937 		.proc_handler	= proc_dointvec,
938 	},
939 	{
940 		.procname	= "perf_event_max_sample_rate",
941 		.data		= &sysctl_perf_event_sample_rate,
942 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
943 		.mode		= 0644,
944 		.proc_handler	= perf_proc_update_handler,
945 	},
946 #endif
947 #ifdef CONFIG_KMEMCHECK
948 	{
949 		.procname	= "kmemcheck",
950 		.data		= &kmemcheck_enabled,
951 		.maxlen		= sizeof(int),
952 		.mode		= 0644,
953 		.proc_handler	= proc_dointvec,
954 	},
955 #endif
956 #ifdef CONFIG_BLOCK
957 	{
958 		.procname	= "blk_iopoll",
959 		.data		= &blk_iopoll_enabled,
960 		.maxlen		= sizeof(int),
961 		.mode		= 0644,
962 		.proc_handler	= proc_dointvec,
963 	},
964 #endif
965 	{ }
966 };
967 
968 static struct ctl_table vm_table[] = {
969 	{
970 		.procname	= "overcommit_memory",
971 		.data		= &sysctl_overcommit_memory,
972 		.maxlen		= sizeof(sysctl_overcommit_memory),
973 		.mode		= 0644,
974 		.proc_handler	= proc_dointvec,
975 	},
976 	{
977 		.procname	= "panic_on_oom",
978 		.data		= &sysctl_panic_on_oom,
979 		.maxlen		= sizeof(sysctl_panic_on_oom),
980 		.mode		= 0644,
981 		.proc_handler	= proc_dointvec,
982 	},
983 	{
984 		.procname	= "oom_kill_allocating_task",
985 		.data		= &sysctl_oom_kill_allocating_task,
986 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
987 		.mode		= 0644,
988 		.proc_handler	= proc_dointvec,
989 	},
990 	{
991 		.procname	= "oom_dump_tasks",
992 		.data		= &sysctl_oom_dump_tasks,
993 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
994 		.mode		= 0644,
995 		.proc_handler	= proc_dointvec,
996 	},
997 	{
998 		.procname	= "overcommit_ratio",
999 		.data		= &sysctl_overcommit_ratio,
1000 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1001 		.mode		= 0644,
1002 		.proc_handler	= proc_dointvec,
1003 	},
1004 	{
1005 		.procname	= "page-cluster",
1006 		.data		= &page_cluster,
1007 		.maxlen		= sizeof(int),
1008 		.mode		= 0644,
1009 		.proc_handler	= proc_dointvec,
1010 	},
1011 	{
1012 		.procname	= "dirty_background_ratio",
1013 		.data		= &dirty_background_ratio,
1014 		.maxlen		= sizeof(dirty_background_ratio),
1015 		.mode		= 0644,
1016 		.proc_handler	= dirty_background_ratio_handler,
1017 		.extra1		= &zero,
1018 		.extra2		= &one_hundred,
1019 	},
1020 	{
1021 		.procname	= "dirty_background_bytes",
1022 		.data		= &dirty_background_bytes,
1023 		.maxlen		= sizeof(dirty_background_bytes),
1024 		.mode		= 0644,
1025 		.proc_handler	= dirty_background_bytes_handler,
1026 		.extra1		= &one_ul,
1027 	},
1028 	{
1029 		.procname	= "dirty_ratio",
1030 		.data		= &vm_dirty_ratio,
1031 		.maxlen		= sizeof(vm_dirty_ratio),
1032 		.mode		= 0644,
1033 		.proc_handler	= dirty_ratio_handler,
1034 		.extra1		= &zero,
1035 		.extra2		= &one_hundred,
1036 	},
1037 	{
1038 		.procname	= "dirty_bytes",
1039 		.data		= &vm_dirty_bytes,
1040 		.maxlen		= sizeof(vm_dirty_bytes),
1041 		.mode		= 0644,
1042 		.proc_handler	= dirty_bytes_handler,
1043 		.extra1		= &dirty_bytes_min,
1044 	},
1045 	{
1046 		.procname	= "dirty_writeback_centisecs",
1047 		.data		= &dirty_writeback_interval,
1048 		.maxlen		= sizeof(dirty_writeback_interval),
1049 		.mode		= 0644,
1050 		.proc_handler	= dirty_writeback_centisecs_handler,
1051 	},
1052 	{
1053 		.procname	= "dirty_expire_centisecs",
1054 		.data		= &dirty_expire_interval,
1055 		.maxlen		= sizeof(dirty_expire_interval),
1056 		.mode		= 0644,
1057 		.proc_handler	= proc_dointvec,
1058 	},
1059 	{
1060 		.procname	= "nr_pdflush_threads",
1061 		.data		= &nr_pdflush_threads,
1062 		.maxlen		= sizeof nr_pdflush_threads,
1063 		.mode		= 0444 /* read-only*/,
1064 		.proc_handler	= proc_dointvec,
1065 	},
1066 	{
1067 		.procname	= "swappiness",
1068 		.data		= &vm_swappiness,
1069 		.maxlen		= sizeof(vm_swappiness),
1070 		.mode		= 0644,
1071 		.proc_handler	= proc_dointvec_minmax,
1072 		.extra1		= &zero,
1073 		.extra2		= &one_hundred,
1074 	},
1075 #ifdef CONFIG_HUGETLB_PAGE
1076 	{
1077 		.procname	= "nr_hugepages",
1078 		.data		= NULL,
1079 		.maxlen		= sizeof(unsigned long),
1080 		.mode		= 0644,
1081 		.proc_handler	= hugetlb_sysctl_handler,
1082 		.extra1		= (void *)&hugetlb_zero,
1083 		.extra2		= (void *)&hugetlb_infinity,
1084 	},
1085 #ifdef CONFIG_NUMA
1086 	{
1087 		.procname       = "nr_hugepages_mempolicy",
1088 		.data           = NULL,
1089 		.maxlen         = sizeof(unsigned long),
1090 		.mode           = 0644,
1091 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1092 		.extra1		= (void *)&hugetlb_zero,
1093 		.extra2		= (void *)&hugetlb_infinity,
1094 	},
1095 #endif
1096 	 {
1097 		.procname	= "hugetlb_shm_group",
1098 		.data		= &sysctl_hugetlb_shm_group,
1099 		.maxlen		= sizeof(gid_t),
1100 		.mode		= 0644,
1101 		.proc_handler	= proc_dointvec,
1102 	 },
1103 	 {
1104 		.procname	= "hugepages_treat_as_movable",
1105 		.data		= &hugepages_treat_as_movable,
1106 		.maxlen		= sizeof(int),
1107 		.mode		= 0644,
1108 		.proc_handler	= hugetlb_treat_movable_handler,
1109 	},
1110 	{
1111 		.procname	= "nr_overcommit_hugepages",
1112 		.data		= NULL,
1113 		.maxlen		= sizeof(unsigned long),
1114 		.mode		= 0644,
1115 		.proc_handler	= hugetlb_overcommit_handler,
1116 		.extra1		= (void *)&hugetlb_zero,
1117 		.extra2		= (void *)&hugetlb_infinity,
1118 	},
1119 #endif
1120 	{
1121 		.procname	= "lowmem_reserve_ratio",
1122 		.data		= &sysctl_lowmem_reserve_ratio,
1123 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1124 		.mode		= 0644,
1125 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1126 	},
1127 	{
1128 		.procname	= "drop_caches",
1129 		.data		= &sysctl_drop_caches,
1130 		.maxlen		= sizeof(int),
1131 		.mode		= 0644,
1132 		.proc_handler	= drop_caches_sysctl_handler,
1133 	},
1134 #ifdef CONFIG_COMPACTION
1135 	{
1136 		.procname	= "compact_memory",
1137 		.data		= &sysctl_compact_memory,
1138 		.maxlen		= sizeof(int),
1139 		.mode		= 0200,
1140 		.proc_handler	= sysctl_compaction_handler,
1141 	},
1142 	{
1143 		.procname	= "extfrag_threshold",
1144 		.data		= &sysctl_extfrag_threshold,
1145 		.maxlen		= sizeof(int),
1146 		.mode		= 0644,
1147 		.proc_handler	= sysctl_extfrag_handler,
1148 		.extra1		= &min_extfrag_threshold,
1149 		.extra2		= &max_extfrag_threshold,
1150 	},
1151 
1152 #endif /* CONFIG_COMPACTION */
1153 	{
1154 		.procname	= "min_free_kbytes",
1155 		.data		= &min_free_kbytes,
1156 		.maxlen		= sizeof(min_free_kbytes),
1157 		.mode		= 0644,
1158 		.proc_handler	= min_free_kbytes_sysctl_handler,
1159 		.extra1		= &zero,
1160 	},
1161 	{
1162 		.procname	= "percpu_pagelist_fraction",
1163 		.data		= &percpu_pagelist_fraction,
1164 		.maxlen		= sizeof(percpu_pagelist_fraction),
1165 		.mode		= 0644,
1166 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1167 		.extra1		= &min_percpu_pagelist_fract,
1168 	},
1169 #ifdef CONFIG_MMU
1170 	{
1171 		.procname	= "max_map_count",
1172 		.data		= &sysctl_max_map_count,
1173 		.maxlen		= sizeof(sysctl_max_map_count),
1174 		.mode		= 0644,
1175 		.proc_handler	= proc_dointvec_minmax,
1176 		.extra1		= &zero,
1177 	},
1178 #else
1179 	{
1180 		.procname	= "nr_trim_pages",
1181 		.data		= &sysctl_nr_trim_pages,
1182 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1183 		.mode		= 0644,
1184 		.proc_handler	= proc_dointvec_minmax,
1185 		.extra1		= &zero,
1186 	},
1187 #endif
1188 	{
1189 		.procname	= "laptop_mode",
1190 		.data		= &laptop_mode,
1191 		.maxlen		= sizeof(laptop_mode),
1192 		.mode		= 0644,
1193 		.proc_handler	= proc_dointvec_jiffies,
1194 	},
1195 	{
1196 		.procname	= "block_dump",
1197 		.data		= &block_dump,
1198 		.maxlen		= sizeof(block_dump),
1199 		.mode		= 0644,
1200 		.proc_handler	= proc_dointvec,
1201 		.extra1		= &zero,
1202 	},
1203 	{
1204 		.procname	= "vfs_cache_pressure",
1205 		.data		= &sysctl_vfs_cache_pressure,
1206 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1207 		.mode		= 0644,
1208 		.proc_handler	= proc_dointvec,
1209 		.extra1		= &zero,
1210 	},
1211 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1212 	{
1213 		.procname	= "legacy_va_layout",
1214 		.data		= &sysctl_legacy_va_layout,
1215 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1216 		.mode		= 0644,
1217 		.proc_handler	= proc_dointvec,
1218 		.extra1		= &zero,
1219 	},
1220 #endif
1221 #ifdef CONFIG_NUMA
1222 	{
1223 		.procname	= "zone_reclaim_mode",
1224 		.data		= &zone_reclaim_mode,
1225 		.maxlen		= sizeof(zone_reclaim_mode),
1226 		.mode		= 0644,
1227 		.proc_handler	= proc_dointvec,
1228 		.extra1		= &zero,
1229 	},
1230 	{
1231 		.procname	= "min_unmapped_ratio",
1232 		.data		= &sysctl_min_unmapped_ratio,
1233 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1234 		.mode		= 0644,
1235 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1236 		.extra1		= &zero,
1237 		.extra2		= &one_hundred,
1238 	},
1239 	{
1240 		.procname	= "min_slab_ratio",
1241 		.data		= &sysctl_min_slab_ratio,
1242 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1243 		.mode		= 0644,
1244 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1245 		.extra1		= &zero,
1246 		.extra2		= &one_hundred,
1247 	},
1248 #endif
1249 #ifdef CONFIG_SMP
1250 	{
1251 		.procname	= "stat_interval",
1252 		.data		= &sysctl_stat_interval,
1253 		.maxlen		= sizeof(sysctl_stat_interval),
1254 		.mode		= 0644,
1255 		.proc_handler	= proc_dointvec_jiffies,
1256 	},
1257 #endif
1258 #ifdef CONFIG_MMU
1259 	{
1260 		.procname	= "mmap_min_addr",
1261 		.data		= &dac_mmap_min_addr,
1262 		.maxlen		= sizeof(unsigned long),
1263 		.mode		= 0644,
1264 		.proc_handler	= mmap_min_addr_handler,
1265 	},
1266 #endif
1267 #ifdef CONFIG_NUMA
1268 	{
1269 		.procname	= "numa_zonelist_order",
1270 		.data		= &numa_zonelist_order,
1271 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1272 		.mode		= 0644,
1273 		.proc_handler	= numa_zonelist_order_handler,
1274 	},
1275 #endif
1276 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1277    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1278 	{
1279 		.procname	= "vdso_enabled",
1280 		.data		= &vdso_enabled,
1281 		.maxlen		= sizeof(vdso_enabled),
1282 		.mode		= 0644,
1283 		.proc_handler	= proc_dointvec,
1284 		.extra1		= &zero,
1285 	},
1286 #endif
1287 #ifdef CONFIG_HIGHMEM
1288 	{
1289 		.procname	= "highmem_is_dirtyable",
1290 		.data		= &vm_highmem_is_dirtyable,
1291 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1292 		.mode		= 0644,
1293 		.proc_handler	= proc_dointvec_minmax,
1294 		.extra1		= &zero,
1295 		.extra2		= &one,
1296 	},
1297 #endif
1298 	{
1299 		.procname	= "scan_unevictable_pages",
1300 		.data		= &scan_unevictable_pages,
1301 		.maxlen		= sizeof(scan_unevictable_pages),
1302 		.mode		= 0644,
1303 		.proc_handler	= scan_unevictable_handler,
1304 	},
1305 #ifdef CONFIG_MEMORY_FAILURE
1306 	{
1307 		.procname	= "memory_failure_early_kill",
1308 		.data		= &sysctl_memory_failure_early_kill,
1309 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1310 		.mode		= 0644,
1311 		.proc_handler	= proc_dointvec_minmax,
1312 		.extra1		= &zero,
1313 		.extra2		= &one,
1314 	},
1315 	{
1316 		.procname	= "memory_failure_recovery",
1317 		.data		= &sysctl_memory_failure_recovery,
1318 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1319 		.mode		= 0644,
1320 		.proc_handler	= proc_dointvec_minmax,
1321 		.extra1		= &zero,
1322 		.extra2		= &one,
1323 	},
1324 #endif
1325 	{ }
1326 };
1327 
1328 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1329 static struct ctl_table binfmt_misc_table[] = {
1330 	{ }
1331 };
1332 #endif
1333 
1334 static struct ctl_table fs_table[] = {
1335 	{
1336 		.procname	= "inode-nr",
1337 		.data		= &inodes_stat,
1338 		.maxlen		= 2*sizeof(int),
1339 		.mode		= 0444,
1340 		.proc_handler	= proc_nr_inodes,
1341 	},
1342 	{
1343 		.procname	= "inode-state",
1344 		.data		= &inodes_stat,
1345 		.maxlen		= 7*sizeof(int),
1346 		.mode		= 0444,
1347 		.proc_handler	= proc_nr_inodes,
1348 	},
1349 	{
1350 		.procname	= "file-nr",
1351 		.data		= &files_stat,
1352 		.maxlen		= sizeof(files_stat),
1353 		.mode		= 0444,
1354 		.proc_handler	= proc_nr_files,
1355 	},
1356 	{
1357 		.procname	= "file-max",
1358 		.data		= &files_stat.max_files,
1359 		.maxlen		= sizeof(files_stat.max_files),
1360 		.mode		= 0644,
1361 		.proc_handler	= proc_doulongvec_minmax,
1362 	},
1363 	{
1364 		.procname	= "nr_open",
1365 		.data		= &sysctl_nr_open,
1366 		.maxlen		= sizeof(int),
1367 		.mode		= 0644,
1368 		.proc_handler	= proc_dointvec_minmax,
1369 		.extra1		= &sysctl_nr_open_min,
1370 		.extra2		= &sysctl_nr_open_max,
1371 	},
1372 	{
1373 		.procname	= "dentry-state",
1374 		.data		= &dentry_stat,
1375 		.maxlen		= 6*sizeof(int),
1376 		.mode		= 0444,
1377 		.proc_handler	= proc_nr_dentry,
1378 	},
1379 	{
1380 		.procname	= "overflowuid",
1381 		.data		= &fs_overflowuid,
1382 		.maxlen		= sizeof(int),
1383 		.mode		= 0644,
1384 		.proc_handler	= proc_dointvec_minmax,
1385 		.extra1		= &minolduid,
1386 		.extra2		= &maxolduid,
1387 	},
1388 	{
1389 		.procname	= "overflowgid",
1390 		.data		= &fs_overflowgid,
1391 		.maxlen		= sizeof(int),
1392 		.mode		= 0644,
1393 		.proc_handler	= proc_dointvec_minmax,
1394 		.extra1		= &minolduid,
1395 		.extra2		= &maxolduid,
1396 	},
1397 #ifdef CONFIG_FILE_LOCKING
1398 	{
1399 		.procname	= "leases-enable",
1400 		.data		= &leases_enable,
1401 		.maxlen		= sizeof(int),
1402 		.mode		= 0644,
1403 		.proc_handler	= proc_dointvec,
1404 	},
1405 #endif
1406 #ifdef CONFIG_DNOTIFY
1407 	{
1408 		.procname	= "dir-notify-enable",
1409 		.data		= &dir_notify_enable,
1410 		.maxlen		= sizeof(int),
1411 		.mode		= 0644,
1412 		.proc_handler	= proc_dointvec,
1413 	},
1414 #endif
1415 #ifdef CONFIG_MMU
1416 #ifdef CONFIG_FILE_LOCKING
1417 	{
1418 		.procname	= "lease-break-time",
1419 		.data		= &lease_break_time,
1420 		.maxlen		= sizeof(int),
1421 		.mode		= 0644,
1422 		.proc_handler	= proc_dointvec,
1423 	},
1424 #endif
1425 #ifdef CONFIG_AIO
1426 	{
1427 		.procname	= "aio-nr",
1428 		.data		= &aio_nr,
1429 		.maxlen		= sizeof(aio_nr),
1430 		.mode		= 0444,
1431 		.proc_handler	= proc_doulongvec_minmax,
1432 	},
1433 	{
1434 		.procname	= "aio-max-nr",
1435 		.data		= &aio_max_nr,
1436 		.maxlen		= sizeof(aio_max_nr),
1437 		.mode		= 0644,
1438 		.proc_handler	= proc_doulongvec_minmax,
1439 	},
1440 #endif /* CONFIG_AIO */
1441 #ifdef CONFIG_INOTIFY_USER
1442 	{
1443 		.procname	= "inotify",
1444 		.mode		= 0555,
1445 		.child		= inotify_table,
1446 	},
1447 #endif
1448 #ifdef CONFIG_EPOLL
1449 	{
1450 		.procname	= "epoll",
1451 		.mode		= 0555,
1452 		.child		= epoll_table,
1453 	},
1454 #endif
1455 #endif
1456 	{
1457 		.procname	= "suid_dumpable",
1458 		.data		= &suid_dumpable,
1459 		.maxlen		= sizeof(int),
1460 		.mode		= 0644,
1461 		.proc_handler	= proc_dointvec_minmax,
1462 		.extra1		= &zero,
1463 		.extra2		= &two,
1464 	},
1465 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1466 	{
1467 		.procname	= "binfmt_misc",
1468 		.mode		= 0555,
1469 		.child		= binfmt_misc_table,
1470 	},
1471 #endif
1472 	{
1473 		.procname	= "pipe-max-size",
1474 		.data		= &pipe_max_size,
1475 		.maxlen		= sizeof(int),
1476 		.mode		= 0644,
1477 		.proc_handler	= &pipe_proc_fn,
1478 		.extra1		= &pipe_min_size,
1479 	},
1480 	{ }
1481 };
1482 
1483 static struct ctl_table debug_table[] = {
1484 #if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
1485     defined(CONFIG_S390)
1486 	{
1487 		.procname	= "exception-trace",
1488 		.data		= &show_unhandled_signals,
1489 		.maxlen		= sizeof(int),
1490 		.mode		= 0644,
1491 		.proc_handler	= proc_dointvec
1492 	},
1493 #endif
1494 #if defined(CONFIG_OPTPROBES)
1495 	{
1496 		.procname	= "kprobes-optimization",
1497 		.data		= &sysctl_kprobes_optimization,
1498 		.maxlen		= sizeof(int),
1499 		.mode		= 0644,
1500 		.proc_handler	= proc_kprobes_optimization_handler,
1501 		.extra1		= &zero,
1502 		.extra2		= &one,
1503 	},
1504 #endif
1505 	{ }
1506 };
1507 
1508 static struct ctl_table dev_table[] = {
1509 	{ }
1510 };
1511 
1512 static DEFINE_SPINLOCK(sysctl_lock);
1513 
1514 /* called under sysctl_lock */
1515 static int use_table(struct ctl_table_header *p)
1516 {
1517 	if (unlikely(p->unregistering))
1518 		return 0;
1519 	p->used++;
1520 	return 1;
1521 }
1522 
1523 /* called under sysctl_lock */
1524 static void unuse_table(struct ctl_table_header *p)
1525 {
1526 	if (!--p->used)
1527 		if (unlikely(p->unregistering))
1528 			complete(p->unregistering);
1529 }
1530 
1531 /* called under sysctl_lock, will reacquire if has to wait */
1532 static void start_unregistering(struct ctl_table_header *p)
1533 {
1534 	/*
1535 	 * if p->used is 0, nobody will ever touch that entry again;
1536 	 * we'll eliminate all paths to it before dropping sysctl_lock
1537 	 */
1538 	if (unlikely(p->used)) {
1539 		struct completion wait;
1540 		init_completion(&wait);
1541 		p->unregistering = &wait;
1542 		spin_unlock(&sysctl_lock);
1543 		wait_for_completion(&wait);
1544 		spin_lock(&sysctl_lock);
1545 	} else {
1546 		/* anything non-NULL; we'll never dereference it */
1547 		p->unregistering = ERR_PTR(-EINVAL);
1548 	}
1549 	/*
1550 	 * do not remove from the list until nobody holds it; walking the
1551 	 * list in do_sysctl() relies on that.
1552 	 */
1553 	list_del_init(&p->ctl_entry);
1554 }
1555 
1556 void sysctl_head_get(struct ctl_table_header *head)
1557 {
1558 	spin_lock(&sysctl_lock);
1559 	head->count++;
1560 	spin_unlock(&sysctl_lock);
1561 }
1562 
1563 static void free_head(struct rcu_head *rcu)
1564 {
1565 	kfree(container_of(rcu, struct ctl_table_header, rcu));
1566 }
1567 
1568 void sysctl_head_put(struct ctl_table_header *head)
1569 {
1570 	spin_lock(&sysctl_lock);
1571 	if (!--head->count)
1572 		call_rcu(&head->rcu, free_head);
1573 	spin_unlock(&sysctl_lock);
1574 }
1575 
1576 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1577 {
1578 	if (!head)
1579 		BUG();
1580 	spin_lock(&sysctl_lock);
1581 	if (!use_table(head))
1582 		head = ERR_PTR(-ENOENT);
1583 	spin_unlock(&sysctl_lock);
1584 	return head;
1585 }
1586 
1587 void sysctl_head_finish(struct ctl_table_header *head)
1588 {
1589 	if (!head)
1590 		return;
1591 	spin_lock(&sysctl_lock);
1592 	unuse_table(head);
1593 	spin_unlock(&sysctl_lock);
1594 }
1595 
1596 static struct ctl_table_set *
1597 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1598 {
1599 	struct ctl_table_set *set = &root->default_set;
1600 	if (root->lookup)
1601 		set = root->lookup(root, namespaces);
1602 	return set;
1603 }
1604 
1605 static struct list_head *
1606 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1607 {
1608 	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1609 	return &set->list;
1610 }
1611 
1612 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1613 					    struct ctl_table_header *prev)
1614 {
1615 	struct ctl_table_root *root;
1616 	struct list_head *header_list;
1617 	struct ctl_table_header *head;
1618 	struct list_head *tmp;
1619 
1620 	spin_lock(&sysctl_lock);
1621 	if (prev) {
1622 		head = prev;
1623 		tmp = &prev->ctl_entry;
1624 		unuse_table(prev);
1625 		goto next;
1626 	}
1627 	tmp = &root_table_header.ctl_entry;
1628 	for (;;) {
1629 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1630 
1631 		if (!use_table(head))
1632 			goto next;
1633 		spin_unlock(&sysctl_lock);
1634 		return head;
1635 	next:
1636 		root = head->root;
1637 		tmp = tmp->next;
1638 		header_list = lookup_header_list(root, namespaces);
1639 		if (tmp != header_list)
1640 			continue;
1641 
1642 		do {
1643 			root = list_entry(root->root_list.next,
1644 					struct ctl_table_root, root_list);
1645 			if (root == &sysctl_table_root)
1646 				goto out;
1647 			header_list = lookup_header_list(root, namespaces);
1648 		} while (list_empty(header_list));
1649 		tmp = header_list->next;
1650 	}
1651 out:
1652 	spin_unlock(&sysctl_lock);
1653 	return NULL;
1654 }
1655 
1656 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1657 {
1658 	return __sysctl_head_next(current->nsproxy, prev);
1659 }
1660 
1661 void register_sysctl_root(struct ctl_table_root *root)
1662 {
1663 	spin_lock(&sysctl_lock);
1664 	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1665 	spin_unlock(&sysctl_lock);
1666 }
1667 
1668 /*
1669  * sysctl_perm does NOT grant the superuser all rights automatically, because
1670  * some sysctl variables are readonly even to root.
1671  */
1672 
1673 static int test_perm(int mode, int op)
1674 {
1675 	if (!current_euid())
1676 		mode >>= 6;
1677 	else if (in_egroup_p(0))
1678 		mode >>= 3;
1679 	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1680 		return 0;
1681 	return -EACCES;
1682 }
1683 
1684 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1685 {
1686 	int mode;
1687 
1688 	if (root->permissions)
1689 		mode = root->permissions(root, current->nsproxy, table);
1690 	else
1691 		mode = table->mode;
1692 
1693 	return test_perm(mode, op);
1694 }
1695 
1696 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1697 {
1698 	for (; table->procname; table++) {
1699 		table->parent = parent;
1700 		if (table->child)
1701 			sysctl_set_parent(table, table->child);
1702 	}
1703 }
1704 
1705 static __init int sysctl_init(void)
1706 {
1707 	sysctl_set_parent(NULL, root_table);
1708 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1709 	sysctl_check_table(current->nsproxy, root_table);
1710 #endif
1711 	return 0;
1712 }
1713 
1714 core_initcall(sysctl_init);
1715 
1716 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1717 				      struct ctl_table *table)
1718 {
1719 	struct ctl_table *p;
1720 	const char *s = branch->procname;
1721 
1722 	/* branch should have named subdirectory as its first element */
1723 	if (!s || !branch->child)
1724 		return NULL;
1725 
1726 	/* ... and nothing else */
1727 	if (branch[1].procname)
1728 		return NULL;
1729 
1730 	/* table should contain subdirectory with the same name */
1731 	for (p = table; p->procname; p++) {
1732 		if (!p->child)
1733 			continue;
1734 		if (p->procname && strcmp(p->procname, s) == 0)
1735 			return p;
1736 	}
1737 	return NULL;
1738 }
1739 
1740 /* see if attaching q to p would be an improvement */
1741 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1742 {
1743 	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1744 	struct ctl_table *next;
1745 	int is_better = 0;
1746 	int not_in_parent = !p->attached_by;
1747 
1748 	while ((next = is_branch_in(by, to)) != NULL) {
1749 		if (by == q->attached_by)
1750 			is_better = 1;
1751 		if (to == p->attached_by)
1752 			not_in_parent = 1;
1753 		by = by->child;
1754 		to = next->child;
1755 	}
1756 
1757 	if (is_better && not_in_parent) {
1758 		q->attached_by = by;
1759 		q->attached_to = to;
1760 		q->parent = p;
1761 	}
1762 }
1763 
1764 /**
1765  * __register_sysctl_paths - register a sysctl hierarchy
1766  * @root: List of sysctl headers to register on
1767  * @namespaces: Data to compute which lists of sysctl entries are visible
1768  * @path: The path to the directory the sysctl table is in.
1769  * @table: the top-level table structure
1770  *
1771  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1772  * array. A completely 0 filled entry terminates the table.
1773  *
1774  * The members of the &struct ctl_table structure are used as follows:
1775  *
1776  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1777  *            enter a sysctl file
1778  *
1779  * data - a pointer to data for use by proc_handler
1780  *
1781  * maxlen - the maximum size in bytes of the data
1782  *
1783  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1784  *
1785  * child - a pointer to the child sysctl table if this entry is a directory, or
1786  *         %NULL.
1787  *
1788  * proc_handler - the text handler routine (described below)
1789  *
1790  * de - for internal use by the sysctl routines
1791  *
1792  * extra1, extra2 - extra pointers usable by the proc handler routines
1793  *
1794  * Leaf nodes in the sysctl tree will be represented by a single file
1795  * under /proc; non-leaf nodes will be represented by directories.
1796  *
1797  * sysctl(2) can automatically manage read and write requests through
1798  * the sysctl table.  The data and maxlen fields of the ctl_table
1799  * struct enable minimal validation of the values being written to be
1800  * performed, and the mode field allows minimal authentication.
1801  *
1802  * There must be a proc_handler routine for any terminal nodes
1803  * mirrored under /proc/sys (non-terminals are handled by a built-in
1804  * directory handler).  Several default handlers are available to
1805  * cover common cases -
1806  *
1807  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1808  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1809  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1810  *
1811  * It is the handler's job to read the input buffer from user memory
1812  * and process it. The handler should return 0 on success.
1813  *
1814  * This routine returns %NULL on a failure to register, and a pointer
1815  * to the table header on success.
1816  */
1817 struct ctl_table_header *__register_sysctl_paths(
1818 	struct ctl_table_root *root,
1819 	struct nsproxy *namespaces,
1820 	const struct ctl_path *path, struct ctl_table *table)
1821 {
1822 	struct ctl_table_header *header;
1823 	struct ctl_table *new, **prevp;
1824 	unsigned int n, npath;
1825 	struct ctl_table_set *set;
1826 
1827 	/* Count the path components */
1828 	for (npath = 0; path[npath].procname; ++npath)
1829 		;
1830 
1831 	/*
1832 	 * For each path component, allocate a 2-element ctl_table array.
1833 	 * The first array element will be filled with the sysctl entry
1834 	 * for this, the second will be the sentinel (procname == 0).
1835 	 *
1836 	 * We allocate everything in one go so that we don't have to
1837 	 * worry about freeing additional memory in unregister_sysctl_table.
1838 	 */
1839 	header = kzalloc(sizeof(struct ctl_table_header) +
1840 			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1841 	if (!header)
1842 		return NULL;
1843 
1844 	new = (struct ctl_table *) (header + 1);
1845 
1846 	/* Now connect the dots */
1847 	prevp = &header->ctl_table;
1848 	for (n = 0; n < npath; ++n, ++path) {
1849 		/* Copy the procname */
1850 		new->procname = path->procname;
1851 		new->mode     = 0555;
1852 
1853 		*prevp = new;
1854 		prevp = &new->child;
1855 
1856 		new += 2;
1857 	}
1858 	*prevp = table;
1859 	header->ctl_table_arg = table;
1860 
1861 	INIT_LIST_HEAD(&header->ctl_entry);
1862 	header->used = 0;
1863 	header->unregistering = NULL;
1864 	header->root = root;
1865 	sysctl_set_parent(NULL, header->ctl_table);
1866 	header->count = 1;
1867 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1868 	if (sysctl_check_table(namespaces, header->ctl_table)) {
1869 		kfree(header);
1870 		return NULL;
1871 	}
1872 #endif
1873 	spin_lock(&sysctl_lock);
1874 	header->set = lookup_header_set(root, namespaces);
1875 	header->attached_by = header->ctl_table;
1876 	header->attached_to = root_table;
1877 	header->parent = &root_table_header;
1878 	for (set = header->set; set; set = set->parent) {
1879 		struct ctl_table_header *p;
1880 		list_for_each_entry(p, &set->list, ctl_entry) {
1881 			if (p->unregistering)
1882 				continue;
1883 			try_attach(p, header);
1884 		}
1885 	}
1886 	header->parent->count++;
1887 	list_add_tail(&header->ctl_entry, &header->set->list);
1888 	spin_unlock(&sysctl_lock);
1889 
1890 	return header;
1891 }
1892 
1893 /**
1894  * register_sysctl_table_path - register a sysctl table hierarchy
1895  * @path: The path to the directory the sysctl table is in.
1896  * @table: the top-level table structure
1897  *
1898  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1899  * array. A completely 0 filled entry terminates the table.
1900  *
1901  * See __register_sysctl_paths for more details.
1902  */
1903 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1904 						struct ctl_table *table)
1905 {
1906 	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1907 					path, table);
1908 }
1909 
1910 /**
1911  * register_sysctl_table - register a sysctl table hierarchy
1912  * @table: the top-level table structure
1913  *
1914  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1915  * array. A completely 0 filled entry terminates the table.
1916  *
1917  * See register_sysctl_paths for more details.
1918  */
1919 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1920 {
1921 	static const struct ctl_path null_path[] = { {} };
1922 
1923 	return register_sysctl_paths(null_path, table);
1924 }
1925 
1926 /**
1927  * unregister_sysctl_table - unregister a sysctl table hierarchy
1928  * @header: the header returned from register_sysctl_table
1929  *
1930  * Unregisters the sysctl table and all children. proc entries may not
1931  * actually be removed until they are no longer used by anyone.
1932  */
1933 void unregister_sysctl_table(struct ctl_table_header * header)
1934 {
1935 	might_sleep();
1936 
1937 	if (header == NULL)
1938 		return;
1939 
1940 	spin_lock(&sysctl_lock);
1941 	start_unregistering(header);
1942 	if (!--header->parent->count) {
1943 		WARN_ON(1);
1944 		call_rcu(&header->parent->rcu, free_head);
1945 	}
1946 	if (!--header->count)
1947 		call_rcu(&header->rcu, free_head);
1948 	spin_unlock(&sysctl_lock);
1949 }
1950 
1951 int sysctl_is_seen(struct ctl_table_header *p)
1952 {
1953 	struct ctl_table_set *set = p->set;
1954 	int res;
1955 	spin_lock(&sysctl_lock);
1956 	if (p->unregistering)
1957 		res = 0;
1958 	else if (!set->is_seen)
1959 		res = 1;
1960 	else
1961 		res = set->is_seen(set);
1962 	spin_unlock(&sysctl_lock);
1963 	return res;
1964 }
1965 
1966 void setup_sysctl_set(struct ctl_table_set *p,
1967 	struct ctl_table_set *parent,
1968 	int (*is_seen)(struct ctl_table_set *))
1969 {
1970 	INIT_LIST_HEAD(&p->list);
1971 	p->parent = parent ? parent : &sysctl_table_root.default_set;
1972 	p->is_seen = is_seen;
1973 }
1974 
1975 #else /* !CONFIG_SYSCTL */
1976 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1977 {
1978 	return NULL;
1979 }
1980 
1981 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1982 						    struct ctl_table *table)
1983 {
1984 	return NULL;
1985 }
1986 
1987 void unregister_sysctl_table(struct ctl_table_header * table)
1988 {
1989 }
1990 
1991 void setup_sysctl_set(struct ctl_table_set *p,
1992 	struct ctl_table_set *parent,
1993 	int (*is_seen)(struct ctl_table_set *))
1994 {
1995 }
1996 
1997 void sysctl_head_put(struct ctl_table_header *head)
1998 {
1999 }
2000 
2001 #endif /* CONFIG_SYSCTL */
2002 
2003 /*
2004  * /proc/sys support
2005  */
2006 
2007 #ifdef CONFIG_PROC_SYSCTL
2008 
2009 static int _proc_do_string(void* data, int maxlen, int write,
2010 			   void __user *buffer,
2011 			   size_t *lenp, loff_t *ppos)
2012 {
2013 	size_t len;
2014 	char __user *p;
2015 	char c;
2016 
2017 	if (!data || !maxlen || !*lenp) {
2018 		*lenp = 0;
2019 		return 0;
2020 	}
2021 
2022 	if (write) {
2023 		len = 0;
2024 		p = buffer;
2025 		while (len < *lenp) {
2026 			if (get_user(c, p++))
2027 				return -EFAULT;
2028 			if (c == 0 || c == '\n')
2029 				break;
2030 			len++;
2031 		}
2032 		if (len >= maxlen)
2033 			len = maxlen-1;
2034 		if(copy_from_user(data, buffer, len))
2035 			return -EFAULT;
2036 		((char *) data)[len] = 0;
2037 		*ppos += *lenp;
2038 	} else {
2039 		len = strlen(data);
2040 		if (len > maxlen)
2041 			len = maxlen;
2042 
2043 		if (*ppos > len) {
2044 			*lenp = 0;
2045 			return 0;
2046 		}
2047 
2048 		data += *ppos;
2049 		len  -= *ppos;
2050 
2051 		if (len > *lenp)
2052 			len = *lenp;
2053 		if (len)
2054 			if(copy_to_user(buffer, data, len))
2055 				return -EFAULT;
2056 		if (len < *lenp) {
2057 			if(put_user('\n', ((char __user *) buffer) + len))
2058 				return -EFAULT;
2059 			len++;
2060 		}
2061 		*lenp = len;
2062 		*ppos += len;
2063 	}
2064 	return 0;
2065 }
2066 
2067 /**
2068  * proc_dostring - read a string sysctl
2069  * @table: the sysctl table
2070  * @write: %TRUE if this is a write to the sysctl file
2071  * @buffer: the user buffer
2072  * @lenp: the size of the user buffer
2073  * @ppos: file position
2074  *
2075  * Reads/writes a string from/to the user buffer. If the kernel
2076  * buffer provided is not large enough to hold the string, the
2077  * string is truncated. The copied string is %NULL-terminated.
2078  * If the string is being read by the user process, it is copied
2079  * and a newline '\n' is added. It is truncated if the buffer is
2080  * not large enough.
2081  *
2082  * Returns 0 on success.
2083  */
2084 int proc_dostring(struct ctl_table *table, int write,
2085 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2086 {
2087 	return _proc_do_string(table->data, table->maxlen, write,
2088 			       buffer, lenp, ppos);
2089 }
2090 
2091 static size_t proc_skip_spaces(char **buf)
2092 {
2093 	size_t ret;
2094 	char *tmp = skip_spaces(*buf);
2095 	ret = tmp - *buf;
2096 	*buf = tmp;
2097 	return ret;
2098 }
2099 
2100 static void proc_skip_char(char **buf, size_t *size, const char v)
2101 {
2102 	while (*size) {
2103 		if (**buf != v)
2104 			break;
2105 		(*size)--;
2106 		(*buf)++;
2107 	}
2108 }
2109 
2110 #define TMPBUFLEN 22
2111 /**
2112  * proc_get_long - reads an ASCII formatted integer from a user buffer
2113  *
2114  * @buf: a kernel buffer
2115  * @size: size of the kernel buffer
2116  * @val: this is where the number will be stored
2117  * @neg: set to %TRUE if number is negative
2118  * @perm_tr: a vector which contains the allowed trailers
2119  * @perm_tr_len: size of the perm_tr vector
2120  * @tr: pointer to store the trailer character
2121  *
2122  * In case of success %0 is returned and @buf and @size are updated with
2123  * the amount of bytes read. If @tr is non-NULL and a trailing
2124  * character exists (size is non-zero after returning from this
2125  * function), @tr is updated with the trailing character.
2126  */
2127 static int proc_get_long(char **buf, size_t *size,
2128 			  unsigned long *val, bool *neg,
2129 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
2130 {
2131 	int len;
2132 	char *p, tmp[TMPBUFLEN];
2133 
2134 	if (!*size)
2135 		return -EINVAL;
2136 
2137 	len = *size;
2138 	if (len > TMPBUFLEN - 1)
2139 		len = TMPBUFLEN - 1;
2140 
2141 	memcpy(tmp, *buf, len);
2142 
2143 	tmp[len] = 0;
2144 	p = tmp;
2145 	if (*p == '-' && *size > 1) {
2146 		*neg = true;
2147 		p++;
2148 	} else
2149 		*neg = false;
2150 	if (!isdigit(*p))
2151 		return -EINVAL;
2152 
2153 	*val = simple_strtoul(p, &p, 0);
2154 
2155 	len = p - tmp;
2156 
2157 	/* We don't know if the next char is whitespace thus we may accept
2158 	 * invalid integers (e.g. 1234...a) or two integers instead of one
2159 	 * (e.g. 123...1). So lets not allow such large numbers. */
2160 	if (len == TMPBUFLEN - 1)
2161 		return -EINVAL;
2162 
2163 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2164 		return -EINVAL;
2165 
2166 	if (tr && (len < *size))
2167 		*tr = *p;
2168 
2169 	*buf += len;
2170 	*size -= len;
2171 
2172 	return 0;
2173 }
2174 
2175 /**
2176  * proc_put_long - converts an integer to a decimal ASCII formatted string
2177  *
2178  * @buf: the user buffer
2179  * @size: the size of the user buffer
2180  * @val: the integer to be converted
2181  * @neg: sign of the number, %TRUE for negative
2182  *
2183  * In case of success %0 is returned and @buf and @size are updated with
2184  * the amount of bytes written.
2185  */
2186 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2187 			  bool neg)
2188 {
2189 	int len;
2190 	char tmp[TMPBUFLEN], *p = tmp;
2191 
2192 	sprintf(p, "%s%lu", neg ? "-" : "", val);
2193 	len = strlen(tmp);
2194 	if (len > *size)
2195 		len = *size;
2196 	if (copy_to_user(*buf, tmp, len))
2197 		return -EFAULT;
2198 	*size -= len;
2199 	*buf += len;
2200 	return 0;
2201 }
2202 #undef TMPBUFLEN
2203 
2204 static int proc_put_char(void __user **buf, size_t *size, char c)
2205 {
2206 	if (*size) {
2207 		char __user **buffer = (char __user **)buf;
2208 		if (put_user(c, *buffer))
2209 			return -EFAULT;
2210 		(*size)--, (*buffer)++;
2211 		*buf = *buffer;
2212 	}
2213 	return 0;
2214 }
2215 
2216 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2217 				 int *valp,
2218 				 int write, void *data)
2219 {
2220 	if (write) {
2221 		*valp = *negp ? -*lvalp : *lvalp;
2222 	} else {
2223 		int val = *valp;
2224 		if (val < 0) {
2225 			*negp = true;
2226 			*lvalp = (unsigned long)-val;
2227 		} else {
2228 			*negp = false;
2229 			*lvalp = (unsigned long)val;
2230 		}
2231 	}
2232 	return 0;
2233 }
2234 
2235 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2236 
2237 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2238 		  int write, void __user *buffer,
2239 		  size_t *lenp, loff_t *ppos,
2240 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2241 			      int write, void *data),
2242 		  void *data)
2243 {
2244 	int *i, vleft, first = 1, err = 0;
2245 	unsigned long page = 0;
2246 	size_t left;
2247 	char *kbuf;
2248 
2249 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2250 		*lenp = 0;
2251 		return 0;
2252 	}
2253 
2254 	i = (int *) tbl_data;
2255 	vleft = table->maxlen / sizeof(*i);
2256 	left = *lenp;
2257 
2258 	if (!conv)
2259 		conv = do_proc_dointvec_conv;
2260 
2261 	if (write) {
2262 		if (left > PAGE_SIZE - 1)
2263 			left = PAGE_SIZE - 1;
2264 		page = __get_free_page(GFP_TEMPORARY);
2265 		kbuf = (char *) page;
2266 		if (!kbuf)
2267 			return -ENOMEM;
2268 		if (copy_from_user(kbuf, buffer, left)) {
2269 			err = -EFAULT;
2270 			goto free;
2271 		}
2272 		kbuf[left] = 0;
2273 	}
2274 
2275 	for (; left && vleft--; i++, first=0) {
2276 		unsigned long lval;
2277 		bool neg;
2278 
2279 		if (write) {
2280 			left -= proc_skip_spaces(&kbuf);
2281 
2282 			if (!left)
2283 				break;
2284 			err = proc_get_long(&kbuf, &left, &lval, &neg,
2285 					     proc_wspace_sep,
2286 					     sizeof(proc_wspace_sep), NULL);
2287 			if (err)
2288 				break;
2289 			if (conv(&neg, &lval, i, 1, data)) {
2290 				err = -EINVAL;
2291 				break;
2292 			}
2293 		} else {
2294 			if (conv(&neg, &lval, i, 0, data)) {
2295 				err = -EINVAL;
2296 				break;
2297 			}
2298 			if (!first)
2299 				err = proc_put_char(&buffer, &left, '\t');
2300 			if (err)
2301 				break;
2302 			err = proc_put_long(&buffer, &left, lval, neg);
2303 			if (err)
2304 				break;
2305 		}
2306 	}
2307 
2308 	if (!write && !first && left && !err)
2309 		err = proc_put_char(&buffer, &left, '\n');
2310 	if (write && !err && left)
2311 		left -= proc_skip_spaces(&kbuf);
2312 free:
2313 	if (write) {
2314 		free_page(page);
2315 		if (first)
2316 			return err ? : -EINVAL;
2317 	}
2318 	*lenp -= left;
2319 	*ppos += *lenp;
2320 	return err;
2321 }
2322 
2323 static int do_proc_dointvec(struct ctl_table *table, int write,
2324 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2325 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2326 			      int write, void *data),
2327 		  void *data)
2328 {
2329 	return __do_proc_dointvec(table->data, table, write,
2330 			buffer, lenp, ppos, conv, data);
2331 }
2332 
2333 /**
2334  * proc_dointvec - read a vector of integers
2335  * @table: the sysctl table
2336  * @write: %TRUE if this is a write to the sysctl file
2337  * @buffer: the user buffer
2338  * @lenp: the size of the user buffer
2339  * @ppos: file position
2340  *
2341  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2342  * values from/to the user buffer, treated as an ASCII string.
2343  *
2344  * Returns 0 on success.
2345  */
2346 int proc_dointvec(struct ctl_table *table, int write,
2347 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2348 {
2349     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2350 		    	    NULL,NULL);
2351 }
2352 
2353 /*
2354  * Taint values can only be increased
2355  * This means we can safely use a temporary.
2356  */
2357 static int proc_taint(struct ctl_table *table, int write,
2358 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2359 {
2360 	struct ctl_table t;
2361 	unsigned long tmptaint = get_taint();
2362 	int err;
2363 
2364 	if (write && !capable(CAP_SYS_ADMIN))
2365 		return -EPERM;
2366 
2367 	t = *table;
2368 	t.data = &tmptaint;
2369 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2370 	if (err < 0)
2371 		return err;
2372 
2373 	if (write) {
2374 		/*
2375 		 * Poor man's atomic or. Not worth adding a primitive
2376 		 * to everyone's atomic.h for this
2377 		 */
2378 		int i;
2379 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2380 			if ((tmptaint >> i) & 1)
2381 				add_taint(i);
2382 		}
2383 	}
2384 
2385 	return err;
2386 }
2387 
2388 struct do_proc_dointvec_minmax_conv_param {
2389 	int *min;
2390 	int *max;
2391 };
2392 
2393 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2394 					int *valp,
2395 					int write, void *data)
2396 {
2397 	struct do_proc_dointvec_minmax_conv_param *param = data;
2398 	if (write) {
2399 		int val = *negp ? -*lvalp : *lvalp;
2400 		if ((param->min && *param->min > val) ||
2401 		    (param->max && *param->max < val))
2402 			return -EINVAL;
2403 		*valp = val;
2404 	} else {
2405 		int val = *valp;
2406 		if (val < 0) {
2407 			*negp = true;
2408 			*lvalp = (unsigned long)-val;
2409 		} else {
2410 			*negp = false;
2411 			*lvalp = (unsigned long)val;
2412 		}
2413 	}
2414 	return 0;
2415 }
2416 
2417 /**
2418  * proc_dointvec_minmax - read a vector of integers with min/max values
2419  * @table: the sysctl table
2420  * @write: %TRUE if this is a write to the sysctl file
2421  * @buffer: the user buffer
2422  * @lenp: the size of the user buffer
2423  * @ppos: file position
2424  *
2425  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2426  * values from/to the user buffer, treated as an ASCII string.
2427  *
2428  * This routine will ensure the values are within the range specified by
2429  * table->extra1 (min) and table->extra2 (max).
2430  *
2431  * Returns 0 on success.
2432  */
2433 int proc_dointvec_minmax(struct ctl_table *table, int write,
2434 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2435 {
2436 	struct do_proc_dointvec_minmax_conv_param param = {
2437 		.min = (int *) table->extra1,
2438 		.max = (int *) table->extra2,
2439 	};
2440 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2441 				do_proc_dointvec_minmax_conv, &param);
2442 }
2443 
2444 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2445 				     void __user *buffer,
2446 				     size_t *lenp, loff_t *ppos,
2447 				     unsigned long convmul,
2448 				     unsigned long convdiv)
2449 {
2450 	unsigned long *i, *min, *max;
2451 	int vleft, first = 1, err = 0;
2452 	unsigned long page = 0;
2453 	size_t left;
2454 	char *kbuf;
2455 
2456 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2457 		*lenp = 0;
2458 		return 0;
2459 	}
2460 
2461 	i = (unsigned long *) data;
2462 	min = (unsigned long *) table->extra1;
2463 	max = (unsigned long *) table->extra2;
2464 	vleft = table->maxlen / sizeof(unsigned long);
2465 	left = *lenp;
2466 
2467 	if (write) {
2468 		if (left > PAGE_SIZE - 1)
2469 			left = PAGE_SIZE - 1;
2470 		page = __get_free_page(GFP_TEMPORARY);
2471 		kbuf = (char *) page;
2472 		if (!kbuf)
2473 			return -ENOMEM;
2474 		if (copy_from_user(kbuf, buffer, left)) {
2475 			err = -EFAULT;
2476 			goto free;
2477 		}
2478 		kbuf[left] = 0;
2479 	}
2480 
2481 	for (; left && vleft--; i++, first = 0) {
2482 		unsigned long val;
2483 
2484 		if (write) {
2485 			bool neg;
2486 
2487 			left -= proc_skip_spaces(&kbuf);
2488 
2489 			err = proc_get_long(&kbuf, &left, &val, &neg,
2490 					     proc_wspace_sep,
2491 					     sizeof(proc_wspace_sep), NULL);
2492 			if (err)
2493 				break;
2494 			if (neg)
2495 				continue;
2496 			if ((min && val < *min) || (max && val > *max))
2497 				continue;
2498 			*i = val;
2499 		} else {
2500 			val = convdiv * (*i) / convmul;
2501 			if (!first)
2502 				err = proc_put_char(&buffer, &left, '\t');
2503 			err = proc_put_long(&buffer, &left, val, false);
2504 			if (err)
2505 				break;
2506 		}
2507 	}
2508 
2509 	if (!write && !first && left && !err)
2510 		err = proc_put_char(&buffer, &left, '\n');
2511 	if (write && !err)
2512 		left -= proc_skip_spaces(&kbuf);
2513 free:
2514 	if (write) {
2515 		free_page(page);
2516 		if (first)
2517 			return err ? : -EINVAL;
2518 	}
2519 	*lenp -= left;
2520 	*ppos += *lenp;
2521 	return err;
2522 }
2523 
2524 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2525 				     void __user *buffer,
2526 				     size_t *lenp, loff_t *ppos,
2527 				     unsigned long convmul,
2528 				     unsigned long convdiv)
2529 {
2530 	return __do_proc_doulongvec_minmax(table->data, table, write,
2531 			buffer, lenp, ppos, convmul, convdiv);
2532 }
2533 
2534 /**
2535  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2536  * @table: the sysctl table
2537  * @write: %TRUE if this is a write to the sysctl file
2538  * @buffer: the user buffer
2539  * @lenp: the size of the user buffer
2540  * @ppos: file position
2541  *
2542  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2543  * values from/to the user buffer, treated as an ASCII string.
2544  *
2545  * This routine will ensure the values are within the range specified by
2546  * table->extra1 (min) and table->extra2 (max).
2547  *
2548  * Returns 0 on success.
2549  */
2550 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2551 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2552 {
2553     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2554 }
2555 
2556 /**
2557  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2558  * @table: the sysctl table
2559  * @write: %TRUE if this is a write to the sysctl file
2560  * @buffer: the user buffer
2561  * @lenp: the size of the user buffer
2562  * @ppos: file position
2563  *
2564  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2565  * values from/to the user buffer, treated as an ASCII string. The values
2566  * are treated as milliseconds, and converted to jiffies when they are stored.
2567  *
2568  * This routine will ensure the values are within the range specified by
2569  * table->extra1 (min) and table->extra2 (max).
2570  *
2571  * Returns 0 on success.
2572  */
2573 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2574 				      void __user *buffer,
2575 				      size_t *lenp, loff_t *ppos)
2576 {
2577     return do_proc_doulongvec_minmax(table, write, buffer,
2578 				     lenp, ppos, HZ, 1000l);
2579 }
2580 
2581 
2582 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2583 					 int *valp,
2584 					 int write, void *data)
2585 {
2586 	if (write) {
2587 		if (*lvalp > LONG_MAX / HZ)
2588 			return 1;
2589 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2590 	} else {
2591 		int val = *valp;
2592 		unsigned long lval;
2593 		if (val < 0) {
2594 			*negp = true;
2595 			lval = (unsigned long)-val;
2596 		} else {
2597 			*negp = false;
2598 			lval = (unsigned long)val;
2599 		}
2600 		*lvalp = lval / HZ;
2601 	}
2602 	return 0;
2603 }
2604 
2605 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2606 						int *valp,
2607 						int write, void *data)
2608 {
2609 	if (write) {
2610 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2611 			return 1;
2612 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2613 	} else {
2614 		int val = *valp;
2615 		unsigned long lval;
2616 		if (val < 0) {
2617 			*negp = true;
2618 			lval = (unsigned long)-val;
2619 		} else {
2620 			*negp = false;
2621 			lval = (unsigned long)val;
2622 		}
2623 		*lvalp = jiffies_to_clock_t(lval);
2624 	}
2625 	return 0;
2626 }
2627 
2628 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2629 					    int *valp,
2630 					    int write, void *data)
2631 {
2632 	if (write) {
2633 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2634 	} else {
2635 		int val = *valp;
2636 		unsigned long lval;
2637 		if (val < 0) {
2638 			*negp = true;
2639 			lval = (unsigned long)-val;
2640 		} else {
2641 			*negp = false;
2642 			lval = (unsigned long)val;
2643 		}
2644 		*lvalp = jiffies_to_msecs(lval);
2645 	}
2646 	return 0;
2647 }
2648 
2649 /**
2650  * proc_dointvec_jiffies - read a vector of integers as seconds
2651  * @table: the sysctl table
2652  * @write: %TRUE if this is a write to the sysctl file
2653  * @buffer: the user buffer
2654  * @lenp: the size of the user buffer
2655  * @ppos: file position
2656  *
2657  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2658  * values from/to the user buffer, treated as an ASCII string.
2659  * The values read are assumed to be in seconds, and are converted into
2660  * jiffies.
2661  *
2662  * Returns 0 on success.
2663  */
2664 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2665 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2666 {
2667     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2668 		    	    do_proc_dointvec_jiffies_conv,NULL);
2669 }
2670 
2671 /**
2672  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2673  * @table: the sysctl table
2674  * @write: %TRUE if this is a write to the sysctl file
2675  * @buffer: the user buffer
2676  * @lenp: the size of the user buffer
2677  * @ppos: pointer to the file position
2678  *
2679  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2680  * values from/to the user buffer, treated as an ASCII string.
2681  * The values read are assumed to be in 1/USER_HZ seconds, and
2682  * are converted into jiffies.
2683  *
2684  * Returns 0 on success.
2685  */
2686 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2687 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2688 {
2689     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2690 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2691 }
2692 
2693 /**
2694  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2695  * @table: the sysctl table
2696  * @write: %TRUE if this is a write to the sysctl file
2697  * @buffer: the user buffer
2698  * @lenp: the size of the user buffer
2699  * @ppos: file position
2700  * @ppos: the current position in the file
2701  *
2702  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2703  * values from/to the user buffer, treated as an ASCII string.
2704  * The values read are assumed to be in 1/1000 seconds, and
2705  * are converted into jiffies.
2706  *
2707  * Returns 0 on success.
2708  */
2709 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2710 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2711 {
2712 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2713 				do_proc_dointvec_ms_jiffies_conv, NULL);
2714 }
2715 
2716 static int proc_do_cad_pid(struct ctl_table *table, int write,
2717 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2718 {
2719 	struct pid *new_pid;
2720 	pid_t tmp;
2721 	int r;
2722 
2723 	tmp = pid_vnr(cad_pid);
2724 
2725 	r = __do_proc_dointvec(&tmp, table, write, buffer,
2726 			       lenp, ppos, NULL, NULL);
2727 	if (r || !write)
2728 		return r;
2729 
2730 	new_pid = find_get_pid(tmp);
2731 	if (!new_pid)
2732 		return -ESRCH;
2733 
2734 	put_pid(xchg(&cad_pid, new_pid));
2735 	return 0;
2736 }
2737 
2738 /**
2739  * proc_do_large_bitmap - read/write from/to a large bitmap
2740  * @table: the sysctl table
2741  * @write: %TRUE if this is a write to the sysctl file
2742  * @buffer: the user buffer
2743  * @lenp: the size of the user buffer
2744  * @ppos: file position
2745  *
2746  * The bitmap is stored at table->data and the bitmap length (in bits)
2747  * in table->maxlen.
2748  *
2749  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2750  * large bitmaps may be represented in a compact manner. Writing into
2751  * the file will clear the bitmap then update it with the given input.
2752  *
2753  * Returns 0 on success.
2754  */
2755 int proc_do_large_bitmap(struct ctl_table *table, int write,
2756 			 void __user *buffer, size_t *lenp, loff_t *ppos)
2757 {
2758 	int err = 0;
2759 	bool first = 1;
2760 	size_t left = *lenp;
2761 	unsigned long bitmap_len = table->maxlen;
2762 	unsigned long *bitmap = (unsigned long *) table->data;
2763 	unsigned long *tmp_bitmap = NULL;
2764 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2765 
2766 	if (!bitmap_len || !left || (*ppos && !write)) {
2767 		*lenp = 0;
2768 		return 0;
2769 	}
2770 
2771 	if (write) {
2772 		unsigned long page = 0;
2773 		char *kbuf;
2774 
2775 		if (left > PAGE_SIZE - 1)
2776 			left = PAGE_SIZE - 1;
2777 
2778 		page = __get_free_page(GFP_TEMPORARY);
2779 		kbuf = (char *) page;
2780 		if (!kbuf)
2781 			return -ENOMEM;
2782 		if (copy_from_user(kbuf, buffer, left)) {
2783 			free_page(page);
2784 			return -EFAULT;
2785                 }
2786 		kbuf[left] = 0;
2787 
2788 		tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2789 				     GFP_KERNEL);
2790 		if (!tmp_bitmap) {
2791 			free_page(page);
2792 			return -ENOMEM;
2793 		}
2794 		proc_skip_char(&kbuf, &left, '\n');
2795 		while (!err && left) {
2796 			unsigned long val_a, val_b;
2797 			bool neg;
2798 
2799 			err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2800 					     sizeof(tr_a), &c);
2801 			if (err)
2802 				break;
2803 			if (val_a >= bitmap_len || neg) {
2804 				err = -EINVAL;
2805 				break;
2806 			}
2807 
2808 			val_b = val_a;
2809 			if (left) {
2810 				kbuf++;
2811 				left--;
2812 			}
2813 
2814 			if (c == '-') {
2815 				err = proc_get_long(&kbuf, &left, &val_b,
2816 						     &neg, tr_b, sizeof(tr_b),
2817 						     &c);
2818 				if (err)
2819 					break;
2820 				if (val_b >= bitmap_len || neg ||
2821 				    val_a > val_b) {
2822 					err = -EINVAL;
2823 					break;
2824 				}
2825 				if (left) {
2826 					kbuf++;
2827 					left--;
2828 				}
2829 			}
2830 
2831 			while (val_a <= val_b)
2832 				set_bit(val_a++, tmp_bitmap);
2833 
2834 			first = 0;
2835 			proc_skip_char(&kbuf, &left, '\n');
2836 		}
2837 		free_page(page);
2838 	} else {
2839 		unsigned long bit_a, bit_b = 0;
2840 
2841 		while (left) {
2842 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2843 			if (bit_a >= bitmap_len)
2844 				break;
2845 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
2846 						   bit_a + 1) - 1;
2847 
2848 			if (!first) {
2849 				err = proc_put_char(&buffer, &left, ',');
2850 				if (err)
2851 					break;
2852 			}
2853 			err = proc_put_long(&buffer, &left, bit_a, false);
2854 			if (err)
2855 				break;
2856 			if (bit_a != bit_b) {
2857 				err = proc_put_char(&buffer, &left, '-');
2858 				if (err)
2859 					break;
2860 				err = proc_put_long(&buffer, &left, bit_b, false);
2861 				if (err)
2862 					break;
2863 			}
2864 
2865 			first = 0; bit_b++;
2866 		}
2867 		if (!err)
2868 			err = proc_put_char(&buffer, &left, '\n');
2869 	}
2870 
2871 	if (!err) {
2872 		if (write) {
2873 			if (*ppos)
2874 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2875 			else
2876 				memcpy(bitmap, tmp_bitmap,
2877 					BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long));
2878 		}
2879 		kfree(tmp_bitmap);
2880 		*lenp -= left;
2881 		*ppos += *lenp;
2882 		return 0;
2883 	} else {
2884 		kfree(tmp_bitmap);
2885 		return err;
2886 	}
2887 }
2888 
2889 #else /* CONFIG_PROC_SYSCTL */
2890 
2891 int proc_dostring(struct ctl_table *table, int write,
2892 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2893 {
2894 	return -ENOSYS;
2895 }
2896 
2897 int proc_dointvec(struct ctl_table *table, int write,
2898 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2899 {
2900 	return -ENOSYS;
2901 }
2902 
2903 int proc_dointvec_minmax(struct ctl_table *table, int write,
2904 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2905 {
2906 	return -ENOSYS;
2907 }
2908 
2909 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2910 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2911 {
2912 	return -ENOSYS;
2913 }
2914 
2915 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2916 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2917 {
2918 	return -ENOSYS;
2919 }
2920 
2921 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2922 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2923 {
2924 	return -ENOSYS;
2925 }
2926 
2927 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2928 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2929 {
2930 	return -ENOSYS;
2931 }
2932 
2933 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2934 				      void __user *buffer,
2935 				      size_t *lenp, loff_t *ppos)
2936 {
2937     return -ENOSYS;
2938 }
2939 
2940 
2941 #endif /* CONFIG_PROC_SYSCTL */
2942 
2943 /*
2944  * No sense putting this after each symbol definition, twice,
2945  * exception granted :-)
2946  */
2947 EXPORT_SYMBOL(proc_dointvec);
2948 EXPORT_SYMBOL(proc_dointvec_jiffies);
2949 EXPORT_SYMBOL(proc_dointvec_minmax);
2950 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2951 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2952 EXPORT_SYMBOL(proc_dostring);
2953 EXPORT_SYMBOL(proc_doulongvec_minmax);
2954 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2955 EXPORT_SYMBOL(register_sysctl_table);
2956 EXPORT_SYMBOL(register_sysctl_paths);
2957 EXPORT_SYMBOL(unregister_sysctl_table);
2958