xref: /linux-6.15/kernel/sysctl.c (revision 22fd411a)
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 static int __sysrq_enabled; /* Note: sysrq code ises it's own private copy */
174 
175 static int sysrq_sysctl_handler(ctl_table *table, int write,
176 				void __user *buffer, size_t *lenp,
177 				loff_t *ppos)
178 {
179 	int error;
180 
181 	error = proc_dointvec(table, write, buffer, lenp, ppos);
182 	if (error)
183 		return error;
184 
185 	if (write)
186 		sysrq_toggle_support(__sysrq_enabled);
187 
188 	return 0;
189 }
190 
191 #endif
192 
193 static struct ctl_table root_table[];
194 static struct ctl_table_root sysctl_table_root;
195 static struct ctl_table_header root_table_header = {
196 	.count = 1,
197 	.ctl_table = root_table,
198 	.ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
199 	.root = &sysctl_table_root,
200 	.set = &sysctl_table_root.default_set,
201 };
202 static struct ctl_table_root sysctl_table_root = {
203 	.root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
204 	.default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
205 };
206 
207 static struct ctl_table kern_table[];
208 static struct ctl_table vm_table[];
209 static struct ctl_table fs_table[];
210 static struct ctl_table debug_table[];
211 static struct ctl_table dev_table[];
212 extern struct ctl_table random_table[];
213 #ifdef CONFIG_EPOLL
214 extern struct ctl_table epoll_table[];
215 #endif
216 
217 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
218 int sysctl_legacy_va_layout;
219 #endif
220 
221 /* The default sysctl tables: */
222 
223 static struct ctl_table root_table[] = {
224 	{
225 		.procname	= "kernel",
226 		.mode		= 0555,
227 		.child		= kern_table,
228 	},
229 	{
230 		.procname	= "vm",
231 		.mode		= 0555,
232 		.child		= vm_table,
233 	},
234 	{
235 		.procname	= "fs",
236 		.mode		= 0555,
237 		.child		= fs_table,
238 	},
239 	{
240 		.procname	= "debug",
241 		.mode		= 0555,
242 		.child		= debug_table,
243 	},
244 	{
245 		.procname	= "dev",
246 		.mode		= 0555,
247 		.child		= dev_table,
248 	},
249 	{ }
250 };
251 
252 #ifdef CONFIG_SCHED_DEBUG
253 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
254 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
255 static int min_wakeup_granularity_ns;			/* 0 usecs */
256 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
257 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
258 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
259 #endif
260 
261 #ifdef CONFIG_COMPACTION
262 static int min_extfrag_threshold;
263 static int max_extfrag_threshold = 1000;
264 #endif
265 
266 static struct ctl_table kern_table[] = {
267 	{
268 		.procname	= "sched_child_runs_first",
269 		.data		= &sysctl_sched_child_runs_first,
270 		.maxlen		= sizeof(unsigned int),
271 		.mode		= 0644,
272 		.proc_handler	= proc_dointvec,
273 	},
274 #ifdef CONFIG_SCHED_DEBUG
275 	{
276 		.procname	= "sched_min_granularity_ns",
277 		.data		= &sysctl_sched_min_granularity,
278 		.maxlen		= sizeof(unsigned int),
279 		.mode		= 0644,
280 		.proc_handler	= sched_proc_update_handler,
281 		.extra1		= &min_sched_granularity_ns,
282 		.extra2		= &max_sched_granularity_ns,
283 	},
284 	{
285 		.procname	= "sched_latency_ns",
286 		.data		= &sysctl_sched_latency,
287 		.maxlen		= sizeof(unsigned int),
288 		.mode		= 0644,
289 		.proc_handler	= sched_proc_update_handler,
290 		.extra1		= &min_sched_granularity_ns,
291 		.extra2		= &max_sched_granularity_ns,
292 	},
293 	{
294 		.procname	= "sched_wakeup_granularity_ns",
295 		.data		= &sysctl_sched_wakeup_granularity,
296 		.maxlen		= sizeof(unsigned int),
297 		.mode		= 0644,
298 		.proc_handler	= sched_proc_update_handler,
299 		.extra1		= &min_wakeup_granularity_ns,
300 		.extra2		= &max_wakeup_granularity_ns,
301 	},
302 	{
303 		.procname	= "sched_tunable_scaling",
304 		.data		= &sysctl_sched_tunable_scaling,
305 		.maxlen		= sizeof(enum sched_tunable_scaling),
306 		.mode		= 0644,
307 		.proc_handler	= sched_proc_update_handler,
308 		.extra1		= &min_sched_tunable_scaling,
309 		.extra2		= &max_sched_tunable_scaling,
310 	},
311 	{
312 		.procname	= "sched_migration_cost",
313 		.data		= &sysctl_sched_migration_cost,
314 		.maxlen		= sizeof(unsigned int),
315 		.mode		= 0644,
316 		.proc_handler	= proc_dointvec,
317 	},
318 	{
319 		.procname	= "sched_nr_migrate",
320 		.data		= &sysctl_sched_nr_migrate,
321 		.maxlen		= sizeof(unsigned int),
322 		.mode		= 0644,
323 		.proc_handler	= proc_dointvec,
324 	},
325 	{
326 		.procname	= "sched_time_avg",
327 		.data		= &sysctl_sched_time_avg,
328 		.maxlen		= sizeof(unsigned int),
329 		.mode		= 0644,
330 		.proc_handler	= proc_dointvec,
331 	},
332 	{
333 		.procname	= "sched_shares_window",
334 		.data		= &sysctl_sched_shares_window,
335 		.maxlen		= sizeof(unsigned int),
336 		.mode		= 0644,
337 		.proc_handler	= proc_dointvec,
338 	},
339 	{
340 		.procname	= "timer_migration",
341 		.data		= &sysctl_timer_migration,
342 		.maxlen		= sizeof(unsigned int),
343 		.mode		= 0644,
344 		.proc_handler	= proc_dointvec_minmax,
345 		.extra1		= &zero,
346 		.extra2		= &one,
347 	},
348 #endif
349 	{
350 		.procname	= "sched_rt_period_us",
351 		.data		= &sysctl_sched_rt_period,
352 		.maxlen		= sizeof(unsigned int),
353 		.mode		= 0644,
354 		.proc_handler	= sched_rt_handler,
355 	},
356 	{
357 		.procname	= "sched_rt_runtime_us",
358 		.data		= &sysctl_sched_rt_runtime,
359 		.maxlen		= sizeof(int),
360 		.mode		= 0644,
361 		.proc_handler	= sched_rt_handler,
362 	},
363 	{
364 		.procname	= "sched_compat_yield",
365 		.data		= &sysctl_sched_compat_yield,
366 		.maxlen		= sizeof(unsigned int),
367 		.mode		= 0644,
368 		.proc_handler	= proc_dointvec,
369 	},
370 #ifdef CONFIG_SCHED_AUTOGROUP
371 	{
372 		.procname	= "sched_autogroup_enabled",
373 		.data		= &sysctl_sched_autogroup_enabled,
374 		.maxlen		= sizeof(unsigned int),
375 		.mode		= 0644,
376 		.proc_handler	= proc_dointvec,
377 		.extra1		= &zero,
378 		.extra2		= &one,
379 	},
380 #endif
381 #ifdef CONFIG_PROVE_LOCKING
382 	{
383 		.procname	= "prove_locking",
384 		.data		= &prove_locking,
385 		.maxlen		= sizeof(int),
386 		.mode		= 0644,
387 		.proc_handler	= proc_dointvec,
388 	},
389 #endif
390 #ifdef CONFIG_LOCK_STAT
391 	{
392 		.procname	= "lock_stat",
393 		.data		= &lock_stat,
394 		.maxlen		= sizeof(int),
395 		.mode		= 0644,
396 		.proc_handler	= proc_dointvec,
397 	},
398 #endif
399 	{
400 		.procname	= "panic",
401 		.data		= &panic_timeout,
402 		.maxlen		= sizeof(int),
403 		.mode		= 0644,
404 		.proc_handler	= proc_dointvec,
405 	},
406 	{
407 		.procname	= "core_uses_pid",
408 		.data		= &core_uses_pid,
409 		.maxlen		= sizeof(int),
410 		.mode		= 0644,
411 		.proc_handler	= proc_dointvec,
412 	},
413 	{
414 		.procname	= "core_pattern",
415 		.data		= core_pattern,
416 		.maxlen		= CORENAME_MAX_SIZE,
417 		.mode		= 0644,
418 		.proc_handler	= proc_dostring,
419 	},
420 	{
421 		.procname	= "core_pipe_limit",
422 		.data		= &core_pipe_limit,
423 		.maxlen		= sizeof(unsigned int),
424 		.mode		= 0644,
425 		.proc_handler	= proc_dointvec,
426 	},
427 #ifdef CONFIG_PROC_SYSCTL
428 	{
429 		.procname	= "tainted",
430 		.maxlen 	= sizeof(long),
431 		.mode		= 0644,
432 		.proc_handler	= proc_taint,
433 	},
434 #endif
435 #ifdef CONFIG_LATENCYTOP
436 	{
437 		.procname	= "latencytop",
438 		.data		= &latencytop_enabled,
439 		.maxlen		= sizeof(int),
440 		.mode		= 0644,
441 		.proc_handler	= proc_dointvec,
442 	},
443 #endif
444 #ifdef CONFIG_BLK_DEV_INITRD
445 	{
446 		.procname	= "real-root-dev",
447 		.data		= &real_root_dev,
448 		.maxlen		= sizeof(int),
449 		.mode		= 0644,
450 		.proc_handler	= proc_dointvec,
451 	},
452 #endif
453 	{
454 		.procname	= "print-fatal-signals",
455 		.data		= &print_fatal_signals,
456 		.maxlen		= sizeof(int),
457 		.mode		= 0644,
458 		.proc_handler	= proc_dointvec,
459 	},
460 #ifdef CONFIG_SPARC
461 	{
462 		.procname	= "reboot-cmd",
463 		.data		= reboot_command,
464 		.maxlen		= 256,
465 		.mode		= 0644,
466 		.proc_handler	= proc_dostring,
467 	},
468 	{
469 		.procname	= "stop-a",
470 		.data		= &stop_a_enabled,
471 		.maxlen		= sizeof (int),
472 		.mode		= 0644,
473 		.proc_handler	= proc_dointvec,
474 	},
475 	{
476 		.procname	= "scons-poweroff",
477 		.data		= &scons_pwroff,
478 		.maxlen		= sizeof (int),
479 		.mode		= 0644,
480 		.proc_handler	= proc_dointvec,
481 	},
482 #endif
483 #ifdef CONFIG_SPARC64
484 	{
485 		.procname	= "tsb-ratio",
486 		.data		= &sysctl_tsb_ratio,
487 		.maxlen		= sizeof (int),
488 		.mode		= 0644,
489 		.proc_handler	= proc_dointvec,
490 	},
491 #endif
492 #ifdef __hppa__
493 	{
494 		.procname	= "soft-power",
495 		.data		= &pwrsw_enabled,
496 		.maxlen		= sizeof (int),
497 	 	.mode		= 0644,
498 		.proc_handler	= proc_dointvec,
499 	},
500 	{
501 		.procname	= "unaligned-trap",
502 		.data		= &unaligned_enabled,
503 		.maxlen		= sizeof (int),
504 		.mode		= 0644,
505 		.proc_handler	= proc_dointvec,
506 	},
507 #endif
508 	{
509 		.procname	= "ctrl-alt-del",
510 		.data		= &C_A_D,
511 		.maxlen		= sizeof(int),
512 		.mode		= 0644,
513 		.proc_handler	= proc_dointvec,
514 	},
515 #ifdef CONFIG_FUNCTION_TRACER
516 	{
517 		.procname	= "ftrace_enabled",
518 		.data		= &ftrace_enabled,
519 		.maxlen		= sizeof(int),
520 		.mode		= 0644,
521 		.proc_handler	= ftrace_enable_sysctl,
522 	},
523 #endif
524 #ifdef CONFIG_STACK_TRACER
525 	{
526 		.procname	= "stack_tracer_enabled",
527 		.data		= &stack_tracer_enabled,
528 		.maxlen		= sizeof(int),
529 		.mode		= 0644,
530 		.proc_handler	= stack_trace_sysctl,
531 	},
532 #endif
533 #ifdef CONFIG_TRACING
534 	{
535 		.procname	= "ftrace_dump_on_oops",
536 		.data		= &ftrace_dump_on_oops,
537 		.maxlen		= sizeof(int),
538 		.mode		= 0644,
539 		.proc_handler	= proc_dointvec,
540 	},
541 #endif
542 #ifdef CONFIG_MODULES
543 	{
544 		.procname	= "modprobe",
545 		.data		= &modprobe_path,
546 		.maxlen		= KMOD_PATH_LEN,
547 		.mode		= 0644,
548 		.proc_handler	= proc_dostring,
549 	},
550 	{
551 		.procname	= "modules_disabled",
552 		.data		= &modules_disabled,
553 		.maxlen		= sizeof(int),
554 		.mode		= 0644,
555 		/* only handle a transition from default "0" to "1" */
556 		.proc_handler	= proc_dointvec_minmax,
557 		.extra1		= &one,
558 		.extra2		= &one,
559 	},
560 #endif
561 #ifdef CONFIG_HOTPLUG
562 	{
563 		.procname	= "hotplug",
564 		.data		= &uevent_helper,
565 		.maxlen		= UEVENT_HELPER_PATH_LEN,
566 		.mode		= 0644,
567 		.proc_handler	= proc_dostring,
568 	},
569 #endif
570 #ifdef CONFIG_CHR_DEV_SG
571 	{
572 		.procname	= "sg-big-buff",
573 		.data		= &sg_big_buff,
574 		.maxlen		= sizeof (int),
575 		.mode		= 0444,
576 		.proc_handler	= proc_dointvec,
577 	},
578 #endif
579 #ifdef CONFIG_BSD_PROCESS_ACCT
580 	{
581 		.procname	= "acct",
582 		.data		= &acct_parm,
583 		.maxlen		= 3*sizeof(int),
584 		.mode		= 0644,
585 		.proc_handler	= proc_dointvec,
586 	},
587 #endif
588 #ifdef CONFIG_MAGIC_SYSRQ
589 	{
590 		.procname	= "sysrq",
591 		.data		= &__sysrq_enabled,
592 		.maxlen		= sizeof (int),
593 		.mode		= 0644,
594 		.proc_handler	= sysrq_sysctl_handler,
595 	},
596 #endif
597 #ifdef CONFIG_PROC_SYSCTL
598 	{
599 		.procname	= "cad_pid",
600 		.data		= NULL,
601 		.maxlen		= sizeof (int),
602 		.mode		= 0600,
603 		.proc_handler	= proc_do_cad_pid,
604 	},
605 #endif
606 	{
607 		.procname	= "threads-max",
608 		.data		= &max_threads,
609 		.maxlen		= sizeof(int),
610 		.mode		= 0644,
611 		.proc_handler	= proc_dointvec,
612 	},
613 	{
614 		.procname	= "random",
615 		.mode		= 0555,
616 		.child		= random_table,
617 	},
618 	{
619 		.procname	= "overflowuid",
620 		.data		= &overflowuid,
621 		.maxlen		= sizeof(int),
622 		.mode		= 0644,
623 		.proc_handler	= proc_dointvec_minmax,
624 		.extra1		= &minolduid,
625 		.extra2		= &maxolduid,
626 	},
627 	{
628 		.procname	= "overflowgid",
629 		.data		= &overflowgid,
630 		.maxlen		= sizeof(int),
631 		.mode		= 0644,
632 		.proc_handler	= proc_dointvec_minmax,
633 		.extra1		= &minolduid,
634 		.extra2		= &maxolduid,
635 	},
636 #ifdef CONFIG_S390
637 #ifdef CONFIG_MATHEMU
638 	{
639 		.procname	= "ieee_emulation_warnings",
640 		.data		= &sysctl_ieee_emulation_warnings,
641 		.maxlen		= sizeof(int),
642 		.mode		= 0644,
643 		.proc_handler	= proc_dointvec,
644 	},
645 #endif
646 	{
647 		.procname	= "userprocess_debug",
648 		.data		= &show_unhandled_signals,
649 		.maxlen		= sizeof(int),
650 		.mode		= 0644,
651 		.proc_handler	= proc_dointvec,
652 	},
653 #endif
654 	{
655 		.procname	= "pid_max",
656 		.data		= &pid_max,
657 		.maxlen		= sizeof (int),
658 		.mode		= 0644,
659 		.proc_handler	= proc_dointvec_minmax,
660 		.extra1		= &pid_max_min,
661 		.extra2		= &pid_max_max,
662 	},
663 	{
664 		.procname	= "panic_on_oops",
665 		.data		= &panic_on_oops,
666 		.maxlen		= sizeof(int),
667 		.mode		= 0644,
668 		.proc_handler	= proc_dointvec,
669 	},
670 #if defined CONFIG_PRINTK
671 	{
672 		.procname	= "printk",
673 		.data		= &console_loglevel,
674 		.maxlen		= 4*sizeof(int),
675 		.mode		= 0644,
676 		.proc_handler	= proc_dointvec,
677 	},
678 	{
679 		.procname	= "printk_ratelimit",
680 		.data		= &printk_ratelimit_state.interval,
681 		.maxlen		= sizeof(int),
682 		.mode		= 0644,
683 		.proc_handler	= proc_dointvec_jiffies,
684 	},
685 	{
686 		.procname	= "printk_ratelimit_burst",
687 		.data		= &printk_ratelimit_state.burst,
688 		.maxlen		= sizeof(int),
689 		.mode		= 0644,
690 		.proc_handler	= proc_dointvec,
691 	},
692 	{
693 		.procname	= "printk_delay",
694 		.data		= &printk_delay_msec,
695 		.maxlen		= sizeof(int),
696 		.mode		= 0644,
697 		.proc_handler	= proc_dointvec_minmax,
698 		.extra1		= &zero,
699 		.extra2		= &ten_thousand,
700 	},
701 	{
702 		.procname	= "dmesg_restrict",
703 		.data		= &dmesg_restrict,
704 		.maxlen		= sizeof(int),
705 		.mode		= 0644,
706 		.proc_handler	= proc_dointvec_minmax,
707 		.extra1		= &zero,
708 		.extra2		= &one,
709 	},
710 	{
711 		.procname	= "kptr_restrict",
712 		.data		= &kptr_restrict,
713 		.maxlen		= sizeof(int),
714 		.mode		= 0644,
715 		.proc_handler	= proc_dointvec_minmax,
716 		.extra1		= &zero,
717 		.extra2		= &two,
718 	},
719 #endif
720 	{
721 		.procname	= "ngroups_max",
722 		.data		= &ngroups_max,
723 		.maxlen		= sizeof (int),
724 		.mode		= 0444,
725 		.proc_handler	= proc_dointvec,
726 	},
727 #if defined(CONFIG_LOCKUP_DETECTOR)
728 	{
729 		.procname       = "watchdog",
730 		.data           = &watchdog_enabled,
731 		.maxlen         = sizeof (int),
732 		.mode           = 0644,
733 		.proc_handler   = proc_dowatchdog_enabled,
734 	},
735 	{
736 		.procname	= "watchdog_thresh",
737 		.data		= &softlockup_thresh,
738 		.maxlen		= sizeof(int),
739 		.mode		= 0644,
740 		.proc_handler	= proc_dowatchdog_thresh,
741 		.extra1		= &neg_one,
742 		.extra2		= &sixty,
743 	},
744 	{
745 		.procname	= "softlockup_panic",
746 		.data		= &softlockup_panic,
747 		.maxlen		= sizeof(int),
748 		.mode		= 0644,
749 		.proc_handler	= proc_dointvec_minmax,
750 		.extra1		= &zero,
751 		.extra2		= &one,
752 	},
753 	{
754 		.procname       = "nmi_watchdog",
755 		.data           = &watchdog_enabled,
756 		.maxlen         = sizeof (int),
757 		.mode           = 0644,
758 		.proc_handler   = proc_dowatchdog_enabled,
759 	},
760 #endif
761 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
762 	{
763 		.procname       = "unknown_nmi_panic",
764 		.data           = &unknown_nmi_panic,
765 		.maxlen         = sizeof (int),
766 		.mode           = 0644,
767 		.proc_handler   = proc_dointvec,
768 	},
769 #endif
770 #if defined(CONFIG_X86)
771 	{
772 		.procname	= "panic_on_unrecovered_nmi",
773 		.data		= &panic_on_unrecovered_nmi,
774 		.maxlen		= sizeof(int),
775 		.mode		= 0644,
776 		.proc_handler	= proc_dointvec,
777 	},
778 	{
779 		.procname	= "panic_on_io_nmi",
780 		.data		= &panic_on_io_nmi,
781 		.maxlen		= sizeof(int),
782 		.mode		= 0644,
783 		.proc_handler	= proc_dointvec,
784 	},
785 	{
786 		.procname	= "bootloader_type",
787 		.data		= &bootloader_type,
788 		.maxlen		= sizeof (int),
789 		.mode		= 0444,
790 		.proc_handler	= proc_dointvec,
791 	},
792 	{
793 		.procname	= "bootloader_version",
794 		.data		= &bootloader_version,
795 		.maxlen		= sizeof (int),
796 		.mode		= 0444,
797 		.proc_handler	= proc_dointvec,
798 	},
799 	{
800 		.procname	= "kstack_depth_to_print",
801 		.data		= &kstack_depth_to_print,
802 		.maxlen		= sizeof(int),
803 		.mode		= 0644,
804 		.proc_handler	= proc_dointvec,
805 	},
806 	{
807 		.procname	= "io_delay_type",
808 		.data		= &io_delay_type,
809 		.maxlen		= sizeof(int),
810 		.mode		= 0644,
811 		.proc_handler	= proc_dointvec,
812 	},
813 #endif
814 #if defined(CONFIG_MMU)
815 	{
816 		.procname	= "randomize_va_space",
817 		.data		= &randomize_va_space,
818 		.maxlen		= sizeof(int),
819 		.mode		= 0644,
820 		.proc_handler	= proc_dointvec,
821 	},
822 #endif
823 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
824 	{
825 		.procname	= "spin_retry",
826 		.data		= &spin_retry,
827 		.maxlen		= sizeof (int),
828 		.mode		= 0644,
829 		.proc_handler	= proc_dointvec,
830 	},
831 #endif
832 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
833 	{
834 		.procname	= "acpi_video_flags",
835 		.data		= &acpi_realmode_flags,
836 		.maxlen		= sizeof (unsigned long),
837 		.mode		= 0644,
838 		.proc_handler	= proc_doulongvec_minmax,
839 	},
840 #endif
841 #ifdef CONFIG_IA64
842 	{
843 		.procname	= "ignore-unaligned-usertrap",
844 		.data		= &no_unaligned_warning,
845 		.maxlen		= sizeof (int),
846 	 	.mode		= 0644,
847 		.proc_handler	= proc_dointvec,
848 	},
849 	{
850 		.procname	= "unaligned-dump-stack",
851 		.data		= &unaligned_dump_stack,
852 		.maxlen		= sizeof (int),
853 		.mode		= 0644,
854 		.proc_handler	= proc_dointvec,
855 	},
856 #endif
857 #ifdef CONFIG_DETECT_HUNG_TASK
858 	{
859 		.procname	= "hung_task_panic",
860 		.data		= &sysctl_hung_task_panic,
861 		.maxlen		= sizeof(int),
862 		.mode		= 0644,
863 		.proc_handler	= proc_dointvec_minmax,
864 		.extra1		= &zero,
865 		.extra2		= &one,
866 	},
867 	{
868 		.procname	= "hung_task_check_count",
869 		.data		= &sysctl_hung_task_check_count,
870 		.maxlen		= sizeof(unsigned long),
871 		.mode		= 0644,
872 		.proc_handler	= proc_doulongvec_minmax,
873 	},
874 	{
875 		.procname	= "hung_task_timeout_secs",
876 		.data		= &sysctl_hung_task_timeout_secs,
877 		.maxlen		= sizeof(unsigned long),
878 		.mode		= 0644,
879 		.proc_handler	= proc_dohung_task_timeout_secs,
880 	},
881 	{
882 		.procname	= "hung_task_warnings",
883 		.data		= &sysctl_hung_task_warnings,
884 		.maxlen		= sizeof(unsigned long),
885 		.mode		= 0644,
886 		.proc_handler	= proc_doulongvec_minmax,
887 	},
888 #endif
889 #ifdef CONFIG_COMPAT
890 	{
891 		.procname	= "compat-log",
892 		.data		= &compat_log,
893 		.maxlen		= sizeof (int),
894 	 	.mode		= 0644,
895 		.proc_handler	= proc_dointvec,
896 	},
897 #endif
898 #ifdef CONFIG_RT_MUTEXES
899 	{
900 		.procname	= "max_lock_depth",
901 		.data		= &max_lock_depth,
902 		.maxlen		= sizeof(int),
903 		.mode		= 0644,
904 		.proc_handler	= proc_dointvec,
905 	},
906 #endif
907 	{
908 		.procname	= "poweroff_cmd",
909 		.data		= &poweroff_cmd,
910 		.maxlen		= POWEROFF_CMD_PATH_LEN,
911 		.mode		= 0644,
912 		.proc_handler	= proc_dostring,
913 	},
914 #ifdef CONFIG_KEYS
915 	{
916 		.procname	= "keys",
917 		.mode		= 0555,
918 		.child		= key_sysctls,
919 	},
920 #endif
921 #ifdef CONFIG_RCU_TORTURE_TEST
922 	{
923 		.procname       = "rcutorture_runnable",
924 		.data           = &rcutorture_runnable,
925 		.maxlen         = sizeof(int),
926 		.mode           = 0644,
927 		.proc_handler	= proc_dointvec,
928 	},
929 #endif
930 #ifdef CONFIG_PERF_EVENTS
931 	{
932 		.procname	= "perf_event_paranoid",
933 		.data		= &sysctl_perf_event_paranoid,
934 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
935 		.mode		= 0644,
936 		.proc_handler	= proc_dointvec,
937 	},
938 	{
939 		.procname	= "perf_event_mlock_kb",
940 		.data		= &sysctl_perf_event_mlock,
941 		.maxlen		= sizeof(sysctl_perf_event_mlock),
942 		.mode		= 0644,
943 		.proc_handler	= proc_dointvec,
944 	},
945 	{
946 		.procname	= "perf_event_max_sample_rate",
947 		.data		= &sysctl_perf_event_sample_rate,
948 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
949 		.mode		= 0644,
950 		.proc_handler	= proc_dointvec,
951 	},
952 #endif
953 #ifdef CONFIG_KMEMCHECK
954 	{
955 		.procname	= "kmemcheck",
956 		.data		= &kmemcheck_enabled,
957 		.maxlen		= sizeof(int),
958 		.mode		= 0644,
959 		.proc_handler	= proc_dointvec,
960 	},
961 #endif
962 #ifdef CONFIG_BLOCK
963 	{
964 		.procname	= "blk_iopoll",
965 		.data		= &blk_iopoll_enabled,
966 		.maxlen		= sizeof(int),
967 		.mode		= 0644,
968 		.proc_handler	= proc_dointvec,
969 	},
970 #endif
971 	{ }
972 };
973 
974 static struct ctl_table vm_table[] = {
975 	{
976 		.procname	= "overcommit_memory",
977 		.data		= &sysctl_overcommit_memory,
978 		.maxlen		= sizeof(sysctl_overcommit_memory),
979 		.mode		= 0644,
980 		.proc_handler	= proc_dointvec,
981 	},
982 	{
983 		.procname	= "panic_on_oom",
984 		.data		= &sysctl_panic_on_oom,
985 		.maxlen		= sizeof(sysctl_panic_on_oom),
986 		.mode		= 0644,
987 		.proc_handler	= proc_dointvec,
988 	},
989 	{
990 		.procname	= "oom_kill_allocating_task",
991 		.data		= &sysctl_oom_kill_allocating_task,
992 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
993 		.mode		= 0644,
994 		.proc_handler	= proc_dointvec,
995 	},
996 	{
997 		.procname	= "oom_dump_tasks",
998 		.data		= &sysctl_oom_dump_tasks,
999 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1000 		.mode		= 0644,
1001 		.proc_handler	= proc_dointvec,
1002 	},
1003 	{
1004 		.procname	= "overcommit_ratio",
1005 		.data		= &sysctl_overcommit_ratio,
1006 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1007 		.mode		= 0644,
1008 		.proc_handler	= proc_dointvec,
1009 	},
1010 	{
1011 		.procname	= "page-cluster",
1012 		.data		= &page_cluster,
1013 		.maxlen		= sizeof(int),
1014 		.mode		= 0644,
1015 		.proc_handler	= proc_dointvec,
1016 	},
1017 	{
1018 		.procname	= "dirty_background_ratio",
1019 		.data		= &dirty_background_ratio,
1020 		.maxlen		= sizeof(dirty_background_ratio),
1021 		.mode		= 0644,
1022 		.proc_handler	= dirty_background_ratio_handler,
1023 		.extra1		= &zero,
1024 		.extra2		= &one_hundred,
1025 	},
1026 	{
1027 		.procname	= "dirty_background_bytes",
1028 		.data		= &dirty_background_bytes,
1029 		.maxlen		= sizeof(dirty_background_bytes),
1030 		.mode		= 0644,
1031 		.proc_handler	= dirty_background_bytes_handler,
1032 		.extra1		= &one_ul,
1033 	},
1034 	{
1035 		.procname	= "dirty_ratio",
1036 		.data		= &vm_dirty_ratio,
1037 		.maxlen		= sizeof(vm_dirty_ratio),
1038 		.mode		= 0644,
1039 		.proc_handler	= dirty_ratio_handler,
1040 		.extra1		= &zero,
1041 		.extra2		= &one_hundred,
1042 	},
1043 	{
1044 		.procname	= "dirty_bytes",
1045 		.data		= &vm_dirty_bytes,
1046 		.maxlen		= sizeof(vm_dirty_bytes),
1047 		.mode		= 0644,
1048 		.proc_handler	= dirty_bytes_handler,
1049 		.extra1		= &dirty_bytes_min,
1050 	},
1051 	{
1052 		.procname	= "dirty_writeback_centisecs",
1053 		.data		= &dirty_writeback_interval,
1054 		.maxlen		= sizeof(dirty_writeback_interval),
1055 		.mode		= 0644,
1056 		.proc_handler	= dirty_writeback_centisecs_handler,
1057 	},
1058 	{
1059 		.procname	= "dirty_expire_centisecs",
1060 		.data		= &dirty_expire_interval,
1061 		.maxlen		= sizeof(dirty_expire_interval),
1062 		.mode		= 0644,
1063 		.proc_handler	= proc_dointvec,
1064 	},
1065 	{
1066 		.procname	= "nr_pdflush_threads",
1067 		.data		= &nr_pdflush_threads,
1068 		.maxlen		= sizeof nr_pdflush_threads,
1069 		.mode		= 0444 /* read-only*/,
1070 		.proc_handler	= proc_dointvec,
1071 	},
1072 	{
1073 		.procname	= "swappiness",
1074 		.data		= &vm_swappiness,
1075 		.maxlen		= sizeof(vm_swappiness),
1076 		.mode		= 0644,
1077 		.proc_handler	= proc_dointvec_minmax,
1078 		.extra1		= &zero,
1079 		.extra2		= &one_hundred,
1080 	},
1081 #ifdef CONFIG_HUGETLB_PAGE
1082 	{
1083 		.procname	= "nr_hugepages",
1084 		.data		= NULL,
1085 		.maxlen		= sizeof(unsigned long),
1086 		.mode		= 0644,
1087 		.proc_handler	= hugetlb_sysctl_handler,
1088 		.extra1		= (void *)&hugetlb_zero,
1089 		.extra2		= (void *)&hugetlb_infinity,
1090 	},
1091 #ifdef CONFIG_NUMA
1092 	{
1093 		.procname       = "nr_hugepages_mempolicy",
1094 		.data           = NULL,
1095 		.maxlen         = sizeof(unsigned long),
1096 		.mode           = 0644,
1097 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1098 		.extra1		= (void *)&hugetlb_zero,
1099 		.extra2		= (void *)&hugetlb_infinity,
1100 	},
1101 #endif
1102 	 {
1103 		.procname	= "hugetlb_shm_group",
1104 		.data		= &sysctl_hugetlb_shm_group,
1105 		.maxlen		= sizeof(gid_t),
1106 		.mode		= 0644,
1107 		.proc_handler	= proc_dointvec,
1108 	 },
1109 	 {
1110 		.procname	= "hugepages_treat_as_movable",
1111 		.data		= &hugepages_treat_as_movable,
1112 		.maxlen		= sizeof(int),
1113 		.mode		= 0644,
1114 		.proc_handler	= hugetlb_treat_movable_handler,
1115 	},
1116 	{
1117 		.procname	= "nr_overcommit_hugepages",
1118 		.data		= NULL,
1119 		.maxlen		= sizeof(unsigned long),
1120 		.mode		= 0644,
1121 		.proc_handler	= hugetlb_overcommit_handler,
1122 		.extra1		= (void *)&hugetlb_zero,
1123 		.extra2		= (void *)&hugetlb_infinity,
1124 	},
1125 #endif
1126 	{
1127 		.procname	= "lowmem_reserve_ratio",
1128 		.data		= &sysctl_lowmem_reserve_ratio,
1129 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1130 		.mode		= 0644,
1131 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1132 	},
1133 	{
1134 		.procname	= "drop_caches",
1135 		.data		= &sysctl_drop_caches,
1136 		.maxlen		= sizeof(int),
1137 		.mode		= 0644,
1138 		.proc_handler	= drop_caches_sysctl_handler,
1139 	},
1140 #ifdef CONFIG_COMPACTION
1141 	{
1142 		.procname	= "compact_memory",
1143 		.data		= &sysctl_compact_memory,
1144 		.maxlen		= sizeof(int),
1145 		.mode		= 0200,
1146 		.proc_handler	= sysctl_compaction_handler,
1147 	},
1148 	{
1149 		.procname	= "extfrag_threshold",
1150 		.data		= &sysctl_extfrag_threshold,
1151 		.maxlen		= sizeof(int),
1152 		.mode		= 0644,
1153 		.proc_handler	= sysctl_extfrag_handler,
1154 		.extra1		= &min_extfrag_threshold,
1155 		.extra2		= &max_extfrag_threshold,
1156 	},
1157 
1158 #endif /* CONFIG_COMPACTION */
1159 	{
1160 		.procname	= "min_free_kbytes",
1161 		.data		= &min_free_kbytes,
1162 		.maxlen		= sizeof(min_free_kbytes),
1163 		.mode		= 0644,
1164 		.proc_handler	= min_free_kbytes_sysctl_handler,
1165 		.extra1		= &zero,
1166 	},
1167 	{
1168 		.procname	= "percpu_pagelist_fraction",
1169 		.data		= &percpu_pagelist_fraction,
1170 		.maxlen		= sizeof(percpu_pagelist_fraction),
1171 		.mode		= 0644,
1172 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1173 		.extra1		= &min_percpu_pagelist_fract,
1174 	},
1175 #ifdef CONFIG_MMU
1176 	{
1177 		.procname	= "max_map_count",
1178 		.data		= &sysctl_max_map_count,
1179 		.maxlen		= sizeof(sysctl_max_map_count),
1180 		.mode		= 0644,
1181 		.proc_handler	= proc_dointvec_minmax,
1182 		.extra1		= &zero,
1183 	},
1184 #else
1185 	{
1186 		.procname	= "nr_trim_pages",
1187 		.data		= &sysctl_nr_trim_pages,
1188 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1189 		.mode		= 0644,
1190 		.proc_handler	= proc_dointvec_minmax,
1191 		.extra1		= &zero,
1192 	},
1193 #endif
1194 	{
1195 		.procname	= "laptop_mode",
1196 		.data		= &laptop_mode,
1197 		.maxlen		= sizeof(laptop_mode),
1198 		.mode		= 0644,
1199 		.proc_handler	= proc_dointvec_jiffies,
1200 	},
1201 	{
1202 		.procname	= "block_dump",
1203 		.data		= &block_dump,
1204 		.maxlen		= sizeof(block_dump),
1205 		.mode		= 0644,
1206 		.proc_handler	= proc_dointvec,
1207 		.extra1		= &zero,
1208 	},
1209 	{
1210 		.procname	= "vfs_cache_pressure",
1211 		.data		= &sysctl_vfs_cache_pressure,
1212 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1213 		.mode		= 0644,
1214 		.proc_handler	= proc_dointvec,
1215 		.extra1		= &zero,
1216 	},
1217 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1218 	{
1219 		.procname	= "legacy_va_layout",
1220 		.data		= &sysctl_legacy_va_layout,
1221 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1222 		.mode		= 0644,
1223 		.proc_handler	= proc_dointvec,
1224 		.extra1		= &zero,
1225 	},
1226 #endif
1227 #ifdef CONFIG_NUMA
1228 	{
1229 		.procname	= "zone_reclaim_mode",
1230 		.data		= &zone_reclaim_mode,
1231 		.maxlen		= sizeof(zone_reclaim_mode),
1232 		.mode		= 0644,
1233 		.proc_handler	= proc_dointvec,
1234 		.extra1		= &zero,
1235 	},
1236 	{
1237 		.procname	= "min_unmapped_ratio",
1238 		.data		= &sysctl_min_unmapped_ratio,
1239 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1240 		.mode		= 0644,
1241 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1242 		.extra1		= &zero,
1243 		.extra2		= &one_hundred,
1244 	},
1245 	{
1246 		.procname	= "min_slab_ratio",
1247 		.data		= &sysctl_min_slab_ratio,
1248 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1249 		.mode		= 0644,
1250 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1251 		.extra1		= &zero,
1252 		.extra2		= &one_hundred,
1253 	},
1254 #endif
1255 #ifdef CONFIG_SMP
1256 	{
1257 		.procname	= "stat_interval",
1258 		.data		= &sysctl_stat_interval,
1259 		.maxlen		= sizeof(sysctl_stat_interval),
1260 		.mode		= 0644,
1261 		.proc_handler	= proc_dointvec_jiffies,
1262 	},
1263 #endif
1264 #ifdef CONFIG_MMU
1265 	{
1266 		.procname	= "mmap_min_addr",
1267 		.data		= &dac_mmap_min_addr,
1268 		.maxlen		= sizeof(unsigned long),
1269 		.mode		= 0644,
1270 		.proc_handler	= mmap_min_addr_handler,
1271 	},
1272 #endif
1273 #ifdef CONFIG_NUMA
1274 	{
1275 		.procname	= "numa_zonelist_order",
1276 		.data		= &numa_zonelist_order,
1277 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1278 		.mode		= 0644,
1279 		.proc_handler	= numa_zonelist_order_handler,
1280 	},
1281 #endif
1282 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1283    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1284 	{
1285 		.procname	= "vdso_enabled",
1286 		.data		= &vdso_enabled,
1287 		.maxlen		= sizeof(vdso_enabled),
1288 		.mode		= 0644,
1289 		.proc_handler	= proc_dointvec,
1290 		.extra1		= &zero,
1291 	},
1292 #endif
1293 #ifdef CONFIG_HIGHMEM
1294 	{
1295 		.procname	= "highmem_is_dirtyable",
1296 		.data		= &vm_highmem_is_dirtyable,
1297 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1298 		.mode		= 0644,
1299 		.proc_handler	= proc_dointvec_minmax,
1300 		.extra1		= &zero,
1301 		.extra2		= &one,
1302 	},
1303 #endif
1304 	{
1305 		.procname	= "scan_unevictable_pages",
1306 		.data		= &scan_unevictable_pages,
1307 		.maxlen		= sizeof(scan_unevictable_pages),
1308 		.mode		= 0644,
1309 		.proc_handler	= scan_unevictable_handler,
1310 	},
1311 #ifdef CONFIG_MEMORY_FAILURE
1312 	{
1313 		.procname	= "memory_failure_early_kill",
1314 		.data		= &sysctl_memory_failure_early_kill,
1315 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1316 		.mode		= 0644,
1317 		.proc_handler	= proc_dointvec_minmax,
1318 		.extra1		= &zero,
1319 		.extra2		= &one,
1320 	},
1321 	{
1322 		.procname	= "memory_failure_recovery",
1323 		.data		= &sysctl_memory_failure_recovery,
1324 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1325 		.mode		= 0644,
1326 		.proc_handler	= proc_dointvec_minmax,
1327 		.extra1		= &zero,
1328 		.extra2		= &one,
1329 	},
1330 #endif
1331 	{ }
1332 };
1333 
1334 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1335 static struct ctl_table binfmt_misc_table[] = {
1336 	{ }
1337 };
1338 #endif
1339 
1340 static struct ctl_table fs_table[] = {
1341 	{
1342 		.procname	= "inode-nr",
1343 		.data		= &inodes_stat,
1344 		.maxlen		= 2*sizeof(int),
1345 		.mode		= 0444,
1346 		.proc_handler	= proc_nr_inodes,
1347 	},
1348 	{
1349 		.procname	= "inode-state",
1350 		.data		= &inodes_stat,
1351 		.maxlen		= 7*sizeof(int),
1352 		.mode		= 0444,
1353 		.proc_handler	= proc_nr_inodes,
1354 	},
1355 	{
1356 		.procname	= "file-nr",
1357 		.data		= &files_stat,
1358 		.maxlen		= sizeof(files_stat),
1359 		.mode		= 0444,
1360 		.proc_handler	= proc_nr_files,
1361 	},
1362 	{
1363 		.procname	= "file-max",
1364 		.data		= &files_stat.max_files,
1365 		.maxlen		= sizeof(files_stat.max_files),
1366 		.mode		= 0644,
1367 		.proc_handler	= proc_doulongvec_minmax,
1368 	},
1369 	{
1370 		.procname	= "nr_open",
1371 		.data		= &sysctl_nr_open,
1372 		.maxlen		= sizeof(int),
1373 		.mode		= 0644,
1374 		.proc_handler	= proc_dointvec_minmax,
1375 		.extra1		= &sysctl_nr_open_min,
1376 		.extra2		= &sysctl_nr_open_max,
1377 	},
1378 	{
1379 		.procname	= "dentry-state",
1380 		.data		= &dentry_stat,
1381 		.maxlen		= 6*sizeof(int),
1382 		.mode		= 0444,
1383 		.proc_handler	= proc_nr_dentry,
1384 	},
1385 	{
1386 		.procname	= "overflowuid",
1387 		.data		= &fs_overflowuid,
1388 		.maxlen		= sizeof(int),
1389 		.mode		= 0644,
1390 		.proc_handler	= proc_dointvec_minmax,
1391 		.extra1		= &minolduid,
1392 		.extra2		= &maxolduid,
1393 	},
1394 	{
1395 		.procname	= "overflowgid",
1396 		.data		= &fs_overflowgid,
1397 		.maxlen		= sizeof(int),
1398 		.mode		= 0644,
1399 		.proc_handler	= proc_dointvec_minmax,
1400 		.extra1		= &minolduid,
1401 		.extra2		= &maxolduid,
1402 	},
1403 #ifdef CONFIG_FILE_LOCKING
1404 	{
1405 		.procname	= "leases-enable",
1406 		.data		= &leases_enable,
1407 		.maxlen		= sizeof(int),
1408 		.mode		= 0644,
1409 		.proc_handler	= proc_dointvec,
1410 	},
1411 #endif
1412 #ifdef CONFIG_DNOTIFY
1413 	{
1414 		.procname	= "dir-notify-enable",
1415 		.data		= &dir_notify_enable,
1416 		.maxlen		= sizeof(int),
1417 		.mode		= 0644,
1418 		.proc_handler	= proc_dointvec,
1419 	},
1420 #endif
1421 #ifdef CONFIG_MMU
1422 #ifdef CONFIG_FILE_LOCKING
1423 	{
1424 		.procname	= "lease-break-time",
1425 		.data		= &lease_break_time,
1426 		.maxlen		= sizeof(int),
1427 		.mode		= 0644,
1428 		.proc_handler	= proc_dointvec,
1429 	},
1430 #endif
1431 #ifdef CONFIG_AIO
1432 	{
1433 		.procname	= "aio-nr",
1434 		.data		= &aio_nr,
1435 		.maxlen		= sizeof(aio_nr),
1436 		.mode		= 0444,
1437 		.proc_handler	= proc_doulongvec_minmax,
1438 	},
1439 	{
1440 		.procname	= "aio-max-nr",
1441 		.data		= &aio_max_nr,
1442 		.maxlen		= sizeof(aio_max_nr),
1443 		.mode		= 0644,
1444 		.proc_handler	= proc_doulongvec_minmax,
1445 	},
1446 #endif /* CONFIG_AIO */
1447 #ifdef CONFIG_INOTIFY_USER
1448 	{
1449 		.procname	= "inotify",
1450 		.mode		= 0555,
1451 		.child		= inotify_table,
1452 	},
1453 #endif
1454 #ifdef CONFIG_EPOLL
1455 	{
1456 		.procname	= "epoll",
1457 		.mode		= 0555,
1458 		.child		= epoll_table,
1459 	},
1460 #endif
1461 #endif
1462 	{
1463 		.procname	= "suid_dumpable",
1464 		.data		= &suid_dumpable,
1465 		.maxlen		= sizeof(int),
1466 		.mode		= 0644,
1467 		.proc_handler	= proc_dointvec_minmax,
1468 		.extra1		= &zero,
1469 		.extra2		= &two,
1470 	},
1471 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1472 	{
1473 		.procname	= "binfmt_misc",
1474 		.mode		= 0555,
1475 		.child		= binfmt_misc_table,
1476 	},
1477 #endif
1478 	{
1479 		.procname	= "pipe-max-size",
1480 		.data		= &pipe_max_size,
1481 		.maxlen		= sizeof(int),
1482 		.mode		= 0644,
1483 		.proc_handler	= &pipe_proc_fn,
1484 		.extra1		= &pipe_min_size,
1485 	},
1486 	{ }
1487 };
1488 
1489 static struct ctl_table debug_table[] = {
1490 #if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
1491     defined(CONFIG_S390)
1492 	{
1493 		.procname	= "exception-trace",
1494 		.data		= &show_unhandled_signals,
1495 		.maxlen		= sizeof(int),
1496 		.mode		= 0644,
1497 		.proc_handler	= proc_dointvec
1498 	},
1499 #endif
1500 #if defined(CONFIG_OPTPROBES)
1501 	{
1502 		.procname	= "kprobes-optimization",
1503 		.data		= &sysctl_kprobes_optimization,
1504 		.maxlen		= sizeof(int),
1505 		.mode		= 0644,
1506 		.proc_handler	= proc_kprobes_optimization_handler,
1507 		.extra1		= &zero,
1508 		.extra2		= &one,
1509 	},
1510 #endif
1511 	{ }
1512 };
1513 
1514 static struct ctl_table dev_table[] = {
1515 	{ }
1516 };
1517 
1518 static DEFINE_SPINLOCK(sysctl_lock);
1519 
1520 /* called under sysctl_lock */
1521 static int use_table(struct ctl_table_header *p)
1522 {
1523 	if (unlikely(p->unregistering))
1524 		return 0;
1525 	p->used++;
1526 	return 1;
1527 }
1528 
1529 /* called under sysctl_lock */
1530 static void unuse_table(struct ctl_table_header *p)
1531 {
1532 	if (!--p->used)
1533 		if (unlikely(p->unregistering))
1534 			complete(p->unregistering);
1535 }
1536 
1537 /* called under sysctl_lock, will reacquire if has to wait */
1538 static void start_unregistering(struct ctl_table_header *p)
1539 {
1540 	/*
1541 	 * if p->used is 0, nobody will ever touch that entry again;
1542 	 * we'll eliminate all paths to it before dropping sysctl_lock
1543 	 */
1544 	if (unlikely(p->used)) {
1545 		struct completion wait;
1546 		init_completion(&wait);
1547 		p->unregistering = &wait;
1548 		spin_unlock(&sysctl_lock);
1549 		wait_for_completion(&wait);
1550 		spin_lock(&sysctl_lock);
1551 	} else {
1552 		/* anything non-NULL; we'll never dereference it */
1553 		p->unregistering = ERR_PTR(-EINVAL);
1554 	}
1555 	/*
1556 	 * do not remove from the list until nobody holds it; walking the
1557 	 * list in do_sysctl() relies on that.
1558 	 */
1559 	list_del_init(&p->ctl_entry);
1560 }
1561 
1562 void sysctl_head_get(struct ctl_table_header *head)
1563 {
1564 	spin_lock(&sysctl_lock);
1565 	head->count++;
1566 	spin_unlock(&sysctl_lock);
1567 }
1568 
1569 void sysctl_head_put(struct ctl_table_header *head)
1570 {
1571 	spin_lock(&sysctl_lock);
1572 	if (!--head->count)
1573 		kfree(head);
1574 	spin_unlock(&sysctl_lock);
1575 }
1576 
1577 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1578 {
1579 	if (!head)
1580 		BUG();
1581 	spin_lock(&sysctl_lock);
1582 	if (!use_table(head))
1583 		head = ERR_PTR(-ENOENT);
1584 	spin_unlock(&sysctl_lock);
1585 	return head;
1586 }
1587 
1588 void sysctl_head_finish(struct ctl_table_header *head)
1589 {
1590 	if (!head)
1591 		return;
1592 	spin_lock(&sysctl_lock);
1593 	unuse_table(head);
1594 	spin_unlock(&sysctl_lock);
1595 }
1596 
1597 static struct ctl_table_set *
1598 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1599 {
1600 	struct ctl_table_set *set = &root->default_set;
1601 	if (root->lookup)
1602 		set = root->lookup(root, namespaces);
1603 	return set;
1604 }
1605 
1606 static struct list_head *
1607 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1608 {
1609 	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1610 	return &set->list;
1611 }
1612 
1613 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1614 					    struct ctl_table_header *prev)
1615 {
1616 	struct ctl_table_root *root;
1617 	struct list_head *header_list;
1618 	struct ctl_table_header *head;
1619 	struct list_head *tmp;
1620 
1621 	spin_lock(&sysctl_lock);
1622 	if (prev) {
1623 		head = prev;
1624 		tmp = &prev->ctl_entry;
1625 		unuse_table(prev);
1626 		goto next;
1627 	}
1628 	tmp = &root_table_header.ctl_entry;
1629 	for (;;) {
1630 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1631 
1632 		if (!use_table(head))
1633 			goto next;
1634 		spin_unlock(&sysctl_lock);
1635 		return head;
1636 	next:
1637 		root = head->root;
1638 		tmp = tmp->next;
1639 		header_list = lookup_header_list(root, namespaces);
1640 		if (tmp != header_list)
1641 			continue;
1642 
1643 		do {
1644 			root = list_entry(root->root_list.next,
1645 					struct ctl_table_root, root_list);
1646 			if (root == &sysctl_table_root)
1647 				goto out;
1648 			header_list = lookup_header_list(root, namespaces);
1649 		} while (list_empty(header_list));
1650 		tmp = header_list->next;
1651 	}
1652 out:
1653 	spin_unlock(&sysctl_lock);
1654 	return NULL;
1655 }
1656 
1657 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1658 {
1659 	return __sysctl_head_next(current->nsproxy, prev);
1660 }
1661 
1662 void register_sysctl_root(struct ctl_table_root *root)
1663 {
1664 	spin_lock(&sysctl_lock);
1665 	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1666 	spin_unlock(&sysctl_lock);
1667 }
1668 
1669 /*
1670  * sysctl_perm does NOT grant the superuser all rights automatically, because
1671  * some sysctl variables are readonly even to root.
1672  */
1673 
1674 static int test_perm(int mode, int op)
1675 {
1676 	if (!current_euid())
1677 		mode >>= 6;
1678 	else if (in_egroup_p(0))
1679 		mode >>= 3;
1680 	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1681 		return 0;
1682 	return -EACCES;
1683 }
1684 
1685 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1686 {
1687 	int error;
1688 	int mode;
1689 
1690 	error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1691 	if (error)
1692 		return error;
1693 
1694 	if (root->permissions)
1695 		mode = root->permissions(root, current->nsproxy, table);
1696 	else
1697 		mode = table->mode;
1698 
1699 	return test_perm(mode, op);
1700 }
1701 
1702 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1703 {
1704 	for (; table->procname; table++) {
1705 		table->parent = parent;
1706 		if (table->child)
1707 			sysctl_set_parent(table, table->child);
1708 	}
1709 }
1710 
1711 static __init int sysctl_init(void)
1712 {
1713 	sysctl_set_parent(NULL, root_table);
1714 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1715 	sysctl_check_table(current->nsproxy, root_table);
1716 #endif
1717 	return 0;
1718 }
1719 
1720 core_initcall(sysctl_init);
1721 
1722 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1723 				      struct ctl_table *table)
1724 {
1725 	struct ctl_table *p;
1726 	const char *s = branch->procname;
1727 
1728 	/* branch should have named subdirectory as its first element */
1729 	if (!s || !branch->child)
1730 		return NULL;
1731 
1732 	/* ... and nothing else */
1733 	if (branch[1].procname)
1734 		return NULL;
1735 
1736 	/* table should contain subdirectory with the same name */
1737 	for (p = table; p->procname; p++) {
1738 		if (!p->child)
1739 			continue;
1740 		if (p->procname && strcmp(p->procname, s) == 0)
1741 			return p;
1742 	}
1743 	return NULL;
1744 }
1745 
1746 /* see if attaching q to p would be an improvement */
1747 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1748 {
1749 	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1750 	struct ctl_table *next;
1751 	int is_better = 0;
1752 	int not_in_parent = !p->attached_by;
1753 
1754 	while ((next = is_branch_in(by, to)) != NULL) {
1755 		if (by == q->attached_by)
1756 			is_better = 1;
1757 		if (to == p->attached_by)
1758 			not_in_parent = 1;
1759 		by = by->child;
1760 		to = next->child;
1761 	}
1762 
1763 	if (is_better && not_in_parent) {
1764 		q->attached_by = by;
1765 		q->attached_to = to;
1766 		q->parent = p;
1767 	}
1768 }
1769 
1770 /**
1771  * __register_sysctl_paths - register a sysctl hierarchy
1772  * @root: List of sysctl headers to register on
1773  * @namespaces: Data to compute which lists of sysctl entries are visible
1774  * @path: The path to the directory the sysctl table is in.
1775  * @table: the top-level table structure
1776  *
1777  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1778  * array. A completely 0 filled entry terminates the table.
1779  *
1780  * The members of the &struct ctl_table structure are used as follows:
1781  *
1782  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1783  *            enter a sysctl file
1784  *
1785  * data - a pointer to data for use by proc_handler
1786  *
1787  * maxlen - the maximum size in bytes of the data
1788  *
1789  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1790  *
1791  * child - a pointer to the child sysctl table if this entry is a directory, or
1792  *         %NULL.
1793  *
1794  * proc_handler - the text handler routine (described below)
1795  *
1796  * de - for internal use by the sysctl routines
1797  *
1798  * extra1, extra2 - extra pointers usable by the proc handler routines
1799  *
1800  * Leaf nodes in the sysctl tree will be represented by a single file
1801  * under /proc; non-leaf nodes will be represented by directories.
1802  *
1803  * sysctl(2) can automatically manage read and write requests through
1804  * the sysctl table.  The data and maxlen fields of the ctl_table
1805  * struct enable minimal validation of the values being written to be
1806  * performed, and the mode field allows minimal authentication.
1807  *
1808  * There must be a proc_handler routine for any terminal nodes
1809  * mirrored under /proc/sys (non-terminals are handled by a built-in
1810  * directory handler).  Several default handlers are available to
1811  * cover common cases -
1812  *
1813  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1814  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1815  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1816  *
1817  * It is the handler's job to read the input buffer from user memory
1818  * and process it. The handler should return 0 on success.
1819  *
1820  * This routine returns %NULL on a failure to register, and a pointer
1821  * to the table header on success.
1822  */
1823 struct ctl_table_header *__register_sysctl_paths(
1824 	struct ctl_table_root *root,
1825 	struct nsproxy *namespaces,
1826 	const struct ctl_path *path, struct ctl_table *table)
1827 {
1828 	struct ctl_table_header *header;
1829 	struct ctl_table *new, **prevp;
1830 	unsigned int n, npath;
1831 	struct ctl_table_set *set;
1832 
1833 	/* Count the path components */
1834 	for (npath = 0; path[npath].procname; ++npath)
1835 		;
1836 
1837 	/*
1838 	 * For each path component, allocate a 2-element ctl_table array.
1839 	 * The first array element will be filled with the sysctl entry
1840 	 * for this, the second will be the sentinel (procname == 0).
1841 	 *
1842 	 * We allocate everything in one go so that we don't have to
1843 	 * worry about freeing additional memory in unregister_sysctl_table.
1844 	 */
1845 	header = kzalloc(sizeof(struct ctl_table_header) +
1846 			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1847 	if (!header)
1848 		return NULL;
1849 
1850 	new = (struct ctl_table *) (header + 1);
1851 
1852 	/* Now connect the dots */
1853 	prevp = &header->ctl_table;
1854 	for (n = 0; n < npath; ++n, ++path) {
1855 		/* Copy the procname */
1856 		new->procname = path->procname;
1857 		new->mode     = 0555;
1858 
1859 		*prevp = new;
1860 		prevp = &new->child;
1861 
1862 		new += 2;
1863 	}
1864 	*prevp = table;
1865 	header->ctl_table_arg = table;
1866 
1867 	INIT_LIST_HEAD(&header->ctl_entry);
1868 	header->used = 0;
1869 	header->unregistering = NULL;
1870 	header->root = root;
1871 	sysctl_set_parent(NULL, header->ctl_table);
1872 	header->count = 1;
1873 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1874 	if (sysctl_check_table(namespaces, header->ctl_table)) {
1875 		kfree(header);
1876 		return NULL;
1877 	}
1878 #endif
1879 	spin_lock(&sysctl_lock);
1880 	header->set = lookup_header_set(root, namespaces);
1881 	header->attached_by = header->ctl_table;
1882 	header->attached_to = root_table;
1883 	header->parent = &root_table_header;
1884 	for (set = header->set; set; set = set->parent) {
1885 		struct ctl_table_header *p;
1886 		list_for_each_entry(p, &set->list, ctl_entry) {
1887 			if (p->unregistering)
1888 				continue;
1889 			try_attach(p, header);
1890 		}
1891 	}
1892 	header->parent->count++;
1893 	list_add_tail(&header->ctl_entry, &header->set->list);
1894 	spin_unlock(&sysctl_lock);
1895 
1896 	return header;
1897 }
1898 
1899 /**
1900  * register_sysctl_table_path - register a sysctl table hierarchy
1901  * @path: The path to the directory the sysctl table is in.
1902  * @table: the top-level table structure
1903  *
1904  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1905  * array. A completely 0 filled entry terminates the table.
1906  *
1907  * See __register_sysctl_paths for more details.
1908  */
1909 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1910 						struct ctl_table *table)
1911 {
1912 	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1913 					path, table);
1914 }
1915 
1916 /**
1917  * register_sysctl_table - register a sysctl table hierarchy
1918  * @table: the top-level table structure
1919  *
1920  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1921  * array. A completely 0 filled entry terminates the table.
1922  *
1923  * See register_sysctl_paths for more details.
1924  */
1925 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1926 {
1927 	static const struct ctl_path null_path[] = { {} };
1928 
1929 	return register_sysctl_paths(null_path, table);
1930 }
1931 
1932 /**
1933  * unregister_sysctl_table - unregister a sysctl table hierarchy
1934  * @header: the header returned from register_sysctl_table
1935  *
1936  * Unregisters the sysctl table and all children. proc entries may not
1937  * actually be removed until they are no longer used by anyone.
1938  */
1939 void unregister_sysctl_table(struct ctl_table_header * header)
1940 {
1941 	might_sleep();
1942 
1943 	if (header == NULL)
1944 		return;
1945 
1946 	spin_lock(&sysctl_lock);
1947 	start_unregistering(header);
1948 	if (!--header->parent->count) {
1949 		WARN_ON(1);
1950 		kfree(header->parent);
1951 	}
1952 	if (!--header->count)
1953 		kfree(header);
1954 	spin_unlock(&sysctl_lock);
1955 }
1956 
1957 int sysctl_is_seen(struct ctl_table_header *p)
1958 {
1959 	struct ctl_table_set *set = p->set;
1960 	int res;
1961 	spin_lock(&sysctl_lock);
1962 	if (p->unregistering)
1963 		res = 0;
1964 	else if (!set->is_seen)
1965 		res = 1;
1966 	else
1967 		res = set->is_seen(set);
1968 	spin_unlock(&sysctl_lock);
1969 	return res;
1970 }
1971 
1972 void setup_sysctl_set(struct ctl_table_set *p,
1973 	struct ctl_table_set *parent,
1974 	int (*is_seen)(struct ctl_table_set *))
1975 {
1976 	INIT_LIST_HEAD(&p->list);
1977 	p->parent = parent ? parent : &sysctl_table_root.default_set;
1978 	p->is_seen = is_seen;
1979 }
1980 
1981 #else /* !CONFIG_SYSCTL */
1982 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1983 {
1984 	return NULL;
1985 }
1986 
1987 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1988 						    struct ctl_table *table)
1989 {
1990 	return NULL;
1991 }
1992 
1993 void unregister_sysctl_table(struct ctl_table_header * table)
1994 {
1995 }
1996 
1997 void setup_sysctl_set(struct ctl_table_set *p,
1998 	struct ctl_table_set *parent,
1999 	int (*is_seen)(struct ctl_table_set *))
2000 {
2001 }
2002 
2003 void sysctl_head_put(struct ctl_table_header *head)
2004 {
2005 }
2006 
2007 #endif /* CONFIG_SYSCTL */
2008 
2009 /*
2010  * /proc/sys support
2011  */
2012 
2013 #ifdef CONFIG_PROC_SYSCTL
2014 
2015 static int _proc_do_string(void* data, int maxlen, int write,
2016 			   void __user *buffer,
2017 			   size_t *lenp, loff_t *ppos)
2018 {
2019 	size_t len;
2020 	char __user *p;
2021 	char c;
2022 
2023 	if (!data || !maxlen || !*lenp) {
2024 		*lenp = 0;
2025 		return 0;
2026 	}
2027 
2028 	if (write) {
2029 		len = 0;
2030 		p = buffer;
2031 		while (len < *lenp) {
2032 			if (get_user(c, p++))
2033 				return -EFAULT;
2034 			if (c == 0 || c == '\n')
2035 				break;
2036 			len++;
2037 		}
2038 		if (len >= maxlen)
2039 			len = maxlen-1;
2040 		if(copy_from_user(data, buffer, len))
2041 			return -EFAULT;
2042 		((char *) data)[len] = 0;
2043 		*ppos += *lenp;
2044 	} else {
2045 		len = strlen(data);
2046 		if (len > maxlen)
2047 			len = maxlen;
2048 
2049 		if (*ppos > len) {
2050 			*lenp = 0;
2051 			return 0;
2052 		}
2053 
2054 		data += *ppos;
2055 		len  -= *ppos;
2056 
2057 		if (len > *lenp)
2058 			len = *lenp;
2059 		if (len)
2060 			if(copy_to_user(buffer, data, len))
2061 				return -EFAULT;
2062 		if (len < *lenp) {
2063 			if(put_user('\n', ((char __user *) buffer) + len))
2064 				return -EFAULT;
2065 			len++;
2066 		}
2067 		*lenp = len;
2068 		*ppos += len;
2069 	}
2070 	return 0;
2071 }
2072 
2073 /**
2074  * proc_dostring - read a string sysctl
2075  * @table: the sysctl table
2076  * @write: %TRUE if this is a write to the sysctl file
2077  * @buffer: the user buffer
2078  * @lenp: the size of the user buffer
2079  * @ppos: file position
2080  *
2081  * Reads/writes a string from/to the user buffer. If the kernel
2082  * buffer provided is not large enough to hold the string, the
2083  * string is truncated. The copied string is %NULL-terminated.
2084  * If the string is being read by the user process, it is copied
2085  * and a newline '\n' is added. It is truncated if the buffer is
2086  * not large enough.
2087  *
2088  * Returns 0 on success.
2089  */
2090 int proc_dostring(struct ctl_table *table, int write,
2091 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2092 {
2093 	return _proc_do_string(table->data, table->maxlen, write,
2094 			       buffer, lenp, ppos);
2095 }
2096 
2097 static size_t proc_skip_spaces(char **buf)
2098 {
2099 	size_t ret;
2100 	char *tmp = skip_spaces(*buf);
2101 	ret = tmp - *buf;
2102 	*buf = tmp;
2103 	return ret;
2104 }
2105 
2106 static void proc_skip_char(char **buf, size_t *size, const char v)
2107 {
2108 	while (*size) {
2109 		if (**buf != v)
2110 			break;
2111 		(*size)--;
2112 		(*buf)++;
2113 	}
2114 }
2115 
2116 #define TMPBUFLEN 22
2117 /**
2118  * proc_get_long - reads an ASCII formatted integer from a user buffer
2119  *
2120  * @buf: a kernel buffer
2121  * @size: size of the kernel buffer
2122  * @val: this is where the number will be stored
2123  * @neg: set to %TRUE if number is negative
2124  * @perm_tr: a vector which contains the allowed trailers
2125  * @perm_tr_len: size of the perm_tr vector
2126  * @tr: pointer to store the trailer character
2127  *
2128  * In case of success %0 is returned and @buf and @size are updated with
2129  * the amount of bytes read. If @tr is non-NULL and a trailing
2130  * character exists (size is non-zero after returning from this
2131  * function), @tr is updated with the trailing character.
2132  */
2133 static int proc_get_long(char **buf, size_t *size,
2134 			  unsigned long *val, bool *neg,
2135 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
2136 {
2137 	int len;
2138 	char *p, tmp[TMPBUFLEN];
2139 
2140 	if (!*size)
2141 		return -EINVAL;
2142 
2143 	len = *size;
2144 	if (len > TMPBUFLEN - 1)
2145 		len = TMPBUFLEN - 1;
2146 
2147 	memcpy(tmp, *buf, len);
2148 
2149 	tmp[len] = 0;
2150 	p = tmp;
2151 	if (*p == '-' && *size > 1) {
2152 		*neg = true;
2153 		p++;
2154 	} else
2155 		*neg = false;
2156 	if (!isdigit(*p))
2157 		return -EINVAL;
2158 
2159 	*val = simple_strtoul(p, &p, 0);
2160 
2161 	len = p - tmp;
2162 
2163 	/* We don't know if the next char is whitespace thus we may accept
2164 	 * invalid integers (e.g. 1234...a) or two integers instead of one
2165 	 * (e.g. 123...1). So lets not allow such large numbers. */
2166 	if (len == TMPBUFLEN - 1)
2167 		return -EINVAL;
2168 
2169 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2170 		return -EINVAL;
2171 
2172 	if (tr && (len < *size))
2173 		*tr = *p;
2174 
2175 	*buf += len;
2176 	*size -= len;
2177 
2178 	return 0;
2179 }
2180 
2181 /**
2182  * proc_put_long - converts an integer to a decimal ASCII formatted string
2183  *
2184  * @buf: the user buffer
2185  * @size: the size of the user buffer
2186  * @val: the integer to be converted
2187  * @neg: sign of the number, %TRUE for negative
2188  *
2189  * In case of success %0 is returned and @buf and @size are updated with
2190  * the amount of bytes written.
2191  */
2192 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2193 			  bool neg)
2194 {
2195 	int len;
2196 	char tmp[TMPBUFLEN], *p = tmp;
2197 
2198 	sprintf(p, "%s%lu", neg ? "-" : "", val);
2199 	len = strlen(tmp);
2200 	if (len > *size)
2201 		len = *size;
2202 	if (copy_to_user(*buf, tmp, len))
2203 		return -EFAULT;
2204 	*size -= len;
2205 	*buf += len;
2206 	return 0;
2207 }
2208 #undef TMPBUFLEN
2209 
2210 static int proc_put_char(void __user **buf, size_t *size, char c)
2211 {
2212 	if (*size) {
2213 		char __user **buffer = (char __user **)buf;
2214 		if (put_user(c, *buffer))
2215 			return -EFAULT;
2216 		(*size)--, (*buffer)++;
2217 		*buf = *buffer;
2218 	}
2219 	return 0;
2220 }
2221 
2222 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2223 				 int *valp,
2224 				 int write, void *data)
2225 {
2226 	if (write) {
2227 		*valp = *negp ? -*lvalp : *lvalp;
2228 	} else {
2229 		int val = *valp;
2230 		if (val < 0) {
2231 			*negp = true;
2232 			*lvalp = (unsigned long)-val;
2233 		} else {
2234 			*negp = false;
2235 			*lvalp = (unsigned long)val;
2236 		}
2237 	}
2238 	return 0;
2239 }
2240 
2241 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2242 
2243 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2244 		  int write, void __user *buffer,
2245 		  size_t *lenp, loff_t *ppos,
2246 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2247 			      int write, void *data),
2248 		  void *data)
2249 {
2250 	int *i, vleft, first = 1, err = 0;
2251 	unsigned long page = 0;
2252 	size_t left;
2253 	char *kbuf;
2254 
2255 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2256 		*lenp = 0;
2257 		return 0;
2258 	}
2259 
2260 	i = (int *) tbl_data;
2261 	vleft = table->maxlen / sizeof(*i);
2262 	left = *lenp;
2263 
2264 	if (!conv)
2265 		conv = do_proc_dointvec_conv;
2266 
2267 	if (write) {
2268 		if (left > PAGE_SIZE - 1)
2269 			left = PAGE_SIZE - 1;
2270 		page = __get_free_page(GFP_TEMPORARY);
2271 		kbuf = (char *) page;
2272 		if (!kbuf)
2273 			return -ENOMEM;
2274 		if (copy_from_user(kbuf, buffer, left)) {
2275 			err = -EFAULT;
2276 			goto free;
2277 		}
2278 		kbuf[left] = 0;
2279 	}
2280 
2281 	for (; left && vleft--; i++, first=0) {
2282 		unsigned long lval;
2283 		bool neg;
2284 
2285 		if (write) {
2286 			left -= proc_skip_spaces(&kbuf);
2287 
2288 			if (!left)
2289 				break;
2290 			err = proc_get_long(&kbuf, &left, &lval, &neg,
2291 					     proc_wspace_sep,
2292 					     sizeof(proc_wspace_sep), NULL);
2293 			if (err)
2294 				break;
2295 			if (conv(&neg, &lval, i, 1, data)) {
2296 				err = -EINVAL;
2297 				break;
2298 			}
2299 		} else {
2300 			if (conv(&neg, &lval, i, 0, data)) {
2301 				err = -EINVAL;
2302 				break;
2303 			}
2304 			if (!first)
2305 				err = proc_put_char(&buffer, &left, '\t');
2306 			if (err)
2307 				break;
2308 			err = proc_put_long(&buffer, &left, lval, neg);
2309 			if (err)
2310 				break;
2311 		}
2312 	}
2313 
2314 	if (!write && !first && left && !err)
2315 		err = proc_put_char(&buffer, &left, '\n');
2316 	if (write && !err && left)
2317 		left -= proc_skip_spaces(&kbuf);
2318 free:
2319 	if (write) {
2320 		free_page(page);
2321 		if (first)
2322 			return err ? : -EINVAL;
2323 	}
2324 	*lenp -= left;
2325 	*ppos += *lenp;
2326 	return err;
2327 }
2328 
2329 static int do_proc_dointvec(struct ctl_table *table, int write,
2330 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2331 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2332 			      int write, void *data),
2333 		  void *data)
2334 {
2335 	return __do_proc_dointvec(table->data, table, write,
2336 			buffer, lenp, ppos, conv, data);
2337 }
2338 
2339 /**
2340  * proc_dointvec - read a vector of integers
2341  * @table: the sysctl table
2342  * @write: %TRUE if this is a write to the sysctl file
2343  * @buffer: the user buffer
2344  * @lenp: the size of the user buffer
2345  * @ppos: file position
2346  *
2347  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2348  * values from/to the user buffer, treated as an ASCII string.
2349  *
2350  * Returns 0 on success.
2351  */
2352 int proc_dointvec(struct ctl_table *table, int write,
2353 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2354 {
2355     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2356 		    	    NULL,NULL);
2357 }
2358 
2359 /*
2360  * Taint values can only be increased
2361  * This means we can safely use a temporary.
2362  */
2363 static int proc_taint(struct ctl_table *table, int write,
2364 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2365 {
2366 	struct ctl_table t;
2367 	unsigned long tmptaint = get_taint();
2368 	int err;
2369 
2370 	if (write && !capable(CAP_SYS_ADMIN))
2371 		return -EPERM;
2372 
2373 	t = *table;
2374 	t.data = &tmptaint;
2375 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2376 	if (err < 0)
2377 		return err;
2378 
2379 	if (write) {
2380 		/*
2381 		 * Poor man's atomic or. Not worth adding a primitive
2382 		 * to everyone's atomic.h for this
2383 		 */
2384 		int i;
2385 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2386 			if ((tmptaint >> i) & 1)
2387 				add_taint(i);
2388 		}
2389 	}
2390 
2391 	return err;
2392 }
2393 
2394 struct do_proc_dointvec_minmax_conv_param {
2395 	int *min;
2396 	int *max;
2397 };
2398 
2399 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2400 					int *valp,
2401 					int write, void *data)
2402 {
2403 	struct do_proc_dointvec_minmax_conv_param *param = data;
2404 	if (write) {
2405 		int val = *negp ? -*lvalp : *lvalp;
2406 		if ((param->min && *param->min > val) ||
2407 		    (param->max && *param->max < val))
2408 			return -EINVAL;
2409 		*valp = val;
2410 	} else {
2411 		int val = *valp;
2412 		if (val < 0) {
2413 			*negp = true;
2414 			*lvalp = (unsigned long)-val;
2415 		} else {
2416 			*negp = false;
2417 			*lvalp = (unsigned long)val;
2418 		}
2419 	}
2420 	return 0;
2421 }
2422 
2423 /**
2424  * proc_dointvec_minmax - read a vector of integers with min/max values
2425  * @table: the sysctl table
2426  * @write: %TRUE if this is a write to the sysctl file
2427  * @buffer: the user buffer
2428  * @lenp: the size of the user buffer
2429  * @ppos: file position
2430  *
2431  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2432  * values from/to the user buffer, treated as an ASCII string.
2433  *
2434  * This routine will ensure the values are within the range specified by
2435  * table->extra1 (min) and table->extra2 (max).
2436  *
2437  * Returns 0 on success.
2438  */
2439 int proc_dointvec_minmax(struct ctl_table *table, int write,
2440 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2441 {
2442 	struct do_proc_dointvec_minmax_conv_param param = {
2443 		.min = (int *) table->extra1,
2444 		.max = (int *) table->extra2,
2445 	};
2446 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2447 				do_proc_dointvec_minmax_conv, &param);
2448 }
2449 
2450 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2451 				     void __user *buffer,
2452 				     size_t *lenp, loff_t *ppos,
2453 				     unsigned long convmul,
2454 				     unsigned long convdiv)
2455 {
2456 	unsigned long *i, *min, *max;
2457 	int vleft, first = 1, err = 0;
2458 	unsigned long page = 0;
2459 	size_t left;
2460 	char *kbuf;
2461 
2462 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2463 		*lenp = 0;
2464 		return 0;
2465 	}
2466 
2467 	i = (unsigned long *) data;
2468 	min = (unsigned long *) table->extra1;
2469 	max = (unsigned long *) table->extra2;
2470 	vleft = table->maxlen / sizeof(unsigned long);
2471 	left = *lenp;
2472 
2473 	if (write) {
2474 		if (left > PAGE_SIZE - 1)
2475 			left = PAGE_SIZE - 1;
2476 		page = __get_free_page(GFP_TEMPORARY);
2477 		kbuf = (char *) page;
2478 		if (!kbuf)
2479 			return -ENOMEM;
2480 		if (copy_from_user(kbuf, buffer, left)) {
2481 			err = -EFAULT;
2482 			goto free;
2483 		}
2484 		kbuf[left] = 0;
2485 	}
2486 
2487 	for (; left && vleft--; i++, first = 0) {
2488 		unsigned long val;
2489 
2490 		if (write) {
2491 			bool neg;
2492 
2493 			left -= proc_skip_spaces(&kbuf);
2494 
2495 			err = proc_get_long(&kbuf, &left, &val, &neg,
2496 					     proc_wspace_sep,
2497 					     sizeof(proc_wspace_sep), NULL);
2498 			if (err)
2499 				break;
2500 			if (neg)
2501 				continue;
2502 			if ((min && val < *min) || (max && val > *max))
2503 				continue;
2504 			*i = val;
2505 		} else {
2506 			val = convdiv * (*i) / convmul;
2507 			if (!first)
2508 				err = proc_put_char(&buffer, &left, '\t');
2509 			err = proc_put_long(&buffer, &left, val, false);
2510 			if (err)
2511 				break;
2512 		}
2513 	}
2514 
2515 	if (!write && !first && left && !err)
2516 		err = proc_put_char(&buffer, &left, '\n');
2517 	if (write && !err)
2518 		left -= proc_skip_spaces(&kbuf);
2519 free:
2520 	if (write) {
2521 		free_page(page);
2522 		if (first)
2523 			return err ? : -EINVAL;
2524 	}
2525 	*lenp -= left;
2526 	*ppos += *lenp;
2527 	return err;
2528 }
2529 
2530 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2531 				     void __user *buffer,
2532 				     size_t *lenp, loff_t *ppos,
2533 				     unsigned long convmul,
2534 				     unsigned long convdiv)
2535 {
2536 	return __do_proc_doulongvec_minmax(table->data, table, write,
2537 			buffer, lenp, ppos, convmul, convdiv);
2538 }
2539 
2540 /**
2541  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2542  * @table: the sysctl table
2543  * @write: %TRUE if this is a write to the sysctl file
2544  * @buffer: the user buffer
2545  * @lenp: the size of the user buffer
2546  * @ppos: file position
2547  *
2548  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2549  * values from/to the user buffer, treated as an ASCII string.
2550  *
2551  * This routine will ensure the values are within the range specified by
2552  * table->extra1 (min) and table->extra2 (max).
2553  *
2554  * Returns 0 on success.
2555  */
2556 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2557 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2558 {
2559     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2560 }
2561 
2562 /**
2563  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2564  * @table: the sysctl table
2565  * @write: %TRUE if this is a write to the sysctl file
2566  * @buffer: the user buffer
2567  * @lenp: the size of the user buffer
2568  * @ppos: file position
2569  *
2570  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2571  * values from/to the user buffer, treated as an ASCII string. The values
2572  * are treated as milliseconds, and converted to jiffies when they are stored.
2573  *
2574  * This routine will ensure the values are within the range specified by
2575  * table->extra1 (min) and table->extra2 (max).
2576  *
2577  * Returns 0 on success.
2578  */
2579 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2580 				      void __user *buffer,
2581 				      size_t *lenp, loff_t *ppos)
2582 {
2583     return do_proc_doulongvec_minmax(table, write, buffer,
2584 				     lenp, ppos, HZ, 1000l);
2585 }
2586 
2587 
2588 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2589 					 int *valp,
2590 					 int write, void *data)
2591 {
2592 	if (write) {
2593 		if (*lvalp > LONG_MAX / HZ)
2594 			return 1;
2595 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2596 	} else {
2597 		int val = *valp;
2598 		unsigned long lval;
2599 		if (val < 0) {
2600 			*negp = true;
2601 			lval = (unsigned long)-val;
2602 		} else {
2603 			*negp = false;
2604 			lval = (unsigned long)val;
2605 		}
2606 		*lvalp = lval / HZ;
2607 	}
2608 	return 0;
2609 }
2610 
2611 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2612 						int *valp,
2613 						int write, void *data)
2614 {
2615 	if (write) {
2616 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2617 			return 1;
2618 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2619 	} else {
2620 		int val = *valp;
2621 		unsigned long lval;
2622 		if (val < 0) {
2623 			*negp = true;
2624 			lval = (unsigned long)-val;
2625 		} else {
2626 			*negp = false;
2627 			lval = (unsigned long)val;
2628 		}
2629 		*lvalp = jiffies_to_clock_t(lval);
2630 	}
2631 	return 0;
2632 }
2633 
2634 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2635 					    int *valp,
2636 					    int write, void *data)
2637 {
2638 	if (write) {
2639 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2640 	} else {
2641 		int val = *valp;
2642 		unsigned long lval;
2643 		if (val < 0) {
2644 			*negp = true;
2645 			lval = (unsigned long)-val;
2646 		} else {
2647 			*negp = false;
2648 			lval = (unsigned long)val;
2649 		}
2650 		*lvalp = jiffies_to_msecs(lval);
2651 	}
2652 	return 0;
2653 }
2654 
2655 /**
2656  * proc_dointvec_jiffies - read a vector of integers as seconds
2657  * @table: the sysctl table
2658  * @write: %TRUE if this is a write to the sysctl file
2659  * @buffer: the user buffer
2660  * @lenp: the size of the user buffer
2661  * @ppos: file position
2662  *
2663  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2664  * values from/to the user buffer, treated as an ASCII string.
2665  * The values read are assumed to be in seconds, and are converted into
2666  * jiffies.
2667  *
2668  * Returns 0 on success.
2669  */
2670 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2671 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2672 {
2673     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2674 		    	    do_proc_dointvec_jiffies_conv,NULL);
2675 }
2676 
2677 /**
2678  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2679  * @table: the sysctl table
2680  * @write: %TRUE if this is a write to the sysctl file
2681  * @buffer: the user buffer
2682  * @lenp: the size of the user buffer
2683  * @ppos: pointer to the file position
2684  *
2685  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2686  * values from/to the user buffer, treated as an ASCII string.
2687  * The values read are assumed to be in 1/USER_HZ seconds, and
2688  * are converted into jiffies.
2689  *
2690  * Returns 0 on success.
2691  */
2692 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2693 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2694 {
2695     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2696 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2697 }
2698 
2699 /**
2700  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2701  * @table: the sysctl table
2702  * @write: %TRUE if this is a write to the sysctl file
2703  * @buffer: the user buffer
2704  * @lenp: the size of the user buffer
2705  * @ppos: file position
2706  * @ppos: the current position in the file
2707  *
2708  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2709  * values from/to the user buffer, treated as an ASCII string.
2710  * The values read are assumed to be in 1/1000 seconds, and
2711  * are converted into jiffies.
2712  *
2713  * Returns 0 on success.
2714  */
2715 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2716 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2717 {
2718 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2719 				do_proc_dointvec_ms_jiffies_conv, NULL);
2720 }
2721 
2722 static int proc_do_cad_pid(struct ctl_table *table, int write,
2723 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2724 {
2725 	struct pid *new_pid;
2726 	pid_t tmp;
2727 	int r;
2728 
2729 	tmp = pid_vnr(cad_pid);
2730 
2731 	r = __do_proc_dointvec(&tmp, table, write, buffer,
2732 			       lenp, ppos, NULL, NULL);
2733 	if (r || !write)
2734 		return r;
2735 
2736 	new_pid = find_get_pid(tmp);
2737 	if (!new_pid)
2738 		return -ESRCH;
2739 
2740 	put_pid(xchg(&cad_pid, new_pid));
2741 	return 0;
2742 }
2743 
2744 /**
2745  * proc_do_large_bitmap - read/write from/to a large bitmap
2746  * @table: the sysctl table
2747  * @write: %TRUE if this is a write to the sysctl file
2748  * @buffer: the user buffer
2749  * @lenp: the size of the user buffer
2750  * @ppos: file position
2751  *
2752  * The bitmap is stored at table->data and the bitmap length (in bits)
2753  * in table->maxlen.
2754  *
2755  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2756  * large bitmaps may be represented in a compact manner. Writing into
2757  * the file will clear the bitmap then update it with the given input.
2758  *
2759  * Returns 0 on success.
2760  */
2761 int proc_do_large_bitmap(struct ctl_table *table, int write,
2762 			 void __user *buffer, size_t *lenp, loff_t *ppos)
2763 {
2764 	int err = 0;
2765 	bool first = 1;
2766 	size_t left = *lenp;
2767 	unsigned long bitmap_len = table->maxlen;
2768 	unsigned long *bitmap = (unsigned long *) table->data;
2769 	unsigned long *tmp_bitmap = NULL;
2770 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2771 
2772 	if (!bitmap_len || !left || (*ppos && !write)) {
2773 		*lenp = 0;
2774 		return 0;
2775 	}
2776 
2777 	if (write) {
2778 		unsigned long page = 0;
2779 		char *kbuf;
2780 
2781 		if (left > PAGE_SIZE - 1)
2782 			left = PAGE_SIZE - 1;
2783 
2784 		page = __get_free_page(GFP_TEMPORARY);
2785 		kbuf = (char *) page;
2786 		if (!kbuf)
2787 			return -ENOMEM;
2788 		if (copy_from_user(kbuf, buffer, left)) {
2789 			free_page(page);
2790 			return -EFAULT;
2791                 }
2792 		kbuf[left] = 0;
2793 
2794 		tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2795 				     GFP_KERNEL);
2796 		if (!tmp_bitmap) {
2797 			free_page(page);
2798 			return -ENOMEM;
2799 		}
2800 		proc_skip_char(&kbuf, &left, '\n');
2801 		while (!err && left) {
2802 			unsigned long val_a, val_b;
2803 			bool neg;
2804 
2805 			err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2806 					     sizeof(tr_a), &c);
2807 			if (err)
2808 				break;
2809 			if (val_a >= bitmap_len || neg) {
2810 				err = -EINVAL;
2811 				break;
2812 			}
2813 
2814 			val_b = val_a;
2815 			if (left) {
2816 				kbuf++;
2817 				left--;
2818 			}
2819 
2820 			if (c == '-') {
2821 				err = proc_get_long(&kbuf, &left, &val_b,
2822 						     &neg, tr_b, sizeof(tr_b),
2823 						     &c);
2824 				if (err)
2825 					break;
2826 				if (val_b >= bitmap_len || neg ||
2827 				    val_a > val_b) {
2828 					err = -EINVAL;
2829 					break;
2830 				}
2831 				if (left) {
2832 					kbuf++;
2833 					left--;
2834 				}
2835 			}
2836 
2837 			while (val_a <= val_b)
2838 				set_bit(val_a++, tmp_bitmap);
2839 
2840 			first = 0;
2841 			proc_skip_char(&kbuf, &left, '\n');
2842 		}
2843 		free_page(page);
2844 	} else {
2845 		unsigned long bit_a, bit_b = 0;
2846 
2847 		while (left) {
2848 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2849 			if (bit_a >= bitmap_len)
2850 				break;
2851 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
2852 						   bit_a + 1) - 1;
2853 
2854 			if (!first) {
2855 				err = proc_put_char(&buffer, &left, ',');
2856 				if (err)
2857 					break;
2858 			}
2859 			err = proc_put_long(&buffer, &left, bit_a, false);
2860 			if (err)
2861 				break;
2862 			if (bit_a != bit_b) {
2863 				err = proc_put_char(&buffer, &left, '-');
2864 				if (err)
2865 					break;
2866 				err = proc_put_long(&buffer, &left, bit_b, false);
2867 				if (err)
2868 					break;
2869 			}
2870 
2871 			first = 0; bit_b++;
2872 		}
2873 		if (!err)
2874 			err = proc_put_char(&buffer, &left, '\n');
2875 	}
2876 
2877 	if (!err) {
2878 		if (write) {
2879 			if (*ppos)
2880 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2881 			else
2882 				memcpy(bitmap, tmp_bitmap,
2883 					BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long));
2884 		}
2885 		kfree(tmp_bitmap);
2886 		*lenp -= left;
2887 		*ppos += *lenp;
2888 		return 0;
2889 	} else {
2890 		kfree(tmp_bitmap);
2891 		return err;
2892 	}
2893 }
2894 
2895 #else /* CONFIG_PROC_SYSCTL */
2896 
2897 int proc_dostring(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(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_minmax(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_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_userhz_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_dointvec_ms_jiffies(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_minmax(struct ctl_table *table, int write,
2934 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2935 {
2936 	return -ENOSYS;
2937 }
2938 
2939 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2940 				      void __user *buffer,
2941 				      size_t *lenp, loff_t *ppos)
2942 {
2943     return -ENOSYS;
2944 }
2945 
2946 
2947 #endif /* CONFIG_PROC_SYSCTL */
2948 
2949 /*
2950  * No sense putting this after each symbol definition, twice,
2951  * exception granted :-)
2952  */
2953 EXPORT_SYMBOL(proc_dointvec);
2954 EXPORT_SYMBOL(proc_dointvec_jiffies);
2955 EXPORT_SYMBOL(proc_dointvec_minmax);
2956 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2957 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2958 EXPORT_SYMBOL(proc_dostring);
2959 EXPORT_SYMBOL(proc_doulongvec_minmax);
2960 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2961 EXPORT_SYMBOL(register_sysctl_table);
2962 EXPORT_SYMBOL(register_sysctl_paths);
2963 EXPORT_SYMBOL(unregister_sysctl_table);
2964