xref: /linux-6.15/kernel/sysctl.c (revision 4dc7ccf7)
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/proc_fs.h>
28 #include <linux/security.h>
29 #include <linux/ctype.h>
30 #include <linux/kmemcheck.h>
31 #include <linux/fs.h>
32 #include <linux/init.h>
33 #include <linux/kernel.h>
34 #include <linux/kobject.h>
35 #include <linux/net.h>
36 #include <linux/sysrq.h>
37 #include <linux/highuid.h>
38 #include <linux/writeback.h>
39 #include <linux/ratelimit.h>
40 #include <linux/hugetlb.h>
41 #include <linux/initrd.h>
42 #include <linux/key.h>
43 #include <linux/times.h>
44 #include <linux/limits.h>
45 #include <linux/dcache.h>
46 #include <linux/syscalls.h>
47 #include <linux/vmstat.h>
48 #include <linux/nfs_fs.h>
49 #include <linux/acpi.h>
50 #include <linux/reboot.h>
51 #include <linux/ftrace.h>
52 #include <linux/slow-work.h>
53 #include <linux/perf_event.h>
54 #include <linux/kprobes.h>
55 
56 #include <asm/uaccess.h>
57 #include <asm/processor.h>
58 
59 #ifdef CONFIG_X86
60 #include <asm/nmi.h>
61 #include <asm/stacktrace.h>
62 #include <asm/io.h>
63 #endif
64 #ifdef CONFIG_BSD_PROCESS_ACCT
65 #include <linux/acct.h>
66 #endif
67 #ifdef CONFIG_RT_MUTEXES
68 #include <linux/rtmutex.h>
69 #endif
70 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
71 #include <linux/lockdep.h>
72 #endif
73 #ifdef CONFIG_CHR_DEV_SG
74 #include <scsi/sg.h>
75 #endif
76 
77 
78 #if defined(CONFIG_SYSCTL)
79 
80 /* External variables not in a header file. */
81 extern int sysctl_overcommit_memory;
82 extern int sysctl_overcommit_ratio;
83 extern int sysctl_panic_on_oom;
84 extern int sysctl_oom_kill_allocating_task;
85 extern int sysctl_oom_dump_tasks;
86 extern int max_threads;
87 extern int core_uses_pid;
88 extern int suid_dumpable;
89 extern char core_pattern[];
90 extern unsigned int core_pipe_limit;
91 extern int pid_max;
92 extern int min_free_kbytes;
93 extern int pid_max_min, pid_max_max;
94 extern int sysctl_drop_caches;
95 extern int percpu_pagelist_fraction;
96 extern int compat_log;
97 extern int latencytop_enabled;
98 extern int sysctl_nr_open_min, sysctl_nr_open_max;
99 #ifndef CONFIG_MMU
100 extern int sysctl_nr_trim_pages;
101 #endif
102 #ifdef CONFIG_BLOCK
103 extern int blk_iopoll_enabled;
104 #endif
105 
106 /* Constants used for minimum and  maximum */
107 #ifdef CONFIG_DETECT_SOFTLOCKUP
108 static int sixty = 60;
109 static int neg_one = -1;
110 #endif
111 
112 static int zero;
113 static int __maybe_unused one = 1;
114 static int __maybe_unused two = 2;
115 static unsigned long one_ul = 1;
116 static int one_hundred = 100;
117 #ifdef CONFIG_PRINTK
118 static int ten_thousand = 10000;
119 #endif
120 
121 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
122 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
123 
124 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
125 static int maxolduid = 65535;
126 static int minolduid;
127 static int min_percpu_pagelist_fract = 8;
128 
129 static int ngroups_max = NGROUPS_MAX;
130 
131 #ifdef CONFIG_SPARC
132 #include <asm/system.h>
133 #endif
134 
135 #ifdef CONFIG_SPARC64
136 extern int sysctl_tsb_ratio;
137 #endif
138 
139 #ifdef __hppa__
140 extern int pwrsw_enabled;
141 extern int unaligned_enabled;
142 #endif
143 
144 #ifdef CONFIG_S390
145 #ifdef CONFIG_MATHEMU
146 extern int sysctl_ieee_emulation_warnings;
147 #endif
148 extern int sysctl_userprocess_debug;
149 extern int spin_retry;
150 #endif
151 
152 #ifdef CONFIG_IA64
153 extern int no_unaligned_warning;
154 extern int unaligned_dump_stack;
155 #endif
156 
157 extern struct ratelimit_state printk_ratelimit_state;
158 
159 #ifdef CONFIG_PROC_SYSCTL
160 static int proc_do_cad_pid(struct ctl_table *table, int write,
161 		  void __user *buffer, size_t *lenp, loff_t *ppos);
162 static int proc_taint(struct ctl_table *table, int write,
163 			       void __user *buffer, size_t *lenp, loff_t *ppos);
164 #endif
165 
166 static struct ctl_table root_table[];
167 static struct ctl_table_root sysctl_table_root;
168 static struct ctl_table_header root_table_header = {
169 	.count = 1,
170 	.ctl_table = root_table,
171 	.ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
172 	.root = &sysctl_table_root,
173 	.set = &sysctl_table_root.default_set,
174 };
175 static struct ctl_table_root sysctl_table_root = {
176 	.root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
177 	.default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
178 };
179 
180 static struct ctl_table kern_table[];
181 static struct ctl_table vm_table[];
182 static struct ctl_table fs_table[];
183 static struct ctl_table debug_table[];
184 static struct ctl_table dev_table[];
185 extern struct ctl_table random_table[];
186 #ifdef CONFIG_INOTIFY_USER
187 extern struct ctl_table inotify_table[];
188 #endif
189 #ifdef CONFIG_EPOLL
190 extern struct ctl_table epoll_table[];
191 #endif
192 
193 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
194 int sysctl_legacy_va_layout;
195 #endif
196 
197 /* The default sysctl tables: */
198 
199 static struct ctl_table root_table[] = {
200 	{
201 		.procname	= "kernel",
202 		.mode		= 0555,
203 		.child		= kern_table,
204 	},
205 	{
206 		.procname	= "vm",
207 		.mode		= 0555,
208 		.child		= vm_table,
209 	},
210 	{
211 		.procname	= "fs",
212 		.mode		= 0555,
213 		.child		= fs_table,
214 	},
215 	{
216 		.procname	= "debug",
217 		.mode		= 0555,
218 		.child		= debug_table,
219 	},
220 	{
221 		.procname	= "dev",
222 		.mode		= 0555,
223 		.child		= dev_table,
224 	},
225 /*
226  * NOTE: do not add new entries to this table unless you have read
227  * Documentation/sysctl/ctl_unnumbered.txt
228  */
229 	{ }
230 };
231 
232 #ifdef CONFIG_SCHED_DEBUG
233 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
234 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
235 static int min_wakeup_granularity_ns;			/* 0 usecs */
236 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
237 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
238 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
239 static int min_sched_shares_ratelimit = 100000; /* 100 usec */
240 static int max_sched_shares_ratelimit = NSEC_PER_SEC; /* 1 second */
241 #endif
242 
243 static struct ctl_table kern_table[] = {
244 	{
245 		.procname	= "sched_child_runs_first",
246 		.data		= &sysctl_sched_child_runs_first,
247 		.maxlen		= sizeof(unsigned int),
248 		.mode		= 0644,
249 		.proc_handler	= proc_dointvec,
250 	},
251 #ifdef CONFIG_SCHED_DEBUG
252 	{
253 		.procname	= "sched_min_granularity_ns",
254 		.data		= &sysctl_sched_min_granularity,
255 		.maxlen		= sizeof(unsigned int),
256 		.mode		= 0644,
257 		.proc_handler	= sched_proc_update_handler,
258 		.extra1		= &min_sched_granularity_ns,
259 		.extra2		= &max_sched_granularity_ns,
260 	},
261 	{
262 		.procname	= "sched_latency_ns",
263 		.data		= &sysctl_sched_latency,
264 		.maxlen		= sizeof(unsigned int),
265 		.mode		= 0644,
266 		.proc_handler	= sched_proc_update_handler,
267 		.extra1		= &min_sched_granularity_ns,
268 		.extra2		= &max_sched_granularity_ns,
269 	},
270 	{
271 		.procname	= "sched_wakeup_granularity_ns",
272 		.data		= &sysctl_sched_wakeup_granularity,
273 		.maxlen		= sizeof(unsigned int),
274 		.mode		= 0644,
275 		.proc_handler	= sched_proc_update_handler,
276 		.extra1		= &min_wakeup_granularity_ns,
277 		.extra2		= &max_wakeup_granularity_ns,
278 	},
279 	{
280 		.procname	= "sched_shares_ratelimit",
281 		.data		= &sysctl_sched_shares_ratelimit,
282 		.maxlen		= sizeof(unsigned int),
283 		.mode		= 0644,
284 		.proc_handler	= sched_proc_update_handler,
285 		.extra1		= &min_sched_shares_ratelimit,
286 		.extra2		= &max_sched_shares_ratelimit,
287 	},
288 	{
289 		.procname	= "sched_tunable_scaling",
290 		.data		= &sysctl_sched_tunable_scaling,
291 		.maxlen		= sizeof(enum sched_tunable_scaling),
292 		.mode		= 0644,
293 		.proc_handler	= sched_proc_update_handler,
294 		.extra1		= &min_sched_tunable_scaling,
295 		.extra2		= &max_sched_tunable_scaling,
296 	},
297 	{
298 		.procname	= "sched_shares_thresh",
299 		.data		= &sysctl_sched_shares_thresh,
300 		.maxlen		= sizeof(unsigned int),
301 		.mode		= 0644,
302 		.proc_handler	= proc_dointvec_minmax,
303 		.extra1		= &zero,
304 	},
305 	{
306 		.procname	= "sched_migration_cost",
307 		.data		= &sysctl_sched_migration_cost,
308 		.maxlen		= sizeof(unsigned int),
309 		.mode		= 0644,
310 		.proc_handler	= proc_dointvec,
311 	},
312 	{
313 		.procname	= "sched_nr_migrate",
314 		.data		= &sysctl_sched_nr_migrate,
315 		.maxlen		= sizeof(unsigned int),
316 		.mode		= 0644,
317 		.proc_handler	= proc_dointvec,
318 	},
319 	{
320 		.procname	= "sched_time_avg",
321 		.data		= &sysctl_sched_time_avg,
322 		.maxlen		= sizeof(unsigned int),
323 		.mode		= 0644,
324 		.proc_handler	= proc_dointvec,
325 	},
326 	{
327 		.procname	= "timer_migration",
328 		.data		= &sysctl_timer_migration,
329 		.maxlen		= sizeof(unsigned int),
330 		.mode		= 0644,
331 		.proc_handler	= proc_dointvec_minmax,
332 		.extra1		= &zero,
333 		.extra2		= &one,
334 	},
335 #endif
336 	{
337 		.procname	= "sched_rt_period_us",
338 		.data		= &sysctl_sched_rt_period,
339 		.maxlen		= sizeof(unsigned int),
340 		.mode		= 0644,
341 		.proc_handler	= sched_rt_handler,
342 	},
343 	{
344 		.procname	= "sched_rt_runtime_us",
345 		.data		= &sysctl_sched_rt_runtime,
346 		.maxlen		= sizeof(int),
347 		.mode		= 0644,
348 		.proc_handler	= sched_rt_handler,
349 	},
350 	{
351 		.procname	= "sched_compat_yield",
352 		.data		= &sysctl_sched_compat_yield,
353 		.maxlen		= sizeof(unsigned int),
354 		.mode		= 0644,
355 		.proc_handler	= proc_dointvec,
356 	},
357 #ifdef CONFIG_PROVE_LOCKING
358 	{
359 		.procname	= "prove_locking",
360 		.data		= &prove_locking,
361 		.maxlen		= sizeof(int),
362 		.mode		= 0644,
363 		.proc_handler	= proc_dointvec,
364 	},
365 #endif
366 #ifdef CONFIG_LOCK_STAT
367 	{
368 		.procname	= "lock_stat",
369 		.data		= &lock_stat,
370 		.maxlen		= sizeof(int),
371 		.mode		= 0644,
372 		.proc_handler	= proc_dointvec,
373 	},
374 #endif
375 	{
376 		.procname	= "panic",
377 		.data		= &panic_timeout,
378 		.maxlen		= sizeof(int),
379 		.mode		= 0644,
380 		.proc_handler	= proc_dointvec,
381 	},
382 	{
383 		.procname	= "core_uses_pid",
384 		.data		= &core_uses_pid,
385 		.maxlen		= sizeof(int),
386 		.mode		= 0644,
387 		.proc_handler	= proc_dointvec,
388 	},
389 	{
390 		.procname	= "core_pattern",
391 		.data		= core_pattern,
392 		.maxlen		= CORENAME_MAX_SIZE,
393 		.mode		= 0644,
394 		.proc_handler	= proc_dostring,
395 	},
396 	{
397 		.procname	= "core_pipe_limit",
398 		.data		= &core_pipe_limit,
399 		.maxlen		= sizeof(unsigned int),
400 		.mode		= 0644,
401 		.proc_handler	= proc_dointvec,
402 	},
403 #ifdef CONFIG_PROC_SYSCTL
404 	{
405 		.procname	= "tainted",
406 		.maxlen 	= sizeof(long),
407 		.mode		= 0644,
408 		.proc_handler	= proc_taint,
409 	},
410 #endif
411 #ifdef CONFIG_LATENCYTOP
412 	{
413 		.procname	= "latencytop",
414 		.data		= &latencytop_enabled,
415 		.maxlen		= sizeof(int),
416 		.mode		= 0644,
417 		.proc_handler	= proc_dointvec,
418 	},
419 #endif
420 #ifdef CONFIG_BLK_DEV_INITRD
421 	{
422 		.procname	= "real-root-dev",
423 		.data		= &real_root_dev,
424 		.maxlen		= sizeof(int),
425 		.mode		= 0644,
426 		.proc_handler	= proc_dointvec,
427 	},
428 #endif
429 	{
430 		.procname	= "print-fatal-signals",
431 		.data		= &print_fatal_signals,
432 		.maxlen		= sizeof(int),
433 		.mode		= 0644,
434 		.proc_handler	= proc_dointvec,
435 	},
436 #ifdef CONFIG_SPARC
437 	{
438 		.procname	= "reboot-cmd",
439 		.data		= reboot_command,
440 		.maxlen		= 256,
441 		.mode		= 0644,
442 		.proc_handler	= proc_dostring,
443 	},
444 	{
445 		.procname	= "stop-a",
446 		.data		= &stop_a_enabled,
447 		.maxlen		= sizeof (int),
448 		.mode		= 0644,
449 		.proc_handler	= proc_dointvec,
450 	},
451 	{
452 		.procname	= "scons-poweroff",
453 		.data		= &scons_pwroff,
454 		.maxlen		= sizeof (int),
455 		.mode		= 0644,
456 		.proc_handler	= proc_dointvec,
457 	},
458 #endif
459 #ifdef CONFIG_SPARC64
460 	{
461 		.procname	= "tsb-ratio",
462 		.data		= &sysctl_tsb_ratio,
463 		.maxlen		= sizeof (int),
464 		.mode		= 0644,
465 		.proc_handler	= proc_dointvec,
466 	},
467 #endif
468 #ifdef __hppa__
469 	{
470 		.procname	= "soft-power",
471 		.data		= &pwrsw_enabled,
472 		.maxlen		= sizeof (int),
473 	 	.mode		= 0644,
474 		.proc_handler	= proc_dointvec,
475 	},
476 	{
477 		.procname	= "unaligned-trap",
478 		.data		= &unaligned_enabled,
479 		.maxlen		= sizeof (int),
480 		.mode		= 0644,
481 		.proc_handler	= proc_dointvec,
482 	},
483 #endif
484 	{
485 		.procname	= "ctrl-alt-del",
486 		.data		= &C_A_D,
487 		.maxlen		= sizeof(int),
488 		.mode		= 0644,
489 		.proc_handler	= proc_dointvec,
490 	},
491 #ifdef CONFIG_FUNCTION_TRACER
492 	{
493 		.procname	= "ftrace_enabled",
494 		.data		= &ftrace_enabled,
495 		.maxlen		= sizeof(int),
496 		.mode		= 0644,
497 		.proc_handler	= ftrace_enable_sysctl,
498 	},
499 #endif
500 #ifdef CONFIG_STACK_TRACER
501 	{
502 		.procname	= "stack_tracer_enabled",
503 		.data		= &stack_tracer_enabled,
504 		.maxlen		= sizeof(int),
505 		.mode		= 0644,
506 		.proc_handler	= stack_trace_sysctl,
507 	},
508 #endif
509 #ifdef CONFIG_TRACING
510 	{
511 		.procname	= "ftrace_dump_on_oops",
512 		.data		= &ftrace_dump_on_oops,
513 		.maxlen		= sizeof(int),
514 		.mode		= 0644,
515 		.proc_handler	= proc_dointvec,
516 	},
517 #endif
518 #ifdef CONFIG_MODULES
519 	{
520 		.procname	= "modprobe",
521 		.data		= &modprobe_path,
522 		.maxlen		= KMOD_PATH_LEN,
523 		.mode		= 0644,
524 		.proc_handler	= proc_dostring,
525 	},
526 	{
527 		.procname	= "modules_disabled",
528 		.data		= &modules_disabled,
529 		.maxlen		= sizeof(int),
530 		.mode		= 0644,
531 		/* only handle a transition from default "0" to "1" */
532 		.proc_handler	= proc_dointvec_minmax,
533 		.extra1		= &one,
534 		.extra2		= &one,
535 	},
536 #endif
537 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
538 	{
539 		.procname	= "hotplug",
540 		.data		= &uevent_helper,
541 		.maxlen		= UEVENT_HELPER_PATH_LEN,
542 		.mode		= 0644,
543 		.proc_handler	= proc_dostring,
544 	},
545 #endif
546 #ifdef CONFIG_CHR_DEV_SG
547 	{
548 		.procname	= "sg-big-buff",
549 		.data		= &sg_big_buff,
550 		.maxlen		= sizeof (int),
551 		.mode		= 0444,
552 		.proc_handler	= proc_dointvec,
553 	},
554 #endif
555 #ifdef CONFIG_BSD_PROCESS_ACCT
556 	{
557 		.procname	= "acct",
558 		.data		= &acct_parm,
559 		.maxlen		= 3*sizeof(int),
560 		.mode		= 0644,
561 		.proc_handler	= proc_dointvec,
562 	},
563 #endif
564 #ifdef CONFIG_MAGIC_SYSRQ
565 	{
566 		.procname	= "sysrq",
567 		.data		= &__sysrq_enabled,
568 		.maxlen		= sizeof (int),
569 		.mode		= 0644,
570 		.proc_handler	= proc_dointvec,
571 	},
572 #endif
573 #ifdef CONFIG_PROC_SYSCTL
574 	{
575 		.procname	= "cad_pid",
576 		.data		= NULL,
577 		.maxlen		= sizeof (int),
578 		.mode		= 0600,
579 		.proc_handler	= proc_do_cad_pid,
580 	},
581 #endif
582 	{
583 		.procname	= "threads-max",
584 		.data		= &max_threads,
585 		.maxlen		= sizeof(int),
586 		.mode		= 0644,
587 		.proc_handler	= proc_dointvec,
588 	},
589 	{
590 		.procname	= "random",
591 		.mode		= 0555,
592 		.child		= random_table,
593 	},
594 	{
595 		.procname	= "overflowuid",
596 		.data		= &overflowuid,
597 		.maxlen		= sizeof(int),
598 		.mode		= 0644,
599 		.proc_handler	= proc_dointvec_minmax,
600 		.extra1		= &minolduid,
601 		.extra2		= &maxolduid,
602 	},
603 	{
604 		.procname	= "overflowgid",
605 		.data		= &overflowgid,
606 		.maxlen		= sizeof(int),
607 		.mode		= 0644,
608 		.proc_handler	= proc_dointvec_minmax,
609 		.extra1		= &minolduid,
610 		.extra2		= &maxolduid,
611 	},
612 #ifdef CONFIG_S390
613 #ifdef CONFIG_MATHEMU
614 	{
615 		.procname	= "ieee_emulation_warnings",
616 		.data		= &sysctl_ieee_emulation_warnings,
617 		.maxlen		= sizeof(int),
618 		.mode		= 0644,
619 		.proc_handler	= proc_dointvec,
620 	},
621 #endif
622 	{
623 		.procname	= "userprocess_debug",
624 		.data		= &sysctl_userprocess_debug,
625 		.maxlen		= sizeof(int),
626 		.mode		= 0644,
627 		.proc_handler	= proc_dointvec,
628 	},
629 #endif
630 	{
631 		.procname	= "pid_max",
632 		.data		= &pid_max,
633 		.maxlen		= sizeof (int),
634 		.mode		= 0644,
635 		.proc_handler	= proc_dointvec_minmax,
636 		.extra1		= &pid_max_min,
637 		.extra2		= &pid_max_max,
638 	},
639 	{
640 		.procname	= "panic_on_oops",
641 		.data		= &panic_on_oops,
642 		.maxlen		= sizeof(int),
643 		.mode		= 0644,
644 		.proc_handler	= proc_dointvec,
645 	},
646 #if defined CONFIG_PRINTK
647 	{
648 		.procname	= "printk",
649 		.data		= &console_loglevel,
650 		.maxlen		= 4*sizeof(int),
651 		.mode		= 0644,
652 		.proc_handler	= proc_dointvec,
653 	},
654 	{
655 		.procname	= "printk_ratelimit",
656 		.data		= &printk_ratelimit_state.interval,
657 		.maxlen		= sizeof(int),
658 		.mode		= 0644,
659 		.proc_handler	= proc_dointvec_jiffies,
660 	},
661 	{
662 		.procname	= "printk_ratelimit_burst",
663 		.data		= &printk_ratelimit_state.burst,
664 		.maxlen		= sizeof(int),
665 		.mode		= 0644,
666 		.proc_handler	= proc_dointvec,
667 	},
668 	{
669 		.procname	= "printk_delay",
670 		.data		= &printk_delay_msec,
671 		.maxlen		= sizeof(int),
672 		.mode		= 0644,
673 		.proc_handler	= proc_dointvec_minmax,
674 		.extra1		= &zero,
675 		.extra2		= &ten_thousand,
676 	},
677 #endif
678 	{
679 		.procname	= "ngroups_max",
680 		.data		= &ngroups_max,
681 		.maxlen		= sizeof (int),
682 		.mode		= 0444,
683 		.proc_handler	= proc_dointvec,
684 	},
685 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
686 	{
687 		.procname       = "unknown_nmi_panic",
688 		.data           = &unknown_nmi_panic,
689 		.maxlen         = sizeof (int),
690 		.mode           = 0644,
691 		.proc_handler   = proc_dointvec,
692 	},
693 	{
694 		.procname       = "nmi_watchdog",
695 		.data           = &nmi_watchdog_enabled,
696 		.maxlen         = sizeof (int),
697 		.mode           = 0644,
698 		.proc_handler   = proc_nmi_enabled,
699 	},
700 #endif
701 #if defined(CONFIG_X86)
702 	{
703 		.procname	= "panic_on_unrecovered_nmi",
704 		.data		= &panic_on_unrecovered_nmi,
705 		.maxlen		= sizeof(int),
706 		.mode		= 0644,
707 		.proc_handler	= proc_dointvec,
708 	},
709 	{
710 		.procname	= "panic_on_io_nmi",
711 		.data		= &panic_on_io_nmi,
712 		.maxlen		= sizeof(int),
713 		.mode		= 0644,
714 		.proc_handler	= proc_dointvec,
715 	},
716 	{
717 		.procname	= "bootloader_type",
718 		.data		= &bootloader_type,
719 		.maxlen		= sizeof (int),
720 		.mode		= 0444,
721 		.proc_handler	= proc_dointvec,
722 	},
723 	{
724 		.procname	= "bootloader_version",
725 		.data		= &bootloader_version,
726 		.maxlen		= sizeof (int),
727 		.mode		= 0444,
728 		.proc_handler	= proc_dointvec,
729 	},
730 	{
731 		.procname	= "kstack_depth_to_print",
732 		.data		= &kstack_depth_to_print,
733 		.maxlen		= sizeof(int),
734 		.mode		= 0644,
735 		.proc_handler	= proc_dointvec,
736 	},
737 	{
738 		.procname	= "io_delay_type",
739 		.data		= &io_delay_type,
740 		.maxlen		= sizeof(int),
741 		.mode		= 0644,
742 		.proc_handler	= proc_dointvec,
743 	},
744 #endif
745 #if defined(CONFIG_MMU)
746 	{
747 		.procname	= "randomize_va_space",
748 		.data		= &randomize_va_space,
749 		.maxlen		= sizeof(int),
750 		.mode		= 0644,
751 		.proc_handler	= proc_dointvec,
752 	},
753 #endif
754 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
755 	{
756 		.procname	= "spin_retry",
757 		.data		= &spin_retry,
758 		.maxlen		= sizeof (int),
759 		.mode		= 0644,
760 		.proc_handler	= proc_dointvec,
761 	},
762 #endif
763 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
764 	{
765 		.procname	= "acpi_video_flags",
766 		.data		= &acpi_realmode_flags,
767 		.maxlen		= sizeof (unsigned long),
768 		.mode		= 0644,
769 		.proc_handler	= proc_doulongvec_minmax,
770 	},
771 #endif
772 #ifdef CONFIG_IA64
773 	{
774 		.procname	= "ignore-unaligned-usertrap",
775 		.data		= &no_unaligned_warning,
776 		.maxlen		= sizeof (int),
777 	 	.mode		= 0644,
778 		.proc_handler	= proc_dointvec,
779 	},
780 	{
781 		.procname	= "unaligned-dump-stack",
782 		.data		= &unaligned_dump_stack,
783 		.maxlen		= sizeof (int),
784 		.mode		= 0644,
785 		.proc_handler	= proc_dointvec,
786 	},
787 #endif
788 #ifdef CONFIG_DETECT_SOFTLOCKUP
789 	{
790 		.procname	= "softlockup_panic",
791 		.data		= &softlockup_panic,
792 		.maxlen		= sizeof(int),
793 		.mode		= 0644,
794 		.proc_handler	= proc_dointvec_minmax,
795 		.extra1		= &zero,
796 		.extra2		= &one,
797 	},
798 	{
799 		.procname	= "softlockup_thresh",
800 		.data		= &softlockup_thresh,
801 		.maxlen		= sizeof(int),
802 		.mode		= 0644,
803 		.proc_handler	= proc_dosoftlockup_thresh,
804 		.extra1		= &neg_one,
805 		.extra2		= &sixty,
806 	},
807 #endif
808 #ifdef CONFIG_DETECT_HUNG_TASK
809 	{
810 		.procname	= "hung_task_panic",
811 		.data		= &sysctl_hung_task_panic,
812 		.maxlen		= sizeof(int),
813 		.mode		= 0644,
814 		.proc_handler	= proc_dointvec_minmax,
815 		.extra1		= &zero,
816 		.extra2		= &one,
817 	},
818 	{
819 		.procname	= "hung_task_check_count",
820 		.data		= &sysctl_hung_task_check_count,
821 		.maxlen		= sizeof(unsigned long),
822 		.mode		= 0644,
823 		.proc_handler	= proc_doulongvec_minmax,
824 	},
825 	{
826 		.procname	= "hung_task_timeout_secs",
827 		.data		= &sysctl_hung_task_timeout_secs,
828 		.maxlen		= sizeof(unsigned long),
829 		.mode		= 0644,
830 		.proc_handler	= proc_dohung_task_timeout_secs,
831 	},
832 	{
833 		.procname	= "hung_task_warnings",
834 		.data		= &sysctl_hung_task_warnings,
835 		.maxlen		= sizeof(unsigned long),
836 		.mode		= 0644,
837 		.proc_handler	= proc_doulongvec_minmax,
838 	},
839 #endif
840 #ifdef CONFIG_COMPAT
841 	{
842 		.procname	= "compat-log",
843 		.data		= &compat_log,
844 		.maxlen		= sizeof (int),
845 	 	.mode		= 0644,
846 		.proc_handler	= proc_dointvec,
847 	},
848 #endif
849 #ifdef CONFIG_RT_MUTEXES
850 	{
851 		.procname	= "max_lock_depth",
852 		.data		= &max_lock_depth,
853 		.maxlen		= sizeof(int),
854 		.mode		= 0644,
855 		.proc_handler	= proc_dointvec,
856 	},
857 #endif
858 	{
859 		.procname	= "poweroff_cmd",
860 		.data		= &poweroff_cmd,
861 		.maxlen		= POWEROFF_CMD_PATH_LEN,
862 		.mode		= 0644,
863 		.proc_handler	= proc_dostring,
864 	},
865 #ifdef CONFIG_KEYS
866 	{
867 		.procname	= "keys",
868 		.mode		= 0555,
869 		.child		= key_sysctls,
870 	},
871 #endif
872 #ifdef CONFIG_RCU_TORTURE_TEST
873 	{
874 		.procname       = "rcutorture_runnable",
875 		.data           = &rcutorture_runnable,
876 		.maxlen         = sizeof(int),
877 		.mode           = 0644,
878 		.proc_handler	= proc_dointvec,
879 	},
880 #endif
881 #ifdef CONFIG_SLOW_WORK
882 	{
883 		.procname	= "slow-work",
884 		.mode		= 0555,
885 		.child		= slow_work_sysctls,
886 	},
887 #endif
888 #ifdef CONFIG_PERF_EVENTS
889 	{
890 		.procname	= "perf_event_paranoid",
891 		.data		= &sysctl_perf_event_paranoid,
892 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
893 		.mode		= 0644,
894 		.proc_handler	= proc_dointvec,
895 	},
896 	{
897 		.procname	= "perf_event_mlock_kb",
898 		.data		= &sysctl_perf_event_mlock,
899 		.maxlen		= sizeof(sysctl_perf_event_mlock),
900 		.mode		= 0644,
901 		.proc_handler	= proc_dointvec,
902 	},
903 	{
904 		.procname	= "perf_event_max_sample_rate",
905 		.data		= &sysctl_perf_event_sample_rate,
906 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
907 		.mode		= 0644,
908 		.proc_handler	= proc_dointvec,
909 	},
910 #endif
911 #ifdef CONFIG_KMEMCHECK
912 	{
913 		.procname	= "kmemcheck",
914 		.data		= &kmemcheck_enabled,
915 		.maxlen		= sizeof(int),
916 		.mode		= 0644,
917 		.proc_handler	= proc_dointvec,
918 	},
919 #endif
920 #ifdef CONFIG_BLOCK
921 	{
922 		.procname	= "blk_iopoll",
923 		.data		= &blk_iopoll_enabled,
924 		.maxlen		= sizeof(int),
925 		.mode		= 0644,
926 		.proc_handler	= proc_dointvec,
927 	},
928 #endif
929 /*
930  * NOTE: do not add new entries to this table unless you have read
931  * Documentation/sysctl/ctl_unnumbered.txt
932  */
933 	{ }
934 };
935 
936 static struct ctl_table vm_table[] = {
937 	{
938 		.procname	= "overcommit_memory",
939 		.data		= &sysctl_overcommit_memory,
940 		.maxlen		= sizeof(sysctl_overcommit_memory),
941 		.mode		= 0644,
942 		.proc_handler	= proc_dointvec,
943 	},
944 	{
945 		.procname	= "panic_on_oom",
946 		.data		= &sysctl_panic_on_oom,
947 		.maxlen		= sizeof(sysctl_panic_on_oom),
948 		.mode		= 0644,
949 		.proc_handler	= proc_dointvec,
950 	},
951 	{
952 		.procname	= "oom_kill_allocating_task",
953 		.data		= &sysctl_oom_kill_allocating_task,
954 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
955 		.mode		= 0644,
956 		.proc_handler	= proc_dointvec,
957 	},
958 	{
959 		.procname	= "oom_dump_tasks",
960 		.data		= &sysctl_oom_dump_tasks,
961 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
962 		.mode		= 0644,
963 		.proc_handler	= proc_dointvec,
964 	},
965 	{
966 		.procname	= "overcommit_ratio",
967 		.data		= &sysctl_overcommit_ratio,
968 		.maxlen		= sizeof(sysctl_overcommit_ratio),
969 		.mode		= 0644,
970 		.proc_handler	= proc_dointvec,
971 	},
972 	{
973 		.procname	= "page-cluster",
974 		.data		= &page_cluster,
975 		.maxlen		= sizeof(int),
976 		.mode		= 0644,
977 		.proc_handler	= proc_dointvec,
978 	},
979 	{
980 		.procname	= "dirty_background_ratio",
981 		.data		= &dirty_background_ratio,
982 		.maxlen		= sizeof(dirty_background_ratio),
983 		.mode		= 0644,
984 		.proc_handler	= dirty_background_ratio_handler,
985 		.extra1		= &zero,
986 		.extra2		= &one_hundred,
987 	},
988 	{
989 		.procname	= "dirty_background_bytes",
990 		.data		= &dirty_background_bytes,
991 		.maxlen		= sizeof(dirty_background_bytes),
992 		.mode		= 0644,
993 		.proc_handler	= dirty_background_bytes_handler,
994 		.extra1		= &one_ul,
995 	},
996 	{
997 		.procname	= "dirty_ratio",
998 		.data		= &vm_dirty_ratio,
999 		.maxlen		= sizeof(vm_dirty_ratio),
1000 		.mode		= 0644,
1001 		.proc_handler	= dirty_ratio_handler,
1002 		.extra1		= &zero,
1003 		.extra2		= &one_hundred,
1004 	},
1005 	{
1006 		.procname	= "dirty_bytes",
1007 		.data		= &vm_dirty_bytes,
1008 		.maxlen		= sizeof(vm_dirty_bytes),
1009 		.mode		= 0644,
1010 		.proc_handler	= dirty_bytes_handler,
1011 		.extra1		= &dirty_bytes_min,
1012 	},
1013 	{
1014 		.procname	= "dirty_writeback_centisecs",
1015 		.data		= &dirty_writeback_interval,
1016 		.maxlen		= sizeof(dirty_writeback_interval),
1017 		.mode		= 0644,
1018 		.proc_handler	= dirty_writeback_centisecs_handler,
1019 	},
1020 	{
1021 		.procname	= "dirty_expire_centisecs",
1022 		.data		= &dirty_expire_interval,
1023 		.maxlen		= sizeof(dirty_expire_interval),
1024 		.mode		= 0644,
1025 		.proc_handler	= proc_dointvec,
1026 	},
1027 	{
1028 		.procname	= "nr_pdflush_threads",
1029 		.data		= &nr_pdflush_threads,
1030 		.maxlen		= sizeof nr_pdflush_threads,
1031 		.mode		= 0444 /* read-only*/,
1032 		.proc_handler	= proc_dointvec,
1033 	},
1034 	{
1035 		.procname	= "swappiness",
1036 		.data		= &vm_swappiness,
1037 		.maxlen		= sizeof(vm_swappiness),
1038 		.mode		= 0644,
1039 		.proc_handler	= proc_dointvec_minmax,
1040 		.extra1		= &zero,
1041 		.extra2		= &one_hundred,
1042 	},
1043 #ifdef CONFIG_HUGETLB_PAGE
1044 	{
1045 		.procname	= "nr_hugepages",
1046 		.data		= NULL,
1047 		.maxlen		= sizeof(unsigned long),
1048 		.mode		= 0644,
1049 		.proc_handler	= hugetlb_sysctl_handler,
1050 		.extra1		= (void *)&hugetlb_zero,
1051 		.extra2		= (void *)&hugetlb_infinity,
1052 	},
1053 #ifdef CONFIG_NUMA
1054 	{
1055 		.procname       = "nr_hugepages_mempolicy",
1056 		.data           = NULL,
1057 		.maxlen         = sizeof(unsigned long),
1058 		.mode           = 0644,
1059 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1060 		.extra1		= (void *)&hugetlb_zero,
1061 		.extra2		= (void *)&hugetlb_infinity,
1062 	},
1063 #endif
1064 	 {
1065 		.procname	= "hugetlb_shm_group",
1066 		.data		= &sysctl_hugetlb_shm_group,
1067 		.maxlen		= sizeof(gid_t),
1068 		.mode		= 0644,
1069 		.proc_handler	= proc_dointvec,
1070 	 },
1071 	 {
1072 		.procname	= "hugepages_treat_as_movable",
1073 		.data		= &hugepages_treat_as_movable,
1074 		.maxlen		= sizeof(int),
1075 		.mode		= 0644,
1076 		.proc_handler	= hugetlb_treat_movable_handler,
1077 	},
1078 	{
1079 		.procname	= "nr_overcommit_hugepages",
1080 		.data		= NULL,
1081 		.maxlen		= sizeof(unsigned long),
1082 		.mode		= 0644,
1083 		.proc_handler	= hugetlb_overcommit_handler,
1084 		.extra1		= (void *)&hugetlb_zero,
1085 		.extra2		= (void *)&hugetlb_infinity,
1086 	},
1087 #endif
1088 	{
1089 		.procname	= "lowmem_reserve_ratio",
1090 		.data		= &sysctl_lowmem_reserve_ratio,
1091 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1092 		.mode		= 0644,
1093 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1094 	},
1095 	{
1096 		.procname	= "drop_caches",
1097 		.data		= &sysctl_drop_caches,
1098 		.maxlen		= sizeof(int),
1099 		.mode		= 0644,
1100 		.proc_handler	= drop_caches_sysctl_handler,
1101 	},
1102 	{
1103 		.procname	= "min_free_kbytes",
1104 		.data		= &min_free_kbytes,
1105 		.maxlen		= sizeof(min_free_kbytes),
1106 		.mode		= 0644,
1107 		.proc_handler	= min_free_kbytes_sysctl_handler,
1108 		.extra1		= &zero,
1109 	},
1110 	{
1111 		.procname	= "percpu_pagelist_fraction",
1112 		.data		= &percpu_pagelist_fraction,
1113 		.maxlen		= sizeof(percpu_pagelist_fraction),
1114 		.mode		= 0644,
1115 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1116 		.extra1		= &min_percpu_pagelist_fract,
1117 	},
1118 #ifdef CONFIG_MMU
1119 	{
1120 		.procname	= "max_map_count",
1121 		.data		= &sysctl_max_map_count,
1122 		.maxlen		= sizeof(sysctl_max_map_count),
1123 		.mode		= 0644,
1124 		.proc_handler	= proc_dointvec_minmax,
1125 		.extra1		= &zero,
1126 	},
1127 #else
1128 	{
1129 		.procname	= "nr_trim_pages",
1130 		.data		= &sysctl_nr_trim_pages,
1131 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1132 		.mode		= 0644,
1133 		.proc_handler	= proc_dointvec_minmax,
1134 		.extra1		= &zero,
1135 	},
1136 #endif
1137 	{
1138 		.procname	= "laptop_mode",
1139 		.data		= &laptop_mode,
1140 		.maxlen		= sizeof(laptop_mode),
1141 		.mode		= 0644,
1142 		.proc_handler	= proc_dointvec_jiffies,
1143 	},
1144 	{
1145 		.procname	= "block_dump",
1146 		.data		= &block_dump,
1147 		.maxlen		= sizeof(block_dump),
1148 		.mode		= 0644,
1149 		.proc_handler	= proc_dointvec,
1150 		.extra1		= &zero,
1151 	},
1152 	{
1153 		.procname	= "vfs_cache_pressure",
1154 		.data		= &sysctl_vfs_cache_pressure,
1155 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1156 		.mode		= 0644,
1157 		.proc_handler	= proc_dointvec,
1158 		.extra1		= &zero,
1159 	},
1160 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1161 	{
1162 		.procname	= "legacy_va_layout",
1163 		.data		= &sysctl_legacy_va_layout,
1164 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1165 		.mode		= 0644,
1166 		.proc_handler	= proc_dointvec,
1167 		.extra1		= &zero,
1168 	},
1169 #endif
1170 #ifdef CONFIG_NUMA
1171 	{
1172 		.procname	= "zone_reclaim_mode",
1173 		.data		= &zone_reclaim_mode,
1174 		.maxlen		= sizeof(zone_reclaim_mode),
1175 		.mode		= 0644,
1176 		.proc_handler	= proc_dointvec,
1177 		.extra1		= &zero,
1178 	},
1179 	{
1180 		.procname	= "min_unmapped_ratio",
1181 		.data		= &sysctl_min_unmapped_ratio,
1182 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1183 		.mode		= 0644,
1184 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1185 		.extra1		= &zero,
1186 		.extra2		= &one_hundred,
1187 	},
1188 	{
1189 		.procname	= "min_slab_ratio",
1190 		.data		= &sysctl_min_slab_ratio,
1191 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1192 		.mode		= 0644,
1193 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1194 		.extra1		= &zero,
1195 		.extra2		= &one_hundred,
1196 	},
1197 #endif
1198 #ifdef CONFIG_SMP
1199 	{
1200 		.procname	= "stat_interval",
1201 		.data		= &sysctl_stat_interval,
1202 		.maxlen		= sizeof(sysctl_stat_interval),
1203 		.mode		= 0644,
1204 		.proc_handler	= proc_dointvec_jiffies,
1205 	},
1206 #endif
1207 #ifdef CONFIG_MMU
1208 	{
1209 		.procname	= "mmap_min_addr",
1210 		.data		= &dac_mmap_min_addr,
1211 		.maxlen		= sizeof(unsigned long),
1212 		.mode		= 0644,
1213 		.proc_handler	= mmap_min_addr_handler,
1214 	},
1215 #endif
1216 #ifdef CONFIG_NUMA
1217 	{
1218 		.procname	= "numa_zonelist_order",
1219 		.data		= &numa_zonelist_order,
1220 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1221 		.mode		= 0644,
1222 		.proc_handler	= numa_zonelist_order_handler,
1223 	},
1224 #endif
1225 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1226    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1227 	{
1228 		.procname	= "vdso_enabled",
1229 		.data		= &vdso_enabled,
1230 		.maxlen		= sizeof(vdso_enabled),
1231 		.mode		= 0644,
1232 		.proc_handler	= proc_dointvec,
1233 		.extra1		= &zero,
1234 	},
1235 #endif
1236 #ifdef CONFIG_HIGHMEM
1237 	{
1238 		.procname	= "highmem_is_dirtyable",
1239 		.data		= &vm_highmem_is_dirtyable,
1240 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1241 		.mode		= 0644,
1242 		.proc_handler	= proc_dointvec_minmax,
1243 		.extra1		= &zero,
1244 		.extra2		= &one,
1245 	},
1246 #endif
1247 	{
1248 		.procname	= "scan_unevictable_pages",
1249 		.data		= &scan_unevictable_pages,
1250 		.maxlen		= sizeof(scan_unevictable_pages),
1251 		.mode		= 0644,
1252 		.proc_handler	= scan_unevictable_handler,
1253 	},
1254 #ifdef CONFIG_MEMORY_FAILURE
1255 	{
1256 		.procname	= "memory_failure_early_kill",
1257 		.data		= &sysctl_memory_failure_early_kill,
1258 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1259 		.mode		= 0644,
1260 		.proc_handler	= proc_dointvec_minmax,
1261 		.extra1		= &zero,
1262 		.extra2		= &one,
1263 	},
1264 	{
1265 		.procname	= "memory_failure_recovery",
1266 		.data		= &sysctl_memory_failure_recovery,
1267 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1268 		.mode		= 0644,
1269 		.proc_handler	= proc_dointvec_minmax,
1270 		.extra1		= &zero,
1271 		.extra2		= &one,
1272 	},
1273 #endif
1274 
1275 /*
1276  * NOTE: do not add new entries to this table unless you have read
1277  * Documentation/sysctl/ctl_unnumbered.txt
1278  */
1279 	{ }
1280 };
1281 
1282 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1283 static struct ctl_table binfmt_misc_table[] = {
1284 	{ }
1285 };
1286 #endif
1287 
1288 static struct ctl_table fs_table[] = {
1289 	{
1290 		.procname	= "inode-nr",
1291 		.data		= &inodes_stat,
1292 		.maxlen		= 2*sizeof(int),
1293 		.mode		= 0444,
1294 		.proc_handler	= proc_dointvec,
1295 	},
1296 	{
1297 		.procname	= "inode-state",
1298 		.data		= &inodes_stat,
1299 		.maxlen		= 7*sizeof(int),
1300 		.mode		= 0444,
1301 		.proc_handler	= proc_dointvec,
1302 	},
1303 	{
1304 		.procname	= "file-nr",
1305 		.data		= &files_stat,
1306 		.maxlen		= 3*sizeof(int),
1307 		.mode		= 0444,
1308 		.proc_handler	= proc_nr_files,
1309 	},
1310 	{
1311 		.procname	= "file-max",
1312 		.data		= &files_stat.max_files,
1313 		.maxlen		= sizeof(int),
1314 		.mode		= 0644,
1315 		.proc_handler	= proc_dointvec,
1316 	},
1317 	{
1318 		.procname	= "nr_open",
1319 		.data		= &sysctl_nr_open,
1320 		.maxlen		= sizeof(int),
1321 		.mode		= 0644,
1322 		.proc_handler	= proc_dointvec_minmax,
1323 		.extra1		= &sysctl_nr_open_min,
1324 		.extra2		= &sysctl_nr_open_max,
1325 	},
1326 	{
1327 		.procname	= "dentry-state",
1328 		.data		= &dentry_stat,
1329 		.maxlen		= 6*sizeof(int),
1330 		.mode		= 0444,
1331 		.proc_handler	= proc_dointvec,
1332 	},
1333 	{
1334 		.procname	= "overflowuid",
1335 		.data		= &fs_overflowuid,
1336 		.maxlen		= sizeof(int),
1337 		.mode		= 0644,
1338 		.proc_handler	= proc_dointvec_minmax,
1339 		.extra1		= &minolduid,
1340 		.extra2		= &maxolduid,
1341 	},
1342 	{
1343 		.procname	= "overflowgid",
1344 		.data		= &fs_overflowgid,
1345 		.maxlen		= sizeof(int),
1346 		.mode		= 0644,
1347 		.proc_handler	= proc_dointvec_minmax,
1348 		.extra1		= &minolduid,
1349 		.extra2		= &maxolduid,
1350 	},
1351 #ifdef CONFIG_FILE_LOCKING
1352 	{
1353 		.procname	= "leases-enable",
1354 		.data		= &leases_enable,
1355 		.maxlen		= sizeof(int),
1356 		.mode		= 0644,
1357 		.proc_handler	= proc_dointvec,
1358 	},
1359 #endif
1360 #ifdef CONFIG_DNOTIFY
1361 	{
1362 		.procname	= "dir-notify-enable",
1363 		.data		= &dir_notify_enable,
1364 		.maxlen		= sizeof(int),
1365 		.mode		= 0644,
1366 		.proc_handler	= proc_dointvec,
1367 	},
1368 #endif
1369 #ifdef CONFIG_MMU
1370 #ifdef CONFIG_FILE_LOCKING
1371 	{
1372 		.procname	= "lease-break-time",
1373 		.data		= &lease_break_time,
1374 		.maxlen		= sizeof(int),
1375 		.mode		= 0644,
1376 		.proc_handler	= proc_dointvec,
1377 	},
1378 #endif
1379 #ifdef CONFIG_AIO
1380 	{
1381 		.procname	= "aio-nr",
1382 		.data		= &aio_nr,
1383 		.maxlen		= sizeof(aio_nr),
1384 		.mode		= 0444,
1385 		.proc_handler	= proc_doulongvec_minmax,
1386 	},
1387 	{
1388 		.procname	= "aio-max-nr",
1389 		.data		= &aio_max_nr,
1390 		.maxlen		= sizeof(aio_max_nr),
1391 		.mode		= 0644,
1392 		.proc_handler	= proc_doulongvec_minmax,
1393 	},
1394 #endif /* CONFIG_AIO */
1395 #ifdef CONFIG_INOTIFY_USER
1396 	{
1397 		.procname	= "inotify",
1398 		.mode		= 0555,
1399 		.child		= inotify_table,
1400 	},
1401 #endif
1402 #ifdef CONFIG_EPOLL
1403 	{
1404 		.procname	= "epoll",
1405 		.mode		= 0555,
1406 		.child		= epoll_table,
1407 	},
1408 #endif
1409 #endif
1410 	{
1411 		.procname	= "suid_dumpable",
1412 		.data		= &suid_dumpable,
1413 		.maxlen		= sizeof(int),
1414 		.mode		= 0644,
1415 		.proc_handler	= proc_dointvec_minmax,
1416 		.extra1		= &zero,
1417 		.extra2		= &two,
1418 	},
1419 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1420 	{
1421 		.procname	= "binfmt_misc",
1422 		.mode		= 0555,
1423 		.child		= binfmt_misc_table,
1424 	},
1425 #endif
1426 /*
1427  * NOTE: do not add new entries to this table unless you have read
1428  * Documentation/sysctl/ctl_unnumbered.txt
1429  */
1430 	{ }
1431 };
1432 
1433 static struct ctl_table debug_table[] = {
1434 #if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC)
1435 	{
1436 		.procname	= "exception-trace",
1437 		.data		= &show_unhandled_signals,
1438 		.maxlen		= sizeof(int),
1439 		.mode		= 0644,
1440 		.proc_handler	= proc_dointvec
1441 	},
1442 #endif
1443 #if defined(CONFIG_OPTPROBES)
1444 	{
1445 		.procname	= "kprobes-optimization",
1446 		.data		= &sysctl_kprobes_optimization,
1447 		.maxlen		= sizeof(int),
1448 		.mode		= 0644,
1449 		.proc_handler	= proc_kprobes_optimization_handler,
1450 		.extra1		= &zero,
1451 		.extra2		= &one,
1452 	},
1453 #endif
1454 	{ }
1455 };
1456 
1457 static struct ctl_table dev_table[] = {
1458 	{ }
1459 };
1460 
1461 static DEFINE_SPINLOCK(sysctl_lock);
1462 
1463 /* called under sysctl_lock */
1464 static int use_table(struct ctl_table_header *p)
1465 {
1466 	if (unlikely(p->unregistering))
1467 		return 0;
1468 	p->used++;
1469 	return 1;
1470 }
1471 
1472 /* called under sysctl_lock */
1473 static void unuse_table(struct ctl_table_header *p)
1474 {
1475 	if (!--p->used)
1476 		if (unlikely(p->unregistering))
1477 			complete(p->unregistering);
1478 }
1479 
1480 /* called under sysctl_lock, will reacquire if has to wait */
1481 static void start_unregistering(struct ctl_table_header *p)
1482 {
1483 	/*
1484 	 * if p->used is 0, nobody will ever touch that entry again;
1485 	 * we'll eliminate all paths to it before dropping sysctl_lock
1486 	 */
1487 	if (unlikely(p->used)) {
1488 		struct completion wait;
1489 		init_completion(&wait);
1490 		p->unregistering = &wait;
1491 		spin_unlock(&sysctl_lock);
1492 		wait_for_completion(&wait);
1493 		spin_lock(&sysctl_lock);
1494 	} else {
1495 		/* anything non-NULL; we'll never dereference it */
1496 		p->unregistering = ERR_PTR(-EINVAL);
1497 	}
1498 	/*
1499 	 * do not remove from the list until nobody holds it; walking the
1500 	 * list in do_sysctl() relies on that.
1501 	 */
1502 	list_del_init(&p->ctl_entry);
1503 }
1504 
1505 void sysctl_head_get(struct ctl_table_header *head)
1506 {
1507 	spin_lock(&sysctl_lock);
1508 	head->count++;
1509 	spin_unlock(&sysctl_lock);
1510 }
1511 
1512 void sysctl_head_put(struct ctl_table_header *head)
1513 {
1514 	spin_lock(&sysctl_lock);
1515 	if (!--head->count)
1516 		kfree(head);
1517 	spin_unlock(&sysctl_lock);
1518 }
1519 
1520 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1521 {
1522 	if (!head)
1523 		BUG();
1524 	spin_lock(&sysctl_lock);
1525 	if (!use_table(head))
1526 		head = ERR_PTR(-ENOENT);
1527 	spin_unlock(&sysctl_lock);
1528 	return head;
1529 }
1530 
1531 void sysctl_head_finish(struct ctl_table_header *head)
1532 {
1533 	if (!head)
1534 		return;
1535 	spin_lock(&sysctl_lock);
1536 	unuse_table(head);
1537 	spin_unlock(&sysctl_lock);
1538 }
1539 
1540 static struct ctl_table_set *
1541 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1542 {
1543 	struct ctl_table_set *set = &root->default_set;
1544 	if (root->lookup)
1545 		set = root->lookup(root, namespaces);
1546 	return set;
1547 }
1548 
1549 static struct list_head *
1550 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1551 {
1552 	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1553 	return &set->list;
1554 }
1555 
1556 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1557 					    struct ctl_table_header *prev)
1558 {
1559 	struct ctl_table_root *root;
1560 	struct list_head *header_list;
1561 	struct ctl_table_header *head;
1562 	struct list_head *tmp;
1563 
1564 	spin_lock(&sysctl_lock);
1565 	if (prev) {
1566 		head = prev;
1567 		tmp = &prev->ctl_entry;
1568 		unuse_table(prev);
1569 		goto next;
1570 	}
1571 	tmp = &root_table_header.ctl_entry;
1572 	for (;;) {
1573 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1574 
1575 		if (!use_table(head))
1576 			goto next;
1577 		spin_unlock(&sysctl_lock);
1578 		return head;
1579 	next:
1580 		root = head->root;
1581 		tmp = tmp->next;
1582 		header_list = lookup_header_list(root, namespaces);
1583 		if (tmp != header_list)
1584 			continue;
1585 
1586 		do {
1587 			root = list_entry(root->root_list.next,
1588 					struct ctl_table_root, root_list);
1589 			if (root == &sysctl_table_root)
1590 				goto out;
1591 			header_list = lookup_header_list(root, namespaces);
1592 		} while (list_empty(header_list));
1593 		tmp = header_list->next;
1594 	}
1595 out:
1596 	spin_unlock(&sysctl_lock);
1597 	return NULL;
1598 }
1599 
1600 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1601 {
1602 	return __sysctl_head_next(current->nsproxy, prev);
1603 }
1604 
1605 void register_sysctl_root(struct ctl_table_root *root)
1606 {
1607 	spin_lock(&sysctl_lock);
1608 	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1609 	spin_unlock(&sysctl_lock);
1610 }
1611 
1612 /*
1613  * sysctl_perm does NOT grant the superuser all rights automatically, because
1614  * some sysctl variables are readonly even to root.
1615  */
1616 
1617 static int test_perm(int mode, int op)
1618 {
1619 	if (!current_euid())
1620 		mode >>= 6;
1621 	else if (in_egroup_p(0))
1622 		mode >>= 3;
1623 	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1624 		return 0;
1625 	return -EACCES;
1626 }
1627 
1628 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1629 {
1630 	int error;
1631 	int mode;
1632 
1633 	error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1634 	if (error)
1635 		return error;
1636 
1637 	if (root->permissions)
1638 		mode = root->permissions(root, current->nsproxy, table);
1639 	else
1640 		mode = table->mode;
1641 
1642 	return test_perm(mode, op);
1643 }
1644 
1645 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1646 {
1647 	for (; table->procname; table++) {
1648 		table->parent = parent;
1649 		if (table->child)
1650 			sysctl_set_parent(table, table->child);
1651 	}
1652 }
1653 
1654 static __init int sysctl_init(void)
1655 {
1656 	sysctl_set_parent(NULL, root_table);
1657 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1658 	{
1659 		int err;
1660 		err = sysctl_check_table(current->nsproxy, root_table);
1661 	}
1662 #endif
1663 	return 0;
1664 }
1665 
1666 core_initcall(sysctl_init);
1667 
1668 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1669 				      struct ctl_table *table)
1670 {
1671 	struct ctl_table *p;
1672 	const char *s = branch->procname;
1673 
1674 	/* branch should have named subdirectory as its first element */
1675 	if (!s || !branch->child)
1676 		return NULL;
1677 
1678 	/* ... and nothing else */
1679 	if (branch[1].procname)
1680 		return NULL;
1681 
1682 	/* table should contain subdirectory with the same name */
1683 	for (p = table; p->procname; p++) {
1684 		if (!p->child)
1685 			continue;
1686 		if (p->procname && strcmp(p->procname, s) == 0)
1687 			return p;
1688 	}
1689 	return NULL;
1690 }
1691 
1692 /* see if attaching q to p would be an improvement */
1693 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1694 {
1695 	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1696 	struct ctl_table *next;
1697 	int is_better = 0;
1698 	int not_in_parent = !p->attached_by;
1699 
1700 	while ((next = is_branch_in(by, to)) != NULL) {
1701 		if (by == q->attached_by)
1702 			is_better = 1;
1703 		if (to == p->attached_by)
1704 			not_in_parent = 1;
1705 		by = by->child;
1706 		to = next->child;
1707 	}
1708 
1709 	if (is_better && not_in_parent) {
1710 		q->attached_by = by;
1711 		q->attached_to = to;
1712 		q->parent = p;
1713 	}
1714 }
1715 
1716 /**
1717  * __register_sysctl_paths - register a sysctl hierarchy
1718  * @root: List of sysctl headers to register on
1719  * @namespaces: Data to compute which lists of sysctl entries are visible
1720  * @path: The path to the directory the sysctl table is in.
1721  * @table: the top-level table structure
1722  *
1723  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1724  * array. A completely 0 filled entry terminates the table.
1725  *
1726  * The members of the &struct ctl_table structure are used as follows:
1727  *
1728  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1729  *            enter a sysctl file
1730  *
1731  * data - a pointer to data for use by proc_handler
1732  *
1733  * maxlen - the maximum size in bytes of the data
1734  *
1735  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1736  *
1737  * child - a pointer to the child sysctl table if this entry is a directory, or
1738  *         %NULL.
1739  *
1740  * proc_handler - the text handler routine (described below)
1741  *
1742  * de - for internal use by the sysctl routines
1743  *
1744  * extra1, extra2 - extra pointers usable by the proc handler routines
1745  *
1746  * Leaf nodes in the sysctl tree will be represented by a single file
1747  * under /proc; non-leaf nodes will be represented by directories.
1748  *
1749  * sysctl(2) can automatically manage read and write requests through
1750  * the sysctl table.  The data and maxlen fields of the ctl_table
1751  * struct enable minimal validation of the values being written to be
1752  * performed, and the mode field allows minimal authentication.
1753  *
1754  * There must be a proc_handler routine for any terminal nodes
1755  * mirrored under /proc/sys (non-terminals are handled by a built-in
1756  * directory handler).  Several default handlers are available to
1757  * cover common cases -
1758  *
1759  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1760  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1761  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1762  *
1763  * It is the handler's job to read the input buffer from user memory
1764  * and process it. The handler should return 0 on success.
1765  *
1766  * This routine returns %NULL on a failure to register, and a pointer
1767  * to the table header on success.
1768  */
1769 struct ctl_table_header *__register_sysctl_paths(
1770 	struct ctl_table_root *root,
1771 	struct nsproxy *namespaces,
1772 	const struct ctl_path *path, struct ctl_table *table)
1773 {
1774 	struct ctl_table_header *header;
1775 	struct ctl_table *new, **prevp;
1776 	unsigned int n, npath;
1777 	struct ctl_table_set *set;
1778 
1779 	/* Count the path components */
1780 	for (npath = 0; path[npath].procname; ++npath)
1781 		;
1782 
1783 	/*
1784 	 * For each path component, allocate a 2-element ctl_table array.
1785 	 * The first array element will be filled with the sysctl entry
1786 	 * for this, the second will be the sentinel (procname == 0).
1787 	 *
1788 	 * We allocate everything in one go so that we don't have to
1789 	 * worry about freeing additional memory in unregister_sysctl_table.
1790 	 */
1791 	header = kzalloc(sizeof(struct ctl_table_header) +
1792 			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1793 	if (!header)
1794 		return NULL;
1795 
1796 	new = (struct ctl_table *) (header + 1);
1797 
1798 	/* Now connect the dots */
1799 	prevp = &header->ctl_table;
1800 	for (n = 0; n < npath; ++n, ++path) {
1801 		/* Copy the procname */
1802 		new->procname = path->procname;
1803 		new->mode     = 0555;
1804 
1805 		*prevp = new;
1806 		prevp = &new->child;
1807 
1808 		new += 2;
1809 	}
1810 	*prevp = table;
1811 	header->ctl_table_arg = table;
1812 
1813 	INIT_LIST_HEAD(&header->ctl_entry);
1814 	header->used = 0;
1815 	header->unregistering = NULL;
1816 	header->root = root;
1817 	sysctl_set_parent(NULL, header->ctl_table);
1818 	header->count = 1;
1819 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1820 	if (sysctl_check_table(namespaces, header->ctl_table)) {
1821 		kfree(header);
1822 		return NULL;
1823 	}
1824 #endif
1825 	spin_lock(&sysctl_lock);
1826 	header->set = lookup_header_set(root, namespaces);
1827 	header->attached_by = header->ctl_table;
1828 	header->attached_to = root_table;
1829 	header->parent = &root_table_header;
1830 	for (set = header->set; set; set = set->parent) {
1831 		struct ctl_table_header *p;
1832 		list_for_each_entry(p, &set->list, ctl_entry) {
1833 			if (p->unregistering)
1834 				continue;
1835 			try_attach(p, header);
1836 		}
1837 	}
1838 	header->parent->count++;
1839 	list_add_tail(&header->ctl_entry, &header->set->list);
1840 	spin_unlock(&sysctl_lock);
1841 
1842 	return header;
1843 }
1844 
1845 /**
1846  * register_sysctl_table_path - register a sysctl table hierarchy
1847  * @path: The path to the directory the sysctl table is in.
1848  * @table: the top-level table structure
1849  *
1850  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1851  * array. A completely 0 filled entry terminates the table.
1852  *
1853  * See __register_sysctl_paths for more details.
1854  */
1855 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1856 						struct ctl_table *table)
1857 {
1858 	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1859 					path, table);
1860 }
1861 
1862 /**
1863  * register_sysctl_table - register a sysctl table hierarchy
1864  * @table: the top-level table structure
1865  *
1866  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1867  * array. A completely 0 filled entry terminates the table.
1868  *
1869  * See register_sysctl_paths for more details.
1870  */
1871 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1872 {
1873 	static const struct ctl_path null_path[] = { {} };
1874 
1875 	return register_sysctl_paths(null_path, table);
1876 }
1877 
1878 /**
1879  * unregister_sysctl_table - unregister a sysctl table hierarchy
1880  * @header: the header returned from register_sysctl_table
1881  *
1882  * Unregisters the sysctl table and all children. proc entries may not
1883  * actually be removed until they are no longer used by anyone.
1884  */
1885 void unregister_sysctl_table(struct ctl_table_header * header)
1886 {
1887 	might_sleep();
1888 
1889 	if (header == NULL)
1890 		return;
1891 
1892 	spin_lock(&sysctl_lock);
1893 	start_unregistering(header);
1894 	if (!--header->parent->count) {
1895 		WARN_ON(1);
1896 		kfree(header->parent);
1897 	}
1898 	if (!--header->count)
1899 		kfree(header);
1900 	spin_unlock(&sysctl_lock);
1901 }
1902 
1903 int sysctl_is_seen(struct ctl_table_header *p)
1904 {
1905 	struct ctl_table_set *set = p->set;
1906 	int res;
1907 	spin_lock(&sysctl_lock);
1908 	if (p->unregistering)
1909 		res = 0;
1910 	else if (!set->is_seen)
1911 		res = 1;
1912 	else
1913 		res = set->is_seen(set);
1914 	spin_unlock(&sysctl_lock);
1915 	return res;
1916 }
1917 
1918 void setup_sysctl_set(struct ctl_table_set *p,
1919 	struct ctl_table_set *parent,
1920 	int (*is_seen)(struct ctl_table_set *))
1921 {
1922 	INIT_LIST_HEAD(&p->list);
1923 	p->parent = parent ? parent : &sysctl_table_root.default_set;
1924 	p->is_seen = is_seen;
1925 }
1926 
1927 #else /* !CONFIG_SYSCTL */
1928 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1929 {
1930 	return NULL;
1931 }
1932 
1933 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1934 						    struct ctl_table *table)
1935 {
1936 	return NULL;
1937 }
1938 
1939 void unregister_sysctl_table(struct ctl_table_header * table)
1940 {
1941 }
1942 
1943 void setup_sysctl_set(struct ctl_table_set *p,
1944 	struct ctl_table_set *parent,
1945 	int (*is_seen)(struct ctl_table_set *))
1946 {
1947 }
1948 
1949 void sysctl_head_put(struct ctl_table_header *head)
1950 {
1951 }
1952 
1953 #endif /* CONFIG_SYSCTL */
1954 
1955 /*
1956  * /proc/sys support
1957  */
1958 
1959 #ifdef CONFIG_PROC_SYSCTL
1960 
1961 static int _proc_do_string(void* data, int maxlen, int write,
1962 			   void __user *buffer,
1963 			   size_t *lenp, loff_t *ppos)
1964 {
1965 	size_t len;
1966 	char __user *p;
1967 	char c;
1968 
1969 	if (!data || !maxlen || !*lenp) {
1970 		*lenp = 0;
1971 		return 0;
1972 	}
1973 
1974 	if (write) {
1975 		len = 0;
1976 		p = buffer;
1977 		while (len < *lenp) {
1978 			if (get_user(c, p++))
1979 				return -EFAULT;
1980 			if (c == 0 || c == '\n')
1981 				break;
1982 			len++;
1983 		}
1984 		if (len >= maxlen)
1985 			len = maxlen-1;
1986 		if(copy_from_user(data, buffer, len))
1987 			return -EFAULT;
1988 		((char *) data)[len] = 0;
1989 		*ppos += *lenp;
1990 	} else {
1991 		len = strlen(data);
1992 		if (len > maxlen)
1993 			len = maxlen;
1994 
1995 		if (*ppos > len) {
1996 			*lenp = 0;
1997 			return 0;
1998 		}
1999 
2000 		data += *ppos;
2001 		len  -= *ppos;
2002 
2003 		if (len > *lenp)
2004 			len = *lenp;
2005 		if (len)
2006 			if(copy_to_user(buffer, data, len))
2007 				return -EFAULT;
2008 		if (len < *lenp) {
2009 			if(put_user('\n', ((char __user *) buffer) + len))
2010 				return -EFAULT;
2011 			len++;
2012 		}
2013 		*lenp = len;
2014 		*ppos += len;
2015 	}
2016 	return 0;
2017 }
2018 
2019 /**
2020  * proc_dostring - read a string sysctl
2021  * @table: the sysctl table
2022  * @write: %TRUE if this is a write to the sysctl file
2023  * @buffer: the user buffer
2024  * @lenp: the size of the user buffer
2025  * @ppos: file position
2026  *
2027  * Reads/writes a string from/to the user buffer. If the kernel
2028  * buffer provided is not large enough to hold the string, the
2029  * string is truncated. The copied string is %NULL-terminated.
2030  * If the string is being read by the user process, it is copied
2031  * and a newline '\n' is added. It is truncated if the buffer is
2032  * not large enough.
2033  *
2034  * Returns 0 on success.
2035  */
2036 int proc_dostring(struct ctl_table *table, int write,
2037 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2038 {
2039 	return _proc_do_string(table->data, table->maxlen, write,
2040 			       buffer, lenp, ppos);
2041 }
2042 
2043 
2044 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2045 				 int *valp,
2046 				 int write, void *data)
2047 {
2048 	if (write) {
2049 		*valp = *negp ? -*lvalp : *lvalp;
2050 	} else {
2051 		int val = *valp;
2052 		if (val < 0) {
2053 			*negp = -1;
2054 			*lvalp = (unsigned long)-val;
2055 		} else {
2056 			*negp = 0;
2057 			*lvalp = (unsigned long)val;
2058 		}
2059 	}
2060 	return 0;
2061 }
2062 
2063 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2064 		  int write, void __user *buffer,
2065 		  size_t *lenp, loff_t *ppos,
2066 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2067 			      int write, void *data),
2068 		  void *data)
2069 {
2070 #define TMPBUFLEN 21
2071 	int *i, vleft, first = 1, neg;
2072 	unsigned long lval;
2073 	size_t left, len;
2074 
2075 	char buf[TMPBUFLEN], *p;
2076 	char __user *s = buffer;
2077 
2078 	if (!tbl_data || !table->maxlen || !*lenp ||
2079 	    (*ppos && !write)) {
2080 		*lenp = 0;
2081 		return 0;
2082 	}
2083 
2084 	i = (int *) tbl_data;
2085 	vleft = table->maxlen / sizeof(*i);
2086 	left = *lenp;
2087 
2088 	if (!conv)
2089 		conv = do_proc_dointvec_conv;
2090 
2091 	for (; left && vleft--; i++, first=0) {
2092 		if (write) {
2093 			while (left) {
2094 				char c;
2095 				if (get_user(c, s))
2096 					return -EFAULT;
2097 				if (!isspace(c))
2098 					break;
2099 				left--;
2100 				s++;
2101 			}
2102 			if (!left)
2103 				break;
2104 			neg = 0;
2105 			len = left;
2106 			if (len > sizeof(buf) - 1)
2107 				len = sizeof(buf) - 1;
2108 			if (copy_from_user(buf, s, len))
2109 				return -EFAULT;
2110 			buf[len] = 0;
2111 			p = buf;
2112 			if (*p == '-' && left > 1) {
2113 				neg = 1;
2114 				p++;
2115 			}
2116 			if (*p < '0' || *p > '9')
2117 				break;
2118 
2119 			lval = simple_strtoul(p, &p, 0);
2120 
2121 			len = p-buf;
2122 			if ((len < left) && *p && !isspace(*p))
2123 				break;
2124 			s += len;
2125 			left -= len;
2126 
2127 			if (conv(&neg, &lval, i, 1, data))
2128 				break;
2129 		} else {
2130 			p = buf;
2131 			if (!first)
2132 				*p++ = '\t';
2133 
2134 			if (conv(&neg, &lval, i, 0, data))
2135 				break;
2136 
2137 			sprintf(p, "%s%lu", neg ? "-" : "", lval);
2138 			len = strlen(buf);
2139 			if (len > left)
2140 				len = left;
2141 			if(copy_to_user(s, buf, len))
2142 				return -EFAULT;
2143 			left -= len;
2144 			s += len;
2145 		}
2146 	}
2147 
2148 	if (!write && !first && left) {
2149 		if(put_user('\n', s))
2150 			return -EFAULT;
2151 		left--, s++;
2152 	}
2153 	if (write) {
2154 		while (left) {
2155 			char c;
2156 			if (get_user(c, s++))
2157 				return -EFAULT;
2158 			if (!isspace(c))
2159 				break;
2160 			left--;
2161 		}
2162 	}
2163 	if (write && first)
2164 		return -EINVAL;
2165 	*lenp -= left;
2166 	*ppos += *lenp;
2167 	return 0;
2168 #undef TMPBUFLEN
2169 }
2170 
2171 static int do_proc_dointvec(struct ctl_table *table, int write,
2172 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2173 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2174 			      int write, void *data),
2175 		  void *data)
2176 {
2177 	return __do_proc_dointvec(table->data, table, write,
2178 			buffer, lenp, ppos, conv, data);
2179 }
2180 
2181 /**
2182  * proc_dointvec - read a vector of integers
2183  * @table: the sysctl table
2184  * @write: %TRUE if this is a write to the sysctl file
2185  * @buffer: the user buffer
2186  * @lenp: the size of the user buffer
2187  * @ppos: file position
2188  *
2189  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2190  * values from/to the user buffer, treated as an ASCII string.
2191  *
2192  * Returns 0 on success.
2193  */
2194 int proc_dointvec(struct ctl_table *table, int write,
2195 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2196 {
2197     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2198 		    	    NULL,NULL);
2199 }
2200 
2201 /*
2202  * Taint values can only be increased
2203  * This means we can safely use a temporary.
2204  */
2205 static int proc_taint(struct ctl_table *table, int write,
2206 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2207 {
2208 	struct ctl_table t;
2209 	unsigned long tmptaint = get_taint();
2210 	int err;
2211 
2212 	if (write && !capable(CAP_SYS_ADMIN))
2213 		return -EPERM;
2214 
2215 	t = *table;
2216 	t.data = &tmptaint;
2217 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2218 	if (err < 0)
2219 		return err;
2220 
2221 	if (write) {
2222 		/*
2223 		 * Poor man's atomic or. Not worth adding a primitive
2224 		 * to everyone's atomic.h for this
2225 		 */
2226 		int i;
2227 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2228 			if ((tmptaint >> i) & 1)
2229 				add_taint(i);
2230 		}
2231 	}
2232 
2233 	return err;
2234 }
2235 
2236 struct do_proc_dointvec_minmax_conv_param {
2237 	int *min;
2238 	int *max;
2239 };
2240 
2241 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2242 					int *valp,
2243 					int write, void *data)
2244 {
2245 	struct do_proc_dointvec_minmax_conv_param *param = data;
2246 	if (write) {
2247 		int val = *negp ? -*lvalp : *lvalp;
2248 		if ((param->min && *param->min > val) ||
2249 		    (param->max && *param->max < val))
2250 			return -EINVAL;
2251 		*valp = val;
2252 	} else {
2253 		int val = *valp;
2254 		if (val < 0) {
2255 			*negp = -1;
2256 			*lvalp = (unsigned long)-val;
2257 		} else {
2258 			*negp = 0;
2259 			*lvalp = (unsigned long)val;
2260 		}
2261 	}
2262 	return 0;
2263 }
2264 
2265 /**
2266  * proc_dointvec_minmax - read a vector of integers with min/max values
2267  * @table: the sysctl table
2268  * @write: %TRUE if this is a write to the sysctl file
2269  * @buffer: the user buffer
2270  * @lenp: the size of the user buffer
2271  * @ppos: file position
2272  *
2273  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2274  * values from/to the user buffer, treated as an ASCII string.
2275  *
2276  * This routine will ensure the values are within the range specified by
2277  * table->extra1 (min) and table->extra2 (max).
2278  *
2279  * Returns 0 on success.
2280  */
2281 int proc_dointvec_minmax(struct ctl_table *table, int write,
2282 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2283 {
2284 	struct do_proc_dointvec_minmax_conv_param param = {
2285 		.min = (int *) table->extra1,
2286 		.max = (int *) table->extra2,
2287 	};
2288 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2289 				do_proc_dointvec_minmax_conv, &param);
2290 }
2291 
2292 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2293 				     void __user *buffer,
2294 				     size_t *lenp, loff_t *ppos,
2295 				     unsigned long convmul,
2296 				     unsigned long convdiv)
2297 {
2298 #define TMPBUFLEN 21
2299 	unsigned long *i, *min, *max, val;
2300 	int vleft, first=1, neg;
2301 	size_t len, left;
2302 	char buf[TMPBUFLEN], *p;
2303 	char __user *s = buffer;
2304 
2305 	if (!data || !table->maxlen || !*lenp ||
2306 	    (*ppos && !write)) {
2307 		*lenp = 0;
2308 		return 0;
2309 	}
2310 
2311 	i = (unsigned long *) data;
2312 	min = (unsigned long *) table->extra1;
2313 	max = (unsigned long *) table->extra2;
2314 	vleft = table->maxlen / sizeof(unsigned long);
2315 	left = *lenp;
2316 
2317 	for (; left && vleft--; i++, min++, max++, first=0) {
2318 		if (write) {
2319 			while (left) {
2320 				char c;
2321 				if (get_user(c, s))
2322 					return -EFAULT;
2323 				if (!isspace(c))
2324 					break;
2325 				left--;
2326 				s++;
2327 			}
2328 			if (!left)
2329 				break;
2330 			neg = 0;
2331 			len = left;
2332 			if (len > TMPBUFLEN-1)
2333 				len = TMPBUFLEN-1;
2334 			if (copy_from_user(buf, s, len))
2335 				return -EFAULT;
2336 			buf[len] = 0;
2337 			p = buf;
2338 			if (*p == '-' && left > 1) {
2339 				neg = 1;
2340 				p++;
2341 			}
2342 			if (*p < '0' || *p > '9')
2343 				break;
2344 			val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2345 			len = p-buf;
2346 			if ((len < left) && *p && !isspace(*p))
2347 				break;
2348 			if (neg)
2349 				val = -val;
2350 			s += len;
2351 			left -= len;
2352 
2353 			if(neg)
2354 				continue;
2355 			if ((min && val < *min) || (max && val > *max))
2356 				continue;
2357 			*i = val;
2358 		} else {
2359 			p = buf;
2360 			if (!first)
2361 				*p++ = '\t';
2362 			sprintf(p, "%lu", convdiv * (*i) / convmul);
2363 			len = strlen(buf);
2364 			if (len > left)
2365 				len = left;
2366 			if(copy_to_user(s, buf, len))
2367 				return -EFAULT;
2368 			left -= len;
2369 			s += len;
2370 		}
2371 	}
2372 
2373 	if (!write && !first && left) {
2374 		if(put_user('\n', s))
2375 			return -EFAULT;
2376 		left--, s++;
2377 	}
2378 	if (write) {
2379 		while (left) {
2380 			char c;
2381 			if (get_user(c, s++))
2382 				return -EFAULT;
2383 			if (!isspace(c))
2384 				break;
2385 			left--;
2386 		}
2387 	}
2388 	if (write && first)
2389 		return -EINVAL;
2390 	*lenp -= left;
2391 	*ppos += *lenp;
2392 	return 0;
2393 #undef TMPBUFLEN
2394 }
2395 
2396 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2397 				     void __user *buffer,
2398 				     size_t *lenp, loff_t *ppos,
2399 				     unsigned long convmul,
2400 				     unsigned long convdiv)
2401 {
2402 	return __do_proc_doulongvec_minmax(table->data, table, write,
2403 			buffer, lenp, ppos, convmul, convdiv);
2404 }
2405 
2406 /**
2407  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2408  * @table: the sysctl table
2409  * @write: %TRUE if this is a write to the sysctl file
2410  * @buffer: the user buffer
2411  * @lenp: the size of the user buffer
2412  * @ppos: file position
2413  *
2414  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2415  * values from/to the user buffer, treated as an ASCII string.
2416  *
2417  * This routine will ensure the values are within the range specified by
2418  * table->extra1 (min) and table->extra2 (max).
2419  *
2420  * Returns 0 on success.
2421  */
2422 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2423 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2424 {
2425     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2426 }
2427 
2428 /**
2429  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2430  * @table: the sysctl table
2431  * @write: %TRUE if this is a write to the sysctl file
2432  * @buffer: the user buffer
2433  * @lenp: the size of the user buffer
2434  * @ppos: file position
2435  *
2436  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2437  * values from/to the user buffer, treated as an ASCII string. The values
2438  * are treated as milliseconds, and converted to jiffies when they are stored.
2439  *
2440  * This routine will ensure the values are within the range specified by
2441  * table->extra1 (min) and table->extra2 (max).
2442  *
2443  * Returns 0 on success.
2444  */
2445 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2446 				      void __user *buffer,
2447 				      size_t *lenp, loff_t *ppos)
2448 {
2449     return do_proc_doulongvec_minmax(table, write, buffer,
2450 				     lenp, ppos, HZ, 1000l);
2451 }
2452 
2453 
2454 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2455 					 int *valp,
2456 					 int write, void *data)
2457 {
2458 	if (write) {
2459 		if (*lvalp > LONG_MAX / HZ)
2460 			return 1;
2461 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2462 	} else {
2463 		int val = *valp;
2464 		unsigned long lval;
2465 		if (val < 0) {
2466 			*negp = -1;
2467 			lval = (unsigned long)-val;
2468 		} else {
2469 			*negp = 0;
2470 			lval = (unsigned long)val;
2471 		}
2472 		*lvalp = lval / HZ;
2473 	}
2474 	return 0;
2475 }
2476 
2477 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2478 						int *valp,
2479 						int write, void *data)
2480 {
2481 	if (write) {
2482 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2483 			return 1;
2484 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2485 	} else {
2486 		int val = *valp;
2487 		unsigned long lval;
2488 		if (val < 0) {
2489 			*negp = -1;
2490 			lval = (unsigned long)-val;
2491 		} else {
2492 			*negp = 0;
2493 			lval = (unsigned long)val;
2494 		}
2495 		*lvalp = jiffies_to_clock_t(lval);
2496 	}
2497 	return 0;
2498 }
2499 
2500 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2501 					    int *valp,
2502 					    int write, void *data)
2503 {
2504 	if (write) {
2505 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2506 	} else {
2507 		int val = *valp;
2508 		unsigned long lval;
2509 		if (val < 0) {
2510 			*negp = -1;
2511 			lval = (unsigned long)-val;
2512 		} else {
2513 			*negp = 0;
2514 			lval = (unsigned long)val;
2515 		}
2516 		*lvalp = jiffies_to_msecs(lval);
2517 	}
2518 	return 0;
2519 }
2520 
2521 /**
2522  * proc_dointvec_jiffies - read a vector of integers as seconds
2523  * @table: the sysctl table
2524  * @write: %TRUE if this is a write to the sysctl file
2525  * @buffer: the user buffer
2526  * @lenp: the size of the user buffer
2527  * @ppos: file position
2528  *
2529  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2530  * values from/to the user buffer, treated as an ASCII string.
2531  * The values read are assumed to be in seconds, and are converted into
2532  * jiffies.
2533  *
2534  * Returns 0 on success.
2535  */
2536 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2537 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2538 {
2539     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2540 		    	    do_proc_dointvec_jiffies_conv,NULL);
2541 }
2542 
2543 /**
2544  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2545  * @table: the sysctl table
2546  * @write: %TRUE if this is a write to the sysctl file
2547  * @buffer: the user buffer
2548  * @lenp: the size of the user buffer
2549  * @ppos: pointer to the file position
2550  *
2551  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2552  * values from/to the user buffer, treated as an ASCII string.
2553  * The values read are assumed to be in 1/USER_HZ seconds, and
2554  * are converted into jiffies.
2555  *
2556  * Returns 0 on success.
2557  */
2558 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2559 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2560 {
2561     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2562 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2563 }
2564 
2565 /**
2566  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2567  * @table: the sysctl table
2568  * @write: %TRUE if this is a write to the sysctl file
2569  * @buffer: the user buffer
2570  * @lenp: the size of the user buffer
2571  * @ppos: file position
2572  * @ppos: the current position in the file
2573  *
2574  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2575  * values from/to the user buffer, treated as an ASCII string.
2576  * The values read are assumed to be in 1/1000 seconds, and
2577  * are converted into jiffies.
2578  *
2579  * Returns 0 on success.
2580  */
2581 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2582 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2583 {
2584 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2585 				do_proc_dointvec_ms_jiffies_conv, NULL);
2586 }
2587 
2588 static int proc_do_cad_pid(struct ctl_table *table, int write,
2589 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2590 {
2591 	struct pid *new_pid;
2592 	pid_t tmp;
2593 	int r;
2594 
2595 	tmp = pid_vnr(cad_pid);
2596 
2597 	r = __do_proc_dointvec(&tmp, table, write, buffer,
2598 			       lenp, ppos, NULL, NULL);
2599 	if (r || !write)
2600 		return r;
2601 
2602 	new_pid = find_get_pid(tmp);
2603 	if (!new_pid)
2604 		return -ESRCH;
2605 
2606 	put_pid(xchg(&cad_pid, new_pid));
2607 	return 0;
2608 }
2609 
2610 #else /* CONFIG_PROC_FS */
2611 
2612 int proc_dostring(struct ctl_table *table, int write,
2613 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2614 {
2615 	return -ENOSYS;
2616 }
2617 
2618 int proc_dointvec(struct ctl_table *table, int write,
2619 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2620 {
2621 	return -ENOSYS;
2622 }
2623 
2624 int proc_dointvec_minmax(struct ctl_table *table, int write,
2625 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2626 {
2627 	return -ENOSYS;
2628 }
2629 
2630 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2631 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2632 {
2633 	return -ENOSYS;
2634 }
2635 
2636 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2637 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2638 {
2639 	return -ENOSYS;
2640 }
2641 
2642 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2643 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2644 {
2645 	return -ENOSYS;
2646 }
2647 
2648 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2649 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2650 {
2651 	return -ENOSYS;
2652 }
2653 
2654 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2655 				      void __user *buffer,
2656 				      size_t *lenp, loff_t *ppos)
2657 {
2658     return -ENOSYS;
2659 }
2660 
2661 
2662 #endif /* CONFIG_PROC_FS */
2663 
2664 /*
2665  * No sense putting this after each symbol definition, twice,
2666  * exception granted :-)
2667  */
2668 EXPORT_SYMBOL(proc_dointvec);
2669 EXPORT_SYMBOL(proc_dointvec_jiffies);
2670 EXPORT_SYMBOL(proc_dointvec_minmax);
2671 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2672 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2673 EXPORT_SYMBOL(proc_dostring);
2674 EXPORT_SYMBOL(proc_doulongvec_minmax);
2675 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2676 EXPORT_SYMBOL(register_sysctl_table);
2677 EXPORT_SYMBOL(register_sysctl_paths);
2678 EXPORT_SYMBOL(unregister_sysctl_table);
2679