xref: /linux/kernel/sysctl.c (revision b8bb76713ec50df2f11efee386e16f93d51e1076)
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/proc_fs.h>
27 #include <linux/security.h>
28 #include <linux/ctype.h>
29 #include <linux/utsname.h>
30 #include <linux/smp_lock.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/hugetlb.h>
40 #include <linux/initrd.h>
41 #include <linux/key.h>
42 #include <linux/times.h>
43 #include <linux/limits.h>
44 #include <linux/dcache.h>
45 #include <linux/syscalls.h>
46 #include <linux/vmstat.h>
47 #include <linux/nfs_fs.h>
48 #include <linux/acpi.h>
49 #include <linux/reboot.h>
50 #include <linux/ftrace.h>
51 
52 #include <asm/uaccess.h>
53 #include <asm/processor.h>
54 
55 #ifdef CONFIG_X86
56 #include <asm/nmi.h>
57 #include <asm/stacktrace.h>
58 #include <asm/io.h>
59 #endif
60 
61 static int deprecated_sysctl_warning(struct __sysctl_args *args);
62 
63 #if defined(CONFIG_SYSCTL)
64 
65 /* External variables not in a header file. */
66 extern int C_A_D;
67 extern int print_fatal_signals;
68 extern int sysctl_overcommit_memory;
69 extern int sysctl_overcommit_ratio;
70 extern int sysctl_panic_on_oom;
71 extern int sysctl_oom_kill_allocating_task;
72 extern int sysctl_oom_dump_tasks;
73 extern int max_threads;
74 extern int core_uses_pid;
75 extern int suid_dumpable;
76 extern char core_pattern[];
77 extern int pid_max;
78 extern int min_free_kbytes;
79 extern int pid_max_min, pid_max_max;
80 extern int sysctl_drop_caches;
81 extern int percpu_pagelist_fraction;
82 extern int compat_log;
83 extern int latencytop_enabled;
84 extern int sysctl_nr_open_min, sysctl_nr_open_max;
85 #ifndef CONFIG_MMU
86 extern int sysctl_nr_trim_pages;
87 #endif
88 #ifdef CONFIG_RCU_TORTURE_TEST
89 extern int rcutorture_runnable;
90 #endif /* #ifdef CONFIG_RCU_TORTURE_TEST */
91 
92 /* Constants used for minimum and  maximum */
93 #ifdef CONFIG_DETECT_SOFTLOCKUP
94 static int sixty = 60;
95 static int neg_one = -1;
96 #endif
97 
98 #if defined(CONFIG_MMU) && defined(CONFIG_FILE_LOCKING)
99 static int two = 2;
100 #endif
101 
102 static int zero;
103 static int one = 1;
104 static unsigned long one_ul = 1;
105 static int one_hundred = 100;
106 
107 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
108 static int maxolduid = 65535;
109 static int minolduid;
110 static int min_percpu_pagelist_fract = 8;
111 
112 static int ngroups_max = NGROUPS_MAX;
113 
114 #ifdef CONFIG_MODULES
115 extern char modprobe_path[];
116 #endif
117 #ifdef CONFIG_CHR_DEV_SG
118 extern int sg_big_buff;
119 #endif
120 
121 #ifdef CONFIG_SPARC
122 #include <asm/system.h>
123 #endif
124 
125 #ifdef CONFIG_SPARC64
126 extern int sysctl_tsb_ratio;
127 #endif
128 
129 #ifdef __hppa__
130 extern int pwrsw_enabled;
131 extern int unaligned_enabled;
132 #endif
133 
134 #ifdef CONFIG_S390
135 #ifdef CONFIG_MATHEMU
136 extern int sysctl_ieee_emulation_warnings;
137 #endif
138 extern int sysctl_userprocess_debug;
139 extern int spin_retry;
140 #endif
141 
142 #ifdef CONFIG_BSD_PROCESS_ACCT
143 extern int acct_parm[];
144 #endif
145 
146 #ifdef CONFIG_IA64
147 extern int no_unaligned_warning;
148 extern int unaligned_dump_stack;
149 #endif
150 
151 #ifdef CONFIG_RT_MUTEXES
152 extern int max_lock_depth;
153 #endif
154 
155 #ifdef CONFIG_PROC_SYSCTL
156 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
157 		  void __user *buffer, size_t *lenp, loff_t *ppos);
158 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
159 			       void __user *buffer, size_t *lenp, loff_t *ppos);
160 #endif
161 
162 static struct ctl_table root_table[];
163 static struct ctl_table_root sysctl_table_root;
164 static struct ctl_table_header root_table_header = {
165 	.count = 1,
166 	.ctl_table = root_table,
167 	.ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
168 	.root = &sysctl_table_root,
169 	.set = &sysctl_table_root.default_set,
170 };
171 static struct ctl_table_root sysctl_table_root = {
172 	.root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
173 	.default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
174 };
175 
176 static struct ctl_table kern_table[];
177 static struct ctl_table vm_table[];
178 static struct ctl_table fs_table[];
179 static struct ctl_table debug_table[];
180 static struct ctl_table dev_table[];
181 extern struct ctl_table random_table[];
182 #ifdef CONFIG_INOTIFY_USER
183 extern struct ctl_table inotify_table[];
184 #endif
185 #ifdef CONFIG_EPOLL
186 extern struct ctl_table epoll_table[];
187 #endif
188 
189 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
190 int sysctl_legacy_va_layout;
191 #endif
192 
193 extern int prove_locking;
194 extern int lock_stat;
195 
196 /* The default sysctl tables: */
197 
198 static struct ctl_table root_table[] = {
199 	{
200 		.ctl_name	= CTL_KERN,
201 		.procname	= "kernel",
202 		.mode		= 0555,
203 		.child		= kern_table,
204 	},
205 	{
206 		.ctl_name	= CTL_VM,
207 		.procname	= "vm",
208 		.mode		= 0555,
209 		.child		= vm_table,
210 	},
211 	{
212 		.ctl_name	= CTL_FS,
213 		.procname	= "fs",
214 		.mode		= 0555,
215 		.child		= fs_table,
216 	},
217 	{
218 		.ctl_name	= CTL_DEBUG,
219 		.procname	= "debug",
220 		.mode		= 0555,
221 		.child		= debug_table,
222 	},
223 	{
224 		.ctl_name	= CTL_DEV,
225 		.procname	= "dev",
226 		.mode		= 0555,
227 		.child		= dev_table,
228 	},
229 /*
230  * NOTE: do not add new entries to this table unless you have read
231  * Documentation/sysctl/ctl_unnumbered.txt
232  */
233 	{ .ctl_name = 0 }
234 };
235 
236 #ifdef CONFIG_SCHED_DEBUG
237 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
238 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
239 static int min_wakeup_granularity_ns;			/* 0 usecs */
240 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
241 #endif
242 
243 static struct ctl_table kern_table[] = {
244 #ifdef CONFIG_SCHED_DEBUG
245 	{
246 		.ctl_name	= CTL_UNNUMBERED,
247 		.procname	= "sched_min_granularity_ns",
248 		.data		= &sysctl_sched_min_granularity,
249 		.maxlen		= sizeof(unsigned int),
250 		.mode		= 0644,
251 		.proc_handler	= &sched_nr_latency_handler,
252 		.strategy	= &sysctl_intvec,
253 		.extra1		= &min_sched_granularity_ns,
254 		.extra2		= &max_sched_granularity_ns,
255 	},
256 	{
257 		.ctl_name	= CTL_UNNUMBERED,
258 		.procname	= "sched_latency_ns",
259 		.data		= &sysctl_sched_latency,
260 		.maxlen		= sizeof(unsigned int),
261 		.mode		= 0644,
262 		.proc_handler	= &sched_nr_latency_handler,
263 		.strategy	= &sysctl_intvec,
264 		.extra1		= &min_sched_granularity_ns,
265 		.extra2		= &max_sched_granularity_ns,
266 	},
267 	{
268 		.ctl_name	= CTL_UNNUMBERED,
269 		.procname	= "sched_wakeup_granularity_ns",
270 		.data		= &sysctl_sched_wakeup_granularity,
271 		.maxlen		= sizeof(unsigned int),
272 		.mode		= 0644,
273 		.proc_handler	= &proc_dointvec_minmax,
274 		.strategy	= &sysctl_intvec,
275 		.extra1		= &min_wakeup_granularity_ns,
276 		.extra2		= &max_wakeup_granularity_ns,
277 	},
278 	{
279 		.ctl_name	= CTL_UNNUMBERED,
280 		.procname	= "sched_shares_ratelimit",
281 		.data		= &sysctl_sched_shares_ratelimit,
282 		.maxlen		= sizeof(unsigned int),
283 		.mode		= 0644,
284 		.proc_handler	= &proc_dointvec,
285 	},
286 	{
287 		.ctl_name	= CTL_UNNUMBERED,
288 		.procname	= "sched_shares_thresh",
289 		.data		= &sysctl_sched_shares_thresh,
290 		.maxlen		= sizeof(unsigned int),
291 		.mode		= 0644,
292 		.proc_handler	= &proc_dointvec_minmax,
293 		.strategy	= &sysctl_intvec,
294 		.extra1		= &zero,
295 	},
296 	{
297 		.ctl_name	= CTL_UNNUMBERED,
298 		.procname	= "sched_child_runs_first",
299 		.data		= &sysctl_sched_child_runs_first,
300 		.maxlen		= sizeof(unsigned int),
301 		.mode		= 0644,
302 		.proc_handler	= &proc_dointvec,
303 	},
304 	{
305 		.ctl_name	= CTL_UNNUMBERED,
306 		.procname	= "sched_features",
307 		.data		= &sysctl_sched_features,
308 		.maxlen		= sizeof(unsigned int),
309 		.mode		= 0644,
310 		.proc_handler	= &proc_dointvec,
311 	},
312 	{
313 		.ctl_name	= CTL_UNNUMBERED,
314 		.procname	= "sched_migration_cost",
315 		.data		= &sysctl_sched_migration_cost,
316 		.maxlen		= sizeof(unsigned int),
317 		.mode		= 0644,
318 		.proc_handler	= &proc_dointvec,
319 	},
320 	{
321 		.ctl_name	= CTL_UNNUMBERED,
322 		.procname	= "sched_nr_migrate",
323 		.data		= &sysctl_sched_nr_migrate,
324 		.maxlen		= sizeof(unsigned int),
325 		.mode		= 0644,
326 		.proc_handler	= &proc_dointvec,
327 	},
328 #endif
329 	{
330 		.ctl_name	= CTL_UNNUMBERED,
331 		.procname	= "sched_rt_period_us",
332 		.data		= &sysctl_sched_rt_period,
333 		.maxlen		= sizeof(unsigned int),
334 		.mode		= 0644,
335 		.proc_handler	= &sched_rt_handler,
336 	},
337 	{
338 		.ctl_name	= CTL_UNNUMBERED,
339 		.procname	= "sched_rt_runtime_us",
340 		.data		= &sysctl_sched_rt_runtime,
341 		.maxlen		= sizeof(int),
342 		.mode		= 0644,
343 		.proc_handler	= &sched_rt_handler,
344 	},
345 	{
346 		.ctl_name	= CTL_UNNUMBERED,
347 		.procname	= "sched_compat_yield",
348 		.data		= &sysctl_sched_compat_yield,
349 		.maxlen		= sizeof(unsigned int),
350 		.mode		= 0644,
351 		.proc_handler	= &proc_dointvec,
352 	},
353 #ifdef CONFIG_PROVE_LOCKING
354 	{
355 		.ctl_name	= CTL_UNNUMBERED,
356 		.procname	= "prove_locking",
357 		.data		= &prove_locking,
358 		.maxlen		= sizeof(int),
359 		.mode		= 0644,
360 		.proc_handler	= &proc_dointvec,
361 	},
362 #endif
363 #ifdef CONFIG_LOCK_STAT
364 	{
365 		.ctl_name	= CTL_UNNUMBERED,
366 		.procname	= "lock_stat",
367 		.data		= &lock_stat,
368 		.maxlen		= sizeof(int),
369 		.mode		= 0644,
370 		.proc_handler	= &proc_dointvec,
371 	},
372 #endif
373 	{
374 		.ctl_name	= KERN_PANIC,
375 		.procname	= "panic",
376 		.data		= &panic_timeout,
377 		.maxlen		= sizeof(int),
378 		.mode		= 0644,
379 		.proc_handler	= &proc_dointvec,
380 	},
381 	{
382 		.ctl_name	= KERN_CORE_USES_PID,
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 		.ctl_name	= KERN_CORE_PATTERN,
391 		.procname	= "core_pattern",
392 		.data		= core_pattern,
393 		.maxlen		= CORENAME_MAX_SIZE,
394 		.mode		= 0644,
395 		.proc_handler	= &proc_dostring,
396 		.strategy	= &sysctl_string,
397 	},
398 #ifdef CONFIG_PROC_SYSCTL
399 	{
400 		.procname	= "tainted",
401 		.maxlen 	= sizeof(long),
402 		.mode		= 0644,
403 		.proc_handler	= &proc_taint,
404 	},
405 #endif
406 #ifdef CONFIG_LATENCYTOP
407 	{
408 		.procname	= "latencytop",
409 		.data		= &latencytop_enabled,
410 		.maxlen		= sizeof(int),
411 		.mode		= 0644,
412 		.proc_handler	= &proc_dointvec,
413 	},
414 #endif
415 #ifdef CONFIG_BLK_DEV_INITRD
416 	{
417 		.ctl_name	= KERN_REALROOTDEV,
418 		.procname	= "real-root-dev",
419 		.data		= &real_root_dev,
420 		.maxlen		= sizeof(int),
421 		.mode		= 0644,
422 		.proc_handler	= &proc_dointvec,
423 	},
424 #endif
425 	{
426 		.ctl_name	= CTL_UNNUMBERED,
427 		.procname	= "print-fatal-signals",
428 		.data		= &print_fatal_signals,
429 		.maxlen		= sizeof(int),
430 		.mode		= 0644,
431 		.proc_handler	= &proc_dointvec,
432 	},
433 #ifdef CONFIG_SPARC
434 	{
435 		.ctl_name	= KERN_SPARC_REBOOT,
436 		.procname	= "reboot-cmd",
437 		.data		= reboot_command,
438 		.maxlen		= 256,
439 		.mode		= 0644,
440 		.proc_handler	= &proc_dostring,
441 		.strategy	= &sysctl_string,
442 	},
443 	{
444 		.ctl_name	= KERN_SPARC_STOP_A,
445 		.procname	= "stop-a",
446 		.data		= &stop_a_enabled,
447 		.maxlen		= sizeof (int),
448 		.mode		= 0644,
449 		.proc_handler	= &proc_dointvec,
450 	},
451 	{
452 		.ctl_name	= KERN_SPARC_SCONS_PWROFF,
453 		.procname	= "scons-poweroff",
454 		.data		= &scons_pwroff,
455 		.maxlen		= sizeof (int),
456 		.mode		= 0644,
457 		.proc_handler	= &proc_dointvec,
458 	},
459 #endif
460 #ifdef CONFIG_SPARC64
461 	{
462 		.ctl_name	= CTL_UNNUMBERED,
463 		.procname	= "tsb-ratio",
464 		.data		= &sysctl_tsb_ratio,
465 		.maxlen		= sizeof (int),
466 		.mode		= 0644,
467 		.proc_handler	= &proc_dointvec,
468 	},
469 #endif
470 #ifdef __hppa__
471 	{
472 		.ctl_name	= KERN_HPPA_PWRSW,
473 		.procname	= "soft-power",
474 		.data		= &pwrsw_enabled,
475 		.maxlen		= sizeof (int),
476 	 	.mode		= 0644,
477 		.proc_handler	= &proc_dointvec,
478 	},
479 	{
480 		.ctl_name	= KERN_HPPA_UNALIGNED,
481 		.procname	= "unaligned-trap",
482 		.data		= &unaligned_enabled,
483 		.maxlen		= sizeof (int),
484 		.mode		= 0644,
485 		.proc_handler	= &proc_dointvec,
486 	},
487 #endif
488 	{
489 		.ctl_name	= KERN_CTLALTDEL,
490 		.procname	= "ctrl-alt-del",
491 		.data		= &C_A_D,
492 		.maxlen		= sizeof(int),
493 		.mode		= 0644,
494 		.proc_handler	= &proc_dointvec,
495 	},
496 #ifdef CONFIG_FUNCTION_TRACER
497 	{
498 		.ctl_name	= CTL_UNNUMBERED,
499 		.procname	= "ftrace_enabled",
500 		.data		= &ftrace_enabled,
501 		.maxlen		= sizeof(int),
502 		.mode		= 0644,
503 		.proc_handler	= &ftrace_enable_sysctl,
504 	},
505 #endif
506 #ifdef CONFIG_STACK_TRACER
507 	{
508 		.ctl_name	= CTL_UNNUMBERED,
509 		.procname	= "stack_tracer_enabled",
510 		.data		= &stack_tracer_enabled,
511 		.maxlen		= sizeof(int),
512 		.mode		= 0644,
513 		.proc_handler	= &stack_trace_sysctl,
514 	},
515 #endif
516 #ifdef CONFIG_TRACING
517 	{
518 		.ctl_name	= CTL_UNNUMBERED,
519 		.procname	= "ftrace_dump_on_oops",
520 		.data		= &ftrace_dump_on_oops,
521 		.maxlen		= sizeof(int),
522 		.mode		= 0644,
523 		.proc_handler	= &proc_dointvec,
524 	},
525 #endif
526 #ifdef CONFIG_MODULES
527 	{
528 		.ctl_name	= KERN_MODPROBE,
529 		.procname	= "modprobe",
530 		.data		= &modprobe_path,
531 		.maxlen		= KMOD_PATH_LEN,
532 		.mode		= 0644,
533 		.proc_handler	= &proc_dostring,
534 		.strategy	= &sysctl_string,
535 	},
536 #endif
537 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
538 	{
539 		.ctl_name	= KERN_HOTPLUG,
540 		.procname	= "hotplug",
541 		.data		= &uevent_helper,
542 		.maxlen		= UEVENT_HELPER_PATH_LEN,
543 		.mode		= 0644,
544 		.proc_handler	= &proc_dostring,
545 		.strategy	= &sysctl_string,
546 	},
547 #endif
548 #ifdef CONFIG_CHR_DEV_SG
549 	{
550 		.ctl_name	= KERN_SG_BIG_BUFF,
551 		.procname	= "sg-big-buff",
552 		.data		= &sg_big_buff,
553 		.maxlen		= sizeof (int),
554 		.mode		= 0444,
555 		.proc_handler	= &proc_dointvec,
556 	},
557 #endif
558 #ifdef CONFIG_BSD_PROCESS_ACCT
559 	{
560 		.ctl_name	= KERN_ACCT,
561 		.procname	= "acct",
562 		.data		= &acct_parm,
563 		.maxlen		= 3*sizeof(int),
564 		.mode		= 0644,
565 		.proc_handler	= &proc_dointvec,
566 	},
567 #endif
568 #ifdef CONFIG_MAGIC_SYSRQ
569 	{
570 		.ctl_name	= KERN_SYSRQ,
571 		.procname	= "sysrq",
572 		.data		= &__sysrq_enabled,
573 		.maxlen		= sizeof (int),
574 		.mode		= 0644,
575 		.proc_handler	= &proc_dointvec,
576 	},
577 #endif
578 #ifdef CONFIG_PROC_SYSCTL
579 	{
580 		.procname	= "cad_pid",
581 		.data		= NULL,
582 		.maxlen		= sizeof (int),
583 		.mode		= 0600,
584 		.proc_handler	= &proc_do_cad_pid,
585 	},
586 #endif
587 	{
588 		.ctl_name	= KERN_MAX_THREADS,
589 		.procname	= "threads-max",
590 		.data		= &max_threads,
591 		.maxlen		= sizeof(int),
592 		.mode		= 0644,
593 		.proc_handler	= &proc_dointvec,
594 	},
595 	{
596 		.ctl_name	= KERN_RANDOM,
597 		.procname	= "random",
598 		.mode		= 0555,
599 		.child		= random_table,
600 	},
601 	{
602 		.ctl_name	= KERN_OVERFLOWUID,
603 		.procname	= "overflowuid",
604 		.data		= &overflowuid,
605 		.maxlen		= sizeof(int),
606 		.mode		= 0644,
607 		.proc_handler	= &proc_dointvec_minmax,
608 		.strategy	= &sysctl_intvec,
609 		.extra1		= &minolduid,
610 		.extra2		= &maxolduid,
611 	},
612 	{
613 		.ctl_name	= KERN_OVERFLOWGID,
614 		.procname	= "overflowgid",
615 		.data		= &overflowgid,
616 		.maxlen		= sizeof(int),
617 		.mode		= 0644,
618 		.proc_handler	= &proc_dointvec_minmax,
619 		.strategy	= &sysctl_intvec,
620 		.extra1		= &minolduid,
621 		.extra2		= &maxolduid,
622 	},
623 #ifdef CONFIG_S390
624 #ifdef CONFIG_MATHEMU
625 	{
626 		.ctl_name	= KERN_IEEE_EMULATION_WARNINGS,
627 		.procname	= "ieee_emulation_warnings",
628 		.data		= &sysctl_ieee_emulation_warnings,
629 		.maxlen		= sizeof(int),
630 		.mode		= 0644,
631 		.proc_handler	= &proc_dointvec,
632 	},
633 #endif
634 	{
635 		.ctl_name	= KERN_S390_USER_DEBUG_LOGGING,
636 		.procname	= "userprocess_debug",
637 		.data		= &sysctl_userprocess_debug,
638 		.maxlen		= sizeof(int),
639 		.mode		= 0644,
640 		.proc_handler	= &proc_dointvec,
641 	},
642 #endif
643 	{
644 		.ctl_name	= KERN_PIDMAX,
645 		.procname	= "pid_max",
646 		.data		= &pid_max,
647 		.maxlen		= sizeof (int),
648 		.mode		= 0644,
649 		.proc_handler	= &proc_dointvec_minmax,
650 		.strategy	= sysctl_intvec,
651 		.extra1		= &pid_max_min,
652 		.extra2		= &pid_max_max,
653 	},
654 	{
655 		.ctl_name	= KERN_PANIC_ON_OOPS,
656 		.procname	= "panic_on_oops",
657 		.data		= &panic_on_oops,
658 		.maxlen		= sizeof(int),
659 		.mode		= 0644,
660 		.proc_handler	= &proc_dointvec,
661 	},
662 #if defined CONFIG_PRINTK
663 	{
664 		.ctl_name	= KERN_PRINTK,
665 		.procname	= "printk",
666 		.data		= &console_loglevel,
667 		.maxlen		= 4*sizeof(int),
668 		.mode		= 0644,
669 		.proc_handler	= &proc_dointvec,
670 	},
671 	{
672 		.ctl_name	= KERN_PRINTK_RATELIMIT,
673 		.procname	= "printk_ratelimit",
674 		.data		= &printk_ratelimit_state.interval,
675 		.maxlen		= sizeof(int),
676 		.mode		= 0644,
677 		.proc_handler	= &proc_dointvec_jiffies,
678 		.strategy	= &sysctl_jiffies,
679 	},
680 	{
681 		.ctl_name	= KERN_PRINTK_RATELIMIT_BURST,
682 		.procname	= "printk_ratelimit_burst",
683 		.data		= &printk_ratelimit_state.burst,
684 		.maxlen		= sizeof(int),
685 		.mode		= 0644,
686 		.proc_handler	= &proc_dointvec,
687 	},
688 #endif
689 	{
690 		.ctl_name	= KERN_NGROUPS_MAX,
691 		.procname	= "ngroups_max",
692 		.data		= &ngroups_max,
693 		.maxlen		= sizeof (int),
694 		.mode		= 0444,
695 		.proc_handler	= &proc_dointvec,
696 	},
697 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
698 	{
699 		.ctl_name       = KERN_UNKNOWN_NMI_PANIC,
700 		.procname       = "unknown_nmi_panic",
701 		.data           = &unknown_nmi_panic,
702 		.maxlen         = sizeof (int),
703 		.mode           = 0644,
704 		.proc_handler   = &proc_dointvec,
705 	},
706 	{
707 		.procname       = "nmi_watchdog",
708 		.data           = &nmi_watchdog_enabled,
709 		.maxlen         = sizeof (int),
710 		.mode           = 0644,
711 		.proc_handler   = &proc_nmi_enabled,
712 	},
713 #endif
714 #if defined(CONFIG_X86)
715 	{
716 		.ctl_name	= KERN_PANIC_ON_NMI,
717 		.procname	= "panic_on_unrecovered_nmi",
718 		.data		= &panic_on_unrecovered_nmi,
719 		.maxlen		= sizeof(int),
720 		.mode		= 0644,
721 		.proc_handler	= &proc_dointvec,
722 	},
723 	{
724 		.ctl_name	= KERN_BOOTLOADER_TYPE,
725 		.procname	= "bootloader_type",
726 		.data		= &bootloader_type,
727 		.maxlen		= sizeof (int),
728 		.mode		= 0444,
729 		.proc_handler	= &proc_dointvec,
730 	},
731 	{
732 		.ctl_name	= CTL_UNNUMBERED,
733 		.procname	= "kstack_depth_to_print",
734 		.data		= &kstack_depth_to_print,
735 		.maxlen		= sizeof(int),
736 		.mode		= 0644,
737 		.proc_handler	= &proc_dointvec,
738 	},
739 	{
740 		.ctl_name	= CTL_UNNUMBERED,
741 		.procname	= "io_delay_type",
742 		.data		= &io_delay_type,
743 		.maxlen		= sizeof(int),
744 		.mode		= 0644,
745 		.proc_handler	= &proc_dointvec,
746 	},
747 #endif
748 #if defined(CONFIG_MMU)
749 	{
750 		.ctl_name	= KERN_RANDOMIZE,
751 		.procname	= "randomize_va_space",
752 		.data		= &randomize_va_space,
753 		.maxlen		= sizeof(int),
754 		.mode		= 0644,
755 		.proc_handler	= &proc_dointvec,
756 	},
757 #endif
758 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
759 	{
760 		.ctl_name	= KERN_SPIN_RETRY,
761 		.procname	= "spin_retry",
762 		.data		= &spin_retry,
763 		.maxlen		= sizeof (int),
764 		.mode		= 0644,
765 		.proc_handler	= &proc_dointvec,
766 	},
767 #endif
768 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
769 	{
770 		.procname	= "acpi_video_flags",
771 		.data		= &acpi_realmode_flags,
772 		.maxlen		= sizeof (unsigned long),
773 		.mode		= 0644,
774 		.proc_handler	= &proc_doulongvec_minmax,
775 	},
776 #endif
777 #ifdef CONFIG_IA64
778 	{
779 		.ctl_name	= KERN_IA64_UNALIGNED,
780 		.procname	= "ignore-unaligned-usertrap",
781 		.data		= &no_unaligned_warning,
782 		.maxlen		= sizeof (int),
783 	 	.mode		= 0644,
784 		.proc_handler	= &proc_dointvec,
785 	},
786 	{
787 		.ctl_name	= CTL_UNNUMBERED,
788 		.procname	= "unaligned-dump-stack",
789 		.data		= &unaligned_dump_stack,
790 		.maxlen		= sizeof (int),
791 		.mode		= 0644,
792 		.proc_handler	= &proc_dointvec,
793 	},
794 #endif
795 #ifdef CONFIG_DETECT_SOFTLOCKUP
796 	{
797 		.ctl_name	= CTL_UNNUMBERED,
798 		.procname	= "softlockup_panic",
799 		.data		= &softlockup_panic,
800 		.maxlen		= sizeof(int),
801 		.mode		= 0644,
802 		.proc_handler	= &proc_dointvec_minmax,
803 		.strategy	= &sysctl_intvec,
804 		.extra1		= &zero,
805 		.extra2		= &one,
806 	},
807 	{
808 		.ctl_name	= CTL_UNNUMBERED,
809 		.procname	= "softlockup_thresh",
810 		.data		= &softlockup_thresh,
811 		.maxlen		= sizeof(int),
812 		.mode		= 0644,
813 		.proc_handler	= &proc_dosoftlockup_thresh,
814 		.strategy	= &sysctl_intvec,
815 		.extra1		= &neg_one,
816 		.extra2		= &sixty,
817 	},
818 	{
819 		.ctl_name	= CTL_UNNUMBERED,
820 		.procname	= "hung_task_check_count",
821 		.data		= &sysctl_hung_task_check_count,
822 		.maxlen		= sizeof(unsigned long),
823 		.mode		= 0644,
824 		.proc_handler	= &proc_doulongvec_minmax,
825 		.strategy	= &sysctl_intvec,
826 	},
827 	{
828 		.ctl_name	= CTL_UNNUMBERED,
829 		.procname	= "hung_task_timeout_secs",
830 		.data		= &sysctl_hung_task_timeout_secs,
831 		.maxlen		= sizeof(unsigned long),
832 		.mode		= 0644,
833 		.proc_handler	= &proc_doulongvec_minmax,
834 		.strategy	= &sysctl_intvec,
835 	},
836 	{
837 		.ctl_name	= CTL_UNNUMBERED,
838 		.procname	= "hung_task_warnings",
839 		.data		= &sysctl_hung_task_warnings,
840 		.maxlen		= sizeof(unsigned long),
841 		.mode		= 0644,
842 		.proc_handler	= &proc_doulongvec_minmax,
843 		.strategy	= &sysctl_intvec,
844 	},
845 #endif
846 #ifdef CONFIG_COMPAT
847 	{
848 		.ctl_name	= KERN_COMPAT_LOG,
849 		.procname	= "compat-log",
850 		.data		= &compat_log,
851 		.maxlen		= sizeof (int),
852 	 	.mode		= 0644,
853 		.proc_handler	= &proc_dointvec,
854 	},
855 #endif
856 #ifdef CONFIG_RT_MUTEXES
857 	{
858 		.ctl_name	= KERN_MAX_LOCK_DEPTH,
859 		.procname	= "max_lock_depth",
860 		.data		= &max_lock_depth,
861 		.maxlen		= sizeof(int),
862 		.mode		= 0644,
863 		.proc_handler	= &proc_dointvec,
864 	},
865 #endif
866 	{
867 		.ctl_name	= CTL_UNNUMBERED,
868 		.procname	= "poweroff_cmd",
869 		.data		= &poweroff_cmd,
870 		.maxlen		= POWEROFF_CMD_PATH_LEN,
871 		.mode		= 0644,
872 		.proc_handler	= &proc_dostring,
873 		.strategy	= &sysctl_string,
874 	},
875 #ifdef CONFIG_KEYS
876 	{
877 		.ctl_name	= CTL_UNNUMBERED,
878 		.procname	= "keys",
879 		.mode		= 0555,
880 		.child		= key_sysctls,
881 	},
882 #endif
883 #ifdef CONFIG_RCU_TORTURE_TEST
884 	{
885 		.ctl_name       = CTL_UNNUMBERED,
886 		.procname       = "rcutorture_runnable",
887 		.data           = &rcutorture_runnable,
888 		.maxlen         = sizeof(int),
889 		.mode           = 0644,
890 		.proc_handler   = &proc_dointvec,
891 	},
892 #endif
893 #ifdef CONFIG_UNEVICTABLE_LRU
894 	{
895 		.ctl_name	= CTL_UNNUMBERED,
896 		.procname	= "scan_unevictable_pages",
897 		.data		= &scan_unevictable_pages,
898 		.maxlen		= sizeof(scan_unevictable_pages),
899 		.mode		= 0644,
900 		.proc_handler	= &scan_unevictable_handler,
901 	},
902 #endif
903 /*
904  * NOTE: do not add new entries to this table unless you have read
905  * Documentation/sysctl/ctl_unnumbered.txt
906  */
907 	{ .ctl_name = 0 }
908 };
909 
910 static struct ctl_table vm_table[] = {
911 	{
912 		.ctl_name	= VM_OVERCOMMIT_MEMORY,
913 		.procname	= "overcommit_memory",
914 		.data		= &sysctl_overcommit_memory,
915 		.maxlen		= sizeof(sysctl_overcommit_memory),
916 		.mode		= 0644,
917 		.proc_handler	= &proc_dointvec,
918 	},
919 	{
920 		.ctl_name	= VM_PANIC_ON_OOM,
921 		.procname	= "panic_on_oom",
922 		.data		= &sysctl_panic_on_oom,
923 		.maxlen		= sizeof(sysctl_panic_on_oom),
924 		.mode		= 0644,
925 		.proc_handler	= &proc_dointvec,
926 	},
927 	{
928 		.ctl_name	= CTL_UNNUMBERED,
929 		.procname	= "oom_kill_allocating_task",
930 		.data		= &sysctl_oom_kill_allocating_task,
931 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
932 		.mode		= 0644,
933 		.proc_handler	= &proc_dointvec,
934 	},
935 	{
936 		.ctl_name	= CTL_UNNUMBERED,
937 		.procname	= "oom_dump_tasks",
938 		.data		= &sysctl_oom_dump_tasks,
939 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
940 		.mode		= 0644,
941 		.proc_handler	= &proc_dointvec,
942 	},
943 	{
944 		.ctl_name	= VM_OVERCOMMIT_RATIO,
945 		.procname	= "overcommit_ratio",
946 		.data		= &sysctl_overcommit_ratio,
947 		.maxlen		= sizeof(sysctl_overcommit_ratio),
948 		.mode		= 0644,
949 		.proc_handler	= &proc_dointvec,
950 	},
951 	{
952 		.ctl_name	= VM_PAGE_CLUSTER,
953 		.procname	= "page-cluster",
954 		.data		= &page_cluster,
955 		.maxlen		= sizeof(int),
956 		.mode		= 0644,
957 		.proc_handler	= &proc_dointvec,
958 	},
959 	{
960 		.ctl_name	= VM_DIRTY_BACKGROUND,
961 		.procname	= "dirty_background_ratio",
962 		.data		= &dirty_background_ratio,
963 		.maxlen		= sizeof(dirty_background_ratio),
964 		.mode		= 0644,
965 		.proc_handler	= &dirty_background_ratio_handler,
966 		.strategy	= &sysctl_intvec,
967 		.extra1		= &zero,
968 		.extra2		= &one_hundred,
969 	},
970 	{
971 		.ctl_name	= CTL_UNNUMBERED,
972 		.procname	= "dirty_background_bytes",
973 		.data		= &dirty_background_bytes,
974 		.maxlen		= sizeof(dirty_background_bytes),
975 		.mode		= 0644,
976 		.proc_handler	= &dirty_background_bytes_handler,
977 		.strategy	= &sysctl_intvec,
978 		.extra1		= &one_ul,
979 	},
980 	{
981 		.ctl_name	= VM_DIRTY_RATIO,
982 		.procname	= "dirty_ratio",
983 		.data		= &vm_dirty_ratio,
984 		.maxlen		= sizeof(vm_dirty_ratio),
985 		.mode		= 0644,
986 		.proc_handler	= &dirty_ratio_handler,
987 		.strategy	= &sysctl_intvec,
988 		.extra1		= &zero,
989 		.extra2		= &one_hundred,
990 	},
991 	{
992 		.ctl_name	= CTL_UNNUMBERED,
993 		.procname	= "dirty_bytes",
994 		.data		= &vm_dirty_bytes,
995 		.maxlen		= sizeof(vm_dirty_bytes),
996 		.mode		= 0644,
997 		.proc_handler	= &dirty_bytes_handler,
998 		.strategy	= &sysctl_intvec,
999 		.extra1		= &one_ul,
1000 	},
1001 	{
1002 		.procname	= "dirty_writeback_centisecs",
1003 		.data		= &dirty_writeback_interval,
1004 		.maxlen		= sizeof(dirty_writeback_interval),
1005 		.mode		= 0644,
1006 		.proc_handler	= &dirty_writeback_centisecs_handler,
1007 	},
1008 	{
1009 		.procname	= "dirty_expire_centisecs",
1010 		.data		= &dirty_expire_interval,
1011 		.maxlen		= sizeof(dirty_expire_interval),
1012 		.mode		= 0644,
1013 		.proc_handler	= &proc_dointvec,
1014 	},
1015 	{
1016 		.ctl_name	= VM_NR_PDFLUSH_THREADS,
1017 		.procname	= "nr_pdflush_threads",
1018 		.data		= &nr_pdflush_threads,
1019 		.maxlen		= sizeof nr_pdflush_threads,
1020 		.mode		= 0444 /* read-only*/,
1021 		.proc_handler	= &proc_dointvec,
1022 	},
1023 	{
1024 		.ctl_name	= VM_SWAPPINESS,
1025 		.procname	= "swappiness",
1026 		.data		= &vm_swappiness,
1027 		.maxlen		= sizeof(vm_swappiness),
1028 		.mode		= 0644,
1029 		.proc_handler	= &proc_dointvec_minmax,
1030 		.strategy	= &sysctl_intvec,
1031 		.extra1		= &zero,
1032 		.extra2		= &one_hundred,
1033 	},
1034 #ifdef CONFIG_HUGETLB_PAGE
1035 	 {
1036 		.procname	= "nr_hugepages",
1037 		.data		= NULL,
1038 		.maxlen		= sizeof(unsigned long),
1039 		.mode		= 0644,
1040 		.proc_handler	= &hugetlb_sysctl_handler,
1041 		.extra1		= (void *)&hugetlb_zero,
1042 		.extra2		= (void *)&hugetlb_infinity,
1043 	 },
1044 	 {
1045 		.ctl_name	= VM_HUGETLB_GROUP,
1046 		.procname	= "hugetlb_shm_group",
1047 		.data		= &sysctl_hugetlb_shm_group,
1048 		.maxlen		= sizeof(gid_t),
1049 		.mode		= 0644,
1050 		.proc_handler	= &proc_dointvec,
1051 	 },
1052 	 {
1053 		.ctl_name	= CTL_UNNUMBERED,
1054 		.procname	= "hugepages_treat_as_movable",
1055 		.data		= &hugepages_treat_as_movable,
1056 		.maxlen		= sizeof(int),
1057 		.mode		= 0644,
1058 		.proc_handler	= &hugetlb_treat_movable_handler,
1059 	},
1060 	{
1061 		.ctl_name	= CTL_UNNUMBERED,
1062 		.procname	= "nr_overcommit_hugepages",
1063 		.data		= NULL,
1064 		.maxlen		= sizeof(unsigned long),
1065 		.mode		= 0644,
1066 		.proc_handler	= &hugetlb_overcommit_handler,
1067 		.extra1		= (void *)&hugetlb_zero,
1068 		.extra2		= (void *)&hugetlb_infinity,
1069 	},
1070 #endif
1071 	{
1072 		.ctl_name	= VM_LOWMEM_RESERVE_RATIO,
1073 		.procname	= "lowmem_reserve_ratio",
1074 		.data		= &sysctl_lowmem_reserve_ratio,
1075 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1076 		.mode		= 0644,
1077 		.proc_handler	= &lowmem_reserve_ratio_sysctl_handler,
1078 		.strategy	= &sysctl_intvec,
1079 	},
1080 	{
1081 		.ctl_name	= VM_DROP_PAGECACHE,
1082 		.procname	= "drop_caches",
1083 		.data		= &sysctl_drop_caches,
1084 		.maxlen		= sizeof(int),
1085 		.mode		= 0644,
1086 		.proc_handler	= drop_caches_sysctl_handler,
1087 		.strategy	= &sysctl_intvec,
1088 	},
1089 	{
1090 		.ctl_name	= VM_MIN_FREE_KBYTES,
1091 		.procname	= "min_free_kbytes",
1092 		.data		= &min_free_kbytes,
1093 		.maxlen		= sizeof(min_free_kbytes),
1094 		.mode		= 0644,
1095 		.proc_handler	= &min_free_kbytes_sysctl_handler,
1096 		.strategy	= &sysctl_intvec,
1097 		.extra1		= &zero,
1098 	},
1099 	{
1100 		.ctl_name	= VM_PERCPU_PAGELIST_FRACTION,
1101 		.procname	= "percpu_pagelist_fraction",
1102 		.data		= &percpu_pagelist_fraction,
1103 		.maxlen		= sizeof(percpu_pagelist_fraction),
1104 		.mode		= 0644,
1105 		.proc_handler	= &percpu_pagelist_fraction_sysctl_handler,
1106 		.strategy	= &sysctl_intvec,
1107 		.extra1		= &min_percpu_pagelist_fract,
1108 	},
1109 #ifdef CONFIG_MMU
1110 	{
1111 		.ctl_name	= VM_MAX_MAP_COUNT,
1112 		.procname	= "max_map_count",
1113 		.data		= &sysctl_max_map_count,
1114 		.maxlen		= sizeof(sysctl_max_map_count),
1115 		.mode		= 0644,
1116 		.proc_handler	= &proc_dointvec
1117 	},
1118 #else
1119 	{
1120 		.ctl_name	= CTL_UNNUMBERED,
1121 		.procname	= "nr_trim_pages",
1122 		.data		= &sysctl_nr_trim_pages,
1123 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1124 		.mode		= 0644,
1125 		.proc_handler	= &proc_dointvec_minmax,
1126 		.strategy	= &sysctl_intvec,
1127 		.extra1		= &zero,
1128 	},
1129 #endif
1130 	{
1131 		.ctl_name	= VM_LAPTOP_MODE,
1132 		.procname	= "laptop_mode",
1133 		.data		= &laptop_mode,
1134 		.maxlen		= sizeof(laptop_mode),
1135 		.mode		= 0644,
1136 		.proc_handler	= &proc_dointvec_jiffies,
1137 		.strategy	= &sysctl_jiffies,
1138 	},
1139 	{
1140 		.ctl_name	= VM_BLOCK_DUMP,
1141 		.procname	= "block_dump",
1142 		.data		= &block_dump,
1143 		.maxlen		= sizeof(block_dump),
1144 		.mode		= 0644,
1145 		.proc_handler	= &proc_dointvec,
1146 		.strategy	= &sysctl_intvec,
1147 		.extra1		= &zero,
1148 	},
1149 	{
1150 		.ctl_name	= VM_VFS_CACHE_PRESSURE,
1151 		.procname	= "vfs_cache_pressure",
1152 		.data		= &sysctl_vfs_cache_pressure,
1153 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1154 		.mode		= 0644,
1155 		.proc_handler	= &proc_dointvec,
1156 		.strategy	= &sysctl_intvec,
1157 		.extra1		= &zero,
1158 	},
1159 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1160 	{
1161 		.ctl_name	= VM_LEGACY_VA_LAYOUT,
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 		.strategy	= &sysctl_intvec,
1168 		.extra1		= &zero,
1169 	},
1170 #endif
1171 #ifdef CONFIG_NUMA
1172 	{
1173 		.ctl_name	= VM_ZONE_RECLAIM_MODE,
1174 		.procname	= "zone_reclaim_mode",
1175 		.data		= &zone_reclaim_mode,
1176 		.maxlen		= sizeof(zone_reclaim_mode),
1177 		.mode		= 0644,
1178 		.proc_handler	= &proc_dointvec,
1179 		.strategy	= &sysctl_intvec,
1180 		.extra1		= &zero,
1181 	},
1182 	{
1183 		.ctl_name	= VM_MIN_UNMAPPED,
1184 		.procname	= "min_unmapped_ratio",
1185 		.data		= &sysctl_min_unmapped_ratio,
1186 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1187 		.mode		= 0644,
1188 		.proc_handler	= &sysctl_min_unmapped_ratio_sysctl_handler,
1189 		.strategy	= &sysctl_intvec,
1190 		.extra1		= &zero,
1191 		.extra2		= &one_hundred,
1192 	},
1193 	{
1194 		.ctl_name	= VM_MIN_SLAB,
1195 		.procname	= "min_slab_ratio",
1196 		.data		= &sysctl_min_slab_ratio,
1197 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1198 		.mode		= 0644,
1199 		.proc_handler	= &sysctl_min_slab_ratio_sysctl_handler,
1200 		.strategy	= &sysctl_intvec,
1201 		.extra1		= &zero,
1202 		.extra2		= &one_hundred,
1203 	},
1204 #endif
1205 #ifdef CONFIG_SMP
1206 	{
1207 		.ctl_name	= CTL_UNNUMBERED,
1208 		.procname	= "stat_interval",
1209 		.data		= &sysctl_stat_interval,
1210 		.maxlen		= sizeof(sysctl_stat_interval),
1211 		.mode		= 0644,
1212 		.proc_handler	= &proc_dointvec_jiffies,
1213 		.strategy	= &sysctl_jiffies,
1214 	},
1215 #endif
1216 #ifdef CONFIG_SECURITY
1217 	{
1218 		.ctl_name	= CTL_UNNUMBERED,
1219 		.procname	= "mmap_min_addr",
1220 		.data		= &mmap_min_addr,
1221 		.maxlen         = sizeof(unsigned long),
1222 		.mode		= 0644,
1223 		.proc_handler	= &proc_doulongvec_minmax,
1224 	},
1225 #endif
1226 #ifdef CONFIG_NUMA
1227 	{
1228 		.ctl_name	= CTL_UNNUMBERED,
1229 		.procname	= "numa_zonelist_order",
1230 		.data		= &numa_zonelist_order,
1231 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1232 		.mode		= 0644,
1233 		.proc_handler	= &numa_zonelist_order_handler,
1234 		.strategy	= &sysctl_string,
1235 	},
1236 #endif
1237 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1238    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1239 	{
1240 		.ctl_name	= VM_VDSO_ENABLED,
1241 		.procname	= "vdso_enabled",
1242 		.data		= &vdso_enabled,
1243 		.maxlen		= sizeof(vdso_enabled),
1244 		.mode		= 0644,
1245 		.proc_handler	= &proc_dointvec,
1246 		.strategy	= &sysctl_intvec,
1247 		.extra1		= &zero,
1248 	},
1249 #endif
1250 #ifdef CONFIG_HIGHMEM
1251 	{
1252 		.ctl_name	= CTL_UNNUMBERED,
1253 		.procname	= "highmem_is_dirtyable",
1254 		.data		= &vm_highmem_is_dirtyable,
1255 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1256 		.mode		= 0644,
1257 		.proc_handler	= &proc_dointvec_minmax,
1258 		.strategy	= &sysctl_intvec,
1259 		.extra1		= &zero,
1260 		.extra2		= &one,
1261 	},
1262 #endif
1263 /*
1264  * NOTE: do not add new entries to this table unless you have read
1265  * Documentation/sysctl/ctl_unnumbered.txt
1266  */
1267 	{ .ctl_name = 0 }
1268 };
1269 
1270 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1271 static struct ctl_table binfmt_misc_table[] = {
1272 	{ .ctl_name = 0 }
1273 };
1274 #endif
1275 
1276 static struct ctl_table fs_table[] = {
1277 	{
1278 		.ctl_name	= FS_NRINODE,
1279 		.procname	= "inode-nr",
1280 		.data		= &inodes_stat,
1281 		.maxlen		= 2*sizeof(int),
1282 		.mode		= 0444,
1283 		.proc_handler	= &proc_dointvec,
1284 	},
1285 	{
1286 		.ctl_name	= FS_STATINODE,
1287 		.procname	= "inode-state",
1288 		.data		= &inodes_stat,
1289 		.maxlen		= 7*sizeof(int),
1290 		.mode		= 0444,
1291 		.proc_handler	= &proc_dointvec,
1292 	},
1293 	{
1294 		.procname	= "file-nr",
1295 		.data		= &files_stat,
1296 		.maxlen		= 3*sizeof(int),
1297 		.mode		= 0444,
1298 		.proc_handler	= &proc_nr_files,
1299 	},
1300 	{
1301 		.ctl_name	= FS_MAXFILE,
1302 		.procname	= "file-max",
1303 		.data		= &files_stat.max_files,
1304 		.maxlen		= sizeof(int),
1305 		.mode		= 0644,
1306 		.proc_handler	= &proc_dointvec,
1307 	},
1308 	{
1309 		.ctl_name	= CTL_UNNUMBERED,
1310 		.procname	= "nr_open",
1311 		.data		= &sysctl_nr_open,
1312 		.maxlen		= sizeof(int),
1313 		.mode		= 0644,
1314 		.proc_handler	= &proc_dointvec_minmax,
1315 		.extra1		= &sysctl_nr_open_min,
1316 		.extra2		= &sysctl_nr_open_max,
1317 	},
1318 	{
1319 		.ctl_name	= FS_DENTRY,
1320 		.procname	= "dentry-state",
1321 		.data		= &dentry_stat,
1322 		.maxlen		= 6*sizeof(int),
1323 		.mode		= 0444,
1324 		.proc_handler	= &proc_dointvec,
1325 	},
1326 	{
1327 		.ctl_name	= FS_OVERFLOWUID,
1328 		.procname	= "overflowuid",
1329 		.data		= &fs_overflowuid,
1330 		.maxlen		= sizeof(int),
1331 		.mode		= 0644,
1332 		.proc_handler	= &proc_dointvec_minmax,
1333 		.strategy	= &sysctl_intvec,
1334 		.extra1		= &minolduid,
1335 		.extra2		= &maxolduid,
1336 	},
1337 	{
1338 		.ctl_name	= FS_OVERFLOWGID,
1339 		.procname	= "overflowgid",
1340 		.data		= &fs_overflowgid,
1341 		.maxlen		= sizeof(int),
1342 		.mode		= 0644,
1343 		.proc_handler	= &proc_dointvec_minmax,
1344 		.strategy	= &sysctl_intvec,
1345 		.extra1		= &minolduid,
1346 		.extra2		= &maxolduid,
1347 	},
1348 #ifdef CONFIG_FILE_LOCKING
1349 	{
1350 		.ctl_name	= FS_LEASES,
1351 		.procname	= "leases-enable",
1352 		.data		= &leases_enable,
1353 		.maxlen		= sizeof(int),
1354 		.mode		= 0644,
1355 		.proc_handler	= &proc_dointvec,
1356 	},
1357 #endif
1358 #ifdef CONFIG_DNOTIFY
1359 	{
1360 		.ctl_name	= FS_DIR_NOTIFY,
1361 		.procname	= "dir-notify-enable",
1362 		.data		= &dir_notify_enable,
1363 		.maxlen		= sizeof(int),
1364 		.mode		= 0644,
1365 		.proc_handler	= &proc_dointvec,
1366 	},
1367 #endif
1368 #ifdef CONFIG_MMU
1369 #ifdef CONFIG_FILE_LOCKING
1370 	{
1371 		.ctl_name	= FS_LEASE_TIME,
1372 		.procname	= "lease-break-time",
1373 		.data		= &lease_break_time,
1374 		.maxlen		= sizeof(int),
1375 		.mode		= 0644,
1376 		.proc_handler	= &proc_dointvec_minmax,
1377 		.strategy	= &sysctl_intvec,
1378 		.extra1		= &zero,
1379 		.extra2		= &two,
1380 	},
1381 #endif
1382 #ifdef CONFIG_AIO
1383 	{
1384 		.procname	= "aio-nr",
1385 		.data		= &aio_nr,
1386 		.maxlen		= sizeof(aio_nr),
1387 		.mode		= 0444,
1388 		.proc_handler	= &proc_doulongvec_minmax,
1389 	},
1390 	{
1391 		.procname	= "aio-max-nr",
1392 		.data		= &aio_max_nr,
1393 		.maxlen		= sizeof(aio_max_nr),
1394 		.mode		= 0644,
1395 		.proc_handler	= &proc_doulongvec_minmax,
1396 	},
1397 #endif /* CONFIG_AIO */
1398 #ifdef CONFIG_INOTIFY_USER
1399 	{
1400 		.ctl_name	= FS_INOTIFY,
1401 		.procname	= "inotify",
1402 		.mode		= 0555,
1403 		.child		= inotify_table,
1404 	},
1405 #endif
1406 #ifdef CONFIG_EPOLL
1407 	{
1408 		.procname	= "epoll",
1409 		.mode		= 0555,
1410 		.child		= epoll_table,
1411 	},
1412 #endif
1413 #endif
1414 	{
1415 		.ctl_name	= KERN_SETUID_DUMPABLE,
1416 		.procname	= "suid_dumpable",
1417 		.data		= &suid_dumpable,
1418 		.maxlen		= sizeof(int),
1419 		.mode		= 0644,
1420 		.proc_handler	= &proc_dointvec,
1421 	},
1422 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1423 	{
1424 		.ctl_name	= CTL_UNNUMBERED,
1425 		.procname	= "binfmt_misc",
1426 		.mode		= 0555,
1427 		.child		= binfmt_misc_table,
1428 	},
1429 #endif
1430 /*
1431  * NOTE: do not add new entries to this table unless you have read
1432  * Documentation/sysctl/ctl_unnumbered.txt
1433  */
1434 	{ .ctl_name = 0 }
1435 };
1436 
1437 static struct ctl_table debug_table[] = {
1438 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1439 	{
1440 		.ctl_name	= CTL_UNNUMBERED,
1441 		.procname	= "exception-trace",
1442 		.data		= &show_unhandled_signals,
1443 		.maxlen		= sizeof(int),
1444 		.mode		= 0644,
1445 		.proc_handler	= proc_dointvec
1446 	},
1447 #endif
1448 	{ .ctl_name = 0 }
1449 };
1450 
1451 static struct ctl_table dev_table[] = {
1452 	{ .ctl_name = 0 }
1453 };
1454 
1455 static DEFINE_SPINLOCK(sysctl_lock);
1456 
1457 /* called under sysctl_lock */
1458 static int use_table(struct ctl_table_header *p)
1459 {
1460 	if (unlikely(p->unregistering))
1461 		return 0;
1462 	p->used++;
1463 	return 1;
1464 }
1465 
1466 /* called under sysctl_lock */
1467 static void unuse_table(struct ctl_table_header *p)
1468 {
1469 	if (!--p->used)
1470 		if (unlikely(p->unregistering))
1471 			complete(p->unregistering);
1472 }
1473 
1474 /* called under sysctl_lock, will reacquire if has to wait */
1475 static void start_unregistering(struct ctl_table_header *p)
1476 {
1477 	/*
1478 	 * if p->used is 0, nobody will ever touch that entry again;
1479 	 * we'll eliminate all paths to it before dropping sysctl_lock
1480 	 */
1481 	if (unlikely(p->used)) {
1482 		struct completion wait;
1483 		init_completion(&wait);
1484 		p->unregistering = &wait;
1485 		spin_unlock(&sysctl_lock);
1486 		wait_for_completion(&wait);
1487 		spin_lock(&sysctl_lock);
1488 	} else {
1489 		/* anything non-NULL; we'll never dereference it */
1490 		p->unregistering = ERR_PTR(-EINVAL);
1491 	}
1492 	/*
1493 	 * do not remove from the list until nobody holds it; walking the
1494 	 * list in do_sysctl() relies on that.
1495 	 */
1496 	list_del_init(&p->ctl_entry);
1497 }
1498 
1499 void sysctl_head_get(struct ctl_table_header *head)
1500 {
1501 	spin_lock(&sysctl_lock);
1502 	head->count++;
1503 	spin_unlock(&sysctl_lock);
1504 }
1505 
1506 void sysctl_head_put(struct ctl_table_header *head)
1507 {
1508 	spin_lock(&sysctl_lock);
1509 	if (!--head->count)
1510 		kfree(head);
1511 	spin_unlock(&sysctl_lock);
1512 }
1513 
1514 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1515 {
1516 	if (!head)
1517 		BUG();
1518 	spin_lock(&sysctl_lock);
1519 	if (!use_table(head))
1520 		head = ERR_PTR(-ENOENT);
1521 	spin_unlock(&sysctl_lock);
1522 	return head;
1523 }
1524 
1525 void sysctl_head_finish(struct ctl_table_header *head)
1526 {
1527 	if (!head)
1528 		return;
1529 	spin_lock(&sysctl_lock);
1530 	unuse_table(head);
1531 	spin_unlock(&sysctl_lock);
1532 }
1533 
1534 static struct ctl_table_set *
1535 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1536 {
1537 	struct ctl_table_set *set = &root->default_set;
1538 	if (root->lookup)
1539 		set = root->lookup(root, namespaces);
1540 	return set;
1541 }
1542 
1543 static struct list_head *
1544 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1545 {
1546 	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1547 	return &set->list;
1548 }
1549 
1550 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1551 					    struct ctl_table_header *prev)
1552 {
1553 	struct ctl_table_root *root;
1554 	struct list_head *header_list;
1555 	struct ctl_table_header *head;
1556 	struct list_head *tmp;
1557 
1558 	spin_lock(&sysctl_lock);
1559 	if (prev) {
1560 		head = prev;
1561 		tmp = &prev->ctl_entry;
1562 		unuse_table(prev);
1563 		goto next;
1564 	}
1565 	tmp = &root_table_header.ctl_entry;
1566 	for (;;) {
1567 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1568 
1569 		if (!use_table(head))
1570 			goto next;
1571 		spin_unlock(&sysctl_lock);
1572 		return head;
1573 	next:
1574 		root = head->root;
1575 		tmp = tmp->next;
1576 		header_list = lookup_header_list(root, namespaces);
1577 		if (tmp != header_list)
1578 			continue;
1579 
1580 		do {
1581 			root = list_entry(root->root_list.next,
1582 					struct ctl_table_root, root_list);
1583 			if (root == &sysctl_table_root)
1584 				goto out;
1585 			header_list = lookup_header_list(root, namespaces);
1586 		} while (list_empty(header_list));
1587 		tmp = header_list->next;
1588 	}
1589 out:
1590 	spin_unlock(&sysctl_lock);
1591 	return NULL;
1592 }
1593 
1594 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1595 {
1596 	return __sysctl_head_next(current->nsproxy, prev);
1597 }
1598 
1599 void register_sysctl_root(struct ctl_table_root *root)
1600 {
1601 	spin_lock(&sysctl_lock);
1602 	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1603 	spin_unlock(&sysctl_lock);
1604 }
1605 
1606 #ifdef CONFIG_SYSCTL_SYSCALL
1607 /* Perform the actual read/write of a sysctl table entry. */
1608 static int do_sysctl_strategy(struct ctl_table_root *root,
1609 			struct ctl_table *table,
1610 			void __user *oldval, size_t __user *oldlenp,
1611 			void __user *newval, size_t newlen)
1612 {
1613 	int op = 0, rc;
1614 
1615 	if (oldval)
1616 		op |= MAY_READ;
1617 	if (newval)
1618 		op |= MAY_WRITE;
1619 	if (sysctl_perm(root, table, op))
1620 		return -EPERM;
1621 
1622 	if (table->strategy) {
1623 		rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1624 		if (rc < 0)
1625 			return rc;
1626 		if (rc > 0)
1627 			return 0;
1628 	}
1629 
1630 	/* If there is no strategy routine, or if the strategy returns
1631 	 * zero, proceed with automatic r/w */
1632 	if (table->data && table->maxlen) {
1633 		rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1634 		if (rc < 0)
1635 			return rc;
1636 	}
1637 	return 0;
1638 }
1639 
1640 static int parse_table(int __user *name, int nlen,
1641 		       void __user *oldval, size_t __user *oldlenp,
1642 		       void __user *newval, size_t newlen,
1643 		       struct ctl_table_root *root,
1644 		       struct ctl_table *table)
1645 {
1646 	int n;
1647 repeat:
1648 	if (!nlen)
1649 		return -ENOTDIR;
1650 	if (get_user(n, name))
1651 		return -EFAULT;
1652 	for ( ; table->ctl_name || table->procname; table++) {
1653 		if (!table->ctl_name)
1654 			continue;
1655 		if (n == table->ctl_name) {
1656 			int error;
1657 			if (table->child) {
1658 				if (sysctl_perm(root, table, MAY_EXEC))
1659 					return -EPERM;
1660 				name++;
1661 				nlen--;
1662 				table = table->child;
1663 				goto repeat;
1664 			}
1665 			error = do_sysctl_strategy(root, table,
1666 						   oldval, oldlenp,
1667 						   newval, newlen);
1668 			return error;
1669 		}
1670 	}
1671 	return -ENOTDIR;
1672 }
1673 
1674 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1675 	       void __user *newval, size_t newlen)
1676 {
1677 	struct ctl_table_header *head;
1678 	int error = -ENOTDIR;
1679 
1680 	if (nlen <= 0 || nlen >= CTL_MAXNAME)
1681 		return -ENOTDIR;
1682 	if (oldval) {
1683 		int old_len;
1684 		if (!oldlenp || get_user(old_len, oldlenp))
1685 			return -EFAULT;
1686 	}
1687 
1688 	for (head = sysctl_head_next(NULL); head;
1689 			head = sysctl_head_next(head)) {
1690 		error = parse_table(name, nlen, oldval, oldlenp,
1691 					newval, newlen,
1692 					head->root, head->ctl_table);
1693 		if (error != -ENOTDIR) {
1694 			sysctl_head_finish(head);
1695 			break;
1696 		}
1697 	}
1698 	return error;
1699 }
1700 
1701 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
1702 {
1703 	struct __sysctl_args tmp;
1704 	int error;
1705 
1706 	if (copy_from_user(&tmp, args, sizeof(tmp)))
1707 		return -EFAULT;
1708 
1709 	error = deprecated_sysctl_warning(&tmp);
1710 	if (error)
1711 		goto out;
1712 
1713 	lock_kernel();
1714 	error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1715 			  tmp.newval, tmp.newlen);
1716 	unlock_kernel();
1717 out:
1718 	return error;
1719 }
1720 #endif /* CONFIG_SYSCTL_SYSCALL */
1721 
1722 /*
1723  * sysctl_perm does NOT grant the superuser all rights automatically, because
1724  * some sysctl variables are readonly even to root.
1725  */
1726 
1727 static int test_perm(int mode, int op)
1728 {
1729 	if (!current_euid())
1730 		mode >>= 6;
1731 	else if (in_egroup_p(0))
1732 		mode >>= 3;
1733 	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1734 		return 0;
1735 	return -EACCES;
1736 }
1737 
1738 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1739 {
1740 	int error;
1741 	int mode;
1742 
1743 	error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1744 	if (error)
1745 		return error;
1746 
1747 	if (root->permissions)
1748 		mode = root->permissions(root, current->nsproxy, table);
1749 	else
1750 		mode = table->mode;
1751 
1752 	return test_perm(mode, op);
1753 }
1754 
1755 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1756 {
1757 	for (; table->ctl_name || table->procname; table++) {
1758 		table->parent = parent;
1759 		if (table->child)
1760 			sysctl_set_parent(table, table->child);
1761 	}
1762 }
1763 
1764 static __init int sysctl_init(void)
1765 {
1766 	sysctl_set_parent(NULL, root_table);
1767 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1768 	{
1769 		int err;
1770 		err = sysctl_check_table(current->nsproxy, root_table);
1771 	}
1772 #endif
1773 	return 0;
1774 }
1775 
1776 core_initcall(sysctl_init);
1777 
1778 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1779 				      struct ctl_table *table)
1780 {
1781 	struct ctl_table *p;
1782 	const char *s = branch->procname;
1783 
1784 	/* branch should have named subdirectory as its first element */
1785 	if (!s || !branch->child)
1786 		return NULL;
1787 
1788 	/* ... and nothing else */
1789 	if (branch[1].procname || branch[1].ctl_name)
1790 		return NULL;
1791 
1792 	/* table should contain subdirectory with the same name */
1793 	for (p = table; p->procname || p->ctl_name; p++) {
1794 		if (!p->child)
1795 			continue;
1796 		if (p->procname && strcmp(p->procname, s) == 0)
1797 			return p;
1798 	}
1799 	return NULL;
1800 }
1801 
1802 /* see if attaching q to p would be an improvement */
1803 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1804 {
1805 	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1806 	struct ctl_table *next;
1807 	int is_better = 0;
1808 	int not_in_parent = !p->attached_by;
1809 
1810 	while ((next = is_branch_in(by, to)) != NULL) {
1811 		if (by == q->attached_by)
1812 			is_better = 1;
1813 		if (to == p->attached_by)
1814 			not_in_parent = 1;
1815 		by = by->child;
1816 		to = next->child;
1817 	}
1818 
1819 	if (is_better && not_in_parent) {
1820 		q->attached_by = by;
1821 		q->attached_to = to;
1822 		q->parent = p;
1823 	}
1824 }
1825 
1826 /**
1827  * __register_sysctl_paths - register a sysctl hierarchy
1828  * @root: List of sysctl headers to register on
1829  * @namespaces: Data to compute which lists of sysctl entries are visible
1830  * @path: The path to the directory the sysctl table is in.
1831  * @table: the top-level table structure
1832  *
1833  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1834  * array. A completely 0 filled entry terminates the table.
1835  *
1836  * The members of the &struct ctl_table structure are used as follows:
1837  *
1838  * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1839  *            must be unique within that level of sysctl
1840  *
1841  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1842  *            enter a sysctl file
1843  *
1844  * data - a pointer to data for use by proc_handler
1845  *
1846  * maxlen - the maximum size in bytes of the data
1847  *
1848  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1849  *
1850  * child - a pointer to the child sysctl table if this entry is a directory, or
1851  *         %NULL.
1852  *
1853  * proc_handler - the text handler routine (described below)
1854  *
1855  * strategy - the strategy routine (described below)
1856  *
1857  * de - for internal use by the sysctl routines
1858  *
1859  * extra1, extra2 - extra pointers usable by the proc handler routines
1860  *
1861  * Leaf nodes in the sysctl tree will be represented by a single file
1862  * under /proc; non-leaf nodes will be represented by directories.
1863  *
1864  * sysctl(2) can automatically manage read and write requests through
1865  * the sysctl table.  The data and maxlen fields of the ctl_table
1866  * struct enable minimal validation of the values being written to be
1867  * performed, and the mode field allows minimal authentication.
1868  *
1869  * More sophisticated management can be enabled by the provision of a
1870  * strategy routine with the table entry.  This will be called before
1871  * any automatic read or write of the data is performed.
1872  *
1873  * The strategy routine may return
1874  *
1875  * < 0 - Error occurred (error is passed to user process)
1876  *
1877  * 0   - OK - proceed with automatic read or write.
1878  *
1879  * > 0 - OK - read or write has been done by the strategy routine, so
1880  *       return immediately.
1881  *
1882  * There must be a proc_handler routine for any terminal nodes
1883  * mirrored under /proc/sys (non-terminals are handled by a built-in
1884  * directory handler).  Several default handlers are available to
1885  * cover common cases -
1886  *
1887  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1888  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1889  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1890  *
1891  * It is the handler's job to read the input buffer from user memory
1892  * and process it. The handler should return 0 on success.
1893  *
1894  * This routine returns %NULL on a failure to register, and a pointer
1895  * to the table header on success.
1896  */
1897 struct ctl_table_header *__register_sysctl_paths(
1898 	struct ctl_table_root *root,
1899 	struct nsproxy *namespaces,
1900 	const struct ctl_path *path, struct ctl_table *table)
1901 {
1902 	struct ctl_table_header *header;
1903 	struct ctl_table *new, **prevp;
1904 	unsigned int n, npath;
1905 	struct ctl_table_set *set;
1906 
1907 	/* Count the path components */
1908 	for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1909 		;
1910 
1911 	/*
1912 	 * For each path component, allocate a 2-element ctl_table array.
1913 	 * The first array element will be filled with the sysctl entry
1914 	 * for this, the second will be the sentinel (ctl_name == 0).
1915 	 *
1916 	 * We allocate everything in one go so that we don't have to
1917 	 * worry about freeing additional memory in unregister_sysctl_table.
1918 	 */
1919 	header = kzalloc(sizeof(struct ctl_table_header) +
1920 			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1921 	if (!header)
1922 		return NULL;
1923 
1924 	new = (struct ctl_table *) (header + 1);
1925 
1926 	/* Now connect the dots */
1927 	prevp = &header->ctl_table;
1928 	for (n = 0; n < npath; ++n, ++path) {
1929 		/* Copy the procname */
1930 		new->procname = path->procname;
1931 		new->ctl_name = path->ctl_name;
1932 		new->mode     = 0555;
1933 
1934 		*prevp = new;
1935 		prevp = &new->child;
1936 
1937 		new += 2;
1938 	}
1939 	*prevp = table;
1940 	header->ctl_table_arg = table;
1941 
1942 	INIT_LIST_HEAD(&header->ctl_entry);
1943 	header->used = 0;
1944 	header->unregistering = NULL;
1945 	header->root = root;
1946 	sysctl_set_parent(NULL, header->ctl_table);
1947 	header->count = 1;
1948 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1949 	if (sysctl_check_table(namespaces, header->ctl_table)) {
1950 		kfree(header);
1951 		return NULL;
1952 	}
1953 #endif
1954 	spin_lock(&sysctl_lock);
1955 	header->set = lookup_header_set(root, namespaces);
1956 	header->attached_by = header->ctl_table;
1957 	header->attached_to = root_table;
1958 	header->parent = &root_table_header;
1959 	for (set = header->set; set; set = set->parent) {
1960 		struct ctl_table_header *p;
1961 		list_for_each_entry(p, &set->list, ctl_entry) {
1962 			if (p->unregistering)
1963 				continue;
1964 			try_attach(p, header);
1965 		}
1966 	}
1967 	header->parent->count++;
1968 	list_add_tail(&header->ctl_entry, &header->set->list);
1969 	spin_unlock(&sysctl_lock);
1970 
1971 	return header;
1972 }
1973 
1974 /**
1975  * register_sysctl_table_path - register a sysctl table hierarchy
1976  * @path: The path to the directory the sysctl table is in.
1977  * @table: the top-level table structure
1978  *
1979  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1980  * array. A completely 0 filled entry terminates the table.
1981  *
1982  * See __register_sysctl_paths for more details.
1983  */
1984 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1985 						struct ctl_table *table)
1986 {
1987 	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1988 					path, table);
1989 }
1990 
1991 /**
1992  * register_sysctl_table - register a sysctl table hierarchy
1993  * @table: the top-level table structure
1994  *
1995  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1996  * array. A completely 0 filled entry terminates the table.
1997  *
1998  * See register_sysctl_paths for more details.
1999  */
2000 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
2001 {
2002 	static const struct ctl_path null_path[] = { {} };
2003 
2004 	return register_sysctl_paths(null_path, table);
2005 }
2006 
2007 /**
2008  * unregister_sysctl_table - unregister a sysctl table hierarchy
2009  * @header: the header returned from register_sysctl_table
2010  *
2011  * Unregisters the sysctl table and all children. proc entries may not
2012  * actually be removed until they are no longer used by anyone.
2013  */
2014 void unregister_sysctl_table(struct ctl_table_header * header)
2015 {
2016 	might_sleep();
2017 
2018 	if (header == NULL)
2019 		return;
2020 
2021 	spin_lock(&sysctl_lock);
2022 	start_unregistering(header);
2023 	if (!--header->parent->count) {
2024 		WARN_ON(1);
2025 		kfree(header->parent);
2026 	}
2027 	if (!--header->count)
2028 		kfree(header);
2029 	spin_unlock(&sysctl_lock);
2030 }
2031 
2032 int sysctl_is_seen(struct ctl_table_header *p)
2033 {
2034 	struct ctl_table_set *set = p->set;
2035 	int res;
2036 	spin_lock(&sysctl_lock);
2037 	if (p->unregistering)
2038 		res = 0;
2039 	else if (!set->is_seen)
2040 		res = 1;
2041 	else
2042 		res = set->is_seen(set);
2043 	spin_unlock(&sysctl_lock);
2044 	return res;
2045 }
2046 
2047 void setup_sysctl_set(struct ctl_table_set *p,
2048 	struct ctl_table_set *parent,
2049 	int (*is_seen)(struct ctl_table_set *))
2050 {
2051 	INIT_LIST_HEAD(&p->list);
2052 	p->parent = parent ? parent : &sysctl_table_root.default_set;
2053 	p->is_seen = is_seen;
2054 }
2055 
2056 #else /* !CONFIG_SYSCTL */
2057 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2058 {
2059 	return NULL;
2060 }
2061 
2062 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2063 						    struct ctl_table *table)
2064 {
2065 	return NULL;
2066 }
2067 
2068 void unregister_sysctl_table(struct ctl_table_header * table)
2069 {
2070 }
2071 
2072 void setup_sysctl_set(struct ctl_table_set *p,
2073 	struct ctl_table_set *parent,
2074 	int (*is_seen)(struct ctl_table_set *))
2075 {
2076 }
2077 
2078 void sysctl_head_put(struct ctl_table_header *head)
2079 {
2080 }
2081 
2082 #endif /* CONFIG_SYSCTL */
2083 
2084 /*
2085  * /proc/sys support
2086  */
2087 
2088 #ifdef CONFIG_PROC_SYSCTL
2089 
2090 static int _proc_do_string(void* data, int maxlen, int write,
2091 			   struct file *filp, void __user *buffer,
2092 			   size_t *lenp, loff_t *ppos)
2093 {
2094 	size_t len;
2095 	char __user *p;
2096 	char c;
2097 
2098 	if (!data || !maxlen || !*lenp) {
2099 		*lenp = 0;
2100 		return 0;
2101 	}
2102 
2103 	if (write) {
2104 		len = 0;
2105 		p = buffer;
2106 		while (len < *lenp) {
2107 			if (get_user(c, p++))
2108 				return -EFAULT;
2109 			if (c == 0 || c == '\n')
2110 				break;
2111 			len++;
2112 		}
2113 		if (len >= maxlen)
2114 			len = maxlen-1;
2115 		if(copy_from_user(data, buffer, len))
2116 			return -EFAULT;
2117 		((char *) data)[len] = 0;
2118 		*ppos += *lenp;
2119 	} else {
2120 		len = strlen(data);
2121 		if (len > maxlen)
2122 			len = maxlen;
2123 
2124 		if (*ppos > len) {
2125 			*lenp = 0;
2126 			return 0;
2127 		}
2128 
2129 		data += *ppos;
2130 		len  -= *ppos;
2131 
2132 		if (len > *lenp)
2133 			len = *lenp;
2134 		if (len)
2135 			if(copy_to_user(buffer, data, len))
2136 				return -EFAULT;
2137 		if (len < *lenp) {
2138 			if(put_user('\n', ((char __user *) buffer) + len))
2139 				return -EFAULT;
2140 			len++;
2141 		}
2142 		*lenp = len;
2143 		*ppos += len;
2144 	}
2145 	return 0;
2146 }
2147 
2148 /**
2149  * proc_dostring - read a string sysctl
2150  * @table: the sysctl table
2151  * @write: %TRUE if this is a write to the sysctl file
2152  * @filp: the file structure
2153  * @buffer: the user buffer
2154  * @lenp: the size of the user buffer
2155  * @ppos: file position
2156  *
2157  * Reads/writes a string from/to the user buffer. If the kernel
2158  * buffer provided is not large enough to hold the string, the
2159  * string is truncated. The copied string is %NULL-terminated.
2160  * If the string is being read by the user process, it is copied
2161  * and a newline '\n' is added. It is truncated if the buffer is
2162  * not large enough.
2163  *
2164  * Returns 0 on success.
2165  */
2166 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2167 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2168 {
2169 	return _proc_do_string(table->data, table->maxlen, write, filp,
2170 			       buffer, lenp, ppos);
2171 }
2172 
2173 
2174 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2175 				 int *valp,
2176 				 int write, void *data)
2177 {
2178 	if (write) {
2179 		*valp = *negp ? -*lvalp : *lvalp;
2180 	} else {
2181 		int val = *valp;
2182 		if (val < 0) {
2183 			*negp = -1;
2184 			*lvalp = (unsigned long)-val;
2185 		} else {
2186 			*negp = 0;
2187 			*lvalp = (unsigned long)val;
2188 		}
2189 	}
2190 	return 0;
2191 }
2192 
2193 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2194 		  int write, struct file *filp, void __user *buffer,
2195 		  size_t *lenp, loff_t *ppos,
2196 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2197 			      int write, void *data),
2198 		  void *data)
2199 {
2200 #define TMPBUFLEN 21
2201 	int *i, vleft, first=1, neg, val;
2202 	unsigned long lval;
2203 	size_t left, len;
2204 
2205 	char buf[TMPBUFLEN], *p;
2206 	char __user *s = buffer;
2207 
2208 	if (!tbl_data || !table->maxlen || !*lenp ||
2209 	    (*ppos && !write)) {
2210 		*lenp = 0;
2211 		return 0;
2212 	}
2213 
2214 	i = (int *) tbl_data;
2215 	vleft = table->maxlen / sizeof(*i);
2216 	left = *lenp;
2217 
2218 	if (!conv)
2219 		conv = do_proc_dointvec_conv;
2220 
2221 	for (; left && vleft--; i++, first=0) {
2222 		if (write) {
2223 			while (left) {
2224 				char c;
2225 				if (get_user(c, s))
2226 					return -EFAULT;
2227 				if (!isspace(c))
2228 					break;
2229 				left--;
2230 				s++;
2231 			}
2232 			if (!left)
2233 				break;
2234 			neg = 0;
2235 			len = left;
2236 			if (len > sizeof(buf) - 1)
2237 				len = sizeof(buf) - 1;
2238 			if (copy_from_user(buf, s, len))
2239 				return -EFAULT;
2240 			buf[len] = 0;
2241 			p = buf;
2242 			if (*p == '-' && left > 1) {
2243 				neg = 1;
2244 				p++;
2245 			}
2246 			if (*p < '0' || *p > '9')
2247 				break;
2248 
2249 			lval = simple_strtoul(p, &p, 0);
2250 
2251 			len = p-buf;
2252 			if ((len < left) && *p && !isspace(*p))
2253 				break;
2254 			if (neg)
2255 				val = -val;
2256 			s += len;
2257 			left -= len;
2258 
2259 			if (conv(&neg, &lval, i, 1, data))
2260 				break;
2261 		} else {
2262 			p = buf;
2263 			if (!first)
2264 				*p++ = '\t';
2265 
2266 			if (conv(&neg, &lval, i, 0, data))
2267 				break;
2268 
2269 			sprintf(p, "%s%lu", neg ? "-" : "", lval);
2270 			len = strlen(buf);
2271 			if (len > left)
2272 				len = left;
2273 			if(copy_to_user(s, buf, len))
2274 				return -EFAULT;
2275 			left -= len;
2276 			s += len;
2277 		}
2278 	}
2279 
2280 	if (!write && !first && left) {
2281 		if(put_user('\n', s))
2282 			return -EFAULT;
2283 		left--, s++;
2284 	}
2285 	if (write) {
2286 		while (left) {
2287 			char c;
2288 			if (get_user(c, s++))
2289 				return -EFAULT;
2290 			if (!isspace(c))
2291 				break;
2292 			left--;
2293 		}
2294 	}
2295 	if (write && first)
2296 		return -EINVAL;
2297 	*lenp -= left;
2298 	*ppos += *lenp;
2299 	return 0;
2300 #undef TMPBUFLEN
2301 }
2302 
2303 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2304 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2305 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2306 			      int write, void *data),
2307 		  void *data)
2308 {
2309 	return __do_proc_dointvec(table->data, table, write, filp,
2310 			buffer, lenp, ppos, conv, data);
2311 }
2312 
2313 /**
2314  * proc_dointvec - read a vector of integers
2315  * @table: the sysctl table
2316  * @write: %TRUE if this is a write to the sysctl file
2317  * @filp: the file structure
2318  * @buffer: the user buffer
2319  * @lenp: the size of the user buffer
2320  * @ppos: file position
2321  *
2322  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2323  * values from/to the user buffer, treated as an ASCII string.
2324  *
2325  * Returns 0 on success.
2326  */
2327 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2328 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2329 {
2330     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2331 		    	    NULL,NULL);
2332 }
2333 
2334 /*
2335  * Taint values can only be increased
2336  * This means we can safely use a temporary.
2337  */
2338 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
2339 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2340 {
2341 	struct ctl_table t;
2342 	unsigned long tmptaint = get_taint();
2343 	int err;
2344 
2345 	if (write && !capable(CAP_SYS_ADMIN))
2346 		return -EPERM;
2347 
2348 	t = *table;
2349 	t.data = &tmptaint;
2350 	err = proc_doulongvec_minmax(&t, write, filp, buffer, lenp, ppos);
2351 	if (err < 0)
2352 		return err;
2353 
2354 	if (write) {
2355 		/*
2356 		 * Poor man's atomic or. Not worth adding a primitive
2357 		 * to everyone's atomic.h for this
2358 		 */
2359 		int i;
2360 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2361 			if ((tmptaint >> i) & 1)
2362 				add_taint(i);
2363 		}
2364 	}
2365 
2366 	return err;
2367 }
2368 
2369 struct do_proc_dointvec_minmax_conv_param {
2370 	int *min;
2371 	int *max;
2372 };
2373 
2374 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2375 					int *valp,
2376 					int write, void *data)
2377 {
2378 	struct do_proc_dointvec_minmax_conv_param *param = data;
2379 	if (write) {
2380 		int val = *negp ? -*lvalp : *lvalp;
2381 		if ((param->min && *param->min > val) ||
2382 		    (param->max && *param->max < val))
2383 			return -EINVAL;
2384 		*valp = val;
2385 	} else {
2386 		int val = *valp;
2387 		if (val < 0) {
2388 			*negp = -1;
2389 			*lvalp = (unsigned long)-val;
2390 		} else {
2391 			*negp = 0;
2392 			*lvalp = (unsigned long)val;
2393 		}
2394 	}
2395 	return 0;
2396 }
2397 
2398 /**
2399  * proc_dointvec_minmax - read a vector of integers with min/max values
2400  * @table: the sysctl table
2401  * @write: %TRUE if this is a write to the sysctl file
2402  * @filp: the file structure
2403  * @buffer: the user buffer
2404  * @lenp: the size of the user buffer
2405  * @ppos: file position
2406  *
2407  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2408  * values from/to the user buffer, treated as an ASCII string.
2409  *
2410  * This routine will ensure the values are within the range specified by
2411  * table->extra1 (min) and table->extra2 (max).
2412  *
2413  * Returns 0 on success.
2414  */
2415 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2416 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2417 {
2418 	struct do_proc_dointvec_minmax_conv_param param = {
2419 		.min = (int *) table->extra1,
2420 		.max = (int *) table->extra2,
2421 	};
2422 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2423 				do_proc_dointvec_minmax_conv, &param);
2424 }
2425 
2426 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2427 				     struct file *filp,
2428 				     void __user *buffer,
2429 				     size_t *lenp, loff_t *ppos,
2430 				     unsigned long convmul,
2431 				     unsigned long convdiv)
2432 {
2433 #define TMPBUFLEN 21
2434 	unsigned long *i, *min, *max, val;
2435 	int vleft, first=1, neg;
2436 	size_t len, left;
2437 	char buf[TMPBUFLEN], *p;
2438 	char __user *s = buffer;
2439 
2440 	if (!data || !table->maxlen || !*lenp ||
2441 	    (*ppos && !write)) {
2442 		*lenp = 0;
2443 		return 0;
2444 	}
2445 
2446 	i = (unsigned long *) data;
2447 	min = (unsigned long *) table->extra1;
2448 	max = (unsigned long *) table->extra2;
2449 	vleft = table->maxlen / sizeof(unsigned long);
2450 	left = *lenp;
2451 
2452 	for (; left && vleft--; i++, min++, max++, first=0) {
2453 		if (write) {
2454 			while (left) {
2455 				char c;
2456 				if (get_user(c, s))
2457 					return -EFAULT;
2458 				if (!isspace(c))
2459 					break;
2460 				left--;
2461 				s++;
2462 			}
2463 			if (!left)
2464 				break;
2465 			neg = 0;
2466 			len = left;
2467 			if (len > TMPBUFLEN-1)
2468 				len = TMPBUFLEN-1;
2469 			if (copy_from_user(buf, s, len))
2470 				return -EFAULT;
2471 			buf[len] = 0;
2472 			p = buf;
2473 			if (*p == '-' && left > 1) {
2474 				neg = 1;
2475 				p++;
2476 			}
2477 			if (*p < '0' || *p > '9')
2478 				break;
2479 			val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2480 			len = p-buf;
2481 			if ((len < left) && *p && !isspace(*p))
2482 				break;
2483 			if (neg)
2484 				val = -val;
2485 			s += len;
2486 			left -= len;
2487 
2488 			if(neg)
2489 				continue;
2490 			if ((min && val < *min) || (max && val > *max))
2491 				continue;
2492 			*i = val;
2493 		} else {
2494 			p = buf;
2495 			if (!first)
2496 				*p++ = '\t';
2497 			sprintf(p, "%lu", convdiv * (*i) / convmul);
2498 			len = strlen(buf);
2499 			if (len > left)
2500 				len = left;
2501 			if(copy_to_user(s, buf, len))
2502 				return -EFAULT;
2503 			left -= len;
2504 			s += len;
2505 		}
2506 	}
2507 
2508 	if (!write && !first && left) {
2509 		if(put_user('\n', s))
2510 			return -EFAULT;
2511 		left--, s++;
2512 	}
2513 	if (write) {
2514 		while (left) {
2515 			char c;
2516 			if (get_user(c, s++))
2517 				return -EFAULT;
2518 			if (!isspace(c))
2519 				break;
2520 			left--;
2521 		}
2522 	}
2523 	if (write && first)
2524 		return -EINVAL;
2525 	*lenp -= left;
2526 	*ppos += *lenp;
2527 	return 0;
2528 #undef TMPBUFLEN
2529 }
2530 
2531 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2532 				     struct file *filp,
2533 				     void __user *buffer,
2534 				     size_t *lenp, loff_t *ppos,
2535 				     unsigned long convmul,
2536 				     unsigned long convdiv)
2537 {
2538 	return __do_proc_doulongvec_minmax(table->data, table, write,
2539 			filp, buffer, lenp, ppos, convmul, convdiv);
2540 }
2541 
2542 /**
2543  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2544  * @table: the sysctl table
2545  * @write: %TRUE if this is a write to the sysctl file
2546  * @filp: the file structure
2547  * @buffer: the user buffer
2548  * @lenp: the size of the user buffer
2549  * @ppos: file position
2550  *
2551  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2552  * values from/to the user buffer, treated as an ASCII string.
2553  *
2554  * This routine will ensure the values are within the range specified by
2555  * table->extra1 (min) and table->extra2 (max).
2556  *
2557  * Returns 0 on success.
2558  */
2559 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2560 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2561 {
2562     return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2563 }
2564 
2565 /**
2566  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2567  * @table: the sysctl table
2568  * @write: %TRUE if this is a write to the sysctl file
2569  * @filp: the file structure
2570  * @buffer: the user buffer
2571  * @lenp: the size of the user buffer
2572  * @ppos: file position
2573  *
2574  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2575  * values from/to the user buffer, treated as an ASCII string. The values
2576  * are treated as milliseconds, and converted to jiffies when they are stored.
2577  *
2578  * This routine will ensure the values are within the range specified by
2579  * table->extra1 (min) and table->extra2 (max).
2580  *
2581  * Returns 0 on success.
2582  */
2583 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2584 				      struct file *filp,
2585 				      void __user *buffer,
2586 				      size_t *lenp, loff_t *ppos)
2587 {
2588     return do_proc_doulongvec_minmax(table, write, filp, buffer,
2589 				     lenp, ppos, HZ, 1000l);
2590 }
2591 
2592 
2593 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2594 					 int *valp,
2595 					 int write, void *data)
2596 {
2597 	if (write) {
2598 		if (*lvalp > LONG_MAX / HZ)
2599 			return 1;
2600 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2601 	} else {
2602 		int val = *valp;
2603 		unsigned long lval;
2604 		if (val < 0) {
2605 			*negp = -1;
2606 			lval = (unsigned long)-val;
2607 		} else {
2608 			*negp = 0;
2609 			lval = (unsigned long)val;
2610 		}
2611 		*lvalp = lval / HZ;
2612 	}
2613 	return 0;
2614 }
2615 
2616 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2617 						int *valp,
2618 						int write, void *data)
2619 {
2620 	if (write) {
2621 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2622 			return 1;
2623 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2624 	} else {
2625 		int val = *valp;
2626 		unsigned long lval;
2627 		if (val < 0) {
2628 			*negp = -1;
2629 			lval = (unsigned long)-val;
2630 		} else {
2631 			*negp = 0;
2632 			lval = (unsigned long)val;
2633 		}
2634 		*lvalp = jiffies_to_clock_t(lval);
2635 	}
2636 	return 0;
2637 }
2638 
2639 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2640 					    int *valp,
2641 					    int write, void *data)
2642 {
2643 	if (write) {
2644 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2645 	} else {
2646 		int val = *valp;
2647 		unsigned long lval;
2648 		if (val < 0) {
2649 			*negp = -1;
2650 			lval = (unsigned long)-val;
2651 		} else {
2652 			*negp = 0;
2653 			lval = (unsigned long)val;
2654 		}
2655 		*lvalp = jiffies_to_msecs(lval);
2656 	}
2657 	return 0;
2658 }
2659 
2660 /**
2661  * proc_dointvec_jiffies - read a vector of integers as seconds
2662  * @table: the sysctl table
2663  * @write: %TRUE if this is a write to the sysctl file
2664  * @filp: the file structure
2665  * @buffer: the user buffer
2666  * @lenp: the size of the user buffer
2667  * @ppos: file position
2668  *
2669  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2670  * values from/to the user buffer, treated as an ASCII string.
2671  * The values read are assumed to be in seconds, and are converted into
2672  * jiffies.
2673  *
2674  * Returns 0 on success.
2675  */
2676 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2677 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2678 {
2679     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2680 		    	    do_proc_dointvec_jiffies_conv,NULL);
2681 }
2682 
2683 /**
2684  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2685  * @table: the sysctl table
2686  * @write: %TRUE if this is a write to the sysctl file
2687  * @filp: the file structure
2688  * @buffer: the user buffer
2689  * @lenp: the size of the user buffer
2690  * @ppos: pointer to the file position
2691  *
2692  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2693  * values from/to the user buffer, treated as an ASCII string.
2694  * The values read are assumed to be in 1/USER_HZ seconds, and
2695  * are converted into jiffies.
2696  *
2697  * Returns 0 on success.
2698  */
2699 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2700 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2701 {
2702     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2703 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2704 }
2705 
2706 /**
2707  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2708  * @table: the sysctl table
2709  * @write: %TRUE if this is a write to the sysctl file
2710  * @filp: the file structure
2711  * @buffer: the user buffer
2712  * @lenp: the size of the user buffer
2713  * @ppos: file position
2714  * @ppos: the current position in the file
2715  *
2716  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2717  * values from/to the user buffer, treated as an ASCII string.
2718  * The values read are assumed to be in 1/1000 seconds, and
2719  * are converted into jiffies.
2720  *
2721  * Returns 0 on success.
2722  */
2723 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2724 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2725 {
2726 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2727 				do_proc_dointvec_ms_jiffies_conv, NULL);
2728 }
2729 
2730 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2731 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2732 {
2733 	struct pid *new_pid;
2734 	pid_t tmp;
2735 	int r;
2736 
2737 	tmp = pid_vnr(cad_pid);
2738 
2739 	r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2740 			       lenp, ppos, NULL, NULL);
2741 	if (r || !write)
2742 		return r;
2743 
2744 	new_pid = find_get_pid(tmp);
2745 	if (!new_pid)
2746 		return -ESRCH;
2747 
2748 	put_pid(xchg(&cad_pid, new_pid));
2749 	return 0;
2750 }
2751 
2752 #else /* CONFIG_PROC_FS */
2753 
2754 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2755 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2756 {
2757 	return -ENOSYS;
2758 }
2759 
2760 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2761 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2762 {
2763 	return -ENOSYS;
2764 }
2765 
2766 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2767 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2768 {
2769 	return -ENOSYS;
2770 }
2771 
2772 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2773 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2774 {
2775 	return -ENOSYS;
2776 }
2777 
2778 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2779 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2780 {
2781 	return -ENOSYS;
2782 }
2783 
2784 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2785 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2786 {
2787 	return -ENOSYS;
2788 }
2789 
2790 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2791 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2792 {
2793 	return -ENOSYS;
2794 }
2795 
2796 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2797 				      struct file *filp,
2798 				      void __user *buffer,
2799 				      size_t *lenp, loff_t *ppos)
2800 {
2801     return -ENOSYS;
2802 }
2803 
2804 
2805 #endif /* CONFIG_PROC_FS */
2806 
2807 
2808 #ifdef CONFIG_SYSCTL_SYSCALL
2809 /*
2810  * General sysctl support routines
2811  */
2812 
2813 /* The generic sysctl data routine (used if no strategy routine supplied) */
2814 int sysctl_data(struct ctl_table *table,
2815 		void __user *oldval, size_t __user *oldlenp,
2816 		void __user *newval, size_t newlen)
2817 {
2818 	size_t len;
2819 
2820 	/* Get out of I don't have a variable */
2821 	if (!table->data || !table->maxlen)
2822 		return -ENOTDIR;
2823 
2824 	if (oldval && oldlenp) {
2825 		if (get_user(len, oldlenp))
2826 			return -EFAULT;
2827 		if (len) {
2828 			if (len > table->maxlen)
2829 				len = table->maxlen;
2830 			if (copy_to_user(oldval, table->data, len))
2831 				return -EFAULT;
2832 			if (put_user(len, oldlenp))
2833 				return -EFAULT;
2834 		}
2835 	}
2836 
2837 	if (newval && newlen) {
2838 		if (newlen > table->maxlen)
2839 			newlen = table->maxlen;
2840 
2841 		if (copy_from_user(table->data, newval, newlen))
2842 			return -EFAULT;
2843 	}
2844 	return 1;
2845 }
2846 
2847 /* The generic string strategy routine: */
2848 int sysctl_string(struct ctl_table *table,
2849 		  void __user *oldval, size_t __user *oldlenp,
2850 		  void __user *newval, size_t newlen)
2851 {
2852 	if (!table->data || !table->maxlen)
2853 		return -ENOTDIR;
2854 
2855 	if (oldval && oldlenp) {
2856 		size_t bufsize;
2857 		if (get_user(bufsize, oldlenp))
2858 			return -EFAULT;
2859 		if (bufsize) {
2860 			size_t len = strlen(table->data), copied;
2861 
2862 			/* This shouldn't trigger for a well-formed sysctl */
2863 			if (len > table->maxlen)
2864 				len = table->maxlen;
2865 
2866 			/* Copy up to a max of bufsize-1 bytes of the string */
2867 			copied = (len >= bufsize) ? bufsize - 1 : len;
2868 
2869 			if (copy_to_user(oldval, table->data, copied) ||
2870 			    put_user(0, (char __user *)(oldval + copied)))
2871 				return -EFAULT;
2872 			if (put_user(len, oldlenp))
2873 				return -EFAULT;
2874 		}
2875 	}
2876 	if (newval && newlen) {
2877 		size_t len = newlen;
2878 		if (len > table->maxlen)
2879 			len = table->maxlen;
2880 		if(copy_from_user(table->data, newval, len))
2881 			return -EFAULT;
2882 		if (len == table->maxlen)
2883 			len--;
2884 		((char *) table->data)[len] = 0;
2885 	}
2886 	return 1;
2887 }
2888 
2889 /*
2890  * This function makes sure that all of the integers in the vector
2891  * are between the minimum and maximum values given in the arrays
2892  * table->extra1 and table->extra2, respectively.
2893  */
2894 int sysctl_intvec(struct ctl_table *table,
2895 		void __user *oldval, size_t __user *oldlenp,
2896 		void __user *newval, size_t newlen)
2897 {
2898 
2899 	if (newval && newlen) {
2900 		int __user *vec = (int __user *) newval;
2901 		int *min = (int *) table->extra1;
2902 		int *max = (int *) table->extra2;
2903 		size_t length;
2904 		int i;
2905 
2906 		if (newlen % sizeof(int) != 0)
2907 			return -EINVAL;
2908 
2909 		if (!table->extra1 && !table->extra2)
2910 			return 0;
2911 
2912 		if (newlen > table->maxlen)
2913 			newlen = table->maxlen;
2914 		length = newlen / sizeof(int);
2915 
2916 		for (i = 0; i < length; i++) {
2917 			int value;
2918 			if (get_user(value, vec + i))
2919 				return -EFAULT;
2920 			if (min && value < min[i])
2921 				return -EINVAL;
2922 			if (max && value > max[i])
2923 				return -EINVAL;
2924 		}
2925 	}
2926 	return 0;
2927 }
2928 
2929 /* Strategy function to convert jiffies to seconds */
2930 int sysctl_jiffies(struct ctl_table *table,
2931 		void __user *oldval, size_t __user *oldlenp,
2932 		void __user *newval, size_t newlen)
2933 {
2934 	if (oldval && oldlenp) {
2935 		size_t olen;
2936 
2937 		if (get_user(olen, oldlenp))
2938 			return -EFAULT;
2939 		if (olen) {
2940 			int val;
2941 
2942 			if (olen < sizeof(int))
2943 				return -EINVAL;
2944 
2945 			val = *(int *)(table->data) / HZ;
2946 			if (put_user(val, (int __user *)oldval))
2947 				return -EFAULT;
2948 			if (put_user(sizeof(int), oldlenp))
2949 				return -EFAULT;
2950 		}
2951 	}
2952 	if (newval && newlen) {
2953 		int new;
2954 		if (newlen != sizeof(int))
2955 			return -EINVAL;
2956 		if (get_user(new, (int __user *)newval))
2957 			return -EFAULT;
2958 		*(int *)(table->data) = new*HZ;
2959 	}
2960 	return 1;
2961 }
2962 
2963 /* Strategy function to convert jiffies to seconds */
2964 int sysctl_ms_jiffies(struct ctl_table *table,
2965 		void __user *oldval, size_t __user *oldlenp,
2966 		void __user *newval, size_t newlen)
2967 {
2968 	if (oldval && oldlenp) {
2969 		size_t olen;
2970 
2971 		if (get_user(olen, oldlenp))
2972 			return -EFAULT;
2973 		if (olen) {
2974 			int val;
2975 
2976 			if (olen < sizeof(int))
2977 				return -EINVAL;
2978 
2979 			val = jiffies_to_msecs(*(int *)(table->data));
2980 			if (put_user(val, (int __user *)oldval))
2981 				return -EFAULT;
2982 			if (put_user(sizeof(int), oldlenp))
2983 				return -EFAULT;
2984 		}
2985 	}
2986 	if (newval && newlen) {
2987 		int new;
2988 		if (newlen != sizeof(int))
2989 			return -EINVAL;
2990 		if (get_user(new, (int __user *)newval))
2991 			return -EFAULT;
2992 		*(int *)(table->data) = msecs_to_jiffies(new);
2993 	}
2994 	return 1;
2995 }
2996 
2997 
2998 
2999 #else /* CONFIG_SYSCTL_SYSCALL */
3000 
3001 
3002 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
3003 {
3004 	struct __sysctl_args tmp;
3005 	int error;
3006 
3007 	if (copy_from_user(&tmp, args, sizeof(tmp)))
3008 		return -EFAULT;
3009 
3010 	error = deprecated_sysctl_warning(&tmp);
3011 
3012 	/* If no error reading the parameters then just -ENOSYS ... */
3013 	if (!error)
3014 		error = -ENOSYS;
3015 
3016 	return error;
3017 }
3018 
3019 int sysctl_data(struct ctl_table *table,
3020 		  void __user *oldval, size_t __user *oldlenp,
3021 		  void __user *newval, size_t newlen)
3022 {
3023 	return -ENOSYS;
3024 }
3025 
3026 int sysctl_string(struct ctl_table *table,
3027 		  void __user *oldval, size_t __user *oldlenp,
3028 		  void __user *newval, size_t newlen)
3029 {
3030 	return -ENOSYS;
3031 }
3032 
3033 int sysctl_intvec(struct ctl_table *table,
3034 		void __user *oldval, size_t __user *oldlenp,
3035 		void __user *newval, size_t newlen)
3036 {
3037 	return -ENOSYS;
3038 }
3039 
3040 int sysctl_jiffies(struct ctl_table *table,
3041 		void __user *oldval, size_t __user *oldlenp,
3042 		void __user *newval, size_t newlen)
3043 {
3044 	return -ENOSYS;
3045 }
3046 
3047 int sysctl_ms_jiffies(struct ctl_table *table,
3048 		void __user *oldval, size_t __user *oldlenp,
3049 		void __user *newval, size_t newlen)
3050 {
3051 	return -ENOSYS;
3052 }
3053 
3054 #endif /* CONFIG_SYSCTL_SYSCALL */
3055 
3056 static int deprecated_sysctl_warning(struct __sysctl_args *args)
3057 {
3058 	static int msg_count;
3059 	int name[CTL_MAXNAME];
3060 	int i;
3061 
3062 	/* Check args->nlen. */
3063 	if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
3064 		return -ENOTDIR;
3065 
3066 	/* Read in the sysctl name for better debug message logging */
3067 	for (i = 0; i < args->nlen; i++)
3068 		if (get_user(name[i], args->name + i))
3069 			return -EFAULT;
3070 
3071 	/* Ignore accesses to kernel.version */
3072 	if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
3073 		return 0;
3074 
3075 	if (msg_count < 5) {
3076 		msg_count++;
3077 		printk(KERN_INFO
3078 			"warning: process `%s' used the deprecated sysctl "
3079 			"system call with ", current->comm);
3080 		for (i = 0; i < args->nlen; i++)
3081 			printk("%d.", name[i]);
3082 		printk("\n");
3083 	}
3084 	return 0;
3085 }
3086 
3087 /*
3088  * No sense putting this after each symbol definition, twice,
3089  * exception granted :-)
3090  */
3091 EXPORT_SYMBOL(proc_dointvec);
3092 EXPORT_SYMBOL(proc_dointvec_jiffies);
3093 EXPORT_SYMBOL(proc_dointvec_minmax);
3094 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3095 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3096 EXPORT_SYMBOL(proc_dostring);
3097 EXPORT_SYMBOL(proc_doulongvec_minmax);
3098 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3099 EXPORT_SYMBOL(register_sysctl_table);
3100 EXPORT_SYMBOL(register_sysctl_paths);
3101 EXPORT_SYMBOL(sysctl_intvec);
3102 EXPORT_SYMBOL(sysctl_jiffies);
3103 EXPORT_SYMBOL(sysctl_ms_jiffies);
3104 EXPORT_SYMBOL(sysctl_string);
3105 EXPORT_SYMBOL(sysctl_data);
3106 EXPORT_SYMBOL(unregister_sysctl_table);
3107