xref: /linux-6.15/kernel/sysctl.c (revision eb4400e3)
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 int one_hundred = 100;
105 
106 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
107 static int maxolduid = 65535;
108 static int minolduid;
109 static int min_percpu_pagelist_fract = 8;
110 
111 static int ngroups_max = NGROUPS_MAX;
112 
113 #ifdef CONFIG_MODULES
114 extern char modprobe_path[];
115 #endif
116 #ifdef CONFIG_CHR_DEV_SG
117 extern int sg_big_buff;
118 #endif
119 
120 #ifdef CONFIG_SPARC
121 #include <asm/system.h>
122 #endif
123 
124 #ifdef CONFIG_SPARC64
125 extern int sysctl_tsb_ratio;
126 #endif
127 
128 #ifdef __hppa__
129 extern int pwrsw_enabled;
130 extern int unaligned_enabled;
131 #endif
132 
133 #ifdef CONFIG_S390
134 #ifdef CONFIG_MATHEMU
135 extern int sysctl_ieee_emulation_warnings;
136 #endif
137 extern int sysctl_userprocess_debug;
138 extern int spin_retry;
139 #endif
140 
141 #ifdef CONFIG_BSD_PROCESS_ACCT
142 extern int acct_parm[];
143 #endif
144 
145 #ifdef CONFIG_IA64
146 extern int no_unaligned_warning;
147 extern int unaligned_dump_stack;
148 #endif
149 
150 #ifdef CONFIG_RT_MUTEXES
151 extern int max_lock_depth;
152 #endif
153 
154 #ifdef CONFIG_PROC_SYSCTL
155 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
156 		  void __user *buffer, size_t *lenp, loff_t *ppos);
157 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
158 			       void __user *buffer, size_t *lenp, loff_t *ppos);
159 #endif
160 
161 static struct ctl_table root_table[];
162 static struct ctl_table_root sysctl_table_root;
163 static struct ctl_table_header root_table_header = {
164 	.count = 1,
165 	.ctl_table = root_table,
166 	.ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
167 	.root = &sysctl_table_root,
168 	.set = &sysctl_table_root.default_set,
169 };
170 static struct ctl_table_root sysctl_table_root = {
171 	.root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
172 	.default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
173 };
174 
175 static struct ctl_table kern_table[];
176 static struct ctl_table vm_table[];
177 static struct ctl_table fs_table[];
178 static struct ctl_table debug_table[];
179 static struct ctl_table dev_table[];
180 extern struct ctl_table random_table[];
181 #ifdef CONFIG_INOTIFY_USER
182 extern struct ctl_table inotify_table[];
183 #endif
184 #ifdef CONFIG_EPOLL
185 extern struct ctl_table epoll_table[];
186 #endif
187 
188 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
189 int sysctl_legacy_va_layout;
190 #endif
191 
192 extern int prove_locking;
193 extern int lock_stat;
194 
195 /* The default sysctl tables: */
196 
197 static struct ctl_table root_table[] = {
198 	{
199 		.ctl_name	= CTL_KERN,
200 		.procname	= "kernel",
201 		.mode		= 0555,
202 		.child		= kern_table,
203 	},
204 	{
205 		.ctl_name	= CTL_VM,
206 		.procname	= "vm",
207 		.mode		= 0555,
208 		.child		= vm_table,
209 	},
210 	{
211 		.ctl_name	= CTL_FS,
212 		.procname	= "fs",
213 		.mode		= 0555,
214 		.child		= fs_table,
215 	},
216 	{
217 		.ctl_name	= CTL_DEBUG,
218 		.procname	= "debug",
219 		.mode		= 0555,
220 		.child		= debug_table,
221 	},
222 	{
223 		.ctl_name	= CTL_DEV,
224 		.procname	= "dev",
225 		.mode		= 0555,
226 		.child		= dev_table,
227 	},
228 /*
229  * NOTE: do not add new entries to this table unless you have read
230  * Documentation/sysctl/ctl_unnumbered.txt
231  */
232 	{ .ctl_name = 0 }
233 };
234 
235 #ifdef CONFIG_SCHED_DEBUG
236 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
237 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
238 static int min_wakeup_granularity_ns;			/* 0 usecs */
239 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
240 #endif
241 
242 static struct ctl_table kern_table[] = {
243 #ifdef CONFIG_SCHED_DEBUG
244 	{
245 		.ctl_name	= CTL_UNNUMBERED,
246 		.procname	= "sched_min_granularity_ns",
247 		.data		= &sysctl_sched_min_granularity,
248 		.maxlen		= sizeof(unsigned int),
249 		.mode		= 0644,
250 		.proc_handler	= &sched_nr_latency_handler,
251 		.strategy	= &sysctl_intvec,
252 		.extra1		= &min_sched_granularity_ns,
253 		.extra2		= &max_sched_granularity_ns,
254 	},
255 	{
256 		.ctl_name	= CTL_UNNUMBERED,
257 		.procname	= "sched_latency_ns",
258 		.data		= &sysctl_sched_latency,
259 		.maxlen		= sizeof(unsigned int),
260 		.mode		= 0644,
261 		.proc_handler	= &sched_nr_latency_handler,
262 		.strategy	= &sysctl_intvec,
263 		.extra1		= &min_sched_granularity_ns,
264 		.extra2		= &max_sched_granularity_ns,
265 	},
266 	{
267 		.ctl_name	= CTL_UNNUMBERED,
268 		.procname	= "sched_wakeup_granularity_ns",
269 		.data		= &sysctl_sched_wakeup_granularity,
270 		.maxlen		= sizeof(unsigned int),
271 		.mode		= 0644,
272 		.proc_handler	= &proc_dointvec_minmax,
273 		.strategy	= &sysctl_intvec,
274 		.extra1		= &min_wakeup_granularity_ns,
275 		.extra2		= &max_wakeup_granularity_ns,
276 	},
277 	{
278 		.ctl_name	= CTL_UNNUMBERED,
279 		.procname	= "sched_shares_ratelimit",
280 		.data		= &sysctl_sched_shares_ratelimit,
281 		.maxlen		= sizeof(unsigned int),
282 		.mode		= 0644,
283 		.proc_handler	= &proc_dointvec,
284 	},
285 	{
286 		.ctl_name	= CTL_UNNUMBERED,
287 		.procname	= "sched_shares_thresh",
288 		.data		= &sysctl_sched_shares_thresh,
289 		.maxlen		= sizeof(unsigned int),
290 		.mode		= 0644,
291 		.proc_handler	= &proc_dointvec_minmax,
292 		.strategy	= &sysctl_intvec,
293 		.extra1		= &zero,
294 	},
295 	{
296 		.ctl_name	= CTL_UNNUMBERED,
297 		.procname	= "sched_child_runs_first",
298 		.data		= &sysctl_sched_child_runs_first,
299 		.maxlen		= sizeof(unsigned int),
300 		.mode		= 0644,
301 		.proc_handler	= &proc_dointvec,
302 	},
303 	{
304 		.ctl_name	= CTL_UNNUMBERED,
305 		.procname	= "sched_features",
306 		.data		= &sysctl_sched_features,
307 		.maxlen		= sizeof(unsigned int),
308 		.mode		= 0644,
309 		.proc_handler	= &proc_dointvec,
310 	},
311 	{
312 		.ctl_name	= CTL_UNNUMBERED,
313 		.procname	= "sched_migration_cost",
314 		.data		= &sysctl_sched_migration_cost,
315 		.maxlen		= sizeof(unsigned int),
316 		.mode		= 0644,
317 		.proc_handler	= &proc_dointvec,
318 	},
319 	{
320 		.ctl_name	= CTL_UNNUMBERED,
321 		.procname	= "sched_nr_migrate",
322 		.data		= &sysctl_sched_nr_migrate,
323 		.maxlen		= sizeof(unsigned int),
324 		.mode		= 0644,
325 		.proc_handler	= &proc_dointvec,
326 	},
327 #endif
328 	{
329 		.ctl_name	= CTL_UNNUMBERED,
330 		.procname	= "sched_rt_period_us",
331 		.data		= &sysctl_sched_rt_period,
332 		.maxlen		= sizeof(unsigned int),
333 		.mode		= 0644,
334 		.proc_handler	= &sched_rt_handler,
335 	},
336 	{
337 		.ctl_name	= CTL_UNNUMBERED,
338 		.procname	= "sched_rt_runtime_us",
339 		.data		= &sysctl_sched_rt_runtime,
340 		.maxlen		= sizeof(int),
341 		.mode		= 0644,
342 		.proc_handler	= &sched_rt_handler,
343 	},
344 	{
345 		.ctl_name	= CTL_UNNUMBERED,
346 		.procname	= "sched_compat_yield",
347 		.data		= &sysctl_sched_compat_yield,
348 		.maxlen		= sizeof(unsigned int),
349 		.mode		= 0644,
350 		.proc_handler	= &proc_dointvec,
351 	},
352 #ifdef CONFIG_PROVE_LOCKING
353 	{
354 		.ctl_name	= CTL_UNNUMBERED,
355 		.procname	= "prove_locking",
356 		.data		= &prove_locking,
357 		.maxlen		= sizeof(int),
358 		.mode		= 0644,
359 		.proc_handler	= &proc_dointvec,
360 	},
361 #endif
362 #ifdef CONFIG_LOCK_STAT
363 	{
364 		.ctl_name	= CTL_UNNUMBERED,
365 		.procname	= "lock_stat",
366 		.data		= &lock_stat,
367 		.maxlen		= sizeof(int),
368 		.mode		= 0644,
369 		.proc_handler	= &proc_dointvec,
370 	},
371 #endif
372 	{
373 		.ctl_name	= KERN_PANIC,
374 		.procname	= "panic",
375 		.data		= &panic_timeout,
376 		.maxlen		= sizeof(int),
377 		.mode		= 0644,
378 		.proc_handler	= &proc_dointvec,
379 	},
380 	{
381 		.ctl_name	= KERN_CORE_USES_PID,
382 		.procname	= "core_uses_pid",
383 		.data		= &core_uses_pid,
384 		.maxlen		= sizeof(int),
385 		.mode		= 0644,
386 		.proc_handler	= &proc_dointvec,
387 	},
388 	{
389 		.ctl_name	= KERN_CORE_PATTERN,
390 		.procname	= "core_pattern",
391 		.data		= core_pattern,
392 		.maxlen		= CORENAME_MAX_SIZE,
393 		.mode		= 0644,
394 		.proc_handler	= &proc_dostring,
395 		.strategy	= &sysctl_string,
396 	},
397 #ifdef CONFIG_PROC_SYSCTL
398 	{
399 		.procname	= "tainted",
400 		.maxlen 	= sizeof(long),
401 		.mode		= 0644,
402 		.proc_handler	= &proc_taint,
403 	},
404 #endif
405 #ifdef CONFIG_LATENCYTOP
406 	{
407 		.procname	= "latencytop",
408 		.data		= &latencytop_enabled,
409 		.maxlen		= sizeof(int),
410 		.mode		= 0644,
411 		.proc_handler	= &proc_dointvec,
412 	},
413 #endif
414 #ifdef CONFIG_BLK_DEV_INITRD
415 	{
416 		.ctl_name	= KERN_REALROOTDEV,
417 		.procname	= "real-root-dev",
418 		.data		= &real_root_dev,
419 		.maxlen		= sizeof(int),
420 		.mode		= 0644,
421 		.proc_handler	= &proc_dointvec,
422 	},
423 #endif
424 	{
425 		.ctl_name	= CTL_UNNUMBERED,
426 		.procname	= "print-fatal-signals",
427 		.data		= &print_fatal_signals,
428 		.maxlen		= sizeof(int),
429 		.mode		= 0644,
430 		.proc_handler	= &proc_dointvec,
431 	},
432 #ifdef CONFIG_SPARC
433 	{
434 		.ctl_name	= KERN_SPARC_REBOOT,
435 		.procname	= "reboot-cmd",
436 		.data		= reboot_command,
437 		.maxlen		= 256,
438 		.mode		= 0644,
439 		.proc_handler	= &proc_dostring,
440 		.strategy	= &sysctl_string,
441 	},
442 	{
443 		.ctl_name	= KERN_SPARC_STOP_A,
444 		.procname	= "stop-a",
445 		.data		= &stop_a_enabled,
446 		.maxlen		= sizeof (int),
447 		.mode		= 0644,
448 		.proc_handler	= &proc_dointvec,
449 	},
450 	{
451 		.ctl_name	= KERN_SPARC_SCONS_PWROFF,
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 		.ctl_name	= CTL_UNNUMBERED,
462 		.procname	= "tsb-ratio",
463 		.data		= &sysctl_tsb_ratio,
464 		.maxlen		= sizeof (int),
465 		.mode		= 0644,
466 		.proc_handler	= &proc_dointvec,
467 	},
468 #endif
469 #ifdef __hppa__
470 	{
471 		.ctl_name	= KERN_HPPA_PWRSW,
472 		.procname	= "soft-power",
473 		.data		= &pwrsw_enabled,
474 		.maxlen		= sizeof (int),
475 	 	.mode		= 0644,
476 		.proc_handler	= &proc_dointvec,
477 	},
478 	{
479 		.ctl_name	= KERN_HPPA_UNALIGNED,
480 		.procname	= "unaligned-trap",
481 		.data		= &unaligned_enabled,
482 		.maxlen		= sizeof (int),
483 		.mode		= 0644,
484 		.proc_handler	= &proc_dointvec,
485 	},
486 #endif
487 	{
488 		.ctl_name	= KERN_CTLALTDEL,
489 		.procname	= "ctrl-alt-del",
490 		.data		= &C_A_D,
491 		.maxlen		= sizeof(int),
492 		.mode		= 0644,
493 		.proc_handler	= &proc_dointvec,
494 	},
495 #ifdef CONFIG_FUNCTION_TRACER
496 	{
497 		.ctl_name	= CTL_UNNUMBERED,
498 		.procname	= "ftrace_enabled",
499 		.data		= &ftrace_enabled,
500 		.maxlen		= sizeof(int),
501 		.mode		= 0644,
502 		.proc_handler	= &ftrace_enable_sysctl,
503 	},
504 #endif
505 #ifdef CONFIG_STACK_TRACER
506 	{
507 		.ctl_name	= CTL_UNNUMBERED,
508 		.procname	= "stack_tracer_enabled",
509 		.data		= &stack_tracer_enabled,
510 		.maxlen		= sizeof(int),
511 		.mode		= 0644,
512 		.proc_handler	= &stack_trace_sysctl,
513 	},
514 #endif
515 #ifdef CONFIG_TRACING
516 	{
517 		.ctl_name	= CTL_UNNUMBERED,
518 		.procname	= "ftrace_dump_on_oops",
519 		.data		= &ftrace_dump_on_oops,
520 		.maxlen		= sizeof(int),
521 		.mode		= 0644,
522 		.proc_handler	= &proc_dointvec,
523 	},
524 #endif
525 #ifdef CONFIG_MODULES
526 	{
527 		.ctl_name	= KERN_MODPROBE,
528 		.procname	= "modprobe",
529 		.data		= &modprobe_path,
530 		.maxlen		= KMOD_PATH_LEN,
531 		.mode		= 0644,
532 		.proc_handler	= &proc_dostring,
533 		.strategy	= &sysctl_string,
534 	},
535 #endif
536 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
537 	{
538 		.ctl_name	= KERN_HOTPLUG,
539 		.procname	= "hotplug",
540 		.data		= &uevent_helper,
541 		.maxlen		= UEVENT_HELPER_PATH_LEN,
542 		.mode		= 0644,
543 		.proc_handler	= &proc_dostring,
544 		.strategy	= &sysctl_string,
545 	},
546 #endif
547 #ifdef CONFIG_CHR_DEV_SG
548 	{
549 		.ctl_name	= KERN_SG_BIG_BUFF,
550 		.procname	= "sg-big-buff",
551 		.data		= &sg_big_buff,
552 		.maxlen		= sizeof (int),
553 		.mode		= 0444,
554 		.proc_handler	= &proc_dointvec,
555 	},
556 #endif
557 #ifdef CONFIG_BSD_PROCESS_ACCT
558 	{
559 		.ctl_name	= KERN_ACCT,
560 		.procname	= "acct",
561 		.data		= &acct_parm,
562 		.maxlen		= 3*sizeof(int),
563 		.mode		= 0644,
564 		.proc_handler	= &proc_dointvec,
565 	},
566 #endif
567 #ifdef CONFIG_MAGIC_SYSRQ
568 	{
569 		.ctl_name	= KERN_SYSRQ,
570 		.procname	= "sysrq",
571 		.data		= &__sysrq_enabled,
572 		.maxlen		= sizeof (int),
573 		.mode		= 0644,
574 		.proc_handler	= &proc_dointvec,
575 	},
576 #endif
577 #ifdef CONFIG_PROC_SYSCTL
578 	{
579 		.procname	= "cad_pid",
580 		.data		= NULL,
581 		.maxlen		= sizeof (int),
582 		.mode		= 0600,
583 		.proc_handler	= &proc_do_cad_pid,
584 	},
585 #endif
586 	{
587 		.ctl_name	= KERN_MAX_THREADS,
588 		.procname	= "threads-max",
589 		.data		= &max_threads,
590 		.maxlen		= sizeof(int),
591 		.mode		= 0644,
592 		.proc_handler	= &proc_dointvec,
593 	},
594 	{
595 		.ctl_name	= KERN_RANDOM,
596 		.procname	= "random",
597 		.mode		= 0555,
598 		.child		= random_table,
599 	},
600 	{
601 		.ctl_name	= KERN_OVERFLOWUID,
602 		.procname	= "overflowuid",
603 		.data		= &overflowuid,
604 		.maxlen		= sizeof(int),
605 		.mode		= 0644,
606 		.proc_handler	= &proc_dointvec_minmax,
607 		.strategy	= &sysctl_intvec,
608 		.extra1		= &minolduid,
609 		.extra2		= &maxolduid,
610 	},
611 	{
612 		.ctl_name	= KERN_OVERFLOWGID,
613 		.procname	= "overflowgid",
614 		.data		= &overflowgid,
615 		.maxlen		= sizeof(int),
616 		.mode		= 0644,
617 		.proc_handler	= &proc_dointvec_minmax,
618 		.strategy	= &sysctl_intvec,
619 		.extra1		= &minolduid,
620 		.extra2		= &maxolduid,
621 	},
622 #ifdef CONFIG_S390
623 #ifdef CONFIG_MATHEMU
624 	{
625 		.ctl_name	= KERN_IEEE_EMULATION_WARNINGS,
626 		.procname	= "ieee_emulation_warnings",
627 		.data		= &sysctl_ieee_emulation_warnings,
628 		.maxlen		= sizeof(int),
629 		.mode		= 0644,
630 		.proc_handler	= &proc_dointvec,
631 	},
632 #endif
633 	{
634 		.ctl_name	= KERN_S390_USER_DEBUG_LOGGING,
635 		.procname	= "userprocess_debug",
636 		.data		= &sysctl_userprocess_debug,
637 		.maxlen		= sizeof(int),
638 		.mode		= 0644,
639 		.proc_handler	= &proc_dointvec,
640 	},
641 #endif
642 	{
643 		.ctl_name	= KERN_PIDMAX,
644 		.procname	= "pid_max",
645 		.data		= &pid_max,
646 		.maxlen		= sizeof (int),
647 		.mode		= 0644,
648 		.proc_handler	= &proc_dointvec_minmax,
649 		.strategy	= sysctl_intvec,
650 		.extra1		= &pid_max_min,
651 		.extra2		= &pid_max_max,
652 	},
653 	{
654 		.ctl_name	= KERN_PANIC_ON_OOPS,
655 		.procname	= "panic_on_oops",
656 		.data		= &panic_on_oops,
657 		.maxlen		= sizeof(int),
658 		.mode		= 0644,
659 		.proc_handler	= &proc_dointvec,
660 	},
661 #if defined CONFIG_PRINTK
662 	{
663 		.ctl_name	= KERN_PRINTK,
664 		.procname	= "printk",
665 		.data		= &console_loglevel,
666 		.maxlen		= 4*sizeof(int),
667 		.mode		= 0644,
668 		.proc_handler	= &proc_dointvec,
669 	},
670 	{
671 		.ctl_name	= KERN_PRINTK_RATELIMIT,
672 		.procname	= "printk_ratelimit",
673 		.data		= &printk_ratelimit_state.interval,
674 		.maxlen		= sizeof(int),
675 		.mode		= 0644,
676 		.proc_handler	= &proc_dointvec_jiffies,
677 		.strategy	= &sysctl_jiffies,
678 	},
679 	{
680 		.ctl_name	= KERN_PRINTK_RATELIMIT_BURST,
681 		.procname	= "printk_ratelimit_burst",
682 		.data		= &printk_ratelimit_state.burst,
683 		.maxlen		= sizeof(int),
684 		.mode		= 0644,
685 		.proc_handler	= &proc_dointvec,
686 	},
687 #endif
688 	{
689 		.ctl_name	= KERN_NGROUPS_MAX,
690 		.procname	= "ngroups_max",
691 		.data		= &ngroups_max,
692 		.maxlen		= sizeof (int),
693 		.mode		= 0444,
694 		.proc_handler	= &proc_dointvec,
695 	},
696 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
697 	{
698 		.ctl_name       = KERN_UNKNOWN_NMI_PANIC,
699 		.procname       = "unknown_nmi_panic",
700 		.data           = &unknown_nmi_panic,
701 		.maxlen         = sizeof (int),
702 		.mode           = 0644,
703 		.proc_handler   = &proc_dointvec,
704 	},
705 	{
706 		.procname       = "nmi_watchdog",
707 		.data           = &nmi_watchdog_enabled,
708 		.maxlen         = sizeof (int),
709 		.mode           = 0644,
710 		.proc_handler   = &proc_nmi_enabled,
711 	},
712 #endif
713 #if defined(CONFIG_X86)
714 	{
715 		.ctl_name	= KERN_PANIC_ON_NMI,
716 		.procname	= "panic_on_unrecovered_nmi",
717 		.data		= &panic_on_unrecovered_nmi,
718 		.maxlen		= sizeof(int),
719 		.mode		= 0644,
720 		.proc_handler	= &proc_dointvec,
721 	},
722 	{
723 		.ctl_name	= KERN_BOOTLOADER_TYPE,
724 		.procname	= "bootloader_type",
725 		.data		= &bootloader_type,
726 		.maxlen		= sizeof (int),
727 		.mode		= 0444,
728 		.proc_handler	= &proc_dointvec,
729 	},
730 	{
731 		.ctl_name	= CTL_UNNUMBERED,
732 		.procname	= "kstack_depth_to_print",
733 		.data		= &kstack_depth_to_print,
734 		.maxlen		= sizeof(int),
735 		.mode		= 0644,
736 		.proc_handler	= &proc_dointvec,
737 	},
738 	{
739 		.ctl_name	= CTL_UNNUMBERED,
740 		.procname	= "io_delay_type",
741 		.data		= &io_delay_type,
742 		.maxlen		= sizeof(int),
743 		.mode		= 0644,
744 		.proc_handler	= &proc_dointvec,
745 	},
746 #endif
747 #if defined(CONFIG_MMU)
748 	{
749 		.ctl_name	= KERN_RANDOMIZE,
750 		.procname	= "randomize_va_space",
751 		.data		= &randomize_va_space,
752 		.maxlen		= sizeof(int),
753 		.mode		= 0644,
754 		.proc_handler	= &proc_dointvec,
755 	},
756 #endif
757 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
758 	{
759 		.ctl_name	= KERN_SPIN_RETRY,
760 		.procname	= "spin_retry",
761 		.data		= &spin_retry,
762 		.maxlen		= sizeof (int),
763 		.mode		= 0644,
764 		.proc_handler	= &proc_dointvec,
765 	},
766 #endif
767 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
768 	{
769 		.procname	= "acpi_video_flags",
770 		.data		= &acpi_realmode_flags,
771 		.maxlen		= sizeof (unsigned long),
772 		.mode		= 0644,
773 		.proc_handler	= &proc_doulongvec_minmax,
774 	},
775 #endif
776 #ifdef CONFIG_IA64
777 	{
778 		.ctl_name	= KERN_IA64_UNALIGNED,
779 		.procname	= "ignore-unaligned-usertrap",
780 		.data		= &no_unaligned_warning,
781 		.maxlen		= sizeof (int),
782 	 	.mode		= 0644,
783 		.proc_handler	= &proc_dointvec,
784 	},
785 	{
786 		.ctl_name	= CTL_UNNUMBERED,
787 		.procname	= "unaligned-dump-stack",
788 		.data		= &unaligned_dump_stack,
789 		.maxlen		= sizeof (int),
790 		.mode		= 0644,
791 		.proc_handler	= &proc_dointvec,
792 	},
793 #endif
794 #ifdef CONFIG_DETECT_SOFTLOCKUP
795 	{
796 		.ctl_name	= CTL_UNNUMBERED,
797 		.procname	= "softlockup_panic",
798 		.data		= &softlockup_panic,
799 		.maxlen		= sizeof(int),
800 		.mode		= 0644,
801 		.proc_handler	= &proc_dointvec_minmax,
802 		.strategy	= &sysctl_intvec,
803 		.extra1		= &zero,
804 		.extra2		= &one,
805 	},
806 	{
807 		.ctl_name	= CTL_UNNUMBERED,
808 		.procname	= "softlockup_thresh",
809 		.data		= &softlockup_thresh,
810 		.maxlen		= sizeof(int),
811 		.mode		= 0644,
812 		.proc_handler	= &proc_dosoftlockup_thresh,
813 		.strategy	= &sysctl_intvec,
814 		.extra1		= &neg_one,
815 		.extra2		= &sixty,
816 	},
817 	{
818 		.ctl_name	= CTL_UNNUMBERED,
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 		.strategy	= &sysctl_intvec,
825 	},
826 	{
827 		.ctl_name	= CTL_UNNUMBERED,
828 		.procname	= "hung_task_timeout_secs",
829 		.data		= &sysctl_hung_task_timeout_secs,
830 		.maxlen		= sizeof(unsigned long),
831 		.mode		= 0644,
832 		.proc_handler	= &proc_doulongvec_minmax,
833 		.strategy	= &sysctl_intvec,
834 	},
835 	{
836 		.ctl_name	= CTL_UNNUMBERED,
837 		.procname	= "hung_task_warnings",
838 		.data		= &sysctl_hung_task_warnings,
839 		.maxlen		= sizeof(unsigned long),
840 		.mode		= 0644,
841 		.proc_handler	= &proc_doulongvec_minmax,
842 		.strategy	= &sysctl_intvec,
843 	},
844 #endif
845 #ifdef CONFIG_COMPAT
846 	{
847 		.ctl_name	= KERN_COMPAT_LOG,
848 		.procname	= "compat-log",
849 		.data		= &compat_log,
850 		.maxlen		= sizeof (int),
851 	 	.mode		= 0644,
852 		.proc_handler	= &proc_dointvec,
853 	},
854 #endif
855 #ifdef CONFIG_RT_MUTEXES
856 	{
857 		.ctl_name	= KERN_MAX_LOCK_DEPTH,
858 		.procname	= "max_lock_depth",
859 		.data		= &max_lock_depth,
860 		.maxlen		= sizeof(int),
861 		.mode		= 0644,
862 		.proc_handler	= &proc_dointvec,
863 	},
864 #endif
865 	{
866 		.ctl_name	= CTL_UNNUMBERED,
867 		.procname	= "poweroff_cmd",
868 		.data		= &poweroff_cmd,
869 		.maxlen		= POWEROFF_CMD_PATH_LEN,
870 		.mode		= 0644,
871 		.proc_handler	= &proc_dostring,
872 		.strategy	= &sysctl_string,
873 	},
874 #ifdef CONFIG_KEYS
875 	{
876 		.ctl_name	= CTL_UNNUMBERED,
877 		.procname	= "keys",
878 		.mode		= 0555,
879 		.child		= key_sysctls,
880 	},
881 #endif
882 #ifdef CONFIG_RCU_TORTURE_TEST
883 	{
884 		.ctl_name       = CTL_UNNUMBERED,
885 		.procname       = "rcutorture_runnable",
886 		.data           = &rcutorture_runnable,
887 		.maxlen         = sizeof(int),
888 		.mode           = 0644,
889 		.proc_handler   = &proc_dointvec,
890 	},
891 #endif
892 #ifdef CONFIG_UNEVICTABLE_LRU
893 	{
894 		.ctl_name	= CTL_UNNUMBERED,
895 		.procname	= "scan_unevictable_pages",
896 		.data		= &scan_unevictable_pages,
897 		.maxlen		= sizeof(scan_unevictable_pages),
898 		.mode		= 0644,
899 		.proc_handler	= &scan_unevictable_handler,
900 	},
901 #endif
902 /*
903  * NOTE: do not add new entries to this table unless you have read
904  * Documentation/sysctl/ctl_unnumbered.txt
905  */
906 	{ .ctl_name = 0 }
907 };
908 
909 static struct ctl_table vm_table[] = {
910 	{
911 		.ctl_name	= VM_OVERCOMMIT_MEMORY,
912 		.procname	= "overcommit_memory",
913 		.data		= &sysctl_overcommit_memory,
914 		.maxlen		= sizeof(sysctl_overcommit_memory),
915 		.mode		= 0644,
916 		.proc_handler	= &proc_dointvec,
917 	},
918 	{
919 		.ctl_name	= VM_PANIC_ON_OOM,
920 		.procname	= "panic_on_oom",
921 		.data		= &sysctl_panic_on_oom,
922 		.maxlen		= sizeof(sysctl_panic_on_oom),
923 		.mode		= 0644,
924 		.proc_handler	= &proc_dointvec,
925 	},
926 	{
927 		.ctl_name	= CTL_UNNUMBERED,
928 		.procname	= "oom_kill_allocating_task",
929 		.data		= &sysctl_oom_kill_allocating_task,
930 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
931 		.mode		= 0644,
932 		.proc_handler	= &proc_dointvec,
933 	},
934 	{
935 		.ctl_name	= CTL_UNNUMBERED,
936 		.procname	= "oom_dump_tasks",
937 		.data		= &sysctl_oom_dump_tasks,
938 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
939 		.mode		= 0644,
940 		.proc_handler	= &proc_dointvec,
941 	},
942 	{
943 		.ctl_name	= VM_OVERCOMMIT_RATIO,
944 		.procname	= "overcommit_ratio",
945 		.data		= &sysctl_overcommit_ratio,
946 		.maxlen		= sizeof(sysctl_overcommit_ratio),
947 		.mode		= 0644,
948 		.proc_handler	= &proc_dointvec,
949 	},
950 	{
951 		.ctl_name	= VM_PAGE_CLUSTER,
952 		.procname	= "page-cluster",
953 		.data		= &page_cluster,
954 		.maxlen		= sizeof(int),
955 		.mode		= 0644,
956 		.proc_handler	= &proc_dointvec,
957 	},
958 	{
959 		.ctl_name	= VM_DIRTY_BACKGROUND,
960 		.procname	= "dirty_background_ratio",
961 		.data		= &dirty_background_ratio,
962 		.maxlen		= sizeof(dirty_background_ratio),
963 		.mode		= 0644,
964 		.proc_handler	= &dirty_background_ratio_handler,
965 		.strategy	= &sysctl_intvec,
966 		.extra1		= &zero,
967 		.extra2		= &one_hundred,
968 	},
969 	{
970 		.ctl_name	= CTL_UNNUMBERED,
971 		.procname	= "dirty_background_bytes",
972 		.data		= &dirty_background_bytes,
973 		.maxlen		= sizeof(dirty_background_bytes),
974 		.mode		= 0644,
975 		.proc_handler	= &dirty_background_bytes_handler,
976 		.strategy	= &sysctl_intvec,
977 		.extra1		= &one,
978 	},
979 	{
980 		.ctl_name	= VM_DIRTY_RATIO,
981 		.procname	= "dirty_ratio",
982 		.data		= &vm_dirty_ratio,
983 		.maxlen		= sizeof(vm_dirty_ratio),
984 		.mode		= 0644,
985 		.proc_handler	= &dirty_ratio_handler,
986 		.strategy	= &sysctl_intvec,
987 		.extra1		= &zero,
988 		.extra2		= &one_hundred,
989 	},
990 	{
991 		.ctl_name	= CTL_UNNUMBERED,
992 		.procname	= "dirty_bytes",
993 		.data		= &vm_dirty_bytes,
994 		.maxlen		= sizeof(vm_dirty_bytes),
995 		.mode		= 0644,
996 		.proc_handler	= &dirty_bytes_handler,
997 		.strategy	= &sysctl_intvec,
998 		.extra1		= &one,
999 	},
1000 	{
1001 		.procname	= "dirty_writeback_centisecs",
1002 		.data		= &dirty_writeback_interval,
1003 		.maxlen		= sizeof(dirty_writeback_interval),
1004 		.mode		= 0644,
1005 		.proc_handler	= &dirty_writeback_centisecs_handler,
1006 	},
1007 	{
1008 		.procname	= "dirty_expire_centisecs",
1009 		.data		= &dirty_expire_interval,
1010 		.maxlen		= sizeof(dirty_expire_interval),
1011 		.mode		= 0644,
1012 		.proc_handler	= &proc_dointvec_userhz_jiffies,
1013 	},
1014 	{
1015 		.ctl_name	= VM_NR_PDFLUSH_THREADS,
1016 		.procname	= "nr_pdflush_threads",
1017 		.data		= &nr_pdflush_threads,
1018 		.maxlen		= sizeof nr_pdflush_threads,
1019 		.mode		= 0444 /* read-only*/,
1020 		.proc_handler	= &proc_dointvec,
1021 	},
1022 	{
1023 		.ctl_name	= VM_SWAPPINESS,
1024 		.procname	= "swappiness",
1025 		.data		= &vm_swappiness,
1026 		.maxlen		= sizeof(vm_swappiness),
1027 		.mode		= 0644,
1028 		.proc_handler	= &proc_dointvec_minmax,
1029 		.strategy	= &sysctl_intvec,
1030 		.extra1		= &zero,
1031 		.extra2		= &one_hundred,
1032 	},
1033 #ifdef CONFIG_HUGETLB_PAGE
1034 	 {
1035 		.procname	= "nr_hugepages",
1036 		.data		= NULL,
1037 		.maxlen		= sizeof(unsigned long),
1038 		.mode		= 0644,
1039 		.proc_handler	= &hugetlb_sysctl_handler,
1040 		.extra1		= (void *)&hugetlb_zero,
1041 		.extra2		= (void *)&hugetlb_infinity,
1042 	 },
1043 	 {
1044 		.ctl_name	= VM_HUGETLB_GROUP,
1045 		.procname	= "hugetlb_shm_group",
1046 		.data		= &sysctl_hugetlb_shm_group,
1047 		.maxlen		= sizeof(gid_t),
1048 		.mode		= 0644,
1049 		.proc_handler	= &proc_dointvec,
1050 	 },
1051 	 {
1052 		.ctl_name	= CTL_UNNUMBERED,
1053 		.procname	= "hugepages_treat_as_movable",
1054 		.data		= &hugepages_treat_as_movable,
1055 		.maxlen		= sizeof(int),
1056 		.mode		= 0644,
1057 		.proc_handler	= &hugetlb_treat_movable_handler,
1058 	},
1059 	{
1060 		.ctl_name	= CTL_UNNUMBERED,
1061 		.procname	= "nr_overcommit_hugepages",
1062 		.data		= NULL,
1063 		.maxlen		= sizeof(unsigned long),
1064 		.mode		= 0644,
1065 		.proc_handler	= &hugetlb_overcommit_handler,
1066 		.extra1		= (void *)&hugetlb_zero,
1067 		.extra2		= (void *)&hugetlb_infinity,
1068 	},
1069 #endif
1070 	{
1071 		.ctl_name	= VM_LOWMEM_RESERVE_RATIO,
1072 		.procname	= "lowmem_reserve_ratio",
1073 		.data		= &sysctl_lowmem_reserve_ratio,
1074 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1075 		.mode		= 0644,
1076 		.proc_handler	= &lowmem_reserve_ratio_sysctl_handler,
1077 		.strategy	= &sysctl_intvec,
1078 	},
1079 	{
1080 		.ctl_name	= VM_DROP_PAGECACHE,
1081 		.procname	= "drop_caches",
1082 		.data		= &sysctl_drop_caches,
1083 		.maxlen		= sizeof(int),
1084 		.mode		= 0644,
1085 		.proc_handler	= drop_caches_sysctl_handler,
1086 		.strategy	= &sysctl_intvec,
1087 	},
1088 	{
1089 		.ctl_name	= VM_MIN_FREE_KBYTES,
1090 		.procname	= "min_free_kbytes",
1091 		.data		= &min_free_kbytes,
1092 		.maxlen		= sizeof(min_free_kbytes),
1093 		.mode		= 0644,
1094 		.proc_handler	= &min_free_kbytes_sysctl_handler,
1095 		.strategy	= &sysctl_intvec,
1096 		.extra1		= &zero,
1097 	},
1098 	{
1099 		.ctl_name	= VM_PERCPU_PAGELIST_FRACTION,
1100 		.procname	= "percpu_pagelist_fraction",
1101 		.data		= &percpu_pagelist_fraction,
1102 		.maxlen		= sizeof(percpu_pagelist_fraction),
1103 		.mode		= 0644,
1104 		.proc_handler	= &percpu_pagelist_fraction_sysctl_handler,
1105 		.strategy	= &sysctl_intvec,
1106 		.extra1		= &min_percpu_pagelist_fract,
1107 	},
1108 #ifdef CONFIG_MMU
1109 	{
1110 		.ctl_name	= VM_MAX_MAP_COUNT,
1111 		.procname	= "max_map_count",
1112 		.data		= &sysctl_max_map_count,
1113 		.maxlen		= sizeof(sysctl_max_map_count),
1114 		.mode		= 0644,
1115 		.proc_handler	= &proc_dointvec
1116 	},
1117 #else
1118 	{
1119 		.ctl_name	= CTL_UNNUMBERED,
1120 		.procname	= "nr_trim_pages",
1121 		.data		= &sysctl_nr_trim_pages,
1122 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1123 		.mode		= 0644,
1124 		.proc_handler	= &proc_dointvec_minmax,
1125 		.strategy	= &sysctl_intvec,
1126 		.extra1		= &zero,
1127 	},
1128 #endif
1129 	{
1130 		.ctl_name	= VM_LAPTOP_MODE,
1131 		.procname	= "laptop_mode",
1132 		.data		= &laptop_mode,
1133 		.maxlen		= sizeof(laptop_mode),
1134 		.mode		= 0644,
1135 		.proc_handler	= &proc_dointvec_jiffies,
1136 		.strategy	= &sysctl_jiffies,
1137 	},
1138 	{
1139 		.ctl_name	= VM_BLOCK_DUMP,
1140 		.procname	= "block_dump",
1141 		.data		= &block_dump,
1142 		.maxlen		= sizeof(block_dump),
1143 		.mode		= 0644,
1144 		.proc_handler	= &proc_dointvec,
1145 		.strategy	= &sysctl_intvec,
1146 		.extra1		= &zero,
1147 	},
1148 	{
1149 		.ctl_name	= VM_VFS_CACHE_PRESSURE,
1150 		.procname	= "vfs_cache_pressure",
1151 		.data		= &sysctl_vfs_cache_pressure,
1152 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1153 		.mode		= 0644,
1154 		.proc_handler	= &proc_dointvec,
1155 		.strategy	= &sysctl_intvec,
1156 		.extra1		= &zero,
1157 	},
1158 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1159 	{
1160 		.ctl_name	= VM_LEGACY_VA_LAYOUT,
1161 		.procname	= "legacy_va_layout",
1162 		.data		= &sysctl_legacy_va_layout,
1163 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1164 		.mode		= 0644,
1165 		.proc_handler	= &proc_dointvec,
1166 		.strategy	= &sysctl_intvec,
1167 		.extra1		= &zero,
1168 	},
1169 #endif
1170 #ifdef CONFIG_NUMA
1171 	{
1172 		.ctl_name	= VM_ZONE_RECLAIM_MODE,
1173 		.procname	= "zone_reclaim_mode",
1174 		.data		= &zone_reclaim_mode,
1175 		.maxlen		= sizeof(zone_reclaim_mode),
1176 		.mode		= 0644,
1177 		.proc_handler	= &proc_dointvec,
1178 		.strategy	= &sysctl_intvec,
1179 		.extra1		= &zero,
1180 	},
1181 	{
1182 		.ctl_name	= VM_MIN_UNMAPPED,
1183 		.procname	= "min_unmapped_ratio",
1184 		.data		= &sysctl_min_unmapped_ratio,
1185 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1186 		.mode		= 0644,
1187 		.proc_handler	= &sysctl_min_unmapped_ratio_sysctl_handler,
1188 		.strategy	= &sysctl_intvec,
1189 		.extra1		= &zero,
1190 		.extra2		= &one_hundred,
1191 	},
1192 	{
1193 		.ctl_name	= VM_MIN_SLAB,
1194 		.procname	= "min_slab_ratio",
1195 		.data		= &sysctl_min_slab_ratio,
1196 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1197 		.mode		= 0644,
1198 		.proc_handler	= &sysctl_min_slab_ratio_sysctl_handler,
1199 		.strategy	= &sysctl_intvec,
1200 		.extra1		= &zero,
1201 		.extra2		= &one_hundred,
1202 	},
1203 #endif
1204 #ifdef CONFIG_SMP
1205 	{
1206 		.ctl_name	= CTL_UNNUMBERED,
1207 		.procname	= "stat_interval",
1208 		.data		= &sysctl_stat_interval,
1209 		.maxlen		= sizeof(sysctl_stat_interval),
1210 		.mode		= 0644,
1211 		.proc_handler	= &proc_dointvec_jiffies,
1212 		.strategy	= &sysctl_jiffies,
1213 	},
1214 #endif
1215 #ifdef CONFIG_SECURITY
1216 	{
1217 		.ctl_name	= CTL_UNNUMBERED,
1218 		.procname	= "mmap_min_addr",
1219 		.data		= &mmap_min_addr,
1220 		.maxlen         = sizeof(unsigned long),
1221 		.mode		= 0644,
1222 		.proc_handler	= &proc_doulongvec_minmax,
1223 	},
1224 #endif
1225 #ifdef CONFIG_NUMA
1226 	{
1227 		.ctl_name	= CTL_UNNUMBERED,
1228 		.procname	= "numa_zonelist_order",
1229 		.data		= &numa_zonelist_order,
1230 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1231 		.mode		= 0644,
1232 		.proc_handler	= &numa_zonelist_order_handler,
1233 		.strategy	= &sysctl_string,
1234 	},
1235 #endif
1236 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1237    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1238 	{
1239 		.ctl_name	= VM_VDSO_ENABLED,
1240 		.procname	= "vdso_enabled",
1241 		.data		= &vdso_enabled,
1242 		.maxlen		= sizeof(vdso_enabled),
1243 		.mode		= 0644,
1244 		.proc_handler	= &proc_dointvec,
1245 		.strategy	= &sysctl_intvec,
1246 		.extra1		= &zero,
1247 	},
1248 #endif
1249 #ifdef CONFIG_HIGHMEM
1250 	{
1251 		.ctl_name	= CTL_UNNUMBERED,
1252 		.procname	= "highmem_is_dirtyable",
1253 		.data		= &vm_highmem_is_dirtyable,
1254 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1255 		.mode		= 0644,
1256 		.proc_handler	= &proc_dointvec_minmax,
1257 		.strategy	= &sysctl_intvec,
1258 		.extra1		= &zero,
1259 		.extra2		= &one,
1260 	},
1261 #endif
1262 /*
1263  * NOTE: do not add new entries to this table unless you have read
1264  * Documentation/sysctl/ctl_unnumbered.txt
1265  */
1266 	{ .ctl_name = 0 }
1267 };
1268 
1269 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1270 static struct ctl_table binfmt_misc_table[] = {
1271 	{ .ctl_name = 0 }
1272 };
1273 #endif
1274 
1275 static struct ctl_table fs_table[] = {
1276 	{
1277 		.ctl_name	= FS_NRINODE,
1278 		.procname	= "inode-nr",
1279 		.data		= &inodes_stat,
1280 		.maxlen		= 2*sizeof(int),
1281 		.mode		= 0444,
1282 		.proc_handler	= &proc_dointvec,
1283 	},
1284 	{
1285 		.ctl_name	= FS_STATINODE,
1286 		.procname	= "inode-state",
1287 		.data		= &inodes_stat,
1288 		.maxlen		= 7*sizeof(int),
1289 		.mode		= 0444,
1290 		.proc_handler	= &proc_dointvec,
1291 	},
1292 	{
1293 		.procname	= "file-nr",
1294 		.data		= &files_stat,
1295 		.maxlen		= 3*sizeof(int),
1296 		.mode		= 0444,
1297 		.proc_handler	= &proc_nr_files,
1298 	},
1299 	{
1300 		.ctl_name	= FS_MAXFILE,
1301 		.procname	= "file-max",
1302 		.data		= &files_stat.max_files,
1303 		.maxlen		= sizeof(int),
1304 		.mode		= 0644,
1305 		.proc_handler	= &proc_dointvec,
1306 	},
1307 	{
1308 		.ctl_name	= CTL_UNNUMBERED,
1309 		.procname	= "nr_open",
1310 		.data		= &sysctl_nr_open,
1311 		.maxlen		= sizeof(int),
1312 		.mode		= 0644,
1313 		.proc_handler	= &proc_dointvec_minmax,
1314 		.extra1		= &sysctl_nr_open_min,
1315 		.extra2		= &sysctl_nr_open_max,
1316 	},
1317 	{
1318 		.ctl_name	= FS_DENTRY,
1319 		.procname	= "dentry-state",
1320 		.data		= &dentry_stat,
1321 		.maxlen		= 6*sizeof(int),
1322 		.mode		= 0444,
1323 		.proc_handler	= &proc_dointvec,
1324 	},
1325 	{
1326 		.ctl_name	= FS_OVERFLOWUID,
1327 		.procname	= "overflowuid",
1328 		.data		= &fs_overflowuid,
1329 		.maxlen		= sizeof(int),
1330 		.mode		= 0644,
1331 		.proc_handler	= &proc_dointvec_minmax,
1332 		.strategy	= &sysctl_intvec,
1333 		.extra1		= &minolduid,
1334 		.extra2		= &maxolduid,
1335 	},
1336 	{
1337 		.ctl_name	= FS_OVERFLOWGID,
1338 		.procname	= "overflowgid",
1339 		.data		= &fs_overflowgid,
1340 		.maxlen		= sizeof(int),
1341 		.mode		= 0644,
1342 		.proc_handler	= &proc_dointvec_minmax,
1343 		.strategy	= &sysctl_intvec,
1344 		.extra1		= &minolduid,
1345 		.extra2		= &maxolduid,
1346 	},
1347 #ifdef CONFIG_FILE_LOCKING
1348 	{
1349 		.ctl_name	= FS_LEASES,
1350 		.procname	= "leases-enable",
1351 		.data		= &leases_enable,
1352 		.maxlen		= sizeof(int),
1353 		.mode		= 0644,
1354 		.proc_handler	= &proc_dointvec,
1355 	},
1356 #endif
1357 #ifdef CONFIG_DNOTIFY
1358 	{
1359 		.ctl_name	= FS_DIR_NOTIFY,
1360 		.procname	= "dir-notify-enable",
1361 		.data		= &dir_notify_enable,
1362 		.maxlen		= sizeof(int),
1363 		.mode		= 0644,
1364 		.proc_handler	= &proc_dointvec,
1365 	},
1366 #endif
1367 #ifdef CONFIG_MMU
1368 #ifdef CONFIG_FILE_LOCKING
1369 	{
1370 		.ctl_name	= FS_LEASE_TIME,
1371 		.procname	= "lease-break-time",
1372 		.data		= &lease_break_time,
1373 		.maxlen		= sizeof(int),
1374 		.mode		= 0644,
1375 		.proc_handler	= &proc_dointvec_minmax,
1376 		.strategy	= &sysctl_intvec,
1377 		.extra1		= &zero,
1378 		.extra2		= &two,
1379 	},
1380 #endif
1381 #ifdef CONFIG_AIO
1382 	{
1383 		.procname	= "aio-nr",
1384 		.data		= &aio_nr,
1385 		.maxlen		= sizeof(aio_nr),
1386 		.mode		= 0444,
1387 		.proc_handler	= &proc_doulongvec_minmax,
1388 	},
1389 	{
1390 		.procname	= "aio-max-nr",
1391 		.data		= &aio_max_nr,
1392 		.maxlen		= sizeof(aio_max_nr),
1393 		.mode		= 0644,
1394 		.proc_handler	= &proc_doulongvec_minmax,
1395 	},
1396 #endif /* CONFIG_AIO */
1397 #ifdef CONFIG_INOTIFY_USER
1398 	{
1399 		.ctl_name	= FS_INOTIFY,
1400 		.procname	= "inotify",
1401 		.mode		= 0555,
1402 		.child		= inotify_table,
1403 	},
1404 #endif
1405 #ifdef CONFIG_EPOLL
1406 	{
1407 		.procname	= "epoll",
1408 		.mode		= 0555,
1409 		.child		= epoll_table,
1410 	},
1411 #endif
1412 #endif
1413 	{
1414 		.ctl_name	= KERN_SETUID_DUMPABLE,
1415 		.procname	= "suid_dumpable",
1416 		.data		= &suid_dumpable,
1417 		.maxlen		= sizeof(int),
1418 		.mode		= 0644,
1419 		.proc_handler	= &proc_dointvec,
1420 	},
1421 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1422 	{
1423 		.ctl_name	= CTL_UNNUMBERED,
1424 		.procname	= "binfmt_misc",
1425 		.mode		= 0555,
1426 		.child		= binfmt_misc_table,
1427 	},
1428 #endif
1429 /*
1430  * NOTE: do not add new entries to this table unless you have read
1431  * Documentation/sysctl/ctl_unnumbered.txt
1432  */
1433 	{ .ctl_name = 0 }
1434 };
1435 
1436 static struct ctl_table debug_table[] = {
1437 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1438 	{
1439 		.ctl_name	= CTL_UNNUMBERED,
1440 		.procname	= "exception-trace",
1441 		.data		= &show_unhandled_signals,
1442 		.maxlen		= sizeof(int),
1443 		.mode		= 0644,
1444 		.proc_handler	= proc_dointvec
1445 	},
1446 #endif
1447 	{ .ctl_name = 0 }
1448 };
1449 
1450 static struct ctl_table dev_table[] = {
1451 	{ .ctl_name = 0 }
1452 };
1453 
1454 static DEFINE_SPINLOCK(sysctl_lock);
1455 
1456 /* called under sysctl_lock */
1457 static int use_table(struct ctl_table_header *p)
1458 {
1459 	if (unlikely(p->unregistering))
1460 		return 0;
1461 	p->used++;
1462 	return 1;
1463 }
1464 
1465 /* called under sysctl_lock */
1466 static void unuse_table(struct ctl_table_header *p)
1467 {
1468 	if (!--p->used)
1469 		if (unlikely(p->unregistering))
1470 			complete(p->unregistering);
1471 }
1472 
1473 /* called under sysctl_lock, will reacquire if has to wait */
1474 static void start_unregistering(struct ctl_table_header *p)
1475 {
1476 	/*
1477 	 * if p->used is 0, nobody will ever touch that entry again;
1478 	 * we'll eliminate all paths to it before dropping sysctl_lock
1479 	 */
1480 	if (unlikely(p->used)) {
1481 		struct completion wait;
1482 		init_completion(&wait);
1483 		p->unregistering = &wait;
1484 		spin_unlock(&sysctl_lock);
1485 		wait_for_completion(&wait);
1486 		spin_lock(&sysctl_lock);
1487 	} else {
1488 		/* anything non-NULL; we'll never dereference it */
1489 		p->unregistering = ERR_PTR(-EINVAL);
1490 	}
1491 	/*
1492 	 * do not remove from the list until nobody holds it; walking the
1493 	 * list in do_sysctl() relies on that.
1494 	 */
1495 	list_del_init(&p->ctl_entry);
1496 }
1497 
1498 void sysctl_head_get(struct ctl_table_header *head)
1499 {
1500 	spin_lock(&sysctl_lock);
1501 	head->count++;
1502 	spin_unlock(&sysctl_lock);
1503 }
1504 
1505 void sysctl_head_put(struct ctl_table_header *head)
1506 {
1507 	spin_lock(&sysctl_lock);
1508 	if (!--head->count)
1509 		kfree(head);
1510 	spin_unlock(&sysctl_lock);
1511 }
1512 
1513 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1514 {
1515 	if (!head)
1516 		BUG();
1517 	spin_lock(&sysctl_lock);
1518 	if (!use_table(head))
1519 		head = ERR_PTR(-ENOENT);
1520 	spin_unlock(&sysctl_lock);
1521 	return head;
1522 }
1523 
1524 void sysctl_head_finish(struct ctl_table_header *head)
1525 {
1526 	if (!head)
1527 		return;
1528 	spin_lock(&sysctl_lock);
1529 	unuse_table(head);
1530 	spin_unlock(&sysctl_lock);
1531 }
1532 
1533 static struct ctl_table_set *
1534 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1535 {
1536 	struct ctl_table_set *set = &root->default_set;
1537 	if (root->lookup)
1538 		set = root->lookup(root, namespaces);
1539 	return set;
1540 }
1541 
1542 static struct list_head *
1543 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1544 {
1545 	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1546 	return &set->list;
1547 }
1548 
1549 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1550 					    struct ctl_table_header *prev)
1551 {
1552 	struct ctl_table_root *root;
1553 	struct list_head *header_list;
1554 	struct ctl_table_header *head;
1555 	struct list_head *tmp;
1556 
1557 	spin_lock(&sysctl_lock);
1558 	if (prev) {
1559 		head = prev;
1560 		tmp = &prev->ctl_entry;
1561 		unuse_table(prev);
1562 		goto next;
1563 	}
1564 	tmp = &root_table_header.ctl_entry;
1565 	for (;;) {
1566 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1567 
1568 		if (!use_table(head))
1569 			goto next;
1570 		spin_unlock(&sysctl_lock);
1571 		return head;
1572 	next:
1573 		root = head->root;
1574 		tmp = tmp->next;
1575 		header_list = lookup_header_list(root, namespaces);
1576 		if (tmp != header_list)
1577 			continue;
1578 
1579 		do {
1580 			root = list_entry(root->root_list.next,
1581 					struct ctl_table_root, root_list);
1582 			if (root == &sysctl_table_root)
1583 				goto out;
1584 			header_list = lookup_header_list(root, namespaces);
1585 		} while (list_empty(header_list));
1586 		tmp = header_list->next;
1587 	}
1588 out:
1589 	spin_unlock(&sysctl_lock);
1590 	return NULL;
1591 }
1592 
1593 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1594 {
1595 	return __sysctl_head_next(current->nsproxy, prev);
1596 }
1597 
1598 void register_sysctl_root(struct ctl_table_root *root)
1599 {
1600 	spin_lock(&sysctl_lock);
1601 	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1602 	spin_unlock(&sysctl_lock);
1603 }
1604 
1605 #ifdef CONFIG_SYSCTL_SYSCALL
1606 /* Perform the actual read/write of a sysctl table entry. */
1607 static int do_sysctl_strategy(struct ctl_table_root *root,
1608 			struct ctl_table *table,
1609 			void __user *oldval, size_t __user *oldlenp,
1610 			void __user *newval, size_t newlen)
1611 {
1612 	int op = 0, rc;
1613 
1614 	if (oldval)
1615 		op |= MAY_READ;
1616 	if (newval)
1617 		op |= MAY_WRITE;
1618 	if (sysctl_perm(root, table, op))
1619 		return -EPERM;
1620 
1621 	if (table->strategy) {
1622 		rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1623 		if (rc < 0)
1624 			return rc;
1625 		if (rc > 0)
1626 			return 0;
1627 	}
1628 
1629 	/* If there is no strategy routine, or if the strategy returns
1630 	 * zero, proceed with automatic r/w */
1631 	if (table->data && table->maxlen) {
1632 		rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1633 		if (rc < 0)
1634 			return rc;
1635 	}
1636 	return 0;
1637 }
1638 
1639 static int parse_table(int __user *name, int nlen,
1640 		       void __user *oldval, size_t __user *oldlenp,
1641 		       void __user *newval, size_t newlen,
1642 		       struct ctl_table_root *root,
1643 		       struct ctl_table *table)
1644 {
1645 	int n;
1646 repeat:
1647 	if (!nlen)
1648 		return -ENOTDIR;
1649 	if (get_user(n, name))
1650 		return -EFAULT;
1651 	for ( ; table->ctl_name || table->procname; table++) {
1652 		if (!table->ctl_name)
1653 			continue;
1654 		if (n == table->ctl_name) {
1655 			int error;
1656 			if (table->child) {
1657 				if (sysctl_perm(root, table, MAY_EXEC))
1658 					return -EPERM;
1659 				name++;
1660 				nlen--;
1661 				table = table->child;
1662 				goto repeat;
1663 			}
1664 			error = do_sysctl_strategy(root, table,
1665 						   oldval, oldlenp,
1666 						   newval, newlen);
1667 			return error;
1668 		}
1669 	}
1670 	return -ENOTDIR;
1671 }
1672 
1673 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1674 	       void __user *newval, size_t newlen)
1675 {
1676 	struct ctl_table_header *head;
1677 	int error = -ENOTDIR;
1678 
1679 	if (nlen <= 0 || nlen >= CTL_MAXNAME)
1680 		return -ENOTDIR;
1681 	if (oldval) {
1682 		int old_len;
1683 		if (!oldlenp || get_user(old_len, oldlenp))
1684 			return -EFAULT;
1685 	}
1686 
1687 	for (head = sysctl_head_next(NULL); head;
1688 			head = sysctl_head_next(head)) {
1689 		error = parse_table(name, nlen, oldval, oldlenp,
1690 					newval, newlen,
1691 					head->root, head->ctl_table);
1692 		if (error != -ENOTDIR) {
1693 			sysctl_head_finish(head);
1694 			break;
1695 		}
1696 	}
1697 	return error;
1698 }
1699 
1700 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
1701 {
1702 	struct __sysctl_args tmp;
1703 	int error;
1704 
1705 	if (copy_from_user(&tmp, args, sizeof(tmp)))
1706 		return -EFAULT;
1707 
1708 	error = deprecated_sysctl_warning(&tmp);
1709 	if (error)
1710 		goto out;
1711 
1712 	lock_kernel();
1713 	error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1714 			  tmp.newval, tmp.newlen);
1715 	unlock_kernel();
1716 out:
1717 	return error;
1718 }
1719 #endif /* CONFIG_SYSCTL_SYSCALL */
1720 
1721 /*
1722  * sysctl_perm does NOT grant the superuser all rights automatically, because
1723  * some sysctl variables are readonly even to root.
1724  */
1725 
1726 static int test_perm(int mode, int op)
1727 {
1728 	if (!current_euid())
1729 		mode >>= 6;
1730 	else if (in_egroup_p(0))
1731 		mode >>= 3;
1732 	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1733 		return 0;
1734 	return -EACCES;
1735 }
1736 
1737 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1738 {
1739 	int error;
1740 	int mode;
1741 
1742 	error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1743 	if (error)
1744 		return error;
1745 
1746 	if (root->permissions)
1747 		mode = root->permissions(root, current->nsproxy, table);
1748 	else
1749 		mode = table->mode;
1750 
1751 	return test_perm(mode, op);
1752 }
1753 
1754 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1755 {
1756 	for (; table->ctl_name || table->procname; table++) {
1757 		table->parent = parent;
1758 		if (table->child)
1759 			sysctl_set_parent(table, table->child);
1760 	}
1761 }
1762 
1763 static __init int sysctl_init(void)
1764 {
1765 	sysctl_set_parent(NULL, root_table);
1766 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1767 	{
1768 		int err;
1769 		err = sysctl_check_table(current->nsproxy, root_table);
1770 	}
1771 #endif
1772 	return 0;
1773 }
1774 
1775 core_initcall(sysctl_init);
1776 
1777 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1778 				      struct ctl_table *table)
1779 {
1780 	struct ctl_table *p;
1781 	const char *s = branch->procname;
1782 
1783 	/* branch should have named subdirectory as its first element */
1784 	if (!s || !branch->child)
1785 		return NULL;
1786 
1787 	/* ... and nothing else */
1788 	if (branch[1].procname || branch[1].ctl_name)
1789 		return NULL;
1790 
1791 	/* table should contain subdirectory with the same name */
1792 	for (p = table; p->procname || p->ctl_name; p++) {
1793 		if (!p->child)
1794 			continue;
1795 		if (p->procname && strcmp(p->procname, s) == 0)
1796 			return p;
1797 	}
1798 	return NULL;
1799 }
1800 
1801 /* see if attaching q to p would be an improvement */
1802 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1803 {
1804 	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1805 	struct ctl_table *next;
1806 	int is_better = 0;
1807 	int not_in_parent = !p->attached_by;
1808 
1809 	while ((next = is_branch_in(by, to)) != NULL) {
1810 		if (by == q->attached_by)
1811 			is_better = 1;
1812 		if (to == p->attached_by)
1813 			not_in_parent = 1;
1814 		by = by->child;
1815 		to = next->child;
1816 	}
1817 
1818 	if (is_better && not_in_parent) {
1819 		q->attached_by = by;
1820 		q->attached_to = to;
1821 		q->parent = p;
1822 	}
1823 }
1824 
1825 /**
1826  * __register_sysctl_paths - register a sysctl hierarchy
1827  * @root: List of sysctl headers to register on
1828  * @namespaces: Data to compute which lists of sysctl entries are visible
1829  * @path: The path to the directory the sysctl table is in.
1830  * @table: the top-level table structure
1831  *
1832  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1833  * array. A completely 0 filled entry terminates the table.
1834  *
1835  * The members of the &struct ctl_table structure are used as follows:
1836  *
1837  * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1838  *            must be unique within that level of sysctl
1839  *
1840  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1841  *            enter a sysctl file
1842  *
1843  * data - a pointer to data for use by proc_handler
1844  *
1845  * maxlen - the maximum size in bytes of the data
1846  *
1847  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1848  *
1849  * child - a pointer to the child sysctl table if this entry is a directory, or
1850  *         %NULL.
1851  *
1852  * proc_handler - the text handler routine (described below)
1853  *
1854  * strategy - the strategy routine (described below)
1855  *
1856  * de - for internal use by the sysctl routines
1857  *
1858  * extra1, extra2 - extra pointers usable by the proc handler routines
1859  *
1860  * Leaf nodes in the sysctl tree will be represented by a single file
1861  * under /proc; non-leaf nodes will be represented by directories.
1862  *
1863  * sysctl(2) can automatically manage read and write requests through
1864  * the sysctl table.  The data and maxlen fields of the ctl_table
1865  * struct enable minimal validation of the values being written to be
1866  * performed, and the mode field allows minimal authentication.
1867  *
1868  * More sophisticated management can be enabled by the provision of a
1869  * strategy routine with the table entry.  This will be called before
1870  * any automatic read or write of the data is performed.
1871  *
1872  * The strategy routine may return
1873  *
1874  * < 0 - Error occurred (error is passed to user process)
1875  *
1876  * 0   - OK - proceed with automatic read or write.
1877  *
1878  * > 0 - OK - read or write has been done by the strategy routine, so
1879  *       return immediately.
1880  *
1881  * There must be a proc_handler routine for any terminal nodes
1882  * mirrored under /proc/sys (non-terminals are handled by a built-in
1883  * directory handler).  Several default handlers are available to
1884  * cover common cases -
1885  *
1886  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1887  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1888  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1889  *
1890  * It is the handler's job to read the input buffer from user memory
1891  * and process it. The handler should return 0 on success.
1892  *
1893  * This routine returns %NULL on a failure to register, and a pointer
1894  * to the table header on success.
1895  */
1896 struct ctl_table_header *__register_sysctl_paths(
1897 	struct ctl_table_root *root,
1898 	struct nsproxy *namespaces,
1899 	const struct ctl_path *path, struct ctl_table *table)
1900 {
1901 	struct ctl_table_header *header;
1902 	struct ctl_table *new, **prevp;
1903 	unsigned int n, npath;
1904 	struct ctl_table_set *set;
1905 
1906 	/* Count the path components */
1907 	for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1908 		;
1909 
1910 	/*
1911 	 * For each path component, allocate a 2-element ctl_table array.
1912 	 * The first array element will be filled with the sysctl entry
1913 	 * for this, the second will be the sentinel (ctl_name == 0).
1914 	 *
1915 	 * We allocate everything in one go so that we don't have to
1916 	 * worry about freeing additional memory in unregister_sysctl_table.
1917 	 */
1918 	header = kzalloc(sizeof(struct ctl_table_header) +
1919 			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1920 	if (!header)
1921 		return NULL;
1922 
1923 	new = (struct ctl_table *) (header + 1);
1924 
1925 	/* Now connect the dots */
1926 	prevp = &header->ctl_table;
1927 	for (n = 0; n < npath; ++n, ++path) {
1928 		/* Copy the procname */
1929 		new->procname = path->procname;
1930 		new->ctl_name = path->ctl_name;
1931 		new->mode     = 0555;
1932 
1933 		*prevp = new;
1934 		prevp = &new->child;
1935 
1936 		new += 2;
1937 	}
1938 	*prevp = table;
1939 	header->ctl_table_arg = table;
1940 
1941 	INIT_LIST_HEAD(&header->ctl_entry);
1942 	header->used = 0;
1943 	header->unregistering = NULL;
1944 	header->root = root;
1945 	sysctl_set_parent(NULL, header->ctl_table);
1946 	header->count = 1;
1947 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1948 	if (sysctl_check_table(namespaces, header->ctl_table)) {
1949 		kfree(header);
1950 		return NULL;
1951 	}
1952 #endif
1953 	spin_lock(&sysctl_lock);
1954 	header->set = lookup_header_set(root, namespaces);
1955 	header->attached_by = header->ctl_table;
1956 	header->attached_to = root_table;
1957 	header->parent = &root_table_header;
1958 	for (set = header->set; set; set = set->parent) {
1959 		struct ctl_table_header *p;
1960 		list_for_each_entry(p, &set->list, ctl_entry) {
1961 			if (p->unregistering)
1962 				continue;
1963 			try_attach(p, header);
1964 		}
1965 	}
1966 	header->parent->count++;
1967 	list_add_tail(&header->ctl_entry, &header->set->list);
1968 	spin_unlock(&sysctl_lock);
1969 
1970 	return header;
1971 }
1972 
1973 /**
1974  * register_sysctl_table_path - register a sysctl table hierarchy
1975  * @path: The path to the directory the sysctl table is in.
1976  * @table: the top-level table structure
1977  *
1978  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1979  * array. A completely 0 filled entry terminates the table.
1980  *
1981  * See __register_sysctl_paths for more details.
1982  */
1983 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1984 						struct ctl_table *table)
1985 {
1986 	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1987 					path, table);
1988 }
1989 
1990 /**
1991  * register_sysctl_table - register a sysctl table hierarchy
1992  * @table: the top-level table structure
1993  *
1994  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1995  * array. A completely 0 filled entry terminates the table.
1996  *
1997  * See register_sysctl_paths for more details.
1998  */
1999 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
2000 {
2001 	static const struct ctl_path null_path[] = { {} };
2002 
2003 	return register_sysctl_paths(null_path, table);
2004 }
2005 
2006 /**
2007  * unregister_sysctl_table - unregister a sysctl table hierarchy
2008  * @header: the header returned from register_sysctl_table
2009  *
2010  * Unregisters the sysctl table and all children. proc entries may not
2011  * actually be removed until they are no longer used by anyone.
2012  */
2013 void unregister_sysctl_table(struct ctl_table_header * header)
2014 {
2015 	might_sleep();
2016 
2017 	if (header == NULL)
2018 		return;
2019 
2020 	spin_lock(&sysctl_lock);
2021 	start_unregistering(header);
2022 	if (!--header->parent->count) {
2023 		WARN_ON(1);
2024 		kfree(header->parent);
2025 	}
2026 	if (!--header->count)
2027 		kfree(header);
2028 	spin_unlock(&sysctl_lock);
2029 }
2030 
2031 int sysctl_is_seen(struct ctl_table_header *p)
2032 {
2033 	struct ctl_table_set *set = p->set;
2034 	int res;
2035 	spin_lock(&sysctl_lock);
2036 	if (p->unregistering)
2037 		res = 0;
2038 	else if (!set->is_seen)
2039 		res = 1;
2040 	else
2041 		res = set->is_seen(set);
2042 	spin_unlock(&sysctl_lock);
2043 	return res;
2044 }
2045 
2046 void setup_sysctl_set(struct ctl_table_set *p,
2047 	struct ctl_table_set *parent,
2048 	int (*is_seen)(struct ctl_table_set *))
2049 {
2050 	INIT_LIST_HEAD(&p->list);
2051 	p->parent = parent ? parent : &sysctl_table_root.default_set;
2052 	p->is_seen = is_seen;
2053 }
2054 
2055 #else /* !CONFIG_SYSCTL */
2056 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2057 {
2058 	return NULL;
2059 }
2060 
2061 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2062 						    struct ctl_table *table)
2063 {
2064 	return NULL;
2065 }
2066 
2067 void unregister_sysctl_table(struct ctl_table_header * table)
2068 {
2069 }
2070 
2071 void setup_sysctl_set(struct ctl_table_set *p,
2072 	struct ctl_table_set *parent,
2073 	int (*is_seen)(struct ctl_table_set *))
2074 {
2075 }
2076 
2077 void sysctl_head_put(struct ctl_table_header *head)
2078 {
2079 }
2080 
2081 #endif /* CONFIG_SYSCTL */
2082 
2083 /*
2084  * /proc/sys support
2085  */
2086 
2087 #ifdef CONFIG_PROC_SYSCTL
2088 
2089 static int _proc_do_string(void* data, int maxlen, int write,
2090 			   struct file *filp, void __user *buffer,
2091 			   size_t *lenp, loff_t *ppos)
2092 {
2093 	size_t len;
2094 	char __user *p;
2095 	char c;
2096 
2097 	if (!data || !maxlen || !*lenp) {
2098 		*lenp = 0;
2099 		return 0;
2100 	}
2101 
2102 	if (write) {
2103 		len = 0;
2104 		p = buffer;
2105 		while (len < *lenp) {
2106 			if (get_user(c, p++))
2107 				return -EFAULT;
2108 			if (c == 0 || c == '\n')
2109 				break;
2110 			len++;
2111 		}
2112 		if (len >= maxlen)
2113 			len = maxlen-1;
2114 		if(copy_from_user(data, buffer, len))
2115 			return -EFAULT;
2116 		((char *) data)[len] = 0;
2117 		*ppos += *lenp;
2118 	} else {
2119 		len = strlen(data);
2120 		if (len > maxlen)
2121 			len = maxlen;
2122 
2123 		if (*ppos > len) {
2124 			*lenp = 0;
2125 			return 0;
2126 		}
2127 
2128 		data += *ppos;
2129 		len  -= *ppos;
2130 
2131 		if (len > *lenp)
2132 			len = *lenp;
2133 		if (len)
2134 			if(copy_to_user(buffer, data, len))
2135 				return -EFAULT;
2136 		if (len < *lenp) {
2137 			if(put_user('\n', ((char __user *) buffer) + len))
2138 				return -EFAULT;
2139 			len++;
2140 		}
2141 		*lenp = len;
2142 		*ppos += len;
2143 	}
2144 	return 0;
2145 }
2146 
2147 /**
2148  * proc_dostring - read a string sysctl
2149  * @table: the sysctl table
2150  * @write: %TRUE if this is a write to the sysctl file
2151  * @filp: the file structure
2152  * @buffer: the user buffer
2153  * @lenp: the size of the user buffer
2154  * @ppos: file position
2155  *
2156  * Reads/writes a string from/to the user buffer. If the kernel
2157  * buffer provided is not large enough to hold the string, the
2158  * string is truncated. The copied string is %NULL-terminated.
2159  * If the string is being read by the user process, it is copied
2160  * and a newline '\n' is added. It is truncated if the buffer is
2161  * not large enough.
2162  *
2163  * Returns 0 on success.
2164  */
2165 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2166 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2167 {
2168 	return _proc_do_string(table->data, table->maxlen, write, filp,
2169 			       buffer, lenp, ppos);
2170 }
2171 
2172 
2173 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2174 				 int *valp,
2175 				 int write, void *data)
2176 {
2177 	if (write) {
2178 		*valp = *negp ? -*lvalp : *lvalp;
2179 	} else {
2180 		int val = *valp;
2181 		if (val < 0) {
2182 			*negp = -1;
2183 			*lvalp = (unsigned long)-val;
2184 		} else {
2185 			*negp = 0;
2186 			*lvalp = (unsigned long)val;
2187 		}
2188 	}
2189 	return 0;
2190 }
2191 
2192 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2193 		  int write, struct file *filp, void __user *buffer,
2194 		  size_t *lenp, loff_t *ppos,
2195 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2196 			      int write, void *data),
2197 		  void *data)
2198 {
2199 #define TMPBUFLEN 21
2200 	int *i, vleft, first=1, neg, val;
2201 	unsigned long lval;
2202 	size_t left, len;
2203 
2204 	char buf[TMPBUFLEN], *p;
2205 	char __user *s = buffer;
2206 
2207 	if (!tbl_data || !table->maxlen || !*lenp ||
2208 	    (*ppos && !write)) {
2209 		*lenp = 0;
2210 		return 0;
2211 	}
2212 
2213 	i = (int *) tbl_data;
2214 	vleft = table->maxlen / sizeof(*i);
2215 	left = *lenp;
2216 
2217 	if (!conv)
2218 		conv = do_proc_dointvec_conv;
2219 
2220 	for (; left && vleft--; i++, first=0) {
2221 		if (write) {
2222 			while (left) {
2223 				char c;
2224 				if (get_user(c, s))
2225 					return -EFAULT;
2226 				if (!isspace(c))
2227 					break;
2228 				left--;
2229 				s++;
2230 			}
2231 			if (!left)
2232 				break;
2233 			neg = 0;
2234 			len = left;
2235 			if (len > sizeof(buf) - 1)
2236 				len = sizeof(buf) - 1;
2237 			if (copy_from_user(buf, s, len))
2238 				return -EFAULT;
2239 			buf[len] = 0;
2240 			p = buf;
2241 			if (*p == '-' && left > 1) {
2242 				neg = 1;
2243 				p++;
2244 			}
2245 			if (*p < '0' || *p > '9')
2246 				break;
2247 
2248 			lval = simple_strtoul(p, &p, 0);
2249 
2250 			len = p-buf;
2251 			if ((len < left) && *p && !isspace(*p))
2252 				break;
2253 			if (neg)
2254 				val = -val;
2255 			s += len;
2256 			left -= len;
2257 
2258 			if (conv(&neg, &lval, i, 1, data))
2259 				break;
2260 		} else {
2261 			p = buf;
2262 			if (!first)
2263 				*p++ = '\t';
2264 
2265 			if (conv(&neg, &lval, i, 0, data))
2266 				break;
2267 
2268 			sprintf(p, "%s%lu", neg ? "-" : "", lval);
2269 			len = strlen(buf);
2270 			if (len > left)
2271 				len = left;
2272 			if(copy_to_user(s, buf, len))
2273 				return -EFAULT;
2274 			left -= len;
2275 			s += len;
2276 		}
2277 	}
2278 
2279 	if (!write && !first && left) {
2280 		if(put_user('\n', s))
2281 			return -EFAULT;
2282 		left--, s++;
2283 	}
2284 	if (write) {
2285 		while (left) {
2286 			char c;
2287 			if (get_user(c, s++))
2288 				return -EFAULT;
2289 			if (!isspace(c))
2290 				break;
2291 			left--;
2292 		}
2293 	}
2294 	if (write && first)
2295 		return -EINVAL;
2296 	*lenp -= left;
2297 	*ppos += *lenp;
2298 	return 0;
2299 #undef TMPBUFLEN
2300 }
2301 
2302 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2303 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2304 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2305 			      int write, void *data),
2306 		  void *data)
2307 {
2308 	return __do_proc_dointvec(table->data, table, write, filp,
2309 			buffer, lenp, ppos, conv, data);
2310 }
2311 
2312 /**
2313  * proc_dointvec - read a vector of integers
2314  * @table: the sysctl table
2315  * @write: %TRUE if this is a write to the sysctl file
2316  * @filp: the file structure
2317  * @buffer: the user buffer
2318  * @lenp: the size of the user buffer
2319  * @ppos: file position
2320  *
2321  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2322  * values from/to the user buffer, treated as an ASCII string.
2323  *
2324  * Returns 0 on success.
2325  */
2326 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2327 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2328 {
2329     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2330 		    	    NULL,NULL);
2331 }
2332 
2333 /*
2334  * Taint values can only be increased
2335  * This means we can safely use a temporary.
2336  */
2337 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
2338 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2339 {
2340 	struct ctl_table t;
2341 	unsigned long tmptaint = get_taint();
2342 	int err;
2343 
2344 	if (write && !capable(CAP_SYS_ADMIN))
2345 		return -EPERM;
2346 
2347 	t = *table;
2348 	t.data = &tmptaint;
2349 	err = proc_doulongvec_minmax(&t, write, filp, buffer, lenp, ppos);
2350 	if (err < 0)
2351 		return err;
2352 
2353 	if (write) {
2354 		/*
2355 		 * Poor man's atomic or. Not worth adding a primitive
2356 		 * to everyone's atomic.h for this
2357 		 */
2358 		int i;
2359 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2360 			if ((tmptaint >> i) & 1)
2361 				add_taint(i);
2362 		}
2363 	}
2364 
2365 	return err;
2366 }
2367 
2368 struct do_proc_dointvec_minmax_conv_param {
2369 	int *min;
2370 	int *max;
2371 };
2372 
2373 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2374 					int *valp,
2375 					int write, void *data)
2376 {
2377 	struct do_proc_dointvec_minmax_conv_param *param = data;
2378 	if (write) {
2379 		int val = *negp ? -*lvalp : *lvalp;
2380 		if ((param->min && *param->min > val) ||
2381 		    (param->max && *param->max < val))
2382 			return -EINVAL;
2383 		*valp = val;
2384 	} else {
2385 		int val = *valp;
2386 		if (val < 0) {
2387 			*negp = -1;
2388 			*lvalp = (unsigned long)-val;
2389 		} else {
2390 			*negp = 0;
2391 			*lvalp = (unsigned long)val;
2392 		}
2393 	}
2394 	return 0;
2395 }
2396 
2397 /**
2398  * proc_dointvec_minmax - read a vector of integers with min/max values
2399  * @table: the sysctl table
2400  * @write: %TRUE if this is a write to the sysctl file
2401  * @filp: the file structure
2402  * @buffer: the user buffer
2403  * @lenp: the size of the user buffer
2404  * @ppos: file position
2405  *
2406  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2407  * values from/to the user buffer, treated as an ASCII string.
2408  *
2409  * This routine will ensure the values are within the range specified by
2410  * table->extra1 (min) and table->extra2 (max).
2411  *
2412  * Returns 0 on success.
2413  */
2414 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2415 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2416 {
2417 	struct do_proc_dointvec_minmax_conv_param param = {
2418 		.min = (int *) table->extra1,
2419 		.max = (int *) table->extra2,
2420 	};
2421 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2422 				do_proc_dointvec_minmax_conv, &param);
2423 }
2424 
2425 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2426 				     struct file *filp,
2427 				     void __user *buffer,
2428 				     size_t *lenp, loff_t *ppos,
2429 				     unsigned long convmul,
2430 				     unsigned long convdiv)
2431 {
2432 #define TMPBUFLEN 21
2433 	unsigned long *i, *min, *max, val;
2434 	int vleft, first=1, neg;
2435 	size_t len, left;
2436 	char buf[TMPBUFLEN], *p;
2437 	char __user *s = buffer;
2438 
2439 	if (!data || !table->maxlen || !*lenp ||
2440 	    (*ppos && !write)) {
2441 		*lenp = 0;
2442 		return 0;
2443 	}
2444 
2445 	i = (unsigned long *) data;
2446 	min = (unsigned long *) table->extra1;
2447 	max = (unsigned long *) table->extra2;
2448 	vleft = table->maxlen / sizeof(unsigned long);
2449 	left = *lenp;
2450 
2451 	for (; left && vleft--; i++, min++, max++, first=0) {
2452 		if (write) {
2453 			while (left) {
2454 				char c;
2455 				if (get_user(c, s))
2456 					return -EFAULT;
2457 				if (!isspace(c))
2458 					break;
2459 				left--;
2460 				s++;
2461 			}
2462 			if (!left)
2463 				break;
2464 			neg = 0;
2465 			len = left;
2466 			if (len > TMPBUFLEN-1)
2467 				len = TMPBUFLEN-1;
2468 			if (copy_from_user(buf, s, len))
2469 				return -EFAULT;
2470 			buf[len] = 0;
2471 			p = buf;
2472 			if (*p == '-' && left > 1) {
2473 				neg = 1;
2474 				p++;
2475 			}
2476 			if (*p < '0' || *p > '9')
2477 				break;
2478 			val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2479 			len = p-buf;
2480 			if ((len < left) && *p && !isspace(*p))
2481 				break;
2482 			if (neg)
2483 				val = -val;
2484 			s += len;
2485 			left -= len;
2486 
2487 			if(neg)
2488 				continue;
2489 			if ((min && val < *min) || (max && val > *max))
2490 				continue;
2491 			*i = val;
2492 		} else {
2493 			p = buf;
2494 			if (!first)
2495 				*p++ = '\t';
2496 			sprintf(p, "%lu", convdiv * (*i) / convmul);
2497 			len = strlen(buf);
2498 			if (len > left)
2499 				len = left;
2500 			if(copy_to_user(s, buf, len))
2501 				return -EFAULT;
2502 			left -= len;
2503 			s += len;
2504 		}
2505 	}
2506 
2507 	if (!write && !first && left) {
2508 		if(put_user('\n', s))
2509 			return -EFAULT;
2510 		left--, s++;
2511 	}
2512 	if (write) {
2513 		while (left) {
2514 			char c;
2515 			if (get_user(c, s++))
2516 				return -EFAULT;
2517 			if (!isspace(c))
2518 				break;
2519 			left--;
2520 		}
2521 	}
2522 	if (write && first)
2523 		return -EINVAL;
2524 	*lenp -= left;
2525 	*ppos += *lenp;
2526 	return 0;
2527 #undef TMPBUFLEN
2528 }
2529 
2530 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2531 				     struct file *filp,
2532 				     void __user *buffer,
2533 				     size_t *lenp, loff_t *ppos,
2534 				     unsigned long convmul,
2535 				     unsigned long convdiv)
2536 {
2537 	return __do_proc_doulongvec_minmax(table->data, table, write,
2538 			filp, buffer, lenp, ppos, convmul, convdiv);
2539 }
2540 
2541 /**
2542  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2543  * @table: the sysctl table
2544  * @write: %TRUE if this is a write to the sysctl file
2545  * @filp: the file structure
2546  * @buffer: the user buffer
2547  * @lenp: the size of the user buffer
2548  * @ppos: file position
2549  *
2550  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2551  * values from/to the user buffer, treated as an ASCII string.
2552  *
2553  * This routine will ensure the values are within the range specified by
2554  * table->extra1 (min) and table->extra2 (max).
2555  *
2556  * Returns 0 on success.
2557  */
2558 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2559 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2560 {
2561     return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2562 }
2563 
2564 /**
2565  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2566  * @table: the sysctl table
2567  * @write: %TRUE if this is a write to the sysctl file
2568  * @filp: the file structure
2569  * @buffer: the user buffer
2570  * @lenp: the size of the user buffer
2571  * @ppos: file position
2572  *
2573  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2574  * values from/to the user buffer, treated as an ASCII string. The values
2575  * are treated as milliseconds, and converted to jiffies when they are stored.
2576  *
2577  * This routine will ensure the values are within the range specified by
2578  * table->extra1 (min) and table->extra2 (max).
2579  *
2580  * Returns 0 on success.
2581  */
2582 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2583 				      struct file *filp,
2584 				      void __user *buffer,
2585 				      size_t *lenp, loff_t *ppos)
2586 {
2587     return do_proc_doulongvec_minmax(table, write, filp, buffer,
2588 				     lenp, ppos, HZ, 1000l);
2589 }
2590 
2591 
2592 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2593 					 int *valp,
2594 					 int write, void *data)
2595 {
2596 	if (write) {
2597 		if (*lvalp > LONG_MAX / HZ)
2598 			return 1;
2599 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2600 	} else {
2601 		int val = *valp;
2602 		unsigned long lval;
2603 		if (val < 0) {
2604 			*negp = -1;
2605 			lval = (unsigned long)-val;
2606 		} else {
2607 			*negp = 0;
2608 			lval = (unsigned long)val;
2609 		}
2610 		*lvalp = lval / HZ;
2611 	}
2612 	return 0;
2613 }
2614 
2615 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2616 						int *valp,
2617 						int write, void *data)
2618 {
2619 	if (write) {
2620 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2621 			return 1;
2622 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2623 	} else {
2624 		int val = *valp;
2625 		unsigned long lval;
2626 		if (val < 0) {
2627 			*negp = -1;
2628 			lval = (unsigned long)-val;
2629 		} else {
2630 			*negp = 0;
2631 			lval = (unsigned long)val;
2632 		}
2633 		*lvalp = jiffies_to_clock_t(lval);
2634 	}
2635 	return 0;
2636 }
2637 
2638 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2639 					    int *valp,
2640 					    int write, void *data)
2641 {
2642 	if (write) {
2643 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2644 	} else {
2645 		int val = *valp;
2646 		unsigned long lval;
2647 		if (val < 0) {
2648 			*negp = -1;
2649 			lval = (unsigned long)-val;
2650 		} else {
2651 			*negp = 0;
2652 			lval = (unsigned long)val;
2653 		}
2654 		*lvalp = jiffies_to_msecs(lval);
2655 	}
2656 	return 0;
2657 }
2658 
2659 /**
2660  * proc_dointvec_jiffies - read a vector of integers as seconds
2661  * @table: the sysctl table
2662  * @write: %TRUE if this is a write to the sysctl file
2663  * @filp: the file structure
2664  * @buffer: the user buffer
2665  * @lenp: the size of the user buffer
2666  * @ppos: file position
2667  *
2668  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2669  * values from/to the user buffer, treated as an ASCII string.
2670  * The values read are assumed to be in seconds, and are converted into
2671  * jiffies.
2672  *
2673  * Returns 0 on success.
2674  */
2675 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2676 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2677 {
2678     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2679 		    	    do_proc_dointvec_jiffies_conv,NULL);
2680 }
2681 
2682 /**
2683  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2684  * @table: the sysctl table
2685  * @write: %TRUE if this is a write to the sysctl file
2686  * @filp: the file structure
2687  * @buffer: the user buffer
2688  * @lenp: the size of the user buffer
2689  * @ppos: pointer to the file position
2690  *
2691  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2692  * values from/to the user buffer, treated as an ASCII string.
2693  * The values read are assumed to be in 1/USER_HZ seconds, and
2694  * are converted into jiffies.
2695  *
2696  * Returns 0 on success.
2697  */
2698 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2699 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2700 {
2701     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2702 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2703 }
2704 
2705 /**
2706  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2707  * @table: the sysctl table
2708  * @write: %TRUE if this is a write to the sysctl file
2709  * @filp: the file structure
2710  * @buffer: the user buffer
2711  * @lenp: the size of the user buffer
2712  * @ppos: file position
2713  * @ppos: the current position in the file
2714  *
2715  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2716  * values from/to the user buffer, treated as an ASCII string.
2717  * The values read are assumed to be in 1/1000 seconds, and
2718  * are converted into jiffies.
2719  *
2720  * Returns 0 on success.
2721  */
2722 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2723 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2724 {
2725 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2726 				do_proc_dointvec_ms_jiffies_conv, NULL);
2727 }
2728 
2729 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2730 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2731 {
2732 	struct pid *new_pid;
2733 	pid_t tmp;
2734 	int r;
2735 
2736 	tmp = pid_vnr(cad_pid);
2737 
2738 	r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2739 			       lenp, ppos, NULL, NULL);
2740 	if (r || !write)
2741 		return r;
2742 
2743 	new_pid = find_get_pid(tmp);
2744 	if (!new_pid)
2745 		return -ESRCH;
2746 
2747 	put_pid(xchg(&cad_pid, new_pid));
2748 	return 0;
2749 }
2750 
2751 #else /* CONFIG_PROC_FS */
2752 
2753 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2754 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2755 {
2756 	return -ENOSYS;
2757 }
2758 
2759 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2760 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2761 {
2762 	return -ENOSYS;
2763 }
2764 
2765 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2766 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2767 {
2768 	return -ENOSYS;
2769 }
2770 
2771 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2772 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2773 {
2774 	return -ENOSYS;
2775 }
2776 
2777 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2778 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2779 {
2780 	return -ENOSYS;
2781 }
2782 
2783 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2784 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2785 {
2786 	return -ENOSYS;
2787 }
2788 
2789 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2790 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2791 {
2792 	return -ENOSYS;
2793 }
2794 
2795 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2796 				      struct file *filp,
2797 				      void __user *buffer,
2798 				      size_t *lenp, loff_t *ppos)
2799 {
2800     return -ENOSYS;
2801 }
2802 
2803 
2804 #endif /* CONFIG_PROC_FS */
2805 
2806 
2807 #ifdef CONFIG_SYSCTL_SYSCALL
2808 /*
2809  * General sysctl support routines
2810  */
2811 
2812 /* The generic sysctl data routine (used if no strategy routine supplied) */
2813 int sysctl_data(struct ctl_table *table,
2814 		void __user *oldval, size_t __user *oldlenp,
2815 		void __user *newval, size_t newlen)
2816 {
2817 	size_t len;
2818 
2819 	/* Get out of I don't have a variable */
2820 	if (!table->data || !table->maxlen)
2821 		return -ENOTDIR;
2822 
2823 	if (oldval && oldlenp) {
2824 		if (get_user(len, oldlenp))
2825 			return -EFAULT;
2826 		if (len) {
2827 			if (len > table->maxlen)
2828 				len = table->maxlen;
2829 			if (copy_to_user(oldval, table->data, len))
2830 				return -EFAULT;
2831 			if (put_user(len, oldlenp))
2832 				return -EFAULT;
2833 		}
2834 	}
2835 
2836 	if (newval && newlen) {
2837 		if (newlen > table->maxlen)
2838 			newlen = table->maxlen;
2839 
2840 		if (copy_from_user(table->data, newval, newlen))
2841 			return -EFAULT;
2842 	}
2843 	return 1;
2844 }
2845 
2846 /* The generic string strategy routine: */
2847 int sysctl_string(struct ctl_table *table,
2848 		  void __user *oldval, size_t __user *oldlenp,
2849 		  void __user *newval, size_t newlen)
2850 {
2851 	if (!table->data || !table->maxlen)
2852 		return -ENOTDIR;
2853 
2854 	if (oldval && oldlenp) {
2855 		size_t bufsize;
2856 		if (get_user(bufsize, oldlenp))
2857 			return -EFAULT;
2858 		if (bufsize) {
2859 			size_t len = strlen(table->data), copied;
2860 
2861 			/* This shouldn't trigger for a well-formed sysctl */
2862 			if (len > table->maxlen)
2863 				len = table->maxlen;
2864 
2865 			/* Copy up to a max of bufsize-1 bytes of the string */
2866 			copied = (len >= bufsize) ? bufsize - 1 : len;
2867 
2868 			if (copy_to_user(oldval, table->data, copied) ||
2869 			    put_user(0, (char __user *)(oldval + copied)))
2870 				return -EFAULT;
2871 			if (put_user(len, oldlenp))
2872 				return -EFAULT;
2873 		}
2874 	}
2875 	if (newval && newlen) {
2876 		size_t len = newlen;
2877 		if (len > table->maxlen)
2878 			len = table->maxlen;
2879 		if(copy_from_user(table->data, newval, len))
2880 			return -EFAULT;
2881 		if (len == table->maxlen)
2882 			len--;
2883 		((char *) table->data)[len] = 0;
2884 	}
2885 	return 1;
2886 }
2887 
2888 /*
2889  * This function makes sure that all of the integers in the vector
2890  * are between the minimum and maximum values given in the arrays
2891  * table->extra1 and table->extra2, respectively.
2892  */
2893 int sysctl_intvec(struct ctl_table *table,
2894 		void __user *oldval, size_t __user *oldlenp,
2895 		void __user *newval, size_t newlen)
2896 {
2897 
2898 	if (newval && newlen) {
2899 		int __user *vec = (int __user *) newval;
2900 		int *min = (int *) table->extra1;
2901 		int *max = (int *) table->extra2;
2902 		size_t length;
2903 		int i;
2904 
2905 		if (newlen % sizeof(int) != 0)
2906 			return -EINVAL;
2907 
2908 		if (!table->extra1 && !table->extra2)
2909 			return 0;
2910 
2911 		if (newlen > table->maxlen)
2912 			newlen = table->maxlen;
2913 		length = newlen / sizeof(int);
2914 
2915 		for (i = 0; i < length; i++) {
2916 			int value;
2917 			if (get_user(value, vec + i))
2918 				return -EFAULT;
2919 			if (min && value < min[i])
2920 				return -EINVAL;
2921 			if (max && value > max[i])
2922 				return -EINVAL;
2923 		}
2924 	}
2925 	return 0;
2926 }
2927 
2928 /* Strategy function to convert jiffies to seconds */
2929 int sysctl_jiffies(struct ctl_table *table,
2930 		void __user *oldval, size_t __user *oldlenp,
2931 		void __user *newval, size_t newlen)
2932 {
2933 	if (oldval && oldlenp) {
2934 		size_t olen;
2935 
2936 		if (get_user(olen, oldlenp))
2937 			return -EFAULT;
2938 		if (olen) {
2939 			int val;
2940 
2941 			if (olen < sizeof(int))
2942 				return -EINVAL;
2943 
2944 			val = *(int *)(table->data) / HZ;
2945 			if (put_user(val, (int __user *)oldval))
2946 				return -EFAULT;
2947 			if (put_user(sizeof(int), oldlenp))
2948 				return -EFAULT;
2949 		}
2950 	}
2951 	if (newval && newlen) {
2952 		int new;
2953 		if (newlen != sizeof(int))
2954 			return -EINVAL;
2955 		if (get_user(new, (int __user *)newval))
2956 			return -EFAULT;
2957 		*(int *)(table->data) = new*HZ;
2958 	}
2959 	return 1;
2960 }
2961 
2962 /* Strategy function to convert jiffies to seconds */
2963 int sysctl_ms_jiffies(struct ctl_table *table,
2964 		void __user *oldval, size_t __user *oldlenp,
2965 		void __user *newval, size_t newlen)
2966 {
2967 	if (oldval && oldlenp) {
2968 		size_t olen;
2969 
2970 		if (get_user(olen, oldlenp))
2971 			return -EFAULT;
2972 		if (olen) {
2973 			int val;
2974 
2975 			if (olen < sizeof(int))
2976 				return -EINVAL;
2977 
2978 			val = jiffies_to_msecs(*(int *)(table->data));
2979 			if (put_user(val, (int __user *)oldval))
2980 				return -EFAULT;
2981 			if (put_user(sizeof(int), oldlenp))
2982 				return -EFAULT;
2983 		}
2984 	}
2985 	if (newval && newlen) {
2986 		int new;
2987 		if (newlen != sizeof(int))
2988 			return -EINVAL;
2989 		if (get_user(new, (int __user *)newval))
2990 			return -EFAULT;
2991 		*(int *)(table->data) = msecs_to_jiffies(new);
2992 	}
2993 	return 1;
2994 }
2995 
2996 
2997 
2998 #else /* CONFIG_SYSCTL_SYSCALL */
2999 
3000 
3001 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
3002 {
3003 	struct __sysctl_args tmp;
3004 	int error;
3005 
3006 	if (copy_from_user(&tmp, args, sizeof(tmp)))
3007 		return -EFAULT;
3008 
3009 	error = deprecated_sysctl_warning(&tmp);
3010 
3011 	/* If no error reading the parameters then just -ENOSYS ... */
3012 	if (!error)
3013 		error = -ENOSYS;
3014 
3015 	return error;
3016 }
3017 
3018 int sysctl_data(struct ctl_table *table,
3019 		  void __user *oldval, size_t __user *oldlenp,
3020 		  void __user *newval, size_t newlen)
3021 {
3022 	return -ENOSYS;
3023 }
3024 
3025 int sysctl_string(struct ctl_table *table,
3026 		  void __user *oldval, size_t __user *oldlenp,
3027 		  void __user *newval, size_t newlen)
3028 {
3029 	return -ENOSYS;
3030 }
3031 
3032 int sysctl_intvec(struct ctl_table *table,
3033 		void __user *oldval, size_t __user *oldlenp,
3034 		void __user *newval, size_t newlen)
3035 {
3036 	return -ENOSYS;
3037 }
3038 
3039 int sysctl_jiffies(struct ctl_table *table,
3040 		void __user *oldval, size_t __user *oldlenp,
3041 		void __user *newval, size_t newlen)
3042 {
3043 	return -ENOSYS;
3044 }
3045 
3046 int sysctl_ms_jiffies(struct ctl_table *table,
3047 		void __user *oldval, size_t __user *oldlenp,
3048 		void __user *newval, size_t newlen)
3049 {
3050 	return -ENOSYS;
3051 }
3052 
3053 #endif /* CONFIG_SYSCTL_SYSCALL */
3054 
3055 static int deprecated_sysctl_warning(struct __sysctl_args *args)
3056 {
3057 	static int msg_count;
3058 	int name[CTL_MAXNAME];
3059 	int i;
3060 
3061 	/* Check args->nlen. */
3062 	if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
3063 		return -ENOTDIR;
3064 
3065 	/* Read in the sysctl name for better debug message logging */
3066 	for (i = 0; i < args->nlen; i++)
3067 		if (get_user(name[i], args->name + i))
3068 			return -EFAULT;
3069 
3070 	/* Ignore accesses to kernel.version */
3071 	if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
3072 		return 0;
3073 
3074 	if (msg_count < 5) {
3075 		msg_count++;
3076 		printk(KERN_INFO
3077 			"warning: process `%s' used the deprecated sysctl "
3078 			"system call with ", current->comm);
3079 		for (i = 0; i < args->nlen; i++)
3080 			printk("%d.", name[i]);
3081 		printk("\n");
3082 	}
3083 	return 0;
3084 }
3085 
3086 /*
3087  * No sense putting this after each symbol definition, twice,
3088  * exception granted :-)
3089  */
3090 EXPORT_SYMBOL(proc_dointvec);
3091 EXPORT_SYMBOL(proc_dointvec_jiffies);
3092 EXPORT_SYMBOL(proc_dointvec_minmax);
3093 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3094 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3095 EXPORT_SYMBOL(proc_dostring);
3096 EXPORT_SYMBOL(proc_doulongvec_minmax);
3097 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3098 EXPORT_SYMBOL(register_sysctl_table);
3099 EXPORT_SYMBOL(register_sysctl_paths);
3100 EXPORT_SYMBOL(sysctl_intvec);
3101 EXPORT_SYMBOL(sysctl_jiffies);
3102 EXPORT_SYMBOL(sysctl_ms_jiffies);
3103 EXPORT_SYMBOL(sysctl_string);
3104 EXPORT_SYMBOL(sysctl_data);
3105 EXPORT_SYMBOL(unregister_sysctl_table);
3106