xref: /linux-6.15/kernel/sysctl.c (revision 2e4c77be)
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 #endif
148 
149 #ifdef CONFIG_RT_MUTEXES
150 extern int max_lock_depth;
151 #endif
152 
153 #ifdef CONFIG_PROC_SYSCTL
154 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
155 		  void __user *buffer, size_t *lenp, loff_t *ppos);
156 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
157 			       void __user *buffer, size_t *lenp, loff_t *ppos);
158 #endif
159 
160 static struct ctl_table root_table[];
161 static struct ctl_table_root sysctl_table_root;
162 static struct ctl_table_header root_table_header = {
163 	.count = 1,
164 	.ctl_table = root_table,
165 	.ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
166 	.root = &sysctl_table_root,
167 	.set = &sysctl_table_root.default_set,
168 };
169 static struct ctl_table_root sysctl_table_root = {
170 	.root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
171 	.default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
172 };
173 
174 static struct ctl_table kern_table[];
175 static struct ctl_table vm_table[];
176 static struct ctl_table fs_table[];
177 static struct ctl_table debug_table[];
178 static struct ctl_table dev_table[];
179 extern struct ctl_table random_table[];
180 #ifdef CONFIG_INOTIFY_USER
181 extern struct ctl_table inotify_table[];
182 #endif
183 #ifdef CONFIG_EPOLL
184 extern struct ctl_table epoll_table[];
185 #endif
186 
187 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
188 int sysctl_legacy_va_layout;
189 #endif
190 
191 extern int prove_locking;
192 extern int lock_stat;
193 
194 /* The default sysctl tables: */
195 
196 static struct ctl_table root_table[] = {
197 	{
198 		.ctl_name	= CTL_KERN,
199 		.procname	= "kernel",
200 		.mode		= 0555,
201 		.child		= kern_table,
202 	},
203 	{
204 		.ctl_name	= CTL_VM,
205 		.procname	= "vm",
206 		.mode		= 0555,
207 		.child		= vm_table,
208 	},
209 	{
210 		.ctl_name	= CTL_FS,
211 		.procname	= "fs",
212 		.mode		= 0555,
213 		.child		= fs_table,
214 	},
215 	{
216 		.ctl_name	= CTL_DEBUG,
217 		.procname	= "debug",
218 		.mode		= 0555,
219 		.child		= debug_table,
220 	},
221 	{
222 		.ctl_name	= CTL_DEV,
223 		.procname	= "dev",
224 		.mode		= 0555,
225 		.child		= dev_table,
226 	},
227 /*
228  * NOTE: do not add new entries to this table unless you have read
229  * Documentation/sysctl/ctl_unnumbered.txt
230  */
231 	{ .ctl_name = 0 }
232 };
233 
234 #ifdef CONFIG_SCHED_DEBUG
235 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
236 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
237 static int min_wakeup_granularity_ns;			/* 0 usecs */
238 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
239 #endif
240 
241 static struct ctl_table kern_table[] = {
242 #ifdef CONFIG_SCHED_DEBUG
243 	{
244 		.ctl_name	= CTL_UNNUMBERED,
245 		.procname	= "sched_min_granularity_ns",
246 		.data		= &sysctl_sched_min_granularity,
247 		.maxlen		= sizeof(unsigned int),
248 		.mode		= 0644,
249 		.proc_handler	= &sched_nr_latency_handler,
250 		.strategy	= &sysctl_intvec,
251 		.extra1		= &min_sched_granularity_ns,
252 		.extra2		= &max_sched_granularity_ns,
253 	},
254 	{
255 		.ctl_name	= CTL_UNNUMBERED,
256 		.procname	= "sched_latency_ns",
257 		.data		= &sysctl_sched_latency,
258 		.maxlen		= sizeof(unsigned int),
259 		.mode		= 0644,
260 		.proc_handler	= &sched_nr_latency_handler,
261 		.strategy	= &sysctl_intvec,
262 		.extra1		= &min_sched_granularity_ns,
263 		.extra2		= &max_sched_granularity_ns,
264 	},
265 	{
266 		.ctl_name	= CTL_UNNUMBERED,
267 		.procname	= "sched_wakeup_granularity_ns",
268 		.data		= &sysctl_sched_wakeup_granularity,
269 		.maxlen		= sizeof(unsigned int),
270 		.mode		= 0644,
271 		.proc_handler	= &proc_dointvec_minmax,
272 		.strategy	= &sysctl_intvec,
273 		.extra1		= &min_wakeup_granularity_ns,
274 		.extra2		= &max_wakeup_granularity_ns,
275 	},
276 	{
277 		.ctl_name	= CTL_UNNUMBERED,
278 		.procname	= "sched_shares_ratelimit",
279 		.data		= &sysctl_sched_shares_ratelimit,
280 		.maxlen		= sizeof(unsigned int),
281 		.mode		= 0644,
282 		.proc_handler	= &proc_dointvec,
283 	},
284 	{
285 		.ctl_name	= CTL_UNNUMBERED,
286 		.procname	= "sched_shares_thresh",
287 		.data		= &sysctl_sched_shares_thresh,
288 		.maxlen		= sizeof(unsigned int),
289 		.mode		= 0644,
290 		.proc_handler	= &proc_dointvec_minmax,
291 		.strategy	= &sysctl_intvec,
292 		.extra1		= &zero,
293 	},
294 	{
295 		.ctl_name	= CTL_UNNUMBERED,
296 		.procname	= "sched_child_runs_first",
297 		.data		= &sysctl_sched_child_runs_first,
298 		.maxlen		= sizeof(unsigned int),
299 		.mode		= 0644,
300 		.proc_handler	= &proc_dointvec,
301 	},
302 	{
303 		.ctl_name	= CTL_UNNUMBERED,
304 		.procname	= "sched_features",
305 		.data		= &sysctl_sched_features,
306 		.maxlen		= sizeof(unsigned int),
307 		.mode		= 0644,
308 		.proc_handler	= &proc_dointvec,
309 	},
310 	{
311 		.ctl_name	= CTL_UNNUMBERED,
312 		.procname	= "sched_migration_cost",
313 		.data		= &sysctl_sched_migration_cost,
314 		.maxlen		= sizeof(unsigned int),
315 		.mode		= 0644,
316 		.proc_handler	= &proc_dointvec,
317 	},
318 	{
319 		.ctl_name	= CTL_UNNUMBERED,
320 		.procname	= "sched_nr_migrate",
321 		.data		= &sysctl_sched_nr_migrate,
322 		.maxlen		= sizeof(unsigned int),
323 		.mode		= 0644,
324 		.proc_handler	= &proc_dointvec,
325 	},
326 #endif
327 	{
328 		.ctl_name	= CTL_UNNUMBERED,
329 		.procname	= "sched_rt_period_us",
330 		.data		= &sysctl_sched_rt_period,
331 		.maxlen		= sizeof(unsigned int),
332 		.mode		= 0644,
333 		.proc_handler	= &sched_rt_handler,
334 	},
335 	{
336 		.ctl_name	= CTL_UNNUMBERED,
337 		.procname	= "sched_rt_runtime_us",
338 		.data		= &sysctl_sched_rt_runtime,
339 		.maxlen		= sizeof(int),
340 		.mode		= 0644,
341 		.proc_handler	= &sched_rt_handler,
342 	},
343 	{
344 		.ctl_name	= CTL_UNNUMBERED,
345 		.procname	= "sched_compat_yield",
346 		.data		= &sysctl_sched_compat_yield,
347 		.maxlen		= sizeof(unsigned int),
348 		.mode		= 0644,
349 		.proc_handler	= &proc_dointvec,
350 	},
351 #ifdef CONFIG_PROVE_LOCKING
352 	{
353 		.ctl_name	= CTL_UNNUMBERED,
354 		.procname	= "prove_locking",
355 		.data		= &prove_locking,
356 		.maxlen		= sizeof(int),
357 		.mode		= 0644,
358 		.proc_handler	= &proc_dointvec,
359 	},
360 #endif
361 #ifdef CONFIG_LOCK_STAT
362 	{
363 		.ctl_name	= CTL_UNNUMBERED,
364 		.procname	= "lock_stat",
365 		.data		= &lock_stat,
366 		.maxlen		= sizeof(int),
367 		.mode		= 0644,
368 		.proc_handler	= &proc_dointvec,
369 	},
370 #endif
371 	{
372 		.ctl_name	= KERN_PANIC,
373 		.procname	= "panic",
374 		.data		= &panic_timeout,
375 		.maxlen		= sizeof(int),
376 		.mode		= 0644,
377 		.proc_handler	= &proc_dointvec,
378 	},
379 	{
380 		.ctl_name	= KERN_CORE_USES_PID,
381 		.procname	= "core_uses_pid",
382 		.data		= &core_uses_pid,
383 		.maxlen		= sizeof(int),
384 		.mode		= 0644,
385 		.proc_handler	= &proc_dointvec,
386 	},
387 	{
388 		.ctl_name	= KERN_CORE_PATTERN,
389 		.procname	= "core_pattern",
390 		.data		= core_pattern,
391 		.maxlen		= CORENAME_MAX_SIZE,
392 		.mode		= 0644,
393 		.proc_handler	= &proc_dostring,
394 		.strategy	= &sysctl_string,
395 	},
396 #ifdef CONFIG_PROC_SYSCTL
397 	{
398 		.procname	= "tainted",
399 		.maxlen 	= sizeof(long),
400 		.mode		= 0644,
401 		.proc_handler	= &proc_taint,
402 	},
403 #endif
404 #ifdef CONFIG_LATENCYTOP
405 	{
406 		.procname	= "latencytop",
407 		.data		= &latencytop_enabled,
408 		.maxlen		= sizeof(int),
409 		.mode		= 0644,
410 		.proc_handler	= &proc_dointvec,
411 	},
412 #endif
413 #ifdef CONFIG_BLK_DEV_INITRD
414 	{
415 		.ctl_name	= KERN_REALROOTDEV,
416 		.procname	= "real-root-dev",
417 		.data		= &real_root_dev,
418 		.maxlen		= sizeof(int),
419 		.mode		= 0644,
420 		.proc_handler	= &proc_dointvec,
421 	},
422 #endif
423 	{
424 		.ctl_name	= CTL_UNNUMBERED,
425 		.procname	= "print-fatal-signals",
426 		.data		= &print_fatal_signals,
427 		.maxlen		= sizeof(int),
428 		.mode		= 0644,
429 		.proc_handler	= &proc_dointvec,
430 	},
431 #ifdef CONFIG_SPARC
432 	{
433 		.ctl_name	= KERN_SPARC_REBOOT,
434 		.procname	= "reboot-cmd",
435 		.data		= reboot_command,
436 		.maxlen		= 256,
437 		.mode		= 0644,
438 		.proc_handler	= &proc_dostring,
439 		.strategy	= &sysctl_string,
440 	},
441 	{
442 		.ctl_name	= KERN_SPARC_STOP_A,
443 		.procname	= "stop-a",
444 		.data		= &stop_a_enabled,
445 		.maxlen		= sizeof (int),
446 		.mode		= 0644,
447 		.proc_handler	= &proc_dointvec,
448 	},
449 	{
450 		.ctl_name	= KERN_SPARC_SCONS_PWROFF,
451 		.procname	= "scons-poweroff",
452 		.data		= &scons_pwroff,
453 		.maxlen		= sizeof (int),
454 		.mode		= 0644,
455 		.proc_handler	= &proc_dointvec,
456 	},
457 #endif
458 #ifdef CONFIG_SPARC64
459 	{
460 		.ctl_name	= CTL_UNNUMBERED,
461 		.procname	= "tsb-ratio",
462 		.data		= &sysctl_tsb_ratio,
463 		.maxlen		= sizeof (int),
464 		.mode		= 0644,
465 		.proc_handler	= &proc_dointvec,
466 	},
467 #endif
468 #ifdef __hppa__
469 	{
470 		.ctl_name	= KERN_HPPA_PWRSW,
471 		.procname	= "soft-power",
472 		.data		= &pwrsw_enabled,
473 		.maxlen		= sizeof (int),
474 	 	.mode		= 0644,
475 		.proc_handler	= &proc_dointvec,
476 	},
477 	{
478 		.ctl_name	= KERN_HPPA_UNALIGNED,
479 		.procname	= "unaligned-trap",
480 		.data		= &unaligned_enabled,
481 		.maxlen		= sizeof (int),
482 		.mode		= 0644,
483 		.proc_handler	= &proc_dointvec,
484 	},
485 #endif
486 	{
487 		.ctl_name	= KERN_CTLALTDEL,
488 		.procname	= "ctrl-alt-del",
489 		.data		= &C_A_D,
490 		.maxlen		= sizeof(int),
491 		.mode		= 0644,
492 		.proc_handler	= &proc_dointvec,
493 	},
494 #ifdef CONFIG_FUNCTION_TRACER
495 	{
496 		.ctl_name	= CTL_UNNUMBERED,
497 		.procname	= "ftrace_enabled",
498 		.data		= &ftrace_enabled,
499 		.maxlen		= sizeof(int),
500 		.mode		= 0644,
501 		.proc_handler	= &ftrace_enable_sysctl,
502 	},
503 #endif
504 #ifdef CONFIG_STACK_TRACER
505 	{
506 		.ctl_name	= CTL_UNNUMBERED,
507 		.procname	= "stack_tracer_enabled",
508 		.data		= &stack_tracer_enabled,
509 		.maxlen		= sizeof(int),
510 		.mode		= 0644,
511 		.proc_handler	= &stack_trace_sysctl,
512 	},
513 #endif
514 #ifdef CONFIG_TRACING
515 	{
516 		.ctl_name	= CTL_UNNUMBERED,
517 		.procname	= "ftrace_dump_on_oops",
518 		.data		= &ftrace_dump_on_oops,
519 		.maxlen		= sizeof(int),
520 		.mode		= 0644,
521 		.proc_handler	= &proc_dointvec,
522 	},
523 #endif
524 #ifdef CONFIG_MODULES
525 	{
526 		.ctl_name	= KERN_MODPROBE,
527 		.procname	= "modprobe",
528 		.data		= &modprobe_path,
529 		.maxlen		= KMOD_PATH_LEN,
530 		.mode		= 0644,
531 		.proc_handler	= &proc_dostring,
532 		.strategy	= &sysctl_string,
533 	},
534 #endif
535 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
536 	{
537 		.ctl_name	= KERN_HOTPLUG,
538 		.procname	= "hotplug",
539 		.data		= &uevent_helper,
540 		.maxlen		= UEVENT_HELPER_PATH_LEN,
541 		.mode		= 0644,
542 		.proc_handler	= &proc_dostring,
543 		.strategy	= &sysctl_string,
544 	},
545 #endif
546 #ifdef CONFIG_CHR_DEV_SG
547 	{
548 		.ctl_name	= KERN_SG_BIG_BUFF,
549 		.procname	= "sg-big-buff",
550 		.data		= &sg_big_buff,
551 		.maxlen		= sizeof (int),
552 		.mode		= 0444,
553 		.proc_handler	= &proc_dointvec,
554 	},
555 #endif
556 #ifdef CONFIG_BSD_PROCESS_ACCT
557 	{
558 		.ctl_name	= KERN_ACCT,
559 		.procname	= "acct",
560 		.data		= &acct_parm,
561 		.maxlen		= 3*sizeof(int),
562 		.mode		= 0644,
563 		.proc_handler	= &proc_dointvec,
564 	},
565 #endif
566 #ifdef CONFIG_MAGIC_SYSRQ
567 	{
568 		.ctl_name	= KERN_SYSRQ,
569 		.procname	= "sysrq",
570 		.data		= &__sysrq_enabled,
571 		.maxlen		= sizeof (int),
572 		.mode		= 0644,
573 		.proc_handler	= &proc_dointvec,
574 	},
575 #endif
576 #ifdef CONFIG_PROC_SYSCTL
577 	{
578 		.procname	= "cad_pid",
579 		.data		= NULL,
580 		.maxlen		= sizeof (int),
581 		.mode		= 0600,
582 		.proc_handler	= &proc_do_cad_pid,
583 	},
584 #endif
585 	{
586 		.ctl_name	= KERN_MAX_THREADS,
587 		.procname	= "threads-max",
588 		.data		= &max_threads,
589 		.maxlen		= sizeof(int),
590 		.mode		= 0644,
591 		.proc_handler	= &proc_dointvec,
592 	},
593 	{
594 		.ctl_name	= KERN_RANDOM,
595 		.procname	= "random",
596 		.mode		= 0555,
597 		.child		= random_table,
598 	},
599 	{
600 		.ctl_name	= KERN_OVERFLOWUID,
601 		.procname	= "overflowuid",
602 		.data		= &overflowuid,
603 		.maxlen		= sizeof(int),
604 		.mode		= 0644,
605 		.proc_handler	= &proc_dointvec_minmax,
606 		.strategy	= &sysctl_intvec,
607 		.extra1		= &minolduid,
608 		.extra2		= &maxolduid,
609 	},
610 	{
611 		.ctl_name	= KERN_OVERFLOWGID,
612 		.procname	= "overflowgid",
613 		.data		= &overflowgid,
614 		.maxlen		= sizeof(int),
615 		.mode		= 0644,
616 		.proc_handler	= &proc_dointvec_minmax,
617 		.strategy	= &sysctl_intvec,
618 		.extra1		= &minolduid,
619 		.extra2		= &maxolduid,
620 	},
621 #ifdef CONFIG_S390
622 #ifdef CONFIG_MATHEMU
623 	{
624 		.ctl_name	= KERN_IEEE_EMULATION_WARNINGS,
625 		.procname	= "ieee_emulation_warnings",
626 		.data		= &sysctl_ieee_emulation_warnings,
627 		.maxlen		= sizeof(int),
628 		.mode		= 0644,
629 		.proc_handler	= &proc_dointvec,
630 	},
631 #endif
632 	{
633 		.ctl_name	= KERN_S390_USER_DEBUG_LOGGING,
634 		.procname	= "userprocess_debug",
635 		.data		= &sysctl_userprocess_debug,
636 		.maxlen		= sizeof(int),
637 		.mode		= 0644,
638 		.proc_handler	= &proc_dointvec,
639 	},
640 #endif
641 	{
642 		.ctl_name	= KERN_PIDMAX,
643 		.procname	= "pid_max",
644 		.data		= &pid_max,
645 		.maxlen		= sizeof (int),
646 		.mode		= 0644,
647 		.proc_handler	= &proc_dointvec_minmax,
648 		.strategy	= sysctl_intvec,
649 		.extra1		= &pid_max_min,
650 		.extra2		= &pid_max_max,
651 	},
652 	{
653 		.ctl_name	= KERN_PANIC_ON_OOPS,
654 		.procname	= "panic_on_oops",
655 		.data		= &panic_on_oops,
656 		.maxlen		= sizeof(int),
657 		.mode		= 0644,
658 		.proc_handler	= &proc_dointvec,
659 	},
660 #if defined CONFIG_PRINTK
661 	{
662 		.ctl_name	= KERN_PRINTK,
663 		.procname	= "printk",
664 		.data		= &console_loglevel,
665 		.maxlen		= 4*sizeof(int),
666 		.mode		= 0644,
667 		.proc_handler	= &proc_dointvec,
668 	},
669 	{
670 		.ctl_name	= KERN_PRINTK_RATELIMIT,
671 		.procname	= "printk_ratelimit",
672 		.data		= &printk_ratelimit_state.interval,
673 		.maxlen		= sizeof(int),
674 		.mode		= 0644,
675 		.proc_handler	= &proc_dointvec_jiffies,
676 		.strategy	= &sysctl_jiffies,
677 	},
678 	{
679 		.ctl_name	= KERN_PRINTK_RATELIMIT_BURST,
680 		.procname	= "printk_ratelimit_burst",
681 		.data		= &printk_ratelimit_state.burst,
682 		.maxlen		= sizeof(int),
683 		.mode		= 0644,
684 		.proc_handler	= &proc_dointvec,
685 	},
686 #endif
687 	{
688 		.ctl_name	= KERN_NGROUPS_MAX,
689 		.procname	= "ngroups_max",
690 		.data		= &ngroups_max,
691 		.maxlen		= sizeof (int),
692 		.mode		= 0444,
693 		.proc_handler	= &proc_dointvec,
694 	},
695 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
696 	{
697 		.ctl_name       = KERN_UNKNOWN_NMI_PANIC,
698 		.procname       = "unknown_nmi_panic",
699 		.data           = &unknown_nmi_panic,
700 		.maxlen         = sizeof (int),
701 		.mode           = 0644,
702 		.proc_handler   = &proc_dointvec,
703 	},
704 	{
705 		.procname       = "nmi_watchdog",
706 		.data           = &nmi_watchdog_enabled,
707 		.maxlen         = sizeof (int),
708 		.mode           = 0644,
709 		.proc_handler   = &proc_nmi_enabled,
710 	},
711 #endif
712 #if defined(CONFIG_X86)
713 	{
714 		.ctl_name	= KERN_PANIC_ON_NMI,
715 		.procname	= "panic_on_unrecovered_nmi",
716 		.data		= &panic_on_unrecovered_nmi,
717 		.maxlen		= sizeof(int),
718 		.mode		= 0644,
719 		.proc_handler	= &proc_dointvec,
720 	},
721 	{
722 		.ctl_name	= KERN_BOOTLOADER_TYPE,
723 		.procname	= "bootloader_type",
724 		.data		= &bootloader_type,
725 		.maxlen		= sizeof (int),
726 		.mode		= 0444,
727 		.proc_handler	= &proc_dointvec,
728 	},
729 	{
730 		.ctl_name	= CTL_UNNUMBERED,
731 		.procname	= "kstack_depth_to_print",
732 		.data		= &kstack_depth_to_print,
733 		.maxlen		= sizeof(int),
734 		.mode		= 0644,
735 		.proc_handler	= &proc_dointvec,
736 	},
737 	{
738 		.ctl_name	= CTL_UNNUMBERED,
739 		.procname	= "io_delay_type",
740 		.data		= &io_delay_type,
741 		.maxlen		= sizeof(int),
742 		.mode		= 0644,
743 		.proc_handler	= &proc_dointvec,
744 	},
745 #endif
746 #if defined(CONFIG_MMU)
747 	{
748 		.ctl_name	= KERN_RANDOMIZE,
749 		.procname	= "randomize_va_space",
750 		.data		= &randomize_va_space,
751 		.maxlen		= sizeof(int),
752 		.mode		= 0644,
753 		.proc_handler	= &proc_dointvec,
754 	},
755 #endif
756 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
757 	{
758 		.ctl_name	= KERN_SPIN_RETRY,
759 		.procname	= "spin_retry",
760 		.data		= &spin_retry,
761 		.maxlen		= sizeof (int),
762 		.mode		= 0644,
763 		.proc_handler	= &proc_dointvec,
764 	},
765 #endif
766 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
767 	{
768 		.procname	= "acpi_video_flags",
769 		.data		= &acpi_realmode_flags,
770 		.maxlen		= sizeof (unsigned long),
771 		.mode		= 0644,
772 		.proc_handler	= &proc_doulongvec_minmax,
773 	},
774 #endif
775 #ifdef CONFIG_IA64
776 	{
777 		.ctl_name	= KERN_IA64_UNALIGNED,
778 		.procname	= "ignore-unaligned-usertrap",
779 		.data		= &no_unaligned_warning,
780 		.maxlen		= sizeof (int),
781 	 	.mode		= 0644,
782 		.proc_handler	= &proc_dointvec,
783 	},
784 #endif
785 #ifdef CONFIG_DETECT_SOFTLOCKUP
786 	{
787 		.ctl_name	= CTL_UNNUMBERED,
788 		.procname	= "softlockup_panic",
789 		.data		= &softlockup_panic,
790 		.maxlen		= sizeof(int),
791 		.mode		= 0644,
792 		.proc_handler	= &proc_dointvec_minmax,
793 		.strategy	= &sysctl_intvec,
794 		.extra1		= &zero,
795 		.extra2		= &one,
796 	},
797 	{
798 		.ctl_name	= CTL_UNNUMBERED,
799 		.procname	= "softlockup_thresh",
800 		.data		= &softlockup_thresh,
801 		.maxlen		= sizeof(int),
802 		.mode		= 0644,
803 		.proc_handler	= &proc_dointvec_minmax,
804 		.strategy	= &sysctl_intvec,
805 		.extra1		= &neg_one,
806 		.extra2		= &sixty,
807 	},
808 	{
809 		.ctl_name	= CTL_UNNUMBERED,
810 		.procname	= "hung_task_check_count",
811 		.data		= &sysctl_hung_task_check_count,
812 		.maxlen		= sizeof(unsigned long),
813 		.mode		= 0644,
814 		.proc_handler	= &proc_doulongvec_minmax,
815 		.strategy	= &sysctl_intvec,
816 	},
817 	{
818 		.ctl_name	= CTL_UNNUMBERED,
819 		.procname	= "hung_task_timeout_secs",
820 		.data		= &sysctl_hung_task_timeout_secs,
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_warnings",
829 		.data		= &sysctl_hung_task_warnings,
830 		.maxlen		= sizeof(unsigned long),
831 		.mode		= 0644,
832 		.proc_handler	= &proc_doulongvec_minmax,
833 		.strategy	= &sysctl_intvec,
834 	},
835 #endif
836 #ifdef CONFIG_COMPAT
837 	{
838 		.ctl_name	= KERN_COMPAT_LOG,
839 		.procname	= "compat-log",
840 		.data		= &compat_log,
841 		.maxlen		= sizeof (int),
842 	 	.mode		= 0644,
843 		.proc_handler	= &proc_dointvec,
844 	},
845 #endif
846 #ifdef CONFIG_RT_MUTEXES
847 	{
848 		.ctl_name	= KERN_MAX_LOCK_DEPTH,
849 		.procname	= "max_lock_depth",
850 		.data		= &max_lock_depth,
851 		.maxlen		= sizeof(int),
852 		.mode		= 0644,
853 		.proc_handler	= &proc_dointvec,
854 	},
855 #endif
856 	{
857 		.ctl_name	= CTL_UNNUMBERED,
858 		.procname	= "poweroff_cmd",
859 		.data		= &poweroff_cmd,
860 		.maxlen		= POWEROFF_CMD_PATH_LEN,
861 		.mode		= 0644,
862 		.proc_handler	= &proc_dostring,
863 		.strategy	= &sysctl_string,
864 	},
865 #ifdef CONFIG_KEYS
866 	{
867 		.ctl_name	= CTL_UNNUMBERED,
868 		.procname	= "keys",
869 		.mode		= 0555,
870 		.child		= key_sysctls,
871 	},
872 #endif
873 #ifdef CONFIG_RCU_TORTURE_TEST
874 	{
875 		.ctl_name       = CTL_UNNUMBERED,
876 		.procname       = "rcutorture_runnable",
877 		.data           = &rcutorture_runnable,
878 		.maxlen         = sizeof(int),
879 		.mode           = 0644,
880 		.proc_handler   = &proc_dointvec,
881 	},
882 #endif
883 #ifdef CONFIG_UNEVICTABLE_LRU
884 	{
885 		.ctl_name	= CTL_UNNUMBERED,
886 		.procname	= "scan_unevictable_pages",
887 		.data		= &scan_unevictable_pages,
888 		.maxlen		= sizeof(scan_unevictable_pages),
889 		.mode		= 0644,
890 		.proc_handler	= &scan_unevictable_handler,
891 	},
892 #endif
893 /*
894  * NOTE: do not add new entries to this table unless you have read
895  * Documentation/sysctl/ctl_unnumbered.txt
896  */
897 	{ .ctl_name = 0 }
898 };
899 
900 static struct ctl_table vm_table[] = {
901 	{
902 		.ctl_name	= VM_OVERCOMMIT_MEMORY,
903 		.procname	= "overcommit_memory",
904 		.data		= &sysctl_overcommit_memory,
905 		.maxlen		= sizeof(sysctl_overcommit_memory),
906 		.mode		= 0644,
907 		.proc_handler	= &proc_dointvec,
908 	},
909 	{
910 		.ctl_name	= VM_PANIC_ON_OOM,
911 		.procname	= "panic_on_oom",
912 		.data		= &sysctl_panic_on_oom,
913 		.maxlen		= sizeof(sysctl_panic_on_oom),
914 		.mode		= 0644,
915 		.proc_handler	= &proc_dointvec,
916 	},
917 	{
918 		.ctl_name	= CTL_UNNUMBERED,
919 		.procname	= "oom_kill_allocating_task",
920 		.data		= &sysctl_oom_kill_allocating_task,
921 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
922 		.mode		= 0644,
923 		.proc_handler	= &proc_dointvec,
924 	},
925 	{
926 		.ctl_name	= CTL_UNNUMBERED,
927 		.procname	= "oom_dump_tasks",
928 		.data		= &sysctl_oom_dump_tasks,
929 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
930 		.mode		= 0644,
931 		.proc_handler	= &proc_dointvec,
932 	},
933 	{
934 		.ctl_name	= VM_OVERCOMMIT_RATIO,
935 		.procname	= "overcommit_ratio",
936 		.data		= &sysctl_overcommit_ratio,
937 		.maxlen		= sizeof(sysctl_overcommit_ratio),
938 		.mode		= 0644,
939 		.proc_handler	= &proc_dointvec,
940 	},
941 	{
942 		.ctl_name	= VM_PAGE_CLUSTER,
943 		.procname	= "page-cluster",
944 		.data		= &page_cluster,
945 		.maxlen		= sizeof(int),
946 		.mode		= 0644,
947 		.proc_handler	= &proc_dointvec,
948 	},
949 	{
950 		.ctl_name	= VM_DIRTY_BACKGROUND,
951 		.procname	= "dirty_background_ratio",
952 		.data		= &dirty_background_ratio,
953 		.maxlen		= sizeof(dirty_background_ratio),
954 		.mode		= 0644,
955 		.proc_handler	= &dirty_background_ratio_handler,
956 		.strategy	= &sysctl_intvec,
957 		.extra1		= &zero,
958 		.extra2		= &one_hundred,
959 	},
960 	{
961 		.ctl_name	= CTL_UNNUMBERED,
962 		.procname	= "dirty_background_bytes",
963 		.data		= &dirty_background_bytes,
964 		.maxlen		= sizeof(dirty_background_bytes),
965 		.mode		= 0644,
966 		.proc_handler	= &dirty_background_bytes_handler,
967 		.strategy	= &sysctl_intvec,
968 		.extra1		= &one,
969 	},
970 	{
971 		.ctl_name	= VM_DIRTY_RATIO,
972 		.procname	= "dirty_ratio",
973 		.data		= &vm_dirty_ratio,
974 		.maxlen		= sizeof(vm_dirty_ratio),
975 		.mode		= 0644,
976 		.proc_handler	= &dirty_ratio_handler,
977 		.strategy	= &sysctl_intvec,
978 		.extra1		= &zero,
979 		.extra2		= &one_hundred,
980 	},
981 	{
982 		.ctl_name	= CTL_UNNUMBERED,
983 		.procname	= "dirty_bytes",
984 		.data		= &vm_dirty_bytes,
985 		.maxlen		= sizeof(vm_dirty_bytes),
986 		.mode		= 0644,
987 		.proc_handler	= &dirty_bytes_handler,
988 		.strategy	= &sysctl_intvec,
989 		.extra1		= &one,
990 	},
991 	{
992 		.procname	= "dirty_writeback_centisecs",
993 		.data		= &dirty_writeback_interval,
994 		.maxlen		= sizeof(dirty_writeback_interval),
995 		.mode		= 0644,
996 		.proc_handler	= &dirty_writeback_centisecs_handler,
997 	},
998 	{
999 		.procname	= "dirty_expire_centisecs",
1000 		.data		= &dirty_expire_interval,
1001 		.maxlen		= sizeof(dirty_expire_interval),
1002 		.mode		= 0644,
1003 		.proc_handler	= &proc_dointvec_userhz_jiffies,
1004 	},
1005 	{
1006 		.ctl_name	= VM_NR_PDFLUSH_THREADS,
1007 		.procname	= "nr_pdflush_threads",
1008 		.data		= &nr_pdflush_threads,
1009 		.maxlen		= sizeof nr_pdflush_threads,
1010 		.mode		= 0444 /* read-only*/,
1011 		.proc_handler	= &proc_dointvec,
1012 	},
1013 	{
1014 		.ctl_name	= VM_SWAPPINESS,
1015 		.procname	= "swappiness",
1016 		.data		= &vm_swappiness,
1017 		.maxlen		= sizeof(vm_swappiness),
1018 		.mode		= 0644,
1019 		.proc_handler	= &proc_dointvec_minmax,
1020 		.strategy	= &sysctl_intvec,
1021 		.extra1		= &zero,
1022 		.extra2		= &one_hundred,
1023 	},
1024 #ifdef CONFIG_HUGETLB_PAGE
1025 	 {
1026 		.procname	= "nr_hugepages",
1027 		.data		= NULL,
1028 		.maxlen		= sizeof(unsigned long),
1029 		.mode		= 0644,
1030 		.proc_handler	= &hugetlb_sysctl_handler,
1031 		.extra1		= (void *)&hugetlb_zero,
1032 		.extra2		= (void *)&hugetlb_infinity,
1033 	 },
1034 	 {
1035 		.ctl_name	= VM_HUGETLB_GROUP,
1036 		.procname	= "hugetlb_shm_group",
1037 		.data		= &sysctl_hugetlb_shm_group,
1038 		.maxlen		= sizeof(gid_t),
1039 		.mode		= 0644,
1040 		.proc_handler	= &proc_dointvec,
1041 	 },
1042 	 {
1043 		.ctl_name	= CTL_UNNUMBERED,
1044 		.procname	= "hugepages_treat_as_movable",
1045 		.data		= &hugepages_treat_as_movable,
1046 		.maxlen		= sizeof(int),
1047 		.mode		= 0644,
1048 		.proc_handler	= &hugetlb_treat_movable_handler,
1049 	},
1050 	{
1051 		.ctl_name	= CTL_UNNUMBERED,
1052 		.procname	= "nr_overcommit_hugepages",
1053 		.data		= NULL,
1054 		.maxlen		= sizeof(unsigned long),
1055 		.mode		= 0644,
1056 		.proc_handler	= &hugetlb_overcommit_handler,
1057 		.extra1		= (void *)&hugetlb_zero,
1058 		.extra2		= (void *)&hugetlb_infinity,
1059 	},
1060 #endif
1061 	{
1062 		.ctl_name	= VM_LOWMEM_RESERVE_RATIO,
1063 		.procname	= "lowmem_reserve_ratio",
1064 		.data		= &sysctl_lowmem_reserve_ratio,
1065 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1066 		.mode		= 0644,
1067 		.proc_handler	= &lowmem_reserve_ratio_sysctl_handler,
1068 		.strategy	= &sysctl_intvec,
1069 	},
1070 	{
1071 		.ctl_name	= VM_DROP_PAGECACHE,
1072 		.procname	= "drop_caches",
1073 		.data		= &sysctl_drop_caches,
1074 		.maxlen		= sizeof(int),
1075 		.mode		= 0644,
1076 		.proc_handler	= drop_caches_sysctl_handler,
1077 		.strategy	= &sysctl_intvec,
1078 	},
1079 	{
1080 		.ctl_name	= VM_MIN_FREE_KBYTES,
1081 		.procname	= "min_free_kbytes",
1082 		.data		= &min_free_kbytes,
1083 		.maxlen		= sizeof(min_free_kbytes),
1084 		.mode		= 0644,
1085 		.proc_handler	= &min_free_kbytes_sysctl_handler,
1086 		.strategy	= &sysctl_intvec,
1087 		.extra1		= &zero,
1088 	},
1089 	{
1090 		.ctl_name	= VM_PERCPU_PAGELIST_FRACTION,
1091 		.procname	= "percpu_pagelist_fraction",
1092 		.data		= &percpu_pagelist_fraction,
1093 		.maxlen		= sizeof(percpu_pagelist_fraction),
1094 		.mode		= 0644,
1095 		.proc_handler	= &percpu_pagelist_fraction_sysctl_handler,
1096 		.strategy	= &sysctl_intvec,
1097 		.extra1		= &min_percpu_pagelist_fract,
1098 	},
1099 #ifdef CONFIG_MMU
1100 	{
1101 		.ctl_name	= VM_MAX_MAP_COUNT,
1102 		.procname	= "max_map_count",
1103 		.data		= &sysctl_max_map_count,
1104 		.maxlen		= sizeof(sysctl_max_map_count),
1105 		.mode		= 0644,
1106 		.proc_handler	= &proc_dointvec
1107 	},
1108 #else
1109 	{
1110 		.ctl_name	= CTL_UNNUMBERED,
1111 		.procname	= "nr_trim_pages",
1112 		.data		= &sysctl_nr_trim_pages,
1113 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1114 		.mode		= 0644,
1115 		.proc_handler	= &proc_dointvec_minmax,
1116 		.strategy	= &sysctl_intvec,
1117 		.extra1		= &zero,
1118 	},
1119 #endif
1120 	{
1121 		.ctl_name	= VM_LAPTOP_MODE,
1122 		.procname	= "laptop_mode",
1123 		.data		= &laptop_mode,
1124 		.maxlen		= sizeof(laptop_mode),
1125 		.mode		= 0644,
1126 		.proc_handler	= &proc_dointvec_jiffies,
1127 		.strategy	= &sysctl_jiffies,
1128 	},
1129 	{
1130 		.ctl_name	= VM_BLOCK_DUMP,
1131 		.procname	= "block_dump",
1132 		.data		= &block_dump,
1133 		.maxlen		= sizeof(block_dump),
1134 		.mode		= 0644,
1135 		.proc_handler	= &proc_dointvec,
1136 		.strategy	= &sysctl_intvec,
1137 		.extra1		= &zero,
1138 	},
1139 	{
1140 		.ctl_name	= VM_VFS_CACHE_PRESSURE,
1141 		.procname	= "vfs_cache_pressure",
1142 		.data		= &sysctl_vfs_cache_pressure,
1143 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1144 		.mode		= 0644,
1145 		.proc_handler	= &proc_dointvec,
1146 		.strategy	= &sysctl_intvec,
1147 		.extra1		= &zero,
1148 	},
1149 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1150 	{
1151 		.ctl_name	= VM_LEGACY_VA_LAYOUT,
1152 		.procname	= "legacy_va_layout",
1153 		.data		= &sysctl_legacy_va_layout,
1154 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1155 		.mode		= 0644,
1156 		.proc_handler	= &proc_dointvec,
1157 		.strategy	= &sysctl_intvec,
1158 		.extra1		= &zero,
1159 	},
1160 #endif
1161 #ifdef CONFIG_NUMA
1162 	{
1163 		.ctl_name	= VM_ZONE_RECLAIM_MODE,
1164 		.procname	= "zone_reclaim_mode",
1165 		.data		= &zone_reclaim_mode,
1166 		.maxlen		= sizeof(zone_reclaim_mode),
1167 		.mode		= 0644,
1168 		.proc_handler	= &proc_dointvec,
1169 		.strategy	= &sysctl_intvec,
1170 		.extra1		= &zero,
1171 	},
1172 	{
1173 		.ctl_name	= VM_MIN_UNMAPPED,
1174 		.procname	= "min_unmapped_ratio",
1175 		.data		= &sysctl_min_unmapped_ratio,
1176 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1177 		.mode		= 0644,
1178 		.proc_handler	= &sysctl_min_unmapped_ratio_sysctl_handler,
1179 		.strategy	= &sysctl_intvec,
1180 		.extra1		= &zero,
1181 		.extra2		= &one_hundred,
1182 	},
1183 	{
1184 		.ctl_name	= VM_MIN_SLAB,
1185 		.procname	= "min_slab_ratio",
1186 		.data		= &sysctl_min_slab_ratio,
1187 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1188 		.mode		= 0644,
1189 		.proc_handler	= &sysctl_min_slab_ratio_sysctl_handler,
1190 		.strategy	= &sysctl_intvec,
1191 		.extra1		= &zero,
1192 		.extra2		= &one_hundred,
1193 	},
1194 #endif
1195 #ifdef CONFIG_SMP
1196 	{
1197 		.ctl_name	= CTL_UNNUMBERED,
1198 		.procname	= "stat_interval",
1199 		.data		= &sysctl_stat_interval,
1200 		.maxlen		= sizeof(sysctl_stat_interval),
1201 		.mode		= 0644,
1202 		.proc_handler	= &proc_dointvec_jiffies,
1203 		.strategy	= &sysctl_jiffies,
1204 	},
1205 #endif
1206 #ifdef CONFIG_SECURITY
1207 	{
1208 		.ctl_name	= CTL_UNNUMBERED,
1209 		.procname	= "mmap_min_addr",
1210 		.data		= &mmap_min_addr,
1211 		.maxlen         = sizeof(unsigned long),
1212 		.mode		= 0644,
1213 		.proc_handler	= &proc_doulongvec_minmax,
1214 	},
1215 #endif
1216 #ifdef CONFIG_NUMA
1217 	{
1218 		.ctl_name	= CTL_UNNUMBERED,
1219 		.procname	= "numa_zonelist_order",
1220 		.data		= &numa_zonelist_order,
1221 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1222 		.mode		= 0644,
1223 		.proc_handler	= &numa_zonelist_order_handler,
1224 		.strategy	= &sysctl_string,
1225 	},
1226 #endif
1227 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1228    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1229 	{
1230 		.ctl_name	= VM_VDSO_ENABLED,
1231 		.procname	= "vdso_enabled",
1232 		.data		= &vdso_enabled,
1233 		.maxlen		= sizeof(vdso_enabled),
1234 		.mode		= 0644,
1235 		.proc_handler	= &proc_dointvec,
1236 		.strategy	= &sysctl_intvec,
1237 		.extra1		= &zero,
1238 	},
1239 #endif
1240 #ifdef CONFIG_HIGHMEM
1241 	{
1242 		.ctl_name	= CTL_UNNUMBERED,
1243 		.procname	= "highmem_is_dirtyable",
1244 		.data		= &vm_highmem_is_dirtyable,
1245 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1246 		.mode		= 0644,
1247 		.proc_handler	= &proc_dointvec_minmax,
1248 		.strategy	= &sysctl_intvec,
1249 		.extra1		= &zero,
1250 		.extra2		= &one,
1251 	},
1252 #endif
1253 /*
1254  * NOTE: do not add new entries to this table unless you have read
1255  * Documentation/sysctl/ctl_unnumbered.txt
1256  */
1257 	{ .ctl_name = 0 }
1258 };
1259 
1260 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1261 static struct ctl_table binfmt_misc_table[] = {
1262 	{ .ctl_name = 0 }
1263 };
1264 #endif
1265 
1266 static struct ctl_table fs_table[] = {
1267 	{
1268 		.ctl_name	= FS_NRINODE,
1269 		.procname	= "inode-nr",
1270 		.data		= &inodes_stat,
1271 		.maxlen		= 2*sizeof(int),
1272 		.mode		= 0444,
1273 		.proc_handler	= &proc_dointvec,
1274 	},
1275 	{
1276 		.ctl_name	= FS_STATINODE,
1277 		.procname	= "inode-state",
1278 		.data		= &inodes_stat,
1279 		.maxlen		= 7*sizeof(int),
1280 		.mode		= 0444,
1281 		.proc_handler	= &proc_dointvec,
1282 	},
1283 	{
1284 		.procname	= "file-nr",
1285 		.data		= &files_stat,
1286 		.maxlen		= 3*sizeof(int),
1287 		.mode		= 0444,
1288 		.proc_handler	= &proc_nr_files,
1289 	},
1290 	{
1291 		.ctl_name	= FS_MAXFILE,
1292 		.procname	= "file-max",
1293 		.data		= &files_stat.max_files,
1294 		.maxlen		= sizeof(int),
1295 		.mode		= 0644,
1296 		.proc_handler	= &proc_dointvec,
1297 	},
1298 	{
1299 		.ctl_name	= CTL_UNNUMBERED,
1300 		.procname	= "nr_open",
1301 		.data		= &sysctl_nr_open,
1302 		.maxlen		= sizeof(int),
1303 		.mode		= 0644,
1304 		.proc_handler	= &proc_dointvec_minmax,
1305 		.extra1		= &sysctl_nr_open_min,
1306 		.extra2		= &sysctl_nr_open_max,
1307 	},
1308 	{
1309 		.ctl_name	= FS_DENTRY,
1310 		.procname	= "dentry-state",
1311 		.data		= &dentry_stat,
1312 		.maxlen		= 6*sizeof(int),
1313 		.mode		= 0444,
1314 		.proc_handler	= &proc_dointvec,
1315 	},
1316 	{
1317 		.ctl_name	= FS_OVERFLOWUID,
1318 		.procname	= "overflowuid",
1319 		.data		= &fs_overflowuid,
1320 		.maxlen		= sizeof(int),
1321 		.mode		= 0644,
1322 		.proc_handler	= &proc_dointvec_minmax,
1323 		.strategy	= &sysctl_intvec,
1324 		.extra1		= &minolduid,
1325 		.extra2		= &maxolduid,
1326 	},
1327 	{
1328 		.ctl_name	= FS_OVERFLOWGID,
1329 		.procname	= "overflowgid",
1330 		.data		= &fs_overflowgid,
1331 		.maxlen		= sizeof(int),
1332 		.mode		= 0644,
1333 		.proc_handler	= &proc_dointvec_minmax,
1334 		.strategy	= &sysctl_intvec,
1335 		.extra1		= &minolduid,
1336 		.extra2		= &maxolduid,
1337 	},
1338 #ifdef CONFIG_FILE_LOCKING
1339 	{
1340 		.ctl_name	= FS_LEASES,
1341 		.procname	= "leases-enable",
1342 		.data		= &leases_enable,
1343 		.maxlen		= sizeof(int),
1344 		.mode		= 0644,
1345 		.proc_handler	= &proc_dointvec,
1346 	},
1347 #endif
1348 #ifdef CONFIG_DNOTIFY
1349 	{
1350 		.ctl_name	= FS_DIR_NOTIFY,
1351 		.procname	= "dir-notify-enable",
1352 		.data		= &dir_notify_enable,
1353 		.maxlen		= sizeof(int),
1354 		.mode		= 0644,
1355 		.proc_handler	= &proc_dointvec,
1356 	},
1357 #endif
1358 #ifdef CONFIG_MMU
1359 #ifdef CONFIG_FILE_LOCKING
1360 	{
1361 		.ctl_name	= FS_LEASE_TIME,
1362 		.procname	= "lease-break-time",
1363 		.data		= &lease_break_time,
1364 		.maxlen		= sizeof(int),
1365 		.mode		= 0644,
1366 		.proc_handler	= &proc_dointvec_minmax,
1367 		.strategy	= &sysctl_intvec,
1368 		.extra1		= &zero,
1369 		.extra2		= &two,
1370 	},
1371 #endif
1372 #ifdef CONFIG_AIO
1373 	{
1374 		.procname	= "aio-nr",
1375 		.data		= &aio_nr,
1376 		.maxlen		= sizeof(aio_nr),
1377 		.mode		= 0444,
1378 		.proc_handler	= &proc_doulongvec_minmax,
1379 	},
1380 	{
1381 		.procname	= "aio-max-nr",
1382 		.data		= &aio_max_nr,
1383 		.maxlen		= sizeof(aio_max_nr),
1384 		.mode		= 0644,
1385 		.proc_handler	= &proc_doulongvec_minmax,
1386 	},
1387 #endif /* CONFIG_AIO */
1388 #ifdef CONFIG_INOTIFY_USER
1389 	{
1390 		.ctl_name	= FS_INOTIFY,
1391 		.procname	= "inotify",
1392 		.mode		= 0555,
1393 		.child		= inotify_table,
1394 	},
1395 #endif
1396 #ifdef CONFIG_EPOLL
1397 	{
1398 		.procname	= "epoll",
1399 		.mode		= 0555,
1400 		.child		= epoll_table,
1401 	},
1402 #endif
1403 #endif
1404 	{
1405 		.ctl_name	= KERN_SETUID_DUMPABLE,
1406 		.procname	= "suid_dumpable",
1407 		.data		= &suid_dumpable,
1408 		.maxlen		= sizeof(int),
1409 		.mode		= 0644,
1410 		.proc_handler	= &proc_dointvec,
1411 	},
1412 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1413 	{
1414 		.ctl_name	= CTL_UNNUMBERED,
1415 		.procname	= "binfmt_misc",
1416 		.mode		= 0555,
1417 		.child		= binfmt_misc_table,
1418 	},
1419 #endif
1420 /*
1421  * NOTE: do not add new entries to this table unless you have read
1422  * Documentation/sysctl/ctl_unnumbered.txt
1423  */
1424 	{ .ctl_name = 0 }
1425 };
1426 
1427 static struct ctl_table debug_table[] = {
1428 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1429 	{
1430 		.ctl_name	= CTL_UNNUMBERED,
1431 		.procname	= "exception-trace",
1432 		.data		= &show_unhandled_signals,
1433 		.maxlen		= sizeof(int),
1434 		.mode		= 0644,
1435 		.proc_handler	= proc_dointvec
1436 	},
1437 #endif
1438 	{ .ctl_name = 0 }
1439 };
1440 
1441 static struct ctl_table dev_table[] = {
1442 	{ .ctl_name = 0 }
1443 };
1444 
1445 static DEFINE_SPINLOCK(sysctl_lock);
1446 
1447 /* called under sysctl_lock */
1448 static int use_table(struct ctl_table_header *p)
1449 {
1450 	if (unlikely(p->unregistering))
1451 		return 0;
1452 	p->used++;
1453 	return 1;
1454 }
1455 
1456 /* called under sysctl_lock */
1457 static void unuse_table(struct ctl_table_header *p)
1458 {
1459 	if (!--p->used)
1460 		if (unlikely(p->unregistering))
1461 			complete(p->unregistering);
1462 }
1463 
1464 /* called under sysctl_lock, will reacquire if has to wait */
1465 static void start_unregistering(struct ctl_table_header *p)
1466 {
1467 	/*
1468 	 * if p->used is 0, nobody will ever touch that entry again;
1469 	 * we'll eliminate all paths to it before dropping sysctl_lock
1470 	 */
1471 	if (unlikely(p->used)) {
1472 		struct completion wait;
1473 		init_completion(&wait);
1474 		p->unregistering = &wait;
1475 		spin_unlock(&sysctl_lock);
1476 		wait_for_completion(&wait);
1477 		spin_lock(&sysctl_lock);
1478 	} else {
1479 		/* anything non-NULL; we'll never dereference it */
1480 		p->unregistering = ERR_PTR(-EINVAL);
1481 	}
1482 	/*
1483 	 * do not remove from the list until nobody holds it; walking the
1484 	 * list in do_sysctl() relies on that.
1485 	 */
1486 	list_del_init(&p->ctl_entry);
1487 }
1488 
1489 void sysctl_head_get(struct ctl_table_header *head)
1490 {
1491 	spin_lock(&sysctl_lock);
1492 	head->count++;
1493 	spin_unlock(&sysctl_lock);
1494 }
1495 
1496 void sysctl_head_put(struct ctl_table_header *head)
1497 {
1498 	spin_lock(&sysctl_lock);
1499 	if (!--head->count)
1500 		kfree(head);
1501 	spin_unlock(&sysctl_lock);
1502 }
1503 
1504 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1505 {
1506 	if (!head)
1507 		BUG();
1508 	spin_lock(&sysctl_lock);
1509 	if (!use_table(head))
1510 		head = ERR_PTR(-ENOENT);
1511 	spin_unlock(&sysctl_lock);
1512 	return head;
1513 }
1514 
1515 void sysctl_head_finish(struct ctl_table_header *head)
1516 {
1517 	if (!head)
1518 		return;
1519 	spin_lock(&sysctl_lock);
1520 	unuse_table(head);
1521 	spin_unlock(&sysctl_lock);
1522 }
1523 
1524 static struct ctl_table_set *
1525 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1526 {
1527 	struct ctl_table_set *set = &root->default_set;
1528 	if (root->lookup)
1529 		set = root->lookup(root, namespaces);
1530 	return set;
1531 }
1532 
1533 static struct list_head *
1534 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1535 {
1536 	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1537 	return &set->list;
1538 }
1539 
1540 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1541 					    struct ctl_table_header *prev)
1542 {
1543 	struct ctl_table_root *root;
1544 	struct list_head *header_list;
1545 	struct ctl_table_header *head;
1546 	struct list_head *tmp;
1547 
1548 	spin_lock(&sysctl_lock);
1549 	if (prev) {
1550 		head = prev;
1551 		tmp = &prev->ctl_entry;
1552 		unuse_table(prev);
1553 		goto next;
1554 	}
1555 	tmp = &root_table_header.ctl_entry;
1556 	for (;;) {
1557 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1558 
1559 		if (!use_table(head))
1560 			goto next;
1561 		spin_unlock(&sysctl_lock);
1562 		return head;
1563 	next:
1564 		root = head->root;
1565 		tmp = tmp->next;
1566 		header_list = lookup_header_list(root, namespaces);
1567 		if (tmp != header_list)
1568 			continue;
1569 
1570 		do {
1571 			root = list_entry(root->root_list.next,
1572 					struct ctl_table_root, root_list);
1573 			if (root == &sysctl_table_root)
1574 				goto out;
1575 			header_list = lookup_header_list(root, namespaces);
1576 		} while (list_empty(header_list));
1577 		tmp = header_list->next;
1578 	}
1579 out:
1580 	spin_unlock(&sysctl_lock);
1581 	return NULL;
1582 }
1583 
1584 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1585 {
1586 	return __sysctl_head_next(current->nsproxy, prev);
1587 }
1588 
1589 void register_sysctl_root(struct ctl_table_root *root)
1590 {
1591 	spin_lock(&sysctl_lock);
1592 	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1593 	spin_unlock(&sysctl_lock);
1594 }
1595 
1596 #ifdef CONFIG_SYSCTL_SYSCALL
1597 /* Perform the actual read/write of a sysctl table entry. */
1598 static int do_sysctl_strategy(struct ctl_table_root *root,
1599 			struct ctl_table *table,
1600 			void __user *oldval, size_t __user *oldlenp,
1601 			void __user *newval, size_t newlen)
1602 {
1603 	int op = 0, rc;
1604 
1605 	if (oldval)
1606 		op |= MAY_READ;
1607 	if (newval)
1608 		op |= MAY_WRITE;
1609 	if (sysctl_perm(root, table, op))
1610 		return -EPERM;
1611 
1612 	if (table->strategy) {
1613 		rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1614 		if (rc < 0)
1615 			return rc;
1616 		if (rc > 0)
1617 			return 0;
1618 	}
1619 
1620 	/* If there is no strategy routine, or if the strategy returns
1621 	 * zero, proceed with automatic r/w */
1622 	if (table->data && table->maxlen) {
1623 		rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1624 		if (rc < 0)
1625 			return rc;
1626 	}
1627 	return 0;
1628 }
1629 
1630 static int parse_table(int __user *name, int nlen,
1631 		       void __user *oldval, size_t __user *oldlenp,
1632 		       void __user *newval, size_t newlen,
1633 		       struct ctl_table_root *root,
1634 		       struct ctl_table *table)
1635 {
1636 	int n;
1637 repeat:
1638 	if (!nlen)
1639 		return -ENOTDIR;
1640 	if (get_user(n, name))
1641 		return -EFAULT;
1642 	for ( ; table->ctl_name || table->procname; table++) {
1643 		if (!table->ctl_name)
1644 			continue;
1645 		if (n == table->ctl_name) {
1646 			int error;
1647 			if (table->child) {
1648 				if (sysctl_perm(root, table, MAY_EXEC))
1649 					return -EPERM;
1650 				name++;
1651 				nlen--;
1652 				table = table->child;
1653 				goto repeat;
1654 			}
1655 			error = do_sysctl_strategy(root, table,
1656 						   oldval, oldlenp,
1657 						   newval, newlen);
1658 			return error;
1659 		}
1660 	}
1661 	return -ENOTDIR;
1662 }
1663 
1664 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1665 	       void __user *newval, size_t newlen)
1666 {
1667 	struct ctl_table_header *head;
1668 	int error = -ENOTDIR;
1669 
1670 	if (nlen <= 0 || nlen >= CTL_MAXNAME)
1671 		return -ENOTDIR;
1672 	if (oldval) {
1673 		int old_len;
1674 		if (!oldlenp || get_user(old_len, oldlenp))
1675 			return -EFAULT;
1676 	}
1677 
1678 	for (head = sysctl_head_next(NULL); head;
1679 			head = sysctl_head_next(head)) {
1680 		error = parse_table(name, nlen, oldval, oldlenp,
1681 					newval, newlen,
1682 					head->root, head->ctl_table);
1683 		if (error != -ENOTDIR) {
1684 			sysctl_head_finish(head);
1685 			break;
1686 		}
1687 	}
1688 	return error;
1689 }
1690 
1691 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1692 {
1693 	struct __sysctl_args tmp;
1694 	int error;
1695 
1696 	if (copy_from_user(&tmp, args, sizeof(tmp)))
1697 		return -EFAULT;
1698 
1699 	error = deprecated_sysctl_warning(&tmp);
1700 	if (error)
1701 		goto out;
1702 
1703 	lock_kernel();
1704 	error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1705 			  tmp.newval, tmp.newlen);
1706 	unlock_kernel();
1707 out:
1708 	return error;
1709 }
1710 #endif /* CONFIG_SYSCTL_SYSCALL */
1711 
1712 /*
1713  * sysctl_perm does NOT grant the superuser all rights automatically, because
1714  * some sysctl variables are readonly even to root.
1715  */
1716 
1717 static int test_perm(int mode, int op)
1718 {
1719 	if (!current_euid())
1720 		mode >>= 6;
1721 	else if (in_egroup_p(0))
1722 		mode >>= 3;
1723 	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1724 		return 0;
1725 	return -EACCES;
1726 }
1727 
1728 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1729 {
1730 	int error;
1731 	int mode;
1732 
1733 	error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1734 	if (error)
1735 		return error;
1736 
1737 	if (root->permissions)
1738 		mode = root->permissions(root, current->nsproxy, table);
1739 	else
1740 		mode = table->mode;
1741 
1742 	return test_perm(mode, op);
1743 }
1744 
1745 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1746 {
1747 	for (; table->ctl_name || table->procname; table++) {
1748 		table->parent = parent;
1749 		if (table->child)
1750 			sysctl_set_parent(table, table->child);
1751 	}
1752 }
1753 
1754 static __init int sysctl_init(void)
1755 {
1756 	sysctl_set_parent(NULL, root_table);
1757 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1758 	{
1759 		int err;
1760 		err = sysctl_check_table(current->nsproxy, root_table);
1761 	}
1762 #endif
1763 	return 0;
1764 }
1765 
1766 core_initcall(sysctl_init);
1767 
1768 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1769 				      struct ctl_table *table)
1770 {
1771 	struct ctl_table *p;
1772 	const char *s = branch->procname;
1773 
1774 	/* branch should have named subdirectory as its first element */
1775 	if (!s || !branch->child)
1776 		return NULL;
1777 
1778 	/* ... and nothing else */
1779 	if (branch[1].procname || branch[1].ctl_name)
1780 		return NULL;
1781 
1782 	/* table should contain subdirectory with the same name */
1783 	for (p = table; p->procname || p->ctl_name; p++) {
1784 		if (!p->child)
1785 			continue;
1786 		if (p->procname && strcmp(p->procname, s) == 0)
1787 			return p;
1788 	}
1789 	return NULL;
1790 }
1791 
1792 /* see if attaching q to p would be an improvement */
1793 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1794 {
1795 	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1796 	struct ctl_table *next;
1797 	int is_better = 0;
1798 	int not_in_parent = !p->attached_by;
1799 
1800 	while ((next = is_branch_in(by, to)) != NULL) {
1801 		if (by == q->attached_by)
1802 			is_better = 1;
1803 		if (to == p->attached_by)
1804 			not_in_parent = 1;
1805 		by = by->child;
1806 		to = next->child;
1807 	}
1808 
1809 	if (is_better && not_in_parent) {
1810 		q->attached_by = by;
1811 		q->attached_to = to;
1812 		q->parent = p;
1813 	}
1814 }
1815 
1816 /**
1817  * __register_sysctl_paths - register a sysctl hierarchy
1818  * @root: List of sysctl headers to register on
1819  * @namespaces: Data to compute which lists of sysctl entries are visible
1820  * @path: The path to the directory the sysctl table is in.
1821  * @table: the top-level table structure
1822  *
1823  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1824  * array. A completely 0 filled entry terminates the table.
1825  *
1826  * The members of the &struct ctl_table structure are used as follows:
1827  *
1828  * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1829  *            must be unique within that level of sysctl
1830  *
1831  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1832  *            enter a sysctl file
1833  *
1834  * data - a pointer to data for use by proc_handler
1835  *
1836  * maxlen - the maximum size in bytes of the data
1837  *
1838  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1839  *
1840  * child - a pointer to the child sysctl table if this entry is a directory, or
1841  *         %NULL.
1842  *
1843  * proc_handler - the text handler routine (described below)
1844  *
1845  * strategy - the strategy routine (described below)
1846  *
1847  * de - for internal use by the sysctl routines
1848  *
1849  * extra1, extra2 - extra pointers usable by the proc handler routines
1850  *
1851  * Leaf nodes in the sysctl tree will be represented by a single file
1852  * under /proc; non-leaf nodes will be represented by directories.
1853  *
1854  * sysctl(2) can automatically manage read and write requests through
1855  * the sysctl table.  The data and maxlen fields of the ctl_table
1856  * struct enable minimal validation of the values being written to be
1857  * performed, and the mode field allows minimal authentication.
1858  *
1859  * More sophisticated management can be enabled by the provision of a
1860  * strategy routine with the table entry.  This will be called before
1861  * any automatic read or write of the data is performed.
1862  *
1863  * The strategy routine may return
1864  *
1865  * < 0 - Error occurred (error is passed to user process)
1866  *
1867  * 0   - OK - proceed with automatic read or write.
1868  *
1869  * > 0 - OK - read or write has been done by the strategy routine, so
1870  *       return immediately.
1871  *
1872  * There must be a proc_handler routine for any terminal nodes
1873  * mirrored under /proc/sys (non-terminals are handled by a built-in
1874  * directory handler).  Several default handlers are available to
1875  * cover common cases -
1876  *
1877  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1878  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1879  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1880  *
1881  * It is the handler's job to read the input buffer from user memory
1882  * and process it. The handler should return 0 on success.
1883  *
1884  * This routine returns %NULL on a failure to register, and a pointer
1885  * to the table header on success.
1886  */
1887 struct ctl_table_header *__register_sysctl_paths(
1888 	struct ctl_table_root *root,
1889 	struct nsproxy *namespaces,
1890 	const struct ctl_path *path, struct ctl_table *table)
1891 {
1892 	struct ctl_table_header *header;
1893 	struct ctl_table *new, **prevp;
1894 	unsigned int n, npath;
1895 	struct ctl_table_set *set;
1896 
1897 	/* Count the path components */
1898 	for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1899 		;
1900 
1901 	/*
1902 	 * For each path component, allocate a 2-element ctl_table array.
1903 	 * The first array element will be filled with the sysctl entry
1904 	 * for this, the second will be the sentinel (ctl_name == 0).
1905 	 *
1906 	 * We allocate everything in one go so that we don't have to
1907 	 * worry about freeing additional memory in unregister_sysctl_table.
1908 	 */
1909 	header = kzalloc(sizeof(struct ctl_table_header) +
1910 			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1911 	if (!header)
1912 		return NULL;
1913 
1914 	new = (struct ctl_table *) (header + 1);
1915 
1916 	/* Now connect the dots */
1917 	prevp = &header->ctl_table;
1918 	for (n = 0; n < npath; ++n, ++path) {
1919 		/* Copy the procname */
1920 		new->procname = path->procname;
1921 		new->ctl_name = path->ctl_name;
1922 		new->mode     = 0555;
1923 
1924 		*prevp = new;
1925 		prevp = &new->child;
1926 
1927 		new += 2;
1928 	}
1929 	*prevp = table;
1930 	header->ctl_table_arg = table;
1931 
1932 	INIT_LIST_HEAD(&header->ctl_entry);
1933 	header->used = 0;
1934 	header->unregistering = NULL;
1935 	header->root = root;
1936 	sysctl_set_parent(NULL, header->ctl_table);
1937 	header->count = 1;
1938 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1939 	if (sysctl_check_table(namespaces, header->ctl_table)) {
1940 		kfree(header);
1941 		return NULL;
1942 	}
1943 #endif
1944 	spin_lock(&sysctl_lock);
1945 	header->set = lookup_header_set(root, namespaces);
1946 	header->attached_by = header->ctl_table;
1947 	header->attached_to = root_table;
1948 	header->parent = &root_table_header;
1949 	for (set = header->set; set; set = set->parent) {
1950 		struct ctl_table_header *p;
1951 		list_for_each_entry(p, &set->list, ctl_entry) {
1952 			if (p->unregistering)
1953 				continue;
1954 			try_attach(p, header);
1955 		}
1956 	}
1957 	header->parent->count++;
1958 	list_add_tail(&header->ctl_entry, &header->set->list);
1959 	spin_unlock(&sysctl_lock);
1960 
1961 	return header;
1962 }
1963 
1964 /**
1965  * register_sysctl_table_path - register a sysctl table hierarchy
1966  * @path: The path to the directory the sysctl table is in.
1967  * @table: the top-level table structure
1968  *
1969  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1970  * array. A completely 0 filled entry terminates the table.
1971  *
1972  * See __register_sysctl_paths for more details.
1973  */
1974 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1975 						struct ctl_table *table)
1976 {
1977 	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1978 					path, table);
1979 }
1980 
1981 /**
1982  * register_sysctl_table - register a sysctl table hierarchy
1983  * @table: the top-level table structure
1984  *
1985  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1986  * array. A completely 0 filled entry terminates the table.
1987  *
1988  * See register_sysctl_paths for more details.
1989  */
1990 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1991 {
1992 	static const struct ctl_path null_path[] = { {} };
1993 
1994 	return register_sysctl_paths(null_path, table);
1995 }
1996 
1997 /**
1998  * unregister_sysctl_table - unregister a sysctl table hierarchy
1999  * @header: the header returned from register_sysctl_table
2000  *
2001  * Unregisters the sysctl table and all children. proc entries may not
2002  * actually be removed until they are no longer used by anyone.
2003  */
2004 void unregister_sysctl_table(struct ctl_table_header * header)
2005 {
2006 	might_sleep();
2007 
2008 	if (header == NULL)
2009 		return;
2010 
2011 	spin_lock(&sysctl_lock);
2012 	start_unregistering(header);
2013 	if (!--header->parent->count) {
2014 		WARN_ON(1);
2015 		kfree(header->parent);
2016 	}
2017 	if (!--header->count)
2018 		kfree(header);
2019 	spin_unlock(&sysctl_lock);
2020 }
2021 
2022 int sysctl_is_seen(struct ctl_table_header *p)
2023 {
2024 	struct ctl_table_set *set = p->set;
2025 	int res;
2026 	spin_lock(&sysctl_lock);
2027 	if (p->unregistering)
2028 		res = 0;
2029 	else if (!set->is_seen)
2030 		res = 1;
2031 	else
2032 		res = set->is_seen(set);
2033 	spin_unlock(&sysctl_lock);
2034 	return res;
2035 }
2036 
2037 void setup_sysctl_set(struct ctl_table_set *p,
2038 	struct ctl_table_set *parent,
2039 	int (*is_seen)(struct ctl_table_set *))
2040 {
2041 	INIT_LIST_HEAD(&p->list);
2042 	p->parent = parent ? parent : &sysctl_table_root.default_set;
2043 	p->is_seen = is_seen;
2044 }
2045 
2046 #else /* !CONFIG_SYSCTL */
2047 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2048 {
2049 	return NULL;
2050 }
2051 
2052 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2053 						    struct ctl_table *table)
2054 {
2055 	return NULL;
2056 }
2057 
2058 void unregister_sysctl_table(struct ctl_table_header * table)
2059 {
2060 }
2061 
2062 void setup_sysctl_set(struct ctl_table_set *p,
2063 	struct ctl_table_set *parent,
2064 	int (*is_seen)(struct ctl_table_set *))
2065 {
2066 }
2067 
2068 void sysctl_head_put(struct ctl_table_header *head)
2069 {
2070 }
2071 
2072 #endif /* CONFIG_SYSCTL */
2073 
2074 /*
2075  * /proc/sys support
2076  */
2077 
2078 #ifdef CONFIG_PROC_SYSCTL
2079 
2080 static int _proc_do_string(void* data, int maxlen, int write,
2081 			   struct file *filp, void __user *buffer,
2082 			   size_t *lenp, loff_t *ppos)
2083 {
2084 	size_t len;
2085 	char __user *p;
2086 	char c;
2087 
2088 	if (!data || !maxlen || !*lenp) {
2089 		*lenp = 0;
2090 		return 0;
2091 	}
2092 
2093 	if (write) {
2094 		len = 0;
2095 		p = buffer;
2096 		while (len < *lenp) {
2097 			if (get_user(c, p++))
2098 				return -EFAULT;
2099 			if (c == 0 || c == '\n')
2100 				break;
2101 			len++;
2102 		}
2103 		if (len >= maxlen)
2104 			len = maxlen-1;
2105 		if(copy_from_user(data, buffer, len))
2106 			return -EFAULT;
2107 		((char *) data)[len] = 0;
2108 		*ppos += *lenp;
2109 	} else {
2110 		len = strlen(data);
2111 		if (len > maxlen)
2112 			len = maxlen;
2113 
2114 		if (*ppos > len) {
2115 			*lenp = 0;
2116 			return 0;
2117 		}
2118 
2119 		data += *ppos;
2120 		len  -= *ppos;
2121 
2122 		if (len > *lenp)
2123 			len = *lenp;
2124 		if (len)
2125 			if(copy_to_user(buffer, data, len))
2126 				return -EFAULT;
2127 		if (len < *lenp) {
2128 			if(put_user('\n', ((char __user *) buffer) + len))
2129 				return -EFAULT;
2130 			len++;
2131 		}
2132 		*lenp = len;
2133 		*ppos += len;
2134 	}
2135 	return 0;
2136 }
2137 
2138 /**
2139  * proc_dostring - read a string sysctl
2140  * @table: the sysctl table
2141  * @write: %TRUE if this is a write to the sysctl file
2142  * @filp: the file structure
2143  * @buffer: the user buffer
2144  * @lenp: the size of the user buffer
2145  * @ppos: file position
2146  *
2147  * Reads/writes a string from/to the user buffer. If the kernel
2148  * buffer provided is not large enough to hold the string, the
2149  * string is truncated. The copied string is %NULL-terminated.
2150  * If the string is being read by the user process, it is copied
2151  * and a newline '\n' is added. It is truncated if the buffer is
2152  * not large enough.
2153  *
2154  * Returns 0 on success.
2155  */
2156 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2157 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2158 {
2159 	return _proc_do_string(table->data, table->maxlen, write, filp,
2160 			       buffer, lenp, ppos);
2161 }
2162 
2163 
2164 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2165 				 int *valp,
2166 				 int write, void *data)
2167 {
2168 	if (write) {
2169 		*valp = *negp ? -*lvalp : *lvalp;
2170 	} else {
2171 		int val = *valp;
2172 		if (val < 0) {
2173 			*negp = -1;
2174 			*lvalp = (unsigned long)-val;
2175 		} else {
2176 			*negp = 0;
2177 			*lvalp = (unsigned long)val;
2178 		}
2179 	}
2180 	return 0;
2181 }
2182 
2183 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2184 		  int write, struct file *filp, void __user *buffer,
2185 		  size_t *lenp, loff_t *ppos,
2186 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2187 			      int write, void *data),
2188 		  void *data)
2189 {
2190 #define TMPBUFLEN 21
2191 	int *i, vleft, first=1, neg, val;
2192 	unsigned long lval;
2193 	size_t left, len;
2194 
2195 	char buf[TMPBUFLEN], *p;
2196 	char __user *s = buffer;
2197 
2198 	if (!tbl_data || !table->maxlen || !*lenp ||
2199 	    (*ppos && !write)) {
2200 		*lenp = 0;
2201 		return 0;
2202 	}
2203 
2204 	i = (int *) tbl_data;
2205 	vleft = table->maxlen / sizeof(*i);
2206 	left = *lenp;
2207 
2208 	if (!conv)
2209 		conv = do_proc_dointvec_conv;
2210 
2211 	for (; left && vleft--; i++, first=0) {
2212 		if (write) {
2213 			while (left) {
2214 				char c;
2215 				if (get_user(c, s))
2216 					return -EFAULT;
2217 				if (!isspace(c))
2218 					break;
2219 				left--;
2220 				s++;
2221 			}
2222 			if (!left)
2223 				break;
2224 			neg = 0;
2225 			len = left;
2226 			if (len > sizeof(buf) - 1)
2227 				len = sizeof(buf) - 1;
2228 			if (copy_from_user(buf, s, len))
2229 				return -EFAULT;
2230 			buf[len] = 0;
2231 			p = buf;
2232 			if (*p == '-' && left > 1) {
2233 				neg = 1;
2234 				p++;
2235 			}
2236 			if (*p < '0' || *p > '9')
2237 				break;
2238 
2239 			lval = simple_strtoul(p, &p, 0);
2240 
2241 			len = p-buf;
2242 			if ((len < left) && *p && !isspace(*p))
2243 				break;
2244 			if (neg)
2245 				val = -val;
2246 			s += len;
2247 			left -= len;
2248 
2249 			if (conv(&neg, &lval, i, 1, data))
2250 				break;
2251 		} else {
2252 			p = buf;
2253 			if (!first)
2254 				*p++ = '\t';
2255 
2256 			if (conv(&neg, &lval, i, 0, data))
2257 				break;
2258 
2259 			sprintf(p, "%s%lu", neg ? "-" : "", lval);
2260 			len = strlen(buf);
2261 			if (len > left)
2262 				len = left;
2263 			if(copy_to_user(s, buf, len))
2264 				return -EFAULT;
2265 			left -= len;
2266 			s += len;
2267 		}
2268 	}
2269 
2270 	if (!write && !first && left) {
2271 		if(put_user('\n', s))
2272 			return -EFAULT;
2273 		left--, s++;
2274 	}
2275 	if (write) {
2276 		while (left) {
2277 			char c;
2278 			if (get_user(c, s++))
2279 				return -EFAULT;
2280 			if (!isspace(c))
2281 				break;
2282 			left--;
2283 		}
2284 	}
2285 	if (write && first)
2286 		return -EINVAL;
2287 	*lenp -= left;
2288 	*ppos += *lenp;
2289 	return 0;
2290 #undef TMPBUFLEN
2291 }
2292 
2293 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2294 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2295 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2296 			      int write, void *data),
2297 		  void *data)
2298 {
2299 	return __do_proc_dointvec(table->data, table, write, filp,
2300 			buffer, lenp, ppos, conv, data);
2301 }
2302 
2303 /**
2304  * proc_dointvec - read a vector of integers
2305  * @table: the sysctl table
2306  * @write: %TRUE if this is a write to the sysctl file
2307  * @filp: the file structure
2308  * @buffer: the user buffer
2309  * @lenp: the size of the user buffer
2310  * @ppos: file position
2311  *
2312  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2313  * values from/to the user buffer, treated as an ASCII string.
2314  *
2315  * Returns 0 on success.
2316  */
2317 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2318 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2319 {
2320     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2321 		    	    NULL,NULL);
2322 }
2323 
2324 /*
2325  * Taint values can only be increased
2326  * This means we can safely use a temporary.
2327  */
2328 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
2329 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2330 {
2331 	struct ctl_table t;
2332 	unsigned long tmptaint = get_taint();
2333 	int err;
2334 
2335 	if (write && !capable(CAP_SYS_ADMIN))
2336 		return -EPERM;
2337 
2338 	t = *table;
2339 	t.data = &tmptaint;
2340 	err = proc_doulongvec_minmax(&t, write, filp, buffer, lenp, ppos);
2341 	if (err < 0)
2342 		return err;
2343 
2344 	if (write) {
2345 		/*
2346 		 * Poor man's atomic or. Not worth adding a primitive
2347 		 * to everyone's atomic.h for this
2348 		 */
2349 		int i;
2350 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2351 			if ((tmptaint >> i) & 1)
2352 				add_taint(i);
2353 		}
2354 	}
2355 
2356 	return err;
2357 }
2358 
2359 struct do_proc_dointvec_minmax_conv_param {
2360 	int *min;
2361 	int *max;
2362 };
2363 
2364 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2365 					int *valp,
2366 					int write, void *data)
2367 {
2368 	struct do_proc_dointvec_minmax_conv_param *param = data;
2369 	if (write) {
2370 		int val = *negp ? -*lvalp : *lvalp;
2371 		if ((param->min && *param->min > val) ||
2372 		    (param->max && *param->max < val))
2373 			return -EINVAL;
2374 		*valp = val;
2375 	} else {
2376 		int val = *valp;
2377 		if (val < 0) {
2378 			*negp = -1;
2379 			*lvalp = (unsigned long)-val;
2380 		} else {
2381 			*negp = 0;
2382 			*lvalp = (unsigned long)val;
2383 		}
2384 	}
2385 	return 0;
2386 }
2387 
2388 /**
2389  * proc_dointvec_minmax - read a vector of integers with min/max values
2390  * @table: the sysctl table
2391  * @write: %TRUE if this is a write to the sysctl file
2392  * @filp: the file structure
2393  * @buffer: the user buffer
2394  * @lenp: the size of the user buffer
2395  * @ppos: file position
2396  *
2397  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2398  * values from/to the user buffer, treated as an ASCII string.
2399  *
2400  * This routine will ensure the values are within the range specified by
2401  * table->extra1 (min) and table->extra2 (max).
2402  *
2403  * Returns 0 on success.
2404  */
2405 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2406 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2407 {
2408 	struct do_proc_dointvec_minmax_conv_param param = {
2409 		.min = (int *) table->extra1,
2410 		.max = (int *) table->extra2,
2411 	};
2412 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2413 				do_proc_dointvec_minmax_conv, &param);
2414 }
2415 
2416 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2417 				     struct file *filp,
2418 				     void __user *buffer,
2419 				     size_t *lenp, loff_t *ppos,
2420 				     unsigned long convmul,
2421 				     unsigned long convdiv)
2422 {
2423 #define TMPBUFLEN 21
2424 	unsigned long *i, *min, *max, val;
2425 	int vleft, first=1, neg;
2426 	size_t len, left;
2427 	char buf[TMPBUFLEN], *p;
2428 	char __user *s = buffer;
2429 
2430 	if (!data || !table->maxlen || !*lenp ||
2431 	    (*ppos && !write)) {
2432 		*lenp = 0;
2433 		return 0;
2434 	}
2435 
2436 	i = (unsigned long *) data;
2437 	min = (unsigned long *) table->extra1;
2438 	max = (unsigned long *) table->extra2;
2439 	vleft = table->maxlen / sizeof(unsigned long);
2440 	left = *lenp;
2441 
2442 	for (; left && vleft--; i++, min++, max++, first=0) {
2443 		if (write) {
2444 			while (left) {
2445 				char c;
2446 				if (get_user(c, s))
2447 					return -EFAULT;
2448 				if (!isspace(c))
2449 					break;
2450 				left--;
2451 				s++;
2452 			}
2453 			if (!left)
2454 				break;
2455 			neg = 0;
2456 			len = left;
2457 			if (len > TMPBUFLEN-1)
2458 				len = TMPBUFLEN-1;
2459 			if (copy_from_user(buf, s, len))
2460 				return -EFAULT;
2461 			buf[len] = 0;
2462 			p = buf;
2463 			if (*p == '-' && left > 1) {
2464 				neg = 1;
2465 				p++;
2466 			}
2467 			if (*p < '0' || *p > '9')
2468 				break;
2469 			val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2470 			len = p-buf;
2471 			if ((len < left) && *p && !isspace(*p))
2472 				break;
2473 			if (neg)
2474 				val = -val;
2475 			s += len;
2476 			left -= len;
2477 
2478 			if(neg)
2479 				continue;
2480 			if ((min && val < *min) || (max && val > *max))
2481 				continue;
2482 			*i = val;
2483 		} else {
2484 			p = buf;
2485 			if (!first)
2486 				*p++ = '\t';
2487 			sprintf(p, "%lu", convdiv * (*i) / convmul);
2488 			len = strlen(buf);
2489 			if (len > left)
2490 				len = left;
2491 			if(copy_to_user(s, buf, len))
2492 				return -EFAULT;
2493 			left -= len;
2494 			s += len;
2495 		}
2496 	}
2497 
2498 	if (!write && !first && left) {
2499 		if(put_user('\n', s))
2500 			return -EFAULT;
2501 		left--, s++;
2502 	}
2503 	if (write) {
2504 		while (left) {
2505 			char c;
2506 			if (get_user(c, s++))
2507 				return -EFAULT;
2508 			if (!isspace(c))
2509 				break;
2510 			left--;
2511 		}
2512 	}
2513 	if (write && first)
2514 		return -EINVAL;
2515 	*lenp -= left;
2516 	*ppos += *lenp;
2517 	return 0;
2518 #undef TMPBUFLEN
2519 }
2520 
2521 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2522 				     struct file *filp,
2523 				     void __user *buffer,
2524 				     size_t *lenp, loff_t *ppos,
2525 				     unsigned long convmul,
2526 				     unsigned long convdiv)
2527 {
2528 	return __do_proc_doulongvec_minmax(table->data, table, write,
2529 			filp, buffer, lenp, ppos, convmul, convdiv);
2530 }
2531 
2532 /**
2533  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2534  * @table: the sysctl table
2535  * @write: %TRUE if this is a write to the sysctl file
2536  * @filp: the file structure
2537  * @buffer: the user buffer
2538  * @lenp: the size of the user buffer
2539  * @ppos: file position
2540  *
2541  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2542  * values from/to the user buffer, treated as an ASCII string.
2543  *
2544  * This routine will ensure the values are within the range specified by
2545  * table->extra1 (min) and table->extra2 (max).
2546  *
2547  * Returns 0 on success.
2548  */
2549 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2550 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2551 {
2552     return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2553 }
2554 
2555 /**
2556  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2557  * @table: the sysctl table
2558  * @write: %TRUE if this is a write to the sysctl file
2559  * @filp: the file structure
2560  * @buffer: the user buffer
2561  * @lenp: the size of the user buffer
2562  * @ppos: file position
2563  *
2564  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2565  * values from/to the user buffer, treated as an ASCII string. The values
2566  * are treated as milliseconds, and converted to jiffies when they are stored.
2567  *
2568  * This routine will ensure the values are within the range specified by
2569  * table->extra1 (min) and table->extra2 (max).
2570  *
2571  * Returns 0 on success.
2572  */
2573 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2574 				      struct file *filp,
2575 				      void __user *buffer,
2576 				      size_t *lenp, loff_t *ppos)
2577 {
2578     return do_proc_doulongvec_minmax(table, write, filp, buffer,
2579 				     lenp, ppos, HZ, 1000l);
2580 }
2581 
2582 
2583 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2584 					 int *valp,
2585 					 int write, void *data)
2586 {
2587 	if (write) {
2588 		if (*lvalp > LONG_MAX / HZ)
2589 			return 1;
2590 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2591 	} else {
2592 		int val = *valp;
2593 		unsigned long lval;
2594 		if (val < 0) {
2595 			*negp = -1;
2596 			lval = (unsigned long)-val;
2597 		} else {
2598 			*negp = 0;
2599 			lval = (unsigned long)val;
2600 		}
2601 		*lvalp = lval / HZ;
2602 	}
2603 	return 0;
2604 }
2605 
2606 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2607 						int *valp,
2608 						int write, void *data)
2609 {
2610 	if (write) {
2611 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2612 			return 1;
2613 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2614 	} else {
2615 		int val = *valp;
2616 		unsigned long lval;
2617 		if (val < 0) {
2618 			*negp = -1;
2619 			lval = (unsigned long)-val;
2620 		} else {
2621 			*negp = 0;
2622 			lval = (unsigned long)val;
2623 		}
2624 		*lvalp = jiffies_to_clock_t(lval);
2625 	}
2626 	return 0;
2627 }
2628 
2629 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2630 					    int *valp,
2631 					    int write, void *data)
2632 {
2633 	if (write) {
2634 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2635 	} else {
2636 		int val = *valp;
2637 		unsigned long lval;
2638 		if (val < 0) {
2639 			*negp = -1;
2640 			lval = (unsigned long)-val;
2641 		} else {
2642 			*negp = 0;
2643 			lval = (unsigned long)val;
2644 		}
2645 		*lvalp = jiffies_to_msecs(lval);
2646 	}
2647 	return 0;
2648 }
2649 
2650 /**
2651  * proc_dointvec_jiffies - read a vector of integers as seconds
2652  * @table: the sysctl table
2653  * @write: %TRUE if this is a write to the sysctl file
2654  * @filp: the file structure
2655  * @buffer: the user buffer
2656  * @lenp: the size of the user buffer
2657  * @ppos: file position
2658  *
2659  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2660  * values from/to the user buffer, treated as an ASCII string.
2661  * The values read are assumed to be in seconds, and are converted into
2662  * jiffies.
2663  *
2664  * Returns 0 on success.
2665  */
2666 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2667 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2668 {
2669     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2670 		    	    do_proc_dointvec_jiffies_conv,NULL);
2671 }
2672 
2673 /**
2674  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2675  * @table: the sysctl table
2676  * @write: %TRUE if this is a write to the sysctl file
2677  * @filp: the file structure
2678  * @buffer: the user buffer
2679  * @lenp: the size of the user buffer
2680  * @ppos: pointer to the file position
2681  *
2682  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2683  * values from/to the user buffer, treated as an ASCII string.
2684  * The values read are assumed to be in 1/USER_HZ seconds, and
2685  * are converted into jiffies.
2686  *
2687  * Returns 0 on success.
2688  */
2689 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2690 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2691 {
2692     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2693 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2694 }
2695 
2696 /**
2697  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2698  * @table: the sysctl table
2699  * @write: %TRUE if this is a write to the sysctl file
2700  * @filp: the file structure
2701  * @buffer: the user buffer
2702  * @lenp: the size of the user buffer
2703  * @ppos: file position
2704  * @ppos: the current position in the file
2705  *
2706  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2707  * values from/to the user buffer, treated as an ASCII string.
2708  * The values read are assumed to be in 1/1000 seconds, and
2709  * are converted into jiffies.
2710  *
2711  * Returns 0 on success.
2712  */
2713 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2714 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2715 {
2716 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2717 				do_proc_dointvec_ms_jiffies_conv, NULL);
2718 }
2719 
2720 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2721 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2722 {
2723 	struct pid *new_pid;
2724 	pid_t tmp;
2725 	int r;
2726 
2727 	tmp = pid_vnr(cad_pid);
2728 
2729 	r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2730 			       lenp, ppos, NULL, NULL);
2731 	if (r || !write)
2732 		return r;
2733 
2734 	new_pid = find_get_pid(tmp);
2735 	if (!new_pid)
2736 		return -ESRCH;
2737 
2738 	put_pid(xchg(&cad_pid, new_pid));
2739 	return 0;
2740 }
2741 
2742 #else /* CONFIG_PROC_FS */
2743 
2744 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2745 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2746 {
2747 	return -ENOSYS;
2748 }
2749 
2750 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2751 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2752 {
2753 	return -ENOSYS;
2754 }
2755 
2756 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2757 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2758 {
2759 	return -ENOSYS;
2760 }
2761 
2762 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2763 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2764 {
2765 	return -ENOSYS;
2766 }
2767 
2768 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2769 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2770 {
2771 	return -ENOSYS;
2772 }
2773 
2774 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2775 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2776 {
2777 	return -ENOSYS;
2778 }
2779 
2780 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2781 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2782 {
2783 	return -ENOSYS;
2784 }
2785 
2786 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2787 				      struct file *filp,
2788 				      void __user *buffer,
2789 				      size_t *lenp, loff_t *ppos)
2790 {
2791     return -ENOSYS;
2792 }
2793 
2794 
2795 #endif /* CONFIG_PROC_FS */
2796 
2797 
2798 #ifdef CONFIG_SYSCTL_SYSCALL
2799 /*
2800  * General sysctl support routines
2801  */
2802 
2803 /* The generic sysctl data routine (used if no strategy routine supplied) */
2804 int sysctl_data(struct ctl_table *table,
2805 		void __user *oldval, size_t __user *oldlenp,
2806 		void __user *newval, size_t newlen)
2807 {
2808 	size_t len;
2809 
2810 	/* Get out of I don't have a variable */
2811 	if (!table->data || !table->maxlen)
2812 		return -ENOTDIR;
2813 
2814 	if (oldval && oldlenp) {
2815 		if (get_user(len, oldlenp))
2816 			return -EFAULT;
2817 		if (len) {
2818 			if (len > table->maxlen)
2819 				len = table->maxlen;
2820 			if (copy_to_user(oldval, table->data, len))
2821 				return -EFAULT;
2822 			if (put_user(len, oldlenp))
2823 				return -EFAULT;
2824 		}
2825 	}
2826 
2827 	if (newval && newlen) {
2828 		if (newlen > table->maxlen)
2829 			newlen = table->maxlen;
2830 
2831 		if (copy_from_user(table->data, newval, newlen))
2832 			return -EFAULT;
2833 	}
2834 	return 1;
2835 }
2836 
2837 /* The generic string strategy routine: */
2838 int sysctl_string(struct ctl_table *table,
2839 		  void __user *oldval, size_t __user *oldlenp,
2840 		  void __user *newval, size_t newlen)
2841 {
2842 	if (!table->data || !table->maxlen)
2843 		return -ENOTDIR;
2844 
2845 	if (oldval && oldlenp) {
2846 		size_t bufsize;
2847 		if (get_user(bufsize, oldlenp))
2848 			return -EFAULT;
2849 		if (bufsize) {
2850 			size_t len = strlen(table->data), copied;
2851 
2852 			/* This shouldn't trigger for a well-formed sysctl */
2853 			if (len > table->maxlen)
2854 				len = table->maxlen;
2855 
2856 			/* Copy up to a max of bufsize-1 bytes of the string */
2857 			copied = (len >= bufsize) ? bufsize - 1 : len;
2858 
2859 			if (copy_to_user(oldval, table->data, copied) ||
2860 			    put_user(0, (char __user *)(oldval + copied)))
2861 				return -EFAULT;
2862 			if (put_user(len, oldlenp))
2863 				return -EFAULT;
2864 		}
2865 	}
2866 	if (newval && newlen) {
2867 		size_t len = newlen;
2868 		if (len > table->maxlen)
2869 			len = table->maxlen;
2870 		if(copy_from_user(table->data, newval, len))
2871 			return -EFAULT;
2872 		if (len == table->maxlen)
2873 			len--;
2874 		((char *) table->data)[len] = 0;
2875 	}
2876 	return 1;
2877 }
2878 
2879 /*
2880  * This function makes sure that all of the integers in the vector
2881  * are between the minimum and maximum values given in the arrays
2882  * table->extra1 and table->extra2, respectively.
2883  */
2884 int sysctl_intvec(struct ctl_table *table,
2885 		void __user *oldval, size_t __user *oldlenp,
2886 		void __user *newval, size_t newlen)
2887 {
2888 
2889 	if (newval && newlen) {
2890 		int __user *vec = (int __user *) newval;
2891 		int *min = (int *) table->extra1;
2892 		int *max = (int *) table->extra2;
2893 		size_t length;
2894 		int i;
2895 
2896 		if (newlen % sizeof(int) != 0)
2897 			return -EINVAL;
2898 
2899 		if (!table->extra1 && !table->extra2)
2900 			return 0;
2901 
2902 		if (newlen > table->maxlen)
2903 			newlen = table->maxlen;
2904 		length = newlen / sizeof(int);
2905 
2906 		for (i = 0; i < length; i++) {
2907 			int value;
2908 			if (get_user(value, vec + i))
2909 				return -EFAULT;
2910 			if (min && value < min[i])
2911 				return -EINVAL;
2912 			if (max && value > max[i])
2913 				return -EINVAL;
2914 		}
2915 	}
2916 	return 0;
2917 }
2918 
2919 /* Strategy function to convert jiffies to seconds */
2920 int sysctl_jiffies(struct ctl_table *table,
2921 		void __user *oldval, size_t __user *oldlenp,
2922 		void __user *newval, size_t newlen)
2923 {
2924 	if (oldval && oldlenp) {
2925 		size_t olen;
2926 
2927 		if (get_user(olen, oldlenp))
2928 			return -EFAULT;
2929 		if (olen) {
2930 			int val;
2931 
2932 			if (olen < sizeof(int))
2933 				return -EINVAL;
2934 
2935 			val = *(int *)(table->data) / HZ;
2936 			if (put_user(val, (int __user *)oldval))
2937 				return -EFAULT;
2938 			if (put_user(sizeof(int), oldlenp))
2939 				return -EFAULT;
2940 		}
2941 	}
2942 	if (newval && newlen) {
2943 		int new;
2944 		if (newlen != sizeof(int))
2945 			return -EINVAL;
2946 		if (get_user(new, (int __user *)newval))
2947 			return -EFAULT;
2948 		*(int *)(table->data) = new*HZ;
2949 	}
2950 	return 1;
2951 }
2952 
2953 /* Strategy function to convert jiffies to seconds */
2954 int sysctl_ms_jiffies(struct ctl_table *table,
2955 		void __user *oldval, size_t __user *oldlenp,
2956 		void __user *newval, size_t newlen)
2957 {
2958 	if (oldval && oldlenp) {
2959 		size_t olen;
2960 
2961 		if (get_user(olen, oldlenp))
2962 			return -EFAULT;
2963 		if (olen) {
2964 			int val;
2965 
2966 			if (olen < sizeof(int))
2967 				return -EINVAL;
2968 
2969 			val = jiffies_to_msecs(*(int *)(table->data));
2970 			if (put_user(val, (int __user *)oldval))
2971 				return -EFAULT;
2972 			if (put_user(sizeof(int), oldlenp))
2973 				return -EFAULT;
2974 		}
2975 	}
2976 	if (newval && newlen) {
2977 		int new;
2978 		if (newlen != sizeof(int))
2979 			return -EINVAL;
2980 		if (get_user(new, (int __user *)newval))
2981 			return -EFAULT;
2982 		*(int *)(table->data) = msecs_to_jiffies(new);
2983 	}
2984 	return 1;
2985 }
2986 
2987 
2988 
2989 #else /* CONFIG_SYSCTL_SYSCALL */
2990 
2991 
2992 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2993 {
2994 	struct __sysctl_args tmp;
2995 	int error;
2996 
2997 	if (copy_from_user(&tmp, args, sizeof(tmp)))
2998 		return -EFAULT;
2999 
3000 	error = deprecated_sysctl_warning(&tmp);
3001 
3002 	/* If no error reading the parameters then just -ENOSYS ... */
3003 	if (!error)
3004 		error = -ENOSYS;
3005 
3006 	return error;
3007 }
3008 
3009 int sysctl_data(struct ctl_table *table,
3010 		  void __user *oldval, size_t __user *oldlenp,
3011 		  void __user *newval, size_t newlen)
3012 {
3013 	return -ENOSYS;
3014 }
3015 
3016 int sysctl_string(struct ctl_table *table,
3017 		  void __user *oldval, size_t __user *oldlenp,
3018 		  void __user *newval, size_t newlen)
3019 {
3020 	return -ENOSYS;
3021 }
3022 
3023 int sysctl_intvec(struct ctl_table *table,
3024 		void __user *oldval, size_t __user *oldlenp,
3025 		void __user *newval, size_t newlen)
3026 {
3027 	return -ENOSYS;
3028 }
3029 
3030 int sysctl_jiffies(struct ctl_table *table,
3031 		void __user *oldval, size_t __user *oldlenp,
3032 		void __user *newval, size_t newlen)
3033 {
3034 	return -ENOSYS;
3035 }
3036 
3037 int sysctl_ms_jiffies(struct ctl_table *table,
3038 		void __user *oldval, size_t __user *oldlenp,
3039 		void __user *newval, size_t newlen)
3040 {
3041 	return -ENOSYS;
3042 }
3043 
3044 #endif /* CONFIG_SYSCTL_SYSCALL */
3045 
3046 static int deprecated_sysctl_warning(struct __sysctl_args *args)
3047 {
3048 	static int msg_count;
3049 	int name[CTL_MAXNAME];
3050 	int i;
3051 
3052 	/* Check args->nlen. */
3053 	if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
3054 		return -ENOTDIR;
3055 
3056 	/* Read in the sysctl name for better debug message logging */
3057 	for (i = 0; i < args->nlen; i++)
3058 		if (get_user(name[i], args->name + i))
3059 			return -EFAULT;
3060 
3061 	/* Ignore accesses to kernel.version */
3062 	if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
3063 		return 0;
3064 
3065 	if (msg_count < 5) {
3066 		msg_count++;
3067 		printk(KERN_INFO
3068 			"warning: process `%s' used the deprecated sysctl "
3069 			"system call with ", current->comm);
3070 		for (i = 0; i < args->nlen; i++)
3071 			printk("%d.", name[i]);
3072 		printk("\n");
3073 	}
3074 	return 0;
3075 }
3076 
3077 /*
3078  * No sense putting this after each symbol definition, twice,
3079  * exception granted :-)
3080  */
3081 EXPORT_SYMBOL(proc_dointvec);
3082 EXPORT_SYMBOL(proc_dointvec_jiffies);
3083 EXPORT_SYMBOL(proc_dointvec_minmax);
3084 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3085 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3086 EXPORT_SYMBOL(proc_dostring);
3087 EXPORT_SYMBOL(proc_doulongvec_minmax);
3088 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3089 EXPORT_SYMBOL(register_sysctl_table);
3090 EXPORT_SYMBOL(register_sysctl_paths);
3091 EXPORT_SYMBOL(sysctl_intvec);
3092 EXPORT_SYMBOL(sysctl_jiffies);
3093 EXPORT_SYMBOL(sysctl_ms_jiffies);
3094 EXPORT_SYMBOL(sysctl_string);
3095 EXPORT_SYMBOL(sysctl_data);
3096 EXPORT_SYMBOL(unregister_sysctl_table);
3097