xref: /openbmc/linux/kernel/sysctl.c (revision b04b4f78)
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 #include <linux/slow-work.h>
52 
53 #include <asm/uaccess.h>
54 #include <asm/processor.h>
55 
56 #ifdef CONFIG_X86
57 #include <asm/nmi.h>
58 #include <asm/stacktrace.h>
59 #include <asm/io.h>
60 #endif
61 
62 static int deprecated_sysctl_warning(struct __sysctl_args *args);
63 
64 #if defined(CONFIG_SYSCTL)
65 
66 /* External variables not in a header file. */
67 extern int C_A_D;
68 extern int print_fatal_signals;
69 extern int sysctl_overcommit_memory;
70 extern int sysctl_overcommit_ratio;
71 extern int sysctl_panic_on_oom;
72 extern int sysctl_oom_kill_allocating_task;
73 extern int sysctl_oom_dump_tasks;
74 extern int max_threads;
75 extern int core_uses_pid;
76 extern int suid_dumpable;
77 extern char core_pattern[];
78 extern int pid_max;
79 extern int min_free_kbytes;
80 extern int pid_max_min, pid_max_max;
81 extern int sysctl_drop_caches;
82 extern int percpu_pagelist_fraction;
83 extern int compat_log;
84 extern int latencytop_enabled;
85 extern int sysctl_nr_open_min, sysctl_nr_open_max;
86 #ifndef CONFIG_MMU
87 extern int sysctl_nr_trim_pages;
88 #endif
89 #ifdef CONFIG_RCU_TORTURE_TEST
90 extern int rcutorture_runnable;
91 #endif /* #ifdef CONFIG_RCU_TORTURE_TEST */
92 
93 /* Constants used for minimum and  maximum */
94 #ifdef CONFIG_DETECT_SOFTLOCKUP
95 static int sixty = 60;
96 static int neg_one = -1;
97 #endif
98 
99 static int zero;
100 static int __maybe_unused one = 1;
101 static int __maybe_unused two = 2;
102 static unsigned long one_ul = 1;
103 static int one_hundred = 100;
104 static int one_thousand = 1000;
105 
106 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
107 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
108 
109 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
110 static int maxolduid = 65535;
111 static int minolduid;
112 static int min_percpu_pagelist_fract = 8;
113 
114 static int ngroups_max = NGROUPS_MAX;
115 
116 #ifdef CONFIG_MODULES
117 extern char modprobe_path[];
118 #endif
119 #ifdef CONFIG_CHR_DEV_SG
120 extern int sg_big_buff;
121 #endif
122 
123 #ifdef CONFIG_SPARC
124 #include <asm/system.h>
125 #endif
126 
127 #ifdef CONFIG_SPARC64
128 extern int sysctl_tsb_ratio;
129 #endif
130 
131 #ifdef __hppa__
132 extern int pwrsw_enabled;
133 extern int unaligned_enabled;
134 #endif
135 
136 #ifdef CONFIG_S390
137 #ifdef CONFIG_MATHEMU
138 extern int sysctl_ieee_emulation_warnings;
139 #endif
140 extern int sysctl_userprocess_debug;
141 extern int spin_retry;
142 #endif
143 
144 #ifdef CONFIG_BSD_PROCESS_ACCT
145 extern int acct_parm[];
146 #endif
147 
148 #ifdef CONFIG_IA64
149 extern int no_unaligned_warning;
150 extern int unaligned_dump_stack;
151 #endif
152 
153 #ifdef CONFIG_RT_MUTEXES
154 extern int max_lock_depth;
155 #endif
156 
157 #ifdef CONFIG_PROC_SYSCTL
158 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
159 		  void __user *buffer, size_t *lenp, loff_t *ppos);
160 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
161 			       void __user *buffer, size_t *lenp, loff_t *ppos);
162 #endif
163 
164 static struct ctl_table root_table[];
165 static struct ctl_table_root sysctl_table_root;
166 static struct ctl_table_header root_table_header = {
167 	.count = 1,
168 	.ctl_table = root_table,
169 	.ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
170 	.root = &sysctl_table_root,
171 	.set = &sysctl_table_root.default_set,
172 };
173 static struct ctl_table_root sysctl_table_root = {
174 	.root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
175 	.default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
176 };
177 
178 static struct ctl_table kern_table[];
179 static struct ctl_table vm_table[];
180 static struct ctl_table fs_table[];
181 static struct ctl_table debug_table[];
182 static struct ctl_table dev_table[];
183 extern struct ctl_table random_table[];
184 #ifdef CONFIG_INOTIFY_USER
185 extern struct ctl_table inotify_table[];
186 #endif
187 #ifdef CONFIG_EPOLL
188 extern struct ctl_table epoll_table[];
189 #endif
190 
191 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
192 int sysctl_legacy_va_layout;
193 #endif
194 
195 extern int prove_locking;
196 extern int lock_stat;
197 
198 /* The default sysctl tables: */
199 
200 static struct ctl_table root_table[] = {
201 	{
202 		.ctl_name	= CTL_KERN,
203 		.procname	= "kernel",
204 		.mode		= 0555,
205 		.child		= kern_table,
206 	},
207 	{
208 		.ctl_name	= CTL_VM,
209 		.procname	= "vm",
210 		.mode		= 0555,
211 		.child		= vm_table,
212 	},
213 	{
214 		.ctl_name	= CTL_FS,
215 		.procname	= "fs",
216 		.mode		= 0555,
217 		.child		= fs_table,
218 	},
219 	{
220 		.ctl_name	= CTL_DEBUG,
221 		.procname	= "debug",
222 		.mode		= 0555,
223 		.child		= debug_table,
224 	},
225 	{
226 		.ctl_name	= CTL_DEV,
227 		.procname	= "dev",
228 		.mode		= 0555,
229 		.child		= dev_table,
230 	},
231 /*
232  * NOTE: do not add new entries to this table unless you have read
233  * Documentation/sysctl/ctl_unnumbered.txt
234  */
235 	{ .ctl_name = 0 }
236 };
237 
238 #ifdef CONFIG_SCHED_DEBUG
239 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
240 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
241 static int min_wakeup_granularity_ns;			/* 0 usecs */
242 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
243 #endif
244 
245 static struct ctl_table kern_table[] = {
246 #ifdef CONFIG_SCHED_DEBUG
247 	{
248 		.ctl_name	= CTL_UNNUMBERED,
249 		.procname	= "sched_min_granularity_ns",
250 		.data		= &sysctl_sched_min_granularity,
251 		.maxlen		= sizeof(unsigned int),
252 		.mode		= 0644,
253 		.proc_handler	= &sched_nr_latency_handler,
254 		.strategy	= &sysctl_intvec,
255 		.extra1		= &min_sched_granularity_ns,
256 		.extra2		= &max_sched_granularity_ns,
257 	},
258 	{
259 		.ctl_name	= CTL_UNNUMBERED,
260 		.procname	= "sched_latency_ns",
261 		.data		= &sysctl_sched_latency,
262 		.maxlen		= sizeof(unsigned int),
263 		.mode		= 0644,
264 		.proc_handler	= &sched_nr_latency_handler,
265 		.strategy	= &sysctl_intvec,
266 		.extra1		= &min_sched_granularity_ns,
267 		.extra2		= &max_sched_granularity_ns,
268 	},
269 	{
270 		.ctl_name	= CTL_UNNUMBERED,
271 		.procname	= "sched_wakeup_granularity_ns",
272 		.data		= &sysctl_sched_wakeup_granularity,
273 		.maxlen		= sizeof(unsigned int),
274 		.mode		= 0644,
275 		.proc_handler	= &proc_dointvec_minmax,
276 		.strategy	= &sysctl_intvec,
277 		.extra1		= &min_wakeup_granularity_ns,
278 		.extra2		= &max_wakeup_granularity_ns,
279 	},
280 	{
281 		.ctl_name	= CTL_UNNUMBERED,
282 		.procname	= "sched_shares_ratelimit",
283 		.data		= &sysctl_sched_shares_ratelimit,
284 		.maxlen		= sizeof(unsigned int),
285 		.mode		= 0644,
286 		.proc_handler	= &proc_dointvec,
287 	},
288 	{
289 		.ctl_name	= CTL_UNNUMBERED,
290 		.procname	= "sched_shares_thresh",
291 		.data		= &sysctl_sched_shares_thresh,
292 		.maxlen		= sizeof(unsigned int),
293 		.mode		= 0644,
294 		.proc_handler	= &proc_dointvec_minmax,
295 		.strategy	= &sysctl_intvec,
296 		.extra1		= &zero,
297 	},
298 	{
299 		.ctl_name	= CTL_UNNUMBERED,
300 		.procname	= "sched_child_runs_first",
301 		.data		= &sysctl_sched_child_runs_first,
302 		.maxlen		= sizeof(unsigned int),
303 		.mode		= 0644,
304 		.proc_handler	= &proc_dointvec,
305 	},
306 	{
307 		.ctl_name	= CTL_UNNUMBERED,
308 		.procname	= "sched_features",
309 		.data		= &sysctl_sched_features,
310 		.maxlen		= sizeof(unsigned int),
311 		.mode		= 0644,
312 		.proc_handler	= &proc_dointvec,
313 	},
314 	{
315 		.ctl_name	= CTL_UNNUMBERED,
316 		.procname	= "sched_migration_cost",
317 		.data		= &sysctl_sched_migration_cost,
318 		.maxlen		= sizeof(unsigned int),
319 		.mode		= 0644,
320 		.proc_handler	= &proc_dointvec,
321 	},
322 	{
323 		.ctl_name	= CTL_UNNUMBERED,
324 		.procname	= "sched_nr_migrate",
325 		.data		= &sysctl_sched_nr_migrate,
326 		.maxlen		= sizeof(unsigned int),
327 		.mode		= 0644,
328 		.proc_handler	= &proc_dointvec,
329 	},
330 #endif
331 	{
332 		.ctl_name	= CTL_UNNUMBERED,
333 		.procname	= "sched_rt_period_us",
334 		.data		= &sysctl_sched_rt_period,
335 		.maxlen		= sizeof(unsigned int),
336 		.mode		= 0644,
337 		.proc_handler	= &sched_rt_handler,
338 	},
339 	{
340 		.ctl_name	= CTL_UNNUMBERED,
341 		.procname	= "sched_rt_runtime_us",
342 		.data		= &sysctl_sched_rt_runtime,
343 		.maxlen		= sizeof(int),
344 		.mode		= 0644,
345 		.proc_handler	= &sched_rt_handler,
346 	},
347 	{
348 		.ctl_name	= CTL_UNNUMBERED,
349 		.procname	= "sched_compat_yield",
350 		.data		= &sysctl_sched_compat_yield,
351 		.maxlen		= sizeof(unsigned int),
352 		.mode		= 0644,
353 		.proc_handler	= &proc_dointvec,
354 	},
355 #ifdef CONFIG_PROVE_LOCKING
356 	{
357 		.ctl_name	= CTL_UNNUMBERED,
358 		.procname	= "prove_locking",
359 		.data		= &prove_locking,
360 		.maxlen		= sizeof(int),
361 		.mode		= 0644,
362 		.proc_handler	= &proc_dointvec,
363 	},
364 #endif
365 #ifdef CONFIG_LOCK_STAT
366 	{
367 		.ctl_name	= CTL_UNNUMBERED,
368 		.procname	= "lock_stat",
369 		.data		= &lock_stat,
370 		.maxlen		= sizeof(int),
371 		.mode		= 0644,
372 		.proc_handler	= &proc_dointvec,
373 	},
374 #endif
375 	{
376 		.ctl_name	= KERN_PANIC,
377 		.procname	= "panic",
378 		.data		= &panic_timeout,
379 		.maxlen		= sizeof(int),
380 		.mode		= 0644,
381 		.proc_handler	= &proc_dointvec,
382 	},
383 	{
384 		.ctl_name	= KERN_CORE_USES_PID,
385 		.procname	= "core_uses_pid",
386 		.data		= &core_uses_pid,
387 		.maxlen		= sizeof(int),
388 		.mode		= 0644,
389 		.proc_handler	= &proc_dointvec,
390 	},
391 	{
392 		.ctl_name	= KERN_CORE_PATTERN,
393 		.procname	= "core_pattern",
394 		.data		= core_pattern,
395 		.maxlen		= CORENAME_MAX_SIZE,
396 		.mode		= 0644,
397 		.proc_handler	= &proc_dostring,
398 		.strategy	= &sysctl_string,
399 	},
400 #ifdef CONFIG_PROC_SYSCTL
401 	{
402 		.procname	= "tainted",
403 		.maxlen 	= sizeof(long),
404 		.mode		= 0644,
405 		.proc_handler	= &proc_taint,
406 	},
407 #endif
408 #ifdef CONFIG_LATENCYTOP
409 	{
410 		.procname	= "latencytop",
411 		.data		= &latencytop_enabled,
412 		.maxlen		= sizeof(int),
413 		.mode		= 0644,
414 		.proc_handler	= &proc_dointvec,
415 	},
416 #endif
417 #ifdef CONFIG_BLK_DEV_INITRD
418 	{
419 		.ctl_name	= KERN_REALROOTDEV,
420 		.procname	= "real-root-dev",
421 		.data		= &real_root_dev,
422 		.maxlen		= sizeof(int),
423 		.mode		= 0644,
424 		.proc_handler	= &proc_dointvec,
425 	},
426 #endif
427 	{
428 		.ctl_name	= CTL_UNNUMBERED,
429 		.procname	= "print-fatal-signals",
430 		.data		= &print_fatal_signals,
431 		.maxlen		= sizeof(int),
432 		.mode		= 0644,
433 		.proc_handler	= &proc_dointvec,
434 	},
435 #ifdef CONFIG_SPARC
436 	{
437 		.ctl_name	= KERN_SPARC_REBOOT,
438 		.procname	= "reboot-cmd",
439 		.data		= reboot_command,
440 		.maxlen		= 256,
441 		.mode		= 0644,
442 		.proc_handler	= &proc_dostring,
443 		.strategy	= &sysctl_string,
444 	},
445 	{
446 		.ctl_name	= KERN_SPARC_STOP_A,
447 		.procname	= "stop-a",
448 		.data		= &stop_a_enabled,
449 		.maxlen		= sizeof (int),
450 		.mode		= 0644,
451 		.proc_handler	= &proc_dointvec,
452 	},
453 	{
454 		.ctl_name	= KERN_SPARC_SCONS_PWROFF,
455 		.procname	= "scons-poweroff",
456 		.data		= &scons_pwroff,
457 		.maxlen		= sizeof (int),
458 		.mode		= 0644,
459 		.proc_handler	= &proc_dointvec,
460 	},
461 #endif
462 #ifdef CONFIG_SPARC64
463 	{
464 		.ctl_name	= CTL_UNNUMBERED,
465 		.procname	= "tsb-ratio",
466 		.data		= &sysctl_tsb_ratio,
467 		.maxlen		= sizeof (int),
468 		.mode		= 0644,
469 		.proc_handler	= &proc_dointvec,
470 	},
471 #endif
472 #ifdef __hppa__
473 	{
474 		.ctl_name	= KERN_HPPA_PWRSW,
475 		.procname	= "soft-power",
476 		.data		= &pwrsw_enabled,
477 		.maxlen		= sizeof (int),
478 	 	.mode		= 0644,
479 		.proc_handler	= &proc_dointvec,
480 	},
481 	{
482 		.ctl_name	= KERN_HPPA_UNALIGNED,
483 		.procname	= "unaligned-trap",
484 		.data		= &unaligned_enabled,
485 		.maxlen		= sizeof (int),
486 		.mode		= 0644,
487 		.proc_handler	= &proc_dointvec,
488 	},
489 #endif
490 	{
491 		.ctl_name	= KERN_CTLALTDEL,
492 		.procname	= "ctrl-alt-del",
493 		.data		= &C_A_D,
494 		.maxlen		= sizeof(int),
495 		.mode		= 0644,
496 		.proc_handler	= &proc_dointvec,
497 	},
498 #ifdef CONFIG_FUNCTION_TRACER
499 	{
500 		.ctl_name	= CTL_UNNUMBERED,
501 		.procname	= "ftrace_enabled",
502 		.data		= &ftrace_enabled,
503 		.maxlen		= sizeof(int),
504 		.mode		= 0644,
505 		.proc_handler	= &ftrace_enable_sysctl,
506 	},
507 #endif
508 #ifdef CONFIG_STACK_TRACER
509 	{
510 		.ctl_name	= CTL_UNNUMBERED,
511 		.procname	= "stack_tracer_enabled",
512 		.data		= &stack_tracer_enabled,
513 		.maxlen		= sizeof(int),
514 		.mode		= 0644,
515 		.proc_handler	= &stack_trace_sysctl,
516 	},
517 #endif
518 #ifdef CONFIG_TRACING
519 	{
520 		.ctl_name	= CTL_UNNUMBERED,
521 		.procname	= "ftrace_dump_on_oops",
522 		.data		= &ftrace_dump_on_oops,
523 		.maxlen		= sizeof(int),
524 		.mode		= 0644,
525 		.proc_handler	= &proc_dointvec,
526 	},
527 #endif
528 #ifdef CONFIG_MODULES
529 	{
530 		.ctl_name	= KERN_MODPROBE,
531 		.procname	= "modprobe",
532 		.data		= &modprobe_path,
533 		.maxlen		= KMOD_PATH_LEN,
534 		.mode		= 0644,
535 		.proc_handler	= &proc_dostring,
536 		.strategy	= &sysctl_string,
537 	},
538 #endif
539 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
540 	{
541 		.ctl_name	= KERN_HOTPLUG,
542 		.procname	= "hotplug",
543 		.data		= &uevent_helper,
544 		.maxlen		= UEVENT_HELPER_PATH_LEN,
545 		.mode		= 0644,
546 		.proc_handler	= &proc_dostring,
547 		.strategy	= &sysctl_string,
548 	},
549 #endif
550 #ifdef CONFIG_CHR_DEV_SG
551 	{
552 		.ctl_name	= KERN_SG_BIG_BUFF,
553 		.procname	= "sg-big-buff",
554 		.data		= &sg_big_buff,
555 		.maxlen		= sizeof (int),
556 		.mode		= 0444,
557 		.proc_handler	= &proc_dointvec,
558 	},
559 #endif
560 #ifdef CONFIG_BSD_PROCESS_ACCT
561 	{
562 		.ctl_name	= KERN_ACCT,
563 		.procname	= "acct",
564 		.data		= &acct_parm,
565 		.maxlen		= 3*sizeof(int),
566 		.mode		= 0644,
567 		.proc_handler	= &proc_dointvec,
568 	},
569 #endif
570 #ifdef CONFIG_MAGIC_SYSRQ
571 	{
572 		.ctl_name	= KERN_SYSRQ,
573 		.procname	= "sysrq",
574 		.data		= &__sysrq_enabled,
575 		.maxlen		= sizeof (int),
576 		.mode		= 0644,
577 		.proc_handler	= &proc_dointvec,
578 	},
579 #endif
580 #ifdef CONFIG_PROC_SYSCTL
581 	{
582 		.procname	= "cad_pid",
583 		.data		= NULL,
584 		.maxlen		= sizeof (int),
585 		.mode		= 0600,
586 		.proc_handler	= &proc_do_cad_pid,
587 	},
588 #endif
589 	{
590 		.ctl_name	= KERN_MAX_THREADS,
591 		.procname	= "threads-max",
592 		.data		= &max_threads,
593 		.maxlen		= sizeof(int),
594 		.mode		= 0644,
595 		.proc_handler	= &proc_dointvec,
596 	},
597 	{
598 		.ctl_name	= KERN_RANDOM,
599 		.procname	= "random",
600 		.mode		= 0555,
601 		.child		= random_table,
602 	},
603 	{
604 		.ctl_name	= KERN_OVERFLOWUID,
605 		.procname	= "overflowuid",
606 		.data		= &overflowuid,
607 		.maxlen		= sizeof(int),
608 		.mode		= 0644,
609 		.proc_handler	= &proc_dointvec_minmax,
610 		.strategy	= &sysctl_intvec,
611 		.extra1		= &minolduid,
612 		.extra2		= &maxolduid,
613 	},
614 	{
615 		.ctl_name	= KERN_OVERFLOWGID,
616 		.procname	= "overflowgid",
617 		.data		= &overflowgid,
618 		.maxlen		= sizeof(int),
619 		.mode		= 0644,
620 		.proc_handler	= &proc_dointvec_minmax,
621 		.strategy	= &sysctl_intvec,
622 		.extra1		= &minolduid,
623 		.extra2		= &maxolduid,
624 	},
625 #ifdef CONFIG_S390
626 #ifdef CONFIG_MATHEMU
627 	{
628 		.ctl_name	= KERN_IEEE_EMULATION_WARNINGS,
629 		.procname	= "ieee_emulation_warnings",
630 		.data		= &sysctl_ieee_emulation_warnings,
631 		.maxlen		= sizeof(int),
632 		.mode		= 0644,
633 		.proc_handler	= &proc_dointvec,
634 	},
635 #endif
636 	{
637 		.ctl_name	= KERN_S390_USER_DEBUG_LOGGING,
638 		.procname	= "userprocess_debug",
639 		.data		= &sysctl_userprocess_debug,
640 		.maxlen		= sizeof(int),
641 		.mode		= 0644,
642 		.proc_handler	= &proc_dointvec,
643 	},
644 #endif
645 	{
646 		.ctl_name	= KERN_PIDMAX,
647 		.procname	= "pid_max",
648 		.data		= &pid_max,
649 		.maxlen		= sizeof (int),
650 		.mode		= 0644,
651 		.proc_handler	= &proc_dointvec_minmax,
652 		.strategy	= sysctl_intvec,
653 		.extra1		= &pid_max_min,
654 		.extra2		= &pid_max_max,
655 	},
656 	{
657 		.ctl_name	= KERN_PANIC_ON_OOPS,
658 		.procname	= "panic_on_oops",
659 		.data		= &panic_on_oops,
660 		.maxlen		= sizeof(int),
661 		.mode		= 0644,
662 		.proc_handler	= &proc_dointvec,
663 	},
664 #if defined CONFIG_PRINTK
665 	{
666 		.ctl_name	= KERN_PRINTK,
667 		.procname	= "printk",
668 		.data		= &console_loglevel,
669 		.maxlen		= 4*sizeof(int),
670 		.mode		= 0644,
671 		.proc_handler	= &proc_dointvec,
672 	},
673 	{
674 		.ctl_name	= KERN_PRINTK_RATELIMIT,
675 		.procname	= "printk_ratelimit",
676 		.data		= &printk_ratelimit_state.interval,
677 		.maxlen		= sizeof(int),
678 		.mode		= 0644,
679 		.proc_handler	= &proc_dointvec_jiffies,
680 		.strategy	= &sysctl_jiffies,
681 	},
682 	{
683 		.ctl_name	= KERN_PRINTK_RATELIMIT_BURST,
684 		.procname	= "printk_ratelimit_burst",
685 		.data		= &printk_ratelimit_state.burst,
686 		.maxlen		= sizeof(int),
687 		.mode		= 0644,
688 		.proc_handler	= &proc_dointvec,
689 	},
690 #endif
691 	{
692 		.ctl_name	= KERN_NGROUPS_MAX,
693 		.procname	= "ngroups_max",
694 		.data		= &ngroups_max,
695 		.maxlen		= sizeof (int),
696 		.mode		= 0444,
697 		.proc_handler	= &proc_dointvec,
698 	},
699 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
700 	{
701 		.ctl_name       = KERN_UNKNOWN_NMI_PANIC,
702 		.procname       = "unknown_nmi_panic",
703 		.data           = &unknown_nmi_panic,
704 		.maxlen         = sizeof (int),
705 		.mode           = 0644,
706 		.proc_handler   = &proc_dointvec,
707 	},
708 	{
709 		.procname       = "nmi_watchdog",
710 		.data           = &nmi_watchdog_enabled,
711 		.maxlen         = sizeof (int),
712 		.mode           = 0644,
713 		.proc_handler   = &proc_nmi_enabled,
714 	},
715 #endif
716 #if defined(CONFIG_X86)
717 	{
718 		.ctl_name	= KERN_PANIC_ON_NMI,
719 		.procname	= "panic_on_unrecovered_nmi",
720 		.data		= &panic_on_unrecovered_nmi,
721 		.maxlen		= sizeof(int),
722 		.mode		= 0644,
723 		.proc_handler	= &proc_dointvec,
724 	},
725 	{
726 		.ctl_name	= KERN_BOOTLOADER_TYPE,
727 		.procname	= "bootloader_type",
728 		.data		= &bootloader_type,
729 		.maxlen		= sizeof (int),
730 		.mode		= 0444,
731 		.proc_handler	= &proc_dointvec,
732 	},
733 	{
734 		.ctl_name	= CTL_UNNUMBERED,
735 		.procname	= "kstack_depth_to_print",
736 		.data		= &kstack_depth_to_print,
737 		.maxlen		= sizeof(int),
738 		.mode		= 0644,
739 		.proc_handler	= &proc_dointvec,
740 	},
741 	{
742 		.ctl_name	= CTL_UNNUMBERED,
743 		.procname	= "io_delay_type",
744 		.data		= &io_delay_type,
745 		.maxlen		= sizeof(int),
746 		.mode		= 0644,
747 		.proc_handler	= &proc_dointvec,
748 	},
749 #endif
750 #if defined(CONFIG_MMU)
751 	{
752 		.ctl_name	= KERN_RANDOMIZE,
753 		.procname	= "randomize_va_space",
754 		.data		= &randomize_va_space,
755 		.maxlen		= sizeof(int),
756 		.mode		= 0644,
757 		.proc_handler	= &proc_dointvec,
758 	},
759 #endif
760 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
761 	{
762 		.ctl_name	= KERN_SPIN_RETRY,
763 		.procname	= "spin_retry",
764 		.data		= &spin_retry,
765 		.maxlen		= sizeof (int),
766 		.mode		= 0644,
767 		.proc_handler	= &proc_dointvec,
768 	},
769 #endif
770 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
771 	{
772 		.procname	= "acpi_video_flags",
773 		.data		= &acpi_realmode_flags,
774 		.maxlen		= sizeof (unsigned long),
775 		.mode		= 0644,
776 		.proc_handler	= &proc_doulongvec_minmax,
777 	},
778 #endif
779 #ifdef CONFIG_IA64
780 	{
781 		.ctl_name	= KERN_IA64_UNALIGNED,
782 		.procname	= "ignore-unaligned-usertrap",
783 		.data		= &no_unaligned_warning,
784 		.maxlen		= sizeof (int),
785 	 	.mode		= 0644,
786 		.proc_handler	= &proc_dointvec,
787 	},
788 	{
789 		.ctl_name	= CTL_UNNUMBERED,
790 		.procname	= "unaligned-dump-stack",
791 		.data		= &unaligned_dump_stack,
792 		.maxlen		= sizeof (int),
793 		.mode		= 0644,
794 		.proc_handler	= &proc_dointvec,
795 	},
796 #endif
797 #ifdef CONFIG_DETECT_SOFTLOCKUP
798 	{
799 		.ctl_name	= CTL_UNNUMBERED,
800 		.procname	= "softlockup_panic",
801 		.data		= &softlockup_panic,
802 		.maxlen		= sizeof(int),
803 		.mode		= 0644,
804 		.proc_handler	= &proc_dointvec_minmax,
805 		.strategy	= &sysctl_intvec,
806 		.extra1		= &zero,
807 		.extra2		= &one,
808 	},
809 	{
810 		.ctl_name	= CTL_UNNUMBERED,
811 		.procname	= "softlockup_thresh",
812 		.data		= &softlockup_thresh,
813 		.maxlen		= sizeof(int),
814 		.mode		= 0644,
815 		.proc_handler	= &proc_dosoftlockup_thresh,
816 		.strategy	= &sysctl_intvec,
817 		.extra1		= &neg_one,
818 		.extra2		= &sixty,
819 	},
820 #endif
821 #ifdef CONFIG_DETECT_HUNG_TASK
822 	{
823 		.ctl_name	= CTL_UNNUMBERED,
824 		.procname	= "hung_task_panic",
825 		.data		= &sysctl_hung_task_panic,
826 		.maxlen		= sizeof(int),
827 		.mode		= 0644,
828 		.proc_handler	= &proc_dointvec_minmax,
829 		.strategy	= &sysctl_intvec,
830 		.extra1		= &zero,
831 		.extra2		= &one,
832 	},
833 	{
834 		.ctl_name	= CTL_UNNUMBERED,
835 		.procname	= "hung_task_check_count",
836 		.data		= &sysctl_hung_task_check_count,
837 		.maxlen		= sizeof(unsigned long),
838 		.mode		= 0644,
839 		.proc_handler	= &proc_doulongvec_minmax,
840 		.strategy	= &sysctl_intvec,
841 	},
842 	{
843 		.ctl_name	= CTL_UNNUMBERED,
844 		.procname	= "hung_task_timeout_secs",
845 		.data		= &sysctl_hung_task_timeout_secs,
846 		.maxlen		= sizeof(unsigned long),
847 		.mode		= 0644,
848 		.proc_handler	= &proc_dohung_task_timeout_secs,
849 		.strategy	= &sysctl_intvec,
850 	},
851 	{
852 		.ctl_name	= CTL_UNNUMBERED,
853 		.procname	= "hung_task_warnings",
854 		.data		= &sysctl_hung_task_warnings,
855 		.maxlen		= sizeof(unsigned long),
856 		.mode		= 0644,
857 		.proc_handler	= &proc_doulongvec_minmax,
858 		.strategy	= &sysctl_intvec,
859 	},
860 #endif
861 #ifdef CONFIG_COMPAT
862 	{
863 		.ctl_name	= KERN_COMPAT_LOG,
864 		.procname	= "compat-log",
865 		.data		= &compat_log,
866 		.maxlen		= sizeof (int),
867 	 	.mode		= 0644,
868 		.proc_handler	= &proc_dointvec,
869 	},
870 #endif
871 #ifdef CONFIG_RT_MUTEXES
872 	{
873 		.ctl_name	= KERN_MAX_LOCK_DEPTH,
874 		.procname	= "max_lock_depth",
875 		.data		= &max_lock_depth,
876 		.maxlen		= sizeof(int),
877 		.mode		= 0644,
878 		.proc_handler	= &proc_dointvec,
879 	},
880 #endif
881 	{
882 		.ctl_name	= CTL_UNNUMBERED,
883 		.procname	= "poweroff_cmd",
884 		.data		= &poweroff_cmd,
885 		.maxlen		= POWEROFF_CMD_PATH_LEN,
886 		.mode		= 0644,
887 		.proc_handler	= &proc_dostring,
888 		.strategy	= &sysctl_string,
889 	},
890 #ifdef CONFIG_KEYS
891 	{
892 		.ctl_name	= CTL_UNNUMBERED,
893 		.procname	= "keys",
894 		.mode		= 0555,
895 		.child		= key_sysctls,
896 	},
897 #endif
898 #ifdef CONFIG_RCU_TORTURE_TEST
899 	{
900 		.ctl_name       = CTL_UNNUMBERED,
901 		.procname       = "rcutorture_runnable",
902 		.data           = &rcutorture_runnable,
903 		.maxlen         = sizeof(int),
904 		.mode           = 0644,
905 		.proc_handler   = &proc_dointvec,
906 	},
907 #endif
908 #ifdef CONFIG_SLOW_WORK
909 	{
910 		.ctl_name	= CTL_UNNUMBERED,
911 		.procname	= "slow-work",
912 		.mode		= 0555,
913 		.child		= slow_work_sysctls,
914 	},
915 #endif
916 /*
917  * NOTE: do not add new entries to this table unless you have read
918  * Documentation/sysctl/ctl_unnumbered.txt
919  */
920 	{ .ctl_name = 0 }
921 };
922 
923 static struct ctl_table vm_table[] = {
924 	{
925 		.ctl_name	= VM_OVERCOMMIT_MEMORY,
926 		.procname	= "overcommit_memory",
927 		.data		= &sysctl_overcommit_memory,
928 		.maxlen		= sizeof(sysctl_overcommit_memory),
929 		.mode		= 0644,
930 		.proc_handler	= &proc_dointvec,
931 	},
932 	{
933 		.ctl_name	= VM_PANIC_ON_OOM,
934 		.procname	= "panic_on_oom",
935 		.data		= &sysctl_panic_on_oom,
936 		.maxlen		= sizeof(sysctl_panic_on_oom),
937 		.mode		= 0644,
938 		.proc_handler	= &proc_dointvec,
939 	},
940 	{
941 		.ctl_name	= CTL_UNNUMBERED,
942 		.procname	= "oom_kill_allocating_task",
943 		.data		= &sysctl_oom_kill_allocating_task,
944 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
945 		.mode		= 0644,
946 		.proc_handler	= &proc_dointvec,
947 	},
948 	{
949 		.ctl_name	= CTL_UNNUMBERED,
950 		.procname	= "oom_dump_tasks",
951 		.data		= &sysctl_oom_dump_tasks,
952 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
953 		.mode		= 0644,
954 		.proc_handler	= &proc_dointvec,
955 	},
956 	{
957 		.ctl_name	= VM_OVERCOMMIT_RATIO,
958 		.procname	= "overcommit_ratio",
959 		.data		= &sysctl_overcommit_ratio,
960 		.maxlen		= sizeof(sysctl_overcommit_ratio),
961 		.mode		= 0644,
962 		.proc_handler	= &proc_dointvec,
963 	},
964 	{
965 		.ctl_name	= VM_PAGE_CLUSTER,
966 		.procname	= "page-cluster",
967 		.data		= &page_cluster,
968 		.maxlen		= sizeof(int),
969 		.mode		= 0644,
970 		.proc_handler	= &proc_dointvec,
971 	},
972 	{
973 		.ctl_name	= VM_DIRTY_BACKGROUND,
974 		.procname	= "dirty_background_ratio",
975 		.data		= &dirty_background_ratio,
976 		.maxlen		= sizeof(dirty_background_ratio),
977 		.mode		= 0644,
978 		.proc_handler	= &dirty_background_ratio_handler,
979 		.strategy	= &sysctl_intvec,
980 		.extra1		= &zero,
981 		.extra2		= &one_hundred,
982 	},
983 	{
984 		.ctl_name	= CTL_UNNUMBERED,
985 		.procname	= "dirty_background_bytes",
986 		.data		= &dirty_background_bytes,
987 		.maxlen		= sizeof(dirty_background_bytes),
988 		.mode		= 0644,
989 		.proc_handler	= &dirty_background_bytes_handler,
990 		.strategy	= &sysctl_intvec,
991 		.extra1		= &one_ul,
992 	},
993 	{
994 		.ctl_name	= VM_DIRTY_RATIO,
995 		.procname	= "dirty_ratio",
996 		.data		= &vm_dirty_ratio,
997 		.maxlen		= sizeof(vm_dirty_ratio),
998 		.mode		= 0644,
999 		.proc_handler	= &dirty_ratio_handler,
1000 		.strategy	= &sysctl_intvec,
1001 		.extra1		= &zero,
1002 		.extra2		= &one_hundred,
1003 	},
1004 	{
1005 		.ctl_name	= CTL_UNNUMBERED,
1006 		.procname	= "dirty_bytes",
1007 		.data		= &vm_dirty_bytes,
1008 		.maxlen		= sizeof(vm_dirty_bytes),
1009 		.mode		= 0644,
1010 		.proc_handler	= &dirty_bytes_handler,
1011 		.strategy	= &sysctl_intvec,
1012 		.extra1		= &dirty_bytes_min,
1013 	},
1014 	{
1015 		.procname	= "dirty_writeback_centisecs",
1016 		.data		= &dirty_writeback_interval,
1017 		.maxlen		= sizeof(dirty_writeback_interval),
1018 		.mode		= 0644,
1019 		.proc_handler	= &dirty_writeback_centisecs_handler,
1020 	},
1021 	{
1022 		.procname	= "dirty_expire_centisecs",
1023 		.data		= &dirty_expire_interval,
1024 		.maxlen		= sizeof(dirty_expire_interval),
1025 		.mode		= 0644,
1026 		.proc_handler	= &proc_dointvec,
1027 	},
1028 	{
1029 		.ctl_name	= VM_NR_PDFLUSH_THREADS,
1030 		.procname	= "nr_pdflush_threads",
1031 		.data		= &nr_pdflush_threads,
1032 		.maxlen		= sizeof nr_pdflush_threads,
1033 		.mode		= 0444 /* read-only*/,
1034 		.proc_handler	= &proc_dointvec,
1035 	},
1036 	{
1037 		.ctl_name	= CTL_UNNUMBERED,
1038 		.procname	= "nr_pdflush_threads_min",
1039 		.data		= &nr_pdflush_threads_min,
1040 		.maxlen		= sizeof nr_pdflush_threads_min,
1041 		.mode		= 0644 /* read-write */,
1042 		.proc_handler	= &proc_dointvec_minmax,
1043 		.strategy	= &sysctl_intvec,
1044 		.extra1		= &one,
1045 		.extra2		= &nr_pdflush_threads_max,
1046 	},
1047 	{
1048 		.ctl_name	= CTL_UNNUMBERED,
1049 		.procname	= "nr_pdflush_threads_max",
1050 		.data		= &nr_pdflush_threads_max,
1051 		.maxlen		= sizeof nr_pdflush_threads_max,
1052 		.mode		= 0644 /* read-write */,
1053 		.proc_handler	= &proc_dointvec_minmax,
1054 		.strategy	= &sysctl_intvec,
1055 		.extra1		= &nr_pdflush_threads_min,
1056 		.extra2		= &one_thousand,
1057 	},
1058 	{
1059 		.ctl_name	= VM_SWAPPINESS,
1060 		.procname	= "swappiness",
1061 		.data		= &vm_swappiness,
1062 		.maxlen		= sizeof(vm_swappiness),
1063 		.mode		= 0644,
1064 		.proc_handler	= &proc_dointvec_minmax,
1065 		.strategy	= &sysctl_intvec,
1066 		.extra1		= &zero,
1067 		.extra2		= &one_hundred,
1068 	},
1069 #ifdef CONFIG_HUGETLB_PAGE
1070 	 {
1071 		.procname	= "nr_hugepages",
1072 		.data		= NULL,
1073 		.maxlen		= sizeof(unsigned long),
1074 		.mode		= 0644,
1075 		.proc_handler	= &hugetlb_sysctl_handler,
1076 		.extra1		= (void *)&hugetlb_zero,
1077 		.extra2		= (void *)&hugetlb_infinity,
1078 	 },
1079 	 {
1080 		.ctl_name	= VM_HUGETLB_GROUP,
1081 		.procname	= "hugetlb_shm_group",
1082 		.data		= &sysctl_hugetlb_shm_group,
1083 		.maxlen		= sizeof(gid_t),
1084 		.mode		= 0644,
1085 		.proc_handler	= &proc_dointvec,
1086 	 },
1087 	 {
1088 		.ctl_name	= CTL_UNNUMBERED,
1089 		.procname	= "hugepages_treat_as_movable",
1090 		.data		= &hugepages_treat_as_movable,
1091 		.maxlen		= sizeof(int),
1092 		.mode		= 0644,
1093 		.proc_handler	= &hugetlb_treat_movable_handler,
1094 	},
1095 	{
1096 		.ctl_name	= CTL_UNNUMBERED,
1097 		.procname	= "nr_overcommit_hugepages",
1098 		.data		= NULL,
1099 		.maxlen		= sizeof(unsigned long),
1100 		.mode		= 0644,
1101 		.proc_handler	= &hugetlb_overcommit_handler,
1102 		.extra1		= (void *)&hugetlb_zero,
1103 		.extra2		= (void *)&hugetlb_infinity,
1104 	},
1105 #endif
1106 	{
1107 		.ctl_name	= VM_LOWMEM_RESERVE_RATIO,
1108 		.procname	= "lowmem_reserve_ratio",
1109 		.data		= &sysctl_lowmem_reserve_ratio,
1110 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1111 		.mode		= 0644,
1112 		.proc_handler	= &lowmem_reserve_ratio_sysctl_handler,
1113 		.strategy	= &sysctl_intvec,
1114 	},
1115 	{
1116 		.ctl_name	= VM_DROP_PAGECACHE,
1117 		.procname	= "drop_caches",
1118 		.data		= &sysctl_drop_caches,
1119 		.maxlen		= sizeof(int),
1120 		.mode		= 0644,
1121 		.proc_handler	= drop_caches_sysctl_handler,
1122 		.strategy	= &sysctl_intvec,
1123 	},
1124 	{
1125 		.ctl_name	= VM_MIN_FREE_KBYTES,
1126 		.procname	= "min_free_kbytes",
1127 		.data		= &min_free_kbytes,
1128 		.maxlen		= sizeof(min_free_kbytes),
1129 		.mode		= 0644,
1130 		.proc_handler	= &min_free_kbytes_sysctl_handler,
1131 		.strategy	= &sysctl_intvec,
1132 		.extra1		= &zero,
1133 	},
1134 	{
1135 		.ctl_name	= VM_PERCPU_PAGELIST_FRACTION,
1136 		.procname	= "percpu_pagelist_fraction",
1137 		.data		= &percpu_pagelist_fraction,
1138 		.maxlen		= sizeof(percpu_pagelist_fraction),
1139 		.mode		= 0644,
1140 		.proc_handler	= &percpu_pagelist_fraction_sysctl_handler,
1141 		.strategy	= &sysctl_intvec,
1142 		.extra1		= &min_percpu_pagelist_fract,
1143 	},
1144 #ifdef CONFIG_MMU
1145 	{
1146 		.ctl_name	= VM_MAX_MAP_COUNT,
1147 		.procname	= "max_map_count",
1148 		.data		= &sysctl_max_map_count,
1149 		.maxlen		= sizeof(sysctl_max_map_count),
1150 		.mode		= 0644,
1151 		.proc_handler	= &proc_dointvec
1152 	},
1153 #else
1154 	{
1155 		.ctl_name	= CTL_UNNUMBERED,
1156 		.procname	= "nr_trim_pages",
1157 		.data		= &sysctl_nr_trim_pages,
1158 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1159 		.mode		= 0644,
1160 		.proc_handler	= &proc_dointvec_minmax,
1161 		.strategy	= &sysctl_intvec,
1162 		.extra1		= &zero,
1163 	},
1164 #endif
1165 	{
1166 		.ctl_name	= VM_LAPTOP_MODE,
1167 		.procname	= "laptop_mode",
1168 		.data		= &laptop_mode,
1169 		.maxlen		= sizeof(laptop_mode),
1170 		.mode		= 0644,
1171 		.proc_handler	= &proc_dointvec_jiffies,
1172 		.strategy	= &sysctl_jiffies,
1173 	},
1174 	{
1175 		.ctl_name	= VM_BLOCK_DUMP,
1176 		.procname	= "block_dump",
1177 		.data		= &block_dump,
1178 		.maxlen		= sizeof(block_dump),
1179 		.mode		= 0644,
1180 		.proc_handler	= &proc_dointvec,
1181 		.strategy	= &sysctl_intvec,
1182 		.extra1		= &zero,
1183 	},
1184 	{
1185 		.ctl_name	= VM_VFS_CACHE_PRESSURE,
1186 		.procname	= "vfs_cache_pressure",
1187 		.data		= &sysctl_vfs_cache_pressure,
1188 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1189 		.mode		= 0644,
1190 		.proc_handler	= &proc_dointvec,
1191 		.strategy	= &sysctl_intvec,
1192 		.extra1		= &zero,
1193 	},
1194 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1195 	{
1196 		.ctl_name	= VM_LEGACY_VA_LAYOUT,
1197 		.procname	= "legacy_va_layout",
1198 		.data		= &sysctl_legacy_va_layout,
1199 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1200 		.mode		= 0644,
1201 		.proc_handler	= &proc_dointvec,
1202 		.strategy	= &sysctl_intvec,
1203 		.extra1		= &zero,
1204 	},
1205 #endif
1206 #ifdef CONFIG_NUMA
1207 	{
1208 		.ctl_name	= VM_ZONE_RECLAIM_MODE,
1209 		.procname	= "zone_reclaim_mode",
1210 		.data		= &zone_reclaim_mode,
1211 		.maxlen		= sizeof(zone_reclaim_mode),
1212 		.mode		= 0644,
1213 		.proc_handler	= &proc_dointvec,
1214 		.strategy	= &sysctl_intvec,
1215 		.extra1		= &zero,
1216 	},
1217 	{
1218 		.ctl_name	= VM_MIN_UNMAPPED,
1219 		.procname	= "min_unmapped_ratio",
1220 		.data		= &sysctl_min_unmapped_ratio,
1221 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1222 		.mode		= 0644,
1223 		.proc_handler	= &sysctl_min_unmapped_ratio_sysctl_handler,
1224 		.strategy	= &sysctl_intvec,
1225 		.extra1		= &zero,
1226 		.extra2		= &one_hundred,
1227 	},
1228 	{
1229 		.ctl_name	= VM_MIN_SLAB,
1230 		.procname	= "min_slab_ratio",
1231 		.data		= &sysctl_min_slab_ratio,
1232 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1233 		.mode		= 0644,
1234 		.proc_handler	= &sysctl_min_slab_ratio_sysctl_handler,
1235 		.strategy	= &sysctl_intvec,
1236 		.extra1		= &zero,
1237 		.extra2		= &one_hundred,
1238 	},
1239 #endif
1240 #ifdef CONFIG_SMP
1241 	{
1242 		.ctl_name	= CTL_UNNUMBERED,
1243 		.procname	= "stat_interval",
1244 		.data		= &sysctl_stat_interval,
1245 		.maxlen		= sizeof(sysctl_stat_interval),
1246 		.mode		= 0644,
1247 		.proc_handler	= &proc_dointvec_jiffies,
1248 		.strategy	= &sysctl_jiffies,
1249 	},
1250 #endif
1251 #ifdef CONFIG_SECURITY
1252 	{
1253 		.ctl_name	= CTL_UNNUMBERED,
1254 		.procname	= "mmap_min_addr",
1255 		.data		= &mmap_min_addr,
1256 		.maxlen         = sizeof(unsigned long),
1257 		.mode		= 0644,
1258 		.proc_handler	= &proc_doulongvec_minmax,
1259 	},
1260 #endif
1261 #ifdef CONFIG_NUMA
1262 	{
1263 		.ctl_name	= CTL_UNNUMBERED,
1264 		.procname	= "numa_zonelist_order",
1265 		.data		= &numa_zonelist_order,
1266 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1267 		.mode		= 0644,
1268 		.proc_handler	= &numa_zonelist_order_handler,
1269 		.strategy	= &sysctl_string,
1270 	},
1271 #endif
1272 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1273    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1274 	{
1275 		.ctl_name	= VM_VDSO_ENABLED,
1276 		.procname	= "vdso_enabled",
1277 		.data		= &vdso_enabled,
1278 		.maxlen		= sizeof(vdso_enabled),
1279 		.mode		= 0644,
1280 		.proc_handler	= &proc_dointvec,
1281 		.strategy	= &sysctl_intvec,
1282 		.extra1		= &zero,
1283 	},
1284 #endif
1285 #ifdef CONFIG_HIGHMEM
1286 	{
1287 		.ctl_name	= CTL_UNNUMBERED,
1288 		.procname	= "highmem_is_dirtyable",
1289 		.data		= &vm_highmem_is_dirtyable,
1290 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1291 		.mode		= 0644,
1292 		.proc_handler	= &proc_dointvec_minmax,
1293 		.strategy	= &sysctl_intvec,
1294 		.extra1		= &zero,
1295 		.extra2		= &one,
1296 	},
1297 #endif
1298 #ifdef CONFIG_UNEVICTABLE_LRU
1299 	{
1300 		.ctl_name	= CTL_UNNUMBERED,
1301 		.procname	= "scan_unevictable_pages",
1302 		.data		= &scan_unevictable_pages,
1303 		.maxlen		= sizeof(scan_unevictable_pages),
1304 		.mode		= 0644,
1305 		.proc_handler	= &scan_unevictable_handler,
1306 	},
1307 #endif
1308 /*
1309  * NOTE: do not add new entries to this table unless you have read
1310  * Documentation/sysctl/ctl_unnumbered.txt
1311  */
1312 	{ .ctl_name = 0 }
1313 };
1314 
1315 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1316 static struct ctl_table binfmt_misc_table[] = {
1317 	{ .ctl_name = 0 }
1318 };
1319 #endif
1320 
1321 static struct ctl_table fs_table[] = {
1322 	{
1323 		.ctl_name	= FS_NRINODE,
1324 		.procname	= "inode-nr",
1325 		.data		= &inodes_stat,
1326 		.maxlen		= 2*sizeof(int),
1327 		.mode		= 0444,
1328 		.proc_handler	= &proc_dointvec,
1329 	},
1330 	{
1331 		.ctl_name	= FS_STATINODE,
1332 		.procname	= "inode-state",
1333 		.data		= &inodes_stat,
1334 		.maxlen		= 7*sizeof(int),
1335 		.mode		= 0444,
1336 		.proc_handler	= &proc_dointvec,
1337 	},
1338 	{
1339 		.procname	= "file-nr",
1340 		.data		= &files_stat,
1341 		.maxlen		= 3*sizeof(int),
1342 		.mode		= 0444,
1343 		.proc_handler	= &proc_nr_files,
1344 	},
1345 	{
1346 		.ctl_name	= FS_MAXFILE,
1347 		.procname	= "file-max",
1348 		.data		= &files_stat.max_files,
1349 		.maxlen		= sizeof(int),
1350 		.mode		= 0644,
1351 		.proc_handler	= &proc_dointvec,
1352 	},
1353 	{
1354 		.ctl_name	= CTL_UNNUMBERED,
1355 		.procname	= "nr_open",
1356 		.data		= &sysctl_nr_open,
1357 		.maxlen		= sizeof(int),
1358 		.mode		= 0644,
1359 		.proc_handler	= &proc_dointvec_minmax,
1360 		.extra1		= &sysctl_nr_open_min,
1361 		.extra2		= &sysctl_nr_open_max,
1362 	},
1363 	{
1364 		.ctl_name	= FS_DENTRY,
1365 		.procname	= "dentry-state",
1366 		.data		= &dentry_stat,
1367 		.maxlen		= 6*sizeof(int),
1368 		.mode		= 0444,
1369 		.proc_handler	= &proc_dointvec,
1370 	},
1371 	{
1372 		.ctl_name	= FS_OVERFLOWUID,
1373 		.procname	= "overflowuid",
1374 		.data		= &fs_overflowuid,
1375 		.maxlen		= sizeof(int),
1376 		.mode		= 0644,
1377 		.proc_handler	= &proc_dointvec_minmax,
1378 		.strategy	= &sysctl_intvec,
1379 		.extra1		= &minolduid,
1380 		.extra2		= &maxolduid,
1381 	},
1382 	{
1383 		.ctl_name	= FS_OVERFLOWGID,
1384 		.procname	= "overflowgid",
1385 		.data		= &fs_overflowgid,
1386 		.maxlen		= sizeof(int),
1387 		.mode		= 0644,
1388 		.proc_handler	= &proc_dointvec_minmax,
1389 		.strategy	= &sysctl_intvec,
1390 		.extra1		= &minolduid,
1391 		.extra2		= &maxolduid,
1392 	},
1393 #ifdef CONFIG_FILE_LOCKING
1394 	{
1395 		.ctl_name	= FS_LEASES,
1396 		.procname	= "leases-enable",
1397 		.data		= &leases_enable,
1398 		.maxlen		= sizeof(int),
1399 		.mode		= 0644,
1400 		.proc_handler	= &proc_dointvec,
1401 	},
1402 #endif
1403 #ifdef CONFIG_DNOTIFY
1404 	{
1405 		.ctl_name	= FS_DIR_NOTIFY,
1406 		.procname	= "dir-notify-enable",
1407 		.data		= &dir_notify_enable,
1408 		.maxlen		= sizeof(int),
1409 		.mode		= 0644,
1410 		.proc_handler	= &proc_dointvec,
1411 	},
1412 #endif
1413 #ifdef CONFIG_MMU
1414 #ifdef CONFIG_FILE_LOCKING
1415 	{
1416 		.ctl_name	= FS_LEASE_TIME,
1417 		.procname	= "lease-break-time",
1418 		.data		= &lease_break_time,
1419 		.maxlen		= sizeof(int),
1420 		.mode		= 0644,
1421 		.proc_handler	= &proc_dointvec,
1422 	},
1423 #endif
1424 #ifdef CONFIG_AIO
1425 	{
1426 		.procname	= "aio-nr",
1427 		.data		= &aio_nr,
1428 		.maxlen		= sizeof(aio_nr),
1429 		.mode		= 0444,
1430 		.proc_handler	= &proc_doulongvec_minmax,
1431 	},
1432 	{
1433 		.procname	= "aio-max-nr",
1434 		.data		= &aio_max_nr,
1435 		.maxlen		= sizeof(aio_max_nr),
1436 		.mode		= 0644,
1437 		.proc_handler	= &proc_doulongvec_minmax,
1438 	},
1439 #endif /* CONFIG_AIO */
1440 #ifdef CONFIG_INOTIFY_USER
1441 	{
1442 		.ctl_name	= FS_INOTIFY,
1443 		.procname	= "inotify",
1444 		.mode		= 0555,
1445 		.child		= inotify_table,
1446 	},
1447 #endif
1448 #ifdef CONFIG_EPOLL
1449 	{
1450 		.procname	= "epoll",
1451 		.mode		= 0555,
1452 		.child		= epoll_table,
1453 	},
1454 #endif
1455 #endif
1456 	{
1457 		.ctl_name	= KERN_SETUID_DUMPABLE,
1458 		.procname	= "suid_dumpable",
1459 		.data		= &suid_dumpable,
1460 		.maxlen		= sizeof(int),
1461 		.mode		= 0644,
1462 		.proc_handler	= &proc_dointvec_minmax,
1463 		.strategy	= &sysctl_intvec,
1464 		.extra1		= &zero,
1465 		.extra2		= &two,
1466 	},
1467 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1468 	{
1469 		.ctl_name	= CTL_UNNUMBERED,
1470 		.procname	= "binfmt_misc",
1471 		.mode		= 0555,
1472 		.child		= binfmt_misc_table,
1473 	},
1474 #endif
1475 /*
1476  * NOTE: do not add new entries to this table unless you have read
1477  * Documentation/sysctl/ctl_unnumbered.txt
1478  */
1479 	{ .ctl_name = 0 }
1480 };
1481 
1482 static struct ctl_table debug_table[] = {
1483 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1484 	{
1485 		.ctl_name	= CTL_UNNUMBERED,
1486 		.procname	= "exception-trace",
1487 		.data		= &show_unhandled_signals,
1488 		.maxlen		= sizeof(int),
1489 		.mode		= 0644,
1490 		.proc_handler	= proc_dointvec
1491 	},
1492 #endif
1493 	{ .ctl_name = 0 }
1494 };
1495 
1496 static struct ctl_table dev_table[] = {
1497 	{ .ctl_name = 0 }
1498 };
1499 
1500 static DEFINE_SPINLOCK(sysctl_lock);
1501 
1502 /* called under sysctl_lock */
1503 static int use_table(struct ctl_table_header *p)
1504 {
1505 	if (unlikely(p->unregistering))
1506 		return 0;
1507 	p->used++;
1508 	return 1;
1509 }
1510 
1511 /* called under sysctl_lock */
1512 static void unuse_table(struct ctl_table_header *p)
1513 {
1514 	if (!--p->used)
1515 		if (unlikely(p->unregistering))
1516 			complete(p->unregistering);
1517 }
1518 
1519 /* called under sysctl_lock, will reacquire if has to wait */
1520 static void start_unregistering(struct ctl_table_header *p)
1521 {
1522 	/*
1523 	 * if p->used is 0, nobody will ever touch that entry again;
1524 	 * we'll eliminate all paths to it before dropping sysctl_lock
1525 	 */
1526 	if (unlikely(p->used)) {
1527 		struct completion wait;
1528 		init_completion(&wait);
1529 		p->unregistering = &wait;
1530 		spin_unlock(&sysctl_lock);
1531 		wait_for_completion(&wait);
1532 		spin_lock(&sysctl_lock);
1533 	} else {
1534 		/* anything non-NULL; we'll never dereference it */
1535 		p->unregistering = ERR_PTR(-EINVAL);
1536 	}
1537 	/*
1538 	 * do not remove from the list until nobody holds it; walking the
1539 	 * list in do_sysctl() relies on that.
1540 	 */
1541 	list_del_init(&p->ctl_entry);
1542 }
1543 
1544 void sysctl_head_get(struct ctl_table_header *head)
1545 {
1546 	spin_lock(&sysctl_lock);
1547 	head->count++;
1548 	spin_unlock(&sysctl_lock);
1549 }
1550 
1551 void sysctl_head_put(struct ctl_table_header *head)
1552 {
1553 	spin_lock(&sysctl_lock);
1554 	if (!--head->count)
1555 		kfree(head);
1556 	spin_unlock(&sysctl_lock);
1557 }
1558 
1559 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1560 {
1561 	if (!head)
1562 		BUG();
1563 	spin_lock(&sysctl_lock);
1564 	if (!use_table(head))
1565 		head = ERR_PTR(-ENOENT);
1566 	spin_unlock(&sysctl_lock);
1567 	return head;
1568 }
1569 
1570 void sysctl_head_finish(struct ctl_table_header *head)
1571 {
1572 	if (!head)
1573 		return;
1574 	spin_lock(&sysctl_lock);
1575 	unuse_table(head);
1576 	spin_unlock(&sysctl_lock);
1577 }
1578 
1579 static struct ctl_table_set *
1580 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1581 {
1582 	struct ctl_table_set *set = &root->default_set;
1583 	if (root->lookup)
1584 		set = root->lookup(root, namespaces);
1585 	return set;
1586 }
1587 
1588 static struct list_head *
1589 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1590 {
1591 	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1592 	return &set->list;
1593 }
1594 
1595 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1596 					    struct ctl_table_header *prev)
1597 {
1598 	struct ctl_table_root *root;
1599 	struct list_head *header_list;
1600 	struct ctl_table_header *head;
1601 	struct list_head *tmp;
1602 
1603 	spin_lock(&sysctl_lock);
1604 	if (prev) {
1605 		head = prev;
1606 		tmp = &prev->ctl_entry;
1607 		unuse_table(prev);
1608 		goto next;
1609 	}
1610 	tmp = &root_table_header.ctl_entry;
1611 	for (;;) {
1612 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1613 
1614 		if (!use_table(head))
1615 			goto next;
1616 		spin_unlock(&sysctl_lock);
1617 		return head;
1618 	next:
1619 		root = head->root;
1620 		tmp = tmp->next;
1621 		header_list = lookup_header_list(root, namespaces);
1622 		if (tmp != header_list)
1623 			continue;
1624 
1625 		do {
1626 			root = list_entry(root->root_list.next,
1627 					struct ctl_table_root, root_list);
1628 			if (root == &sysctl_table_root)
1629 				goto out;
1630 			header_list = lookup_header_list(root, namespaces);
1631 		} while (list_empty(header_list));
1632 		tmp = header_list->next;
1633 	}
1634 out:
1635 	spin_unlock(&sysctl_lock);
1636 	return NULL;
1637 }
1638 
1639 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1640 {
1641 	return __sysctl_head_next(current->nsproxy, prev);
1642 }
1643 
1644 void register_sysctl_root(struct ctl_table_root *root)
1645 {
1646 	spin_lock(&sysctl_lock);
1647 	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1648 	spin_unlock(&sysctl_lock);
1649 }
1650 
1651 #ifdef CONFIG_SYSCTL_SYSCALL
1652 /* Perform the actual read/write of a sysctl table entry. */
1653 static int do_sysctl_strategy(struct ctl_table_root *root,
1654 			struct ctl_table *table,
1655 			void __user *oldval, size_t __user *oldlenp,
1656 			void __user *newval, size_t newlen)
1657 {
1658 	int op = 0, rc;
1659 
1660 	if (oldval)
1661 		op |= MAY_READ;
1662 	if (newval)
1663 		op |= MAY_WRITE;
1664 	if (sysctl_perm(root, table, op))
1665 		return -EPERM;
1666 
1667 	if (table->strategy) {
1668 		rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1669 		if (rc < 0)
1670 			return rc;
1671 		if (rc > 0)
1672 			return 0;
1673 	}
1674 
1675 	/* If there is no strategy routine, or if the strategy returns
1676 	 * zero, proceed with automatic r/w */
1677 	if (table->data && table->maxlen) {
1678 		rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1679 		if (rc < 0)
1680 			return rc;
1681 	}
1682 	return 0;
1683 }
1684 
1685 static int parse_table(int __user *name, int nlen,
1686 		       void __user *oldval, size_t __user *oldlenp,
1687 		       void __user *newval, size_t newlen,
1688 		       struct ctl_table_root *root,
1689 		       struct ctl_table *table)
1690 {
1691 	int n;
1692 repeat:
1693 	if (!nlen)
1694 		return -ENOTDIR;
1695 	if (get_user(n, name))
1696 		return -EFAULT;
1697 	for ( ; table->ctl_name || table->procname; table++) {
1698 		if (!table->ctl_name)
1699 			continue;
1700 		if (n == table->ctl_name) {
1701 			int error;
1702 			if (table->child) {
1703 				if (sysctl_perm(root, table, MAY_EXEC))
1704 					return -EPERM;
1705 				name++;
1706 				nlen--;
1707 				table = table->child;
1708 				goto repeat;
1709 			}
1710 			error = do_sysctl_strategy(root, table,
1711 						   oldval, oldlenp,
1712 						   newval, newlen);
1713 			return error;
1714 		}
1715 	}
1716 	return -ENOTDIR;
1717 }
1718 
1719 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1720 	       void __user *newval, size_t newlen)
1721 {
1722 	struct ctl_table_header *head;
1723 	int error = -ENOTDIR;
1724 
1725 	if (nlen <= 0 || nlen >= CTL_MAXNAME)
1726 		return -ENOTDIR;
1727 	if (oldval) {
1728 		int old_len;
1729 		if (!oldlenp || get_user(old_len, oldlenp))
1730 			return -EFAULT;
1731 	}
1732 
1733 	for (head = sysctl_head_next(NULL); head;
1734 			head = sysctl_head_next(head)) {
1735 		error = parse_table(name, nlen, oldval, oldlenp,
1736 					newval, newlen,
1737 					head->root, head->ctl_table);
1738 		if (error != -ENOTDIR) {
1739 			sysctl_head_finish(head);
1740 			break;
1741 		}
1742 	}
1743 	return error;
1744 }
1745 
1746 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
1747 {
1748 	struct __sysctl_args tmp;
1749 	int error;
1750 
1751 	if (copy_from_user(&tmp, args, sizeof(tmp)))
1752 		return -EFAULT;
1753 
1754 	error = deprecated_sysctl_warning(&tmp);
1755 	if (error)
1756 		goto out;
1757 
1758 	lock_kernel();
1759 	error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1760 			  tmp.newval, tmp.newlen);
1761 	unlock_kernel();
1762 out:
1763 	return error;
1764 }
1765 #endif /* CONFIG_SYSCTL_SYSCALL */
1766 
1767 /*
1768  * sysctl_perm does NOT grant the superuser all rights automatically, because
1769  * some sysctl variables are readonly even to root.
1770  */
1771 
1772 static int test_perm(int mode, int op)
1773 {
1774 	if (!current_euid())
1775 		mode >>= 6;
1776 	else if (in_egroup_p(0))
1777 		mode >>= 3;
1778 	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1779 		return 0;
1780 	return -EACCES;
1781 }
1782 
1783 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1784 {
1785 	int error;
1786 	int mode;
1787 
1788 	error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1789 	if (error)
1790 		return error;
1791 
1792 	if (root->permissions)
1793 		mode = root->permissions(root, current->nsproxy, table);
1794 	else
1795 		mode = table->mode;
1796 
1797 	return test_perm(mode, op);
1798 }
1799 
1800 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1801 {
1802 	for (; table->ctl_name || table->procname; table++) {
1803 		table->parent = parent;
1804 		if (table->child)
1805 			sysctl_set_parent(table, table->child);
1806 	}
1807 }
1808 
1809 static __init int sysctl_init(void)
1810 {
1811 	sysctl_set_parent(NULL, root_table);
1812 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1813 	{
1814 		int err;
1815 		err = sysctl_check_table(current->nsproxy, root_table);
1816 	}
1817 #endif
1818 	return 0;
1819 }
1820 
1821 core_initcall(sysctl_init);
1822 
1823 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1824 				      struct ctl_table *table)
1825 {
1826 	struct ctl_table *p;
1827 	const char *s = branch->procname;
1828 
1829 	/* branch should have named subdirectory as its first element */
1830 	if (!s || !branch->child)
1831 		return NULL;
1832 
1833 	/* ... and nothing else */
1834 	if (branch[1].procname || branch[1].ctl_name)
1835 		return NULL;
1836 
1837 	/* table should contain subdirectory with the same name */
1838 	for (p = table; p->procname || p->ctl_name; p++) {
1839 		if (!p->child)
1840 			continue;
1841 		if (p->procname && strcmp(p->procname, s) == 0)
1842 			return p;
1843 	}
1844 	return NULL;
1845 }
1846 
1847 /* see if attaching q to p would be an improvement */
1848 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1849 {
1850 	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1851 	struct ctl_table *next;
1852 	int is_better = 0;
1853 	int not_in_parent = !p->attached_by;
1854 
1855 	while ((next = is_branch_in(by, to)) != NULL) {
1856 		if (by == q->attached_by)
1857 			is_better = 1;
1858 		if (to == p->attached_by)
1859 			not_in_parent = 1;
1860 		by = by->child;
1861 		to = next->child;
1862 	}
1863 
1864 	if (is_better && not_in_parent) {
1865 		q->attached_by = by;
1866 		q->attached_to = to;
1867 		q->parent = p;
1868 	}
1869 }
1870 
1871 /**
1872  * __register_sysctl_paths - register a sysctl hierarchy
1873  * @root: List of sysctl headers to register on
1874  * @namespaces: Data to compute which lists of sysctl entries are visible
1875  * @path: The path to the directory the sysctl table is in.
1876  * @table: the top-level table structure
1877  *
1878  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1879  * array. A completely 0 filled entry terminates the table.
1880  *
1881  * The members of the &struct ctl_table structure are used as follows:
1882  *
1883  * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1884  *            must be unique within that level of sysctl
1885  *
1886  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1887  *            enter a sysctl file
1888  *
1889  * data - a pointer to data for use by proc_handler
1890  *
1891  * maxlen - the maximum size in bytes of the data
1892  *
1893  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1894  *
1895  * child - a pointer to the child sysctl table if this entry is a directory, or
1896  *         %NULL.
1897  *
1898  * proc_handler - the text handler routine (described below)
1899  *
1900  * strategy - the strategy routine (described below)
1901  *
1902  * de - for internal use by the sysctl routines
1903  *
1904  * extra1, extra2 - extra pointers usable by the proc handler routines
1905  *
1906  * Leaf nodes in the sysctl tree will be represented by a single file
1907  * under /proc; non-leaf nodes will be represented by directories.
1908  *
1909  * sysctl(2) can automatically manage read and write requests through
1910  * the sysctl table.  The data and maxlen fields of the ctl_table
1911  * struct enable minimal validation of the values being written to be
1912  * performed, and the mode field allows minimal authentication.
1913  *
1914  * More sophisticated management can be enabled by the provision of a
1915  * strategy routine with the table entry.  This will be called before
1916  * any automatic read or write of the data is performed.
1917  *
1918  * The strategy routine may return
1919  *
1920  * < 0 - Error occurred (error is passed to user process)
1921  *
1922  * 0   - OK - proceed with automatic read or write.
1923  *
1924  * > 0 - OK - read or write has been done by the strategy routine, so
1925  *       return immediately.
1926  *
1927  * There must be a proc_handler routine for any terminal nodes
1928  * mirrored under /proc/sys (non-terminals are handled by a built-in
1929  * directory handler).  Several default handlers are available to
1930  * cover common cases -
1931  *
1932  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1933  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1934  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1935  *
1936  * It is the handler's job to read the input buffer from user memory
1937  * and process it. The handler should return 0 on success.
1938  *
1939  * This routine returns %NULL on a failure to register, and a pointer
1940  * to the table header on success.
1941  */
1942 struct ctl_table_header *__register_sysctl_paths(
1943 	struct ctl_table_root *root,
1944 	struct nsproxy *namespaces,
1945 	const struct ctl_path *path, struct ctl_table *table)
1946 {
1947 	struct ctl_table_header *header;
1948 	struct ctl_table *new, **prevp;
1949 	unsigned int n, npath;
1950 	struct ctl_table_set *set;
1951 
1952 	/* Count the path components */
1953 	for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1954 		;
1955 
1956 	/*
1957 	 * For each path component, allocate a 2-element ctl_table array.
1958 	 * The first array element will be filled with the sysctl entry
1959 	 * for this, the second will be the sentinel (ctl_name == 0).
1960 	 *
1961 	 * We allocate everything in one go so that we don't have to
1962 	 * worry about freeing additional memory in unregister_sysctl_table.
1963 	 */
1964 	header = kzalloc(sizeof(struct ctl_table_header) +
1965 			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1966 	if (!header)
1967 		return NULL;
1968 
1969 	new = (struct ctl_table *) (header + 1);
1970 
1971 	/* Now connect the dots */
1972 	prevp = &header->ctl_table;
1973 	for (n = 0; n < npath; ++n, ++path) {
1974 		/* Copy the procname */
1975 		new->procname = path->procname;
1976 		new->ctl_name = path->ctl_name;
1977 		new->mode     = 0555;
1978 
1979 		*prevp = new;
1980 		prevp = &new->child;
1981 
1982 		new += 2;
1983 	}
1984 	*prevp = table;
1985 	header->ctl_table_arg = table;
1986 
1987 	INIT_LIST_HEAD(&header->ctl_entry);
1988 	header->used = 0;
1989 	header->unregistering = NULL;
1990 	header->root = root;
1991 	sysctl_set_parent(NULL, header->ctl_table);
1992 	header->count = 1;
1993 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1994 	if (sysctl_check_table(namespaces, header->ctl_table)) {
1995 		kfree(header);
1996 		return NULL;
1997 	}
1998 #endif
1999 	spin_lock(&sysctl_lock);
2000 	header->set = lookup_header_set(root, namespaces);
2001 	header->attached_by = header->ctl_table;
2002 	header->attached_to = root_table;
2003 	header->parent = &root_table_header;
2004 	for (set = header->set; set; set = set->parent) {
2005 		struct ctl_table_header *p;
2006 		list_for_each_entry(p, &set->list, ctl_entry) {
2007 			if (p->unregistering)
2008 				continue;
2009 			try_attach(p, header);
2010 		}
2011 	}
2012 	header->parent->count++;
2013 	list_add_tail(&header->ctl_entry, &header->set->list);
2014 	spin_unlock(&sysctl_lock);
2015 
2016 	return header;
2017 }
2018 
2019 /**
2020  * register_sysctl_table_path - register a sysctl table hierarchy
2021  * @path: The path to the directory the sysctl table is in.
2022  * @table: the top-level table structure
2023  *
2024  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2025  * array. A completely 0 filled entry terminates the table.
2026  *
2027  * See __register_sysctl_paths for more details.
2028  */
2029 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2030 						struct ctl_table *table)
2031 {
2032 	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
2033 					path, table);
2034 }
2035 
2036 /**
2037  * register_sysctl_table - register a sysctl table hierarchy
2038  * @table: the top-level table structure
2039  *
2040  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2041  * array. A completely 0 filled entry terminates the table.
2042  *
2043  * See register_sysctl_paths for more details.
2044  */
2045 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
2046 {
2047 	static const struct ctl_path null_path[] = { {} };
2048 
2049 	return register_sysctl_paths(null_path, table);
2050 }
2051 
2052 /**
2053  * unregister_sysctl_table - unregister a sysctl table hierarchy
2054  * @header: the header returned from register_sysctl_table
2055  *
2056  * Unregisters the sysctl table and all children. proc entries may not
2057  * actually be removed until they are no longer used by anyone.
2058  */
2059 void unregister_sysctl_table(struct ctl_table_header * header)
2060 {
2061 	might_sleep();
2062 
2063 	if (header == NULL)
2064 		return;
2065 
2066 	spin_lock(&sysctl_lock);
2067 	start_unregistering(header);
2068 	if (!--header->parent->count) {
2069 		WARN_ON(1);
2070 		kfree(header->parent);
2071 	}
2072 	if (!--header->count)
2073 		kfree(header);
2074 	spin_unlock(&sysctl_lock);
2075 }
2076 
2077 int sysctl_is_seen(struct ctl_table_header *p)
2078 {
2079 	struct ctl_table_set *set = p->set;
2080 	int res;
2081 	spin_lock(&sysctl_lock);
2082 	if (p->unregistering)
2083 		res = 0;
2084 	else if (!set->is_seen)
2085 		res = 1;
2086 	else
2087 		res = set->is_seen(set);
2088 	spin_unlock(&sysctl_lock);
2089 	return res;
2090 }
2091 
2092 void setup_sysctl_set(struct ctl_table_set *p,
2093 	struct ctl_table_set *parent,
2094 	int (*is_seen)(struct ctl_table_set *))
2095 {
2096 	INIT_LIST_HEAD(&p->list);
2097 	p->parent = parent ? parent : &sysctl_table_root.default_set;
2098 	p->is_seen = is_seen;
2099 }
2100 
2101 #else /* !CONFIG_SYSCTL */
2102 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2103 {
2104 	return NULL;
2105 }
2106 
2107 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2108 						    struct ctl_table *table)
2109 {
2110 	return NULL;
2111 }
2112 
2113 void unregister_sysctl_table(struct ctl_table_header * table)
2114 {
2115 }
2116 
2117 void setup_sysctl_set(struct ctl_table_set *p,
2118 	struct ctl_table_set *parent,
2119 	int (*is_seen)(struct ctl_table_set *))
2120 {
2121 }
2122 
2123 void sysctl_head_put(struct ctl_table_header *head)
2124 {
2125 }
2126 
2127 #endif /* CONFIG_SYSCTL */
2128 
2129 /*
2130  * /proc/sys support
2131  */
2132 
2133 #ifdef CONFIG_PROC_SYSCTL
2134 
2135 static int _proc_do_string(void* data, int maxlen, int write,
2136 			   struct file *filp, void __user *buffer,
2137 			   size_t *lenp, loff_t *ppos)
2138 {
2139 	size_t len;
2140 	char __user *p;
2141 	char c;
2142 
2143 	if (!data || !maxlen || !*lenp) {
2144 		*lenp = 0;
2145 		return 0;
2146 	}
2147 
2148 	if (write) {
2149 		len = 0;
2150 		p = buffer;
2151 		while (len < *lenp) {
2152 			if (get_user(c, p++))
2153 				return -EFAULT;
2154 			if (c == 0 || c == '\n')
2155 				break;
2156 			len++;
2157 		}
2158 		if (len >= maxlen)
2159 			len = maxlen-1;
2160 		if(copy_from_user(data, buffer, len))
2161 			return -EFAULT;
2162 		((char *) data)[len] = 0;
2163 		*ppos += *lenp;
2164 	} else {
2165 		len = strlen(data);
2166 		if (len > maxlen)
2167 			len = maxlen;
2168 
2169 		if (*ppos > len) {
2170 			*lenp = 0;
2171 			return 0;
2172 		}
2173 
2174 		data += *ppos;
2175 		len  -= *ppos;
2176 
2177 		if (len > *lenp)
2178 			len = *lenp;
2179 		if (len)
2180 			if(copy_to_user(buffer, data, len))
2181 				return -EFAULT;
2182 		if (len < *lenp) {
2183 			if(put_user('\n', ((char __user *) buffer) + len))
2184 				return -EFAULT;
2185 			len++;
2186 		}
2187 		*lenp = len;
2188 		*ppos += len;
2189 	}
2190 	return 0;
2191 }
2192 
2193 /**
2194  * proc_dostring - read a string sysctl
2195  * @table: the sysctl table
2196  * @write: %TRUE if this is a write to the sysctl file
2197  * @filp: the file structure
2198  * @buffer: the user buffer
2199  * @lenp: the size of the user buffer
2200  * @ppos: file position
2201  *
2202  * Reads/writes a string from/to the user buffer. If the kernel
2203  * buffer provided is not large enough to hold the string, the
2204  * string is truncated. The copied string is %NULL-terminated.
2205  * If the string is being read by the user process, it is copied
2206  * and a newline '\n' is added. It is truncated if the buffer is
2207  * not large enough.
2208  *
2209  * Returns 0 on success.
2210  */
2211 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2212 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2213 {
2214 	return _proc_do_string(table->data, table->maxlen, write, filp,
2215 			       buffer, lenp, ppos);
2216 }
2217 
2218 
2219 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2220 				 int *valp,
2221 				 int write, void *data)
2222 {
2223 	if (write) {
2224 		*valp = *negp ? -*lvalp : *lvalp;
2225 	} else {
2226 		int val = *valp;
2227 		if (val < 0) {
2228 			*negp = -1;
2229 			*lvalp = (unsigned long)-val;
2230 		} else {
2231 			*negp = 0;
2232 			*lvalp = (unsigned long)val;
2233 		}
2234 	}
2235 	return 0;
2236 }
2237 
2238 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2239 		  int write, struct file *filp, void __user *buffer,
2240 		  size_t *lenp, loff_t *ppos,
2241 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2242 			      int write, void *data),
2243 		  void *data)
2244 {
2245 #define TMPBUFLEN 21
2246 	int *i, vleft, first=1, neg, val;
2247 	unsigned long lval;
2248 	size_t left, len;
2249 
2250 	char buf[TMPBUFLEN], *p;
2251 	char __user *s = buffer;
2252 
2253 	if (!tbl_data || !table->maxlen || !*lenp ||
2254 	    (*ppos && !write)) {
2255 		*lenp = 0;
2256 		return 0;
2257 	}
2258 
2259 	i = (int *) tbl_data;
2260 	vleft = table->maxlen / sizeof(*i);
2261 	left = *lenp;
2262 
2263 	if (!conv)
2264 		conv = do_proc_dointvec_conv;
2265 
2266 	for (; left && vleft--; i++, first=0) {
2267 		if (write) {
2268 			while (left) {
2269 				char c;
2270 				if (get_user(c, s))
2271 					return -EFAULT;
2272 				if (!isspace(c))
2273 					break;
2274 				left--;
2275 				s++;
2276 			}
2277 			if (!left)
2278 				break;
2279 			neg = 0;
2280 			len = left;
2281 			if (len > sizeof(buf) - 1)
2282 				len = sizeof(buf) - 1;
2283 			if (copy_from_user(buf, s, len))
2284 				return -EFAULT;
2285 			buf[len] = 0;
2286 			p = buf;
2287 			if (*p == '-' && left > 1) {
2288 				neg = 1;
2289 				p++;
2290 			}
2291 			if (*p < '0' || *p > '9')
2292 				break;
2293 
2294 			lval = simple_strtoul(p, &p, 0);
2295 
2296 			len = p-buf;
2297 			if ((len < left) && *p && !isspace(*p))
2298 				break;
2299 			if (neg)
2300 				val = -val;
2301 			s += len;
2302 			left -= len;
2303 
2304 			if (conv(&neg, &lval, i, 1, data))
2305 				break;
2306 		} else {
2307 			p = buf;
2308 			if (!first)
2309 				*p++ = '\t';
2310 
2311 			if (conv(&neg, &lval, i, 0, data))
2312 				break;
2313 
2314 			sprintf(p, "%s%lu", neg ? "-" : "", lval);
2315 			len = strlen(buf);
2316 			if (len > left)
2317 				len = left;
2318 			if(copy_to_user(s, buf, len))
2319 				return -EFAULT;
2320 			left -= len;
2321 			s += len;
2322 		}
2323 	}
2324 
2325 	if (!write && !first && left) {
2326 		if(put_user('\n', s))
2327 			return -EFAULT;
2328 		left--, s++;
2329 	}
2330 	if (write) {
2331 		while (left) {
2332 			char c;
2333 			if (get_user(c, s++))
2334 				return -EFAULT;
2335 			if (!isspace(c))
2336 				break;
2337 			left--;
2338 		}
2339 	}
2340 	if (write && first)
2341 		return -EINVAL;
2342 	*lenp -= left;
2343 	*ppos += *lenp;
2344 	return 0;
2345 #undef TMPBUFLEN
2346 }
2347 
2348 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2349 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2350 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2351 			      int write, void *data),
2352 		  void *data)
2353 {
2354 	return __do_proc_dointvec(table->data, table, write, filp,
2355 			buffer, lenp, ppos, conv, data);
2356 }
2357 
2358 /**
2359  * proc_dointvec - read a vector of integers
2360  * @table: the sysctl table
2361  * @write: %TRUE if this is a write to the sysctl file
2362  * @filp: the file structure
2363  * @buffer: the user buffer
2364  * @lenp: the size of the user buffer
2365  * @ppos: file position
2366  *
2367  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2368  * values from/to the user buffer, treated as an ASCII string.
2369  *
2370  * Returns 0 on success.
2371  */
2372 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2373 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2374 {
2375     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2376 		    	    NULL,NULL);
2377 }
2378 
2379 /*
2380  * Taint values can only be increased
2381  * This means we can safely use a temporary.
2382  */
2383 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
2384 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2385 {
2386 	struct ctl_table t;
2387 	unsigned long tmptaint = get_taint();
2388 	int err;
2389 
2390 	if (write && !capable(CAP_SYS_ADMIN))
2391 		return -EPERM;
2392 
2393 	t = *table;
2394 	t.data = &tmptaint;
2395 	err = proc_doulongvec_minmax(&t, write, filp, buffer, lenp, ppos);
2396 	if (err < 0)
2397 		return err;
2398 
2399 	if (write) {
2400 		/*
2401 		 * Poor man's atomic or. Not worth adding a primitive
2402 		 * to everyone's atomic.h for this
2403 		 */
2404 		int i;
2405 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2406 			if ((tmptaint >> i) & 1)
2407 				add_taint(i);
2408 		}
2409 	}
2410 
2411 	return err;
2412 }
2413 
2414 struct do_proc_dointvec_minmax_conv_param {
2415 	int *min;
2416 	int *max;
2417 };
2418 
2419 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2420 					int *valp,
2421 					int write, void *data)
2422 {
2423 	struct do_proc_dointvec_minmax_conv_param *param = data;
2424 	if (write) {
2425 		int val = *negp ? -*lvalp : *lvalp;
2426 		if ((param->min && *param->min > val) ||
2427 		    (param->max && *param->max < val))
2428 			return -EINVAL;
2429 		*valp = val;
2430 	} else {
2431 		int val = *valp;
2432 		if (val < 0) {
2433 			*negp = -1;
2434 			*lvalp = (unsigned long)-val;
2435 		} else {
2436 			*negp = 0;
2437 			*lvalp = (unsigned long)val;
2438 		}
2439 	}
2440 	return 0;
2441 }
2442 
2443 /**
2444  * proc_dointvec_minmax - read a vector of integers with min/max values
2445  * @table: the sysctl table
2446  * @write: %TRUE if this is a write to the sysctl file
2447  * @filp: the file structure
2448  * @buffer: the user buffer
2449  * @lenp: the size of the user buffer
2450  * @ppos: file position
2451  *
2452  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2453  * values from/to the user buffer, treated as an ASCII string.
2454  *
2455  * This routine will ensure the values are within the range specified by
2456  * table->extra1 (min) and table->extra2 (max).
2457  *
2458  * Returns 0 on success.
2459  */
2460 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2461 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2462 {
2463 	struct do_proc_dointvec_minmax_conv_param param = {
2464 		.min = (int *) table->extra1,
2465 		.max = (int *) table->extra2,
2466 	};
2467 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2468 				do_proc_dointvec_minmax_conv, &param);
2469 }
2470 
2471 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2472 				     struct file *filp,
2473 				     void __user *buffer,
2474 				     size_t *lenp, loff_t *ppos,
2475 				     unsigned long convmul,
2476 				     unsigned long convdiv)
2477 {
2478 #define TMPBUFLEN 21
2479 	unsigned long *i, *min, *max, val;
2480 	int vleft, first=1, neg;
2481 	size_t len, left;
2482 	char buf[TMPBUFLEN], *p;
2483 	char __user *s = buffer;
2484 
2485 	if (!data || !table->maxlen || !*lenp ||
2486 	    (*ppos && !write)) {
2487 		*lenp = 0;
2488 		return 0;
2489 	}
2490 
2491 	i = (unsigned long *) data;
2492 	min = (unsigned long *) table->extra1;
2493 	max = (unsigned long *) table->extra2;
2494 	vleft = table->maxlen / sizeof(unsigned long);
2495 	left = *lenp;
2496 
2497 	for (; left && vleft--; i++, min++, max++, first=0) {
2498 		if (write) {
2499 			while (left) {
2500 				char c;
2501 				if (get_user(c, s))
2502 					return -EFAULT;
2503 				if (!isspace(c))
2504 					break;
2505 				left--;
2506 				s++;
2507 			}
2508 			if (!left)
2509 				break;
2510 			neg = 0;
2511 			len = left;
2512 			if (len > TMPBUFLEN-1)
2513 				len = TMPBUFLEN-1;
2514 			if (copy_from_user(buf, s, len))
2515 				return -EFAULT;
2516 			buf[len] = 0;
2517 			p = buf;
2518 			if (*p == '-' && left > 1) {
2519 				neg = 1;
2520 				p++;
2521 			}
2522 			if (*p < '0' || *p > '9')
2523 				break;
2524 			val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2525 			len = p-buf;
2526 			if ((len < left) && *p && !isspace(*p))
2527 				break;
2528 			if (neg)
2529 				val = -val;
2530 			s += len;
2531 			left -= len;
2532 
2533 			if(neg)
2534 				continue;
2535 			if ((min && val < *min) || (max && val > *max))
2536 				continue;
2537 			*i = val;
2538 		} else {
2539 			p = buf;
2540 			if (!first)
2541 				*p++ = '\t';
2542 			sprintf(p, "%lu", convdiv * (*i) / convmul);
2543 			len = strlen(buf);
2544 			if (len > left)
2545 				len = left;
2546 			if(copy_to_user(s, buf, len))
2547 				return -EFAULT;
2548 			left -= len;
2549 			s += len;
2550 		}
2551 	}
2552 
2553 	if (!write && !first && left) {
2554 		if(put_user('\n', s))
2555 			return -EFAULT;
2556 		left--, s++;
2557 	}
2558 	if (write) {
2559 		while (left) {
2560 			char c;
2561 			if (get_user(c, s++))
2562 				return -EFAULT;
2563 			if (!isspace(c))
2564 				break;
2565 			left--;
2566 		}
2567 	}
2568 	if (write && first)
2569 		return -EINVAL;
2570 	*lenp -= left;
2571 	*ppos += *lenp;
2572 	return 0;
2573 #undef TMPBUFLEN
2574 }
2575 
2576 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2577 				     struct file *filp,
2578 				     void __user *buffer,
2579 				     size_t *lenp, loff_t *ppos,
2580 				     unsigned long convmul,
2581 				     unsigned long convdiv)
2582 {
2583 	return __do_proc_doulongvec_minmax(table->data, table, write,
2584 			filp, buffer, lenp, ppos, convmul, convdiv);
2585 }
2586 
2587 /**
2588  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2589  * @table: the sysctl table
2590  * @write: %TRUE if this is a write to the sysctl file
2591  * @filp: the file structure
2592  * @buffer: the user buffer
2593  * @lenp: the size of the user buffer
2594  * @ppos: file position
2595  *
2596  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2597  * values from/to the user buffer, treated as an ASCII string.
2598  *
2599  * This routine will ensure the values are within the range specified by
2600  * table->extra1 (min) and table->extra2 (max).
2601  *
2602  * Returns 0 on success.
2603  */
2604 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2605 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2606 {
2607     return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2608 }
2609 
2610 /**
2611  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2612  * @table: the sysctl table
2613  * @write: %TRUE if this is a write to the sysctl file
2614  * @filp: the file structure
2615  * @buffer: the user buffer
2616  * @lenp: the size of the user buffer
2617  * @ppos: file position
2618  *
2619  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2620  * values from/to the user buffer, treated as an ASCII string. The values
2621  * are treated as milliseconds, and converted to jiffies when they are stored.
2622  *
2623  * This routine will ensure the values are within the range specified by
2624  * table->extra1 (min) and table->extra2 (max).
2625  *
2626  * Returns 0 on success.
2627  */
2628 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2629 				      struct file *filp,
2630 				      void __user *buffer,
2631 				      size_t *lenp, loff_t *ppos)
2632 {
2633     return do_proc_doulongvec_minmax(table, write, filp, buffer,
2634 				     lenp, ppos, HZ, 1000l);
2635 }
2636 
2637 
2638 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2639 					 int *valp,
2640 					 int write, void *data)
2641 {
2642 	if (write) {
2643 		if (*lvalp > LONG_MAX / HZ)
2644 			return 1;
2645 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2646 	} else {
2647 		int val = *valp;
2648 		unsigned long lval;
2649 		if (val < 0) {
2650 			*negp = -1;
2651 			lval = (unsigned long)-val;
2652 		} else {
2653 			*negp = 0;
2654 			lval = (unsigned long)val;
2655 		}
2656 		*lvalp = lval / HZ;
2657 	}
2658 	return 0;
2659 }
2660 
2661 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2662 						int *valp,
2663 						int write, void *data)
2664 {
2665 	if (write) {
2666 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2667 			return 1;
2668 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2669 	} else {
2670 		int val = *valp;
2671 		unsigned long lval;
2672 		if (val < 0) {
2673 			*negp = -1;
2674 			lval = (unsigned long)-val;
2675 		} else {
2676 			*negp = 0;
2677 			lval = (unsigned long)val;
2678 		}
2679 		*lvalp = jiffies_to_clock_t(lval);
2680 	}
2681 	return 0;
2682 }
2683 
2684 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2685 					    int *valp,
2686 					    int write, void *data)
2687 {
2688 	if (write) {
2689 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2690 	} else {
2691 		int val = *valp;
2692 		unsigned long lval;
2693 		if (val < 0) {
2694 			*negp = -1;
2695 			lval = (unsigned long)-val;
2696 		} else {
2697 			*negp = 0;
2698 			lval = (unsigned long)val;
2699 		}
2700 		*lvalp = jiffies_to_msecs(lval);
2701 	}
2702 	return 0;
2703 }
2704 
2705 /**
2706  * proc_dointvec_jiffies - read a vector of integers as seconds
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  *
2714  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2715  * values from/to the user buffer, treated as an ASCII string.
2716  * The values read are assumed to be in seconds, and are converted into
2717  * jiffies.
2718  *
2719  * Returns 0 on success.
2720  */
2721 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2722 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2723 {
2724     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2725 		    	    do_proc_dointvec_jiffies_conv,NULL);
2726 }
2727 
2728 /**
2729  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2730  * @table: the sysctl table
2731  * @write: %TRUE if this is a write to the sysctl file
2732  * @filp: the file structure
2733  * @buffer: the user buffer
2734  * @lenp: the size of the user buffer
2735  * @ppos: pointer to the file position
2736  *
2737  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2738  * values from/to the user buffer, treated as an ASCII string.
2739  * The values read are assumed to be in 1/USER_HZ seconds, and
2740  * are converted into jiffies.
2741  *
2742  * Returns 0 on success.
2743  */
2744 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2745 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2746 {
2747     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2748 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2749 }
2750 
2751 /**
2752  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2753  * @table: the sysctl table
2754  * @write: %TRUE if this is a write to the sysctl file
2755  * @filp: the file structure
2756  * @buffer: the user buffer
2757  * @lenp: the size of the user buffer
2758  * @ppos: file position
2759  * @ppos: the current position in the file
2760  *
2761  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2762  * values from/to the user buffer, treated as an ASCII string.
2763  * The values read are assumed to be in 1/1000 seconds, and
2764  * are converted into jiffies.
2765  *
2766  * Returns 0 on success.
2767  */
2768 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2769 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2770 {
2771 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2772 				do_proc_dointvec_ms_jiffies_conv, NULL);
2773 }
2774 
2775 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2776 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2777 {
2778 	struct pid *new_pid;
2779 	pid_t tmp;
2780 	int r;
2781 
2782 	tmp = pid_vnr(cad_pid);
2783 
2784 	r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2785 			       lenp, ppos, NULL, NULL);
2786 	if (r || !write)
2787 		return r;
2788 
2789 	new_pid = find_get_pid(tmp);
2790 	if (!new_pid)
2791 		return -ESRCH;
2792 
2793 	put_pid(xchg(&cad_pid, new_pid));
2794 	return 0;
2795 }
2796 
2797 #else /* CONFIG_PROC_FS */
2798 
2799 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2800 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2801 {
2802 	return -ENOSYS;
2803 }
2804 
2805 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2806 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2807 {
2808 	return -ENOSYS;
2809 }
2810 
2811 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2812 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2813 {
2814 	return -ENOSYS;
2815 }
2816 
2817 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2818 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2819 {
2820 	return -ENOSYS;
2821 }
2822 
2823 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2824 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2825 {
2826 	return -ENOSYS;
2827 }
2828 
2829 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2830 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2831 {
2832 	return -ENOSYS;
2833 }
2834 
2835 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2836 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2837 {
2838 	return -ENOSYS;
2839 }
2840 
2841 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2842 				      struct file *filp,
2843 				      void __user *buffer,
2844 				      size_t *lenp, loff_t *ppos)
2845 {
2846     return -ENOSYS;
2847 }
2848 
2849 
2850 #endif /* CONFIG_PROC_FS */
2851 
2852 
2853 #ifdef CONFIG_SYSCTL_SYSCALL
2854 /*
2855  * General sysctl support routines
2856  */
2857 
2858 /* The generic sysctl data routine (used if no strategy routine supplied) */
2859 int sysctl_data(struct ctl_table *table,
2860 		void __user *oldval, size_t __user *oldlenp,
2861 		void __user *newval, size_t newlen)
2862 {
2863 	size_t len;
2864 
2865 	/* Get out of I don't have a variable */
2866 	if (!table->data || !table->maxlen)
2867 		return -ENOTDIR;
2868 
2869 	if (oldval && oldlenp) {
2870 		if (get_user(len, oldlenp))
2871 			return -EFAULT;
2872 		if (len) {
2873 			if (len > table->maxlen)
2874 				len = table->maxlen;
2875 			if (copy_to_user(oldval, table->data, len))
2876 				return -EFAULT;
2877 			if (put_user(len, oldlenp))
2878 				return -EFAULT;
2879 		}
2880 	}
2881 
2882 	if (newval && newlen) {
2883 		if (newlen > table->maxlen)
2884 			newlen = table->maxlen;
2885 
2886 		if (copy_from_user(table->data, newval, newlen))
2887 			return -EFAULT;
2888 	}
2889 	return 1;
2890 }
2891 
2892 /* The generic string strategy routine: */
2893 int sysctl_string(struct ctl_table *table,
2894 		  void __user *oldval, size_t __user *oldlenp,
2895 		  void __user *newval, size_t newlen)
2896 {
2897 	if (!table->data || !table->maxlen)
2898 		return -ENOTDIR;
2899 
2900 	if (oldval && oldlenp) {
2901 		size_t bufsize;
2902 		if (get_user(bufsize, oldlenp))
2903 			return -EFAULT;
2904 		if (bufsize) {
2905 			size_t len = strlen(table->data), copied;
2906 
2907 			/* This shouldn't trigger for a well-formed sysctl */
2908 			if (len > table->maxlen)
2909 				len = table->maxlen;
2910 
2911 			/* Copy up to a max of bufsize-1 bytes of the string */
2912 			copied = (len >= bufsize) ? bufsize - 1 : len;
2913 
2914 			if (copy_to_user(oldval, table->data, copied) ||
2915 			    put_user(0, (char __user *)(oldval + copied)))
2916 				return -EFAULT;
2917 			if (put_user(len, oldlenp))
2918 				return -EFAULT;
2919 		}
2920 	}
2921 	if (newval && newlen) {
2922 		size_t len = newlen;
2923 		if (len > table->maxlen)
2924 			len = table->maxlen;
2925 		if(copy_from_user(table->data, newval, len))
2926 			return -EFAULT;
2927 		if (len == table->maxlen)
2928 			len--;
2929 		((char *) table->data)[len] = 0;
2930 	}
2931 	return 1;
2932 }
2933 
2934 /*
2935  * This function makes sure that all of the integers in the vector
2936  * are between the minimum and maximum values given in the arrays
2937  * table->extra1 and table->extra2, respectively.
2938  */
2939 int sysctl_intvec(struct ctl_table *table,
2940 		void __user *oldval, size_t __user *oldlenp,
2941 		void __user *newval, size_t newlen)
2942 {
2943 
2944 	if (newval && newlen) {
2945 		int __user *vec = (int __user *) newval;
2946 		int *min = (int *) table->extra1;
2947 		int *max = (int *) table->extra2;
2948 		size_t length;
2949 		int i;
2950 
2951 		if (newlen % sizeof(int) != 0)
2952 			return -EINVAL;
2953 
2954 		if (!table->extra1 && !table->extra2)
2955 			return 0;
2956 
2957 		if (newlen > table->maxlen)
2958 			newlen = table->maxlen;
2959 		length = newlen / sizeof(int);
2960 
2961 		for (i = 0; i < length; i++) {
2962 			int value;
2963 			if (get_user(value, vec + i))
2964 				return -EFAULT;
2965 			if (min && value < min[i])
2966 				return -EINVAL;
2967 			if (max && value > max[i])
2968 				return -EINVAL;
2969 		}
2970 	}
2971 	return 0;
2972 }
2973 
2974 /* Strategy function to convert jiffies to seconds */
2975 int sysctl_jiffies(struct ctl_table *table,
2976 		void __user *oldval, size_t __user *oldlenp,
2977 		void __user *newval, size_t newlen)
2978 {
2979 	if (oldval && oldlenp) {
2980 		size_t olen;
2981 
2982 		if (get_user(olen, oldlenp))
2983 			return -EFAULT;
2984 		if (olen) {
2985 			int val;
2986 
2987 			if (olen < sizeof(int))
2988 				return -EINVAL;
2989 
2990 			val = *(int *)(table->data) / HZ;
2991 			if (put_user(val, (int __user *)oldval))
2992 				return -EFAULT;
2993 			if (put_user(sizeof(int), oldlenp))
2994 				return -EFAULT;
2995 		}
2996 	}
2997 	if (newval && newlen) {
2998 		int new;
2999 		if (newlen != sizeof(int))
3000 			return -EINVAL;
3001 		if (get_user(new, (int __user *)newval))
3002 			return -EFAULT;
3003 		*(int *)(table->data) = new*HZ;
3004 	}
3005 	return 1;
3006 }
3007 
3008 /* Strategy function to convert jiffies to seconds */
3009 int sysctl_ms_jiffies(struct ctl_table *table,
3010 		void __user *oldval, size_t __user *oldlenp,
3011 		void __user *newval, size_t newlen)
3012 {
3013 	if (oldval && oldlenp) {
3014 		size_t olen;
3015 
3016 		if (get_user(olen, oldlenp))
3017 			return -EFAULT;
3018 		if (olen) {
3019 			int val;
3020 
3021 			if (olen < sizeof(int))
3022 				return -EINVAL;
3023 
3024 			val = jiffies_to_msecs(*(int *)(table->data));
3025 			if (put_user(val, (int __user *)oldval))
3026 				return -EFAULT;
3027 			if (put_user(sizeof(int), oldlenp))
3028 				return -EFAULT;
3029 		}
3030 	}
3031 	if (newval && newlen) {
3032 		int new;
3033 		if (newlen != sizeof(int))
3034 			return -EINVAL;
3035 		if (get_user(new, (int __user *)newval))
3036 			return -EFAULT;
3037 		*(int *)(table->data) = msecs_to_jiffies(new);
3038 	}
3039 	return 1;
3040 }
3041 
3042 
3043 
3044 #else /* CONFIG_SYSCTL_SYSCALL */
3045 
3046 
3047 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
3048 {
3049 	struct __sysctl_args tmp;
3050 	int error;
3051 
3052 	if (copy_from_user(&tmp, args, sizeof(tmp)))
3053 		return -EFAULT;
3054 
3055 	error = deprecated_sysctl_warning(&tmp);
3056 
3057 	/* If no error reading the parameters then just -ENOSYS ... */
3058 	if (!error)
3059 		error = -ENOSYS;
3060 
3061 	return error;
3062 }
3063 
3064 int sysctl_data(struct ctl_table *table,
3065 		  void __user *oldval, size_t __user *oldlenp,
3066 		  void __user *newval, size_t newlen)
3067 {
3068 	return -ENOSYS;
3069 }
3070 
3071 int sysctl_string(struct ctl_table *table,
3072 		  void __user *oldval, size_t __user *oldlenp,
3073 		  void __user *newval, size_t newlen)
3074 {
3075 	return -ENOSYS;
3076 }
3077 
3078 int sysctl_intvec(struct ctl_table *table,
3079 		void __user *oldval, size_t __user *oldlenp,
3080 		void __user *newval, size_t newlen)
3081 {
3082 	return -ENOSYS;
3083 }
3084 
3085 int sysctl_jiffies(struct ctl_table *table,
3086 		void __user *oldval, size_t __user *oldlenp,
3087 		void __user *newval, size_t newlen)
3088 {
3089 	return -ENOSYS;
3090 }
3091 
3092 int sysctl_ms_jiffies(struct ctl_table *table,
3093 		void __user *oldval, size_t __user *oldlenp,
3094 		void __user *newval, size_t newlen)
3095 {
3096 	return -ENOSYS;
3097 }
3098 
3099 #endif /* CONFIG_SYSCTL_SYSCALL */
3100 
3101 static int deprecated_sysctl_warning(struct __sysctl_args *args)
3102 {
3103 	static int msg_count;
3104 	int name[CTL_MAXNAME];
3105 	int i;
3106 
3107 	/* Check args->nlen. */
3108 	if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
3109 		return -ENOTDIR;
3110 
3111 	/* Read in the sysctl name for better debug message logging */
3112 	for (i = 0; i < args->nlen; i++)
3113 		if (get_user(name[i], args->name + i))
3114 			return -EFAULT;
3115 
3116 	/* Ignore accesses to kernel.version */
3117 	if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
3118 		return 0;
3119 
3120 	if (msg_count < 5) {
3121 		msg_count++;
3122 		printk(KERN_INFO
3123 			"warning: process `%s' used the deprecated sysctl "
3124 			"system call with ", current->comm);
3125 		for (i = 0; i < args->nlen; i++)
3126 			printk("%d.", name[i]);
3127 		printk("\n");
3128 	}
3129 	return 0;
3130 }
3131 
3132 /*
3133  * No sense putting this after each symbol definition, twice,
3134  * exception granted :-)
3135  */
3136 EXPORT_SYMBOL(proc_dointvec);
3137 EXPORT_SYMBOL(proc_dointvec_jiffies);
3138 EXPORT_SYMBOL(proc_dointvec_minmax);
3139 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3140 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3141 EXPORT_SYMBOL(proc_dostring);
3142 EXPORT_SYMBOL(proc_doulongvec_minmax);
3143 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3144 EXPORT_SYMBOL(register_sysctl_table);
3145 EXPORT_SYMBOL(register_sysctl_paths);
3146 EXPORT_SYMBOL(sysctl_intvec);
3147 EXPORT_SYMBOL(sysctl_jiffies);
3148 EXPORT_SYMBOL(sysctl_ms_jiffies);
3149 EXPORT_SYMBOL(sysctl_string);
3150 EXPORT_SYMBOL(sysctl_data);
3151 EXPORT_SYMBOL(unregister_sysctl_table);
3152