xref: /linux/kernel/sysctl.c (revision 44eeab67416711db9b84610ef18c99a60415dff8)
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/kmemcheck.h>
31 #include <linux/smp_lock.h>
32 #include <linux/fs.h>
33 #include <linux/init.h>
34 #include <linux/kernel.h>
35 #include <linux/kobject.h>
36 #include <linux/net.h>
37 #include <linux/sysrq.h>
38 #include <linux/highuid.h>
39 #include <linux/writeback.h>
40 #include <linux/hugetlb.h>
41 #include <linux/initrd.h>
42 #include <linux/key.h>
43 #include <linux/times.h>
44 #include <linux/limits.h>
45 #include <linux/dcache.h>
46 #include <linux/syscalls.h>
47 #include <linux/vmstat.h>
48 #include <linux/nfs_fs.h>
49 #include <linux/acpi.h>
50 #include <linux/reboot.h>
51 #include <linux/ftrace.h>
52 #include <linux/slow-work.h>
53 #include <linux/perf_counter.h>
54 
55 #include <asm/uaccess.h>
56 #include <asm/processor.h>
57 
58 #ifdef CONFIG_X86
59 #include <asm/nmi.h>
60 #include <asm/stacktrace.h>
61 #include <asm/io.h>
62 #endif
63 
64 static int deprecated_sysctl_warning(struct __sysctl_args *args);
65 
66 #if defined(CONFIG_SYSCTL)
67 
68 /* External variables not in a header file. */
69 extern int C_A_D;
70 extern int print_fatal_signals;
71 extern int sysctl_overcommit_memory;
72 extern int sysctl_overcommit_ratio;
73 extern int sysctl_panic_on_oom;
74 extern int sysctl_oom_kill_allocating_task;
75 extern int sysctl_oom_dump_tasks;
76 extern int max_threads;
77 extern int core_uses_pid;
78 extern int suid_dumpable;
79 extern char core_pattern[];
80 extern int pid_max;
81 extern int min_free_kbytes;
82 extern int pid_max_min, pid_max_max;
83 extern int sysctl_drop_caches;
84 extern int percpu_pagelist_fraction;
85 extern int compat_log;
86 extern int latencytop_enabled;
87 extern int sysctl_nr_open_min, sysctl_nr_open_max;
88 #ifndef CONFIG_MMU
89 extern int sysctl_nr_trim_pages;
90 #endif
91 #ifdef CONFIG_RCU_TORTURE_TEST
92 extern int rcutorture_runnable;
93 #endif /* #ifdef CONFIG_RCU_TORTURE_TEST */
94 
95 /* Constants used for minimum and  maximum */
96 #ifdef CONFIG_DETECT_SOFTLOCKUP
97 static int sixty = 60;
98 static int neg_one = -1;
99 #endif
100 
101 static int zero;
102 static int __maybe_unused one = 1;
103 static int __maybe_unused two = 2;
104 static unsigned long one_ul = 1;
105 static int one_hundred = 100;
106 
107 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
108 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
109 
110 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
111 static int maxolduid = 65535;
112 static int minolduid;
113 static int min_percpu_pagelist_fract = 8;
114 
115 static int ngroups_max = NGROUPS_MAX;
116 
117 #ifdef CONFIG_MODULES
118 extern char modprobe_path[];
119 extern int modules_disabled;
120 #endif
121 #ifdef CONFIG_CHR_DEV_SG
122 extern int sg_big_buff;
123 #endif
124 
125 #ifdef CONFIG_SPARC
126 #include <asm/system.h>
127 #endif
128 
129 #ifdef CONFIG_SPARC64
130 extern int sysctl_tsb_ratio;
131 #endif
132 
133 #ifdef __hppa__
134 extern int pwrsw_enabled;
135 extern int unaligned_enabled;
136 #endif
137 
138 #ifdef CONFIG_S390
139 #ifdef CONFIG_MATHEMU
140 extern int sysctl_ieee_emulation_warnings;
141 #endif
142 extern int sysctl_userprocess_debug;
143 extern int spin_retry;
144 #endif
145 
146 #ifdef CONFIG_BSD_PROCESS_ACCT
147 extern int acct_parm[];
148 #endif
149 
150 #ifdef CONFIG_IA64
151 extern int no_unaligned_warning;
152 extern int unaligned_dump_stack;
153 #endif
154 
155 #ifdef CONFIG_RT_MUTEXES
156 extern int max_lock_depth;
157 #endif
158 
159 #ifdef CONFIG_PROC_SYSCTL
160 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
161 		  void __user *buffer, size_t *lenp, loff_t *ppos);
162 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
163 			       void __user *buffer, size_t *lenp, loff_t *ppos);
164 #endif
165 
166 static struct ctl_table root_table[];
167 static struct ctl_table_root sysctl_table_root;
168 static struct ctl_table_header root_table_header = {
169 	.count = 1,
170 	.ctl_table = root_table,
171 	.ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
172 	.root = &sysctl_table_root,
173 	.set = &sysctl_table_root.default_set,
174 };
175 static struct ctl_table_root sysctl_table_root = {
176 	.root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
177 	.default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
178 };
179 
180 static struct ctl_table kern_table[];
181 static struct ctl_table vm_table[];
182 static struct ctl_table fs_table[];
183 static struct ctl_table debug_table[];
184 static struct ctl_table dev_table[];
185 extern struct ctl_table random_table[];
186 #ifdef CONFIG_INOTIFY_USER
187 extern struct ctl_table inotify_table[];
188 #endif
189 #ifdef CONFIG_EPOLL
190 extern struct ctl_table epoll_table[];
191 #endif
192 
193 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
194 int sysctl_legacy_va_layout;
195 #endif
196 
197 extern int prove_locking;
198 extern int lock_stat;
199 
200 /* The default sysctl tables: */
201 
202 static struct ctl_table root_table[] = {
203 	{
204 		.ctl_name	= CTL_KERN,
205 		.procname	= "kernel",
206 		.mode		= 0555,
207 		.child		= kern_table,
208 	},
209 	{
210 		.ctl_name	= CTL_VM,
211 		.procname	= "vm",
212 		.mode		= 0555,
213 		.child		= vm_table,
214 	},
215 	{
216 		.ctl_name	= CTL_FS,
217 		.procname	= "fs",
218 		.mode		= 0555,
219 		.child		= fs_table,
220 	},
221 	{
222 		.ctl_name	= CTL_DEBUG,
223 		.procname	= "debug",
224 		.mode		= 0555,
225 		.child		= debug_table,
226 	},
227 	{
228 		.ctl_name	= CTL_DEV,
229 		.procname	= "dev",
230 		.mode		= 0555,
231 		.child		= dev_table,
232 	},
233 /*
234  * NOTE: do not add new entries to this table unless you have read
235  * Documentation/sysctl/ctl_unnumbered.txt
236  */
237 	{ .ctl_name = 0 }
238 };
239 
240 #ifdef CONFIG_SCHED_DEBUG
241 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
242 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
243 static int min_wakeup_granularity_ns;			/* 0 usecs */
244 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
245 #endif
246 
247 static struct ctl_table kern_table[] = {
248 #ifdef CONFIG_SCHED_DEBUG
249 	{
250 		.ctl_name	= CTL_UNNUMBERED,
251 		.procname	= "sched_min_granularity_ns",
252 		.data		= &sysctl_sched_min_granularity,
253 		.maxlen		= sizeof(unsigned int),
254 		.mode		= 0644,
255 		.proc_handler	= &sched_nr_latency_handler,
256 		.strategy	= &sysctl_intvec,
257 		.extra1		= &min_sched_granularity_ns,
258 		.extra2		= &max_sched_granularity_ns,
259 	},
260 	{
261 		.ctl_name	= CTL_UNNUMBERED,
262 		.procname	= "sched_latency_ns",
263 		.data		= &sysctl_sched_latency,
264 		.maxlen		= sizeof(unsigned int),
265 		.mode		= 0644,
266 		.proc_handler	= &sched_nr_latency_handler,
267 		.strategy	= &sysctl_intvec,
268 		.extra1		= &min_sched_granularity_ns,
269 		.extra2		= &max_sched_granularity_ns,
270 	},
271 	{
272 		.ctl_name	= CTL_UNNUMBERED,
273 		.procname	= "sched_wakeup_granularity_ns",
274 		.data		= &sysctl_sched_wakeup_granularity,
275 		.maxlen		= sizeof(unsigned int),
276 		.mode		= 0644,
277 		.proc_handler	= &proc_dointvec_minmax,
278 		.strategy	= &sysctl_intvec,
279 		.extra1		= &min_wakeup_granularity_ns,
280 		.extra2		= &max_wakeup_granularity_ns,
281 	},
282 	{
283 		.ctl_name	= CTL_UNNUMBERED,
284 		.procname	= "sched_shares_ratelimit",
285 		.data		= &sysctl_sched_shares_ratelimit,
286 		.maxlen		= sizeof(unsigned int),
287 		.mode		= 0644,
288 		.proc_handler	= &proc_dointvec,
289 	},
290 	{
291 		.ctl_name	= CTL_UNNUMBERED,
292 		.procname	= "sched_shares_thresh",
293 		.data		= &sysctl_sched_shares_thresh,
294 		.maxlen		= sizeof(unsigned int),
295 		.mode		= 0644,
296 		.proc_handler	= &proc_dointvec_minmax,
297 		.strategy	= &sysctl_intvec,
298 		.extra1		= &zero,
299 	},
300 	{
301 		.ctl_name	= CTL_UNNUMBERED,
302 		.procname	= "sched_child_runs_first",
303 		.data		= &sysctl_sched_child_runs_first,
304 		.maxlen		= sizeof(unsigned int),
305 		.mode		= 0644,
306 		.proc_handler	= &proc_dointvec,
307 	},
308 	{
309 		.ctl_name	= CTL_UNNUMBERED,
310 		.procname	= "sched_features",
311 		.data		= &sysctl_sched_features,
312 		.maxlen		= sizeof(unsigned int),
313 		.mode		= 0644,
314 		.proc_handler	= &proc_dointvec,
315 	},
316 	{
317 		.ctl_name	= CTL_UNNUMBERED,
318 		.procname	= "sched_migration_cost",
319 		.data		= &sysctl_sched_migration_cost,
320 		.maxlen		= sizeof(unsigned int),
321 		.mode		= 0644,
322 		.proc_handler	= &proc_dointvec,
323 	},
324 	{
325 		.ctl_name	= CTL_UNNUMBERED,
326 		.procname	= "sched_nr_migrate",
327 		.data		= &sysctl_sched_nr_migrate,
328 		.maxlen		= sizeof(unsigned int),
329 		.mode		= 0644,
330 		.proc_handler	= &proc_dointvec,
331 	},
332 	{
333 		.ctl_name	= CTL_UNNUMBERED,
334 		.procname	= "timer_migration",
335 		.data		= &sysctl_timer_migration,
336 		.maxlen		= sizeof(unsigned int),
337 		.mode		= 0644,
338 		.proc_handler	= &proc_dointvec,
339 	},
340 #endif
341 	{
342 		.ctl_name	= CTL_UNNUMBERED,
343 		.procname	= "sched_rt_period_us",
344 		.data		= &sysctl_sched_rt_period,
345 		.maxlen		= sizeof(unsigned int),
346 		.mode		= 0644,
347 		.proc_handler	= &sched_rt_handler,
348 	},
349 	{
350 		.ctl_name	= CTL_UNNUMBERED,
351 		.procname	= "sched_rt_runtime_us",
352 		.data		= &sysctl_sched_rt_runtime,
353 		.maxlen		= sizeof(int),
354 		.mode		= 0644,
355 		.proc_handler	= &sched_rt_handler,
356 	},
357 	{
358 		.ctl_name	= CTL_UNNUMBERED,
359 		.procname	= "sched_compat_yield",
360 		.data		= &sysctl_sched_compat_yield,
361 		.maxlen		= sizeof(unsigned int),
362 		.mode		= 0644,
363 		.proc_handler	= &proc_dointvec,
364 	},
365 #ifdef CONFIG_PROVE_LOCKING
366 	{
367 		.ctl_name	= CTL_UNNUMBERED,
368 		.procname	= "prove_locking",
369 		.data		= &prove_locking,
370 		.maxlen		= sizeof(int),
371 		.mode		= 0644,
372 		.proc_handler	= &proc_dointvec,
373 	},
374 #endif
375 #ifdef CONFIG_LOCK_STAT
376 	{
377 		.ctl_name	= CTL_UNNUMBERED,
378 		.procname	= "lock_stat",
379 		.data		= &lock_stat,
380 		.maxlen		= sizeof(int),
381 		.mode		= 0644,
382 		.proc_handler	= &proc_dointvec,
383 	},
384 #endif
385 	{
386 		.ctl_name	= KERN_PANIC,
387 		.procname	= "panic",
388 		.data		= &panic_timeout,
389 		.maxlen		= sizeof(int),
390 		.mode		= 0644,
391 		.proc_handler	= &proc_dointvec,
392 	},
393 	{
394 		.ctl_name	= KERN_CORE_USES_PID,
395 		.procname	= "core_uses_pid",
396 		.data		= &core_uses_pid,
397 		.maxlen		= sizeof(int),
398 		.mode		= 0644,
399 		.proc_handler	= &proc_dointvec,
400 	},
401 	{
402 		.ctl_name	= KERN_CORE_PATTERN,
403 		.procname	= "core_pattern",
404 		.data		= core_pattern,
405 		.maxlen		= CORENAME_MAX_SIZE,
406 		.mode		= 0644,
407 		.proc_handler	= &proc_dostring,
408 		.strategy	= &sysctl_string,
409 	},
410 #ifdef CONFIG_PROC_SYSCTL
411 	{
412 		.procname	= "tainted",
413 		.maxlen 	= sizeof(long),
414 		.mode		= 0644,
415 		.proc_handler	= &proc_taint,
416 	},
417 #endif
418 #ifdef CONFIG_LATENCYTOP
419 	{
420 		.procname	= "latencytop",
421 		.data		= &latencytop_enabled,
422 		.maxlen		= sizeof(int),
423 		.mode		= 0644,
424 		.proc_handler	= &proc_dointvec,
425 	},
426 #endif
427 #ifdef CONFIG_BLK_DEV_INITRD
428 	{
429 		.ctl_name	= KERN_REALROOTDEV,
430 		.procname	= "real-root-dev",
431 		.data		= &real_root_dev,
432 		.maxlen		= sizeof(int),
433 		.mode		= 0644,
434 		.proc_handler	= &proc_dointvec,
435 	},
436 #endif
437 	{
438 		.ctl_name	= CTL_UNNUMBERED,
439 		.procname	= "print-fatal-signals",
440 		.data		= &print_fatal_signals,
441 		.maxlen		= sizeof(int),
442 		.mode		= 0644,
443 		.proc_handler	= &proc_dointvec,
444 	},
445 #ifdef CONFIG_SPARC
446 	{
447 		.ctl_name	= KERN_SPARC_REBOOT,
448 		.procname	= "reboot-cmd",
449 		.data		= reboot_command,
450 		.maxlen		= 256,
451 		.mode		= 0644,
452 		.proc_handler	= &proc_dostring,
453 		.strategy	= &sysctl_string,
454 	},
455 	{
456 		.ctl_name	= KERN_SPARC_STOP_A,
457 		.procname	= "stop-a",
458 		.data		= &stop_a_enabled,
459 		.maxlen		= sizeof (int),
460 		.mode		= 0644,
461 		.proc_handler	= &proc_dointvec,
462 	},
463 	{
464 		.ctl_name	= KERN_SPARC_SCONS_PWROFF,
465 		.procname	= "scons-poweroff",
466 		.data		= &scons_pwroff,
467 		.maxlen		= sizeof (int),
468 		.mode		= 0644,
469 		.proc_handler	= &proc_dointvec,
470 	},
471 #endif
472 #ifdef CONFIG_SPARC64
473 	{
474 		.ctl_name	= CTL_UNNUMBERED,
475 		.procname	= "tsb-ratio",
476 		.data		= &sysctl_tsb_ratio,
477 		.maxlen		= sizeof (int),
478 		.mode		= 0644,
479 		.proc_handler	= &proc_dointvec,
480 	},
481 #endif
482 #ifdef __hppa__
483 	{
484 		.ctl_name	= KERN_HPPA_PWRSW,
485 		.procname	= "soft-power",
486 		.data		= &pwrsw_enabled,
487 		.maxlen		= sizeof (int),
488 	 	.mode		= 0644,
489 		.proc_handler	= &proc_dointvec,
490 	},
491 	{
492 		.ctl_name	= KERN_HPPA_UNALIGNED,
493 		.procname	= "unaligned-trap",
494 		.data		= &unaligned_enabled,
495 		.maxlen		= sizeof (int),
496 		.mode		= 0644,
497 		.proc_handler	= &proc_dointvec,
498 	},
499 #endif
500 	{
501 		.ctl_name	= KERN_CTLALTDEL,
502 		.procname	= "ctrl-alt-del",
503 		.data		= &C_A_D,
504 		.maxlen		= sizeof(int),
505 		.mode		= 0644,
506 		.proc_handler	= &proc_dointvec,
507 	},
508 #ifdef CONFIG_FUNCTION_TRACER
509 	{
510 		.ctl_name	= CTL_UNNUMBERED,
511 		.procname	= "ftrace_enabled",
512 		.data		= &ftrace_enabled,
513 		.maxlen		= sizeof(int),
514 		.mode		= 0644,
515 		.proc_handler	= &ftrace_enable_sysctl,
516 	},
517 #endif
518 #ifdef CONFIG_STACK_TRACER
519 	{
520 		.ctl_name	= CTL_UNNUMBERED,
521 		.procname	= "stack_tracer_enabled",
522 		.data		= &stack_tracer_enabled,
523 		.maxlen		= sizeof(int),
524 		.mode		= 0644,
525 		.proc_handler	= &stack_trace_sysctl,
526 	},
527 #endif
528 #ifdef CONFIG_TRACING
529 	{
530 		.ctl_name	= CTL_UNNUMBERED,
531 		.procname	= "ftrace_dump_on_oops",
532 		.data		= &ftrace_dump_on_oops,
533 		.maxlen		= sizeof(int),
534 		.mode		= 0644,
535 		.proc_handler	= &proc_dointvec,
536 	},
537 #endif
538 #ifdef CONFIG_MODULES
539 	{
540 		.ctl_name	= KERN_MODPROBE,
541 		.procname	= "modprobe",
542 		.data		= &modprobe_path,
543 		.maxlen		= KMOD_PATH_LEN,
544 		.mode		= 0644,
545 		.proc_handler	= &proc_dostring,
546 		.strategy	= &sysctl_string,
547 	},
548 	{
549 		.ctl_name	= CTL_UNNUMBERED,
550 		.procname	= "modules_disabled",
551 		.data		= &modules_disabled,
552 		.maxlen		= sizeof(int),
553 		.mode		= 0644,
554 		/* only handle a transition from default "0" to "1" */
555 		.proc_handler	= &proc_dointvec_minmax,
556 		.extra1		= &one,
557 		.extra2		= &one,
558 	},
559 #endif
560 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
561 	{
562 		.ctl_name	= KERN_HOTPLUG,
563 		.procname	= "hotplug",
564 		.data		= &uevent_helper,
565 		.maxlen		= UEVENT_HELPER_PATH_LEN,
566 		.mode		= 0644,
567 		.proc_handler	= &proc_dostring,
568 		.strategy	= &sysctl_string,
569 	},
570 #endif
571 #ifdef CONFIG_CHR_DEV_SG
572 	{
573 		.ctl_name	= KERN_SG_BIG_BUFF,
574 		.procname	= "sg-big-buff",
575 		.data		= &sg_big_buff,
576 		.maxlen		= sizeof (int),
577 		.mode		= 0444,
578 		.proc_handler	= &proc_dointvec,
579 	},
580 #endif
581 #ifdef CONFIG_BSD_PROCESS_ACCT
582 	{
583 		.ctl_name	= KERN_ACCT,
584 		.procname	= "acct",
585 		.data		= &acct_parm,
586 		.maxlen		= 3*sizeof(int),
587 		.mode		= 0644,
588 		.proc_handler	= &proc_dointvec,
589 	},
590 #endif
591 #ifdef CONFIG_MAGIC_SYSRQ
592 	{
593 		.ctl_name	= KERN_SYSRQ,
594 		.procname	= "sysrq",
595 		.data		= &__sysrq_enabled,
596 		.maxlen		= sizeof (int),
597 		.mode		= 0644,
598 		.proc_handler	= &proc_dointvec,
599 	},
600 #endif
601 #ifdef CONFIG_PROC_SYSCTL
602 	{
603 		.procname	= "cad_pid",
604 		.data		= NULL,
605 		.maxlen		= sizeof (int),
606 		.mode		= 0600,
607 		.proc_handler	= &proc_do_cad_pid,
608 	},
609 #endif
610 	{
611 		.ctl_name	= KERN_MAX_THREADS,
612 		.procname	= "threads-max",
613 		.data		= &max_threads,
614 		.maxlen		= sizeof(int),
615 		.mode		= 0644,
616 		.proc_handler	= &proc_dointvec,
617 	},
618 	{
619 		.ctl_name	= KERN_RANDOM,
620 		.procname	= "random",
621 		.mode		= 0555,
622 		.child		= random_table,
623 	},
624 	{
625 		.ctl_name	= KERN_OVERFLOWUID,
626 		.procname	= "overflowuid",
627 		.data		= &overflowuid,
628 		.maxlen		= sizeof(int),
629 		.mode		= 0644,
630 		.proc_handler	= &proc_dointvec_minmax,
631 		.strategy	= &sysctl_intvec,
632 		.extra1		= &minolduid,
633 		.extra2		= &maxolduid,
634 	},
635 	{
636 		.ctl_name	= KERN_OVERFLOWGID,
637 		.procname	= "overflowgid",
638 		.data		= &overflowgid,
639 		.maxlen		= sizeof(int),
640 		.mode		= 0644,
641 		.proc_handler	= &proc_dointvec_minmax,
642 		.strategy	= &sysctl_intvec,
643 		.extra1		= &minolduid,
644 		.extra2		= &maxolduid,
645 	},
646 #ifdef CONFIG_S390
647 #ifdef CONFIG_MATHEMU
648 	{
649 		.ctl_name	= KERN_IEEE_EMULATION_WARNINGS,
650 		.procname	= "ieee_emulation_warnings",
651 		.data		= &sysctl_ieee_emulation_warnings,
652 		.maxlen		= sizeof(int),
653 		.mode		= 0644,
654 		.proc_handler	= &proc_dointvec,
655 	},
656 #endif
657 	{
658 		.ctl_name	= KERN_S390_USER_DEBUG_LOGGING,
659 		.procname	= "userprocess_debug",
660 		.data		= &sysctl_userprocess_debug,
661 		.maxlen		= sizeof(int),
662 		.mode		= 0644,
663 		.proc_handler	= &proc_dointvec,
664 	},
665 #endif
666 	{
667 		.ctl_name	= KERN_PIDMAX,
668 		.procname	= "pid_max",
669 		.data		= &pid_max,
670 		.maxlen		= sizeof (int),
671 		.mode		= 0644,
672 		.proc_handler	= &proc_dointvec_minmax,
673 		.strategy	= sysctl_intvec,
674 		.extra1		= &pid_max_min,
675 		.extra2		= &pid_max_max,
676 	},
677 	{
678 		.ctl_name	= KERN_PANIC_ON_OOPS,
679 		.procname	= "panic_on_oops",
680 		.data		= &panic_on_oops,
681 		.maxlen		= sizeof(int),
682 		.mode		= 0644,
683 		.proc_handler	= &proc_dointvec,
684 	},
685 #if defined CONFIG_PRINTK
686 	{
687 		.ctl_name	= KERN_PRINTK,
688 		.procname	= "printk",
689 		.data		= &console_loglevel,
690 		.maxlen		= 4*sizeof(int),
691 		.mode		= 0644,
692 		.proc_handler	= &proc_dointvec,
693 	},
694 	{
695 		.ctl_name	= KERN_PRINTK_RATELIMIT,
696 		.procname	= "printk_ratelimit",
697 		.data		= &printk_ratelimit_state.interval,
698 		.maxlen		= sizeof(int),
699 		.mode		= 0644,
700 		.proc_handler	= &proc_dointvec_jiffies,
701 		.strategy	= &sysctl_jiffies,
702 	},
703 	{
704 		.ctl_name	= KERN_PRINTK_RATELIMIT_BURST,
705 		.procname	= "printk_ratelimit_burst",
706 		.data		= &printk_ratelimit_state.burst,
707 		.maxlen		= sizeof(int),
708 		.mode		= 0644,
709 		.proc_handler	= &proc_dointvec,
710 	},
711 #endif
712 	{
713 		.ctl_name	= KERN_NGROUPS_MAX,
714 		.procname	= "ngroups_max",
715 		.data		= &ngroups_max,
716 		.maxlen		= sizeof (int),
717 		.mode		= 0444,
718 		.proc_handler	= &proc_dointvec,
719 	},
720 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
721 	{
722 		.ctl_name       = KERN_UNKNOWN_NMI_PANIC,
723 		.procname       = "unknown_nmi_panic",
724 		.data           = &unknown_nmi_panic,
725 		.maxlen         = sizeof (int),
726 		.mode           = 0644,
727 		.proc_handler   = &proc_dointvec,
728 	},
729 	{
730 		.procname       = "nmi_watchdog",
731 		.data           = &nmi_watchdog_enabled,
732 		.maxlen         = sizeof (int),
733 		.mode           = 0644,
734 		.proc_handler   = &proc_nmi_enabled,
735 	},
736 #endif
737 #if defined(CONFIG_X86)
738 	{
739 		.ctl_name	= KERN_PANIC_ON_NMI,
740 		.procname	= "panic_on_unrecovered_nmi",
741 		.data		= &panic_on_unrecovered_nmi,
742 		.maxlen		= sizeof(int),
743 		.mode		= 0644,
744 		.proc_handler	= &proc_dointvec,
745 	},
746 	{
747 		.ctl_name	= KERN_BOOTLOADER_TYPE,
748 		.procname	= "bootloader_type",
749 		.data		= &bootloader_type,
750 		.maxlen		= sizeof (int),
751 		.mode		= 0444,
752 		.proc_handler	= &proc_dointvec,
753 	},
754 	{
755 		.ctl_name	= CTL_UNNUMBERED,
756 		.procname	= "bootloader_version",
757 		.data		= &bootloader_version,
758 		.maxlen		= sizeof (int),
759 		.mode		= 0444,
760 		.proc_handler	= &proc_dointvec,
761 	},
762 	{
763 		.ctl_name	= CTL_UNNUMBERED,
764 		.procname	= "kstack_depth_to_print",
765 		.data		= &kstack_depth_to_print,
766 		.maxlen		= sizeof(int),
767 		.mode		= 0644,
768 		.proc_handler	= &proc_dointvec,
769 	},
770 	{
771 		.ctl_name	= CTL_UNNUMBERED,
772 		.procname	= "io_delay_type",
773 		.data		= &io_delay_type,
774 		.maxlen		= sizeof(int),
775 		.mode		= 0644,
776 		.proc_handler	= &proc_dointvec,
777 	},
778 #endif
779 #if defined(CONFIG_MMU)
780 	{
781 		.ctl_name	= KERN_RANDOMIZE,
782 		.procname	= "randomize_va_space",
783 		.data		= &randomize_va_space,
784 		.maxlen		= sizeof(int),
785 		.mode		= 0644,
786 		.proc_handler	= &proc_dointvec,
787 	},
788 #endif
789 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
790 	{
791 		.ctl_name	= KERN_SPIN_RETRY,
792 		.procname	= "spin_retry",
793 		.data		= &spin_retry,
794 		.maxlen		= sizeof (int),
795 		.mode		= 0644,
796 		.proc_handler	= &proc_dointvec,
797 	},
798 #endif
799 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
800 	{
801 		.procname	= "acpi_video_flags",
802 		.data		= &acpi_realmode_flags,
803 		.maxlen		= sizeof (unsigned long),
804 		.mode		= 0644,
805 		.proc_handler	= &proc_doulongvec_minmax,
806 	},
807 #endif
808 #ifdef CONFIG_IA64
809 	{
810 		.ctl_name	= KERN_IA64_UNALIGNED,
811 		.procname	= "ignore-unaligned-usertrap",
812 		.data		= &no_unaligned_warning,
813 		.maxlen		= sizeof (int),
814 	 	.mode		= 0644,
815 		.proc_handler	= &proc_dointvec,
816 	},
817 	{
818 		.ctl_name	= CTL_UNNUMBERED,
819 		.procname	= "unaligned-dump-stack",
820 		.data		= &unaligned_dump_stack,
821 		.maxlen		= sizeof (int),
822 		.mode		= 0644,
823 		.proc_handler	= &proc_dointvec,
824 	},
825 #endif
826 #ifdef CONFIG_DETECT_SOFTLOCKUP
827 	{
828 		.ctl_name	= CTL_UNNUMBERED,
829 		.procname	= "softlockup_panic",
830 		.data		= &softlockup_panic,
831 		.maxlen		= sizeof(int),
832 		.mode		= 0644,
833 		.proc_handler	= &proc_dointvec_minmax,
834 		.strategy	= &sysctl_intvec,
835 		.extra1		= &zero,
836 		.extra2		= &one,
837 	},
838 	{
839 		.ctl_name	= CTL_UNNUMBERED,
840 		.procname	= "softlockup_thresh",
841 		.data		= &softlockup_thresh,
842 		.maxlen		= sizeof(int),
843 		.mode		= 0644,
844 		.proc_handler	= &proc_dosoftlockup_thresh,
845 		.strategy	= &sysctl_intvec,
846 		.extra1		= &neg_one,
847 		.extra2		= &sixty,
848 	},
849 #endif
850 #ifdef CONFIG_DETECT_HUNG_TASK
851 	{
852 		.ctl_name	= CTL_UNNUMBERED,
853 		.procname	= "hung_task_panic",
854 		.data		= &sysctl_hung_task_panic,
855 		.maxlen		= sizeof(int),
856 		.mode		= 0644,
857 		.proc_handler	= &proc_dointvec_minmax,
858 		.strategy	= &sysctl_intvec,
859 		.extra1		= &zero,
860 		.extra2		= &one,
861 	},
862 	{
863 		.ctl_name	= CTL_UNNUMBERED,
864 		.procname	= "hung_task_check_count",
865 		.data		= &sysctl_hung_task_check_count,
866 		.maxlen		= sizeof(unsigned long),
867 		.mode		= 0644,
868 		.proc_handler	= &proc_doulongvec_minmax,
869 		.strategy	= &sysctl_intvec,
870 	},
871 	{
872 		.ctl_name	= CTL_UNNUMBERED,
873 		.procname	= "hung_task_timeout_secs",
874 		.data		= &sysctl_hung_task_timeout_secs,
875 		.maxlen		= sizeof(unsigned long),
876 		.mode		= 0644,
877 		.proc_handler	= &proc_dohung_task_timeout_secs,
878 		.strategy	= &sysctl_intvec,
879 	},
880 	{
881 		.ctl_name	= CTL_UNNUMBERED,
882 		.procname	= "hung_task_warnings",
883 		.data		= &sysctl_hung_task_warnings,
884 		.maxlen		= sizeof(unsigned long),
885 		.mode		= 0644,
886 		.proc_handler	= &proc_doulongvec_minmax,
887 		.strategy	= &sysctl_intvec,
888 	},
889 #endif
890 #ifdef CONFIG_COMPAT
891 	{
892 		.ctl_name	= KERN_COMPAT_LOG,
893 		.procname	= "compat-log",
894 		.data		= &compat_log,
895 		.maxlen		= sizeof (int),
896 	 	.mode		= 0644,
897 		.proc_handler	= &proc_dointvec,
898 	},
899 #endif
900 #ifdef CONFIG_RT_MUTEXES
901 	{
902 		.ctl_name	= KERN_MAX_LOCK_DEPTH,
903 		.procname	= "max_lock_depth",
904 		.data		= &max_lock_depth,
905 		.maxlen		= sizeof(int),
906 		.mode		= 0644,
907 		.proc_handler	= &proc_dointvec,
908 	},
909 #endif
910 	{
911 		.ctl_name	= CTL_UNNUMBERED,
912 		.procname	= "poweroff_cmd",
913 		.data		= &poweroff_cmd,
914 		.maxlen		= POWEROFF_CMD_PATH_LEN,
915 		.mode		= 0644,
916 		.proc_handler	= &proc_dostring,
917 		.strategy	= &sysctl_string,
918 	},
919 #ifdef CONFIG_KEYS
920 	{
921 		.ctl_name	= CTL_UNNUMBERED,
922 		.procname	= "keys",
923 		.mode		= 0555,
924 		.child		= key_sysctls,
925 	},
926 #endif
927 #ifdef CONFIG_RCU_TORTURE_TEST
928 	{
929 		.ctl_name       = CTL_UNNUMBERED,
930 		.procname       = "rcutorture_runnable",
931 		.data           = &rcutorture_runnable,
932 		.maxlen         = sizeof(int),
933 		.mode           = 0644,
934 		.proc_handler   = &proc_dointvec,
935 	},
936 #endif
937 #ifdef CONFIG_SLOW_WORK
938 	{
939 		.ctl_name	= CTL_UNNUMBERED,
940 		.procname	= "slow-work",
941 		.mode		= 0555,
942 		.child		= slow_work_sysctls,
943 	},
944 #endif
945 #ifdef CONFIG_PERF_COUNTERS
946 	{
947 		.ctl_name	= CTL_UNNUMBERED,
948 		.procname	= "perf_counter_paranoid",
949 		.data		= &sysctl_perf_counter_paranoid,
950 		.maxlen		= sizeof(sysctl_perf_counter_paranoid),
951 		.mode		= 0644,
952 		.proc_handler	= &proc_dointvec,
953 	},
954 	{
955 		.ctl_name	= CTL_UNNUMBERED,
956 		.procname	= "perf_counter_mlock_kb",
957 		.data		= &sysctl_perf_counter_mlock,
958 		.maxlen		= sizeof(sysctl_perf_counter_mlock),
959 		.mode		= 0644,
960 		.proc_handler	= &proc_dointvec,
961 	},
962 	{
963 		.ctl_name	= CTL_UNNUMBERED,
964 		.procname	= "perf_counter_max_sample_rate",
965 		.data		= &sysctl_perf_counter_sample_rate,
966 		.maxlen		= sizeof(sysctl_perf_counter_sample_rate),
967 		.mode		= 0644,
968 		.proc_handler	= &proc_dointvec,
969 	},
970 #endif
971 #ifdef CONFIG_KMEMCHECK
972 	{
973 		.ctl_name	= CTL_UNNUMBERED,
974 		.procname	= "kmemcheck",
975 		.data		= &kmemcheck_enabled,
976 		.maxlen		= sizeof(int),
977 		.mode		= 0644,
978 		.proc_handler	= &proc_dointvec,
979 	},
980 #endif
981 
982 /*
983  * NOTE: do not add new entries to this table unless you have read
984  * Documentation/sysctl/ctl_unnumbered.txt
985  */
986 	{ .ctl_name = 0 }
987 };
988 
989 static struct ctl_table vm_table[] = {
990 	{
991 		.ctl_name	= VM_OVERCOMMIT_MEMORY,
992 		.procname	= "overcommit_memory",
993 		.data		= &sysctl_overcommit_memory,
994 		.maxlen		= sizeof(sysctl_overcommit_memory),
995 		.mode		= 0644,
996 		.proc_handler	= &proc_dointvec,
997 	},
998 	{
999 		.ctl_name	= VM_PANIC_ON_OOM,
1000 		.procname	= "panic_on_oom",
1001 		.data		= &sysctl_panic_on_oom,
1002 		.maxlen		= sizeof(sysctl_panic_on_oom),
1003 		.mode		= 0644,
1004 		.proc_handler	= &proc_dointvec,
1005 	},
1006 	{
1007 		.ctl_name	= CTL_UNNUMBERED,
1008 		.procname	= "oom_kill_allocating_task",
1009 		.data		= &sysctl_oom_kill_allocating_task,
1010 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1011 		.mode		= 0644,
1012 		.proc_handler	= &proc_dointvec,
1013 	},
1014 	{
1015 		.ctl_name	= CTL_UNNUMBERED,
1016 		.procname	= "oom_dump_tasks",
1017 		.data		= &sysctl_oom_dump_tasks,
1018 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1019 		.mode		= 0644,
1020 		.proc_handler	= &proc_dointvec,
1021 	},
1022 	{
1023 		.ctl_name	= VM_OVERCOMMIT_RATIO,
1024 		.procname	= "overcommit_ratio",
1025 		.data		= &sysctl_overcommit_ratio,
1026 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1027 		.mode		= 0644,
1028 		.proc_handler	= &proc_dointvec,
1029 	},
1030 	{
1031 		.ctl_name	= VM_PAGE_CLUSTER,
1032 		.procname	= "page-cluster",
1033 		.data		= &page_cluster,
1034 		.maxlen		= sizeof(int),
1035 		.mode		= 0644,
1036 		.proc_handler	= &proc_dointvec,
1037 	},
1038 	{
1039 		.ctl_name	= VM_DIRTY_BACKGROUND,
1040 		.procname	= "dirty_background_ratio",
1041 		.data		= &dirty_background_ratio,
1042 		.maxlen		= sizeof(dirty_background_ratio),
1043 		.mode		= 0644,
1044 		.proc_handler	= &dirty_background_ratio_handler,
1045 		.strategy	= &sysctl_intvec,
1046 		.extra1		= &zero,
1047 		.extra2		= &one_hundred,
1048 	},
1049 	{
1050 		.ctl_name	= CTL_UNNUMBERED,
1051 		.procname	= "dirty_background_bytes",
1052 		.data		= &dirty_background_bytes,
1053 		.maxlen		= sizeof(dirty_background_bytes),
1054 		.mode		= 0644,
1055 		.proc_handler	= &dirty_background_bytes_handler,
1056 		.strategy	= &sysctl_intvec,
1057 		.extra1		= &one_ul,
1058 	},
1059 	{
1060 		.ctl_name	= VM_DIRTY_RATIO,
1061 		.procname	= "dirty_ratio",
1062 		.data		= &vm_dirty_ratio,
1063 		.maxlen		= sizeof(vm_dirty_ratio),
1064 		.mode		= 0644,
1065 		.proc_handler	= &dirty_ratio_handler,
1066 		.strategy	= &sysctl_intvec,
1067 		.extra1		= &zero,
1068 		.extra2		= &one_hundred,
1069 	},
1070 	{
1071 		.ctl_name	= CTL_UNNUMBERED,
1072 		.procname	= "dirty_bytes",
1073 		.data		= &vm_dirty_bytes,
1074 		.maxlen		= sizeof(vm_dirty_bytes),
1075 		.mode		= 0644,
1076 		.proc_handler	= &dirty_bytes_handler,
1077 		.strategy	= &sysctl_intvec,
1078 		.extra1		= &dirty_bytes_min,
1079 	},
1080 	{
1081 		.procname	= "dirty_writeback_centisecs",
1082 		.data		= &dirty_writeback_interval,
1083 		.maxlen		= sizeof(dirty_writeback_interval),
1084 		.mode		= 0644,
1085 		.proc_handler	= &dirty_writeback_centisecs_handler,
1086 	},
1087 	{
1088 		.procname	= "dirty_expire_centisecs",
1089 		.data		= &dirty_expire_interval,
1090 		.maxlen		= sizeof(dirty_expire_interval),
1091 		.mode		= 0644,
1092 		.proc_handler	= &proc_dointvec,
1093 	},
1094 	{
1095 		.ctl_name	= VM_NR_PDFLUSH_THREADS,
1096 		.procname	= "nr_pdflush_threads",
1097 		.data		= &nr_pdflush_threads,
1098 		.maxlen		= sizeof nr_pdflush_threads,
1099 		.mode		= 0444 /* read-only*/,
1100 		.proc_handler	= &proc_dointvec,
1101 	},
1102 	{
1103 		.ctl_name	= VM_SWAPPINESS,
1104 		.procname	= "swappiness",
1105 		.data		= &vm_swappiness,
1106 		.maxlen		= sizeof(vm_swappiness),
1107 		.mode		= 0644,
1108 		.proc_handler	= &proc_dointvec_minmax,
1109 		.strategy	= &sysctl_intvec,
1110 		.extra1		= &zero,
1111 		.extra2		= &one_hundred,
1112 	},
1113 #ifdef CONFIG_HUGETLB_PAGE
1114 	 {
1115 		.procname	= "nr_hugepages",
1116 		.data		= NULL,
1117 		.maxlen		= sizeof(unsigned long),
1118 		.mode		= 0644,
1119 		.proc_handler	= &hugetlb_sysctl_handler,
1120 		.extra1		= (void *)&hugetlb_zero,
1121 		.extra2		= (void *)&hugetlb_infinity,
1122 	 },
1123 	 {
1124 		.ctl_name	= VM_HUGETLB_GROUP,
1125 		.procname	= "hugetlb_shm_group",
1126 		.data		= &sysctl_hugetlb_shm_group,
1127 		.maxlen		= sizeof(gid_t),
1128 		.mode		= 0644,
1129 		.proc_handler	= &proc_dointvec,
1130 	 },
1131 	 {
1132 		.ctl_name	= CTL_UNNUMBERED,
1133 		.procname	= "hugepages_treat_as_movable",
1134 		.data		= &hugepages_treat_as_movable,
1135 		.maxlen		= sizeof(int),
1136 		.mode		= 0644,
1137 		.proc_handler	= &hugetlb_treat_movable_handler,
1138 	},
1139 	{
1140 		.ctl_name	= CTL_UNNUMBERED,
1141 		.procname	= "nr_overcommit_hugepages",
1142 		.data		= NULL,
1143 		.maxlen		= sizeof(unsigned long),
1144 		.mode		= 0644,
1145 		.proc_handler	= &hugetlb_overcommit_handler,
1146 		.extra1		= (void *)&hugetlb_zero,
1147 		.extra2		= (void *)&hugetlb_infinity,
1148 	},
1149 #endif
1150 	{
1151 		.ctl_name	= VM_LOWMEM_RESERVE_RATIO,
1152 		.procname	= "lowmem_reserve_ratio",
1153 		.data		= &sysctl_lowmem_reserve_ratio,
1154 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1155 		.mode		= 0644,
1156 		.proc_handler	= &lowmem_reserve_ratio_sysctl_handler,
1157 		.strategy	= &sysctl_intvec,
1158 	},
1159 	{
1160 		.ctl_name	= VM_DROP_PAGECACHE,
1161 		.procname	= "drop_caches",
1162 		.data		= &sysctl_drop_caches,
1163 		.maxlen		= sizeof(int),
1164 		.mode		= 0644,
1165 		.proc_handler	= drop_caches_sysctl_handler,
1166 		.strategy	= &sysctl_intvec,
1167 	},
1168 	{
1169 		.ctl_name	= VM_MIN_FREE_KBYTES,
1170 		.procname	= "min_free_kbytes",
1171 		.data		= &min_free_kbytes,
1172 		.maxlen		= sizeof(min_free_kbytes),
1173 		.mode		= 0644,
1174 		.proc_handler	= &min_free_kbytes_sysctl_handler,
1175 		.strategy	= &sysctl_intvec,
1176 		.extra1		= &zero,
1177 	},
1178 	{
1179 		.ctl_name	= VM_PERCPU_PAGELIST_FRACTION,
1180 		.procname	= "percpu_pagelist_fraction",
1181 		.data		= &percpu_pagelist_fraction,
1182 		.maxlen		= sizeof(percpu_pagelist_fraction),
1183 		.mode		= 0644,
1184 		.proc_handler	= &percpu_pagelist_fraction_sysctl_handler,
1185 		.strategy	= &sysctl_intvec,
1186 		.extra1		= &min_percpu_pagelist_fract,
1187 	},
1188 #ifdef CONFIG_MMU
1189 	{
1190 		.ctl_name	= VM_MAX_MAP_COUNT,
1191 		.procname	= "max_map_count",
1192 		.data		= &sysctl_max_map_count,
1193 		.maxlen		= sizeof(sysctl_max_map_count),
1194 		.mode		= 0644,
1195 		.proc_handler	= &proc_dointvec
1196 	},
1197 #else
1198 	{
1199 		.ctl_name	= CTL_UNNUMBERED,
1200 		.procname	= "nr_trim_pages",
1201 		.data		= &sysctl_nr_trim_pages,
1202 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1203 		.mode		= 0644,
1204 		.proc_handler	= &proc_dointvec_minmax,
1205 		.strategy	= &sysctl_intvec,
1206 		.extra1		= &zero,
1207 	},
1208 #endif
1209 	{
1210 		.ctl_name	= VM_LAPTOP_MODE,
1211 		.procname	= "laptop_mode",
1212 		.data		= &laptop_mode,
1213 		.maxlen		= sizeof(laptop_mode),
1214 		.mode		= 0644,
1215 		.proc_handler	= &proc_dointvec_jiffies,
1216 		.strategy	= &sysctl_jiffies,
1217 	},
1218 	{
1219 		.ctl_name	= VM_BLOCK_DUMP,
1220 		.procname	= "block_dump",
1221 		.data		= &block_dump,
1222 		.maxlen		= sizeof(block_dump),
1223 		.mode		= 0644,
1224 		.proc_handler	= &proc_dointvec,
1225 		.strategy	= &sysctl_intvec,
1226 		.extra1		= &zero,
1227 	},
1228 	{
1229 		.ctl_name	= VM_VFS_CACHE_PRESSURE,
1230 		.procname	= "vfs_cache_pressure",
1231 		.data		= &sysctl_vfs_cache_pressure,
1232 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1233 		.mode		= 0644,
1234 		.proc_handler	= &proc_dointvec,
1235 		.strategy	= &sysctl_intvec,
1236 		.extra1		= &zero,
1237 	},
1238 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1239 	{
1240 		.ctl_name	= VM_LEGACY_VA_LAYOUT,
1241 		.procname	= "legacy_va_layout",
1242 		.data		= &sysctl_legacy_va_layout,
1243 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1244 		.mode		= 0644,
1245 		.proc_handler	= &proc_dointvec,
1246 		.strategy	= &sysctl_intvec,
1247 		.extra1		= &zero,
1248 	},
1249 #endif
1250 #ifdef CONFIG_NUMA
1251 	{
1252 		.ctl_name	= VM_ZONE_RECLAIM_MODE,
1253 		.procname	= "zone_reclaim_mode",
1254 		.data		= &zone_reclaim_mode,
1255 		.maxlen		= sizeof(zone_reclaim_mode),
1256 		.mode		= 0644,
1257 		.proc_handler	= &proc_dointvec,
1258 		.strategy	= &sysctl_intvec,
1259 		.extra1		= &zero,
1260 	},
1261 	{
1262 		.ctl_name	= VM_MIN_UNMAPPED,
1263 		.procname	= "min_unmapped_ratio",
1264 		.data		= &sysctl_min_unmapped_ratio,
1265 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1266 		.mode		= 0644,
1267 		.proc_handler	= &sysctl_min_unmapped_ratio_sysctl_handler,
1268 		.strategy	= &sysctl_intvec,
1269 		.extra1		= &zero,
1270 		.extra2		= &one_hundred,
1271 	},
1272 	{
1273 		.ctl_name	= VM_MIN_SLAB,
1274 		.procname	= "min_slab_ratio",
1275 		.data		= &sysctl_min_slab_ratio,
1276 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1277 		.mode		= 0644,
1278 		.proc_handler	= &sysctl_min_slab_ratio_sysctl_handler,
1279 		.strategy	= &sysctl_intvec,
1280 		.extra1		= &zero,
1281 		.extra2		= &one_hundred,
1282 	},
1283 #endif
1284 #ifdef CONFIG_SMP
1285 	{
1286 		.ctl_name	= CTL_UNNUMBERED,
1287 		.procname	= "stat_interval",
1288 		.data		= &sysctl_stat_interval,
1289 		.maxlen		= sizeof(sysctl_stat_interval),
1290 		.mode		= 0644,
1291 		.proc_handler	= &proc_dointvec_jiffies,
1292 		.strategy	= &sysctl_jiffies,
1293 	},
1294 #endif
1295 	{
1296 		.ctl_name	= CTL_UNNUMBERED,
1297 		.procname	= "mmap_min_addr",
1298 		.data		= &mmap_min_addr,
1299 		.maxlen         = sizeof(unsigned long),
1300 		.mode		= 0644,
1301 		.proc_handler	= &proc_doulongvec_minmax,
1302 	},
1303 #ifdef CONFIG_NUMA
1304 	{
1305 		.ctl_name	= CTL_UNNUMBERED,
1306 		.procname	= "numa_zonelist_order",
1307 		.data		= &numa_zonelist_order,
1308 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1309 		.mode		= 0644,
1310 		.proc_handler	= &numa_zonelist_order_handler,
1311 		.strategy	= &sysctl_string,
1312 	},
1313 #endif
1314 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1315    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1316 	{
1317 		.ctl_name	= VM_VDSO_ENABLED,
1318 		.procname	= "vdso_enabled",
1319 		.data		= &vdso_enabled,
1320 		.maxlen		= sizeof(vdso_enabled),
1321 		.mode		= 0644,
1322 		.proc_handler	= &proc_dointvec,
1323 		.strategy	= &sysctl_intvec,
1324 		.extra1		= &zero,
1325 	},
1326 #endif
1327 #ifdef CONFIG_HIGHMEM
1328 	{
1329 		.ctl_name	= CTL_UNNUMBERED,
1330 		.procname	= "highmem_is_dirtyable",
1331 		.data		= &vm_highmem_is_dirtyable,
1332 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1333 		.mode		= 0644,
1334 		.proc_handler	= &proc_dointvec_minmax,
1335 		.strategy	= &sysctl_intvec,
1336 		.extra1		= &zero,
1337 		.extra2		= &one,
1338 	},
1339 #endif
1340 	{
1341 		.ctl_name	= CTL_UNNUMBERED,
1342 		.procname	= "scan_unevictable_pages",
1343 		.data		= &scan_unevictable_pages,
1344 		.maxlen		= sizeof(scan_unevictable_pages),
1345 		.mode		= 0644,
1346 		.proc_handler	= &scan_unevictable_handler,
1347 	},
1348 /*
1349  * NOTE: do not add new entries to this table unless you have read
1350  * Documentation/sysctl/ctl_unnumbered.txt
1351  */
1352 	{ .ctl_name = 0 }
1353 };
1354 
1355 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1356 static struct ctl_table binfmt_misc_table[] = {
1357 	{ .ctl_name = 0 }
1358 };
1359 #endif
1360 
1361 static struct ctl_table fs_table[] = {
1362 	{
1363 		.ctl_name	= FS_NRINODE,
1364 		.procname	= "inode-nr",
1365 		.data		= &inodes_stat,
1366 		.maxlen		= 2*sizeof(int),
1367 		.mode		= 0444,
1368 		.proc_handler	= &proc_dointvec,
1369 	},
1370 	{
1371 		.ctl_name	= FS_STATINODE,
1372 		.procname	= "inode-state",
1373 		.data		= &inodes_stat,
1374 		.maxlen		= 7*sizeof(int),
1375 		.mode		= 0444,
1376 		.proc_handler	= &proc_dointvec,
1377 	},
1378 	{
1379 		.procname	= "file-nr",
1380 		.data		= &files_stat,
1381 		.maxlen		= 3*sizeof(int),
1382 		.mode		= 0444,
1383 		.proc_handler	= &proc_nr_files,
1384 	},
1385 	{
1386 		.ctl_name	= FS_MAXFILE,
1387 		.procname	= "file-max",
1388 		.data		= &files_stat.max_files,
1389 		.maxlen		= sizeof(int),
1390 		.mode		= 0644,
1391 		.proc_handler	= &proc_dointvec,
1392 	},
1393 	{
1394 		.ctl_name	= CTL_UNNUMBERED,
1395 		.procname	= "nr_open",
1396 		.data		= &sysctl_nr_open,
1397 		.maxlen		= sizeof(int),
1398 		.mode		= 0644,
1399 		.proc_handler	= &proc_dointvec_minmax,
1400 		.extra1		= &sysctl_nr_open_min,
1401 		.extra2		= &sysctl_nr_open_max,
1402 	},
1403 	{
1404 		.ctl_name	= FS_DENTRY,
1405 		.procname	= "dentry-state",
1406 		.data		= &dentry_stat,
1407 		.maxlen		= 6*sizeof(int),
1408 		.mode		= 0444,
1409 		.proc_handler	= &proc_dointvec,
1410 	},
1411 	{
1412 		.ctl_name	= FS_OVERFLOWUID,
1413 		.procname	= "overflowuid",
1414 		.data		= &fs_overflowuid,
1415 		.maxlen		= sizeof(int),
1416 		.mode		= 0644,
1417 		.proc_handler	= &proc_dointvec_minmax,
1418 		.strategy	= &sysctl_intvec,
1419 		.extra1		= &minolduid,
1420 		.extra2		= &maxolduid,
1421 	},
1422 	{
1423 		.ctl_name	= FS_OVERFLOWGID,
1424 		.procname	= "overflowgid",
1425 		.data		= &fs_overflowgid,
1426 		.maxlen		= sizeof(int),
1427 		.mode		= 0644,
1428 		.proc_handler	= &proc_dointvec_minmax,
1429 		.strategy	= &sysctl_intvec,
1430 		.extra1		= &minolduid,
1431 		.extra2		= &maxolduid,
1432 	},
1433 #ifdef CONFIG_FILE_LOCKING
1434 	{
1435 		.ctl_name	= FS_LEASES,
1436 		.procname	= "leases-enable",
1437 		.data		= &leases_enable,
1438 		.maxlen		= sizeof(int),
1439 		.mode		= 0644,
1440 		.proc_handler	= &proc_dointvec,
1441 	},
1442 #endif
1443 #ifdef CONFIG_DNOTIFY
1444 	{
1445 		.ctl_name	= FS_DIR_NOTIFY,
1446 		.procname	= "dir-notify-enable",
1447 		.data		= &dir_notify_enable,
1448 		.maxlen		= sizeof(int),
1449 		.mode		= 0644,
1450 		.proc_handler	= &proc_dointvec,
1451 	},
1452 #endif
1453 #ifdef CONFIG_MMU
1454 #ifdef CONFIG_FILE_LOCKING
1455 	{
1456 		.ctl_name	= FS_LEASE_TIME,
1457 		.procname	= "lease-break-time",
1458 		.data		= &lease_break_time,
1459 		.maxlen		= sizeof(int),
1460 		.mode		= 0644,
1461 		.proc_handler	= &proc_dointvec,
1462 	},
1463 #endif
1464 #ifdef CONFIG_AIO
1465 	{
1466 		.procname	= "aio-nr",
1467 		.data		= &aio_nr,
1468 		.maxlen		= sizeof(aio_nr),
1469 		.mode		= 0444,
1470 		.proc_handler	= &proc_doulongvec_minmax,
1471 	},
1472 	{
1473 		.procname	= "aio-max-nr",
1474 		.data		= &aio_max_nr,
1475 		.maxlen		= sizeof(aio_max_nr),
1476 		.mode		= 0644,
1477 		.proc_handler	= &proc_doulongvec_minmax,
1478 	},
1479 #endif /* CONFIG_AIO */
1480 #ifdef CONFIG_INOTIFY_USER
1481 	{
1482 		.ctl_name	= FS_INOTIFY,
1483 		.procname	= "inotify",
1484 		.mode		= 0555,
1485 		.child		= inotify_table,
1486 	},
1487 #endif
1488 #ifdef CONFIG_EPOLL
1489 	{
1490 		.procname	= "epoll",
1491 		.mode		= 0555,
1492 		.child		= epoll_table,
1493 	},
1494 #endif
1495 #endif
1496 	{
1497 		.ctl_name	= KERN_SETUID_DUMPABLE,
1498 		.procname	= "suid_dumpable",
1499 		.data		= &suid_dumpable,
1500 		.maxlen		= sizeof(int),
1501 		.mode		= 0644,
1502 		.proc_handler	= &proc_dointvec_minmax,
1503 		.strategy	= &sysctl_intvec,
1504 		.extra1		= &zero,
1505 		.extra2		= &two,
1506 	},
1507 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1508 	{
1509 		.ctl_name	= CTL_UNNUMBERED,
1510 		.procname	= "binfmt_misc",
1511 		.mode		= 0555,
1512 		.child		= binfmt_misc_table,
1513 	},
1514 #endif
1515 /*
1516  * NOTE: do not add new entries to this table unless you have read
1517  * Documentation/sysctl/ctl_unnumbered.txt
1518  */
1519 	{ .ctl_name = 0 }
1520 };
1521 
1522 static struct ctl_table debug_table[] = {
1523 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1524 	{
1525 		.ctl_name	= CTL_UNNUMBERED,
1526 		.procname	= "exception-trace",
1527 		.data		= &show_unhandled_signals,
1528 		.maxlen		= sizeof(int),
1529 		.mode		= 0644,
1530 		.proc_handler	= proc_dointvec
1531 	},
1532 #endif
1533 	{ .ctl_name = 0 }
1534 };
1535 
1536 static struct ctl_table dev_table[] = {
1537 	{ .ctl_name = 0 }
1538 };
1539 
1540 static DEFINE_SPINLOCK(sysctl_lock);
1541 
1542 /* called under sysctl_lock */
1543 static int use_table(struct ctl_table_header *p)
1544 {
1545 	if (unlikely(p->unregistering))
1546 		return 0;
1547 	p->used++;
1548 	return 1;
1549 }
1550 
1551 /* called under sysctl_lock */
1552 static void unuse_table(struct ctl_table_header *p)
1553 {
1554 	if (!--p->used)
1555 		if (unlikely(p->unregistering))
1556 			complete(p->unregistering);
1557 }
1558 
1559 /* called under sysctl_lock, will reacquire if has to wait */
1560 static void start_unregistering(struct ctl_table_header *p)
1561 {
1562 	/*
1563 	 * if p->used is 0, nobody will ever touch that entry again;
1564 	 * we'll eliminate all paths to it before dropping sysctl_lock
1565 	 */
1566 	if (unlikely(p->used)) {
1567 		struct completion wait;
1568 		init_completion(&wait);
1569 		p->unregistering = &wait;
1570 		spin_unlock(&sysctl_lock);
1571 		wait_for_completion(&wait);
1572 		spin_lock(&sysctl_lock);
1573 	} else {
1574 		/* anything non-NULL; we'll never dereference it */
1575 		p->unregistering = ERR_PTR(-EINVAL);
1576 	}
1577 	/*
1578 	 * do not remove from the list until nobody holds it; walking the
1579 	 * list in do_sysctl() relies on that.
1580 	 */
1581 	list_del_init(&p->ctl_entry);
1582 }
1583 
1584 void sysctl_head_get(struct ctl_table_header *head)
1585 {
1586 	spin_lock(&sysctl_lock);
1587 	head->count++;
1588 	spin_unlock(&sysctl_lock);
1589 }
1590 
1591 void sysctl_head_put(struct ctl_table_header *head)
1592 {
1593 	spin_lock(&sysctl_lock);
1594 	if (!--head->count)
1595 		kfree(head);
1596 	spin_unlock(&sysctl_lock);
1597 }
1598 
1599 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1600 {
1601 	if (!head)
1602 		BUG();
1603 	spin_lock(&sysctl_lock);
1604 	if (!use_table(head))
1605 		head = ERR_PTR(-ENOENT);
1606 	spin_unlock(&sysctl_lock);
1607 	return head;
1608 }
1609 
1610 void sysctl_head_finish(struct ctl_table_header *head)
1611 {
1612 	if (!head)
1613 		return;
1614 	spin_lock(&sysctl_lock);
1615 	unuse_table(head);
1616 	spin_unlock(&sysctl_lock);
1617 }
1618 
1619 static struct ctl_table_set *
1620 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1621 {
1622 	struct ctl_table_set *set = &root->default_set;
1623 	if (root->lookup)
1624 		set = root->lookup(root, namespaces);
1625 	return set;
1626 }
1627 
1628 static struct list_head *
1629 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1630 {
1631 	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1632 	return &set->list;
1633 }
1634 
1635 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1636 					    struct ctl_table_header *prev)
1637 {
1638 	struct ctl_table_root *root;
1639 	struct list_head *header_list;
1640 	struct ctl_table_header *head;
1641 	struct list_head *tmp;
1642 
1643 	spin_lock(&sysctl_lock);
1644 	if (prev) {
1645 		head = prev;
1646 		tmp = &prev->ctl_entry;
1647 		unuse_table(prev);
1648 		goto next;
1649 	}
1650 	tmp = &root_table_header.ctl_entry;
1651 	for (;;) {
1652 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1653 
1654 		if (!use_table(head))
1655 			goto next;
1656 		spin_unlock(&sysctl_lock);
1657 		return head;
1658 	next:
1659 		root = head->root;
1660 		tmp = tmp->next;
1661 		header_list = lookup_header_list(root, namespaces);
1662 		if (tmp != header_list)
1663 			continue;
1664 
1665 		do {
1666 			root = list_entry(root->root_list.next,
1667 					struct ctl_table_root, root_list);
1668 			if (root == &sysctl_table_root)
1669 				goto out;
1670 			header_list = lookup_header_list(root, namespaces);
1671 		} while (list_empty(header_list));
1672 		tmp = header_list->next;
1673 	}
1674 out:
1675 	spin_unlock(&sysctl_lock);
1676 	return NULL;
1677 }
1678 
1679 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1680 {
1681 	return __sysctl_head_next(current->nsproxy, prev);
1682 }
1683 
1684 void register_sysctl_root(struct ctl_table_root *root)
1685 {
1686 	spin_lock(&sysctl_lock);
1687 	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1688 	spin_unlock(&sysctl_lock);
1689 }
1690 
1691 #ifdef CONFIG_SYSCTL_SYSCALL
1692 /* Perform the actual read/write of a sysctl table entry. */
1693 static int do_sysctl_strategy(struct ctl_table_root *root,
1694 			struct ctl_table *table,
1695 			void __user *oldval, size_t __user *oldlenp,
1696 			void __user *newval, size_t newlen)
1697 {
1698 	int op = 0, rc;
1699 
1700 	if (oldval)
1701 		op |= MAY_READ;
1702 	if (newval)
1703 		op |= MAY_WRITE;
1704 	if (sysctl_perm(root, table, op))
1705 		return -EPERM;
1706 
1707 	if (table->strategy) {
1708 		rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1709 		if (rc < 0)
1710 			return rc;
1711 		if (rc > 0)
1712 			return 0;
1713 	}
1714 
1715 	/* If there is no strategy routine, or if the strategy returns
1716 	 * zero, proceed with automatic r/w */
1717 	if (table->data && table->maxlen) {
1718 		rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1719 		if (rc < 0)
1720 			return rc;
1721 	}
1722 	return 0;
1723 }
1724 
1725 static int parse_table(int __user *name, int nlen,
1726 		       void __user *oldval, size_t __user *oldlenp,
1727 		       void __user *newval, size_t newlen,
1728 		       struct ctl_table_root *root,
1729 		       struct ctl_table *table)
1730 {
1731 	int n;
1732 repeat:
1733 	if (!nlen)
1734 		return -ENOTDIR;
1735 	if (get_user(n, name))
1736 		return -EFAULT;
1737 	for ( ; table->ctl_name || table->procname; table++) {
1738 		if (!table->ctl_name)
1739 			continue;
1740 		if (n == table->ctl_name) {
1741 			int error;
1742 			if (table->child) {
1743 				if (sysctl_perm(root, table, MAY_EXEC))
1744 					return -EPERM;
1745 				name++;
1746 				nlen--;
1747 				table = table->child;
1748 				goto repeat;
1749 			}
1750 			error = do_sysctl_strategy(root, table,
1751 						   oldval, oldlenp,
1752 						   newval, newlen);
1753 			return error;
1754 		}
1755 	}
1756 	return -ENOTDIR;
1757 }
1758 
1759 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1760 	       void __user *newval, size_t newlen)
1761 {
1762 	struct ctl_table_header *head;
1763 	int error = -ENOTDIR;
1764 
1765 	if (nlen <= 0 || nlen >= CTL_MAXNAME)
1766 		return -ENOTDIR;
1767 	if (oldval) {
1768 		int old_len;
1769 		if (!oldlenp || get_user(old_len, oldlenp))
1770 			return -EFAULT;
1771 	}
1772 
1773 	for (head = sysctl_head_next(NULL); head;
1774 			head = sysctl_head_next(head)) {
1775 		error = parse_table(name, nlen, oldval, oldlenp,
1776 					newval, newlen,
1777 					head->root, head->ctl_table);
1778 		if (error != -ENOTDIR) {
1779 			sysctl_head_finish(head);
1780 			break;
1781 		}
1782 	}
1783 	return error;
1784 }
1785 
1786 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
1787 {
1788 	struct __sysctl_args tmp;
1789 	int error;
1790 
1791 	if (copy_from_user(&tmp, args, sizeof(tmp)))
1792 		return -EFAULT;
1793 
1794 	error = deprecated_sysctl_warning(&tmp);
1795 	if (error)
1796 		goto out;
1797 
1798 	lock_kernel();
1799 	error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1800 			  tmp.newval, tmp.newlen);
1801 	unlock_kernel();
1802 out:
1803 	return error;
1804 }
1805 #endif /* CONFIG_SYSCTL_SYSCALL */
1806 
1807 /*
1808  * sysctl_perm does NOT grant the superuser all rights automatically, because
1809  * some sysctl variables are readonly even to root.
1810  */
1811 
1812 static int test_perm(int mode, int op)
1813 {
1814 	if (!current_euid())
1815 		mode >>= 6;
1816 	else if (in_egroup_p(0))
1817 		mode >>= 3;
1818 	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1819 		return 0;
1820 	return -EACCES;
1821 }
1822 
1823 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1824 {
1825 	int error;
1826 	int mode;
1827 
1828 	error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1829 	if (error)
1830 		return error;
1831 
1832 	if (root->permissions)
1833 		mode = root->permissions(root, current->nsproxy, table);
1834 	else
1835 		mode = table->mode;
1836 
1837 	return test_perm(mode, op);
1838 }
1839 
1840 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1841 {
1842 	for (; table->ctl_name || table->procname; table++) {
1843 		table->parent = parent;
1844 		if (table->child)
1845 			sysctl_set_parent(table, table->child);
1846 	}
1847 }
1848 
1849 static __init int sysctl_init(void)
1850 {
1851 	sysctl_set_parent(NULL, root_table);
1852 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1853 	{
1854 		int err;
1855 		err = sysctl_check_table(current->nsproxy, root_table);
1856 	}
1857 #endif
1858 	return 0;
1859 }
1860 
1861 core_initcall(sysctl_init);
1862 
1863 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1864 				      struct ctl_table *table)
1865 {
1866 	struct ctl_table *p;
1867 	const char *s = branch->procname;
1868 
1869 	/* branch should have named subdirectory as its first element */
1870 	if (!s || !branch->child)
1871 		return NULL;
1872 
1873 	/* ... and nothing else */
1874 	if (branch[1].procname || branch[1].ctl_name)
1875 		return NULL;
1876 
1877 	/* table should contain subdirectory with the same name */
1878 	for (p = table; p->procname || p->ctl_name; p++) {
1879 		if (!p->child)
1880 			continue;
1881 		if (p->procname && strcmp(p->procname, s) == 0)
1882 			return p;
1883 	}
1884 	return NULL;
1885 }
1886 
1887 /* see if attaching q to p would be an improvement */
1888 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1889 {
1890 	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1891 	struct ctl_table *next;
1892 	int is_better = 0;
1893 	int not_in_parent = !p->attached_by;
1894 
1895 	while ((next = is_branch_in(by, to)) != NULL) {
1896 		if (by == q->attached_by)
1897 			is_better = 1;
1898 		if (to == p->attached_by)
1899 			not_in_parent = 1;
1900 		by = by->child;
1901 		to = next->child;
1902 	}
1903 
1904 	if (is_better && not_in_parent) {
1905 		q->attached_by = by;
1906 		q->attached_to = to;
1907 		q->parent = p;
1908 	}
1909 }
1910 
1911 /**
1912  * __register_sysctl_paths - register a sysctl hierarchy
1913  * @root: List of sysctl headers to register on
1914  * @namespaces: Data to compute which lists of sysctl entries are visible
1915  * @path: The path to the directory the sysctl table is in.
1916  * @table: the top-level table structure
1917  *
1918  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1919  * array. A completely 0 filled entry terminates the table.
1920  *
1921  * The members of the &struct ctl_table structure are used as follows:
1922  *
1923  * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1924  *            must be unique within that level of sysctl
1925  *
1926  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1927  *            enter a sysctl file
1928  *
1929  * data - a pointer to data for use by proc_handler
1930  *
1931  * maxlen - the maximum size in bytes of the data
1932  *
1933  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1934  *
1935  * child - a pointer to the child sysctl table if this entry is a directory, or
1936  *         %NULL.
1937  *
1938  * proc_handler - the text handler routine (described below)
1939  *
1940  * strategy - the strategy routine (described below)
1941  *
1942  * de - for internal use by the sysctl routines
1943  *
1944  * extra1, extra2 - extra pointers usable by the proc handler routines
1945  *
1946  * Leaf nodes in the sysctl tree will be represented by a single file
1947  * under /proc; non-leaf nodes will be represented by directories.
1948  *
1949  * sysctl(2) can automatically manage read and write requests through
1950  * the sysctl table.  The data and maxlen fields of the ctl_table
1951  * struct enable minimal validation of the values being written to be
1952  * performed, and the mode field allows minimal authentication.
1953  *
1954  * More sophisticated management can be enabled by the provision of a
1955  * strategy routine with the table entry.  This will be called before
1956  * any automatic read or write of the data is performed.
1957  *
1958  * The strategy routine may return
1959  *
1960  * < 0 - Error occurred (error is passed to user process)
1961  *
1962  * 0   - OK - proceed with automatic read or write.
1963  *
1964  * > 0 - OK - read or write has been done by the strategy routine, so
1965  *       return immediately.
1966  *
1967  * There must be a proc_handler routine for any terminal nodes
1968  * mirrored under /proc/sys (non-terminals are handled by a built-in
1969  * directory handler).  Several default handlers are available to
1970  * cover common cases -
1971  *
1972  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1973  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1974  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1975  *
1976  * It is the handler's job to read the input buffer from user memory
1977  * and process it. The handler should return 0 on success.
1978  *
1979  * This routine returns %NULL on a failure to register, and a pointer
1980  * to the table header on success.
1981  */
1982 struct ctl_table_header *__register_sysctl_paths(
1983 	struct ctl_table_root *root,
1984 	struct nsproxy *namespaces,
1985 	const struct ctl_path *path, struct ctl_table *table)
1986 {
1987 	struct ctl_table_header *header;
1988 	struct ctl_table *new, **prevp;
1989 	unsigned int n, npath;
1990 	struct ctl_table_set *set;
1991 
1992 	/* Count the path components */
1993 	for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1994 		;
1995 
1996 	/*
1997 	 * For each path component, allocate a 2-element ctl_table array.
1998 	 * The first array element will be filled with the sysctl entry
1999 	 * for this, the second will be the sentinel (ctl_name == 0).
2000 	 *
2001 	 * We allocate everything in one go so that we don't have to
2002 	 * worry about freeing additional memory in unregister_sysctl_table.
2003 	 */
2004 	header = kzalloc(sizeof(struct ctl_table_header) +
2005 			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
2006 	if (!header)
2007 		return NULL;
2008 
2009 	new = (struct ctl_table *) (header + 1);
2010 
2011 	/* Now connect the dots */
2012 	prevp = &header->ctl_table;
2013 	for (n = 0; n < npath; ++n, ++path) {
2014 		/* Copy the procname */
2015 		new->procname = path->procname;
2016 		new->ctl_name = path->ctl_name;
2017 		new->mode     = 0555;
2018 
2019 		*prevp = new;
2020 		prevp = &new->child;
2021 
2022 		new += 2;
2023 	}
2024 	*prevp = table;
2025 	header->ctl_table_arg = table;
2026 
2027 	INIT_LIST_HEAD(&header->ctl_entry);
2028 	header->used = 0;
2029 	header->unregistering = NULL;
2030 	header->root = root;
2031 	sysctl_set_parent(NULL, header->ctl_table);
2032 	header->count = 1;
2033 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
2034 	if (sysctl_check_table(namespaces, header->ctl_table)) {
2035 		kfree(header);
2036 		return NULL;
2037 	}
2038 #endif
2039 	spin_lock(&sysctl_lock);
2040 	header->set = lookup_header_set(root, namespaces);
2041 	header->attached_by = header->ctl_table;
2042 	header->attached_to = root_table;
2043 	header->parent = &root_table_header;
2044 	for (set = header->set; set; set = set->parent) {
2045 		struct ctl_table_header *p;
2046 		list_for_each_entry(p, &set->list, ctl_entry) {
2047 			if (p->unregistering)
2048 				continue;
2049 			try_attach(p, header);
2050 		}
2051 	}
2052 	header->parent->count++;
2053 	list_add_tail(&header->ctl_entry, &header->set->list);
2054 	spin_unlock(&sysctl_lock);
2055 
2056 	return header;
2057 }
2058 
2059 /**
2060  * register_sysctl_table_path - register a sysctl table hierarchy
2061  * @path: The path to the directory the sysctl table is in.
2062  * @table: the top-level table structure
2063  *
2064  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2065  * array. A completely 0 filled entry terminates the table.
2066  *
2067  * See __register_sysctl_paths for more details.
2068  */
2069 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2070 						struct ctl_table *table)
2071 {
2072 	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
2073 					path, table);
2074 }
2075 
2076 /**
2077  * register_sysctl_table - register a sysctl table hierarchy
2078  * @table: the top-level table structure
2079  *
2080  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2081  * array. A completely 0 filled entry terminates the table.
2082  *
2083  * See register_sysctl_paths for more details.
2084  */
2085 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
2086 {
2087 	static const struct ctl_path null_path[] = { {} };
2088 
2089 	return register_sysctl_paths(null_path, table);
2090 }
2091 
2092 /**
2093  * unregister_sysctl_table - unregister a sysctl table hierarchy
2094  * @header: the header returned from register_sysctl_table
2095  *
2096  * Unregisters the sysctl table and all children. proc entries may not
2097  * actually be removed until they are no longer used by anyone.
2098  */
2099 void unregister_sysctl_table(struct ctl_table_header * header)
2100 {
2101 	might_sleep();
2102 
2103 	if (header == NULL)
2104 		return;
2105 
2106 	spin_lock(&sysctl_lock);
2107 	start_unregistering(header);
2108 	if (!--header->parent->count) {
2109 		WARN_ON(1);
2110 		kfree(header->parent);
2111 	}
2112 	if (!--header->count)
2113 		kfree(header);
2114 	spin_unlock(&sysctl_lock);
2115 }
2116 
2117 int sysctl_is_seen(struct ctl_table_header *p)
2118 {
2119 	struct ctl_table_set *set = p->set;
2120 	int res;
2121 	spin_lock(&sysctl_lock);
2122 	if (p->unregistering)
2123 		res = 0;
2124 	else if (!set->is_seen)
2125 		res = 1;
2126 	else
2127 		res = set->is_seen(set);
2128 	spin_unlock(&sysctl_lock);
2129 	return res;
2130 }
2131 
2132 void setup_sysctl_set(struct ctl_table_set *p,
2133 	struct ctl_table_set *parent,
2134 	int (*is_seen)(struct ctl_table_set *))
2135 {
2136 	INIT_LIST_HEAD(&p->list);
2137 	p->parent = parent ? parent : &sysctl_table_root.default_set;
2138 	p->is_seen = is_seen;
2139 }
2140 
2141 #else /* !CONFIG_SYSCTL */
2142 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2143 {
2144 	return NULL;
2145 }
2146 
2147 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2148 						    struct ctl_table *table)
2149 {
2150 	return NULL;
2151 }
2152 
2153 void unregister_sysctl_table(struct ctl_table_header * table)
2154 {
2155 }
2156 
2157 void setup_sysctl_set(struct ctl_table_set *p,
2158 	struct ctl_table_set *parent,
2159 	int (*is_seen)(struct ctl_table_set *))
2160 {
2161 }
2162 
2163 void sysctl_head_put(struct ctl_table_header *head)
2164 {
2165 }
2166 
2167 #endif /* CONFIG_SYSCTL */
2168 
2169 /*
2170  * /proc/sys support
2171  */
2172 
2173 #ifdef CONFIG_PROC_SYSCTL
2174 
2175 static int _proc_do_string(void* data, int maxlen, int write,
2176 			   struct file *filp, void __user *buffer,
2177 			   size_t *lenp, loff_t *ppos)
2178 {
2179 	size_t len;
2180 	char __user *p;
2181 	char c;
2182 
2183 	if (!data || !maxlen || !*lenp) {
2184 		*lenp = 0;
2185 		return 0;
2186 	}
2187 
2188 	if (write) {
2189 		len = 0;
2190 		p = buffer;
2191 		while (len < *lenp) {
2192 			if (get_user(c, p++))
2193 				return -EFAULT;
2194 			if (c == 0 || c == '\n')
2195 				break;
2196 			len++;
2197 		}
2198 		if (len >= maxlen)
2199 			len = maxlen-1;
2200 		if(copy_from_user(data, buffer, len))
2201 			return -EFAULT;
2202 		((char *) data)[len] = 0;
2203 		*ppos += *lenp;
2204 	} else {
2205 		len = strlen(data);
2206 		if (len > maxlen)
2207 			len = maxlen;
2208 
2209 		if (*ppos > len) {
2210 			*lenp = 0;
2211 			return 0;
2212 		}
2213 
2214 		data += *ppos;
2215 		len  -= *ppos;
2216 
2217 		if (len > *lenp)
2218 			len = *lenp;
2219 		if (len)
2220 			if(copy_to_user(buffer, data, len))
2221 				return -EFAULT;
2222 		if (len < *lenp) {
2223 			if(put_user('\n', ((char __user *) buffer) + len))
2224 				return -EFAULT;
2225 			len++;
2226 		}
2227 		*lenp = len;
2228 		*ppos += len;
2229 	}
2230 	return 0;
2231 }
2232 
2233 /**
2234  * proc_dostring - read a string sysctl
2235  * @table: the sysctl table
2236  * @write: %TRUE if this is a write to the sysctl file
2237  * @filp: the file structure
2238  * @buffer: the user buffer
2239  * @lenp: the size of the user buffer
2240  * @ppos: file position
2241  *
2242  * Reads/writes a string from/to the user buffer. If the kernel
2243  * buffer provided is not large enough to hold the string, the
2244  * string is truncated. The copied string is %NULL-terminated.
2245  * If the string is being read by the user process, it is copied
2246  * and a newline '\n' is added. It is truncated if the buffer is
2247  * not large enough.
2248  *
2249  * Returns 0 on success.
2250  */
2251 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2252 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2253 {
2254 	return _proc_do_string(table->data, table->maxlen, write, filp,
2255 			       buffer, lenp, ppos);
2256 }
2257 
2258 
2259 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2260 				 int *valp,
2261 				 int write, void *data)
2262 {
2263 	if (write) {
2264 		*valp = *negp ? -*lvalp : *lvalp;
2265 	} else {
2266 		int val = *valp;
2267 		if (val < 0) {
2268 			*negp = -1;
2269 			*lvalp = (unsigned long)-val;
2270 		} else {
2271 			*negp = 0;
2272 			*lvalp = (unsigned long)val;
2273 		}
2274 	}
2275 	return 0;
2276 }
2277 
2278 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2279 		  int write, struct file *filp, void __user *buffer,
2280 		  size_t *lenp, loff_t *ppos,
2281 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2282 			      int write, void *data),
2283 		  void *data)
2284 {
2285 #define TMPBUFLEN 21
2286 	int *i, vleft, first = 1, neg;
2287 	unsigned long lval;
2288 	size_t left, len;
2289 
2290 	char buf[TMPBUFLEN], *p;
2291 	char __user *s = buffer;
2292 
2293 	if (!tbl_data || !table->maxlen || !*lenp ||
2294 	    (*ppos && !write)) {
2295 		*lenp = 0;
2296 		return 0;
2297 	}
2298 
2299 	i = (int *) tbl_data;
2300 	vleft = table->maxlen / sizeof(*i);
2301 	left = *lenp;
2302 
2303 	if (!conv)
2304 		conv = do_proc_dointvec_conv;
2305 
2306 	for (; left && vleft--; i++, first=0) {
2307 		if (write) {
2308 			while (left) {
2309 				char c;
2310 				if (get_user(c, s))
2311 					return -EFAULT;
2312 				if (!isspace(c))
2313 					break;
2314 				left--;
2315 				s++;
2316 			}
2317 			if (!left)
2318 				break;
2319 			neg = 0;
2320 			len = left;
2321 			if (len > sizeof(buf) - 1)
2322 				len = sizeof(buf) - 1;
2323 			if (copy_from_user(buf, s, len))
2324 				return -EFAULT;
2325 			buf[len] = 0;
2326 			p = buf;
2327 			if (*p == '-' && left > 1) {
2328 				neg = 1;
2329 				p++;
2330 			}
2331 			if (*p < '0' || *p > '9')
2332 				break;
2333 
2334 			lval = simple_strtoul(p, &p, 0);
2335 
2336 			len = p-buf;
2337 			if ((len < left) && *p && !isspace(*p))
2338 				break;
2339 			s += len;
2340 			left -= len;
2341 
2342 			if (conv(&neg, &lval, i, 1, data))
2343 				break;
2344 		} else {
2345 			p = buf;
2346 			if (!first)
2347 				*p++ = '\t';
2348 
2349 			if (conv(&neg, &lval, i, 0, data))
2350 				break;
2351 
2352 			sprintf(p, "%s%lu", neg ? "-" : "", lval);
2353 			len = strlen(buf);
2354 			if (len > left)
2355 				len = left;
2356 			if(copy_to_user(s, buf, len))
2357 				return -EFAULT;
2358 			left -= len;
2359 			s += len;
2360 		}
2361 	}
2362 
2363 	if (!write && !first && left) {
2364 		if(put_user('\n', s))
2365 			return -EFAULT;
2366 		left--, s++;
2367 	}
2368 	if (write) {
2369 		while (left) {
2370 			char c;
2371 			if (get_user(c, s++))
2372 				return -EFAULT;
2373 			if (!isspace(c))
2374 				break;
2375 			left--;
2376 		}
2377 	}
2378 	if (write && first)
2379 		return -EINVAL;
2380 	*lenp -= left;
2381 	*ppos += *lenp;
2382 	return 0;
2383 #undef TMPBUFLEN
2384 }
2385 
2386 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2387 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2388 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2389 			      int write, void *data),
2390 		  void *data)
2391 {
2392 	return __do_proc_dointvec(table->data, table, write, filp,
2393 			buffer, lenp, ppos, conv, data);
2394 }
2395 
2396 /**
2397  * proc_dointvec - read a vector of integers
2398  * @table: the sysctl table
2399  * @write: %TRUE if this is a write to the sysctl file
2400  * @filp: the file structure
2401  * @buffer: the user buffer
2402  * @lenp: the size of the user buffer
2403  * @ppos: file position
2404  *
2405  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2406  * values from/to the user buffer, treated as an ASCII string.
2407  *
2408  * Returns 0 on success.
2409  */
2410 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2411 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2412 {
2413     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2414 		    	    NULL,NULL);
2415 }
2416 
2417 /*
2418  * Taint values can only be increased
2419  * This means we can safely use a temporary.
2420  */
2421 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
2422 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2423 {
2424 	struct ctl_table t;
2425 	unsigned long tmptaint = get_taint();
2426 	int err;
2427 
2428 	if (write && !capable(CAP_SYS_ADMIN))
2429 		return -EPERM;
2430 
2431 	t = *table;
2432 	t.data = &tmptaint;
2433 	err = proc_doulongvec_minmax(&t, write, filp, buffer, lenp, ppos);
2434 	if (err < 0)
2435 		return err;
2436 
2437 	if (write) {
2438 		/*
2439 		 * Poor man's atomic or. Not worth adding a primitive
2440 		 * to everyone's atomic.h for this
2441 		 */
2442 		int i;
2443 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2444 			if ((tmptaint >> i) & 1)
2445 				add_taint(i);
2446 		}
2447 	}
2448 
2449 	return err;
2450 }
2451 
2452 struct do_proc_dointvec_minmax_conv_param {
2453 	int *min;
2454 	int *max;
2455 };
2456 
2457 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2458 					int *valp,
2459 					int write, void *data)
2460 {
2461 	struct do_proc_dointvec_minmax_conv_param *param = data;
2462 	if (write) {
2463 		int val = *negp ? -*lvalp : *lvalp;
2464 		if ((param->min && *param->min > val) ||
2465 		    (param->max && *param->max < val))
2466 			return -EINVAL;
2467 		*valp = val;
2468 	} else {
2469 		int val = *valp;
2470 		if (val < 0) {
2471 			*negp = -1;
2472 			*lvalp = (unsigned long)-val;
2473 		} else {
2474 			*negp = 0;
2475 			*lvalp = (unsigned long)val;
2476 		}
2477 	}
2478 	return 0;
2479 }
2480 
2481 /**
2482  * proc_dointvec_minmax - read a vector of integers with min/max values
2483  * @table: the sysctl table
2484  * @write: %TRUE if this is a write to the sysctl file
2485  * @filp: the file structure
2486  * @buffer: the user buffer
2487  * @lenp: the size of the user buffer
2488  * @ppos: file position
2489  *
2490  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2491  * values from/to the user buffer, treated as an ASCII string.
2492  *
2493  * This routine will ensure the values are within the range specified by
2494  * table->extra1 (min) and table->extra2 (max).
2495  *
2496  * Returns 0 on success.
2497  */
2498 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2499 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2500 {
2501 	struct do_proc_dointvec_minmax_conv_param param = {
2502 		.min = (int *) table->extra1,
2503 		.max = (int *) table->extra2,
2504 	};
2505 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2506 				do_proc_dointvec_minmax_conv, &param);
2507 }
2508 
2509 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2510 				     struct file *filp,
2511 				     void __user *buffer,
2512 				     size_t *lenp, loff_t *ppos,
2513 				     unsigned long convmul,
2514 				     unsigned long convdiv)
2515 {
2516 #define TMPBUFLEN 21
2517 	unsigned long *i, *min, *max, val;
2518 	int vleft, first=1, neg;
2519 	size_t len, left;
2520 	char buf[TMPBUFLEN], *p;
2521 	char __user *s = buffer;
2522 
2523 	if (!data || !table->maxlen || !*lenp ||
2524 	    (*ppos && !write)) {
2525 		*lenp = 0;
2526 		return 0;
2527 	}
2528 
2529 	i = (unsigned long *) data;
2530 	min = (unsigned long *) table->extra1;
2531 	max = (unsigned long *) table->extra2;
2532 	vleft = table->maxlen / sizeof(unsigned long);
2533 	left = *lenp;
2534 
2535 	for (; left && vleft--; i++, min++, max++, first=0) {
2536 		if (write) {
2537 			while (left) {
2538 				char c;
2539 				if (get_user(c, s))
2540 					return -EFAULT;
2541 				if (!isspace(c))
2542 					break;
2543 				left--;
2544 				s++;
2545 			}
2546 			if (!left)
2547 				break;
2548 			neg = 0;
2549 			len = left;
2550 			if (len > TMPBUFLEN-1)
2551 				len = TMPBUFLEN-1;
2552 			if (copy_from_user(buf, s, len))
2553 				return -EFAULT;
2554 			buf[len] = 0;
2555 			p = buf;
2556 			if (*p == '-' && left > 1) {
2557 				neg = 1;
2558 				p++;
2559 			}
2560 			if (*p < '0' || *p > '9')
2561 				break;
2562 			val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2563 			len = p-buf;
2564 			if ((len < left) && *p && !isspace(*p))
2565 				break;
2566 			if (neg)
2567 				val = -val;
2568 			s += len;
2569 			left -= len;
2570 
2571 			if(neg)
2572 				continue;
2573 			if ((min && val < *min) || (max && val > *max))
2574 				continue;
2575 			*i = val;
2576 		} else {
2577 			p = buf;
2578 			if (!first)
2579 				*p++ = '\t';
2580 			sprintf(p, "%lu", convdiv * (*i) / convmul);
2581 			len = strlen(buf);
2582 			if (len > left)
2583 				len = left;
2584 			if(copy_to_user(s, buf, len))
2585 				return -EFAULT;
2586 			left -= len;
2587 			s += len;
2588 		}
2589 	}
2590 
2591 	if (!write && !first && left) {
2592 		if(put_user('\n', s))
2593 			return -EFAULT;
2594 		left--, s++;
2595 	}
2596 	if (write) {
2597 		while (left) {
2598 			char c;
2599 			if (get_user(c, s++))
2600 				return -EFAULT;
2601 			if (!isspace(c))
2602 				break;
2603 			left--;
2604 		}
2605 	}
2606 	if (write && first)
2607 		return -EINVAL;
2608 	*lenp -= left;
2609 	*ppos += *lenp;
2610 	return 0;
2611 #undef TMPBUFLEN
2612 }
2613 
2614 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2615 				     struct file *filp,
2616 				     void __user *buffer,
2617 				     size_t *lenp, loff_t *ppos,
2618 				     unsigned long convmul,
2619 				     unsigned long convdiv)
2620 {
2621 	return __do_proc_doulongvec_minmax(table->data, table, write,
2622 			filp, buffer, lenp, ppos, convmul, convdiv);
2623 }
2624 
2625 /**
2626  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2627  * @table: the sysctl table
2628  * @write: %TRUE if this is a write to the sysctl file
2629  * @filp: the file structure
2630  * @buffer: the user buffer
2631  * @lenp: the size of the user buffer
2632  * @ppos: file position
2633  *
2634  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2635  * values from/to the user buffer, treated as an ASCII string.
2636  *
2637  * This routine will ensure the values are within the range specified by
2638  * table->extra1 (min) and table->extra2 (max).
2639  *
2640  * Returns 0 on success.
2641  */
2642 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2643 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2644 {
2645     return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2646 }
2647 
2648 /**
2649  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2650  * @table: the sysctl table
2651  * @write: %TRUE if this is a write to the sysctl file
2652  * @filp: the file structure
2653  * @buffer: the user buffer
2654  * @lenp: the size of the user buffer
2655  * @ppos: file position
2656  *
2657  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2658  * values from/to the user buffer, treated as an ASCII string. The values
2659  * are treated as milliseconds, and converted to jiffies when they are stored.
2660  *
2661  * This routine will ensure the values are within the range specified by
2662  * table->extra1 (min) and table->extra2 (max).
2663  *
2664  * Returns 0 on success.
2665  */
2666 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2667 				      struct file *filp,
2668 				      void __user *buffer,
2669 				      size_t *lenp, loff_t *ppos)
2670 {
2671     return do_proc_doulongvec_minmax(table, write, filp, buffer,
2672 				     lenp, ppos, HZ, 1000l);
2673 }
2674 
2675 
2676 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2677 					 int *valp,
2678 					 int write, void *data)
2679 {
2680 	if (write) {
2681 		if (*lvalp > LONG_MAX / HZ)
2682 			return 1;
2683 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2684 	} else {
2685 		int val = *valp;
2686 		unsigned long lval;
2687 		if (val < 0) {
2688 			*negp = -1;
2689 			lval = (unsigned long)-val;
2690 		} else {
2691 			*negp = 0;
2692 			lval = (unsigned long)val;
2693 		}
2694 		*lvalp = lval / HZ;
2695 	}
2696 	return 0;
2697 }
2698 
2699 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2700 						int *valp,
2701 						int write, void *data)
2702 {
2703 	if (write) {
2704 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2705 			return 1;
2706 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2707 	} else {
2708 		int val = *valp;
2709 		unsigned long lval;
2710 		if (val < 0) {
2711 			*negp = -1;
2712 			lval = (unsigned long)-val;
2713 		} else {
2714 			*negp = 0;
2715 			lval = (unsigned long)val;
2716 		}
2717 		*lvalp = jiffies_to_clock_t(lval);
2718 	}
2719 	return 0;
2720 }
2721 
2722 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2723 					    int *valp,
2724 					    int write, void *data)
2725 {
2726 	if (write) {
2727 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2728 	} else {
2729 		int val = *valp;
2730 		unsigned long lval;
2731 		if (val < 0) {
2732 			*negp = -1;
2733 			lval = (unsigned long)-val;
2734 		} else {
2735 			*negp = 0;
2736 			lval = (unsigned long)val;
2737 		}
2738 		*lvalp = jiffies_to_msecs(lval);
2739 	}
2740 	return 0;
2741 }
2742 
2743 /**
2744  * proc_dointvec_jiffies - read a vector of integers as seconds
2745  * @table: the sysctl table
2746  * @write: %TRUE if this is a write to the sysctl file
2747  * @filp: the file structure
2748  * @buffer: the user buffer
2749  * @lenp: the size of the user buffer
2750  * @ppos: file position
2751  *
2752  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2753  * values from/to the user buffer, treated as an ASCII string.
2754  * The values read are assumed to be in seconds, and are converted into
2755  * jiffies.
2756  *
2757  * Returns 0 on success.
2758  */
2759 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2760 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2761 {
2762     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2763 		    	    do_proc_dointvec_jiffies_conv,NULL);
2764 }
2765 
2766 /**
2767  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2768  * @table: the sysctl table
2769  * @write: %TRUE if this is a write to the sysctl file
2770  * @filp: the file structure
2771  * @buffer: the user buffer
2772  * @lenp: the size of the user buffer
2773  * @ppos: pointer to the file position
2774  *
2775  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2776  * values from/to the user buffer, treated as an ASCII string.
2777  * The values read are assumed to be in 1/USER_HZ seconds, and
2778  * are converted into jiffies.
2779  *
2780  * Returns 0 on success.
2781  */
2782 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2783 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2784 {
2785     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2786 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2787 }
2788 
2789 /**
2790  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2791  * @table: the sysctl table
2792  * @write: %TRUE if this is a write to the sysctl file
2793  * @filp: the file structure
2794  * @buffer: the user buffer
2795  * @lenp: the size of the user buffer
2796  * @ppos: file position
2797  * @ppos: the current position in the file
2798  *
2799  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2800  * values from/to the user buffer, treated as an ASCII string.
2801  * The values read are assumed to be in 1/1000 seconds, and
2802  * are converted into jiffies.
2803  *
2804  * Returns 0 on success.
2805  */
2806 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2807 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2808 {
2809 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2810 				do_proc_dointvec_ms_jiffies_conv, NULL);
2811 }
2812 
2813 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2814 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2815 {
2816 	struct pid *new_pid;
2817 	pid_t tmp;
2818 	int r;
2819 
2820 	tmp = pid_vnr(cad_pid);
2821 
2822 	r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2823 			       lenp, ppos, NULL, NULL);
2824 	if (r || !write)
2825 		return r;
2826 
2827 	new_pid = find_get_pid(tmp);
2828 	if (!new_pid)
2829 		return -ESRCH;
2830 
2831 	put_pid(xchg(&cad_pid, new_pid));
2832 	return 0;
2833 }
2834 
2835 #else /* CONFIG_PROC_FS */
2836 
2837 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2838 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2839 {
2840 	return -ENOSYS;
2841 }
2842 
2843 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2844 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2845 {
2846 	return -ENOSYS;
2847 }
2848 
2849 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2850 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2851 {
2852 	return -ENOSYS;
2853 }
2854 
2855 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2856 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2857 {
2858 	return -ENOSYS;
2859 }
2860 
2861 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2862 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2863 {
2864 	return -ENOSYS;
2865 }
2866 
2867 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2868 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2869 {
2870 	return -ENOSYS;
2871 }
2872 
2873 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2874 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2875 {
2876 	return -ENOSYS;
2877 }
2878 
2879 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2880 				      struct file *filp,
2881 				      void __user *buffer,
2882 				      size_t *lenp, loff_t *ppos)
2883 {
2884     return -ENOSYS;
2885 }
2886 
2887 
2888 #endif /* CONFIG_PROC_FS */
2889 
2890 
2891 #ifdef CONFIG_SYSCTL_SYSCALL
2892 /*
2893  * General sysctl support routines
2894  */
2895 
2896 /* The generic sysctl data routine (used if no strategy routine supplied) */
2897 int sysctl_data(struct ctl_table *table,
2898 		void __user *oldval, size_t __user *oldlenp,
2899 		void __user *newval, size_t newlen)
2900 {
2901 	size_t len;
2902 
2903 	/* Get out of I don't have a variable */
2904 	if (!table->data || !table->maxlen)
2905 		return -ENOTDIR;
2906 
2907 	if (oldval && oldlenp) {
2908 		if (get_user(len, oldlenp))
2909 			return -EFAULT;
2910 		if (len) {
2911 			if (len > table->maxlen)
2912 				len = table->maxlen;
2913 			if (copy_to_user(oldval, table->data, len))
2914 				return -EFAULT;
2915 			if (put_user(len, oldlenp))
2916 				return -EFAULT;
2917 		}
2918 	}
2919 
2920 	if (newval && newlen) {
2921 		if (newlen > table->maxlen)
2922 			newlen = table->maxlen;
2923 
2924 		if (copy_from_user(table->data, newval, newlen))
2925 			return -EFAULT;
2926 	}
2927 	return 1;
2928 }
2929 
2930 /* The generic string strategy routine: */
2931 int sysctl_string(struct ctl_table *table,
2932 		  void __user *oldval, size_t __user *oldlenp,
2933 		  void __user *newval, size_t newlen)
2934 {
2935 	if (!table->data || !table->maxlen)
2936 		return -ENOTDIR;
2937 
2938 	if (oldval && oldlenp) {
2939 		size_t bufsize;
2940 		if (get_user(bufsize, oldlenp))
2941 			return -EFAULT;
2942 		if (bufsize) {
2943 			size_t len = strlen(table->data), copied;
2944 
2945 			/* This shouldn't trigger for a well-formed sysctl */
2946 			if (len > table->maxlen)
2947 				len = table->maxlen;
2948 
2949 			/* Copy up to a max of bufsize-1 bytes of the string */
2950 			copied = (len >= bufsize) ? bufsize - 1 : len;
2951 
2952 			if (copy_to_user(oldval, table->data, copied) ||
2953 			    put_user(0, (char __user *)(oldval + copied)))
2954 				return -EFAULT;
2955 			if (put_user(len, oldlenp))
2956 				return -EFAULT;
2957 		}
2958 	}
2959 	if (newval && newlen) {
2960 		size_t len = newlen;
2961 		if (len > table->maxlen)
2962 			len = table->maxlen;
2963 		if(copy_from_user(table->data, newval, len))
2964 			return -EFAULT;
2965 		if (len == table->maxlen)
2966 			len--;
2967 		((char *) table->data)[len] = 0;
2968 	}
2969 	return 1;
2970 }
2971 
2972 /*
2973  * This function makes sure that all of the integers in the vector
2974  * are between the minimum and maximum values given in the arrays
2975  * table->extra1 and table->extra2, respectively.
2976  */
2977 int sysctl_intvec(struct ctl_table *table,
2978 		void __user *oldval, size_t __user *oldlenp,
2979 		void __user *newval, size_t newlen)
2980 {
2981 
2982 	if (newval && newlen) {
2983 		int __user *vec = (int __user *) newval;
2984 		int *min = (int *) table->extra1;
2985 		int *max = (int *) table->extra2;
2986 		size_t length;
2987 		int i;
2988 
2989 		if (newlen % sizeof(int) != 0)
2990 			return -EINVAL;
2991 
2992 		if (!table->extra1 && !table->extra2)
2993 			return 0;
2994 
2995 		if (newlen > table->maxlen)
2996 			newlen = table->maxlen;
2997 		length = newlen / sizeof(int);
2998 
2999 		for (i = 0; i < length; i++) {
3000 			int value;
3001 			if (get_user(value, vec + i))
3002 				return -EFAULT;
3003 			if (min && value < min[i])
3004 				return -EINVAL;
3005 			if (max && value > max[i])
3006 				return -EINVAL;
3007 		}
3008 	}
3009 	return 0;
3010 }
3011 
3012 /* Strategy function to convert jiffies to seconds */
3013 int sysctl_jiffies(struct ctl_table *table,
3014 		void __user *oldval, size_t __user *oldlenp,
3015 		void __user *newval, size_t newlen)
3016 {
3017 	if (oldval && oldlenp) {
3018 		size_t olen;
3019 
3020 		if (get_user(olen, oldlenp))
3021 			return -EFAULT;
3022 		if (olen) {
3023 			int val;
3024 
3025 			if (olen < sizeof(int))
3026 				return -EINVAL;
3027 
3028 			val = *(int *)(table->data) / HZ;
3029 			if (put_user(val, (int __user *)oldval))
3030 				return -EFAULT;
3031 			if (put_user(sizeof(int), oldlenp))
3032 				return -EFAULT;
3033 		}
3034 	}
3035 	if (newval && newlen) {
3036 		int new;
3037 		if (newlen != sizeof(int))
3038 			return -EINVAL;
3039 		if (get_user(new, (int __user *)newval))
3040 			return -EFAULT;
3041 		*(int *)(table->data) = new*HZ;
3042 	}
3043 	return 1;
3044 }
3045 
3046 /* Strategy function to convert jiffies to seconds */
3047 int sysctl_ms_jiffies(struct ctl_table *table,
3048 		void __user *oldval, size_t __user *oldlenp,
3049 		void __user *newval, size_t newlen)
3050 {
3051 	if (oldval && oldlenp) {
3052 		size_t olen;
3053 
3054 		if (get_user(olen, oldlenp))
3055 			return -EFAULT;
3056 		if (olen) {
3057 			int val;
3058 
3059 			if (olen < sizeof(int))
3060 				return -EINVAL;
3061 
3062 			val = jiffies_to_msecs(*(int *)(table->data));
3063 			if (put_user(val, (int __user *)oldval))
3064 				return -EFAULT;
3065 			if (put_user(sizeof(int), oldlenp))
3066 				return -EFAULT;
3067 		}
3068 	}
3069 	if (newval && newlen) {
3070 		int new;
3071 		if (newlen != sizeof(int))
3072 			return -EINVAL;
3073 		if (get_user(new, (int __user *)newval))
3074 			return -EFAULT;
3075 		*(int *)(table->data) = msecs_to_jiffies(new);
3076 	}
3077 	return 1;
3078 }
3079 
3080 
3081 
3082 #else /* CONFIG_SYSCTL_SYSCALL */
3083 
3084 
3085 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
3086 {
3087 	struct __sysctl_args tmp;
3088 	int error;
3089 
3090 	if (copy_from_user(&tmp, args, sizeof(tmp)))
3091 		return -EFAULT;
3092 
3093 	error = deprecated_sysctl_warning(&tmp);
3094 
3095 	/* If no error reading the parameters then just -ENOSYS ... */
3096 	if (!error)
3097 		error = -ENOSYS;
3098 
3099 	return error;
3100 }
3101 
3102 int sysctl_data(struct ctl_table *table,
3103 		  void __user *oldval, size_t __user *oldlenp,
3104 		  void __user *newval, size_t newlen)
3105 {
3106 	return -ENOSYS;
3107 }
3108 
3109 int sysctl_string(struct ctl_table *table,
3110 		  void __user *oldval, size_t __user *oldlenp,
3111 		  void __user *newval, size_t newlen)
3112 {
3113 	return -ENOSYS;
3114 }
3115 
3116 int sysctl_intvec(struct ctl_table *table,
3117 		void __user *oldval, size_t __user *oldlenp,
3118 		void __user *newval, size_t newlen)
3119 {
3120 	return -ENOSYS;
3121 }
3122 
3123 int sysctl_jiffies(struct ctl_table *table,
3124 		void __user *oldval, size_t __user *oldlenp,
3125 		void __user *newval, size_t newlen)
3126 {
3127 	return -ENOSYS;
3128 }
3129 
3130 int sysctl_ms_jiffies(struct ctl_table *table,
3131 		void __user *oldval, size_t __user *oldlenp,
3132 		void __user *newval, size_t newlen)
3133 {
3134 	return -ENOSYS;
3135 }
3136 
3137 #endif /* CONFIG_SYSCTL_SYSCALL */
3138 
3139 static int deprecated_sysctl_warning(struct __sysctl_args *args)
3140 {
3141 	static int msg_count;
3142 	int name[CTL_MAXNAME];
3143 	int i;
3144 
3145 	/* Check args->nlen. */
3146 	if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
3147 		return -ENOTDIR;
3148 
3149 	/* Read in the sysctl name for better debug message logging */
3150 	for (i = 0; i < args->nlen; i++)
3151 		if (get_user(name[i], args->name + i))
3152 			return -EFAULT;
3153 
3154 	/* Ignore accesses to kernel.version */
3155 	if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
3156 		return 0;
3157 
3158 	if (msg_count < 5) {
3159 		msg_count++;
3160 		printk(KERN_INFO
3161 			"warning: process `%s' used the deprecated sysctl "
3162 			"system call with ", current->comm);
3163 		for (i = 0; i < args->nlen; i++)
3164 			printk("%d.", name[i]);
3165 		printk("\n");
3166 	}
3167 	return 0;
3168 }
3169 
3170 /*
3171  * No sense putting this after each symbol definition, twice,
3172  * exception granted :-)
3173  */
3174 EXPORT_SYMBOL(proc_dointvec);
3175 EXPORT_SYMBOL(proc_dointvec_jiffies);
3176 EXPORT_SYMBOL(proc_dointvec_minmax);
3177 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3178 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3179 EXPORT_SYMBOL(proc_dostring);
3180 EXPORT_SYMBOL(proc_doulongvec_minmax);
3181 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3182 EXPORT_SYMBOL(register_sysctl_table);
3183 EXPORT_SYMBOL(register_sysctl_paths);
3184 EXPORT_SYMBOL(sysctl_intvec);
3185 EXPORT_SYMBOL(sysctl_jiffies);
3186 EXPORT_SYMBOL(sysctl_ms_jiffies);
3187 EXPORT_SYMBOL(sysctl_string);
3188 EXPORT_SYMBOL(sysctl_data);
3189 EXPORT_SYMBOL(unregister_sysctl_table);
3190