xref: /openbmc/linux/kernel/sysctl.c (revision 5211a242d0cbdded372aee59da18f80552b0a80a)
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	= CTL_UNNUMBERED,
748 		.procname	= "panic_on_io_nmi",
749 		.data		= &panic_on_io_nmi,
750 		.maxlen		= sizeof(int),
751 		.mode		= 0644,
752 		.proc_handler	= &proc_dointvec,
753 	},
754 	{
755 		.ctl_name	= KERN_BOOTLOADER_TYPE,
756 		.procname	= "bootloader_type",
757 		.data		= &bootloader_type,
758 		.maxlen		= sizeof (int),
759 		.mode		= 0444,
760 		.proc_handler	= &proc_dointvec,
761 	},
762 	{
763 		.ctl_name	= CTL_UNNUMBERED,
764 		.procname	= "bootloader_version",
765 		.data		= &bootloader_version,
766 		.maxlen		= sizeof (int),
767 		.mode		= 0444,
768 		.proc_handler	= &proc_dointvec,
769 	},
770 	{
771 		.ctl_name	= CTL_UNNUMBERED,
772 		.procname	= "kstack_depth_to_print",
773 		.data		= &kstack_depth_to_print,
774 		.maxlen		= sizeof(int),
775 		.mode		= 0644,
776 		.proc_handler	= &proc_dointvec,
777 	},
778 	{
779 		.ctl_name	= CTL_UNNUMBERED,
780 		.procname	= "io_delay_type",
781 		.data		= &io_delay_type,
782 		.maxlen		= sizeof(int),
783 		.mode		= 0644,
784 		.proc_handler	= &proc_dointvec,
785 	},
786 #endif
787 #if defined(CONFIG_MMU)
788 	{
789 		.ctl_name	= KERN_RANDOMIZE,
790 		.procname	= "randomize_va_space",
791 		.data		= &randomize_va_space,
792 		.maxlen		= sizeof(int),
793 		.mode		= 0644,
794 		.proc_handler	= &proc_dointvec,
795 	},
796 #endif
797 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
798 	{
799 		.ctl_name	= KERN_SPIN_RETRY,
800 		.procname	= "spin_retry",
801 		.data		= &spin_retry,
802 		.maxlen		= sizeof (int),
803 		.mode		= 0644,
804 		.proc_handler	= &proc_dointvec,
805 	},
806 #endif
807 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
808 	{
809 		.procname	= "acpi_video_flags",
810 		.data		= &acpi_realmode_flags,
811 		.maxlen		= sizeof (unsigned long),
812 		.mode		= 0644,
813 		.proc_handler	= &proc_doulongvec_minmax,
814 	},
815 #endif
816 #ifdef CONFIG_IA64
817 	{
818 		.ctl_name	= KERN_IA64_UNALIGNED,
819 		.procname	= "ignore-unaligned-usertrap",
820 		.data		= &no_unaligned_warning,
821 		.maxlen		= sizeof (int),
822 	 	.mode		= 0644,
823 		.proc_handler	= &proc_dointvec,
824 	},
825 	{
826 		.ctl_name	= CTL_UNNUMBERED,
827 		.procname	= "unaligned-dump-stack",
828 		.data		= &unaligned_dump_stack,
829 		.maxlen		= sizeof (int),
830 		.mode		= 0644,
831 		.proc_handler	= &proc_dointvec,
832 	},
833 #endif
834 #ifdef CONFIG_DETECT_SOFTLOCKUP
835 	{
836 		.ctl_name	= CTL_UNNUMBERED,
837 		.procname	= "softlockup_panic",
838 		.data		= &softlockup_panic,
839 		.maxlen		= sizeof(int),
840 		.mode		= 0644,
841 		.proc_handler	= &proc_dointvec_minmax,
842 		.strategy	= &sysctl_intvec,
843 		.extra1		= &zero,
844 		.extra2		= &one,
845 	},
846 	{
847 		.ctl_name	= CTL_UNNUMBERED,
848 		.procname	= "softlockup_thresh",
849 		.data		= &softlockup_thresh,
850 		.maxlen		= sizeof(int),
851 		.mode		= 0644,
852 		.proc_handler	= &proc_dosoftlockup_thresh,
853 		.strategy	= &sysctl_intvec,
854 		.extra1		= &neg_one,
855 		.extra2		= &sixty,
856 	},
857 #endif
858 #ifdef CONFIG_DETECT_HUNG_TASK
859 	{
860 		.ctl_name	= CTL_UNNUMBERED,
861 		.procname	= "hung_task_panic",
862 		.data		= &sysctl_hung_task_panic,
863 		.maxlen		= sizeof(int),
864 		.mode		= 0644,
865 		.proc_handler	= &proc_dointvec_minmax,
866 		.strategy	= &sysctl_intvec,
867 		.extra1		= &zero,
868 		.extra2		= &one,
869 	},
870 	{
871 		.ctl_name	= CTL_UNNUMBERED,
872 		.procname	= "hung_task_check_count",
873 		.data		= &sysctl_hung_task_check_count,
874 		.maxlen		= sizeof(unsigned long),
875 		.mode		= 0644,
876 		.proc_handler	= &proc_doulongvec_minmax,
877 		.strategy	= &sysctl_intvec,
878 	},
879 	{
880 		.ctl_name	= CTL_UNNUMBERED,
881 		.procname	= "hung_task_timeout_secs",
882 		.data		= &sysctl_hung_task_timeout_secs,
883 		.maxlen		= sizeof(unsigned long),
884 		.mode		= 0644,
885 		.proc_handler	= &proc_dohung_task_timeout_secs,
886 		.strategy	= &sysctl_intvec,
887 	},
888 	{
889 		.ctl_name	= CTL_UNNUMBERED,
890 		.procname	= "hung_task_warnings",
891 		.data		= &sysctl_hung_task_warnings,
892 		.maxlen		= sizeof(unsigned long),
893 		.mode		= 0644,
894 		.proc_handler	= &proc_doulongvec_minmax,
895 		.strategy	= &sysctl_intvec,
896 	},
897 #endif
898 #ifdef CONFIG_COMPAT
899 	{
900 		.ctl_name	= KERN_COMPAT_LOG,
901 		.procname	= "compat-log",
902 		.data		= &compat_log,
903 		.maxlen		= sizeof (int),
904 	 	.mode		= 0644,
905 		.proc_handler	= &proc_dointvec,
906 	},
907 #endif
908 #ifdef CONFIG_RT_MUTEXES
909 	{
910 		.ctl_name	= KERN_MAX_LOCK_DEPTH,
911 		.procname	= "max_lock_depth",
912 		.data		= &max_lock_depth,
913 		.maxlen		= sizeof(int),
914 		.mode		= 0644,
915 		.proc_handler	= &proc_dointvec,
916 	},
917 #endif
918 	{
919 		.ctl_name	= CTL_UNNUMBERED,
920 		.procname	= "poweroff_cmd",
921 		.data		= &poweroff_cmd,
922 		.maxlen		= POWEROFF_CMD_PATH_LEN,
923 		.mode		= 0644,
924 		.proc_handler	= &proc_dostring,
925 		.strategy	= &sysctl_string,
926 	},
927 #ifdef CONFIG_KEYS
928 	{
929 		.ctl_name	= CTL_UNNUMBERED,
930 		.procname	= "keys",
931 		.mode		= 0555,
932 		.child		= key_sysctls,
933 	},
934 #endif
935 #ifdef CONFIG_RCU_TORTURE_TEST
936 	{
937 		.ctl_name       = CTL_UNNUMBERED,
938 		.procname       = "rcutorture_runnable",
939 		.data           = &rcutorture_runnable,
940 		.maxlen         = sizeof(int),
941 		.mode           = 0644,
942 		.proc_handler   = &proc_dointvec,
943 	},
944 #endif
945 #ifdef CONFIG_SLOW_WORK
946 	{
947 		.ctl_name	= CTL_UNNUMBERED,
948 		.procname	= "slow-work",
949 		.mode		= 0555,
950 		.child		= slow_work_sysctls,
951 	},
952 #endif
953 #ifdef CONFIG_PERF_COUNTERS
954 	{
955 		.ctl_name	= CTL_UNNUMBERED,
956 		.procname	= "perf_counter_paranoid",
957 		.data		= &sysctl_perf_counter_paranoid,
958 		.maxlen		= sizeof(sysctl_perf_counter_paranoid),
959 		.mode		= 0644,
960 		.proc_handler	= &proc_dointvec,
961 	},
962 	{
963 		.ctl_name	= CTL_UNNUMBERED,
964 		.procname	= "perf_counter_mlock_kb",
965 		.data		= &sysctl_perf_counter_mlock,
966 		.maxlen		= sizeof(sysctl_perf_counter_mlock),
967 		.mode		= 0644,
968 		.proc_handler	= &proc_dointvec,
969 	},
970 	{
971 		.ctl_name	= CTL_UNNUMBERED,
972 		.procname	= "perf_counter_max_sample_rate",
973 		.data		= &sysctl_perf_counter_sample_rate,
974 		.maxlen		= sizeof(sysctl_perf_counter_sample_rate),
975 		.mode		= 0644,
976 		.proc_handler	= &proc_dointvec,
977 	},
978 #endif
979 #ifdef CONFIG_KMEMCHECK
980 	{
981 		.ctl_name	= CTL_UNNUMBERED,
982 		.procname	= "kmemcheck",
983 		.data		= &kmemcheck_enabled,
984 		.maxlen		= sizeof(int),
985 		.mode		= 0644,
986 		.proc_handler	= &proc_dointvec,
987 	},
988 #endif
989 
990 /*
991  * NOTE: do not add new entries to this table unless you have read
992  * Documentation/sysctl/ctl_unnumbered.txt
993  */
994 	{ .ctl_name = 0 }
995 };
996 
997 static struct ctl_table vm_table[] = {
998 	{
999 		.ctl_name	= VM_OVERCOMMIT_MEMORY,
1000 		.procname	= "overcommit_memory",
1001 		.data		= &sysctl_overcommit_memory,
1002 		.maxlen		= sizeof(sysctl_overcommit_memory),
1003 		.mode		= 0644,
1004 		.proc_handler	= &proc_dointvec,
1005 	},
1006 	{
1007 		.ctl_name	= VM_PANIC_ON_OOM,
1008 		.procname	= "panic_on_oom",
1009 		.data		= &sysctl_panic_on_oom,
1010 		.maxlen		= sizeof(sysctl_panic_on_oom),
1011 		.mode		= 0644,
1012 		.proc_handler	= &proc_dointvec,
1013 	},
1014 	{
1015 		.ctl_name	= CTL_UNNUMBERED,
1016 		.procname	= "oom_kill_allocating_task",
1017 		.data		= &sysctl_oom_kill_allocating_task,
1018 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1019 		.mode		= 0644,
1020 		.proc_handler	= &proc_dointvec,
1021 	},
1022 	{
1023 		.ctl_name	= CTL_UNNUMBERED,
1024 		.procname	= "oom_dump_tasks",
1025 		.data		= &sysctl_oom_dump_tasks,
1026 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1027 		.mode		= 0644,
1028 		.proc_handler	= &proc_dointvec,
1029 	},
1030 	{
1031 		.ctl_name	= VM_OVERCOMMIT_RATIO,
1032 		.procname	= "overcommit_ratio",
1033 		.data		= &sysctl_overcommit_ratio,
1034 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1035 		.mode		= 0644,
1036 		.proc_handler	= &proc_dointvec,
1037 	},
1038 	{
1039 		.ctl_name	= VM_PAGE_CLUSTER,
1040 		.procname	= "page-cluster",
1041 		.data		= &page_cluster,
1042 		.maxlen		= sizeof(int),
1043 		.mode		= 0644,
1044 		.proc_handler	= &proc_dointvec,
1045 	},
1046 	{
1047 		.ctl_name	= VM_DIRTY_BACKGROUND,
1048 		.procname	= "dirty_background_ratio",
1049 		.data		= &dirty_background_ratio,
1050 		.maxlen		= sizeof(dirty_background_ratio),
1051 		.mode		= 0644,
1052 		.proc_handler	= &dirty_background_ratio_handler,
1053 		.strategy	= &sysctl_intvec,
1054 		.extra1		= &zero,
1055 		.extra2		= &one_hundred,
1056 	},
1057 	{
1058 		.ctl_name	= CTL_UNNUMBERED,
1059 		.procname	= "dirty_background_bytes",
1060 		.data		= &dirty_background_bytes,
1061 		.maxlen		= sizeof(dirty_background_bytes),
1062 		.mode		= 0644,
1063 		.proc_handler	= &dirty_background_bytes_handler,
1064 		.strategy	= &sysctl_intvec,
1065 		.extra1		= &one_ul,
1066 	},
1067 	{
1068 		.ctl_name	= VM_DIRTY_RATIO,
1069 		.procname	= "dirty_ratio",
1070 		.data		= &vm_dirty_ratio,
1071 		.maxlen		= sizeof(vm_dirty_ratio),
1072 		.mode		= 0644,
1073 		.proc_handler	= &dirty_ratio_handler,
1074 		.strategy	= &sysctl_intvec,
1075 		.extra1		= &zero,
1076 		.extra2		= &one_hundred,
1077 	},
1078 	{
1079 		.ctl_name	= CTL_UNNUMBERED,
1080 		.procname	= "dirty_bytes",
1081 		.data		= &vm_dirty_bytes,
1082 		.maxlen		= sizeof(vm_dirty_bytes),
1083 		.mode		= 0644,
1084 		.proc_handler	= &dirty_bytes_handler,
1085 		.strategy	= &sysctl_intvec,
1086 		.extra1		= &dirty_bytes_min,
1087 	},
1088 	{
1089 		.procname	= "dirty_writeback_centisecs",
1090 		.data		= &dirty_writeback_interval,
1091 		.maxlen		= sizeof(dirty_writeback_interval),
1092 		.mode		= 0644,
1093 		.proc_handler	= &dirty_writeback_centisecs_handler,
1094 	},
1095 	{
1096 		.procname	= "dirty_expire_centisecs",
1097 		.data		= &dirty_expire_interval,
1098 		.maxlen		= sizeof(dirty_expire_interval),
1099 		.mode		= 0644,
1100 		.proc_handler	= &proc_dointvec,
1101 	},
1102 	{
1103 		.ctl_name	= VM_NR_PDFLUSH_THREADS,
1104 		.procname	= "nr_pdflush_threads",
1105 		.data		= &nr_pdflush_threads,
1106 		.maxlen		= sizeof nr_pdflush_threads,
1107 		.mode		= 0444 /* read-only*/,
1108 		.proc_handler	= &proc_dointvec,
1109 	},
1110 	{
1111 		.ctl_name	= VM_SWAPPINESS,
1112 		.procname	= "swappiness",
1113 		.data		= &vm_swappiness,
1114 		.maxlen		= sizeof(vm_swappiness),
1115 		.mode		= 0644,
1116 		.proc_handler	= &proc_dointvec_minmax,
1117 		.strategy	= &sysctl_intvec,
1118 		.extra1		= &zero,
1119 		.extra2		= &one_hundred,
1120 	},
1121 #ifdef CONFIG_HUGETLB_PAGE
1122 	 {
1123 		.procname	= "nr_hugepages",
1124 		.data		= NULL,
1125 		.maxlen		= sizeof(unsigned long),
1126 		.mode		= 0644,
1127 		.proc_handler	= &hugetlb_sysctl_handler,
1128 		.extra1		= (void *)&hugetlb_zero,
1129 		.extra2		= (void *)&hugetlb_infinity,
1130 	 },
1131 	 {
1132 		.ctl_name	= VM_HUGETLB_GROUP,
1133 		.procname	= "hugetlb_shm_group",
1134 		.data		= &sysctl_hugetlb_shm_group,
1135 		.maxlen		= sizeof(gid_t),
1136 		.mode		= 0644,
1137 		.proc_handler	= &proc_dointvec,
1138 	 },
1139 	 {
1140 		.ctl_name	= CTL_UNNUMBERED,
1141 		.procname	= "hugepages_treat_as_movable",
1142 		.data		= &hugepages_treat_as_movable,
1143 		.maxlen		= sizeof(int),
1144 		.mode		= 0644,
1145 		.proc_handler	= &hugetlb_treat_movable_handler,
1146 	},
1147 	{
1148 		.ctl_name	= CTL_UNNUMBERED,
1149 		.procname	= "nr_overcommit_hugepages",
1150 		.data		= NULL,
1151 		.maxlen		= sizeof(unsigned long),
1152 		.mode		= 0644,
1153 		.proc_handler	= &hugetlb_overcommit_handler,
1154 		.extra1		= (void *)&hugetlb_zero,
1155 		.extra2		= (void *)&hugetlb_infinity,
1156 	},
1157 #endif
1158 	{
1159 		.ctl_name	= VM_LOWMEM_RESERVE_RATIO,
1160 		.procname	= "lowmem_reserve_ratio",
1161 		.data		= &sysctl_lowmem_reserve_ratio,
1162 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1163 		.mode		= 0644,
1164 		.proc_handler	= &lowmem_reserve_ratio_sysctl_handler,
1165 		.strategy	= &sysctl_intvec,
1166 	},
1167 	{
1168 		.ctl_name	= VM_DROP_PAGECACHE,
1169 		.procname	= "drop_caches",
1170 		.data		= &sysctl_drop_caches,
1171 		.maxlen		= sizeof(int),
1172 		.mode		= 0644,
1173 		.proc_handler	= drop_caches_sysctl_handler,
1174 		.strategy	= &sysctl_intvec,
1175 	},
1176 	{
1177 		.ctl_name	= VM_MIN_FREE_KBYTES,
1178 		.procname	= "min_free_kbytes",
1179 		.data		= &min_free_kbytes,
1180 		.maxlen		= sizeof(min_free_kbytes),
1181 		.mode		= 0644,
1182 		.proc_handler	= &min_free_kbytes_sysctl_handler,
1183 		.strategy	= &sysctl_intvec,
1184 		.extra1		= &zero,
1185 	},
1186 	{
1187 		.ctl_name	= VM_PERCPU_PAGELIST_FRACTION,
1188 		.procname	= "percpu_pagelist_fraction",
1189 		.data		= &percpu_pagelist_fraction,
1190 		.maxlen		= sizeof(percpu_pagelist_fraction),
1191 		.mode		= 0644,
1192 		.proc_handler	= &percpu_pagelist_fraction_sysctl_handler,
1193 		.strategy	= &sysctl_intvec,
1194 		.extra1		= &min_percpu_pagelist_fract,
1195 	},
1196 #ifdef CONFIG_MMU
1197 	{
1198 		.ctl_name	= VM_MAX_MAP_COUNT,
1199 		.procname	= "max_map_count",
1200 		.data		= &sysctl_max_map_count,
1201 		.maxlen		= sizeof(sysctl_max_map_count),
1202 		.mode		= 0644,
1203 		.proc_handler	= &proc_dointvec
1204 	},
1205 #else
1206 	{
1207 		.ctl_name	= CTL_UNNUMBERED,
1208 		.procname	= "nr_trim_pages",
1209 		.data		= &sysctl_nr_trim_pages,
1210 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1211 		.mode		= 0644,
1212 		.proc_handler	= &proc_dointvec_minmax,
1213 		.strategy	= &sysctl_intvec,
1214 		.extra1		= &zero,
1215 	},
1216 #endif
1217 	{
1218 		.ctl_name	= VM_LAPTOP_MODE,
1219 		.procname	= "laptop_mode",
1220 		.data		= &laptop_mode,
1221 		.maxlen		= sizeof(laptop_mode),
1222 		.mode		= 0644,
1223 		.proc_handler	= &proc_dointvec_jiffies,
1224 		.strategy	= &sysctl_jiffies,
1225 	},
1226 	{
1227 		.ctl_name	= VM_BLOCK_DUMP,
1228 		.procname	= "block_dump",
1229 		.data		= &block_dump,
1230 		.maxlen		= sizeof(block_dump),
1231 		.mode		= 0644,
1232 		.proc_handler	= &proc_dointvec,
1233 		.strategy	= &sysctl_intvec,
1234 		.extra1		= &zero,
1235 	},
1236 	{
1237 		.ctl_name	= VM_VFS_CACHE_PRESSURE,
1238 		.procname	= "vfs_cache_pressure",
1239 		.data		= &sysctl_vfs_cache_pressure,
1240 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1241 		.mode		= 0644,
1242 		.proc_handler	= &proc_dointvec,
1243 		.strategy	= &sysctl_intvec,
1244 		.extra1		= &zero,
1245 	},
1246 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1247 	{
1248 		.ctl_name	= VM_LEGACY_VA_LAYOUT,
1249 		.procname	= "legacy_va_layout",
1250 		.data		= &sysctl_legacy_va_layout,
1251 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1252 		.mode		= 0644,
1253 		.proc_handler	= &proc_dointvec,
1254 		.strategy	= &sysctl_intvec,
1255 		.extra1		= &zero,
1256 	},
1257 #endif
1258 #ifdef CONFIG_NUMA
1259 	{
1260 		.ctl_name	= VM_ZONE_RECLAIM_MODE,
1261 		.procname	= "zone_reclaim_mode",
1262 		.data		= &zone_reclaim_mode,
1263 		.maxlen		= sizeof(zone_reclaim_mode),
1264 		.mode		= 0644,
1265 		.proc_handler	= &proc_dointvec,
1266 		.strategy	= &sysctl_intvec,
1267 		.extra1		= &zero,
1268 	},
1269 	{
1270 		.ctl_name	= VM_MIN_UNMAPPED,
1271 		.procname	= "min_unmapped_ratio",
1272 		.data		= &sysctl_min_unmapped_ratio,
1273 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1274 		.mode		= 0644,
1275 		.proc_handler	= &sysctl_min_unmapped_ratio_sysctl_handler,
1276 		.strategy	= &sysctl_intvec,
1277 		.extra1		= &zero,
1278 		.extra2		= &one_hundred,
1279 	},
1280 	{
1281 		.ctl_name	= VM_MIN_SLAB,
1282 		.procname	= "min_slab_ratio",
1283 		.data		= &sysctl_min_slab_ratio,
1284 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1285 		.mode		= 0644,
1286 		.proc_handler	= &sysctl_min_slab_ratio_sysctl_handler,
1287 		.strategy	= &sysctl_intvec,
1288 		.extra1		= &zero,
1289 		.extra2		= &one_hundred,
1290 	},
1291 #endif
1292 #ifdef CONFIG_SMP
1293 	{
1294 		.ctl_name	= CTL_UNNUMBERED,
1295 		.procname	= "stat_interval",
1296 		.data		= &sysctl_stat_interval,
1297 		.maxlen		= sizeof(sysctl_stat_interval),
1298 		.mode		= 0644,
1299 		.proc_handler	= &proc_dointvec_jiffies,
1300 		.strategy	= &sysctl_jiffies,
1301 	},
1302 #endif
1303 	{
1304 		.ctl_name	= CTL_UNNUMBERED,
1305 		.procname	= "mmap_min_addr",
1306 		.data		= &mmap_min_addr,
1307 		.maxlen         = sizeof(unsigned long),
1308 		.mode		= 0644,
1309 		.proc_handler	= &proc_doulongvec_minmax,
1310 	},
1311 #ifdef CONFIG_NUMA
1312 	{
1313 		.ctl_name	= CTL_UNNUMBERED,
1314 		.procname	= "numa_zonelist_order",
1315 		.data		= &numa_zonelist_order,
1316 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1317 		.mode		= 0644,
1318 		.proc_handler	= &numa_zonelist_order_handler,
1319 		.strategy	= &sysctl_string,
1320 	},
1321 #endif
1322 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1323    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1324 	{
1325 		.ctl_name	= VM_VDSO_ENABLED,
1326 		.procname	= "vdso_enabled",
1327 		.data		= &vdso_enabled,
1328 		.maxlen		= sizeof(vdso_enabled),
1329 		.mode		= 0644,
1330 		.proc_handler	= &proc_dointvec,
1331 		.strategy	= &sysctl_intvec,
1332 		.extra1		= &zero,
1333 	},
1334 #endif
1335 #ifdef CONFIG_HIGHMEM
1336 	{
1337 		.ctl_name	= CTL_UNNUMBERED,
1338 		.procname	= "highmem_is_dirtyable",
1339 		.data		= &vm_highmem_is_dirtyable,
1340 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1341 		.mode		= 0644,
1342 		.proc_handler	= &proc_dointvec_minmax,
1343 		.strategy	= &sysctl_intvec,
1344 		.extra1		= &zero,
1345 		.extra2		= &one,
1346 	},
1347 #endif
1348 	{
1349 		.ctl_name	= CTL_UNNUMBERED,
1350 		.procname	= "scan_unevictable_pages",
1351 		.data		= &scan_unevictable_pages,
1352 		.maxlen		= sizeof(scan_unevictable_pages),
1353 		.mode		= 0644,
1354 		.proc_handler	= &scan_unevictable_handler,
1355 	},
1356 /*
1357  * NOTE: do not add new entries to this table unless you have read
1358  * Documentation/sysctl/ctl_unnumbered.txt
1359  */
1360 	{ .ctl_name = 0 }
1361 };
1362 
1363 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1364 static struct ctl_table binfmt_misc_table[] = {
1365 	{ .ctl_name = 0 }
1366 };
1367 #endif
1368 
1369 static struct ctl_table fs_table[] = {
1370 	{
1371 		.ctl_name	= FS_NRINODE,
1372 		.procname	= "inode-nr",
1373 		.data		= &inodes_stat,
1374 		.maxlen		= 2*sizeof(int),
1375 		.mode		= 0444,
1376 		.proc_handler	= &proc_dointvec,
1377 	},
1378 	{
1379 		.ctl_name	= FS_STATINODE,
1380 		.procname	= "inode-state",
1381 		.data		= &inodes_stat,
1382 		.maxlen		= 7*sizeof(int),
1383 		.mode		= 0444,
1384 		.proc_handler	= &proc_dointvec,
1385 	},
1386 	{
1387 		.procname	= "file-nr",
1388 		.data		= &files_stat,
1389 		.maxlen		= 3*sizeof(int),
1390 		.mode		= 0444,
1391 		.proc_handler	= &proc_nr_files,
1392 	},
1393 	{
1394 		.ctl_name	= FS_MAXFILE,
1395 		.procname	= "file-max",
1396 		.data		= &files_stat.max_files,
1397 		.maxlen		= sizeof(int),
1398 		.mode		= 0644,
1399 		.proc_handler	= &proc_dointvec,
1400 	},
1401 	{
1402 		.ctl_name	= CTL_UNNUMBERED,
1403 		.procname	= "nr_open",
1404 		.data		= &sysctl_nr_open,
1405 		.maxlen		= sizeof(int),
1406 		.mode		= 0644,
1407 		.proc_handler	= &proc_dointvec_minmax,
1408 		.extra1		= &sysctl_nr_open_min,
1409 		.extra2		= &sysctl_nr_open_max,
1410 	},
1411 	{
1412 		.ctl_name	= FS_DENTRY,
1413 		.procname	= "dentry-state",
1414 		.data		= &dentry_stat,
1415 		.maxlen		= 6*sizeof(int),
1416 		.mode		= 0444,
1417 		.proc_handler	= &proc_dointvec,
1418 	},
1419 	{
1420 		.ctl_name	= FS_OVERFLOWUID,
1421 		.procname	= "overflowuid",
1422 		.data		= &fs_overflowuid,
1423 		.maxlen		= sizeof(int),
1424 		.mode		= 0644,
1425 		.proc_handler	= &proc_dointvec_minmax,
1426 		.strategy	= &sysctl_intvec,
1427 		.extra1		= &minolduid,
1428 		.extra2		= &maxolduid,
1429 	},
1430 	{
1431 		.ctl_name	= FS_OVERFLOWGID,
1432 		.procname	= "overflowgid",
1433 		.data		= &fs_overflowgid,
1434 		.maxlen		= sizeof(int),
1435 		.mode		= 0644,
1436 		.proc_handler	= &proc_dointvec_minmax,
1437 		.strategy	= &sysctl_intvec,
1438 		.extra1		= &minolduid,
1439 		.extra2		= &maxolduid,
1440 	},
1441 #ifdef CONFIG_FILE_LOCKING
1442 	{
1443 		.ctl_name	= FS_LEASES,
1444 		.procname	= "leases-enable",
1445 		.data		= &leases_enable,
1446 		.maxlen		= sizeof(int),
1447 		.mode		= 0644,
1448 		.proc_handler	= &proc_dointvec,
1449 	},
1450 #endif
1451 #ifdef CONFIG_DNOTIFY
1452 	{
1453 		.ctl_name	= FS_DIR_NOTIFY,
1454 		.procname	= "dir-notify-enable",
1455 		.data		= &dir_notify_enable,
1456 		.maxlen		= sizeof(int),
1457 		.mode		= 0644,
1458 		.proc_handler	= &proc_dointvec,
1459 	},
1460 #endif
1461 #ifdef CONFIG_MMU
1462 #ifdef CONFIG_FILE_LOCKING
1463 	{
1464 		.ctl_name	= FS_LEASE_TIME,
1465 		.procname	= "lease-break-time",
1466 		.data		= &lease_break_time,
1467 		.maxlen		= sizeof(int),
1468 		.mode		= 0644,
1469 		.proc_handler	= &proc_dointvec,
1470 	},
1471 #endif
1472 #ifdef CONFIG_AIO
1473 	{
1474 		.procname	= "aio-nr",
1475 		.data		= &aio_nr,
1476 		.maxlen		= sizeof(aio_nr),
1477 		.mode		= 0444,
1478 		.proc_handler	= &proc_doulongvec_minmax,
1479 	},
1480 	{
1481 		.procname	= "aio-max-nr",
1482 		.data		= &aio_max_nr,
1483 		.maxlen		= sizeof(aio_max_nr),
1484 		.mode		= 0644,
1485 		.proc_handler	= &proc_doulongvec_minmax,
1486 	},
1487 #endif /* CONFIG_AIO */
1488 #ifdef CONFIG_INOTIFY_USER
1489 	{
1490 		.ctl_name	= FS_INOTIFY,
1491 		.procname	= "inotify",
1492 		.mode		= 0555,
1493 		.child		= inotify_table,
1494 	},
1495 #endif
1496 #ifdef CONFIG_EPOLL
1497 	{
1498 		.procname	= "epoll",
1499 		.mode		= 0555,
1500 		.child		= epoll_table,
1501 	},
1502 #endif
1503 #endif
1504 	{
1505 		.ctl_name	= KERN_SETUID_DUMPABLE,
1506 		.procname	= "suid_dumpable",
1507 		.data		= &suid_dumpable,
1508 		.maxlen		= sizeof(int),
1509 		.mode		= 0644,
1510 		.proc_handler	= &proc_dointvec_minmax,
1511 		.strategy	= &sysctl_intvec,
1512 		.extra1		= &zero,
1513 		.extra2		= &two,
1514 	},
1515 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1516 	{
1517 		.ctl_name	= CTL_UNNUMBERED,
1518 		.procname	= "binfmt_misc",
1519 		.mode		= 0555,
1520 		.child		= binfmt_misc_table,
1521 	},
1522 #endif
1523 /*
1524  * NOTE: do not add new entries to this table unless you have read
1525  * Documentation/sysctl/ctl_unnumbered.txt
1526  */
1527 	{ .ctl_name = 0 }
1528 };
1529 
1530 static struct ctl_table debug_table[] = {
1531 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1532 	{
1533 		.ctl_name	= CTL_UNNUMBERED,
1534 		.procname	= "exception-trace",
1535 		.data		= &show_unhandled_signals,
1536 		.maxlen		= sizeof(int),
1537 		.mode		= 0644,
1538 		.proc_handler	= proc_dointvec
1539 	},
1540 #endif
1541 	{ .ctl_name = 0 }
1542 };
1543 
1544 static struct ctl_table dev_table[] = {
1545 	{ .ctl_name = 0 }
1546 };
1547 
1548 static DEFINE_SPINLOCK(sysctl_lock);
1549 
1550 /* called under sysctl_lock */
1551 static int use_table(struct ctl_table_header *p)
1552 {
1553 	if (unlikely(p->unregistering))
1554 		return 0;
1555 	p->used++;
1556 	return 1;
1557 }
1558 
1559 /* called under sysctl_lock */
1560 static void unuse_table(struct ctl_table_header *p)
1561 {
1562 	if (!--p->used)
1563 		if (unlikely(p->unregistering))
1564 			complete(p->unregistering);
1565 }
1566 
1567 /* called under sysctl_lock, will reacquire if has to wait */
1568 static void start_unregistering(struct ctl_table_header *p)
1569 {
1570 	/*
1571 	 * if p->used is 0, nobody will ever touch that entry again;
1572 	 * we'll eliminate all paths to it before dropping sysctl_lock
1573 	 */
1574 	if (unlikely(p->used)) {
1575 		struct completion wait;
1576 		init_completion(&wait);
1577 		p->unregistering = &wait;
1578 		spin_unlock(&sysctl_lock);
1579 		wait_for_completion(&wait);
1580 		spin_lock(&sysctl_lock);
1581 	} else {
1582 		/* anything non-NULL; we'll never dereference it */
1583 		p->unregistering = ERR_PTR(-EINVAL);
1584 	}
1585 	/*
1586 	 * do not remove from the list until nobody holds it; walking the
1587 	 * list in do_sysctl() relies on that.
1588 	 */
1589 	list_del_init(&p->ctl_entry);
1590 }
1591 
1592 void sysctl_head_get(struct ctl_table_header *head)
1593 {
1594 	spin_lock(&sysctl_lock);
1595 	head->count++;
1596 	spin_unlock(&sysctl_lock);
1597 }
1598 
1599 void sysctl_head_put(struct ctl_table_header *head)
1600 {
1601 	spin_lock(&sysctl_lock);
1602 	if (!--head->count)
1603 		kfree(head);
1604 	spin_unlock(&sysctl_lock);
1605 }
1606 
1607 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1608 {
1609 	if (!head)
1610 		BUG();
1611 	spin_lock(&sysctl_lock);
1612 	if (!use_table(head))
1613 		head = ERR_PTR(-ENOENT);
1614 	spin_unlock(&sysctl_lock);
1615 	return head;
1616 }
1617 
1618 void sysctl_head_finish(struct ctl_table_header *head)
1619 {
1620 	if (!head)
1621 		return;
1622 	spin_lock(&sysctl_lock);
1623 	unuse_table(head);
1624 	spin_unlock(&sysctl_lock);
1625 }
1626 
1627 static struct ctl_table_set *
1628 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1629 {
1630 	struct ctl_table_set *set = &root->default_set;
1631 	if (root->lookup)
1632 		set = root->lookup(root, namespaces);
1633 	return set;
1634 }
1635 
1636 static struct list_head *
1637 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1638 {
1639 	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1640 	return &set->list;
1641 }
1642 
1643 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1644 					    struct ctl_table_header *prev)
1645 {
1646 	struct ctl_table_root *root;
1647 	struct list_head *header_list;
1648 	struct ctl_table_header *head;
1649 	struct list_head *tmp;
1650 
1651 	spin_lock(&sysctl_lock);
1652 	if (prev) {
1653 		head = prev;
1654 		tmp = &prev->ctl_entry;
1655 		unuse_table(prev);
1656 		goto next;
1657 	}
1658 	tmp = &root_table_header.ctl_entry;
1659 	for (;;) {
1660 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1661 
1662 		if (!use_table(head))
1663 			goto next;
1664 		spin_unlock(&sysctl_lock);
1665 		return head;
1666 	next:
1667 		root = head->root;
1668 		tmp = tmp->next;
1669 		header_list = lookup_header_list(root, namespaces);
1670 		if (tmp != header_list)
1671 			continue;
1672 
1673 		do {
1674 			root = list_entry(root->root_list.next,
1675 					struct ctl_table_root, root_list);
1676 			if (root == &sysctl_table_root)
1677 				goto out;
1678 			header_list = lookup_header_list(root, namespaces);
1679 		} while (list_empty(header_list));
1680 		tmp = header_list->next;
1681 	}
1682 out:
1683 	spin_unlock(&sysctl_lock);
1684 	return NULL;
1685 }
1686 
1687 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1688 {
1689 	return __sysctl_head_next(current->nsproxy, prev);
1690 }
1691 
1692 void register_sysctl_root(struct ctl_table_root *root)
1693 {
1694 	spin_lock(&sysctl_lock);
1695 	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1696 	spin_unlock(&sysctl_lock);
1697 }
1698 
1699 #ifdef CONFIG_SYSCTL_SYSCALL
1700 /* Perform the actual read/write of a sysctl table entry. */
1701 static int do_sysctl_strategy(struct ctl_table_root *root,
1702 			struct ctl_table *table,
1703 			void __user *oldval, size_t __user *oldlenp,
1704 			void __user *newval, size_t newlen)
1705 {
1706 	int op = 0, rc;
1707 
1708 	if (oldval)
1709 		op |= MAY_READ;
1710 	if (newval)
1711 		op |= MAY_WRITE;
1712 	if (sysctl_perm(root, table, op))
1713 		return -EPERM;
1714 
1715 	if (table->strategy) {
1716 		rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1717 		if (rc < 0)
1718 			return rc;
1719 		if (rc > 0)
1720 			return 0;
1721 	}
1722 
1723 	/* If there is no strategy routine, or if the strategy returns
1724 	 * zero, proceed with automatic r/w */
1725 	if (table->data && table->maxlen) {
1726 		rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1727 		if (rc < 0)
1728 			return rc;
1729 	}
1730 	return 0;
1731 }
1732 
1733 static int parse_table(int __user *name, int nlen,
1734 		       void __user *oldval, size_t __user *oldlenp,
1735 		       void __user *newval, size_t newlen,
1736 		       struct ctl_table_root *root,
1737 		       struct ctl_table *table)
1738 {
1739 	int n;
1740 repeat:
1741 	if (!nlen)
1742 		return -ENOTDIR;
1743 	if (get_user(n, name))
1744 		return -EFAULT;
1745 	for ( ; table->ctl_name || table->procname; table++) {
1746 		if (!table->ctl_name)
1747 			continue;
1748 		if (n == table->ctl_name) {
1749 			int error;
1750 			if (table->child) {
1751 				if (sysctl_perm(root, table, MAY_EXEC))
1752 					return -EPERM;
1753 				name++;
1754 				nlen--;
1755 				table = table->child;
1756 				goto repeat;
1757 			}
1758 			error = do_sysctl_strategy(root, table,
1759 						   oldval, oldlenp,
1760 						   newval, newlen);
1761 			return error;
1762 		}
1763 	}
1764 	return -ENOTDIR;
1765 }
1766 
1767 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1768 	       void __user *newval, size_t newlen)
1769 {
1770 	struct ctl_table_header *head;
1771 	int error = -ENOTDIR;
1772 
1773 	if (nlen <= 0 || nlen >= CTL_MAXNAME)
1774 		return -ENOTDIR;
1775 	if (oldval) {
1776 		int old_len;
1777 		if (!oldlenp || get_user(old_len, oldlenp))
1778 			return -EFAULT;
1779 	}
1780 
1781 	for (head = sysctl_head_next(NULL); head;
1782 			head = sysctl_head_next(head)) {
1783 		error = parse_table(name, nlen, oldval, oldlenp,
1784 					newval, newlen,
1785 					head->root, head->ctl_table);
1786 		if (error != -ENOTDIR) {
1787 			sysctl_head_finish(head);
1788 			break;
1789 		}
1790 	}
1791 	return error;
1792 }
1793 
1794 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
1795 {
1796 	struct __sysctl_args tmp;
1797 	int error;
1798 
1799 	if (copy_from_user(&tmp, args, sizeof(tmp)))
1800 		return -EFAULT;
1801 
1802 	error = deprecated_sysctl_warning(&tmp);
1803 	if (error)
1804 		goto out;
1805 
1806 	lock_kernel();
1807 	error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1808 			  tmp.newval, tmp.newlen);
1809 	unlock_kernel();
1810 out:
1811 	return error;
1812 }
1813 #endif /* CONFIG_SYSCTL_SYSCALL */
1814 
1815 /*
1816  * sysctl_perm does NOT grant the superuser all rights automatically, because
1817  * some sysctl variables are readonly even to root.
1818  */
1819 
1820 static int test_perm(int mode, int op)
1821 {
1822 	if (!current_euid())
1823 		mode >>= 6;
1824 	else if (in_egroup_p(0))
1825 		mode >>= 3;
1826 	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1827 		return 0;
1828 	return -EACCES;
1829 }
1830 
1831 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1832 {
1833 	int error;
1834 	int mode;
1835 
1836 	error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1837 	if (error)
1838 		return error;
1839 
1840 	if (root->permissions)
1841 		mode = root->permissions(root, current->nsproxy, table);
1842 	else
1843 		mode = table->mode;
1844 
1845 	return test_perm(mode, op);
1846 }
1847 
1848 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1849 {
1850 	for (; table->ctl_name || table->procname; table++) {
1851 		table->parent = parent;
1852 		if (table->child)
1853 			sysctl_set_parent(table, table->child);
1854 	}
1855 }
1856 
1857 static __init int sysctl_init(void)
1858 {
1859 	sysctl_set_parent(NULL, root_table);
1860 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1861 	{
1862 		int err;
1863 		err = sysctl_check_table(current->nsproxy, root_table);
1864 	}
1865 #endif
1866 	return 0;
1867 }
1868 
1869 core_initcall(sysctl_init);
1870 
1871 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1872 				      struct ctl_table *table)
1873 {
1874 	struct ctl_table *p;
1875 	const char *s = branch->procname;
1876 
1877 	/* branch should have named subdirectory as its first element */
1878 	if (!s || !branch->child)
1879 		return NULL;
1880 
1881 	/* ... and nothing else */
1882 	if (branch[1].procname || branch[1].ctl_name)
1883 		return NULL;
1884 
1885 	/* table should contain subdirectory with the same name */
1886 	for (p = table; p->procname || p->ctl_name; p++) {
1887 		if (!p->child)
1888 			continue;
1889 		if (p->procname && strcmp(p->procname, s) == 0)
1890 			return p;
1891 	}
1892 	return NULL;
1893 }
1894 
1895 /* see if attaching q to p would be an improvement */
1896 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1897 {
1898 	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1899 	struct ctl_table *next;
1900 	int is_better = 0;
1901 	int not_in_parent = !p->attached_by;
1902 
1903 	while ((next = is_branch_in(by, to)) != NULL) {
1904 		if (by == q->attached_by)
1905 			is_better = 1;
1906 		if (to == p->attached_by)
1907 			not_in_parent = 1;
1908 		by = by->child;
1909 		to = next->child;
1910 	}
1911 
1912 	if (is_better && not_in_parent) {
1913 		q->attached_by = by;
1914 		q->attached_to = to;
1915 		q->parent = p;
1916 	}
1917 }
1918 
1919 /**
1920  * __register_sysctl_paths - register a sysctl hierarchy
1921  * @root: List of sysctl headers to register on
1922  * @namespaces: Data to compute which lists of sysctl entries are visible
1923  * @path: The path to the directory the sysctl table is in.
1924  * @table: the top-level table structure
1925  *
1926  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1927  * array. A completely 0 filled entry terminates the table.
1928  *
1929  * The members of the &struct ctl_table structure are used as follows:
1930  *
1931  * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1932  *            must be unique within that level of sysctl
1933  *
1934  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1935  *            enter a sysctl file
1936  *
1937  * data - a pointer to data for use by proc_handler
1938  *
1939  * maxlen - the maximum size in bytes of the data
1940  *
1941  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1942  *
1943  * child - a pointer to the child sysctl table if this entry is a directory, or
1944  *         %NULL.
1945  *
1946  * proc_handler - the text handler routine (described below)
1947  *
1948  * strategy - the strategy routine (described below)
1949  *
1950  * de - for internal use by the sysctl routines
1951  *
1952  * extra1, extra2 - extra pointers usable by the proc handler routines
1953  *
1954  * Leaf nodes in the sysctl tree will be represented by a single file
1955  * under /proc; non-leaf nodes will be represented by directories.
1956  *
1957  * sysctl(2) can automatically manage read and write requests through
1958  * the sysctl table.  The data and maxlen fields of the ctl_table
1959  * struct enable minimal validation of the values being written to be
1960  * performed, and the mode field allows minimal authentication.
1961  *
1962  * More sophisticated management can be enabled by the provision of a
1963  * strategy routine with the table entry.  This will be called before
1964  * any automatic read or write of the data is performed.
1965  *
1966  * The strategy routine may return
1967  *
1968  * < 0 - Error occurred (error is passed to user process)
1969  *
1970  * 0   - OK - proceed with automatic read or write.
1971  *
1972  * > 0 - OK - read or write has been done by the strategy routine, so
1973  *       return immediately.
1974  *
1975  * There must be a proc_handler routine for any terminal nodes
1976  * mirrored under /proc/sys (non-terminals are handled by a built-in
1977  * directory handler).  Several default handlers are available to
1978  * cover common cases -
1979  *
1980  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1981  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1982  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1983  *
1984  * It is the handler's job to read the input buffer from user memory
1985  * and process it. The handler should return 0 on success.
1986  *
1987  * This routine returns %NULL on a failure to register, and a pointer
1988  * to the table header on success.
1989  */
1990 struct ctl_table_header *__register_sysctl_paths(
1991 	struct ctl_table_root *root,
1992 	struct nsproxy *namespaces,
1993 	const struct ctl_path *path, struct ctl_table *table)
1994 {
1995 	struct ctl_table_header *header;
1996 	struct ctl_table *new, **prevp;
1997 	unsigned int n, npath;
1998 	struct ctl_table_set *set;
1999 
2000 	/* Count the path components */
2001 	for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
2002 		;
2003 
2004 	/*
2005 	 * For each path component, allocate a 2-element ctl_table array.
2006 	 * The first array element will be filled with the sysctl entry
2007 	 * for this, the second will be the sentinel (ctl_name == 0).
2008 	 *
2009 	 * We allocate everything in one go so that we don't have to
2010 	 * worry about freeing additional memory in unregister_sysctl_table.
2011 	 */
2012 	header = kzalloc(sizeof(struct ctl_table_header) +
2013 			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
2014 	if (!header)
2015 		return NULL;
2016 
2017 	new = (struct ctl_table *) (header + 1);
2018 
2019 	/* Now connect the dots */
2020 	prevp = &header->ctl_table;
2021 	for (n = 0; n < npath; ++n, ++path) {
2022 		/* Copy the procname */
2023 		new->procname = path->procname;
2024 		new->ctl_name = path->ctl_name;
2025 		new->mode     = 0555;
2026 
2027 		*prevp = new;
2028 		prevp = &new->child;
2029 
2030 		new += 2;
2031 	}
2032 	*prevp = table;
2033 	header->ctl_table_arg = table;
2034 
2035 	INIT_LIST_HEAD(&header->ctl_entry);
2036 	header->used = 0;
2037 	header->unregistering = NULL;
2038 	header->root = root;
2039 	sysctl_set_parent(NULL, header->ctl_table);
2040 	header->count = 1;
2041 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
2042 	if (sysctl_check_table(namespaces, header->ctl_table)) {
2043 		kfree(header);
2044 		return NULL;
2045 	}
2046 #endif
2047 	spin_lock(&sysctl_lock);
2048 	header->set = lookup_header_set(root, namespaces);
2049 	header->attached_by = header->ctl_table;
2050 	header->attached_to = root_table;
2051 	header->parent = &root_table_header;
2052 	for (set = header->set; set; set = set->parent) {
2053 		struct ctl_table_header *p;
2054 		list_for_each_entry(p, &set->list, ctl_entry) {
2055 			if (p->unregistering)
2056 				continue;
2057 			try_attach(p, header);
2058 		}
2059 	}
2060 	header->parent->count++;
2061 	list_add_tail(&header->ctl_entry, &header->set->list);
2062 	spin_unlock(&sysctl_lock);
2063 
2064 	return header;
2065 }
2066 
2067 /**
2068  * register_sysctl_table_path - register a sysctl table hierarchy
2069  * @path: The path to the directory the sysctl table is in.
2070  * @table: the top-level table structure
2071  *
2072  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2073  * array. A completely 0 filled entry terminates the table.
2074  *
2075  * See __register_sysctl_paths for more details.
2076  */
2077 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2078 						struct ctl_table *table)
2079 {
2080 	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
2081 					path, table);
2082 }
2083 
2084 /**
2085  * register_sysctl_table - register a sysctl table hierarchy
2086  * @table: the top-level table structure
2087  *
2088  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2089  * array. A completely 0 filled entry terminates the table.
2090  *
2091  * See register_sysctl_paths for more details.
2092  */
2093 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
2094 {
2095 	static const struct ctl_path null_path[] = { {} };
2096 
2097 	return register_sysctl_paths(null_path, table);
2098 }
2099 
2100 /**
2101  * unregister_sysctl_table - unregister a sysctl table hierarchy
2102  * @header: the header returned from register_sysctl_table
2103  *
2104  * Unregisters the sysctl table and all children. proc entries may not
2105  * actually be removed until they are no longer used by anyone.
2106  */
2107 void unregister_sysctl_table(struct ctl_table_header * header)
2108 {
2109 	might_sleep();
2110 
2111 	if (header == NULL)
2112 		return;
2113 
2114 	spin_lock(&sysctl_lock);
2115 	start_unregistering(header);
2116 	if (!--header->parent->count) {
2117 		WARN_ON(1);
2118 		kfree(header->parent);
2119 	}
2120 	if (!--header->count)
2121 		kfree(header);
2122 	spin_unlock(&sysctl_lock);
2123 }
2124 
2125 int sysctl_is_seen(struct ctl_table_header *p)
2126 {
2127 	struct ctl_table_set *set = p->set;
2128 	int res;
2129 	spin_lock(&sysctl_lock);
2130 	if (p->unregistering)
2131 		res = 0;
2132 	else if (!set->is_seen)
2133 		res = 1;
2134 	else
2135 		res = set->is_seen(set);
2136 	spin_unlock(&sysctl_lock);
2137 	return res;
2138 }
2139 
2140 void setup_sysctl_set(struct ctl_table_set *p,
2141 	struct ctl_table_set *parent,
2142 	int (*is_seen)(struct ctl_table_set *))
2143 {
2144 	INIT_LIST_HEAD(&p->list);
2145 	p->parent = parent ? parent : &sysctl_table_root.default_set;
2146 	p->is_seen = is_seen;
2147 }
2148 
2149 #else /* !CONFIG_SYSCTL */
2150 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2151 {
2152 	return NULL;
2153 }
2154 
2155 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2156 						    struct ctl_table *table)
2157 {
2158 	return NULL;
2159 }
2160 
2161 void unregister_sysctl_table(struct ctl_table_header * table)
2162 {
2163 }
2164 
2165 void setup_sysctl_set(struct ctl_table_set *p,
2166 	struct ctl_table_set *parent,
2167 	int (*is_seen)(struct ctl_table_set *))
2168 {
2169 }
2170 
2171 void sysctl_head_put(struct ctl_table_header *head)
2172 {
2173 }
2174 
2175 #endif /* CONFIG_SYSCTL */
2176 
2177 /*
2178  * /proc/sys support
2179  */
2180 
2181 #ifdef CONFIG_PROC_SYSCTL
2182 
2183 static int _proc_do_string(void* data, int maxlen, int write,
2184 			   struct file *filp, void __user *buffer,
2185 			   size_t *lenp, loff_t *ppos)
2186 {
2187 	size_t len;
2188 	char __user *p;
2189 	char c;
2190 
2191 	if (!data || !maxlen || !*lenp) {
2192 		*lenp = 0;
2193 		return 0;
2194 	}
2195 
2196 	if (write) {
2197 		len = 0;
2198 		p = buffer;
2199 		while (len < *lenp) {
2200 			if (get_user(c, p++))
2201 				return -EFAULT;
2202 			if (c == 0 || c == '\n')
2203 				break;
2204 			len++;
2205 		}
2206 		if (len >= maxlen)
2207 			len = maxlen-1;
2208 		if(copy_from_user(data, buffer, len))
2209 			return -EFAULT;
2210 		((char *) data)[len] = 0;
2211 		*ppos += *lenp;
2212 	} else {
2213 		len = strlen(data);
2214 		if (len > maxlen)
2215 			len = maxlen;
2216 
2217 		if (*ppos > len) {
2218 			*lenp = 0;
2219 			return 0;
2220 		}
2221 
2222 		data += *ppos;
2223 		len  -= *ppos;
2224 
2225 		if (len > *lenp)
2226 			len = *lenp;
2227 		if (len)
2228 			if(copy_to_user(buffer, data, len))
2229 				return -EFAULT;
2230 		if (len < *lenp) {
2231 			if(put_user('\n', ((char __user *) buffer) + len))
2232 				return -EFAULT;
2233 			len++;
2234 		}
2235 		*lenp = len;
2236 		*ppos += len;
2237 	}
2238 	return 0;
2239 }
2240 
2241 /**
2242  * proc_dostring - read a string sysctl
2243  * @table: the sysctl table
2244  * @write: %TRUE if this is a write to the sysctl file
2245  * @filp: the file structure
2246  * @buffer: the user buffer
2247  * @lenp: the size of the user buffer
2248  * @ppos: file position
2249  *
2250  * Reads/writes a string from/to the user buffer. If the kernel
2251  * buffer provided is not large enough to hold the string, the
2252  * string is truncated. The copied string is %NULL-terminated.
2253  * If the string is being read by the user process, it is copied
2254  * and a newline '\n' is added. It is truncated if the buffer is
2255  * not large enough.
2256  *
2257  * Returns 0 on success.
2258  */
2259 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2260 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2261 {
2262 	return _proc_do_string(table->data, table->maxlen, write, filp,
2263 			       buffer, lenp, ppos);
2264 }
2265 
2266 
2267 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2268 				 int *valp,
2269 				 int write, void *data)
2270 {
2271 	if (write) {
2272 		*valp = *negp ? -*lvalp : *lvalp;
2273 	} else {
2274 		int val = *valp;
2275 		if (val < 0) {
2276 			*negp = -1;
2277 			*lvalp = (unsigned long)-val;
2278 		} else {
2279 			*negp = 0;
2280 			*lvalp = (unsigned long)val;
2281 		}
2282 	}
2283 	return 0;
2284 }
2285 
2286 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2287 		  int write, struct file *filp, void __user *buffer,
2288 		  size_t *lenp, loff_t *ppos,
2289 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2290 			      int write, void *data),
2291 		  void *data)
2292 {
2293 #define TMPBUFLEN 21
2294 	int *i, vleft, first = 1, neg;
2295 	unsigned long lval;
2296 	size_t left, len;
2297 
2298 	char buf[TMPBUFLEN], *p;
2299 	char __user *s = buffer;
2300 
2301 	if (!tbl_data || !table->maxlen || !*lenp ||
2302 	    (*ppos && !write)) {
2303 		*lenp = 0;
2304 		return 0;
2305 	}
2306 
2307 	i = (int *) tbl_data;
2308 	vleft = table->maxlen / sizeof(*i);
2309 	left = *lenp;
2310 
2311 	if (!conv)
2312 		conv = do_proc_dointvec_conv;
2313 
2314 	for (; left && vleft--; i++, first=0) {
2315 		if (write) {
2316 			while (left) {
2317 				char c;
2318 				if (get_user(c, s))
2319 					return -EFAULT;
2320 				if (!isspace(c))
2321 					break;
2322 				left--;
2323 				s++;
2324 			}
2325 			if (!left)
2326 				break;
2327 			neg = 0;
2328 			len = left;
2329 			if (len > sizeof(buf) - 1)
2330 				len = sizeof(buf) - 1;
2331 			if (copy_from_user(buf, s, len))
2332 				return -EFAULT;
2333 			buf[len] = 0;
2334 			p = buf;
2335 			if (*p == '-' && left > 1) {
2336 				neg = 1;
2337 				p++;
2338 			}
2339 			if (*p < '0' || *p > '9')
2340 				break;
2341 
2342 			lval = simple_strtoul(p, &p, 0);
2343 
2344 			len = p-buf;
2345 			if ((len < left) && *p && !isspace(*p))
2346 				break;
2347 			s += len;
2348 			left -= len;
2349 
2350 			if (conv(&neg, &lval, i, 1, data))
2351 				break;
2352 		} else {
2353 			p = buf;
2354 			if (!first)
2355 				*p++ = '\t';
2356 
2357 			if (conv(&neg, &lval, i, 0, data))
2358 				break;
2359 
2360 			sprintf(p, "%s%lu", neg ? "-" : "", lval);
2361 			len = strlen(buf);
2362 			if (len > left)
2363 				len = left;
2364 			if(copy_to_user(s, buf, len))
2365 				return -EFAULT;
2366 			left -= len;
2367 			s += len;
2368 		}
2369 	}
2370 
2371 	if (!write && !first && left) {
2372 		if(put_user('\n', s))
2373 			return -EFAULT;
2374 		left--, s++;
2375 	}
2376 	if (write) {
2377 		while (left) {
2378 			char c;
2379 			if (get_user(c, s++))
2380 				return -EFAULT;
2381 			if (!isspace(c))
2382 				break;
2383 			left--;
2384 		}
2385 	}
2386 	if (write && first)
2387 		return -EINVAL;
2388 	*lenp -= left;
2389 	*ppos += *lenp;
2390 	return 0;
2391 #undef TMPBUFLEN
2392 }
2393 
2394 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2395 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2396 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2397 			      int write, void *data),
2398 		  void *data)
2399 {
2400 	return __do_proc_dointvec(table->data, table, write, filp,
2401 			buffer, lenp, ppos, conv, data);
2402 }
2403 
2404 /**
2405  * proc_dointvec - read a vector of integers
2406  * @table: the sysctl table
2407  * @write: %TRUE if this is a write to the sysctl file
2408  * @filp: the file structure
2409  * @buffer: the user buffer
2410  * @lenp: the size of the user buffer
2411  * @ppos: file position
2412  *
2413  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2414  * values from/to the user buffer, treated as an ASCII string.
2415  *
2416  * Returns 0 on success.
2417  */
2418 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2419 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2420 {
2421     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2422 		    	    NULL,NULL);
2423 }
2424 
2425 /*
2426  * Taint values can only be increased
2427  * This means we can safely use a temporary.
2428  */
2429 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
2430 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2431 {
2432 	struct ctl_table t;
2433 	unsigned long tmptaint = get_taint();
2434 	int err;
2435 
2436 	if (write && !capable(CAP_SYS_ADMIN))
2437 		return -EPERM;
2438 
2439 	t = *table;
2440 	t.data = &tmptaint;
2441 	err = proc_doulongvec_minmax(&t, write, filp, buffer, lenp, ppos);
2442 	if (err < 0)
2443 		return err;
2444 
2445 	if (write) {
2446 		/*
2447 		 * Poor man's atomic or. Not worth adding a primitive
2448 		 * to everyone's atomic.h for this
2449 		 */
2450 		int i;
2451 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2452 			if ((tmptaint >> i) & 1)
2453 				add_taint(i);
2454 		}
2455 	}
2456 
2457 	return err;
2458 }
2459 
2460 struct do_proc_dointvec_minmax_conv_param {
2461 	int *min;
2462 	int *max;
2463 };
2464 
2465 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2466 					int *valp,
2467 					int write, void *data)
2468 {
2469 	struct do_proc_dointvec_minmax_conv_param *param = data;
2470 	if (write) {
2471 		int val = *negp ? -*lvalp : *lvalp;
2472 		if ((param->min && *param->min > val) ||
2473 		    (param->max && *param->max < val))
2474 			return -EINVAL;
2475 		*valp = val;
2476 	} else {
2477 		int val = *valp;
2478 		if (val < 0) {
2479 			*negp = -1;
2480 			*lvalp = (unsigned long)-val;
2481 		} else {
2482 			*negp = 0;
2483 			*lvalp = (unsigned long)val;
2484 		}
2485 	}
2486 	return 0;
2487 }
2488 
2489 /**
2490  * proc_dointvec_minmax - read a vector of integers with min/max values
2491  * @table: the sysctl table
2492  * @write: %TRUE if this is a write to the sysctl file
2493  * @filp: the file structure
2494  * @buffer: the user buffer
2495  * @lenp: the size of the user buffer
2496  * @ppos: file position
2497  *
2498  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2499  * values from/to the user buffer, treated as an ASCII string.
2500  *
2501  * This routine will ensure the values are within the range specified by
2502  * table->extra1 (min) and table->extra2 (max).
2503  *
2504  * Returns 0 on success.
2505  */
2506 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2507 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2508 {
2509 	struct do_proc_dointvec_minmax_conv_param param = {
2510 		.min = (int *) table->extra1,
2511 		.max = (int *) table->extra2,
2512 	};
2513 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2514 				do_proc_dointvec_minmax_conv, &param);
2515 }
2516 
2517 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2518 				     struct file *filp,
2519 				     void __user *buffer,
2520 				     size_t *lenp, loff_t *ppos,
2521 				     unsigned long convmul,
2522 				     unsigned long convdiv)
2523 {
2524 #define TMPBUFLEN 21
2525 	unsigned long *i, *min, *max, val;
2526 	int vleft, first=1, neg;
2527 	size_t len, left;
2528 	char buf[TMPBUFLEN], *p;
2529 	char __user *s = buffer;
2530 
2531 	if (!data || !table->maxlen || !*lenp ||
2532 	    (*ppos && !write)) {
2533 		*lenp = 0;
2534 		return 0;
2535 	}
2536 
2537 	i = (unsigned long *) data;
2538 	min = (unsigned long *) table->extra1;
2539 	max = (unsigned long *) table->extra2;
2540 	vleft = table->maxlen / sizeof(unsigned long);
2541 	left = *lenp;
2542 
2543 	for (; left && vleft--; i++, min++, max++, first=0) {
2544 		if (write) {
2545 			while (left) {
2546 				char c;
2547 				if (get_user(c, s))
2548 					return -EFAULT;
2549 				if (!isspace(c))
2550 					break;
2551 				left--;
2552 				s++;
2553 			}
2554 			if (!left)
2555 				break;
2556 			neg = 0;
2557 			len = left;
2558 			if (len > TMPBUFLEN-1)
2559 				len = TMPBUFLEN-1;
2560 			if (copy_from_user(buf, s, len))
2561 				return -EFAULT;
2562 			buf[len] = 0;
2563 			p = buf;
2564 			if (*p == '-' && left > 1) {
2565 				neg = 1;
2566 				p++;
2567 			}
2568 			if (*p < '0' || *p > '9')
2569 				break;
2570 			val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2571 			len = p-buf;
2572 			if ((len < left) && *p && !isspace(*p))
2573 				break;
2574 			if (neg)
2575 				val = -val;
2576 			s += len;
2577 			left -= len;
2578 
2579 			if(neg)
2580 				continue;
2581 			if ((min && val < *min) || (max && val > *max))
2582 				continue;
2583 			*i = val;
2584 		} else {
2585 			p = buf;
2586 			if (!first)
2587 				*p++ = '\t';
2588 			sprintf(p, "%lu", convdiv * (*i) / convmul);
2589 			len = strlen(buf);
2590 			if (len > left)
2591 				len = left;
2592 			if(copy_to_user(s, buf, len))
2593 				return -EFAULT;
2594 			left -= len;
2595 			s += len;
2596 		}
2597 	}
2598 
2599 	if (!write && !first && left) {
2600 		if(put_user('\n', s))
2601 			return -EFAULT;
2602 		left--, s++;
2603 	}
2604 	if (write) {
2605 		while (left) {
2606 			char c;
2607 			if (get_user(c, s++))
2608 				return -EFAULT;
2609 			if (!isspace(c))
2610 				break;
2611 			left--;
2612 		}
2613 	}
2614 	if (write && first)
2615 		return -EINVAL;
2616 	*lenp -= left;
2617 	*ppos += *lenp;
2618 	return 0;
2619 #undef TMPBUFLEN
2620 }
2621 
2622 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2623 				     struct file *filp,
2624 				     void __user *buffer,
2625 				     size_t *lenp, loff_t *ppos,
2626 				     unsigned long convmul,
2627 				     unsigned long convdiv)
2628 {
2629 	return __do_proc_doulongvec_minmax(table->data, table, write,
2630 			filp, buffer, lenp, ppos, convmul, convdiv);
2631 }
2632 
2633 /**
2634  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2635  * @table: the sysctl table
2636  * @write: %TRUE if this is a write to the sysctl file
2637  * @filp: the file structure
2638  * @buffer: the user buffer
2639  * @lenp: the size of the user buffer
2640  * @ppos: file position
2641  *
2642  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2643  * values from/to the user buffer, treated as an ASCII string.
2644  *
2645  * This routine will ensure the values are within the range specified by
2646  * table->extra1 (min) and table->extra2 (max).
2647  *
2648  * Returns 0 on success.
2649  */
2650 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2651 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2652 {
2653     return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2654 }
2655 
2656 /**
2657  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2658  * @table: the sysctl table
2659  * @write: %TRUE if this is a write to the sysctl file
2660  * @filp: the file structure
2661  * @buffer: the user buffer
2662  * @lenp: the size of the user buffer
2663  * @ppos: file position
2664  *
2665  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2666  * values from/to the user buffer, treated as an ASCII string. The values
2667  * are treated as milliseconds, and converted to jiffies when they are stored.
2668  *
2669  * This routine will ensure the values are within the range specified by
2670  * table->extra1 (min) and table->extra2 (max).
2671  *
2672  * Returns 0 on success.
2673  */
2674 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2675 				      struct file *filp,
2676 				      void __user *buffer,
2677 				      size_t *lenp, loff_t *ppos)
2678 {
2679     return do_proc_doulongvec_minmax(table, write, filp, buffer,
2680 				     lenp, ppos, HZ, 1000l);
2681 }
2682 
2683 
2684 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2685 					 int *valp,
2686 					 int write, void *data)
2687 {
2688 	if (write) {
2689 		if (*lvalp > LONG_MAX / HZ)
2690 			return 1;
2691 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2692 	} else {
2693 		int val = *valp;
2694 		unsigned long lval;
2695 		if (val < 0) {
2696 			*negp = -1;
2697 			lval = (unsigned long)-val;
2698 		} else {
2699 			*negp = 0;
2700 			lval = (unsigned long)val;
2701 		}
2702 		*lvalp = lval / HZ;
2703 	}
2704 	return 0;
2705 }
2706 
2707 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2708 						int *valp,
2709 						int write, void *data)
2710 {
2711 	if (write) {
2712 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2713 			return 1;
2714 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2715 	} else {
2716 		int val = *valp;
2717 		unsigned long lval;
2718 		if (val < 0) {
2719 			*negp = -1;
2720 			lval = (unsigned long)-val;
2721 		} else {
2722 			*negp = 0;
2723 			lval = (unsigned long)val;
2724 		}
2725 		*lvalp = jiffies_to_clock_t(lval);
2726 	}
2727 	return 0;
2728 }
2729 
2730 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2731 					    int *valp,
2732 					    int write, void *data)
2733 {
2734 	if (write) {
2735 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2736 	} else {
2737 		int val = *valp;
2738 		unsigned long lval;
2739 		if (val < 0) {
2740 			*negp = -1;
2741 			lval = (unsigned long)-val;
2742 		} else {
2743 			*negp = 0;
2744 			lval = (unsigned long)val;
2745 		}
2746 		*lvalp = jiffies_to_msecs(lval);
2747 	}
2748 	return 0;
2749 }
2750 
2751 /**
2752  * proc_dointvec_jiffies - read a vector of integers as seconds
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  *
2760  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2761  * values from/to the user buffer, treated as an ASCII string.
2762  * The values read are assumed to be in seconds, and are converted into
2763  * jiffies.
2764  *
2765  * Returns 0 on success.
2766  */
2767 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2768 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2769 {
2770     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2771 		    	    do_proc_dointvec_jiffies_conv,NULL);
2772 }
2773 
2774 /**
2775  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2776  * @table: the sysctl table
2777  * @write: %TRUE if this is a write to the sysctl file
2778  * @filp: the file structure
2779  * @buffer: the user buffer
2780  * @lenp: the size of the user buffer
2781  * @ppos: pointer to the file position
2782  *
2783  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2784  * values from/to the user buffer, treated as an ASCII string.
2785  * The values read are assumed to be in 1/USER_HZ seconds, and
2786  * are converted into jiffies.
2787  *
2788  * Returns 0 on success.
2789  */
2790 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2791 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2792 {
2793     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2794 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2795 }
2796 
2797 /**
2798  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2799  * @table: the sysctl table
2800  * @write: %TRUE if this is a write to the sysctl file
2801  * @filp: the file structure
2802  * @buffer: the user buffer
2803  * @lenp: the size of the user buffer
2804  * @ppos: file position
2805  * @ppos: the current position in the file
2806  *
2807  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2808  * values from/to the user buffer, treated as an ASCII string.
2809  * The values read are assumed to be in 1/1000 seconds, and
2810  * are converted into jiffies.
2811  *
2812  * Returns 0 on success.
2813  */
2814 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2815 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2816 {
2817 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2818 				do_proc_dointvec_ms_jiffies_conv, NULL);
2819 }
2820 
2821 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2822 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2823 {
2824 	struct pid *new_pid;
2825 	pid_t tmp;
2826 	int r;
2827 
2828 	tmp = pid_vnr(cad_pid);
2829 
2830 	r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2831 			       lenp, ppos, NULL, NULL);
2832 	if (r || !write)
2833 		return r;
2834 
2835 	new_pid = find_get_pid(tmp);
2836 	if (!new_pid)
2837 		return -ESRCH;
2838 
2839 	put_pid(xchg(&cad_pid, new_pid));
2840 	return 0;
2841 }
2842 
2843 #else /* CONFIG_PROC_FS */
2844 
2845 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2846 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2847 {
2848 	return -ENOSYS;
2849 }
2850 
2851 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2852 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2853 {
2854 	return -ENOSYS;
2855 }
2856 
2857 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2858 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2859 {
2860 	return -ENOSYS;
2861 }
2862 
2863 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2864 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2865 {
2866 	return -ENOSYS;
2867 }
2868 
2869 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2870 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2871 {
2872 	return -ENOSYS;
2873 }
2874 
2875 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2876 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2877 {
2878 	return -ENOSYS;
2879 }
2880 
2881 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2882 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2883 {
2884 	return -ENOSYS;
2885 }
2886 
2887 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2888 				      struct file *filp,
2889 				      void __user *buffer,
2890 				      size_t *lenp, loff_t *ppos)
2891 {
2892     return -ENOSYS;
2893 }
2894 
2895 
2896 #endif /* CONFIG_PROC_FS */
2897 
2898 
2899 #ifdef CONFIG_SYSCTL_SYSCALL
2900 /*
2901  * General sysctl support routines
2902  */
2903 
2904 /* The generic sysctl data routine (used if no strategy routine supplied) */
2905 int sysctl_data(struct ctl_table *table,
2906 		void __user *oldval, size_t __user *oldlenp,
2907 		void __user *newval, size_t newlen)
2908 {
2909 	size_t len;
2910 
2911 	/* Get out of I don't have a variable */
2912 	if (!table->data || !table->maxlen)
2913 		return -ENOTDIR;
2914 
2915 	if (oldval && oldlenp) {
2916 		if (get_user(len, oldlenp))
2917 			return -EFAULT;
2918 		if (len) {
2919 			if (len > table->maxlen)
2920 				len = table->maxlen;
2921 			if (copy_to_user(oldval, table->data, len))
2922 				return -EFAULT;
2923 			if (put_user(len, oldlenp))
2924 				return -EFAULT;
2925 		}
2926 	}
2927 
2928 	if (newval && newlen) {
2929 		if (newlen > table->maxlen)
2930 			newlen = table->maxlen;
2931 
2932 		if (copy_from_user(table->data, newval, newlen))
2933 			return -EFAULT;
2934 	}
2935 	return 1;
2936 }
2937 
2938 /* The generic string strategy routine: */
2939 int sysctl_string(struct ctl_table *table,
2940 		  void __user *oldval, size_t __user *oldlenp,
2941 		  void __user *newval, size_t newlen)
2942 {
2943 	if (!table->data || !table->maxlen)
2944 		return -ENOTDIR;
2945 
2946 	if (oldval && oldlenp) {
2947 		size_t bufsize;
2948 		if (get_user(bufsize, oldlenp))
2949 			return -EFAULT;
2950 		if (bufsize) {
2951 			size_t len = strlen(table->data), copied;
2952 
2953 			/* This shouldn't trigger for a well-formed sysctl */
2954 			if (len > table->maxlen)
2955 				len = table->maxlen;
2956 
2957 			/* Copy up to a max of bufsize-1 bytes of the string */
2958 			copied = (len >= bufsize) ? bufsize - 1 : len;
2959 
2960 			if (copy_to_user(oldval, table->data, copied) ||
2961 			    put_user(0, (char __user *)(oldval + copied)))
2962 				return -EFAULT;
2963 			if (put_user(len, oldlenp))
2964 				return -EFAULT;
2965 		}
2966 	}
2967 	if (newval && newlen) {
2968 		size_t len = newlen;
2969 		if (len > table->maxlen)
2970 			len = table->maxlen;
2971 		if(copy_from_user(table->data, newval, len))
2972 			return -EFAULT;
2973 		if (len == table->maxlen)
2974 			len--;
2975 		((char *) table->data)[len] = 0;
2976 	}
2977 	return 1;
2978 }
2979 
2980 /*
2981  * This function makes sure that all of the integers in the vector
2982  * are between the minimum and maximum values given in the arrays
2983  * table->extra1 and table->extra2, respectively.
2984  */
2985 int sysctl_intvec(struct ctl_table *table,
2986 		void __user *oldval, size_t __user *oldlenp,
2987 		void __user *newval, size_t newlen)
2988 {
2989 
2990 	if (newval && newlen) {
2991 		int __user *vec = (int __user *) newval;
2992 		int *min = (int *) table->extra1;
2993 		int *max = (int *) table->extra2;
2994 		size_t length;
2995 		int i;
2996 
2997 		if (newlen % sizeof(int) != 0)
2998 			return -EINVAL;
2999 
3000 		if (!table->extra1 && !table->extra2)
3001 			return 0;
3002 
3003 		if (newlen > table->maxlen)
3004 			newlen = table->maxlen;
3005 		length = newlen / sizeof(int);
3006 
3007 		for (i = 0; i < length; i++) {
3008 			int value;
3009 			if (get_user(value, vec + i))
3010 				return -EFAULT;
3011 			if (min && value < min[i])
3012 				return -EINVAL;
3013 			if (max && value > max[i])
3014 				return -EINVAL;
3015 		}
3016 	}
3017 	return 0;
3018 }
3019 
3020 /* Strategy function to convert jiffies to seconds */
3021 int sysctl_jiffies(struct ctl_table *table,
3022 		void __user *oldval, size_t __user *oldlenp,
3023 		void __user *newval, size_t newlen)
3024 {
3025 	if (oldval && oldlenp) {
3026 		size_t olen;
3027 
3028 		if (get_user(olen, oldlenp))
3029 			return -EFAULT;
3030 		if (olen) {
3031 			int val;
3032 
3033 			if (olen < sizeof(int))
3034 				return -EINVAL;
3035 
3036 			val = *(int *)(table->data) / HZ;
3037 			if (put_user(val, (int __user *)oldval))
3038 				return -EFAULT;
3039 			if (put_user(sizeof(int), oldlenp))
3040 				return -EFAULT;
3041 		}
3042 	}
3043 	if (newval && newlen) {
3044 		int new;
3045 		if (newlen != sizeof(int))
3046 			return -EINVAL;
3047 		if (get_user(new, (int __user *)newval))
3048 			return -EFAULT;
3049 		*(int *)(table->data) = new*HZ;
3050 	}
3051 	return 1;
3052 }
3053 
3054 /* Strategy function to convert jiffies to seconds */
3055 int sysctl_ms_jiffies(struct ctl_table *table,
3056 		void __user *oldval, size_t __user *oldlenp,
3057 		void __user *newval, size_t newlen)
3058 {
3059 	if (oldval && oldlenp) {
3060 		size_t olen;
3061 
3062 		if (get_user(olen, oldlenp))
3063 			return -EFAULT;
3064 		if (olen) {
3065 			int val;
3066 
3067 			if (olen < sizeof(int))
3068 				return -EINVAL;
3069 
3070 			val = jiffies_to_msecs(*(int *)(table->data));
3071 			if (put_user(val, (int __user *)oldval))
3072 				return -EFAULT;
3073 			if (put_user(sizeof(int), oldlenp))
3074 				return -EFAULT;
3075 		}
3076 	}
3077 	if (newval && newlen) {
3078 		int new;
3079 		if (newlen != sizeof(int))
3080 			return -EINVAL;
3081 		if (get_user(new, (int __user *)newval))
3082 			return -EFAULT;
3083 		*(int *)(table->data) = msecs_to_jiffies(new);
3084 	}
3085 	return 1;
3086 }
3087 
3088 
3089 
3090 #else /* CONFIG_SYSCTL_SYSCALL */
3091 
3092 
3093 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
3094 {
3095 	struct __sysctl_args tmp;
3096 	int error;
3097 
3098 	if (copy_from_user(&tmp, args, sizeof(tmp)))
3099 		return -EFAULT;
3100 
3101 	error = deprecated_sysctl_warning(&tmp);
3102 
3103 	/* If no error reading the parameters then just -ENOSYS ... */
3104 	if (!error)
3105 		error = -ENOSYS;
3106 
3107 	return error;
3108 }
3109 
3110 int sysctl_data(struct ctl_table *table,
3111 		  void __user *oldval, size_t __user *oldlenp,
3112 		  void __user *newval, size_t newlen)
3113 {
3114 	return -ENOSYS;
3115 }
3116 
3117 int sysctl_string(struct ctl_table *table,
3118 		  void __user *oldval, size_t __user *oldlenp,
3119 		  void __user *newval, size_t newlen)
3120 {
3121 	return -ENOSYS;
3122 }
3123 
3124 int sysctl_intvec(struct ctl_table *table,
3125 		void __user *oldval, size_t __user *oldlenp,
3126 		void __user *newval, size_t newlen)
3127 {
3128 	return -ENOSYS;
3129 }
3130 
3131 int sysctl_jiffies(struct ctl_table *table,
3132 		void __user *oldval, size_t __user *oldlenp,
3133 		void __user *newval, size_t newlen)
3134 {
3135 	return -ENOSYS;
3136 }
3137 
3138 int sysctl_ms_jiffies(struct ctl_table *table,
3139 		void __user *oldval, size_t __user *oldlenp,
3140 		void __user *newval, size_t newlen)
3141 {
3142 	return -ENOSYS;
3143 }
3144 
3145 #endif /* CONFIG_SYSCTL_SYSCALL */
3146 
3147 static int deprecated_sysctl_warning(struct __sysctl_args *args)
3148 {
3149 	static int msg_count;
3150 	int name[CTL_MAXNAME];
3151 	int i;
3152 
3153 	/* Check args->nlen. */
3154 	if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
3155 		return -ENOTDIR;
3156 
3157 	/* Read in the sysctl name for better debug message logging */
3158 	for (i = 0; i < args->nlen; i++)
3159 		if (get_user(name[i], args->name + i))
3160 			return -EFAULT;
3161 
3162 	/* Ignore accesses to kernel.version */
3163 	if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
3164 		return 0;
3165 
3166 	if (msg_count < 5) {
3167 		msg_count++;
3168 		printk(KERN_INFO
3169 			"warning: process `%s' used the deprecated sysctl "
3170 			"system call with ", current->comm);
3171 		for (i = 0; i < args->nlen; i++)
3172 			printk("%d.", name[i]);
3173 		printk("\n");
3174 	}
3175 	return 0;
3176 }
3177 
3178 /*
3179  * No sense putting this after each symbol definition, twice,
3180  * exception granted :-)
3181  */
3182 EXPORT_SYMBOL(proc_dointvec);
3183 EXPORT_SYMBOL(proc_dointvec_jiffies);
3184 EXPORT_SYMBOL(proc_dointvec_minmax);
3185 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3186 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3187 EXPORT_SYMBOL(proc_dostring);
3188 EXPORT_SYMBOL(proc_doulongvec_minmax);
3189 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3190 EXPORT_SYMBOL(register_sysctl_table);
3191 EXPORT_SYMBOL(register_sysctl_paths);
3192 EXPORT_SYMBOL(sysctl_intvec);
3193 EXPORT_SYMBOL(sysctl_jiffies);
3194 EXPORT_SYMBOL(sysctl_ms_jiffies);
3195 EXPORT_SYMBOL(sysctl_string);
3196 EXPORT_SYMBOL(sysctl_data);
3197 EXPORT_SYMBOL(unregister_sysctl_table);
3198