xref: /linux/kernel/sysctl.c (revision 5bdef865eb358b6f3760e25e591ae115e9eeddef)
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_minmax,
339 		.strategy	= &sysctl_intvec,
340 		.extra1		= &zero,
341 		.extra2		= &one,
342 	},
343 #endif
344 	{
345 		.ctl_name	= CTL_UNNUMBERED,
346 		.procname	= "sched_rt_period_us",
347 		.data		= &sysctl_sched_rt_period,
348 		.maxlen		= sizeof(unsigned int),
349 		.mode		= 0644,
350 		.proc_handler	= &sched_rt_handler,
351 	},
352 	{
353 		.ctl_name	= CTL_UNNUMBERED,
354 		.procname	= "sched_rt_runtime_us",
355 		.data		= &sysctl_sched_rt_runtime,
356 		.maxlen		= sizeof(int),
357 		.mode		= 0644,
358 		.proc_handler	= &sched_rt_handler,
359 	},
360 	{
361 		.ctl_name	= CTL_UNNUMBERED,
362 		.procname	= "sched_compat_yield",
363 		.data		= &sysctl_sched_compat_yield,
364 		.maxlen		= sizeof(unsigned int),
365 		.mode		= 0644,
366 		.proc_handler	= &proc_dointvec,
367 	},
368 #ifdef CONFIG_PROVE_LOCKING
369 	{
370 		.ctl_name	= CTL_UNNUMBERED,
371 		.procname	= "prove_locking",
372 		.data		= &prove_locking,
373 		.maxlen		= sizeof(int),
374 		.mode		= 0644,
375 		.proc_handler	= &proc_dointvec,
376 	},
377 #endif
378 #ifdef CONFIG_LOCK_STAT
379 	{
380 		.ctl_name	= CTL_UNNUMBERED,
381 		.procname	= "lock_stat",
382 		.data		= &lock_stat,
383 		.maxlen		= sizeof(int),
384 		.mode		= 0644,
385 		.proc_handler	= &proc_dointvec,
386 	},
387 #endif
388 	{
389 		.ctl_name	= KERN_PANIC,
390 		.procname	= "panic",
391 		.data		= &panic_timeout,
392 		.maxlen		= sizeof(int),
393 		.mode		= 0644,
394 		.proc_handler	= &proc_dointvec,
395 	},
396 	{
397 		.ctl_name	= KERN_CORE_USES_PID,
398 		.procname	= "core_uses_pid",
399 		.data		= &core_uses_pid,
400 		.maxlen		= sizeof(int),
401 		.mode		= 0644,
402 		.proc_handler	= &proc_dointvec,
403 	},
404 	{
405 		.ctl_name	= KERN_CORE_PATTERN,
406 		.procname	= "core_pattern",
407 		.data		= core_pattern,
408 		.maxlen		= CORENAME_MAX_SIZE,
409 		.mode		= 0644,
410 		.proc_handler	= &proc_dostring,
411 		.strategy	= &sysctl_string,
412 	},
413 #ifdef CONFIG_PROC_SYSCTL
414 	{
415 		.procname	= "tainted",
416 		.maxlen 	= sizeof(long),
417 		.mode		= 0644,
418 		.proc_handler	= &proc_taint,
419 	},
420 #endif
421 #ifdef CONFIG_LATENCYTOP
422 	{
423 		.procname	= "latencytop",
424 		.data		= &latencytop_enabled,
425 		.maxlen		= sizeof(int),
426 		.mode		= 0644,
427 		.proc_handler	= &proc_dointvec,
428 	},
429 #endif
430 #ifdef CONFIG_BLK_DEV_INITRD
431 	{
432 		.ctl_name	= KERN_REALROOTDEV,
433 		.procname	= "real-root-dev",
434 		.data		= &real_root_dev,
435 		.maxlen		= sizeof(int),
436 		.mode		= 0644,
437 		.proc_handler	= &proc_dointvec,
438 	},
439 #endif
440 	{
441 		.ctl_name	= CTL_UNNUMBERED,
442 		.procname	= "print-fatal-signals",
443 		.data		= &print_fatal_signals,
444 		.maxlen		= sizeof(int),
445 		.mode		= 0644,
446 		.proc_handler	= &proc_dointvec,
447 	},
448 #ifdef CONFIG_SPARC
449 	{
450 		.ctl_name	= KERN_SPARC_REBOOT,
451 		.procname	= "reboot-cmd",
452 		.data		= reboot_command,
453 		.maxlen		= 256,
454 		.mode		= 0644,
455 		.proc_handler	= &proc_dostring,
456 		.strategy	= &sysctl_string,
457 	},
458 	{
459 		.ctl_name	= KERN_SPARC_STOP_A,
460 		.procname	= "stop-a",
461 		.data		= &stop_a_enabled,
462 		.maxlen		= sizeof (int),
463 		.mode		= 0644,
464 		.proc_handler	= &proc_dointvec,
465 	},
466 	{
467 		.ctl_name	= KERN_SPARC_SCONS_PWROFF,
468 		.procname	= "scons-poweroff",
469 		.data		= &scons_pwroff,
470 		.maxlen		= sizeof (int),
471 		.mode		= 0644,
472 		.proc_handler	= &proc_dointvec,
473 	},
474 #endif
475 #ifdef CONFIG_SPARC64
476 	{
477 		.ctl_name	= CTL_UNNUMBERED,
478 		.procname	= "tsb-ratio",
479 		.data		= &sysctl_tsb_ratio,
480 		.maxlen		= sizeof (int),
481 		.mode		= 0644,
482 		.proc_handler	= &proc_dointvec,
483 	},
484 #endif
485 #ifdef __hppa__
486 	{
487 		.ctl_name	= KERN_HPPA_PWRSW,
488 		.procname	= "soft-power",
489 		.data		= &pwrsw_enabled,
490 		.maxlen		= sizeof (int),
491 	 	.mode		= 0644,
492 		.proc_handler	= &proc_dointvec,
493 	},
494 	{
495 		.ctl_name	= KERN_HPPA_UNALIGNED,
496 		.procname	= "unaligned-trap",
497 		.data		= &unaligned_enabled,
498 		.maxlen		= sizeof (int),
499 		.mode		= 0644,
500 		.proc_handler	= &proc_dointvec,
501 	},
502 #endif
503 	{
504 		.ctl_name	= KERN_CTLALTDEL,
505 		.procname	= "ctrl-alt-del",
506 		.data		= &C_A_D,
507 		.maxlen		= sizeof(int),
508 		.mode		= 0644,
509 		.proc_handler	= &proc_dointvec,
510 	},
511 #ifdef CONFIG_FUNCTION_TRACER
512 	{
513 		.ctl_name	= CTL_UNNUMBERED,
514 		.procname	= "ftrace_enabled",
515 		.data		= &ftrace_enabled,
516 		.maxlen		= sizeof(int),
517 		.mode		= 0644,
518 		.proc_handler	= &ftrace_enable_sysctl,
519 	},
520 #endif
521 #ifdef CONFIG_STACK_TRACER
522 	{
523 		.ctl_name	= CTL_UNNUMBERED,
524 		.procname	= "stack_tracer_enabled",
525 		.data		= &stack_tracer_enabled,
526 		.maxlen		= sizeof(int),
527 		.mode		= 0644,
528 		.proc_handler	= &stack_trace_sysctl,
529 	},
530 #endif
531 #ifdef CONFIG_TRACING
532 	{
533 		.ctl_name	= CTL_UNNUMBERED,
534 		.procname	= "ftrace_dump_on_oops",
535 		.data		= &ftrace_dump_on_oops,
536 		.maxlen		= sizeof(int),
537 		.mode		= 0644,
538 		.proc_handler	= &proc_dointvec,
539 	},
540 #endif
541 #ifdef CONFIG_MODULES
542 	{
543 		.ctl_name	= KERN_MODPROBE,
544 		.procname	= "modprobe",
545 		.data		= &modprobe_path,
546 		.maxlen		= KMOD_PATH_LEN,
547 		.mode		= 0644,
548 		.proc_handler	= &proc_dostring,
549 		.strategy	= &sysctl_string,
550 	},
551 	{
552 		.ctl_name	= CTL_UNNUMBERED,
553 		.procname	= "modules_disabled",
554 		.data		= &modules_disabled,
555 		.maxlen		= sizeof(int),
556 		.mode		= 0644,
557 		/* only handle a transition from default "0" to "1" */
558 		.proc_handler	= &proc_dointvec_minmax,
559 		.extra1		= &one,
560 		.extra2		= &one,
561 	},
562 #endif
563 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
564 	{
565 		.ctl_name	= KERN_HOTPLUG,
566 		.procname	= "hotplug",
567 		.data		= &uevent_helper,
568 		.maxlen		= UEVENT_HELPER_PATH_LEN,
569 		.mode		= 0644,
570 		.proc_handler	= &proc_dostring,
571 		.strategy	= &sysctl_string,
572 	},
573 #endif
574 #ifdef CONFIG_CHR_DEV_SG
575 	{
576 		.ctl_name	= KERN_SG_BIG_BUFF,
577 		.procname	= "sg-big-buff",
578 		.data		= &sg_big_buff,
579 		.maxlen		= sizeof (int),
580 		.mode		= 0444,
581 		.proc_handler	= &proc_dointvec,
582 	},
583 #endif
584 #ifdef CONFIG_BSD_PROCESS_ACCT
585 	{
586 		.ctl_name	= KERN_ACCT,
587 		.procname	= "acct",
588 		.data		= &acct_parm,
589 		.maxlen		= 3*sizeof(int),
590 		.mode		= 0644,
591 		.proc_handler	= &proc_dointvec,
592 	},
593 #endif
594 #ifdef CONFIG_MAGIC_SYSRQ
595 	{
596 		.ctl_name	= KERN_SYSRQ,
597 		.procname	= "sysrq",
598 		.data		= &__sysrq_enabled,
599 		.maxlen		= sizeof (int),
600 		.mode		= 0644,
601 		.proc_handler	= &proc_dointvec,
602 	},
603 #endif
604 #ifdef CONFIG_PROC_SYSCTL
605 	{
606 		.procname	= "cad_pid",
607 		.data		= NULL,
608 		.maxlen		= sizeof (int),
609 		.mode		= 0600,
610 		.proc_handler	= &proc_do_cad_pid,
611 	},
612 #endif
613 	{
614 		.ctl_name	= KERN_MAX_THREADS,
615 		.procname	= "threads-max",
616 		.data		= &max_threads,
617 		.maxlen		= sizeof(int),
618 		.mode		= 0644,
619 		.proc_handler	= &proc_dointvec,
620 	},
621 	{
622 		.ctl_name	= KERN_RANDOM,
623 		.procname	= "random",
624 		.mode		= 0555,
625 		.child		= random_table,
626 	},
627 	{
628 		.ctl_name	= KERN_OVERFLOWUID,
629 		.procname	= "overflowuid",
630 		.data		= &overflowuid,
631 		.maxlen		= sizeof(int),
632 		.mode		= 0644,
633 		.proc_handler	= &proc_dointvec_minmax,
634 		.strategy	= &sysctl_intvec,
635 		.extra1		= &minolduid,
636 		.extra2		= &maxolduid,
637 	},
638 	{
639 		.ctl_name	= KERN_OVERFLOWGID,
640 		.procname	= "overflowgid",
641 		.data		= &overflowgid,
642 		.maxlen		= sizeof(int),
643 		.mode		= 0644,
644 		.proc_handler	= &proc_dointvec_minmax,
645 		.strategy	= &sysctl_intvec,
646 		.extra1		= &minolduid,
647 		.extra2		= &maxolduid,
648 	},
649 #ifdef CONFIG_S390
650 #ifdef CONFIG_MATHEMU
651 	{
652 		.ctl_name	= KERN_IEEE_EMULATION_WARNINGS,
653 		.procname	= "ieee_emulation_warnings",
654 		.data		= &sysctl_ieee_emulation_warnings,
655 		.maxlen		= sizeof(int),
656 		.mode		= 0644,
657 		.proc_handler	= &proc_dointvec,
658 	},
659 #endif
660 	{
661 		.ctl_name	= KERN_S390_USER_DEBUG_LOGGING,
662 		.procname	= "userprocess_debug",
663 		.data		= &sysctl_userprocess_debug,
664 		.maxlen		= sizeof(int),
665 		.mode		= 0644,
666 		.proc_handler	= &proc_dointvec,
667 	},
668 #endif
669 	{
670 		.ctl_name	= KERN_PIDMAX,
671 		.procname	= "pid_max",
672 		.data		= &pid_max,
673 		.maxlen		= sizeof (int),
674 		.mode		= 0644,
675 		.proc_handler	= &proc_dointvec_minmax,
676 		.strategy	= sysctl_intvec,
677 		.extra1		= &pid_max_min,
678 		.extra2		= &pid_max_max,
679 	},
680 	{
681 		.ctl_name	= KERN_PANIC_ON_OOPS,
682 		.procname	= "panic_on_oops",
683 		.data		= &panic_on_oops,
684 		.maxlen		= sizeof(int),
685 		.mode		= 0644,
686 		.proc_handler	= &proc_dointvec,
687 	},
688 #if defined CONFIG_PRINTK
689 	{
690 		.ctl_name	= KERN_PRINTK,
691 		.procname	= "printk",
692 		.data		= &console_loglevel,
693 		.maxlen		= 4*sizeof(int),
694 		.mode		= 0644,
695 		.proc_handler	= &proc_dointvec,
696 	},
697 	{
698 		.ctl_name	= KERN_PRINTK_RATELIMIT,
699 		.procname	= "printk_ratelimit",
700 		.data		= &printk_ratelimit_state.interval,
701 		.maxlen		= sizeof(int),
702 		.mode		= 0644,
703 		.proc_handler	= &proc_dointvec_jiffies,
704 		.strategy	= &sysctl_jiffies,
705 	},
706 	{
707 		.ctl_name	= KERN_PRINTK_RATELIMIT_BURST,
708 		.procname	= "printk_ratelimit_burst",
709 		.data		= &printk_ratelimit_state.burst,
710 		.maxlen		= sizeof(int),
711 		.mode		= 0644,
712 		.proc_handler	= &proc_dointvec,
713 	},
714 #endif
715 	{
716 		.ctl_name	= KERN_NGROUPS_MAX,
717 		.procname	= "ngroups_max",
718 		.data		= &ngroups_max,
719 		.maxlen		= sizeof (int),
720 		.mode		= 0444,
721 		.proc_handler	= &proc_dointvec,
722 	},
723 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
724 	{
725 		.ctl_name       = KERN_UNKNOWN_NMI_PANIC,
726 		.procname       = "unknown_nmi_panic",
727 		.data           = &unknown_nmi_panic,
728 		.maxlen         = sizeof (int),
729 		.mode           = 0644,
730 		.proc_handler   = &proc_dointvec,
731 	},
732 	{
733 		.procname       = "nmi_watchdog",
734 		.data           = &nmi_watchdog_enabled,
735 		.maxlen         = sizeof (int),
736 		.mode           = 0644,
737 		.proc_handler   = &proc_nmi_enabled,
738 	},
739 #endif
740 #if defined(CONFIG_X86)
741 	{
742 		.ctl_name	= KERN_PANIC_ON_NMI,
743 		.procname	= "panic_on_unrecovered_nmi",
744 		.data		= &panic_on_unrecovered_nmi,
745 		.maxlen		= sizeof(int),
746 		.mode		= 0644,
747 		.proc_handler	= &proc_dointvec,
748 	},
749 	{
750 		.ctl_name	= CTL_UNNUMBERED,
751 		.procname	= "panic_on_io_nmi",
752 		.data		= &panic_on_io_nmi,
753 		.maxlen		= sizeof(int),
754 		.mode		= 0644,
755 		.proc_handler	= &proc_dointvec,
756 	},
757 	{
758 		.ctl_name	= KERN_BOOTLOADER_TYPE,
759 		.procname	= "bootloader_type",
760 		.data		= &bootloader_type,
761 		.maxlen		= sizeof (int),
762 		.mode		= 0444,
763 		.proc_handler	= &proc_dointvec,
764 	},
765 	{
766 		.ctl_name	= CTL_UNNUMBERED,
767 		.procname	= "bootloader_version",
768 		.data		= &bootloader_version,
769 		.maxlen		= sizeof (int),
770 		.mode		= 0444,
771 		.proc_handler	= &proc_dointvec,
772 	},
773 	{
774 		.ctl_name	= CTL_UNNUMBERED,
775 		.procname	= "kstack_depth_to_print",
776 		.data		= &kstack_depth_to_print,
777 		.maxlen		= sizeof(int),
778 		.mode		= 0644,
779 		.proc_handler	= &proc_dointvec,
780 	},
781 	{
782 		.ctl_name	= CTL_UNNUMBERED,
783 		.procname	= "io_delay_type",
784 		.data		= &io_delay_type,
785 		.maxlen		= sizeof(int),
786 		.mode		= 0644,
787 		.proc_handler	= &proc_dointvec,
788 	},
789 #endif
790 #if defined(CONFIG_MMU)
791 	{
792 		.ctl_name	= KERN_RANDOMIZE,
793 		.procname	= "randomize_va_space",
794 		.data		= &randomize_va_space,
795 		.maxlen		= sizeof(int),
796 		.mode		= 0644,
797 		.proc_handler	= &proc_dointvec,
798 	},
799 #endif
800 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
801 	{
802 		.ctl_name	= KERN_SPIN_RETRY,
803 		.procname	= "spin_retry",
804 		.data		= &spin_retry,
805 		.maxlen		= sizeof (int),
806 		.mode		= 0644,
807 		.proc_handler	= &proc_dointvec,
808 	},
809 #endif
810 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
811 	{
812 		.procname	= "acpi_video_flags",
813 		.data		= &acpi_realmode_flags,
814 		.maxlen		= sizeof (unsigned long),
815 		.mode		= 0644,
816 		.proc_handler	= &proc_doulongvec_minmax,
817 	},
818 #endif
819 #ifdef CONFIG_IA64
820 	{
821 		.ctl_name	= KERN_IA64_UNALIGNED,
822 		.procname	= "ignore-unaligned-usertrap",
823 		.data		= &no_unaligned_warning,
824 		.maxlen		= sizeof (int),
825 	 	.mode		= 0644,
826 		.proc_handler	= &proc_dointvec,
827 	},
828 	{
829 		.ctl_name	= CTL_UNNUMBERED,
830 		.procname	= "unaligned-dump-stack",
831 		.data		= &unaligned_dump_stack,
832 		.maxlen		= sizeof (int),
833 		.mode		= 0644,
834 		.proc_handler	= &proc_dointvec,
835 	},
836 #endif
837 #ifdef CONFIG_DETECT_SOFTLOCKUP
838 	{
839 		.ctl_name	= CTL_UNNUMBERED,
840 		.procname	= "softlockup_panic",
841 		.data		= &softlockup_panic,
842 		.maxlen		= sizeof(int),
843 		.mode		= 0644,
844 		.proc_handler	= &proc_dointvec_minmax,
845 		.strategy	= &sysctl_intvec,
846 		.extra1		= &zero,
847 		.extra2		= &one,
848 	},
849 	{
850 		.ctl_name	= CTL_UNNUMBERED,
851 		.procname	= "softlockup_thresh",
852 		.data		= &softlockup_thresh,
853 		.maxlen		= sizeof(int),
854 		.mode		= 0644,
855 		.proc_handler	= &proc_dosoftlockup_thresh,
856 		.strategy	= &sysctl_intvec,
857 		.extra1		= &neg_one,
858 		.extra2		= &sixty,
859 	},
860 #endif
861 #ifdef CONFIG_DETECT_HUNG_TASK
862 	{
863 		.ctl_name	= CTL_UNNUMBERED,
864 		.procname	= "hung_task_panic",
865 		.data		= &sysctl_hung_task_panic,
866 		.maxlen		= sizeof(int),
867 		.mode		= 0644,
868 		.proc_handler	= &proc_dointvec_minmax,
869 		.strategy	= &sysctl_intvec,
870 		.extra1		= &zero,
871 		.extra2		= &one,
872 	},
873 	{
874 		.ctl_name	= CTL_UNNUMBERED,
875 		.procname	= "hung_task_check_count",
876 		.data		= &sysctl_hung_task_check_count,
877 		.maxlen		= sizeof(unsigned long),
878 		.mode		= 0644,
879 		.proc_handler	= &proc_doulongvec_minmax,
880 		.strategy	= &sysctl_intvec,
881 	},
882 	{
883 		.ctl_name	= CTL_UNNUMBERED,
884 		.procname	= "hung_task_timeout_secs",
885 		.data		= &sysctl_hung_task_timeout_secs,
886 		.maxlen		= sizeof(unsigned long),
887 		.mode		= 0644,
888 		.proc_handler	= &proc_dohung_task_timeout_secs,
889 		.strategy	= &sysctl_intvec,
890 	},
891 	{
892 		.ctl_name	= CTL_UNNUMBERED,
893 		.procname	= "hung_task_warnings",
894 		.data		= &sysctl_hung_task_warnings,
895 		.maxlen		= sizeof(unsigned long),
896 		.mode		= 0644,
897 		.proc_handler	= &proc_doulongvec_minmax,
898 		.strategy	= &sysctl_intvec,
899 	},
900 #endif
901 #ifdef CONFIG_COMPAT
902 	{
903 		.ctl_name	= KERN_COMPAT_LOG,
904 		.procname	= "compat-log",
905 		.data		= &compat_log,
906 		.maxlen		= sizeof (int),
907 	 	.mode		= 0644,
908 		.proc_handler	= &proc_dointvec,
909 	},
910 #endif
911 #ifdef CONFIG_RT_MUTEXES
912 	{
913 		.ctl_name	= KERN_MAX_LOCK_DEPTH,
914 		.procname	= "max_lock_depth",
915 		.data		= &max_lock_depth,
916 		.maxlen		= sizeof(int),
917 		.mode		= 0644,
918 		.proc_handler	= &proc_dointvec,
919 	},
920 #endif
921 	{
922 		.ctl_name	= CTL_UNNUMBERED,
923 		.procname	= "poweroff_cmd",
924 		.data		= &poweroff_cmd,
925 		.maxlen		= POWEROFF_CMD_PATH_LEN,
926 		.mode		= 0644,
927 		.proc_handler	= &proc_dostring,
928 		.strategy	= &sysctl_string,
929 	},
930 #ifdef CONFIG_KEYS
931 	{
932 		.ctl_name	= CTL_UNNUMBERED,
933 		.procname	= "keys",
934 		.mode		= 0555,
935 		.child		= key_sysctls,
936 	},
937 #endif
938 #ifdef CONFIG_RCU_TORTURE_TEST
939 	{
940 		.ctl_name       = CTL_UNNUMBERED,
941 		.procname       = "rcutorture_runnable",
942 		.data           = &rcutorture_runnable,
943 		.maxlen         = sizeof(int),
944 		.mode           = 0644,
945 		.proc_handler   = &proc_dointvec,
946 	},
947 #endif
948 #ifdef CONFIG_SLOW_WORK
949 	{
950 		.ctl_name	= CTL_UNNUMBERED,
951 		.procname	= "slow-work",
952 		.mode		= 0555,
953 		.child		= slow_work_sysctls,
954 	},
955 #endif
956 #ifdef CONFIG_PERF_COUNTERS
957 	{
958 		.ctl_name	= CTL_UNNUMBERED,
959 		.procname	= "perf_counter_paranoid",
960 		.data		= &sysctl_perf_counter_paranoid,
961 		.maxlen		= sizeof(sysctl_perf_counter_paranoid),
962 		.mode		= 0644,
963 		.proc_handler	= &proc_dointvec,
964 	},
965 	{
966 		.ctl_name	= CTL_UNNUMBERED,
967 		.procname	= "perf_counter_mlock_kb",
968 		.data		= &sysctl_perf_counter_mlock,
969 		.maxlen		= sizeof(sysctl_perf_counter_mlock),
970 		.mode		= 0644,
971 		.proc_handler	= &proc_dointvec,
972 	},
973 	{
974 		.ctl_name	= CTL_UNNUMBERED,
975 		.procname	= "perf_counter_max_sample_rate",
976 		.data		= &sysctl_perf_counter_sample_rate,
977 		.maxlen		= sizeof(sysctl_perf_counter_sample_rate),
978 		.mode		= 0644,
979 		.proc_handler	= &proc_dointvec,
980 	},
981 #endif
982 #ifdef CONFIG_KMEMCHECK
983 	{
984 		.ctl_name	= CTL_UNNUMBERED,
985 		.procname	= "kmemcheck",
986 		.data		= &kmemcheck_enabled,
987 		.maxlen		= sizeof(int),
988 		.mode		= 0644,
989 		.proc_handler	= &proc_dointvec,
990 	},
991 #endif
992 
993 /*
994  * NOTE: do not add new entries to this table unless you have read
995  * Documentation/sysctl/ctl_unnumbered.txt
996  */
997 	{ .ctl_name = 0 }
998 };
999 
1000 static struct ctl_table vm_table[] = {
1001 	{
1002 		.ctl_name	= VM_OVERCOMMIT_MEMORY,
1003 		.procname	= "overcommit_memory",
1004 		.data		= &sysctl_overcommit_memory,
1005 		.maxlen		= sizeof(sysctl_overcommit_memory),
1006 		.mode		= 0644,
1007 		.proc_handler	= &proc_dointvec,
1008 	},
1009 	{
1010 		.ctl_name	= VM_PANIC_ON_OOM,
1011 		.procname	= "panic_on_oom",
1012 		.data		= &sysctl_panic_on_oom,
1013 		.maxlen		= sizeof(sysctl_panic_on_oom),
1014 		.mode		= 0644,
1015 		.proc_handler	= &proc_dointvec,
1016 	},
1017 	{
1018 		.ctl_name	= CTL_UNNUMBERED,
1019 		.procname	= "oom_kill_allocating_task",
1020 		.data		= &sysctl_oom_kill_allocating_task,
1021 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1022 		.mode		= 0644,
1023 		.proc_handler	= &proc_dointvec,
1024 	},
1025 	{
1026 		.ctl_name	= CTL_UNNUMBERED,
1027 		.procname	= "oom_dump_tasks",
1028 		.data		= &sysctl_oom_dump_tasks,
1029 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1030 		.mode		= 0644,
1031 		.proc_handler	= &proc_dointvec,
1032 	},
1033 	{
1034 		.ctl_name	= VM_OVERCOMMIT_RATIO,
1035 		.procname	= "overcommit_ratio",
1036 		.data		= &sysctl_overcommit_ratio,
1037 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1038 		.mode		= 0644,
1039 		.proc_handler	= &proc_dointvec,
1040 	},
1041 	{
1042 		.ctl_name	= VM_PAGE_CLUSTER,
1043 		.procname	= "page-cluster",
1044 		.data		= &page_cluster,
1045 		.maxlen		= sizeof(int),
1046 		.mode		= 0644,
1047 		.proc_handler	= &proc_dointvec,
1048 	},
1049 	{
1050 		.ctl_name	= VM_DIRTY_BACKGROUND,
1051 		.procname	= "dirty_background_ratio",
1052 		.data		= &dirty_background_ratio,
1053 		.maxlen		= sizeof(dirty_background_ratio),
1054 		.mode		= 0644,
1055 		.proc_handler	= &dirty_background_ratio_handler,
1056 		.strategy	= &sysctl_intvec,
1057 		.extra1		= &zero,
1058 		.extra2		= &one_hundred,
1059 	},
1060 	{
1061 		.ctl_name	= CTL_UNNUMBERED,
1062 		.procname	= "dirty_background_bytes",
1063 		.data		= &dirty_background_bytes,
1064 		.maxlen		= sizeof(dirty_background_bytes),
1065 		.mode		= 0644,
1066 		.proc_handler	= &dirty_background_bytes_handler,
1067 		.strategy	= &sysctl_intvec,
1068 		.extra1		= &one_ul,
1069 	},
1070 	{
1071 		.ctl_name	= VM_DIRTY_RATIO,
1072 		.procname	= "dirty_ratio",
1073 		.data		= &vm_dirty_ratio,
1074 		.maxlen		= sizeof(vm_dirty_ratio),
1075 		.mode		= 0644,
1076 		.proc_handler	= &dirty_ratio_handler,
1077 		.strategy	= &sysctl_intvec,
1078 		.extra1		= &zero,
1079 		.extra2		= &one_hundred,
1080 	},
1081 	{
1082 		.ctl_name	= CTL_UNNUMBERED,
1083 		.procname	= "dirty_bytes",
1084 		.data		= &vm_dirty_bytes,
1085 		.maxlen		= sizeof(vm_dirty_bytes),
1086 		.mode		= 0644,
1087 		.proc_handler	= &dirty_bytes_handler,
1088 		.strategy	= &sysctl_intvec,
1089 		.extra1		= &dirty_bytes_min,
1090 	},
1091 	{
1092 		.procname	= "dirty_writeback_centisecs",
1093 		.data		= &dirty_writeback_interval,
1094 		.maxlen		= sizeof(dirty_writeback_interval),
1095 		.mode		= 0644,
1096 		.proc_handler	= &dirty_writeback_centisecs_handler,
1097 	},
1098 	{
1099 		.procname	= "dirty_expire_centisecs",
1100 		.data		= &dirty_expire_interval,
1101 		.maxlen		= sizeof(dirty_expire_interval),
1102 		.mode		= 0644,
1103 		.proc_handler	= &proc_dointvec,
1104 	},
1105 	{
1106 		.ctl_name	= VM_NR_PDFLUSH_THREADS,
1107 		.procname	= "nr_pdflush_threads",
1108 		.data		= &nr_pdflush_threads,
1109 		.maxlen		= sizeof nr_pdflush_threads,
1110 		.mode		= 0444 /* read-only*/,
1111 		.proc_handler	= &proc_dointvec,
1112 	},
1113 	{
1114 		.ctl_name	= VM_SWAPPINESS,
1115 		.procname	= "swappiness",
1116 		.data		= &vm_swappiness,
1117 		.maxlen		= sizeof(vm_swappiness),
1118 		.mode		= 0644,
1119 		.proc_handler	= &proc_dointvec_minmax,
1120 		.strategy	= &sysctl_intvec,
1121 		.extra1		= &zero,
1122 		.extra2		= &one_hundred,
1123 	},
1124 #ifdef CONFIG_HUGETLB_PAGE
1125 	 {
1126 		.procname	= "nr_hugepages",
1127 		.data		= NULL,
1128 		.maxlen		= sizeof(unsigned long),
1129 		.mode		= 0644,
1130 		.proc_handler	= &hugetlb_sysctl_handler,
1131 		.extra1		= (void *)&hugetlb_zero,
1132 		.extra2		= (void *)&hugetlb_infinity,
1133 	 },
1134 	 {
1135 		.ctl_name	= VM_HUGETLB_GROUP,
1136 		.procname	= "hugetlb_shm_group",
1137 		.data		= &sysctl_hugetlb_shm_group,
1138 		.maxlen		= sizeof(gid_t),
1139 		.mode		= 0644,
1140 		.proc_handler	= &proc_dointvec,
1141 	 },
1142 	 {
1143 		.ctl_name	= CTL_UNNUMBERED,
1144 		.procname	= "hugepages_treat_as_movable",
1145 		.data		= &hugepages_treat_as_movable,
1146 		.maxlen		= sizeof(int),
1147 		.mode		= 0644,
1148 		.proc_handler	= &hugetlb_treat_movable_handler,
1149 	},
1150 	{
1151 		.ctl_name	= CTL_UNNUMBERED,
1152 		.procname	= "nr_overcommit_hugepages",
1153 		.data		= NULL,
1154 		.maxlen		= sizeof(unsigned long),
1155 		.mode		= 0644,
1156 		.proc_handler	= &hugetlb_overcommit_handler,
1157 		.extra1		= (void *)&hugetlb_zero,
1158 		.extra2		= (void *)&hugetlb_infinity,
1159 	},
1160 #endif
1161 	{
1162 		.ctl_name	= VM_LOWMEM_RESERVE_RATIO,
1163 		.procname	= "lowmem_reserve_ratio",
1164 		.data		= &sysctl_lowmem_reserve_ratio,
1165 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1166 		.mode		= 0644,
1167 		.proc_handler	= &lowmem_reserve_ratio_sysctl_handler,
1168 		.strategy	= &sysctl_intvec,
1169 	},
1170 	{
1171 		.ctl_name	= VM_DROP_PAGECACHE,
1172 		.procname	= "drop_caches",
1173 		.data		= &sysctl_drop_caches,
1174 		.maxlen		= sizeof(int),
1175 		.mode		= 0644,
1176 		.proc_handler	= drop_caches_sysctl_handler,
1177 		.strategy	= &sysctl_intvec,
1178 	},
1179 	{
1180 		.ctl_name	= VM_MIN_FREE_KBYTES,
1181 		.procname	= "min_free_kbytes",
1182 		.data		= &min_free_kbytes,
1183 		.maxlen		= sizeof(min_free_kbytes),
1184 		.mode		= 0644,
1185 		.proc_handler	= &min_free_kbytes_sysctl_handler,
1186 		.strategy	= &sysctl_intvec,
1187 		.extra1		= &zero,
1188 	},
1189 	{
1190 		.ctl_name	= VM_PERCPU_PAGELIST_FRACTION,
1191 		.procname	= "percpu_pagelist_fraction",
1192 		.data		= &percpu_pagelist_fraction,
1193 		.maxlen		= sizeof(percpu_pagelist_fraction),
1194 		.mode		= 0644,
1195 		.proc_handler	= &percpu_pagelist_fraction_sysctl_handler,
1196 		.strategy	= &sysctl_intvec,
1197 		.extra1		= &min_percpu_pagelist_fract,
1198 	},
1199 #ifdef CONFIG_MMU
1200 	{
1201 		.ctl_name	= VM_MAX_MAP_COUNT,
1202 		.procname	= "max_map_count",
1203 		.data		= &sysctl_max_map_count,
1204 		.maxlen		= sizeof(sysctl_max_map_count),
1205 		.mode		= 0644,
1206 		.proc_handler	= &proc_dointvec
1207 	},
1208 #else
1209 	{
1210 		.ctl_name	= CTL_UNNUMBERED,
1211 		.procname	= "nr_trim_pages",
1212 		.data		= &sysctl_nr_trim_pages,
1213 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1214 		.mode		= 0644,
1215 		.proc_handler	= &proc_dointvec_minmax,
1216 		.strategy	= &sysctl_intvec,
1217 		.extra1		= &zero,
1218 	},
1219 #endif
1220 	{
1221 		.ctl_name	= VM_LAPTOP_MODE,
1222 		.procname	= "laptop_mode",
1223 		.data		= &laptop_mode,
1224 		.maxlen		= sizeof(laptop_mode),
1225 		.mode		= 0644,
1226 		.proc_handler	= &proc_dointvec_jiffies,
1227 		.strategy	= &sysctl_jiffies,
1228 	},
1229 	{
1230 		.ctl_name	= VM_BLOCK_DUMP,
1231 		.procname	= "block_dump",
1232 		.data		= &block_dump,
1233 		.maxlen		= sizeof(block_dump),
1234 		.mode		= 0644,
1235 		.proc_handler	= &proc_dointvec,
1236 		.strategy	= &sysctl_intvec,
1237 		.extra1		= &zero,
1238 	},
1239 	{
1240 		.ctl_name	= VM_VFS_CACHE_PRESSURE,
1241 		.procname	= "vfs_cache_pressure",
1242 		.data		= &sysctl_vfs_cache_pressure,
1243 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1244 		.mode		= 0644,
1245 		.proc_handler	= &proc_dointvec,
1246 		.strategy	= &sysctl_intvec,
1247 		.extra1		= &zero,
1248 	},
1249 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1250 	{
1251 		.ctl_name	= VM_LEGACY_VA_LAYOUT,
1252 		.procname	= "legacy_va_layout",
1253 		.data		= &sysctl_legacy_va_layout,
1254 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1255 		.mode		= 0644,
1256 		.proc_handler	= &proc_dointvec,
1257 		.strategy	= &sysctl_intvec,
1258 		.extra1		= &zero,
1259 	},
1260 #endif
1261 #ifdef CONFIG_NUMA
1262 	{
1263 		.ctl_name	= VM_ZONE_RECLAIM_MODE,
1264 		.procname	= "zone_reclaim_mode",
1265 		.data		= &zone_reclaim_mode,
1266 		.maxlen		= sizeof(zone_reclaim_mode),
1267 		.mode		= 0644,
1268 		.proc_handler	= &proc_dointvec,
1269 		.strategy	= &sysctl_intvec,
1270 		.extra1		= &zero,
1271 	},
1272 	{
1273 		.ctl_name	= VM_MIN_UNMAPPED,
1274 		.procname	= "min_unmapped_ratio",
1275 		.data		= &sysctl_min_unmapped_ratio,
1276 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1277 		.mode		= 0644,
1278 		.proc_handler	= &sysctl_min_unmapped_ratio_sysctl_handler,
1279 		.strategy	= &sysctl_intvec,
1280 		.extra1		= &zero,
1281 		.extra2		= &one_hundred,
1282 	},
1283 	{
1284 		.ctl_name	= VM_MIN_SLAB,
1285 		.procname	= "min_slab_ratio",
1286 		.data		= &sysctl_min_slab_ratio,
1287 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1288 		.mode		= 0644,
1289 		.proc_handler	= &sysctl_min_slab_ratio_sysctl_handler,
1290 		.strategy	= &sysctl_intvec,
1291 		.extra1		= &zero,
1292 		.extra2		= &one_hundred,
1293 	},
1294 #endif
1295 #ifdef CONFIG_SMP
1296 	{
1297 		.ctl_name	= CTL_UNNUMBERED,
1298 		.procname	= "stat_interval",
1299 		.data		= &sysctl_stat_interval,
1300 		.maxlen		= sizeof(sysctl_stat_interval),
1301 		.mode		= 0644,
1302 		.proc_handler	= &proc_dointvec_jiffies,
1303 		.strategy	= &sysctl_jiffies,
1304 	},
1305 #endif
1306 	{
1307 		.ctl_name	= CTL_UNNUMBERED,
1308 		.procname	= "mmap_min_addr",
1309 		.data		= &mmap_min_addr,
1310 		.maxlen         = sizeof(unsigned long),
1311 		.mode		= 0644,
1312 		.proc_handler	= &proc_doulongvec_minmax,
1313 	},
1314 #ifdef CONFIG_NUMA
1315 	{
1316 		.ctl_name	= CTL_UNNUMBERED,
1317 		.procname	= "numa_zonelist_order",
1318 		.data		= &numa_zonelist_order,
1319 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1320 		.mode		= 0644,
1321 		.proc_handler	= &numa_zonelist_order_handler,
1322 		.strategy	= &sysctl_string,
1323 	},
1324 #endif
1325 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1326    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1327 	{
1328 		.ctl_name	= VM_VDSO_ENABLED,
1329 		.procname	= "vdso_enabled",
1330 		.data		= &vdso_enabled,
1331 		.maxlen		= sizeof(vdso_enabled),
1332 		.mode		= 0644,
1333 		.proc_handler	= &proc_dointvec,
1334 		.strategy	= &sysctl_intvec,
1335 		.extra1		= &zero,
1336 	},
1337 #endif
1338 #ifdef CONFIG_HIGHMEM
1339 	{
1340 		.ctl_name	= CTL_UNNUMBERED,
1341 		.procname	= "highmem_is_dirtyable",
1342 		.data		= &vm_highmem_is_dirtyable,
1343 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1344 		.mode		= 0644,
1345 		.proc_handler	= &proc_dointvec_minmax,
1346 		.strategy	= &sysctl_intvec,
1347 		.extra1		= &zero,
1348 		.extra2		= &one,
1349 	},
1350 #endif
1351 	{
1352 		.ctl_name	= CTL_UNNUMBERED,
1353 		.procname	= "scan_unevictable_pages",
1354 		.data		= &scan_unevictable_pages,
1355 		.maxlen		= sizeof(scan_unevictable_pages),
1356 		.mode		= 0644,
1357 		.proc_handler	= &scan_unevictable_handler,
1358 	},
1359 /*
1360  * NOTE: do not add new entries to this table unless you have read
1361  * Documentation/sysctl/ctl_unnumbered.txt
1362  */
1363 	{ .ctl_name = 0 }
1364 };
1365 
1366 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1367 static struct ctl_table binfmt_misc_table[] = {
1368 	{ .ctl_name = 0 }
1369 };
1370 #endif
1371 
1372 static struct ctl_table fs_table[] = {
1373 	{
1374 		.ctl_name	= FS_NRINODE,
1375 		.procname	= "inode-nr",
1376 		.data		= &inodes_stat,
1377 		.maxlen		= 2*sizeof(int),
1378 		.mode		= 0444,
1379 		.proc_handler	= &proc_dointvec,
1380 	},
1381 	{
1382 		.ctl_name	= FS_STATINODE,
1383 		.procname	= "inode-state",
1384 		.data		= &inodes_stat,
1385 		.maxlen		= 7*sizeof(int),
1386 		.mode		= 0444,
1387 		.proc_handler	= &proc_dointvec,
1388 	},
1389 	{
1390 		.procname	= "file-nr",
1391 		.data		= &files_stat,
1392 		.maxlen		= 3*sizeof(int),
1393 		.mode		= 0444,
1394 		.proc_handler	= &proc_nr_files,
1395 	},
1396 	{
1397 		.ctl_name	= FS_MAXFILE,
1398 		.procname	= "file-max",
1399 		.data		= &files_stat.max_files,
1400 		.maxlen		= sizeof(int),
1401 		.mode		= 0644,
1402 		.proc_handler	= &proc_dointvec,
1403 	},
1404 	{
1405 		.ctl_name	= CTL_UNNUMBERED,
1406 		.procname	= "nr_open",
1407 		.data		= &sysctl_nr_open,
1408 		.maxlen		= sizeof(int),
1409 		.mode		= 0644,
1410 		.proc_handler	= &proc_dointvec_minmax,
1411 		.extra1		= &sysctl_nr_open_min,
1412 		.extra2		= &sysctl_nr_open_max,
1413 	},
1414 	{
1415 		.ctl_name	= FS_DENTRY,
1416 		.procname	= "dentry-state",
1417 		.data		= &dentry_stat,
1418 		.maxlen		= 6*sizeof(int),
1419 		.mode		= 0444,
1420 		.proc_handler	= &proc_dointvec,
1421 	},
1422 	{
1423 		.ctl_name	= FS_OVERFLOWUID,
1424 		.procname	= "overflowuid",
1425 		.data		= &fs_overflowuid,
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 	{
1434 		.ctl_name	= FS_OVERFLOWGID,
1435 		.procname	= "overflowgid",
1436 		.data		= &fs_overflowgid,
1437 		.maxlen		= sizeof(int),
1438 		.mode		= 0644,
1439 		.proc_handler	= &proc_dointvec_minmax,
1440 		.strategy	= &sysctl_intvec,
1441 		.extra1		= &minolduid,
1442 		.extra2		= &maxolduid,
1443 	},
1444 #ifdef CONFIG_FILE_LOCKING
1445 	{
1446 		.ctl_name	= FS_LEASES,
1447 		.procname	= "leases-enable",
1448 		.data		= &leases_enable,
1449 		.maxlen		= sizeof(int),
1450 		.mode		= 0644,
1451 		.proc_handler	= &proc_dointvec,
1452 	},
1453 #endif
1454 #ifdef CONFIG_DNOTIFY
1455 	{
1456 		.ctl_name	= FS_DIR_NOTIFY,
1457 		.procname	= "dir-notify-enable",
1458 		.data		= &dir_notify_enable,
1459 		.maxlen		= sizeof(int),
1460 		.mode		= 0644,
1461 		.proc_handler	= &proc_dointvec,
1462 	},
1463 #endif
1464 #ifdef CONFIG_MMU
1465 #ifdef CONFIG_FILE_LOCKING
1466 	{
1467 		.ctl_name	= FS_LEASE_TIME,
1468 		.procname	= "lease-break-time",
1469 		.data		= &lease_break_time,
1470 		.maxlen		= sizeof(int),
1471 		.mode		= 0644,
1472 		.proc_handler	= &proc_dointvec,
1473 	},
1474 #endif
1475 #ifdef CONFIG_AIO
1476 	{
1477 		.procname	= "aio-nr",
1478 		.data		= &aio_nr,
1479 		.maxlen		= sizeof(aio_nr),
1480 		.mode		= 0444,
1481 		.proc_handler	= &proc_doulongvec_minmax,
1482 	},
1483 	{
1484 		.procname	= "aio-max-nr",
1485 		.data		= &aio_max_nr,
1486 		.maxlen		= sizeof(aio_max_nr),
1487 		.mode		= 0644,
1488 		.proc_handler	= &proc_doulongvec_minmax,
1489 	},
1490 #endif /* CONFIG_AIO */
1491 #ifdef CONFIG_INOTIFY_USER
1492 	{
1493 		.ctl_name	= FS_INOTIFY,
1494 		.procname	= "inotify",
1495 		.mode		= 0555,
1496 		.child		= inotify_table,
1497 	},
1498 #endif
1499 #ifdef CONFIG_EPOLL
1500 	{
1501 		.procname	= "epoll",
1502 		.mode		= 0555,
1503 		.child		= epoll_table,
1504 	},
1505 #endif
1506 #endif
1507 	{
1508 		.ctl_name	= KERN_SETUID_DUMPABLE,
1509 		.procname	= "suid_dumpable",
1510 		.data		= &suid_dumpable,
1511 		.maxlen		= sizeof(int),
1512 		.mode		= 0644,
1513 		.proc_handler	= &proc_dointvec_minmax,
1514 		.strategy	= &sysctl_intvec,
1515 		.extra1		= &zero,
1516 		.extra2		= &two,
1517 	},
1518 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1519 	{
1520 		.ctl_name	= CTL_UNNUMBERED,
1521 		.procname	= "binfmt_misc",
1522 		.mode		= 0555,
1523 		.child		= binfmt_misc_table,
1524 	},
1525 #endif
1526 /*
1527  * NOTE: do not add new entries to this table unless you have read
1528  * Documentation/sysctl/ctl_unnumbered.txt
1529  */
1530 	{ .ctl_name = 0 }
1531 };
1532 
1533 static struct ctl_table debug_table[] = {
1534 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1535 	{
1536 		.ctl_name	= CTL_UNNUMBERED,
1537 		.procname	= "exception-trace",
1538 		.data		= &show_unhandled_signals,
1539 		.maxlen		= sizeof(int),
1540 		.mode		= 0644,
1541 		.proc_handler	= proc_dointvec
1542 	},
1543 #endif
1544 	{ .ctl_name = 0 }
1545 };
1546 
1547 static struct ctl_table dev_table[] = {
1548 	{ .ctl_name = 0 }
1549 };
1550 
1551 static DEFINE_SPINLOCK(sysctl_lock);
1552 
1553 /* called under sysctl_lock */
1554 static int use_table(struct ctl_table_header *p)
1555 {
1556 	if (unlikely(p->unregistering))
1557 		return 0;
1558 	p->used++;
1559 	return 1;
1560 }
1561 
1562 /* called under sysctl_lock */
1563 static void unuse_table(struct ctl_table_header *p)
1564 {
1565 	if (!--p->used)
1566 		if (unlikely(p->unregistering))
1567 			complete(p->unregistering);
1568 }
1569 
1570 /* called under sysctl_lock, will reacquire if has to wait */
1571 static void start_unregistering(struct ctl_table_header *p)
1572 {
1573 	/*
1574 	 * if p->used is 0, nobody will ever touch that entry again;
1575 	 * we'll eliminate all paths to it before dropping sysctl_lock
1576 	 */
1577 	if (unlikely(p->used)) {
1578 		struct completion wait;
1579 		init_completion(&wait);
1580 		p->unregistering = &wait;
1581 		spin_unlock(&sysctl_lock);
1582 		wait_for_completion(&wait);
1583 		spin_lock(&sysctl_lock);
1584 	} else {
1585 		/* anything non-NULL; we'll never dereference it */
1586 		p->unregistering = ERR_PTR(-EINVAL);
1587 	}
1588 	/*
1589 	 * do not remove from the list until nobody holds it; walking the
1590 	 * list in do_sysctl() relies on that.
1591 	 */
1592 	list_del_init(&p->ctl_entry);
1593 }
1594 
1595 void sysctl_head_get(struct ctl_table_header *head)
1596 {
1597 	spin_lock(&sysctl_lock);
1598 	head->count++;
1599 	spin_unlock(&sysctl_lock);
1600 }
1601 
1602 void sysctl_head_put(struct ctl_table_header *head)
1603 {
1604 	spin_lock(&sysctl_lock);
1605 	if (!--head->count)
1606 		kfree(head);
1607 	spin_unlock(&sysctl_lock);
1608 }
1609 
1610 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1611 {
1612 	if (!head)
1613 		BUG();
1614 	spin_lock(&sysctl_lock);
1615 	if (!use_table(head))
1616 		head = ERR_PTR(-ENOENT);
1617 	spin_unlock(&sysctl_lock);
1618 	return head;
1619 }
1620 
1621 void sysctl_head_finish(struct ctl_table_header *head)
1622 {
1623 	if (!head)
1624 		return;
1625 	spin_lock(&sysctl_lock);
1626 	unuse_table(head);
1627 	spin_unlock(&sysctl_lock);
1628 }
1629 
1630 static struct ctl_table_set *
1631 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1632 {
1633 	struct ctl_table_set *set = &root->default_set;
1634 	if (root->lookup)
1635 		set = root->lookup(root, namespaces);
1636 	return set;
1637 }
1638 
1639 static struct list_head *
1640 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1641 {
1642 	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1643 	return &set->list;
1644 }
1645 
1646 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1647 					    struct ctl_table_header *prev)
1648 {
1649 	struct ctl_table_root *root;
1650 	struct list_head *header_list;
1651 	struct ctl_table_header *head;
1652 	struct list_head *tmp;
1653 
1654 	spin_lock(&sysctl_lock);
1655 	if (prev) {
1656 		head = prev;
1657 		tmp = &prev->ctl_entry;
1658 		unuse_table(prev);
1659 		goto next;
1660 	}
1661 	tmp = &root_table_header.ctl_entry;
1662 	for (;;) {
1663 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1664 
1665 		if (!use_table(head))
1666 			goto next;
1667 		spin_unlock(&sysctl_lock);
1668 		return head;
1669 	next:
1670 		root = head->root;
1671 		tmp = tmp->next;
1672 		header_list = lookup_header_list(root, namespaces);
1673 		if (tmp != header_list)
1674 			continue;
1675 
1676 		do {
1677 			root = list_entry(root->root_list.next,
1678 					struct ctl_table_root, root_list);
1679 			if (root == &sysctl_table_root)
1680 				goto out;
1681 			header_list = lookup_header_list(root, namespaces);
1682 		} while (list_empty(header_list));
1683 		tmp = header_list->next;
1684 	}
1685 out:
1686 	spin_unlock(&sysctl_lock);
1687 	return NULL;
1688 }
1689 
1690 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1691 {
1692 	return __sysctl_head_next(current->nsproxy, prev);
1693 }
1694 
1695 void register_sysctl_root(struct ctl_table_root *root)
1696 {
1697 	spin_lock(&sysctl_lock);
1698 	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1699 	spin_unlock(&sysctl_lock);
1700 }
1701 
1702 #ifdef CONFIG_SYSCTL_SYSCALL
1703 /* Perform the actual read/write of a sysctl table entry. */
1704 static int do_sysctl_strategy(struct ctl_table_root *root,
1705 			struct ctl_table *table,
1706 			void __user *oldval, size_t __user *oldlenp,
1707 			void __user *newval, size_t newlen)
1708 {
1709 	int op = 0, rc;
1710 
1711 	if (oldval)
1712 		op |= MAY_READ;
1713 	if (newval)
1714 		op |= MAY_WRITE;
1715 	if (sysctl_perm(root, table, op))
1716 		return -EPERM;
1717 
1718 	if (table->strategy) {
1719 		rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1720 		if (rc < 0)
1721 			return rc;
1722 		if (rc > 0)
1723 			return 0;
1724 	}
1725 
1726 	/* If there is no strategy routine, or if the strategy returns
1727 	 * zero, proceed with automatic r/w */
1728 	if (table->data && table->maxlen) {
1729 		rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1730 		if (rc < 0)
1731 			return rc;
1732 	}
1733 	return 0;
1734 }
1735 
1736 static int parse_table(int __user *name, int nlen,
1737 		       void __user *oldval, size_t __user *oldlenp,
1738 		       void __user *newval, size_t newlen,
1739 		       struct ctl_table_root *root,
1740 		       struct ctl_table *table)
1741 {
1742 	int n;
1743 repeat:
1744 	if (!nlen)
1745 		return -ENOTDIR;
1746 	if (get_user(n, name))
1747 		return -EFAULT;
1748 	for ( ; table->ctl_name || table->procname; table++) {
1749 		if (!table->ctl_name)
1750 			continue;
1751 		if (n == table->ctl_name) {
1752 			int error;
1753 			if (table->child) {
1754 				if (sysctl_perm(root, table, MAY_EXEC))
1755 					return -EPERM;
1756 				name++;
1757 				nlen--;
1758 				table = table->child;
1759 				goto repeat;
1760 			}
1761 			error = do_sysctl_strategy(root, table,
1762 						   oldval, oldlenp,
1763 						   newval, newlen);
1764 			return error;
1765 		}
1766 	}
1767 	return -ENOTDIR;
1768 }
1769 
1770 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1771 	       void __user *newval, size_t newlen)
1772 {
1773 	struct ctl_table_header *head;
1774 	int error = -ENOTDIR;
1775 
1776 	if (nlen <= 0 || nlen >= CTL_MAXNAME)
1777 		return -ENOTDIR;
1778 	if (oldval) {
1779 		int old_len;
1780 		if (!oldlenp || get_user(old_len, oldlenp))
1781 			return -EFAULT;
1782 	}
1783 
1784 	for (head = sysctl_head_next(NULL); head;
1785 			head = sysctl_head_next(head)) {
1786 		error = parse_table(name, nlen, oldval, oldlenp,
1787 					newval, newlen,
1788 					head->root, head->ctl_table);
1789 		if (error != -ENOTDIR) {
1790 			sysctl_head_finish(head);
1791 			break;
1792 		}
1793 	}
1794 	return error;
1795 }
1796 
1797 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
1798 {
1799 	struct __sysctl_args tmp;
1800 	int error;
1801 
1802 	if (copy_from_user(&tmp, args, sizeof(tmp)))
1803 		return -EFAULT;
1804 
1805 	error = deprecated_sysctl_warning(&tmp);
1806 	if (error)
1807 		goto out;
1808 
1809 	lock_kernel();
1810 	error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1811 			  tmp.newval, tmp.newlen);
1812 	unlock_kernel();
1813 out:
1814 	return error;
1815 }
1816 #endif /* CONFIG_SYSCTL_SYSCALL */
1817 
1818 /*
1819  * sysctl_perm does NOT grant the superuser all rights automatically, because
1820  * some sysctl variables are readonly even to root.
1821  */
1822 
1823 static int test_perm(int mode, int op)
1824 {
1825 	if (!current_euid())
1826 		mode >>= 6;
1827 	else if (in_egroup_p(0))
1828 		mode >>= 3;
1829 	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1830 		return 0;
1831 	return -EACCES;
1832 }
1833 
1834 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1835 {
1836 	int error;
1837 	int mode;
1838 
1839 	error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1840 	if (error)
1841 		return error;
1842 
1843 	if (root->permissions)
1844 		mode = root->permissions(root, current->nsproxy, table);
1845 	else
1846 		mode = table->mode;
1847 
1848 	return test_perm(mode, op);
1849 }
1850 
1851 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1852 {
1853 	for (; table->ctl_name || table->procname; table++) {
1854 		table->parent = parent;
1855 		if (table->child)
1856 			sysctl_set_parent(table, table->child);
1857 	}
1858 }
1859 
1860 static __init int sysctl_init(void)
1861 {
1862 	sysctl_set_parent(NULL, root_table);
1863 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1864 	{
1865 		int err;
1866 		err = sysctl_check_table(current->nsproxy, root_table);
1867 	}
1868 #endif
1869 	return 0;
1870 }
1871 
1872 core_initcall(sysctl_init);
1873 
1874 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1875 				      struct ctl_table *table)
1876 {
1877 	struct ctl_table *p;
1878 	const char *s = branch->procname;
1879 
1880 	/* branch should have named subdirectory as its first element */
1881 	if (!s || !branch->child)
1882 		return NULL;
1883 
1884 	/* ... and nothing else */
1885 	if (branch[1].procname || branch[1].ctl_name)
1886 		return NULL;
1887 
1888 	/* table should contain subdirectory with the same name */
1889 	for (p = table; p->procname || p->ctl_name; p++) {
1890 		if (!p->child)
1891 			continue;
1892 		if (p->procname && strcmp(p->procname, s) == 0)
1893 			return p;
1894 	}
1895 	return NULL;
1896 }
1897 
1898 /* see if attaching q to p would be an improvement */
1899 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1900 {
1901 	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1902 	struct ctl_table *next;
1903 	int is_better = 0;
1904 	int not_in_parent = !p->attached_by;
1905 
1906 	while ((next = is_branch_in(by, to)) != NULL) {
1907 		if (by == q->attached_by)
1908 			is_better = 1;
1909 		if (to == p->attached_by)
1910 			not_in_parent = 1;
1911 		by = by->child;
1912 		to = next->child;
1913 	}
1914 
1915 	if (is_better && not_in_parent) {
1916 		q->attached_by = by;
1917 		q->attached_to = to;
1918 		q->parent = p;
1919 	}
1920 }
1921 
1922 /**
1923  * __register_sysctl_paths - register a sysctl hierarchy
1924  * @root: List of sysctl headers to register on
1925  * @namespaces: Data to compute which lists of sysctl entries are visible
1926  * @path: The path to the directory the sysctl table is in.
1927  * @table: the top-level table structure
1928  *
1929  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1930  * array. A completely 0 filled entry terminates the table.
1931  *
1932  * The members of the &struct ctl_table structure are used as follows:
1933  *
1934  * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1935  *            must be unique within that level of sysctl
1936  *
1937  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1938  *            enter a sysctl file
1939  *
1940  * data - a pointer to data for use by proc_handler
1941  *
1942  * maxlen - the maximum size in bytes of the data
1943  *
1944  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1945  *
1946  * child - a pointer to the child sysctl table if this entry is a directory, or
1947  *         %NULL.
1948  *
1949  * proc_handler - the text handler routine (described below)
1950  *
1951  * strategy - the strategy routine (described below)
1952  *
1953  * de - for internal use by the sysctl routines
1954  *
1955  * extra1, extra2 - extra pointers usable by the proc handler routines
1956  *
1957  * Leaf nodes in the sysctl tree will be represented by a single file
1958  * under /proc; non-leaf nodes will be represented by directories.
1959  *
1960  * sysctl(2) can automatically manage read and write requests through
1961  * the sysctl table.  The data and maxlen fields of the ctl_table
1962  * struct enable minimal validation of the values being written to be
1963  * performed, and the mode field allows minimal authentication.
1964  *
1965  * More sophisticated management can be enabled by the provision of a
1966  * strategy routine with the table entry.  This will be called before
1967  * any automatic read or write of the data is performed.
1968  *
1969  * The strategy routine may return
1970  *
1971  * < 0 - Error occurred (error is passed to user process)
1972  *
1973  * 0   - OK - proceed with automatic read or write.
1974  *
1975  * > 0 - OK - read or write has been done by the strategy routine, so
1976  *       return immediately.
1977  *
1978  * There must be a proc_handler routine for any terminal nodes
1979  * mirrored under /proc/sys (non-terminals are handled by a built-in
1980  * directory handler).  Several default handlers are available to
1981  * cover common cases -
1982  *
1983  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1984  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1985  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1986  *
1987  * It is the handler's job to read the input buffer from user memory
1988  * and process it. The handler should return 0 on success.
1989  *
1990  * This routine returns %NULL on a failure to register, and a pointer
1991  * to the table header on success.
1992  */
1993 struct ctl_table_header *__register_sysctl_paths(
1994 	struct ctl_table_root *root,
1995 	struct nsproxy *namespaces,
1996 	const struct ctl_path *path, struct ctl_table *table)
1997 {
1998 	struct ctl_table_header *header;
1999 	struct ctl_table *new, **prevp;
2000 	unsigned int n, npath;
2001 	struct ctl_table_set *set;
2002 
2003 	/* Count the path components */
2004 	for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
2005 		;
2006 
2007 	/*
2008 	 * For each path component, allocate a 2-element ctl_table array.
2009 	 * The first array element will be filled with the sysctl entry
2010 	 * for this, the second will be the sentinel (ctl_name == 0).
2011 	 *
2012 	 * We allocate everything in one go so that we don't have to
2013 	 * worry about freeing additional memory in unregister_sysctl_table.
2014 	 */
2015 	header = kzalloc(sizeof(struct ctl_table_header) +
2016 			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
2017 	if (!header)
2018 		return NULL;
2019 
2020 	new = (struct ctl_table *) (header + 1);
2021 
2022 	/* Now connect the dots */
2023 	prevp = &header->ctl_table;
2024 	for (n = 0; n < npath; ++n, ++path) {
2025 		/* Copy the procname */
2026 		new->procname = path->procname;
2027 		new->ctl_name = path->ctl_name;
2028 		new->mode     = 0555;
2029 
2030 		*prevp = new;
2031 		prevp = &new->child;
2032 
2033 		new += 2;
2034 	}
2035 	*prevp = table;
2036 	header->ctl_table_arg = table;
2037 
2038 	INIT_LIST_HEAD(&header->ctl_entry);
2039 	header->used = 0;
2040 	header->unregistering = NULL;
2041 	header->root = root;
2042 	sysctl_set_parent(NULL, header->ctl_table);
2043 	header->count = 1;
2044 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
2045 	if (sysctl_check_table(namespaces, header->ctl_table)) {
2046 		kfree(header);
2047 		return NULL;
2048 	}
2049 #endif
2050 	spin_lock(&sysctl_lock);
2051 	header->set = lookup_header_set(root, namespaces);
2052 	header->attached_by = header->ctl_table;
2053 	header->attached_to = root_table;
2054 	header->parent = &root_table_header;
2055 	for (set = header->set; set; set = set->parent) {
2056 		struct ctl_table_header *p;
2057 		list_for_each_entry(p, &set->list, ctl_entry) {
2058 			if (p->unregistering)
2059 				continue;
2060 			try_attach(p, header);
2061 		}
2062 	}
2063 	header->parent->count++;
2064 	list_add_tail(&header->ctl_entry, &header->set->list);
2065 	spin_unlock(&sysctl_lock);
2066 
2067 	return header;
2068 }
2069 
2070 /**
2071  * register_sysctl_table_path - register a sysctl table hierarchy
2072  * @path: The path to the directory the sysctl table is in.
2073  * @table: the top-level table structure
2074  *
2075  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2076  * array. A completely 0 filled entry terminates the table.
2077  *
2078  * See __register_sysctl_paths for more details.
2079  */
2080 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2081 						struct ctl_table *table)
2082 {
2083 	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
2084 					path, table);
2085 }
2086 
2087 /**
2088  * register_sysctl_table - register a sysctl table hierarchy
2089  * @table: the top-level table structure
2090  *
2091  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2092  * array. A completely 0 filled entry terminates the table.
2093  *
2094  * See register_sysctl_paths for more details.
2095  */
2096 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
2097 {
2098 	static const struct ctl_path null_path[] = { {} };
2099 
2100 	return register_sysctl_paths(null_path, table);
2101 }
2102 
2103 /**
2104  * unregister_sysctl_table - unregister a sysctl table hierarchy
2105  * @header: the header returned from register_sysctl_table
2106  *
2107  * Unregisters the sysctl table and all children. proc entries may not
2108  * actually be removed until they are no longer used by anyone.
2109  */
2110 void unregister_sysctl_table(struct ctl_table_header * header)
2111 {
2112 	might_sleep();
2113 
2114 	if (header == NULL)
2115 		return;
2116 
2117 	spin_lock(&sysctl_lock);
2118 	start_unregistering(header);
2119 	if (!--header->parent->count) {
2120 		WARN_ON(1);
2121 		kfree(header->parent);
2122 	}
2123 	if (!--header->count)
2124 		kfree(header);
2125 	spin_unlock(&sysctl_lock);
2126 }
2127 
2128 int sysctl_is_seen(struct ctl_table_header *p)
2129 {
2130 	struct ctl_table_set *set = p->set;
2131 	int res;
2132 	spin_lock(&sysctl_lock);
2133 	if (p->unregistering)
2134 		res = 0;
2135 	else if (!set->is_seen)
2136 		res = 1;
2137 	else
2138 		res = set->is_seen(set);
2139 	spin_unlock(&sysctl_lock);
2140 	return res;
2141 }
2142 
2143 void setup_sysctl_set(struct ctl_table_set *p,
2144 	struct ctl_table_set *parent,
2145 	int (*is_seen)(struct ctl_table_set *))
2146 {
2147 	INIT_LIST_HEAD(&p->list);
2148 	p->parent = parent ? parent : &sysctl_table_root.default_set;
2149 	p->is_seen = is_seen;
2150 }
2151 
2152 #else /* !CONFIG_SYSCTL */
2153 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2154 {
2155 	return NULL;
2156 }
2157 
2158 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2159 						    struct ctl_table *table)
2160 {
2161 	return NULL;
2162 }
2163 
2164 void unregister_sysctl_table(struct ctl_table_header * table)
2165 {
2166 }
2167 
2168 void setup_sysctl_set(struct ctl_table_set *p,
2169 	struct ctl_table_set *parent,
2170 	int (*is_seen)(struct ctl_table_set *))
2171 {
2172 }
2173 
2174 void sysctl_head_put(struct ctl_table_header *head)
2175 {
2176 }
2177 
2178 #endif /* CONFIG_SYSCTL */
2179 
2180 /*
2181  * /proc/sys support
2182  */
2183 
2184 #ifdef CONFIG_PROC_SYSCTL
2185 
2186 static int _proc_do_string(void* data, int maxlen, int write,
2187 			   struct file *filp, void __user *buffer,
2188 			   size_t *lenp, loff_t *ppos)
2189 {
2190 	size_t len;
2191 	char __user *p;
2192 	char c;
2193 
2194 	if (!data || !maxlen || !*lenp) {
2195 		*lenp = 0;
2196 		return 0;
2197 	}
2198 
2199 	if (write) {
2200 		len = 0;
2201 		p = buffer;
2202 		while (len < *lenp) {
2203 			if (get_user(c, p++))
2204 				return -EFAULT;
2205 			if (c == 0 || c == '\n')
2206 				break;
2207 			len++;
2208 		}
2209 		if (len >= maxlen)
2210 			len = maxlen-1;
2211 		if(copy_from_user(data, buffer, len))
2212 			return -EFAULT;
2213 		((char *) data)[len] = 0;
2214 		*ppos += *lenp;
2215 	} else {
2216 		len = strlen(data);
2217 		if (len > maxlen)
2218 			len = maxlen;
2219 
2220 		if (*ppos > len) {
2221 			*lenp = 0;
2222 			return 0;
2223 		}
2224 
2225 		data += *ppos;
2226 		len  -= *ppos;
2227 
2228 		if (len > *lenp)
2229 			len = *lenp;
2230 		if (len)
2231 			if(copy_to_user(buffer, data, len))
2232 				return -EFAULT;
2233 		if (len < *lenp) {
2234 			if(put_user('\n', ((char __user *) buffer) + len))
2235 				return -EFAULT;
2236 			len++;
2237 		}
2238 		*lenp = len;
2239 		*ppos += len;
2240 	}
2241 	return 0;
2242 }
2243 
2244 /**
2245  * proc_dostring - read a string sysctl
2246  * @table: the sysctl table
2247  * @write: %TRUE if this is a write to the sysctl file
2248  * @filp: the file structure
2249  * @buffer: the user buffer
2250  * @lenp: the size of the user buffer
2251  * @ppos: file position
2252  *
2253  * Reads/writes a string from/to the user buffer. If the kernel
2254  * buffer provided is not large enough to hold the string, the
2255  * string is truncated. The copied string is %NULL-terminated.
2256  * If the string is being read by the user process, it is copied
2257  * and a newline '\n' is added. It is truncated if the buffer is
2258  * not large enough.
2259  *
2260  * Returns 0 on success.
2261  */
2262 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2263 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2264 {
2265 	return _proc_do_string(table->data, table->maxlen, write, filp,
2266 			       buffer, lenp, ppos);
2267 }
2268 
2269 
2270 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2271 				 int *valp,
2272 				 int write, void *data)
2273 {
2274 	if (write) {
2275 		*valp = *negp ? -*lvalp : *lvalp;
2276 	} else {
2277 		int val = *valp;
2278 		if (val < 0) {
2279 			*negp = -1;
2280 			*lvalp = (unsigned long)-val;
2281 		} else {
2282 			*negp = 0;
2283 			*lvalp = (unsigned long)val;
2284 		}
2285 	}
2286 	return 0;
2287 }
2288 
2289 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2290 		  int write, struct file *filp, void __user *buffer,
2291 		  size_t *lenp, loff_t *ppos,
2292 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2293 			      int write, void *data),
2294 		  void *data)
2295 {
2296 #define TMPBUFLEN 21
2297 	int *i, vleft, first = 1, neg;
2298 	unsigned long lval;
2299 	size_t left, len;
2300 
2301 	char buf[TMPBUFLEN], *p;
2302 	char __user *s = buffer;
2303 
2304 	if (!tbl_data || !table->maxlen || !*lenp ||
2305 	    (*ppos && !write)) {
2306 		*lenp = 0;
2307 		return 0;
2308 	}
2309 
2310 	i = (int *) tbl_data;
2311 	vleft = table->maxlen / sizeof(*i);
2312 	left = *lenp;
2313 
2314 	if (!conv)
2315 		conv = do_proc_dointvec_conv;
2316 
2317 	for (; left && vleft--; i++, first=0) {
2318 		if (write) {
2319 			while (left) {
2320 				char c;
2321 				if (get_user(c, s))
2322 					return -EFAULT;
2323 				if (!isspace(c))
2324 					break;
2325 				left--;
2326 				s++;
2327 			}
2328 			if (!left)
2329 				break;
2330 			neg = 0;
2331 			len = left;
2332 			if (len > sizeof(buf) - 1)
2333 				len = sizeof(buf) - 1;
2334 			if (copy_from_user(buf, s, len))
2335 				return -EFAULT;
2336 			buf[len] = 0;
2337 			p = buf;
2338 			if (*p == '-' && left > 1) {
2339 				neg = 1;
2340 				p++;
2341 			}
2342 			if (*p < '0' || *p > '9')
2343 				break;
2344 
2345 			lval = simple_strtoul(p, &p, 0);
2346 
2347 			len = p-buf;
2348 			if ((len < left) && *p && !isspace(*p))
2349 				break;
2350 			s += len;
2351 			left -= len;
2352 
2353 			if (conv(&neg, &lval, i, 1, data))
2354 				break;
2355 		} else {
2356 			p = buf;
2357 			if (!first)
2358 				*p++ = '\t';
2359 
2360 			if (conv(&neg, &lval, i, 0, data))
2361 				break;
2362 
2363 			sprintf(p, "%s%lu", neg ? "-" : "", lval);
2364 			len = strlen(buf);
2365 			if (len > left)
2366 				len = left;
2367 			if(copy_to_user(s, buf, len))
2368 				return -EFAULT;
2369 			left -= len;
2370 			s += len;
2371 		}
2372 	}
2373 
2374 	if (!write && !first && left) {
2375 		if(put_user('\n', s))
2376 			return -EFAULT;
2377 		left--, s++;
2378 	}
2379 	if (write) {
2380 		while (left) {
2381 			char c;
2382 			if (get_user(c, s++))
2383 				return -EFAULT;
2384 			if (!isspace(c))
2385 				break;
2386 			left--;
2387 		}
2388 	}
2389 	if (write && first)
2390 		return -EINVAL;
2391 	*lenp -= left;
2392 	*ppos += *lenp;
2393 	return 0;
2394 #undef TMPBUFLEN
2395 }
2396 
2397 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2398 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2399 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2400 			      int write, void *data),
2401 		  void *data)
2402 {
2403 	return __do_proc_dointvec(table->data, table, write, filp,
2404 			buffer, lenp, ppos, conv, data);
2405 }
2406 
2407 /**
2408  * proc_dointvec - read a vector of integers
2409  * @table: the sysctl table
2410  * @write: %TRUE if this is a write to the sysctl file
2411  * @filp: the file structure
2412  * @buffer: the user buffer
2413  * @lenp: the size of the user buffer
2414  * @ppos: file position
2415  *
2416  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2417  * values from/to the user buffer, treated as an ASCII string.
2418  *
2419  * Returns 0 on success.
2420  */
2421 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2422 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2423 {
2424     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2425 		    	    NULL,NULL);
2426 }
2427 
2428 /*
2429  * Taint values can only be increased
2430  * This means we can safely use a temporary.
2431  */
2432 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
2433 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2434 {
2435 	struct ctl_table t;
2436 	unsigned long tmptaint = get_taint();
2437 	int err;
2438 
2439 	if (write && !capable(CAP_SYS_ADMIN))
2440 		return -EPERM;
2441 
2442 	t = *table;
2443 	t.data = &tmptaint;
2444 	err = proc_doulongvec_minmax(&t, write, filp, buffer, lenp, ppos);
2445 	if (err < 0)
2446 		return err;
2447 
2448 	if (write) {
2449 		/*
2450 		 * Poor man's atomic or. Not worth adding a primitive
2451 		 * to everyone's atomic.h for this
2452 		 */
2453 		int i;
2454 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2455 			if ((tmptaint >> i) & 1)
2456 				add_taint(i);
2457 		}
2458 	}
2459 
2460 	return err;
2461 }
2462 
2463 struct do_proc_dointvec_minmax_conv_param {
2464 	int *min;
2465 	int *max;
2466 };
2467 
2468 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2469 					int *valp,
2470 					int write, void *data)
2471 {
2472 	struct do_proc_dointvec_minmax_conv_param *param = data;
2473 	if (write) {
2474 		int val = *negp ? -*lvalp : *lvalp;
2475 		if ((param->min && *param->min > val) ||
2476 		    (param->max && *param->max < val))
2477 			return -EINVAL;
2478 		*valp = val;
2479 	} else {
2480 		int val = *valp;
2481 		if (val < 0) {
2482 			*negp = -1;
2483 			*lvalp = (unsigned long)-val;
2484 		} else {
2485 			*negp = 0;
2486 			*lvalp = (unsigned long)val;
2487 		}
2488 	}
2489 	return 0;
2490 }
2491 
2492 /**
2493  * proc_dointvec_minmax - read a vector of integers with min/max values
2494  * @table: the sysctl table
2495  * @write: %TRUE if this is a write to the sysctl file
2496  * @filp: the file structure
2497  * @buffer: the user buffer
2498  * @lenp: the size of the user buffer
2499  * @ppos: file position
2500  *
2501  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2502  * values from/to the user buffer, treated as an ASCII string.
2503  *
2504  * This routine will ensure the values are within the range specified by
2505  * table->extra1 (min) and table->extra2 (max).
2506  *
2507  * Returns 0 on success.
2508  */
2509 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2510 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2511 {
2512 	struct do_proc_dointvec_minmax_conv_param param = {
2513 		.min = (int *) table->extra1,
2514 		.max = (int *) table->extra2,
2515 	};
2516 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2517 				do_proc_dointvec_minmax_conv, &param);
2518 }
2519 
2520 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2521 				     struct file *filp,
2522 				     void __user *buffer,
2523 				     size_t *lenp, loff_t *ppos,
2524 				     unsigned long convmul,
2525 				     unsigned long convdiv)
2526 {
2527 #define TMPBUFLEN 21
2528 	unsigned long *i, *min, *max, val;
2529 	int vleft, first=1, neg;
2530 	size_t len, left;
2531 	char buf[TMPBUFLEN], *p;
2532 	char __user *s = buffer;
2533 
2534 	if (!data || !table->maxlen || !*lenp ||
2535 	    (*ppos && !write)) {
2536 		*lenp = 0;
2537 		return 0;
2538 	}
2539 
2540 	i = (unsigned long *) data;
2541 	min = (unsigned long *) table->extra1;
2542 	max = (unsigned long *) table->extra2;
2543 	vleft = table->maxlen / sizeof(unsigned long);
2544 	left = *lenp;
2545 
2546 	for (; left && vleft--; i++, min++, max++, first=0) {
2547 		if (write) {
2548 			while (left) {
2549 				char c;
2550 				if (get_user(c, s))
2551 					return -EFAULT;
2552 				if (!isspace(c))
2553 					break;
2554 				left--;
2555 				s++;
2556 			}
2557 			if (!left)
2558 				break;
2559 			neg = 0;
2560 			len = left;
2561 			if (len > TMPBUFLEN-1)
2562 				len = TMPBUFLEN-1;
2563 			if (copy_from_user(buf, s, len))
2564 				return -EFAULT;
2565 			buf[len] = 0;
2566 			p = buf;
2567 			if (*p == '-' && left > 1) {
2568 				neg = 1;
2569 				p++;
2570 			}
2571 			if (*p < '0' || *p > '9')
2572 				break;
2573 			val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2574 			len = p-buf;
2575 			if ((len < left) && *p && !isspace(*p))
2576 				break;
2577 			if (neg)
2578 				val = -val;
2579 			s += len;
2580 			left -= len;
2581 
2582 			if(neg)
2583 				continue;
2584 			if ((min && val < *min) || (max && val > *max))
2585 				continue;
2586 			*i = val;
2587 		} else {
2588 			p = buf;
2589 			if (!first)
2590 				*p++ = '\t';
2591 			sprintf(p, "%lu", convdiv * (*i) / convmul);
2592 			len = strlen(buf);
2593 			if (len > left)
2594 				len = left;
2595 			if(copy_to_user(s, buf, len))
2596 				return -EFAULT;
2597 			left -= len;
2598 			s += len;
2599 		}
2600 	}
2601 
2602 	if (!write && !first && left) {
2603 		if(put_user('\n', s))
2604 			return -EFAULT;
2605 		left--, s++;
2606 	}
2607 	if (write) {
2608 		while (left) {
2609 			char c;
2610 			if (get_user(c, s++))
2611 				return -EFAULT;
2612 			if (!isspace(c))
2613 				break;
2614 			left--;
2615 		}
2616 	}
2617 	if (write && first)
2618 		return -EINVAL;
2619 	*lenp -= left;
2620 	*ppos += *lenp;
2621 	return 0;
2622 #undef TMPBUFLEN
2623 }
2624 
2625 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2626 				     struct file *filp,
2627 				     void __user *buffer,
2628 				     size_t *lenp, loff_t *ppos,
2629 				     unsigned long convmul,
2630 				     unsigned long convdiv)
2631 {
2632 	return __do_proc_doulongvec_minmax(table->data, table, write,
2633 			filp, buffer, lenp, ppos, convmul, convdiv);
2634 }
2635 
2636 /**
2637  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2638  * @table: the sysctl table
2639  * @write: %TRUE if this is a write to the sysctl file
2640  * @filp: the file structure
2641  * @buffer: the user buffer
2642  * @lenp: the size of the user buffer
2643  * @ppos: file position
2644  *
2645  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2646  * values from/to the user buffer, treated as an ASCII string.
2647  *
2648  * This routine will ensure the values are within the range specified by
2649  * table->extra1 (min) and table->extra2 (max).
2650  *
2651  * Returns 0 on success.
2652  */
2653 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2654 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2655 {
2656     return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2657 }
2658 
2659 /**
2660  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2661  * @table: the sysctl table
2662  * @write: %TRUE if this is a write to the sysctl file
2663  * @filp: the file structure
2664  * @buffer: the user buffer
2665  * @lenp: the size of the user buffer
2666  * @ppos: file position
2667  *
2668  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2669  * values from/to the user buffer, treated as an ASCII string. The values
2670  * are treated as milliseconds, and converted to jiffies when they are stored.
2671  *
2672  * This routine will ensure the values are within the range specified by
2673  * table->extra1 (min) and table->extra2 (max).
2674  *
2675  * Returns 0 on success.
2676  */
2677 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2678 				      struct file *filp,
2679 				      void __user *buffer,
2680 				      size_t *lenp, loff_t *ppos)
2681 {
2682     return do_proc_doulongvec_minmax(table, write, filp, buffer,
2683 				     lenp, ppos, HZ, 1000l);
2684 }
2685 
2686 
2687 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2688 					 int *valp,
2689 					 int write, void *data)
2690 {
2691 	if (write) {
2692 		if (*lvalp > LONG_MAX / HZ)
2693 			return 1;
2694 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2695 	} else {
2696 		int val = *valp;
2697 		unsigned long lval;
2698 		if (val < 0) {
2699 			*negp = -1;
2700 			lval = (unsigned long)-val;
2701 		} else {
2702 			*negp = 0;
2703 			lval = (unsigned long)val;
2704 		}
2705 		*lvalp = lval / HZ;
2706 	}
2707 	return 0;
2708 }
2709 
2710 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2711 						int *valp,
2712 						int write, void *data)
2713 {
2714 	if (write) {
2715 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2716 			return 1;
2717 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2718 	} else {
2719 		int val = *valp;
2720 		unsigned long lval;
2721 		if (val < 0) {
2722 			*negp = -1;
2723 			lval = (unsigned long)-val;
2724 		} else {
2725 			*negp = 0;
2726 			lval = (unsigned long)val;
2727 		}
2728 		*lvalp = jiffies_to_clock_t(lval);
2729 	}
2730 	return 0;
2731 }
2732 
2733 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2734 					    int *valp,
2735 					    int write, void *data)
2736 {
2737 	if (write) {
2738 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2739 	} else {
2740 		int val = *valp;
2741 		unsigned long lval;
2742 		if (val < 0) {
2743 			*negp = -1;
2744 			lval = (unsigned long)-val;
2745 		} else {
2746 			*negp = 0;
2747 			lval = (unsigned long)val;
2748 		}
2749 		*lvalp = jiffies_to_msecs(lval);
2750 	}
2751 	return 0;
2752 }
2753 
2754 /**
2755  * proc_dointvec_jiffies - read a vector of integers as seconds
2756  * @table: the sysctl table
2757  * @write: %TRUE if this is a write to the sysctl file
2758  * @filp: the file structure
2759  * @buffer: the user buffer
2760  * @lenp: the size of the user buffer
2761  * @ppos: file position
2762  *
2763  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2764  * values from/to the user buffer, treated as an ASCII string.
2765  * The values read are assumed to be in seconds, and are converted into
2766  * jiffies.
2767  *
2768  * Returns 0 on success.
2769  */
2770 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2771 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2772 {
2773     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2774 		    	    do_proc_dointvec_jiffies_conv,NULL);
2775 }
2776 
2777 /**
2778  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2779  * @table: the sysctl table
2780  * @write: %TRUE if this is a write to the sysctl file
2781  * @filp: the file structure
2782  * @buffer: the user buffer
2783  * @lenp: the size of the user buffer
2784  * @ppos: pointer to the file position
2785  *
2786  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2787  * values from/to the user buffer, treated as an ASCII string.
2788  * The values read are assumed to be in 1/USER_HZ seconds, and
2789  * are converted into jiffies.
2790  *
2791  * Returns 0 on success.
2792  */
2793 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2794 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2795 {
2796     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2797 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2798 }
2799 
2800 /**
2801  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2802  * @table: the sysctl table
2803  * @write: %TRUE if this is a write to the sysctl file
2804  * @filp: the file structure
2805  * @buffer: the user buffer
2806  * @lenp: the size of the user buffer
2807  * @ppos: file position
2808  * @ppos: the current position in the file
2809  *
2810  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2811  * values from/to the user buffer, treated as an ASCII string.
2812  * The values read are assumed to be in 1/1000 seconds, and
2813  * are converted into jiffies.
2814  *
2815  * Returns 0 on success.
2816  */
2817 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2818 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2819 {
2820 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2821 				do_proc_dointvec_ms_jiffies_conv, NULL);
2822 }
2823 
2824 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2825 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2826 {
2827 	struct pid *new_pid;
2828 	pid_t tmp;
2829 	int r;
2830 
2831 	tmp = pid_vnr(cad_pid);
2832 
2833 	r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2834 			       lenp, ppos, NULL, NULL);
2835 	if (r || !write)
2836 		return r;
2837 
2838 	new_pid = find_get_pid(tmp);
2839 	if (!new_pid)
2840 		return -ESRCH;
2841 
2842 	put_pid(xchg(&cad_pid, new_pid));
2843 	return 0;
2844 }
2845 
2846 #else /* CONFIG_PROC_FS */
2847 
2848 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2849 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2850 {
2851 	return -ENOSYS;
2852 }
2853 
2854 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2855 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2856 {
2857 	return -ENOSYS;
2858 }
2859 
2860 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2861 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2862 {
2863 	return -ENOSYS;
2864 }
2865 
2866 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2867 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2868 {
2869 	return -ENOSYS;
2870 }
2871 
2872 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2873 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2874 {
2875 	return -ENOSYS;
2876 }
2877 
2878 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2879 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2880 {
2881 	return -ENOSYS;
2882 }
2883 
2884 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2885 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2886 {
2887 	return -ENOSYS;
2888 }
2889 
2890 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2891 				      struct file *filp,
2892 				      void __user *buffer,
2893 				      size_t *lenp, loff_t *ppos)
2894 {
2895     return -ENOSYS;
2896 }
2897 
2898 
2899 #endif /* CONFIG_PROC_FS */
2900 
2901 
2902 #ifdef CONFIG_SYSCTL_SYSCALL
2903 /*
2904  * General sysctl support routines
2905  */
2906 
2907 /* The generic sysctl data routine (used if no strategy routine supplied) */
2908 int sysctl_data(struct ctl_table *table,
2909 		void __user *oldval, size_t __user *oldlenp,
2910 		void __user *newval, size_t newlen)
2911 {
2912 	size_t len;
2913 
2914 	/* Get out of I don't have a variable */
2915 	if (!table->data || !table->maxlen)
2916 		return -ENOTDIR;
2917 
2918 	if (oldval && oldlenp) {
2919 		if (get_user(len, oldlenp))
2920 			return -EFAULT;
2921 		if (len) {
2922 			if (len > table->maxlen)
2923 				len = table->maxlen;
2924 			if (copy_to_user(oldval, table->data, len))
2925 				return -EFAULT;
2926 			if (put_user(len, oldlenp))
2927 				return -EFAULT;
2928 		}
2929 	}
2930 
2931 	if (newval && newlen) {
2932 		if (newlen > table->maxlen)
2933 			newlen = table->maxlen;
2934 
2935 		if (copy_from_user(table->data, newval, newlen))
2936 			return -EFAULT;
2937 	}
2938 	return 1;
2939 }
2940 
2941 /* The generic string strategy routine: */
2942 int sysctl_string(struct ctl_table *table,
2943 		  void __user *oldval, size_t __user *oldlenp,
2944 		  void __user *newval, size_t newlen)
2945 {
2946 	if (!table->data || !table->maxlen)
2947 		return -ENOTDIR;
2948 
2949 	if (oldval && oldlenp) {
2950 		size_t bufsize;
2951 		if (get_user(bufsize, oldlenp))
2952 			return -EFAULT;
2953 		if (bufsize) {
2954 			size_t len = strlen(table->data), copied;
2955 
2956 			/* This shouldn't trigger for a well-formed sysctl */
2957 			if (len > table->maxlen)
2958 				len = table->maxlen;
2959 
2960 			/* Copy up to a max of bufsize-1 bytes of the string */
2961 			copied = (len >= bufsize) ? bufsize - 1 : len;
2962 
2963 			if (copy_to_user(oldval, table->data, copied) ||
2964 			    put_user(0, (char __user *)(oldval + copied)))
2965 				return -EFAULT;
2966 			if (put_user(len, oldlenp))
2967 				return -EFAULT;
2968 		}
2969 	}
2970 	if (newval && newlen) {
2971 		size_t len = newlen;
2972 		if (len > table->maxlen)
2973 			len = table->maxlen;
2974 		if(copy_from_user(table->data, newval, len))
2975 			return -EFAULT;
2976 		if (len == table->maxlen)
2977 			len--;
2978 		((char *) table->data)[len] = 0;
2979 	}
2980 	return 1;
2981 }
2982 
2983 /*
2984  * This function makes sure that all of the integers in the vector
2985  * are between the minimum and maximum values given in the arrays
2986  * table->extra1 and table->extra2, respectively.
2987  */
2988 int sysctl_intvec(struct ctl_table *table,
2989 		void __user *oldval, size_t __user *oldlenp,
2990 		void __user *newval, size_t newlen)
2991 {
2992 
2993 	if (newval && newlen) {
2994 		int __user *vec = (int __user *) newval;
2995 		int *min = (int *) table->extra1;
2996 		int *max = (int *) table->extra2;
2997 		size_t length;
2998 		int i;
2999 
3000 		if (newlen % sizeof(int) != 0)
3001 			return -EINVAL;
3002 
3003 		if (!table->extra1 && !table->extra2)
3004 			return 0;
3005 
3006 		if (newlen > table->maxlen)
3007 			newlen = table->maxlen;
3008 		length = newlen / sizeof(int);
3009 
3010 		for (i = 0; i < length; i++) {
3011 			int value;
3012 			if (get_user(value, vec + i))
3013 				return -EFAULT;
3014 			if (min && value < min[i])
3015 				return -EINVAL;
3016 			if (max && value > max[i])
3017 				return -EINVAL;
3018 		}
3019 	}
3020 	return 0;
3021 }
3022 
3023 /* Strategy function to convert jiffies to seconds */
3024 int sysctl_jiffies(struct ctl_table *table,
3025 		void __user *oldval, size_t __user *oldlenp,
3026 		void __user *newval, size_t newlen)
3027 {
3028 	if (oldval && oldlenp) {
3029 		size_t olen;
3030 
3031 		if (get_user(olen, oldlenp))
3032 			return -EFAULT;
3033 		if (olen) {
3034 			int val;
3035 
3036 			if (olen < sizeof(int))
3037 				return -EINVAL;
3038 
3039 			val = *(int *)(table->data) / HZ;
3040 			if (put_user(val, (int __user *)oldval))
3041 				return -EFAULT;
3042 			if (put_user(sizeof(int), oldlenp))
3043 				return -EFAULT;
3044 		}
3045 	}
3046 	if (newval && newlen) {
3047 		int new;
3048 		if (newlen != sizeof(int))
3049 			return -EINVAL;
3050 		if (get_user(new, (int __user *)newval))
3051 			return -EFAULT;
3052 		*(int *)(table->data) = new*HZ;
3053 	}
3054 	return 1;
3055 }
3056 
3057 /* Strategy function to convert jiffies to seconds */
3058 int sysctl_ms_jiffies(struct ctl_table *table,
3059 		void __user *oldval, size_t __user *oldlenp,
3060 		void __user *newval, size_t newlen)
3061 {
3062 	if (oldval && oldlenp) {
3063 		size_t olen;
3064 
3065 		if (get_user(olen, oldlenp))
3066 			return -EFAULT;
3067 		if (olen) {
3068 			int val;
3069 
3070 			if (olen < sizeof(int))
3071 				return -EINVAL;
3072 
3073 			val = jiffies_to_msecs(*(int *)(table->data));
3074 			if (put_user(val, (int __user *)oldval))
3075 				return -EFAULT;
3076 			if (put_user(sizeof(int), oldlenp))
3077 				return -EFAULT;
3078 		}
3079 	}
3080 	if (newval && newlen) {
3081 		int new;
3082 		if (newlen != sizeof(int))
3083 			return -EINVAL;
3084 		if (get_user(new, (int __user *)newval))
3085 			return -EFAULT;
3086 		*(int *)(table->data) = msecs_to_jiffies(new);
3087 	}
3088 	return 1;
3089 }
3090 
3091 
3092 
3093 #else /* CONFIG_SYSCTL_SYSCALL */
3094 
3095 
3096 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
3097 {
3098 	struct __sysctl_args tmp;
3099 	int error;
3100 
3101 	if (copy_from_user(&tmp, args, sizeof(tmp)))
3102 		return -EFAULT;
3103 
3104 	error = deprecated_sysctl_warning(&tmp);
3105 
3106 	/* If no error reading the parameters then just -ENOSYS ... */
3107 	if (!error)
3108 		error = -ENOSYS;
3109 
3110 	return error;
3111 }
3112 
3113 int sysctl_data(struct ctl_table *table,
3114 		  void __user *oldval, size_t __user *oldlenp,
3115 		  void __user *newval, size_t newlen)
3116 {
3117 	return -ENOSYS;
3118 }
3119 
3120 int sysctl_string(struct ctl_table *table,
3121 		  void __user *oldval, size_t __user *oldlenp,
3122 		  void __user *newval, size_t newlen)
3123 {
3124 	return -ENOSYS;
3125 }
3126 
3127 int sysctl_intvec(struct ctl_table *table,
3128 		void __user *oldval, size_t __user *oldlenp,
3129 		void __user *newval, size_t newlen)
3130 {
3131 	return -ENOSYS;
3132 }
3133 
3134 int sysctl_jiffies(struct ctl_table *table,
3135 		void __user *oldval, size_t __user *oldlenp,
3136 		void __user *newval, size_t newlen)
3137 {
3138 	return -ENOSYS;
3139 }
3140 
3141 int sysctl_ms_jiffies(struct ctl_table *table,
3142 		void __user *oldval, size_t __user *oldlenp,
3143 		void __user *newval, size_t newlen)
3144 {
3145 	return -ENOSYS;
3146 }
3147 
3148 #endif /* CONFIG_SYSCTL_SYSCALL */
3149 
3150 static int deprecated_sysctl_warning(struct __sysctl_args *args)
3151 {
3152 	static int msg_count;
3153 	int name[CTL_MAXNAME];
3154 	int i;
3155 
3156 	/* Check args->nlen. */
3157 	if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
3158 		return -ENOTDIR;
3159 
3160 	/* Read in the sysctl name for better debug message logging */
3161 	for (i = 0; i < args->nlen; i++)
3162 		if (get_user(name[i], args->name + i))
3163 			return -EFAULT;
3164 
3165 	/* Ignore accesses to kernel.version */
3166 	if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
3167 		return 0;
3168 
3169 	if (msg_count < 5) {
3170 		msg_count++;
3171 		printk(KERN_INFO
3172 			"warning: process `%s' used the deprecated sysctl "
3173 			"system call with ", current->comm);
3174 		for (i = 0; i < args->nlen; i++)
3175 			printk("%d.", name[i]);
3176 		printk("\n");
3177 	}
3178 	return 0;
3179 }
3180 
3181 /*
3182  * No sense putting this after each symbol definition, twice,
3183  * exception granted :-)
3184  */
3185 EXPORT_SYMBOL(proc_dointvec);
3186 EXPORT_SYMBOL(proc_dointvec_jiffies);
3187 EXPORT_SYMBOL(proc_dointvec_minmax);
3188 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3189 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3190 EXPORT_SYMBOL(proc_dostring);
3191 EXPORT_SYMBOL(proc_doulongvec_minmax);
3192 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3193 EXPORT_SYMBOL(register_sysctl_table);
3194 EXPORT_SYMBOL(register_sysctl_paths);
3195 EXPORT_SYMBOL(sysctl_intvec);
3196 EXPORT_SYMBOL(sysctl_jiffies);
3197 EXPORT_SYMBOL(sysctl_ms_jiffies);
3198 EXPORT_SYMBOL(sysctl_string);
3199 EXPORT_SYMBOL(sysctl_data);
3200 EXPORT_SYMBOL(unregister_sysctl_table);
3201