xref: /linux-6.15/kernel/sysctl.c (revision 5c725138)
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_event.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 #ifdef CONFIG_BLOCK
95 extern int blk_iopoll_enabled;
96 #endif
97 
98 /* Constants used for minimum and  maximum */
99 #ifdef CONFIG_DETECT_SOFTLOCKUP
100 static int sixty = 60;
101 static int neg_one = -1;
102 #endif
103 
104 static int zero;
105 static int __maybe_unused one = 1;
106 static int __maybe_unused two = 2;
107 static unsigned long one_ul = 1;
108 static int one_hundred = 100;
109 #ifdef CONFIG_PRINTK
110 static int ten_thousand = 10000;
111 #endif
112 
113 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
114 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
115 
116 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
117 static int maxolduid = 65535;
118 static int minolduid;
119 static int min_percpu_pagelist_fract = 8;
120 
121 static int ngroups_max = NGROUPS_MAX;
122 
123 #ifdef CONFIG_MODULES
124 extern char modprobe_path[];
125 extern int modules_disabled;
126 #endif
127 #ifdef CONFIG_CHR_DEV_SG
128 extern int sg_big_buff;
129 #endif
130 
131 #ifdef CONFIG_SPARC
132 #include <asm/system.h>
133 #endif
134 
135 #ifdef CONFIG_SPARC64
136 extern int sysctl_tsb_ratio;
137 #endif
138 
139 #ifdef __hppa__
140 extern int pwrsw_enabled;
141 extern int unaligned_enabled;
142 #endif
143 
144 #ifdef CONFIG_S390
145 #ifdef CONFIG_MATHEMU
146 extern int sysctl_ieee_emulation_warnings;
147 #endif
148 extern int sysctl_userprocess_debug;
149 extern int spin_retry;
150 #endif
151 
152 #ifdef CONFIG_BSD_PROCESS_ACCT
153 extern int acct_parm[];
154 #endif
155 
156 #ifdef CONFIG_IA64
157 extern int no_unaligned_warning;
158 extern int unaligned_dump_stack;
159 #endif
160 
161 #ifdef CONFIG_RT_MUTEXES
162 extern int max_lock_depth;
163 #endif
164 
165 #ifdef CONFIG_PROC_SYSCTL
166 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
167 		  void __user *buffer, size_t *lenp, loff_t *ppos);
168 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
169 			       void __user *buffer, size_t *lenp, loff_t *ppos);
170 #endif
171 
172 static struct ctl_table root_table[];
173 static struct ctl_table_root sysctl_table_root;
174 static struct ctl_table_header root_table_header = {
175 	.count = 1,
176 	.ctl_table = root_table,
177 	.ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
178 	.root = &sysctl_table_root,
179 	.set = &sysctl_table_root.default_set,
180 };
181 static struct ctl_table_root sysctl_table_root = {
182 	.root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
183 	.default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
184 };
185 
186 static struct ctl_table kern_table[];
187 static struct ctl_table vm_table[];
188 static struct ctl_table fs_table[];
189 static struct ctl_table debug_table[];
190 static struct ctl_table dev_table[];
191 extern struct ctl_table random_table[];
192 #ifdef CONFIG_INOTIFY_USER
193 extern struct ctl_table inotify_table[];
194 #endif
195 #ifdef CONFIG_EPOLL
196 extern struct ctl_table epoll_table[];
197 #endif
198 
199 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
200 int sysctl_legacy_va_layout;
201 #endif
202 
203 extern int prove_locking;
204 extern int lock_stat;
205 
206 /* The default sysctl tables: */
207 
208 static struct ctl_table root_table[] = {
209 	{
210 		.ctl_name	= CTL_KERN,
211 		.procname	= "kernel",
212 		.mode		= 0555,
213 		.child		= kern_table,
214 	},
215 	{
216 		.ctl_name	= CTL_VM,
217 		.procname	= "vm",
218 		.mode		= 0555,
219 		.child		= vm_table,
220 	},
221 	{
222 		.ctl_name	= CTL_FS,
223 		.procname	= "fs",
224 		.mode		= 0555,
225 		.child		= fs_table,
226 	},
227 	{
228 		.ctl_name	= CTL_DEBUG,
229 		.procname	= "debug",
230 		.mode		= 0555,
231 		.child		= debug_table,
232 	},
233 	{
234 		.ctl_name	= CTL_DEV,
235 		.procname	= "dev",
236 		.mode		= 0555,
237 		.child		= dev_table,
238 	},
239 /*
240  * NOTE: do not add new entries to this table unless you have read
241  * Documentation/sysctl/ctl_unnumbered.txt
242  */
243 	{ .ctl_name = 0 }
244 };
245 
246 #ifdef CONFIG_SCHED_DEBUG
247 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
248 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
249 static int min_wakeup_granularity_ns;			/* 0 usecs */
250 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
251 #endif
252 
253 static struct ctl_table kern_table[] = {
254 	{
255 		.ctl_name	= CTL_UNNUMBERED,
256 		.procname	= "sched_child_runs_first",
257 		.data		= &sysctl_sched_child_runs_first,
258 		.maxlen		= sizeof(unsigned int),
259 		.mode		= 0644,
260 		.proc_handler	= &proc_dointvec,
261 	},
262 #ifdef CONFIG_SCHED_DEBUG
263 	{
264 		.ctl_name	= CTL_UNNUMBERED,
265 		.procname	= "sched_min_granularity_ns",
266 		.data		= &sysctl_sched_min_granularity,
267 		.maxlen		= sizeof(unsigned int),
268 		.mode		= 0644,
269 		.proc_handler	= &sched_nr_latency_handler,
270 		.strategy	= &sysctl_intvec,
271 		.extra1		= &min_sched_granularity_ns,
272 		.extra2		= &max_sched_granularity_ns,
273 	},
274 	{
275 		.ctl_name	= CTL_UNNUMBERED,
276 		.procname	= "sched_latency_ns",
277 		.data		= &sysctl_sched_latency,
278 		.maxlen		= sizeof(unsigned int),
279 		.mode		= 0644,
280 		.proc_handler	= &sched_nr_latency_handler,
281 		.strategy	= &sysctl_intvec,
282 		.extra1		= &min_sched_granularity_ns,
283 		.extra2		= &max_sched_granularity_ns,
284 	},
285 	{
286 		.ctl_name	= CTL_UNNUMBERED,
287 		.procname	= "sched_wakeup_granularity_ns",
288 		.data		= &sysctl_sched_wakeup_granularity,
289 		.maxlen		= sizeof(unsigned int),
290 		.mode		= 0644,
291 		.proc_handler	= &proc_dointvec_minmax,
292 		.strategy	= &sysctl_intvec,
293 		.extra1		= &min_wakeup_granularity_ns,
294 		.extra2		= &max_wakeup_granularity_ns,
295 	},
296 	{
297 		.ctl_name	= CTL_UNNUMBERED,
298 		.procname	= "sched_shares_ratelimit",
299 		.data		= &sysctl_sched_shares_ratelimit,
300 		.maxlen		= sizeof(unsigned int),
301 		.mode		= 0644,
302 		.proc_handler	= &proc_dointvec,
303 	},
304 	{
305 		.ctl_name	= CTL_UNNUMBERED,
306 		.procname	= "sched_shares_thresh",
307 		.data		= &sysctl_sched_shares_thresh,
308 		.maxlen		= sizeof(unsigned int),
309 		.mode		= 0644,
310 		.proc_handler	= &proc_dointvec_minmax,
311 		.strategy	= &sysctl_intvec,
312 		.extra1		= &zero,
313 	},
314 	{
315 		.ctl_name	= CTL_UNNUMBERED,
316 		.procname	= "sched_features",
317 		.data		= &sysctl_sched_features,
318 		.maxlen		= sizeof(unsigned int),
319 		.mode		= 0644,
320 		.proc_handler	= &proc_dointvec,
321 	},
322 	{
323 		.ctl_name	= CTL_UNNUMBERED,
324 		.procname	= "sched_migration_cost",
325 		.data		= &sysctl_sched_migration_cost,
326 		.maxlen		= sizeof(unsigned int),
327 		.mode		= 0644,
328 		.proc_handler	= &proc_dointvec,
329 	},
330 	{
331 		.ctl_name	= CTL_UNNUMBERED,
332 		.procname	= "sched_nr_migrate",
333 		.data		= &sysctl_sched_nr_migrate,
334 		.maxlen		= sizeof(unsigned int),
335 		.mode		= 0644,
336 		.proc_handler	= &proc_dointvec,
337 	},
338 	{
339 		.ctl_name	= CTL_UNNUMBERED,
340 		.procname	= "sched_time_avg",
341 		.data		= &sysctl_sched_time_avg,
342 		.maxlen		= sizeof(unsigned int),
343 		.mode		= 0644,
344 		.proc_handler	= &proc_dointvec,
345 	},
346 	{
347 		.ctl_name	= CTL_UNNUMBERED,
348 		.procname	= "timer_migration",
349 		.data		= &sysctl_timer_migration,
350 		.maxlen		= sizeof(unsigned int),
351 		.mode		= 0644,
352 		.proc_handler	= &proc_dointvec_minmax,
353 		.strategy	= &sysctl_intvec,
354 		.extra1		= &zero,
355 		.extra2		= &one,
356 	},
357 #endif
358 	{
359 		.ctl_name	= CTL_UNNUMBERED,
360 		.procname	= "sched_rt_period_us",
361 		.data		= &sysctl_sched_rt_period,
362 		.maxlen		= sizeof(unsigned int),
363 		.mode		= 0644,
364 		.proc_handler	= &sched_rt_handler,
365 	},
366 	{
367 		.ctl_name	= CTL_UNNUMBERED,
368 		.procname	= "sched_rt_runtime_us",
369 		.data		= &sysctl_sched_rt_runtime,
370 		.maxlen		= sizeof(int),
371 		.mode		= 0644,
372 		.proc_handler	= &sched_rt_handler,
373 	},
374 	{
375 		.ctl_name	= CTL_UNNUMBERED,
376 		.procname	= "sched_compat_yield",
377 		.data		= &sysctl_sched_compat_yield,
378 		.maxlen		= sizeof(unsigned int),
379 		.mode		= 0644,
380 		.proc_handler	= &proc_dointvec,
381 	},
382 #ifdef CONFIG_PROVE_LOCKING
383 	{
384 		.ctl_name	= CTL_UNNUMBERED,
385 		.procname	= "prove_locking",
386 		.data		= &prove_locking,
387 		.maxlen		= sizeof(int),
388 		.mode		= 0644,
389 		.proc_handler	= &proc_dointvec,
390 	},
391 #endif
392 #ifdef CONFIG_LOCK_STAT
393 	{
394 		.ctl_name	= CTL_UNNUMBERED,
395 		.procname	= "lock_stat",
396 		.data		= &lock_stat,
397 		.maxlen		= sizeof(int),
398 		.mode		= 0644,
399 		.proc_handler	= &proc_dointvec,
400 	},
401 #endif
402 	{
403 		.ctl_name	= KERN_PANIC,
404 		.procname	= "panic",
405 		.data		= &panic_timeout,
406 		.maxlen		= sizeof(int),
407 		.mode		= 0644,
408 		.proc_handler	= &proc_dointvec,
409 	},
410 	{
411 		.ctl_name	= KERN_CORE_USES_PID,
412 		.procname	= "core_uses_pid",
413 		.data		= &core_uses_pid,
414 		.maxlen		= sizeof(int),
415 		.mode		= 0644,
416 		.proc_handler	= &proc_dointvec,
417 	},
418 	{
419 		.ctl_name	= KERN_CORE_PATTERN,
420 		.procname	= "core_pattern",
421 		.data		= core_pattern,
422 		.maxlen		= CORENAME_MAX_SIZE,
423 		.mode		= 0644,
424 		.proc_handler	= &proc_dostring,
425 		.strategy	= &sysctl_string,
426 	},
427 #ifdef CONFIG_PROC_SYSCTL
428 	{
429 		.procname	= "tainted",
430 		.maxlen 	= sizeof(long),
431 		.mode		= 0644,
432 		.proc_handler	= &proc_taint,
433 	},
434 #endif
435 #ifdef CONFIG_LATENCYTOP
436 	{
437 		.procname	= "latencytop",
438 		.data		= &latencytop_enabled,
439 		.maxlen		= sizeof(int),
440 		.mode		= 0644,
441 		.proc_handler	= &proc_dointvec,
442 	},
443 #endif
444 #ifdef CONFIG_BLK_DEV_INITRD
445 	{
446 		.ctl_name	= KERN_REALROOTDEV,
447 		.procname	= "real-root-dev",
448 		.data		= &real_root_dev,
449 		.maxlen		= sizeof(int),
450 		.mode		= 0644,
451 		.proc_handler	= &proc_dointvec,
452 	},
453 #endif
454 	{
455 		.ctl_name	= CTL_UNNUMBERED,
456 		.procname	= "print-fatal-signals",
457 		.data		= &print_fatal_signals,
458 		.maxlen		= sizeof(int),
459 		.mode		= 0644,
460 		.proc_handler	= &proc_dointvec,
461 	},
462 #ifdef CONFIG_SPARC
463 	{
464 		.ctl_name	= KERN_SPARC_REBOOT,
465 		.procname	= "reboot-cmd",
466 		.data		= reboot_command,
467 		.maxlen		= 256,
468 		.mode		= 0644,
469 		.proc_handler	= &proc_dostring,
470 		.strategy	= &sysctl_string,
471 	},
472 	{
473 		.ctl_name	= KERN_SPARC_STOP_A,
474 		.procname	= "stop-a",
475 		.data		= &stop_a_enabled,
476 		.maxlen		= sizeof (int),
477 		.mode		= 0644,
478 		.proc_handler	= &proc_dointvec,
479 	},
480 	{
481 		.ctl_name	= KERN_SPARC_SCONS_PWROFF,
482 		.procname	= "scons-poweroff",
483 		.data		= &scons_pwroff,
484 		.maxlen		= sizeof (int),
485 		.mode		= 0644,
486 		.proc_handler	= &proc_dointvec,
487 	},
488 #endif
489 #ifdef CONFIG_SPARC64
490 	{
491 		.ctl_name	= CTL_UNNUMBERED,
492 		.procname	= "tsb-ratio",
493 		.data		= &sysctl_tsb_ratio,
494 		.maxlen		= sizeof (int),
495 		.mode		= 0644,
496 		.proc_handler	= &proc_dointvec,
497 	},
498 #endif
499 #ifdef __hppa__
500 	{
501 		.ctl_name	= KERN_HPPA_PWRSW,
502 		.procname	= "soft-power",
503 		.data		= &pwrsw_enabled,
504 		.maxlen		= sizeof (int),
505 	 	.mode		= 0644,
506 		.proc_handler	= &proc_dointvec,
507 	},
508 	{
509 		.ctl_name	= KERN_HPPA_UNALIGNED,
510 		.procname	= "unaligned-trap",
511 		.data		= &unaligned_enabled,
512 		.maxlen		= sizeof (int),
513 		.mode		= 0644,
514 		.proc_handler	= &proc_dointvec,
515 	},
516 #endif
517 	{
518 		.ctl_name	= KERN_CTLALTDEL,
519 		.procname	= "ctrl-alt-del",
520 		.data		= &C_A_D,
521 		.maxlen		= sizeof(int),
522 		.mode		= 0644,
523 		.proc_handler	= &proc_dointvec,
524 	},
525 #ifdef CONFIG_FUNCTION_TRACER
526 	{
527 		.ctl_name	= CTL_UNNUMBERED,
528 		.procname	= "ftrace_enabled",
529 		.data		= &ftrace_enabled,
530 		.maxlen		= sizeof(int),
531 		.mode		= 0644,
532 		.proc_handler	= &ftrace_enable_sysctl,
533 	},
534 #endif
535 #ifdef CONFIG_STACK_TRACER
536 	{
537 		.ctl_name	= CTL_UNNUMBERED,
538 		.procname	= "stack_tracer_enabled",
539 		.data		= &stack_tracer_enabled,
540 		.maxlen		= sizeof(int),
541 		.mode		= 0644,
542 		.proc_handler	= &stack_trace_sysctl,
543 	},
544 #endif
545 #ifdef CONFIG_TRACING
546 	{
547 		.ctl_name	= CTL_UNNUMBERED,
548 		.procname	= "ftrace_dump_on_oops",
549 		.data		= &ftrace_dump_on_oops,
550 		.maxlen		= sizeof(int),
551 		.mode		= 0644,
552 		.proc_handler	= &proc_dointvec,
553 	},
554 #endif
555 #ifdef CONFIG_MODULES
556 	{
557 		.ctl_name	= KERN_MODPROBE,
558 		.procname	= "modprobe",
559 		.data		= &modprobe_path,
560 		.maxlen		= KMOD_PATH_LEN,
561 		.mode		= 0644,
562 		.proc_handler	= &proc_dostring,
563 		.strategy	= &sysctl_string,
564 	},
565 	{
566 		.ctl_name	= CTL_UNNUMBERED,
567 		.procname	= "modules_disabled",
568 		.data		= &modules_disabled,
569 		.maxlen		= sizeof(int),
570 		.mode		= 0644,
571 		/* only handle a transition from default "0" to "1" */
572 		.proc_handler	= &proc_dointvec_minmax,
573 		.extra1		= &one,
574 		.extra2		= &one,
575 	},
576 #endif
577 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
578 	{
579 		.ctl_name	= KERN_HOTPLUG,
580 		.procname	= "hotplug",
581 		.data		= &uevent_helper,
582 		.maxlen		= UEVENT_HELPER_PATH_LEN,
583 		.mode		= 0644,
584 		.proc_handler	= &proc_dostring,
585 		.strategy	= &sysctl_string,
586 	},
587 #endif
588 #ifdef CONFIG_CHR_DEV_SG
589 	{
590 		.ctl_name	= KERN_SG_BIG_BUFF,
591 		.procname	= "sg-big-buff",
592 		.data		= &sg_big_buff,
593 		.maxlen		= sizeof (int),
594 		.mode		= 0444,
595 		.proc_handler	= &proc_dointvec,
596 	},
597 #endif
598 #ifdef CONFIG_BSD_PROCESS_ACCT
599 	{
600 		.ctl_name	= KERN_ACCT,
601 		.procname	= "acct",
602 		.data		= &acct_parm,
603 		.maxlen		= 3*sizeof(int),
604 		.mode		= 0644,
605 		.proc_handler	= &proc_dointvec,
606 	},
607 #endif
608 #ifdef CONFIG_MAGIC_SYSRQ
609 	{
610 		.ctl_name	= KERN_SYSRQ,
611 		.procname	= "sysrq",
612 		.data		= &__sysrq_enabled,
613 		.maxlen		= sizeof (int),
614 		.mode		= 0644,
615 		.proc_handler	= &proc_dointvec,
616 	},
617 #endif
618 #ifdef CONFIG_PROC_SYSCTL
619 	{
620 		.procname	= "cad_pid",
621 		.data		= NULL,
622 		.maxlen		= sizeof (int),
623 		.mode		= 0600,
624 		.proc_handler	= &proc_do_cad_pid,
625 	},
626 #endif
627 	{
628 		.ctl_name	= KERN_MAX_THREADS,
629 		.procname	= "threads-max",
630 		.data		= &max_threads,
631 		.maxlen		= sizeof(int),
632 		.mode		= 0644,
633 		.proc_handler	= &proc_dointvec,
634 	},
635 	{
636 		.ctl_name	= KERN_RANDOM,
637 		.procname	= "random",
638 		.mode		= 0555,
639 		.child		= random_table,
640 	},
641 	{
642 		.ctl_name	= KERN_OVERFLOWUID,
643 		.procname	= "overflowuid",
644 		.data		= &overflowuid,
645 		.maxlen		= sizeof(int),
646 		.mode		= 0644,
647 		.proc_handler	= &proc_dointvec_minmax,
648 		.strategy	= &sysctl_intvec,
649 		.extra1		= &minolduid,
650 		.extra2		= &maxolduid,
651 	},
652 	{
653 		.ctl_name	= KERN_OVERFLOWGID,
654 		.procname	= "overflowgid",
655 		.data		= &overflowgid,
656 		.maxlen		= sizeof(int),
657 		.mode		= 0644,
658 		.proc_handler	= &proc_dointvec_minmax,
659 		.strategy	= &sysctl_intvec,
660 		.extra1		= &minolduid,
661 		.extra2		= &maxolduid,
662 	},
663 #ifdef CONFIG_S390
664 #ifdef CONFIG_MATHEMU
665 	{
666 		.ctl_name	= KERN_IEEE_EMULATION_WARNINGS,
667 		.procname	= "ieee_emulation_warnings",
668 		.data		= &sysctl_ieee_emulation_warnings,
669 		.maxlen		= sizeof(int),
670 		.mode		= 0644,
671 		.proc_handler	= &proc_dointvec,
672 	},
673 #endif
674 	{
675 		.ctl_name	= KERN_S390_USER_DEBUG_LOGGING,
676 		.procname	= "userprocess_debug",
677 		.data		= &sysctl_userprocess_debug,
678 		.maxlen		= sizeof(int),
679 		.mode		= 0644,
680 		.proc_handler	= &proc_dointvec,
681 	},
682 #endif
683 	{
684 		.ctl_name	= KERN_PIDMAX,
685 		.procname	= "pid_max",
686 		.data		= &pid_max,
687 		.maxlen		= sizeof (int),
688 		.mode		= 0644,
689 		.proc_handler	= &proc_dointvec_minmax,
690 		.strategy	= sysctl_intvec,
691 		.extra1		= &pid_max_min,
692 		.extra2		= &pid_max_max,
693 	},
694 	{
695 		.ctl_name	= KERN_PANIC_ON_OOPS,
696 		.procname	= "panic_on_oops",
697 		.data		= &panic_on_oops,
698 		.maxlen		= sizeof(int),
699 		.mode		= 0644,
700 		.proc_handler	= &proc_dointvec,
701 	},
702 #if defined CONFIG_PRINTK
703 	{
704 		.ctl_name	= KERN_PRINTK,
705 		.procname	= "printk",
706 		.data		= &console_loglevel,
707 		.maxlen		= 4*sizeof(int),
708 		.mode		= 0644,
709 		.proc_handler	= &proc_dointvec,
710 	},
711 	{
712 		.ctl_name	= KERN_PRINTK_RATELIMIT,
713 		.procname	= "printk_ratelimit",
714 		.data		= &printk_ratelimit_state.interval,
715 		.maxlen		= sizeof(int),
716 		.mode		= 0644,
717 		.proc_handler	= &proc_dointvec_jiffies,
718 		.strategy	= &sysctl_jiffies,
719 	},
720 	{
721 		.ctl_name	= KERN_PRINTK_RATELIMIT_BURST,
722 		.procname	= "printk_ratelimit_burst",
723 		.data		= &printk_ratelimit_state.burst,
724 		.maxlen		= sizeof(int),
725 		.mode		= 0644,
726 		.proc_handler	= &proc_dointvec,
727 	},
728 	{
729 		.ctl_name	= CTL_UNNUMBERED,
730 		.procname	= "printk_delay",
731 		.data		= &printk_delay_msec,
732 		.maxlen		= sizeof(int),
733 		.mode		= 0644,
734 		.proc_handler	= &proc_dointvec_minmax,
735 		.strategy	= &sysctl_intvec,
736 		.extra1		= &zero,
737 		.extra2		= &ten_thousand,
738 	},
739 #endif
740 	{
741 		.ctl_name	= KERN_NGROUPS_MAX,
742 		.procname	= "ngroups_max",
743 		.data		= &ngroups_max,
744 		.maxlen		= sizeof (int),
745 		.mode		= 0444,
746 		.proc_handler	= &proc_dointvec,
747 	},
748 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
749 	{
750 		.ctl_name       = KERN_UNKNOWN_NMI_PANIC,
751 		.procname       = "unknown_nmi_panic",
752 		.data           = &unknown_nmi_panic,
753 		.maxlen         = sizeof (int),
754 		.mode           = 0644,
755 		.proc_handler   = &proc_dointvec,
756 	},
757 	{
758 		.procname       = "nmi_watchdog",
759 		.data           = &nmi_watchdog_enabled,
760 		.maxlen         = sizeof (int),
761 		.mode           = 0644,
762 		.proc_handler   = &proc_nmi_enabled,
763 	},
764 #endif
765 #if defined(CONFIG_X86)
766 	{
767 		.ctl_name	= KERN_PANIC_ON_NMI,
768 		.procname	= "panic_on_unrecovered_nmi",
769 		.data		= &panic_on_unrecovered_nmi,
770 		.maxlen		= sizeof(int),
771 		.mode		= 0644,
772 		.proc_handler	= &proc_dointvec,
773 	},
774 	{
775 		.ctl_name	= CTL_UNNUMBERED,
776 		.procname	= "panic_on_io_nmi",
777 		.data		= &panic_on_io_nmi,
778 		.maxlen		= sizeof(int),
779 		.mode		= 0644,
780 		.proc_handler	= &proc_dointvec,
781 	},
782 	{
783 		.ctl_name	= KERN_BOOTLOADER_TYPE,
784 		.procname	= "bootloader_type",
785 		.data		= &bootloader_type,
786 		.maxlen		= sizeof (int),
787 		.mode		= 0444,
788 		.proc_handler	= &proc_dointvec,
789 	},
790 	{
791 		.ctl_name	= CTL_UNNUMBERED,
792 		.procname	= "bootloader_version",
793 		.data		= &bootloader_version,
794 		.maxlen		= sizeof (int),
795 		.mode		= 0444,
796 		.proc_handler	= &proc_dointvec,
797 	},
798 	{
799 		.ctl_name	= CTL_UNNUMBERED,
800 		.procname	= "kstack_depth_to_print",
801 		.data		= &kstack_depth_to_print,
802 		.maxlen		= sizeof(int),
803 		.mode		= 0644,
804 		.proc_handler	= &proc_dointvec,
805 	},
806 	{
807 		.ctl_name	= CTL_UNNUMBERED,
808 		.procname	= "io_delay_type",
809 		.data		= &io_delay_type,
810 		.maxlen		= sizeof(int),
811 		.mode		= 0644,
812 		.proc_handler	= &proc_dointvec,
813 	},
814 #endif
815 #if defined(CONFIG_MMU)
816 	{
817 		.ctl_name	= KERN_RANDOMIZE,
818 		.procname	= "randomize_va_space",
819 		.data		= &randomize_va_space,
820 		.maxlen		= sizeof(int),
821 		.mode		= 0644,
822 		.proc_handler	= &proc_dointvec,
823 	},
824 #endif
825 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
826 	{
827 		.ctl_name	= KERN_SPIN_RETRY,
828 		.procname	= "spin_retry",
829 		.data		= &spin_retry,
830 		.maxlen		= sizeof (int),
831 		.mode		= 0644,
832 		.proc_handler	= &proc_dointvec,
833 	},
834 #endif
835 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
836 	{
837 		.procname	= "acpi_video_flags",
838 		.data		= &acpi_realmode_flags,
839 		.maxlen		= sizeof (unsigned long),
840 		.mode		= 0644,
841 		.proc_handler	= &proc_doulongvec_minmax,
842 	},
843 #endif
844 #ifdef CONFIG_IA64
845 	{
846 		.ctl_name	= KERN_IA64_UNALIGNED,
847 		.procname	= "ignore-unaligned-usertrap",
848 		.data		= &no_unaligned_warning,
849 		.maxlen		= sizeof (int),
850 	 	.mode		= 0644,
851 		.proc_handler	= &proc_dointvec,
852 	},
853 	{
854 		.ctl_name	= CTL_UNNUMBERED,
855 		.procname	= "unaligned-dump-stack",
856 		.data		= &unaligned_dump_stack,
857 		.maxlen		= sizeof (int),
858 		.mode		= 0644,
859 		.proc_handler	= &proc_dointvec,
860 	},
861 #endif
862 #ifdef CONFIG_DETECT_SOFTLOCKUP
863 	{
864 		.ctl_name	= CTL_UNNUMBERED,
865 		.procname	= "softlockup_panic",
866 		.data		= &softlockup_panic,
867 		.maxlen		= sizeof(int),
868 		.mode		= 0644,
869 		.proc_handler	= &proc_dointvec_minmax,
870 		.strategy	= &sysctl_intvec,
871 		.extra1		= &zero,
872 		.extra2		= &one,
873 	},
874 	{
875 		.ctl_name	= CTL_UNNUMBERED,
876 		.procname	= "softlockup_thresh",
877 		.data		= &softlockup_thresh,
878 		.maxlen		= sizeof(int),
879 		.mode		= 0644,
880 		.proc_handler	= &proc_dosoftlockup_thresh,
881 		.strategy	= &sysctl_intvec,
882 		.extra1		= &neg_one,
883 		.extra2		= &sixty,
884 	},
885 #endif
886 #ifdef CONFIG_DETECT_HUNG_TASK
887 	{
888 		.ctl_name	= CTL_UNNUMBERED,
889 		.procname	= "hung_task_panic",
890 		.data		= &sysctl_hung_task_panic,
891 		.maxlen		= sizeof(int),
892 		.mode		= 0644,
893 		.proc_handler	= &proc_dointvec_minmax,
894 		.strategy	= &sysctl_intvec,
895 		.extra1		= &zero,
896 		.extra2		= &one,
897 	},
898 	{
899 		.ctl_name	= CTL_UNNUMBERED,
900 		.procname	= "hung_task_check_count",
901 		.data		= &sysctl_hung_task_check_count,
902 		.maxlen		= sizeof(unsigned long),
903 		.mode		= 0644,
904 		.proc_handler	= &proc_doulongvec_minmax,
905 		.strategy	= &sysctl_intvec,
906 	},
907 	{
908 		.ctl_name	= CTL_UNNUMBERED,
909 		.procname	= "hung_task_timeout_secs",
910 		.data		= &sysctl_hung_task_timeout_secs,
911 		.maxlen		= sizeof(unsigned long),
912 		.mode		= 0644,
913 		.proc_handler	= &proc_dohung_task_timeout_secs,
914 		.strategy	= &sysctl_intvec,
915 	},
916 	{
917 		.ctl_name	= CTL_UNNUMBERED,
918 		.procname	= "hung_task_warnings",
919 		.data		= &sysctl_hung_task_warnings,
920 		.maxlen		= sizeof(unsigned long),
921 		.mode		= 0644,
922 		.proc_handler	= &proc_doulongvec_minmax,
923 		.strategy	= &sysctl_intvec,
924 	},
925 #endif
926 #ifdef CONFIG_COMPAT
927 	{
928 		.ctl_name	= KERN_COMPAT_LOG,
929 		.procname	= "compat-log",
930 		.data		= &compat_log,
931 		.maxlen		= sizeof (int),
932 	 	.mode		= 0644,
933 		.proc_handler	= &proc_dointvec,
934 	},
935 #endif
936 #ifdef CONFIG_RT_MUTEXES
937 	{
938 		.ctl_name	= KERN_MAX_LOCK_DEPTH,
939 		.procname	= "max_lock_depth",
940 		.data		= &max_lock_depth,
941 		.maxlen		= sizeof(int),
942 		.mode		= 0644,
943 		.proc_handler	= &proc_dointvec,
944 	},
945 #endif
946 	{
947 		.ctl_name	= CTL_UNNUMBERED,
948 		.procname	= "poweroff_cmd",
949 		.data		= &poweroff_cmd,
950 		.maxlen		= POWEROFF_CMD_PATH_LEN,
951 		.mode		= 0644,
952 		.proc_handler	= &proc_dostring,
953 		.strategy	= &sysctl_string,
954 	},
955 #ifdef CONFIG_KEYS
956 	{
957 		.ctl_name	= CTL_UNNUMBERED,
958 		.procname	= "keys",
959 		.mode		= 0555,
960 		.child		= key_sysctls,
961 	},
962 #endif
963 #ifdef CONFIG_RCU_TORTURE_TEST
964 	{
965 		.ctl_name       = CTL_UNNUMBERED,
966 		.procname       = "rcutorture_runnable",
967 		.data           = &rcutorture_runnable,
968 		.maxlen         = sizeof(int),
969 		.mode           = 0644,
970 		.proc_handler   = &proc_dointvec,
971 	},
972 #endif
973 #ifdef CONFIG_SLOW_WORK
974 	{
975 		.ctl_name	= CTL_UNNUMBERED,
976 		.procname	= "slow-work",
977 		.mode		= 0555,
978 		.child		= slow_work_sysctls,
979 	},
980 #endif
981 #ifdef CONFIG_PERF_EVENTS
982 	{
983 		.ctl_name	= CTL_UNNUMBERED,
984 		.procname	= "perf_event_paranoid",
985 		.data		= &sysctl_perf_event_paranoid,
986 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
987 		.mode		= 0644,
988 		.proc_handler	= &proc_dointvec,
989 	},
990 	{
991 		.ctl_name	= CTL_UNNUMBERED,
992 		.procname	= "perf_event_mlock_kb",
993 		.data		= &sysctl_perf_event_mlock,
994 		.maxlen		= sizeof(sysctl_perf_event_mlock),
995 		.mode		= 0644,
996 		.proc_handler	= &proc_dointvec,
997 	},
998 	{
999 		.ctl_name	= CTL_UNNUMBERED,
1000 		.procname	= "perf_event_max_sample_rate",
1001 		.data		= &sysctl_perf_event_sample_rate,
1002 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
1003 		.mode		= 0644,
1004 		.proc_handler	= &proc_dointvec,
1005 	},
1006 #endif
1007 #ifdef CONFIG_KMEMCHECK
1008 	{
1009 		.ctl_name	= CTL_UNNUMBERED,
1010 		.procname	= "kmemcheck",
1011 		.data		= &kmemcheck_enabled,
1012 		.maxlen		= sizeof(int),
1013 		.mode		= 0644,
1014 		.proc_handler	= &proc_dointvec,
1015 	},
1016 #endif
1017 #ifdef CONFIG_BLOCK
1018 	{
1019 		.ctl_name	= CTL_UNNUMBERED,
1020 		.procname	= "blk_iopoll",
1021 		.data		= &blk_iopoll_enabled,
1022 		.maxlen		= sizeof(int),
1023 		.mode		= 0644,
1024 		.proc_handler	= &proc_dointvec,
1025 	},
1026 #endif
1027 /*
1028  * NOTE: do not add new entries to this table unless you have read
1029  * Documentation/sysctl/ctl_unnumbered.txt
1030  */
1031 	{ .ctl_name = 0 }
1032 };
1033 
1034 static struct ctl_table vm_table[] = {
1035 	{
1036 		.ctl_name	= VM_OVERCOMMIT_MEMORY,
1037 		.procname	= "overcommit_memory",
1038 		.data		= &sysctl_overcommit_memory,
1039 		.maxlen		= sizeof(sysctl_overcommit_memory),
1040 		.mode		= 0644,
1041 		.proc_handler	= &proc_dointvec,
1042 	},
1043 	{
1044 		.ctl_name	= VM_PANIC_ON_OOM,
1045 		.procname	= "panic_on_oom",
1046 		.data		= &sysctl_panic_on_oom,
1047 		.maxlen		= sizeof(sysctl_panic_on_oom),
1048 		.mode		= 0644,
1049 		.proc_handler	= &proc_dointvec,
1050 	},
1051 	{
1052 		.ctl_name	= CTL_UNNUMBERED,
1053 		.procname	= "oom_kill_allocating_task",
1054 		.data		= &sysctl_oom_kill_allocating_task,
1055 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1056 		.mode		= 0644,
1057 		.proc_handler	= &proc_dointvec,
1058 	},
1059 	{
1060 		.ctl_name	= CTL_UNNUMBERED,
1061 		.procname	= "oom_dump_tasks",
1062 		.data		= &sysctl_oom_dump_tasks,
1063 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1064 		.mode		= 0644,
1065 		.proc_handler	= &proc_dointvec,
1066 	},
1067 	{
1068 		.ctl_name	= VM_OVERCOMMIT_RATIO,
1069 		.procname	= "overcommit_ratio",
1070 		.data		= &sysctl_overcommit_ratio,
1071 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1072 		.mode		= 0644,
1073 		.proc_handler	= &proc_dointvec,
1074 	},
1075 	{
1076 		.ctl_name	= VM_PAGE_CLUSTER,
1077 		.procname	= "page-cluster",
1078 		.data		= &page_cluster,
1079 		.maxlen		= sizeof(int),
1080 		.mode		= 0644,
1081 		.proc_handler	= &proc_dointvec,
1082 	},
1083 	{
1084 		.ctl_name	= VM_DIRTY_BACKGROUND,
1085 		.procname	= "dirty_background_ratio",
1086 		.data		= &dirty_background_ratio,
1087 		.maxlen		= sizeof(dirty_background_ratio),
1088 		.mode		= 0644,
1089 		.proc_handler	= &dirty_background_ratio_handler,
1090 		.strategy	= &sysctl_intvec,
1091 		.extra1		= &zero,
1092 		.extra2		= &one_hundred,
1093 	},
1094 	{
1095 		.ctl_name	= CTL_UNNUMBERED,
1096 		.procname	= "dirty_background_bytes",
1097 		.data		= &dirty_background_bytes,
1098 		.maxlen		= sizeof(dirty_background_bytes),
1099 		.mode		= 0644,
1100 		.proc_handler	= &dirty_background_bytes_handler,
1101 		.strategy	= &sysctl_intvec,
1102 		.extra1		= &one_ul,
1103 	},
1104 	{
1105 		.ctl_name	= VM_DIRTY_RATIO,
1106 		.procname	= "dirty_ratio",
1107 		.data		= &vm_dirty_ratio,
1108 		.maxlen		= sizeof(vm_dirty_ratio),
1109 		.mode		= 0644,
1110 		.proc_handler	= &dirty_ratio_handler,
1111 		.strategy	= &sysctl_intvec,
1112 		.extra1		= &zero,
1113 		.extra2		= &one_hundred,
1114 	},
1115 	{
1116 		.ctl_name	= CTL_UNNUMBERED,
1117 		.procname	= "dirty_bytes",
1118 		.data		= &vm_dirty_bytes,
1119 		.maxlen		= sizeof(vm_dirty_bytes),
1120 		.mode		= 0644,
1121 		.proc_handler	= &dirty_bytes_handler,
1122 		.strategy	= &sysctl_intvec,
1123 		.extra1		= &dirty_bytes_min,
1124 	},
1125 	{
1126 		.procname	= "dirty_writeback_centisecs",
1127 		.data		= &dirty_writeback_interval,
1128 		.maxlen		= sizeof(dirty_writeback_interval),
1129 		.mode		= 0644,
1130 		.proc_handler	= &dirty_writeback_centisecs_handler,
1131 	},
1132 	{
1133 		.procname	= "dirty_expire_centisecs",
1134 		.data		= &dirty_expire_interval,
1135 		.maxlen		= sizeof(dirty_expire_interval),
1136 		.mode		= 0644,
1137 		.proc_handler	= &proc_dointvec,
1138 	},
1139 	{
1140 		.ctl_name	= VM_NR_PDFLUSH_THREADS,
1141 		.procname	= "nr_pdflush_threads",
1142 		.data		= &nr_pdflush_threads,
1143 		.maxlen		= sizeof nr_pdflush_threads,
1144 		.mode		= 0444 /* read-only*/,
1145 		.proc_handler	= &proc_dointvec,
1146 	},
1147 	{
1148 		.ctl_name	= VM_SWAPPINESS,
1149 		.procname	= "swappiness",
1150 		.data		= &vm_swappiness,
1151 		.maxlen		= sizeof(vm_swappiness),
1152 		.mode		= 0644,
1153 		.proc_handler	= &proc_dointvec_minmax,
1154 		.strategy	= &sysctl_intvec,
1155 		.extra1		= &zero,
1156 		.extra2		= &one_hundred,
1157 	},
1158 #ifdef CONFIG_HUGETLB_PAGE
1159 	 {
1160 		.procname	= "nr_hugepages",
1161 		.data		= NULL,
1162 		.maxlen		= sizeof(unsigned long),
1163 		.mode		= 0644,
1164 		.proc_handler	= &hugetlb_sysctl_handler,
1165 		.extra1		= (void *)&hugetlb_zero,
1166 		.extra2		= (void *)&hugetlb_infinity,
1167 	 },
1168 	 {
1169 		.ctl_name	= VM_HUGETLB_GROUP,
1170 		.procname	= "hugetlb_shm_group",
1171 		.data		= &sysctl_hugetlb_shm_group,
1172 		.maxlen		= sizeof(gid_t),
1173 		.mode		= 0644,
1174 		.proc_handler	= &proc_dointvec,
1175 	 },
1176 	 {
1177 		.ctl_name	= CTL_UNNUMBERED,
1178 		.procname	= "hugepages_treat_as_movable",
1179 		.data		= &hugepages_treat_as_movable,
1180 		.maxlen		= sizeof(int),
1181 		.mode		= 0644,
1182 		.proc_handler	= &hugetlb_treat_movable_handler,
1183 	},
1184 	{
1185 		.ctl_name	= CTL_UNNUMBERED,
1186 		.procname	= "nr_overcommit_hugepages",
1187 		.data		= NULL,
1188 		.maxlen		= sizeof(unsigned long),
1189 		.mode		= 0644,
1190 		.proc_handler	= &hugetlb_overcommit_handler,
1191 		.extra1		= (void *)&hugetlb_zero,
1192 		.extra2		= (void *)&hugetlb_infinity,
1193 	},
1194 #endif
1195 	{
1196 		.ctl_name	= VM_LOWMEM_RESERVE_RATIO,
1197 		.procname	= "lowmem_reserve_ratio",
1198 		.data		= &sysctl_lowmem_reserve_ratio,
1199 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1200 		.mode		= 0644,
1201 		.proc_handler	= &lowmem_reserve_ratio_sysctl_handler,
1202 		.strategy	= &sysctl_intvec,
1203 	},
1204 	{
1205 		.ctl_name	= VM_DROP_PAGECACHE,
1206 		.procname	= "drop_caches",
1207 		.data		= &sysctl_drop_caches,
1208 		.maxlen		= sizeof(int),
1209 		.mode		= 0644,
1210 		.proc_handler	= drop_caches_sysctl_handler,
1211 		.strategy	= &sysctl_intvec,
1212 	},
1213 	{
1214 		.ctl_name	= VM_MIN_FREE_KBYTES,
1215 		.procname	= "min_free_kbytes",
1216 		.data		= &min_free_kbytes,
1217 		.maxlen		= sizeof(min_free_kbytes),
1218 		.mode		= 0644,
1219 		.proc_handler	= &min_free_kbytes_sysctl_handler,
1220 		.strategy	= &sysctl_intvec,
1221 		.extra1		= &zero,
1222 	},
1223 	{
1224 		.ctl_name	= VM_PERCPU_PAGELIST_FRACTION,
1225 		.procname	= "percpu_pagelist_fraction",
1226 		.data		= &percpu_pagelist_fraction,
1227 		.maxlen		= sizeof(percpu_pagelist_fraction),
1228 		.mode		= 0644,
1229 		.proc_handler	= &percpu_pagelist_fraction_sysctl_handler,
1230 		.strategy	= &sysctl_intvec,
1231 		.extra1		= &min_percpu_pagelist_fract,
1232 	},
1233 #ifdef CONFIG_MMU
1234 	{
1235 		.ctl_name	= VM_MAX_MAP_COUNT,
1236 		.procname	= "max_map_count",
1237 		.data		= &sysctl_max_map_count,
1238 		.maxlen		= sizeof(sysctl_max_map_count),
1239 		.mode		= 0644,
1240 		.proc_handler	= &proc_dointvec
1241 	},
1242 #else
1243 	{
1244 		.ctl_name	= CTL_UNNUMBERED,
1245 		.procname	= "nr_trim_pages",
1246 		.data		= &sysctl_nr_trim_pages,
1247 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1248 		.mode		= 0644,
1249 		.proc_handler	= &proc_dointvec_minmax,
1250 		.strategy	= &sysctl_intvec,
1251 		.extra1		= &zero,
1252 	},
1253 #endif
1254 	{
1255 		.ctl_name	= VM_LAPTOP_MODE,
1256 		.procname	= "laptop_mode",
1257 		.data		= &laptop_mode,
1258 		.maxlen		= sizeof(laptop_mode),
1259 		.mode		= 0644,
1260 		.proc_handler	= &proc_dointvec_jiffies,
1261 		.strategy	= &sysctl_jiffies,
1262 	},
1263 	{
1264 		.ctl_name	= VM_BLOCK_DUMP,
1265 		.procname	= "block_dump",
1266 		.data		= &block_dump,
1267 		.maxlen		= sizeof(block_dump),
1268 		.mode		= 0644,
1269 		.proc_handler	= &proc_dointvec,
1270 		.strategy	= &sysctl_intvec,
1271 		.extra1		= &zero,
1272 	},
1273 	{
1274 		.ctl_name	= VM_VFS_CACHE_PRESSURE,
1275 		.procname	= "vfs_cache_pressure",
1276 		.data		= &sysctl_vfs_cache_pressure,
1277 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1278 		.mode		= 0644,
1279 		.proc_handler	= &proc_dointvec,
1280 		.strategy	= &sysctl_intvec,
1281 		.extra1		= &zero,
1282 	},
1283 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1284 	{
1285 		.ctl_name	= VM_LEGACY_VA_LAYOUT,
1286 		.procname	= "legacy_va_layout",
1287 		.data		= &sysctl_legacy_va_layout,
1288 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1289 		.mode		= 0644,
1290 		.proc_handler	= &proc_dointvec,
1291 		.strategy	= &sysctl_intvec,
1292 		.extra1		= &zero,
1293 	},
1294 #endif
1295 #ifdef CONFIG_NUMA
1296 	{
1297 		.ctl_name	= VM_ZONE_RECLAIM_MODE,
1298 		.procname	= "zone_reclaim_mode",
1299 		.data		= &zone_reclaim_mode,
1300 		.maxlen		= sizeof(zone_reclaim_mode),
1301 		.mode		= 0644,
1302 		.proc_handler	= &proc_dointvec,
1303 		.strategy	= &sysctl_intvec,
1304 		.extra1		= &zero,
1305 	},
1306 	{
1307 		.ctl_name	= VM_MIN_UNMAPPED,
1308 		.procname	= "min_unmapped_ratio",
1309 		.data		= &sysctl_min_unmapped_ratio,
1310 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1311 		.mode		= 0644,
1312 		.proc_handler	= &sysctl_min_unmapped_ratio_sysctl_handler,
1313 		.strategy	= &sysctl_intvec,
1314 		.extra1		= &zero,
1315 		.extra2		= &one_hundred,
1316 	},
1317 	{
1318 		.ctl_name	= VM_MIN_SLAB,
1319 		.procname	= "min_slab_ratio",
1320 		.data		= &sysctl_min_slab_ratio,
1321 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1322 		.mode		= 0644,
1323 		.proc_handler	= &sysctl_min_slab_ratio_sysctl_handler,
1324 		.strategy	= &sysctl_intvec,
1325 		.extra1		= &zero,
1326 		.extra2		= &one_hundred,
1327 	},
1328 #endif
1329 #ifdef CONFIG_SMP
1330 	{
1331 		.ctl_name	= CTL_UNNUMBERED,
1332 		.procname	= "stat_interval",
1333 		.data		= &sysctl_stat_interval,
1334 		.maxlen		= sizeof(sysctl_stat_interval),
1335 		.mode		= 0644,
1336 		.proc_handler	= &proc_dointvec_jiffies,
1337 		.strategy	= &sysctl_jiffies,
1338 	},
1339 #endif
1340 	{
1341 		.ctl_name	= CTL_UNNUMBERED,
1342 		.procname	= "mmap_min_addr",
1343 		.data		= &dac_mmap_min_addr,
1344 		.maxlen		= sizeof(unsigned long),
1345 		.mode		= 0644,
1346 		.proc_handler	= &mmap_min_addr_handler,
1347 	},
1348 #ifdef CONFIG_NUMA
1349 	{
1350 		.ctl_name	= CTL_UNNUMBERED,
1351 		.procname	= "numa_zonelist_order",
1352 		.data		= &numa_zonelist_order,
1353 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1354 		.mode		= 0644,
1355 		.proc_handler	= &numa_zonelist_order_handler,
1356 		.strategy	= &sysctl_string,
1357 	},
1358 #endif
1359 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1360    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1361 	{
1362 		.ctl_name	= VM_VDSO_ENABLED,
1363 		.procname	= "vdso_enabled",
1364 		.data		= &vdso_enabled,
1365 		.maxlen		= sizeof(vdso_enabled),
1366 		.mode		= 0644,
1367 		.proc_handler	= &proc_dointvec,
1368 		.strategy	= &sysctl_intvec,
1369 		.extra1		= &zero,
1370 	},
1371 #endif
1372 #ifdef CONFIG_HIGHMEM
1373 	{
1374 		.ctl_name	= CTL_UNNUMBERED,
1375 		.procname	= "highmem_is_dirtyable",
1376 		.data		= &vm_highmem_is_dirtyable,
1377 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1378 		.mode		= 0644,
1379 		.proc_handler	= &proc_dointvec_minmax,
1380 		.strategy	= &sysctl_intvec,
1381 		.extra1		= &zero,
1382 		.extra2		= &one,
1383 	},
1384 #endif
1385 	{
1386 		.ctl_name	= CTL_UNNUMBERED,
1387 		.procname	= "scan_unevictable_pages",
1388 		.data		= &scan_unevictable_pages,
1389 		.maxlen		= sizeof(scan_unevictable_pages),
1390 		.mode		= 0644,
1391 		.proc_handler	= &scan_unevictable_handler,
1392 	},
1393 /*
1394  * NOTE: do not add new entries to this table unless you have read
1395  * Documentation/sysctl/ctl_unnumbered.txt
1396  */
1397 	{ .ctl_name = 0 }
1398 };
1399 
1400 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1401 static struct ctl_table binfmt_misc_table[] = {
1402 	{ .ctl_name = 0 }
1403 };
1404 #endif
1405 
1406 static struct ctl_table fs_table[] = {
1407 	{
1408 		.ctl_name	= FS_NRINODE,
1409 		.procname	= "inode-nr",
1410 		.data		= &inodes_stat,
1411 		.maxlen		= 2*sizeof(int),
1412 		.mode		= 0444,
1413 		.proc_handler	= &proc_dointvec,
1414 	},
1415 	{
1416 		.ctl_name	= FS_STATINODE,
1417 		.procname	= "inode-state",
1418 		.data		= &inodes_stat,
1419 		.maxlen		= 7*sizeof(int),
1420 		.mode		= 0444,
1421 		.proc_handler	= &proc_dointvec,
1422 	},
1423 	{
1424 		.procname	= "file-nr",
1425 		.data		= &files_stat,
1426 		.maxlen		= 3*sizeof(int),
1427 		.mode		= 0444,
1428 		.proc_handler	= &proc_nr_files,
1429 	},
1430 	{
1431 		.ctl_name	= FS_MAXFILE,
1432 		.procname	= "file-max",
1433 		.data		= &files_stat.max_files,
1434 		.maxlen		= sizeof(int),
1435 		.mode		= 0644,
1436 		.proc_handler	= &proc_dointvec,
1437 	},
1438 	{
1439 		.ctl_name	= CTL_UNNUMBERED,
1440 		.procname	= "nr_open",
1441 		.data		= &sysctl_nr_open,
1442 		.maxlen		= sizeof(int),
1443 		.mode		= 0644,
1444 		.proc_handler	= &proc_dointvec_minmax,
1445 		.extra1		= &sysctl_nr_open_min,
1446 		.extra2		= &sysctl_nr_open_max,
1447 	},
1448 	{
1449 		.ctl_name	= FS_DENTRY,
1450 		.procname	= "dentry-state",
1451 		.data		= &dentry_stat,
1452 		.maxlen		= 6*sizeof(int),
1453 		.mode		= 0444,
1454 		.proc_handler	= &proc_dointvec,
1455 	},
1456 	{
1457 		.ctl_name	= FS_OVERFLOWUID,
1458 		.procname	= "overflowuid",
1459 		.data		= &fs_overflowuid,
1460 		.maxlen		= sizeof(int),
1461 		.mode		= 0644,
1462 		.proc_handler	= &proc_dointvec_minmax,
1463 		.strategy	= &sysctl_intvec,
1464 		.extra1		= &minolduid,
1465 		.extra2		= &maxolduid,
1466 	},
1467 	{
1468 		.ctl_name	= FS_OVERFLOWGID,
1469 		.procname	= "overflowgid",
1470 		.data		= &fs_overflowgid,
1471 		.maxlen		= sizeof(int),
1472 		.mode		= 0644,
1473 		.proc_handler	= &proc_dointvec_minmax,
1474 		.strategy	= &sysctl_intvec,
1475 		.extra1		= &minolduid,
1476 		.extra2		= &maxolduid,
1477 	},
1478 #ifdef CONFIG_FILE_LOCKING
1479 	{
1480 		.ctl_name	= FS_LEASES,
1481 		.procname	= "leases-enable",
1482 		.data		= &leases_enable,
1483 		.maxlen		= sizeof(int),
1484 		.mode		= 0644,
1485 		.proc_handler	= &proc_dointvec,
1486 	},
1487 #endif
1488 #ifdef CONFIG_DNOTIFY
1489 	{
1490 		.ctl_name	= FS_DIR_NOTIFY,
1491 		.procname	= "dir-notify-enable",
1492 		.data		= &dir_notify_enable,
1493 		.maxlen		= sizeof(int),
1494 		.mode		= 0644,
1495 		.proc_handler	= &proc_dointvec,
1496 	},
1497 #endif
1498 #ifdef CONFIG_MMU
1499 #ifdef CONFIG_FILE_LOCKING
1500 	{
1501 		.ctl_name	= FS_LEASE_TIME,
1502 		.procname	= "lease-break-time",
1503 		.data		= &lease_break_time,
1504 		.maxlen		= sizeof(int),
1505 		.mode		= 0644,
1506 		.proc_handler	= &proc_dointvec,
1507 	},
1508 #endif
1509 #ifdef CONFIG_AIO
1510 	{
1511 		.procname	= "aio-nr",
1512 		.data		= &aio_nr,
1513 		.maxlen		= sizeof(aio_nr),
1514 		.mode		= 0444,
1515 		.proc_handler	= &proc_doulongvec_minmax,
1516 	},
1517 	{
1518 		.procname	= "aio-max-nr",
1519 		.data		= &aio_max_nr,
1520 		.maxlen		= sizeof(aio_max_nr),
1521 		.mode		= 0644,
1522 		.proc_handler	= &proc_doulongvec_minmax,
1523 	},
1524 #endif /* CONFIG_AIO */
1525 #ifdef CONFIG_INOTIFY_USER
1526 	{
1527 		.ctl_name	= FS_INOTIFY,
1528 		.procname	= "inotify",
1529 		.mode		= 0555,
1530 		.child		= inotify_table,
1531 	},
1532 #endif
1533 #ifdef CONFIG_EPOLL
1534 	{
1535 		.procname	= "epoll",
1536 		.mode		= 0555,
1537 		.child		= epoll_table,
1538 	},
1539 #endif
1540 #endif
1541 	{
1542 		.ctl_name	= KERN_SETUID_DUMPABLE,
1543 		.procname	= "suid_dumpable",
1544 		.data		= &suid_dumpable,
1545 		.maxlen		= sizeof(int),
1546 		.mode		= 0644,
1547 		.proc_handler	= &proc_dointvec_minmax,
1548 		.strategy	= &sysctl_intvec,
1549 		.extra1		= &zero,
1550 		.extra2		= &two,
1551 	},
1552 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1553 	{
1554 		.ctl_name	= CTL_UNNUMBERED,
1555 		.procname	= "binfmt_misc",
1556 		.mode		= 0555,
1557 		.child		= binfmt_misc_table,
1558 	},
1559 #endif
1560 /*
1561  * NOTE: do not add new entries to this table unless you have read
1562  * Documentation/sysctl/ctl_unnumbered.txt
1563  */
1564 	{ .ctl_name = 0 }
1565 };
1566 
1567 static struct ctl_table debug_table[] = {
1568 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1569 	{
1570 		.ctl_name	= CTL_UNNUMBERED,
1571 		.procname	= "exception-trace",
1572 		.data		= &show_unhandled_signals,
1573 		.maxlen		= sizeof(int),
1574 		.mode		= 0644,
1575 		.proc_handler	= proc_dointvec
1576 	},
1577 #endif
1578 	{ .ctl_name = 0 }
1579 };
1580 
1581 static struct ctl_table dev_table[] = {
1582 	{ .ctl_name = 0 }
1583 };
1584 
1585 static DEFINE_SPINLOCK(sysctl_lock);
1586 
1587 /* called under sysctl_lock */
1588 static int use_table(struct ctl_table_header *p)
1589 {
1590 	if (unlikely(p->unregistering))
1591 		return 0;
1592 	p->used++;
1593 	return 1;
1594 }
1595 
1596 /* called under sysctl_lock */
1597 static void unuse_table(struct ctl_table_header *p)
1598 {
1599 	if (!--p->used)
1600 		if (unlikely(p->unregistering))
1601 			complete(p->unregistering);
1602 }
1603 
1604 /* called under sysctl_lock, will reacquire if has to wait */
1605 static void start_unregistering(struct ctl_table_header *p)
1606 {
1607 	/*
1608 	 * if p->used is 0, nobody will ever touch that entry again;
1609 	 * we'll eliminate all paths to it before dropping sysctl_lock
1610 	 */
1611 	if (unlikely(p->used)) {
1612 		struct completion wait;
1613 		init_completion(&wait);
1614 		p->unregistering = &wait;
1615 		spin_unlock(&sysctl_lock);
1616 		wait_for_completion(&wait);
1617 		spin_lock(&sysctl_lock);
1618 	} else {
1619 		/* anything non-NULL; we'll never dereference it */
1620 		p->unregistering = ERR_PTR(-EINVAL);
1621 	}
1622 	/*
1623 	 * do not remove from the list until nobody holds it; walking the
1624 	 * list in do_sysctl() relies on that.
1625 	 */
1626 	list_del_init(&p->ctl_entry);
1627 }
1628 
1629 void sysctl_head_get(struct ctl_table_header *head)
1630 {
1631 	spin_lock(&sysctl_lock);
1632 	head->count++;
1633 	spin_unlock(&sysctl_lock);
1634 }
1635 
1636 void sysctl_head_put(struct ctl_table_header *head)
1637 {
1638 	spin_lock(&sysctl_lock);
1639 	if (!--head->count)
1640 		kfree(head);
1641 	spin_unlock(&sysctl_lock);
1642 }
1643 
1644 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1645 {
1646 	if (!head)
1647 		BUG();
1648 	spin_lock(&sysctl_lock);
1649 	if (!use_table(head))
1650 		head = ERR_PTR(-ENOENT);
1651 	spin_unlock(&sysctl_lock);
1652 	return head;
1653 }
1654 
1655 void sysctl_head_finish(struct ctl_table_header *head)
1656 {
1657 	if (!head)
1658 		return;
1659 	spin_lock(&sysctl_lock);
1660 	unuse_table(head);
1661 	spin_unlock(&sysctl_lock);
1662 }
1663 
1664 static struct ctl_table_set *
1665 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1666 {
1667 	struct ctl_table_set *set = &root->default_set;
1668 	if (root->lookup)
1669 		set = root->lookup(root, namespaces);
1670 	return set;
1671 }
1672 
1673 static struct list_head *
1674 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1675 {
1676 	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1677 	return &set->list;
1678 }
1679 
1680 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1681 					    struct ctl_table_header *prev)
1682 {
1683 	struct ctl_table_root *root;
1684 	struct list_head *header_list;
1685 	struct ctl_table_header *head;
1686 	struct list_head *tmp;
1687 
1688 	spin_lock(&sysctl_lock);
1689 	if (prev) {
1690 		head = prev;
1691 		tmp = &prev->ctl_entry;
1692 		unuse_table(prev);
1693 		goto next;
1694 	}
1695 	tmp = &root_table_header.ctl_entry;
1696 	for (;;) {
1697 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1698 
1699 		if (!use_table(head))
1700 			goto next;
1701 		spin_unlock(&sysctl_lock);
1702 		return head;
1703 	next:
1704 		root = head->root;
1705 		tmp = tmp->next;
1706 		header_list = lookup_header_list(root, namespaces);
1707 		if (tmp != header_list)
1708 			continue;
1709 
1710 		do {
1711 			root = list_entry(root->root_list.next,
1712 					struct ctl_table_root, root_list);
1713 			if (root == &sysctl_table_root)
1714 				goto out;
1715 			header_list = lookup_header_list(root, namespaces);
1716 		} while (list_empty(header_list));
1717 		tmp = header_list->next;
1718 	}
1719 out:
1720 	spin_unlock(&sysctl_lock);
1721 	return NULL;
1722 }
1723 
1724 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1725 {
1726 	return __sysctl_head_next(current->nsproxy, prev);
1727 }
1728 
1729 void register_sysctl_root(struct ctl_table_root *root)
1730 {
1731 	spin_lock(&sysctl_lock);
1732 	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1733 	spin_unlock(&sysctl_lock);
1734 }
1735 
1736 #ifdef CONFIG_SYSCTL_SYSCALL
1737 /* Perform the actual read/write of a sysctl table entry. */
1738 static int do_sysctl_strategy(struct ctl_table_root *root,
1739 			struct ctl_table *table,
1740 			void __user *oldval, size_t __user *oldlenp,
1741 			void __user *newval, size_t newlen)
1742 {
1743 	int op = 0, rc;
1744 
1745 	if (oldval)
1746 		op |= MAY_READ;
1747 	if (newval)
1748 		op |= MAY_WRITE;
1749 	if (sysctl_perm(root, table, op))
1750 		return -EPERM;
1751 
1752 	if (table->strategy) {
1753 		rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1754 		if (rc < 0)
1755 			return rc;
1756 		if (rc > 0)
1757 			return 0;
1758 	}
1759 
1760 	/* If there is no strategy routine, or if the strategy returns
1761 	 * zero, proceed with automatic r/w */
1762 	if (table->data && table->maxlen) {
1763 		rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1764 		if (rc < 0)
1765 			return rc;
1766 	}
1767 	return 0;
1768 }
1769 
1770 static int parse_table(int __user *name, int nlen,
1771 		       void __user *oldval, size_t __user *oldlenp,
1772 		       void __user *newval, size_t newlen,
1773 		       struct ctl_table_root *root,
1774 		       struct ctl_table *table)
1775 {
1776 	int n;
1777 repeat:
1778 	if (!nlen)
1779 		return -ENOTDIR;
1780 	if (get_user(n, name))
1781 		return -EFAULT;
1782 	for ( ; table->ctl_name || table->procname; table++) {
1783 		if (!table->ctl_name)
1784 			continue;
1785 		if (n == table->ctl_name) {
1786 			int error;
1787 			if (table->child) {
1788 				if (sysctl_perm(root, table, MAY_EXEC))
1789 					return -EPERM;
1790 				name++;
1791 				nlen--;
1792 				table = table->child;
1793 				goto repeat;
1794 			}
1795 			error = do_sysctl_strategy(root, table,
1796 						   oldval, oldlenp,
1797 						   newval, newlen);
1798 			return error;
1799 		}
1800 	}
1801 	return -ENOTDIR;
1802 }
1803 
1804 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1805 	       void __user *newval, size_t newlen)
1806 {
1807 	struct ctl_table_header *head;
1808 	int error = -ENOTDIR;
1809 
1810 	if (nlen <= 0 || nlen >= CTL_MAXNAME)
1811 		return -ENOTDIR;
1812 	if (oldval) {
1813 		int old_len;
1814 		if (!oldlenp || get_user(old_len, oldlenp))
1815 			return -EFAULT;
1816 	}
1817 
1818 	for (head = sysctl_head_next(NULL); head;
1819 			head = sysctl_head_next(head)) {
1820 		error = parse_table(name, nlen, oldval, oldlenp,
1821 					newval, newlen,
1822 					head->root, head->ctl_table);
1823 		if (error != -ENOTDIR) {
1824 			sysctl_head_finish(head);
1825 			break;
1826 		}
1827 	}
1828 	return error;
1829 }
1830 
1831 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
1832 {
1833 	struct __sysctl_args tmp;
1834 	int error;
1835 
1836 	if (copy_from_user(&tmp, args, sizeof(tmp)))
1837 		return -EFAULT;
1838 
1839 	error = deprecated_sysctl_warning(&tmp);
1840 	if (error)
1841 		goto out;
1842 
1843 	lock_kernel();
1844 	error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1845 			  tmp.newval, tmp.newlen);
1846 	unlock_kernel();
1847 out:
1848 	return error;
1849 }
1850 #endif /* CONFIG_SYSCTL_SYSCALL */
1851 
1852 /*
1853  * sysctl_perm does NOT grant the superuser all rights automatically, because
1854  * some sysctl variables are readonly even to root.
1855  */
1856 
1857 static int test_perm(int mode, int op)
1858 {
1859 	if (!current_euid())
1860 		mode >>= 6;
1861 	else if (in_egroup_p(0))
1862 		mode >>= 3;
1863 	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1864 		return 0;
1865 	return -EACCES;
1866 }
1867 
1868 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1869 {
1870 	int error;
1871 	int mode;
1872 
1873 	error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1874 	if (error)
1875 		return error;
1876 
1877 	if (root->permissions)
1878 		mode = root->permissions(root, current->nsproxy, table);
1879 	else
1880 		mode = table->mode;
1881 
1882 	return test_perm(mode, op);
1883 }
1884 
1885 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1886 {
1887 	for (; table->ctl_name || table->procname; table++) {
1888 		table->parent = parent;
1889 		if (table->child)
1890 			sysctl_set_parent(table, table->child);
1891 	}
1892 }
1893 
1894 static __init int sysctl_init(void)
1895 {
1896 	sysctl_set_parent(NULL, root_table);
1897 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1898 	{
1899 		int err;
1900 		err = sysctl_check_table(current->nsproxy, root_table);
1901 	}
1902 #endif
1903 	return 0;
1904 }
1905 
1906 core_initcall(sysctl_init);
1907 
1908 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1909 				      struct ctl_table *table)
1910 {
1911 	struct ctl_table *p;
1912 	const char *s = branch->procname;
1913 
1914 	/* branch should have named subdirectory as its first element */
1915 	if (!s || !branch->child)
1916 		return NULL;
1917 
1918 	/* ... and nothing else */
1919 	if (branch[1].procname || branch[1].ctl_name)
1920 		return NULL;
1921 
1922 	/* table should contain subdirectory with the same name */
1923 	for (p = table; p->procname || p->ctl_name; p++) {
1924 		if (!p->child)
1925 			continue;
1926 		if (p->procname && strcmp(p->procname, s) == 0)
1927 			return p;
1928 	}
1929 	return NULL;
1930 }
1931 
1932 /* see if attaching q to p would be an improvement */
1933 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1934 {
1935 	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1936 	struct ctl_table *next;
1937 	int is_better = 0;
1938 	int not_in_parent = !p->attached_by;
1939 
1940 	while ((next = is_branch_in(by, to)) != NULL) {
1941 		if (by == q->attached_by)
1942 			is_better = 1;
1943 		if (to == p->attached_by)
1944 			not_in_parent = 1;
1945 		by = by->child;
1946 		to = next->child;
1947 	}
1948 
1949 	if (is_better && not_in_parent) {
1950 		q->attached_by = by;
1951 		q->attached_to = to;
1952 		q->parent = p;
1953 	}
1954 }
1955 
1956 /**
1957  * __register_sysctl_paths - register a sysctl hierarchy
1958  * @root: List of sysctl headers to register on
1959  * @namespaces: Data to compute which lists of sysctl entries are visible
1960  * @path: The path to the directory the sysctl table is in.
1961  * @table: the top-level table structure
1962  *
1963  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1964  * array. A completely 0 filled entry terminates the table.
1965  *
1966  * The members of the &struct ctl_table structure are used as follows:
1967  *
1968  * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1969  *            must be unique within that level of sysctl
1970  *
1971  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1972  *            enter a sysctl file
1973  *
1974  * data - a pointer to data for use by proc_handler
1975  *
1976  * maxlen - the maximum size in bytes of the data
1977  *
1978  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1979  *
1980  * child - a pointer to the child sysctl table if this entry is a directory, or
1981  *         %NULL.
1982  *
1983  * proc_handler - the text handler routine (described below)
1984  *
1985  * strategy - the strategy routine (described below)
1986  *
1987  * de - for internal use by the sysctl routines
1988  *
1989  * extra1, extra2 - extra pointers usable by the proc handler routines
1990  *
1991  * Leaf nodes in the sysctl tree will be represented by a single file
1992  * under /proc; non-leaf nodes will be represented by directories.
1993  *
1994  * sysctl(2) can automatically manage read and write requests through
1995  * the sysctl table.  The data and maxlen fields of the ctl_table
1996  * struct enable minimal validation of the values being written to be
1997  * performed, and the mode field allows minimal authentication.
1998  *
1999  * More sophisticated management can be enabled by the provision of a
2000  * strategy routine with the table entry.  This will be called before
2001  * any automatic read or write of the data is performed.
2002  *
2003  * The strategy routine may return
2004  *
2005  * < 0 - Error occurred (error is passed to user process)
2006  *
2007  * 0   - OK - proceed with automatic read or write.
2008  *
2009  * > 0 - OK - read or write has been done by the strategy routine, so
2010  *       return immediately.
2011  *
2012  * There must be a proc_handler routine for any terminal nodes
2013  * mirrored under /proc/sys (non-terminals are handled by a built-in
2014  * directory handler).  Several default handlers are available to
2015  * cover common cases -
2016  *
2017  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
2018  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
2019  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
2020  *
2021  * It is the handler's job to read the input buffer from user memory
2022  * and process it. The handler should return 0 on success.
2023  *
2024  * This routine returns %NULL on a failure to register, and a pointer
2025  * to the table header on success.
2026  */
2027 struct ctl_table_header *__register_sysctl_paths(
2028 	struct ctl_table_root *root,
2029 	struct nsproxy *namespaces,
2030 	const struct ctl_path *path, struct ctl_table *table)
2031 {
2032 	struct ctl_table_header *header;
2033 	struct ctl_table *new, **prevp;
2034 	unsigned int n, npath;
2035 	struct ctl_table_set *set;
2036 
2037 	/* Count the path components */
2038 	for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
2039 		;
2040 
2041 	/*
2042 	 * For each path component, allocate a 2-element ctl_table array.
2043 	 * The first array element will be filled with the sysctl entry
2044 	 * for this, the second will be the sentinel (ctl_name == 0).
2045 	 *
2046 	 * We allocate everything in one go so that we don't have to
2047 	 * worry about freeing additional memory in unregister_sysctl_table.
2048 	 */
2049 	header = kzalloc(sizeof(struct ctl_table_header) +
2050 			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
2051 	if (!header)
2052 		return NULL;
2053 
2054 	new = (struct ctl_table *) (header + 1);
2055 
2056 	/* Now connect the dots */
2057 	prevp = &header->ctl_table;
2058 	for (n = 0; n < npath; ++n, ++path) {
2059 		/* Copy the procname */
2060 		new->procname = path->procname;
2061 		new->ctl_name = path->ctl_name;
2062 		new->mode     = 0555;
2063 
2064 		*prevp = new;
2065 		prevp = &new->child;
2066 
2067 		new += 2;
2068 	}
2069 	*prevp = table;
2070 	header->ctl_table_arg = table;
2071 
2072 	INIT_LIST_HEAD(&header->ctl_entry);
2073 	header->used = 0;
2074 	header->unregistering = NULL;
2075 	header->root = root;
2076 	sysctl_set_parent(NULL, header->ctl_table);
2077 	header->count = 1;
2078 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
2079 	if (sysctl_check_table(namespaces, header->ctl_table)) {
2080 		kfree(header);
2081 		return NULL;
2082 	}
2083 #endif
2084 	spin_lock(&sysctl_lock);
2085 	header->set = lookup_header_set(root, namespaces);
2086 	header->attached_by = header->ctl_table;
2087 	header->attached_to = root_table;
2088 	header->parent = &root_table_header;
2089 	for (set = header->set; set; set = set->parent) {
2090 		struct ctl_table_header *p;
2091 		list_for_each_entry(p, &set->list, ctl_entry) {
2092 			if (p->unregistering)
2093 				continue;
2094 			try_attach(p, header);
2095 		}
2096 	}
2097 	header->parent->count++;
2098 	list_add_tail(&header->ctl_entry, &header->set->list);
2099 	spin_unlock(&sysctl_lock);
2100 
2101 	return header;
2102 }
2103 
2104 /**
2105  * register_sysctl_table_path - register a sysctl table hierarchy
2106  * @path: The path to the directory the sysctl table is in.
2107  * @table: the top-level table structure
2108  *
2109  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2110  * array. A completely 0 filled entry terminates the table.
2111  *
2112  * See __register_sysctl_paths for more details.
2113  */
2114 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2115 						struct ctl_table *table)
2116 {
2117 	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
2118 					path, table);
2119 }
2120 
2121 /**
2122  * register_sysctl_table - register a sysctl table hierarchy
2123  * @table: the top-level table structure
2124  *
2125  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2126  * array. A completely 0 filled entry terminates the table.
2127  *
2128  * See register_sysctl_paths for more details.
2129  */
2130 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
2131 {
2132 	static const struct ctl_path null_path[] = { {} };
2133 
2134 	return register_sysctl_paths(null_path, table);
2135 }
2136 
2137 /**
2138  * unregister_sysctl_table - unregister a sysctl table hierarchy
2139  * @header: the header returned from register_sysctl_table
2140  *
2141  * Unregisters the sysctl table and all children. proc entries may not
2142  * actually be removed until they are no longer used by anyone.
2143  */
2144 void unregister_sysctl_table(struct ctl_table_header * header)
2145 {
2146 	might_sleep();
2147 
2148 	if (header == NULL)
2149 		return;
2150 
2151 	spin_lock(&sysctl_lock);
2152 	start_unregistering(header);
2153 	if (!--header->parent->count) {
2154 		WARN_ON(1);
2155 		kfree(header->parent);
2156 	}
2157 	if (!--header->count)
2158 		kfree(header);
2159 	spin_unlock(&sysctl_lock);
2160 }
2161 
2162 int sysctl_is_seen(struct ctl_table_header *p)
2163 {
2164 	struct ctl_table_set *set = p->set;
2165 	int res;
2166 	spin_lock(&sysctl_lock);
2167 	if (p->unregistering)
2168 		res = 0;
2169 	else if (!set->is_seen)
2170 		res = 1;
2171 	else
2172 		res = set->is_seen(set);
2173 	spin_unlock(&sysctl_lock);
2174 	return res;
2175 }
2176 
2177 void setup_sysctl_set(struct ctl_table_set *p,
2178 	struct ctl_table_set *parent,
2179 	int (*is_seen)(struct ctl_table_set *))
2180 {
2181 	INIT_LIST_HEAD(&p->list);
2182 	p->parent = parent ? parent : &sysctl_table_root.default_set;
2183 	p->is_seen = is_seen;
2184 }
2185 
2186 #else /* !CONFIG_SYSCTL */
2187 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2188 {
2189 	return NULL;
2190 }
2191 
2192 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2193 						    struct ctl_table *table)
2194 {
2195 	return NULL;
2196 }
2197 
2198 void unregister_sysctl_table(struct ctl_table_header * table)
2199 {
2200 }
2201 
2202 void setup_sysctl_set(struct ctl_table_set *p,
2203 	struct ctl_table_set *parent,
2204 	int (*is_seen)(struct ctl_table_set *))
2205 {
2206 }
2207 
2208 void sysctl_head_put(struct ctl_table_header *head)
2209 {
2210 }
2211 
2212 #endif /* CONFIG_SYSCTL */
2213 
2214 /*
2215  * /proc/sys support
2216  */
2217 
2218 #ifdef CONFIG_PROC_SYSCTL
2219 
2220 static int _proc_do_string(void* data, int maxlen, int write,
2221 			   struct file *filp, void __user *buffer,
2222 			   size_t *lenp, loff_t *ppos)
2223 {
2224 	size_t len;
2225 	char __user *p;
2226 	char c;
2227 
2228 	if (!data || !maxlen || !*lenp) {
2229 		*lenp = 0;
2230 		return 0;
2231 	}
2232 
2233 	if (write) {
2234 		len = 0;
2235 		p = buffer;
2236 		while (len < *lenp) {
2237 			if (get_user(c, p++))
2238 				return -EFAULT;
2239 			if (c == 0 || c == '\n')
2240 				break;
2241 			len++;
2242 		}
2243 		if (len >= maxlen)
2244 			len = maxlen-1;
2245 		if(copy_from_user(data, buffer, len))
2246 			return -EFAULT;
2247 		((char *) data)[len] = 0;
2248 		*ppos += *lenp;
2249 	} else {
2250 		len = strlen(data);
2251 		if (len > maxlen)
2252 			len = maxlen;
2253 
2254 		if (*ppos > len) {
2255 			*lenp = 0;
2256 			return 0;
2257 		}
2258 
2259 		data += *ppos;
2260 		len  -= *ppos;
2261 
2262 		if (len > *lenp)
2263 			len = *lenp;
2264 		if (len)
2265 			if(copy_to_user(buffer, data, len))
2266 				return -EFAULT;
2267 		if (len < *lenp) {
2268 			if(put_user('\n', ((char __user *) buffer) + len))
2269 				return -EFAULT;
2270 			len++;
2271 		}
2272 		*lenp = len;
2273 		*ppos += len;
2274 	}
2275 	return 0;
2276 }
2277 
2278 /**
2279  * proc_dostring - read a string sysctl
2280  * @table: the sysctl table
2281  * @write: %TRUE if this is a write to the sysctl file
2282  * @filp: the file structure
2283  * @buffer: the user buffer
2284  * @lenp: the size of the user buffer
2285  * @ppos: file position
2286  *
2287  * Reads/writes a string from/to the user buffer. If the kernel
2288  * buffer provided is not large enough to hold the string, the
2289  * string is truncated. The copied string is %NULL-terminated.
2290  * If the string is being read by the user process, it is copied
2291  * and a newline '\n' is added. It is truncated if the buffer is
2292  * not large enough.
2293  *
2294  * Returns 0 on success.
2295  */
2296 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2297 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2298 {
2299 	return _proc_do_string(table->data, table->maxlen, write, filp,
2300 			       buffer, lenp, ppos);
2301 }
2302 
2303 
2304 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2305 				 int *valp,
2306 				 int write, void *data)
2307 {
2308 	if (write) {
2309 		*valp = *negp ? -*lvalp : *lvalp;
2310 	} else {
2311 		int val = *valp;
2312 		if (val < 0) {
2313 			*negp = -1;
2314 			*lvalp = (unsigned long)-val;
2315 		} else {
2316 			*negp = 0;
2317 			*lvalp = (unsigned long)val;
2318 		}
2319 	}
2320 	return 0;
2321 }
2322 
2323 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2324 		  int write, struct file *filp, void __user *buffer,
2325 		  size_t *lenp, loff_t *ppos,
2326 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2327 			      int write, void *data),
2328 		  void *data)
2329 {
2330 #define TMPBUFLEN 21
2331 	int *i, vleft, first = 1, neg;
2332 	unsigned long lval;
2333 	size_t left, len;
2334 
2335 	char buf[TMPBUFLEN], *p;
2336 	char __user *s = buffer;
2337 
2338 	if (!tbl_data || !table->maxlen || !*lenp ||
2339 	    (*ppos && !write)) {
2340 		*lenp = 0;
2341 		return 0;
2342 	}
2343 
2344 	i = (int *) tbl_data;
2345 	vleft = table->maxlen / sizeof(*i);
2346 	left = *lenp;
2347 
2348 	if (!conv)
2349 		conv = do_proc_dointvec_conv;
2350 
2351 	for (; left && vleft--; i++, first=0) {
2352 		if (write) {
2353 			while (left) {
2354 				char c;
2355 				if (get_user(c, s))
2356 					return -EFAULT;
2357 				if (!isspace(c))
2358 					break;
2359 				left--;
2360 				s++;
2361 			}
2362 			if (!left)
2363 				break;
2364 			neg = 0;
2365 			len = left;
2366 			if (len > sizeof(buf) - 1)
2367 				len = sizeof(buf) - 1;
2368 			if (copy_from_user(buf, s, len))
2369 				return -EFAULT;
2370 			buf[len] = 0;
2371 			p = buf;
2372 			if (*p == '-' && left > 1) {
2373 				neg = 1;
2374 				p++;
2375 			}
2376 			if (*p < '0' || *p > '9')
2377 				break;
2378 
2379 			lval = simple_strtoul(p, &p, 0);
2380 
2381 			len = p-buf;
2382 			if ((len < left) && *p && !isspace(*p))
2383 				break;
2384 			s += len;
2385 			left -= len;
2386 
2387 			if (conv(&neg, &lval, i, 1, data))
2388 				break;
2389 		} else {
2390 			p = buf;
2391 			if (!first)
2392 				*p++ = '\t';
2393 
2394 			if (conv(&neg, &lval, i, 0, data))
2395 				break;
2396 
2397 			sprintf(p, "%s%lu", neg ? "-" : "", lval);
2398 			len = strlen(buf);
2399 			if (len > left)
2400 				len = left;
2401 			if(copy_to_user(s, buf, len))
2402 				return -EFAULT;
2403 			left -= len;
2404 			s += len;
2405 		}
2406 	}
2407 
2408 	if (!write && !first && left) {
2409 		if(put_user('\n', s))
2410 			return -EFAULT;
2411 		left--, s++;
2412 	}
2413 	if (write) {
2414 		while (left) {
2415 			char c;
2416 			if (get_user(c, s++))
2417 				return -EFAULT;
2418 			if (!isspace(c))
2419 				break;
2420 			left--;
2421 		}
2422 	}
2423 	if (write && first)
2424 		return -EINVAL;
2425 	*lenp -= left;
2426 	*ppos += *lenp;
2427 	return 0;
2428 #undef TMPBUFLEN
2429 }
2430 
2431 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2432 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2433 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2434 			      int write, void *data),
2435 		  void *data)
2436 {
2437 	return __do_proc_dointvec(table->data, table, write, filp,
2438 			buffer, lenp, ppos, conv, data);
2439 }
2440 
2441 /**
2442  * proc_dointvec - read a vector of integers
2443  * @table: the sysctl table
2444  * @write: %TRUE if this is a write to the sysctl file
2445  * @filp: the file structure
2446  * @buffer: the user buffer
2447  * @lenp: the size of the user buffer
2448  * @ppos: file position
2449  *
2450  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2451  * values from/to the user buffer, treated as an ASCII string.
2452  *
2453  * Returns 0 on success.
2454  */
2455 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2456 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2457 {
2458     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2459 		    	    NULL,NULL);
2460 }
2461 
2462 /*
2463  * Taint values can only be increased
2464  * This means we can safely use a temporary.
2465  */
2466 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
2467 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2468 {
2469 	struct ctl_table t;
2470 	unsigned long tmptaint = get_taint();
2471 	int err;
2472 
2473 	if (write && !capable(CAP_SYS_ADMIN))
2474 		return -EPERM;
2475 
2476 	t = *table;
2477 	t.data = &tmptaint;
2478 	err = proc_doulongvec_minmax(&t, write, filp, buffer, lenp, ppos);
2479 	if (err < 0)
2480 		return err;
2481 
2482 	if (write) {
2483 		/*
2484 		 * Poor man's atomic or. Not worth adding a primitive
2485 		 * to everyone's atomic.h for this
2486 		 */
2487 		int i;
2488 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2489 			if ((tmptaint >> i) & 1)
2490 				add_taint(i);
2491 		}
2492 	}
2493 
2494 	return err;
2495 }
2496 
2497 struct do_proc_dointvec_minmax_conv_param {
2498 	int *min;
2499 	int *max;
2500 };
2501 
2502 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2503 					int *valp,
2504 					int write, void *data)
2505 {
2506 	struct do_proc_dointvec_minmax_conv_param *param = data;
2507 	if (write) {
2508 		int val = *negp ? -*lvalp : *lvalp;
2509 		if ((param->min && *param->min > val) ||
2510 		    (param->max && *param->max < val))
2511 			return -EINVAL;
2512 		*valp = val;
2513 	} else {
2514 		int val = *valp;
2515 		if (val < 0) {
2516 			*negp = -1;
2517 			*lvalp = (unsigned long)-val;
2518 		} else {
2519 			*negp = 0;
2520 			*lvalp = (unsigned long)val;
2521 		}
2522 	}
2523 	return 0;
2524 }
2525 
2526 /**
2527  * proc_dointvec_minmax - read a vector of integers with min/max values
2528  * @table: the sysctl table
2529  * @write: %TRUE if this is a write to the sysctl file
2530  * @filp: the file structure
2531  * @buffer: the user buffer
2532  * @lenp: the size of the user buffer
2533  * @ppos: file position
2534  *
2535  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2536  * values from/to the user buffer, treated as an ASCII string.
2537  *
2538  * This routine will ensure the values are within the range specified by
2539  * table->extra1 (min) and table->extra2 (max).
2540  *
2541  * Returns 0 on success.
2542  */
2543 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2544 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2545 {
2546 	struct do_proc_dointvec_minmax_conv_param param = {
2547 		.min = (int *) table->extra1,
2548 		.max = (int *) table->extra2,
2549 	};
2550 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2551 				do_proc_dointvec_minmax_conv, &param);
2552 }
2553 
2554 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2555 				     struct file *filp,
2556 				     void __user *buffer,
2557 				     size_t *lenp, loff_t *ppos,
2558 				     unsigned long convmul,
2559 				     unsigned long convdiv)
2560 {
2561 #define TMPBUFLEN 21
2562 	unsigned long *i, *min, *max, val;
2563 	int vleft, first=1, neg;
2564 	size_t len, left;
2565 	char buf[TMPBUFLEN], *p;
2566 	char __user *s = buffer;
2567 
2568 	if (!data || !table->maxlen || !*lenp ||
2569 	    (*ppos && !write)) {
2570 		*lenp = 0;
2571 		return 0;
2572 	}
2573 
2574 	i = (unsigned long *) data;
2575 	min = (unsigned long *) table->extra1;
2576 	max = (unsigned long *) table->extra2;
2577 	vleft = table->maxlen / sizeof(unsigned long);
2578 	left = *lenp;
2579 
2580 	for (; left && vleft--; i++, min++, max++, first=0) {
2581 		if (write) {
2582 			while (left) {
2583 				char c;
2584 				if (get_user(c, s))
2585 					return -EFAULT;
2586 				if (!isspace(c))
2587 					break;
2588 				left--;
2589 				s++;
2590 			}
2591 			if (!left)
2592 				break;
2593 			neg = 0;
2594 			len = left;
2595 			if (len > TMPBUFLEN-1)
2596 				len = TMPBUFLEN-1;
2597 			if (copy_from_user(buf, s, len))
2598 				return -EFAULT;
2599 			buf[len] = 0;
2600 			p = buf;
2601 			if (*p == '-' && left > 1) {
2602 				neg = 1;
2603 				p++;
2604 			}
2605 			if (*p < '0' || *p > '9')
2606 				break;
2607 			val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2608 			len = p-buf;
2609 			if ((len < left) && *p && !isspace(*p))
2610 				break;
2611 			if (neg)
2612 				val = -val;
2613 			s += len;
2614 			left -= len;
2615 
2616 			if(neg)
2617 				continue;
2618 			if ((min && val < *min) || (max && val > *max))
2619 				continue;
2620 			*i = val;
2621 		} else {
2622 			p = buf;
2623 			if (!first)
2624 				*p++ = '\t';
2625 			sprintf(p, "%lu", convdiv * (*i) / convmul);
2626 			len = strlen(buf);
2627 			if (len > left)
2628 				len = left;
2629 			if(copy_to_user(s, buf, len))
2630 				return -EFAULT;
2631 			left -= len;
2632 			s += len;
2633 		}
2634 	}
2635 
2636 	if (!write && !first && left) {
2637 		if(put_user('\n', s))
2638 			return -EFAULT;
2639 		left--, s++;
2640 	}
2641 	if (write) {
2642 		while (left) {
2643 			char c;
2644 			if (get_user(c, s++))
2645 				return -EFAULT;
2646 			if (!isspace(c))
2647 				break;
2648 			left--;
2649 		}
2650 	}
2651 	if (write && first)
2652 		return -EINVAL;
2653 	*lenp -= left;
2654 	*ppos += *lenp;
2655 	return 0;
2656 #undef TMPBUFLEN
2657 }
2658 
2659 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2660 				     struct file *filp,
2661 				     void __user *buffer,
2662 				     size_t *lenp, loff_t *ppos,
2663 				     unsigned long convmul,
2664 				     unsigned long convdiv)
2665 {
2666 	return __do_proc_doulongvec_minmax(table->data, table, write,
2667 			filp, buffer, lenp, ppos, convmul, convdiv);
2668 }
2669 
2670 /**
2671  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2672  * @table: the sysctl table
2673  * @write: %TRUE if this is a write to the sysctl file
2674  * @filp: the file structure
2675  * @buffer: the user buffer
2676  * @lenp: the size of the user buffer
2677  * @ppos: file position
2678  *
2679  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2680  * values from/to the user buffer, treated as an ASCII string.
2681  *
2682  * This routine will ensure the values are within the range specified by
2683  * table->extra1 (min) and table->extra2 (max).
2684  *
2685  * Returns 0 on success.
2686  */
2687 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2688 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2689 {
2690     return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2691 }
2692 
2693 /**
2694  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2695  * @table: the sysctl table
2696  * @write: %TRUE if this is a write to the sysctl file
2697  * @filp: the file structure
2698  * @buffer: the user buffer
2699  * @lenp: the size of the user buffer
2700  * @ppos: file position
2701  *
2702  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2703  * values from/to the user buffer, treated as an ASCII string. The values
2704  * are treated as milliseconds, and converted to jiffies when they are stored.
2705  *
2706  * This routine will ensure the values are within the range specified by
2707  * table->extra1 (min) and table->extra2 (max).
2708  *
2709  * Returns 0 on success.
2710  */
2711 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2712 				      struct file *filp,
2713 				      void __user *buffer,
2714 				      size_t *lenp, loff_t *ppos)
2715 {
2716     return do_proc_doulongvec_minmax(table, write, filp, buffer,
2717 				     lenp, ppos, HZ, 1000l);
2718 }
2719 
2720 
2721 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2722 					 int *valp,
2723 					 int write, void *data)
2724 {
2725 	if (write) {
2726 		if (*lvalp > LONG_MAX / HZ)
2727 			return 1;
2728 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2729 	} else {
2730 		int val = *valp;
2731 		unsigned long lval;
2732 		if (val < 0) {
2733 			*negp = -1;
2734 			lval = (unsigned long)-val;
2735 		} else {
2736 			*negp = 0;
2737 			lval = (unsigned long)val;
2738 		}
2739 		*lvalp = lval / HZ;
2740 	}
2741 	return 0;
2742 }
2743 
2744 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2745 						int *valp,
2746 						int write, void *data)
2747 {
2748 	if (write) {
2749 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2750 			return 1;
2751 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2752 	} else {
2753 		int val = *valp;
2754 		unsigned long lval;
2755 		if (val < 0) {
2756 			*negp = -1;
2757 			lval = (unsigned long)-val;
2758 		} else {
2759 			*negp = 0;
2760 			lval = (unsigned long)val;
2761 		}
2762 		*lvalp = jiffies_to_clock_t(lval);
2763 	}
2764 	return 0;
2765 }
2766 
2767 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2768 					    int *valp,
2769 					    int write, void *data)
2770 {
2771 	if (write) {
2772 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2773 	} else {
2774 		int val = *valp;
2775 		unsigned long lval;
2776 		if (val < 0) {
2777 			*negp = -1;
2778 			lval = (unsigned long)-val;
2779 		} else {
2780 			*negp = 0;
2781 			lval = (unsigned long)val;
2782 		}
2783 		*lvalp = jiffies_to_msecs(lval);
2784 	}
2785 	return 0;
2786 }
2787 
2788 /**
2789  * proc_dointvec_jiffies - read a vector of integers as seconds
2790  * @table: the sysctl table
2791  * @write: %TRUE if this is a write to the sysctl file
2792  * @filp: the file structure
2793  * @buffer: the user buffer
2794  * @lenp: the size of the user buffer
2795  * @ppos: file position
2796  *
2797  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2798  * values from/to the user buffer, treated as an ASCII string.
2799  * The values read are assumed to be in seconds, and are converted into
2800  * jiffies.
2801  *
2802  * Returns 0 on success.
2803  */
2804 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2805 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2806 {
2807     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2808 		    	    do_proc_dointvec_jiffies_conv,NULL);
2809 }
2810 
2811 /**
2812  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2813  * @table: the sysctl table
2814  * @write: %TRUE if this is a write to the sysctl file
2815  * @filp: the file structure
2816  * @buffer: the user buffer
2817  * @lenp: the size of the user buffer
2818  * @ppos: pointer to the file position
2819  *
2820  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2821  * values from/to the user buffer, treated as an ASCII string.
2822  * The values read are assumed to be in 1/USER_HZ seconds, and
2823  * are converted into jiffies.
2824  *
2825  * Returns 0 on success.
2826  */
2827 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2828 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2829 {
2830     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2831 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2832 }
2833 
2834 /**
2835  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2836  * @table: the sysctl table
2837  * @write: %TRUE if this is a write to the sysctl file
2838  * @filp: the file structure
2839  * @buffer: the user buffer
2840  * @lenp: the size of the user buffer
2841  * @ppos: file position
2842  * @ppos: the current position in the file
2843  *
2844  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2845  * values from/to the user buffer, treated as an ASCII string.
2846  * The values read are assumed to be in 1/1000 seconds, and
2847  * are converted into jiffies.
2848  *
2849  * Returns 0 on success.
2850  */
2851 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2852 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2853 {
2854 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2855 				do_proc_dointvec_ms_jiffies_conv, NULL);
2856 }
2857 
2858 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2859 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2860 {
2861 	struct pid *new_pid;
2862 	pid_t tmp;
2863 	int r;
2864 
2865 	tmp = pid_vnr(cad_pid);
2866 
2867 	r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2868 			       lenp, ppos, NULL, NULL);
2869 	if (r || !write)
2870 		return r;
2871 
2872 	new_pid = find_get_pid(tmp);
2873 	if (!new_pid)
2874 		return -ESRCH;
2875 
2876 	put_pid(xchg(&cad_pid, new_pid));
2877 	return 0;
2878 }
2879 
2880 #else /* CONFIG_PROC_FS */
2881 
2882 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2883 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2884 {
2885 	return -ENOSYS;
2886 }
2887 
2888 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2889 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2890 {
2891 	return -ENOSYS;
2892 }
2893 
2894 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2895 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2896 {
2897 	return -ENOSYS;
2898 }
2899 
2900 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2901 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2902 {
2903 	return -ENOSYS;
2904 }
2905 
2906 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2907 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2908 {
2909 	return -ENOSYS;
2910 }
2911 
2912 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2913 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2914 {
2915 	return -ENOSYS;
2916 }
2917 
2918 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2919 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2920 {
2921 	return -ENOSYS;
2922 }
2923 
2924 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2925 				      struct file *filp,
2926 				      void __user *buffer,
2927 				      size_t *lenp, loff_t *ppos)
2928 {
2929     return -ENOSYS;
2930 }
2931 
2932 
2933 #endif /* CONFIG_PROC_FS */
2934 
2935 
2936 #ifdef CONFIG_SYSCTL_SYSCALL
2937 /*
2938  * General sysctl support routines
2939  */
2940 
2941 /* The generic sysctl data routine (used if no strategy routine supplied) */
2942 int sysctl_data(struct ctl_table *table,
2943 		void __user *oldval, size_t __user *oldlenp,
2944 		void __user *newval, size_t newlen)
2945 {
2946 	size_t len;
2947 
2948 	/* Get out of I don't have a variable */
2949 	if (!table->data || !table->maxlen)
2950 		return -ENOTDIR;
2951 
2952 	if (oldval && oldlenp) {
2953 		if (get_user(len, oldlenp))
2954 			return -EFAULT;
2955 		if (len) {
2956 			if (len > table->maxlen)
2957 				len = table->maxlen;
2958 			if (copy_to_user(oldval, table->data, len))
2959 				return -EFAULT;
2960 			if (put_user(len, oldlenp))
2961 				return -EFAULT;
2962 		}
2963 	}
2964 
2965 	if (newval && newlen) {
2966 		if (newlen > table->maxlen)
2967 			newlen = table->maxlen;
2968 
2969 		if (copy_from_user(table->data, newval, newlen))
2970 			return -EFAULT;
2971 	}
2972 	return 1;
2973 }
2974 
2975 /* The generic string strategy routine: */
2976 int sysctl_string(struct ctl_table *table,
2977 		  void __user *oldval, size_t __user *oldlenp,
2978 		  void __user *newval, size_t newlen)
2979 {
2980 	if (!table->data || !table->maxlen)
2981 		return -ENOTDIR;
2982 
2983 	if (oldval && oldlenp) {
2984 		size_t bufsize;
2985 		if (get_user(bufsize, oldlenp))
2986 			return -EFAULT;
2987 		if (bufsize) {
2988 			size_t len = strlen(table->data), copied;
2989 
2990 			/* This shouldn't trigger for a well-formed sysctl */
2991 			if (len > table->maxlen)
2992 				len = table->maxlen;
2993 
2994 			/* Copy up to a max of bufsize-1 bytes of the string */
2995 			copied = (len >= bufsize) ? bufsize - 1 : len;
2996 
2997 			if (copy_to_user(oldval, table->data, copied) ||
2998 			    put_user(0, (char __user *)(oldval + copied)))
2999 				return -EFAULT;
3000 			if (put_user(len, oldlenp))
3001 				return -EFAULT;
3002 		}
3003 	}
3004 	if (newval && newlen) {
3005 		size_t len = newlen;
3006 		if (len > table->maxlen)
3007 			len = table->maxlen;
3008 		if(copy_from_user(table->data, newval, len))
3009 			return -EFAULT;
3010 		if (len == table->maxlen)
3011 			len--;
3012 		((char *) table->data)[len] = 0;
3013 	}
3014 	return 1;
3015 }
3016 
3017 /*
3018  * This function makes sure that all of the integers in the vector
3019  * are between the minimum and maximum values given in the arrays
3020  * table->extra1 and table->extra2, respectively.
3021  */
3022 int sysctl_intvec(struct ctl_table *table,
3023 		void __user *oldval, size_t __user *oldlenp,
3024 		void __user *newval, size_t newlen)
3025 {
3026 
3027 	if (newval && newlen) {
3028 		int __user *vec = (int __user *) newval;
3029 		int *min = (int *) table->extra1;
3030 		int *max = (int *) table->extra2;
3031 		size_t length;
3032 		int i;
3033 
3034 		if (newlen % sizeof(int) != 0)
3035 			return -EINVAL;
3036 
3037 		if (!table->extra1 && !table->extra2)
3038 			return 0;
3039 
3040 		if (newlen > table->maxlen)
3041 			newlen = table->maxlen;
3042 		length = newlen / sizeof(int);
3043 
3044 		for (i = 0; i < length; i++) {
3045 			int value;
3046 			if (get_user(value, vec + i))
3047 				return -EFAULT;
3048 			if (min && value < min[i])
3049 				return -EINVAL;
3050 			if (max && value > max[i])
3051 				return -EINVAL;
3052 		}
3053 	}
3054 	return 0;
3055 }
3056 
3057 /* Strategy function to convert jiffies to seconds */
3058 int sysctl_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 = *(int *)(table->data) / HZ;
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) = new*HZ;
3087 	}
3088 	return 1;
3089 }
3090 
3091 /* Strategy function to convert jiffies to seconds */
3092 int sysctl_ms_jiffies(struct ctl_table *table,
3093 		void __user *oldval, size_t __user *oldlenp,
3094 		void __user *newval, size_t newlen)
3095 {
3096 	if (oldval && oldlenp) {
3097 		size_t olen;
3098 
3099 		if (get_user(olen, oldlenp))
3100 			return -EFAULT;
3101 		if (olen) {
3102 			int val;
3103 
3104 			if (olen < sizeof(int))
3105 				return -EINVAL;
3106 
3107 			val = jiffies_to_msecs(*(int *)(table->data));
3108 			if (put_user(val, (int __user *)oldval))
3109 				return -EFAULT;
3110 			if (put_user(sizeof(int), oldlenp))
3111 				return -EFAULT;
3112 		}
3113 	}
3114 	if (newval && newlen) {
3115 		int new;
3116 		if (newlen != sizeof(int))
3117 			return -EINVAL;
3118 		if (get_user(new, (int __user *)newval))
3119 			return -EFAULT;
3120 		*(int *)(table->data) = msecs_to_jiffies(new);
3121 	}
3122 	return 1;
3123 }
3124 
3125 
3126 
3127 #else /* CONFIG_SYSCTL_SYSCALL */
3128 
3129 
3130 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
3131 {
3132 	struct __sysctl_args tmp;
3133 	int error;
3134 
3135 	if (copy_from_user(&tmp, args, sizeof(tmp)))
3136 		return -EFAULT;
3137 
3138 	error = deprecated_sysctl_warning(&tmp);
3139 
3140 	/* If no error reading the parameters then just -ENOSYS ... */
3141 	if (!error)
3142 		error = -ENOSYS;
3143 
3144 	return error;
3145 }
3146 
3147 int sysctl_data(struct ctl_table *table,
3148 		  void __user *oldval, size_t __user *oldlenp,
3149 		  void __user *newval, size_t newlen)
3150 {
3151 	return -ENOSYS;
3152 }
3153 
3154 int sysctl_string(struct ctl_table *table,
3155 		  void __user *oldval, size_t __user *oldlenp,
3156 		  void __user *newval, size_t newlen)
3157 {
3158 	return -ENOSYS;
3159 }
3160 
3161 int sysctl_intvec(struct ctl_table *table,
3162 		void __user *oldval, size_t __user *oldlenp,
3163 		void __user *newval, size_t newlen)
3164 {
3165 	return -ENOSYS;
3166 }
3167 
3168 int sysctl_jiffies(struct ctl_table *table,
3169 		void __user *oldval, size_t __user *oldlenp,
3170 		void __user *newval, size_t newlen)
3171 {
3172 	return -ENOSYS;
3173 }
3174 
3175 int sysctl_ms_jiffies(struct ctl_table *table,
3176 		void __user *oldval, size_t __user *oldlenp,
3177 		void __user *newval, size_t newlen)
3178 {
3179 	return -ENOSYS;
3180 }
3181 
3182 #endif /* CONFIG_SYSCTL_SYSCALL */
3183 
3184 static int deprecated_sysctl_warning(struct __sysctl_args *args)
3185 {
3186 	static int msg_count;
3187 	int name[CTL_MAXNAME];
3188 	int i;
3189 
3190 	/* Check args->nlen. */
3191 	if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
3192 		return -ENOTDIR;
3193 
3194 	/* Read in the sysctl name for better debug message logging */
3195 	for (i = 0; i < args->nlen; i++)
3196 		if (get_user(name[i], args->name + i))
3197 			return -EFAULT;
3198 
3199 	/* Ignore accesses to kernel.version */
3200 	if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
3201 		return 0;
3202 
3203 	if (msg_count < 5) {
3204 		msg_count++;
3205 		printk(KERN_INFO
3206 			"warning: process `%s' used the deprecated sysctl "
3207 			"system call with ", current->comm);
3208 		for (i = 0; i < args->nlen; i++)
3209 			printk("%d.", name[i]);
3210 		printk("\n");
3211 	}
3212 	return 0;
3213 }
3214 
3215 /*
3216  * No sense putting this after each symbol definition, twice,
3217  * exception granted :-)
3218  */
3219 EXPORT_SYMBOL(proc_dointvec);
3220 EXPORT_SYMBOL(proc_dointvec_jiffies);
3221 EXPORT_SYMBOL(proc_dointvec_minmax);
3222 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3223 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3224 EXPORT_SYMBOL(proc_dostring);
3225 EXPORT_SYMBOL(proc_doulongvec_minmax);
3226 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3227 EXPORT_SYMBOL(register_sysctl_table);
3228 EXPORT_SYMBOL(register_sysctl_paths);
3229 EXPORT_SYMBOL(sysctl_intvec);
3230 EXPORT_SYMBOL(sysctl_jiffies);
3231 EXPORT_SYMBOL(sysctl_ms_jiffies);
3232 EXPORT_SYMBOL(sysctl_string);
3233 EXPORT_SYMBOL(sysctl_data);
3234 EXPORT_SYMBOL(unregister_sysctl_table);
3235