xref: /linux-6.15/kernel/sysctl.c (revision f7275650)
1 /*
2  * sysctl.c: General linux system control interface
3  *
4  * Begun 24 March 1995, Stephen Tweedie
5  * Added /proc support, Dec 1995
6  * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7  * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8  * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9  * Dynamic registration fixes, Stephen Tweedie.
10  * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11  * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12  *  Horn.
13  * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14  * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15  * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16  *  Wendling.
17  * The list_for_each() macro wasn't appropriate for the sysctl loop.
18  *  Removed it and replaced it with older style, 03/23/00, Bill Wendling
19  */
20 
21 #include <linux/module.h>
22 #include <linux/mm.h>
23 #include <linux/swap.h>
24 #include <linux/slab.h>
25 #include <linux/sysctl.h>
26 #include <linux/proc_fs.h>
27 #include <linux/security.h>
28 #include <linux/ctype.h>
29 #include <linux/utsname.h>
30 #include <linux/smp_lock.h>
31 #include <linux/fs.h>
32 #include <linux/init.h>
33 #include <linux/kernel.h>
34 #include <linux/kobject.h>
35 #include <linux/net.h>
36 #include <linux/sysrq.h>
37 #include <linux/highuid.h>
38 #include <linux/writeback.h>
39 #include <linux/hugetlb.h>
40 #include <linux/initrd.h>
41 #include <linux/key.h>
42 #include <linux/times.h>
43 #include <linux/limits.h>
44 #include <linux/dcache.h>
45 #include <linux/syscalls.h>
46 #include <linux/vmstat.h>
47 #include <linux/nfs_fs.h>
48 #include <linux/acpi.h>
49 #include <linux/reboot.h>
50 #include <linux/ftrace.h>
51 
52 #include <asm/uaccess.h>
53 #include <asm/processor.h>
54 
55 #ifdef CONFIG_X86
56 #include <asm/nmi.h>
57 #include <asm/stacktrace.h>
58 #include <asm/io.h>
59 #endif
60 
61 static int deprecated_sysctl_warning(struct __sysctl_args *args);
62 
63 #if defined(CONFIG_SYSCTL)
64 
65 /* External variables not in a header file. */
66 extern int C_A_D;
67 extern int print_fatal_signals;
68 extern int sysctl_overcommit_memory;
69 extern int sysctl_overcommit_ratio;
70 extern int sysctl_panic_on_oom;
71 extern int sysctl_oom_kill_allocating_task;
72 extern int sysctl_oom_dump_tasks;
73 extern int max_threads;
74 extern int core_uses_pid;
75 extern int suid_dumpable;
76 extern char core_pattern[];
77 extern int pid_max;
78 extern int min_free_kbytes;
79 extern int pid_max_min, pid_max_max;
80 extern int sysctl_drop_caches;
81 extern int percpu_pagelist_fraction;
82 extern int compat_log;
83 extern int latencytop_enabled;
84 extern int sysctl_nr_open_min, sysctl_nr_open_max;
85 #ifdef CONFIG_RCU_TORTURE_TEST
86 extern int rcutorture_runnable;
87 #endif /* #ifdef CONFIG_RCU_TORTURE_TEST */
88 
89 /* Constants used for minimum and  maximum */
90 #if defined(CONFIG_HIGHMEM) || defined(CONFIG_DETECT_SOFTLOCKUP)
91 static int one = 1;
92 #endif
93 
94 #ifdef CONFIG_DETECT_SOFTLOCKUP
95 static int sixty = 60;
96 static int neg_one = -1;
97 #endif
98 
99 #if defined(CONFIG_MMU) && defined(CONFIG_FILE_LOCKING)
100 static int two = 2;
101 #endif
102 
103 static int zero;
104 static int one_hundred = 100;
105 
106 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
107 static int maxolduid = 65535;
108 static int minolduid;
109 static int min_percpu_pagelist_fract = 8;
110 
111 static int ngroups_max = NGROUPS_MAX;
112 
113 #ifdef CONFIG_MODULES
114 extern char modprobe_path[];
115 #endif
116 #ifdef CONFIG_CHR_DEV_SG
117 extern int sg_big_buff;
118 #endif
119 
120 #ifdef CONFIG_SPARC
121 #include <asm/system.h>
122 #endif
123 
124 #ifdef __hppa__
125 extern int pwrsw_enabled;
126 extern int unaligned_enabled;
127 #endif
128 
129 #ifdef CONFIG_S390
130 #ifdef CONFIG_MATHEMU
131 extern int sysctl_ieee_emulation_warnings;
132 #endif
133 extern int sysctl_userprocess_debug;
134 extern int spin_retry;
135 #endif
136 
137 #ifdef CONFIG_BSD_PROCESS_ACCT
138 extern int acct_parm[];
139 #endif
140 
141 #ifdef CONFIG_IA64
142 extern int no_unaligned_warning;
143 #endif
144 
145 #ifdef CONFIG_RT_MUTEXES
146 extern int max_lock_depth;
147 #endif
148 
149 #ifdef CONFIG_PROC_SYSCTL
150 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
151 		  void __user *buffer, size_t *lenp, loff_t *ppos);
152 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
153 			       void __user *buffer, size_t *lenp, loff_t *ppos);
154 #endif
155 
156 static struct ctl_table root_table[];
157 static struct ctl_table_root sysctl_table_root;
158 static struct ctl_table_header root_table_header = {
159 	.count = 1,
160 	.ctl_table = root_table,
161 	.ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
162 	.root = &sysctl_table_root,
163 	.set = &sysctl_table_root.default_set,
164 };
165 static struct ctl_table_root sysctl_table_root = {
166 	.root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
167 	.default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
168 };
169 
170 static struct ctl_table kern_table[];
171 static struct ctl_table vm_table[];
172 static struct ctl_table fs_table[];
173 static struct ctl_table debug_table[];
174 static struct ctl_table dev_table[];
175 extern struct ctl_table random_table[];
176 #ifdef CONFIG_INOTIFY_USER
177 extern struct ctl_table inotify_table[];
178 #endif
179 
180 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
181 int sysctl_legacy_va_layout;
182 #endif
183 
184 extern int prove_locking;
185 extern int lock_stat;
186 
187 /* The default sysctl tables: */
188 
189 static struct ctl_table root_table[] = {
190 	{
191 		.ctl_name	= CTL_KERN,
192 		.procname	= "kernel",
193 		.mode		= 0555,
194 		.child		= kern_table,
195 	},
196 	{
197 		.ctl_name	= CTL_VM,
198 		.procname	= "vm",
199 		.mode		= 0555,
200 		.child		= vm_table,
201 	},
202 	{
203 		.ctl_name	= CTL_FS,
204 		.procname	= "fs",
205 		.mode		= 0555,
206 		.child		= fs_table,
207 	},
208 	{
209 		.ctl_name	= CTL_DEBUG,
210 		.procname	= "debug",
211 		.mode		= 0555,
212 		.child		= debug_table,
213 	},
214 	{
215 		.ctl_name	= CTL_DEV,
216 		.procname	= "dev",
217 		.mode		= 0555,
218 		.child		= dev_table,
219 	},
220 /*
221  * NOTE: do not add new entries to this table unless you have read
222  * Documentation/sysctl/ctl_unnumbered.txt
223  */
224 	{ .ctl_name = 0 }
225 };
226 
227 #ifdef CONFIG_SCHED_DEBUG
228 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
229 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
230 static int min_wakeup_granularity_ns;			/* 0 usecs */
231 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
232 #endif
233 
234 static struct ctl_table kern_table[] = {
235 #ifdef CONFIG_SCHED_DEBUG
236 	{
237 		.ctl_name	= CTL_UNNUMBERED,
238 		.procname	= "sched_min_granularity_ns",
239 		.data		= &sysctl_sched_min_granularity,
240 		.maxlen		= sizeof(unsigned int),
241 		.mode		= 0644,
242 		.proc_handler	= &sched_nr_latency_handler,
243 		.strategy	= &sysctl_intvec,
244 		.extra1		= &min_sched_granularity_ns,
245 		.extra2		= &max_sched_granularity_ns,
246 	},
247 	{
248 		.ctl_name	= CTL_UNNUMBERED,
249 		.procname	= "sched_latency_ns",
250 		.data		= &sysctl_sched_latency,
251 		.maxlen		= sizeof(unsigned int),
252 		.mode		= 0644,
253 		.proc_handler	= &sched_nr_latency_handler,
254 		.strategy	= &sysctl_intvec,
255 		.extra1		= &min_sched_granularity_ns,
256 		.extra2		= &max_sched_granularity_ns,
257 	},
258 	{
259 		.ctl_name	= CTL_UNNUMBERED,
260 		.procname	= "sched_wakeup_granularity_ns",
261 		.data		= &sysctl_sched_wakeup_granularity,
262 		.maxlen		= sizeof(unsigned int),
263 		.mode		= 0644,
264 		.proc_handler	= &proc_dointvec_minmax,
265 		.strategy	= &sysctl_intvec,
266 		.extra1		= &min_wakeup_granularity_ns,
267 		.extra2		= &max_wakeup_granularity_ns,
268 	},
269 	{
270 		.ctl_name	= CTL_UNNUMBERED,
271 		.procname	= "sched_shares_ratelimit",
272 		.data		= &sysctl_sched_shares_ratelimit,
273 		.maxlen		= sizeof(unsigned int),
274 		.mode		= 0644,
275 		.proc_handler	= &proc_dointvec,
276 	},
277 	{
278 		.ctl_name	= CTL_UNNUMBERED,
279 		.procname	= "sched_child_runs_first",
280 		.data		= &sysctl_sched_child_runs_first,
281 		.maxlen		= sizeof(unsigned int),
282 		.mode		= 0644,
283 		.proc_handler	= &proc_dointvec,
284 	},
285 	{
286 		.ctl_name	= CTL_UNNUMBERED,
287 		.procname	= "sched_features",
288 		.data		= &sysctl_sched_features,
289 		.maxlen		= sizeof(unsigned int),
290 		.mode		= 0644,
291 		.proc_handler	= &proc_dointvec,
292 	},
293 	{
294 		.ctl_name	= CTL_UNNUMBERED,
295 		.procname	= "sched_migration_cost",
296 		.data		= &sysctl_sched_migration_cost,
297 		.maxlen		= sizeof(unsigned int),
298 		.mode		= 0644,
299 		.proc_handler	= &proc_dointvec,
300 	},
301 	{
302 		.ctl_name	= CTL_UNNUMBERED,
303 		.procname	= "sched_nr_migrate",
304 		.data		= &sysctl_sched_nr_migrate,
305 		.maxlen		= sizeof(unsigned int),
306 		.mode		= 0644,
307 		.proc_handler	= &proc_dointvec,
308 	},
309 #endif
310 	{
311 		.ctl_name	= CTL_UNNUMBERED,
312 		.procname	= "sched_rt_period_us",
313 		.data		= &sysctl_sched_rt_period,
314 		.maxlen		= sizeof(unsigned int),
315 		.mode		= 0644,
316 		.proc_handler	= &sched_rt_handler,
317 	},
318 	{
319 		.ctl_name	= CTL_UNNUMBERED,
320 		.procname	= "sched_rt_runtime_us",
321 		.data		= &sysctl_sched_rt_runtime,
322 		.maxlen		= sizeof(int),
323 		.mode		= 0644,
324 		.proc_handler	= &sched_rt_handler,
325 	},
326 	{
327 		.ctl_name	= CTL_UNNUMBERED,
328 		.procname	= "sched_compat_yield",
329 		.data		= &sysctl_sched_compat_yield,
330 		.maxlen		= sizeof(unsigned int),
331 		.mode		= 0644,
332 		.proc_handler	= &proc_dointvec,
333 	},
334 #ifdef CONFIG_PROVE_LOCKING
335 	{
336 		.ctl_name	= CTL_UNNUMBERED,
337 		.procname	= "prove_locking",
338 		.data		= &prove_locking,
339 		.maxlen		= sizeof(int),
340 		.mode		= 0644,
341 		.proc_handler	= &proc_dointvec,
342 	},
343 #endif
344 #ifdef CONFIG_LOCK_STAT
345 	{
346 		.ctl_name	= CTL_UNNUMBERED,
347 		.procname	= "lock_stat",
348 		.data		= &lock_stat,
349 		.maxlen		= sizeof(int),
350 		.mode		= 0644,
351 		.proc_handler	= &proc_dointvec,
352 	},
353 #endif
354 	{
355 		.ctl_name	= KERN_PANIC,
356 		.procname	= "panic",
357 		.data		= &panic_timeout,
358 		.maxlen		= sizeof(int),
359 		.mode		= 0644,
360 		.proc_handler	= &proc_dointvec,
361 	},
362 	{
363 		.ctl_name	= KERN_CORE_USES_PID,
364 		.procname	= "core_uses_pid",
365 		.data		= &core_uses_pid,
366 		.maxlen		= sizeof(int),
367 		.mode		= 0644,
368 		.proc_handler	= &proc_dointvec,
369 	},
370 	{
371 		.ctl_name	= KERN_CORE_PATTERN,
372 		.procname	= "core_pattern",
373 		.data		= core_pattern,
374 		.maxlen		= CORENAME_MAX_SIZE,
375 		.mode		= 0644,
376 		.proc_handler	= &proc_dostring,
377 		.strategy	= &sysctl_string,
378 	},
379 #ifdef CONFIG_PROC_SYSCTL
380 	{
381 		.procname	= "tainted",
382 		.maxlen 	= sizeof(long),
383 		.mode		= 0644,
384 		.proc_handler	= &proc_taint,
385 	},
386 #endif
387 #ifdef CONFIG_LATENCYTOP
388 	{
389 		.procname	= "latencytop",
390 		.data		= &latencytop_enabled,
391 		.maxlen		= sizeof(int),
392 		.mode		= 0644,
393 		.proc_handler	= &proc_dointvec,
394 	},
395 #endif
396 #ifdef CONFIG_BLK_DEV_INITRD
397 	{
398 		.ctl_name	= KERN_REALROOTDEV,
399 		.procname	= "real-root-dev",
400 		.data		= &real_root_dev,
401 		.maxlen		= sizeof(int),
402 		.mode		= 0644,
403 		.proc_handler	= &proc_dointvec,
404 	},
405 #endif
406 	{
407 		.ctl_name	= CTL_UNNUMBERED,
408 		.procname	= "print-fatal-signals",
409 		.data		= &print_fatal_signals,
410 		.maxlen		= sizeof(int),
411 		.mode		= 0644,
412 		.proc_handler	= &proc_dointvec,
413 	},
414 #ifdef CONFIG_SPARC
415 	{
416 		.ctl_name	= KERN_SPARC_REBOOT,
417 		.procname	= "reboot-cmd",
418 		.data		= reboot_command,
419 		.maxlen		= 256,
420 		.mode		= 0644,
421 		.proc_handler	= &proc_dostring,
422 		.strategy	= &sysctl_string,
423 	},
424 	{
425 		.ctl_name	= KERN_SPARC_STOP_A,
426 		.procname	= "stop-a",
427 		.data		= &stop_a_enabled,
428 		.maxlen		= sizeof (int),
429 		.mode		= 0644,
430 		.proc_handler	= &proc_dointvec,
431 	},
432 	{
433 		.ctl_name	= KERN_SPARC_SCONS_PWROFF,
434 		.procname	= "scons-poweroff",
435 		.data		= &scons_pwroff,
436 		.maxlen		= sizeof (int),
437 		.mode		= 0644,
438 		.proc_handler	= &proc_dointvec,
439 	},
440 #endif
441 #ifdef __hppa__
442 	{
443 		.ctl_name	= KERN_HPPA_PWRSW,
444 		.procname	= "soft-power",
445 		.data		= &pwrsw_enabled,
446 		.maxlen		= sizeof (int),
447 	 	.mode		= 0644,
448 		.proc_handler	= &proc_dointvec,
449 	},
450 	{
451 		.ctl_name	= KERN_HPPA_UNALIGNED,
452 		.procname	= "unaligned-trap",
453 		.data		= &unaligned_enabled,
454 		.maxlen		= sizeof (int),
455 		.mode		= 0644,
456 		.proc_handler	= &proc_dointvec,
457 	},
458 #endif
459 	{
460 		.ctl_name	= KERN_CTLALTDEL,
461 		.procname	= "ctrl-alt-del",
462 		.data		= &C_A_D,
463 		.maxlen		= sizeof(int),
464 		.mode		= 0644,
465 		.proc_handler	= &proc_dointvec,
466 	},
467 #ifdef CONFIG_FTRACE
468 	{
469 		.ctl_name	= CTL_UNNUMBERED,
470 		.procname	= "ftrace_enabled",
471 		.data		= &ftrace_enabled,
472 		.maxlen		= sizeof(int),
473 		.mode		= 0644,
474 		.proc_handler	= &ftrace_enable_sysctl,
475 	},
476 #endif
477 #ifdef CONFIG_MODULES
478 	{
479 		.ctl_name	= KERN_MODPROBE,
480 		.procname	= "modprobe",
481 		.data		= &modprobe_path,
482 		.maxlen		= KMOD_PATH_LEN,
483 		.mode		= 0644,
484 		.proc_handler	= &proc_dostring,
485 		.strategy	= &sysctl_string,
486 	},
487 #endif
488 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
489 	{
490 		.ctl_name	= KERN_HOTPLUG,
491 		.procname	= "hotplug",
492 		.data		= &uevent_helper,
493 		.maxlen		= UEVENT_HELPER_PATH_LEN,
494 		.mode		= 0644,
495 		.proc_handler	= &proc_dostring,
496 		.strategy	= &sysctl_string,
497 	},
498 #endif
499 #ifdef CONFIG_CHR_DEV_SG
500 	{
501 		.ctl_name	= KERN_SG_BIG_BUFF,
502 		.procname	= "sg-big-buff",
503 		.data		= &sg_big_buff,
504 		.maxlen		= sizeof (int),
505 		.mode		= 0444,
506 		.proc_handler	= &proc_dointvec,
507 	},
508 #endif
509 #ifdef CONFIG_BSD_PROCESS_ACCT
510 	{
511 		.ctl_name	= KERN_ACCT,
512 		.procname	= "acct",
513 		.data		= &acct_parm,
514 		.maxlen		= 3*sizeof(int),
515 		.mode		= 0644,
516 		.proc_handler	= &proc_dointvec,
517 	},
518 #endif
519 #ifdef CONFIG_MAGIC_SYSRQ
520 	{
521 		.ctl_name	= KERN_SYSRQ,
522 		.procname	= "sysrq",
523 		.data		= &__sysrq_enabled,
524 		.maxlen		= sizeof (int),
525 		.mode		= 0644,
526 		.proc_handler	= &proc_dointvec,
527 	},
528 #endif
529 #ifdef CONFIG_PROC_SYSCTL
530 	{
531 		.procname	= "cad_pid",
532 		.data		= NULL,
533 		.maxlen		= sizeof (int),
534 		.mode		= 0600,
535 		.proc_handler	= &proc_do_cad_pid,
536 	},
537 #endif
538 	{
539 		.ctl_name	= KERN_MAX_THREADS,
540 		.procname	= "threads-max",
541 		.data		= &max_threads,
542 		.maxlen		= sizeof(int),
543 		.mode		= 0644,
544 		.proc_handler	= &proc_dointvec,
545 	},
546 	{
547 		.ctl_name	= KERN_RANDOM,
548 		.procname	= "random",
549 		.mode		= 0555,
550 		.child		= random_table,
551 	},
552 	{
553 		.ctl_name	= KERN_OVERFLOWUID,
554 		.procname	= "overflowuid",
555 		.data		= &overflowuid,
556 		.maxlen		= sizeof(int),
557 		.mode		= 0644,
558 		.proc_handler	= &proc_dointvec_minmax,
559 		.strategy	= &sysctl_intvec,
560 		.extra1		= &minolduid,
561 		.extra2		= &maxolduid,
562 	},
563 	{
564 		.ctl_name	= KERN_OVERFLOWGID,
565 		.procname	= "overflowgid",
566 		.data		= &overflowgid,
567 		.maxlen		= sizeof(int),
568 		.mode		= 0644,
569 		.proc_handler	= &proc_dointvec_minmax,
570 		.strategy	= &sysctl_intvec,
571 		.extra1		= &minolduid,
572 		.extra2		= &maxolduid,
573 	},
574 #ifdef CONFIG_S390
575 #ifdef CONFIG_MATHEMU
576 	{
577 		.ctl_name	= KERN_IEEE_EMULATION_WARNINGS,
578 		.procname	= "ieee_emulation_warnings",
579 		.data		= &sysctl_ieee_emulation_warnings,
580 		.maxlen		= sizeof(int),
581 		.mode		= 0644,
582 		.proc_handler	= &proc_dointvec,
583 	},
584 #endif
585 	{
586 		.ctl_name	= KERN_S390_USER_DEBUG_LOGGING,
587 		.procname	= "userprocess_debug",
588 		.data		= &sysctl_userprocess_debug,
589 		.maxlen		= sizeof(int),
590 		.mode		= 0644,
591 		.proc_handler	= &proc_dointvec,
592 	},
593 #endif
594 	{
595 		.ctl_name	= KERN_PIDMAX,
596 		.procname	= "pid_max",
597 		.data		= &pid_max,
598 		.maxlen		= sizeof (int),
599 		.mode		= 0644,
600 		.proc_handler	= &proc_dointvec_minmax,
601 		.strategy	= sysctl_intvec,
602 		.extra1		= &pid_max_min,
603 		.extra2		= &pid_max_max,
604 	},
605 	{
606 		.ctl_name	= KERN_PANIC_ON_OOPS,
607 		.procname	= "panic_on_oops",
608 		.data		= &panic_on_oops,
609 		.maxlen		= sizeof(int),
610 		.mode		= 0644,
611 		.proc_handler	= &proc_dointvec,
612 	},
613 #if defined CONFIG_PRINTK
614 	{
615 		.ctl_name	= KERN_PRINTK,
616 		.procname	= "printk",
617 		.data		= &console_loglevel,
618 		.maxlen		= 4*sizeof(int),
619 		.mode		= 0644,
620 		.proc_handler	= &proc_dointvec,
621 	},
622 	{
623 		.ctl_name	= KERN_PRINTK_RATELIMIT,
624 		.procname	= "printk_ratelimit",
625 		.data		= &printk_ratelimit_state.interval,
626 		.maxlen		= sizeof(int),
627 		.mode		= 0644,
628 		.proc_handler	= &proc_dointvec_jiffies,
629 		.strategy	= &sysctl_jiffies,
630 	},
631 	{
632 		.ctl_name	= KERN_PRINTK_RATELIMIT_BURST,
633 		.procname	= "printk_ratelimit_burst",
634 		.data		= &printk_ratelimit_state.burst,
635 		.maxlen		= sizeof(int),
636 		.mode		= 0644,
637 		.proc_handler	= &proc_dointvec,
638 	},
639 #endif
640 	{
641 		.ctl_name	= KERN_NGROUPS_MAX,
642 		.procname	= "ngroups_max",
643 		.data		= &ngroups_max,
644 		.maxlen		= sizeof (int),
645 		.mode		= 0444,
646 		.proc_handler	= &proc_dointvec,
647 	},
648 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
649 	{
650 		.ctl_name       = KERN_UNKNOWN_NMI_PANIC,
651 		.procname       = "unknown_nmi_panic",
652 		.data           = &unknown_nmi_panic,
653 		.maxlen         = sizeof (int),
654 		.mode           = 0644,
655 		.proc_handler   = &proc_dointvec,
656 	},
657 	{
658 		.procname       = "nmi_watchdog",
659 		.data           = &nmi_watchdog_enabled,
660 		.maxlen         = sizeof (int),
661 		.mode           = 0644,
662 		.proc_handler   = &proc_nmi_enabled,
663 	},
664 #endif
665 #if defined(CONFIG_X86)
666 	{
667 		.ctl_name	= KERN_PANIC_ON_NMI,
668 		.procname	= "panic_on_unrecovered_nmi",
669 		.data		= &panic_on_unrecovered_nmi,
670 		.maxlen		= sizeof(int),
671 		.mode		= 0644,
672 		.proc_handler	= &proc_dointvec,
673 	},
674 	{
675 		.ctl_name	= KERN_BOOTLOADER_TYPE,
676 		.procname	= "bootloader_type",
677 		.data		= &bootloader_type,
678 		.maxlen		= sizeof (int),
679 		.mode		= 0444,
680 		.proc_handler	= &proc_dointvec,
681 	},
682 	{
683 		.ctl_name	= CTL_UNNUMBERED,
684 		.procname	= "kstack_depth_to_print",
685 		.data		= &kstack_depth_to_print,
686 		.maxlen		= sizeof(int),
687 		.mode		= 0644,
688 		.proc_handler	= &proc_dointvec,
689 	},
690 	{
691 		.ctl_name	= CTL_UNNUMBERED,
692 		.procname	= "io_delay_type",
693 		.data		= &io_delay_type,
694 		.maxlen		= sizeof(int),
695 		.mode		= 0644,
696 		.proc_handler	= &proc_dointvec,
697 	},
698 #endif
699 #if defined(CONFIG_MMU)
700 	{
701 		.ctl_name	= KERN_RANDOMIZE,
702 		.procname	= "randomize_va_space",
703 		.data		= &randomize_va_space,
704 		.maxlen		= sizeof(int),
705 		.mode		= 0644,
706 		.proc_handler	= &proc_dointvec,
707 	},
708 #endif
709 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
710 	{
711 		.ctl_name	= KERN_SPIN_RETRY,
712 		.procname	= "spin_retry",
713 		.data		= &spin_retry,
714 		.maxlen		= sizeof (int),
715 		.mode		= 0644,
716 		.proc_handler	= &proc_dointvec,
717 	},
718 #endif
719 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
720 	{
721 		.procname	= "acpi_video_flags",
722 		.data		= &acpi_realmode_flags,
723 		.maxlen		= sizeof (unsigned long),
724 		.mode		= 0644,
725 		.proc_handler	= &proc_doulongvec_minmax,
726 	},
727 #endif
728 #ifdef CONFIG_IA64
729 	{
730 		.ctl_name	= KERN_IA64_UNALIGNED,
731 		.procname	= "ignore-unaligned-usertrap",
732 		.data		= &no_unaligned_warning,
733 		.maxlen		= sizeof (int),
734 	 	.mode		= 0644,
735 		.proc_handler	= &proc_dointvec,
736 	},
737 #endif
738 #ifdef CONFIG_DETECT_SOFTLOCKUP
739 	{
740 		.ctl_name	= CTL_UNNUMBERED,
741 		.procname	= "softlockup_panic",
742 		.data		= &softlockup_panic,
743 		.maxlen		= sizeof(int),
744 		.mode		= 0644,
745 		.proc_handler	= &proc_dointvec_minmax,
746 		.strategy	= &sysctl_intvec,
747 		.extra1		= &zero,
748 		.extra2		= &one,
749 	},
750 	{
751 		.ctl_name	= CTL_UNNUMBERED,
752 		.procname	= "softlockup_thresh",
753 		.data		= &softlockup_thresh,
754 		.maxlen		= sizeof(int),
755 		.mode		= 0644,
756 		.proc_handler	= &proc_dointvec_minmax,
757 		.strategy	= &sysctl_intvec,
758 		.extra1		= &neg_one,
759 		.extra2		= &sixty,
760 	},
761 	{
762 		.ctl_name	= CTL_UNNUMBERED,
763 		.procname	= "hung_task_check_count",
764 		.data		= &sysctl_hung_task_check_count,
765 		.maxlen		= sizeof(unsigned long),
766 		.mode		= 0644,
767 		.proc_handler	= &proc_doulongvec_minmax,
768 		.strategy	= &sysctl_intvec,
769 	},
770 	{
771 		.ctl_name	= CTL_UNNUMBERED,
772 		.procname	= "hung_task_timeout_secs",
773 		.data		= &sysctl_hung_task_timeout_secs,
774 		.maxlen		= sizeof(unsigned long),
775 		.mode		= 0644,
776 		.proc_handler	= &proc_doulongvec_minmax,
777 		.strategy	= &sysctl_intvec,
778 	},
779 	{
780 		.ctl_name	= CTL_UNNUMBERED,
781 		.procname	= "hung_task_warnings",
782 		.data		= &sysctl_hung_task_warnings,
783 		.maxlen		= sizeof(unsigned long),
784 		.mode		= 0644,
785 		.proc_handler	= &proc_doulongvec_minmax,
786 		.strategy	= &sysctl_intvec,
787 	},
788 #endif
789 #ifdef CONFIG_COMPAT
790 	{
791 		.ctl_name	= KERN_COMPAT_LOG,
792 		.procname	= "compat-log",
793 		.data		= &compat_log,
794 		.maxlen		= sizeof (int),
795 	 	.mode		= 0644,
796 		.proc_handler	= &proc_dointvec,
797 	},
798 #endif
799 #ifdef CONFIG_RT_MUTEXES
800 	{
801 		.ctl_name	= KERN_MAX_LOCK_DEPTH,
802 		.procname	= "max_lock_depth",
803 		.data		= &max_lock_depth,
804 		.maxlen		= sizeof(int),
805 		.mode		= 0644,
806 		.proc_handler	= &proc_dointvec,
807 	},
808 #endif
809 	{
810 		.ctl_name	= CTL_UNNUMBERED,
811 		.procname	= "poweroff_cmd",
812 		.data		= &poweroff_cmd,
813 		.maxlen		= POWEROFF_CMD_PATH_LEN,
814 		.mode		= 0644,
815 		.proc_handler	= &proc_dostring,
816 		.strategy	= &sysctl_string,
817 	},
818 #ifdef CONFIG_KEYS
819 	{
820 		.ctl_name	= CTL_UNNUMBERED,
821 		.procname	= "keys",
822 		.mode		= 0555,
823 		.child		= key_sysctls,
824 	},
825 #endif
826 #ifdef CONFIG_RCU_TORTURE_TEST
827 	{
828 		.ctl_name       = CTL_UNNUMBERED,
829 		.procname       = "rcutorture_runnable",
830 		.data           = &rcutorture_runnable,
831 		.maxlen         = sizeof(int),
832 		.mode           = 0644,
833 		.proc_handler   = &proc_dointvec,
834 	},
835 #endif
836 /*
837  * NOTE: do not add new entries to this table unless you have read
838  * Documentation/sysctl/ctl_unnumbered.txt
839  */
840 	{ .ctl_name = 0 }
841 };
842 
843 static struct ctl_table vm_table[] = {
844 	{
845 		.ctl_name	= VM_OVERCOMMIT_MEMORY,
846 		.procname	= "overcommit_memory",
847 		.data		= &sysctl_overcommit_memory,
848 		.maxlen		= sizeof(sysctl_overcommit_memory),
849 		.mode		= 0644,
850 		.proc_handler	= &proc_dointvec,
851 	},
852 	{
853 		.ctl_name	= VM_PANIC_ON_OOM,
854 		.procname	= "panic_on_oom",
855 		.data		= &sysctl_panic_on_oom,
856 		.maxlen		= sizeof(sysctl_panic_on_oom),
857 		.mode		= 0644,
858 		.proc_handler	= &proc_dointvec,
859 	},
860 	{
861 		.ctl_name	= CTL_UNNUMBERED,
862 		.procname	= "oom_kill_allocating_task",
863 		.data		= &sysctl_oom_kill_allocating_task,
864 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
865 		.mode		= 0644,
866 		.proc_handler	= &proc_dointvec,
867 	},
868 	{
869 		.ctl_name	= CTL_UNNUMBERED,
870 		.procname	= "oom_dump_tasks",
871 		.data		= &sysctl_oom_dump_tasks,
872 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
873 		.mode		= 0644,
874 		.proc_handler	= &proc_dointvec,
875 	},
876 	{
877 		.ctl_name	= VM_OVERCOMMIT_RATIO,
878 		.procname	= "overcommit_ratio",
879 		.data		= &sysctl_overcommit_ratio,
880 		.maxlen		= sizeof(sysctl_overcommit_ratio),
881 		.mode		= 0644,
882 		.proc_handler	= &proc_dointvec,
883 	},
884 	{
885 		.ctl_name	= VM_PAGE_CLUSTER,
886 		.procname	= "page-cluster",
887 		.data		= &page_cluster,
888 		.maxlen		= sizeof(int),
889 		.mode		= 0644,
890 		.proc_handler	= &proc_dointvec,
891 	},
892 	{
893 		.ctl_name	= VM_DIRTY_BACKGROUND,
894 		.procname	= "dirty_background_ratio",
895 		.data		= &dirty_background_ratio,
896 		.maxlen		= sizeof(dirty_background_ratio),
897 		.mode		= 0644,
898 		.proc_handler	= &proc_dointvec_minmax,
899 		.strategy	= &sysctl_intvec,
900 		.extra1		= &zero,
901 		.extra2		= &one_hundred,
902 	},
903 	{
904 		.ctl_name	= VM_DIRTY_RATIO,
905 		.procname	= "dirty_ratio",
906 		.data		= &vm_dirty_ratio,
907 		.maxlen		= sizeof(vm_dirty_ratio),
908 		.mode		= 0644,
909 		.proc_handler	= &dirty_ratio_handler,
910 		.strategy	= &sysctl_intvec,
911 		.extra1		= &zero,
912 		.extra2		= &one_hundred,
913 	},
914 	{
915 		.procname	= "dirty_writeback_centisecs",
916 		.data		= &dirty_writeback_interval,
917 		.maxlen		= sizeof(dirty_writeback_interval),
918 		.mode		= 0644,
919 		.proc_handler	= &dirty_writeback_centisecs_handler,
920 	},
921 	{
922 		.procname	= "dirty_expire_centisecs",
923 		.data		= &dirty_expire_interval,
924 		.maxlen		= sizeof(dirty_expire_interval),
925 		.mode		= 0644,
926 		.proc_handler	= &proc_dointvec_userhz_jiffies,
927 	},
928 	{
929 		.ctl_name	= VM_NR_PDFLUSH_THREADS,
930 		.procname	= "nr_pdflush_threads",
931 		.data		= &nr_pdflush_threads,
932 		.maxlen		= sizeof nr_pdflush_threads,
933 		.mode		= 0444 /* read-only*/,
934 		.proc_handler	= &proc_dointvec,
935 	},
936 	{
937 		.ctl_name	= VM_SWAPPINESS,
938 		.procname	= "swappiness",
939 		.data		= &vm_swappiness,
940 		.maxlen		= sizeof(vm_swappiness),
941 		.mode		= 0644,
942 		.proc_handler	= &proc_dointvec_minmax,
943 		.strategy	= &sysctl_intvec,
944 		.extra1		= &zero,
945 		.extra2		= &one_hundred,
946 	},
947 #ifdef CONFIG_HUGETLB_PAGE
948 	 {
949 		.procname	= "nr_hugepages",
950 		.data		= NULL,
951 		.maxlen		= sizeof(unsigned long),
952 		.mode		= 0644,
953 		.proc_handler	= &hugetlb_sysctl_handler,
954 		.extra1		= (void *)&hugetlb_zero,
955 		.extra2		= (void *)&hugetlb_infinity,
956 	 },
957 	 {
958 		.ctl_name	= VM_HUGETLB_GROUP,
959 		.procname	= "hugetlb_shm_group",
960 		.data		= &sysctl_hugetlb_shm_group,
961 		.maxlen		= sizeof(gid_t),
962 		.mode		= 0644,
963 		.proc_handler	= &proc_dointvec,
964 	 },
965 	 {
966 		.ctl_name	= CTL_UNNUMBERED,
967 		.procname	= "hugepages_treat_as_movable",
968 		.data		= &hugepages_treat_as_movable,
969 		.maxlen		= sizeof(int),
970 		.mode		= 0644,
971 		.proc_handler	= &hugetlb_treat_movable_handler,
972 	},
973 	{
974 		.ctl_name	= CTL_UNNUMBERED,
975 		.procname	= "nr_overcommit_hugepages",
976 		.data		= NULL,
977 		.maxlen		= sizeof(unsigned long),
978 		.mode		= 0644,
979 		.proc_handler	= &hugetlb_overcommit_handler,
980 		.extra1		= (void *)&hugetlb_zero,
981 		.extra2		= (void *)&hugetlb_infinity,
982 	},
983 #endif
984 	{
985 		.ctl_name	= VM_LOWMEM_RESERVE_RATIO,
986 		.procname	= "lowmem_reserve_ratio",
987 		.data		= &sysctl_lowmem_reserve_ratio,
988 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
989 		.mode		= 0644,
990 		.proc_handler	= &lowmem_reserve_ratio_sysctl_handler,
991 		.strategy	= &sysctl_intvec,
992 	},
993 	{
994 		.ctl_name	= VM_DROP_PAGECACHE,
995 		.procname	= "drop_caches",
996 		.data		= &sysctl_drop_caches,
997 		.maxlen		= sizeof(int),
998 		.mode		= 0644,
999 		.proc_handler	= drop_caches_sysctl_handler,
1000 		.strategy	= &sysctl_intvec,
1001 	},
1002 	{
1003 		.ctl_name	= VM_MIN_FREE_KBYTES,
1004 		.procname	= "min_free_kbytes",
1005 		.data		= &min_free_kbytes,
1006 		.maxlen		= sizeof(min_free_kbytes),
1007 		.mode		= 0644,
1008 		.proc_handler	= &min_free_kbytes_sysctl_handler,
1009 		.strategy	= &sysctl_intvec,
1010 		.extra1		= &zero,
1011 	},
1012 	{
1013 		.ctl_name	= VM_PERCPU_PAGELIST_FRACTION,
1014 		.procname	= "percpu_pagelist_fraction",
1015 		.data		= &percpu_pagelist_fraction,
1016 		.maxlen		= sizeof(percpu_pagelist_fraction),
1017 		.mode		= 0644,
1018 		.proc_handler	= &percpu_pagelist_fraction_sysctl_handler,
1019 		.strategy	= &sysctl_intvec,
1020 		.extra1		= &min_percpu_pagelist_fract,
1021 	},
1022 #ifdef CONFIG_MMU
1023 	{
1024 		.ctl_name	= VM_MAX_MAP_COUNT,
1025 		.procname	= "max_map_count",
1026 		.data		= &sysctl_max_map_count,
1027 		.maxlen		= sizeof(sysctl_max_map_count),
1028 		.mode		= 0644,
1029 		.proc_handler	= &proc_dointvec
1030 	},
1031 #endif
1032 	{
1033 		.ctl_name	= VM_LAPTOP_MODE,
1034 		.procname	= "laptop_mode",
1035 		.data		= &laptop_mode,
1036 		.maxlen		= sizeof(laptop_mode),
1037 		.mode		= 0644,
1038 		.proc_handler	= &proc_dointvec_jiffies,
1039 		.strategy	= &sysctl_jiffies,
1040 	},
1041 	{
1042 		.ctl_name	= VM_BLOCK_DUMP,
1043 		.procname	= "block_dump",
1044 		.data		= &block_dump,
1045 		.maxlen		= sizeof(block_dump),
1046 		.mode		= 0644,
1047 		.proc_handler	= &proc_dointvec,
1048 		.strategy	= &sysctl_intvec,
1049 		.extra1		= &zero,
1050 	},
1051 	{
1052 		.ctl_name	= VM_VFS_CACHE_PRESSURE,
1053 		.procname	= "vfs_cache_pressure",
1054 		.data		= &sysctl_vfs_cache_pressure,
1055 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1056 		.mode		= 0644,
1057 		.proc_handler	= &proc_dointvec,
1058 		.strategy	= &sysctl_intvec,
1059 		.extra1		= &zero,
1060 	},
1061 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1062 	{
1063 		.ctl_name	= VM_LEGACY_VA_LAYOUT,
1064 		.procname	= "legacy_va_layout",
1065 		.data		= &sysctl_legacy_va_layout,
1066 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1067 		.mode		= 0644,
1068 		.proc_handler	= &proc_dointvec,
1069 		.strategy	= &sysctl_intvec,
1070 		.extra1		= &zero,
1071 	},
1072 #endif
1073 #ifdef CONFIG_NUMA
1074 	{
1075 		.ctl_name	= VM_ZONE_RECLAIM_MODE,
1076 		.procname	= "zone_reclaim_mode",
1077 		.data		= &zone_reclaim_mode,
1078 		.maxlen		= sizeof(zone_reclaim_mode),
1079 		.mode		= 0644,
1080 		.proc_handler	= &proc_dointvec,
1081 		.strategy	= &sysctl_intvec,
1082 		.extra1		= &zero,
1083 	},
1084 	{
1085 		.ctl_name	= VM_MIN_UNMAPPED,
1086 		.procname	= "min_unmapped_ratio",
1087 		.data		= &sysctl_min_unmapped_ratio,
1088 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1089 		.mode		= 0644,
1090 		.proc_handler	= &sysctl_min_unmapped_ratio_sysctl_handler,
1091 		.strategy	= &sysctl_intvec,
1092 		.extra1		= &zero,
1093 		.extra2		= &one_hundred,
1094 	},
1095 	{
1096 		.ctl_name	= VM_MIN_SLAB,
1097 		.procname	= "min_slab_ratio",
1098 		.data		= &sysctl_min_slab_ratio,
1099 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1100 		.mode		= 0644,
1101 		.proc_handler	= &sysctl_min_slab_ratio_sysctl_handler,
1102 		.strategy	= &sysctl_intvec,
1103 		.extra1		= &zero,
1104 		.extra2		= &one_hundred,
1105 	},
1106 #endif
1107 #ifdef CONFIG_SMP
1108 	{
1109 		.ctl_name	= CTL_UNNUMBERED,
1110 		.procname	= "stat_interval",
1111 		.data		= &sysctl_stat_interval,
1112 		.maxlen		= sizeof(sysctl_stat_interval),
1113 		.mode		= 0644,
1114 		.proc_handler	= &proc_dointvec_jiffies,
1115 		.strategy	= &sysctl_jiffies,
1116 	},
1117 #endif
1118 #ifdef CONFIG_SECURITY
1119 	{
1120 		.ctl_name	= CTL_UNNUMBERED,
1121 		.procname	= "mmap_min_addr",
1122 		.data		= &mmap_min_addr,
1123 		.maxlen         = sizeof(unsigned long),
1124 		.mode		= 0644,
1125 		.proc_handler	= &proc_doulongvec_minmax,
1126 	},
1127 #endif
1128 #ifdef CONFIG_NUMA
1129 	{
1130 		.ctl_name	= CTL_UNNUMBERED,
1131 		.procname	= "numa_zonelist_order",
1132 		.data		= &numa_zonelist_order,
1133 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1134 		.mode		= 0644,
1135 		.proc_handler	= &numa_zonelist_order_handler,
1136 		.strategy	= &sysctl_string,
1137 	},
1138 #endif
1139 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1140    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1141 	{
1142 		.ctl_name	= VM_VDSO_ENABLED,
1143 		.procname	= "vdso_enabled",
1144 		.data		= &vdso_enabled,
1145 		.maxlen		= sizeof(vdso_enabled),
1146 		.mode		= 0644,
1147 		.proc_handler	= &proc_dointvec,
1148 		.strategy	= &sysctl_intvec,
1149 		.extra1		= &zero,
1150 	},
1151 #endif
1152 #ifdef CONFIG_HIGHMEM
1153 	{
1154 		.ctl_name	= CTL_UNNUMBERED,
1155 		.procname	= "highmem_is_dirtyable",
1156 		.data		= &vm_highmem_is_dirtyable,
1157 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1158 		.mode		= 0644,
1159 		.proc_handler	= &proc_dointvec_minmax,
1160 		.strategy	= &sysctl_intvec,
1161 		.extra1		= &zero,
1162 		.extra2		= &one,
1163 	},
1164 #endif
1165 /*
1166  * NOTE: do not add new entries to this table unless you have read
1167  * Documentation/sysctl/ctl_unnumbered.txt
1168  */
1169 	{ .ctl_name = 0 }
1170 };
1171 
1172 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1173 static struct ctl_table binfmt_misc_table[] = {
1174 	{ .ctl_name = 0 }
1175 };
1176 #endif
1177 
1178 static struct ctl_table fs_table[] = {
1179 	{
1180 		.ctl_name	= FS_NRINODE,
1181 		.procname	= "inode-nr",
1182 		.data		= &inodes_stat,
1183 		.maxlen		= 2*sizeof(int),
1184 		.mode		= 0444,
1185 		.proc_handler	= &proc_dointvec,
1186 	},
1187 	{
1188 		.ctl_name	= FS_STATINODE,
1189 		.procname	= "inode-state",
1190 		.data		= &inodes_stat,
1191 		.maxlen		= 7*sizeof(int),
1192 		.mode		= 0444,
1193 		.proc_handler	= &proc_dointvec,
1194 	},
1195 	{
1196 		.procname	= "file-nr",
1197 		.data		= &files_stat,
1198 		.maxlen		= 3*sizeof(int),
1199 		.mode		= 0444,
1200 		.proc_handler	= &proc_nr_files,
1201 	},
1202 	{
1203 		.ctl_name	= FS_MAXFILE,
1204 		.procname	= "file-max",
1205 		.data		= &files_stat.max_files,
1206 		.maxlen		= sizeof(int),
1207 		.mode		= 0644,
1208 		.proc_handler	= &proc_dointvec,
1209 	},
1210 	{
1211 		.ctl_name	= CTL_UNNUMBERED,
1212 		.procname	= "nr_open",
1213 		.data		= &sysctl_nr_open,
1214 		.maxlen		= sizeof(int),
1215 		.mode		= 0644,
1216 		.proc_handler	= &proc_dointvec_minmax,
1217 		.extra1		= &sysctl_nr_open_min,
1218 		.extra2		= &sysctl_nr_open_max,
1219 	},
1220 	{
1221 		.ctl_name	= FS_DENTRY,
1222 		.procname	= "dentry-state",
1223 		.data		= &dentry_stat,
1224 		.maxlen		= 6*sizeof(int),
1225 		.mode		= 0444,
1226 		.proc_handler	= &proc_dointvec,
1227 	},
1228 	{
1229 		.ctl_name	= FS_OVERFLOWUID,
1230 		.procname	= "overflowuid",
1231 		.data		= &fs_overflowuid,
1232 		.maxlen		= sizeof(int),
1233 		.mode		= 0644,
1234 		.proc_handler	= &proc_dointvec_minmax,
1235 		.strategy	= &sysctl_intvec,
1236 		.extra1		= &minolduid,
1237 		.extra2		= &maxolduid,
1238 	},
1239 	{
1240 		.ctl_name	= FS_OVERFLOWGID,
1241 		.procname	= "overflowgid",
1242 		.data		= &fs_overflowgid,
1243 		.maxlen		= sizeof(int),
1244 		.mode		= 0644,
1245 		.proc_handler	= &proc_dointvec_minmax,
1246 		.strategy	= &sysctl_intvec,
1247 		.extra1		= &minolduid,
1248 		.extra2		= &maxolduid,
1249 	},
1250 #ifdef CONFIG_FILE_LOCKING
1251 	{
1252 		.ctl_name	= FS_LEASES,
1253 		.procname	= "leases-enable",
1254 		.data		= &leases_enable,
1255 		.maxlen		= sizeof(int),
1256 		.mode		= 0644,
1257 		.proc_handler	= &proc_dointvec,
1258 	},
1259 #endif
1260 #ifdef CONFIG_DNOTIFY
1261 	{
1262 		.ctl_name	= FS_DIR_NOTIFY,
1263 		.procname	= "dir-notify-enable",
1264 		.data		= &dir_notify_enable,
1265 		.maxlen		= sizeof(int),
1266 		.mode		= 0644,
1267 		.proc_handler	= &proc_dointvec,
1268 	},
1269 #endif
1270 #ifdef CONFIG_MMU
1271 #ifdef CONFIG_FILE_LOCKING
1272 	{
1273 		.ctl_name	= FS_LEASE_TIME,
1274 		.procname	= "lease-break-time",
1275 		.data		= &lease_break_time,
1276 		.maxlen		= sizeof(int),
1277 		.mode		= 0644,
1278 		.proc_handler	= &proc_dointvec_minmax,
1279 		.strategy	= &sysctl_intvec,
1280 		.extra1		= &zero,
1281 		.extra2		= &two,
1282 	},
1283 #endif
1284 #ifdef CONFIG_AIO
1285 	{
1286 		.procname	= "aio-nr",
1287 		.data		= &aio_nr,
1288 		.maxlen		= sizeof(aio_nr),
1289 		.mode		= 0444,
1290 		.proc_handler	= &proc_doulongvec_minmax,
1291 	},
1292 	{
1293 		.procname	= "aio-max-nr",
1294 		.data		= &aio_max_nr,
1295 		.maxlen		= sizeof(aio_max_nr),
1296 		.mode		= 0644,
1297 		.proc_handler	= &proc_doulongvec_minmax,
1298 	},
1299 #endif /* CONFIG_AIO */
1300 #ifdef CONFIG_INOTIFY_USER
1301 	{
1302 		.ctl_name	= FS_INOTIFY,
1303 		.procname	= "inotify",
1304 		.mode		= 0555,
1305 		.child		= inotify_table,
1306 	},
1307 #endif
1308 #endif
1309 	{
1310 		.ctl_name	= KERN_SETUID_DUMPABLE,
1311 		.procname	= "suid_dumpable",
1312 		.data		= &suid_dumpable,
1313 		.maxlen		= sizeof(int),
1314 		.mode		= 0644,
1315 		.proc_handler	= &proc_dointvec,
1316 	},
1317 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1318 	{
1319 		.ctl_name	= CTL_UNNUMBERED,
1320 		.procname	= "binfmt_misc",
1321 		.mode		= 0555,
1322 		.child		= binfmt_misc_table,
1323 	},
1324 #endif
1325 /*
1326  * NOTE: do not add new entries to this table unless you have read
1327  * Documentation/sysctl/ctl_unnumbered.txt
1328  */
1329 	{ .ctl_name = 0 }
1330 };
1331 
1332 static struct ctl_table debug_table[] = {
1333 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1334 	{
1335 		.ctl_name	= CTL_UNNUMBERED,
1336 		.procname	= "exception-trace",
1337 		.data		= &show_unhandled_signals,
1338 		.maxlen		= sizeof(int),
1339 		.mode		= 0644,
1340 		.proc_handler	= proc_dointvec
1341 	},
1342 #endif
1343 	{ .ctl_name = 0 }
1344 };
1345 
1346 static struct ctl_table dev_table[] = {
1347 	{ .ctl_name = 0 }
1348 };
1349 
1350 static DEFINE_SPINLOCK(sysctl_lock);
1351 
1352 /* called under sysctl_lock */
1353 static int use_table(struct ctl_table_header *p)
1354 {
1355 	if (unlikely(p->unregistering))
1356 		return 0;
1357 	p->used++;
1358 	return 1;
1359 }
1360 
1361 /* called under sysctl_lock */
1362 static void unuse_table(struct ctl_table_header *p)
1363 {
1364 	if (!--p->used)
1365 		if (unlikely(p->unregistering))
1366 			complete(p->unregistering);
1367 }
1368 
1369 /* called under sysctl_lock, will reacquire if has to wait */
1370 static void start_unregistering(struct ctl_table_header *p)
1371 {
1372 	/*
1373 	 * if p->used is 0, nobody will ever touch that entry again;
1374 	 * we'll eliminate all paths to it before dropping sysctl_lock
1375 	 */
1376 	if (unlikely(p->used)) {
1377 		struct completion wait;
1378 		init_completion(&wait);
1379 		p->unregistering = &wait;
1380 		spin_unlock(&sysctl_lock);
1381 		wait_for_completion(&wait);
1382 		spin_lock(&sysctl_lock);
1383 	} else {
1384 		/* anything non-NULL; we'll never dereference it */
1385 		p->unregistering = ERR_PTR(-EINVAL);
1386 	}
1387 	/*
1388 	 * do not remove from the list until nobody holds it; walking the
1389 	 * list in do_sysctl() relies on that.
1390 	 */
1391 	list_del_init(&p->ctl_entry);
1392 }
1393 
1394 void sysctl_head_get(struct ctl_table_header *head)
1395 {
1396 	spin_lock(&sysctl_lock);
1397 	head->count++;
1398 	spin_unlock(&sysctl_lock);
1399 }
1400 
1401 void sysctl_head_put(struct ctl_table_header *head)
1402 {
1403 	spin_lock(&sysctl_lock);
1404 	if (!--head->count)
1405 		kfree(head);
1406 	spin_unlock(&sysctl_lock);
1407 }
1408 
1409 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1410 {
1411 	if (!head)
1412 		BUG();
1413 	spin_lock(&sysctl_lock);
1414 	if (!use_table(head))
1415 		head = ERR_PTR(-ENOENT);
1416 	spin_unlock(&sysctl_lock);
1417 	return head;
1418 }
1419 
1420 void sysctl_head_finish(struct ctl_table_header *head)
1421 {
1422 	if (!head)
1423 		return;
1424 	spin_lock(&sysctl_lock);
1425 	unuse_table(head);
1426 	spin_unlock(&sysctl_lock);
1427 }
1428 
1429 static struct ctl_table_set *
1430 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1431 {
1432 	struct ctl_table_set *set = &root->default_set;
1433 	if (root->lookup)
1434 		set = root->lookup(root, namespaces);
1435 	return set;
1436 }
1437 
1438 static struct list_head *
1439 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1440 {
1441 	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1442 	return &set->list;
1443 }
1444 
1445 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1446 					    struct ctl_table_header *prev)
1447 {
1448 	struct ctl_table_root *root;
1449 	struct list_head *header_list;
1450 	struct ctl_table_header *head;
1451 	struct list_head *tmp;
1452 
1453 	spin_lock(&sysctl_lock);
1454 	if (prev) {
1455 		head = prev;
1456 		tmp = &prev->ctl_entry;
1457 		unuse_table(prev);
1458 		goto next;
1459 	}
1460 	tmp = &root_table_header.ctl_entry;
1461 	for (;;) {
1462 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1463 
1464 		if (!use_table(head))
1465 			goto next;
1466 		spin_unlock(&sysctl_lock);
1467 		return head;
1468 	next:
1469 		root = head->root;
1470 		tmp = tmp->next;
1471 		header_list = lookup_header_list(root, namespaces);
1472 		if (tmp != header_list)
1473 			continue;
1474 
1475 		do {
1476 			root = list_entry(root->root_list.next,
1477 					struct ctl_table_root, root_list);
1478 			if (root == &sysctl_table_root)
1479 				goto out;
1480 			header_list = lookup_header_list(root, namespaces);
1481 		} while (list_empty(header_list));
1482 		tmp = header_list->next;
1483 	}
1484 out:
1485 	spin_unlock(&sysctl_lock);
1486 	return NULL;
1487 }
1488 
1489 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1490 {
1491 	return __sysctl_head_next(current->nsproxy, prev);
1492 }
1493 
1494 void register_sysctl_root(struct ctl_table_root *root)
1495 {
1496 	spin_lock(&sysctl_lock);
1497 	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1498 	spin_unlock(&sysctl_lock);
1499 }
1500 
1501 #ifdef CONFIG_SYSCTL_SYSCALL
1502 /* Perform the actual read/write of a sysctl table entry. */
1503 static int do_sysctl_strategy(struct ctl_table_root *root,
1504 			struct ctl_table *table,
1505 			void __user *oldval, size_t __user *oldlenp,
1506 			void __user *newval, size_t newlen)
1507 {
1508 	int op = 0, rc;
1509 
1510 	if (oldval)
1511 		op |= MAY_READ;
1512 	if (newval)
1513 		op |= MAY_WRITE;
1514 	if (sysctl_perm(root, table, op))
1515 		return -EPERM;
1516 
1517 	if (table->strategy) {
1518 		rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1519 		if (rc < 0)
1520 			return rc;
1521 		if (rc > 0)
1522 			return 0;
1523 	}
1524 
1525 	/* If there is no strategy routine, or if the strategy returns
1526 	 * zero, proceed with automatic r/w */
1527 	if (table->data && table->maxlen) {
1528 		rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1529 		if (rc < 0)
1530 			return rc;
1531 	}
1532 	return 0;
1533 }
1534 
1535 static int parse_table(int __user *name, int nlen,
1536 		       void __user *oldval, size_t __user *oldlenp,
1537 		       void __user *newval, size_t newlen,
1538 		       struct ctl_table_root *root,
1539 		       struct ctl_table *table)
1540 {
1541 	int n;
1542 repeat:
1543 	if (!nlen)
1544 		return -ENOTDIR;
1545 	if (get_user(n, name))
1546 		return -EFAULT;
1547 	for ( ; table->ctl_name || table->procname; table++) {
1548 		if (!table->ctl_name)
1549 			continue;
1550 		if (n == table->ctl_name) {
1551 			int error;
1552 			if (table->child) {
1553 				if (sysctl_perm(root, table, MAY_EXEC))
1554 					return -EPERM;
1555 				name++;
1556 				nlen--;
1557 				table = table->child;
1558 				goto repeat;
1559 			}
1560 			error = do_sysctl_strategy(root, table,
1561 						   oldval, oldlenp,
1562 						   newval, newlen);
1563 			return error;
1564 		}
1565 	}
1566 	return -ENOTDIR;
1567 }
1568 
1569 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1570 	       void __user *newval, size_t newlen)
1571 {
1572 	struct ctl_table_header *head;
1573 	int error = -ENOTDIR;
1574 
1575 	if (nlen <= 0 || nlen >= CTL_MAXNAME)
1576 		return -ENOTDIR;
1577 	if (oldval) {
1578 		int old_len;
1579 		if (!oldlenp || get_user(old_len, oldlenp))
1580 			return -EFAULT;
1581 	}
1582 
1583 	for (head = sysctl_head_next(NULL); head;
1584 			head = sysctl_head_next(head)) {
1585 		error = parse_table(name, nlen, oldval, oldlenp,
1586 					newval, newlen,
1587 					head->root, head->ctl_table);
1588 		if (error != -ENOTDIR) {
1589 			sysctl_head_finish(head);
1590 			break;
1591 		}
1592 	}
1593 	return error;
1594 }
1595 
1596 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1597 {
1598 	struct __sysctl_args tmp;
1599 	int error;
1600 
1601 	if (copy_from_user(&tmp, args, sizeof(tmp)))
1602 		return -EFAULT;
1603 
1604 	error = deprecated_sysctl_warning(&tmp);
1605 	if (error)
1606 		goto out;
1607 
1608 	lock_kernel();
1609 	error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1610 			  tmp.newval, tmp.newlen);
1611 	unlock_kernel();
1612 out:
1613 	return error;
1614 }
1615 #endif /* CONFIG_SYSCTL_SYSCALL */
1616 
1617 /*
1618  * sysctl_perm does NOT grant the superuser all rights automatically, because
1619  * some sysctl variables are readonly even to root.
1620  */
1621 
1622 static int test_perm(int mode, int op)
1623 {
1624 	if (!current->euid)
1625 		mode >>= 6;
1626 	else if (in_egroup_p(0))
1627 		mode >>= 3;
1628 	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1629 		return 0;
1630 	return -EACCES;
1631 }
1632 
1633 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1634 {
1635 	int error;
1636 	int mode;
1637 
1638 	error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1639 	if (error)
1640 		return error;
1641 
1642 	if (root->permissions)
1643 		mode = root->permissions(root, current->nsproxy, table);
1644 	else
1645 		mode = table->mode;
1646 
1647 	return test_perm(mode, op);
1648 }
1649 
1650 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1651 {
1652 	for (; table->ctl_name || table->procname; table++) {
1653 		table->parent = parent;
1654 		if (table->child)
1655 			sysctl_set_parent(table, table->child);
1656 	}
1657 }
1658 
1659 static __init int sysctl_init(void)
1660 {
1661 	sysctl_set_parent(NULL, root_table);
1662 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1663 	{
1664 		int err;
1665 		err = sysctl_check_table(current->nsproxy, root_table);
1666 	}
1667 #endif
1668 	return 0;
1669 }
1670 
1671 core_initcall(sysctl_init);
1672 
1673 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1674 				      struct ctl_table *table)
1675 {
1676 	struct ctl_table *p;
1677 	const char *s = branch->procname;
1678 
1679 	/* branch should have named subdirectory as its first element */
1680 	if (!s || !branch->child)
1681 		return NULL;
1682 
1683 	/* ... and nothing else */
1684 	if (branch[1].procname || branch[1].ctl_name)
1685 		return NULL;
1686 
1687 	/* table should contain subdirectory with the same name */
1688 	for (p = table; p->procname || p->ctl_name; p++) {
1689 		if (!p->child)
1690 			continue;
1691 		if (p->procname && strcmp(p->procname, s) == 0)
1692 			return p;
1693 	}
1694 	return NULL;
1695 }
1696 
1697 /* see if attaching q to p would be an improvement */
1698 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1699 {
1700 	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1701 	struct ctl_table *next;
1702 	int is_better = 0;
1703 	int not_in_parent = !p->attached_by;
1704 
1705 	while ((next = is_branch_in(by, to)) != NULL) {
1706 		if (by == q->attached_by)
1707 			is_better = 1;
1708 		if (to == p->attached_by)
1709 			not_in_parent = 1;
1710 		by = by->child;
1711 		to = next->child;
1712 	}
1713 
1714 	if (is_better && not_in_parent) {
1715 		q->attached_by = by;
1716 		q->attached_to = to;
1717 		q->parent = p;
1718 	}
1719 }
1720 
1721 /**
1722  * __register_sysctl_paths - register a sysctl hierarchy
1723  * @root: List of sysctl headers to register on
1724  * @namespaces: Data to compute which lists of sysctl entries are visible
1725  * @path: The path to the directory the sysctl table is in.
1726  * @table: the top-level table structure
1727  *
1728  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1729  * array. A completely 0 filled entry terminates the table.
1730  *
1731  * The members of the &struct ctl_table structure are used as follows:
1732  *
1733  * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1734  *            must be unique within that level of sysctl
1735  *
1736  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1737  *            enter a sysctl file
1738  *
1739  * data - a pointer to data for use by proc_handler
1740  *
1741  * maxlen - the maximum size in bytes of the data
1742  *
1743  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1744  *
1745  * child - a pointer to the child sysctl table if this entry is a directory, or
1746  *         %NULL.
1747  *
1748  * proc_handler - the text handler routine (described below)
1749  *
1750  * strategy - the strategy routine (described below)
1751  *
1752  * de - for internal use by the sysctl routines
1753  *
1754  * extra1, extra2 - extra pointers usable by the proc handler routines
1755  *
1756  * Leaf nodes in the sysctl tree will be represented by a single file
1757  * under /proc; non-leaf nodes will be represented by directories.
1758  *
1759  * sysctl(2) can automatically manage read and write requests through
1760  * the sysctl table.  The data and maxlen fields of the ctl_table
1761  * struct enable minimal validation of the values being written to be
1762  * performed, and the mode field allows minimal authentication.
1763  *
1764  * More sophisticated management can be enabled by the provision of a
1765  * strategy routine with the table entry.  This will be called before
1766  * any automatic read or write of the data is performed.
1767  *
1768  * The strategy routine may return
1769  *
1770  * < 0 - Error occurred (error is passed to user process)
1771  *
1772  * 0   - OK - proceed with automatic read or write.
1773  *
1774  * > 0 - OK - read or write has been done by the strategy routine, so
1775  *       return immediately.
1776  *
1777  * There must be a proc_handler routine for any terminal nodes
1778  * mirrored under /proc/sys (non-terminals are handled by a built-in
1779  * directory handler).  Several default handlers are available to
1780  * cover common cases -
1781  *
1782  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1783  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1784  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1785  *
1786  * It is the handler's job to read the input buffer from user memory
1787  * and process it. The handler should return 0 on success.
1788  *
1789  * This routine returns %NULL on a failure to register, and a pointer
1790  * to the table header on success.
1791  */
1792 struct ctl_table_header *__register_sysctl_paths(
1793 	struct ctl_table_root *root,
1794 	struct nsproxy *namespaces,
1795 	const struct ctl_path *path, struct ctl_table *table)
1796 {
1797 	struct ctl_table_header *header;
1798 	struct ctl_table *new, **prevp;
1799 	unsigned int n, npath;
1800 	struct ctl_table_set *set;
1801 
1802 	/* Count the path components */
1803 	for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1804 		;
1805 
1806 	/*
1807 	 * For each path component, allocate a 2-element ctl_table array.
1808 	 * The first array element will be filled with the sysctl entry
1809 	 * for this, the second will be the sentinel (ctl_name == 0).
1810 	 *
1811 	 * We allocate everything in one go so that we don't have to
1812 	 * worry about freeing additional memory in unregister_sysctl_table.
1813 	 */
1814 	header = kzalloc(sizeof(struct ctl_table_header) +
1815 			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1816 	if (!header)
1817 		return NULL;
1818 
1819 	new = (struct ctl_table *) (header + 1);
1820 
1821 	/* Now connect the dots */
1822 	prevp = &header->ctl_table;
1823 	for (n = 0; n < npath; ++n, ++path) {
1824 		/* Copy the procname */
1825 		new->procname = path->procname;
1826 		new->ctl_name = path->ctl_name;
1827 		new->mode     = 0555;
1828 
1829 		*prevp = new;
1830 		prevp = &new->child;
1831 
1832 		new += 2;
1833 	}
1834 	*prevp = table;
1835 	header->ctl_table_arg = table;
1836 
1837 	INIT_LIST_HEAD(&header->ctl_entry);
1838 	header->used = 0;
1839 	header->unregistering = NULL;
1840 	header->root = root;
1841 	sysctl_set_parent(NULL, header->ctl_table);
1842 	header->count = 1;
1843 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1844 	if (sysctl_check_table(namespaces, header->ctl_table)) {
1845 		kfree(header);
1846 		return NULL;
1847 	}
1848 #endif
1849 	spin_lock(&sysctl_lock);
1850 	header->set = lookup_header_set(root, namespaces);
1851 	header->attached_by = header->ctl_table;
1852 	header->attached_to = root_table;
1853 	header->parent = &root_table_header;
1854 	for (set = header->set; set; set = set->parent) {
1855 		struct ctl_table_header *p;
1856 		list_for_each_entry(p, &set->list, ctl_entry) {
1857 			if (p->unregistering)
1858 				continue;
1859 			try_attach(p, header);
1860 		}
1861 	}
1862 	header->parent->count++;
1863 	list_add_tail(&header->ctl_entry, &header->set->list);
1864 	spin_unlock(&sysctl_lock);
1865 
1866 	return header;
1867 }
1868 
1869 /**
1870  * register_sysctl_table_path - register a sysctl table hierarchy
1871  * @path: The path to the directory the sysctl table is in.
1872  * @table: the top-level table structure
1873  *
1874  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1875  * array. A completely 0 filled entry terminates the table.
1876  *
1877  * See __register_sysctl_paths for more details.
1878  */
1879 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1880 						struct ctl_table *table)
1881 {
1882 	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1883 					path, table);
1884 }
1885 
1886 /**
1887  * register_sysctl_table - register a sysctl table hierarchy
1888  * @table: the top-level table structure
1889  *
1890  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1891  * array. A completely 0 filled entry terminates the table.
1892  *
1893  * See register_sysctl_paths for more details.
1894  */
1895 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1896 {
1897 	static const struct ctl_path null_path[] = { {} };
1898 
1899 	return register_sysctl_paths(null_path, table);
1900 }
1901 
1902 /**
1903  * unregister_sysctl_table - unregister a sysctl table hierarchy
1904  * @header: the header returned from register_sysctl_table
1905  *
1906  * Unregisters the sysctl table and all children. proc entries may not
1907  * actually be removed until they are no longer used by anyone.
1908  */
1909 void unregister_sysctl_table(struct ctl_table_header * header)
1910 {
1911 	might_sleep();
1912 
1913 	if (header == NULL)
1914 		return;
1915 
1916 	spin_lock(&sysctl_lock);
1917 	start_unregistering(header);
1918 	if (!--header->parent->count) {
1919 		WARN_ON(1);
1920 		kfree(header->parent);
1921 	}
1922 	if (!--header->count)
1923 		kfree(header);
1924 	spin_unlock(&sysctl_lock);
1925 }
1926 
1927 int sysctl_is_seen(struct ctl_table_header *p)
1928 {
1929 	struct ctl_table_set *set = p->set;
1930 	int res;
1931 	spin_lock(&sysctl_lock);
1932 	if (p->unregistering)
1933 		res = 0;
1934 	else if (!set->is_seen)
1935 		res = 1;
1936 	else
1937 		res = set->is_seen(set);
1938 	spin_unlock(&sysctl_lock);
1939 	return res;
1940 }
1941 
1942 void setup_sysctl_set(struct ctl_table_set *p,
1943 	struct ctl_table_set *parent,
1944 	int (*is_seen)(struct ctl_table_set *))
1945 {
1946 	INIT_LIST_HEAD(&p->list);
1947 	p->parent = parent ? parent : &sysctl_table_root.default_set;
1948 	p->is_seen = is_seen;
1949 }
1950 
1951 #else /* !CONFIG_SYSCTL */
1952 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1953 {
1954 	return NULL;
1955 }
1956 
1957 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1958 						    struct ctl_table *table)
1959 {
1960 	return NULL;
1961 }
1962 
1963 void unregister_sysctl_table(struct ctl_table_header * table)
1964 {
1965 }
1966 
1967 void setup_sysctl_set(struct ctl_table_set *p,
1968 	struct ctl_table_set *parent,
1969 	int (*is_seen)(struct ctl_table_set *))
1970 {
1971 }
1972 
1973 void sysctl_head_put(struct ctl_table_header *head)
1974 {
1975 }
1976 
1977 #endif /* CONFIG_SYSCTL */
1978 
1979 /*
1980  * /proc/sys support
1981  */
1982 
1983 #ifdef CONFIG_PROC_SYSCTL
1984 
1985 static int _proc_do_string(void* data, int maxlen, int write,
1986 			   struct file *filp, void __user *buffer,
1987 			   size_t *lenp, loff_t *ppos)
1988 {
1989 	size_t len;
1990 	char __user *p;
1991 	char c;
1992 
1993 	if (!data || !maxlen || !*lenp) {
1994 		*lenp = 0;
1995 		return 0;
1996 	}
1997 
1998 	if (write) {
1999 		len = 0;
2000 		p = buffer;
2001 		while (len < *lenp) {
2002 			if (get_user(c, p++))
2003 				return -EFAULT;
2004 			if (c == 0 || c == '\n')
2005 				break;
2006 			len++;
2007 		}
2008 		if (len >= maxlen)
2009 			len = maxlen-1;
2010 		if(copy_from_user(data, buffer, len))
2011 			return -EFAULT;
2012 		((char *) data)[len] = 0;
2013 		*ppos += *lenp;
2014 	} else {
2015 		len = strlen(data);
2016 		if (len > maxlen)
2017 			len = maxlen;
2018 
2019 		if (*ppos > len) {
2020 			*lenp = 0;
2021 			return 0;
2022 		}
2023 
2024 		data += *ppos;
2025 		len  -= *ppos;
2026 
2027 		if (len > *lenp)
2028 			len = *lenp;
2029 		if (len)
2030 			if(copy_to_user(buffer, data, len))
2031 				return -EFAULT;
2032 		if (len < *lenp) {
2033 			if(put_user('\n', ((char __user *) buffer) + len))
2034 				return -EFAULT;
2035 			len++;
2036 		}
2037 		*lenp = len;
2038 		*ppos += len;
2039 	}
2040 	return 0;
2041 }
2042 
2043 /**
2044  * proc_dostring - read a string sysctl
2045  * @table: the sysctl table
2046  * @write: %TRUE if this is a write to the sysctl file
2047  * @filp: the file structure
2048  * @buffer: the user buffer
2049  * @lenp: the size of the user buffer
2050  * @ppos: file position
2051  *
2052  * Reads/writes a string from/to the user buffer. If the kernel
2053  * buffer provided is not large enough to hold the string, the
2054  * string is truncated. The copied string is %NULL-terminated.
2055  * If the string is being read by the user process, it is copied
2056  * and a newline '\n' is added. It is truncated if the buffer is
2057  * not large enough.
2058  *
2059  * Returns 0 on success.
2060  */
2061 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2062 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2063 {
2064 	return _proc_do_string(table->data, table->maxlen, write, filp,
2065 			       buffer, lenp, ppos);
2066 }
2067 
2068 
2069 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2070 				 int *valp,
2071 				 int write, void *data)
2072 {
2073 	if (write) {
2074 		*valp = *negp ? -*lvalp : *lvalp;
2075 	} else {
2076 		int val = *valp;
2077 		if (val < 0) {
2078 			*negp = -1;
2079 			*lvalp = (unsigned long)-val;
2080 		} else {
2081 			*negp = 0;
2082 			*lvalp = (unsigned long)val;
2083 		}
2084 	}
2085 	return 0;
2086 }
2087 
2088 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2089 		  int write, struct file *filp, void __user *buffer,
2090 		  size_t *lenp, loff_t *ppos,
2091 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2092 			      int write, void *data),
2093 		  void *data)
2094 {
2095 #define TMPBUFLEN 21
2096 	int *i, vleft, first=1, neg, val;
2097 	unsigned long lval;
2098 	size_t left, len;
2099 
2100 	char buf[TMPBUFLEN], *p;
2101 	char __user *s = buffer;
2102 
2103 	if (!tbl_data || !table->maxlen || !*lenp ||
2104 	    (*ppos && !write)) {
2105 		*lenp = 0;
2106 		return 0;
2107 	}
2108 
2109 	i = (int *) tbl_data;
2110 	vleft = table->maxlen / sizeof(*i);
2111 	left = *lenp;
2112 
2113 	if (!conv)
2114 		conv = do_proc_dointvec_conv;
2115 
2116 	for (; left && vleft--; i++, first=0) {
2117 		if (write) {
2118 			while (left) {
2119 				char c;
2120 				if (get_user(c, s))
2121 					return -EFAULT;
2122 				if (!isspace(c))
2123 					break;
2124 				left--;
2125 				s++;
2126 			}
2127 			if (!left)
2128 				break;
2129 			neg = 0;
2130 			len = left;
2131 			if (len > sizeof(buf) - 1)
2132 				len = sizeof(buf) - 1;
2133 			if (copy_from_user(buf, s, len))
2134 				return -EFAULT;
2135 			buf[len] = 0;
2136 			p = buf;
2137 			if (*p == '-' && left > 1) {
2138 				neg = 1;
2139 				p++;
2140 			}
2141 			if (*p < '0' || *p > '9')
2142 				break;
2143 
2144 			lval = simple_strtoul(p, &p, 0);
2145 
2146 			len = p-buf;
2147 			if ((len < left) && *p && !isspace(*p))
2148 				break;
2149 			if (neg)
2150 				val = -val;
2151 			s += len;
2152 			left -= len;
2153 
2154 			if (conv(&neg, &lval, i, 1, data))
2155 				break;
2156 		} else {
2157 			p = buf;
2158 			if (!first)
2159 				*p++ = '\t';
2160 
2161 			if (conv(&neg, &lval, i, 0, data))
2162 				break;
2163 
2164 			sprintf(p, "%s%lu", neg ? "-" : "", lval);
2165 			len = strlen(buf);
2166 			if (len > left)
2167 				len = left;
2168 			if(copy_to_user(s, buf, len))
2169 				return -EFAULT;
2170 			left -= len;
2171 			s += len;
2172 		}
2173 	}
2174 
2175 	if (!write && !first && left) {
2176 		if(put_user('\n', s))
2177 			return -EFAULT;
2178 		left--, s++;
2179 	}
2180 	if (write) {
2181 		while (left) {
2182 			char c;
2183 			if (get_user(c, s++))
2184 				return -EFAULT;
2185 			if (!isspace(c))
2186 				break;
2187 			left--;
2188 		}
2189 	}
2190 	if (write && first)
2191 		return -EINVAL;
2192 	*lenp -= left;
2193 	*ppos += *lenp;
2194 	return 0;
2195 #undef TMPBUFLEN
2196 }
2197 
2198 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2199 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2200 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2201 			      int write, void *data),
2202 		  void *data)
2203 {
2204 	return __do_proc_dointvec(table->data, table, write, filp,
2205 			buffer, lenp, ppos, conv, data);
2206 }
2207 
2208 /**
2209  * proc_dointvec - read a vector of integers
2210  * @table: the sysctl table
2211  * @write: %TRUE if this is a write to the sysctl file
2212  * @filp: the file structure
2213  * @buffer: the user buffer
2214  * @lenp: the size of the user buffer
2215  * @ppos: file position
2216  *
2217  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2218  * values from/to the user buffer, treated as an ASCII string.
2219  *
2220  * Returns 0 on success.
2221  */
2222 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2223 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2224 {
2225     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2226 		    	    NULL,NULL);
2227 }
2228 
2229 /*
2230  * Taint values can only be increased
2231  * This means we can safely use a temporary.
2232  */
2233 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
2234 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2235 {
2236 	struct ctl_table t;
2237 	unsigned long tmptaint = get_taint();
2238 	int err;
2239 
2240 	if (write && !capable(CAP_SYS_ADMIN))
2241 		return -EPERM;
2242 
2243 	t = *table;
2244 	t.data = &tmptaint;
2245 	err = proc_doulongvec_minmax(&t, write, filp, buffer, lenp, ppos);
2246 	if (err < 0)
2247 		return err;
2248 
2249 	if (write) {
2250 		/*
2251 		 * Poor man's atomic or. Not worth adding a primitive
2252 		 * to everyone's atomic.h for this
2253 		 */
2254 		int i;
2255 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2256 			if ((tmptaint >> i) & 1)
2257 				add_taint(i);
2258 		}
2259 	}
2260 
2261 	return err;
2262 }
2263 
2264 struct do_proc_dointvec_minmax_conv_param {
2265 	int *min;
2266 	int *max;
2267 };
2268 
2269 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2270 					int *valp,
2271 					int write, void *data)
2272 {
2273 	struct do_proc_dointvec_minmax_conv_param *param = data;
2274 	if (write) {
2275 		int val = *negp ? -*lvalp : *lvalp;
2276 		if ((param->min && *param->min > val) ||
2277 		    (param->max && *param->max < val))
2278 			return -EINVAL;
2279 		*valp = val;
2280 	} else {
2281 		int val = *valp;
2282 		if (val < 0) {
2283 			*negp = -1;
2284 			*lvalp = (unsigned long)-val;
2285 		} else {
2286 			*negp = 0;
2287 			*lvalp = (unsigned long)val;
2288 		}
2289 	}
2290 	return 0;
2291 }
2292 
2293 /**
2294  * proc_dointvec_minmax - read a vector of integers with min/max values
2295  * @table: the sysctl table
2296  * @write: %TRUE if this is a write to the sysctl file
2297  * @filp: the file structure
2298  * @buffer: the user buffer
2299  * @lenp: the size of the user buffer
2300  * @ppos: file position
2301  *
2302  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2303  * values from/to the user buffer, treated as an ASCII string.
2304  *
2305  * This routine will ensure the values are within the range specified by
2306  * table->extra1 (min) and table->extra2 (max).
2307  *
2308  * Returns 0 on success.
2309  */
2310 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2311 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2312 {
2313 	struct do_proc_dointvec_minmax_conv_param param = {
2314 		.min = (int *) table->extra1,
2315 		.max = (int *) table->extra2,
2316 	};
2317 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2318 				do_proc_dointvec_minmax_conv, &param);
2319 }
2320 
2321 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2322 				     struct file *filp,
2323 				     void __user *buffer,
2324 				     size_t *lenp, loff_t *ppos,
2325 				     unsigned long convmul,
2326 				     unsigned long convdiv)
2327 {
2328 #define TMPBUFLEN 21
2329 	unsigned long *i, *min, *max, val;
2330 	int vleft, first=1, neg;
2331 	size_t len, left;
2332 	char buf[TMPBUFLEN], *p;
2333 	char __user *s = buffer;
2334 
2335 	if (!data || !table->maxlen || !*lenp ||
2336 	    (*ppos && !write)) {
2337 		*lenp = 0;
2338 		return 0;
2339 	}
2340 
2341 	i = (unsigned long *) data;
2342 	min = (unsigned long *) table->extra1;
2343 	max = (unsigned long *) table->extra2;
2344 	vleft = table->maxlen / sizeof(unsigned long);
2345 	left = *lenp;
2346 
2347 	for (; left && vleft--; i++, min++, max++, first=0) {
2348 		if (write) {
2349 			while (left) {
2350 				char c;
2351 				if (get_user(c, s))
2352 					return -EFAULT;
2353 				if (!isspace(c))
2354 					break;
2355 				left--;
2356 				s++;
2357 			}
2358 			if (!left)
2359 				break;
2360 			neg = 0;
2361 			len = left;
2362 			if (len > TMPBUFLEN-1)
2363 				len = TMPBUFLEN-1;
2364 			if (copy_from_user(buf, s, len))
2365 				return -EFAULT;
2366 			buf[len] = 0;
2367 			p = buf;
2368 			if (*p == '-' && left > 1) {
2369 				neg = 1;
2370 				p++;
2371 			}
2372 			if (*p < '0' || *p > '9')
2373 				break;
2374 			val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2375 			len = p-buf;
2376 			if ((len < left) && *p && !isspace(*p))
2377 				break;
2378 			if (neg)
2379 				val = -val;
2380 			s += len;
2381 			left -= len;
2382 
2383 			if(neg)
2384 				continue;
2385 			if ((min && val < *min) || (max && val > *max))
2386 				continue;
2387 			*i = val;
2388 		} else {
2389 			p = buf;
2390 			if (!first)
2391 				*p++ = '\t';
2392 			sprintf(p, "%lu", convdiv * (*i) / convmul);
2393 			len = strlen(buf);
2394 			if (len > left)
2395 				len = left;
2396 			if(copy_to_user(s, buf, len))
2397 				return -EFAULT;
2398 			left -= len;
2399 			s += len;
2400 		}
2401 	}
2402 
2403 	if (!write && !first && left) {
2404 		if(put_user('\n', s))
2405 			return -EFAULT;
2406 		left--, s++;
2407 	}
2408 	if (write) {
2409 		while (left) {
2410 			char c;
2411 			if (get_user(c, s++))
2412 				return -EFAULT;
2413 			if (!isspace(c))
2414 				break;
2415 			left--;
2416 		}
2417 	}
2418 	if (write && first)
2419 		return -EINVAL;
2420 	*lenp -= left;
2421 	*ppos += *lenp;
2422 	return 0;
2423 #undef TMPBUFLEN
2424 }
2425 
2426 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2427 				     struct file *filp,
2428 				     void __user *buffer,
2429 				     size_t *lenp, loff_t *ppos,
2430 				     unsigned long convmul,
2431 				     unsigned long convdiv)
2432 {
2433 	return __do_proc_doulongvec_minmax(table->data, table, write,
2434 			filp, buffer, lenp, ppos, convmul, convdiv);
2435 }
2436 
2437 /**
2438  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2439  * @table: the sysctl table
2440  * @write: %TRUE if this is a write to the sysctl file
2441  * @filp: the file structure
2442  * @buffer: the user buffer
2443  * @lenp: the size of the user buffer
2444  * @ppos: file position
2445  *
2446  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2447  * values from/to the user buffer, treated as an ASCII string.
2448  *
2449  * This routine will ensure the values are within the range specified by
2450  * table->extra1 (min) and table->extra2 (max).
2451  *
2452  * Returns 0 on success.
2453  */
2454 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2455 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2456 {
2457     return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2458 }
2459 
2460 /**
2461  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2462  * @table: the sysctl table
2463  * @write: %TRUE if this is a write to the sysctl file
2464  * @filp: the file structure
2465  * @buffer: the user buffer
2466  * @lenp: the size of the user buffer
2467  * @ppos: file position
2468  *
2469  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2470  * values from/to the user buffer, treated as an ASCII string. The values
2471  * are treated as milliseconds, and converted to jiffies when they are stored.
2472  *
2473  * This routine will ensure the values are within the range specified by
2474  * table->extra1 (min) and table->extra2 (max).
2475  *
2476  * Returns 0 on success.
2477  */
2478 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2479 				      struct file *filp,
2480 				      void __user *buffer,
2481 				      size_t *lenp, loff_t *ppos)
2482 {
2483     return do_proc_doulongvec_minmax(table, write, filp, buffer,
2484 				     lenp, ppos, HZ, 1000l);
2485 }
2486 
2487 
2488 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2489 					 int *valp,
2490 					 int write, void *data)
2491 {
2492 	if (write) {
2493 		if (*lvalp > LONG_MAX / HZ)
2494 			return 1;
2495 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2496 	} else {
2497 		int val = *valp;
2498 		unsigned long lval;
2499 		if (val < 0) {
2500 			*negp = -1;
2501 			lval = (unsigned long)-val;
2502 		} else {
2503 			*negp = 0;
2504 			lval = (unsigned long)val;
2505 		}
2506 		*lvalp = lval / HZ;
2507 	}
2508 	return 0;
2509 }
2510 
2511 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2512 						int *valp,
2513 						int write, void *data)
2514 {
2515 	if (write) {
2516 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2517 			return 1;
2518 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2519 	} else {
2520 		int val = *valp;
2521 		unsigned long lval;
2522 		if (val < 0) {
2523 			*negp = -1;
2524 			lval = (unsigned long)-val;
2525 		} else {
2526 			*negp = 0;
2527 			lval = (unsigned long)val;
2528 		}
2529 		*lvalp = jiffies_to_clock_t(lval);
2530 	}
2531 	return 0;
2532 }
2533 
2534 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2535 					    int *valp,
2536 					    int write, void *data)
2537 {
2538 	if (write) {
2539 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2540 	} else {
2541 		int val = *valp;
2542 		unsigned long lval;
2543 		if (val < 0) {
2544 			*negp = -1;
2545 			lval = (unsigned long)-val;
2546 		} else {
2547 			*negp = 0;
2548 			lval = (unsigned long)val;
2549 		}
2550 		*lvalp = jiffies_to_msecs(lval);
2551 	}
2552 	return 0;
2553 }
2554 
2555 /**
2556  * proc_dointvec_jiffies - read a vector of integers as seconds
2557  * @table: the sysctl table
2558  * @write: %TRUE if this is a write to the sysctl file
2559  * @filp: the file structure
2560  * @buffer: the user buffer
2561  * @lenp: the size of the user buffer
2562  * @ppos: file position
2563  *
2564  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2565  * values from/to the user buffer, treated as an ASCII string.
2566  * The values read are assumed to be in seconds, and are converted into
2567  * jiffies.
2568  *
2569  * Returns 0 on success.
2570  */
2571 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2572 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2573 {
2574     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2575 		    	    do_proc_dointvec_jiffies_conv,NULL);
2576 }
2577 
2578 /**
2579  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2580  * @table: the sysctl table
2581  * @write: %TRUE if this is a write to the sysctl file
2582  * @filp: the file structure
2583  * @buffer: the user buffer
2584  * @lenp: the size of the user buffer
2585  * @ppos: pointer to the file position
2586  *
2587  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2588  * values from/to the user buffer, treated as an ASCII string.
2589  * The values read are assumed to be in 1/USER_HZ seconds, and
2590  * are converted into jiffies.
2591  *
2592  * Returns 0 on success.
2593  */
2594 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2595 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2596 {
2597     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2598 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2599 }
2600 
2601 /**
2602  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2603  * @table: the sysctl table
2604  * @write: %TRUE if this is a write to the sysctl file
2605  * @filp: the file structure
2606  * @buffer: the user buffer
2607  * @lenp: the size of the user buffer
2608  * @ppos: file position
2609  * @ppos: the current position in the file
2610  *
2611  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2612  * values from/to the user buffer, treated as an ASCII string.
2613  * The values read are assumed to be in 1/1000 seconds, and
2614  * are converted into jiffies.
2615  *
2616  * Returns 0 on success.
2617  */
2618 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2619 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2620 {
2621 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2622 				do_proc_dointvec_ms_jiffies_conv, NULL);
2623 }
2624 
2625 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2626 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2627 {
2628 	struct pid *new_pid;
2629 	pid_t tmp;
2630 	int r;
2631 
2632 	tmp = pid_vnr(cad_pid);
2633 
2634 	r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2635 			       lenp, ppos, NULL, NULL);
2636 	if (r || !write)
2637 		return r;
2638 
2639 	new_pid = find_get_pid(tmp);
2640 	if (!new_pid)
2641 		return -ESRCH;
2642 
2643 	put_pid(xchg(&cad_pid, new_pid));
2644 	return 0;
2645 }
2646 
2647 #else /* CONFIG_PROC_FS */
2648 
2649 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2650 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2651 {
2652 	return -ENOSYS;
2653 }
2654 
2655 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2656 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2657 {
2658 	return -ENOSYS;
2659 }
2660 
2661 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2662 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2663 {
2664 	return -ENOSYS;
2665 }
2666 
2667 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2668 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2669 {
2670 	return -ENOSYS;
2671 }
2672 
2673 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2674 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2675 {
2676 	return -ENOSYS;
2677 }
2678 
2679 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2680 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2681 {
2682 	return -ENOSYS;
2683 }
2684 
2685 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2686 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2687 {
2688 	return -ENOSYS;
2689 }
2690 
2691 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2692 				      struct file *filp,
2693 				      void __user *buffer,
2694 				      size_t *lenp, loff_t *ppos)
2695 {
2696     return -ENOSYS;
2697 }
2698 
2699 
2700 #endif /* CONFIG_PROC_FS */
2701 
2702 
2703 #ifdef CONFIG_SYSCTL_SYSCALL
2704 /*
2705  * General sysctl support routines
2706  */
2707 
2708 /* The generic sysctl data routine (used if no strategy routine supplied) */
2709 int sysctl_data(struct ctl_table *table,
2710 		void __user *oldval, size_t __user *oldlenp,
2711 		void __user *newval, size_t newlen)
2712 {
2713 	size_t len;
2714 
2715 	/* Get out of I don't have a variable */
2716 	if (!table->data || !table->maxlen)
2717 		return -ENOTDIR;
2718 
2719 	if (oldval && oldlenp) {
2720 		if (get_user(len, oldlenp))
2721 			return -EFAULT;
2722 		if (len) {
2723 			if (len > table->maxlen)
2724 				len = table->maxlen;
2725 			if (copy_to_user(oldval, table->data, len))
2726 				return -EFAULT;
2727 			if (put_user(len, oldlenp))
2728 				return -EFAULT;
2729 		}
2730 	}
2731 
2732 	if (newval && newlen) {
2733 		if (newlen > table->maxlen)
2734 			newlen = table->maxlen;
2735 
2736 		if (copy_from_user(table->data, newval, newlen))
2737 			return -EFAULT;
2738 	}
2739 	return 1;
2740 }
2741 
2742 /* The generic string strategy routine: */
2743 int sysctl_string(struct ctl_table *table,
2744 		  void __user *oldval, size_t __user *oldlenp,
2745 		  void __user *newval, size_t newlen)
2746 {
2747 	if (!table->data || !table->maxlen)
2748 		return -ENOTDIR;
2749 
2750 	if (oldval && oldlenp) {
2751 		size_t bufsize;
2752 		if (get_user(bufsize, oldlenp))
2753 			return -EFAULT;
2754 		if (bufsize) {
2755 			size_t len = strlen(table->data), copied;
2756 
2757 			/* This shouldn't trigger for a well-formed sysctl */
2758 			if (len > table->maxlen)
2759 				len = table->maxlen;
2760 
2761 			/* Copy up to a max of bufsize-1 bytes of the string */
2762 			copied = (len >= bufsize) ? bufsize - 1 : len;
2763 
2764 			if (copy_to_user(oldval, table->data, copied) ||
2765 			    put_user(0, (char __user *)(oldval + copied)))
2766 				return -EFAULT;
2767 			if (put_user(len, oldlenp))
2768 				return -EFAULT;
2769 		}
2770 	}
2771 	if (newval && newlen) {
2772 		size_t len = newlen;
2773 		if (len > table->maxlen)
2774 			len = table->maxlen;
2775 		if(copy_from_user(table->data, newval, len))
2776 			return -EFAULT;
2777 		if (len == table->maxlen)
2778 			len--;
2779 		((char *) table->data)[len] = 0;
2780 	}
2781 	return 1;
2782 }
2783 
2784 /*
2785  * This function makes sure that all of the integers in the vector
2786  * are between the minimum and maximum values given in the arrays
2787  * table->extra1 and table->extra2, respectively.
2788  */
2789 int sysctl_intvec(struct ctl_table *table,
2790 		void __user *oldval, size_t __user *oldlenp,
2791 		void __user *newval, size_t newlen)
2792 {
2793 
2794 	if (newval && newlen) {
2795 		int __user *vec = (int __user *) newval;
2796 		int *min = (int *) table->extra1;
2797 		int *max = (int *) table->extra2;
2798 		size_t length;
2799 		int i;
2800 
2801 		if (newlen % sizeof(int) != 0)
2802 			return -EINVAL;
2803 
2804 		if (!table->extra1 && !table->extra2)
2805 			return 0;
2806 
2807 		if (newlen > table->maxlen)
2808 			newlen = table->maxlen;
2809 		length = newlen / sizeof(int);
2810 
2811 		for (i = 0; i < length; i++) {
2812 			int value;
2813 			if (get_user(value, vec + i))
2814 				return -EFAULT;
2815 			if (min && value < min[i])
2816 				return -EINVAL;
2817 			if (max && value > max[i])
2818 				return -EINVAL;
2819 		}
2820 	}
2821 	return 0;
2822 }
2823 
2824 /* Strategy function to convert jiffies to seconds */
2825 int sysctl_jiffies(struct ctl_table *table,
2826 		void __user *oldval, size_t __user *oldlenp,
2827 		void __user *newval, size_t newlen)
2828 {
2829 	if (oldval && oldlenp) {
2830 		size_t olen;
2831 
2832 		if (get_user(olen, oldlenp))
2833 			return -EFAULT;
2834 		if (olen) {
2835 			int val;
2836 
2837 			if (olen < sizeof(int))
2838 				return -EINVAL;
2839 
2840 			val = *(int *)(table->data) / HZ;
2841 			if (put_user(val, (int __user *)oldval))
2842 				return -EFAULT;
2843 			if (put_user(sizeof(int), oldlenp))
2844 				return -EFAULT;
2845 		}
2846 	}
2847 	if (newval && newlen) {
2848 		int new;
2849 		if (newlen != sizeof(int))
2850 			return -EINVAL;
2851 		if (get_user(new, (int __user *)newval))
2852 			return -EFAULT;
2853 		*(int *)(table->data) = new*HZ;
2854 	}
2855 	return 1;
2856 }
2857 
2858 /* Strategy function to convert jiffies to seconds */
2859 int sysctl_ms_jiffies(struct ctl_table *table,
2860 		void __user *oldval, size_t __user *oldlenp,
2861 		void __user *newval, size_t newlen)
2862 {
2863 	if (oldval && oldlenp) {
2864 		size_t olen;
2865 
2866 		if (get_user(olen, oldlenp))
2867 			return -EFAULT;
2868 		if (olen) {
2869 			int val;
2870 
2871 			if (olen < sizeof(int))
2872 				return -EINVAL;
2873 
2874 			val = jiffies_to_msecs(*(int *)(table->data));
2875 			if (put_user(val, (int __user *)oldval))
2876 				return -EFAULT;
2877 			if (put_user(sizeof(int), oldlenp))
2878 				return -EFAULT;
2879 		}
2880 	}
2881 	if (newval && newlen) {
2882 		int new;
2883 		if (newlen != sizeof(int))
2884 			return -EINVAL;
2885 		if (get_user(new, (int __user *)newval))
2886 			return -EFAULT;
2887 		*(int *)(table->data) = msecs_to_jiffies(new);
2888 	}
2889 	return 1;
2890 }
2891 
2892 
2893 
2894 #else /* CONFIG_SYSCTL_SYSCALL */
2895 
2896 
2897 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2898 {
2899 	struct __sysctl_args tmp;
2900 	int error;
2901 
2902 	if (copy_from_user(&tmp, args, sizeof(tmp)))
2903 		return -EFAULT;
2904 
2905 	error = deprecated_sysctl_warning(&tmp);
2906 
2907 	/* If no error reading the parameters then just -ENOSYS ... */
2908 	if (!error)
2909 		error = -ENOSYS;
2910 
2911 	return error;
2912 }
2913 
2914 int sysctl_data(struct ctl_table *table,
2915 		  void __user *oldval, size_t __user *oldlenp,
2916 		  void __user *newval, size_t newlen)
2917 {
2918 	return -ENOSYS;
2919 }
2920 
2921 int sysctl_string(struct ctl_table *table,
2922 		  void __user *oldval, size_t __user *oldlenp,
2923 		  void __user *newval, size_t newlen)
2924 {
2925 	return -ENOSYS;
2926 }
2927 
2928 int sysctl_intvec(struct ctl_table *table,
2929 		void __user *oldval, size_t __user *oldlenp,
2930 		void __user *newval, size_t newlen)
2931 {
2932 	return -ENOSYS;
2933 }
2934 
2935 int sysctl_jiffies(struct ctl_table *table,
2936 		void __user *oldval, size_t __user *oldlenp,
2937 		void __user *newval, size_t newlen)
2938 {
2939 	return -ENOSYS;
2940 }
2941 
2942 int sysctl_ms_jiffies(struct ctl_table *table,
2943 		void __user *oldval, size_t __user *oldlenp,
2944 		void __user *newval, size_t newlen)
2945 {
2946 	return -ENOSYS;
2947 }
2948 
2949 #endif /* CONFIG_SYSCTL_SYSCALL */
2950 
2951 static int deprecated_sysctl_warning(struct __sysctl_args *args)
2952 {
2953 	static int msg_count;
2954 	int name[CTL_MAXNAME];
2955 	int i;
2956 
2957 	/* Check args->nlen. */
2958 	if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
2959 		return -ENOTDIR;
2960 
2961 	/* Read in the sysctl name for better debug message logging */
2962 	for (i = 0; i < args->nlen; i++)
2963 		if (get_user(name[i], args->name + i))
2964 			return -EFAULT;
2965 
2966 	/* Ignore accesses to kernel.version */
2967 	if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2968 		return 0;
2969 
2970 	if (msg_count < 5) {
2971 		msg_count++;
2972 		printk(KERN_INFO
2973 			"warning: process `%s' used the deprecated sysctl "
2974 			"system call with ", current->comm);
2975 		for (i = 0; i < args->nlen; i++)
2976 			printk("%d.", name[i]);
2977 		printk("\n");
2978 	}
2979 	return 0;
2980 }
2981 
2982 /*
2983  * No sense putting this after each symbol definition, twice,
2984  * exception granted :-)
2985  */
2986 EXPORT_SYMBOL(proc_dointvec);
2987 EXPORT_SYMBOL(proc_dointvec_jiffies);
2988 EXPORT_SYMBOL(proc_dointvec_minmax);
2989 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2990 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2991 EXPORT_SYMBOL(proc_dostring);
2992 EXPORT_SYMBOL(proc_doulongvec_minmax);
2993 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2994 EXPORT_SYMBOL(register_sysctl_table);
2995 EXPORT_SYMBOL(register_sysctl_paths);
2996 EXPORT_SYMBOL(sysctl_intvec);
2997 EXPORT_SYMBOL(sysctl_jiffies);
2998 EXPORT_SYMBOL(sysctl_ms_jiffies);
2999 EXPORT_SYMBOL(sysctl_string);
3000 EXPORT_SYMBOL(sysctl_data);
3001 EXPORT_SYMBOL(unregister_sysctl_table);
3002