xref: /linux-6.15/kernel/sysctl.c (revision 7ec7fb39)
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 #ifdef CONFIG_DETECT_SOFTLOCKUP
91 static int sixty = 60;
92 static int neg_one = -1;
93 #endif
94 
95 #if defined(CONFIG_MMU) && defined(CONFIG_FILE_LOCKING)
96 static int two = 2;
97 #endif
98 
99 static int zero;
100 static int one = 1;
101 static int one_hundred = 100;
102 
103 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
104 static int maxolduid = 65535;
105 static int minolduid;
106 static int min_percpu_pagelist_fract = 8;
107 
108 static int ngroups_max = NGROUPS_MAX;
109 
110 #ifdef CONFIG_MODULES
111 extern char modprobe_path[];
112 #endif
113 #ifdef CONFIG_CHR_DEV_SG
114 extern int sg_big_buff;
115 #endif
116 
117 #ifdef CONFIG_SPARC
118 #include <asm/system.h>
119 #endif
120 
121 #ifdef CONFIG_SPARC64
122 extern int sysctl_tsb_ratio;
123 #endif
124 
125 #ifdef __hppa__
126 extern int pwrsw_enabled;
127 extern int unaligned_enabled;
128 #endif
129 
130 #ifdef CONFIG_S390
131 #ifdef CONFIG_MATHEMU
132 extern int sysctl_ieee_emulation_warnings;
133 #endif
134 extern int sysctl_userprocess_debug;
135 extern int spin_retry;
136 #endif
137 
138 #ifdef CONFIG_BSD_PROCESS_ACCT
139 extern int acct_parm[];
140 #endif
141 
142 #ifdef CONFIG_IA64
143 extern int no_unaligned_warning;
144 #endif
145 
146 #ifdef CONFIG_RT_MUTEXES
147 extern int max_lock_depth;
148 #endif
149 
150 #ifdef CONFIG_PROC_SYSCTL
151 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
152 		  void __user *buffer, size_t *lenp, loff_t *ppos);
153 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
154 			       void __user *buffer, size_t *lenp, loff_t *ppos);
155 #endif
156 
157 static struct ctl_table root_table[];
158 static struct ctl_table_root sysctl_table_root;
159 static struct ctl_table_header root_table_header = {
160 	.count = 1,
161 	.ctl_table = root_table,
162 	.ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
163 	.root = &sysctl_table_root,
164 	.set = &sysctl_table_root.default_set,
165 };
166 static struct ctl_table_root sysctl_table_root = {
167 	.root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
168 	.default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
169 };
170 
171 static struct ctl_table kern_table[];
172 static struct ctl_table vm_table[];
173 static struct ctl_table fs_table[];
174 static struct ctl_table debug_table[];
175 static struct ctl_table dev_table[];
176 extern struct ctl_table random_table[];
177 #ifdef CONFIG_INOTIFY_USER
178 extern struct ctl_table inotify_table[];
179 #endif
180 #ifdef CONFIG_EPOLL
181 extern struct ctl_table epoll_table[];
182 #endif
183 
184 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
185 int sysctl_legacy_va_layout;
186 #endif
187 
188 extern int prove_locking;
189 extern int lock_stat;
190 
191 /* The default sysctl tables: */
192 
193 static struct ctl_table root_table[] = {
194 	{
195 		.ctl_name	= CTL_KERN,
196 		.procname	= "kernel",
197 		.mode		= 0555,
198 		.child		= kern_table,
199 	},
200 	{
201 		.ctl_name	= CTL_VM,
202 		.procname	= "vm",
203 		.mode		= 0555,
204 		.child		= vm_table,
205 	},
206 	{
207 		.ctl_name	= CTL_FS,
208 		.procname	= "fs",
209 		.mode		= 0555,
210 		.child		= fs_table,
211 	},
212 	{
213 		.ctl_name	= CTL_DEBUG,
214 		.procname	= "debug",
215 		.mode		= 0555,
216 		.child		= debug_table,
217 	},
218 	{
219 		.ctl_name	= CTL_DEV,
220 		.procname	= "dev",
221 		.mode		= 0555,
222 		.child		= dev_table,
223 	},
224 /*
225  * NOTE: do not add new entries to this table unless you have read
226  * Documentation/sysctl/ctl_unnumbered.txt
227  */
228 	{ .ctl_name = 0 }
229 };
230 
231 #ifdef CONFIG_SCHED_DEBUG
232 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
233 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
234 static int min_wakeup_granularity_ns;			/* 0 usecs */
235 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
236 #endif
237 
238 static struct ctl_table kern_table[] = {
239 #ifdef CONFIG_SCHED_DEBUG
240 	{
241 		.ctl_name	= CTL_UNNUMBERED,
242 		.procname	= "sched_min_granularity_ns",
243 		.data		= &sysctl_sched_min_granularity,
244 		.maxlen		= sizeof(unsigned int),
245 		.mode		= 0644,
246 		.proc_handler	= &sched_nr_latency_handler,
247 		.strategy	= &sysctl_intvec,
248 		.extra1		= &min_sched_granularity_ns,
249 		.extra2		= &max_sched_granularity_ns,
250 	},
251 	{
252 		.ctl_name	= CTL_UNNUMBERED,
253 		.procname	= "sched_latency_ns",
254 		.data		= &sysctl_sched_latency,
255 		.maxlen		= sizeof(unsigned int),
256 		.mode		= 0644,
257 		.proc_handler	= &sched_nr_latency_handler,
258 		.strategy	= &sysctl_intvec,
259 		.extra1		= &min_sched_granularity_ns,
260 		.extra2		= &max_sched_granularity_ns,
261 	},
262 	{
263 		.ctl_name	= CTL_UNNUMBERED,
264 		.procname	= "sched_wakeup_granularity_ns",
265 		.data		= &sysctl_sched_wakeup_granularity,
266 		.maxlen		= sizeof(unsigned int),
267 		.mode		= 0644,
268 		.proc_handler	= &proc_dointvec_minmax,
269 		.strategy	= &sysctl_intvec,
270 		.extra1		= &min_wakeup_granularity_ns,
271 		.extra2		= &max_wakeup_granularity_ns,
272 	},
273 	{
274 		.ctl_name	= CTL_UNNUMBERED,
275 		.procname	= "sched_shares_ratelimit",
276 		.data		= &sysctl_sched_shares_ratelimit,
277 		.maxlen		= sizeof(unsigned int),
278 		.mode		= 0644,
279 		.proc_handler	= &proc_dointvec,
280 	},
281 	{
282 		.ctl_name	= CTL_UNNUMBERED,
283 		.procname	= "sched_shares_thresh",
284 		.data		= &sysctl_sched_shares_thresh,
285 		.maxlen		= sizeof(unsigned int),
286 		.mode		= 0644,
287 		.proc_handler	= &proc_dointvec_minmax,
288 		.strategy	= &sysctl_intvec,
289 		.extra1		= &zero,
290 	},
291 	{
292 		.ctl_name	= CTL_UNNUMBERED,
293 		.procname	= "sched_child_runs_first",
294 		.data		= &sysctl_sched_child_runs_first,
295 		.maxlen		= sizeof(unsigned int),
296 		.mode		= 0644,
297 		.proc_handler	= &proc_dointvec,
298 	},
299 	{
300 		.ctl_name	= CTL_UNNUMBERED,
301 		.procname	= "sched_features",
302 		.data		= &sysctl_sched_features,
303 		.maxlen		= sizeof(unsigned int),
304 		.mode		= 0644,
305 		.proc_handler	= &proc_dointvec,
306 	},
307 	{
308 		.ctl_name	= CTL_UNNUMBERED,
309 		.procname	= "sched_migration_cost",
310 		.data		= &sysctl_sched_migration_cost,
311 		.maxlen		= sizeof(unsigned int),
312 		.mode		= 0644,
313 		.proc_handler	= &proc_dointvec,
314 	},
315 	{
316 		.ctl_name	= CTL_UNNUMBERED,
317 		.procname	= "sched_nr_migrate",
318 		.data		= &sysctl_sched_nr_migrate,
319 		.maxlen		= sizeof(unsigned int),
320 		.mode		= 0644,
321 		.proc_handler	= &proc_dointvec,
322 	},
323 #endif
324 	{
325 		.ctl_name	= CTL_UNNUMBERED,
326 		.procname	= "sched_rt_period_us",
327 		.data		= &sysctl_sched_rt_period,
328 		.maxlen		= sizeof(unsigned int),
329 		.mode		= 0644,
330 		.proc_handler	= &sched_rt_handler,
331 	},
332 	{
333 		.ctl_name	= CTL_UNNUMBERED,
334 		.procname	= "sched_rt_runtime_us",
335 		.data		= &sysctl_sched_rt_runtime,
336 		.maxlen		= sizeof(int),
337 		.mode		= 0644,
338 		.proc_handler	= &sched_rt_handler,
339 	},
340 	{
341 		.ctl_name	= CTL_UNNUMBERED,
342 		.procname	= "sched_compat_yield",
343 		.data		= &sysctl_sched_compat_yield,
344 		.maxlen		= sizeof(unsigned int),
345 		.mode		= 0644,
346 		.proc_handler	= &proc_dointvec,
347 	},
348 #ifdef CONFIG_PROVE_LOCKING
349 	{
350 		.ctl_name	= CTL_UNNUMBERED,
351 		.procname	= "prove_locking",
352 		.data		= &prove_locking,
353 		.maxlen		= sizeof(int),
354 		.mode		= 0644,
355 		.proc_handler	= &proc_dointvec,
356 	},
357 #endif
358 #ifdef CONFIG_LOCK_STAT
359 	{
360 		.ctl_name	= CTL_UNNUMBERED,
361 		.procname	= "lock_stat",
362 		.data		= &lock_stat,
363 		.maxlen		= sizeof(int),
364 		.mode		= 0644,
365 		.proc_handler	= &proc_dointvec,
366 	},
367 #endif
368 	{
369 		.ctl_name	= KERN_PANIC,
370 		.procname	= "panic",
371 		.data		= &panic_timeout,
372 		.maxlen		= sizeof(int),
373 		.mode		= 0644,
374 		.proc_handler	= &proc_dointvec,
375 	},
376 	{
377 		.ctl_name	= KERN_CORE_USES_PID,
378 		.procname	= "core_uses_pid",
379 		.data		= &core_uses_pid,
380 		.maxlen		= sizeof(int),
381 		.mode		= 0644,
382 		.proc_handler	= &proc_dointvec,
383 	},
384 	{
385 		.ctl_name	= KERN_CORE_PATTERN,
386 		.procname	= "core_pattern",
387 		.data		= core_pattern,
388 		.maxlen		= CORENAME_MAX_SIZE,
389 		.mode		= 0644,
390 		.proc_handler	= &proc_dostring,
391 		.strategy	= &sysctl_string,
392 	},
393 #ifdef CONFIG_PROC_SYSCTL
394 	{
395 		.procname	= "tainted",
396 		.maxlen 	= sizeof(long),
397 		.mode		= 0644,
398 		.proc_handler	= &proc_taint,
399 	},
400 #endif
401 #ifdef CONFIG_LATENCYTOP
402 	{
403 		.procname	= "latencytop",
404 		.data		= &latencytop_enabled,
405 		.maxlen		= sizeof(int),
406 		.mode		= 0644,
407 		.proc_handler	= &proc_dointvec,
408 	},
409 #endif
410 #ifdef CONFIG_BLK_DEV_INITRD
411 	{
412 		.ctl_name	= KERN_REALROOTDEV,
413 		.procname	= "real-root-dev",
414 		.data		= &real_root_dev,
415 		.maxlen		= sizeof(int),
416 		.mode		= 0644,
417 		.proc_handler	= &proc_dointvec,
418 	},
419 #endif
420 	{
421 		.ctl_name	= CTL_UNNUMBERED,
422 		.procname	= "print-fatal-signals",
423 		.data		= &print_fatal_signals,
424 		.maxlen		= sizeof(int),
425 		.mode		= 0644,
426 		.proc_handler	= &proc_dointvec,
427 	},
428 #ifdef CONFIG_SPARC
429 	{
430 		.ctl_name	= KERN_SPARC_REBOOT,
431 		.procname	= "reboot-cmd",
432 		.data		= reboot_command,
433 		.maxlen		= 256,
434 		.mode		= 0644,
435 		.proc_handler	= &proc_dostring,
436 		.strategy	= &sysctl_string,
437 	},
438 	{
439 		.ctl_name	= KERN_SPARC_STOP_A,
440 		.procname	= "stop-a",
441 		.data		= &stop_a_enabled,
442 		.maxlen		= sizeof (int),
443 		.mode		= 0644,
444 		.proc_handler	= &proc_dointvec,
445 	},
446 	{
447 		.ctl_name	= KERN_SPARC_SCONS_PWROFF,
448 		.procname	= "scons-poweroff",
449 		.data		= &scons_pwroff,
450 		.maxlen		= sizeof (int),
451 		.mode		= 0644,
452 		.proc_handler	= &proc_dointvec,
453 	},
454 #endif
455 #ifdef CONFIG_SPARC64
456 	{
457 		.ctl_name	= CTL_UNNUMBERED,
458 		.procname	= "tsb-ratio",
459 		.data		= &sysctl_tsb_ratio,
460 		.maxlen		= sizeof (int),
461 		.mode		= 0644,
462 		.proc_handler	= &proc_dointvec,
463 	},
464 #endif
465 #ifdef __hppa__
466 	{
467 		.ctl_name	= KERN_HPPA_PWRSW,
468 		.procname	= "soft-power",
469 		.data		= &pwrsw_enabled,
470 		.maxlen		= sizeof (int),
471 	 	.mode		= 0644,
472 		.proc_handler	= &proc_dointvec,
473 	},
474 	{
475 		.ctl_name	= KERN_HPPA_UNALIGNED,
476 		.procname	= "unaligned-trap",
477 		.data		= &unaligned_enabled,
478 		.maxlen		= sizeof (int),
479 		.mode		= 0644,
480 		.proc_handler	= &proc_dointvec,
481 	},
482 #endif
483 	{
484 		.ctl_name	= KERN_CTLALTDEL,
485 		.procname	= "ctrl-alt-del",
486 		.data		= &C_A_D,
487 		.maxlen		= sizeof(int),
488 		.mode		= 0644,
489 		.proc_handler	= &proc_dointvec,
490 	},
491 #ifdef CONFIG_FUNCTION_TRACER
492 	{
493 		.ctl_name	= CTL_UNNUMBERED,
494 		.procname	= "ftrace_enabled",
495 		.data		= &ftrace_enabled,
496 		.maxlen		= sizeof(int),
497 		.mode		= 0644,
498 		.proc_handler	= &ftrace_enable_sysctl,
499 	},
500 #endif
501 #ifdef CONFIG_STACK_TRACER
502 	{
503 		.ctl_name	= CTL_UNNUMBERED,
504 		.procname	= "stack_tracer_enabled",
505 		.data		= &stack_tracer_enabled,
506 		.maxlen		= sizeof(int),
507 		.mode		= 0644,
508 		.proc_handler	= &stack_trace_sysctl,
509 	},
510 #endif
511 #ifdef CONFIG_TRACING
512 	{
513 		.ctl_name	= CTL_UNNUMBERED,
514 		.procname	= "ftrace_dump_on_oops",
515 		.data		= &ftrace_dump_on_oops,
516 		.maxlen		= sizeof(int),
517 		.mode		= 0644,
518 		.proc_handler	= &proc_dointvec,
519 	},
520 #endif
521 #ifdef CONFIG_MODULES
522 	{
523 		.ctl_name	= KERN_MODPROBE,
524 		.procname	= "modprobe",
525 		.data		= &modprobe_path,
526 		.maxlen		= KMOD_PATH_LEN,
527 		.mode		= 0644,
528 		.proc_handler	= &proc_dostring,
529 		.strategy	= &sysctl_string,
530 	},
531 #endif
532 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
533 	{
534 		.ctl_name	= KERN_HOTPLUG,
535 		.procname	= "hotplug",
536 		.data		= &uevent_helper,
537 		.maxlen		= UEVENT_HELPER_PATH_LEN,
538 		.mode		= 0644,
539 		.proc_handler	= &proc_dostring,
540 		.strategy	= &sysctl_string,
541 	},
542 #endif
543 #ifdef CONFIG_CHR_DEV_SG
544 	{
545 		.ctl_name	= KERN_SG_BIG_BUFF,
546 		.procname	= "sg-big-buff",
547 		.data		= &sg_big_buff,
548 		.maxlen		= sizeof (int),
549 		.mode		= 0444,
550 		.proc_handler	= &proc_dointvec,
551 	},
552 #endif
553 #ifdef CONFIG_BSD_PROCESS_ACCT
554 	{
555 		.ctl_name	= KERN_ACCT,
556 		.procname	= "acct",
557 		.data		= &acct_parm,
558 		.maxlen		= 3*sizeof(int),
559 		.mode		= 0644,
560 		.proc_handler	= &proc_dointvec,
561 	},
562 #endif
563 #ifdef CONFIG_MAGIC_SYSRQ
564 	{
565 		.ctl_name	= KERN_SYSRQ,
566 		.procname	= "sysrq",
567 		.data		= &__sysrq_enabled,
568 		.maxlen		= sizeof (int),
569 		.mode		= 0644,
570 		.proc_handler	= &proc_dointvec,
571 	},
572 #endif
573 #ifdef CONFIG_PROC_SYSCTL
574 	{
575 		.procname	= "cad_pid",
576 		.data		= NULL,
577 		.maxlen		= sizeof (int),
578 		.mode		= 0600,
579 		.proc_handler	= &proc_do_cad_pid,
580 	},
581 #endif
582 	{
583 		.ctl_name	= KERN_MAX_THREADS,
584 		.procname	= "threads-max",
585 		.data		= &max_threads,
586 		.maxlen		= sizeof(int),
587 		.mode		= 0644,
588 		.proc_handler	= &proc_dointvec,
589 	},
590 	{
591 		.ctl_name	= KERN_RANDOM,
592 		.procname	= "random",
593 		.mode		= 0555,
594 		.child		= random_table,
595 	},
596 	{
597 		.ctl_name	= KERN_OVERFLOWUID,
598 		.procname	= "overflowuid",
599 		.data		= &overflowuid,
600 		.maxlen		= sizeof(int),
601 		.mode		= 0644,
602 		.proc_handler	= &proc_dointvec_minmax,
603 		.strategy	= &sysctl_intvec,
604 		.extra1		= &minolduid,
605 		.extra2		= &maxolduid,
606 	},
607 	{
608 		.ctl_name	= KERN_OVERFLOWGID,
609 		.procname	= "overflowgid",
610 		.data		= &overflowgid,
611 		.maxlen		= sizeof(int),
612 		.mode		= 0644,
613 		.proc_handler	= &proc_dointvec_minmax,
614 		.strategy	= &sysctl_intvec,
615 		.extra1		= &minolduid,
616 		.extra2		= &maxolduid,
617 	},
618 #ifdef CONFIG_S390
619 #ifdef CONFIG_MATHEMU
620 	{
621 		.ctl_name	= KERN_IEEE_EMULATION_WARNINGS,
622 		.procname	= "ieee_emulation_warnings",
623 		.data		= &sysctl_ieee_emulation_warnings,
624 		.maxlen		= sizeof(int),
625 		.mode		= 0644,
626 		.proc_handler	= &proc_dointvec,
627 	},
628 #endif
629 	{
630 		.ctl_name	= KERN_S390_USER_DEBUG_LOGGING,
631 		.procname	= "userprocess_debug",
632 		.data		= &sysctl_userprocess_debug,
633 		.maxlen		= sizeof(int),
634 		.mode		= 0644,
635 		.proc_handler	= &proc_dointvec,
636 	},
637 #endif
638 	{
639 		.ctl_name	= KERN_PIDMAX,
640 		.procname	= "pid_max",
641 		.data		= &pid_max,
642 		.maxlen		= sizeof (int),
643 		.mode		= 0644,
644 		.proc_handler	= &proc_dointvec_minmax,
645 		.strategy	= sysctl_intvec,
646 		.extra1		= &pid_max_min,
647 		.extra2		= &pid_max_max,
648 	},
649 	{
650 		.ctl_name	= KERN_PANIC_ON_OOPS,
651 		.procname	= "panic_on_oops",
652 		.data		= &panic_on_oops,
653 		.maxlen		= sizeof(int),
654 		.mode		= 0644,
655 		.proc_handler	= &proc_dointvec,
656 	},
657 #if defined CONFIG_PRINTK
658 	{
659 		.ctl_name	= KERN_PRINTK,
660 		.procname	= "printk",
661 		.data		= &console_loglevel,
662 		.maxlen		= 4*sizeof(int),
663 		.mode		= 0644,
664 		.proc_handler	= &proc_dointvec,
665 	},
666 	{
667 		.ctl_name	= KERN_PRINTK_RATELIMIT,
668 		.procname	= "printk_ratelimit",
669 		.data		= &printk_ratelimit_state.interval,
670 		.maxlen		= sizeof(int),
671 		.mode		= 0644,
672 		.proc_handler	= &proc_dointvec_jiffies,
673 		.strategy	= &sysctl_jiffies,
674 	},
675 	{
676 		.ctl_name	= KERN_PRINTK_RATELIMIT_BURST,
677 		.procname	= "printk_ratelimit_burst",
678 		.data		= &printk_ratelimit_state.burst,
679 		.maxlen		= sizeof(int),
680 		.mode		= 0644,
681 		.proc_handler	= &proc_dointvec,
682 	},
683 #endif
684 	{
685 		.ctl_name	= KERN_NGROUPS_MAX,
686 		.procname	= "ngroups_max",
687 		.data		= &ngroups_max,
688 		.maxlen		= sizeof (int),
689 		.mode		= 0444,
690 		.proc_handler	= &proc_dointvec,
691 	},
692 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
693 	{
694 		.ctl_name       = KERN_UNKNOWN_NMI_PANIC,
695 		.procname       = "unknown_nmi_panic",
696 		.data           = &unknown_nmi_panic,
697 		.maxlen         = sizeof (int),
698 		.mode           = 0644,
699 		.proc_handler   = &proc_dointvec,
700 	},
701 	{
702 		.procname       = "nmi_watchdog",
703 		.data           = &nmi_watchdog_enabled,
704 		.maxlen         = sizeof (int),
705 		.mode           = 0644,
706 		.proc_handler   = &proc_nmi_enabled,
707 	},
708 #endif
709 #if defined(CONFIG_X86)
710 	{
711 		.ctl_name	= KERN_PANIC_ON_NMI,
712 		.procname	= "panic_on_unrecovered_nmi",
713 		.data		= &panic_on_unrecovered_nmi,
714 		.maxlen		= sizeof(int),
715 		.mode		= 0644,
716 		.proc_handler	= &proc_dointvec,
717 	},
718 	{
719 		.ctl_name	= KERN_BOOTLOADER_TYPE,
720 		.procname	= "bootloader_type",
721 		.data		= &bootloader_type,
722 		.maxlen		= sizeof (int),
723 		.mode		= 0444,
724 		.proc_handler	= &proc_dointvec,
725 	},
726 	{
727 		.ctl_name	= CTL_UNNUMBERED,
728 		.procname	= "kstack_depth_to_print",
729 		.data		= &kstack_depth_to_print,
730 		.maxlen		= sizeof(int),
731 		.mode		= 0644,
732 		.proc_handler	= &proc_dointvec,
733 	},
734 	{
735 		.ctl_name	= CTL_UNNUMBERED,
736 		.procname	= "io_delay_type",
737 		.data		= &io_delay_type,
738 		.maxlen		= sizeof(int),
739 		.mode		= 0644,
740 		.proc_handler	= &proc_dointvec,
741 	},
742 #endif
743 #if defined(CONFIG_MMU)
744 	{
745 		.ctl_name	= KERN_RANDOMIZE,
746 		.procname	= "randomize_va_space",
747 		.data		= &randomize_va_space,
748 		.maxlen		= sizeof(int),
749 		.mode		= 0644,
750 		.proc_handler	= &proc_dointvec,
751 	},
752 #endif
753 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
754 	{
755 		.ctl_name	= KERN_SPIN_RETRY,
756 		.procname	= "spin_retry",
757 		.data		= &spin_retry,
758 		.maxlen		= sizeof (int),
759 		.mode		= 0644,
760 		.proc_handler	= &proc_dointvec,
761 	},
762 #endif
763 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
764 	{
765 		.procname	= "acpi_video_flags",
766 		.data		= &acpi_realmode_flags,
767 		.maxlen		= sizeof (unsigned long),
768 		.mode		= 0644,
769 		.proc_handler	= &proc_doulongvec_minmax,
770 	},
771 #endif
772 #ifdef CONFIG_IA64
773 	{
774 		.ctl_name	= KERN_IA64_UNALIGNED,
775 		.procname	= "ignore-unaligned-usertrap",
776 		.data		= &no_unaligned_warning,
777 		.maxlen		= sizeof (int),
778 	 	.mode		= 0644,
779 		.proc_handler	= &proc_dointvec,
780 	},
781 #endif
782 #ifdef CONFIG_DETECT_SOFTLOCKUP
783 	{
784 		.ctl_name	= CTL_UNNUMBERED,
785 		.procname	= "softlockup_panic",
786 		.data		= &softlockup_panic,
787 		.maxlen		= sizeof(int),
788 		.mode		= 0644,
789 		.proc_handler	= &proc_dointvec_minmax,
790 		.strategy	= &sysctl_intvec,
791 		.extra1		= &zero,
792 		.extra2		= &one,
793 	},
794 	{
795 		.ctl_name	= CTL_UNNUMBERED,
796 		.procname	= "softlockup_thresh",
797 		.data		= &softlockup_thresh,
798 		.maxlen		= sizeof(int),
799 		.mode		= 0644,
800 		.proc_handler	= &proc_dointvec_minmax,
801 		.strategy	= &sysctl_intvec,
802 		.extra1		= &neg_one,
803 		.extra2		= &sixty,
804 	},
805 	{
806 		.ctl_name	= CTL_UNNUMBERED,
807 		.procname	= "hung_task_check_count",
808 		.data		= &sysctl_hung_task_check_count,
809 		.maxlen		= sizeof(unsigned long),
810 		.mode		= 0644,
811 		.proc_handler	= &proc_doulongvec_minmax,
812 		.strategy	= &sysctl_intvec,
813 	},
814 	{
815 		.ctl_name	= CTL_UNNUMBERED,
816 		.procname	= "hung_task_timeout_secs",
817 		.data		= &sysctl_hung_task_timeout_secs,
818 		.maxlen		= sizeof(unsigned long),
819 		.mode		= 0644,
820 		.proc_handler	= &proc_doulongvec_minmax,
821 		.strategy	= &sysctl_intvec,
822 	},
823 	{
824 		.ctl_name	= CTL_UNNUMBERED,
825 		.procname	= "hung_task_warnings",
826 		.data		= &sysctl_hung_task_warnings,
827 		.maxlen		= sizeof(unsigned long),
828 		.mode		= 0644,
829 		.proc_handler	= &proc_doulongvec_minmax,
830 		.strategy	= &sysctl_intvec,
831 	},
832 #endif
833 #ifdef CONFIG_COMPAT
834 	{
835 		.ctl_name	= KERN_COMPAT_LOG,
836 		.procname	= "compat-log",
837 		.data		= &compat_log,
838 		.maxlen		= sizeof (int),
839 	 	.mode		= 0644,
840 		.proc_handler	= &proc_dointvec,
841 	},
842 #endif
843 #ifdef CONFIG_RT_MUTEXES
844 	{
845 		.ctl_name	= KERN_MAX_LOCK_DEPTH,
846 		.procname	= "max_lock_depth",
847 		.data		= &max_lock_depth,
848 		.maxlen		= sizeof(int),
849 		.mode		= 0644,
850 		.proc_handler	= &proc_dointvec,
851 	},
852 #endif
853 	{
854 		.ctl_name	= CTL_UNNUMBERED,
855 		.procname	= "poweroff_cmd",
856 		.data		= &poweroff_cmd,
857 		.maxlen		= POWEROFF_CMD_PATH_LEN,
858 		.mode		= 0644,
859 		.proc_handler	= &proc_dostring,
860 		.strategy	= &sysctl_string,
861 	},
862 #ifdef CONFIG_KEYS
863 	{
864 		.ctl_name	= CTL_UNNUMBERED,
865 		.procname	= "keys",
866 		.mode		= 0555,
867 		.child		= key_sysctls,
868 	},
869 #endif
870 #ifdef CONFIG_RCU_TORTURE_TEST
871 	{
872 		.ctl_name       = CTL_UNNUMBERED,
873 		.procname       = "rcutorture_runnable",
874 		.data           = &rcutorture_runnable,
875 		.maxlen         = sizeof(int),
876 		.mode           = 0644,
877 		.proc_handler   = &proc_dointvec,
878 	},
879 #endif
880 #ifdef CONFIG_UNEVICTABLE_LRU
881 	{
882 		.ctl_name	= CTL_UNNUMBERED,
883 		.procname	= "scan_unevictable_pages",
884 		.data		= &scan_unevictable_pages,
885 		.maxlen		= sizeof(scan_unevictable_pages),
886 		.mode		= 0644,
887 		.proc_handler	= &scan_unevictable_handler,
888 	},
889 #endif
890 /*
891  * NOTE: do not add new entries to this table unless you have read
892  * Documentation/sysctl/ctl_unnumbered.txt
893  */
894 	{ .ctl_name = 0 }
895 };
896 
897 static struct ctl_table vm_table[] = {
898 	{
899 		.ctl_name	= VM_OVERCOMMIT_MEMORY,
900 		.procname	= "overcommit_memory",
901 		.data		= &sysctl_overcommit_memory,
902 		.maxlen		= sizeof(sysctl_overcommit_memory),
903 		.mode		= 0644,
904 		.proc_handler	= &proc_dointvec,
905 	},
906 	{
907 		.ctl_name	= VM_PANIC_ON_OOM,
908 		.procname	= "panic_on_oom",
909 		.data		= &sysctl_panic_on_oom,
910 		.maxlen		= sizeof(sysctl_panic_on_oom),
911 		.mode		= 0644,
912 		.proc_handler	= &proc_dointvec,
913 	},
914 	{
915 		.ctl_name	= CTL_UNNUMBERED,
916 		.procname	= "oom_kill_allocating_task",
917 		.data		= &sysctl_oom_kill_allocating_task,
918 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
919 		.mode		= 0644,
920 		.proc_handler	= &proc_dointvec,
921 	},
922 	{
923 		.ctl_name	= CTL_UNNUMBERED,
924 		.procname	= "oom_dump_tasks",
925 		.data		= &sysctl_oom_dump_tasks,
926 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
927 		.mode		= 0644,
928 		.proc_handler	= &proc_dointvec,
929 	},
930 	{
931 		.ctl_name	= VM_OVERCOMMIT_RATIO,
932 		.procname	= "overcommit_ratio",
933 		.data		= &sysctl_overcommit_ratio,
934 		.maxlen		= sizeof(sysctl_overcommit_ratio),
935 		.mode		= 0644,
936 		.proc_handler	= &proc_dointvec,
937 	},
938 	{
939 		.ctl_name	= VM_PAGE_CLUSTER,
940 		.procname	= "page-cluster",
941 		.data		= &page_cluster,
942 		.maxlen		= sizeof(int),
943 		.mode		= 0644,
944 		.proc_handler	= &proc_dointvec,
945 	},
946 	{
947 		.ctl_name	= VM_DIRTY_BACKGROUND,
948 		.procname	= "dirty_background_ratio",
949 		.data		= &dirty_background_ratio,
950 		.maxlen		= sizeof(dirty_background_ratio),
951 		.mode		= 0644,
952 		.proc_handler	= &dirty_background_ratio_handler,
953 		.strategy	= &sysctl_intvec,
954 		.extra1		= &zero,
955 		.extra2		= &one_hundred,
956 	},
957 	{
958 		.ctl_name	= CTL_UNNUMBERED,
959 		.procname	= "dirty_background_bytes",
960 		.data		= &dirty_background_bytes,
961 		.maxlen		= sizeof(dirty_background_bytes),
962 		.mode		= 0644,
963 		.proc_handler	= &dirty_background_bytes_handler,
964 		.strategy	= &sysctl_intvec,
965 		.extra1		= &one,
966 	},
967 	{
968 		.ctl_name	= VM_DIRTY_RATIO,
969 		.procname	= "dirty_ratio",
970 		.data		= &vm_dirty_ratio,
971 		.maxlen		= sizeof(vm_dirty_ratio),
972 		.mode		= 0644,
973 		.proc_handler	= &dirty_ratio_handler,
974 		.strategy	= &sysctl_intvec,
975 		.extra1		= &zero,
976 		.extra2		= &one_hundred,
977 	},
978 	{
979 		.ctl_name	= CTL_UNNUMBERED,
980 		.procname	= "dirty_bytes",
981 		.data		= &vm_dirty_bytes,
982 		.maxlen		= sizeof(vm_dirty_bytes),
983 		.mode		= 0644,
984 		.proc_handler	= &dirty_bytes_handler,
985 		.strategy	= &sysctl_intvec,
986 		.extra1		= &one,
987 	},
988 	{
989 		.procname	= "dirty_writeback_centisecs",
990 		.data		= &dirty_writeback_interval,
991 		.maxlen		= sizeof(dirty_writeback_interval),
992 		.mode		= 0644,
993 		.proc_handler	= &dirty_writeback_centisecs_handler,
994 	},
995 	{
996 		.procname	= "dirty_expire_centisecs",
997 		.data		= &dirty_expire_interval,
998 		.maxlen		= sizeof(dirty_expire_interval),
999 		.mode		= 0644,
1000 		.proc_handler	= &proc_dointvec_userhz_jiffies,
1001 	},
1002 	{
1003 		.ctl_name	= VM_NR_PDFLUSH_THREADS,
1004 		.procname	= "nr_pdflush_threads",
1005 		.data		= &nr_pdflush_threads,
1006 		.maxlen		= sizeof nr_pdflush_threads,
1007 		.mode		= 0444 /* read-only*/,
1008 		.proc_handler	= &proc_dointvec,
1009 	},
1010 	{
1011 		.ctl_name	= VM_SWAPPINESS,
1012 		.procname	= "swappiness",
1013 		.data		= &vm_swappiness,
1014 		.maxlen		= sizeof(vm_swappiness),
1015 		.mode		= 0644,
1016 		.proc_handler	= &proc_dointvec_minmax,
1017 		.strategy	= &sysctl_intvec,
1018 		.extra1		= &zero,
1019 		.extra2		= &one_hundred,
1020 	},
1021 #ifdef CONFIG_HUGETLB_PAGE
1022 	 {
1023 		.procname	= "nr_hugepages",
1024 		.data		= NULL,
1025 		.maxlen		= sizeof(unsigned long),
1026 		.mode		= 0644,
1027 		.proc_handler	= &hugetlb_sysctl_handler,
1028 		.extra1		= (void *)&hugetlb_zero,
1029 		.extra2		= (void *)&hugetlb_infinity,
1030 	 },
1031 	 {
1032 		.ctl_name	= VM_HUGETLB_GROUP,
1033 		.procname	= "hugetlb_shm_group",
1034 		.data		= &sysctl_hugetlb_shm_group,
1035 		.maxlen		= sizeof(gid_t),
1036 		.mode		= 0644,
1037 		.proc_handler	= &proc_dointvec,
1038 	 },
1039 	 {
1040 		.ctl_name	= CTL_UNNUMBERED,
1041 		.procname	= "hugepages_treat_as_movable",
1042 		.data		= &hugepages_treat_as_movable,
1043 		.maxlen		= sizeof(int),
1044 		.mode		= 0644,
1045 		.proc_handler	= &hugetlb_treat_movable_handler,
1046 	},
1047 	{
1048 		.ctl_name	= CTL_UNNUMBERED,
1049 		.procname	= "nr_overcommit_hugepages",
1050 		.data		= NULL,
1051 		.maxlen		= sizeof(unsigned long),
1052 		.mode		= 0644,
1053 		.proc_handler	= &hugetlb_overcommit_handler,
1054 		.extra1		= (void *)&hugetlb_zero,
1055 		.extra2		= (void *)&hugetlb_infinity,
1056 	},
1057 #endif
1058 	{
1059 		.ctl_name	= VM_LOWMEM_RESERVE_RATIO,
1060 		.procname	= "lowmem_reserve_ratio",
1061 		.data		= &sysctl_lowmem_reserve_ratio,
1062 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1063 		.mode		= 0644,
1064 		.proc_handler	= &lowmem_reserve_ratio_sysctl_handler,
1065 		.strategy	= &sysctl_intvec,
1066 	},
1067 	{
1068 		.ctl_name	= VM_DROP_PAGECACHE,
1069 		.procname	= "drop_caches",
1070 		.data		= &sysctl_drop_caches,
1071 		.maxlen		= sizeof(int),
1072 		.mode		= 0644,
1073 		.proc_handler	= drop_caches_sysctl_handler,
1074 		.strategy	= &sysctl_intvec,
1075 	},
1076 	{
1077 		.ctl_name	= VM_MIN_FREE_KBYTES,
1078 		.procname	= "min_free_kbytes",
1079 		.data		= &min_free_kbytes,
1080 		.maxlen		= sizeof(min_free_kbytes),
1081 		.mode		= 0644,
1082 		.proc_handler	= &min_free_kbytes_sysctl_handler,
1083 		.strategy	= &sysctl_intvec,
1084 		.extra1		= &zero,
1085 	},
1086 	{
1087 		.ctl_name	= VM_PERCPU_PAGELIST_FRACTION,
1088 		.procname	= "percpu_pagelist_fraction",
1089 		.data		= &percpu_pagelist_fraction,
1090 		.maxlen		= sizeof(percpu_pagelist_fraction),
1091 		.mode		= 0644,
1092 		.proc_handler	= &percpu_pagelist_fraction_sysctl_handler,
1093 		.strategy	= &sysctl_intvec,
1094 		.extra1		= &min_percpu_pagelist_fract,
1095 	},
1096 #ifdef CONFIG_MMU
1097 	{
1098 		.ctl_name	= VM_MAX_MAP_COUNT,
1099 		.procname	= "max_map_count",
1100 		.data		= &sysctl_max_map_count,
1101 		.maxlen		= sizeof(sysctl_max_map_count),
1102 		.mode		= 0644,
1103 		.proc_handler	= &proc_dointvec
1104 	},
1105 #endif
1106 	{
1107 		.ctl_name	= VM_LAPTOP_MODE,
1108 		.procname	= "laptop_mode",
1109 		.data		= &laptop_mode,
1110 		.maxlen		= sizeof(laptop_mode),
1111 		.mode		= 0644,
1112 		.proc_handler	= &proc_dointvec_jiffies,
1113 		.strategy	= &sysctl_jiffies,
1114 	},
1115 	{
1116 		.ctl_name	= VM_BLOCK_DUMP,
1117 		.procname	= "block_dump",
1118 		.data		= &block_dump,
1119 		.maxlen		= sizeof(block_dump),
1120 		.mode		= 0644,
1121 		.proc_handler	= &proc_dointvec,
1122 		.strategy	= &sysctl_intvec,
1123 		.extra1		= &zero,
1124 	},
1125 	{
1126 		.ctl_name	= VM_VFS_CACHE_PRESSURE,
1127 		.procname	= "vfs_cache_pressure",
1128 		.data		= &sysctl_vfs_cache_pressure,
1129 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1130 		.mode		= 0644,
1131 		.proc_handler	= &proc_dointvec,
1132 		.strategy	= &sysctl_intvec,
1133 		.extra1		= &zero,
1134 	},
1135 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1136 	{
1137 		.ctl_name	= VM_LEGACY_VA_LAYOUT,
1138 		.procname	= "legacy_va_layout",
1139 		.data		= &sysctl_legacy_va_layout,
1140 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1141 		.mode		= 0644,
1142 		.proc_handler	= &proc_dointvec,
1143 		.strategy	= &sysctl_intvec,
1144 		.extra1		= &zero,
1145 	},
1146 #endif
1147 #ifdef CONFIG_NUMA
1148 	{
1149 		.ctl_name	= VM_ZONE_RECLAIM_MODE,
1150 		.procname	= "zone_reclaim_mode",
1151 		.data		= &zone_reclaim_mode,
1152 		.maxlen		= sizeof(zone_reclaim_mode),
1153 		.mode		= 0644,
1154 		.proc_handler	= &proc_dointvec,
1155 		.strategy	= &sysctl_intvec,
1156 		.extra1		= &zero,
1157 	},
1158 	{
1159 		.ctl_name	= VM_MIN_UNMAPPED,
1160 		.procname	= "min_unmapped_ratio",
1161 		.data		= &sysctl_min_unmapped_ratio,
1162 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1163 		.mode		= 0644,
1164 		.proc_handler	= &sysctl_min_unmapped_ratio_sysctl_handler,
1165 		.strategy	= &sysctl_intvec,
1166 		.extra1		= &zero,
1167 		.extra2		= &one_hundred,
1168 	},
1169 	{
1170 		.ctl_name	= VM_MIN_SLAB,
1171 		.procname	= "min_slab_ratio",
1172 		.data		= &sysctl_min_slab_ratio,
1173 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1174 		.mode		= 0644,
1175 		.proc_handler	= &sysctl_min_slab_ratio_sysctl_handler,
1176 		.strategy	= &sysctl_intvec,
1177 		.extra1		= &zero,
1178 		.extra2		= &one_hundred,
1179 	},
1180 #endif
1181 #ifdef CONFIG_SMP
1182 	{
1183 		.ctl_name	= CTL_UNNUMBERED,
1184 		.procname	= "stat_interval",
1185 		.data		= &sysctl_stat_interval,
1186 		.maxlen		= sizeof(sysctl_stat_interval),
1187 		.mode		= 0644,
1188 		.proc_handler	= &proc_dointvec_jiffies,
1189 		.strategy	= &sysctl_jiffies,
1190 	},
1191 #endif
1192 #ifdef CONFIG_SECURITY
1193 	{
1194 		.ctl_name	= CTL_UNNUMBERED,
1195 		.procname	= "mmap_min_addr",
1196 		.data		= &mmap_min_addr,
1197 		.maxlen         = sizeof(unsigned long),
1198 		.mode		= 0644,
1199 		.proc_handler	= &proc_doulongvec_minmax,
1200 	},
1201 #endif
1202 #ifdef CONFIG_NUMA
1203 	{
1204 		.ctl_name	= CTL_UNNUMBERED,
1205 		.procname	= "numa_zonelist_order",
1206 		.data		= &numa_zonelist_order,
1207 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1208 		.mode		= 0644,
1209 		.proc_handler	= &numa_zonelist_order_handler,
1210 		.strategy	= &sysctl_string,
1211 	},
1212 #endif
1213 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1214    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1215 	{
1216 		.ctl_name	= VM_VDSO_ENABLED,
1217 		.procname	= "vdso_enabled",
1218 		.data		= &vdso_enabled,
1219 		.maxlen		= sizeof(vdso_enabled),
1220 		.mode		= 0644,
1221 		.proc_handler	= &proc_dointvec,
1222 		.strategy	= &sysctl_intvec,
1223 		.extra1		= &zero,
1224 	},
1225 #endif
1226 #ifdef CONFIG_HIGHMEM
1227 	{
1228 		.ctl_name	= CTL_UNNUMBERED,
1229 		.procname	= "highmem_is_dirtyable",
1230 		.data		= &vm_highmem_is_dirtyable,
1231 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1232 		.mode		= 0644,
1233 		.proc_handler	= &proc_dointvec_minmax,
1234 		.strategy	= &sysctl_intvec,
1235 		.extra1		= &zero,
1236 		.extra2		= &one,
1237 	},
1238 #endif
1239 /*
1240  * NOTE: do not add new entries to this table unless you have read
1241  * Documentation/sysctl/ctl_unnumbered.txt
1242  */
1243 	{ .ctl_name = 0 }
1244 };
1245 
1246 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1247 static struct ctl_table binfmt_misc_table[] = {
1248 	{ .ctl_name = 0 }
1249 };
1250 #endif
1251 
1252 static struct ctl_table fs_table[] = {
1253 	{
1254 		.ctl_name	= FS_NRINODE,
1255 		.procname	= "inode-nr",
1256 		.data		= &inodes_stat,
1257 		.maxlen		= 2*sizeof(int),
1258 		.mode		= 0444,
1259 		.proc_handler	= &proc_dointvec,
1260 	},
1261 	{
1262 		.ctl_name	= FS_STATINODE,
1263 		.procname	= "inode-state",
1264 		.data		= &inodes_stat,
1265 		.maxlen		= 7*sizeof(int),
1266 		.mode		= 0444,
1267 		.proc_handler	= &proc_dointvec,
1268 	},
1269 	{
1270 		.procname	= "file-nr",
1271 		.data		= &files_stat,
1272 		.maxlen		= 3*sizeof(int),
1273 		.mode		= 0444,
1274 		.proc_handler	= &proc_nr_files,
1275 	},
1276 	{
1277 		.ctl_name	= FS_MAXFILE,
1278 		.procname	= "file-max",
1279 		.data		= &files_stat.max_files,
1280 		.maxlen		= sizeof(int),
1281 		.mode		= 0644,
1282 		.proc_handler	= &proc_dointvec,
1283 	},
1284 	{
1285 		.ctl_name	= CTL_UNNUMBERED,
1286 		.procname	= "nr_open",
1287 		.data		= &sysctl_nr_open,
1288 		.maxlen		= sizeof(int),
1289 		.mode		= 0644,
1290 		.proc_handler	= &proc_dointvec_minmax,
1291 		.extra1		= &sysctl_nr_open_min,
1292 		.extra2		= &sysctl_nr_open_max,
1293 	},
1294 	{
1295 		.ctl_name	= FS_DENTRY,
1296 		.procname	= "dentry-state",
1297 		.data		= &dentry_stat,
1298 		.maxlen		= 6*sizeof(int),
1299 		.mode		= 0444,
1300 		.proc_handler	= &proc_dointvec,
1301 	},
1302 	{
1303 		.ctl_name	= FS_OVERFLOWUID,
1304 		.procname	= "overflowuid",
1305 		.data		= &fs_overflowuid,
1306 		.maxlen		= sizeof(int),
1307 		.mode		= 0644,
1308 		.proc_handler	= &proc_dointvec_minmax,
1309 		.strategy	= &sysctl_intvec,
1310 		.extra1		= &minolduid,
1311 		.extra2		= &maxolduid,
1312 	},
1313 	{
1314 		.ctl_name	= FS_OVERFLOWGID,
1315 		.procname	= "overflowgid",
1316 		.data		= &fs_overflowgid,
1317 		.maxlen		= sizeof(int),
1318 		.mode		= 0644,
1319 		.proc_handler	= &proc_dointvec_minmax,
1320 		.strategy	= &sysctl_intvec,
1321 		.extra1		= &minolduid,
1322 		.extra2		= &maxolduid,
1323 	},
1324 #ifdef CONFIG_FILE_LOCKING
1325 	{
1326 		.ctl_name	= FS_LEASES,
1327 		.procname	= "leases-enable",
1328 		.data		= &leases_enable,
1329 		.maxlen		= sizeof(int),
1330 		.mode		= 0644,
1331 		.proc_handler	= &proc_dointvec,
1332 	},
1333 #endif
1334 #ifdef CONFIG_DNOTIFY
1335 	{
1336 		.ctl_name	= FS_DIR_NOTIFY,
1337 		.procname	= "dir-notify-enable",
1338 		.data		= &dir_notify_enable,
1339 		.maxlen		= sizeof(int),
1340 		.mode		= 0644,
1341 		.proc_handler	= &proc_dointvec,
1342 	},
1343 #endif
1344 #ifdef CONFIG_MMU
1345 #ifdef CONFIG_FILE_LOCKING
1346 	{
1347 		.ctl_name	= FS_LEASE_TIME,
1348 		.procname	= "lease-break-time",
1349 		.data		= &lease_break_time,
1350 		.maxlen		= sizeof(int),
1351 		.mode		= 0644,
1352 		.proc_handler	= &proc_dointvec_minmax,
1353 		.strategy	= &sysctl_intvec,
1354 		.extra1		= &zero,
1355 		.extra2		= &two,
1356 	},
1357 #endif
1358 #ifdef CONFIG_AIO
1359 	{
1360 		.procname	= "aio-nr",
1361 		.data		= &aio_nr,
1362 		.maxlen		= sizeof(aio_nr),
1363 		.mode		= 0444,
1364 		.proc_handler	= &proc_doulongvec_minmax,
1365 	},
1366 	{
1367 		.procname	= "aio-max-nr",
1368 		.data		= &aio_max_nr,
1369 		.maxlen		= sizeof(aio_max_nr),
1370 		.mode		= 0644,
1371 		.proc_handler	= &proc_doulongvec_minmax,
1372 	},
1373 #endif /* CONFIG_AIO */
1374 #ifdef CONFIG_INOTIFY_USER
1375 	{
1376 		.ctl_name	= FS_INOTIFY,
1377 		.procname	= "inotify",
1378 		.mode		= 0555,
1379 		.child		= inotify_table,
1380 	},
1381 #endif
1382 #ifdef CONFIG_EPOLL
1383 	{
1384 		.procname	= "epoll",
1385 		.mode		= 0555,
1386 		.child		= epoll_table,
1387 	},
1388 #endif
1389 #endif
1390 	{
1391 		.ctl_name	= KERN_SETUID_DUMPABLE,
1392 		.procname	= "suid_dumpable",
1393 		.data		= &suid_dumpable,
1394 		.maxlen		= sizeof(int),
1395 		.mode		= 0644,
1396 		.proc_handler	= &proc_dointvec,
1397 	},
1398 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1399 	{
1400 		.ctl_name	= CTL_UNNUMBERED,
1401 		.procname	= "binfmt_misc",
1402 		.mode		= 0555,
1403 		.child		= binfmt_misc_table,
1404 	},
1405 #endif
1406 /*
1407  * NOTE: do not add new entries to this table unless you have read
1408  * Documentation/sysctl/ctl_unnumbered.txt
1409  */
1410 	{ .ctl_name = 0 }
1411 };
1412 
1413 static struct ctl_table debug_table[] = {
1414 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1415 	{
1416 		.ctl_name	= CTL_UNNUMBERED,
1417 		.procname	= "exception-trace",
1418 		.data		= &show_unhandled_signals,
1419 		.maxlen		= sizeof(int),
1420 		.mode		= 0644,
1421 		.proc_handler	= proc_dointvec
1422 	},
1423 #endif
1424 	{ .ctl_name = 0 }
1425 };
1426 
1427 static struct ctl_table dev_table[] = {
1428 	{ .ctl_name = 0 }
1429 };
1430 
1431 static DEFINE_SPINLOCK(sysctl_lock);
1432 
1433 /* called under sysctl_lock */
1434 static int use_table(struct ctl_table_header *p)
1435 {
1436 	if (unlikely(p->unregistering))
1437 		return 0;
1438 	p->used++;
1439 	return 1;
1440 }
1441 
1442 /* called under sysctl_lock */
1443 static void unuse_table(struct ctl_table_header *p)
1444 {
1445 	if (!--p->used)
1446 		if (unlikely(p->unregistering))
1447 			complete(p->unregistering);
1448 }
1449 
1450 /* called under sysctl_lock, will reacquire if has to wait */
1451 static void start_unregistering(struct ctl_table_header *p)
1452 {
1453 	/*
1454 	 * if p->used is 0, nobody will ever touch that entry again;
1455 	 * we'll eliminate all paths to it before dropping sysctl_lock
1456 	 */
1457 	if (unlikely(p->used)) {
1458 		struct completion wait;
1459 		init_completion(&wait);
1460 		p->unregistering = &wait;
1461 		spin_unlock(&sysctl_lock);
1462 		wait_for_completion(&wait);
1463 		spin_lock(&sysctl_lock);
1464 	} else {
1465 		/* anything non-NULL; we'll never dereference it */
1466 		p->unregistering = ERR_PTR(-EINVAL);
1467 	}
1468 	/*
1469 	 * do not remove from the list until nobody holds it; walking the
1470 	 * list in do_sysctl() relies on that.
1471 	 */
1472 	list_del_init(&p->ctl_entry);
1473 }
1474 
1475 void sysctl_head_get(struct ctl_table_header *head)
1476 {
1477 	spin_lock(&sysctl_lock);
1478 	head->count++;
1479 	spin_unlock(&sysctl_lock);
1480 }
1481 
1482 void sysctl_head_put(struct ctl_table_header *head)
1483 {
1484 	spin_lock(&sysctl_lock);
1485 	if (!--head->count)
1486 		kfree(head);
1487 	spin_unlock(&sysctl_lock);
1488 }
1489 
1490 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1491 {
1492 	if (!head)
1493 		BUG();
1494 	spin_lock(&sysctl_lock);
1495 	if (!use_table(head))
1496 		head = ERR_PTR(-ENOENT);
1497 	spin_unlock(&sysctl_lock);
1498 	return head;
1499 }
1500 
1501 void sysctl_head_finish(struct ctl_table_header *head)
1502 {
1503 	if (!head)
1504 		return;
1505 	spin_lock(&sysctl_lock);
1506 	unuse_table(head);
1507 	spin_unlock(&sysctl_lock);
1508 }
1509 
1510 static struct ctl_table_set *
1511 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1512 {
1513 	struct ctl_table_set *set = &root->default_set;
1514 	if (root->lookup)
1515 		set = root->lookup(root, namespaces);
1516 	return set;
1517 }
1518 
1519 static struct list_head *
1520 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1521 {
1522 	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1523 	return &set->list;
1524 }
1525 
1526 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1527 					    struct ctl_table_header *prev)
1528 {
1529 	struct ctl_table_root *root;
1530 	struct list_head *header_list;
1531 	struct ctl_table_header *head;
1532 	struct list_head *tmp;
1533 
1534 	spin_lock(&sysctl_lock);
1535 	if (prev) {
1536 		head = prev;
1537 		tmp = &prev->ctl_entry;
1538 		unuse_table(prev);
1539 		goto next;
1540 	}
1541 	tmp = &root_table_header.ctl_entry;
1542 	for (;;) {
1543 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1544 
1545 		if (!use_table(head))
1546 			goto next;
1547 		spin_unlock(&sysctl_lock);
1548 		return head;
1549 	next:
1550 		root = head->root;
1551 		tmp = tmp->next;
1552 		header_list = lookup_header_list(root, namespaces);
1553 		if (tmp != header_list)
1554 			continue;
1555 
1556 		do {
1557 			root = list_entry(root->root_list.next,
1558 					struct ctl_table_root, root_list);
1559 			if (root == &sysctl_table_root)
1560 				goto out;
1561 			header_list = lookup_header_list(root, namespaces);
1562 		} while (list_empty(header_list));
1563 		tmp = header_list->next;
1564 	}
1565 out:
1566 	spin_unlock(&sysctl_lock);
1567 	return NULL;
1568 }
1569 
1570 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1571 {
1572 	return __sysctl_head_next(current->nsproxy, prev);
1573 }
1574 
1575 void register_sysctl_root(struct ctl_table_root *root)
1576 {
1577 	spin_lock(&sysctl_lock);
1578 	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1579 	spin_unlock(&sysctl_lock);
1580 }
1581 
1582 #ifdef CONFIG_SYSCTL_SYSCALL
1583 /* Perform the actual read/write of a sysctl table entry. */
1584 static int do_sysctl_strategy(struct ctl_table_root *root,
1585 			struct ctl_table *table,
1586 			void __user *oldval, size_t __user *oldlenp,
1587 			void __user *newval, size_t newlen)
1588 {
1589 	int op = 0, rc;
1590 
1591 	if (oldval)
1592 		op |= MAY_READ;
1593 	if (newval)
1594 		op |= MAY_WRITE;
1595 	if (sysctl_perm(root, table, op))
1596 		return -EPERM;
1597 
1598 	if (table->strategy) {
1599 		rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1600 		if (rc < 0)
1601 			return rc;
1602 		if (rc > 0)
1603 			return 0;
1604 	}
1605 
1606 	/* If there is no strategy routine, or if the strategy returns
1607 	 * zero, proceed with automatic r/w */
1608 	if (table->data && table->maxlen) {
1609 		rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1610 		if (rc < 0)
1611 			return rc;
1612 	}
1613 	return 0;
1614 }
1615 
1616 static int parse_table(int __user *name, int nlen,
1617 		       void __user *oldval, size_t __user *oldlenp,
1618 		       void __user *newval, size_t newlen,
1619 		       struct ctl_table_root *root,
1620 		       struct ctl_table *table)
1621 {
1622 	int n;
1623 repeat:
1624 	if (!nlen)
1625 		return -ENOTDIR;
1626 	if (get_user(n, name))
1627 		return -EFAULT;
1628 	for ( ; table->ctl_name || table->procname; table++) {
1629 		if (!table->ctl_name)
1630 			continue;
1631 		if (n == table->ctl_name) {
1632 			int error;
1633 			if (table->child) {
1634 				if (sysctl_perm(root, table, MAY_EXEC))
1635 					return -EPERM;
1636 				name++;
1637 				nlen--;
1638 				table = table->child;
1639 				goto repeat;
1640 			}
1641 			error = do_sysctl_strategy(root, table,
1642 						   oldval, oldlenp,
1643 						   newval, newlen);
1644 			return error;
1645 		}
1646 	}
1647 	return -ENOTDIR;
1648 }
1649 
1650 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1651 	       void __user *newval, size_t newlen)
1652 {
1653 	struct ctl_table_header *head;
1654 	int error = -ENOTDIR;
1655 
1656 	if (nlen <= 0 || nlen >= CTL_MAXNAME)
1657 		return -ENOTDIR;
1658 	if (oldval) {
1659 		int old_len;
1660 		if (!oldlenp || get_user(old_len, oldlenp))
1661 			return -EFAULT;
1662 	}
1663 
1664 	for (head = sysctl_head_next(NULL); head;
1665 			head = sysctl_head_next(head)) {
1666 		error = parse_table(name, nlen, oldval, oldlenp,
1667 					newval, newlen,
1668 					head->root, head->ctl_table);
1669 		if (error != -ENOTDIR) {
1670 			sysctl_head_finish(head);
1671 			break;
1672 		}
1673 	}
1674 	return error;
1675 }
1676 
1677 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1678 {
1679 	struct __sysctl_args tmp;
1680 	int error;
1681 
1682 	if (copy_from_user(&tmp, args, sizeof(tmp)))
1683 		return -EFAULT;
1684 
1685 	error = deprecated_sysctl_warning(&tmp);
1686 	if (error)
1687 		goto out;
1688 
1689 	lock_kernel();
1690 	error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1691 			  tmp.newval, tmp.newlen);
1692 	unlock_kernel();
1693 out:
1694 	return error;
1695 }
1696 #endif /* CONFIG_SYSCTL_SYSCALL */
1697 
1698 /*
1699  * sysctl_perm does NOT grant the superuser all rights automatically, because
1700  * some sysctl variables are readonly even to root.
1701  */
1702 
1703 static int test_perm(int mode, int op)
1704 {
1705 	if (!current_euid())
1706 		mode >>= 6;
1707 	else if (in_egroup_p(0))
1708 		mode >>= 3;
1709 	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1710 		return 0;
1711 	return -EACCES;
1712 }
1713 
1714 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1715 {
1716 	int error;
1717 	int mode;
1718 
1719 	error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1720 	if (error)
1721 		return error;
1722 
1723 	if (root->permissions)
1724 		mode = root->permissions(root, current->nsproxy, table);
1725 	else
1726 		mode = table->mode;
1727 
1728 	return test_perm(mode, op);
1729 }
1730 
1731 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1732 {
1733 	for (; table->ctl_name || table->procname; table++) {
1734 		table->parent = parent;
1735 		if (table->child)
1736 			sysctl_set_parent(table, table->child);
1737 	}
1738 }
1739 
1740 static __init int sysctl_init(void)
1741 {
1742 	sysctl_set_parent(NULL, root_table);
1743 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1744 	{
1745 		int err;
1746 		err = sysctl_check_table(current->nsproxy, root_table);
1747 	}
1748 #endif
1749 	return 0;
1750 }
1751 
1752 core_initcall(sysctl_init);
1753 
1754 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1755 				      struct ctl_table *table)
1756 {
1757 	struct ctl_table *p;
1758 	const char *s = branch->procname;
1759 
1760 	/* branch should have named subdirectory as its first element */
1761 	if (!s || !branch->child)
1762 		return NULL;
1763 
1764 	/* ... and nothing else */
1765 	if (branch[1].procname || branch[1].ctl_name)
1766 		return NULL;
1767 
1768 	/* table should contain subdirectory with the same name */
1769 	for (p = table; p->procname || p->ctl_name; p++) {
1770 		if (!p->child)
1771 			continue;
1772 		if (p->procname && strcmp(p->procname, s) == 0)
1773 			return p;
1774 	}
1775 	return NULL;
1776 }
1777 
1778 /* see if attaching q to p would be an improvement */
1779 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1780 {
1781 	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1782 	struct ctl_table *next;
1783 	int is_better = 0;
1784 	int not_in_parent = !p->attached_by;
1785 
1786 	while ((next = is_branch_in(by, to)) != NULL) {
1787 		if (by == q->attached_by)
1788 			is_better = 1;
1789 		if (to == p->attached_by)
1790 			not_in_parent = 1;
1791 		by = by->child;
1792 		to = next->child;
1793 	}
1794 
1795 	if (is_better && not_in_parent) {
1796 		q->attached_by = by;
1797 		q->attached_to = to;
1798 		q->parent = p;
1799 	}
1800 }
1801 
1802 /**
1803  * __register_sysctl_paths - register a sysctl hierarchy
1804  * @root: List of sysctl headers to register on
1805  * @namespaces: Data to compute which lists of sysctl entries are visible
1806  * @path: The path to the directory the sysctl table is in.
1807  * @table: the top-level table structure
1808  *
1809  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1810  * array. A completely 0 filled entry terminates the table.
1811  *
1812  * The members of the &struct ctl_table structure are used as follows:
1813  *
1814  * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1815  *            must be unique within that level of sysctl
1816  *
1817  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1818  *            enter a sysctl file
1819  *
1820  * data - a pointer to data for use by proc_handler
1821  *
1822  * maxlen - the maximum size in bytes of the data
1823  *
1824  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1825  *
1826  * child - a pointer to the child sysctl table if this entry is a directory, or
1827  *         %NULL.
1828  *
1829  * proc_handler - the text handler routine (described below)
1830  *
1831  * strategy - the strategy routine (described below)
1832  *
1833  * de - for internal use by the sysctl routines
1834  *
1835  * extra1, extra2 - extra pointers usable by the proc handler routines
1836  *
1837  * Leaf nodes in the sysctl tree will be represented by a single file
1838  * under /proc; non-leaf nodes will be represented by directories.
1839  *
1840  * sysctl(2) can automatically manage read and write requests through
1841  * the sysctl table.  The data and maxlen fields of the ctl_table
1842  * struct enable minimal validation of the values being written to be
1843  * performed, and the mode field allows minimal authentication.
1844  *
1845  * More sophisticated management can be enabled by the provision of a
1846  * strategy routine with the table entry.  This will be called before
1847  * any automatic read or write of the data is performed.
1848  *
1849  * The strategy routine may return
1850  *
1851  * < 0 - Error occurred (error is passed to user process)
1852  *
1853  * 0   - OK - proceed with automatic read or write.
1854  *
1855  * > 0 - OK - read or write has been done by the strategy routine, so
1856  *       return immediately.
1857  *
1858  * There must be a proc_handler routine for any terminal nodes
1859  * mirrored under /proc/sys (non-terminals are handled by a built-in
1860  * directory handler).  Several default handlers are available to
1861  * cover common cases -
1862  *
1863  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1864  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1865  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1866  *
1867  * It is the handler's job to read the input buffer from user memory
1868  * and process it. The handler should return 0 on success.
1869  *
1870  * This routine returns %NULL on a failure to register, and a pointer
1871  * to the table header on success.
1872  */
1873 struct ctl_table_header *__register_sysctl_paths(
1874 	struct ctl_table_root *root,
1875 	struct nsproxy *namespaces,
1876 	const struct ctl_path *path, struct ctl_table *table)
1877 {
1878 	struct ctl_table_header *header;
1879 	struct ctl_table *new, **prevp;
1880 	unsigned int n, npath;
1881 	struct ctl_table_set *set;
1882 
1883 	/* Count the path components */
1884 	for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1885 		;
1886 
1887 	/*
1888 	 * For each path component, allocate a 2-element ctl_table array.
1889 	 * The first array element will be filled with the sysctl entry
1890 	 * for this, the second will be the sentinel (ctl_name == 0).
1891 	 *
1892 	 * We allocate everything in one go so that we don't have to
1893 	 * worry about freeing additional memory in unregister_sysctl_table.
1894 	 */
1895 	header = kzalloc(sizeof(struct ctl_table_header) +
1896 			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1897 	if (!header)
1898 		return NULL;
1899 
1900 	new = (struct ctl_table *) (header + 1);
1901 
1902 	/* Now connect the dots */
1903 	prevp = &header->ctl_table;
1904 	for (n = 0; n < npath; ++n, ++path) {
1905 		/* Copy the procname */
1906 		new->procname = path->procname;
1907 		new->ctl_name = path->ctl_name;
1908 		new->mode     = 0555;
1909 
1910 		*prevp = new;
1911 		prevp = &new->child;
1912 
1913 		new += 2;
1914 	}
1915 	*prevp = table;
1916 	header->ctl_table_arg = table;
1917 
1918 	INIT_LIST_HEAD(&header->ctl_entry);
1919 	header->used = 0;
1920 	header->unregistering = NULL;
1921 	header->root = root;
1922 	sysctl_set_parent(NULL, header->ctl_table);
1923 	header->count = 1;
1924 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1925 	if (sysctl_check_table(namespaces, header->ctl_table)) {
1926 		kfree(header);
1927 		return NULL;
1928 	}
1929 #endif
1930 	spin_lock(&sysctl_lock);
1931 	header->set = lookup_header_set(root, namespaces);
1932 	header->attached_by = header->ctl_table;
1933 	header->attached_to = root_table;
1934 	header->parent = &root_table_header;
1935 	for (set = header->set; set; set = set->parent) {
1936 		struct ctl_table_header *p;
1937 		list_for_each_entry(p, &set->list, ctl_entry) {
1938 			if (p->unregistering)
1939 				continue;
1940 			try_attach(p, header);
1941 		}
1942 	}
1943 	header->parent->count++;
1944 	list_add_tail(&header->ctl_entry, &header->set->list);
1945 	spin_unlock(&sysctl_lock);
1946 
1947 	return header;
1948 }
1949 
1950 /**
1951  * register_sysctl_table_path - register a sysctl table hierarchy
1952  * @path: The path to the directory the sysctl table is in.
1953  * @table: the top-level table structure
1954  *
1955  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1956  * array. A completely 0 filled entry terminates the table.
1957  *
1958  * See __register_sysctl_paths for more details.
1959  */
1960 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1961 						struct ctl_table *table)
1962 {
1963 	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1964 					path, table);
1965 }
1966 
1967 /**
1968  * register_sysctl_table - register a sysctl table hierarchy
1969  * @table: the top-level table structure
1970  *
1971  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1972  * array. A completely 0 filled entry terminates the table.
1973  *
1974  * See register_sysctl_paths for more details.
1975  */
1976 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1977 {
1978 	static const struct ctl_path null_path[] = { {} };
1979 
1980 	return register_sysctl_paths(null_path, table);
1981 }
1982 
1983 /**
1984  * unregister_sysctl_table - unregister a sysctl table hierarchy
1985  * @header: the header returned from register_sysctl_table
1986  *
1987  * Unregisters the sysctl table and all children. proc entries may not
1988  * actually be removed until they are no longer used by anyone.
1989  */
1990 void unregister_sysctl_table(struct ctl_table_header * header)
1991 {
1992 	might_sleep();
1993 
1994 	if (header == NULL)
1995 		return;
1996 
1997 	spin_lock(&sysctl_lock);
1998 	start_unregistering(header);
1999 	if (!--header->parent->count) {
2000 		WARN_ON(1);
2001 		kfree(header->parent);
2002 	}
2003 	if (!--header->count)
2004 		kfree(header);
2005 	spin_unlock(&sysctl_lock);
2006 }
2007 
2008 int sysctl_is_seen(struct ctl_table_header *p)
2009 {
2010 	struct ctl_table_set *set = p->set;
2011 	int res;
2012 	spin_lock(&sysctl_lock);
2013 	if (p->unregistering)
2014 		res = 0;
2015 	else if (!set->is_seen)
2016 		res = 1;
2017 	else
2018 		res = set->is_seen(set);
2019 	spin_unlock(&sysctl_lock);
2020 	return res;
2021 }
2022 
2023 void setup_sysctl_set(struct ctl_table_set *p,
2024 	struct ctl_table_set *parent,
2025 	int (*is_seen)(struct ctl_table_set *))
2026 {
2027 	INIT_LIST_HEAD(&p->list);
2028 	p->parent = parent ? parent : &sysctl_table_root.default_set;
2029 	p->is_seen = is_seen;
2030 }
2031 
2032 #else /* !CONFIG_SYSCTL */
2033 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2034 {
2035 	return NULL;
2036 }
2037 
2038 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2039 						    struct ctl_table *table)
2040 {
2041 	return NULL;
2042 }
2043 
2044 void unregister_sysctl_table(struct ctl_table_header * table)
2045 {
2046 }
2047 
2048 void setup_sysctl_set(struct ctl_table_set *p,
2049 	struct ctl_table_set *parent,
2050 	int (*is_seen)(struct ctl_table_set *))
2051 {
2052 }
2053 
2054 void sysctl_head_put(struct ctl_table_header *head)
2055 {
2056 }
2057 
2058 #endif /* CONFIG_SYSCTL */
2059 
2060 /*
2061  * /proc/sys support
2062  */
2063 
2064 #ifdef CONFIG_PROC_SYSCTL
2065 
2066 static int _proc_do_string(void* data, int maxlen, int write,
2067 			   struct file *filp, void __user *buffer,
2068 			   size_t *lenp, loff_t *ppos)
2069 {
2070 	size_t len;
2071 	char __user *p;
2072 	char c;
2073 
2074 	if (!data || !maxlen || !*lenp) {
2075 		*lenp = 0;
2076 		return 0;
2077 	}
2078 
2079 	if (write) {
2080 		len = 0;
2081 		p = buffer;
2082 		while (len < *lenp) {
2083 			if (get_user(c, p++))
2084 				return -EFAULT;
2085 			if (c == 0 || c == '\n')
2086 				break;
2087 			len++;
2088 		}
2089 		if (len >= maxlen)
2090 			len = maxlen-1;
2091 		if(copy_from_user(data, buffer, len))
2092 			return -EFAULT;
2093 		((char *) data)[len] = 0;
2094 		*ppos += *lenp;
2095 	} else {
2096 		len = strlen(data);
2097 		if (len > maxlen)
2098 			len = maxlen;
2099 
2100 		if (*ppos > len) {
2101 			*lenp = 0;
2102 			return 0;
2103 		}
2104 
2105 		data += *ppos;
2106 		len  -= *ppos;
2107 
2108 		if (len > *lenp)
2109 			len = *lenp;
2110 		if (len)
2111 			if(copy_to_user(buffer, data, len))
2112 				return -EFAULT;
2113 		if (len < *lenp) {
2114 			if(put_user('\n', ((char __user *) buffer) + len))
2115 				return -EFAULT;
2116 			len++;
2117 		}
2118 		*lenp = len;
2119 		*ppos += len;
2120 	}
2121 	return 0;
2122 }
2123 
2124 /**
2125  * proc_dostring - read a string sysctl
2126  * @table: the sysctl table
2127  * @write: %TRUE if this is a write to the sysctl file
2128  * @filp: the file structure
2129  * @buffer: the user buffer
2130  * @lenp: the size of the user buffer
2131  * @ppos: file position
2132  *
2133  * Reads/writes a string from/to the user buffer. If the kernel
2134  * buffer provided is not large enough to hold the string, the
2135  * string is truncated. The copied string is %NULL-terminated.
2136  * If the string is being read by the user process, it is copied
2137  * and a newline '\n' is added. It is truncated if the buffer is
2138  * not large enough.
2139  *
2140  * Returns 0 on success.
2141  */
2142 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2143 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2144 {
2145 	return _proc_do_string(table->data, table->maxlen, write, filp,
2146 			       buffer, lenp, ppos);
2147 }
2148 
2149 
2150 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2151 				 int *valp,
2152 				 int write, void *data)
2153 {
2154 	if (write) {
2155 		*valp = *negp ? -*lvalp : *lvalp;
2156 	} else {
2157 		int val = *valp;
2158 		if (val < 0) {
2159 			*negp = -1;
2160 			*lvalp = (unsigned long)-val;
2161 		} else {
2162 			*negp = 0;
2163 			*lvalp = (unsigned long)val;
2164 		}
2165 	}
2166 	return 0;
2167 }
2168 
2169 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2170 		  int write, struct file *filp, void __user *buffer,
2171 		  size_t *lenp, loff_t *ppos,
2172 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2173 			      int write, void *data),
2174 		  void *data)
2175 {
2176 #define TMPBUFLEN 21
2177 	int *i, vleft, first=1, neg, val;
2178 	unsigned long lval;
2179 	size_t left, len;
2180 
2181 	char buf[TMPBUFLEN], *p;
2182 	char __user *s = buffer;
2183 
2184 	if (!tbl_data || !table->maxlen || !*lenp ||
2185 	    (*ppos && !write)) {
2186 		*lenp = 0;
2187 		return 0;
2188 	}
2189 
2190 	i = (int *) tbl_data;
2191 	vleft = table->maxlen / sizeof(*i);
2192 	left = *lenp;
2193 
2194 	if (!conv)
2195 		conv = do_proc_dointvec_conv;
2196 
2197 	for (; left && vleft--; i++, first=0) {
2198 		if (write) {
2199 			while (left) {
2200 				char c;
2201 				if (get_user(c, s))
2202 					return -EFAULT;
2203 				if (!isspace(c))
2204 					break;
2205 				left--;
2206 				s++;
2207 			}
2208 			if (!left)
2209 				break;
2210 			neg = 0;
2211 			len = left;
2212 			if (len > sizeof(buf) - 1)
2213 				len = sizeof(buf) - 1;
2214 			if (copy_from_user(buf, s, len))
2215 				return -EFAULT;
2216 			buf[len] = 0;
2217 			p = buf;
2218 			if (*p == '-' && left > 1) {
2219 				neg = 1;
2220 				p++;
2221 			}
2222 			if (*p < '0' || *p > '9')
2223 				break;
2224 
2225 			lval = simple_strtoul(p, &p, 0);
2226 
2227 			len = p-buf;
2228 			if ((len < left) && *p && !isspace(*p))
2229 				break;
2230 			if (neg)
2231 				val = -val;
2232 			s += len;
2233 			left -= len;
2234 
2235 			if (conv(&neg, &lval, i, 1, data))
2236 				break;
2237 		} else {
2238 			p = buf;
2239 			if (!first)
2240 				*p++ = '\t';
2241 
2242 			if (conv(&neg, &lval, i, 0, data))
2243 				break;
2244 
2245 			sprintf(p, "%s%lu", neg ? "-" : "", lval);
2246 			len = strlen(buf);
2247 			if (len > left)
2248 				len = left;
2249 			if(copy_to_user(s, buf, len))
2250 				return -EFAULT;
2251 			left -= len;
2252 			s += len;
2253 		}
2254 	}
2255 
2256 	if (!write && !first && left) {
2257 		if(put_user('\n', s))
2258 			return -EFAULT;
2259 		left--, s++;
2260 	}
2261 	if (write) {
2262 		while (left) {
2263 			char c;
2264 			if (get_user(c, s++))
2265 				return -EFAULT;
2266 			if (!isspace(c))
2267 				break;
2268 			left--;
2269 		}
2270 	}
2271 	if (write && first)
2272 		return -EINVAL;
2273 	*lenp -= left;
2274 	*ppos += *lenp;
2275 	return 0;
2276 #undef TMPBUFLEN
2277 }
2278 
2279 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2280 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2281 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2282 			      int write, void *data),
2283 		  void *data)
2284 {
2285 	return __do_proc_dointvec(table->data, table, write, filp,
2286 			buffer, lenp, ppos, conv, data);
2287 }
2288 
2289 /**
2290  * proc_dointvec - read a vector of integers
2291  * @table: the sysctl table
2292  * @write: %TRUE if this is a write to the sysctl file
2293  * @filp: the file structure
2294  * @buffer: the user buffer
2295  * @lenp: the size of the user buffer
2296  * @ppos: file position
2297  *
2298  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2299  * values from/to the user buffer, treated as an ASCII string.
2300  *
2301  * Returns 0 on success.
2302  */
2303 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2304 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2305 {
2306     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2307 		    	    NULL,NULL);
2308 }
2309 
2310 /*
2311  * Taint values can only be increased
2312  * This means we can safely use a temporary.
2313  */
2314 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
2315 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2316 {
2317 	struct ctl_table t;
2318 	unsigned long tmptaint = get_taint();
2319 	int err;
2320 
2321 	if (write && !capable(CAP_SYS_ADMIN))
2322 		return -EPERM;
2323 
2324 	t = *table;
2325 	t.data = &tmptaint;
2326 	err = proc_doulongvec_minmax(&t, write, filp, buffer, lenp, ppos);
2327 	if (err < 0)
2328 		return err;
2329 
2330 	if (write) {
2331 		/*
2332 		 * Poor man's atomic or. Not worth adding a primitive
2333 		 * to everyone's atomic.h for this
2334 		 */
2335 		int i;
2336 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2337 			if ((tmptaint >> i) & 1)
2338 				add_taint(i);
2339 		}
2340 	}
2341 
2342 	return err;
2343 }
2344 
2345 struct do_proc_dointvec_minmax_conv_param {
2346 	int *min;
2347 	int *max;
2348 };
2349 
2350 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2351 					int *valp,
2352 					int write, void *data)
2353 {
2354 	struct do_proc_dointvec_minmax_conv_param *param = data;
2355 	if (write) {
2356 		int val = *negp ? -*lvalp : *lvalp;
2357 		if ((param->min && *param->min > val) ||
2358 		    (param->max && *param->max < val))
2359 			return -EINVAL;
2360 		*valp = val;
2361 	} else {
2362 		int val = *valp;
2363 		if (val < 0) {
2364 			*negp = -1;
2365 			*lvalp = (unsigned long)-val;
2366 		} else {
2367 			*negp = 0;
2368 			*lvalp = (unsigned long)val;
2369 		}
2370 	}
2371 	return 0;
2372 }
2373 
2374 /**
2375  * proc_dointvec_minmax - read a vector of integers with min/max values
2376  * @table: the sysctl table
2377  * @write: %TRUE if this is a write to the sysctl file
2378  * @filp: the file structure
2379  * @buffer: the user buffer
2380  * @lenp: the size of the user buffer
2381  * @ppos: file position
2382  *
2383  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2384  * values from/to the user buffer, treated as an ASCII string.
2385  *
2386  * This routine will ensure the values are within the range specified by
2387  * table->extra1 (min) and table->extra2 (max).
2388  *
2389  * Returns 0 on success.
2390  */
2391 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2392 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2393 {
2394 	struct do_proc_dointvec_minmax_conv_param param = {
2395 		.min = (int *) table->extra1,
2396 		.max = (int *) table->extra2,
2397 	};
2398 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2399 				do_proc_dointvec_minmax_conv, &param);
2400 }
2401 
2402 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2403 				     struct file *filp,
2404 				     void __user *buffer,
2405 				     size_t *lenp, loff_t *ppos,
2406 				     unsigned long convmul,
2407 				     unsigned long convdiv)
2408 {
2409 #define TMPBUFLEN 21
2410 	unsigned long *i, *min, *max, val;
2411 	int vleft, first=1, neg;
2412 	size_t len, left;
2413 	char buf[TMPBUFLEN], *p;
2414 	char __user *s = buffer;
2415 
2416 	if (!data || !table->maxlen || !*lenp ||
2417 	    (*ppos && !write)) {
2418 		*lenp = 0;
2419 		return 0;
2420 	}
2421 
2422 	i = (unsigned long *) data;
2423 	min = (unsigned long *) table->extra1;
2424 	max = (unsigned long *) table->extra2;
2425 	vleft = table->maxlen / sizeof(unsigned long);
2426 	left = *lenp;
2427 
2428 	for (; left && vleft--; i++, min++, max++, first=0) {
2429 		if (write) {
2430 			while (left) {
2431 				char c;
2432 				if (get_user(c, s))
2433 					return -EFAULT;
2434 				if (!isspace(c))
2435 					break;
2436 				left--;
2437 				s++;
2438 			}
2439 			if (!left)
2440 				break;
2441 			neg = 0;
2442 			len = left;
2443 			if (len > TMPBUFLEN-1)
2444 				len = TMPBUFLEN-1;
2445 			if (copy_from_user(buf, s, len))
2446 				return -EFAULT;
2447 			buf[len] = 0;
2448 			p = buf;
2449 			if (*p == '-' && left > 1) {
2450 				neg = 1;
2451 				p++;
2452 			}
2453 			if (*p < '0' || *p > '9')
2454 				break;
2455 			val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2456 			len = p-buf;
2457 			if ((len < left) && *p && !isspace(*p))
2458 				break;
2459 			if (neg)
2460 				val = -val;
2461 			s += len;
2462 			left -= len;
2463 
2464 			if(neg)
2465 				continue;
2466 			if ((min && val < *min) || (max && val > *max))
2467 				continue;
2468 			*i = val;
2469 		} else {
2470 			p = buf;
2471 			if (!first)
2472 				*p++ = '\t';
2473 			sprintf(p, "%lu", convdiv * (*i) / convmul);
2474 			len = strlen(buf);
2475 			if (len > left)
2476 				len = left;
2477 			if(copy_to_user(s, buf, len))
2478 				return -EFAULT;
2479 			left -= len;
2480 			s += len;
2481 		}
2482 	}
2483 
2484 	if (!write && !first && left) {
2485 		if(put_user('\n', s))
2486 			return -EFAULT;
2487 		left--, s++;
2488 	}
2489 	if (write) {
2490 		while (left) {
2491 			char c;
2492 			if (get_user(c, s++))
2493 				return -EFAULT;
2494 			if (!isspace(c))
2495 				break;
2496 			left--;
2497 		}
2498 	}
2499 	if (write && first)
2500 		return -EINVAL;
2501 	*lenp -= left;
2502 	*ppos += *lenp;
2503 	return 0;
2504 #undef TMPBUFLEN
2505 }
2506 
2507 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2508 				     struct file *filp,
2509 				     void __user *buffer,
2510 				     size_t *lenp, loff_t *ppos,
2511 				     unsigned long convmul,
2512 				     unsigned long convdiv)
2513 {
2514 	return __do_proc_doulongvec_minmax(table->data, table, write,
2515 			filp, buffer, lenp, ppos, convmul, convdiv);
2516 }
2517 
2518 /**
2519  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2520  * @table: the sysctl table
2521  * @write: %TRUE if this is a write to the sysctl file
2522  * @filp: the file structure
2523  * @buffer: the user buffer
2524  * @lenp: the size of the user buffer
2525  * @ppos: file position
2526  *
2527  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2528  * values from/to the user buffer, treated as an ASCII string.
2529  *
2530  * This routine will ensure the values are within the range specified by
2531  * table->extra1 (min) and table->extra2 (max).
2532  *
2533  * Returns 0 on success.
2534  */
2535 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2536 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2537 {
2538     return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2539 }
2540 
2541 /**
2542  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2543  * @table: the sysctl table
2544  * @write: %TRUE if this is a write to the sysctl file
2545  * @filp: the file structure
2546  * @buffer: the user buffer
2547  * @lenp: the size of the user buffer
2548  * @ppos: file position
2549  *
2550  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2551  * values from/to the user buffer, treated as an ASCII string. The values
2552  * are treated as milliseconds, and converted to jiffies when they are stored.
2553  *
2554  * This routine will ensure the values are within the range specified by
2555  * table->extra1 (min) and table->extra2 (max).
2556  *
2557  * Returns 0 on success.
2558  */
2559 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2560 				      struct file *filp,
2561 				      void __user *buffer,
2562 				      size_t *lenp, loff_t *ppos)
2563 {
2564     return do_proc_doulongvec_minmax(table, write, filp, buffer,
2565 				     lenp, ppos, HZ, 1000l);
2566 }
2567 
2568 
2569 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2570 					 int *valp,
2571 					 int write, void *data)
2572 {
2573 	if (write) {
2574 		if (*lvalp > LONG_MAX / HZ)
2575 			return 1;
2576 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2577 	} else {
2578 		int val = *valp;
2579 		unsigned long lval;
2580 		if (val < 0) {
2581 			*negp = -1;
2582 			lval = (unsigned long)-val;
2583 		} else {
2584 			*negp = 0;
2585 			lval = (unsigned long)val;
2586 		}
2587 		*lvalp = lval / HZ;
2588 	}
2589 	return 0;
2590 }
2591 
2592 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2593 						int *valp,
2594 						int write, void *data)
2595 {
2596 	if (write) {
2597 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2598 			return 1;
2599 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2600 	} else {
2601 		int val = *valp;
2602 		unsigned long lval;
2603 		if (val < 0) {
2604 			*negp = -1;
2605 			lval = (unsigned long)-val;
2606 		} else {
2607 			*negp = 0;
2608 			lval = (unsigned long)val;
2609 		}
2610 		*lvalp = jiffies_to_clock_t(lval);
2611 	}
2612 	return 0;
2613 }
2614 
2615 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2616 					    int *valp,
2617 					    int write, void *data)
2618 {
2619 	if (write) {
2620 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2621 	} else {
2622 		int val = *valp;
2623 		unsigned long lval;
2624 		if (val < 0) {
2625 			*negp = -1;
2626 			lval = (unsigned long)-val;
2627 		} else {
2628 			*negp = 0;
2629 			lval = (unsigned long)val;
2630 		}
2631 		*lvalp = jiffies_to_msecs(lval);
2632 	}
2633 	return 0;
2634 }
2635 
2636 /**
2637  * proc_dointvec_jiffies - read a vector of integers as seconds
2638  * @table: the sysctl table
2639  * @write: %TRUE if this is a write to the sysctl file
2640  * @filp: the file structure
2641  * @buffer: the user buffer
2642  * @lenp: the size of the user buffer
2643  * @ppos: file position
2644  *
2645  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2646  * values from/to the user buffer, treated as an ASCII string.
2647  * The values read are assumed to be in seconds, and are converted into
2648  * jiffies.
2649  *
2650  * Returns 0 on success.
2651  */
2652 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2653 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2654 {
2655     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2656 		    	    do_proc_dointvec_jiffies_conv,NULL);
2657 }
2658 
2659 /**
2660  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2661  * @table: the sysctl table
2662  * @write: %TRUE if this is a write to the sysctl file
2663  * @filp: the file structure
2664  * @buffer: the user buffer
2665  * @lenp: the size of the user buffer
2666  * @ppos: pointer to the file position
2667  *
2668  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2669  * values from/to the user buffer, treated as an ASCII string.
2670  * The values read are assumed to be in 1/USER_HZ seconds, and
2671  * are converted into jiffies.
2672  *
2673  * Returns 0 on success.
2674  */
2675 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2676 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2677 {
2678     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2679 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2680 }
2681 
2682 /**
2683  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2684  * @table: the sysctl table
2685  * @write: %TRUE if this is a write to the sysctl file
2686  * @filp: the file structure
2687  * @buffer: the user buffer
2688  * @lenp: the size of the user buffer
2689  * @ppos: file position
2690  * @ppos: the current position in the file
2691  *
2692  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2693  * values from/to the user buffer, treated as an ASCII string.
2694  * The values read are assumed to be in 1/1000 seconds, and
2695  * are converted into jiffies.
2696  *
2697  * Returns 0 on success.
2698  */
2699 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2700 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2701 {
2702 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2703 				do_proc_dointvec_ms_jiffies_conv, NULL);
2704 }
2705 
2706 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2707 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2708 {
2709 	struct pid *new_pid;
2710 	pid_t tmp;
2711 	int r;
2712 
2713 	tmp = pid_vnr(cad_pid);
2714 
2715 	r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2716 			       lenp, ppos, NULL, NULL);
2717 	if (r || !write)
2718 		return r;
2719 
2720 	new_pid = find_get_pid(tmp);
2721 	if (!new_pid)
2722 		return -ESRCH;
2723 
2724 	put_pid(xchg(&cad_pid, new_pid));
2725 	return 0;
2726 }
2727 
2728 #else /* CONFIG_PROC_FS */
2729 
2730 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2731 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2732 {
2733 	return -ENOSYS;
2734 }
2735 
2736 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2737 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2738 {
2739 	return -ENOSYS;
2740 }
2741 
2742 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2743 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2744 {
2745 	return -ENOSYS;
2746 }
2747 
2748 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2749 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2750 {
2751 	return -ENOSYS;
2752 }
2753 
2754 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2755 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2756 {
2757 	return -ENOSYS;
2758 }
2759 
2760 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2761 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2762 {
2763 	return -ENOSYS;
2764 }
2765 
2766 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2767 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2768 {
2769 	return -ENOSYS;
2770 }
2771 
2772 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2773 				      struct file *filp,
2774 				      void __user *buffer,
2775 				      size_t *lenp, loff_t *ppos)
2776 {
2777     return -ENOSYS;
2778 }
2779 
2780 
2781 #endif /* CONFIG_PROC_FS */
2782 
2783 
2784 #ifdef CONFIG_SYSCTL_SYSCALL
2785 /*
2786  * General sysctl support routines
2787  */
2788 
2789 /* The generic sysctl data routine (used if no strategy routine supplied) */
2790 int sysctl_data(struct ctl_table *table,
2791 		void __user *oldval, size_t __user *oldlenp,
2792 		void __user *newval, size_t newlen)
2793 {
2794 	size_t len;
2795 
2796 	/* Get out of I don't have a variable */
2797 	if (!table->data || !table->maxlen)
2798 		return -ENOTDIR;
2799 
2800 	if (oldval && oldlenp) {
2801 		if (get_user(len, oldlenp))
2802 			return -EFAULT;
2803 		if (len) {
2804 			if (len > table->maxlen)
2805 				len = table->maxlen;
2806 			if (copy_to_user(oldval, table->data, len))
2807 				return -EFAULT;
2808 			if (put_user(len, oldlenp))
2809 				return -EFAULT;
2810 		}
2811 	}
2812 
2813 	if (newval && newlen) {
2814 		if (newlen > table->maxlen)
2815 			newlen = table->maxlen;
2816 
2817 		if (copy_from_user(table->data, newval, newlen))
2818 			return -EFAULT;
2819 	}
2820 	return 1;
2821 }
2822 
2823 /* The generic string strategy routine: */
2824 int sysctl_string(struct ctl_table *table,
2825 		  void __user *oldval, size_t __user *oldlenp,
2826 		  void __user *newval, size_t newlen)
2827 {
2828 	if (!table->data || !table->maxlen)
2829 		return -ENOTDIR;
2830 
2831 	if (oldval && oldlenp) {
2832 		size_t bufsize;
2833 		if (get_user(bufsize, oldlenp))
2834 			return -EFAULT;
2835 		if (bufsize) {
2836 			size_t len = strlen(table->data), copied;
2837 
2838 			/* This shouldn't trigger for a well-formed sysctl */
2839 			if (len > table->maxlen)
2840 				len = table->maxlen;
2841 
2842 			/* Copy up to a max of bufsize-1 bytes of the string */
2843 			copied = (len >= bufsize) ? bufsize - 1 : len;
2844 
2845 			if (copy_to_user(oldval, table->data, copied) ||
2846 			    put_user(0, (char __user *)(oldval + copied)))
2847 				return -EFAULT;
2848 			if (put_user(len, oldlenp))
2849 				return -EFAULT;
2850 		}
2851 	}
2852 	if (newval && newlen) {
2853 		size_t len = newlen;
2854 		if (len > table->maxlen)
2855 			len = table->maxlen;
2856 		if(copy_from_user(table->data, newval, len))
2857 			return -EFAULT;
2858 		if (len == table->maxlen)
2859 			len--;
2860 		((char *) table->data)[len] = 0;
2861 	}
2862 	return 1;
2863 }
2864 
2865 /*
2866  * This function makes sure that all of the integers in the vector
2867  * are between the minimum and maximum values given in the arrays
2868  * table->extra1 and table->extra2, respectively.
2869  */
2870 int sysctl_intvec(struct ctl_table *table,
2871 		void __user *oldval, size_t __user *oldlenp,
2872 		void __user *newval, size_t newlen)
2873 {
2874 
2875 	if (newval && newlen) {
2876 		int __user *vec = (int __user *) newval;
2877 		int *min = (int *) table->extra1;
2878 		int *max = (int *) table->extra2;
2879 		size_t length;
2880 		int i;
2881 
2882 		if (newlen % sizeof(int) != 0)
2883 			return -EINVAL;
2884 
2885 		if (!table->extra1 && !table->extra2)
2886 			return 0;
2887 
2888 		if (newlen > table->maxlen)
2889 			newlen = table->maxlen;
2890 		length = newlen / sizeof(int);
2891 
2892 		for (i = 0; i < length; i++) {
2893 			int value;
2894 			if (get_user(value, vec + i))
2895 				return -EFAULT;
2896 			if (min && value < min[i])
2897 				return -EINVAL;
2898 			if (max && value > max[i])
2899 				return -EINVAL;
2900 		}
2901 	}
2902 	return 0;
2903 }
2904 
2905 /* Strategy function to convert jiffies to seconds */
2906 int sysctl_jiffies(struct ctl_table *table,
2907 		void __user *oldval, size_t __user *oldlenp,
2908 		void __user *newval, size_t newlen)
2909 {
2910 	if (oldval && oldlenp) {
2911 		size_t olen;
2912 
2913 		if (get_user(olen, oldlenp))
2914 			return -EFAULT;
2915 		if (olen) {
2916 			int val;
2917 
2918 			if (olen < sizeof(int))
2919 				return -EINVAL;
2920 
2921 			val = *(int *)(table->data) / HZ;
2922 			if (put_user(val, (int __user *)oldval))
2923 				return -EFAULT;
2924 			if (put_user(sizeof(int), oldlenp))
2925 				return -EFAULT;
2926 		}
2927 	}
2928 	if (newval && newlen) {
2929 		int new;
2930 		if (newlen != sizeof(int))
2931 			return -EINVAL;
2932 		if (get_user(new, (int __user *)newval))
2933 			return -EFAULT;
2934 		*(int *)(table->data) = new*HZ;
2935 	}
2936 	return 1;
2937 }
2938 
2939 /* Strategy function to convert jiffies to seconds */
2940 int sysctl_ms_jiffies(struct ctl_table *table,
2941 		void __user *oldval, size_t __user *oldlenp,
2942 		void __user *newval, size_t newlen)
2943 {
2944 	if (oldval && oldlenp) {
2945 		size_t olen;
2946 
2947 		if (get_user(olen, oldlenp))
2948 			return -EFAULT;
2949 		if (olen) {
2950 			int val;
2951 
2952 			if (olen < sizeof(int))
2953 				return -EINVAL;
2954 
2955 			val = jiffies_to_msecs(*(int *)(table->data));
2956 			if (put_user(val, (int __user *)oldval))
2957 				return -EFAULT;
2958 			if (put_user(sizeof(int), oldlenp))
2959 				return -EFAULT;
2960 		}
2961 	}
2962 	if (newval && newlen) {
2963 		int new;
2964 		if (newlen != sizeof(int))
2965 			return -EINVAL;
2966 		if (get_user(new, (int __user *)newval))
2967 			return -EFAULT;
2968 		*(int *)(table->data) = msecs_to_jiffies(new);
2969 	}
2970 	return 1;
2971 }
2972 
2973 
2974 
2975 #else /* CONFIG_SYSCTL_SYSCALL */
2976 
2977 
2978 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2979 {
2980 	struct __sysctl_args tmp;
2981 	int error;
2982 
2983 	if (copy_from_user(&tmp, args, sizeof(tmp)))
2984 		return -EFAULT;
2985 
2986 	error = deprecated_sysctl_warning(&tmp);
2987 
2988 	/* If no error reading the parameters then just -ENOSYS ... */
2989 	if (!error)
2990 		error = -ENOSYS;
2991 
2992 	return error;
2993 }
2994 
2995 int sysctl_data(struct ctl_table *table,
2996 		  void __user *oldval, size_t __user *oldlenp,
2997 		  void __user *newval, size_t newlen)
2998 {
2999 	return -ENOSYS;
3000 }
3001 
3002 int sysctl_string(struct ctl_table *table,
3003 		  void __user *oldval, size_t __user *oldlenp,
3004 		  void __user *newval, size_t newlen)
3005 {
3006 	return -ENOSYS;
3007 }
3008 
3009 int sysctl_intvec(struct ctl_table *table,
3010 		void __user *oldval, size_t __user *oldlenp,
3011 		void __user *newval, size_t newlen)
3012 {
3013 	return -ENOSYS;
3014 }
3015 
3016 int sysctl_jiffies(struct ctl_table *table,
3017 		void __user *oldval, size_t __user *oldlenp,
3018 		void __user *newval, size_t newlen)
3019 {
3020 	return -ENOSYS;
3021 }
3022 
3023 int sysctl_ms_jiffies(struct ctl_table *table,
3024 		void __user *oldval, size_t __user *oldlenp,
3025 		void __user *newval, size_t newlen)
3026 {
3027 	return -ENOSYS;
3028 }
3029 
3030 #endif /* CONFIG_SYSCTL_SYSCALL */
3031 
3032 static int deprecated_sysctl_warning(struct __sysctl_args *args)
3033 {
3034 	static int msg_count;
3035 	int name[CTL_MAXNAME];
3036 	int i;
3037 
3038 	/* Check args->nlen. */
3039 	if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
3040 		return -ENOTDIR;
3041 
3042 	/* Read in the sysctl name for better debug message logging */
3043 	for (i = 0; i < args->nlen; i++)
3044 		if (get_user(name[i], args->name + i))
3045 			return -EFAULT;
3046 
3047 	/* Ignore accesses to kernel.version */
3048 	if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
3049 		return 0;
3050 
3051 	if (msg_count < 5) {
3052 		msg_count++;
3053 		printk(KERN_INFO
3054 			"warning: process `%s' used the deprecated sysctl "
3055 			"system call with ", current->comm);
3056 		for (i = 0; i < args->nlen; i++)
3057 			printk("%d.", name[i]);
3058 		printk("\n");
3059 	}
3060 	return 0;
3061 }
3062 
3063 /*
3064  * No sense putting this after each symbol definition, twice,
3065  * exception granted :-)
3066  */
3067 EXPORT_SYMBOL(proc_dointvec);
3068 EXPORT_SYMBOL(proc_dointvec_jiffies);
3069 EXPORT_SYMBOL(proc_dointvec_minmax);
3070 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3071 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3072 EXPORT_SYMBOL(proc_dostring);
3073 EXPORT_SYMBOL(proc_doulongvec_minmax);
3074 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3075 EXPORT_SYMBOL(register_sysctl_table);
3076 EXPORT_SYMBOL(register_sysctl_paths);
3077 EXPORT_SYMBOL(sysctl_intvec);
3078 EXPORT_SYMBOL(sysctl_jiffies);
3079 EXPORT_SYMBOL(sysctl_ms_jiffies);
3080 EXPORT_SYMBOL(sysctl_string);
3081 EXPORT_SYMBOL(sysctl_data);
3082 EXPORT_SYMBOL(unregister_sysctl_table);
3083