xref: /linux-6.15/kernel/sysctl.c (revision 5b25ab29)
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 #include <linux/slow-work.h>
52 
53 #include <asm/uaccess.h>
54 #include <asm/processor.h>
55 
56 #ifdef CONFIG_X86
57 #include <asm/nmi.h>
58 #include <asm/stacktrace.h>
59 #include <asm/io.h>
60 #endif
61 
62 static int deprecated_sysctl_warning(struct __sysctl_args *args);
63 
64 #if defined(CONFIG_SYSCTL)
65 
66 /* External variables not in a header file. */
67 extern int C_A_D;
68 extern int print_fatal_signals;
69 extern int sysctl_overcommit_memory;
70 extern int sysctl_overcommit_ratio;
71 extern int sysctl_panic_on_oom;
72 extern int sysctl_oom_kill_allocating_task;
73 extern int sysctl_oom_dump_tasks;
74 extern int max_threads;
75 extern int core_uses_pid;
76 extern int suid_dumpable;
77 extern char core_pattern[];
78 extern int pid_max;
79 extern int min_free_kbytes;
80 extern int pid_max_min, pid_max_max;
81 extern int sysctl_drop_caches;
82 extern int percpu_pagelist_fraction;
83 extern int compat_log;
84 extern int latencytop_enabled;
85 extern int sysctl_nr_open_min, sysctl_nr_open_max;
86 #ifndef CONFIG_MMU
87 extern int sysctl_nr_trim_pages;
88 #endif
89 #ifdef CONFIG_RCU_TORTURE_TEST
90 extern int rcutorture_runnable;
91 #endif /* #ifdef CONFIG_RCU_TORTURE_TEST */
92 
93 /* Constants used for minimum and  maximum */
94 #ifdef CONFIG_DETECT_SOFTLOCKUP
95 static int sixty = 60;
96 static int neg_one = -1;
97 #endif
98 
99 static int zero;
100 static int __maybe_unused one = 1;
101 static int __maybe_unused two = 2;
102 static unsigned long one_ul = 1;
103 static int one_hundred = 100;
104 
105 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
106 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
107 
108 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
109 static int maxolduid = 65535;
110 static int minolduid;
111 static int min_percpu_pagelist_fract = 8;
112 
113 static int ngroups_max = NGROUPS_MAX;
114 
115 #ifdef CONFIG_MODULES
116 extern char modprobe_path[];
117 #endif
118 #ifdef CONFIG_CHR_DEV_SG
119 extern int sg_big_buff;
120 #endif
121 
122 #ifdef CONFIG_SPARC
123 #include <asm/system.h>
124 #endif
125 
126 #ifdef CONFIG_SPARC64
127 extern int sysctl_tsb_ratio;
128 #endif
129 
130 #ifdef __hppa__
131 extern int pwrsw_enabled;
132 extern int unaligned_enabled;
133 #endif
134 
135 #ifdef CONFIG_S390
136 #ifdef CONFIG_MATHEMU
137 extern int sysctl_ieee_emulation_warnings;
138 #endif
139 extern int sysctl_userprocess_debug;
140 extern int spin_retry;
141 #endif
142 
143 #ifdef CONFIG_BSD_PROCESS_ACCT
144 extern int acct_parm[];
145 #endif
146 
147 #ifdef CONFIG_IA64
148 extern int no_unaligned_warning;
149 extern int unaligned_dump_stack;
150 #endif
151 
152 #ifdef CONFIG_RT_MUTEXES
153 extern int max_lock_depth;
154 #endif
155 
156 #ifdef CONFIG_PROC_SYSCTL
157 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
158 		  void __user *buffer, size_t *lenp, loff_t *ppos);
159 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
160 			       void __user *buffer, size_t *lenp, loff_t *ppos);
161 #endif
162 
163 static struct ctl_table root_table[];
164 static struct ctl_table_root sysctl_table_root;
165 static struct ctl_table_header root_table_header = {
166 	.count = 1,
167 	.ctl_table = root_table,
168 	.ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
169 	.root = &sysctl_table_root,
170 	.set = &sysctl_table_root.default_set,
171 };
172 static struct ctl_table_root sysctl_table_root = {
173 	.root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
174 	.default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
175 };
176 
177 static struct ctl_table kern_table[];
178 static struct ctl_table vm_table[];
179 static struct ctl_table fs_table[];
180 static struct ctl_table debug_table[];
181 static struct ctl_table dev_table[];
182 extern struct ctl_table random_table[];
183 #ifdef CONFIG_INOTIFY_USER
184 extern struct ctl_table inotify_table[];
185 #endif
186 #ifdef CONFIG_EPOLL
187 extern struct ctl_table epoll_table[];
188 #endif
189 
190 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
191 int sysctl_legacy_va_layout;
192 #endif
193 
194 extern int prove_locking;
195 extern int lock_stat;
196 
197 /* The default sysctl tables: */
198 
199 static struct ctl_table root_table[] = {
200 	{
201 		.ctl_name	= CTL_KERN,
202 		.procname	= "kernel",
203 		.mode		= 0555,
204 		.child		= kern_table,
205 	},
206 	{
207 		.ctl_name	= CTL_VM,
208 		.procname	= "vm",
209 		.mode		= 0555,
210 		.child		= vm_table,
211 	},
212 	{
213 		.ctl_name	= CTL_FS,
214 		.procname	= "fs",
215 		.mode		= 0555,
216 		.child		= fs_table,
217 	},
218 	{
219 		.ctl_name	= CTL_DEBUG,
220 		.procname	= "debug",
221 		.mode		= 0555,
222 		.child		= debug_table,
223 	},
224 	{
225 		.ctl_name	= CTL_DEV,
226 		.procname	= "dev",
227 		.mode		= 0555,
228 		.child		= dev_table,
229 	},
230 /*
231  * NOTE: do not add new entries to this table unless you have read
232  * Documentation/sysctl/ctl_unnumbered.txt
233  */
234 	{ .ctl_name = 0 }
235 };
236 
237 #ifdef CONFIG_SCHED_DEBUG
238 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
239 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
240 static int min_wakeup_granularity_ns;			/* 0 usecs */
241 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
242 #endif
243 
244 static struct ctl_table kern_table[] = {
245 #ifdef CONFIG_SCHED_DEBUG
246 	{
247 		.ctl_name	= CTL_UNNUMBERED,
248 		.procname	= "sched_min_granularity_ns",
249 		.data		= &sysctl_sched_min_granularity,
250 		.maxlen		= sizeof(unsigned int),
251 		.mode		= 0644,
252 		.proc_handler	= &sched_nr_latency_handler,
253 		.strategy	= &sysctl_intvec,
254 		.extra1		= &min_sched_granularity_ns,
255 		.extra2		= &max_sched_granularity_ns,
256 	},
257 	{
258 		.ctl_name	= CTL_UNNUMBERED,
259 		.procname	= "sched_latency_ns",
260 		.data		= &sysctl_sched_latency,
261 		.maxlen		= sizeof(unsigned int),
262 		.mode		= 0644,
263 		.proc_handler	= &sched_nr_latency_handler,
264 		.strategy	= &sysctl_intvec,
265 		.extra1		= &min_sched_granularity_ns,
266 		.extra2		= &max_sched_granularity_ns,
267 	},
268 	{
269 		.ctl_name	= CTL_UNNUMBERED,
270 		.procname	= "sched_wakeup_granularity_ns",
271 		.data		= &sysctl_sched_wakeup_granularity,
272 		.maxlen		= sizeof(unsigned int),
273 		.mode		= 0644,
274 		.proc_handler	= &proc_dointvec_minmax,
275 		.strategy	= &sysctl_intvec,
276 		.extra1		= &min_wakeup_granularity_ns,
277 		.extra2		= &max_wakeup_granularity_ns,
278 	},
279 	{
280 		.ctl_name	= CTL_UNNUMBERED,
281 		.procname	= "sched_shares_ratelimit",
282 		.data		= &sysctl_sched_shares_ratelimit,
283 		.maxlen		= sizeof(unsigned int),
284 		.mode		= 0644,
285 		.proc_handler	= &proc_dointvec,
286 	},
287 	{
288 		.ctl_name	= CTL_UNNUMBERED,
289 		.procname	= "sched_shares_thresh",
290 		.data		= &sysctl_sched_shares_thresh,
291 		.maxlen		= sizeof(unsigned int),
292 		.mode		= 0644,
293 		.proc_handler	= &proc_dointvec_minmax,
294 		.strategy	= &sysctl_intvec,
295 		.extra1		= &zero,
296 	},
297 	{
298 		.ctl_name	= CTL_UNNUMBERED,
299 		.procname	= "sched_child_runs_first",
300 		.data		= &sysctl_sched_child_runs_first,
301 		.maxlen		= sizeof(unsigned int),
302 		.mode		= 0644,
303 		.proc_handler	= &proc_dointvec,
304 	},
305 	{
306 		.ctl_name	= CTL_UNNUMBERED,
307 		.procname	= "sched_features",
308 		.data		= &sysctl_sched_features,
309 		.maxlen		= sizeof(unsigned int),
310 		.mode		= 0644,
311 		.proc_handler	= &proc_dointvec,
312 	},
313 	{
314 		.ctl_name	= CTL_UNNUMBERED,
315 		.procname	= "sched_migration_cost",
316 		.data		= &sysctl_sched_migration_cost,
317 		.maxlen		= sizeof(unsigned int),
318 		.mode		= 0644,
319 		.proc_handler	= &proc_dointvec,
320 	},
321 	{
322 		.ctl_name	= CTL_UNNUMBERED,
323 		.procname	= "sched_nr_migrate",
324 		.data		= &sysctl_sched_nr_migrate,
325 		.maxlen		= sizeof(unsigned int),
326 		.mode		= 0644,
327 		.proc_handler	= &proc_dointvec,
328 	},
329 #endif
330 	{
331 		.ctl_name	= CTL_UNNUMBERED,
332 		.procname	= "sched_rt_period_us",
333 		.data		= &sysctl_sched_rt_period,
334 		.maxlen		= sizeof(unsigned int),
335 		.mode		= 0644,
336 		.proc_handler	= &sched_rt_handler,
337 	},
338 	{
339 		.ctl_name	= CTL_UNNUMBERED,
340 		.procname	= "sched_rt_runtime_us",
341 		.data		= &sysctl_sched_rt_runtime,
342 		.maxlen		= sizeof(int),
343 		.mode		= 0644,
344 		.proc_handler	= &sched_rt_handler,
345 	},
346 	{
347 		.ctl_name	= CTL_UNNUMBERED,
348 		.procname	= "sched_compat_yield",
349 		.data		= &sysctl_sched_compat_yield,
350 		.maxlen		= sizeof(unsigned int),
351 		.mode		= 0644,
352 		.proc_handler	= &proc_dointvec,
353 	},
354 #ifdef CONFIG_PROVE_LOCKING
355 	{
356 		.ctl_name	= CTL_UNNUMBERED,
357 		.procname	= "prove_locking",
358 		.data		= &prove_locking,
359 		.maxlen		= sizeof(int),
360 		.mode		= 0644,
361 		.proc_handler	= &proc_dointvec,
362 	},
363 #endif
364 #ifdef CONFIG_LOCK_STAT
365 	{
366 		.ctl_name	= CTL_UNNUMBERED,
367 		.procname	= "lock_stat",
368 		.data		= &lock_stat,
369 		.maxlen		= sizeof(int),
370 		.mode		= 0644,
371 		.proc_handler	= &proc_dointvec,
372 	},
373 #endif
374 	{
375 		.ctl_name	= KERN_PANIC,
376 		.procname	= "panic",
377 		.data		= &panic_timeout,
378 		.maxlen		= sizeof(int),
379 		.mode		= 0644,
380 		.proc_handler	= &proc_dointvec,
381 	},
382 	{
383 		.ctl_name	= KERN_CORE_USES_PID,
384 		.procname	= "core_uses_pid",
385 		.data		= &core_uses_pid,
386 		.maxlen		= sizeof(int),
387 		.mode		= 0644,
388 		.proc_handler	= &proc_dointvec,
389 	},
390 	{
391 		.ctl_name	= KERN_CORE_PATTERN,
392 		.procname	= "core_pattern",
393 		.data		= core_pattern,
394 		.maxlen		= CORENAME_MAX_SIZE,
395 		.mode		= 0644,
396 		.proc_handler	= &proc_dostring,
397 		.strategy	= &sysctl_string,
398 	},
399 #ifdef CONFIG_PROC_SYSCTL
400 	{
401 		.procname	= "tainted",
402 		.maxlen 	= sizeof(long),
403 		.mode		= 0644,
404 		.proc_handler	= &proc_taint,
405 	},
406 #endif
407 #ifdef CONFIG_LATENCYTOP
408 	{
409 		.procname	= "latencytop",
410 		.data		= &latencytop_enabled,
411 		.maxlen		= sizeof(int),
412 		.mode		= 0644,
413 		.proc_handler	= &proc_dointvec,
414 	},
415 #endif
416 #ifdef CONFIG_BLK_DEV_INITRD
417 	{
418 		.ctl_name	= KERN_REALROOTDEV,
419 		.procname	= "real-root-dev",
420 		.data		= &real_root_dev,
421 		.maxlen		= sizeof(int),
422 		.mode		= 0644,
423 		.proc_handler	= &proc_dointvec,
424 	},
425 #endif
426 	{
427 		.ctl_name	= CTL_UNNUMBERED,
428 		.procname	= "print-fatal-signals",
429 		.data		= &print_fatal_signals,
430 		.maxlen		= sizeof(int),
431 		.mode		= 0644,
432 		.proc_handler	= &proc_dointvec,
433 	},
434 #ifdef CONFIG_SPARC
435 	{
436 		.ctl_name	= KERN_SPARC_REBOOT,
437 		.procname	= "reboot-cmd",
438 		.data		= reboot_command,
439 		.maxlen		= 256,
440 		.mode		= 0644,
441 		.proc_handler	= &proc_dostring,
442 		.strategy	= &sysctl_string,
443 	},
444 	{
445 		.ctl_name	= KERN_SPARC_STOP_A,
446 		.procname	= "stop-a",
447 		.data		= &stop_a_enabled,
448 		.maxlen		= sizeof (int),
449 		.mode		= 0644,
450 		.proc_handler	= &proc_dointvec,
451 	},
452 	{
453 		.ctl_name	= KERN_SPARC_SCONS_PWROFF,
454 		.procname	= "scons-poweroff",
455 		.data		= &scons_pwroff,
456 		.maxlen		= sizeof (int),
457 		.mode		= 0644,
458 		.proc_handler	= &proc_dointvec,
459 	},
460 #endif
461 #ifdef CONFIG_SPARC64
462 	{
463 		.ctl_name	= CTL_UNNUMBERED,
464 		.procname	= "tsb-ratio",
465 		.data		= &sysctl_tsb_ratio,
466 		.maxlen		= sizeof (int),
467 		.mode		= 0644,
468 		.proc_handler	= &proc_dointvec,
469 	},
470 #endif
471 #ifdef __hppa__
472 	{
473 		.ctl_name	= KERN_HPPA_PWRSW,
474 		.procname	= "soft-power",
475 		.data		= &pwrsw_enabled,
476 		.maxlen		= sizeof (int),
477 	 	.mode		= 0644,
478 		.proc_handler	= &proc_dointvec,
479 	},
480 	{
481 		.ctl_name	= KERN_HPPA_UNALIGNED,
482 		.procname	= "unaligned-trap",
483 		.data		= &unaligned_enabled,
484 		.maxlen		= sizeof (int),
485 		.mode		= 0644,
486 		.proc_handler	= &proc_dointvec,
487 	},
488 #endif
489 	{
490 		.ctl_name	= KERN_CTLALTDEL,
491 		.procname	= "ctrl-alt-del",
492 		.data		= &C_A_D,
493 		.maxlen		= sizeof(int),
494 		.mode		= 0644,
495 		.proc_handler	= &proc_dointvec,
496 	},
497 #ifdef CONFIG_FUNCTION_TRACER
498 	{
499 		.ctl_name	= CTL_UNNUMBERED,
500 		.procname	= "ftrace_enabled",
501 		.data		= &ftrace_enabled,
502 		.maxlen		= sizeof(int),
503 		.mode		= 0644,
504 		.proc_handler	= &ftrace_enable_sysctl,
505 	},
506 #endif
507 #ifdef CONFIG_STACK_TRACER
508 	{
509 		.ctl_name	= CTL_UNNUMBERED,
510 		.procname	= "stack_tracer_enabled",
511 		.data		= &stack_tracer_enabled,
512 		.maxlen		= sizeof(int),
513 		.mode		= 0644,
514 		.proc_handler	= &stack_trace_sysctl,
515 	},
516 #endif
517 #ifdef CONFIG_TRACING
518 	{
519 		.ctl_name	= CTL_UNNUMBERED,
520 		.procname	= "ftrace_dump_on_oops",
521 		.data		= &ftrace_dump_on_oops,
522 		.maxlen		= sizeof(int),
523 		.mode		= 0644,
524 		.proc_handler	= &proc_dointvec,
525 	},
526 #endif
527 #ifdef CONFIG_MODULES
528 	{
529 		.ctl_name	= KERN_MODPROBE,
530 		.procname	= "modprobe",
531 		.data		= &modprobe_path,
532 		.maxlen		= KMOD_PATH_LEN,
533 		.mode		= 0644,
534 		.proc_handler	= &proc_dostring,
535 		.strategy	= &sysctl_string,
536 	},
537 #endif
538 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
539 	{
540 		.ctl_name	= KERN_HOTPLUG,
541 		.procname	= "hotplug",
542 		.data		= &uevent_helper,
543 		.maxlen		= UEVENT_HELPER_PATH_LEN,
544 		.mode		= 0644,
545 		.proc_handler	= &proc_dostring,
546 		.strategy	= &sysctl_string,
547 	},
548 #endif
549 #ifdef CONFIG_CHR_DEV_SG
550 	{
551 		.ctl_name	= KERN_SG_BIG_BUFF,
552 		.procname	= "sg-big-buff",
553 		.data		= &sg_big_buff,
554 		.maxlen		= sizeof (int),
555 		.mode		= 0444,
556 		.proc_handler	= &proc_dointvec,
557 	},
558 #endif
559 #ifdef CONFIG_BSD_PROCESS_ACCT
560 	{
561 		.ctl_name	= KERN_ACCT,
562 		.procname	= "acct",
563 		.data		= &acct_parm,
564 		.maxlen		= 3*sizeof(int),
565 		.mode		= 0644,
566 		.proc_handler	= &proc_dointvec,
567 	},
568 #endif
569 #ifdef CONFIG_MAGIC_SYSRQ
570 	{
571 		.ctl_name	= KERN_SYSRQ,
572 		.procname	= "sysrq",
573 		.data		= &__sysrq_enabled,
574 		.maxlen		= sizeof (int),
575 		.mode		= 0644,
576 		.proc_handler	= &proc_dointvec,
577 	},
578 #endif
579 #ifdef CONFIG_PROC_SYSCTL
580 	{
581 		.procname	= "cad_pid",
582 		.data		= NULL,
583 		.maxlen		= sizeof (int),
584 		.mode		= 0600,
585 		.proc_handler	= &proc_do_cad_pid,
586 	},
587 #endif
588 	{
589 		.ctl_name	= KERN_MAX_THREADS,
590 		.procname	= "threads-max",
591 		.data		= &max_threads,
592 		.maxlen		= sizeof(int),
593 		.mode		= 0644,
594 		.proc_handler	= &proc_dointvec,
595 	},
596 	{
597 		.ctl_name	= KERN_RANDOM,
598 		.procname	= "random",
599 		.mode		= 0555,
600 		.child		= random_table,
601 	},
602 	{
603 		.ctl_name	= KERN_OVERFLOWUID,
604 		.procname	= "overflowuid",
605 		.data		= &overflowuid,
606 		.maxlen		= sizeof(int),
607 		.mode		= 0644,
608 		.proc_handler	= &proc_dointvec_minmax,
609 		.strategy	= &sysctl_intvec,
610 		.extra1		= &minolduid,
611 		.extra2		= &maxolduid,
612 	},
613 	{
614 		.ctl_name	= KERN_OVERFLOWGID,
615 		.procname	= "overflowgid",
616 		.data		= &overflowgid,
617 		.maxlen		= sizeof(int),
618 		.mode		= 0644,
619 		.proc_handler	= &proc_dointvec_minmax,
620 		.strategy	= &sysctl_intvec,
621 		.extra1		= &minolduid,
622 		.extra2		= &maxolduid,
623 	},
624 #ifdef CONFIG_S390
625 #ifdef CONFIG_MATHEMU
626 	{
627 		.ctl_name	= KERN_IEEE_EMULATION_WARNINGS,
628 		.procname	= "ieee_emulation_warnings",
629 		.data		= &sysctl_ieee_emulation_warnings,
630 		.maxlen		= sizeof(int),
631 		.mode		= 0644,
632 		.proc_handler	= &proc_dointvec,
633 	},
634 #endif
635 	{
636 		.ctl_name	= KERN_S390_USER_DEBUG_LOGGING,
637 		.procname	= "userprocess_debug",
638 		.data		= &sysctl_userprocess_debug,
639 		.maxlen		= sizeof(int),
640 		.mode		= 0644,
641 		.proc_handler	= &proc_dointvec,
642 	},
643 #endif
644 	{
645 		.ctl_name	= KERN_PIDMAX,
646 		.procname	= "pid_max",
647 		.data		= &pid_max,
648 		.maxlen		= sizeof (int),
649 		.mode		= 0644,
650 		.proc_handler	= &proc_dointvec_minmax,
651 		.strategy	= sysctl_intvec,
652 		.extra1		= &pid_max_min,
653 		.extra2		= &pid_max_max,
654 	},
655 	{
656 		.ctl_name	= KERN_PANIC_ON_OOPS,
657 		.procname	= "panic_on_oops",
658 		.data		= &panic_on_oops,
659 		.maxlen		= sizeof(int),
660 		.mode		= 0644,
661 		.proc_handler	= &proc_dointvec,
662 	},
663 #if defined CONFIG_PRINTK
664 	{
665 		.ctl_name	= KERN_PRINTK,
666 		.procname	= "printk",
667 		.data		= &console_loglevel,
668 		.maxlen		= 4*sizeof(int),
669 		.mode		= 0644,
670 		.proc_handler	= &proc_dointvec,
671 	},
672 	{
673 		.ctl_name	= KERN_PRINTK_RATELIMIT,
674 		.procname	= "printk_ratelimit",
675 		.data		= &printk_ratelimit_state.interval,
676 		.maxlen		= sizeof(int),
677 		.mode		= 0644,
678 		.proc_handler	= &proc_dointvec_jiffies,
679 		.strategy	= &sysctl_jiffies,
680 	},
681 	{
682 		.ctl_name	= KERN_PRINTK_RATELIMIT_BURST,
683 		.procname	= "printk_ratelimit_burst",
684 		.data		= &printk_ratelimit_state.burst,
685 		.maxlen		= sizeof(int),
686 		.mode		= 0644,
687 		.proc_handler	= &proc_dointvec,
688 	},
689 #endif
690 	{
691 		.ctl_name	= KERN_NGROUPS_MAX,
692 		.procname	= "ngroups_max",
693 		.data		= &ngroups_max,
694 		.maxlen		= sizeof (int),
695 		.mode		= 0444,
696 		.proc_handler	= &proc_dointvec,
697 	},
698 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
699 	{
700 		.ctl_name       = KERN_UNKNOWN_NMI_PANIC,
701 		.procname       = "unknown_nmi_panic",
702 		.data           = &unknown_nmi_panic,
703 		.maxlen         = sizeof (int),
704 		.mode           = 0644,
705 		.proc_handler   = &proc_dointvec,
706 	},
707 	{
708 		.procname       = "nmi_watchdog",
709 		.data           = &nmi_watchdog_enabled,
710 		.maxlen         = sizeof (int),
711 		.mode           = 0644,
712 		.proc_handler   = &proc_nmi_enabled,
713 	},
714 #endif
715 #if defined(CONFIG_X86)
716 	{
717 		.ctl_name	= KERN_PANIC_ON_NMI,
718 		.procname	= "panic_on_unrecovered_nmi",
719 		.data		= &panic_on_unrecovered_nmi,
720 		.maxlen		= sizeof(int),
721 		.mode		= 0644,
722 		.proc_handler	= &proc_dointvec,
723 	},
724 	{
725 		.ctl_name	= KERN_BOOTLOADER_TYPE,
726 		.procname	= "bootloader_type",
727 		.data		= &bootloader_type,
728 		.maxlen		= sizeof (int),
729 		.mode		= 0444,
730 		.proc_handler	= &proc_dointvec,
731 	},
732 	{
733 		.ctl_name	= CTL_UNNUMBERED,
734 		.procname	= "kstack_depth_to_print",
735 		.data		= &kstack_depth_to_print,
736 		.maxlen		= sizeof(int),
737 		.mode		= 0644,
738 		.proc_handler	= &proc_dointvec,
739 	},
740 	{
741 		.ctl_name	= CTL_UNNUMBERED,
742 		.procname	= "io_delay_type",
743 		.data		= &io_delay_type,
744 		.maxlen		= sizeof(int),
745 		.mode		= 0644,
746 		.proc_handler	= &proc_dointvec,
747 	},
748 #endif
749 #if defined(CONFIG_MMU)
750 	{
751 		.ctl_name	= KERN_RANDOMIZE,
752 		.procname	= "randomize_va_space",
753 		.data		= &randomize_va_space,
754 		.maxlen		= sizeof(int),
755 		.mode		= 0644,
756 		.proc_handler	= &proc_dointvec,
757 	},
758 #endif
759 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
760 	{
761 		.ctl_name	= KERN_SPIN_RETRY,
762 		.procname	= "spin_retry",
763 		.data		= &spin_retry,
764 		.maxlen		= sizeof (int),
765 		.mode		= 0644,
766 		.proc_handler	= &proc_dointvec,
767 	},
768 #endif
769 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
770 	{
771 		.procname	= "acpi_video_flags",
772 		.data		= &acpi_realmode_flags,
773 		.maxlen		= sizeof (unsigned long),
774 		.mode		= 0644,
775 		.proc_handler	= &proc_doulongvec_minmax,
776 	},
777 #endif
778 #ifdef CONFIG_IA64
779 	{
780 		.ctl_name	= KERN_IA64_UNALIGNED,
781 		.procname	= "ignore-unaligned-usertrap",
782 		.data		= &no_unaligned_warning,
783 		.maxlen		= sizeof (int),
784 	 	.mode		= 0644,
785 		.proc_handler	= &proc_dointvec,
786 	},
787 	{
788 		.ctl_name	= CTL_UNNUMBERED,
789 		.procname	= "unaligned-dump-stack",
790 		.data		= &unaligned_dump_stack,
791 		.maxlen		= sizeof (int),
792 		.mode		= 0644,
793 		.proc_handler	= &proc_dointvec,
794 	},
795 #endif
796 #ifdef CONFIG_DETECT_SOFTLOCKUP
797 	{
798 		.ctl_name	= CTL_UNNUMBERED,
799 		.procname	= "softlockup_panic",
800 		.data		= &softlockup_panic,
801 		.maxlen		= sizeof(int),
802 		.mode		= 0644,
803 		.proc_handler	= &proc_dointvec_minmax,
804 		.strategy	= &sysctl_intvec,
805 		.extra1		= &zero,
806 		.extra2		= &one,
807 	},
808 	{
809 		.ctl_name	= CTL_UNNUMBERED,
810 		.procname	= "softlockup_thresh",
811 		.data		= &softlockup_thresh,
812 		.maxlen		= sizeof(int),
813 		.mode		= 0644,
814 		.proc_handler	= &proc_dosoftlockup_thresh,
815 		.strategy	= &sysctl_intvec,
816 		.extra1		= &neg_one,
817 		.extra2		= &sixty,
818 	},
819 #endif
820 #ifdef CONFIG_DETECT_HUNG_TASK
821 	{
822 		.ctl_name	= CTL_UNNUMBERED,
823 		.procname	= "hung_task_panic",
824 		.data		= &sysctl_hung_task_panic,
825 		.maxlen		= sizeof(int),
826 		.mode		= 0644,
827 		.proc_handler	= &proc_dointvec_minmax,
828 		.strategy	= &sysctl_intvec,
829 		.extra1		= &zero,
830 		.extra2		= &one,
831 	},
832 	{
833 		.ctl_name	= CTL_UNNUMBERED,
834 		.procname	= "hung_task_check_count",
835 		.data		= &sysctl_hung_task_check_count,
836 		.maxlen		= sizeof(unsigned long),
837 		.mode		= 0644,
838 		.proc_handler	= &proc_doulongvec_minmax,
839 		.strategy	= &sysctl_intvec,
840 	},
841 	{
842 		.ctl_name	= CTL_UNNUMBERED,
843 		.procname	= "hung_task_timeout_secs",
844 		.data		= &sysctl_hung_task_timeout_secs,
845 		.maxlen		= sizeof(unsigned long),
846 		.mode		= 0644,
847 		.proc_handler	= &proc_dohung_task_timeout_secs,
848 		.strategy	= &sysctl_intvec,
849 	},
850 	{
851 		.ctl_name	= CTL_UNNUMBERED,
852 		.procname	= "hung_task_warnings",
853 		.data		= &sysctl_hung_task_warnings,
854 		.maxlen		= sizeof(unsigned long),
855 		.mode		= 0644,
856 		.proc_handler	= &proc_doulongvec_minmax,
857 		.strategy	= &sysctl_intvec,
858 	},
859 #endif
860 #ifdef CONFIG_COMPAT
861 	{
862 		.ctl_name	= KERN_COMPAT_LOG,
863 		.procname	= "compat-log",
864 		.data		= &compat_log,
865 		.maxlen		= sizeof (int),
866 	 	.mode		= 0644,
867 		.proc_handler	= &proc_dointvec,
868 	},
869 #endif
870 #ifdef CONFIG_RT_MUTEXES
871 	{
872 		.ctl_name	= KERN_MAX_LOCK_DEPTH,
873 		.procname	= "max_lock_depth",
874 		.data		= &max_lock_depth,
875 		.maxlen		= sizeof(int),
876 		.mode		= 0644,
877 		.proc_handler	= &proc_dointvec,
878 	},
879 #endif
880 	{
881 		.ctl_name	= CTL_UNNUMBERED,
882 		.procname	= "poweroff_cmd",
883 		.data		= &poweroff_cmd,
884 		.maxlen		= POWEROFF_CMD_PATH_LEN,
885 		.mode		= 0644,
886 		.proc_handler	= &proc_dostring,
887 		.strategy	= &sysctl_string,
888 	},
889 #ifdef CONFIG_KEYS
890 	{
891 		.ctl_name	= CTL_UNNUMBERED,
892 		.procname	= "keys",
893 		.mode		= 0555,
894 		.child		= key_sysctls,
895 	},
896 #endif
897 #ifdef CONFIG_RCU_TORTURE_TEST
898 	{
899 		.ctl_name       = CTL_UNNUMBERED,
900 		.procname       = "rcutorture_runnable",
901 		.data           = &rcutorture_runnable,
902 		.maxlen         = sizeof(int),
903 		.mode           = 0644,
904 		.proc_handler   = &proc_dointvec,
905 	},
906 #endif
907 #ifdef CONFIG_SLOW_WORK
908 	{
909 		.ctl_name	= CTL_UNNUMBERED,
910 		.procname	= "slow-work",
911 		.mode		= 0555,
912 		.child		= slow_work_sysctls,
913 	},
914 #endif
915 /*
916  * NOTE: do not add new entries to this table unless you have read
917  * Documentation/sysctl/ctl_unnumbered.txt
918  */
919 	{ .ctl_name = 0 }
920 };
921 
922 static struct ctl_table vm_table[] = {
923 	{
924 		.ctl_name	= VM_OVERCOMMIT_MEMORY,
925 		.procname	= "overcommit_memory",
926 		.data		= &sysctl_overcommit_memory,
927 		.maxlen		= sizeof(sysctl_overcommit_memory),
928 		.mode		= 0644,
929 		.proc_handler	= &proc_dointvec,
930 	},
931 	{
932 		.ctl_name	= VM_PANIC_ON_OOM,
933 		.procname	= "panic_on_oom",
934 		.data		= &sysctl_panic_on_oom,
935 		.maxlen		= sizeof(sysctl_panic_on_oom),
936 		.mode		= 0644,
937 		.proc_handler	= &proc_dointvec,
938 	},
939 	{
940 		.ctl_name	= CTL_UNNUMBERED,
941 		.procname	= "oom_kill_allocating_task",
942 		.data		= &sysctl_oom_kill_allocating_task,
943 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
944 		.mode		= 0644,
945 		.proc_handler	= &proc_dointvec,
946 	},
947 	{
948 		.ctl_name	= CTL_UNNUMBERED,
949 		.procname	= "oom_dump_tasks",
950 		.data		= &sysctl_oom_dump_tasks,
951 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
952 		.mode		= 0644,
953 		.proc_handler	= &proc_dointvec,
954 	},
955 	{
956 		.ctl_name	= VM_OVERCOMMIT_RATIO,
957 		.procname	= "overcommit_ratio",
958 		.data		= &sysctl_overcommit_ratio,
959 		.maxlen		= sizeof(sysctl_overcommit_ratio),
960 		.mode		= 0644,
961 		.proc_handler	= &proc_dointvec,
962 	},
963 	{
964 		.ctl_name	= VM_PAGE_CLUSTER,
965 		.procname	= "page-cluster",
966 		.data		= &page_cluster,
967 		.maxlen		= sizeof(int),
968 		.mode		= 0644,
969 		.proc_handler	= &proc_dointvec,
970 	},
971 	{
972 		.ctl_name	= VM_DIRTY_BACKGROUND,
973 		.procname	= "dirty_background_ratio",
974 		.data		= &dirty_background_ratio,
975 		.maxlen		= sizeof(dirty_background_ratio),
976 		.mode		= 0644,
977 		.proc_handler	= &dirty_background_ratio_handler,
978 		.strategy	= &sysctl_intvec,
979 		.extra1		= &zero,
980 		.extra2		= &one_hundred,
981 	},
982 	{
983 		.ctl_name	= CTL_UNNUMBERED,
984 		.procname	= "dirty_background_bytes",
985 		.data		= &dirty_background_bytes,
986 		.maxlen		= sizeof(dirty_background_bytes),
987 		.mode		= 0644,
988 		.proc_handler	= &dirty_background_bytes_handler,
989 		.strategy	= &sysctl_intvec,
990 		.extra1		= &one_ul,
991 	},
992 	{
993 		.ctl_name	= VM_DIRTY_RATIO,
994 		.procname	= "dirty_ratio",
995 		.data		= &vm_dirty_ratio,
996 		.maxlen		= sizeof(vm_dirty_ratio),
997 		.mode		= 0644,
998 		.proc_handler	= &dirty_ratio_handler,
999 		.strategy	= &sysctl_intvec,
1000 		.extra1		= &zero,
1001 		.extra2		= &one_hundred,
1002 	},
1003 	{
1004 		.ctl_name	= CTL_UNNUMBERED,
1005 		.procname	= "dirty_bytes",
1006 		.data		= &vm_dirty_bytes,
1007 		.maxlen		= sizeof(vm_dirty_bytes),
1008 		.mode		= 0644,
1009 		.proc_handler	= &dirty_bytes_handler,
1010 		.strategy	= &sysctl_intvec,
1011 		.extra1		= &dirty_bytes_min,
1012 	},
1013 	{
1014 		.procname	= "dirty_writeback_centisecs",
1015 		.data		= &dirty_writeback_interval,
1016 		.maxlen		= sizeof(dirty_writeback_interval),
1017 		.mode		= 0644,
1018 		.proc_handler	= &dirty_writeback_centisecs_handler,
1019 	},
1020 	{
1021 		.procname	= "dirty_expire_centisecs",
1022 		.data		= &dirty_expire_interval,
1023 		.maxlen		= sizeof(dirty_expire_interval),
1024 		.mode		= 0644,
1025 		.proc_handler	= &proc_dointvec,
1026 	},
1027 	{
1028 		.ctl_name	= VM_NR_PDFLUSH_THREADS,
1029 		.procname	= "nr_pdflush_threads",
1030 		.data		= &nr_pdflush_threads,
1031 		.maxlen		= sizeof nr_pdflush_threads,
1032 		.mode		= 0444 /* read-only*/,
1033 		.proc_handler	= &proc_dointvec,
1034 	},
1035 	{
1036 		.ctl_name	= VM_SWAPPINESS,
1037 		.procname	= "swappiness",
1038 		.data		= &vm_swappiness,
1039 		.maxlen		= sizeof(vm_swappiness),
1040 		.mode		= 0644,
1041 		.proc_handler	= &proc_dointvec_minmax,
1042 		.strategy	= &sysctl_intvec,
1043 		.extra1		= &zero,
1044 		.extra2		= &one_hundred,
1045 	},
1046 #ifdef CONFIG_HUGETLB_PAGE
1047 	 {
1048 		.procname	= "nr_hugepages",
1049 		.data		= NULL,
1050 		.maxlen		= sizeof(unsigned long),
1051 		.mode		= 0644,
1052 		.proc_handler	= &hugetlb_sysctl_handler,
1053 		.extra1		= (void *)&hugetlb_zero,
1054 		.extra2		= (void *)&hugetlb_infinity,
1055 	 },
1056 	 {
1057 		.ctl_name	= VM_HUGETLB_GROUP,
1058 		.procname	= "hugetlb_shm_group",
1059 		.data		= &sysctl_hugetlb_shm_group,
1060 		.maxlen		= sizeof(gid_t),
1061 		.mode		= 0644,
1062 		.proc_handler	= &proc_dointvec,
1063 	 },
1064 	 {
1065 		.ctl_name	= CTL_UNNUMBERED,
1066 		.procname	= "hugepages_treat_as_movable",
1067 		.data		= &hugepages_treat_as_movable,
1068 		.maxlen		= sizeof(int),
1069 		.mode		= 0644,
1070 		.proc_handler	= &hugetlb_treat_movable_handler,
1071 	},
1072 	{
1073 		.ctl_name	= CTL_UNNUMBERED,
1074 		.procname	= "nr_overcommit_hugepages",
1075 		.data		= NULL,
1076 		.maxlen		= sizeof(unsigned long),
1077 		.mode		= 0644,
1078 		.proc_handler	= &hugetlb_overcommit_handler,
1079 		.extra1		= (void *)&hugetlb_zero,
1080 		.extra2		= (void *)&hugetlb_infinity,
1081 	},
1082 #endif
1083 	{
1084 		.ctl_name	= VM_LOWMEM_RESERVE_RATIO,
1085 		.procname	= "lowmem_reserve_ratio",
1086 		.data		= &sysctl_lowmem_reserve_ratio,
1087 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1088 		.mode		= 0644,
1089 		.proc_handler	= &lowmem_reserve_ratio_sysctl_handler,
1090 		.strategy	= &sysctl_intvec,
1091 	},
1092 	{
1093 		.ctl_name	= VM_DROP_PAGECACHE,
1094 		.procname	= "drop_caches",
1095 		.data		= &sysctl_drop_caches,
1096 		.maxlen		= sizeof(int),
1097 		.mode		= 0644,
1098 		.proc_handler	= drop_caches_sysctl_handler,
1099 		.strategy	= &sysctl_intvec,
1100 	},
1101 	{
1102 		.ctl_name	= VM_MIN_FREE_KBYTES,
1103 		.procname	= "min_free_kbytes",
1104 		.data		= &min_free_kbytes,
1105 		.maxlen		= sizeof(min_free_kbytes),
1106 		.mode		= 0644,
1107 		.proc_handler	= &min_free_kbytes_sysctl_handler,
1108 		.strategy	= &sysctl_intvec,
1109 		.extra1		= &zero,
1110 	},
1111 	{
1112 		.ctl_name	= VM_PERCPU_PAGELIST_FRACTION,
1113 		.procname	= "percpu_pagelist_fraction",
1114 		.data		= &percpu_pagelist_fraction,
1115 		.maxlen		= sizeof(percpu_pagelist_fraction),
1116 		.mode		= 0644,
1117 		.proc_handler	= &percpu_pagelist_fraction_sysctl_handler,
1118 		.strategy	= &sysctl_intvec,
1119 		.extra1		= &min_percpu_pagelist_fract,
1120 	},
1121 #ifdef CONFIG_MMU
1122 	{
1123 		.ctl_name	= VM_MAX_MAP_COUNT,
1124 		.procname	= "max_map_count",
1125 		.data		= &sysctl_max_map_count,
1126 		.maxlen		= sizeof(sysctl_max_map_count),
1127 		.mode		= 0644,
1128 		.proc_handler	= &proc_dointvec
1129 	},
1130 #else
1131 	{
1132 		.ctl_name	= CTL_UNNUMBERED,
1133 		.procname	= "nr_trim_pages",
1134 		.data		= &sysctl_nr_trim_pages,
1135 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1136 		.mode		= 0644,
1137 		.proc_handler	= &proc_dointvec_minmax,
1138 		.strategy	= &sysctl_intvec,
1139 		.extra1		= &zero,
1140 	},
1141 #endif
1142 	{
1143 		.ctl_name	= VM_LAPTOP_MODE,
1144 		.procname	= "laptop_mode",
1145 		.data		= &laptop_mode,
1146 		.maxlen		= sizeof(laptop_mode),
1147 		.mode		= 0644,
1148 		.proc_handler	= &proc_dointvec_jiffies,
1149 		.strategy	= &sysctl_jiffies,
1150 	},
1151 	{
1152 		.ctl_name	= VM_BLOCK_DUMP,
1153 		.procname	= "block_dump",
1154 		.data		= &block_dump,
1155 		.maxlen		= sizeof(block_dump),
1156 		.mode		= 0644,
1157 		.proc_handler	= &proc_dointvec,
1158 		.strategy	= &sysctl_intvec,
1159 		.extra1		= &zero,
1160 	},
1161 	{
1162 		.ctl_name	= VM_VFS_CACHE_PRESSURE,
1163 		.procname	= "vfs_cache_pressure",
1164 		.data		= &sysctl_vfs_cache_pressure,
1165 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1166 		.mode		= 0644,
1167 		.proc_handler	= &proc_dointvec,
1168 		.strategy	= &sysctl_intvec,
1169 		.extra1		= &zero,
1170 	},
1171 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1172 	{
1173 		.ctl_name	= VM_LEGACY_VA_LAYOUT,
1174 		.procname	= "legacy_va_layout",
1175 		.data		= &sysctl_legacy_va_layout,
1176 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1177 		.mode		= 0644,
1178 		.proc_handler	= &proc_dointvec,
1179 		.strategy	= &sysctl_intvec,
1180 		.extra1		= &zero,
1181 	},
1182 #endif
1183 #ifdef CONFIG_NUMA
1184 	{
1185 		.ctl_name	= VM_ZONE_RECLAIM_MODE,
1186 		.procname	= "zone_reclaim_mode",
1187 		.data		= &zone_reclaim_mode,
1188 		.maxlen		= sizeof(zone_reclaim_mode),
1189 		.mode		= 0644,
1190 		.proc_handler	= &proc_dointvec,
1191 		.strategy	= &sysctl_intvec,
1192 		.extra1		= &zero,
1193 	},
1194 	{
1195 		.ctl_name	= VM_MIN_UNMAPPED,
1196 		.procname	= "min_unmapped_ratio",
1197 		.data		= &sysctl_min_unmapped_ratio,
1198 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1199 		.mode		= 0644,
1200 		.proc_handler	= &sysctl_min_unmapped_ratio_sysctl_handler,
1201 		.strategy	= &sysctl_intvec,
1202 		.extra1		= &zero,
1203 		.extra2		= &one_hundred,
1204 	},
1205 	{
1206 		.ctl_name	= VM_MIN_SLAB,
1207 		.procname	= "min_slab_ratio",
1208 		.data		= &sysctl_min_slab_ratio,
1209 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1210 		.mode		= 0644,
1211 		.proc_handler	= &sysctl_min_slab_ratio_sysctl_handler,
1212 		.strategy	= &sysctl_intvec,
1213 		.extra1		= &zero,
1214 		.extra2		= &one_hundred,
1215 	},
1216 #endif
1217 #ifdef CONFIG_SMP
1218 	{
1219 		.ctl_name	= CTL_UNNUMBERED,
1220 		.procname	= "stat_interval",
1221 		.data		= &sysctl_stat_interval,
1222 		.maxlen		= sizeof(sysctl_stat_interval),
1223 		.mode		= 0644,
1224 		.proc_handler	= &proc_dointvec_jiffies,
1225 		.strategy	= &sysctl_jiffies,
1226 	},
1227 #endif
1228 #ifdef CONFIG_SECURITY
1229 	{
1230 		.ctl_name	= CTL_UNNUMBERED,
1231 		.procname	= "mmap_min_addr",
1232 		.data		= &mmap_min_addr,
1233 		.maxlen         = sizeof(unsigned long),
1234 		.mode		= 0644,
1235 		.proc_handler	= &proc_doulongvec_minmax,
1236 	},
1237 #endif
1238 #ifdef CONFIG_NUMA
1239 	{
1240 		.ctl_name	= CTL_UNNUMBERED,
1241 		.procname	= "numa_zonelist_order",
1242 		.data		= &numa_zonelist_order,
1243 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1244 		.mode		= 0644,
1245 		.proc_handler	= &numa_zonelist_order_handler,
1246 		.strategy	= &sysctl_string,
1247 	},
1248 #endif
1249 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1250    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1251 	{
1252 		.ctl_name	= VM_VDSO_ENABLED,
1253 		.procname	= "vdso_enabled",
1254 		.data		= &vdso_enabled,
1255 		.maxlen		= sizeof(vdso_enabled),
1256 		.mode		= 0644,
1257 		.proc_handler	= &proc_dointvec,
1258 		.strategy	= &sysctl_intvec,
1259 		.extra1		= &zero,
1260 	},
1261 #endif
1262 #ifdef CONFIG_HIGHMEM
1263 	{
1264 		.ctl_name	= CTL_UNNUMBERED,
1265 		.procname	= "highmem_is_dirtyable",
1266 		.data		= &vm_highmem_is_dirtyable,
1267 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1268 		.mode		= 0644,
1269 		.proc_handler	= &proc_dointvec_minmax,
1270 		.strategy	= &sysctl_intvec,
1271 		.extra1		= &zero,
1272 		.extra2		= &one,
1273 	},
1274 #endif
1275 #ifdef CONFIG_UNEVICTABLE_LRU
1276 	{
1277 		.ctl_name	= CTL_UNNUMBERED,
1278 		.procname	= "scan_unevictable_pages",
1279 		.data		= &scan_unevictable_pages,
1280 		.maxlen		= sizeof(scan_unevictable_pages),
1281 		.mode		= 0644,
1282 		.proc_handler	= &scan_unevictable_handler,
1283 	},
1284 #endif
1285 /*
1286  * NOTE: do not add new entries to this table unless you have read
1287  * Documentation/sysctl/ctl_unnumbered.txt
1288  */
1289 	{ .ctl_name = 0 }
1290 };
1291 
1292 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1293 static struct ctl_table binfmt_misc_table[] = {
1294 	{ .ctl_name = 0 }
1295 };
1296 #endif
1297 
1298 static struct ctl_table fs_table[] = {
1299 	{
1300 		.ctl_name	= FS_NRINODE,
1301 		.procname	= "inode-nr",
1302 		.data		= &inodes_stat,
1303 		.maxlen		= 2*sizeof(int),
1304 		.mode		= 0444,
1305 		.proc_handler	= &proc_dointvec,
1306 	},
1307 	{
1308 		.ctl_name	= FS_STATINODE,
1309 		.procname	= "inode-state",
1310 		.data		= &inodes_stat,
1311 		.maxlen		= 7*sizeof(int),
1312 		.mode		= 0444,
1313 		.proc_handler	= &proc_dointvec,
1314 	},
1315 	{
1316 		.procname	= "file-nr",
1317 		.data		= &files_stat,
1318 		.maxlen		= 3*sizeof(int),
1319 		.mode		= 0444,
1320 		.proc_handler	= &proc_nr_files,
1321 	},
1322 	{
1323 		.ctl_name	= FS_MAXFILE,
1324 		.procname	= "file-max",
1325 		.data		= &files_stat.max_files,
1326 		.maxlen		= sizeof(int),
1327 		.mode		= 0644,
1328 		.proc_handler	= &proc_dointvec,
1329 	},
1330 	{
1331 		.ctl_name	= CTL_UNNUMBERED,
1332 		.procname	= "nr_open",
1333 		.data		= &sysctl_nr_open,
1334 		.maxlen		= sizeof(int),
1335 		.mode		= 0644,
1336 		.proc_handler	= &proc_dointvec_minmax,
1337 		.extra1		= &sysctl_nr_open_min,
1338 		.extra2		= &sysctl_nr_open_max,
1339 	},
1340 	{
1341 		.ctl_name	= FS_DENTRY,
1342 		.procname	= "dentry-state",
1343 		.data		= &dentry_stat,
1344 		.maxlen		= 6*sizeof(int),
1345 		.mode		= 0444,
1346 		.proc_handler	= &proc_dointvec,
1347 	},
1348 	{
1349 		.ctl_name	= FS_OVERFLOWUID,
1350 		.procname	= "overflowuid",
1351 		.data		= &fs_overflowuid,
1352 		.maxlen		= sizeof(int),
1353 		.mode		= 0644,
1354 		.proc_handler	= &proc_dointvec_minmax,
1355 		.strategy	= &sysctl_intvec,
1356 		.extra1		= &minolduid,
1357 		.extra2		= &maxolduid,
1358 	},
1359 	{
1360 		.ctl_name	= FS_OVERFLOWGID,
1361 		.procname	= "overflowgid",
1362 		.data		= &fs_overflowgid,
1363 		.maxlen		= sizeof(int),
1364 		.mode		= 0644,
1365 		.proc_handler	= &proc_dointvec_minmax,
1366 		.strategy	= &sysctl_intvec,
1367 		.extra1		= &minolduid,
1368 		.extra2		= &maxolduid,
1369 	},
1370 #ifdef CONFIG_FILE_LOCKING
1371 	{
1372 		.ctl_name	= FS_LEASES,
1373 		.procname	= "leases-enable",
1374 		.data		= &leases_enable,
1375 		.maxlen		= sizeof(int),
1376 		.mode		= 0644,
1377 		.proc_handler	= &proc_dointvec,
1378 	},
1379 #endif
1380 #ifdef CONFIG_DNOTIFY
1381 	{
1382 		.ctl_name	= FS_DIR_NOTIFY,
1383 		.procname	= "dir-notify-enable",
1384 		.data		= &dir_notify_enable,
1385 		.maxlen		= sizeof(int),
1386 		.mode		= 0644,
1387 		.proc_handler	= &proc_dointvec,
1388 	},
1389 #endif
1390 #ifdef CONFIG_MMU
1391 #ifdef CONFIG_FILE_LOCKING
1392 	{
1393 		.ctl_name	= FS_LEASE_TIME,
1394 		.procname	= "lease-break-time",
1395 		.data		= &lease_break_time,
1396 		.maxlen		= sizeof(int),
1397 		.mode		= 0644,
1398 		.proc_handler	= &proc_dointvec,
1399 	},
1400 #endif
1401 #ifdef CONFIG_AIO
1402 	{
1403 		.procname	= "aio-nr",
1404 		.data		= &aio_nr,
1405 		.maxlen		= sizeof(aio_nr),
1406 		.mode		= 0444,
1407 		.proc_handler	= &proc_doulongvec_minmax,
1408 	},
1409 	{
1410 		.procname	= "aio-max-nr",
1411 		.data		= &aio_max_nr,
1412 		.maxlen		= sizeof(aio_max_nr),
1413 		.mode		= 0644,
1414 		.proc_handler	= &proc_doulongvec_minmax,
1415 	},
1416 #endif /* CONFIG_AIO */
1417 #ifdef CONFIG_INOTIFY_USER
1418 	{
1419 		.ctl_name	= FS_INOTIFY,
1420 		.procname	= "inotify",
1421 		.mode		= 0555,
1422 		.child		= inotify_table,
1423 	},
1424 #endif
1425 #ifdef CONFIG_EPOLL
1426 	{
1427 		.procname	= "epoll",
1428 		.mode		= 0555,
1429 		.child		= epoll_table,
1430 	},
1431 #endif
1432 #endif
1433 	{
1434 		.ctl_name	= KERN_SETUID_DUMPABLE,
1435 		.procname	= "suid_dumpable",
1436 		.data		= &suid_dumpable,
1437 		.maxlen		= sizeof(int),
1438 		.mode		= 0644,
1439 		.proc_handler	= &proc_dointvec_minmax,
1440 		.strategy	= &sysctl_intvec,
1441 		.extra1		= &zero,
1442 		.extra2		= &two,
1443 	},
1444 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1445 	{
1446 		.ctl_name	= CTL_UNNUMBERED,
1447 		.procname	= "binfmt_misc",
1448 		.mode		= 0555,
1449 		.child		= binfmt_misc_table,
1450 	},
1451 #endif
1452 /*
1453  * NOTE: do not add new entries to this table unless you have read
1454  * Documentation/sysctl/ctl_unnumbered.txt
1455  */
1456 	{ .ctl_name = 0 }
1457 };
1458 
1459 static struct ctl_table debug_table[] = {
1460 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1461 	{
1462 		.ctl_name	= CTL_UNNUMBERED,
1463 		.procname	= "exception-trace",
1464 		.data		= &show_unhandled_signals,
1465 		.maxlen		= sizeof(int),
1466 		.mode		= 0644,
1467 		.proc_handler	= proc_dointvec
1468 	},
1469 #endif
1470 	{ .ctl_name = 0 }
1471 };
1472 
1473 static struct ctl_table dev_table[] = {
1474 	{ .ctl_name = 0 }
1475 };
1476 
1477 static DEFINE_SPINLOCK(sysctl_lock);
1478 
1479 /* called under sysctl_lock */
1480 static int use_table(struct ctl_table_header *p)
1481 {
1482 	if (unlikely(p->unregistering))
1483 		return 0;
1484 	p->used++;
1485 	return 1;
1486 }
1487 
1488 /* called under sysctl_lock */
1489 static void unuse_table(struct ctl_table_header *p)
1490 {
1491 	if (!--p->used)
1492 		if (unlikely(p->unregistering))
1493 			complete(p->unregistering);
1494 }
1495 
1496 /* called under sysctl_lock, will reacquire if has to wait */
1497 static void start_unregistering(struct ctl_table_header *p)
1498 {
1499 	/*
1500 	 * if p->used is 0, nobody will ever touch that entry again;
1501 	 * we'll eliminate all paths to it before dropping sysctl_lock
1502 	 */
1503 	if (unlikely(p->used)) {
1504 		struct completion wait;
1505 		init_completion(&wait);
1506 		p->unregistering = &wait;
1507 		spin_unlock(&sysctl_lock);
1508 		wait_for_completion(&wait);
1509 		spin_lock(&sysctl_lock);
1510 	} else {
1511 		/* anything non-NULL; we'll never dereference it */
1512 		p->unregistering = ERR_PTR(-EINVAL);
1513 	}
1514 	/*
1515 	 * do not remove from the list until nobody holds it; walking the
1516 	 * list in do_sysctl() relies on that.
1517 	 */
1518 	list_del_init(&p->ctl_entry);
1519 }
1520 
1521 void sysctl_head_get(struct ctl_table_header *head)
1522 {
1523 	spin_lock(&sysctl_lock);
1524 	head->count++;
1525 	spin_unlock(&sysctl_lock);
1526 }
1527 
1528 void sysctl_head_put(struct ctl_table_header *head)
1529 {
1530 	spin_lock(&sysctl_lock);
1531 	if (!--head->count)
1532 		kfree(head);
1533 	spin_unlock(&sysctl_lock);
1534 }
1535 
1536 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1537 {
1538 	if (!head)
1539 		BUG();
1540 	spin_lock(&sysctl_lock);
1541 	if (!use_table(head))
1542 		head = ERR_PTR(-ENOENT);
1543 	spin_unlock(&sysctl_lock);
1544 	return head;
1545 }
1546 
1547 void sysctl_head_finish(struct ctl_table_header *head)
1548 {
1549 	if (!head)
1550 		return;
1551 	spin_lock(&sysctl_lock);
1552 	unuse_table(head);
1553 	spin_unlock(&sysctl_lock);
1554 }
1555 
1556 static struct ctl_table_set *
1557 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1558 {
1559 	struct ctl_table_set *set = &root->default_set;
1560 	if (root->lookup)
1561 		set = root->lookup(root, namespaces);
1562 	return set;
1563 }
1564 
1565 static struct list_head *
1566 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1567 {
1568 	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1569 	return &set->list;
1570 }
1571 
1572 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1573 					    struct ctl_table_header *prev)
1574 {
1575 	struct ctl_table_root *root;
1576 	struct list_head *header_list;
1577 	struct ctl_table_header *head;
1578 	struct list_head *tmp;
1579 
1580 	spin_lock(&sysctl_lock);
1581 	if (prev) {
1582 		head = prev;
1583 		tmp = &prev->ctl_entry;
1584 		unuse_table(prev);
1585 		goto next;
1586 	}
1587 	tmp = &root_table_header.ctl_entry;
1588 	for (;;) {
1589 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1590 
1591 		if (!use_table(head))
1592 			goto next;
1593 		spin_unlock(&sysctl_lock);
1594 		return head;
1595 	next:
1596 		root = head->root;
1597 		tmp = tmp->next;
1598 		header_list = lookup_header_list(root, namespaces);
1599 		if (tmp != header_list)
1600 			continue;
1601 
1602 		do {
1603 			root = list_entry(root->root_list.next,
1604 					struct ctl_table_root, root_list);
1605 			if (root == &sysctl_table_root)
1606 				goto out;
1607 			header_list = lookup_header_list(root, namespaces);
1608 		} while (list_empty(header_list));
1609 		tmp = header_list->next;
1610 	}
1611 out:
1612 	spin_unlock(&sysctl_lock);
1613 	return NULL;
1614 }
1615 
1616 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1617 {
1618 	return __sysctl_head_next(current->nsproxy, prev);
1619 }
1620 
1621 void register_sysctl_root(struct ctl_table_root *root)
1622 {
1623 	spin_lock(&sysctl_lock);
1624 	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1625 	spin_unlock(&sysctl_lock);
1626 }
1627 
1628 #ifdef CONFIG_SYSCTL_SYSCALL
1629 /* Perform the actual read/write of a sysctl table entry. */
1630 static int do_sysctl_strategy(struct ctl_table_root *root,
1631 			struct ctl_table *table,
1632 			void __user *oldval, size_t __user *oldlenp,
1633 			void __user *newval, size_t newlen)
1634 {
1635 	int op = 0, rc;
1636 
1637 	if (oldval)
1638 		op |= MAY_READ;
1639 	if (newval)
1640 		op |= MAY_WRITE;
1641 	if (sysctl_perm(root, table, op))
1642 		return -EPERM;
1643 
1644 	if (table->strategy) {
1645 		rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1646 		if (rc < 0)
1647 			return rc;
1648 		if (rc > 0)
1649 			return 0;
1650 	}
1651 
1652 	/* If there is no strategy routine, or if the strategy returns
1653 	 * zero, proceed with automatic r/w */
1654 	if (table->data && table->maxlen) {
1655 		rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1656 		if (rc < 0)
1657 			return rc;
1658 	}
1659 	return 0;
1660 }
1661 
1662 static int parse_table(int __user *name, int nlen,
1663 		       void __user *oldval, size_t __user *oldlenp,
1664 		       void __user *newval, size_t newlen,
1665 		       struct ctl_table_root *root,
1666 		       struct ctl_table *table)
1667 {
1668 	int n;
1669 repeat:
1670 	if (!nlen)
1671 		return -ENOTDIR;
1672 	if (get_user(n, name))
1673 		return -EFAULT;
1674 	for ( ; table->ctl_name || table->procname; table++) {
1675 		if (!table->ctl_name)
1676 			continue;
1677 		if (n == table->ctl_name) {
1678 			int error;
1679 			if (table->child) {
1680 				if (sysctl_perm(root, table, MAY_EXEC))
1681 					return -EPERM;
1682 				name++;
1683 				nlen--;
1684 				table = table->child;
1685 				goto repeat;
1686 			}
1687 			error = do_sysctl_strategy(root, table,
1688 						   oldval, oldlenp,
1689 						   newval, newlen);
1690 			return error;
1691 		}
1692 	}
1693 	return -ENOTDIR;
1694 }
1695 
1696 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1697 	       void __user *newval, size_t newlen)
1698 {
1699 	struct ctl_table_header *head;
1700 	int error = -ENOTDIR;
1701 
1702 	if (nlen <= 0 || nlen >= CTL_MAXNAME)
1703 		return -ENOTDIR;
1704 	if (oldval) {
1705 		int old_len;
1706 		if (!oldlenp || get_user(old_len, oldlenp))
1707 			return -EFAULT;
1708 	}
1709 
1710 	for (head = sysctl_head_next(NULL); head;
1711 			head = sysctl_head_next(head)) {
1712 		error = parse_table(name, nlen, oldval, oldlenp,
1713 					newval, newlen,
1714 					head->root, head->ctl_table);
1715 		if (error != -ENOTDIR) {
1716 			sysctl_head_finish(head);
1717 			break;
1718 		}
1719 	}
1720 	return error;
1721 }
1722 
1723 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
1724 {
1725 	struct __sysctl_args tmp;
1726 	int error;
1727 
1728 	if (copy_from_user(&tmp, args, sizeof(tmp)))
1729 		return -EFAULT;
1730 
1731 	error = deprecated_sysctl_warning(&tmp);
1732 	if (error)
1733 		goto out;
1734 
1735 	lock_kernel();
1736 	error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1737 			  tmp.newval, tmp.newlen);
1738 	unlock_kernel();
1739 out:
1740 	return error;
1741 }
1742 #endif /* CONFIG_SYSCTL_SYSCALL */
1743 
1744 /*
1745  * sysctl_perm does NOT grant the superuser all rights automatically, because
1746  * some sysctl variables are readonly even to root.
1747  */
1748 
1749 static int test_perm(int mode, int op)
1750 {
1751 	if (!current_euid())
1752 		mode >>= 6;
1753 	else if (in_egroup_p(0))
1754 		mode >>= 3;
1755 	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1756 		return 0;
1757 	return -EACCES;
1758 }
1759 
1760 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1761 {
1762 	int error;
1763 	int mode;
1764 
1765 	error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1766 	if (error)
1767 		return error;
1768 
1769 	if (root->permissions)
1770 		mode = root->permissions(root, current->nsproxy, table);
1771 	else
1772 		mode = table->mode;
1773 
1774 	return test_perm(mode, op);
1775 }
1776 
1777 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1778 {
1779 	for (; table->ctl_name || table->procname; table++) {
1780 		table->parent = parent;
1781 		if (table->child)
1782 			sysctl_set_parent(table, table->child);
1783 	}
1784 }
1785 
1786 static __init int sysctl_init(void)
1787 {
1788 	sysctl_set_parent(NULL, root_table);
1789 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1790 	{
1791 		int err;
1792 		err = sysctl_check_table(current->nsproxy, root_table);
1793 	}
1794 #endif
1795 	return 0;
1796 }
1797 
1798 core_initcall(sysctl_init);
1799 
1800 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1801 				      struct ctl_table *table)
1802 {
1803 	struct ctl_table *p;
1804 	const char *s = branch->procname;
1805 
1806 	/* branch should have named subdirectory as its first element */
1807 	if (!s || !branch->child)
1808 		return NULL;
1809 
1810 	/* ... and nothing else */
1811 	if (branch[1].procname || branch[1].ctl_name)
1812 		return NULL;
1813 
1814 	/* table should contain subdirectory with the same name */
1815 	for (p = table; p->procname || p->ctl_name; p++) {
1816 		if (!p->child)
1817 			continue;
1818 		if (p->procname && strcmp(p->procname, s) == 0)
1819 			return p;
1820 	}
1821 	return NULL;
1822 }
1823 
1824 /* see if attaching q to p would be an improvement */
1825 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1826 {
1827 	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1828 	struct ctl_table *next;
1829 	int is_better = 0;
1830 	int not_in_parent = !p->attached_by;
1831 
1832 	while ((next = is_branch_in(by, to)) != NULL) {
1833 		if (by == q->attached_by)
1834 			is_better = 1;
1835 		if (to == p->attached_by)
1836 			not_in_parent = 1;
1837 		by = by->child;
1838 		to = next->child;
1839 	}
1840 
1841 	if (is_better && not_in_parent) {
1842 		q->attached_by = by;
1843 		q->attached_to = to;
1844 		q->parent = p;
1845 	}
1846 }
1847 
1848 /**
1849  * __register_sysctl_paths - register a sysctl hierarchy
1850  * @root: List of sysctl headers to register on
1851  * @namespaces: Data to compute which lists of sysctl entries are visible
1852  * @path: The path to the directory the sysctl table is in.
1853  * @table: the top-level table structure
1854  *
1855  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1856  * array. A completely 0 filled entry terminates the table.
1857  *
1858  * The members of the &struct ctl_table structure are used as follows:
1859  *
1860  * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1861  *            must be unique within that level of sysctl
1862  *
1863  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1864  *            enter a sysctl file
1865  *
1866  * data - a pointer to data for use by proc_handler
1867  *
1868  * maxlen - the maximum size in bytes of the data
1869  *
1870  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1871  *
1872  * child - a pointer to the child sysctl table if this entry is a directory, or
1873  *         %NULL.
1874  *
1875  * proc_handler - the text handler routine (described below)
1876  *
1877  * strategy - the strategy routine (described below)
1878  *
1879  * de - for internal use by the sysctl routines
1880  *
1881  * extra1, extra2 - extra pointers usable by the proc handler routines
1882  *
1883  * Leaf nodes in the sysctl tree will be represented by a single file
1884  * under /proc; non-leaf nodes will be represented by directories.
1885  *
1886  * sysctl(2) can automatically manage read and write requests through
1887  * the sysctl table.  The data and maxlen fields of the ctl_table
1888  * struct enable minimal validation of the values being written to be
1889  * performed, and the mode field allows minimal authentication.
1890  *
1891  * More sophisticated management can be enabled by the provision of a
1892  * strategy routine with the table entry.  This will be called before
1893  * any automatic read or write of the data is performed.
1894  *
1895  * The strategy routine may return
1896  *
1897  * < 0 - Error occurred (error is passed to user process)
1898  *
1899  * 0   - OK - proceed with automatic read or write.
1900  *
1901  * > 0 - OK - read or write has been done by the strategy routine, so
1902  *       return immediately.
1903  *
1904  * There must be a proc_handler routine for any terminal nodes
1905  * mirrored under /proc/sys (non-terminals are handled by a built-in
1906  * directory handler).  Several default handlers are available to
1907  * cover common cases -
1908  *
1909  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1910  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1911  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1912  *
1913  * It is the handler's job to read the input buffer from user memory
1914  * and process it. The handler should return 0 on success.
1915  *
1916  * This routine returns %NULL on a failure to register, and a pointer
1917  * to the table header on success.
1918  */
1919 struct ctl_table_header *__register_sysctl_paths(
1920 	struct ctl_table_root *root,
1921 	struct nsproxy *namespaces,
1922 	const struct ctl_path *path, struct ctl_table *table)
1923 {
1924 	struct ctl_table_header *header;
1925 	struct ctl_table *new, **prevp;
1926 	unsigned int n, npath;
1927 	struct ctl_table_set *set;
1928 
1929 	/* Count the path components */
1930 	for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1931 		;
1932 
1933 	/*
1934 	 * For each path component, allocate a 2-element ctl_table array.
1935 	 * The first array element will be filled with the sysctl entry
1936 	 * for this, the second will be the sentinel (ctl_name == 0).
1937 	 *
1938 	 * We allocate everything in one go so that we don't have to
1939 	 * worry about freeing additional memory in unregister_sysctl_table.
1940 	 */
1941 	header = kzalloc(sizeof(struct ctl_table_header) +
1942 			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1943 	if (!header)
1944 		return NULL;
1945 
1946 	new = (struct ctl_table *) (header + 1);
1947 
1948 	/* Now connect the dots */
1949 	prevp = &header->ctl_table;
1950 	for (n = 0; n < npath; ++n, ++path) {
1951 		/* Copy the procname */
1952 		new->procname = path->procname;
1953 		new->ctl_name = path->ctl_name;
1954 		new->mode     = 0555;
1955 
1956 		*prevp = new;
1957 		prevp = &new->child;
1958 
1959 		new += 2;
1960 	}
1961 	*prevp = table;
1962 	header->ctl_table_arg = table;
1963 
1964 	INIT_LIST_HEAD(&header->ctl_entry);
1965 	header->used = 0;
1966 	header->unregistering = NULL;
1967 	header->root = root;
1968 	sysctl_set_parent(NULL, header->ctl_table);
1969 	header->count = 1;
1970 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1971 	if (sysctl_check_table(namespaces, header->ctl_table)) {
1972 		kfree(header);
1973 		return NULL;
1974 	}
1975 #endif
1976 	spin_lock(&sysctl_lock);
1977 	header->set = lookup_header_set(root, namespaces);
1978 	header->attached_by = header->ctl_table;
1979 	header->attached_to = root_table;
1980 	header->parent = &root_table_header;
1981 	for (set = header->set; set; set = set->parent) {
1982 		struct ctl_table_header *p;
1983 		list_for_each_entry(p, &set->list, ctl_entry) {
1984 			if (p->unregistering)
1985 				continue;
1986 			try_attach(p, header);
1987 		}
1988 	}
1989 	header->parent->count++;
1990 	list_add_tail(&header->ctl_entry, &header->set->list);
1991 	spin_unlock(&sysctl_lock);
1992 
1993 	return header;
1994 }
1995 
1996 /**
1997  * register_sysctl_table_path - register a sysctl table hierarchy
1998  * @path: The path to the directory the sysctl table is in.
1999  * @table: the top-level table structure
2000  *
2001  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2002  * array. A completely 0 filled entry terminates the table.
2003  *
2004  * See __register_sysctl_paths for more details.
2005  */
2006 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2007 						struct ctl_table *table)
2008 {
2009 	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
2010 					path, table);
2011 }
2012 
2013 /**
2014  * register_sysctl_table - register a sysctl table hierarchy
2015  * @table: the top-level table structure
2016  *
2017  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2018  * array. A completely 0 filled entry terminates the table.
2019  *
2020  * See register_sysctl_paths for more details.
2021  */
2022 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
2023 {
2024 	static const struct ctl_path null_path[] = { {} };
2025 
2026 	return register_sysctl_paths(null_path, table);
2027 }
2028 
2029 /**
2030  * unregister_sysctl_table - unregister a sysctl table hierarchy
2031  * @header: the header returned from register_sysctl_table
2032  *
2033  * Unregisters the sysctl table and all children. proc entries may not
2034  * actually be removed until they are no longer used by anyone.
2035  */
2036 void unregister_sysctl_table(struct ctl_table_header * header)
2037 {
2038 	might_sleep();
2039 
2040 	if (header == NULL)
2041 		return;
2042 
2043 	spin_lock(&sysctl_lock);
2044 	start_unregistering(header);
2045 	if (!--header->parent->count) {
2046 		WARN_ON(1);
2047 		kfree(header->parent);
2048 	}
2049 	if (!--header->count)
2050 		kfree(header);
2051 	spin_unlock(&sysctl_lock);
2052 }
2053 
2054 int sysctl_is_seen(struct ctl_table_header *p)
2055 {
2056 	struct ctl_table_set *set = p->set;
2057 	int res;
2058 	spin_lock(&sysctl_lock);
2059 	if (p->unregistering)
2060 		res = 0;
2061 	else if (!set->is_seen)
2062 		res = 1;
2063 	else
2064 		res = set->is_seen(set);
2065 	spin_unlock(&sysctl_lock);
2066 	return res;
2067 }
2068 
2069 void setup_sysctl_set(struct ctl_table_set *p,
2070 	struct ctl_table_set *parent,
2071 	int (*is_seen)(struct ctl_table_set *))
2072 {
2073 	INIT_LIST_HEAD(&p->list);
2074 	p->parent = parent ? parent : &sysctl_table_root.default_set;
2075 	p->is_seen = is_seen;
2076 }
2077 
2078 #else /* !CONFIG_SYSCTL */
2079 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2080 {
2081 	return NULL;
2082 }
2083 
2084 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2085 						    struct ctl_table *table)
2086 {
2087 	return NULL;
2088 }
2089 
2090 void unregister_sysctl_table(struct ctl_table_header * table)
2091 {
2092 }
2093 
2094 void setup_sysctl_set(struct ctl_table_set *p,
2095 	struct ctl_table_set *parent,
2096 	int (*is_seen)(struct ctl_table_set *))
2097 {
2098 }
2099 
2100 void sysctl_head_put(struct ctl_table_header *head)
2101 {
2102 }
2103 
2104 #endif /* CONFIG_SYSCTL */
2105 
2106 /*
2107  * /proc/sys support
2108  */
2109 
2110 #ifdef CONFIG_PROC_SYSCTL
2111 
2112 static int _proc_do_string(void* data, int maxlen, int write,
2113 			   struct file *filp, void __user *buffer,
2114 			   size_t *lenp, loff_t *ppos)
2115 {
2116 	size_t len;
2117 	char __user *p;
2118 	char c;
2119 
2120 	if (!data || !maxlen || !*lenp) {
2121 		*lenp = 0;
2122 		return 0;
2123 	}
2124 
2125 	if (write) {
2126 		len = 0;
2127 		p = buffer;
2128 		while (len < *lenp) {
2129 			if (get_user(c, p++))
2130 				return -EFAULT;
2131 			if (c == 0 || c == '\n')
2132 				break;
2133 			len++;
2134 		}
2135 		if (len >= maxlen)
2136 			len = maxlen-1;
2137 		if(copy_from_user(data, buffer, len))
2138 			return -EFAULT;
2139 		((char *) data)[len] = 0;
2140 		*ppos += *lenp;
2141 	} else {
2142 		len = strlen(data);
2143 		if (len > maxlen)
2144 			len = maxlen;
2145 
2146 		if (*ppos > len) {
2147 			*lenp = 0;
2148 			return 0;
2149 		}
2150 
2151 		data += *ppos;
2152 		len  -= *ppos;
2153 
2154 		if (len > *lenp)
2155 			len = *lenp;
2156 		if (len)
2157 			if(copy_to_user(buffer, data, len))
2158 				return -EFAULT;
2159 		if (len < *lenp) {
2160 			if(put_user('\n', ((char __user *) buffer) + len))
2161 				return -EFAULT;
2162 			len++;
2163 		}
2164 		*lenp = len;
2165 		*ppos += len;
2166 	}
2167 	return 0;
2168 }
2169 
2170 /**
2171  * proc_dostring - read a string sysctl
2172  * @table: the sysctl table
2173  * @write: %TRUE if this is a write to the sysctl file
2174  * @filp: the file structure
2175  * @buffer: the user buffer
2176  * @lenp: the size of the user buffer
2177  * @ppos: file position
2178  *
2179  * Reads/writes a string from/to the user buffer. If the kernel
2180  * buffer provided is not large enough to hold the string, the
2181  * string is truncated. The copied string is %NULL-terminated.
2182  * If the string is being read by the user process, it is copied
2183  * and a newline '\n' is added. It is truncated if the buffer is
2184  * not large enough.
2185  *
2186  * Returns 0 on success.
2187  */
2188 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2189 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2190 {
2191 	return _proc_do_string(table->data, table->maxlen, write, filp,
2192 			       buffer, lenp, ppos);
2193 }
2194 
2195 
2196 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2197 				 int *valp,
2198 				 int write, void *data)
2199 {
2200 	if (write) {
2201 		*valp = *negp ? -*lvalp : *lvalp;
2202 	} else {
2203 		int val = *valp;
2204 		if (val < 0) {
2205 			*negp = -1;
2206 			*lvalp = (unsigned long)-val;
2207 		} else {
2208 			*negp = 0;
2209 			*lvalp = (unsigned long)val;
2210 		}
2211 	}
2212 	return 0;
2213 }
2214 
2215 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2216 		  int write, struct file *filp, void __user *buffer,
2217 		  size_t *lenp, loff_t *ppos,
2218 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2219 			      int write, void *data),
2220 		  void *data)
2221 {
2222 #define TMPBUFLEN 21
2223 	int *i, vleft, first=1, neg, val;
2224 	unsigned long lval;
2225 	size_t left, len;
2226 
2227 	char buf[TMPBUFLEN], *p;
2228 	char __user *s = buffer;
2229 
2230 	if (!tbl_data || !table->maxlen || !*lenp ||
2231 	    (*ppos && !write)) {
2232 		*lenp = 0;
2233 		return 0;
2234 	}
2235 
2236 	i = (int *) tbl_data;
2237 	vleft = table->maxlen / sizeof(*i);
2238 	left = *lenp;
2239 
2240 	if (!conv)
2241 		conv = do_proc_dointvec_conv;
2242 
2243 	for (; left && vleft--; i++, first=0) {
2244 		if (write) {
2245 			while (left) {
2246 				char c;
2247 				if (get_user(c, s))
2248 					return -EFAULT;
2249 				if (!isspace(c))
2250 					break;
2251 				left--;
2252 				s++;
2253 			}
2254 			if (!left)
2255 				break;
2256 			neg = 0;
2257 			len = left;
2258 			if (len > sizeof(buf) - 1)
2259 				len = sizeof(buf) - 1;
2260 			if (copy_from_user(buf, s, len))
2261 				return -EFAULT;
2262 			buf[len] = 0;
2263 			p = buf;
2264 			if (*p == '-' && left > 1) {
2265 				neg = 1;
2266 				p++;
2267 			}
2268 			if (*p < '0' || *p > '9')
2269 				break;
2270 
2271 			lval = simple_strtoul(p, &p, 0);
2272 
2273 			len = p-buf;
2274 			if ((len < left) && *p && !isspace(*p))
2275 				break;
2276 			if (neg)
2277 				val = -val;
2278 			s += len;
2279 			left -= len;
2280 
2281 			if (conv(&neg, &lval, i, 1, data))
2282 				break;
2283 		} else {
2284 			p = buf;
2285 			if (!first)
2286 				*p++ = '\t';
2287 
2288 			if (conv(&neg, &lval, i, 0, data))
2289 				break;
2290 
2291 			sprintf(p, "%s%lu", neg ? "-" : "", lval);
2292 			len = strlen(buf);
2293 			if (len > left)
2294 				len = left;
2295 			if(copy_to_user(s, buf, len))
2296 				return -EFAULT;
2297 			left -= len;
2298 			s += len;
2299 		}
2300 	}
2301 
2302 	if (!write && !first && left) {
2303 		if(put_user('\n', s))
2304 			return -EFAULT;
2305 		left--, s++;
2306 	}
2307 	if (write) {
2308 		while (left) {
2309 			char c;
2310 			if (get_user(c, s++))
2311 				return -EFAULT;
2312 			if (!isspace(c))
2313 				break;
2314 			left--;
2315 		}
2316 	}
2317 	if (write && first)
2318 		return -EINVAL;
2319 	*lenp -= left;
2320 	*ppos += *lenp;
2321 	return 0;
2322 #undef TMPBUFLEN
2323 }
2324 
2325 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2326 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2327 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2328 			      int write, void *data),
2329 		  void *data)
2330 {
2331 	return __do_proc_dointvec(table->data, table, write, filp,
2332 			buffer, lenp, ppos, conv, data);
2333 }
2334 
2335 /**
2336  * proc_dointvec - read a vector of integers
2337  * @table: the sysctl table
2338  * @write: %TRUE if this is a write to the sysctl file
2339  * @filp: the file structure
2340  * @buffer: the user buffer
2341  * @lenp: the size of the user buffer
2342  * @ppos: file position
2343  *
2344  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2345  * values from/to the user buffer, treated as an ASCII string.
2346  *
2347  * Returns 0 on success.
2348  */
2349 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2350 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2351 {
2352     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2353 		    	    NULL,NULL);
2354 }
2355 
2356 /*
2357  * Taint values can only be increased
2358  * This means we can safely use a temporary.
2359  */
2360 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
2361 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2362 {
2363 	struct ctl_table t;
2364 	unsigned long tmptaint = get_taint();
2365 	int err;
2366 
2367 	if (write && !capable(CAP_SYS_ADMIN))
2368 		return -EPERM;
2369 
2370 	t = *table;
2371 	t.data = &tmptaint;
2372 	err = proc_doulongvec_minmax(&t, write, filp, buffer, lenp, ppos);
2373 	if (err < 0)
2374 		return err;
2375 
2376 	if (write) {
2377 		/*
2378 		 * Poor man's atomic or. Not worth adding a primitive
2379 		 * to everyone's atomic.h for this
2380 		 */
2381 		int i;
2382 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2383 			if ((tmptaint >> i) & 1)
2384 				add_taint(i);
2385 		}
2386 	}
2387 
2388 	return err;
2389 }
2390 
2391 struct do_proc_dointvec_minmax_conv_param {
2392 	int *min;
2393 	int *max;
2394 };
2395 
2396 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2397 					int *valp,
2398 					int write, void *data)
2399 {
2400 	struct do_proc_dointvec_minmax_conv_param *param = data;
2401 	if (write) {
2402 		int val = *negp ? -*lvalp : *lvalp;
2403 		if ((param->min && *param->min > val) ||
2404 		    (param->max && *param->max < val))
2405 			return -EINVAL;
2406 		*valp = val;
2407 	} else {
2408 		int val = *valp;
2409 		if (val < 0) {
2410 			*negp = -1;
2411 			*lvalp = (unsigned long)-val;
2412 		} else {
2413 			*negp = 0;
2414 			*lvalp = (unsigned long)val;
2415 		}
2416 	}
2417 	return 0;
2418 }
2419 
2420 /**
2421  * proc_dointvec_minmax - read a vector of integers with min/max values
2422  * @table: the sysctl table
2423  * @write: %TRUE if this is a write to the sysctl file
2424  * @filp: the file structure
2425  * @buffer: the user buffer
2426  * @lenp: the size of the user buffer
2427  * @ppos: file position
2428  *
2429  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2430  * values from/to the user buffer, treated as an ASCII string.
2431  *
2432  * This routine will ensure the values are within the range specified by
2433  * table->extra1 (min) and table->extra2 (max).
2434  *
2435  * Returns 0 on success.
2436  */
2437 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2438 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2439 {
2440 	struct do_proc_dointvec_minmax_conv_param param = {
2441 		.min = (int *) table->extra1,
2442 		.max = (int *) table->extra2,
2443 	};
2444 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2445 				do_proc_dointvec_minmax_conv, &param);
2446 }
2447 
2448 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2449 				     struct file *filp,
2450 				     void __user *buffer,
2451 				     size_t *lenp, loff_t *ppos,
2452 				     unsigned long convmul,
2453 				     unsigned long convdiv)
2454 {
2455 #define TMPBUFLEN 21
2456 	unsigned long *i, *min, *max, val;
2457 	int vleft, first=1, neg;
2458 	size_t len, left;
2459 	char buf[TMPBUFLEN], *p;
2460 	char __user *s = buffer;
2461 
2462 	if (!data || !table->maxlen || !*lenp ||
2463 	    (*ppos && !write)) {
2464 		*lenp = 0;
2465 		return 0;
2466 	}
2467 
2468 	i = (unsigned long *) data;
2469 	min = (unsigned long *) table->extra1;
2470 	max = (unsigned long *) table->extra2;
2471 	vleft = table->maxlen / sizeof(unsigned long);
2472 	left = *lenp;
2473 
2474 	for (; left && vleft--; i++, min++, max++, first=0) {
2475 		if (write) {
2476 			while (left) {
2477 				char c;
2478 				if (get_user(c, s))
2479 					return -EFAULT;
2480 				if (!isspace(c))
2481 					break;
2482 				left--;
2483 				s++;
2484 			}
2485 			if (!left)
2486 				break;
2487 			neg = 0;
2488 			len = left;
2489 			if (len > TMPBUFLEN-1)
2490 				len = TMPBUFLEN-1;
2491 			if (copy_from_user(buf, s, len))
2492 				return -EFAULT;
2493 			buf[len] = 0;
2494 			p = buf;
2495 			if (*p == '-' && left > 1) {
2496 				neg = 1;
2497 				p++;
2498 			}
2499 			if (*p < '0' || *p > '9')
2500 				break;
2501 			val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2502 			len = p-buf;
2503 			if ((len < left) && *p && !isspace(*p))
2504 				break;
2505 			if (neg)
2506 				val = -val;
2507 			s += len;
2508 			left -= len;
2509 
2510 			if(neg)
2511 				continue;
2512 			if ((min && val < *min) || (max && val > *max))
2513 				continue;
2514 			*i = val;
2515 		} else {
2516 			p = buf;
2517 			if (!first)
2518 				*p++ = '\t';
2519 			sprintf(p, "%lu", convdiv * (*i) / convmul);
2520 			len = strlen(buf);
2521 			if (len > left)
2522 				len = left;
2523 			if(copy_to_user(s, buf, len))
2524 				return -EFAULT;
2525 			left -= len;
2526 			s += len;
2527 		}
2528 	}
2529 
2530 	if (!write && !first && left) {
2531 		if(put_user('\n', s))
2532 			return -EFAULT;
2533 		left--, s++;
2534 	}
2535 	if (write) {
2536 		while (left) {
2537 			char c;
2538 			if (get_user(c, s++))
2539 				return -EFAULT;
2540 			if (!isspace(c))
2541 				break;
2542 			left--;
2543 		}
2544 	}
2545 	if (write && first)
2546 		return -EINVAL;
2547 	*lenp -= left;
2548 	*ppos += *lenp;
2549 	return 0;
2550 #undef TMPBUFLEN
2551 }
2552 
2553 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2554 				     struct file *filp,
2555 				     void __user *buffer,
2556 				     size_t *lenp, loff_t *ppos,
2557 				     unsigned long convmul,
2558 				     unsigned long convdiv)
2559 {
2560 	return __do_proc_doulongvec_minmax(table->data, table, write,
2561 			filp, buffer, lenp, ppos, convmul, convdiv);
2562 }
2563 
2564 /**
2565  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2566  * @table: the sysctl table
2567  * @write: %TRUE if this is a write to the sysctl file
2568  * @filp: the file structure
2569  * @buffer: the user buffer
2570  * @lenp: the size of the user buffer
2571  * @ppos: file position
2572  *
2573  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2574  * values from/to the user buffer, treated as an ASCII string.
2575  *
2576  * This routine will ensure the values are within the range specified by
2577  * table->extra1 (min) and table->extra2 (max).
2578  *
2579  * Returns 0 on success.
2580  */
2581 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2582 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2583 {
2584     return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2585 }
2586 
2587 /**
2588  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2589  * @table: the sysctl table
2590  * @write: %TRUE if this is a write to the sysctl file
2591  * @filp: the file structure
2592  * @buffer: the user buffer
2593  * @lenp: the size of the user buffer
2594  * @ppos: file position
2595  *
2596  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2597  * values from/to the user buffer, treated as an ASCII string. The values
2598  * are treated as milliseconds, and converted to jiffies when they are stored.
2599  *
2600  * This routine will ensure the values are within the range specified by
2601  * table->extra1 (min) and table->extra2 (max).
2602  *
2603  * Returns 0 on success.
2604  */
2605 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2606 				      struct file *filp,
2607 				      void __user *buffer,
2608 				      size_t *lenp, loff_t *ppos)
2609 {
2610     return do_proc_doulongvec_minmax(table, write, filp, buffer,
2611 				     lenp, ppos, HZ, 1000l);
2612 }
2613 
2614 
2615 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2616 					 int *valp,
2617 					 int write, void *data)
2618 {
2619 	if (write) {
2620 		if (*lvalp > LONG_MAX / HZ)
2621 			return 1;
2622 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2623 	} else {
2624 		int val = *valp;
2625 		unsigned long lval;
2626 		if (val < 0) {
2627 			*negp = -1;
2628 			lval = (unsigned long)-val;
2629 		} else {
2630 			*negp = 0;
2631 			lval = (unsigned long)val;
2632 		}
2633 		*lvalp = lval / HZ;
2634 	}
2635 	return 0;
2636 }
2637 
2638 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2639 						int *valp,
2640 						int write, void *data)
2641 {
2642 	if (write) {
2643 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2644 			return 1;
2645 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2646 	} else {
2647 		int val = *valp;
2648 		unsigned long lval;
2649 		if (val < 0) {
2650 			*negp = -1;
2651 			lval = (unsigned long)-val;
2652 		} else {
2653 			*negp = 0;
2654 			lval = (unsigned long)val;
2655 		}
2656 		*lvalp = jiffies_to_clock_t(lval);
2657 	}
2658 	return 0;
2659 }
2660 
2661 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2662 					    int *valp,
2663 					    int write, void *data)
2664 {
2665 	if (write) {
2666 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2667 	} else {
2668 		int val = *valp;
2669 		unsigned long lval;
2670 		if (val < 0) {
2671 			*negp = -1;
2672 			lval = (unsigned long)-val;
2673 		} else {
2674 			*negp = 0;
2675 			lval = (unsigned long)val;
2676 		}
2677 		*lvalp = jiffies_to_msecs(lval);
2678 	}
2679 	return 0;
2680 }
2681 
2682 /**
2683  * proc_dointvec_jiffies - read a vector of integers as seconds
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  *
2691  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2692  * values from/to the user buffer, treated as an ASCII string.
2693  * The values read are assumed to be in seconds, and are converted into
2694  * jiffies.
2695  *
2696  * Returns 0 on success.
2697  */
2698 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2699 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2700 {
2701     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2702 		    	    do_proc_dointvec_jiffies_conv,NULL);
2703 }
2704 
2705 /**
2706  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2707  * @table: the sysctl table
2708  * @write: %TRUE if this is a write to the sysctl file
2709  * @filp: the file structure
2710  * @buffer: the user buffer
2711  * @lenp: the size of the user buffer
2712  * @ppos: pointer to the file position
2713  *
2714  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2715  * values from/to the user buffer, treated as an ASCII string.
2716  * The values read are assumed to be in 1/USER_HZ seconds, and
2717  * are converted into jiffies.
2718  *
2719  * Returns 0 on success.
2720  */
2721 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2722 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2723 {
2724     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2725 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2726 }
2727 
2728 /**
2729  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2730  * @table: the sysctl table
2731  * @write: %TRUE if this is a write to the sysctl file
2732  * @filp: the file structure
2733  * @buffer: the user buffer
2734  * @lenp: the size of the user buffer
2735  * @ppos: file position
2736  * @ppos: the current position in the file
2737  *
2738  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2739  * values from/to the user buffer, treated as an ASCII string.
2740  * The values read are assumed to be in 1/1000 seconds, and
2741  * are converted into jiffies.
2742  *
2743  * Returns 0 on success.
2744  */
2745 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2746 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2747 {
2748 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2749 				do_proc_dointvec_ms_jiffies_conv, NULL);
2750 }
2751 
2752 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2753 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2754 {
2755 	struct pid *new_pid;
2756 	pid_t tmp;
2757 	int r;
2758 
2759 	tmp = pid_vnr(cad_pid);
2760 
2761 	r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2762 			       lenp, ppos, NULL, NULL);
2763 	if (r || !write)
2764 		return r;
2765 
2766 	new_pid = find_get_pid(tmp);
2767 	if (!new_pid)
2768 		return -ESRCH;
2769 
2770 	put_pid(xchg(&cad_pid, new_pid));
2771 	return 0;
2772 }
2773 
2774 #else /* CONFIG_PROC_FS */
2775 
2776 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2777 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2778 {
2779 	return -ENOSYS;
2780 }
2781 
2782 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2783 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2784 {
2785 	return -ENOSYS;
2786 }
2787 
2788 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2789 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2790 {
2791 	return -ENOSYS;
2792 }
2793 
2794 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2795 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2796 {
2797 	return -ENOSYS;
2798 }
2799 
2800 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2801 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2802 {
2803 	return -ENOSYS;
2804 }
2805 
2806 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2807 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2808 {
2809 	return -ENOSYS;
2810 }
2811 
2812 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2813 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2814 {
2815 	return -ENOSYS;
2816 }
2817 
2818 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2819 				      struct file *filp,
2820 				      void __user *buffer,
2821 				      size_t *lenp, loff_t *ppos)
2822 {
2823     return -ENOSYS;
2824 }
2825 
2826 
2827 #endif /* CONFIG_PROC_FS */
2828 
2829 
2830 #ifdef CONFIG_SYSCTL_SYSCALL
2831 /*
2832  * General sysctl support routines
2833  */
2834 
2835 /* The generic sysctl data routine (used if no strategy routine supplied) */
2836 int sysctl_data(struct ctl_table *table,
2837 		void __user *oldval, size_t __user *oldlenp,
2838 		void __user *newval, size_t newlen)
2839 {
2840 	size_t len;
2841 
2842 	/* Get out of I don't have a variable */
2843 	if (!table->data || !table->maxlen)
2844 		return -ENOTDIR;
2845 
2846 	if (oldval && oldlenp) {
2847 		if (get_user(len, oldlenp))
2848 			return -EFAULT;
2849 		if (len) {
2850 			if (len > table->maxlen)
2851 				len = table->maxlen;
2852 			if (copy_to_user(oldval, table->data, len))
2853 				return -EFAULT;
2854 			if (put_user(len, oldlenp))
2855 				return -EFAULT;
2856 		}
2857 	}
2858 
2859 	if (newval && newlen) {
2860 		if (newlen > table->maxlen)
2861 			newlen = table->maxlen;
2862 
2863 		if (copy_from_user(table->data, newval, newlen))
2864 			return -EFAULT;
2865 	}
2866 	return 1;
2867 }
2868 
2869 /* The generic string strategy routine: */
2870 int sysctl_string(struct ctl_table *table,
2871 		  void __user *oldval, size_t __user *oldlenp,
2872 		  void __user *newval, size_t newlen)
2873 {
2874 	if (!table->data || !table->maxlen)
2875 		return -ENOTDIR;
2876 
2877 	if (oldval && oldlenp) {
2878 		size_t bufsize;
2879 		if (get_user(bufsize, oldlenp))
2880 			return -EFAULT;
2881 		if (bufsize) {
2882 			size_t len = strlen(table->data), copied;
2883 
2884 			/* This shouldn't trigger for a well-formed sysctl */
2885 			if (len > table->maxlen)
2886 				len = table->maxlen;
2887 
2888 			/* Copy up to a max of bufsize-1 bytes of the string */
2889 			copied = (len >= bufsize) ? bufsize - 1 : len;
2890 
2891 			if (copy_to_user(oldval, table->data, copied) ||
2892 			    put_user(0, (char __user *)(oldval + copied)))
2893 				return -EFAULT;
2894 			if (put_user(len, oldlenp))
2895 				return -EFAULT;
2896 		}
2897 	}
2898 	if (newval && newlen) {
2899 		size_t len = newlen;
2900 		if (len > table->maxlen)
2901 			len = table->maxlen;
2902 		if(copy_from_user(table->data, newval, len))
2903 			return -EFAULT;
2904 		if (len == table->maxlen)
2905 			len--;
2906 		((char *) table->data)[len] = 0;
2907 	}
2908 	return 1;
2909 }
2910 
2911 /*
2912  * This function makes sure that all of the integers in the vector
2913  * are between the minimum and maximum values given in the arrays
2914  * table->extra1 and table->extra2, respectively.
2915  */
2916 int sysctl_intvec(struct ctl_table *table,
2917 		void __user *oldval, size_t __user *oldlenp,
2918 		void __user *newval, size_t newlen)
2919 {
2920 
2921 	if (newval && newlen) {
2922 		int __user *vec = (int __user *) newval;
2923 		int *min = (int *) table->extra1;
2924 		int *max = (int *) table->extra2;
2925 		size_t length;
2926 		int i;
2927 
2928 		if (newlen % sizeof(int) != 0)
2929 			return -EINVAL;
2930 
2931 		if (!table->extra1 && !table->extra2)
2932 			return 0;
2933 
2934 		if (newlen > table->maxlen)
2935 			newlen = table->maxlen;
2936 		length = newlen / sizeof(int);
2937 
2938 		for (i = 0; i < length; i++) {
2939 			int value;
2940 			if (get_user(value, vec + i))
2941 				return -EFAULT;
2942 			if (min && value < min[i])
2943 				return -EINVAL;
2944 			if (max && value > max[i])
2945 				return -EINVAL;
2946 		}
2947 	}
2948 	return 0;
2949 }
2950 
2951 /* Strategy function to convert jiffies to seconds */
2952 int sysctl_jiffies(struct ctl_table *table,
2953 		void __user *oldval, size_t __user *oldlenp,
2954 		void __user *newval, size_t newlen)
2955 {
2956 	if (oldval && oldlenp) {
2957 		size_t olen;
2958 
2959 		if (get_user(olen, oldlenp))
2960 			return -EFAULT;
2961 		if (olen) {
2962 			int val;
2963 
2964 			if (olen < sizeof(int))
2965 				return -EINVAL;
2966 
2967 			val = *(int *)(table->data) / HZ;
2968 			if (put_user(val, (int __user *)oldval))
2969 				return -EFAULT;
2970 			if (put_user(sizeof(int), oldlenp))
2971 				return -EFAULT;
2972 		}
2973 	}
2974 	if (newval && newlen) {
2975 		int new;
2976 		if (newlen != sizeof(int))
2977 			return -EINVAL;
2978 		if (get_user(new, (int __user *)newval))
2979 			return -EFAULT;
2980 		*(int *)(table->data) = new*HZ;
2981 	}
2982 	return 1;
2983 }
2984 
2985 /* Strategy function to convert jiffies to seconds */
2986 int sysctl_ms_jiffies(struct ctl_table *table,
2987 		void __user *oldval, size_t __user *oldlenp,
2988 		void __user *newval, size_t newlen)
2989 {
2990 	if (oldval && oldlenp) {
2991 		size_t olen;
2992 
2993 		if (get_user(olen, oldlenp))
2994 			return -EFAULT;
2995 		if (olen) {
2996 			int val;
2997 
2998 			if (olen < sizeof(int))
2999 				return -EINVAL;
3000 
3001 			val = jiffies_to_msecs(*(int *)(table->data));
3002 			if (put_user(val, (int __user *)oldval))
3003 				return -EFAULT;
3004 			if (put_user(sizeof(int), oldlenp))
3005 				return -EFAULT;
3006 		}
3007 	}
3008 	if (newval && newlen) {
3009 		int new;
3010 		if (newlen != sizeof(int))
3011 			return -EINVAL;
3012 		if (get_user(new, (int __user *)newval))
3013 			return -EFAULT;
3014 		*(int *)(table->data) = msecs_to_jiffies(new);
3015 	}
3016 	return 1;
3017 }
3018 
3019 
3020 
3021 #else /* CONFIG_SYSCTL_SYSCALL */
3022 
3023 
3024 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
3025 {
3026 	struct __sysctl_args tmp;
3027 	int error;
3028 
3029 	if (copy_from_user(&tmp, args, sizeof(tmp)))
3030 		return -EFAULT;
3031 
3032 	error = deprecated_sysctl_warning(&tmp);
3033 
3034 	/* If no error reading the parameters then just -ENOSYS ... */
3035 	if (!error)
3036 		error = -ENOSYS;
3037 
3038 	return error;
3039 }
3040 
3041 int sysctl_data(struct ctl_table *table,
3042 		  void __user *oldval, size_t __user *oldlenp,
3043 		  void __user *newval, size_t newlen)
3044 {
3045 	return -ENOSYS;
3046 }
3047 
3048 int sysctl_string(struct ctl_table *table,
3049 		  void __user *oldval, size_t __user *oldlenp,
3050 		  void __user *newval, size_t newlen)
3051 {
3052 	return -ENOSYS;
3053 }
3054 
3055 int sysctl_intvec(struct ctl_table *table,
3056 		void __user *oldval, size_t __user *oldlenp,
3057 		void __user *newval, size_t newlen)
3058 {
3059 	return -ENOSYS;
3060 }
3061 
3062 int sysctl_jiffies(struct ctl_table *table,
3063 		void __user *oldval, size_t __user *oldlenp,
3064 		void __user *newval, size_t newlen)
3065 {
3066 	return -ENOSYS;
3067 }
3068 
3069 int sysctl_ms_jiffies(struct ctl_table *table,
3070 		void __user *oldval, size_t __user *oldlenp,
3071 		void __user *newval, size_t newlen)
3072 {
3073 	return -ENOSYS;
3074 }
3075 
3076 #endif /* CONFIG_SYSCTL_SYSCALL */
3077 
3078 static int deprecated_sysctl_warning(struct __sysctl_args *args)
3079 {
3080 	static int msg_count;
3081 	int name[CTL_MAXNAME];
3082 	int i;
3083 
3084 	/* Check args->nlen. */
3085 	if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
3086 		return -ENOTDIR;
3087 
3088 	/* Read in the sysctl name for better debug message logging */
3089 	for (i = 0; i < args->nlen; i++)
3090 		if (get_user(name[i], args->name + i))
3091 			return -EFAULT;
3092 
3093 	/* Ignore accesses to kernel.version */
3094 	if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
3095 		return 0;
3096 
3097 	if (msg_count < 5) {
3098 		msg_count++;
3099 		printk(KERN_INFO
3100 			"warning: process `%s' used the deprecated sysctl "
3101 			"system call with ", current->comm);
3102 		for (i = 0; i < args->nlen; i++)
3103 			printk("%d.", name[i]);
3104 		printk("\n");
3105 	}
3106 	return 0;
3107 }
3108 
3109 /*
3110  * No sense putting this after each symbol definition, twice,
3111  * exception granted :-)
3112  */
3113 EXPORT_SYMBOL(proc_dointvec);
3114 EXPORT_SYMBOL(proc_dointvec_jiffies);
3115 EXPORT_SYMBOL(proc_dointvec_minmax);
3116 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3117 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3118 EXPORT_SYMBOL(proc_dostring);
3119 EXPORT_SYMBOL(proc_doulongvec_minmax);
3120 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3121 EXPORT_SYMBOL(register_sysctl_table);
3122 EXPORT_SYMBOL(register_sysctl_paths);
3123 EXPORT_SYMBOL(sysctl_intvec);
3124 EXPORT_SYMBOL(sysctl_jiffies);
3125 EXPORT_SYMBOL(sysctl_ms_jiffies);
3126 EXPORT_SYMBOL(sysctl_string);
3127 EXPORT_SYMBOL(sysctl_data);
3128 EXPORT_SYMBOL(unregister_sysctl_table);
3129