xref: /linux-6.15/kernel/sysctl.c (revision 7fcab099)
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 #include <linux/perf_counter.h>
53 
54 #include <asm/uaccess.h>
55 #include <asm/processor.h>
56 
57 #ifdef CONFIG_X86
58 #include <asm/nmi.h>
59 #include <asm/stacktrace.h>
60 #include <asm/io.h>
61 #endif
62 
63 static int deprecated_sysctl_warning(struct __sysctl_args *args);
64 
65 #if defined(CONFIG_SYSCTL)
66 
67 /* External variables not in a header file. */
68 extern int C_A_D;
69 extern int print_fatal_signals;
70 extern int sysctl_overcommit_memory;
71 extern int sysctl_overcommit_ratio;
72 extern int sysctl_panic_on_oom;
73 extern int sysctl_oom_kill_allocating_task;
74 extern int sysctl_oom_dump_tasks;
75 extern int max_threads;
76 extern int core_uses_pid;
77 extern int suid_dumpable;
78 extern char core_pattern[];
79 extern int pid_max;
80 extern int min_free_kbytes;
81 extern int pid_max_min, pid_max_max;
82 extern int sysctl_drop_caches;
83 extern int percpu_pagelist_fraction;
84 extern int compat_log;
85 extern int latencytop_enabled;
86 extern int sysctl_nr_open_min, sysctl_nr_open_max;
87 #ifndef CONFIG_MMU
88 extern int sysctl_nr_trim_pages;
89 #endif
90 #ifdef CONFIG_RCU_TORTURE_TEST
91 extern int rcutorture_runnable;
92 #endif /* #ifdef CONFIG_RCU_TORTURE_TEST */
93 
94 /* Constants used for minimum and  maximum */
95 #ifdef CONFIG_DETECT_SOFTLOCKUP
96 static int sixty = 60;
97 static int neg_one = -1;
98 #endif
99 
100 static int zero;
101 static int __maybe_unused one = 1;
102 static int __maybe_unused two = 2;
103 static unsigned long one_ul = 1;
104 static int one_hundred = 100;
105 
106 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
107 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
108 
109 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
110 static int maxolduid = 65535;
111 static int minolduid;
112 static int min_percpu_pagelist_fract = 8;
113 
114 static int ngroups_max = NGROUPS_MAX;
115 
116 #ifdef CONFIG_MODULES
117 extern char modprobe_path[];
118 extern int modules_disabled;
119 #endif
120 #ifdef CONFIG_CHR_DEV_SG
121 extern int sg_big_buff;
122 #endif
123 
124 #ifdef CONFIG_SPARC
125 #include <asm/system.h>
126 #endif
127 
128 #ifdef CONFIG_SPARC64
129 extern int sysctl_tsb_ratio;
130 #endif
131 
132 #ifdef __hppa__
133 extern int pwrsw_enabled;
134 extern int unaligned_enabled;
135 #endif
136 
137 #ifdef CONFIG_S390
138 #ifdef CONFIG_MATHEMU
139 extern int sysctl_ieee_emulation_warnings;
140 #endif
141 extern int sysctl_userprocess_debug;
142 extern int spin_retry;
143 #endif
144 
145 #ifdef CONFIG_BSD_PROCESS_ACCT
146 extern int acct_parm[];
147 #endif
148 
149 #ifdef CONFIG_IA64
150 extern int no_unaligned_warning;
151 extern int unaligned_dump_stack;
152 #endif
153 
154 #ifdef CONFIG_RT_MUTEXES
155 extern int max_lock_depth;
156 #endif
157 
158 #ifdef CONFIG_PROC_SYSCTL
159 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
160 		  void __user *buffer, size_t *lenp, loff_t *ppos);
161 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
162 			       void __user *buffer, size_t *lenp, loff_t *ppos);
163 #endif
164 
165 static struct ctl_table root_table[];
166 static struct ctl_table_root sysctl_table_root;
167 static struct ctl_table_header root_table_header = {
168 	.count = 1,
169 	.ctl_table = root_table,
170 	.ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
171 	.root = &sysctl_table_root,
172 	.set = &sysctl_table_root.default_set,
173 };
174 static struct ctl_table_root sysctl_table_root = {
175 	.root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
176 	.default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
177 };
178 
179 static struct ctl_table kern_table[];
180 static struct ctl_table vm_table[];
181 static struct ctl_table fs_table[];
182 static struct ctl_table debug_table[];
183 static struct ctl_table dev_table[];
184 extern struct ctl_table random_table[];
185 #ifdef CONFIG_INOTIFY_USER
186 extern struct ctl_table inotify_table[];
187 #endif
188 #ifdef CONFIG_EPOLL
189 extern struct ctl_table epoll_table[];
190 #endif
191 
192 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
193 int sysctl_legacy_va_layout;
194 #endif
195 
196 extern int prove_locking;
197 extern int lock_stat;
198 
199 /* The default sysctl tables: */
200 
201 static struct ctl_table root_table[] = {
202 	{
203 		.ctl_name	= CTL_KERN,
204 		.procname	= "kernel",
205 		.mode		= 0555,
206 		.child		= kern_table,
207 	},
208 	{
209 		.ctl_name	= CTL_VM,
210 		.procname	= "vm",
211 		.mode		= 0555,
212 		.child		= vm_table,
213 	},
214 	{
215 		.ctl_name	= CTL_FS,
216 		.procname	= "fs",
217 		.mode		= 0555,
218 		.child		= fs_table,
219 	},
220 	{
221 		.ctl_name	= CTL_DEBUG,
222 		.procname	= "debug",
223 		.mode		= 0555,
224 		.child		= debug_table,
225 	},
226 	{
227 		.ctl_name	= CTL_DEV,
228 		.procname	= "dev",
229 		.mode		= 0555,
230 		.child		= dev_table,
231 	},
232 /*
233  * NOTE: do not add new entries to this table unless you have read
234  * Documentation/sysctl/ctl_unnumbered.txt
235  */
236 	{ .ctl_name = 0 }
237 };
238 
239 #ifdef CONFIG_SCHED_DEBUG
240 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
241 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
242 static int min_wakeup_granularity_ns;			/* 0 usecs */
243 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
244 #endif
245 
246 static struct ctl_table kern_table[] = {
247 #ifdef CONFIG_SCHED_DEBUG
248 	{
249 		.ctl_name	= CTL_UNNUMBERED,
250 		.procname	= "sched_min_granularity_ns",
251 		.data		= &sysctl_sched_min_granularity,
252 		.maxlen		= sizeof(unsigned int),
253 		.mode		= 0644,
254 		.proc_handler	= &sched_nr_latency_handler,
255 		.strategy	= &sysctl_intvec,
256 		.extra1		= &min_sched_granularity_ns,
257 		.extra2		= &max_sched_granularity_ns,
258 	},
259 	{
260 		.ctl_name	= CTL_UNNUMBERED,
261 		.procname	= "sched_latency_ns",
262 		.data		= &sysctl_sched_latency,
263 		.maxlen		= sizeof(unsigned int),
264 		.mode		= 0644,
265 		.proc_handler	= &sched_nr_latency_handler,
266 		.strategy	= &sysctl_intvec,
267 		.extra1		= &min_sched_granularity_ns,
268 		.extra2		= &max_sched_granularity_ns,
269 	},
270 	{
271 		.ctl_name	= CTL_UNNUMBERED,
272 		.procname	= "sched_wakeup_granularity_ns",
273 		.data		= &sysctl_sched_wakeup_granularity,
274 		.maxlen		= sizeof(unsigned int),
275 		.mode		= 0644,
276 		.proc_handler	= &proc_dointvec_minmax,
277 		.strategy	= &sysctl_intvec,
278 		.extra1		= &min_wakeup_granularity_ns,
279 		.extra2		= &max_wakeup_granularity_ns,
280 	},
281 	{
282 		.ctl_name	= CTL_UNNUMBERED,
283 		.procname	= "sched_shares_ratelimit",
284 		.data		= &sysctl_sched_shares_ratelimit,
285 		.maxlen		= sizeof(unsigned int),
286 		.mode		= 0644,
287 		.proc_handler	= &proc_dointvec,
288 	},
289 	{
290 		.ctl_name	= CTL_UNNUMBERED,
291 		.procname	= "sched_shares_thresh",
292 		.data		= &sysctl_sched_shares_thresh,
293 		.maxlen		= sizeof(unsigned int),
294 		.mode		= 0644,
295 		.proc_handler	= &proc_dointvec_minmax,
296 		.strategy	= &sysctl_intvec,
297 		.extra1		= &zero,
298 	},
299 	{
300 		.ctl_name	= CTL_UNNUMBERED,
301 		.procname	= "sched_child_runs_first",
302 		.data		= &sysctl_sched_child_runs_first,
303 		.maxlen		= sizeof(unsigned int),
304 		.mode		= 0644,
305 		.proc_handler	= &proc_dointvec,
306 	},
307 	{
308 		.ctl_name	= CTL_UNNUMBERED,
309 		.procname	= "sched_features",
310 		.data		= &sysctl_sched_features,
311 		.maxlen		= sizeof(unsigned int),
312 		.mode		= 0644,
313 		.proc_handler	= &proc_dointvec,
314 	},
315 	{
316 		.ctl_name	= CTL_UNNUMBERED,
317 		.procname	= "sched_migration_cost",
318 		.data		= &sysctl_sched_migration_cost,
319 		.maxlen		= sizeof(unsigned int),
320 		.mode		= 0644,
321 		.proc_handler	= &proc_dointvec,
322 	},
323 	{
324 		.ctl_name	= CTL_UNNUMBERED,
325 		.procname	= "sched_nr_migrate",
326 		.data		= &sysctl_sched_nr_migrate,
327 		.maxlen		= sizeof(unsigned int),
328 		.mode		= 0644,
329 		.proc_handler	= &proc_dointvec,
330 	},
331 	{
332 		.ctl_name	= CTL_UNNUMBERED,
333 		.procname	= "timer_migration",
334 		.data		= &sysctl_timer_migration,
335 		.maxlen		= sizeof(unsigned int),
336 		.mode		= 0644,
337 		.proc_handler	= &proc_dointvec,
338 	},
339 #endif
340 	{
341 		.ctl_name	= CTL_UNNUMBERED,
342 		.procname	= "sched_rt_period_us",
343 		.data		= &sysctl_sched_rt_period,
344 		.maxlen		= sizeof(unsigned int),
345 		.mode		= 0644,
346 		.proc_handler	= &sched_rt_handler,
347 	},
348 	{
349 		.ctl_name	= CTL_UNNUMBERED,
350 		.procname	= "sched_rt_runtime_us",
351 		.data		= &sysctl_sched_rt_runtime,
352 		.maxlen		= sizeof(int),
353 		.mode		= 0644,
354 		.proc_handler	= &sched_rt_handler,
355 	},
356 	{
357 		.ctl_name	= CTL_UNNUMBERED,
358 		.procname	= "sched_compat_yield",
359 		.data		= &sysctl_sched_compat_yield,
360 		.maxlen		= sizeof(unsigned int),
361 		.mode		= 0644,
362 		.proc_handler	= &proc_dointvec,
363 	},
364 #ifdef CONFIG_PROVE_LOCKING
365 	{
366 		.ctl_name	= CTL_UNNUMBERED,
367 		.procname	= "prove_locking",
368 		.data		= &prove_locking,
369 		.maxlen		= sizeof(int),
370 		.mode		= 0644,
371 		.proc_handler	= &proc_dointvec,
372 	},
373 #endif
374 #ifdef CONFIG_LOCK_STAT
375 	{
376 		.ctl_name	= CTL_UNNUMBERED,
377 		.procname	= "lock_stat",
378 		.data		= &lock_stat,
379 		.maxlen		= sizeof(int),
380 		.mode		= 0644,
381 		.proc_handler	= &proc_dointvec,
382 	},
383 #endif
384 	{
385 		.ctl_name	= KERN_PANIC,
386 		.procname	= "panic",
387 		.data		= &panic_timeout,
388 		.maxlen		= sizeof(int),
389 		.mode		= 0644,
390 		.proc_handler	= &proc_dointvec,
391 	},
392 	{
393 		.ctl_name	= KERN_CORE_USES_PID,
394 		.procname	= "core_uses_pid",
395 		.data		= &core_uses_pid,
396 		.maxlen		= sizeof(int),
397 		.mode		= 0644,
398 		.proc_handler	= &proc_dointvec,
399 	},
400 	{
401 		.ctl_name	= KERN_CORE_PATTERN,
402 		.procname	= "core_pattern",
403 		.data		= core_pattern,
404 		.maxlen		= CORENAME_MAX_SIZE,
405 		.mode		= 0644,
406 		.proc_handler	= &proc_dostring,
407 		.strategy	= &sysctl_string,
408 	},
409 #ifdef CONFIG_PROC_SYSCTL
410 	{
411 		.procname	= "tainted",
412 		.maxlen 	= sizeof(long),
413 		.mode		= 0644,
414 		.proc_handler	= &proc_taint,
415 	},
416 #endif
417 #ifdef CONFIG_LATENCYTOP
418 	{
419 		.procname	= "latencytop",
420 		.data		= &latencytop_enabled,
421 		.maxlen		= sizeof(int),
422 		.mode		= 0644,
423 		.proc_handler	= &proc_dointvec,
424 	},
425 #endif
426 #ifdef CONFIG_BLK_DEV_INITRD
427 	{
428 		.ctl_name	= KERN_REALROOTDEV,
429 		.procname	= "real-root-dev",
430 		.data		= &real_root_dev,
431 		.maxlen		= sizeof(int),
432 		.mode		= 0644,
433 		.proc_handler	= &proc_dointvec,
434 	},
435 #endif
436 	{
437 		.ctl_name	= CTL_UNNUMBERED,
438 		.procname	= "print-fatal-signals",
439 		.data		= &print_fatal_signals,
440 		.maxlen		= sizeof(int),
441 		.mode		= 0644,
442 		.proc_handler	= &proc_dointvec,
443 	},
444 #ifdef CONFIG_SPARC
445 	{
446 		.ctl_name	= KERN_SPARC_REBOOT,
447 		.procname	= "reboot-cmd",
448 		.data		= reboot_command,
449 		.maxlen		= 256,
450 		.mode		= 0644,
451 		.proc_handler	= &proc_dostring,
452 		.strategy	= &sysctl_string,
453 	},
454 	{
455 		.ctl_name	= KERN_SPARC_STOP_A,
456 		.procname	= "stop-a",
457 		.data		= &stop_a_enabled,
458 		.maxlen		= sizeof (int),
459 		.mode		= 0644,
460 		.proc_handler	= &proc_dointvec,
461 	},
462 	{
463 		.ctl_name	= KERN_SPARC_SCONS_PWROFF,
464 		.procname	= "scons-poweroff",
465 		.data		= &scons_pwroff,
466 		.maxlen		= sizeof (int),
467 		.mode		= 0644,
468 		.proc_handler	= &proc_dointvec,
469 	},
470 #endif
471 #ifdef CONFIG_SPARC64
472 	{
473 		.ctl_name	= CTL_UNNUMBERED,
474 		.procname	= "tsb-ratio",
475 		.data		= &sysctl_tsb_ratio,
476 		.maxlen		= sizeof (int),
477 		.mode		= 0644,
478 		.proc_handler	= &proc_dointvec,
479 	},
480 #endif
481 #ifdef __hppa__
482 	{
483 		.ctl_name	= KERN_HPPA_PWRSW,
484 		.procname	= "soft-power",
485 		.data		= &pwrsw_enabled,
486 		.maxlen		= sizeof (int),
487 	 	.mode		= 0644,
488 		.proc_handler	= &proc_dointvec,
489 	},
490 	{
491 		.ctl_name	= KERN_HPPA_UNALIGNED,
492 		.procname	= "unaligned-trap",
493 		.data		= &unaligned_enabled,
494 		.maxlen		= sizeof (int),
495 		.mode		= 0644,
496 		.proc_handler	= &proc_dointvec,
497 	},
498 #endif
499 	{
500 		.ctl_name	= KERN_CTLALTDEL,
501 		.procname	= "ctrl-alt-del",
502 		.data		= &C_A_D,
503 		.maxlen		= sizeof(int),
504 		.mode		= 0644,
505 		.proc_handler	= &proc_dointvec,
506 	},
507 #ifdef CONFIG_FUNCTION_TRACER
508 	{
509 		.ctl_name	= CTL_UNNUMBERED,
510 		.procname	= "ftrace_enabled",
511 		.data		= &ftrace_enabled,
512 		.maxlen		= sizeof(int),
513 		.mode		= 0644,
514 		.proc_handler	= &ftrace_enable_sysctl,
515 	},
516 #endif
517 #ifdef CONFIG_STACK_TRACER
518 	{
519 		.ctl_name	= CTL_UNNUMBERED,
520 		.procname	= "stack_tracer_enabled",
521 		.data		= &stack_tracer_enabled,
522 		.maxlen		= sizeof(int),
523 		.mode		= 0644,
524 		.proc_handler	= &stack_trace_sysctl,
525 	},
526 #endif
527 #ifdef CONFIG_TRACING
528 	{
529 		.ctl_name	= CTL_UNNUMBERED,
530 		.procname	= "ftrace_dump_on_oops",
531 		.data		= &ftrace_dump_on_oops,
532 		.maxlen		= sizeof(int),
533 		.mode		= 0644,
534 		.proc_handler	= &proc_dointvec,
535 	},
536 #endif
537 #ifdef CONFIG_MODULES
538 	{
539 		.ctl_name	= KERN_MODPROBE,
540 		.procname	= "modprobe",
541 		.data		= &modprobe_path,
542 		.maxlen		= KMOD_PATH_LEN,
543 		.mode		= 0644,
544 		.proc_handler	= &proc_dostring,
545 		.strategy	= &sysctl_string,
546 	},
547 	{
548 		.ctl_name	= CTL_UNNUMBERED,
549 		.procname	= "modules_disabled",
550 		.data		= &modules_disabled,
551 		.maxlen		= sizeof(int),
552 		.mode		= 0644,
553 		/* only handle a transition from default "0" to "1" */
554 		.proc_handler	= &proc_dointvec_minmax,
555 		.extra1		= &one,
556 		.extra2		= &one,
557 	},
558 #endif
559 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
560 	{
561 		.ctl_name	= KERN_HOTPLUG,
562 		.procname	= "hotplug",
563 		.data		= &uevent_helper,
564 		.maxlen		= UEVENT_HELPER_PATH_LEN,
565 		.mode		= 0644,
566 		.proc_handler	= &proc_dostring,
567 		.strategy	= &sysctl_string,
568 	},
569 #endif
570 #ifdef CONFIG_CHR_DEV_SG
571 	{
572 		.ctl_name	= KERN_SG_BIG_BUFF,
573 		.procname	= "sg-big-buff",
574 		.data		= &sg_big_buff,
575 		.maxlen		= sizeof (int),
576 		.mode		= 0444,
577 		.proc_handler	= &proc_dointvec,
578 	},
579 #endif
580 #ifdef CONFIG_BSD_PROCESS_ACCT
581 	{
582 		.ctl_name	= KERN_ACCT,
583 		.procname	= "acct",
584 		.data		= &acct_parm,
585 		.maxlen		= 3*sizeof(int),
586 		.mode		= 0644,
587 		.proc_handler	= &proc_dointvec,
588 	},
589 #endif
590 #ifdef CONFIG_MAGIC_SYSRQ
591 	{
592 		.ctl_name	= KERN_SYSRQ,
593 		.procname	= "sysrq",
594 		.data		= &__sysrq_enabled,
595 		.maxlen		= sizeof (int),
596 		.mode		= 0644,
597 		.proc_handler	= &proc_dointvec,
598 	},
599 #endif
600 #ifdef CONFIG_PROC_SYSCTL
601 	{
602 		.procname	= "cad_pid",
603 		.data		= NULL,
604 		.maxlen		= sizeof (int),
605 		.mode		= 0600,
606 		.proc_handler	= &proc_do_cad_pid,
607 	},
608 #endif
609 	{
610 		.ctl_name	= KERN_MAX_THREADS,
611 		.procname	= "threads-max",
612 		.data		= &max_threads,
613 		.maxlen		= sizeof(int),
614 		.mode		= 0644,
615 		.proc_handler	= &proc_dointvec,
616 	},
617 	{
618 		.ctl_name	= KERN_RANDOM,
619 		.procname	= "random",
620 		.mode		= 0555,
621 		.child		= random_table,
622 	},
623 	{
624 		.ctl_name	= KERN_OVERFLOWUID,
625 		.procname	= "overflowuid",
626 		.data		= &overflowuid,
627 		.maxlen		= sizeof(int),
628 		.mode		= 0644,
629 		.proc_handler	= &proc_dointvec_minmax,
630 		.strategy	= &sysctl_intvec,
631 		.extra1		= &minolduid,
632 		.extra2		= &maxolduid,
633 	},
634 	{
635 		.ctl_name	= KERN_OVERFLOWGID,
636 		.procname	= "overflowgid",
637 		.data		= &overflowgid,
638 		.maxlen		= sizeof(int),
639 		.mode		= 0644,
640 		.proc_handler	= &proc_dointvec_minmax,
641 		.strategy	= &sysctl_intvec,
642 		.extra1		= &minolduid,
643 		.extra2		= &maxolduid,
644 	},
645 #ifdef CONFIG_S390
646 #ifdef CONFIG_MATHEMU
647 	{
648 		.ctl_name	= KERN_IEEE_EMULATION_WARNINGS,
649 		.procname	= "ieee_emulation_warnings",
650 		.data		= &sysctl_ieee_emulation_warnings,
651 		.maxlen		= sizeof(int),
652 		.mode		= 0644,
653 		.proc_handler	= &proc_dointvec,
654 	},
655 #endif
656 	{
657 		.ctl_name	= KERN_S390_USER_DEBUG_LOGGING,
658 		.procname	= "userprocess_debug",
659 		.data		= &sysctl_userprocess_debug,
660 		.maxlen		= sizeof(int),
661 		.mode		= 0644,
662 		.proc_handler	= &proc_dointvec,
663 	},
664 #endif
665 	{
666 		.ctl_name	= KERN_PIDMAX,
667 		.procname	= "pid_max",
668 		.data		= &pid_max,
669 		.maxlen		= sizeof (int),
670 		.mode		= 0644,
671 		.proc_handler	= &proc_dointvec_minmax,
672 		.strategy	= sysctl_intvec,
673 		.extra1		= &pid_max_min,
674 		.extra2		= &pid_max_max,
675 	},
676 	{
677 		.ctl_name	= KERN_PANIC_ON_OOPS,
678 		.procname	= "panic_on_oops",
679 		.data		= &panic_on_oops,
680 		.maxlen		= sizeof(int),
681 		.mode		= 0644,
682 		.proc_handler	= &proc_dointvec,
683 	},
684 #if defined CONFIG_PRINTK
685 	{
686 		.ctl_name	= KERN_PRINTK,
687 		.procname	= "printk",
688 		.data		= &console_loglevel,
689 		.maxlen		= 4*sizeof(int),
690 		.mode		= 0644,
691 		.proc_handler	= &proc_dointvec,
692 	},
693 	{
694 		.ctl_name	= KERN_PRINTK_RATELIMIT,
695 		.procname	= "printk_ratelimit",
696 		.data		= &printk_ratelimit_state.interval,
697 		.maxlen		= sizeof(int),
698 		.mode		= 0644,
699 		.proc_handler	= &proc_dointvec_jiffies,
700 		.strategy	= &sysctl_jiffies,
701 	},
702 	{
703 		.ctl_name	= KERN_PRINTK_RATELIMIT_BURST,
704 		.procname	= "printk_ratelimit_burst",
705 		.data		= &printk_ratelimit_state.burst,
706 		.maxlen		= sizeof(int),
707 		.mode		= 0644,
708 		.proc_handler	= &proc_dointvec,
709 	},
710 #endif
711 	{
712 		.ctl_name	= KERN_NGROUPS_MAX,
713 		.procname	= "ngroups_max",
714 		.data		= &ngroups_max,
715 		.maxlen		= sizeof (int),
716 		.mode		= 0444,
717 		.proc_handler	= &proc_dointvec,
718 	},
719 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
720 	{
721 		.ctl_name       = KERN_UNKNOWN_NMI_PANIC,
722 		.procname       = "unknown_nmi_panic",
723 		.data           = &unknown_nmi_panic,
724 		.maxlen         = sizeof (int),
725 		.mode           = 0644,
726 		.proc_handler   = &proc_dointvec,
727 	},
728 	{
729 		.procname       = "nmi_watchdog",
730 		.data           = &nmi_watchdog_enabled,
731 		.maxlen         = sizeof (int),
732 		.mode           = 0644,
733 		.proc_handler   = &proc_nmi_enabled,
734 	},
735 #endif
736 #if defined(CONFIG_X86)
737 	{
738 		.ctl_name	= KERN_PANIC_ON_NMI,
739 		.procname	= "panic_on_unrecovered_nmi",
740 		.data		= &panic_on_unrecovered_nmi,
741 		.maxlen		= sizeof(int),
742 		.mode		= 0644,
743 		.proc_handler	= &proc_dointvec,
744 	},
745 	{
746 		.ctl_name	= KERN_BOOTLOADER_TYPE,
747 		.procname	= "bootloader_type",
748 		.data		= &bootloader_type,
749 		.maxlen		= sizeof (int),
750 		.mode		= 0444,
751 		.proc_handler	= &proc_dointvec,
752 	},
753 	{
754 		.ctl_name	= CTL_UNNUMBERED,
755 		.procname	= "bootloader_version",
756 		.data		= &bootloader_version,
757 		.maxlen		= sizeof (int),
758 		.mode		= 0444,
759 		.proc_handler	= &proc_dointvec,
760 	},
761 	{
762 		.ctl_name	= CTL_UNNUMBERED,
763 		.procname	= "kstack_depth_to_print",
764 		.data		= &kstack_depth_to_print,
765 		.maxlen		= sizeof(int),
766 		.mode		= 0644,
767 		.proc_handler	= &proc_dointvec,
768 	},
769 	{
770 		.ctl_name	= CTL_UNNUMBERED,
771 		.procname	= "io_delay_type",
772 		.data		= &io_delay_type,
773 		.maxlen		= sizeof(int),
774 		.mode		= 0644,
775 		.proc_handler	= &proc_dointvec,
776 	},
777 #endif
778 #if defined(CONFIG_MMU)
779 	{
780 		.ctl_name	= KERN_RANDOMIZE,
781 		.procname	= "randomize_va_space",
782 		.data		= &randomize_va_space,
783 		.maxlen		= sizeof(int),
784 		.mode		= 0644,
785 		.proc_handler	= &proc_dointvec,
786 	},
787 #endif
788 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
789 	{
790 		.ctl_name	= KERN_SPIN_RETRY,
791 		.procname	= "spin_retry",
792 		.data		= &spin_retry,
793 		.maxlen		= sizeof (int),
794 		.mode		= 0644,
795 		.proc_handler	= &proc_dointvec,
796 	},
797 #endif
798 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
799 	{
800 		.procname	= "acpi_video_flags",
801 		.data		= &acpi_realmode_flags,
802 		.maxlen		= sizeof (unsigned long),
803 		.mode		= 0644,
804 		.proc_handler	= &proc_doulongvec_minmax,
805 	},
806 #endif
807 #ifdef CONFIG_IA64
808 	{
809 		.ctl_name	= KERN_IA64_UNALIGNED,
810 		.procname	= "ignore-unaligned-usertrap",
811 		.data		= &no_unaligned_warning,
812 		.maxlen		= sizeof (int),
813 	 	.mode		= 0644,
814 		.proc_handler	= &proc_dointvec,
815 	},
816 	{
817 		.ctl_name	= CTL_UNNUMBERED,
818 		.procname	= "unaligned-dump-stack",
819 		.data		= &unaligned_dump_stack,
820 		.maxlen		= sizeof (int),
821 		.mode		= 0644,
822 		.proc_handler	= &proc_dointvec,
823 	},
824 #endif
825 #ifdef CONFIG_DETECT_SOFTLOCKUP
826 	{
827 		.ctl_name	= CTL_UNNUMBERED,
828 		.procname	= "softlockup_panic",
829 		.data		= &softlockup_panic,
830 		.maxlen		= sizeof(int),
831 		.mode		= 0644,
832 		.proc_handler	= &proc_dointvec_minmax,
833 		.strategy	= &sysctl_intvec,
834 		.extra1		= &zero,
835 		.extra2		= &one,
836 	},
837 	{
838 		.ctl_name	= CTL_UNNUMBERED,
839 		.procname	= "softlockup_thresh",
840 		.data		= &softlockup_thresh,
841 		.maxlen		= sizeof(int),
842 		.mode		= 0644,
843 		.proc_handler	= &proc_dosoftlockup_thresh,
844 		.strategy	= &sysctl_intvec,
845 		.extra1		= &neg_one,
846 		.extra2		= &sixty,
847 	},
848 #endif
849 #ifdef CONFIG_DETECT_HUNG_TASK
850 	{
851 		.ctl_name	= CTL_UNNUMBERED,
852 		.procname	= "hung_task_panic",
853 		.data		= &sysctl_hung_task_panic,
854 		.maxlen		= sizeof(int),
855 		.mode		= 0644,
856 		.proc_handler	= &proc_dointvec_minmax,
857 		.strategy	= &sysctl_intvec,
858 		.extra1		= &zero,
859 		.extra2		= &one,
860 	},
861 	{
862 		.ctl_name	= CTL_UNNUMBERED,
863 		.procname	= "hung_task_check_count",
864 		.data		= &sysctl_hung_task_check_count,
865 		.maxlen		= sizeof(unsigned long),
866 		.mode		= 0644,
867 		.proc_handler	= &proc_doulongvec_minmax,
868 		.strategy	= &sysctl_intvec,
869 	},
870 	{
871 		.ctl_name	= CTL_UNNUMBERED,
872 		.procname	= "hung_task_timeout_secs",
873 		.data		= &sysctl_hung_task_timeout_secs,
874 		.maxlen		= sizeof(unsigned long),
875 		.mode		= 0644,
876 		.proc_handler	= &proc_dohung_task_timeout_secs,
877 		.strategy	= &sysctl_intvec,
878 	},
879 	{
880 		.ctl_name	= CTL_UNNUMBERED,
881 		.procname	= "hung_task_warnings",
882 		.data		= &sysctl_hung_task_warnings,
883 		.maxlen		= sizeof(unsigned long),
884 		.mode		= 0644,
885 		.proc_handler	= &proc_doulongvec_minmax,
886 		.strategy	= &sysctl_intvec,
887 	},
888 #endif
889 #ifdef CONFIG_COMPAT
890 	{
891 		.ctl_name	= KERN_COMPAT_LOG,
892 		.procname	= "compat-log",
893 		.data		= &compat_log,
894 		.maxlen		= sizeof (int),
895 	 	.mode		= 0644,
896 		.proc_handler	= &proc_dointvec,
897 	},
898 #endif
899 #ifdef CONFIG_RT_MUTEXES
900 	{
901 		.ctl_name	= KERN_MAX_LOCK_DEPTH,
902 		.procname	= "max_lock_depth",
903 		.data		= &max_lock_depth,
904 		.maxlen		= sizeof(int),
905 		.mode		= 0644,
906 		.proc_handler	= &proc_dointvec,
907 	},
908 #endif
909 	{
910 		.ctl_name	= CTL_UNNUMBERED,
911 		.procname	= "poweroff_cmd",
912 		.data		= &poweroff_cmd,
913 		.maxlen		= POWEROFF_CMD_PATH_LEN,
914 		.mode		= 0644,
915 		.proc_handler	= &proc_dostring,
916 		.strategy	= &sysctl_string,
917 	},
918 #ifdef CONFIG_KEYS
919 	{
920 		.ctl_name	= CTL_UNNUMBERED,
921 		.procname	= "keys",
922 		.mode		= 0555,
923 		.child		= key_sysctls,
924 	},
925 #endif
926 #ifdef CONFIG_RCU_TORTURE_TEST
927 	{
928 		.ctl_name       = CTL_UNNUMBERED,
929 		.procname       = "rcutorture_runnable",
930 		.data           = &rcutorture_runnable,
931 		.maxlen         = sizeof(int),
932 		.mode           = 0644,
933 		.proc_handler   = &proc_dointvec,
934 	},
935 #endif
936 #ifdef CONFIG_SLOW_WORK
937 	{
938 		.ctl_name	= CTL_UNNUMBERED,
939 		.procname	= "slow-work",
940 		.mode		= 0555,
941 		.child		= slow_work_sysctls,
942 	},
943 #endif
944 #ifdef CONFIG_PERF_COUNTERS
945 	{
946 		.ctl_name	= CTL_UNNUMBERED,
947 		.procname	= "perf_counter_paranoid",
948 		.data		= &sysctl_perf_counter_paranoid,
949 		.maxlen		= sizeof(sysctl_perf_counter_paranoid),
950 		.mode		= 0644,
951 		.proc_handler	= &proc_dointvec,
952 	},
953 	{
954 		.ctl_name	= CTL_UNNUMBERED,
955 		.procname	= "perf_counter_mlock_kb",
956 		.data		= &sysctl_perf_counter_mlock,
957 		.maxlen		= sizeof(sysctl_perf_counter_mlock),
958 		.mode		= 0644,
959 		.proc_handler	= &proc_dointvec,
960 	},
961 	{
962 		.ctl_name	= CTL_UNNUMBERED,
963 		.procname	= "perf_counter_max_sample_rate",
964 		.data		= &sysctl_perf_counter_sample_rate,
965 		.maxlen		= sizeof(sysctl_perf_counter_sample_rate),
966 		.mode		= 0644,
967 		.proc_handler	= &proc_dointvec,
968 	},
969 #endif
970 /*
971  * NOTE: do not add new entries to this table unless you have read
972  * Documentation/sysctl/ctl_unnumbered.txt
973  */
974 	{ .ctl_name = 0 }
975 };
976 
977 static struct ctl_table vm_table[] = {
978 	{
979 		.ctl_name	= VM_OVERCOMMIT_MEMORY,
980 		.procname	= "overcommit_memory",
981 		.data		= &sysctl_overcommit_memory,
982 		.maxlen		= sizeof(sysctl_overcommit_memory),
983 		.mode		= 0644,
984 		.proc_handler	= &proc_dointvec,
985 	},
986 	{
987 		.ctl_name	= VM_PANIC_ON_OOM,
988 		.procname	= "panic_on_oom",
989 		.data		= &sysctl_panic_on_oom,
990 		.maxlen		= sizeof(sysctl_panic_on_oom),
991 		.mode		= 0644,
992 		.proc_handler	= &proc_dointvec,
993 	},
994 	{
995 		.ctl_name	= CTL_UNNUMBERED,
996 		.procname	= "oom_kill_allocating_task",
997 		.data		= &sysctl_oom_kill_allocating_task,
998 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
999 		.mode		= 0644,
1000 		.proc_handler	= &proc_dointvec,
1001 	},
1002 	{
1003 		.ctl_name	= CTL_UNNUMBERED,
1004 		.procname	= "oom_dump_tasks",
1005 		.data		= &sysctl_oom_dump_tasks,
1006 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1007 		.mode		= 0644,
1008 		.proc_handler	= &proc_dointvec,
1009 	},
1010 	{
1011 		.ctl_name	= VM_OVERCOMMIT_RATIO,
1012 		.procname	= "overcommit_ratio",
1013 		.data		= &sysctl_overcommit_ratio,
1014 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1015 		.mode		= 0644,
1016 		.proc_handler	= &proc_dointvec,
1017 	},
1018 	{
1019 		.ctl_name	= VM_PAGE_CLUSTER,
1020 		.procname	= "page-cluster",
1021 		.data		= &page_cluster,
1022 		.maxlen		= sizeof(int),
1023 		.mode		= 0644,
1024 		.proc_handler	= &proc_dointvec,
1025 	},
1026 	{
1027 		.ctl_name	= VM_DIRTY_BACKGROUND,
1028 		.procname	= "dirty_background_ratio",
1029 		.data		= &dirty_background_ratio,
1030 		.maxlen		= sizeof(dirty_background_ratio),
1031 		.mode		= 0644,
1032 		.proc_handler	= &dirty_background_ratio_handler,
1033 		.strategy	= &sysctl_intvec,
1034 		.extra1		= &zero,
1035 		.extra2		= &one_hundred,
1036 	},
1037 	{
1038 		.ctl_name	= CTL_UNNUMBERED,
1039 		.procname	= "dirty_background_bytes",
1040 		.data		= &dirty_background_bytes,
1041 		.maxlen		= sizeof(dirty_background_bytes),
1042 		.mode		= 0644,
1043 		.proc_handler	= &dirty_background_bytes_handler,
1044 		.strategy	= &sysctl_intvec,
1045 		.extra1		= &one_ul,
1046 	},
1047 	{
1048 		.ctl_name	= VM_DIRTY_RATIO,
1049 		.procname	= "dirty_ratio",
1050 		.data		= &vm_dirty_ratio,
1051 		.maxlen		= sizeof(vm_dirty_ratio),
1052 		.mode		= 0644,
1053 		.proc_handler	= &dirty_ratio_handler,
1054 		.strategy	= &sysctl_intvec,
1055 		.extra1		= &zero,
1056 		.extra2		= &one_hundred,
1057 	},
1058 	{
1059 		.ctl_name	= CTL_UNNUMBERED,
1060 		.procname	= "dirty_bytes",
1061 		.data		= &vm_dirty_bytes,
1062 		.maxlen		= sizeof(vm_dirty_bytes),
1063 		.mode		= 0644,
1064 		.proc_handler	= &dirty_bytes_handler,
1065 		.strategy	= &sysctl_intvec,
1066 		.extra1		= &dirty_bytes_min,
1067 	},
1068 	{
1069 		.procname	= "dirty_writeback_centisecs",
1070 		.data		= &dirty_writeback_interval,
1071 		.maxlen		= sizeof(dirty_writeback_interval),
1072 		.mode		= 0644,
1073 		.proc_handler	= &dirty_writeback_centisecs_handler,
1074 	},
1075 	{
1076 		.procname	= "dirty_expire_centisecs",
1077 		.data		= &dirty_expire_interval,
1078 		.maxlen		= sizeof(dirty_expire_interval),
1079 		.mode		= 0644,
1080 		.proc_handler	= &proc_dointvec,
1081 	},
1082 	{
1083 		.ctl_name	= VM_NR_PDFLUSH_THREADS,
1084 		.procname	= "nr_pdflush_threads",
1085 		.data		= &nr_pdflush_threads,
1086 		.maxlen		= sizeof nr_pdflush_threads,
1087 		.mode		= 0444 /* read-only*/,
1088 		.proc_handler	= &proc_dointvec,
1089 	},
1090 	{
1091 		.ctl_name	= VM_SWAPPINESS,
1092 		.procname	= "swappiness",
1093 		.data		= &vm_swappiness,
1094 		.maxlen		= sizeof(vm_swappiness),
1095 		.mode		= 0644,
1096 		.proc_handler	= &proc_dointvec_minmax,
1097 		.strategy	= &sysctl_intvec,
1098 		.extra1		= &zero,
1099 		.extra2		= &one_hundred,
1100 	},
1101 #ifdef CONFIG_HUGETLB_PAGE
1102 	 {
1103 		.procname	= "nr_hugepages",
1104 		.data		= NULL,
1105 		.maxlen		= sizeof(unsigned long),
1106 		.mode		= 0644,
1107 		.proc_handler	= &hugetlb_sysctl_handler,
1108 		.extra1		= (void *)&hugetlb_zero,
1109 		.extra2		= (void *)&hugetlb_infinity,
1110 	 },
1111 	 {
1112 		.ctl_name	= VM_HUGETLB_GROUP,
1113 		.procname	= "hugetlb_shm_group",
1114 		.data		= &sysctl_hugetlb_shm_group,
1115 		.maxlen		= sizeof(gid_t),
1116 		.mode		= 0644,
1117 		.proc_handler	= &proc_dointvec,
1118 	 },
1119 	 {
1120 		.ctl_name	= CTL_UNNUMBERED,
1121 		.procname	= "hugepages_treat_as_movable",
1122 		.data		= &hugepages_treat_as_movable,
1123 		.maxlen		= sizeof(int),
1124 		.mode		= 0644,
1125 		.proc_handler	= &hugetlb_treat_movable_handler,
1126 	},
1127 	{
1128 		.ctl_name	= CTL_UNNUMBERED,
1129 		.procname	= "nr_overcommit_hugepages",
1130 		.data		= NULL,
1131 		.maxlen		= sizeof(unsigned long),
1132 		.mode		= 0644,
1133 		.proc_handler	= &hugetlb_overcommit_handler,
1134 		.extra1		= (void *)&hugetlb_zero,
1135 		.extra2		= (void *)&hugetlb_infinity,
1136 	},
1137 #endif
1138 	{
1139 		.ctl_name	= VM_LOWMEM_RESERVE_RATIO,
1140 		.procname	= "lowmem_reserve_ratio",
1141 		.data		= &sysctl_lowmem_reserve_ratio,
1142 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1143 		.mode		= 0644,
1144 		.proc_handler	= &lowmem_reserve_ratio_sysctl_handler,
1145 		.strategy	= &sysctl_intvec,
1146 	},
1147 	{
1148 		.ctl_name	= VM_DROP_PAGECACHE,
1149 		.procname	= "drop_caches",
1150 		.data		= &sysctl_drop_caches,
1151 		.maxlen		= sizeof(int),
1152 		.mode		= 0644,
1153 		.proc_handler	= drop_caches_sysctl_handler,
1154 		.strategy	= &sysctl_intvec,
1155 	},
1156 	{
1157 		.ctl_name	= VM_MIN_FREE_KBYTES,
1158 		.procname	= "min_free_kbytes",
1159 		.data		= &min_free_kbytes,
1160 		.maxlen		= sizeof(min_free_kbytes),
1161 		.mode		= 0644,
1162 		.proc_handler	= &min_free_kbytes_sysctl_handler,
1163 		.strategy	= &sysctl_intvec,
1164 		.extra1		= &zero,
1165 	},
1166 	{
1167 		.ctl_name	= VM_PERCPU_PAGELIST_FRACTION,
1168 		.procname	= "percpu_pagelist_fraction",
1169 		.data		= &percpu_pagelist_fraction,
1170 		.maxlen		= sizeof(percpu_pagelist_fraction),
1171 		.mode		= 0644,
1172 		.proc_handler	= &percpu_pagelist_fraction_sysctl_handler,
1173 		.strategy	= &sysctl_intvec,
1174 		.extra1		= &min_percpu_pagelist_fract,
1175 	},
1176 #ifdef CONFIG_MMU
1177 	{
1178 		.ctl_name	= VM_MAX_MAP_COUNT,
1179 		.procname	= "max_map_count",
1180 		.data		= &sysctl_max_map_count,
1181 		.maxlen		= sizeof(sysctl_max_map_count),
1182 		.mode		= 0644,
1183 		.proc_handler	= &proc_dointvec
1184 	},
1185 #else
1186 	{
1187 		.ctl_name	= CTL_UNNUMBERED,
1188 		.procname	= "nr_trim_pages",
1189 		.data		= &sysctl_nr_trim_pages,
1190 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1191 		.mode		= 0644,
1192 		.proc_handler	= &proc_dointvec_minmax,
1193 		.strategy	= &sysctl_intvec,
1194 		.extra1		= &zero,
1195 	},
1196 #endif
1197 	{
1198 		.ctl_name	= VM_LAPTOP_MODE,
1199 		.procname	= "laptop_mode",
1200 		.data		= &laptop_mode,
1201 		.maxlen		= sizeof(laptop_mode),
1202 		.mode		= 0644,
1203 		.proc_handler	= &proc_dointvec_jiffies,
1204 		.strategy	= &sysctl_jiffies,
1205 	},
1206 	{
1207 		.ctl_name	= VM_BLOCK_DUMP,
1208 		.procname	= "block_dump",
1209 		.data		= &block_dump,
1210 		.maxlen		= sizeof(block_dump),
1211 		.mode		= 0644,
1212 		.proc_handler	= &proc_dointvec,
1213 		.strategy	= &sysctl_intvec,
1214 		.extra1		= &zero,
1215 	},
1216 	{
1217 		.ctl_name	= VM_VFS_CACHE_PRESSURE,
1218 		.procname	= "vfs_cache_pressure",
1219 		.data		= &sysctl_vfs_cache_pressure,
1220 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1221 		.mode		= 0644,
1222 		.proc_handler	= &proc_dointvec,
1223 		.strategy	= &sysctl_intvec,
1224 		.extra1		= &zero,
1225 	},
1226 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1227 	{
1228 		.ctl_name	= VM_LEGACY_VA_LAYOUT,
1229 		.procname	= "legacy_va_layout",
1230 		.data		= &sysctl_legacy_va_layout,
1231 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1232 		.mode		= 0644,
1233 		.proc_handler	= &proc_dointvec,
1234 		.strategy	= &sysctl_intvec,
1235 		.extra1		= &zero,
1236 	},
1237 #endif
1238 #ifdef CONFIG_NUMA
1239 	{
1240 		.ctl_name	= VM_ZONE_RECLAIM_MODE,
1241 		.procname	= "zone_reclaim_mode",
1242 		.data		= &zone_reclaim_mode,
1243 		.maxlen		= sizeof(zone_reclaim_mode),
1244 		.mode		= 0644,
1245 		.proc_handler	= &proc_dointvec,
1246 		.strategy	= &sysctl_intvec,
1247 		.extra1		= &zero,
1248 	},
1249 	{
1250 		.ctl_name	= VM_MIN_UNMAPPED,
1251 		.procname	= "min_unmapped_ratio",
1252 		.data		= &sysctl_min_unmapped_ratio,
1253 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1254 		.mode		= 0644,
1255 		.proc_handler	= &sysctl_min_unmapped_ratio_sysctl_handler,
1256 		.strategy	= &sysctl_intvec,
1257 		.extra1		= &zero,
1258 		.extra2		= &one_hundred,
1259 	},
1260 	{
1261 		.ctl_name	= VM_MIN_SLAB,
1262 		.procname	= "min_slab_ratio",
1263 		.data		= &sysctl_min_slab_ratio,
1264 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1265 		.mode		= 0644,
1266 		.proc_handler	= &sysctl_min_slab_ratio_sysctl_handler,
1267 		.strategy	= &sysctl_intvec,
1268 		.extra1		= &zero,
1269 		.extra2		= &one_hundred,
1270 	},
1271 #endif
1272 #ifdef CONFIG_SMP
1273 	{
1274 		.ctl_name	= CTL_UNNUMBERED,
1275 		.procname	= "stat_interval",
1276 		.data		= &sysctl_stat_interval,
1277 		.maxlen		= sizeof(sysctl_stat_interval),
1278 		.mode		= 0644,
1279 		.proc_handler	= &proc_dointvec_jiffies,
1280 		.strategy	= &sysctl_jiffies,
1281 	},
1282 #endif
1283 	{
1284 		.ctl_name	= CTL_UNNUMBERED,
1285 		.procname	= "mmap_min_addr",
1286 		.data		= &mmap_min_addr,
1287 		.maxlen         = sizeof(unsigned long),
1288 		.mode		= 0644,
1289 		.proc_handler	= &proc_doulongvec_minmax,
1290 	},
1291 #ifdef CONFIG_NUMA
1292 	{
1293 		.ctl_name	= CTL_UNNUMBERED,
1294 		.procname	= "numa_zonelist_order",
1295 		.data		= &numa_zonelist_order,
1296 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1297 		.mode		= 0644,
1298 		.proc_handler	= &numa_zonelist_order_handler,
1299 		.strategy	= &sysctl_string,
1300 	},
1301 #endif
1302 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1303    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1304 	{
1305 		.ctl_name	= VM_VDSO_ENABLED,
1306 		.procname	= "vdso_enabled",
1307 		.data		= &vdso_enabled,
1308 		.maxlen		= sizeof(vdso_enabled),
1309 		.mode		= 0644,
1310 		.proc_handler	= &proc_dointvec,
1311 		.strategy	= &sysctl_intvec,
1312 		.extra1		= &zero,
1313 	},
1314 #endif
1315 #ifdef CONFIG_HIGHMEM
1316 	{
1317 		.ctl_name	= CTL_UNNUMBERED,
1318 		.procname	= "highmem_is_dirtyable",
1319 		.data		= &vm_highmem_is_dirtyable,
1320 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1321 		.mode		= 0644,
1322 		.proc_handler	= &proc_dointvec_minmax,
1323 		.strategy	= &sysctl_intvec,
1324 		.extra1		= &zero,
1325 		.extra2		= &one,
1326 	},
1327 #endif
1328 #ifdef CONFIG_UNEVICTABLE_LRU
1329 	{
1330 		.ctl_name	= CTL_UNNUMBERED,
1331 		.procname	= "scan_unevictable_pages",
1332 		.data		= &scan_unevictable_pages,
1333 		.maxlen		= sizeof(scan_unevictable_pages),
1334 		.mode		= 0644,
1335 		.proc_handler	= &scan_unevictable_handler,
1336 	},
1337 #endif
1338 /*
1339  * NOTE: do not add new entries to this table unless you have read
1340  * Documentation/sysctl/ctl_unnumbered.txt
1341  */
1342 	{ .ctl_name = 0 }
1343 };
1344 
1345 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1346 static struct ctl_table binfmt_misc_table[] = {
1347 	{ .ctl_name = 0 }
1348 };
1349 #endif
1350 
1351 static struct ctl_table fs_table[] = {
1352 	{
1353 		.ctl_name	= FS_NRINODE,
1354 		.procname	= "inode-nr",
1355 		.data		= &inodes_stat,
1356 		.maxlen		= 2*sizeof(int),
1357 		.mode		= 0444,
1358 		.proc_handler	= &proc_dointvec,
1359 	},
1360 	{
1361 		.ctl_name	= FS_STATINODE,
1362 		.procname	= "inode-state",
1363 		.data		= &inodes_stat,
1364 		.maxlen		= 7*sizeof(int),
1365 		.mode		= 0444,
1366 		.proc_handler	= &proc_dointvec,
1367 	},
1368 	{
1369 		.procname	= "file-nr",
1370 		.data		= &files_stat,
1371 		.maxlen		= 3*sizeof(int),
1372 		.mode		= 0444,
1373 		.proc_handler	= &proc_nr_files,
1374 	},
1375 	{
1376 		.ctl_name	= FS_MAXFILE,
1377 		.procname	= "file-max",
1378 		.data		= &files_stat.max_files,
1379 		.maxlen		= sizeof(int),
1380 		.mode		= 0644,
1381 		.proc_handler	= &proc_dointvec,
1382 	},
1383 	{
1384 		.ctl_name	= CTL_UNNUMBERED,
1385 		.procname	= "nr_open",
1386 		.data		= &sysctl_nr_open,
1387 		.maxlen		= sizeof(int),
1388 		.mode		= 0644,
1389 		.proc_handler	= &proc_dointvec_minmax,
1390 		.extra1		= &sysctl_nr_open_min,
1391 		.extra2		= &sysctl_nr_open_max,
1392 	},
1393 	{
1394 		.ctl_name	= FS_DENTRY,
1395 		.procname	= "dentry-state",
1396 		.data		= &dentry_stat,
1397 		.maxlen		= 6*sizeof(int),
1398 		.mode		= 0444,
1399 		.proc_handler	= &proc_dointvec,
1400 	},
1401 	{
1402 		.ctl_name	= FS_OVERFLOWUID,
1403 		.procname	= "overflowuid",
1404 		.data		= &fs_overflowuid,
1405 		.maxlen		= sizeof(int),
1406 		.mode		= 0644,
1407 		.proc_handler	= &proc_dointvec_minmax,
1408 		.strategy	= &sysctl_intvec,
1409 		.extra1		= &minolduid,
1410 		.extra2		= &maxolduid,
1411 	},
1412 	{
1413 		.ctl_name	= FS_OVERFLOWGID,
1414 		.procname	= "overflowgid",
1415 		.data		= &fs_overflowgid,
1416 		.maxlen		= sizeof(int),
1417 		.mode		= 0644,
1418 		.proc_handler	= &proc_dointvec_minmax,
1419 		.strategy	= &sysctl_intvec,
1420 		.extra1		= &minolduid,
1421 		.extra2		= &maxolduid,
1422 	},
1423 #ifdef CONFIG_FILE_LOCKING
1424 	{
1425 		.ctl_name	= FS_LEASES,
1426 		.procname	= "leases-enable",
1427 		.data		= &leases_enable,
1428 		.maxlen		= sizeof(int),
1429 		.mode		= 0644,
1430 		.proc_handler	= &proc_dointvec,
1431 	},
1432 #endif
1433 #ifdef CONFIG_DNOTIFY
1434 	{
1435 		.ctl_name	= FS_DIR_NOTIFY,
1436 		.procname	= "dir-notify-enable",
1437 		.data		= &dir_notify_enable,
1438 		.maxlen		= sizeof(int),
1439 		.mode		= 0644,
1440 		.proc_handler	= &proc_dointvec,
1441 	},
1442 #endif
1443 #ifdef CONFIG_MMU
1444 #ifdef CONFIG_FILE_LOCKING
1445 	{
1446 		.ctl_name	= FS_LEASE_TIME,
1447 		.procname	= "lease-break-time",
1448 		.data		= &lease_break_time,
1449 		.maxlen		= sizeof(int),
1450 		.mode		= 0644,
1451 		.proc_handler	= &proc_dointvec,
1452 	},
1453 #endif
1454 #ifdef CONFIG_AIO
1455 	{
1456 		.procname	= "aio-nr",
1457 		.data		= &aio_nr,
1458 		.maxlen		= sizeof(aio_nr),
1459 		.mode		= 0444,
1460 		.proc_handler	= &proc_doulongvec_minmax,
1461 	},
1462 	{
1463 		.procname	= "aio-max-nr",
1464 		.data		= &aio_max_nr,
1465 		.maxlen		= sizeof(aio_max_nr),
1466 		.mode		= 0644,
1467 		.proc_handler	= &proc_doulongvec_minmax,
1468 	},
1469 #endif /* CONFIG_AIO */
1470 #ifdef CONFIG_INOTIFY_USER
1471 	{
1472 		.ctl_name	= FS_INOTIFY,
1473 		.procname	= "inotify",
1474 		.mode		= 0555,
1475 		.child		= inotify_table,
1476 	},
1477 #endif
1478 #ifdef CONFIG_EPOLL
1479 	{
1480 		.procname	= "epoll",
1481 		.mode		= 0555,
1482 		.child		= epoll_table,
1483 	},
1484 #endif
1485 #endif
1486 	{
1487 		.ctl_name	= KERN_SETUID_DUMPABLE,
1488 		.procname	= "suid_dumpable",
1489 		.data		= &suid_dumpable,
1490 		.maxlen		= sizeof(int),
1491 		.mode		= 0644,
1492 		.proc_handler	= &proc_dointvec_minmax,
1493 		.strategy	= &sysctl_intvec,
1494 		.extra1		= &zero,
1495 		.extra2		= &two,
1496 	},
1497 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1498 	{
1499 		.ctl_name	= CTL_UNNUMBERED,
1500 		.procname	= "binfmt_misc",
1501 		.mode		= 0555,
1502 		.child		= binfmt_misc_table,
1503 	},
1504 #endif
1505 /*
1506  * NOTE: do not add new entries to this table unless you have read
1507  * Documentation/sysctl/ctl_unnumbered.txt
1508  */
1509 	{ .ctl_name = 0 }
1510 };
1511 
1512 static struct ctl_table debug_table[] = {
1513 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1514 	{
1515 		.ctl_name	= CTL_UNNUMBERED,
1516 		.procname	= "exception-trace",
1517 		.data		= &show_unhandled_signals,
1518 		.maxlen		= sizeof(int),
1519 		.mode		= 0644,
1520 		.proc_handler	= proc_dointvec
1521 	},
1522 #endif
1523 	{ .ctl_name = 0 }
1524 };
1525 
1526 static struct ctl_table dev_table[] = {
1527 	{ .ctl_name = 0 }
1528 };
1529 
1530 static DEFINE_SPINLOCK(sysctl_lock);
1531 
1532 /* called under sysctl_lock */
1533 static int use_table(struct ctl_table_header *p)
1534 {
1535 	if (unlikely(p->unregistering))
1536 		return 0;
1537 	p->used++;
1538 	return 1;
1539 }
1540 
1541 /* called under sysctl_lock */
1542 static void unuse_table(struct ctl_table_header *p)
1543 {
1544 	if (!--p->used)
1545 		if (unlikely(p->unregistering))
1546 			complete(p->unregistering);
1547 }
1548 
1549 /* called under sysctl_lock, will reacquire if has to wait */
1550 static void start_unregistering(struct ctl_table_header *p)
1551 {
1552 	/*
1553 	 * if p->used is 0, nobody will ever touch that entry again;
1554 	 * we'll eliminate all paths to it before dropping sysctl_lock
1555 	 */
1556 	if (unlikely(p->used)) {
1557 		struct completion wait;
1558 		init_completion(&wait);
1559 		p->unregistering = &wait;
1560 		spin_unlock(&sysctl_lock);
1561 		wait_for_completion(&wait);
1562 		spin_lock(&sysctl_lock);
1563 	} else {
1564 		/* anything non-NULL; we'll never dereference it */
1565 		p->unregistering = ERR_PTR(-EINVAL);
1566 	}
1567 	/*
1568 	 * do not remove from the list until nobody holds it; walking the
1569 	 * list in do_sysctl() relies on that.
1570 	 */
1571 	list_del_init(&p->ctl_entry);
1572 }
1573 
1574 void sysctl_head_get(struct ctl_table_header *head)
1575 {
1576 	spin_lock(&sysctl_lock);
1577 	head->count++;
1578 	spin_unlock(&sysctl_lock);
1579 }
1580 
1581 void sysctl_head_put(struct ctl_table_header *head)
1582 {
1583 	spin_lock(&sysctl_lock);
1584 	if (!--head->count)
1585 		kfree(head);
1586 	spin_unlock(&sysctl_lock);
1587 }
1588 
1589 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1590 {
1591 	if (!head)
1592 		BUG();
1593 	spin_lock(&sysctl_lock);
1594 	if (!use_table(head))
1595 		head = ERR_PTR(-ENOENT);
1596 	spin_unlock(&sysctl_lock);
1597 	return head;
1598 }
1599 
1600 void sysctl_head_finish(struct ctl_table_header *head)
1601 {
1602 	if (!head)
1603 		return;
1604 	spin_lock(&sysctl_lock);
1605 	unuse_table(head);
1606 	spin_unlock(&sysctl_lock);
1607 }
1608 
1609 static struct ctl_table_set *
1610 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1611 {
1612 	struct ctl_table_set *set = &root->default_set;
1613 	if (root->lookup)
1614 		set = root->lookup(root, namespaces);
1615 	return set;
1616 }
1617 
1618 static struct list_head *
1619 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1620 {
1621 	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1622 	return &set->list;
1623 }
1624 
1625 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1626 					    struct ctl_table_header *prev)
1627 {
1628 	struct ctl_table_root *root;
1629 	struct list_head *header_list;
1630 	struct ctl_table_header *head;
1631 	struct list_head *tmp;
1632 
1633 	spin_lock(&sysctl_lock);
1634 	if (prev) {
1635 		head = prev;
1636 		tmp = &prev->ctl_entry;
1637 		unuse_table(prev);
1638 		goto next;
1639 	}
1640 	tmp = &root_table_header.ctl_entry;
1641 	for (;;) {
1642 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1643 
1644 		if (!use_table(head))
1645 			goto next;
1646 		spin_unlock(&sysctl_lock);
1647 		return head;
1648 	next:
1649 		root = head->root;
1650 		tmp = tmp->next;
1651 		header_list = lookup_header_list(root, namespaces);
1652 		if (tmp != header_list)
1653 			continue;
1654 
1655 		do {
1656 			root = list_entry(root->root_list.next,
1657 					struct ctl_table_root, root_list);
1658 			if (root == &sysctl_table_root)
1659 				goto out;
1660 			header_list = lookup_header_list(root, namespaces);
1661 		} while (list_empty(header_list));
1662 		tmp = header_list->next;
1663 	}
1664 out:
1665 	spin_unlock(&sysctl_lock);
1666 	return NULL;
1667 }
1668 
1669 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1670 {
1671 	return __sysctl_head_next(current->nsproxy, prev);
1672 }
1673 
1674 void register_sysctl_root(struct ctl_table_root *root)
1675 {
1676 	spin_lock(&sysctl_lock);
1677 	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1678 	spin_unlock(&sysctl_lock);
1679 }
1680 
1681 #ifdef CONFIG_SYSCTL_SYSCALL
1682 /* Perform the actual read/write of a sysctl table entry. */
1683 static int do_sysctl_strategy(struct ctl_table_root *root,
1684 			struct ctl_table *table,
1685 			void __user *oldval, size_t __user *oldlenp,
1686 			void __user *newval, size_t newlen)
1687 {
1688 	int op = 0, rc;
1689 
1690 	if (oldval)
1691 		op |= MAY_READ;
1692 	if (newval)
1693 		op |= MAY_WRITE;
1694 	if (sysctl_perm(root, table, op))
1695 		return -EPERM;
1696 
1697 	if (table->strategy) {
1698 		rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1699 		if (rc < 0)
1700 			return rc;
1701 		if (rc > 0)
1702 			return 0;
1703 	}
1704 
1705 	/* If there is no strategy routine, or if the strategy returns
1706 	 * zero, proceed with automatic r/w */
1707 	if (table->data && table->maxlen) {
1708 		rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1709 		if (rc < 0)
1710 			return rc;
1711 	}
1712 	return 0;
1713 }
1714 
1715 static int parse_table(int __user *name, int nlen,
1716 		       void __user *oldval, size_t __user *oldlenp,
1717 		       void __user *newval, size_t newlen,
1718 		       struct ctl_table_root *root,
1719 		       struct ctl_table *table)
1720 {
1721 	int n;
1722 repeat:
1723 	if (!nlen)
1724 		return -ENOTDIR;
1725 	if (get_user(n, name))
1726 		return -EFAULT;
1727 	for ( ; table->ctl_name || table->procname; table++) {
1728 		if (!table->ctl_name)
1729 			continue;
1730 		if (n == table->ctl_name) {
1731 			int error;
1732 			if (table->child) {
1733 				if (sysctl_perm(root, table, MAY_EXEC))
1734 					return -EPERM;
1735 				name++;
1736 				nlen--;
1737 				table = table->child;
1738 				goto repeat;
1739 			}
1740 			error = do_sysctl_strategy(root, table,
1741 						   oldval, oldlenp,
1742 						   newval, newlen);
1743 			return error;
1744 		}
1745 	}
1746 	return -ENOTDIR;
1747 }
1748 
1749 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1750 	       void __user *newval, size_t newlen)
1751 {
1752 	struct ctl_table_header *head;
1753 	int error = -ENOTDIR;
1754 
1755 	if (nlen <= 0 || nlen >= CTL_MAXNAME)
1756 		return -ENOTDIR;
1757 	if (oldval) {
1758 		int old_len;
1759 		if (!oldlenp || get_user(old_len, oldlenp))
1760 			return -EFAULT;
1761 	}
1762 
1763 	for (head = sysctl_head_next(NULL); head;
1764 			head = sysctl_head_next(head)) {
1765 		error = parse_table(name, nlen, oldval, oldlenp,
1766 					newval, newlen,
1767 					head->root, head->ctl_table);
1768 		if (error != -ENOTDIR) {
1769 			sysctl_head_finish(head);
1770 			break;
1771 		}
1772 	}
1773 	return error;
1774 }
1775 
1776 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
1777 {
1778 	struct __sysctl_args tmp;
1779 	int error;
1780 
1781 	if (copy_from_user(&tmp, args, sizeof(tmp)))
1782 		return -EFAULT;
1783 
1784 	error = deprecated_sysctl_warning(&tmp);
1785 	if (error)
1786 		goto out;
1787 
1788 	lock_kernel();
1789 	error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1790 			  tmp.newval, tmp.newlen);
1791 	unlock_kernel();
1792 out:
1793 	return error;
1794 }
1795 #endif /* CONFIG_SYSCTL_SYSCALL */
1796 
1797 /*
1798  * sysctl_perm does NOT grant the superuser all rights automatically, because
1799  * some sysctl variables are readonly even to root.
1800  */
1801 
1802 static int test_perm(int mode, int op)
1803 {
1804 	if (!current_euid())
1805 		mode >>= 6;
1806 	else if (in_egroup_p(0))
1807 		mode >>= 3;
1808 	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1809 		return 0;
1810 	return -EACCES;
1811 }
1812 
1813 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1814 {
1815 	int error;
1816 	int mode;
1817 
1818 	error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1819 	if (error)
1820 		return error;
1821 
1822 	if (root->permissions)
1823 		mode = root->permissions(root, current->nsproxy, table);
1824 	else
1825 		mode = table->mode;
1826 
1827 	return test_perm(mode, op);
1828 }
1829 
1830 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1831 {
1832 	for (; table->ctl_name || table->procname; table++) {
1833 		table->parent = parent;
1834 		if (table->child)
1835 			sysctl_set_parent(table, table->child);
1836 	}
1837 }
1838 
1839 static __init int sysctl_init(void)
1840 {
1841 	sysctl_set_parent(NULL, root_table);
1842 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1843 	{
1844 		int err;
1845 		err = sysctl_check_table(current->nsproxy, root_table);
1846 	}
1847 #endif
1848 	return 0;
1849 }
1850 
1851 core_initcall(sysctl_init);
1852 
1853 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1854 				      struct ctl_table *table)
1855 {
1856 	struct ctl_table *p;
1857 	const char *s = branch->procname;
1858 
1859 	/* branch should have named subdirectory as its first element */
1860 	if (!s || !branch->child)
1861 		return NULL;
1862 
1863 	/* ... and nothing else */
1864 	if (branch[1].procname || branch[1].ctl_name)
1865 		return NULL;
1866 
1867 	/* table should contain subdirectory with the same name */
1868 	for (p = table; p->procname || p->ctl_name; p++) {
1869 		if (!p->child)
1870 			continue;
1871 		if (p->procname && strcmp(p->procname, s) == 0)
1872 			return p;
1873 	}
1874 	return NULL;
1875 }
1876 
1877 /* see if attaching q to p would be an improvement */
1878 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1879 {
1880 	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1881 	struct ctl_table *next;
1882 	int is_better = 0;
1883 	int not_in_parent = !p->attached_by;
1884 
1885 	while ((next = is_branch_in(by, to)) != NULL) {
1886 		if (by == q->attached_by)
1887 			is_better = 1;
1888 		if (to == p->attached_by)
1889 			not_in_parent = 1;
1890 		by = by->child;
1891 		to = next->child;
1892 	}
1893 
1894 	if (is_better && not_in_parent) {
1895 		q->attached_by = by;
1896 		q->attached_to = to;
1897 		q->parent = p;
1898 	}
1899 }
1900 
1901 /**
1902  * __register_sysctl_paths - register a sysctl hierarchy
1903  * @root: List of sysctl headers to register on
1904  * @namespaces: Data to compute which lists of sysctl entries are visible
1905  * @path: The path to the directory the sysctl table is in.
1906  * @table: the top-level table structure
1907  *
1908  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1909  * array. A completely 0 filled entry terminates the table.
1910  *
1911  * The members of the &struct ctl_table structure are used as follows:
1912  *
1913  * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1914  *            must be unique within that level of sysctl
1915  *
1916  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1917  *            enter a sysctl file
1918  *
1919  * data - a pointer to data for use by proc_handler
1920  *
1921  * maxlen - the maximum size in bytes of the data
1922  *
1923  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1924  *
1925  * child - a pointer to the child sysctl table if this entry is a directory, or
1926  *         %NULL.
1927  *
1928  * proc_handler - the text handler routine (described below)
1929  *
1930  * strategy - the strategy routine (described below)
1931  *
1932  * de - for internal use by the sysctl routines
1933  *
1934  * extra1, extra2 - extra pointers usable by the proc handler routines
1935  *
1936  * Leaf nodes in the sysctl tree will be represented by a single file
1937  * under /proc; non-leaf nodes will be represented by directories.
1938  *
1939  * sysctl(2) can automatically manage read and write requests through
1940  * the sysctl table.  The data and maxlen fields of the ctl_table
1941  * struct enable minimal validation of the values being written to be
1942  * performed, and the mode field allows minimal authentication.
1943  *
1944  * More sophisticated management can be enabled by the provision of a
1945  * strategy routine with the table entry.  This will be called before
1946  * any automatic read or write of the data is performed.
1947  *
1948  * The strategy routine may return
1949  *
1950  * < 0 - Error occurred (error is passed to user process)
1951  *
1952  * 0   - OK - proceed with automatic read or write.
1953  *
1954  * > 0 - OK - read or write has been done by the strategy routine, so
1955  *       return immediately.
1956  *
1957  * There must be a proc_handler routine for any terminal nodes
1958  * mirrored under /proc/sys (non-terminals are handled by a built-in
1959  * directory handler).  Several default handlers are available to
1960  * cover common cases -
1961  *
1962  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1963  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1964  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1965  *
1966  * It is the handler's job to read the input buffer from user memory
1967  * and process it. The handler should return 0 on success.
1968  *
1969  * This routine returns %NULL on a failure to register, and a pointer
1970  * to the table header on success.
1971  */
1972 struct ctl_table_header *__register_sysctl_paths(
1973 	struct ctl_table_root *root,
1974 	struct nsproxy *namespaces,
1975 	const struct ctl_path *path, struct ctl_table *table)
1976 {
1977 	struct ctl_table_header *header;
1978 	struct ctl_table *new, **prevp;
1979 	unsigned int n, npath;
1980 	struct ctl_table_set *set;
1981 
1982 	/* Count the path components */
1983 	for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1984 		;
1985 
1986 	/*
1987 	 * For each path component, allocate a 2-element ctl_table array.
1988 	 * The first array element will be filled with the sysctl entry
1989 	 * for this, the second will be the sentinel (ctl_name == 0).
1990 	 *
1991 	 * We allocate everything in one go so that we don't have to
1992 	 * worry about freeing additional memory in unregister_sysctl_table.
1993 	 */
1994 	header = kzalloc(sizeof(struct ctl_table_header) +
1995 			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1996 	if (!header)
1997 		return NULL;
1998 
1999 	new = (struct ctl_table *) (header + 1);
2000 
2001 	/* Now connect the dots */
2002 	prevp = &header->ctl_table;
2003 	for (n = 0; n < npath; ++n, ++path) {
2004 		/* Copy the procname */
2005 		new->procname = path->procname;
2006 		new->ctl_name = path->ctl_name;
2007 		new->mode     = 0555;
2008 
2009 		*prevp = new;
2010 		prevp = &new->child;
2011 
2012 		new += 2;
2013 	}
2014 	*prevp = table;
2015 	header->ctl_table_arg = table;
2016 
2017 	INIT_LIST_HEAD(&header->ctl_entry);
2018 	header->used = 0;
2019 	header->unregistering = NULL;
2020 	header->root = root;
2021 	sysctl_set_parent(NULL, header->ctl_table);
2022 	header->count = 1;
2023 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
2024 	if (sysctl_check_table(namespaces, header->ctl_table)) {
2025 		kfree(header);
2026 		return NULL;
2027 	}
2028 #endif
2029 	spin_lock(&sysctl_lock);
2030 	header->set = lookup_header_set(root, namespaces);
2031 	header->attached_by = header->ctl_table;
2032 	header->attached_to = root_table;
2033 	header->parent = &root_table_header;
2034 	for (set = header->set; set; set = set->parent) {
2035 		struct ctl_table_header *p;
2036 		list_for_each_entry(p, &set->list, ctl_entry) {
2037 			if (p->unregistering)
2038 				continue;
2039 			try_attach(p, header);
2040 		}
2041 	}
2042 	header->parent->count++;
2043 	list_add_tail(&header->ctl_entry, &header->set->list);
2044 	spin_unlock(&sysctl_lock);
2045 
2046 	return header;
2047 }
2048 
2049 /**
2050  * register_sysctl_table_path - register a sysctl table hierarchy
2051  * @path: The path to the directory the sysctl table is in.
2052  * @table: the top-level table structure
2053  *
2054  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2055  * array. A completely 0 filled entry terminates the table.
2056  *
2057  * See __register_sysctl_paths for more details.
2058  */
2059 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2060 						struct ctl_table *table)
2061 {
2062 	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
2063 					path, table);
2064 }
2065 
2066 /**
2067  * register_sysctl_table - register a sysctl table hierarchy
2068  * @table: the top-level table structure
2069  *
2070  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2071  * array. A completely 0 filled entry terminates the table.
2072  *
2073  * See register_sysctl_paths for more details.
2074  */
2075 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
2076 {
2077 	static const struct ctl_path null_path[] = { {} };
2078 
2079 	return register_sysctl_paths(null_path, table);
2080 }
2081 
2082 /**
2083  * unregister_sysctl_table - unregister a sysctl table hierarchy
2084  * @header: the header returned from register_sysctl_table
2085  *
2086  * Unregisters the sysctl table and all children. proc entries may not
2087  * actually be removed until they are no longer used by anyone.
2088  */
2089 void unregister_sysctl_table(struct ctl_table_header * header)
2090 {
2091 	might_sleep();
2092 
2093 	if (header == NULL)
2094 		return;
2095 
2096 	spin_lock(&sysctl_lock);
2097 	start_unregistering(header);
2098 	if (!--header->parent->count) {
2099 		WARN_ON(1);
2100 		kfree(header->parent);
2101 	}
2102 	if (!--header->count)
2103 		kfree(header);
2104 	spin_unlock(&sysctl_lock);
2105 }
2106 
2107 int sysctl_is_seen(struct ctl_table_header *p)
2108 {
2109 	struct ctl_table_set *set = p->set;
2110 	int res;
2111 	spin_lock(&sysctl_lock);
2112 	if (p->unregistering)
2113 		res = 0;
2114 	else if (!set->is_seen)
2115 		res = 1;
2116 	else
2117 		res = set->is_seen(set);
2118 	spin_unlock(&sysctl_lock);
2119 	return res;
2120 }
2121 
2122 void setup_sysctl_set(struct ctl_table_set *p,
2123 	struct ctl_table_set *parent,
2124 	int (*is_seen)(struct ctl_table_set *))
2125 {
2126 	INIT_LIST_HEAD(&p->list);
2127 	p->parent = parent ? parent : &sysctl_table_root.default_set;
2128 	p->is_seen = is_seen;
2129 }
2130 
2131 #else /* !CONFIG_SYSCTL */
2132 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2133 {
2134 	return NULL;
2135 }
2136 
2137 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2138 						    struct ctl_table *table)
2139 {
2140 	return NULL;
2141 }
2142 
2143 void unregister_sysctl_table(struct ctl_table_header * table)
2144 {
2145 }
2146 
2147 void setup_sysctl_set(struct ctl_table_set *p,
2148 	struct ctl_table_set *parent,
2149 	int (*is_seen)(struct ctl_table_set *))
2150 {
2151 }
2152 
2153 void sysctl_head_put(struct ctl_table_header *head)
2154 {
2155 }
2156 
2157 #endif /* CONFIG_SYSCTL */
2158 
2159 /*
2160  * /proc/sys support
2161  */
2162 
2163 #ifdef CONFIG_PROC_SYSCTL
2164 
2165 static int _proc_do_string(void* data, int maxlen, int write,
2166 			   struct file *filp, void __user *buffer,
2167 			   size_t *lenp, loff_t *ppos)
2168 {
2169 	size_t len;
2170 	char __user *p;
2171 	char c;
2172 
2173 	if (!data || !maxlen || !*lenp) {
2174 		*lenp = 0;
2175 		return 0;
2176 	}
2177 
2178 	if (write) {
2179 		len = 0;
2180 		p = buffer;
2181 		while (len < *lenp) {
2182 			if (get_user(c, p++))
2183 				return -EFAULT;
2184 			if (c == 0 || c == '\n')
2185 				break;
2186 			len++;
2187 		}
2188 		if (len >= maxlen)
2189 			len = maxlen-1;
2190 		if(copy_from_user(data, buffer, len))
2191 			return -EFAULT;
2192 		((char *) data)[len] = 0;
2193 		*ppos += *lenp;
2194 	} else {
2195 		len = strlen(data);
2196 		if (len > maxlen)
2197 			len = maxlen;
2198 
2199 		if (*ppos > len) {
2200 			*lenp = 0;
2201 			return 0;
2202 		}
2203 
2204 		data += *ppos;
2205 		len  -= *ppos;
2206 
2207 		if (len > *lenp)
2208 			len = *lenp;
2209 		if (len)
2210 			if(copy_to_user(buffer, data, len))
2211 				return -EFAULT;
2212 		if (len < *lenp) {
2213 			if(put_user('\n', ((char __user *) buffer) + len))
2214 				return -EFAULT;
2215 			len++;
2216 		}
2217 		*lenp = len;
2218 		*ppos += len;
2219 	}
2220 	return 0;
2221 }
2222 
2223 /**
2224  * proc_dostring - read a string sysctl
2225  * @table: the sysctl table
2226  * @write: %TRUE if this is a write to the sysctl file
2227  * @filp: the file structure
2228  * @buffer: the user buffer
2229  * @lenp: the size of the user buffer
2230  * @ppos: file position
2231  *
2232  * Reads/writes a string from/to the user buffer. If the kernel
2233  * buffer provided is not large enough to hold the string, the
2234  * string is truncated. The copied string is %NULL-terminated.
2235  * If the string is being read by the user process, it is copied
2236  * and a newline '\n' is added. It is truncated if the buffer is
2237  * not large enough.
2238  *
2239  * Returns 0 on success.
2240  */
2241 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2242 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2243 {
2244 	return _proc_do_string(table->data, table->maxlen, write, filp,
2245 			       buffer, lenp, ppos);
2246 }
2247 
2248 
2249 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2250 				 int *valp,
2251 				 int write, void *data)
2252 {
2253 	if (write) {
2254 		*valp = *negp ? -*lvalp : *lvalp;
2255 	} else {
2256 		int val = *valp;
2257 		if (val < 0) {
2258 			*negp = -1;
2259 			*lvalp = (unsigned long)-val;
2260 		} else {
2261 			*negp = 0;
2262 			*lvalp = (unsigned long)val;
2263 		}
2264 	}
2265 	return 0;
2266 }
2267 
2268 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2269 		  int write, struct file *filp, void __user *buffer,
2270 		  size_t *lenp, loff_t *ppos,
2271 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2272 			      int write, void *data),
2273 		  void *data)
2274 {
2275 #define TMPBUFLEN 21
2276 	int *i, vleft, first=1, neg, val;
2277 	unsigned long lval;
2278 	size_t left, len;
2279 
2280 	char buf[TMPBUFLEN], *p;
2281 	char __user *s = buffer;
2282 
2283 	if (!tbl_data || !table->maxlen || !*lenp ||
2284 	    (*ppos && !write)) {
2285 		*lenp = 0;
2286 		return 0;
2287 	}
2288 
2289 	i = (int *) tbl_data;
2290 	vleft = table->maxlen / sizeof(*i);
2291 	left = *lenp;
2292 
2293 	if (!conv)
2294 		conv = do_proc_dointvec_conv;
2295 
2296 	for (; left && vleft--; i++, first=0) {
2297 		if (write) {
2298 			while (left) {
2299 				char c;
2300 				if (get_user(c, s))
2301 					return -EFAULT;
2302 				if (!isspace(c))
2303 					break;
2304 				left--;
2305 				s++;
2306 			}
2307 			if (!left)
2308 				break;
2309 			neg = 0;
2310 			len = left;
2311 			if (len > sizeof(buf) - 1)
2312 				len = sizeof(buf) - 1;
2313 			if (copy_from_user(buf, s, len))
2314 				return -EFAULT;
2315 			buf[len] = 0;
2316 			p = buf;
2317 			if (*p == '-' && left > 1) {
2318 				neg = 1;
2319 				p++;
2320 			}
2321 			if (*p < '0' || *p > '9')
2322 				break;
2323 
2324 			lval = simple_strtoul(p, &p, 0);
2325 
2326 			len = p-buf;
2327 			if ((len < left) && *p && !isspace(*p))
2328 				break;
2329 			if (neg)
2330 				val = -val;
2331 			s += len;
2332 			left -= len;
2333 
2334 			if (conv(&neg, &lval, i, 1, data))
2335 				break;
2336 		} else {
2337 			p = buf;
2338 			if (!first)
2339 				*p++ = '\t';
2340 
2341 			if (conv(&neg, &lval, i, 0, data))
2342 				break;
2343 
2344 			sprintf(p, "%s%lu", neg ? "-" : "", lval);
2345 			len = strlen(buf);
2346 			if (len > left)
2347 				len = left;
2348 			if(copy_to_user(s, buf, len))
2349 				return -EFAULT;
2350 			left -= len;
2351 			s += len;
2352 		}
2353 	}
2354 
2355 	if (!write && !first && left) {
2356 		if(put_user('\n', s))
2357 			return -EFAULT;
2358 		left--, s++;
2359 	}
2360 	if (write) {
2361 		while (left) {
2362 			char c;
2363 			if (get_user(c, s++))
2364 				return -EFAULT;
2365 			if (!isspace(c))
2366 				break;
2367 			left--;
2368 		}
2369 	}
2370 	if (write && first)
2371 		return -EINVAL;
2372 	*lenp -= left;
2373 	*ppos += *lenp;
2374 	return 0;
2375 #undef TMPBUFLEN
2376 }
2377 
2378 static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2379 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2380 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2381 			      int write, void *data),
2382 		  void *data)
2383 {
2384 	return __do_proc_dointvec(table->data, table, write, filp,
2385 			buffer, lenp, ppos, conv, data);
2386 }
2387 
2388 /**
2389  * proc_dointvec - read a vector of integers
2390  * @table: the sysctl table
2391  * @write: %TRUE if this is a write to the sysctl file
2392  * @filp: the file structure
2393  * @buffer: the user buffer
2394  * @lenp: the size of the user buffer
2395  * @ppos: file position
2396  *
2397  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2398  * values from/to the user buffer, treated as an ASCII string.
2399  *
2400  * Returns 0 on success.
2401  */
2402 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2403 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2404 {
2405     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2406 		    	    NULL,NULL);
2407 }
2408 
2409 /*
2410  * Taint values can only be increased
2411  * This means we can safely use a temporary.
2412  */
2413 static int proc_taint(struct ctl_table *table, int write, struct file *filp,
2414 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2415 {
2416 	struct ctl_table t;
2417 	unsigned long tmptaint = get_taint();
2418 	int err;
2419 
2420 	if (write && !capable(CAP_SYS_ADMIN))
2421 		return -EPERM;
2422 
2423 	t = *table;
2424 	t.data = &tmptaint;
2425 	err = proc_doulongvec_minmax(&t, write, filp, buffer, lenp, ppos);
2426 	if (err < 0)
2427 		return err;
2428 
2429 	if (write) {
2430 		/*
2431 		 * Poor man's atomic or. Not worth adding a primitive
2432 		 * to everyone's atomic.h for this
2433 		 */
2434 		int i;
2435 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2436 			if ((tmptaint >> i) & 1)
2437 				add_taint(i);
2438 		}
2439 	}
2440 
2441 	return err;
2442 }
2443 
2444 struct do_proc_dointvec_minmax_conv_param {
2445 	int *min;
2446 	int *max;
2447 };
2448 
2449 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2450 					int *valp,
2451 					int write, void *data)
2452 {
2453 	struct do_proc_dointvec_minmax_conv_param *param = data;
2454 	if (write) {
2455 		int val = *negp ? -*lvalp : *lvalp;
2456 		if ((param->min && *param->min > val) ||
2457 		    (param->max && *param->max < val))
2458 			return -EINVAL;
2459 		*valp = val;
2460 	} else {
2461 		int val = *valp;
2462 		if (val < 0) {
2463 			*negp = -1;
2464 			*lvalp = (unsigned long)-val;
2465 		} else {
2466 			*negp = 0;
2467 			*lvalp = (unsigned long)val;
2468 		}
2469 	}
2470 	return 0;
2471 }
2472 
2473 /**
2474  * proc_dointvec_minmax - read a vector of integers with min/max values
2475  * @table: the sysctl table
2476  * @write: %TRUE if this is a write to the sysctl file
2477  * @filp: the file structure
2478  * @buffer: the user buffer
2479  * @lenp: the size of the user buffer
2480  * @ppos: file position
2481  *
2482  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2483  * values from/to the user buffer, treated as an ASCII string.
2484  *
2485  * This routine will ensure the values are within the range specified by
2486  * table->extra1 (min) and table->extra2 (max).
2487  *
2488  * Returns 0 on success.
2489  */
2490 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2491 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2492 {
2493 	struct do_proc_dointvec_minmax_conv_param param = {
2494 		.min = (int *) table->extra1,
2495 		.max = (int *) table->extra2,
2496 	};
2497 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2498 				do_proc_dointvec_minmax_conv, &param);
2499 }
2500 
2501 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2502 				     struct file *filp,
2503 				     void __user *buffer,
2504 				     size_t *lenp, loff_t *ppos,
2505 				     unsigned long convmul,
2506 				     unsigned long convdiv)
2507 {
2508 #define TMPBUFLEN 21
2509 	unsigned long *i, *min, *max, val;
2510 	int vleft, first=1, neg;
2511 	size_t len, left;
2512 	char buf[TMPBUFLEN], *p;
2513 	char __user *s = buffer;
2514 
2515 	if (!data || !table->maxlen || !*lenp ||
2516 	    (*ppos && !write)) {
2517 		*lenp = 0;
2518 		return 0;
2519 	}
2520 
2521 	i = (unsigned long *) data;
2522 	min = (unsigned long *) table->extra1;
2523 	max = (unsigned long *) table->extra2;
2524 	vleft = table->maxlen / sizeof(unsigned long);
2525 	left = *lenp;
2526 
2527 	for (; left && vleft--; i++, min++, max++, first=0) {
2528 		if (write) {
2529 			while (left) {
2530 				char c;
2531 				if (get_user(c, s))
2532 					return -EFAULT;
2533 				if (!isspace(c))
2534 					break;
2535 				left--;
2536 				s++;
2537 			}
2538 			if (!left)
2539 				break;
2540 			neg = 0;
2541 			len = left;
2542 			if (len > TMPBUFLEN-1)
2543 				len = TMPBUFLEN-1;
2544 			if (copy_from_user(buf, s, len))
2545 				return -EFAULT;
2546 			buf[len] = 0;
2547 			p = buf;
2548 			if (*p == '-' && left > 1) {
2549 				neg = 1;
2550 				p++;
2551 			}
2552 			if (*p < '0' || *p > '9')
2553 				break;
2554 			val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2555 			len = p-buf;
2556 			if ((len < left) && *p && !isspace(*p))
2557 				break;
2558 			if (neg)
2559 				val = -val;
2560 			s += len;
2561 			left -= len;
2562 
2563 			if(neg)
2564 				continue;
2565 			if ((min && val < *min) || (max && val > *max))
2566 				continue;
2567 			*i = val;
2568 		} else {
2569 			p = buf;
2570 			if (!first)
2571 				*p++ = '\t';
2572 			sprintf(p, "%lu", convdiv * (*i) / convmul);
2573 			len = strlen(buf);
2574 			if (len > left)
2575 				len = left;
2576 			if(copy_to_user(s, buf, len))
2577 				return -EFAULT;
2578 			left -= len;
2579 			s += len;
2580 		}
2581 	}
2582 
2583 	if (!write && !first && left) {
2584 		if(put_user('\n', s))
2585 			return -EFAULT;
2586 		left--, s++;
2587 	}
2588 	if (write) {
2589 		while (left) {
2590 			char c;
2591 			if (get_user(c, s++))
2592 				return -EFAULT;
2593 			if (!isspace(c))
2594 				break;
2595 			left--;
2596 		}
2597 	}
2598 	if (write && first)
2599 		return -EINVAL;
2600 	*lenp -= left;
2601 	*ppos += *lenp;
2602 	return 0;
2603 #undef TMPBUFLEN
2604 }
2605 
2606 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2607 				     struct file *filp,
2608 				     void __user *buffer,
2609 				     size_t *lenp, loff_t *ppos,
2610 				     unsigned long convmul,
2611 				     unsigned long convdiv)
2612 {
2613 	return __do_proc_doulongvec_minmax(table->data, table, write,
2614 			filp, buffer, lenp, ppos, convmul, convdiv);
2615 }
2616 
2617 /**
2618  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2619  * @table: the sysctl table
2620  * @write: %TRUE if this is a write to the sysctl file
2621  * @filp: the file structure
2622  * @buffer: the user buffer
2623  * @lenp: the size of the user buffer
2624  * @ppos: file position
2625  *
2626  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2627  * values from/to the user buffer, treated as an ASCII string.
2628  *
2629  * This routine will ensure the values are within the range specified by
2630  * table->extra1 (min) and table->extra2 (max).
2631  *
2632  * Returns 0 on success.
2633  */
2634 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2635 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2636 {
2637     return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2638 }
2639 
2640 /**
2641  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2642  * @table: the sysctl table
2643  * @write: %TRUE if this is a write to the sysctl file
2644  * @filp: the file structure
2645  * @buffer: the user buffer
2646  * @lenp: the size of the user buffer
2647  * @ppos: file position
2648  *
2649  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2650  * values from/to the user buffer, treated as an ASCII string. The values
2651  * are treated as milliseconds, and converted to jiffies when they are stored.
2652  *
2653  * This routine will ensure the values are within the range specified by
2654  * table->extra1 (min) and table->extra2 (max).
2655  *
2656  * Returns 0 on success.
2657  */
2658 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2659 				      struct file *filp,
2660 				      void __user *buffer,
2661 				      size_t *lenp, loff_t *ppos)
2662 {
2663     return do_proc_doulongvec_minmax(table, write, filp, buffer,
2664 				     lenp, ppos, HZ, 1000l);
2665 }
2666 
2667 
2668 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2669 					 int *valp,
2670 					 int write, void *data)
2671 {
2672 	if (write) {
2673 		if (*lvalp > LONG_MAX / HZ)
2674 			return 1;
2675 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2676 	} else {
2677 		int val = *valp;
2678 		unsigned long lval;
2679 		if (val < 0) {
2680 			*negp = -1;
2681 			lval = (unsigned long)-val;
2682 		} else {
2683 			*negp = 0;
2684 			lval = (unsigned long)val;
2685 		}
2686 		*lvalp = lval / HZ;
2687 	}
2688 	return 0;
2689 }
2690 
2691 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2692 						int *valp,
2693 						int write, void *data)
2694 {
2695 	if (write) {
2696 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2697 			return 1;
2698 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2699 	} else {
2700 		int val = *valp;
2701 		unsigned long lval;
2702 		if (val < 0) {
2703 			*negp = -1;
2704 			lval = (unsigned long)-val;
2705 		} else {
2706 			*negp = 0;
2707 			lval = (unsigned long)val;
2708 		}
2709 		*lvalp = jiffies_to_clock_t(lval);
2710 	}
2711 	return 0;
2712 }
2713 
2714 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2715 					    int *valp,
2716 					    int write, void *data)
2717 {
2718 	if (write) {
2719 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2720 	} else {
2721 		int val = *valp;
2722 		unsigned long lval;
2723 		if (val < 0) {
2724 			*negp = -1;
2725 			lval = (unsigned long)-val;
2726 		} else {
2727 			*negp = 0;
2728 			lval = (unsigned long)val;
2729 		}
2730 		*lvalp = jiffies_to_msecs(lval);
2731 	}
2732 	return 0;
2733 }
2734 
2735 /**
2736  * proc_dointvec_jiffies - read a vector of integers as seconds
2737  * @table: the sysctl table
2738  * @write: %TRUE if this is a write to the sysctl file
2739  * @filp: the file structure
2740  * @buffer: the user buffer
2741  * @lenp: the size of the user buffer
2742  * @ppos: file position
2743  *
2744  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2745  * values from/to the user buffer, treated as an ASCII string.
2746  * The values read are assumed to be in seconds, and are converted into
2747  * jiffies.
2748  *
2749  * Returns 0 on success.
2750  */
2751 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2752 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2753 {
2754     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2755 		    	    do_proc_dointvec_jiffies_conv,NULL);
2756 }
2757 
2758 /**
2759  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2760  * @table: the sysctl table
2761  * @write: %TRUE if this is a write to the sysctl file
2762  * @filp: the file structure
2763  * @buffer: the user buffer
2764  * @lenp: the size of the user buffer
2765  * @ppos: pointer to the file position
2766  *
2767  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2768  * values from/to the user buffer, treated as an ASCII string.
2769  * The values read are assumed to be in 1/USER_HZ seconds, and
2770  * are converted into jiffies.
2771  *
2772  * Returns 0 on success.
2773  */
2774 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2775 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2776 {
2777     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2778 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2779 }
2780 
2781 /**
2782  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2783  * @table: the sysctl table
2784  * @write: %TRUE if this is a write to the sysctl file
2785  * @filp: the file structure
2786  * @buffer: the user buffer
2787  * @lenp: the size of the user buffer
2788  * @ppos: file position
2789  * @ppos: the current position in the file
2790  *
2791  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2792  * values from/to the user buffer, treated as an ASCII string.
2793  * The values read are assumed to be in 1/1000 seconds, and
2794  * are converted into jiffies.
2795  *
2796  * Returns 0 on success.
2797  */
2798 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2799 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2800 {
2801 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2802 				do_proc_dointvec_ms_jiffies_conv, NULL);
2803 }
2804 
2805 static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2806 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2807 {
2808 	struct pid *new_pid;
2809 	pid_t tmp;
2810 	int r;
2811 
2812 	tmp = pid_vnr(cad_pid);
2813 
2814 	r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2815 			       lenp, ppos, NULL, NULL);
2816 	if (r || !write)
2817 		return r;
2818 
2819 	new_pid = find_get_pid(tmp);
2820 	if (!new_pid)
2821 		return -ESRCH;
2822 
2823 	put_pid(xchg(&cad_pid, new_pid));
2824 	return 0;
2825 }
2826 
2827 #else /* CONFIG_PROC_FS */
2828 
2829 int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2830 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2831 {
2832 	return -ENOSYS;
2833 }
2834 
2835 int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2836 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2837 {
2838 	return -ENOSYS;
2839 }
2840 
2841 int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2842 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2843 {
2844 	return -ENOSYS;
2845 }
2846 
2847 int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2848 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2849 {
2850 	return -ENOSYS;
2851 }
2852 
2853 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2854 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2855 {
2856 	return -ENOSYS;
2857 }
2858 
2859 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2860 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2861 {
2862 	return -ENOSYS;
2863 }
2864 
2865 int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2866 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2867 {
2868 	return -ENOSYS;
2869 }
2870 
2871 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2872 				      struct file *filp,
2873 				      void __user *buffer,
2874 				      size_t *lenp, loff_t *ppos)
2875 {
2876     return -ENOSYS;
2877 }
2878 
2879 
2880 #endif /* CONFIG_PROC_FS */
2881 
2882 
2883 #ifdef CONFIG_SYSCTL_SYSCALL
2884 /*
2885  * General sysctl support routines
2886  */
2887 
2888 /* The generic sysctl data routine (used if no strategy routine supplied) */
2889 int sysctl_data(struct ctl_table *table,
2890 		void __user *oldval, size_t __user *oldlenp,
2891 		void __user *newval, size_t newlen)
2892 {
2893 	size_t len;
2894 
2895 	/* Get out of I don't have a variable */
2896 	if (!table->data || !table->maxlen)
2897 		return -ENOTDIR;
2898 
2899 	if (oldval && oldlenp) {
2900 		if (get_user(len, oldlenp))
2901 			return -EFAULT;
2902 		if (len) {
2903 			if (len > table->maxlen)
2904 				len = table->maxlen;
2905 			if (copy_to_user(oldval, table->data, len))
2906 				return -EFAULT;
2907 			if (put_user(len, oldlenp))
2908 				return -EFAULT;
2909 		}
2910 	}
2911 
2912 	if (newval && newlen) {
2913 		if (newlen > table->maxlen)
2914 			newlen = table->maxlen;
2915 
2916 		if (copy_from_user(table->data, newval, newlen))
2917 			return -EFAULT;
2918 	}
2919 	return 1;
2920 }
2921 
2922 /* The generic string strategy routine: */
2923 int sysctl_string(struct ctl_table *table,
2924 		  void __user *oldval, size_t __user *oldlenp,
2925 		  void __user *newval, size_t newlen)
2926 {
2927 	if (!table->data || !table->maxlen)
2928 		return -ENOTDIR;
2929 
2930 	if (oldval && oldlenp) {
2931 		size_t bufsize;
2932 		if (get_user(bufsize, oldlenp))
2933 			return -EFAULT;
2934 		if (bufsize) {
2935 			size_t len = strlen(table->data), copied;
2936 
2937 			/* This shouldn't trigger for a well-formed sysctl */
2938 			if (len > table->maxlen)
2939 				len = table->maxlen;
2940 
2941 			/* Copy up to a max of bufsize-1 bytes of the string */
2942 			copied = (len >= bufsize) ? bufsize - 1 : len;
2943 
2944 			if (copy_to_user(oldval, table->data, copied) ||
2945 			    put_user(0, (char __user *)(oldval + copied)))
2946 				return -EFAULT;
2947 			if (put_user(len, oldlenp))
2948 				return -EFAULT;
2949 		}
2950 	}
2951 	if (newval && newlen) {
2952 		size_t len = newlen;
2953 		if (len > table->maxlen)
2954 			len = table->maxlen;
2955 		if(copy_from_user(table->data, newval, len))
2956 			return -EFAULT;
2957 		if (len == table->maxlen)
2958 			len--;
2959 		((char *) table->data)[len] = 0;
2960 	}
2961 	return 1;
2962 }
2963 
2964 /*
2965  * This function makes sure that all of the integers in the vector
2966  * are between the minimum and maximum values given in the arrays
2967  * table->extra1 and table->extra2, respectively.
2968  */
2969 int sysctl_intvec(struct ctl_table *table,
2970 		void __user *oldval, size_t __user *oldlenp,
2971 		void __user *newval, size_t newlen)
2972 {
2973 
2974 	if (newval && newlen) {
2975 		int __user *vec = (int __user *) newval;
2976 		int *min = (int *) table->extra1;
2977 		int *max = (int *) table->extra2;
2978 		size_t length;
2979 		int i;
2980 
2981 		if (newlen % sizeof(int) != 0)
2982 			return -EINVAL;
2983 
2984 		if (!table->extra1 && !table->extra2)
2985 			return 0;
2986 
2987 		if (newlen > table->maxlen)
2988 			newlen = table->maxlen;
2989 		length = newlen / sizeof(int);
2990 
2991 		for (i = 0; i < length; i++) {
2992 			int value;
2993 			if (get_user(value, vec + i))
2994 				return -EFAULT;
2995 			if (min && value < min[i])
2996 				return -EINVAL;
2997 			if (max && value > max[i])
2998 				return -EINVAL;
2999 		}
3000 	}
3001 	return 0;
3002 }
3003 
3004 /* Strategy function to convert jiffies to seconds */
3005 int sysctl_jiffies(struct ctl_table *table,
3006 		void __user *oldval, size_t __user *oldlenp,
3007 		void __user *newval, size_t newlen)
3008 {
3009 	if (oldval && oldlenp) {
3010 		size_t olen;
3011 
3012 		if (get_user(olen, oldlenp))
3013 			return -EFAULT;
3014 		if (olen) {
3015 			int val;
3016 
3017 			if (olen < sizeof(int))
3018 				return -EINVAL;
3019 
3020 			val = *(int *)(table->data) / HZ;
3021 			if (put_user(val, (int __user *)oldval))
3022 				return -EFAULT;
3023 			if (put_user(sizeof(int), oldlenp))
3024 				return -EFAULT;
3025 		}
3026 	}
3027 	if (newval && newlen) {
3028 		int new;
3029 		if (newlen != sizeof(int))
3030 			return -EINVAL;
3031 		if (get_user(new, (int __user *)newval))
3032 			return -EFAULT;
3033 		*(int *)(table->data) = new*HZ;
3034 	}
3035 	return 1;
3036 }
3037 
3038 /* Strategy function to convert jiffies to seconds */
3039 int sysctl_ms_jiffies(struct ctl_table *table,
3040 		void __user *oldval, size_t __user *oldlenp,
3041 		void __user *newval, size_t newlen)
3042 {
3043 	if (oldval && oldlenp) {
3044 		size_t olen;
3045 
3046 		if (get_user(olen, oldlenp))
3047 			return -EFAULT;
3048 		if (olen) {
3049 			int val;
3050 
3051 			if (olen < sizeof(int))
3052 				return -EINVAL;
3053 
3054 			val = jiffies_to_msecs(*(int *)(table->data));
3055 			if (put_user(val, (int __user *)oldval))
3056 				return -EFAULT;
3057 			if (put_user(sizeof(int), oldlenp))
3058 				return -EFAULT;
3059 		}
3060 	}
3061 	if (newval && newlen) {
3062 		int new;
3063 		if (newlen != sizeof(int))
3064 			return -EINVAL;
3065 		if (get_user(new, (int __user *)newval))
3066 			return -EFAULT;
3067 		*(int *)(table->data) = msecs_to_jiffies(new);
3068 	}
3069 	return 1;
3070 }
3071 
3072 
3073 
3074 #else /* CONFIG_SYSCTL_SYSCALL */
3075 
3076 
3077 SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
3078 {
3079 	struct __sysctl_args tmp;
3080 	int error;
3081 
3082 	if (copy_from_user(&tmp, args, sizeof(tmp)))
3083 		return -EFAULT;
3084 
3085 	error = deprecated_sysctl_warning(&tmp);
3086 
3087 	/* If no error reading the parameters then just -ENOSYS ... */
3088 	if (!error)
3089 		error = -ENOSYS;
3090 
3091 	return error;
3092 }
3093 
3094 int sysctl_data(struct ctl_table *table,
3095 		  void __user *oldval, size_t __user *oldlenp,
3096 		  void __user *newval, size_t newlen)
3097 {
3098 	return -ENOSYS;
3099 }
3100 
3101 int sysctl_string(struct ctl_table *table,
3102 		  void __user *oldval, size_t __user *oldlenp,
3103 		  void __user *newval, size_t newlen)
3104 {
3105 	return -ENOSYS;
3106 }
3107 
3108 int sysctl_intvec(struct ctl_table *table,
3109 		void __user *oldval, size_t __user *oldlenp,
3110 		void __user *newval, size_t newlen)
3111 {
3112 	return -ENOSYS;
3113 }
3114 
3115 int sysctl_jiffies(struct ctl_table *table,
3116 		void __user *oldval, size_t __user *oldlenp,
3117 		void __user *newval, size_t newlen)
3118 {
3119 	return -ENOSYS;
3120 }
3121 
3122 int sysctl_ms_jiffies(struct ctl_table *table,
3123 		void __user *oldval, size_t __user *oldlenp,
3124 		void __user *newval, size_t newlen)
3125 {
3126 	return -ENOSYS;
3127 }
3128 
3129 #endif /* CONFIG_SYSCTL_SYSCALL */
3130 
3131 static int deprecated_sysctl_warning(struct __sysctl_args *args)
3132 {
3133 	static int msg_count;
3134 	int name[CTL_MAXNAME];
3135 	int i;
3136 
3137 	/* Check args->nlen. */
3138 	if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
3139 		return -ENOTDIR;
3140 
3141 	/* Read in the sysctl name for better debug message logging */
3142 	for (i = 0; i < args->nlen; i++)
3143 		if (get_user(name[i], args->name + i))
3144 			return -EFAULT;
3145 
3146 	/* Ignore accesses to kernel.version */
3147 	if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
3148 		return 0;
3149 
3150 	if (msg_count < 5) {
3151 		msg_count++;
3152 		printk(KERN_INFO
3153 			"warning: process `%s' used the deprecated sysctl "
3154 			"system call with ", current->comm);
3155 		for (i = 0; i < args->nlen; i++)
3156 			printk("%d.", name[i]);
3157 		printk("\n");
3158 	}
3159 	return 0;
3160 }
3161 
3162 /*
3163  * No sense putting this after each symbol definition, twice,
3164  * exception granted :-)
3165  */
3166 EXPORT_SYMBOL(proc_dointvec);
3167 EXPORT_SYMBOL(proc_dointvec_jiffies);
3168 EXPORT_SYMBOL(proc_dointvec_minmax);
3169 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3170 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3171 EXPORT_SYMBOL(proc_dostring);
3172 EXPORT_SYMBOL(proc_doulongvec_minmax);
3173 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3174 EXPORT_SYMBOL(register_sysctl_table);
3175 EXPORT_SYMBOL(register_sysctl_paths);
3176 EXPORT_SYMBOL(sysctl_intvec);
3177 EXPORT_SYMBOL(sysctl_jiffies);
3178 EXPORT_SYMBOL(sysctl_ms_jiffies);
3179 EXPORT_SYMBOL(sysctl_string);
3180 EXPORT_SYMBOL(sysctl_data);
3181 EXPORT_SYMBOL(unregister_sysctl_table);
3182