xref: /openbmc/linux/kernel/sysctl.c (revision c21b37f6)
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/capability.h>
28 #include <linux/ctype.h>
29 #include <linux/utsname.h>
30 #include <linux/capability.h>
31 #include <linux/smp_lock.h>
32 #include <linux/fs.h>
33 #include <linux/init.h>
34 #include <linux/kernel.h>
35 #include <linux/kobject.h>
36 #include <linux/net.h>
37 #include <linux/sysrq.h>
38 #include <linux/highuid.h>
39 #include <linux/writeback.h>
40 #include <linux/hugetlb.h>
41 #include <linux/security.h>
42 #include <linux/initrd.h>
43 #include <linux/times.h>
44 #include <linux/limits.h>
45 #include <linux/dcache.h>
46 #include <linux/syscalls.h>
47 #include <linux/nfs_fs.h>
48 #include <linux/acpi.h>
49 #include <linux/reboot.h>
50 
51 #include <asm/uaccess.h>
52 #include <asm/processor.h>
53 
54 #ifdef CONFIG_X86
55 #include <asm/nmi.h>
56 #include <asm/stacktrace.h>
57 #endif
58 
59 #if defined(CONFIG_SYSCTL)
60 
61 /* External variables not in a header file. */
62 extern int C_A_D;
63 extern int print_fatal_signals;
64 extern int sysctl_overcommit_memory;
65 extern int sysctl_overcommit_ratio;
66 extern int sysctl_panic_on_oom;
67 extern int max_threads;
68 extern int core_uses_pid;
69 extern int suid_dumpable;
70 extern char core_pattern[];
71 extern int pid_max;
72 extern int min_free_kbytes;
73 extern int printk_ratelimit_jiffies;
74 extern int printk_ratelimit_burst;
75 extern int pid_max_min, pid_max_max;
76 extern int sysctl_drop_caches;
77 extern int percpu_pagelist_fraction;
78 extern int compat_log;
79 extern int maps_protect;
80 extern int sysctl_stat_interval;
81 extern int audit_argv_kb;
82 
83 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
84 static int maxolduid = 65535;
85 static int minolduid;
86 static int min_percpu_pagelist_fract = 8;
87 
88 static int ngroups_max = NGROUPS_MAX;
89 
90 #ifdef CONFIG_KMOD
91 extern char modprobe_path[];
92 #endif
93 #ifdef CONFIG_CHR_DEV_SG
94 extern int sg_big_buff;
95 #endif
96 
97 #ifdef __sparc__
98 extern char reboot_command [];
99 extern int stop_a_enabled;
100 extern int scons_pwroff;
101 #endif
102 
103 #ifdef __hppa__
104 extern int pwrsw_enabled;
105 extern int unaligned_enabled;
106 #endif
107 
108 #ifdef CONFIG_S390
109 #ifdef CONFIG_MATHEMU
110 extern int sysctl_ieee_emulation_warnings;
111 #endif
112 extern int sysctl_userprocess_debug;
113 extern int spin_retry;
114 #endif
115 
116 extern int sysctl_hz_timer;
117 
118 #ifdef CONFIG_BSD_PROCESS_ACCT
119 extern int acct_parm[];
120 #endif
121 
122 #ifdef CONFIG_IA64
123 extern int no_unaligned_warning;
124 #endif
125 
126 #ifdef CONFIG_RT_MUTEXES
127 extern int max_lock_depth;
128 #endif
129 
130 #ifdef CONFIG_SYSCTL_SYSCALL
131 static int parse_table(int __user *, int, void __user *, size_t __user *,
132 		void __user *, size_t, ctl_table *);
133 #endif
134 
135 
136 #ifdef CONFIG_PROC_SYSCTL
137 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
138 		  void __user *buffer, size_t *lenp, loff_t *ppos);
139 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
140 			       void __user *buffer, size_t *lenp, loff_t *ppos);
141 #endif
142 
143 static ctl_table root_table[];
144 static struct ctl_table_header root_table_header =
145 	{ root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
146 
147 static ctl_table kern_table[];
148 static ctl_table vm_table[];
149 static ctl_table fs_table[];
150 static ctl_table debug_table[];
151 static ctl_table dev_table[];
152 extern ctl_table random_table[];
153 #ifdef CONFIG_UNIX98_PTYS
154 extern ctl_table pty_table[];
155 #endif
156 #ifdef CONFIG_INOTIFY_USER
157 extern ctl_table inotify_table[];
158 #endif
159 
160 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
161 int sysctl_legacy_va_layout;
162 #endif
163 
164 extern int prove_locking;
165 extern int lock_stat;
166 
167 /* The default sysctl tables: */
168 
169 static ctl_table root_table[] = {
170 	{
171 		.ctl_name	= CTL_KERN,
172 		.procname	= "kernel",
173 		.mode		= 0555,
174 		.child		= kern_table,
175 	},
176 	{
177 		.ctl_name	= CTL_VM,
178 		.procname	= "vm",
179 		.mode		= 0555,
180 		.child		= vm_table,
181 	},
182 #ifdef CONFIG_NET
183 	{
184 		.ctl_name	= CTL_NET,
185 		.procname	= "net",
186 		.mode		= 0555,
187 		.child		= net_table,
188 	},
189 #endif
190 	{
191 		.ctl_name	= CTL_FS,
192 		.procname	= "fs",
193 		.mode		= 0555,
194 		.child		= fs_table,
195 	},
196 	{
197 		.ctl_name	= CTL_DEBUG,
198 		.procname	= "debug",
199 		.mode		= 0555,
200 		.child		= debug_table,
201 	},
202 	{
203 		.ctl_name	= CTL_DEV,
204 		.procname	= "dev",
205 		.mode		= 0555,
206 		.child		= dev_table,
207 	},
208 /*
209  * NOTE: do not add new entries to this table unless you have read
210  * Documentation/sysctl/ctl_unnumbered.txt
211  */
212 	{ .ctl_name = 0 }
213 };
214 
215 #ifdef CONFIG_SCHED_DEBUG
216 static unsigned long min_sched_granularity_ns = 100000;		/* 100 usecs */
217 static unsigned long max_sched_granularity_ns = 1000000000;	/* 1 second */
218 static unsigned long min_wakeup_granularity_ns;			/* 0 usecs */
219 static unsigned long max_wakeup_granularity_ns = 1000000000;	/* 1 second */
220 #endif
221 
222 static ctl_table kern_table[] = {
223 #ifdef CONFIG_SCHED_DEBUG
224 	{
225 		.ctl_name	= CTL_UNNUMBERED,
226 		.procname	= "sched_granularity_ns",
227 		.data		= &sysctl_sched_granularity,
228 		.maxlen		= sizeof(unsigned int),
229 		.mode		= 0644,
230 		.proc_handler	= &proc_dointvec_minmax,
231 		.strategy	= &sysctl_intvec,
232 		.extra1		= &min_sched_granularity_ns,
233 		.extra2		= &max_sched_granularity_ns,
234 	},
235 	{
236 		.ctl_name	= CTL_UNNUMBERED,
237 		.procname	= "sched_wakeup_granularity_ns",
238 		.data		= &sysctl_sched_wakeup_granularity,
239 		.maxlen		= sizeof(unsigned int),
240 		.mode		= 0644,
241 		.proc_handler	= &proc_dointvec_minmax,
242 		.strategy	= &sysctl_intvec,
243 		.extra1		= &min_wakeup_granularity_ns,
244 		.extra2		= &max_wakeup_granularity_ns,
245 	},
246 	{
247 		.ctl_name	= CTL_UNNUMBERED,
248 		.procname	= "sched_batch_wakeup_granularity_ns",
249 		.data		= &sysctl_sched_batch_wakeup_granularity,
250 		.maxlen		= sizeof(unsigned int),
251 		.mode		= 0644,
252 		.proc_handler	= &proc_dointvec_minmax,
253 		.strategy	= &sysctl_intvec,
254 		.extra1		= &min_wakeup_granularity_ns,
255 		.extra2		= &max_wakeup_granularity_ns,
256 	},
257 	{
258 		.ctl_name	= CTL_UNNUMBERED,
259 		.procname	= "sched_stat_granularity_ns",
260 		.data		= &sysctl_sched_stat_granularity,
261 		.maxlen		= sizeof(unsigned int),
262 		.mode		= 0644,
263 		.proc_handler	= &proc_dointvec_minmax,
264 		.strategy	= &sysctl_intvec,
265 		.extra1		= &min_wakeup_granularity_ns,
266 		.extra2		= &max_wakeup_granularity_ns,
267 	},
268 	{
269 		.ctl_name	= CTL_UNNUMBERED,
270 		.procname	= "sched_runtime_limit_ns",
271 		.data		= &sysctl_sched_runtime_limit,
272 		.maxlen		= sizeof(unsigned int),
273 		.mode		= 0644,
274 		.proc_handler	= &proc_dointvec_minmax,
275 		.strategy	= &sysctl_intvec,
276 		.extra1		= &min_sched_granularity_ns,
277 		.extra2		= &max_sched_granularity_ns,
278 	},
279 	{
280 		.ctl_name	= CTL_UNNUMBERED,
281 		.procname	= "sched_child_runs_first",
282 		.data		= &sysctl_sched_child_runs_first,
283 		.maxlen		= sizeof(unsigned int),
284 		.mode		= 0644,
285 		.proc_handler	= &proc_dointvec,
286 	},
287 #ifdef CONFIG_PROVE_LOCKING
288 	{
289 		.ctl_name	= CTL_UNNUMBERED,
290 		.procname	= "prove_locking",
291 		.data		= &prove_locking,
292 		.maxlen		= sizeof(int),
293 		.mode		= 0644,
294 		.proc_handler	= &proc_dointvec,
295 	},
296 #endif
297 #ifdef CONFIG_LOCK_STAT
298 	{
299 		.ctl_name	= CTL_UNNUMBERED,
300 		.procname	= "lock_stat",
301 		.data		= &lock_stat,
302 		.maxlen		= sizeof(int),
303 		.mode		= 0644,
304 		.proc_handler	= &proc_dointvec,
305 	},
306 #endif
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 #endif
316 	{
317 		.ctl_name	= KERN_PANIC,
318 		.procname	= "panic",
319 		.data		= &panic_timeout,
320 		.maxlen		= sizeof(int),
321 		.mode		= 0644,
322 		.proc_handler	= &proc_dointvec,
323 	},
324 	{
325 		.ctl_name	= KERN_CORE_USES_PID,
326 		.procname	= "core_uses_pid",
327 		.data		= &core_uses_pid,
328 		.maxlen		= sizeof(int),
329 		.mode		= 0644,
330 		.proc_handler	= &proc_dointvec,
331 	},
332 #ifdef CONFIG_AUDITSYSCALL
333 	{
334 		.ctl_name	= CTL_UNNUMBERED,
335 		.procname	= "audit_argv_kb",
336 		.data		= &audit_argv_kb,
337 		.maxlen		= sizeof(int),
338 		.mode		= 0644,
339 		.proc_handler	= &proc_dointvec,
340 	},
341 #endif
342 	{
343 		.ctl_name	= KERN_CORE_PATTERN,
344 		.procname	= "core_pattern",
345 		.data		= core_pattern,
346 		.maxlen		= CORENAME_MAX_SIZE,
347 		.mode		= 0644,
348 		.proc_handler	= &proc_dostring,
349 		.strategy	= &sysctl_string,
350 	},
351 #ifdef CONFIG_PROC_SYSCTL
352 	{
353 		.ctl_name	= KERN_TAINTED,
354 		.procname	= "tainted",
355 		.data		= &tainted,
356 		.maxlen		= sizeof(int),
357 		.mode		= 0644,
358 		.proc_handler	= &proc_dointvec_taint,
359 	},
360 #endif
361 	{
362 		.ctl_name	= KERN_CAP_BSET,
363 		.procname	= "cap-bound",
364 		.data		= &cap_bset,
365 		.maxlen		= sizeof(kernel_cap_t),
366 		.mode		= 0600,
367 		.proc_handler	= &proc_dointvec_bset,
368 	},
369 #ifdef CONFIG_BLK_DEV_INITRD
370 	{
371 		.ctl_name	= KERN_REALROOTDEV,
372 		.procname	= "real-root-dev",
373 		.data		= &real_root_dev,
374 		.maxlen		= sizeof(int),
375 		.mode		= 0644,
376 		.proc_handler	= &proc_dointvec,
377 	},
378 #endif
379 	{
380 		.ctl_name	= CTL_UNNUMBERED,
381 		.procname	= "print-fatal-signals",
382 		.data		= &print_fatal_signals,
383 		.maxlen		= sizeof(int),
384 		.mode		= 0644,
385 		.proc_handler	= &proc_dointvec,
386 	},
387 #ifdef __sparc__
388 	{
389 		.ctl_name	= KERN_SPARC_REBOOT,
390 		.procname	= "reboot-cmd",
391 		.data		= reboot_command,
392 		.maxlen		= 256,
393 		.mode		= 0644,
394 		.proc_handler	= &proc_dostring,
395 		.strategy	= &sysctl_string,
396 	},
397 	{
398 		.ctl_name	= KERN_SPARC_STOP_A,
399 		.procname	= "stop-a",
400 		.data		= &stop_a_enabled,
401 		.maxlen		= sizeof (int),
402 		.mode		= 0644,
403 		.proc_handler	= &proc_dointvec,
404 	},
405 	{
406 		.ctl_name	= KERN_SPARC_SCONS_PWROFF,
407 		.procname	= "scons-poweroff",
408 		.data		= &scons_pwroff,
409 		.maxlen		= sizeof (int),
410 		.mode		= 0644,
411 		.proc_handler	= &proc_dointvec,
412 	},
413 #endif
414 #ifdef __hppa__
415 	{
416 		.ctl_name	= KERN_HPPA_PWRSW,
417 		.procname	= "soft-power",
418 		.data		= &pwrsw_enabled,
419 		.maxlen		= sizeof (int),
420 	 	.mode		= 0644,
421 		.proc_handler	= &proc_dointvec,
422 	},
423 	{
424 		.ctl_name	= KERN_HPPA_UNALIGNED,
425 		.procname	= "unaligned-trap",
426 		.data		= &unaligned_enabled,
427 		.maxlen		= sizeof (int),
428 		.mode		= 0644,
429 		.proc_handler	= &proc_dointvec,
430 	},
431 #endif
432 	{
433 		.ctl_name	= KERN_CTLALTDEL,
434 		.procname	= "ctrl-alt-del",
435 		.data		= &C_A_D,
436 		.maxlen		= sizeof(int),
437 		.mode		= 0644,
438 		.proc_handler	= &proc_dointvec,
439 	},
440 	{
441 		.ctl_name	= KERN_PRINTK,
442 		.procname	= "printk",
443 		.data		= &console_loglevel,
444 		.maxlen		= 4*sizeof(int),
445 		.mode		= 0644,
446 		.proc_handler	= &proc_dointvec,
447 	},
448 #ifdef CONFIG_KMOD
449 	{
450 		.ctl_name	= KERN_MODPROBE,
451 		.procname	= "modprobe",
452 		.data		= &modprobe_path,
453 		.maxlen		= KMOD_PATH_LEN,
454 		.mode		= 0644,
455 		.proc_handler	= &proc_dostring,
456 		.strategy	= &sysctl_string,
457 	},
458 #endif
459 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
460 	{
461 		.ctl_name	= KERN_HOTPLUG,
462 		.procname	= "hotplug",
463 		.data		= &uevent_helper,
464 		.maxlen		= UEVENT_HELPER_PATH_LEN,
465 		.mode		= 0644,
466 		.proc_handler	= &proc_dostring,
467 		.strategy	= &sysctl_string,
468 	},
469 #endif
470 #ifdef CONFIG_CHR_DEV_SG
471 	{
472 		.ctl_name	= KERN_SG_BIG_BUFF,
473 		.procname	= "sg-big-buff",
474 		.data		= &sg_big_buff,
475 		.maxlen		= sizeof (int),
476 		.mode		= 0444,
477 		.proc_handler	= &proc_dointvec,
478 	},
479 #endif
480 #ifdef CONFIG_BSD_PROCESS_ACCT
481 	{
482 		.ctl_name	= KERN_ACCT,
483 		.procname	= "acct",
484 		.data		= &acct_parm,
485 		.maxlen		= 3*sizeof(int),
486 		.mode		= 0644,
487 		.proc_handler	= &proc_dointvec,
488 	},
489 #endif
490 #ifdef CONFIG_MAGIC_SYSRQ
491 	{
492 		.ctl_name	= KERN_SYSRQ,
493 		.procname	= "sysrq",
494 		.data		= &__sysrq_enabled,
495 		.maxlen		= sizeof (int),
496 		.mode		= 0644,
497 		.proc_handler	= &proc_dointvec,
498 	},
499 #endif
500 #ifdef CONFIG_PROC_SYSCTL
501 	{
502 		.ctl_name	= KERN_CADPID,
503 		.procname	= "cad_pid",
504 		.data		= NULL,
505 		.maxlen		= sizeof (int),
506 		.mode		= 0600,
507 		.proc_handler	= &proc_do_cad_pid,
508 	},
509 #endif
510 	{
511 		.ctl_name	= KERN_MAX_THREADS,
512 		.procname	= "threads-max",
513 		.data		= &max_threads,
514 		.maxlen		= sizeof(int),
515 		.mode		= 0644,
516 		.proc_handler	= &proc_dointvec,
517 	},
518 	{
519 		.ctl_name	= KERN_RANDOM,
520 		.procname	= "random",
521 		.mode		= 0555,
522 		.child		= random_table,
523 	},
524 #ifdef CONFIG_UNIX98_PTYS
525 	{
526 		.ctl_name	= KERN_PTY,
527 		.procname	= "pty",
528 		.mode		= 0555,
529 		.child		= pty_table,
530 	},
531 #endif
532 	{
533 		.ctl_name	= KERN_OVERFLOWUID,
534 		.procname	= "overflowuid",
535 		.data		= &overflowuid,
536 		.maxlen		= sizeof(int),
537 		.mode		= 0644,
538 		.proc_handler	= &proc_dointvec_minmax,
539 		.strategy	= &sysctl_intvec,
540 		.extra1		= &minolduid,
541 		.extra2		= &maxolduid,
542 	},
543 	{
544 		.ctl_name	= KERN_OVERFLOWGID,
545 		.procname	= "overflowgid",
546 		.data		= &overflowgid,
547 		.maxlen		= sizeof(int),
548 		.mode		= 0644,
549 		.proc_handler	= &proc_dointvec_minmax,
550 		.strategy	= &sysctl_intvec,
551 		.extra1		= &minolduid,
552 		.extra2		= &maxolduid,
553 	},
554 #ifdef CONFIG_S390
555 #ifdef CONFIG_MATHEMU
556 	{
557 		.ctl_name	= KERN_IEEE_EMULATION_WARNINGS,
558 		.procname	= "ieee_emulation_warnings",
559 		.data		= &sysctl_ieee_emulation_warnings,
560 		.maxlen		= sizeof(int),
561 		.mode		= 0644,
562 		.proc_handler	= &proc_dointvec,
563 	},
564 #endif
565 #ifdef CONFIG_NO_IDLE_HZ
566 	{
567 		.ctl_name       = KERN_HZ_TIMER,
568 		.procname       = "hz_timer",
569 		.data           = &sysctl_hz_timer,
570 		.maxlen         = sizeof(int),
571 		.mode           = 0644,
572 		.proc_handler   = &proc_dointvec,
573 	},
574 #endif
575 	{
576 		.ctl_name	= KERN_S390_USER_DEBUG_LOGGING,
577 		.procname	= "userprocess_debug",
578 		.data		= &sysctl_userprocess_debug,
579 		.maxlen		= sizeof(int),
580 		.mode		= 0644,
581 		.proc_handler	= &proc_dointvec,
582 	},
583 #endif
584 	{
585 		.ctl_name	= KERN_PIDMAX,
586 		.procname	= "pid_max",
587 		.data		= &pid_max,
588 		.maxlen		= sizeof (int),
589 		.mode		= 0644,
590 		.proc_handler	= &proc_dointvec_minmax,
591 		.strategy	= sysctl_intvec,
592 		.extra1		= &pid_max_min,
593 		.extra2		= &pid_max_max,
594 	},
595 	{
596 		.ctl_name	= KERN_PANIC_ON_OOPS,
597 		.procname	= "panic_on_oops",
598 		.data		= &panic_on_oops,
599 		.maxlen		= sizeof(int),
600 		.mode		= 0644,
601 		.proc_handler	= &proc_dointvec,
602 	},
603 	{
604 		.ctl_name	= KERN_PRINTK_RATELIMIT,
605 		.procname	= "printk_ratelimit",
606 		.data		= &printk_ratelimit_jiffies,
607 		.maxlen		= sizeof(int),
608 		.mode		= 0644,
609 		.proc_handler	= &proc_dointvec_jiffies,
610 		.strategy	= &sysctl_jiffies,
611 	},
612 	{
613 		.ctl_name	= KERN_PRINTK_RATELIMIT_BURST,
614 		.procname	= "printk_ratelimit_burst",
615 		.data		= &printk_ratelimit_burst,
616 		.maxlen		= sizeof(int),
617 		.mode		= 0644,
618 		.proc_handler	= &proc_dointvec,
619 	},
620 	{
621 		.ctl_name	= KERN_NGROUPS_MAX,
622 		.procname	= "ngroups_max",
623 		.data		= &ngroups_max,
624 		.maxlen		= sizeof (int),
625 		.mode		= 0444,
626 		.proc_handler	= &proc_dointvec,
627 	},
628 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
629 	{
630 		.ctl_name       = KERN_UNKNOWN_NMI_PANIC,
631 		.procname       = "unknown_nmi_panic",
632 		.data           = &unknown_nmi_panic,
633 		.maxlen         = sizeof (int),
634 		.mode           = 0644,
635 		.proc_handler   = &proc_dointvec,
636 	},
637 	{
638 		.ctl_name       = KERN_NMI_WATCHDOG,
639 		.procname       = "nmi_watchdog",
640 		.data           = &nmi_watchdog_enabled,
641 		.maxlen         = sizeof (int),
642 		.mode           = 0644,
643 		.proc_handler   = &proc_nmi_enabled,
644 	},
645 #endif
646 #if defined(CONFIG_X86)
647 	{
648 		.ctl_name	= KERN_PANIC_ON_NMI,
649 		.procname	= "panic_on_unrecovered_nmi",
650 		.data		= &panic_on_unrecovered_nmi,
651 		.maxlen		= sizeof(int),
652 		.mode		= 0644,
653 		.proc_handler	= &proc_dointvec,
654 	},
655 	{
656 		.ctl_name	= KERN_BOOTLOADER_TYPE,
657 		.procname	= "bootloader_type",
658 		.data		= &bootloader_type,
659 		.maxlen		= sizeof (int),
660 		.mode		= 0444,
661 		.proc_handler	= &proc_dointvec,
662 	},
663 	{
664 		.ctl_name	= CTL_UNNUMBERED,
665 		.procname	= "kstack_depth_to_print",
666 		.data		= &kstack_depth_to_print,
667 		.maxlen		= sizeof(int),
668 		.mode		= 0644,
669 		.proc_handler	= &proc_dointvec,
670 	},
671 #endif
672 #if defined(CONFIG_MMU)
673 	{
674 		.ctl_name	= KERN_RANDOMIZE,
675 		.procname	= "randomize_va_space",
676 		.data		= &randomize_va_space,
677 		.maxlen		= sizeof(int),
678 		.mode		= 0644,
679 		.proc_handler	= &proc_dointvec,
680 	},
681 #endif
682 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
683 	{
684 		.ctl_name	= KERN_SPIN_RETRY,
685 		.procname	= "spin_retry",
686 		.data		= &spin_retry,
687 		.maxlen		= sizeof (int),
688 		.mode		= 0644,
689 		.proc_handler	= &proc_dointvec,
690 	},
691 #endif
692 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
693 	{
694 		.ctl_name	= KERN_ACPI_VIDEO_FLAGS,
695 		.procname	= "acpi_video_flags",
696 		.data		= &acpi_realmode_flags,
697 		.maxlen		= sizeof (unsigned long),
698 		.mode		= 0644,
699 		.proc_handler	= &proc_doulongvec_minmax,
700 	},
701 #endif
702 #ifdef CONFIG_IA64
703 	{
704 		.ctl_name	= KERN_IA64_UNALIGNED,
705 		.procname	= "ignore-unaligned-usertrap",
706 		.data		= &no_unaligned_warning,
707 		.maxlen		= sizeof (int),
708 	 	.mode		= 0644,
709 		.proc_handler	= &proc_dointvec,
710 	},
711 #endif
712 #ifdef CONFIG_COMPAT
713 	{
714 		.ctl_name	= KERN_COMPAT_LOG,
715 		.procname	= "compat-log",
716 		.data		= &compat_log,
717 		.maxlen		= sizeof (int),
718 	 	.mode		= 0644,
719 		.proc_handler	= &proc_dointvec,
720 	},
721 #endif
722 #ifdef CONFIG_RT_MUTEXES
723 	{
724 		.ctl_name	= KERN_MAX_LOCK_DEPTH,
725 		.procname	= "max_lock_depth",
726 		.data		= &max_lock_depth,
727 		.maxlen		= sizeof(int),
728 		.mode		= 0644,
729 		.proc_handler	= &proc_dointvec,
730 	},
731 #endif
732 #ifdef CONFIG_PROC_FS
733 	{
734 		.ctl_name       = CTL_UNNUMBERED,
735 		.procname       = "maps_protect",
736 		.data           = &maps_protect,
737 		.maxlen         = sizeof(int),
738 		.mode           = 0644,
739 		.proc_handler   = &proc_dointvec,
740 	},
741 #endif
742 	{
743 		.ctl_name	= CTL_UNNUMBERED,
744 		.procname	= "poweroff_cmd",
745 		.data		= &poweroff_cmd,
746 		.maxlen		= POWEROFF_CMD_PATH_LEN,
747 		.mode		= 0644,
748 		.proc_handler	= &proc_dostring,
749 		.strategy	= &sysctl_string,
750 	},
751 /*
752  * NOTE: do not add new entries to this table unless you have read
753  * Documentation/sysctl/ctl_unnumbered.txt
754  */
755 	{ .ctl_name = 0 }
756 };
757 
758 /* Constants for minimum and maximum testing in vm_table.
759    We use these as one-element integer vectors. */
760 static int zero;
761 static int two = 2;
762 static int one_hundred = 100;
763 
764 
765 static ctl_table vm_table[] = {
766 	{
767 		.ctl_name	= VM_OVERCOMMIT_MEMORY,
768 		.procname	= "overcommit_memory",
769 		.data		= &sysctl_overcommit_memory,
770 		.maxlen		= sizeof(sysctl_overcommit_memory),
771 		.mode		= 0644,
772 		.proc_handler	= &proc_dointvec,
773 	},
774 	{
775 		.ctl_name	= VM_PANIC_ON_OOM,
776 		.procname	= "panic_on_oom",
777 		.data		= &sysctl_panic_on_oom,
778 		.maxlen		= sizeof(sysctl_panic_on_oom),
779 		.mode		= 0644,
780 		.proc_handler	= &proc_dointvec,
781 	},
782 	{
783 		.ctl_name	= VM_OVERCOMMIT_RATIO,
784 		.procname	= "overcommit_ratio",
785 		.data		= &sysctl_overcommit_ratio,
786 		.maxlen		= sizeof(sysctl_overcommit_ratio),
787 		.mode		= 0644,
788 		.proc_handler	= &proc_dointvec,
789 	},
790 	{
791 		.ctl_name	= VM_PAGE_CLUSTER,
792 		.procname	= "page-cluster",
793 		.data		= &page_cluster,
794 		.maxlen		= sizeof(int),
795 		.mode		= 0644,
796 		.proc_handler	= &proc_dointvec,
797 	},
798 	{
799 		.ctl_name	= VM_DIRTY_BACKGROUND,
800 		.procname	= "dirty_background_ratio",
801 		.data		= &dirty_background_ratio,
802 		.maxlen		= sizeof(dirty_background_ratio),
803 		.mode		= 0644,
804 		.proc_handler	= &proc_dointvec_minmax,
805 		.strategy	= &sysctl_intvec,
806 		.extra1		= &zero,
807 		.extra2		= &one_hundred,
808 	},
809 	{
810 		.ctl_name	= VM_DIRTY_RATIO,
811 		.procname	= "dirty_ratio",
812 		.data		= &vm_dirty_ratio,
813 		.maxlen		= sizeof(vm_dirty_ratio),
814 		.mode		= 0644,
815 		.proc_handler	= &proc_dointvec_minmax,
816 		.strategy	= &sysctl_intvec,
817 		.extra1		= &zero,
818 		.extra2		= &one_hundred,
819 	},
820 	{
821 		.ctl_name	= VM_DIRTY_WB_CS,
822 		.procname	= "dirty_writeback_centisecs",
823 		.data		= &dirty_writeback_interval,
824 		.maxlen		= sizeof(dirty_writeback_interval),
825 		.mode		= 0644,
826 		.proc_handler	= &dirty_writeback_centisecs_handler,
827 	},
828 	{
829 		.ctl_name	= VM_DIRTY_EXPIRE_CS,
830 		.procname	= "dirty_expire_centisecs",
831 		.data		= &dirty_expire_interval,
832 		.maxlen		= sizeof(dirty_expire_interval),
833 		.mode		= 0644,
834 		.proc_handler	= &proc_dointvec_userhz_jiffies,
835 	},
836 	{
837 		.ctl_name	= VM_NR_PDFLUSH_THREADS,
838 		.procname	= "nr_pdflush_threads",
839 		.data		= &nr_pdflush_threads,
840 		.maxlen		= sizeof nr_pdflush_threads,
841 		.mode		= 0444 /* read-only*/,
842 		.proc_handler	= &proc_dointvec,
843 	},
844 	{
845 		.ctl_name	= VM_SWAPPINESS,
846 		.procname	= "swappiness",
847 		.data		= &vm_swappiness,
848 		.maxlen		= sizeof(vm_swappiness),
849 		.mode		= 0644,
850 		.proc_handler	= &proc_dointvec_minmax,
851 		.strategy	= &sysctl_intvec,
852 		.extra1		= &zero,
853 		.extra2		= &one_hundred,
854 	},
855 #ifdef CONFIG_HUGETLB_PAGE
856 	 {
857 		.ctl_name	= VM_HUGETLB_PAGES,
858 		.procname	= "nr_hugepages",
859 		.data		= &max_huge_pages,
860 		.maxlen		= sizeof(unsigned long),
861 		.mode		= 0644,
862 		.proc_handler	= &hugetlb_sysctl_handler,
863 		.extra1		= (void *)&hugetlb_zero,
864 		.extra2		= (void *)&hugetlb_infinity,
865 	 },
866 	 {
867 		.ctl_name	= VM_HUGETLB_GROUP,
868 		.procname	= "hugetlb_shm_group",
869 		.data		= &sysctl_hugetlb_shm_group,
870 		.maxlen		= sizeof(gid_t),
871 		.mode		= 0644,
872 		.proc_handler	= &proc_dointvec,
873 	 },
874 	 {
875 		.ctl_name	= CTL_UNNUMBERED,
876 		.procname	= "hugepages_treat_as_movable",
877 		.data		= &hugepages_treat_as_movable,
878 		.maxlen		= sizeof(int),
879 		.mode		= 0644,
880 		.proc_handler	= &hugetlb_treat_movable_handler,
881 	},
882 #endif
883 	{
884 		.ctl_name	= VM_LOWMEM_RESERVE_RATIO,
885 		.procname	= "lowmem_reserve_ratio",
886 		.data		= &sysctl_lowmem_reserve_ratio,
887 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
888 		.mode		= 0644,
889 		.proc_handler	= &lowmem_reserve_ratio_sysctl_handler,
890 		.strategy	= &sysctl_intvec,
891 	},
892 	{
893 		.ctl_name	= VM_DROP_PAGECACHE,
894 		.procname	= "drop_caches",
895 		.data		= &sysctl_drop_caches,
896 		.maxlen		= sizeof(int),
897 		.mode		= 0644,
898 		.proc_handler	= drop_caches_sysctl_handler,
899 		.strategy	= &sysctl_intvec,
900 	},
901 	{
902 		.ctl_name	= VM_MIN_FREE_KBYTES,
903 		.procname	= "min_free_kbytes",
904 		.data		= &min_free_kbytes,
905 		.maxlen		= sizeof(min_free_kbytes),
906 		.mode		= 0644,
907 		.proc_handler	= &min_free_kbytes_sysctl_handler,
908 		.strategy	= &sysctl_intvec,
909 		.extra1		= &zero,
910 	},
911 	{
912 		.ctl_name	= VM_PERCPU_PAGELIST_FRACTION,
913 		.procname	= "percpu_pagelist_fraction",
914 		.data		= &percpu_pagelist_fraction,
915 		.maxlen		= sizeof(percpu_pagelist_fraction),
916 		.mode		= 0644,
917 		.proc_handler	= &percpu_pagelist_fraction_sysctl_handler,
918 		.strategy	= &sysctl_intvec,
919 		.extra1		= &min_percpu_pagelist_fract,
920 	},
921 #ifdef CONFIG_MMU
922 	{
923 		.ctl_name	= VM_MAX_MAP_COUNT,
924 		.procname	= "max_map_count",
925 		.data		= &sysctl_max_map_count,
926 		.maxlen		= sizeof(sysctl_max_map_count),
927 		.mode		= 0644,
928 		.proc_handler	= &proc_dointvec
929 	},
930 #endif
931 	{
932 		.ctl_name	= VM_LAPTOP_MODE,
933 		.procname	= "laptop_mode",
934 		.data		= &laptop_mode,
935 		.maxlen		= sizeof(laptop_mode),
936 		.mode		= 0644,
937 		.proc_handler	= &proc_dointvec_jiffies,
938 		.strategy	= &sysctl_jiffies,
939 	},
940 	{
941 		.ctl_name	= VM_BLOCK_DUMP,
942 		.procname	= "block_dump",
943 		.data		= &block_dump,
944 		.maxlen		= sizeof(block_dump),
945 		.mode		= 0644,
946 		.proc_handler	= &proc_dointvec,
947 		.strategy	= &sysctl_intvec,
948 		.extra1		= &zero,
949 	},
950 	{
951 		.ctl_name	= VM_VFS_CACHE_PRESSURE,
952 		.procname	= "vfs_cache_pressure",
953 		.data		= &sysctl_vfs_cache_pressure,
954 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
955 		.mode		= 0644,
956 		.proc_handler	= &proc_dointvec,
957 		.strategy	= &sysctl_intvec,
958 		.extra1		= &zero,
959 	},
960 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
961 	{
962 		.ctl_name	= VM_LEGACY_VA_LAYOUT,
963 		.procname	= "legacy_va_layout",
964 		.data		= &sysctl_legacy_va_layout,
965 		.maxlen		= sizeof(sysctl_legacy_va_layout),
966 		.mode		= 0644,
967 		.proc_handler	= &proc_dointvec,
968 		.strategy	= &sysctl_intvec,
969 		.extra1		= &zero,
970 	},
971 #endif
972 #ifdef CONFIG_NUMA
973 	{
974 		.ctl_name	= VM_ZONE_RECLAIM_MODE,
975 		.procname	= "zone_reclaim_mode",
976 		.data		= &zone_reclaim_mode,
977 		.maxlen		= sizeof(zone_reclaim_mode),
978 		.mode		= 0644,
979 		.proc_handler	= &proc_dointvec,
980 		.strategy	= &sysctl_intvec,
981 		.extra1		= &zero,
982 	},
983 	{
984 		.ctl_name	= VM_MIN_UNMAPPED,
985 		.procname	= "min_unmapped_ratio",
986 		.data		= &sysctl_min_unmapped_ratio,
987 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
988 		.mode		= 0644,
989 		.proc_handler	= &sysctl_min_unmapped_ratio_sysctl_handler,
990 		.strategy	= &sysctl_intvec,
991 		.extra1		= &zero,
992 		.extra2		= &one_hundred,
993 	},
994 	{
995 		.ctl_name	= VM_MIN_SLAB,
996 		.procname	= "min_slab_ratio",
997 		.data		= &sysctl_min_slab_ratio,
998 		.maxlen		= sizeof(sysctl_min_slab_ratio),
999 		.mode		= 0644,
1000 		.proc_handler	= &sysctl_min_slab_ratio_sysctl_handler,
1001 		.strategy	= &sysctl_intvec,
1002 		.extra1		= &zero,
1003 		.extra2		= &one_hundred,
1004 	},
1005 #endif
1006 #ifdef CONFIG_SMP
1007 	{
1008 		.ctl_name	= CTL_UNNUMBERED,
1009 		.procname	= "stat_interval",
1010 		.data		= &sysctl_stat_interval,
1011 		.maxlen		= sizeof(sysctl_stat_interval),
1012 		.mode		= 0644,
1013 		.proc_handler	= &proc_dointvec_jiffies,
1014 		.strategy	= &sysctl_jiffies,
1015 	},
1016 #endif
1017 #ifdef CONFIG_SECURITY
1018 	{
1019 		.ctl_name	= CTL_UNNUMBERED,
1020 		.procname	= "mmap_min_addr",
1021 		.data		= &mmap_min_addr,
1022 		.maxlen         = sizeof(unsigned long),
1023 		.mode		= 0644,
1024 		.proc_handler	= &proc_doulongvec_minmax,
1025 	},
1026 #ifdef CONFIG_NUMA
1027 	{
1028 		.ctl_name	= CTL_UNNUMBERED,
1029 		.procname	= "numa_zonelist_order",
1030 		.data		= &numa_zonelist_order,
1031 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1032 		.mode		= 0644,
1033 		.proc_handler	= &numa_zonelist_order_handler,
1034 		.strategy	= &sysctl_string,
1035 	},
1036 #endif
1037 #endif
1038 #if defined(CONFIG_X86_32) || \
1039    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1040 	{
1041 		.ctl_name	= VM_VDSO_ENABLED,
1042 		.procname	= "vdso_enabled",
1043 		.data		= &vdso_enabled,
1044 		.maxlen		= sizeof(vdso_enabled),
1045 		.mode		= 0644,
1046 		.proc_handler	= &proc_dointvec,
1047 		.strategy	= &sysctl_intvec,
1048 		.extra1		= &zero,
1049 	},
1050 #endif
1051 /*
1052  * NOTE: do not add new entries to this table unless you have read
1053  * Documentation/sysctl/ctl_unnumbered.txt
1054  */
1055 	{ .ctl_name = 0 }
1056 };
1057 
1058 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1059 static ctl_table binfmt_misc_table[] = {
1060 	{ .ctl_name = 0 }
1061 };
1062 #endif
1063 
1064 static ctl_table fs_table[] = {
1065 	{
1066 		.ctl_name	= FS_NRINODE,
1067 		.procname	= "inode-nr",
1068 		.data		= &inodes_stat,
1069 		.maxlen		= 2*sizeof(int),
1070 		.mode		= 0444,
1071 		.proc_handler	= &proc_dointvec,
1072 	},
1073 	{
1074 		.ctl_name	= FS_STATINODE,
1075 		.procname	= "inode-state",
1076 		.data		= &inodes_stat,
1077 		.maxlen		= 7*sizeof(int),
1078 		.mode		= 0444,
1079 		.proc_handler	= &proc_dointvec,
1080 	},
1081 	{
1082 		.ctl_name	= FS_NRFILE,
1083 		.procname	= "file-nr",
1084 		.data		= &files_stat,
1085 		.maxlen		= 3*sizeof(int),
1086 		.mode		= 0444,
1087 		.proc_handler	= &proc_nr_files,
1088 	},
1089 	{
1090 		.ctl_name	= FS_MAXFILE,
1091 		.procname	= "file-max",
1092 		.data		= &files_stat.max_files,
1093 		.maxlen		= sizeof(int),
1094 		.mode		= 0644,
1095 		.proc_handler	= &proc_dointvec,
1096 	},
1097 	{
1098 		.ctl_name	= FS_DENTRY,
1099 		.procname	= "dentry-state",
1100 		.data		= &dentry_stat,
1101 		.maxlen		= 6*sizeof(int),
1102 		.mode		= 0444,
1103 		.proc_handler	= &proc_dointvec,
1104 	},
1105 	{
1106 		.ctl_name	= FS_OVERFLOWUID,
1107 		.procname	= "overflowuid",
1108 		.data		= &fs_overflowuid,
1109 		.maxlen		= sizeof(int),
1110 		.mode		= 0644,
1111 		.proc_handler	= &proc_dointvec_minmax,
1112 		.strategy	= &sysctl_intvec,
1113 		.extra1		= &minolduid,
1114 		.extra2		= &maxolduid,
1115 	},
1116 	{
1117 		.ctl_name	= FS_OVERFLOWGID,
1118 		.procname	= "overflowgid",
1119 		.data		= &fs_overflowgid,
1120 		.maxlen		= sizeof(int),
1121 		.mode		= 0644,
1122 		.proc_handler	= &proc_dointvec_minmax,
1123 		.strategy	= &sysctl_intvec,
1124 		.extra1		= &minolduid,
1125 		.extra2		= &maxolduid,
1126 	},
1127 	{
1128 		.ctl_name	= FS_LEASES,
1129 		.procname	= "leases-enable",
1130 		.data		= &leases_enable,
1131 		.maxlen		= sizeof(int),
1132 		.mode		= 0644,
1133 		.proc_handler	= &proc_dointvec,
1134 	},
1135 #ifdef CONFIG_DNOTIFY
1136 	{
1137 		.ctl_name	= FS_DIR_NOTIFY,
1138 		.procname	= "dir-notify-enable",
1139 		.data		= &dir_notify_enable,
1140 		.maxlen		= sizeof(int),
1141 		.mode		= 0644,
1142 		.proc_handler	= &proc_dointvec,
1143 	},
1144 #endif
1145 #ifdef CONFIG_MMU
1146 	{
1147 		.ctl_name	= FS_LEASE_TIME,
1148 		.procname	= "lease-break-time",
1149 		.data		= &lease_break_time,
1150 		.maxlen		= sizeof(int),
1151 		.mode		= 0644,
1152 		.proc_handler	= &proc_dointvec_minmax,
1153 		.strategy	= &sysctl_intvec,
1154 		.extra1		= &zero,
1155 		.extra2		= &two,
1156 	},
1157 	{
1158 		.ctl_name	= FS_AIO_NR,
1159 		.procname	= "aio-nr",
1160 		.data		= &aio_nr,
1161 		.maxlen		= sizeof(aio_nr),
1162 		.mode		= 0444,
1163 		.proc_handler	= &proc_doulongvec_minmax,
1164 	},
1165 	{
1166 		.ctl_name	= FS_AIO_MAX_NR,
1167 		.procname	= "aio-max-nr",
1168 		.data		= &aio_max_nr,
1169 		.maxlen		= sizeof(aio_max_nr),
1170 		.mode		= 0644,
1171 		.proc_handler	= &proc_doulongvec_minmax,
1172 	},
1173 #ifdef CONFIG_INOTIFY_USER
1174 	{
1175 		.ctl_name	= FS_INOTIFY,
1176 		.procname	= "inotify",
1177 		.mode		= 0555,
1178 		.child		= inotify_table,
1179 	},
1180 #endif
1181 #endif
1182 	{
1183 		.ctl_name	= KERN_SETUID_DUMPABLE,
1184 		.procname	= "suid_dumpable",
1185 		.data		= &suid_dumpable,
1186 		.maxlen		= sizeof(int),
1187 		.mode		= 0644,
1188 		.proc_handler	= &proc_dointvec,
1189 	},
1190 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1191 	{
1192 		.ctl_name	= CTL_UNNUMBERED,
1193 		.procname	= "binfmt_misc",
1194 		.mode		= 0555,
1195 		.child		= binfmt_misc_table,
1196 	},
1197 #endif
1198 /*
1199  * NOTE: do not add new entries to this table unless you have read
1200  * Documentation/sysctl/ctl_unnumbered.txt
1201  */
1202 	{ .ctl_name = 0 }
1203 };
1204 
1205 static ctl_table debug_table[] = {
1206 #ifdef CONFIG_X86
1207 	{
1208 		.ctl_name	= CTL_UNNUMBERED,
1209 		.procname	= "exception-trace",
1210 		.data		= &show_unhandled_signals,
1211 		.maxlen		= sizeof(int),
1212 		.mode		= 0644,
1213 		.proc_handler	= proc_dointvec
1214 	},
1215 #endif
1216 	{ .ctl_name = 0 }
1217 };
1218 
1219 static ctl_table dev_table[] = {
1220 	{ .ctl_name = 0 }
1221 };
1222 
1223 static DEFINE_SPINLOCK(sysctl_lock);
1224 
1225 /* called under sysctl_lock */
1226 static int use_table(struct ctl_table_header *p)
1227 {
1228 	if (unlikely(p->unregistering))
1229 		return 0;
1230 	p->used++;
1231 	return 1;
1232 }
1233 
1234 /* called under sysctl_lock */
1235 static void unuse_table(struct ctl_table_header *p)
1236 {
1237 	if (!--p->used)
1238 		if (unlikely(p->unregistering))
1239 			complete(p->unregistering);
1240 }
1241 
1242 /* called under sysctl_lock, will reacquire if has to wait */
1243 static void start_unregistering(struct ctl_table_header *p)
1244 {
1245 	/*
1246 	 * if p->used is 0, nobody will ever touch that entry again;
1247 	 * we'll eliminate all paths to it before dropping sysctl_lock
1248 	 */
1249 	if (unlikely(p->used)) {
1250 		struct completion wait;
1251 		init_completion(&wait);
1252 		p->unregistering = &wait;
1253 		spin_unlock(&sysctl_lock);
1254 		wait_for_completion(&wait);
1255 		spin_lock(&sysctl_lock);
1256 	}
1257 	/*
1258 	 * do not remove from the list until nobody holds it; walking the
1259 	 * list in do_sysctl() relies on that.
1260 	 */
1261 	list_del_init(&p->ctl_entry);
1262 }
1263 
1264 void sysctl_head_finish(struct ctl_table_header *head)
1265 {
1266 	if (!head)
1267 		return;
1268 	spin_lock(&sysctl_lock);
1269 	unuse_table(head);
1270 	spin_unlock(&sysctl_lock);
1271 }
1272 
1273 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1274 {
1275 	struct ctl_table_header *head;
1276 	struct list_head *tmp;
1277 	spin_lock(&sysctl_lock);
1278 	if (prev) {
1279 		tmp = &prev->ctl_entry;
1280 		unuse_table(prev);
1281 		goto next;
1282 	}
1283 	tmp = &root_table_header.ctl_entry;
1284 	for (;;) {
1285 		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1286 
1287 		if (!use_table(head))
1288 			goto next;
1289 		spin_unlock(&sysctl_lock);
1290 		return head;
1291 	next:
1292 		tmp = tmp->next;
1293 		if (tmp == &root_table_header.ctl_entry)
1294 			break;
1295 	}
1296 	spin_unlock(&sysctl_lock);
1297 	return NULL;
1298 }
1299 
1300 #ifdef CONFIG_SYSCTL_SYSCALL
1301 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1302 	       void __user *newval, size_t newlen)
1303 {
1304 	struct ctl_table_header *head;
1305 	int error = -ENOTDIR;
1306 
1307 	if (nlen <= 0 || nlen >= CTL_MAXNAME)
1308 		return -ENOTDIR;
1309 	if (oldval) {
1310 		int old_len;
1311 		if (!oldlenp || get_user(old_len, oldlenp))
1312 			return -EFAULT;
1313 	}
1314 
1315 	for (head = sysctl_head_next(NULL); head;
1316 			head = sysctl_head_next(head)) {
1317 		error = parse_table(name, nlen, oldval, oldlenp,
1318 					newval, newlen, head->ctl_table);
1319 		if (error != -ENOTDIR) {
1320 			sysctl_head_finish(head);
1321 			break;
1322 		}
1323 	}
1324 	return error;
1325 }
1326 
1327 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1328 {
1329 	struct __sysctl_args tmp;
1330 	int error;
1331 
1332 	if (copy_from_user(&tmp, args, sizeof(tmp)))
1333 		return -EFAULT;
1334 
1335 	lock_kernel();
1336 	error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1337 			  tmp.newval, tmp.newlen);
1338 	unlock_kernel();
1339 	return error;
1340 }
1341 #endif /* CONFIG_SYSCTL_SYSCALL */
1342 
1343 /*
1344  * sysctl_perm does NOT grant the superuser all rights automatically, because
1345  * some sysctl variables are readonly even to root.
1346  */
1347 
1348 static int test_perm(int mode, int op)
1349 {
1350 	if (!current->euid)
1351 		mode >>= 6;
1352 	else if (in_egroup_p(0))
1353 		mode >>= 3;
1354 	if ((mode & op & 0007) == op)
1355 		return 0;
1356 	return -EACCES;
1357 }
1358 
1359 int sysctl_perm(ctl_table *table, int op)
1360 {
1361 	int error;
1362 	error = security_sysctl(table, op);
1363 	if (error)
1364 		return error;
1365 	return test_perm(table->mode, op);
1366 }
1367 
1368 #ifdef CONFIG_SYSCTL_SYSCALL
1369 static int parse_table(int __user *name, int nlen,
1370 		       void __user *oldval, size_t __user *oldlenp,
1371 		       void __user *newval, size_t newlen,
1372 		       ctl_table *table)
1373 {
1374 	int n;
1375 repeat:
1376 	if (!nlen)
1377 		return -ENOTDIR;
1378 	if (get_user(n, name))
1379 		return -EFAULT;
1380 	for ( ; table->ctl_name || table->procname; table++) {
1381 		if (!table->ctl_name)
1382 			continue;
1383 		if (n == table->ctl_name) {
1384 			int error;
1385 			if (table->child) {
1386 				if (sysctl_perm(table, 001))
1387 					return -EPERM;
1388 				name++;
1389 				nlen--;
1390 				table = table->child;
1391 				goto repeat;
1392 			}
1393 			error = do_sysctl_strategy(table, name, nlen,
1394 						   oldval, oldlenp,
1395 						   newval, newlen);
1396 			return error;
1397 		}
1398 	}
1399 	return -ENOTDIR;
1400 }
1401 
1402 /* Perform the actual read/write of a sysctl table entry. */
1403 int do_sysctl_strategy (ctl_table *table,
1404 			int __user *name, int nlen,
1405 			void __user *oldval, size_t __user *oldlenp,
1406 			void __user *newval, size_t newlen)
1407 {
1408 	int op = 0, rc;
1409 	size_t len;
1410 
1411 	if (oldval)
1412 		op |= 004;
1413 	if (newval)
1414 		op |= 002;
1415 	if (sysctl_perm(table, op))
1416 		return -EPERM;
1417 
1418 	if (table->strategy) {
1419 		rc = table->strategy(table, name, nlen, oldval, oldlenp,
1420 				     newval, newlen);
1421 		if (rc < 0)
1422 			return rc;
1423 		if (rc > 0)
1424 			return 0;
1425 	}
1426 
1427 	/* If there is no strategy routine, or if the strategy returns
1428 	 * zero, proceed with automatic r/w */
1429 	if (table->data && table->maxlen) {
1430 		if (oldval && oldlenp) {
1431 			if (get_user(len, oldlenp))
1432 				return -EFAULT;
1433 			if (len) {
1434 				if (len > table->maxlen)
1435 					len = table->maxlen;
1436 				if(copy_to_user(oldval, table->data, len))
1437 					return -EFAULT;
1438 				if(put_user(len, oldlenp))
1439 					return -EFAULT;
1440 			}
1441 		}
1442 		if (newval && newlen) {
1443 			len = newlen;
1444 			if (len > table->maxlen)
1445 				len = table->maxlen;
1446 			if(copy_from_user(table->data, newval, len))
1447 				return -EFAULT;
1448 		}
1449 	}
1450 	return 0;
1451 }
1452 #endif /* CONFIG_SYSCTL_SYSCALL */
1453 
1454 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1455 {
1456 	for (; table->ctl_name || table->procname; table++) {
1457 		table->parent = parent;
1458 		if (table->child)
1459 			sysctl_set_parent(table, table->child);
1460 	}
1461 }
1462 
1463 static __init int sysctl_init(void)
1464 {
1465 	sysctl_set_parent(NULL, root_table);
1466 	return 0;
1467 }
1468 
1469 core_initcall(sysctl_init);
1470 
1471 /**
1472  * register_sysctl_table - register a sysctl hierarchy
1473  * @table: the top-level table structure
1474  *
1475  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1476  * array. An entry with a ctl_name of 0 terminates the table.
1477  *
1478  * The members of the &ctl_table structure are used as follows:
1479  *
1480  * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1481  *            must be unique within that level of sysctl
1482  *
1483  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1484  *            enter a sysctl file
1485  *
1486  * data - a pointer to data for use by proc_handler
1487  *
1488  * maxlen - the maximum size in bytes of the data
1489  *
1490  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1491  *
1492  * child - a pointer to the child sysctl table if this entry is a directory, or
1493  *         %NULL.
1494  *
1495  * proc_handler - the text handler routine (described below)
1496  *
1497  * strategy - the strategy routine (described below)
1498  *
1499  * de - for internal use by the sysctl routines
1500  *
1501  * extra1, extra2 - extra pointers usable by the proc handler routines
1502  *
1503  * Leaf nodes in the sysctl tree will be represented by a single file
1504  * under /proc; non-leaf nodes will be represented by directories.
1505  *
1506  * sysctl(2) can automatically manage read and write requests through
1507  * the sysctl table.  The data and maxlen fields of the ctl_table
1508  * struct enable minimal validation of the values being written to be
1509  * performed, and the mode field allows minimal authentication.
1510  *
1511  * More sophisticated management can be enabled by the provision of a
1512  * strategy routine with the table entry.  This will be called before
1513  * any automatic read or write of the data is performed.
1514  *
1515  * The strategy routine may return
1516  *
1517  * < 0 - Error occurred (error is passed to user process)
1518  *
1519  * 0   - OK - proceed with automatic read or write.
1520  *
1521  * > 0 - OK - read or write has been done by the strategy routine, so
1522  *       return immediately.
1523  *
1524  * There must be a proc_handler routine for any terminal nodes
1525  * mirrored under /proc/sys (non-terminals are handled by a built-in
1526  * directory handler).  Several default handlers are available to
1527  * cover common cases -
1528  *
1529  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1530  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1531  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1532  *
1533  * It is the handler's job to read the input buffer from user memory
1534  * and process it. The handler should return 0 on success.
1535  *
1536  * This routine returns %NULL on a failure to register, and a pointer
1537  * to the table header on success.
1538  */
1539 struct ctl_table_header *register_sysctl_table(ctl_table * table)
1540 {
1541 	struct ctl_table_header *tmp;
1542 	tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1543 	if (!tmp)
1544 		return NULL;
1545 	tmp->ctl_table = table;
1546 	INIT_LIST_HEAD(&tmp->ctl_entry);
1547 	tmp->used = 0;
1548 	tmp->unregistering = NULL;
1549 	sysctl_set_parent(NULL, table);
1550 	spin_lock(&sysctl_lock);
1551 	list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
1552 	spin_unlock(&sysctl_lock);
1553 	return tmp;
1554 }
1555 
1556 /**
1557  * unregister_sysctl_table - unregister a sysctl table hierarchy
1558  * @header: the header returned from register_sysctl_table
1559  *
1560  * Unregisters the sysctl table and all children. proc entries may not
1561  * actually be removed until they are no longer used by anyone.
1562  */
1563 void unregister_sysctl_table(struct ctl_table_header * header)
1564 {
1565 	might_sleep();
1566 	spin_lock(&sysctl_lock);
1567 	start_unregistering(header);
1568 	spin_unlock(&sysctl_lock);
1569 	kfree(header);
1570 }
1571 
1572 #else /* !CONFIG_SYSCTL */
1573 struct ctl_table_header *register_sysctl_table(ctl_table * table)
1574 {
1575 	return NULL;
1576 }
1577 
1578 void unregister_sysctl_table(struct ctl_table_header * table)
1579 {
1580 }
1581 
1582 #endif /* CONFIG_SYSCTL */
1583 
1584 /*
1585  * /proc/sys support
1586  */
1587 
1588 #ifdef CONFIG_PROC_SYSCTL
1589 
1590 static int _proc_do_string(void* data, int maxlen, int write,
1591 			   struct file *filp, void __user *buffer,
1592 			   size_t *lenp, loff_t *ppos)
1593 {
1594 	size_t len;
1595 	char __user *p;
1596 	char c;
1597 
1598 	if (!data || !maxlen || !*lenp) {
1599 		*lenp = 0;
1600 		return 0;
1601 	}
1602 
1603 	if (write) {
1604 		len = 0;
1605 		p = buffer;
1606 		while (len < *lenp) {
1607 			if (get_user(c, p++))
1608 				return -EFAULT;
1609 			if (c == 0 || c == '\n')
1610 				break;
1611 			len++;
1612 		}
1613 		if (len >= maxlen)
1614 			len = maxlen-1;
1615 		if(copy_from_user(data, buffer, len))
1616 			return -EFAULT;
1617 		((char *) data)[len] = 0;
1618 		*ppos += *lenp;
1619 	} else {
1620 		len = strlen(data);
1621 		if (len > maxlen)
1622 			len = maxlen;
1623 
1624 		if (*ppos > len) {
1625 			*lenp = 0;
1626 			return 0;
1627 		}
1628 
1629 		data += *ppos;
1630 		len  -= *ppos;
1631 
1632 		if (len > *lenp)
1633 			len = *lenp;
1634 		if (len)
1635 			if(copy_to_user(buffer, data, len))
1636 				return -EFAULT;
1637 		if (len < *lenp) {
1638 			if(put_user('\n', ((char __user *) buffer) + len))
1639 				return -EFAULT;
1640 			len++;
1641 		}
1642 		*lenp = len;
1643 		*ppos += len;
1644 	}
1645 	return 0;
1646 }
1647 
1648 /**
1649  * proc_dostring - read a string sysctl
1650  * @table: the sysctl table
1651  * @write: %TRUE if this is a write to the sysctl file
1652  * @filp: the file structure
1653  * @buffer: the user buffer
1654  * @lenp: the size of the user buffer
1655  * @ppos: file position
1656  *
1657  * Reads/writes a string from/to the user buffer. If the kernel
1658  * buffer provided is not large enough to hold the string, the
1659  * string is truncated. The copied string is %NULL-terminated.
1660  * If the string is being read by the user process, it is copied
1661  * and a newline '\n' is added. It is truncated if the buffer is
1662  * not large enough.
1663  *
1664  * Returns 0 on success.
1665  */
1666 int proc_dostring(ctl_table *table, int write, struct file *filp,
1667 		  void __user *buffer, size_t *lenp, loff_t *ppos)
1668 {
1669 	return _proc_do_string(table->data, table->maxlen, write, filp,
1670 			       buffer, lenp, ppos);
1671 }
1672 
1673 
1674 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1675 				 int *valp,
1676 				 int write, void *data)
1677 {
1678 	if (write) {
1679 		*valp = *negp ? -*lvalp : *lvalp;
1680 	} else {
1681 		int val = *valp;
1682 		if (val < 0) {
1683 			*negp = -1;
1684 			*lvalp = (unsigned long)-val;
1685 		} else {
1686 			*negp = 0;
1687 			*lvalp = (unsigned long)val;
1688 		}
1689 	}
1690 	return 0;
1691 }
1692 
1693 static int __do_proc_dointvec(void *tbl_data, ctl_table *table,
1694 		  int write, struct file *filp, void __user *buffer,
1695 		  size_t *lenp, loff_t *ppos,
1696 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1697 			      int write, void *data),
1698 		  void *data)
1699 {
1700 #define TMPBUFLEN 21
1701 	int *i, vleft, first=1, neg, val;
1702 	unsigned long lval;
1703 	size_t left, len;
1704 
1705 	char buf[TMPBUFLEN], *p;
1706 	char __user *s = buffer;
1707 
1708 	if (!tbl_data || !table->maxlen || !*lenp ||
1709 	    (*ppos && !write)) {
1710 		*lenp = 0;
1711 		return 0;
1712 	}
1713 
1714 	i = (int *) tbl_data;
1715 	vleft = table->maxlen / sizeof(*i);
1716 	left = *lenp;
1717 
1718 	if (!conv)
1719 		conv = do_proc_dointvec_conv;
1720 
1721 	for (; left && vleft--; i++, first=0) {
1722 		if (write) {
1723 			while (left) {
1724 				char c;
1725 				if (get_user(c, s))
1726 					return -EFAULT;
1727 				if (!isspace(c))
1728 					break;
1729 				left--;
1730 				s++;
1731 			}
1732 			if (!left)
1733 				break;
1734 			neg = 0;
1735 			len = left;
1736 			if (len > sizeof(buf) - 1)
1737 				len = sizeof(buf) - 1;
1738 			if (copy_from_user(buf, s, len))
1739 				return -EFAULT;
1740 			buf[len] = 0;
1741 			p = buf;
1742 			if (*p == '-' && left > 1) {
1743 				neg = 1;
1744 				p++;
1745 			}
1746 			if (*p < '0' || *p > '9')
1747 				break;
1748 
1749 			lval = simple_strtoul(p, &p, 0);
1750 
1751 			len = p-buf;
1752 			if ((len < left) && *p && !isspace(*p))
1753 				break;
1754 			if (neg)
1755 				val = -val;
1756 			s += len;
1757 			left -= len;
1758 
1759 			if (conv(&neg, &lval, i, 1, data))
1760 				break;
1761 		} else {
1762 			p = buf;
1763 			if (!first)
1764 				*p++ = '\t';
1765 
1766 			if (conv(&neg, &lval, i, 0, data))
1767 				break;
1768 
1769 			sprintf(p, "%s%lu", neg ? "-" : "", lval);
1770 			len = strlen(buf);
1771 			if (len > left)
1772 				len = left;
1773 			if(copy_to_user(s, buf, len))
1774 				return -EFAULT;
1775 			left -= len;
1776 			s += len;
1777 		}
1778 	}
1779 
1780 	if (!write && !first && left) {
1781 		if(put_user('\n', s))
1782 			return -EFAULT;
1783 		left--, s++;
1784 	}
1785 	if (write) {
1786 		while (left) {
1787 			char c;
1788 			if (get_user(c, s++))
1789 				return -EFAULT;
1790 			if (!isspace(c))
1791 				break;
1792 			left--;
1793 		}
1794 	}
1795 	if (write && first)
1796 		return -EINVAL;
1797 	*lenp -= left;
1798 	*ppos += *lenp;
1799 	return 0;
1800 #undef TMPBUFLEN
1801 }
1802 
1803 static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1804 		  void __user *buffer, size_t *lenp, loff_t *ppos,
1805 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1806 			      int write, void *data),
1807 		  void *data)
1808 {
1809 	return __do_proc_dointvec(table->data, table, write, filp,
1810 			buffer, lenp, ppos, conv, data);
1811 }
1812 
1813 /**
1814  * proc_dointvec - read a vector of integers
1815  * @table: the sysctl table
1816  * @write: %TRUE if this is a write to the sysctl file
1817  * @filp: the file structure
1818  * @buffer: the user buffer
1819  * @lenp: the size of the user buffer
1820  * @ppos: file position
1821  *
1822  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1823  * values from/to the user buffer, treated as an ASCII string.
1824  *
1825  * Returns 0 on success.
1826  */
1827 int proc_dointvec(ctl_table *table, int write, struct file *filp,
1828 		     void __user *buffer, size_t *lenp, loff_t *ppos)
1829 {
1830     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1831 		    	    NULL,NULL);
1832 }
1833 
1834 #define OP_SET	0
1835 #define OP_AND	1
1836 #define OP_OR	2
1837 
1838 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1839 				      int *valp,
1840 				      int write, void *data)
1841 {
1842 	int op = *(int *)data;
1843 	if (write) {
1844 		int val = *negp ? -*lvalp : *lvalp;
1845 		switch(op) {
1846 		case OP_SET:	*valp = val; break;
1847 		case OP_AND:	*valp &= val; break;
1848 		case OP_OR:	*valp |= val; break;
1849 		}
1850 	} else {
1851 		int val = *valp;
1852 		if (val < 0) {
1853 			*negp = -1;
1854 			*lvalp = (unsigned long)-val;
1855 		} else {
1856 			*negp = 0;
1857 			*lvalp = (unsigned long)val;
1858 		}
1859 	}
1860 	return 0;
1861 }
1862 
1863 /*
1864  *	init may raise the set.
1865  */
1866 
1867 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1868 			void __user *buffer, size_t *lenp, loff_t *ppos)
1869 {
1870 	int op;
1871 
1872 	if (write && !capable(CAP_SYS_MODULE)) {
1873 		return -EPERM;
1874 	}
1875 
1876 	op = is_init(current) ? OP_SET : OP_AND;
1877 	return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1878 				do_proc_dointvec_bset_conv,&op);
1879 }
1880 
1881 /*
1882  *	Taint values can only be increased
1883  */
1884 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
1885 			       void __user *buffer, size_t *lenp, loff_t *ppos)
1886 {
1887 	int op;
1888 
1889 	if (write && !capable(CAP_SYS_ADMIN))
1890 		return -EPERM;
1891 
1892 	op = OP_OR;
1893 	return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1894 				do_proc_dointvec_bset_conv,&op);
1895 }
1896 
1897 struct do_proc_dointvec_minmax_conv_param {
1898 	int *min;
1899 	int *max;
1900 };
1901 
1902 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1903 					int *valp,
1904 					int write, void *data)
1905 {
1906 	struct do_proc_dointvec_minmax_conv_param *param = data;
1907 	if (write) {
1908 		int val = *negp ? -*lvalp : *lvalp;
1909 		if ((param->min && *param->min > val) ||
1910 		    (param->max && *param->max < val))
1911 			return -EINVAL;
1912 		*valp = val;
1913 	} else {
1914 		int val = *valp;
1915 		if (val < 0) {
1916 			*negp = -1;
1917 			*lvalp = (unsigned long)-val;
1918 		} else {
1919 			*negp = 0;
1920 			*lvalp = (unsigned long)val;
1921 		}
1922 	}
1923 	return 0;
1924 }
1925 
1926 /**
1927  * proc_dointvec_minmax - read a vector of integers with min/max values
1928  * @table: the sysctl table
1929  * @write: %TRUE if this is a write to the sysctl file
1930  * @filp: the file structure
1931  * @buffer: the user buffer
1932  * @lenp: the size of the user buffer
1933  * @ppos: file position
1934  *
1935  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1936  * values from/to the user buffer, treated as an ASCII string.
1937  *
1938  * This routine will ensure the values are within the range specified by
1939  * table->extra1 (min) and table->extra2 (max).
1940  *
1941  * Returns 0 on success.
1942  */
1943 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1944 		  void __user *buffer, size_t *lenp, loff_t *ppos)
1945 {
1946 	struct do_proc_dointvec_minmax_conv_param param = {
1947 		.min = (int *) table->extra1,
1948 		.max = (int *) table->extra2,
1949 	};
1950 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1951 				do_proc_dointvec_minmax_conv, &param);
1952 }
1953 
1954 static int __do_proc_doulongvec_minmax(void *data, ctl_table *table, int write,
1955 				     struct file *filp,
1956 				     void __user *buffer,
1957 				     size_t *lenp, loff_t *ppos,
1958 				     unsigned long convmul,
1959 				     unsigned long convdiv)
1960 {
1961 #define TMPBUFLEN 21
1962 	unsigned long *i, *min, *max, val;
1963 	int vleft, first=1, neg;
1964 	size_t len, left;
1965 	char buf[TMPBUFLEN], *p;
1966 	char __user *s = buffer;
1967 
1968 	if (!data || !table->maxlen || !*lenp ||
1969 	    (*ppos && !write)) {
1970 		*lenp = 0;
1971 		return 0;
1972 	}
1973 
1974 	i = (unsigned long *) data;
1975 	min = (unsigned long *) table->extra1;
1976 	max = (unsigned long *) table->extra2;
1977 	vleft = table->maxlen / sizeof(unsigned long);
1978 	left = *lenp;
1979 
1980 	for (; left && vleft--; i++, min++, max++, first=0) {
1981 		if (write) {
1982 			while (left) {
1983 				char c;
1984 				if (get_user(c, s))
1985 					return -EFAULT;
1986 				if (!isspace(c))
1987 					break;
1988 				left--;
1989 				s++;
1990 			}
1991 			if (!left)
1992 				break;
1993 			neg = 0;
1994 			len = left;
1995 			if (len > TMPBUFLEN-1)
1996 				len = TMPBUFLEN-1;
1997 			if (copy_from_user(buf, s, len))
1998 				return -EFAULT;
1999 			buf[len] = 0;
2000 			p = buf;
2001 			if (*p == '-' && left > 1) {
2002 				neg = 1;
2003 				p++;
2004 			}
2005 			if (*p < '0' || *p > '9')
2006 				break;
2007 			val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2008 			len = p-buf;
2009 			if ((len < left) && *p && !isspace(*p))
2010 				break;
2011 			if (neg)
2012 				val = -val;
2013 			s += len;
2014 			left -= len;
2015 
2016 			if(neg)
2017 				continue;
2018 			if ((min && val < *min) || (max && val > *max))
2019 				continue;
2020 			*i = val;
2021 		} else {
2022 			p = buf;
2023 			if (!first)
2024 				*p++ = '\t';
2025 			sprintf(p, "%lu", convdiv * (*i) / convmul);
2026 			len = strlen(buf);
2027 			if (len > left)
2028 				len = left;
2029 			if(copy_to_user(s, buf, len))
2030 				return -EFAULT;
2031 			left -= len;
2032 			s += len;
2033 		}
2034 	}
2035 
2036 	if (!write && !first && left) {
2037 		if(put_user('\n', s))
2038 			return -EFAULT;
2039 		left--, s++;
2040 	}
2041 	if (write) {
2042 		while (left) {
2043 			char c;
2044 			if (get_user(c, s++))
2045 				return -EFAULT;
2046 			if (!isspace(c))
2047 				break;
2048 			left--;
2049 		}
2050 	}
2051 	if (write && first)
2052 		return -EINVAL;
2053 	*lenp -= left;
2054 	*ppos += *lenp;
2055 	return 0;
2056 #undef TMPBUFLEN
2057 }
2058 
2059 static int do_proc_doulongvec_minmax(ctl_table *table, int write,
2060 				     struct file *filp,
2061 				     void __user *buffer,
2062 				     size_t *lenp, loff_t *ppos,
2063 				     unsigned long convmul,
2064 				     unsigned long convdiv)
2065 {
2066 	return __do_proc_doulongvec_minmax(table->data, table, write,
2067 			filp, buffer, lenp, ppos, convmul, convdiv);
2068 }
2069 
2070 /**
2071  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2072  * @table: the sysctl table
2073  * @write: %TRUE if this is a write to the sysctl file
2074  * @filp: the file structure
2075  * @buffer: the user buffer
2076  * @lenp: the size of the user buffer
2077  * @ppos: file position
2078  *
2079  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2080  * values from/to the user buffer, treated as an ASCII string.
2081  *
2082  * This routine will ensure the values are within the range specified by
2083  * table->extra1 (min) and table->extra2 (max).
2084  *
2085  * Returns 0 on success.
2086  */
2087 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2088 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2089 {
2090     return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2091 }
2092 
2093 /**
2094  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2095  * @table: the sysctl table
2096  * @write: %TRUE if this is a write to the sysctl file
2097  * @filp: the file structure
2098  * @buffer: the user buffer
2099  * @lenp: the size of the user buffer
2100  * @ppos: file position
2101  *
2102  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2103  * values from/to the user buffer, treated as an ASCII string. The values
2104  * are treated as milliseconds, and converted to jiffies when they are stored.
2105  *
2106  * This routine will ensure the values are within the range specified by
2107  * table->extra1 (min) and table->extra2 (max).
2108  *
2109  * Returns 0 on success.
2110  */
2111 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2112 				      struct file *filp,
2113 				      void __user *buffer,
2114 				      size_t *lenp, loff_t *ppos)
2115 {
2116     return do_proc_doulongvec_minmax(table, write, filp, buffer,
2117 				     lenp, ppos, HZ, 1000l);
2118 }
2119 
2120 
2121 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2122 					 int *valp,
2123 					 int write, void *data)
2124 {
2125 	if (write) {
2126 		if (*lvalp > LONG_MAX / HZ)
2127 			return 1;
2128 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2129 	} else {
2130 		int val = *valp;
2131 		unsigned long lval;
2132 		if (val < 0) {
2133 			*negp = -1;
2134 			lval = (unsigned long)-val;
2135 		} else {
2136 			*negp = 0;
2137 			lval = (unsigned long)val;
2138 		}
2139 		*lvalp = lval / HZ;
2140 	}
2141 	return 0;
2142 }
2143 
2144 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2145 						int *valp,
2146 						int write, void *data)
2147 {
2148 	if (write) {
2149 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2150 			return 1;
2151 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2152 	} else {
2153 		int val = *valp;
2154 		unsigned long lval;
2155 		if (val < 0) {
2156 			*negp = -1;
2157 			lval = (unsigned long)-val;
2158 		} else {
2159 			*negp = 0;
2160 			lval = (unsigned long)val;
2161 		}
2162 		*lvalp = jiffies_to_clock_t(lval);
2163 	}
2164 	return 0;
2165 }
2166 
2167 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2168 					    int *valp,
2169 					    int write, void *data)
2170 {
2171 	if (write) {
2172 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2173 	} else {
2174 		int val = *valp;
2175 		unsigned long lval;
2176 		if (val < 0) {
2177 			*negp = -1;
2178 			lval = (unsigned long)-val;
2179 		} else {
2180 			*negp = 0;
2181 			lval = (unsigned long)val;
2182 		}
2183 		*lvalp = jiffies_to_msecs(lval);
2184 	}
2185 	return 0;
2186 }
2187 
2188 /**
2189  * proc_dointvec_jiffies - read a vector of integers as seconds
2190  * @table: the sysctl table
2191  * @write: %TRUE if this is a write to the sysctl file
2192  * @filp: the file structure
2193  * @buffer: the user buffer
2194  * @lenp: the size of the user buffer
2195  * @ppos: file position
2196  *
2197  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2198  * values from/to the user buffer, treated as an ASCII string.
2199  * The values read are assumed to be in seconds, and are converted into
2200  * jiffies.
2201  *
2202  * Returns 0 on success.
2203  */
2204 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2205 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2206 {
2207     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2208 		    	    do_proc_dointvec_jiffies_conv,NULL);
2209 }
2210 
2211 /**
2212  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2213  * @table: the sysctl table
2214  * @write: %TRUE if this is a write to the sysctl file
2215  * @filp: the file structure
2216  * @buffer: the user buffer
2217  * @lenp: the size of the user buffer
2218  * @ppos: pointer to the file position
2219  *
2220  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2221  * values from/to the user buffer, treated as an ASCII string.
2222  * The values read are assumed to be in 1/USER_HZ seconds, and
2223  * are converted into jiffies.
2224  *
2225  * Returns 0 on success.
2226  */
2227 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2228 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2229 {
2230     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2231 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2232 }
2233 
2234 /**
2235  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2236  * @table: the sysctl table
2237  * @write: %TRUE if this is a write to the sysctl file
2238  * @filp: the file structure
2239  * @buffer: the user buffer
2240  * @lenp: the size of the user buffer
2241  * @ppos: file position
2242  * @ppos: the current position in the file
2243  *
2244  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2245  * values from/to the user buffer, treated as an ASCII string.
2246  * The values read are assumed to be in 1/1000 seconds, and
2247  * are converted into jiffies.
2248  *
2249  * Returns 0 on success.
2250  */
2251 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2252 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2253 {
2254 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2255 				do_proc_dointvec_ms_jiffies_conv, NULL);
2256 }
2257 
2258 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
2259 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2260 {
2261 	struct pid *new_pid;
2262 	pid_t tmp;
2263 	int r;
2264 
2265 	tmp = pid_nr(cad_pid);
2266 
2267 	r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2268 			       lenp, ppos, NULL, NULL);
2269 	if (r || !write)
2270 		return r;
2271 
2272 	new_pid = find_get_pid(tmp);
2273 	if (!new_pid)
2274 		return -ESRCH;
2275 
2276 	put_pid(xchg(&cad_pid, new_pid));
2277 	return 0;
2278 }
2279 
2280 #else /* CONFIG_PROC_FS */
2281 
2282 int proc_dostring(ctl_table *table, int write, struct file *filp,
2283 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2284 {
2285 	return -ENOSYS;
2286 }
2287 
2288 int proc_dointvec(ctl_table *table, int write, struct file *filp,
2289 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2290 {
2291 	return -ENOSYS;
2292 }
2293 
2294 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2295 			void __user *buffer, size_t *lenp, loff_t *ppos)
2296 {
2297 	return -ENOSYS;
2298 }
2299 
2300 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2301 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2302 {
2303 	return -ENOSYS;
2304 }
2305 
2306 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2307 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2308 {
2309 	return -ENOSYS;
2310 }
2311 
2312 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2313 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2314 {
2315 	return -ENOSYS;
2316 }
2317 
2318 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2319 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2320 {
2321 	return -ENOSYS;
2322 }
2323 
2324 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2325 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2326 {
2327 	return -ENOSYS;
2328 }
2329 
2330 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2331 				      struct file *filp,
2332 				      void __user *buffer,
2333 				      size_t *lenp, loff_t *ppos)
2334 {
2335     return -ENOSYS;
2336 }
2337 
2338 
2339 #endif /* CONFIG_PROC_FS */
2340 
2341 
2342 #ifdef CONFIG_SYSCTL_SYSCALL
2343 /*
2344  * General sysctl support routines
2345  */
2346 
2347 /* The generic string strategy routine: */
2348 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2349 		  void __user *oldval, size_t __user *oldlenp,
2350 		  void __user *newval, size_t newlen)
2351 {
2352 	if (!table->data || !table->maxlen)
2353 		return -ENOTDIR;
2354 
2355 	if (oldval && oldlenp) {
2356 		size_t bufsize;
2357 		if (get_user(bufsize, oldlenp))
2358 			return -EFAULT;
2359 		if (bufsize) {
2360 			size_t len = strlen(table->data), copied;
2361 
2362 			/* This shouldn't trigger for a well-formed sysctl */
2363 			if (len > table->maxlen)
2364 				len = table->maxlen;
2365 
2366 			/* Copy up to a max of bufsize-1 bytes of the string */
2367 			copied = (len >= bufsize) ? bufsize - 1 : len;
2368 
2369 			if (copy_to_user(oldval, table->data, copied) ||
2370 			    put_user(0, (char __user *)(oldval + copied)))
2371 				return -EFAULT;
2372 			if (put_user(len, oldlenp))
2373 				return -EFAULT;
2374 		}
2375 	}
2376 	if (newval && newlen) {
2377 		size_t len = newlen;
2378 		if (len > table->maxlen)
2379 			len = table->maxlen;
2380 		if(copy_from_user(table->data, newval, len))
2381 			return -EFAULT;
2382 		if (len == table->maxlen)
2383 			len--;
2384 		((char *) table->data)[len] = 0;
2385 	}
2386 	return 1;
2387 }
2388 
2389 /*
2390  * This function makes sure that all of the integers in the vector
2391  * are between the minimum and maximum values given in the arrays
2392  * table->extra1 and table->extra2, respectively.
2393  */
2394 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2395 		void __user *oldval, size_t __user *oldlenp,
2396 		void __user *newval, size_t newlen)
2397 {
2398 
2399 	if (newval && newlen) {
2400 		int __user *vec = (int __user *) newval;
2401 		int *min = (int *) table->extra1;
2402 		int *max = (int *) table->extra2;
2403 		size_t length;
2404 		int i;
2405 
2406 		if (newlen % sizeof(int) != 0)
2407 			return -EINVAL;
2408 
2409 		if (!table->extra1 && !table->extra2)
2410 			return 0;
2411 
2412 		if (newlen > table->maxlen)
2413 			newlen = table->maxlen;
2414 		length = newlen / sizeof(int);
2415 
2416 		for (i = 0; i < length; i++) {
2417 			int value;
2418 			if (get_user(value, vec + i))
2419 				return -EFAULT;
2420 			if (min && value < min[i])
2421 				return -EINVAL;
2422 			if (max && value > max[i])
2423 				return -EINVAL;
2424 		}
2425 	}
2426 	return 0;
2427 }
2428 
2429 /* Strategy function to convert jiffies to seconds */
2430 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2431 		void __user *oldval, size_t __user *oldlenp,
2432 		void __user *newval, size_t newlen)
2433 {
2434 	if (oldval && oldlenp) {
2435 		size_t olen;
2436 
2437 		if (get_user(olen, oldlenp))
2438 			return -EFAULT;
2439 		if (olen) {
2440 			int val;
2441 
2442 			if (olen < sizeof(int))
2443 				return -EINVAL;
2444 
2445 			val = *(int *)(table->data) / HZ;
2446 			if (put_user(val, (int __user *)oldval))
2447 				return -EFAULT;
2448 			if (put_user(sizeof(int), oldlenp))
2449 				return -EFAULT;
2450 		}
2451 	}
2452 	if (newval && newlen) {
2453 		int new;
2454 		if (newlen != sizeof(int))
2455 			return -EINVAL;
2456 		if (get_user(new, (int __user *)newval))
2457 			return -EFAULT;
2458 		*(int *)(table->data) = new*HZ;
2459 	}
2460 	return 1;
2461 }
2462 
2463 /* Strategy function to convert jiffies to seconds */
2464 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2465 		void __user *oldval, size_t __user *oldlenp,
2466 		void __user *newval, size_t newlen)
2467 {
2468 	if (oldval && oldlenp) {
2469 		size_t olen;
2470 
2471 		if (get_user(olen, oldlenp))
2472 			return -EFAULT;
2473 		if (olen) {
2474 			int val;
2475 
2476 			if (olen < sizeof(int))
2477 				return -EINVAL;
2478 
2479 			val = jiffies_to_msecs(*(int *)(table->data));
2480 			if (put_user(val, (int __user *)oldval))
2481 				return -EFAULT;
2482 			if (put_user(sizeof(int), oldlenp))
2483 				return -EFAULT;
2484 		}
2485 	}
2486 	if (newval && newlen) {
2487 		int new;
2488 		if (newlen != sizeof(int))
2489 			return -EINVAL;
2490 		if (get_user(new, (int __user *)newval))
2491 			return -EFAULT;
2492 		*(int *)(table->data) = msecs_to_jiffies(new);
2493 	}
2494 	return 1;
2495 }
2496 
2497 
2498 
2499 #else /* CONFIG_SYSCTL_SYSCALL */
2500 
2501 
2502 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2503 {
2504 	static int msg_count;
2505 	struct __sysctl_args tmp;
2506 	int name[CTL_MAXNAME];
2507 	int i;
2508 
2509 	/* Read in the sysctl name for better debug message logging */
2510 	if (copy_from_user(&tmp, args, sizeof(tmp)))
2511 		return -EFAULT;
2512 	if (tmp.nlen <= 0 || tmp.nlen >= CTL_MAXNAME)
2513 		return -ENOTDIR;
2514 	for (i = 0; i < tmp.nlen; i++)
2515 		if (get_user(name[i], tmp.name + i))
2516 			return -EFAULT;
2517 
2518 	/* Ignore accesses to kernel.version */
2519 	if ((tmp.nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2520 		goto out;
2521 
2522 	if (msg_count < 5) {
2523 		msg_count++;
2524 		printk(KERN_INFO
2525 			"warning: process `%s' used the removed sysctl "
2526 			"system call with ", current->comm);
2527 		for (i = 0; i < tmp.nlen; i++)
2528 			printk("%d.", name[i]);
2529 		printk("\n");
2530 	}
2531 out:
2532 	return -ENOSYS;
2533 }
2534 
2535 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2536 		  void __user *oldval, size_t __user *oldlenp,
2537 		  void __user *newval, size_t newlen)
2538 {
2539 	return -ENOSYS;
2540 }
2541 
2542 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2543 		void __user *oldval, size_t __user *oldlenp,
2544 		void __user *newval, size_t newlen)
2545 {
2546 	return -ENOSYS;
2547 }
2548 
2549 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2550 		void __user *oldval, size_t __user *oldlenp,
2551 		void __user *newval, size_t newlen)
2552 {
2553 	return -ENOSYS;
2554 }
2555 
2556 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2557 		void __user *oldval, size_t __user *oldlenp,
2558 		void __user *newval, size_t newlen)
2559 {
2560 	return -ENOSYS;
2561 }
2562 
2563 #endif /* CONFIG_SYSCTL_SYSCALL */
2564 
2565 /*
2566  * No sense putting this after each symbol definition, twice,
2567  * exception granted :-)
2568  */
2569 EXPORT_SYMBOL(proc_dointvec);
2570 EXPORT_SYMBOL(proc_dointvec_jiffies);
2571 EXPORT_SYMBOL(proc_dointvec_minmax);
2572 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2573 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2574 EXPORT_SYMBOL(proc_dostring);
2575 EXPORT_SYMBOL(proc_doulongvec_minmax);
2576 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2577 EXPORT_SYMBOL(register_sysctl_table);
2578 EXPORT_SYMBOL(sysctl_intvec);
2579 EXPORT_SYMBOL(sysctl_jiffies);
2580 EXPORT_SYMBOL(sysctl_ms_jiffies);
2581 EXPORT_SYMBOL(sysctl_string);
2582 EXPORT_SYMBOL(unregister_sysctl_table);
2583