xref: /linux/kernel/sysctl.c (revision 14b42963f64b98ab61fa9723c03d71aa5ef4f862)
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/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/security.h>
41 #include <linux/initrd.h>
42 #include <linux/times.h>
43 #include <linux/limits.h>
44 #include <linux/dcache.h>
45 #include <linux/syscalls.h>
46 #include <linux/nfs_fs.h>
47 #include <linux/acpi.h>
48 
49 #include <asm/uaccess.h>
50 #include <asm/processor.h>
51 
52 extern int proc_nr_files(ctl_table *table, int write, struct file *filp,
53                      void __user *buffer, size_t *lenp, loff_t *ppos);
54 
55 #if defined(CONFIG_SYSCTL)
56 
57 /* External variables not in a header file. */
58 extern int C_A_D;
59 extern int sysctl_overcommit_memory;
60 extern int sysctl_overcommit_ratio;
61 extern int sysctl_panic_on_oom;
62 extern int max_threads;
63 extern int sysrq_enabled;
64 extern int core_uses_pid;
65 extern int suid_dumpable;
66 extern char core_pattern[];
67 extern int cad_pid;
68 extern int pid_max;
69 extern int min_free_kbytes;
70 extern int printk_ratelimit_jiffies;
71 extern int printk_ratelimit_burst;
72 extern int pid_max_min, pid_max_max;
73 extern int sysctl_drop_caches;
74 extern int percpu_pagelist_fraction;
75 extern int compat_log;
76 
77 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
78 int unknown_nmi_panic;
79 extern int proc_unknown_nmi_panic(ctl_table *, int, struct file *,
80 				  void __user *, size_t *, loff_t *);
81 #endif
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 #ifdef CONFIG_SYSVIPC
97 extern size_t shm_ctlmax;
98 extern size_t shm_ctlall;
99 extern int shm_ctlmni;
100 extern int msg_ctlmax;
101 extern int msg_ctlmnb;
102 extern int msg_ctlmni;
103 extern int sem_ctls[];
104 #endif
105 
106 #ifdef __sparc__
107 extern char reboot_command [];
108 extern int stop_a_enabled;
109 extern int scons_pwroff;
110 #endif
111 
112 #ifdef __hppa__
113 extern int pwrsw_enabled;
114 extern int unaligned_enabled;
115 #endif
116 
117 #ifdef CONFIG_S390
118 #ifdef CONFIG_MATHEMU
119 extern int sysctl_ieee_emulation_warnings;
120 #endif
121 extern int sysctl_userprocess_debug;
122 extern int spin_retry;
123 #endif
124 
125 extern int sysctl_hz_timer;
126 
127 #ifdef CONFIG_BSD_PROCESS_ACCT
128 extern int acct_parm[];
129 #endif
130 
131 #ifdef CONFIG_IA64
132 extern int no_unaligned_warning;
133 #endif
134 
135 #ifdef CONFIG_RT_MUTEXES
136 extern int max_lock_depth;
137 #endif
138 
139 static int parse_table(int __user *, int, void __user *, size_t __user *, void __user *, size_t,
140 		       ctl_table *, void **);
141 static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
142 		  void __user *buffer, size_t *lenp, loff_t *ppos);
143 
144 static ctl_table root_table[];
145 static struct ctl_table_header root_table_header =
146 	{ root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
147 
148 static ctl_table kern_table[];
149 static ctl_table vm_table[];
150 static ctl_table fs_table[];
151 static ctl_table debug_table[];
152 static ctl_table dev_table[];
153 extern ctl_table random_table[];
154 #ifdef CONFIG_UNIX98_PTYS
155 extern ctl_table pty_table[];
156 #endif
157 #ifdef CONFIG_INOTIFY_USER
158 extern ctl_table inotify_table[];
159 #endif
160 
161 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
162 int sysctl_legacy_va_layout;
163 #endif
164 
165 /* /proc declarations: */
166 
167 #ifdef CONFIG_PROC_FS
168 
169 static ssize_t proc_readsys(struct file *, char __user *, size_t, loff_t *);
170 static ssize_t proc_writesys(struct file *, const char __user *, size_t, loff_t *);
171 static int proc_opensys(struct inode *, struct file *);
172 
173 struct file_operations proc_sys_file_operations = {
174 	.open		= proc_opensys,
175 	.read		= proc_readsys,
176 	.write		= proc_writesys,
177 };
178 
179 extern struct proc_dir_entry *proc_sys_root;
180 
181 static void register_proc_table(ctl_table *, struct proc_dir_entry *, void *);
182 static void unregister_proc_table(ctl_table *, struct proc_dir_entry *);
183 #endif
184 
185 /* The default sysctl tables: */
186 
187 static ctl_table root_table[] = {
188 	{
189 		.ctl_name	= CTL_KERN,
190 		.procname	= "kernel",
191 		.mode		= 0555,
192 		.child		= kern_table,
193 	},
194 	{
195 		.ctl_name	= CTL_VM,
196 		.procname	= "vm",
197 		.mode		= 0555,
198 		.child		= vm_table,
199 	},
200 #ifdef CONFIG_NET
201 	{
202 		.ctl_name	= CTL_NET,
203 		.procname	= "net",
204 		.mode		= 0555,
205 		.child		= net_table,
206 	},
207 #endif
208 	{
209 		.ctl_name	= CTL_FS,
210 		.procname	= "fs",
211 		.mode		= 0555,
212 		.child		= fs_table,
213 	},
214 	{
215 		.ctl_name	= CTL_DEBUG,
216 		.procname	= "debug",
217 		.mode		= 0555,
218 		.child		= debug_table,
219 	},
220 	{
221 		.ctl_name	= CTL_DEV,
222 		.procname	= "dev",
223 		.mode		= 0555,
224 		.child		= dev_table,
225 	},
226 
227 	{ .ctl_name = 0 }
228 };
229 
230 static ctl_table kern_table[] = {
231 	{
232 		.ctl_name	= KERN_OSTYPE,
233 		.procname	= "ostype",
234 		.data		= system_utsname.sysname,
235 		.maxlen		= sizeof(system_utsname.sysname),
236 		.mode		= 0444,
237 		.proc_handler	= &proc_doutsstring,
238 		.strategy	= &sysctl_string,
239 	},
240 	{
241 		.ctl_name	= KERN_OSRELEASE,
242 		.procname	= "osrelease",
243 		.data		= system_utsname.release,
244 		.maxlen		= sizeof(system_utsname.release),
245 		.mode		= 0444,
246 		.proc_handler	= &proc_doutsstring,
247 		.strategy	= &sysctl_string,
248 	},
249 	{
250 		.ctl_name	= KERN_VERSION,
251 		.procname	= "version",
252 		.data		= system_utsname.version,
253 		.maxlen		= sizeof(system_utsname.version),
254 		.mode		= 0444,
255 		.proc_handler	= &proc_doutsstring,
256 		.strategy	= &sysctl_string,
257 	},
258 	{
259 		.ctl_name	= KERN_NODENAME,
260 		.procname	= "hostname",
261 		.data		= system_utsname.nodename,
262 		.maxlen		= sizeof(system_utsname.nodename),
263 		.mode		= 0644,
264 		.proc_handler	= &proc_doutsstring,
265 		.strategy	= &sysctl_string,
266 	},
267 	{
268 		.ctl_name	= KERN_DOMAINNAME,
269 		.procname	= "domainname",
270 		.data		= system_utsname.domainname,
271 		.maxlen		= sizeof(system_utsname.domainname),
272 		.mode		= 0644,
273 		.proc_handler	= &proc_doutsstring,
274 		.strategy	= &sysctl_string,
275 	},
276 	{
277 		.ctl_name	= KERN_PANIC,
278 		.procname	= "panic",
279 		.data		= &panic_timeout,
280 		.maxlen		= sizeof(int),
281 		.mode		= 0644,
282 		.proc_handler	= &proc_dointvec,
283 	},
284 	{
285 		.ctl_name	= KERN_CORE_USES_PID,
286 		.procname	= "core_uses_pid",
287 		.data		= &core_uses_pid,
288 		.maxlen		= sizeof(int),
289 		.mode		= 0644,
290 		.proc_handler	= &proc_dointvec,
291 	},
292 	{
293 		.ctl_name	= KERN_CORE_PATTERN,
294 		.procname	= "core_pattern",
295 		.data		= core_pattern,
296 		.maxlen		= 64,
297 		.mode		= 0644,
298 		.proc_handler	= &proc_dostring,
299 		.strategy	= &sysctl_string,
300 	},
301 	{
302 		.ctl_name	= KERN_TAINTED,
303 		.procname	= "tainted",
304 		.data		= &tainted,
305 		.maxlen		= sizeof(int),
306 		.mode		= 0444,
307 		.proc_handler	= &proc_dointvec,
308 	},
309 	{
310 		.ctl_name	= KERN_CAP_BSET,
311 		.procname	= "cap-bound",
312 		.data		= &cap_bset,
313 		.maxlen		= sizeof(kernel_cap_t),
314 		.mode		= 0600,
315 		.proc_handler	= &proc_dointvec_bset,
316 	},
317 #ifdef CONFIG_BLK_DEV_INITRD
318 	{
319 		.ctl_name	= KERN_REALROOTDEV,
320 		.procname	= "real-root-dev",
321 		.data		= &real_root_dev,
322 		.maxlen		= sizeof(int),
323 		.mode		= 0644,
324 		.proc_handler	= &proc_dointvec,
325 	},
326 #endif
327 #ifdef __sparc__
328 	{
329 		.ctl_name	= KERN_SPARC_REBOOT,
330 		.procname	= "reboot-cmd",
331 		.data		= reboot_command,
332 		.maxlen		= 256,
333 		.mode		= 0644,
334 		.proc_handler	= &proc_dostring,
335 		.strategy	= &sysctl_string,
336 	},
337 	{
338 		.ctl_name	= KERN_SPARC_STOP_A,
339 		.procname	= "stop-a",
340 		.data		= &stop_a_enabled,
341 		.maxlen		= sizeof (int),
342 		.mode		= 0644,
343 		.proc_handler	= &proc_dointvec,
344 	},
345 	{
346 		.ctl_name	= KERN_SPARC_SCONS_PWROFF,
347 		.procname	= "scons-poweroff",
348 		.data		= &scons_pwroff,
349 		.maxlen		= sizeof (int),
350 		.mode		= 0644,
351 		.proc_handler	= &proc_dointvec,
352 	},
353 #endif
354 #ifdef __hppa__
355 	{
356 		.ctl_name	= KERN_HPPA_PWRSW,
357 		.procname	= "soft-power",
358 		.data		= &pwrsw_enabled,
359 		.maxlen		= sizeof (int),
360 	 	.mode		= 0644,
361 		.proc_handler	= &proc_dointvec,
362 	},
363 	{
364 		.ctl_name	= KERN_HPPA_UNALIGNED,
365 		.procname	= "unaligned-trap",
366 		.data		= &unaligned_enabled,
367 		.maxlen		= sizeof (int),
368 		.mode		= 0644,
369 		.proc_handler	= &proc_dointvec,
370 	},
371 #endif
372 	{
373 		.ctl_name	= KERN_CTLALTDEL,
374 		.procname	= "ctrl-alt-del",
375 		.data		= &C_A_D,
376 		.maxlen		= sizeof(int),
377 		.mode		= 0644,
378 		.proc_handler	= &proc_dointvec,
379 	},
380 	{
381 		.ctl_name	= KERN_PRINTK,
382 		.procname	= "printk",
383 		.data		= &console_loglevel,
384 		.maxlen		= 4*sizeof(int),
385 		.mode		= 0644,
386 		.proc_handler	= &proc_dointvec,
387 	},
388 #ifdef CONFIG_KMOD
389 	{
390 		.ctl_name	= KERN_MODPROBE,
391 		.procname	= "modprobe",
392 		.data		= &modprobe_path,
393 		.maxlen		= KMOD_PATH_LEN,
394 		.mode		= 0644,
395 		.proc_handler	= &proc_dostring,
396 		.strategy	= &sysctl_string,
397 	},
398 #endif
399 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
400 	{
401 		.ctl_name	= KERN_HOTPLUG,
402 		.procname	= "hotplug",
403 		.data		= &uevent_helper,
404 		.maxlen		= UEVENT_HELPER_PATH_LEN,
405 		.mode		= 0644,
406 		.proc_handler	= &proc_dostring,
407 		.strategy	= &sysctl_string,
408 	},
409 #endif
410 #ifdef CONFIG_CHR_DEV_SG
411 	{
412 		.ctl_name	= KERN_SG_BIG_BUFF,
413 		.procname	= "sg-big-buff",
414 		.data		= &sg_big_buff,
415 		.maxlen		= sizeof (int),
416 		.mode		= 0444,
417 		.proc_handler	= &proc_dointvec,
418 	},
419 #endif
420 #ifdef CONFIG_BSD_PROCESS_ACCT
421 	{
422 		.ctl_name	= KERN_ACCT,
423 		.procname	= "acct",
424 		.data		= &acct_parm,
425 		.maxlen		= 3*sizeof(int),
426 		.mode		= 0644,
427 		.proc_handler	= &proc_dointvec,
428 	},
429 #endif
430 #ifdef CONFIG_SYSVIPC
431 	{
432 		.ctl_name	= KERN_SHMMAX,
433 		.procname	= "shmmax",
434 		.data		= &shm_ctlmax,
435 		.maxlen		= sizeof (size_t),
436 		.mode		= 0644,
437 		.proc_handler	= &proc_doulongvec_minmax,
438 	},
439 	{
440 		.ctl_name	= KERN_SHMALL,
441 		.procname	= "shmall",
442 		.data		= &shm_ctlall,
443 		.maxlen		= sizeof (size_t),
444 		.mode		= 0644,
445 		.proc_handler	= &proc_doulongvec_minmax,
446 	},
447 	{
448 		.ctl_name	= KERN_SHMMNI,
449 		.procname	= "shmmni",
450 		.data		= &shm_ctlmni,
451 		.maxlen		= sizeof (int),
452 		.mode		= 0644,
453 		.proc_handler	= &proc_dointvec,
454 	},
455 	{
456 		.ctl_name	= KERN_MSGMAX,
457 		.procname	= "msgmax",
458 		.data		= &msg_ctlmax,
459 		.maxlen		= sizeof (int),
460 		.mode		= 0644,
461 		.proc_handler	= &proc_dointvec,
462 	},
463 	{
464 		.ctl_name	= KERN_MSGMNI,
465 		.procname	= "msgmni",
466 		.data		= &msg_ctlmni,
467 		.maxlen		= sizeof (int),
468 		.mode		= 0644,
469 		.proc_handler	= &proc_dointvec,
470 	},
471 	{
472 		.ctl_name	= KERN_MSGMNB,
473 		.procname	=  "msgmnb",
474 		.data		= &msg_ctlmnb,
475 		.maxlen		= sizeof (int),
476 		.mode		= 0644,
477 		.proc_handler	= &proc_dointvec,
478 	},
479 	{
480 		.ctl_name	= KERN_SEM,
481 		.procname	= "sem",
482 		.data		= &sem_ctls,
483 		.maxlen		= 4*sizeof (int),
484 		.mode		= 0644,
485 		.proc_handler	= &proc_dointvec,
486 	},
487 #endif
488 #ifdef CONFIG_MAGIC_SYSRQ
489 	{
490 		.ctl_name	= KERN_SYSRQ,
491 		.procname	= "sysrq",
492 		.data		= &sysrq_enabled,
493 		.maxlen		= sizeof (int),
494 		.mode		= 0644,
495 		.proc_handler	= &proc_dointvec,
496 	},
497 #endif
498 	{
499 		.ctl_name	= KERN_CADPID,
500 		.procname	= "cad_pid",
501 		.data		= &cad_pid,
502 		.maxlen		= sizeof (int),
503 		.mode		= 0600,
504 		.proc_handler	= &proc_dointvec,
505 	},
506 	{
507 		.ctl_name	= KERN_MAX_THREADS,
508 		.procname	= "threads-max",
509 		.data		= &max_threads,
510 		.maxlen		= sizeof(int),
511 		.mode		= 0644,
512 		.proc_handler	= &proc_dointvec,
513 	},
514 	{
515 		.ctl_name	= KERN_RANDOM,
516 		.procname	= "random",
517 		.mode		= 0555,
518 		.child		= random_table,
519 	},
520 #ifdef CONFIG_UNIX98_PTYS
521 	{
522 		.ctl_name	= KERN_PTY,
523 		.procname	= "pty",
524 		.mode		= 0555,
525 		.child		= pty_table,
526 	},
527 #endif
528 	{
529 		.ctl_name	= KERN_OVERFLOWUID,
530 		.procname	= "overflowuid",
531 		.data		= &overflowuid,
532 		.maxlen		= sizeof(int),
533 		.mode		= 0644,
534 		.proc_handler	= &proc_dointvec_minmax,
535 		.strategy	= &sysctl_intvec,
536 		.extra1		= &minolduid,
537 		.extra2		= &maxolduid,
538 	},
539 	{
540 		.ctl_name	= KERN_OVERFLOWGID,
541 		.procname	= "overflowgid",
542 		.data		= &overflowgid,
543 		.maxlen		= sizeof(int),
544 		.mode		= 0644,
545 		.proc_handler	= &proc_dointvec_minmax,
546 		.strategy	= &sysctl_intvec,
547 		.extra1		= &minolduid,
548 		.extra2		= &maxolduid,
549 	},
550 #ifdef CONFIG_S390
551 #ifdef CONFIG_MATHEMU
552 	{
553 		.ctl_name	= KERN_IEEE_EMULATION_WARNINGS,
554 		.procname	= "ieee_emulation_warnings",
555 		.data		= &sysctl_ieee_emulation_warnings,
556 		.maxlen		= sizeof(int),
557 		.mode		= 0644,
558 		.proc_handler	= &proc_dointvec,
559 	},
560 #endif
561 #ifdef CONFIG_NO_IDLE_HZ
562 	{
563 		.ctl_name       = KERN_HZ_TIMER,
564 		.procname       = "hz_timer",
565 		.data           = &sysctl_hz_timer,
566 		.maxlen         = sizeof(int),
567 		.mode           = 0644,
568 		.proc_handler   = &proc_dointvec,
569 	},
570 #endif
571 	{
572 		.ctl_name	= KERN_S390_USER_DEBUG_LOGGING,
573 		.procname	= "userprocess_debug",
574 		.data		= &sysctl_userprocess_debug,
575 		.maxlen		= sizeof(int),
576 		.mode		= 0644,
577 		.proc_handler	= &proc_dointvec,
578 	},
579 #endif
580 	{
581 		.ctl_name	= KERN_PIDMAX,
582 		.procname	= "pid_max",
583 		.data		= &pid_max,
584 		.maxlen		= sizeof (int),
585 		.mode		= 0644,
586 		.proc_handler	= &proc_dointvec_minmax,
587 		.strategy	= sysctl_intvec,
588 		.extra1		= &pid_max_min,
589 		.extra2		= &pid_max_max,
590 	},
591 	{
592 		.ctl_name	= KERN_PANIC_ON_OOPS,
593 		.procname	= "panic_on_oops",
594 		.data		= &panic_on_oops,
595 		.maxlen		= sizeof(int),
596 		.mode		= 0644,
597 		.proc_handler	= &proc_dointvec,
598 	},
599 	{
600 		.ctl_name	= KERN_PRINTK_RATELIMIT,
601 		.procname	= "printk_ratelimit",
602 		.data		= &printk_ratelimit_jiffies,
603 		.maxlen		= sizeof(int),
604 		.mode		= 0644,
605 		.proc_handler	= &proc_dointvec_jiffies,
606 		.strategy	= &sysctl_jiffies,
607 	},
608 	{
609 		.ctl_name	= KERN_PRINTK_RATELIMIT_BURST,
610 		.procname	= "printk_ratelimit_burst",
611 		.data		= &printk_ratelimit_burst,
612 		.maxlen		= sizeof(int),
613 		.mode		= 0644,
614 		.proc_handler	= &proc_dointvec,
615 	},
616 	{
617 		.ctl_name	= KERN_NGROUPS_MAX,
618 		.procname	= "ngroups_max",
619 		.data		= &ngroups_max,
620 		.maxlen		= sizeof (int),
621 		.mode		= 0444,
622 		.proc_handler	= &proc_dointvec,
623 	},
624 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
625 	{
626 		.ctl_name       = KERN_UNKNOWN_NMI_PANIC,
627 		.procname       = "unknown_nmi_panic",
628 		.data           = &unknown_nmi_panic,
629 		.maxlen         = sizeof (int),
630 		.mode           = 0644,
631 		.proc_handler   = &proc_unknown_nmi_panic,
632 	},
633 #endif
634 #if defined(CONFIG_X86)
635 	{
636 		.ctl_name	= KERN_BOOTLOADER_TYPE,
637 		.procname	= "bootloader_type",
638 		.data		= &bootloader_type,
639 		.maxlen		= sizeof (int),
640 		.mode		= 0444,
641 		.proc_handler	= &proc_dointvec,
642 	},
643 #endif
644 #if defined(CONFIG_MMU)
645 	{
646 		.ctl_name	= KERN_RANDOMIZE,
647 		.procname	= "randomize_va_space",
648 		.data		= &randomize_va_space,
649 		.maxlen		= sizeof(int),
650 		.mode		= 0644,
651 		.proc_handler	= &proc_dointvec,
652 	},
653 #endif
654 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
655 	{
656 		.ctl_name	= KERN_SPIN_RETRY,
657 		.procname	= "spin_retry",
658 		.data		= &spin_retry,
659 		.maxlen		= sizeof (int),
660 		.mode		= 0644,
661 		.proc_handler	= &proc_dointvec,
662 	},
663 #endif
664 #ifdef CONFIG_ACPI_SLEEP
665 	{
666 		.ctl_name	= KERN_ACPI_VIDEO_FLAGS,
667 		.procname	= "acpi_video_flags",
668 		.data		= &acpi_video_flags,
669 		.maxlen		= sizeof (unsigned long),
670 		.mode		= 0644,
671 		.proc_handler	= &proc_doulongvec_minmax,
672 	},
673 #endif
674 #ifdef CONFIG_IA64
675 	{
676 		.ctl_name	= KERN_IA64_UNALIGNED,
677 		.procname	= "ignore-unaligned-usertrap",
678 		.data		= &no_unaligned_warning,
679 		.maxlen		= sizeof (int),
680 	 	.mode		= 0644,
681 		.proc_handler	= &proc_dointvec,
682 	},
683 #endif
684 #ifdef CONFIG_COMPAT
685 	{
686 		.ctl_name	= KERN_COMPAT_LOG,
687 		.procname	= "compat-log",
688 		.data		= &compat_log,
689 		.maxlen		= sizeof (int),
690 	 	.mode		= 0644,
691 		.proc_handler	= &proc_dointvec,
692 	},
693 #endif
694 #ifdef CONFIG_RT_MUTEXES
695 	{
696 		.ctl_name	= KERN_MAX_LOCK_DEPTH,
697 		.procname	= "max_lock_depth",
698 		.data		= &max_lock_depth,
699 		.maxlen		= sizeof(int),
700 		.mode		= 0644,
701 		.proc_handler	= &proc_dointvec,
702 	},
703 #endif
704 
705 	{ .ctl_name = 0 }
706 };
707 
708 /* Constants for minimum and maximum testing in vm_table.
709    We use these as one-element integer vectors. */
710 static int zero;
711 static int one_hundred = 100;
712 
713 
714 static ctl_table vm_table[] = {
715 	{
716 		.ctl_name	= VM_OVERCOMMIT_MEMORY,
717 		.procname	= "overcommit_memory",
718 		.data		= &sysctl_overcommit_memory,
719 		.maxlen		= sizeof(sysctl_overcommit_memory),
720 		.mode		= 0644,
721 		.proc_handler	= &proc_dointvec,
722 	},
723 	{
724 		.ctl_name	= VM_PANIC_ON_OOM,
725 		.procname	= "panic_on_oom",
726 		.data		= &sysctl_panic_on_oom,
727 		.maxlen		= sizeof(sysctl_panic_on_oom),
728 		.mode		= 0644,
729 		.proc_handler	= &proc_dointvec,
730 	},
731 	{
732 		.ctl_name	= VM_OVERCOMMIT_RATIO,
733 		.procname	= "overcommit_ratio",
734 		.data		= &sysctl_overcommit_ratio,
735 		.maxlen		= sizeof(sysctl_overcommit_ratio),
736 		.mode		= 0644,
737 		.proc_handler	= &proc_dointvec,
738 	},
739 	{
740 		.ctl_name	= VM_PAGE_CLUSTER,
741 		.procname	= "page-cluster",
742 		.data		= &page_cluster,
743 		.maxlen		= sizeof(int),
744 		.mode		= 0644,
745 		.proc_handler	= &proc_dointvec,
746 	},
747 	{
748 		.ctl_name	= VM_DIRTY_BACKGROUND,
749 		.procname	= "dirty_background_ratio",
750 		.data		= &dirty_background_ratio,
751 		.maxlen		= sizeof(dirty_background_ratio),
752 		.mode		= 0644,
753 		.proc_handler	= &proc_dointvec_minmax,
754 		.strategy	= &sysctl_intvec,
755 		.extra1		= &zero,
756 		.extra2		= &one_hundred,
757 	},
758 	{
759 		.ctl_name	= VM_DIRTY_RATIO,
760 		.procname	= "dirty_ratio",
761 		.data		= &vm_dirty_ratio,
762 		.maxlen		= sizeof(vm_dirty_ratio),
763 		.mode		= 0644,
764 		.proc_handler	= &proc_dointvec_minmax,
765 		.strategy	= &sysctl_intvec,
766 		.extra1		= &zero,
767 		.extra2		= &one_hundred,
768 	},
769 	{
770 		.ctl_name	= VM_DIRTY_WB_CS,
771 		.procname	= "dirty_writeback_centisecs",
772 		.data		= &dirty_writeback_interval,
773 		.maxlen		= sizeof(dirty_writeback_interval),
774 		.mode		= 0644,
775 		.proc_handler	= &dirty_writeback_centisecs_handler,
776 	},
777 	{
778 		.ctl_name	= VM_DIRTY_EXPIRE_CS,
779 		.procname	= "dirty_expire_centisecs",
780 		.data		= &dirty_expire_interval,
781 		.maxlen		= sizeof(dirty_expire_interval),
782 		.mode		= 0644,
783 		.proc_handler	= &proc_dointvec_userhz_jiffies,
784 	},
785 	{
786 		.ctl_name	= VM_NR_PDFLUSH_THREADS,
787 		.procname	= "nr_pdflush_threads",
788 		.data		= &nr_pdflush_threads,
789 		.maxlen		= sizeof nr_pdflush_threads,
790 		.mode		= 0444 /* read-only*/,
791 		.proc_handler	= &proc_dointvec,
792 	},
793 	{
794 		.ctl_name	= VM_SWAPPINESS,
795 		.procname	= "swappiness",
796 		.data		= &vm_swappiness,
797 		.maxlen		= sizeof(vm_swappiness),
798 		.mode		= 0644,
799 		.proc_handler	= &proc_dointvec_minmax,
800 		.strategy	= &sysctl_intvec,
801 		.extra1		= &zero,
802 		.extra2		= &one_hundred,
803 	},
804 #ifdef CONFIG_HUGETLB_PAGE
805 	 {
806 		.ctl_name	= VM_HUGETLB_PAGES,
807 		.procname	= "nr_hugepages",
808 		.data		= &max_huge_pages,
809 		.maxlen		= sizeof(unsigned long),
810 		.mode		= 0644,
811 		.proc_handler	= &hugetlb_sysctl_handler,
812 		.extra1		= (void *)&hugetlb_zero,
813 		.extra2		= (void *)&hugetlb_infinity,
814 	 },
815 	 {
816 		.ctl_name	= VM_HUGETLB_GROUP,
817 		.procname	= "hugetlb_shm_group",
818 		.data		= &sysctl_hugetlb_shm_group,
819 		.maxlen		= sizeof(gid_t),
820 		.mode		= 0644,
821 		.proc_handler	= &proc_dointvec,
822 	 },
823 #endif
824 	{
825 		.ctl_name	= VM_LOWMEM_RESERVE_RATIO,
826 		.procname	= "lowmem_reserve_ratio",
827 		.data		= &sysctl_lowmem_reserve_ratio,
828 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
829 		.mode		= 0644,
830 		.proc_handler	= &lowmem_reserve_ratio_sysctl_handler,
831 		.strategy	= &sysctl_intvec,
832 	},
833 	{
834 		.ctl_name	= VM_DROP_PAGECACHE,
835 		.procname	= "drop_caches",
836 		.data		= &sysctl_drop_caches,
837 		.maxlen		= sizeof(int),
838 		.mode		= 0644,
839 		.proc_handler	= drop_caches_sysctl_handler,
840 		.strategy	= &sysctl_intvec,
841 	},
842 	{
843 		.ctl_name	= VM_MIN_FREE_KBYTES,
844 		.procname	= "min_free_kbytes",
845 		.data		= &min_free_kbytes,
846 		.maxlen		= sizeof(min_free_kbytes),
847 		.mode		= 0644,
848 		.proc_handler	= &min_free_kbytes_sysctl_handler,
849 		.strategy	= &sysctl_intvec,
850 		.extra1		= &zero,
851 	},
852 	{
853 		.ctl_name	= VM_PERCPU_PAGELIST_FRACTION,
854 		.procname	= "percpu_pagelist_fraction",
855 		.data		= &percpu_pagelist_fraction,
856 		.maxlen		= sizeof(percpu_pagelist_fraction),
857 		.mode		= 0644,
858 		.proc_handler	= &percpu_pagelist_fraction_sysctl_handler,
859 		.strategy	= &sysctl_intvec,
860 		.extra1		= &min_percpu_pagelist_fract,
861 	},
862 #ifdef CONFIG_MMU
863 	{
864 		.ctl_name	= VM_MAX_MAP_COUNT,
865 		.procname	= "max_map_count",
866 		.data		= &sysctl_max_map_count,
867 		.maxlen		= sizeof(sysctl_max_map_count),
868 		.mode		= 0644,
869 		.proc_handler	= &proc_dointvec
870 	},
871 #endif
872 	{
873 		.ctl_name	= VM_LAPTOP_MODE,
874 		.procname	= "laptop_mode",
875 		.data		= &laptop_mode,
876 		.maxlen		= sizeof(laptop_mode),
877 		.mode		= 0644,
878 		.proc_handler	= &proc_dointvec_jiffies,
879 		.strategy	= &sysctl_jiffies,
880 	},
881 	{
882 		.ctl_name	= VM_BLOCK_DUMP,
883 		.procname	= "block_dump",
884 		.data		= &block_dump,
885 		.maxlen		= sizeof(block_dump),
886 		.mode		= 0644,
887 		.proc_handler	= &proc_dointvec,
888 		.strategy	= &sysctl_intvec,
889 		.extra1		= &zero,
890 	},
891 	{
892 		.ctl_name	= VM_VFS_CACHE_PRESSURE,
893 		.procname	= "vfs_cache_pressure",
894 		.data		= &sysctl_vfs_cache_pressure,
895 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
896 		.mode		= 0644,
897 		.proc_handler	= &proc_dointvec,
898 		.strategy	= &sysctl_intvec,
899 		.extra1		= &zero,
900 	},
901 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
902 	{
903 		.ctl_name	= VM_LEGACY_VA_LAYOUT,
904 		.procname	= "legacy_va_layout",
905 		.data		= &sysctl_legacy_va_layout,
906 		.maxlen		= sizeof(sysctl_legacy_va_layout),
907 		.mode		= 0644,
908 		.proc_handler	= &proc_dointvec,
909 		.strategy	= &sysctl_intvec,
910 		.extra1		= &zero,
911 	},
912 #endif
913 #ifdef CONFIG_SWAP
914 	{
915 		.ctl_name	= VM_SWAP_TOKEN_TIMEOUT,
916 		.procname	= "swap_token_timeout",
917 		.data		= &swap_token_default_timeout,
918 		.maxlen		= sizeof(swap_token_default_timeout),
919 		.mode		= 0644,
920 		.proc_handler	= &proc_dointvec_jiffies,
921 		.strategy	= &sysctl_jiffies,
922 	},
923 #endif
924 #ifdef CONFIG_NUMA
925 	{
926 		.ctl_name	= VM_ZONE_RECLAIM_MODE,
927 		.procname	= "zone_reclaim_mode",
928 		.data		= &zone_reclaim_mode,
929 		.maxlen		= sizeof(zone_reclaim_mode),
930 		.mode		= 0644,
931 		.proc_handler	= &proc_dointvec,
932 		.strategy	= &sysctl_intvec,
933 		.extra1		= &zero,
934 	},
935 	{
936 		.ctl_name	= VM_MIN_UNMAPPED,
937 		.procname	= "min_unmapped_ratio",
938 		.data		= &sysctl_min_unmapped_ratio,
939 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
940 		.mode		= 0644,
941 		.proc_handler	= &sysctl_min_unmapped_ratio_sysctl_handler,
942 		.strategy	= &sysctl_intvec,
943 		.extra1		= &zero,
944 		.extra2		= &one_hundred,
945 	},
946 #endif
947 #ifdef CONFIG_X86_32
948 	{
949 		.ctl_name	= VM_VDSO_ENABLED,
950 		.procname	= "vdso_enabled",
951 		.data		= &vdso_enabled,
952 		.maxlen		= sizeof(vdso_enabled),
953 		.mode		= 0644,
954 		.proc_handler	= &proc_dointvec,
955 		.strategy	= &sysctl_intvec,
956 		.extra1		= &zero,
957 	},
958 #endif
959 	{ .ctl_name = 0 }
960 };
961 
962 static ctl_table fs_table[] = {
963 	{
964 		.ctl_name	= FS_NRINODE,
965 		.procname	= "inode-nr",
966 		.data		= &inodes_stat,
967 		.maxlen		= 2*sizeof(int),
968 		.mode		= 0444,
969 		.proc_handler	= &proc_dointvec,
970 	},
971 	{
972 		.ctl_name	= FS_STATINODE,
973 		.procname	= "inode-state",
974 		.data		= &inodes_stat,
975 		.maxlen		= 7*sizeof(int),
976 		.mode		= 0444,
977 		.proc_handler	= &proc_dointvec,
978 	},
979 	{
980 		.ctl_name	= FS_NRFILE,
981 		.procname	= "file-nr",
982 		.data		= &files_stat,
983 		.maxlen		= 3*sizeof(int),
984 		.mode		= 0444,
985 		.proc_handler	= &proc_nr_files,
986 	},
987 	{
988 		.ctl_name	= FS_MAXFILE,
989 		.procname	= "file-max",
990 		.data		= &files_stat.max_files,
991 		.maxlen		= sizeof(int),
992 		.mode		= 0644,
993 		.proc_handler	= &proc_dointvec,
994 	},
995 	{
996 		.ctl_name	= FS_DENTRY,
997 		.procname	= "dentry-state",
998 		.data		= &dentry_stat,
999 		.maxlen		= 6*sizeof(int),
1000 		.mode		= 0444,
1001 		.proc_handler	= &proc_dointvec,
1002 	},
1003 	{
1004 		.ctl_name	= FS_OVERFLOWUID,
1005 		.procname	= "overflowuid",
1006 		.data		= &fs_overflowuid,
1007 		.maxlen		= sizeof(int),
1008 		.mode		= 0644,
1009 		.proc_handler	= &proc_dointvec_minmax,
1010 		.strategy	= &sysctl_intvec,
1011 		.extra1		= &minolduid,
1012 		.extra2		= &maxolduid,
1013 	},
1014 	{
1015 		.ctl_name	= FS_OVERFLOWGID,
1016 		.procname	= "overflowgid",
1017 		.data		= &fs_overflowgid,
1018 		.maxlen		= sizeof(int),
1019 		.mode		= 0644,
1020 		.proc_handler	= &proc_dointvec_minmax,
1021 		.strategy	= &sysctl_intvec,
1022 		.extra1		= &minolduid,
1023 		.extra2		= &maxolduid,
1024 	},
1025 	{
1026 		.ctl_name	= FS_LEASES,
1027 		.procname	= "leases-enable",
1028 		.data		= &leases_enable,
1029 		.maxlen		= sizeof(int),
1030 		.mode		= 0644,
1031 		.proc_handler	= &proc_dointvec,
1032 	},
1033 #ifdef CONFIG_DNOTIFY
1034 	{
1035 		.ctl_name	= FS_DIR_NOTIFY,
1036 		.procname	= "dir-notify-enable",
1037 		.data		= &dir_notify_enable,
1038 		.maxlen		= sizeof(int),
1039 		.mode		= 0644,
1040 		.proc_handler	= &proc_dointvec,
1041 	},
1042 #endif
1043 #ifdef CONFIG_MMU
1044 	{
1045 		.ctl_name	= FS_LEASE_TIME,
1046 		.procname	= "lease-break-time",
1047 		.data		= &lease_break_time,
1048 		.maxlen		= sizeof(int),
1049 		.mode		= 0644,
1050 		.proc_handler	= &proc_dointvec,
1051 	},
1052 	{
1053 		.ctl_name	= FS_AIO_NR,
1054 		.procname	= "aio-nr",
1055 		.data		= &aio_nr,
1056 		.maxlen		= sizeof(aio_nr),
1057 		.mode		= 0444,
1058 		.proc_handler	= &proc_doulongvec_minmax,
1059 	},
1060 	{
1061 		.ctl_name	= FS_AIO_MAX_NR,
1062 		.procname	= "aio-max-nr",
1063 		.data		= &aio_max_nr,
1064 		.maxlen		= sizeof(aio_max_nr),
1065 		.mode		= 0644,
1066 		.proc_handler	= &proc_doulongvec_minmax,
1067 	},
1068 #ifdef CONFIG_INOTIFY_USER
1069 	{
1070 		.ctl_name	= FS_INOTIFY,
1071 		.procname	= "inotify",
1072 		.mode		= 0555,
1073 		.child		= inotify_table,
1074 	},
1075 #endif
1076 #endif
1077 	{
1078 		.ctl_name	= KERN_SETUID_DUMPABLE,
1079 		.procname	= "suid_dumpable",
1080 		.data		= &suid_dumpable,
1081 		.maxlen		= sizeof(int),
1082 		.mode		= 0644,
1083 		.proc_handler	= &proc_dointvec,
1084 	},
1085 	{ .ctl_name = 0 }
1086 };
1087 
1088 static ctl_table debug_table[] = {
1089 	{ .ctl_name = 0 }
1090 };
1091 
1092 static ctl_table dev_table[] = {
1093 	{ .ctl_name = 0 }
1094 };
1095 
1096 extern void init_irq_proc (void);
1097 
1098 static DEFINE_SPINLOCK(sysctl_lock);
1099 
1100 /* called under sysctl_lock */
1101 static int use_table(struct ctl_table_header *p)
1102 {
1103 	if (unlikely(p->unregistering))
1104 		return 0;
1105 	p->used++;
1106 	return 1;
1107 }
1108 
1109 /* called under sysctl_lock */
1110 static void unuse_table(struct ctl_table_header *p)
1111 {
1112 	if (!--p->used)
1113 		if (unlikely(p->unregistering))
1114 			complete(p->unregistering);
1115 }
1116 
1117 /* called under sysctl_lock, will reacquire if has to wait */
1118 static void start_unregistering(struct ctl_table_header *p)
1119 {
1120 	/*
1121 	 * if p->used is 0, nobody will ever touch that entry again;
1122 	 * we'll eliminate all paths to it before dropping sysctl_lock
1123 	 */
1124 	if (unlikely(p->used)) {
1125 		struct completion wait;
1126 		init_completion(&wait);
1127 		p->unregistering = &wait;
1128 		spin_unlock(&sysctl_lock);
1129 		wait_for_completion(&wait);
1130 		spin_lock(&sysctl_lock);
1131 	}
1132 	/*
1133 	 * do not remove from the list until nobody holds it; walking the
1134 	 * list in do_sysctl() relies on that.
1135 	 */
1136 	list_del_init(&p->ctl_entry);
1137 }
1138 
1139 void __init sysctl_init(void)
1140 {
1141 #ifdef CONFIG_PROC_FS
1142 	register_proc_table(root_table, proc_sys_root, &root_table_header);
1143 	init_irq_proc();
1144 #endif
1145 }
1146 
1147 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1148 	       void __user *newval, size_t newlen)
1149 {
1150 	struct list_head *tmp;
1151 	int error = -ENOTDIR;
1152 
1153 	if (nlen <= 0 || nlen >= CTL_MAXNAME)
1154 		return -ENOTDIR;
1155 	if (oldval) {
1156 		int old_len;
1157 		if (!oldlenp || get_user(old_len, oldlenp))
1158 			return -EFAULT;
1159 	}
1160 	spin_lock(&sysctl_lock);
1161 	tmp = &root_table_header.ctl_entry;
1162 	do {
1163 		struct ctl_table_header *head =
1164 			list_entry(tmp, struct ctl_table_header, ctl_entry);
1165 		void *context = NULL;
1166 
1167 		if (!use_table(head))
1168 			continue;
1169 
1170 		spin_unlock(&sysctl_lock);
1171 
1172 		error = parse_table(name, nlen, oldval, oldlenp,
1173 					newval, newlen, head->ctl_table,
1174 					&context);
1175 		kfree(context);
1176 
1177 		spin_lock(&sysctl_lock);
1178 		unuse_table(head);
1179 		if (error != -ENOTDIR)
1180 			break;
1181 	} while ((tmp = tmp->next) != &root_table_header.ctl_entry);
1182 	spin_unlock(&sysctl_lock);
1183 	return error;
1184 }
1185 
1186 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1187 {
1188 	struct __sysctl_args tmp;
1189 	int error;
1190 
1191 	if (copy_from_user(&tmp, args, sizeof(tmp)))
1192 		return -EFAULT;
1193 
1194 	lock_kernel();
1195 	error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1196 			  tmp.newval, tmp.newlen);
1197 	unlock_kernel();
1198 	return error;
1199 }
1200 
1201 /*
1202  * ctl_perm does NOT grant the superuser all rights automatically, because
1203  * some sysctl variables are readonly even to root.
1204  */
1205 
1206 static int test_perm(int mode, int op)
1207 {
1208 	if (!current->euid)
1209 		mode >>= 6;
1210 	else if (in_egroup_p(0))
1211 		mode >>= 3;
1212 	if ((mode & op & 0007) == op)
1213 		return 0;
1214 	return -EACCES;
1215 }
1216 
1217 static inline int ctl_perm(ctl_table *table, int op)
1218 {
1219 	int error;
1220 	error = security_sysctl(table, op);
1221 	if (error)
1222 		return error;
1223 	return test_perm(table->mode, op);
1224 }
1225 
1226 static int parse_table(int __user *name, int nlen,
1227 		       void __user *oldval, size_t __user *oldlenp,
1228 		       void __user *newval, size_t newlen,
1229 		       ctl_table *table, void **context)
1230 {
1231 	int n;
1232 repeat:
1233 	if (!nlen)
1234 		return -ENOTDIR;
1235 	if (get_user(n, name))
1236 		return -EFAULT;
1237 	for ( ; table->ctl_name; table++) {
1238 		if (n == table->ctl_name || table->ctl_name == CTL_ANY) {
1239 			int error;
1240 			if (table->child) {
1241 				if (ctl_perm(table, 001))
1242 					return -EPERM;
1243 				if (table->strategy) {
1244 					error = table->strategy(
1245 						table, name, nlen,
1246 						oldval, oldlenp,
1247 						newval, newlen, context);
1248 					if (error)
1249 						return error;
1250 				}
1251 				name++;
1252 				nlen--;
1253 				table = table->child;
1254 				goto repeat;
1255 			}
1256 			error = do_sysctl_strategy(table, name, nlen,
1257 						   oldval, oldlenp,
1258 						   newval, newlen, context);
1259 			return error;
1260 		}
1261 	}
1262 	return -ENOTDIR;
1263 }
1264 
1265 /* Perform the actual read/write of a sysctl table entry. */
1266 int do_sysctl_strategy (ctl_table *table,
1267 			int __user *name, int nlen,
1268 			void __user *oldval, size_t __user *oldlenp,
1269 			void __user *newval, size_t newlen, void **context)
1270 {
1271 	int op = 0, rc;
1272 	size_t len;
1273 
1274 	if (oldval)
1275 		op |= 004;
1276 	if (newval)
1277 		op |= 002;
1278 	if (ctl_perm(table, op))
1279 		return -EPERM;
1280 
1281 	if (table->strategy) {
1282 		rc = table->strategy(table, name, nlen, oldval, oldlenp,
1283 				     newval, newlen, context);
1284 		if (rc < 0)
1285 			return rc;
1286 		if (rc > 0)
1287 			return 0;
1288 	}
1289 
1290 	/* If there is no strategy routine, or if the strategy returns
1291 	 * zero, proceed with automatic r/w */
1292 	if (table->data && table->maxlen) {
1293 		if (oldval && oldlenp) {
1294 			if (get_user(len, oldlenp))
1295 				return -EFAULT;
1296 			if (len) {
1297 				if (len > table->maxlen)
1298 					len = table->maxlen;
1299 				if(copy_to_user(oldval, table->data, len))
1300 					return -EFAULT;
1301 				if(put_user(len, oldlenp))
1302 					return -EFAULT;
1303 			}
1304 		}
1305 		if (newval && newlen) {
1306 			len = newlen;
1307 			if (len > table->maxlen)
1308 				len = table->maxlen;
1309 			if(copy_from_user(table->data, newval, len))
1310 				return -EFAULT;
1311 		}
1312 	}
1313 	return 0;
1314 }
1315 
1316 /**
1317  * register_sysctl_table - register a sysctl hierarchy
1318  * @table: the top-level table structure
1319  * @insert_at_head: whether the entry should be inserted in front or at the end
1320  *
1321  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1322  * array. An entry with a ctl_name of 0 terminates the table.
1323  *
1324  * The members of the &ctl_table structure are used as follows:
1325  *
1326  * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1327  *            must be unique within that level of sysctl
1328  *
1329  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1330  *            enter a sysctl file
1331  *
1332  * data - a pointer to data for use by proc_handler
1333  *
1334  * maxlen - the maximum size in bytes of the data
1335  *
1336  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1337  *
1338  * child - a pointer to the child sysctl table if this entry is a directory, or
1339  *         %NULL.
1340  *
1341  * proc_handler - the text handler routine (described below)
1342  *
1343  * strategy - the strategy routine (described below)
1344  *
1345  * de - for internal use by the sysctl routines
1346  *
1347  * extra1, extra2 - extra pointers usable by the proc handler routines
1348  *
1349  * Leaf nodes in the sysctl tree will be represented by a single file
1350  * under /proc; non-leaf nodes will be represented by directories.
1351  *
1352  * sysctl(2) can automatically manage read and write requests through
1353  * the sysctl table.  The data and maxlen fields of the ctl_table
1354  * struct enable minimal validation of the values being written to be
1355  * performed, and the mode field allows minimal authentication.
1356  *
1357  * More sophisticated management can be enabled by the provision of a
1358  * strategy routine with the table entry.  This will be called before
1359  * any automatic read or write of the data is performed.
1360  *
1361  * The strategy routine may return
1362  *
1363  * < 0 - Error occurred (error is passed to user process)
1364  *
1365  * 0   - OK - proceed with automatic read or write.
1366  *
1367  * > 0 - OK - read or write has been done by the strategy routine, so
1368  *       return immediately.
1369  *
1370  * There must be a proc_handler routine for any terminal nodes
1371  * mirrored under /proc/sys (non-terminals are handled by a built-in
1372  * directory handler).  Several default handlers are available to
1373  * cover common cases -
1374  *
1375  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1376  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1377  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1378  *
1379  * It is the handler's job to read the input buffer from user memory
1380  * and process it. The handler should return 0 on success.
1381  *
1382  * This routine returns %NULL on a failure to register, and a pointer
1383  * to the table header on success.
1384  */
1385 struct ctl_table_header *register_sysctl_table(ctl_table * table,
1386 					       int insert_at_head)
1387 {
1388 	struct ctl_table_header *tmp;
1389 	tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1390 	if (!tmp)
1391 		return NULL;
1392 	tmp->ctl_table = table;
1393 	INIT_LIST_HEAD(&tmp->ctl_entry);
1394 	tmp->used = 0;
1395 	tmp->unregistering = NULL;
1396 	spin_lock(&sysctl_lock);
1397 	if (insert_at_head)
1398 		list_add(&tmp->ctl_entry, &root_table_header.ctl_entry);
1399 	else
1400 		list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
1401 	spin_unlock(&sysctl_lock);
1402 #ifdef CONFIG_PROC_FS
1403 	register_proc_table(table, proc_sys_root, tmp);
1404 #endif
1405 	return tmp;
1406 }
1407 
1408 /**
1409  * unregister_sysctl_table - unregister a sysctl table hierarchy
1410  * @header: the header returned from register_sysctl_table
1411  *
1412  * Unregisters the sysctl table and all children. proc entries may not
1413  * actually be removed until they are no longer used by anyone.
1414  */
1415 void unregister_sysctl_table(struct ctl_table_header * header)
1416 {
1417 	might_sleep();
1418 	spin_lock(&sysctl_lock);
1419 	start_unregistering(header);
1420 #ifdef CONFIG_PROC_FS
1421 	unregister_proc_table(header->ctl_table, proc_sys_root);
1422 #endif
1423 	spin_unlock(&sysctl_lock);
1424 	kfree(header);
1425 }
1426 
1427 /*
1428  * /proc/sys support
1429  */
1430 
1431 #ifdef CONFIG_PROC_FS
1432 
1433 /* Scan the sysctl entries in table and add them all into /proc */
1434 static void register_proc_table(ctl_table * table, struct proc_dir_entry *root, void *set)
1435 {
1436 	struct proc_dir_entry *de;
1437 	int len;
1438 	mode_t mode;
1439 
1440 	for (; table->ctl_name; table++) {
1441 		/* Can't do anything without a proc name. */
1442 		if (!table->procname)
1443 			continue;
1444 		/* Maybe we can't do anything with it... */
1445 		if (!table->proc_handler && !table->child) {
1446 			printk(KERN_WARNING "SYSCTL: Can't register %s\n",
1447 				table->procname);
1448 			continue;
1449 		}
1450 
1451 		len = strlen(table->procname);
1452 		mode = table->mode;
1453 
1454 		de = NULL;
1455 		if (table->proc_handler)
1456 			mode |= S_IFREG;
1457 		else {
1458 			mode |= S_IFDIR;
1459 			for (de = root->subdir; de; de = de->next) {
1460 				if (proc_match(len, table->procname, de))
1461 					break;
1462 			}
1463 			/* If the subdir exists already, de is non-NULL */
1464 		}
1465 
1466 		if (!de) {
1467 			de = create_proc_entry(table->procname, mode, root);
1468 			if (!de)
1469 				continue;
1470 			de->set = set;
1471 			de->data = (void *) table;
1472 			if (table->proc_handler)
1473 				de->proc_fops = &proc_sys_file_operations;
1474 		}
1475 		table->de = de;
1476 		if (de->mode & S_IFDIR)
1477 			register_proc_table(table->child, de, set);
1478 	}
1479 }
1480 
1481 /*
1482  * Unregister a /proc sysctl table and any subdirectories.
1483  */
1484 static void unregister_proc_table(ctl_table * table, struct proc_dir_entry *root)
1485 {
1486 	struct proc_dir_entry *de;
1487 	for (; table->ctl_name; table++) {
1488 		if (!(de = table->de))
1489 			continue;
1490 		if (de->mode & S_IFDIR) {
1491 			if (!table->child) {
1492 				printk (KERN_ALERT "Help - malformed sysctl tree on free\n");
1493 				continue;
1494 			}
1495 			unregister_proc_table(table->child, de);
1496 
1497 			/* Don't unregister directories which still have entries.. */
1498 			if (de->subdir)
1499 				continue;
1500 		}
1501 
1502 		/*
1503 		 * In any case, mark the entry as goner; we'll keep it
1504 		 * around if it's busy, but we'll know to do nothing with
1505 		 * its fields.  We are under sysctl_lock here.
1506 		 */
1507 		de->data = NULL;
1508 
1509 		/* Don't unregister proc entries that are still being used.. */
1510 		if (atomic_read(&de->count))
1511 			continue;
1512 
1513 		table->de = NULL;
1514 		remove_proc_entry(table->procname, root);
1515 	}
1516 }
1517 
1518 static ssize_t do_rw_proc(int write, struct file * file, char __user * buf,
1519 			  size_t count, loff_t *ppos)
1520 {
1521 	int op;
1522 	struct proc_dir_entry *de = PDE(file->f_dentry->d_inode);
1523 	struct ctl_table *table;
1524 	size_t res;
1525 	ssize_t error = -ENOTDIR;
1526 
1527 	spin_lock(&sysctl_lock);
1528 	if (de && de->data && use_table(de->set)) {
1529 		/*
1530 		 * at that point we know that sysctl was not unregistered
1531 		 * and won't be until we finish
1532 		 */
1533 		spin_unlock(&sysctl_lock);
1534 		table = (struct ctl_table *) de->data;
1535 		if (!table || !table->proc_handler)
1536 			goto out;
1537 		error = -EPERM;
1538 		op = (write ? 002 : 004);
1539 		if (ctl_perm(table, op))
1540 			goto out;
1541 
1542 		/* careful: calling conventions are nasty here */
1543 		res = count;
1544 		error = (*table->proc_handler)(table, write, file,
1545 						buf, &res, ppos);
1546 		if (!error)
1547 			error = res;
1548 	out:
1549 		spin_lock(&sysctl_lock);
1550 		unuse_table(de->set);
1551 	}
1552 	spin_unlock(&sysctl_lock);
1553 	return error;
1554 }
1555 
1556 static int proc_opensys(struct inode *inode, struct file *file)
1557 {
1558 	if (file->f_mode & FMODE_WRITE) {
1559 		/*
1560 		 * sysctl entries that are not writable,
1561 		 * are _NOT_ writable, capabilities or not.
1562 		 */
1563 		if (!(inode->i_mode & S_IWUSR))
1564 			return -EPERM;
1565 	}
1566 
1567 	return 0;
1568 }
1569 
1570 static ssize_t proc_readsys(struct file * file, char __user * buf,
1571 			    size_t count, loff_t *ppos)
1572 {
1573 	return do_rw_proc(0, file, buf, count, ppos);
1574 }
1575 
1576 static ssize_t proc_writesys(struct file * file, const char __user * buf,
1577 			     size_t count, loff_t *ppos)
1578 {
1579 	return do_rw_proc(1, file, (char __user *) buf, count, ppos);
1580 }
1581 
1582 /**
1583  * proc_dostring - read a string sysctl
1584  * @table: the sysctl table
1585  * @write: %TRUE if this is a write to the sysctl file
1586  * @filp: the file structure
1587  * @buffer: the user buffer
1588  * @lenp: the size of the user buffer
1589  * @ppos: file position
1590  *
1591  * Reads/writes a string from/to the user buffer. If the kernel
1592  * buffer provided is not large enough to hold the string, the
1593  * string is truncated. The copied string is %NULL-terminated.
1594  * If the string is being read by the user process, it is copied
1595  * and a newline '\n' is added. It is truncated if the buffer is
1596  * not large enough.
1597  *
1598  * Returns 0 on success.
1599  */
1600 int proc_dostring(ctl_table *table, int write, struct file *filp,
1601 		  void __user *buffer, size_t *lenp, loff_t *ppos)
1602 {
1603 	size_t len;
1604 	char __user *p;
1605 	char c;
1606 
1607 	if (!table->data || !table->maxlen || !*lenp ||
1608 	    (*ppos && !write)) {
1609 		*lenp = 0;
1610 		return 0;
1611 	}
1612 
1613 	if (write) {
1614 		len = 0;
1615 		p = buffer;
1616 		while (len < *lenp) {
1617 			if (get_user(c, p++))
1618 				return -EFAULT;
1619 			if (c == 0 || c == '\n')
1620 				break;
1621 			len++;
1622 		}
1623 		if (len >= table->maxlen)
1624 			len = table->maxlen-1;
1625 		if(copy_from_user(table->data, buffer, len))
1626 			return -EFAULT;
1627 		((char *) table->data)[len] = 0;
1628 		*ppos += *lenp;
1629 	} else {
1630 		len = strlen(table->data);
1631 		if (len > table->maxlen)
1632 			len = table->maxlen;
1633 		if (len > *lenp)
1634 			len = *lenp;
1635 		if (len)
1636 			if(copy_to_user(buffer, table->data, len))
1637 				return -EFAULT;
1638 		if (len < *lenp) {
1639 			if(put_user('\n', ((char __user *) buffer) + len))
1640 				return -EFAULT;
1641 			len++;
1642 		}
1643 		*lenp = len;
1644 		*ppos += len;
1645 	}
1646 	return 0;
1647 }
1648 
1649 /*
1650  *	Special case of dostring for the UTS structure. This has locks
1651  *	to observe. Should this be in kernel/sys.c ????
1652  */
1653 
1654 static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
1655 		  void __user *buffer, size_t *lenp, loff_t *ppos)
1656 {
1657 	int r;
1658 
1659 	if (!write) {
1660 		down_read(&uts_sem);
1661 		r=proc_dostring(table,0,filp,buffer,lenp, ppos);
1662 		up_read(&uts_sem);
1663 	} else {
1664 		down_write(&uts_sem);
1665 		r=proc_dostring(table,1,filp,buffer,lenp, ppos);
1666 		up_write(&uts_sem);
1667 	}
1668 	return r;
1669 }
1670 
1671 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1672 				 int *valp,
1673 				 int write, void *data)
1674 {
1675 	if (write) {
1676 		*valp = *negp ? -*lvalp : *lvalp;
1677 	} else {
1678 		int val = *valp;
1679 		if (val < 0) {
1680 			*negp = -1;
1681 			*lvalp = (unsigned long)-val;
1682 		} else {
1683 			*negp = 0;
1684 			*lvalp = (unsigned long)val;
1685 		}
1686 	}
1687 	return 0;
1688 }
1689 
1690 static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1691 		  void __user *buffer, size_t *lenp, loff_t *ppos,
1692 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1693 			      int write, void *data),
1694 		  void *data)
1695 {
1696 #define TMPBUFLEN 21
1697 	int *i, vleft, first=1, neg, val;
1698 	unsigned long lval;
1699 	size_t left, len;
1700 
1701 	char buf[TMPBUFLEN], *p;
1702 	char __user *s = buffer;
1703 
1704 	if (!table->data || !table->maxlen || !*lenp ||
1705 	    (*ppos && !write)) {
1706 		*lenp = 0;
1707 		return 0;
1708 	}
1709 
1710 	i = (int *) table->data;
1711 	vleft = table->maxlen / sizeof(*i);
1712 	left = *lenp;
1713 
1714 	if (!conv)
1715 		conv = do_proc_dointvec_conv;
1716 
1717 	for (; left && vleft--; i++, first=0) {
1718 		if (write) {
1719 			while (left) {
1720 				char c;
1721 				if (get_user(c, s))
1722 					return -EFAULT;
1723 				if (!isspace(c))
1724 					break;
1725 				left--;
1726 				s++;
1727 			}
1728 			if (!left)
1729 				break;
1730 			neg = 0;
1731 			len = left;
1732 			if (len > sizeof(buf) - 1)
1733 				len = sizeof(buf) - 1;
1734 			if (copy_from_user(buf, s, len))
1735 				return -EFAULT;
1736 			buf[len] = 0;
1737 			p = buf;
1738 			if (*p == '-' && left > 1) {
1739 				neg = 1;
1740 				left--, p++;
1741 			}
1742 			if (*p < '0' || *p > '9')
1743 				break;
1744 
1745 			lval = simple_strtoul(p, &p, 0);
1746 
1747 			len = p-buf;
1748 			if ((len < left) && *p && !isspace(*p))
1749 				break;
1750 			if (neg)
1751 				val = -val;
1752 			s += len;
1753 			left -= len;
1754 
1755 			if (conv(&neg, &lval, i, 1, data))
1756 				break;
1757 		} else {
1758 			p = buf;
1759 			if (!first)
1760 				*p++ = '\t';
1761 
1762 			if (conv(&neg, &lval, i, 0, data))
1763 				break;
1764 
1765 			sprintf(p, "%s%lu", neg ? "-" : "", lval);
1766 			len = strlen(buf);
1767 			if (len > left)
1768 				len = left;
1769 			if(copy_to_user(s, buf, len))
1770 				return -EFAULT;
1771 			left -= len;
1772 			s += len;
1773 		}
1774 	}
1775 
1776 	if (!write && !first && left) {
1777 		if(put_user('\n', s))
1778 			return -EFAULT;
1779 		left--, s++;
1780 	}
1781 	if (write) {
1782 		while (left) {
1783 			char c;
1784 			if (get_user(c, s++))
1785 				return -EFAULT;
1786 			if (!isspace(c))
1787 				break;
1788 			left--;
1789 		}
1790 	}
1791 	if (write && first)
1792 		return -EINVAL;
1793 	*lenp -= left;
1794 	*ppos += *lenp;
1795 	return 0;
1796 #undef TMPBUFLEN
1797 }
1798 
1799 /**
1800  * proc_dointvec - read a vector of integers
1801  * @table: the sysctl table
1802  * @write: %TRUE if this is a write to the sysctl file
1803  * @filp: the file structure
1804  * @buffer: the user buffer
1805  * @lenp: the size of the user buffer
1806  * @ppos: file position
1807  *
1808  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1809  * values from/to the user buffer, treated as an ASCII string.
1810  *
1811  * Returns 0 on success.
1812  */
1813 int proc_dointvec(ctl_table *table, int write, struct file *filp,
1814 		     void __user *buffer, size_t *lenp, loff_t *ppos)
1815 {
1816     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1817 		    	    NULL,NULL);
1818 }
1819 
1820 #define OP_SET	0
1821 #define OP_AND	1
1822 #define OP_OR	2
1823 #define OP_MAX	3
1824 #define OP_MIN	4
1825 
1826 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1827 				      int *valp,
1828 				      int write, void *data)
1829 {
1830 	int op = *(int *)data;
1831 	if (write) {
1832 		int val = *negp ? -*lvalp : *lvalp;
1833 		switch(op) {
1834 		case OP_SET:	*valp = val; break;
1835 		case OP_AND:	*valp &= val; break;
1836 		case OP_OR:	*valp |= val; break;
1837 		case OP_MAX:	if(*valp < val)
1838 					*valp = val;
1839 				break;
1840 		case OP_MIN:	if(*valp > val)
1841 				*valp = val;
1842 				break;
1843 		}
1844 	} else {
1845 		int val = *valp;
1846 		if (val < 0) {
1847 			*negp = -1;
1848 			*lvalp = (unsigned long)-val;
1849 		} else {
1850 			*negp = 0;
1851 			*lvalp = (unsigned long)val;
1852 		}
1853 	}
1854 	return 0;
1855 }
1856 
1857 /*
1858  *	init may raise the set.
1859  */
1860 
1861 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1862 			void __user *buffer, size_t *lenp, loff_t *ppos)
1863 {
1864 	int op;
1865 
1866 	if (!capable(CAP_SYS_MODULE)) {
1867 		return -EPERM;
1868 	}
1869 
1870 	op = (current->pid == 1) ? OP_SET : OP_AND;
1871 	return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1872 				do_proc_dointvec_bset_conv,&op);
1873 }
1874 
1875 struct do_proc_dointvec_minmax_conv_param {
1876 	int *min;
1877 	int *max;
1878 };
1879 
1880 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1881 					int *valp,
1882 					int write, void *data)
1883 {
1884 	struct do_proc_dointvec_minmax_conv_param *param = data;
1885 	if (write) {
1886 		int val = *negp ? -*lvalp : *lvalp;
1887 		if ((param->min && *param->min > val) ||
1888 		    (param->max && *param->max < val))
1889 			return -EINVAL;
1890 		*valp = val;
1891 	} else {
1892 		int val = *valp;
1893 		if (val < 0) {
1894 			*negp = -1;
1895 			*lvalp = (unsigned long)-val;
1896 		} else {
1897 			*negp = 0;
1898 			*lvalp = (unsigned long)val;
1899 		}
1900 	}
1901 	return 0;
1902 }
1903 
1904 /**
1905  * proc_dointvec_minmax - read a vector of integers with min/max values
1906  * @table: the sysctl table
1907  * @write: %TRUE if this is a write to the sysctl file
1908  * @filp: the file structure
1909  * @buffer: the user buffer
1910  * @lenp: the size of the user buffer
1911  * @ppos: file position
1912  *
1913  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1914  * values from/to the user buffer, treated as an ASCII string.
1915  *
1916  * This routine will ensure the values are within the range specified by
1917  * table->extra1 (min) and table->extra2 (max).
1918  *
1919  * Returns 0 on success.
1920  */
1921 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1922 		  void __user *buffer, size_t *lenp, loff_t *ppos)
1923 {
1924 	struct do_proc_dointvec_minmax_conv_param param = {
1925 		.min = (int *) table->extra1,
1926 		.max = (int *) table->extra2,
1927 	};
1928 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1929 				do_proc_dointvec_minmax_conv, &param);
1930 }
1931 
1932 static int do_proc_doulongvec_minmax(ctl_table *table, int write,
1933 				     struct file *filp,
1934 				     void __user *buffer,
1935 				     size_t *lenp, loff_t *ppos,
1936 				     unsigned long convmul,
1937 				     unsigned long convdiv)
1938 {
1939 #define TMPBUFLEN 21
1940 	unsigned long *i, *min, *max, val;
1941 	int vleft, first=1, neg;
1942 	size_t len, left;
1943 	char buf[TMPBUFLEN], *p;
1944 	char __user *s = buffer;
1945 
1946 	if (!table->data || !table->maxlen || !*lenp ||
1947 	    (*ppos && !write)) {
1948 		*lenp = 0;
1949 		return 0;
1950 	}
1951 
1952 	i = (unsigned long *) table->data;
1953 	min = (unsigned long *) table->extra1;
1954 	max = (unsigned long *) table->extra2;
1955 	vleft = table->maxlen / sizeof(unsigned long);
1956 	left = *lenp;
1957 
1958 	for (; left && vleft--; i++, min++, max++, first=0) {
1959 		if (write) {
1960 			while (left) {
1961 				char c;
1962 				if (get_user(c, s))
1963 					return -EFAULT;
1964 				if (!isspace(c))
1965 					break;
1966 				left--;
1967 				s++;
1968 			}
1969 			if (!left)
1970 				break;
1971 			neg = 0;
1972 			len = left;
1973 			if (len > TMPBUFLEN-1)
1974 				len = TMPBUFLEN-1;
1975 			if (copy_from_user(buf, s, len))
1976 				return -EFAULT;
1977 			buf[len] = 0;
1978 			p = buf;
1979 			if (*p == '-' && left > 1) {
1980 				neg = 1;
1981 				left--, p++;
1982 			}
1983 			if (*p < '0' || *p > '9')
1984 				break;
1985 			val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
1986 			len = p-buf;
1987 			if ((len < left) && *p && !isspace(*p))
1988 				break;
1989 			if (neg)
1990 				val = -val;
1991 			s += len;
1992 			left -= len;
1993 
1994 			if(neg)
1995 				continue;
1996 			if ((min && val < *min) || (max && val > *max))
1997 				continue;
1998 			*i = val;
1999 		} else {
2000 			p = buf;
2001 			if (!first)
2002 				*p++ = '\t';
2003 			sprintf(p, "%lu", convdiv * (*i) / convmul);
2004 			len = strlen(buf);
2005 			if (len > left)
2006 				len = left;
2007 			if(copy_to_user(s, buf, len))
2008 				return -EFAULT;
2009 			left -= len;
2010 			s += len;
2011 		}
2012 	}
2013 
2014 	if (!write && !first && left) {
2015 		if(put_user('\n', s))
2016 			return -EFAULT;
2017 		left--, s++;
2018 	}
2019 	if (write) {
2020 		while (left) {
2021 			char c;
2022 			if (get_user(c, s++))
2023 				return -EFAULT;
2024 			if (!isspace(c))
2025 				break;
2026 			left--;
2027 		}
2028 	}
2029 	if (write && first)
2030 		return -EINVAL;
2031 	*lenp -= left;
2032 	*ppos += *lenp;
2033 	return 0;
2034 #undef TMPBUFLEN
2035 }
2036 
2037 /**
2038  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2039  * @table: the sysctl table
2040  * @write: %TRUE if this is a write to the sysctl file
2041  * @filp: the file structure
2042  * @buffer: the user buffer
2043  * @lenp: the size of the user buffer
2044  * @ppos: file position
2045  *
2046  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2047  * values from/to the user buffer, treated as an ASCII string.
2048  *
2049  * This routine will ensure the values are within the range specified by
2050  * table->extra1 (min) and table->extra2 (max).
2051  *
2052  * Returns 0 on success.
2053  */
2054 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2055 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2056 {
2057     return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2058 }
2059 
2060 /**
2061  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2062  * @table: the sysctl table
2063  * @write: %TRUE if this is a write to the sysctl file
2064  * @filp: the file structure
2065  * @buffer: the user buffer
2066  * @lenp: the size of the user buffer
2067  * @ppos: file position
2068  *
2069  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2070  * values from/to the user buffer, treated as an ASCII string. The values
2071  * are treated as milliseconds, and converted to jiffies when they are stored.
2072  *
2073  * This routine will ensure the values are within the range specified by
2074  * table->extra1 (min) and table->extra2 (max).
2075  *
2076  * Returns 0 on success.
2077  */
2078 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2079 				      struct file *filp,
2080 				      void __user *buffer,
2081 				      size_t *lenp, loff_t *ppos)
2082 {
2083     return do_proc_doulongvec_minmax(table, write, filp, buffer,
2084 				     lenp, ppos, HZ, 1000l);
2085 }
2086 
2087 
2088 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2089 					 int *valp,
2090 					 int write, void *data)
2091 {
2092 	if (write) {
2093 		if (*lvalp > LONG_MAX / HZ)
2094 			return 1;
2095 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2096 	} else {
2097 		int val = *valp;
2098 		unsigned long lval;
2099 		if (val < 0) {
2100 			*negp = -1;
2101 			lval = (unsigned long)-val;
2102 		} else {
2103 			*negp = 0;
2104 			lval = (unsigned long)val;
2105 		}
2106 		*lvalp = lval / HZ;
2107 	}
2108 	return 0;
2109 }
2110 
2111 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2112 						int *valp,
2113 						int write, void *data)
2114 {
2115 	if (write) {
2116 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2117 			return 1;
2118 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2119 	} else {
2120 		int val = *valp;
2121 		unsigned long lval;
2122 		if (val < 0) {
2123 			*negp = -1;
2124 			lval = (unsigned long)-val;
2125 		} else {
2126 			*negp = 0;
2127 			lval = (unsigned long)val;
2128 		}
2129 		*lvalp = jiffies_to_clock_t(lval);
2130 	}
2131 	return 0;
2132 }
2133 
2134 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2135 					    int *valp,
2136 					    int write, void *data)
2137 {
2138 	if (write) {
2139 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2140 	} else {
2141 		int val = *valp;
2142 		unsigned long lval;
2143 		if (val < 0) {
2144 			*negp = -1;
2145 			lval = (unsigned long)-val;
2146 		} else {
2147 			*negp = 0;
2148 			lval = (unsigned long)val;
2149 		}
2150 		*lvalp = jiffies_to_msecs(lval);
2151 	}
2152 	return 0;
2153 }
2154 
2155 /**
2156  * proc_dointvec_jiffies - read a vector of integers as seconds
2157  * @table: the sysctl table
2158  * @write: %TRUE if this is a write to the sysctl file
2159  * @filp: the file structure
2160  * @buffer: the user buffer
2161  * @lenp: the size of the user buffer
2162  * @ppos: file position
2163  *
2164  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2165  * values from/to the user buffer, treated as an ASCII string.
2166  * The values read are assumed to be in seconds, and are converted into
2167  * jiffies.
2168  *
2169  * Returns 0 on success.
2170  */
2171 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2172 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2173 {
2174     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2175 		    	    do_proc_dointvec_jiffies_conv,NULL);
2176 }
2177 
2178 /**
2179  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2180  * @table: the sysctl table
2181  * @write: %TRUE if this is a write to the sysctl file
2182  * @filp: the file structure
2183  * @buffer: the user buffer
2184  * @lenp: the size of the user buffer
2185  * @ppos: pointer to the file position
2186  *
2187  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2188  * values from/to the user buffer, treated as an ASCII string.
2189  * The values read are assumed to be in 1/USER_HZ seconds, and
2190  * are converted into jiffies.
2191  *
2192  * Returns 0 on success.
2193  */
2194 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2195 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2196 {
2197     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2198 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2199 }
2200 
2201 /**
2202  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2203  * @table: the sysctl table
2204  * @write: %TRUE if this is a write to the sysctl file
2205  * @filp: the file structure
2206  * @buffer: the user buffer
2207  * @lenp: the size of the user buffer
2208  * @ppos: file position
2209  * @ppos: the current position in the file
2210  *
2211  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2212  * values from/to the user buffer, treated as an ASCII string.
2213  * The values read are assumed to be in 1/1000 seconds, and
2214  * are converted into jiffies.
2215  *
2216  * Returns 0 on success.
2217  */
2218 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2219 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2220 {
2221 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2222 				do_proc_dointvec_ms_jiffies_conv, NULL);
2223 }
2224 
2225 #else /* CONFIG_PROC_FS */
2226 
2227 int proc_dostring(ctl_table *table, int write, struct file *filp,
2228 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2229 {
2230 	return -ENOSYS;
2231 }
2232 
2233 static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
2234 			    void __user *buffer, size_t *lenp, loff_t *ppos)
2235 {
2236 	return -ENOSYS;
2237 }
2238 
2239 int proc_dointvec(ctl_table *table, int write, struct file *filp,
2240 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2241 {
2242 	return -ENOSYS;
2243 }
2244 
2245 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2246 			void __user *buffer, size_t *lenp, loff_t *ppos)
2247 {
2248 	return -ENOSYS;
2249 }
2250 
2251 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2252 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2253 {
2254 	return -ENOSYS;
2255 }
2256 
2257 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2258 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2259 {
2260 	return -ENOSYS;
2261 }
2262 
2263 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2264 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2265 {
2266 	return -ENOSYS;
2267 }
2268 
2269 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2270 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2271 {
2272 	return -ENOSYS;
2273 }
2274 
2275 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2276 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2277 {
2278 	return -ENOSYS;
2279 }
2280 
2281 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2282 				      struct file *filp,
2283 				      void __user *buffer,
2284 				      size_t *lenp, loff_t *ppos)
2285 {
2286     return -ENOSYS;
2287 }
2288 
2289 
2290 #endif /* CONFIG_PROC_FS */
2291 
2292 
2293 /*
2294  * General sysctl support routines
2295  */
2296 
2297 /* The generic string strategy routine: */
2298 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2299 		  void __user *oldval, size_t __user *oldlenp,
2300 		  void __user *newval, size_t newlen, void **context)
2301 {
2302 	if (!table->data || !table->maxlen)
2303 		return -ENOTDIR;
2304 
2305 	if (oldval && oldlenp) {
2306 		size_t bufsize;
2307 		if (get_user(bufsize, oldlenp))
2308 			return -EFAULT;
2309 		if (bufsize) {
2310 			size_t len = strlen(table->data), copied;
2311 
2312 			/* This shouldn't trigger for a well-formed sysctl */
2313 			if (len > table->maxlen)
2314 				len = table->maxlen;
2315 
2316 			/* Copy up to a max of bufsize-1 bytes of the string */
2317 			copied = (len >= bufsize) ? bufsize - 1 : len;
2318 
2319 			if (copy_to_user(oldval, table->data, copied) ||
2320 			    put_user(0, (char __user *)(oldval + copied)))
2321 				return -EFAULT;
2322 			if (put_user(len, oldlenp))
2323 				return -EFAULT;
2324 		}
2325 	}
2326 	if (newval && newlen) {
2327 		size_t len = newlen;
2328 		if (len > table->maxlen)
2329 			len = table->maxlen;
2330 		if(copy_from_user(table->data, newval, len))
2331 			return -EFAULT;
2332 		if (len == table->maxlen)
2333 			len--;
2334 		((char *) table->data)[len] = 0;
2335 	}
2336 	return 1;
2337 }
2338 
2339 /*
2340  * This function makes sure that all of the integers in the vector
2341  * are between the minimum and maximum values given in the arrays
2342  * table->extra1 and table->extra2, respectively.
2343  */
2344 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2345 		void __user *oldval, size_t __user *oldlenp,
2346 		void __user *newval, size_t newlen, void **context)
2347 {
2348 
2349 	if (newval && newlen) {
2350 		int __user *vec = (int __user *) newval;
2351 		int *min = (int *) table->extra1;
2352 		int *max = (int *) table->extra2;
2353 		size_t length;
2354 		int i;
2355 
2356 		if (newlen % sizeof(int) != 0)
2357 			return -EINVAL;
2358 
2359 		if (!table->extra1 && !table->extra2)
2360 			return 0;
2361 
2362 		if (newlen > table->maxlen)
2363 			newlen = table->maxlen;
2364 		length = newlen / sizeof(int);
2365 
2366 		for (i = 0; i < length; i++) {
2367 			int value;
2368 			if (get_user(value, vec + i))
2369 				return -EFAULT;
2370 			if (min && value < min[i])
2371 				return -EINVAL;
2372 			if (max && value > max[i])
2373 				return -EINVAL;
2374 		}
2375 	}
2376 	return 0;
2377 }
2378 
2379 /* Strategy function to convert jiffies to seconds */
2380 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2381 		void __user *oldval, size_t __user *oldlenp,
2382 		void __user *newval, size_t newlen, void **context)
2383 {
2384 	if (oldval) {
2385 		size_t olen;
2386 		if (oldlenp) {
2387 			if (get_user(olen, oldlenp))
2388 				return -EFAULT;
2389 			if (olen!=sizeof(int))
2390 				return -EINVAL;
2391 		}
2392 		if (put_user(*(int *)(table->data)/HZ, (int __user *)oldval) ||
2393 		    (oldlenp && put_user(sizeof(int),oldlenp)))
2394 			return -EFAULT;
2395 	}
2396 	if (newval && newlen) {
2397 		int new;
2398 		if (newlen != sizeof(int))
2399 			return -EINVAL;
2400 		if (get_user(new, (int __user *)newval))
2401 			return -EFAULT;
2402 		*(int *)(table->data) = new*HZ;
2403 	}
2404 	return 1;
2405 }
2406 
2407 /* Strategy function to convert jiffies to seconds */
2408 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2409 		void __user *oldval, size_t __user *oldlenp,
2410 		void __user *newval, size_t newlen, void **context)
2411 {
2412 	if (oldval) {
2413 		size_t olen;
2414 		if (oldlenp) {
2415 			if (get_user(olen, oldlenp))
2416 				return -EFAULT;
2417 			if (olen!=sizeof(int))
2418 				return -EINVAL;
2419 		}
2420 		if (put_user(jiffies_to_msecs(*(int *)(table->data)), (int __user *)oldval) ||
2421 		    (oldlenp && put_user(sizeof(int),oldlenp)))
2422 			return -EFAULT;
2423 	}
2424 	if (newval && newlen) {
2425 		int new;
2426 		if (newlen != sizeof(int))
2427 			return -EINVAL;
2428 		if (get_user(new, (int __user *)newval))
2429 			return -EFAULT;
2430 		*(int *)(table->data) = msecs_to_jiffies(new);
2431 	}
2432 	return 1;
2433 }
2434 
2435 #else /* CONFIG_SYSCTL */
2436 
2437 
2438 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2439 {
2440 	return -ENOSYS;
2441 }
2442 
2443 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2444 		  void __user *oldval, size_t __user *oldlenp,
2445 		  void __user *newval, size_t newlen, void **context)
2446 {
2447 	return -ENOSYS;
2448 }
2449 
2450 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2451 		void __user *oldval, size_t __user *oldlenp,
2452 		void __user *newval, size_t newlen, void **context)
2453 {
2454 	return -ENOSYS;
2455 }
2456 
2457 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2458 		void __user *oldval, size_t __user *oldlenp,
2459 		void __user *newval, size_t newlen, void **context)
2460 {
2461 	return -ENOSYS;
2462 }
2463 
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, void **context)
2467 {
2468 	return -ENOSYS;
2469 }
2470 
2471 int proc_dostring(ctl_table *table, int write, struct file *filp,
2472 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2473 {
2474 	return -ENOSYS;
2475 }
2476 
2477 int proc_dointvec(ctl_table *table, int write, struct file *filp,
2478 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2479 {
2480 	return -ENOSYS;
2481 }
2482 
2483 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2484 			void __user *buffer, size_t *lenp, loff_t *ppos)
2485 {
2486 	return -ENOSYS;
2487 }
2488 
2489 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2490 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2491 {
2492 	return -ENOSYS;
2493 }
2494 
2495 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2496 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2497 {
2498 	return -ENOSYS;
2499 }
2500 
2501 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2502 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2503 {
2504 	return -ENOSYS;
2505 }
2506 
2507 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2508 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2509 {
2510 	return -ENOSYS;
2511 }
2512 
2513 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2514 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2515 {
2516 	return -ENOSYS;
2517 }
2518 
2519 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2520 				      struct file *filp,
2521 				      void __user *buffer,
2522 				      size_t *lenp, loff_t *ppos)
2523 {
2524     return -ENOSYS;
2525 }
2526 
2527 struct ctl_table_header * register_sysctl_table(ctl_table * table,
2528 						int insert_at_head)
2529 {
2530 	return NULL;
2531 }
2532 
2533 void unregister_sysctl_table(struct ctl_table_header * table)
2534 {
2535 }
2536 
2537 #endif /* CONFIG_SYSCTL */
2538 
2539 /*
2540  * No sense putting this after each symbol definition, twice,
2541  * exception granted :-)
2542  */
2543 EXPORT_SYMBOL(proc_dointvec);
2544 EXPORT_SYMBOL(proc_dointvec_jiffies);
2545 EXPORT_SYMBOL(proc_dointvec_minmax);
2546 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2547 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2548 EXPORT_SYMBOL(proc_dostring);
2549 EXPORT_SYMBOL(proc_doulongvec_minmax);
2550 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2551 EXPORT_SYMBOL(register_sysctl_table);
2552 EXPORT_SYMBOL(sysctl_intvec);
2553 EXPORT_SYMBOL(sysctl_jiffies);
2554 EXPORT_SYMBOL(sysctl_ms_jiffies);
2555 EXPORT_SYMBOL(sysctl_string);
2556 EXPORT_SYMBOL(unregister_sysctl_table);
2557