xref: /linux/kernel/sysctl.c (revision 9ce7677cfd7cd871adb457c80bea3b581b839641)
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/config.h>
22 #include <linux/module.h>
23 #include <linux/mm.h>
24 #include <linux/swap.h>
25 #include <linux/slab.h>
26 #include <linux/sysctl.h>
27 #include <linux/proc_fs.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/net.h>
35 #include <linux/sysrq.h>
36 #include <linux/highuid.h>
37 #include <linux/writeback.h>
38 #include <linux/hugetlb.h>
39 #include <linux/security.h>
40 #include <linux/initrd.h>
41 #include <linux/times.h>
42 #include <linux/limits.h>
43 #include <linux/dcache.h>
44 #include <linux/syscalls.h>
45 
46 #include <asm/uaccess.h>
47 #include <asm/processor.h>
48 
49 #ifdef CONFIG_ROOT_NFS
50 #include <linux/nfs_fs.h>
51 #endif
52 
53 #if defined(CONFIG_SYSCTL)
54 
55 /* External variables not in a header file. */
56 extern int C_A_D;
57 extern int sysctl_overcommit_memory;
58 extern int sysctl_overcommit_ratio;
59 extern int max_threads;
60 extern int sysrq_enabled;
61 extern int core_uses_pid;
62 extern int suid_dumpable;
63 extern char core_pattern[];
64 extern int cad_pid;
65 extern int pid_max;
66 extern int min_free_kbytes;
67 extern int printk_ratelimit_jiffies;
68 extern int printk_ratelimit_burst;
69 extern int pid_max_min, pid_max_max;
70 
71 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
72 int unknown_nmi_panic;
73 extern int proc_unknown_nmi_panic(ctl_table *, int, struct file *,
74 				  void __user *, size_t *, loff_t *);
75 #endif
76 
77 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
78 static int maxolduid = 65535;
79 static int minolduid;
80 
81 static int ngroups_max = NGROUPS_MAX;
82 
83 #ifdef CONFIG_KMOD
84 extern char modprobe_path[];
85 #endif
86 #ifdef CONFIG_HOTPLUG
87 extern char hotplug_path[];
88 #endif
89 #ifdef CONFIG_CHR_DEV_SG
90 extern int sg_big_buff;
91 #endif
92 #ifdef CONFIG_SYSVIPC
93 extern size_t shm_ctlmax;
94 extern size_t shm_ctlall;
95 extern int shm_ctlmni;
96 extern int msg_ctlmax;
97 extern int msg_ctlmnb;
98 extern int msg_ctlmni;
99 extern int sem_ctls[];
100 #endif
101 
102 #ifdef __sparc__
103 extern char reboot_command [];
104 extern int stop_a_enabled;
105 extern int scons_pwroff;
106 #endif
107 
108 #ifdef __hppa__
109 extern int pwrsw_enabled;
110 extern int unaligned_enabled;
111 #endif
112 
113 #ifdef CONFIG_ARCH_S390
114 #ifdef CONFIG_MATHEMU
115 extern int sysctl_ieee_emulation_warnings;
116 #endif
117 extern int sysctl_userprocess_debug;
118 extern int spin_retry;
119 #endif
120 
121 extern int sysctl_hz_timer;
122 
123 #ifdef CONFIG_BSD_PROCESS_ACCT
124 extern int acct_parm[];
125 #endif
126 
127 int randomize_va_space = 1;
128 
129 static int parse_table(int __user *, int, void __user *, size_t __user *, void __user *, size_t,
130 		       ctl_table *, void **);
131 static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
132 		  void __user *buffer, size_t *lenp, loff_t *ppos);
133 
134 static ctl_table root_table[];
135 static struct ctl_table_header root_table_header =
136 	{ root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
137 
138 static ctl_table kern_table[];
139 static ctl_table vm_table[];
140 static ctl_table proc_table[];
141 static ctl_table fs_table[];
142 static ctl_table debug_table[];
143 static ctl_table dev_table[];
144 extern ctl_table random_table[];
145 #ifdef CONFIG_UNIX98_PTYS
146 extern ctl_table pty_table[];
147 #endif
148 #ifdef CONFIG_INOTIFY
149 extern ctl_table inotify_table[];
150 #endif
151 
152 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
153 int sysctl_legacy_va_layout;
154 #endif
155 
156 /* /proc declarations: */
157 
158 #ifdef CONFIG_PROC_FS
159 
160 static ssize_t proc_readsys(struct file *, char __user *, size_t, loff_t *);
161 static ssize_t proc_writesys(struct file *, const char __user *, size_t, loff_t *);
162 static int proc_opensys(struct inode *, struct file *);
163 
164 struct file_operations proc_sys_file_operations = {
165 	.open		= proc_opensys,
166 	.read		= proc_readsys,
167 	.write		= proc_writesys,
168 };
169 
170 extern struct proc_dir_entry *proc_sys_root;
171 
172 static void register_proc_table(ctl_table *, struct proc_dir_entry *, void *);
173 static void unregister_proc_table(ctl_table *, struct proc_dir_entry *);
174 #endif
175 
176 /* The default sysctl tables: */
177 
178 static ctl_table root_table[] = {
179 	{
180 		.ctl_name	= CTL_KERN,
181 		.procname	= "kernel",
182 		.mode		= 0555,
183 		.child		= kern_table,
184 	},
185 	{
186 		.ctl_name	= CTL_VM,
187 		.procname	= "vm",
188 		.mode		= 0555,
189 		.child		= vm_table,
190 	},
191 #ifdef CONFIG_NET
192 	{
193 		.ctl_name	= CTL_NET,
194 		.procname	= "net",
195 		.mode		= 0555,
196 		.child		= net_table,
197 	},
198 #endif
199 	{
200 		.ctl_name	= CTL_PROC,
201 		.procname	= "proc",
202 		.mode		= 0555,
203 		.child		= proc_table,
204 	},
205 	{
206 		.ctl_name	= CTL_FS,
207 		.procname	= "fs",
208 		.mode		= 0555,
209 		.child		= fs_table,
210 	},
211 	{
212 		.ctl_name	= CTL_DEBUG,
213 		.procname	= "debug",
214 		.mode		= 0555,
215 		.child		= debug_table,
216 	},
217 	{
218 		.ctl_name	= CTL_DEV,
219 		.procname	= "dev",
220 		.mode		= 0555,
221 		.child		= dev_table,
222 	},
223 
224 	{ .ctl_name = 0 }
225 };
226 
227 static ctl_table kern_table[] = {
228 	{
229 		.ctl_name	= KERN_OSTYPE,
230 		.procname	= "ostype",
231 		.data		= system_utsname.sysname,
232 		.maxlen		= sizeof(system_utsname.sysname),
233 		.mode		= 0444,
234 		.proc_handler	= &proc_doutsstring,
235 		.strategy	= &sysctl_string,
236 	},
237 	{
238 		.ctl_name	= KERN_OSRELEASE,
239 		.procname	= "osrelease",
240 		.data		= system_utsname.release,
241 		.maxlen		= sizeof(system_utsname.release),
242 		.mode		= 0444,
243 		.proc_handler	= &proc_doutsstring,
244 		.strategy	= &sysctl_string,
245 	},
246 	{
247 		.ctl_name	= KERN_VERSION,
248 		.procname	= "version",
249 		.data		= system_utsname.version,
250 		.maxlen		= sizeof(system_utsname.version),
251 		.mode		= 0444,
252 		.proc_handler	= &proc_doutsstring,
253 		.strategy	= &sysctl_string,
254 	},
255 	{
256 		.ctl_name	= KERN_NODENAME,
257 		.procname	= "hostname",
258 		.data		= system_utsname.nodename,
259 		.maxlen		= sizeof(system_utsname.nodename),
260 		.mode		= 0644,
261 		.proc_handler	= &proc_doutsstring,
262 		.strategy	= &sysctl_string,
263 	},
264 	{
265 		.ctl_name	= KERN_DOMAINNAME,
266 		.procname	= "domainname",
267 		.data		= system_utsname.domainname,
268 		.maxlen		= sizeof(system_utsname.domainname),
269 		.mode		= 0644,
270 		.proc_handler	= &proc_doutsstring,
271 		.strategy	= &sysctl_string,
272 	},
273 	{
274 		.ctl_name	= KERN_PANIC,
275 		.procname	= "panic",
276 		.data		= &panic_timeout,
277 		.maxlen		= sizeof(int),
278 		.mode		= 0644,
279 		.proc_handler	= &proc_dointvec,
280 	},
281 	{
282 		.ctl_name	= KERN_CORE_USES_PID,
283 		.procname	= "core_uses_pid",
284 		.data		= &core_uses_pid,
285 		.maxlen		= sizeof(int),
286 		.mode		= 0644,
287 		.proc_handler	= &proc_dointvec,
288 	},
289 	{
290 		.ctl_name	= KERN_CORE_PATTERN,
291 		.procname	= "core_pattern",
292 		.data		= core_pattern,
293 		.maxlen		= 64,
294 		.mode		= 0644,
295 		.proc_handler	= &proc_dostring,
296 		.strategy	= &sysctl_string,
297 	},
298 	{
299 		.ctl_name	= KERN_TAINTED,
300 		.procname	= "tainted",
301 		.data		= &tainted,
302 		.maxlen		= sizeof(int),
303 		.mode		= 0444,
304 		.proc_handler	= &proc_dointvec,
305 	},
306 	{
307 		.ctl_name	= KERN_CAP_BSET,
308 		.procname	= "cap-bound",
309 		.data		= &cap_bset,
310 		.maxlen		= sizeof(kernel_cap_t),
311 		.mode		= 0600,
312 		.proc_handler	= &proc_dointvec_bset,
313 	},
314 #ifdef CONFIG_BLK_DEV_INITRD
315 	{
316 		.ctl_name	= KERN_REALROOTDEV,
317 		.procname	= "real-root-dev",
318 		.data		= &real_root_dev,
319 		.maxlen		= sizeof(int),
320 		.mode		= 0644,
321 		.proc_handler	= &proc_dointvec,
322 	},
323 #endif
324 #ifdef __sparc__
325 	{
326 		.ctl_name	= KERN_SPARC_REBOOT,
327 		.procname	= "reboot-cmd",
328 		.data		= reboot_command,
329 		.maxlen		= 256,
330 		.mode		= 0644,
331 		.proc_handler	= &proc_dostring,
332 		.strategy	= &sysctl_string,
333 	},
334 	{
335 		.ctl_name	= KERN_SPARC_STOP_A,
336 		.procname	= "stop-a",
337 		.data		= &stop_a_enabled,
338 		.maxlen		= sizeof (int),
339 		.mode		= 0644,
340 		.proc_handler	= &proc_dointvec,
341 	},
342 	{
343 		.ctl_name	= KERN_SPARC_SCONS_PWROFF,
344 		.procname	= "scons-poweroff",
345 		.data		= &scons_pwroff,
346 		.maxlen		= sizeof (int),
347 		.mode		= 0644,
348 		.proc_handler	= &proc_dointvec,
349 	},
350 #endif
351 #ifdef __hppa__
352 	{
353 		.ctl_name	= KERN_HPPA_PWRSW,
354 		.procname	= "soft-power",
355 		.data		= &pwrsw_enabled,
356 		.maxlen		= sizeof (int),
357 	 	.mode		= 0644,
358 		.proc_handler	= &proc_dointvec,
359 	},
360 	{
361 		.ctl_name	= KERN_HPPA_UNALIGNED,
362 		.procname	= "unaligned-trap",
363 		.data		= &unaligned_enabled,
364 		.maxlen		= sizeof (int),
365 		.mode		= 0644,
366 		.proc_handler	= &proc_dointvec,
367 	},
368 #endif
369 	{
370 		.ctl_name	= KERN_CTLALTDEL,
371 		.procname	= "ctrl-alt-del",
372 		.data		= &C_A_D,
373 		.maxlen		= sizeof(int),
374 		.mode		= 0644,
375 		.proc_handler	= &proc_dointvec,
376 	},
377 	{
378 		.ctl_name	= KERN_PRINTK,
379 		.procname	= "printk",
380 		.data		= &console_loglevel,
381 		.maxlen		= 4*sizeof(int),
382 		.mode		= 0644,
383 		.proc_handler	= &proc_dointvec,
384 	},
385 #ifdef CONFIG_KMOD
386 	{
387 		.ctl_name	= KERN_MODPROBE,
388 		.procname	= "modprobe",
389 		.data		= &modprobe_path,
390 		.maxlen		= KMOD_PATH_LEN,
391 		.mode		= 0644,
392 		.proc_handler	= &proc_dostring,
393 		.strategy	= &sysctl_string,
394 	},
395 #endif
396 #ifdef CONFIG_HOTPLUG
397 	{
398 		.ctl_name	= KERN_HOTPLUG,
399 		.procname	= "hotplug",
400 		.data		= &hotplug_path,
401 		.maxlen		= HOTPLUG_PATH_LEN,
402 		.mode		= 0644,
403 		.proc_handler	= &proc_dostring,
404 		.strategy	= &sysctl_string,
405 	},
406 #endif
407 #ifdef CONFIG_CHR_DEV_SG
408 	{
409 		.ctl_name	= KERN_SG_BIG_BUFF,
410 		.procname	= "sg-big-buff",
411 		.data		= &sg_big_buff,
412 		.maxlen		= sizeof (int),
413 		.mode		= 0444,
414 		.proc_handler	= &proc_dointvec,
415 	},
416 #endif
417 #ifdef CONFIG_BSD_PROCESS_ACCT
418 	{
419 		.ctl_name	= KERN_ACCT,
420 		.procname	= "acct",
421 		.data		= &acct_parm,
422 		.maxlen		= 3*sizeof(int),
423 		.mode		= 0644,
424 		.proc_handler	= &proc_dointvec,
425 	},
426 #endif
427 #ifdef CONFIG_SYSVIPC
428 	{
429 		.ctl_name	= KERN_SHMMAX,
430 		.procname	= "shmmax",
431 		.data		= &shm_ctlmax,
432 		.maxlen		= sizeof (size_t),
433 		.mode		= 0644,
434 		.proc_handler	= &proc_doulongvec_minmax,
435 	},
436 	{
437 		.ctl_name	= KERN_SHMALL,
438 		.procname	= "shmall",
439 		.data		= &shm_ctlall,
440 		.maxlen		= sizeof (size_t),
441 		.mode		= 0644,
442 		.proc_handler	= &proc_doulongvec_minmax,
443 	},
444 	{
445 		.ctl_name	= KERN_SHMMNI,
446 		.procname	= "shmmni",
447 		.data		= &shm_ctlmni,
448 		.maxlen		= sizeof (int),
449 		.mode		= 0644,
450 		.proc_handler	= &proc_dointvec,
451 	},
452 	{
453 		.ctl_name	= KERN_MSGMAX,
454 		.procname	= "msgmax",
455 		.data		= &msg_ctlmax,
456 		.maxlen		= sizeof (int),
457 		.mode		= 0644,
458 		.proc_handler	= &proc_dointvec,
459 	},
460 	{
461 		.ctl_name	= KERN_MSGMNI,
462 		.procname	= "msgmni",
463 		.data		= &msg_ctlmni,
464 		.maxlen		= sizeof (int),
465 		.mode		= 0644,
466 		.proc_handler	= &proc_dointvec,
467 	},
468 	{
469 		.ctl_name	= KERN_MSGMNB,
470 		.procname	=  "msgmnb",
471 		.data		= &msg_ctlmnb,
472 		.maxlen		= sizeof (int),
473 		.mode		= 0644,
474 		.proc_handler	= &proc_dointvec,
475 	},
476 	{
477 		.ctl_name	= KERN_SEM,
478 		.procname	= "sem",
479 		.data		= &sem_ctls,
480 		.maxlen		= 4*sizeof (int),
481 		.mode		= 0644,
482 		.proc_handler	= &proc_dointvec,
483 	},
484 #endif
485 #ifdef CONFIG_MAGIC_SYSRQ
486 	{
487 		.ctl_name	= KERN_SYSRQ,
488 		.procname	= "sysrq",
489 		.data		= &sysrq_enabled,
490 		.maxlen		= sizeof (int),
491 		.mode		= 0644,
492 		.proc_handler	= &proc_dointvec,
493 	},
494 #endif
495 	{
496 		.ctl_name	= KERN_CADPID,
497 		.procname	= "cad_pid",
498 		.data		= &cad_pid,
499 		.maxlen		= sizeof (int),
500 		.mode		= 0600,
501 		.proc_handler	= &proc_dointvec,
502 	},
503 	{
504 		.ctl_name	= KERN_MAX_THREADS,
505 		.procname	= "threads-max",
506 		.data		= &max_threads,
507 		.maxlen		= sizeof(int),
508 		.mode		= 0644,
509 		.proc_handler	= &proc_dointvec,
510 	},
511 	{
512 		.ctl_name	= KERN_RANDOM,
513 		.procname	= "random",
514 		.mode		= 0555,
515 		.child		= random_table,
516 	},
517 #ifdef CONFIG_UNIX98_PTYS
518 	{
519 		.ctl_name	= KERN_PTY,
520 		.procname	= "pty",
521 		.mode		= 0555,
522 		.child		= pty_table,
523 	},
524 #endif
525 	{
526 		.ctl_name	= KERN_OVERFLOWUID,
527 		.procname	= "overflowuid",
528 		.data		= &overflowuid,
529 		.maxlen		= sizeof(int),
530 		.mode		= 0644,
531 		.proc_handler	= &proc_dointvec_minmax,
532 		.strategy	= &sysctl_intvec,
533 		.extra1		= &minolduid,
534 		.extra2		= &maxolduid,
535 	},
536 	{
537 		.ctl_name	= KERN_OVERFLOWGID,
538 		.procname	= "overflowgid",
539 		.data		= &overflowgid,
540 		.maxlen		= sizeof(int),
541 		.mode		= 0644,
542 		.proc_handler	= &proc_dointvec_minmax,
543 		.strategy	= &sysctl_intvec,
544 		.extra1		= &minolduid,
545 		.extra2		= &maxolduid,
546 	},
547 #ifdef CONFIG_ARCH_S390
548 #ifdef CONFIG_MATHEMU
549 	{
550 		.ctl_name	= KERN_IEEE_EMULATION_WARNINGS,
551 		.procname	= "ieee_emulation_warnings",
552 		.data		= &sysctl_ieee_emulation_warnings,
553 		.maxlen		= sizeof(int),
554 		.mode		= 0644,
555 		.proc_handler	= &proc_dointvec,
556 	},
557 #endif
558 #ifdef CONFIG_NO_IDLE_HZ
559 	{
560 		.ctl_name       = KERN_HZ_TIMER,
561 		.procname       = "hz_timer",
562 		.data           = &sysctl_hz_timer,
563 		.maxlen         = sizeof(int),
564 		.mode           = 0644,
565 		.proc_handler   = &proc_dointvec,
566 	},
567 #endif
568 	{
569 		.ctl_name	= KERN_S390_USER_DEBUG_LOGGING,
570 		.procname	= "userprocess_debug",
571 		.data		= &sysctl_userprocess_debug,
572 		.maxlen		= sizeof(int),
573 		.mode		= 0644,
574 		.proc_handler	= &proc_dointvec,
575 	},
576 #endif
577 	{
578 		.ctl_name	= KERN_PIDMAX,
579 		.procname	= "pid_max",
580 		.data		= &pid_max,
581 		.maxlen		= sizeof (int),
582 		.mode		= 0644,
583 		.proc_handler	= &proc_dointvec_minmax,
584 		.strategy	= sysctl_intvec,
585 		.extra1		= &pid_max_min,
586 		.extra2		= &pid_max_max,
587 	},
588 	{
589 		.ctl_name	= KERN_PANIC_ON_OOPS,
590 		.procname	= "panic_on_oops",
591 		.data		= &panic_on_oops,
592 		.maxlen		= sizeof(int),
593 		.mode		= 0644,
594 		.proc_handler	= &proc_dointvec,
595 	},
596 	{
597 		.ctl_name	= KERN_PRINTK_RATELIMIT,
598 		.procname	= "printk_ratelimit",
599 		.data		= &printk_ratelimit_jiffies,
600 		.maxlen		= sizeof(int),
601 		.mode		= 0644,
602 		.proc_handler	= &proc_dointvec_jiffies,
603 		.strategy	= &sysctl_jiffies,
604 	},
605 	{
606 		.ctl_name	= KERN_PRINTK_RATELIMIT_BURST,
607 		.procname	= "printk_ratelimit_burst",
608 		.data		= &printk_ratelimit_burst,
609 		.maxlen		= sizeof(int),
610 		.mode		= 0644,
611 		.proc_handler	= &proc_dointvec,
612 	},
613 	{
614 		.ctl_name	= KERN_NGROUPS_MAX,
615 		.procname	= "ngroups_max",
616 		.data		= &ngroups_max,
617 		.maxlen		= sizeof (int),
618 		.mode		= 0444,
619 		.proc_handler	= &proc_dointvec,
620 	},
621 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
622 	{
623 		.ctl_name       = KERN_UNKNOWN_NMI_PANIC,
624 		.procname       = "unknown_nmi_panic",
625 		.data           = &unknown_nmi_panic,
626 		.maxlen         = sizeof (int),
627 		.mode           = 0644,
628 		.proc_handler   = &proc_unknown_nmi_panic,
629 	},
630 #endif
631 #if defined(CONFIG_X86)
632 	{
633 		.ctl_name	= KERN_BOOTLOADER_TYPE,
634 		.procname	= "bootloader_type",
635 		.data		= &bootloader_type,
636 		.maxlen		= sizeof (int),
637 		.mode		= 0444,
638 		.proc_handler	= &proc_dointvec,
639 	},
640 #endif
641 	{
642 		.ctl_name	= KERN_RANDOMIZE,
643 		.procname	= "randomize_va_space",
644 		.data		= &randomize_va_space,
645 		.maxlen		= sizeof(int),
646 		.mode		= 0644,
647 		.proc_handler	= &proc_dointvec,
648 	},
649 #if defined(CONFIG_ARCH_S390)
650 	{
651 		.ctl_name	= KERN_SPIN_RETRY,
652 		.procname	= "spin_retry",
653 		.data		= &spin_retry,
654 		.maxlen		= sizeof (int),
655 		.mode		= 0644,
656 		.proc_handler	= &proc_dointvec,
657 	},
658 #endif
659 	{ .ctl_name = 0 }
660 };
661 
662 /* Constants for minimum and maximum testing in vm_table.
663    We use these as one-element integer vectors. */
664 static int zero;
665 static int one_hundred = 100;
666 
667 
668 static ctl_table vm_table[] = {
669 	{
670 		.ctl_name	= VM_OVERCOMMIT_MEMORY,
671 		.procname	= "overcommit_memory",
672 		.data		= &sysctl_overcommit_memory,
673 		.maxlen		= sizeof(sysctl_overcommit_memory),
674 		.mode		= 0644,
675 		.proc_handler	= &proc_dointvec,
676 	},
677 	{
678 		.ctl_name	= VM_OVERCOMMIT_RATIO,
679 		.procname	= "overcommit_ratio",
680 		.data		= &sysctl_overcommit_ratio,
681 		.maxlen		= sizeof(sysctl_overcommit_ratio),
682 		.mode		= 0644,
683 		.proc_handler	= &proc_dointvec,
684 	},
685 	{
686 		.ctl_name	= VM_PAGE_CLUSTER,
687 		.procname	= "page-cluster",
688 		.data		= &page_cluster,
689 		.maxlen		= sizeof(int),
690 		.mode		= 0644,
691 		.proc_handler	= &proc_dointvec,
692 	},
693 	{
694 		.ctl_name	= VM_DIRTY_BACKGROUND,
695 		.procname	= "dirty_background_ratio",
696 		.data		= &dirty_background_ratio,
697 		.maxlen		= sizeof(dirty_background_ratio),
698 		.mode		= 0644,
699 		.proc_handler	= &proc_dointvec_minmax,
700 		.strategy	= &sysctl_intvec,
701 		.extra1		= &zero,
702 		.extra2		= &one_hundred,
703 	},
704 	{
705 		.ctl_name	= VM_DIRTY_RATIO,
706 		.procname	= "dirty_ratio",
707 		.data		= &vm_dirty_ratio,
708 		.maxlen		= sizeof(vm_dirty_ratio),
709 		.mode		= 0644,
710 		.proc_handler	= &proc_dointvec_minmax,
711 		.strategy	= &sysctl_intvec,
712 		.extra1		= &zero,
713 		.extra2		= &one_hundred,
714 	},
715 	{
716 		.ctl_name	= VM_DIRTY_WB_CS,
717 		.procname	= "dirty_writeback_centisecs",
718 		.data		= &dirty_writeback_centisecs,
719 		.maxlen		= sizeof(dirty_writeback_centisecs),
720 		.mode		= 0644,
721 		.proc_handler	= &dirty_writeback_centisecs_handler,
722 	},
723 	{
724 		.ctl_name	= VM_DIRTY_EXPIRE_CS,
725 		.procname	= "dirty_expire_centisecs",
726 		.data		= &dirty_expire_centisecs,
727 		.maxlen		= sizeof(dirty_expire_centisecs),
728 		.mode		= 0644,
729 		.proc_handler	= &proc_dointvec,
730 	},
731 	{
732 		.ctl_name	= VM_NR_PDFLUSH_THREADS,
733 		.procname	= "nr_pdflush_threads",
734 		.data		= &nr_pdflush_threads,
735 		.maxlen		= sizeof nr_pdflush_threads,
736 		.mode		= 0444 /* read-only*/,
737 		.proc_handler	= &proc_dointvec,
738 	},
739 	{
740 		.ctl_name	= VM_SWAPPINESS,
741 		.procname	= "swappiness",
742 		.data		= &vm_swappiness,
743 		.maxlen		= sizeof(vm_swappiness),
744 		.mode		= 0644,
745 		.proc_handler	= &proc_dointvec_minmax,
746 		.strategy	= &sysctl_intvec,
747 		.extra1		= &zero,
748 		.extra2		= &one_hundred,
749 	},
750 #ifdef CONFIG_HUGETLB_PAGE
751 	 {
752 		.ctl_name	= VM_HUGETLB_PAGES,
753 		.procname	= "nr_hugepages",
754 		.data		= &max_huge_pages,
755 		.maxlen		= sizeof(unsigned long),
756 		.mode		= 0644,
757 		.proc_handler	= &hugetlb_sysctl_handler,
758 		.extra1		= (void *)&hugetlb_zero,
759 		.extra2		= (void *)&hugetlb_infinity,
760 	 },
761 	 {
762 		.ctl_name	= VM_HUGETLB_GROUP,
763 		.procname	= "hugetlb_shm_group",
764 		.data		= &sysctl_hugetlb_shm_group,
765 		.maxlen		= sizeof(gid_t),
766 		.mode		= 0644,
767 		.proc_handler	= &proc_dointvec,
768 	 },
769 #endif
770 	{
771 		.ctl_name	= VM_LOWMEM_RESERVE_RATIO,
772 		.procname	= "lowmem_reserve_ratio",
773 		.data		= &sysctl_lowmem_reserve_ratio,
774 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
775 		.mode		= 0644,
776 		.proc_handler	= &lowmem_reserve_ratio_sysctl_handler,
777 		.strategy	= &sysctl_intvec,
778 	},
779 	{
780 		.ctl_name	= VM_MIN_FREE_KBYTES,
781 		.procname	= "min_free_kbytes",
782 		.data		= &min_free_kbytes,
783 		.maxlen		= sizeof(min_free_kbytes),
784 		.mode		= 0644,
785 		.proc_handler	= &min_free_kbytes_sysctl_handler,
786 		.strategy	= &sysctl_intvec,
787 		.extra1		= &zero,
788 	},
789 #ifdef CONFIG_MMU
790 	{
791 		.ctl_name	= VM_MAX_MAP_COUNT,
792 		.procname	= "max_map_count",
793 		.data		= &sysctl_max_map_count,
794 		.maxlen		= sizeof(sysctl_max_map_count),
795 		.mode		= 0644,
796 		.proc_handler	= &proc_dointvec
797 	},
798 #endif
799 	{
800 		.ctl_name	= VM_LAPTOP_MODE,
801 		.procname	= "laptop_mode",
802 		.data		= &laptop_mode,
803 		.maxlen		= sizeof(laptop_mode),
804 		.mode		= 0644,
805 		.proc_handler	= &proc_dointvec,
806 		.strategy	= &sysctl_intvec,
807 		.extra1		= &zero,
808 	},
809 	{
810 		.ctl_name	= VM_BLOCK_DUMP,
811 		.procname	= "block_dump",
812 		.data		= &block_dump,
813 		.maxlen		= sizeof(block_dump),
814 		.mode		= 0644,
815 		.proc_handler	= &proc_dointvec,
816 		.strategy	= &sysctl_intvec,
817 		.extra1		= &zero,
818 	},
819 	{
820 		.ctl_name	= VM_VFS_CACHE_PRESSURE,
821 		.procname	= "vfs_cache_pressure",
822 		.data		= &sysctl_vfs_cache_pressure,
823 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
824 		.mode		= 0644,
825 		.proc_handler	= &proc_dointvec,
826 		.strategy	= &sysctl_intvec,
827 		.extra1		= &zero,
828 	},
829 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
830 	{
831 		.ctl_name	= VM_LEGACY_VA_LAYOUT,
832 		.procname	= "legacy_va_layout",
833 		.data		= &sysctl_legacy_va_layout,
834 		.maxlen		= sizeof(sysctl_legacy_va_layout),
835 		.mode		= 0644,
836 		.proc_handler	= &proc_dointvec,
837 		.strategy	= &sysctl_intvec,
838 		.extra1		= &zero,
839 	},
840 #endif
841 #ifdef CONFIG_SWAP
842 	{
843 		.ctl_name	= VM_SWAP_TOKEN_TIMEOUT,
844 		.procname	= "swap_token_timeout",
845 		.data		= &swap_token_default_timeout,
846 		.maxlen		= sizeof(swap_token_default_timeout),
847 		.mode		= 0644,
848 		.proc_handler	= &proc_dointvec_jiffies,
849 		.strategy	= &sysctl_jiffies,
850 	},
851 #endif
852 	{ .ctl_name = 0 }
853 };
854 
855 static ctl_table proc_table[] = {
856 	{ .ctl_name = 0 }
857 };
858 
859 static ctl_table fs_table[] = {
860 	{
861 		.ctl_name	= FS_NRINODE,
862 		.procname	= "inode-nr",
863 		.data		= &inodes_stat,
864 		.maxlen		= 2*sizeof(int),
865 		.mode		= 0444,
866 		.proc_handler	= &proc_dointvec,
867 	},
868 	{
869 		.ctl_name	= FS_STATINODE,
870 		.procname	= "inode-state",
871 		.data		= &inodes_stat,
872 		.maxlen		= 7*sizeof(int),
873 		.mode		= 0444,
874 		.proc_handler	= &proc_dointvec,
875 	},
876 	{
877 		.ctl_name	= FS_NRFILE,
878 		.procname	= "file-nr",
879 		.data		= &files_stat,
880 		.maxlen		= 3*sizeof(int),
881 		.mode		= 0444,
882 		.proc_handler	= &proc_dointvec,
883 	},
884 	{
885 		.ctl_name	= FS_MAXFILE,
886 		.procname	= "file-max",
887 		.data		= &files_stat.max_files,
888 		.maxlen		= sizeof(int),
889 		.mode		= 0644,
890 		.proc_handler	= &proc_dointvec,
891 	},
892 	{
893 		.ctl_name	= FS_DENTRY,
894 		.procname	= "dentry-state",
895 		.data		= &dentry_stat,
896 		.maxlen		= 6*sizeof(int),
897 		.mode		= 0444,
898 		.proc_handler	= &proc_dointvec,
899 	},
900 	{
901 		.ctl_name	= FS_OVERFLOWUID,
902 		.procname	= "overflowuid",
903 		.data		= &fs_overflowuid,
904 		.maxlen		= sizeof(int),
905 		.mode		= 0644,
906 		.proc_handler	= &proc_dointvec_minmax,
907 		.strategy	= &sysctl_intvec,
908 		.extra1		= &minolduid,
909 		.extra2		= &maxolduid,
910 	},
911 	{
912 		.ctl_name	= FS_OVERFLOWGID,
913 		.procname	= "overflowgid",
914 		.data		= &fs_overflowgid,
915 		.maxlen		= sizeof(int),
916 		.mode		= 0644,
917 		.proc_handler	= &proc_dointvec_minmax,
918 		.strategy	= &sysctl_intvec,
919 		.extra1		= &minolduid,
920 		.extra2		= &maxolduid,
921 	},
922 	{
923 		.ctl_name	= FS_LEASES,
924 		.procname	= "leases-enable",
925 		.data		= &leases_enable,
926 		.maxlen		= sizeof(int),
927 		.mode		= 0644,
928 		.proc_handler	= &proc_dointvec,
929 	},
930 #ifdef CONFIG_DNOTIFY
931 	{
932 		.ctl_name	= FS_DIR_NOTIFY,
933 		.procname	= "dir-notify-enable",
934 		.data		= &dir_notify_enable,
935 		.maxlen		= sizeof(int),
936 		.mode		= 0644,
937 		.proc_handler	= &proc_dointvec,
938 	},
939 #endif
940 #ifdef CONFIG_MMU
941 	{
942 		.ctl_name	= FS_LEASE_TIME,
943 		.procname	= "lease-break-time",
944 		.data		= &lease_break_time,
945 		.maxlen		= sizeof(int),
946 		.mode		= 0644,
947 		.proc_handler	= &proc_dointvec,
948 	},
949 	{
950 		.ctl_name	= FS_AIO_NR,
951 		.procname	= "aio-nr",
952 		.data		= &aio_nr,
953 		.maxlen		= sizeof(aio_nr),
954 		.mode		= 0444,
955 		.proc_handler	= &proc_doulongvec_minmax,
956 	},
957 	{
958 		.ctl_name	= FS_AIO_MAX_NR,
959 		.procname	= "aio-max-nr",
960 		.data		= &aio_max_nr,
961 		.maxlen		= sizeof(aio_max_nr),
962 		.mode		= 0644,
963 		.proc_handler	= &proc_doulongvec_minmax,
964 	},
965 #ifdef CONFIG_INOTIFY
966 	{
967 		.ctl_name	= FS_INOTIFY,
968 		.procname	= "inotify",
969 		.mode		= 0555,
970 		.child		= inotify_table,
971 	},
972 #endif
973 #endif
974 	{
975 		.ctl_name	= KERN_SETUID_DUMPABLE,
976 		.procname	= "suid_dumpable",
977 		.data		= &suid_dumpable,
978 		.maxlen		= sizeof(int),
979 		.mode		= 0644,
980 		.proc_handler	= &proc_dointvec,
981 	},
982 	{ .ctl_name = 0 }
983 };
984 
985 static ctl_table debug_table[] = {
986 	{ .ctl_name = 0 }
987 };
988 
989 static ctl_table dev_table[] = {
990 	{ .ctl_name = 0 }
991 };
992 
993 extern void init_irq_proc (void);
994 
995 static DEFINE_SPINLOCK(sysctl_lock);
996 
997 /* called under sysctl_lock */
998 static int use_table(struct ctl_table_header *p)
999 {
1000 	if (unlikely(p->unregistering))
1001 		return 0;
1002 	p->used++;
1003 	return 1;
1004 }
1005 
1006 /* called under sysctl_lock */
1007 static void unuse_table(struct ctl_table_header *p)
1008 {
1009 	if (!--p->used)
1010 		if (unlikely(p->unregistering))
1011 			complete(p->unregistering);
1012 }
1013 
1014 /* called under sysctl_lock, will reacquire if has to wait */
1015 static void start_unregistering(struct ctl_table_header *p)
1016 {
1017 	/*
1018 	 * if p->used is 0, nobody will ever touch that entry again;
1019 	 * we'll eliminate all paths to it before dropping sysctl_lock
1020 	 */
1021 	if (unlikely(p->used)) {
1022 		struct completion wait;
1023 		init_completion(&wait);
1024 		p->unregistering = &wait;
1025 		spin_unlock(&sysctl_lock);
1026 		wait_for_completion(&wait);
1027 		spin_lock(&sysctl_lock);
1028 	}
1029 	/*
1030 	 * do not remove from the list until nobody holds it; walking the
1031 	 * list in do_sysctl() relies on that.
1032 	 */
1033 	list_del_init(&p->ctl_entry);
1034 }
1035 
1036 void __init sysctl_init(void)
1037 {
1038 #ifdef CONFIG_PROC_FS
1039 	register_proc_table(root_table, proc_sys_root, &root_table_header);
1040 	init_irq_proc();
1041 #endif
1042 }
1043 
1044 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1045 	       void __user *newval, size_t newlen)
1046 {
1047 	struct list_head *tmp;
1048 	int error = -ENOTDIR;
1049 
1050 	if (nlen <= 0 || nlen >= CTL_MAXNAME)
1051 		return -ENOTDIR;
1052 	if (oldval) {
1053 		int old_len;
1054 		if (!oldlenp || get_user(old_len, oldlenp))
1055 			return -EFAULT;
1056 	}
1057 	spin_lock(&sysctl_lock);
1058 	tmp = &root_table_header.ctl_entry;
1059 	do {
1060 		struct ctl_table_header *head =
1061 			list_entry(tmp, struct ctl_table_header, ctl_entry);
1062 		void *context = NULL;
1063 
1064 		if (!use_table(head))
1065 			continue;
1066 
1067 		spin_unlock(&sysctl_lock);
1068 
1069 		error = parse_table(name, nlen, oldval, oldlenp,
1070 					newval, newlen, head->ctl_table,
1071 					&context);
1072 		kfree(context);
1073 
1074 		spin_lock(&sysctl_lock);
1075 		unuse_table(head);
1076 		if (error != -ENOTDIR)
1077 			break;
1078 	} while ((tmp = tmp->next) != &root_table_header.ctl_entry);
1079 	spin_unlock(&sysctl_lock);
1080 	return error;
1081 }
1082 
1083 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1084 {
1085 	struct __sysctl_args tmp;
1086 	int error;
1087 
1088 	if (copy_from_user(&tmp, args, sizeof(tmp)))
1089 		return -EFAULT;
1090 
1091 	lock_kernel();
1092 	error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1093 			  tmp.newval, tmp.newlen);
1094 	unlock_kernel();
1095 	return error;
1096 }
1097 
1098 /*
1099  * ctl_perm does NOT grant the superuser all rights automatically, because
1100  * some sysctl variables are readonly even to root.
1101  */
1102 
1103 static int test_perm(int mode, int op)
1104 {
1105 	if (!current->euid)
1106 		mode >>= 6;
1107 	else if (in_egroup_p(0))
1108 		mode >>= 3;
1109 	if ((mode & op & 0007) == op)
1110 		return 0;
1111 	return -EACCES;
1112 }
1113 
1114 static inline int ctl_perm(ctl_table *table, int op)
1115 {
1116 	int error;
1117 	error = security_sysctl(table, op);
1118 	if (error)
1119 		return error;
1120 	return test_perm(table->mode, op);
1121 }
1122 
1123 static int parse_table(int __user *name, int nlen,
1124 		       void __user *oldval, size_t __user *oldlenp,
1125 		       void __user *newval, size_t newlen,
1126 		       ctl_table *table, void **context)
1127 {
1128 	int n;
1129 repeat:
1130 	if (!nlen)
1131 		return -ENOTDIR;
1132 	if (get_user(n, name))
1133 		return -EFAULT;
1134 	for ( ; table->ctl_name; table++) {
1135 		if (n == table->ctl_name || table->ctl_name == CTL_ANY) {
1136 			int error;
1137 			if (table->child) {
1138 				if (ctl_perm(table, 001))
1139 					return -EPERM;
1140 				if (table->strategy) {
1141 					error = table->strategy(
1142 						table, name, nlen,
1143 						oldval, oldlenp,
1144 						newval, newlen, context);
1145 					if (error)
1146 						return error;
1147 				}
1148 				name++;
1149 				nlen--;
1150 				table = table->child;
1151 				goto repeat;
1152 			}
1153 			error = do_sysctl_strategy(table, name, nlen,
1154 						   oldval, oldlenp,
1155 						   newval, newlen, context);
1156 			return error;
1157 		}
1158 	}
1159 	return -ENOTDIR;
1160 }
1161 
1162 /* Perform the actual read/write of a sysctl table entry. */
1163 int do_sysctl_strategy (ctl_table *table,
1164 			int __user *name, int nlen,
1165 			void __user *oldval, size_t __user *oldlenp,
1166 			void __user *newval, size_t newlen, void **context)
1167 {
1168 	int op = 0, rc;
1169 	size_t len;
1170 
1171 	if (oldval)
1172 		op |= 004;
1173 	if (newval)
1174 		op |= 002;
1175 	if (ctl_perm(table, op))
1176 		return -EPERM;
1177 
1178 	if (table->strategy) {
1179 		rc = table->strategy(table, name, nlen, oldval, oldlenp,
1180 				     newval, newlen, context);
1181 		if (rc < 0)
1182 			return rc;
1183 		if (rc > 0)
1184 			return 0;
1185 	}
1186 
1187 	/* If there is no strategy routine, or if the strategy returns
1188 	 * zero, proceed with automatic r/w */
1189 	if (table->data && table->maxlen) {
1190 		if (oldval && oldlenp) {
1191 			if (get_user(len, oldlenp))
1192 				return -EFAULT;
1193 			if (len) {
1194 				if (len > table->maxlen)
1195 					len = table->maxlen;
1196 				if(copy_to_user(oldval, table->data, len))
1197 					return -EFAULT;
1198 				if(put_user(len, oldlenp))
1199 					return -EFAULT;
1200 			}
1201 		}
1202 		if (newval && newlen) {
1203 			len = newlen;
1204 			if (len > table->maxlen)
1205 				len = table->maxlen;
1206 			if(copy_from_user(table->data, newval, len))
1207 				return -EFAULT;
1208 		}
1209 	}
1210 	return 0;
1211 }
1212 
1213 /**
1214  * register_sysctl_table - register a sysctl hierarchy
1215  * @table: the top-level table structure
1216  * @insert_at_head: whether the entry should be inserted in front or at the end
1217  *
1218  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1219  * array. An entry with a ctl_name of 0 terminates the table.
1220  *
1221  * The members of the &ctl_table structure are used as follows:
1222  *
1223  * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1224  *            must be unique within that level of sysctl
1225  *
1226  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1227  *            enter a sysctl file
1228  *
1229  * data - a pointer to data for use by proc_handler
1230  *
1231  * maxlen - the maximum size in bytes of the data
1232  *
1233  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1234  *
1235  * child - a pointer to the child sysctl table if this entry is a directory, or
1236  *         %NULL.
1237  *
1238  * proc_handler - the text handler routine (described below)
1239  *
1240  * strategy - the strategy routine (described below)
1241  *
1242  * de - for internal use by the sysctl routines
1243  *
1244  * extra1, extra2 - extra pointers usable by the proc handler routines
1245  *
1246  * Leaf nodes in the sysctl tree will be represented by a single file
1247  * under /proc; non-leaf nodes will be represented by directories.
1248  *
1249  * sysctl(2) can automatically manage read and write requests through
1250  * the sysctl table.  The data and maxlen fields of the ctl_table
1251  * struct enable minimal validation of the values being written to be
1252  * performed, and the mode field allows minimal authentication.
1253  *
1254  * More sophisticated management can be enabled by the provision of a
1255  * strategy routine with the table entry.  This will be called before
1256  * any automatic read or write of the data is performed.
1257  *
1258  * The strategy routine may return
1259  *
1260  * < 0 - Error occurred (error is passed to user process)
1261  *
1262  * 0   - OK - proceed with automatic read or write.
1263  *
1264  * > 0 - OK - read or write has been done by the strategy routine, so
1265  *       return immediately.
1266  *
1267  * There must be a proc_handler routine for any terminal nodes
1268  * mirrored under /proc/sys (non-terminals are handled by a built-in
1269  * directory handler).  Several default handlers are available to
1270  * cover common cases -
1271  *
1272  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1273  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1274  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1275  *
1276  * It is the handler's job to read the input buffer from user memory
1277  * and process it. The handler should return 0 on success.
1278  *
1279  * This routine returns %NULL on a failure to register, and a pointer
1280  * to the table header on success.
1281  */
1282 struct ctl_table_header *register_sysctl_table(ctl_table * table,
1283 					       int insert_at_head)
1284 {
1285 	struct ctl_table_header *tmp;
1286 	tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1287 	if (!tmp)
1288 		return NULL;
1289 	tmp->ctl_table = table;
1290 	INIT_LIST_HEAD(&tmp->ctl_entry);
1291 	tmp->used = 0;
1292 	tmp->unregistering = NULL;
1293 	spin_lock(&sysctl_lock);
1294 	if (insert_at_head)
1295 		list_add(&tmp->ctl_entry, &root_table_header.ctl_entry);
1296 	else
1297 		list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
1298 	spin_unlock(&sysctl_lock);
1299 #ifdef CONFIG_PROC_FS
1300 	register_proc_table(table, proc_sys_root, tmp);
1301 #endif
1302 	return tmp;
1303 }
1304 
1305 /**
1306  * unregister_sysctl_table - unregister a sysctl table hierarchy
1307  * @header: the header returned from register_sysctl_table
1308  *
1309  * Unregisters the sysctl table and all children. proc entries may not
1310  * actually be removed until they are no longer used by anyone.
1311  */
1312 void unregister_sysctl_table(struct ctl_table_header * header)
1313 {
1314 	might_sleep();
1315 	spin_lock(&sysctl_lock);
1316 	start_unregistering(header);
1317 #ifdef CONFIG_PROC_FS
1318 	unregister_proc_table(header->ctl_table, proc_sys_root);
1319 #endif
1320 	spin_unlock(&sysctl_lock);
1321 	kfree(header);
1322 }
1323 
1324 /*
1325  * /proc/sys support
1326  */
1327 
1328 #ifdef CONFIG_PROC_FS
1329 
1330 /* Scan the sysctl entries in table and add them all into /proc */
1331 static void register_proc_table(ctl_table * table, struct proc_dir_entry *root, void *set)
1332 {
1333 	struct proc_dir_entry *de;
1334 	int len;
1335 	mode_t mode;
1336 
1337 	for (; table->ctl_name; table++) {
1338 		/* Can't do anything without a proc name. */
1339 		if (!table->procname)
1340 			continue;
1341 		/* Maybe we can't do anything with it... */
1342 		if (!table->proc_handler && !table->child) {
1343 			printk(KERN_WARNING "SYSCTL: Can't register %s\n",
1344 				table->procname);
1345 			continue;
1346 		}
1347 
1348 		len = strlen(table->procname);
1349 		mode = table->mode;
1350 
1351 		de = NULL;
1352 		if (table->proc_handler)
1353 			mode |= S_IFREG;
1354 		else {
1355 			mode |= S_IFDIR;
1356 			for (de = root->subdir; de; de = de->next) {
1357 				if (proc_match(len, table->procname, de))
1358 					break;
1359 			}
1360 			/* If the subdir exists already, de is non-NULL */
1361 		}
1362 
1363 		if (!de) {
1364 			de = create_proc_entry(table->procname, mode, root);
1365 			if (!de)
1366 				continue;
1367 			de->set = set;
1368 			de->data = (void *) table;
1369 			if (table->proc_handler)
1370 				de->proc_fops = &proc_sys_file_operations;
1371 		}
1372 		table->de = de;
1373 		if (de->mode & S_IFDIR)
1374 			register_proc_table(table->child, de, set);
1375 	}
1376 }
1377 
1378 /*
1379  * Unregister a /proc sysctl table and any subdirectories.
1380  */
1381 static void unregister_proc_table(ctl_table * table, struct proc_dir_entry *root)
1382 {
1383 	struct proc_dir_entry *de;
1384 	for (; table->ctl_name; table++) {
1385 		if (!(de = table->de))
1386 			continue;
1387 		if (de->mode & S_IFDIR) {
1388 			if (!table->child) {
1389 				printk (KERN_ALERT "Help - malformed sysctl tree on free\n");
1390 				continue;
1391 			}
1392 			unregister_proc_table(table->child, de);
1393 
1394 			/* Don't unregister directories which still have entries.. */
1395 			if (de->subdir)
1396 				continue;
1397 		}
1398 
1399 		/*
1400 		 * In any case, mark the entry as goner; we'll keep it
1401 		 * around if it's busy, but we'll know to do nothing with
1402 		 * its fields.  We are under sysctl_lock here.
1403 		 */
1404 		de->data = NULL;
1405 
1406 		/* Don't unregister proc entries that are still being used.. */
1407 		if (atomic_read(&de->count))
1408 			continue;
1409 
1410 		table->de = NULL;
1411 		remove_proc_entry(table->procname, root);
1412 	}
1413 }
1414 
1415 static ssize_t do_rw_proc(int write, struct file * file, char __user * buf,
1416 			  size_t count, loff_t *ppos)
1417 {
1418 	int op;
1419 	struct proc_dir_entry *de = PDE(file->f_dentry->d_inode);
1420 	struct ctl_table *table;
1421 	size_t res;
1422 	ssize_t error = -ENOTDIR;
1423 
1424 	spin_lock(&sysctl_lock);
1425 	if (de && de->data && use_table(de->set)) {
1426 		/*
1427 		 * at that point we know that sysctl was not unregistered
1428 		 * and won't be until we finish
1429 		 */
1430 		spin_unlock(&sysctl_lock);
1431 		table = (struct ctl_table *) de->data;
1432 		if (!table || !table->proc_handler)
1433 			goto out;
1434 		error = -EPERM;
1435 		op = (write ? 002 : 004);
1436 		if (ctl_perm(table, op))
1437 			goto out;
1438 
1439 		/* careful: calling conventions are nasty here */
1440 		res = count;
1441 		error = (*table->proc_handler)(table, write, file,
1442 						buf, &res, ppos);
1443 		if (!error)
1444 			error = res;
1445 	out:
1446 		spin_lock(&sysctl_lock);
1447 		unuse_table(de->set);
1448 	}
1449 	spin_unlock(&sysctl_lock);
1450 	return error;
1451 }
1452 
1453 static int proc_opensys(struct inode *inode, struct file *file)
1454 {
1455 	if (file->f_mode & FMODE_WRITE) {
1456 		/*
1457 		 * sysctl entries that are not writable,
1458 		 * are _NOT_ writable, capabilities or not.
1459 		 */
1460 		if (!(inode->i_mode & S_IWUSR))
1461 			return -EPERM;
1462 	}
1463 
1464 	return 0;
1465 }
1466 
1467 static ssize_t proc_readsys(struct file * file, char __user * buf,
1468 			    size_t count, loff_t *ppos)
1469 {
1470 	return do_rw_proc(0, file, buf, count, ppos);
1471 }
1472 
1473 static ssize_t proc_writesys(struct file * file, const char __user * buf,
1474 			     size_t count, loff_t *ppos)
1475 {
1476 	return do_rw_proc(1, file, (char __user *) buf, count, ppos);
1477 }
1478 
1479 /**
1480  * proc_dostring - read a string sysctl
1481  * @table: the sysctl table
1482  * @write: %TRUE if this is a write to the sysctl file
1483  * @filp: the file structure
1484  * @buffer: the user buffer
1485  * @lenp: the size of the user buffer
1486  * @ppos: file position
1487  *
1488  * Reads/writes a string from/to the user buffer. If the kernel
1489  * buffer provided is not large enough to hold the string, the
1490  * string is truncated. The copied string is %NULL-terminated.
1491  * If the string is being read by the user process, it is copied
1492  * and a newline '\n' is added. It is truncated if the buffer is
1493  * not large enough.
1494  *
1495  * Returns 0 on success.
1496  */
1497 int proc_dostring(ctl_table *table, int write, struct file *filp,
1498 		  void __user *buffer, size_t *lenp, loff_t *ppos)
1499 {
1500 	size_t len;
1501 	char __user *p;
1502 	char c;
1503 
1504 	if (!table->data || !table->maxlen || !*lenp ||
1505 	    (*ppos && !write)) {
1506 		*lenp = 0;
1507 		return 0;
1508 	}
1509 
1510 	if (write) {
1511 		len = 0;
1512 		p = buffer;
1513 		while (len < *lenp) {
1514 			if (get_user(c, p++))
1515 				return -EFAULT;
1516 			if (c == 0 || c == '\n')
1517 				break;
1518 			len++;
1519 		}
1520 		if (len >= table->maxlen)
1521 			len = table->maxlen-1;
1522 		if(copy_from_user(table->data, buffer, len))
1523 			return -EFAULT;
1524 		((char *) table->data)[len] = 0;
1525 		*ppos += *lenp;
1526 	} else {
1527 		len = strlen(table->data);
1528 		if (len > table->maxlen)
1529 			len = table->maxlen;
1530 		if (len > *lenp)
1531 			len = *lenp;
1532 		if (len)
1533 			if(copy_to_user(buffer, table->data, len))
1534 				return -EFAULT;
1535 		if (len < *lenp) {
1536 			if(put_user('\n', ((char __user *) buffer) + len))
1537 				return -EFAULT;
1538 			len++;
1539 		}
1540 		*lenp = len;
1541 		*ppos += len;
1542 	}
1543 	return 0;
1544 }
1545 
1546 /*
1547  *	Special case of dostring for the UTS structure. This has locks
1548  *	to observe. Should this be in kernel/sys.c ????
1549  */
1550 
1551 static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
1552 		  void __user *buffer, size_t *lenp, loff_t *ppos)
1553 {
1554 	int r;
1555 
1556 	if (!write) {
1557 		down_read(&uts_sem);
1558 		r=proc_dostring(table,0,filp,buffer,lenp, ppos);
1559 		up_read(&uts_sem);
1560 	} else {
1561 		down_write(&uts_sem);
1562 		r=proc_dostring(table,1,filp,buffer,lenp, ppos);
1563 		up_write(&uts_sem);
1564 	}
1565 	return r;
1566 }
1567 
1568 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1569 				 int *valp,
1570 				 int write, void *data)
1571 {
1572 	if (write) {
1573 		*valp = *negp ? -*lvalp : *lvalp;
1574 	} else {
1575 		int val = *valp;
1576 		if (val < 0) {
1577 			*negp = -1;
1578 			*lvalp = (unsigned long)-val;
1579 		} else {
1580 			*negp = 0;
1581 			*lvalp = (unsigned long)val;
1582 		}
1583 	}
1584 	return 0;
1585 }
1586 
1587 static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1588 		  void __user *buffer, size_t *lenp, loff_t *ppos,
1589 		  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1590 			      int write, void *data),
1591 		  void *data)
1592 {
1593 #define TMPBUFLEN 21
1594 	int *i, vleft, first=1, neg, val;
1595 	unsigned long lval;
1596 	size_t left, len;
1597 
1598 	char buf[TMPBUFLEN], *p;
1599 	char __user *s = buffer;
1600 
1601 	if (!table->data || !table->maxlen || !*lenp ||
1602 	    (*ppos && !write)) {
1603 		*lenp = 0;
1604 		return 0;
1605 	}
1606 
1607 	i = (int *) table->data;
1608 	vleft = table->maxlen / sizeof(*i);
1609 	left = *lenp;
1610 
1611 	if (!conv)
1612 		conv = do_proc_dointvec_conv;
1613 
1614 	for (; left && vleft--; i++, first=0) {
1615 		if (write) {
1616 			while (left) {
1617 				char c;
1618 				if (get_user(c, s))
1619 					return -EFAULT;
1620 				if (!isspace(c))
1621 					break;
1622 				left--;
1623 				s++;
1624 			}
1625 			if (!left)
1626 				break;
1627 			neg = 0;
1628 			len = left;
1629 			if (len > sizeof(buf) - 1)
1630 				len = sizeof(buf) - 1;
1631 			if (copy_from_user(buf, s, len))
1632 				return -EFAULT;
1633 			buf[len] = 0;
1634 			p = buf;
1635 			if (*p == '-' && left > 1) {
1636 				neg = 1;
1637 				left--, p++;
1638 			}
1639 			if (*p < '0' || *p > '9')
1640 				break;
1641 
1642 			lval = simple_strtoul(p, &p, 0);
1643 
1644 			len = p-buf;
1645 			if ((len < left) && *p && !isspace(*p))
1646 				break;
1647 			if (neg)
1648 				val = -val;
1649 			s += len;
1650 			left -= len;
1651 
1652 			if (conv(&neg, &lval, i, 1, data))
1653 				break;
1654 		} else {
1655 			p = buf;
1656 			if (!first)
1657 				*p++ = '\t';
1658 
1659 			if (conv(&neg, &lval, i, 0, data))
1660 				break;
1661 
1662 			sprintf(p, "%s%lu", neg ? "-" : "", lval);
1663 			len = strlen(buf);
1664 			if (len > left)
1665 				len = left;
1666 			if(copy_to_user(s, buf, len))
1667 				return -EFAULT;
1668 			left -= len;
1669 			s += len;
1670 		}
1671 	}
1672 
1673 	if (!write && !first && left) {
1674 		if(put_user('\n', s))
1675 			return -EFAULT;
1676 		left--, s++;
1677 	}
1678 	if (write) {
1679 		while (left) {
1680 			char c;
1681 			if (get_user(c, s++))
1682 				return -EFAULT;
1683 			if (!isspace(c))
1684 				break;
1685 			left--;
1686 		}
1687 	}
1688 	if (write && first)
1689 		return -EINVAL;
1690 	*lenp -= left;
1691 	*ppos += *lenp;
1692 	return 0;
1693 #undef TMPBUFLEN
1694 }
1695 
1696 /**
1697  * proc_dointvec - read a vector of integers
1698  * @table: the sysctl table
1699  * @write: %TRUE if this is a write to the sysctl file
1700  * @filp: the file structure
1701  * @buffer: the user buffer
1702  * @lenp: the size of the user buffer
1703  * @ppos: file position
1704  *
1705  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1706  * values from/to the user buffer, treated as an ASCII string.
1707  *
1708  * Returns 0 on success.
1709  */
1710 int proc_dointvec(ctl_table *table, int write, struct file *filp,
1711 		     void __user *buffer, size_t *lenp, loff_t *ppos)
1712 {
1713     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1714 		    	    NULL,NULL);
1715 }
1716 
1717 #define OP_SET	0
1718 #define OP_AND	1
1719 #define OP_OR	2
1720 #define OP_MAX	3
1721 #define OP_MIN	4
1722 
1723 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1724 				      int *valp,
1725 				      int write, void *data)
1726 {
1727 	int op = *(int *)data;
1728 	if (write) {
1729 		int val = *negp ? -*lvalp : *lvalp;
1730 		switch(op) {
1731 		case OP_SET:	*valp = val; break;
1732 		case OP_AND:	*valp &= val; break;
1733 		case OP_OR:	*valp |= val; break;
1734 		case OP_MAX:	if(*valp < val)
1735 					*valp = val;
1736 				break;
1737 		case OP_MIN:	if(*valp > val)
1738 				*valp = val;
1739 				break;
1740 		}
1741 	} else {
1742 		int val = *valp;
1743 		if (val < 0) {
1744 			*negp = -1;
1745 			*lvalp = (unsigned long)-val;
1746 		} else {
1747 			*negp = 0;
1748 			*lvalp = (unsigned long)val;
1749 		}
1750 	}
1751 	return 0;
1752 }
1753 
1754 /*
1755  *	init may raise the set.
1756  */
1757 
1758 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1759 			void __user *buffer, size_t *lenp, loff_t *ppos)
1760 {
1761 	int op;
1762 
1763 	if (!capable(CAP_SYS_MODULE)) {
1764 		return -EPERM;
1765 	}
1766 
1767 	op = (current->pid == 1) ? OP_SET : OP_AND;
1768 	return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1769 				do_proc_dointvec_bset_conv,&op);
1770 }
1771 
1772 struct do_proc_dointvec_minmax_conv_param {
1773 	int *min;
1774 	int *max;
1775 };
1776 
1777 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1778 					int *valp,
1779 					int write, void *data)
1780 {
1781 	struct do_proc_dointvec_minmax_conv_param *param = data;
1782 	if (write) {
1783 		int val = *negp ? -*lvalp : *lvalp;
1784 		if ((param->min && *param->min > val) ||
1785 		    (param->max && *param->max < val))
1786 			return -EINVAL;
1787 		*valp = val;
1788 	} else {
1789 		int val = *valp;
1790 		if (val < 0) {
1791 			*negp = -1;
1792 			*lvalp = (unsigned long)-val;
1793 		} else {
1794 			*negp = 0;
1795 			*lvalp = (unsigned long)val;
1796 		}
1797 	}
1798 	return 0;
1799 }
1800 
1801 /**
1802  * proc_dointvec_minmax - read a vector of integers with min/max values
1803  * @table: the sysctl table
1804  * @write: %TRUE if this is a write to the sysctl file
1805  * @filp: the file structure
1806  * @buffer: the user buffer
1807  * @lenp: the size of the user buffer
1808  * @ppos: file position
1809  *
1810  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1811  * values from/to the user buffer, treated as an ASCII string.
1812  *
1813  * This routine will ensure the values are within the range specified by
1814  * table->extra1 (min) and table->extra2 (max).
1815  *
1816  * Returns 0 on success.
1817  */
1818 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1819 		  void __user *buffer, size_t *lenp, loff_t *ppos)
1820 {
1821 	struct do_proc_dointvec_minmax_conv_param param = {
1822 		.min = (int *) table->extra1,
1823 		.max = (int *) table->extra2,
1824 	};
1825 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1826 				do_proc_dointvec_minmax_conv, &param);
1827 }
1828 
1829 static int do_proc_doulongvec_minmax(ctl_table *table, int write,
1830 				     struct file *filp,
1831 				     void __user *buffer,
1832 				     size_t *lenp, loff_t *ppos,
1833 				     unsigned long convmul,
1834 				     unsigned long convdiv)
1835 {
1836 #define TMPBUFLEN 21
1837 	unsigned long *i, *min, *max, val;
1838 	int vleft, first=1, neg;
1839 	size_t len, left;
1840 	char buf[TMPBUFLEN], *p;
1841 	char __user *s = buffer;
1842 
1843 	if (!table->data || !table->maxlen || !*lenp ||
1844 	    (*ppos && !write)) {
1845 		*lenp = 0;
1846 		return 0;
1847 	}
1848 
1849 	i = (unsigned long *) table->data;
1850 	min = (unsigned long *) table->extra1;
1851 	max = (unsigned long *) table->extra2;
1852 	vleft = table->maxlen / sizeof(unsigned long);
1853 	left = *lenp;
1854 
1855 	for (; left && vleft--; i++, min++, max++, first=0) {
1856 		if (write) {
1857 			while (left) {
1858 				char c;
1859 				if (get_user(c, s))
1860 					return -EFAULT;
1861 				if (!isspace(c))
1862 					break;
1863 				left--;
1864 				s++;
1865 			}
1866 			if (!left)
1867 				break;
1868 			neg = 0;
1869 			len = left;
1870 			if (len > TMPBUFLEN-1)
1871 				len = TMPBUFLEN-1;
1872 			if (copy_from_user(buf, s, len))
1873 				return -EFAULT;
1874 			buf[len] = 0;
1875 			p = buf;
1876 			if (*p == '-' && left > 1) {
1877 				neg = 1;
1878 				left--, p++;
1879 			}
1880 			if (*p < '0' || *p > '9')
1881 				break;
1882 			val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
1883 			len = p-buf;
1884 			if ((len < left) && *p && !isspace(*p))
1885 				break;
1886 			if (neg)
1887 				val = -val;
1888 			s += len;
1889 			left -= len;
1890 
1891 			if(neg)
1892 				continue;
1893 			if ((min && val < *min) || (max && val > *max))
1894 				continue;
1895 			*i = val;
1896 		} else {
1897 			p = buf;
1898 			if (!first)
1899 				*p++ = '\t';
1900 			sprintf(p, "%lu", convdiv * (*i) / convmul);
1901 			len = strlen(buf);
1902 			if (len > left)
1903 				len = left;
1904 			if(copy_to_user(s, buf, len))
1905 				return -EFAULT;
1906 			left -= len;
1907 			s += len;
1908 		}
1909 	}
1910 
1911 	if (!write && !first && left) {
1912 		if(put_user('\n', s))
1913 			return -EFAULT;
1914 		left--, s++;
1915 	}
1916 	if (write) {
1917 		while (left) {
1918 			char c;
1919 			if (get_user(c, s++))
1920 				return -EFAULT;
1921 			if (!isspace(c))
1922 				break;
1923 			left--;
1924 		}
1925 	}
1926 	if (write && first)
1927 		return -EINVAL;
1928 	*lenp -= left;
1929 	*ppos += *lenp;
1930 	return 0;
1931 #undef TMPBUFLEN
1932 }
1933 
1934 /**
1935  * proc_doulongvec_minmax - read a vector of long integers with min/max values
1936  * @table: the sysctl table
1937  * @write: %TRUE if this is a write to the sysctl file
1938  * @filp: the file structure
1939  * @buffer: the user buffer
1940  * @lenp: the size of the user buffer
1941  * @ppos: file position
1942  *
1943  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1944  * values from/to the user buffer, treated as an ASCII string.
1945  *
1946  * This routine will ensure the values are within the range specified by
1947  * table->extra1 (min) and table->extra2 (max).
1948  *
1949  * Returns 0 on success.
1950  */
1951 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
1952 			   void __user *buffer, size_t *lenp, loff_t *ppos)
1953 {
1954     return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
1955 }
1956 
1957 /**
1958  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
1959  * @table: the sysctl table
1960  * @write: %TRUE if this is a write to the sysctl file
1961  * @filp: the file structure
1962  * @buffer: the user buffer
1963  * @lenp: the size of the user buffer
1964  * @ppos: file position
1965  *
1966  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1967  * values from/to the user buffer, treated as an ASCII string. The values
1968  * are treated as milliseconds, and converted to jiffies when they are stored.
1969  *
1970  * This routine will ensure the values are within the range specified by
1971  * table->extra1 (min) and table->extra2 (max).
1972  *
1973  * Returns 0 on success.
1974  */
1975 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
1976 				      struct file *filp,
1977 				      void __user *buffer,
1978 				      size_t *lenp, loff_t *ppos)
1979 {
1980     return do_proc_doulongvec_minmax(table, write, filp, buffer,
1981 				     lenp, ppos, HZ, 1000l);
1982 }
1983 
1984 
1985 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
1986 					 int *valp,
1987 					 int write, void *data)
1988 {
1989 	if (write) {
1990 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
1991 	} else {
1992 		int val = *valp;
1993 		unsigned long lval;
1994 		if (val < 0) {
1995 			*negp = -1;
1996 			lval = (unsigned long)-val;
1997 		} else {
1998 			*negp = 0;
1999 			lval = (unsigned long)val;
2000 		}
2001 		*lvalp = lval / HZ;
2002 	}
2003 	return 0;
2004 }
2005 
2006 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2007 						int *valp,
2008 						int write, void *data)
2009 {
2010 	if (write) {
2011 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2012 	} else {
2013 		int val = *valp;
2014 		unsigned long lval;
2015 		if (val < 0) {
2016 			*negp = -1;
2017 			lval = (unsigned long)-val;
2018 		} else {
2019 			*negp = 0;
2020 			lval = (unsigned long)val;
2021 		}
2022 		*lvalp = jiffies_to_clock_t(lval);
2023 	}
2024 	return 0;
2025 }
2026 
2027 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2028 					    int *valp,
2029 					    int write, void *data)
2030 {
2031 	if (write) {
2032 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2033 	} else {
2034 		int val = *valp;
2035 		unsigned long lval;
2036 		if (val < 0) {
2037 			*negp = -1;
2038 			lval = (unsigned long)-val;
2039 		} else {
2040 			*negp = 0;
2041 			lval = (unsigned long)val;
2042 		}
2043 		*lvalp = jiffies_to_msecs(lval);
2044 	}
2045 	return 0;
2046 }
2047 
2048 /**
2049  * proc_dointvec_jiffies - read a vector of integers as seconds
2050  * @table: the sysctl table
2051  * @write: %TRUE if this is a write to the sysctl file
2052  * @filp: the file structure
2053  * @buffer: the user buffer
2054  * @lenp: the size of the user buffer
2055  * @ppos: file position
2056  *
2057  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2058  * values from/to the user buffer, treated as an ASCII string.
2059  * The values read are assumed to be in seconds, and are converted into
2060  * jiffies.
2061  *
2062  * Returns 0 on success.
2063  */
2064 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2065 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2066 {
2067     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2068 		    	    do_proc_dointvec_jiffies_conv,NULL);
2069 }
2070 
2071 /**
2072  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2073  * @table: the sysctl table
2074  * @write: %TRUE if this is a write to the sysctl file
2075  * @filp: the file structure
2076  * @buffer: the user buffer
2077  * @lenp: the size of the user buffer
2078  * @ppos: pointer to the file position
2079  *
2080  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2081  * values from/to the user buffer, treated as an ASCII string.
2082  * The values read are assumed to be in 1/USER_HZ seconds, and
2083  * are converted into jiffies.
2084  *
2085  * Returns 0 on success.
2086  */
2087 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2088 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2089 {
2090     return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2091 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2092 }
2093 
2094 /**
2095  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2096  * @table: the sysctl table
2097  * @write: %TRUE if this is a write to the sysctl file
2098  * @filp: the file structure
2099  * @buffer: the user buffer
2100  * @lenp: the size of the user buffer
2101  * @ppos: file position
2102  * @ppos: the current position in the file
2103  *
2104  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2105  * values from/to the user buffer, treated as an ASCII string.
2106  * The values read are assumed to be in 1/1000 seconds, and
2107  * are converted into jiffies.
2108  *
2109  * Returns 0 on success.
2110  */
2111 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2112 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2113 {
2114 	return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2115 				do_proc_dointvec_ms_jiffies_conv, NULL);
2116 }
2117 
2118 #else /* CONFIG_PROC_FS */
2119 
2120 int proc_dostring(ctl_table *table, int write, struct file *filp,
2121 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2122 {
2123 	return -ENOSYS;
2124 }
2125 
2126 static int proc_doutsstring(ctl_table *table, int write, struct file *filp,
2127 			    void __user *buffer, size_t *lenp, loff_t *ppos)
2128 {
2129 	return -ENOSYS;
2130 }
2131 
2132 int proc_dointvec(ctl_table *table, int write, struct file *filp,
2133 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2134 {
2135 	return -ENOSYS;
2136 }
2137 
2138 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2139 			void __user *buffer, size_t *lenp, loff_t *ppos)
2140 {
2141 	return -ENOSYS;
2142 }
2143 
2144 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2145 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2146 {
2147 	return -ENOSYS;
2148 }
2149 
2150 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2151 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2152 {
2153 	return -ENOSYS;
2154 }
2155 
2156 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2157 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2158 {
2159 	return -ENOSYS;
2160 }
2161 
2162 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2163 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2164 {
2165 	return -ENOSYS;
2166 }
2167 
2168 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2169 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2170 {
2171 	return -ENOSYS;
2172 }
2173 
2174 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2175 				      struct file *filp,
2176 				      void __user *buffer,
2177 				      size_t *lenp, loff_t *ppos)
2178 {
2179     return -ENOSYS;
2180 }
2181 
2182 
2183 #endif /* CONFIG_PROC_FS */
2184 
2185 
2186 /*
2187  * General sysctl support routines
2188  */
2189 
2190 /* The generic string strategy routine: */
2191 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2192 		  void __user *oldval, size_t __user *oldlenp,
2193 		  void __user *newval, size_t newlen, void **context)
2194 {
2195 	size_t l, len;
2196 
2197 	if (!table->data || !table->maxlen)
2198 		return -ENOTDIR;
2199 
2200 	if (oldval && oldlenp) {
2201 		if (get_user(len, oldlenp))
2202 			return -EFAULT;
2203 		if (len) {
2204 			l = strlen(table->data);
2205 			if (len > l) len = l;
2206 			if (len >= table->maxlen)
2207 				len = table->maxlen;
2208 			if(copy_to_user(oldval, table->data, len))
2209 				return -EFAULT;
2210 			if(put_user(0, ((char __user *) oldval) + len))
2211 				return -EFAULT;
2212 			if(put_user(len, oldlenp))
2213 				return -EFAULT;
2214 		}
2215 	}
2216 	if (newval && newlen) {
2217 		len = newlen;
2218 		if (len > table->maxlen)
2219 			len = table->maxlen;
2220 		if(copy_from_user(table->data, newval, len))
2221 			return -EFAULT;
2222 		if (len == table->maxlen)
2223 			len--;
2224 		((char *) table->data)[len] = 0;
2225 	}
2226 	return 0;
2227 }
2228 
2229 /*
2230  * This function makes sure that all of the integers in the vector
2231  * are between the minimum and maximum values given in the arrays
2232  * table->extra1 and table->extra2, respectively.
2233  */
2234 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2235 		void __user *oldval, size_t __user *oldlenp,
2236 		void __user *newval, size_t newlen, void **context)
2237 {
2238 
2239 	if (newval && newlen) {
2240 		int __user *vec = (int __user *) newval;
2241 		int *min = (int *) table->extra1;
2242 		int *max = (int *) table->extra2;
2243 		size_t length;
2244 		int i;
2245 
2246 		if (newlen % sizeof(int) != 0)
2247 			return -EINVAL;
2248 
2249 		if (!table->extra1 && !table->extra2)
2250 			return 0;
2251 
2252 		if (newlen > table->maxlen)
2253 			newlen = table->maxlen;
2254 		length = newlen / sizeof(int);
2255 
2256 		for (i = 0; i < length; i++) {
2257 			int value;
2258 			if (get_user(value, vec + i))
2259 				return -EFAULT;
2260 			if (min && value < min[i])
2261 				return -EINVAL;
2262 			if (max && value > max[i])
2263 				return -EINVAL;
2264 		}
2265 	}
2266 	return 0;
2267 }
2268 
2269 /* Strategy function to convert jiffies to seconds */
2270 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2271 		void __user *oldval, size_t __user *oldlenp,
2272 		void __user *newval, size_t newlen, void **context)
2273 {
2274 	if (oldval) {
2275 		size_t olen;
2276 		if (oldlenp) {
2277 			if (get_user(olen, oldlenp))
2278 				return -EFAULT;
2279 			if (olen!=sizeof(int))
2280 				return -EINVAL;
2281 		}
2282 		if (put_user(*(int *)(table->data)/HZ, (int __user *)oldval) ||
2283 		    (oldlenp && put_user(sizeof(int),oldlenp)))
2284 			return -EFAULT;
2285 	}
2286 	if (newval && newlen) {
2287 		int new;
2288 		if (newlen != sizeof(int))
2289 			return -EINVAL;
2290 		if (get_user(new, (int __user *)newval))
2291 			return -EFAULT;
2292 		*(int *)(table->data) = new*HZ;
2293 	}
2294 	return 1;
2295 }
2296 
2297 /* Strategy function to convert jiffies to seconds */
2298 int sysctl_ms_jiffies(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 (oldval) {
2303 		size_t olen;
2304 		if (oldlenp) {
2305 			if (get_user(olen, oldlenp))
2306 				return -EFAULT;
2307 			if (olen!=sizeof(int))
2308 				return -EINVAL;
2309 		}
2310 		if (put_user(jiffies_to_msecs(*(int *)(table->data)), (int __user *)oldval) ||
2311 		    (oldlenp && put_user(sizeof(int),oldlenp)))
2312 			return -EFAULT;
2313 	}
2314 	if (newval && newlen) {
2315 		int new;
2316 		if (newlen != sizeof(int))
2317 			return -EINVAL;
2318 		if (get_user(new, (int __user *)newval))
2319 			return -EFAULT;
2320 		*(int *)(table->data) = msecs_to_jiffies(new);
2321 	}
2322 	return 1;
2323 }
2324 
2325 #else /* CONFIG_SYSCTL */
2326 
2327 
2328 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2329 {
2330 	return -ENOSYS;
2331 }
2332 
2333 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2334 		  void __user *oldval, size_t __user *oldlenp,
2335 		  void __user *newval, size_t newlen, void **context)
2336 {
2337 	return -ENOSYS;
2338 }
2339 
2340 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2341 		void __user *oldval, size_t __user *oldlenp,
2342 		void __user *newval, size_t newlen, void **context)
2343 {
2344 	return -ENOSYS;
2345 }
2346 
2347 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2348 		void __user *oldval, size_t __user *oldlenp,
2349 		void __user *newval, size_t newlen, void **context)
2350 {
2351 	return -ENOSYS;
2352 }
2353 
2354 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2355 		void __user *oldval, size_t __user *oldlenp,
2356 		void __user *newval, size_t newlen, void **context)
2357 {
2358 	return -ENOSYS;
2359 }
2360 
2361 int proc_dostring(ctl_table *table, int write, struct file *filp,
2362 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2363 {
2364 	return -ENOSYS;
2365 }
2366 
2367 int proc_dointvec(ctl_table *table, int write, struct file *filp,
2368 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2369 {
2370 	return -ENOSYS;
2371 }
2372 
2373 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2374 			void __user *buffer, size_t *lenp, loff_t *ppos)
2375 {
2376 	return -ENOSYS;
2377 }
2378 
2379 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2380 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2381 {
2382 	return -ENOSYS;
2383 }
2384 
2385 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2386 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2387 {
2388 	return -ENOSYS;
2389 }
2390 
2391 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2392 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2393 {
2394 	return -ENOSYS;
2395 }
2396 
2397 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2398 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2399 {
2400 	return -ENOSYS;
2401 }
2402 
2403 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2404 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2405 {
2406 	return -ENOSYS;
2407 }
2408 
2409 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2410 				      struct file *filp,
2411 				      void __user *buffer,
2412 				      size_t *lenp, loff_t *ppos)
2413 {
2414     return -ENOSYS;
2415 }
2416 
2417 struct ctl_table_header * register_sysctl_table(ctl_table * table,
2418 						int insert_at_head)
2419 {
2420 	return NULL;
2421 }
2422 
2423 void unregister_sysctl_table(struct ctl_table_header * table)
2424 {
2425 }
2426 
2427 #endif /* CONFIG_SYSCTL */
2428 
2429 /*
2430  * No sense putting this after each symbol definition, twice,
2431  * exception granted :-)
2432  */
2433 EXPORT_SYMBOL(proc_dointvec);
2434 EXPORT_SYMBOL(proc_dointvec_jiffies);
2435 EXPORT_SYMBOL(proc_dointvec_minmax);
2436 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2437 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2438 EXPORT_SYMBOL(proc_dostring);
2439 EXPORT_SYMBOL(proc_doulongvec_minmax);
2440 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2441 EXPORT_SYMBOL(register_sysctl_table);
2442 EXPORT_SYMBOL(sysctl_intvec);
2443 EXPORT_SYMBOL(sysctl_jiffies);
2444 EXPORT_SYMBOL(sysctl_ms_jiffies);
2445 EXPORT_SYMBOL(sysctl_string);
2446 EXPORT_SYMBOL(unregister_sysctl_table);
2447