?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
[PATCH] sysctl: move utsname sysctls to their own file
[linux-2.6/x86.git] / kernel / sysctl.c
blobeff1d36c06871d2cf24febf733bb010ea049bedb
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 #ifdef CONFIG_X86
56 #include <asm/nmi.h>
57 #include <asm/stacktrace.h>
58 #endif
59
60 #if defined(CONFIG_SYSCTL)
61
62 /* External variables not in a header file. */
63 extern int C_A_D;
64 extern int sysctl_overcommit_memory;
65 extern int sysctl_overcommit_ratio;
66 extern int sysctl_panic_on_oom;
67 extern int max_threads;
68 extern int core_uses_pid;
69 extern int suid_dumpable;
70 extern char core_pattern[];
71 extern int pid_max;
72 extern int min_free_kbytes;
73 extern int printk_ratelimit_jiffies;
74 extern int printk_ratelimit_burst;
75 extern int pid_max_min, pid_max_max;
76 extern int sysctl_drop_caches;
77 extern int percpu_pagelist_fraction;
78 extern int compat_log;
79
80 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
81 static int maxolduid = 65535;
82 static int minolduid;
83 static int min_percpu_pagelist_fract = 8;
84
85 static int ngroups_max = NGROUPS_MAX;
86
87 #ifdef CONFIG_KMOD
88 extern char modprobe_path[];
89 #endif
90 #ifdef CONFIG_CHR_DEV_SG
91 extern int sg_big_buff;
92 #endif
93 #ifdef CONFIG_SYSVIPC
94 static int proc_ipc_dointvec(ctl_table *table, int write, struct file *filp,
95 void __user *buffer, size_t *lenp, loff_t *ppos);
96 static int proc_ipc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
97 void __user *buffer, size_t *lenp, loff_t *ppos);
98 #endif
99
100 #ifdef __sparc__
101 extern char reboot_command [];
102 extern int stop_a_enabled;
103 extern int scons_pwroff;
104 #endif
105
106 #ifdef __hppa__
107 extern int pwrsw_enabled;
108 extern int unaligned_enabled;
109 #endif
110
111 #ifdef CONFIG_S390
112 #ifdef CONFIG_MATHEMU
113 extern int sysctl_ieee_emulation_warnings;
114 #endif
115 extern int sysctl_userprocess_debug;
116 extern int spin_retry;
117 #endif
118
119 extern int sysctl_hz_timer;
120
121 #ifdef CONFIG_BSD_PROCESS_ACCT
122 extern int acct_parm[];
123 #endif
124
125 #ifdef CONFIG_IA64
126 extern int no_unaligned_warning;
127 #endif
128
129 #ifdef CONFIG_RT_MUTEXES
130 extern int max_lock_depth;
131 #endif
132
133 #ifdef CONFIG_SYSCTL_SYSCALL
134 static int parse_table(int __user *, int, void __user *, size_t __user *,
135 void __user *, size_t, ctl_table *);
136 #endif
137
138 #ifdef CONFIG_SYSVIPC
139 static int sysctl_ipc_data(ctl_table *table, int __user *name, int nlen,
140 void __user *oldval, size_t __user *oldlenp,
141 void __user *newval, size_t newlen);
142 #endif
143
144 #ifdef CONFIG_PROC_SYSCTL
145 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
146 void __user *buffer, size_t *lenp, loff_t *ppos);
147 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
148 void __user *buffer, size_t *lenp, loff_t *ppos);
149 #endif
150
151 static ctl_table root_table[];
152 static struct ctl_table_header root_table_header =
153 { root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
154
155 static ctl_table kern_table[];
156 static ctl_table vm_table[];
157 static ctl_table fs_table[];
158 static ctl_table debug_table[];
159 static ctl_table dev_table[];
160 extern ctl_table random_table[];
161 #ifdef CONFIG_UNIX98_PTYS
162 extern ctl_table pty_table[];
163 #endif
164 #ifdef CONFIG_INOTIFY_USER
165 extern ctl_table inotify_table[];
166 #endif
167
168 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
169 int sysctl_legacy_va_layout;
170 #endif
171
172 #ifdef CONFIG_SYSVIPC
173 static void *get_ipc(ctl_table *table, int write)
174 {
175 char *which = table->data;
176 struct ipc_namespace *ipc_ns = current->nsproxy->ipc_ns;
177 which = (which - (char *)&init_ipc_ns) + (char *)ipc_ns;
178 return which;
179 }
180 #else
181 #define get_ipc(T,W) ((T)->data)
182 #endif
183
184 /* /proc declarations: */
185
186 #ifdef CONFIG_PROC_SYSCTL
187
188 static ssize_t proc_readsys(struct file *, char __user *, size_t, loff_t *);
189 static ssize_t proc_writesys(struct file *, const char __user *, size_t, loff_t *);
190 static int proc_opensys(struct inode *, struct file *);
191
192 const struct file_operations proc_sys_file_operations = {
193 .open = proc_opensys,
194 .read = proc_readsys,
195 .write = proc_writesys,
196 };
197
198 extern struct proc_dir_entry *proc_sys_root;
199
200 static void register_proc_table(ctl_table *, struct proc_dir_entry *, void *);
201 static void unregister_proc_table(ctl_table *, struct proc_dir_entry *);
202 #endif
203
204 /* The default sysctl tables: */
205
206 static ctl_table root_table[] = {
207 {
208 .ctl_name = CTL_KERN,
209 .procname = "kernel",
210 .mode = 0555,
211 .child = kern_table,
212 },
213 {
214 .ctl_name = CTL_VM,
215 .procname = "vm",
216 .mode = 0555,
217 .child = vm_table,
218 },
219 #ifdef CONFIG_NET
220 {
221 .ctl_name = CTL_NET,
222 .procname = "net",
223 .mode = 0555,
224 .child = net_table,
225 },
226 #endif
227 {
228 .ctl_name = CTL_FS,
229 .procname = "fs",
230 .mode = 0555,
231 .child = fs_table,
232 },
233 {
234 .ctl_name = CTL_DEBUG,
235 .procname = "debug",
236 .mode = 0555,
237 .child = debug_table,
238 },
239 {
240 .ctl_name = CTL_DEV,
241 .procname = "dev",
242 .mode = 0555,
243 .child = dev_table,
244 },
245
246 { .ctl_name = 0 }
247 };
248
249 static ctl_table kern_table[] = {
250 {
251 .ctl_name = KERN_PANIC,
252 .procname = "panic",
253 .data = &panic_timeout,
254 .maxlen = sizeof(int),
255 .mode = 0644,
256 .proc_handler = &proc_dointvec,
257 },
258 {
259 .ctl_name = KERN_CORE_USES_PID,
260 .procname = "core_uses_pid",
261 .data = &core_uses_pid,
262 .maxlen = sizeof(int),
263 .mode = 0644,
264 .proc_handler = &proc_dointvec,
265 },
266 {
267 .ctl_name = KERN_CORE_PATTERN,
268 .procname = "core_pattern",
269 .data = core_pattern,
270 .maxlen = 128,
271 .mode = 0644,
272 .proc_handler = &proc_dostring,
273 .strategy = &sysctl_string,
274 },
275 #ifdef CONFIG_PROC_SYSCTL
276 {
277 .ctl_name = KERN_TAINTED,
278 .procname = "tainted",
279 .data = &tainted,
280 .maxlen = sizeof(int),
281 .mode = 0644,
282 .proc_handler = &proc_dointvec_taint,
283 },
284 #endif
285 {
286 .ctl_name = KERN_CAP_BSET,
287 .procname = "cap-bound",
288 .data = &cap_bset,
289 .maxlen = sizeof(kernel_cap_t),
290 .mode = 0600,
291 .proc_handler = &proc_dointvec_bset,
292 },
293 #ifdef CONFIG_BLK_DEV_INITRD
294 {
295 .ctl_name = KERN_REALROOTDEV,
296 .procname = "real-root-dev",
297 .data = &real_root_dev,
298 .maxlen = sizeof(int),
299 .mode = 0644,
300 .proc_handler = &proc_dointvec,
301 },
302 #endif
303 #ifdef __sparc__
304 {
305 .ctl_name = KERN_SPARC_REBOOT,
306 .procname = "reboot-cmd",
307 .data = reboot_command,
308 .maxlen = 256,
309 .mode = 0644,
310 .proc_handler = &proc_dostring,
311 .strategy = &sysctl_string,
312 },
313 {
314 .ctl_name = KERN_SPARC_STOP_A,
315 .procname = "stop-a",
316 .data = &stop_a_enabled,
317 .maxlen = sizeof (int),
318 .mode = 0644,
319 .proc_handler = &proc_dointvec,
320 },
321 {
322 .ctl_name = KERN_SPARC_SCONS_PWROFF,
323 .procname = "scons-poweroff",
324 .data = &scons_pwroff,
325 .maxlen = sizeof (int),
326 .mode = 0644,
327 .proc_handler = &proc_dointvec,
328 },
329 #endif
330 #ifdef __hppa__
331 {
332 .ctl_name = KERN_HPPA_PWRSW,
333 .procname = "soft-power",
334 .data = &pwrsw_enabled,
335 .maxlen = sizeof (int),
336 .mode = 0644,
337 .proc_handler = &proc_dointvec,
338 },
339 {
340 .ctl_name = KERN_HPPA_UNALIGNED,
341 .procname = "unaligned-trap",
342 .data = &unaligned_enabled,
343 .maxlen = sizeof (int),
344 .mode = 0644,
345 .proc_handler = &proc_dointvec,
346 },
347 #endif
348 {
349 .ctl_name = KERN_CTLALTDEL,
350 .procname = "ctrl-alt-del",
351 .data = &C_A_D,
352 .maxlen = sizeof(int),
353 .mode = 0644,
354 .proc_handler = &proc_dointvec,
355 },
356 {
357 .ctl_name = KERN_PRINTK,
358 .procname = "printk",
359 .data = &console_loglevel,
360 .maxlen = 4*sizeof(int),
361 .mode = 0644,
362 .proc_handler = &proc_dointvec,
363 },
364 #ifdef CONFIG_KMOD
365 {
366 .ctl_name = KERN_MODPROBE,
367 .procname = "modprobe",
368 .data = &modprobe_path,
369 .maxlen = KMOD_PATH_LEN,
370 .mode = 0644,
371 .proc_handler = &proc_dostring,
372 .strategy = &sysctl_string,
373 },
374 #endif
375 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
376 {
377 .ctl_name = KERN_HOTPLUG,
378 .procname = "hotplug",
379 .data = &uevent_helper,
380 .maxlen = UEVENT_HELPER_PATH_LEN,
381 .mode = 0644,
382 .proc_handler = &proc_dostring,
383 .strategy = &sysctl_string,
384 },
385 #endif
386 #ifdef CONFIG_CHR_DEV_SG
387 {
388 .ctl_name = KERN_SG_BIG_BUFF,
389 .procname = "sg-big-buff",
390 .data = &sg_big_buff,
391 .maxlen = sizeof (int),
392 .mode = 0444,
393 .proc_handler = &proc_dointvec,
394 },
395 #endif
396 #ifdef CONFIG_BSD_PROCESS_ACCT
397 {
398 .ctl_name = KERN_ACCT,
399 .procname = "acct",
400 .data = &acct_parm,
401 .maxlen = 3*sizeof(int),
402 .mode = 0644,
403 .proc_handler = &proc_dointvec,
404 },
405 #endif
406 #ifdef CONFIG_SYSVIPC
407 {
408 .ctl_name = KERN_SHMMAX,
409 .procname = "shmmax",
410 .data = &init_ipc_ns.shm_ctlmax,
411 .maxlen = sizeof (init_ipc_ns.shm_ctlmax),
412 .mode = 0644,
413 .proc_handler = &proc_ipc_doulongvec_minmax,
414 .strategy = sysctl_ipc_data,
415 },
416 {
417 .ctl_name = KERN_SHMALL,
418 .procname = "shmall",
419 .data = &init_ipc_ns.shm_ctlall,
420 .maxlen = sizeof (init_ipc_ns.shm_ctlall),
421 .mode = 0644,
422 .proc_handler = &proc_ipc_doulongvec_minmax,
423 .strategy = sysctl_ipc_data,
424 },
425 {
426 .ctl_name = KERN_SHMMNI,
427 .procname = "shmmni",
428 .data = &init_ipc_ns.shm_ctlmni,
429 .maxlen = sizeof (init_ipc_ns.shm_ctlmni),
430 .mode = 0644,
431 .proc_handler = &proc_ipc_dointvec,
432 .strategy = sysctl_ipc_data,
433 },
434 {
435 .ctl_name = KERN_MSGMAX,
436 .procname = "msgmax",
437 .data = &init_ipc_ns.msg_ctlmax,
438 .maxlen = sizeof (init_ipc_ns.msg_ctlmax),
439 .mode = 0644,
440 .proc_handler = &proc_ipc_dointvec,
441 .strategy = sysctl_ipc_data,
442 },
443 {
444 .ctl_name = KERN_MSGMNI,
445 .procname = "msgmni",
446 .data = &init_ipc_ns.msg_ctlmni,
447 .maxlen = sizeof (init_ipc_ns.msg_ctlmni),
448 .mode = 0644,
449 .proc_handler = &proc_ipc_dointvec,
450 .strategy = sysctl_ipc_data,
451 },
452 {
453 .ctl_name = KERN_MSGMNB,
454 .procname = "msgmnb",
455 .data = &init_ipc_ns.msg_ctlmnb,
456 .maxlen = sizeof (init_ipc_ns.msg_ctlmnb),
457 .mode = 0644,
458 .proc_handler = &proc_ipc_dointvec,
459 .strategy = sysctl_ipc_data,
460 },
461 {
462 .ctl_name = KERN_SEM,
463 .procname = "sem",
464 .data = &init_ipc_ns.sem_ctls,
465 .maxlen = 4*sizeof (int),
466 .mode = 0644,
467 .proc_handler = &proc_ipc_dointvec,
468 .strategy = sysctl_ipc_data,
469 },
470 #endif
471 #ifdef CONFIG_MAGIC_SYSRQ
472 {
473 .ctl_name = KERN_SYSRQ,
474 .procname = "sysrq",
475 .data = &__sysrq_enabled,
476 .maxlen = sizeof (int),
477 .mode = 0644,
478 .proc_handler = &proc_dointvec,
479 },
480 #endif
481 #ifdef CONFIG_PROC_SYSCTL
482 {
483 .ctl_name = KERN_CADPID,
484 .procname = "cad_pid",
485 .data = NULL,
486 .maxlen = sizeof (int),
487 .mode = 0600,
488 .proc_handler = &proc_do_cad_pid,
489 },
490 #endif
491 {
492 .ctl_name = KERN_MAX_THREADS,
493 .procname = "threads-max",
494 .data = &max_threads,
495 .maxlen = sizeof(int),
496 .mode = 0644,
497 .proc_handler = &proc_dointvec,
498 },
499 {
500 .ctl_name = KERN_RANDOM,
501 .procname = "random",
502 .mode = 0555,
503 .child = random_table,
504 },
505 #ifdef CONFIG_UNIX98_PTYS
506 {
507 .ctl_name = KERN_PTY,
508 .procname = "pty",
509 .mode = 0555,
510 .child = pty_table,
511 },
512 #endif
513 {
514 .ctl_name = KERN_OVERFLOWUID,
515 .procname = "overflowuid",
516 .data = &overflowuid,
517 .maxlen = sizeof(int),
518 .mode = 0644,
519 .proc_handler = &proc_dointvec_minmax,
520 .strategy = &sysctl_intvec,
521 .extra1 = &minolduid,
522 .extra2 = &maxolduid,
523 },
524 {
525 .ctl_name = KERN_OVERFLOWGID,
526 .procname = "overflowgid",
527 .data = &overflowgid,
528 .maxlen = sizeof(int),
529 .mode = 0644,
530 .proc_handler = &proc_dointvec_minmax,
531 .strategy = &sysctl_intvec,
532 .extra1 = &minolduid,
533 .extra2 = &maxolduid,
534 },
535 #ifdef CONFIG_S390
536 #ifdef CONFIG_MATHEMU
537 {
538 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
539 .procname = "ieee_emulation_warnings",
540 .data = &sysctl_ieee_emulation_warnings,
541 .maxlen = sizeof(int),
542 .mode = 0644,
543 .proc_handler = &proc_dointvec,
544 },
545 #endif
546 #ifdef CONFIG_NO_IDLE_HZ
547 {
548 .ctl_name = KERN_HZ_TIMER,
549 .procname = "hz_timer",
550 .data = &sysctl_hz_timer,
551 .maxlen = sizeof(int),
552 .mode = 0644,
553 .proc_handler = &proc_dointvec,
554 },
555 #endif
556 {
557 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
558 .procname = "userprocess_debug",
559 .data = &sysctl_userprocess_debug,
560 .maxlen = sizeof(int),
561 .mode = 0644,
562 .proc_handler = &proc_dointvec,
563 },
564 #endif
565 {
566 .ctl_name = KERN_PIDMAX,
567 .procname = "pid_max",
568 .data = &pid_max,
569 .maxlen = sizeof (int),
570 .mode = 0644,
571 .proc_handler = &proc_dointvec_minmax,
572 .strategy = sysctl_intvec,
573 .extra1 = &pid_max_min,
574 .extra2 = &pid_max_max,
575 },
576 {
577 .ctl_name = KERN_PANIC_ON_OOPS,
578 .procname = "panic_on_oops",
579 .data = &panic_on_oops,
580 .maxlen = sizeof(int),
581 .mode = 0644,
582 .proc_handler = &proc_dointvec,
583 },
584 {
585 .ctl_name = KERN_PRINTK_RATELIMIT,
586 .procname = "printk_ratelimit",
587 .data = &printk_ratelimit_jiffies,
588 .maxlen = sizeof(int),
589 .mode = 0644,
590 .proc_handler = &proc_dointvec_jiffies,
591 .strategy = &sysctl_jiffies,
592 },
593 {
594 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
595 .procname = "printk_ratelimit_burst",
596 .data = &printk_ratelimit_burst,
597 .maxlen = sizeof(int),
598 .mode = 0644,
599 .proc_handler = &proc_dointvec,
600 },
601 {
602 .ctl_name = KERN_NGROUPS_MAX,
603 .procname = "ngroups_max",
604 .data = &ngroups_max,
605 .maxlen = sizeof (int),
606 .mode = 0444,
607 .proc_handler = &proc_dointvec,
608 },
609 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
610 {
611 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
612 .procname = "unknown_nmi_panic",
613 .data = &unknown_nmi_panic,
614 .maxlen = sizeof (int),
615 .mode = 0644,
616 .proc_handler = &proc_dointvec,
617 },
618 {
619 .ctl_name = KERN_NMI_WATCHDOG,
620 .procname = "nmi_watchdog",
621 .data = &nmi_watchdog_enabled,
622 .maxlen = sizeof (int),
623 .mode = 0644,
624 .proc_handler = &proc_nmi_enabled,
625 },
626 #endif
627 #if defined(CONFIG_X86)
628 {
629 .ctl_name = KERN_PANIC_ON_NMI,
630 .procname = "panic_on_unrecovered_nmi",
631 .data = &panic_on_unrecovered_nmi,
632 .maxlen = sizeof(int),
633 .mode = 0644,
634 .proc_handler = &proc_dointvec,
635 },
636 {
637 .ctl_name = KERN_BOOTLOADER_TYPE,
638 .procname = "bootloader_type",
639 .data = &bootloader_type,
640 .maxlen = sizeof (int),
641 .mode = 0444,
642 .proc_handler = &proc_dointvec,
643 },
644 {
645 .ctl_name = CTL_UNNUMBERED,
646 .procname = "kstack_depth_to_print",
647 .data = &kstack_depth_to_print,
648 .maxlen = sizeof(int),
649 .mode = 0644,
650 .proc_handler = &proc_dointvec,
651 },
652 #endif
653 #if defined(CONFIG_MMU)
654 {
655 .ctl_name = KERN_RANDOMIZE,
656 .procname = "randomize_va_space",
657 .data = &randomize_va_space,
658 .maxlen = sizeof(int),
659 .mode = 0644,
660 .proc_handler = &proc_dointvec,
661 },
662 #endif
663 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
664 {
665 .ctl_name = KERN_SPIN_RETRY,
666 .procname = "spin_retry",
667 .data = &spin_retry,
668 .maxlen = sizeof (int),
669 .mode = 0644,
670 .proc_handler = &proc_dointvec,
671 },
672 #endif
673 #ifdef CONFIG_ACPI_SLEEP
674 {
675 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
676 .procname = "acpi_video_flags",
677 .data = &acpi_video_flags,
678 .maxlen = sizeof (unsigned long),
679 .mode = 0644,
680 .proc_handler = &proc_doulongvec_minmax,
681 },
682 #endif
683 #ifdef CONFIG_IA64
684 {
685 .ctl_name = KERN_IA64_UNALIGNED,
686 .procname = "ignore-unaligned-usertrap",
687 .data = &no_unaligned_warning,
688 .maxlen = sizeof (int),
689 .mode = 0644,
690 .proc_handler = &proc_dointvec,
691 },
692 #endif
693 #ifdef CONFIG_COMPAT
694 {
695 .ctl_name = KERN_COMPAT_LOG,
696 .procname = "compat-log",
697 .data = &compat_log,
698 .maxlen = sizeof (int),
699 .mode = 0644,
700 .proc_handler = &proc_dointvec,
701 },
702 #endif
703 #ifdef CONFIG_RT_MUTEXES
704 {
705 .ctl_name = KERN_MAX_LOCK_DEPTH,
706 .procname = "max_lock_depth",
707 .data = &max_lock_depth,
708 .maxlen = sizeof(int),
709 .mode = 0644,
710 .proc_handler = &proc_dointvec,
711 },
712 #endif
713
714 { .ctl_name = 0 }
715 };
716
717 /* Constants for minimum and maximum testing in vm_table.
718 We use these as one-element integer vectors. */
719 static int zero;
720 static int one_hundred = 100;
721
722
723 static ctl_table vm_table[] = {
724 {
725 .ctl_name = VM_OVERCOMMIT_MEMORY,
726 .procname = "overcommit_memory",
727 .data = &sysctl_overcommit_memory,
728 .maxlen = sizeof(sysctl_overcommit_memory),
729 .mode = 0644,
730 .proc_handler = &proc_dointvec,
731 },
732 {
733 .ctl_name = VM_PANIC_ON_OOM,
734 .procname = "panic_on_oom",
735 .data = &sysctl_panic_on_oom,
736 .maxlen = sizeof(sysctl_panic_on_oom),
737 .mode = 0644,
738 .proc_handler = &proc_dointvec,
739 },
740 {
741 .ctl_name = VM_OVERCOMMIT_RATIO,
742 .procname = "overcommit_ratio",
743 .data = &sysctl_overcommit_ratio,
744 .maxlen = sizeof(sysctl_overcommit_ratio),
745 .mode = 0644,
746 .proc_handler = &proc_dointvec,
747 },
748 {
749 .ctl_name = VM_PAGE_CLUSTER,
750 .procname = "page-cluster",
751 .data = &page_cluster,
752 .maxlen = sizeof(int),
753 .mode = 0644,
754 .proc_handler = &proc_dointvec,
755 },
756 {
757 .ctl_name = VM_DIRTY_BACKGROUND,
758 .procname = "dirty_background_ratio",
759 .data = &dirty_background_ratio,
760 .maxlen = sizeof(dirty_background_ratio),
761 .mode = 0644,
762 .proc_handler = &proc_dointvec_minmax,
763 .strategy = &sysctl_intvec,
764 .extra1 = &zero,
765 .extra2 = &one_hundred,
766 },
767 {
768 .ctl_name = VM_DIRTY_RATIO,
769 .procname = "dirty_ratio",
770 .data = &vm_dirty_ratio,
771 .maxlen = sizeof(vm_dirty_ratio),
772 .mode = 0644,
773 .proc_handler = &proc_dointvec_minmax,
774 .strategy = &sysctl_intvec,
775 .extra1 = &zero,
776 .extra2 = &one_hundred,
777 },
778 {
779 .ctl_name = VM_DIRTY_WB_CS,
780 .procname = "dirty_writeback_centisecs",
781 .data = &dirty_writeback_interval,
782 .maxlen = sizeof(dirty_writeback_interval),
783 .mode = 0644,
784 .proc_handler = &dirty_writeback_centisecs_handler,
785 },
786 {
787 .ctl_name = VM_DIRTY_EXPIRE_CS,
788 .procname = "dirty_expire_centisecs",
789 .data = &dirty_expire_interval,
790 .maxlen = sizeof(dirty_expire_interval),
791 .mode = 0644,
792 .proc_handler = &proc_dointvec_userhz_jiffies,
793 },
794 {
795 .ctl_name = VM_NR_PDFLUSH_THREADS,
796 .procname = "nr_pdflush_threads",
797 .data = &nr_pdflush_threads,
798 .maxlen = sizeof nr_pdflush_threads,
799 .mode = 0444 /* read-only*/,
800 .proc_handler = &proc_dointvec,
801 },
802 {
803 .ctl_name = VM_SWAPPINESS,
804 .procname = "swappiness",
805 .data = &vm_swappiness,
806 .maxlen = sizeof(vm_swappiness),
807 .mode = 0644,
808 .proc_handler = &proc_dointvec_minmax,
809 .strategy = &sysctl_intvec,
810 .extra1 = &zero,
811 .extra2 = &one_hundred,
812 },
813 #ifdef CONFIG_HUGETLB_PAGE
814 {
815 .ctl_name = VM_HUGETLB_PAGES,
816 .procname = "nr_hugepages",
817 .data = &max_huge_pages,
818 .maxlen = sizeof(unsigned long),
819 .mode = 0644,
820 .proc_handler = &hugetlb_sysctl_handler,
821 .extra1 = (void *)&hugetlb_zero,
822 .extra2 = (void *)&hugetlb_infinity,
823 },
824 {
825 .ctl_name = VM_HUGETLB_GROUP,
826 .procname = "hugetlb_shm_group",
827 .data = &sysctl_hugetlb_shm_group,
828 .maxlen = sizeof(gid_t),
829 .mode = 0644,
830 .proc_handler = &proc_dointvec,
831 },
832 #endif
833 {
834 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
835 .procname = "lowmem_reserve_ratio",
836 .data = &sysctl_lowmem_reserve_ratio,
837 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
838 .mode = 0644,
839 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
840 .strategy = &sysctl_intvec,
841 },
842 {
843 .ctl_name = VM_DROP_PAGECACHE,
844 .procname = "drop_caches",
845 .data = &sysctl_drop_caches,
846 .maxlen = sizeof(int),
847 .mode = 0644,
848 .proc_handler = drop_caches_sysctl_handler,
849 .strategy = &sysctl_intvec,
850 },
851 {
852 .ctl_name = VM_MIN_FREE_KBYTES,
853 .procname = "min_free_kbytes",
854 .data = &min_free_kbytes,
855 .maxlen = sizeof(min_free_kbytes),
856 .mode = 0644,
857 .proc_handler = &min_free_kbytes_sysctl_handler,
858 .strategy = &sysctl_intvec,
859 .extra1 = &zero,
860 },
861 {
862 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
863 .procname = "percpu_pagelist_fraction",
864 .data = &percpu_pagelist_fraction,
865 .maxlen = sizeof(percpu_pagelist_fraction),
866 .mode = 0644,
867 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
868 .strategy = &sysctl_intvec,
869 .extra1 = &min_percpu_pagelist_fract,
870 },
871 #ifdef CONFIG_MMU
872 {
873 .ctl_name = VM_MAX_MAP_COUNT,
874 .procname = "max_map_count",
875 .data = &sysctl_max_map_count,
876 .maxlen = sizeof(sysctl_max_map_count),
877 .mode = 0644,
878 .proc_handler = &proc_dointvec
879 },
880 #endif
881 {
882 .ctl_name = VM_LAPTOP_MODE,
883 .procname = "laptop_mode",
884 .data = &laptop_mode,
885 .maxlen = sizeof(laptop_mode),
886 .mode = 0644,
887 .proc_handler = &proc_dointvec_jiffies,
888 .strategy = &sysctl_jiffies,
889 },
890 {
891 .ctl_name = VM_BLOCK_DUMP,
892 .procname = "block_dump",
893 .data = &block_dump,
894 .maxlen = sizeof(block_dump),
895 .mode = 0644,
896 .proc_handler = &proc_dointvec,
897 .strategy = &sysctl_intvec,
898 .extra1 = &zero,
899 },
900 {
901 .ctl_name = VM_VFS_CACHE_PRESSURE,
902 .procname = "vfs_cache_pressure",
903 .data = &sysctl_vfs_cache_pressure,
904 .maxlen = sizeof(sysctl_vfs_cache_pressure),
905 .mode = 0644,
906 .proc_handler = &proc_dointvec,
907 .strategy = &sysctl_intvec,
908 .extra1 = &zero,
909 },
910 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
911 {
912 .ctl_name = VM_LEGACY_VA_LAYOUT,
913 .procname = "legacy_va_layout",
914 .data = &sysctl_legacy_va_layout,
915 .maxlen = sizeof(sysctl_legacy_va_layout),
916 .mode = 0644,
917 .proc_handler = &proc_dointvec,
918 .strategy = &sysctl_intvec,
919 .extra1 = &zero,
920 },
921 #endif
922 #ifdef CONFIG_NUMA
923 {
924 .ctl_name = VM_ZONE_RECLAIM_MODE,
925 .procname = "zone_reclaim_mode",
926 .data = &zone_reclaim_mode,
927 .maxlen = sizeof(zone_reclaim_mode),
928 .mode = 0644,
929 .proc_handler = &proc_dointvec,
930 .strategy = &sysctl_intvec,
931 .extra1 = &zero,
932 },
933 {
934 .ctl_name = VM_MIN_UNMAPPED,
935 .procname = "min_unmapped_ratio",
936 .data = &sysctl_min_unmapped_ratio,
937 .maxlen = sizeof(sysctl_min_unmapped_ratio),
938 .mode = 0644,
939 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
940 .strategy = &sysctl_intvec,
941 .extra1 = &zero,
942 .extra2 = &one_hundred,
943 },
944 {
945 .ctl_name = VM_MIN_SLAB,
946 .procname = "min_slab_ratio",
947 .data = &sysctl_min_slab_ratio,
948 .maxlen = sizeof(sysctl_min_slab_ratio),
949 .mode = 0644,
950 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
951 .strategy = &sysctl_intvec,
952 .extra1 = &zero,
953 .extra2 = &one_hundred,
954 },
955 #endif
956 #ifdef CONFIG_X86_32
957 {
958 .ctl_name = VM_VDSO_ENABLED,
959 .procname = "vdso_enabled",
960 .data = &vdso_enabled,
961 .maxlen = sizeof(vdso_enabled),
962 .mode = 0644,
963 .proc_handler = &proc_dointvec,
964 .strategy = &sysctl_intvec,
965 .extra1 = &zero,
966 },
967 #endif
968 { .ctl_name = 0 }
969 };
970
971 static ctl_table fs_table[] = {
972 {
973 .ctl_name = FS_NRINODE,
974 .procname = "inode-nr",
975 .data = &inodes_stat,
976 .maxlen = 2*sizeof(int),
977 .mode = 0444,
978 .proc_handler = &proc_dointvec,
979 },
980 {
981 .ctl_name = FS_STATINODE,
982 .procname = "inode-state",
983 .data = &inodes_stat,
984 .maxlen = 7*sizeof(int),
985 .mode = 0444,
986 .proc_handler = &proc_dointvec,
987 },
988 {
989 .ctl_name = FS_NRFILE,
990 .procname = "file-nr",
991 .data = &files_stat,
992 .maxlen = 3*sizeof(int),
993 .mode = 0444,
994 .proc_handler = &proc_nr_files,
995 },
996 {
997 .ctl_name = FS_MAXFILE,
998 .procname = "file-max",
999 .data = &files_stat.max_files,
1000 .maxlen = sizeof(int),
1001 .mode = 0644,
1002 .proc_handler = &proc_dointvec,
1003 },
1004 {
1005 .ctl_name = FS_DENTRY,
1006 .procname = "dentry-state",
1007 .data = &dentry_stat,
1008 .maxlen = 6*sizeof(int),
1009 .mode = 0444,
1010 .proc_handler = &proc_dointvec,
1011 },
1012 {
1013 .ctl_name = FS_OVERFLOWUID,
1014 .procname = "overflowuid",
1015 .data = &fs_overflowuid,
1016 .maxlen = sizeof(int),
1017 .mode = 0644,
1018 .proc_handler = &proc_dointvec_minmax,
1019 .strategy = &sysctl_intvec,
1020 .extra1 = &minolduid,
1021 .extra2 = &maxolduid,
1022 },
1023 {
1024 .ctl_name = FS_OVERFLOWGID,
1025 .procname = "overflowgid",
1026 .data = &fs_overflowgid,
1027 .maxlen = sizeof(int),
1028 .mode = 0644,
1029 .proc_handler = &proc_dointvec_minmax,
1030 .strategy = &sysctl_intvec,
1031 .extra1 = &minolduid,
1032 .extra2 = &maxolduid,
1033 },
1034 {
1035 .ctl_name = FS_LEASES,
1036 .procname = "leases-enable",
1037 .data = &leases_enable,
1038 .maxlen = sizeof(int),
1039 .mode = 0644,
1040 .proc_handler = &proc_dointvec,
1041 },
1042 #ifdef CONFIG_DNOTIFY
1043 {
1044 .ctl_name = FS_DIR_NOTIFY,
1045 .procname = "dir-notify-enable",
1046 .data = &dir_notify_enable,
1047 .maxlen = sizeof(int),
1048 .mode = 0644,
1049 .proc_handler = &proc_dointvec,
1050 },
1051 #endif
1052 #ifdef CONFIG_MMU
1053 {
1054 .ctl_name = FS_LEASE_TIME,
1055 .procname = "lease-break-time",
1056 .data = &lease_break_time,
1057 .maxlen = sizeof(int),
1058 .mode = 0644,
1059 .proc_handler = &proc_dointvec,
1060 },
1061 {
1062 .ctl_name = FS_AIO_NR,
1063 .procname = "aio-nr",
1064 .data = &aio_nr,
1065 .maxlen = sizeof(aio_nr),
1066 .mode = 0444,
1067 .proc_handler = &proc_doulongvec_minmax,
1068 },
1069 {
1070 .ctl_name = FS_AIO_MAX_NR,
1071 .procname = "aio-max-nr",
1072 .data = &aio_max_nr,
1073 .maxlen = sizeof(aio_max_nr),
1074 .mode = 0644,
1075 .proc_handler = &proc_doulongvec_minmax,
1076 },
1077 #ifdef CONFIG_INOTIFY_USER
1078 {
1079 .ctl_name = FS_INOTIFY,
1080 .procname = "inotify",
1081 .mode = 0555,
1082 .child = inotify_table,
1083 },
1084 #endif
1085 #endif
1086 {
1087 .ctl_name = KERN_SETUID_DUMPABLE,
1088 .procname = "suid_dumpable",
1089 .data = &suid_dumpable,
1090 .maxlen = sizeof(int),
1091 .mode = 0644,
1092 .proc_handler = &proc_dointvec,
1093 },
1094 { .ctl_name = 0 }
1095 };
1096
1097 static ctl_table debug_table[] = {
1098 { .ctl_name = 0 }
1099 };
1100
1101 static ctl_table dev_table[] = {
1102 { .ctl_name = 0 }
1103 };
1104
1105 static DEFINE_SPINLOCK(sysctl_lock);
1106
1107 /* called under sysctl_lock */
1108 static int use_table(struct ctl_table_header *p)
1109 {
1110 if (unlikely(p->unregistering))
1111 return 0;
1112 p->used++;
1113 return 1;
1114 }
1115
1116 /* called under sysctl_lock */
1117 static void unuse_table(struct ctl_table_header *p)
1118 {
1119 if (!--p->used)
1120 if (unlikely(p->unregistering))
1121 complete(p->unregistering);
1122 }
1123
1124 /* called under sysctl_lock, will reacquire if has to wait */
1125 static void start_unregistering(struct ctl_table_header *p)
1126 {
1127 /*
1128 * if p->used is 0, nobody will ever touch that entry again;
1129 * we'll eliminate all paths to it before dropping sysctl_lock
1130 */
1131 if (unlikely(p->used)) {
1132 struct completion wait;
1133 init_completion(&wait);
1134 p->unregistering = &wait;
1135 spin_unlock(&sysctl_lock);
1136 wait_for_completion(&wait);
1137 spin_lock(&sysctl_lock);
1138 }
1139 /*
1140 * do not remove from the list until nobody holds it; walking the
1141 * list in do_sysctl() relies on that.
1142 */
1143 list_del_init(&p->ctl_entry);
1144 }
1145
1146 void __init sysctl_init(void)
1147 {
1148 #ifdef CONFIG_PROC_SYSCTL
1149 register_proc_table(root_table, proc_sys_root, &root_table_header);
1150 #endif
1151 }
1152
1153 #ifdef CONFIG_SYSCTL_SYSCALL
1154 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1155 void __user *newval, size_t newlen)
1156 {
1157 struct list_head *tmp;
1158 int error = -ENOTDIR;
1159
1160 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1161 return -ENOTDIR;
1162 if (oldval) {
1163 int old_len;
1164 if (!oldlenp || get_user(old_len, oldlenp))
1165 return -EFAULT;
1166 }
1167 spin_lock(&sysctl_lock);
1168 tmp = &root_table_header.ctl_entry;
1169 do {
1170 struct ctl_table_header *head =
1171 list_entry(tmp, struct ctl_table_header, ctl_entry);
1172
1173 if (!use_table(head))
1174 continue;
1175
1176 spin_unlock(&sysctl_lock);
1177
1178 error = parse_table(name, nlen, oldval, oldlenp,
1179 newval, newlen, head->ctl_table);
1180
1181 spin_lock(&sysctl_lock);
1182 unuse_table(head);
1183 if (error != -ENOTDIR)
1184 break;
1185 } while ((tmp = tmp->next) != &root_table_header.ctl_entry);
1186 spin_unlock(&sysctl_lock);
1187 return error;
1188 }
1189
1190 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1191 {
1192 struct __sysctl_args tmp;
1193 int error;
1194
1195 if (copy_from_user(&tmp, args, sizeof(tmp)))
1196 return -EFAULT;
1197
1198 lock_kernel();
1199 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1200 tmp.newval, tmp.newlen);
1201 unlock_kernel();
1202 return error;
1203 }
1204 #endif /* CONFIG_SYSCTL_SYSCALL */
1205
1206 /*
1207 * ctl_perm does NOT grant the superuser all rights automatically, because
1208 * some sysctl variables are readonly even to root.
1209 */
1210
1211 static int test_perm(int mode, int op)
1212 {
1213 if (!current->euid)
1214 mode >>= 6;
1215 else if (in_egroup_p(0))
1216 mode >>= 3;
1217 if ((mode & op & 0007) == op)
1218 return 0;
1219 return -EACCES;
1220 }
1221
1222 static inline int ctl_perm(ctl_table *table, int op)
1223 {
1224 int error;
1225 error = security_sysctl(table, op);
1226 if (error)
1227 return error;
1228 return test_perm(table->mode, op);
1229 }
1230
1231 #ifdef CONFIG_SYSCTL_SYSCALL
1232 static int parse_table(int __user *name, int nlen,
1233 void __user *oldval, size_t __user *oldlenp,
1234 void __user *newval, size_t newlen,
1235 ctl_table *table)
1236 {
1237 int n;
1238 repeat:
1239 if (!nlen)
1240 return -ENOTDIR;
1241 if (get_user(n, name))
1242 return -EFAULT;
1243 for ( ; table->ctl_name || table->procname; table++) {
1244 if (!table->ctl_name)
1245 continue;
1246 if (n == table->ctl_name || table->ctl_name == CTL_ANY) {
1247 int error;
1248 if (table->child) {
1249 if (ctl_perm(table, 001))
1250 return -EPERM;
1251 if (table->strategy) {
1252 error = table->strategy(
1253 table, name, nlen,
1254 oldval, oldlenp,
1255 newval, newlen);
1256 if (error)
1257 return error;
1258 }
1259 name++;
1260 nlen--;
1261 table = table->child;
1262 goto repeat;
1263 }
1264 error = do_sysctl_strategy(table, name, nlen,
1265 oldval, oldlenp,
1266 newval, newlen);
1267 return error;
1268 }
1269 }
1270 return -ENOTDIR;
1271 }
1272
1273 /* Perform the actual read/write of a sysctl table entry. */
1274 int do_sysctl_strategy (ctl_table *table,
1275 int __user *name, int nlen,
1276 void __user *oldval, size_t __user *oldlenp,
1277 void __user *newval, size_t newlen)
1278 {
1279 int op = 0, rc;
1280 size_t len;
1281
1282 if (oldval)
1283 op |= 004;
1284 if (newval)
1285 op |= 002;
1286 if (ctl_perm(table, op))
1287 return -EPERM;
1288
1289 if (table->strategy) {
1290 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1291 newval, newlen);
1292 if (rc < 0)
1293 return rc;
1294 if (rc > 0)
1295 return 0;
1296 }
1297
1298 /* If there is no strategy routine, or if the strategy returns
1299 * zero, proceed with automatic r/w */
1300 if (table->data && table->maxlen) {
1301 if (oldval && oldlenp) {
1302 if (get_user(len, oldlenp))
1303 return -EFAULT;
1304 if (len) {
1305 if (len > table->maxlen)
1306 len = table->maxlen;
1307 if(copy_to_user(oldval, table->data, len))
1308 return -EFAULT;
1309 if(put_user(len, oldlenp))
1310 return -EFAULT;
1311 }
1312 }
1313 if (newval && newlen) {
1314 len = newlen;
1315 if (len > table->maxlen)
1316 len = table->maxlen;
1317 if(copy_from_user(table->data, newval, len))
1318 return -EFAULT;
1319 }
1320 }
1321 return 0;
1322 }
1323 #endif /* CONFIG_SYSCTL_SYSCALL */
1324
1325 /**
1326 * register_sysctl_table - register a sysctl hierarchy
1327 * @table: the top-level table structure
1328 * @insert_at_head: whether the entry should be inserted in front or at the end
1329 *
1330 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1331 * array. An entry with a ctl_name of 0 terminates the table.
1332 *
1333 * The members of the &ctl_table structure are used as follows:
1334 *
1335 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1336 * must be unique within that level of sysctl
1337 *
1338 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1339 * enter a sysctl file
1340 *
1341 * data - a pointer to data for use by proc_handler
1342 *
1343 * maxlen - the maximum size in bytes of the data
1344 *
1345 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1346 *
1347 * child - a pointer to the child sysctl table if this entry is a directory, or
1348 * %NULL.
1349 *
1350 * proc_handler - the text handler routine (described below)
1351 *
1352 * strategy - the strategy routine (described below)
1353 *
1354 * de - for internal use by the sysctl routines
1355 *
1356 * extra1, extra2 - extra pointers usable by the proc handler routines
1357 *
1358 * Leaf nodes in the sysctl tree will be represented by a single file
1359 * under /proc; non-leaf nodes will be represented by directories.
1360 *
1361 * sysctl(2) can automatically manage read and write requests through
1362 * the sysctl table. The data and maxlen fields of the ctl_table
1363 * struct enable minimal validation of the values being written to be
1364 * performed, and the mode field allows minimal authentication.
1365 *
1366 * More sophisticated management can be enabled by the provision of a
1367 * strategy routine with the table entry. This will be called before
1368 * any automatic read or write of the data is performed.
1369 *
1370 * The strategy routine may return
1371 *
1372 * < 0 - Error occurred (error is passed to user process)
1373 *
1374 * 0 - OK - proceed with automatic read or write.
1375 *
1376 * > 0 - OK - read or write has been done by the strategy routine, so
1377 * return immediately.
1378 *
1379 * There must be a proc_handler routine for any terminal nodes
1380 * mirrored under /proc/sys (non-terminals are handled by a built-in
1381 * directory handler). Several default handlers are available to
1382 * cover common cases -
1383 *
1384 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1385 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1386 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1387 *
1388 * It is the handler's job to read the input buffer from user memory
1389 * and process it. The handler should return 0 on success.
1390 *
1391 * This routine returns %NULL on a failure to register, and a pointer
1392 * to the table header on success.
1393 */
1394 struct ctl_table_header *register_sysctl_table(ctl_table * table,
1395 int insert_at_head)
1396 {
1397 struct ctl_table_header *tmp;
1398 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1399 if (!tmp)
1400 return NULL;
1401 tmp->ctl_table = table;
1402 INIT_LIST_HEAD(&tmp->ctl_entry);
1403 tmp->used = 0;
1404 tmp->unregistering = NULL;
1405 spin_lock(&sysctl_lock);
1406 if (insert_at_head)
1407 list_add(&tmp->ctl_entry, &root_table_header.ctl_entry);
1408 else
1409 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
1410 spin_unlock(&sysctl_lock);
1411 #ifdef CONFIG_PROC_SYSCTL
1412 register_proc_table(table, proc_sys_root, tmp);
1413 #endif
1414 return tmp;
1415 }
1416
1417 /**
1418 * unregister_sysctl_table - unregister a sysctl table hierarchy
1419 * @header: the header returned from register_sysctl_table
1420 *
1421 * Unregisters the sysctl table and all children. proc entries may not
1422 * actually be removed until they are no longer used by anyone.
1423 */
1424 void unregister_sysctl_table(struct ctl_table_header * header)
1425 {
1426 might_sleep();
1427 spin_lock(&sysctl_lock);
1428 start_unregistering(header);
1429 #ifdef CONFIG_PROC_SYSCTL
1430 unregister_proc_table(header->ctl_table, proc_sys_root);
1431 #endif
1432 spin_unlock(&sysctl_lock);
1433 kfree(header);
1434 }
1435
1436 #else /* !CONFIG_SYSCTL */
1437 struct ctl_table_header * register_sysctl_table(ctl_table * table,
1438 int insert_at_head)
1439 {
1440 return NULL;
1441 }
1442
1443 void unregister_sysctl_table(struct ctl_table_header * table)
1444 {
1445 }
1446
1447 #endif /* CONFIG_SYSCTL */
1448
1449 /*
1450 * /proc/sys support
1451 */
1452
1453 #ifdef CONFIG_PROC_SYSCTL
1454
1455 /* Scan the sysctl entries in table and add them all into /proc */
1456 static void register_proc_table(ctl_table * table, struct proc_dir_entry *root, void *set)
1457 {
1458 struct proc_dir_entry *de;
1459 int len;
1460 mode_t mode;
1461
1462 for (; table->ctl_name || table->procname; table++) {
1463 /* Can't do anything without a proc name. */
1464 if (!table->procname)
1465 continue;
1466 /* Maybe we can't do anything with it... */
1467 if (!table->proc_handler && !table->child) {
1468 printk(KERN_WARNING "SYSCTL: Can't register %s\n",
1469 table->procname);
1470 continue;
1471 }
1472
1473 len = strlen(table->procname);
1474 mode = table->mode;
1475
1476 de = NULL;
1477 if (table->proc_handler)
1478 mode |= S_IFREG;
1479 else {
1480 mode |= S_IFDIR;
1481 for (de = root->subdir; de; de = de->next) {
1482 if (proc_match(len, table->procname, de))
1483 break;
1484 }
1485 /* If the subdir exists already, de is non-NULL */
1486 }
1487
1488 if (!de) {
1489 de = create_proc_entry(table->procname, mode, root);
1490 if (!de)
1491 continue;
1492 de->set = set;
1493 de->data = (void *) table;
1494 if (table->proc_handler)
1495 de->proc_fops = &proc_sys_file_operations;
1496 }
1497 table->de = de;
1498 if (de->mode & S_IFDIR)
1499 register_proc_table(table->child, de, set);
1500 }
1501 }
1502
1503 /*
1504 * Unregister a /proc sysctl table and any subdirectories.
1505 */
1506 static void unregister_proc_table(ctl_table * table, struct proc_dir_entry *root)
1507 {
1508 struct proc_dir_entry *de;
1509 for (; table->ctl_name || table->procname; table++) {
1510 if (!(de = table->de))
1511 continue;
1512 if (de->mode & S_IFDIR) {
1513 if (!table->child) {
1514 printk (KERN_ALERT "Help - malformed sysctl tree on free\n");
1515 continue;
1516 }
1517 unregister_proc_table(table->child, de);
1518
1519 /* Don't unregister directories which still have entries.. */
1520 if (de->subdir)
1521 continue;
1522 }
1523
1524 /*
1525 * In any case, mark the entry as goner; we'll keep it
1526 * around if it's busy, but we'll know to do nothing with
1527 * its fields. We are under sysctl_lock here.
1528 */
1529 de->data = NULL;
1530
1531 /* Don't unregister proc entries that are still being used.. */
1532 if (atomic_read(&de->count))
1533 continue;
1534
1535 table->de = NULL;
1536 remove_proc_entry(table->procname, root);
1537 }
1538 }
1539
1540 static ssize_t do_rw_proc(int write, struct file * file, char __user * buf,
1541 size_t count, loff_t *ppos)
1542 {
1543 int op;
1544 struct proc_dir_entry *de = PDE(file->f_path.dentry->d_inode);
1545 struct ctl_table *table;
1546 size_t res;
1547 ssize_t error = -ENOTDIR;
1548
1549 spin_lock(&sysctl_lock);
1550 if (de && de->data && use_table(de->set)) {
1551 /*
1552 * at that point we know that sysctl was not unregistered
1553 * and won't be until we finish
1554 */
1555 spin_unlock(&sysctl_lock);
1556 table = (struct ctl_table *) de->data;
1557 if (!table || !table->proc_handler)
1558 goto out;
1559 error = -EPERM;
1560 op = (write ? 002 : 004);
1561 if (ctl_perm(table, op))
1562 goto out;
1563
1564 /* careful: calling conventions are nasty here */
1565 res = count;
1566 error = (*table->proc_handler)(table, write, file,
1567 buf, &res, ppos);
1568 if (!error)
1569 error = res;
1570 out:
1571 spin_lock(&sysctl_lock);
1572 unuse_table(de->set);
1573 }
1574 spin_unlock(&sysctl_lock);
1575 return error;
1576 }
1577
1578 static int proc_opensys(struct inode *inode, struct file *file)
1579 {
1580 if (file->f_mode & FMODE_WRITE) {
1581 /*
1582 * sysctl entries that are not writable,
1583 * are _NOT_ writable, capabilities or not.
1584 */
1585 if (!(inode->i_mode & S_IWUSR))
1586 return -EPERM;
1587 }
1588
1589 return 0;
1590 }
1591
1592 static ssize_t proc_readsys(struct file * file, char __user * buf,
1593 size_t count, loff_t *ppos)
1594 {
1595 return do_rw_proc(0, file, buf, count, ppos);
1596 }
1597
1598 static ssize_t proc_writesys(struct file * file, const char __user * buf,
1599 size_t count, loff_t *ppos)
1600 {
1601 return do_rw_proc(1, file, (char __user *) buf, count, ppos);
1602 }
1603
1604 static int _proc_do_string(void* data, int maxlen, int write,
1605 struct file *filp, void __user *buffer,
1606 size_t *lenp, loff_t *ppos)
1607 {
1608 size_t len;
1609 char __user *p;
1610 char c;
1611
1612 if (!data || !maxlen || !*lenp) {
1613 *lenp = 0;
1614 return 0;
1615 }
1616
1617 if (write) {
1618 len = 0;
1619 p = buffer;
1620 while (len < *lenp) {
1621 if (get_user(c, p++))
1622 return -EFAULT;
1623 if (c == 0 || c == '\n')
1624 break;
1625 len++;
1626 }
1627 if (len >= maxlen)
1628 len = maxlen-1;
1629 if(copy_from_user(data, buffer, len))
1630 return -EFAULT;
1631 ((char *) data)[len] = 0;
1632 *ppos += *lenp;
1633 } else {
1634 len = strlen(data);
1635 if (len > maxlen)
1636 len = maxlen;
1637
1638 if (*ppos > len) {
1639 *lenp = 0;
1640 return 0;
1641 }
1642
1643 data += *ppos;
1644 len -= *ppos;
1645
1646 if (len > *lenp)
1647 len = *lenp;
1648 if (len)
1649 if(copy_to_user(buffer, data, len))
1650 return -EFAULT;
1651 if (len < *lenp) {
1652 if(put_user('\n', ((char __user *) buffer) + len))
1653 return -EFAULT;
1654 len++;
1655 }
1656 *lenp = len;
1657 *ppos += len;
1658 }
1659 return 0;
1660 }
1661
1662 /**
1663 * proc_dostring - read a string sysctl
1664 * @table: the sysctl table
1665 * @write: %TRUE if this is a write to the sysctl file
1666 * @filp: the file structure
1667 * @buffer: the user buffer
1668 * @lenp: the size of the user buffer
1669 * @ppos: file position
1670 *
1671 * Reads/writes a string from/to the user buffer. If the kernel
1672 * buffer provided is not large enough to hold the string, the
1673 * string is truncated. The copied string is %NULL-terminated.
1674 * If the string is being read by the user process, it is copied
1675 * and a newline '\n' is added. It is truncated if the buffer is
1676 * not large enough.
1677 *
1678 * Returns 0 on success.
1679 */
1680 int proc_dostring(ctl_table *table, int write, struct file *filp,
1681 void __user *buffer, size_t *lenp, loff_t *ppos)
1682 {
1683 return _proc_do_string(table->data, table->maxlen, write, filp,
1684 buffer, lenp, ppos);
1685 }
1686
1687
1688 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1689 int *valp,
1690 int write, void *data)
1691 {
1692 if (write) {
1693 *valp = *negp ? -*lvalp : *lvalp;
1694 } else {
1695 int val = *valp;
1696 if (val < 0) {
1697 *negp = -1;
1698 *lvalp = (unsigned long)-val;
1699 } else {
1700 *negp = 0;
1701 *lvalp = (unsigned long)val;
1702 }
1703 }
1704 return 0;
1705 }
1706
1707 static int __do_proc_dointvec(void *tbl_data, ctl_table *table,
1708 int write, struct file *filp, void __user *buffer,
1709 size_t *lenp, loff_t *ppos,
1710 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1711 int write, void *data),
1712 void *data)
1713 {
1714 #define TMPBUFLEN 21
1715 int *i, vleft, first=1, neg, val;
1716 unsigned long lval;
1717 size_t left, len;
1718
1719 char buf[TMPBUFLEN], *p;
1720 char __user *s = buffer;
1721
1722 if (!tbl_data || !table->maxlen || !*lenp ||
1723 (*ppos && !write)) {
1724 *lenp = 0;
1725 return 0;
1726 }
1727
1728 i = (int *) tbl_data;
1729 vleft = table->maxlen / sizeof(*i);
1730 left = *lenp;
1731
1732 if (!conv)
1733 conv = do_proc_dointvec_conv;
1734
1735 for (; left && vleft--; i++, first=0) {
1736 if (write) {
1737 while (left) {
1738 char c;
1739 if (get_user(c, s))
1740 return -EFAULT;
1741 if (!isspace(c))
1742 break;
1743 left--;
1744 s++;
1745 }
1746 if (!left)
1747 break;
1748 neg = 0;
1749 len = left;
1750 if (len > sizeof(buf) - 1)
1751 len = sizeof(buf) - 1;
1752 if (copy_from_user(buf, s, len))
1753 return -EFAULT;
1754 buf[len] = 0;
1755 p = buf;
1756 if (*p == '-' && left > 1) {
1757 neg = 1;
1758 p++;
1759 }
1760 if (*p < '0' || *p > '9')
1761 break;
1762
1763 lval = simple_strtoul(p, &p, 0);
1764
1765 len = p-buf;
1766 if ((len < left) && *p && !isspace(*p))
1767 break;
1768 if (neg)
1769 val = -val;
1770 s += len;
1771 left -= len;
1772
1773 if (conv(&neg, &lval, i, 1, data))
1774 break;
1775 } else {
1776 p = buf;
1777 if (!first)
1778 *p++ = '\t';
1779
1780 if (conv(&neg, &lval, i, 0, data))
1781 break;
1782
1783 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1784 len = strlen(buf);
1785 if (len > left)
1786 len = left;
1787 if(copy_to_user(s, buf, len))
1788 return -EFAULT;
1789 left -= len;
1790 s += len;
1791 }
1792 }
1793
1794 if (!write && !first && left) {
1795 if(put_user('\n', s))
1796 return -EFAULT;
1797 left--, s++;
1798 }
1799 if (write) {
1800 while (left) {
1801 char c;
1802 if (get_user(c, s++))
1803 return -EFAULT;
1804 if (!isspace(c))
1805 break;
1806 left--;
1807 }
1808 }
1809 if (write && first)
1810 return -EINVAL;
1811 *lenp -= left;
1812 *ppos += *lenp;
1813 return 0;
1814 #undef TMPBUFLEN
1815 }
1816
1817 static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1818 void __user *buffer, size_t *lenp, loff_t *ppos,
1819 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1820 int write, void *data),
1821 void *data)
1822 {
1823 return __do_proc_dointvec(table->data, table, write, filp,
1824 buffer, lenp, ppos, conv, data);
1825 }
1826
1827 /**
1828 * proc_dointvec - read a vector of integers
1829 * @table: the sysctl table
1830 * @write: %TRUE if this is a write to the sysctl file
1831 * @filp: the file structure
1832 * @buffer: the user buffer
1833 * @lenp: the size of the user buffer
1834 * @ppos: file position
1835 *
1836 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1837 * values from/to the user buffer, treated as an ASCII string.
1838 *
1839 * Returns 0 on success.
1840 */
1841 int proc_dointvec(ctl_table *table, int write, struct file *filp,
1842 void __user *buffer, size_t *lenp, loff_t *ppos)
1843 {
1844 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1845 NULL,NULL);
1846 }
1847
1848 #define OP_SET 0
1849 #define OP_AND 1
1850 #define OP_OR 2
1851
1852 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1853 int *valp,
1854 int write, void *data)
1855 {
1856 int op = *(int *)data;
1857 if (write) {
1858 int val = *negp ? -*lvalp : *lvalp;
1859 switch(op) {
1860 case OP_SET: *valp = val; break;
1861 case OP_AND: *valp &= val; break;
1862 case OP_OR: *valp |= val; break;
1863 }
1864 } else {
1865 int val = *valp;
1866 if (val < 0) {
1867 *negp = -1;
1868 *lvalp = (unsigned long)-val;
1869 } else {
1870 *negp = 0;
1871 *lvalp = (unsigned long)val;
1872 }
1873 }
1874 return 0;
1875 }
1876
1877 /*
1878 * init may raise the set.
1879 */
1880
1881 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1882 void __user *buffer, size_t *lenp, loff_t *ppos)
1883 {
1884 int op;
1885
1886 if (write && !capable(CAP_SYS_MODULE)) {
1887 return -EPERM;
1888 }
1889
1890 op = is_init(current) ? OP_SET : OP_AND;
1891 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1892 do_proc_dointvec_bset_conv,&op);
1893 }
1894
1895 /*
1896 * Taint values can only be increased
1897 */
1898 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
1899 void __user *buffer, size_t *lenp, loff_t *ppos)
1900 {
1901 int op;
1902
1903 if (!capable(CAP_SYS_ADMIN))
1904 return -EPERM;
1905
1906 op = OP_OR;
1907 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1908 do_proc_dointvec_bset_conv,&op);
1909 }
1910
1911 struct do_proc_dointvec_minmax_conv_param {
1912 int *min;
1913 int *max;
1914 };
1915
1916 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1917 int *valp,
1918 int write, void *data)
1919 {
1920 struct do_proc_dointvec_minmax_conv_param *param = data;
1921 if (write) {
1922 int val = *negp ? -*lvalp : *lvalp;
1923 if ((param->min && *param->min > val) ||
1924 (param->max && *param->max < val))
1925 return -EINVAL;
1926 *valp = val;
1927 } else {
1928 int val = *valp;
1929 if (val < 0) {
1930 *negp = -1;
1931 *lvalp = (unsigned long)-val;
1932 } else {
1933 *negp = 0;
1934 *lvalp = (unsigned long)val;
1935 }
1936 }
1937 return 0;
1938 }
1939
1940 /**
1941 * proc_dointvec_minmax - read a vector of integers with min/max values
1942 * @table: the sysctl table
1943 * @write: %TRUE if this is a write to the sysctl file
1944 * @filp: the file structure
1945 * @buffer: the user buffer
1946 * @lenp: the size of the user buffer
1947 * @ppos: file position
1948 *
1949 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1950 * values from/to the user buffer, treated as an ASCII string.
1951 *
1952 * This routine will ensure the values are within the range specified by
1953 * table->extra1 (min) and table->extra2 (max).
1954 *
1955 * Returns 0 on success.
1956 */
1957 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1958 void __user *buffer, size_t *lenp, loff_t *ppos)
1959 {
1960 struct do_proc_dointvec_minmax_conv_param param = {
1961 .min = (int *) table->extra1,
1962 .max = (int *) table->extra2,
1963 };
1964 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1965 do_proc_dointvec_minmax_conv, &param);
1966 }
1967
1968 static int __do_proc_doulongvec_minmax(void *data, ctl_table *table, int write,
1969 struct file *filp,
1970 void __user *buffer,
1971 size_t *lenp, loff_t *ppos,
1972 unsigned long convmul,
1973 unsigned long convdiv)
1974 {
1975 #define TMPBUFLEN 21
1976 unsigned long *i, *min, *max, val;
1977 int vleft, first=1, neg;
1978 size_t len, left;
1979 char buf[TMPBUFLEN], *p;
1980 char __user *s = buffer;
1981
1982 if (!data || !table->maxlen || !*lenp ||
1983 (*ppos && !write)) {
1984 *lenp = 0;
1985 return 0;
1986 }
1987
1988 i = (unsigned long *) data;
1989 min = (unsigned long *) table->extra1;
1990 max = (unsigned long *) table->extra2;
1991 vleft = table->maxlen / sizeof(unsigned long);
1992 left = *lenp;
1993
1994 for (; left && vleft--; i++, min++, max++, first=0) {
1995 if (write) {
1996 while (left) {
1997 char c;
1998 if (get_user(c, s))
1999 return -EFAULT;
2000 if (!isspace(c))
2001 break;
2002 left--;
2003 s++;
2004 }
2005 if (!left)
2006 break;
2007 neg = 0;
2008 len = left;
2009 if (len > TMPBUFLEN-1)
2010 len = TMPBUFLEN-1;
2011 if (copy_from_user(buf, s, len))
2012 return -EFAULT;
2013 buf[len] = 0;
2014 p = buf;
2015 if (*p == '-' && left > 1) {
2016 neg = 1;
2017 p++;
2018 }
2019 if (*p < '0' || *p > '9')
2020 break;
2021 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2022 len = p-buf;
2023 if ((len < left) && *p && !isspace(*p))
2024 break;
2025 if (neg)
2026 val = -val;
2027 s += len;
2028 left -= len;
2029
2030 if(neg)
2031 continue;
2032 if ((min && val < *min) || (max && val > *max))
2033 continue;
2034 *i = val;
2035 } else {
2036 p = buf;
2037 if (!first)
2038 *p++ = '\t';
2039 sprintf(p, "%lu", convdiv * (*i) / convmul);
2040 len = strlen(buf);
2041 if (len > left)
2042 len = left;
2043 if(copy_to_user(s, buf, len))
2044 return -EFAULT;
2045 left -= len;
2046 s += len;
2047 }
2048 }
2049
2050 if (!write && !first && left) {
2051 if(put_user('\n', s))
2052 return -EFAULT;
2053 left--, s++;
2054 }
2055 if (write) {
2056 while (left) {
2057 char c;
2058 if (get_user(c, s++))
2059 return -EFAULT;
2060 if (!isspace(c))
2061 break;
2062 left--;
2063 }
2064 }
2065 if (write && first)
2066 return -EINVAL;
2067 *lenp -= left;
2068 *ppos += *lenp;
2069 return 0;
2070 #undef TMPBUFLEN
2071 }
2072
2073 static int do_proc_doulongvec_minmax(ctl_table *table, int write,
2074 struct file *filp,
2075 void __user *buffer,
2076 size_t *lenp, loff_t *ppos,
2077 unsigned long convmul,
2078 unsigned long convdiv)
2079 {
2080 return __do_proc_doulongvec_minmax(table->data, table, write,
2081 filp, buffer, lenp, ppos, convmul, convdiv);
2082 }
2083
2084 /**
2085 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2086 * @table: the sysctl table
2087 * @write: %TRUE if this is a write to the sysctl file
2088 * @filp: the file structure
2089 * @buffer: the user buffer
2090 * @lenp: the size of the user buffer
2091 * @ppos: file position
2092 *
2093 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2094 * values from/to the user buffer, treated as an ASCII string.
2095 *
2096 * This routine will ensure the values are within the range specified by
2097 * table->extra1 (min) and table->extra2 (max).
2098 *
2099 * Returns 0 on success.
2100 */
2101 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2102 void __user *buffer, size_t *lenp, loff_t *ppos)
2103 {
2104 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2105 }
2106
2107 /**
2108 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2109 * @table: the sysctl table
2110 * @write: %TRUE if this is a write to the sysctl file
2111 * @filp: the file structure
2112 * @buffer: the user buffer
2113 * @lenp: the size of the user buffer
2114 * @ppos: file position
2115 *
2116 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2117 * values from/to the user buffer, treated as an ASCII string. The values
2118 * are treated as milliseconds, and converted to jiffies when they are stored.
2119 *
2120 * This routine will ensure the values are within the range specified by
2121 * table->extra1 (min) and table->extra2 (max).
2122 *
2123 * Returns 0 on success.
2124 */
2125 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2126 struct file *filp,
2127 void __user *buffer,
2128 size_t *lenp, loff_t *ppos)
2129 {
2130 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2131 lenp, ppos, HZ, 1000l);
2132 }
2133
2134
2135 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2136 int *valp,
2137 int write, void *data)
2138 {
2139 if (write) {
2140 if (*lvalp > LONG_MAX / HZ)
2141 return 1;
2142 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2143 } else {
2144 int val = *valp;
2145 unsigned long lval;
2146 if (val < 0) {
2147 *negp = -1;
2148 lval = (unsigned long)-val;
2149 } else {
2150 *negp = 0;
2151 lval = (unsigned long)val;
2152 }
2153 *lvalp = lval / HZ;
2154 }
2155 return 0;
2156 }
2157
2158 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2159 int *valp,
2160 int write, void *data)
2161 {
2162 if (write) {
2163 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2164 return 1;
2165 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2166 } else {
2167 int val = *valp;
2168 unsigned long lval;
2169 if (val < 0) {
2170 *negp = -1;
2171 lval = (unsigned long)-val;
2172 } else {
2173 *negp = 0;
2174 lval = (unsigned long)val;
2175 }
2176 *lvalp = jiffies_to_clock_t(lval);
2177 }
2178 return 0;
2179 }
2180
2181 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2182 int *valp,
2183 int write, void *data)
2184 {
2185 if (write) {
2186 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2187 } else {
2188 int val = *valp;
2189 unsigned long lval;
2190 if (val < 0) {
2191 *negp = -1;
2192 lval = (unsigned long)-val;
2193 } else {
2194 *negp = 0;
2195 lval = (unsigned long)val;
2196 }
2197 *lvalp = jiffies_to_msecs(lval);
2198 }
2199 return 0;
2200 }
2201
2202 /**
2203 * proc_dointvec_jiffies - read a vector of integers as seconds
2204 * @table: the sysctl table
2205 * @write: %TRUE if this is a write to the sysctl file
2206 * @filp: the file structure
2207 * @buffer: the user buffer
2208 * @lenp: the size of the user buffer
2209 * @ppos: file position
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 seconds, and are converted into
2214 * jiffies.
2215 *
2216 * Returns 0 on success.
2217 */
2218 int proc_dointvec_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_jiffies_conv,NULL);
2223 }
2224
2225 /**
2226 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2227 * @table: the sysctl table
2228 * @write: %TRUE if this is a write to the sysctl file
2229 * @filp: the file structure
2230 * @buffer: the user buffer
2231 * @lenp: the size of the user buffer
2232 * @ppos: pointer to the file position
2233 *
2234 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2235 * values from/to the user buffer, treated as an ASCII string.
2236 * The values read are assumed to be in 1/USER_HZ seconds, and
2237 * are converted into jiffies.
2238 *
2239 * Returns 0 on success.
2240 */
2241 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2242 void __user *buffer, size_t *lenp, loff_t *ppos)
2243 {
2244 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2245 do_proc_dointvec_userhz_jiffies_conv,NULL);
2246 }
2247
2248 /**
2249 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2250 * @table: the sysctl table
2251 * @write: %TRUE if this is a write to the sysctl file
2252 * @filp: the file structure
2253 * @buffer: the user buffer
2254 * @lenp: the size of the user buffer
2255 * @ppos: file position
2256 * @ppos: the current position in the file
2257 *
2258 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2259 * values from/to the user buffer, treated as an ASCII string.
2260 * The values read are assumed to be in 1/1000 seconds, and
2261 * are converted into jiffies.
2262 *
2263 * Returns 0 on success.
2264 */
2265 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2266 void __user *buffer, size_t *lenp, loff_t *ppos)
2267 {
2268 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2269 do_proc_dointvec_ms_jiffies_conv, NULL);
2270 }
2271
2272 #ifdef CONFIG_SYSVIPC
2273 static int proc_ipc_dointvec(ctl_table *table, int write, struct file *filp,
2274 void __user *buffer, size_t *lenp, loff_t *ppos)
2275 {
2276 void *which;
2277 which = get_ipc(table, write);
2278 return __do_proc_dointvec(which, table, write, filp, buffer,
2279 lenp, ppos, NULL, NULL);
2280 }
2281
2282 static int proc_ipc_doulongvec_minmax(ctl_table *table, int write,
2283 struct file *filp, void __user *buffer, size_t *lenp, loff_t *ppos)
2284 {
2285 void *which;
2286 which = get_ipc(table, write);
2287 return __do_proc_doulongvec_minmax(which, table, write, filp, buffer,
2288 lenp, ppos, 1l, 1l);
2289 }
2290
2291 #endif
2292
2293 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
2294 void __user *buffer, size_t *lenp, loff_t *ppos)
2295 {
2296 struct pid *new_pid;
2297 pid_t tmp;
2298 int r;
2299
2300 tmp = pid_nr(cad_pid);
2301
2302 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2303 lenp, ppos, NULL, NULL);
2304 if (r || !write)
2305 return r;
2306
2307 new_pid = find_get_pid(tmp);
2308 if (!new_pid)
2309 return -ESRCH;
2310
2311 put_pid(xchg(&cad_pid, new_pid));
2312 return 0;
2313 }
2314
2315 #else /* CONFIG_PROC_FS */
2316
2317 int proc_dostring(ctl_table *table, int write, struct file *filp,
2318 void __user *buffer, size_t *lenp, loff_t *ppos)
2319 {
2320 return -ENOSYS;
2321 }
2322
2323 #ifdef CONFIG_SYSVIPC
2324 static int proc_do_ipc_string(ctl_table *table, int write, struct file *filp,
2325 void __user *buffer, size_t *lenp, loff_t *ppos)
2326 {
2327 return -ENOSYS;
2328 }
2329 static int proc_ipc_dointvec(ctl_table *table, int write, struct file *filp,
2330 void __user *buffer, size_t *lenp, loff_t *ppos)
2331 {
2332 return -ENOSYS;
2333 }
2334 static int proc_ipc_doulongvec_minmax(ctl_table *table, int write,
2335 struct file *filp, void __user *buffer,
2336 size_t *lenp, loff_t *ppos)
2337 {
2338 return -ENOSYS;
2339 }
2340 #endif
2341
2342 int proc_dointvec(ctl_table *table, int write, struct file *filp,
2343 void __user *buffer, size_t *lenp, loff_t *ppos)
2344 {
2345 return -ENOSYS;
2346 }
2347
2348 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2349 void __user *buffer, size_t *lenp, loff_t *ppos)
2350 {
2351 return -ENOSYS;
2352 }
2353
2354 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2355 void __user *buffer, size_t *lenp, loff_t *ppos)
2356 {
2357 return -ENOSYS;
2358 }
2359
2360 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2361 void __user *buffer, size_t *lenp, loff_t *ppos)
2362 {
2363 return -ENOSYS;
2364 }
2365
2366 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2367 void __user *buffer, size_t *lenp, loff_t *ppos)
2368 {
2369 return -ENOSYS;
2370 }
2371
2372 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2373 void __user *buffer, size_t *lenp, loff_t *ppos)
2374 {
2375 return -ENOSYS;
2376 }
2377
2378 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2379 void __user *buffer, size_t *lenp, loff_t *ppos)
2380 {
2381 return -ENOSYS;
2382 }
2383
2384 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2385 struct file *filp,
2386 void __user *buffer,
2387 size_t *lenp, loff_t *ppos)
2388 {
2389 return -ENOSYS;
2390 }
2391
2392
2393 #endif /* CONFIG_PROC_FS */
2394
2395
2396 #ifdef CONFIG_SYSCTL_SYSCALL
2397 /*
2398 * General sysctl support routines
2399 */
2400
2401 /* The generic string strategy routine: */
2402 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2403 void __user *oldval, size_t __user *oldlenp,
2404 void __user *newval, size_t newlen)
2405 {
2406 if (!table->data || !table->maxlen)
2407 return -ENOTDIR;
2408
2409 if (oldval && oldlenp) {
2410 size_t bufsize;
2411 if (get_user(bufsize, oldlenp))
2412 return -EFAULT;
2413 if (bufsize) {
2414 size_t len = strlen(table->data), copied;
2415
2416 /* This shouldn't trigger for a well-formed sysctl */
2417 if (len > table->maxlen)
2418 len = table->maxlen;
2419
2420 /* Copy up to a max of bufsize-1 bytes of the string */
2421 copied = (len >= bufsize) ? bufsize - 1 : len;
2422
2423 if (copy_to_user(oldval, table->data, copied) ||
2424 put_user(0, (char __user *)(oldval + copied)))
2425 return -EFAULT;
2426 if (put_user(len, oldlenp))
2427 return -EFAULT;
2428 }
2429 }
2430 if (newval && newlen) {
2431 size_t len = newlen;
2432 if (len > table->maxlen)
2433 len = table->maxlen;
2434 if(copy_from_user(table->data, newval, len))
2435 return -EFAULT;
2436 if (len == table->maxlen)
2437 len--;
2438 ((char *) table->data)[len] = 0;
2439 }
2440 return 1;
2441 }
2442
2443 /*
2444 * This function makes sure that all of the integers in the vector
2445 * are between the minimum and maximum values given in the arrays
2446 * table->extra1 and table->extra2, respectively.
2447 */
2448 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2449 void __user *oldval, size_t __user *oldlenp,
2450 void __user *newval, size_t newlen)
2451 {
2452
2453 if (newval && newlen) {
2454 int __user *vec = (int __user *) newval;
2455 int *min = (int *) table->extra1;
2456 int *max = (int *) table->extra2;
2457 size_t length;
2458 int i;
2459
2460 if (newlen % sizeof(int) != 0)
2461 return -EINVAL;
2462
2463 if (!table->extra1 && !table->extra2)
2464 return 0;
2465
2466 if (newlen > table->maxlen)
2467 newlen = table->maxlen;
2468 length = newlen / sizeof(int);
2469
2470 for (i = 0; i < length; i++) {
2471 int value;
2472 if (get_user(value, vec + i))
2473 return -EFAULT;
2474 if (min && value < min[i])
2475 return -EINVAL;
2476 if (max && value > max[i])
2477 return -EINVAL;
2478 }
2479 }
2480 return 0;
2481 }
2482
2483 /* Strategy function to convert jiffies to seconds */
2484 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2485 void __user *oldval, size_t __user *oldlenp,
2486 void __user *newval, size_t newlen)
2487 {
2488 if (oldval && oldlenp) {
2489 size_t olen;
2490
2491 if (get_user(olen, oldlenp))
2492 return -EFAULT;
2493 if (olen) {
2494 int val;
2495
2496 if (olen < sizeof(int))
2497 return -EINVAL;
2498
2499 val = *(int *)(table->data) / HZ;
2500 if (put_user(val, (int __user *)oldval))
2501 return -EFAULT;
2502 if (put_user(sizeof(int), oldlenp))
2503 return -EFAULT;
2504 }
2505 }
2506 if (newval && newlen) {
2507 int new;
2508 if (newlen != sizeof(int))
2509 return -EINVAL;
2510 if (get_user(new, (int __user *)newval))
2511 return -EFAULT;
2512 *(int *)(table->data) = new*HZ;
2513 }
2514 return 1;
2515 }
2516
2517 /* Strategy function to convert jiffies to seconds */
2518 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2519 void __user *oldval, size_t __user *oldlenp,
2520 void __user *newval, size_t newlen)
2521 {
2522 if (oldval && oldlenp) {
2523 size_t olen;
2524
2525 if (get_user(olen, oldlenp))
2526 return -EFAULT;
2527 if (olen) {
2528 int val;
2529
2530 if (olen < sizeof(int))
2531 return -EINVAL;
2532
2533 val = jiffies_to_msecs(*(int *)(table->data));
2534 if (put_user(val, (int __user *)oldval))
2535 return -EFAULT;
2536 if (put_user(sizeof(int), oldlenp))
2537 return -EFAULT;
2538 }
2539 }
2540 if (newval && newlen) {
2541 int new;
2542 if (newlen != sizeof(int))
2543 return -EINVAL;
2544 if (get_user(new, (int __user *)newval))
2545 return -EFAULT;
2546 *(int *)(table->data) = msecs_to_jiffies(new);
2547 }
2548 return 1;
2549 }
2550
2551
2552
2553 #ifdef CONFIG_SYSVIPC
2554 /* The generic sysctl ipc data routine. */
2555 static int sysctl_ipc_data(ctl_table *table, int __user *name, int nlen,
2556 void __user *oldval, size_t __user *oldlenp,
2557 void __user *newval, size_t newlen)
2558 {
2559 size_t len;
2560 void *data;
2561
2562 /* Get out of I don't have a variable */
2563 if (!table->data || !table->maxlen)
2564 return -ENOTDIR;
2565
2566 data = get_ipc(table, 1);
2567 if (!data)
2568 return -ENOTDIR;
2569
2570 if (oldval && oldlenp) {
2571 if (get_user(len, oldlenp))
2572 return -EFAULT;
2573 if (len) {
2574 if (len > table->maxlen)
2575 len = table->maxlen;
2576 if (copy_to_user(oldval, data, len))
2577 return -EFAULT;
2578 if (put_user(len, oldlenp))
2579 return -EFAULT;
2580 }
2581 }
2582
2583 if (newval && newlen) {
2584 if (newlen > table->maxlen)
2585 newlen = table->maxlen;
2586
2587 if (copy_from_user(data, newval, newlen))
2588 return -EFAULT;
2589 }
2590 return 1;
2591 }
2592 #endif
2593
2594 #else /* CONFIG_SYSCTL_SYSCALL */
2595
2596
2597 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2598 {
2599 static int msg_count;
2600 struct __sysctl_args tmp;
2601 int name[CTL_MAXNAME];
2602 int i;
2603
2604 /* Read in the sysctl name for better debug message logging */
2605 if (copy_from_user(&tmp, args, sizeof(tmp)))
2606 return -EFAULT;
2607 if (tmp.nlen <= 0 || tmp.nlen >= CTL_MAXNAME)
2608 return -ENOTDIR;
2609 for (i = 0; i < tmp.nlen; i++)
2610 if (get_user(name[i], tmp.name + i))
2611 return -EFAULT;
2612
2613 /* Ignore accesses to kernel.version */
2614 if ((tmp.nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2615 goto out;
2616
2617 if (msg_count < 5) {
2618 msg_count++;
2619 printk(KERN_INFO
2620 "warning: process `%s' used the removed sysctl "
2621 "system call with ", current->comm);
2622 for (i = 0; i < tmp.nlen; i++)
2623 printk("%d.", name[i]);
2624 printk("\n");
2625 }
2626 out:
2627 return -ENOSYS;
2628 }
2629
2630 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2631 void __user *oldval, size_t __user *oldlenp,
2632 void __user *newval, size_t newlen)
2633 {
2634 return -ENOSYS;
2635 }
2636
2637 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2638 void __user *oldval, size_t __user *oldlenp,
2639 void __user *newval, size_t newlen)
2640 {
2641 return -ENOSYS;
2642 }
2643
2644 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2645 void __user *oldval, size_t __user *oldlenp,
2646 void __user *newval, size_t newlen)
2647 {
2648 return -ENOSYS;
2649 }
2650
2651 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2652 void __user *oldval, size_t __user *oldlenp,
2653 void __user *newval, size_t newlen)
2654 {
2655 return -ENOSYS;
2656 }
2657
2658 #ifdef CONFIG_SYSVIPC
2659 static int sysctl_ipc_data(ctl_table *table, int __user *name, int nlen,
2660 void __user *oldval, size_t __user *oldlenp,
2661 void __user *newval, size_t newlen)
2662 {
2663 return -ENOSYS;
2664 }
2665 #endif
2666 #endif /* CONFIG_SYSCTL_SYSCALL */
2667
2668 /*
2669 * No sense putting this after each symbol definition, twice,
2670 * exception granted :-)
2671 */
2672 EXPORT_SYMBOL(proc_dointvec);
2673 EXPORT_SYMBOL(proc_dointvec_jiffies);
2674 EXPORT_SYMBOL(proc_dointvec_minmax);
2675 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2676 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2677 EXPORT_SYMBOL(proc_dostring);
2678 EXPORT_SYMBOL(proc_doulongvec_minmax);
2679 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2680 EXPORT_SYMBOL(register_sysctl_table);
2681 EXPORT_SYMBOL(sysctl_intvec);
2682 EXPORT_SYMBOL(sysctl_jiffies);
2683 EXPORT_SYMBOL(sysctl_ms_jiffies);
2684 EXPORT_SYMBOL(sysctl_string);
2685 EXPORT_SYMBOL(unregister_sysctl_table);
x86 "subsystem" repository