projects / linux-2.6-block.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
[PATCH] sysctl: remove insert_at_head from register_sysctl
[linux-2.6-block.git] / kernel / sysctl.c
CommitLineData
1da177e4
LT		 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
1da177e4
LT		 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>
c59ede7b 27#include <linux/capability.h>
1da177e4
LT		 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>
0296b228 34#include <linux/kobject.h>
20380731 35#include <linux/net.h>
1da177e4
LT		 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>
c255d844
PM		 46#include <linux/nfs_fs.h>
47#include <linux/acpi.h>
1da177e4
LT		 48
49#include <asm/uaccess.h>
50#include <asm/processor.h>
51
529bf6be
DS		 52extern int proc_nr_files(ctl_table *table, int write, struct file *filp,
53 void __user *buffer, size_t *lenp, loff_t *ppos);
54
29cbc78b
AK		 55#ifdef CONFIG_X86
56#include <asm/nmi.h>
0741f4d2 57#include <asm/stacktrace.h>
29cbc78b
AK		 58#endif
59
1da177e4
LT		 60#if defined(CONFIG_SYSCTL)
61
62/* External variables not in a header file. */
63extern int C_A_D;
64extern int sysctl_overcommit_memory;
65extern int sysctl_overcommit_ratio;
fadd8fbd 66extern int sysctl_panic_on_oom;
1da177e4 67extern int max_threads;
1da177e4 68extern int core_uses_pid;
d6e71144 69extern int suid_dumpable;
1da177e4 70extern char core_pattern[];
1da177e4
LT		 71extern int pid_max;
72extern int min_free_kbytes;
73extern int printk_ratelimit_jiffies;
74extern int printk_ratelimit_burst;
75extern int pid_max_min, pid_max_max;
9d0243bc 76extern int sysctl_drop_caches;
8ad4b1fb 77extern int percpu_pagelist_fraction;
bebfa101 78extern int compat_log;
1da177e4 79
1da177e4
LT		 80/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
81static int maxolduid = 65535;
82static int minolduid;
8ad4b1fb 83static int min_percpu_pagelist_fract = 8;
1da177e4
LT		 84
85static int ngroups_max = NGROUPS_MAX;
86
87#ifdef CONFIG_KMOD
88extern char modprobe_path[];
89#endif
1da177e4
LT		 90#ifdef CONFIG_CHR_DEV_SG
91extern int sg_big_buff;
92#endif
1da177e4
LT		 93
94#ifdef __sparc__
95extern char reboot_command [];
96extern int stop_a_enabled;
97extern int scons_pwroff;
98#endif
99
100#ifdef __hppa__
101extern int pwrsw_enabled;
102extern int unaligned_enabled;
103#endif
104
347a8dc3 105#ifdef CONFIG_S390
1da177e4
LT		 106#ifdef CONFIG_MATHEMU
107extern int sysctl_ieee_emulation_warnings;
108#endif
109extern int sysctl_userprocess_debug;
951f22d5 110extern int spin_retry;
1da177e4
LT		 111#endif
112
113extern int sysctl_hz_timer;
114
115#ifdef CONFIG_BSD_PROCESS_ACCT
116extern int acct_parm[];
117#endif
118
d2b176ed
JS		 119#ifdef CONFIG_IA64
120extern int no_unaligned_warning;
121#endif
122
23f78d4a
IM		 123#ifdef CONFIG_RT_MUTEXES
124extern int max_lock_depth;
125#endif
126
b89a8171
EB		 127#ifdef CONFIG_SYSCTL_SYSCALL
128static int parse_table(int __user *, int, void __user *, size_t __user *,
1f29bcd7 129 void __user *, size_t, ctl_table *);
b89a8171
EB		 130#endif
131
6b49a257 132
d6f8ff73 133#ifdef CONFIG_PROC_SYSCTL
9ec52099
CLG		 134static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
135 void __user *buffer, size_t *lenp, loff_t *ppos);
34f5a398
TT		 136static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
137 void __user *buffer, size_t *lenp, loff_t *ppos);
d6f8ff73 138#endif
9ec52099 139
1da177e4
LT		 140static ctl_table root_table[];
141static struct ctl_table_header root_table_header =
142 { root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
143
144static ctl_table kern_table[];
145static ctl_table vm_table[];
1da177e4
LT		 146static ctl_table fs_table[];
147static ctl_table debug_table[];
148static ctl_table dev_table[];
149extern ctl_table random_table[];
150#ifdef CONFIG_UNIX98_PTYS
151extern ctl_table pty_table[];
152#endif
2d9048e2 153#ifdef CONFIG_INOTIFY_USER
0399cb08
RL		 154extern ctl_table inotify_table[];
155#endif
1da177e4
LT		 156
157#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
158int sysctl_legacy_va_layout;
159#endif
160
9bc9a6bd 161
1da177e4
LT		 162/* /proc declarations: */
163
b89a8171 164#ifdef CONFIG_PROC_SYSCTL
1da177e4
LT		 165
166static ssize_t proc_readsys(struct file *, char __user *, size_t, loff_t *);
167static ssize_t proc_writesys(struct file *, const char __user *, size_t, loff_t *);
168static int proc_opensys(struct inode *, struct file *);
169
15ad7cdc 170const struct file_operations proc_sys_file_operations = {
1da177e4
LT		 171 .open = proc_opensys,
172 .read = proc_readsys,
173 .write = proc_writesys,
174};
175
176extern struct proc_dir_entry *proc_sys_root;
177
330d57fb 178static void register_proc_table(ctl_table *, struct proc_dir_entry *, void *);
1da177e4
LT		 179static void unregister_proc_table(ctl_table *, struct proc_dir_entry *);
180#endif
181
182/* The default sysctl tables: */
183
184static ctl_table root_table[] = {
185 {
186 .ctl_name = CTL_KERN,
187 .procname = "kernel",
188 .mode = 0555,
189 .child = kern_table,
190 },
191 {
192 .ctl_name = CTL_VM,
193 .procname = "vm",
194 .mode = 0555,
195 .child = vm_table,
196 },
197#ifdef CONFIG_NET
198 {
199 .ctl_name = CTL_NET,
200 .procname = "net",
201 .mode = 0555,
202 .child = net_table,
203 },
204#endif
1da177e4
LT		 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 },
0eeca283 223
1da177e4
LT		 224 { .ctl_name = 0 }
225};
226
227static ctl_table kern_table[] = {
1da177e4
LT		 228 {
229 .ctl_name = KERN_PANIC,
230 .procname = "panic",
231 .data = &panic_timeout,
232 .maxlen = sizeof(int),
233 .mode = 0644,
234 .proc_handler = &proc_dointvec,
235 },
236 {
237 .ctl_name = KERN_CORE_USES_PID,
238 .procname = "core_uses_pid",
239 .data = &core_uses_pid,
240 .maxlen = sizeof(int),
241 .mode = 0644,
242 .proc_handler = &proc_dointvec,
243 },
244 {
245 .ctl_name = KERN_CORE_PATTERN,
246 .procname = "core_pattern",
247 .data = core_pattern,
d025c9db 248 .maxlen = 128,
1da177e4
LT		 249 .mode = 0644,
250 .proc_handler = &proc_dostring,
251 .strategy = &sysctl_string,
252 },
34f5a398 253#ifdef CONFIG_PROC_SYSCTL
1da177e4
LT		 254 {
255 .ctl_name = KERN_TAINTED,
256 .procname = "tainted",
257 .data = &tainted,
258 .maxlen = sizeof(int),
34f5a398
TT		 259 .mode = 0644,
260 .proc_handler = &proc_dointvec_taint,
1da177e4 261 },
34f5a398 262#endif
1da177e4
LT		 263 {
264 .ctl_name = KERN_CAP_BSET,
265 .procname = "cap-bound",
266 .data = &cap_bset,
267 .maxlen = sizeof(kernel_cap_t),
268 .mode = 0600,
269 .proc_handler = &proc_dointvec_bset,
270 },
271#ifdef CONFIG_BLK_DEV_INITRD
272 {
273 .ctl_name = KERN_REALROOTDEV,
274 .procname = "real-root-dev",
275 .data = &real_root_dev,
276 .maxlen = sizeof(int),
277 .mode = 0644,
278 .proc_handler = &proc_dointvec,
279 },
280#endif
281#ifdef __sparc__
282 {
283 .ctl_name = KERN_SPARC_REBOOT,
284 .procname = "reboot-cmd",
285 .data = reboot_command,
286 .maxlen = 256,
287 .mode = 0644,
288 .proc_handler = &proc_dostring,
289 .strategy = &sysctl_string,
290 },
291 {
292 .ctl_name = KERN_SPARC_STOP_A,
293 .procname = "stop-a",
294 .data = &stop_a_enabled,
295 .maxlen = sizeof (int),
296 .mode = 0644,
297 .proc_handler = &proc_dointvec,
298 },
299 {
300 .ctl_name = KERN_SPARC_SCONS_PWROFF,
301 .procname = "scons-poweroff",
302 .data = &scons_pwroff,
303 .maxlen = sizeof (int),
304 .mode = 0644,
305 .proc_handler = &proc_dointvec,
306 },
307#endif
308#ifdef __hppa__
309 {
310 .ctl_name = KERN_HPPA_PWRSW,
311 .procname = "soft-power",
312 .data = &pwrsw_enabled,
313 .maxlen = sizeof (int),
314 .mode = 0644,
315 .proc_handler = &proc_dointvec,
316 },
317 {
318 .ctl_name = KERN_HPPA_UNALIGNED,
319 .procname = "unaligned-trap",
320 .data = &unaligned_enabled,
321 .maxlen = sizeof (int),
322 .mode = 0644,
323 .proc_handler = &proc_dointvec,
324 },
325#endif
326 {
327 .ctl_name = KERN_CTLALTDEL,
328 .procname = "ctrl-alt-del",
329 .data = &C_A_D,
330 .maxlen = sizeof(int),
331 .mode = 0644,
332 .proc_handler = &proc_dointvec,
333 },
334 {
335 .ctl_name = KERN_PRINTK,
336 .procname = "printk",
337 .data = &console_loglevel,
338 .maxlen = 4*sizeof(int),
339 .mode = 0644,
340 .proc_handler = &proc_dointvec,
341 },
342#ifdef CONFIG_KMOD
343 {
344 .ctl_name = KERN_MODPROBE,
345 .procname = "modprobe",
346 .data = &modprobe_path,
347 .maxlen = KMOD_PATH_LEN,
348 .mode = 0644,
349 .proc_handler = &proc_dostring,
350 .strategy = &sysctl_string,
351 },
352#endif
57ae2508 353#if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
1da177e4
LT		 354 {
355 .ctl_name = KERN_HOTPLUG,
356 .procname = "hotplug",
312c004d
KS		 357 .data = &uevent_helper,
358 .maxlen = UEVENT_HELPER_PATH_LEN,
1da177e4
LT		 359 .mode = 0644,
360 .proc_handler = &proc_dostring,
361 .strategy = &sysctl_string,
362 },
363#endif
364#ifdef CONFIG_CHR_DEV_SG
365 {
366 .ctl_name = KERN_SG_BIG_BUFF,
367 .procname = "sg-big-buff",
368 .data = &sg_big_buff,
369 .maxlen = sizeof (int),
370 .mode = 0444,
371 .proc_handler = &proc_dointvec,
372 },
373#endif
374#ifdef CONFIG_BSD_PROCESS_ACCT
375 {
376 .ctl_name = KERN_ACCT,
377 .procname = "acct",
378 .data = &acct_parm,
379 .maxlen = 3*sizeof(int),
380 .mode = 0644,
381 .proc_handler = &proc_dointvec,
382 },
383#endif
1da177e4
LT		 384#ifdef CONFIG_MAGIC_SYSRQ
385 {
386 .ctl_name = KERN_SYSRQ,
387 .procname = "sysrq",
5d6f647f 388 .data = &__sysrq_enabled,
1da177e4
LT		 389 .maxlen = sizeof (int),
390 .mode = 0644,
391 .proc_handler = &proc_dointvec,
392 },
393#endif
d6f8ff73 394#ifdef CONFIG_PROC_SYSCTL
1da177e4
LT		 395 {
396 .ctl_name = KERN_CADPID,
397 .procname = "cad_pid",
9ec52099 398 .data = NULL,
1da177e4
LT		 399 .maxlen = sizeof (int),
400 .mode = 0600,
9ec52099 401 .proc_handler = &proc_do_cad_pid,
1da177e4 402 },
d6f8ff73 403#endif
1da177e4
LT		 404 {
405 .ctl_name = KERN_MAX_THREADS,
406 .procname = "threads-max",
407 .data = &max_threads,
408 .maxlen = sizeof(int),
409 .mode = 0644,
410 .proc_handler = &proc_dointvec,
411 },
412 {
413 .ctl_name = KERN_RANDOM,
414 .procname = "random",
415 .mode = 0555,
416 .child = random_table,
417 },
418#ifdef CONFIG_UNIX98_PTYS
419 {
420 .ctl_name = KERN_PTY,
421 .procname = "pty",
422 .mode = 0555,
423 .child = pty_table,
424 },
425#endif
426 {
427 .ctl_name = KERN_OVERFLOWUID,
428 .procname = "overflowuid",
429 .data = &overflowuid,
430 .maxlen = sizeof(int),
431 .mode = 0644,
432 .proc_handler = &proc_dointvec_minmax,
433 .strategy = &sysctl_intvec,
434 .extra1 = &minolduid,
435 .extra2 = &maxolduid,
436 },
437 {
438 .ctl_name = KERN_OVERFLOWGID,
439 .procname = "overflowgid",
440 .data = &overflowgid,
441 .maxlen = sizeof(int),
442 .mode = 0644,
443 .proc_handler = &proc_dointvec_minmax,
444 .strategy = &sysctl_intvec,
445 .extra1 = &minolduid,
446 .extra2 = &maxolduid,
447 },
347a8dc3 448#ifdef CONFIG_S390
1da177e4
LT		 449#ifdef CONFIG_MATHEMU
450 {
451 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
452 .procname = "ieee_emulation_warnings",
453 .data = &sysctl_ieee_emulation_warnings,
454 .maxlen = sizeof(int),
455 .mode = 0644,
456 .proc_handler = &proc_dointvec,
457 },
458#endif
459#ifdef CONFIG_NO_IDLE_HZ
460 {
461 .ctl_name = KERN_HZ_TIMER,
462 .procname = "hz_timer",
463 .data = &sysctl_hz_timer,
464 .maxlen = sizeof(int),
465 .mode = 0644,
466 .proc_handler = &proc_dointvec,
467 },
468#endif
469 {
470 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
471 .procname = "userprocess_debug",
472 .data = &sysctl_userprocess_debug,
473 .maxlen = sizeof(int),
474 .mode = 0644,
475 .proc_handler = &proc_dointvec,
476 },
477#endif
478 {
479 .ctl_name = KERN_PIDMAX,
480 .procname = "pid_max",
481 .data = &pid_max,
482 .maxlen = sizeof (int),
483 .mode = 0644,
484 .proc_handler = &proc_dointvec_minmax,
485 .strategy = sysctl_intvec,
486 .extra1 = &pid_max_min,
487 .extra2 = &pid_max_max,
488 },
489 {
490 .ctl_name = KERN_PANIC_ON_OOPS,
491 .procname = "panic_on_oops",
492 .data = &panic_on_oops,
493 .maxlen = sizeof(int),
494 .mode = 0644,
495 .proc_handler = &proc_dointvec,
496 },
497 {
498 .ctl_name = KERN_PRINTK_RATELIMIT,
499 .procname = "printk_ratelimit",
500 .data = &printk_ratelimit_jiffies,
501 .maxlen = sizeof(int),
502 .mode = 0644,
503 .proc_handler = &proc_dointvec_jiffies,
504 .strategy = &sysctl_jiffies,
505 },
506 {
507 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
508 .procname = "printk_ratelimit_burst",
509 .data = &printk_ratelimit_burst,
510 .maxlen = sizeof(int),
511 .mode = 0644,
512 .proc_handler = &proc_dointvec,
513 },
514 {
515 .ctl_name = KERN_NGROUPS_MAX,
516 .procname = "ngroups_max",
517 .data = &ngroups_max,
518 .maxlen = sizeof (int),
519 .mode = 0444,
520 .proc_handler = &proc_dointvec,
521 },
522#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
523 {
524 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
525 .procname = "unknown_nmi_panic",
526 .data = &unknown_nmi_panic,
527 .maxlen = sizeof (int),
528 .mode = 0644,
2fbe7b25 529 .proc_handler = &proc_dointvec,
1da177e4 530 },
407984f1
DZ		 531 {
532 .ctl_name = KERN_NMI_WATCHDOG,
533 .procname = "nmi_watchdog",
534 .data = &nmi_watchdog_enabled,
535 .maxlen = sizeof (int),
536 .mode = 0644,
537 .proc_handler = &proc_nmi_enabled,
1da177e4
LT		 538 },
539#endif
540#if defined(CONFIG_X86)
8da5adda
DZ		 541 {
542 .ctl_name = KERN_PANIC_ON_NMI,
543 .procname = "panic_on_unrecovered_nmi",
544 .data = &panic_on_unrecovered_nmi,
545 .maxlen = sizeof(int),
546 .mode = 0644,
547 .proc_handler = &proc_dointvec,
548 },
1da177e4
LT		 549 {
550 .ctl_name = KERN_BOOTLOADER_TYPE,
551 .procname = "bootloader_type",
552 .data = &bootloader_type,
553 .maxlen = sizeof (int),
554 .mode = 0444,
555 .proc_handler = &proc_dointvec,
556 },
0741f4d2
CE		 557 {
558 .ctl_name = CTL_UNNUMBERED,
559 .procname = "kstack_depth_to_print",
560 .data = &kstack_depth_to_print,
561 .maxlen = sizeof(int),
562 .mode = 0644,
563 .proc_handler = &proc_dointvec,
564 },
1da177e4 565#endif
7a9166e3 566#if defined(CONFIG_MMU)
1da177e4
LT		 567 {
568 .ctl_name = KERN_RANDOMIZE,
569 .procname = "randomize_va_space",
570 .data = &randomize_va_space,
571 .maxlen = sizeof(int),
572 .mode = 0644,
573 .proc_handler = &proc_dointvec,
574 },
7a9166e3 575#endif
0152fb37 576#if defined(CONFIG_S390) && defined(CONFIG_SMP)
951f22d5
MS		 577 {
578 .ctl_name = KERN_SPIN_RETRY,
579 .procname = "spin_retry",
580 .data = &spin_retry,
581 .maxlen = sizeof (int),
582 .mode = 0644,
583 .proc_handler = &proc_dointvec,
584 },
c255d844
PM		 585#endif
586#ifdef CONFIG_ACPI_SLEEP
587 {
588 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
589 .procname = "acpi_video_flags",
590 .data = &acpi_video_flags,
591 .maxlen = sizeof (unsigned long),
592 .mode = 0644,
7f99f06f 593 .proc_handler = &proc_doulongvec_minmax,
c255d844 594 },
d2b176ed
JS		 595#endif
596#ifdef CONFIG_IA64
597 {
598 .ctl_name = KERN_IA64_UNALIGNED,
599 .procname = "ignore-unaligned-usertrap",
600 .data = &no_unaligned_warning,
601 .maxlen = sizeof (int),
602 .mode = 0644,
603 .proc_handler = &proc_dointvec,
604 },
bebfa101
AK		 605#endif
606#ifdef CONFIG_COMPAT
607 {
608 .ctl_name = KERN_COMPAT_LOG,
609 .procname = "compat-log",
610 .data = &compat_log,
611 .maxlen = sizeof (int),
612 .mode = 0644,
613 .proc_handler = &proc_dointvec,
614 },
951f22d5 615#endif
23f78d4a
IM		 616#ifdef CONFIG_RT_MUTEXES
617 {
618 .ctl_name = KERN_MAX_LOCK_DEPTH,
619 .procname = "max_lock_depth",
620 .data = &max_lock_depth,
621 .maxlen = sizeof(int),
622 .mode = 0644,
623 .proc_handler = &proc_dointvec,
624 },
625#endif
626
1da177e4
LT		 627 { .ctl_name = 0 }
628};
629
630/* Constants for minimum and maximum testing in vm_table.
631 We use these as one-element integer vectors. */
632static int zero;
633static int one_hundred = 100;
634
635
636static ctl_table vm_table[] = {
637 {
638 .ctl_name = VM_OVERCOMMIT_MEMORY,
639 .procname = "overcommit_memory",
640 .data = &sysctl_overcommit_memory,
641 .maxlen = sizeof(sysctl_overcommit_memory),
642 .mode = 0644,
643 .proc_handler = &proc_dointvec,
644 },
fadd8fbd
KH		 645 {
646 .ctl_name = VM_PANIC_ON_OOM,
647 .procname = "panic_on_oom",
648 .data = &sysctl_panic_on_oom,
649 .maxlen = sizeof(sysctl_panic_on_oom),
650 .mode = 0644,
651 .proc_handler = &proc_dointvec,
652 },
1da177e4
LT		 653 {
654 .ctl_name = VM_OVERCOMMIT_RATIO,
655 .procname = "overcommit_ratio",
656 .data = &sysctl_overcommit_ratio,
657 .maxlen = sizeof(sysctl_overcommit_ratio),
658 .mode = 0644,
659 .proc_handler = &proc_dointvec,
660 },
661 {
662 .ctl_name = VM_PAGE_CLUSTER,
663 .procname = "page-cluster",
664 .data = &page_cluster,
665 .maxlen = sizeof(int),
666 .mode = 0644,
667 .proc_handler = &proc_dointvec,
668 },
669 {
670 .ctl_name = VM_DIRTY_BACKGROUND,
671 .procname = "dirty_background_ratio",
672 .data = &dirty_background_ratio,
673 .maxlen = sizeof(dirty_background_ratio),
674 .mode = 0644,
675 .proc_handler = &proc_dointvec_minmax,
676 .strategy = &sysctl_intvec,
677 .extra1 = &zero,
678 .extra2 = &one_hundred,
679 },
680 {
681 .ctl_name = VM_DIRTY_RATIO,
682 .procname = "dirty_ratio",
683 .data = &vm_dirty_ratio,
684 .maxlen = sizeof(vm_dirty_ratio),
685 .mode = 0644,
686 .proc_handler = &proc_dointvec_minmax,
687 .strategy = &sysctl_intvec,
688 .extra1 = &zero,
689 .extra2 = &one_hundred,
690 },
691 {
692 .ctl_name = VM_DIRTY_WB_CS,
693 .procname = "dirty_writeback_centisecs",
f6ef9438
BS		 694 .data = &dirty_writeback_interval,
695 .maxlen = sizeof(dirty_writeback_interval),
1da177e4
LT		 696 .mode = 0644,
697 .proc_handler = &dirty_writeback_centisecs_handler,
698 },
699 {
700 .ctl_name = VM_DIRTY_EXPIRE_CS,
701 .procname = "dirty_expire_centisecs",
f6ef9438
BS		 702 .data = &dirty_expire_interval,
703 .maxlen = sizeof(dirty_expire_interval),
1da177e4 704 .mode = 0644,
f6ef9438 705 .proc_handler = &proc_dointvec_userhz_jiffies,
1da177e4
LT		 706 },
707 {
708 .ctl_name = VM_NR_PDFLUSH_THREADS,
709 .procname = "nr_pdflush_threads",
710 .data = &nr_pdflush_threads,
711 .maxlen = sizeof nr_pdflush_threads,
712 .mode = 0444 /* read-only*/,
713 .proc_handler = &proc_dointvec,
714 },
715 {
716 .ctl_name = VM_SWAPPINESS,
717 .procname = "swappiness",
718 .data = &vm_swappiness,
719 .maxlen = sizeof(vm_swappiness),
720 .mode = 0644,
721 .proc_handler = &proc_dointvec_minmax,
722 .strategy = &sysctl_intvec,
723 .extra1 = &zero,
724 .extra2 = &one_hundred,
725 },
726#ifdef CONFIG_HUGETLB_PAGE
727 {
728 .ctl_name = VM_HUGETLB_PAGES,
729 .procname = "nr_hugepages",
730 .data = &max_huge_pages,
731 .maxlen = sizeof(unsigned long),
732 .mode = 0644,
733 .proc_handler = &hugetlb_sysctl_handler,
734 .extra1 = (void *)&hugetlb_zero,
735 .extra2 = (void *)&hugetlb_infinity,
736 },
737 {
738 .ctl_name = VM_HUGETLB_GROUP,
739 .procname = "hugetlb_shm_group",
740 .data = &sysctl_hugetlb_shm_group,
741 .maxlen = sizeof(gid_t),
742 .mode = 0644,
743 .proc_handler = &proc_dointvec,
744 },
745#endif
746 {
747 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
748 .procname = "lowmem_reserve_ratio",
749 .data = &sysctl_lowmem_reserve_ratio,
750 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
751 .mode = 0644,
752 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
753 .strategy = &sysctl_intvec,
754 },
9d0243bc
AM		 755 {
756 .ctl_name = VM_DROP_PAGECACHE,
757 .procname = "drop_caches",
758 .data = &sysctl_drop_caches,
759 .maxlen = sizeof(int),
760 .mode = 0644,
761 .proc_handler = drop_caches_sysctl_handler,
762 .strategy = &sysctl_intvec,
763 },
1da177e4
LT		 764 {
765 .ctl_name = VM_MIN_FREE_KBYTES,
766 .procname = "min_free_kbytes",
767 .data = &min_free_kbytes,
768 .maxlen = sizeof(min_free_kbytes),
769 .mode = 0644,
770 .proc_handler = &min_free_kbytes_sysctl_handler,
771 .strategy = &sysctl_intvec,
772 .extra1 = &zero,
773 },
8ad4b1fb
RS		 774 {
775 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
776 .procname = "percpu_pagelist_fraction",
777 .data = &percpu_pagelist_fraction,
778 .maxlen = sizeof(percpu_pagelist_fraction),
779 .mode = 0644,
780 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
781 .strategy = &sysctl_intvec,
782 .extra1 = &min_percpu_pagelist_fract,
783 },
1da177e4
LT		 784#ifdef CONFIG_MMU
785 {
786 .ctl_name = VM_MAX_MAP_COUNT,
787 .procname = "max_map_count",
788 .data = &sysctl_max_map_count,
789 .maxlen = sizeof(sysctl_max_map_count),
790 .mode = 0644,
791 .proc_handler = &proc_dointvec
792 },
793#endif
794 {
795 .ctl_name = VM_LAPTOP_MODE,
796 .procname = "laptop_mode",
797 .data = &laptop_mode,
798 .maxlen = sizeof(laptop_mode),
799 .mode = 0644,
ed5b43f1
BS		 800 .proc_handler = &proc_dointvec_jiffies,
801 .strategy = &sysctl_jiffies,
1da177e4
LT		 802 },
803 {
804 .ctl_name = VM_BLOCK_DUMP,
805 .procname = "block_dump",
806 .data = &block_dump,
807 .maxlen = sizeof(block_dump),
808 .mode = 0644,
809 .proc_handler = &proc_dointvec,
810 .strategy = &sysctl_intvec,
811 .extra1 = &zero,
812 },
813 {
814 .ctl_name = VM_VFS_CACHE_PRESSURE,
815 .procname = "vfs_cache_pressure",
816 .data = &sysctl_vfs_cache_pressure,
817 .maxlen = sizeof(sysctl_vfs_cache_pressure),
818 .mode = 0644,
819 .proc_handler = &proc_dointvec,
820 .strategy = &sysctl_intvec,
821 .extra1 = &zero,
822 },
823#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
824 {
825 .ctl_name = VM_LEGACY_VA_LAYOUT,
826 .procname = "legacy_va_layout",
827 .data = &sysctl_legacy_va_layout,
828 .maxlen = sizeof(sysctl_legacy_va_layout),
829 .mode = 0644,
830 .proc_handler = &proc_dointvec,
831 .strategy = &sysctl_intvec,
832 .extra1 = &zero,
833 },
834#endif
1743660b
CL		 835#ifdef CONFIG_NUMA
836 {
837 .ctl_name = VM_ZONE_RECLAIM_MODE,
838 .procname = "zone_reclaim_mode",
839 .data = &zone_reclaim_mode,
840 .maxlen = sizeof(zone_reclaim_mode),
841 .mode = 0644,
842 .proc_handler = &proc_dointvec,
c84db23c
CL		 843 .strategy = &sysctl_intvec,
844 .extra1 = &zero,
1743660b 845 },
9614634f
CL		 846 {
847 .ctl_name = VM_MIN_UNMAPPED,
848 .procname = "min_unmapped_ratio",
849 .data = &sysctl_min_unmapped_ratio,
850 .maxlen = sizeof(sysctl_min_unmapped_ratio),
851 .mode = 0644,
852 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
853 .strategy = &sysctl_intvec,
854 .extra1 = &zero,
855 .extra2 = &one_hundred,
856 },
0ff38490
CL		 857 {
858 .ctl_name = VM_MIN_SLAB,
859 .procname = "min_slab_ratio",
860 .data = &sysctl_min_slab_ratio,
861 .maxlen = sizeof(sysctl_min_slab_ratio),
862 .mode = 0644,
863 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
864 .strategy = &sysctl_intvec,
865 .extra1 = &zero,
866 .extra2 = &one_hundred,
867 },
e6e5494c
IM		 868#endif
869#ifdef CONFIG_X86_32
870 {
871 .ctl_name = VM_VDSO_ENABLED,
872 .procname = "vdso_enabled",
873 .data = &vdso_enabled,
874 .maxlen = sizeof(vdso_enabled),
875 .mode = 0644,
876 .proc_handler = &proc_dointvec,
877 .strategy = &sysctl_intvec,
878 .extra1 = &zero,
879 },
1da177e4
LT		 880#endif
881 { .ctl_name = 0 }
882};
883
2abc26fc
EB		 884#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
885static ctl_table binfmt_misc_table[] = {
886 { .ctl_name = 0 }
887};
888#endif
889
1da177e4
LT		 890static ctl_table fs_table[] = {
891 {
892 .ctl_name = FS_NRINODE,
893 .procname = "inode-nr",
894 .data = &inodes_stat,
895 .maxlen = 2*sizeof(int),
896 .mode = 0444,
897 .proc_handler = &proc_dointvec,
898 },
899 {
900 .ctl_name = FS_STATINODE,
901 .procname = "inode-state",
902 .data = &inodes_stat,
903 .maxlen = 7*sizeof(int),
904 .mode = 0444,
905 .proc_handler = &proc_dointvec,
906 },
907 {
908 .ctl_name = FS_NRFILE,
909 .procname = "file-nr",
910 .data = &files_stat,
911 .maxlen = 3*sizeof(int),
912 .mode = 0444,
529bf6be 913 .proc_handler = &proc_nr_files,
1da177e4
LT		 914 },
915 {
916 .ctl_name = FS_MAXFILE,
917 .procname = "file-max",
918 .data = &files_stat.max_files,
919 .maxlen = sizeof(int),
920 .mode = 0644,
921 .proc_handler = &proc_dointvec,
922 },
923 {
924 .ctl_name = FS_DENTRY,
925 .procname = "dentry-state",
926 .data = &dentry_stat,
927 .maxlen = 6*sizeof(int),
928 .mode = 0444,
929 .proc_handler = &proc_dointvec,
930 },
931 {
932 .ctl_name = FS_OVERFLOWUID,
933 .procname = "overflowuid",
934 .data = &fs_overflowuid,
935 .maxlen = sizeof(int),
936 .mode = 0644,
937 .proc_handler = &proc_dointvec_minmax,
938 .strategy = &sysctl_intvec,
939 .extra1 = &minolduid,
940 .extra2 = &maxolduid,
941 },
942 {
943 .ctl_name = FS_OVERFLOWGID,
944 .procname = "overflowgid",
945 .data = &fs_overflowgid,
946 .maxlen = sizeof(int),
947 .mode = 0644,
948 .proc_handler = &proc_dointvec_minmax,
949 .strategy = &sysctl_intvec,
950 .extra1 = &minolduid,
951 .extra2 = &maxolduid,
952 },
953 {
954 .ctl_name = FS_LEASES,
955 .procname = "leases-enable",
956 .data = &leases_enable,
957 .maxlen = sizeof(int),
958 .mode = 0644,
959 .proc_handler = &proc_dointvec,
960 },
961#ifdef CONFIG_DNOTIFY
962 {
963 .ctl_name = FS_DIR_NOTIFY,
964 .procname = "dir-notify-enable",
965 .data = &dir_notify_enable,
966 .maxlen = sizeof(int),
967 .mode = 0644,
968 .proc_handler = &proc_dointvec,
969 },
970#endif
971#ifdef CONFIG_MMU
972 {
973 .ctl_name = FS_LEASE_TIME,
974 .procname = "lease-break-time",
975 .data = &lease_break_time,
976 .maxlen = sizeof(int),
977 .mode = 0644,
978 .proc_handler = &proc_dointvec,
979 },
980 {
981 .ctl_name = FS_AIO_NR,
982 .procname = "aio-nr",
983 .data = &aio_nr,
984 .maxlen = sizeof(aio_nr),
985 .mode = 0444,
d55b5fda 986 .proc_handler = &proc_doulongvec_minmax,
1da177e4
LT		 987 },
988 {
989 .ctl_name = FS_AIO_MAX_NR,
990 .procname = "aio-max-nr",
991 .data = &aio_max_nr,
992 .maxlen = sizeof(aio_max_nr),
993 .mode = 0644,
d55b5fda 994 .proc_handler = &proc_doulongvec_minmax,
1da177e4 995 },
2d9048e2 996#ifdef CONFIG_INOTIFY_USER
0399cb08
RL		 997 {
998 .ctl_name = FS_INOTIFY,
999 .procname = "inotify",
1000 .mode = 0555,
1001 .child = inotify_table,
1002 },
1003#endif
1da177e4 1004#endif
d6e71144
AC		 1005 {
1006 .ctl_name = KERN_SETUID_DUMPABLE,
1007 .procname = "suid_dumpable",
1008 .data = &suid_dumpable,
1009 .maxlen = sizeof(int),
1010 .mode = 0644,
1011 .proc_handler = &proc_dointvec,
1012 },
2abc26fc
EB		 1013#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1014 {
1015 .ctl_name = CTL_UNNUMBERED,
1016 .procname = "binfmt_misc",
1017 .mode = 0555,
1018 .child = binfmt_misc_table,
1019 },
1020#endif
1da177e4
LT		 1021 { .ctl_name = 0 }
1022};
1023
1024static ctl_table debug_table[] = {
1025 { .ctl_name = 0 }
1026};
1027
1028static ctl_table dev_table[] = {
1029 { .ctl_name = 0 }
0eeca283 1030};
1da177e4 1031
330d57fb
AV		 1032static DEFINE_SPINLOCK(sysctl_lock);
1033
1034/* called under sysctl_lock */
1035static int use_table(struct ctl_table_header *p)
1036{
1037 if (unlikely(p->unregistering))
1038 return 0;
1039 p->used++;
1040 return 1;
1041}
1042
1043/* called under sysctl_lock */
1044static void unuse_table(struct ctl_table_header *p)
1045{
1046 if (!--p->used)
1047 if (unlikely(p->unregistering))
1048 complete(p->unregistering);
1049}
1050
1051/* called under sysctl_lock, will reacquire if has to wait */
1052static void start_unregistering(struct ctl_table_header *p)
1053{
1054 /*
1055 * if p->used is 0, nobody will ever touch that entry again;
1056 * we'll eliminate all paths to it before dropping sysctl_lock
1057 */
1058 if (unlikely(p->used)) {
1059 struct completion wait;
1060 init_completion(&wait);
1061 p->unregistering = &wait;
1062 spin_unlock(&sysctl_lock);
1063 wait_for_completion(&wait);
1064 spin_lock(&sysctl_lock);
1065 }
1066 /*
1067 * do not remove from the list until nobody holds it; walking the
1068 * list in do_sysctl() relies on that.
1069 */
1070 list_del_init(&p->ctl_entry);
1071}
1072
1da177e4
LT		 1073void __init sysctl_init(void)
1074{
b89a8171 1075#ifdef CONFIG_PROC_SYSCTL
330d57fb 1076 register_proc_table(root_table, proc_sys_root, &root_table_header);
1da177e4
LT		 1077#endif
1078}
1079
b89a8171 1080#ifdef CONFIG_SYSCTL_SYSCALL
1da177e4
LT		 1081int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1082 void __user *newval, size_t newlen)
1083{
1084 struct list_head *tmp;
330d57fb 1085 int error = -ENOTDIR;
1da177e4
LT		 1086
1087 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1088 return -ENOTDIR;
1089 if (oldval) {
1090 int old_len;
1091 if (!oldlenp || get_user(old_len, oldlenp))
1092 return -EFAULT;
1093 }
330d57fb 1094 spin_lock(&sysctl_lock);
1da177e4
LT		 1095 tmp = &root_table_header.ctl_entry;
1096 do {
1097 struct ctl_table_header *head =
1098 list_entry(tmp, struct ctl_table_header, ctl_entry);
330d57fb
AV		 1099
1100 if (!use_table(head))
1101 continue;
1102
1103 spin_unlock(&sysctl_lock);
1104
1105 error = parse_table(name, nlen, oldval, oldlenp,
1f29bcd7 1106 newval, newlen, head->ctl_table);
330d57fb
AV		 1107
1108 spin_lock(&sysctl_lock);
1109 unuse_table(head);
1da177e4 1110 if (error != -ENOTDIR)
330d57fb
AV		 1111 break;
1112 } while ((tmp = tmp->next) != &root_table_header.ctl_entry);
1113 spin_unlock(&sysctl_lock);
1114 return error;
1da177e4
LT		 1115}
1116
1117asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1118{
1119 struct __sysctl_args tmp;
1120 int error;
1121
1122 if (copy_from_user(&tmp, args, sizeof(tmp)))
1123 return -EFAULT;
1124
1125 lock_kernel();
1126 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1127 tmp.newval, tmp.newlen);
1128 unlock_kernel();
1129 return error;
1130}
b89a8171 1131#endif /* CONFIG_SYSCTL_SYSCALL */
1da177e4
LT		 1132
1133/*
1134 * ctl_perm does NOT grant the superuser all rights automatically, because
1135 * some sysctl variables are readonly even to root.
1136 */
1137
1138static int test_perm(int mode, int op)
1139{
1140 if (!current->euid)
1141 mode >>= 6;
1142 else if (in_egroup_p(0))
1143 mode >>= 3;
1144 if ((mode & op & 0007) == op)
1145 return 0;
1146 return -EACCES;
1147}
1148
1149static inline int ctl_perm(ctl_table *table, int op)
1150{
1151 int error;
1152 error = security_sysctl(table, op);
1153 if (error)
1154 return error;
1155 return test_perm(table->mode, op);
1156}
1157
b89a8171 1158#ifdef CONFIG_SYSCTL_SYSCALL
1da177e4
LT		 1159static int parse_table(int __user *name, int nlen,
1160 void __user *oldval, size_t __user *oldlenp,
1161 void __user *newval, size_t newlen,
1f29bcd7 1162 ctl_table *table)
1da177e4
LT		 1163{
1164 int n;
1165repeat:
1166 if (!nlen)
1167 return -ENOTDIR;
1168 if (get_user(n, name))
1169 return -EFAULT;
d99f160a
EB		 1170 for ( ; table->ctl_name || table->procname; table++) {
1171 if (!table->ctl_name)
1172 continue;
6703ddfc 1173 if (n == table->ctl_name) {
1da177e4
LT		 1174 int error;
1175 if (table->child) {
1176 if (ctl_perm(table, 001))
1177 return -EPERM;
1da177e4
LT		 1178 name++;
1179 nlen--;
1180 table = table->child;
1181 goto repeat;
1182 }
1183 error = do_sysctl_strategy(table, name, nlen,
1184 oldval, oldlenp,
1f29bcd7 1185 newval, newlen);
1da177e4
LT		 1186 return error;
1187 }
1188 }
1189 return -ENOTDIR;
1190}
1191
1192/* Perform the actual read/write of a sysctl table entry. */
1193int do_sysctl_strategy (ctl_table *table,
1194 int __user *name, int nlen,
1195 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 1196 void __user *newval, size_t newlen)
1da177e4
LT		 1197{
1198 int op = 0, rc;
1199 size_t len;
1200
1201 if (oldval)
1202 op |= 004;
1203 if (newval)
1204 op |= 002;
1205 if (ctl_perm(table, op))
1206 return -EPERM;
1207
1208 if (table->strategy) {
1209 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1f29bcd7 1210 newval, newlen);
1da177e4
LT		 1211 if (rc < 0)
1212 return rc;
1213 if (rc > 0)
1214 return 0;
1215 }
1216
1217 /* If there is no strategy routine, or if the strategy returns
1218 * zero, proceed with automatic r/w */
1219 if (table->data && table->maxlen) {
1220 if (oldval && oldlenp) {
1221 if (get_user(len, oldlenp))
1222 return -EFAULT;
1223 if (len) {
1224 if (len > table->maxlen)
1225 len = table->maxlen;
1226 if(copy_to_user(oldval, table->data, len))
1227 return -EFAULT;
1228 if(put_user(len, oldlenp))
1229 return -EFAULT;
1230 }
1231 }
1232 if (newval && newlen) {
1233 len = newlen;
1234 if (len > table->maxlen)
1235 len = table->maxlen;
1236 if(copy_from_user(table->data, newval, len))
1237 return -EFAULT;
1238 }
1239 }
1240 return 0;
1241}
b89a8171 1242#endif /* CONFIG_SYSCTL_SYSCALL */
1da177e4
LT		 1243
1244/**
1245 * register_sysctl_table - register a sysctl hierarchy
1246 * @table: the top-level table structure
1da177e4
LT		 1247 *
1248 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1249 * array. An entry with a ctl_name of 0 terminates the table.
1250 *
1251 * The members of the &ctl_table structure are used as follows:
1252 *
1253 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1254 * must be unique within that level of sysctl
1255 *
1256 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1257 * enter a sysctl file
1258 *
1259 * data - a pointer to data for use by proc_handler
1260 *
1261 * maxlen - the maximum size in bytes of the data
1262 *
1263 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1264 *
1265 * child - a pointer to the child sysctl table if this entry is a directory, or
1266 * %NULL.
1267 *
1268 * proc_handler - the text handler routine (described below)
1269 *
1270 * strategy - the strategy routine (described below)
1271 *
1272 * de - for internal use by the sysctl routines
1273 *
1274 * extra1, extra2 - extra pointers usable by the proc handler routines
1275 *
1276 * Leaf nodes in the sysctl tree will be represented by a single file
1277 * under /proc; non-leaf nodes will be represented by directories.
1278 *
1279 * sysctl(2) can automatically manage read and write requests through
1280 * the sysctl table. The data and maxlen fields of the ctl_table
1281 * struct enable minimal validation of the values being written to be
1282 * performed, and the mode field allows minimal authentication.
1283 *
1284 * More sophisticated management can be enabled by the provision of a
1285 * strategy routine with the table entry. This will be called before
1286 * any automatic read or write of the data is performed.
1287 *
1288 * The strategy routine may return
1289 *
1290 * < 0 - Error occurred (error is passed to user process)
1291 *
1292 * 0 - OK - proceed with automatic read or write.
1293 *
1294 * > 0 - OK - read or write has been done by the strategy routine, so
1295 * return immediately.
1296 *
1297 * There must be a proc_handler routine for any terminal nodes
1298 * mirrored under /proc/sys (non-terminals are handled by a built-in
1299 * directory handler). Several default handlers are available to
1300 * cover common cases -
1301 *
1302 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1303 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1304 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1305 *
1306 * It is the handler's job to read the input buffer from user memory
1307 * and process it. The handler should return 0 on success.
1308 *
1309 * This routine returns %NULL on a failure to register, and a pointer
1310 * to the table header on success.
1311 */
0b4d4147 1312struct ctl_table_header *register_sysctl_table(ctl_table * table)
1da177e4
LT		 1313{
1314 struct ctl_table_header *tmp;
1315 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1316 if (!tmp)
1317 return NULL;
1318 tmp->ctl_table = table;
1319 INIT_LIST_HEAD(&tmp->ctl_entry);
330d57fb
AV		 1320 tmp->used = 0;
1321 tmp->unregistering = NULL;
1322 spin_lock(&sysctl_lock);
0b4d4147 1323 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
330d57fb 1324 spin_unlock(&sysctl_lock);
b89a8171 1325#ifdef CONFIG_PROC_SYSCTL
330d57fb 1326 register_proc_table(table, proc_sys_root, tmp);
1da177e4
LT		 1327#endif
1328 return tmp;
1329}
1330
1331/**
1332 * unregister_sysctl_table - unregister a sysctl table hierarchy
1333 * @header: the header returned from register_sysctl_table
1334 *
1335 * Unregisters the sysctl table and all children. proc entries may not
1336 * actually be removed until they are no longer used by anyone.
1337 */
1338void unregister_sysctl_table(struct ctl_table_header * header)
1339{
330d57fb
AV		 1340 might_sleep();
1341 spin_lock(&sysctl_lock);
1342 start_unregistering(header);
b89a8171 1343#ifdef CONFIG_PROC_SYSCTL
1da177e4
LT		 1344 unregister_proc_table(header->ctl_table, proc_sys_root);
1345#endif
330d57fb 1346 spin_unlock(&sysctl_lock);
1da177e4
LT		 1347 kfree(header);
1348}
1349
b89a8171
EB		 1350#else /* !CONFIG_SYSCTL */
1351struct ctl_table_header * register_sysctl_table(ctl_table * table,
1352 int insert_at_head)
1353{
1354 return NULL;
1355}
1356
1357void unregister_sysctl_table(struct ctl_table_header * table)
1358{
1359}
1360
1361#endif /* CONFIG_SYSCTL */
1362
1da177e4
LT		 1363/*
1364 * /proc/sys support
1365 */
1366
b89a8171 1367#ifdef CONFIG_PROC_SYSCTL
1da177e4
LT		 1368
1369/* Scan the sysctl entries in table and add them all into /proc */
330d57fb 1370static void register_proc_table(ctl_table * table, struct proc_dir_entry *root, void *set)
1da177e4
LT		 1371{
1372 struct proc_dir_entry *de;
1373 int len;
1374 mode_t mode;
1375
d99f160a 1376 for (; table->ctl_name || table->procname; table++) {
1da177e4
LT		 1377 /* Can't do anything without a proc name. */
1378 if (!table->procname)
1379 continue;
1380 /* Maybe we can't do anything with it... */
1381 if (!table->proc_handler && !table->child) {
1382 printk(KERN_WARNING "SYSCTL: Can't register %s\n",
1383 table->procname);
1384 continue;
1385 }
1386
1387 len = strlen(table->procname);
1388 mode = table->mode;
1389
1390 de = NULL;
1391 if (table->proc_handler)
1392 mode |= S_IFREG;
1393 else {
1394 mode |= S_IFDIR;
1395 for (de = root->subdir; de; de = de->next) {
1396 if (proc_match(len, table->procname, de))
1397 break;
1398 }
1399 /* If the subdir exists already, de is non-NULL */
1400 }
1401
1402 if (!de) {
1403 de = create_proc_entry(table->procname, mode, root);
1404 if (!de)
1405 continue;
330d57fb 1406 de->set = set;
1da177e4
LT		 1407 de->data = (void *) table;
1408 if (table->proc_handler)
1409 de->proc_fops = &proc_sys_file_operations;
1410 }
1411 table->de = de;
1412 if (de->mode & S_IFDIR)
330d57fb 1413 register_proc_table(table->child, de, set);
1da177e4
LT		 1414 }
1415}
1416
1417/*
1418 * Unregister a /proc sysctl table and any subdirectories.
1419 */
1420static void unregister_proc_table(ctl_table * table, struct proc_dir_entry *root)
1421{
1422 struct proc_dir_entry *de;
d99f160a 1423 for (; table->ctl_name || table->procname; table++) {
1da177e4
LT		 1424 if (!(de = table->de))
1425 continue;
1426 if (de->mode & S_IFDIR) {
1427 if (!table->child) {
1428 printk (KERN_ALERT "Help - malformed sysctl tree on free\n");
1429 continue;
1430 }
1431 unregister_proc_table(table->child, de);
1432
1433 /* Don't unregister directories which still have entries.. */
1434 if (de->subdir)
1435 continue;
1436 }
1437
330d57fb
AV		 1438 /*
1439 * In any case, mark the entry as goner; we'll keep it
1440 * around if it's busy, but we'll know to do nothing with
1441 * its fields. We are under sysctl_lock here.
1442 */
1443 de->data = NULL;
1444
1da177e4
LT		 1445 /* Don't unregister proc entries that are still being used.. */
1446 if (atomic_read(&de->count))
1447 continue;
1448
1449 table->de = NULL;
1450 remove_proc_entry(table->procname, root);
1451 }
1452}
1453
1454static ssize_t do_rw_proc(int write, struct file * file, char __user * buf,
1455 size_t count, loff_t *ppos)
1456{
1457 int op;
f3a43f3f 1458 struct proc_dir_entry *de = PDE(file->f_path.dentry->d_inode);
1da177e4
LT		 1459 struct ctl_table *table;
1460 size_t res;
330d57fb 1461 ssize_t error = -ENOTDIR;
1da177e4 1462
330d57fb
AV		 1463 spin_lock(&sysctl_lock);
1464 if (de && de->data && use_table(de->set)) {
1465 /*
1466 * at that point we know that sysctl was not unregistered
1467 * and won't be until we finish
1468 */
1469 spin_unlock(&sysctl_lock);
1470 table = (struct ctl_table *) de->data;
1471 if (!table || !table->proc_handler)
1472 goto out;
1473 error = -EPERM;
1474 op = (write ? 002 : 004);
1475 if (ctl_perm(table, op))
1476 goto out;
1477
1478 /* careful: calling conventions are nasty here */
1479 res = count;
1480 error = (*table->proc_handler)(table, write, file,
1481 buf, &res, ppos);
1482 if (!error)
1483 error = res;
1484 out:
1485 spin_lock(&sysctl_lock);
1486 unuse_table(de->set);
1487 }
1488 spin_unlock(&sysctl_lock);
1489 return error;
1da177e4
LT		 1490}
1491
1492static int proc_opensys(struct inode *inode, struct file *file)
1493{
1494 if (file->f_mode & FMODE_WRITE) {
1495 /*
1496 * sysctl entries that are not writable,
1497 * are _NOT_ writable, capabilities or not.
1498 */
1499 if (!(inode->i_mode & S_IWUSR))
1500 return -EPERM;
1501 }
1502
1503 return 0;
1504}
1505
1506static ssize_t proc_readsys(struct file * file, char __user * buf,
1507 size_t count, loff_t *ppos)
1508{
1509 return do_rw_proc(0, file, buf, count, ppos);
1510}
1511
1512static ssize_t proc_writesys(struct file * file, const char __user * buf,
1513 size_t count, loff_t *ppos)
1514{
1515 return do_rw_proc(1, file, (char __user *) buf, count, ppos);
1516}
1517
b1ba4ddd
AB		 1518static int _proc_do_string(void* data, int maxlen, int write,
1519 struct file *filp, void __user *buffer,
1520 size_t *lenp, loff_t *ppos)
1da177e4
LT		 1521{
1522 size_t len;
1523 char __user *p;
1524 char c;
8d060877
ON		 1525
1526 if (!data || !maxlen || !*lenp) {
1da177e4
LT		 1527 *lenp = 0;
1528 return 0;
1529 }
8d060877 1530
1da177e4
LT		 1531 if (write) {
1532 len = 0;
1533 p = buffer;
1534 while (len < *lenp) {
1535 if (get_user(c, p++))
1536 return -EFAULT;
1537 if (c == 0 || c == '\n')
1538 break;
1539 len++;
1540 }
f5dd3d6f
SV		 1541 if (len >= maxlen)
1542 len = maxlen-1;
1543 if(copy_from_user(data, buffer, len))
1da177e4 1544 return -EFAULT;
f5dd3d6f 1545 ((char *) data)[len] = 0;
1da177e4
LT		 1546 *ppos += *lenp;
1547 } else {
f5dd3d6f
SV		 1548 len = strlen(data);
1549 if (len > maxlen)
1550 len = maxlen;
8d060877
ON		 1551
1552 if (*ppos > len) {
1553 *lenp = 0;
1554 return 0;
1555 }
1556
1557 data += *ppos;
1558 len -= *ppos;
1559
1da177e4
LT		 1560 if (len > *lenp)
1561 len = *lenp;
1562 if (len)
f5dd3d6f 1563 if(copy_to_user(buffer, data, len))
1da177e4
LT		 1564 return -EFAULT;
1565 if (len < *lenp) {
1566 if(put_user('\n', ((char __user *) buffer) + len))
1567 return -EFAULT;
1568 len++;
1569 }
1570 *lenp = len;
1571 *ppos += len;
1572 }
1573 return 0;
1574}
1575
f5dd3d6f
SV		 1576/**
1577 * proc_dostring - read a string sysctl
1578 * @table: the sysctl table
1579 * @write: %TRUE if this is a write to the sysctl file
1580 * @filp: the file structure
1581 * @buffer: the user buffer
1582 * @lenp: the size of the user buffer
1583 * @ppos: file position
1584 *
1585 * Reads/writes a string from/to the user buffer. If the kernel
1586 * buffer provided is not large enough to hold the string, the
1587 * string is truncated. The copied string is %NULL-terminated.
1588 * If the string is being read by the user process, it is copied
1589 * and a newline '\n' is added. It is truncated if the buffer is
1590 * not large enough.
1591 *
1592 * Returns 0 on success.
1593 */
1594int proc_dostring(ctl_table *table, int write, struct file *filp,
1595 void __user *buffer, size_t *lenp, loff_t *ppos)
1596{
1597 return _proc_do_string(table->data, table->maxlen, write, filp,
1598 buffer, lenp, ppos);
1599}
1600
1da177e4
LT		 1601
1602static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1603 int *valp,
1604 int write, void *data)
1605{
1606 if (write) {
1607 *valp = *negp ? -*lvalp : *lvalp;
1608 } else {
1609 int val = *valp;
1610 if (val < 0) {
1611 *negp = -1;
1612 *lvalp = (unsigned long)-val;
1613 } else {
1614 *negp = 0;
1615 *lvalp = (unsigned long)val;
1616 }
1617 }
1618 return 0;
1619}
1620
fcfbd547
KK		 1621static int __do_proc_dointvec(void *tbl_data, ctl_table *table,
1622 int write, struct file *filp, void __user *buffer,
1623 size_t *lenp, loff_t *ppos,
1da177e4
LT		 1624 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1625 int write, void *data),
1626 void *data)
1627{
1628#define TMPBUFLEN 21
1629 int *i, vleft, first=1, neg, val;
1630 unsigned long lval;
1631 size_t left, len;
1632
1633 char buf[TMPBUFLEN], *p;
1634 char __user *s = buffer;
1635
fcfbd547 1636 if (!tbl_data || !table->maxlen || !*lenp ||
1da177e4
LT		 1637 (*ppos && !write)) {
1638 *lenp = 0;
1639 return 0;
1640 }
1641
fcfbd547 1642 i = (int *) tbl_data;
1da177e4
LT		 1643 vleft = table->maxlen / sizeof(*i);
1644 left = *lenp;
1645
1646 if (!conv)
1647 conv = do_proc_dointvec_conv;
1648
1649 for (; left && vleft--; i++, first=0) {
1650 if (write) {
1651 while (left) {
1652 char c;
1653 if (get_user(c, s))
1654 return -EFAULT;
1655 if (!isspace(c))
1656 break;
1657 left--;
1658 s++;
1659 }
1660 if (!left)
1661 break;
1662 neg = 0;
1663 len = left;
1664 if (len > sizeof(buf) - 1)
1665 len = sizeof(buf) - 1;
1666 if (copy_from_user(buf, s, len))
1667 return -EFAULT;
1668 buf[len] = 0;
1669 p = buf;
1670 if (*p == '-' && left > 1) {
1671 neg = 1;
bd9b0bac 1672 p++;
1da177e4
LT		 1673 }
1674 if (*p < '0' || *p > '9')
1675 break;
1676
1677 lval = simple_strtoul(p, &p, 0);
1678
1679 len = p-buf;
1680 if ((len < left) && *p && !isspace(*p))
1681 break;
1682 if (neg)
1683 val = -val;
1684 s += len;
1685 left -= len;
1686
1687 if (conv(&neg, &lval, i, 1, data))
1688 break;
1689 } else {
1690 p = buf;
1691 if (!first)
1692 *p++ = '\t';
1693
1694 if (conv(&neg, &lval, i, 0, data))
1695 break;
1696
1697 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1698 len = strlen(buf);
1699 if (len > left)
1700 len = left;
1701 if(copy_to_user(s, buf, len))
1702 return -EFAULT;
1703 left -= len;
1704 s += len;
1705 }
1706 }
1707
1708 if (!write && !first && left) {
1709 if(put_user('\n', s))
1710 return -EFAULT;
1711 left--, s++;
1712 }
1713 if (write) {
1714 while (left) {
1715 char c;
1716 if (get_user(c, s++))
1717 return -EFAULT;
1718 if (!isspace(c))
1719 break;
1720 left--;
1721 }
1722 }
1723 if (write && first)
1724 return -EINVAL;
1725 *lenp -= left;
1726 *ppos += *lenp;
1727 return 0;
1728#undef TMPBUFLEN
1729}
1730
fcfbd547
KK		 1731static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1732 void __user *buffer, size_t *lenp, loff_t *ppos,
1733 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1734 int write, void *data),
1735 void *data)
1736{
1737 return __do_proc_dointvec(table->data, table, write, filp,
1738 buffer, lenp, ppos, conv, data);
1739}
1740
1da177e4
LT		 1741/**
1742 * proc_dointvec - read a vector of integers
1743 * @table: the sysctl table
1744 * @write: %TRUE if this is a write to the sysctl file
1745 * @filp: the file structure
1746 * @buffer: the user buffer
1747 * @lenp: the size of the user buffer
1748 * @ppos: file position
1749 *
1750 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1751 * values from/to the user buffer, treated as an ASCII string.
1752 *
1753 * Returns 0 on success.
1754 */
1755int proc_dointvec(ctl_table *table, int write, struct file *filp,
1756 void __user *buffer, size_t *lenp, loff_t *ppos)
1757{
1758 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1759 NULL,NULL);
1760}
1761
1762#define OP_SET 0
1763#define OP_AND 1
34f5a398 1764#define OP_OR 2
1da177e4
LT		 1765
1766static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1767 int *valp,
1768 int write, void *data)
1769{
1770 int op = *(int *)data;
1771 if (write) {
1772 int val = *negp ? -*lvalp : *lvalp;
1773 switch(op) {
1774 case OP_SET: *valp = val; break;
1775 case OP_AND: *valp &= val; break;
34f5a398 1776 case OP_OR: *valp |= val; break;
1da177e4
LT		 1777 }
1778 } else {
1779 int val = *valp;
1780 if (val < 0) {
1781 *negp = -1;
1782 *lvalp = (unsigned long)-val;
1783 } else {
1784 *negp = 0;
1785 *lvalp = (unsigned long)val;
1786 }
1787 }
1788 return 0;
1789}
1790
1791/*
1792 * init may raise the set.
1793 */
1794
1795int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1796 void __user *buffer, size_t *lenp, loff_t *ppos)
1797{
1798 int op;
1799
6ff1b442 1800 if (write && !capable(CAP_SYS_MODULE)) {
1da177e4
LT		 1801 return -EPERM;
1802 }
1803
f400e198 1804 op = is_init(current) ? OP_SET : OP_AND;
1da177e4
LT		 1805 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1806 do_proc_dointvec_bset_conv,&op);
1807}
1808
34f5a398
TT		 1809/*
1810 * Taint values can only be increased
1811 */
1812static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
1813 void __user *buffer, size_t *lenp, loff_t *ppos)
1814{
1815 int op;
1816
1817 if (!capable(CAP_SYS_ADMIN))
1818 return -EPERM;
1819
1820 op = OP_OR;
1821 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1822 do_proc_dointvec_bset_conv,&op);
1823}
1824
1da177e4
LT		 1825struct do_proc_dointvec_minmax_conv_param {
1826 int *min;
1827 int *max;
1828};
1829
1830static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1831 int *valp,
1832 int write, void *data)
1833{
1834 struct do_proc_dointvec_minmax_conv_param *param = data;
1835 if (write) {
1836 int val = *negp ? -*lvalp : *lvalp;
1837 if ((param->min && *param->min > val) ||
1838 (param->max && *param->max < val))
1839 return -EINVAL;
1840 *valp = val;
1841 } else {
1842 int val = *valp;
1843 if (val < 0) {
1844 *negp = -1;
1845 *lvalp = (unsigned long)-val;
1846 } else {
1847 *negp = 0;
1848 *lvalp = (unsigned long)val;
1849 }
1850 }
1851 return 0;
1852}
1853
1854/**
1855 * proc_dointvec_minmax - read a vector of integers with min/max values
1856 * @table: the sysctl table
1857 * @write: %TRUE if this is a write to the sysctl file
1858 * @filp: the file structure
1859 * @buffer: the user buffer
1860 * @lenp: the size of the user buffer
1861 * @ppos: file position
1862 *
1863 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1864 * values from/to the user buffer, treated as an ASCII string.
1865 *
1866 * This routine will ensure the values are within the range specified by
1867 * table->extra1 (min) and table->extra2 (max).
1868 *
1869 * Returns 0 on success.
1870 */
1871int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1872 void __user *buffer, size_t *lenp, loff_t *ppos)
1873{
1874 struct do_proc_dointvec_minmax_conv_param param = {
1875 .min = (int *) table->extra1,
1876 .max = (int *) table->extra2,
1877 };
1878 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1879 do_proc_dointvec_minmax_conv, &param);
1880}
1881
fcfbd547 1882static int __do_proc_doulongvec_minmax(void *data, ctl_table *table, int write,
1da177e4
LT		 1883 struct file *filp,
1884 void __user *buffer,
1885 size_t *lenp, loff_t *ppos,
1886 unsigned long convmul,
1887 unsigned long convdiv)
1888{
1889#define TMPBUFLEN 21
1890 unsigned long *i, *min, *max, val;
1891 int vleft, first=1, neg;
1892 size_t len, left;
1893 char buf[TMPBUFLEN], *p;
1894 char __user *s = buffer;
1895
fcfbd547 1896 if (!data || !table->maxlen || !*lenp ||
1da177e4
LT		 1897 (*ppos && !write)) {
1898 *lenp = 0;
1899 return 0;
1900 }
1901
fcfbd547 1902 i = (unsigned long *) data;
1da177e4
LT		 1903 min = (unsigned long *) table->extra1;
1904 max = (unsigned long *) table->extra2;
1905 vleft = table->maxlen / sizeof(unsigned long);
1906 left = *lenp;
1907
1908 for (; left && vleft--; i++, min++, max++, first=0) {
1909 if (write) {
1910 while (left) {
1911 char c;
1912 if (get_user(c, s))
1913 return -EFAULT;
1914 if (!isspace(c))
1915 break;
1916 left--;
1917 s++;
1918 }
1919 if (!left)
1920 break;
1921 neg = 0;
1922 len = left;
1923 if (len > TMPBUFLEN-1)
1924 len = TMPBUFLEN-1;
1925 if (copy_from_user(buf, s, len))
1926 return -EFAULT;
1927 buf[len] = 0;
1928 p = buf;
1929 if (*p == '-' && left > 1) {
1930 neg = 1;
bd9b0bac 1931 p++;
1da177e4
LT		 1932 }
1933 if (*p < '0' || *p > '9')
1934 break;
1935 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
1936 len = p-buf;
1937 if ((len < left) && *p && !isspace(*p))
1938 break;
1939 if (neg)
1940 val = -val;
1941 s += len;
1942 left -= len;
1943
1944 if(neg)
1945 continue;
1946 if ((min && val < *min) || (max && val > *max))
1947 continue;
1948 *i = val;
1949 } else {
1950 p = buf;
1951 if (!first)
1952 *p++ = '\t';
1953 sprintf(p, "%lu", convdiv * (*i) / convmul);
1954 len = strlen(buf);
1955 if (len > left)
1956 len = left;
1957 if(copy_to_user(s, buf, len))
1958 return -EFAULT;
1959 left -= len;
1960 s += len;
1961 }
1962 }
1963
1964 if (!write && !first && left) {
1965 if(put_user('\n', s))
1966 return -EFAULT;
1967 left--, s++;
1968 }
1969 if (write) {
1970 while (left) {
1971 char c;
1972 if (get_user(c, s++))
1973 return -EFAULT;
1974 if (!isspace(c))
1975 break;
1976 left--;
1977 }
1978 }
1979 if (write && first)
1980 return -EINVAL;
1981 *lenp -= left;
1982 *ppos += *lenp;
1983 return 0;
1984#undef TMPBUFLEN
1985}
1986
fcfbd547
KK		 1987static int do_proc_doulongvec_minmax(ctl_table *table, int write,
1988 struct file *filp,
1989 void __user *buffer,
1990 size_t *lenp, loff_t *ppos,
1991 unsigned long convmul,
1992 unsigned long convdiv)
1993{
1994 return __do_proc_doulongvec_minmax(table->data, table, write,
1995 filp, buffer, lenp, ppos, convmul, convdiv);
1996}
1997
1da177e4
LT		 1998/**
1999 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2000 * @table: the sysctl table
2001 * @write: %TRUE if this is a write to the sysctl file
2002 * @filp: the file structure
2003 * @buffer: the user buffer
2004 * @lenp: the size of the user buffer
2005 * @ppos: file position
2006 *
2007 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2008 * values from/to the user buffer, treated as an ASCII string.
2009 *
2010 * This routine will ensure the values are within the range specified by
2011 * table->extra1 (min) and table->extra2 (max).
2012 *
2013 * Returns 0 on success.
2014 */
2015int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2016 void __user *buffer, size_t *lenp, loff_t *ppos)
2017{
2018 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2019}
2020
2021/**
2022 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2023 * @table: the sysctl table
2024 * @write: %TRUE if this is a write to the sysctl file
2025 * @filp: the file structure
2026 * @buffer: the user buffer
2027 * @lenp: the size of the user buffer
2028 * @ppos: file position
2029 *
2030 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2031 * values from/to the user buffer, treated as an ASCII string. The values
2032 * are treated as milliseconds, and converted to jiffies when they are stored.
2033 *
2034 * This routine will ensure the values are within the range specified by
2035 * table->extra1 (min) and table->extra2 (max).
2036 *
2037 * Returns 0 on success.
2038 */
2039int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2040 struct file *filp,
2041 void __user *buffer,
2042 size_t *lenp, loff_t *ppos)
2043{
2044 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2045 lenp, ppos, HZ, 1000l);
2046}
2047
2048
2049static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2050 int *valp,
2051 int write, void *data)
2052{
2053 if (write) {
cba9f33d
BS		 2054 if (*lvalp > LONG_MAX / HZ)
2055 return 1;
1da177e4
LT		 2056 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2057 } else {
2058 int val = *valp;
2059 unsigned long lval;
2060 if (val < 0) {
2061 *negp = -1;
2062 lval = (unsigned long)-val;
2063 } else {
2064 *negp = 0;
2065 lval = (unsigned long)val;
2066 }
2067 *lvalp = lval / HZ;
2068 }
2069 return 0;
2070}
2071
2072static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2073 int *valp,
2074 int write, void *data)
2075{
2076 if (write) {
cba9f33d
BS		 2077 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2078 return 1;
1da177e4
LT		 2079 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2080 } else {
2081 int val = *valp;
2082 unsigned long lval;
2083 if (val < 0) {
2084 *negp = -1;
2085 lval = (unsigned long)-val;
2086 } else {
2087 *negp = 0;
2088 lval = (unsigned long)val;
2089 }
2090 *lvalp = jiffies_to_clock_t(lval);
2091 }
2092 return 0;
2093}
2094
2095static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2096 int *valp,
2097 int write, void *data)
2098{
2099 if (write) {
2100 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2101 } else {
2102 int val = *valp;
2103 unsigned long lval;
2104 if (val < 0) {
2105 *negp = -1;
2106 lval = (unsigned long)-val;
2107 } else {
2108 *negp = 0;
2109 lval = (unsigned long)val;
2110 }
2111 *lvalp = jiffies_to_msecs(lval);
2112 }
2113 return 0;
2114}
2115
2116/**
2117 * proc_dointvec_jiffies - read a vector of integers as seconds
2118 * @table: the sysctl table
2119 * @write: %TRUE if this is a write to the sysctl file
2120 * @filp: the file structure
2121 * @buffer: the user buffer
2122 * @lenp: the size of the user buffer
2123 * @ppos: file position
2124 *
2125 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2126 * values from/to the user buffer, treated as an ASCII string.
2127 * The values read are assumed to be in seconds, and are converted into
2128 * jiffies.
2129 *
2130 * Returns 0 on success.
2131 */
2132int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2133 void __user *buffer, size_t *lenp, loff_t *ppos)
2134{
2135 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2136 do_proc_dointvec_jiffies_conv,NULL);
2137}
2138
2139/**
2140 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2141 * @table: the sysctl table
2142 * @write: %TRUE if this is a write to the sysctl file
2143 * @filp: the file structure
2144 * @buffer: the user buffer
2145 * @lenp: the size of the user buffer
1e5d5331 2146 * @ppos: pointer to the file position
1da177e4
LT		 2147 *
2148 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2149 * values from/to the user buffer, treated as an ASCII string.
2150 * The values read are assumed to be in 1/USER_HZ seconds, and
2151 * are converted into jiffies.
2152 *
2153 * Returns 0 on success.
2154 */
2155int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2156 void __user *buffer, size_t *lenp, loff_t *ppos)
2157{
2158 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2159 do_proc_dointvec_userhz_jiffies_conv,NULL);
2160}
2161
2162/**
2163 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2164 * @table: the sysctl table
2165 * @write: %TRUE if this is a write to the sysctl file
2166 * @filp: the file structure
2167 * @buffer: the user buffer
2168 * @lenp: the size of the user buffer
67be2dd1
MW		 2169 * @ppos: file position
2170 * @ppos: the current position in the file
1da177e4
LT		 2171 *
2172 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2173 * values from/to the user buffer, treated as an ASCII string.
2174 * The values read are assumed to be in 1/1000 seconds, and
2175 * are converted into jiffies.
2176 *
2177 * Returns 0 on success.
2178 */
2179int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2180 void __user *buffer, size_t *lenp, loff_t *ppos)
2181{
2182 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2183 do_proc_dointvec_ms_jiffies_conv, NULL);
2184}
2185
9ec52099
CLG		 2186static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
2187 void __user *buffer, size_t *lenp, loff_t *ppos)
2188{
2189 struct pid *new_pid;
2190 pid_t tmp;
2191 int r;
2192
2193 tmp = pid_nr(cad_pid);
2194
2195 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2196 lenp, ppos, NULL, NULL);
2197 if (r || !write)
2198 return r;
2199
2200 new_pid = find_get_pid(tmp);
2201 if (!new_pid)
2202 return -ESRCH;
2203
2204 put_pid(xchg(&cad_pid, new_pid));
2205 return 0;
2206}
2207
1da177e4
LT		 2208#else /* CONFIG_PROC_FS */
2209
2210int proc_dostring(ctl_table *table, int write, struct file *filp,
2211 void __user *buffer, size_t *lenp, loff_t *ppos)
2212{
2213 return -ENOSYS;
2214}
2215
1da177e4
LT		 2216int proc_dointvec(ctl_table *table, int write, struct file *filp,
2217 void __user *buffer, size_t *lenp, loff_t *ppos)
2218{
2219 return -ENOSYS;
2220}
2221
2222int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2223 void __user *buffer, size_t *lenp, loff_t *ppos)
2224{
2225 return -ENOSYS;
2226}
2227
2228int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2229 void __user *buffer, size_t *lenp, loff_t *ppos)
2230{
2231 return -ENOSYS;
2232}
2233
2234int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2235 void __user *buffer, size_t *lenp, loff_t *ppos)
2236{
2237 return -ENOSYS;
2238}
2239
2240int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2241 void __user *buffer, size_t *lenp, loff_t *ppos)
2242{
2243 return -ENOSYS;
2244}
2245
2246int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2247 void __user *buffer, size_t *lenp, loff_t *ppos)
2248{
2249 return -ENOSYS;
2250}
2251
2252int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2253 void __user *buffer, size_t *lenp, loff_t *ppos)
2254{
2255 return -ENOSYS;
2256}
2257
2258int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2259 struct file *filp,
2260 void __user *buffer,
2261 size_t *lenp, loff_t *ppos)
2262{
2263 return -ENOSYS;
2264}
2265
2266
2267#endif /* CONFIG_PROC_FS */
2268
2269
b89a8171 2270#ifdef CONFIG_SYSCTL_SYSCALL
1da177e4
LT		 2271/*
2272 * General sysctl support routines
2273 */
2274
2275/* The generic string strategy routine: */
2276int sysctl_string(ctl_table *table, int __user *name, int nlen,
2277 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2278 void __user *newval, size_t newlen)
1da177e4 2279{
1da177e4
LT		 2280 if (!table->data || !table->maxlen)
2281 return -ENOTDIR;
2282
2283 if (oldval && oldlenp) {
de9e007d
LT		 2284 size_t bufsize;
2285 if (get_user(bufsize, oldlenp))
1da177e4 2286 return -EFAULT;
de9e007d
LT		 2287 if (bufsize) {
2288 size_t len = strlen(table->data), copied;
2289
2290 /* This shouldn't trigger for a well-formed sysctl */
2291 if (len > table->maxlen)
1da177e4 2292 len = table->maxlen;
de9e007d
LT		 2293
2294 /* Copy up to a max of bufsize-1 bytes of the string */
2295 copied = (len >= bufsize) ? bufsize - 1 : len;
2296
2297 if (copy_to_user(oldval, table->data, copied) ||
2298 put_user(0, (char __user *)(oldval + copied)))
1da177e4 2299 return -EFAULT;
de9e007d 2300 if (put_user(len, oldlenp))
1da177e4
LT		 2301 return -EFAULT;
2302 }
2303 }
2304 if (newval && newlen) {
de9e007d 2305 size_t len = newlen;
1da177e4
LT		 2306 if (len > table->maxlen)
2307 len = table->maxlen;
2308 if(copy_from_user(table->data, newval, len))
2309 return -EFAULT;
2310 if (len == table->maxlen)
2311 len--;
2312 ((char *) table->data)[len] = 0;
2313 }
82c9df82 2314 return 1;
1da177e4
LT		 2315}
2316
2317/*
2318 * This function makes sure that all of the integers in the vector
2319 * are between the minimum and maximum values given in the arrays
2320 * table->extra1 and table->extra2, respectively.
2321 */
2322int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2323 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2324 void __user *newval, size_t newlen)
1da177e4
LT		 2325{
2326
2327 if (newval && newlen) {
2328 int __user *vec = (int __user *) newval;
2329 int *min = (int *) table->extra1;
2330 int *max = (int *) table->extra2;
2331 size_t length;
2332 int i;
2333
2334 if (newlen % sizeof(int) != 0)
2335 return -EINVAL;
2336
2337 if (!table->extra1 && !table->extra2)
2338 return 0;
2339
2340 if (newlen > table->maxlen)
2341 newlen = table->maxlen;
2342 length = newlen / sizeof(int);
2343
2344 for (i = 0; i < length; i++) {
2345 int value;
2346 if (get_user(value, vec + i))
2347 return -EFAULT;
2348 if (min && value < min[i])
2349 return -EINVAL;
2350 if (max && value > max[i])
2351 return -EINVAL;
2352 }
2353 }
2354 return 0;
2355}
2356
2357/* Strategy function to convert jiffies to seconds */
2358int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2359 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2360 void __user *newval, size_t newlen)
1da177e4 2361{
3ee75ac3 2362 if (oldval && oldlenp) {
1da177e4 2363 size_t olen;
3ee75ac3
AD		 2364
2365 if (get_user(olen, oldlenp))
2366 return -EFAULT;
2367 if (olen) {
2368 int val;
2369
2370 if (olen < sizeof(int))
2371 return -EINVAL;
2372
2373 val = *(int *)(table->data) / HZ;
2374 if (put_user(val, (int __user *)oldval))
2375 return -EFAULT;
2376 if (put_user(sizeof(int), oldlenp))
1da177e4 2377 return -EFAULT;
1da177e4 2378 }
1da177e4
LT		 2379 }
2380 if (newval && newlen) {
2381 int new;
2382 if (newlen != sizeof(int))
2383 return -EINVAL;
2384 if (get_user(new, (int __user *)newval))
2385 return -EFAULT;
2386 *(int *)(table->data) = new*HZ;
2387 }
2388 return 1;
2389}
2390
2391/* Strategy function to convert jiffies to seconds */
2392int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2393 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2394 void __user *newval, size_t newlen)
1da177e4 2395{
3ee75ac3 2396 if (oldval && oldlenp) {
1da177e4 2397 size_t olen;
3ee75ac3
AD		 2398
2399 if (get_user(olen, oldlenp))
2400 return -EFAULT;
2401 if (olen) {
2402 int val;
2403
2404 if (olen < sizeof(int))
2405 return -EINVAL;
2406
2407 val = jiffies_to_msecs(*(int *)(table->data));
2408 if (put_user(val, (int __user *)oldval))
2409 return -EFAULT;
2410 if (put_user(sizeof(int), oldlenp))
1da177e4 2411 return -EFAULT;
1da177e4 2412 }
1da177e4
LT		 2413 }
2414 if (newval && newlen) {
2415 int new;
2416 if (newlen != sizeof(int))
2417 return -EINVAL;
2418 if (get_user(new, (int __user *)newval))
2419 return -EFAULT;
2420 *(int *)(table->data) = msecs_to_jiffies(new);
2421 }
2422 return 1;
2423}
2424
c4b8b769 2425
c4b8b769 2426
b89a8171 2427#else /* CONFIG_SYSCTL_SYSCALL */
1da177e4
LT		 2428
2429
2430asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2431{
b89a8171 2432 static int msg_count;
0e009be8
EB		 2433 struct __sysctl_args tmp;
2434 int name[CTL_MAXNAME];
2435 int i;
2436
2437 /* Read in the sysctl name for better debug message logging */
2438 if (copy_from_user(&tmp, args, sizeof(tmp)))
2439 return -EFAULT;
2440 if (tmp.nlen <= 0 || tmp.nlen >= CTL_MAXNAME)
2441 return -ENOTDIR;
2442 for (i = 0; i < tmp.nlen; i++)
2443 if (get_user(name[i], tmp.name + i))
2444 return -EFAULT;
2445
2446 /* Ignore accesses to kernel.version */
2447 if ((tmp.nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2448 goto out;
b89a8171
EB		 2449
2450 if (msg_count < 5) {
2451 msg_count++;
2452 printk(KERN_INFO
2453 "warning: process `%s' used the removed sysctl "
0e009be8
EB		 2454 "system call with ", current->comm);
2455 for (i = 0; i < tmp.nlen; i++)
2456 printk("%d.", name[i]);
2457 printk("\n");
b89a8171 2458 }
0e009be8 2459out:
1da177e4
LT		 2460 return -ENOSYS;
2461}
2462
2463int sysctl_string(ctl_table *table, int __user *name, int nlen,
2464 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2465 void __user *newval, size_t newlen)
1da177e4
LT		 2466{
2467 return -ENOSYS;
2468}
2469
2470int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2471 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2472 void __user *newval, size_t newlen)
1da177e4
LT		 2473{
2474 return -ENOSYS;
2475}
2476
2477int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2478 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2479 void __user *newval, size_t newlen)
1da177e4
LT		 2480{
2481 return -ENOSYS;
2482}
2483
2484int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2485 void __user *oldval, size_t __user *oldlenp,
1f29bcd7 2486 void __user *newval, size_t newlen)
1da177e4
LT		 2487{
2488 return -ENOSYS;
2489}
2490
b89a8171 2491#endif /* CONFIG_SYSCTL_SYSCALL */
1da177e4
LT		 2492
2493/*
2494 * No sense putting this after each symbol definition, twice,
2495 * exception granted :-)
2496 */
2497EXPORT_SYMBOL(proc_dointvec);
2498EXPORT_SYMBOL(proc_dointvec_jiffies);
2499EXPORT_SYMBOL(proc_dointvec_minmax);
2500EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2501EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2502EXPORT_SYMBOL(proc_dostring);
2503EXPORT_SYMBOL(proc_doulongvec_minmax);
2504EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2505EXPORT_SYMBOL(register_sysctl_table);
2506EXPORT_SYMBOL(sysctl_intvec);
2507EXPORT_SYMBOL(sysctl_jiffies);
2508EXPORT_SYMBOL(sysctl_ms_jiffies);
2509EXPORT_SYMBOL(sysctl_string);
2510EXPORT_SYMBOL(unregister_sysctl_table);
Linux 6.x block layer and io_uring tree(s)