]> git.proxmox.com Git - mirror_ubuntu-zesty-kernel.git/blame - kernel/sysctl.c
projects / mirror_ubuntu-zesty-kernel.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
[PATCH] introduce get_task_pid() to fix unsafe get_pid()
[mirror_ubuntu-zesty-kernel.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>
57#endif
58
1da177e4
LT		 59#if defined(CONFIG_SYSCTL)
60
61/* External variables not in a header file. */
62extern int C_A_D;
63extern int sysctl_overcommit_memory;
64extern int sysctl_overcommit_ratio;
fadd8fbd 65extern int sysctl_panic_on_oom;
1da177e4
LT		 66extern int max_threads;
67extern int sysrq_enabled;
68extern int core_uses_pid;
d6e71144 69extern int suid_dumpable;
1da177e4
LT		 70extern char core_pattern[];
71extern int cad_pid;
72extern int pid_max;
73extern int min_free_kbytes;
74extern int printk_ratelimit_jiffies;
75extern int printk_ratelimit_burst;
76extern int pid_max_min, pid_max_max;
9d0243bc 77extern int sysctl_drop_caches;
8ad4b1fb 78extern int percpu_pagelist_fraction;
bebfa101 79extern int compat_log;
1da177e4 80
1da177e4
LT		 81/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
82static int maxolduid = 65535;
83static int minolduid;
8ad4b1fb 84static int min_percpu_pagelist_fract = 8;
1da177e4
LT		 85
86static int ngroups_max = NGROUPS_MAX;
87
88#ifdef CONFIG_KMOD
89extern char modprobe_path[];
90#endif
1da177e4
LT		 91#ifdef CONFIG_CHR_DEV_SG
92extern int sg_big_buff;
93#endif
94#ifdef CONFIG_SYSVIPC
fcfbd547
KK		 95static int proc_do_ipc_string(ctl_table *table, int write, struct file *filp,
96 void __user *buffer, size_t *lenp, loff_t *ppos);
1da177e4
LT		 97#endif
98
99#ifdef __sparc__
100extern char reboot_command [];
101extern int stop_a_enabled;
102extern int scons_pwroff;
103#endif
104
105#ifdef __hppa__
106extern int pwrsw_enabled;
107extern int unaligned_enabled;
108#endif
109
347a8dc3 110#ifdef CONFIG_S390
1da177e4
LT		 111#ifdef CONFIG_MATHEMU
112extern int sysctl_ieee_emulation_warnings;
113#endif
114extern int sysctl_userprocess_debug;
951f22d5 115extern int spin_retry;
1da177e4
LT		 116#endif
117
118extern int sysctl_hz_timer;
119
120#ifdef CONFIG_BSD_PROCESS_ACCT
121extern int acct_parm[];
122#endif
123
d2b176ed
JS		 124#ifdef CONFIG_IA64
125extern int no_unaligned_warning;
126#endif
127
23f78d4a
IM		 128#ifdef CONFIG_RT_MUTEXES
129extern int max_lock_depth;
130#endif
131
b89a8171
EB		 132#ifdef CONFIG_SYSCTL_SYSCALL
133static int parse_table(int __user *, int, void __user *, size_t __user *,
134 void __user *, size_t, ctl_table *, void **);
135#endif
136
8218c74c 137static int proc_do_uts_string(ctl_table *table, int write, struct file *filp,
1da177e4
LT		 138 void __user *buffer, size_t *lenp, loff_t *ppos);
139
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
161/* /proc declarations: */
162
b89a8171 163#ifdef CONFIG_PROC_SYSCTL
1da177e4
LT		 164
165static ssize_t proc_readsys(struct file *, char __user *, size_t, loff_t *);
166static ssize_t proc_writesys(struct file *, const char __user *, size_t, loff_t *);
167static int proc_opensys(struct inode *, struct file *);
168
169struct file_operations proc_sys_file_operations = {
170 .open = proc_opensys,
171 .read = proc_readsys,
172 .write = proc_writesys,
173};
174
175extern struct proc_dir_entry *proc_sys_root;
176
330d57fb 177static void register_proc_table(ctl_table *, struct proc_dir_entry *, void *);
1da177e4
LT		 178static void unregister_proc_table(ctl_table *, struct proc_dir_entry *);
179#endif
180
181/* The default sysctl tables: */
182
183static ctl_table root_table[] = {
184 {
185 .ctl_name = CTL_KERN,
186 .procname = "kernel",
187 .mode = 0555,
188 .child = kern_table,
189 },
190 {
191 .ctl_name = CTL_VM,
192 .procname = "vm",
193 .mode = 0555,
194 .child = vm_table,
195 },
196#ifdef CONFIG_NET
197 {
198 .ctl_name = CTL_NET,
199 .procname = "net",
200 .mode = 0555,
201 .child = net_table,
202 },
203#endif
1da177e4
LT		 204 {
205 .ctl_name = CTL_FS,
206 .procname = "fs",
207 .mode = 0555,
208 .child = fs_table,
209 },
210 {
211 .ctl_name = CTL_DEBUG,
212 .procname = "debug",
213 .mode = 0555,
214 .child = debug_table,
215 },
216 {
217 .ctl_name = CTL_DEV,
218 .procname = "dev",
219 .mode = 0555,
220 .child = dev_table,
221 },
0eeca283 222
1da177e4
LT		 223 { .ctl_name = 0 }
224};
225
226static ctl_table kern_table[] = {
8218c74c 227#ifndef CONFIG_UTS_NS
1da177e4
LT		 228 {
229 .ctl_name = KERN_OSTYPE,
230 .procname = "ostype",
8218c74c
SH		 231 .data = init_uts_ns.name.sysname,
232 .maxlen = sizeof(init_uts_ns.name.sysname),
1da177e4 233 .mode = 0444,
8218c74c 234 .proc_handler = &proc_do_uts_string,
1da177e4
LT		 235 .strategy = &sysctl_string,
236 },
237 {
238 .ctl_name = KERN_OSRELEASE,
239 .procname = "osrelease",
8218c74c
SH		 240 .data = init_uts_ns.name.release,
241 .maxlen = sizeof(init_uts_ns.name.release),
1da177e4 242 .mode = 0444,
8218c74c 243 .proc_handler = &proc_do_uts_string,
1da177e4
LT		 244 .strategy = &sysctl_string,
245 },
246 {
247 .ctl_name = KERN_VERSION,
248 .procname = "version",
8218c74c
SH		 249 .data = init_uts_ns.name.version,
250 .maxlen = sizeof(init_uts_ns.name.version),
1da177e4 251 .mode = 0444,
8218c74c 252 .proc_handler = &proc_do_uts_string,
1da177e4
LT		 253 .strategy = &sysctl_string,
254 },
255 {
256 .ctl_name = KERN_NODENAME,
257 .procname = "hostname",
8218c74c
SH		 258 .data = init_uts_ns.name.nodename,
259 .maxlen = sizeof(init_uts_ns.name.nodename),
1da177e4 260 .mode = 0644,
8218c74c 261 .proc_handler = &proc_do_uts_string,
1da177e4
LT		 262 .strategy = &sysctl_string,
263 },
264 {
265 .ctl_name = KERN_DOMAINNAME,
266 .procname = "domainname",
8218c74c
SH		 267 .data = init_uts_ns.name.domainname,
268 .maxlen = sizeof(init_uts_ns.name.domainname),
1da177e4 269 .mode = 0644,
8218c74c 270 .proc_handler = &proc_do_uts_string,
1da177e4
LT		 271 .strategy = &sysctl_string,
272 },
8218c74c
SH		 273#else /* !CONFIG_UTS_NS */
274 {
275 .ctl_name = KERN_OSTYPE,
276 .procname = "ostype",
277 .data = NULL,
278 /* could maybe use __NEW_UTS_LEN here? */
279 .maxlen = FIELD_SIZEOF(struct new_utsname, sysname),
280 .mode = 0444,
281 .proc_handler = &proc_do_uts_string,
282 .strategy = &sysctl_string,
283 },
284 {
285 .ctl_name = KERN_OSRELEASE,
286 .procname = "osrelease",
287 .data = NULL,
288 .maxlen = FIELD_SIZEOF(struct new_utsname, release),
289 .mode = 0444,
290 .proc_handler = &proc_do_uts_string,
291 .strategy = &sysctl_string,
292 },
293 {
294 .ctl_name = KERN_VERSION,
295 .procname = "version",
296 .data = NULL,
297 .maxlen = FIELD_SIZEOF(struct new_utsname, version),
298 .mode = 0444,
299 .proc_handler = &proc_do_uts_string,
300 .strategy = &sysctl_string,
301 },
302 {
303 .ctl_name = KERN_NODENAME,
304 .procname = "hostname",
305 .data = NULL,
306 .maxlen = FIELD_SIZEOF(struct new_utsname, nodename),
307 .mode = 0644,
308 .proc_handler = &proc_do_uts_string,
309 .strategy = &sysctl_string,
310 },
311 {
312 .ctl_name = KERN_DOMAINNAME,
313 .procname = "domainname",
314 .data = NULL,
315 .maxlen = FIELD_SIZEOF(struct new_utsname, domainname),
316 .mode = 0644,
317 .proc_handler = &proc_do_uts_string,
318 .strategy = &sysctl_string,
319 },
320#endif /* !CONFIG_UTS_NS */
1da177e4
LT		 321 {
322 .ctl_name = KERN_PANIC,
323 .procname = "panic",
324 .data = &panic_timeout,
325 .maxlen = sizeof(int),
326 .mode = 0644,
327 .proc_handler = &proc_dointvec,
328 },
329 {
330 .ctl_name = KERN_CORE_USES_PID,
331 .procname = "core_uses_pid",
332 .data = &core_uses_pid,
333 .maxlen = sizeof(int),
334 .mode = 0644,
335 .proc_handler = &proc_dointvec,
336 },
337 {
338 .ctl_name = KERN_CORE_PATTERN,
339 .procname = "core_pattern",
340 .data = core_pattern,
d025c9db 341 .maxlen = 128,
1da177e4
LT		 342 .mode = 0644,
343 .proc_handler = &proc_dostring,
344 .strategy = &sysctl_string,
345 },
346 {
347 .ctl_name = KERN_TAINTED,
348 .procname = "tainted",
349 .data = &tainted,
350 .maxlen = sizeof(int),
351 .mode = 0444,
352 .proc_handler = &proc_dointvec,
353 },
354 {
355 .ctl_name = KERN_CAP_BSET,
356 .procname = "cap-bound",
357 .data = &cap_bset,
358 .maxlen = sizeof(kernel_cap_t),
359 .mode = 0600,
360 .proc_handler = &proc_dointvec_bset,
361 },
362#ifdef CONFIG_BLK_DEV_INITRD
363 {
364 .ctl_name = KERN_REALROOTDEV,
365 .procname = "real-root-dev",
366 .data = &real_root_dev,
367 .maxlen = sizeof(int),
368 .mode = 0644,
369 .proc_handler = &proc_dointvec,
370 },
371#endif
372#ifdef __sparc__
373 {
374 .ctl_name = KERN_SPARC_REBOOT,
375 .procname = "reboot-cmd",
376 .data = reboot_command,
377 .maxlen = 256,
378 .mode = 0644,
379 .proc_handler = &proc_dostring,
380 .strategy = &sysctl_string,
381 },
382 {
383 .ctl_name = KERN_SPARC_STOP_A,
384 .procname = "stop-a",
385 .data = &stop_a_enabled,
386 .maxlen = sizeof (int),
387 .mode = 0644,
388 .proc_handler = &proc_dointvec,
389 },
390 {
391 .ctl_name = KERN_SPARC_SCONS_PWROFF,
392 .procname = "scons-poweroff",
393 .data = &scons_pwroff,
394 .maxlen = sizeof (int),
395 .mode = 0644,
396 .proc_handler = &proc_dointvec,
397 },
398#endif
399#ifdef __hppa__
400 {
401 .ctl_name = KERN_HPPA_PWRSW,
402 .procname = "soft-power",
403 .data = &pwrsw_enabled,
404 .maxlen = sizeof (int),
405 .mode = 0644,
406 .proc_handler = &proc_dointvec,
407 },
408 {
409 .ctl_name = KERN_HPPA_UNALIGNED,
410 .procname = "unaligned-trap",
411 .data = &unaligned_enabled,
412 .maxlen = sizeof (int),
413 .mode = 0644,
414 .proc_handler = &proc_dointvec,
415 },
416#endif
417 {
418 .ctl_name = KERN_CTLALTDEL,
419 .procname = "ctrl-alt-del",
420 .data = &C_A_D,
421 .maxlen = sizeof(int),
422 .mode = 0644,
423 .proc_handler = &proc_dointvec,
424 },
425 {
426 .ctl_name = KERN_PRINTK,
427 .procname = "printk",
428 .data = &console_loglevel,
429 .maxlen = 4*sizeof(int),
430 .mode = 0644,
431 .proc_handler = &proc_dointvec,
432 },
433#ifdef CONFIG_KMOD
434 {
435 .ctl_name = KERN_MODPROBE,
436 .procname = "modprobe",
437 .data = &modprobe_path,
438 .maxlen = KMOD_PATH_LEN,
439 .mode = 0644,
440 .proc_handler = &proc_dostring,
441 .strategy = &sysctl_string,
442 },
443#endif
57ae2508 444#if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
1da177e4
LT		 445 {
446 .ctl_name = KERN_HOTPLUG,
447 .procname = "hotplug",
312c004d
KS		 448 .data = &uevent_helper,
449 .maxlen = UEVENT_HELPER_PATH_LEN,
1da177e4
LT		 450 .mode = 0644,
451 .proc_handler = &proc_dostring,
452 .strategy = &sysctl_string,
453 },
454#endif
455#ifdef CONFIG_CHR_DEV_SG
456 {
457 .ctl_name = KERN_SG_BIG_BUFF,
458 .procname = "sg-big-buff",
459 .data = &sg_big_buff,
460 .maxlen = sizeof (int),
461 .mode = 0444,
462 .proc_handler = &proc_dointvec,
463 },
464#endif
465#ifdef CONFIG_BSD_PROCESS_ACCT
466 {
467 .ctl_name = KERN_ACCT,
468 .procname = "acct",
469 .data = &acct_parm,
470 .maxlen = 3*sizeof(int),
471 .mode = 0644,
472 .proc_handler = &proc_dointvec,
473 },
474#endif
475#ifdef CONFIG_SYSVIPC
476 {
477 .ctl_name = KERN_SHMMAX,
478 .procname = "shmmax",
fcfbd547 479 .data = NULL,
1da177e4
LT		 480 .maxlen = sizeof (size_t),
481 .mode = 0644,
fcfbd547 482 .proc_handler = &proc_do_ipc_string,
1da177e4
LT		 483 },
484 {
485 .ctl_name = KERN_SHMALL,
486 .procname = "shmall",
fcfbd547 487 .data = NULL,
1da177e4
LT		 488 .maxlen = sizeof (size_t),
489 .mode = 0644,
fcfbd547 490 .proc_handler = &proc_do_ipc_string,
1da177e4
LT		 491 },
492 {
493 .ctl_name = KERN_SHMMNI,
494 .procname = "shmmni",
fcfbd547 495 .data = NULL,
1da177e4
LT		 496 .maxlen = sizeof (int),
497 .mode = 0644,
fcfbd547 498 .proc_handler = &proc_do_ipc_string,
1da177e4
LT		 499 },
500 {
501 .ctl_name = KERN_MSGMAX,
502 .procname = "msgmax",
fcfbd547 503 .data = NULL,
1da177e4
LT		 504 .maxlen = sizeof (int),
505 .mode = 0644,
fcfbd547 506 .proc_handler = &proc_do_ipc_string,
1da177e4
LT		 507 },
508 {
509 .ctl_name = KERN_MSGMNI,
510 .procname = "msgmni",
fcfbd547 511 .data = NULL,
1da177e4
LT		 512 .maxlen = sizeof (int),
513 .mode = 0644,
fcfbd547 514 .proc_handler = &proc_do_ipc_string,
1da177e4
LT		 515 },
516 {
517 .ctl_name = KERN_MSGMNB,
518 .procname = "msgmnb",
fcfbd547 519 .data = NULL,
1da177e4
LT		 520 .maxlen = sizeof (int),
521 .mode = 0644,
fcfbd547 522 .proc_handler = &proc_do_ipc_string,
1da177e4
LT		 523 },
524 {
525 .ctl_name = KERN_SEM,
526 .procname = "sem",
fcfbd547 527 .data = NULL,
1da177e4
LT		 528 .maxlen = 4*sizeof (int),
529 .mode = 0644,
fcfbd547 530 .proc_handler = &proc_do_ipc_string,
1da177e4
LT		 531 },
532#endif
533#ifdef CONFIG_MAGIC_SYSRQ
534 {
535 .ctl_name = KERN_SYSRQ,
536 .procname = "sysrq",
537 .data = &sysrq_enabled,
538 .maxlen = sizeof (int),
539 .mode = 0644,
540 .proc_handler = &proc_dointvec,
541 },
542#endif
543 {
544 .ctl_name = KERN_CADPID,
545 .procname = "cad_pid",
546 .data = &cad_pid,
547 .maxlen = sizeof (int),
548 .mode = 0600,
549 .proc_handler = &proc_dointvec,
550 },
551 {
552 .ctl_name = KERN_MAX_THREADS,
553 .procname = "threads-max",
554 .data = &max_threads,
555 .maxlen = sizeof(int),
556 .mode = 0644,
557 .proc_handler = &proc_dointvec,
558 },
559 {
560 .ctl_name = KERN_RANDOM,
561 .procname = "random",
562 .mode = 0555,
563 .child = random_table,
564 },
565#ifdef CONFIG_UNIX98_PTYS
566 {
567 .ctl_name = KERN_PTY,
568 .procname = "pty",
569 .mode = 0555,
570 .child = pty_table,
571 },
572#endif
573 {
574 .ctl_name = KERN_OVERFLOWUID,
575 .procname = "overflowuid",
576 .data = &overflowuid,
577 .maxlen = sizeof(int),
578 .mode = 0644,
579 .proc_handler = &proc_dointvec_minmax,
580 .strategy = &sysctl_intvec,
581 .extra1 = &minolduid,
582 .extra2 = &maxolduid,
583 },
584 {
585 .ctl_name = KERN_OVERFLOWGID,
586 .procname = "overflowgid",
587 .data = &overflowgid,
588 .maxlen = sizeof(int),
589 .mode = 0644,
590 .proc_handler = &proc_dointvec_minmax,
591 .strategy = &sysctl_intvec,
592 .extra1 = &minolduid,
593 .extra2 = &maxolduid,
594 },
347a8dc3 595#ifdef CONFIG_S390
1da177e4
LT		 596#ifdef CONFIG_MATHEMU
597 {
598 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
599 .procname = "ieee_emulation_warnings",
600 .data = &sysctl_ieee_emulation_warnings,
601 .maxlen = sizeof(int),
602 .mode = 0644,
603 .proc_handler = &proc_dointvec,
604 },
605#endif
606#ifdef CONFIG_NO_IDLE_HZ
607 {
608 .ctl_name = KERN_HZ_TIMER,
609 .procname = "hz_timer",
610 .data = &sysctl_hz_timer,
611 .maxlen = sizeof(int),
612 .mode = 0644,
613 .proc_handler = &proc_dointvec,
614 },
615#endif
616 {
617 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
618 .procname = "userprocess_debug",
619 .data = &sysctl_userprocess_debug,
620 .maxlen = sizeof(int),
621 .mode = 0644,
622 .proc_handler = &proc_dointvec,
623 },
624#endif
625 {
626 .ctl_name = KERN_PIDMAX,
627 .procname = "pid_max",
628 .data = &pid_max,
629 .maxlen = sizeof (int),
630 .mode = 0644,
631 .proc_handler = &proc_dointvec_minmax,
632 .strategy = sysctl_intvec,
633 .extra1 = &pid_max_min,
634 .extra2 = &pid_max_max,
635 },
636 {
637 .ctl_name = KERN_PANIC_ON_OOPS,
638 .procname = "panic_on_oops",
639 .data = &panic_on_oops,
640 .maxlen = sizeof(int),
641 .mode = 0644,
642 .proc_handler = &proc_dointvec,
643 },
644 {
645 .ctl_name = KERN_PRINTK_RATELIMIT,
646 .procname = "printk_ratelimit",
647 .data = &printk_ratelimit_jiffies,
648 .maxlen = sizeof(int),
649 .mode = 0644,
650 .proc_handler = &proc_dointvec_jiffies,
651 .strategy = &sysctl_jiffies,
652 },
653 {
654 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
655 .procname = "printk_ratelimit_burst",
656 .data = &printk_ratelimit_burst,
657 .maxlen = sizeof(int),
658 .mode = 0644,
659 .proc_handler = &proc_dointvec,
660 },
661 {
662 .ctl_name = KERN_NGROUPS_MAX,
663 .procname = "ngroups_max",
664 .data = &ngroups_max,
665 .maxlen = sizeof (int),
666 .mode = 0444,
667 .proc_handler = &proc_dointvec,
668 },
669#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
670 {
671 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
672 .procname = "unknown_nmi_panic",
673 .data = &unknown_nmi_panic,
674 .maxlen = sizeof (int),
675 .mode = 0644,
2fbe7b25 676 .proc_handler = &proc_dointvec,
1da177e4 677 },
407984f1
DZ		 678 {
679 .ctl_name = KERN_NMI_WATCHDOG,
680 .procname = "nmi_watchdog",
681 .data = &nmi_watchdog_enabled,
682 .maxlen = sizeof (int),
683 .mode = 0644,
684 .proc_handler = &proc_nmi_enabled,
1da177e4
LT		 685 },
686#endif
687#if defined(CONFIG_X86)
8da5adda
DZ		 688 {
689 .ctl_name = KERN_PANIC_ON_NMI,
690 .procname = "panic_on_unrecovered_nmi",
691 .data = &panic_on_unrecovered_nmi,
692 .maxlen = sizeof(int),
693 .mode = 0644,
694 .proc_handler = &proc_dointvec,
695 },
1da177e4
LT		 696 {
697 .ctl_name = KERN_BOOTLOADER_TYPE,
698 .procname = "bootloader_type",
699 .data = &bootloader_type,
700 .maxlen = sizeof (int),
701 .mode = 0444,
702 .proc_handler = &proc_dointvec,
703 },
704#endif
7a9166e3 705#if defined(CONFIG_MMU)
1da177e4
LT		 706 {
707 .ctl_name = KERN_RANDOMIZE,
708 .procname = "randomize_va_space",
709 .data = &randomize_va_space,
710 .maxlen = sizeof(int),
711 .mode = 0644,
712 .proc_handler = &proc_dointvec,
713 },
7a9166e3 714#endif
0152fb37 715#if defined(CONFIG_S390) && defined(CONFIG_SMP)
951f22d5
MS		 716 {
717 .ctl_name = KERN_SPIN_RETRY,
718 .procname = "spin_retry",
719 .data = &spin_retry,
720 .maxlen = sizeof (int),
721 .mode = 0644,
722 .proc_handler = &proc_dointvec,
723 },
c255d844
PM		 724#endif
725#ifdef CONFIG_ACPI_SLEEP
726 {
727 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
728 .procname = "acpi_video_flags",
729 .data = &acpi_video_flags,
730 .maxlen = sizeof (unsigned long),
731 .mode = 0644,
7f99f06f 732 .proc_handler = &proc_doulongvec_minmax,
c255d844 733 },
d2b176ed
JS		 734#endif
735#ifdef CONFIG_IA64
736 {
737 .ctl_name = KERN_IA64_UNALIGNED,
738 .procname = "ignore-unaligned-usertrap",
739 .data = &no_unaligned_warning,
740 .maxlen = sizeof (int),
741 .mode = 0644,
742 .proc_handler = &proc_dointvec,
743 },
bebfa101
AK		 744#endif
745#ifdef CONFIG_COMPAT
746 {
747 .ctl_name = KERN_COMPAT_LOG,
748 .procname = "compat-log",
749 .data = &compat_log,
750 .maxlen = sizeof (int),
751 .mode = 0644,
752 .proc_handler = &proc_dointvec,
753 },
951f22d5 754#endif
23f78d4a
IM		 755#ifdef CONFIG_RT_MUTEXES
756 {
757 .ctl_name = KERN_MAX_LOCK_DEPTH,
758 .procname = "max_lock_depth",
759 .data = &max_lock_depth,
760 .maxlen = sizeof(int),
761 .mode = 0644,
762 .proc_handler = &proc_dointvec,
763 },
764#endif
765
1da177e4
LT		 766 { .ctl_name = 0 }
767};
768
769/* Constants for minimum and maximum testing in vm_table.
770 We use these as one-element integer vectors. */
771static int zero;
772static int one_hundred = 100;
773
774
775static ctl_table vm_table[] = {
776 {
777 .ctl_name = VM_OVERCOMMIT_MEMORY,
778 .procname = "overcommit_memory",
779 .data = &sysctl_overcommit_memory,
780 .maxlen = sizeof(sysctl_overcommit_memory),
781 .mode = 0644,
782 .proc_handler = &proc_dointvec,
783 },
fadd8fbd
KH		 784 {
785 .ctl_name = VM_PANIC_ON_OOM,
786 .procname = "panic_on_oom",
787 .data = &sysctl_panic_on_oom,
788 .maxlen = sizeof(sysctl_panic_on_oom),
789 .mode = 0644,
790 .proc_handler = &proc_dointvec,
791 },
1da177e4
LT		 792 {
793 .ctl_name = VM_OVERCOMMIT_RATIO,
794 .procname = "overcommit_ratio",
795 .data = &sysctl_overcommit_ratio,
796 .maxlen = sizeof(sysctl_overcommit_ratio),
797 .mode = 0644,
798 .proc_handler = &proc_dointvec,
799 },
800 {
801 .ctl_name = VM_PAGE_CLUSTER,
802 .procname = "page-cluster",
803 .data = &page_cluster,
804 .maxlen = sizeof(int),
805 .mode = 0644,
806 .proc_handler = &proc_dointvec,
807 },
808 {
809 .ctl_name = VM_DIRTY_BACKGROUND,
810 .procname = "dirty_background_ratio",
811 .data = &dirty_background_ratio,
812 .maxlen = sizeof(dirty_background_ratio),
813 .mode = 0644,
814 .proc_handler = &proc_dointvec_minmax,
815 .strategy = &sysctl_intvec,
816 .extra1 = &zero,
817 .extra2 = &one_hundred,
818 },
819 {
820 .ctl_name = VM_DIRTY_RATIO,
821 .procname = "dirty_ratio",
822 .data = &vm_dirty_ratio,
823 .maxlen = sizeof(vm_dirty_ratio),
824 .mode = 0644,
825 .proc_handler = &proc_dointvec_minmax,
826 .strategy = &sysctl_intvec,
827 .extra1 = &zero,
828 .extra2 = &one_hundred,
829 },
830 {
831 .ctl_name = VM_DIRTY_WB_CS,
832 .procname = "dirty_writeback_centisecs",
f6ef9438
BS		 833 .data = &dirty_writeback_interval,
834 .maxlen = sizeof(dirty_writeback_interval),
1da177e4
LT		 835 .mode = 0644,
836 .proc_handler = &dirty_writeback_centisecs_handler,
837 },
838 {
839 .ctl_name = VM_DIRTY_EXPIRE_CS,
840 .procname = "dirty_expire_centisecs",
f6ef9438
BS		 841 .data = &dirty_expire_interval,
842 .maxlen = sizeof(dirty_expire_interval),
1da177e4 843 .mode = 0644,
f6ef9438 844 .proc_handler = &proc_dointvec_userhz_jiffies,
1da177e4
LT		 845 },
846 {
847 .ctl_name = VM_NR_PDFLUSH_THREADS,
848 .procname = "nr_pdflush_threads",
849 .data = &nr_pdflush_threads,
850 .maxlen = sizeof nr_pdflush_threads,
851 .mode = 0444 /* read-only*/,
852 .proc_handler = &proc_dointvec,
853 },
854 {
855 .ctl_name = VM_SWAPPINESS,
856 .procname = "swappiness",
857 .data = &vm_swappiness,
858 .maxlen = sizeof(vm_swappiness),
859 .mode = 0644,
860 .proc_handler = &proc_dointvec_minmax,
861 .strategy = &sysctl_intvec,
862 .extra1 = &zero,
863 .extra2 = &one_hundred,
864 },
865#ifdef CONFIG_HUGETLB_PAGE
866 {
867 .ctl_name = VM_HUGETLB_PAGES,
868 .procname = "nr_hugepages",
869 .data = &max_huge_pages,
870 .maxlen = sizeof(unsigned long),
871 .mode = 0644,
872 .proc_handler = &hugetlb_sysctl_handler,
873 .extra1 = (void *)&hugetlb_zero,
874 .extra2 = (void *)&hugetlb_infinity,
875 },
876 {
877 .ctl_name = VM_HUGETLB_GROUP,
878 .procname = "hugetlb_shm_group",
879 .data = &sysctl_hugetlb_shm_group,
880 .maxlen = sizeof(gid_t),
881 .mode = 0644,
882 .proc_handler = &proc_dointvec,
883 },
884#endif
885 {
886 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
887 .procname = "lowmem_reserve_ratio",
888 .data = &sysctl_lowmem_reserve_ratio,
889 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
890 .mode = 0644,
891 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
892 .strategy = &sysctl_intvec,
893 },
9d0243bc
AM		 894 {
895 .ctl_name = VM_DROP_PAGECACHE,
896 .procname = "drop_caches",
897 .data = &sysctl_drop_caches,
898 .maxlen = sizeof(int),
899 .mode = 0644,
900 .proc_handler = drop_caches_sysctl_handler,
901 .strategy = &sysctl_intvec,
902 },
1da177e4
LT		 903 {
904 .ctl_name = VM_MIN_FREE_KBYTES,
905 .procname = "min_free_kbytes",
906 .data = &min_free_kbytes,
907 .maxlen = sizeof(min_free_kbytes),
908 .mode = 0644,
909 .proc_handler = &min_free_kbytes_sysctl_handler,
910 .strategy = &sysctl_intvec,
911 .extra1 = &zero,
912 },
8ad4b1fb
RS		 913 {
914 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
915 .procname = "percpu_pagelist_fraction",
916 .data = &percpu_pagelist_fraction,
917 .maxlen = sizeof(percpu_pagelist_fraction),
918 .mode = 0644,
919 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
920 .strategy = &sysctl_intvec,
921 .extra1 = &min_percpu_pagelist_fract,
922 },
1da177e4
LT		 923#ifdef CONFIG_MMU
924 {
925 .ctl_name = VM_MAX_MAP_COUNT,
926 .procname = "max_map_count",
927 .data = &sysctl_max_map_count,
928 .maxlen = sizeof(sysctl_max_map_count),
929 .mode = 0644,
930 .proc_handler = &proc_dointvec
931 },
932#endif
933 {
934 .ctl_name = VM_LAPTOP_MODE,
935 .procname = "laptop_mode",
936 .data = &laptop_mode,
937 .maxlen = sizeof(laptop_mode),
938 .mode = 0644,
ed5b43f1
BS		 939 .proc_handler = &proc_dointvec_jiffies,
940 .strategy = &sysctl_jiffies,
1da177e4
LT		 941 },
942 {
943 .ctl_name = VM_BLOCK_DUMP,
944 .procname = "block_dump",
945 .data = &block_dump,
946 .maxlen = sizeof(block_dump),
947 .mode = 0644,
948 .proc_handler = &proc_dointvec,
949 .strategy = &sysctl_intvec,
950 .extra1 = &zero,
951 },
952 {
953 .ctl_name = VM_VFS_CACHE_PRESSURE,
954 .procname = "vfs_cache_pressure",
955 .data = &sysctl_vfs_cache_pressure,
956 .maxlen = sizeof(sysctl_vfs_cache_pressure),
957 .mode = 0644,
958 .proc_handler = &proc_dointvec,
959 .strategy = &sysctl_intvec,
960 .extra1 = &zero,
961 },
962#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
963 {
964 .ctl_name = VM_LEGACY_VA_LAYOUT,
965 .procname = "legacy_va_layout",
966 .data = &sysctl_legacy_va_layout,
967 .maxlen = sizeof(sysctl_legacy_va_layout),
968 .mode = 0644,
969 .proc_handler = &proc_dointvec,
970 .strategy = &sysctl_intvec,
971 .extra1 = &zero,
972 },
973#endif
974#ifdef CONFIG_SWAP
975 {
976 .ctl_name = VM_SWAP_TOKEN_TIMEOUT,
977 .procname = "swap_token_timeout",
978 .data = &swap_token_default_timeout,
979 .maxlen = sizeof(swap_token_default_timeout),
980 .mode = 0644,
981 .proc_handler = &proc_dointvec_jiffies,
982 .strategy = &sysctl_jiffies,
983 },
1743660b
CL		 984#endif
985#ifdef CONFIG_NUMA
986 {
987 .ctl_name = VM_ZONE_RECLAIM_MODE,
988 .procname = "zone_reclaim_mode",
989 .data = &zone_reclaim_mode,
990 .maxlen = sizeof(zone_reclaim_mode),
991 .mode = 0644,
992 .proc_handler = &proc_dointvec,
c84db23c
CL		 993 .strategy = &sysctl_intvec,
994 .extra1 = &zero,
1743660b 995 },
9614634f
CL		 996 {
997 .ctl_name = VM_MIN_UNMAPPED,
998 .procname = "min_unmapped_ratio",
999 .data = &sysctl_min_unmapped_ratio,
1000 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1001 .mode = 0644,
1002 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
1003 .strategy = &sysctl_intvec,
1004 .extra1 = &zero,
1005 .extra2 = &one_hundred,
1006 },
0ff38490
CL		 1007 {
1008 .ctl_name = VM_MIN_SLAB,
1009 .procname = "min_slab_ratio",
1010 .data = &sysctl_min_slab_ratio,
1011 .maxlen = sizeof(sysctl_min_slab_ratio),
1012 .mode = 0644,
1013 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1014 .strategy = &sysctl_intvec,
1015 .extra1 = &zero,
1016 .extra2 = &one_hundred,
1017 },
e6e5494c
IM		 1018#endif
1019#ifdef CONFIG_X86_32
1020 {
1021 .ctl_name = VM_VDSO_ENABLED,
1022 .procname = "vdso_enabled",
1023 .data = &vdso_enabled,
1024 .maxlen = sizeof(vdso_enabled),
1025 .mode = 0644,
1026 .proc_handler = &proc_dointvec,
1027 .strategy = &sysctl_intvec,
1028 .extra1 = &zero,
1029 },
1da177e4
LT		 1030#endif
1031 { .ctl_name = 0 }
1032};
1033
1da177e4
LT		 1034static ctl_table fs_table[] = {
1035 {
1036 .ctl_name = FS_NRINODE,
1037 .procname = "inode-nr",
1038 .data = &inodes_stat,
1039 .maxlen = 2*sizeof(int),
1040 .mode = 0444,
1041 .proc_handler = &proc_dointvec,
1042 },
1043 {
1044 .ctl_name = FS_STATINODE,
1045 .procname = "inode-state",
1046 .data = &inodes_stat,
1047 .maxlen = 7*sizeof(int),
1048 .mode = 0444,
1049 .proc_handler = &proc_dointvec,
1050 },
1051 {
1052 .ctl_name = FS_NRFILE,
1053 .procname = "file-nr",
1054 .data = &files_stat,
1055 .maxlen = 3*sizeof(int),
1056 .mode = 0444,
529bf6be 1057 .proc_handler = &proc_nr_files,
1da177e4
LT		 1058 },
1059 {
1060 .ctl_name = FS_MAXFILE,
1061 .procname = "file-max",
1062 .data = &files_stat.max_files,
1063 .maxlen = sizeof(int),
1064 .mode = 0644,
1065 .proc_handler = &proc_dointvec,
1066 },
1067 {
1068 .ctl_name = FS_DENTRY,
1069 .procname = "dentry-state",
1070 .data = &dentry_stat,
1071 .maxlen = 6*sizeof(int),
1072 .mode = 0444,
1073 .proc_handler = &proc_dointvec,
1074 },
1075 {
1076 .ctl_name = FS_OVERFLOWUID,
1077 .procname = "overflowuid",
1078 .data = &fs_overflowuid,
1079 .maxlen = sizeof(int),
1080 .mode = 0644,
1081 .proc_handler = &proc_dointvec_minmax,
1082 .strategy = &sysctl_intvec,
1083 .extra1 = &minolduid,
1084 .extra2 = &maxolduid,
1085 },
1086 {
1087 .ctl_name = FS_OVERFLOWGID,
1088 .procname = "overflowgid",
1089 .data = &fs_overflowgid,
1090 .maxlen = sizeof(int),
1091 .mode = 0644,
1092 .proc_handler = &proc_dointvec_minmax,
1093 .strategy = &sysctl_intvec,
1094 .extra1 = &minolduid,
1095 .extra2 = &maxolduid,
1096 },
1097 {
1098 .ctl_name = FS_LEASES,
1099 .procname = "leases-enable",
1100 .data = &leases_enable,
1101 .maxlen = sizeof(int),
1102 .mode = 0644,
1103 .proc_handler = &proc_dointvec,
1104 },
1105#ifdef CONFIG_DNOTIFY
1106 {
1107 .ctl_name = FS_DIR_NOTIFY,
1108 .procname = "dir-notify-enable",
1109 .data = &dir_notify_enable,
1110 .maxlen = sizeof(int),
1111 .mode = 0644,
1112 .proc_handler = &proc_dointvec,
1113 },
1114#endif
1115#ifdef CONFIG_MMU
1116 {
1117 .ctl_name = FS_LEASE_TIME,
1118 .procname = "lease-break-time",
1119 .data = &lease_break_time,
1120 .maxlen = sizeof(int),
1121 .mode = 0644,
1122 .proc_handler = &proc_dointvec,
1123 },
1124 {
1125 .ctl_name = FS_AIO_NR,
1126 .procname = "aio-nr",
1127 .data = &aio_nr,
1128 .maxlen = sizeof(aio_nr),
1129 .mode = 0444,
d55b5fda 1130 .proc_handler = &proc_doulongvec_minmax,
1da177e4
LT		 1131 },
1132 {
1133 .ctl_name = FS_AIO_MAX_NR,
1134 .procname = "aio-max-nr",
1135 .data = &aio_max_nr,
1136 .maxlen = sizeof(aio_max_nr),
1137 .mode = 0644,
d55b5fda 1138 .proc_handler = &proc_doulongvec_minmax,
1da177e4 1139 },
2d9048e2 1140#ifdef CONFIG_INOTIFY_USER
0399cb08
RL		 1141 {
1142 .ctl_name = FS_INOTIFY,
1143 .procname = "inotify",
1144 .mode = 0555,
1145 .child = inotify_table,
1146 },
1147#endif
1da177e4 1148#endif
d6e71144
AC		 1149 {
1150 .ctl_name = KERN_SETUID_DUMPABLE,
1151 .procname = "suid_dumpable",
1152 .data = &suid_dumpable,
1153 .maxlen = sizeof(int),
1154 .mode = 0644,
1155 .proc_handler = &proc_dointvec,
1156 },
1da177e4
LT		 1157 { .ctl_name = 0 }
1158};
1159
1160static ctl_table debug_table[] = {
1161 { .ctl_name = 0 }
1162};
1163
1164static ctl_table dev_table[] = {
1165 { .ctl_name = 0 }
0eeca283 1166};
1da177e4
LT		 1167
1168extern void init_irq_proc (void);
1169
330d57fb
AV		 1170static DEFINE_SPINLOCK(sysctl_lock);
1171
1172/* called under sysctl_lock */
1173static int use_table(struct ctl_table_header *p)
1174{
1175 if (unlikely(p->unregistering))
1176 return 0;
1177 p->used++;
1178 return 1;
1179}
1180
1181/* called under sysctl_lock */
1182static void unuse_table(struct ctl_table_header *p)
1183{
1184 if (!--p->used)
1185 if (unlikely(p->unregistering))
1186 complete(p->unregistering);
1187}
1188
1189/* called under sysctl_lock, will reacquire if has to wait */
1190static void start_unregistering(struct ctl_table_header *p)
1191{
1192 /*
1193 * if p->used is 0, nobody will ever touch that entry again;
1194 * we'll eliminate all paths to it before dropping sysctl_lock
1195 */
1196 if (unlikely(p->used)) {
1197 struct completion wait;
1198 init_completion(&wait);
1199 p->unregistering = &wait;
1200 spin_unlock(&sysctl_lock);
1201 wait_for_completion(&wait);
1202 spin_lock(&sysctl_lock);
1203 }
1204 /*
1205 * do not remove from the list until nobody holds it; walking the
1206 * list in do_sysctl() relies on that.
1207 */
1208 list_del_init(&p->ctl_entry);
1209}
1210
1da177e4
LT		 1211void __init sysctl_init(void)
1212{
b89a8171 1213#ifdef CONFIG_PROC_SYSCTL
330d57fb 1214 register_proc_table(root_table, proc_sys_root, &root_table_header);
1da177e4
LT		 1215 init_irq_proc();
1216#endif
1217}
1218
b89a8171 1219#ifdef CONFIG_SYSCTL_SYSCALL
1da177e4
LT		 1220int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1221 void __user *newval, size_t newlen)
1222{
1223 struct list_head *tmp;
330d57fb 1224 int error = -ENOTDIR;
1da177e4
LT		 1225
1226 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1227 return -ENOTDIR;
1228 if (oldval) {
1229 int old_len;
1230 if (!oldlenp || get_user(old_len, oldlenp))
1231 return -EFAULT;
1232 }
330d57fb 1233 spin_lock(&sysctl_lock);
1da177e4
LT		 1234 tmp = &root_table_header.ctl_entry;
1235 do {
1236 struct ctl_table_header *head =
1237 list_entry(tmp, struct ctl_table_header, ctl_entry);
1238 void *context = NULL;
330d57fb
AV		 1239
1240 if (!use_table(head))
1241 continue;
1242
1243 spin_unlock(&sysctl_lock);
1244
1245 error = parse_table(name, nlen, oldval, oldlenp,
1da177e4
LT		 1246 newval, newlen, head->ctl_table,
1247 &context);
5a6b454f 1248 kfree(context);
330d57fb
AV		 1249
1250 spin_lock(&sysctl_lock);
1251 unuse_table(head);
1da177e4 1252 if (error != -ENOTDIR)
330d57fb
AV		 1253 break;
1254 } while ((tmp = tmp->next) != &root_table_header.ctl_entry);
1255 spin_unlock(&sysctl_lock);
1256 return error;
1da177e4
LT		 1257}
1258
1259asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1260{
1261 struct __sysctl_args tmp;
1262 int error;
1263
1264 if (copy_from_user(&tmp, args, sizeof(tmp)))
1265 return -EFAULT;
1266
1267 lock_kernel();
1268 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1269 tmp.newval, tmp.newlen);
1270 unlock_kernel();
1271 return error;
1272}
b89a8171 1273#endif /* CONFIG_SYSCTL_SYSCALL */
1da177e4
LT		 1274
1275/*
1276 * ctl_perm does NOT grant the superuser all rights automatically, because
1277 * some sysctl variables are readonly even to root.
1278 */
1279
1280static int test_perm(int mode, int op)
1281{
1282 if (!current->euid)
1283 mode >>= 6;
1284 else if (in_egroup_p(0))
1285 mode >>= 3;
1286 if ((mode & op & 0007) == op)
1287 return 0;
1288 return -EACCES;
1289}
1290
1291static inline int ctl_perm(ctl_table *table, int op)
1292{
1293 int error;
1294 error = security_sysctl(table, op);
1295 if (error)
1296 return error;
1297 return test_perm(table->mode, op);
1298}
1299
b89a8171 1300#ifdef CONFIG_SYSCTL_SYSCALL
1da177e4
LT		 1301static int parse_table(int __user *name, int nlen,
1302 void __user *oldval, size_t __user *oldlenp,
1303 void __user *newval, size_t newlen,
1304 ctl_table *table, void **context)
1305{
1306 int n;
1307repeat:
1308 if (!nlen)
1309 return -ENOTDIR;
1310 if (get_user(n, name))
1311 return -EFAULT;
1312 for ( ; table->ctl_name; table++) {
1313 if (n == table->ctl_name || table->ctl_name == CTL_ANY) {
1314 int error;
1315 if (table->child) {
1316 if (ctl_perm(table, 001))
1317 return -EPERM;
1318 if (table->strategy) {
1319 error = table->strategy(
1320 table, name, nlen,
1321 oldval, oldlenp,
1322 newval, newlen, context);
1323 if (error)
1324 return error;
1325 }
1326 name++;
1327 nlen--;
1328 table = table->child;
1329 goto repeat;
1330 }
1331 error = do_sysctl_strategy(table, name, nlen,
1332 oldval, oldlenp,
1333 newval, newlen, context);
1334 return error;
1335 }
1336 }
1337 return -ENOTDIR;
1338}
1339
1340/* Perform the actual read/write of a sysctl table entry. */
1341int do_sysctl_strategy (ctl_table *table,
1342 int __user *name, int nlen,
1343 void __user *oldval, size_t __user *oldlenp,
1344 void __user *newval, size_t newlen, void **context)
1345{
1346 int op = 0, rc;
1347 size_t len;
1348
1349 if (oldval)
1350 op |= 004;
1351 if (newval)
1352 op |= 002;
1353 if (ctl_perm(table, op))
1354 return -EPERM;
1355
1356 if (table->strategy) {
1357 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1358 newval, newlen, context);
1359 if (rc < 0)
1360 return rc;
1361 if (rc > 0)
1362 return 0;
1363 }
1364
1365 /* If there is no strategy routine, or if the strategy returns
1366 * zero, proceed with automatic r/w */
1367 if (table->data && table->maxlen) {
1368 if (oldval && oldlenp) {
1369 if (get_user(len, oldlenp))
1370 return -EFAULT;
1371 if (len) {
1372 if (len > table->maxlen)
1373 len = table->maxlen;
1374 if(copy_to_user(oldval, table->data, len))
1375 return -EFAULT;
1376 if(put_user(len, oldlenp))
1377 return -EFAULT;
1378 }
1379 }
1380 if (newval && newlen) {
1381 len = newlen;
1382 if (len > table->maxlen)
1383 len = table->maxlen;
1384 if(copy_from_user(table->data, newval, len))
1385 return -EFAULT;
1386 }
1387 }
1388 return 0;
1389}
b89a8171 1390#endif /* CONFIG_SYSCTL_SYSCALL */
1da177e4
LT		 1391
1392/**
1393 * register_sysctl_table - register a sysctl hierarchy
1394 * @table: the top-level table structure
1395 * @insert_at_head: whether the entry should be inserted in front or at the end
1396 *
1397 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1398 * array. An entry with a ctl_name of 0 terminates the table.
1399 *
1400 * The members of the &ctl_table structure are used as follows:
1401 *
1402 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1403 * must be unique within that level of sysctl
1404 *
1405 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1406 * enter a sysctl file
1407 *
1408 * data - a pointer to data for use by proc_handler
1409 *
1410 * maxlen - the maximum size in bytes of the data
1411 *
1412 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1413 *
1414 * child - a pointer to the child sysctl table if this entry is a directory, or
1415 * %NULL.
1416 *
1417 * proc_handler - the text handler routine (described below)
1418 *
1419 * strategy - the strategy routine (described below)
1420 *
1421 * de - for internal use by the sysctl routines
1422 *
1423 * extra1, extra2 - extra pointers usable by the proc handler routines
1424 *
1425 * Leaf nodes in the sysctl tree will be represented by a single file
1426 * under /proc; non-leaf nodes will be represented by directories.
1427 *
1428 * sysctl(2) can automatically manage read and write requests through
1429 * the sysctl table. The data and maxlen fields of the ctl_table
1430 * struct enable minimal validation of the values being written to be
1431 * performed, and the mode field allows minimal authentication.
1432 *
1433 * More sophisticated management can be enabled by the provision of a
1434 * strategy routine with the table entry. This will be called before
1435 * any automatic read or write of the data is performed.
1436 *
1437 * The strategy routine may return
1438 *
1439 * < 0 - Error occurred (error is passed to user process)
1440 *
1441 * 0 - OK - proceed with automatic read or write.
1442 *
1443 * > 0 - OK - read or write has been done by the strategy routine, so
1444 * return immediately.
1445 *
1446 * There must be a proc_handler routine for any terminal nodes
1447 * mirrored under /proc/sys (non-terminals are handled by a built-in
1448 * directory handler). Several default handlers are available to
1449 * cover common cases -
1450 *
1451 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1452 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1453 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1454 *
1455 * It is the handler's job to read the input buffer from user memory
1456 * and process it. The handler should return 0 on success.
1457 *
1458 * This routine returns %NULL on a failure to register, and a pointer
1459 * to the table header on success.
1460 */
1461struct ctl_table_header *register_sysctl_table(ctl_table * table,
1462 int insert_at_head)
1463{
1464 struct ctl_table_header *tmp;
1465 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1466 if (!tmp)
1467 return NULL;
1468 tmp->ctl_table = table;
1469 INIT_LIST_HEAD(&tmp->ctl_entry);
330d57fb
AV		 1470 tmp->used = 0;
1471 tmp->unregistering = NULL;
1472 spin_lock(&sysctl_lock);
1da177e4
LT		 1473 if (insert_at_head)
1474 list_add(&tmp->ctl_entry, &root_table_header.ctl_entry);
1475 else
1476 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
330d57fb 1477 spin_unlock(&sysctl_lock);
b89a8171 1478#ifdef CONFIG_PROC_SYSCTL
330d57fb 1479 register_proc_table(table, proc_sys_root, tmp);
1da177e4
LT		 1480#endif
1481 return tmp;
1482}
1483
1484/**
1485 * unregister_sysctl_table - unregister a sysctl table hierarchy
1486 * @header: the header returned from register_sysctl_table
1487 *
1488 * Unregisters the sysctl table and all children. proc entries may not
1489 * actually be removed until they are no longer used by anyone.
1490 */
1491void unregister_sysctl_table(struct ctl_table_header * header)
1492{
330d57fb
AV		 1493 might_sleep();
1494 spin_lock(&sysctl_lock);
1495 start_unregistering(header);
b89a8171 1496#ifdef CONFIG_PROC_SYSCTL
1da177e4
LT		 1497 unregister_proc_table(header->ctl_table, proc_sys_root);
1498#endif
330d57fb 1499 spin_unlock(&sysctl_lock);
1da177e4
LT		 1500 kfree(header);
1501}
1502
b89a8171
EB		 1503#else /* !CONFIG_SYSCTL */
1504struct ctl_table_header * register_sysctl_table(ctl_table * table,
1505 int insert_at_head)
1506{
1507 return NULL;
1508}
1509
1510void unregister_sysctl_table(struct ctl_table_header * table)
1511{
1512}
1513
1514#endif /* CONFIG_SYSCTL */
1515
1da177e4
LT		 1516/*
1517 * /proc/sys support
1518 */
1519
b89a8171 1520#ifdef CONFIG_PROC_SYSCTL
1da177e4
LT		 1521
1522/* Scan the sysctl entries in table and add them all into /proc */
330d57fb 1523static void register_proc_table(ctl_table * table, struct proc_dir_entry *root, void *set)
1da177e4
LT		 1524{
1525 struct proc_dir_entry *de;
1526 int len;
1527 mode_t mode;
1528
1529 for (; table->ctl_name; table++) {
1530 /* Can't do anything without a proc name. */
1531 if (!table->procname)
1532 continue;
1533 /* Maybe we can't do anything with it... */
1534 if (!table->proc_handler && !table->child) {
1535 printk(KERN_WARNING "SYSCTL: Can't register %s\n",
1536 table->procname);
1537 continue;
1538 }
1539
1540 len = strlen(table->procname);
1541 mode = table->mode;
1542
1543 de = NULL;
1544 if (table->proc_handler)
1545 mode |= S_IFREG;
1546 else {
1547 mode |= S_IFDIR;
1548 for (de = root->subdir; de; de = de->next) {
1549 if (proc_match(len, table->procname, de))
1550 break;
1551 }
1552 /* If the subdir exists already, de is non-NULL */
1553 }
1554
1555 if (!de) {
1556 de = create_proc_entry(table->procname, mode, root);
1557 if (!de)
1558 continue;
330d57fb 1559 de->set = set;
1da177e4
LT		 1560 de->data = (void *) table;
1561 if (table->proc_handler)
1562 de->proc_fops = &proc_sys_file_operations;
1563 }
1564 table->de = de;
1565 if (de->mode & S_IFDIR)
330d57fb 1566 register_proc_table(table->child, de, set);
1da177e4
LT		 1567 }
1568}
1569
1570/*
1571 * Unregister a /proc sysctl table and any subdirectories.
1572 */
1573static void unregister_proc_table(ctl_table * table, struct proc_dir_entry *root)
1574{
1575 struct proc_dir_entry *de;
1576 for (; table->ctl_name; table++) {
1577 if (!(de = table->de))
1578 continue;
1579 if (de->mode & S_IFDIR) {
1580 if (!table->child) {
1581 printk (KERN_ALERT "Help - malformed sysctl tree on free\n");
1582 continue;
1583 }
1584 unregister_proc_table(table->child, de);
1585
1586 /* Don't unregister directories which still have entries.. */
1587 if (de->subdir)
1588 continue;
1589 }
1590
330d57fb
AV		 1591 /*
1592 * In any case, mark the entry as goner; we'll keep it
1593 * around if it's busy, but we'll know to do nothing with
1594 * its fields. We are under sysctl_lock here.
1595 */
1596 de->data = NULL;
1597
1da177e4
LT		 1598 /* Don't unregister proc entries that are still being used.. */
1599 if (atomic_read(&de->count))
1600 continue;
1601
1602 table->de = NULL;
1603 remove_proc_entry(table->procname, root);
1604 }
1605}
1606
1607static ssize_t do_rw_proc(int write, struct file * file, char __user * buf,
1608 size_t count, loff_t *ppos)
1609{
1610 int op;
330d57fb 1611 struct proc_dir_entry *de = PDE(file->f_dentry->d_inode);
1da177e4
LT		 1612 struct ctl_table *table;
1613 size_t res;
330d57fb 1614 ssize_t error = -ENOTDIR;
1da177e4 1615
330d57fb
AV		 1616 spin_lock(&sysctl_lock);
1617 if (de && de->data && use_table(de->set)) {
1618 /*
1619 * at that point we know that sysctl was not unregistered
1620 * and won't be until we finish
1621 */
1622 spin_unlock(&sysctl_lock);
1623 table = (struct ctl_table *) de->data;
1624 if (!table || !table->proc_handler)
1625 goto out;
1626 error = -EPERM;
1627 op = (write ? 002 : 004);
1628 if (ctl_perm(table, op))
1629 goto out;
1630
1631 /* careful: calling conventions are nasty here */
1632 res = count;
1633 error = (*table->proc_handler)(table, write, file,
1634 buf, &res, ppos);
1635 if (!error)
1636 error = res;
1637 out:
1638 spin_lock(&sysctl_lock);
1639 unuse_table(de->set);
1640 }
1641 spin_unlock(&sysctl_lock);
1642 return error;
1da177e4
LT		 1643}
1644
1645static int proc_opensys(struct inode *inode, struct file *file)
1646{
1647 if (file->f_mode & FMODE_WRITE) {
1648 /*
1649 * sysctl entries that are not writable,
1650 * are _NOT_ writable, capabilities or not.
1651 */
1652 if (!(inode->i_mode & S_IWUSR))
1653 return -EPERM;
1654 }
1655
1656 return 0;
1657}
1658
1659static ssize_t proc_readsys(struct file * file, char __user * buf,
1660 size_t count, loff_t *ppos)
1661{
1662 return do_rw_proc(0, file, buf, count, ppos);
1663}
1664
1665static ssize_t proc_writesys(struct file * file, const char __user * buf,
1666 size_t count, loff_t *ppos)
1667{
1668 return do_rw_proc(1, file, (char __user *) buf, count, ppos);
1669}
1670
b1ba4ddd
AB		 1671static int _proc_do_string(void* data, int maxlen, int write,
1672 struct file *filp, void __user *buffer,
1673 size_t *lenp, loff_t *ppos)
1da177e4
LT		 1674{
1675 size_t len;
1676 char __user *p;
1677 char c;
1678
f5dd3d6f 1679 if (!data || !maxlen || !*lenp ||
1da177e4
LT		 1680 (*ppos && !write)) {
1681 *lenp = 0;
1682 return 0;
1683 }
1684
1685 if (write) {
1686 len = 0;
1687 p = buffer;
1688 while (len < *lenp) {
1689 if (get_user(c, p++))
1690 return -EFAULT;
1691 if (c == 0 || c == '\n')
1692 break;
1693 len++;
1694 }
f5dd3d6f
SV		 1695 if (len >= maxlen)
1696 len = maxlen-1;
1697 if(copy_from_user(data, buffer, len))
1da177e4 1698 return -EFAULT;
f5dd3d6f 1699 ((char *) data)[len] = 0;
1da177e4
LT		 1700 *ppos += *lenp;
1701 } else {
f5dd3d6f
SV		 1702 len = strlen(data);
1703 if (len > maxlen)
1704 len = maxlen;
1da177e4
LT		 1705 if (len > *lenp)
1706 len = *lenp;
1707 if (len)
f5dd3d6f 1708 if(copy_to_user(buffer, data, len))
1da177e4
LT		 1709 return -EFAULT;
1710 if (len < *lenp) {
1711 if(put_user('\n', ((char __user *) buffer) + len))
1712 return -EFAULT;
1713 len++;
1714 }
1715 *lenp = len;
1716 *ppos += len;
1717 }
1718 return 0;
1719}
1720
f5dd3d6f
SV		 1721/**
1722 * proc_dostring - read a string sysctl
1723 * @table: the sysctl table
1724 * @write: %TRUE if this is a write to the sysctl file
1725 * @filp: the file structure
1726 * @buffer: the user buffer
1727 * @lenp: the size of the user buffer
1728 * @ppos: file position
1729 *
1730 * Reads/writes a string from/to the user buffer. If the kernel
1731 * buffer provided is not large enough to hold the string, the
1732 * string is truncated. The copied string is %NULL-terminated.
1733 * If the string is being read by the user process, it is copied
1734 * and a newline '\n' is added. It is truncated if the buffer is
1735 * not large enough.
1736 *
1737 * Returns 0 on success.
1738 */
1739int proc_dostring(ctl_table *table, int write, struct file *filp,
1740 void __user *buffer, size_t *lenp, loff_t *ppos)
1741{
1742 return _proc_do_string(table->data, table->maxlen, write, filp,
1743 buffer, lenp, ppos);
1744}
1745
1da177e4
LT		 1746/*
1747 * Special case of dostring for the UTS structure. This has locks
1748 * to observe. Should this be in kernel/sys.c ????
1749 */
1750
8218c74c
SH		 1751#ifndef CONFIG_UTS_NS
1752static int proc_do_uts_string(ctl_table *table, int write, struct file *filp,
1da177e4
LT		 1753 void __user *buffer, size_t *lenp, loff_t *ppos)
1754{
1755 int r;
1756
1757 if (!write) {
1758 down_read(&uts_sem);
1759 r=proc_dostring(table,0,filp,buffer,lenp, ppos);
1760 up_read(&uts_sem);
1761 } else {
1762 down_write(&uts_sem);
1763 r=proc_dostring(table,1,filp,buffer,lenp, ppos);
1764 up_write(&uts_sem);
1765 }
1766 return r;
1767}
8218c74c
SH		 1768#else /* !CONFIG_UTS_NS */
1769static int proc_do_uts_string(ctl_table *table, int write, struct file *filp,
1770 void __user *buffer, size_t *lenp, loff_t *ppos)
1771{
1772 int r;
1773 struct uts_namespace* uts_ns = current->nsproxy->uts_ns;
1774 char* which;
1775
1776 switch (table->ctl_name) {
1777 case KERN_OSTYPE:
1778 which = uts_ns->name.sysname;
1779 break;
1780 case KERN_NODENAME:
1781 which = uts_ns->name.nodename;
1782 break;
1783 case KERN_OSRELEASE:
1784 which = uts_ns->name.release;
1785 break;
1786 case KERN_VERSION:
1787 which = uts_ns->name.version;
1788 break;
1789 case KERN_DOMAINNAME:
1790 which = uts_ns->name.domainname;
1791 break;
1792 default:
1793 r = -EINVAL;
1794 goto out;
1795 }
1796
1797 if (!write) {
1798 down_read(&uts_sem);
1799 r=_proc_do_string(which,table->maxlen,0,filp,buffer,lenp, ppos);
1800 up_read(&uts_sem);
1801 } else {
1802 down_write(&uts_sem);
1803 r=_proc_do_string(which,table->maxlen,1,filp,buffer,lenp, ppos);
1804 up_write(&uts_sem);
1805 }
1806 out:
1807 return r;
1808}
1809#endif /* !CONFIG_UTS_NS */
1da177e4
LT		 1810
1811static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1812 int *valp,
1813 int write, void *data)
1814{
1815 if (write) {
1816 *valp = *negp ? -*lvalp : *lvalp;
1817 } else {
1818 int val = *valp;
1819 if (val < 0) {
1820 *negp = -1;
1821 *lvalp = (unsigned long)-val;
1822 } else {
1823 *negp = 0;
1824 *lvalp = (unsigned long)val;
1825 }
1826 }
1827 return 0;
1828}
1829
fcfbd547
KK		 1830static int __do_proc_dointvec(void *tbl_data, ctl_table *table,
1831 int write, struct file *filp, void __user *buffer,
1832 size_t *lenp, loff_t *ppos,
1da177e4
LT		 1833 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1834 int write, void *data),
1835 void *data)
1836{
1837#define TMPBUFLEN 21
1838 int *i, vleft, first=1, neg, val;
1839 unsigned long lval;
1840 size_t left, len;
1841
1842 char buf[TMPBUFLEN], *p;
1843 char __user *s = buffer;
1844
fcfbd547 1845 if (!tbl_data || !table->maxlen || !*lenp ||
1da177e4
LT		 1846 (*ppos && !write)) {
1847 *lenp = 0;
1848 return 0;
1849 }
1850
fcfbd547 1851 i = (int *) tbl_data;
1da177e4
LT		 1852 vleft = table->maxlen / sizeof(*i);
1853 left = *lenp;
1854
1855 if (!conv)
1856 conv = do_proc_dointvec_conv;
1857
1858 for (; left && vleft--; i++, first=0) {
1859 if (write) {
1860 while (left) {
1861 char c;
1862 if (get_user(c, s))
1863 return -EFAULT;
1864 if (!isspace(c))
1865 break;
1866 left--;
1867 s++;
1868 }
1869 if (!left)
1870 break;
1871 neg = 0;
1872 len = left;
1873 if (len > sizeof(buf) - 1)
1874 len = sizeof(buf) - 1;
1875 if (copy_from_user(buf, s, len))
1876 return -EFAULT;
1877 buf[len] = 0;
1878 p = buf;
1879 if (*p == '-' && left > 1) {
1880 neg = 1;
1881 left--, p++;
1882 }
1883 if (*p < '0' || *p > '9')
1884 break;
1885
1886 lval = simple_strtoul(p, &p, 0);
1887
1888 len = p-buf;
1889 if ((len < left) && *p && !isspace(*p))
1890 break;
1891 if (neg)
1892 val = -val;
1893 s += len;
1894 left -= len;
1895
1896 if (conv(&neg, &lval, i, 1, data))
1897 break;
1898 } else {
1899 p = buf;
1900 if (!first)
1901 *p++ = '\t';
1902
1903 if (conv(&neg, &lval, i, 0, data))
1904 break;
1905
1906 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1907 len = strlen(buf);
1908 if (len > left)
1909 len = left;
1910 if(copy_to_user(s, buf, len))
1911 return -EFAULT;
1912 left -= len;
1913 s += len;
1914 }
1915 }
1916
1917 if (!write && !first && left) {
1918 if(put_user('\n', s))
1919 return -EFAULT;
1920 left--, s++;
1921 }
1922 if (write) {
1923 while (left) {
1924 char c;
1925 if (get_user(c, s++))
1926 return -EFAULT;
1927 if (!isspace(c))
1928 break;
1929 left--;
1930 }
1931 }
1932 if (write && first)
1933 return -EINVAL;
1934 *lenp -= left;
1935 *ppos += *lenp;
1936 return 0;
1937#undef TMPBUFLEN
1938}
1939
fcfbd547
KK		 1940static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1941 void __user *buffer, size_t *lenp, loff_t *ppos,
1942 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1943 int write, void *data),
1944 void *data)
1945{
1946 return __do_proc_dointvec(table->data, table, write, filp,
1947 buffer, lenp, ppos, conv, data);
1948}
1949
1da177e4
LT		 1950/**
1951 * proc_dointvec - read a vector of integers
1952 * @table: the sysctl table
1953 * @write: %TRUE if this is a write to the sysctl file
1954 * @filp: the file structure
1955 * @buffer: the user buffer
1956 * @lenp: the size of the user buffer
1957 * @ppos: file position
1958 *
1959 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1960 * values from/to the user buffer, treated as an ASCII string.
1961 *
1962 * Returns 0 on success.
1963 */
1964int proc_dointvec(ctl_table *table, int write, struct file *filp,
1965 void __user *buffer, size_t *lenp, loff_t *ppos)
1966{
1967 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1968 NULL,NULL);
1969}
1970
1971#define OP_SET 0
1972#define OP_AND 1
1973#define OP_OR 2
1974#define OP_MAX 3
1975#define OP_MIN 4
1976
1977static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1978 int *valp,
1979 int write, void *data)
1980{
1981 int op = *(int *)data;
1982 if (write) {
1983 int val = *negp ? -*lvalp : *lvalp;
1984 switch(op) {
1985 case OP_SET: *valp = val; break;
1986 case OP_AND: *valp &= val; break;
1987 case OP_OR: *valp |= val; break;
1988 case OP_MAX: if(*valp < val)
1989 *valp = val;
1990 break;
1991 case OP_MIN: if(*valp > val)
1992 *valp = val;
1993 break;
1994 }
1995 } else {
1996 int val = *valp;
1997 if (val < 0) {
1998 *negp = -1;
1999 *lvalp = (unsigned long)-val;
2000 } else {
2001 *negp = 0;
2002 *lvalp = (unsigned long)val;
2003 }
2004 }
2005 return 0;
2006}
2007
2008/*
2009 * init may raise the set.
2010 */
2011
2012int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2013 void __user *buffer, size_t *lenp, loff_t *ppos)
2014{
2015 int op;
2016
2017 if (!capable(CAP_SYS_MODULE)) {
2018 return -EPERM;
2019 }
2020
f400e198 2021 op = is_init(current) ? OP_SET : OP_AND;
1da177e4
LT		 2022 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2023 do_proc_dointvec_bset_conv,&op);
2024}
2025
2026struct do_proc_dointvec_minmax_conv_param {
2027 int *min;
2028 int *max;
2029};
2030
2031static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2032 int *valp,
2033 int write, void *data)
2034{
2035 struct do_proc_dointvec_minmax_conv_param *param = data;
2036 if (write) {
2037 int val = *negp ? -*lvalp : *lvalp;
2038 if ((param->min && *param->min > val) ||
2039 (param->max && *param->max < val))
2040 return -EINVAL;
2041 *valp = val;
2042 } else {
2043 int val = *valp;
2044 if (val < 0) {
2045 *negp = -1;
2046 *lvalp = (unsigned long)-val;
2047 } else {
2048 *negp = 0;
2049 *lvalp = (unsigned long)val;
2050 }
2051 }
2052 return 0;
2053}
2054
2055/**
2056 * proc_dointvec_minmax - read a vector of integers with min/max values
2057 * @table: the sysctl table
2058 * @write: %TRUE if this is a write to the sysctl file
2059 * @filp: the file structure
2060 * @buffer: the user buffer
2061 * @lenp: the size of the user buffer
2062 * @ppos: file position
2063 *
2064 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2065 * values from/to the user buffer, treated as an ASCII string.
2066 *
2067 * This routine will ensure the values are within the range specified by
2068 * table->extra1 (min) and table->extra2 (max).
2069 *
2070 * Returns 0 on success.
2071 */
2072int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2073 void __user *buffer, size_t *lenp, loff_t *ppos)
2074{
2075 struct do_proc_dointvec_minmax_conv_param param = {
2076 .min = (int *) table->extra1,
2077 .max = (int *) table->extra2,
2078 };
2079 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2080 do_proc_dointvec_minmax_conv, &param);
2081}
2082
fcfbd547 2083static int __do_proc_doulongvec_minmax(void *data, ctl_table *table, int write,
1da177e4
LT		 2084 struct file *filp,
2085 void __user *buffer,
2086 size_t *lenp, loff_t *ppos,
2087 unsigned long convmul,
2088 unsigned long convdiv)
2089{
2090#define TMPBUFLEN 21
2091 unsigned long *i, *min, *max, val;
2092 int vleft, first=1, neg;
2093 size_t len, left;
2094 char buf[TMPBUFLEN], *p;
2095 char __user *s = buffer;
2096
fcfbd547 2097 if (!data || !table->maxlen || !*lenp ||
1da177e4
LT		 2098 (*ppos && !write)) {
2099 *lenp = 0;
2100 return 0;
2101 }
2102
fcfbd547 2103 i = (unsigned long *) data;
1da177e4
LT		 2104 min = (unsigned long *) table->extra1;
2105 max = (unsigned long *) table->extra2;
2106 vleft = table->maxlen / sizeof(unsigned long);
2107 left = *lenp;
2108
2109 for (; left && vleft--; i++, min++, max++, first=0) {
2110 if (write) {
2111 while (left) {
2112 char c;
2113 if (get_user(c, s))
2114 return -EFAULT;
2115 if (!isspace(c))
2116 break;
2117 left--;
2118 s++;
2119 }
2120 if (!left)
2121 break;
2122 neg = 0;
2123 len = left;
2124 if (len > TMPBUFLEN-1)
2125 len = TMPBUFLEN-1;
2126 if (copy_from_user(buf, s, len))
2127 return -EFAULT;
2128 buf[len] = 0;
2129 p = buf;
2130 if (*p == '-' && left > 1) {
2131 neg = 1;
2132 left--, p++;
2133 }
2134 if (*p < '0' || *p > '9')
2135 break;
2136 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2137 len = p-buf;
2138 if ((len < left) && *p && !isspace(*p))
2139 break;
2140 if (neg)
2141 val = -val;
2142 s += len;
2143 left -= len;
2144
2145 if(neg)
2146 continue;
2147 if ((min && val < *min) || (max && val > *max))
2148 continue;
2149 *i = val;
2150 } else {
2151 p = buf;
2152 if (!first)
2153 *p++ = '\t';
2154 sprintf(p, "%lu", convdiv * (*i) / convmul);
2155 len = strlen(buf);
2156 if (len > left)
2157 len = left;
2158 if(copy_to_user(s, buf, len))
2159 return -EFAULT;
2160 left -= len;
2161 s += len;
2162 }
2163 }
2164
2165 if (!write && !first && left) {
2166 if(put_user('\n', s))
2167 return -EFAULT;
2168 left--, s++;
2169 }
2170 if (write) {
2171 while (left) {
2172 char c;
2173 if (get_user(c, s++))
2174 return -EFAULT;
2175 if (!isspace(c))
2176 break;
2177 left--;
2178 }
2179 }
2180 if (write && first)
2181 return -EINVAL;
2182 *lenp -= left;
2183 *ppos += *lenp;
2184 return 0;
2185#undef TMPBUFLEN
2186}
2187
fcfbd547
KK		 2188static int do_proc_doulongvec_minmax(ctl_table *table, int write,
2189 struct file *filp,
2190 void __user *buffer,
2191 size_t *lenp, loff_t *ppos,
2192 unsigned long convmul,
2193 unsigned long convdiv)
2194{
2195 return __do_proc_doulongvec_minmax(table->data, table, write,
2196 filp, buffer, lenp, ppos, convmul, convdiv);
2197}
2198
1da177e4
LT		 2199/**
2200 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2201 * @table: the sysctl table
2202 * @write: %TRUE if this is a write to the sysctl file
2203 * @filp: the file structure
2204 * @buffer: the user buffer
2205 * @lenp: the size of the user buffer
2206 * @ppos: file position
2207 *
2208 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2209 * values from/to the user buffer, treated as an ASCII string.
2210 *
2211 * This routine will ensure the values are within the range specified by
2212 * table->extra1 (min) and table->extra2 (max).
2213 *
2214 * Returns 0 on success.
2215 */
2216int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2217 void __user *buffer, size_t *lenp, loff_t *ppos)
2218{
2219 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2220}
2221
2222/**
2223 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2224 * @table: the sysctl table
2225 * @write: %TRUE if this is a write to the sysctl file
2226 * @filp: the file structure
2227 * @buffer: the user buffer
2228 * @lenp: the size of the user buffer
2229 * @ppos: file position
2230 *
2231 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2232 * values from/to the user buffer, treated as an ASCII string. The values
2233 * are treated as milliseconds, and converted to jiffies when they are stored.
2234 *
2235 * This routine will ensure the values are within the range specified by
2236 * table->extra1 (min) and table->extra2 (max).
2237 *
2238 * Returns 0 on success.
2239 */
2240int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2241 struct file *filp,
2242 void __user *buffer,
2243 size_t *lenp, loff_t *ppos)
2244{
2245 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2246 lenp, ppos, HZ, 1000l);
2247}
2248
2249
2250static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2251 int *valp,
2252 int write, void *data)
2253{
2254 if (write) {
cba9f33d
BS		 2255 if (*lvalp > LONG_MAX / HZ)
2256 return 1;
1da177e4
LT		 2257 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2258 } else {
2259 int val = *valp;
2260 unsigned long lval;
2261 if (val < 0) {
2262 *negp = -1;
2263 lval = (unsigned long)-val;
2264 } else {
2265 *negp = 0;
2266 lval = (unsigned long)val;
2267 }
2268 *lvalp = lval / HZ;
2269 }
2270 return 0;
2271}
2272
2273static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2274 int *valp,
2275 int write, void *data)
2276{
2277 if (write) {
cba9f33d
BS		 2278 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2279 return 1;
1da177e4
LT		 2280 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2281 } else {
2282 int val = *valp;
2283 unsigned long lval;
2284 if (val < 0) {
2285 *negp = -1;
2286 lval = (unsigned long)-val;
2287 } else {
2288 *negp = 0;
2289 lval = (unsigned long)val;
2290 }
2291 *lvalp = jiffies_to_clock_t(lval);
2292 }
2293 return 0;
2294}
2295
2296static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2297 int *valp,
2298 int write, void *data)
2299{
2300 if (write) {
2301 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2302 } else {
2303 int val = *valp;
2304 unsigned long lval;
2305 if (val < 0) {
2306 *negp = -1;
2307 lval = (unsigned long)-val;
2308 } else {
2309 *negp = 0;
2310 lval = (unsigned long)val;
2311 }
2312 *lvalp = jiffies_to_msecs(lval);
2313 }
2314 return 0;
2315}
2316
2317/**
2318 * proc_dointvec_jiffies - read a vector of integers as seconds
2319 * @table: the sysctl table
2320 * @write: %TRUE if this is a write to the sysctl file
2321 * @filp: the file structure
2322 * @buffer: the user buffer
2323 * @lenp: the size of the user buffer
2324 * @ppos: file position
2325 *
2326 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2327 * values from/to the user buffer, treated as an ASCII string.
2328 * The values read are assumed to be in seconds, and are converted into
2329 * jiffies.
2330 *
2331 * Returns 0 on success.
2332 */
2333int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2334 void __user *buffer, size_t *lenp, loff_t *ppos)
2335{
2336 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2337 do_proc_dointvec_jiffies_conv,NULL);
2338}
2339
2340/**
2341 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2342 * @table: the sysctl table
2343 * @write: %TRUE if this is a write to the sysctl file
2344 * @filp: the file structure
2345 * @buffer: the user buffer
2346 * @lenp: the size of the user buffer
1e5d5331 2347 * @ppos: pointer to the file position
1da177e4
LT		 2348 *
2349 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2350 * values from/to the user buffer, treated as an ASCII string.
2351 * The values read are assumed to be in 1/USER_HZ seconds, and
2352 * are converted into jiffies.
2353 *
2354 * Returns 0 on success.
2355 */
2356int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2357 void __user *buffer, size_t *lenp, loff_t *ppos)
2358{
2359 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2360 do_proc_dointvec_userhz_jiffies_conv,NULL);
2361}
2362
2363/**
2364 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2365 * @table: the sysctl table
2366 * @write: %TRUE if this is a write to the sysctl file
2367 * @filp: the file structure
2368 * @buffer: the user buffer
2369 * @lenp: the size of the user buffer
67be2dd1
MW		 2370 * @ppos: file position
2371 * @ppos: the current position in the file
1da177e4
LT		 2372 *
2373 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2374 * values from/to the user buffer, treated as an ASCII string.
2375 * The values read are assumed to be in 1/1000 seconds, and
2376 * are converted into jiffies.
2377 *
2378 * Returns 0 on success.
2379 */
2380int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2381 void __user *buffer, size_t *lenp, loff_t *ppos)
2382{
2383 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2384 do_proc_dointvec_ms_jiffies_conv, NULL);
2385}
2386
fcfbd547
KK		 2387#ifdef CONFIG_SYSVIPC
2388static int proc_do_ipc_string(ctl_table *table, int write, struct file *filp,
2389 void __user *buffer, size_t *lenp, loff_t *ppos)
2390{
2391 void *data;
2392 struct ipc_namespace *ns;
2393
2394 ns = current->nsproxy->ipc_ns;
2395
2396 switch (table->ctl_name) {
2397 case KERN_SHMMAX:
2398 data = &ns->shm_ctlmax;
2399 goto proc_minmax;
2400 case KERN_SHMALL:
2401 data = &ns->shm_ctlall;
2402 goto proc_minmax;
2403 case KERN_SHMMNI:
2404 data = &ns->shm_ctlmni;
2405 break;
2406 case KERN_MSGMAX:
2407 data = &ns->msg_ctlmax;
2408 break;
2409 case KERN_MSGMNI:
2410 data = &ns->msg_ctlmni;
2411 break;
2412 case KERN_MSGMNB:
2413 data = &ns->msg_ctlmnb;
2414 break;
2415 case KERN_SEM:
2416 data = &ns->sem_ctls;
2417 break;
2418 default:
2419 return -EINVAL;
2420 }
2421
2422 return __do_proc_dointvec(data, table, write, filp, buffer,
2423 lenp, ppos, NULL, NULL);
2424proc_minmax:
2425 return __do_proc_doulongvec_minmax(data, table, write, filp, buffer,
2426 lenp, ppos, 1l, 1l);
2427}
2428#endif
2429
1da177e4
LT		 2430#else /* CONFIG_PROC_FS */
2431
2432int proc_dostring(ctl_table *table, int write, struct file *filp,
2433 void __user *buffer, size_t *lenp, loff_t *ppos)
2434{
2435 return -ENOSYS;
2436}
2437
8218c74c
SH		 2438static int proc_do_uts_string(ctl_table *table, int write, struct file *filp,
2439 void __user *buffer, size_t *lenp, loff_t *ppos)
1da177e4
LT		 2440{
2441 return -ENOSYS;
2442}
2443
8218c74c
SH		 2444#ifdef CONFIG_SYSVIPC
2445static int proc_do_ipc_string(ctl_table *table, int write, struct file *filp,
2446 void __user *buffer, size_t *lenp, loff_t *ppos)
2447{
2448 return -ENOSYS;
2449}
2450#endif
2451
1da177e4
LT		 2452int proc_dointvec(ctl_table *table, int write, struct file *filp,
2453 void __user *buffer, size_t *lenp, loff_t *ppos)
2454{
2455 return -ENOSYS;
2456}
2457
2458int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2459 void __user *buffer, size_t *lenp, loff_t *ppos)
2460{
2461 return -ENOSYS;
2462}
2463
2464int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2465 void __user *buffer, size_t *lenp, loff_t *ppos)
2466{
2467 return -ENOSYS;
2468}
2469
2470int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2471 void __user *buffer, size_t *lenp, loff_t *ppos)
2472{
2473 return -ENOSYS;
2474}
2475
2476int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2477 void __user *buffer, size_t *lenp, loff_t *ppos)
2478{
2479 return -ENOSYS;
2480}
2481
2482int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2483 void __user *buffer, size_t *lenp, loff_t *ppos)
2484{
2485 return -ENOSYS;
2486}
2487
2488int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2489 void __user *buffer, size_t *lenp, loff_t *ppos)
2490{
2491 return -ENOSYS;
2492}
2493
2494int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2495 struct file *filp,
2496 void __user *buffer,
2497 size_t *lenp, loff_t *ppos)
2498{
2499 return -ENOSYS;
2500}
2501
2502
2503#endif /* CONFIG_PROC_FS */
2504
2505
b89a8171 2506#ifdef CONFIG_SYSCTL_SYSCALL
1da177e4
LT		 2507/*
2508 * General sysctl support routines
2509 */
2510
2511/* The generic string strategy routine: */
2512int sysctl_string(ctl_table *table, int __user *name, int nlen,
2513 void __user *oldval, size_t __user *oldlenp,
2514 void __user *newval, size_t newlen, void **context)
2515{
1da177e4
LT		 2516 if (!table->data || !table->maxlen)
2517 return -ENOTDIR;
2518
2519 if (oldval && oldlenp) {
de9e007d
LT		 2520 size_t bufsize;
2521 if (get_user(bufsize, oldlenp))
1da177e4 2522 return -EFAULT;
de9e007d
LT		 2523 if (bufsize) {
2524 size_t len = strlen(table->data), copied;
2525
2526 /* This shouldn't trigger for a well-formed sysctl */
2527 if (len > table->maxlen)
1da177e4 2528 len = table->maxlen;
de9e007d
LT		 2529
2530 /* Copy up to a max of bufsize-1 bytes of the string */
2531 copied = (len >= bufsize) ? bufsize - 1 : len;
2532
2533 if (copy_to_user(oldval, table->data, copied) ||
2534 put_user(0, (char __user *)(oldval + copied)))
1da177e4 2535 return -EFAULT;
de9e007d 2536 if (put_user(len, oldlenp))
1da177e4
LT		 2537 return -EFAULT;
2538 }
2539 }
2540 if (newval && newlen) {
de9e007d 2541 size_t len = newlen;
1da177e4
LT		 2542 if (len > table->maxlen)
2543 len = table->maxlen;
2544 if(copy_from_user(table->data, newval, len))
2545 return -EFAULT;
2546 if (len == table->maxlen)
2547 len--;
2548 ((char *) table->data)[len] = 0;
2549 }
82c9df82 2550 return 1;
1da177e4
LT		 2551}
2552
2553/*
2554 * This function makes sure that all of the integers in the vector
2555 * are between the minimum and maximum values given in the arrays
2556 * table->extra1 and table->extra2, respectively.
2557 */
2558int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2559 void __user *oldval, size_t __user *oldlenp,
2560 void __user *newval, size_t newlen, void **context)
2561{
2562
2563 if (newval && newlen) {
2564 int __user *vec = (int __user *) newval;
2565 int *min = (int *) table->extra1;
2566 int *max = (int *) table->extra2;
2567 size_t length;
2568 int i;
2569
2570 if (newlen % sizeof(int) != 0)
2571 return -EINVAL;
2572
2573 if (!table->extra1 && !table->extra2)
2574 return 0;
2575
2576 if (newlen > table->maxlen)
2577 newlen = table->maxlen;
2578 length = newlen / sizeof(int);
2579
2580 for (i = 0; i < length; i++) {
2581 int value;
2582 if (get_user(value, vec + i))
2583 return -EFAULT;
2584 if (min && value < min[i])
2585 return -EINVAL;
2586 if (max && value > max[i])
2587 return -EINVAL;
2588 }
2589 }
2590 return 0;
2591}
2592
2593/* Strategy function to convert jiffies to seconds */
2594int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2595 void __user *oldval, size_t __user *oldlenp,
2596 void __user *newval, size_t newlen, void **context)
2597{
2598 if (oldval) {
2599 size_t olen;
2600 if (oldlenp) {
2601 if (get_user(olen, oldlenp))
2602 return -EFAULT;
2603 if (olen!=sizeof(int))
2604 return -EINVAL;
2605 }
2606 if (put_user(*(int *)(table->data)/HZ, (int __user *)oldval) ||
2607 (oldlenp && put_user(sizeof(int),oldlenp)))
2608 return -EFAULT;
2609 }
2610 if (newval && newlen) {
2611 int new;
2612 if (newlen != sizeof(int))
2613 return -EINVAL;
2614 if (get_user(new, (int __user *)newval))
2615 return -EFAULT;
2616 *(int *)(table->data) = new*HZ;
2617 }
2618 return 1;
2619}
2620
2621/* Strategy function to convert jiffies to seconds */
2622int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2623 void __user *oldval, size_t __user *oldlenp,
2624 void __user *newval, size_t newlen, void **context)
2625{
2626 if (oldval) {
2627 size_t olen;
2628 if (oldlenp) {
2629 if (get_user(olen, oldlenp))
2630 return -EFAULT;
2631 if (olen!=sizeof(int))
2632 return -EINVAL;
2633 }
2634 if (put_user(jiffies_to_msecs(*(int *)(table->data)), (int __user *)oldval) ||
2635 (oldlenp && put_user(sizeof(int),oldlenp)))
2636 return -EFAULT;
2637 }
2638 if (newval && newlen) {
2639 int new;
2640 if (newlen != sizeof(int))
2641 return -EINVAL;
2642 if (get_user(new, (int __user *)newval))
2643 return -EFAULT;
2644 *(int *)(table->data) = msecs_to_jiffies(new);
2645 }
2646 return 1;
2647}
2648
b89a8171 2649#else /* CONFIG_SYSCTL_SYSCALL */
1da177e4
LT		 2650
2651
2652asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2653{
b89a8171
EB		 2654 static int msg_count;
2655
2656 if (msg_count < 5) {
2657 msg_count++;
2658 printk(KERN_INFO
2659 "warning: process `%s' used the removed sysctl "
2660 "system call\n", current->comm);
2661 }
1da177e4
LT		 2662 return -ENOSYS;
2663}
2664
2665int sysctl_string(ctl_table *table, int __user *name, int nlen,
2666 void __user *oldval, size_t __user *oldlenp,
2667 void __user *newval, size_t newlen, void **context)
2668{
2669 return -ENOSYS;
2670}
2671
2672int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2673 void __user *oldval, size_t __user *oldlenp,
2674 void __user *newval, size_t newlen, void **context)
2675{
2676 return -ENOSYS;
2677}
2678
2679int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2680 void __user *oldval, size_t __user *oldlenp,
2681 void __user *newval, size_t newlen, void **context)
2682{
2683 return -ENOSYS;
2684}
2685
2686int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2687 void __user *oldval, size_t __user *oldlenp,
2688 void __user *newval, size_t newlen, void **context)
2689{
2690 return -ENOSYS;
2691}
2692
b89a8171 2693#endif /* CONFIG_SYSCTL_SYSCALL */
1da177e4
LT		 2694
2695/*
2696 * No sense putting this after each symbol definition, twice,
2697 * exception granted :-)
2698 */
2699EXPORT_SYMBOL(proc_dointvec);
2700EXPORT_SYMBOL(proc_dointvec_jiffies);
2701EXPORT_SYMBOL(proc_dointvec_minmax);
2702EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2703EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2704EXPORT_SYMBOL(proc_dostring);
2705EXPORT_SYMBOL(proc_doulongvec_minmax);
2706EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2707EXPORT_SYMBOL(register_sysctl_table);
2708EXPORT_SYMBOL(sysctl_intvec);
2709EXPORT_SYMBOL(sysctl_jiffies);
2710EXPORT_SYMBOL(sysctl_ms_jiffies);
2711EXPORT_SYMBOL(sysctl_string);
2712EXPORT_SYMBOL(unregister_sysctl_table);
ubuntu-zesty-kernel mirror