Loading...
1// SPDX-License-Identifier: GPL-2.0-only
2/*
3 * sysctl.c: General linux system control interface
4 *
5 * Begun 24 March 1995, Stephen Tweedie
6 * Added /proc support, Dec 1995
7 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
8 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
9 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
10 * Dynamic registration fixes, Stephen Tweedie.
11 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
12 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
13 * Horn.
14 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
15 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
16 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
17 * Wendling.
18 * The list_for_each() macro wasn't appropriate for the sysctl loop.
19 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
20 */
21
22#include <linux/module.h>
23#include <linux/aio.h>
24#include <linux/mm.h>
25#include <linux/swap.h>
26#include <linux/slab.h>
27#include <linux/sysctl.h>
28#include <linux/bitmap.h>
29#include <linux/signal.h>
30#include <linux/panic.h>
31#include <linux/printk.h>
32#include <linux/proc_fs.h>
33#include <linux/security.h>
34#include <linux/ctype.h>
35#include <linux/kmemleak.h>
36#include <linux/fs.h>
37#include <linux/init.h>
38#include <linux/kernel.h>
39#include <linux/kobject.h>
40#include <linux/net.h>
41#include <linux/sysrq.h>
42#include <linux/highuid.h>
43#include <linux/writeback.h>
44#include <linux/ratelimit.h>
45#include <linux/compaction.h>
46#include <linux/hugetlb.h>
47#include <linux/initrd.h>
48#include <linux/key.h>
49#include <linux/times.h>
50#include <linux/limits.h>
51#include <linux/dcache.h>
52#include <linux/dnotify.h>
53#include <linux/syscalls.h>
54#include <linux/vmstat.h>
55#include <linux/nfs_fs.h>
56#include <linux/acpi.h>
57#include <linux/reboot.h>
58#include <linux/ftrace.h>
59#include <linux/perf_event.h>
60#include <linux/kprobes.h>
61#include <linux/pipe_fs_i.h>
62#include <linux/oom.h>
63#include <linux/kmod.h>
64#include <linux/capability.h>
65#include <linux/binfmts.h>
66#include <linux/sched/sysctl.h>
67#include <linux/sched/coredump.h>
68#include <linux/kexec.h>
69#include <linux/bpf.h>
70#include <linux/mount.h>
71#include <linux/userfaultfd_k.h>
72#include <linux/coredump.h>
73#include <linux/latencytop.h>
74#include <linux/pid.h>
75#include <linux/delayacct.h>
76
77#include "../lib/kstrtox.h"
78
79#include <linux/uaccess.h>
80#include <asm/processor.h>
81
82#ifdef CONFIG_X86
83#include <asm/nmi.h>
84#include <asm/stacktrace.h>
85#include <asm/io.h>
86#endif
87#ifdef CONFIG_SPARC
88#include <asm/setup.h>
89#endif
90#ifdef CONFIG_BSD_PROCESS_ACCT
91#include <linux/acct.h>
92#endif
93#ifdef CONFIG_RT_MUTEXES
94#include <linux/rtmutex.h>
95#endif
96#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
97#include <linux/lockdep.h>
98#endif
99#ifdef CONFIG_CHR_DEV_SG
100#include <scsi/sg.h>
101#endif
102#ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
103#include <linux/stackleak.h>
104#endif
105#ifdef CONFIG_LOCKUP_DETECTOR
106#include <linux/nmi.h>
107#endif
108
109#if defined(CONFIG_SYSCTL)
110
111/* Constants used for minimum and maximum */
112#ifdef CONFIG_LOCKUP_DETECTOR
113static int sixty = 60;
114#endif
115
116static int __maybe_unused neg_one = -1;
117static int __maybe_unused two = 2;
118static int __maybe_unused four = 4;
119static unsigned long zero_ul;
120static unsigned long one_ul = 1;
121static unsigned long long_max = LONG_MAX;
122static int one_hundred = 100;
123static int two_hundred = 200;
124static int one_thousand = 1000;
125#ifdef CONFIG_PRINTK
126static int ten_thousand = 10000;
127#endif
128#ifdef CONFIG_PERF_EVENTS
129static int six_hundred_forty_kb = 640 * 1024;
130#endif
131
132/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
133static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
134
135/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
136static int maxolduid = 65535;
137static int minolduid;
138
139static int ngroups_max = NGROUPS_MAX;
140static const int cap_last_cap = CAP_LAST_CAP;
141
142/*
143 * This is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs
144 * and hung_task_check_interval_secs
145 */
146#ifdef CONFIG_DETECT_HUNG_TASK
147static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
148#endif
149
150#ifdef CONFIG_INOTIFY_USER
151#include <linux/inotify.h>
152#endif
153#ifdef CONFIG_FANOTIFY
154#include <linux/fanotify.h>
155#endif
156
157#ifdef CONFIG_PROC_SYSCTL
158
159/**
160 * enum sysctl_writes_mode - supported sysctl write modes
161 *
162 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
163 * to be written, and multiple writes on the same sysctl file descriptor
164 * will rewrite the sysctl value, regardless of file position. No warning
165 * is issued when the initial position is not 0.
166 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
167 * not 0.
168 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
169 * file position 0 and the value must be fully contained in the buffer
170 * sent to the write syscall. If dealing with strings respect the file
171 * position, but restrict this to the max length of the buffer, anything
172 * passed the max length will be ignored. Multiple writes will append
173 * to the buffer.
174 *
175 * These write modes control how current file position affects the behavior of
176 * updating sysctl values through the proc interface on each write.
177 */
178enum sysctl_writes_mode {
179 SYSCTL_WRITES_LEGACY = -1,
180 SYSCTL_WRITES_WARN = 0,
181 SYSCTL_WRITES_STRICT = 1,
182};
183
184static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
185#endif /* CONFIG_PROC_SYSCTL */
186
187#if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
188 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
189int sysctl_legacy_va_layout;
190#endif
191
192#ifdef CONFIG_COMPACTION
193static int min_extfrag_threshold;
194static int max_extfrag_threshold = 1000;
195#endif
196
197#endif /* CONFIG_SYSCTL */
198
199#if defined(CONFIG_BPF_SYSCALL) && defined(CONFIG_SYSCTL)
200static int bpf_stats_handler(struct ctl_table *table, int write,
201 void *buffer, size_t *lenp, loff_t *ppos)
202{
203 struct static_key *key = (struct static_key *)table->data;
204 static int saved_val;
205 int val, ret;
206 struct ctl_table tmp = {
207 .data = &val,
208 .maxlen = sizeof(val),
209 .mode = table->mode,
210 .extra1 = SYSCTL_ZERO,
211 .extra2 = SYSCTL_ONE,
212 };
213
214 if (write && !capable(CAP_SYS_ADMIN))
215 return -EPERM;
216
217 mutex_lock(&bpf_stats_enabled_mutex);
218 val = saved_val;
219 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
220 if (write && !ret && val != saved_val) {
221 if (val)
222 static_key_slow_inc(key);
223 else
224 static_key_slow_dec(key);
225 saved_val = val;
226 }
227 mutex_unlock(&bpf_stats_enabled_mutex);
228 return ret;
229}
230
231static int bpf_unpriv_handler(struct ctl_table *table, int write,
232 void *buffer, size_t *lenp, loff_t *ppos)
233{
234 int ret, unpriv_enable = *(int *)table->data;
235 bool locked_state = unpriv_enable == 1;
236 struct ctl_table tmp = *table;
237
238 if (write && !capable(CAP_SYS_ADMIN))
239 return -EPERM;
240
241 tmp.data = &unpriv_enable;
242 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
243 if (write && !ret) {
244 if (locked_state && unpriv_enable != 1)
245 return -EPERM;
246 *(int *)table->data = unpriv_enable;
247 }
248 return ret;
249}
250#endif /* CONFIG_BPF_SYSCALL && CONFIG_SYSCTL */
251
252/*
253 * /proc/sys support
254 */
255
256#ifdef CONFIG_PROC_SYSCTL
257
258static int _proc_do_string(char *data, int maxlen, int write,
259 char *buffer, size_t *lenp, loff_t *ppos)
260{
261 size_t len;
262 char c, *p;
263
264 if (!data || !maxlen || !*lenp) {
265 *lenp = 0;
266 return 0;
267 }
268
269 if (write) {
270 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
271 /* Only continue writes not past the end of buffer. */
272 len = strlen(data);
273 if (len > maxlen - 1)
274 len = maxlen - 1;
275
276 if (*ppos > len)
277 return 0;
278 len = *ppos;
279 } else {
280 /* Start writing from beginning of buffer. */
281 len = 0;
282 }
283
284 *ppos += *lenp;
285 p = buffer;
286 while ((p - buffer) < *lenp && len < maxlen - 1) {
287 c = *(p++);
288 if (c == 0 || c == '\n')
289 break;
290 data[len++] = c;
291 }
292 data[len] = 0;
293 } else {
294 len = strlen(data);
295 if (len > maxlen)
296 len = maxlen;
297
298 if (*ppos > len) {
299 *lenp = 0;
300 return 0;
301 }
302
303 data += *ppos;
304 len -= *ppos;
305
306 if (len > *lenp)
307 len = *lenp;
308 if (len)
309 memcpy(buffer, data, len);
310 if (len < *lenp) {
311 buffer[len] = '\n';
312 len++;
313 }
314 *lenp = len;
315 *ppos += len;
316 }
317 return 0;
318}
319
320static void warn_sysctl_write(struct ctl_table *table)
321{
322 pr_warn_once("%s wrote to %s when file position was not 0!\n"
323 "This will not be supported in the future. To silence this\n"
324 "warning, set kernel.sysctl_writes_strict = -1\n",
325 current->comm, table->procname);
326}
327
328/**
329 * proc_first_pos_non_zero_ignore - check if first position is allowed
330 * @ppos: file position
331 * @table: the sysctl table
332 *
333 * Returns true if the first position is non-zero and the sysctl_writes_strict
334 * mode indicates this is not allowed for numeric input types. String proc
335 * handlers can ignore the return value.
336 */
337static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
338 struct ctl_table *table)
339{
340 if (!*ppos)
341 return false;
342
343 switch (sysctl_writes_strict) {
344 case SYSCTL_WRITES_STRICT:
345 return true;
346 case SYSCTL_WRITES_WARN:
347 warn_sysctl_write(table);
348 return false;
349 default:
350 return false;
351 }
352}
353
354/**
355 * proc_dostring - read a string sysctl
356 * @table: the sysctl table
357 * @write: %TRUE if this is a write to the sysctl file
358 * @buffer: the user buffer
359 * @lenp: the size of the user buffer
360 * @ppos: file position
361 *
362 * Reads/writes a string from/to the user buffer. If the kernel
363 * buffer provided is not large enough to hold the string, the
364 * string is truncated. The copied string is %NULL-terminated.
365 * If the string is being read by the user process, it is copied
366 * and a newline '\n' is added. It is truncated if the buffer is
367 * not large enough.
368 *
369 * Returns 0 on success.
370 */
371int proc_dostring(struct ctl_table *table, int write,
372 void *buffer, size_t *lenp, loff_t *ppos)
373{
374 if (write)
375 proc_first_pos_non_zero_ignore(ppos, table);
376
377 return _proc_do_string(table->data, table->maxlen, write, buffer, lenp,
378 ppos);
379}
380
381static size_t proc_skip_spaces(char **buf)
382{
383 size_t ret;
384 char *tmp = skip_spaces(*buf);
385 ret = tmp - *buf;
386 *buf = tmp;
387 return ret;
388}
389
390static void proc_skip_char(char **buf, size_t *size, const char v)
391{
392 while (*size) {
393 if (**buf != v)
394 break;
395 (*size)--;
396 (*buf)++;
397 }
398}
399
400/**
401 * strtoul_lenient - parse an ASCII formatted integer from a buffer and only
402 * fail on overflow
403 *
404 * @cp: kernel buffer containing the string to parse
405 * @endp: pointer to store the trailing characters
406 * @base: the base to use
407 * @res: where the parsed integer will be stored
408 *
409 * In case of success 0 is returned and @res will contain the parsed integer,
410 * @endp will hold any trailing characters.
411 * This function will fail the parse on overflow. If there wasn't an overflow
412 * the function will defer the decision what characters count as invalid to the
413 * caller.
414 */
415static int strtoul_lenient(const char *cp, char **endp, unsigned int base,
416 unsigned long *res)
417{
418 unsigned long long result;
419 unsigned int rv;
420
421 cp = _parse_integer_fixup_radix(cp, &base);
422 rv = _parse_integer(cp, base, &result);
423 if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result))
424 return -ERANGE;
425
426 cp += rv;
427
428 if (endp)
429 *endp = (char *)cp;
430
431 *res = (unsigned long)result;
432 return 0;
433}
434
435#define TMPBUFLEN 22
436/**
437 * proc_get_long - reads an ASCII formatted integer from a user buffer
438 *
439 * @buf: a kernel buffer
440 * @size: size of the kernel buffer
441 * @val: this is where the number will be stored
442 * @neg: set to %TRUE if number is negative
443 * @perm_tr: a vector which contains the allowed trailers
444 * @perm_tr_len: size of the perm_tr vector
445 * @tr: pointer to store the trailer character
446 *
447 * In case of success %0 is returned and @buf and @size are updated with
448 * the amount of bytes read. If @tr is non-NULL and a trailing
449 * character exists (size is non-zero after returning from this
450 * function), @tr is updated with the trailing character.
451 */
452static int proc_get_long(char **buf, size_t *size,
453 unsigned long *val, bool *neg,
454 const char *perm_tr, unsigned perm_tr_len, char *tr)
455{
456 int len;
457 char *p, tmp[TMPBUFLEN];
458
459 if (!*size)
460 return -EINVAL;
461
462 len = *size;
463 if (len > TMPBUFLEN - 1)
464 len = TMPBUFLEN - 1;
465
466 memcpy(tmp, *buf, len);
467
468 tmp[len] = 0;
469 p = tmp;
470 if (*p == '-' && *size > 1) {
471 *neg = true;
472 p++;
473 } else
474 *neg = false;
475 if (!isdigit(*p))
476 return -EINVAL;
477
478 if (strtoul_lenient(p, &p, 0, val))
479 return -EINVAL;
480
481 len = p - tmp;
482
483 /* We don't know if the next char is whitespace thus we may accept
484 * invalid integers (e.g. 1234...a) or two integers instead of one
485 * (e.g. 123...1). So lets not allow such large numbers. */
486 if (len == TMPBUFLEN - 1)
487 return -EINVAL;
488
489 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
490 return -EINVAL;
491
492 if (tr && (len < *size))
493 *tr = *p;
494
495 *buf += len;
496 *size -= len;
497
498 return 0;
499}
500
501/**
502 * proc_put_long - converts an integer to a decimal ASCII formatted string
503 *
504 * @buf: the user buffer
505 * @size: the size of the user buffer
506 * @val: the integer to be converted
507 * @neg: sign of the number, %TRUE for negative
508 *
509 * In case of success @buf and @size are updated with the amount of bytes
510 * written.
511 */
512static void proc_put_long(void **buf, size_t *size, unsigned long val, bool neg)
513{
514 int len;
515 char tmp[TMPBUFLEN], *p = tmp;
516
517 sprintf(p, "%s%lu", neg ? "-" : "", val);
518 len = strlen(tmp);
519 if (len > *size)
520 len = *size;
521 memcpy(*buf, tmp, len);
522 *size -= len;
523 *buf += len;
524}
525#undef TMPBUFLEN
526
527static void proc_put_char(void **buf, size_t *size, char c)
528{
529 if (*size) {
530 char **buffer = (char **)buf;
531 **buffer = c;
532
533 (*size)--;
534 (*buffer)++;
535 *buf = *buffer;
536 }
537}
538
539static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
540 int *valp,
541 int write, void *data)
542{
543 if (write) {
544 if (*negp) {
545 if (*lvalp > (unsigned long) INT_MAX + 1)
546 return -EINVAL;
547 *valp = -*lvalp;
548 } else {
549 if (*lvalp > (unsigned long) INT_MAX)
550 return -EINVAL;
551 *valp = *lvalp;
552 }
553 } else {
554 int val = *valp;
555 if (val < 0) {
556 *negp = true;
557 *lvalp = -(unsigned long)val;
558 } else {
559 *negp = false;
560 *lvalp = (unsigned long)val;
561 }
562 }
563 return 0;
564}
565
566static int do_proc_douintvec_conv(unsigned long *lvalp,
567 unsigned int *valp,
568 int write, void *data)
569{
570 if (write) {
571 if (*lvalp > UINT_MAX)
572 return -EINVAL;
573 *valp = *lvalp;
574 } else {
575 unsigned int val = *valp;
576 *lvalp = (unsigned long)val;
577 }
578 return 0;
579}
580
581static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
582
583static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
584 int write, void *buffer,
585 size_t *lenp, loff_t *ppos,
586 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
587 int write, void *data),
588 void *data)
589{
590 int *i, vleft, first = 1, err = 0;
591 size_t left;
592 char *p;
593
594 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
595 *lenp = 0;
596 return 0;
597 }
598
599 i = (int *) tbl_data;
600 vleft = table->maxlen / sizeof(*i);
601 left = *lenp;
602
603 if (!conv)
604 conv = do_proc_dointvec_conv;
605
606 if (write) {
607 if (proc_first_pos_non_zero_ignore(ppos, table))
608 goto out;
609
610 if (left > PAGE_SIZE - 1)
611 left = PAGE_SIZE - 1;
612 p = buffer;
613 }
614
615 for (; left && vleft--; i++, first=0) {
616 unsigned long lval;
617 bool neg;
618
619 if (write) {
620 left -= proc_skip_spaces(&p);
621
622 if (!left)
623 break;
624 err = proc_get_long(&p, &left, &lval, &neg,
625 proc_wspace_sep,
626 sizeof(proc_wspace_sep), NULL);
627 if (err)
628 break;
629 if (conv(&neg, &lval, i, 1, data)) {
630 err = -EINVAL;
631 break;
632 }
633 } else {
634 if (conv(&neg, &lval, i, 0, data)) {
635 err = -EINVAL;
636 break;
637 }
638 if (!first)
639 proc_put_char(&buffer, &left, '\t');
640 proc_put_long(&buffer, &left, lval, neg);
641 }
642 }
643
644 if (!write && !first && left && !err)
645 proc_put_char(&buffer, &left, '\n');
646 if (write && !err && left)
647 left -= proc_skip_spaces(&p);
648 if (write && first)
649 return err ? : -EINVAL;
650 *lenp -= left;
651out:
652 *ppos += *lenp;
653 return err;
654}
655
656static int do_proc_dointvec(struct ctl_table *table, int write,
657 void *buffer, size_t *lenp, loff_t *ppos,
658 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
659 int write, void *data),
660 void *data)
661{
662 return __do_proc_dointvec(table->data, table, write,
663 buffer, lenp, ppos, conv, data);
664}
665
666static int do_proc_douintvec_w(unsigned int *tbl_data,
667 struct ctl_table *table,
668 void *buffer,
669 size_t *lenp, loff_t *ppos,
670 int (*conv)(unsigned long *lvalp,
671 unsigned int *valp,
672 int write, void *data),
673 void *data)
674{
675 unsigned long lval;
676 int err = 0;
677 size_t left;
678 bool neg;
679 char *p = buffer;
680
681 left = *lenp;
682
683 if (proc_first_pos_non_zero_ignore(ppos, table))
684 goto bail_early;
685
686 if (left > PAGE_SIZE - 1)
687 left = PAGE_SIZE - 1;
688
689 left -= proc_skip_spaces(&p);
690 if (!left) {
691 err = -EINVAL;
692 goto out_free;
693 }
694
695 err = proc_get_long(&p, &left, &lval, &neg,
696 proc_wspace_sep,
697 sizeof(proc_wspace_sep), NULL);
698 if (err || neg) {
699 err = -EINVAL;
700 goto out_free;
701 }
702
703 if (conv(&lval, tbl_data, 1, data)) {
704 err = -EINVAL;
705 goto out_free;
706 }
707
708 if (!err && left)
709 left -= proc_skip_spaces(&p);
710
711out_free:
712 if (err)
713 return -EINVAL;
714
715 return 0;
716
717 /* This is in keeping with old __do_proc_dointvec() */
718bail_early:
719 *ppos += *lenp;
720 return err;
721}
722
723static int do_proc_douintvec_r(unsigned int *tbl_data, void *buffer,
724 size_t *lenp, loff_t *ppos,
725 int (*conv)(unsigned long *lvalp,
726 unsigned int *valp,
727 int write, void *data),
728 void *data)
729{
730 unsigned long lval;
731 int err = 0;
732 size_t left;
733
734 left = *lenp;
735
736 if (conv(&lval, tbl_data, 0, data)) {
737 err = -EINVAL;
738 goto out;
739 }
740
741 proc_put_long(&buffer, &left, lval, false);
742 if (!left)
743 goto out;
744
745 proc_put_char(&buffer, &left, '\n');
746
747out:
748 *lenp -= left;
749 *ppos += *lenp;
750
751 return err;
752}
753
754static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
755 int write, void *buffer,
756 size_t *lenp, loff_t *ppos,
757 int (*conv)(unsigned long *lvalp,
758 unsigned int *valp,
759 int write, void *data),
760 void *data)
761{
762 unsigned int *i, vleft;
763
764 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
765 *lenp = 0;
766 return 0;
767 }
768
769 i = (unsigned int *) tbl_data;
770 vleft = table->maxlen / sizeof(*i);
771
772 /*
773 * Arrays are not supported, keep this simple. *Do not* add
774 * support for them.
775 */
776 if (vleft != 1) {
777 *lenp = 0;
778 return -EINVAL;
779 }
780
781 if (!conv)
782 conv = do_proc_douintvec_conv;
783
784 if (write)
785 return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
786 conv, data);
787 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
788}
789
790static int do_proc_douintvec(struct ctl_table *table, int write,
791 void *buffer, size_t *lenp, loff_t *ppos,
792 int (*conv)(unsigned long *lvalp,
793 unsigned int *valp,
794 int write, void *data),
795 void *data)
796{
797 return __do_proc_douintvec(table->data, table, write,
798 buffer, lenp, ppos, conv, data);
799}
800
801/**
802 * proc_dointvec - read a vector of integers
803 * @table: the sysctl table
804 * @write: %TRUE if this is a write to the sysctl file
805 * @buffer: the user buffer
806 * @lenp: the size of the user buffer
807 * @ppos: file position
808 *
809 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
810 * values from/to the user buffer, treated as an ASCII string.
811 *
812 * Returns 0 on success.
813 */
814int proc_dointvec(struct ctl_table *table, int write, void *buffer,
815 size_t *lenp, loff_t *ppos)
816{
817 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
818}
819
820#ifdef CONFIG_COMPACTION
821static int proc_dointvec_minmax_warn_RT_change(struct ctl_table *table,
822 int write, void *buffer, size_t *lenp, loff_t *ppos)
823{
824 int ret, old;
825
826 if (!IS_ENABLED(CONFIG_PREEMPT_RT) || !write)
827 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
828
829 old = *(int *)table->data;
830 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
831 if (ret)
832 return ret;
833 if (old != *(int *)table->data)
834 pr_warn_once("sysctl attribute %s changed by %s[%d]\n",
835 table->procname, current->comm,
836 task_pid_nr(current));
837 return ret;
838}
839#endif
840
841/**
842 * proc_douintvec - read a vector of unsigned integers
843 * @table: the sysctl table
844 * @write: %TRUE if this is a write to the sysctl file
845 * @buffer: the user buffer
846 * @lenp: the size of the user buffer
847 * @ppos: file position
848 *
849 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
850 * values from/to the user buffer, treated as an ASCII string.
851 *
852 * Returns 0 on success.
853 */
854int proc_douintvec(struct ctl_table *table, int write, void *buffer,
855 size_t *lenp, loff_t *ppos)
856{
857 return do_proc_douintvec(table, write, buffer, lenp, ppos,
858 do_proc_douintvec_conv, NULL);
859}
860
861/*
862 * Taint values can only be increased
863 * This means we can safely use a temporary.
864 */
865static int proc_taint(struct ctl_table *table, int write,
866 void *buffer, size_t *lenp, loff_t *ppos)
867{
868 struct ctl_table t;
869 unsigned long tmptaint = get_taint();
870 int err;
871
872 if (write && !capable(CAP_SYS_ADMIN))
873 return -EPERM;
874
875 t = *table;
876 t.data = &tmptaint;
877 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
878 if (err < 0)
879 return err;
880
881 if (write) {
882 int i;
883
884 /*
885 * If we are relying on panic_on_taint not producing
886 * false positives due to userspace input, bail out
887 * before setting the requested taint flags.
888 */
889 if (panic_on_taint_nousertaint && (tmptaint & panic_on_taint))
890 return -EINVAL;
891
892 /*
893 * Poor man's atomic or. Not worth adding a primitive
894 * to everyone's atomic.h for this
895 */
896 for (i = 0; i < TAINT_FLAGS_COUNT; i++)
897 if ((1UL << i) & tmptaint)
898 add_taint(i, LOCKDEP_STILL_OK);
899 }
900
901 return err;
902}
903
904#ifdef CONFIG_PRINTK
905static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
906 void *buffer, size_t *lenp, loff_t *ppos)
907{
908 if (write && !capable(CAP_SYS_ADMIN))
909 return -EPERM;
910
911 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
912}
913#endif
914
915/**
916 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
917 * @min: pointer to minimum allowable value
918 * @max: pointer to maximum allowable value
919 *
920 * The do_proc_dointvec_minmax_conv_param structure provides the
921 * minimum and maximum values for doing range checking for those sysctl
922 * parameters that use the proc_dointvec_minmax() handler.
923 */
924struct do_proc_dointvec_minmax_conv_param {
925 int *min;
926 int *max;
927};
928
929static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
930 int *valp,
931 int write, void *data)
932{
933 int tmp, ret;
934 struct do_proc_dointvec_minmax_conv_param *param = data;
935 /*
936 * If writing, first do so via a temporary local int so we can
937 * bounds-check it before touching *valp.
938 */
939 int *ip = write ? &tmp : valp;
940
941 ret = do_proc_dointvec_conv(negp, lvalp, ip, write, data);
942 if (ret)
943 return ret;
944
945 if (write) {
946 if ((param->min && *param->min > tmp) ||
947 (param->max && *param->max < tmp))
948 return -EINVAL;
949 *valp = tmp;
950 }
951
952 return 0;
953}
954
955/**
956 * proc_dointvec_minmax - read a vector of integers with min/max values
957 * @table: the sysctl table
958 * @write: %TRUE if this is a write to the sysctl file
959 * @buffer: the user buffer
960 * @lenp: the size of the user buffer
961 * @ppos: file position
962 *
963 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
964 * values from/to the user buffer, treated as an ASCII string.
965 *
966 * This routine will ensure the values are within the range specified by
967 * table->extra1 (min) and table->extra2 (max).
968 *
969 * Returns 0 on success or -EINVAL on write when the range check fails.
970 */
971int proc_dointvec_minmax(struct ctl_table *table, int write,
972 void *buffer, size_t *lenp, loff_t *ppos)
973{
974 struct do_proc_dointvec_minmax_conv_param param = {
975 .min = (int *) table->extra1,
976 .max = (int *) table->extra2,
977 };
978 return do_proc_dointvec(table, write, buffer, lenp, ppos,
979 do_proc_dointvec_minmax_conv, ¶m);
980}
981
982/**
983 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
984 * @min: pointer to minimum allowable value
985 * @max: pointer to maximum allowable value
986 *
987 * The do_proc_douintvec_minmax_conv_param structure provides the
988 * minimum and maximum values for doing range checking for those sysctl
989 * parameters that use the proc_douintvec_minmax() handler.
990 */
991struct do_proc_douintvec_minmax_conv_param {
992 unsigned int *min;
993 unsigned int *max;
994};
995
996static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
997 unsigned int *valp,
998 int write, void *data)
999{
1000 int ret;
1001 unsigned int tmp;
1002 struct do_proc_douintvec_minmax_conv_param *param = data;
1003 /* write via temporary local uint for bounds-checking */
1004 unsigned int *up = write ? &tmp : valp;
1005
1006 ret = do_proc_douintvec_conv(lvalp, up, write, data);
1007 if (ret)
1008 return ret;
1009
1010 if (write) {
1011 if ((param->min && *param->min > tmp) ||
1012 (param->max && *param->max < tmp))
1013 return -ERANGE;
1014
1015 *valp = tmp;
1016 }
1017
1018 return 0;
1019}
1020
1021/**
1022 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
1023 * @table: the sysctl table
1024 * @write: %TRUE if this is a write to the sysctl file
1025 * @buffer: the user buffer
1026 * @lenp: the size of the user buffer
1027 * @ppos: file position
1028 *
1029 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
1030 * values from/to the user buffer, treated as an ASCII string. Negative
1031 * strings are not allowed.
1032 *
1033 * This routine will ensure the values are within the range specified by
1034 * table->extra1 (min) and table->extra2 (max). There is a final sanity
1035 * check for UINT_MAX to avoid having to support wrap around uses from
1036 * userspace.
1037 *
1038 * Returns 0 on success or -ERANGE on write when the range check fails.
1039 */
1040int proc_douintvec_minmax(struct ctl_table *table, int write,
1041 void *buffer, size_t *lenp, loff_t *ppos)
1042{
1043 struct do_proc_douintvec_minmax_conv_param param = {
1044 .min = (unsigned int *) table->extra1,
1045 .max = (unsigned int *) table->extra2,
1046 };
1047 return do_proc_douintvec(table, write, buffer, lenp, ppos,
1048 do_proc_douintvec_minmax_conv, ¶m);
1049}
1050
1051/**
1052 * proc_dou8vec_minmax - read a vector of unsigned chars with min/max values
1053 * @table: the sysctl table
1054 * @write: %TRUE if this is a write to the sysctl file
1055 * @buffer: the user buffer
1056 * @lenp: the size of the user buffer
1057 * @ppos: file position
1058 *
1059 * Reads/writes up to table->maxlen/sizeof(u8) unsigned chars
1060 * values from/to the user buffer, treated as an ASCII string. Negative
1061 * strings are not allowed.
1062 *
1063 * This routine will ensure the values are within the range specified by
1064 * table->extra1 (min) and table->extra2 (max).
1065 *
1066 * Returns 0 on success or an error on write when the range check fails.
1067 */
1068int proc_dou8vec_minmax(struct ctl_table *table, int write,
1069 void *buffer, size_t *lenp, loff_t *ppos)
1070{
1071 struct ctl_table tmp;
1072 unsigned int min = 0, max = 255U, val;
1073 u8 *data = table->data;
1074 struct do_proc_douintvec_minmax_conv_param param = {
1075 .min = &min,
1076 .max = &max,
1077 };
1078 int res;
1079
1080 /* Do not support arrays yet. */
1081 if (table->maxlen != sizeof(u8))
1082 return -EINVAL;
1083
1084 if (table->extra1) {
1085 min = *(unsigned int *) table->extra1;
1086 if (min > 255U)
1087 return -EINVAL;
1088 }
1089 if (table->extra2) {
1090 max = *(unsigned int *) table->extra2;
1091 if (max > 255U)
1092 return -EINVAL;
1093 }
1094
1095 tmp = *table;
1096
1097 tmp.maxlen = sizeof(val);
1098 tmp.data = &val;
1099 val = *data;
1100 res = do_proc_douintvec(&tmp, write, buffer, lenp, ppos,
1101 do_proc_douintvec_minmax_conv, ¶m);
1102 if (res)
1103 return res;
1104 if (write)
1105 *data = val;
1106 return 0;
1107}
1108EXPORT_SYMBOL_GPL(proc_dou8vec_minmax);
1109
1110static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
1111 unsigned int *valp,
1112 int write, void *data)
1113{
1114 if (write) {
1115 unsigned int val;
1116
1117 val = round_pipe_size(*lvalp);
1118 if (val == 0)
1119 return -EINVAL;
1120
1121 *valp = val;
1122 } else {
1123 unsigned int val = *valp;
1124 *lvalp = (unsigned long) val;
1125 }
1126
1127 return 0;
1128}
1129
1130static int proc_dopipe_max_size(struct ctl_table *table, int write,
1131 void *buffer, size_t *lenp, loff_t *ppos)
1132{
1133 return do_proc_douintvec(table, write, buffer, lenp, ppos,
1134 do_proc_dopipe_max_size_conv, NULL);
1135}
1136
1137static void validate_coredump_safety(void)
1138{
1139#ifdef CONFIG_COREDUMP
1140 if (suid_dumpable == SUID_DUMP_ROOT &&
1141 core_pattern[0] != '/' && core_pattern[0] != '|') {
1142 printk(KERN_WARNING
1143"Unsafe core_pattern used with fs.suid_dumpable=2.\n"
1144"Pipe handler or fully qualified core dump path required.\n"
1145"Set kernel.core_pattern before fs.suid_dumpable.\n"
1146 );
1147 }
1148#endif
1149}
1150
1151static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
1152 void *buffer, size_t *lenp, loff_t *ppos)
1153{
1154 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
1155 if (!error)
1156 validate_coredump_safety();
1157 return error;
1158}
1159
1160#ifdef CONFIG_COREDUMP
1161static int proc_dostring_coredump(struct ctl_table *table, int write,
1162 void *buffer, size_t *lenp, loff_t *ppos)
1163{
1164 int error = proc_dostring(table, write, buffer, lenp, ppos);
1165 if (!error)
1166 validate_coredump_safety();
1167 return error;
1168}
1169#endif
1170
1171#ifdef CONFIG_MAGIC_SYSRQ
1172static int sysrq_sysctl_handler(struct ctl_table *table, int write,
1173 void *buffer, size_t *lenp, loff_t *ppos)
1174{
1175 int tmp, ret;
1176
1177 tmp = sysrq_mask();
1178
1179 ret = __do_proc_dointvec(&tmp, table, write, buffer,
1180 lenp, ppos, NULL, NULL);
1181 if (ret || !write)
1182 return ret;
1183
1184 if (write)
1185 sysrq_toggle_support(tmp);
1186
1187 return 0;
1188}
1189#endif
1190
1191static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table,
1192 int write, void *buffer, size_t *lenp, loff_t *ppos,
1193 unsigned long convmul, unsigned long convdiv)
1194{
1195 unsigned long *i, *min, *max;
1196 int vleft, first = 1, err = 0;
1197 size_t left;
1198 char *p;
1199
1200 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
1201 *lenp = 0;
1202 return 0;
1203 }
1204
1205 i = (unsigned long *) data;
1206 min = (unsigned long *) table->extra1;
1207 max = (unsigned long *) table->extra2;
1208 vleft = table->maxlen / sizeof(unsigned long);
1209 left = *lenp;
1210
1211 if (write) {
1212 if (proc_first_pos_non_zero_ignore(ppos, table))
1213 goto out;
1214
1215 if (left > PAGE_SIZE - 1)
1216 left = PAGE_SIZE - 1;
1217 p = buffer;
1218 }
1219
1220 for (; left && vleft--; i++, first = 0) {
1221 unsigned long val;
1222
1223 if (write) {
1224 bool neg;
1225
1226 left -= proc_skip_spaces(&p);
1227 if (!left)
1228 break;
1229
1230 err = proc_get_long(&p, &left, &val, &neg,
1231 proc_wspace_sep,
1232 sizeof(proc_wspace_sep), NULL);
1233 if (err)
1234 break;
1235 if (neg)
1236 continue;
1237 val = convmul * val / convdiv;
1238 if ((min && val < *min) || (max && val > *max)) {
1239 err = -EINVAL;
1240 break;
1241 }
1242 *i = val;
1243 } else {
1244 val = convdiv * (*i) / convmul;
1245 if (!first)
1246 proc_put_char(&buffer, &left, '\t');
1247 proc_put_long(&buffer, &left, val, false);
1248 }
1249 }
1250
1251 if (!write && !first && left && !err)
1252 proc_put_char(&buffer, &left, '\n');
1253 if (write && !err)
1254 left -= proc_skip_spaces(&p);
1255 if (write && first)
1256 return err ? : -EINVAL;
1257 *lenp -= left;
1258out:
1259 *ppos += *lenp;
1260 return err;
1261}
1262
1263static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
1264 void *buffer, size_t *lenp, loff_t *ppos, unsigned long convmul,
1265 unsigned long convdiv)
1266{
1267 return __do_proc_doulongvec_minmax(table->data, table, write,
1268 buffer, lenp, ppos, convmul, convdiv);
1269}
1270
1271/**
1272 * proc_doulongvec_minmax - read a vector of long integers with min/max values
1273 * @table: the sysctl table
1274 * @write: %TRUE if this is a write to the sysctl file
1275 * @buffer: the user buffer
1276 * @lenp: the size of the user buffer
1277 * @ppos: file position
1278 *
1279 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1280 * values from/to the user buffer, treated as an ASCII string.
1281 *
1282 * This routine will ensure the values are within the range specified by
1283 * table->extra1 (min) and table->extra2 (max).
1284 *
1285 * Returns 0 on success.
1286 */
1287int proc_doulongvec_minmax(struct ctl_table *table, int write,
1288 void *buffer, size_t *lenp, loff_t *ppos)
1289{
1290 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
1291}
1292
1293/**
1294 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
1295 * @table: the sysctl table
1296 * @write: %TRUE if this is a write to the sysctl file
1297 * @buffer: the user buffer
1298 * @lenp: the size of the user buffer
1299 * @ppos: file position
1300 *
1301 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1302 * values from/to the user buffer, treated as an ASCII string. The values
1303 * are treated as milliseconds, and converted to jiffies when they are stored.
1304 *
1305 * This routine will ensure the values are within the range specified by
1306 * table->extra1 (min) and table->extra2 (max).
1307 *
1308 * Returns 0 on success.
1309 */
1310int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1311 void *buffer, size_t *lenp, loff_t *ppos)
1312{
1313 return do_proc_doulongvec_minmax(table, write, buffer,
1314 lenp, ppos, HZ, 1000l);
1315}
1316
1317
1318static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
1319 int *valp,
1320 int write, void *data)
1321{
1322 if (write) {
1323 if (*lvalp > INT_MAX / HZ)
1324 return 1;
1325 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
1326 } else {
1327 int val = *valp;
1328 unsigned long lval;
1329 if (val < 0) {
1330 *negp = true;
1331 lval = -(unsigned long)val;
1332 } else {
1333 *negp = false;
1334 lval = (unsigned long)val;
1335 }
1336 *lvalp = lval / HZ;
1337 }
1338 return 0;
1339}
1340
1341static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
1342 int *valp,
1343 int write, void *data)
1344{
1345 if (write) {
1346 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
1347 return 1;
1348 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
1349 } else {
1350 int val = *valp;
1351 unsigned long lval;
1352 if (val < 0) {
1353 *negp = true;
1354 lval = -(unsigned long)val;
1355 } else {
1356 *negp = false;
1357 lval = (unsigned long)val;
1358 }
1359 *lvalp = jiffies_to_clock_t(lval);
1360 }
1361 return 0;
1362}
1363
1364static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
1365 int *valp,
1366 int write, void *data)
1367{
1368 if (write) {
1369 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
1370
1371 if (jif > INT_MAX)
1372 return 1;
1373 *valp = (int)jif;
1374 } else {
1375 int val = *valp;
1376 unsigned long lval;
1377 if (val < 0) {
1378 *negp = true;
1379 lval = -(unsigned long)val;
1380 } else {
1381 *negp = false;
1382 lval = (unsigned long)val;
1383 }
1384 *lvalp = jiffies_to_msecs(lval);
1385 }
1386 return 0;
1387}
1388
1389/**
1390 * proc_dointvec_jiffies - read a vector of integers as seconds
1391 * @table: the sysctl table
1392 * @write: %TRUE if this is a write to the sysctl file
1393 * @buffer: the user buffer
1394 * @lenp: the size of the user buffer
1395 * @ppos: file position
1396 *
1397 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1398 * values from/to the user buffer, treated as an ASCII string.
1399 * The values read are assumed to be in seconds, and are converted into
1400 * jiffies.
1401 *
1402 * Returns 0 on success.
1403 */
1404int proc_dointvec_jiffies(struct ctl_table *table, int write,
1405 void *buffer, size_t *lenp, loff_t *ppos)
1406{
1407 return do_proc_dointvec(table,write,buffer,lenp,ppos,
1408 do_proc_dointvec_jiffies_conv,NULL);
1409}
1410
1411/**
1412 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
1413 * @table: the sysctl table
1414 * @write: %TRUE if this is a write to the sysctl file
1415 * @buffer: the user buffer
1416 * @lenp: the size of the user buffer
1417 * @ppos: pointer to the file position
1418 *
1419 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1420 * values from/to the user buffer, treated as an ASCII string.
1421 * The values read are assumed to be in 1/USER_HZ seconds, and
1422 * are converted into jiffies.
1423 *
1424 * Returns 0 on success.
1425 */
1426int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1427 void *buffer, size_t *lenp, loff_t *ppos)
1428{
1429 return do_proc_dointvec(table,write,buffer,lenp,ppos,
1430 do_proc_dointvec_userhz_jiffies_conv,NULL);
1431}
1432
1433/**
1434 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
1435 * @table: the sysctl table
1436 * @write: %TRUE if this is a write to the sysctl file
1437 * @buffer: the user buffer
1438 * @lenp: the size of the user buffer
1439 * @ppos: file position
1440 * @ppos: the current position in the file
1441 *
1442 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1443 * values from/to the user buffer, treated as an ASCII string.
1444 * The values read are assumed to be in 1/1000 seconds, and
1445 * are converted into jiffies.
1446 *
1447 * Returns 0 on success.
1448 */
1449int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, void *buffer,
1450 size_t *lenp, loff_t *ppos)
1451{
1452 return do_proc_dointvec(table, write, buffer, lenp, ppos,
1453 do_proc_dointvec_ms_jiffies_conv, NULL);
1454}
1455
1456static int proc_do_cad_pid(struct ctl_table *table, int write, void *buffer,
1457 size_t *lenp, loff_t *ppos)
1458{
1459 struct pid *new_pid;
1460 pid_t tmp;
1461 int r;
1462
1463 tmp = pid_vnr(cad_pid);
1464
1465 r = __do_proc_dointvec(&tmp, table, write, buffer,
1466 lenp, ppos, NULL, NULL);
1467 if (r || !write)
1468 return r;
1469
1470 new_pid = find_get_pid(tmp);
1471 if (!new_pid)
1472 return -ESRCH;
1473
1474 put_pid(xchg(&cad_pid, new_pid));
1475 return 0;
1476}
1477
1478/**
1479 * proc_do_large_bitmap - read/write from/to a large bitmap
1480 * @table: the sysctl table
1481 * @write: %TRUE if this is a write to the sysctl file
1482 * @buffer: the user buffer
1483 * @lenp: the size of the user buffer
1484 * @ppos: file position
1485 *
1486 * The bitmap is stored at table->data and the bitmap length (in bits)
1487 * in table->maxlen.
1488 *
1489 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
1490 * large bitmaps may be represented in a compact manner. Writing into
1491 * the file will clear the bitmap then update it with the given input.
1492 *
1493 * Returns 0 on success.
1494 */
1495int proc_do_large_bitmap(struct ctl_table *table, int write,
1496 void *buffer, size_t *lenp, loff_t *ppos)
1497{
1498 int err = 0;
1499 size_t left = *lenp;
1500 unsigned long bitmap_len = table->maxlen;
1501 unsigned long *bitmap = *(unsigned long **) table->data;
1502 unsigned long *tmp_bitmap = NULL;
1503 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
1504
1505 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
1506 *lenp = 0;
1507 return 0;
1508 }
1509
1510 if (write) {
1511 char *p = buffer;
1512 size_t skipped = 0;
1513
1514 if (left > PAGE_SIZE - 1) {
1515 left = PAGE_SIZE - 1;
1516 /* How much of the buffer we'll skip this pass */
1517 skipped = *lenp - left;
1518 }
1519
1520 tmp_bitmap = bitmap_zalloc(bitmap_len, GFP_KERNEL);
1521 if (!tmp_bitmap)
1522 return -ENOMEM;
1523 proc_skip_char(&p, &left, '\n');
1524 while (!err && left) {
1525 unsigned long val_a, val_b;
1526 bool neg;
1527 size_t saved_left;
1528
1529 /* In case we stop parsing mid-number, we can reset */
1530 saved_left = left;
1531 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
1532 sizeof(tr_a), &c);
1533 /*
1534 * If we consumed the entirety of a truncated buffer or
1535 * only one char is left (may be a "-"), then stop here,
1536 * reset, & come back for more.
1537 */
1538 if ((left <= 1) && skipped) {
1539 left = saved_left;
1540 break;
1541 }
1542
1543 if (err)
1544 break;
1545 if (val_a >= bitmap_len || neg) {
1546 err = -EINVAL;
1547 break;
1548 }
1549
1550 val_b = val_a;
1551 if (left) {
1552 p++;
1553 left--;
1554 }
1555
1556 if (c == '-') {
1557 err = proc_get_long(&p, &left, &val_b,
1558 &neg, tr_b, sizeof(tr_b),
1559 &c);
1560 /*
1561 * If we consumed all of a truncated buffer or
1562 * then stop here, reset, & come back for more.
1563 */
1564 if (!left && skipped) {
1565 left = saved_left;
1566 break;
1567 }
1568
1569 if (err)
1570 break;
1571 if (val_b >= bitmap_len || neg ||
1572 val_a > val_b) {
1573 err = -EINVAL;
1574 break;
1575 }
1576 if (left) {
1577 p++;
1578 left--;
1579 }
1580 }
1581
1582 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
1583 proc_skip_char(&p, &left, '\n');
1584 }
1585 left += skipped;
1586 } else {
1587 unsigned long bit_a, bit_b = 0;
1588 bool first = 1;
1589
1590 while (left) {
1591 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
1592 if (bit_a >= bitmap_len)
1593 break;
1594 bit_b = find_next_zero_bit(bitmap, bitmap_len,
1595 bit_a + 1) - 1;
1596
1597 if (!first)
1598 proc_put_char(&buffer, &left, ',');
1599 proc_put_long(&buffer, &left, bit_a, false);
1600 if (bit_a != bit_b) {
1601 proc_put_char(&buffer, &left, '-');
1602 proc_put_long(&buffer, &left, bit_b, false);
1603 }
1604
1605 first = 0; bit_b++;
1606 }
1607 proc_put_char(&buffer, &left, '\n');
1608 }
1609
1610 if (!err) {
1611 if (write) {
1612 if (*ppos)
1613 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
1614 else
1615 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
1616 }
1617 *lenp -= left;
1618 *ppos += *lenp;
1619 }
1620
1621 bitmap_free(tmp_bitmap);
1622 return err;
1623}
1624
1625#else /* CONFIG_PROC_SYSCTL */
1626
1627int proc_dostring(struct ctl_table *table, int write,
1628 void *buffer, size_t *lenp, loff_t *ppos)
1629{
1630 return -ENOSYS;
1631}
1632
1633int proc_dointvec(struct ctl_table *table, int write,
1634 void *buffer, size_t *lenp, loff_t *ppos)
1635{
1636 return -ENOSYS;
1637}
1638
1639int proc_douintvec(struct ctl_table *table, int write,
1640 void *buffer, size_t *lenp, loff_t *ppos)
1641{
1642 return -ENOSYS;
1643}
1644
1645int proc_dointvec_minmax(struct ctl_table *table, int write,
1646 void *buffer, size_t *lenp, loff_t *ppos)
1647{
1648 return -ENOSYS;
1649}
1650
1651int proc_douintvec_minmax(struct ctl_table *table, int write,
1652 void *buffer, size_t *lenp, loff_t *ppos)
1653{
1654 return -ENOSYS;
1655}
1656
1657int proc_dou8vec_minmax(struct ctl_table *table, int write,
1658 void *buffer, size_t *lenp, loff_t *ppos)
1659{
1660 return -ENOSYS;
1661}
1662
1663int proc_dointvec_jiffies(struct ctl_table *table, int write,
1664 void *buffer, size_t *lenp, loff_t *ppos)
1665{
1666 return -ENOSYS;
1667}
1668
1669int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1670 void *buffer, size_t *lenp, loff_t *ppos)
1671{
1672 return -ENOSYS;
1673}
1674
1675int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
1676 void *buffer, size_t *lenp, loff_t *ppos)
1677{
1678 return -ENOSYS;
1679}
1680
1681int proc_doulongvec_minmax(struct ctl_table *table, int write,
1682 void *buffer, size_t *lenp, loff_t *ppos)
1683{
1684 return -ENOSYS;
1685}
1686
1687int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1688 void *buffer, size_t *lenp, loff_t *ppos)
1689{
1690 return -ENOSYS;
1691}
1692
1693int proc_do_large_bitmap(struct ctl_table *table, int write,
1694 void *buffer, size_t *lenp, loff_t *ppos)
1695{
1696 return -ENOSYS;
1697}
1698
1699#endif /* CONFIG_PROC_SYSCTL */
1700
1701#if defined(CONFIG_SYSCTL)
1702int proc_do_static_key(struct ctl_table *table, int write,
1703 void *buffer, size_t *lenp, loff_t *ppos)
1704{
1705 struct static_key *key = (struct static_key *)table->data;
1706 static DEFINE_MUTEX(static_key_mutex);
1707 int val, ret;
1708 struct ctl_table tmp = {
1709 .data = &val,
1710 .maxlen = sizeof(val),
1711 .mode = table->mode,
1712 .extra1 = SYSCTL_ZERO,
1713 .extra2 = SYSCTL_ONE,
1714 };
1715
1716 if (write && !capable(CAP_SYS_ADMIN))
1717 return -EPERM;
1718
1719 mutex_lock(&static_key_mutex);
1720 val = static_key_enabled(key);
1721 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
1722 if (write && !ret) {
1723 if (val)
1724 static_key_enable(key);
1725 else
1726 static_key_disable(key);
1727 }
1728 mutex_unlock(&static_key_mutex);
1729 return ret;
1730}
1731
1732static struct ctl_table kern_table[] = {
1733 {
1734 .procname = "sched_child_runs_first",
1735 .data = &sysctl_sched_child_runs_first,
1736 .maxlen = sizeof(unsigned int),
1737 .mode = 0644,
1738 .proc_handler = proc_dointvec,
1739 },
1740#ifdef CONFIG_SCHEDSTATS
1741 {
1742 .procname = "sched_schedstats",
1743 .data = NULL,
1744 .maxlen = sizeof(unsigned int),
1745 .mode = 0644,
1746 .proc_handler = sysctl_schedstats,
1747 .extra1 = SYSCTL_ZERO,
1748 .extra2 = SYSCTL_ONE,
1749 },
1750#endif /* CONFIG_SCHEDSTATS */
1751#ifdef CONFIG_TASK_DELAY_ACCT
1752 {
1753 .procname = "task_delayacct",
1754 .data = NULL,
1755 .maxlen = sizeof(unsigned int),
1756 .mode = 0644,
1757 .proc_handler = sysctl_delayacct,
1758 .extra1 = SYSCTL_ZERO,
1759 .extra2 = SYSCTL_ONE,
1760 },
1761#endif /* CONFIG_TASK_DELAY_ACCT */
1762#ifdef CONFIG_NUMA_BALANCING
1763 {
1764 .procname = "numa_balancing",
1765 .data = NULL, /* filled in by handler */
1766 .maxlen = sizeof(unsigned int),
1767 .mode = 0644,
1768 .proc_handler = sysctl_numa_balancing,
1769 .extra1 = SYSCTL_ZERO,
1770 .extra2 = SYSCTL_ONE,
1771 },
1772#endif /* CONFIG_NUMA_BALANCING */
1773 {
1774 .procname = "sched_rt_period_us",
1775 .data = &sysctl_sched_rt_period,
1776 .maxlen = sizeof(unsigned int),
1777 .mode = 0644,
1778 .proc_handler = sched_rt_handler,
1779 },
1780 {
1781 .procname = "sched_rt_runtime_us",
1782 .data = &sysctl_sched_rt_runtime,
1783 .maxlen = sizeof(int),
1784 .mode = 0644,
1785 .proc_handler = sched_rt_handler,
1786 },
1787 {
1788 .procname = "sched_deadline_period_max_us",
1789 .data = &sysctl_sched_dl_period_max,
1790 .maxlen = sizeof(unsigned int),
1791 .mode = 0644,
1792 .proc_handler = proc_dointvec,
1793 },
1794 {
1795 .procname = "sched_deadline_period_min_us",
1796 .data = &sysctl_sched_dl_period_min,
1797 .maxlen = sizeof(unsigned int),
1798 .mode = 0644,
1799 .proc_handler = proc_dointvec,
1800 },
1801 {
1802 .procname = "sched_rr_timeslice_ms",
1803 .data = &sysctl_sched_rr_timeslice,
1804 .maxlen = sizeof(int),
1805 .mode = 0644,
1806 .proc_handler = sched_rr_handler,
1807 },
1808#ifdef CONFIG_UCLAMP_TASK
1809 {
1810 .procname = "sched_util_clamp_min",
1811 .data = &sysctl_sched_uclamp_util_min,
1812 .maxlen = sizeof(unsigned int),
1813 .mode = 0644,
1814 .proc_handler = sysctl_sched_uclamp_handler,
1815 },
1816 {
1817 .procname = "sched_util_clamp_max",
1818 .data = &sysctl_sched_uclamp_util_max,
1819 .maxlen = sizeof(unsigned int),
1820 .mode = 0644,
1821 .proc_handler = sysctl_sched_uclamp_handler,
1822 },
1823 {
1824 .procname = "sched_util_clamp_min_rt_default",
1825 .data = &sysctl_sched_uclamp_util_min_rt_default,
1826 .maxlen = sizeof(unsigned int),
1827 .mode = 0644,
1828 .proc_handler = sysctl_sched_uclamp_handler,
1829 },
1830#endif
1831#ifdef CONFIG_SCHED_AUTOGROUP
1832 {
1833 .procname = "sched_autogroup_enabled",
1834 .data = &sysctl_sched_autogroup_enabled,
1835 .maxlen = sizeof(unsigned int),
1836 .mode = 0644,
1837 .proc_handler = proc_dointvec_minmax,
1838 .extra1 = SYSCTL_ZERO,
1839 .extra2 = SYSCTL_ONE,
1840 },
1841#endif
1842#ifdef CONFIG_CFS_BANDWIDTH
1843 {
1844 .procname = "sched_cfs_bandwidth_slice_us",
1845 .data = &sysctl_sched_cfs_bandwidth_slice,
1846 .maxlen = sizeof(unsigned int),
1847 .mode = 0644,
1848 .proc_handler = proc_dointvec_minmax,
1849 .extra1 = SYSCTL_ONE,
1850 },
1851#endif
1852#if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_CPU_FREQ_GOV_SCHEDUTIL)
1853 {
1854 .procname = "sched_energy_aware",
1855 .data = &sysctl_sched_energy_aware,
1856 .maxlen = sizeof(unsigned int),
1857 .mode = 0644,
1858 .proc_handler = sched_energy_aware_handler,
1859 .extra1 = SYSCTL_ZERO,
1860 .extra2 = SYSCTL_ONE,
1861 },
1862#endif
1863#ifdef CONFIG_PROVE_LOCKING
1864 {
1865 .procname = "prove_locking",
1866 .data = &prove_locking,
1867 .maxlen = sizeof(int),
1868 .mode = 0644,
1869 .proc_handler = proc_dointvec,
1870 },
1871#endif
1872#ifdef CONFIG_LOCK_STAT
1873 {
1874 .procname = "lock_stat",
1875 .data = &lock_stat,
1876 .maxlen = sizeof(int),
1877 .mode = 0644,
1878 .proc_handler = proc_dointvec,
1879 },
1880#endif
1881 {
1882 .procname = "panic",
1883 .data = &panic_timeout,
1884 .maxlen = sizeof(int),
1885 .mode = 0644,
1886 .proc_handler = proc_dointvec,
1887 },
1888#ifdef CONFIG_COREDUMP
1889 {
1890 .procname = "core_uses_pid",
1891 .data = &core_uses_pid,
1892 .maxlen = sizeof(int),
1893 .mode = 0644,
1894 .proc_handler = proc_dointvec,
1895 },
1896 {
1897 .procname = "core_pattern",
1898 .data = core_pattern,
1899 .maxlen = CORENAME_MAX_SIZE,
1900 .mode = 0644,
1901 .proc_handler = proc_dostring_coredump,
1902 },
1903 {
1904 .procname = "core_pipe_limit",
1905 .data = &core_pipe_limit,
1906 .maxlen = sizeof(unsigned int),
1907 .mode = 0644,
1908 .proc_handler = proc_dointvec,
1909 },
1910#endif
1911#ifdef CONFIG_PROC_SYSCTL
1912 {
1913 .procname = "tainted",
1914 .maxlen = sizeof(long),
1915 .mode = 0644,
1916 .proc_handler = proc_taint,
1917 },
1918 {
1919 .procname = "sysctl_writes_strict",
1920 .data = &sysctl_writes_strict,
1921 .maxlen = sizeof(int),
1922 .mode = 0644,
1923 .proc_handler = proc_dointvec_minmax,
1924 .extra1 = &neg_one,
1925 .extra2 = SYSCTL_ONE,
1926 },
1927#endif
1928#ifdef CONFIG_LATENCYTOP
1929 {
1930 .procname = "latencytop",
1931 .data = &latencytop_enabled,
1932 .maxlen = sizeof(int),
1933 .mode = 0644,
1934 .proc_handler = sysctl_latencytop,
1935 },
1936#endif
1937#ifdef CONFIG_BLK_DEV_INITRD
1938 {
1939 .procname = "real-root-dev",
1940 .data = &real_root_dev,
1941 .maxlen = sizeof(int),
1942 .mode = 0644,
1943 .proc_handler = proc_dointvec,
1944 },
1945#endif
1946 {
1947 .procname = "print-fatal-signals",
1948 .data = &print_fatal_signals,
1949 .maxlen = sizeof(int),
1950 .mode = 0644,
1951 .proc_handler = proc_dointvec,
1952 },
1953#ifdef CONFIG_SPARC
1954 {
1955 .procname = "reboot-cmd",
1956 .data = reboot_command,
1957 .maxlen = 256,
1958 .mode = 0644,
1959 .proc_handler = proc_dostring,
1960 },
1961 {
1962 .procname = "stop-a",
1963 .data = &stop_a_enabled,
1964 .maxlen = sizeof (int),
1965 .mode = 0644,
1966 .proc_handler = proc_dointvec,
1967 },
1968 {
1969 .procname = "scons-poweroff",
1970 .data = &scons_pwroff,
1971 .maxlen = sizeof (int),
1972 .mode = 0644,
1973 .proc_handler = proc_dointvec,
1974 },
1975#endif
1976#ifdef CONFIG_SPARC64
1977 {
1978 .procname = "tsb-ratio",
1979 .data = &sysctl_tsb_ratio,
1980 .maxlen = sizeof (int),
1981 .mode = 0644,
1982 .proc_handler = proc_dointvec,
1983 },
1984#endif
1985#ifdef CONFIG_PARISC
1986 {
1987 .procname = "soft-power",
1988 .data = &pwrsw_enabled,
1989 .maxlen = sizeof (int),
1990 .mode = 0644,
1991 .proc_handler = proc_dointvec,
1992 },
1993#endif
1994#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
1995 {
1996 .procname = "unaligned-trap",
1997 .data = &unaligned_enabled,
1998 .maxlen = sizeof (int),
1999 .mode = 0644,
2000 .proc_handler = proc_dointvec,
2001 },
2002#endif
2003 {
2004 .procname = "ctrl-alt-del",
2005 .data = &C_A_D,
2006 .maxlen = sizeof(int),
2007 .mode = 0644,
2008 .proc_handler = proc_dointvec,
2009 },
2010#ifdef CONFIG_FUNCTION_TRACER
2011 {
2012 .procname = "ftrace_enabled",
2013 .data = &ftrace_enabled,
2014 .maxlen = sizeof(int),
2015 .mode = 0644,
2016 .proc_handler = ftrace_enable_sysctl,
2017 },
2018#endif
2019#ifdef CONFIG_STACK_TRACER
2020 {
2021 .procname = "stack_tracer_enabled",
2022 .data = &stack_tracer_enabled,
2023 .maxlen = sizeof(int),
2024 .mode = 0644,
2025 .proc_handler = stack_trace_sysctl,
2026 },
2027#endif
2028#ifdef CONFIG_TRACING
2029 {
2030 .procname = "ftrace_dump_on_oops",
2031 .data = &ftrace_dump_on_oops,
2032 .maxlen = sizeof(int),
2033 .mode = 0644,
2034 .proc_handler = proc_dointvec,
2035 },
2036 {
2037 .procname = "traceoff_on_warning",
2038 .data = &__disable_trace_on_warning,
2039 .maxlen = sizeof(__disable_trace_on_warning),
2040 .mode = 0644,
2041 .proc_handler = proc_dointvec,
2042 },
2043 {
2044 .procname = "tracepoint_printk",
2045 .data = &tracepoint_printk,
2046 .maxlen = sizeof(tracepoint_printk),
2047 .mode = 0644,
2048 .proc_handler = tracepoint_printk_sysctl,
2049 },
2050#endif
2051#ifdef CONFIG_KEXEC_CORE
2052 {
2053 .procname = "kexec_load_disabled",
2054 .data = &kexec_load_disabled,
2055 .maxlen = sizeof(int),
2056 .mode = 0644,
2057 /* only handle a transition from default "0" to "1" */
2058 .proc_handler = proc_dointvec_minmax,
2059 .extra1 = SYSCTL_ONE,
2060 .extra2 = SYSCTL_ONE,
2061 },
2062#endif
2063#ifdef CONFIG_MODULES
2064 {
2065 .procname = "modprobe",
2066 .data = &modprobe_path,
2067 .maxlen = KMOD_PATH_LEN,
2068 .mode = 0644,
2069 .proc_handler = proc_dostring,
2070 },
2071 {
2072 .procname = "modules_disabled",
2073 .data = &modules_disabled,
2074 .maxlen = sizeof(int),
2075 .mode = 0644,
2076 /* only handle a transition from default "0" to "1" */
2077 .proc_handler = proc_dointvec_minmax,
2078 .extra1 = SYSCTL_ONE,
2079 .extra2 = SYSCTL_ONE,
2080 },
2081#endif
2082#ifdef CONFIG_UEVENT_HELPER
2083 {
2084 .procname = "hotplug",
2085 .data = &uevent_helper,
2086 .maxlen = UEVENT_HELPER_PATH_LEN,
2087 .mode = 0644,
2088 .proc_handler = proc_dostring,
2089 },
2090#endif
2091#ifdef CONFIG_CHR_DEV_SG
2092 {
2093 .procname = "sg-big-buff",
2094 .data = &sg_big_buff,
2095 .maxlen = sizeof (int),
2096 .mode = 0444,
2097 .proc_handler = proc_dointvec,
2098 },
2099#endif
2100#ifdef CONFIG_BSD_PROCESS_ACCT
2101 {
2102 .procname = "acct",
2103 .data = &acct_parm,
2104 .maxlen = 3*sizeof(int),
2105 .mode = 0644,
2106 .proc_handler = proc_dointvec,
2107 },
2108#endif
2109#ifdef CONFIG_MAGIC_SYSRQ
2110 {
2111 .procname = "sysrq",
2112 .data = NULL,
2113 .maxlen = sizeof (int),
2114 .mode = 0644,
2115 .proc_handler = sysrq_sysctl_handler,
2116 },
2117#endif
2118#ifdef CONFIG_PROC_SYSCTL
2119 {
2120 .procname = "cad_pid",
2121 .data = NULL,
2122 .maxlen = sizeof (int),
2123 .mode = 0600,
2124 .proc_handler = proc_do_cad_pid,
2125 },
2126#endif
2127 {
2128 .procname = "threads-max",
2129 .data = NULL,
2130 .maxlen = sizeof(int),
2131 .mode = 0644,
2132 .proc_handler = sysctl_max_threads,
2133 },
2134 {
2135 .procname = "random",
2136 .mode = 0555,
2137 .child = random_table,
2138 },
2139 {
2140 .procname = "usermodehelper",
2141 .mode = 0555,
2142 .child = usermodehelper_table,
2143 },
2144#ifdef CONFIG_FW_LOADER_USER_HELPER
2145 {
2146 .procname = "firmware_config",
2147 .mode = 0555,
2148 .child = firmware_config_table,
2149 },
2150#endif
2151 {
2152 .procname = "overflowuid",
2153 .data = &overflowuid,
2154 .maxlen = sizeof(int),
2155 .mode = 0644,
2156 .proc_handler = proc_dointvec_minmax,
2157 .extra1 = &minolduid,
2158 .extra2 = &maxolduid,
2159 },
2160 {
2161 .procname = "overflowgid",
2162 .data = &overflowgid,
2163 .maxlen = sizeof(int),
2164 .mode = 0644,
2165 .proc_handler = proc_dointvec_minmax,
2166 .extra1 = &minolduid,
2167 .extra2 = &maxolduid,
2168 },
2169#ifdef CONFIG_S390
2170 {
2171 .procname = "userprocess_debug",
2172 .data = &show_unhandled_signals,
2173 .maxlen = sizeof(int),
2174 .mode = 0644,
2175 .proc_handler = proc_dointvec,
2176 },
2177#endif
2178#ifdef CONFIG_SMP
2179 {
2180 .procname = "oops_all_cpu_backtrace",
2181 .data = &sysctl_oops_all_cpu_backtrace,
2182 .maxlen = sizeof(int),
2183 .mode = 0644,
2184 .proc_handler = proc_dointvec_minmax,
2185 .extra1 = SYSCTL_ZERO,
2186 .extra2 = SYSCTL_ONE,
2187 },
2188#endif /* CONFIG_SMP */
2189 {
2190 .procname = "pid_max",
2191 .data = &pid_max,
2192 .maxlen = sizeof (int),
2193 .mode = 0644,
2194 .proc_handler = proc_dointvec_minmax,
2195 .extra1 = &pid_max_min,
2196 .extra2 = &pid_max_max,
2197 },
2198 {
2199 .procname = "panic_on_oops",
2200 .data = &panic_on_oops,
2201 .maxlen = sizeof(int),
2202 .mode = 0644,
2203 .proc_handler = proc_dointvec,
2204 },
2205 {
2206 .procname = "panic_print",
2207 .data = &panic_print,
2208 .maxlen = sizeof(unsigned long),
2209 .mode = 0644,
2210 .proc_handler = proc_doulongvec_minmax,
2211 },
2212#if defined CONFIG_PRINTK
2213 {
2214 .procname = "printk",
2215 .data = &console_loglevel,
2216 .maxlen = 4*sizeof(int),
2217 .mode = 0644,
2218 .proc_handler = proc_dointvec,
2219 },
2220 {
2221 .procname = "printk_ratelimit",
2222 .data = &printk_ratelimit_state.interval,
2223 .maxlen = sizeof(int),
2224 .mode = 0644,
2225 .proc_handler = proc_dointvec_jiffies,
2226 },
2227 {
2228 .procname = "printk_ratelimit_burst",
2229 .data = &printk_ratelimit_state.burst,
2230 .maxlen = sizeof(int),
2231 .mode = 0644,
2232 .proc_handler = proc_dointvec,
2233 },
2234 {
2235 .procname = "printk_delay",
2236 .data = &printk_delay_msec,
2237 .maxlen = sizeof(int),
2238 .mode = 0644,
2239 .proc_handler = proc_dointvec_minmax,
2240 .extra1 = SYSCTL_ZERO,
2241 .extra2 = &ten_thousand,
2242 },
2243 {
2244 .procname = "printk_devkmsg",
2245 .data = devkmsg_log_str,
2246 .maxlen = DEVKMSG_STR_MAX_SIZE,
2247 .mode = 0644,
2248 .proc_handler = devkmsg_sysctl_set_loglvl,
2249 },
2250 {
2251 .procname = "dmesg_restrict",
2252 .data = &dmesg_restrict,
2253 .maxlen = sizeof(int),
2254 .mode = 0644,
2255 .proc_handler = proc_dointvec_minmax_sysadmin,
2256 .extra1 = SYSCTL_ZERO,
2257 .extra2 = SYSCTL_ONE,
2258 },
2259 {
2260 .procname = "kptr_restrict",
2261 .data = &kptr_restrict,
2262 .maxlen = sizeof(int),
2263 .mode = 0644,
2264 .proc_handler = proc_dointvec_minmax_sysadmin,
2265 .extra1 = SYSCTL_ZERO,
2266 .extra2 = &two,
2267 },
2268#endif
2269 {
2270 .procname = "ngroups_max",
2271 .data = &ngroups_max,
2272 .maxlen = sizeof (int),
2273 .mode = 0444,
2274 .proc_handler = proc_dointvec,
2275 },
2276 {
2277 .procname = "cap_last_cap",
2278 .data = (void *)&cap_last_cap,
2279 .maxlen = sizeof(int),
2280 .mode = 0444,
2281 .proc_handler = proc_dointvec,
2282 },
2283#if defined(CONFIG_LOCKUP_DETECTOR)
2284 {
2285 .procname = "watchdog",
2286 .data = &watchdog_user_enabled,
2287 .maxlen = sizeof(int),
2288 .mode = 0644,
2289 .proc_handler = proc_watchdog,
2290 .extra1 = SYSCTL_ZERO,
2291 .extra2 = SYSCTL_ONE,
2292 },
2293 {
2294 .procname = "watchdog_thresh",
2295 .data = &watchdog_thresh,
2296 .maxlen = sizeof(int),
2297 .mode = 0644,
2298 .proc_handler = proc_watchdog_thresh,
2299 .extra1 = SYSCTL_ZERO,
2300 .extra2 = &sixty,
2301 },
2302 {
2303 .procname = "nmi_watchdog",
2304 .data = &nmi_watchdog_user_enabled,
2305 .maxlen = sizeof(int),
2306 .mode = NMI_WATCHDOG_SYSCTL_PERM,
2307 .proc_handler = proc_nmi_watchdog,
2308 .extra1 = SYSCTL_ZERO,
2309 .extra2 = SYSCTL_ONE,
2310 },
2311 {
2312 .procname = "watchdog_cpumask",
2313 .data = &watchdog_cpumask_bits,
2314 .maxlen = NR_CPUS,
2315 .mode = 0644,
2316 .proc_handler = proc_watchdog_cpumask,
2317 },
2318#ifdef CONFIG_SOFTLOCKUP_DETECTOR
2319 {
2320 .procname = "soft_watchdog",
2321 .data = &soft_watchdog_user_enabled,
2322 .maxlen = sizeof(int),
2323 .mode = 0644,
2324 .proc_handler = proc_soft_watchdog,
2325 .extra1 = SYSCTL_ZERO,
2326 .extra2 = SYSCTL_ONE,
2327 },
2328 {
2329 .procname = "softlockup_panic",
2330 .data = &softlockup_panic,
2331 .maxlen = sizeof(int),
2332 .mode = 0644,
2333 .proc_handler = proc_dointvec_minmax,
2334 .extra1 = SYSCTL_ZERO,
2335 .extra2 = SYSCTL_ONE,
2336 },
2337#ifdef CONFIG_SMP
2338 {
2339 .procname = "softlockup_all_cpu_backtrace",
2340 .data = &sysctl_softlockup_all_cpu_backtrace,
2341 .maxlen = sizeof(int),
2342 .mode = 0644,
2343 .proc_handler = proc_dointvec_minmax,
2344 .extra1 = SYSCTL_ZERO,
2345 .extra2 = SYSCTL_ONE,
2346 },
2347#endif /* CONFIG_SMP */
2348#endif
2349#ifdef CONFIG_HARDLOCKUP_DETECTOR
2350 {
2351 .procname = "hardlockup_panic",
2352 .data = &hardlockup_panic,
2353 .maxlen = sizeof(int),
2354 .mode = 0644,
2355 .proc_handler = proc_dointvec_minmax,
2356 .extra1 = SYSCTL_ZERO,
2357 .extra2 = SYSCTL_ONE,
2358 },
2359#ifdef CONFIG_SMP
2360 {
2361 .procname = "hardlockup_all_cpu_backtrace",
2362 .data = &sysctl_hardlockup_all_cpu_backtrace,
2363 .maxlen = sizeof(int),
2364 .mode = 0644,
2365 .proc_handler = proc_dointvec_minmax,
2366 .extra1 = SYSCTL_ZERO,
2367 .extra2 = SYSCTL_ONE,
2368 },
2369#endif /* CONFIG_SMP */
2370#endif
2371#endif
2372
2373#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
2374 {
2375 .procname = "unknown_nmi_panic",
2376 .data = &unknown_nmi_panic,
2377 .maxlen = sizeof (int),
2378 .mode = 0644,
2379 .proc_handler = proc_dointvec,
2380 },
2381#endif
2382
2383#if (defined(CONFIG_X86_32) || defined(CONFIG_PARISC)) && \
2384 defined(CONFIG_DEBUG_STACKOVERFLOW)
2385 {
2386 .procname = "panic_on_stackoverflow",
2387 .data = &sysctl_panic_on_stackoverflow,
2388 .maxlen = sizeof(int),
2389 .mode = 0644,
2390 .proc_handler = proc_dointvec,
2391 },
2392#endif
2393#if defined(CONFIG_X86)
2394 {
2395 .procname = "panic_on_unrecovered_nmi",
2396 .data = &panic_on_unrecovered_nmi,
2397 .maxlen = sizeof(int),
2398 .mode = 0644,
2399 .proc_handler = proc_dointvec,
2400 },
2401 {
2402 .procname = "panic_on_io_nmi",
2403 .data = &panic_on_io_nmi,
2404 .maxlen = sizeof(int),
2405 .mode = 0644,
2406 .proc_handler = proc_dointvec,
2407 },
2408 {
2409 .procname = "bootloader_type",
2410 .data = &bootloader_type,
2411 .maxlen = sizeof (int),
2412 .mode = 0444,
2413 .proc_handler = proc_dointvec,
2414 },
2415 {
2416 .procname = "bootloader_version",
2417 .data = &bootloader_version,
2418 .maxlen = sizeof (int),
2419 .mode = 0444,
2420 .proc_handler = proc_dointvec,
2421 },
2422 {
2423 .procname = "io_delay_type",
2424 .data = &io_delay_type,
2425 .maxlen = sizeof(int),
2426 .mode = 0644,
2427 .proc_handler = proc_dointvec,
2428 },
2429#endif
2430#if defined(CONFIG_MMU)
2431 {
2432 .procname = "randomize_va_space",
2433 .data = &randomize_va_space,
2434 .maxlen = sizeof(int),
2435 .mode = 0644,
2436 .proc_handler = proc_dointvec,
2437 },
2438#endif
2439#if defined(CONFIG_S390) && defined(CONFIG_SMP)
2440 {
2441 .procname = "spin_retry",
2442 .data = &spin_retry,
2443 .maxlen = sizeof (int),
2444 .mode = 0644,
2445 .proc_handler = proc_dointvec,
2446 },
2447#endif
2448#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
2449 {
2450 .procname = "acpi_video_flags",
2451 .data = &acpi_realmode_flags,
2452 .maxlen = sizeof (unsigned long),
2453 .mode = 0644,
2454 .proc_handler = proc_doulongvec_minmax,
2455 },
2456#endif
2457#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
2458 {
2459 .procname = "ignore-unaligned-usertrap",
2460 .data = &no_unaligned_warning,
2461 .maxlen = sizeof (int),
2462 .mode = 0644,
2463 .proc_handler = proc_dointvec,
2464 },
2465#endif
2466#ifdef CONFIG_IA64
2467 {
2468 .procname = "unaligned-dump-stack",
2469 .data = &unaligned_dump_stack,
2470 .maxlen = sizeof (int),
2471 .mode = 0644,
2472 .proc_handler = proc_dointvec,
2473 },
2474#endif
2475#ifdef CONFIG_DETECT_HUNG_TASK
2476#ifdef CONFIG_SMP
2477 {
2478 .procname = "hung_task_all_cpu_backtrace",
2479 .data = &sysctl_hung_task_all_cpu_backtrace,
2480 .maxlen = sizeof(int),
2481 .mode = 0644,
2482 .proc_handler = proc_dointvec_minmax,
2483 .extra1 = SYSCTL_ZERO,
2484 .extra2 = SYSCTL_ONE,
2485 },
2486#endif /* CONFIG_SMP */
2487 {
2488 .procname = "hung_task_panic",
2489 .data = &sysctl_hung_task_panic,
2490 .maxlen = sizeof(int),
2491 .mode = 0644,
2492 .proc_handler = proc_dointvec_minmax,
2493 .extra1 = SYSCTL_ZERO,
2494 .extra2 = SYSCTL_ONE,
2495 },
2496 {
2497 .procname = "hung_task_check_count",
2498 .data = &sysctl_hung_task_check_count,
2499 .maxlen = sizeof(int),
2500 .mode = 0644,
2501 .proc_handler = proc_dointvec_minmax,
2502 .extra1 = SYSCTL_ZERO,
2503 },
2504 {
2505 .procname = "hung_task_timeout_secs",
2506 .data = &sysctl_hung_task_timeout_secs,
2507 .maxlen = sizeof(unsigned long),
2508 .mode = 0644,
2509 .proc_handler = proc_dohung_task_timeout_secs,
2510 .extra2 = &hung_task_timeout_max,
2511 },
2512 {
2513 .procname = "hung_task_check_interval_secs",
2514 .data = &sysctl_hung_task_check_interval_secs,
2515 .maxlen = sizeof(unsigned long),
2516 .mode = 0644,
2517 .proc_handler = proc_dohung_task_timeout_secs,
2518 .extra2 = &hung_task_timeout_max,
2519 },
2520 {
2521 .procname = "hung_task_warnings",
2522 .data = &sysctl_hung_task_warnings,
2523 .maxlen = sizeof(int),
2524 .mode = 0644,
2525 .proc_handler = proc_dointvec_minmax,
2526 .extra1 = &neg_one,
2527 },
2528#endif
2529#ifdef CONFIG_RT_MUTEXES
2530 {
2531 .procname = "max_lock_depth",
2532 .data = &max_lock_depth,
2533 .maxlen = sizeof(int),
2534 .mode = 0644,
2535 .proc_handler = proc_dointvec,
2536 },
2537#endif
2538 {
2539 .procname = "poweroff_cmd",
2540 .data = &poweroff_cmd,
2541 .maxlen = POWEROFF_CMD_PATH_LEN,
2542 .mode = 0644,
2543 .proc_handler = proc_dostring,
2544 },
2545#ifdef CONFIG_KEYS
2546 {
2547 .procname = "keys",
2548 .mode = 0555,
2549 .child = key_sysctls,
2550 },
2551#endif
2552#ifdef CONFIG_PERF_EVENTS
2553 /*
2554 * User-space scripts rely on the existence of this file
2555 * as a feature check for perf_events being enabled.
2556 *
2557 * So it's an ABI, do not remove!
2558 */
2559 {
2560 .procname = "perf_event_paranoid",
2561 .data = &sysctl_perf_event_paranoid,
2562 .maxlen = sizeof(sysctl_perf_event_paranoid),
2563 .mode = 0644,
2564 .proc_handler = proc_dointvec,
2565 },
2566 {
2567 .procname = "perf_event_mlock_kb",
2568 .data = &sysctl_perf_event_mlock,
2569 .maxlen = sizeof(sysctl_perf_event_mlock),
2570 .mode = 0644,
2571 .proc_handler = proc_dointvec,
2572 },
2573 {
2574 .procname = "perf_event_max_sample_rate",
2575 .data = &sysctl_perf_event_sample_rate,
2576 .maxlen = sizeof(sysctl_perf_event_sample_rate),
2577 .mode = 0644,
2578 .proc_handler = perf_proc_update_handler,
2579 .extra1 = SYSCTL_ONE,
2580 },
2581 {
2582 .procname = "perf_cpu_time_max_percent",
2583 .data = &sysctl_perf_cpu_time_max_percent,
2584 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
2585 .mode = 0644,
2586 .proc_handler = perf_cpu_time_max_percent_handler,
2587 .extra1 = SYSCTL_ZERO,
2588 .extra2 = &one_hundred,
2589 },
2590 {
2591 .procname = "perf_event_max_stack",
2592 .data = &sysctl_perf_event_max_stack,
2593 .maxlen = sizeof(sysctl_perf_event_max_stack),
2594 .mode = 0644,
2595 .proc_handler = perf_event_max_stack_handler,
2596 .extra1 = SYSCTL_ZERO,
2597 .extra2 = &six_hundred_forty_kb,
2598 },
2599 {
2600 .procname = "perf_event_max_contexts_per_stack",
2601 .data = &sysctl_perf_event_max_contexts_per_stack,
2602 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
2603 .mode = 0644,
2604 .proc_handler = perf_event_max_stack_handler,
2605 .extra1 = SYSCTL_ZERO,
2606 .extra2 = &one_thousand,
2607 },
2608#endif
2609 {
2610 .procname = "panic_on_warn",
2611 .data = &panic_on_warn,
2612 .maxlen = sizeof(int),
2613 .mode = 0644,
2614 .proc_handler = proc_dointvec_minmax,
2615 .extra1 = SYSCTL_ZERO,
2616 .extra2 = SYSCTL_ONE,
2617 },
2618#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
2619 {
2620 .procname = "timer_migration",
2621 .data = &sysctl_timer_migration,
2622 .maxlen = sizeof(unsigned int),
2623 .mode = 0644,
2624 .proc_handler = timer_migration_handler,
2625 .extra1 = SYSCTL_ZERO,
2626 .extra2 = SYSCTL_ONE,
2627 },
2628#endif
2629#ifdef CONFIG_BPF_SYSCALL
2630 {
2631 .procname = "unprivileged_bpf_disabled",
2632 .data = &sysctl_unprivileged_bpf_disabled,
2633 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
2634 .mode = 0644,
2635 .proc_handler = bpf_unpriv_handler,
2636 .extra1 = SYSCTL_ZERO,
2637 .extra2 = &two,
2638 },
2639 {
2640 .procname = "bpf_stats_enabled",
2641 .data = &bpf_stats_enabled_key.key,
2642 .maxlen = sizeof(bpf_stats_enabled_key),
2643 .mode = 0644,
2644 .proc_handler = bpf_stats_handler,
2645 },
2646#endif
2647#if defined(CONFIG_TREE_RCU)
2648 {
2649 .procname = "panic_on_rcu_stall",
2650 .data = &sysctl_panic_on_rcu_stall,
2651 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
2652 .mode = 0644,
2653 .proc_handler = proc_dointvec_minmax,
2654 .extra1 = SYSCTL_ZERO,
2655 .extra2 = SYSCTL_ONE,
2656 },
2657#endif
2658#if defined(CONFIG_TREE_RCU)
2659 {
2660 .procname = "max_rcu_stall_to_panic",
2661 .data = &sysctl_max_rcu_stall_to_panic,
2662 .maxlen = sizeof(sysctl_max_rcu_stall_to_panic),
2663 .mode = 0644,
2664 .proc_handler = proc_dointvec_minmax,
2665 .extra1 = SYSCTL_ONE,
2666 .extra2 = SYSCTL_INT_MAX,
2667 },
2668#endif
2669#ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
2670 {
2671 .procname = "stack_erasing",
2672 .data = NULL,
2673 .maxlen = sizeof(int),
2674 .mode = 0600,
2675 .proc_handler = stack_erasing_sysctl,
2676 .extra1 = SYSCTL_ZERO,
2677 .extra2 = SYSCTL_ONE,
2678 },
2679#endif
2680 { }
2681};
2682
2683static struct ctl_table vm_table[] = {
2684 {
2685 .procname = "overcommit_memory",
2686 .data = &sysctl_overcommit_memory,
2687 .maxlen = sizeof(sysctl_overcommit_memory),
2688 .mode = 0644,
2689 .proc_handler = overcommit_policy_handler,
2690 .extra1 = SYSCTL_ZERO,
2691 .extra2 = &two,
2692 },
2693 {
2694 .procname = "panic_on_oom",
2695 .data = &sysctl_panic_on_oom,
2696 .maxlen = sizeof(sysctl_panic_on_oom),
2697 .mode = 0644,
2698 .proc_handler = proc_dointvec_minmax,
2699 .extra1 = SYSCTL_ZERO,
2700 .extra2 = &two,
2701 },
2702 {
2703 .procname = "oom_kill_allocating_task",
2704 .data = &sysctl_oom_kill_allocating_task,
2705 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
2706 .mode = 0644,
2707 .proc_handler = proc_dointvec,
2708 },
2709 {
2710 .procname = "oom_dump_tasks",
2711 .data = &sysctl_oom_dump_tasks,
2712 .maxlen = sizeof(sysctl_oom_dump_tasks),
2713 .mode = 0644,
2714 .proc_handler = proc_dointvec,
2715 },
2716 {
2717 .procname = "overcommit_ratio",
2718 .data = &sysctl_overcommit_ratio,
2719 .maxlen = sizeof(sysctl_overcommit_ratio),
2720 .mode = 0644,
2721 .proc_handler = overcommit_ratio_handler,
2722 },
2723 {
2724 .procname = "overcommit_kbytes",
2725 .data = &sysctl_overcommit_kbytes,
2726 .maxlen = sizeof(sysctl_overcommit_kbytes),
2727 .mode = 0644,
2728 .proc_handler = overcommit_kbytes_handler,
2729 },
2730 {
2731 .procname = "page-cluster",
2732 .data = &page_cluster,
2733 .maxlen = sizeof(int),
2734 .mode = 0644,
2735 .proc_handler = proc_dointvec_minmax,
2736 .extra1 = SYSCTL_ZERO,
2737 },
2738 {
2739 .procname = "dirty_background_ratio",
2740 .data = &dirty_background_ratio,
2741 .maxlen = sizeof(dirty_background_ratio),
2742 .mode = 0644,
2743 .proc_handler = dirty_background_ratio_handler,
2744 .extra1 = SYSCTL_ZERO,
2745 .extra2 = &one_hundred,
2746 },
2747 {
2748 .procname = "dirty_background_bytes",
2749 .data = &dirty_background_bytes,
2750 .maxlen = sizeof(dirty_background_bytes),
2751 .mode = 0644,
2752 .proc_handler = dirty_background_bytes_handler,
2753 .extra1 = &one_ul,
2754 },
2755 {
2756 .procname = "dirty_ratio",
2757 .data = &vm_dirty_ratio,
2758 .maxlen = sizeof(vm_dirty_ratio),
2759 .mode = 0644,
2760 .proc_handler = dirty_ratio_handler,
2761 .extra1 = SYSCTL_ZERO,
2762 .extra2 = &one_hundred,
2763 },
2764 {
2765 .procname = "dirty_bytes",
2766 .data = &vm_dirty_bytes,
2767 .maxlen = sizeof(vm_dirty_bytes),
2768 .mode = 0644,
2769 .proc_handler = dirty_bytes_handler,
2770 .extra1 = &dirty_bytes_min,
2771 },
2772 {
2773 .procname = "dirty_writeback_centisecs",
2774 .data = &dirty_writeback_interval,
2775 .maxlen = sizeof(dirty_writeback_interval),
2776 .mode = 0644,
2777 .proc_handler = dirty_writeback_centisecs_handler,
2778 },
2779 {
2780 .procname = "dirty_expire_centisecs",
2781 .data = &dirty_expire_interval,
2782 .maxlen = sizeof(dirty_expire_interval),
2783 .mode = 0644,
2784 .proc_handler = proc_dointvec_minmax,
2785 .extra1 = SYSCTL_ZERO,
2786 },
2787 {
2788 .procname = "dirtytime_expire_seconds",
2789 .data = &dirtytime_expire_interval,
2790 .maxlen = sizeof(dirtytime_expire_interval),
2791 .mode = 0644,
2792 .proc_handler = dirtytime_interval_handler,
2793 .extra1 = SYSCTL_ZERO,
2794 },
2795 {
2796 .procname = "swappiness",
2797 .data = &vm_swappiness,
2798 .maxlen = sizeof(vm_swappiness),
2799 .mode = 0644,
2800 .proc_handler = proc_dointvec_minmax,
2801 .extra1 = SYSCTL_ZERO,
2802 .extra2 = &two_hundred,
2803 },
2804#ifdef CONFIG_HUGETLB_PAGE
2805 {
2806 .procname = "nr_hugepages",
2807 .data = NULL,
2808 .maxlen = sizeof(unsigned long),
2809 .mode = 0644,
2810 .proc_handler = hugetlb_sysctl_handler,
2811 },
2812#ifdef CONFIG_NUMA
2813 {
2814 .procname = "nr_hugepages_mempolicy",
2815 .data = NULL,
2816 .maxlen = sizeof(unsigned long),
2817 .mode = 0644,
2818 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
2819 },
2820 {
2821 .procname = "numa_stat",
2822 .data = &sysctl_vm_numa_stat,
2823 .maxlen = sizeof(int),
2824 .mode = 0644,
2825 .proc_handler = sysctl_vm_numa_stat_handler,
2826 .extra1 = SYSCTL_ZERO,
2827 .extra2 = SYSCTL_ONE,
2828 },
2829#endif
2830 {
2831 .procname = "hugetlb_shm_group",
2832 .data = &sysctl_hugetlb_shm_group,
2833 .maxlen = sizeof(gid_t),
2834 .mode = 0644,
2835 .proc_handler = proc_dointvec,
2836 },
2837 {
2838 .procname = "nr_overcommit_hugepages",
2839 .data = NULL,
2840 .maxlen = sizeof(unsigned long),
2841 .mode = 0644,
2842 .proc_handler = hugetlb_overcommit_handler,
2843 },
2844#endif
2845 {
2846 .procname = "lowmem_reserve_ratio",
2847 .data = &sysctl_lowmem_reserve_ratio,
2848 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
2849 .mode = 0644,
2850 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
2851 },
2852 {
2853 .procname = "drop_caches",
2854 .data = &sysctl_drop_caches,
2855 .maxlen = sizeof(int),
2856 .mode = 0200,
2857 .proc_handler = drop_caches_sysctl_handler,
2858 .extra1 = SYSCTL_ONE,
2859 .extra2 = &four,
2860 },
2861#ifdef CONFIG_COMPACTION
2862 {
2863 .procname = "compact_memory",
2864 .data = NULL,
2865 .maxlen = sizeof(int),
2866 .mode = 0200,
2867 .proc_handler = sysctl_compaction_handler,
2868 },
2869 {
2870 .procname = "compaction_proactiveness",
2871 .data = &sysctl_compaction_proactiveness,
2872 .maxlen = sizeof(sysctl_compaction_proactiveness),
2873 .mode = 0644,
2874 .proc_handler = proc_dointvec_minmax,
2875 .extra1 = SYSCTL_ZERO,
2876 .extra2 = &one_hundred,
2877 },
2878 {
2879 .procname = "extfrag_threshold",
2880 .data = &sysctl_extfrag_threshold,
2881 .maxlen = sizeof(int),
2882 .mode = 0644,
2883 .proc_handler = proc_dointvec_minmax,
2884 .extra1 = &min_extfrag_threshold,
2885 .extra2 = &max_extfrag_threshold,
2886 },
2887 {
2888 .procname = "compact_unevictable_allowed",
2889 .data = &sysctl_compact_unevictable_allowed,
2890 .maxlen = sizeof(int),
2891 .mode = 0644,
2892 .proc_handler = proc_dointvec_minmax_warn_RT_change,
2893 .extra1 = SYSCTL_ZERO,
2894 .extra2 = SYSCTL_ONE,
2895 },
2896
2897#endif /* CONFIG_COMPACTION */
2898 {
2899 .procname = "min_free_kbytes",
2900 .data = &min_free_kbytes,
2901 .maxlen = sizeof(min_free_kbytes),
2902 .mode = 0644,
2903 .proc_handler = min_free_kbytes_sysctl_handler,
2904 .extra1 = SYSCTL_ZERO,
2905 },
2906 {
2907 .procname = "watermark_boost_factor",
2908 .data = &watermark_boost_factor,
2909 .maxlen = sizeof(watermark_boost_factor),
2910 .mode = 0644,
2911 .proc_handler = proc_dointvec_minmax,
2912 .extra1 = SYSCTL_ZERO,
2913 },
2914 {
2915 .procname = "watermark_scale_factor",
2916 .data = &watermark_scale_factor,
2917 .maxlen = sizeof(watermark_scale_factor),
2918 .mode = 0644,
2919 .proc_handler = watermark_scale_factor_sysctl_handler,
2920 .extra1 = SYSCTL_ONE,
2921 .extra2 = &one_thousand,
2922 },
2923 {
2924 .procname = "percpu_pagelist_high_fraction",
2925 .data = &percpu_pagelist_high_fraction,
2926 .maxlen = sizeof(percpu_pagelist_high_fraction),
2927 .mode = 0644,
2928 .proc_handler = percpu_pagelist_high_fraction_sysctl_handler,
2929 .extra1 = SYSCTL_ZERO,
2930 },
2931 {
2932 .procname = "page_lock_unfairness",
2933 .data = &sysctl_page_lock_unfairness,
2934 .maxlen = sizeof(sysctl_page_lock_unfairness),
2935 .mode = 0644,
2936 .proc_handler = proc_dointvec_minmax,
2937 .extra1 = SYSCTL_ZERO,
2938 },
2939#ifdef CONFIG_MMU
2940 {
2941 .procname = "max_map_count",
2942 .data = &sysctl_max_map_count,
2943 .maxlen = sizeof(sysctl_max_map_count),
2944 .mode = 0644,
2945 .proc_handler = proc_dointvec_minmax,
2946 .extra1 = SYSCTL_ZERO,
2947 },
2948#else
2949 {
2950 .procname = "nr_trim_pages",
2951 .data = &sysctl_nr_trim_pages,
2952 .maxlen = sizeof(sysctl_nr_trim_pages),
2953 .mode = 0644,
2954 .proc_handler = proc_dointvec_minmax,
2955 .extra1 = SYSCTL_ZERO,
2956 },
2957#endif
2958 {
2959 .procname = "laptop_mode",
2960 .data = &laptop_mode,
2961 .maxlen = sizeof(laptop_mode),
2962 .mode = 0644,
2963 .proc_handler = proc_dointvec_jiffies,
2964 },
2965 {
2966 .procname = "vfs_cache_pressure",
2967 .data = &sysctl_vfs_cache_pressure,
2968 .maxlen = sizeof(sysctl_vfs_cache_pressure),
2969 .mode = 0644,
2970 .proc_handler = proc_dointvec_minmax,
2971 .extra1 = SYSCTL_ZERO,
2972 },
2973#if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
2974 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
2975 {
2976 .procname = "legacy_va_layout",
2977 .data = &sysctl_legacy_va_layout,
2978 .maxlen = sizeof(sysctl_legacy_va_layout),
2979 .mode = 0644,
2980 .proc_handler = proc_dointvec_minmax,
2981 .extra1 = SYSCTL_ZERO,
2982 },
2983#endif
2984#ifdef CONFIG_NUMA
2985 {
2986 .procname = "zone_reclaim_mode",
2987 .data = &node_reclaim_mode,
2988 .maxlen = sizeof(node_reclaim_mode),
2989 .mode = 0644,
2990 .proc_handler = proc_dointvec_minmax,
2991 .extra1 = SYSCTL_ZERO,
2992 },
2993 {
2994 .procname = "min_unmapped_ratio",
2995 .data = &sysctl_min_unmapped_ratio,
2996 .maxlen = sizeof(sysctl_min_unmapped_ratio),
2997 .mode = 0644,
2998 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
2999 .extra1 = SYSCTL_ZERO,
3000 .extra2 = &one_hundred,
3001 },
3002 {
3003 .procname = "min_slab_ratio",
3004 .data = &sysctl_min_slab_ratio,
3005 .maxlen = sizeof(sysctl_min_slab_ratio),
3006 .mode = 0644,
3007 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
3008 .extra1 = SYSCTL_ZERO,
3009 .extra2 = &one_hundred,
3010 },
3011#endif
3012#ifdef CONFIG_SMP
3013 {
3014 .procname = "stat_interval",
3015 .data = &sysctl_stat_interval,
3016 .maxlen = sizeof(sysctl_stat_interval),
3017 .mode = 0644,
3018 .proc_handler = proc_dointvec_jiffies,
3019 },
3020 {
3021 .procname = "stat_refresh",
3022 .data = NULL,
3023 .maxlen = 0,
3024 .mode = 0600,
3025 .proc_handler = vmstat_refresh,
3026 },
3027#endif
3028#ifdef CONFIG_MMU
3029 {
3030 .procname = "mmap_min_addr",
3031 .data = &dac_mmap_min_addr,
3032 .maxlen = sizeof(unsigned long),
3033 .mode = 0644,
3034 .proc_handler = mmap_min_addr_handler,
3035 },
3036#endif
3037#ifdef CONFIG_NUMA
3038 {
3039 .procname = "numa_zonelist_order",
3040 .data = &numa_zonelist_order,
3041 .maxlen = NUMA_ZONELIST_ORDER_LEN,
3042 .mode = 0644,
3043 .proc_handler = numa_zonelist_order_handler,
3044 },
3045#endif
3046#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
3047 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
3048 {
3049 .procname = "vdso_enabled",
3050#ifdef CONFIG_X86_32
3051 .data = &vdso32_enabled,
3052 .maxlen = sizeof(vdso32_enabled),
3053#else
3054 .data = &vdso_enabled,
3055 .maxlen = sizeof(vdso_enabled),
3056#endif
3057 .mode = 0644,
3058 .proc_handler = proc_dointvec,
3059 .extra1 = SYSCTL_ZERO,
3060 },
3061#endif
3062#ifdef CONFIG_HIGHMEM
3063 {
3064 .procname = "highmem_is_dirtyable",
3065 .data = &vm_highmem_is_dirtyable,
3066 .maxlen = sizeof(vm_highmem_is_dirtyable),
3067 .mode = 0644,
3068 .proc_handler = proc_dointvec_minmax,
3069 .extra1 = SYSCTL_ZERO,
3070 .extra2 = SYSCTL_ONE,
3071 },
3072#endif
3073#ifdef CONFIG_MEMORY_FAILURE
3074 {
3075 .procname = "memory_failure_early_kill",
3076 .data = &sysctl_memory_failure_early_kill,
3077 .maxlen = sizeof(sysctl_memory_failure_early_kill),
3078 .mode = 0644,
3079 .proc_handler = proc_dointvec_minmax,
3080 .extra1 = SYSCTL_ZERO,
3081 .extra2 = SYSCTL_ONE,
3082 },
3083 {
3084 .procname = "memory_failure_recovery",
3085 .data = &sysctl_memory_failure_recovery,
3086 .maxlen = sizeof(sysctl_memory_failure_recovery),
3087 .mode = 0644,
3088 .proc_handler = proc_dointvec_minmax,
3089 .extra1 = SYSCTL_ZERO,
3090 .extra2 = SYSCTL_ONE,
3091 },
3092#endif
3093 {
3094 .procname = "user_reserve_kbytes",
3095 .data = &sysctl_user_reserve_kbytes,
3096 .maxlen = sizeof(sysctl_user_reserve_kbytes),
3097 .mode = 0644,
3098 .proc_handler = proc_doulongvec_minmax,
3099 },
3100 {
3101 .procname = "admin_reserve_kbytes",
3102 .data = &sysctl_admin_reserve_kbytes,
3103 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
3104 .mode = 0644,
3105 .proc_handler = proc_doulongvec_minmax,
3106 },
3107#ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
3108 {
3109 .procname = "mmap_rnd_bits",
3110 .data = &mmap_rnd_bits,
3111 .maxlen = sizeof(mmap_rnd_bits),
3112 .mode = 0600,
3113 .proc_handler = proc_dointvec_minmax,
3114 .extra1 = (void *)&mmap_rnd_bits_min,
3115 .extra2 = (void *)&mmap_rnd_bits_max,
3116 },
3117#endif
3118#ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
3119 {
3120 .procname = "mmap_rnd_compat_bits",
3121 .data = &mmap_rnd_compat_bits,
3122 .maxlen = sizeof(mmap_rnd_compat_bits),
3123 .mode = 0600,
3124 .proc_handler = proc_dointvec_minmax,
3125 .extra1 = (void *)&mmap_rnd_compat_bits_min,
3126 .extra2 = (void *)&mmap_rnd_compat_bits_max,
3127 },
3128#endif
3129#ifdef CONFIG_USERFAULTFD
3130 {
3131 .procname = "unprivileged_userfaultfd",
3132 .data = &sysctl_unprivileged_userfaultfd,
3133 .maxlen = sizeof(sysctl_unprivileged_userfaultfd),
3134 .mode = 0644,
3135 .proc_handler = proc_dointvec_minmax,
3136 .extra1 = SYSCTL_ZERO,
3137 .extra2 = SYSCTL_ONE,
3138 },
3139#endif
3140 { }
3141};
3142
3143static struct ctl_table fs_table[] = {
3144 {
3145 .procname = "inode-nr",
3146 .data = &inodes_stat,
3147 .maxlen = 2*sizeof(long),
3148 .mode = 0444,
3149 .proc_handler = proc_nr_inodes,
3150 },
3151 {
3152 .procname = "inode-state",
3153 .data = &inodes_stat,
3154 .maxlen = 7*sizeof(long),
3155 .mode = 0444,
3156 .proc_handler = proc_nr_inodes,
3157 },
3158 {
3159 .procname = "file-nr",
3160 .data = &files_stat,
3161 .maxlen = sizeof(files_stat),
3162 .mode = 0444,
3163 .proc_handler = proc_nr_files,
3164 },
3165 {
3166 .procname = "file-max",
3167 .data = &files_stat.max_files,
3168 .maxlen = sizeof(files_stat.max_files),
3169 .mode = 0644,
3170 .proc_handler = proc_doulongvec_minmax,
3171 .extra1 = &zero_ul,
3172 .extra2 = &long_max,
3173 },
3174 {
3175 .procname = "nr_open",
3176 .data = &sysctl_nr_open,
3177 .maxlen = sizeof(unsigned int),
3178 .mode = 0644,
3179 .proc_handler = proc_dointvec_minmax,
3180 .extra1 = &sysctl_nr_open_min,
3181 .extra2 = &sysctl_nr_open_max,
3182 },
3183 {
3184 .procname = "dentry-state",
3185 .data = &dentry_stat,
3186 .maxlen = 6*sizeof(long),
3187 .mode = 0444,
3188 .proc_handler = proc_nr_dentry,
3189 },
3190 {
3191 .procname = "overflowuid",
3192 .data = &fs_overflowuid,
3193 .maxlen = sizeof(int),
3194 .mode = 0644,
3195 .proc_handler = proc_dointvec_minmax,
3196 .extra1 = &minolduid,
3197 .extra2 = &maxolduid,
3198 },
3199 {
3200 .procname = "overflowgid",
3201 .data = &fs_overflowgid,
3202 .maxlen = sizeof(int),
3203 .mode = 0644,
3204 .proc_handler = proc_dointvec_minmax,
3205 .extra1 = &minolduid,
3206 .extra2 = &maxolduid,
3207 },
3208#ifdef CONFIG_FILE_LOCKING
3209 {
3210 .procname = "leases-enable",
3211 .data = &leases_enable,
3212 .maxlen = sizeof(int),
3213 .mode = 0644,
3214 .proc_handler = proc_dointvec,
3215 },
3216#endif
3217#ifdef CONFIG_DNOTIFY
3218 {
3219 .procname = "dir-notify-enable",
3220 .data = &dir_notify_enable,
3221 .maxlen = sizeof(int),
3222 .mode = 0644,
3223 .proc_handler = proc_dointvec,
3224 },
3225#endif
3226#ifdef CONFIG_MMU
3227#ifdef CONFIG_FILE_LOCKING
3228 {
3229 .procname = "lease-break-time",
3230 .data = &lease_break_time,
3231 .maxlen = sizeof(int),
3232 .mode = 0644,
3233 .proc_handler = proc_dointvec,
3234 },
3235#endif
3236#ifdef CONFIG_AIO
3237 {
3238 .procname = "aio-nr",
3239 .data = &aio_nr,
3240 .maxlen = sizeof(aio_nr),
3241 .mode = 0444,
3242 .proc_handler = proc_doulongvec_minmax,
3243 },
3244 {
3245 .procname = "aio-max-nr",
3246 .data = &aio_max_nr,
3247 .maxlen = sizeof(aio_max_nr),
3248 .mode = 0644,
3249 .proc_handler = proc_doulongvec_minmax,
3250 },
3251#endif /* CONFIG_AIO */
3252#ifdef CONFIG_INOTIFY_USER
3253 {
3254 .procname = "inotify",
3255 .mode = 0555,
3256 .child = inotify_table,
3257 },
3258#endif
3259#ifdef CONFIG_FANOTIFY
3260 {
3261 .procname = "fanotify",
3262 .mode = 0555,
3263 .child = fanotify_table,
3264 },
3265#endif
3266#ifdef CONFIG_EPOLL
3267 {
3268 .procname = "epoll",
3269 .mode = 0555,
3270 .child = epoll_table,
3271 },
3272#endif
3273#endif
3274 {
3275 .procname = "protected_symlinks",
3276 .data = &sysctl_protected_symlinks,
3277 .maxlen = sizeof(int),
3278 .mode = 0600,
3279 .proc_handler = proc_dointvec_minmax,
3280 .extra1 = SYSCTL_ZERO,
3281 .extra2 = SYSCTL_ONE,
3282 },
3283 {
3284 .procname = "protected_hardlinks",
3285 .data = &sysctl_protected_hardlinks,
3286 .maxlen = sizeof(int),
3287 .mode = 0600,
3288 .proc_handler = proc_dointvec_minmax,
3289 .extra1 = SYSCTL_ZERO,
3290 .extra2 = SYSCTL_ONE,
3291 },
3292 {
3293 .procname = "protected_fifos",
3294 .data = &sysctl_protected_fifos,
3295 .maxlen = sizeof(int),
3296 .mode = 0600,
3297 .proc_handler = proc_dointvec_minmax,
3298 .extra1 = SYSCTL_ZERO,
3299 .extra2 = &two,
3300 },
3301 {
3302 .procname = "protected_regular",
3303 .data = &sysctl_protected_regular,
3304 .maxlen = sizeof(int),
3305 .mode = 0600,
3306 .proc_handler = proc_dointvec_minmax,
3307 .extra1 = SYSCTL_ZERO,
3308 .extra2 = &two,
3309 },
3310 {
3311 .procname = "suid_dumpable",
3312 .data = &suid_dumpable,
3313 .maxlen = sizeof(int),
3314 .mode = 0644,
3315 .proc_handler = proc_dointvec_minmax_coredump,
3316 .extra1 = SYSCTL_ZERO,
3317 .extra2 = &two,
3318 },
3319#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
3320 {
3321 .procname = "binfmt_misc",
3322 .mode = 0555,
3323 .child = sysctl_mount_point,
3324 },
3325#endif
3326 {
3327 .procname = "pipe-max-size",
3328 .data = &pipe_max_size,
3329 .maxlen = sizeof(pipe_max_size),
3330 .mode = 0644,
3331 .proc_handler = proc_dopipe_max_size,
3332 },
3333 {
3334 .procname = "pipe-user-pages-hard",
3335 .data = &pipe_user_pages_hard,
3336 .maxlen = sizeof(pipe_user_pages_hard),
3337 .mode = 0644,
3338 .proc_handler = proc_doulongvec_minmax,
3339 },
3340 {
3341 .procname = "pipe-user-pages-soft",
3342 .data = &pipe_user_pages_soft,
3343 .maxlen = sizeof(pipe_user_pages_soft),
3344 .mode = 0644,
3345 .proc_handler = proc_doulongvec_minmax,
3346 },
3347 {
3348 .procname = "mount-max",
3349 .data = &sysctl_mount_max,
3350 .maxlen = sizeof(unsigned int),
3351 .mode = 0644,
3352 .proc_handler = proc_dointvec_minmax,
3353 .extra1 = SYSCTL_ONE,
3354 },
3355 { }
3356};
3357
3358static struct ctl_table debug_table[] = {
3359#ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
3360 {
3361 .procname = "exception-trace",
3362 .data = &show_unhandled_signals,
3363 .maxlen = sizeof(int),
3364 .mode = 0644,
3365 .proc_handler = proc_dointvec
3366 },
3367#endif
3368#if defined(CONFIG_OPTPROBES)
3369 {
3370 .procname = "kprobes-optimization",
3371 .data = &sysctl_kprobes_optimization,
3372 .maxlen = sizeof(int),
3373 .mode = 0644,
3374 .proc_handler = proc_kprobes_optimization_handler,
3375 .extra1 = SYSCTL_ZERO,
3376 .extra2 = SYSCTL_ONE,
3377 },
3378#endif
3379 { }
3380};
3381
3382static struct ctl_table dev_table[] = {
3383 { }
3384};
3385
3386static struct ctl_table sysctl_base_table[] = {
3387 {
3388 .procname = "kernel",
3389 .mode = 0555,
3390 .child = kern_table,
3391 },
3392 {
3393 .procname = "vm",
3394 .mode = 0555,
3395 .child = vm_table,
3396 },
3397 {
3398 .procname = "fs",
3399 .mode = 0555,
3400 .child = fs_table,
3401 },
3402 {
3403 .procname = "debug",
3404 .mode = 0555,
3405 .child = debug_table,
3406 },
3407 {
3408 .procname = "dev",
3409 .mode = 0555,
3410 .child = dev_table,
3411 },
3412 { }
3413};
3414
3415int __init sysctl_init(void)
3416{
3417 struct ctl_table_header *hdr;
3418
3419 hdr = register_sysctl_table(sysctl_base_table);
3420 kmemleak_not_leak(hdr);
3421 return 0;
3422}
3423#endif /* CONFIG_SYSCTL */
3424/*
3425 * No sense putting this after each symbol definition, twice,
3426 * exception granted :-)
3427 */
3428EXPORT_SYMBOL(proc_dointvec);
3429EXPORT_SYMBOL(proc_douintvec);
3430EXPORT_SYMBOL(proc_dointvec_jiffies);
3431EXPORT_SYMBOL(proc_dointvec_minmax);
3432EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3433EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3434EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3435EXPORT_SYMBOL(proc_dostring);
3436EXPORT_SYMBOL(proc_doulongvec_minmax);
3437EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3438EXPORT_SYMBOL(proc_do_large_bitmap);
1/*
2 * sysctl.c: General linux system control interface
3 *
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12 * Horn.
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16 * Wendling.
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19 */
20
21#include <linux/module.h>
22#include <linux/mm.h>
23#include <linux/swap.h>
24#include <linux/slab.h>
25#include <linux/sysctl.h>
26#include <linux/signal.h>
27#include <linux/printk.h>
28#include <linux/proc_fs.h>
29#include <linux/security.h>
30#include <linux/ctype.h>
31#include <linux/kmemcheck.h>
32#include <linux/fs.h>
33#include <linux/init.h>
34#include <linux/kernel.h>
35#include <linux/kobject.h>
36#include <linux/net.h>
37#include <linux/sysrq.h>
38#include <linux/highuid.h>
39#include <linux/writeback.h>
40#include <linux/ratelimit.h>
41#include <linux/compaction.h>
42#include <linux/hugetlb.h>
43#include <linux/initrd.h>
44#include <linux/key.h>
45#include <linux/times.h>
46#include <linux/limits.h>
47#include <linux/dcache.h>
48#include <linux/dnotify.h>
49#include <linux/syscalls.h>
50#include <linux/vmstat.h>
51#include <linux/nfs_fs.h>
52#include <linux/acpi.h>
53#include <linux/reboot.h>
54#include <linux/ftrace.h>
55#include <linux/perf_event.h>
56#include <linux/kprobes.h>
57#include <linux/pipe_fs_i.h>
58#include <linux/oom.h>
59#include <linux/kmod.h>
60
61#include <asm/uaccess.h>
62#include <asm/processor.h>
63
64#ifdef CONFIG_X86
65#include <asm/nmi.h>
66#include <asm/stacktrace.h>
67#include <asm/io.h>
68#endif
69#ifdef CONFIG_BSD_PROCESS_ACCT
70#include <linux/acct.h>
71#endif
72#ifdef CONFIG_RT_MUTEXES
73#include <linux/rtmutex.h>
74#endif
75#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
76#include <linux/lockdep.h>
77#endif
78#ifdef CONFIG_CHR_DEV_SG
79#include <scsi/sg.h>
80#endif
81
82#ifdef CONFIG_LOCKUP_DETECTOR
83#include <linux/nmi.h>
84#endif
85
86
87#if defined(CONFIG_SYSCTL)
88
89/* External variables not in a header file. */
90extern int sysctl_overcommit_memory;
91extern int sysctl_overcommit_ratio;
92extern int max_threads;
93extern int core_uses_pid;
94extern int suid_dumpable;
95extern char core_pattern[];
96extern unsigned int core_pipe_limit;
97extern int pid_max;
98extern int min_free_kbytes;
99extern int pid_max_min, pid_max_max;
100extern int sysctl_drop_caches;
101extern int percpu_pagelist_fraction;
102extern int compat_log;
103extern int latencytop_enabled;
104extern int sysctl_nr_open_min, sysctl_nr_open_max;
105#ifndef CONFIG_MMU
106extern int sysctl_nr_trim_pages;
107#endif
108#ifdef CONFIG_BLOCK
109extern int blk_iopoll_enabled;
110#endif
111
112/* Constants used for minimum and maximum */
113#ifdef CONFIG_LOCKUP_DETECTOR
114static int sixty = 60;
115static int neg_one = -1;
116#endif
117
118static int zero;
119static int __maybe_unused one = 1;
120static int __maybe_unused two = 2;
121static int __maybe_unused three = 3;
122static unsigned long one_ul = 1;
123static int one_hundred = 100;
124#ifdef CONFIG_PRINTK
125static int ten_thousand = 10000;
126#endif
127
128/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
129static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
130
131/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
132static int maxolduid = 65535;
133static int minolduid;
134static int min_percpu_pagelist_fract = 8;
135
136static int ngroups_max = NGROUPS_MAX;
137
138#ifdef CONFIG_INOTIFY_USER
139#include <linux/inotify.h>
140#endif
141#ifdef CONFIG_SPARC
142#include <asm/system.h>
143#endif
144
145#ifdef CONFIG_SPARC64
146extern int sysctl_tsb_ratio;
147#endif
148
149#ifdef __hppa__
150extern int pwrsw_enabled;
151extern int unaligned_enabled;
152#endif
153
154#ifdef CONFIG_S390
155#ifdef CONFIG_MATHEMU
156extern int sysctl_ieee_emulation_warnings;
157#endif
158extern int sysctl_userprocess_debug;
159extern int spin_retry;
160#endif
161
162#ifdef CONFIG_IA64
163extern int no_unaligned_warning;
164extern int unaligned_dump_stack;
165#endif
166
167#ifdef CONFIG_PROC_SYSCTL
168static int proc_do_cad_pid(struct ctl_table *table, int write,
169 void __user *buffer, size_t *lenp, loff_t *ppos);
170static int proc_taint(struct ctl_table *table, int write,
171 void __user *buffer, size_t *lenp, loff_t *ppos);
172#endif
173
174#ifdef CONFIG_PRINTK
175static int proc_dmesg_restrict(struct ctl_table *table, int write,
176 void __user *buffer, size_t *lenp, loff_t *ppos);
177#endif
178
179#ifdef CONFIG_MAGIC_SYSRQ
180/* Note: sysrq code uses it's own private copy */
181static int __sysrq_enabled = SYSRQ_DEFAULT_ENABLE;
182
183static int sysrq_sysctl_handler(ctl_table *table, int write,
184 void __user *buffer, size_t *lenp,
185 loff_t *ppos)
186{
187 int error;
188
189 error = proc_dointvec(table, write, buffer, lenp, ppos);
190 if (error)
191 return error;
192
193 if (write)
194 sysrq_toggle_support(__sysrq_enabled);
195
196 return 0;
197}
198
199#endif
200
201static struct ctl_table root_table[];
202static struct ctl_table_root sysctl_table_root;
203static struct ctl_table_header root_table_header = {
204 {{.count = 1,
205 .ctl_table = root_table,
206 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),}},
207 .root = &sysctl_table_root,
208 .set = &sysctl_table_root.default_set,
209};
210static struct ctl_table_root sysctl_table_root = {
211 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
212 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
213};
214
215static struct ctl_table kern_table[];
216static struct ctl_table vm_table[];
217static struct ctl_table fs_table[];
218static struct ctl_table debug_table[];
219static struct ctl_table dev_table[];
220extern struct ctl_table random_table[];
221#ifdef CONFIG_EPOLL
222extern struct ctl_table epoll_table[];
223#endif
224
225#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
226int sysctl_legacy_va_layout;
227#endif
228
229/* The default sysctl tables: */
230
231static struct ctl_table root_table[] = {
232 {
233 .procname = "kernel",
234 .mode = 0555,
235 .child = kern_table,
236 },
237 {
238 .procname = "vm",
239 .mode = 0555,
240 .child = vm_table,
241 },
242 {
243 .procname = "fs",
244 .mode = 0555,
245 .child = fs_table,
246 },
247 {
248 .procname = "debug",
249 .mode = 0555,
250 .child = debug_table,
251 },
252 {
253 .procname = "dev",
254 .mode = 0555,
255 .child = dev_table,
256 },
257 { }
258};
259
260#ifdef CONFIG_SCHED_DEBUG
261static int min_sched_granularity_ns = 100000; /* 100 usecs */
262static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
263static int min_wakeup_granularity_ns; /* 0 usecs */
264static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
265static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
266static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
267#endif
268
269#ifdef CONFIG_COMPACTION
270static int min_extfrag_threshold;
271static int max_extfrag_threshold = 1000;
272#endif
273
274static struct ctl_table kern_table[] = {
275 {
276 .procname = "sched_child_runs_first",
277 .data = &sysctl_sched_child_runs_first,
278 .maxlen = sizeof(unsigned int),
279 .mode = 0644,
280 .proc_handler = proc_dointvec,
281 },
282#ifdef CONFIG_SCHED_DEBUG
283 {
284 .procname = "sched_min_granularity_ns",
285 .data = &sysctl_sched_min_granularity,
286 .maxlen = sizeof(unsigned int),
287 .mode = 0644,
288 .proc_handler = sched_proc_update_handler,
289 .extra1 = &min_sched_granularity_ns,
290 .extra2 = &max_sched_granularity_ns,
291 },
292 {
293 .procname = "sched_latency_ns",
294 .data = &sysctl_sched_latency,
295 .maxlen = sizeof(unsigned int),
296 .mode = 0644,
297 .proc_handler = sched_proc_update_handler,
298 .extra1 = &min_sched_granularity_ns,
299 .extra2 = &max_sched_granularity_ns,
300 },
301 {
302 .procname = "sched_wakeup_granularity_ns",
303 .data = &sysctl_sched_wakeup_granularity,
304 .maxlen = sizeof(unsigned int),
305 .mode = 0644,
306 .proc_handler = sched_proc_update_handler,
307 .extra1 = &min_wakeup_granularity_ns,
308 .extra2 = &max_wakeup_granularity_ns,
309 },
310 {
311 .procname = "sched_tunable_scaling",
312 .data = &sysctl_sched_tunable_scaling,
313 .maxlen = sizeof(enum sched_tunable_scaling),
314 .mode = 0644,
315 .proc_handler = sched_proc_update_handler,
316 .extra1 = &min_sched_tunable_scaling,
317 .extra2 = &max_sched_tunable_scaling,
318 },
319 {
320 .procname = "sched_migration_cost",
321 .data = &sysctl_sched_migration_cost,
322 .maxlen = sizeof(unsigned int),
323 .mode = 0644,
324 .proc_handler = proc_dointvec,
325 },
326 {
327 .procname = "sched_nr_migrate",
328 .data = &sysctl_sched_nr_migrate,
329 .maxlen = sizeof(unsigned int),
330 .mode = 0644,
331 .proc_handler = proc_dointvec,
332 },
333 {
334 .procname = "sched_time_avg",
335 .data = &sysctl_sched_time_avg,
336 .maxlen = sizeof(unsigned int),
337 .mode = 0644,
338 .proc_handler = proc_dointvec,
339 },
340 {
341 .procname = "sched_shares_window",
342 .data = &sysctl_sched_shares_window,
343 .maxlen = sizeof(unsigned int),
344 .mode = 0644,
345 .proc_handler = proc_dointvec,
346 },
347 {
348 .procname = "timer_migration",
349 .data = &sysctl_timer_migration,
350 .maxlen = sizeof(unsigned int),
351 .mode = 0644,
352 .proc_handler = proc_dointvec_minmax,
353 .extra1 = &zero,
354 .extra2 = &one,
355 },
356#endif
357 {
358 .procname = "sched_rt_period_us",
359 .data = &sysctl_sched_rt_period,
360 .maxlen = sizeof(unsigned int),
361 .mode = 0644,
362 .proc_handler = sched_rt_handler,
363 },
364 {
365 .procname = "sched_rt_runtime_us",
366 .data = &sysctl_sched_rt_runtime,
367 .maxlen = sizeof(int),
368 .mode = 0644,
369 .proc_handler = sched_rt_handler,
370 },
371#ifdef CONFIG_SCHED_AUTOGROUP
372 {
373 .procname = "sched_autogroup_enabled",
374 .data = &sysctl_sched_autogroup_enabled,
375 .maxlen = sizeof(unsigned int),
376 .mode = 0644,
377 .proc_handler = proc_dointvec_minmax,
378 .extra1 = &zero,
379 .extra2 = &one,
380 },
381#endif
382#ifdef CONFIG_PROVE_LOCKING
383 {
384 .procname = "prove_locking",
385 .data = &prove_locking,
386 .maxlen = sizeof(int),
387 .mode = 0644,
388 .proc_handler = proc_dointvec,
389 },
390#endif
391#ifdef CONFIG_LOCK_STAT
392 {
393 .procname = "lock_stat",
394 .data = &lock_stat,
395 .maxlen = sizeof(int),
396 .mode = 0644,
397 .proc_handler = proc_dointvec,
398 },
399#endif
400 {
401 .procname = "panic",
402 .data = &panic_timeout,
403 .maxlen = sizeof(int),
404 .mode = 0644,
405 .proc_handler = proc_dointvec,
406 },
407 {
408 .procname = "core_uses_pid",
409 .data = &core_uses_pid,
410 .maxlen = sizeof(int),
411 .mode = 0644,
412 .proc_handler = proc_dointvec,
413 },
414 {
415 .procname = "core_pattern",
416 .data = core_pattern,
417 .maxlen = CORENAME_MAX_SIZE,
418 .mode = 0644,
419 .proc_handler = proc_dostring,
420 },
421 {
422 .procname = "core_pipe_limit",
423 .data = &core_pipe_limit,
424 .maxlen = sizeof(unsigned int),
425 .mode = 0644,
426 .proc_handler = proc_dointvec,
427 },
428#ifdef CONFIG_PROC_SYSCTL
429 {
430 .procname = "tainted",
431 .maxlen = sizeof(long),
432 .mode = 0644,
433 .proc_handler = proc_taint,
434 },
435#endif
436#ifdef CONFIG_LATENCYTOP
437 {
438 .procname = "latencytop",
439 .data = &latencytop_enabled,
440 .maxlen = sizeof(int),
441 .mode = 0644,
442 .proc_handler = proc_dointvec,
443 },
444#endif
445#ifdef CONFIG_BLK_DEV_INITRD
446 {
447 .procname = "real-root-dev",
448 .data = &real_root_dev,
449 .maxlen = sizeof(int),
450 .mode = 0644,
451 .proc_handler = proc_dointvec,
452 },
453#endif
454 {
455 .procname = "print-fatal-signals",
456 .data = &print_fatal_signals,
457 .maxlen = sizeof(int),
458 .mode = 0644,
459 .proc_handler = proc_dointvec,
460 },
461#ifdef CONFIG_SPARC
462 {
463 .procname = "reboot-cmd",
464 .data = reboot_command,
465 .maxlen = 256,
466 .mode = 0644,
467 .proc_handler = proc_dostring,
468 },
469 {
470 .procname = "stop-a",
471 .data = &stop_a_enabled,
472 .maxlen = sizeof (int),
473 .mode = 0644,
474 .proc_handler = proc_dointvec,
475 },
476 {
477 .procname = "scons-poweroff",
478 .data = &scons_pwroff,
479 .maxlen = sizeof (int),
480 .mode = 0644,
481 .proc_handler = proc_dointvec,
482 },
483#endif
484#ifdef CONFIG_SPARC64
485 {
486 .procname = "tsb-ratio",
487 .data = &sysctl_tsb_ratio,
488 .maxlen = sizeof (int),
489 .mode = 0644,
490 .proc_handler = proc_dointvec,
491 },
492#endif
493#ifdef __hppa__
494 {
495 .procname = "soft-power",
496 .data = &pwrsw_enabled,
497 .maxlen = sizeof (int),
498 .mode = 0644,
499 .proc_handler = proc_dointvec,
500 },
501 {
502 .procname = "unaligned-trap",
503 .data = &unaligned_enabled,
504 .maxlen = sizeof (int),
505 .mode = 0644,
506 .proc_handler = proc_dointvec,
507 },
508#endif
509 {
510 .procname = "ctrl-alt-del",
511 .data = &C_A_D,
512 .maxlen = sizeof(int),
513 .mode = 0644,
514 .proc_handler = proc_dointvec,
515 },
516#ifdef CONFIG_FUNCTION_TRACER
517 {
518 .procname = "ftrace_enabled",
519 .data = &ftrace_enabled,
520 .maxlen = sizeof(int),
521 .mode = 0644,
522 .proc_handler = ftrace_enable_sysctl,
523 },
524#endif
525#ifdef CONFIG_STACK_TRACER
526 {
527 .procname = "stack_tracer_enabled",
528 .data = &stack_tracer_enabled,
529 .maxlen = sizeof(int),
530 .mode = 0644,
531 .proc_handler = stack_trace_sysctl,
532 },
533#endif
534#ifdef CONFIG_TRACING
535 {
536 .procname = "ftrace_dump_on_oops",
537 .data = &ftrace_dump_on_oops,
538 .maxlen = sizeof(int),
539 .mode = 0644,
540 .proc_handler = proc_dointvec,
541 },
542#endif
543#ifdef CONFIG_MODULES
544 {
545 .procname = "modprobe",
546 .data = &modprobe_path,
547 .maxlen = KMOD_PATH_LEN,
548 .mode = 0644,
549 .proc_handler = proc_dostring,
550 },
551 {
552 .procname = "modules_disabled",
553 .data = &modules_disabled,
554 .maxlen = sizeof(int),
555 .mode = 0644,
556 /* only handle a transition from default "0" to "1" */
557 .proc_handler = proc_dointvec_minmax,
558 .extra1 = &one,
559 .extra2 = &one,
560 },
561#endif
562#ifdef CONFIG_HOTPLUG
563 {
564 .procname = "hotplug",
565 .data = &uevent_helper,
566 .maxlen = UEVENT_HELPER_PATH_LEN,
567 .mode = 0644,
568 .proc_handler = proc_dostring,
569 },
570#endif
571#ifdef CONFIG_CHR_DEV_SG
572 {
573 .procname = "sg-big-buff",
574 .data = &sg_big_buff,
575 .maxlen = sizeof (int),
576 .mode = 0444,
577 .proc_handler = proc_dointvec,
578 },
579#endif
580#ifdef CONFIG_BSD_PROCESS_ACCT
581 {
582 .procname = "acct",
583 .data = &acct_parm,
584 .maxlen = 3*sizeof(int),
585 .mode = 0644,
586 .proc_handler = proc_dointvec,
587 },
588#endif
589#ifdef CONFIG_MAGIC_SYSRQ
590 {
591 .procname = "sysrq",
592 .data = &__sysrq_enabled,
593 .maxlen = sizeof (int),
594 .mode = 0644,
595 .proc_handler = sysrq_sysctl_handler,
596 },
597#endif
598#ifdef CONFIG_PROC_SYSCTL
599 {
600 .procname = "cad_pid",
601 .data = NULL,
602 .maxlen = sizeof (int),
603 .mode = 0600,
604 .proc_handler = proc_do_cad_pid,
605 },
606#endif
607 {
608 .procname = "threads-max",
609 .data = &max_threads,
610 .maxlen = sizeof(int),
611 .mode = 0644,
612 .proc_handler = proc_dointvec,
613 },
614 {
615 .procname = "random",
616 .mode = 0555,
617 .child = random_table,
618 },
619 {
620 .procname = "usermodehelper",
621 .mode = 0555,
622 .child = usermodehelper_table,
623 },
624 {
625 .procname = "overflowuid",
626 .data = &overflowuid,
627 .maxlen = sizeof(int),
628 .mode = 0644,
629 .proc_handler = proc_dointvec_minmax,
630 .extra1 = &minolduid,
631 .extra2 = &maxolduid,
632 },
633 {
634 .procname = "overflowgid",
635 .data = &overflowgid,
636 .maxlen = sizeof(int),
637 .mode = 0644,
638 .proc_handler = proc_dointvec_minmax,
639 .extra1 = &minolduid,
640 .extra2 = &maxolduid,
641 },
642#ifdef CONFIG_S390
643#ifdef CONFIG_MATHEMU
644 {
645 .procname = "ieee_emulation_warnings",
646 .data = &sysctl_ieee_emulation_warnings,
647 .maxlen = sizeof(int),
648 .mode = 0644,
649 .proc_handler = proc_dointvec,
650 },
651#endif
652 {
653 .procname = "userprocess_debug",
654 .data = &show_unhandled_signals,
655 .maxlen = sizeof(int),
656 .mode = 0644,
657 .proc_handler = proc_dointvec,
658 },
659#endif
660 {
661 .procname = "pid_max",
662 .data = &pid_max,
663 .maxlen = sizeof (int),
664 .mode = 0644,
665 .proc_handler = proc_dointvec_minmax,
666 .extra1 = &pid_max_min,
667 .extra2 = &pid_max_max,
668 },
669 {
670 .procname = "panic_on_oops",
671 .data = &panic_on_oops,
672 .maxlen = sizeof(int),
673 .mode = 0644,
674 .proc_handler = proc_dointvec,
675 },
676#if defined CONFIG_PRINTK
677 {
678 .procname = "printk",
679 .data = &console_loglevel,
680 .maxlen = 4*sizeof(int),
681 .mode = 0644,
682 .proc_handler = proc_dointvec,
683 },
684 {
685 .procname = "printk_ratelimit",
686 .data = &printk_ratelimit_state.interval,
687 .maxlen = sizeof(int),
688 .mode = 0644,
689 .proc_handler = proc_dointvec_jiffies,
690 },
691 {
692 .procname = "printk_ratelimit_burst",
693 .data = &printk_ratelimit_state.burst,
694 .maxlen = sizeof(int),
695 .mode = 0644,
696 .proc_handler = proc_dointvec,
697 },
698 {
699 .procname = "printk_delay",
700 .data = &printk_delay_msec,
701 .maxlen = sizeof(int),
702 .mode = 0644,
703 .proc_handler = proc_dointvec_minmax,
704 .extra1 = &zero,
705 .extra2 = &ten_thousand,
706 },
707 {
708 .procname = "dmesg_restrict",
709 .data = &dmesg_restrict,
710 .maxlen = sizeof(int),
711 .mode = 0644,
712 .proc_handler = proc_dointvec_minmax,
713 .extra1 = &zero,
714 .extra2 = &one,
715 },
716 {
717 .procname = "kptr_restrict",
718 .data = &kptr_restrict,
719 .maxlen = sizeof(int),
720 .mode = 0644,
721 .proc_handler = proc_dmesg_restrict,
722 .extra1 = &zero,
723 .extra2 = &two,
724 },
725#endif
726 {
727 .procname = "ngroups_max",
728 .data = &ngroups_max,
729 .maxlen = sizeof (int),
730 .mode = 0444,
731 .proc_handler = proc_dointvec,
732 },
733#if defined(CONFIG_LOCKUP_DETECTOR)
734 {
735 .procname = "watchdog",
736 .data = &watchdog_enabled,
737 .maxlen = sizeof (int),
738 .mode = 0644,
739 .proc_handler = proc_dowatchdog,
740 .extra1 = &zero,
741 .extra2 = &one,
742 },
743 {
744 .procname = "watchdog_thresh",
745 .data = &watchdog_thresh,
746 .maxlen = sizeof(int),
747 .mode = 0644,
748 .proc_handler = proc_dowatchdog,
749 .extra1 = &neg_one,
750 .extra2 = &sixty,
751 },
752 {
753 .procname = "softlockup_panic",
754 .data = &softlockup_panic,
755 .maxlen = sizeof(int),
756 .mode = 0644,
757 .proc_handler = proc_dointvec_minmax,
758 .extra1 = &zero,
759 .extra2 = &one,
760 },
761 {
762 .procname = "nmi_watchdog",
763 .data = &watchdog_enabled,
764 .maxlen = sizeof (int),
765 .mode = 0644,
766 .proc_handler = proc_dowatchdog,
767 .extra1 = &zero,
768 .extra2 = &one,
769 },
770#endif
771#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
772 {
773 .procname = "unknown_nmi_panic",
774 .data = &unknown_nmi_panic,
775 .maxlen = sizeof (int),
776 .mode = 0644,
777 .proc_handler = proc_dointvec,
778 },
779#endif
780#if defined(CONFIG_X86)
781 {
782 .procname = "panic_on_unrecovered_nmi",
783 .data = &panic_on_unrecovered_nmi,
784 .maxlen = sizeof(int),
785 .mode = 0644,
786 .proc_handler = proc_dointvec,
787 },
788 {
789 .procname = "panic_on_io_nmi",
790 .data = &panic_on_io_nmi,
791 .maxlen = sizeof(int),
792 .mode = 0644,
793 .proc_handler = proc_dointvec,
794 },
795 {
796 .procname = "bootloader_type",
797 .data = &bootloader_type,
798 .maxlen = sizeof (int),
799 .mode = 0444,
800 .proc_handler = proc_dointvec,
801 },
802 {
803 .procname = "bootloader_version",
804 .data = &bootloader_version,
805 .maxlen = sizeof (int),
806 .mode = 0444,
807 .proc_handler = proc_dointvec,
808 },
809 {
810 .procname = "kstack_depth_to_print",
811 .data = &kstack_depth_to_print,
812 .maxlen = sizeof(int),
813 .mode = 0644,
814 .proc_handler = proc_dointvec,
815 },
816 {
817 .procname = "io_delay_type",
818 .data = &io_delay_type,
819 .maxlen = sizeof(int),
820 .mode = 0644,
821 .proc_handler = proc_dointvec,
822 },
823#endif
824#if defined(CONFIG_MMU)
825 {
826 .procname = "randomize_va_space",
827 .data = &randomize_va_space,
828 .maxlen = sizeof(int),
829 .mode = 0644,
830 .proc_handler = proc_dointvec,
831 },
832#endif
833#if defined(CONFIG_S390) && defined(CONFIG_SMP)
834 {
835 .procname = "spin_retry",
836 .data = &spin_retry,
837 .maxlen = sizeof (int),
838 .mode = 0644,
839 .proc_handler = proc_dointvec,
840 },
841#endif
842#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
843 {
844 .procname = "acpi_video_flags",
845 .data = &acpi_realmode_flags,
846 .maxlen = sizeof (unsigned long),
847 .mode = 0644,
848 .proc_handler = proc_doulongvec_minmax,
849 },
850#endif
851#ifdef CONFIG_IA64
852 {
853 .procname = "ignore-unaligned-usertrap",
854 .data = &no_unaligned_warning,
855 .maxlen = sizeof (int),
856 .mode = 0644,
857 .proc_handler = proc_dointvec,
858 },
859 {
860 .procname = "unaligned-dump-stack",
861 .data = &unaligned_dump_stack,
862 .maxlen = sizeof (int),
863 .mode = 0644,
864 .proc_handler = proc_dointvec,
865 },
866#endif
867#ifdef CONFIG_DETECT_HUNG_TASK
868 {
869 .procname = "hung_task_panic",
870 .data = &sysctl_hung_task_panic,
871 .maxlen = sizeof(int),
872 .mode = 0644,
873 .proc_handler = proc_dointvec_minmax,
874 .extra1 = &zero,
875 .extra2 = &one,
876 },
877 {
878 .procname = "hung_task_check_count",
879 .data = &sysctl_hung_task_check_count,
880 .maxlen = sizeof(unsigned long),
881 .mode = 0644,
882 .proc_handler = proc_doulongvec_minmax,
883 },
884 {
885 .procname = "hung_task_timeout_secs",
886 .data = &sysctl_hung_task_timeout_secs,
887 .maxlen = sizeof(unsigned long),
888 .mode = 0644,
889 .proc_handler = proc_dohung_task_timeout_secs,
890 },
891 {
892 .procname = "hung_task_warnings",
893 .data = &sysctl_hung_task_warnings,
894 .maxlen = sizeof(unsigned long),
895 .mode = 0644,
896 .proc_handler = proc_doulongvec_minmax,
897 },
898#endif
899#ifdef CONFIG_COMPAT
900 {
901 .procname = "compat-log",
902 .data = &compat_log,
903 .maxlen = sizeof (int),
904 .mode = 0644,
905 .proc_handler = proc_dointvec,
906 },
907#endif
908#ifdef CONFIG_RT_MUTEXES
909 {
910 .procname = "max_lock_depth",
911 .data = &max_lock_depth,
912 .maxlen = sizeof(int),
913 .mode = 0644,
914 .proc_handler = proc_dointvec,
915 },
916#endif
917 {
918 .procname = "poweroff_cmd",
919 .data = &poweroff_cmd,
920 .maxlen = POWEROFF_CMD_PATH_LEN,
921 .mode = 0644,
922 .proc_handler = proc_dostring,
923 },
924#ifdef CONFIG_KEYS
925 {
926 .procname = "keys",
927 .mode = 0555,
928 .child = key_sysctls,
929 },
930#endif
931#ifdef CONFIG_RCU_TORTURE_TEST
932 {
933 .procname = "rcutorture_runnable",
934 .data = &rcutorture_runnable,
935 .maxlen = sizeof(int),
936 .mode = 0644,
937 .proc_handler = proc_dointvec,
938 },
939#endif
940#ifdef CONFIG_PERF_EVENTS
941 /*
942 * User-space scripts rely on the existence of this file
943 * as a feature check for perf_events being enabled.
944 *
945 * So it's an ABI, do not remove!
946 */
947 {
948 .procname = "perf_event_paranoid",
949 .data = &sysctl_perf_event_paranoid,
950 .maxlen = sizeof(sysctl_perf_event_paranoid),
951 .mode = 0644,
952 .proc_handler = proc_dointvec,
953 },
954 {
955 .procname = "perf_event_mlock_kb",
956 .data = &sysctl_perf_event_mlock,
957 .maxlen = sizeof(sysctl_perf_event_mlock),
958 .mode = 0644,
959 .proc_handler = proc_dointvec,
960 },
961 {
962 .procname = "perf_event_max_sample_rate",
963 .data = &sysctl_perf_event_sample_rate,
964 .maxlen = sizeof(sysctl_perf_event_sample_rate),
965 .mode = 0644,
966 .proc_handler = perf_proc_update_handler,
967 },
968#endif
969#ifdef CONFIG_KMEMCHECK
970 {
971 .procname = "kmemcheck",
972 .data = &kmemcheck_enabled,
973 .maxlen = sizeof(int),
974 .mode = 0644,
975 .proc_handler = proc_dointvec,
976 },
977#endif
978#ifdef CONFIG_BLOCK
979 {
980 .procname = "blk_iopoll",
981 .data = &blk_iopoll_enabled,
982 .maxlen = sizeof(int),
983 .mode = 0644,
984 .proc_handler = proc_dointvec,
985 },
986#endif
987 { }
988};
989
990static struct ctl_table vm_table[] = {
991 {
992 .procname = "overcommit_memory",
993 .data = &sysctl_overcommit_memory,
994 .maxlen = sizeof(sysctl_overcommit_memory),
995 .mode = 0644,
996 .proc_handler = proc_dointvec_minmax,
997 .extra1 = &zero,
998 .extra2 = &two,
999 },
1000 {
1001 .procname = "panic_on_oom",
1002 .data = &sysctl_panic_on_oom,
1003 .maxlen = sizeof(sysctl_panic_on_oom),
1004 .mode = 0644,
1005 .proc_handler = proc_dointvec_minmax,
1006 .extra1 = &zero,
1007 .extra2 = &two,
1008 },
1009 {
1010 .procname = "oom_kill_allocating_task",
1011 .data = &sysctl_oom_kill_allocating_task,
1012 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1013 .mode = 0644,
1014 .proc_handler = proc_dointvec,
1015 },
1016 {
1017 .procname = "oom_dump_tasks",
1018 .data = &sysctl_oom_dump_tasks,
1019 .maxlen = sizeof(sysctl_oom_dump_tasks),
1020 .mode = 0644,
1021 .proc_handler = proc_dointvec,
1022 },
1023 {
1024 .procname = "overcommit_ratio",
1025 .data = &sysctl_overcommit_ratio,
1026 .maxlen = sizeof(sysctl_overcommit_ratio),
1027 .mode = 0644,
1028 .proc_handler = proc_dointvec,
1029 },
1030 {
1031 .procname = "page-cluster",
1032 .data = &page_cluster,
1033 .maxlen = sizeof(int),
1034 .mode = 0644,
1035 .proc_handler = proc_dointvec_minmax,
1036 .extra1 = &zero,
1037 },
1038 {
1039 .procname = "dirty_background_ratio",
1040 .data = &dirty_background_ratio,
1041 .maxlen = sizeof(dirty_background_ratio),
1042 .mode = 0644,
1043 .proc_handler = dirty_background_ratio_handler,
1044 .extra1 = &zero,
1045 .extra2 = &one_hundred,
1046 },
1047 {
1048 .procname = "dirty_background_bytes",
1049 .data = &dirty_background_bytes,
1050 .maxlen = sizeof(dirty_background_bytes),
1051 .mode = 0644,
1052 .proc_handler = dirty_background_bytes_handler,
1053 .extra1 = &one_ul,
1054 },
1055 {
1056 .procname = "dirty_ratio",
1057 .data = &vm_dirty_ratio,
1058 .maxlen = sizeof(vm_dirty_ratio),
1059 .mode = 0644,
1060 .proc_handler = dirty_ratio_handler,
1061 .extra1 = &zero,
1062 .extra2 = &one_hundred,
1063 },
1064 {
1065 .procname = "dirty_bytes",
1066 .data = &vm_dirty_bytes,
1067 .maxlen = sizeof(vm_dirty_bytes),
1068 .mode = 0644,
1069 .proc_handler = dirty_bytes_handler,
1070 .extra1 = &dirty_bytes_min,
1071 },
1072 {
1073 .procname = "dirty_writeback_centisecs",
1074 .data = &dirty_writeback_interval,
1075 .maxlen = sizeof(dirty_writeback_interval),
1076 .mode = 0644,
1077 .proc_handler = dirty_writeback_centisecs_handler,
1078 },
1079 {
1080 .procname = "dirty_expire_centisecs",
1081 .data = &dirty_expire_interval,
1082 .maxlen = sizeof(dirty_expire_interval),
1083 .mode = 0644,
1084 .proc_handler = proc_dointvec_minmax,
1085 .extra1 = &zero,
1086 },
1087 {
1088 .procname = "nr_pdflush_threads",
1089 .data = &nr_pdflush_threads,
1090 .maxlen = sizeof nr_pdflush_threads,
1091 .mode = 0444 /* read-only*/,
1092 .proc_handler = proc_dointvec,
1093 },
1094 {
1095 .procname = "swappiness",
1096 .data = &vm_swappiness,
1097 .maxlen = sizeof(vm_swappiness),
1098 .mode = 0644,
1099 .proc_handler = proc_dointvec_minmax,
1100 .extra1 = &zero,
1101 .extra2 = &one_hundred,
1102 },
1103#ifdef CONFIG_HUGETLB_PAGE
1104 {
1105 .procname = "nr_hugepages",
1106 .data = NULL,
1107 .maxlen = sizeof(unsigned long),
1108 .mode = 0644,
1109 .proc_handler = hugetlb_sysctl_handler,
1110 .extra1 = (void *)&hugetlb_zero,
1111 .extra2 = (void *)&hugetlb_infinity,
1112 },
1113#ifdef CONFIG_NUMA
1114 {
1115 .procname = "nr_hugepages_mempolicy",
1116 .data = NULL,
1117 .maxlen = sizeof(unsigned long),
1118 .mode = 0644,
1119 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1120 .extra1 = (void *)&hugetlb_zero,
1121 .extra2 = (void *)&hugetlb_infinity,
1122 },
1123#endif
1124 {
1125 .procname = "hugetlb_shm_group",
1126 .data = &sysctl_hugetlb_shm_group,
1127 .maxlen = sizeof(gid_t),
1128 .mode = 0644,
1129 .proc_handler = proc_dointvec,
1130 },
1131 {
1132 .procname = "hugepages_treat_as_movable",
1133 .data = &hugepages_treat_as_movable,
1134 .maxlen = sizeof(int),
1135 .mode = 0644,
1136 .proc_handler = hugetlb_treat_movable_handler,
1137 },
1138 {
1139 .procname = "nr_overcommit_hugepages",
1140 .data = NULL,
1141 .maxlen = sizeof(unsigned long),
1142 .mode = 0644,
1143 .proc_handler = hugetlb_overcommit_handler,
1144 .extra1 = (void *)&hugetlb_zero,
1145 .extra2 = (void *)&hugetlb_infinity,
1146 },
1147#endif
1148 {
1149 .procname = "lowmem_reserve_ratio",
1150 .data = &sysctl_lowmem_reserve_ratio,
1151 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1152 .mode = 0644,
1153 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1154 },
1155 {
1156 .procname = "drop_caches",
1157 .data = &sysctl_drop_caches,
1158 .maxlen = sizeof(int),
1159 .mode = 0644,
1160 .proc_handler = drop_caches_sysctl_handler,
1161 .extra1 = &one,
1162 .extra2 = &three,
1163 },
1164#ifdef CONFIG_COMPACTION
1165 {
1166 .procname = "compact_memory",
1167 .data = &sysctl_compact_memory,
1168 .maxlen = sizeof(int),
1169 .mode = 0200,
1170 .proc_handler = sysctl_compaction_handler,
1171 },
1172 {
1173 .procname = "extfrag_threshold",
1174 .data = &sysctl_extfrag_threshold,
1175 .maxlen = sizeof(int),
1176 .mode = 0644,
1177 .proc_handler = sysctl_extfrag_handler,
1178 .extra1 = &min_extfrag_threshold,
1179 .extra2 = &max_extfrag_threshold,
1180 },
1181
1182#endif /* CONFIG_COMPACTION */
1183 {
1184 .procname = "min_free_kbytes",
1185 .data = &min_free_kbytes,
1186 .maxlen = sizeof(min_free_kbytes),
1187 .mode = 0644,
1188 .proc_handler = min_free_kbytes_sysctl_handler,
1189 .extra1 = &zero,
1190 },
1191 {
1192 .procname = "percpu_pagelist_fraction",
1193 .data = &percpu_pagelist_fraction,
1194 .maxlen = sizeof(percpu_pagelist_fraction),
1195 .mode = 0644,
1196 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1197 .extra1 = &min_percpu_pagelist_fract,
1198 },
1199#ifdef CONFIG_MMU
1200 {
1201 .procname = "max_map_count",
1202 .data = &sysctl_max_map_count,
1203 .maxlen = sizeof(sysctl_max_map_count),
1204 .mode = 0644,
1205 .proc_handler = proc_dointvec_minmax,
1206 .extra1 = &zero,
1207 },
1208#else
1209 {
1210 .procname = "nr_trim_pages",
1211 .data = &sysctl_nr_trim_pages,
1212 .maxlen = sizeof(sysctl_nr_trim_pages),
1213 .mode = 0644,
1214 .proc_handler = proc_dointvec_minmax,
1215 .extra1 = &zero,
1216 },
1217#endif
1218 {
1219 .procname = "laptop_mode",
1220 .data = &laptop_mode,
1221 .maxlen = sizeof(laptop_mode),
1222 .mode = 0644,
1223 .proc_handler = proc_dointvec_jiffies,
1224 },
1225 {
1226 .procname = "block_dump",
1227 .data = &block_dump,
1228 .maxlen = sizeof(block_dump),
1229 .mode = 0644,
1230 .proc_handler = proc_dointvec,
1231 .extra1 = &zero,
1232 },
1233 {
1234 .procname = "vfs_cache_pressure",
1235 .data = &sysctl_vfs_cache_pressure,
1236 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1237 .mode = 0644,
1238 .proc_handler = proc_dointvec,
1239 .extra1 = &zero,
1240 },
1241#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1242 {
1243 .procname = "legacy_va_layout",
1244 .data = &sysctl_legacy_va_layout,
1245 .maxlen = sizeof(sysctl_legacy_va_layout),
1246 .mode = 0644,
1247 .proc_handler = proc_dointvec,
1248 .extra1 = &zero,
1249 },
1250#endif
1251#ifdef CONFIG_NUMA
1252 {
1253 .procname = "zone_reclaim_mode",
1254 .data = &zone_reclaim_mode,
1255 .maxlen = sizeof(zone_reclaim_mode),
1256 .mode = 0644,
1257 .proc_handler = proc_dointvec,
1258 .extra1 = &zero,
1259 },
1260 {
1261 .procname = "min_unmapped_ratio",
1262 .data = &sysctl_min_unmapped_ratio,
1263 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1264 .mode = 0644,
1265 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1266 .extra1 = &zero,
1267 .extra2 = &one_hundred,
1268 },
1269 {
1270 .procname = "min_slab_ratio",
1271 .data = &sysctl_min_slab_ratio,
1272 .maxlen = sizeof(sysctl_min_slab_ratio),
1273 .mode = 0644,
1274 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1275 .extra1 = &zero,
1276 .extra2 = &one_hundred,
1277 },
1278#endif
1279#ifdef CONFIG_SMP
1280 {
1281 .procname = "stat_interval",
1282 .data = &sysctl_stat_interval,
1283 .maxlen = sizeof(sysctl_stat_interval),
1284 .mode = 0644,
1285 .proc_handler = proc_dointvec_jiffies,
1286 },
1287#endif
1288#ifdef CONFIG_MMU
1289 {
1290 .procname = "mmap_min_addr",
1291 .data = &dac_mmap_min_addr,
1292 .maxlen = sizeof(unsigned long),
1293 .mode = 0644,
1294 .proc_handler = mmap_min_addr_handler,
1295 },
1296#endif
1297#ifdef CONFIG_NUMA
1298 {
1299 .procname = "numa_zonelist_order",
1300 .data = &numa_zonelist_order,
1301 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1302 .mode = 0644,
1303 .proc_handler = numa_zonelist_order_handler,
1304 },
1305#endif
1306#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1307 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1308 {
1309 .procname = "vdso_enabled",
1310 .data = &vdso_enabled,
1311 .maxlen = sizeof(vdso_enabled),
1312 .mode = 0644,
1313 .proc_handler = proc_dointvec,
1314 .extra1 = &zero,
1315 },
1316#endif
1317#ifdef CONFIG_HIGHMEM
1318 {
1319 .procname = "highmem_is_dirtyable",
1320 .data = &vm_highmem_is_dirtyable,
1321 .maxlen = sizeof(vm_highmem_is_dirtyable),
1322 .mode = 0644,
1323 .proc_handler = proc_dointvec_minmax,
1324 .extra1 = &zero,
1325 .extra2 = &one,
1326 },
1327#endif
1328 {
1329 .procname = "scan_unevictable_pages",
1330 .data = &scan_unevictable_pages,
1331 .maxlen = sizeof(scan_unevictable_pages),
1332 .mode = 0644,
1333 .proc_handler = scan_unevictable_handler,
1334 },
1335#ifdef CONFIG_MEMORY_FAILURE
1336 {
1337 .procname = "memory_failure_early_kill",
1338 .data = &sysctl_memory_failure_early_kill,
1339 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1340 .mode = 0644,
1341 .proc_handler = proc_dointvec_minmax,
1342 .extra1 = &zero,
1343 .extra2 = &one,
1344 },
1345 {
1346 .procname = "memory_failure_recovery",
1347 .data = &sysctl_memory_failure_recovery,
1348 .maxlen = sizeof(sysctl_memory_failure_recovery),
1349 .mode = 0644,
1350 .proc_handler = proc_dointvec_minmax,
1351 .extra1 = &zero,
1352 .extra2 = &one,
1353 },
1354#endif
1355 { }
1356};
1357
1358#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1359static struct ctl_table binfmt_misc_table[] = {
1360 { }
1361};
1362#endif
1363
1364static struct ctl_table fs_table[] = {
1365 {
1366 .procname = "inode-nr",
1367 .data = &inodes_stat,
1368 .maxlen = 2*sizeof(int),
1369 .mode = 0444,
1370 .proc_handler = proc_nr_inodes,
1371 },
1372 {
1373 .procname = "inode-state",
1374 .data = &inodes_stat,
1375 .maxlen = 7*sizeof(int),
1376 .mode = 0444,
1377 .proc_handler = proc_nr_inodes,
1378 },
1379 {
1380 .procname = "file-nr",
1381 .data = &files_stat,
1382 .maxlen = sizeof(files_stat),
1383 .mode = 0444,
1384 .proc_handler = proc_nr_files,
1385 },
1386 {
1387 .procname = "file-max",
1388 .data = &files_stat.max_files,
1389 .maxlen = sizeof(files_stat.max_files),
1390 .mode = 0644,
1391 .proc_handler = proc_doulongvec_minmax,
1392 },
1393 {
1394 .procname = "nr_open",
1395 .data = &sysctl_nr_open,
1396 .maxlen = sizeof(int),
1397 .mode = 0644,
1398 .proc_handler = proc_dointvec_minmax,
1399 .extra1 = &sysctl_nr_open_min,
1400 .extra2 = &sysctl_nr_open_max,
1401 },
1402 {
1403 .procname = "dentry-state",
1404 .data = &dentry_stat,
1405 .maxlen = 6*sizeof(int),
1406 .mode = 0444,
1407 .proc_handler = proc_nr_dentry,
1408 },
1409 {
1410 .procname = "overflowuid",
1411 .data = &fs_overflowuid,
1412 .maxlen = sizeof(int),
1413 .mode = 0644,
1414 .proc_handler = proc_dointvec_minmax,
1415 .extra1 = &minolduid,
1416 .extra2 = &maxolduid,
1417 },
1418 {
1419 .procname = "overflowgid",
1420 .data = &fs_overflowgid,
1421 .maxlen = sizeof(int),
1422 .mode = 0644,
1423 .proc_handler = proc_dointvec_minmax,
1424 .extra1 = &minolduid,
1425 .extra2 = &maxolduid,
1426 },
1427#ifdef CONFIG_FILE_LOCKING
1428 {
1429 .procname = "leases-enable",
1430 .data = &leases_enable,
1431 .maxlen = sizeof(int),
1432 .mode = 0644,
1433 .proc_handler = proc_dointvec,
1434 },
1435#endif
1436#ifdef CONFIG_DNOTIFY
1437 {
1438 .procname = "dir-notify-enable",
1439 .data = &dir_notify_enable,
1440 .maxlen = sizeof(int),
1441 .mode = 0644,
1442 .proc_handler = proc_dointvec,
1443 },
1444#endif
1445#ifdef CONFIG_MMU
1446#ifdef CONFIG_FILE_LOCKING
1447 {
1448 .procname = "lease-break-time",
1449 .data = &lease_break_time,
1450 .maxlen = sizeof(int),
1451 .mode = 0644,
1452 .proc_handler = proc_dointvec,
1453 },
1454#endif
1455#ifdef CONFIG_AIO
1456 {
1457 .procname = "aio-nr",
1458 .data = &aio_nr,
1459 .maxlen = sizeof(aio_nr),
1460 .mode = 0444,
1461 .proc_handler = proc_doulongvec_minmax,
1462 },
1463 {
1464 .procname = "aio-max-nr",
1465 .data = &aio_max_nr,
1466 .maxlen = sizeof(aio_max_nr),
1467 .mode = 0644,
1468 .proc_handler = proc_doulongvec_minmax,
1469 },
1470#endif /* CONFIG_AIO */
1471#ifdef CONFIG_INOTIFY_USER
1472 {
1473 .procname = "inotify",
1474 .mode = 0555,
1475 .child = inotify_table,
1476 },
1477#endif
1478#ifdef CONFIG_EPOLL
1479 {
1480 .procname = "epoll",
1481 .mode = 0555,
1482 .child = epoll_table,
1483 },
1484#endif
1485#endif
1486 {
1487 .procname = "suid_dumpable",
1488 .data = &suid_dumpable,
1489 .maxlen = sizeof(int),
1490 .mode = 0644,
1491 .proc_handler = proc_dointvec_minmax,
1492 .extra1 = &zero,
1493 .extra2 = &two,
1494 },
1495#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1496 {
1497 .procname = "binfmt_misc",
1498 .mode = 0555,
1499 .child = binfmt_misc_table,
1500 },
1501#endif
1502 {
1503 .procname = "pipe-max-size",
1504 .data = &pipe_max_size,
1505 .maxlen = sizeof(int),
1506 .mode = 0644,
1507 .proc_handler = &pipe_proc_fn,
1508 .extra1 = &pipe_min_size,
1509 },
1510 { }
1511};
1512
1513static struct ctl_table debug_table[] = {
1514#if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
1515 defined(CONFIG_S390) || defined(CONFIG_TILE)
1516 {
1517 .procname = "exception-trace",
1518 .data = &show_unhandled_signals,
1519 .maxlen = sizeof(int),
1520 .mode = 0644,
1521 .proc_handler = proc_dointvec
1522 },
1523#endif
1524#if defined(CONFIG_OPTPROBES)
1525 {
1526 .procname = "kprobes-optimization",
1527 .data = &sysctl_kprobes_optimization,
1528 .maxlen = sizeof(int),
1529 .mode = 0644,
1530 .proc_handler = proc_kprobes_optimization_handler,
1531 .extra1 = &zero,
1532 .extra2 = &one,
1533 },
1534#endif
1535 { }
1536};
1537
1538static struct ctl_table dev_table[] = {
1539 { }
1540};
1541
1542static DEFINE_SPINLOCK(sysctl_lock);
1543
1544/* called under sysctl_lock */
1545static int use_table(struct ctl_table_header *p)
1546{
1547 if (unlikely(p->unregistering))
1548 return 0;
1549 p->used++;
1550 return 1;
1551}
1552
1553/* called under sysctl_lock */
1554static void unuse_table(struct ctl_table_header *p)
1555{
1556 if (!--p->used)
1557 if (unlikely(p->unregistering))
1558 complete(p->unregistering);
1559}
1560
1561/* called under sysctl_lock, will reacquire if has to wait */
1562static void start_unregistering(struct ctl_table_header *p)
1563{
1564 /*
1565 * if p->used is 0, nobody will ever touch that entry again;
1566 * we'll eliminate all paths to it before dropping sysctl_lock
1567 */
1568 if (unlikely(p->used)) {
1569 struct completion wait;
1570 init_completion(&wait);
1571 p->unregistering = &wait;
1572 spin_unlock(&sysctl_lock);
1573 wait_for_completion(&wait);
1574 spin_lock(&sysctl_lock);
1575 } else {
1576 /* anything non-NULL; we'll never dereference it */
1577 p->unregistering = ERR_PTR(-EINVAL);
1578 }
1579 /*
1580 * do not remove from the list until nobody holds it; walking the
1581 * list in do_sysctl() relies on that.
1582 */
1583 list_del_init(&p->ctl_entry);
1584}
1585
1586void sysctl_head_get(struct ctl_table_header *head)
1587{
1588 spin_lock(&sysctl_lock);
1589 head->count++;
1590 spin_unlock(&sysctl_lock);
1591}
1592
1593void sysctl_head_put(struct ctl_table_header *head)
1594{
1595 spin_lock(&sysctl_lock);
1596 if (!--head->count)
1597 kfree_rcu(head, rcu);
1598 spin_unlock(&sysctl_lock);
1599}
1600
1601struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1602{
1603 if (!head)
1604 BUG();
1605 spin_lock(&sysctl_lock);
1606 if (!use_table(head))
1607 head = ERR_PTR(-ENOENT);
1608 spin_unlock(&sysctl_lock);
1609 return head;
1610}
1611
1612void sysctl_head_finish(struct ctl_table_header *head)
1613{
1614 if (!head)
1615 return;
1616 spin_lock(&sysctl_lock);
1617 unuse_table(head);
1618 spin_unlock(&sysctl_lock);
1619}
1620
1621static struct ctl_table_set *
1622lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1623{
1624 struct ctl_table_set *set = &root->default_set;
1625 if (root->lookup)
1626 set = root->lookup(root, namespaces);
1627 return set;
1628}
1629
1630static struct list_head *
1631lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1632{
1633 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1634 return &set->list;
1635}
1636
1637struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1638 struct ctl_table_header *prev)
1639{
1640 struct ctl_table_root *root;
1641 struct list_head *header_list;
1642 struct ctl_table_header *head;
1643 struct list_head *tmp;
1644
1645 spin_lock(&sysctl_lock);
1646 if (prev) {
1647 head = prev;
1648 tmp = &prev->ctl_entry;
1649 unuse_table(prev);
1650 goto next;
1651 }
1652 tmp = &root_table_header.ctl_entry;
1653 for (;;) {
1654 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1655
1656 if (!use_table(head))
1657 goto next;
1658 spin_unlock(&sysctl_lock);
1659 return head;
1660 next:
1661 root = head->root;
1662 tmp = tmp->next;
1663 header_list = lookup_header_list(root, namespaces);
1664 if (tmp != header_list)
1665 continue;
1666
1667 do {
1668 root = list_entry(root->root_list.next,
1669 struct ctl_table_root, root_list);
1670 if (root == &sysctl_table_root)
1671 goto out;
1672 header_list = lookup_header_list(root, namespaces);
1673 } while (list_empty(header_list));
1674 tmp = header_list->next;
1675 }
1676out:
1677 spin_unlock(&sysctl_lock);
1678 return NULL;
1679}
1680
1681struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1682{
1683 return __sysctl_head_next(current->nsproxy, prev);
1684}
1685
1686void register_sysctl_root(struct ctl_table_root *root)
1687{
1688 spin_lock(&sysctl_lock);
1689 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1690 spin_unlock(&sysctl_lock);
1691}
1692
1693/*
1694 * sysctl_perm does NOT grant the superuser all rights automatically, because
1695 * some sysctl variables are readonly even to root.
1696 */
1697
1698static int test_perm(int mode, int op)
1699{
1700 if (!current_euid())
1701 mode >>= 6;
1702 else if (in_egroup_p(0))
1703 mode >>= 3;
1704 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1705 return 0;
1706 return -EACCES;
1707}
1708
1709int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1710{
1711 int mode;
1712
1713 if (root->permissions)
1714 mode = root->permissions(root, current->nsproxy, table);
1715 else
1716 mode = table->mode;
1717
1718 return test_perm(mode, op);
1719}
1720
1721static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1722{
1723 for (; table->procname; table++) {
1724 table->parent = parent;
1725 if (table->child)
1726 sysctl_set_parent(table, table->child);
1727 }
1728}
1729
1730static __init int sysctl_init(void)
1731{
1732 sysctl_set_parent(NULL, root_table);
1733#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1734 sysctl_check_table(current->nsproxy, root_table);
1735#endif
1736 return 0;
1737}
1738
1739core_initcall(sysctl_init);
1740
1741static struct ctl_table *is_branch_in(struct ctl_table *branch,
1742 struct ctl_table *table)
1743{
1744 struct ctl_table *p;
1745 const char *s = branch->procname;
1746
1747 /* branch should have named subdirectory as its first element */
1748 if (!s || !branch->child)
1749 return NULL;
1750
1751 /* ... and nothing else */
1752 if (branch[1].procname)
1753 return NULL;
1754
1755 /* table should contain subdirectory with the same name */
1756 for (p = table; p->procname; p++) {
1757 if (!p->child)
1758 continue;
1759 if (p->procname && strcmp(p->procname, s) == 0)
1760 return p;
1761 }
1762 return NULL;
1763}
1764
1765/* see if attaching q to p would be an improvement */
1766static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1767{
1768 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1769 struct ctl_table *next;
1770 int is_better = 0;
1771 int not_in_parent = !p->attached_by;
1772
1773 while ((next = is_branch_in(by, to)) != NULL) {
1774 if (by == q->attached_by)
1775 is_better = 1;
1776 if (to == p->attached_by)
1777 not_in_parent = 1;
1778 by = by->child;
1779 to = next->child;
1780 }
1781
1782 if (is_better && not_in_parent) {
1783 q->attached_by = by;
1784 q->attached_to = to;
1785 q->parent = p;
1786 }
1787}
1788
1789/**
1790 * __register_sysctl_paths - register a sysctl hierarchy
1791 * @root: List of sysctl headers to register on
1792 * @namespaces: Data to compute which lists of sysctl entries are visible
1793 * @path: The path to the directory the sysctl table is in.
1794 * @table: the top-level table structure
1795 *
1796 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1797 * array. A completely 0 filled entry terminates the table.
1798 *
1799 * The members of the &struct ctl_table structure are used as follows:
1800 *
1801 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1802 * enter a sysctl file
1803 *
1804 * data - a pointer to data for use by proc_handler
1805 *
1806 * maxlen - the maximum size in bytes of the data
1807 *
1808 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1809 *
1810 * child - a pointer to the child sysctl table if this entry is a directory, or
1811 * %NULL.
1812 *
1813 * proc_handler - the text handler routine (described below)
1814 *
1815 * de - for internal use by the sysctl routines
1816 *
1817 * extra1, extra2 - extra pointers usable by the proc handler routines
1818 *
1819 * Leaf nodes in the sysctl tree will be represented by a single file
1820 * under /proc; non-leaf nodes will be represented by directories.
1821 *
1822 * sysctl(2) can automatically manage read and write requests through
1823 * the sysctl table. The data and maxlen fields of the ctl_table
1824 * struct enable minimal validation of the values being written to be
1825 * performed, and the mode field allows minimal authentication.
1826 *
1827 * There must be a proc_handler routine for any terminal nodes
1828 * mirrored under /proc/sys (non-terminals are handled by a built-in
1829 * directory handler). Several default handlers are available to
1830 * cover common cases -
1831 *
1832 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1833 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1834 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1835 *
1836 * It is the handler's job to read the input buffer from user memory
1837 * and process it. The handler should return 0 on success.
1838 *
1839 * This routine returns %NULL on a failure to register, and a pointer
1840 * to the table header on success.
1841 */
1842struct ctl_table_header *__register_sysctl_paths(
1843 struct ctl_table_root *root,
1844 struct nsproxy *namespaces,
1845 const struct ctl_path *path, struct ctl_table *table)
1846{
1847 struct ctl_table_header *header;
1848 struct ctl_table *new, **prevp;
1849 unsigned int n, npath;
1850 struct ctl_table_set *set;
1851
1852 /* Count the path components */
1853 for (npath = 0; path[npath].procname; ++npath)
1854 ;
1855
1856 /*
1857 * For each path component, allocate a 2-element ctl_table array.
1858 * The first array element will be filled with the sysctl entry
1859 * for this, the second will be the sentinel (procname == 0).
1860 *
1861 * We allocate everything in one go so that we don't have to
1862 * worry about freeing additional memory in unregister_sysctl_table.
1863 */
1864 header = kzalloc(sizeof(struct ctl_table_header) +
1865 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1866 if (!header)
1867 return NULL;
1868
1869 new = (struct ctl_table *) (header + 1);
1870
1871 /* Now connect the dots */
1872 prevp = &header->ctl_table;
1873 for (n = 0; n < npath; ++n, ++path) {
1874 /* Copy the procname */
1875 new->procname = path->procname;
1876 new->mode = 0555;
1877
1878 *prevp = new;
1879 prevp = &new->child;
1880
1881 new += 2;
1882 }
1883 *prevp = table;
1884 header->ctl_table_arg = table;
1885
1886 INIT_LIST_HEAD(&header->ctl_entry);
1887 header->used = 0;
1888 header->unregistering = NULL;
1889 header->root = root;
1890 sysctl_set_parent(NULL, header->ctl_table);
1891 header->count = 1;
1892#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1893 if (sysctl_check_table(namespaces, header->ctl_table)) {
1894 kfree(header);
1895 return NULL;
1896 }
1897#endif
1898 spin_lock(&sysctl_lock);
1899 header->set = lookup_header_set(root, namespaces);
1900 header->attached_by = header->ctl_table;
1901 header->attached_to = root_table;
1902 header->parent = &root_table_header;
1903 for (set = header->set; set; set = set->parent) {
1904 struct ctl_table_header *p;
1905 list_for_each_entry(p, &set->list, ctl_entry) {
1906 if (p->unregistering)
1907 continue;
1908 try_attach(p, header);
1909 }
1910 }
1911 header->parent->count++;
1912 list_add_tail(&header->ctl_entry, &header->set->list);
1913 spin_unlock(&sysctl_lock);
1914
1915 return header;
1916}
1917
1918/**
1919 * register_sysctl_table_path - register a sysctl table hierarchy
1920 * @path: The path to the directory the sysctl table is in.
1921 * @table: the top-level table structure
1922 *
1923 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1924 * array. A completely 0 filled entry terminates the table.
1925 *
1926 * See __register_sysctl_paths for more details.
1927 */
1928struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1929 struct ctl_table *table)
1930{
1931 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1932 path, table);
1933}
1934
1935/**
1936 * register_sysctl_table - register a sysctl table hierarchy
1937 * @table: the top-level table structure
1938 *
1939 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1940 * array. A completely 0 filled entry terminates the table.
1941 *
1942 * See register_sysctl_paths for more details.
1943 */
1944struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1945{
1946 static const struct ctl_path null_path[] = { {} };
1947
1948 return register_sysctl_paths(null_path, table);
1949}
1950
1951/**
1952 * unregister_sysctl_table - unregister a sysctl table hierarchy
1953 * @header: the header returned from register_sysctl_table
1954 *
1955 * Unregisters the sysctl table and all children. proc entries may not
1956 * actually be removed until they are no longer used by anyone.
1957 */
1958void unregister_sysctl_table(struct ctl_table_header * header)
1959{
1960 might_sleep();
1961
1962 if (header == NULL)
1963 return;
1964
1965 spin_lock(&sysctl_lock);
1966 start_unregistering(header);
1967 if (!--header->parent->count) {
1968 WARN_ON(1);
1969 kfree_rcu(header->parent, rcu);
1970 }
1971 if (!--header->count)
1972 kfree_rcu(header, rcu);
1973 spin_unlock(&sysctl_lock);
1974}
1975
1976int sysctl_is_seen(struct ctl_table_header *p)
1977{
1978 struct ctl_table_set *set = p->set;
1979 int res;
1980 spin_lock(&sysctl_lock);
1981 if (p->unregistering)
1982 res = 0;
1983 else if (!set->is_seen)
1984 res = 1;
1985 else
1986 res = set->is_seen(set);
1987 spin_unlock(&sysctl_lock);
1988 return res;
1989}
1990
1991void setup_sysctl_set(struct ctl_table_set *p,
1992 struct ctl_table_set *parent,
1993 int (*is_seen)(struct ctl_table_set *))
1994{
1995 INIT_LIST_HEAD(&p->list);
1996 p->parent = parent ? parent : &sysctl_table_root.default_set;
1997 p->is_seen = is_seen;
1998}
1999
2000#else /* !CONFIG_SYSCTL */
2001struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2002{
2003 return NULL;
2004}
2005
2006struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2007 struct ctl_table *table)
2008{
2009 return NULL;
2010}
2011
2012void unregister_sysctl_table(struct ctl_table_header * table)
2013{
2014}
2015
2016void setup_sysctl_set(struct ctl_table_set *p,
2017 struct ctl_table_set *parent,
2018 int (*is_seen)(struct ctl_table_set *))
2019{
2020}
2021
2022void sysctl_head_put(struct ctl_table_header *head)
2023{
2024}
2025
2026#endif /* CONFIG_SYSCTL */
2027
2028/*
2029 * /proc/sys support
2030 */
2031
2032#ifdef CONFIG_PROC_SYSCTL
2033
2034static int _proc_do_string(void* data, int maxlen, int write,
2035 void __user *buffer,
2036 size_t *lenp, loff_t *ppos)
2037{
2038 size_t len;
2039 char __user *p;
2040 char c;
2041
2042 if (!data || !maxlen || !*lenp) {
2043 *lenp = 0;
2044 return 0;
2045 }
2046
2047 if (write) {
2048 len = 0;
2049 p = buffer;
2050 while (len < *lenp) {
2051 if (get_user(c, p++))
2052 return -EFAULT;
2053 if (c == 0 || c == '\n')
2054 break;
2055 len++;
2056 }
2057 if (len >= maxlen)
2058 len = maxlen-1;
2059 if(copy_from_user(data, buffer, len))
2060 return -EFAULT;
2061 ((char *) data)[len] = 0;
2062 *ppos += *lenp;
2063 } else {
2064 len = strlen(data);
2065 if (len > maxlen)
2066 len = maxlen;
2067
2068 if (*ppos > len) {
2069 *lenp = 0;
2070 return 0;
2071 }
2072
2073 data += *ppos;
2074 len -= *ppos;
2075
2076 if (len > *lenp)
2077 len = *lenp;
2078 if (len)
2079 if(copy_to_user(buffer, data, len))
2080 return -EFAULT;
2081 if (len < *lenp) {
2082 if(put_user('\n', ((char __user *) buffer) + len))
2083 return -EFAULT;
2084 len++;
2085 }
2086 *lenp = len;
2087 *ppos += len;
2088 }
2089 return 0;
2090}
2091
2092/**
2093 * proc_dostring - read a string sysctl
2094 * @table: the sysctl table
2095 * @write: %TRUE if this is a write to the sysctl file
2096 * @buffer: the user buffer
2097 * @lenp: the size of the user buffer
2098 * @ppos: file position
2099 *
2100 * Reads/writes a string from/to the user buffer. If the kernel
2101 * buffer provided is not large enough to hold the string, the
2102 * string is truncated. The copied string is %NULL-terminated.
2103 * If the string is being read by the user process, it is copied
2104 * and a newline '\n' is added. It is truncated if the buffer is
2105 * not large enough.
2106 *
2107 * Returns 0 on success.
2108 */
2109int proc_dostring(struct ctl_table *table, int write,
2110 void __user *buffer, size_t *lenp, loff_t *ppos)
2111{
2112 return _proc_do_string(table->data, table->maxlen, write,
2113 buffer, lenp, ppos);
2114}
2115
2116static size_t proc_skip_spaces(char **buf)
2117{
2118 size_t ret;
2119 char *tmp = skip_spaces(*buf);
2120 ret = tmp - *buf;
2121 *buf = tmp;
2122 return ret;
2123}
2124
2125static void proc_skip_char(char **buf, size_t *size, const char v)
2126{
2127 while (*size) {
2128 if (**buf != v)
2129 break;
2130 (*size)--;
2131 (*buf)++;
2132 }
2133}
2134
2135#define TMPBUFLEN 22
2136/**
2137 * proc_get_long - reads an ASCII formatted integer from a user buffer
2138 *
2139 * @buf: a kernel buffer
2140 * @size: size of the kernel buffer
2141 * @val: this is where the number will be stored
2142 * @neg: set to %TRUE if number is negative
2143 * @perm_tr: a vector which contains the allowed trailers
2144 * @perm_tr_len: size of the perm_tr vector
2145 * @tr: pointer to store the trailer character
2146 *
2147 * In case of success %0 is returned and @buf and @size are updated with
2148 * the amount of bytes read. If @tr is non-NULL and a trailing
2149 * character exists (size is non-zero after returning from this
2150 * function), @tr is updated with the trailing character.
2151 */
2152static int proc_get_long(char **buf, size_t *size,
2153 unsigned long *val, bool *neg,
2154 const char *perm_tr, unsigned perm_tr_len, char *tr)
2155{
2156 int len;
2157 char *p, tmp[TMPBUFLEN];
2158
2159 if (!*size)
2160 return -EINVAL;
2161
2162 len = *size;
2163 if (len > TMPBUFLEN - 1)
2164 len = TMPBUFLEN - 1;
2165
2166 memcpy(tmp, *buf, len);
2167
2168 tmp[len] = 0;
2169 p = tmp;
2170 if (*p == '-' && *size > 1) {
2171 *neg = true;
2172 p++;
2173 } else
2174 *neg = false;
2175 if (!isdigit(*p))
2176 return -EINVAL;
2177
2178 *val = simple_strtoul(p, &p, 0);
2179
2180 len = p - tmp;
2181
2182 /* We don't know if the next char is whitespace thus we may accept
2183 * invalid integers (e.g. 1234...a) or two integers instead of one
2184 * (e.g. 123...1). So lets not allow such large numbers. */
2185 if (len == TMPBUFLEN - 1)
2186 return -EINVAL;
2187
2188 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2189 return -EINVAL;
2190
2191 if (tr && (len < *size))
2192 *tr = *p;
2193
2194 *buf += len;
2195 *size -= len;
2196
2197 return 0;
2198}
2199
2200/**
2201 * proc_put_long - converts an integer to a decimal ASCII formatted string
2202 *
2203 * @buf: the user buffer
2204 * @size: the size of the user buffer
2205 * @val: the integer to be converted
2206 * @neg: sign of the number, %TRUE for negative
2207 *
2208 * In case of success %0 is returned and @buf and @size are updated with
2209 * the amount of bytes written.
2210 */
2211static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2212 bool neg)
2213{
2214 int len;
2215 char tmp[TMPBUFLEN], *p = tmp;
2216
2217 sprintf(p, "%s%lu", neg ? "-" : "", val);
2218 len = strlen(tmp);
2219 if (len > *size)
2220 len = *size;
2221 if (copy_to_user(*buf, tmp, len))
2222 return -EFAULT;
2223 *size -= len;
2224 *buf += len;
2225 return 0;
2226}
2227#undef TMPBUFLEN
2228
2229static int proc_put_char(void __user **buf, size_t *size, char c)
2230{
2231 if (*size) {
2232 char __user **buffer = (char __user **)buf;
2233 if (put_user(c, *buffer))
2234 return -EFAULT;
2235 (*size)--, (*buffer)++;
2236 *buf = *buffer;
2237 }
2238 return 0;
2239}
2240
2241static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2242 int *valp,
2243 int write, void *data)
2244{
2245 if (write) {
2246 *valp = *negp ? -*lvalp : *lvalp;
2247 } else {
2248 int val = *valp;
2249 if (val < 0) {
2250 *negp = true;
2251 *lvalp = (unsigned long)-val;
2252 } else {
2253 *negp = false;
2254 *lvalp = (unsigned long)val;
2255 }
2256 }
2257 return 0;
2258}
2259
2260static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2261
2262static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2263 int write, void __user *buffer,
2264 size_t *lenp, loff_t *ppos,
2265 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2266 int write, void *data),
2267 void *data)
2268{
2269 int *i, vleft, first = 1, err = 0;
2270 unsigned long page = 0;
2271 size_t left;
2272 char *kbuf;
2273
2274 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2275 *lenp = 0;
2276 return 0;
2277 }
2278
2279 i = (int *) tbl_data;
2280 vleft = table->maxlen / sizeof(*i);
2281 left = *lenp;
2282
2283 if (!conv)
2284 conv = do_proc_dointvec_conv;
2285
2286 if (write) {
2287 if (left > PAGE_SIZE - 1)
2288 left = PAGE_SIZE - 1;
2289 page = __get_free_page(GFP_TEMPORARY);
2290 kbuf = (char *) page;
2291 if (!kbuf)
2292 return -ENOMEM;
2293 if (copy_from_user(kbuf, buffer, left)) {
2294 err = -EFAULT;
2295 goto free;
2296 }
2297 kbuf[left] = 0;
2298 }
2299
2300 for (; left && vleft--; i++, first=0) {
2301 unsigned long lval;
2302 bool neg;
2303
2304 if (write) {
2305 left -= proc_skip_spaces(&kbuf);
2306
2307 if (!left)
2308 break;
2309 err = proc_get_long(&kbuf, &left, &lval, &neg,
2310 proc_wspace_sep,
2311 sizeof(proc_wspace_sep), NULL);
2312 if (err)
2313 break;
2314 if (conv(&neg, &lval, i, 1, data)) {
2315 err = -EINVAL;
2316 break;
2317 }
2318 } else {
2319 if (conv(&neg, &lval, i, 0, data)) {
2320 err = -EINVAL;
2321 break;
2322 }
2323 if (!first)
2324 err = proc_put_char(&buffer, &left, '\t');
2325 if (err)
2326 break;
2327 err = proc_put_long(&buffer, &left, lval, neg);
2328 if (err)
2329 break;
2330 }
2331 }
2332
2333 if (!write && !first && left && !err)
2334 err = proc_put_char(&buffer, &left, '\n');
2335 if (write && !err && left)
2336 left -= proc_skip_spaces(&kbuf);
2337free:
2338 if (write) {
2339 free_page(page);
2340 if (first)
2341 return err ? : -EINVAL;
2342 }
2343 *lenp -= left;
2344 *ppos += *lenp;
2345 return err;
2346}
2347
2348static int do_proc_dointvec(struct ctl_table *table, int write,
2349 void __user *buffer, size_t *lenp, loff_t *ppos,
2350 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2351 int write, void *data),
2352 void *data)
2353{
2354 return __do_proc_dointvec(table->data, table, write,
2355 buffer, lenp, ppos, conv, data);
2356}
2357
2358/**
2359 * proc_dointvec - read a vector of integers
2360 * @table: the sysctl table
2361 * @write: %TRUE if this is a write to the sysctl file
2362 * @buffer: the user buffer
2363 * @lenp: the size of the user buffer
2364 * @ppos: file position
2365 *
2366 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2367 * values from/to the user buffer, treated as an ASCII string.
2368 *
2369 * Returns 0 on success.
2370 */
2371int proc_dointvec(struct ctl_table *table, int write,
2372 void __user *buffer, size_t *lenp, loff_t *ppos)
2373{
2374 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2375 NULL,NULL);
2376}
2377
2378/*
2379 * Taint values can only be increased
2380 * This means we can safely use a temporary.
2381 */
2382static int proc_taint(struct ctl_table *table, int write,
2383 void __user *buffer, size_t *lenp, loff_t *ppos)
2384{
2385 struct ctl_table t;
2386 unsigned long tmptaint = get_taint();
2387 int err;
2388
2389 if (write && !capable(CAP_SYS_ADMIN))
2390 return -EPERM;
2391
2392 t = *table;
2393 t.data = &tmptaint;
2394 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2395 if (err < 0)
2396 return err;
2397
2398 if (write) {
2399 /*
2400 * Poor man's atomic or. Not worth adding a primitive
2401 * to everyone's atomic.h for this
2402 */
2403 int i;
2404 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2405 if ((tmptaint >> i) & 1)
2406 add_taint(i);
2407 }
2408 }
2409
2410 return err;
2411}
2412
2413#ifdef CONFIG_PRINTK
2414static int proc_dmesg_restrict(struct ctl_table *table, int write,
2415 void __user *buffer, size_t *lenp, loff_t *ppos)
2416{
2417 if (write && !capable(CAP_SYS_ADMIN))
2418 return -EPERM;
2419
2420 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2421}
2422#endif
2423
2424struct do_proc_dointvec_minmax_conv_param {
2425 int *min;
2426 int *max;
2427};
2428
2429static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2430 int *valp,
2431 int write, void *data)
2432{
2433 struct do_proc_dointvec_minmax_conv_param *param = data;
2434 if (write) {
2435 int val = *negp ? -*lvalp : *lvalp;
2436 if ((param->min && *param->min > val) ||
2437 (param->max && *param->max < val))
2438 return -EINVAL;
2439 *valp = val;
2440 } else {
2441 int val = *valp;
2442 if (val < 0) {
2443 *negp = true;
2444 *lvalp = (unsigned long)-val;
2445 } else {
2446 *negp = false;
2447 *lvalp = (unsigned long)val;
2448 }
2449 }
2450 return 0;
2451}
2452
2453/**
2454 * proc_dointvec_minmax - read a vector of integers with min/max values
2455 * @table: the sysctl table
2456 * @write: %TRUE if this is a write to the sysctl file
2457 * @buffer: the user buffer
2458 * @lenp: the size of the user buffer
2459 * @ppos: file position
2460 *
2461 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2462 * values from/to the user buffer, treated as an ASCII string.
2463 *
2464 * This routine will ensure the values are within the range specified by
2465 * table->extra1 (min) and table->extra2 (max).
2466 *
2467 * Returns 0 on success.
2468 */
2469int proc_dointvec_minmax(struct ctl_table *table, int write,
2470 void __user *buffer, size_t *lenp, loff_t *ppos)
2471{
2472 struct do_proc_dointvec_minmax_conv_param param = {
2473 .min = (int *) table->extra1,
2474 .max = (int *) table->extra2,
2475 };
2476 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2477 do_proc_dointvec_minmax_conv, ¶m);
2478}
2479
2480static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2481 void __user *buffer,
2482 size_t *lenp, loff_t *ppos,
2483 unsigned long convmul,
2484 unsigned long convdiv)
2485{
2486 unsigned long *i, *min, *max;
2487 int vleft, first = 1, err = 0;
2488 unsigned long page = 0;
2489 size_t left;
2490 char *kbuf;
2491
2492 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2493 *lenp = 0;
2494 return 0;
2495 }
2496
2497 i = (unsigned long *) data;
2498 min = (unsigned long *) table->extra1;
2499 max = (unsigned long *) table->extra2;
2500 vleft = table->maxlen / sizeof(unsigned long);
2501 left = *lenp;
2502
2503 if (write) {
2504 if (left > PAGE_SIZE - 1)
2505 left = PAGE_SIZE - 1;
2506 page = __get_free_page(GFP_TEMPORARY);
2507 kbuf = (char *) page;
2508 if (!kbuf)
2509 return -ENOMEM;
2510 if (copy_from_user(kbuf, buffer, left)) {
2511 err = -EFAULT;
2512 goto free;
2513 }
2514 kbuf[left] = 0;
2515 }
2516
2517 for (; left && vleft--; i++, first = 0) {
2518 unsigned long val;
2519
2520 if (write) {
2521 bool neg;
2522
2523 left -= proc_skip_spaces(&kbuf);
2524
2525 err = proc_get_long(&kbuf, &left, &val, &neg,
2526 proc_wspace_sep,
2527 sizeof(proc_wspace_sep), NULL);
2528 if (err)
2529 break;
2530 if (neg)
2531 continue;
2532 if ((min && val < *min) || (max && val > *max))
2533 continue;
2534 *i = val;
2535 } else {
2536 val = convdiv * (*i) / convmul;
2537 if (!first)
2538 err = proc_put_char(&buffer, &left, '\t');
2539 err = proc_put_long(&buffer, &left, val, false);
2540 if (err)
2541 break;
2542 }
2543 }
2544
2545 if (!write && !first && left && !err)
2546 err = proc_put_char(&buffer, &left, '\n');
2547 if (write && !err)
2548 left -= proc_skip_spaces(&kbuf);
2549free:
2550 if (write) {
2551 free_page(page);
2552 if (first)
2553 return err ? : -EINVAL;
2554 }
2555 *lenp -= left;
2556 *ppos += *lenp;
2557 return err;
2558}
2559
2560static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2561 void __user *buffer,
2562 size_t *lenp, loff_t *ppos,
2563 unsigned long convmul,
2564 unsigned long convdiv)
2565{
2566 return __do_proc_doulongvec_minmax(table->data, table, write,
2567 buffer, lenp, ppos, convmul, convdiv);
2568}
2569
2570/**
2571 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2572 * @table: the sysctl table
2573 * @write: %TRUE if this is a write to the sysctl file
2574 * @buffer: the user buffer
2575 * @lenp: the size of the user buffer
2576 * @ppos: file position
2577 *
2578 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2579 * values from/to the user buffer, treated as an ASCII string.
2580 *
2581 * This routine will ensure the values are within the range specified by
2582 * table->extra1 (min) and table->extra2 (max).
2583 *
2584 * Returns 0 on success.
2585 */
2586int proc_doulongvec_minmax(struct ctl_table *table, int write,
2587 void __user *buffer, size_t *lenp, loff_t *ppos)
2588{
2589 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2590}
2591
2592/**
2593 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2594 * @table: the sysctl table
2595 * @write: %TRUE if this is a write to the sysctl file
2596 * @buffer: the user buffer
2597 * @lenp: the size of the user buffer
2598 * @ppos: file position
2599 *
2600 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2601 * values from/to the user buffer, treated as an ASCII string. The values
2602 * are treated as milliseconds, and converted to jiffies when they are stored.
2603 *
2604 * This routine will ensure the values are within the range specified by
2605 * table->extra1 (min) and table->extra2 (max).
2606 *
2607 * Returns 0 on success.
2608 */
2609int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2610 void __user *buffer,
2611 size_t *lenp, loff_t *ppos)
2612{
2613 return do_proc_doulongvec_minmax(table, write, buffer,
2614 lenp, ppos, HZ, 1000l);
2615}
2616
2617
2618static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2619 int *valp,
2620 int write, void *data)
2621{
2622 if (write) {
2623 if (*lvalp > LONG_MAX / HZ)
2624 return 1;
2625 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2626 } else {
2627 int val = *valp;
2628 unsigned long lval;
2629 if (val < 0) {
2630 *negp = true;
2631 lval = (unsigned long)-val;
2632 } else {
2633 *negp = false;
2634 lval = (unsigned long)val;
2635 }
2636 *lvalp = lval / HZ;
2637 }
2638 return 0;
2639}
2640
2641static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2642 int *valp,
2643 int write, void *data)
2644{
2645 if (write) {
2646 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2647 return 1;
2648 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2649 } else {
2650 int val = *valp;
2651 unsigned long lval;
2652 if (val < 0) {
2653 *negp = true;
2654 lval = (unsigned long)-val;
2655 } else {
2656 *negp = false;
2657 lval = (unsigned long)val;
2658 }
2659 *lvalp = jiffies_to_clock_t(lval);
2660 }
2661 return 0;
2662}
2663
2664static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2665 int *valp,
2666 int write, void *data)
2667{
2668 if (write) {
2669 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2670 } else {
2671 int val = *valp;
2672 unsigned long lval;
2673 if (val < 0) {
2674 *negp = true;
2675 lval = (unsigned long)-val;
2676 } else {
2677 *negp = false;
2678 lval = (unsigned long)val;
2679 }
2680 *lvalp = jiffies_to_msecs(lval);
2681 }
2682 return 0;
2683}
2684
2685/**
2686 * proc_dointvec_jiffies - read a vector of integers as seconds
2687 * @table: the sysctl table
2688 * @write: %TRUE if this is a write to the sysctl file
2689 * @buffer: the user buffer
2690 * @lenp: the size of the user buffer
2691 * @ppos: file position
2692 *
2693 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2694 * values from/to the user buffer, treated as an ASCII string.
2695 * The values read are assumed to be in seconds, and are converted into
2696 * jiffies.
2697 *
2698 * Returns 0 on success.
2699 */
2700int proc_dointvec_jiffies(struct ctl_table *table, int write,
2701 void __user *buffer, size_t *lenp, loff_t *ppos)
2702{
2703 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2704 do_proc_dointvec_jiffies_conv,NULL);
2705}
2706
2707/**
2708 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2709 * @table: the sysctl table
2710 * @write: %TRUE if this is a write to the sysctl file
2711 * @buffer: the user buffer
2712 * @lenp: the size of the user buffer
2713 * @ppos: pointer to the file position
2714 *
2715 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2716 * values from/to the user buffer, treated as an ASCII string.
2717 * The values read are assumed to be in 1/USER_HZ seconds, and
2718 * are converted into jiffies.
2719 *
2720 * Returns 0 on success.
2721 */
2722int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2723 void __user *buffer, size_t *lenp, loff_t *ppos)
2724{
2725 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2726 do_proc_dointvec_userhz_jiffies_conv,NULL);
2727}
2728
2729/**
2730 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2731 * @table: the sysctl table
2732 * @write: %TRUE if this is a write to the sysctl file
2733 * @buffer: the user buffer
2734 * @lenp: the size of the user buffer
2735 * @ppos: file position
2736 * @ppos: the current position in the file
2737 *
2738 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2739 * values from/to the user buffer, treated as an ASCII string.
2740 * The values read are assumed to be in 1/1000 seconds, and
2741 * are converted into jiffies.
2742 *
2743 * Returns 0 on success.
2744 */
2745int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2746 void __user *buffer, size_t *lenp, loff_t *ppos)
2747{
2748 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2749 do_proc_dointvec_ms_jiffies_conv, NULL);
2750}
2751
2752static int proc_do_cad_pid(struct ctl_table *table, int write,
2753 void __user *buffer, size_t *lenp, loff_t *ppos)
2754{
2755 struct pid *new_pid;
2756 pid_t tmp;
2757 int r;
2758
2759 tmp = pid_vnr(cad_pid);
2760
2761 r = __do_proc_dointvec(&tmp, table, write, buffer,
2762 lenp, ppos, NULL, NULL);
2763 if (r || !write)
2764 return r;
2765
2766 new_pid = find_get_pid(tmp);
2767 if (!new_pid)
2768 return -ESRCH;
2769
2770 put_pid(xchg(&cad_pid, new_pid));
2771 return 0;
2772}
2773
2774/**
2775 * proc_do_large_bitmap - read/write from/to a large bitmap
2776 * @table: the sysctl table
2777 * @write: %TRUE if this is a write to the sysctl file
2778 * @buffer: the user buffer
2779 * @lenp: the size of the user buffer
2780 * @ppos: file position
2781 *
2782 * The bitmap is stored at table->data and the bitmap length (in bits)
2783 * in table->maxlen.
2784 *
2785 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2786 * large bitmaps may be represented in a compact manner. Writing into
2787 * the file will clear the bitmap then update it with the given input.
2788 *
2789 * Returns 0 on success.
2790 */
2791int proc_do_large_bitmap(struct ctl_table *table, int write,
2792 void __user *buffer, size_t *lenp, loff_t *ppos)
2793{
2794 int err = 0;
2795 bool first = 1;
2796 size_t left = *lenp;
2797 unsigned long bitmap_len = table->maxlen;
2798 unsigned long *bitmap = (unsigned long *) table->data;
2799 unsigned long *tmp_bitmap = NULL;
2800 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2801
2802 if (!bitmap_len || !left || (*ppos && !write)) {
2803 *lenp = 0;
2804 return 0;
2805 }
2806
2807 if (write) {
2808 unsigned long page = 0;
2809 char *kbuf;
2810
2811 if (left > PAGE_SIZE - 1)
2812 left = PAGE_SIZE - 1;
2813
2814 page = __get_free_page(GFP_TEMPORARY);
2815 kbuf = (char *) page;
2816 if (!kbuf)
2817 return -ENOMEM;
2818 if (copy_from_user(kbuf, buffer, left)) {
2819 free_page(page);
2820 return -EFAULT;
2821 }
2822 kbuf[left] = 0;
2823
2824 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2825 GFP_KERNEL);
2826 if (!tmp_bitmap) {
2827 free_page(page);
2828 return -ENOMEM;
2829 }
2830 proc_skip_char(&kbuf, &left, '\n');
2831 while (!err && left) {
2832 unsigned long val_a, val_b;
2833 bool neg;
2834
2835 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2836 sizeof(tr_a), &c);
2837 if (err)
2838 break;
2839 if (val_a >= bitmap_len || neg) {
2840 err = -EINVAL;
2841 break;
2842 }
2843
2844 val_b = val_a;
2845 if (left) {
2846 kbuf++;
2847 left--;
2848 }
2849
2850 if (c == '-') {
2851 err = proc_get_long(&kbuf, &left, &val_b,
2852 &neg, tr_b, sizeof(tr_b),
2853 &c);
2854 if (err)
2855 break;
2856 if (val_b >= bitmap_len || neg ||
2857 val_a > val_b) {
2858 err = -EINVAL;
2859 break;
2860 }
2861 if (left) {
2862 kbuf++;
2863 left--;
2864 }
2865 }
2866
2867 while (val_a <= val_b)
2868 set_bit(val_a++, tmp_bitmap);
2869
2870 first = 0;
2871 proc_skip_char(&kbuf, &left, '\n');
2872 }
2873 free_page(page);
2874 } else {
2875 unsigned long bit_a, bit_b = 0;
2876
2877 while (left) {
2878 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2879 if (bit_a >= bitmap_len)
2880 break;
2881 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2882 bit_a + 1) - 1;
2883
2884 if (!first) {
2885 err = proc_put_char(&buffer, &left, ',');
2886 if (err)
2887 break;
2888 }
2889 err = proc_put_long(&buffer, &left, bit_a, false);
2890 if (err)
2891 break;
2892 if (bit_a != bit_b) {
2893 err = proc_put_char(&buffer, &left, '-');
2894 if (err)
2895 break;
2896 err = proc_put_long(&buffer, &left, bit_b, false);
2897 if (err)
2898 break;
2899 }
2900
2901 first = 0; bit_b++;
2902 }
2903 if (!err)
2904 err = proc_put_char(&buffer, &left, '\n');
2905 }
2906
2907 if (!err) {
2908 if (write) {
2909 if (*ppos)
2910 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2911 else
2912 memcpy(bitmap, tmp_bitmap,
2913 BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long));
2914 }
2915 kfree(tmp_bitmap);
2916 *lenp -= left;
2917 *ppos += *lenp;
2918 return 0;
2919 } else {
2920 kfree(tmp_bitmap);
2921 return err;
2922 }
2923}
2924
2925#else /* CONFIG_PROC_SYSCTL */
2926
2927int proc_dostring(struct ctl_table *table, int write,
2928 void __user *buffer, size_t *lenp, loff_t *ppos)
2929{
2930 return -ENOSYS;
2931}
2932
2933int proc_dointvec(struct ctl_table *table, int write,
2934 void __user *buffer, size_t *lenp, loff_t *ppos)
2935{
2936 return -ENOSYS;
2937}
2938
2939int proc_dointvec_minmax(struct ctl_table *table, int write,
2940 void __user *buffer, size_t *lenp, loff_t *ppos)
2941{
2942 return -ENOSYS;
2943}
2944
2945int proc_dointvec_jiffies(struct ctl_table *table, int write,
2946 void __user *buffer, size_t *lenp, loff_t *ppos)
2947{
2948 return -ENOSYS;
2949}
2950
2951int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2952 void __user *buffer, size_t *lenp, loff_t *ppos)
2953{
2954 return -ENOSYS;
2955}
2956
2957int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2958 void __user *buffer, size_t *lenp, loff_t *ppos)
2959{
2960 return -ENOSYS;
2961}
2962
2963int proc_doulongvec_minmax(struct ctl_table *table, int write,
2964 void __user *buffer, size_t *lenp, loff_t *ppos)
2965{
2966 return -ENOSYS;
2967}
2968
2969int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2970 void __user *buffer,
2971 size_t *lenp, loff_t *ppos)
2972{
2973 return -ENOSYS;
2974}
2975
2976
2977#endif /* CONFIG_PROC_SYSCTL */
2978
2979/*
2980 * No sense putting this after each symbol definition, twice,
2981 * exception granted :-)
2982 */
2983EXPORT_SYMBOL(proc_dointvec);
2984EXPORT_SYMBOL(proc_dointvec_jiffies);
2985EXPORT_SYMBOL(proc_dointvec_minmax);
2986EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2987EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2988EXPORT_SYMBOL(proc_dostring);
2989EXPORT_SYMBOL(proc_doulongvec_minmax);
2990EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2991EXPORT_SYMBOL(register_sysctl_table);
2992EXPORT_SYMBOL(register_sysctl_paths);
2993EXPORT_SYMBOL(unregister_sysctl_table);