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