]> git.proxmox.com Git - mirror_ubuntu-jammy-kernel.git/blob - kernel/sysctl.c
Merge tag 'block-5.13-2021-06-03' of git://git.kernel.dk/linux-block
[mirror_ubuntu-jammy-kernel.git] / kernel / sysctl.c
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
111 static int sixty = 60;
112 #endif
113
114 static int __maybe_unused neg_one = -1;
115 static int __maybe_unused two = 2;
116 static int __maybe_unused four = 4;
117 static unsigned long zero_ul;
118 static unsigned long one_ul = 1;
119 static unsigned long long_max = LONG_MAX;
120 static int one_hundred = 100;
121 static int two_hundred = 200;
122 static int one_thousand = 1000;
123 #ifdef CONFIG_PRINTK
124 static int ten_thousand = 10000;
125 #endif
126 #ifdef CONFIG_PERF_EVENTS
127 static int six_hundred_forty_kb = 640 * 1024;
128 #endif
129
130 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
131 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
132
133 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
134 static int maxolduid = 65535;
135 static int minolduid;
136
137 static int ngroups_max = NGROUPS_MAX;
138 static 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
145 static 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 #ifdef CONFIG_FANOTIFY
152 #include <linux/fanotify.h>
153 #endif
154
155 #ifdef CONFIG_PROC_SYSCTL
156
157 /**
158 * enum sysctl_writes_mode - supported sysctl write modes
159 *
160 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
161 * to be written, and multiple writes on the same sysctl file descriptor
162 * will rewrite the sysctl value, regardless of file position. No warning
163 * is issued when the initial position is not 0.
164 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
165 * not 0.
166 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
167 * file position 0 and the value must be fully contained in the buffer
168 * sent to the write syscall. If dealing with strings respect the file
169 * position, but restrict this to the max length of the buffer, anything
170 * passed the max length will be ignored. Multiple writes will append
171 * to the buffer.
172 *
173 * These write modes control how current file position affects the behavior of
174 * updating sysctl values through the proc interface on each write.
175 */
176 enum sysctl_writes_mode {
177 SYSCTL_WRITES_LEGACY = -1,
178 SYSCTL_WRITES_WARN = 0,
179 SYSCTL_WRITES_STRICT = 1,
180 };
181
182 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
183 #endif /* CONFIG_PROC_SYSCTL */
184
185 #if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
186 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
187 int sysctl_legacy_va_layout;
188 #endif
189
190 #ifdef CONFIG_COMPACTION
191 static int min_extfrag_threshold;
192 static int max_extfrag_threshold = 1000;
193 #endif
194
195 #endif /* CONFIG_SYSCTL */
196
197 #if defined(CONFIG_BPF_SYSCALL) && defined(CONFIG_SYSCTL)
198 static int bpf_stats_handler(struct ctl_table *table, int write,
199 void *buffer, size_t *lenp, loff_t *ppos)
200 {
201 struct static_key *key = (struct static_key *)table->data;
202 static int saved_val;
203 int val, ret;
204 struct ctl_table tmp = {
205 .data = &val,
206 .maxlen = sizeof(val),
207 .mode = table->mode,
208 .extra1 = SYSCTL_ZERO,
209 .extra2 = SYSCTL_ONE,
210 };
211
212 if (write && !capable(CAP_SYS_ADMIN))
213 return -EPERM;
214
215 mutex_lock(&bpf_stats_enabled_mutex);
216 val = saved_val;
217 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
218 if (write && !ret && val != saved_val) {
219 if (val)
220 static_key_slow_inc(key);
221 else
222 static_key_slow_dec(key);
223 saved_val = val;
224 }
225 mutex_unlock(&bpf_stats_enabled_mutex);
226 return ret;
227 }
228
229 static int bpf_unpriv_handler(struct ctl_table *table, int write,
230 void *buffer, size_t *lenp, loff_t *ppos)
231 {
232 int ret, unpriv_enable = *(int *)table->data;
233 bool locked_state = unpriv_enable == 1;
234 struct ctl_table tmp = *table;
235
236 if (write && !capable(CAP_SYS_ADMIN))
237 return -EPERM;
238
239 tmp.data = &unpriv_enable;
240 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
241 if (write && !ret) {
242 if (locked_state && unpriv_enable != 1)
243 return -EPERM;
244 *(int *)table->data = unpriv_enable;
245 }
246 return ret;
247 }
248 #endif /* CONFIG_BPF_SYSCALL && CONFIG_SYSCTL */
249
250 /*
251 * /proc/sys support
252 */
253
254 #ifdef CONFIG_PROC_SYSCTL
255
256 static int _proc_do_string(char *data, int maxlen, int write,
257 char *buffer, size_t *lenp, loff_t *ppos)
258 {
259 size_t len;
260 char c, *p;
261
262 if (!data || !maxlen || !*lenp) {
263 *lenp = 0;
264 return 0;
265 }
266
267 if (write) {
268 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
269 /* Only continue writes not past the end of buffer. */
270 len = strlen(data);
271 if (len > maxlen - 1)
272 len = maxlen - 1;
273
274 if (*ppos > len)
275 return 0;
276 len = *ppos;
277 } else {
278 /* Start writing from beginning of buffer. */
279 len = 0;
280 }
281
282 *ppos += *lenp;
283 p = buffer;
284 while ((p - buffer) < *lenp && len < maxlen - 1) {
285 c = *(p++);
286 if (c == 0 || c == '\n')
287 break;
288 data[len++] = c;
289 }
290 data[len] = 0;
291 } else {
292 len = strlen(data);
293 if (len > maxlen)
294 len = maxlen;
295
296 if (*ppos > len) {
297 *lenp = 0;
298 return 0;
299 }
300
301 data += *ppos;
302 len -= *ppos;
303
304 if (len > *lenp)
305 len = *lenp;
306 if (len)
307 memcpy(buffer, data, len);
308 if (len < *lenp) {
309 buffer[len] = '\n';
310 len++;
311 }
312 *lenp = len;
313 *ppos += len;
314 }
315 return 0;
316 }
317
318 static void warn_sysctl_write(struct ctl_table *table)
319 {
320 pr_warn_once("%s wrote to %s when file position was not 0!\n"
321 "This will not be supported in the future. To silence this\n"
322 "warning, set kernel.sysctl_writes_strict = -1\n",
323 current->comm, table->procname);
324 }
325
326 /**
327 * proc_first_pos_non_zero_ignore - check if first position is allowed
328 * @ppos: file position
329 * @table: the sysctl table
330 *
331 * Returns true if the first position is non-zero and the sysctl_writes_strict
332 * mode indicates this is not allowed for numeric input types. String proc
333 * handlers can ignore the return value.
334 */
335 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
336 struct ctl_table *table)
337 {
338 if (!*ppos)
339 return false;
340
341 switch (sysctl_writes_strict) {
342 case SYSCTL_WRITES_STRICT:
343 return true;
344 case SYSCTL_WRITES_WARN:
345 warn_sysctl_write(table);
346 return false;
347 default:
348 return false;
349 }
350 }
351
352 /**
353 * proc_dostring - read a string sysctl
354 * @table: the sysctl table
355 * @write: %TRUE if this is a write to the sysctl file
356 * @buffer: the user buffer
357 * @lenp: the size of the user buffer
358 * @ppos: file position
359 *
360 * Reads/writes a string from/to the user buffer. If the kernel
361 * buffer provided is not large enough to hold the string, the
362 * string is truncated. The copied string is %NULL-terminated.
363 * If the string is being read by the user process, it is copied
364 * and a newline '\n' is added. It is truncated if the buffer is
365 * not large enough.
366 *
367 * Returns 0 on success.
368 */
369 int proc_dostring(struct ctl_table *table, int write,
370 void *buffer, size_t *lenp, loff_t *ppos)
371 {
372 if (write)
373 proc_first_pos_non_zero_ignore(ppos, table);
374
375 return _proc_do_string(table->data, table->maxlen, write, buffer, lenp,
376 ppos);
377 }
378
379 static size_t proc_skip_spaces(char **buf)
380 {
381 size_t ret;
382 char *tmp = skip_spaces(*buf);
383 ret = tmp - *buf;
384 *buf = tmp;
385 return ret;
386 }
387
388 static void proc_skip_char(char **buf, size_t *size, const char v)
389 {
390 while (*size) {
391 if (**buf != v)
392 break;
393 (*size)--;
394 (*buf)++;
395 }
396 }
397
398 /**
399 * strtoul_lenient - parse an ASCII formatted integer from a buffer and only
400 * fail on overflow
401 *
402 * @cp: kernel buffer containing the string to parse
403 * @endp: pointer to store the trailing characters
404 * @base: the base to use
405 * @res: where the parsed integer will be stored
406 *
407 * In case of success 0 is returned and @res will contain the parsed integer,
408 * @endp will hold any trailing characters.
409 * This function will fail the parse on overflow. If there wasn't an overflow
410 * the function will defer the decision what characters count as invalid to the
411 * caller.
412 */
413 static int strtoul_lenient(const char *cp, char **endp, unsigned int base,
414 unsigned long *res)
415 {
416 unsigned long long result;
417 unsigned int rv;
418
419 cp = _parse_integer_fixup_radix(cp, &base);
420 rv = _parse_integer(cp, base, &result);
421 if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result))
422 return -ERANGE;
423
424 cp += rv;
425
426 if (endp)
427 *endp = (char *)cp;
428
429 *res = (unsigned long)result;
430 return 0;
431 }
432
433 #define TMPBUFLEN 22
434 /**
435 * proc_get_long - reads an ASCII formatted integer from a user buffer
436 *
437 * @buf: a kernel buffer
438 * @size: size of the kernel buffer
439 * @val: this is where the number will be stored
440 * @neg: set to %TRUE if number is negative
441 * @perm_tr: a vector which contains the allowed trailers
442 * @perm_tr_len: size of the perm_tr vector
443 * @tr: pointer to store the trailer character
444 *
445 * In case of success %0 is returned and @buf and @size are updated with
446 * the amount of bytes read. If @tr is non-NULL and a trailing
447 * character exists (size is non-zero after returning from this
448 * function), @tr is updated with the trailing character.
449 */
450 static int proc_get_long(char **buf, size_t *size,
451 unsigned long *val, bool *neg,
452 const char *perm_tr, unsigned perm_tr_len, char *tr)
453 {
454 int len;
455 char *p, tmp[TMPBUFLEN];
456
457 if (!*size)
458 return -EINVAL;
459
460 len = *size;
461 if (len > TMPBUFLEN - 1)
462 len = TMPBUFLEN - 1;
463
464 memcpy(tmp, *buf, len);
465
466 tmp[len] = 0;
467 p = tmp;
468 if (*p == '-' && *size > 1) {
469 *neg = true;
470 p++;
471 } else
472 *neg = false;
473 if (!isdigit(*p))
474 return -EINVAL;
475
476 if (strtoul_lenient(p, &p, 0, val))
477 return -EINVAL;
478
479 len = p - tmp;
480
481 /* We don't know if the next char is whitespace thus we may accept
482 * invalid integers (e.g. 1234...a) or two integers instead of one
483 * (e.g. 123...1). So lets not allow such large numbers. */
484 if (len == TMPBUFLEN - 1)
485 return -EINVAL;
486
487 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
488 return -EINVAL;
489
490 if (tr && (len < *size))
491 *tr = *p;
492
493 *buf += len;
494 *size -= len;
495
496 return 0;
497 }
498
499 /**
500 * proc_put_long - converts an integer to a decimal ASCII formatted string
501 *
502 * @buf: the user buffer
503 * @size: the size of the user buffer
504 * @val: the integer to be converted
505 * @neg: sign of the number, %TRUE for negative
506 *
507 * In case of success @buf and @size are updated with the amount of bytes
508 * written.
509 */
510 static void proc_put_long(void **buf, size_t *size, unsigned long val, bool neg)
511 {
512 int len;
513 char tmp[TMPBUFLEN], *p = tmp;
514
515 sprintf(p, "%s%lu", neg ? "-" : "", val);
516 len = strlen(tmp);
517 if (len > *size)
518 len = *size;
519 memcpy(*buf, tmp, len);
520 *size -= len;
521 *buf += len;
522 }
523 #undef TMPBUFLEN
524
525 static void proc_put_char(void **buf, size_t *size, char c)
526 {
527 if (*size) {
528 char **buffer = (char **)buf;
529 **buffer = c;
530
531 (*size)--;
532 (*buffer)++;
533 *buf = *buffer;
534 }
535 }
536
537 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
538 int *valp,
539 int write, void *data)
540 {
541 if (write) {
542 if (*negp) {
543 if (*lvalp > (unsigned long) INT_MAX + 1)
544 return -EINVAL;
545 *valp = -*lvalp;
546 } else {
547 if (*lvalp > (unsigned long) INT_MAX)
548 return -EINVAL;
549 *valp = *lvalp;
550 }
551 } else {
552 int val = *valp;
553 if (val < 0) {
554 *negp = true;
555 *lvalp = -(unsigned long)val;
556 } else {
557 *negp = false;
558 *lvalp = (unsigned long)val;
559 }
560 }
561 return 0;
562 }
563
564 static int do_proc_douintvec_conv(unsigned long *lvalp,
565 unsigned int *valp,
566 int write, void *data)
567 {
568 if (write) {
569 if (*lvalp > UINT_MAX)
570 return -EINVAL;
571 *valp = *lvalp;
572 } else {
573 unsigned int val = *valp;
574 *lvalp = (unsigned long)val;
575 }
576 return 0;
577 }
578
579 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
580
581 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
582 int write, void *buffer,
583 size_t *lenp, loff_t *ppos,
584 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
585 int write, void *data),
586 void *data)
587 {
588 int *i, vleft, first = 1, err = 0;
589 size_t left;
590 char *p;
591
592 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
593 *lenp = 0;
594 return 0;
595 }
596
597 i = (int *) tbl_data;
598 vleft = table->maxlen / sizeof(*i);
599 left = *lenp;
600
601 if (!conv)
602 conv = do_proc_dointvec_conv;
603
604 if (write) {
605 if (proc_first_pos_non_zero_ignore(ppos, table))
606 goto out;
607
608 if (left > PAGE_SIZE - 1)
609 left = PAGE_SIZE - 1;
610 p = buffer;
611 }
612
613 for (; left && vleft--; i++, first=0) {
614 unsigned long lval;
615 bool neg;
616
617 if (write) {
618 left -= proc_skip_spaces(&p);
619
620 if (!left)
621 break;
622 err = proc_get_long(&p, &left, &lval, &neg,
623 proc_wspace_sep,
624 sizeof(proc_wspace_sep), NULL);
625 if (err)
626 break;
627 if (conv(&neg, &lval, i, 1, data)) {
628 err = -EINVAL;
629 break;
630 }
631 } else {
632 if (conv(&neg, &lval, i, 0, data)) {
633 err = -EINVAL;
634 break;
635 }
636 if (!first)
637 proc_put_char(&buffer, &left, '\t');
638 proc_put_long(&buffer, &left, lval, neg);
639 }
640 }
641
642 if (!write && !first && left && !err)
643 proc_put_char(&buffer, &left, '\n');
644 if (write && !err && left)
645 left -= proc_skip_spaces(&p);
646 if (write && first)
647 return err ? : -EINVAL;
648 *lenp -= left;
649 out:
650 *ppos += *lenp;
651 return err;
652 }
653
654 static int do_proc_dointvec(struct ctl_table *table, int write,
655 void *buffer, size_t *lenp, loff_t *ppos,
656 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
657 int write, void *data),
658 void *data)
659 {
660 return __do_proc_dointvec(table->data, table, write,
661 buffer, lenp, ppos, conv, data);
662 }
663
664 static int do_proc_douintvec_w(unsigned int *tbl_data,
665 struct ctl_table *table,
666 void *buffer,
667 size_t *lenp, loff_t *ppos,
668 int (*conv)(unsigned long *lvalp,
669 unsigned int *valp,
670 int write, void *data),
671 void *data)
672 {
673 unsigned long lval;
674 int err = 0;
675 size_t left;
676 bool neg;
677 char *p = buffer;
678
679 left = *lenp;
680
681 if (proc_first_pos_non_zero_ignore(ppos, table))
682 goto bail_early;
683
684 if (left > PAGE_SIZE - 1)
685 left = PAGE_SIZE - 1;
686
687 left -= proc_skip_spaces(&p);
688 if (!left) {
689 err = -EINVAL;
690 goto out_free;
691 }
692
693 err = proc_get_long(&p, &left, &lval, &neg,
694 proc_wspace_sep,
695 sizeof(proc_wspace_sep), NULL);
696 if (err || neg) {
697 err = -EINVAL;
698 goto out_free;
699 }
700
701 if (conv(&lval, tbl_data, 1, data)) {
702 err = -EINVAL;
703 goto out_free;
704 }
705
706 if (!err && left)
707 left -= proc_skip_spaces(&p);
708
709 out_free:
710 if (err)
711 return -EINVAL;
712
713 return 0;
714
715 /* This is in keeping with old __do_proc_dointvec() */
716 bail_early:
717 *ppos += *lenp;
718 return err;
719 }
720
721 static int do_proc_douintvec_r(unsigned int *tbl_data, void *buffer,
722 size_t *lenp, loff_t *ppos,
723 int (*conv)(unsigned long *lvalp,
724 unsigned int *valp,
725 int write, void *data),
726 void *data)
727 {
728 unsigned long lval;
729 int err = 0;
730 size_t left;
731
732 left = *lenp;
733
734 if (conv(&lval, tbl_data, 0, data)) {
735 err = -EINVAL;
736 goto out;
737 }
738
739 proc_put_long(&buffer, &left, lval, false);
740 if (!left)
741 goto out;
742
743 proc_put_char(&buffer, &left, '\n');
744
745 out:
746 *lenp -= left;
747 *ppos += *lenp;
748
749 return err;
750 }
751
752 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
753 int write, void *buffer,
754 size_t *lenp, loff_t *ppos,
755 int (*conv)(unsigned long *lvalp,
756 unsigned int *valp,
757 int write, void *data),
758 void *data)
759 {
760 unsigned int *i, vleft;
761
762 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
763 *lenp = 0;
764 return 0;
765 }
766
767 i = (unsigned int *) tbl_data;
768 vleft = table->maxlen / sizeof(*i);
769
770 /*
771 * Arrays are not supported, keep this simple. *Do not* add
772 * support for them.
773 */
774 if (vleft != 1) {
775 *lenp = 0;
776 return -EINVAL;
777 }
778
779 if (!conv)
780 conv = do_proc_douintvec_conv;
781
782 if (write)
783 return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
784 conv, data);
785 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
786 }
787
788 static int do_proc_douintvec(struct ctl_table *table, int write,
789 void *buffer, size_t *lenp, loff_t *ppos,
790 int (*conv)(unsigned long *lvalp,
791 unsigned int *valp,
792 int write, void *data),
793 void *data)
794 {
795 return __do_proc_douintvec(table->data, table, write,
796 buffer, lenp, ppos, conv, data);
797 }
798
799 /**
800 * proc_dointvec - read a vector of integers
801 * @table: the sysctl table
802 * @write: %TRUE if this is a write to the sysctl file
803 * @buffer: the user buffer
804 * @lenp: the size of the user buffer
805 * @ppos: file position
806 *
807 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
808 * values from/to the user buffer, treated as an ASCII string.
809 *
810 * Returns 0 on success.
811 */
812 int proc_dointvec(struct ctl_table *table, int write, void *buffer,
813 size_t *lenp, loff_t *ppos)
814 {
815 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
816 }
817
818 #ifdef CONFIG_COMPACTION
819 static int proc_dointvec_minmax_warn_RT_change(struct ctl_table *table,
820 int write, void *buffer, size_t *lenp, loff_t *ppos)
821 {
822 int ret, old;
823
824 if (!IS_ENABLED(CONFIG_PREEMPT_RT) || !write)
825 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
826
827 old = *(int *)table->data;
828 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
829 if (ret)
830 return ret;
831 if (old != *(int *)table->data)
832 pr_warn_once("sysctl attribute %s changed by %s[%d]\n",
833 table->procname, current->comm,
834 task_pid_nr(current));
835 return ret;
836 }
837 #endif
838
839 /**
840 * proc_douintvec - read a vector of unsigned integers
841 * @table: the sysctl table
842 * @write: %TRUE if this is a write to the sysctl file
843 * @buffer: the user buffer
844 * @lenp: the size of the user buffer
845 * @ppos: file position
846 *
847 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
848 * values from/to the user buffer, treated as an ASCII string.
849 *
850 * Returns 0 on success.
851 */
852 int proc_douintvec(struct ctl_table *table, int write, void *buffer,
853 size_t *lenp, loff_t *ppos)
854 {
855 return do_proc_douintvec(table, write, buffer, lenp, ppos,
856 do_proc_douintvec_conv, NULL);
857 }
858
859 /*
860 * Taint values can only be increased
861 * This means we can safely use a temporary.
862 */
863 static int proc_taint(struct ctl_table *table, int write,
864 void *buffer, size_t *lenp, loff_t *ppos)
865 {
866 struct ctl_table t;
867 unsigned long tmptaint = get_taint();
868 int err;
869
870 if (write && !capable(CAP_SYS_ADMIN))
871 return -EPERM;
872
873 t = *table;
874 t.data = &tmptaint;
875 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
876 if (err < 0)
877 return err;
878
879 if (write) {
880 int i;
881
882 /*
883 * If we are relying on panic_on_taint not producing
884 * false positives due to userspace input, bail out
885 * before setting the requested taint flags.
886 */
887 if (panic_on_taint_nousertaint && (tmptaint & panic_on_taint))
888 return -EINVAL;
889
890 /*
891 * Poor man's atomic or. Not worth adding a primitive
892 * to everyone's atomic.h for this
893 */
894 for (i = 0; i < TAINT_FLAGS_COUNT; i++)
895 if ((1UL << i) & tmptaint)
896 add_taint(i, LOCKDEP_STILL_OK);
897 }
898
899 return err;
900 }
901
902 #ifdef CONFIG_PRINTK
903 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
904 void *buffer, size_t *lenp, loff_t *ppos)
905 {
906 if (write && !capable(CAP_SYS_ADMIN))
907 return -EPERM;
908
909 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
910 }
911 #endif
912
913 /**
914 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
915 * @min: pointer to minimum allowable value
916 * @max: pointer to maximum allowable value
917 *
918 * The do_proc_dointvec_minmax_conv_param structure provides the
919 * minimum and maximum values for doing range checking for those sysctl
920 * parameters that use the proc_dointvec_minmax() handler.
921 */
922 struct do_proc_dointvec_minmax_conv_param {
923 int *min;
924 int *max;
925 };
926
927 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
928 int *valp,
929 int write, void *data)
930 {
931 int tmp, ret;
932 struct do_proc_dointvec_minmax_conv_param *param = data;
933 /*
934 * If writing, first do so via a temporary local int so we can
935 * bounds-check it before touching *valp.
936 */
937 int *ip = write ? &tmp : valp;
938
939 ret = do_proc_dointvec_conv(negp, lvalp, ip, write, data);
940 if (ret)
941 return ret;
942
943 if (write) {
944 if ((param->min && *param->min > tmp) ||
945 (param->max && *param->max < tmp))
946 return -EINVAL;
947 *valp = tmp;
948 }
949
950 return 0;
951 }
952
953 /**
954 * proc_dointvec_minmax - read a vector of integers with min/max values
955 * @table: the sysctl table
956 * @write: %TRUE if this is a write to the sysctl file
957 * @buffer: the user buffer
958 * @lenp: the size of the user buffer
959 * @ppos: file position
960 *
961 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
962 * values from/to the user buffer, treated as an ASCII string.
963 *
964 * This routine will ensure the values are within the range specified by
965 * table->extra1 (min) and table->extra2 (max).
966 *
967 * Returns 0 on success or -EINVAL on write when the range check fails.
968 */
969 int proc_dointvec_minmax(struct ctl_table *table, int write,
970 void *buffer, size_t *lenp, loff_t *ppos)
971 {
972 struct do_proc_dointvec_minmax_conv_param param = {
973 .min = (int *) table->extra1,
974 .max = (int *) table->extra2,
975 };
976 return do_proc_dointvec(table, write, buffer, lenp, ppos,
977 do_proc_dointvec_minmax_conv, &param);
978 }
979
980 /**
981 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
982 * @min: pointer to minimum allowable value
983 * @max: pointer to maximum allowable value
984 *
985 * The do_proc_douintvec_minmax_conv_param structure provides the
986 * minimum and maximum values for doing range checking for those sysctl
987 * parameters that use the proc_douintvec_minmax() handler.
988 */
989 struct do_proc_douintvec_minmax_conv_param {
990 unsigned int *min;
991 unsigned int *max;
992 };
993
994 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
995 unsigned int *valp,
996 int write, void *data)
997 {
998 int ret;
999 unsigned int tmp;
1000 struct do_proc_douintvec_minmax_conv_param *param = data;
1001 /* write via temporary local uint for bounds-checking */
1002 unsigned int *up = write ? &tmp : valp;
1003
1004 ret = do_proc_douintvec_conv(lvalp, up, write, data);
1005 if (ret)
1006 return ret;
1007
1008 if (write) {
1009 if ((param->min && *param->min > tmp) ||
1010 (param->max && *param->max < tmp))
1011 return -ERANGE;
1012
1013 *valp = tmp;
1014 }
1015
1016 return 0;
1017 }
1018
1019 /**
1020 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
1021 * @table: the sysctl table
1022 * @write: %TRUE if this is a write to the sysctl file
1023 * @buffer: the user buffer
1024 * @lenp: the size of the user buffer
1025 * @ppos: file position
1026 *
1027 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
1028 * values from/to the user buffer, treated as an ASCII string. Negative
1029 * strings are not allowed.
1030 *
1031 * This routine will ensure the values are within the range specified by
1032 * table->extra1 (min) and table->extra2 (max). There is a final sanity
1033 * check for UINT_MAX to avoid having to support wrap around uses from
1034 * userspace.
1035 *
1036 * Returns 0 on success or -ERANGE on write when the range check fails.
1037 */
1038 int proc_douintvec_minmax(struct ctl_table *table, int write,
1039 void *buffer, size_t *lenp, loff_t *ppos)
1040 {
1041 struct do_proc_douintvec_minmax_conv_param param = {
1042 .min = (unsigned int *) table->extra1,
1043 .max = (unsigned int *) table->extra2,
1044 };
1045 return do_proc_douintvec(table, write, buffer, lenp, ppos,
1046 do_proc_douintvec_minmax_conv, &param);
1047 }
1048
1049 /**
1050 * proc_dou8vec_minmax - read a vector of unsigned chars with min/max values
1051 * @table: the sysctl table
1052 * @write: %TRUE if this is a write to the sysctl file
1053 * @buffer: the user buffer
1054 * @lenp: the size of the user buffer
1055 * @ppos: file position
1056 *
1057 * Reads/writes up to table->maxlen/sizeof(u8) unsigned chars
1058 * values from/to the user buffer, treated as an ASCII string. Negative
1059 * strings are not allowed.
1060 *
1061 * This routine will ensure the values are within the range specified by
1062 * table->extra1 (min) and table->extra2 (max).
1063 *
1064 * Returns 0 on success or an error on write when the range check fails.
1065 */
1066 int proc_dou8vec_minmax(struct ctl_table *table, int write,
1067 void *buffer, size_t *lenp, loff_t *ppos)
1068 {
1069 struct ctl_table tmp;
1070 unsigned int min = 0, max = 255U, val;
1071 u8 *data = table->data;
1072 struct do_proc_douintvec_minmax_conv_param param = {
1073 .min = &min,
1074 .max = &max,
1075 };
1076 int res;
1077
1078 /* Do not support arrays yet. */
1079 if (table->maxlen != sizeof(u8))
1080 return -EINVAL;
1081
1082 if (table->extra1) {
1083 min = *(unsigned int *) table->extra1;
1084 if (min > 255U)
1085 return -EINVAL;
1086 }
1087 if (table->extra2) {
1088 max = *(unsigned int *) table->extra2;
1089 if (max > 255U)
1090 return -EINVAL;
1091 }
1092
1093 tmp = *table;
1094
1095 tmp.maxlen = sizeof(val);
1096 tmp.data = &val;
1097 val = *data;
1098 res = do_proc_douintvec(&tmp, write, buffer, lenp, ppos,
1099 do_proc_douintvec_minmax_conv, &param);
1100 if (res)
1101 return res;
1102 if (write)
1103 *data = val;
1104 return 0;
1105 }
1106 EXPORT_SYMBOL_GPL(proc_dou8vec_minmax);
1107
1108 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
1109 unsigned int *valp,
1110 int write, void *data)
1111 {
1112 if (write) {
1113 unsigned int val;
1114
1115 val = round_pipe_size(*lvalp);
1116 if (val == 0)
1117 return -EINVAL;
1118
1119 *valp = val;
1120 } else {
1121 unsigned int val = *valp;
1122 *lvalp = (unsigned long) val;
1123 }
1124
1125 return 0;
1126 }
1127
1128 static int proc_dopipe_max_size(struct ctl_table *table, int write,
1129 void *buffer, size_t *lenp, loff_t *ppos)
1130 {
1131 return do_proc_douintvec(table, write, buffer, lenp, ppos,
1132 do_proc_dopipe_max_size_conv, NULL);
1133 }
1134
1135 static void validate_coredump_safety(void)
1136 {
1137 #ifdef CONFIG_COREDUMP
1138 if (suid_dumpable == SUID_DUMP_ROOT &&
1139 core_pattern[0] != '/' && core_pattern[0] != '|') {
1140 printk(KERN_WARNING
1141 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
1142 "Pipe handler or fully qualified core dump path required.\n"
1143 "Set kernel.core_pattern before fs.suid_dumpable.\n"
1144 );
1145 }
1146 #endif
1147 }
1148
1149 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
1150 void *buffer, size_t *lenp, loff_t *ppos)
1151 {
1152 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
1153 if (!error)
1154 validate_coredump_safety();
1155 return error;
1156 }
1157
1158 #ifdef CONFIG_COREDUMP
1159 static int proc_dostring_coredump(struct ctl_table *table, int write,
1160 void *buffer, size_t *lenp, loff_t *ppos)
1161 {
1162 int error = proc_dostring(table, write, buffer, lenp, ppos);
1163 if (!error)
1164 validate_coredump_safety();
1165 return error;
1166 }
1167 #endif
1168
1169 #ifdef CONFIG_MAGIC_SYSRQ
1170 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
1171 void *buffer, size_t *lenp, loff_t *ppos)
1172 {
1173 int tmp, ret;
1174
1175 tmp = sysrq_mask();
1176
1177 ret = __do_proc_dointvec(&tmp, table, write, buffer,
1178 lenp, ppos, NULL, NULL);
1179 if (ret || !write)
1180 return ret;
1181
1182 if (write)
1183 sysrq_toggle_support(tmp);
1184
1185 return 0;
1186 }
1187 #endif
1188
1189 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table,
1190 int write, void *buffer, size_t *lenp, loff_t *ppos,
1191 unsigned long convmul, unsigned long convdiv)
1192 {
1193 unsigned long *i, *min, *max;
1194 int vleft, first = 1, err = 0;
1195 size_t left;
1196 char *p;
1197
1198 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
1199 *lenp = 0;
1200 return 0;
1201 }
1202
1203 i = (unsigned long *) data;
1204 min = (unsigned long *) table->extra1;
1205 max = (unsigned long *) table->extra2;
1206 vleft = table->maxlen / sizeof(unsigned long);
1207 left = *lenp;
1208
1209 if (write) {
1210 if (proc_first_pos_non_zero_ignore(ppos, table))
1211 goto out;
1212
1213 if (left > PAGE_SIZE - 1)
1214 left = PAGE_SIZE - 1;
1215 p = buffer;
1216 }
1217
1218 for (; left && vleft--; i++, first = 0) {
1219 unsigned long val;
1220
1221 if (write) {
1222 bool neg;
1223
1224 left -= proc_skip_spaces(&p);
1225 if (!left)
1226 break;
1227
1228 err = proc_get_long(&p, &left, &val, &neg,
1229 proc_wspace_sep,
1230 sizeof(proc_wspace_sep), NULL);
1231 if (err)
1232 break;
1233 if (neg)
1234 continue;
1235 val = convmul * val / convdiv;
1236 if ((min && val < *min) || (max && val > *max)) {
1237 err = -EINVAL;
1238 break;
1239 }
1240 *i = val;
1241 } else {
1242 val = convdiv * (*i) / convmul;
1243 if (!first)
1244 proc_put_char(&buffer, &left, '\t');
1245 proc_put_long(&buffer, &left, val, false);
1246 }
1247 }
1248
1249 if (!write && !first && left && !err)
1250 proc_put_char(&buffer, &left, '\n');
1251 if (write && !err)
1252 left -= proc_skip_spaces(&p);
1253 if (write && first)
1254 return err ? : -EINVAL;
1255 *lenp -= left;
1256 out:
1257 *ppos += *lenp;
1258 return err;
1259 }
1260
1261 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
1262 void *buffer, size_t *lenp, loff_t *ppos, unsigned long convmul,
1263 unsigned long convdiv)
1264 {
1265 return __do_proc_doulongvec_minmax(table->data, table, write,
1266 buffer, lenp, ppos, convmul, convdiv);
1267 }
1268
1269 /**
1270 * proc_doulongvec_minmax - read a vector of long integers with min/max values
1271 * @table: the sysctl table
1272 * @write: %TRUE if this is a write to the sysctl file
1273 * @buffer: the user buffer
1274 * @lenp: the size of the user buffer
1275 * @ppos: file position
1276 *
1277 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1278 * values from/to the user buffer, treated as an ASCII string.
1279 *
1280 * This routine will ensure the values are within the range specified by
1281 * table->extra1 (min) and table->extra2 (max).
1282 *
1283 * Returns 0 on success.
1284 */
1285 int proc_doulongvec_minmax(struct ctl_table *table, int write,
1286 void *buffer, size_t *lenp, loff_t *ppos)
1287 {
1288 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
1289 }
1290
1291 /**
1292 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
1293 * @table: the sysctl table
1294 * @write: %TRUE if this is a write to the sysctl file
1295 * @buffer: the user buffer
1296 * @lenp: the size of the user buffer
1297 * @ppos: file position
1298 *
1299 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1300 * values from/to the user buffer, treated as an ASCII string. The values
1301 * are treated as milliseconds, and converted to jiffies when they are stored.
1302 *
1303 * This routine will ensure the values are within the range specified by
1304 * table->extra1 (min) and table->extra2 (max).
1305 *
1306 * Returns 0 on success.
1307 */
1308 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1309 void *buffer, size_t *lenp, loff_t *ppos)
1310 {
1311 return do_proc_doulongvec_minmax(table, write, buffer,
1312 lenp, ppos, HZ, 1000l);
1313 }
1314
1315
1316 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
1317 int *valp,
1318 int write, void *data)
1319 {
1320 if (write) {
1321 if (*lvalp > INT_MAX / HZ)
1322 return 1;
1323 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
1324 } else {
1325 int val = *valp;
1326 unsigned long lval;
1327 if (val < 0) {
1328 *negp = true;
1329 lval = -(unsigned long)val;
1330 } else {
1331 *negp = false;
1332 lval = (unsigned long)val;
1333 }
1334 *lvalp = lval / HZ;
1335 }
1336 return 0;
1337 }
1338
1339 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
1340 int *valp,
1341 int write, void *data)
1342 {
1343 if (write) {
1344 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
1345 return 1;
1346 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
1347 } else {
1348 int val = *valp;
1349 unsigned long lval;
1350 if (val < 0) {
1351 *negp = true;
1352 lval = -(unsigned long)val;
1353 } else {
1354 *negp = false;
1355 lval = (unsigned long)val;
1356 }
1357 *lvalp = jiffies_to_clock_t(lval);
1358 }
1359 return 0;
1360 }
1361
1362 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
1363 int *valp,
1364 int write, void *data)
1365 {
1366 if (write) {
1367 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
1368
1369 if (jif > INT_MAX)
1370 return 1;
1371 *valp = (int)jif;
1372 } else {
1373 int val = *valp;
1374 unsigned long lval;
1375 if (val < 0) {
1376 *negp = true;
1377 lval = -(unsigned long)val;
1378 } else {
1379 *negp = false;
1380 lval = (unsigned long)val;
1381 }
1382 *lvalp = jiffies_to_msecs(lval);
1383 }
1384 return 0;
1385 }
1386
1387 /**
1388 * proc_dointvec_jiffies - read a vector of integers as seconds
1389 * @table: the sysctl table
1390 * @write: %TRUE if this is a write to the sysctl file
1391 * @buffer: the user buffer
1392 * @lenp: the size of the user buffer
1393 * @ppos: file position
1394 *
1395 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1396 * values from/to the user buffer, treated as an ASCII string.
1397 * The values read are assumed to be in seconds, and are converted into
1398 * jiffies.
1399 *
1400 * Returns 0 on success.
1401 */
1402 int proc_dointvec_jiffies(struct ctl_table *table, int write,
1403 void *buffer, size_t *lenp, loff_t *ppos)
1404 {
1405 return do_proc_dointvec(table,write,buffer,lenp,ppos,
1406 do_proc_dointvec_jiffies_conv,NULL);
1407 }
1408
1409 /**
1410 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
1411 * @table: the sysctl table
1412 * @write: %TRUE if this is a write to the sysctl file
1413 * @buffer: the user buffer
1414 * @lenp: the size of the user buffer
1415 * @ppos: pointer to the file position
1416 *
1417 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1418 * values from/to the user buffer, treated as an ASCII string.
1419 * The values read are assumed to be in 1/USER_HZ seconds, and
1420 * are converted into jiffies.
1421 *
1422 * Returns 0 on success.
1423 */
1424 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1425 void *buffer, size_t *lenp, loff_t *ppos)
1426 {
1427 return do_proc_dointvec(table,write,buffer,lenp,ppos,
1428 do_proc_dointvec_userhz_jiffies_conv,NULL);
1429 }
1430
1431 /**
1432 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
1433 * @table: the sysctl table
1434 * @write: %TRUE if this is a write to the sysctl file
1435 * @buffer: the user buffer
1436 * @lenp: the size of the user buffer
1437 * @ppos: file position
1438 * @ppos: the current position in the file
1439 *
1440 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1441 * values from/to the user buffer, treated as an ASCII string.
1442 * The values read are assumed to be in 1/1000 seconds, and
1443 * are converted into jiffies.
1444 *
1445 * Returns 0 on success.
1446 */
1447 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, void *buffer,
1448 size_t *lenp, loff_t *ppos)
1449 {
1450 return do_proc_dointvec(table, write, buffer, lenp, ppos,
1451 do_proc_dointvec_ms_jiffies_conv, NULL);
1452 }
1453
1454 static int proc_do_cad_pid(struct ctl_table *table, int write, void *buffer,
1455 size_t *lenp, loff_t *ppos)
1456 {
1457 struct pid *new_pid;
1458 pid_t tmp;
1459 int r;
1460
1461 tmp = pid_vnr(cad_pid);
1462
1463 r = __do_proc_dointvec(&tmp, table, write, buffer,
1464 lenp, ppos, NULL, NULL);
1465 if (r || !write)
1466 return r;
1467
1468 new_pid = find_get_pid(tmp);
1469 if (!new_pid)
1470 return -ESRCH;
1471
1472 put_pid(xchg(&cad_pid, new_pid));
1473 return 0;
1474 }
1475
1476 /**
1477 * proc_do_large_bitmap - read/write from/to a large bitmap
1478 * @table: the sysctl table
1479 * @write: %TRUE if this is a write to the sysctl file
1480 * @buffer: the user buffer
1481 * @lenp: the size of the user buffer
1482 * @ppos: file position
1483 *
1484 * The bitmap is stored at table->data and the bitmap length (in bits)
1485 * in table->maxlen.
1486 *
1487 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
1488 * large bitmaps may be represented in a compact manner. Writing into
1489 * the file will clear the bitmap then update it with the given input.
1490 *
1491 * Returns 0 on success.
1492 */
1493 int proc_do_large_bitmap(struct ctl_table *table, int write,
1494 void *buffer, size_t *lenp, loff_t *ppos)
1495 {
1496 int err = 0;
1497 bool first = 1;
1498 size_t left = *lenp;
1499 unsigned long bitmap_len = table->maxlen;
1500 unsigned long *bitmap = *(unsigned long **) table->data;
1501 unsigned long *tmp_bitmap = NULL;
1502 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
1503
1504 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
1505 *lenp = 0;
1506 return 0;
1507 }
1508
1509 if (write) {
1510 char *p = buffer;
1511 size_t skipped = 0;
1512
1513 if (left > PAGE_SIZE - 1) {
1514 left = PAGE_SIZE - 1;
1515 /* How much of the buffer we'll skip this pass */
1516 skipped = *lenp - left;
1517 }
1518
1519 tmp_bitmap = bitmap_zalloc(bitmap_len, GFP_KERNEL);
1520 if (!tmp_bitmap)
1521 return -ENOMEM;
1522 proc_skip_char(&p, &left, '\n');
1523 while (!err && left) {
1524 unsigned long val_a, val_b;
1525 bool neg;
1526 size_t saved_left;
1527
1528 /* In case we stop parsing mid-number, we can reset */
1529 saved_left = left;
1530 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
1531 sizeof(tr_a), &c);
1532 /*
1533 * If we consumed the entirety of a truncated buffer or
1534 * only one char is left (may be a "-"), then stop here,
1535 * reset, & come back for more.
1536 */
1537 if ((left <= 1) && skipped) {
1538 left = saved_left;
1539 break;
1540 }
1541
1542 if (err)
1543 break;
1544 if (val_a >= bitmap_len || neg) {
1545 err = -EINVAL;
1546 break;
1547 }
1548
1549 val_b = val_a;
1550 if (left) {
1551 p++;
1552 left--;
1553 }
1554
1555 if (c == '-') {
1556 err = proc_get_long(&p, &left, &val_b,
1557 &neg, tr_b, sizeof(tr_b),
1558 &c);
1559 /*
1560 * If we consumed all of a truncated buffer or
1561 * then stop here, reset, & come back for more.
1562 */
1563 if (!left && skipped) {
1564 left = saved_left;
1565 break;
1566 }
1567
1568 if (err)
1569 break;
1570 if (val_b >= bitmap_len || neg ||
1571 val_a > val_b) {
1572 err = -EINVAL;
1573 break;
1574 }
1575 if (left) {
1576 p++;
1577 left--;
1578 }
1579 }
1580
1581 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
1582 first = 0;
1583 proc_skip_char(&p, &left, '\n');
1584 }
1585 left += skipped;
1586 } else {
1587 unsigned long bit_a, bit_b = 0;
1588
1589 while (left) {
1590 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
1591 if (bit_a >= bitmap_len)
1592 break;
1593 bit_b = find_next_zero_bit(bitmap, bitmap_len,
1594 bit_a + 1) - 1;
1595
1596 if (!first)
1597 proc_put_char(&buffer, &left, ',');
1598 proc_put_long(&buffer, &left, bit_a, false);
1599 if (bit_a != bit_b) {
1600 proc_put_char(&buffer, &left, '-');
1601 proc_put_long(&buffer, &left, bit_b, false);
1602 }
1603
1604 first = 0; bit_b++;
1605 }
1606 proc_put_char(&buffer, &left, '\n');
1607 }
1608
1609 if (!err) {
1610 if (write) {
1611 if (*ppos)
1612 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
1613 else
1614 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
1615 }
1616 *lenp -= left;
1617 *ppos += *lenp;
1618 }
1619
1620 bitmap_free(tmp_bitmap);
1621 return err;
1622 }
1623
1624 #else /* CONFIG_PROC_SYSCTL */
1625
1626 int proc_dostring(struct ctl_table *table, int write,
1627 void *buffer, size_t *lenp, loff_t *ppos)
1628 {
1629 return -ENOSYS;
1630 }
1631
1632 int proc_dointvec(struct ctl_table *table, int write,
1633 void *buffer, size_t *lenp, loff_t *ppos)
1634 {
1635 return -ENOSYS;
1636 }
1637
1638 int proc_douintvec(struct ctl_table *table, int write,
1639 void *buffer, size_t *lenp, loff_t *ppos)
1640 {
1641 return -ENOSYS;
1642 }
1643
1644 int proc_dointvec_minmax(struct ctl_table *table, int write,
1645 void *buffer, size_t *lenp, loff_t *ppos)
1646 {
1647 return -ENOSYS;
1648 }
1649
1650 int proc_douintvec_minmax(struct ctl_table *table, int write,
1651 void *buffer, size_t *lenp, loff_t *ppos)
1652 {
1653 return -ENOSYS;
1654 }
1655
1656 int proc_dou8vec_minmax(struct ctl_table *table, int write,
1657 void *buffer, size_t *lenp, loff_t *ppos)
1658 {
1659 return -ENOSYS;
1660 }
1661
1662 int proc_dointvec_jiffies(struct ctl_table *table, int write,
1663 void *buffer, size_t *lenp, loff_t *ppos)
1664 {
1665 return -ENOSYS;
1666 }
1667
1668 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1669 void *buffer, size_t *lenp, loff_t *ppos)
1670 {
1671 return -ENOSYS;
1672 }
1673
1674 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
1675 void *buffer, size_t *lenp, loff_t *ppos)
1676 {
1677 return -ENOSYS;
1678 }
1679
1680 int proc_doulongvec_minmax(struct ctl_table *table, int write,
1681 void *buffer, size_t *lenp, loff_t *ppos)
1682 {
1683 return -ENOSYS;
1684 }
1685
1686 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1687 void *buffer, size_t *lenp, loff_t *ppos)
1688 {
1689 return -ENOSYS;
1690 }
1691
1692 int proc_do_large_bitmap(struct ctl_table *table, int write,
1693 void *buffer, size_t *lenp, loff_t *ppos)
1694 {
1695 return -ENOSYS;
1696 }
1697
1698 #endif /* CONFIG_PROC_SYSCTL */
1699
1700 #if defined(CONFIG_SYSCTL)
1701 int proc_do_static_key(struct ctl_table *table, int write,
1702 void *buffer, size_t *lenp, loff_t *ppos)
1703 {
1704 struct static_key *key = (struct static_key *)table->data;
1705 static DEFINE_MUTEX(static_key_mutex);
1706 int val, ret;
1707 struct ctl_table tmp = {
1708 .data = &val,
1709 .maxlen = sizeof(val),
1710 .mode = table->mode,
1711 .extra1 = SYSCTL_ZERO,
1712 .extra2 = SYSCTL_ONE,
1713 };
1714
1715 if (write && !capable(CAP_SYS_ADMIN))
1716 return -EPERM;
1717
1718 mutex_lock(&static_key_mutex);
1719 val = static_key_enabled(key);
1720 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
1721 if (write && !ret) {
1722 if (val)
1723 static_key_enable(key);
1724 else
1725 static_key_disable(key);
1726 }
1727 mutex_unlock(&static_key_mutex);
1728 return ret;
1729 }
1730
1731 static struct ctl_table kern_table[] = {
1732 {
1733 .procname = "sched_child_runs_first",
1734 .data = &sysctl_sched_child_runs_first,
1735 .maxlen = sizeof(unsigned int),
1736 .mode = 0644,
1737 .proc_handler = proc_dointvec,
1738 },
1739 #ifdef CONFIG_SCHEDSTATS
1740 {
1741 .procname = "sched_schedstats",
1742 .data = NULL,
1743 .maxlen = sizeof(unsigned int),
1744 .mode = 0644,
1745 .proc_handler = sysctl_schedstats,
1746 .extra1 = SYSCTL_ZERO,
1747 .extra2 = SYSCTL_ONE,
1748 },
1749 #endif /* CONFIG_SCHEDSTATS */
1750 #ifdef CONFIG_NUMA_BALANCING
1751 {
1752 .procname = "numa_balancing",
1753 .data = NULL, /* filled in by handler */
1754 .maxlen = sizeof(unsigned int),
1755 .mode = 0644,
1756 .proc_handler = sysctl_numa_balancing,
1757 .extra1 = SYSCTL_ZERO,
1758 .extra2 = SYSCTL_ONE,
1759 },
1760 #endif /* CONFIG_NUMA_BALANCING */
1761 {
1762 .procname = "sched_rt_period_us",
1763 .data = &sysctl_sched_rt_period,
1764 .maxlen = sizeof(unsigned int),
1765 .mode = 0644,
1766 .proc_handler = sched_rt_handler,
1767 },
1768 {
1769 .procname = "sched_rt_runtime_us",
1770 .data = &sysctl_sched_rt_runtime,
1771 .maxlen = sizeof(int),
1772 .mode = 0644,
1773 .proc_handler = sched_rt_handler,
1774 },
1775 {
1776 .procname = "sched_deadline_period_max_us",
1777 .data = &sysctl_sched_dl_period_max,
1778 .maxlen = sizeof(unsigned int),
1779 .mode = 0644,
1780 .proc_handler = proc_dointvec,
1781 },
1782 {
1783 .procname = "sched_deadline_period_min_us",
1784 .data = &sysctl_sched_dl_period_min,
1785 .maxlen = sizeof(unsigned int),
1786 .mode = 0644,
1787 .proc_handler = proc_dointvec,
1788 },
1789 {
1790 .procname = "sched_rr_timeslice_ms",
1791 .data = &sysctl_sched_rr_timeslice,
1792 .maxlen = sizeof(int),
1793 .mode = 0644,
1794 .proc_handler = sched_rr_handler,
1795 },
1796 #ifdef CONFIG_UCLAMP_TASK
1797 {
1798 .procname = "sched_util_clamp_min",
1799 .data = &sysctl_sched_uclamp_util_min,
1800 .maxlen = sizeof(unsigned int),
1801 .mode = 0644,
1802 .proc_handler = sysctl_sched_uclamp_handler,
1803 },
1804 {
1805 .procname = "sched_util_clamp_max",
1806 .data = &sysctl_sched_uclamp_util_max,
1807 .maxlen = sizeof(unsigned int),
1808 .mode = 0644,
1809 .proc_handler = sysctl_sched_uclamp_handler,
1810 },
1811 {
1812 .procname = "sched_util_clamp_min_rt_default",
1813 .data = &sysctl_sched_uclamp_util_min_rt_default,
1814 .maxlen = sizeof(unsigned int),
1815 .mode = 0644,
1816 .proc_handler = sysctl_sched_uclamp_handler,
1817 },
1818 #endif
1819 #ifdef CONFIG_SCHED_AUTOGROUP
1820 {
1821 .procname = "sched_autogroup_enabled",
1822 .data = &sysctl_sched_autogroup_enabled,
1823 .maxlen = sizeof(unsigned int),
1824 .mode = 0644,
1825 .proc_handler = proc_dointvec_minmax,
1826 .extra1 = SYSCTL_ZERO,
1827 .extra2 = SYSCTL_ONE,
1828 },
1829 #endif
1830 #ifdef CONFIG_CFS_BANDWIDTH
1831 {
1832 .procname = "sched_cfs_bandwidth_slice_us",
1833 .data = &sysctl_sched_cfs_bandwidth_slice,
1834 .maxlen = sizeof(unsigned int),
1835 .mode = 0644,
1836 .proc_handler = proc_dointvec_minmax,
1837 .extra1 = SYSCTL_ONE,
1838 },
1839 #endif
1840 #if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_CPU_FREQ_GOV_SCHEDUTIL)
1841 {
1842 .procname = "sched_energy_aware",
1843 .data = &sysctl_sched_energy_aware,
1844 .maxlen = sizeof(unsigned int),
1845 .mode = 0644,
1846 .proc_handler = sched_energy_aware_handler,
1847 .extra1 = SYSCTL_ZERO,
1848 .extra2 = SYSCTL_ONE,
1849 },
1850 #endif
1851 #ifdef CONFIG_PROVE_LOCKING
1852 {
1853 .procname = "prove_locking",
1854 .data = &prove_locking,
1855 .maxlen = sizeof(int),
1856 .mode = 0644,
1857 .proc_handler = proc_dointvec,
1858 },
1859 #endif
1860 #ifdef CONFIG_LOCK_STAT
1861 {
1862 .procname = "lock_stat",
1863 .data = &lock_stat,
1864 .maxlen = sizeof(int),
1865 .mode = 0644,
1866 .proc_handler = proc_dointvec,
1867 },
1868 #endif
1869 {
1870 .procname = "panic",
1871 .data = &panic_timeout,
1872 .maxlen = sizeof(int),
1873 .mode = 0644,
1874 .proc_handler = proc_dointvec,
1875 },
1876 #ifdef CONFIG_COREDUMP
1877 {
1878 .procname = "core_uses_pid",
1879 .data = &core_uses_pid,
1880 .maxlen = sizeof(int),
1881 .mode = 0644,
1882 .proc_handler = proc_dointvec,
1883 },
1884 {
1885 .procname = "core_pattern",
1886 .data = core_pattern,
1887 .maxlen = CORENAME_MAX_SIZE,
1888 .mode = 0644,
1889 .proc_handler = proc_dostring_coredump,
1890 },
1891 {
1892 .procname = "core_pipe_limit",
1893 .data = &core_pipe_limit,
1894 .maxlen = sizeof(unsigned int),
1895 .mode = 0644,
1896 .proc_handler = proc_dointvec,
1897 },
1898 #endif
1899 #ifdef CONFIG_PROC_SYSCTL
1900 {
1901 .procname = "tainted",
1902 .maxlen = sizeof(long),
1903 .mode = 0644,
1904 .proc_handler = proc_taint,
1905 },
1906 {
1907 .procname = "sysctl_writes_strict",
1908 .data = &sysctl_writes_strict,
1909 .maxlen = sizeof(int),
1910 .mode = 0644,
1911 .proc_handler = proc_dointvec_minmax,
1912 .extra1 = &neg_one,
1913 .extra2 = SYSCTL_ONE,
1914 },
1915 #endif
1916 #ifdef CONFIG_LATENCYTOP
1917 {
1918 .procname = "latencytop",
1919 .data = &latencytop_enabled,
1920 .maxlen = sizeof(int),
1921 .mode = 0644,
1922 .proc_handler = sysctl_latencytop,
1923 },
1924 #endif
1925 #ifdef CONFIG_BLK_DEV_INITRD
1926 {
1927 .procname = "real-root-dev",
1928 .data = &real_root_dev,
1929 .maxlen = sizeof(int),
1930 .mode = 0644,
1931 .proc_handler = proc_dointvec,
1932 },
1933 #endif
1934 {
1935 .procname = "print-fatal-signals",
1936 .data = &print_fatal_signals,
1937 .maxlen = sizeof(int),
1938 .mode = 0644,
1939 .proc_handler = proc_dointvec,
1940 },
1941 #ifdef CONFIG_SPARC
1942 {
1943 .procname = "reboot-cmd",
1944 .data = reboot_command,
1945 .maxlen = 256,
1946 .mode = 0644,
1947 .proc_handler = proc_dostring,
1948 },
1949 {
1950 .procname = "stop-a",
1951 .data = &stop_a_enabled,
1952 .maxlen = sizeof (int),
1953 .mode = 0644,
1954 .proc_handler = proc_dointvec,
1955 },
1956 {
1957 .procname = "scons-poweroff",
1958 .data = &scons_pwroff,
1959 .maxlen = sizeof (int),
1960 .mode = 0644,
1961 .proc_handler = proc_dointvec,
1962 },
1963 #endif
1964 #ifdef CONFIG_SPARC64
1965 {
1966 .procname = "tsb-ratio",
1967 .data = &sysctl_tsb_ratio,
1968 .maxlen = sizeof (int),
1969 .mode = 0644,
1970 .proc_handler = proc_dointvec,
1971 },
1972 #endif
1973 #ifdef CONFIG_PARISC
1974 {
1975 .procname = "soft-power",
1976 .data = &pwrsw_enabled,
1977 .maxlen = sizeof (int),
1978 .mode = 0644,
1979 .proc_handler = proc_dointvec,
1980 },
1981 #endif
1982 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
1983 {
1984 .procname = "unaligned-trap",
1985 .data = &unaligned_enabled,
1986 .maxlen = sizeof (int),
1987 .mode = 0644,
1988 .proc_handler = proc_dointvec,
1989 },
1990 #endif
1991 {
1992 .procname = "ctrl-alt-del",
1993 .data = &C_A_D,
1994 .maxlen = sizeof(int),
1995 .mode = 0644,
1996 .proc_handler = proc_dointvec,
1997 },
1998 #ifdef CONFIG_FUNCTION_TRACER
1999 {
2000 .procname = "ftrace_enabled",
2001 .data = &ftrace_enabled,
2002 .maxlen = sizeof(int),
2003 .mode = 0644,
2004 .proc_handler = ftrace_enable_sysctl,
2005 },
2006 #endif
2007 #ifdef CONFIG_STACK_TRACER
2008 {
2009 .procname = "stack_tracer_enabled",
2010 .data = &stack_tracer_enabled,
2011 .maxlen = sizeof(int),
2012 .mode = 0644,
2013 .proc_handler = stack_trace_sysctl,
2014 },
2015 #endif
2016 #ifdef CONFIG_TRACING
2017 {
2018 .procname = "ftrace_dump_on_oops",
2019 .data = &ftrace_dump_on_oops,
2020 .maxlen = sizeof(int),
2021 .mode = 0644,
2022 .proc_handler = proc_dointvec,
2023 },
2024 {
2025 .procname = "traceoff_on_warning",
2026 .data = &__disable_trace_on_warning,
2027 .maxlen = sizeof(__disable_trace_on_warning),
2028 .mode = 0644,
2029 .proc_handler = proc_dointvec,
2030 },
2031 {
2032 .procname = "tracepoint_printk",
2033 .data = &tracepoint_printk,
2034 .maxlen = sizeof(tracepoint_printk),
2035 .mode = 0644,
2036 .proc_handler = tracepoint_printk_sysctl,
2037 },
2038 #endif
2039 #ifdef CONFIG_KEXEC_CORE
2040 {
2041 .procname = "kexec_load_disabled",
2042 .data = &kexec_load_disabled,
2043 .maxlen = sizeof(int),
2044 .mode = 0644,
2045 /* only handle a transition from default "0" to "1" */
2046 .proc_handler = proc_dointvec_minmax,
2047 .extra1 = SYSCTL_ONE,
2048 .extra2 = SYSCTL_ONE,
2049 },
2050 #endif
2051 #ifdef CONFIG_MODULES
2052 {
2053 .procname = "modprobe",
2054 .data = &modprobe_path,
2055 .maxlen = KMOD_PATH_LEN,
2056 .mode = 0644,
2057 .proc_handler = proc_dostring,
2058 },
2059 {
2060 .procname = "modules_disabled",
2061 .data = &modules_disabled,
2062 .maxlen = sizeof(int),
2063 .mode = 0644,
2064 /* only handle a transition from default "0" to "1" */
2065 .proc_handler = proc_dointvec_minmax,
2066 .extra1 = SYSCTL_ONE,
2067 .extra2 = SYSCTL_ONE,
2068 },
2069 #endif
2070 #ifdef CONFIG_UEVENT_HELPER
2071 {
2072 .procname = "hotplug",
2073 .data = &uevent_helper,
2074 .maxlen = UEVENT_HELPER_PATH_LEN,
2075 .mode = 0644,
2076 .proc_handler = proc_dostring,
2077 },
2078 #endif
2079 #ifdef CONFIG_CHR_DEV_SG
2080 {
2081 .procname = "sg-big-buff",
2082 .data = &sg_big_buff,
2083 .maxlen = sizeof (int),
2084 .mode = 0444,
2085 .proc_handler = proc_dointvec,
2086 },
2087 #endif
2088 #ifdef CONFIG_BSD_PROCESS_ACCT
2089 {
2090 .procname = "acct",
2091 .data = &acct_parm,
2092 .maxlen = 3*sizeof(int),
2093 .mode = 0644,
2094 .proc_handler = proc_dointvec,
2095 },
2096 #endif
2097 #ifdef CONFIG_MAGIC_SYSRQ
2098 {
2099 .procname = "sysrq",
2100 .data = NULL,
2101 .maxlen = sizeof (int),
2102 .mode = 0644,
2103 .proc_handler = sysrq_sysctl_handler,
2104 },
2105 #endif
2106 #ifdef CONFIG_PROC_SYSCTL
2107 {
2108 .procname = "cad_pid",
2109 .data = NULL,
2110 .maxlen = sizeof (int),
2111 .mode = 0600,
2112 .proc_handler = proc_do_cad_pid,
2113 },
2114 #endif
2115 {
2116 .procname = "threads-max",
2117 .data = NULL,
2118 .maxlen = sizeof(int),
2119 .mode = 0644,
2120 .proc_handler = sysctl_max_threads,
2121 },
2122 {
2123 .procname = "random",
2124 .mode = 0555,
2125 .child = random_table,
2126 },
2127 {
2128 .procname = "usermodehelper",
2129 .mode = 0555,
2130 .child = usermodehelper_table,
2131 },
2132 #ifdef CONFIG_FW_LOADER_USER_HELPER
2133 {
2134 .procname = "firmware_config",
2135 .mode = 0555,
2136 .child = firmware_config_table,
2137 },
2138 #endif
2139 {
2140 .procname = "overflowuid",
2141 .data = &overflowuid,
2142 .maxlen = sizeof(int),
2143 .mode = 0644,
2144 .proc_handler = proc_dointvec_minmax,
2145 .extra1 = &minolduid,
2146 .extra2 = &maxolduid,
2147 },
2148 {
2149 .procname = "overflowgid",
2150 .data = &overflowgid,
2151 .maxlen = sizeof(int),
2152 .mode = 0644,
2153 .proc_handler = proc_dointvec_minmax,
2154 .extra1 = &minolduid,
2155 .extra2 = &maxolduid,
2156 },
2157 #ifdef CONFIG_S390
2158 {
2159 .procname = "userprocess_debug",
2160 .data = &show_unhandled_signals,
2161 .maxlen = sizeof(int),
2162 .mode = 0644,
2163 .proc_handler = proc_dointvec,
2164 },
2165 #endif
2166 #ifdef CONFIG_SMP
2167 {
2168 .procname = "oops_all_cpu_backtrace",
2169 .data = &sysctl_oops_all_cpu_backtrace,
2170 .maxlen = sizeof(int),
2171 .mode = 0644,
2172 .proc_handler = proc_dointvec_minmax,
2173 .extra1 = SYSCTL_ZERO,
2174 .extra2 = SYSCTL_ONE,
2175 },
2176 #endif /* CONFIG_SMP */
2177 {
2178 .procname = "pid_max",
2179 .data = &pid_max,
2180 .maxlen = sizeof (int),
2181 .mode = 0644,
2182 .proc_handler = proc_dointvec_minmax,
2183 .extra1 = &pid_max_min,
2184 .extra2 = &pid_max_max,
2185 },
2186 {
2187 .procname = "panic_on_oops",
2188 .data = &panic_on_oops,
2189 .maxlen = sizeof(int),
2190 .mode = 0644,
2191 .proc_handler = proc_dointvec,
2192 },
2193 {
2194 .procname = "panic_print",
2195 .data = &panic_print,
2196 .maxlen = sizeof(unsigned long),
2197 .mode = 0644,
2198 .proc_handler = proc_doulongvec_minmax,
2199 },
2200 #if defined CONFIG_PRINTK
2201 {
2202 .procname = "printk",
2203 .data = &console_loglevel,
2204 .maxlen = 4*sizeof(int),
2205 .mode = 0644,
2206 .proc_handler = proc_dointvec,
2207 },
2208 {
2209 .procname = "printk_ratelimit",
2210 .data = &printk_ratelimit_state.interval,
2211 .maxlen = sizeof(int),
2212 .mode = 0644,
2213 .proc_handler = proc_dointvec_jiffies,
2214 },
2215 {
2216 .procname = "printk_ratelimit_burst",
2217 .data = &printk_ratelimit_state.burst,
2218 .maxlen = sizeof(int),
2219 .mode = 0644,
2220 .proc_handler = proc_dointvec,
2221 },
2222 {
2223 .procname = "printk_delay",
2224 .data = &printk_delay_msec,
2225 .maxlen = sizeof(int),
2226 .mode = 0644,
2227 .proc_handler = proc_dointvec_minmax,
2228 .extra1 = SYSCTL_ZERO,
2229 .extra2 = &ten_thousand,
2230 },
2231 {
2232 .procname = "printk_devkmsg",
2233 .data = devkmsg_log_str,
2234 .maxlen = DEVKMSG_STR_MAX_SIZE,
2235 .mode = 0644,
2236 .proc_handler = devkmsg_sysctl_set_loglvl,
2237 },
2238 {
2239 .procname = "dmesg_restrict",
2240 .data = &dmesg_restrict,
2241 .maxlen = sizeof(int),
2242 .mode = 0644,
2243 .proc_handler = proc_dointvec_minmax_sysadmin,
2244 .extra1 = SYSCTL_ZERO,
2245 .extra2 = SYSCTL_ONE,
2246 },
2247 {
2248 .procname = "kptr_restrict",
2249 .data = &kptr_restrict,
2250 .maxlen = sizeof(int),
2251 .mode = 0644,
2252 .proc_handler = proc_dointvec_minmax_sysadmin,
2253 .extra1 = SYSCTL_ZERO,
2254 .extra2 = &two,
2255 },
2256 #endif
2257 {
2258 .procname = "ngroups_max",
2259 .data = &ngroups_max,
2260 .maxlen = sizeof (int),
2261 .mode = 0444,
2262 .proc_handler = proc_dointvec,
2263 },
2264 {
2265 .procname = "cap_last_cap",
2266 .data = (void *)&cap_last_cap,
2267 .maxlen = sizeof(int),
2268 .mode = 0444,
2269 .proc_handler = proc_dointvec,
2270 },
2271 #if defined(CONFIG_LOCKUP_DETECTOR)
2272 {
2273 .procname = "watchdog",
2274 .data = &watchdog_user_enabled,
2275 .maxlen = sizeof(int),
2276 .mode = 0644,
2277 .proc_handler = proc_watchdog,
2278 .extra1 = SYSCTL_ZERO,
2279 .extra2 = SYSCTL_ONE,
2280 },
2281 {
2282 .procname = "watchdog_thresh",
2283 .data = &watchdog_thresh,
2284 .maxlen = sizeof(int),
2285 .mode = 0644,
2286 .proc_handler = proc_watchdog_thresh,
2287 .extra1 = SYSCTL_ZERO,
2288 .extra2 = &sixty,
2289 },
2290 {
2291 .procname = "nmi_watchdog",
2292 .data = &nmi_watchdog_user_enabled,
2293 .maxlen = sizeof(int),
2294 .mode = NMI_WATCHDOG_SYSCTL_PERM,
2295 .proc_handler = proc_nmi_watchdog,
2296 .extra1 = SYSCTL_ZERO,
2297 .extra2 = SYSCTL_ONE,
2298 },
2299 {
2300 .procname = "watchdog_cpumask",
2301 .data = &watchdog_cpumask_bits,
2302 .maxlen = NR_CPUS,
2303 .mode = 0644,
2304 .proc_handler = proc_watchdog_cpumask,
2305 },
2306 #ifdef CONFIG_SOFTLOCKUP_DETECTOR
2307 {
2308 .procname = "soft_watchdog",
2309 .data = &soft_watchdog_user_enabled,
2310 .maxlen = sizeof(int),
2311 .mode = 0644,
2312 .proc_handler = proc_soft_watchdog,
2313 .extra1 = SYSCTL_ZERO,
2314 .extra2 = SYSCTL_ONE,
2315 },
2316 {
2317 .procname = "softlockup_panic",
2318 .data = &softlockup_panic,
2319 .maxlen = sizeof(int),
2320 .mode = 0644,
2321 .proc_handler = proc_dointvec_minmax,
2322 .extra1 = SYSCTL_ZERO,
2323 .extra2 = SYSCTL_ONE,
2324 },
2325 #ifdef CONFIG_SMP
2326 {
2327 .procname = "softlockup_all_cpu_backtrace",
2328 .data = &sysctl_softlockup_all_cpu_backtrace,
2329 .maxlen = sizeof(int),
2330 .mode = 0644,
2331 .proc_handler = proc_dointvec_minmax,
2332 .extra1 = SYSCTL_ZERO,
2333 .extra2 = SYSCTL_ONE,
2334 },
2335 #endif /* CONFIG_SMP */
2336 #endif
2337 #ifdef CONFIG_HARDLOCKUP_DETECTOR
2338 {
2339 .procname = "hardlockup_panic",
2340 .data = &hardlockup_panic,
2341 .maxlen = sizeof(int),
2342 .mode = 0644,
2343 .proc_handler = proc_dointvec_minmax,
2344 .extra1 = SYSCTL_ZERO,
2345 .extra2 = SYSCTL_ONE,
2346 },
2347 #ifdef CONFIG_SMP
2348 {
2349 .procname = "hardlockup_all_cpu_backtrace",
2350 .data = &sysctl_hardlockup_all_cpu_backtrace,
2351 .maxlen = sizeof(int),
2352 .mode = 0644,
2353 .proc_handler = proc_dointvec_minmax,
2354 .extra1 = SYSCTL_ZERO,
2355 .extra2 = SYSCTL_ONE,
2356 },
2357 #endif /* CONFIG_SMP */
2358 #endif
2359 #endif
2360
2361 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
2362 {
2363 .procname = "unknown_nmi_panic",
2364 .data = &unknown_nmi_panic,
2365 .maxlen = sizeof (int),
2366 .mode = 0644,
2367 .proc_handler = proc_dointvec,
2368 },
2369 #endif
2370
2371 #if (defined(CONFIG_X86_32) || defined(CONFIG_PARISC)) && \
2372 defined(CONFIG_DEBUG_STACKOVERFLOW)
2373 {
2374 .procname = "panic_on_stackoverflow",
2375 .data = &sysctl_panic_on_stackoverflow,
2376 .maxlen = sizeof(int),
2377 .mode = 0644,
2378 .proc_handler = proc_dointvec,
2379 },
2380 #endif
2381 #if defined(CONFIG_X86)
2382 {
2383 .procname = "panic_on_unrecovered_nmi",
2384 .data = &panic_on_unrecovered_nmi,
2385 .maxlen = sizeof(int),
2386 .mode = 0644,
2387 .proc_handler = proc_dointvec,
2388 },
2389 {
2390 .procname = "panic_on_io_nmi",
2391 .data = &panic_on_io_nmi,
2392 .maxlen = sizeof(int),
2393 .mode = 0644,
2394 .proc_handler = proc_dointvec,
2395 },
2396 {
2397 .procname = "bootloader_type",
2398 .data = &bootloader_type,
2399 .maxlen = sizeof (int),
2400 .mode = 0444,
2401 .proc_handler = proc_dointvec,
2402 },
2403 {
2404 .procname = "bootloader_version",
2405 .data = &bootloader_version,
2406 .maxlen = sizeof (int),
2407 .mode = 0444,
2408 .proc_handler = proc_dointvec,
2409 },
2410 {
2411 .procname = "io_delay_type",
2412 .data = &io_delay_type,
2413 .maxlen = sizeof(int),
2414 .mode = 0644,
2415 .proc_handler = proc_dointvec,
2416 },
2417 #endif
2418 #if defined(CONFIG_MMU)
2419 {
2420 .procname = "randomize_va_space",
2421 .data = &randomize_va_space,
2422 .maxlen = sizeof(int),
2423 .mode = 0644,
2424 .proc_handler = proc_dointvec,
2425 },
2426 #endif
2427 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
2428 {
2429 .procname = "spin_retry",
2430 .data = &spin_retry,
2431 .maxlen = sizeof (int),
2432 .mode = 0644,
2433 .proc_handler = proc_dointvec,
2434 },
2435 #endif
2436 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
2437 {
2438 .procname = "acpi_video_flags",
2439 .data = &acpi_realmode_flags,
2440 .maxlen = sizeof (unsigned long),
2441 .mode = 0644,
2442 .proc_handler = proc_doulongvec_minmax,
2443 },
2444 #endif
2445 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
2446 {
2447 .procname = "ignore-unaligned-usertrap",
2448 .data = &no_unaligned_warning,
2449 .maxlen = sizeof (int),
2450 .mode = 0644,
2451 .proc_handler = proc_dointvec,
2452 },
2453 #endif
2454 #ifdef CONFIG_IA64
2455 {
2456 .procname = "unaligned-dump-stack",
2457 .data = &unaligned_dump_stack,
2458 .maxlen = sizeof (int),
2459 .mode = 0644,
2460 .proc_handler = proc_dointvec,
2461 },
2462 #endif
2463 #ifdef CONFIG_DETECT_HUNG_TASK
2464 #ifdef CONFIG_SMP
2465 {
2466 .procname = "hung_task_all_cpu_backtrace",
2467 .data = &sysctl_hung_task_all_cpu_backtrace,
2468 .maxlen = sizeof(int),
2469 .mode = 0644,
2470 .proc_handler = proc_dointvec_minmax,
2471 .extra1 = SYSCTL_ZERO,
2472 .extra2 = SYSCTL_ONE,
2473 },
2474 #endif /* CONFIG_SMP */
2475 {
2476 .procname = "hung_task_panic",
2477 .data = &sysctl_hung_task_panic,
2478 .maxlen = sizeof(int),
2479 .mode = 0644,
2480 .proc_handler = proc_dointvec_minmax,
2481 .extra1 = SYSCTL_ZERO,
2482 .extra2 = SYSCTL_ONE,
2483 },
2484 {
2485 .procname = "hung_task_check_count",
2486 .data = &sysctl_hung_task_check_count,
2487 .maxlen = sizeof(int),
2488 .mode = 0644,
2489 .proc_handler = proc_dointvec_minmax,
2490 .extra1 = SYSCTL_ZERO,
2491 },
2492 {
2493 .procname = "hung_task_timeout_secs",
2494 .data = &sysctl_hung_task_timeout_secs,
2495 .maxlen = sizeof(unsigned long),
2496 .mode = 0644,
2497 .proc_handler = proc_dohung_task_timeout_secs,
2498 .extra2 = &hung_task_timeout_max,
2499 },
2500 {
2501 .procname = "hung_task_check_interval_secs",
2502 .data = &sysctl_hung_task_check_interval_secs,
2503 .maxlen = sizeof(unsigned long),
2504 .mode = 0644,
2505 .proc_handler = proc_dohung_task_timeout_secs,
2506 .extra2 = &hung_task_timeout_max,
2507 },
2508 {
2509 .procname = "hung_task_warnings",
2510 .data = &sysctl_hung_task_warnings,
2511 .maxlen = sizeof(int),
2512 .mode = 0644,
2513 .proc_handler = proc_dointvec_minmax,
2514 .extra1 = &neg_one,
2515 },
2516 #endif
2517 #ifdef CONFIG_RT_MUTEXES
2518 {
2519 .procname = "max_lock_depth",
2520 .data = &max_lock_depth,
2521 .maxlen = sizeof(int),
2522 .mode = 0644,
2523 .proc_handler = proc_dointvec,
2524 },
2525 #endif
2526 {
2527 .procname = "poweroff_cmd",
2528 .data = &poweroff_cmd,
2529 .maxlen = POWEROFF_CMD_PATH_LEN,
2530 .mode = 0644,
2531 .proc_handler = proc_dostring,
2532 },
2533 #ifdef CONFIG_KEYS
2534 {
2535 .procname = "keys",
2536 .mode = 0555,
2537 .child = key_sysctls,
2538 },
2539 #endif
2540 #ifdef CONFIG_PERF_EVENTS
2541 /*
2542 * User-space scripts rely on the existence of this file
2543 * as a feature check for perf_events being enabled.
2544 *
2545 * So it's an ABI, do not remove!
2546 */
2547 {
2548 .procname = "perf_event_paranoid",
2549 .data = &sysctl_perf_event_paranoid,
2550 .maxlen = sizeof(sysctl_perf_event_paranoid),
2551 .mode = 0644,
2552 .proc_handler = proc_dointvec,
2553 },
2554 {
2555 .procname = "perf_event_mlock_kb",
2556 .data = &sysctl_perf_event_mlock,
2557 .maxlen = sizeof(sysctl_perf_event_mlock),
2558 .mode = 0644,
2559 .proc_handler = proc_dointvec,
2560 },
2561 {
2562 .procname = "perf_event_max_sample_rate",
2563 .data = &sysctl_perf_event_sample_rate,
2564 .maxlen = sizeof(sysctl_perf_event_sample_rate),
2565 .mode = 0644,
2566 .proc_handler = perf_proc_update_handler,
2567 .extra1 = SYSCTL_ONE,
2568 },
2569 {
2570 .procname = "perf_cpu_time_max_percent",
2571 .data = &sysctl_perf_cpu_time_max_percent,
2572 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
2573 .mode = 0644,
2574 .proc_handler = perf_cpu_time_max_percent_handler,
2575 .extra1 = SYSCTL_ZERO,
2576 .extra2 = &one_hundred,
2577 },
2578 {
2579 .procname = "perf_event_max_stack",
2580 .data = &sysctl_perf_event_max_stack,
2581 .maxlen = sizeof(sysctl_perf_event_max_stack),
2582 .mode = 0644,
2583 .proc_handler = perf_event_max_stack_handler,
2584 .extra1 = SYSCTL_ZERO,
2585 .extra2 = &six_hundred_forty_kb,
2586 },
2587 {
2588 .procname = "perf_event_max_contexts_per_stack",
2589 .data = &sysctl_perf_event_max_contexts_per_stack,
2590 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
2591 .mode = 0644,
2592 .proc_handler = perf_event_max_stack_handler,
2593 .extra1 = SYSCTL_ZERO,
2594 .extra2 = &one_thousand,
2595 },
2596 #endif
2597 {
2598 .procname = "panic_on_warn",
2599 .data = &panic_on_warn,
2600 .maxlen = sizeof(int),
2601 .mode = 0644,
2602 .proc_handler = proc_dointvec_minmax,
2603 .extra1 = SYSCTL_ZERO,
2604 .extra2 = SYSCTL_ONE,
2605 },
2606 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
2607 {
2608 .procname = "timer_migration",
2609 .data = &sysctl_timer_migration,
2610 .maxlen = sizeof(unsigned int),
2611 .mode = 0644,
2612 .proc_handler = timer_migration_handler,
2613 .extra1 = SYSCTL_ZERO,
2614 .extra2 = SYSCTL_ONE,
2615 },
2616 #endif
2617 #ifdef CONFIG_BPF_SYSCALL
2618 {
2619 .procname = "unprivileged_bpf_disabled",
2620 .data = &sysctl_unprivileged_bpf_disabled,
2621 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
2622 .mode = 0644,
2623 .proc_handler = bpf_unpriv_handler,
2624 .extra1 = SYSCTL_ZERO,
2625 .extra2 = &two,
2626 },
2627 {
2628 .procname = "bpf_stats_enabled",
2629 .data = &bpf_stats_enabled_key.key,
2630 .maxlen = sizeof(bpf_stats_enabled_key),
2631 .mode = 0644,
2632 .proc_handler = bpf_stats_handler,
2633 },
2634 #endif
2635 #if defined(CONFIG_TREE_RCU)
2636 {
2637 .procname = "panic_on_rcu_stall",
2638 .data = &sysctl_panic_on_rcu_stall,
2639 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
2640 .mode = 0644,
2641 .proc_handler = proc_dointvec_minmax,
2642 .extra1 = SYSCTL_ZERO,
2643 .extra2 = SYSCTL_ONE,
2644 },
2645 #endif
2646 #if defined(CONFIG_TREE_RCU)
2647 {
2648 .procname = "max_rcu_stall_to_panic",
2649 .data = &sysctl_max_rcu_stall_to_panic,
2650 .maxlen = sizeof(sysctl_max_rcu_stall_to_panic),
2651 .mode = 0644,
2652 .proc_handler = proc_dointvec_minmax,
2653 .extra1 = SYSCTL_ONE,
2654 .extra2 = SYSCTL_INT_MAX,
2655 },
2656 #endif
2657 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
2658 {
2659 .procname = "stack_erasing",
2660 .data = NULL,
2661 .maxlen = sizeof(int),
2662 .mode = 0600,
2663 .proc_handler = stack_erasing_sysctl,
2664 .extra1 = SYSCTL_ZERO,
2665 .extra2 = SYSCTL_ONE,
2666 },
2667 #endif
2668 { }
2669 };
2670
2671 static struct ctl_table vm_table[] = {
2672 {
2673 .procname = "overcommit_memory",
2674 .data = &sysctl_overcommit_memory,
2675 .maxlen = sizeof(sysctl_overcommit_memory),
2676 .mode = 0644,
2677 .proc_handler = overcommit_policy_handler,
2678 .extra1 = SYSCTL_ZERO,
2679 .extra2 = &two,
2680 },
2681 {
2682 .procname = "panic_on_oom",
2683 .data = &sysctl_panic_on_oom,
2684 .maxlen = sizeof(sysctl_panic_on_oom),
2685 .mode = 0644,
2686 .proc_handler = proc_dointvec_minmax,
2687 .extra1 = SYSCTL_ZERO,
2688 .extra2 = &two,
2689 },
2690 {
2691 .procname = "oom_kill_allocating_task",
2692 .data = &sysctl_oom_kill_allocating_task,
2693 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
2694 .mode = 0644,
2695 .proc_handler = proc_dointvec,
2696 },
2697 {
2698 .procname = "oom_dump_tasks",
2699 .data = &sysctl_oom_dump_tasks,
2700 .maxlen = sizeof(sysctl_oom_dump_tasks),
2701 .mode = 0644,
2702 .proc_handler = proc_dointvec,
2703 },
2704 {
2705 .procname = "overcommit_ratio",
2706 .data = &sysctl_overcommit_ratio,
2707 .maxlen = sizeof(sysctl_overcommit_ratio),
2708 .mode = 0644,
2709 .proc_handler = overcommit_ratio_handler,
2710 },
2711 {
2712 .procname = "overcommit_kbytes",
2713 .data = &sysctl_overcommit_kbytes,
2714 .maxlen = sizeof(sysctl_overcommit_kbytes),
2715 .mode = 0644,
2716 .proc_handler = overcommit_kbytes_handler,
2717 },
2718 {
2719 .procname = "page-cluster",
2720 .data = &page_cluster,
2721 .maxlen = sizeof(int),
2722 .mode = 0644,
2723 .proc_handler = proc_dointvec_minmax,
2724 .extra1 = SYSCTL_ZERO,
2725 },
2726 {
2727 .procname = "dirty_background_ratio",
2728 .data = &dirty_background_ratio,
2729 .maxlen = sizeof(dirty_background_ratio),
2730 .mode = 0644,
2731 .proc_handler = dirty_background_ratio_handler,
2732 .extra1 = SYSCTL_ZERO,
2733 .extra2 = &one_hundred,
2734 },
2735 {
2736 .procname = "dirty_background_bytes",
2737 .data = &dirty_background_bytes,
2738 .maxlen = sizeof(dirty_background_bytes),
2739 .mode = 0644,
2740 .proc_handler = dirty_background_bytes_handler,
2741 .extra1 = &one_ul,
2742 },
2743 {
2744 .procname = "dirty_ratio",
2745 .data = &vm_dirty_ratio,
2746 .maxlen = sizeof(vm_dirty_ratio),
2747 .mode = 0644,
2748 .proc_handler = dirty_ratio_handler,
2749 .extra1 = SYSCTL_ZERO,
2750 .extra2 = &one_hundred,
2751 },
2752 {
2753 .procname = "dirty_bytes",
2754 .data = &vm_dirty_bytes,
2755 .maxlen = sizeof(vm_dirty_bytes),
2756 .mode = 0644,
2757 .proc_handler = dirty_bytes_handler,
2758 .extra1 = &dirty_bytes_min,
2759 },
2760 {
2761 .procname = "dirty_writeback_centisecs",
2762 .data = &dirty_writeback_interval,
2763 .maxlen = sizeof(dirty_writeback_interval),
2764 .mode = 0644,
2765 .proc_handler = dirty_writeback_centisecs_handler,
2766 },
2767 {
2768 .procname = "dirty_expire_centisecs",
2769 .data = &dirty_expire_interval,
2770 .maxlen = sizeof(dirty_expire_interval),
2771 .mode = 0644,
2772 .proc_handler = proc_dointvec_minmax,
2773 .extra1 = SYSCTL_ZERO,
2774 },
2775 {
2776 .procname = "dirtytime_expire_seconds",
2777 .data = &dirtytime_expire_interval,
2778 .maxlen = sizeof(dirtytime_expire_interval),
2779 .mode = 0644,
2780 .proc_handler = dirtytime_interval_handler,
2781 .extra1 = SYSCTL_ZERO,
2782 },
2783 {
2784 .procname = "swappiness",
2785 .data = &vm_swappiness,
2786 .maxlen = sizeof(vm_swappiness),
2787 .mode = 0644,
2788 .proc_handler = proc_dointvec_minmax,
2789 .extra1 = SYSCTL_ZERO,
2790 .extra2 = &two_hundred,
2791 },
2792 #ifdef CONFIG_HUGETLB_PAGE
2793 {
2794 .procname = "nr_hugepages",
2795 .data = NULL,
2796 .maxlen = sizeof(unsigned long),
2797 .mode = 0644,
2798 .proc_handler = hugetlb_sysctl_handler,
2799 },
2800 #ifdef CONFIG_NUMA
2801 {
2802 .procname = "nr_hugepages_mempolicy",
2803 .data = NULL,
2804 .maxlen = sizeof(unsigned long),
2805 .mode = 0644,
2806 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
2807 },
2808 {
2809 .procname = "numa_stat",
2810 .data = &sysctl_vm_numa_stat,
2811 .maxlen = sizeof(int),
2812 .mode = 0644,
2813 .proc_handler = sysctl_vm_numa_stat_handler,
2814 .extra1 = SYSCTL_ZERO,
2815 .extra2 = SYSCTL_ONE,
2816 },
2817 #endif
2818 {
2819 .procname = "hugetlb_shm_group",
2820 .data = &sysctl_hugetlb_shm_group,
2821 .maxlen = sizeof(gid_t),
2822 .mode = 0644,
2823 .proc_handler = proc_dointvec,
2824 },
2825 {
2826 .procname = "nr_overcommit_hugepages",
2827 .data = NULL,
2828 .maxlen = sizeof(unsigned long),
2829 .mode = 0644,
2830 .proc_handler = hugetlb_overcommit_handler,
2831 },
2832 #endif
2833 {
2834 .procname = "lowmem_reserve_ratio",
2835 .data = &sysctl_lowmem_reserve_ratio,
2836 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
2837 .mode = 0644,
2838 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
2839 },
2840 {
2841 .procname = "drop_caches",
2842 .data = &sysctl_drop_caches,
2843 .maxlen = sizeof(int),
2844 .mode = 0200,
2845 .proc_handler = drop_caches_sysctl_handler,
2846 .extra1 = SYSCTL_ONE,
2847 .extra2 = &four,
2848 },
2849 #ifdef CONFIG_COMPACTION
2850 {
2851 .procname = "compact_memory",
2852 .data = NULL,
2853 .maxlen = sizeof(int),
2854 .mode = 0200,
2855 .proc_handler = sysctl_compaction_handler,
2856 },
2857 {
2858 .procname = "compaction_proactiveness",
2859 .data = &sysctl_compaction_proactiveness,
2860 .maxlen = sizeof(sysctl_compaction_proactiveness),
2861 .mode = 0644,
2862 .proc_handler = proc_dointvec_minmax,
2863 .extra1 = SYSCTL_ZERO,
2864 .extra2 = &one_hundred,
2865 },
2866 {
2867 .procname = "extfrag_threshold",
2868 .data = &sysctl_extfrag_threshold,
2869 .maxlen = sizeof(int),
2870 .mode = 0644,
2871 .proc_handler = proc_dointvec_minmax,
2872 .extra1 = &min_extfrag_threshold,
2873 .extra2 = &max_extfrag_threshold,
2874 },
2875 {
2876 .procname = "compact_unevictable_allowed",
2877 .data = &sysctl_compact_unevictable_allowed,
2878 .maxlen = sizeof(int),
2879 .mode = 0644,
2880 .proc_handler = proc_dointvec_minmax_warn_RT_change,
2881 .extra1 = SYSCTL_ZERO,
2882 .extra2 = SYSCTL_ONE,
2883 },
2884
2885 #endif /* CONFIG_COMPACTION */
2886 {
2887 .procname = "min_free_kbytes",
2888 .data = &min_free_kbytes,
2889 .maxlen = sizeof(min_free_kbytes),
2890 .mode = 0644,
2891 .proc_handler = min_free_kbytes_sysctl_handler,
2892 .extra1 = SYSCTL_ZERO,
2893 },
2894 {
2895 .procname = "watermark_boost_factor",
2896 .data = &watermark_boost_factor,
2897 .maxlen = sizeof(watermark_boost_factor),
2898 .mode = 0644,
2899 .proc_handler = proc_dointvec_minmax,
2900 .extra1 = SYSCTL_ZERO,
2901 },
2902 {
2903 .procname = "watermark_scale_factor",
2904 .data = &watermark_scale_factor,
2905 .maxlen = sizeof(watermark_scale_factor),
2906 .mode = 0644,
2907 .proc_handler = watermark_scale_factor_sysctl_handler,
2908 .extra1 = SYSCTL_ONE,
2909 .extra2 = &one_thousand,
2910 },
2911 {
2912 .procname = "percpu_pagelist_fraction",
2913 .data = &percpu_pagelist_fraction,
2914 .maxlen = sizeof(percpu_pagelist_fraction),
2915 .mode = 0644,
2916 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
2917 .extra1 = SYSCTL_ZERO,
2918 },
2919 {
2920 .procname = "page_lock_unfairness",
2921 .data = &sysctl_page_lock_unfairness,
2922 .maxlen = sizeof(sysctl_page_lock_unfairness),
2923 .mode = 0644,
2924 .proc_handler = proc_dointvec_minmax,
2925 .extra1 = SYSCTL_ZERO,
2926 },
2927 #ifdef CONFIG_MMU
2928 {
2929 .procname = "max_map_count",
2930 .data = &sysctl_max_map_count,
2931 .maxlen = sizeof(sysctl_max_map_count),
2932 .mode = 0644,
2933 .proc_handler = proc_dointvec_minmax,
2934 .extra1 = SYSCTL_ZERO,
2935 },
2936 #else
2937 {
2938 .procname = "nr_trim_pages",
2939 .data = &sysctl_nr_trim_pages,
2940 .maxlen = sizeof(sysctl_nr_trim_pages),
2941 .mode = 0644,
2942 .proc_handler = proc_dointvec_minmax,
2943 .extra1 = SYSCTL_ZERO,
2944 },
2945 #endif
2946 {
2947 .procname = "laptop_mode",
2948 .data = &laptop_mode,
2949 .maxlen = sizeof(laptop_mode),
2950 .mode = 0644,
2951 .proc_handler = proc_dointvec_jiffies,
2952 },
2953 {
2954 .procname = "block_dump",
2955 .data = &block_dump,
2956 .maxlen = sizeof(block_dump),
2957 .mode = 0644,
2958 .proc_handler = proc_dointvec_minmax,
2959 .extra1 = SYSCTL_ZERO,
2960 },
2961 {
2962 .procname = "vfs_cache_pressure",
2963 .data = &sysctl_vfs_cache_pressure,
2964 .maxlen = sizeof(sysctl_vfs_cache_pressure),
2965 .mode = 0644,
2966 .proc_handler = proc_dointvec_minmax,
2967 .extra1 = SYSCTL_ZERO,
2968 },
2969 #if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
2970 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
2971 {
2972 .procname = "legacy_va_layout",
2973 .data = &sysctl_legacy_va_layout,
2974 .maxlen = sizeof(sysctl_legacy_va_layout),
2975 .mode = 0644,
2976 .proc_handler = proc_dointvec_minmax,
2977 .extra1 = SYSCTL_ZERO,
2978 },
2979 #endif
2980 #ifdef CONFIG_NUMA
2981 {
2982 .procname = "zone_reclaim_mode",
2983 .data = &node_reclaim_mode,
2984 .maxlen = sizeof(node_reclaim_mode),
2985 .mode = 0644,
2986 .proc_handler = proc_dointvec_minmax,
2987 .extra1 = SYSCTL_ZERO,
2988 },
2989 {
2990 .procname = "min_unmapped_ratio",
2991 .data = &sysctl_min_unmapped_ratio,
2992 .maxlen = sizeof(sysctl_min_unmapped_ratio),
2993 .mode = 0644,
2994 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
2995 .extra1 = SYSCTL_ZERO,
2996 .extra2 = &one_hundred,
2997 },
2998 {
2999 .procname = "min_slab_ratio",
3000 .data = &sysctl_min_slab_ratio,
3001 .maxlen = sizeof(sysctl_min_slab_ratio),
3002 .mode = 0644,
3003 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
3004 .extra1 = SYSCTL_ZERO,
3005 .extra2 = &one_hundred,
3006 },
3007 #endif
3008 #ifdef CONFIG_SMP
3009 {
3010 .procname = "stat_interval",
3011 .data = &sysctl_stat_interval,
3012 .maxlen = sizeof(sysctl_stat_interval),
3013 .mode = 0644,
3014 .proc_handler = proc_dointvec_jiffies,
3015 },
3016 {
3017 .procname = "stat_refresh",
3018 .data = NULL,
3019 .maxlen = 0,
3020 .mode = 0600,
3021 .proc_handler = vmstat_refresh,
3022 },
3023 #endif
3024 #ifdef CONFIG_MMU
3025 {
3026 .procname = "mmap_min_addr",
3027 .data = &dac_mmap_min_addr,
3028 .maxlen = sizeof(unsigned long),
3029 .mode = 0644,
3030 .proc_handler = mmap_min_addr_handler,
3031 },
3032 #endif
3033 #ifdef CONFIG_NUMA
3034 {
3035 .procname = "numa_zonelist_order",
3036 .data = &numa_zonelist_order,
3037 .maxlen = NUMA_ZONELIST_ORDER_LEN,
3038 .mode = 0644,
3039 .proc_handler = numa_zonelist_order_handler,
3040 },
3041 #endif
3042 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
3043 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
3044 {
3045 .procname = "vdso_enabled",
3046 #ifdef CONFIG_X86_32
3047 .data = &vdso32_enabled,
3048 .maxlen = sizeof(vdso32_enabled),
3049 #else
3050 .data = &vdso_enabled,
3051 .maxlen = sizeof(vdso_enabled),
3052 #endif
3053 .mode = 0644,
3054 .proc_handler = proc_dointvec,
3055 .extra1 = SYSCTL_ZERO,
3056 },
3057 #endif
3058 #ifdef CONFIG_HIGHMEM
3059 {
3060 .procname = "highmem_is_dirtyable",
3061 .data = &vm_highmem_is_dirtyable,
3062 .maxlen = sizeof(vm_highmem_is_dirtyable),
3063 .mode = 0644,
3064 .proc_handler = proc_dointvec_minmax,
3065 .extra1 = SYSCTL_ZERO,
3066 .extra2 = SYSCTL_ONE,
3067 },
3068 #endif
3069 #ifdef CONFIG_MEMORY_FAILURE
3070 {
3071 .procname = "memory_failure_early_kill",
3072 .data = &sysctl_memory_failure_early_kill,
3073 .maxlen = sizeof(sysctl_memory_failure_early_kill),
3074 .mode = 0644,
3075 .proc_handler = proc_dointvec_minmax,
3076 .extra1 = SYSCTL_ZERO,
3077 .extra2 = SYSCTL_ONE,
3078 },
3079 {
3080 .procname = "memory_failure_recovery",
3081 .data = &sysctl_memory_failure_recovery,
3082 .maxlen = sizeof(sysctl_memory_failure_recovery),
3083 .mode = 0644,
3084 .proc_handler = proc_dointvec_minmax,
3085 .extra1 = SYSCTL_ZERO,
3086 .extra2 = SYSCTL_ONE,
3087 },
3088 #endif
3089 {
3090 .procname = "user_reserve_kbytes",
3091 .data = &sysctl_user_reserve_kbytes,
3092 .maxlen = sizeof(sysctl_user_reserve_kbytes),
3093 .mode = 0644,
3094 .proc_handler = proc_doulongvec_minmax,
3095 },
3096 {
3097 .procname = "admin_reserve_kbytes",
3098 .data = &sysctl_admin_reserve_kbytes,
3099 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
3100 .mode = 0644,
3101 .proc_handler = proc_doulongvec_minmax,
3102 },
3103 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
3104 {
3105 .procname = "mmap_rnd_bits",
3106 .data = &mmap_rnd_bits,
3107 .maxlen = sizeof(mmap_rnd_bits),
3108 .mode = 0600,
3109 .proc_handler = proc_dointvec_minmax,
3110 .extra1 = (void *)&mmap_rnd_bits_min,
3111 .extra2 = (void *)&mmap_rnd_bits_max,
3112 },
3113 #endif
3114 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
3115 {
3116 .procname = "mmap_rnd_compat_bits",
3117 .data = &mmap_rnd_compat_bits,
3118 .maxlen = sizeof(mmap_rnd_compat_bits),
3119 .mode = 0600,
3120 .proc_handler = proc_dointvec_minmax,
3121 .extra1 = (void *)&mmap_rnd_compat_bits_min,
3122 .extra2 = (void *)&mmap_rnd_compat_bits_max,
3123 },
3124 #endif
3125 #ifdef CONFIG_USERFAULTFD
3126 {
3127 .procname = "unprivileged_userfaultfd",
3128 .data = &sysctl_unprivileged_userfaultfd,
3129 .maxlen = sizeof(sysctl_unprivileged_userfaultfd),
3130 .mode = 0644,
3131 .proc_handler = proc_dointvec_minmax,
3132 .extra1 = SYSCTL_ZERO,
3133 .extra2 = SYSCTL_ONE,
3134 },
3135 #endif
3136 { }
3137 };
3138
3139 static struct ctl_table fs_table[] = {
3140 {
3141 .procname = "inode-nr",
3142 .data = &inodes_stat,
3143 .maxlen = 2*sizeof(long),
3144 .mode = 0444,
3145 .proc_handler = proc_nr_inodes,
3146 },
3147 {
3148 .procname = "inode-state",
3149 .data = &inodes_stat,
3150 .maxlen = 7*sizeof(long),
3151 .mode = 0444,
3152 .proc_handler = proc_nr_inodes,
3153 },
3154 {
3155 .procname = "file-nr",
3156 .data = &files_stat,
3157 .maxlen = sizeof(files_stat),
3158 .mode = 0444,
3159 .proc_handler = proc_nr_files,
3160 },
3161 {
3162 .procname = "file-max",
3163 .data = &files_stat.max_files,
3164 .maxlen = sizeof(files_stat.max_files),
3165 .mode = 0644,
3166 .proc_handler = proc_doulongvec_minmax,
3167 .extra1 = &zero_ul,
3168 .extra2 = &long_max,
3169 },
3170 {
3171 .procname = "nr_open",
3172 .data = &sysctl_nr_open,
3173 .maxlen = sizeof(unsigned int),
3174 .mode = 0644,
3175 .proc_handler = proc_dointvec_minmax,
3176 .extra1 = &sysctl_nr_open_min,
3177 .extra2 = &sysctl_nr_open_max,
3178 },
3179 {
3180 .procname = "dentry-state",
3181 .data = &dentry_stat,
3182 .maxlen = 6*sizeof(long),
3183 .mode = 0444,
3184 .proc_handler = proc_nr_dentry,
3185 },
3186 {
3187 .procname = "overflowuid",
3188 .data = &fs_overflowuid,
3189 .maxlen = sizeof(int),
3190 .mode = 0644,
3191 .proc_handler = proc_dointvec_minmax,
3192 .extra1 = &minolduid,
3193 .extra2 = &maxolduid,
3194 },
3195 {
3196 .procname = "overflowgid",
3197 .data = &fs_overflowgid,
3198 .maxlen = sizeof(int),
3199 .mode = 0644,
3200 .proc_handler = proc_dointvec_minmax,
3201 .extra1 = &minolduid,
3202 .extra2 = &maxolduid,
3203 },
3204 #ifdef CONFIG_FILE_LOCKING
3205 {
3206 .procname = "leases-enable",
3207 .data = &leases_enable,
3208 .maxlen = sizeof(int),
3209 .mode = 0644,
3210 .proc_handler = proc_dointvec,
3211 },
3212 #endif
3213 #ifdef CONFIG_DNOTIFY
3214 {
3215 .procname = "dir-notify-enable",
3216 .data = &dir_notify_enable,
3217 .maxlen = sizeof(int),
3218 .mode = 0644,
3219 .proc_handler = proc_dointvec,
3220 },
3221 #endif
3222 #ifdef CONFIG_MMU
3223 #ifdef CONFIG_FILE_LOCKING
3224 {
3225 .procname = "lease-break-time",
3226 .data = &lease_break_time,
3227 .maxlen = sizeof(int),
3228 .mode = 0644,
3229 .proc_handler = proc_dointvec,
3230 },
3231 #endif
3232 #ifdef CONFIG_AIO
3233 {
3234 .procname = "aio-nr",
3235 .data = &aio_nr,
3236 .maxlen = sizeof(aio_nr),
3237 .mode = 0444,
3238 .proc_handler = proc_doulongvec_minmax,
3239 },
3240 {
3241 .procname = "aio-max-nr",
3242 .data = &aio_max_nr,
3243 .maxlen = sizeof(aio_max_nr),
3244 .mode = 0644,
3245 .proc_handler = proc_doulongvec_minmax,
3246 },
3247 #endif /* CONFIG_AIO */
3248 #ifdef CONFIG_INOTIFY_USER
3249 {
3250 .procname = "inotify",
3251 .mode = 0555,
3252 .child = inotify_table,
3253 },
3254 #endif
3255 #ifdef CONFIG_FANOTIFY
3256 {
3257 .procname = "fanotify",
3258 .mode = 0555,
3259 .child = fanotify_table,
3260 },
3261 #endif
3262 #ifdef CONFIG_EPOLL
3263 {
3264 .procname = "epoll",
3265 .mode = 0555,
3266 .child = epoll_table,
3267 },
3268 #endif
3269 #endif
3270 {
3271 .procname = "protected_symlinks",
3272 .data = &sysctl_protected_symlinks,
3273 .maxlen = sizeof(int),
3274 .mode = 0600,
3275 .proc_handler = proc_dointvec_minmax,
3276 .extra1 = SYSCTL_ZERO,
3277 .extra2 = SYSCTL_ONE,
3278 },
3279 {
3280 .procname = "protected_hardlinks",
3281 .data = &sysctl_protected_hardlinks,
3282 .maxlen = sizeof(int),
3283 .mode = 0600,
3284 .proc_handler = proc_dointvec_minmax,
3285 .extra1 = SYSCTL_ZERO,
3286 .extra2 = SYSCTL_ONE,
3287 },
3288 {
3289 .procname = "protected_fifos",
3290 .data = &sysctl_protected_fifos,
3291 .maxlen = sizeof(int),
3292 .mode = 0600,
3293 .proc_handler = proc_dointvec_minmax,
3294 .extra1 = SYSCTL_ZERO,
3295 .extra2 = &two,
3296 },
3297 {
3298 .procname = "protected_regular",
3299 .data = &sysctl_protected_regular,
3300 .maxlen = sizeof(int),
3301 .mode = 0600,
3302 .proc_handler = proc_dointvec_minmax,
3303 .extra1 = SYSCTL_ZERO,
3304 .extra2 = &two,
3305 },
3306 {
3307 .procname = "suid_dumpable",
3308 .data = &suid_dumpable,
3309 .maxlen = sizeof(int),
3310 .mode = 0644,
3311 .proc_handler = proc_dointvec_minmax_coredump,
3312 .extra1 = SYSCTL_ZERO,
3313 .extra2 = &two,
3314 },
3315 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
3316 {
3317 .procname = "binfmt_misc",
3318 .mode = 0555,
3319 .child = sysctl_mount_point,
3320 },
3321 #endif
3322 {
3323 .procname = "pipe-max-size",
3324 .data = &pipe_max_size,
3325 .maxlen = sizeof(pipe_max_size),
3326 .mode = 0644,
3327 .proc_handler = proc_dopipe_max_size,
3328 },
3329 {
3330 .procname = "pipe-user-pages-hard",
3331 .data = &pipe_user_pages_hard,
3332 .maxlen = sizeof(pipe_user_pages_hard),
3333 .mode = 0644,
3334 .proc_handler = proc_doulongvec_minmax,
3335 },
3336 {
3337 .procname = "pipe-user-pages-soft",
3338 .data = &pipe_user_pages_soft,
3339 .maxlen = sizeof(pipe_user_pages_soft),
3340 .mode = 0644,
3341 .proc_handler = proc_doulongvec_minmax,
3342 },
3343 {
3344 .procname = "mount-max",
3345 .data = &sysctl_mount_max,
3346 .maxlen = sizeof(unsigned int),
3347 .mode = 0644,
3348 .proc_handler = proc_dointvec_minmax,
3349 .extra1 = SYSCTL_ONE,
3350 },
3351 { }
3352 };
3353
3354 static struct ctl_table debug_table[] = {
3355 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
3356 {
3357 .procname = "exception-trace",
3358 .data = &show_unhandled_signals,
3359 .maxlen = sizeof(int),
3360 .mode = 0644,
3361 .proc_handler = proc_dointvec
3362 },
3363 #endif
3364 #if defined(CONFIG_OPTPROBES)
3365 {
3366 .procname = "kprobes-optimization",
3367 .data = &sysctl_kprobes_optimization,
3368 .maxlen = sizeof(int),
3369 .mode = 0644,
3370 .proc_handler = proc_kprobes_optimization_handler,
3371 .extra1 = SYSCTL_ZERO,
3372 .extra2 = SYSCTL_ONE,
3373 },
3374 #endif
3375 { }
3376 };
3377
3378 static struct ctl_table dev_table[] = {
3379 { }
3380 };
3381
3382 static struct ctl_table sysctl_base_table[] = {
3383 {
3384 .procname = "kernel",
3385 .mode = 0555,
3386 .child = kern_table,
3387 },
3388 {
3389 .procname = "vm",
3390 .mode = 0555,
3391 .child = vm_table,
3392 },
3393 {
3394 .procname = "fs",
3395 .mode = 0555,
3396 .child = fs_table,
3397 },
3398 {
3399 .procname = "debug",
3400 .mode = 0555,
3401 .child = debug_table,
3402 },
3403 {
3404 .procname = "dev",
3405 .mode = 0555,
3406 .child = dev_table,
3407 },
3408 { }
3409 };
3410
3411 int __init sysctl_init(void)
3412 {
3413 struct ctl_table_header *hdr;
3414
3415 hdr = register_sysctl_table(sysctl_base_table);
3416 kmemleak_not_leak(hdr);
3417 return 0;
3418 }
3419 #endif /* CONFIG_SYSCTL */
3420 /*
3421 * No sense putting this after each symbol definition, twice,
3422 * exception granted :-)
3423 */
3424 EXPORT_SYMBOL(proc_dointvec);
3425 EXPORT_SYMBOL(proc_douintvec);
3426 EXPORT_SYMBOL(proc_dointvec_jiffies);
3427 EXPORT_SYMBOL(proc_dointvec_minmax);
3428 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3429 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3430 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3431 EXPORT_SYMBOL(proc_dostring);
3432 EXPORT_SYMBOL(proc_doulongvec_minmax);
3433 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3434 EXPORT_SYMBOL(proc_do_large_bitmap);