projects/linux·blob v5.14-rc4 · refs

3438 lines · 80.2 KiB

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