xref: /openbmc/linux/kernel/sysctl.c (revision e565a8ed1ee4b481539b66cd6f54df9ecf1e9861)
1 // SPDX-License-Identifier: GPL-2.0-only
2 /*
3  * sysctl.c: General linux system control interface
4  *
5  * Begun 24 March 1995, Stephen Tweedie
6  * Added /proc support, Dec 1995
7  * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
8  * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
9  * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
10  * Dynamic registration fixes, Stephen Tweedie.
11  * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
12  * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
13  *  Horn.
14  * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
15  * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
16  * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
17  *  Wendling.
18  * The list_for_each() macro wasn't appropriate for the sysctl loop.
19  *  Removed it and replaced it with older style, 03/23/00, Bill Wendling
20  */
21 
22 #include <linux/module.h>
23 #include <linux/mm.h>
24 #include <linux/swap.h>
25 #include <linux/slab.h>
26 #include <linux/sysctl.h>
27 #include <linux/bitmap.h>
28 #include <linux/signal.h>
29 #include <linux/panic.h>
30 #include <linux/printk.h>
31 #include <linux/proc_fs.h>
32 #include <linux/security.h>
33 #include <linux/ctype.h>
34 #include <linux/kmemleak.h>
35 #include <linux/filter.h>
36 #include <linux/fs.h>
37 #include <linux/init.h>
38 #include <linux/kernel.h>
39 #include <linux/kobject.h>
40 #include <linux/net.h>
41 #include <linux/sysrq.h>
42 #include <linux/highuid.h>
43 #include <linux/writeback.h>
44 #include <linux/ratelimit.h>
45 #include <linux/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/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/oom.h>
60 #include <linux/kmod.h>
61 #include <linux/capability.h>
62 #include <linux/binfmts.h>
63 #include <linux/sched/sysctl.h>
64 #include <linux/kexec.h>
65 #include <linux/bpf.h>
66 #include <linux/mount.h>
67 #include <linux/userfaultfd_k.h>
68 #include <linux/latencytop.h>
69 #include <linux/pid.h>
70 #include <linux/delayacct.h>
71 
72 #include "../lib/kstrtox.h"
73 
74 #include <linux/uaccess.h>
75 #include <asm/processor.h>
76 
77 #ifdef CONFIG_X86
78 #include <asm/nmi.h>
79 #include <asm/stacktrace.h>
80 #include <asm/io.h>
81 #endif
82 #ifdef CONFIG_SPARC
83 #include <asm/setup.h>
84 #endif
85 #ifdef CONFIG_BSD_PROCESS_ACCT
86 #include <linux/acct.h>
87 #endif
88 #ifdef CONFIG_RT_MUTEXES
89 #include <linux/rtmutex.h>
90 #endif
91 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
92 #include <linux/lockdep.h>
93 #endif
94 
95 #if defined(CONFIG_SYSCTL)
96 
97 /* Constants used for minimum and  maximum */
98 
99 #ifdef CONFIG_PERF_EVENTS
100 static const int six_hundred_forty_kb = 640 * 1024;
101 #endif
102 
103 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
104 static const unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
105 
106 static const int ngroups_max = NGROUPS_MAX;
107 static const int cap_last_cap = CAP_LAST_CAP;
108 
109 #ifdef CONFIG_PROC_SYSCTL
110 
111 /**
112  * enum sysctl_writes_mode - supported sysctl write modes
113  *
114  * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
115  *	to be written, and multiple writes on the same sysctl file descriptor
116  *	will rewrite the sysctl value, regardless of file position. No warning
117  *	is issued when the initial position is not 0.
118  * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
119  *	not 0.
120  * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
121  *	file position 0 and the value must be fully contained in the buffer
122  *	sent to the write syscall. If dealing with strings respect the file
123  *	position, but restrict this to the max length of the buffer, anything
124  *	passed the max length will be ignored. Multiple writes will append
125  *	to the buffer.
126  *
127  * These write modes control how current file position affects the behavior of
128  * updating sysctl values through the proc interface on each write.
129  */
130 enum sysctl_writes_mode {
131 	SYSCTL_WRITES_LEGACY		= -1,
132 	SYSCTL_WRITES_WARN		= 0,
133 	SYSCTL_WRITES_STRICT		= 1,
134 };
135 
136 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
137 #endif /* CONFIG_PROC_SYSCTL */
138 
139 #if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
140     defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
141 int sysctl_legacy_va_layout;
142 #endif
143 
144 #ifdef CONFIG_COMPACTION
145 /* min_extfrag_threshold is SYSCTL_ZERO */;
146 static const int max_extfrag_threshold = 1000;
147 #endif
148 
149 #endif /* CONFIG_SYSCTL */
150 
151 #if defined(CONFIG_BPF_SYSCALL) && defined(CONFIG_SYSCTL)
152 static int bpf_stats_handler(struct ctl_table *table, int write,
153 			     void *buffer, size_t *lenp, loff_t *ppos)
154 {
155 	struct static_key *key = (struct static_key *)table->data;
156 	static int saved_val;
157 	int val, ret;
158 	struct ctl_table tmp = {
159 		.data   = &val,
160 		.maxlen = sizeof(val),
161 		.mode   = table->mode,
162 		.extra1 = SYSCTL_ZERO,
163 		.extra2 = SYSCTL_ONE,
164 	};
165 
166 	if (write && !capable(CAP_SYS_ADMIN))
167 		return -EPERM;
168 
169 	mutex_lock(&bpf_stats_enabled_mutex);
170 	val = saved_val;
171 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
172 	if (write && !ret && val != saved_val) {
173 		if (val)
174 			static_key_slow_inc(key);
175 		else
176 			static_key_slow_dec(key);
177 		saved_val = val;
178 	}
179 	mutex_unlock(&bpf_stats_enabled_mutex);
180 	return ret;
181 }
182 
183 static int bpf_unpriv_handler(struct ctl_table *table, int write,
184 			      void *buffer, size_t *lenp, loff_t *ppos)
185 {
186 	int ret, unpriv_enable = *(int *)table->data;
187 	bool locked_state = unpriv_enable == 1;
188 	struct ctl_table tmp = *table;
189 
190 	if (write && !capable(CAP_SYS_ADMIN))
191 		return -EPERM;
192 
193 	tmp.data = &unpriv_enable;
194 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
195 	if (write && !ret) {
196 		if (locked_state && unpriv_enable != 1)
197 			return -EPERM;
198 		*(int *)table->data = unpriv_enable;
199 	}
200 	return ret;
201 }
202 #endif /* CONFIG_BPF_SYSCALL && CONFIG_SYSCTL */
203 
204 /*
205  * /proc/sys support
206  */
207 
208 #ifdef CONFIG_PROC_SYSCTL
209 
210 static int _proc_do_string(char *data, int maxlen, int write,
211 		char *buffer, size_t *lenp, loff_t *ppos)
212 {
213 	size_t len;
214 	char c, *p;
215 
216 	if (!data || !maxlen || !*lenp) {
217 		*lenp = 0;
218 		return 0;
219 	}
220 
221 	if (write) {
222 		if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
223 			/* Only continue writes not past the end of buffer. */
224 			len = strlen(data);
225 			if (len > maxlen - 1)
226 				len = maxlen - 1;
227 
228 			if (*ppos > len)
229 				return 0;
230 			len = *ppos;
231 		} else {
232 			/* Start writing from beginning of buffer. */
233 			len = 0;
234 		}
235 
236 		*ppos += *lenp;
237 		p = buffer;
238 		while ((p - buffer) < *lenp && len < maxlen - 1) {
239 			c = *(p++);
240 			if (c == 0 || c == '\n')
241 				break;
242 			data[len++] = c;
243 		}
244 		data[len] = 0;
245 	} else {
246 		len = strlen(data);
247 		if (len > maxlen)
248 			len = maxlen;
249 
250 		if (*ppos > len) {
251 			*lenp = 0;
252 			return 0;
253 		}
254 
255 		data += *ppos;
256 		len  -= *ppos;
257 
258 		if (len > *lenp)
259 			len = *lenp;
260 		if (len)
261 			memcpy(buffer, data, len);
262 		if (len < *lenp) {
263 			buffer[len] = '\n';
264 			len++;
265 		}
266 		*lenp = len;
267 		*ppos += len;
268 	}
269 	return 0;
270 }
271 
272 static void warn_sysctl_write(struct ctl_table *table)
273 {
274 	pr_warn_once("%s wrote to %s when file position was not 0!\n"
275 		"This will not be supported in the future. To silence this\n"
276 		"warning, set kernel.sysctl_writes_strict = -1\n",
277 		current->comm, table->procname);
278 }
279 
280 /**
281  * proc_first_pos_non_zero_ignore - check if first position is allowed
282  * @ppos: file position
283  * @table: the sysctl table
284  *
285  * Returns true if the first position is non-zero and the sysctl_writes_strict
286  * mode indicates this is not allowed for numeric input types. String proc
287  * handlers can ignore the return value.
288  */
289 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
290 					   struct ctl_table *table)
291 {
292 	if (!*ppos)
293 		return false;
294 
295 	switch (sysctl_writes_strict) {
296 	case SYSCTL_WRITES_STRICT:
297 		return true;
298 	case SYSCTL_WRITES_WARN:
299 		warn_sysctl_write(table);
300 		return false;
301 	default:
302 		return false;
303 	}
304 }
305 
306 /**
307  * proc_dostring - read a string sysctl
308  * @table: the sysctl table
309  * @write: %TRUE if this is a write to the sysctl file
310  * @buffer: the user buffer
311  * @lenp: the size of the user buffer
312  * @ppos: file position
313  *
314  * Reads/writes a string from/to the user buffer. If the kernel
315  * buffer provided is not large enough to hold the string, the
316  * string is truncated. The copied string is %NULL-terminated.
317  * If the string is being read by the user process, it is copied
318  * and a newline '\n' is added. It is truncated if the buffer is
319  * not large enough.
320  *
321  * Returns 0 on success.
322  */
323 int proc_dostring(struct ctl_table *table, int write,
324 		  void *buffer, size_t *lenp, loff_t *ppos)
325 {
326 	if (write)
327 		proc_first_pos_non_zero_ignore(ppos, table);
328 
329 	return _proc_do_string(table->data, table->maxlen, write, buffer, lenp,
330 			ppos);
331 }
332 
333 static size_t proc_skip_spaces(char **buf)
334 {
335 	size_t ret;
336 	char *tmp = skip_spaces(*buf);
337 	ret = tmp - *buf;
338 	*buf = tmp;
339 	return ret;
340 }
341 
342 static void proc_skip_char(char **buf, size_t *size, const char v)
343 {
344 	while (*size) {
345 		if (**buf != v)
346 			break;
347 		(*size)--;
348 		(*buf)++;
349 	}
350 }
351 
352 /**
353  * strtoul_lenient - parse an ASCII formatted integer from a buffer and only
354  *                   fail on overflow
355  *
356  * @cp: kernel buffer containing the string to parse
357  * @endp: pointer to store the trailing characters
358  * @base: the base to use
359  * @res: where the parsed integer will be stored
360  *
361  * In case of success 0 is returned and @res will contain the parsed integer,
362  * @endp will hold any trailing characters.
363  * This function will fail the parse on overflow. If there wasn't an overflow
364  * the function will defer the decision what characters count as invalid to the
365  * caller.
366  */
367 static int strtoul_lenient(const char *cp, char **endp, unsigned int base,
368 			   unsigned long *res)
369 {
370 	unsigned long long result;
371 	unsigned int rv;
372 
373 	cp = _parse_integer_fixup_radix(cp, &base);
374 	rv = _parse_integer(cp, base, &result);
375 	if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result))
376 		return -ERANGE;
377 
378 	cp += rv;
379 
380 	if (endp)
381 		*endp = (char *)cp;
382 
383 	*res = (unsigned long)result;
384 	return 0;
385 }
386 
387 #define TMPBUFLEN 22
388 /**
389  * proc_get_long - reads an ASCII formatted integer from a user buffer
390  *
391  * @buf: a kernel buffer
392  * @size: size of the kernel buffer
393  * @val: this is where the number will be stored
394  * @neg: set to %TRUE if number is negative
395  * @perm_tr: a vector which contains the allowed trailers
396  * @perm_tr_len: size of the perm_tr vector
397  * @tr: pointer to store the trailer character
398  *
399  * In case of success %0 is returned and @buf and @size are updated with
400  * the amount of bytes read. If @tr is non-NULL and a trailing
401  * character exists (size is non-zero after returning from this
402  * function), @tr is updated with the trailing character.
403  */
404 static int proc_get_long(char **buf, size_t *size,
405 			  unsigned long *val, bool *neg,
406 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
407 {
408 	int len;
409 	char *p, tmp[TMPBUFLEN];
410 
411 	if (!*size)
412 		return -EINVAL;
413 
414 	len = *size;
415 	if (len > TMPBUFLEN - 1)
416 		len = TMPBUFLEN - 1;
417 
418 	memcpy(tmp, *buf, len);
419 
420 	tmp[len] = 0;
421 	p = tmp;
422 	if (*p == '-' && *size > 1) {
423 		*neg = true;
424 		p++;
425 	} else
426 		*neg = false;
427 	if (!isdigit(*p))
428 		return -EINVAL;
429 
430 	if (strtoul_lenient(p, &p, 0, val))
431 		return -EINVAL;
432 
433 	len = p - tmp;
434 
435 	/* We don't know if the next char is whitespace thus we may accept
436 	 * invalid integers (e.g. 1234...a) or two integers instead of one
437 	 * (e.g. 123...1). So lets not allow such large numbers. */
438 	if (len == TMPBUFLEN - 1)
439 		return -EINVAL;
440 
441 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
442 		return -EINVAL;
443 
444 	if (tr && (len < *size))
445 		*tr = *p;
446 
447 	*buf += len;
448 	*size -= len;
449 
450 	return 0;
451 }
452 
453 /**
454  * proc_put_long - converts an integer to a decimal ASCII formatted string
455  *
456  * @buf: the user buffer
457  * @size: the size of the user buffer
458  * @val: the integer to be converted
459  * @neg: sign of the number, %TRUE for negative
460  *
461  * In case of success @buf and @size are updated with the amount of bytes
462  * written.
463  */
464 static void proc_put_long(void **buf, size_t *size, unsigned long val, bool neg)
465 {
466 	int len;
467 	char tmp[TMPBUFLEN], *p = tmp;
468 
469 	sprintf(p, "%s%lu", neg ? "-" : "", val);
470 	len = strlen(tmp);
471 	if (len > *size)
472 		len = *size;
473 	memcpy(*buf, tmp, len);
474 	*size -= len;
475 	*buf += len;
476 }
477 #undef TMPBUFLEN
478 
479 static void proc_put_char(void **buf, size_t *size, char c)
480 {
481 	if (*size) {
482 		char **buffer = (char **)buf;
483 		**buffer = c;
484 
485 		(*size)--;
486 		(*buffer)++;
487 		*buf = *buffer;
488 	}
489 }
490 
491 static int do_proc_dobool_conv(bool *negp, unsigned long *lvalp,
492 				int *valp,
493 				int write, void *data)
494 {
495 	if (write) {
496 		*(bool *)valp = *lvalp;
497 	} else {
498 		int val = *(bool *)valp;
499 
500 		*lvalp = (unsigned long)val;
501 		*negp = false;
502 	}
503 	return 0;
504 }
505 
506 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
507 				 int *valp,
508 				 int write, void *data)
509 {
510 	if (write) {
511 		if (*negp) {
512 			if (*lvalp > (unsigned long) INT_MAX + 1)
513 				return -EINVAL;
514 			*valp = -*lvalp;
515 		} else {
516 			if (*lvalp > (unsigned long) INT_MAX)
517 				return -EINVAL;
518 			*valp = *lvalp;
519 		}
520 	} else {
521 		int val = *valp;
522 		if (val < 0) {
523 			*negp = true;
524 			*lvalp = -(unsigned long)val;
525 		} else {
526 			*negp = false;
527 			*lvalp = (unsigned long)val;
528 		}
529 	}
530 	return 0;
531 }
532 
533 static int do_proc_douintvec_conv(unsigned long *lvalp,
534 				  unsigned int *valp,
535 				  int write, void *data)
536 {
537 	if (write) {
538 		if (*lvalp > UINT_MAX)
539 			return -EINVAL;
540 		*valp = *lvalp;
541 	} else {
542 		unsigned int val = *valp;
543 		*lvalp = (unsigned long)val;
544 	}
545 	return 0;
546 }
547 
548 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
549 
550 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
551 		  int write, void *buffer,
552 		  size_t *lenp, loff_t *ppos,
553 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
554 			      int write, void *data),
555 		  void *data)
556 {
557 	int *i, vleft, first = 1, err = 0;
558 	size_t left;
559 	char *p;
560 
561 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
562 		*lenp = 0;
563 		return 0;
564 	}
565 
566 	i = (int *) tbl_data;
567 	vleft = table->maxlen / sizeof(*i);
568 	left = *lenp;
569 
570 	if (!conv)
571 		conv = do_proc_dointvec_conv;
572 
573 	if (write) {
574 		if (proc_first_pos_non_zero_ignore(ppos, table))
575 			goto out;
576 
577 		if (left > PAGE_SIZE - 1)
578 			left = PAGE_SIZE - 1;
579 		p = buffer;
580 	}
581 
582 	for (; left && vleft--; i++, first=0) {
583 		unsigned long lval;
584 		bool neg;
585 
586 		if (write) {
587 			left -= proc_skip_spaces(&p);
588 
589 			if (!left)
590 				break;
591 			err = proc_get_long(&p, &left, &lval, &neg,
592 					     proc_wspace_sep,
593 					     sizeof(proc_wspace_sep), NULL);
594 			if (err)
595 				break;
596 			if (conv(&neg, &lval, i, 1, data)) {
597 				err = -EINVAL;
598 				break;
599 			}
600 		} else {
601 			if (conv(&neg, &lval, i, 0, data)) {
602 				err = -EINVAL;
603 				break;
604 			}
605 			if (!first)
606 				proc_put_char(&buffer, &left, '\t');
607 			proc_put_long(&buffer, &left, lval, neg);
608 		}
609 	}
610 
611 	if (!write && !first && left && !err)
612 		proc_put_char(&buffer, &left, '\n');
613 	if (write && !err && left)
614 		left -= proc_skip_spaces(&p);
615 	if (write && first)
616 		return err ? : -EINVAL;
617 	*lenp -= left;
618 out:
619 	*ppos += *lenp;
620 	return err;
621 }
622 
623 static int do_proc_dointvec(struct ctl_table *table, int write,
624 		  void *buffer, size_t *lenp, loff_t *ppos,
625 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
626 			      int write, void *data),
627 		  void *data)
628 {
629 	return __do_proc_dointvec(table->data, table, write,
630 			buffer, lenp, ppos, conv, data);
631 }
632 
633 static int do_proc_douintvec_w(unsigned int *tbl_data,
634 			       struct ctl_table *table,
635 			       void *buffer,
636 			       size_t *lenp, loff_t *ppos,
637 			       int (*conv)(unsigned long *lvalp,
638 					   unsigned int *valp,
639 					   int write, void *data),
640 			       void *data)
641 {
642 	unsigned long lval;
643 	int err = 0;
644 	size_t left;
645 	bool neg;
646 	char *p = buffer;
647 
648 	left = *lenp;
649 
650 	if (proc_first_pos_non_zero_ignore(ppos, table))
651 		goto bail_early;
652 
653 	if (left > PAGE_SIZE - 1)
654 		left = PAGE_SIZE - 1;
655 
656 	left -= proc_skip_spaces(&p);
657 	if (!left) {
658 		err = -EINVAL;
659 		goto out_free;
660 	}
661 
662 	err = proc_get_long(&p, &left, &lval, &neg,
663 			     proc_wspace_sep,
664 			     sizeof(proc_wspace_sep), NULL);
665 	if (err || neg) {
666 		err = -EINVAL;
667 		goto out_free;
668 	}
669 
670 	if (conv(&lval, tbl_data, 1, data)) {
671 		err = -EINVAL;
672 		goto out_free;
673 	}
674 
675 	if (!err && left)
676 		left -= proc_skip_spaces(&p);
677 
678 out_free:
679 	if (err)
680 		return -EINVAL;
681 
682 	return 0;
683 
684 	/* This is in keeping with old __do_proc_dointvec() */
685 bail_early:
686 	*ppos += *lenp;
687 	return err;
688 }
689 
690 static int do_proc_douintvec_r(unsigned int *tbl_data, void *buffer,
691 			       size_t *lenp, loff_t *ppos,
692 			       int (*conv)(unsigned long *lvalp,
693 					   unsigned int *valp,
694 					   int write, void *data),
695 			       void *data)
696 {
697 	unsigned long lval;
698 	int err = 0;
699 	size_t left;
700 
701 	left = *lenp;
702 
703 	if (conv(&lval, tbl_data, 0, data)) {
704 		err = -EINVAL;
705 		goto out;
706 	}
707 
708 	proc_put_long(&buffer, &left, lval, false);
709 	if (!left)
710 		goto out;
711 
712 	proc_put_char(&buffer, &left, '\n');
713 
714 out:
715 	*lenp -= left;
716 	*ppos += *lenp;
717 
718 	return err;
719 }
720 
721 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
722 			       int write, void *buffer,
723 			       size_t *lenp, loff_t *ppos,
724 			       int (*conv)(unsigned long *lvalp,
725 					   unsigned int *valp,
726 					   int write, void *data),
727 			       void *data)
728 {
729 	unsigned int *i, vleft;
730 
731 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
732 		*lenp = 0;
733 		return 0;
734 	}
735 
736 	i = (unsigned int *) tbl_data;
737 	vleft = table->maxlen / sizeof(*i);
738 
739 	/*
740 	 * Arrays are not supported, keep this simple. *Do not* add
741 	 * support for them.
742 	 */
743 	if (vleft != 1) {
744 		*lenp = 0;
745 		return -EINVAL;
746 	}
747 
748 	if (!conv)
749 		conv = do_proc_douintvec_conv;
750 
751 	if (write)
752 		return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
753 					   conv, data);
754 	return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
755 }
756 
757 int do_proc_douintvec(struct ctl_table *table, int write,
758 		      void *buffer, size_t *lenp, loff_t *ppos,
759 		      int (*conv)(unsigned long *lvalp,
760 				  unsigned int *valp,
761 				  int write, void *data),
762 		      void *data)
763 {
764 	return __do_proc_douintvec(table->data, table, write,
765 				   buffer, lenp, ppos, conv, data);
766 }
767 
768 /**
769  * proc_dobool - read/write a bool
770  * @table: the sysctl table
771  * @write: %TRUE if this is a write to the sysctl file
772  * @buffer: the user buffer
773  * @lenp: the size of the user buffer
774  * @ppos: file position
775  *
776  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
777  * values from/to the user buffer, treated as an ASCII string.
778  *
779  * Returns 0 on success.
780  */
781 int proc_dobool(struct ctl_table *table, int write, void *buffer,
782 		size_t *lenp, loff_t *ppos)
783 {
784 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
785 				do_proc_dobool_conv, NULL);
786 }
787 
788 /**
789  * proc_dointvec - read a vector of integers
790  * @table: the sysctl table
791  * @write: %TRUE if this is a write to the sysctl file
792  * @buffer: the user buffer
793  * @lenp: the size of the user buffer
794  * @ppos: file position
795  *
796  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
797  * values from/to the user buffer, treated as an ASCII string.
798  *
799  * Returns 0 on success.
800  */
801 int proc_dointvec(struct ctl_table *table, int write, void *buffer,
802 		  size_t *lenp, loff_t *ppos)
803 {
804 	return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
805 }
806 
807 #ifdef CONFIG_COMPACTION
808 static int proc_dointvec_minmax_warn_RT_change(struct ctl_table *table,
809 		int write, void *buffer, size_t *lenp, loff_t *ppos)
810 {
811 	int ret, old;
812 
813 	if (!IS_ENABLED(CONFIG_PREEMPT_RT) || !write)
814 		return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
815 
816 	old = *(int *)table->data;
817 	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
818 	if (ret)
819 		return ret;
820 	if (old != *(int *)table->data)
821 		pr_warn_once("sysctl attribute %s changed by %s[%d]\n",
822 			     table->procname, current->comm,
823 			     task_pid_nr(current));
824 	return ret;
825 }
826 #endif
827 
828 /**
829  * proc_douintvec - read a vector of unsigned integers
830  * @table: the sysctl table
831  * @write: %TRUE if this is a write to the sysctl file
832  * @buffer: the user buffer
833  * @lenp: the size of the user buffer
834  * @ppos: file position
835  *
836  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
837  * values from/to the user buffer, treated as an ASCII string.
838  *
839  * Returns 0 on success.
840  */
841 int proc_douintvec(struct ctl_table *table, int write, void *buffer,
842 		size_t *lenp, loff_t *ppos)
843 {
844 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
845 				 do_proc_douintvec_conv, NULL);
846 }
847 
848 /*
849  * Taint values can only be increased
850  * This means we can safely use a temporary.
851  */
852 static int proc_taint(struct ctl_table *table, int write,
853 			       void *buffer, size_t *lenp, loff_t *ppos)
854 {
855 	struct ctl_table t;
856 	unsigned long tmptaint = get_taint();
857 	int err;
858 
859 	if (write && !capable(CAP_SYS_ADMIN))
860 		return -EPERM;
861 
862 	t = *table;
863 	t.data = &tmptaint;
864 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
865 	if (err < 0)
866 		return err;
867 
868 	if (write) {
869 		int i;
870 
871 		/*
872 		 * If we are relying on panic_on_taint not producing
873 		 * false positives due to userspace input, bail out
874 		 * before setting the requested taint flags.
875 		 */
876 		if (panic_on_taint_nousertaint && (tmptaint & panic_on_taint))
877 			return -EINVAL;
878 
879 		/*
880 		 * Poor man's atomic or. Not worth adding a primitive
881 		 * to everyone's atomic.h for this
882 		 */
883 		for (i = 0; i < TAINT_FLAGS_COUNT; i++)
884 			if ((1UL << i) & tmptaint)
885 				add_taint(i, LOCKDEP_STILL_OK);
886 	}
887 
888 	return err;
889 }
890 
891 /**
892  * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
893  * @min: pointer to minimum allowable value
894  * @max: pointer to maximum allowable value
895  *
896  * The do_proc_dointvec_minmax_conv_param structure provides the
897  * minimum and maximum values for doing range checking for those sysctl
898  * parameters that use the proc_dointvec_minmax() handler.
899  */
900 struct do_proc_dointvec_minmax_conv_param {
901 	int *min;
902 	int *max;
903 };
904 
905 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
906 					int *valp,
907 					int write, void *data)
908 {
909 	int tmp, ret;
910 	struct do_proc_dointvec_minmax_conv_param *param = data;
911 	/*
912 	 * If writing, first do so via a temporary local int so we can
913 	 * bounds-check it before touching *valp.
914 	 */
915 	int *ip = write ? &tmp : valp;
916 
917 	ret = do_proc_dointvec_conv(negp, lvalp, ip, write, data);
918 	if (ret)
919 		return ret;
920 
921 	if (write) {
922 		if ((param->min && *param->min > tmp) ||
923 		    (param->max && *param->max < tmp))
924 			return -EINVAL;
925 		*valp = tmp;
926 	}
927 
928 	return 0;
929 }
930 
931 /**
932  * proc_dointvec_minmax - read a vector of integers with min/max values
933  * @table: the sysctl table
934  * @write: %TRUE if this is a write to the sysctl file
935  * @buffer: the user buffer
936  * @lenp: the size of the user buffer
937  * @ppos: file position
938  *
939  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
940  * values from/to the user buffer, treated as an ASCII string.
941  *
942  * This routine will ensure the values are within the range specified by
943  * table->extra1 (min) and table->extra2 (max).
944  *
945  * Returns 0 on success or -EINVAL on write when the range check fails.
946  */
947 int proc_dointvec_minmax(struct ctl_table *table, int write,
948 		  void *buffer, size_t *lenp, loff_t *ppos)
949 {
950 	struct do_proc_dointvec_minmax_conv_param param = {
951 		.min = (int *) table->extra1,
952 		.max = (int *) table->extra2,
953 	};
954 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
955 				do_proc_dointvec_minmax_conv, &param);
956 }
957 
958 /**
959  * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
960  * @min: pointer to minimum allowable value
961  * @max: pointer to maximum allowable value
962  *
963  * The do_proc_douintvec_minmax_conv_param structure provides the
964  * minimum and maximum values for doing range checking for those sysctl
965  * parameters that use the proc_douintvec_minmax() handler.
966  */
967 struct do_proc_douintvec_minmax_conv_param {
968 	unsigned int *min;
969 	unsigned int *max;
970 };
971 
972 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
973 					 unsigned int *valp,
974 					 int write, void *data)
975 {
976 	int ret;
977 	unsigned int tmp;
978 	struct do_proc_douintvec_minmax_conv_param *param = data;
979 	/* write via temporary local uint for bounds-checking */
980 	unsigned int *up = write ? &tmp : valp;
981 
982 	ret = do_proc_douintvec_conv(lvalp, up, write, data);
983 	if (ret)
984 		return ret;
985 
986 	if (write) {
987 		if ((param->min && *param->min > tmp) ||
988 		    (param->max && *param->max < tmp))
989 			return -ERANGE;
990 
991 		*valp = tmp;
992 	}
993 
994 	return 0;
995 }
996 
997 /**
998  * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
999  * @table: the sysctl table
1000  * @write: %TRUE if this is a write to the sysctl file
1001  * @buffer: the user buffer
1002  * @lenp: the size of the user buffer
1003  * @ppos: file position
1004  *
1005  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
1006  * values from/to the user buffer, treated as an ASCII string. Negative
1007  * strings are not allowed.
1008  *
1009  * This routine will ensure the values are within the range specified by
1010  * table->extra1 (min) and table->extra2 (max). There is a final sanity
1011  * check for UINT_MAX to avoid having to support wrap around uses from
1012  * userspace.
1013  *
1014  * Returns 0 on success or -ERANGE on write when the range check fails.
1015  */
1016 int proc_douintvec_minmax(struct ctl_table *table, int write,
1017 			  void *buffer, size_t *lenp, loff_t *ppos)
1018 {
1019 	struct do_proc_douintvec_minmax_conv_param param = {
1020 		.min = (unsigned int *) table->extra1,
1021 		.max = (unsigned int *) table->extra2,
1022 	};
1023 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
1024 				 do_proc_douintvec_minmax_conv, &param);
1025 }
1026 
1027 /**
1028  * proc_dou8vec_minmax - read a vector of unsigned chars with min/max values
1029  * @table: the sysctl table
1030  * @write: %TRUE if this is a write to the sysctl file
1031  * @buffer: the user buffer
1032  * @lenp: the size of the user buffer
1033  * @ppos: file position
1034  *
1035  * Reads/writes up to table->maxlen/sizeof(u8) unsigned chars
1036  * values from/to the user buffer, treated as an ASCII string. Negative
1037  * strings are not allowed.
1038  *
1039  * This routine will ensure the values are within the range specified by
1040  * table->extra1 (min) and table->extra2 (max).
1041  *
1042  * Returns 0 on success or an error on write when the range check fails.
1043  */
1044 int proc_dou8vec_minmax(struct ctl_table *table, int write,
1045 			void *buffer, size_t *lenp, loff_t *ppos)
1046 {
1047 	struct ctl_table tmp;
1048 	unsigned int min = 0, max = 255U, val;
1049 	u8 *data = table->data;
1050 	struct do_proc_douintvec_minmax_conv_param param = {
1051 		.min = &min,
1052 		.max = &max,
1053 	};
1054 	int res;
1055 
1056 	/* Do not support arrays yet. */
1057 	if (table->maxlen != sizeof(u8))
1058 		return -EINVAL;
1059 
1060 	if (table->extra1) {
1061 		min = *(unsigned int *) table->extra1;
1062 		if (min > 255U)
1063 			return -EINVAL;
1064 	}
1065 	if (table->extra2) {
1066 		max = *(unsigned int *) table->extra2;
1067 		if (max > 255U)
1068 			return -EINVAL;
1069 	}
1070 
1071 	tmp = *table;
1072 
1073 	tmp.maxlen = sizeof(val);
1074 	tmp.data = &val;
1075 	val = *data;
1076 	res = do_proc_douintvec(&tmp, write, buffer, lenp, ppos,
1077 				do_proc_douintvec_minmax_conv, &param);
1078 	if (res)
1079 		return res;
1080 	if (write)
1081 		*data = val;
1082 	return 0;
1083 }
1084 EXPORT_SYMBOL_GPL(proc_dou8vec_minmax);
1085 
1086 #ifdef CONFIG_MAGIC_SYSRQ
1087 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
1088 				void *buffer, size_t *lenp, loff_t *ppos)
1089 {
1090 	int tmp, ret;
1091 
1092 	tmp = sysrq_mask();
1093 
1094 	ret = __do_proc_dointvec(&tmp, table, write, buffer,
1095 			       lenp, ppos, NULL, NULL);
1096 	if (ret || !write)
1097 		return ret;
1098 
1099 	if (write)
1100 		sysrq_toggle_support(tmp);
1101 
1102 	return 0;
1103 }
1104 #endif
1105 
1106 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table,
1107 		int write, void *buffer, size_t *lenp, loff_t *ppos,
1108 		unsigned long convmul, unsigned long convdiv)
1109 {
1110 	unsigned long *i, *min, *max;
1111 	int vleft, first = 1, err = 0;
1112 	size_t left;
1113 	char *p;
1114 
1115 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
1116 		*lenp = 0;
1117 		return 0;
1118 	}
1119 
1120 	i = (unsigned long *) data;
1121 	min = (unsigned long *) table->extra1;
1122 	max = (unsigned long *) table->extra2;
1123 	vleft = table->maxlen / sizeof(unsigned long);
1124 	left = *lenp;
1125 
1126 	if (write) {
1127 		if (proc_first_pos_non_zero_ignore(ppos, table))
1128 			goto out;
1129 
1130 		if (left > PAGE_SIZE - 1)
1131 			left = PAGE_SIZE - 1;
1132 		p = buffer;
1133 	}
1134 
1135 	for (; left && vleft--; i++, first = 0) {
1136 		unsigned long val;
1137 
1138 		if (write) {
1139 			bool neg;
1140 
1141 			left -= proc_skip_spaces(&p);
1142 			if (!left)
1143 				break;
1144 
1145 			err = proc_get_long(&p, &left, &val, &neg,
1146 					     proc_wspace_sep,
1147 					     sizeof(proc_wspace_sep), NULL);
1148 			if (err)
1149 				break;
1150 			if (neg)
1151 				continue;
1152 			val = convmul * val / convdiv;
1153 			if ((min && val < *min) || (max && val > *max)) {
1154 				err = -EINVAL;
1155 				break;
1156 			}
1157 			*i = val;
1158 		} else {
1159 			val = convdiv * (*i) / convmul;
1160 			if (!first)
1161 				proc_put_char(&buffer, &left, '\t');
1162 			proc_put_long(&buffer, &left, val, false);
1163 		}
1164 	}
1165 
1166 	if (!write && !first && left && !err)
1167 		proc_put_char(&buffer, &left, '\n');
1168 	if (write && !err)
1169 		left -= proc_skip_spaces(&p);
1170 	if (write && first)
1171 		return err ? : -EINVAL;
1172 	*lenp -= left;
1173 out:
1174 	*ppos += *lenp;
1175 	return err;
1176 }
1177 
1178 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
1179 		void *buffer, size_t *lenp, loff_t *ppos, unsigned long convmul,
1180 		unsigned long convdiv)
1181 {
1182 	return __do_proc_doulongvec_minmax(table->data, table, write,
1183 			buffer, lenp, ppos, convmul, convdiv);
1184 }
1185 
1186 /**
1187  * proc_doulongvec_minmax - read a vector of long integers with min/max values
1188  * @table: the sysctl table
1189  * @write: %TRUE if this is a write to the sysctl file
1190  * @buffer: the user buffer
1191  * @lenp: the size of the user buffer
1192  * @ppos: file position
1193  *
1194  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1195  * values from/to the user buffer, treated as an ASCII string.
1196  *
1197  * This routine will ensure the values are within the range specified by
1198  * table->extra1 (min) and table->extra2 (max).
1199  *
1200  * Returns 0 on success.
1201  */
1202 int proc_doulongvec_minmax(struct ctl_table *table, int write,
1203 			   void *buffer, size_t *lenp, loff_t *ppos)
1204 {
1205     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
1206 }
1207 
1208 /**
1209  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
1210  * @table: the sysctl table
1211  * @write: %TRUE if this is a write to the sysctl file
1212  * @buffer: the user buffer
1213  * @lenp: the size of the user buffer
1214  * @ppos: file position
1215  *
1216  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1217  * values from/to the user buffer, treated as an ASCII string. The values
1218  * are treated as milliseconds, and converted to jiffies when they are stored.
1219  *
1220  * This routine will ensure the values are within the range specified by
1221  * table->extra1 (min) and table->extra2 (max).
1222  *
1223  * Returns 0 on success.
1224  */
1225 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1226 				      void *buffer, size_t *lenp, loff_t *ppos)
1227 {
1228     return do_proc_doulongvec_minmax(table, write, buffer,
1229 				     lenp, ppos, HZ, 1000l);
1230 }
1231 
1232 
1233 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
1234 					 int *valp,
1235 					 int write, void *data)
1236 {
1237 	if (write) {
1238 		if (*lvalp > INT_MAX / HZ)
1239 			return 1;
1240 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
1241 	} else {
1242 		int val = *valp;
1243 		unsigned long lval;
1244 		if (val < 0) {
1245 			*negp = true;
1246 			lval = -(unsigned long)val;
1247 		} else {
1248 			*negp = false;
1249 			lval = (unsigned long)val;
1250 		}
1251 		*lvalp = lval / HZ;
1252 	}
1253 	return 0;
1254 }
1255 
1256 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
1257 						int *valp,
1258 						int write, void *data)
1259 {
1260 	if (write) {
1261 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
1262 			return 1;
1263 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
1264 	} else {
1265 		int val = *valp;
1266 		unsigned long lval;
1267 		if (val < 0) {
1268 			*negp = true;
1269 			lval = -(unsigned long)val;
1270 		} else {
1271 			*negp = false;
1272 			lval = (unsigned long)val;
1273 		}
1274 		*lvalp = jiffies_to_clock_t(lval);
1275 	}
1276 	return 0;
1277 }
1278 
1279 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
1280 					    int *valp,
1281 					    int write, void *data)
1282 {
1283 	if (write) {
1284 		unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
1285 
1286 		if (jif > INT_MAX)
1287 			return 1;
1288 		*valp = (int)jif;
1289 	} else {
1290 		int val = *valp;
1291 		unsigned long lval;
1292 		if (val < 0) {
1293 			*negp = true;
1294 			lval = -(unsigned long)val;
1295 		} else {
1296 			*negp = false;
1297 			lval = (unsigned long)val;
1298 		}
1299 		*lvalp = jiffies_to_msecs(lval);
1300 	}
1301 	return 0;
1302 }
1303 
1304 /**
1305  * proc_dointvec_jiffies - read a vector of integers as seconds
1306  * @table: the sysctl table
1307  * @write: %TRUE if this is a write to the sysctl file
1308  * @buffer: the user buffer
1309  * @lenp: the size of the user buffer
1310  * @ppos: file position
1311  *
1312  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1313  * values from/to the user buffer, treated as an ASCII string.
1314  * The values read are assumed to be in seconds, and are converted into
1315  * jiffies.
1316  *
1317  * Returns 0 on success.
1318  */
1319 int proc_dointvec_jiffies(struct ctl_table *table, int write,
1320 			  void *buffer, size_t *lenp, loff_t *ppos)
1321 {
1322     return do_proc_dointvec(table,write,buffer,lenp,ppos,
1323 		    	    do_proc_dointvec_jiffies_conv,NULL);
1324 }
1325 
1326 /**
1327  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
1328  * @table: the sysctl table
1329  * @write: %TRUE if this is a write to the sysctl file
1330  * @buffer: the user buffer
1331  * @lenp: the size of the user buffer
1332  * @ppos: pointer to the file position
1333  *
1334  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1335  * values from/to the user buffer, treated as an ASCII string.
1336  * The values read are assumed to be in 1/USER_HZ seconds, and
1337  * are converted into jiffies.
1338  *
1339  * Returns 0 on success.
1340  */
1341 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1342 				 void *buffer, size_t *lenp, loff_t *ppos)
1343 {
1344     return do_proc_dointvec(table,write,buffer,lenp,ppos,
1345 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
1346 }
1347 
1348 /**
1349  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
1350  * @table: the sysctl table
1351  * @write: %TRUE if this is a write to the sysctl file
1352  * @buffer: the user buffer
1353  * @lenp: the size of the user buffer
1354  * @ppos: file position
1355  * @ppos: the current position in the file
1356  *
1357  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1358  * values from/to the user buffer, treated as an ASCII string.
1359  * The values read are assumed to be in 1/1000 seconds, and
1360  * are converted into jiffies.
1361  *
1362  * Returns 0 on success.
1363  */
1364 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, void *buffer,
1365 		size_t *lenp, loff_t *ppos)
1366 {
1367 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
1368 				do_proc_dointvec_ms_jiffies_conv, NULL);
1369 }
1370 
1371 static int proc_do_cad_pid(struct ctl_table *table, int write, void *buffer,
1372 		size_t *lenp, loff_t *ppos)
1373 {
1374 	struct pid *new_pid;
1375 	pid_t tmp;
1376 	int r;
1377 
1378 	tmp = pid_vnr(cad_pid);
1379 
1380 	r = __do_proc_dointvec(&tmp, table, write, buffer,
1381 			       lenp, ppos, NULL, NULL);
1382 	if (r || !write)
1383 		return r;
1384 
1385 	new_pid = find_get_pid(tmp);
1386 	if (!new_pid)
1387 		return -ESRCH;
1388 
1389 	put_pid(xchg(&cad_pid, new_pid));
1390 	return 0;
1391 }
1392 
1393 /**
1394  * proc_do_large_bitmap - read/write from/to a large bitmap
1395  * @table: the sysctl table
1396  * @write: %TRUE if this is a write to the sysctl file
1397  * @buffer: the user buffer
1398  * @lenp: the size of the user buffer
1399  * @ppos: file position
1400  *
1401  * The bitmap is stored at table->data and the bitmap length (in bits)
1402  * in table->maxlen.
1403  *
1404  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
1405  * large bitmaps may be represented in a compact manner. Writing into
1406  * the file will clear the bitmap then update it with the given input.
1407  *
1408  * Returns 0 on success.
1409  */
1410 int proc_do_large_bitmap(struct ctl_table *table, int write,
1411 			 void *buffer, size_t *lenp, loff_t *ppos)
1412 {
1413 	int err = 0;
1414 	size_t left = *lenp;
1415 	unsigned long bitmap_len = table->maxlen;
1416 	unsigned long *bitmap = *(unsigned long **) table->data;
1417 	unsigned long *tmp_bitmap = NULL;
1418 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
1419 
1420 	if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
1421 		*lenp = 0;
1422 		return 0;
1423 	}
1424 
1425 	if (write) {
1426 		char *p = buffer;
1427 		size_t skipped = 0;
1428 
1429 		if (left > PAGE_SIZE - 1) {
1430 			left = PAGE_SIZE - 1;
1431 			/* How much of the buffer we'll skip this pass */
1432 			skipped = *lenp - left;
1433 		}
1434 
1435 		tmp_bitmap = bitmap_zalloc(bitmap_len, GFP_KERNEL);
1436 		if (!tmp_bitmap)
1437 			return -ENOMEM;
1438 		proc_skip_char(&p, &left, '\n');
1439 		while (!err && left) {
1440 			unsigned long val_a, val_b;
1441 			bool neg;
1442 			size_t saved_left;
1443 
1444 			/* In case we stop parsing mid-number, we can reset */
1445 			saved_left = left;
1446 			err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
1447 					     sizeof(tr_a), &c);
1448 			/*
1449 			 * If we consumed the entirety of a truncated buffer or
1450 			 * only one char is left (may be a "-"), then stop here,
1451 			 * reset, & come back for more.
1452 			 */
1453 			if ((left <= 1) && skipped) {
1454 				left = saved_left;
1455 				break;
1456 			}
1457 
1458 			if (err)
1459 				break;
1460 			if (val_a >= bitmap_len || neg) {
1461 				err = -EINVAL;
1462 				break;
1463 			}
1464 
1465 			val_b = val_a;
1466 			if (left) {
1467 				p++;
1468 				left--;
1469 			}
1470 
1471 			if (c == '-') {
1472 				err = proc_get_long(&p, &left, &val_b,
1473 						     &neg, tr_b, sizeof(tr_b),
1474 						     &c);
1475 				/*
1476 				 * If we consumed all of a truncated buffer or
1477 				 * then stop here, reset, & come back for more.
1478 				 */
1479 				if (!left && skipped) {
1480 					left = saved_left;
1481 					break;
1482 				}
1483 
1484 				if (err)
1485 					break;
1486 				if (val_b >= bitmap_len || neg ||
1487 				    val_a > val_b) {
1488 					err = -EINVAL;
1489 					break;
1490 				}
1491 				if (left) {
1492 					p++;
1493 					left--;
1494 				}
1495 			}
1496 
1497 			bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
1498 			proc_skip_char(&p, &left, '\n');
1499 		}
1500 		left += skipped;
1501 	} else {
1502 		unsigned long bit_a, bit_b = 0;
1503 		bool first = 1;
1504 
1505 		while (left) {
1506 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
1507 			if (bit_a >= bitmap_len)
1508 				break;
1509 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
1510 						   bit_a + 1) - 1;
1511 
1512 			if (!first)
1513 				proc_put_char(&buffer, &left, ',');
1514 			proc_put_long(&buffer, &left, bit_a, false);
1515 			if (bit_a != bit_b) {
1516 				proc_put_char(&buffer, &left, '-');
1517 				proc_put_long(&buffer, &left, bit_b, false);
1518 			}
1519 
1520 			first = 0; bit_b++;
1521 		}
1522 		proc_put_char(&buffer, &left, '\n');
1523 	}
1524 
1525 	if (!err) {
1526 		if (write) {
1527 			if (*ppos)
1528 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
1529 			else
1530 				bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
1531 		}
1532 		*lenp -= left;
1533 		*ppos += *lenp;
1534 	}
1535 
1536 	bitmap_free(tmp_bitmap);
1537 	return err;
1538 }
1539 
1540 #else /* CONFIG_PROC_SYSCTL */
1541 
1542 int proc_dostring(struct ctl_table *table, int write,
1543 		  void *buffer, size_t *lenp, loff_t *ppos)
1544 {
1545 	return -ENOSYS;
1546 }
1547 
1548 int proc_dobool(struct ctl_table *table, int write,
1549 		void *buffer, size_t *lenp, loff_t *ppos)
1550 {
1551 	return -ENOSYS;
1552 }
1553 
1554 int proc_dointvec(struct ctl_table *table, int write,
1555 		  void *buffer, size_t *lenp, loff_t *ppos)
1556 {
1557 	return -ENOSYS;
1558 }
1559 
1560 int proc_douintvec(struct ctl_table *table, int write,
1561 		  void *buffer, size_t *lenp, loff_t *ppos)
1562 {
1563 	return -ENOSYS;
1564 }
1565 
1566 int proc_dointvec_minmax(struct ctl_table *table, int write,
1567 		    void *buffer, size_t *lenp, loff_t *ppos)
1568 {
1569 	return -ENOSYS;
1570 }
1571 
1572 int proc_douintvec_minmax(struct ctl_table *table, int write,
1573 			  void *buffer, size_t *lenp, loff_t *ppos)
1574 {
1575 	return -ENOSYS;
1576 }
1577 
1578 int proc_dou8vec_minmax(struct ctl_table *table, int write,
1579 			void *buffer, size_t *lenp, loff_t *ppos)
1580 {
1581 	return -ENOSYS;
1582 }
1583 
1584 int proc_dointvec_jiffies(struct ctl_table *table, int write,
1585 		    void *buffer, size_t *lenp, loff_t *ppos)
1586 {
1587 	return -ENOSYS;
1588 }
1589 
1590 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1591 		    void *buffer, size_t *lenp, loff_t *ppos)
1592 {
1593 	return -ENOSYS;
1594 }
1595 
1596 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
1597 			     void *buffer, size_t *lenp, loff_t *ppos)
1598 {
1599 	return -ENOSYS;
1600 }
1601 
1602 int proc_doulongvec_minmax(struct ctl_table *table, int write,
1603 		    void *buffer, size_t *lenp, loff_t *ppos)
1604 {
1605 	return -ENOSYS;
1606 }
1607 
1608 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1609 				      void *buffer, size_t *lenp, loff_t *ppos)
1610 {
1611 	return -ENOSYS;
1612 }
1613 
1614 int proc_do_large_bitmap(struct ctl_table *table, int write,
1615 			 void *buffer, size_t *lenp, loff_t *ppos)
1616 {
1617 	return -ENOSYS;
1618 }
1619 
1620 #endif /* CONFIG_PROC_SYSCTL */
1621 
1622 #if defined(CONFIG_SYSCTL)
1623 int proc_do_static_key(struct ctl_table *table, int write,
1624 		       void *buffer, size_t *lenp, loff_t *ppos)
1625 {
1626 	struct static_key *key = (struct static_key *)table->data;
1627 	static DEFINE_MUTEX(static_key_mutex);
1628 	int val, ret;
1629 	struct ctl_table tmp = {
1630 		.data   = &val,
1631 		.maxlen = sizeof(val),
1632 		.mode   = table->mode,
1633 		.extra1 = SYSCTL_ZERO,
1634 		.extra2 = SYSCTL_ONE,
1635 	};
1636 
1637 	if (write && !capable(CAP_SYS_ADMIN))
1638 		return -EPERM;
1639 
1640 	mutex_lock(&static_key_mutex);
1641 	val = static_key_enabled(key);
1642 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
1643 	if (write && !ret) {
1644 		if (val)
1645 			static_key_enable(key);
1646 		else
1647 			static_key_disable(key);
1648 	}
1649 	mutex_unlock(&static_key_mutex);
1650 	return ret;
1651 }
1652 
1653 static struct ctl_table kern_table[] = {
1654 	{
1655 		.procname	= "sched_child_runs_first",
1656 		.data		= &sysctl_sched_child_runs_first,
1657 		.maxlen		= sizeof(unsigned int),
1658 		.mode		= 0644,
1659 		.proc_handler	= proc_dointvec,
1660 	},
1661 #ifdef CONFIG_SCHEDSTATS
1662 	{
1663 		.procname	= "sched_schedstats",
1664 		.data		= NULL,
1665 		.maxlen		= sizeof(unsigned int),
1666 		.mode		= 0644,
1667 		.proc_handler	= sysctl_schedstats,
1668 		.extra1		= SYSCTL_ZERO,
1669 		.extra2		= SYSCTL_ONE,
1670 	},
1671 #endif /* CONFIG_SCHEDSTATS */
1672 #ifdef CONFIG_TASK_DELAY_ACCT
1673 	{
1674 		.procname	= "task_delayacct",
1675 		.data		= NULL,
1676 		.maxlen		= sizeof(unsigned int),
1677 		.mode		= 0644,
1678 		.proc_handler	= sysctl_delayacct,
1679 		.extra1		= SYSCTL_ZERO,
1680 		.extra2		= SYSCTL_ONE,
1681 	},
1682 #endif /* CONFIG_TASK_DELAY_ACCT */
1683 #ifdef CONFIG_NUMA_BALANCING
1684 	{
1685 		.procname	= "numa_balancing",
1686 		.data		= NULL, /* filled in by handler */
1687 		.maxlen		= sizeof(unsigned int),
1688 		.mode		= 0644,
1689 		.proc_handler	= sysctl_numa_balancing,
1690 		.extra1		= SYSCTL_ZERO,
1691 		.extra2		= SYSCTL_ONE,
1692 	},
1693 #endif /* CONFIG_NUMA_BALANCING */
1694 	{
1695 		.procname	= "sched_rt_period_us",
1696 		.data		= &sysctl_sched_rt_period,
1697 		.maxlen		= sizeof(unsigned int),
1698 		.mode		= 0644,
1699 		.proc_handler	= sched_rt_handler,
1700 	},
1701 	{
1702 		.procname	= "sched_rt_runtime_us",
1703 		.data		= &sysctl_sched_rt_runtime,
1704 		.maxlen		= sizeof(int),
1705 		.mode		= 0644,
1706 		.proc_handler	= sched_rt_handler,
1707 	},
1708 	{
1709 		.procname	= "sched_deadline_period_max_us",
1710 		.data		= &sysctl_sched_dl_period_max,
1711 		.maxlen		= sizeof(unsigned int),
1712 		.mode		= 0644,
1713 		.proc_handler	= proc_dointvec,
1714 	},
1715 	{
1716 		.procname	= "sched_deadline_period_min_us",
1717 		.data		= &sysctl_sched_dl_period_min,
1718 		.maxlen		= sizeof(unsigned int),
1719 		.mode		= 0644,
1720 		.proc_handler	= proc_dointvec,
1721 	},
1722 	{
1723 		.procname	= "sched_rr_timeslice_ms",
1724 		.data		= &sysctl_sched_rr_timeslice,
1725 		.maxlen		= sizeof(int),
1726 		.mode		= 0644,
1727 		.proc_handler	= sched_rr_handler,
1728 	},
1729 #ifdef CONFIG_UCLAMP_TASK
1730 	{
1731 		.procname	= "sched_util_clamp_min",
1732 		.data		= &sysctl_sched_uclamp_util_min,
1733 		.maxlen		= sizeof(unsigned int),
1734 		.mode		= 0644,
1735 		.proc_handler	= sysctl_sched_uclamp_handler,
1736 	},
1737 	{
1738 		.procname	= "sched_util_clamp_max",
1739 		.data		= &sysctl_sched_uclamp_util_max,
1740 		.maxlen		= sizeof(unsigned int),
1741 		.mode		= 0644,
1742 		.proc_handler	= sysctl_sched_uclamp_handler,
1743 	},
1744 	{
1745 		.procname	= "sched_util_clamp_min_rt_default",
1746 		.data		= &sysctl_sched_uclamp_util_min_rt_default,
1747 		.maxlen		= sizeof(unsigned int),
1748 		.mode		= 0644,
1749 		.proc_handler	= sysctl_sched_uclamp_handler,
1750 	},
1751 #endif
1752 #ifdef CONFIG_SCHED_AUTOGROUP
1753 	{
1754 		.procname	= "sched_autogroup_enabled",
1755 		.data		= &sysctl_sched_autogroup_enabled,
1756 		.maxlen		= sizeof(unsigned int),
1757 		.mode		= 0644,
1758 		.proc_handler	= proc_dointvec_minmax,
1759 		.extra1		= SYSCTL_ZERO,
1760 		.extra2		= SYSCTL_ONE,
1761 	},
1762 #endif
1763 #ifdef CONFIG_CFS_BANDWIDTH
1764 	{
1765 		.procname	= "sched_cfs_bandwidth_slice_us",
1766 		.data		= &sysctl_sched_cfs_bandwidth_slice,
1767 		.maxlen		= sizeof(unsigned int),
1768 		.mode		= 0644,
1769 		.proc_handler	= proc_dointvec_minmax,
1770 		.extra1		= SYSCTL_ONE,
1771 	},
1772 #endif
1773 #if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_CPU_FREQ_GOV_SCHEDUTIL)
1774 	{
1775 		.procname	= "sched_energy_aware",
1776 		.data		= &sysctl_sched_energy_aware,
1777 		.maxlen		= sizeof(unsigned int),
1778 		.mode		= 0644,
1779 		.proc_handler	= sched_energy_aware_handler,
1780 		.extra1		= SYSCTL_ZERO,
1781 		.extra2		= SYSCTL_ONE,
1782 	},
1783 #endif
1784 #ifdef CONFIG_PROVE_LOCKING
1785 	{
1786 		.procname	= "prove_locking",
1787 		.data		= &prove_locking,
1788 		.maxlen		= sizeof(int),
1789 		.mode		= 0644,
1790 		.proc_handler	= proc_dointvec,
1791 	},
1792 #endif
1793 #ifdef CONFIG_LOCK_STAT
1794 	{
1795 		.procname	= "lock_stat",
1796 		.data		= &lock_stat,
1797 		.maxlen		= sizeof(int),
1798 		.mode		= 0644,
1799 		.proc_handler	= proc_dointvec,
1800 	},
1801 #endif
1802 	{
1803 		.procname	= "panic",
1804 		.data		= &panic_timeout,
1805 		.maxlen		= sizeof(int),
1806 		.mode		= 0644,
1807 		.proc_handler	= proc_dointvec,
1808 	},
1809 #ifdef CONFIG_PROC_SYSCTL
1810 	{
1811 		.procname	= "tainted",
1812 		.maxlen 	= sizeof(long),
1813 		.mode		= 0644,
1814 		.proc_handler	= proc_taint,
1815 	},
1816 	{
1817 		.procname	= "sysctl_writes_strict",
1818 		.data		= &sysctl_writes_strict,
1819 		.maxlen		= sizeof(int),
1820 		.mode		= 0644,
1821 		.proc_handler	= proc_dointvec_minmax,
1822 		.extra1		= SYSCTL_NEG_ONE,
1823 		.extra2		= SYSCTL_ONE,
1824 	},
1825 #endif
1826 #ifdef CONFIG_LATENCYTOP
1827 	{
1828 		.procname	= "latencytop",
1829 		.data		= &latencytop_enabled,
1830 		.maxlen		= sizeof(int),
1831 		.mode		= 0644,
1832 		.proc_handler	= sysctl_latencytop,
1833 	},
1834 #endif
1835 #ifdef CONFIG_BLK_DEV_INITRD
1836 	{
1837 		.procname	= "real-root-dev",
1838 		.data		= &real_root_dev,
1839 		.maxlen		= sizeof(int),
1840 		.mode		= 0644,
1841 		.proc_handler	= proc_dointvec,
1842 	},
1843 #endif
1844 	{
1845 		.procname	= "print-fatal-signals",
1846 		.data		= &print_fatal_signals,
1847 		.maxlen		= sizeof(int),
1848 		.mode		= 0644,
1849 		.proc_handler	= proc_dointvec,
1850 	},
1851 #ifdef CONFIG_SPARC
1852 	{
1853 		.procname	= "reboot-cmd",
1854 		.data		= reboot_command,
1855 		.maxlen		= 256,
1856 		.mode		= 0644,
1857 		.proc_handler	= proc_dostring,
1858 	},
1859 	{
1860 		.procname	= "stop-a",
1861 		.data		= &stop_a_enabled,
1862 		.maxlen		= sizeof (int),
1863 		.mode		= 0644,
1864 		.proc_handler	= proc_dointvec,
1865 	},
1866 	{
1867 		.procname	= "scons-poweroff",
1868 		.data		= &scons_pwroff,
1869 		.maxlen		= sizeof (int),
1870 		.mode		= 0644,
1871 		.proc_handler	= proc_dointvec,
1872 	},
1873 #endif
1874 #ifdef CONFIG_SPARC64
1875 	{
1876 		.procname	= "tsb-ratio",
1877 		.data		= &sysctl_tsb_ratio,
1878 		.maxlen		= sizeof (int),
1879 		.mode		= 0644,
1880 		.proc_handler	= proc_dointvec,
1881 	},
1882 #endif
1883 #ifdef CONFIG_PARISC
1884 	{
1885 		.procname	= "soft-power",
1886 		.data		= &pwrsw_enabled,
1887 		.maxlen		= sizeof (int),
1888 		.mode		= 0644,
1889 		.proc_handler	= proc_dointvec,
1890 	},
1891 #endif
1892 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
1893 	{
1894 		.procname	= "unaligned-trap",
1895 		.data		= &unaligned_enabled,
1896 		.maxlen		= sizeof (int),
1897 		.mode		= 0644,
1898 		.proc_handler	= proc_dointvec,
1899 	},
1900 #endif
1901 	{
1902 		.procname	= "ctrl-alt-del",
1903 		.data		= &C_A_D,
1904 		.maxlen		= sizeof(int),
1905 		.mode		= 0644,
1906 		.proc_handler	= proc_dointvec,
1907 	},
1908 #ifdef CONFIG_FUNCTION_TRACER
1909 	{
1910 		.procname	= "ftrace_enabled",
1911 		.data		= &ftrace_enabled,
1912 		.maxlen		= sizeof(int),
1913 		.mode		= 0644,
1914 		.proc_handler	= ftrace_enable_sysctl,
1915 	},
1916 #endif
1917 #ifdef CONFIG_STACK_TRACER
1918 	{
1919 		.procname	= "stack_tracer_enabled",
1920 		.data		= &stack_tracer_enabled,
1921 		.maxlen		= sizeof(int),
1922 		.mode		= 0644,
1923 		.proc_handler	= stack_trace_sysctl,
1924 	},
1925 #endif
1926 #ifdef CONFIG_TRACING
1927 	{
1928 		.procname	= "ftrace_dump_on_oops",
1929 		.data		= &ftrace_dump_on_oops,
1930 		.maxlen		= sizeof(int),
1931 		.mode		= 0644,
1932 		.proc_handler	= proc_dointvec,
1933 	},
1934 	{
1935 		.procname	= "traceoff_on_warning",
1936 		.data		= &__disable_trace_on_warning,
1937 		.maxlen		= sizeof(__disable_trace_on_warning),
1938 		.mode		= 0644,
1939 		.proc_handler	= proc_dointvec,
1940 	},
1941 	{
1942 		.procname	= "tracepoint_printk",
1943 		.data		= &tracepoint_printk,
1944 		.maxlen		= sizeof(tracepoint_printk),
1945 		.mode		= 0644,
1946 		.proc_handler	= tracepoint_printk_sysctl,
1947 	},
1948 #endif
1949 #ifdef CONFIG_KEXEC_CORE
1950 	{
1951 		.procname	= "kexec_load_disabled",
1952 		.data		= &kexec_load_disabled,
1953 		.maxlen		= sizeof(int),
1954 		.mode		= 0644,
1955 		/* only handle a transition from default "0" to "1" */
1956 		.proc_handler	= proc_dointvec_minmax,
1957 		.extra1		= SYSCTL_ONE,
1958 		.extra2		= SYSCTL_ONE,
1959 	},
1960 #endif
1961 #ifdef CONFIG_MODULES
1962 	{
1963 		.procname	= "modprobe",
1964 		.data		= &modprobe_path,
1965 		.maxlen		= KMOD_PATH_LEN,
1966 		.mode		= 0644,
1967 		.proc_handler	= proc_dostring,
1968 	},
1969 	{
1970 		.procname	= "modules_disabled",
1971 		.data		= &modules_disabled,
1972 		.maxlen		= sizeof(int),
1973 		.mode		= 0644,
1974 		/* only handle a transition from default "0" to "1" */
1975 		.proc_handler	= proc_dointvec_minmax,
1976 		.extra1		= SYSCTL_ONE,
1977 		.extra2		= SYSCTL_ONE,
1978 	},
1979 #endif
1980 #ifdef CONFIG_UEVENT_HELPER
1981 	{
1982 		.procname	= "hotplug",
1983 		.data		= &uevent_helper,
1984 		.maxlen		= UEVENT_HELPER_PATH_LEN,
1985 		.mode		= 0644,
1986 		.proc_handler	= proc_dostring,
1987 	},
1988 #endif
1989 #ifdef CONFIG_BSD_PROCESS_ACCT
1990 	{
1991 		.procname	= "acct",
1992 		.data		= &acct_parm,
1993 		.maxlen		= 3*sizeof(int),
1994 		.mode		= 0644,
1995 		.proc_handler	= proc_dointvec,
1996 	},
1997 #endif
1998 #ifdef CONFIG_MAGIC_SYSRQ
1999 	{
2000 		.procname	= "sysrq",
2001 		.data		= NULL,
2002 		.maxlen		= sizeof (int),
2003 		.mode		= 0644,
2004 		.proc_handler	= sysrq_sysctl_handler,
2005 	},
2006 #endif
2007 #ifdef CONFIG_PROC_SYSCTL
2008 	{
2009 		.procname	= "cad_pid",
2010 		.data		= NULL,
2011 		.maxlen		= sizeof (int),
2012 		.mode		= 0600,
2013 		.proc_handler	= proc_do_cad_pid,
2014 	},
2015 #endif
2016 	{
2017 		.procname	= "threads-max",
2018 		.data		= NULL,
2019 		.maxlen		= sizeof(int),
2020 		.mode		= 0644,
2021 		.proc_handler	= sysctl_max_threads,
2022 	},
2023 	{
2024 		.procname	= "usermodehelper",
2025 		.mode		= 0555,
2026 		.child		= usermodehelper_table,
2027 	},
2028 	{
2029 		.procname	= "overflowuid",
2030 		.data		= &overflowuid,
2031 		.maxlen		= sizeof(int),
2032 		.mode		= 0644,
2033 		.proc_handler	= proc_dointvec_minmax,
2034 		.extra1		= SYSCTL_ZERO,
2035 		.extra2		= SYSCTL_MAXOLDUID,
2036 	},
2037 	{
2038 		.procname	= "overflowgid",
2039 		.data		= &overflowgid,
2040 		.maxlen		= sizeof(int),
2041 		.mode		= 0644,
2042 		.proc_handler	= proc_dointvec_minmax,
2043 		.extra1		= SYSCTL_ZERO,
2044 		.extra2		= SYSCTL_MAXOLDUID,
2045 	},
2046 #ifdef CONFIG_S390
2047 	{
2048 		.procname	= "userprocess_debug",
2049 		.data		= &show_unhandled_signals,
2050 		.maxlen		= sizeof(int),
2051 		.mode		= 0644,
2052 		.proc_handler	= proc_dointvec,
2053 	},
2054 #endif
2055 #ifdef CONFIG_SMP
2056 	{
2057 		.procname	= "oops_all_cpu_backtrace",
2058 		.data		= &sysctl_oops_all_cpu_backtrace,
2059 		.maxlen		= sizeof(int),
2060 		.mode		= 0644,
2061 		.proc_handler	= proc_dointvec_minmax,
2062 		.extra1		= SYSCTL_ZERO,
2063 		.extra2		= SYSCTL_ONE,
2064 	},
2065 #endif /* CONFIG_SMP */
2066 	{
2067 		.procname	= "pid_max",
2068 		.data		= &pid_max,
2069 		.maxlen		= sizeof (int),
2070 		.mode		= 0644,
2071 		.proc_handler	= proc_dointvec_minmax,
2072 		.extra1		= &pid_max_min,
2073 		.extra2		= &pid_max_max,
2074 	},
2075 	{
2076 		.procname	= "panic_on_oops",
2077 		.data		= &panic_on_oops,
2078 		.maxlen		= sizeof(int),
2079 		.mode		= 0644,
2080 		.proc_handler	= proc_dointvec,
2081 	},
2082 	{
2083 		.procname	= "panic_print",
2084 		.data		= &panic_print,
2085 		.maxlen		= sizeof(unsigned long),
2086 		.mode		= 0644,
2087 		.proc_handler	= proc_doulongvec_minmax,
2088 	},
2089 	{
2090 		.procname	= "ngroups_max",
2091 		.data		= (void *)&ngroups_max,
2092 		.maxlen		= sizeof (int),
2093 		.mode		= 0444,
2094 		.proc_handler	= proc_dointvec,
2095 	},
2096 	{
2097 		.procname	= "cap_last_cap",
2098 		.data		= (void *)&cap_last_cap,
2099 		.maxlen		= sizeof(int),
2100 		.mode		= 0444,
2101 		.proc_handler	= proc_dointvec,
2102 	},
2103 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
2104 	{
2105 		.procname       = "unknown_nmi_panic",
2106 		.data           = &unknown_nmi_panic,
2107 		.maxlen         = sizeof (int),
2108 		.mode           = 0644,
2109 		.proc_handler   = proc_dointvec,
2110 	},
2111 #endif
2112 
2113 #if (defined(CONFIG_X86_32) || defined(CONFIG_PARISC)) && \
2114 	defined(CONFIG_DEBUG_STACKOVERFLOW)
2115 	{
2116 		.procname	= "panic_on_stackoverflow",
2117 		.data		= &sysctl_panic_on_stackoverflow,
2118 		.maxlen		= sizeof(int),
2119 		.mode		= 0644,
2120 		.proc_handler	= proc_dointvec,
2121 	},
2122 #endif
2123 #if defined(CONFIG_X86)
2124 	{
2125 		.procname	= "panic_on_unrecovered_nmi",
2126 		.data		= &panic_on_unrecovered_nmi,
2127 		.maxlen		= sizeof(int),
2128 		.mode		= 0644,
2129 		.proc_handler	= proc_dointvec,
2130 	},
2131 	{
2132 		.procname	= "panic_on_io_nmi",
2133 		.data		= &panic_on_io_nmi,
2134 		.maxlen		= sizeof(int),
2135 		.mode		= 0644,
2136 		.proc_handler	= proc_dointvec,
2137 	},
2138 	{
2139 		.procname	= "bootloader_type",
2140 		.data		= &bootloader_type,
2141 		.maxlen		= sizeof (int),
2142 		.mode		= 0444,
2143 		.proc_handler	= proc_dointvec,
2144 	},
2145 	{
2146 		.procname	= "bootloader_version",
2147 		.data		= &bootloader_version,
2148 		.maxlen		= sizeof (int),
2149 		.mode		= 0444,
2150 		.proc_handler	= proc_dointvec,
2151 	},
2152 	{
2153 		.procname	= "io_delay_type",
2154 		.data		= &io_delay_type,
2155 		.maxlen		= sizeof(int),
2156 		.mode		= 0644,
2157 		.proc_handler	= proc_dointvec,
2158 	},
2159 #endif
2160 #if defined(CONFIG_MMU)
2161 	{
2162 		.procname	= "randomize_va_space",
2163 		.data		= &randomize_va_space,
2164 		.maxlen		= sizeof(int),
2165 		.mode		= 0644,
2166 		.proc_handler	= proc_dointvec,
2167 	},
2168 #endif
2169 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
2170 	{
2171 		.procname	= "spin_retry",
2172 		.data		= &spin_retry,
2173 		.maxlen		= sizeof (int),
2174 		.mode		= 0644,
2175 		.proc_handler	= proc_dointvec,
2176 	},
2177 #endif
2178 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
2179 	{
2180 		.procname	= "acpi_video_flags",
2181 		.data		= &acpi_realmode_flags,
2182 		.maxlen		= sizeof (unsigned long),
2183 		.mode		= 0644,
2184 		.proc_handler	= proc_doulongvec_minmax,
2185 	},
2186 #endif
2187 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
2188 	{
2189 		.procname	= "ignore-unaligned-usertrap",
2190 		.data		= &no_unaligned_warning,
2191 		.maxlen		= sizeof (int),
2192 		.mode		= 0644,
2193 		.proc_handler	= proc_dointvec,
2194 	},
2195 #endif
2196 #ifdef CONFIG_IA64
2197 	{
2198 		.procname	= "unaligned-dump-stack",
2199 		.data		= &unaligned_dump_stack,
2200 		.maxlen		= sizeof (int),
2201 		.mode		= 0644,
2202 		.proc_handler	= proc_dointvec,
2203 	},
2204 #endif
2205 #ifdef CONFIG_RT_MUTEXES
2206 	{
2207 		.procname	= "max_lock_depth",
2208 		.data		= &max_lock_depth,
2209 		.maxlen		= sizeof(int),
2210 		.mode		= 0644,
2211 		.proc_handler	= proc_dointvec,
2212 	},
2213 #endif
2214 	{
2215 		.procname	= "poweroff_cmd",
2216 		.data		= &poweroff_cmd,
2217 		.maxlen		= POWEROFF_CMD_PATH_LEN,
2218 		.mode		= 0644,
2219 		.proc_handler	= proc_dostring,
2220 	},
2221 #ifdef CONFIG_KEYS
2222 	{
2223 		.procname	= "keys",
2224 		.mode		= 0555,
2225 		.child		= key_sysctls,
2226 	},
2227 #endif
2228 #ifdef CONFIG_PERF_EVENTS
2229 	/*
2230 	 * User-space scripts rely on the existence of this file
2231 	 * as a feature check for perf_events being enabled.
2232 	 *
2233 	 * So it's an ABI, do not remove!
2234 	 */
2235 	{
2236 		.procname	= "perf_event_paranoid",
2237 		.data		= &sysctl_perf_event_paranoid,
2238 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
2239 		.mode		= 0644,
2240 		.proc_handler	= proc_dointvec,
2241 	},
2242 	{
2243 		.procname	= "perf_event_mlock_kb",
2244 		.data		= &sysctl_perf_event_mlock,
2245 		.maxlen		= sizeof(sysctl_perf_event_mlock),
2246 		.mode		= 0644,
2247 		.proc_handler	= proc_dointvec,
2248 	},
2249 	{
2250 		.procname	= "perf_event_max_sample_rate",
2251 		.data		= &sysctl_perf_event_sample_rate,
2252 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
2253 		.mode		= 0644,
2254 		.proc_handler	= perf_proc_update_handler,
2255 		.extra1		= SYSCTL_ONE,
2256 	},
2257 	{
2258 		.procname	= "perf_cpu_time_max_percent",
2259 		.data		= &sysctl_perf_cpu_time_max_percent,
2260 		.maxlen		= sizeof(sysctl_perf_cpu_time_max_percent),
2261 		.mode		= 0644,
2262 		.proc_handler	= perf_cpu_time_max_percent_handler,
2263 		.extra1		= SYSCTL_ZERO,
2264 		.extra2		= SYSCTL_ONE_HUNDRED,
2265 	},
2266 	{
2267 		.procname	= "perf_event_max_stack",
2268 		.data		= &sysctl_perf_event_max_stack,
2269 		.maxlen		= sizeof(sysctl_perf_event_max_stack),
2270 		.mode		= 0644,
2271 		.proc_handler	= perf_event_max_stack_handler,
2272 		.extra1		= SYSCTL_ZERO,
2273 		.extra2		= (void *)&six_hundred_forty_kb,
2274 	},
2275 	{
2276 		.procname	= "perf_event_max_contexts_per_stack",
2277 		.data		= &sysctl_perf_event_max_contexts_per_stack,
2278 		.maxlen		= sizeof(sysctl_perf_event_max_contexts_per_stack),
2279 		.mode		= 0644,
2280 		.proc_handler	= perf_event_max_stack_handler,
2281 		.extra1		= SYSCTL_ZERO,
2282 		.extra2		= SYSCTL_ONE_THOUSAND,
2283 	},
2284 #endif
2285 	{
2286 		.procname	= "panic_on_warn",
2287 		.data		= &panic_on_warn,
2288 		.maxlen		= sizeof(int),
2289 		.mode		= 0644,
2290 		.proc_handler	= proc_dointvec_minmax,
2291 		.extra1		= SYSCTL_ZERO,
2292 		.extra2		= SYSCTL_ONE,
2293 	},
2294 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
2295 	{
2296 		.procname	= "timer_migration",
2297 		.data		= &sysctl_timer_migration,
2298 		.maxlen		= sizeof(unsigned int),
2299 		.mode		= 0644,
2300 		.proc_handler	= timer_migration_handler,
2301 		.extra1		= SYSCTL_ZERO,
2302 		.extra2		= SYSCTL_ONE,
2303 	},
2304 #endif
2305 #ifdef CONFIG_BPF_SYSCALL
2306 	{
2307 		.procname	= "unprivileged_bpf_disabled",
2308 		.data		= &sysctl_unprivileged_bpf_disabled,
2309 		.maxlen		= sizeof(sysctl_unprivileged_bpf_disabled),
2310 		.mode		= 0644,
2311 		.proc_handler	= bpf_unpriv_handler,
2312 		.extra1		= SYSCTL_ZERO,
2313 		.extra2		= SYSCTL_TWO,
2314 	},
2315 	{
2316 		.procname	= "bpf_stats_enabled",
2317 		.data		= &bpf_stats_enabled_key.key,
2318 		.maxlen		= sizeof(bpf_stats_enabled_key),
2319 		.mode		= 0644,
2320 		.proc_handler	= bpf_stats_handler,
2321 	},
2322 #endif
2323 #if defined(CONFIG_TREE_RCU)
2324 	{
2325 		.procname	= "panic_on_rcu_stall",
2326 		.data		= &sysctl_panic_on_rcu_stall,
2327 		.maxlen		= sizeof(sysctl_panic_on_rcu_stall),
2328 		.mode		= 0644,
2329 		.proc_handler	= proc_dointvec_minmax,
2330 		.extra1		= SYSCTL_ZERO,
2331 		.extra2		= SYSCTL_ONE,
2332 	},
2333 #endif
2334 #if defined(CONFIG_TREE_RCU)
2335 	{
2336 		.procname	= "max_rcu_stall_to_panic",
2337 		.data		= &sysctl_max_rcu_stall_to_panic,
2338 		.maxlen		= sizeof(sysctl_max_rcu_stall_to_panic),
2339 		.mode		= 0644,
2340 		.proc_handler	= proc_dointvec_minmax,
2341 		.extra1		= SYSCTL_ONE,
2342 		.extra2		= SYSCTL_INT_MAX,
2343 	},
2344 #endif
2345 	{ }
2346 };
2347 
2348 static struct ctl_table vm_table[] = {
2349 	{
2350 		.procname	= "overcommit_memory",
2351 		.data		= &sysctl_overcommit_memory,
2352 		.maxlen		= sizeof(sysctl_overcommit_memory),
2353 		.mode		= 0644,
2354 		.proc_handler	= overcommit_policy_handler,
2355 		.extra1		= SYSCTL_ZERO,
2356 		.extra2		= SYSCTL_TWO,
2357 	},
2358 	{
2359 		.procname	= "panic_on_oom",
2360 		.data		= &sysctl_panic_on_oom,
2361 		.maxlen		= sizeof(sysctl_panic_on_oom),
2362 		.mode		= 0644,
2363 		.proc_handler	= proc_dointvec_minmax,
2364 		.extra1		= SYSCTL_ZERO,
2365 		.extra2		= SYSCTL_TWO,
2366 	},
2367 	{
2368 		.procname	= "oom_kill_allocating_task",
2369 		.data		= &sysctl_oom_kill_allocating_task,
2370 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
2371 		.mode		= 0644,
2372 		.proc_handler	= proc_dointvec,
2373 	},
2374 	{
2375 		.procname	= "oom_dump_tasks",
2376 		.data		= &sysctl_oom_dump_tasks,
2377 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
2378 		.mode		= 0644,
2379 		.proc_handler	= proc_dointvec,
2380 	},
2381 	{
2382 		.procname	= "overcommit_ratio",
2383 		.data		= &sysctl_overcommit_ratio,
2384 		.maxlen		= sizeof(sysctl_overcommit_ratio),
2385 		.mode		= 0644,
2386 		.proc_handler	= overcommit_ratio_handler,
2387 	},
2388 	{
2389 		.procname	= "overcommit_kbytes",
2390 		.data		= &sysctl_overcommit_kbytes,
2391 		.maxlen		= sizeof(sysctl_overcommit_kbytes),
2392 		.mode		= 0644,
2393 		.proc_handler	= overcommit_kbytes_handler,
2394 	},
2395 	{
2396 		.procname	= "page-cluster",
2397 		.data		= &page_cluster,
2398 		.maxlen		= sizeof(int),
2399 		.mode		= 0644,
2400 		.proc_handler	= proc_dointvec_minmax,
2401 		.extra1		= SYSCTL_ZERO,
2402 	},
2403 	{
2404 		.procname	= "dirty_background_ratio",
2405 		.data		= &dirty_background_ratio,
2406 		.maxlen		= sizeof(dirty_background_ratio),
2407 		.mode		= 0644,
2408 		.proc_handler	= dirty_background_ratio_handler,
2409 		.extra1		= SYSCTL_ZERO,
2410 		.extra2		= SYSCTL_ONE_HUNDRED,
2411 	},
2412 	{
2413 		.procname	= "dirty_background_bytes",
2414 		.data		= &dirty_background_bytes,
2415 		.maxlen		= sizeof(dirty_background_bytes),
2416 		.mode		= 0644,
2417 		.proc_handler	= dirty_background_bytes_handler,
2418 		.extra1		= SYSCTL_LONG_ONE,
2419 	},
2420 	{
2421 		.procname	= "dirty_ratio",
2422 		.data		= &vm_dirty_ratio,
2423 		.maxlen		= sizeof(vm_dirty_ratio),
2424 		.mode		= 0644,
2425 		.proc_handler	= dirty_ratio_handler,
2426 		.extra1		= SYSCTL_ZERO,
2427 		.extra2		= SYSCTL_ONE_HUNDRED,
2428 	},
2429 	{
2430 		.procname	= "dirty_bytes",
2431 		.data		= &vm_dirty_bytes,
2432 		.maxlen		= sizeof(vm_dirty_bytes),
2433 		.mode		= 0644,
2434 		.proc_handler	= dirty_bytes_handler,
2435 		.extra1		= (void *)&dirty_bytes_min,
2436 	},
2437 	{
2438 		.procname	= "dirty_writeback_centisecs",
2439 		.data		= &dirty_writeback_interval,
2440 		.maxlen		= sizeof(dirty_writeback_interval),
2441 		.mode		= 0644,
2442 		.proc_handler	= dirty_writeback_centisecs_handler,
2443 	},
2444 	{
2445 		.procname	= "dirty_expire_centisecs",
2446 		.data		= &dirty_expire_interval,
2447 		.maxlen		= sizeof(dirty_expire_interval),
2448 		.mode		= 0644,
2449 		.proc_handler	= proc_dointvec_minmax,
2450 		.extra1		= SYSCTL_ZERO,
2451 	},
2452 	{
2453 		.procname	= "dirtytime_expire_seconds",
2454 		.data		= &dirtytime_expire_interval,
2455 		.maxlen		= sizeof(dirtytime_expire_interval),
2456 		.mode		= 0644,
2457 		.proc_handler	= dirtytime_interval_handler,
2458 		.extra1		= SYSCTL_ZERO,
2459 	},
2460 	{
2461 		.procname	= "swappiness",
2462 		.data		= &vm_swappiness,
2463 		.maxlen		= sizeof(vm_swappiness),
2464 		.mode		= 0644,
2465 		.proc_handler	= proc_dointvec_minmax,
2466 		.extra1		= SYSCTL_ZERO,
2467 		.extra2		= SYSCTL_TWO_HUNDRED,
2468 	},
2469 #ifdef CONFIG_HUGETLB_PAGE
2470 	{
2471 		.procname	= "nr_hugepages",
2472 		.data		= NULL,
2473 		.maxlen		= sizeof(unsigned long),
2474 		.mode		= 0644,
2475 		.proc_handler	= hugetlb_sysctl_handler,
2476 	},
2477 #ifdef CONFIG_NUMA
2478 	{
2479 		.procname       = "nr_hugepages_mempolicy",
2480 		.data           = NULL,
2481 		.maxlen         = sizeof(unsigned long),
2482 		.mode           = 0644,
2483 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
2484 	},
2485 	{
2486 		.procname		= "numa_stat",
2487 		.data			= &sysctl_vm_numa_stat,
2488 		.maxlen			= sizeof(int),
2489 		.mode			= 0644,
2490 		.proc_handler	= sysctl_vm_numa_stat_handler,
2491 		.extra1			= SYSCTL_ZERO,
2492 		.extra2			= SYSCTL_ONE,
2493 	},
2494 #endif
2495 	 {
2496 		.procname	= "hugetlb_shm_group",
2497 		.data		= &sysctl_hugetlb_shm_group,
2498 		.maxlen		= sizeof(gid_t),
2499 		.mode		= 0644,
2500 		.proc_handler	= proc_dointvec,
2501 	 },
2502 	{
2503 		.procname	= "nr_overcommit_hugepages",
2504 		.data		= NULL,
2505 		.maxlen		= sizeof(unsigned long),
2506 		.mode		= 0644,
2507 		.proc_handler	= hugetlb_overcommit_handler,
2508 	},
2509 #endif
2510 	{
2511 		.procname	= "lowmem_reserve_ratio",
2512 		.data		= &sysctl_lowmem_reserve_ratio,
2513 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
2514 		.mode		= 0644,
2515 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
2516 	},
2517 	{
2518 		.procname	= "drop_caches",
2519 		.data		= &sysctl_drop_caches,
2520 		.maxlen		= sizeof(int),
2521 		.mode		= 0200,
2522 		.proc_handler	= drop_caches_sysctl_handler,
2523 		.extra1		= SYSCTL_ONE,
2524 		.extra2		= SYSCTL_FOUR,
2525 	},
2526 #ifdef CONFIG_COMPACTION
2527 	{
2528 		.procname	= "compact_memory",
2529 		.data		= NULL,
2530 		.maxlen		= sizeof(int),
2531 		.mode		= 0200,
2532 		.proc_handler	= sysctl_compaction_handler,
2533 	},
2534 	{
2535 		.procname	= "compaction_proactiveness",
2536 		.data		= &sysctl_compaction_proactiveness,
2537 		.maxlen		= sizeof(sysctl_compaction_proactiveness),
2538 		.mode		= 0644,
2539 		.proc_handler	= compaction_proactiveness_sysctl_handler,
2540 		.extra1		= SYSCTL_ZERO,
2541 		.extra2		= SYSCTL_ONE_HUNDRED,
2542 	},
2543 	{
2544 		.procname	= "extfrag_threshold",
2545 		.data		= &sysctl_extfrag_threshold,
2546 		.maxlen		= sizeof(int),
2547 		.mode		= 0644,
2548 		.proc_handler	= proc_dointvec_minmax,
2549 		.extra1		= SYSCTL_ZERO,
2550 		.extra2		= (void *)&max_extfrag_threshold,
2551 	},
2552 	{
2553 		.procname	= "compact_unevictable_allowed",
2554 		.data		= &sysctl_compact_unevictable_allowed,
2555 		.maxlen		= sizeof(int),
2556 		.mode		= 0644,
2557 		.proc_handler	= proc_dointvec_minmax_warn_RT_change,
2558 		.extra1		= SYSCTL_ZERO,
2559 		.extra2		= SYSCTL_ONE,
2560 	},
2561 
2562 #endif /* CONFIG_COMPACTION */
2563 	{
2564 		.procname	= "min_free_kbytes",
2565 		.data		= &min_free_kbytes,
2566 		.maxlen		= sizeof(min_free_kbytes),
2567 		.mode		= 0644,
2568 		.proc_handler	= min_free_kbytes_sysctl_handler,
2569 		.extra1		= SYSCTL_ZERO,
2570 	},
2571 	{
2572 		.procname	= "watermark_boost_factor",
2573 		.data		= &watermark_boost_factor,
2574 		.maxlen		= sizeof(watermark_boost_factor),
2575 		.mode		= 0644,
2576 		.proc_handler	= proc_dointvec_minmax,
2577 		.extra1		= SYSCTL_ZERO,
2578 	},
2579 	{
2580 		.procname	= "watermark_scale_factor",
2581 		.data		= &watermark_scale_factor,
2582 		.maxlen		= sizeof(watermark_scale_factor),
2583 		.mode		= 0644,
2584 		.proc_handler	= watermark_scale_factor_sysctl_handler,
2585 		.extra1		= SYSCTL_ONE,
2586 		.extra2		= SYSCTL_THREE_THOUSAND,
2587 	},
2588 	{
2589 		.procname	= "percpu_pagelist_high_fraction",
2590 		.data		= &percpu_pagelist_high_fraction,
2591 		.maxlen		= sizeof(percpu_pagelist_high_fraction),
2592 		.mode		= 0644,
2593 		.proc_handler	= percpu_pagelist_high_fraction_sysctl_handler,
2594 		.extra1		= SYSCTL_ZERO,
2595 	},
2596 	{
2597 		.procname	= "page_lock_unfairness",
2598 		.data		= &sysctl_page_lock_unfairness,
2599 		.maxlen		= sizeof(sysctl_page_lock_unfairness),
2600 		.mode		= 0644,
2601 		.proc_handler	= proc_dointvec_minmax,
2602 		.extra1		= SYSCTL_ZERO,
2603 	},
2604 #ifdef CONFIG_MMU
2605 	{
2606 		.procname	= "max_map_count",
2607 		.data		= &sysctl_max_map_count,
2608 		.maxlen		= sizeof(sysctl_max_map_count),
2609 		.mode		= 0644,
2610 		.proc_handler	= proc_dointvec_minmax,
2611 		.extra1		= SYSCTL_ZERO,
2612 	},
2613 #else
2614 	{
2615 		.procname	= "nr_trim_pages",
2616 		.data		= &sysctl_nr_trim_pages,
2617 		.maxlen		= sizeof(sysctl_nr_trim_pages),
2618 		.mode		= 0644,
2619 		.proc_handler	= proc_dointvec_minmax,
2620 		.extra1		= SYSCTL_ZERO,
2621 	},
2622 #endif
2623 	{
2624 		.procname	= "laptop_mode",
2625 		.data		= &laptop_mode,
2626 		.maxlen		= sizeof(laptop_mode),
2627 		.mode		= 0644,
2628 		.proc_handler	= proc_dointvec_jiffies,
2629 	},
2630 	{
2631 		.procname	= "vfs_cache_pressure",
2632 		.data		= &sysctl_vfs_cache_pressure,
2633 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
2634 		.mode		= 0644,
2635 		.proc_handler	= proc_dointvec_minmax,
2636 		.extra1		= SYSCTL_ZERO,
2637 	},
2638 #if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
2639     defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
2640 	{
2641 		.procname	= "legacy_va_layout",
2642 		.data		= &sysctl_legacy_va_layout,
2643 		.maxlen		= sizeof(sysctl_legacy_va_layout),
2644 		.mode		= 0644,
2645 		.proc_handler	= proc_dointvec_minmax,
2646 		.extra1		= SYSCTL_ZERO,
2647 	},
2648 #endif
2649 #ifdef CONFIG_NUMA
2650 	{
2651 		.procname	= "zone_reclaim_mode",
2652 		.data		= &node_reclaim_mode,
2653 		.maxlen		= sizeof(node_reclaim_mode),
2654 		.mode		= 0644,
2655 		.proc_handler	= proc_dointvec_minmax,
2656 		.extra1		= SYSCTL_ZERO,
2657 	},
2658 	{
2659 		.procname	= "min_unmapped_ratio",
2660 		.data		= &sysctl_min_unmapped_ratio,
2661 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
2662 		.mode		= 0644,
2663 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
2664 		.extra1		= SYSCTL_ZERO,
2665 		.extra2		= SYSCTL_ONE_HUNDRED,
2666 	},
2667 	{
2668 		.procname	= "min_slab_ratio",
2669 		.data		= &sysctl_min_slab_ratio,
2670 		.maxlen		= sizeof(sysctl_min_slab_ratio),
2671 		.mode		= 0644,
2672 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
2673 		.extra1		= SYSCTL_ZERO,
2674 		.extra2		= SYSCTL_ONE_HUNDRED,
2675 	},
2676 #endif
2677 #ifdef CONFIG_SMP
2678 	{
2679 		.procname	= "stat_interval",
2680 		.data		= &sysctl_stat_interval,
2681 		.maxlen		= sizeof(sysctl_stat_interval),
2682 		.mode		= 0644,
2683 		.proc_handler	= proc_dointvec_jiffies,
2684 	},
2685 	{
2686 		.procname	= "stat_refresh",
2687 		.data		= NULL,
2688 		.maxlen		= 0,
2689 		.mode		= 0600,
2690 		.proc_handler	= vmstat_refresh,
2691 	},
2692 #endif
2693 #ifdef CONFIG_MMU
2694 	{
2695 		.procname	= "mmap_min_addr",
2696 		.data		= &dac_mmap_min_addr,
2697 		.maxlen		= sizeof(unsigned long),
2698 		.mode		= 0644,
2699 		.proc_handler	= mmap_min_addr_handler,
2700 	},
2701 #endif
2702 #ifdef CONFIG_NUMA
2703 	{
2704 		.procname	= "numa_zonelist_order",
2705 		.data		= &numa_zonelist_order,
2706 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
2707 		.mode		= 0644,
2708 		.proc_handler	= numa_zonelist_order_handler,
2709 	},
2710 #endif
2711 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
2712    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
2713 	{
2714 		.procname	= "vdso_enabled",
2715 #ifdef CONFIG_X86_32
2716 		.data		= &vdso32_enabled,
2717 		.maxlen		= sizeof(vdso32_enabled),
2718 #else
2719 		.data		= &vdso_enabled,
2720 		.maxlen		= sizeof(vdso_enabled),
2721 #endif
2722 		.mode		= 0644,
2723 		.proc_handler	= proc_dointvec,
2724 		.extra1		= SYSCTL_ZERO,
2725 	},
2726 #endif
2727 #ifdef CONFIG_HIGHMEM
2728 	{
2729 		.procname	= "highmem_is_dirtyable",
2730 		.data		= &vm_highmem_is_dirtyable,
2731 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
2732 		.mode		= 0644,
2733 		.proc_handler	= proc_dointvec_minmax,
2734 		.extra1		= SYSCTL_ZERO,
2735 		.extra2		= SYSCTL_ONE,
2736 	},
2737 #endif
2738 #ifdef CONFIG_MEMORY_FAILURE
2739 	{
2740 		.procname	= "memory_failure_early_kill",
2741 		.data		= &sysctl_memory_failure_early_kill,
2742 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
2743 		.mode		= 0644,
2744 		.proc_handler	= proc_dointvec_minmax,
2745 		.extra1		= SYSCTL_ZERO,
2746 		.extra2		= SYSCTL_ONE,
2747 	},
2748 	{
2749 		.procname	= "memory_failure_recovery",
2750 		.data		= &sysctl_memory_failure_recovery,
2751 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
2752 		.mode		= 0644,
2753 		.proc_handler	= proc_dointvec_minmax,
2754 		.extra1		= SYSCTL_ZERO,
2755 		.extra2		= SYSCTL_ONE,
2756 	},
2757 #endif
2758 	{
2759 		.procname	= "user_reserve_kbytes",
2760 		.data		= &sysctl_user_reserve_kbytes,
2761 		.maxlen		= sizeof(sysctl_user_reserve_kbytes),
2762 		.mode		= 0644,
2763 		.proc_handler	= proc_doulongvec_minmax,
2764 	},
2765 	{
2766 		.procname	= "admin_reserve_kbytes",
2767 		.data		= &sysctl_admin_reserve_kbytes,
2768 		.maxlen		= sizeof(sysctl_admin_reserve_kbytes),
2769 		.mode		= 0644,
2770 		.proc_handler	= proc_doulongvec_minmax,
2771 	},
2772 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
2773 	{
2774 		.procname	= "mmap_rnd_bits",
2775 		.data		= &mmap_rnd_bits,
2776 		.maxlen		= sizeof(mmap_rnd_bits),
2777 		.mode		= 0600,
2778 		.proc_handler	= proc_dointvec_minmax,
2779 		.extra1		= (void *)&mmap_rnd_bits_min,
2780 		.extra2		= (void *)&mmap_rnd_bits_max,
2781 	},
2782 #endif
2783 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
2784 	{
2785 		.procname	= "mmap_rnd_compat_bits",
2786 		.data		= &mmap_rnd_compat_bits,
2787 		.maxlen		= sizeof(mmap_rnd_compat_bits),
2788 		.mode		= 0600,
2789 		.proc_handler	= proc_dointvec_minmax,
2790 		.extra1		= (void *)&mmap_rnd_compat_bits_min,
2791 		.extra2		= (void *)&mmap_rnd_compat_bits_max,
2792 	},
2793 #endif
2794 #ifdef CONFIG_USERFAULTFD
2795 	{
2796 		.procname	= "unprivileged_userfaultfd",
2797 		.data		= &sysctl_unprivileged_userfaultfd,
2798 		.maxlen		= sizeof(sysctl_unprivileged_userfaultfd),
2799 		.mode		= 0644,
2800 		.proc_handler	= proc_dointvec_minmax,
2801 		.extra1		= SYSCTL_ZERO,
2802 		.extra2		= SYSCTL_ONE,
2803 	},
2804 #endif
2805 	{ }
2806 };
2807 
2808 static struct ctl_table debug_table[] = {
2809 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
2810 	{
2811 		.procname	= "exception-trace",
2812 		.data		= &show_unhandled_signals,
2813 		.maxlen		= sizeof(int),
2814 		.mode		= 0644,
2815 		.proc_handler	= proc_dointvec
2816 	},
2817 #endif
2818 	{ }
2819 };
2820 
2821 static struct ctl_table dev_table[] = {
2822 	{ }
2823 };
2824 
2825 DECLARE_SYSCTL_BASE(kernel, kern_table);
2826 DECLARE_SYSCTL_BASE(vm, vm_table);
2827 DECLARE_SYSCTL_BASE(debug, debug_table);
2828 DECLARE_SYSCTL_BASE(dev, dev_table);
2829 
2830 int __init sysctl_init_bases(void)
2831 {
2832 	register_sysctl_base(kernel);
2833 	register_sysctl_base(vm);
2834 	register_sysctl_base(debug);
2835 	register_sysctl_base(dev);
2836 
2837 	return 0;
2838 }
2839 #endif /* CONFIG_SYSCTL */
2840 /*
2841  * No sense putting this after each symbol definition, twice,
2842  * exception granted :-)
2843  */
2844 EXPORT_SYMBOL(proc_dobool);
2845 EXPORT_SYMBOL(proc_dointvec);
2846 EXPORT_SYMBOL(proc_douintvec);
2847 EXPORT_SYMBOL(proc_dointvec_jiffies);
2848 EXPORT_SYMBOL(proc_dointvec_minmax);
2849 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
2850 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2851 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2852 EXPORT_SYMBOL(proc_dostring);
2853 EXPORT_SYMBOL(proc_doulongvec_minmax);
2854 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2855 EXPORT_SYMBOL(proc_do_large_bitmap);
2856