xref: /linux/kernel/sysctl.c (revision 3a755ebcc2557e22b895b8976257f682c653db1d)
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 void __weak unpriv_ebpf_notify(int new_state)
184 {
185 }
186 
187 static int bpf_unpriv_handler(struct ctl_table *table, int write,
188 			      void *buffer, size_t *lenp, loff_t *ppos)
189 {
190 	int ret, unpriv_enable = *(int *)table->data;
191 	bool locked_state = unpriv_enable == 1;
192 	struct ctl_table tmp = *table;
193 
194 	if (write && !capable(CAP_SYS_ADMIN))
195 		return -EPERM;
196 
197 	tmp.data = &unpriv_enable;
198 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
199 	if (write && !ret) {
200 		if (locked_state && unpriv_enable != 1)
201 			return -EPERM;
202 		*(int *)table->data = unpriv_enable;
203 	}
204 
205 	unpriv_ebpf_notify(unpriv_enable);
206 
207 	return ret;
208 }
209 #endif /* CONFIG_BPF_SYSCALL && CONFIG_SYSCTL */
210 
211 /*
212  * /proc/sys support
213  */
214 
215 #ifdef CONFIG_PROC_SYSCTL
216 
217 static int _proc_do_string(char *data, int maxlen, int write,
218 		char *buffer, size_t *lenp, loff_t *ppos)
219 {
220 	size_t len;
221 	char c, *p;
222 
223 	if (!data || !maxlen || !*lenp) {
224 		*lenp = 0;
225 		return 0;
226 	}
227 
228 	if (write) {
229 		if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
230 			/* Only continue writes not past the end of buffer. */
231 			len = strlen(data);
232 			if (len > maxlen - 1)
233 				len = maxlen - 1;
234 
235 			if (*ppos > len)
236 				return 0;
237 			len = *ppos;
238 		} else {
239 			/* Start writing from beginning of buffer. */
240 			len = 0;
241 		}
242 
243 		*ppos += *lenp;
244 		p = buffer;
245 		while ((p - buffer) < *lenp && len < maxlen - 1) {
246 			c = *(p++);
247 			if (c == 0 || c == '\n')
248 				break;
249 			data[len++] = c;
250 		}
251 		data[len] = 0;
252 	} else {
253 		len = strlen(data);
254 		if (len > maxlen)
255 			len = maxlen;
256 
257 		if (*ppos > len) {
258 			*lenp = 0;
259 			return 0;
260 		}
261 
262 		data += *ppos;
263 		len  -= *ppos;
264 
265 		if (len > *lenp)
266 			len = *lenp;
267 		if (len)
268 			memcpy(buffer, data, len);
269 		if (len < *lenp) {
270 			buffer[len] = '\n';
271 			len++;
272 		}
273 		*lenp = len;
274 		*ppos += len;
275 	}
276 	return 0;
277 }
278 
279 static void warn_sysctl_write(struct ctl_table *table)
280 {
281 	pr_warn_once("%s wrote to %s when file position was not 0!\n"
282 		"This will not be supported in the future. To silence this\n"
283 		"warning, set kernel.sysctl_writes_strict = -1\n",
284 		current->comm, table->procname);
285 }
286 
287 /**
288  * proc_first_pos_non_zero_ignore - check if first position is allowed
289  * @ppos: file position
290  * @table: the sysctl table
291  *
292  * Returns true if the first position is non-zero and the sysctl_writes_strict
293  * mode indicates this is not allowed for numeric input types. String proc
294  * handlers can ignore the return value.
295  */
296 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
297 					   struct ctl_table *table)
298 {
299 	if (!*ppos)
300 		return false;
301 
302 	switch (sysctl_writes_strict) {
303 	case SYSCTL_WRITES_STRICT:
304 		return true;
305 	case SYSCTL_WRITES_WARN:
306 		warn_sysctl_write(table);
307 		return false;
308 	default:
309 		return false;
310 	}
311 }
312 
313 /**
314  * proc_dostring - read a string sysctl
315  * @table: the sysctl table
316  * @write: %TRUE if this is a write to the sysctl file
317  * @buffer: the user buffer
318  * @lenp: the size of the user buffer
319  * @ppos: file position
320  *
321  * Reads/writes a string from/to the user buffer. If the kernel
322  * buffer provided is not large enough to hold the string, the
323  * string is truncated. The copied string is %NULL-terminated.
324  * If the string is being read by the user process, it is copied
325  * and a newline '\n' is added. It is truncated if the buffer is
326  * not large enough.
327  *
328  * Returns 0 on success.
329  */
330 int proc_dostring(struct ctl_table *table, int write,
331 		  void *buffer, size_t *lenp, loff_t *ppos)
332 {
333 	if (write)
334 		proc_first_pos_non_zero_ignore(ppos, table);
335 
336 	return _proc_do_string(table->data, table->maxlen, write, buffer, lenp,
337 			ppos);
338 }
339 
340 static size_t proc_skip_spaces(char **buf)
341 {
342 	size_t ret;
343 	char *tmp = skip_spaces(*buf);
344 	ret = tmp - *buf;
345 	*buf = tmp;
346 	return ret;
347 }
348 
349 static void proc_skip_char(char **buf, size_t *size, const char v)
350 {
351 	while (*size) {
352 		if (**buf != v)
353 			break;
354 		(*size)--;
355 		(*buf)++;
356 	}
357 }
358 
359 /**
360  * strtoul_lenient - parse an ASCII formatted integer from a buffer and only
361  *                   fail on overflow
362  *
363  * @cp: kernel buffer containing the string to parse
364  * @endp: pointer to store the trailing characters
365  * @base: the base to use
366  * @res: where the parsed integer will be stored
367  *
368  * In case of success 0 is returned and @res will contain the parsed integer,
369  * @endp will hold any trailing characters.
370  * This function will fail the parse on overflow. If there wasn't an overflow
371  * the function will defer the decision what characters count as invalid to the
372  * caller.
373  */
374 static int strtoul_lenient(const char *cp, char **endp, unsigned int base,
375 			   unsigned long *res)
376 {
377 	unsigned long long result;
378 	unsigned int rv;
379 
380 	cp = _parse_integer_fixup_radix(cp, &base);
381 	rv = _parse_integer(cp, base, &result);
382 	if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result))
383 		return -ERANGE;
384 
385 	cp += rv;
386 
387 	if (endp)
388 		*endp = (char *)cp;
389 
390 	*res = (unsigned long)result;
391 	return 0;
392 }
393 
394 #define TMPBUFLEN 22
395 /**
396  * proc_get_long - reads an ASCII formatted integer from a user buffer
397  *
398  * @buf: a kernel buffer
399  * @size: size of the kernel buffer
400  * @val: this is where the number will be stored
401  * @neg: set to %TRUE if number is negative
402  * @perm_tr: a vector which contains the allowed trailers
403  * @perm_tr_len: size of the perm_tr vector
404  * @tr: pointer to store the trailer character
405  *
406  * In case of success %0 is returned and @buf and @size are updated with
407  * the amount of bytes read. If @tr is non-NULL and a trailing
408  * character exists (size is non-zero after returning from this
409  * function), @tr is updated with the trailing character.
410  */
411 static int proc_get_long(char **buf, size_t *size,
412 			  unsigned long *val, bool *neg,
413 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
414 {
415 	int len;
416 	char *p, tmp[TMPBUFLEN];
417 
418 	if (!*size)
419 		return -EINVAL;
420 
421 	len = *size;
422 	if (len > TMPBUFLEN - 1)
423 		len = TMPBUFLEN - 1;
424 
425 	memcpy(tmp, *buf, len);
426 
427 	tmp[len] = 0;
428 	p = tmp;
429 	if (*p == '-' && *size > 1) {
430 		*neg = true;
431 		p++;
432 	} else
433 		*neg = false;
434 	if (!isdigit(*p))
435 		return -EINVAL;
436 
437 	if (strtoul_lenient(p, &p, 0, val))
438 		return -EINVAL;
439 
440 	len = p - tmp;
441 
442 	/* We don't know if the next char is whitespace thus we may accept
443 	 * invalid integers (e.g. 1234...a) or two integers instead of one
444 	 * (e.g. 123...1). So lets not allow such large numbers. */
445 	if (len == TMPBUFLEN - 1)
446 		return -EINVAL;
447 
448 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
449 		return -EINVAL;
450 
451 	if (tr && (len < *size))
452 		*tr = *p;
453 
454 	*buf += len;
455 	*size -= len;
456 
457 	return 0;
458 }
459 
460 /**
461  * proc_put_long - converts an integer to a decimal ASCII formatted string
462  *
463  * @buf: the user buffer
464  * @size: the size of the user buffer
465  * @val: the integer to be converted
466  * @neg: sign of the number, %TRUE for negative
467  *
468  * In case of success @buf and @size are updated with the amount of bytes
469  * written.
470  */
471 static void proc_put_long(void **buf, size_t *size, unsigned long val, bool neg)
472 {
473 	int len;
474 	char tmp[TMPBUFLEN], *p = tmp;
475 
476 	sprintf(p, "%s%lu", neg ? "-" : "", val);
477 	len = strlen(tmp);
478 	if (len > *size)
479 		len = *size;
480 	memcpy(*buf, tmp, len);
481 	*size -= len;
482 	*buf += len;
483 }
484 #undef TMPBUFLEN
485 
486 static void proc_put_char(void **buf, size_t *size, char c)
487 {
488 	if (*size) {
489 		char **buffer = (char **)buf;
490 		**buffer = c;
491 
492 		(*size)--;
493 		(*buffer)++;
494 		*buf = *buffer;
495 	}
496 }
497 
498 static int do_proc_dobool_conv(bool *negp, unsigned long *lvalp,
499 				int *valp,
500 				int write, void *data)
501 {
502 	if (write) {
503 		*(bool *)valp = *lvalp;
504 	} else {
505 		int val = *(bool *)valp;
506 
507 		*lvalp = (unsigned long)val;
508 		*negp = false;
509 	}
510 	return 0;
511 }
512 
513 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
514 				 int *valp,
515 				 int write, void *data)
516 {
517 	if (write) {
518 		if (*negp) {
519 			if (*lvalp > (unsigned long) INT_MAX + 1)
520 				return -EINVAL;
521 			*valp = -*lvalp;
522 		} else {
523 			if (*lvalp > (unsigned long) INT_MAX)
524 				return -EINVAL;
525 			*valp = *lvalp;
526 		}
527 	} else {
528 		int val = *valp;
529 		if (val < 0) {
530 			*negp = true;
531 			*lvalp = -(unsigned long)val;
532 		} else {
533 			*negp = false;
534 			*lvalp = (unsigned long)val;
535 		}
536 	}
537 	return 0;
538 }
539 
540 static int do_proc_douintvec_conv(unsigned long *lvalp,
541 				  unsigned int *valp,
542 				  int write, void *data)
543 {
544 	if (write) {
545 		if (*lvalp > UINT_MAX)
546 			return -EINVAL;
547 		*valp = *lvalp;
548 	} else {
549 		unsigned int val = *valp;
550 		*lvalp = (unsigned long)val;
551 	}
552 	return 0;
553 }
554 
555 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
556 
557 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
558 		  int write, void *buffer,
559 		  size_t *lenp, loff_t *ppos,
560 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
561 			      int write, void *data),
562 		  void *data)
563 {
564 	int *i, vleft, first = 1, err = 0;
565 	size_t left;
566 	char *p;
567 
568 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
569 		*lenp = 0;
570 		return 0;
571 	}
572 
573 	i = (int *) tbl_data;
574 	vleft = table->maxlen / sizeof(*i);
575 	left = *lenp;
576 
577 	if (!conv)
578 		conv = do_proc_dointvec_conv;
579 
580 	if (write) {
581 		if (proc_first_pos_non_zero_ignore(ppos, table))
582 			goto out;
583 
584 		if (left > PAGE_SIZE - 1)
585 			left = PAGE_SIZE - 1;
586 		p = buffer;
587 	}
588 
589 	for (; left && vleft--; i++, first=0) {
590 		unsigned long lval;
591 		bool neg;
592 
593 		if (write) {
594 			left -= proc_skip_spaces(&p);
595 
596 			if (!left)
597 				break;
598 			err = proc_get_long(&p, &left, &lval, &neg,
599 					     proc_wspace_sep,
600 					     sizeof(proc_wspace_sep), NULL);
601 			if (err)
602 				break;
603 			if (conv(&neg, &lval, i, 1, data)) {
604 				err = -EINVAL;
605 				break;
606 			}
607 		} else {
608 			if (conv(&neg, &lval, i, 0, data)) {
609 				err = -EINVAL;
610 				break;
611 			}
612 			if (!first)
613 				proc_put_char(&buffer, &left, '\t');
614 			proc_put_long(&buffer, &left, lval, neg);
615 		}
616 	}
617 
618 	if (!write && !first && left && !err)
619 		proc_put_char(&buffer, &left, '\n');
620 	if (write && !err && left)
621 		left -= proc_skip_spaces(&p);
622 	if (write && first)
623 		return err ? : -EINVAL;
624 	*lenp -= left;
625 out:
626 	*ppos += *lenp;
627 	return err;
628 }
629 
630 static int do_proc_dointvec(struct ctl_table *table, int write,
631 		  void *buffer, size_t *lenp, loff_t *ppos,
632 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
633 			      int write, void *data),
634 		  void *data)
635 {
636 	return __do_proc_dointvec(table->data, table, write,
637 			buffer, lenp, ppos, conv, data);
638 }
639 
640 static int do_proc_douintvec_w(unsigned int *tbl_data,
641 			       struct ctl_table *table,
642 			       void *buffer,
643 			       size_t *lenp, loff_t *ppos,
644 			       int (*conv)(unsigned long *lvalp,
645 					   unsigned int *valp,
646 					   int write, void *data),
647 			       void *data)
648 {
649 	unsigned long lval;
650 	int err = 0;
651 	size_t left;
652 	bool neg;
653 	char *p = buffer;
654 
655 	left = *lenp;
656 
657 	if (proc_first_pos_non_zero_ignore(ppos, table))
658 		goto bail_early;
659 
660 	if (left > PAGE_SIZE - 1)
661 		left = PAGE_SIZE - 1;
662 
663 	left -= proc_skip_spaces(&p);
664 	if (!left) {
665 		err = -EINVAL;
666 		goto out_free;
667 	}
668 
669 	err = proc_get_long(&p, &left, &lval, &neg,
670 			     proc_wspace_sep,
671 			     sizeof(proc_wspace_sep), NULL);
672 	if (err || neg) {
673 		err = -EINVAL;
674 		goto out_free;
675 	}
676 
677 	if (conv(&lval, tbl_data, 1, data)) {
678 		err = -EINVAL;
679 		goto out_free;
680 	}
681 
682 	if (!err && left)
683 		left -= proc_skip_spaces(&p);
684 
685 out_free:
686 	if (err)
687 		return -EINVAL;
688 
689 	return 0;
690 
691 	/* This is in keeping with old __do_proc_dointvec() */
692 bail_early:
693 	*ppos += *lenp;
694 	return err;
695 }
696 
697 static int do_proc_douintvec_r(unsigned int *tbl_data, void *buffer,
698 			       size_t *lenp, loff_t *ppos,
699 			       int (*conv)(unsigned long *lvalp,
700 					   unsigned int *valp,
701 					   int write, void *data),
702 			       void *data)
703 {
704 	unsigned long lval;
705 	int err = 0;
706 	size_t left;
707 
708 	left = *lenp;
709 
710 	if (conv(&lval, tbl_data, 0, data)) {
711 		err = -EINVAL;
712 		goto out;
713 	}
714 
715 	proc_put_long(&buffer, &left, lval, false);
716 	if (!left)
717 		goto out;
718 
719 	proc_put_char(&buffer, &left, '\n');
720 
721 out:
722 	*lenp -= left;
723 	*ppos += *lenp;
724 
725 	return err;
726 }
727 
728 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
729 			       int write, void *buffer,
730 			       size_t *lenp, loff_t *ppos,
731 			       int (*conv)(unsigned long *lvalp,
732 					   unsigned int *valp,
733 					   int write, void *data),
734 			       void *data)
735 {
736 	unsigned int *i, vleft;
737 
738 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
739 		*lenp = 0;
740 		return 0;
741 	}
742 
743 	i = (unsigned int *) tbl_data;
744 	vleft = table->maxlen / sizeof(*i);
745 
746 	/*
747 	 * Arrays are not supported, keep this simple. *Do not* add
748 	 * support for them.
749 	 */
750 	if (vleft != 1) {
751 		*lenp = 0;
752 		return -EINVAL;
753 	}
754 
755 	if (!conv)
756 		conv = do_proc_douintvec_conv;
757 
758 	if (write)
759 		return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
760 					   conv, data);
761 	return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
762 }
763 
764 int do_proc_douintvec(struct ctl_table *table, int write,
765 		      void *buffer, size_t *lenp, loff_t *ppos,
766 		      int (*conv)(unsigned long *lvalp,
767 				  unsigned int *valp,
768 				  int write, void *data),
769 		      void *data)
770 {
771 	return __do_proc_douintvec(table->data, table, write,
772 				   buffer, lenp, ppos, conv, data);
773 }
774 
775 /**
776  * proc_dobool - read/write a bool
777  * @table: the sysctl table
778  * @write: %TRUE if this is a write to the sysctl file
779  * @buffer: the user buffer
780  * @lenp: the size of the user buffer
781  * @ppos: file position
782  *
783  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
784  * values from/to the user buffer, treated as an ASCII string.
785  *
786  * Returns 0 on success.
787  */
788 int proc_dobool(struct ctl_table *table, int write, void *buffer,
789 		size_t *lenp, loff_t *ppos)
790 {
791 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
792 				do_proc_dobool_conv, NULL);
793 }
794 
795 /**
796  * proc_dointvec - read a vector of integers
797  * @table: the sysctl table
798  * @write: %TRUE if this is a write to the sysctl file
799  * @buffer: the user buffer
800  * @lenp: the size of the user buffer
801  * @ppos: file position
802  *
803  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
804  * values from/to the user buffer, treated as an ASCII string.
805  *
806  * Returns 0 on success.
807  */
808 int proc_dointvec(struct ctl_table *table, int write, void *buffer,
809 		  size_t *lenp, loff_t *ppos)
810 {
811 	return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
812 }
813 
814 #ifdef CONFIG_COMPACTION
815 static int proc_dointvec_minmax_warn_RT_change(struct ctl_table *table,
816 		int write, void *buffer, size_t *lenp, loff_t *ppos)
817 {
818 	int ret, old;
819 
820 	if (!IS_ENABLED(CONFIG_PREEMPT_RT) || !write)
821 		return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
822 
823 	old = *(int *)table->data;
824 	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
825 	if (ret)
826 		return ret;
827 	if (old != *(int *)table->data)
828 		pr_warn_once("sysctl attribute %s changed by %s[%d]\n",
829 			     table->procname, current->comm,
830 			     task_pid_nr(current));
831 	return ret;
832 }
833 #endif
834 
835 /**
836  * proc_douintvec - read a vector of unsigned integers
837  * @table: the sysctl table
838  * @write: %TRUE if this is a write to the sysctl file
839  * @buffer: the user buffer
840  * @lenp: the size of the user buffer
841  * @ppos: file position
842  *
843  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
844  * values from/to the user buffer, treated as an ASCII string.
845  *
846  * Returns 0 on success.
847  */
848 int proc_douintvec(struct ctl_table *table, int write, void *buffer,
849 		size_t *lenp, loff_t *ppos)
850 {
851 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
852 				 do_proc_douintvec_conv, NULL);
853 }
854 
855 /*
856  * Taint values can only be increased
857  * This means we can safely use a temporary.
858  */
859 static int proc_taint(struct ctl_table *table, int write,
860 			       void *buffer, size_t *lenp, loff_t *ppos)
861 {
862 	struct ctl_table t;
863 	unsigned long tmptaint = get_taint();
864 	int err;
865 
866 	if (write && !capable(CAP_SYS_ADMIN))
867 		return -EPERM;
868 
869 	t = *table;
870 	t.data = &tmptaint;
871 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
872 	if (err < 0)
873 		return err;
874 
875 	if (write) {
876 		int i;
877 
878 		/*
879 		 * If we are relying on panic_on_taint not producing
880 		 * false positives due to userspace input, bail out
881 		 * before setting the requested taint flags.
882 		 */
883 		if (panic_on_taint_nousertaint && (tmptaint & panic_on_taint))
884 			return -EINVAL;
885 
886 		/*
887 		 * Poor man's atomic or. Not worth adding a primitive
888 		 * to everyone's atomic.h for this
889 		 */
890 		for (i = 0; i < TAINT_FLAGS_COUNT; i++)
891 			if ((1UL << i) & tmptaint)
892 				add_taint(i, LOCKDEP_STILL_OK);
893 	}
894 
895 	return err;
896 }
897 
898 /**
899  * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
900  * @min: pointer to minimum allowable value
901  * @max: pointer to maximum allowable value
902  *
903  * The do_proc_dointvec_minmax_conv_param structure provides the
904  * minimum and maximum values for doing range checking for those sysctl
905  * parameters that use the proc_dointvec_minmax() handler.
906  */
907 struct do_proc_dointvec_minmax_conv_param {
908 	int *min;
909 	int *max;
910 };
911 
912 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
913 					int *valp,
914 					int write, void *data)
915 {
916 	int tmp, ret;
917 	struct do_proc_dointvec_minmax_conv_param *param = data;
918 	/*
919 	 * If writing, first do so via a temporary local int so we can
920 	 * bounds-check it before touching *valp.
921 	 */
922 	int *ip = write ? &tmp : valp;
923 
924 	ret = do_proc_dointvec_conv(negp, lvalp, ip, write, data);
925 	if (ret)
926 		return ret;
927 
928 	if (write) {
929 		if ((param->min && *param->min > tmp) ||
930 		    (param->max && *param->max < tmp))
931 			return -EINVAL;
932 		*valp = tmp;
933 	}
934 
935 	return 0;
936 }
937 
938 /**
939  * proc_dointvec_minmax - read a vector of integers with min/max values
940  * @table: the sysctl table
941  * @write: %TRUE if this is a write to the sysctl file
942  * @buffer: the user buffer
943  * @lenp: the size of the user buffer
944  * @ppos: file position
945  *
946  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
947  * values from/to the user buffer, treated as an ASCII string.
948  *
949  * This routine will ensure the values are within the range specified by
950  * table->extra1 (min) and table->extra2 (max).
951  *
952  * Returns 0 on success or -EINVAL on write when the range check fails.
953  */
954 int proc_dointvec_minmax(struct ctl_table *table, int write,
955 		  void *buffer, size_t *lenp, loff_t *ppos)
956 {
957 	struct do_proc_dointvec_minmax_conv_param param = {
958 		.min = (int *) table->extra1,
959 		.max = (int *) table->extra2,
960 	};
961 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
962 				do_proc_dointvec_minmax_conv, &param);
963 }
964 
965 /**
966  * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
967  * @min: pointer to minimum allowable value
968  * @max: pointer to maximum allowable value
969  *
970  * The do_proc_douintvec_minmax_conv_param structure provides the
971  * minimum and maximum values for doing range checking for those sysctl
972  * parameters that use the proc_douintvec_minmax() handler.
973  */
974 struct do_proc_douintvec_minmax_conv_param {
975 	unsigned int *min;
976 	unsigned int *max;
977 };
978 
979 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
980 					 unsigned int *valp,
981 					 int write, void *data)
982 {
983 	int ret;
984 	unsigned int tmp;
985 	struct do_proc_douintvec_minmax_conv_param *param = data;
986 	/* write via temporary local uint for bounds-checking */
987 	unsigned int *up = write ? &tmp : valp;
988 
989 	ret = do_proc_douintvec_conv(lvalp, up, write, data);
990 	if (ret)
991 		return ret;
992 
993 	if (write) {
994 		if ((param->min && *param->min > tmp) ||
995 		    (param->max && *param->max < tmp))
996 			return -ERANGE;
997 
998 		*valp = tmp;
999 	}
1000 
1001 	return 0;
1002 }
1003 
1004 /**
1005  * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
1006  * @table: the sysctl table
1007  * @write: %TRUE if this is a write to the sysctl file
1008  * @buffer: the user buffer
1009  * @lenp: the size of the user buffer
1010  * @ppos: file position
1011  *
1012  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
1013  * values from/to the user buffer, treated as an ASCII string. Negative
1014  * strings are not allowed.
1015  *
1016  * This routine will ensure the values are within the range specified by
1017  * table->extra1 (min) and table->extra2 (max). There is a final sanity
1018  * check for UINT_MAX to avoid having to support wrap around uses from
1019  * userspace.
1020  *
1021  * Returns 0 on success or -ERANGE on write when the range check fails.
1022  */
1023 int proc_douintvec_minmax(struct ctl_table *table, int write,
1024 			  void *buffer, size_t *lenp, loff_t *ppos)
1025 {
1026 	struct do_proc_douintvec_minmax_conv_param param = {
1027 		.min = (unsigned int *) table->extra1,
1028 		.max = (unsigned int *) table->extra2,
1029 	};
1030 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
1031 				 do_proc_douintvec_minmax_conv, &param);
1032 }
1033 
1034 /**
1035  * proc_dou8vec_minmax - read a vector of unsigned chars with min/max values
1036  * @table: the sysctl table
1037  * @write: %TRUE if this is a write to the sysctl file
1038  * @buffer: the user buffer
1039  * @lenp: the size of the user buffer
1040  * @ppos: file position
1041  *
1042  * Reads/writes up to table->maxlen/sizeof(u8) unsigned chars
1043  * values from/to the user buffer, treated as an ASCII string. Negative
1044  * strings are not allowed.
1045  *
1046  * This routine will ensure the values are within the range specified by
1047  * table->extra1 (min) and table->extra2 (max).
1048  *
1049  * Returns 0 on success or an error on write when the range check fails.
1050  */
1051 int proc_dou8vec_minmax(struct ctl_table *table, int write,
1052 			void *buffer, size_t *lenp, loff_t *ppos)
1053 {
1054 	struct ctl_table tmp;
1055 	unsigned int min = 0, max = 255U, val;
1056 	u8 *data = table->data;
1057 	struct do_proc_douintvec_minmax_conv_param param = {
1058 		.min = &min,
1059 		.max = &max,
1060 	};
1061 	int res;
1062 
1063 	/* Do not support arrays yet. */
1064 	if (table->maxlen != sizeof(u8))
1065 		return -EINVAL;
1066 
1067 	if (table->extra1) {
1068 		min = *(unsigned int *) table->extra1;
1069 		if (min > 255U)
1070 			return -EINVAL;
1071 	}
1072 	if (table->extra2) {
1073 		max = *(unsigned int *) table->extra2;
1074 		if (max > 255U)
1075 			return -EINVAL;
1076 	}
1077 
1078 	tmp = *table;
1079 
1080 	tmp.maxlen = sizeof(val);
1081 	tmp.data = &val;
1082 	val = *data;
1083 	res = do_proc_douintvec(&tmp, write, buffer, lenp, ppos,
1084 				do_proc_douintvec_minmax_conv, &param);
1085 	if (res)
1086 		return res;
1087 	if (write)
1088 		*data = val;
1089 	return 0;
1090 }
1091 EXPORT_SYMBOL_GPL(proc_dou8vec_minmax);
1092 
1093 #ifdef CONFIG_MAGIC_SYSRQ
1094 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
1095 				void *buffer, size_t *lenp, loff_t *ppos)
1096 {
1097 	int tmp, ret;
1098 
1099 	tmp = sysrq_mask();
1100 
1101 	ret = __do_proc_dointvec(&tmp, table, write, buffer,
1102 			       lenp, ppos, NULL, NULL);
1103 	if (ret || !write)
1104 		return ret;
1105 
1106 	if (write)
1107 		sysrq_toggle_support(tmp);
1108 
1109 	return 0;
1110 }
1111 #endif
1112 
1113 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table,
1114 		int write, void *buffer, size_t *lenp, loff_t *ppos,
1115 		unsigned long convmul, unsigned long convdiv)
1116 {
1117 	unsigned long *i, *min, *max;
1118 	int vleft, first = 1, err = 0;
1119 	size_t left;
1120 	char *p;
1121 
1122 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
1123 		*lenp = 0;
1124 		return 0;
1125 	}
1126 
1127 	i = (unsigned long *) data;
1128 	min = (unsigned long *) table->extra1;
1129 	max = (unsigned long *) table->extra2;
1130 	vleft = table->maxlen / sizeof(unsigned long);
1131 	left = *lenp;
1132 
1133 	if (write) {
1134 		if (proc_first_pos_non_zero_ignore(ppos, table))
1135 			goto out;
1136 
1137 		if (left > PAGE_SIZE - 1)
1138 			left = PAGE_SIZE - 1;
1139 		p = buffer;
1140 	}
1141 
1142 	for (; left && vleft--; i++, first = 0) {
1143 		unsigned long val;
1144 
1145 		if (write) {
1146 			bool neg;
1147 
1148 			left -= proc_skip_spaces(&p);
1149 			if (!left)
1150 				break;
1151 
1152 			err = proc_get_long(&p, &left, &val, &neg,
1153 					     proc_wspace_sep,
1154 					     sizeof(proc_wspace_sep), NULL);
1155 			if (err || neg) {
1156 				err = -EINVAL;
1157 				break;
1158 			}
1159 
1160 			val = convmul * val / convdiv;
1161 			if ((min && val < *min) || (max && val > *max)) {
1162 				err = -EINVAL;
1163 				break;
1164 			}
1165 			*i = val;
1166 		} else {
1167 			val = convdiv * (*i) / convmul;
1168 			if (!first)
1169 				proc_put_char(&buffer, &left, '\t');
1170 			proc_put_long(&buffer, &left, val, false);
1171 		}
1172 	}
1173 
1174 	if (!write && !first && left && !err)
1175 		proc_put_char(&buffer, &left, '\n');
1176 	if (write && !err)
1177 		left -= proc_skip_spaces(&p);
1178 	if (write && first)
1179 		return err ? : -EINVAL;
1180 	*lenp -= left;
1181 out:
1182 	*ppos += *lenp;
1183 	return err;
1184 }
1185 
1186 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
1187 		void *buffer, size_t *lenp, loff_t *ppos, unsigned long convmul,
1188 		unsigned long convdiv)
1189 {
1190 	return __do_proc_doulongvec_minmax(table->data, table, write,
1191 			buffer, lenp, ppos, convmul, convdiv);
1192 }
1193 
1194 /**
1195  * proc_doulongvec_minmax - read a vector of long integers with min/max values
1196  * @table: the sysctl table
1197  * @write: %TRUE if this is a write to the sysctl file
1198  * @buffer: the user buffer
1199  * @lenp: the size of the user buffer
1200  * @ppos: file position
1201  *
1202  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1203  * values from/to the user buffer, treated as an ASCII string.
1204  *
1205  * This routine will ensure the values are within the range specified by
1206  * table->extra1 (min) and table->extra2 (max).
1207  *
1208  * Returns 0 on success.
1209  */
1210 int proc_doulongvec_minmax(struct ctl_table *table, int write,
1211 			   void *buffer, size_t *lenp, loff_t *ppos)
1212 {
1213     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
1214 }
1215 
1216 /**
1217  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
1218  * @table: the sysctl table
1219  * @write: %TRUE if this is a write to the sysctl file
1220  * @buffer: the user buffer
1221  * @lenp: the size of the user buffer
1222  * @ppos: file position
1223  *
1224  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1225  * values from/to the user buffer, treated as an ASCII string. The values
1226  * are treated as milliseconds, and converted to jiffies when they are stored.
1227  *
1228  * This routine will ensure the values are within the range specified by
1229  * table->extra1 (min) and table->extra2 (max).
1230  *
1231  * Returns 0 on success.
1232  */
1233 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1234 				      void *buffer, size_t *lenp, loff_t *ppos)
1235 {
1236     return do_proc_doulongvec_minmax(table, write, buffer,
1237 				     lenp, ppos, HZ, 1000l);
1238 }
1239 
1240 
1241 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
1242 					 int *valp,
1243 					 int write, void *data)
1244 {
1245 	if (write) {
1246 		if (*lvalp > INT_MAX / HZ)
1247 			return 1;
1248 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
1249 	} else {
1250 		int val = *valp;
1251 		unsigned long lval;
1252 		if (val < 0) {
1253 			*negp = true;
1254 			lval = -(unsigned long)val;
1255 		} else {
1256 			*negp = false;
1257 			lval = (unsigned long)val;
1258 		}
1259 		*lvalp = lval / HZ;
1260 	}
1261 	return 0;
1262 }
1263 
1264 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
1265 						int *valp,
1266 						int write, void *data)
1267 {
1268 	if (write) {
1269 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
1270 			return 1;
1271 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
1272 	} else {
1273 		int val = *valp;
1274 		unsigned long lval;
1275 		if (val < 0) {
1276 			*negp = true;
1277 			lval = -(unsigned long)val;
1278 		} else {
1279 			*negp = false;
1280 			lval = (unsigned long)val;
1281 		}
1282 		*lvalp = jiffies_to_clock_t(lval);
1283 	}
1284 	return 0;
1285 }
1286 
1287 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
1288 					    int *valp,
1289 					    int write, void *data)
1290 {
1291 	if (write) {
1292 		unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
1293 
1294 		if (jif > INT_MAX)
1295 			return 1;
1296 		*valp = (int)jif;
1297 	} else {
1298 		int val = *valp;
1299 		unsigned long lval;
1300 		if (val < 0) {
1301 			*negp = true;
1302 			lval = -(unsigned long)val;
1303 		} else {
1304 			*negp = false;
1305 			lval = (unsigned long)val;
1306 		}
1307 		*lvalp = jiffies_to_msecs(lval);
1308 	}
1309 	return 0;
1310 }
1311 
1312 /**
1313  * proc_dointvec_jiffies - read a vector of integers as seconds
1314  * @table: the sysctl table
1315  * @write: %TRUE if this is a write to the sysctl file
1316  * @buffer: the user buffer
1317  * @lenp: the size of the user buffer
1318  * @ppos: file position
1319  *
1320  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1321  * values from/to the user buffer, treated as an ASCII string.
1322  * The values read are assumed to be in seconds, and are converted into
1323  * jiffies.
1324  *
1325  * Returns 0 on success.
1326  */
1327 int proc_dointvec_jiffies(struct ctl_table *table, int write,
1328 			  void *buffer, size_t *lenp, loff_t *ppos)
1329 {
1330     return do_proc_dointvec(table,write,buffer,lenp,ppos,
1331 		    	    do_proc_dointvec_jiffies_conv,NULL);
1332 }
1333 
1334 /**
1335  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
1336  * @table: the sysctl table
1337  * @write: %TRUE if this is a write to the sysctl file
1338  * @buffer: the user buffer
1339  * @lenp: the size of the user buffer
1340  * @ppos: pointer to the file position
1341  *
1342  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1343  * values from/to the user buffer, treated as an ASCII string.
1344  * The values read are assumed to be in 1/USER_HZ seconds, and
1345  * are converted into jiffies.
1346  *
1347  * Returns 0 on success.
1348  */
1349 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1350 				 void *buffer, size_t *lenp, loff_t *ppos)
1351 {
1352     return do_proc_dointvec(table,write,buffer,lenp,ppos,
1353 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
1354 }
1355 
1356 /**
1357  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
1358  * @table: the sysctl table
1359  * @write: %TRUE if this is a write to the sysctl file
1360  * @buffer: the user buffer
1361  * @lenp: the size of the user buffer
1362  * @ppos: file position
1363  * @ppos: the current position in the file
1364  *
1365  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1366  * values from/to the user buffer, treated as an ASCII string.
1367  * The values read are assumed to be in 1/1000 seconds, and
1368  * are converted into jiffies.
1369  *
1370  * Returns 0 on success.
1371  */
1372 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, void *buffer,
1373 		size_t *lenp, loff_t *ppos)
1374 {
1375 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
1376 				do_proc_dointvec_ms_jiffies_conv, NULL);
1377 }
1378 
1379 static int proc_do_cad_pid(struct ctl_table *table, int write, void *buffer,
1380 		size_t *lenp, loff_t *ppos)
1381 {
1382 	struct pid *new_pid;
1383 	pid_t tmp;
1384 	int r;
1385 
1386 	tmp = pid_vnr(cad_pid);
1387 
1388 	r = __do_proc_dointvec(&tmp, table, write, buffer,
1389 			       lenp, ppos, NULL, NULL);
1390 	if (r || !write)
1391 		return r;
1392 
1393 	new_pid = find_get_pid(tmp);
1394 	if (!new_pid)
1395 		return -ESRCH;
1396 
1397 	put_pid(xchg(&cad_pid, new_pid));
1398 	return 0;
1399 }
1400 
1401 /**
1402  * proc_do_large_bitmap - read/write from/to a large bitmap
1403  * @table: the sysctl table
1404  * @write: %TRUE if this is a write to the sysctl file
1405  * @buffer: the user buffer
1406  * @lenp: the size of the user buffer
1407  * @ppos: file position
1408  *
1409  * The bitmap is stored at table->data and the bitmap length (in bits)
1410  * in table->maxlen.
1411  *
1412  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
1413  * large bitmaps may be represented in a compact manner. Writing into
1414  * the file will clear the bitmap then update it with the given input.
1415  *
1416  * Returns 0 on success.
1417  */
1418 int proc_do_large_bitmap(struct ctl_table *table, int write,
1419 			 void *buffer, size_t *lenp, loff_t *ppos)
1420 {
1421 	int err = 0;
1422 	size_t left = *lenp;
1423 	unsigned long bitmap_len = table->maxlen;
1424 	unsigned long *bitmap = *(unsigned long **) table->data;
1425 	unsigned long *tmp_bitmap = NULL;
1426 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
1427 
1428 	if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
1429 		*lenp = 0;
1430 		return 0;
1431 	}
1432 
1433 	if (write) {
1434 		char *p = buffer;
1435 		size_t skipped = 0;
1436 
1437 		if (left > PAGE_SIZE - 1) {
1438 			left = PAGE_SIZE - 1;
1439 			/* How much of the buffer we'll skip this pass */
1440 			skipped = *lenp - left;
1441 		}
1442 
1443 		tmp_bitmap = bitmap_zalloc(bitmap_len, GFP_KERNEL);
1444 		if (!tmp_bitmap)
1445 			return -ENOMEM;
1446 		proc_skip_char(&p, &left, '\n');
1447 		while (!err && left) {
1448 			unsigned long val_a, val_b;
1449 			bool neg;
1450 			size_t saved_left;
1451 
1452 			/* In case we stop parsing mid-number, we can reset */
1453 			saved_left = left;
1454 			err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
1455 					     sizeof(tr_a), &c);
1456 			/*
1457 			 * If we consumed the entirety of a truncated buffer or
1458 			 * only one char is left (may be a "-"), then stop here,
1459 			 * reset, & come back for more.
1460 			 */
1461 			if ((left <= 1) && skipped) {
1462 				left = saved_left;
1463 				break;
1464 			}
1465 
1466 			if (err)
1467 				break;
1468 			if (val_a >= bitmap_len || neg) {
1469 				err = -EINVAL;
1470 				break;
1471 			}
1472 
1473 			val_b = val_a;
1474 			if (left) {
1475 				p++;
1476 				left--;
1477 			}
1478 
1479 			if (c == '-') {
1480 				err = proc_get_long(&p, &left, &val_b,
1481 						     &neg, tr_b, sizeof(tr_b),
1482 						     &c);
1483 				/*
1484 				 * If we consumed all of a truncated buffer or
1485 				 * then stop here, reset, & come back for more.
1486 				 */
1487 				if (!left && skipped) {
1488 					left = saved_left;
1489 					break;
1490 				}
1491 
1492 				if (err)
1493 					break;
1494 				if (val_b >= bitmap_len || neg ||
1495 				    val_a > val_b) {
1496 					err = -EINVAL;
1497 					break;
1498 				}
1499 				if (left) {
1500 					p++;
1501 					left--;
1502 				}
1503 			}
1504 
1505 			bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
1506 			proc_skip_char(&p, &left, '\n');
1507 		}
1508 		left += skipped;
1509 	} else {
1510 		unsigned long bit_a, bit_b = 0;
1511 		bool first = 1;
1512 
1513 		while (left) {
1514 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
1515 			if (bit_a >= bitmap_len)
1516 				break;
1517 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
1518 						   bit_a + 1) - 1;
1519 
1520 			if (!first)
1521 				proc_put_char(&buffer, &left, ',');
1522 			proc_put_long(&buffer, &left, bit_a, false);
1523 			if (bit_a != bit_b) {
1524 				proc_put_char(&buffer, &left, '-');
1525 				proc_put_long(&buffer, &left, bit_b, false);
1526 			}
1527 
1528 			first = 0; bit_b++;
1529 		}
1530 		proc_put_char(&buffer, &left, '\n');
1531 	}
1532 
1533 	if (!err) {
1534 		if (write) {
1535 			if (*ppos)
1536 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
1537 			else
1538 				bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
1539 		}
1540 		*lenp -= left;
1541 		*ppos += *lenp;
1542 	}
1543 
1544 	bitmap_free(tmp_bitmap);
1545 	return err;
1546 }
1547 
1548 #else /* CONFIG_PROC_SYSCTL */
1549 
1550 int proc_dostring(struct ctl_table *table, int write,
1551 		  void *buffer, size_t *lenp, loff_t *ppos)
1552 {
1553 	return -ENOSYS;
1554 }
1555 
1556 int proc_dobool(struct ctl_table *table, int write,
1557 		void *buffer, size_t *lenp, loff_t *ppos)
1558 {
1559 	return -ENOSYS;
1560 }
1561 
1562 int proc_dointvec(struct ctl_table *table, int write,
1563 		  void *buffer, size_t *lenp, loff_t *ppos)
1564 {
1565 	return -ENOSYS;
1566 }
1567 
1568 int proc_douintvec(struct ctl_table *table, int write,
1569 		  void *buffer, size_t *lenp, loff_t *ppos)
1570 {
1571 	return -ENOSYS;
1572 }
1573 
1574 int proc_dointvec_minmax(struct ctl_table *table, int write,
1575 		    void *buffer, size_t *lenp, loff_t *ppos)
1576 {
1577 	return -ENOSYS;
1578 }
1579 
1580 int proc_douintvec_minmax(struct ctl_table *table, int write,
1581 			  void *buffer, size_t *lenp, loff_t *ppos)
1582 {
1583 	return -ENOSYS;
1584 }
1585 
1586 int proc_dou8vec_minmax(struct ctl_table *table, int write,
1587 			void *buffer, size_t *lenp, loff_t *ppos)
1588 {
1589 	return -ENOSYS;
1590 }
1591 
1592 int proc_dointvec_jiffies(struct ctl_table *table, int write,
1593 		    void *buffer, size_t *lenp, loff_t *ppos)
1594 {
1595 	return -ENOSYS;
1596 }
1597 
1598 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1599 		    void *buffer, size_t *lenp, loff_t *ppos)
1600 {
1601 	return -ENOSYS;
1602 }
1603 
1604 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
1605 			     void *buffer, size_t *lenp, loff_t *ppos)
1606 {
1607 	return -ENOSYS;
1608 }
1609 
1610 int proc_doulongvec_minmax(struct ctl_table *table, int write,
1611 		    void *buffer, size_t *lenp, loff_t *ppos)
1612 {
1613 	return -ENOSYS;
1614 }
1615 
1616 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1617 				      void *buffer, size_t *lenp, loff_t *ppos)
1618 {
1619 	return -ENOSYS;
1620 }
1621 
1622 int proc_do_large_bitmap(struct ctl_table *table, int write,
1623 			 void *buffer, size_t *lenp, loff_t *ppos)
1624 {
1625 	return -ENOSYS;
1626 }
1627 
1628 #endif /* CONFIG_PROC_SYSCTL */
1629 
1630 #if defined(CONFIG_SYSCTL)
1631 int proc_do_static_key(struct ctl_table *table, int write,
1632 		       void *buffer, size_t *lenp, loff_t *ppos)
1633 {
1634 	struct static_key *key = (struct static_key *)table->data;
1635 	static DEFINE_MUTEX(static_key_mutex);
1636 	int val, ret;
1637 	struct ctl_table tmp = {
1638 		.data   = &val,
1639 		.maxlen = sizeof(val),
1640 		.mode   = table->mode,
1641 		.extra1 = SYSCTL_ZERO,
1642 		.extra2 = SYSCTL_ONE,
1643 	};
1644 
1645 	if (write && !capable(CAP_SYS_ADMIN))
1646 		return -EPERM;
1647 
1648 	mutex_lock(&static_key_mutex);
1649 	val = static_key_enabled(key);
1650 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
1651 	if (write && !ret) {
1652 		if (val)
1653 			static_key_enable(key);
1654 		else
1655 			static_key_disable(key);
1656 	}
1657 	mutex_unlock(&static_key_mutex);
1658 	return ret;
1659 }
1660 
1661 static struct ctl_table kern_table[] = {
1662 	{
1663 		.procname	= "sched_child_runs_first",
1664 		.data		= &sysctl_sched_child_runs_first,
1665 		.maxlen		= sizeof(unsigned int),
1666 		.mode		= 0644,
1667 		.proc_handler	= proc_dointvec,
1668 	},
1669 #ifdef CONFIG_SCHEDSTATS
1670 	{
1671 		.procname	= "sched_schedstats",
1672 		.data		= NULL,
1673 		.maxlen		= sizeof(unsigned int),
1674 		.mode		= 0644,
1675 		.proc_handler	= sysctl_schedstats,
1676 		.extra1		= SYSCTL_ZERO,
1677 		.extra2		= SYSCTL_ONE,
1678 	},
1679 #endif /* CONFIG_SCHEDSTATS */
1680 #ifdef CONFIG_TASK_DELAY_ACCT
1681 	{
1682 		.procname	= "task_delayacct",
1683 		.data		= NULL,
1684 		.maxlen		= sizeof(unsigned int),
1685 		.mode		= 0644,
1686 		.proc_handler	= sysctl_delayacct,
1687 		.extra1		= SYSCTL_ZERO,
1688 		.extra2		= SYSCTL_ONE,
1689 	},
1690 #endif /* CONFIG_TASK_DELAY_ACCT */
1691 #ifdef CONFIG_NUMA_BALANCING
1692 	{
1693 		.procname	= "numa_balancing",
1694 		.data		= NULL, /* filled in by handler */
1695 		.maxlen		= sizeof(unsigned int),
1696 		.mode		= 0644,
1697 		.proc_handler	= sysctl_numa_balancing,
1698 		.extra1		= SYSCTL_ZERO,
1699 		.extra2		= SYSCTL_FOUR,
1700 	},
1701 #endif /* CONFIG_NUMA_BALANCING */
1702 	{
1703 		.procname	= "sched_rt_period_us",
1704 		.data		= &sysctl_sched_rt_period,
1705 		.maxlen		= sizeof(unsigned int),
1706 		.mode		= 0644,
1707 		.proc_handler	= sched_rt_handler,
1708 	},
1709 	{
1710 		.procname	= "sched_rt_runtime_us",
1711 		.data		= &sysctl_sched_rt_runtime,
1712 		.maxlen		= sizeof(int),
1713 		.mode		= 0644,
1714 		.proc_handler	= sched_rt_handler,
1715 	},
1716 	{
1717 		.procname	= "sched_deadline_period_max_us",
1718 		.data		= &sysctl_sched_dl_period_max,
1719 		.maxlen		= sizeof(unsigned int),
1720 		.mode		= 0644,
1721 		.proc_handler	= proc_dointvec,
1722 	},
1723 	{
1724 		.procname	= "sched_deadline_period_min_us",
1725 		.data		= &sysctl_sched_dl_period_min,
1726 		.maxlen		= sizeof(unsigned int),
1727 		.mode		= 0644,
1728 		.proc_handler	= proc_dointvec,
1729 	},
1730 	{
1731 		.procname	= "sched_rr_timeslice_ms",
1732 		.data		= &sysctl_sched_rr_timeslice,
1733 		.maxlen		= sizeof(int),
1734 		.mode		= 0644,
1735 		.proc_handler	= sched_rr_handler,
1736 	},
1737 #ifdef CONFIG_UCLAMP_TASK
1738 	{
1739 		.procname	= "sched_util_clamp_min",
1740 		.data		= &sysctl_sched_uclamp_util_min,
1741 		.maxlen		= sizeof(unsigned int),
1742 		.mode		= 0644,
1743 		.proc_handler	= sysctl_sched_uclamp_handler,
1744 	},
1745 	{
1746 		.procname	= "sched_util_clamp_max",
1747 		.data		= &sysctl_sched_uclamp_util_max,
1748 		.maxlen		= sizeof(unsigned int),
1749 		.mode		= 0644,
1750 		.proc_handler	= sysctl_sched_uclamp_handler,
1751 	},
1752 	{
1753 		.procname	= "sched_util_clamp_min_rt_default",
1754 		.data		= &sysctl_sched_uclamp_util_min_rt_default,
1755 		.maxlen		= sizeof(unsigned int),
1756 		.mode		= 0644,
1757 		.proc_handler	= sysctl_sched_uclamp_handler,
1758 	},
1759 #endif
1760 #ifdef CONFIG_CFS_BANDWIDTH
1761 	{
1762 		.procname	= "sched_cfs_bandwidth_slice_us",
1763 		.data		= &sysctl_sched_cfs_bandwidth_slice,
1764 		.maxlen		= sizeof(unsigned int),
1765 		.mode		= 0644,
1766 		.proc_handler	= proc_dointvec_minmax,
1767 		.extra1		= SYSCTL_ONE,
1768 	},
1769 #endif
1770 #if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_CPU_FREQ_GOV_SCHEDUTIL)
1771 	{
1772 		.procname	= "sched_energy_aware",
1773 		.data		= &sysctl_sched_energy_aware,
1774 		.maxlen		= sizeof(unsigned int),
1775 		.mode		= 0644,
1776 		.proc_handler	= sched_energy_aware_handler,
1777 		.extra1		= SYSCTL_ZERO,
1778 		.extra2		= SYSCTL_ONE,
1779 	},
1780 #endif
1781 #ifdef CONFIG_PROVE_LOCKING
1782 	{
1783 		.procname	= "prove_locking",
1784 		.data		= &prove_locking,
1785 		.maxlen		= sizeof(int),
1786 		.mode		= 0644,
1787 		.proc_handler	= proc_dointvec,
1788 	},
1789 #endif
1790 #ifdef CONFIG_LOCK_STAT
1791 	{
1792 		.procname	= "lock_stat",
1793 		.data		= &lock_stat,
1794 		.maxlen		= sizeof(int),
1795 		.mode		= 0644,
1796 		.proc_handler	= proc_dointvec,
1797 	},
1798 #endif
1799 	{
1800 		.procname	= "panic",
1801 		.data		= &panic_timeout,
1802 		.maxlen		= sizeof(int),
1803 		.mode		= 0644,
1804 		.proc_handler	= proc_dointvec,
1805 	},
1806 #ifdef CONFIG_PROC_SYSCTL
1807 	{
1808 		.procname	= "tainted",
1809 		.maxlen 	= sizeof(long),
1810 		.mode		= 0644,
1811 		.proc_handler	= proc_taint,
1812 	},
1813 	{
1814 		.procname	= "sysctl_writes_strict",
1815 		.data		= &sysctl_writes_strict,
1816 		.maxlen		= sizeof(int),
1817 		.mode		= 0644,
1818 		.proc_handler	= proc_dointvec_minmax,
1819 		.extra1		= SYSCTL_NEG_ONE,
1820 		.extra2		= SYSCTL_ONE,
1821 	},
1822 #endif
1823 #ifdef CONFIG_LATENCYTOP
1824 	{
1825 		.procname	= "latencytop",
1826 		.data		= &latencytop_enabled,
1827 		.maxlen		= sizeof(int),
1828 		.mode		= 0644,
1829 		.proc_handler	= sysctl_latencytop,
1830 	},
1831 #endif
1832 #ifdef CONFIG_BLK_DEV_INITRD
1833 	{
1834 		.procname	= "real-root-dev",
1835 		.data		= &real_root_dev,
1836 		.maxlen		= sizeof(int),
1837 		.mode		= 0644,
1838 		.proc_handler	= proc_dointvec,
1839 	},
1840 #endif
1841 	{
1842 		.procname	= "print-fatal-signals",
1843 		.data		= &print_fatal_signals,
1844 		.maxlen		= sizeof(int),
1845 		.mode		= 0644,
1846 		.proc_handler	= proc_dointvec,
1847 	},
1848 #ifdef CONFIG_SPARC
1849 	{
1850 		.procname	= "reboot-cmd",
1851 		.data		= reboot_command,
1852 		.maxlen		= 256,
1853 		.mode		= 0644,
1854 		.proc_handler	= proc_dostring,
1855 	},
1856 	{
1857 		.procname	= "stop-a",
1858 		.data		= &stop_a_enabled,
1859 		.maxlen		= sizeof (int),
1860 		.mode		= 0644,
1861 		.proc_handler	= proc_dointvec,
1862 	},
1863 	{
1864 		.procname	= "scons-poweroff",
1865 		.data		= &scons_pwroff,
1866 		.maxlen		= sizeof (int),
1867 		.mode		= 0644,
1868 		.proc_handler	= proc_dointvec,
1869 	},
1870 #endif
1871 #ifdef CONFIG_SPARC64
1872 	{
1873 		.procname	= "tsb-ratio",
1874 		.data		= &sysctl_tsb_ratio,
1875 		.maxlen		= sizeof (int),
1876 		.mode		= 0644,
1877 		.proc_handler	= proc_dointvec,
1878 	},
1879 #endif
1880 #ifdef CONFIG_PARISC
1881 	{
1882 		.procname	= "soft-power",
1883 		.data		= &pwrsw_enabled,
1884 		.maxlen		= sizeof (int),
1885 		.mode		= 0644,
1886 		.proc_handler	= proc_dointvec,
1887 	},
1888 #endif
1889 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
1890 	{
1891 		.procname	= "unaligned-trap",
1892 		.data		= &unaligned_enabled,
1893 		.maxlen		= sizeof (int),
1894 		.mode		= 0644,
1895 		.proc_handler	= proc_dointvec,
1896 	},
1897 #endif
1898 	{
1899 		.procname	= "ctrl-alt-del",
1900 		.data		= &C_A_D,
1901 		.maxlen		= sizeof(int),
1902 		.mode		= 0644,
1903 		.proc_handler	= proc_dointvec,
1904 	},
1905 #ifdef CONFIG_FUNCTION_TRACER
1906 	{
1907 		.procname	= "ftrace_enabled",
1908 		.data		= &ftrace_enabled,
1909 		.maxlen		= sizeof(int),
1910 		.mode		= 0644,
1911 		.proc_handler	= ftrace_enable_sysctl,
1912 	},
1913 #endif
1914 #ifdef CONFIG_STACK_TRACER
1915 	{
1916 		.procname	= "stack_tracer_enabled",
1917 		.data		= &stack_tracer_enabled,
1918 		.maxlen		= sizeof(int),
1919 		.mode		= 0644,
1920 		.proc_handler	= stack_trace_sysctl,
1921 	},
1922 #endif
1923 #ifdef CONFIG_TRACING
1924 	{
1925 		.procname	= "ftrace_dump_on_oops",
1926 		.data		= &ftrace_dump_on_oops,
1927 		.maxlen		= sizeof(int),
1928 		.mode		= 0644,
1929 		.proc_handler	= proc_dointvec,
1930 	},
1931 	{
1932 		.procname	= "traceoff_on_warning",
1933 		.data		= &__disable_trace_on_warning,
1934 		.maxlen		= sizeof(__disable_trace_on_warning),
1935 		.mode		= 0644,
1936 		.proc_handler	= proc_dointvec,
1937 	},
1938 	{
1939 		.procname	= "tracepoint_printk",
1940 		.data		= &tracepoint_printk,
1941 		.maxlen		= sizeof(tracepoint_printk),
1942 		.mode		= 0644,
1943 		.proc_handler	= tracepoint_printk_sysctl,
1944 	},
1945 #endif
1946 #ifdef CONFIG_KEXEC_CORE
1947 	{
1948 		.procname	= "kexec_load_disabled",
1949 		.data		= &kexec_load_disabled,
1950 		.maxlen		= sizeof(int),
1951 		.mode		= 0644,
1952 		/* only handle a transition from default "0" to "1" */
1953 		.proc_handler	= proc_dointvec_minmax,
1954 		.extra1		= SYSCTL_ONE,
1955 		.extra2		= SYSCTL_ONE,
1956 	},
1957 #endif
1958 #ifdef CONFIG_MODULES
1959 	{
1960 		.procname	= "modprobe",
1961 		.data		= &modprobe_path,
1962 		.maxlen		= KMOD_PATH_LEN,
1963 		.mode		= 0644,
1964 		.proc_handler	= proc_dostring,
1965 	},
1966 	{
1967 		.procname	= "modules_disabled",
1968 		.data		= &modules_disabled,
1969 		.maxlen		= sizeof(int),
1970 		.mode		= 0644,
1971 		/* only handle a transition from default "0" to "1" */
1972 		.proc_handler	= proc_dointvec_minmax,
1973 		.extra1		= SYSCTL_ONE,
1974 		.extra2		= SYSCTL_ONE,
1975 	},
1976 #endif
1977 #ifdef CONFIG_UEVENT_HELPER
1978 	{
1979 		.procname	= "hotplug",
1980 		.data		= &uevent_helper,
1981 		.maxlen		= UEVENT_HELPER_PATH_LEN,
1982 		.mode		= 0644,
1983 		.proc_handler	= proc_dostring,
1984 	},
1985 #endif
1986 #ifdef CONFIG_BSD_PROCESS_ACCT
1987 	{
1988 		.procname	= "acct",
1989 		.data		= &acct_parm,
1990 		.maxlen		= 3*sizeof(int),
1991 		.mode		= 0644,
1992 		.proc_handler	= proc_dointvec,
1993 	},
1994 #endif
1995 #ifdef CONFIG_MAGIC_SYSRQ
1996 	{
1997 		.procname	= "sysrq",
1998 		.data		= NULL,
1999 		.maxlen		= sizeof (int),
2000 		.mode		= 0644,
2001 		.proc_handler	= sysrq_sysctl_handler,
2002 	},
2003 #endif
2004 #ifdef CONFIG_PROC_SYSCTL
2005 	{
2006 		.procname	= "cad_pid",
2007 		.data		= NULL,
2008 		.maxlen		= sizeof (int),
2009 		.mode		= 0600,
2010 		.proc_handler	= proc_do_cad_pid,
2011 	},
2012 #endif
2013 	{
2014 		.procname	= "threads-max",
2015 		.data		= NULL,
2016 		.maxlen		= sizeof(int),
2017 		.mode		= 0644,
2018 		.proc_handler	= sysctl_max_threads,
2019 	},
2020 	{
2021 		.procname	= "usermodehelper",
2022 		.mode		= 0555,
2023 		.child		= usermodehelper_table,
2024 	},
2025 	{
2026 		.procname	= "overflowuid",
2027 		.data		= &overflowuid,
2028 		.maxlen		= sizeof(int),
2029 		.mode		= 0644,
2030 		.proc_handler	= proc_dointvec_minmax,
2031 		.extra1		= SYSCTL_ZERO,
2032 		.extra2		= SYSCTL_MAXOLDUID,
2033 	},
2034 	{
2035 		.procname	= "overflowgid",
2036 		.data		= &overflowgid,
2037 		.maxlen		= sizeof(int),
2038 		.mode		= 0644,
2039 		.proc_handler	= proc_dointvec_minmax,
2040 		.extra1		= SYSCTL_ZERO,
2041 		.extra2		= SYSCTL_MAXOLDUID,
2042 	},
2043 #ifdef CONFIG_S390
2044 	{
2045 		.procname	= "userprocess_debug",
2046 		.data		= &show_unhandled_signals,
2047 		.maxlen		= sizeof(int),
2048 		.mode		= 0644,
2049 		.proc_handler	= proc_dointvec,
2050 	},
2051 #endif
2052 #ifdef CONFIG_SMP
2053 	{
2054 		.procname	= "oops_all_cpu_backtrace",
2055 		.data		= &sysctl_oops_all_cpu_backtrace,
2056 		.maxlen		= sizeof(int),
2057 		.mode		= 0644,
2058 		.proc_handler	= proc_dointvec_minmax,
2059 		.extra1		= SYSCTL_ZERO,
2060 		.extra2		= SYSCTL_ONE,
2061 	},
2062 #endif /* CONFIG_SMP */
2063 	{
2064 		.procname	= "pid_max",
2065 		.data		= &pid_max,
2066 		.maxlen		= sizeof (int),
2067 		.mode		= 0644,
2068 		.proc_handler	= proc_dointvec_minmax,
2069 		.extra1		= &pid_max_min,
2070 		.extra2		= &pid_max_max,
2071 	},
2072 	{
2073 		.procname	= "panic_on_oops",
2074 		.data		= &panic_on_oops,
2075 		.maxlen		= sizeof(int),
2076 		.mode		= 0644,
2077 		.proc_handler	= proc_dointvec,
2078 	},
2079 	{
2080 		.procname	= "panic_print",
2081 		.data		= &panic_print,
2082 		.maxlen		= sizeof(unsigned long),
2083 		.mode		= 0644,
2084 		.proc_handler	= proc_doulongvec_minmax,
2085 	},
2086 	{
2087 		.procname	= "ngroups_max",
2088 		.data		= (void *)&ngroups_max,
2089 		.maxlen		= sizeof (int),
2090 		.mode		= 0444,
2091 		.proc_handler	= proc_dointvec,
2092 	},
2093 	{
2094 		.procname	= "cap_last_cap",
2095 		.data		= (void *)&cap_last_cap,
2096 		.maxlen		= sizeof(int),
2097 		.mode		= 0444,
2098 		.proc_handler	= proc_dointvec,
2099 	},
2100 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
2101 	{
2102 		.procname       = "unknown_nmi_panic",
2103 		.data           = &unknown_nmi_panic,
2104 		.maxlen         = sizeof (int),
2105 		.mode           = 0644,
2106 		.proc_handler   = proc_dointvec,
2107 	},
2108 #endif
2109 
2110 #if (defined(CONFIG_X86_32) || defined(CONFIG_PARISC)) && \
2111 	defined(CONFIG_DEBUG_STACKOVERFLOW)
2112 	{
2113 		.procname	= "panic_on_stackoverflow",
2114 		.data		= &sysctl_panic_on_stackoverflow,
2115 		.maxlen		= sizeof(int),
2116 		.mode		= 0644,
2117 		.proc_handler	= proc_dointvec,
2118 	},
2119 #endif
2120 #if defined(CONFIG_X86)
2121 	{
2122 		.procname	= "panic_on_unrecovered_nmi",
2123 		.data		= &panic_on_unrecovered_nmi,
2124 		.maxlen		= sizeof(int),
2125 		.mode		= 0644,
2126 		.proc_handler	= proc_dointvec,
2127 	},
2128 	{
2129 		.procname	= "panic_on_io_nmi",
2130 		.data		= &panic_on_io_nmi,
2131 		.maxlen		= sizeof(int),
2132 		.mode		= 0644,
2133 		.proc_handler	= proc_dointvec,
2134 	},
2135 	{
2136 		.procname	= "bootloader_type",
2137 		.data		= &bootloader_type,
2138 		.maxlen		= sizeof (int),
2139 		.mode		= 0444,
2140 		.proc_handler	= proc_dointvec,
2141 	},
2142 	{
2143 		.procname	= "bootloader_version",
2144 		.data		= &bootloader_version,
2145 		.maxlen		= sizeof (int),
2146 		.mode		= 0444,
2147 		.proc_handler	= proc_dointvec,
2148 	},
2149 	{
2150 		.procname	= "io_delay_type",
2151 		.data		= &io_delay_type,
2152 		.maxlen		= sizeof(int),
2153 		.mode		= 0644,
2154 		.proc_handler	= proc_dointvec,
2155 	},
2156 #endif
2157 #if defined(CONFIG_MMU)
2158 	{
2159 		.procname	= "randomize_va_space",
2160 		.data		= &randomize_va_space,
2161 		.maxlen		= sizeof(int),
2162 		.mode		= 0644,
2163 		.proc_handler	= proc_dointvec,
2164 	},
2165 #endif
2166 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
2167 	{
2168 		.procname	= "spin_retry",
2169 		.data		= &spin_retry,
2170 		.maxlen		= sizeof (int),
2171 		.mode		= 0644,
2172 		.proc_handler	= proc_dointvec,
2173 	},
2174 #endif
2175 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
2176 	{
2177 		.procname	= "acpi_video_flags",
2178 		.data		= &acpi_realmode_flags,
2179 		.maxlen		= sizeof (unsigned long),
2180 		.mode		= 0644,
2181 		.proc_handler	= proc_doulongvec_minmax,
2182 	},
2183 #endif
2184 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
2185 	{
2186 		.procname	= "ignore-unaligned-usertrap",
2187 		.data		= &no_unaligned_warning,
2188 		.maxlen		= sizeof (int),
2189 		.mode		= 0644,
2190 		.proc_handler	= proc_dointvec,
2191 	},
2192 #endif
2193 #ifdef CONFIG_IA64
2194 	{
2195 		.procname	= "unaligned-dump-stack",
2196 		.data		= &unaligned_dump_stack,
2197 		.maxlen		= sizeof (int),
2198 		.mode		= 0644,
2199 		.proc_handler	= proc_dointvec,
2200 	},
2201 #endif
2202 #ifdef CONFIG_RT_MUTEXES
2203 	{
2204 		.procname	= "max_lock_depth",
2205 		.data		= &max_lock_depth,
2206 		.maxlen		= sizeof(int),
2207 		.mode		= 0644,
2208 		.proc_handler	= proc_dointvec,
2209 	},
2210 #endif
2211 	{
2212 		.procname	= "poweroff_cmd",
2213 		.data		= &poweroff_cmd,
2214 		.maxlen		= POWEROFF_CMD_PATH_LEN,
2215 		.mode		= 0644,
2216 		.proc_handler	= proc_dostring,
2217 	},
2218 #ifdef CONFIG_KEYS
2219 	{
2220 		.procname	= "keys",
2221 		.mode		= 0555,
2222 		.child		= key_sysctls,
2223 	},
2224 #endif
2225 #ifdef CONFIG_PERF_EVENTS
2226 	/*
2227 	 * User-space scripts rely on the existence of this file
2228 	 * as a feature check for perf_events being enabled.
2229 	 *
2230 	 * So it's an ABI, do not remove!
2231 	 */
2232 	{
2233 		.procname	= "perf_event_paranoid",
2234 		.data		= &sysctl_perf_event_paranoid,
2235 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
2236 		.mode		= 0644,
2237 		.proc_handler	= proc_dointvec,
2238 	},
2239 	{
2240 		.procname	= "perf_event_mlock_kb",
2241 		.data		= &sysctl_perf_event_mlock,
2242 		.maxlen		= sizeof(sysctl_perf_event_mlock),
2243 		.mode		= 0644,
2244 		.proc_handler	= proc_dointvec,
2245 	},
2246 	{
2247 		.procname	= "perf_event_max_sample_rate",
2248 		.data		= &sysctl_perf_event_sample_rate,
2249 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
2250 		.mode		= 0644,
2251 		.proc_handler	= perf_proc_update_handler,
2252 		.extra1		= SYSCTL_ONE,
2253 	},
2254 	{
2255 		.procname	= "perf_cpu_time_max_percent",
2256 		.data		= &sysctl_perf_cpu_time_max_percent,
2257 		.maxlen		= sizeof(sysctl_perf_cpu_time_max_percent),
2258 		.mode		= 0644,
2259 		.proc_handler	= perf_cpu_time_max_percent_handler,
2260 		.extra1		= SYSCTL_ZERO,
2261 		.extra2		= SYSCTL_ONE_HUNDRED,
2262 	},
2263 	{
2264 		.procname	= "perf_event_max_stack",
2265 		.data		= &sysctl_perf_event_max_stack,
2266 		.maxlen		= sizeof(sysctl_perf_event_max_stack),
2267 		.mode		= 0644,
2268 		.proc_handler	= perf_event_max_stack_handler,
2269 		.extra1		= SYSCTL_ZERO,
2270 		.extra2		= (void *)&six_hundred_forty_kb,
2271 	},
2272 	{
2273 		.procname	= "perf_event_max_contexts_per_stack",
2274 		.data		= &sysctl_perf_event_max_contexts_per_stack,
2275 		.maxlen		= sizeof(sysctl_perf_event_max_contexts_per_stack),
2276 		.mode		= 0644,
2277 		.proc_handler	= perf_event_max_stack_handler,
2278 		.extra1		= SYSCTL_ZERO,
2279 		.extra2		= SYSCTL_ONE_THOUSAND,
2280 	},
2281 #endif
2282 	{
2283 		.procname	= "panic_on_warn",
2284 		.data		= &panic_on_warn,
2285 		.maxlen		= sizeof(int),
2286 		.mode		= 0644,
2287 		.proc_handler	= proc_dointvec_minmax,
2288 		.extra1		= SYSCTL_ZERO,
2289 		.extra2		= SYSCTL_ONE,
2290 	},
2291 #ifdef CONFIG_BPF_SYSCALL
2292 	{
2293 		.procname	= "unprivileged_bpf_disabled",
2294 		.data		= &sysctl_unprivileged_bpf_disabled,
2295 		.maxlen		= sizeof(sysctl_unprivileged_bpf_disabled),
2296 		.mode		= 0644,
2297 		.proc_handler	= bpf_unpriv_handler,
2298 		.extra1		= SYSCTL_ZERO,
2299 		.extra2		= SYSCTL_TWO,
2300 	},
2301 	{
2302 		.procname	= "bpf_stats_enabled",
2303 		.data		= &bpf_stats_enabled_key.key,
2304 		.maxlen		= sizeof(bpf_stats_enabled_key),
2305 		.mode		= 0644,
2306 		.proc_handler	= bpf_stats_handler,
2307 	},
2308 #endif
2309 #if defined(CONFIG_TREE_RCU)
2310 	{
2311 		.procname	= "panic_on_rcu_stall",
2312 		.data		= &sysctl_panic_on_rcu_stall,
2313 		.maxlen		= sizeof(sysctl_panic_on_rcu_stall),
2314 		.mode		= 0644,
2315 		.proc_handler	= proc_dointvec_minmax,
2316 		.extra1		= SYSCTL_ZERO,
2317 		.extra2		= SYSCTL_ONE,
2318 	},
2319 #endif
2320 #if defined(CONFIG_TREE_RCU)
2321 	{
2322 		.procname	= "max_rcu_stall_to_panic",
2323 		.data		= &sysctl_max_rcu_stall_to_panic,
2324 		.maxlen		= sizeof(sysctl_max_rcu_stall_to_panic),
2325 		.mode		= 0644,
2326 		.proc_handler	= proc_dointvec_minmax,
2327 		.extra1		= SYSCTL_ONE,
2328 		.extra2		= SYSCTL_INT_MAX,
2329 	},
2330 #endif
2331 	{ }
2332 };
2333 
2334 static struct ctl_table vm_table[] = {
2335 	{
2336 		.procname	= "overcommit_memory",
2337 		.data		= &sysctl_overcommit_memory,
2338 		.maxlen		= sizeof(sysctl_overcommit_memory),
2339 		.mode		= 0644,
2340 		.proc_handler	= overcommit_policy_handler,
2341 		.extra1		= SYSCTL_ZERO,
2342 		.extra2		= SYSCTL_TWO,
2343 	},
2344 	{
2345 		.procname	= "panic_on_oom",
2346 		.data		= &sysctl_panic_on_oom,
2347 		.maxlen		= sizeof(sysctl_panic_on_oom),
2348 		.mode		= 0644,
2349 		.proc_handler	= proc_dointvec_minmax,
2350 		.extra1		= SYSCTL_ZERO,
2351 		.extra2		= SYSCTL_TWO,
2352 	},
2353 	{
2354 		.procname	= "oom_kill_allocating_task",
2355 		.data		= &sysctl_oom_kill_allocating_task,
2356 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
2357 		.mode		= 0644,
2358 		.proc_handler	= proc_dointvec,
2359 	},
2360 	{
2361 		.procname	= "oom_dump_tasks",
2362 		.data		= &sysctl_oom_dump_tasks,
2363 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
2364 		.mode		= 0644,
2365 		.proc_handler	= proc_dointvec,
2366 	},
2367 	{
2368 		.procname	= "overcommit_ratio",
2369 		.data		= &sysctl_overcommit_ratio,
2370 		.maxlen		= sizeof(sysctl_overcommit_ratio),
2371 		.mode		= 0644,
2372 		.proc_handler	= overcommit_ratio_handler,
2373 	},
2374 	{
2375 		.procname	= "overcommit_kbytes",
2376 		.data		= &sysctl_overcommit_kbytes,
2377 		.maxlen		= sizeof(sysctl_overcommit_kbytes),
2378 		.mode		= 0644,
2379 		.proc_handler	= overcommit_kbytes_handler,
2380 	},
2381 	{
2382 		.procname	= "page-cluster",
2383 		.data		= &page_cluster,
2384 		.maxlen		= sizeof(int),
2385 		.mode		= 0644,
2386 		.proc_handler	= proc_dointvec_minmax,
2387 		.extra1		= SYSCTL_ZERO,
2388 	},
2389 	{
2390 		.procname	= "dirty_background_ratio",
2391 		.data		= &dirty_background_ratio,
2392 		.maxlen		= sizeof(dirty_background_ratio),
2393 		.mode		= 0644,
2394 		.proc_handler	= dirty_background_ratio_handler,
2395 		.extra1		= SYSCTL_ZERO,
2396 		.extra2		= SYSCTL_ONE_HUNDRED,
2397 	},
2398 	{
2399 		.procname	= "dirty_background_bytes",
2400 		.data		= &dirty_background_bytes,
2401 		.maxlen		= sizeof(dirty_background_bytes),
2402 		.mode		= 0644,
2403 		.proc_handler	= dirty_background_bytes_handler,
2404 		.extra1		= SYSCTL_LONG_ONE,
2405 	},
2406 	{
2407 		.procname	= "dirty_ratio",
2408 		.data		= &vm_dirty_ratio,
2409 		.maxlen		= sizeof(vm_dirty_ratio),
2410 		.mode		= 0644,
2411 		.proc_handler	= dirty_ratio_handler,
2412 		.extra1		= SYSCTL_ZERO,
2413 		.extra2		= SYSCTL_ONE_HUNDRED,
2414 	},
2415 	{
2416 		.procname	= "dirty_bytes",
2417 		.data		= &vm_dirty_bytes,
2418 		.maxlen		= sizeof(vm_dirty_bytes),
2419 		.mode		= 0644,
2420 		.proc_handler	= dirty_bytes_handler,
2421 		.extra1		= (void *)&dirty_bytes_min,
2422 	},
2423 	{
2424 		.procname	= "dirty_writeback_centisecs",
2425 		.data		= &dirty_writeback_interval,
2426 		.maxlen		= sizeof(dirty_writeback_interval),
2427 		.mode		= 0644,
2428 		.proc_handler	= dirty_writeback_centisecs_handler,
2429 	},
2430 	{
2431 		.procname	= "dirty_expire_centisecs",
2432 		.data		= &dirty_expire_interval,
2433 		.maxlen		= sizeof(dirty_expire_interval),
2434 		.mode		= 0644,
2435 		.proc_handler	= proc_dointvec_minmax,
2436 		.extra1		= SYSCTL_ZERO,
2437 	},
2438 	{
2439 		.procname	= "dirtytime_expire_seconds",
2440 		.data		= &dirtytime_expire_interval,
2441 		.maxlen		= sizeof(dirtytime_expire_interval),
2442 		.mode		= 0644,
2443 		.proc_handler	= dirtytime_interval_handler,
2444 		.extra1		= SYSCTL_ZERO,
2445 	},
2446 	{
2447 		.procname	= "swappiness",
2448 		.data		= &vm_swappiness,
2449 		.maxlen		= sizeof(vm_swappiness),
2450 		.mode		= 0644,
2451 		.proc_handler	= proc_dointvec_minmax,
2452 		.extra1		= SYSCTL_ZERO,
2453 		.extra2		= SYSCTL_TWO_HUNDRED,
2454 	},
2455 #ifdef CONFIG_HUGETLB_PAGE
2456 	{
2457 		.procname	= "nr_hugepages",
2458 		.data		= NULL,
2459 		.maxlen		= sizeof(unsigned long),
2460 		.mode		= 0644,
2461 		.proc_handler	= hugetlb_sysctl_handler,
2462 	},
2463 #ifdef CONFIG_NUMA
2464 	{
2465 		.procname       = "nr_hugepages_mempolicy",
2466 		.data           = NULL,
2467 		.maxlen         = sizeof(unsigned long),
2468 		.mode           = 0644,
2469 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
2470 	},
2471 	{
2472 		.procname		= "numa_stat",
2473 		.data			= &sysctl_vm_numa_stat,
2474 		.maxlen			= sizeof(int),
2475 		.mode			= 0644,
2476 		.proc_handler	= sysctl_vm_numa_stat_handler,
2477 		.extra1			= SYSCTL_ZERO,
2478 		.extra2			= SYSCTL_ONE,
2479 	},
2480 #endif
2481 	 {
2482 		.procname	= "hugetlb_shm_group",
2483 		.data		= &sysctl_hugetlb_shm_group,
2484 		.maxlen		= sizeof(gid_t),
2485 		.mode		= 0644,
2486 		.proc_handler	= proc_dointvec,
2487 	 },
2488 	{
2489 		.procname	= "nr_overcommit_hugepages",
2490 		.data		= NULL,
2491 		.maxlen		= sizeof(unsigned long),
2492 		.mode		= 0644,
2493 		.proc_handler	= hugetlb_overcommit_handler,
2494 	},
2495 #endif
2496 	{
2497 		.procname	= "lowmem_reserve_ratio",
2498 		.data		= &sysctl_lowmem_reserve_ratio,
2499 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
2500 		.mode		= 0644,
2501 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
2502 	},
2503 	{
2504 		.procname	= "drop_caches",
2505 		.data		= &sysctl_drop_caches,
2506 		.maxlen		= sizeof(int),
2507 		.mode		= 0200,
2508 		.proc_handler	= drop_caches_sysctl_handler,
2509 		.extra1		= SYSCTL_ONE,
2510 		.extra2		= SYSCTL_FOUR,
2511 	},
2512 #ifdef CONFIG_COMPACTION
2513 	{
2514 		.procname	= "compact_memory",
2515 		.data		= NULL,
2516 		.maxlen		= sizeof(int),
2517 		.mode		= 0200,
2518 		.proc_handler	= sysctl_compaction_handler,
2519 	},
2520 	{
2521 		.procname	= "compaction_proactiveness",
2522 		.data		= &sysctl_compaction_proactiveness,
2523 		.maxlen		= sizeof(sysctl_compaction_proactiveness),
2524 		.mode		= 0644,
2525 		.proc_handler	= compaction_proactiveness_sysctl_handler,
2526 		.extra1		= SYSCTL_ZERO,
2527 		.extra2		= SYSCTL_ONE_HUNDRED,
2528 	},
2529 	{
2530 		.procname	= "extfrag_threshold",
2531 		.data		= &sysctl_extfrag_threshold,
2532 		.maxlen		= sizeof(int),
2533 		.mode		= 0644,
2534 		.proc_handler	= proc_dointvec_minmax,
2535 		.extra1		= SYSCTL_ZERO,
2536 		.extra2		= (void *)&max_extfrag_threshold,
2537 	},
2538 	{
2539 		.procname	= "compact_unevictable_allowed",
2540 		.data		= &sysctl_compact_unevictable_allowed,
2541 		.maxlen		= sizeof(int),
2542 		.mode		= 0644,
2543 		.proc_handler	= proc_dointvec_minmax_warn_RT_change,
2544 		.extra1		= SYSCTL_ZERO,
2545 		.extra2		= SYSCTL_ONE,
2546 	},
2547 
2548 #endif /* CONFIG_COMPACTION */
2549 	{
2550 		.procname	= "min_free_kbytes",
2551 		.data		= &min_free_kbytes,
2552 		.maxlen		= sizeof(min_free_kbytes),
2553 		.mode		= 0644,
2554 		.proc_handler	= min_free_kbytes_sysctl_handler,
2555 		.extra1		= SYSCTL_ZERO,
2556 	},
2557 	{
2558 		.procname	= "watermark_boost_factor",
2559 		.data		= &watermark_boost_factor,
2560 		.maxlen		= sizeof(watermark_boost_factor),
2561 		.mode		= 0644,
2562 		.proc_handler	= proc_dointvec_minmax,
2563 		.extra1		= SYSCTL_ZERO,
2564 	},
2565 	{
2566 		.procname	= "watermark_scale_factor",
2567 		.data		= &watermark_scale_factor,
2568 		.maxlen		= sizeof(watermark_scale_factor),
2569 		.mode		= 0644,
2570 		.proc_handler	= watermark_scale_factor_sysctl_handler,
2571 		.extra1		= SYSCTL_ONE,
2572 		.extra2		= SYSCTL_THREE_THOUSAND,
2573 	},
2574 	{
2575 		.procname	= "percpu_pagelist_high_fraction",
2576 		.data		= &percpu_pagelist_high_fraction,
2577 		.maxlen		= sizeof(percpu_pagelist_high_fraction),
2578 		.mode		= 0644,
2579 		.proc_handler	= percpu_pagelist_high_fraction_sysctl_handler,
2580 		.extra1		= SYSCTL_ZERO,
2581 	},
2582 	{
2583 		.procname	= "page_lock_unfairness",
2584 		.data		= &sysctl_page_lock_unfairness,
2585 		.maxlen		= sizeof(sysctl_page_lock_unfairness),
2586 		.mode		= 0644,
2587 		.proc_handler	= proc_dointvec_minmax,
2588 		.extra1		= SYSCTL_ZERO,
2589 	},
2590 #ifdef CONFIG_MMU
2591 	{
2592 		.procname	= "max_map_count",
2593 		.data		= &sysctl_max_map_count,
2594 		.maxlen		= sizeof(sysctl_max_map_count),
2595 		.mode		= 0644,
2596 		.proc_handler	= proc_dointvec_minmax,
2597 		.extra1		= SYSCTL_ZERO,
2598 	},
2599 #else
2600 	{
2601 		.procname	= "nr_trim_pages",
2602 		.data		= &sysctl_nr_trim_pages,
2603 		.maxlen		= sizeof(sysctl_nr_trim_pages),
2604 		.mode		= 0644,
2605 		.proc_handler	= proc_dointvec_minmax,
2606 		.extra1		= SYSCTL_ZERO,
2607 	},
2608 #endif
2609 	{
2610 		.procname	= "laptop_mode",
2611 		.data		= &laptop_mode,
2612 		.maxlen		= sizeof(laptop_mode),
2613 		.mode		= 0644,
2614 		.proc_handler	= proc_dointvec_jiffies,
2615 	},
2616 	{
2617 		.procname	= "vfs_cache_pressure",
2618 		.data		= &sysctl_vfs_cache_pressure,
2619 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
2620 		.mode		= 0644,
2621 		.proc_handler	= proc_dointvec_minmax,
2622 		.extra1		= SYSCTL_ZERO,
2623 	},
2624 #if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
2625     defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
2626 	{
2627 		.procname	= "legacy_va_layout",
2628 		.data		= &sysctl_legacy_va_layout,
2629 		.maxlen		= sizeof(sysctl_legacy_va_layout),
2630 		.mode		= 0644,
2631 		.proc_handler	= proc_dointvec_minmax,
2632 		.extra1		= SYSCTL_ZERO,
2633 	},
2634 #endif
2635 #ifdef CONFIG_NUMA
2636 	{
2637 		.procname	= "zone_reclaim_mode",
2638 		.data		= &node_reclaim_mode,
2639 		.maxlen		= sizeof(node_reclaim_mode),
2640 		.mode		= 0644,
2641 		.proc_handler	= proc_dointvec_minmax,
2642 		.extra1		= SYSCTL_ZERO,
2643 	},
2644 	{
2645 		.procname	= "min_unmapped_ratio",
2646 		.data		= &sysctl_min_unmapped_ratio,
2647 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
2648 		.mode		= 0644,
2649 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
2650 		.extra1		= SYSCTL_ZERO,
2651 		.extra2		= SYSCTL_ONE_HUNDRED,
2652 	},
2653 	{
2654 		.procname	= "min_slab_ratio",
2655 		.data		= &sysctl_min_slab_ratio,
2656 		.maxlen		= sizeof(sysctl_min_slab_ratio),
2657 		.mode		= 0644,
2658 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
2659 		.extra1		= SYSCTL_ZERO,
2660 		.extra2		= SYSCTL_ONE_HUNDRED,
2661 	},
2662 #endif
2663 #ifdef CONFIG_SMP
2664 	{
2665 		.procname	= "stat_interval",
2666 		.data		= &sysctl_stat_interval,
2667 		.maxlen		= sizeof(sysctl_stat_interval),
2668 		.mode		= 0644,
2669 		.proc_handler	= proc_dointvec_jiffies,
2670 	},
2671 	{
2672 		.procname	= "stat_refresh",
2673 		.data		= NULL,
2674 		.maxlen		= 0,
2675 		.mode		= 0600,
2676 		.proc_handler	= vmstat_refresh,
2677 	},
2678 #endif
2679 #ifdef CONFIG_MMU
2680 	{
2681 		.procname	= "mmap_min_addr",
2682 		.data		= &dac_mmap_min_addr,
2683 		.maxlen		= sizeof(unsigned long),
2684 		.mode		= 0644,
2685 		.proc_handler	= mmap_min_addr_handler,
2686 	},
2687 #endif
2688 #ifdef CONFIG_NUMA
2689 	{
2690 		.procname	= "numa_zonelist_order",
2691 		.data		= &numa_zonelist_order,
2692 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
2693 		.mode		= 0644,
2694 		.proc_handler	= numa_zonelist_order_handler,
2695 	},
2696 #endif
2697 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
2698    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
2699 	{
2700 		.procname	= "vdso_enabled",
2701 #ifdef CONFIG_X86_32
2702 		.data		= &vdso32_enabled,
2703 		.maxlen		= sizeof(vdso32_enabled),
2704 #else
2705 		.data		= &vdso_enabled,
2706 		.maxlen		= sizeof(vdso_enabled),
2707 #endif
2708 		.mode		= 0644,
2709 		.proc_handler	= proc_dointvec,
2710 		.extra1		= SYSCTL_ZERO,
2711 	},
2712 #endif
2713 #ifdef CONFIG_HIGHMEM
2714 	{
2715 		.procname	= "highmem_is_dirtyable",
2716 		.data		= &vm_highmem_is_dirtyable,
2717 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
2718 		.mode		= 0644,
2719 		.proc_handler	= proc_dointvec_minmax,
2720 		.extra1		= SYSCTL_ZERO,
2721 		.extra2		= SYSCTL_ONE,
2722 	},
2723 #endif
2724 #ifdef CONFIG_MEMORY_FAILURE
2725 	{
2726 		.procname	= "memory_failure_early_kill",
2727 		.data		= &sysctl_memory_failure_early_kill,
2728 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
2729 		.mode		= 0644,
2730 		.proc_handler	= proc_dointvec_minmax,
2731 		.extra1		= SYSCTL_ZERO,
2732 		.extra2		= SYSCTL_ONE,
2733 	},
2734 	{
2735 		.procname	= "memory_failure_recovery",
2736 		.data		= &sysctl_memory_failure_recovery,
2737 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
2738 		.mode		= 0644,
2739 		.proc_handler	= proc_dointvec_minmax,
2740 		.extra1		= SYSCTL_ZERO,
2741 		.extra2		= SYSCTL_ONE,
2742 	},
2743 #endif
2744 	{
2745 		.procname	= "user_reserve_kbytes",
2746 		.data		= &sysctl_user_reserve_kbytes,
2747 		.maxlen		= sizeof(sysctl_user_reserve_kbytes),
2748 		.mode		= 0644,
2749 		.proc_handler	= proc_doulongvec_minmax,
2750 	},
2751 	{
2752 		.procname	= "admin_reserve_kbytes",
2753 		.data		= &sysctl_admin_reserve_kbytes,
2754 		.maxlen		= sizeof(sysctl_admin_reserve_kbytes),
2755 		.mode		= 0644,
2756 		.proc_handler	= proc_doulongvec_minmax,
2757 	},
2758 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
2759 	{
2760 		.procname	= "mmap_rnd_bits",
2761 		.data		= &mmap_rnd_bits,
2762 		.maxlen		= sizeof(mmap_rnd_bits),
2763 		.mode		= 0600,
2764 		.proc_handler	= proc_dointvec_minmax,
2765 		.extra1		= (void *)&mmap_rnd_bits_min,
2766 		.extra2		= (void *)&mmap_rnd_bits_max,
2767 	},
2768 #endif
2769 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
2770 	{
2771 		.procname	= "mmap_rnd_compat_bits",
2772 		.data		= &mmap_rnd_compat_bits,
2773 		.maxlen		= sizeof(mmap_rnd_compat_bits),
2774 		.mode		= 0600,
2775 		.proc_handler	= proc_dointvec_minmax,
2776 		.extra1		= (void *)&mmap_rnd_compat_bits_min,
2777 		.extra2		= (void *)&mmap_rnd_compat_bits_max,
2778 	},
2779 #endif
2780 #ifdef CONFIG_USERFAULTFD
2781 	{
2782 		.procname	= "unprivileged_userfaultfd",
2783 		.data		= &sysctl_unprivileged_userfaultfd,
2784 		.maxlen		= sizeof(sysctl_unprivileged_userfaultfd),
2785 		.mode		= 0644,
2786 		.proc_handler	= proc_dointvec_minmax,
2787 		.extra1		= SYSCTL_ZERO,
2788 		.extra2		= SYSCTL_ONE,
2789 	},
2790 #endif
2791 	{ }
2792 };
2793 
2794 static struct ctl_table debug_table[] = {
2795 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
2796 	{
2797 		.procname	= "exception-trace",
2798 		.data		= &show_unhandled_signals,
2799 		.maxlen		= sizeof(int),
2800 		.mode		= 0644,
2801 		.proc_handler	= proc_dointvec
2802 	},
2803 #endif
2804 	{ }
2805 };
2806 
2807 static struct ctl_table dev_table[] = {
2808 	{ }
2809 };
2810 
2811 DECLARE_SYSCTL_BASE(kernel, kern_table);
2812 DECLARE_SYSCTL_BASE(vm, vm_table);
2813 DECLARE_SYSCTL_BASE(debug, debug_table);
2814 DECLARE_SYSCTL_BASE(dev, dev_table);
2815 
2816 int __init sysctl_init_bases(void)
2817 {
2818 	register_sysctl_base(kernel);
2819 	register_sysctl_base(vm);
2820 	register_sysctl_base(debug);
2821 	register_sysctl_base(dev);
2822 
2823 	return 0;
2824 }
2825 #endif /* CONFIG_SYSCTL */
2826 /*
2827  * No sense putting this after each symbol definition, twice,
2828  * exception granted :-)
2829  */
2830 EXPORT_SYMBOL(proc_dobool);
2831 EXPORT_SYMBOL(proc_dointvec);
2832 EXPORT_SYMBOL(proc_douintvec);
2833 EXPORT_SYMBOL(proc_dointvec_jiffies);
2834 EXPORT_SYMBOL(proc_dointvec_minmax);
2835 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
2836 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2837 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2838 EXPORT_SYMBOL(proc_dostring);
2839 EXPORT_SYMBOL(proc_doulongvec_minmax);
2840 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2841 EXPORT_SYMBOL(proc_do_large_bitmap);
2842