xref: /openbmc/linux/kernel/sysctl.c (revision 52cdded0)
1 // SPDX-License-Identifier: GPL-2.0-only
2 /*
3  * sysctl.c: General linux system control interface
4  *
5  * Begun 24 March 1995, Stephen Tweedie
6  * Added /proc support, Dec 1995
7  * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
8  * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
9  * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
10  * Dynamic registration fixes, Stephen Tweedie.
11  * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
12  * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
13  *  Horn.
14  * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
15  * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
16  * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
17  *  Wendling.
18  * The list_for_each() macro wasn't appropriate for the sysctl loop.
19  *  Removed it and replaced it with older style, 03/23/00, Bill Wendling
20  */
21 
22 #include <linux/module.h>
23 #include <linux/aio.h>
24 #include <linux/mm.h>
25 #include <linux/swap.h>
26 #include <linux/slab.h>
27 #include <linux/sysctl.h>
28 #include <linux/bitmap.h>
29 #include <linux/signal.h>
30 #include <linux/printk.h>
31 #include <linux/proc_fs.h>
32 #include <linux/security.h>
33 #include <linux/ctype.h>
34 #include <linux/kmemleak.h>
35 #include <linux/fs.h>
36 #include <linux/init.h>
37 #include <linux/kernel.h>
38 #include <linux/kobject.h>
39 #include <linux/net.h>
40 #include <linux/sysrq.h>
41 #include <linux/highuid.h>
42 #include <linux/writeback.h>
43 #include <linux/ratelimit.h>
44 #include <linux/compaction.h>
45 #include <linux/hugetlb.h>
46 #include <linux/initrd.h>
47 #include <linux/key.h>
48 #include <linux/times.h>
49 #include <linux/limits.h>
50 #include <linux/dcache.h>
51 #include <linux/dnotify.h>
52 #include <linux/syscalls.h>
53 #include <linux/vmstat.h>
54 #include <linux/nfs_fs.h>
55 #include <linux/acpi.h>
56 #include <linux/reboot.h>
57 #include <linux/ftrace.h>
58 #include <linux/perf_event.h>
59 #include <linux/kprobes.h>
60 #include <linux/pipe_fs_i.h>
61 #include <linux/oom.h>
62 #include <linux/kmod.h>
63 #include <linux/capability.h>
64 #include <linux/binfmts.h>
65 #include <linux/sched/sysctl.h>
66 #include <linux/sched/coredump.h>
67 #include <linux/kexec.h>
68 #include <linux/bpf.h>
69 #include <linux/mount.h>
70 #include <linux/userfaultfd_k.h>
71 #include <linux/coredump.h>
72 #include <linux/latencytop.h>
73 #include <linux/pid.h>
74 
75 #include "../lib/kstrtox.h"
76 
77 #include <linux/uaccess.h>
78 #include <asm/processor.h>
79 
80 #ifdef CONFIG_X86
81 #include <asm/nmi.h>
82 #include <asm/stacktrace.h>
83 #include <asm/io.h>
84 #endif
85 #ifdef CONFIG_SPARC
86 #include <asm/setup.h>
87 #endif
88 #ifdef CONFIG_BSD_PROCESS_ACCT
89 #include <linux/acct.h>
90 #endif
91 #ifdef CONFIG_RT_MUTEXES
92 #include <linux/rtmutex.h>
93 #endif
94 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
95 #include <linux/lockdep.h>
96 #endif
97 #ifdef CONFIG_CHR_DEV_SG
98 #include <scsi/sg.h>
99 #endif
100 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
101 #include <linux/stackleak.h>
102 #endif
103 #ifdef CONFIG_LOCKUP_DETECTOR
104 #include <linux/nmi.h>
105 #endif
106 
107 #if defined(CONFIG_SYSCTL)
108 
109 /* Constants used for minimum and  maximum */
110 #ifdef CONFIG_LOCKUP_DETECTOR
111 static int sixty = 60;
112 #endif
113 
114 static int __maybe_unused neg_one = -1;
115 static int __maybe_unused two = 2;
116 static int __maybe_unused four = 4;
117 static unsigned long zero_ul;
118 static unsigned long one_ul = 1;
119 static unsigned long long_max = LONG_MAX;
120 static int one_hundred = 100;
121 static int two_hundred = 200;
122 static int one_thousand = 1000;
123 #ifdef CONFIG_PRINTK
124 static int ten_thousand = 10000;
125 #endif
126 #ifdef CONFIG_PERF_EVENTS
127 static int six_hundred_forty_kb = 640 * 1024;
128 #endif
129 
130 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
131 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
132 
133 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
134 static int maxolduid = 65535;
135 static int minolduid;
136 
137 static int ngroups_max = NGROUPS_MAX;
138 static const int cap_last_cap = CAP_LAST_CAP;
139 
140 /*
141  * This is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs
142  * and hung_task_check_interval_secs
143  */
144 #ifdef CONFIG_DETECT_HUNG_TASK
145 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
146 #endif
147 
148 #ifdef CONFIG_INOTIFY_USER
149 #include <linux/inotify.h>
150 #endif
151 
152 #ifdef CONFIG_PROC_SYSCTL
153 
154 /**
155  * enum sysctl_writes_mode - supported sysctl write modes
156  *
157  * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
158  *	to be written, and multiple writes on the same sysctl file descriptor
159  *	will rewrite the sysctl value, regardless of file position. No warning
160  *	is issued when the initial position is not 0.
161  * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
162  *	not 0.
163  * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
164  *	file position 0 and the value must be fully contained in the buffer
165  *	sent to the write syscall. If dealing with strings respect the file
166  *	position, but restrict this to the max length of the buffer, anything
167  *	passed the max length will be ignored. Multiple writes will append
168  *	to the buffer.
169  *
170  * These write modes control how current file position affects the behavior of
171  * updating sysctl values through the proc interface on each write.
172  */
173 enum sysctl_writes_mode {
174 	SYSCTL_WRITES_LEGACY		= -1,
175 	SYSCTL_WRITES_WARN		= 0,
176 	SYSCTL_WRITES_STRICT		= 1,
177 };
178 
179 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
180 #endif /* CONFIG_PROC_SYSCTL */
181 
182 #if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
183     defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
184 int sysctl_legacy_va_layout;
185 #endif
186 
187 #ifdef CONFIG_SCHED_DEBUG
188 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
189 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
190 static int min_wakeup_granularity_ns;			/* 0 usecs */
191 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
192 #ifdef CONFIG_SMP
193 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
194 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
195 #endif /* CONFIG_SMP */
196 #endif /* CONFIG_SCHED_DEBUG */
197 
198 #ifdef CONFIG_COMPACTION
199 static int min_extfrag_threshold;
200 static int max_extfrag_threshold = 1000;
201 #endif
202 
203 #endif /* CONFIG_SYSCTL */
204 
205 #if defined(CONFIG_BPF_SYSCALL) && defined(CONFIG_SYSCTL)
206 static int bpf_stats_handler(struct ctl_table *table, int write,
207 			     void __user *buffer, size_t *lenp,
208 			     loff_t *ppos)
209 {
210 	struct static_key *key = (struct static_key *)table->data;
211 	static int saved_val;
212 	int val, ret;
213 	struct ctl_table tmp = {
214 		.data   = &val,
215 		.maxlen = sizeof(val),
216 		.mode   = table->mode,
217 		.extra1 = SYSCTL_ZERO,
218 		.extra2 = SYSCTL_ONE,
219 	};
220 
221 	if (write && !capable(CAP_SYS_ADMIN))
222 		return -EPERM;
223 
224 	mutex_lock(&bpf_stats_enabled_mutex);
225 	val = saved_val;
226 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
227 	if (write && !ret && val != saved_val) {
228 		if (val)
229 			static_key_slow_inc(key);
230 		else
231 			static_key_slow_dec(key);
232 		saved_val = val;
233 	}
234 	mutex_unlock(&bpf_stats_enabled_mutex);
235 	return ret;
236 }
237 #endif
238 
239 /*
240  * /proc/sys support
241  */
242 
243 #ifdef CONFIG_PROC_SYSCTL
244 
245 static int _proc_do_string(char *data, int maxlen, int write,
246 		char *buffer, size_t *lenp, loff_t *ppos)
247 {
248 	size_t len;
249 	char c, *p;
250 
251 	if (!data || !maxlen || !*lenp) {
252 		*lenp = 0;
253 		return 0;
254 	}
255 
256 	if (write) {
257 		if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
258 			/* Only continue writes not past the end of buffer. */
259 			len = strlen(data);
260 			if (len > maxlen - 1)
261 				len = maxlen - 1;
262 
263 			if (*ppos > len)
264 				return 0;
265 			len = *ppos;
266 		} else {
267 			/* Start writing from beginning of buffer. */
268 			len = 0;
269 		}
270 
271 		*ppos += *lenp;
272 		p = buffer;
273 		while ((p - buffer) < *lenp && len < maxlen - 1) {
274 			c = *(p++);
275 			if (c == 0 || c == '\n')
276 				break;
277 			data[len++] = c;
278 		}
279 		data[len] = 0;
280 	} else {
281 		len = strlen(data);
282 		if (len > maxlen)
283 			len = maxlen;
284 
285 		if (*ppos > len) {
286 			*lenp = 0;
287 			return 0;
288 		}
289 
290 		data += *ppos;
291 		len  -= *ppos;
292 
293 		if (len > *lenp)
294 			len = *lenp;
295 		if (len)
296 			memcpy(buffer, data, len);
297 		if (len < *lenp) {
298 			buffer[len] = '\n';
299 			len++;
300 		}
301 		*lenp = len;
302 		*ppos += len;
303 	}
304 	return 0;
305 }
306 
307 static void warn_sysctl_write(struct ctl_table *table)
308 {
309 	pr_warn_once("%s wrote to %s when file position was not 0!\n"
310 		"This will not be supported in the future. To silence this\n"
311 		"warning, set kernel.sysctl_writes_strict = -1\n",
312 		current->comm, table->procname);
313 }
314 
315 /**
316  * proc_first_pos_non_zero_ignore - check if first position is allowed
317  * @ppos: file position
318  * @table: the sysctl table
319  *
320  * Returns true if the first position is non-zero and the sysctl_writes_strict
321  * mode indicates this is not allowed for numeric input types. String proc
322  * handlers can ignore the return value.
323  */
324 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
325 					   struct ctl_table *table)
326 {
327 	if (!*ppos)
328 		return false;
329 
330 	switch (sysctl_writes_strict) {
331 	case SYSCTL_WRITES_STRICT:
332 		return true;
333 	case SYSCTL_WRITES_WARN:
334 		warn_sysctl_write(table);
335 		return false;
336 	default:
337 		return false;
338 	}
339 }
340 
341 /**
342  * proc_dostring - read a string sysctl
343  * @table: the sysctl table
344  * @write: %TRUE if this is a write to the sysctl file
345  * @buffer: the user buffer
346  * @lenp: the size of the user buffer
347  * @ppos: file position
348  *
349  * Reads/writes a string from/to the user buffer. If the kernel
350  * buffer provided is not large enough to hold the string, the
351  * string is truncated. The copied string is %NULL-terminated.
352  * If the string is being read by the user process, it is copied
353  * and a newline '\n' is added. It is truncated if the buffer is
354  * not large enough.
355  *
356  * Returns 0 on success.
357  */
358 int proc_dostring(struct ctl_table *table, int write,
359 		  void *buffer, size_t *lenp, loff_t *ppos)
360 {
361 	if (write)
362 		proc_first_pos_non_zero_ignore(ppos, table);
363 
364 	return _proc_do_string(table->data, table->maxlen, write, buffer, lenp,
365 			ppos);
366 }
367 
368 static size_t proc_skip_spaces(char **buf)
369 {
370 	size_t ret;
371 	char *tmp = skip_spaces(*buf);
372 	ret = tmp - *buf;
373 	*buf = tmp;
374 	return ret;
375 }
376 
377 static void proc_skip_char(char **buf, size_t *size, const char v)
378 {
379 	while (*size) {
380 		if (**buf != v)
381 			break;
382 		(*size)--;
383 		(*buf)++;
384 	}
385 }
386 
387 /**
388  * strtoul_lenient - parse an ASCII formatted integer from a buffer and only
389  *                   fail on overflow
390  *
391  * @cp: kernel buffer containing the string to parse
392  * @endp: pointer to store the trailing characters
393  * @base: the base to use
394  * @res: where the parsed integer will be stored
395  *
396  * In case of success 0 is returned and @res will contain the parsed integer,
397  * @endp will hold any trailing characters.
398  * This function will fail the parse on overflow. If there wasn't an overflow
399  * the function will defer the decision what characters count as invalid to the
400  * caller.
401  */
402 static int strtoul_lenient(const char *cp, char **endp, unsigned int base,
403 			   unsigned long *res)
404 {
405 	unsigned long long result;
406 	unsigned int rv;
407 
408 	cp = _parse_integer_fixup_radix(cp, &base);
409 	rv = _parse_integer(cp, base, &result);
410 	if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result))
411 		return -ERANGE;
412 
413 	cp += rv;
414 
415 	if (endp)
416 		*endp = (char *)cp;
417 
418 	*res = (unsigned long)result;
419 	return 0;
420 }
421 
422 #define TMPBUFLEN 22
423 /**
424  * proc_get_long - reads an ASCII formatted integer from a user buffer
425  *
426  * @buf: a kernel buffer
427  * @size: size of the kernel buffer
428  * @val: this is where the number will be stored
429  * @neg: set to %TRUE if number is negative
430  * @perm_tr: a vector which contains the allowed trailers
431  * @perm_tr_len: size of the perm_tr vector
432  * @tr: pointer to store the trailer character
433  *
434  * In case of success %0 is returned and @buf and @size are updated with
435  * the amount of bytes read. If @tr is non-NULL and a trailing
436  * character exists (size is non-zero after returning from this
437  * function), @tr is updated with the trailing character.
438  */
439 static int proc_get_long(char **buf, size_t *size,
440 			  unsigned long *val, bool *neg,
441 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
442 {
443 	int len;
444 	char *p, tmp[TMPBUFLEN];
445 
446 	if (!*size)
447 		return -EINVAL;
448 
449 	len = *size;
450 	if (len > TMPBUFLEN - 1)
451 		len = TMPBUFLEN - 1;
452 
453 	memcpy(tmp, *buf, len);
454 
455 	tmp[len] = 0;
456 	p = tmp;
457 	if (*p == '-' && *size > 1) {
458 		*neg = true;
459 		p++;
460 	} else
461 		*neg = false;
462 	if (!isdigit(*p))
463 		return -EINVAL;
464 
465 	if (strtoul_lenient(p, &p, 0, val))
466 		return -EINVAL;
467 
468 	len = p - tmp;
469 
470 	/* We don't know if the next char is whitespace thus we may accept
471 	 * invalid integers (e.g. 1234...a) or two integers instead of one
472 	 * (e.g. 123...1). So lets not allow such large numbers. */
473 	if (len == TMPBUFLEN - 1)
474 		return -EINVAL;
475 
476 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
477 		return -EINVAL;
478 
479 	if (tr && (len < *size))
480 		*tr = *p;
481 
482 	*buf += len;
483 	*size -= len;
484 
485 	return 0;
486 }
487 
488 /**
489  * proc_put_long - converts an integer to a decimal ASCII formatted string
490  *
491  * @buf: the user buffer
492  * @size: the size of the user buffer
493  * @val: the integer to be converted
494  * @neg: sign of the number, %TRUE for negative
495  *
496  * In case of success @buf and @size are updated with the amount of bytes
497  * written.
498  */
499 static void proc_put_long(void **buf, size_t *size, unsigned long val, bool neg)
500 {
501 	int len;
502 	char tmp[TMPBUFLEN], *p = tmp;
503 
504 	sprintf(p, "%s%lu", neg ? "-" : "", val);
505 	len = strlen(tmp);
506 	if (len > *size)
507 		len = *size;
508 	memcpy(*buf, tmp, len);
509 	*size -= len;
510 	*buf += len;
511 }
512 #undef TMPBUFLEN
513 
514 static void proc_put_char(void **buf, size_t *size, char c)
515 {
516 	if (*size) {
517 		char **buffer = (char **)buf;
518 		**buffer = c;
519 
520 		(*size)--;
521 		(*buffer)++;
522 		*buf = *buffer;
523 	}
524 }
525 
526 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
527 				 int *valp,
528 				 int write, void *data)
529 {
530 	if (write) {
531 		if (*negp) {
532 			if (*lvalp > (unsigned long) INT_MAX + 1)
533 				return -EINVAL;
534 			*valp = -*lvalp;
535 		} else {
536 			if (*lvalp > (unsigned long) INT_MAX)
537 				return -EINVAL;
538 			*valp = *lvalp;
539 		}
540 	} else {
541 		int val = *valp;
542 		if (val < 0) {
543 			*negp = true;
544 			*lvalp = -(unsigned long)val;
545 		} else {
546 			*negp = false;
547 			*lvalp = (unsigned long)val;
548 		}
549 	}
550 	return 0;
551 }
552 
553 static int do_proc_douintvec_conv(unsigned long *lvalp,
554 				  unsigned int *valp,
555 				  int write, void *data)
556 {
557 	if (write) {
558 		if (*lvalp > UINT_MAX)
559 			return -EINVAL;
560 		*valp = *lvalp;
561 	} else {
562 		unsigned int val = *valp;
563 		*lvalp = (unsigned long)val;
564 	}
565 	return 0;
566 }
567 
568 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
569 
570 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
571 		  int write, void *buffer,
572 		  size_t *lenp, loff_t *ppos,
573 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
574 			      int write, void *data),
575 		  void *data)
576 {
577 	int *i, vleft, first = 1, err = 0;
578 	size_t left;
579 	char *p;
580 
581 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
582 		*lenp = 0;
583 		return 0;
584 	}
585 
586 	i = (int *) tbl_data;
587 	vleft = table->maxlen / sizeof(*i);
588 	left = *lenp;
589 
590 	if (!conv)
591 		conv = do_proc_dointvec_conv;
592 
593 	if (write) {
594 		if (proc_first_pos_non_zero_ignore(ppos, table))
595 			goto out;
596 
597 		if (left > PAGE_SIZE - 1)
598 			left = PAGE_SIZE - 1;
599 		p = buffer;
600 	}
601 
602 	for (; left && vleft--; i++, first=0) {
603 		unsigned long lval;
604 		bool neg;
605 
606 		if (write) {
607 			left -= proc_skip_spaces(&p);
608 
609 			if (!left)
610 				break;
611 			err = proc_get_long(&p, &left, &lval, &neg,
612 					     proc_wspace_sep,
613 					     sizeof(proc_wspace_sep), NULL);
614 			if (err)
615 				break;
616 			if (conv(&neg, &lval, i, 1, data)) {
617 				err = -EINVAL;
618 				break;
619 			}
620 		} else {
621 			if (conv(&neg, &lval, i, 0, data)) {
622 				err = -EINVAL;
623 				break;
624 			}
625 			if (!first)
626 				proc_put_char(&buffer, &left, '\t');
627 			proc_put_long(&buffer, &left, lval, neg);
628 		}
629 	}
630 
631 	if (!write && !first && left && !err)
632 		proc_put_char(&buffer, &left, '\n');
633 	if (write && !err && left)
634 		left -= proc_skip_spaces(&p);
635 	if (write && first)
636 		return err ? : -EINVAL;
637 	*lenp -= left;
638 out:
639 	*ppos += *lenp;
640 	return err;
641 }
642 
643 static int do_proc_dointvec(struct ctl_table *table, int write,
644 		  void *buffer, size_t *lenp, loff_t *ppos,
645 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
646 			      int write, void *data),
647 		  void *data)
648 {
649 	return __do_proc_dointvec(table->data, table, write,
650 			buffer, lenp, ppos, conv, data);
651 }
652 
653 static int do_proc_douintvec_w(unsigned int *tbl_data,
654 			       struct ctl_table *table,
655 			       void *buffer,
656 			       size_t *lenp, loff_t *ppos,
657 			       int (*conv)(unsigned long *lvalp,
658 					   unsigned int *valp,
659 					   int write, void *data),
660 			       void *data)
661 {
662 	unsigned long lval;
663 	int err = 0;
664 	size_t left;
665 	bool neg;
666 	char *p = buffer;
667 
668 	left = *lenp;
669 
670 	if (proc_first_pos_non_zero_ignore(ppos, table))
671 		goto bail_early;
672 
673 	if (left > PAGE_SIZE - 1)
674 		left = PAGE_SIZE - 1;
675 
676 	left -= proc_skip_spaces(&p);
677 	if (!left) {
678 		err = -EINVAL;
679 		goto out_free;
680 	}
681 
682 	err = proc_get_long(&p, &left, &lval, &neg,
683 			     proc_wspace_sep,
684 			     sizeof(proc_wspace_sep), NULL);
685 	if (err || neg) {
686 		err = -EINVAL;
687 		goto out_free;
688 	}
689 
690 	if (conv(&lval, tbl_data, 1, data)) {
691 		err = -EINVAL;
692 		goto out_free;
693 	}
694 
695 	if (!err && left)
696 		left -= proc_skip_spaces(&p);
697 
698 out_free:
699 	if (err)
700 		return -EINVAL;
701 
702 	return 0;
703 
704 	/* This is in keeping with old __do_proc_dointvec() */
705 bail_early:
706 	*ppos += *lenp;
707 	return err;
708 }
709 
710 static int do_proc_douintvec_r(unsigned int *tbl_data, void *buffer,
711 			       size_t *lenp, loff_t *ppos,
712 			       int (*conv)(unsigned long *lvalp,
713 					   unsigned int *valp,
714 					   int write, void *data),
715 			       void *data)
716 {
717 	unsigned long lval;
718 	int err = 0;
719 	size_t left;
720 
721 	left = *lenp;
722 
723 	if (conv(&lval, tbl_data, 0, data)) {
724 		err = -EINVAL;
725 		goto out;
726 	}
727 
728 	proc_put_long(&buffer, &left, lval, false);
729 	if (!left)
730 		goto out;
731 
732 	proc_put_char(&buffer, &left, '\n');
733 
734 out:
735 	*lenp -= left;
736 	*ppos += *lenp;
737 
738 	return err;
739 }
740 
741 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
742 			       int write, void *buffer,
743 			       size_t *lenp, loff_t *ppos,
744 			       int (*conv)(unsigned long *lvalp,
745 					   unsigned int *valp,
746 					   int write, void *data),
747 			       void *data)
748 {
749 	unsigned int *i, vleft;
750 
751 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
752 		*lenp = 0;
753 		return 0;
754 	}
755 
756 	i = (unsigned int *) tbl_data;
757 	vleft = table->maxlen / sizeof(*i);
758 
759 	/*
760 	 * Arrays are not supported, keep this simple. *Do not* add
761 	 * support for them.
762 	 */
763 	if (vleft != 1) {
764 		*lenp = 0;
765 		return -EINVAL;
766 	}
767 
768 	if (!conv)
769 		conv = do_proc_douintvec_conv;
770 
771 	if (write)
772 		return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
773 					   conv, data);
774 	return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
775 }
776 
777 static int do_proc_douintvec(struct ctl_table *table, int write,
778 			     void *buffer, size_t *lenp, loff_t *ppos,
779 			     int (*conv)(unsigned long *lvalp,
780 					 unsigned int *valp,
781 					 int write, void *data),
782 			     void *data)
783 {
784 	return __do_proc_douintvec(table->data, table, write,
785 				   buffer, lenp, ppos, conv, data);
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 #ifdef CONFIG_PRINTK
892 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
893 				void *buffer, size_t *lenp, loff_t *ppos)
894 {
895 	if (write && !capable(CAP_SYS_ADMIN))
896 		return -EPERM;
897 
898 	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
899 }
900 #endif
901 
902 /**
903  * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
904  * @min: pointer to minimum allowable value
905  * @max: pointer to maximum allowable value
906  *
907  * The do_proc_dointvec_minmax_conv_param structure provides the
908  * minimum and maximum values for doing range checking for those sysctl
909  * parameters that use the proc_dointvec_minmax() handler.
910  */
911 struct do_proc_dointvec_minmax_conv_param {
912 	int *min;
913 	int *max;
914 };
915 
916 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
917 					int *valp,
918 					int write, void *data)
919 {
920 	int tmp, ret;
921 	struct do_proc_dointvec_minmax_conv_param *param = data;
922 	/*
923 	 * If writing, first do so via a temporary local int so we can
924 	 * bounds-check it before touching *valp.
925 	 */
926 	int *ip = write ? &tmp : valp;
927 
928 	ret = do_proc_dointvec_conv(negp, lvalp, ip, write, data);
929 	if (ret)
930 		return ret;
931 
932 	if (write) {
933 		if ((param->min && *param->min > tmp) ||
934 		    (param->max && *param->max < tmp))
935 			return -EINVAL;
936 		*valp = tmp;
937 	}
938 
939 	return 0;
940 }
941 
942 /**
943  * proc_dointvec_minmax - read a vector of integers with min/max values
944  * @table: the sysctl table
945  * @write: %TRUE if this is a write to the sysctl file
946  * @buffer: the user buffer
947  * @lenp: the size of the user buffer
948  * @ppos: file position
949  *
950  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
951  * values from/to the user buffer, treated as an ASCII string.
952  *
953  * This routine will ensure the values are within the range specified by
954  * table->extra1 (min) and table->extra2 (max).
955  *
956  * Returns 0 on success or -EINVAL on write when the range check fails.
957  */
958 int proc_dointvec_minmax(struct ctl_table *table, int write,
959 		  void *buffer, size_t *lenp, loff_t *ppos)
960 {
961 	struct do_proc_dointvec_minmax_conv_param param = {
962 		.min = (int *) table->extra1,
963 		.max = (int *) table->extra2,
964 	};
965 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
966 				do_proc_dointvec_minmax_conv, &param);
967 }
968 
969 /**
970  * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
971  * @min: pointer to minimum allowable value
972  * @max: pointer to maximum allowable value
973  *
974  * The do_proc_douintvec_minmax_conv_param structure provides the
975  * minimum and maximum values for doing range checking for those sysctl
976  * parameters that use the proc_douintvec_minmax() handler.
977  */
978 struct do_proc_douintvec_minmax_conv_param {
979 	unsigned int *min;
980 	unsigned int *max;
981 };
982 
983 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
984 					 unsigned int *valp,
985 					 int write, void *data)
986 {
987 	int ret;
988 	unsigned int tmp;
989 	struct do_proc_douintvec_minmax_conv_param *param = data;
990 	/* write via temporary local uint for bounds-checking */
991 	unsigned int *up = write ? &tmp : valp;
992 
993 	ret = do_proc_douintvec_conv(lvalp, up, write, data);
994 	if (ret)
995 		return ret;
996 
997 	if (write) {
998 		if ((param->min && *param->min > tmp) ||
999 		    (param->max && *param->max < tmp))
1000 			return -ERANGE;
1001 
1002 		*valp = tmp;
1003 	}
1004 
1005 	return 0;
1006 }
1007 
1008 /**
1009  * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
1010  * @table: the sysctl table
1011  * @write: %TRUE if this is a write to the sysctl file
1012  * @buffer: the user buffer
1013  * @lenp: the size of the user buffer
1014  * @ppos: file position
1015  *
1016  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
1017  * values from/to the user buffer, treated as an ASCII string. Negative
1018  * strings are not allowed.
1019  *
1020  * This routine will ensure the values are within the range specified by
1021  * table->extra1 (min) and table->extra2 (max). There is a final sanity
1022  * check for UINT_MAX to avoid having to support wrap around uses from
1023  * userspace.
1024  *
1025  * Returns 0 on success or -ERANGE on write when the range check fails.
1026  */
1027 int proc_douintvec_minmax(struct ctl_table *table, int write,
1028 			  void *buffer, size_t *lenp, loff_t *ppos)
1029 {
1030 	struct do_proc_douintvec_minmax_conv_param param = {
1031 		.min = (unsigned int *) table->extra1,
1032 		.max = (unsigned int *) table->extra2,
1033 	};
1034 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
1035 				 do_proc_douintvec_minmax_conv, &param);
1036 }
1037 
1038 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
1039 					unsigned int *valp,
1040 					int write, void *data)
1041 {
1042 	if (write) {
1043 		unsigned int val;
1044 
1045 		val = round_pipe_size(*lvalp);
1046 		if (val == 0)
1047 			return -EINVAL;
1048 
1049 		*valp = val;
1050 	} else {
1051 		unsigned int val = *valp;
1052 		*lvalp = (unsigned long) val;
1053 	}
1054 
1055 	return 0;
1056 }
1057 
1058 static int proc_dopipe_max_size(struct ctl_table *table, int write,
1059 				void *buffer, size_t *lenp, loff_t *ppos)
1060 {
1061 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
1062 				 do_proc_dopipe_max_size_conv, NULL);
1063 }
1064 
1065 static void validate_coredump_safety(void)
1066 {
1067 #ifdef CONFIG_COREDUMP
1068 	if (suid_dumpable == SUID_DUMP_ROOT &&
1069 	    core_pattern[0] != '/' && core_pattern[0] != '|') {
1070 		printk(KERN_WARNING
1071 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
1072 "Pipe handler or fully qualified core dump path required.\n"
1073 "Set kernel.core_pattern before fs.suid_dumpable.\n"
1074 		);
1075 	}
1076 #endif
1077 }
1078 
1079 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
1080 		void *buffer, size_t *lenp, loff_t *ppos)
1081 {
1082 	int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
1083 	if (!error)
1084 		validate_coredump_safety();
1085 	return error;
1086 }
1087 
1088 #ifdef CONFIG_COREDUMP
1089 static int proc_dostring_coredump(struct ctl_table *table, int write,
1090 		  void *buffer, size_t *lenp, loff_t *ppos)
1091 {
1092 	int error = proc_dostring(table, write, buffer, lenp, ppos);
1093 	if (!error)
1094 		validate_coredump_safety();
1095 	return error;
1096 }
1097 #endif
1098 
1099 #ifdef CONFIG_MAGIC_SYSRQ
1100 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
1101 				void *buffer, size_t *lenp, loff_t *ppos)
1102 {
1103 	int tmp, ret;
1104 
1105 	tmp = sysrq_mask();
1106 
1107 	ret = __do_proc_dointvec(&tmp, table, write, buffer,
1108 			       lenp, ppos, NULL, NULL);
1109 	if (ret || !write)
1110 		return ret;
1111 
1112 	if (write)
1113 		sysrq_toggle_support(tmp);
1114 
1115 	return 0;
1116 }
1117 #endif
1118 
1119 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table,
1120 		int write, void *buffer, size_t *lenp, loff_t *ppos,
1121 		unsigned long convmul, unsigned long convdiv)
1122 {
1123 	unsigned long *i, *min, *max;
1124 	int vleft, first = 1, err = 0;
1125 	size_t left;
1126 	char *p;
1127 
1128 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
1129 		*lenp = 0;
1130 		return 0;
1131 	}
1132 
1133 	i = (unsigned long *) data;
1134 	min = (unsigned long *) table->extra1;
1135 	max = (unsigned long *) table->extra2;
1136 	vleft = table->maxlen / sizeof(unsigned long);
1137 	left = *lenp;
1138 
1139 	if (write) {
1140 		if (proc_first_pos_non_zero_ignore(ppos, table))
1141 			goto out;
1142 
1143 		if (left > PAGE_SIZE - 1)
1144 			left = PAGE_SIZE - 1;
1145 		p = buffer;
1146 	}
1147 
1148 	for (; left && vleft--; i++, first = 0) {
1149 		unsigned long val;
1150 
1151 		if (write) {
1152 			bool neg;
1153 
1154 			left -= proc_skip_spaces(&p);
1155 			if (!left)
1156 				break;
1157 
1158 			err = proc_get_long(&p, &left, &val, &neg,
1159 					     proc_wspace_sep,
1160 					     sizeof(proc_wspace_sep), NULL);
1161 			if (err)
1162 				break;
1163 			if (neg)
1164 				continue;
1165 			val = convmul * val / convdiv;
1166 			if ((min && val < *min) || (max && val > *max)) {
1167 				err = -EINVAL;
1168 				break;
1169 			}
1170 			*i = val;
1171 		} else {
1172 			val = convdiv * (*i) / convmul;
1173 			if (!first)
1174 				proc_put_char(&buffer, &left, '\t');
1175 			proc_put_long(&buffer, &left, val, false);
1176 		}
1177 	}
1178 
1179 	if (!write && !first && left && !err)
1180 		proc_put_char(&buffer, &left, '\n');
1181 	if (write && !err)
1182 		left -= proc_skip_spaces(&p);
1183 	if (write && first)
1184 		return err ? : -EINVAL;
1185 	*lenp -= left;
1186 out:
1187 	*ppos += *lenp;
1188 	return err;
1189 }
1190 
1191 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
1192 		void *buffer, size_t *lenp, loff_t *ppos, unsigned long convmul,
1193 		unsigned long convdiv)
1194 {
1195 	return __do_proc_doulongvec_minmax(table->data, table, write,
1196 			buffer, lenp, ppos, convmul, convdiv);
1197 }
1198 
1199 /**
1200  * proc_doulongvec_minmax - read a vector of long integers with min/max values
1201  * @table: the sysctl table
1202  * @write: %TRUE if this is a write to the sysctl file
1203  * @buffer: the user buffer
1204  * @lenp: the size of the user buffer
1205  * @ppos: file position
1206  *
1207  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1208  * values from/to the user buffer, treated as an ASCII string.
1209  *
1210  * This routine will ensure the values are within the range specified by
1211  * table->extra1 (min) and table->extra2 (max).
1212  *
1213  * Returns 0 on success.
1214  */
1215 int proc_doulongvec_minmax(struct ctl_table *table, int write,
1216 			   void *buffer, size_t *lenp, loff_t *ppos)
1217 {
1218     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
1219 }
1220 
1221 /**
1222  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
1223  * @table: the sysctl table
1224  * @write: %TRUE if this is a write to the sysctl file
1225  * @buffer: the user buffer
1226  * @lenp: the size of the user buffer
1227  * @ppos: file position
1228  *
1229  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1230  * values from/to the user buffer, treated as an ASCII string. The values
1231  * are treated as milliseconds, and converted to jiffies when they are stored.
1232  *
1233  * This routine will ensure the values are within the range specified by
1234  * table->extra1 (min) and table->extra2 (max).
1235  *
1236  * Returns 0 on success.
1237  */
1238 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1239 				      void *buffer, size_t *lenp, loff_t *ppos)
1240 {
1241     return do_proc_doulongvec_minmax(table, write, buffer,
1242 				     lenp, ppos, HZ, 1000l);
1243 }
1244 
1245 
1246 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
1247 					 int *valp,
1248 					 int write, void *data)
1249 {
1250 	if (write) {
1251 		if (*lvalp > INT_MAX / HZ)
1252 			return 1;
1253 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
1254 	} else {
1255 		int val = *valp;
1256 		unsigned long lval;
1257 		if (val < 0) {
1258 			*negp = true;
1259 			lval = -(unsigned long)val;
1260 		} else {
1261 			*negp = false;
1262 			lval = (unsigned long)val;
1263 		}
1264 		*lvalp = lval / HZ;
1265 	}
1266 	return 0;
1267 }
1268 
1269 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
1270 						int *valp,
1271 						int write, void *data)
1272 {
1273 	if (write) {
1274 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
1275 			return 1;
1276 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
1277 	} else {
1278 		int val = *valp;
1279 		unsigned long lval;
1280 		if (val < 0) {
1281 			*negp = true;
1282 			lval = -(unsigned long)val;
1283 		} else {
1284 			*negp = false;
1285 			lval = (unsigned long)val;
1286 		}
1287 		*lvalp = jiffies_to_clock_t(lval);
1288 	}
1289 	return 0;
1290 }
1291 
1292 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
1293 					    int *valp,
1294 					    int write, void *data)
1295 {
1296 	if (write) {
1297 		unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
1298 
1299 		if (jif > INT_MAX)
1300 			return 1;
1301 		*valp = (int)jif;
1302 	} else {
1303 		int val = *valp;
1304 		unsigned long lval;
1305 		if (val < 0) {
1306 			*negp = true;
1307 			lval = -(unsigned long)val;
1308 		} else {
1309 			*negp = false;
1310 			lval = (unsigned long)val;
1311 		}
1312 		*lvalp = jiffies_to_msecs(lval);
1313 	}
1314 	return 0;
1315 }
1316 
1317 /**
1318  * proc_dointvec_jiffies - read a vector of integers as seconds
1319  * @table: the sysctl table
1320  * @write: %TRUE if this is a write to the sysctl file
1321  * @buffer: the user buffer
1322  * @lenp: the size of the user buffer
1323  * @ppos: file position
1324  *
1325  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1326  * values from/to the user buffer, treated as an ASCII string.
1327  * The values read are assumed to be in seconds, and are converted into
1328  * jiffies.
1329  *
1330  * Returns 0 on success.
1331  */
1332 int proc_dointvec_jiffies(struct ctl_table *table, int write,
1333 			  void *buffer, size_t *lenp, loff_t *ppos)
1334 {
1335     return do_proc_dointvec(table,write,buffer,lenp,ppos,
1336 		    	    do_proc_dointvec_jiffies_conv,NULL);
1337 }
1338 
1339 /**
1340  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
1341  * @table: the sysctl table
1342  * @write: %TRUE if this is a write to the sysctl file
1343  * @buffer: the user buffer
1344  * @lenp: the size of the user buffer
1345  * @ppos: pointer to the file position
1346  *
1347  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1348  * values from/to the user buffer, treated as an ASCII string.
1349  * The values read are assumed to be in 1/USER_HZ seconds, and
1350  * are converted into jiffies.
1351  *
1352  * Returns 0 on success.
1353  */
1354 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1355 				 void *buffer, size_t *lenp, loff_t *ppos)
1356 {
1357     return do_proc_dointvec(table,write,buffer,lenp,ppos,
1358 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
1359 }
1360 
1361 /**
1362  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
1363  * @table: the sysctl table
1364  * @write: %TRUE if this is a write to the sysctl file
1365  * @buffer: the user buffer
1366  * @lenp: the size of the user buffer
1367  * @ppos: file position
1368  * @ppos: the current position in the file
1369  *
1370  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1371  * values from/to the user buffer, treated as an ASCII string.
1372  * The values read are assumed to be in 1/1000 seconds, and
1373  * are converted into jiffies.
1374  *
1375  * Returns 0 on success.
1376  */
1377 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, void *buffer,
1378 		size_t *lenp, loff_t *ppos)
1379 {
1380 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
1381 				do_proc_dointvec_ms_jiffies_conv, NULL);
1382 }
1383 
1384 static int proc_do_cad_pid(struct ctl_table *table, int write, void *buffer,
1385 		size_t *lenp, loff_t *ppos)
1386 {
1387 	struct pid *new_pid;
1388 	pid_t tmp;
1389 	int r;
1390 
1391 	tmp = pid_vnr(cad_pid);
1392 
1393 	r = __do_proc_dointvec(&tmp, table, write, buffer,
1394 			       lenp, ppos, NULL, NULL);
1395 	if (r || !write)
1396 		return r;
1397 
1398 	new_pid = find_get_pid(tmp);
1399 	if (!new_pid)
1400 		return -ESRCH;
1401 
1402 	put_pid(xchg(&cad_pid, new_pid));
1403 	return 0;
1404 }
1405 
1406 /**
1407  * proc_do_large_bitmap - read/write from/to a large bitmap
1408  * @table: the sysctl table
1409  * @write: %TRUE if this is a write to the sysctl file
1410  * @buffer: the user buffer
1411  * @lenp: the size of the user buffer
1412  * @ppos: file position
1413  *
1414  * The bitmap is stored at table->data and the bitmap length (in bits)
1415  * in table->maxlen.
1416  *
1417  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
1418  * large bitmaps may be represented in a compact manner. Writing into
1419  * the file will clear the bitmap then update it with the given input.
1420  *
1421  * Returns 0 on success.
1422  */
1423 int proc_do_large_bitmap(struct ctl_table *table, int write,
1424 			 void *buffer, size_t *lenp, loff_t *ppos)
1425 {
1426 	int err = 0;
1427 	bool first = 1;
1428 	size_t left = *lenp;
1429 	unsigned long bitmap_len = table->maxlen;
1430 	unsigned long *bitmap = *(unsigned long **) table->data;
1431 	unsigned long *tmp_bitmap = NULL;
1432 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
1433 
1434 	if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
1435 		*lenp = 0;
1436 		return 0;
1437 	}
1438 
1439 	if (write) {
1440 		char *p = buffer;
1441 		size_t skipped = 0;
1442 
1443 		if (left > PAGE_SIZE - 1) {
1444 			left = PAGE_SIZE - 1;
1445 			/* How much of the buffer we'll skip this pass */
1446 			skipped = *lenp - left;
1447 		}
1448 
1449 		tmp_bitmap = bitmap_zalloc(bitmap_len, GFP_KERNEL);
1450 		if (!tmp_bitmap)
1451 			return -ENOMEM;
1452 		proc_skip_char(&p, &left, '\n');
1453 		while (!err && left) {
1454 			unsigned long val_a, val_b;
1455 			bool neg;
1456 			size_t saved_left;
1457 
1458 			/* In case we stop parsing mid-number, we can reset */
1459 			saved_left = left;
1460 			err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
1461 					     sizeof(tr_a), &c);
1462 			/*
1463 			 * If we consumed the entirety of a truncated buffer or
1464 			 * only one char is left (may be a "-"), then stop here,
1465 			 * reset, & come back for more.
1466 			 */
1467 			if ((left <= 1) && skipped) {
1468 				left = saved_left;
1469 				break;
1470 			}
1471 
1472 			if (err)
1473 				break;
1474 			if (val_a >= bitmap_len || neg) {
1475 				err = -EINVAL;
1476 				break;
1477 			}
1478 
1479 			val_b = val_a;
1480 			if (left) {
1481 				p++;
1482 				left--;
1483 			}
1484 
1485 			if (c == '-') {
1486 				err = proc_get_long(&p, &left, &val_b,
1487 						     &neg, tr_b, sizeof(tr_b),
1488 						     &c);
1489 				/*
1490 				 * If we consumed all of a truncated buffer or
1491 				 * then stop here, reset, & come back for more.
1492 				 */
1493 				if (!left && skipped) {
1494 					left = saved_left;
1495 					break;
1496 				}
1497 
1498 				if (err)
1499 					break;
1500 				if (val_b >= bitmap_len || neg ||
1501 				    val_a > val_b) {
1502 					err = -EINVAL;
1503 					break;
1504 				}
1505 				if (left) {
1506 					p++;
1507 					left--;
1508 				}
1509 			}
1510 
1511 			bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
1512 			first = 0;
1513 			proc_skip_char(&p, &left, '\n');
1514 		}
1515 		left += skipped;
1516 	} else {
1517 		unsigned long bit_a, bit_b = 0;
1518 
1519 		while (left) {
1520 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
1521 			if (bit_a >= bitmap_len)
1522 				break;
1523 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
1524 						   bit_a + 1) - 1;
1525 
1526 			if (!first)
1527 				proc_put_char(&buffer, &left, ',');
1528 			proc_put_long(&buffer, &left, bit_a, false);
1529 			if (bit_a != bit_b) {
1530 				proc_put_char(&buffer, &left, '-');
1531 				proc_put_long(&buffer, &left, bit_b, false);
1532 			}
1533 
1534 			first = 0; bit_b++;
1535 		}
1536 		proc_put_char(&buffer, &left, '\n');
1537 	}
1538 
1539 	if (!err) {
1540 		if (write) {
1541 			if (*ppos)
1542 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
1543 			else
1544 				bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
1545 		}
1546 		*lenp -= left;
1547 		*ppos += *lenp;
1548 	}
1549 
1550 	bitmap_free(tmp_bitmap);
1551 	return err;
1552 }
1553 
1554 #else /* CONFIG_PROC_SYSCTL */
1555 
1556 int proc_dostring(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_dointvec_jiffies(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_userhz_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_ms_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_doulongvec_minmax(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_ms_jiffies_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_do_large_bitmap(struct ctl_table *table, int write,
1617 			 void *buffer, size_t *lenp, loff_t *ppos)
1618 {
1619 	return -ENOSYS;
1620 }
1621 
1622 #endif /* CONFIG_PROC_SYSCTL */
1623 
1624 #if defined(CONFIG_SYSCTL)
1625 int proc_do_static_key(struct ctl_table *table, int write,
1626 		       void *buffer, size_t *lenp, loff_t *ppos)
1627 {
1628 	struct static_key *key = (struct static_key *)table->data;
1629 	static DEFINE_MUTEX(static_key_mutex);
1630 	int val, ret;
1631 	struct ctl_table tmp = {
1632 		.data   = &val,
1633 		.maxlen = sizeof(val),
1634 		.mode   = table->mode,
1635 		.extra1 = SYSCTL_ZERO,
1636 		.extra2 = SYSCTL_ONE,
1637 	};
1638 
1639 	if (write && !capable(CAP_SYS_ADMIN))
1640 		return -EPERM;
1641 
1642 	mutex_lock(&static_key_mutex);
1643 	val = static_key_enabled(key);
1644 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
1645 	if (write && !ret) {
1646 		if (val)
1647 			static_key_enable(key);
1648 		else
1649 			static_key_disable(key);
1650 	}
1651 	mutex_unlock(&static_key_mutex);
1652 	return ret;
1653 }
1654 
1655 static struct ctl_table kern_table[] = {
1656 	{
1657 		.procname	= "sched_child_runs_first",
1658 		.data		= &sysctl_sched_child_runs_first,
1659 		.maxlen		= sizeof(unsigned int),
1660 		.mode		= 0644,
1661 		.proc_handler	= proc_dointvec,
1662 	},
1663 #ifdef CONFIG_SCHED_DEBUG
1664 	{
1665 		.procname	= "sched_min_granularity_ns",
1666 		.data		= &sysctl_sched_min_granularity,
1667 		.maxlen		= sizeof(unsigned int),
1668 		.mode		= 0644,
1669 		.proc_handler	= sched_proc_update_handler,
1670 		.extra1		= &min_sched_granularity_ns,
1671 		.extra2		= &max_sched_granularity_ns,
1672 	},
1673 	{
1674 		.procname	= "sched_latency_ns",
1675 		.data		= &sysctl_sched_latency,
1676 		.maxlen		= sizeof(unsigned int),
1677 		.mode		= 0644,
1678 		.proc_handler	= sched_proc_update_handler,
1679 		.extra1		= &min_sched_granularity_ns,
1680 		.extra2		= &max_sched_granularity_ns,
1681 	},
1682 	{
1683 		.procname	= "sched_wakeup_granularity_ns",
1684 		.data		= &sysctl_sched_wakeup_granularity,
1685 		.maxlen		= sizeof(unsigned int),
1686 		.mode		= 0644,
1687 		.proc_handler	= sched_proc_update_handler,
1688 		.extra1		= &min_wakeup_granularity_ns,
1689 		.extra2		= &max_wakeup_granularity_ns,
1690 	},
1691 #ifdef CONFIG_SMP
1692 	{
1693 		.procname	= "sched_tunable_scaling",
1694 		.data		= &sysctl_sched_tunable_scaling,
1695 		.maxlen		= sizeof(enum sched_tunable_scaling),
1696 		.mode		= 0644,
1697 		.proc_handler	= sched_proc_update_handler,
1698 		.extra1		= &min_sched_tunable_scaling,
1699 		.extra2		= &max_sched_tunable_scaling,
1700 	},
1701 	{
1702 		.procname	= "sched_migration_cost_ns",
1703 		.data		= &sysctl_sched_migration_cost,
1704 		.maxlen		= sizeof(unsigned int),
1705 		.mode		= 0644,
1706 		.proc_handler	= proc_dointvec,
1707 	},
1708 	{
1709 		.procname	= "sched_nr_migrate",
1710 		.data		= &sysctl_sched_nr_migrate,
1711 		.maxlen		= sizeof(unsigned int),
1712 		.mode		= 0644,
1713 		.proc_handler	= proc_dointvec,
1714 	},
1715 #ifdef CONFIG_SCHEDSTATS
1716 	{
1717 		.procname	= "sched_schedstats",
1718 		.data		= NULL,
1719 		.maxlen		= sizeof(unsigned int),
1720 		.mode		= 0644,
1721 		.proc_handler	= sysctl_schedstats,
1722 		.extra1		= SYSCTL_ZERO,
1723 		.extra2		= SYSCTL_ONE,
1724 	},
1725 #endif /* CONFIG_SCHEDSTATS */
1726 #endif /* CONFIG_SMP */
1727 #ifdef CONFIG_NUMA_BALANCING
1728 	{
1729 		.procname	= "numa_balancing_scan_delay_ms",
1730 		.data		= &sysctl_numa_balancing_scan_delay,
1731 		.maxlen		= sizeof(unsigned int),
1732 		.mode		= 0644,
1733 		.proc_handler	= proc_dointvec,
1734 	},
1735 	{
1736 		.procname	= "numa_balancing_scan_period_min_ms",
1737 		.data		= &sysctl_numa_balancing_scan_period_min,
1738 		.maxlen		= sizeof(unsigned int),
1739 		.mode		= 0644,
1740 		.proc_handler	= proc_dointvec,
1741 	},
1742 	{
1743 		.procname	= "numa_balancing_scan_period_max_ms",
1744 		.data		= &sysctl_numa_balancing_scan_period_max,
1745 		.maxlen		= sizeof(unsigned int),
1746 		.mode		= 0644,
1747 		.proc_handler	= proc_dointvec,
1748 	},
1749 	{
1750 		.procname	= "numa_balancing_scan_size_mb",
1751 		.data		= &sysctl_numa_balancing_scan_size,
1752 		.maxlen		= sizeof(unsigned int),
1753 		.mode		= 0644,
1754 		.proc_handler	= proc_dointvec_minmax,
1755 		.extra1		= SYSCTL_ONE,
1756 	},
1757 	{
1758 		.procname	= "numa_balancing",
1759 		.data		= NULL, /* filled in by handler */
1760 		.maxlen		= sizeof(unsigned int),
1761 		.mode		= 0644,
1762 		.proc_handler	= sysctl_numa_balancing,
1763 		.extra1		= SYSCTL_ZERO,
1764 		.extra2		= SYSCTL_ONE,
1765 	},
1766 #endif /* CONFIG_NUMA_BALANCING */
1767 #endif /* CONFIG_SCHED_DEBUG */
1768 	{
1769 		.procname	= "sched_rt_period_us",
1770 		.data		= &sysctl_sched_rt_period,
1771 		.maxlen		= sizeof(unsigned int),
1772 		.mode		= 0644,
1773 		.proc_handler	= sched_rt_handler,
1774 	},
1775 	{
1776 		.procname	= "sched_rt_runtime_us",
1777 		.data		= &sysctl_sched_rt_runtime,
1778 		.maxlen		= sizeof(int),
1779 		.mode		= 0644,
1780 		.proc_handler	= sched_rt_handler,
1781 	},
1782 	{
1783 		.procname	= "sched_deadline_period_max_us",
1784 		.data		= &sysctl_sched_dl_period_max,
1785 		.maxlen		= sizeof(unsigned int),
1786 		.mode		= 0644,
1787 		.proc_handler	= proc_dointvec,
1788 	},
1789 	{
1790 		.procname	= "sched_deadline_period_min_us",
1791 		.data		= &sysctl_sched_dl_period_min,
1792 		.maxlen		= sizeof(unsigned int),
1793 		.mode		= 0644,
1794 		.proc_handler	= proc_dointvec,
1795 	},
1796 	{
1797 		.procname	= "sched_rr_timeslice_ms",
1798 		.data		= &sysctl_sched_rr_timeslice,
1799 		.maxlen		= sizeof(int),
1800 		.mode		= 0644,
1801 		.proc_handler	= sched_rr_handler,
1802 	},
1803 #ifdef CONFIG_UCLAMP_TASK
1804 	{
1805 		.procname	= "sched_util_clamp_min",
1806 		.data		= &sysctl_sched_uclamp_util_min,
1807 		.maxlen		= sizeof(unsigned int),
1808 		.mode		= 0644,
1809 		.proc_handler	= sysctl_sched_uclamp_handler,
1810 	},
1811 	{
1812 		.procname	= "sched_util_clamp_max",
1813 		.data		= &sysctl_sched_uclamp_util_max,
1814 		.maxlen		= sizeof(unsigned int),
1815 		.mode		= 0644,
1816 		.proc_handler	= sysctl_sched_uclamp_handler,
1817 	},
1818 	{
1819 		.procname	= "sched_util_clamp_min_rt_default",
1820 		.data		= &sysctl_sched_uclamp_util_min_rt_default,
1821 		.maxlen		= sizeof(unsigned int),
1822 		.mode		= 0644,
1823 		.proc_handler	= sysctl_sched_uclamp_handler,
1824 	},
1825 #endif
1826 #ifdef CONFIG_SCHED_AUTOGROUP
1827 	{
1828 		.procname	= "sched_autogroup_enabled",
1829 		.data		= &sysctl_sched_autogroup_enabled,
1830 		.maxlen		= sizeof(unsigned int),
1831 		.mode		= 0644,
1832 		.proc_handler	= proc_dointvec_minmax,
1833 		.extra1		= SYSCTL_ZERO,
1834 		.extra2		= SYSCTL_ONE,
1835 	},
1836 #endif
1837 #ifdef CONFIG_CFS_BANDWIDTH
1838 	{
1839 		.procname	= "sched_cfs_bandwidth_slice_us",
1840 		.data		= &sysctl_sched_cfs_bandwidth_slice,
1841 		.maxlen		= sizeof(unsigned int),
1842 		.mode		= 0644,
1843 		.proc_handler	= proc_dointvec_minmax,
1844 		.extra1		= SYSCTL_ONE,
1845 	},
1846 #endif
1847 #if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_CPU_FREQ_GOV_SCHEDUTIL)
1848 	{
1849 		.procname	= "sched_energy_aware",
1850 		.data		= &sysctl_sched_energy_aware,
1851 		.maxlen		= sizeof(unsigned int),
1852 		.mode		= 0644,
1853 		.proc_handler	= sched_energy_aware_handler,
1854 		.extra1		= SYSCTL_ZERO,
1855 		.extra2		= SYSCTL_ONE,
1856 	},
1857 #endif
1858 #ifdef CONFIG_PROVE_LOCKING
1859 	{
1860 		.procname	= "prove_locking",
1861 		.data		= &prove_locking,
1862 		.maxlen		= sizeof(int),
1863 		.mode		= 0644,
1864 		.proc_handler	= proc_dointvec,
1865 	},
1866 #endif
1867 #ifdef CONFIG_LOCK_STAT
1868 	{
1869 		.procname	= "lock_stat",
1870 		.data		= &lock_stat,
1871 		.maxlen		= sizeof(int),
1872 		.mode		= 0644,
1873 		.proc_handler	= proc_dointvec,
1874 	},
1875 #endif
1876 	{
1877 		.procname	= "panic",
1878 		.data		= &panic_timeout,
1879 		.maxlen		= sizeof(int),
1880 		.mode		= 0644,
1881 		.proc_handler	= proc_dointvec,
1882 	},
1883 #ifdef CONFIG_COREDUMP
1884 	{
1885 		.procname	= "core_uses_pid",
1886 		.data		= &core_uses_pid,
1887 		.maxlen		= sizeof(int),
1888 		.mode		= 0644,
1889 		.proc_handler	= proc_dointvec,
1890 	},
1891 	{
1892 		.procname	= "core_pattern",
1893 		.data		= core_pattern,
1894 		.maxlen		= CORENAME_MAX_SIZE,
1895 		.mode		= 0644,
1896 		.proc_handler	= proc_dostring_coredump,
1897 	},
1898 	{
1899 		.procname	= "core_pipe_limit",
1900 		.data		= &core_pipe_limit,
1901 		.maxlen		= sizeof(unsigned int),
1902 		.mode		= 0644,
1903 		.proc_handler	= proc_dointvec,
1904 	},
1905 #endif
1906 #ifdef CONFIG_PROC_SYSCTL
1907 	{
1908 		.procname	= "tainted",
1909 		.maxlen 	= sizeof(long),
1910 		.mode		= 0644,
1911 		.proc_handler	= proc_taint,
1912 	},
1913 	{
1914 		.procname	= "sysctl_writes_strict",
1915 		.data		= &sysctl_writes_strict,
1916 		.maxlen		= sizeof(int),
1917 		.mode		= 0644,
1918 		.proc_handler	= proc_dointvec_minmax,
1919 		.extra1		= &neg_one,
1920 		.extra2		= SYSCTL_ONE,
1921 	},
1922 #endif
1923 #ifdef CONFIG_LATENCYTOP
1924 	{
1925 		.procname	= "latencytop",
1926 		.data		= &latencytop_enabled,
1927 		.maxlen		= sizeof(int),
1928 		.mode		= 0644,
1929 		.proc_handler	= sysctl_latencytop,
1930 	},
1931 #endif
1932 #ifdef CONFIG_BLK_DEV_INITRD
1933 	{
1934 		.procname	= "real-root-dev",
1935 		.data		= &real_root_dev,
1936 		.maxlen		= sizeof(int),
1937 		.mode		= 0644,
1938 		.proc_handler	= proc_dointvec,
1939 	},
1940 #endif
1941 	{
1942 		.procname	= "print-fatal-signals",
1943 		.data		= &print_fatal_signals,
1944 		.maxlen		= sizeof(int),
1945 		.mode		= 0644,
1946 		.proc_handler	= proc_dointvec,
1947 	},
1948 #ifdef CONFIG_SPARC
1949 	{
1950 		.procname	= "reboot-cmd",
1951 		.data		= reboot_command,
1952 		.maxlen		= 256,
1953 		.mode		= 0644,
1954 		.proc_handler	= proc_dostring,
1955 	},
1956 	{
1957 		.procname	= "stop-a",
1958 		.data		= &stop_a_enabled,
1959 		.maxlen		= sizeof (int),
1960 		.mode		= 0644,
1961 		.proc_handler	= proc_dointvec,
1962 	},
1963 	{
1964 		.procname	= "scons-poweroff",
1965 		.data		= &scons_pwroff,
1966 		.maxlen		= sizeof (int),
1967 		.mode		= 0644,
1968 		.proc_handler	= proc_dointvec,
1969 	},
1970 #endif
1971 #ifdef CONFIG_SPARC64
1972 	{
1973 		.procname	= "tsb-ratio",
1974 		.data		= &sysctl_tsb_ratio,
1975 		.maxlen		= sizeof (int),
1976 		.mode		= 0644,
1977 		.proc_handler	= proc_dointvec,
1978 	},
1979 #endif
1980 #ifdef CONFIG_PARISC
1981 	{
1982 		.procname	= "soft-power",
1983 		.data		= &pwrsw_enabled,
1984 		.maxlen		= sizeof (int),
1985 		.mode		= 0644,
1986 		.proc_handler	= proc_dointvec,
1987 	},
1988 #endif
1989 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
1990 	{
1991 		.procname	= "unaligned-trap",
1992 		.data		= &unaligned_enabled,
1993 		.maxlen		= sizeof (int),
1994 		.mode		= 0644,
1995 		.proc_handler	= proc_dointvec,
1996 	},
1997 #endif
1998 	{
1999 		.procname	= "ctrl-alt-del",
2000 		.data		= &C_A_D,
2001 		.maxlen		= sizeof(int),
2002 		.mode		= 0644,
2003 		.proc_handler	= proc_dointvec,
2004 	},
2005 #ifdef CONFIG_FUNCTION_TRACER
2006 	{
2007 		.procname	= "ftrace_enabled",
2008 		.data		= &ftrace_enabled,
2009 		.maxlen		= sizeof(int),
2010 		.mode		= 0644,
2011 		.proc_handler	= ftrace_enable_sysctl,
2012 	},
2013 #endif
2014 #ifdef CONFIG_STACK_TRACER
2015 	{
2016 		.procname	= "stack_tracer_enabled",
2017 		.data		= &stack_tracer_enabled,
2018 		.maxlen		= sizeof(int),
2019 		.mode		= 0644,
2020 		.proc_handler	= stack_trace_sysctl,
2021 	},
2022 #endif
2023 #ifdef CONFIG_TRACING
2024 	{
2025 		.procname	= "ftrace_dump_on_oops",
2026 		.data		= &ftrace_dump_on_oops,
2027 		.maxlen		= sizeof(int),
2028 		.mode		= 0644,
2029 		.proc_handler	= proc_dointvec,
2030 	},
2031 	{
2032 		.procname	= "traceoff_on_warning",
2033 		.data		= &__disable_trace_on_warning,
2034 		.maxlen		= sizeof(__disable_trace_on_warning),
2035 		.mode		= 0644,
2036 		.proc_handler	= proc_dointvec,
2037 	},
2038 	{
2039 		.procname	= "tracepoint_printk",
2040 		.data		= &tracepoint_printk,
2041 		.maxlen		= sizeof(tracepoint_printk),
2042 		.mode		= 0644,
2043 		.proc_handler	= tracepoint_printk_sysctl,
2044 	},
2045 #endif
2046 #ifdef CONFIG_KEXEC_CORE
2047 	{
2048 		.procname	= "kexec_load_disabled",
2049 		.data		= &kexec_load_disabled,
2050 		.maxlen		= sizeof(int),
2051 		.mode		= 0644,
2052 		/* only handle a transition from default "0" to "1" */
2053 		.proc_handler	= proc_dointvec_minmax,
2054 		.extra1		= SYSCTL_ONE,
2055 		.extra2		= SYSCTL_ONE,
2056 	},
2057 #endif
2058 #ifdef CONFIG_MODULES
2059 	{
2060 		.procname	= "modprobe",
2061 		.data		= &modprobe_path,
2062 		.maxlen		= KMOD_PATH_LEN,
2063 		.mode		= 0644,
2064 		.proc_handler	= proc_dostring,
2065 	},
2066 	{
2067 		.procname	= "modules_disabled",
2068 		.data		= &modules_disabled,
2069 		.maxlen		= sizeof(int),
2070 		.mode		= 0644,
2071 		/* only handle a transition from default "0" to "1" */
2072 		.proc_handler	= proc_dointvec_minmax,
2073 		.extra1		= SYSCTL_ONE,
2074 		.extra2		= SYSCTL_ONE,
2075 	},
2076 #endif
2077 #ifdef CONFIG_UEVENT_HELPER
2078 	{
2079 		.procname	= "hotplug",
2080 		.data		= &uevent_helper,
2081 		.maxlen		= UEVENT_HELPER_PATH_LEN,
2082 		.mode		= 0644,
2083 		.proc_handler	= proc_dostring,
2084 	},
2085 #endif
2086 #ifdef CONFIG_CHR_DEV_SG
2087 	{
2088 		.procname	= "sg-big-buff",
2089 		.data		= &sg_big_buff,
2090 		.maxlen		= sizeof (int),
2091 		.mode		= 0444,
2092 		.proc_handler	= proc_dointvec,
2093 	},
2094 #endif
2095 #ifdef CONFIG_BSD_PROCESS_ACCT
2096 	{
2097 		.procname	= "acct",
2098 		.data		= &acct_parm,
2099 		.maxlen		= 3*sizeof(int),
2100 		.mode		= 0644,
2101 		.proc_handler	= proc_dointvec,
2102 	},
2103 #endif
2104 #ifdef CONFIG_MAGIC_SYSRQ
2105 	{
2106 		.procname	= "sysrq",
2107 		.data		= NULL,
2108 		.maxlen		= sizeof (int),
2109 		.mode		= 0644,
2110 		.proc_handler	= sysrq_sysctl_handler,
2111 	},
2112 #endif
2113 #ifdef CONFIG_PROC_SYSCTL
2114 	{
2115 		.procname	= "cad_pid",
2116 		.data		= NULL,
2117 		.maxlen		= sizeof (int),
2118 		.mode		= 0600,
2119 		.proc_handler	= proc_do_cad_pid,
2120 	},
2121 #endif
2122 	{
2123 		.procname	= "threads-max",
2124 		.data		= NULL,
2125 		.maxlen		= sizeof(int),
2126 		.mode		= 0644,
2127 		.proc_handler	= sysctl_max_threads,
2128 	},
2129 	{
2130 		.procname	= "random",
2131 		.mode		= 0555,
2132 		.child		= random_table,
2133 	},
2134 	{
2135 		.procname	= "usermodehelper",
2136 		.mode		= 0555,
2137 		.child		= usermodehelper_table,
2138 	},
2139 #ifdef CONFIG_FW_LOADER_USER_HELPER
2140 	{
2141 		.procname	= "firmware_config",
2142 		.mode		= 0555,
2143 		.child		= firmware_config_table,
2144 	},
2145 #endif
2146 	{
2147 		.procname	= "overflowuid",
2148 		.data		= &overflowuid,
2149 		.maxlen		= sizeof(int),
2150 		.mode		= 0644,
2151 		.proc_handler	= proc_dointvec_minmax,
2152 		.extra1		= &minolduid,
2153 		.extra2		= &maxolduid,
2154 	},
2155 	{
2156 		.procname	= "overflowgid",
2157 		.data		= &overflowgid,
2158 		.maxlen		= sizeof(int),
2159 		.mode		= 0644,
2160 		.proc_handler	= proc_dointvec_minmax,
2161 		.extra1		= &minolduid,
2162 		.extra2		= &maxolduid,
2163 	},
2164 #ifdef CONFIG_S390
2165 	{
2166 		.procname	= "userprocess_debug",
2167 		.data		= &show_unhandled_signals,
2168 		.maxlen		= sizeof(int),
2169 		.mode		= 0644,
2170 		.proc_handler	= proc_dointvec,
2171 	},
2172 #endif
2173 #ifdef CONFIG_SMP
2174 	{
2175 		.procname	= "oops_all_cpu_backtrace",
2176 		.data		= &sysctl_oops_all_cpu_backtrace,
2177 		.maxlen		= sizeof(int),
2178 		.mode		= 0644,
2179 		.proc_handler	= proc_dointvec_minmax,
2180 		.extra1		= SYSCTL_ZERO,
2181 		.extra2		= SYSCTL_ONE,
2182 	},
2183 #endif /* CONFIG_SMP */
2184 	{
2185 		.procname	= "pid_max",
2186 		.data		= &pid_max,
2187 		.maxlen		= sizeof (int),
2188 		.mode		= 0644,
2189 		.proc_handler	= proc_dointvec_minmax,
2190 		.extra1		= &pid_max_min,
2191 		.extra2		= &pid_max_max,
2192 	},
2193 	{
2194 		.procname	= "panic_on_oops",
2195 		.data		= &panic_on_oops,
2196 		.maxlen		= sizeof(int),
2197 		.mode		= 0644,
2198 		.proc_handler	= proc_dointvec,
2199 	},
2200 	{
2201 		.procname	= "panic_print",
2202 		.data		= &panic_print,
2203 		.maxlen		= sizeof(unsigned long),
2204 		.mode		= 0644,
2205 		.proc_handler	= proc_doulongvec_minmax,
2206 	},
2207 #if defined CONFIG_PRINTK
2208 	{
2209 		.procname	= "printk",
2210 		.data		= &console_loglevel,
2211 		.maxlen		= 4*sizeof(int),
2212 		.mode		= 0644,
2213 		.proc_handler	= proc_dointvec,
2214 	},
2215 	{
2216 		.procname	= "printk_ratelimit",
2217 		.data		= &printk_ratelimit_state.interval,
2218 		.maxlen		= sizeof(int),
2219 		.mode		= 0644,
2220 		.proc_handler	= proc_dointvec_jiffies,
2221 	},
2222 	{
2223 		.procname	= "printk_ratelimit_burst",
2224 		.data		= &printk_ratelimit_state.burst,
2225 		.maxlen		= sizeof(int),
2226 		.mode		= 0644,
2227 		.proc_handler	= proc_dointvec,
2228 	},
2229 	{
2230 		.procname	= "printk_delay",
2231 		.data		= &printk_delay_msec,
2232 		.maxlen		= sizeof(int),
2233 		.mode		= 0644,
2234 		.proc_handler	= proc_dointvec_minmax,
2235 		.extra1		= SYSCTL_ZERO,
2236 		.extra2		= &ten_thousand,
2237 	},
2238 	{
2239 		.procname	= "printk_devkmsg",
2240 		.data		= devkmsg_log_str,
2241 		.maxlen		= DEVKMSG_STR_MAX_SIZE,
2242 		.mode		= 0644,
2243 		.proc_handler	= devkmsg_sysctl_set_loglvl,
2244 	},
2245 	{
2246 		.procname	= "dmesg_restrict",
2247 		.data		= &dmesg_restrict,
2248 		.maxlen		= sizeof(int),
2249 		.mode		= 0644,
2250 		.proc_handler	= proc_dointvec_minmax_sysadmin,
2251 		.extra1		= SYSCTL_ZERO,
2252 		.extra2		= SYSCTL_ONE,
2253 	},
2254 	{
2255 		.procname	= "kptr_restrict",
2256 		.data		= &kptr_restrict,
2257 		.maxlen		= sizeof(int),
2258 		.mode		= 0644,
2259 		.proc_handler	= proc_dointvec_minmax_sysadmin,
2260 		.extra1		= SYSCTL_ZERO,
2261 		.extra2		= &two,
2262 	},
2263 #endif
2264 	{
2265 		.procname	= "ngroups_max",
2266 		.data		= &ngroups_max,
2267 		.maxlen		= sizeof (int),
2268 		.mode		= 0444,
2269 		.proc_handler	= proc_dointvec,
2270 	},
2271 	{
2272 		.procname	= "cap_last_cap",
2273 		.data		= (void *)&cap_last_cap,
2274 		.maxlen		= sizeof(int),
2275 		.mode		= 0444,
2276 		.proc_handler	= proc_dointvec,
2277 	},
2278 #if defined(CONFIG_LOCKUP_DETECTOR)
2279 	{
2280 		.procname       = "watchdog",
2281 		.data		= &watchdog_user_enabled,
2282 		.maxlen		= sizeof(int),
2283 		.mode		= 0644,
2284 		.proc_handler   = proc_watchdog,
2285 		.extra1		= SYSCTL_ZERO,
2286 		.extra2		= SYSCTL_ONE,
2287 	},
2288 	{
2289 		.procname	= "watchdog_thresh",
2290 		.data		= &watchdog_thresh,
2291 		.maxlen		= sizeof(int),
2292 		.mode		= 0644,
2293 		.proc_handler	= proc_watchdog_thresh,
2294 		.extra1		= SYSCTL_ZERO,
2295 		.extra2		= &sixty,
2296 	},
2297 	{
2298 		.procname       = "nmi_watchdog",
2299 		.data		= &nmi_watchdog_user_enabled,
2300 		.maxlen		= sizeof(int),
2301 		.mode		= NMI_WATCHDOG_SYSCTL_PERM,
2302 		.proc_handler   = proc_nmi_watchdog,
2303 		.extra1		= SYSCTL_ZERO,
2304 		.extra2		= SYSCTL_ONE,
2305 	},
2306 	{
2307 		.procname	= "watchdog_cpumask",
2308 		.data		= &watchdog_cpumask_bits,
2309 		.maxlen		= NR_CPUS,
2310 		.mode		= 0644,
2311 		.proc_handler	= proc_watchdog_cpumask,
2312 	},
2313 #ifdef CONFIG_SOFTLOCKUP_DETECTOR
2314 	{
2315 		.procname       = "soft_watchdog",
2316 		.data		= &soft_watchdog_user_enabled,
2317 		.maxlen		= sizeof(int),
2318 		.mode		= 0644,
2319 		.proc_handler   = proc_soft_watchdog,
2320 		.extra1		= SYSCTL_ZERO,
2321 		.extra2		= SYSCTL_ONE,
2322 	},
2323 	{
2324 		.procname	= "softlockup_panic",
2325 		.data		= &softlockup_panic,
2326 		.maxlen		= sizeof(int),
2327 		.mode		= 0644,
2328 		.proc_handler	= proc_dointvec_minmax,
2329 		.extra1		= SYSCTL_ZERO,
2330 		.extra2		= SYSCTL_ONE,
2331 	},
2332 #ifdef CONFIG_SMP
2333 	{
2334 		.procname	= "softlockup_all_cpu_backtrace",
2335 		.data		= &sysctl_softlockup_all_cpu_backtrace,
2336 		.maxlen		= sizeof(int),
2337 		.mode		= 0644,
2338 		.proc_handler	= proc_dointvec_minmax,
2339 		.extra1		= SYSCTL_ZERO,
2340 		.extra2		= SYSCTL_ONE,
2341 	},
2342 #endif /* CONFIG_SMP */
2343 #endif
2344 #ifdef CONFIG_HARDLOCKUP_DETECTOR
2345 	{
2346 		.procname	= "hardlockup_panic",
2347 		.data		= &hardlockup_panic,
2348 		.maxlen		= sizeof(int),
2349 		.mode		= 0644,
2350 		.proc_handler	= proc_dointvec_minmax,
2351 		.extra1		= SYSCTL_ZERO,
2352 		.extra2		= SYSCTL_ONE,
2353 	},
2354 #ifdef CONFIG_SMP
2355 	{
2356 		.procname	= "hardlockup_all_cpu_backtrace",
2357 		.data		= &sysctl_hardlockup_all_cpu_backtrace,
2358 		.maxlen		= sizeof(int),
2359 		.mode		= 0644,
2360 		.proc_handler	= proc_dointvec_minmax,
2361 		.extra1		= SYSCTL_ZERO,
2362 		.extra2		= SYSCTL_ONE,
2363 	},
2364 #endif /* CONFIG_SMP */
2365 #endif
2366 #endif
2367 
2368 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
2369 	{
2370 		.procname       = "unknown_nmi_panic",
2371 		.data           = &unknown_nmi_panic,
2372 		.maxlen         = sizeof (int),
2373 		.mode           = 0644,
2374 		.proc_handler   = proc_dointvec,
2375 	},
2376 #endif
2377 
2378 #if (defined(CONFIG_X86_32) || defined(CONFIG_PARISC)) && \
2379 	defined(CONFIG_DEBUG_STACKOVERFLOW)
2380 	{
2381 		.procname	= "panic_on_stackoverflow",
2382 		.data		= &sysctl_panic_on_stackoverflow,
2383 		.maxlen		= sizeof(int),
2384 		.mode		= 0644,
2385 		.proc_handler	= proc_dointvec,
2386 	},
2387 #endif
2388 #if defined(CONFIG_X86)
2389 	{
2390 		.procname	= "panic_on_unrecovered_nmi",
2391 		.data		= &panic_on_unrecovered_nmi,
2392 		.maxlen		= sizeof(int),
2393 		.mode		= 0644,
2394 		.proc_handler	= proc_dointvec,
2395 	},
2396 	{
2397 		.procname	= "panic_on_io_nmi",
2398 		.data		= &panic_on_io_nmi,
2399 		.maxlen		= sizeof(int),
2400 		.mode		= 0644,
2401 		.proc_handler	= proc_dointvec,
2402 	},
2403 	{
2404 		.procname	= "bootloader_type",
2405 		.data		= &bootloader_type,
2406 		.maxlen		= sizeof (int),
2407 		.mode		= 0444,
2408 		.proc_handler	= proc_dointvec,
2409 	},
2410 	{
2411 		.procname	= "bootloader_version",
2412 		.data		= &bootloader_version,
2413 		.maxlen		= sizeof (int),
2414 		.mode		= 0444,
2415 		.proc_handler	= proc_dointvec,
2416 	},
2417 	{
2418 		.procname	= "io_delay_type",
2419 		.data		= &io_delay_type,
2420 		.maxlen		= sizeof(int),
2421 		.mode		= 0644,
2422 		.proc_handler	= proc_dointvec,
2423 	},
2424 #endif
2425 #if defined(CONFIG_MMU)
2426 	{
2427 		.procname	= "randomize_va_space",
2428 		.data		= &randomize_va_space,
2429 		.maxlen		= sizeof(int),
2430 		.mode		= 0644,
2431 		.proc_handler	= proc_dointvec,
2432 	},
2433 #endif
2434 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
2435 	{
2436 		.procname	= "spin_retry",
2437 		.data		= &spin_retry,
2438 		.maxlen		= sizeof (int),
2439 		.mode		= 0644,
2440 		.proc_handler	= proc_dointvec,
2441 	},
2442 #endif
2443 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
2444 	{
2445 		.procname	= "acpi_video_flags",
2446 		.data		= &acpi_realmode_flags,
2447 		.maxlen		= sizeof (unsigned long),
2448 		.mode		= 0644,
2449 		.proc_handler	= proc_doulongvec_minmax,
2450 	},
2451 #endif
2452 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
2453 	{
2454 		.procname	= "ignore-unaligned-usertrap",
2455 		.data		= &no_unaligned_warning,
2456 		.maxlen		= sizeof (int),
2457 		.mode		= 0644,
2458 		.proc_handler	= proc_dointvec,
2459 	},
2460 #endif
2461 #ifdef CONFIG_IA64
2462 	{
2463 		.procname	= "unaligned-dump-stack",
2464 		.data		= &unaligned_dump_stack,
2465 		.maxlen		= sizeof (int),
2466 		.mode		= 0644,
2467 		.proc_handler	= proc_dointvec,
2468 	},
2469 #endif
2470 #ifdef CONFIG_DETECT_HUNG_TASK
2471 #ifdef CONFIG_SMP
2472 	{
2473 		.procname	= "hung_task_all_cpu_backtrace",
2474 		.data		= &sysctl_hung_task_all_cpu_backtrace,
2475 		.maxlen		= sizeof(int),
2476 		.mode		= 0644,
2477 		.proc_handler	= proc_dointvec_minmax,
2478 		.extra1		= SYSCTL_ZERO,
2479 		.extra2		= SYSCTL_ONE,
2480 	},
2481 #endif /* CONFIG_SMP */
2482 	{
2483 		.procname	= "hung_task_panic",
2484 		.data		= &sysctl_hung_task_panic,
2485 		.maxlen		= sizeof(int),
2486 		.mode		= 0644,
2487 		.proc_handler	= proc_dointvec_minmax,
2488 		.extra1		= SYSCTL_ZERO,
2489 		.extra2		= SYSCTL_ONE,
2490 	},
2491 	{
2492 		.procname	= "hung_task_check_count",
2493 		.data		= &sysctl_hung_task_check_count,
2494 		.maxlen		= sizeof(int),
2495 		.mode		= 0644,
2496 		.proc_handler	= proc_dointvec_minmax,
2497 		.extra1		= SYSCTL_ZERO,
2498 	},
2499 	{
2500 		.procname	= "hung_task_timeout_secs",
2501 		.data		= &sysctl_hung_task_timeout_secs,
2502 		.maxlen		= sizeof(unsigned long),
2503 		.mode		= 0644,
2504 		.proc_handler	= proc_dohung_task_timeout_secs,
2505 		.extra2		= &hung_task_timeout_max,
2506 	},
2507 	{
2508 		.procname	= "hung_task_check_interval_secs",
2509 		.data		= &sysctl_hung_task_check_interval_secs,
2510 		.maxlen		= sizeof(unsigned long),
2511 		.mode		= 0644,
2512 		.proc_handler	= proc_dohung_task_timeout_secs,
2513 		.extra2		= &hung_task_timeout_max,
2514 	},
2515 	{
2516 		.procname	= "hung_task_warnings",
2517 		.data		= &sysctl_hung_task_warnings,
2518 		.maxlen		= sizeof(int),
2519 		.mode		= 0644,
2520 		.proc_handler	= proc_dointvec_minmax,
2521 		.extra1		= &neg_one,
2522 	},
2523 #endif
2524 #ifdef CONFIG_RT_MUTEXES
2525 	{
2526 		.procname	= "max_lock_depth",
2527 		.data		= &max_lock_depth,
2528 		.maxlen		= sizeof(int),
2529 		.mode		= 0644,
2530 		.proc_handler	= proc_dointvec,
2531 	},
2532 #endif
2533 	{
2534 		.procname	= "poweroff_cmd",
2535 		.data		= &poweroff_cmd,
2536 		.maxlen		= POWEROFF_CMD_PATH_LEN,
2537 		.mode		= 0644,
2538 		.proc_handler	= proc_dostring,
2539 	},
2540 #ifdef CONFIG_KEYS
2541 	{
2542 		.procname	= "keys",
2543 		.mode		= 0555,
2544 		.child		= key_sysctls,
2545 	},
2546 #endif
2547 #ifdef CONFIG_PERF_EVENTS
2548 	/*
2549 	 * User-space scripts rely on the existence of this file
2550 	 * as a feature check for perf_events being enabled.
2551 	 *
2552 	 * So it's an ABI, do not remove!
2553 	 */
2554 	{
2555 		.procname	= "perf_event_paranoid",
2556 		.data		= &sysctl_perf_event_paranoid,
2557 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
2558 		.mode		= 0644,
2559 		.proc_handler	= proc_dointvec,
2560 	},
2561 	{
2562 		.procname	= "perf_event_mlock_kb",
2563 		.data		= &sysctl_perf_event_mlock,
2564 		.maxlen		= sizeof(sysctl_perf_event_mlock),
2565 		.mode		= 0644,
2566 		.proc_handler	= proc_dointvec,
2567 	},
2568 	{
2569 		.procname	= "perf_event_max_sample_rate",
2570 		.data		= &sysctl_perf_event_sample_rate,
2571 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
2572 		.mode		= 0644,
2573 		.proc_handler	= perf_proc_update_handler,
2574 		.extra1		= SYSCTL_ONE,
2575 	},
2576 	{
2577 		.procname	= "perf_cpu_time_max_percent",
2578 		.data		= &sysctl_perf_cpu_time_max_percent,
2579 		.maxlen		= sizeof(sysctl_perf_cpu_time_max_percent),
2580 		.mode		= 0644,
2581 		.proc_handler	= perf_cpu_time_max_percent_handler,
2582 		.extra1		= SYSCTL_ZERO,
2583 		.extra2		= &one_hundred,
2584 	},
2585 	{
2586 		.procname	= "perf_event_max_stack",
2587 		.data		= &sysctl_perf_event_max_stack,
2588 		.maxlen		= sizeof(sysctl_perf_event_max_stack),
2589 		.mode		= 0644,
2590 		.proc_handler	= perf_event_max_stack_handler,
2591 		.extra1		= SYSCTL_ZERO,
2592 		.extra2		= &six_hundred_forty_kb,
2593 	},
2594 	{
2595 		.procname	= "perf_event_max_contexts_per_stack",
2596 		.data		= &sysctl_perf_event_max_contexts_per_stack,
2597 		.maxlen		= sizeof(sysctl_perf_event_max_contexts_per_stack),
2598 		.mode		= 0644,
2599 		.proc_handler	= perf_event_max_stack_handler,
2600 		.extra1		= SYSCTL_ZERO,
2601 		.extra2		= &one_thousand,
2602 	},
2603 #endif
2604 	{
2605 		.procname	= "panic_on_warn",
2606 		.data		= &panic_on_warn,
2607 		.maxlen		= sizeof(int),
2608 		.mode		= 0644,
2609 		.proc_handler	= proc_dointvec_minmax,
2610 		.extra1		= SYSCTL_ZERO,
2611 		.extra2		= SYSCTL_ONE,
2612 	},
2613 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
2614 	{
2615 		.procname	= "timer_migration",
2616 		.data		= &sysctl_timer_migration,
2617 		.maxlen		= sizeof(unsigned int),
2618 		.mode		= 0644,
2619 		.proc_handler	= timer_migration_handler,
2620 		.extra1		= SYSCTL_ZERO,
2621 		.extra2		= SYSCTL_ONE,
2622 	},
2623 #endif
2624 #ifdef CONFIG_BPF_SYSCALL
2625 	{
2626 		.procname	= "unprivileged_bpf_disabled",
2627 		.data		= &sysctl_unprivileged_bpf_disabled,
2628 		.maxlen		= sizeof(sysctl_unprivileged_bpf_disabled),
2629 		.mode		= 0644,
2630 		/* only handle a transition from default "0" to "1" */
2631 		.proc_handler	= proc_dointvec_minmax,
2632 		.extra1		= SYSCTL_ONE,
2633 		.extra2		= SYSCTL_ONE,
2634 	},
2635 	{
2636 		.procname	= "bpf_stats_enabled",
2637 		.data		= &bpf_stats_enabled_key.key,
2638 		.maxlen		= sizeof(bpf_stats_enabled_key),
2639 		.mode		= 0644,
2640 		.proc_handler	= bpf_stats_handler,
2641 	},
2642 #endif
2643 #if defined(CONFIG_TREE_RCU)
2644 	{
2645 		.procname	= "panic_on_rcu_stall",
2646 		.data		= &sysctl_panic_on_rcu_stall,
2647 		.maxlen		= sizeof(sysctl_panic_on_rcu_stall),
2648 		.mode		= 0644,
2649 		.proc_handler	= proc_dointvec_minmax,
2650 		.extra1		= SYSCTL_ZERO,
2651 		.extra2		= SYSCTL_ONE,
2652 	},
2653 #endif
2654 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
2655 	{
2656 		.procname	= "stack_erasing",
2657 		.data		= NULL,
2658 		.maxlen		= sizeof(int),
2659 		.mode		= 0600,
2660 		.proc_handler	= stack_erasing_sysctl,
2661 		.extra1		= SYSCTL_ZERO,
2662 		.extra2		= SYSCTL_ONE,
2663 	},
2664 #endif
2665 	{ }
2666 };
2667 
2668 static struct ctl_table vm_table[] = {
2669 	{
2670 		.procname	= "overcommit_memory",
2671 		.data		= &sysctl_overcommit_memory,
2672 		.maxlen		= sizeof(sysctl_overcommit_memory),
2673 		.mode		= 0644,
2674 		.proc_handler	= overcommit_policy_handler,
2675 		.extra1		= SYSCTL_ZERO,
2676 		.extra2		= &two,
2677 	},
2678 	{
2679 		.procname	= "panic_on_oom",
2680 		.data		= &sysctl_panic_on_oom,
2681 		.maxlen		= sizeof(sysctl_panic_on_oom),
2682 		.mode		= 0644,
2683 		.proc_handler	= proc_dointvec_minmax,
2684 		.extra1		= SYSCTL_ZERO,
2685 		.extra2		= &two,
2686 	},
2687 	{
2688 		.procname	= "oom_kill_allocating_task",
2689 		.data		= &sysctl_oom_kill_allocating_task,
2690 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
2691 		.mode		= 0644,
2692 		.proc_handler	= proc_dointvec,
2693 	},
2694 	{
2695 		.procname	= "oom_dump_tasks",
2696 		.data		= &sysctl_oom_dump_tasks,
2697 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
2698 		.mode		= 0644,
2699 		.proc_handler	= proc_dointvec,
2700 	},
2701 	{
2702 		.procname	= "overcommit_ratio",
2703 		.data		= &sysctl_overcommit_ratio,
2704 		.maxlen		= sizeof(sysctl_overcommit_ratio),
2705 		.mode		= 0644,
2706 		.proc_handler	= overcommit_ratio_handler,
2707 	},
2708 	{
2709 		.procname	= "overcommit_kbytes",
2710 		.data		= &sysctl_overcommit_kbytes,
2711 		.maxlen		= sizeof(sysctl_overcommit_kbytes),
2712 		.mode		= 0644,
2713 		.proc_handler	= overcommit_kbytes_handler,
2714 	},
2715 	{
2716 		.procname	= "page-cluster",
2717 		.data		= &page_cluster,
2718 		.maxlen		= sizeof(int),
2719 		.mode		= 0644,
2720 		.proc_handler	= proc_dointvec_minmax,
2721 		.extra1		= SYSCTL_ZERO,
2722 	},
2723 	{
2724 		.procname	= "dirty_background_ratio",
2725 		.data		= &dirty_background_ratio,
2726 		.maxlen		= sizeof(dirty_background_ratio),
2727 		.mode		= 0644,
2728 		.proc_handler	= dirty_background_ratio_handler,
2729 		.extra1		= SYSCTL_ZERO,
2730 		.extra2		= &one_hundred,
2731 	},
2732 	{
2733 		.procname	= "dirty_background_bytes",
2734 		.data		= &dirty_background_bytes,
2735 		.maxlen		= sizeof(dirty_background_bytes),
2736 		.mode		= 0644,
2737 		.proc_handler	= dirty_background_bytes_handler,
2738 		.extra1		= &one_ul,
2739 	},
2740 	{
2741 		.procname	= "dirty_ratio",
2742 		.data		= &vm_dirty_ratio,
2743 		.maxlen		= sizeof(vm_dirty_ratio),
2744 		.mode		= 0644,
2745 		.proc_handler	= dirty_ratio_handler,
2746 		.extra1		= SYSCTL_ZERO,
2747 		.extra2		= &one_hundred,
2748 	},
2749 	{
2750 		.procname	= "dirty_bytes",
2751 		.data		= &vm_dirty_bytes,
2752 		.maxlen		= sizeof(vm_dirty_bytes),
2753 		.mode		= 0644,
2754 		.proc_handler	= dirty_bytes_handler,
2755 		.extra1		= &dirty_bytes_min,
2756 	},
2757 	{
2758 		.procname	= "dirty_writeback_centisecs",
2759 		.data		= &dirty_writeback_interval,
2760 		.maxlen		= sizeof(dirty_writeback_interval),
2761 		.mode		= 0644,
2762 		.proc_handler	= dirty_writeback_centisecs_handler,
2763 	},
2764 	{
2765 		.procname	= "dirty_expire_centisecs",
2766 		.data		= &dirty_expire_interval,
2767 		.maxlen		= sizeof(dirty_expire_interval),
2768 		.mode		= 0644,
2769 		.proc_handler	= proc_dointvec_minmax,
2770 		.extra1		= SYSCTL_ZERO,
2771 	},
2772 	{
2773 		.procname	= "dirtytime_expire_seconds",
2774 		.data		= &dirtytime_expire_interval,
2775 		.maxlen		= sizeof(dirtytime_expire_interval),
2776 		.mode		= 0644,
2777 		.proc_handler	= dirtytime_interval_handler,
2778 		.extra1		= SYSCTL_ZERO,
2779 	},
2780 	{
2781 		.procname	= "swappiness",
2782 		.data		= &vm_swappiness,
2783 		.maxlen		= sizeof(vm_swappiness),
2784 		.mode		= 0644,
2785 		.proc_handler	= proc_dointvec_minmax,
2786 		.extra1		= SYSCTL_ZERO,
2787 		.extra2		= &two_hundred,
2788 	},
2789 #ifdef CONFIG_HUGETLB_PAGE
2790 	{
2791 		.procname	= "nr_hugepages",
2792 		.data		= NULL,
2793 		.maxlen		= sizeof(unsigned long),
2794 		.mode		= 0644,
2795 		.proc_handler	= hugetlb_sysctl_handler,
2796 	},
2797 #ifdef CONFIG_NUMA
2798 	{
2799 		.procname       = "nr_hugepages_mempolicy",
2800 		.data           = NULL,
2801 		.maxlen         = sizeof(unsigned long),
2802 		.mode           = 0644,
2803 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
2804 	},
2805 	{
2806 		.procname		= "numa_stat",
2807 		.data			= &sysctl_vm_numa_stat,
2808 		.maxlen			= sizeof(int),
2809 		.mode			= 0644,
2810 		.proc_handler	= sysctl_vm_numa_stat_handler,
2811 		.extra1			= SYSCTL_ZERO,
2812 		.extra2			= SYSCTL_ONE,
2813 	},
2814 #endif
2815 	 {
2816 		.procname	= "hugetlb_shm_group",
2817 		.data		= &sysctl_hugetlb_shm_group,
2818 		.maxlen		= sizeof(gid_t),
2819 		.mode		= 0644,
2820 		.proc_handler	= proc_dointvec,
2821 	 },
2822 	{
2823 		.procname	= "nr_overcommit_hugepages",
2824 		.data		= NULL,
2825 		.maxlen		= sizeof(unsigned long),
2826 		.mode		= 0644,
2827 		.proc_handler	= hugetlb_overcommit_handler,
2828 	},
2829 #endif
2830 	{
2831 		.procname	= "lowmem_reserve_ratio",
2832 		.data		= &sysctl_lowmem_reserve_ratio,
2833 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
2834 		.mode		= 0644,
2835 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
2836 	},
2837 	{
2838 		.procname	= "drop_caches",
2839 		.data		= &sysctl_drop_caches,
2840 		.maxlen		= sizeof(int),
2841 		.mode		= 0200,
2842 		.proc_handler	= drop_caches_sysctl_handler,
2843 		.extra1		= SYSCTL_ONE,
2844 		.extra2		= &four,
2845 	},
2846 #ifdef CONFIG_COMPACTION
2847 	{
2848 		.procname	= "compact_memory",
2849 		.data		= &sysctl_compact_memory,
2850 		.maxlen		= sizeof(int),
2851 		.mode		= 0200,
2852 		.proc_handler	= sysctl_compaction_handler,
2853 	},
2854 	{
2855 		.procname	= "compaction_proactiveness",
2856 		.data		= &sysctl_compaction_proactiveness,
2857 		.maxlen		= sizeof(sysctl_compaction_proactiveness),
2858 		.mode		= 0644,
2859 		.proc_handler	= proc_dointvec_minmax,
2860 		.extra1		= SYSCTL_ZERO,
2861 		.extra2		= &one_hundred,
2862 	},
2863 	{
2864 		.procname	= "extfrag_threshold",
2865 		.data		= &sysctl_extfrag_threshold,
2866 		.maxlen		= sizeof(int),
2867 		.mode		= 0644,
2868 		.proc_handler	= proc_dointvec_minmax,
2869 		.extra1		= &min_extfrag_threshold,
2870 		.extra2		= &max_extfrag_threshold,
2871 	},
2872 	{
2873 		.procname	= "compact_unevictable_allowed",
2874 		.data		= &sysctl_compact_unevictable_allowed,
2875 		.maxlen		= sizeof(int),
2876 		.mode		= 0644,
2877 		.proc_handler	= proc_dointvec_minmax_warn_RT_change,
2878 		.extra1		= SYSCTL_ZERO,
2879 		.extra2		= SYSCTL_ONE,
2880 	},
2881 
2882 #endif /* CONFIG_COMPACTION */
2883 	{
2884 		.procname	= "min_free_kbytes",
2885 		.data		= &min_free_kbytes,
2886 		.maxlen		= sizeof(min_free_kbytes),
2887 		.mode		= 0644,
2888 		.proc_handler	= min_free_kbytes_sysctl_handler,
2889 		.extra1		= SYSCTL_ZERO,
2890 	},
2891 	{
2892 		.procname	= "watermark_boost_factor",
2893 		.data		= &watermark_boost_factor,
2894 		.maxlen		= sizeof(watermark_boost_factor),
2895 		.mode		= 0644,
2896 		.proc_handler	= proc_dointvec_minmax,
2897 		.extra1		= SYSCTL_ZERO,
2898 	},
2899 	{
2900 		.procname	= "watermark_scale_factor",
2901 		.data		= &watermark_scale_factor,
2902 		.maxlen		= sizeof(watermark_scale_factor),
2903 		.mode		= 0644,
2904 		.proc_handler	= watermark_scale_factor_sysctl_handler,
2905 		.extra1		= SYSCTL_ONE,
2906 		.extra2		= &one_thousand,
2907 	},
2908 	{
2909 		.procname	= "percpu_pagelist_fraction",
2910 		.data		= &percpu_pagelist_fraction,
2911 		.maxlen		= sizeof(percpu_pagelist_fraction),
2912 		.mode		= 0644,
2913 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
2914 		.extra1		= SYSCTL_ZERO,
2915 	},
2916 #ifdef CONFIG_MMU
2917 	{
2918 		.procname	= "max_map_count",
2919 		.data		= &sysctl_max_map_count,
2920 		.maxlen		= sizeof(sysctl_max_map_count),
2921 		.mode		= 0644,
2922 		.proc_handler	= proc_dointvec_minmax,
2923 		.extra1		= SYSCTL_ZERO,
2924 	},
2925 #else
2926 	{
2927 		.procname	= "nr_trim_pages",
2928 		.data		= &sysctl_nr_trim_pages,
2929 		.maxlen		= sizeof(sysctl_nr_trim_pages),
2930 		.mode		= 0644,
2931 		.proc_handler	= proc_dointvec_minmax,
2932 		.extra1		= SYSCTL_ZERO,
2933 	},
2934 #endif
2935 	{
2936 		.procname	= "laptop_mode",
2937 		.data		= &laptop_mode,
2938 		.maxlen		= sizeof(laptop_mode),
2939 		.mode		= 0644,
2940 		.proc_handler	= proc_dointvec_jiffies,
2941 	},
2942 	{
2943 		.procname	= "block_dump",
2944 		.data		= &block_dump,
2945 		.maxlen		= sizeof(block_dump),
2946 		.mode		= 0644,
2947 		.proc_handler	= proc_dointvec,
2948 		.extra1		= SYSCTL_ZERO,
2949 	},
2950 	{
2951 		.procname	= "vfs_cache_pressure",
2952 		.data		= &sysctl_vfs_cache_pressure,
2953 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
2954 		.mode		= 0644,
2955 		.proc_handler	= proc_dointvec,
2956 		.extra1		= SYSCTL_ZERO,
2957 	},
2958 #if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
2959     defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
2960 	{
2961 		.procname	= "legacy_va_layout",
2962 		.data		= &sysctl_legacy_va_layout,
2963 		.maxlen		= sizeof(sysctl_legacy_va_layout),
2964 		.mode		= 0644,
2965 		.proc_handler	= proc_dointvec,
2966 		.extra1		= SYSCTL_ZERO,
2967 	},
2968 #endif
2969 #ifdef CONFIG_NUMA
2970 	{
2971 		.procname	= "zone_reclaim_mode",
2972 		.data		= &node_reclaim_mode,
2973 		.maxlen		= sizeof(node_reclaim_mode),
2974 		.mode		= 0644,
2975 		.proc_handler	= proc_dointvec,
2976 		.extra1		= SYSCTL_ZERO,
2977 	},
2978 	{
2979 		.procname	= "min_unmapped_ratio",
2980 		.data		= &sysctl_min_unmapped_ratio,
2981 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
2982 		.mode		= 0644,
2983 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
2984 		.extra1		= SYSCTL_ZERO,
2985 		.extra2		= &one_hundred,
2986 	},
2987 	{
2988 		.procname	= "min_slab_ratio",
2989 		.data		= &sysctl_min_slab_ratio,
2990 		.maxlen		= sizeof(sysctl_min_slab_ratio),
2991 		.mode		= 0644,
2992 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
2993 		.extra1		= SYSCTL_ZERO,
2994 		.extra2		= &one_hundred,
2995 	},
2996 #endif
2997 #ifdef CONFIG_SMP
2998 	{
2999 		.procname	= "stat_interval",
3000 		.data		= &sysctl_stat_interval,
3001 		.maxlen		= sizeof(sysctl_stat_interval),
3002 		.mode		= 0644,
3003 		.proc_handler	= proc_dointvec_jiffies,
3004 	},
3005 	{
3006 		.procname	= "stat_refresh",
3007 		.data		= NULL,
3008 		.maxlen		= 0,
3009 		.mode		= 0600,
3010 		.proc_handler	= vmstat_refresh,
3011 	},
3012 #endif
3013 #ifdef CONFIG_MMU
3014 	{
3015 		.procname	= "mmap_min_addr",
3016 		.data		= &dac_mmap_min_addr,
3017 		.maxlen		= sizeof(unsigned long),
3018 		.mode		= 0644,
3019 		.proc_handler	= mmap_min_addr_handler,
3020 	},
3021 #endif
3022 #ifdef CONFIG_NUMA
3023 	{
3024 		.procname	= "numa_zonelist_order",
3025 		.data		= &numa_zonelist_order,
3026 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
3027 		.mode		= 0644,
3028 		.proc_handler	= numa_zonelist_order_handler,
3029 	},
3030 #endif
3031 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
3032    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
3033 	{
3034 		.procname	= "vdso_enabled",
3035 #ifdef CONFIG_X86_32
3036 		.data		= &vdso32_enabled,
3037 		.maxlen		= sizeof(vdso32_enabled),
3038 #else
3039 		.data		= &vdso_enabled,
3040 		.maxlen		= sizeof(vdso_enabled),
3041 #endif
3042 		.mode		= 0644,
3043 		.proc_handler	= proc_dointvec,
3044 		.extra1		= SYSCTL_ZERO,
3045 	},
3046 #endif
3047 #ifdef CONFIG_HIGHMEM
3048 	{
3049 		.procname	= "highmem_is_dirtyable",
3050 		.data		= &vm_highmem_is_dirtyable,
3051 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
3052 		.mode		= 0644,
3053 		.proc_handler	= proc_dointvec_minmax,
3054 		.extra1		= SYSCTL_ZERO,
3055 		.extra2		= SYSCTL_ONE,
3056 	},
3057 #endif
3058 #ifdef CONFIG_MEMORY_FAILURE
3059 	{
3060 		.procname	= "memory_failure_early_kill",
3061 		.data		= &sysctl_memory_failure_early_kill,
3062 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
3063 		.mode		= 0644,
3064 		.proc_handler	= proc_dointvec_minmax,
3065 		.extra1		= SYSCTL_ZERO,
3066 		.extra2		= SYSCTL_ONE,
3067 	},
3068 	{
3069 		.procname	= "memory_failure_recovery",
3070 		.data		= &sysctl_memory_failure_recovery,
3071 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
3072 		.mode		= 0644,
3073 		.proc_handler	= proc_dointvec_minmax,
3074 		.extra1		= SYSCTL_ZERO,
3075 		.extra2		= SYSCTL_ONE,
3076 	},
3077 #endif
3078 	{
3079 		.procname	= "user_reserve_kbytes",
3080 		.data		= &sysctl_user_reserve_kbytes,
3081 		.maxlen		= sizeof(sysctl_user_reserve_kbytes),
3082 		.mode		= 0644,
3083 		.proc_handler	= proc_doulongvec_minmax,
3084 	},
3085 	{
3086 		.procname	= "admin_reserve_kbytes",
3087 		.data		= &sysctl_admin_reserve_kbytes,
3088 		.maxlen		= sizeof(sysctl_admin_reserve_kbytes),
3089 		.mode		= 0644,
3090 		.proc_handler	= proc_doulongvec_minmax,
3091 	},
3092 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
3093 	{
3094 		.procname	= "mmap_rnd_bits",
3095 		.data		= &mmap_rnd_bits,
3096 		.maxlen		= sizeof(mmap_rnd_bits),
3097 		.mode		= 0600,
3098 		.proc_handler	= proc_dointvec_minmax,
3099 		.extra1		= (void *)&mmap_rnd_bits_min,
3100 		.extra2		= (void *)&mmap_rnd_bits_max,
3101 	},
3102 #endif
3103 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
3104 	{
3105 		.procname	= "mmap_rnd_compat_bits",
3106 		.data		= &mmap_rnd_compat_bits,
3107 		.maxlen		= sizeof(mmap_rnd_compat_bits),
3108 		.mode		= 0600,
3109 		.proc_handler	= proc_dointvec_minmax,
3110 		.extra1		= (void *)&mmap_rnd_compat_bits_min,
3111 		.extra2		= (void *)&mmap_rnd_compat_bits_max,
3112 	},
3113 #endif
3114 #ifdef CONFIG_USERFAULTFD
3115 	{
3116 		.procname	= "unprivileged_userfaultfd",
3117 		.data		= &sysctl_unprivileged_userfaultfd,
3118 		.maxlen		= sizeof(sysctl_unprivileged_userfaultfd),
3119 		.mode		= 0644,
3120 		.proc_handler	= proc_dointvec_minmax,
3121 		.extra1		= SYSCTL_ZERO,
3122 		.extra2		= SYSCTL_ONE,
3123 	},
3124 #endif
3125 	{ }
3126 };
3127 
3128 static struct ctl_table fs_table[] = {
3129 	{
3130 		.procname	= "inode-nr",
3131 		.data		= &inodes_stat,
3132 		.maxlen		= 2*sizeof(long),
3133 		.mode		= 0444,
3134 		.proc_handler	= proc_nr_inodes,
3135 	},
3136 	{
3137 		.procname	= "inode-state",
3138 		.data		= &inodes_stat,
3139 		.maxlen		= 7*sizeof(long),
3140 		.mode		= 0444,
3141 		.proc_handler	= proc_nr_inodes,
3142 	},
3143 	{
3144 		.procname	= "file-nr",
3145 		.data		= &files_stat,
3146 		.maxlen		= sizeof(files_stat),
3147 		.mode		= 0444,
3148 		.proc_handler	= proc_nr_files,
3149 	},
3150 	{
3151 		.procname	= "file-max",
3152 		.data		= &files_stat.max_files,
3153 		.maxlen		= sizeof(files_stat.max_files),
3154 		.mode		= 0644,
3155 		.proc_handler	= proc_doulongvec_minmax,
3156 		.extra1		= &zero_ul,
3157 		.extra2		= &long_max,
3158 	},
3159 	{
3160 		.procname	= "nr_open",
3161 		.data		= &sysctl_nr_open,
3162 		.maxlen		= sizeof(unsigned int),
3163 		.mode		= 0644,
3164 		.proc_handler	= proc_dointvec_minmax,
3165 		.extra1		= &sysctl_nr_open_min,
3166 		.extra2		= &sysctl_nr_open_max,
3167 	},
3168 	{
3169 		.procname	= "dentry-state",
3170 		.data		= &dentry_stat,
3171 		.maxlen		= 6*sizeof(long),
3172 		.mode		= 0444,
3173 		.proc_handler	= proc_nr_dentry,
3174 	},
3175 	{
3176 		.procname	= "overflowuid",
3177 		.data		= &fs_overflowuid,
3178 		.maxlen		= sizeof(int),
3179 		.mode		= 0644,
3180 		.proc_handler	= proc_dointvec_minmax,
3181 		.extra1		= &minolduid,
3182 		.extra2		= &maxolduid,
3183 	},
3184 	{
3185 		.procname	= "overflowgid",
3186 		.data		= &fs_overflowgid,
3187 		.maxlen		= sizeof(int),
3188 		.mode		= 0644,
3189 		.proc_handler	= proc_dointvec_minmax,
3190 		.extra1		= &minolduid,
3191 		.extra2		= &maxolduid,
3192 	},
3193 #ifdef CONFIG_FILE_LOCKING
3194 	{
3195 		.procname	= "leases-enable",
3196 		.data		= &leases_enable,
3197 		.maxlen		= sizeof(int),
3198 		.mode		= 0644,
3199 		.proc_handler	= proc_dointvec,
3200 	},
3201 #endif
3202 #ifdef CONFIG_DNOTIFY
3203 	{
3204 		.procname	= "dir-notify-enable",
3205 		.data		= &dir_notify_enable,
3206 		.maxlen		= sizeof(int),
3207 		.mode		= 0644,
3208 		.proc_handler	= proc_dointvec,
3209 	},
3210 #endif
3211 #ifdef CONFIG_MMU
3212 #ifdef CONFIG_FILE_LOCKING
3213 	{
3214 		.procname	= "lease-break-time",
3215 		.data		= &lease_break_time,
3216 		.maxlen		= sizeof(int),
3217 		.mode		= 0644,
3218 		.proc_handler	= proc_dointvec,
3219 	},
3220 #endif
3221 #ifdef CONFIG_AIO
3222 	{
3223 		.procname	= "aio-nr",
3224 		.data		= &aio_nr,
3225 		.maxlen		= sizeof(aio_nr),
3226 		.mode		= 0444,
3227 		.proc_handler	= proc_doulongvec_minmax,
3228 	},
3229 	{
3230 		.procname	= "aio-max-nr",
3231 		.data		= &aio_max_nr,
3232 		.maxlen		= sizeof(aio_max_nr),
3233 		.mode		= 0644,
3234 		.proc_handler	= proc_doulongvec_minmax,
3235 	},
3236 #endif /* CONFIG_AIO */
3237 #ifdef CONFIG_INOTIFY_USER
3238 	{
3239 		.procname	= "inotify",
3240 		.mode		= 0555,
3241 		.child		= inotify_table,
3242 	},
3243 #endif
3244 #ifdef CONFIG_EPOLL
3245 	{
3246 		.procname	= "epoll",
3247 		.mode		= 0555,
3248 		.child		= epoll_table,
3249 	},
3250 #endif
3251 #endif
3252 	{
3253 		.procname	= "protected_symlinks",
3254 		.data		= &sysctl_protected_symlinks,
3255 		.maxlen		= sizeof(int),
3256 		.mode		= 0600,
3257 		.proc_handler	= proc_dointvec_minmax,
3258 		.extra1		= SYSCTL_ZERO,
3259 		.extra2		= SYSCTL_ONE,
3260 	},
3261 	{
3262 		.procname	= "protected_hardlinks",
3263 		.data		= &sysctl_protected_hardlinks,
3264 		.maxlen		= sizeof(int),
3265 		.mode		= 0600,
3266 		.proc_handler	= proc_dointvec_minmax,
3267 		.extra1		= SYSCTL_ZERO,
3268 		.extra2		= SYSCTL_ONE,
3269 	},
3270 	{
3271 		.procname	= "protected_fifos",
3272 		.data		= &sysctl_protected_fifos,
3273 		.maxlen		= sizeof(int),
3274 		.mode		= 0600,
3275 		.proc_handler	= proc_dointvec_minmax,
3276 		.extra1		= SYSCTL_ZERO,
3277 		.extra2		= &two,
3278 	},
3279 	{
3280 		.procname	= "protected_regular",
3281 		.data		= &sysctl_protected_regular,
3282 		.maxlen		= sizeof(int),
3283 		.mode		= 0600,
3284 		.proc_handler	= proc_dointvec_minmax,
3285 		.extra1		= SYSCTL_ZERO,
3286 		.extra2		= &two,
3287 	},
3288 	{
3289 		.procname	= "suid_dumpable",
3290 		.data		= &suid_dumpable,
3291 		.maxlen		= sizeof(int),
3292 		.mode		= 0644,
3293 		.proc_handler	= proc_dointvec_minmax_coredump,
3294 		.extra1		= SYSCTL_ZERO,
3295 		.extra2		= &two,
3296 	},
3297 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
3298 	{
3299 		.procname	= "binfmt_misc",
3300 		.mode		= 0555,
3301 		.child		= sysctl_mount_point,
3302 	},
3303 #endif
3304 	{
3305 		.procname	= "pipe-max-size",
3306 		.data		= &pipe_max_size,
3307 		.maxlen		= sizeof(pipe_max_size),
3308 		.mode		= 0644,
3309 		.proc_handler	= proc_dopipe_max_size,
3310 	},
3311 	{
3312 		.procname	= "pipe-user-pages-hard",
3313 		.data		= &pipe_user_pages_hard,
3314 		.maxlen		= sizeof(pipe_user_pages_hard),
3315 		.mode		= 0644,
3316 		.proc_handler	= proc_doulongvec_minmax,
3317 	},
3318 	{
3319 		.procname	= "pipe-user-pages-soft",
3320 		.data		= &pipe_user_pages_soft,
3321 		.maxlen		= sizeof(pipe_user_pages_soft),
3322 		.mode		= 0644,
3323 		.proc_handler	= proc_doulongvec_minmax,
3324 	},
3325 	{
3326 		.procname	= "mount-max",
3327 		.data		= &sysctl_mount_max,
3328 		.maxlen		= sizeof(unsigned int),
3329 		.mode		= 0644,
3330 		.proc_handler	= proc_dointvec_minmax,
3331 		.extra1		= SYSCTL_ONE,
3332 	},
3333 	{ }
3334 };
3335 
3336 static struct ctl_table debug_table[] = {
3337 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
3338 	{
3339 		.procname	= "exception-trace",
3340 		.data		= &show_unhandled_signals,
3341 		.maxlen		= sizeof(int),
3342 		.mode		= 0644,
3343 		.proc_handler	= proc_dointvec
3344 	},
3345 #endif
3346 #if defined(CONFIG_OPTPROBES)
3347 	{
3348 		.procname	= "kprobes-optimization",
3349 		.data		= &sysctl_kprobes_optimization,
3350 		.maxlen		= sizeof(int),
3351 		.mode		= 0644,
3352 		.proc_handler	= proc_kprobes_optimization_handler,
3353 		.extra1		= SYSCTL_ZERO,
3354 		.extra2		= SYSCTL_ONE,
3355 	},
3356 #endif
3357 	{ }
3358 };
3359 
3360 static struct ctl_table dev_table[] = {
3361 	{ }
3362 };
3363 
3364 static struct ctl_table sysctl_base_table[] = {
3365 	{
3366 		.procname	= "kernel",
3367 		.mode		= 0555,
3368 		.child		= kern_table,
3369 	},
3370 	{
3371 		.procname	= "vm",
3372 		.mode		= 0555,
3373 		.child		= vm_table,
3374 	},
3375 	{
3376 		.procname	= "fs",
3377 		.mode		= 0555,
3378 		.child		= fs_table,
3379 	},
3380 	{
3381 		.procname	= "debug",
3382 		.mode		= 0555,
3383 		.child		= debug_table,
3384 	},
3385 	{
3386 		.procname	= "dev",
3387 		.mode		= 0555,
3388 		.child		= dev_table,
3389 	},
3390 	{ }
3391 };
3392 
3393 int __init sysctl_init(void)
3394 {
3395 	struct ctl_table_header *hdr;
3396 
3397 	hdr = register_sysctl_table(sysctl_base_table);
3398 	kmemleak_not_leak(hdr);
3399 	return 0;
3400 }
3401 #endif /* CONFIG_SYSCTL */
3402 /*
3403  * No sense putting this after each symbol definition, twice,
3404  * exception granted :-)
3405  */
3406 EXPORT_SYMBOL(proc_dointvec);
3407 EXPORT_SYMBOL(proc_douintvec);
3408 EXPORT_SYMBOL(proc_dointvec_jiffies);
3409 EXPORT_SYMBOL(proc_dointvec_minmax);
3410 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3411 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3412 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3413 EXPORT_SYMBOL(proc_dostring);
3414 EXPORT_SYMBOL(proc_doulongvec_minmax);
3415 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3416 EXPORT_SYMBOL(proc_do_large_bitmap);
3417