Loading...
1// SPDX-License-Identifier: GPL-2.0-only
2/*
3 * sysctl.c: General linux system control interface
4 *
5 * Begun 24 March 1995, Stephen Tweedie
6 * Added /proc support, Dec 1995
7 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
8 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
9 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
10 * Dynamic registration fixes, Stephen Tweedie.
11 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
12 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
13 * Horn.
14 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
15 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
16 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
17 * Wendling.
18 * The list_for_each() macro wasn't appropriate for the sysctl loop.
19 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
20 */
21
22#include <linux/module.h>
23#include <linux/mm.h>
24#include <linux/swap.h>
25#include <linux/slab.h>
26#include <linux/sysctl.h>
27#include <linux/bitmap.h>
28#include <linux/signal.h>
29#include <linux/panic.h>
30#include <linux/printk.h>
31#include <linux/proc_fs.h>
32#include <linux/security.h>
33#include <linux/ctype.h>
34#include <linux/kmemleak.h>
35#include <linux/filter.h>
36#include <linux/fs.h>
37#include <linux/init.h>
38#include <linux/kernel.h>
39#include <linux/kobject.h>
40#include <linux/net.h>
41#include <linux/sysrq.h>
42#include <linux/highuid.h>
43#include <linux/writeback.h>
44#include <linux/ratelimit.h>
45#include <linux/compaction.h>
46#include <linux/hugetlb.h>
47#include <linux/initrd.h>
48#include <linux/key.h>
49#include <linux/times.h>
50#include <linux/limits.h>
51#include <linux/dcache.h>
52#include <linux/syscalls.h>
53#include <linux/vmstat.h>
54#include <linux/nfs_fs.h>
55#include <linux/acpi.h>
56#include <linux/reboot.h>
57#include <linux/ftrace.h>
58#include <linux/perf_event.h>
59#include <linux/oom.h>
60#include <linux/kmod.h>
61#include <linux/capability.h>
62#include <linux/binfmts.h>
63#include <linux/sched/sysctl.h>
64#include <linux/mount.h>
65#include <linux/userfaultfd_k.h>
66#include <linux/pid.h>
67
68#include "../lib/kstrtox.h"
69
70#include <linux/uaccess.h>
71#include <asm/processor.h>
72
73#ifdef CONFIG_X86
74#include <asm/nmi.h>
75#include <asm/stacktrace.h>
76#include <asm/io.h>
77#endif
78#ifdef CONFIG_SPARC
79#include <asm/setup.h>
80#endif
81#ifdef CONFIG_RT_MUTEXES
82#include <linux/rtmutex.h>
83#endif
84
85/* shared constants to be used in various sysctls */
86const int sysctl_vals[] = { 0, 1, 2, 3, 4, 100, 200, 1000, 3000, INT_MAX, 65535, -1 };
87EXPORT_SYMBOL(sysctl_vals);
88
89const unsigned long sysctl_long_vals[] = { 0, 1, LONG_MAX };
90EXPORT_SYMBOL_GPL(sysctl_long_vals);
91
92#if defined(CONFIG_SYSCTL)
93
94/* Constants used for minimum and maximum */
95
96#ifdef CONFIG_PERF_EVENTS
97static const int six_hundred_forty_kb = 640 * 1024;
98#endif
99
100
101static const int ngroups_max = NGROUPS_MAX;
102static const int cap_last_cap = CAP_LAST_CAP;
103
104#ifdef CONFIG_PROC_SYSCTL
105
106/**
107 * enum sysctl_writes_mode - supported sysctl write modes
108 *
109 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
110 * to be written, and multiple writes on the same sysctl file descriptor
111 * will rewrite the sysctl value, regardless of file position. No warning
112 * is issued when the initial position is not 0.
113 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
114 * not 0.
115 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
116 * file position 0 and the value must be fully contained in the buffer
117 * sent to the write syscall. If dealing with strings respect the file
118 * position, but restrict this to the max length of the buffer, anything
119 * passed the max length will be ignored. Multiple writes will append
120 * to the buffer.
121 *
122 * These write modes control how current file position affects the behavior of
123 * updating sysctl values through the proc interface on each write.
124 */
125enum sysctl_writes_mode {
126 SYSCTL_WRITES_LEGACY = -1,
127 SYSCTL_WRITES_WARN = 0,
128 SYSCTL_WRITES_STRICT = 1,
129};
130
131static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
132#endif /* CONFIG_PROC_SYSCTL */
133
134#if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
135 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
136int sysctl_legacy_va_layout;
137#endif
138
139#endif /* CONFIG_SYSCTL */
140
141/*
142 * /proc/sys support
143 */
144
145#ifdef CONFIG_PROC_SYSCTL
146
147static int _proc_do_string(char *data, int maxlen, int write,
148 char *buffer, size_t *lenp, loff_t *ppos)
149{
150 size_t len;
151 char c, *p;
152
153 if (!data || !maxlen || !*lenp) {
154 *lenp = 0;
155 return 0;
156 }
157
158 if (write) {
159 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
160 /* Only continue writes not past the end of buffer. */
161 len = strlen(data);
162 if (len > maxlen - 1)
163 len = maxlen - 1;
164
165 if (*ppos > len)
166 return 0;
167 len = *ppos;
168 } else {
169 /* Start writing from beginning of buffer. */
170 len = 0;
171 }
172
173 *ppos += *lenp;
174 p = buffer;
175 while ((p - buffer) < *lenp && len < maxlen - 1) {
176 c = *(p++);
177 if (c == 0 || c == '\n')
178 break;
179 data[len++] = c;
180 }
181 data[len] = 0;
182 } else {
183 len = strlen(data);
184 if (len > maxlen)
185 len = maxlen;
186
187 if (*ppos > len) {
188 *lenp = 0;
189 return 0;
190 }
191
192 data += *ppos;
193 len -= *ppos;
194
195 if (len > *lenp)
196 len = *lenp;
197 if (len)
198 memcpy(buffer, data, len);
199 if (len < *lenp) {
200 buffer[len] = '\n';
201 len++;
202 }
203 *lenp = len;
204 *ppos += len;
205 }
206 return 0;
207}
208
209static void warn_sysctl_write(struct ctl_table *table)
210{
211 pr_warn_once("%s wrote to %s when file position was not 0!\n"
212 "This will not be supported in the future. To silence this\n"
213 "warning, set kernel.sysctl_writes_strict = -1\n",
214 current->comm, table->procname);
215}
216
217/**
218 * proc_first_pos_non_zero_ignore - check if first position is allowed
219 * @ppos: file position
220 * @table: the sysctl table
221 *
222 * Returns true if the first position is non-zero and the sysctl_writes_strict
223 * mode indicates this is not allowed for numeric input types. String proc
224 * handlers can ignore the return value.
225 */
226static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
227 struct ctl_table *table)
228{
229 if (!*ppos)
230 return false;
231
232 switch (sysctl_writes_strict) {
233 case SYSCTL_WRITES_STRICT:
234 return true;
235 case SYSCTL_WRITES_WARN:
236 warn_sysctl_write(table);
237 return false;
238 default:
239 return false;
240 }
241}
242
243/**
244 * proc_dostring - read a string sysctl
245 * @table: the sysctl table
246 * @write: %TRUE if this is a write to the sysctl file
247 * @buffer: the user buffer
248 * @lenp: the size of the user buffer
249 * @ppos: file position
250 *
251 * Reads/writes a string from/to the user buffer. If the kernel
252 * buffer provided is not large enough to hold the string, the
253 * string is truncated. The copied string is %NULL-terminated.
254 * If the string is being read by the user process, it is copied
255 * and a newline '\n' is added. It is truncated if the buffer is
256 * not large enough.
257 *
258 * Returns 0 on success.
259 */
260int proc_dostring(struct ctl_table *table, int write,
261 void *buffer, size_t *lenp, loff_t *ppos)
262{
263 if (write)
264 proc_first_pos_non_zero_ignore(ppos, table);
265
266 return _proc_do_string(table->data, table->maxlen, write, buffer, lenp,
267 ppos);
268}
269
270static void proc_skip_spaces(char **buf, size_t *size)
271{
272 while (*size) {
273 if (!isspace(**buf))
274 break;
275 (*size)--;
276 (*buf)++;
277 }
278}
279
280static void proc_skip_char(char **buf, size_t *size, const char v)
281{
282 while (*size) {
283 if (**buf != v)
284 break;
285 (*size)--;
286 (*buf)++;
287 }
288}
289
290/**
291 * strtoul_lenient - parse an ASCII formatted integer from a buffer and only
292 * fail on overflow
293 *
294 * @cp: kernel buffer containing the string to parse
295 * @endp: pointer to store the trailing characters
296 * @base: the base to use
297 * @res: where the parsed integer will be stored
298 *
299 * In case of success 0 is returned and @res will contain the parsed integer,
300 * @endp will hold any trailing characters.
301 * This function will fail the parse on overflow. If there wasn't an overflow
302 * the function will defer the decision what characters count as invalid to the
303 * caller.
304 */
305static int strtoul_lenient(const char *cp, char **endp, unsigned int base,
306 unsigned long *res)
307{
308 unsigned long long result;
309 unsigned int rv;
310
311 cp = _parse_integer_fixup_radix(cp, &base);
312 rv = _parse_integer(cp, base, &result);
313 if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result))
314 return -ERANGE;
315
316 cp += rv;
317
318 if (endp)
319 *endp = (char *)cp;
320
321 *res = (unsigned long)result;
322 return 0;
323}
324
325#define TMPBUFLEN 22
326/**
327 * proc_get_long - reads an ASCII formatted integer from a user buffer
328 *
329 * @buf: a kernel buffer
330 * @size: size of the kernel buffer
331 * @val: this is where the number will be stored
332 * @neg: set to %TRUE if number is negative
333 * @perm_tr: a vector which contains the allowed trailers
334 * @perm_tr_len: size of the perm_tr vector
335 * @tr: pointer to store the trailer character
336 *
337 * In case of success %0 is returned and @buf and @size are updated with
338 * the amount of bytes read. If @tr is non-NULL and a trailing
339 * character exists (size is non-zero after returning from this
340 * function), @tr is updated with the trailing character.
341 */
342static int proc_get_long(char **buf, size_t *size,
343 unsigned long *val, bool *neg,
344 const char *perm_tr, unsigned perm_tr_len, char *tr)
345{
346 char *p, tmp[TMPBUFLEN];
347 ssize_t len = *size;
348
349 if (len <= 0)
350 return -EINVAL;
351
352 if (len > TMPBUFLEN - 1)
353 len = TMPBUFLEN - 1;
354
355 memcpy(tmp, *buf, len);
356
357 tmp[len] = 0;
358 p = tmp;
359 if (*p == '-' && *size > 1) {
360 *neg = true;
361 p++;
362 } else
363 *neg = false;
364 if (!isdigit(*p))
365 return -EINVAL;
366
367 if (strtoul_lenient(p, &p, 0, val))
368 return -EINVAL;
369
370 len = p - tmp;
371
372 /* We don't know if the next char is whitespace thus we may accept
373 * invalid integers (e.g. 1234...a) or two integers instead of one
374 * (e.g. 123...1). So lets not allow such large numbers. */
375 if (len == TMPBUFLEN - 1)
376 return -EINVAL;
377
378 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
379 return -EINVAL;
380
381 if (tr && (len < *size))
382 *tr = *p;
383
384 *buf += len;
385 *size -= len;
386
387 return 0;
388}
389
390/**
391 * proc_put_long - converts an integer to a decimal ASCII formatted string
392 *
393 * @buf: the user buffer
394 * @size: the size of the user buffer
395 * @val: the integer to be converted
396 * @neg: sign of the number, %TRUE for negative
397 *
398 * In case of success @buf and @size are updated with the amount of bytes
399 * written.
400 */
401static void proc_put_long(void **buf, size_t *size, unsigned long val, bool neg)
402{
403 int len;
404 char tmp[TMPBUFLEN], *p = tmp;
405
406 sprintf(p, "%s%lu", neg ? "-" : "", val);
407 len = strlen(tmp);
408 if (len > *size)
409 len = *size;
410 memcpy(*buf, tmp, len);
411 *size -= len;
412 *buf += len;
413}
414#undef TMPBUFLEN
415
416static void proc_put_char(void **buf, size_t *size, char c)
417{
418 if (*size) {
419 char **buffer = (char **)buf;
420 **buffer = c;
421
422 (*size)--;
423 (*buffer)++;
424 *buf = *buffer;
425 }
426}
427
428static int do_proc_dobool_conv(bool *negp, unsigned long *lvalp,
429 int *valp,
430 int write, void *data)
431{
432 if (write) {
433 *(bool *)valp = *lvalp;
434 } else {
435 int val = *(bool *)valp;
436
437 *lvalp = (unsigned long)val;
438 *negp = false;
439 }
440 return 0;
441}
442
443static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
444 int *valp,
445 int write, void *data)
446{
447 if (write) {
448 if (*negp) {
449 if (*lvalp > (unsigned long) INT_MAX + 1)
450 return -EINVAL;
451 WRITE_ONCE(*valp, -*lvalp);
452 } else {
453 if (*lvalp > (unsigned long) INT_MAX)
454 return -EINVAL;
455 WRITE_ONCE(*valp, *lvalp);
456 }
457 } else {
458 int val = READ_ONCE(*valp);
459 if (val < 0) {
460 *negp = true;
461 *lvalp = -(unsigned long)val;
462 } else {
463 *negp = false;
464 *lvalp = (unsigned long)val;
465 }
466 }
467 return 0;
468}
469
470static int do_proc_douintvec_conv(unsigned long *lvalp,
471 unsigned int *valp,
472 int write, void *data)
473{
474 if (write) {
475 if (*lvalp > UINT_MAX)
476 return -EINVAL;
477 WRITE_ONCE(*valp, *lvalp);
478 } else {
479 unsigned int val = READ_ONCE(*valp);
480 *lvalp = (unsigned long)val;
481 }
482 return 0;
483}
484
485static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
486
487static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
488 int write, void *buffer,
489 size_t *lenp, loff_t *ppos,
490 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
491 int write, void *data),
492 void *data)
493{
494 int *i, vleft, first = 1, err = 0;
495 size_t left;
496 char *p;
497
498 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
499 *lenp = 0;
500 return 0;
501 }
502
503 i = (int *) tbl_data;
504 vleft = table->maxlen / sizeof(*i);
505 left = *lenp;
506
507 if (!conv)
508 conv = do_proc_dointvec_conv;
509
510 if (write) {
511 if (proc_first_pos_non_zero_ignore(ppos, table))
512 goto out;
513
514 if (left > PAGE_SIZE - 1)
515 left = PAGE_SIZE - 1;
516 p = buffer;
517 }
518
519 for (; left && vleft--; i++, first=0) {
520 unsigned long lval;
521 bool neg;
522
523 if (write) {
524 proc_skip_spaces(&p, &left);
525
526 if (!left)
527 break;
528 err = proc_get_long(&p, &left, &lval, &neg,
529 proc_wspace_sep,
530 sizeof(proc_wspace_sep), NULL);
531 if (err)
532 break;
533 if (conv(&neg, &lval, i, 1, data)) {
534 err = -EINVAL;
535 break;
536 }
537 } else {
538 if (conv(&neg, &lval, i, 0, data)) {
539 err = -EINVAL;
540 break;
541 }
542 if (!first)
543 proc_put_char(&buffer, &left, '\t');
544 proc_put_long(&buffer, &left, lval, neg);
545 }
546 }
547
548 if (!write && !first && left && !err)
549 proc_put_char(&buffer, &left, '\n');
550 if (write && !err && left)
551 proc_skip_spaces(&p, &left);
552 if (write && first)
553 return err ? : -EINVAL;
554 *lenp -= left;
555out:
556 *ppos += *lenp;
557 return err;
558}
559
560static int do_proc_dointvec(struct ctl_table *table, int write,
561 void *buffer, size_t *lenp, loff_t *ppos,
562 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
563 int write, void *data),
564 void *data)
565{
566 return __do_proc_dointvec(table->data, table, write,
567 buffer, lenp, ppos, conv, data);
568}
569
570static int do_proc_douintvec_w(unsigned int *tbl_data,
571 struct ctl_table *table,
572 void *buffer,
573 size_t *lenp, loff_t *ppos,
574 int (*conv)(unsigned long *lvalp,
575 unsigned int *valp,
576 int write, void *data),
577 void *data)
578{
579 unsigned long lval;
580 int err = 0;
581 size_t left;
582 bool neg;
583 char *p = buffer;
584
585 left = *lenp;
586
587 if (proc_first_pos_non_zero_ignore(ppos, table))
588 goto bail_early;
589
590 if (left > PAGE_SIZE - 1)
591 left = PAGE_SIZE - 1;
592
593 proc_skip_spaces(&p, &left);
594 if (!left) {
595 err = -EINVAL;
596 goto out_free;
597 }
598
599 err = proc_get_long(&p, &left, &lval, &neg,
600 proc_wspace_sep,
601 sizeof(proc_wspace_sep), NULL);
602 if (err || neg) {
603 err = -EINVAL;
604 goto out_free;
605 }
606
607 if (conv(&lval, tbl_data, 1, data)) {
608 err = -EINVAL;
609 goto out_free;
610 }
611
612 if (!err && left)
613 proc_skip_spaces(&p, &left);
614
615out_free:
616 if (err)
617 return -EINVAL;
618
619 return 0;
620
621 /* This is in keeping with old __do_proc_dointvec() */
622bail_early:
623 *ppos += *lenp;
624 return err;
625}
626
627static int do_proc_douintvec_r(unsigned int *tbl_data, void *buffer,
628 size_t *lenp, loff_t *ppos,
629 int (*conv)(unsigned long *lvalp,
630 unsigned int *valp,
631 int write, void *data),
632 void *data)
633{
634 unsigned long lval;
635 int err = 0;
636 size_t left;
637
638 left = *lenp;
639
640 if (conv(&lval, tbl_data, 0, data)) {
641 err = -EINVAL;
642 goto out;
643 }
644
645 proc_put_long(&buffer, &left, lval, false);
646 if (!left)
647 goto out;
648
649 proc_put_char(&buffer, &left, '\n');
650
651out:
652 *lenp -= left;
653 *ppos += *lenp;
654
655 return err;
656}
657
658static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
659 int write, void *buffer,
660 size_t *lenp, loff_t *ppos,
661 int (*conv)(unsigned long *lvalp,
662 unsigned int *valp,
663 int write, void *data),
664 void *data)
665{
666 unsigned int *i, vleft;
667
668 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
669 *lenp = 0;
670 return 0;
671 }
672
673 i = (unsigned int *) tbl_data;
674 vleft = table->maxlen / sizeof(*i);
675
676 /*
677 * Arrays are not supported, keep this simple. *Do not* add
678 * support for them.
679 */
680 if (vleft != 1) {
681 *lenp = 0;
682 return -EINVAL;
683 }
684
685 if (!conv)
686 conv = do_proc_douintvec_conv;
687
688 if (write)
689 return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
690 conv, data);
691 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
692}
693
694int do_proc_douintvec(struct ctl_table *table, int write,
695 void *buffer, size_t *lenp, loff_t *ppos,
696 int (*conv)(unsigned long *lvalp,
697 unsigned int *valp,
698 int write, void *data),
699 void *data)
700{
701 return __do_proc_douintvec(table->data, table, write,
702 buffer, lenp, ppos, conv, data);
703}
704
705/**
706 * proc_dobool - read/write a bool
707 * @table: the sysctl table
708 * @write: %TRUE if this is a write to the sysctl file
709 * @buffer: the user buffer
710 * @lenp: the size of the user buffer
711 * @ppos: file position
712 *
713 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
714 * values from/to the user buffer, treated as an ASCII string.
715 *
716 * Returns 0 on success.
717 */
718int proc_dobool(struct ctl_table *table, int write, void *buffer,
719 size_t *lenp, loff_t *ppos)
720{
721 return do_proc_dointvec(table, write, buffer, lenp, ppos,
722 do_proc_dobool_conv, NULL);
723}
724
725/**
726 * proc_dointvec - read a vector of integers
727 * @table: the sysctl table
728 * @write: %TRUE if this is a write to the sysctl file
729 * @buffer: the user buffer
730 * @lenp: the size of the user buffer
731 * @ppos: file position
732 *
733 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
734 * values from/to the user buffer, treated as an ASCII string.
735 *
736 * Returns 0 on success.
737 */
738int proc_dointvec(struct ctl_table *table, int write, void *buffer,
739 size_t *lenp, loff_t *ppos)
740{
741 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
742}
743
744#ifdef CONFIG_COMPACTION
745static int proc_dointvec_minmax_warn_RT_change(struct ctl_table *table,
746 int write, void *buffer, size_t *lenp, loff_t *ppos)
747{
748 int ret, old;
749
750 if (!IS_ENABLED(CONFIG_PREEMPT_RT) || !write)
751 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
752
753 old = *(int *)table->data;
754 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
755 if (ret)
756 return ret;
757 if (old != *(int *)table->data)
758 pr_warn_once("sysctl attribute %s changed by %s[%d]\n",
759 table->procname, current->comm,
760 task_pid_nr(current));
761 return ret;
762}
763#endif
764
765/**
766 * proc_douintvec - read a vector of unsigned integers
767 * @table: the sysctl table
768 * @write: %TRUE if this is a write to the sysctl file
769 * @buffer: the user buffer
770 * @lenp: the size of the user buffer
771 * @ppos: file position
772 *
773 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
774 * values from/to the user buffer, treated as an ASCII string.
775 *
776 * Returns 0 on success.
777 */
778int proc_douintvec(struct ctl_table *table, int write, void *buffer,
779 size_t *lenp, loff_t *ppos)
780{
781 return do_proc_douintvec(table, write, buffer, lenp, ppos,
782 do_proc_douintvec_conv, NULL);
783}
784
785/*
786 * Taint values can only be increased
787 * This means we can safely use a temporary.
788 */
789static int proc_taint(struct ctl_table *table, int write,
790 void *buffer, size_t *lenp, loff_t *ppos)
791{
792 struct ctl_table t;
793 unsigned long tmptaint = get_taint();
794 int err;
795
796 if (write && !capable(CAP_SYS_ADMIN))
797 return -EPERM;
798
799 t = *table;
800 t.data = &tmptaint;
801 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
802 if (err < 0)
803 return err;
804
805 if (write) {
806 int i;
807
808 /*
809 * If we are relying on panic_on_taint not producing
810 * false positives due to userspace input, bail out
811 * before setting the requested taint flags.
812 */
813 if (panic_on_taint_nousertaint && (tmptaint & panic_on_taint))
814 return -EINVAL;
815
816 /*
817 * Poor man's atomic or. Not worth adding a primitive
818 * to everyone's atomic.h for this
819 */
820 for (i = 0; i < TAINT_FLAGS_COUNT; i++)
821 if ((1UL << i) & tmptaint)
822 add_taint(i, LOCKDEP_STILL_OK);
823 }
824
825 return err;
826}
827
828/**
829 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
830 * @min: pointer to minimum allowable value
831 * @max: pointer to maximum allowable value
832 *
833 * The do_proc_dointvec_minmax_conv_param structure provides the
834 * minimum and maximum values for doing range checking for those sysctl
835 * parameters that use the proc_dointvec_minmax() handler.
836 */
837struct do_proc_dointvec_minmax_conv_param {
838 int *min;
839 int *max;
840};
841
842static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
843 int *valp,
844 int write, void *data)
845{
846 int tmp, ret;
847 struct do_proc_dointvec_minmax_conv_param *param = data;
848 /*
849 * If writing, first do so via a temporary local int so we can
850 * bounds-check it before touching *valp.
851 */
852 int *ip = write ? &tmp : valp;
853
854 ret = do_proc_dointvec_conv(negp, lvalp, ip, write, data);
855 if (ret)
856 return ret;
857
858 if (write) {
859 if ((param->min && *param->min > tmp) ||
860 (param->max && *param->max < tmp))
861 return -EINVAL;
862 WRITE_ONCE(*valp, tmp);
863 }
864
865 return 0;
866}
867
868/**
869 * proc_dointvec_minmax - read a vector of integers with min/max values
870 * @table: the sysctl table
871 * @write: %TRUE if this is a write to the sysctl file
872 * @buffer: the user buffer
873 * @lenp: the size of the user buffer
874 * @ppos: file position
875 *
876 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
877 * values from/to the user buffer, treated as an ASCII string.
878 *
879 * This routine will ensure the values are within the range specified by
880 * table->extra1 (min) and table->extra2 (max).
881 *
882 * Returns 0 on success or -EINVAL on write when the range check fails.
883 */
884int proc_dointvec_minmax(struct ctl_table *table, int write,
885 void *buffer, size_t *lenp, loff_t *ppos)
886{
887 struct do_proc_dointvec_minmax_conv_param param = {
888 .min = (int *) table->extra1,
889 .max = (int *) table->extra2,
890 };
891 return do_proc_dointvec(table, write, buffer, lenp, ppos,
892 do_proc_dointvec_minmax_conv, ¶m);
893}
894
895/**
896 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
897 * @min: pointer to minimum allowable value
898 * @max: pointer to maximum allowable value
899 *
900 * The do_proc_douintvec_minmax_conv_param structure provides the
901 * minimum and maximum values for doing range checking for those sysctl
902 * parameters that use the proc_douintvec_minmax() handler.
903 */
904struct do_proc_douintvec_minmax_conv_param {
905 unsigned int *min;
906 unsigned int *max;
907};
908
909static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
910 unsigned int *valp,
911 int write, void *data)
912{
913 int ret;
914 unsigned int tmp;
915 struct do_proc_douintvec_minmax_conv_param *param = data;
916 /* write via temporary local uint for bounds-checking */
917 unsigned int *up = write ? &tmp : valp;
918
919 ret = do_proc_douintvec_conv(lvalp, up, write, data);
920 if (ret)
921 return ret;
922
923 if (write) {
924 if ((param->min && *param->min > tmp) ||
925 (param->max && *param->max < tmp))
926 return -ERANGE;
927
928 WRITE_ONCE(*valp, tmp);
929 }
930
931 return 0;
932}
933
934/**
935 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
936 * @table: the sysctl table
937 * @write: %TRUE if this is a write to the sysctl file
938 * @buffer: the user buffer
939 * @lenp: the size of the user buffer
940 * @ppos: file position
941 *
942 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
943 * values from/to the user buffer, treated as an ASCII string. Negative
944 * strings are not allowed.
945 *
946 * This routine will ensure the values are within the range specified by
947 * table->extra1 (min) and table->extra2 (max). There is a final sanity
948 * check for UINT_MAX to avoid having to support wrap around uses from
949 * userspace.
950 *
951 * Returns 0 on success or -ERANGE on write when the range check fails.
952 */
953int proc_douintvec_minmax(struct ctl_table *table, int write,
954 void *buffer, size_t *lenp, loff_t *ppos)
955{
956 struct do_proc_douintvec_minmax_conv_param param = {
957 .min = (unsigned int *) table->extra1,
958 .max = (unsigned int *) table->extra2,
959 };
960 return do_proc_douintvec(table, write, buffer, lenp, ppos,
961 do_proc_douintvec_minmax_conv, ¶m);
962}
963
964/**
965 * proc_dou8vec_minmax - read a vector of unsigned chars with min/max values
966 * @table: the sysctl table
967 * @write: %TRUE if this is a write to the sysctl file
968 * @buffer: the user buffer
969 * @lenp: the size of the user buffer
970 * @ppos: file position
971 *
972 * Reads/writes up to table->maxlen/sizeof(u8) unsigned chars
973 * values from/to the user buffer, treated as an ASCII string. Negative
974 * strings are not allowed.
975 *
976 * This routine will ensure the values are within the range specified by
977 * table->extra1 (min) and table->extra2 (max).
978 *
979 * Returns 0 on success or an error on write when the range check fails.
980 */
981int proc_dou8vec_minmax(struct ctl_table *table, int write,
982 void *buffer, size_t *lenp, loff_t *ppos)
983{
984 struct ctl_table tmp;
985 unsigned int min = 0, max = 255U, val;
986 u8 *data = table->data;
987 struct do_proc_douintvec_minmax_conv_param param = {
988 .min = &min,
989 .max = &max,
990 };
991 int res;
992
993 /* Do not support arrays yet. */
994 if (table->maxlen != sizeof(u8))
995 return -EINVAL;
996
997 if (table->extra1) {
998 min = *(unsigned int *) table->extra1;
999 if (min > 255U)
1000 return -EINVAL;
1001 }
1002 if (table->extra2) {
1003 max = *(unsigned int *) table->extra2;
1004 if (max > 255U)
1005 return -EINVAL;
1006 }
1007
1008 tmp = *table;
1009
1010 tmp.maxlen = sizeof(val);
1011 tmp.data = &val;
1012 val = READ_ONCE(*data);
1013 res = do_proc_douintvec(&tmp, write, buffer, lenp, ppos,
1014 do_proc_douintvec_minmax_conv, ¶m);
1015 if (res)
1016 return res;
1017 if (write)
1018 WRITE_ONCE(*data, val);
1019 return 0;
1020}
1021EXPORT_SYMBOL_GPL(proc_dou8vec_minmax);
1022
1023#ifdef CONFIG_MAGIC_SYSRQ
1024static int sysrq_sysctl_handler(struct ctl_table *table, int write,
1025 void *buffer, size_t *lenp, loff_t *ppos)
1026{
1027 int tmp, ret;
1028
1029 tmp = sysrq_mask();
1030
1031 ret = __do_proc_dointvec(&tmp, table, write, buffer,
1032 lenp, ppos, NULL, NULL);
1033 if (ret || !write)
1034 return ret;
1035
1036 if (write)
1037 sysrq_toggle_support(tmp);
1038
1039 return 0;
1040}
1041#endif
1042
1043static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table,
1044 int write, void *buffer, size_t *lenp, loff_t *ppos,
1045 unsigned long convmul, unsigned long convdiv)
1046{
1047 unsigned long *i, *min, *max;
1048 int vleft, first = 1, err = 0;
1049 size_t left;
1050 char *p;
1051
1052 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
1053 *lenp = 0;
1054 return 0;
1055 }
1056
1057 i = data;
1058 min = table->extra1;
1059 max = table->extra2;
1060 vleft = table->maxlen / sizeof(unsigned long);
1061 left = *lenp;
1062
1063 if (write) {
1064 if (proc_first_pos_non_zero_ignore(ppos, table))
1065 goto out;
1066
1067 if (left > PAGE_SIZE - 1)
1068 left = PAGE_SIZE - 1;
1069 p = buffer;
1070 }
1071
1072 for (; left && vleft--; i++, first = 0) {
1073 unsigned long val;
1074
1075 if (write) {
1076 bool neg;
1077
1078 proc_skip_spaces(&p, &left);
1079 if (!left)
1080 break;
1081
1082 err = proc_get_long(&p, &left, &val, &neg,
1083 proc_wspace_sep,
1084 sizeof(proc_wspace_sep), NULL);
1085 if (err || neg) {
1086 err = -EINVAL;
1087 break;
1088 }
1089
1090 val = convmul * val / convdiv;
1091 if ((min && val < *min) || (max && val > *max)) {
1092 err = -EINVAL;
1093 break;
1094 }
1095 WRITE_ONCE(*i, val);
1096 } else {
1097 val = convdiv * READ_ONCE(*i) / convmul;
1098 if (!first)
1099 proc_put_char(&buffer, &left, '\t');
1100 proc_put_long(&buffer, &left, val, false);
1101 }
1102 }
1103
1104 if (!write && !first && left && !err)
1105 proc_put_char(&buffer, &left, '\n');
1106 if (write && !err)
1107 proc_skip_spaces(&p, &left);
1108 if (write && first)
1109 return err ? : -EINVAL;
1110 *lenp -= left;
1111out:
1112 *ppos += *lenp;
1113 return err;
1114}
1115
1116static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
1117 void *buffer, size_t *lenp, loff_t *ppos, unsigned long convmul,
1118 unsigned long convdiv)
1119{
1120 return __do_proc_doulongvec_minmax(table->data, table, write,
1121 buffer, lenp, ppos, convmul, convdiv);
1122}
1123
1124/**
1125 * proc_doulongvec_minmax - read a vector of long integers with min/max values
1126 * @table: the sysctl table
1127 * @write: %TRUE if this is a write to the sysctl file
1128 * @buffer: the user buffer
1129 * @lenp: the size of the user buffer
1130 * @ppos: file position
1131 *
1132 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1133 * values from/to the user buffer, treated as an ASCII string.
1134 *
1135 * This routine will ensure the values are within the range specified by
1136 * table->extra1 (min) and table->extra2 (max).
1137 *
1138 * Returns 0 on success.
1139 */
1140int proc_doulongvec_minmax(struct ctl_table *table, int write,
1141 void *buffer, size_t *lenp, loff_t *ppos)
1142{
1143 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
1144}
1145
1146/**
1147 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
1148 * @table: the sysctl table
1149 * @write: %TRUE if this is a write to the sysctl file
1150 * @buffer: the user buffer
1151 * @lenp: the size of the user buffer
1152 * @ppos: file position
1153 *
1154 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1155 * values from/to the user buffer, treated as an ASCII string. The values
1156 * are treated as milliseconds, and converted to jiffies when they are stored.
1157 *
1158 * This routine will ensure the values are within the range specified by
1159 * table->extra1 (min) and table->extra2 (max).
1160 *
1161 * Returns 0 on success.
1162 */
1163int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1164 void *buffer, size_t *lenp, loff_t *ppos)
1165{
1166 return do_proc_doulongvec_minmax(table, write, buffer,
1167 lenp, ppos, HZ, 1000l);
1168}
1169
1170
1171static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
1172 int *valp,
1173 int write, void *data)
1174{
1175 if (write) {
1176 if (*lvalp > INT_MAX / HZ)
1177 return 1;
1178 if (*negp)
1179 WRITE_ONCE(*valp, -*lvalp * HZ);
1180 else
1181 WRITE_ONCE(*valp, *lvalp * HZ);
1182 } else {
1183 int val = READ_ONCE(*valp);
1184 unsigned long lval;
1185 if (val < 0) {
1186 *negp = true;
1187 lval = -(unsigned long)val;
1188 } else {
1189 *negp = false;
1190 lval = (unsigned long)val;
1191 }
1192 *lvalp = lval / HZ;
1193 }
1194 return 0;
1195}
1196
1197static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
1198 int *valp,
1199 int write, void *data)
1200{
1201 if (write) {
1202 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
1203 return 1;
1204 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
1205 } else {
1206 int val = *valp;
1207 unsigned long lval;
1208 if (val < 0) {
1209 *negp = true;
1210 lval = -(unsigned long)val;
1211 } else {
1212 *negp = false;
1213 lval = (unsigned long)val;
1214 }
1215 *lvalp = jiffies_to_clock_t(lval);
1216 }
1217 return 0;
1218}
1219
1220static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
1221 int *valp,
1222 int write, void *data)
1223{
1224 if (write) {
1225 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
1226
1227 if (jif > INT_MAX)
1228 return 1;
1229 WRITE_ONCE(*valp, (int)jif);
1230 } else {
1231 int val = READ_ONCE(*valp);
1232 unsigned long lval;
1233 if (val < 0) {
1234 *negp = true;
1235 lval = -(unsigned long)val;
1236 } else {
1237 *negp = false;
1238 lval = (unsigned long)val;
1239 }
1240 *lvalp = jiffies_to_msecs(lval);
1241 }
1242 return 0;
1243}
1244
1245static int do_proc_dointvec_ms_jiffies_minmax_conv(bool *negp, unsigned long *lvalp,
1246 int *valp, int write, void *data)
1247{
1248 int tmp, ret;
1249 struct do_proc_dointvec_minmax_conv_param *param = data;
1250 /*
1251 * If writing, first do so via a temporary local int so we can
1252 * bounds-check it before touching *valp.
1253 */
1254 int *ip = write ? &tmp : valp;
1255
1256 ret = do_proc_dointvec_ms_jiffies_conv(negp, lvalp, ip, write, data);
1257 if (ret)
1258 return ret;
1259
1260 if (write) {
1261 if ((param->min && *param->min > tmp) ||
1262 (param->max && *param->max < tmp))
1263 return -EINVAL;
1264 *valp = tmp;
1265 }
1266 return 0;
1267}
1268
1269/**
1270 * proc_dointvec_jiffies - read a vector of integers as seconds
1271 * @table: the sysctl table
1272 * @write: %TRUE if this is a write to the sysctl file
1273 * @buffer: the user buffer
1274 * @lenp: the size of the user buffer
1275 * @ppos: file position
1276 *
1277 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1278 * values from/to the user buffer, treated as an ASCII string.
1279 * The values read are assumed to be in seconds, and are converted into
1280 * jiffies.
1281 *
1282 * Returns 0 on success.
1283 */
1284int proc_dointvec_jiffies(struct ctl_table *table, int write,
1285 void *buffer, size_t *lenp, loff_t *ppos)
1286{
1287 return do_proc_dointvec(table,write,buffer,lenp,ppos,
1288 do_proc_dointvec_jiffies_conv,NULL);
1289}
1290
1291int proc_dointvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1292 void *buffer, size_t *lenp, loff_t *ppos)
1293{
1294 struct do_proc_dointvec_minmax_conv_param param = {
1295 .min = (int *) table->extra1,
1296 .max = (int *) table->extra2,
1297 };
1298 return do_proc_dointvec(table, write, buffer, lenp, ppos,
1299 do_proc_dointvec_ms_jiffies_minmax_conv, ¶m);
1300}
1301
1302/**
1303 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
1304 * @table: the sysctl table
1305 * @write: %TRUE if this is a write to the sysctl file
1306 * @buffer: the user buffer
1307 * @lenp: the size of the user buffer
1308 * @ppos: pointer to the file position
1309 *
1310 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1311 * values from/to the user buffer, treated as an ASCII string.
1312 * The values read are assumed to be in 1/USER_HZ seconds, and
1313 * are converted into jiffies.
1314 *
1315 * Returns 0 on success.
1316 */
1317int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1318 void *buffer, size_t *lenp, loff_t *ppos)
1319{
1320 return do_proc_dointvec(table, write, buffer, lenp, ppos,
1321 do_proc_dointvec_userhz_jiffies_conv, NULL);
1322}
1323
1324/**
1325 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
1326 * @table: the sysctl table
1327 * @write: %TRUE if this is a write to the sysctl file
1328 * @buffer: the user buffer
1329 * @lenp: the size of the user buffer
1330 * @ppos: file position
1331 * @ppos: the current position in the file
1332 *
1333 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1334 * values from/to the user buffer, treated as an ASCII string.
1335 * The values read are assumed to be in 1/1000 seconds, and
1336 * are converted into jiffies.
1337 *
1338 * Returns 0 on success.
1339 */
1340int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, void *buffer,
1341 size_t *lenp, loff_t *ppos)
1342{
1343 return do_proc_dointvec(table, write, buffer, lenp, ppos,
1344 do_proc_dointvec_ms_jiffies_conv, NULL);
1345}
1346
1347static int proc_do_cad_pid(struct ctl_table *table, int write, void *buffer,
1348 size_t *lenp, loff_t *ppos)
1349{
1350 struct pid *new_pid;
1351 pid_t tmp;
1352 int r;
1353
1354 tmp = pid_vnr(cad_pid);
1355
1356 r = __do_proc_dointvec(&tmp, table, write, buffer,
1357 lenp, ppos, NULL, NULL);
1358 if (r || !write)
1359 return r;
1360
1361 new_pid = find_get_pid(tmp);
1362 if (!new_pid)
1363 return -ESRCH;
1364
1365 put_pid(xchg(&cad_pid, new_pid));
1366 return 0;
1367}
1368
1369/**
1370 * proc_do_large_bitmap - read/write from/to a large bitmap
1371 * @table: the sysctl table
1372 * @write: %TRUE if this is a write to the sysctl file
1373 * @buffer: the user buffer
1374 * @lenp: the size of the user buffer
1375 * @ppos: file position
1376 *
1377 * The bitmap is stored at table->data and the bitmap length (in bits)
1378 * in table->maxlen.
1379 *
1380 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
1381 * large bitmaps may be represented in a compact manner. Writing into
1382 * the file will clear the bitmap then update it with the given input.
1383 *
1384 * Returns 0 on success.
1385 */
1386int proc_do_large_bitmap(struct ctl_table *table, int write,
1387 void *buffer, size_t *lenp, loff_t *ppos)
1388{
1389 int err = 0;
1390 size_t left = *lenp;
1391 unsigned long bitmap_len = table->maxlen;
1392 unsigned long *bitmap = *(unsigned long **) table->data;
1393 unsigned long *tmp_bitmap = NULL;
1394 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
1395
1396 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
1397 *lenp = 0;
1398 return 0;
1399 }
1400
1401 if (write) {
1402 char *p = buffer;
1403 size_t skipped = 0;
1404
1405 if (left > PAGE_SIZE - 1) {
1406 left = PAGE_SIZE - 1;
1407 /* How much of the buffer we'll skip this pass */
1408 skipped = *lenp - left;
1409 }
1410
1411 tmp_bitmap = bitmap_zalloc(bitmap_len, GFP_KERNEL);
1412 if (!tmp_bitmap)
1413 return -ENOMEM;
1414 proc_skip_char(&p, &left, '\n');
1415 while (!err && left) {
1416 unsigned long val_a, val_b;
1417 bool neg;
1418 size_t saved_left;
1419
1420 /* In case we stop parsing mid-number, we can reset */
1421 saved_left = left;
1422 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
1423 sizeof(tr_a), &c);
1424 /*
1425 * If we consumed the entirety of a truncated buffer or
1426 * only one char is left (may be a "-"), then stop here,
1427 * reset, & come back for more.
1428 */
1429 if ((left <= 1) && skipped) {
1430 left = saved_left;
1431 break;
1432 }
1433
1434 if (err)
1435 break;
1436 if (val_a >= bitmap_len || neg) {
1437 err = -EINVAL;
1438 break;
1439 }
1440
1441 val_b = val_a;
1442 if (left) {
1443 p++;
1444 left--;
1445 }
1446
1447 if (c == '-') {
1448 err = proc_get_long(&p, &left, &val_b,
1449 &neg, tr_b, sizeof(tr_b),
1450 &c);
1451 /*
1452 * If we consumed all of a truncated buffer or
1453 * then stop here, reset, & come back for more.
1454 */
1455 if (!left && skipped) {
1456 left = saved_left;
1457 break;
1458 }
1459
1460 if (err)
1461 break;
1462 if (val_b >= bitmap_len || neg ||
1463 val_a > val_b) {
1464 err = -EINVAL;
1465 break;
1466 }
1467 if (left) {
1468 p++;
1469 left--;
1470 }
1471 }
1472
1473 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
1474 proc_skip_char(&p, &left, '\n');
1475 }
1476 left += skipped;
1477 } else {
1478 unsigned long bit_a, bit_b = 0;
1479 bool first = 1;
1480
1481 while (left) {
1482 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
1483 if (bit_a >= bitmap_len)
1484 break;
1485 bit_b = find_next_zero_bit(bitmap, bitmap_len,
1486 bit_a + 1) - 1;
1487
1488 if (!first)
1489 proc_put_char(&buffer, &left, ',');
1490 proc_put_long(&buffer, &left, bit_a, false);
1491 if (bit_a != bit_b) {
1492 proc_put_char(&buffer, &left, '-');
1493 proc_put_long(&buffer, &left, bit_b, false);
1494 }
1495
1496 first = 0; bit_b++;
1497 }
1498 proc_put_char(&buffer, &left, '\n');
1499 }
1500
1501 if (!err) {
1502 if (write) {
1503 if (*ppos)
1504 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
1505 else
1506 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
1507 }
1508 *lenp -= left;
1509 *ppos += *lenp;
1510 }
1511
1512 bitmap_free(tmp_bitmap);
1513 return err;
1514}
1515
1516#else /* CONFIG_PROC_SYSCTL */
1517
1518int proc_dostring(struct ctl_table *table, int write,
1519 void *buffer, size_t *lenp, loff_t *ppos)
1520{
1521 return -ENOSYS;
1522}
1523
1524int proc_dobool(struct ctl_table *table, int write,
1525 void *buffer, size_t *lenp, loff_t *ppos)
1526{
1527 return -ENOSYS;
1528}
1529
1530int proc_dointvec(struct ctl_table *table, int write,
1531 void *buffer, size_t *lenp, loff_t *ppos)
1532{
1533 return -ENOSYS;
1534}
1535
1536int proc_douintvec(struct ctl_table *table, int write,
1537 void *buffer, size_t *lenp, loff_t *ppos)
1538{
1539 return -ENOSYS;
1540}
1541
1542int proc_dointvec_minmax(struct ctl_table *table, int write,
1543 void *buffer, size_t *lenp, loff_t *ppos)
1544{
1545 return -ENOSYS;
1546}
1547
1548int proc_douintvec_minmax(struct ctl_table *table, int write,
1549 void *buffer, size_t *lenp, loff_t *ppos)
1550{
1551 return -ENOSYS;
1552}
1553
1554int proc_dou8vec_minmax(struct ctl_table *table, int write,
1555 void *buffer, size_t *lenp, loff_t *ppos)
1556{
1557 return -ENOSYS;
1558}
1559
1560int proc_dointvec_jiffies(struct ctl_table *table, int write,
1561 void *buffer, size_t *lenp, loff_t *ppos)
1562{
1563 return -ENOSYS;
1564}
1565
1566int proc_dointvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1567 void *buffer, size_t *lenp, loff_t *ppos)
1568{
1569 return -ENOSYS;
1570}
1571
1572int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1573 void *buffer, size_t *lenp, loff_t *ppos)
1574{
1575 return -ENOSYS;
1576}
1577
1578int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
1579 void *buffer, size_t *lenp, loff_t *ppos)
1580{
1581 return -ENOSYS;
1582}
1583
1584int proc_doulongvec_minmax(struct ctl_table *table, int write,
1585 void *buffer, size_t *lenp, loff_t *ppos)
1586{
1587 return -ENOSYS;
1588}
1589
1590int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1591 void *buffer, size_t *lenp, loff_t *ppos)
1592{
1593 return -ENOSYS;
1594}
1595
1596int proc_do_large_bitmap(struct ctl_table *table, int write,
1597 void *buffer, size_t *lenp, loff_t *ppos)
1598{
1599 return -ENOSYS;
1600}
1601
1602#endif /* CONFIG_PROC_SYSCTL */
1603
1604#if defined(CONFIG_SYSCTL)
1605int proc_do_static_key(struct ctl_table *table, int write,
1606 void *buffer, size_t *lenp, loff_t *ppos)
1607{
1608 struct static_key *key = (struct static_key *)table->data;
1609 static DEFINE_MUTEX(static_key_mutex);
1610 int val, ret;
1611 struct ctl_table tmp = {
1612 .data = &val,
1613 .maxlen = sizeof(val),
1614 .mode = table->mode,
1615 .extra1 = SYSCTL_ZERO,
1616 .extra2 = SYSCTL_ONE,
1617 };
1618
1619 if (write && !capable(CAP_SYS_ADMIN))
1620 return -EPERM;
1621
1622 mutex_lock(&static_key_mutex);
1623 val = static_key_enabled(key);
1624 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
1625 if (write && !ret) {
1626 if (val)
1627 static_key_enable(key);
1628 else
1629 static_key_disable(key);
1630 }
1631 mutex_unlock(&static_key_mutex);
1632 return ret;
1633}
1634
1635static struct ctl_table kern_table[] = {
1636 {
1637 .procname = "panic",
1638 .data = &panic_timeout,
1639 .maxlen = sizeof(int),
1640 .mode = 0644,
1641 .proc_handler = proc_dointvec,
1642 },
1643#ifdef CONFIG_PROC_SYSCTL
1644 {
1645 .procname = "tainted",
1646 .maxlen = sizeof(long),
1647 .mode = 0644,
1648 .proc_handler = proc_taint,
1649 },
1650 {
1651 .procname = "sysctl_writes_strict",
1652 .data = &sysctl_writes_strict,
1653 .maxlen = sizeof(int),
1654 .mode = 0644,
1655 .proc_handler = proc_dointvec_minmax,
1656 .extra1 = SYSCTL_NEG_ONE,
1657 .extra2 = SYSCTL_ONE,
1658 },
1659#endif
1660 {
1661 .procname = "print-fatal-signals",
1662 .data = &print_fatal_signals,
1663 .maxlen = sizeof(int),
1664 .mode = 0644,
1665 .proc_handler = proc_dointvec,
1666 },
1667#ifdef CONFIG_SPARC
1668 {
1669 .procname = "reboot-cmd",
1670 .data = reboot_command,
1671 .maxlen = 256,
1672 .mode = 0644,
1673 .proc_handler = proc_dostring,
1674 },
1675 {
1676 .procname = "stop-a",
1677 .data = &stop_a_enabled,
1678 .maxlen = sizeof (int),
1679 .mode = 0644,
1680 .proc_handler = proc_dointvec,
1681 },
1682 {
1683 .procname = "scons-poweroff",
1684 .data = &scons_pwroff,
1685 .maxlen = sizeof (int),
1686 .mode = 0644,
1687 .proc_handler = proc_dointvec,
1688 },
1689#endif
1690#ifdef CONFIG_SPARC64
1691 {
1692 .procname = "tsb-ratio",
1693 .data = &sysctl_tsb_ratio,
1694 .maxlen = sizeof (int),
1695 .mode = 0644,
1696 .proc_handler = proc_dointvec,
1697 },
1698#endif
1699#ifdef CONFIG_PARISC
1700 {
1701 .procname = "soft-power",
1702 .data = &pwrsw_enabled,
1703 .maxlen = sizeof (int),
1704 .mode = 0644,
1705 .proc_handler = proc_dointvec,
1706 },
1707#endif
1708#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
1709 {
1710 .procname = "unaligned-trap",
1711 .data = &unaligned_enabled,
1712 .maxlen = sizeof (int),
1713 .mode = 0644,
1714 .proc_handler = proc_dointvec,
1715 },
1716#endif
1717#ifdef CONFIG_STACK_TRACER
1718 {
1719 .procname = "stack_tracer_enabled",
1720 .data = &stack_tracer_enabled,
1721 .maxlen = sizeof(int),
1722 .mode = 0644,
1723 .proc_handler = stack_trace_sysctl,
1724 },
1725#endif
1726#ifdef CONFIG_TRACING
1727 {
1728 .procname = "ftrace_dump_on_oops",
1729 .data = &ftrace_dump_on_oops,
1730 .maxlen = sizeof(int),
1731 .mode = 0644,
1732 .proc_handler = proc_dointvec,
1733 },
1734 {
1735 .procname = "traceoff_on_warning",
1736 .data = &__disable_trace_on_warning,
1737 .maxlen = sizeof(__disable_trace_on_warning),
1738 .mode = 0644,
1739 .proc_handler = proc_dointvec,
1740 },
1741 {
1742 .procname = "tracepoint_printk",
1743 .data = &tracepoint_printk,
1744 .maxlen = sizeof(tracepoint_printk),
1745 .mode = 0644,
1746 .proc_handler = tracepoint_printk_sysctl,
1747 },
1748#endif
1749#ifdef CONFIG_MODULES
1750 {
1751 .procname = "modprobe",
1752 .data = &modprobe_path,
1753 .maxlen = KMOD_PATH_LEN,
1754 .mode = 0644,
1755 .proc_handler = proc_dostring,
1756 },
1757 {
1758 .procname = "modules_disabled",
1759 .data = &modules_disabled,
1760 .maxlen = sizeof(int),
1761 .mode = 0644,
1762 /* only handle a transition from default "0" to "1" */
1763 .proc_handler = proc_dointvec_minmax,
1764 .extra1 = SYSCTL_ONE,
1765 .extra2 = SYSCTL_ONE,
1766 },
1767#endif
1768#ifdef CONFIG_UEVENT_HELPER
1769 {
1770 .procname = "hotplug",
1771 .data = &uevent_helper,
1772 .maxlen = UEVENT_HELPER_PATH_LEN,
1773 .mode = 0644,
1774 .proc_handler = proc_dostring,
1775 },
1776#endif
1777#ifdef CONFIG_MAGIC_SYSRQ
1778 {
1779 .procname = "sysrq",
1780 .data = NULL,
1781 .maxlen = sizeof (int),
1782 .mode = 0644,
1783 .proc_handler = sysrq_sysctl_handler,
1784 },
1785#endif
1786#ifdef CONFIG_PROC_SYSCTL
1787 {
1788 .procname = "cad_pid",
1789 .data = NULL,
1790 .maxlen = sizeof (int),
1791 .mode = 0600,
1792 .proc_handler = proc_do_cad_pid,
1793 },
1794#endif
1795 {
1796 .procname = "threads-max",
1797 .data = NULL,
1798 .maxlen = sizeof(int),
1799 .mode = 0644,
1800 .proc_handler = sysctl_max_threads,
1801 },
1802 {
1803 .procname = "usermodehelper",
1804 .mode = 0555,
1805 .child = usermodehelper_table,
1806 },
1807 {
1808 .procname = "overflowuid",
1809 .data = &overflowuid,
1810 .maxlen = sizeof(int),
1811 .mode = 0644,
1812 .proc_handler = proc_dointvec_minmax,
1813 .extra1 = SYSCTL_ZERO,
1814 .extra2 = SYSCTL_MAXOLDUID,
1815 },
1816 {
1817 .procname = "overflowgid",
1818 .data = &overflowgid,
1819 .maxlen = sizeof(int),
1820 .mode = 0644,
1821 .proc_handler = proc_dointvec_minmax,
1822 .extra1 = SYSCTL_ZERO,
1823 .extra2 = SYSCTL_MAXOLDUID,
1824 },
1825#ifdef CONFIG_S390
1826 {
1827 .procname = "userprocess_debug",
1828 .data = &show_unhandled_signals,
1829 .maxlen = sizeof(int),
1830 .mode = 0644,
1831 .proc_handler = proc_dointvec,
1832 },
1833#endif
1834 {
1835 .procname = "pid_max",
1836 .data = &pid_max,
1837 .maxlen = sizeof (int),
1838 .mode = 0644,
1839 .proc_handler = proc_dointvec_minmax,
1840 .extra1 = &pid_max_min,
1841 .extra2 = &pid_max_max,
1842 },
1843 {
1844 .procname = "panic_on_oops",
1845 .data = &panic_on_oops,
1846 .maxlen = sizeof(int),
1847 .mode = 0644,
1848 .proc_handler = proc_dointvec,
1849 },
1850 {
1851 .procname = "panic_print",
1852 .data = &panic_print,
1853 .maxlen = sizeof(unsigned long),
1854 .mode = 0644,
1855 .proc_handler = proc_doulongvec_minmax,
1856 },
1857 {
1858 .procname = "ngroups_max",
1859 .data = (void *)&ngroups_max,
1860 .maxlen = sizeof (int),
1861 .mode = 0444,
1862 .proc_handler = proc_dointvec,
1863 },
1864 {
1865 .procname = "cap_last_cap",
1866 .data = (void *)&cap_last_cap,
1867 .maxlen = sizeof(int),
1868 .mode = 0444,
1869 .proc_handler = proc_dointvec,
1870 },
1871#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
1872 {
1873 .procname = "unknown_nmi_panic",
1874 .data = &unknown_nmi_panic,
1875 .maxlen = sizeof (int),
1876 .mode = 0644,
1877 .proc_handler = proc_dointvec,
1878 },
1879#endif
1880
1881#if (defined(CONFIG_X86_32) || defined(CONFIG_PARISC)) && \
1882 defined(CONFIG_DEBUG_STACKOVERFLOW)
1883 {
1884 .procname = "panic_on_stackoverflow",
1885 .data = &sysctl_panic_on_stackoverflow,
1886 .maxlen = sizeof(int),
1887 .mode = 0644,
1888 .proc_handler = proc_dointvec,
1889 },
1890#endif
1891#if defined(CONFIG_X86)
1892 {
1893 .procname = "panic_on_unrecovered_nmi",
1894 .data = &panic_on_unrecovered_nmi,
1895 .maxlen = sizeof(int),
1896 .mode = 0644,
1897 .proc_handler = proc_dointvec,
1898 },
1899 {
1900 .procname = "panic_on_io_nmi",
1901 .data = &panic_on_io_nmi,
1902 .maxlen = sizeof(int),
1903 .mode = 0644,
1904 .proc_handler = proc_dointvec,
1905 },
1906 {
1907 .procname = "bootloader_type",
1908 .data = &bootloader_type,
1909 .maxlen = sizeof (int),
1910 .mode = 0444,
1911 .proc_handler = proc_dointvec,
1912 },
1913 {
1914 .procname = "bootloader_version",
1915 .data = &bootloader_version,
1916 .maxlen = sizeof (int),
1917 .mode = 0444,
1918 .proc_handler = proc_dointvec,
1919 },
1920 {
1921 .procname = "io_delay_type",
1922 .data = &io_delay_type,
1923 .maxlen = sizeof(int),
1924 .mode = 0644,
1925 .proc_handler = proc_dointvec,
1926 },
1927#endif
1928#if defined(CONFIG_MMU)
1929 {
1930 .procname = "randomize_va_space",
1931 .data = &randomize_va_space,
1932 .maxlen = sizeof(int),
1933 .mode = 0644,
1934 .proc_handler = proc_dointvec,
1935 },
1936#endif
1937#if defined(CONFIG_S390) && defined(CONFIG_SMP)
1938 {
1939 .procname = "spin_retry",
1940 .data = &spin_retry,
1941 .maxlen = sizeof (int),
1942 .mode = 0644,
1943 .proc_handler = proc_dointvec,
1944 },
1945#endif
1946#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1947 {
1948 .procname = "acpi_video_flags",
1949 .data = &acpi_realmode_flags,
1950 .maxlen = sizeof (unsigned long),
1951 .mode = 0644,
1952 .proc_handler = proc_doulongvec_minmax,
1953 },
1954#endif
1955#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1956 {
1957 .procname = "ignore-unaligned-usertrap",
1958 .data = &no_unaligned_warning,
1959 .maxlen = sizeof (int),
1960 .mode = 0644,
1961 .proc_handler = proc_dointvec,
1962 },
1963#endif
1964#ifdef CONFIG_IA64
1965 {
1966 .procname = "unaligned-dump-stack",
1967 .data = &unaligned_dump_stack,
1968 .maxlen = sizeof (int),
1969 .mode = 0644,
1970 .proc_handler = proc_dointvec,
1971 },
1972#endif
1973#ifdef CONFIG_RT_MUTEXES
1974 {
1975 .procname = "max_lock_depth",
1976 .data = &max_lock_depth,
1977 .maxlen = sizeof(int),
1978 .mode = 0644,
1979 .proc_handler = proc_dointvec,
1980 },
1981#endif
1982#ifdef CONFIG_KEYS
1983 {
1984 .procname = "keys",
1985 .mode = 0555,
1986 .child = key_sysctls,
1987 },
1988#endif
1989#ifdef CONFIG_PERF_EVENTS
1990 /*
1991 * User-space scripts rely on the existence of this file
1992 * as a feature check for perf_events being enabled.
1993 *
1994 * So it's an ABI, do not remove!
1995 */
1996 {
1997 .procname = "perf_event_paranoid",
1998 .data = &sysctl_perf_event_paranoid,
1999 .maxlen = sizeof(sysctl_perf_event_paranoid),
2000 .mode = 0644,
2001 .proc_handler = proc_dointvec,
2002 },
2003 {
2004 .procname = "perf_event_mlock_kb",
2005 .data = &sysctl_perf_event_mlock,
2006 .maxlen = sizeof(sysctl_perf_event_mlock),
2007 .mode = 0644,
2008 .proc_handler = proc_dointvec,
2009 },
2010 {
2011 .procname = "perf_event_max_sample_rate",
2012 .data = &sysctl_perf_event_sample_rate,
2013 .maxlen = sizeof(sysctl_perf_event_sample_rate),
2014 .mode = 0644,
2015 .proc_handler = perf_proc_update_handler,
2016 .extra1 = SYSCTL_ONE,
2017 },
2018 {
2019 .procname = "perf_cpu_time_max_percent",
2020 .data = &sysctl_perf_cpu_time_max_percent,
2021 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
2022 .mode = 0644,
2023 .proc_handler = perf_cpu_time_max_percent_handler,
2024 .extra1 = SYSCTL_ZERO,
2025 .extra2 = SYSCTL_ONE_HUNDRED,
2026 },
2027 {
2028 .procname = "perf_event_max_stack",
2029 .data = &sysctl_perf_event_max_stack,
2030 .maxlen = sizeof(sysctl_perf_event_max_stack),
2031 .mode = 0644,
2032 .proc_handler = perf_event_max_stack_handler,
2033 .extra1 = SYSCTL_ZERO,
2034 .extra2 = (void *)&six_hundred_forty_kb,
2035 },
2036 {
2037 .procname = "perf_event_max_contexts_per_stack",
2038 .data = &sysctl_perf_event_max_contexts_per_stack,
2039 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
2040 .mode = 0644,
2041 .proc_handler = perf_event_max_stack_handler,
2042 .extra1 = SYSCTL_ZERO,
2043 .extra2 = SYSCTL_ONE_THOUSAND,
2044 },
2045#endif
2046 {
2047 .procname = "panic_on_warn",
2048 .data = &panic_on_warn,
2049 .maxlen = sizeof(int),
2050 .mode = 0644,
2051 .proc_handler = proc_dointvec_minmax,
2052 .extra1 = SYSCTL_ZERO,
2053 .extra2 = SYSCTL_ONE,
2054 },
2055#ifdef CONFIG_TREE_RCU
2056 {
2057 .procname = "panic_on_rcu_stall",
2058 .data = &sysctl_panic_on_rcu_stall,
2059 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
2060 .mode = 0644,
2061 .proc_handler = proc_dointvec_minmax,
2062 .extra1 = SYSCTL_ZERO,
2063 .extra2 = SYSCTL_ONE,
2064 },
2065 {
2066 .procname = "max_rcu_stall_to_panic",
2067 .data = &sysctl_max_rcu_stall_to_panic,
2068 .maxlen = sizeof(sysctl_max_rcu_stall_to_panic),
2069 .mode = 0644,
2070 .proc_handler = proc_dointvec_minmax,
2071 .extra1 = SYSCTL_ONE,
2072 .extra2 = SYSCTL_INT_MAX,
2073 },
2074#endif
2075 { }
2076};
2077
2078static struct ctl_table vm_table[] = {
2079 {
2080 .procname = "overcommit_memory",
2081 .data = &sysctl_overcommit_memory,
2082 .maxlen = sizeof(sysctl_overcommit_memory),
2083 .mode = 0644,
2084 .proc_handler = overcommit_policy_handler,
2085 .extra1 = SYSCTL_ZERO,
2086 .extra2 = SYSCTL_TWO,
2087 },
2088 {
2089 .procname = "overcommit_ratio",
2090 .data = &sysctl_overcommit_ratio,
2091 .maxlen = sizeof(sysctl_overcommit_ratio),
2092 .mode = 0644,
2093 .proc_handler = overcommit_ratio_handler,
2094 },
2095 {
2096 .procname = "overcommit_kbytes",
2097 .data = &sysctl_overcommit_kbytes,
2098 .maxlen = sizeof(sysctl_overcommit_kbytes),
2099 .mode = 0644,
2100 .proc_handler = overcommit_kbytes_handler,
2101 },
2102 {
2103 .procname = "page-cluster",
2104 .data = &page_cluster,
2105 .maxlen = sizeof(int),
2106 .mode = 0644,
2107 .proc_handler = proc_dointvec_minmax,
2108 .extra1 = SYSCTL_ZERO,
2109 .extra2 = (void *)&page_cluster_max,
2110 },
2111 {
2112 .procname = "dirtytime_expire_seconds",
2113 .data = &dirtytime_expire_interval,
2114 .maxlen = sizeof(dirtytime_expire_interval),
2115 .mode = 0644,
2116 .proc_handler = dirtytime_interval_handler,
2117 .extra1 = SYSCTL_ZERO,
2118 },
2119 {
2120 .procname = "swappiness",
2121 .data = &vm_swappiness,
2122 .maxlen = sizeof(vm_swappiness),
2123 .mode = 0644,
2124 .proc_handler = proc_dointvec_minmax,
2125 .extra1 = SYSCTL_ZERO,
2126 .extra2 = SYSCTL_TWO_HUNDRED,
2127 },
2128#ifdef CONFIG_NUMA
2129 {
2130 .procname = "numa_stat",
2131 .data = &sysctl_vm_numa_stat,
2132 .maxlen = sizeof(int),
2133 .mode = 0644,
2134 .proc_handler = sysctl_vm_numa_stat_handler,
2135 .extra1 = SYSCTL_ZERO,
2136 .extra2 = SYSCTL_ONE,
2137 },
2138#endif
2139#ifdef CONFIG_HUGETLB_PAGE
2140 {
2141 .procname = "nr_hugepages",
2142 .data = NULL,
2143 .maxlen = sizeof(unsigned long),
2144 .mode = 0644,
2145 .proc_handler = hugetlb_sysctl_handler,
2146 },
2147#ifdef CONFIG_NUMA
2148 {
2149 .procname = "nr_hugepages_mempolicy",
2150 .data = NULL,
2151 .maxlen = sizeof(unsigned long),
2152 .mode = 0644,
2153 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
2154 },
2155#endif
2156 {
2157 .procname = "hugetlb_shm_group",
2158 .data = &sysctl_hugetlb_shm_group,
2159 .maxlen = sizeof(gid_t),
2160 .mode = 0644,
2161 .proc_handler = proc_dointvec,
2162 },
2163 {
2164 .procname = "nr_overcommit_hugepages",
2165 .data = NULL,
2166 .maxlen = sizeof(unsigned long),
2167 .mode = 0644,
2168 .proc_handler = hugetlb_overcommit_handler,
2169 },
2170#endif
2171 {
2172 .procname = "lowmem_reserve_ratio",
2173 .data = &sysctl_lowmem_reserve_ratio,
2174 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
2175 .mode = 0644,
2176 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
2177 },
2178 {
2179 .procname = "drop_caches",
2180 .data = &sysctl_drop_caches,
2181 .maxlen = sizeof(int),
2182 .mode = 0200,
2183 .proc_handler = drop_caches_sysctl_handler,
2184 .extra1 = SYSCTL_ONE,
2185 .extra2 = SYSCTL_FOUR,
2186 },
2187#ifdef CONFIG_COMPACTION
2188 {
2189 .procname = "compact_memory",
2190 .data = NULL,
2191 .maxlen = sizeof(int),
2192 .mode = 0200,
2193 .proc_handler = sysctl_compaction_handler,
2194 },
2195 {
2196 .procname = "compaction_proactiveness",
2197 .data = &sysctl_compaction_proactiveness,
2198 .maxlen = sizeof(sysctl_compaction_proactiveness),
2199 .mode = 0644,
2200 .proc_handler = compaction_proactiveness_sysctl_handler,
2201 .extra1 = SYSCTL_ZERO,
2202 .extra2 = SYSCTL_ONE_HUNDRED,
2203 },
2204 {
2205 .procname = "extfrag_threshold",
2206 .data = &sysctl_extfrag_threshold,
2207 .maxlen = sizeof(int),
2208 .mode = 0644,
2209 .proc_handler = proc_dointvec_minmax,
2210 .extra1 = SYSCTL_ZERO,
2211 .extra2 = SYSCTL_ONE_THOUSAND,
2212 },
2213 {
2214 .procname = "compact_unevictable_allowed",
2215 .data = &sysctl_compact_unevictable_allowed,
2216 .maxlen = sizeof(int),
2217 .mode = 0644,
2218 .proc_handler = proc_dointvec_minmax_warn_RT_change,
2219 .extra1 = SYSCTL_ZERO,
2220 .extra2 = SYSCTL_ONE,
2221 },
2222
2223#endif /* CONFIG_COMPACTION */
2224 {
2225 .procname = "min_free_kbytes",
2226 .data = &min_free_kbytes,
2227 .maxlen = sizeof(min_free_kbytes),
2228 .mode = 0644,
2229 .proc_handler = min_free_kbytes_sysctl_handler,
2230 .extra1 = SYSCTL_ZERO,
2231 },
2232 {
2233 .procname = "watermark_boost_factor",
2234 .data = &watermark_boost_factor,
2235 .maxlen = sizeof(watermark_boost_factor),
2236 .mode = 0644,
2237 .proc_handler = proc_dointvec_minmax,
2238 .extra1 = SYSCTL_ZERO,
2239 },
2240 {
2241 .procname = "watermark_scale_factor",
2242 .data = &watermark_scale_factor,
2243 .maxlen = sizeof(watermark_scale_factor),
2244 .mode = 0644,
2245 .proc_handler = watermark_scale_factor_sysctl_handler,
2246 .extra1 = SYSCTL_ONE,
2247 .extra2 = SYSCTL_THREE_THOUSAND,
2248 },
2249 {
2250 .procname = "percpu_pagelist_high_fraction",
2251 .data = &percpu_pagelist_high_fraction,
2252 .maxlen = sizeof(percpu_pagelist_high_fraction),
2253 .mode = 0644,
2254 .proc_handler = percpu_pagelist_high_fraction_sysctl_handler,
2255 .extra1 = SYSCTL_ZERO,
2256 },
2257 {
2258 .procname = "page_lock_unfairness",
2259 .data = &sysctl_page_lock_unfairness,
2260 .maxlen = sizeof(sysctl_page_lock_unfairness),
2261 .mode = 0644,
2262 .proc_handler = proc_dointvec_minmax,
2263 .extra1 = SYSCTL_ZERO,
2264 },
2265#ifdef CONFIG_MMU
2266 {
2267 .procname = "max_map_count",
2268 .data = &sysctl_max_map_count,
2269 .maxlen = sizeof(sysctl_max_map_count),
2270 .mode = 0644,
2271 .proc_handler = proc_dointvec_minmax,
2272 .extra1 = SYSCTL_ZERO,
2273 },
2274#else
2275 {
2276 .procname = "nr_trim_pages",
2277 .data = &sysctl_nr_trim_pages,
2278 .maxlen = sizeof(sysctl_nr_trim_pages),
2279 .mode = 0644,
2280 .proc_handler = proc_dointvec_minmax,
2281 .extra1 = SYSCTL_ZERO,
2282 },
2283#endif
2284 {
2285 .procname = "vfs_cache_pressure",
2286 .data = &sysctl_vfs_cache_pressure,
2287 .maxlen = sizeof(sysctl_vfs_cache_pressure),
2288 .mode = 0644,
2289 .proc_handler = proc_dointvec_minmax,
2290 .extra1 = SYSCTL_ZERO,
2291 },
2292#if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
2293 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
2294 {
2295 .procname = "legacy_va_layout",
2296 .data = &sysctl_legacy_va_layout,
2297 .maxlen = sizeof(sysctl_legacy_va_layout),
2298 .mode = 0644,
2299 .proc_handler = proc_dointvec_minmax,
2300 .extra1 = SYSCTL_ZERO,
2301 },
2302#endif
2303#ifdef CONFIG_NUMA
2304 {
2305 .procname = "zone_reclaim_mode",
2306 .data = &node_reclaim_mode,
2307 .maxlen = sizeof(node_reclaim_mode),
2308 .mode = 0644,
2309 .proc_handler = proc_dointvec_minmax,
2310 .extra1 = SYSCTL_ZERO,
2311 },
2312 {
2313 .procname = "min_unmapped_ratio",
2314 .data = &sysctl_min_unmapped_ratio,
2315 .maxlen = sizeof(sysctl_min_unmapped_ratio),
2316 .mode = 0644,
2317 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
2318 .extra1 = SYSCTL_ZERO,
2319 .extra2 = SYSCTL_ONE_HUNDRED,
2320 },
2321 {
2322 .procname = "min_slab_ratio",
2323 .data = &sysctl_min_slab_ratio,
2324 .maxlen = sizeof(sysctl_min_slab_ratio),
2325 .mode = 0644,
2326 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
2327 .extra1 = SYSCTL_ZERO,
2328 .extra2 = SYSCTL_ONE_HUNDRED,
2329 },
2330#endif
2331#ifdef CONFIG_SMP
2332 {
2333 .procname = "stat_interval",
2334 .data = &sysctl_stat_interval,
2335 .maxlen = sizeof(sysctl_stat_interval),
2336 .mode = 0644,
2337 .proc_handler = proc_dointvec_jiffies,
2338 },
2339 {
2340 .procname = "stat_refresh",
2341 .data = NULL,
2342 .maxlen = 0,
2343 .mode = 0600,
2344 .proc_handler = vmstat_refresh,
2345 },
2346#endif
2347#ifdef CONFIG_MMU
2348 {
2349 .procname = "mmap_min_addr",
2350 .data = &dac_mmap_min_addr,
2351 .maxlen = sizeof(unsigned long),
2352 .mode = 0644,
2353 .proc_handler = mmap_min_addr_handler,
2354 },
2355#endif
2356#ifdef CONFIG_NUMA
2357 {
2358 .procname = "numa_zonelist_order",
2359 .data = &numa_zonelist_order,
2360 .maxlen = NUMA_ZONELIST_ORDER_LEN,
2361 .mode = 0644,
2362 .proc_handler = numa_zonelist_order_handler,
2363 },
2364#endif
2365#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
2366 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
2367 {
2368 .procname = "vdso_enabled",
2369#ifdef CONFIG_X86_32
2370 .data = &vdso32_enabled,
2371 .maxlen = sizeof(vdso32_enabled),
2372#else
2373 .data = &vdso_enabled,
2374 .maxlen = sizeof(vdso_enabled),
2375#endif
2376 .mode = 0644,
2377 .proc_handler = proc_dointvec,
2378 .extra1 = SYSCTL_ZERO,
2379 },
2380#endif
2381#ifdef CONFIG_MEMORY_FAILURE
2382 {
2383 .procname = "memory_failure_early_kill",
2384 .data = &sysctl_memory_failure_early_kill,
2385 .maxlen = sizeof(sysctl_memory_failure_early_kill),
2386 .mode = 0644,
2387 .proc_handler = proc_dointvec_minmax,
2388 .extra1 = SYSCTL_ZERO,
2389 .extra2 = SYSCTL_ONE,
2390 },
2391 {
2392 .procname = "memory_failure_recovery",
2393 .data = &sysctl_memory_failure_recovery,
2394 .maxlen = sizeof(sysctl_memory_failure_recovery),
2395 .mode = 0644,
2396 .proc_handler = proc_dointvec_minmax,
2397 .extra1 = SYSCTL_ZERO,
2398 .extra2 = SYSCTL_ONE,
2399 },
2400#endif
2401 {
2402 .procname = "user_reserve_kbytes",
2403 .data = &sysctl_user_reserve_kbytes,
2404 .maxlen = sizeof(sysctl_user_reserve_kbytes),
2405 .mode = 0644,
2406 .proc_handler = proc_doulongvec_minmax,
2407 },
2408 {
2409 .procname = "admin_reserve_kbytes",
2410 .data = &sysctl_admin_reserve_kbytes,
2411 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
2412 .mode = 0644,
2413 .proc_handler = proc_doulongvec_minmax,
2414 },
2415#ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
2416 {
2417 .procname = "mmap_rnd_bits",
2418 .data = &mmap_rnd_bits,
2419 .maxlen = sizeof(mmap_rnd_bits),
2420 .mode = 0600,
2421 .proc_handler = proc_dointvec_minmax,
2422 .extra1 = (void *)&mmap_rnd_bits_min,
2423 .extra2 = (void *)&mmap_rnd_bits_max,
2424 },
2425#endif
2426#ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
2427 {
2428 .procname = "mmap_rnd_compat_bits",
2429 .data = &mmap_rnd_compat_bits,
2430 .maxlen = sizeof(mmap_rnd_compat_bits),
2431 .mode = 0600,
2432 .proc_handler = proc_dointvec_minmax,
2433 .extra1 = (void *)&mmap_rnd_compat_bits_min,
2434 .extra2 = (void *)&mmap_rnd_compat_bits_max,
2435 },
2436#endif
2437#ifdef CONFIG_USERFAULTFD
2438 {
2439 .procname = "unprivileged_userfaultfd",
2440 .data = &sysctl_unprivileged_userfaultfd,
2441 .maxlen = sizeof(sysctl_unprivileged_userfaultfd),
2442 .mode = 0644,
2443 .proc_handler = proc_dointvec_minmax,
2444 .extra1 = SYSCTL_ZERO,
2445 .extra2 = SYSCTL_ONE,
2446 },
2447#endif
2448 { }
2449};
2450
2451static struct ctl_table debug_table[] = {
2452#ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
2453 {
2454 .procname = "exception-trace",
2455 .data = &show_unhandled_signals,
2456 .maxlen = sizeof(int),
2457 .mode = 0644,
2458 .proc_handler = proc_dointvec
2459 },
2460#endif
2461 { }
2462};
2463
2464static struct ctl_table dev_table[] = {
2465 { }
2466};
2467
2468DECLARE_SYSCTL_BASE(kernel, kern_table);
2469DECLARE_SYSCTL_BASE(vm, vm_table);
2470DECLARE_SYSCTL_BASE(debug, debug_table);
2471DECLARE_SYSCTL_BASE(dev, dev_table);
2472
2473int __init sysctl_init_bases(void)
2474{
2475 register_sysctl_base(kernel);
2476 register_sysctl_base(vm);
2477 register_sysctl_base(debug);
2478 register_sysctl_base(dev);
2479
2480 return 0;
2481}
2482#endif /* CONFIG_SYSCTL */
2483/*
2484 * No sense putting this after each symbol definition, twice,
2485 * exception granted :-)
2486 */
2487EXPORT_SYMBOL(proc_dobool);
2488EXPORT_SYMBOL(proc_dointvec);
2489EXPORT_SYMBOL(proc_douintvec);
2490EXPORT_SYMBOL(proc_dointvec_jiffies);
2491EXPORT_SYMBOL(proc_dointvec_minmax);
2492EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
2493EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2494EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2495EXPORT_SYMBOL(proc_dostring);
2496EXPORT_SYMBOL(proc_doulongvec_minmax);
2497EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2498EXPORT_SYMBOL(proc_do_large_bitmap);
1/*
2 * sysctl.c: General linux system control interface
3 *
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12 * Horn.
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16 * Wendling.
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19 */
20
21#include <linux/module.h>
22#include <linux/mm.h>
23#include <linux/swap.h>
24#include <linux/slab.h>
25#include <linux/sysctl.h>
26#include <linux/signal.h>
27#include <linux/printk.h>
28#include <linux/proc_fs.h>
29#include <linux/security.h>
30#include <linux/ctype.h>
31#include <linux/kmemcheck.h>
32#include <linux/fs.h>
33#include <linux/init.h>
34#include <linux/kernel.h>
35#include <linux/kobject.h>
36#include <linux/net.h>
37#include <linux/sysrq.h>
38#include <linux/highuid.h>
39#include <linux/writeback.h>
40#include <linux/ratelimit.h>
41#include <linux/compaction.h>
42#include <linux/hugetlb.h>
43#include <linux/initrd.h>
44#include <linux/key.h>
45#include <linux/times.h>
46#include <linux/limits.h>
47#include <linux/dcache.h>
48#include <linux/dnotify.h>
49#include <linux/syscalls.h>
50#include <linux/vmstat.h>
51#include <linux/nfs_fs.h>
52#include <linux/acpi.h>
53#include <linux/reboot.h>
54#include <linux/ftrace.h>
55#include <linux/perf_event.h>
56#include <linux/kprobes.h>
57#include <linux/pipe_fs_i.h>
58#include <linux/oom.h>
59#include <linux/kmod.h>
60
61#include <asm/uaccess.h>
62#include <asm/processor.h>
63
64#ifdef CONFIG_X86
65#include <asm/nmi.h>
66#include <asm/stacktrace.h>
67#include <asm/io.h>
68#endif
69#ifdef CONFIG_BSD_PROCESS_ACCT
70#include <linux/acct.h>
71#endif
72#ifdef CONFIG_RT_MUTEXES
73#include <linux/rtmutex.h>
74#endif
75#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
76#include <linux/lockdep.h>
77#endif
78#ifdef CONFIG_CHR_DEV_SG
79#include <scsi/sg.h>
80#endif
81
82#ifdef CONFIG_LOCKUP_DETECTOR
83#include <linux/nmi.h>
84#endif
85
86
87#if defined(CONFIG_SYSCTL)
88
89/* External variables not in a header file. */
90extern int sysctl_overcommit_memory;
91extern int sysctl_overcommit_ratio;
92extern int max_threads;
93extern int core_uses_pid;
94extern int suid_dumpable;
95extern char core_pattern[];
96extern unsigned int core_pipe_limit;
97extern int pid_max;
98extern int min_free_kbytes;
99extern int pid_max_min, pid_max_max;
100extern int sysctl_drop_caches;
101extern int percpu_pagelist_fraction;
102extern int compat_log;
103extern int latencytop_enabled;
104extern int sysctl_nr_open_min, sysctl_nr_open_max;
105#ifndef CONFIG_MMU
106extern int sysctl_nr_trim_pages;
107#endif
108#ifdef CONFIG_BLOCK
109extern int blk_iopoll_enabled;
110#endif
111
112/* Constants used for minimum and maximum */
113#ifdef CONFIG_LOCKUP_DETECTOR
114static int sixty = 60;
115static int neg_one = -1;
116#endif
117
118static int zero;
119static int __maybe_unused one = 1;
120static int __maybe_unused two = 2;
121static int __maybe_unused three = 3;
122static unsigned long one_ul = 1;
123static int one_hundred = 100;
124#ifdef CONFIG_PRINTK
125static int ten_thousand = 10000;
126#endif
127
128/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
129static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
130
131/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
132static int maxolduid = 65535;
133static int minolduid;
134static int min_percpu_pagelist_fract = 8;
135
136static int ngroups_max = NGROUPS_MAX;
137
138#ifdef CONFIG_INOTIFY_USER
139#include <linux/inotify.h>
140#endif
141#ifdef CONFIG_SPARC
142#include <asm/system.h>
143#endif
144
145#ifdef CONFIG_SPARC64
146extern int sysctl_tsb_ratio;
147#endif
148
149#ifdef __hppa__
150extern int pwrsw_enabled;
151extern int unaligned_enabled;
152#endif
153
154#ifdef CONFIG_S390
155#ifdef CONFIG_MATHEMU
156extern int sysctl_ieee_emulation_warnings;
157#endif
158extern int sysctl_userprocess_debug;
159extern int spin_retry;
160#endif
161
162#ifdef CONFIG_IA64
163extern int no_unaligned_warning;
164extern int unaligned_dump_stack;
165#endif
166
167#ifdef CONFIG_PROC_SYSCTL
168static int proc_do_cad_pid(struct ctl_table *table, int write,
169 void __user *buffer, size_t *lenp, loff_t *ppos);
170static int proc_taint(struct ctl_table *table, int write,
171 void __user *buffer, size_t *lenp, loff_t *ppos);
172#endif
173
174#ifdef CONFIG_PRINTK
175static int proc_dmesg_restrict(struct ctl_table *table, int write,
176 void __user *buffer, size_t *lenp, loff_t *ppos);
177#endif
178
179#ifdef CONFIG_MAGIC_SYSRQ
180/* Note: sysrq code uses it's own private copy */
181static int __sysrq_enabled = SYSRQ_DEFAULT_ENABLE;
182
183static int sysrq_sysctl_handler(ctl_table *table, int write,
184 void __user *buffer, size_t *lenp,
185 loff_t *ppos)
186{
187 int error;
188
189 error = proc_dointvec(table, write, buffer, lenp, ppos);
190 if (error)
191 return error;
192
193 if (write)
194 sysrq_toggle_support(__sysrq_enabled);
195
196 return 0;
197}
198
199#endif
200
201static struct ctl_table root_table[];
202static struct ctl_table_root sysctl_table_root;
203static struct ctl_table_header root_table_header = {
204 {{.count = 1,
205 .ctl_table = root_table,
206 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),}},
207 .root = &sysctl_table_root,
208 .set = &sysctl_table_root.default_set,
209};
210static struct ctl_table_root sysctl_table_root = {
211 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
212 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
213};
214
215static struct ctl_table kern_table[];
216static struct ctl_table vm_table[];
217static struct ctl_table fs_table[];
218static struct ctl_table debug_table[];
219static struct ctl_table dev_table[];
220extern struct ctl_table random_table[];
221#ifdef CONFIG_EPOLL
222extern struct ctl_table epoll_table[];
223#endif
224
225#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
226int sysctl_legacy_va_layout;
227#endif
228
229/* The default sysctl tables: */
230
231static struct ctl_table root_table[] = {
232 {
233 .procname = "kernel",
234 .mode = 0555,
235 .child = kern_table,
236 },
237 {
238 .procname = "vm",
239 .mode = 0555,
240 .child = vm_table,
241 },
242 {
243 .procname = "fs",
244 .mode = 0555,
245 .child = fs_table,
246 },
247 {
248 .procname = "debug",
249 .mode = 0555,
250 .child = debug_table,
251 },
252 {
253 .procname = "dev",
254 .mode = 0555,
255 .child = dev_table,
256 },
257 { }
258};
259
260#ifdef CONFIG_SCHED_DEBUG
261static int min_sched_granularity_ns = 100000; /* 100 usecs */
262static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
263static int min_wakeup_granularity_ns; /* 0 usecs */
264static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
265static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
266static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
267#endif
268
269#ifdef CONFIG_COMPACTION
270static int min_extfrag_threshold;
271static int max_extfrag_threshold = 1000;
272#endif
273
274static struct ctl_table kern_table[] = {
275 {
276 .procname = "sched_child_runs_first",
277 .data = &sysctl_sched_child_runs_first,
278 .maxlen = sizeof(unsigned int),
279 .mode = 0644,
280 .proc_handler = proc_dointvec,
281 },
282#ifdef CONFIG_SCHED_DEBUG
283 {
284 .procname = "sched_min_granularity_ns",
285 .data = &sysctl_sched_min_granularity,
286 .maxlen = sizeof(unsigned int),
287 .mode = 0644,
288 .proc_handler = sched_proc_update_handler,
289 .extra1 = &min_sched_granularity_ns,
290 .extra2 = &max_sched_granularity_ns,
291 },
292 {
293 .procname = "sched_latency_ns",
294 .data = &sysctl_sched_latency,
295 .maxlen = sizeof(unsigned int),
296 .mode = 0644,
297 .proc_handler = sched_proc_update_handler,
298 .extra1 = &min_sched_granularity_ns,
299 .extra2 = &max_sched_granularity_ns,
300 },
301 {
302 .procname = "sched_wakeup_granularity_ns",
303 .data = &sysctl_sched_wakeup_granularity,
304 .maxlen = sizeof(unsigned int),
305 .mode = 0644,
306 .proc_handler = sched_proc_update_handler,
307 .extra1 = &min_wakeup_granularity_ns,
308 .extra2 = &max_wakeup_granularity_ns,
309 },
310 {
311 .procname = "sched_tunable_scaling",
312 .data = &sysctl_sched_tunable_scaling,
313 .maxlen = sizeof(enum sched_tunable_scaling),
314 .mode = 0644,
315 .proc_handler = sched_proc_update_handler,
316 .extra1 = &min_sched_tunable_scaling,
317 .extra2 = &max_sched_tunable_scaling,
318 },
319 {
320 .procname = "sched_migration_cost",
321 .data = &sysctl_sched_migration_cost,
322 .maxlen = sizeof(unsigned int),
323 .mode = 0644,
324 .proc_handler = proc_dointvec,
325 },
326 {
327 .procname = "sched_nr_migrate",
328 .data = &sysctl_sched_nr_migrate,
329 .maxlen = sizeof(unsigned int),
330 .mode = 0644,
331 .proc_handler = proc_dointvec,
332 },
333 {
334 .procname = "sched_time_avg",
335 .data = &sysctl_sched_time_avg,
336 .maxlen = sizeof(unsigned int),
337 .mode = 0644,
338 .proc_handler = proc_dointvec,
339 },
340 {
341 .procname = "sched_shares_window",
342 .data = &sysctl_sched_shares_window,
343 .maxlen = sizeof(unsigned int),
344 .mode = 0644,
345 .proc_handler = proc_dointvec,
346 },
347 {
348 .procname = "timer_migration",
349 .data = &sysctl_timer_migration,
350 .maxlen = sizeof(unsigned int),
351 .mode = 0644,
352 .proc_handler = proc_dointvec_minmax,
353 .extra1 = &zero,
354 .extra2 = &one,
355 },
356#endif
357 {
358 .procname = "sched_rt_period_us",
359 .data = &sysctl_sched_rt_period,
360 .maxlen = sizeof(unsigned int),
361 .mode = 0644,
362 .proc_handler = sched_rt_handler,
363 },
364 {
365 .procname = "sched_rt_runtime_us",
366 .data = &sysctl_sched_rt_runtime,
367 .maxlen = sizeof(int),
368 .mode = 0644,
369 .proc_handler = sched_rt_handler,
370 },
371#ifdef CONFIG_SCHED_AUTOGROUP
372 {
373 .procname = "sched_autogroup_enabled",
374 .data = &sysctl_sched_autogroup_enabled,
375 .maxlen = sizeof(unsigned int),
376 .mode = 0644,
377 .proc_handler = proc_dointvec_minmax,
378 .extra1 = &zero,
379 .extra2 = &one,
380 },
381#endif
382#ifdef CONFIG_PROVE_LOCKING
383 {
384 .procname = "prove_locking",
385 .data = &prove_locking,
386 .maxlen = sizeof(int),
387 .mode = 0644,
388 .proc_handler = proc_dointvec,
389 },
390#endif
391#ifdef CONFIG_LOCK_STAT
392 {
393 .procname = "lock_stat",
394 .data = &lock_stat,
395 .maxlen = sizeof(int),
396 .mode = 0644,
397 .proc_handler = proc_dointvec,
398 },
399#endif
400 {
401 .procname = "panic",
402 .data = &panic_timeout,
403 .maxlen = sizeof(int),
404 .mode = 0644,
405 .proc_handler = proc_dointvec,
406 },
407 {
408 .procname = "core_uses_pid",
409 .data = &core_uses_pid,
410 .maxlen = sizeof(int),
411 .mode = 0644,
412 .proc_handler = proc_dointvec,
413 },
414 {
415 .procname = "core_pattern",
416 .data = core_pattern,
417 .maxlen = CORENAME_MAX_SIZE,
418 .mode = 0644,
419 .proc_handler = proc_dostring,
420 },
421 {
422 .procname = "core_pipe_limit",
423 .data = &core_pipe_limit,
424 .maxlen = sizeof(unsigned int),
425 .mode = 0644,
426 .proc_handler = proc_dointvec,
427 },
428#ifdef CONFIG_PROC_SYSCTL
429 {
430 .procname = "tainted",
431 .maxlen = sizeof(long),
432 .mode = 0644,
433 .proc_handler = proc_taint,
434 },
435#endif
436#ifdef CONFIG_LATENCYTOP
437 {
438 .procname = "latencytop",
439 .data = &latencytop_enabled,
440 .maxlen = sizeof(int),
441 .mode = 0644,
442 .proc_handler = proc_dointvec,
443 },
444#endif
445#ifdef CONFIG_BLK_DEV_INITRD
446 {
447 .procname = "real-root-dev",
448 .data = &real_root_dev,
449 .maxlen = sizeof(int),
450 .mode = 0644,
451 .proc_handler = proc_dointvec,
452 },
453#endif
454 {
455 .procname = "print-fatal-signals",
456 .data = &print_fatal_signals,
457 .maxlen = sizeof(int),
458 .mode = 0644,
459 .proc_handler = proc_dointvec,
460 },
461#ifdef CONFIG_SPARC
462 {
463 .procname = "reboot-cmd",
464 .data = reboot_command,
465 .maxlen = 256,
466 .mode = 0644,
467 .proc_handler = proc_dostring,
468 },
469 {
470 .procname = "stop-a",
471 .data = &stop_a_enabled,
472 .maxlen = sizeof (int),
473 .mode = 0644,
474 .proc_handler = proc_dointvec,
475 },
476 {
477 .procname = "scons-poweroff",
478 .data = &scons_pwroff,
479 .maxlen = sizeof (int),
480 .mode = 0644,
481 .proc_handler = proc_dointvec,
482 },
483#endif
484#ifdef CONFIG_SPARC64
485 {
486 .procname = "tsb-ratio",
487 .data = &sysctl_tsb_ratio,
488 .maxlen = sizeof (int),
489 .mode = 0644,
490 .proc_handler = proc_dointvec,
491 },
492#endif
493#ifdef __hppa__
494 {
495 .procname = "soft-power",
496 .data = &pwrsw_enabled,
497 .maxlen = sizeof (int),
498 .mode = 0644,
499 .proc_handler = proc_dointvec,
500 },
501 {
502 .procname = "unaligned-trap",
503 .data = &unaligned_enabled,
504 .maxlen = sizeof (int),
505 .mode = 0644,
506 .proc_handler = proc_dointvec,
507 },
508#endif
509 {
510 .procname = "ctrl-alt-del",
511 .data = &C_A_D,
512 .maxlen = sizeof(int),
513 .mode = 0644,
514 .proc_handler = proc_dointvec,
515 },
516#ifdef CONFIG_FUNCTION_TRACER
517 {
518 .procname = "ftrace_enabled",
519 .data = &ftrace_enabled,
520 .maxlen = sizeof(int),
521 .mode = 0644,
522 .proc_handler = ftrace_enable_sysctl,
523 },
524#endif
525#ifdef CONFIG_STACK_TRACER
526 {
527 .procname = "stack_tracer_enabled",
528 .data = &stack_tracer_enabled,
529 .maxlen = sizeof(int),
530 .mode = 0644,
531 .proc_handler = stack_trace_sysctl,
532 },
533#endif
534#ifdef CONFIG_TRACING
535 {
536 .procname = "ftrace_dump_on_oops",
537 .data = &ftrace_dump_on_oops,
538 .maxlen = sizeof(int),
539 .mode = 0644,
540 .proc_handler = proc_dointvec,
541 },
542#endif
543#ifdef CONFIG_MODULES
544 {
545 .procname = "modprobe",
546 .data = &modprobe_path,
547 .maxlen = KMOD_PATH_LEN,
548 .mode = 0644,
549 .proc_handler = proc_dostring,
550 },
551 {
552 .procname = "modules_disabled",
553 .data = &modules_disabled,
554 .maxlen = sizeof(int),
555 .mode = 0644,
556 /* only handle a transition from default "0" to "1" */
557 .proc_handler = proc_dointvec_minmax,
558 .extra1 = &one,
559 .extra2 = &one,
560 },
561#endif
562#ifdef CONFIG_HOTPLUG
563 {
564 .procname = "hotplug",
565 .data = &uevent_helper,
566 .maxlen = UEVENT_HELPER_PATH_LEN,
567 .mode = 0644,
568 .proc_handler = proc_dostring,
569 },
570#endif
571#ifdef CONFIG_CHR_DEV_SG
572 {
573 .procname = "sg-big-buff",
574 .data = &sg_big_buff,
575 .maxlen = sizeof (int),
576 .mode = 0444,
577 .proc_handler = proc_dointvec,
578 },
579#endif
580#ifdef CONFIG_BSD_PROCESS_ACCT
581 {
582 .procname = "acct",
583 .data = &acct_parm,
584 .maxlen = 3*sizeof(int),
585 .mode = 0644,
586 .proc_handler = proc_dointvec,
587 },
588#endif
589#ifdef CONFIG_MAGIC_SYSRQ
590 {
591 .procname = "sysrq",
592 .data = &__sysrq_enabled,
593 .maxlen = sizeof (int),
594 .mode = 0644,
595 .proc_handler = sysrq_sysctl_handler,
596 },
597#endif
598#ifdef CONFIG_PROC_SYSCTL
599 {
600 .procname = "cad_pid",
601 .data = NULL,
602 .maxlen = sizeof (int),
603 .mode = 0600,
604 .proc_handler = proc_do_cad_pid,
605 },
606#endif
607 {
608 .procname = "threads-max",
609 .data = &max_threads,
610 .maxlen = sizeof(int),
611 .mode = 0644,
612 .proc_handler = proc_dointvec,
613 },
614 {
615 .procname = "random",
616 .mode = 0555,
617 .child = random_table,
618 },
619 {
620 .procname = "usermodehelper",
621 .mode = 0555,
622 .child = usermodehelper_table,
623 },
624 {
625 .procname = "overflowuid",
626 .data = &overflowuid,
627 .maxlen = sizeof(int),
628 .mode = 0644,
629 .proc_handler = proc_dointvec_minmax,
630 .extra1 = &minolduid,
631 .extra2 = &maxolduid,
632 },
633 {
634 .procname = "overflowgid",
635 .data = &overflowgid,
636 .maxlen = sizeof(int),
637 .mode = 0644,
638 .proc_handler = proc_dointvec_minmax,
639 .extra1 = &minolduid,
640 .extra2 = &maxolduid,
641 },
642#ifdef CONFIG_S390
643#ifdef CONFIG_MATHEMU
644 {
645 .procname = "ieee_emulation_warnings",
646 .data = &sysctl_ieee_emulation_warnings,
647 .maxlen = sizeof(int),
648 .mode = 0644,
649 .proc_handler = proc_dointvec,
650 },
651#endif
652 {
653 .procname = "userprocess_debug",
654 .data = &show_unhandled_signals,
655 .maxlen = sizeof(int),
656 .mode = 0644,
657 .proc_handler = proc_dointvec,
658 },
659#endif
660 {
661 .procname = "pid_max",
662 .data = &pid_max,
663 .maxlen = sizeof (int),
664 .mode = 0644,
665 .proc_handler = proc_dointvec_minmax,
666 .extra1 = &pid_max_min,
667 .extra2 = &pid_max_max,
668 },
669 {
670 .procname = "panic_on_oops",
671 .data = &panic_on_oops,
672 .maxlen = sizeof(int),
673 .mode = 0644,
674 .proc_handler = proc_dointvec,
675 },
676#if defined CONFIG_PRINTK
677 {
678 .procname = "printk",
679 .data = &console_loglevel,
680 .maxlen = 4*sizeof(int),
681 .mode = 0644,
682 .proc_handler = proc_dointvec,
683 },
684 {
685 .procname = "printk_ratelimit",
686 .data = &printk_ratelimit_state.interval,
687 .maxlen = sizeof(int),
688 .mode = 0644,
689 .proc_handler = proc_dointvec_jiffies,
690 },
691 {
692 .procname = "printk_ratelimit_burst",
693 .data = &printk_ratelimit_state.burst,
694 .maxlen = sizeof(int),
695 .mode = 0644,
696 .proc_handler = proc_dointvec,
697 },
698 {
699 .procname = "printk_delay",
700 .data = &printk_delay_msec,
701 .maxlen = sizeof(int),
702 .mode = 0644,
703 .proc_handler = proc_dointvec_minmax,
704 .extra1 = &zero,
705 .extra2 = &ten_thousand,
706 },
707 {
708 .procname = "dmesg_restrict",
709 .data = &dmesg_restrict,
710 .maxlen = sizeof(int),
711 .mode = 0644,
712 .proc_handler = proc_dointvec_minmax,
713 .extra1 = &zero,
714 .extra2 = &one,
715 },
716 {
717 .procname = "kptr_restrict",
718 .data = &kptr_restrict,
719 .maxlen = sizeof(int),
720 .mode = 0644,
721 .proc_handler = proc_dmesg_restrict,
722 .extra1 = &zero,
723 .extra2 = &two,
724 },
725#endif
726 {
727 .procname = "ngroups_max",
728 .data = &ngroups_max,
729 .maxlen = sizeof (int),
730 .mode = 0444,
731 .proc_handler = proc_dointvec,
732 },
733#if defined(CONFIG_LOCKUP_DETECTOR)
734 {
735 .procname = "watchdog",
736 .data = &watchdog_enabled,
737 .maxlen = sizeof (int),
738 .mode = 0644,
739 .proc_handler = proc_dowatchdog,
740 .extra1 = &zero,
741 .extra2 = &one,
742 },
743 {
744 .procname = "watchdog_thresh",
745 .data = &watchdog_thresh,
746 .maxlen = sizeof(int),
747 .mode = 0644,
748 .proc_handler = proc_dowatchdog,
749 .extra1 = &neg_one,
750 .extra2 = &sixty,
751 },
752 {
753 .procname = "softlockup_panic",
754 .data = &softlockup_panic,
755 .maxlen = sizeof(int),
756 .mode = 0644,
757 .proc_handler = proc_dointvec_minmax,
758 .extra1 = &zero,
759 .extra2 = &one,
760 },
761 {
762 .procname = "nmi_watchdog",
763 .data = &watchdog_enabled,
764 .maxlen = sizeof (int),
765 .mode = 0644,
766 .proc_handler = proc_dowatchdog,
767 .extra1 = &zero,
768 .extra2 = &one,
769 },
770#endif
771#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
772 {
773 .procname = "unknown_nmi_panic",
774 .data = &unknown_nmi_panic,
775 .maxlen = sizeof (int),
776 .mode = 0644,
777 .proc_handler = proc_dointvec,
778 },
779#endif
780#if defined(CONFIG_X86)
781 {
782 .procname = "panic_on_unrecovered_nmi",
783 .data = &panic_on_unrecovered_nmi,
784 .maxlen = sizeof(int),
785 .mode = 0644,
786 .proc_handler = proc_dointvec,
787 },
788 {
789 .procname = "panic_on_io_nmi",
790 .data = &panic_on_io_nmi,
791 .maxlen = sizeof(int),
792 .mode = 0644,
793 .proc_handler = proc_dointvec,
794 },
795 {
796 .procname = "bootloader_type",
797 .data = &bootloader_type,
798 .maxlen = sizeof (int),
799 .mode = 0444,
800 .proc_handler = proc_dointvec,
801 },
802 {
803 .procname = "bootloader_version",
804 .data = &bootloader_version,
805 .maxlen = sizeof (int),
806 .mode = 0444,
807 .proc_handler = proc_dointvec,
808 },
809 {
810 .procname = "kstack_depth_to_print",
811 .data = &kstack_depth_to_print,
812 .maxlen = sizeof(int),
813 .mode = 0644,
814 .proc_handler = proc_dointvec,
815 },
816 {
817 .procname = "io_delay_type",
818 .data = &io_delay_type,
819 .maxlen = sizeof(int),
820 .mode = 0644,
821 .proc_handler = proc_dointvec,
822 },
823#endif
824#if defined(CONFIG_MMU)
825 {
826 .procname = "randomize_va_space",
827 .data = &randomize_va_space,
828 .maxlen = sizeof(int),
829 .mode = 0644,
830 .proc_handler = proc_dointvec,
831 },
832#endif
833#if defined(CONFIG_S390) && defined(CONFIG_SMP)
834 {
835 .procname = "spin_retry",
836 .data = &spin_retry,
837 .maxlen = sizeof (int),
838 .mode = 0644,
839 .proc_handler = proc_dointvec,
840 },
841#endif
842#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
843 {
844 .procname = "acpi_video_flags",
845 .data = &acpi_realmode_flags,
846 .maxlen = sizeof (unsigned long),
847 .mode = 0644,
848 .proc_handler = proc_doulongvec_minmax,
849 },
850#endif
851#ifdef CONFIG_IA64
852 {
853 .procname = "ignore-unaligned-usertrap",
854 .data = &no_unaligned_warning,
855 .maxlen = sizeof (int),
856 .mode = 0644,
857 .proc_handler = proc_dointvec,
858 },
859 {
860 .procname = "unaligned-dump-stack",
861 .data = &unaligned_dump_stack,
862 .maxlen = sizeof (int),
863 .mode = 0644,
864 .proc_handler = proc_dointvec,
865 },
866#endif
867#ifdef CONFIG_DETECT_HUNG_TASK
868 {
869 .procname = "hung_task_panic",
870 .data = &sysctl_hung_task_panic,
871 .maxlen = sizeof(int),
872 .mode = 0644,
873 .proc_handler = proc_dointvec_minmax,
874 .extra1 = &zero,
875 .extra2 = &one,
876 },
877 {
878 .procname = "hung_task_check_count",
879 .data = &sysctl_hung_task_check_count,
880 .maxlen = sizeof(unsigned long),
881 .mode = 0644,
882 .proc_handler = proc_doulongvec_minmax,
883 },
884 {
885 .procname = "hung_task_timeout_secs",
886 .data = &sysctl_hung_task_timeout_secs,
887 .maxlen = sizeof(unsigned long),
888 .mode = 0644,
889 .proc_handler = proc_dohung_task_timeout_secs,
890 },
891 {
892 .procname = "hung_task_warnings",
893 .data = &sysctl_hung_task_warnings,
894 .maxlen = sizeof(unsigned long),
895 .mode = 0644,
896 .proc_handler = proc_doulongvec_minmax,
897 },
898#endif
899#ifdef CONFIG_COMPAT
900 {
901 .procname = "compat-log",
902 .data = &compat_log,
903 .maxlen = sizeof (int),
904 .mode = 0644,
905 .proc_handler = proc_dointvec,
906 },
907#endif
908#ifdef CONFIG_RT_MUTEXES
909 {
910 .procname = "max_lock_depth",
911 .data = &max_lock_depth,
912 .maxlen = sizeof(int),
913 .mode = 0644,
914 .proc_handler = proc_dointvec,
915 },
916#endif
917 {
918 .procname = "poweroff_cmd",
919 .data = &poweroff_cmd,
920 .maxlen = POWEROFF_CMD_PATH_LEN,
921 .mode = 0644,
922 .proc_handler = proc_dostring,
923 },
924#ifdef CONFIG_KEYS
925 {
926 .procname = "keys",
927 .mode = 0555,
928 .child = key_sysctls,
929 },
930#endif
931#ifdef CONFIG_RCU_TORTURE_TEST
932 {
933 .procname = "rcutorture_runnable",
934 .data = &rcutorture_runnable,
935 .maxlen = sizeof(int),
936 .mode = 0644,
937 .proc_handler = proc_dointvec,
938 },
939#endif
940#ifdef CONFIG_PERF_EVENTS
941 /*
942 * User-space scripts rely on the existence of this file
943 * as a feature check for perf_events being enabled.
944 *
945 * So it's an ABI, do not remove!
946 */
947 {
948 .procname = "perf_event_paranoid",
949 .data = &sysctl_perf_event_paranoid,
950 .maxlen = sizeof(sysctl_perf_event_paranoid),
951 .mode = 0644,
952 .proc_handler = proc_dointvec,
953 },
954 {
955 .procname = "perf_event_mlock_kb",
956 .data = &sysctl_perf_event_mlock,
957 .maxlen = sizeof(sysctl_perf_event_mlock),
958 .mode = 0644,
959 .proc_handler = proc_dointvec,
960 },
961 {
962 .procname = "perf_event_max_sample_rate",
963 .data = &sysctl_perf_event_sample_rate,
964 .maxlen = sizeof(sysctl_perf_event_sample_rate),
965 .mode = 0644,
966 .proc_handler = perf_proc_update_handler,
967 },
968#endif
969#ifdef CONFIG_KMEMCHECK
970 {
971 .procname = "kmemcheck",
972 .data = &kmemcheck_enabled,
973 .maxlen = sizeof(int),
974 .mode = 0644,
975 .proc_handler = proc_dointvec,
976 },
977#endif
978#ifdef CONFIG_BLOCK
979 {
980 .procname = "blk_iopoll",
981 .data = &blk_iopoll_enabled,
982 .maxlen = sizeof(int),
983 .mode = 0644,
984 .proc_handler = proc_dointvec,
985 },
986#endif
987 { }
988};
989
990static struct ctl_table vm_table[] = {
991 {
992 .procname = "overcommit_memory",
993 .data = &sysctl_overcommit_memory,
994 .maxlen = sizeof(sysctl_overcommit_memory),
995 .mode = 0644,
996 .proc_handler = proc_dointvec_minmax,
997 .extra1 = &zero,
998 .extra2 = &two,
999 },
1000 {
1001 .procname = "panic_on_oom",
1002 .data = &sysctl_panic_on_oom,
1003 .maxlen = sizeof(sysctl_panic_on_oom),
1004 .mode = 0644,
1005 .proc_handler = proc_dointvec_minmax,
1006 .extra1 = &zero,
1007 .extra2 = &two,
1008 },
1009 {
1010 .procname = "oom_kill_allocating_task",
1011 .data = &sysctl_oom_kill_allocating_task,
1012 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1013 .mode = 0644,
1014 .proc_handler = proc_dointvec,
1015 },
1016 {
1017 .procname = "oom_dump_tasks",
1018 .data = &sysctl_oom_dump_tasks,
1019 .maxlen = sizeof(sysctl_oom_dump_tasks),
1020 .mode = 0644,
1021 .proc_handler = proc_dointvec,
1022 },
1023 {
1024 .procname = "overcommit_ratio",
1025 .data = &sysctl_overcommit_ratio,
1026 .maxlen = sizeof(sysctl_overcommit_ratio),
1027 .mode = 0644,
1028 .proc_handler = proc_dointvec,
1029 },
1030 {
1031 .procname = "page-cluster",
1032 .data = &page_cluster,
1033 .maxlen = sizeof(int),
1034 .mode = 0644,
1035 .proc_handler = proc_dointvec_minmax,
1036 .extra1 = &zero,
1037 },
1038 {
1039 .procname = "dirty_background_ratio",
1040 .data = &dirty_background_ratio,
1041 .maxlen = sizeof(dirty_background_ratio),
1042 .mode = 0644,
1043 .proc_handler = dirty_background_ratio_handler,
1044 .extra1 = &zero,
1045 .extra2 = &one_hundred,
1046 },
1047 {
1048 .procname = "dirty_background_bytes",
1049 .data = &dirty_background_bytes,
1050 .maxlen = sizeof(dirty_background_bytes),
1051 .mode = 0644,
1052 .proc_handler = dirty_background_bytes_handler,
1053 .extra1 = &one_ul,
1054 },
1055 {
1056 .procname = "dirty_ratio",
1057 .data = &vm_dirty_ratio,
1058 .maxlen = sizeof(vm_dirty_ratio),
1059 .mode = 0644,
1060 .proc_handler = dirty_ratio_handler,
1061 .extra1 = &zero,
1062 .extra2 = &one_hundred,
1063 },
1064 {
1065 .procname = "dirty_bytes",
1066 .data = &vm_dirty_bytes,
1067 .maxlen = sizeof(vm_dirty_bytes),
1068 .mode = 0644,
1069 .proc_handler = dirty_bytes_handler,
1070 .extra1 = &dirty_bytes_min,
1071 },
1072 {
1073 .procname = "dirty_writeback_centisecs",
1074 .data = &dirty_writeback_interval,
1075 .maxlen = sizeof(dirty_writeback_interval),
1076 .mode = 0644,
1077 .proc_handler = dirty_writeback_centisecs_handler,
1078 },
1079 {
1080 .procname = "dirty_expire_centisecs",
1081 .data = &dirty_expire_interval,
1082 .maxlen = sizeof(dirty_expire_interval),
1083 .mode = 0644,
1084 .proc_handler = proc_dointvec_minmax,
1085 .extra1 = &zero,
1086 },
1087 {
1088 .procname = "nr_pdflush_threads",
1089 .data = &nr_pdflush_threads,
1090 .maxlen = sizeof nr_pdflush_threads,
1091 .mode = 0444 /* read-only*/,
1092 .proc_handler = proc_dointvec,
1093 },
1094 {
1095 .procname = "swappiness",
1096 .data = &vm_swappiness,
1097 .maxlen = sizeof(vm_swappiness),
1098 .mode = 0644,
1099 .proc_handler = proc_dointvec_minmax,
1100 .extra1 = &zero,
1101 .extra2 = &one_hundred,
1102 },
1103#ifdef CONFIG_HUGETLB_PAGE
1104 {
1105 .procname = "nr_hugepages",
1106 .data = NULL,
1107 .maxlen = sizeof(unsigned long),
1108 .mode = 0644,
1109 .proc_handler = hugetlb_sysctl_handler,
1110 .extra1 = (void *)&hugetlb_zero,
1111 .extra2 = (void *)&hugetlb_infinity,
1112 },
1113#ifdef CONFIG_NUMA
1114 {
1115 .procname = "nr_hugepages_mempolicy",
1116 .data = NULL,
1117 .maxlen = sizeof(unsigned long),
1118 .mode = 0644,
1119 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1120 .extra1 = (void *)&hugetlb_zero,
1121 .extra2 = (void *)&hugetlb_infinity,
1122 },
1123#endif
1124 {
1125 .procname = "hugetlb_shm_group",
1126 .data = &sysctl_hugetlb_shm_group,
1127 .maxlen = sizeof(gid_t),
1128 .mode = 0644,
1129 .proc_handler = proc_dointvec,
1130 },
1131 {
1132 .procname = "hugepages_treat_as_movable",
1133 .data = &hugepages_treat_as_movable,
1134 .maxlen = sizeof(int),
1135 .mode = 0644,
1136 .proc_handler = hugetlb_treat_movable_handler,
1137 },
1138 {
1139 .procname = "nr_overcommit_hugepages",
1140 .data = NULL,
1141 .maxlen = sizeof(unsigned long),
1142 .mode = 0644,
1143 .proc_handler = hugetlb_overcommit_handler,
1144 .extra1 = (void *)&hugetlb_zero,
1145 .extra2 = (void *)&hugetlb_infinity,
1146 },
1147#endif
1148 {
1149 .procname = "lowmem_reserve_ratio",
1150 .data = &sysctl_lowmem_reserve_ratio,
1151 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1152 .mode = 0644,
1153 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1154 },
1155 {
1156 .procname = "drop_caches",
1157 .data = &sysctl_drop_caches,
1158 .maxlen = sizeof(int),
1159 .mode = 0644,
1160 .proc_handler = drop_caches_sysctl_handler,
1161 .extra1 = &one,
1162 .extra2 = &three,
1163 },
1164#ifdef CONFIG_COMPACTION
1165 {
1166 .procname = "compact_memory",
1167 .data = &sysctl_compact_memory,
1168 .maxlen = sizeof(int),
1169 .mode = 0200,
1170 .proc_handler = sysctl_compaction_handler,
1171 },
1172 {
1173 .procname = "extfrag_threshold",
1174 .data = &sysctl_extfrag_threshold,
1175 .maxlen = sizeof(int),
1176 .mode = 0644,
1177 .proc_handler = sysctl_extfrag_handler,
1178 .extra1 = &min_extfrag_threshold,
1179 .extra2 = &max_extfrag_threshold,
1180 },
1181
1182#endif /* CONFIG_COMPACTION */
1183 {
1184 .procname = "min_free_kbytes",
1185 .data = &min_free_kbytes,
1186 .maxlen = sizeof(min_free_kbytes),
1187 .mode = 0644,
1188 .proc_handler = min_free_kbytes_sysctl_handler,
1189 .extra1 = &zero,
1190 },
1191 {
1192 .procname = "percpu_pagelist_fraction",
1193 .data = &percpu_pagelist_fraction,
1194 .maxlen = sizeof(percpu_pagelist_fraction),
1195 .mode = 0644,
1196 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1197 .extra1 = &min_percpu_pagelist_fract,
1198 },
1199#ifdef CONFIG_MMU
1200 {
1201 .procname = "max_map_count",
1202 .data = &sysctl_max_map_count,
1203 .maxlen = sizeof(sysctl_max_map_count),
1204 .mode = 0644,
1205 .proc_handler = proc_dointvec_minmax,
1206 .extra1 = &zero,
1207 },
1208#else
1209 {
1210 .procname = "nr_trim_pages",
1211 .data = &sysctl_nr_trim_pages,
1212 .maxlen = sizeof(sysctl_nr_trim_pages),
1213 .mode = 0644,
1214 .proc_handler = proc_dointvec_minmax,
1215 .extra1 = &zero,
1216 },
1217#endif
1218 {
1219 .procname = "laptop_mode",
1220 .data = &laptop_mode,
1221 .maxlen = sizeof(laptop_mode),
1222 .mode = 0644,
1223 .proc_handler = proc_dointvec_jiffies,
1224 },
1225 {
1226 .procname = "block_dump",
1227 .data = &block_dump,
1228 .maxlen = sizeof(block_dump),
1229 .mode = 0644,
1230 .proc_handler = proc_dointvec,
1231 .extra1 = &zero,
1232 },
1233 {
1234 .procname = "vfs_cache_pressure",
1235 .data = &sysctl_vfs_cache_pressure,
1236 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1237 .mode = 0644,
1238 .proc_handler = proc_dointvec,
1239 .extra1 = &zero,
1240 },
1241#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1242 {
1243 .procname = "legacy_va_layout",
1244 .data = &sysctl_legacy_va_layout,
1245 .maxlen = sizeof(sysctl_legacy_va_layout),
1246 .mode = 0644,
1247 .proc_handler = proc_dointvec,
1248 .extra1 = &zero,
1249 },
1250#endif
1251#ifdef CONFIG_NUMA
1252 {
1253 .procname = "zone_reclaim_mode",
1254 .data = &zone_reclaim_mode,
1255 .maxlen = sizeof(zone_reclaim_mode),
1256 .mode = 0644,
1257 .proc_handler = proc_dointvec,
1258 .extra1 = &zero,
1259 },
1260 {
1261 .procname = "min_unmapped_ratio",
1262 .data = &sysctl_min_unmapped_ratio,
1263 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1264 .mode = 0644,
1265 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1266 .extra1 = &zero,
1267 .extra2 = &one_hundred,
1268 },
1269 {
1270 .procname = "min_slab_ratio",
1271 .data = &sysctl_min_slab_ratio,
1272 .maxlen = sizeof(sysctl_min_slab_ratio),
1273 .mode = 0644,
1274 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1275 .extra1 = &zero,
1276 .extra2 = &one_hundred,
1277 },
1278#endif
1279#ifdef CONFIG_SMP
1280 {
1281 .procname = "stat_interval",
1282 .data = &sysctl_stat_interval,
1283 .maxlen = sizeof(sysctl_stat_interval),
1284 .mode = 0644,
1285 .proc_handler = proc_dointvec_jiffies,
1286 },
1287#endif
1288#ifdef CONFIG_MMU
1289 {
1290 .procname = "mmap_min_addr",
1291 .data = &dac_mmap_min_addr,
1292 .maxlen = sizeof(unsigned long),
1293 .mode = 0644,
1294 .proc_handler = mmap_min_addr_handler,
1295 },
1296#endif
1297#ifdef CONFIG_NUMA
1298 {
1299 .procname = "numa_zonelist_order",
1300 .data = &numa_zonelist_order,
1301 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1302 .mode = 0644,
1303 .proc_handler = numa_zonelist_order_handler,
1304 },
1305#endif
1306#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1307 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1308 {
1309 .procname = "vdso_enabled",
1310 .data = &vdso_enabled,
1311 .maxlen = sizeof(vdso_enabled),
1312 .mode = 0644,
1313 .proc_handler = proc_dointvec,
1314 .extra1 = &zero,
1315 },
1316#endif
1317#ifdef CONFIG_HIGHMEM
1318 {
1319 .procname = "highmem_is_dirtyable",
1320 .data = &vm_highmem_is_dirtyable,
1321 .maxlen = sizeof(vm_highmem_is_dirtyable),
1322 .mode = 0644,
1323 .proc_handler = proc_dointvec_minmax,
1324 .extra1 = &zero,
1325 .extra2 = &one,
1326 },
1327#endif
1328 {
1329 .procname = "scan_unevictable_pages",
1330 .data = &scan_unevictable_pages,
1331 .maxlen = sizeof(scan_unevictable_pages),
1332 .mode = 0644,
1333 .proc_handler = scan_unevictable_handler,
1334 },
1335#ifdef CONFIG_MEMORY_FAILURE
1336 {
1337 .procname = "memory_failure_early_kill",
1338 .data = &sysctl_memory_failure_early_kill,
1339 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1340 .mode = 0644,
1341 .proc_handler = proc_dointvec_minmax,
1342 .extra1 = &zero,
1343 .extra2 = &one,
1344 },
1345 {
1346 .procname = "memory_failure_recovery",
1347 .data = &sysctl_memory_failure_recovery,
1348 .maxlen = sizeof(sysctl_memory_failure_recovery),
1349 .mode = 0644,
1350 .proc_handler = proc_dointvec_minmax,
1351 .extra1 = &zero,
1352 .extra2 = &one,
1353 },
1354#endif
1355 { }
1356};
1357
1358#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1359static struct ctl_table binfmt_misc_table[] = {
1360 { }
1361};
1362#endif
1363
1364static struct ctl_table fs_table[] = {
1365 {
1366 .procname = "inode-nr",
1367 .data = &inodes_stat,
1368 .maxlen = 2*sizeof(int),
1369 .mode = 0444,
1370 .proc_handler = proc_nr_inodes,
1371 },
1372 {
1373 .procname = "inode-state",
1374 .data = &inodes_stat,
1375 .maxlen = 7*sizeof(int),
1376 .mode = 0444,
1377 .proc_handler = proc_nr_inodes,
1378 },
1379 {
1380 .procname = "file-nr",
1381 .data = &files_stat,
1382 .maxlen = sizeof(files_stat),
1383 .mode = 0444,
1384 .proc_handler = proc_nr_files,
1385 },
1386 {
1387 .procname = "file-max",
1388 .data = &files_stat.max_files,
1389 .maxlen = sizeof(files_stat.max_files),
1390 .mode = 0644,
1391 .proc_handler = proc_doulongvec_minmax,
1392 },
1393 {
1394 .procname = "nr_open",
1395 .data = &sysctl_nr_open,
1396 .maxlen = sizeof(int),
1397 .mode = 0644,
1398 .proc_handler = proc_dointvec_minmax,
1399 .extra1 = &sysctl_nr_open_min,
1400 .extra2 = &sysctl_nr_open_max,
1401 },
1402 {
1403 .procname = "dentry-state",
1404 .data = &dentry_stat,
1405 .maxlen = 6*sizeof(int),
1406 .mode = 0444,
1407 .proc_handler = proc_nr_dentry,
1408 },
1409 {
1410 .procname = "overflowuid",
1411 .data = &fs_overflowuid,
1412 .maxlen = sizeof(int),
1413 .mode = 0644,
1414 .proc_handler = proc_dointvec_minmax,
1415 .extra1 = &minolduid,
1416 .extra2 = &maxolduid,
1417 },
1418 {
1419 .procname = "overflowgid",
1420 .data = &fs_overflowgid,
1421 .maxlen = sizeof(int),
1422 .mode = 0644,
1423 .proc_handler = proc_dointvec_minmax,
1424 .extra1 = &minolduid,
1425 .extra2 = &maxolduid,
1426 },
1427#ifdef CONFIG_FILE_LOCKING
1428 {
1429 .procname = "leases-enable",
1430 .data = &leases_enable,
1431 .maxlen = sizeof(int),
1432 .mode = 0644,
1433 .proc_handler = proc_dointvec,
1434 },
1435#endif
1436#ifdef CONFIG_DNOTIFY
1437 {
1438 .procname = "dir-notify-enable",
1439 .data = &dir_notify_enable,
1440 .maxlen = sizeof(int),
1441 .mode = 0644,
1442 .proc_handler = proc_dointvec,
1443 },
1444#endif
1445#ifdef CONFIG_MMU
1446#ifdef CONFIG_FILE_LOCKING
1447 {
1448 .procname = "lease-break-time",
1449 .data = &lease_break_time,
1450 .maxlen = sizeof(int),
1451 .mode = 0644,
1452 .proc_handler = proc_dointvec,
1453 },
1454#endif
1455#ifdef CONFIG_AIO
1456 {
1457 .procname = "aio-nr",
1458 .data = &aio_nr,
1459 .maxlen = sizeof(aio_nr),
1460 .mode = 0444,
1461 .proc_handler = proc_doulongvec_minmax,
1462 },
1463 {
1464 .procname = "aio-max-nr",
1465 .data = &aio_max_nr,
1466 .maxlen = sizeof(aio_max_nr),
1467 .mode = 0644,
1468 .proc_handler = proc_doulongvec_minmax,
1469 },
1470#endif /* CONFIG_AIO */
1471#ifdef CONFIG_INOTIFY_USER
1472 {
1473 .procname = "inotify",
1474 .mode = 0555,
1475 .child = inotify_table,
1476 },
1477#endif
1478#ifdef CONFIG_EPOLL
1479 {
1480 .procname = "epoll",
1481 .mode = 0555,
1482 .child = epoll_table,
1483 },
1484#endif
1485#endif
1486 {
1487 .procname = "suid_dumpable",
1488 .data = &suid_dumpable,
1489 .maxlen = sizeof(int),
1490 .mode = 0644,
1491 .proc_handler = proc_dointvec_minmax,
1492 .extra1 = &zero,
1493 .extra2 = &two,
1494 },
1495#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1496 {
1497 .procname = "binfmt_misc",
1498 .mode = 0555,
1499 .child = binfmt_misc_table,
1500 },
1501#endif
1502 {
1503 .procname = "pipe-max-size",
1504 .data = &pipe_max_size,
1505 .maxlen = sizeof(int),
1506 .mode = 0644,
1507 .proc_handler = &pipe_proc_fn,
1508 .extra1 = &pipe_min_size,
1509 },
1510 { }
1511};
1512
1513static struct ctl_table debug_table[] = {
1514#if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
1515 defined(CONFIG_S390) || defined(CONFIG_TILE)
1516 {
1517 .procname = "exception-trace",
1518 .data = &show_unhandled_signals,
1519 .maxlen = sizeof(int),
1520 .mode = 0644,
1521 .proc_handler = proc_dointvec
1522 },
1523#endif
1524#if defined(CONFIG_OPTPROBES)
1525 {
1526 .procname = "kprobes-optimization",
1527 .data = &sysctl_kprobes_optimization,
1528 .maxlen = sizeof(int),
1529 .mode = 0644,
1530 .proc_handler = proc_kprobes_optimization_handler,
1531 .extra1 = &zero,
1532 .extra2 = &one,
1533 },
1534#endif
1535 { }
1536};
1537
1538static struct ctl_table dev_table[] = {
1539 { }
1540};
1541
1542static DEFINE_SPINLOCK(sysctl_lock);
1543
1544/* called under sysctl_lock */
1545static int use_table(struct ctl_table_header *p)
1546{
1547 if (unlikely(p->unregistering))
1548 return 0;
1549 p->used++;
1550 return 1;
1551}
1552
1553/* called under sysctl_lock */
1554static void unuse_table(struct ctl_table_header *p)
1555{
1556 if (!--p->used)
1557 if (unlikely(p->unregistering))
1558 complete(p->unregistering);
1559}
1560
1561/* called under sysctl_lock, will reacquire if has to wait */
1562static void start_unregistering(struct ctl_table_header *p)
1563{
1564 /*
1565 * if p->used is 0, nobody will ever touch that entry again;
1566 * we'll eliminate all paths to it before dropping sysctl_lock
1567 */
1568 if (unlikely(p->used)) {
1569 struct completion wait;
1570 init_completion(&wait);
1571 p->unregistering = &wait;
1572 spin_unlock(&sysctl_lock);
1573 wait_for_completion(&wait);
1574 spin_lock(&sysctl_lock);
1575 } else {
1576 /* anything non-NULL; we'll never dereference it */
1577 p->unregistering = ERR_PTR(-EINVAL);
1578 }
1579 /*
1580 * do not remove from the list until nobody holds it; walking the
1581 * list in do_sysctl() relies on that.
1582 */
1583 list_del_init(&p->ctl_entry);
1584}
1585
1586void sysctl_head_get(struct ctl_table_header *head)
1587{
1588 spin_lock(&sysctl_lock);
1589 head->count++;
1590 spin_unlock(&sysctl_lock);
1591}
1592
1593void sysctl_head_put(struct ctl_table_header *head)
1594{
1595 spin_lock(&sysctl_lock);
1596 if (!--head->count)
1597 kfree_rcu(head, rcu);
1598 spin_unlock(&sysctl_lock);
1599}
1600
1601struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1602{
1603 if (!head)
1604 BUG();
1605 spin_lock(&sysctl_lock);
1606 if (!use_table(head))
1607 head = ERR_PTR(-ENOENT);
1608 spin_unlock(&sysctl_lock);
1609 return head;
1610}
1611
1612void sysctl_head_finish(struct ctl_table_header *head)
1613{
1614 if (!head)
1615 return;
1616 spin_lock(&sysctl_lock);
1617 unuse_table(head);
1618 spin_unlock(&sysctl_lock);
1619}
1620
1621static struct ctl_table_set *
1622lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1623{
1624 struct ctl_table_set *set = &root->default_set;
1625 if (root->lookup)
1626 set = root->lookup(root, namespaces);
1627 return set;
1628}
1629
1630static struct list_head *
1631lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1632{
1633 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1634 return &set->list;
1635}
1636
1637struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1638 struct ctl_table_header *prev)
1639{
1640 struct ctl_table_root *root;
1641 struct list_head *header_list;
1642 struct ctl_table_header *head;
1643 struct list_head *tmp;
1644
1645 spin_lock(&sysctl_lock);
1646 if (prev) {
1647 head = prev;
1648 tmp = &prev->ctl_entry;
1649 unuse_table(prev);
1650 goto next;
1651 }
1652 tmp = &root_table_header.ctl_entry;
1653 for (;;) {
1654 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1655
1656 if (!use_table(head))
1657 goto next;
1658 spin_unlock(&sysctl_lock);
1659 return head;
1660 next:
1661 root = head->root;
1662 tmp = tmp->next;
1663 header_list = lookup_header_list(root, namespaces);
1664 if (tmp != header_list)
1665 continue;
1666
1667 do {
1668 root = list_entry(root->root_list.next,
1669 struct ctl_table_root, root_list);
1670 if (root == &sysctl_table_root)
1671 goto out;
1672 header_list = lookup_header_list(root, namespaces);
1673 } while (list_empty(header_list));
1674 tmp = header_list->next;
1675 }
1676out:
1677 spin_unlock(&sysctl_lock);
1678 return NULL;
1679}
1680
1681struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1682{
1683 return __sysctl_head_next(current->nsproxy, prev);
1684}
1685
1686void register_sysctl_root(struct ctl_table_root *root)
1687{
1688 spin_lock(&sysctl_lock);
1689 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1690 spin_unlock(&sysctl_lock);
1691}
1692
1693/*
1694 * sysctl_perm does NOT grant the superuser all rights automatically, because
1695 * some sysctl variables are readonly even to root.
1696 */
1697
1698static int test_perm(int mode, int op)
1699{
1700 if (!current_euid())
1701 mode >>= 6;
1702 else if (in_egroup_p(0))
1703 mode >>= 3;
1704 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1705 return 0;
1706 return -EACCES;
1707}
1708
1709int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1710{
1711 int mode;
1712
1713 if (root->permissions)
1714 mode = root->permissions(root, current->nsproxy, table);
1715 else
1716 mode = table->mode;
1717
1718 return test_perm(mode, op);
1719}
1720
1721static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1722{
1723 for (; table->procname; table++) {
1724 table->parent = parent;
1725 if (table->child)
1726 sysctl_set_parent(table, table->child);
1727 }
1728}
1729
1730static __init int sysctl_init(void)
1731{
1732 sysctl_set_parent(NULL, root_table);
1733#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1734 sysctl_check_table(current->nsproxy, root_table);
1735#endif
1736 return 0;
1737}
1738
1739core_initcall(sysctl_init);
1740
1741static struct ctl_table *is_branch_in(struct ctl_table *branch,
1742 struct ctl_table *table)
1743{
1744 struct ctl_table *p;
1745 const char *s = branch->procname;
1746
1747 /* branch should have named subdirectory as its first element */
1748 if (!s || !branch->child)
1749 return NULL;
1750
1751 /* ... and nothing else */
1752 if (branch[1].procname)
1753 return NULL;
1754
1755 /* table should contain subdirectory with the same name */
1756 for (p = table; p->procname; p++) {
1757 if (!p->child)
1758 continue;
1759 if (p->procname && strcmp(p->procname, s) == 0)
1760 return p;
1761 }
1762 return NULL;
1763}
1764
1765/* see if attaching q to p would be an improvement */
1766static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1767{
1768 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1769 struct ctl_table *next;
1770 int is_better = 0;
1771 int not_in_parent = !p->attached_by;
1772
1773 while ((next = is_branch_in(by, to)) != NULL) {
1774 if (by == q->attached_by)
1775 is_better = 1;
1776 if (to == p->attached_by)
1777 not_in_parent = 1;
1778 by = by->child;
1779 to = next->child;
1780 }
1781
1782 if (is_better && not_in_parent) {
1783 q->attached_by = by;
1784 q->attached_to = to;
1785 q->parent = p;
1786 }
1787}
1788
1789/**
1790 * __register_sysctl_paths - register a sysctl hierarchy
1791 * @root: List of sysctl headers to register on
1792 * @namespaces: Data to compute which lists of sysctl entries are visible
1793 * @path: The path to the directory the sysctl table is in.
1794 * @table: the top-level table structure
1795 *
1796 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1797 * array. A completely 0 filled entry terminates the table.
1798 *
1799 * The members of the &struct ctl_table structure are used as follows:
1800 *
1801 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1802 * enter a sysctl file
1803 *
1804 * data - a pointer to data for use by proc_handler
1805 *
1806 * maxlen - the maximum size in bytes of the data
1807 *
1808 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1809 *
1810 * child - a pointer to the child sysctl table if this entry is a directory, or
1811 * %NULL.
1812 *
1813 * proc_handler - the text handler routine (described below)
1814 *
1815 * de - for internal use by the sysctl routines
1816 *
1817 * extra1, extra2 - extra pointers usable by the proc handler routines
1818 *
1819 * Leaf nodes in the sysctl tree will be represented by a single file
1820 * under /proc; non-leaf nodes will be represented by directories.
1821 *
1822 * sysctl(2) can automatically manage read and write requests through
1823 * the sysctl table. The data and maxlen fields of the ctl_table
1824 * struct enable minimal validation of the values being written to be
1825 * performed, and the mode field allows minimal authentication.
1826 *
1827 * There must be a proc_handler routine for any terminal nodes
1828 * mirrored under /proc/sys (non-terminals are handled by a built-in
1829 * directory handler). Several default handlers are available to
1830 * cover common cases -
1831 *
1832 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1833 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1834 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1835 *
1836 * It is the handler's job to read the input buffer from user memory
1837 * and process it. The handler should return 0 on success.
1838 *
1839 * This routine returns %NULL on a failure to register, and a pointer
1840 * to the table header on success.
1841 */
1842struct ctl_table_header *__register_sysctl_paths(
1843 struct ctl_table_root *root,
1844 struct nsproxy *namespaces,
1845 const struct ctl_path *path, struct ctl_table *table)
1846{
1847 struct ctl_table_header *header;
1848 struct ctl_table *new, **prevp;
1849 unsigned int n, npath;
1850 struct ctl_table_set *set;
1851
1852 /* Count the path components */
1853 for (npath = 0; path[npath].procname; ++npath)
1854 ;
1855
1856 /*
1857 * For each path component, allocate a 2-element ctl_table array.
1858 * The first array element will be filled with the sysctl entry
1859 * for this, the second will be the sentinel (procname == 0).
1860 *
1861 * We allocate everything in one go so that we don't have to
1862 * worry about freeing additional memory in unregister_sysctl_table.
1863 */
1864 header = kzalloc(sizeof(struct ctl_table_header) +
1865 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1866 if (!header)
1867 return NULL;
1868
1869 new = (struct ctl_table *) (header + 1);
1870
1871 /* Now connect the dots */
1872 prevp = &header->ctl_table;
1873 for (n = 0; n < npath; ++n, ++path) {
1874 /* Copy the procname */
1875 new->procname = path->procname;
1876 new->mode = 0555;
1877
1878 *prevp = new;
1879 prevp = &new->child;
1880
1881 new += 2;
1882 }
1883 *prevp = table;
1884 header->ctl_table_arg = table;
1885
1886 INIT_LIST_HEAD(&header->ctl_entry);
1887 header->used = 0;
1888 header->unregistering = NULL;
1889 header->root = root;
1890 sysctl_set_parent(NULL, header->ctl_table);
1891 header->count = 1;
1892#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1893 if (sysctl_check_table(namespaces, header->ctl_table)) {
1894 kfree(header);
1895 return NULL;
1896 }
1897#endif
1898 spin_lock(&sysctl_lock);
1899 header->set = lookup_header_set(root, namespaces);
1900 header->attached_by = header->ctl_table;
1901 header->attached_to = root_table;
1902 header->parent = &root_table_header;
1903 for (set = header->set; set; set = set->parent) {
1904 struct ctl_table_header *p;
1905 list_for_each_entry(p, &set->list, ctl_entry) {
1906 if (p->unregistering)
1907 continue;
1908 try_attach(p, header);
1909 }
1910 }
1911 header->parent->count++;
1912 list_add_tail(&header->ctl_entry, &header->set->list);
1913 spin_unlock(&sysctl_lock);
1914
1915 return header;
1916}
1917
1918/**
1919 * register_sysctl_table_path - register a sysctl table hierarchy
1920 * @path: The path to the directory the sysctl table is in.
1921 * @table: the top-level table structure
1922 *
1923 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1924 * array. A completely 0 filled entry terminates the table.
1925 *
1926 * See __register_sysctl_paths for more details.
1927 */
1928struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1929 struct ctl_table *table)
1930{
1931 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1932 path, table);
1933}
1934
1935/**
1936 * register_sysctl_table - register a sysctl table hierarchy
1937 * @table: the top-level table structure
1938 *
1939 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1940 * array. A completely 0 filled entry terminates the table.
1941 *
1942 * See register_sysctl_paths for more details.
1943 */
1944struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1945{
1946 static const struct ctl_path null_path[] = { {} };
1947
1948 return register_sysctl_paths(null_path, table);
1949}
1950
1951/**
1952 * unregister_sysctl_table - unregister a sysctl table hierarchy
1953 * @header: the header returned from register_sysctl_table
1954 *
1955 * Unregisters the sysctl table and all children. proc entries may not
1956 * actually be removed until they are no longer used by anyone.
1957 */
1958void unregister_sysctl_table(struct ctl_table_header * header)
1959{
1960 might_sleep();
1961
1962 if (header == NULL)
1963 return;
1964
1965 spin_lock(&sysctl_lock);
1966 start_unregistering(header);
1967 if (!--header->parent->count) {
1968 WARN_ON(1);
1969 kfree_rcu(header->parent, rcu);
1970 }
1971 if (!--header->count)
1972 kfree_rcu(header, rcu);
1973 spin_unlock(&sysctl_lock);
1974}
1975
1976int sysctl_is_seen(struct ctl_table_header *p)
1977{
1978 struct ctl_table_set *set = p->set;
1979 int res;
1980 spin_lock(&sysctl_lock);
1981 if (p->unregistering)
1982 res = 0;
1983 else if (!set->is_seen)
1984 res = 1;
1985 else
1986 res = set->is_seen(set);
1987 spin_unlock(&sysctl_lock);
1988 return res;
1989}
1990
1991void setup_sysctl_set(struct ctl_table_set *p,
1992 struct ctl_table_set *parent,
1993 int (*is_seen)(struct ctl_table_set *))
1994{
1995 INIT_LIST_HEAD(&p->list);
1996 p->parent = parent ? parent : &sysctl_table_root.default_set;
1997 p->is_seen = is_seen;
1998}
1999
2000#else /* !CONFIG_SYSCTL */
2001struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2002{
2003 return NULL;
2004}
2005
2006struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2007 struct ctl_table *table)
2008{
2009 return NULL;
2010}
2011
2012void unregister_sysctl_table(struct ctl_table_header * table)
2013{
2014}
2015
2016void setup_sysctl_set(struct ctl_table_set *p,
2017 struct ctl_table_set *parent,
2018 int (*is_seen)(struct ctl_table_set *))
2019{
2020}
2021
2022void sysctl_head_put(struct ctl_table_header *head)
2023{
2024}
2025
2026#endif /* CONFIG_SYSCTL */
2027
2028/*
2029 * /proc/sys support
2030 */
2031
2032#ifdef CONFIG_PROC_SYSCTL
2033
2034static int _proc_do_string(void* data, int maxlen, int write,
2035 void __user *buffer,
2036 size_t *lenp, loff_t *ppos)
2037{
2038 size_t len;
2039 char __user *p;
2040 char c;
2041
2042 if (!data || !maxlen || !*lenp) {
2043 *lenp = 0;
2044 return 0;
2045 }
2046
2047 if (write) {
2048 len = 0;
2049 p = buffer;
2050 while (len < *lenp) {
2051 if (get_user(c, p++))
2052 return -EFAULT;
2053 if (c == 0 || c == '\n')
2054 break;
2055 len++;
2056 }
2057 if (len >= maxlen)
2058 len = maxlen-1;
2059 if(copy_from_user(data, buffer, len))
2060 return -EFAULT;
2061 ((char *) data)[len] = 0;
2062 *ppos += *lenp;
2063 } else {
2064 len = strlen(data);
2065 if (len > maxlen)
2066 len = maxlen;
2067
2068 if (*ppos > len) {
2069 *lenp = 0;
2070 return 0;
2071 }
2072
2073 data += *ppos;
2074 len -= *ppos;
2075
2076 if (len > *lenp)
2077 len = *lenp;
2078 if (len)
2079 if(copy_to_user(buffer, data, len))
2080 return -EFAULT;
2081 if (len < *lenp) {
2082 if(put_user('\n', ((char __user *) buffer) + len))
2083 return -EFAULT;
2084 len++;
2085 }
2086 *lenp = len;
2087 *ppos += len;
2088 }
2089 return 0;
2090}
2091
2092/**
2093 * proc_dostring - read a string sysctl
2094 * @table: the sysctl table
2095 * @write: %TRUE if this is a write to the sysctl file
2096 * @buffer: the user buffer
2097 * @lenp: the size of the user buffer
2098 * @ppos: file position
2099 *
2100 * Reads/writes a string from/to the user buffer. If the kernel
2101 * buffer provided is not large enough to hold the string, the
2102 * string is truncated. The copied string is %NULL-terminated.
2103 * If the string is being read by the user process, it is copied
2104 * and a newline '\n' is added. It is truncated if the buffer is
2105 * not large enough.
2106 *
2107 * Returns 0 on success.
2108 */
2109int proc_dostring(struct ctl_table *table, int write,
2110 void __user *buffer, size_t *lenp, loff_t *ppos)
2111{
2112 return _proc_do_string(table->data, table->maxlen, write,
2113 buffer, lenp, ppos);
2114}
2115
2116static size_t proc_skip_spaces(char **buf)
2117{
2118 size_t ret;
2119 char *tmp = skip_spaces(*buf);
2120 ret = tmp - *buf;
2121 *buf = tmp;
2122 return ret;
2123}
2124
2125static void proc_skip_char(char **buf, size_t *size, const char v)
2126{
2127 while (*size) {
2128 if (**buf != v)
2129 break;
2130 (*size)--;
2131 (*buf)++;
2132 }
2133}
2134
2135#define TMPBUFLEN 22
2136/**
2137 * proc_get_long - reads an ASCII formatted integer from a user buffer
2138 *
2139 * @buf: a kernel buffer
2140 * @size: size of the kernel buffer
2141 * @val: this is where the number will be stored
2142 * @neg: set to %TRUE if number is negative
2143 * @perm_tr: a vector which contains the allowed trailers
2144 * @perm_tr_len: size of the perm_tr vector
2145 * @tr: pointer to store the trailer character
2146 *
2147 * In case of success %0 is returned and @buf and @size are updated with
2148 * the amount of bytes read. If @tr is non-NULL and a trailing
2149 * character exists (size is non-zero after returning from this
2150 * function), @tr is updated with the trailing character.
2151 */
2152static int proc_get_long(char **buf, size_t *size,
2153 unsigned long *val, bool *neg,
2154 const char *perm_tr, unsigned perm_tr_len, char *tr)
2155{
2156 int len;
2157 char *p, tmp[TMPBUFLEN];
2158
2159 if (!*size)
2160 return -EINVAL;
2161
2162 len = *size;
2163 if (len > TMPBUFLEN - 1)
2164 len = TMPBUFLEN - 1;
2165
2166 memcpy(tmp, *buf, len);
2167
2168 tmp[len] = 0;
2169 p = tmp;
2170 if (*p == '-' && *size > 1) {
2171 *neg = true;
2172 p++;
2173 } else
2174 *neg = false;
2175 if (!isdigit(*p))
2176 return -EINVAL;
2177
2178 *val = simple_strtoul(p, &p, 0);
2179
2180 len = p - tmp;
2181
2182 /* We don't know if the next char is whitespace thus we may accept
2183 * invalid integers (e.g. 1234...a) or two integers instead of one
2184 * (e.g. 123...1). So lets not allow such large numbers. */
2185 if (len == TMPBUFLEN - 1)
2186 return -EINVAL;
2187
2188 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2189 return -EINVAL;
2190
2191 if (tr && (len < *size))
2192 *tr = *p;
2193
2194 *buf += len;
2195 *size -= len;
2196
2197 return 0;
2198}
2199
2200/**
2201 * proc_put_long - converts an integer to a decimal ASCII formatted string
2202 *
2203 * @buf: the user buffer
2204 * @size: the size of the user buffer
2205 * @val: the integer to be converted
2206 * @neg: sign of the number, %TRUE for negative
2207 *
2208 * In case of success %0 is returned and @buf and @size are updated with
2209 * the amount of bytes written.
2210 */
2211static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2212 bool neg)
2213{
2214 int len;
2215 char tmp[TMPBUFLEN], *p = tmp;
2216
2217 sprintf(p, "%s%lu", neg ? "-" : "", val);
2218 len = strlen(tmp);
2219 if (len > *size)
2220 len = *size;
2221 if (copy_to_user(*buf, tmp, len))
2222 return -EFAULT;
2223 *size -= len;
2224 *buf += len;
2225 return 0;
2226}
2227#undef TMPBUFLEN
2228
2229static int proc_put_char(void __user **buf, size_t *size, char c)
2230{
2231 if (*size) {
2232 char __user **buffer = (char __user **)buf;
2233 if (put_user(c, *buffer))
2234 return -EFAULT;
2235 (*size)--, (*buffer)++;
2236 *buf = *buffer;
2237 }
2238 return 0;
2239}
2240
2241static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2242 int *valp,
2243 int write, void *data)
2244{
2245 if (write) {
2246 *valp = *negp ? -*lvalp : *lvalp;
2247 } else {
2248 int val = *valp;
2249 if (val < 0) {
2250 *negp = true;
2251 *lvalp = (unsigned long)-val;
2252 } else {
2253 *negp = false;
2254 *lvalp = (unsigned long)val;
2255 }
2256 }
2257 return 0;
2258}
2259
2260static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2261
2262static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2263 int write, void __user *buffer,
2264 size_t *lenp, loff_t *ppos,
2265 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2266 int write, void *data),
2267 void *data)
2268{
2269 int *i, vleft, first = 1, err = 0;
2270 unsigned long page = 0;
2271 size_t left;
2272 char *kbuf;
2273
2274 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2275 *lenp = 0;
2276 return 0;
2277 }
2278
2279 i = (int *) tbl_data;
2280 vleft = table->maxlen / sizeof(*i);
2281 left = *lenp;
2282
2283 if (!conv)
2284 conv = do_proc_dointvec_conv;
2285
2286 if (write) {
2287 if (left > PAGE_SIZE - 1)
2288 left = PAGE_SIZE - 1;
2289 page = __get_free_page(GFP_TEMPORARY);
2290 kbuf = (char *) page;
2291 if (!kbuf)
2292 return -ENOMEM;
2293 if (copy_from_user(kbuf, buffer, left)) {
2294 err = -EFAULT;
2295 goto free;
2296 }
2297 kbuf[left] = 0;
2298 }
2299
2300 for (; left && vleft--; i++, first=0) {
2301 unsigned long lval;
2302 bool neg;
2303
2304 if (write) {
2305 left -= proc_skip_spaces(&kbuf);
2306
2307 if (!left)
2308 break;
2309 err = proc_get_long(&kbuf, &left, &lval, &neg,
2310 proc_wspace_sep,
2311 sizeof(proc_wspace_sep), NULL);
2312 if (err)
2313 break;
2314 if (conv(&neg, &lval, i, 1, data)) {
2315 err = -EINVAL;
2316 break;
2317 }
2318 } else {
2319 if (conv(&neg, &lval, i, 0, data)) {
2320 err = -EINVAL;
2321 break;
2322 }
2323 if (!first)
2324 err = proc_put_char(&buffer, &left, '\t');
2325 if (err)
2326 break;
2327 err = proc_put_long(&buffer, &left, lval, neg);
2328 if (err)
2329 break;
2330 }
2331 }
2332
2333 if (!write && !first && left && !err)
2334 err = proc_put_char(&buffer, &left, '\n');
2335 if (write && !err && left)
2336 left -= proc_skip_spaces(&kbuf);
2337free:
2338 if (write) {
2339 free_page(page);
2340 if (first)
2341 return err ? : -EINVAL;
2342 }
2343 *lenp -= left;
2344 *ppos += *lenp;
2345 return err;
2346}
2347
2348static int do_proc_dointvec(struct ctl_table *table, int write,
2349 void __user *buffer, size_t *lenp, loff_t *ppos,
2350 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2351 int write, void *data),
2352 void *data)
2353{
2354 return __do_proc_dointvec(table->data, table, write,
2355 buffer, lenp, ppos, conv, data);
2356}
2357
2358/**
2359 * proc_dointvec - read a vector of integers
2360 * @table: the sysctl table
2361 * @write: %TRUE if this is a write to the sysctl file
2362 * @buffer: the user buffer
2363 * @lenp: the size of the user buffer
2364 * @ppos: file position
2365 *
2366 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2367 * values from/to the user buffer, treated as an ASCII string.
2368 *
2369 * Returns 0 on success.
2370 */
2371int proc_dointvec(struct ctl_table *table, int write,
2372 void __user *buffer, size_t *lenp, loff_t *ppos)
2373{
2374 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2375 NULL,NULL);
2376}
2377
2378/*
2379 * Taint values can only be increased
2380 * This means we can safely use a temporary.
2381 */
2382static int proc_taint(struct ctl_table *table, int write,
2383 void __user *buffer, size_t *lenp, loff_t *ppos)
2384{
2385 struct ctl_table t;
2386 unsigned long tmptaint = get_taint();
2387 int err;
2388
2389 if (write && !capable(CAP_SYS_ADMIN))
2390 return -EPERM;
2391
2392 t = *table;
2393 t.data = &tmptaint;
2394 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2395 if (err < 0)
2396 return err;
2397
2398 if (write) {
2399 /*
2400 * Poor man's atomic or. Not worth adding a primitive
2401 * to everyone's atomic.h for this
2402 */
2403 int i;
2404 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2405 if ((tmptaint >> i) & 1)
2406 add_taint(i);
2407 }
2408 }
2409
2410 return err;
2411}
2412
2413#ifdef CONFIG_PRINTK
2414static int proc_dmesg_restrict(struct ctl_table *table, int write,
2415 void __user *buffer, size_t *lenp, loff_t *ppos)
2416{
2417 if (write && !capable(CAP_SYS_ADMIN))
2418 return -EPERM;
2419
2420 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2421}
2422#endif
2423
2424struct do_proc_dointvec_minmax_conv_param {
2425 int *min;
2426 int *max;
2427};
2428
2429static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2430 int *valp,
2431 int write, void *data)
2432{
2433 struct do_proc_dointvec_minmax_conv_param *param = data;
2434 if (write) {
2435 int val = *negp ? -*lvalp : *lvalp;
2436 if ((param->min && *param->min > val) ||
2437 (param->max && *param->max < val))
2438 return -EINVAL;
2439 *valp = val;
2440 } else {
2441 int val = *valp;
2442 if (val < 0) {
2443 *negp = true;
2444 *lvalp = (unsigned long)-val;
2445 } else {
2446 *negp = false;
2447 *lvalp = (unsigned long)val;
2448 }
2449 }
2450 return 0;
2451}
2452
2453/**
2454 * proc_dointvec_minmax - read a vector of integers with min/max values
2455 * @table: the sysctl table
2456 * @write: %TRUE if this is a write to the sysctl file
2457 * @buffer: the user buffer
2458 * @lenp: the size of the user buffer
2459 * @ppos: file position
2460 *
2461 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2462 * values from/to the user buffer, treated as an ASCII string.
2463 *
2464 * This routine will ensure the values are within the range specified by
2465 * table->extra1 (min) and table->extra2 (max).
2466 *
2467 * Returns 0 on success.
2468 */
2469int proc_dointvec_minmax(struct ctl_table *table, int write,
2470 void __user *buffer, size_t *lenp, loff_t *ppos)
2471{
2472 struct do_proc_dointvec_minmax_conv_param param = {
2473 .min = (int *) table->extra1,
2474 .max = (int *) table->extra2,
2475 };
2476 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2477 do_proc_dointvec_minmax_conv, ¶m);
2478}
2479
2480static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2481 void __user *buffer,
2482 size_t *lenp, loff_t *ppos,
2483 unsigned long convmul,
2484 unsigned long convdiv)
2485{
2486 unsigned long *i, *min, *max;
2487 int vleft, first = 1, err = 0;
2488 unsigned long page = 0;
2489 size_t left;
2490 char *kbuf;
2491
2492 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2493 *lenp = 0;
2494 return 0;
2495 }
2496
2497 i = (unsigned long *) data;
2498 min = (unsigned long *) table->extra1;
2499 max = (unsigned long *) table->extra2;
2500 vleft = table->maxlen / sizeof(unsigned long);
2501 left = *lenp;
2502
2503 if (write) {
2504 if (left > PAGE_SIZE - 1)
2505 left = PAGE_SIZE - 1;
2506 page = __get_free_page(GFP_TEMPORARY);
2507 kbuf = (char *) page;
2508 if (!kbuf)
2509 return -ENOMEM;
2510 if (copy_from_user(kbuf, buffer, left)) {
2511 err = -EFAULT;
2512 goto free;
2513 }
2514 kbuf[left] = 0;
2515 }
2516
2517 for (; left && vleft--; i++, first = 0) {
2518 unsigned long val;
2519
2520 if (write) {
2521 bool neg;
2522
2523 left -= proc_skip_spaces(&kbuf);
2524
2525 err = proc_get_long(&kbuf, &left, &val, &neg,
2526 proc_wspace_sep,
2527 sizeof(proc_wspace_sep), NULL);
2528 if (err)
2529 break;
2530 if (neg)
2531 continue;
2532 if ((min && val < *min) || (max && val > *max))
2533 continue;
2534 *i = val;
2535 } else {
2536 val = convdiv * (*i) / convmul;
2537 if (!first)
2538 err = proc_put_char(&buffer, &left, '\t');
2539 err = proc_put_long(&buffer, &left, val, false);
2540 if (err)
2541 break;
2542 }
2543 }
2544
2545 if (!write && !first && left && !err)
2546 err = proc_put_char(&buffer, &left, '\n');
2547 if (write && !err)
2548 left -= proc_skip_spaces(&kbuf);
2549free:
2550 if (write) {
2551 free_page(page);
2552 if (first)
2553 return err ? : -EINVAL;
2554 }
2555 *lenp -= left;
2556 *ppos += *lenp;
2557 return err;
2558}
2559
2560static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2561 void __user *buffer,
2562 size_t *lenp, loff_t *ppos,
2563 unsigned long convmul,
2564 unsigned long convdiv)
2565{
2566 return __do_proc_doulongvec_minmax(table->data, table, write,
2567 buffer, lenp, ppos, convmul, convdiv);
2568}
2569
2570/**
2571 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2572 * @table: the sysctl table
2573 * @write: %TRUE if this is a write to the sysctl file
2574 * @buffer: the user buffer
2575 * @lenp: the size of the user buffer
2576 * @ppos: file position
2577 *
2578 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2579 * values from/to the user buffer, treated as an ASCII string.
2580 *
2581 * This routine will ensure the values are within the range specified by
2582 * table->extra1 (min) and table->extra2 (max).
2583 *
2584 * Returns 0 on success.
2585 */
2586int proc_doulongvec_minmax(struct ctl_table *table, int write,
2587 void __user *buffer, size_t *lenp, loff_t *ppos)
2588{
2589 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2590}
2591
2592/**
2593 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2594 * @table: the sysctl table
2595 * @write: %TRUE if this is a write to the sysctl file
2596 * @buffer: the user buffer
2597 * @lenp: the size of the user buffer
2598 * @ppos: file position
2599 *
2600 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2601 * values from/to the user buffer, treated as an ASCII string. The values
2602 * are treated as milliseconds, and converted to jiffies when they are stored.
2603 *
2604 * This routine will ensure the values are within the range specified by
2605 * table->extra1 (min) and table->extra2 (max).
2606 *
2607 * Returns 0 on success.
2608 */
2609int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2610 void __user *buffer,
2611 size_t *lenp, loff_t *ppos)
2612{
2613 return do_proc_doulongvec_minmax(table, write, buffer,
2614 lenp, ppos, HZ, 1000l);
2615}
2616
2617
2618static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2619 int *valp,
2620 int write, void *data)
2621{
2622 if (write) {
2623 if (*lvalp > LONG_MAX / HZ)
2624 return 1;
2625 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2626 } else {
2627 int val = *valp;
2628 unsigned long lval;
2629 if (val < 0) {
2630 *negp = true;
2631 lval = (unsigned long)-val;
2632 } else {
2633 *negp = false;
2634 lval = (unsigned long)val;
2635 }
2636 *lvalp = lval / HZ;
2637 }
2638 return 0;
2639}
2640
2641static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2642 int *valp,
2643 int write, void *data)
2644{
2645 if (write) {
2646 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2647 return 1;
2648 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2649 } else {
2650 int val = *valp;
2651 unsigned long lval;
2652 if (val < 0) {
2653 *negp = true;
2654 lval = (unsigned long)-val;
2655 } else {
2656 *negp = false;
2657 lval = (unsigned long)val;
2658 }
2659 *lvalp = jiffies_to_clock_t(lval);
2660 }
2661 return 0;
2662}
2663
2664static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2665 int *valp,
2666 int write, void *data)
2667{
2668 if (write) {
2669 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2670 } else {
2671 int val = *valp;
2672 unsigned long lval;
2673 if (val < 0) {
2674 *negp = true;
2675 lval = (unsigned long)-val;
2676 } else {
2677 *negp = false;
2678 lval = (unsigned long)val;
2679 }
2680 *lvalp = jiffies_to_msecs(lval);
2681 }
2682 return 0;
2683}
2684
2685/**
2686 * proc_dointvec_jiffies - read a vector of integers as seconds
2687 * @table: the sysctl table
2688 * @write: %TRUE if this is a write to the sysctl file
2689 * @buffer: the user buffer
2690 * @lenp: the size of the user buffer
2691 * @ppos: file position
2692 *
2693 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2694 * values from/to the user buffer, treated as an ASCII string.
2695 * The values read are assumed to be in seconds, and are converted into
2696 * jiffies.
2697 *
2698 * Returns 0 on success.
2699 */
2700int proc_dointvec_jiffies(struct ctl_table *table, int write,
2701 void __user *buffer, size_t *lenp, loff_t *ppos)
2702{
2703 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2704 do_proc_dointvec_jiffies_conv,NULL);
2705}
2706
2707/**
2708 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2709 * @table: the sysctl table
2710 * @write: %TRUE if this is a write to the sysctl file
2711 * @buffer: the user buffer
2712 * @lenp: the size of the user buffer
2713 * @ppos: pointer to the file position
2714 *
2715 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2716 * values from/to the user buffer, treated as an ASCII string.
2717 * The values read are assumed to be in 1/USER_HZ seconds, and
2718 * are converted into jiffies.
2719 *
2720 * Returns 0 on success.
2721 */
2722int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2723 void __user *buffer, size_t *lenp, loff_t *ppos)
2724{
2725 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2726 do_proc_dointvec_userhz_jiffies_conv,NULL);
2727}
2728
2729/**
2730 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2731 * @table: the sysctl table
2732 * @write: %TRUE if this is a write to the sysctl file
2733 * @buffer: the user buffer
2734 * @lenp: the size of the user buffer
2735 * @ppos: file position
2736 * @ppos: the current position in the file
2737 *
2738 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2739 * values from/to the user buffer, treated as an ASCII string.
2740 * The values read are assumed to be in 1/1000 seconds, and
2741 * are converted into jiffies.
2742 *
2743 * Returns 0 on success.
2744 */
2745int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2746 void __user *buffer, size_t *lenp, loff_t *ppos)
2747{
2748 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2749 do_proc_dointvec_ms_jiffies_conv, NULL);
2750}
2751
2752static int proc_do_cad_pid(struct ctl_table *table, int write,
2753 void __user *buffer, size_t *lenp, loff_t *ppos)
2754{
2755 struct pid *new_pid;
2756 pid_t tmp;
2757 int r;
2758
2759 tmp = pid_vnr(cad_pid);
2760
2761 r = __do_proc_dointvec(&tmp, table, write, buffer,
2762 lenp, ppos, NULL, NULL);
2763 if (r || !write)
2764 return r;
2765
2766 new_pid = find_get_pid(tmp);
2767 if (!new_pid)
2768 return -ESRCH;
2769
2770 put_pid(xchg(&cad_pid, new_pid));
2771 return 0;
2772}
2773
2774/**
2775 * proc_do_large_bitmap - read/write from/to a large bitmap
2776 * @table: the sysctl table
2777 * @write: %TRUE if this is a write to the sysctl file
2778 * @buffer: the user buffer
2779 * @lenp: the size of the user buffer
2780 * @ppos: file position
2781 *
2782 * The bitmap is stored at table->data and the bitmap length (in bits)
2783 * in table->maxlen.
2784 *
2785 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2786 * large bitmaps may be represented in a compact manner. Writing into
2787 * the file will clear the bitmap then update it with the given input.
2788 *
2789 * Returns 0 on success.
2790 */
2791int proc_do_large_bitmap(struct ctl_table *table, int write,
2792 void __user *buffer, size_t *lenp, loff_t *ppos)
2793{
2794 int err = 0;
2795 bool first = 1;
2796 size_t left = *lenp;
2797 unsigned long bitmap_len = table->maxlen;
2798 unsigned long *bitmap = (unsigned long *) table->data;
2799 unsigned long *tmp_bitmap = NULL;
2800 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2801
2802 if (!bitmap_len || !left || (*ppos && !write)) {
2803 *lenp = 0;
2804 return 0;
2805 }
2806
2807 if (write) {
2808 unsigned long page = 0;
2809 char *kbuf;
2810
2811 if (left > PAGE_SIZE - 1)
2812 left = PAGE_SIZE - 1;
2813
2814 page = __get_free_page(GFP_TEMPORARY);
2815 kbuf = (char *) page;
2816 if (!kbuf)
2817 return -ENOMEM;
2818 if (copy_from_user(kbuf, buffer, left)) {
2819 free_page(page);
2820 return -EFAULT;
2821 }
2822 kbuf[left] = 0;
2823
2824 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2825 GFP_KERNEL);
2826 if (!tmp_bitmap) {
2827 free_page(page);
2828 return -ENOMEM;
2829 }
2830 proc_skip_char(&kbuf, &left, '\n');
2831 while (!err && left) {
2832 unsigned long val_a, val_b;
2833 bool neg;
2834
2835 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2836 sizeof(tr_a), &c);
2837 if (err)
2838 break;
2839 if (val_a >= bitmap_len || neg) {
2840 err = -EINVAL;
2841 break;
2842 }
2843
2844 val_b = val_a;
2845 if (left) {
2846 kbuf++;
2847 left--;
2848 }
2849
2850 if (c == '-') {
2851 err = proc_get_long(&kbuf, &left, &val_b,
2852 &neg, tr_b, sizeof(tr_b),
2853 &c);
2854 if (err)
2855 break;
2856 if (val_b >= bitmap_len || neg ||
2857 val_a > val_b) {
2858 err = -EINVAL;
2859 break;
2860 }
2861 if (left) {
2862 kbuf++;
2863 left--;
2864 }
2865 }
2866
2867 while (val_a <= val_b)
2868 set_bit(val_a++, tmp_bitmap);
2869
2870 first = 0;
2871 proc_skip_char(&kbuf, &left, '\n');
2872 }
2873 free_page(page);
2874 } else {
2875 unsigned long bit_a, bit_b = 0;
2876
2877 while (left) {
2878 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2879 if (bit_a >= bitmap_len)
2880 break;
2881 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2882 bit_a + 1) - 1;
2883
2884 if (!first) {
2885 err = proc_put_char(&buffer, &left, ',');
2886 if (err)
2887 break;
2888 }
2889 err = proc_put_long(&buffer, &left, bit_a, false);
2890 if (err)
2891 break;
2892 if (bit_a != bit_b) {
2893 err = proc_put_char(&buffer, &left, '-');
2894 if (err)
2895 break;
2896 err = proc_put_long(&buffer, &left, bit_b, false);
2897 if (err)
2898 break;
2899 }
2900
2901 first = 0; bit_b++;
2902 }
2903 if (!err)
2904 err = proc_put_char(&buffer, &left, '\n');
2905 }
2906
2907 if (!err) {
2908 if (write) {
2909 if (*ppos)
2910 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2911 else
2912 memcpy(bitmap, tmp_bitmap,
2913 BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long));
2914 }
2915 kfree(tmp_bitmap);
2916 *lenp -= left;
2917 *ppos += *lenp;
2918 return 0;
2919 } else {
2920 kfree(tmp_bitmap);
2921 return err;
2922 }
2923}
2924
2925#else /* CONFIG_PROC_SYSCTL */
2926
2927int proc_dostring(struct ctl_table *table, int write,
2928 void __user *buffer, size_t *lenp, loff_t *ppos)
2929{
2930 return -ENOSYS;
2931}
2932
2933int proc_dointvec(struct ctl_table *table, int write,
2934 void __user *buffer, size_t *lenp, loff_t *ppos)
2935{
2936 return -ENOSYS;
2937}
2938
2939int proc_dointvec_minmax(struct ctl_table *table, int write,
2940 void __user *buffer, size_t *lenp, loff_t *ppos)
2941{
2942 return -ENOSYS;
2943}
2944
2945int proc_dointvec_jiffies(struct ctl_table *table, int write,
2946 void __user *buffer, size_t *lenp, loff_t *ppos)
2947{
2948 return -ENOSYS;
2949}
2950
2951int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2952 void __user *buffer, size_t *lenp, loff_t *ppos)
2953{
2954 return -ENOSYS;
2955}
2956
2957int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2958 void __user *buffer, size_t *lenp, loff_t *ppos)
2959{
2960 return -ENOSYS;
2961}
2962
2963int proc_doulongvec_minmax(struct ctl_table *table, int write,
2964 void __user *buffer, size_t *lenp, loff_t *ppos)
2965{
2966 return -ENOSYS;
2967}
2968
2969int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2970 void __user *buffer,
2971 size_t *lenp, loff_t *ppos)
2972{
2973 return -ENOSYS;
2974}
2975
2976
2977#endif /* CONFIG_PROC_SYSCTL */
2978
2979/*
2980 * No sense putting this after each symbol definition, twice,
2981 * exception granted :-)
2982 */
2983EXPORT_SYMBOL(proc_dointvec);
2984EXPORT_SYMBOL(proc_dointvec_jiffies);
2985EXPORT_SYMBOL(proc_dointvec_minmax);
2986EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2987EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2988EXPORT_SYMBOL(proc_dostring);
2989EXPORT_SYMBOL(proc_doulongvec_minmax);
2990EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2991EXPORT_SYMBOL(register_sysctl_table);
2992EXPORT_SYMBOL(register_sysctl_paths);
2993EXPORT_SYMBOL(unregister_sysctl_table);