1 // SPDX-License-Identifier: GPL-2.0-only 2 /* 3 * sysctl.c: General linux system control interface 4 * 5 * Begun 24 March 1995, Stephen Tweedie 6 * Added /proc support, Dec 1995 7 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas. 8 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver. 9 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver. 10 * Dynamic registration fixes, Stephen Tweedie. 11 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn. 12 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris 13 * Horn. 14 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer. 15 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer. 16 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill 17 * Wendling. 18 * The list_for_each() macro wasn't appropriate for the sysctl loop. 19 * Removed it and replaced it with older style, 03/23/00, Bill Wendling 20 */ 21 22 #include <linux/module.h> 23 #include <linux/aio.h> 24 #include <linux/mm.h> 25 #include <linux/swap.h> 26 #include <linux/slab.h> 27 #include <linux/sysctl.h> 28 #include <linux/bitmap.h> 29 #include <linux/signal.h> 30 #include <linux/printk.h> 31 #include <linux/proc_fs.h> 32 #include <linux/security.h> 33 #include <linux/ctype.h> 34 #include <linux/kmemleak.h> 35 #include <linux/fs.h> 36 #include <linux/init.h> 37 #include <linux/kernel.h> 38 #include <linux/kobject.h> 39 #include <linux/net.h> 40 #include <linux/sysrq.h> 41 #include <linux/highuid.h> 42 #include <linux/writeback.h> 43 #include <linux/ratelimit.h> 44 #include <linux/compaction.h> 45 #include <linux/hugetlb.h> 46 #include <linux/initrd.h> 47 #include <linux/key.h> 48 #include <linux/times.h> 49 #include <linux/limits.h> 50 #include <linux/dcache.h> 51 #include <linux/dnotify.h> 52 #include <linux/syscalls.h> 53 #include <linux/vmstat.h> 54 #include <linux/nfs_fs.h> 55 #include <linux/acpi.h> 56 #include <linux/reboot.h> 57 #include <linux/ftrace.h> 58 #include <linux/perf_event.h> 59 #include <linux/kprobes.h> 60 #include <linux/pipe_fs_i.h> 61 #include <linux/oom.h> 62 #include <linux/kmod.h> 63 #include <linux/capability.h> 64 #include <linux/binfmts.h> 65 #include <linux/sched/sysctl.h> 66 #include <linux/sched/coredump.h> 67 #include <linux/kexec.h> 68 #include <linux/bpf.h> 69 #include <linux/mount.h> 70 #include <linux/userfaultfd_k.h> 71 #include <linux/coredump.h> 72 #include <linux/latencytop.h> 73 #include <linux/pid.h> 74 75 #include "../lib/kstrtox.h" 76 77 #include <linux/uaccess.h> 78 #include <asm/processor.h> 79 80 #ifdef CONFIG_X86 81 #include <asm/nmi.h> 82 #include <asm/stacktrace.h> 83 #include <asm/io.h> 84 #endif 85 #ifdef CONFIG_SPARC 86 #include <asm/setup.h> 87 #endif 88 #ifdef CONFIG_BSD_PROCESS_ACCT 89 #include <linux/acct.h> 90 #endif 91 #ifdef CONFIG_RT_MUTEXES 92 #include <linux/rtmutex.h> 93 #endif 94 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT) 95 #include <linux/lockdep.h> 96 #endif 97 #ifdef CONFIG_CHR_DEV_SG 98 #include <scsi/sg.h> 99 #endif 100 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE 101 #include <linux/stackleak.h> 102 #endif 103 #ifdef CONFIG_LOCKUP_DETECTOR 104 #include <linux/nmi.h> 105 #endif 106 107 #if defined(CONFIG_SYSCTL) 108 109 /* Constants used for minimum and maximum */ 110 #ifdef CONFIG_LOCKUP_DETECTOR 111 static int sixty = 60; 112 #endif 113 114 static int __maybe_unused neg_one = -1; 115 static int __maybe_unused two = 2; 116 static int __maybe_unused four = 4; 117 static unsigned long zero_ul; 118 static unsigned long one_ul = 1; 119 static unsigned long long_max = LONG_MAX; 120 static int one_hundred = 100; 121 static int one_thousand = 1000; 122 #ifdef CONFIG_PRINTK 123 static int ten_thousand = 10000; 124 #endif 125 #ifdef CONFIG_PERF_EVENTS 126 static int six_hundred_forty_kb = 640 * 1024; 127 #endif 128 129 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */ 130 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE; 131 132 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */ 133 static int maxolduid = 65535; 134 static int minolduid; 135 136 static int ngroups_max = NGROUPS_MAX; 137 static const int cap_last_cap = CAP_LAST_CAP; 138 139 /* 140 * This is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs 141 * and hung_task_check_interval_secs 142 */ 143 #ifdef CONFIG_DETECT_HUNG_TASK 144 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ); 145 #endif 146 147 #ifdef CONFIG_INOTIFY_USER 148 #include <linux/inotify.h> 149 #endif 150 151 #ifdef CONFIG_PROC_SYSCTL 152 153 /** 154 * enum sysctl_writes_mode - supported sysctl write modes 155 * 156 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value 157 * to be written, and multiple writes on the same sysctl file descriptor 158 * will rewrite the sysctl value, regardless of file position. No warning 159 * is issued when the initial position is not 0. 160 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is 161 * not 0. 162 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at 163 * file position 0 and the value must be fully contained in the buffer 164 * sent to the write syscall. If dealing with strings respect the file 165 * position, but restrict this to the max length of the buffer, anything 166 * passed the max length will be ignored. Multiple writes will append 167 * to the buffer. 168 * 169 * These write modes control how current file position affects the behavior of 170 * updating sysctl values through the proc interface on each write. 171 */ 172 enum sysctl_writes_mode { 173 SYSCTL_WRITES_LEGACY = -1, 174 SYSCTL_WRITES_WARN = 0, 175 SYSCTL_WRITES_STRICT = 1, 176 }; 177 178 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT; 179 #endif /* CONFIG_PROC_SYSCTL */ 180 181 #if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \ 182 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT) 183 int sysctl_legacy_va_layout; 184 #endif 185 186 #ifdef CONFIG_SCHED_DEBUG 187 static int min_sched_granularity_ns = 100000; /* 100 usecs */ 188 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */ 189 static int min_wakeup_granularity_ns; /* 0 usecs */ 190 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */ 191 #ifdef CONFIG_SMP 192 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE; 193 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1; 194 #endif /* CONFIG_SMP */ 195 #endif /* CONFIG_SCHED_DEBUG */ 196 197 #ifdef CONFIG_COMPACTION 198 static int min_extfrag_threshold; 199 static int max_extfrag_threshold = 1000; 200 #endif 201 202 #endif /* CONFIG_SYSCTL */ 203 204 /* 205 * /proc/sys support 206 */ 207 208 #ifdef CONFIG_PROC_SYSCTL 209 210 static int _proc_do_string(char *data, int maxlen, int write, 211 char *buffer, size_t *lenp, loff_t *ppos) 212 { 213 size_t len; 214 char c, *p; 215 216 if (!data || !maxlen || !*lenp) { 217 *lenp = 0; 218 return 0; 219 } 220 221 if (write) { 222 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) { 223 /* Only continue writes not past the end of buffer. */ 224 len = strlen(data); 225 if (len > maxlen - 1) 226 len = maxlen - 1; 227 228 if (*ppos > len) 229 return 0; 230 len = *ppos; 231 } else { 232 /* Start writing from beginning of buffer. */ 233 len = 0; 234 } 235 236 *ppos += *lenp; 237 p = buffer; 238 while ((p - buffer) < *lenp && len < maxlen - 1) { 239 c = *(p++); 240 if (c == 0 || c == '\n') 241 break; 242 data[len++] = c; 243 } 244 data[len] = 0; 245 } else { 246 len = strlen(data); 247 if (len > maxlen) 248 len = maxlen; 249 250 if (*ppos > len) { 251 *lenp = 0; 252 return 0; 253 } 254 255 data += *ppos; 256 len -= *ppos; 257 258 if (len > *lenp) 259 len = *lenp; 260 if (len) 261 memcpy(buffer, data, len); 262 if (len < *lenp) { 263 buffer[len] = '\n'; 264 len++; 265 } 266 *lenp = len; 267 *ppos += len; 268 } 269 return 0; 270 } 271 272 static void warn_sysctl_write(struct ctl_table *table) 273 { 274 pr_warn_once("%s wrote to %s when file position was not 0!\n" 275 "This will not be supported in the future. To silence this\n" 276 "warning, set kernel.sysctl_writes_strict = -1\n", 277 current->comm, table->procname); 278 } 279 280 /** 281 * proc_first_pos_non_zero_ignore - check if first position is allowed 282 * @ppos: file position 283 * @table: the sysctl table 284 * 285 * Returns true if the first position is non-zero and the sysctl_writes_strict 286 * mode indicates this is not allowed for numeric input types. String proc 287 * handlers can ignore the return value. 288 */ 289 static bool proc_first_pos_non_zero_ignore(loff_t *ppos, 290 struct ctl_table *table) 291 { 292 if (!*ppos) 293 return false; 294 295 switch (sysctl_writes_strict) { 296 case SYSCTL_WRITES_STRICT: 297 return true; 298 case SYSCTL_WRITES_WARN: 299 warn_sysctl_write(table); 300 return false; 301 default: 302 return false; 303 } 304 } 305 306 /** 307 * proc_dostring - read a string sysctl 308 * @table: the sysctl table 309 * @write: %TRUE if this is a write to the sysctl file 310 * @buffer: the user buffer 311 * @lenp: the size of the user buffer 312 * @ppos: file position 313 * 314 * Reads/writes a string from/to the user buffer. If the kernel 315 * buffer provided is not large enough to hold the string, the 316 * string is truncated. The copied string is %NULL-terminated. 317 * If the string is being read by the user process, it is copied 318 * and a newline '\n' is added. It is truncated if the buffer is 319 * not large enough. 320 * 321 * Returns 0 on success. 322 */ 323 int proc_dostring(struct ctl_table *table, int write, 324 void *buffer, size_t *lenp, loff_t *ppos) 325 { 326 if (write) 327 proc_first_pos_non_zero_ignore(ppos, table); 328 329 return _proc_do_string(table->data, table->maxlen, write, buffer, lenp, 330 ppos); 331 } 332 333 static size_t proc_skip_spaces(char **buf) 334 { 335 size_t ret; 336 char *tmp = skip_spaces(*buf); 337 ret = tmp - *buf; 338 *buf = tmp; 339 return ret; 340 } 341 342 static void proc_skip_char(char **buf, size_t *size, const char v) 343 { 344 while (*size) { 345 if (**buf != v) 346 break; 347 (*size)--; 348 (*buf)++; 349 } 350 } 351 352 /** 353 * strtoul_lenient - parse an ASCII formatted integer from a buffer and only 354 * fail on overflow 355 * 356 * @cp: kernel buffer containing the string to parse 357 * @endp: pointer to store the trailing characters 358 * @base: the base to use 359 * @res: where the parsed integer will be stored 360 * 361 * In case of success 0 is returned and @res will contain the parsed integer, 362 * @endp will hold any trailing characters. 363 * This function will fail the parse on overflow. If there wasn't an overflow 364 * the function will defer the decision what characters count as invalid to the 365 * caller. 366 */ 367 static int strtoul_lenient(const char *cp, char **endp, unsigned int base, 368 unsigned long *res) 369 { 370 unsigned long long result; 371 unsigned int rv; 372 373 cp = _parse_integer_fixup_radix(cp, &base); 374 rv = _parse_integer(cp, base, &result); 375 if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result)) 376 return -ERANGE; 377 378 cp += rv; 379 380 if (endp) 381 *endp = (char *)cp; 382 383 *res = (unsigned long)result; 384 return 0; 385 } 386 387 #define TMPBUFLEN 22 388 /** 389 * proc_get_long - reads an ASCII formatted integer from a user buffer 390 * 391 * @buf: a kernel buffer 392 * @size: size of the kernel buffer 393 * @val: this is where the number will be stored 394 * @neg: set to %TRUE if number is negative 395 * @perm_tr: a vector which contains the allowed trailers 396 * @perm_tr_len: size of the perm_tr vector 397 * @tr: pointer to store the trailer character 398 * 399 * In case of success %0 is returned and @buf and @size are updated with 400 * the amount of bytes read. If @tr is non-NULL and a trailing 401 * character exists (size is non-zero after returning from this 402 * function), @tr is updated with the trailing character. 403 */ 404 static int proc_get_long(char **buf, size_t *size, 405 unsigned long *val, bool *neg, 406 const char *perm_tr, unsigned perm_tr_len, char *tr) 407 { 408 int len; 409 char *p, tmp[TMPBUFLEN]; 410 411 if (!*size) 412 return -EINVAL; 413 414 len = *size; 415 if (len > TMPBUFLEN - 1) 416 len = TMPBUFLEN - 1; 417 418 memcpy(tmp, *buf, len); 419 420 tmp[len] = 0; 421 p = tmp; 422 if (*p == '-' && *size > 1) { 423 *neg = true; 424 p++; 425 } else 426 *neg = false; 427 if (!isdigit(*p)) 428 return -EINVAL; 429 430 if (strtoul_lenient(p, &p, 0, val)) 431 return -EINVAL; 432 433 len = p - tmp; 434 435 /* We don't know if the next char is whitespace thus we may accept 436 * invalid integers (e.g. 1234...a) or two integers instead of one 437 * (e.g. 123...1). So lets not allow such large numbers. */ 438 if (len == TMPBUFLEN - 1) 439 return -EINVAL; 440 441 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len)) 442 return -EINVAL; 443 444 if (tr && (len < *size)) 445 *tr = *p; 446 447 *buf += len; 448 *size -= len; 449 450 return 0; 451 } 452 453 /** 454 * proc_put_long - converts an integer to a decimal ASCII formatted string 455 * 456 * @buf: the user buffer 457 * @size: the size of the user buffer 458 * @val: the integer to be converted 459 * @neg: sign of the number, %TRUE for negative 460 * 461 * In case of success @buf and @size are updated with the amount of bytes 462 * written. 463 */ 464 static void proc_put_long(void **buf, size_t *size, unsigned long val, bool neg) 465 { 466 int len; 467 char tmp[TMPBUFLEN], *p = tmp; 468 469 sprintf(p, "%s%lu", neg ? "-" : "", val); 470 len = strlen(tmp); 471 if (len > *size) 472 len = *size; 473 memcpy(*buf, tmp, len); 474 *size -= len; 475 *buf += len; 476 } 477 #undef TMPBUFLEN 478 479 static void proc_put_char(void **buf, size_t *size, char c) 480 { 481 if (*size) { 482 char **buffer = (char **)buf; 483 **buffer = c; 484 485 (*size)--; 486 (*buffer)++; 487 *buf = *buffer; 488 } 489 } 490 491 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp, 492 int *valp, 493 int write, void *data) 494 { 495 if (write) { 496 if (*negp) { 497 if (*lvalp > (unsigned long) INT_MAX + 1) 498 return -EINVAL; 499 *valp = -*lvalp; 500 } else { 501 if (*lvalp > (unsigned long) INT_MAX) 502 return -EINVAL; 503 *valp = *lvalp; 504 } 505 } else { 506 int val = *valp; 507 if (val < 0) { 508 *negp = true; 509 *lvalp = -(unsigned long)val; 510 } else { 511 *negp = false; 512 *lvalp = (unsigned long)val; 513 } 514 } 515 return 0; 516 } 517 518 static int do_proc_douintvec_conv(unsigned long *lvalp, 519 unsigned int *valp, 520 int write, void *data) 521 { 522 if (write) { 523 if (*lvalp > UINT_MAX) 524 return -EINVAL; 525 *valp = *lvalp; 526 } else { 527 unsigned int val = *valp; 528 *lvalp = (unsigned long)val; 529 } 530 return 0; 531 } 532 533 static const char proc_wspace_sep[] = { ' ', '\t', '\n' }; 534 535 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table, 536 int write, void *buffer, 537 size_t *lenp, loff_t *ppos, 538 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 539 int write, void *data), 540 void *data) 541 { 542 int *i, vleft, first = 1, err = 0; 543 size_t left; 544 char *p; 545 546 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) { 547 *lenp = 0; 548 return 0; 549 } 550 551 i = (int *) tbl_data; 552 vleft = table->maxlen / sizeof(*i); 553 left = *lenp; 554 555 if (!conv) 556 conv = do_proc_dointvec_conv; 557 558 if (write) { 559 if (proc_first_pos_non_zero_ignore(ppos, table)) 560 goto out; 561 562 if (left > PAGE_SIZE - 1) 563 left = PAGE_SIZE - 1; 564 p = buffer; 565 } 566 567 for (; left && vleft--; i++, first=0) { 568 unsigned long lval; 569 bool neg; 570 571 if (write) { 572 left -= proc_skip_spaces(&p); 573 574 if (!left) 575 break; 576 err = proc_get_long(&p, &left, &lval, &neg, 577 proc_wspace_sep, 578 sizeof(proc_wspace_sep), NULL); 579 if (err) 580 break; 581 if (conv(&neg, &lval, i, 1, data)) { 582 err = -EINVAL; 583 break; 584 } 585 } else { 586 if (conv(&neg, &lval, i, 0, data)) { 587 err = -EINVAL; 588 break; 589 } 590 if (!first) 591 proc_put_char(&buffer, &left, '\t'); 592 proc_put_long(&buffer, &left, lval, neg); 593 } 594 } 595 596 if (!write && !first && left && !err) 597 proc_put_char(&buffer, &left, '\n'); 598 if (write && !err && left) 599 left -= proc_skip_spaces(&p); 600 if (write && first) 601 return err ? : -EINVAL; 602 *lenp -= left; 603 out: 604 *ppos += *lenp; 605 return err; 606 } 607 608 static int do_proc_dointvec(struct ctl_table *table, int write, 609 void *buffer, size_t *lenp, loff_t *ppos, 610 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 611 int write, void *data), 612 void *data) 613 { 614 return __do_proc_dointvec(table->data, table, write, 615 buffer, lenp, ppos, conv, data); 616 } 617 618 static int do_proc_douintvec_w(unsigned int *tbl_data, 619 struct ctl_table *table, 620 void *buffer, 621 size_t *lenp, loff_t *ppos, 622 int (*conv)(unsigned long *lvalp, 623 unsigned int *valp, 624 int write, void *data), 625 void *data) 626 { 627 unsigned long lval; 628 int err = 0; 629 size_t left; 630 bool neg; 631 char *p = buffer; 632 633 left = *lenp; 634 635 if (proc_first_pos_non_zero_ignore(ppos, table)) 636 goto bail_early; 637 638 if (left > PAGE_SIZE - 1) 639 left = PAGE_SIZE - 1; 640 641 left -= proc_skip_spaces(&p); 642 if (!left) { 643 err = -EINVAL; 644 goto out_free; 645 } 646 647 err = proc_get_long(&p, &left, &lval, &neg, 648 proc_wspace_sep, 649 sizeof(proc_wspace_sep), NULL); 650 if (err || neg) { 651 err = -EINVAL; 652 goto out_free; 653 } 654 655 if (conv(&lval, tbl_data, 1, data)) { 656 err = -EINVAL; 657 goto out_free; 658 } 659 660 if (!err && left) 661 left -= proc_skip_spaces(&p); 662 663 out_free: 664 if (err) 665 return -EINVAL; 666 667 return 0; 668 669 /* This is in keeping with old __do_proc_dointvec() */ 670 bail_early: 671 *ppos += *lenp; 672 return err; 673 } 674 675 static int do_proc_douintvec_r(unsigned int *tbl_data, void *buffer, 676 size_t *lenp, loff_t *ppos, 677 int (*conv)(unsigned long *lvalp, 678 unsigned int *valp, 679 int write, void *data), 680 void *data) 681 { 682 unsigned long lval; 683 int err = 0; 684 size_t left; 685 686 left = *lenp; 687 688 if (conv(&lval, tbl_data, 0, data)) { 689 err = -EINVAL; 690 goto out; 691 } 692 693 proc_put_long(&buffer, &left, lval, false); 694 if (!left) 695 goto out; 696 697 proc_put_char(&buffer, &left, '\n'); 698 699 out: 700 *lenp -= left; 701 *ppos += *lenp; 702 703 return err; 704 } 705 706 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table, 707 int write, void *buffer, 708 size_t *lenp, loff_t *ppos, 709 int (*conv)(unsigned long *lvalp, 710 unsigned int *valp, 711 int write, void *data), 712 void *data) 713 { 714 unsigned int *i, vleft; 715 716 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) { 717 *lenp = 0; 718 return 0; 719 } 720 721 i = (unsigned int *) tbl_data; 722 vleft = table->maxlen / sizeof(*i); 723 724 /* 725 * Arrays are not supported, keep this simple. *Do not* add 726 * support for them. 727 */ 728 if (vleft != 1) { 729 *lenp = 0; 730 return -EINVAL; 731 } 732 733 if (!conv) 734 conv = do_proc_douintvec_conv; 735 736 if (write) 737 return do_proc_douintvec_w(i, table, buffer, lenp, ppos, 738 conv, data); 739 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data); 740 } 741 742 static int do_proc_douintvec(struct ctl_table *table, int write, 743 void *buffer, size_t *lenp, loff_t *ppos, 744 int (*conv)(unsigned long *lvalp, 745 unsigned int *valp, 746 int write, void *data), 747 void *data) 748 { 749 return __do_proc_douintvec(table->data, table, write, 750 buffer, lenp, ppos, conv, data); 751 } 752 753 /** 754 * proc_dointvec - read a vector of integers 755 * @table: the sysctl table 756 * @write: %TRUE if this is a write to the sysctl file 757 * @buffer: the user buffer 758 * @lenp: the size of the user buffer 759 * @ppos: file position 760 * 761 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 762 * values from/to the user buffer, treated as an ASCII string. 763 * 764 * Returns 0 on success. 765 */ 766 int proc_dointvec(struct ctl_table *table, int write, void *buffer, 767 size_t *lenp, loff_t *ppos) 768 { 769 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL); 770 } 771 772 #ifdef CONFIG_COMPACTION 773 static int proc_dointvec_minmax_warn_RT_change(struct ctl_table *table, 774 int write, void *buffer, size_t *lenp, loff_t *ppos) 775 { 776 int ret, old; 777 778 if (!IS_ENABLED(CONFIG_PREEMPT_RT) || !write) 779 return proc_dointvec_minmax(table, write, buffer, lenp, ppos); 780 781 old = *(int *)table->data; 782 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos); 783 if (ret) 784 return ret; 785 if (old != *(int *)table->data) 786 pr_warn_once("sysctl attribute %s changed by %s[%d]\n", 787 table->procname, current->comm, 788 task_pid_nr(current)); 789 return ret; 790 } 791 #endif 792 793 /** 794 * proc_douintvec - read a vector of unsigned integers 795 * @table: the sysctl table 796 * @write: %TRUE if this is a write to the sysctl file 797 * @buffer: the user buffer 798 * @lenp: the size of the user buffer 799 * @ppos: file position 800 * 801 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer 802 * values from/to the user buffer, treated as an ASCII string. 803 * 804 * Returns 0 on success. 805 */ 806 int proc_douintvec(struct ctl_table *table, int write, void *buffer, 807 size_t *lenp, loff_t *ppos) 808 { 809 return do_proc_douintvec(table, write, buffer, lenp, ppos, 810 do_proc_douintvec_conv, NULL); 811 } 812 813 /* 814 * Taint values can only be increased 815 * This means we can safely use a temporary. 816 */ 817 static int proc_taint(struct ctl_table *table, int write, 818 void *buffer, size_t *lenp, loff_t *ppos) 819 { 820 struct ctl_table t; 821 unsigned long tmptaint = get_taint(); 822 int err; 823 824 if (write && !capable(CAP_SYS_ADMIN)) 825 return -EPERM; 826 827 t = *table; 828 t.data = &tmptaint; 829 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos); 830 if (err < 0) 831 return err; 832 833 if (write) { 834 /* 835 * Poor man's atomic or. Not worth adding a primitive 836 * to everyone's atomic.h for this 837 */ 838 int i; 839 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) { 840 if ((tmptaint >> i) & 1) 841 add_taint(i, LOCKDEP_STILL_OK); 842 } 843 } 844 845 return err; 846 } 847 848 #ifdef CONFIG_PRINTK 849 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write, 850 void *buffer, size_t *lenp, loff_t *ppos) 851 { 852 if (write && !capable(CAP_SYS_ADMIN)) 853 return -EPERM; 854 855 return proc_dointvec_minmax(table, write, buffer, lenp, ppos); 856 } 857 #endif 858 859 /** 860 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure 861 * @min: pointer to minimum allowable value 862 * @max: pointer to maximum allowable value 863 * 864 * The do_proc_dointvec_minmax_conv_param structure provides the 865 * minimum and maximum values for doing range checking for those sysctl 866 * parameters that use the proc_dointvec_minmax() handler. 867 */ 868 struct do_proc_dointvec_minmax_conv_param { 869 int *min; 870 int *max; 871 }; 872 873 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp, 874 int *valp, 875 int write, void *data) 876 { 877 int tmp, ret; 878 struct do_proc_dointvec_minmax_conv_param *param = data; 879 /* 880 * If writing, first do so via a temporary local int so we can 881 * bounds-check it before touching *valp. 882 */ 883 int *ip = write ? &tmp : valp; 884 885 ret = do_proc_dointvec_conv(negp, lvalp, ip, write, data); 886 if (ret) 887 return ret; 888 889 if (write) { 890 if ((param->min && *param->min > tmp) || 891 (param->max && *param->max < tmp)) 892 return -EINVAL; 893 *valp = tmp; 894 } 895 896 return 0; 897 } 898 899 /** 900 * proc_dointvec_minmax - read a vector of integers with min/max values 901 * @table: the sysctl table 902 * @write: %TRUE if this is a write to the sysctl file 903 * @buffer: the user buffer 904 * @lenp: the size of the user buffer 905 * @ppos: file position 906 * 907 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 908 * values from/to the user buffer, treated as an ASCII string. 909 * 910 * This routine will ensure the values are within the range specified by 911 * table->extra1 (min) and table->extra2 (max). 912 * 913 * Returns 0 on success or -EINVAL on write when the range check fails. 914 */ 915 int proc_dointvec_minmax(struct ctl_table *table, int write, 916 void *buffer, size_t *lenp, loff_t *ppos) 917 { 918 struct do_proc_dointvec_minmax_conv_param param = { 919 .min = (int *) table->extra1, 920 .max = (int *) table->extra2, 921 }; 922 return do_proc_dointvec(table, write, buffer, lenp, ppos, 923 do_proc_dointvec_minmax_conv, ¶m); 924 } 925 926 /** 927 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure 928 * @min: pointer to minimum allowable value 929 * @max: pointer to maximum allowable value 930 * 931 * The do_proc_douintvec_minmax_conv_param structure provides the 932 * minimum and maximum values for doing range checking for those sysctl 933 * parameters that use the proc_douintvec_minmax() handler. 934 */ 935 struct do_proc_douintvec_minmax_conv_param { 936 unsigned int *min; 937 unsigned int *max; 938 }; 939 940 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp, 941 unsigned int *valp, 942 int write, void *data) 943 { 944 int ret; 945 unsigned int tmp; 946 struct do_proc_douintvec_minmax_conv_param *param = data; 947 /* write via temporary local uint for bounds-checking */ 948 unsigned int *up = write ? &tmp : valp; 949 950 ret = do_proc_douintvec_conv(lvalp, up, write, data); 951 if (ret) 952 return ret; 953 954 if (write) { 955 if ((param->min && *param->min > tmp) || 956 (param->max && *param->max < tmp)) 957 return -ERANGE; 958 959 *valp = tmp; 960 } 961 962 return 0; 963 } 964 965 /** 966 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values 967 * @table: the sysctl table 968 * @write: %TRUE if this is a write to the sysctl file 969 * @buffer: the user buffer 970 * @lenp: the size of the user buffer 971 * @ppos: file position 972 * 973 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer 974 * values from/to the user buffer, treated as an ASCII string. Negative 975 * strings are not allowed. 976 * 977 * This routine will ensure the values are within the range specified by 978 * table->extra1 (min) and table->extra2 (max). There is a final sanity 979 * check for UINT_MAX to avoid having to support wrap around uses from 980 * userspace. 981 * 982 * Returns 0 on success or -ERANGE on write when the range check fails. 983 */ 984 int proc_douintvec_minmax(struct ctl_table *table, int write, 985 void *buffer, size_t *lenp, loff_t *ppos) 986 { 987 struct do_proc_douintvec_minmax_conv_param param = { 988 .min = (unsigned int *) table->extra1, 989 .max = (unsigned int *) table->extra2, 990 }; 991 return do_proc_douintvec(table, write, buffer, lenp, ppos, 992 do_proc_douintvec_minmax_conv, ¶m); 993 } 994 995 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp, 996 unsigned int *valp, 997 int write, void *data) 998 { 999 if (write) { 1000 unsigned int val; 1001 1002 val = round_pipe_size(*lvalp); 1003 if (val == 0) 1004 return -EINVAL; 1005 1006 *valp = val; 1007 } else { 1008 unsigned int val = *valp; 1009 *lvalp = (unsigned long) val; 1010 } 1011 1012 return 0; 1013 } 1014 1015 static int proc_dopipe_max_size(struct ctl_table *table, int write, 1016 void *buffer, size_t *lenp, loff_t *ppos) 1017 { 1018 return do_proc_douintvec(table, write, buffer, lenp, ppos, 1019 do_proc_dopipe_max_size_conv, NULL); 1020 } 1021 1022 static void validate_coredump_safety(void) 1023 { 1024 #ifdef CONFIG_COREDUMP 1025 if (suid_dumpable == SUID_DUMP_ROOT && 1026 core_pattern[0] != '/' && core_pattern[0] != '|') { 1027 printk(KERN_WARNING 1028 "Unsafe core_pattern used with fs.suid_dumpable=2.\n" 1029 "Pipe handler or fully qualified core dump path required.\n" 1030 "Set kernel.core_pattern before fs.suid_dumpable.\n" 1031 ); 1032 } 1033 #endif 1034 } 1035 1036 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write, 1037 void *buffer, size_t *lenp, loff_t *ppos) 1038 { 1039 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos); 1040 if (!error) 1041 validate_coredump_safety(); 1042 return error; 1043 } 1044 1045 #ifdef CONFIG_COREDUMP 1046 static int proc_dostring_coredump(struct ctl_table *table, int write, 1047 void *buffer, size_t *lenp, loff_t *ppos) 1048 { 1049 int error = proc_dostring(table, write, buffer, lenp, ppos); 1050 if (!error) 1051 validate_coredump_safety(); 1052 return error; 1053 } 1054 #endif 1055 1056 #ifdef CONFIG_MAGIC_SYSRQ 1057 static int sysrq_sysctl_handler(struct ctl_table *table, int write, 1058 void *buffer, size_t *lenp, loff_t *ppos) 1059 { 1060 int tmp, ret; 1061 1062 tmp = sysrq_mask(); 1063 1064 ret = __do_proc_dointvec(&tmp, table, write, buffer, 1065 lenp, ppos, NULL, NULL); 1066 if (ret || !write) 1067 return ret; 1068 1069 if (write) 1070 sysrq_toggle_support(tmp); 1071 1072 return 0; 1073 } 1074 #endif 1075 1076 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, 1077 int write, void *buffer, size_t *lenp, loff_t *ppos, 1078 unsigned long convmul, unsigned long convdiv) 1079 { 1080 unsigned long *i, *min, *max; 1081 int vleft, first = 1, err = 0; 1082 size_t left; 1083 char *p; 1084 1085 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) { 1086 *lenp = 0; 1087 return 0; 1088 } 1089 1090 i = (unsigned long *) data; 1091 min = (unsigned long *) table->extra1; 1092 max = (unsigned long *) table->extra2; 1093 vleft = table->maxlen / sizeof(unsigned long); 1094 left = *lenp; 1095 1096 if (write) { 1097 if (proc_first_pos_non_zero_ignore(ppos, table)) 1098 goto out; 1099 1100 if (left > PAGE_SIZE - 1) 1101 left = PAGE_SIZE - 1; 1102 p = buffer; 1103 } 1104 1105 for (; left && vleft--; i++, first = 0) { 1106 unsigned long val; 1107 1108 if (write) { 1109 bool neg; 1110 1111 left -= proc_skip_spaces(&p); 1112 if (!left) 1113 break; 1114 1115 err = proc_get_long(&p, &left, &val, &neg, 1116 proc_wspace_sep, 1117 sizeof(proc_wspace_sep), NULL); 1118 if (err) 1119 break; 1120 if (neg) 1121 continue; 1122 val = convmul * val / convdiv; 1123 if ((min && val < *min) || (max && val > *max)) { 1124 err = -EINVAL; 1125 break; 1126 } 1127 *i = val; 1128 } else { 1129 val = convdiv * (*i) / convmul; 1130 if (!first) 1131 proc_put_char(&buffer, &left, '\t'); 1132 proc_put_long(&buffer, &left, val, false); 1133 } 1134 } 1135 1136 if (!write && !first && left && !err) 1137 proc_put_char(&buffer, &left, '\n'); 1138 if (write && !err) 1139 left -= proc_skip_spaces(&p); 1140 if (write && first) 1141 return err ? : -EINVAL; 1142 *lenp -= left; 1143 out: 1144 *ppos += *lenp; 1145 return err; 1146 } 1147 1148 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write, 1149 void *buffer, size_t *lenp, loff_t *ppos, unsigned long convmul, 1150 unsigned long convdiv) 1151 { 1152 return __do_proc_doulongvec_minmax(table->data, table, write, 1153 buffer, lenp, ppos, convmul, convdiv); 1154 } 1155 1156 /** 1157 * proc_doulongvec_minmax - read a vector of long integers with min/max values 1158 * @table: the sysctl table 1159 * @write: %TRUE if this is a write to the sysctl file 1160 * @buffer: the user buffer 1161 * @lenp: the size of the user buffer 1162 * @ppos: file position 1163 * 1164 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 1165 * values from/to the user buffer, treated as an ASCII string. 1166 * 1167 * This routine will ensure the values are within the range specified by 1168 * table->extra1 (min) and table->extra2 (max). 1169 * 1170 * Returns 0 on success. 1171 */ 1172 int proc_doulongvec_minmax(struct ctl_table *table, int write, 1173 void *buffer, size_t *lenp, loff_t *ppos) 1174 { 1175 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l); 1176 } 1177 1178 /** 1179 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values 1180 * @table: the sysctl table 1181 * @write: %TRUE if this is a write to the sysctl file 1182 * @buffer: the user buffer 1183 * @lenp: the size of the user buffer 1184 * @ppos: file position 1185 * 1186 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 1187 * values from/to the user buffer, treated as an ASCII string. The values 1188 * are treated as milliseconds, and converted to jiffies when they are stored. 1189 * 1190 * This routine will ensure the values are within the range specified by 1191 * table->extra1 (min) and table->extra2 (max). 1192 * 1193 * Returns 0 on success. 1194 */ 1195 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 1196 void *buffer, size_t *lenp, loff_t *ppos) 1197 { 1198 return do_proc_doulongvec_minmax(table, write, buffer, 1199 lenp, ppos, HZ, 1000l); 1200 } 1201 1202 1203 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp, 1204 int *valp, 1205 int write, void *data) 1206 { 1207 if (write) { 1208 if (*lvalp > INT_MAX / HZ) 1209 return 1; 1210 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ); 1211 } else { 1212 int val = *valp; 1213 unsigned long lval; 1214 if (val < 0) { 1215 *negp = true; 1216 lval = -(unsigned long)val; 1217 } else { 1218 *negp = false; 1219 lval = (unsigned long)val; 1220 } 1221 *lvalp = lval / HZ; 1222 } 1223 return 0; 1224 } 1225 1226 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp, 1227 int *valp, 1228 int write, void *data) 1229 { 1230 if (write) { 1231 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ) 1232 return 1; 1233 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp); 1234 } else { 1235 int val = *valp; 1236 unsigned long lval; 1237 if (val < 0) { 1238 *negp = true; 1239 lval = -(unsigned long)val; 1240 } else { 1241 *negp = false; 1242 lval = (unsigned long)val; 1243 } 1244 *lvalp = jiffies_to_clock_t(lval); 1245 } 1246 return 0; 1247 } 1248 1249 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp, 1250 int *valp, 1251 int write, void *data) 1252 { 1253 if (write) { 1254 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp); 1255 1256 if (jif > INT_MAX) 1257 return 1; 1258 *valp = (int)jif; 1259 } else { 1260 int val = *valp; 1261 unsigned long lval; 1262 if (val < 0) { 1263 *negp = true; 1264 lval = -(unsigned long)val; 1265 } else { 1266 *negp = false; 1267 lval = (unsigned long)val; 1268 } 1269 *lvalp = jiffies_to_msecs(lval); 1270 } 1271 return 0; 1272 } 1273 1274 /** 1275 * proc_dointvec_jiffies - read a vector of integers as seconds 1276 * @table: the sysctl table 1277 * @write: %TRUE if this is a write to the sysctl file 1278 * @buffer: the user buffer 1279 * @lenp: the size of the user buffer 1280 * @ppos: file position 1281 * 1282 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 1283 * values from/to the user buffer, treated as an ASCII string. 1284 * The values read are assumed to be in seconds, and are converted into 1285 * jiffies. 1286 * 1287 * Returns 0 on success. 1288 */ 1289 int proc_dointvec_jiffies(struct ctl_table *table, int write, 1290 void *buffer, size_t *lenp, loff_t *ppos) 1291 { 1292 return do_proc_dointvec(table,write,buffer,lenp,ppos, 1293 do_proc_dointvec_jiffies_conv,NULL); 1294 } 1295 1296 /** 1297 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds 1298 * @table: the sysctl table 1299 * @write: %TRUE if this is a write to the sysctl file 1300 * @buffer: the user buffer 1301 * @lenp: the size of the user buffer 1302 * @ppos: pointer to the file position 1303 * 1304 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 1305 * values from/to the user buffer, treated as an ASCII string. 1306 * The values read are assumed to be in 1/USER_HZ seconds, and 1307 * are converted into jiffies. 1308 * 1309 * Returns 0 on success. 1310 */ 1311 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 1312 void *buffer, size_t *lenp, loff_t *ppos) 1313 { 1314 return do_proc_dointvec(table,write,buffer,lenp,ppos, 1315 do_proc_dointvec_userhz_jiffies_conv,NULL); 1316 } 1317 1318 /** 1319 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds 1320 * @table: the sysctl table 1321 * @write: %TRUE if this is a write to the sysctl file 1322 * @buffer: the user buffer 1323 * @lenp: the size of the user buffer 1324 * @ppos: file position 1325 * @ppos: the current position in the file 1326 * 1327 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 1328 * values from/to the user buffer, treated as an ASCII string. 1329 * The values read are assumed to be in 1/1000 seconds, and 1330 * are converted into jiffies. 1331 * 1332 * Returns 0 on success. 1333 */ 1334 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, void *buffer, 1335 size_t *lenp, loff_t *ppos) 1336 { 1337 return do_proc_dointvec(table, write, buffer, lenp, ppos, 1338 do_proc_dointvec_ms_jiffies_conv, NULL); 1339 } 1340 1341 static int proc_do_cad_pid(struct ctl_table *table, int write, void *buffer, 1342 size_t *lenp, loff_t *ppos) 1343 { 1344 struct pid *new_pid; 1345 pid_t tmp; 1346 int r; 1347 1348 tmp = pid_vnr(cad_pid); 1349 1350 r = __do_proc_dointvec(&tmp, table, write, buffer, 1351 lenp, ppos, NULL, NULL); 1352 if (r || !write) 1353 return r; 1354 1355 new_pid = find_get_pid(tmp); 1356 if (!new_pid) 1357 return -ESRCH; 1358 1359 put_pid(xchg(&cad_pid, new_pid)); 1360 return 0; 1361 } 1362 1363 /** 1364 * proc_do_large_bitmap - read/write from/to a large bitmap 1365 * @table: the sysctl table 1366 * @write: %TRUE if this is a write to the sysctl file 1367 * @buffer: the user buffer 1368 * @lenp: the size of the user buffer 1369 * @ppos: file position 1370 * 1371 * The bitmap is stored at table->data and the bitmap length (in bits) 1372 * in table->maxlen. 1373 * 1374 * We use a range comma separated format (e.g. 1,3-4,10-10) so that 1375 * large bitmaps may be represented in a compact manner. Writing into 1376 * the file will clear the bitmap then update it with the given input. 1377 * 1378 * Returns 0 on success. 1379 */ 1380 int proc_do_large_bitmap(struct ctl_table *table, int write, 1381 void *buffer, size_t *lenp, loff_t *ppos) 1382 { 1383 int err = 0; 1384 bool first = 1; 1385 size_t left = *lenp; 1386 unsigned long bitmap_len = table->maxlen; 1387 unsigned long *bitmap = *(unsigned long **) table->data; 1388 unsigned long *tmp_bitmap = NULL; 1389 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c; 1390 1391 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) { 1392 *lenp = 0; 1393 return 0; 1394 } 1395 1396 if (write) { 1397 char *p = buffer; 1398 size_t skipped = 0; 1399 1400 if (left > PAGE_SIZE - 1) { 1401 left = PAGE_SIZE - 1; 1402 /* How much of the buffer we'll skip this pass */ 1403 skipped = *lenp - left; 1404 } 1405 1406 tmp_bitmap = bitmap_zalloc(bitmap_len, GFP_KERNEL); 1407 if (!tmp_bitmap) 1408 return -ENOMEM; 1409 proc_skip_char(&p, &left, '\n'); 1410 while (!err && left) { 1411 unsigned long val_a, val_b; 1412 bool neg; 1413 size_t saved_left; 1414 1415 /* In case we stop parsing mid-number, we can reset */ 1416 saved_left = left; 1417 err = proc_get_long(&p, &left, &val_a, &neg, tr_a, 1418 sizeof(tr_a), &c); 1419 /* 1420 * If we consumed the entirety of a truncated buffer or 1421 * only one char is left (may be a "-"), then stop here, 1422 * reset, & come back for more. 1423 */ 1424 if ((left <= 1) && skipped) { 1425 left = saved_left; 1426 break; 1427 } 1428 1429 if (err) 1430 break; 1431 if (val_a >= bitmap_len || neg) { 1432 err = -EINVAL; 1433 break; 1434 } 1435 1436 val_b = val_a; 1437 if (left) { 1438 p++; 1439 left--; 1440 } 1441 1442 if (c == '-') { 1443 err = proc_get_long(&p, &left, &val_b, 1444 &neg, tr_b, sizeof(tr_b), 1445 &c); 1446 /* 1447 * If we consumed all of a truncated buffer or 1448 * then stop here, reset, & come back for more. 1449 */ 1450 if (!left && skipped) { 1451 left = saved_left; 1452 break; 1453 } 1454 1455 if (err) 1456 break; 1457 if (val_b >= bitmap_len || neg || 1458 val_a > val_b) { 1459 err = -EINVAL; 1460 break; 1461 } 1462 if (left) { 1463 p++; 1464 left--; 1465 } 1466 } 1467 1468 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1); 1469 first = 0; 1470 proc_skip_char(&p, &left, '\n'); 1471 } 1472 left += skipped; 1473 } else { 1474 unsigned long bit_a, bit_b = 0; 1475 1476 while (left) { 1477 bit_a = find_next_bit(bitmap, bitmap_len, bit_b); 1478 if (bit_a >= bitmap_len) 1479 break; 1480 bit_b = find_next_zero_bit(bitmap, bitmap_len, 1481 bit_a + 1) - 1; 1482 1483 if (!first) 1484 proc_put_char(&buffer, &left, ','); 1485 proc_put_long(&buffer, &left, bit_a, false); 1486 if (bit_a != bit_b) { 1487 proc_put_char(&buffer, &left, '-'); 1488 proc_put_long(&buffer, &left, bit_b, false); 1489 } 1490 1491 first = 0; bit_b++; 1492 } 1493 proc_put_char(&buffer, &left, '\n'); 1494 } 1495 1496 if (!err) { 1497 if (write) { 1498 if (*ppos) 1499 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len); 1500 else 1501 bitmap_copy(bitmap, tmp_bitmap, bitmap_len); 1502 } 1503 *lenp -= left; 1504 *ppos += *lenp; 1505 } 1506 1507 bitmap_free(tmp_bitmap); 1508 return err; 1509 } 1510 1511 #else /* CONFIG_PROC_SYSCTL */ 1512 1513 int proc_dostring(struct ctl_table *table, int write, 1514 void *buffer, size_t *lenp, loff_t *ppos) 1515 { 1516 return -ENOSYS; 1517 } 1518 1519 int proc_dointvec(struct ctl_table *table, int write, 1520 void *buffer, size_t *lenp, loff_t *ppos) 1521 { 1522 return -ENOSYS; 1523 } 1524 1525 int proc_douintvec(struct ctl_table *table, int write, 1526 void *buffer, size_t *lenp, loff_t *ppos) 1527 { 1528 return -ENOSYS; 1529 } 1530 1531 int proc_dointvec_minmax(struct ctl_table *table, int write, 1532 void *buffer, size_t *lenp, loff_t *ppos) 1533 { 1534 return -ENOSYS; 1535 } 1536 1537 int proc_douintvec_minmax(struct ctl_table *table, int write, 1538 void *buffer, size_t *lenp, loff_t *ppos) 1539 { 1540 return -ENOSYS; 1541 } 1542 1543 int proc_dointvec_jiffies(struct ctl_table *table, int write, 1544 void *buffer, size_t *lenp, loff_t *ppos) 1545 { 1546 return -ENOSYS; 1547 } 1548 1549 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 1550 void *buffer, size_t *lenp, loff_t *ppos) 1551 { 1552 return -ENOSYS; 1553 } 1554 1555 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, 1556 void *buffer, size_t *lenp, loff_t *ppos) 1557 { 1558 return -ENOSYS; 1559 } 1560 1561 int proc_doulongvec_minmax(struct ctl_table *table, int write, 1562 void *buffer, size_t *lenp, loff_t *ppos) 1563 { 1564 return -ENOSYS; 1565 } 1566 1567 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 1568 void *buffer, size_t *lenp, loff_t *ppos) 1569 { 1570 return -ENOSYS; 1571 } 1572 1573 int proc_do_large_bitmap(struct ctl_table *table, int write, 1574 void *buffer, size_t *lenp, loff_t *ppos) 1575 { 1576 return -ENOSYS; 1577 } 1578 1579 #endif /* CONFIG_PROC_SYSCTL */ 1580 1581 #if defined(CONFIG_SYSCTL) 1582 int proc_do_static_key(struct ctl_table *table, int write, 1583 void *buffer, size_t *lenp, loff_t *ppos) 1584 { 1585 struct static_key *key = (struct static_key *)table->data; 1586 static DEFINE_MUTEX(static_key_mutex); 1587 int val, ret; 1588 struct ctl_table tmp = { 1589 .data = &val, 1590 .maxlen = sizeof(val), 1591 .mode = table->mode, 1592 .extra1 = SYSCTL_ZERO, 1593 .extra2 = SYSCTL_ONE, 1594 }; 1595 1596 if (write && !capable(CAP_SYS_ADMIN)) 1597 return -EPERM; 1598 1599 mutex_lock(&static_key_mutex); 1600 val = static_key_enabled(key); 1601 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos); 1602 if (write && !ret) { 1603 if (val) 1604 static_key_enable(key); 1605 else 1606 static_key_disable(key); 1607 } 1608 mutex_unlock(&static_key_mutex); 1609 return ret; 1610 } 1611 1612 static struct ctl_table kern_table[] = { 1613 { 1614 .procname = "sched_child_runs_first", 1615 .data = &sysctl_sched_child_runs_first, 1616 .maxlen = sizeof(unsigned int), 1617 .mode = 0644, 1618 .proc_handler = proc_dointvec, 1619 }, 1620 #ifdef CONFIG_SCHED_DEBUG 1621 { 1622 .procname = "sched_min_granularity_ns", 1623 .data = &sysctl_sched_min_granularity, 1624 .maxlen = sizeof(unsigned int), 1625 .mode = 0644, 1626 .proc_handler = sched_proc_update_handler, 1627 .extra1 = &min_sched_granularity_ns, 1628 .extra2 = &max_sched_granularity_ns, 1629 }, 1630 { 1631 .procname = "sched_latency_ns", 1632 .data = &sysctl_sched_latency, 1633 .maxlen = sizeof(unsigned int), 1634 .mode = 0644, 1635 .proc_handler = sched_proc_update_handler, 1636 .extra1 = &min_sched_granularity_ns, 1637 .extra2 = &max_sched_granularity_ns, 1638 }, 1639 { 1640 .procname = "sched_wakeup_granularity_ns", 1641 .data = &sysctl_sched_wakeup_granularity, 1642 .maxlen = sizeof(unsigned int), 1643 .mode = 0644, 1644 .proc_handler = sched_proc_update_handler, 1645 .extra1 = &min_wakeup_granularity_ns, 1646 .extra2 = &max_wakeup_granularity_ns, 1647 }, 1648 #ifdef CONFIG_SMP 1649 { 1650 .procname = "sched_tunable_scaling", 1651 .data = &sysctl_sched_tunable_scaling, 1652 .maxlen = sizeof(enum sched_tunable_scaling), 1653 .mode = 0644, 1654 .proc_handler = sched_proc_update_handler, 1655 .extra1 = &min_sched_tunable_scaling, 1656 .extra2 = &max_sched_tunable_scaling, 1657 }, 1658 { 1659 .procname = "sched_migration_cost_ns", 1660 .data = &sysctl_sched_migration_cost, 1661 .maxlen = sizeof(unsigned int), 1662 .mode = 0644, 1663 .proc_handler = proc_dointvec, 1664 }, 1665 { 1666 .procname = "sched_nr_migrate", 1667 .data = &sysctl_sched_nr_migrate, 1668 .maxlen = sizeof(unsigned int), 1669 .mode = 0644, 1670 .proc_handler = proc_dointvec, 1671 }, 1672 #ifdef CONFIG_SCHEDSTATS 1673 { 1674 .procname = "sched_schedstats", 1675 .data = NULL, 1676 .maxlen = sizeof(unsigned int), 1677 .mode = 0644, 1678 .proc_handler = sysctl_schedstats, 1679 .extra1 = SYSCTL_ZERO, 1680 .extra2 = SYSCTL_ONE, 1681 }, 1682 #endif /* CONFIG_SCHEDSTATS */ 1683 #endif /* CONFIG_SMP */ 1684 #ifdef CONFIG_NUMA_BALANCING 1685 { 1686 .procname = "numa_balancing_scan_delay_ms", 1687 .data = &sysctl_numa_balancing_scan_delay, 1688 .maxlen = sizeof(unsigned int), 1689 .mode = 0644, 1690 .proc_handler = proc_dointvec, 1691 }, 1692 { 1693 .procname = "numa_balancing_scan_period_min_ms", 1694 .data = &sysctl_numa_balancing_scan_period_min, 1695 .maxlen = sizeof(unsigned int), 1696 .mode = 0644, 1697 .proc_handler = proc_dointvec, 1698 }, 1699 { 1700 .procname = "numa_balancing_scan_period_max_ms", 1701 .data = &sysctl_numa_balancing_scan_period_max, 1702 .maxlen = sizeof(unsigned int), 1703 .mode = 0644, 1704 .proc_handler = proc_dointvec, 1705 }, 1706 { 1707 .procname = "numa_balancing_scan_size_mb", 1708 .data = &sysctl_numa_balancing_scan_size, 1709 .maxlen = sizeof(unsigned int), 1710 .mode = 0644, 1711 .proc_handler = proc_dointvec_minmax, 1712 .extra1 = SYSCTL_ONE, 1713 }, 1714 { 1715 .procname = "numa_balancing", 1716 .data = NULL, /* filled in by handler */ 1717 .maxlen = sizeof(unsigned int), 1718 .mode = 0644, 1719 .proc_handler = sysctl_numa_balancing, 1720 .extra1 = SYSCTL_ZERO, 1721 .extra2 = SYSCTL_ONE, 1722 }, 1723 #endif /* CONFIG_NUMA_BALANCING */ 1724 #endif /* CONFIG_SCHED_DEBUG */ 1725 { 1726 .procname = "sched_rt_period_us", 1727 .data = &sysctl_sched_rt_period, 1728 .maxlen = sizeof(unsigned int), 1729 .mode = 0644, 1730 .proc_handler = sched_rt_handler, 1731 }, 1732 { 1733 .procname = "sched_rt_runtime_us", 1734 .data = &sysctl_sched_rt_runtime, 1735 .maxlen = sizeof(int), 1736 .mode = 0644, 1737 .proc_handler = sched_rt_handler, 1738 }, 1739 { 1740 .procname = "sched_rr_timeslice_ms", 1741 .data = &sysctl_sched_rr_timeslice, 1742 .maxlen = sizeof(int), 1743 .mode = 0644, 1744 .proc_handler = sched_rr_handler, 1745 }, 1746 #ifdef CONFIG_UCLAMP_TASK 1747 { 1748 .procname = "sched_util_clamp_min", 1749 .data = &sysctl_sched_uclamp_util_min, 1750 .maxlen = sizeof(unsigned int), 1751 .mode = 0644, 1752 .proc_handler = sysctl_sched_uclamp_handler, 1753 }, 1754 { 1755 .procname = "sched_util_clamp_max", 1756 .data = &sysctl_sched_uclamp_util_max, 1757 .maxlen = sizeof(unsigned int), 1758 .mode = 0644, 1759 .proc_handler = sysctl_sched_uclamp_handler, 1760 }, 1761 #endif 1762 #ifdef CONFIG_SCHED_AUTOGROUP 1763 { 1764 .procname = "sched_autogroup_enabled", 1765 .data = &sysctl_sched_autogroup_enabled, 1766 .maxlen = sizeof(unsigned int), 1767 .mode = 0644, 1768 .proc_handler = proc_dointvec_minmax, 1769 .extra1 = SYSCTL_ZERO, 1770 .extra2 = SYSCTL_ONE, 1771 }, 1772 #endif 1773 #ifdef CONFIG_CFS_BANDWIDTH 1774 { 1775 .procname = "sched_cfs_bandwidth_slice_us", 1776 .data = &sysctl_sched_cfs_bandwidth_slice, 1777 .maxlen = sizeof(unsigned int), 1778 .mode = 0644, 1779 .proc_handler = proc_dointvec_minmax, 1780 .extra1 = SYSCTL_ONE, 1781 }, 1782 #endif 1783 #if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_CPU_FREQ_GOV_SCHEDUTIL) 1784 { 1785 .procname = "sched_energy_aware", 1786 .data = &sysctl_sched_energy_aware, 1787 .maxlen = sizeof(unsigned int), 1788 .mode = 0644, 1789 .proc_handler = sched_energy_aware_handler, 1790 .extra1 = SYSCTL_ZERO, 1791 .extra2 = SYSCTL_ONE, 1792 }, 1793 #endif 1794 #ifdef CONFIG_PROVE_LOCKING 1795 { 1796 .procname = "prove_locking", 1797 .data = &prove_locking, 1798 .maxlen = sizeof(int), 1799 .mode = 0644, 1800 .proc_handler = proc_dointvec, 1801 }, 1802 #endif 1803 #ifdef CONFIG_LOCK_STAT 1804 { 1805 .procname = "lock_stat", 1806 .data = &lock_stat, 1807 .maxlen = sizeof(int), 1808 .mode = 0644, 1809 .proc_handler = proc_dointvec, 1810 }, 1811 #endif 1812 { 1813 .procname = "panic", 1814 .data = &panic_timeout, 1815 .maxlen = sizeof(int), 1816 .mode = 0644, 1817 .proc_handler = proc_dointvec, 1818 }, 1819 #ifdef CONFIG_COREDUMP 1820 { 1821 .procname = "core_uses_pid", 1822 .data = &core_uses_pid, 1823 .maxlen = sizeof(int), 1824 .mode = 0644, 1825 .proc_handler = proc_dointvec, 1826 }, 1827 { 1828 .procname = "core_pattern", 1829 .data = core_pattern, 1830 .maxlen = CORENAME_MAX_SIZE, 1831 .mode = 0644, 1832 .proc_handler = proc_dostring_coredump, 1833 }, 1834 { 1835 .procname = "core_pipe_limit", 1836 .data = &core_pipe_limit, 1837 .maxlen = sizeof(unsigned int), 1838 .mode = 0644, 1839 .proc_handler = proc_dointvec, 1840 }, 1841 #endif 1842 #ifdef CONFIG_PROC_SYSCTL 1843 { 1844 .procname = "tainted", 1845 .maxlen = sizeof(long), 1846 .mode = 0644, 1847 .proc_handler = proc_taint, 1848 }, 1849 { 1850 .procname = "sysctl_writes_strict", 1851 .data = &sysctl_writes_strict, 1852 .maxlen = sizeof(int), 1853 .mode = 0644, 1854 .proc_handler = proc_dointvec_minmax, 1855 .extra1 = &neg_one, 1856 .extra2 = SYSCTL_ONE, 1857 }, 1858 #endif 1859 #ifdef CONFIG_LATENCYTOP 1860 { 1861 .procname = "latencytop", 1862 .data = &latencytop_enabled, 1863 .maxlen = sizeof(int), 1864 .mode = 0644, 1865 .proc_handler = sysctl_latencytop, 1866 }, 1867 #endif 1868 #ifdef CONFIG_BLK_DEV_INITRD 1869 { 1870 .procname = "real-root-dev", 1871 .data = &real_root_dev, 1872 .maxlen = sizeof(int), 1873 .mode = 0644, 1874 .proc_handler = proc_dointvec, 1875 }, 1876 #endif 1877 { 1878 .procname = "print-fatal-signals", 1879 .data = &print_fatal_signals, 1880 .maxlen = sizeof(int), 1881 .mode = 0644, 1882 .proc_handler = proc_dointvec, 1883 }, 1884 #ifdef CONFIG_SPARC 1885 { 1886 .procname = "reboot-cmd", 1887 .data = reboot_command, 1888 .maxlen = 256, 1889 .mode = 0644, 1890 .proc_handler = proc_dostring, 1891 }, 1892 { 1893 .procname = "stop-a", 1894 .data = &stop_a_enabled, 1895 .maxlen = sizeof (int), 1896 .mode = 0644, 1897 .proc_handler = proc_dointvec, 1898 }, 1899 { 1900 .procname = "scons-poweroff", 1901 .data = &scons_pwroff, 1902 .maxlen = sizeof (int), 1903 .mode = 0644, 1904 .proc_handler = proc_dointvec, 1905 }, 1906 #endif 1907 #ifdef CONFIG_SPARC64 1908 { 1909 .procname = "tsb-ratio", 1910 .data = &sysctl_tsb_ratio, 1911 .maxlen = sizeof (int), 1912 .mode = 0644, 1913 .proc_handler = proc_dointvec, 1914 }, 1915 #endif 1916 #ifdef CONFIG_PARISC 1917 { 1918 .procname = "soft-power", 1919 .data = &pwrsw_enabled, 1920 .maxlen = sizeof (int), 1921 .mode = 0644, 1922 .proc_handler = proc_dointvec, 1923 }, 1924 #endif 1925 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW 1926 { 1927 .procname = "unaligned-trap", 1928 .data = &unaligned_enabled, 1929 .maxlen = sizeof (int), 1930 .mode = 0644, 1931 .proc_handler = proc_dointvec, 1932 }, 1933 #endif 1934 { 1935 .procname = "ctrl-alt-del", 1936 .data = &C_A_D, 1937 .maxlen = sizeof(int), 1938 .mode = 0644, 1939 .proc_handler = proc_dointvec, 1940 }, 1941 #ifdef CONFIG_FUNCTION_TRACER 1942 { 1943 .procname = "ftrace_enabled", 1944 .data = &ftrace_enabled, 1945 .maxlen = sizeof(int), 1946 .mode = 0644, 1947 .proc_handler = ftrace_enable_sysctl, 1948 }, 1949 #endif 1950 #ifdef CONFIG_STACK_TRACER 1951 { 1952 .procname = "stack_tracer_enabled", 1953 .data = &stack_tracer_enabled, 1954 .maxlen = sizeof(int), 1955 .mode = 0644, 1956 .proc_handler = stack_trace_sysctl, 1957 }, 1958 #endif 1959 #ifdef CONFIG_TRACING 1960 { 1961 .procname = "ftrace_dump_on_oops", 1962 .data = &ftrace_dump_on_oops, 1963 .maxlen = sizeof(int), 1964 .mode = 0644, 1965 .proc_handler = proc_dointvec, 1966 }, 1967 { 1968 .procname = "traceoff_on_warning", 1969 .data = &__disable_trace_on_warning, 1970 .maxlen = sizeof(__disable_trace_on_warning), 1971 .mode = 0644, 1972 .proc_handler = proc_dointvec, 1973 }, 1974 { 1975 .procname = "tracepoint_printk", 1976 .data = &tracepoint_printk, 1977 .maxlen = sizeof(tracepoint_printk), 1978 .mode = 0644, 1979 .proc_handler = tracepoint_printk_sysctl, 1980 }, 1981 #endif 1982 #ifdef CONFIG_KEXEC_CORE 1983 { 1984 .procname = "kexec_load_disabled", 1985 .data = &kexec_load_disabled, 1986 .maxlen = sizeof(int), 1987 .mode = 0644, 1988 /* only handle a transition from default "0" to "1" */ 1989 .proc_handler = proc_dointvec_minmax, 1990 .extra1 = SYSCTL_ONE, 1991 .extra2 = SYSCTL_ONE, 1992 }, 1993 #endif 1994 #ifdef CONFIG_MODULES 1995 { 1996 .procname = "modprobe", 1997 .data = &modprobe_path, 1998 .maxlen = KMOD_PATH_LEN, 1999 .mode = 0644, 2000 .proc_handler = proc_dostring, 2001 }, 2002 { 2003 .procname = "modules_disabled", 2004 .data = &modules_disabled, 2005 .maxlen = sizeof(int), 2006 .mode = 0644, 2007 /* only handle a transition from default "0" to "1" */ 2008 .proc_handler = proc_dointvec_minmax, 2009 .extra1 = SYSCTL_ONE, 2010 .extra2 = SYSCTL_ONE, 2011 }, 2012 #endif 2013 #ifdef CONFIG_UEVENT_HELPER 2014 { 2015 .procname = "hotplug", 2016 .data = &uevent_helper, 2017 .maxlen = UEVENT_HELPER_PATH_LEN, 2018 .mode = 0644, 2019 .proc_handler = proc_dostring, 2020 }, 2021 #endif 2022 #ifdef CONFIG_CHR_DEV_SG 2023 { 2024 .procname = "sg-big-buff", 2025 .data = &sg_big_buff, 2026 .maxlen = sizeof (int), 2027 .mode = 0444, 2028 .proc_handler = proc_dointvec, 2029 }, 2030 #endif 2031 #ifdef CONFIG_BSD_PROCESS_ACCT 2032 { 2033 .procname = "acct", 2034 .data = &acct_parm, 2035 .maxlen = 3*sizeof(int), 2036 .mode = 0644, 2037 .proc_handler = proc_dointvec, 2038 }, 2039 #endif 2040 #ifdef CONFIG_MAGIC_SYSRQ 2041 { 2042 .procname = "sysrq", 2043 .data = NULL, 2044 .maxlen = sizeof (int), 2045 .mode = 0644, 2046 .proc_handler = sysrq_sysctl_handler, 2047 }, 2048 #endif 2049 #ifdef CONFIG_PROC_SYSCTL 2050 { 2051 .procname = "cad_pid", 2052 .data = NULL, 2053 .maxlen = sizeof (int), 2054 .mode = 0600, 2055 .proc_handler = proc_do_cad_pid, 2056 }, 2057 #endif 2058 { 2059 .procname = "threads-max", 2060 .data = NULL, 2061 .maxlen = sizeof(int), 2062 .mode = 0644, 2063 .proc_handler = sysctl_max_threads, 2064 }, 2065 { 2066 .procname = "random", 2067 .mode = 0555, 2068 .child = random_table, 2069 }, 2070 { 2071 .procname = "usermodehelper", 2072 .mode = 0555, 2073 .child = usermodehelper_table, 2074 }, 2075 #ifdef CONFIG_FW_LOADER_USER_HELPER 2076 { 2077 .procname = "firmware_config", 2078 .mode = 0555, 2079 .child = firmware_config_table, 2080 }, 2081 #endif 2082 { 2083 .procname = "overflowuid", 2084 .data = &overflowuid, 2085 .maxlen = sizeof(int), 2086 .mode = 0644, 2087 .proc_handler = proc_dointvec_minmax, 2088 .extra1 = &minolduid, 2089 .extra2 = &maxolduid, 2090 }, 2091 { 2092 .procname = "overflowgid", 2093 .data = &overflowgid, 2094 .maxlen = sizeof(int), 2095 .mode = 0644, 2096 .proc_handler = proc_dointvec_minmax, 2097 .extra1 = &minolduid, 2098 .extra2 = &maxolduid, 2099 }, 2100 #ifdef CONFIG_S390 2101 { 2102 .procname = "userprocess_debug", 2103 .data = &show_unhandled_signals, 2104 .maxlen = sizeof(int), 2105 .mode = 0644, 2106 .proc_handler = proc_dointvec, 2107 }, 2108 #endif 2109 { 2110 .procname = "pid_max", 2111 .data = &pid_max, 2112 .maxlen = sizeof (int), 2113 .mode = 0644, 2114 .proc_handler = proc_dointvec_minmax, 2115 .extra1 = &pid_max_min, 2116 .extra2 = &pid_max_max, 2117 }, 2118 { 2119 .procname = "panic_on_oops", 2120 .data = &panic_on_oops, 2121 .maxlen = sizeof(int), 2122 .mode = 0644, 2123 .proc_handler = proc_dointvec, 2124 }, 2125 { 2126 .procname = "panic_print", 2127 .data = &panic_print, 2128 .maxlen = sizeof(unsigned long), 2129 .mode = 0644, 2130 .proc_handler = proc_doulongvec_minmax, 2131 }, 2132 #if defined CONFIG_PRINTK 2133 { 2134 .procname = "printk", 2135 .data = &console_loglevel, 2136 .maxlen = 4*sizeof(int), 2137 .mode = 0644, 2138 .proc_handler = proc_dointvec, 2139 }, 2140 { 2141 .procname = "printk_ratelimit", 2142 .data = &printk_ratelimit_state.interval, 2143 .maxlen = sizeof(int), 2144 .mode = 0644, 2145 .proc_handler = proc_dointvec_jiffies, 2146 }, 2147 { 2148 .procname = "printk_ratelimit_burst", 2149 .data = &printk_ratelimit_state.burst, 2150 .maxlen = sizeof(int), 2151 .mode = 0644, 2152 .proc_handler = proc_dointvec, 2153 }, 2154 { 2155 .procname = "printk_delay", 2156 .data = &printk_delay_msec, 2157 .maxlen = sizeof(int), 2158 .mode = 0644, 2159 .proc_handler = proc_dointvec_minmax, 2160 .extra1 = SYSCTL_ZERO, 2161 .extra2 = &ten_thousand, 2162 }, 2163 { 2164 .procname = "printk_devkmsg", 2165 .data = devkmsg_log_str, 2166 .maxlen = DEVKMSG_STR_MAX_SIZE, 2167 .mode = 0644, 2168 .proc_handler = devkmsg_sysctl_set_loglvl, 2169 }, 2170 { 2171 .procname = "dmesg_restrict", 2172 .data = &dmesg_restrict, 2173 .maxlen = sizeof(int), 2174 .mode = 0644, 2175 .proc_handler = proc_dointvec_minmax_sysadmin, 2176 .extra1 = SYSCTL_ZERO, 2177 .extra2 = SYSCTL_ONE, 2178 }, 2179 { 2180 .procname = "kptr_restrict", 2181 .data = &kptr_restrict, 2182 .maxlen = sizeof(int), 2183 .mode = 0644, 2184 .proc_handler = proc_dointvec_minmax_sysadmin, 2185 .extra1 = SYSCTL_ZERO, 2186 .extra2 = &two, 2187 }, 2188 #endif 2189 { 2190 .procname = "ngroups_max", 2191 .data = &ngroups_max, 2192 .maxlen = sizeof (int), 2193 .mode = 0444, 2194 .proc_handler = proc_dointvec, 2195 }, 2196 { 2197 .procname = "cap_last_cap", 2198 .data = (void *)&cap_last_cap, 2199 .maxlen = sizeof(int), 2200 .mode = 0444, 2201 .proc_handler = proc_dointvec, 2202 }, 2203 #if defined(CONFIG_LOCKUP_DETECTOR) 2204 { 2205 .procname = "watchdog", 2206 .data = &watchdog_user_enabled, 2207 .maxlen = sizeof(int), 2208 .mode = 0644, 2209 .proc_handler = proc_watchdog, 2210 .extra1 = SYSCTL_ZERO, 2211 .extra2 = SYSCTL_ONE, 2212 }, 2213 { 2214 .procname = "watchdog_thresh", 2215 .data = &watchdog_thresh, 2216 .maxlen = sizeof(int), 2217 .mode = 0644, 2218 .proc_handler = proc_watchdog_thresh, 2219 .extra1 = SYSCTL_ZERO, 2220 .extra2 = &sixty, 2221 }, 2222 { 2223 .procname = "nmi_watchdog", 2224 .data = &nmi_watchdog_user_enabled, 2225 .maxlen = sizeof(int), 2226 .mode = NMI_WATCHDOG_SYSCTL_PERM, 2227 .proc_handler = proc_nmi_watchdog, 2228 .extra1 = SYSCTL_ZERO, 2229 .extra2 = SYSCTL_ONE, 2230 }, 2231 { 2232 .procname = "watchdog_cpumask", 2233 .data = &watchdog_cpumask_bits, 2234 .maxlen = NR_CPUS, 2235 .mode = 0644, 2236 .proc_handler = proc_watchdog_cpumask, 2237 }, 2238 #ifdef CONFIG_SOFTLOCKUP_DETECTOR 2239 { 2240 .procname = "soft_watchdog", 2241 .data = &soft_watchdog_user_enabled, 2242 .maxlen = sizeof(int), 2243 .mode = 0644, 2244 .proc_handler = proc_soft_watchdog, 2245 .extra1 = SYSCTL_ZERO, 2246 .extra2 = SYSCTL_ONE, 2247 }, 2248 { 2249 .procname = "softlockup_panic", 2250 .data = &softlockup_panic, 2251 .maxlen = sizeof(int), 2252 .mode = 0644, 2253 .proc_handler = proc_dointvec_minmax, 2254 .extra1 = SYSCTL_ZERO, 2255 .extra2 = SYSCTL_ONE, 2256 }, 2257 #ifdef CONFIG_SMP 2258 { 2259 .procname = "softlockup_all_cpu_backtrace", 2260 .data = &sysctl_softlockup_all_cpu_backtrace, 2261 .maxlen = sizeof(int), 2262 .mode = 0644, 2263 .proc_handler = proc_dointvec_minmax, 2264 .extra1 = SYSCTL_ZERO, 2265 .extra2 = SYSCTL_ONE, 2266 }, 2267 #endif /* CONFIG_SMP */ 2268 #endif 2269 #ifdef CONFIG_HARDLOCKUP_DETECTOR 2270 { 2271 .procname = "hardlockup_panic", 2272 .data = &hardlockup_panic, 2273 .maxlen = sizeof(int), 2274 .mode = 0644, 2275 .proc_handler = proc_dointvec_minmax, 2276 .extra1 = SYSCTL_ZERO, 2277 .extra2 = SYSCTL_ONE, 2278 }, 2279 #ifdef CONFIG_SMP 2280 { 2281 .procname = "hardlockup_all_cpu_backtrace", 2282 .data = &sysctl_hardlockup_all_cpu_backtrace, 2283 .maxlen = sizeof(int), 2284 .mode = 0644, 2285 .proc_handler = proc_dointvec_minmax, 2286 .extra1 = SYSCTL_ZERO, 2287 .extra2 = SYSCTL_ONE, 2288 }, 2289 #endif /* CONFIG_SMP */ 2290 #endif 2291 #endif 2292 2293 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86) 2294 { 2295 .procname = "unknown_nmi_panic", 2296 .data = &unknown_nmi_panic, 2297 .maxlen = sizeof (int), 2298 .mode = 0644, 2299 .proc_handler = proc_dointvec, 2300 }, 2301 #endif 2302 #if defined(CONFIG_X86) 2303 { 2304 .procname = "panic_on_unrecovered_nmi", 2305 .data = &panic_on_unrecovered_nmi, 2306 .maxlen = sizeof(int), 2307 .mode = 0644, 2308 .proc_handler = proc_dointvec, 2309 }, 2310 { 2311 .procname = "panic_on_io_nmi", 2312 .data = &panic_on_io_nmi, 2313 .maxlen = sizeof(int), 2314 .mode = 0644, 2315 .proc_handler = proc_dointvec, 2316 }, 2317 #ifdef CONFIG_DEBUG_STACKOVERFLOW 2318 { 2319 .procname = "panic_on_stackoverflow", 2320 .data = &sysctl_panic_on_stackoverflow, 2321 .maxlen = sizeof(int), 2322 .mode = 0644, 2323 .proc_handler = proc_dointvec, 2324 }, 2325 #endif 2326 { 2327 .procname = "bootloader_type", 2328 .data = &bootloader_type, 2329 .maxlen = sizeof (int), 2330 .mode = 0444, 2331 .proc_handler = proc_dointvec, 2332 }, 2333 { 2334 .procname = "bootloader_version", 2335 .data = &bootloader_version, 2336 .maxlen = sizeof (int), 2337 .mode = 0444, 2338 .proc_handler = proc_dointvec, 2339 }, 2340 { 2341 .procname = "io_delay_type", 2342 .data = &io_delay_type, 2343 .maxlen = sizeof(int), 2344 .mode = 0644, 2345 .proc_handler = proc_dointvec, 2346 }, 2347 #endif 2348 #if defined(CONFIG_MMU) 2349 { 2350 .procname = "randomize_va_space", 2351 .data = &randomize_va_space, 2352 .maxlen = sizeof(int), 2353 .mode = 0644, 2354 .proc_handler = proc_dointvec, 2355 }, 2356 #endif 2357 #if defined(CONFIG_S390) && defined(CONFIG_SMP) 2358 { 2359 .procname = "spin_retry", 2360 .data = &spin_retry, 2361 .maxlen = sizeof (int), 2362 .mode = 0644, 2363 .proc_handler = proc_dointvec, 2364 }, 2365 #endif 2366 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86) 2367 { 2368 .procname = "acpi_video_flags", 2369 .data = &acpi_realmode_flags, 2370 .maxlen = sizeof (unsigned long), 2371 .mode = 0644, 2372 .proc_handler = proc_doulongvec_minmax, 2373 }, 2374 #endif 2375 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN 2376 { 2377 .procname = "ignore-unaligned-usertrap", 2378 .data = &no_unaligned_warning, 2379 .maxlen = sizeof (int), 2380 .mode = 0644, 2381 .proc_handler = proc_dointvec, 2382 }, 2383 #endif 2384 #ifdef CONFIG_IA64 2385 { 2386 .procname = "unaligned-dump-stack", 2387 .data = &unaligned_dump_stack, 2388 .maxlen = sizeof (int), 2389 .mode = 0644, 2390 .proc_handler = proc_dointvec, 2391 }, 2392 #endif 2393 #ifdef CONFIG_DETECT_HUNG_TASK 2394 { 2395 .procname = "hung_task_panic", 2396 .data = &sysctl_hung_task_panic, 2397 .maxlen = sizeof(int), 2398 .mode = 0644, 2399 .proc_handler = proc_dointvec_minmax, 2400 .extra1 = SYSCTL_ZERO, 2401 .extra2 = SYSCTL_ONE, 2402 }, 2403 { 2404 .procname = "hung_task_check_count", 2405 .data = &sysctl_hung_task_check_count, 2406 .maxlen = sizeof(int), 2407 .mode = 0644, 2408 .proc_handler = proc_dointvec_minmax, 2409 .extra1 = SYSCTL_ZERO, 2410 }, 2411 { 2412 .procname = "hung_task_timeout_secs", 2413 .data = &sysctl_hung_task_timeout_secs, 2414 .maxlen = sizeof(unsigned long), 2415 .mode = 0644, 2416 .proc_handler = proc_dohung_task_timeout_secs, 2417 .extra2 = &hung_task_timeout_max, 2418 }, 2419 { 2420 .procname = "hung_task_check_interval_secs", 2421 .data = &sysctl_hung_task_check_interval_secs, 2422 .maxlen = sizeof(unsigned long), 2423 .mode = 0644, 2424 .proc_handler = proc_dohung_task_timeout_secs, 2425 .extra2 = &hung_task_timeout_max, 2426 }, 2427 { 2428 .procname = "hung_task_warnings", 2429 .data = &sysctl_hung_task_warnings, 2430 .maxlen = sizeof(int), 2431 .mode = 0644, 2432 .proc_handler = proc_dointvec_minmax, 2433 .extra1 = &neg_one, 2434 }, 2435 #endif 2436 #ifdef CONFIG_RT_MUTEXES 2437 { 2438 .procname = "max_lock_depth", 2439 .data = &max_lock_depth, 2440 .maxlen = sizeof(int), 2441 .mode = 0644, 2442 .proc_handler = proc_dointvec, 2443 }, 2444 #endif 2445 { 2446 .procname = "poweroff_cmd", 2447 .data = &poweroff_cmd, 2448 .maxlen = POWEROFF_CMD_PATH_LEN, 2449 .mode = 0644, 2450 .proc_handler = proc_dostring, 2451 }, 2452 #ifdef CONFIG_KEYS 2453 { 2454 .procname = "keys", 2455 .mode = 0555, 2456 .child = key_sysctls, 2457 }, 2458 #endif 2459 #ifdef CONFIG_PERF_EVENTS 2460 /* 2461 * User-space scripts rely on the existence of this file 2462 * as a feature check for perf_events being enabled. 2463 * 2464 * So it's an ABI, do not remove! 2465 */ 2466 { 2467 .procname = "perf_event_paranoid", 2468 .data = &sysctl_perf_event_paranoid, 2469 .maxlen = sizeof(sysctl_perf_event_paranoid), 2470 .mode = 0644, 2471 .proc_handler = proc_dointvec, 2472 }, 2473 { 2474 .procname = "perf_event_mlock_kb", 2475 .data = &sysctl_perf_event_mlock, 2476 .maxlen = sizeof(sysctl_perf_event_mlock), 2477 .mode = 0644, 2478 .proc_handler = proc_dointvec, 2479 }, 2480 { 2481 .procname = "perf_event_max_sample_rate", 2482 .data = &sysctl_perf_event_sample_rate, 2483 .maxlen = sizeof(sysctl_perf_event_sample_rate), 2484 .mode = 0644, 2485 .proc_handler = perf_proc_update_handler, 2486 .extra1 = SYSCTL_ONE, 2487 }, 2488 { 2489 .procname = "perf_cpu_time_max_percent", 2490 .data = &sysctl_perf_cpu_time_max_percent, 2491 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent), 2492 .mode = 0644, 2493 .proc_handler = perf_cpu_time_max_percent_handler, 2494 .extra1 = SYSCTL_ZERO, 2495 .extra2 = &one_hundred, 2496 }, 2497 { 2498 .procname = "perf_event_max_stack", 2499 .data = &sysctl_perf_event_max_stack, 2500 .maxlen = sizeof(sysctl_perf_event_max_stack), 2501 .mode = 0644, 2502 .proc_handler = perf_event_max_stack_handler, 2503 .extra1 = SYSCTL_ZERO, 2504 .extra2 = &six_hundred_forty_kb, 2505 }, 2506 { 2507 .procname = "perf_event_max_contexts_per_stack", 2508 .data = &sysctl_perf_event_max_contexts_per_stack, 2509 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack), 2510 .mode = 0644, 2511 .proc_handler = perf_event_max_stack_handler, 2512 .extra1 = SYSCTL_ZERO, 2513 .extra2 = &one_thousand, 2514 }, 2515 #endif 2516 { 2517 .procname = "panic_on_warn", 2518 .data = &panic_on_warn, 2519 .maxlen = sizeof(int), 2520 .mode = 0644, 2521 .proc_handler = proc_dointvec_minmax, 2522 .extra1 = SYSCTL_ZERO, 2523 .extra2 = SYSCTL_ONE, 2524 }, 2525 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) 2526 { 2527 .procname = "timer_migration", 2528 .data = &sysctl_timer_migration, 2529 .maxlen = sizeof(unsigned int), 2530 .mode = 0644, 2531 .proc_handler = timer_migration_handler, 2532 .extra1 = SYSCTL_ZERO, 2533 .extra2 = SYSCTL_ONE, 2534 }, 2535 #endif 2536 #ifdef CONFIG_BPF_SYSCALL 2537 { 2538 .procname = "unprivileged_bpf_disabled", 2539 .data = &sysctl_unprivileged_bpf_disabled, 2540 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled), 2541 .mode = 0644, 2542 /* only handle a transition from default "0" to "1" */ 2543 .proc_handler = proc_dointvec_minmax, 2544 .extra1 = SYSCTL_ONE, 2545 .extra2 = SYSCTL_ONE, 2546 }, 2547 { 2548 .procname = "bpf_stats_enabled", 2549 .data = &bpf_stats_enabled_key.key, 2550 .maxlen = sizeof(bpf_stats_enabled_key), 2551 .mode = 0644, 2552 .proc_handler = proc_do_static_key, 2553 }, 2554 #endif 2555 #if defined(CONFIG_TREE_RCU) 2556 { 2557 .procname = "panic_on_rcu_stall", 2558 .data = &sysctl_panic_on_rcu_stall, 2559 .maxlen = sizeof(sysctl_panic_on_rcu_stall), 2560 .mode = 0644, 2561 .proc_handler = proc_dointvec_minmax, 2562 .extra1 = SYSCTL_ZERO, 2563 .extra2 = SYSCTL_ONE, 2564 }, 2565 #endif 2566 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE 2567 { 2568 .procname = "stack_erasing", 2569 .data = NULL, 2570 .maxlen = sizeof(int), 2571 .mode = 0600, 2572 .proc_handler = stack_erasing_sysctl, 2573 .extra1 = SYSCTL_ZERO, 2574 .extra2 = SYSCTL_ONE, 2575 }, 2576 #endif 2577 { } 2578 }; 2579 2580 static struct ctl_table vm_table[] = { 2581 { 2582 .procname = "overcommit_memory", 2583 .data = &sysctl_overcommit_memory, 2584 .maxlen = sizeof(sysctl_overcommit_memory), 2585 .mode = 0644, 2586 .proc_handler = proc_dointvec_minmax, 2587 .extra1 = SYSCTL_ZERO, 2588 .extra2 = &two, 2589 }, 2590 { 2591 .procname = "panic_on_oom", 2592 .data = &sysctl_panic_on_oom, 2593 .maxlen = sizeof(sysctl_panic_on_oom), 2594 .mode = 0644, 2595 .proc_handler = proc_dointvec_minmax, 2596 .extra1 = SYSCTL_ZERO, 2597 .extra2 = &two, 2598 }, 2599 { 2600 .procname = "oom_kill_allocating_task", 2601 .data = &sysctl_oom_kill_allocating_task, 2602 .maxlen = sizeof(sysctl_oom_kill_allocating_task), 2603 .mode = 0644, 2604 .proc_handler = proc_dointvec, 2605 }, 2606 { 2607 .procname = "oom_dump_tasks", 2608 .data = &sysctl_oom_dump_tasks, 2609 .maxlen = sizeof(sysctl_oom_dump_tasks), 2610 .mode = 0644, 2611 .proc_handler = proc_dointvec, 2612 }, 2613 { 2614 .procname = "overcommit_ratio", 2615 .data = &sysctl_overcommit_ratio, 2616 .maxlen = sizeof(sysctl_overcommit_ratio), 2617 .mode = 0644, 2618 .proc_handler = overcommit_ratio_handler, 2619 }, 2620 { 2621 .procname = "overcommit_kbytes", 2622 .data = &sysctl_overcommit_kbytes, 2623 .maxlen = sizeof(sysctl_overcommit_kbytes), 2624 .mode = 0644, 2625 .proc_handler = overcommit_kbytes_handler, 2626 }, 2627 { 2628 .procname = "page-cluster", 2629 .data = &page_cluster, 2630 .maxlen = sizeof(int), 2631 .mode = 0644, 2632 .proc_handler = proc_dointvec_minmax, 2633 .extra1 = SYSCTL_ZERO, 2634 }, 2635 { 2636 .procname = "dirty_background_ratio", 2637 .data = &dirty_background_ratio, 2638 .maxlen = sizeof(dirty_background_ratio), 2639 .mode = 0644, 2640 .proc_handler = dirty_background_ratio_handler, 2641 .extra1 = SYSCTL_ZERO, 2642 .extra2 = &one_hundred, 2643 }, 2644 { 2645 .procname = "dirty_background_bytes", 2646 .data = &dirty_background_bytes, 2647 .maxlen = sizeof(dirty_background_bytes), 2648 .mode = 0644, 2649 .proc_handler = dirty_background_bytes_handler, 2650 .extra1 = &one_ul, 2651 }, 2652 { 2653 .procname = "dirty_ratio", 2654 .data = &vm_dirty_ratio, 2655 .maxlen = sizeof(vm_dirty_ratio), 2656 .mode = 0644, 2657 .proc_handler = dirty_ratio_handler, 2658 .extra1 = SYSCTL_ZERO, 2659 .extra2 = &one_hundred, 2660 }, 2661 { 2662 .procname = "dirty_bytes", 2663 .data = &vm_dirty_bytes, 2664 .maxlen = sizeof(vm_dirty_bytes), 2665 .mode = 0644, 2666 .proc_handler = dirty_bytes_handler, 2667 .extra1 = &dirty_bytes_min, 2668 }, 2669 { 2670 .procname = "dirty_writeback_centisecs", 2671 .data = &dirty_writeback_interval, 2672 .maxlen = sizeof(dirty_writeback_interval), 2673 .mode = 0644, 2674 .proc_handler = dirty_writeback_centisecs_handler, 2675 }, 2676 { 2677 .procname = "dirty_expire_centisecs", 2678 .data = &dirty_expire_interval, 2679 .maxlen = sizeof(dirty_expire_interval), 2680 .mode = 0644, 2681 .proc_handler = proc_dointvec_minmax, 2682 .extra1 = SYSCTL_ZERO, 2683 }, 2684 { 2685 .procname = "dirtytime_expire_seconds", 2686 .data = &dirtytime_expire_interval, 2687 .maxlen = sizeof(dirtytime_expire_interval), 2688 .mode = 0644, 2689 .proc_handler = dirtytime_interval_handler, 2690 .extra1 = SYSCTL_ZERO, 2691 }, 2692 { 2693 .procname = "swappiness", 2694 .data = &vm_swappiness, 2695 .maxlen = sizeof(vm_swappiness), 2696 .mode = 0644, 2697 .proc_handler = proc_dointvec_minmax, 2698 .extra1 = SYSCTL_ZERO, 2699 .extra2 = &one_hundred, 2700 }, 2701 #ifdef CONFIG_HUGETLB_PAGE 2702 { 2703 .procname = "nr_hugepages", 2704 .data = NULL, 2705 .maxlen = sizeof(unsigned long), 2706 .mode = 0644, 2707 .proc_handler = hugetlb_sysctl_handler, 2708 }, 2709 #ifdef CONFIG_NUMA 2710 { 2711 .procname = "nr_hugepages_mempolicy", 2712 .data = NULL, 2713 .maxlen = sizeof(unsigned long), 2714 .mode = 0644, 2715 .proc_handler = &hugetlb_mempolicy_sysctl_handler, 2716 }, 2717 { 2718 .procname = "numa_stat", 2719 .data = &sysctl_vm_numa_stat, 2720 .maxlen = sizeof(int), 2721 .mode = 0644, 2722 .proc_handler = sysctl_vm_numa_stat_handler, 2723 .extra1 = SYSCTL_ZERO, 2724 .extra2 = SYSCTL_ONE, 2725 }, 2726 #endif 2727 { 2728 .procname = "hugetlb_shm_group", 2729 .data = &sysctl_hugetlb_shm_group, 2730 .maxlen = sizeof(gid_t), 2731 .mode = 0644, 2732 .proc_handler = proc_dointvec, 2733 }, 2734 { 2735 .procname = "nr_overcommit_hugepages", 2736 .data = NULL, 2737 .maxlen = sizeof(unsigned long), 2738 .mode = 0644, 2739 .proc_handler = hugetlb_overcommit_handler, 2740 }, 2741 #endif 2742 { 2743 .procname = "lowmem_reserve_ratio", 2744 .data = &sysctl_lowmem_reserve_ratio, 2745 .maxlen = sizeof(sysctl_lowmem_reserve_ratio), 2746 .mode = 0644, 2747 .proc_handler = lowmem_reserve_ratio_sysctl_handler, 2748 }, 2749 { 2750 .procname = "drop_caches", 2751 .data = &sysctl_drop_caches, 2752 .maxlen = sizeof(int), 2753 .mode = 0200, 2754 .proc_handler = drop_caches_sysctl_handler, 2755 .extra1 = SYSCTL_ONE, 2756 .extra2 = &four, 2757 }, 2758 #ifdef CONFIG_COMPACTION 2759 { 2760 .procname = "compact_memory", 2761 .data = &sysctl_compact_memory, 2762 .maxlen = sizeof(int), 2763 .mode = 0200, 2764 .proc_handler = sysctl_compaction_handler, 2765 }, 2766 { 2767 .procname = "extfrag_threshold", 2768 .data = &sysctl_extfrag_threshold, 2769 .maxlen = sizeof(int), 2770 .mode = 0644, 2771 .proc_handler = proc_dointvec_minmax, 2772 .extra1 = &min_extfrag_threshold, 2773 .extra2 = &max_extfrag_threshold, 2774 }, 2775 { 2776 .procname = "compact_unevictable_allowed", 2777 .data = &sysctl_compact_unevictable_allowed, 2778 .maxlen = sizeof(int), 2779 .mode = 0644, 2780 .proc_handler = proc_dointvec_minmax_warn_RT_change, 2781 .extra1 = SYSCTL_ZERO, 2782 .extra2 = SYSCTL_ONE, 2783 }, 2784 2785 #endif /* CONFIG_COMPACTION */ 2786 { 2787 .procname = "min_free_kbytes", 2788 .data = &min_free_kbytes, 2789 .maxlen = sizeof(min_free_kbytes), 2790 .mode = 0644, 2791 .proc_handler = min_free_kbytes_sysctl_handler, 2792 .extra1 = SYSCTL_ZERO, 2793 }, 2794 { 2795 .procname = "watermark_boost_factor", 2796 .data = &watermark_boost_factor, 2797 .maxlen = sizeof(watermark_boost_factor), 2798 .mode = 0644, 2799 .proc_handler = proc_dointvec_minmax, 2800 .extra1 = SYSCTL_ZERO, 2801 }, 2802 { 2803 .procname = "watermark_scale_factor", 2804 .data = &watermark_scale_factor, 2805 .maxlen = sizeof(watermark_scale_factor), 2806 .mode = 0644, 2807 .proc_handler = watermark_scale_factor_sysctl_handler, 2808 .extra1 = SYSCTL_ONE, 2809 .extra2 = &one_thousand, 2810 }, 2811 { 2812 .procname = "percpu_pagelist_fraction", 2813 .data = &percpu_pagelist_fraction, 2814 .maxlen = sizeof(percpu_pagelist_fraction), 2815 .mode = 0644, 2816 .proc_handler = percpu_pagelist_fraction_sysctl_handler, 2817 .extra1 = SYSCTL_ZERO, 2818 }, 2819 #ifdef CONFIG_MMU 2820 { 2821 .procname = "max_map_count", 2822 .data = &sysctl_max_map_count, 2823 .maxlen = sizeof(sysctl_max_map_count), 2824 .mode = 0644, 2825 .proc_handler = proc_dointvec_minmax, 2826 .extra1 = SYSCTL_ZERO, 2827 }, 2828 #else 2829 { 2830 .procname = "nr_trim_pages", 2831 .data = &sysctl_nr_trim_pages, 2832 .maxlen = sizeof(sysctl_nr_trim_pages), 2833 .mode = 0644, 2834 .proc_handler = proc_dointvec_minmax, 2835 .extra1 = SYSCTL_ZERO, 2836 }, 2837 #endif 2838 { 2839 .procname = "laptop_mode", 2840 .data = &laptop_mode, 2841 .maxlen = sizeof(laptop_mode), 2842 .mode = 0644, 2843 .proc_handler = proc_dointvec_jiffies, 2844 }, 2845 { 2846 .procname = "block_dump", 2847 .data = &block_dump, 2848 .maxlen = sizeof(block_dump), 2849 .mode = 0644, 2850 .proc_handler = proc_dointvec, 2851 .extra1 = SYSCTL_ZERO, 2852 }, 2853 { 2854 .procname = "vfs_cache_pressure", 2855 .data = &sysctl_vfs_cache_pressure, 2856 .maxlen = sizeof(sysctl_vfs_cache_pressure), 2857 .mode = 0644, 2858 .proc_handler = proc_dointvec, 2859 .extra1 = SYSCTL_ZERO, 2860 }, 2861 #if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \ 2862 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT) 2863 { 2864 .procname = "legacy_va_layout", 2865 .data = &sysctl_legacy_va_layout, 2866 .maxlen = sizeof(sysctl_legacy_va_layout), 2867 .mode = 0644, 2868 .proc_handler = proc_dointvec, 2869 .extra1 = SYSCTL_ZERO, 2870 }, 2871 #endif 2872 #ifdef CONFIG_NUMA 2873 { 2874 .procname = "zone_reclaim_mode", 2875 .data = &node_reclaim_mode, 2876 .maxlen = sizeof(node_reclaim_mode), 2877 .mode = 0644, 2878 .proc_handler = proc_dointvec, 2879 .extra1 = SYSCTL_ZERO, 2880 }, 2881 { 2882 .procname = "min_unmapped_ratio", 2883 .data = &sysctl_min_unmapped_ratio, 2884 .maxlen = sizeof(sysctl_min_unmapped_ratio), 2885 .mode = 0644, 2886 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler, 2887 .extra1 = SYSCTL_ZERO, 2888 .extra2 = &one_hundred, 2889 }, 2890 { 2891 .procname = "min_slab_ratio", 2892 .data = &sysctl_min_slab_ratio, 2893 .maxlen = sizeof(sysctl_min_slab_ratio), 2894 .mode = 0644, 2895 .proc_handler = sysctl_min_slab_ratio_sysctl_handler, 2896 .extra1 = SYSCTL_ZERO, 2897 .extra2 = &one_hundred, 2898 }, 2899 #endif 2900 #ifdef CONFIG_SMP 2901 { 2902 .procname = "stat_interval", 2903 .data = &sysctl_stat_interval, 2904 .maxlen = sizeof(sysctl_stat_interval), 2905 .mode = 0644, 2906 .proc_handler = proc_dointvec_jiffies, 2907 }, 2908 { 2909 .procname = "stat_refresh", 2910 .data = NULL, 2911 .maxlen = 0, 2912 .mode = 0600, 2913 .proc_handler = vmstat_refresh, 2914 }, 2915 #endif 2916 #ifdef CONFIG_MMU 2917 { 2918 .procname = "mmap_min_addr", 2919 .data = &dac_mmap_min_addr, 2920 .maxlen = sizeof(unsigned long), 2921 .mode = 0644, 2922 .proc_handler = mmap_min_addr_handler, 2923 }, 2924 #endif 2925 #ifdef CONFIG_NUMA 2926 { 2927 .procname = "numa_zonelist_order", 2928 .data = &numa_zonelist_order, 2929 .maxlen = NUMA_ZONELIST_ORDER_LEN, 2930 .mode = 0644, 2931 .proc_handler = numa_zonelist_order_handler, 2932 }, 2933 #endif 2934 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \ 2935 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL)) 2936 { 2937 .procname = "vdso_enabled", 2938 #ifdef CONFIG_X86_32 2939 .data = &vdso32_enabled, 2940 .maxlen = sizeof(vdso32_enabled), 2941 #else 2942 .data = &vdso_enabled, 2943 .maxlen = sizeof(vdso_enabled), 2944 #endif 2945 .mode = 0644, 2946 .proc_handler = proc_dointvec, 2947 .extra1 = SYSCTL_ZERO, 2948 }, 2949 #endif 2950 #ifdef CONFIG_HIGHMEM 2951 { 2952 .procname = "highmem_is_dirtyable", 2953 .data = &vm_highmem_is_dirtyable, 2954 .maxlen = sizeof(vm_highmem_is_dirtyable), 2955 .mode = 0644, 2956 .proc_handler = proc_dointvec_minmax, 2957 .extra1 = SYSCTL_ZERO, 2958 .extra2 = SYSCTL_ONE, 2959 }, 2960 #endif 2961 #ifdef CONFIG_MEMORY_FAILURE 2962 { 2963 .procname = "memory_failure_early_kill", 2964 .data = &sysctl_memory_failure_early_kill, 2965 .maxlen = sizeof(sysctl_memory_failure_early_kill), 2966 .mode = 0644, 2967 .proc_handler = proc_dointvec_minmax, 2968 .extra1 = SYSCTL_ZERO, 2969 .extra2 = SYSCTL_ONE, 2970 }, 2971 { 2972 .procname = "memory_failure_recovery", 2973 .data = &sysctl_memory_failure_recovery, 2974 .maxlen = sizeof(sysctl_memory_failure_recovery), 2975 .mode = 0644, 2976 .proc_handler = proc_dointvec_minmax, 2977 .extra1 = SYSCTL_ZERO, 2978 .extra2 = SYSCTL_ONE, 2979 }, 2980 #endif 2981 { 2982 .procname = "user_reserve_kbytes", 2983 .data = &sysctl_user_reserve_kbytes, 2984 .maxlen = sizeof(sysctl_user_reserve_kbytes), 2985 .mode = 0644, 2986 .proc_handler = proc_doulongvec_minmax, 2987 }, 2988 { 2989 .procname = "admin_reserve_kbytes", 2990 .data = &sysctl_admin_reserve_kbytes, 2991 .maxlen = sizeof(sysctl_admin_reserve_kbytes), 2992 .mode = 0644, 2993 .proc_handler = proc_doulongvec_minmax, 2994 }, 2995 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS 2996 { 2997 .procname = "mmap_rnd_bits", 2998 .data = &mmap_rnd_bits, 2999 .maxlen = sizeof(mmap_rnd_bits), 3000 .mode = 0600, 3001 .proc_handler = proc_dointvec_minmax, 3002 .extra1 = (void *)&mmap_rnd_bits_min, 3003 .extra2 = (void *)&mmap_rnd_bits_max, 3004 }, 3005 #endif 3006 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS 3007 { 3008 .procname = "mmap_rnd_compat_bits", 3009 .data = &mmap_rnd_compat_bits, 3010 .maxlen = sizeof(mmap_rnd_compat_bits), 3011 .mode = 0600, 3012 .proc_handler = proc_dointvec_minmax, 3013 .extra1 = (void *)&mmap_rnd_compat_bits_min, 3014 .extra2 = (void *)&mmap_rnd_compat_bits_max, 3015 }, 3016 #endif 3017 #ifdef CONFIG_USERFAULTFD 3018 { 3019 .procname = "unprivileged_userfaultfd", 3020 .data = &sysctl_unprivileged_userfaultfd, 3021 .maxlen = sizeof(sysctl_unprivileged_userfaultfd), 3022 .mode = 0644, 3023 .proc_handler = proc_dointvec_minmax, 3024 .extra1 = SYSCTL_ZERO, 3025 .extra2 = SYSCTL_ONE, 3026 }, 3027 #endif 3028 { } 3029 }; 3030 3031 static struct ctl_table fs_table[] = { 3032 { 3033 .procname = "inode-nr", 3034 .data = &inodes_stat, 3035 .maxlen = 2*sizeof(long), 3036 .mode = 0444, 3037 .proc_handler = proc_nr_inodes, 3038 }, 3039 { 3040 .procname = "inode-state", 3041 .data = &inodes_stat, 3042 .maxlen = 7*sizeof(long), 3043 .mode = 0444, 3044 .proc_handler = proc_nr_inodes, 3045 }, 3046 { 3047 .procname = "file-nr", 3048 .data = &files_stat, 3049 .maxlen = sizeof(files_stat), 3050 .mode = 0444, 3051 .proc_handler = proc_nr_files, 3052 }, 3053 { 3054 .procname = "file-max", 3055 .data = &files_stat.max_files, 3056 .maxlen = sizeof(files_stat.max_files), 3057 .mode = 0644, 3058 .proc_handler = proc_doulongvec_minmax, 3059 .extra1 = &zero_ul, 3060 .extra2 = &long_max, 3061 }, 3062 { 3063 .procname = "nr_open", 3064 .data = &sysctl_nr_open, 3065 .maxlen = sizeof(unsigned int), 3066 .mode = 0644, 3067 .proc_handler = proc_dointvec_minmax, 3068 .extra1 = &sysctl_nr_open_min, 3069 .extra2 = &sysctl_nr_open_max, 3070 }, 3071 { 3072 .procname = "dentry-state", 3073 .data = &dentry_stat, 3074 .maxlen = 6*sizeof(long), 3075 .mode = 0444, 3076 .proc_handler = proc_nr_dentry, 3077 }, 3078 { 3079 .procname = "overflowuid", 3080 .data = &fs_overflowuid, 3081 .maxlen = sizeof(int), 3082 .mode = 0644, 3083 .proc_handler = proc_dointvec_minmax, 3084 .extra1 = &minolduid, 3085 .extra2 = &maxolduid, 3086 }, 3087 { 3088 .procname = "overflowgid", 3089 .data = &fs_overflowgid, 3090 .maxlen = sizeof(int), 3091 .mode = 0644, 3092 .proc_handler = proc_dointvec_minmax, 3093 .extra1 = &minolduid, 3094 .extra2 = &maxolduid, 3095 }, 3096 #ifdef CONFIG_FILE_LOCKING 3097 { 3098 .procname = "leases-enable", 3099 .data = &leases_enable, 3100 .maxlen = sizeof(int), 3101 .mode = 0644, 3102 .proc_handler = proc_dointvec, 3103 }, 3104 #endif 3105 #ifdef CONFIG_DNOTIFY 3106 { 3107 .procname = "dir-notify-enable", 3108 .data = &dir_notify_enable, 3109 .maxlen = sizeof(int), 3110 .mode = 0644, 3111 .proc_handler = proc_dointvec, 3112 }, 3113 #endif 3114 #ifdef CONFIG_MMU 3115 #ifdef CONFIG_FILE_LOCKING 3116 { 3117 .procname = "lease-break-time", 3118 .data = &lease_break_time, 3119 .maxlen = sizeof(int), 3120 .mode = 0644, 3121 .proc_handler = proc_dointvec, 3122 }, 3123 #endif 3124 #ifdef CONFIG_AIO 3125 { 3126 .procname = "aio-nr", 3127 .data = &aio_nr, 3128 .maxlen = sizeof(aio_nr), 3129 .mode = 0444, 3130 .proc_handler = proc_doulongvec_minmax, 3131 }, 3132 { 3133 .procname = "aio-max-nr", 3134 .data = &aio_max_nr, 3135 .maxlen = sizeof(aio_max_nr), 3136 .mode = 0644, 3137 .proc_handler = proc_doulongvec_minmax, 3138 }, 3139 #endif /* CONFIG_AIO */ 3140 #ifdef CONFIG_INOTIFY_USER 3141 { 3142 .procname = "inotify", 3143 .mode = 0555, 3144 .child = inotify_table, 3145 }, 3146 #endif 3147 #ifdef CONFIG_EPOLL 3148 { 3149 .procname = "epoll", 3150 .mode = 0555, 3151 .child = epoll_table, 3152 }, 3153 #endif 3154 #endif 3155 { 3156 .procname = "protected_symlinks", 3157 .data = &sysctl_protected_symlinks, 3158 .maxlen = sizeof(int), 3159 .mode = 0600, 3160 .proc_handler = proc_dointvec_minmax, 3161 .extra1 = SYSCTL_ZERO, 3162 .extra2 = SYSCTL_ONE, 3163 }, 3164 { 3165 .procname = "protected_hardlinks", 3166 .data = &sysctl_protected_hardlinks, 3167 .maxlen = sizeof(int), 3168 .mode = 0600, 3169 .proc_handler = proc_dointvec_minmax, 3170 .extra1 = SYSCTL_ZERO, 3171 .extra2 = SYSCTL_ONE, 3172 }, 3173 { 3174 .procname = "protected_fifos", 3175 .data = &sysctl_protected_fifos, 3176 .maxlen = sizeof(int), 3177 .mode = 0600, 3178 .proc_handler = proc_dointvec_minmax, 3179 .extra1 = SYSCTL_ZERO, 3180 .extra2 = &two, 3181 }, 3182 { 3183 .procname = "protected_regular", 3184 .data = &sysctl_protected_regular, 3185 .maxlen = sizeof(int), 3186 .mode = 0600, 3187 .proc_handler = proc_dointvec_minmax, 3188 .extra1 = SYSCTL_ZERO, 3189 .extra2 = &two, 3190 }, 3191 { 3192 .procname = "suid_dumpable", 3193 .data = &suid_dumpable, 3194 .maxlen = sizeof(int), 3195 .mode = 0644, 3196 .proc_handler = proc_dointvec_minmax_coredump, 3197 .extra1 = SYSCTL_ZERO, 3198 .extra2 = &two, 3199 }, 3200 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE) 3201 { 3202 .procname = "binfmt_misc", 3203 .mode = 0555, 3204 .child = sysctl_mount_point, 3205 }, 3206 #endif 3207 { 3208 .procname = "pipe-max-size", 3209 .data = &pipe_max_size, 3210 .maxlen = sizeof(pipe_max_size), 3211 .mode = 0644, 3212 .proc_handler = proc_dopipe_max_size, 3213 }, 3214 { 3215 .procname = "pipe-user-pages-hard", 3216 .data = &pipe_user_pages_hard, 3217 .maxlen = sizeof(pipe_user_pages_hard), 3218 .mode = 0644, 3219 .proc_handler = proc_doulongvec_minmax, 3220 }, 3221 { 3222 .procname = "pipe-user-pages-soft", 3223 .data = &pipe_user_pages_soft, 3224 .maxlen = sizeof(pipe_user_pages_soft), 3225 .mode = 0644, 3226 .proc_handler = proc_doulongvec_minmax, 3227 }, 3228 { 3229 .procname = "mount-max", 3230 .data = &sysctl_mount_max, 3231 .maxlen = sizeof(unsigned int), 3232 .mode = 0644, 3233 .proc_handler = proc_dointvec_minmax, 3234 .extra1 = SYSCTL_ONE, 3235 }, 3236 { } 3237 }; 3238 3239 static struct ctl_table debug_table[] = { 3240 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE 3241 { 3242 .procname = "exception-trace", 3243 .data = &show_unhandled_signals, 3244 .maxlen = sizeof(int), 3245 .mode = 0644, 3246 .proc_handler = proc_dointvec 3247 }, 3248 #endif 3249 #if defined(CONFIG_OPTPROBES) 3250 { 3251 .procname = "kprobes-optimization", 3252 .data = &sysctl_kprobes_optimization, 3253 .maxlen = sizeof(int), 3254 .mode = 0644, 3255 .proc_handler = proc_kprobes_optimization_handler, 3256 .extra1 = SYSCTL_ZERO, 3257 .extra2 = SYSCTL_ONE, 3258 }, 3259 #endif 3260 { } 3261 }; 3262 3263 static struct ctl_table dev_table[] = { 3264 { } 3265 }; 3266 3267 static struct ctl_table sysctl_base_table[] = { 3268 { 3269 .procname = "kernel", 3270 .mode = 0555, 3271 .child = kern_table, 3272 }, 3273 { 3274 .procname = "vm", 3275 .mode = 0555, 3276 .child = vm_table, 3277 }, 3278 { 3279 .procname = "fs", 3280 .mode = 0555, 3281 .child = fs_table, 3282 }, 3283 { 3284 .procname = "debug", 3285 .mode = 0555, 3286 .child = debug_table, 3287 }, 3288 { 3289 .procname = "dev", 3290 .mode = 0555, 3291 .child = dev_table, 3292 }, 3293 { } 3294 }; 3295 3296 int __init sysctl_init(void) 3297 { 3298 struct ctl_table_header *hdr; 3299 3300 hdr = register_sysctl_table(sysctl_base_table); 3301 kmemleak_not_leak(hdr); 3302 return 0; 3303 } 3304 #endif /* CONFIG_SYSCTL */ 3305 /* 3306 * No sense putting this after each symbol definition, twice, 3307 * exception granted :-) 3308 */ 3309 EXPORT_SYMBOL(proc_dointvec); 3310 EXPORT_SYMBOL(proc_douintvec); 3311 EXPORT_SYMBOL(proc_dointvec_jiffies); 3312 EXPORT_SYMBOL(proc_dointvec_minmax); 3313 EXPORT_SYMBOL_GPL(proc_douintvec_minmax); 3314 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies); 3315 EXPORT_SYMBOL(proc_dointvec_ms_jiffies); 3316 EXPORT_SYMBOL(proc_dostring); 3317 EXPORT_SYMBOL(proc_doulongvec_minmax); 3318 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax); 3319 EXPORT_SYMBOL(proc_do_large_bitmap); 3320