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 #if defined(CONFIG_BPF_SYSCALL) && defined(CONFIG_SYSCTL) 205 static int bpf_stats_handler(struct ctl_table *table, int write, 206 void __user *buffer, size_t *lenp, 207 loff_t *ppos) 208 { 209 struct static_key *key = (struct static_key *)table->data; 210 static int saved_val; 211 int val, ret; 212 struct ctl_table tmp = { 213 .data = &val, 214 .maxlen = sizeof(val), 215 .mode = table->mode, 216 .extra1 = SYSCTL_ZERO, 217 .extra2 = SYSCTL_ONE, 218 }; 219 220 if (write && !capable(CAP_SYS_ADMIN)) 221 return -EPERM; 222 223 mutex_lock(&bpf_stats_enabled_mutex); 224 val = saved_val; 225 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos); 226 if (write && !ret && val != saved_val) { 227 if (val) 228 static_key_slow_inc(key); 229 else 230 static_key_slow_dec(key); 231 saved_val = val; 232 } 233 mutex_unlock(&bpf_stats_enabled_mutex); 234 return ret; 235 } 236 #endif 237 238 /* 239 * /proc/sys support 240 */ 241 242 #ifdef CONFIG_PROC_SYSCTL 243 244 static int _proc_do_string(char *data, int maxlen, int write, 245 char *buffer, size_t *lenp, loff_t *ppos) 246 { 247 size_t len; 248 char c, *p; 249 250 if (!data || !maxlen || !*lenp) { 251 *lenp = 0; 252 return 0; 253 } 254 255 if (write) { 256 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) { 257 /* Only continue writes not past the end of buffer. */ 258 len = strlen(data); 259 if (len > maxlen - 1) 260 len = maxlen - 1; 261 262 if (*ppos > len) 263 return 0; 264 len = *ppos; 265 } else { 266 /* Start writing from beginning of buffer. */ 267 len = 0; 268 } 269 270 *ppos += *lenp; 271 p = buffer; 272 while ((p - buffer) < *lenp && len < maxlen - 1) { 273 c = *(p++); 274 if (c == 0 || c == '\n') 275 break; 276 data[len++] = c; 277 } 278 data[len] = 0; 279 } else { 280 len = strlen(data); 281 if (len > maxlen) 282 len = maxlen; 283 284 if (*ppos > len) { 285 *lenp = 0; 286 return 0; 287 } 288 289 data += *ppos; 290 len -= *ppos; 291 292 if (len > *lenp) 293 len = *lenp; 294 if (len) 295 memcpy(buffer, data, len); 296 if (len < *lenp) { 297 buffer[len] = '\n'; 298 len++; 299 } 300 *lenp = len; 301 *ppos += len; 302 } 303 return 0; 304 } 305 306 static void warn_sysctl_write(struct ctl_table *table) 307 { 308 pr_warn_once("%s wrote to %s when file position was not 0!\n" 309 "This will not be supported in the future. To silence this\n" 310 "warning, set kernel.sysctl_writes_strict = -1\n", 311 current->comm, table->procname); 312 } 313 314 /** 315 * proc_first_pos_non_zero_ignore - check if first position is allowed 316 * @ppos: file position 317 * @table: the sysctl table 318 * 319 * Returns true if the first position is non-zero and the sysctl_writes_strict 320 * mode indicates this is not allowed for numeric input types. String proc 321 * handlers can ignore the return value. 322 */ 323 static bool proc_first_pos_non_zero_ignore(loff_t *ppos, 324 struct ctl_table *table) 325 { 326 if (!*ppos) 327 return false; 328 329 switch (sysctl_writes_strict) { 330 case SYSCTL_WRITES_STRICT: 331 return true; 332 case SYSCTL_WRITES_WARN: 333 warn_sysctl_write(table); 334 return false; 335 default: 336 return false; 337 } 338 } 339 340 /** 341 * proc_dostring - read a string sysctl 342 * @table: the sysctl table 343 * @write: %TRUE if this is a write to the sysctl file 344 * @buffer: the user buffer 345 * @lenp: the size of the user buffer 346 * @ppos: file position 347 * 348 * Reads/writes a string from/to the user buffer. If the kernel 349 * buffer provided is not large enough to hold the string, the 350 * string is truncated. The copied string is %NULL-terminated. 351 * If the string is being read by the user process, it is copied 352 * and a newline '\n' is added. It is truncated if the buffer is 353 * not large enough. 354 * 355 * Returns 0 on success. 356 */ 357 int proc_dostring(struct ctl_table *table, int write, 358 void *buffer, size_t *lenp, loff_t *ppos) 359 { 360 if (write) 361 proc_first_pos_non_zero_ignore(ppos, table); 362 363 return _proc_do_string(table->data, table->maxlen, write, buffer, lenp, 364 ppos); 365 } 366 367 static size_t proc_skip_spaces(char **buf) 368 { 369 size_t ret; 370 char *tmp = skip_spaces(*buf); 371 ret = tmp - *buf; 372 *buf = tmp; 373 return ret; 374 } 375 376 static void proc_skip_char(char **buf, size_t *size, const char v) 377 { 378 while (*size) { 379 if (**buf != v) 380 break; 381 (*size)--; 382 (*buf)++; 383 } 384 } 385 386 /** 387 * strtoul_lenient - parse an ASCII formatted integer from a buffer and only 388 * fail on overflow 389 * 390 * @cp: kernel buffer containing the string to parse 391 * @endp: pointer to store the trailing characters 392 * @base: the base to use 393 * @res: where the parsed integer will be stored 394 * 395 * In case of success 0 is returned and @res will contain the parsed integer, 396 * @endp will hold any trailing characters. 397 * This function will fail the parse on overflow. If there wasn't an overflow 398 * the function will defer the decision what characters count as invalid to the 399 * caller. 400 */ 401 static int strtoul_lenient(const char *cp, char **endp, unsigned int base, 402 unsigned long *res) 403 { 404 unsigned long long result; 405 unsigned int rv; 406 407 cp = _parse_integer_fixup_radix(cp, &base); 408 rv = _parse_integer(cp, base, &result); 409 if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result)) 410 return -ERANGE; 411 412 cp += rv; 413 414 if (endp) 415 *endp = (char *)cp; 416 417 *res = (unsigned long)result; 418 return 0; 419 } 420 421 #define TMPBUFLEN 22 422 /** 423 * proc_get_long - reads an ASCII formatted integer from a user buffer 424 * 425 * @buf: a kernel buffer 426 * @size: size of the kernel buffer 427 * @val: this is where the number will be stored 428 * @neg: set to %TRUE if number is negative 429 * @perm_tr: a vector which contains the allowed trailers 430 * @perm_tr_len: size of the perm_tr vector 431 * @tr: pointer to store the trailer character 432 * 433 * In case of success %0 is returned and @buf and @size are updated with 434 * the amount of bytes read. If @tr is non-NULL and a trailing 435 * character exists (size is non-zero after returning from this 436 * function), @tr is updated with the trailing character. 437 */ 438 static int proc_get_long(char **buf, size_t *size, 439 unsigned long *val, bool *neg, 440 const char *perm_tr, unsigned perm_tr_len, char *tr) 441 { 442 int len; 443 char *p, tmp[TMPBUFLEN]; 444 445 if (!*size) 446 return -EINVAL; 447 448 len = *size; 449 if (len > TMPBUFLEN - 1) 450 len = TMPBUFLEN - 1; 451 452 memcpy(tmp, *buf, len); 453 454 tmp[len] = 0; 455 p = tmp; 456 if (*p == '-' && *size > 1) { 457 *neg = true; 458 p++; 459 } else 460 *neg = false; 461 if (!isdigit(*p)) 462 return -EINVAL; 463 464 if (strtoul_lenient(p, &p, 0, val)) 465 return -EINVAL; 466 467 len = p - tmp; 468 469 /* We don't know if the next char is whitespace thus we may accept 470 * invalid integers (e.g. 1234...a) or two integers instead of one 471 * (e.g. 123...1). So lets not allow such large numbers. */ 472 if (len == TMPBUFLEN - 1) 473 return -EINVAL; 474 475 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len)) 476 return -EINVAL; 477 478 if (tr && (len < *size)) 479 *tr = *p; 480 481 *buf += len; 482 *size -= len; 483 484 return 0; 485 } 486 487 /** 488 * proc_put_long - converts an integer to a decimal ASCII formatted string 489 * 490 * @buf: the user buffer 491 * @size: the size of the user buffer 492 * @val: the integer to be converted 493 * @neg: sign of the number, %TRUE for negative 494 * 495 * In case of success @buf and @size are updated with the amount of bytes 496 * written. 497 */ 498 static void proc_put_long(void **buf, size_t *size, unsigned long val, bool neg) 499 { 500 int len; 501 char tmp[TMPBUFLEN], *p = tmp; 502 503 sprintf(p, "%s%lu", neg ? "-" : "", val); 504 len = strlen(tmp); 505 if (len > *size) 506 len = *size; 507 memcpy(*buf, tmp, len); 508 *size -= len; 509 *buf += len; 510 } 511 #undef TMPBUFLEN 512 513 static void proc_put_char(void **buf, size_t *size, char c) 514 { 515 if (*size) { 516 char **buffer = (char **)buf; 517 **buffer = c; 518 519 (*size)--; 520 (*buffer)++; 521 *buf = *buffer; 522 } 523 } 524 525 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp, 526 int *valp, 527 int write, void *data) 528 { 529 if (write) { 530 if (*negp) { 531 if (*lvalp > (unsigned long) INT_MAX + 1) 532 return -EINVAL; 533 *valp = -*lvalp; 534 } else { 535 if (*lvalp > (unsigned long) INT_MAX) 536 return -EINVAL; 537 *valp = *lvalp; 538 } 539 } else { 540 int val = *valp; 541 if (val < 0) { 542 *negp = true; 543 *lvalp = -(unsigned long)val; 544 } else { 545 *negp = false; 546 *lvalp = (unsigned long)val; 547 } 548 } 549 return 0; 550 } 551 552 static int do_proc_douintvec_conv(unsigned long *lvalp, 553 unsigned int *valp, 554 int write, void *data) 555 { 556 if (write) { 557 if (*lvalp > UINT_MAX) 558 return -EINVAL; 559 *valp = *lvalp; 560 } else { 561 unsigned int val = *valp; 562 *lvalp = (unsigned long)val; 563 } 564 return 0; 565 } 566 567 static const char proc_wspace_sep[] = { ' ', '\t', '\n' }; 568 569 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table, 570 int write, void *buffer, 571 size_t *lenp, loff_t *ppos, 572 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 573 int write, void *data), 574 void *data) 575 { 576 int *i, vleft, first = 1, err = 0; 577 size_t left; 578 char *p; 579 580 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) { 581 *lenp = 0; 582 return 0; 583 } 584 585 i = (int *) tbl_data; 586 vleft = table->maxlen / sizeof(*i); 587 left = *lenp; 588 589 if (!conv) 590 conv = do_proc_dointvec_conv; 591 592 if (write) { 593 if (proc_first_pos_non_zero_ignore(ppos, table)) 594 goto out; 595 596 if (left > PAGE_SIZE - 1) 597 left = PAGE_SIZE - 1; 598 p = buffer; 599 } 600 601 for (; left && vleft--; i++, first=0) { 602 unsigned long lval; 603 bool neg; 604 605 if (write) { 606 left -= proc_skip_spaces(&p); 607 608 if (!left) 609 break; 610 err = proc_get_long(&p, &left, &lval, &neg, 611 proc_wspace_sep, 612 sizeof(proc_wspace_sep), NULL); 613 if (err) 614 break; 615 if (conv(&neg, &lval, i, 1, data)) { 616 err = -EINVAL; 617 break; 618 } 619 } else { 620 if (conv(&neg, &lval, i, 0, data)) { 621 err = -EINVAL; 622 break; 623 } 624 if (!first) 625 proc_put_char(&buffer, &left, '\t'); 626 proc_put_long(&buffer, &left, lval, neg); 627 } 628 } 629 630 if (!write && !first && left && !err) 631 proc_put_char(&buffer, &left, '\n'); 632 if (write && !err && left) 633 left -= proc_skip_spaces(&p); 634 if (write && first) 635 return err ? : -EINVAL; 636 *lenp -= left; 637 out: 638 *ppos += *lenp; 639 return err; 640 } 641 642 static int do_proc_dointvec(struct ctl_table *table, int write, 643 void *buffer, size_t *lenp, loff_t *ppos, 644 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 645 int write, void *data), 646 void *data) 647 { 648 return __do_proc_dointvec(table->data, table, write, 649 buffer, lenp, ppos, conv, data); 650 } 651 652 static int do_proc_douintvec_w(unsigned int *tbl_data, 653 struct ctl_table *table, 654 void *buffer, 655 size_t *lenp, loff_t *ppos, 656 int (*conv)(unsigned long *lvalp, 657 unsigned int *valp, 658 int write, void *data), 659 void *data) 660 { 661 unsigned long lval; 662 int err = 0; 663 size_t left; 664 bool neg; 665 char *p = buffer; 666 667 left = *lenp; 668 669 if (proc_first_pos_non_zero_ignore(ppos, table)) 670 goto bail_early; 671 672 if (left > PAGE_SIZE - 1) 673 left = PAGE_SIZE - 1; 674 675 left -= proc_skip_spaces(&p); 676 if (!left) { 677 err = -EINVAL; 678 goto out_free; 679 } 680 681 err = proc_get_long(&p, &left, &lval, &neg, 682 proc_wspace_sep, 683 sizeof(proc_wspace_sep), NULL); 684 if (err || neg) { 685 err = -EINVAL; 686 goto out_free; 687 } 688 689 if (conv(&lval, tbl_data, 1, data)) { 690 err = -EINVAL; 691 goto out_free; 692 } 693 694 if (!err && left) 695 left -= proc_skip_spaces(&p); 696 697 out_free: 698 if (err) 699 return -EINVAL; 700 701 return 0; 702 703 /* This is in keeping with old __do_proc_dointvec() */ 704 bail_early: 705 *ppos += *lenp; 706 return err; 707 } 708 709 static int do_proc_douintvec_r(unsigned int *tbl_data, void *buffer, 710 size_t *lenp, loff_t *ppos, 711 int (*conv)(unsigned long *lvalp, 712 unsigned int *valp, 713 int write, void *data), 714 void *data) 715 { 716 unsigned long lval; 717 int err = 0; 718 size_t left; 719 720 left = *lenp; 721 722 if (conv(&lval, tbl_data, 0, data)) { 723 err = -EINVAL; 724 goto out; 725 } 726 727 proc_put_long(&buffer, &left, lval, false); 728 if (!left) 729 goto out; 730 731 proc_put_char(&buffer, &left, '\n'); 732 733 out: 734 *lenp -= left; 735 *ppos += *lenp; 736 737 return err; 738 } 739 740 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table, 741 int write, void *buffer, 742 size_t *lenp, loff_t *ppos, 743 int (*conv)(unsigned long *lvalp, 744 unsigned int *valp, 745 int write, void *data), 746 void *data) 747 { 748 unsigned int *i, vleft; 749 750 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) { 751 *lenp = 0; 752 return 0; 753 } 754 755 i = (unsigned int *) tbl_data; 756 vleft = table->maxlen / sizeof(*i); 757 758 /* 759 * Arrays are not supported, keep this simple. *Do not* add 760 * support for them. 761 */ 762 if (vleft != 1) { 763 *lenp = 0; 764 return -EINVAL; 765 } 766 767 if (!conv) 768 conv = do_proc_douintvec_conv; 769 770 if (write) 771 return do_proc_douintvec_w(i, table, buffer, lenp, ppos, 772 conv, data); 773 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data); 774 } 775 776 static int do_proc_douintvec(struct ctl_table *table, int write, 777 void *buffer, size_t *lenp, loff_t *ppos, 778 int (*conv)(unsigned long *lvalp, 779 unsigned int *valp, 780 int write, void *data), 781 void *data) 782 { 783 return __do_proc_douintvec(table->data, table, write, 784 buffer, lenp, ppos, conv, data); 785 } 786 787 /** 788 * proc_dointvec - read a vector of integers 789 * @table: the sysctl table 790 * @write: %TRUE if this is a write to the sysctl file 791 * @buffer: the user buffer 792 * @lenp: the size of the user buffer 793 * @ppos: file position 794 * 795 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 796 * values from/to the user buffer, treated as an ASCII string. 797 * 798 * Returns 0 on success. 799 */ 800 int proc_dointvec(struct ctl_table *table, int write, void *buffer, 801 size_t *lenp, loff_t *ppos) 802 { 803 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL); 804 } 805 806 #ifdef CONFIG_COMPACTION 807 static int proc_dointvec_minmax_warn_RT_change(struct ctl_table *table, 808 int write, void *buffer, size_t *lenp, loff_t *ppos) 809 { 810 int ret, old; 811 812 if (!IS_ENABLED(CONFIG_PREEMPT_RT) || !write) 813 return proc_dointvec_minmax(table, write, buffer, lenp, ppos); 814 815 old = *(int *)table->data; 816 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos); 817 if (ret) 818 return ret; 819 if (old != *(int *)table->data) 820 pr_warn_once("sysctl attribute %s changed by %s[%d]\n", 821 table->procname, current->comm, 822 task_pid_nr(current)); 823 return ret; 824 } 825 #endif 826 827 /** 828 * proc_douintvec - read a vector of unsigned integers 829 * @table: the sysctl table 830 * @write: %TRUE if this is a write to the sysctl file 831 * @buffer: the user buffer 832 * @lenp: the size of the user buffer 833 * @ppos: file position 834 * 835 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer 836 * values from/to the user buffer, treated as an ASCII string. 837 * 838 * Returns 0 on success. 839 */ 840 int proc_douintvec(struct ctl_table *table, int write, void *buffer, 841 size_t *lenp, loff_t *ppos) 842 { 843 return do_proc_douintvec(table, write, buffer, lenp, ppos, 844 do_proc_douintvec_conv, NULL); 845 } 846 847 /* 848 * Taint values can only be increased 849 * This means we can safely use a temporary. 850 */ 851 static int proc_taint(struct ctl_table *table, int write, 852 void *buffer, size_t *lenp, loff_t *ppos) 853 { 854 struct ctl_table t; 855 unsigned long tmptaint = get_taint(); 856 int err; 857 858 if (write && !capable(CAP_SYS_ADMIN)) 859 return -EPERM; 860 861 t = *table; 862 t.data = &tmptaint; 863 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos); 864 if (err < 0) 865 return err; 866 867 if (write) { 868 /* 869 * Poor man's atomic or. Not worth adding a primitive 870 * to everyone's atomic.h for this 871 */ 872 int i; 873 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) { 874 if ((tmptaint >> i) & 1) 875 add_taint(i, LOCKDEP_STILL_OK); 876 } 877 } 878 879 return err; 880 } 881 882 #ifdef CONFIG_PRINTK 883 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write, 884 void *buffer, size_t *lenp, loff_t *ppos) 885 { 886 if (write && !capable(CAP_SYS_ADMIN)) 887 return -EPERM; 888 889 return proc_dointvec_minmax(table, write, buffer, lenp, ppos); 890 } 891 #endif 892 893 /** 894 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure 895 * @min: pointer to minimum allowable value 896 * @max: pointer to maximum allowable value 897 * 898 * The do_proc_dointvec_minmax_conv_param structure provides the 899 * minimum and maximum values for doing range checking for those sysctl 900 * parameters that use the proc_dointvec_minmax() handler. 901 */ 902 struct do_proc_dointvec_minmax_conv_param { 903 int *min; 904 int *max; 905 }; 906 907 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp, 908 int *valp, 909 int write, void *data) 910 { 911 int tmp, ret; 912 struct do_proc_dointvec_minmax_conv_param *param = data; 913 /* 914 * If writing, first do so via a temporary local int so we can 915 * bounds-check it before touching *valp. 916 */ 917 int *ip = write ? &tmp : valp; 918 919 ret = do_proc_dointvec_conv(negp, lvalp, ip, 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 -EINVAL; 927 *valp = tmp; 928 } 929 930 return 0; 931 } 932 933 /** 934 * proc_dointvec_minmax - read a vector of integers with min/max values 935 * @table: the sysctl table 936 * @write: %TRUE if this is a write to the sysctl file 937 * @buffer: the user buffer 938 * @lenp: the size of the user buffer 939 * @ppos: file position 940 * 941 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 942 * values from/to the user buffer, treated as an ASCII string. 943 * 944 * This routine will ensure the values are within the range specified by 945 * table->extra1 (min) and table->extra2 (max). 946 * 947 * Returns 0 on success or -EINVAL on write when the range check fails. 948 */ 949 int proc_dointvec_minmax(struct ctl_table *table, int write, 950 void *buffer, size_t *lenp, loff_t *ppos) 951 { 952 struct do_proc_dointvec_minmax_conv_param param = { 953 .min = (int *) table->extra1, 954 .max = (int *) table->extra2, 955 }; 956 return do_proc_dointvec(table, write, buffer, lenp, ppos, 957 do_proc_dointvec_minmax_conv, ¶m); 958 } 959 960 /** 961 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure 962 * @min: pointer to minimum allowable value 963 * @max: pointer to maximum allowable value 964 * 965 * The do_proc_douintvec_minmax_conv_param structure provides the 966 * minimum and maximum values for doing range checking for those sysctl 967 * parameters that use the proc_douintvec_minmax() handler. 968 */ 969 struct do_proc_douintvec_minmax_conv_param { 970 unsigned int *min; 971 unsigned int *max; 972 }; 973 974 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp, 975 unsigned int *valp, 976 int write, void *data) 977 { 978 int ret; 979 unsigned int tmp; 980 struct do_proc_douintvec_minmax_conv_param *param = data; 981 /* write via temporary local uint for bounds-checking */ 982 unsigned int *up = write ? &tmp : valp; 983 984 ret = do_proc_douintvec_conv(lvalp, up, write, data); 985 if (ret) 986 return ret; 987 988 if (write) { 989 if ((param->min && *param->min > tmp) || 990 (param->max && *param->max < tmp)) 991 return -ERANGE; 992 993 *valp = tmp; 994 } 995 996 return 0; 997 } 998 999 /** 1000 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values 1001 * @table: the sysctl table 1002 * @write: %TRUE if this is a write to the sysctl file 1003 * @buffer: the user buffer 1004 * @lenp: the size of the user buffer 1005 * @ppos: file position 1006 * 1007 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer 1008 * values from/to the user buffer, treated as an ASCII string. Negative 1009 * strings are not allowed. 1010 * 1011 * This routine will ensure the values are within the range specified by 1012 * table->extra1 (min) and table->extra2 (max). There is a final sanity 1013 * check for UINT_MAX to avoid having to support wrap around uses from 1014 * userspace. 1015 * 1016 * Returns 0 on success or -ERANGE on write when the range check fails. 1017 */ 1018 int proc_douintvec_minmax(struct ctl_table *table, int write, 1019 void *buffer, size_t *lenp, loff_t *ppos) 1020 { 1021 struct do_proc_douintvec_minmax_conv_param param = { 1022 .min = (unsigned int *) table->extra1, 1023 .max = (unsigned int *) table->extra2, 1024 }; 1025 return do_proc_douintvec(table, write, buffer, lenp, ppos, 1026 do_proc_douintvec_minmax_conv, ¶m); 1027 } 1028 1029 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp, 1030 unsigned int *valp, 1031 int write, void *data) 1032 { 1033 if (write) { 1034 unsigned int val; 1035 1036 val = round_pipe_size(*lvalp); 1037 if (val == 0) 1038 return -EINVAL; 1039 1040 *valp = val; 1041 } else { 1042 unsigned int val = *valp; 1043 *lvalp = (unsigned long) val; 1044 } 1045 1046 return 0; 1047 } 1048 1049 static int proc_dopipe_max_size(struct ctl_table *table, int write, 1050 void *buffer, size_t *lenp, loff_t *ppos) 1051 { 1052 return do_proc_douintvec(table, write, buffer, lenp, ppos, 1053 do_proc_dopipe_max_size_conv, NULL); 1054 } 1055 1056 static void validate_coredump_safety(void) 1057 { 1058 #ifdef CONFIG_COREDUMP 1059 if (suid_dumpable == SUID_DUMP_ROOT && 1060 core_pattern[0] != '/' && core_pattern[0] != '|') { 1061 printk(KERN_WARNING 1062 "Unsafe core_pattern used with fs.suid_dumpable=2.\n" 1063 "Pipe handler or fully qualified core dump path required.\n" 1064 "Set kernel.core_pattern before fs.suid_dumpable.\n" 1065 ); 1066 } 1067 #endif 1068 } 1069 1070 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write, 1071 void *buffer, size_t *lenp, loff_t *ppos) 1072 { 1073 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos); 1074 if (!error) 1075 validate_coredump_safety(); 1076 return error; 1077 } 1078 1079 #ifdef CONFIG_COREDUMP 1080 static int proc_dostring_coredump(struct ctl_table *table, int write, 1081 void *buffer, size_t *lenp, loff_t *ppos) 1082 { 1083 int error = proc_dostring(table, write, buffer, lenp, ppos); 1084 if (!error) 1085 validate_coredump_safety(); 1086 return error; 1087 } 1088 #endif 1089 1090 #ifdef CONFIG_MAGIC_SYSRQ 1091 static int sysrq_sysctl_handler(struct ctl_table *table, int write, 1092 void *buffer, size_t *lenp, loff_t *ppos) 1093 { 1094 int tmp, ret; 1095 1096 tmp = sysrq_mask(); 1097 1098 ret = __do_proc_dointvec(&tmp, table, write, buffer, 1099 lenp, ppos, NULL, NULL); 1100 if (ret || !write) 1101 return ret; 1102 1103 if (write) 1104 sysrq_toggle_support(tmp); 1105 1106 return 0; 1107 } 1108 #endif 1109 1110 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, 1111 int write, void *buffer, size_t *lenp, loff_t *ppos, 1112 unsigned long convmul, unsigned long convdiv) 1113 { 1114 unsigned long *i, *min, *max; 1115 int vleft, first = 1, err = 0; 1116 size_t left; 1117 char *p; 1118 1119 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) { 1120 *lenp = 0; 1121 return 0; 1122 } 1123 1124 i = (unsigned long *) data; 1125 min = (unsigned long *) table->extra1; 1126 max = (unsigned long *) table->extra2; 1127 vleft = table->maxlen / sizeof(unsigned long); 1128 left = *lenp; 1129 1130 if (write) { 1131 if (proc_first_pos_non_zero_ignore(ppos, table)) 1132 goto out; 1133 1134 if (left > PAGE_SIZE - 1) 1135 left = PAGE_SIZE - 1; 1136 p = buffer; 1137 } 1138 1139 for (; left && vleft--; i++, first = 0) { 1140 unsigned long val; 1141 1142 if (write) { 1143 bool neg; 1144 1145 left -= proc_skip_spaces(&p); 1146 if (!left) 1147 break; 1148 1149 err = proc_get_long(&p, &left, &val, &neg, 1150 proc_wspace_sep, 1151 sizeof(proc_wspace_sep), NULL); 1152 if (err) 1153 break; 1154 if (neg) 1155 continue; 1156 val = convmul * val / convdiv; 1157 if ((min && val < *min) || (max && val > *max)) { 1158 err = -EINVAL; 1159 break; 1160 } 1161 *i = val; 1162 } else { 1163 val = convdiv * (*i) / convmul; 1164 if (!first) 1165 proc_put_char(&buffer, &left, '\t'); 1166 proc_put_long(&buffer, &left, val, false); 1167 } 1168 } 1169 1170 if (!write && !first && left && !err) 1171 proc_put_char(&buffer, &left, '\n'); 1172 if (write && !err) 1173 left -= proc_skip_spaces(&p); 1174 if (write && first) 1175 return err ? : -EINVAL; 1176 *lenp -= left; 1177 out: 1178 *ppos += *lenp; 1179 return err; 1180 } 1181 1182 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write, 1183 void *buffer, size_t *lenp, loff_t *ppos, unsigned long convmul, 1184 unsigned long convdiv) 1185 { 1186 return __do_proc_doulongvec_minmax(table->data, table, write, 1187 buffer, lenp, ppos, convmul, convdiv); 1188 } 1189 1190 /** 1191 * proc_doulongvec_minmax - read a vector of long integers with min/max values 1192 * @table: the sysctl table 1193 * @write: %TRUE if this is a write to the sysctl file 1194 * @buffer: the user buffer 1195 * @lenp: the size of the user buffer 1196 * @ppos: file position 1197 * 1198 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 1199 * values from/to the user buffer, treated as an ASCII string. 1200 * 1201 * This routine will ensure the values are within the range specified by 1202 * table->extra1 (min) and table->extra2 (max). 1203 * 1204 * Returns 0 on success. 1205 */ 1206 int proc_doulongvec_minmax(struct ctl_table *table, int write, 1207 void *buffer, size_t *lenp, loff_t *ppos) 1208 { 1209 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l); 1210 } 1211 1212 /** 1213 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values 1214 * @table: the sysctl table 1215 * @write: %TRUE if this is a write to the sysctl file 1216 * @buffer: the user buffer 1217 * @lenp: the size of the user buffer 1218 * @ppos: file position 1219 * 1220 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 1221 * values from/to the user buffer, treated as an ASCII string. The values 1222 * are treated as milliseconds, and converted to jiffies when they are stored. 1223 * 1224 * This routine will ensure the values are within the range specified by 1225 * table->extra1 (min) and table->extra2 (max). 1226 * 1227 * Returns 0 on success. 1228 */ 1229 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 1230 void *buffer, size_t *lenp, loff_t *ppos) 1231 { 1232 return do_proc_doulongvec_minmax(table, write, buffer, 1233 lenp, ppos, HZ, 1000l); 1234 } 1235 1236 1237 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp, 1238 int *valp, 1239 int write, void *data) 1240 { 1241 if (write) { 1242 if (*lvalp > INT_MAX / HZ) 1243 return 1; 1244 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ); 1245 } else { 1246 int val = *valp; 1247 unsigned long lval; 1248 if (val < 0) { 1249 *negp = true; 1250 lval = -(unsigned long)val; 1251 } else { 1252 *negp = false; 1253 lval = (unsigned long)val; 1254 } 1255 *lvalp = lval / HZ; 1256 } 1257 return 0; 1258 } 1259 1260 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp, 1261 int *valp, 1262 int write, void *data) 1263 { 1264 if (write) { 1265 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ) 1266 return 1; 1267 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp); 1268 } else { 1269 int val = *valp; 1270 unsigned long lval; 1271 if (val < 0) { 1272 *negp = true; 1273 lval = -(unsigned long)val; 1274 } else { 1275 *negp = false; 1276 lval = (unsigned long)val; 1277 } 1278 *lvalp = jiffies_to_clock_t(lval); 1279 } 1280 return 0; 1281 } 1282 1283 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp, 1284 int *valp, 1285 int write, void *data) 1286 { 1287 if (write) { 1288 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp); 1289 1290 if (jif > INT_MAX) 1291 return 1; 1292 *valp = (int)jif; 1293 } else { 1294 int val = *valp; 1295 unsigned long lval; 1296 if (val < 0) { 1297 *negp = true; 1298 lval = -(unsigned long)val; 1299 } else { 1300 *negp = false; 1301 lval = (unsigned long)val; 1302 } 1303 *lvalp = jiffies_to_msecs(lval); 1304 } 1305 return 0; 1306 } 1307 1308 /** 1309 * proc_dointvec_jiffies - read a vector of integers as seconds 1310 * @table: the sysctl table 1311 * @write: %TRUE if this is a write to the sysctl file 1312 * @buffer: the user buffer 1313 * @lenp: the size of the user buffer 1314 * @ppos: file position 1315 * 1316 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 1317 * values from/to the user buffer, treated as an ASCII string. 1318 * The values read are assumed to be in seconds, and are converted into 1319 * jiffies. 1320 * 1321 * Returns 0 on success. 1322 */ 1323 int proc_dointvec_jiffies(struct ctl_table *table, int write, 1324 void *buffer, size_t *lenp, loff_t *ppos) 1325 { 1326 return do_proc_dointvec(table,write,buffer,lenp,ppos, 1327 do_proc_dointvec_jiffies_conv,NULL); 1328 } 1329 1330 /** 1331 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds 1332 * @table: the sysctl table 1333 * @write: %TRUE if this is a write to the sysctl file 1334 * @buffer: the user buffer 1335 * @lenp: the size of the user buffer 1336 * @ppos: pointer to the file position 1337 * 1338 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 1339 * values from/to the user buffer, treated as an ASCII string. 1340 * The values read are assumed to be in 1/USER_HZ seconds, and 1341 * are converted into jiffies. 1342 * 1343 * Returns 0 on success. 1344 */ 1345 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 1346 void *buffer, size_t *lenp, loff_t *ppos) 1347 { 1348 return do_proc_dointvec(table,write,buffer,lenp,ppos, 1349 do_proc_dointvec_userhz_jiffies_conv,NULL); 1350 } 1351 1352 /** 1353 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds 1354 * @table: the sysctl table 1355 * @write: %TRUE if this is a write to the sysctl file 1356 * @buffer: the user buffer 1357 * @lenp: the size of the user buffer 1358 * @ppos: file position 1359 * @ppos: the current position in the file 1360 * 1361 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 1362 * values from/to the user buffer, treated as an ASCII string. 1363 * The values read are assumed to be in 1/1000 seconds, and 1364 * are converted into jiffies. 1365 * 1366 * Returns 0 on success. 1367 */ 1368 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, void *buffer, 1369 size_t *lenp, loff_t *ppos) 1370 { 1371 return do_proc_dointvec(table, write, buffer, lenp, ppos, 1372 do_proc_dointvec_ms_jiffies_conv, NULL); 1373 } 1374 1375 static int proc_do_cad_pid(struct ctl_table *table, int write, void *buffer, 1376 size_t *lenp, loff_t *ppos) 1377 { 1378 struct pid *new_pid; 1379 pid_t tmp; 1380 int r; 1381 1382 tmp = pid_vnr(cad_pid); 1383 1384 r = __do_proc_dointvec(&tmp, table, write, buffer, 1385 lenp, ppos, NULL, NULL); 1386 if (r || !write) 1387 return r; 1388 1389 new_pid = find_get_pid(tmp); 1390 if (!new_pid) 1391 return -ESRCH; 1392 1393 put_pid(xchg(&cad_pid, new_pid)); 1394 return 0; 1395 } 1396 1397 /** 1398 * proc_do_large_bitmap - read/write from/to a large bitmap 1399 * @table: the sysctl table 1400 * @write: %TRUE if this is a write to the sysctl file 1401 * @buffer: the user buffer 1402 * @lenp: the size of the user buffer 1403 * @ppos: file position 1404 * 1405 * The bitmap is stored at table->data and the bitmap length (in bits) 1406 * in table->maxlen. 1407 * 1408 * We use a range comma separated format (e.g. 1,3-4,10-10) so that 1409 * large bitmaps may be represented in a compact manner. Writing into 1410 * the file will clear the bitmap then update it with the given input. 1411 * 1412 * Returns 0 on success. 1413 */ 1414 int proc_do_large_bitmap(struct ctl_table *table, int write, 1415 void *buffer, size_t *lenp, loff_t *ppos) 1416 { 1417 int err = 0; 1418 bool first = 1; 1419 size_t left = *lenp; 1420 unsigned long bitmap_len = table->maxlen; 1421 unsigned long *bitmap = *(unsigned long **) table->data; 1422 unsigned long *tmp_bitmap = NULL; 1423 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c; 1424 1425 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) { 1426 *lenp = 0; 1427 return 0; 1428 } 1429 1430 if (write) { 1431 char *p = buffer; 1432 size_t skipped = 0; 1433 1434 if (left > PAGE_SIZE - 1) { 1435 left = PAGE_SIZE - 1; 1436 /* How much of the buffer we'll skip this pass */ 1437 skipped = *lenp - left; 1438 } 1439 1440 tmp_bitmap = bitmap_zalloc(bitmap_len, GFP_KERNEL); 1441 if (!tmp_bitmap) 1442 return -ENOMEM; 1443 proc_skip_char(&p, &left, '\n'); 1444 while (!err && left) { 1445 unsigned long val_a, val_b; 1446 bool neg; 1447 size_t saved_left; 1448 1449 /* In case we stop parsing mid-number, we can reset */ 1450 saved_left = left; 1451 err = proc_get_long(&p, &left, &val_a, &neg, tr_a, 1452 sizeof(tr_a), &c); 1453 /* 1454 * If we consumed the entirety of a truncated buffer or 1455 * only one char is left (may be a "-"), then stop here, 1456 * reset, & come back for more. 1457 */ 1458 if ((left <= 1) && skipped) { 1459 left = saved_left; 1460 break; 1461 } 1462 1463 if (err) 1464 break; 1465 if (val_a >= bitmap_len || neg) { 1466 err = -EINVAL; 1467 break; 1468 } 1469 1470 val_b = val_a; 1471 if (left) { 1472 p++; 1473 left--; 1474 } 1475 1476 if (c == '-') { 1477 err = proc_get_long(&p, &left, &val_b, 1478 &neg, tr_b, sizeof(tr_b), 1479 &c); 1480 /* 1481 * If we consumed all of a truncated buffer or 1482 * then stop here, reset, & come back for more. 1483 */ 1484 if (!left && skipped) { 1485 left = saved_left; 1486 break; 1487 } 1488 1489 if (err) 1490 break; 1491 if (val_b >= bitmap_len || neg || 1492 val_a > val_b) { 1493 err = -EINVAL; 1494 break; 1495 } 1496 if (left) { 1497 p++; 1498 left--; 1499 } 1500 } 1501 1502 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1); 1503 first = 0; 1504 proc_skip_char(&p, &left, '\n'); 1505 } 1506 left += skipped; 1507 } else { 1508 unsigned long bit_a, bit_b = 0; 1509 1510 while (left) { 1511 bit_a = find_next_bit(bitmap, bitmap_len, bit_b); 1512 if (bit_a >= bitmap_len) 1513 break; 1514 bit_b = find_next_zero_bit(bitmap, bitmap_len, 1515 bit_a + 1) - 1; 1516 1517 if (!first) 1518 proc_put_char(&buffer, &left, ','); 1519 proc_put_long(&buffer, &left, bit_a, false); 1520 if (bit_a != bit_b) { 1521 proc_put_char(&buffer, &left, '-'); 1522 proc_put_long(&buffer, &left, bit_b, false); 1523 } 1524 1525 first = 0; bit_b++; 1526 } 1527 proc_put_char(&buffer, &left, '\n'); 1528 } 1529 1530 if (!err) { 1531 if (write) { 1532 if (*ppos) 1533 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len); 1534 else 1535 bitmap_copy(bitmap, tmp_bitmap, bitmap_len); 1536 } 1537 *lenp -= left; 1538 *ppos += *lenp; 1539 } 1540 1541 bitmap_free(tmp_bitmap); 1542 return err; 1543 } 1544 1545 #else /* CONFIG_PROC_SYSCTL */ 1546 1547 int proc_dostring(struct ctl_table *table, int write, 1548 void *buffer, size_t *lenp, loff_t *ppos) 1549 { 1550 return -ENOSYS; 1551 } 1552 1553 int proc_dointvec(struct ctl_table *table, int write, 1554 void *buffer, size_t *lenp, loff_t *ppos) 1555 { 1556 return -ENOSYS; 1557 } 1558 1559 int proc_douintvec(struct ctl_table *table, int write, 1560 void *buffer, size_t *lenp, loff_t *ppos) 1561 { 1562 return -ENOSYS; 1563 } 1564 1565 int proc_dointvec_minmax(struct ctl_table *table, int write, 1566 void *buffer, size_t *lenp, loff_t *ppos) 1567 { 1568 return -ENOSYS; 1569 } 1570 1571 int proc_douintvec_minmax(struct ctl_table *table, int write, 1572 void *buffer, size_t *lenp, loff_t *ppos) 1573 { 1574 return -ENOSYS; 1575 } 1576 1577 int proc_dointvec_jiffies(struct ctl_table *table, int write, 1578 void *buffer, size_t *lenp, loff_t *ppos) 1579 { 1580 return -ENOSYS; 1581 } 1582 1583 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 1584 void *buffer, size_t *lenp, loff_t *ppos) 1585 { 1586 return -ENOSYS; 1587 } 1588 1589 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, 1590 void *buffer, size_t *lenp, loff_t *ppos) 1591 { 1592 return -ENOSYS; 1593 } 1594 1595 int proc_doulongvec_minmax(struct ctl_table *table, int write, 1596 void *buffer, size_t *lenp, loff_t *ppos) 1597 { 1598 return -ENOSYS; 1599 } 1600 1601 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 1602 void *buffer, size_t *lenp, loff_t *ppos) 1603 { 1604 return -ENOSYS; 1605 } 1606 1607 int proc_do_large_bitmap(struct ctl_table *table, int write, 1608 void *buffer, size_t *lenp, loff_t *ppos) 1609 { 1610 return -ENOSYS; 1611 } 1612 1613 #endif /* CONFIG_PROC_SYSCTL */ 1614 1615 #if defined(CONFIG_SYSCTL) 1616 int proc_do_static_key(struct ctl_table *table, int write, 1617 void *buffer, size_t *lenp, loff_t *ppos) 1618 { 1619 struct static_key *key = (struct static_key *)table->data; 1620 static DEFINE_MUTEX(static_key_mutex); 1621 int val, ret; 1622 struct ctl_table tmp = { 1623 .data = &val, 1624 .maxlen = sizeof(val), 1625 .mode = table->mode, 1626 .extra1 = SYSCTL_ZERO, 1627 .extra2 = SYSCTL_ONE, 1628 }; 1629 1630 if (write && !capable(CAP_SYS_ADMIN)) 1631 return -EPERM; 1632 1633 mutex_lock(&static_key_mutex); 1634 val = static_key_enabled(key); 1635 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos); 1636 if (write && !ret) { 1637 if (val) 1638 static_key_enable(key); 1639 else 1640 static_key_disable(key); 1641 } 1642 mutex_unlock(&static_key_mutex); 1643 return ret; 1644 } 1645 1646 static struct ctl_table kern_table[] = { 1647 { 1648 .procname = "sched_child_runs_first", 1649 .data = &sysctl_sched_child_runs_first, 1650 .maxlen = sizeof(unsigned int), 1651 .mode = 0644, 1652 .proc_handler = proc_dointvec, 1653 }, 1654 #ifdef CONFIG_SCHED_DEBUG 1655 { 1656 .procname = "sched_min_granularity_ns", 1657 .data = &sysctl_sched_min_granularity, 1658 .maxlen = sizeof(unsigned int), 1659 .mode = 0644, 1660 .proc_handler = sched_proc_update_handler, 1661 .extra1 = &min_sched_granularity_ns, 1662 .extra2 = &max_sched_granularity_ns, 1663 }, 1664 { 1665 .procname = "sched_latency_ns", 1666 .data = &sysctl_sched_latency, 1667 .maxlen = sizeof(unsigned int), 1668 .mode = 0644, 1669 .proc_handler = sched_proc_update_handler, 1670 .extra1 = &min_sched_granularity_ns, 1671 .extra2 = &max_sched_granularity_ns, 1672 }, 1673 { 1674 .procname = "sched_wakeup_granularity_ns", 1675 .data = &sysctl_sched_wakeup_granularity, 1676 .maxlen = sizeof(unsigned int), 1677 .mode = 0644, 1678 .proc_handler = sched_proc_update_handler, 1679 .extra1 = &min_wakeup_granularity_ns, 1680 .extra2 = &max_wakeup_granularity_ns, 1681 }, 1682 #ifdef CONFIG_SMP 1683 { 1684 .procname = "sched_tunable_scaling", 1685 .data = &sysctl_sched_tunable_scaling, 1686 .maxlen = sizeof(enum sched_tunable_scaling), 1687 .mode = 0644, 1688 .proc_handler = sched_proc_update_handler, 1689 .extra1 = &min_sched_tunable_scaling, 1690 .extra2 = &max_sched_tunable_scaling, 1691 }, 1692 { 1693 .procname = "sched_migration_cost_ns", 1694 .data = &sysctl_sched_migration_cost, 1695 .maxlen = sizeof(unsigned int), 1696 .mode = 0644, 1697 .proc_handler = proc_dointvec, 1698 }, 1699 { 1700 .procname = "sched_nr_migrate", 1701 .data = &sysctl_sched_nr_migrate, 1702 .maxlen = sizeof(unsigned int), 1703 .mode = 0644, 1704 .proc_handler = proc_dointvec, 1705 }, 1706 #ifdef CONFIG_SCHEDSTATS 1707 { 1708 .procname = "sched_schedstats", 1709 .data = NULL, 1710 .maxlen = sizeof(unsigned int), 1711 .mode = 0644, 1712 .proc_handler = sysctl_schedstats, 1713 .extra1 = SYSCTL_ZERO, 1714 .extra2 = SYSCTL_ONE, 1715 }, 1716 #endif /* CONFIG_SCHEDSTATS */ 1717 #endif /* CONFIG_SMP */ 1718 #ifdef CONFIG_NUMA_BALANCING 1719 { 1720 .procname = "numa_balancing_scan_delay_ms", 1721 .data = &sysctl_numa_balancing_scan_delay, 1722 .maxlen = sizeof(unsigned int), 1723 .mode = 0644, 1724 .proc_handler = proc_dointvec, 1725 }, 1726 { 1727 .procname = "numa_balancing_scan_period_min_ms", 1728 .data = &sysctl_numa_balancing_scan_period_min, 1729 .maxlen = sizeof(unsigned int), 1730 .mode = 0644, 1731 .proc_handler = proc_dointvec, 1732 }, 1733 { 1734 .procname = "numa_balancing_scan_period_max_ms", 1735 .data = &sysctl_numa_balancing_scan_period_max, 1736 .maxlen = sizeof(unsigned int), 1737 .mode = 0644, 1738 .proc_handler = proc_dointvec, 1739 }, 1740 { 1741 .procname = "numa_balancing_scan_size_mb", 1742 .data = &sysctl_numa_balancing_scan_size, 1743 .maxlen = sizeof(unsigned int), 1744 .mode = 0644, 1745 .proc_handler = proc_dointvec_minmax, 1746 .extra1 = SYSCTL_ONE, 1747 }, 1748 { 1749 .procname = "numa_balancing", 1750 .data = NULL, /* filled in by handler */ 1751 .maxlen = sizeof(unsigned int), 1752 .mode = 0644, 1753 .proc_handler = sysctl_numa_balancing, 1754 .extra1 = SYSCTL_ZERO, 1755 .extra2 = SYSCTL_ONE, 1756 }, 1757 #endif /* CONFIG_NUMA_BALANCING */ 1758 #endif /* CONFIG_SCHED_DEBUG */ 1759 { 1760 .procname = "sched_rt_period_us", 1761 .data = &sysctl_sched_rt_period, 1762 .maxlen = sizeof(unsigned int), 1763 .mode = 0644, 1764 .proc_handler = sched_rt_handler, 1765 }, 1766 { 1767 .procname = "sched_rt_runtime_us", 1768 .data = &sysctl_sched_rt_runtime, 1769 .maxlen = sizeof(int), 1770 .mode = 0644, 1771 .proc_handler = sched_rt_handler, 1772 }, 1773 { 1774 .procname = "sched_rr_timeslice_ms", 1775 .data = &sysctl_sched_rr_timeslice, 1776 .maxlen = sizeof(int), 1777 .mode = 0644, 1778 .proc_handler = sched_rr_handler, 1779 }, 1780 #ifdef CONFIG_UCLAMP_TASK 1781 { 1782 .procname = "sched_util_clamp_min", 1783 .data = &sysctl_sched_uclamp_util_min, 1784 .maxlen = sizeof(unsigned int), 1785 .mode = 0644, 1786 .proc_handler = sysctl_sched_uclamp_handler, 1787 }, 1788 { 1789 .procname = "sched_util_clamp_max", 1790 .data = &sysctl_sched_uclamp_util_max, 1791 .maxlen = sizeof(unsigned int), 1792 .mode = 0644, 1793 .proc_handler = sysctl_sched_uclamp_handler, 1794 }, 1795 #endif 1796 #ifdef CONFIG_SCHED_AUTOGROUP 1797 { 1798 .procname = "sched_autogroup_enabled", 1799 .data = &sysctl_sched_autogroup_enabled, 1800 .maxlen = sizeof(unsigned int), 1801 .mode = 0644, 1802 .proc_handler = proc_dointvec_minmax, 1803 .extra1 = SYSCTL_ZERO, 1804 .extra2 = SYSCTL_ONE, 1805 }, 1806 #endif 1807 #ifdef CONFIG_CFS_BANDWIDTH 1808 { 1809 .procname = "sched_cfs_bandwidth_slice_us", 1810 .data = &sysctl_sched_cfs_bandwidth_slice, 1811 .maxlen = sizeof(unsigned int), 1812 .mode = 0644, 1813 .proc_handler = proc_dointvec_minmax, 1814 .extra1 = SYSCTL_ONE, 1815 }, 1816 #endif 1817 #if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_CPU_FREQ_GOV_SCHEDUTIL) 1818 { 1819 .procname = "sched_energy_aware", 1820 .data = &sysctl_sched_energy_aware, 1821 .maxlen = sizeof(unsigned int), 1822 .mode = 0644, 1823 .proc_handler = sched_energy_aware_handler, 1824 .extra1 = SYSCTL_ZERO, 1825 .extra2 = SYSCTL_ONE, 1826 }, 1827 #endif 1828 #ifdef CONFIG_PROVE_LOCKING 1829 { 1830 .procname = "prove_locking", 1831 .data = &prove_locking, 1832 .maxlen = sizeof(int), 1833 .mode = 0644, 1834 .proc_handler = proc_dointvec, 1835 }, 1836 #endif 1837 #ifdef CONFIG_LOCK_STAT 1838 { 1839 .procname = "lock_stat", 1840 .data = &lock_stat, 1841 .maxlen = sizeof(int), 1842 .mode = 0644, 1843 .proc_handler = proc_dointvec, 1844 }, 1845 #endif 1846 { 1847 .procname = "panic", 1848 .data = &panic_timeout, 1849 .maxlen = sizeof(int), 1850 .mode = 0644, 1851 .proc_handler = proc_dointvec, 1852 }, 1853 #ifdef CONFIG_COREDUMP 1854 { 1855 .procname = "core_uses_pid", 1856 .data = &core_uses_pid, 1857 .maxlen = sizeof(int), 1858 .mode = 0644, 1859 .proc_handler = proc_dointvec, 1860 }, 1861 { 1862 .procname = "core_pattern", 1863 .data = core_pattern, 1864 .maxlen = CORENAME_MAX_SIZE, 1865 .mode = 0644, 1866 .proc_handler = proc_dostring_coredump, 1867 }, 1868 { 1869 .procname = "core_pipe_limit", 1870 .data = &core_pipe_limit, 1871 .maxlen = sizeof(unsigned int), 1872 .mode = 0644, 1873 .proc_handler = proc_dointvec, 1874 }, 1875 #endif 1876 #ifdef CONFIG_PROC_SYSCTL 1877 { 1878 .procname = "tainted", 1879 .maxlen = sizeof(long), 1880 .mode = 0644, 1881 .proc_handler = proc_taint, 1882 }, 1883 { 1884 .procname = "sysctl_writes_strict", 1885 .data = &sysctl_writes_strict, 1886 .maxlen = sizeof(int), 1887 .mode = 0644, 1888 .proc_handler = proc_dointvec_minmax, 1889 .extra1 = &neg_one, 1890 .extra2 = SYSCTL_ONE, 1891 }, 1892 #endif 1893 #ifdef CONFIG_LATENCYTOP 1894 { 1895 .procname = "latencytop", 1896 .data = &latencytop_enabled, 1897 .maxlen = sizeof(int), 1898 .mode = 0644, 1899 .proc_handler = sysctl_latencytop, 1900 }, 1901 #endif 1902 #ifdef CONFIG_BLK_DEV_INITRD 1903 { 1904 .procname = "real-root-dev", 1905 .data = &real_root_dev, 1906 .maxlen = sizeof(int), 1907 .mode = 0644, 1908 .proc_handler = proc_dointvec, 1909 }, 1910 #endif 1911 { 1912 .procname = "print-fatal-signals", 1913 .data = &print_fatal_signals, 1914 .maxlen = sizeof(int), 1915 .mode = 0644, 1916 .proc_handler = proc_dointvec, 1917 }, 1918 #ifdef CONFIG_SPARC 1919 { 1920 .procname = "reboot-cmd", 1921 .data = reboot_command, 1922 .maxlen = 256, 1923 .mode = 0644, 1924 .proc_handler = proc_dostring, 1925 }, 1926 { 1927 .procname = "stop-a", 1928 .data = &stop_a_enabled, 1929 .maxlen = sizeof (int), 1930 .mode = 0644, 1931 .proc_handler = proc_dointvec, 1932 }, 1933 { 1934 .procname = "scons-poweroff", 1935 .data = &scons_pwroff, 1936 .maxlen = sizeof (int), 1937 .mode = 0644, 1938 .proc_handler = proc_dointvec, 1939 }, 1940 #endif 1941 #ifdef CONFIG_SPARC64 1942 { 1943 .procname = "tsb-ratio", 1944 .data = &sysctl_tsb_ratio, 1945 .maxlen = sizeof (int), 1946 .mode = 0644, 1947 .proc_handler = proc_dointvec, 1948 }, 1949 #endif 1950 #ifdef CONFIG_PARISC 1951 { 1952 .procname = "soft-power", 1953 .data = &pwrsw_enabled, 1954 .maxlen = sizeof (int), 1955 .mode = 0644, 1956 .proc_handler = proc_dointvec, 1957 }, 1958 #endif 1959 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW 1960 { 1961 .procname = "unaligned-trap", 1962 .data = &unaligned_enabled, 1963 .maxlen = sizeof (int), 1964 .mode = 0644, 1965 .proc_handler = proc_dointvec, 1966 }, 1967 #endif 1968 { 1969 .procname = "ctrl-alt-del", 1970 .data = &C_A_D, 1971 .maxlen = sizeof(int), 1972 .mode = 0644, 1973 .proc_handler = proc_dointvec, 1974 }, 1975 #ifdef CONFIG_FUNCTION_TRACER 1976 { 1977 .procname = "ftrace_enabled", 1978 .data = &ftrace_enabled, 1979 .maxlen = sizeof(int), 1980 .mode = 0644, 1981 .proc_handler = ftrace_enable_sysctl, 1982 }, 1983 #endif 1984 #ifdef CONFIG_STACK_TRACER 1985 { 1986 .procname = "stack_tracer_enabled", 1987 .data = &stack_tracer_enabled, 1988 .maxlen = sizeof(int), 1989 .mode = 0644, 1990 .proc_handler = stack_trace_sysctl, 1991 }, 1992 #endif 1993 #ifdef CONFIG_TRACING 1994 { 1995 .procname = "ftrace_dump_on_oops", 1996 .data = &ftrace_dump_on_oops, 1997 .maxlen = sizeof(int), 1998 .mode = 0644, 1999 .proc_handler = proc_dointvec, 2000 }, 2001 { 2002 .procname = "traceoff_on_warning", 2003 .data = &__disable_trace_on_warning, 2004 .maxlen = sizeof(__disable_trace_on_warning), 2005 .mode = 0644, 2006 .proc_handler = proc_dointvec, 2007 }, 2008 { 2009 .procname = "tracepoint_printk", 2010 .data = &tracepoint_printk, 2011 .maxlen = sizeof(tracepoint_printk), 2012 .mode = 0644, 2013 .proc_handler = tracepoint_printk_sysctl, 2014 }, 2015 #endif 2016 #ifdef CONFIG_KEXEC_CORE 2017 { 2018 .procname = "kexec_load_disabled", 2019 .data = &kexec_load_disabled, 2020 .maxlen = sizeof(int), 2021 .mode = 0644, 2022 /* only handle a transition from default "0" to "1" */ 2023 .proc_handler = proc_dointvec_minmax, 2024 .extra1 = SYSCTL_ONE, 2025 .extra2 = SYSCTL_ONE, 2026 }, 2027 #endif 2028 #ifdef CONFIG_MODULES 2029 { 2030 .procname = "modprobe", 2031 .data = &modprobe_path, 2032 .maxlen = KMOD_PATH_LEN, 2033 .mode = 0644, 2034 .proc_handler = proc_dostring, 2035 }, 2036 { 2037 .procname = "modules_disabled", 2038 .data = &modules_disabled, 2039 .maxlen = sizeof(int), 2040 .mode = 0644, 2041 /* only handle a transition from default "0" to "1" */ 2042 .proc_handler = proc_dointvec_minmax, 2043 .extra1 = SYSCTL_ONE, 2044 .extra2 = SYSCTL_ONE, 2045 }, 2046 #endif 2047 #ifdef CONFIG_UEVENT_HELPER 2048 { 2049 .procname = "hotplug", 2050 .data = &uevent_helper, 2051 .maxlen = UEVENT_HELPER_PATH_LEN, 2052 .mode = 0644, 2053 .proc_handler = proc_dostring, 2054 }, 2055 #endif 2056 #ifdef CONFIG_CHR_DEV_SG 2057 { 2058 .procname = "sg-big-buff", 2059 .data = &sg_big_buff, 2060 .maxlen = sizeof (int), 2061 .mode = 0444, 2062 .proc_handler = proc_dointvec, 2063 }, 2064 #endif 2065 #ifdef CONFIG_BSD_PROCESS_ACCT 2066 { 2067 .procname = "acct", 2068 .data = &acct_parm, 2069 .maxlen = 3*sizeof(int), 2070 .mode = 0644, 2071 .proc_handler = proc_dointvec, 2072 }, 2073 #endif 2074 #ifdef CONFIG_MAGIC_SYSRQ 2075 { 2076 .procname = "sysrq", 2077 .data = NULL, 2078 .maxlen = sizeof (int), 2079 .mode = 0644, 2080 .proc_handler = sysrq_sysctl_handler, 2081 }, 2082 #endif 2083 #ifdef CONFIG_PROC_SYSCTL 2084 { 2085 .procname = "cad_pid", 2086 .data = NULL, 2087 .maxlen = sizeof (int), 2088 .mode = 0600, 2089 .proc_handler = proc_do_cad_pid, 2090 }, 2091 #endif 2092 { 2093 .procname = "threads-max", 2094 .data = NULL, 2095 .maxlen = sizeof(int), 2096 .mode = 0644, 2097 .proc_handler = sysctl_max_threads, 2098 }, 2099 { 2100 .procname = "random", 2101 .mode = 0555, 2102 .child = random_table, 2103 }, 2104 { 2105 .procname = "usermodehelper", 2106 .mode = 0555, 2107 .child = usermodehelper_table, 2108 }, 2109 #ifdef CONFIG_FW_LOADER_USER_HELPER 2110 { 2111 .procname = "firmware_config", 2112 .mode = 0555, 2113 .child = firmware_config_table, 2114 }, 2115 #endif 2116 { 2117 .procname = "overflowuid", 2118 .data = &overflowuid, 2119 .maxlen = sizeof(int), 2120 .mode = 0644, 2121 .proc_handler = proc_dointvec_minmax, 2122 .extra1 = &minolduid, 2123 .extra2 = &maxolduid, 2124 }, 2125 { 2126 .procname = "overflowgid", 2127 .data = &overflowgid, 2128 .maxlen = sizeof(int), 2129 .mode = 0644, 2130 .proc_handler = proc_dointvec_minmax, 2131 .extra1 = &minolduid, 2132 .extra2 = &maxolduid, 2133 }, 2134 #ifdef CONFIG_S390 2135 { 2136 .procname = "userprocess_debug", 2137 .data = &show_unhandled_signals, 2138 .maxlen = sizeof(int), 2139 .mode = 0644, 2140 .proc_handler = proc_dointvec, 2141 }, 2142 #endif 2143 { 2144 .procname = "pid_max", 2145 .data = &pid_max, 2146 .maxlen = sizeof (int), 2147 .mode = 0644, 2148 .proc_handler = proc_dointvec_minmax, 2149 .extra1 = &pid_max_min, 2150 .extra2 = &pid_max_max, 2151 }, 2152 { 2153 .procname = "panic_on_oops", 2154 .data = &panic_on_oops, 2155 .maxlen = sizeof(int), 2156 .mode = 0644, 2157 .proc_handler = proc_dointvec, 2158 }, 2159 { 2160 .procname = "panic_print", 2161 .data = &panic_print, 2162 .maxlen = sizeof(unsigned long), 2163 .mode = 0644, 2164 .proc_handler = proc_doulongvec_minmax, 2165 }, 2166 #if defined CONFIG_PRINTK 2167 { 2168 .procname = "printk", 2169 .data = &console_loglevel, 2170 .maxlen = 4*sizeof(int), 2171 .mode = 0644, 2172 .proc_handler = proc_dointvec, 2173 }, 2174 { 2175 .procname = "printk_ratelimit", 2176 .data = &printk_ratelimit_state.interval, 2177 .maxlen = sizeof(int), 2178 .mode = 0644, 2179 .proc_handler = proc_dointvec_jiffies, 2180 }, 2181 { 2182 .procname = "printk_ratelimit_burst", 2183 .data = &printk_ratelimit_state.burst, 2184 .maxlen = sizeof(int), 2185 .mode = 0644, 2186 .proc_handler = proc_dointvec, 2187 }, 2188 { 2189 .procname = "printk_delay", 2190 .data = &printk_delay_msec, 2191 .maxlen = sizeof(int), 2192 .mode = 0644, 2193 .proc_handler = proc_dointvec_minmax, 2194 .extra1 = SYSCTL_ZERO, 2195 .extra2 = &ten_thousand, 2196 }, 2197 { 2198 .procname = "printk_devkmsg", 2199 .data = devkmsg_log_str, 2200 .maxlen = DEVKMSG_STR_MAX_SIZE, 2201 .mode = 0644, 2202 .proc_handler = devkmsg_sysctl_set_loglvl, 2203 }, 2204 { 2205 .procname = "dmesg_restrict", 2206 .data = &dmesg_restrict, 2207 .maxlen = sizeof(int), 2208 .mode = 0644, 2209 .proc_handler = proc_dointvec_minmax_sysadmin, 2210 .extra1 = SYSCTL_ZERO, 2211 .extra2 = SYSCTL_ONE, 2212 }, 2213 { 2214 .procname = "kptr_restrict", 2215 .data = &kptr_restrict, 2216 .maxlen = sizeof(int), 2217 .mode = 0644, 2218 .proc_handler = proc_dointvec_minmax_sysadmin, 2219 .extra1 = SYSCTL_ZERO, 2220 .extra2 = &two, 2221 }, 2222 #endif 2223 { 2224 .procname = "ngroups_max", 2225 .data = &ngroups_max, 2226 .maxlen = sizeof (int), 2227 .mode = 0444, 2228 .proc_handler = proc_dointvec, 2229 }, 2230 { 2231 .procname = "cap_last_cap", 2232 .data = (void *)&cap_last_cap, 2233 .maxlen = sizeof(int), 2234 .mode = 0444, 2235 .proc_handler = proc_dointvec, 2236 }, 2237 #if defined(CONFIG_LOCKUP_DETECTOR) 2238 { 2239 .procname = "watchdog", 2240 .data = &watchdog_user_enabled, 2241 .maxlen = sizeof(int), 2242 .mode = 0644, 2243 .proc_handler = proc_watchdog, 2244 .extra1 = SYSCTL_ZERO, 2245 .extra2 = SYSCTL_ONE, 2246 }, 2247 { 2248 .procname = "watchdog_thresh", 2249 .data = &watchdog_thresh, 2250 .maxlen = sizeof(int), 2251 .mode = 0644, 2252 .proc_handler = proc_watchdog_thresh, 2253 .extra1 = SYSCTL_ZERO, 2254 .extra2 = &sixty, 2255 }, 2256 { 2257 .procname = "nmi_watchdog", 2258 .data = &nmi_watchdog_user_enabled, 2259 .maxlen = sizeof(int), 2260 .mode = NMI_WATCHDOG_SYSCTL_PERM, 2261 .proc_handler = proc_nmi_watchdog, 2262 .extra1 = SYSCTL_ZERO, 2263 .extra2 = SYSCTL_ONE, 2264 }, 2265 { 2266 .procname = "watchdog_cpumask", 2267 .data = &watchdog_cpumask_bits, 2268 .maxlen = NR_CPUS, 2269 .mode = 0644, 2270 .proc_handler = proc_watchdog_cpumask, 2271 }, 2272 #ifdef CONFIG_SOFTLOCKUP_DETECTOR 2273 { 2274 .procname = "soft_watchdog", 2275 .data = &soft_watchdog_user_enabled, 2276 .maxlen = sizeof(int), 2277 .mode = 0644, 2278 .proc_handler = proc_soft_watchdog, 2279 .extra1 = SYSCTL_ZERO, 2280 .extra2 = SYSCTL_ONE, 2281 }, 2282 { 2283 .procname = "softlockup_panic", 2284 .data = &softlockup_panic, 2285 .maxlen = sizeof(int), 2286 .mode = 0644, 2287 .proc_handler = proc_dointvec_minmax, 2288 .extra1 = SYSCTL_ZERO, 2289 .extra2 = SYSCTL_ONE, 2290 }, 2291 #ifdef CONFIG_SMP 2292 { 2293 .procname = "softlockup_all_cpu_backtrace", 2294 .data = &sysctl_softlockup_all_cpu_backtrace, 2295 .maxlen = sizeof(int), 2296 .mode = 0644, 2297 .proc_handler = proc_dointvec_minmax, 2298 .extra1 = SYSCTL_ZERO, 2299 .extra2 = SYSCTL_ONE, 2300 }, 2301 #endif /* CONFIG_SMP */ 2302 #endif 2303 #ifdef CONFIG_HARDLOCKUP_DETECTOR 2304 { 2305 .procname = "hardlockup_panic", 2306 .data = &hardlockup_panic, 2307 .maxlen = sizeof(int), 2308 .mode = 0644, 2309 .proc_handler = proc_dointvec_minmax, 2310 .extra1 = SYSCTL_ZERO, 2311 .extra2 = SYSCTL_ONE, 2312 }, 2313 #ifdef CONFIG_SMP 2314 { 2315 .procname = "hardlockup_all_cpu_backtrace", 2316 .data = &sysctl_hardlockup_all_cpu_backtrace, 2317 .maxlen = sizeof(int), 2318 .mode = 0644, 2319 .proc_handler = proc_dointvec_minmax, 2320 .extra1 = SYSCTL_ZERO, 2321 .extra2 = SYSCTL_ONE, 2322 }, 2323 #endif /* CONFIG_SMP */ 2324 #endif 2325 #endif 2326 2327 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86) 2328 { 2329 .procname = "unknown_nmi_panic", 2330 .data = &unknown_nmi_panic, 2331 .maxlen = sizeof (int), 2332 .mode = 0644, 2333 .proc_handler = proc_dointvec, 2334 }, 2335 #endif 2336 #if defined(CONFIG_X86) 2337 { 2338 .procname = "panic_on_unrecovered_nmi", 2339 .data = &panic_on_unrecovered_nmi, 2340 .maxlen = sizeof(int), 2341 .mode = 0644, 2342 .proc_handler = proc_dointvec, 2343 }, 2344 { 2345 .procname = "panic_on_io_nmi", 2346 .data = &panic_on_io_nmi, 2347 .maxlen = sizeof(int), 2348 .mode = 0644, 2349 .proc_handler = proc_dointvec, 2350 }, 2351 #ifdef CONFIG_DEBUG_STACKOVERFLOW 2352 { 2353 .procname = "panic_on_stackoverflow", 2354 .data = &sysctl_panic_on_stackoverflow, 2355 .maxlen = sizeof(int), 2356 .mode = 0644, 2357 .proc_handler = proc_dointvec, 2358 }, 2359 #endif 2360 { 2361 .procname = "bootloader_type", 2362 .data = &bootloader_type, 2363 .maxlen = sizeof (int), 2364 .mode = 0444, 2365 .proc_handler = proc_dointvec, 2366 }, 2367 { 2368 .procname = "bootloader_version", 2369 .data = &bootloader_version, 2370 .maxlen = sizeof (int), 2371 .mode = 0444, 2372 .proc_handler = proc_dointvec, 2373 }, 2374 { 2375 .procname = "io_delay_type", 2376 .data = &io_delay_type, 2377 .maxlen = sizeof(int), 2378 .mode = 0644, 2379 .proc_handler = proc_dointvec, 2380 }, 2381 #endif 2382 #if defined(CONFIG_MMU) 2383 { 2384 .procname = "randomize_va_space", 2385 .data = &randomize_va_space, 2386 .maxlen = sizeof(int), 2387 .mode = 0644, 2388 .proc_handler = proc_dointvec, 2389 }, 2390 #endif 2391 #if defined(CONFIG_S390) && defined(CONFIG_SMP) 2392 { 2393 .procname = "spin_retry", 2394 .data = &spin_retry, 2395 .maxlen = sizeof (int), 2396 .mode = 0644, 2397 .proc_handler = proc_dointvec, 2398 }, 2399 #endif 2400 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86) 2401 { 2402 .procname = "acpi_video_flags", 2403 .data = &acpi_realmode_flags, 2404 .maxlen = sizeof (unsigned long), 2405 .mode = 0644, 2406 .proc_handler = proc_doulongvec_minmax, 2407 }, 2408 #endif 2409 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN 2410 { 2411 .procname = "ignore-unaligned-usertrap", 2412 .data = &no_unaligned_warning, 2413 .maxlen = sizeof (int), 2414 .mode = 0644, 2415 .proc_handler = proc_dointvec, 2416 }, 2417 #endif 2418 #ifdef CONFIG_IA64 2419 { 2420 .procname = "unaligned-dump-stack", 2421 .data = &unaligned_dump_stack, 2422 .maxlen = sizeof (int), 2423 .mode = 0644, 2424 .proc_handler = proc_dointvec, 2425 }, 2426 #endif 2427 #ifdef CONFIG_DETECT_HUNG_TASK 2428 { 2429 .procname = "hung_task_panic", 2430 .data = &sysctl_hung_task_panic, 2431 .maxlen = sizeof(int), 2432 .mode = 0644, 2433 .proc_handler = proc_dointvec_minmax, 2434 .extra1 = SYSCTL_ZERO, 2435 .extra2 = SYSCTL_ONE, 2436 }, 2437 { 2438 .procname = "hung_task_check_count", 2439 .data = &sysctl_hung_task_check_count, 2440 .maxlen = sizeof(int), 2441 .mode = 0644, 2442 .proc_handler = proc_dointvec_minmax, 2443 .extra1 = SYSCTL_ZERO, 2444 }, 2445 { 2446 .procname = "hung_task_timeout_secs", 2447 .data = &sysctl_hung_task_timeout_secs, 2448 .maxlen = sizeof(unsigned long), 2449 .mode = 0644, 2450 .proc_handler = proc_dohung_task_timeout_secs, 2451 .extra2 = &hung_task_timeout_max, 2452 }, 2453 { 2454 .procname = "hung_task_check_interval_secs", 2455 .data = &sysctl_hung_task_check_interval_secs, 2456 .maxlen = sizeof(unsigned long), 2457 .mode = 0644, 2458 .proc_handler = proc_dohung_task_timeout_secs, 2459 .extra2 = &hung_task_timeout_max, 2460 }, 2461 { 2462 .procname = "hung_task_warnings", 2463 .data = &sysctl_hung_task_warnings, 2464 .maxlen = sizeof(int), 2465 .mode = 0644, 2466 .proc_handler = proc_dointvec_minmax, 2467 .extra1 = &neg_one, 2468 }, 2469 #endif 2470 #ifdef CONFIG_RT_MUTEXES 2471 { 2472 .procname = "max_lock_depth", 2473 .data = &max_lock_depth, 2474 .maxlen = sizeof(int), 2475 .mode = 0644, 2476 .proc_handler = proc_dointvec, 2477 }, 2478 #endif 2479 { 2480 .procname = "poweroff_cmd", 2481 .data = &poweroff_cmd, 2482 .maxlen = POWEROFF_CMD_PATH_LEN, 2483 .mode = 0644, 2484 .proc_handler = proc_dostring, 2485 }, 2486 #ifdef CONFIG_KEYS 2487 { 2488 .procname = "keys", 2489 .mode = 0555, 2490 .child = key_sysctls, 2491 }, 2492 #endif 2493 #ifdef CONFIG_PERF_EVENTS 2494 /* 2495 * User-space scripts rely on the existence of this file 2496 * as a feature check for perf_events being enabled. 2497 * 2498 * So it's an ABI, do not remove! 2499 */ 2500 { 2501 .procname = "perf_event_paranoid", 2502 .data = &sysctl_perf_event_paranoid, 2503 .maxlen = sizeof(sysctl_perf_event_paranoid), 2504 .mode = 0644, 2505 .proc_handler = proc_dointvec, 2506 }, 2507 { 2508 .procname = "perf_event_mlock_kb", 2509 .data = &sysctl_perf_event_mlock, 2510 .maxlen = sizeof(sysctl_perf_event_mlock), 2511 .mode = 0644, 2512 .proc_handler = proc_dointvec, 2513 }, 2514 { 2515 .procname = "perf_event_max_sample_rate", 2516 .data = &sysctl_perf_event_sample_rate, 2517 .maxlen = sizeof(sysctl_perf_event_sample_rate), 2518 .mode = 0644, 2519 .proc_handler = perf_proc_update_handler, 2520 .extra1 = SYSCTL_ONE, 2521 }, 2522 { 2523 .procname = "perf_cpu_time_max_percent", 2524 .data = &sysctl_perf_cpu_time_max_percent, 2525 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent), 2526 .mode = 0644, 2527 .proc_handler = perf_cpu_time_max_percent_handler, 2528 .extra1 = SYSCTL_ZERO, 2529 .extra2 = &one_hundred, 2530 }, 2531 { 2532 .procname = "perf_event_max_stack", 2533 .data = &sysctl_perf_event_max_stack, 2534 .maxlen = sizeof(sysctl_perf_event_max_stack), 2535 .mode = 0644, 2536 .proc_handler = perf_event_max_stack_handler, 2537 .extra1 = SYSCTL_ZERO, 2538 .extra2 = &six_hundred_forty_kb, 2539 }, 2540 { 2541 .procname = "perf_event_max_contexts_per_stack", 2542 .data = &sysctl_perf_event_max_contexts_per_stack, 2543 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack), 2544 .mode = 0644, 2545 .proc_handler = perf_event_max_stack_handler, 2546 .extra1 = SYSCTL_ZERO, 2547 .extra2 = &one_thousand, 2548 }, 2549 #endif 2550 { 2551 .procname = "panic_on_warn", 2552 .data = &panic_on_warn, 2553 .maxlen = sizeof(int), 2554 .mode = 0644, 2555 .proc_handler = proc_dointvec_minmax, 2556 .extra1 = SYSCTL_ZERO, 2557 .extra2 = SYSCTL_ONE, 2558 }, 2559 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) 2560 { 2561 .procname = "timer_migration", 2562 .data = &sysctl_timer_migration, 2563 .maxlen = sizeof(unsigned int), 2564 .mode = 0644, 2565 .proc_handler = timer_migration_handler, 2566 .extra1 = SYSCTL_ZERO, 2567 .extra2 = SYSCTL_ONE, 2568 }, 2569 #endif 2570 #ifdef CONFIG_BPF_SYSCALL 2571 { 2572 .procname = "unprivileged_bpf_disabled", 2573 .data = &sysctl_unprivileged_bpf_disabled, 2574 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled), 2575 .mode = 0644, 2576 /* only handle a transition from default "0" to "1" */ 2577 .proc_handler = proc_dointvec_minmax, 2578 .extra1 = SYSCTL_ONE, 2579 .extra2 = SYSCTL_ONE, 2580 }, 2581 { 2582 .procname = "bpf_stats_enabled", 2583 .data = &bpf_stats_enabled_key.key, 2584 .maxlen = sizeof(bpf_stats_enabled_key), 2585 .mode = 0644, 2586 .proc_handler = bpf_stats_handler, 2587 }, 2588 #endif 2589 #if defined(CONFIG_TREE_RCU) 2590 { 2591 .procname = "panic_on_rcu_stall", 2592 .data = &sysctl_panic_on_rcu_stall, 2593 .maxlen = sizeof(sysctl_panic_on_rcu_stall), 2594 .mode = 0644, 2595 .proc_handler = proc_dointvec_minmax, 2596 .extra1 = SYSCTL_ZERO, 2597 .extra2 = SYSCTL_ONE, 2598 }, 2599 #endif 2600 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE 2601 { 2602 .procname = "stack_erasing", 2603 .data = NULL, 2604 .maxlen = sizeof(int), 2605 .mode = 0600, 2606 .proc_handler = stack_erasing_sysctl, 2607 .extra1 = SYSCTL_ZERO, 2608 .extra2 = SYSCTL_ONE, 2609 }, 2610 #endif 2611 { } 2612 }; 2613 2614 static struct ctl_table vm_table[] = { 2615 { 2616 .procname = "overcommit_memory", 2617 .data = &sysctl_overcommit_memory, 2618 .maxlen = sizeof(sysctl_overcommit_memory), 2619 .mode = 0644, 2620 .proc_handler = proc_dointvec_minmax, 2621 .extra1 = SYSCTL_ZERO, 2622 .extra2 = &two, 2623 }, 2624 { 2625 .procname = "panic_on_oom", 2626 .data = &sysctl_panic_on_oom, 2627 .maxlen = sizeof(sysctl_panic_on_oom), 2628 .mode = 0644, 2629 .proc_handler = proc_dointvec_minmax, 2630 .extra1 = SYSCTL_ZERO, 2631 .extra2 = &two, 2632 }, 2633 { 2634 .procname = "oom_kill_allocating_task", 2635 .data = &sysctl_oom_kill_allocating_task, 2636 .maxlen = sizeof(sysctl_oom_kill_allocating_task), 2637 .mode = 0644, 2638 .proc_handler = proc_dointvec, 2639 }, 2640 { 2641 .procname = "oom_dump_tasks", 2642 .data = &sysctl_oom_dump_tasks, 2643 .maxlen = sizeof(sysctl_oom_dump_tasks), 2644 .mode = 0644, 2645 .proc_handler = proc_dointvec, 2646 }, 2647 { 2648 .procname = "overcommit_ratio", 2649 .data = &sysctl_overcommit_ratio, 2650 .maxlen = sizeof(sysctl_overcommit_ratio), 2651 .mode = 0644, 2652 .proc_handler = overcommit_ratio_handler, 2653 }, 2654 { 2655 .procname = "overcommit_kbytes", 2656 .data = &sysctl_overcommit_kbytes, 2657 .maxlen = sizeof(sysctl_overcommit_kbytes), 2658 .mode = 0644, 2659 .proc_handler = overcommit_kbytes_handler, 2660 }, 2661 { 2662 .procname = "page-cluster", 2663 .data = &page_cluster, 2664 .maxlen = sizeof(int), 2665 .mode = 0644, 2666 .proc_handler = proc_dointvec_minmax, 2667 .extra1 = SYSCTL_ZERO, 2668 }, 2669 { 2670 .procname = "dirty_background_ratio", 2671 .data = &dirty_background_ratio, 2672 .maxlen = sizeof(dirty_background_ratio), 2673 .mode = 0644, 2674 .proc_handler = dirty_background_ratio_handler, 2675 .extra1 = SYSCTL_ZERO, 2676 .extra2 = &one_hundred, 2677 }, 2678 { 2679 .procname = "dirty_background_bytes", 2680 .data = &dirty_background_bytes, 2681 .maxlen = sizeof(dirty_background_bytes), 2682 .mode = 0644, 2683 .proc_handler = dirty_background_bytes_handler, 2684 .extra1 = &one_ul, 2685 }, 2686 { 2687 .procname = "dirty_ratio", 2688 .data = &vm_dirty_ratio, 2689 .maxlen = sizeof(vm_dirty_ratio), 2690 .mode = 0644, 2691 .proc_handler = dirty_ratio_handler, 2692 .extra1 = SYSCTL_ZERO, 2693 .extra2 = &one_hundred, 2694 }, 2695 { 2696 .procname = "dirty_bytes", 2697 .data = &vm_dirty_bytes, 2698 .maxlen = sizeof(vm_dirty_bytes), 2699 .mode = 0644, 2700 .proc_handler = dirty_bytes_handler, 2701 .extra1 = &dirty_bytes_min, 2702 }, 2703 { 2704 .procname = "dirty_writeback_centisecs", 2705 .data = &dirty_writeback_interval, 2706 .maxlen = sizeof(dirty_writeback_interval), 2707 .mode = 0644, 2708 .proc_handler = dirty_writeback_centisecs_handler, 2709 }, 2710 { 2711 .procname = "dirty_expire_centisecs", 2712 .data = &dirty_expire_interval, 2713 .maxlen = sizeof(dirty_expire_interval), 2714 .mode = 0644, 2715 .proc_handler = proc_dointvec_minmax, 2716 .extra1 = SYSCTL_ZERO, 2717 }, 2718 { 2719 .procname = "dirtytime_expire_seconds", 2720 .data = &dirtytime_expire_interval, 2721 .maxlen = sizeof(dirtytime_expire_interval), 2722 .mode = 0644, 2723 .proc_handler = dirtytime_interval_handler, 2724 .extra1 = SYSCTL_ZERO, 2725 }, 2726 { 2727 .procname = "swappiness", 2728 .data = &vm_swappiness, 2729 .maxlen = sizeof(vm_swappiness), 2730 .mode = 0644, 2731 .proc_handler = proc_dointvec_minmax, 2732 .extra1 = SYSCTL_ZERO, 2733 .extra2 = &one_hundred, 2734 }, 2735 #ifdef CONFIG_HUGETLB_PAGE 2736 { 2737 .procname = "nr_hugepages", 2738 .data = NULL, 2739 .maxlen = sizeof(unsigned long), 2740 .mode = 0644, 2741 .proc_handler = hugetlb_sysctl_handler, 2742 }, 2743 #ifdef CONFIG_NUMA 2744 { 2745 .procname = "nr_hugepages_mempolicy", 2746 .data = NULL, 2747 .maxlen = sizeof(unsigned long), 2748 .mode = 0644, 2749 .proc_handler = &hugetlb_mempolicy_sysctl_handler, 2750 }, 2751 { 2752 .procname = "numa_stat", 2753 .data = &sysctl_vm_numa_stat, 2754 .maxlen = sizeof(int), 2755 .mode = 0644, 2756 .proc_handler = sysctl_vm_numa_stat_handler, 2757 .extra1 = SYSCTL_ZERO, 2758 .extra2 = SYSCTL_ONE, 2759 }, 2760 #endif 2761 { 2762 .procname = "hugetlb_shm_group", 2763 .data = &sysctl_hugetlb_shm_group, 2764 .maxlen = sizeof(gid_t), 2765 .mode = 0644, 2766 .proc_handler = proc_dointvec, 2767 }, 2768 { 2769 .procname = "nr_overcommit_hugepages", 2770 .data = NULL, 2771 .maxlen = sizeof(unsigned long), 2772 .mode = 0644, 2773 .proc_handler = hugetlb_overcommit_handler, 2774 }, 2775 #endif 2776 { 2777 .procname = "lowmem_reserve_ratio", 2778 .data = &sysctl_lowmem_reserve_ratio, 2779 .maxlen = sizeof(sysctl_lowmem_reserve_ratio), 2780 .mode = 0644, 2781 .proc_handler = lowmem_reserve_ratio_sysctl_handler, 2782 }, 2783 { 2784 .procname = "drop_caches", 2785 .data = &sysctl_drop_caches, 2786 .maxlen = sizeof(int), 2787 .mode = 0200, 2788 .proc_handler = drop_caches_sysctl_handler, 2789 .extra1 = SYSCTL_ONE, 2790 .extra2 = &four, 2791 }, 2792 #ifdef CONFIG_COMPACTION 2793 { 2794 .procname = "compact_memory", 2795 .data = &sysctl_compact_memory, 2796 .maxlen = sizeof(int), 2797 .mode = 0200, 2798 .proc_handler = sysctl_compaction_handler, 2799 }, 2800 { 2801 .procname = "extfrag_threshold", 2802 .data = &sysctl_extfrag_threshold, 2803 .maxlen = sizeof(int), 2804 .mode = 0644, 2805 .proc_handler = proc_dointvec_minmax, 2806 .extra1 = &min_extfrag_threshold, 2807 .extra2 = &max_extfrag_threshold, 2808 }, 2809 { 2810 .procname = "compact_unevictable_allowed", 2811 .data = &sysctl_compact_unevictable_allowed, 2812 .maxlen = sizeof(int), 2813 .mode = 0644, 2814 .proc_handler = proc_dointvec_minmax_warn_RT_change, 2815 .extra1 = SYSCTL_ZERO, 2816 .extra2 = SYSCTL_ONE, 2817 }, 2818 2819 #endif /* CONFIG_COMPACTION */ 2820 { 2821 .procname = "min_free_kbytes", 2822 .data = &min_free_kbytes, 2823 .maxlen = sizeof(min_free_kbytes), 2824 .mode = 0644, 2825 .proc_handler = min_free_kbytes_sysctl_handler, 2826 .extra1 = SYSCTL_ZERO, 2827 }, 2828 { 2829 .procname = "watermark_boost_factor", 2830 .data = &watermark_boost_factor, 2831 .maxlen = sizeof(watermark_boost_factor), 2832 .mode = 0644, 2833 .proc_handler = proc_dointvec_minmax, 2834 .extra1 = SYSCTL_ZERO, 2835 }, 2836 { 2837 .procname = "watermark_scale_factor", 2838 .data = &watermark_scale_factor, 2839 .maxlen = sizeof(watermark_scale_factor), 2840 .mode = 0644, 2841 .proc_handler = watermark_scale_factor_sysctl_handler, 2842 .extra1 = SYSCTL_ONE, 2843 .extra2 = &one_thousand, 2844 }, 2845 { 2846 .procname = "percpu_pagelist_fraction", 2847 .data = &percpu_pagelist_fraction, 2848 .maxlen = sizeof(percpu_pagelist_fraction), 2849 .mode = 0644, 2850 .proc_handler = percpu_pagelist_fraction_sysctl_handler, 2851 .extra1 = SYSCTL_ZERO, 2852 }, 2853 #ifdef CONFIG_MMU 2854 { 2855 .procname = "max_map_count", 2856 .data = &sysctl_max_map_count, 2857 .maxlen = sizeof(sysctl_max_map_count), 2858 .mode = 0644, 2859 .proc_handler = proc_dointvec_minmax, 2860 .extra1 = SYSCTL_ZERO, 2861 }, 2862 #else 2863 { 2864 .procname = "nr_trim_pages", 2865 .data = &sysctl_nr_trim_pages, 2866 .maxlen = sizeof(sysctl_nr_trim_pages), 2867 .mode = 0644, 2868 .proc_handler = proc_dointvec_minmax, 2869 .extra1 = SYSCTL_ZERO, 2870 }, 2871 #endif 2872 { 2873 .procname = "laptop_mode", 2874 .data = &laptop_mode, 2875 .maxlen = sizeof(laptop_mode), 2876 .mode = 0644, 2877 .proc_handler = proc_dointvec_jiffies, 2878 }, 2879 { 2880 .procname = "block_dump", 2881 .data = &block_dump, 2882 .maxlen = sizeof(block_dump), 2883 .mode = 0644, 2884 .proc_handler = proc_dointvec, 2885 .extra1 = SYSCTL_ZERO, 2886 }, 2887 { 2888 .procname = "vfs_cache_pressure", 2889 .data = &sysctl_vfs_cache_pressure, 2890 .maxlen = sizeof(sysctl_vfs_cache_pressure), 2891 .mode = 0644, 2892 .proc_handler = proc_dointvec, 2893 .extra1 = SYSCTL_ZERO, 2894 }, 2895 #if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \ 2896 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT) 2897 { 2898 .procname = "legacy_va_layout", 2899 .data = &sysctl_legacy_va_layout, 2900 .maxlen = sizeof(sysctl_legacy_va_layout), 2901 .mode = 0644, 2902 .proc_handler = proc_dointvec, 2903 .extra1 = SYSCTL_ZERO, 2904 }, 2905 #endif 2906 #ifdef CONFIG_NUMA 2907 { 2908 .procname = "zone_reclaim_mode", 2909 .data = &node_reclaim_mode, 2910 .maxlen = sizeof(node_reclaim_mode), 2911 .mode = 0644, 2912 .proc_handler = proc_dointvec, 2913 .extra1 = SYSCTL_ZERO, 2914 }, 2915 { 2916 .procname = "min_unmapped_ratio", 2917 .data = &sysctl_min_unmapped_ratio, 2918 .maxlen = sizeof(sysctl_min_unmapped_ratio), 2919 .mode = 0644, 2920 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler, 2921 .extra1 = SYSCTL_ZERO, 2922 .extra2 = &one_hundred, 2923 }, 2924 { 2925 .procname = "min_slab_ratio", 2926 .data = &sysctl_min_slab_ratio, 2927 .maxlen = sizeof(sysctl_min_slab_ratio), 2928 .mode = 0644, 2929 .proc_handler = sysctl_min_slab_ratio_sysctl_handler, 2930 .extra1 = SYSCTL_ZERO, 2931 .extra2 = &one_hundred, 2932 }, 2933 #endif 2934 #ifdef CONFIG_SMP 2935 { 2936 .procname = "stat_interval", 2937 .data = &sysctl_stat_interval, 2938 .maxlen = sizeof(sysctl_stat_interval), 2939 .mode = 0644, 2940 .proc_handler = proc_dointvec_jiffies, 2941 }, 2942 { 2943 .procname = "stat_refresh", 2944 .data = NULL, 2945 .maxlen = 0, 2946 .mode = 0600, 2947 .proc_handler = vmstat_refresh, 2948 }, 2949 #endif 2950 #ifdef CONFIG_MMU 2951 { 2952 .procname = "mmap_min_addr", 2953 .data = &dac_mmap_min_addr, 2954 .maxlen = sizeof(unsigned long), 2955 .mode = 0644, 2956 .proc_handler = mmap_min_addr_handler, 2957 }, 2958 #endif 2959 #ifdef CONFIG_NUMA 2960 { 2961 .procname = "numa_zonelist_order", 2962 .data = &numa_zonelist_order, 2963 .maxlen = NUMA_ZONELIST_ORDER_LEN, 2964 .mode = 0644, 2965 .proc_handler = numa_zonelist_order_handler, 2966 }, 2967 #endif 2968 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \ 2969 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL)) 2970 { 2971 .procname = "vdso_enabled", 2972 #ifdef CONFIG_X86_32 2973 .data = &vdso32_enabled, 2974 .maxlen = sizeof(vdso32_enabled), 2975 #else 2976 .data = &vdso_enabled, 2977 .maxlen = sizeof(vdso_enabled), 2978 #endif 2979 .mode = 0644, 2980 .proc_handler = proc_dointvec, 2981 .extra1 = SYSCTL_ZERO, 2982 }, 2983 #endif 2984 #ifdef CONFIG_HIGHMEM 2985 { 2986 .procname = "highmem_is_dirtyable", 2987 .data = &vm_highmem_is_dirtyable, 2988 .maxlen = sizeof(vm_highmem_is_dirtyable), 2989 .mode = 0644, 2990 .proc_handler = proc_dointvec_minmax, 2991 .extra1 = SYSCTL_ZERO, 2992 .extra2 = SYSCTL_ONE, 2993 }, 2994 #endif 2995 #ifdef CONFIG_MEMORY_FAILURE 2996 { 2997 .procname = "memory_failure_early_kill", 2998 .data = &sysctl_memory_failure_early_kill, 2999 .maxlen = sizeof(sysctl_memory_failure_early_kill), 3000 .mode = 0644, 3001 .proc_handler = proc_dointvec_minmax, 3002 .extra1 = SYSCTL_ZERO, 3003 .extra2 = SYSCTL_ONE, 3004 }, 3005 { 3006 .procname = "memory_failure_recovery", 3007 .data = &sysctl_memory_failure_recovery, 3008 .maxlen = sizeof(sysctl_memory_failure_recovery), 3009 .mode = 0644, 3010 .proc_handler = proc_dointvec_minmax, 3011 .extra1 = SYSCTL_ZERO, 3012 .extra2 = SYSCTL_ONE, 3013 }, 3014 #endif 3015 { 3016 .procname = "user_reserve_kbytes", 3017 .data = &sysctl_user_reserve_kbytes, 3018 .maxlen = sizeof(sysctl_user_reserve_kbytes), 3019 .mode = 0644, 3020 .proc_handler = proc_doulongvec_minmax, 3021 }, 3022 { 3023 .procname = "admin_reserve_kbytes", 3024 .data = &sysctl_admin_reserve_kbytes, 3025 .maxlen = sizeof(sysctl_admin_reserve_kbytes), 3026 .mode = 0644, 3027 .proc_handler = proc_doulongvec_minmax, 3028 }, 3029 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS 3030 { 3031 .procname = "mmap_rnd_bits", 3032 .data = &mmap_rnd_bits, 3033 .maxlen = sizeof(mmap_rnd_bits), 3034 .mode = 0600, 3035 .proc_handler = proc_dointvec_minmax, 3036 .extra1 = (void *)&mmap_rnd_bits_min, 3037 .extra2 = (void *)&mmap_rnd_bits_max, 3038 }, 3039 #endif 3040 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS 3041 { 3042 .procname = "mmap_rnd_compat_bits", 3043 .data = &mmap_rnd_compat_bits, 3044 .maxlen = sizeof(mmap_rnd_compat_bits), 3045 .mode = 0600, 3046 .proc_handler = proc_dointvec_minmax, 3047 .extra1 = (void *)&mmap_rnd_compat_bits_min, 3048 .extra2 = (void *)&mmap_rnd_compat_bits_max, 3049 }, 3050 #endif 3051 #ifdef CONFIG_USERFAULTFD 3052 { 3053 .procname = "unprivileged_userfaultfd", 3054 .data = &sysctl_unprivileged_userfaultfd, 3055 .maxlen = sizeof(sysctl_unprivileged_userfaultfd), 3056 .mode = 0644, 3057 .proc_handler = proc_dointvec_minmax, 3058 .extra1 = SYSCTL_ZERO, 3059 .extra2 = SYSCTL_ONE, 3060 }, 3061 #endif 3062 { } 3063 }; 3064 3065 static struct ctl_table fs_table[] = { 3066 { 3067 .procname = "inode-nr", 3068 .data = &inodes_stat, 3069 .maxlen = 2*sizeof(long), 3070 .mode = 0444, 3071 .proc_handler = proc_nr_inodes, 3072 }, 3073 { 3074 .procname = "inode-state", 3075 .data = &inodes_stat, 3076 .maxlen = 7*sizeof(long), 3077 .mode = 0444, 3078 .proc_handler = proc_nr_inodes, 3079 }, 3080 { 3081 .procname = "file-nr", 3082 .data = &files_stat, 3083 .maxlen = sizeof(files_stat), 3084 .mode = 0444, 3085 .proc_handler = proc_nr_files, 3086 }, 3087 { 3088 .procname = "file-max", 3089 .data = &files_stat.max_files, 3090 .maxlen = sizeof(files_stat.max_files), 3091 .mode = 0644, 3092 .proc_handler = proc_doulongvec_minmax, 3093 .extra1 = &zero_ul, 3094 .extra2 = &long_max, 3095 }, 3096 { 3097 .procname = "nr_open", 3098 .data = &sysctl_nr_open, 3099 .maxlen = sizeof(unsigned int), 3100 .mode = 0644, 3101 .proc_handler = proc_dointvec_minmax, 3102 .extra1 = &sysctl_nr_open_min, 3103 .extra2 = &sysctl_nr_open_max, 3104 }, 3105 { 3106 .procname = "dentry-state", 3107 .data = &dentry_stat, 3108 .maxlen = 6*sizeof(long), 3109 .mode = 0444, 3110 .proc_handler = proc_nr_dentry, 3111 }, 3112 { 3113 .procname = "overflowuid", 3114 .data = &fs_overflowuid, 3115 .maxlen = sizeof(int), 3116 .mode = 0644, 3117 .proc_handler = proc_dointvec_minmax, 3118 .extra1 = &minolduid, 3119 .extra2 = &maxolduid, 3120 }, 3121 { 3122 .procname = "overflowgid", 3123 .data = &fs_overflowgid, 3124 .maxlen = sizeof(int), 3125 .mode = 0644, 3126 .proc_handler = proc_dointvec_minmax, 3127 .extra1 = &minolduid, 3128 .extra2 = &maxolduid, 3129 }, 3130 #ifdef CONFIG_FILE_LOCKING 3131 { 3132 .procname = "leases-enable", 3133 .data = &leases_enable, 3134 .maxlen = sizeof(int), 3135 .mode = 0644, 3136 .proc_handler = proc_dointvec, 3137 }, 3138 #endif 3139 #ifdef CONFIG_DNOTIFY 3140 { 3141 .procname = "dir-notify-enable", 3142 .data = &dir_notify_enable, 3143 .maxlen = sizeof(int), 3144 .mode = 0644, 3145 .proc_handler = proc_dointvec, 3146 }, 3147 #endif 3148 #ifdef CONFIG_MMU 3149 #ifdef CONFIG_FILE_LOCKING 3150 { 3151 .procname = "lease-break-time", 3152 .data = &lease_break_time, 3153 .maxlen = sizeof(int), 3154 .mode = 0644, 3155 .proc_handler = proc_dointvec, 3156 }, 3157 #endif 3158 #ifdef CONFIG_AIO 3159 { 3160 .procname = "aio-nr", 3161 .data = &aio_nr, 3162 .maxlen = sizeof(aio_nr), 3163 .mode = 0444, 3164 .proc_handler = proc_doulongvec_minmax, 3165 }, 3166 { 3167 .procname = "aio-max-nr", 3168 .data = &aio_max_nr, 3169 .maxlen = sizeof(aio_max_nr), 3170 .mode = 0644, 3171 .proc_handler = proc_doulongvec_minmax, 3172 }, 3173 #endif /* CONFIG_AIO */ 3174 #ifdef CONFIG_INOTIFY_USER 3175 { 3176 .procname = "inotify", 3177 .mode = 0555, 3178 .child = inotify_table, 3179 }, 3180 #endif 3181 #ifdef CONFIG_EPOLL 3182 { 3183 .procname = "epoll", 3184 .mode = 0555, 3185 .child = epoll_table, 3186 }, 3187 #endif 3188 #endif 3189 { 3190 .procname = "protected_symlinks", 3191 .data = &sysctl_protected_symlinks, 3192 .maxlen = sizeof(int), 3193 .mode = 0600, 3194 .proc_handler = proc_dointvec_minmax, 3195 .extra1 = SYSCTL_ZERO, 3196 .extra2 = SYSCTL_ONE, 3197 }, 3198 { 3199 .procname = "protected_hardlinks", 3200 .data = &sysctl_protected_hardlinks, 3201 .maxlen = sizeof(int), 3202 .mode = 0600, 3203 .proc_handler = proc_dointvec_minmax, 3204 .extra1 = SYSCTL_ZERO, 3205 .extra2 = SYSCTL_ONE, 3206 }, 3207 { 3208 .procname = "protected_fifos", 3209 .data = &sysctl_protected_fifos, 3210 .maxlen = sizeof(int), 3211 .mode = 0600, 3212 .proc_handler = proc_dointvec_minmax, 3213 .extra1 = SYSCTL_ZERO, 3214 .extra2 = &two, 3215 }, 3216 { 3217 .procname = "protected_regular", 3218 .data = &sysctl_protected_regular, 3219 .maxlen = sizeof(int), 3220 .mode = 0600, 3221 .proc_handler = proc_dointvec_minmax, 3222 .extra1 = SYSCTL_ZERO, 3223 .extra2 = &two, 3224 }, 3225 { 3226 .procname = "suid_dumpable", 3227 .data = &suid_dumpable, 3228 .maxlen = sizeof(int), 3229 .mode = 0644, 3230 .proc_handler = proc_dointvec_minmax_coredump, 3231 .extra1 = SYSCTL_ZERO, 3232 .extra2 = &two, 3233 }, 3234 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE) 3235 { 3236 .procname = "binfmt_misc", 3237 .mode = 0555, 3238 .child = sysctl_mount_point, 3239 }, 3240 #endif 3241 { 3242 .procname = "pipe-max-size", 3243 .data = &pipe_max_size, 3244 .maxlen = sizeof(pipe_max_size), 3245 .mode = 0644, 3246 .proc_handler = proc_dopipe_max_size, 3247 }, 3248 { 3249 .procname = "pipe-user-pages-hard", 3250 .data = &pipe_user_pages_hard, 3251 .maxlen = sizeof(pipe_user_pages_hard), 3252 .mode = 0644, 3253 .proc_handler = proc_doulongvec_minmax, 3254 }, 3255 { 3256 .procname = "pipe-user-pages-soft", 3257 .data = &pipe_user_pages_soft, 3258 .maxlen = sizeof(pipe_user_pages_soft), 3259 .mode = 0644, 3260 .proc_handler = proc_doulongvec_minmax, 3261 }, 3262 { 3263 .procname = "mount-max", 3264 .data = &sysctl_mount_max, 3265 .maxlen = sizeof(unsigned int), 3266 .mode = 0644, 3267 .proc_handler = proc_dointvec_minmax, 3268 .extra1 = SYSCTL_ONE, 3269 }, 3270 { } 3271 }; 3272 3273 static struct ctl_table debug_table[] = { 3274 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE 3275 { 3276 .procname = "exception-trace", 3277 .data = &show_unhandled_signals, 3278 .maxlen = sizeof(int), 3279 .mode = 0644, 3280 .proc_handler = proc_dointvec 3281 }, 3282 #endif 3283 #if defined(CONFIG_OPTPROBES) 3284 { 3285 .procname = "kprobes-optimization", 3286 .data = &sysctl_kprobes_optimization, 3287 .maxlen = sizeof(int), 3288 .mode = 0644, 3289 .proc_handler = proc_kprobes_optimization_handler, 3290 .extra1 = SYSCTL_ZERO, 3291 .extra2 = SYSCTL_ONE, 3292 }, 3293 #endif 3294 { } 3295 }; 3296 3297 static struct ctl_table dev_table[] = { 3298 { } 3299 }; 3300 3301 static struct ctl_table sysctl_base_table[] = { 3302 { 3303 .procname = "kernel", 3304 .mode = 0555, 3305 .child = kern_table, 3306 }, 3307 { 3308 .procname = "vm", 3309 .mode = 0555, 3310 .child = vm_table, 3311 }, 3312 { 3313 .procname = "fs", 3314 .mode = 0555, 3315 .child = fs_table, 3316 }, 3317 { 3318 .procname = "debug", 3319 .mode = 0555, 3320 .child = debug_table, 3321 }, 3322 { 3323 .procname = "dev", 3324 .mode = 0555, 3325 .child = dev_table, 3326 }, 3327 { } 3328 }; 3329 3330 int __init sysctl_init(void) 3331 { 3332 struct ctl_table_header *hdr; 3333 3334 hdr = register_sysctl_table(sysctl_base_table); 3335 kmemleak_not_leak(hdr); 3336 return 0; 3337 } 3338 #endif /* CONFIG_SYSCTL */ 3339 /* 3340 * No sense putting this after each symbol definition, twice, 3341 * exception granted :-) 3342 */ 3343 EXPORT_SYMBOL(proc_dointvec); 3344 EXPORT_SYMBOL(proc_douintvec); 3345 EXPORT_SYMBOL(proc_dointvec_jiffies); 3346 EXPORT_SYMBOL(proc_dointvec_minmax); 3347 EXPORT_SYMBOL_GPL(proc_douintvec_minmax); 3348 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies); 3349 EXPORT_SYMBOL(proc_dointvec_ms_jiffies); 3350 EXPORT_SYMBOL(proc_dostring); 3351 EXPORT_SYMBOL(proc_doulongvec_minmax); 3352 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax); 3353 EXPORT_SYMBOL(proc_do_large_bitmap); 3354