1 // SPDX-License-Identifier: GPL-2.0-only 2 /* 3 * sysctl.c: General linux system control interface 4 * 5 * Begun 24 March 1995, Stephen Tweedie 6 * Added /proc support, Dec 1995 7 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas. 8 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver. 9 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver. 10 * Dynamic registration fixes, Stephen Tweedie. 11 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn. 12 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris 13 * Horn. 14 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer. 15 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer. 16 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill 17 * Wendling. 18 * The list_for_each() macro wasn't appropriate for the sysctl loop. 19 * Removed it and replaced it with older style, 03/23/00, Bill Wendling 20 */ 21 22 #include <linux/module.h> 23 #include <linux/aio.h> 24 #include <linux/mm.h> 25 #include <linux/swap.h> 26 #include <linux/slab.h> 27 #include <linux/sysctl.h> 28 #include <linux/bitmap.h> 29 #include <linux/signal.h> 30 #include <linux/printk.h> 31 #include <linux/proc_fs.h> 32 #include <linux/security.h> 33 #include <linux/ctype.h> 34 #include <linux/kmemleak.h> 35 #include <linux/fs.h> 36 #include <linux/init.h> 37 #include <linux/kernel.h> 38 #include <linux/kobject.h> 39 #include <linux/net.h> 40 #include <linux/sysrq.h> 41 #include <linux/highuid.h> 42 #include <linux/writeback.h> 43 #include <linux/ratelimit.h> 44 #include <linux/compaction.h> 45 #include <linux/hugetlb.h> 46 #include <linux/initrd.h> 47 #include <linux/key.h> 48 #include <linux/times.h> 49 #include <linux/limits.h> 50 #include <linux/dcache.h> 51 #include <linux/dnotify.h> 52 #include <linux/syscalls.h> 53 #include <linux/vmstat.h> 54 #include <linux/nfs_fs.h> 55 #include <linux/acpi.h> 56 #include <linux/reboot.h> 57 #include <linux/ftrace.h> 58 #include <linux/perf_event.h> 59 #include <linux/kprobes.h> 60 #include <linux/pipe_fs_i.h> 61 #include <linux/oom.h> 62 #include <linux/kmod.h> 63 #include <linux/capability.h> 64 #include <linux/binfmts.h> 65 #include <linux/sched/sysctl.h> 66 #include <linux/sched/coredump.h> 67 #include <linux/kexec.h> 68 #include <linux/bpf.h> 69 #include <linux/mount.h> 70 #include <linux/userfaultfd_k.h> 71 #include <linux/coredump.h> 72 #include <linux/latencytop.h> 73 #include <linux/pid.h> 74 75 #include "../lib/kstrtox.h" 76 77 #include <linux/uaccess.h> 78 #include <asm/processor.h> 79 80 #ifdef CONFIG_X86 81 #include <asm/nmi.h> 82 #include <asm/stacktrace.h> 83 #include <asm/io.h> 84 #endif 85 #ifdef CONFIG_SPARC 86 #include <asm/setup.h> 87 #endif 88 #ifdef CONFIG_BSD_PROCESS_ACCT 89 #include <linux/acct.h> 90 #endif 91 #ifdef CONFIG_RT_MUTEXES 92 #include <linux/rtmutex.h> 93 #endif 94 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT) 95 #include <linux/lockdep.h> 96 #endif 97 #ifdef CONFIG_CHR_DEV_SG 98 #include <scsi/sg.h> 99 #endif 100 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE 101 #include <linux/stackleak.h> 102 #endif 103 #ifdef CONFIG_LOCKUP_DETECTOR 104 #include <linux/nmi.h> 105 #endif 106 107 #if defined(CONFIG_SYSCTL) 108 109 /* Constants used for minimum and maximum */ 110 #ifdef CONFIG_LOCKUP_DETECTOR 111 static int sixty = 60; 112 #endif 113 114 static int __maybe_unused neg_one = -1; 115 static int __maybe_unused two = 2; 116 static int __maybe_unused four = 4; 117 static unsigned long zero_ul; 118 static unsigned long one_ul = 1; 119 static unsigned long long_max = LONG_MAX; 120 static int one_hundred = 100; 121 static int two_hundred = 200; 122 static int one_thousand = 1000; 123 #ifdef CONFIG_PRINTK 124 static int ten_thousand = 10000; 125 #endif 126 #ifdef CONFIG_PERF_EVENTS 127 static int six_hundred_forty_kb = 640 * 1024; 128 #endif 129 130 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */ 131 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE; 132 133 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */ 134 static int maxolduid = 65535; 135 static int minolduid; 136 137 static int ngroups_max = NGROUPS_MAX; 138 static const int cap_last_cap = CAP_LAST_CAP; 139 140 /* 141 * This is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs 142 * and hung_task_check_interval_secs 143 */ 144 #ifdef CONFIG_DETECT_HUNG_TASK 145 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ); 146 #endif 147 148 #ifdef CONFIG_INOTIFY_USER 149 #include <linux/inotify.h> 150 #endif 151 152 #ifdef CONFIG_PROC_SYSCTL 153 154 /** 155 * enum sysctl_writes_mode - supported sysctl write modes 156 * 157 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value 158 * to be written, and multiple writes on the same sysctl file descriptor 159 * will rewrite the sysctl value, regardless of file position. No warning 160 * is issued when the initial position is not 0. 161 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is 162 * not 0. 163 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at 164 * file position 0 and the value must be fully contained in the buffer 165 * sent to the write syscall. If dealing with strings respect the file 166 * position, but restrict this to the max length of the buffer, anything 167 * passed the max length will be ignored. Multiple writes will append 168 * to the buffer. 169 * 170 * These write modes control how current file position affects the behavior of 171 * updating sysctl values through the proc interface on each write. 172 */ 173 enum sysctl_writes_mode { 174 SYSCTL_WRITES_LEGACY = -1, 175 SYSCTL_WRITES_WARN = 0, 176 SYSCTL_WRITES_STRICT = 1, 177 }; 178 179 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT; 180 #endif /* CONFIG_PROC_SYSCTL */ 181 182 #if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \ 183 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT) 184 int sysctl_legacy_va_layout; 185 #endif 186 187 #ifdef CONFIG_SCHED_DEBUG 188 static int min_sched_granularity_ns = 100000; /* 100 usecs */ 189 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */ 190 static int min_wakeup_granularity_ns; /* 0 usecs */ 191 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */ 192 #ifdef CONFIG_SMP 193 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE; 194 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1; 195 #endif /* CONFIG_SMP */ 196 #endif /* CONFIG_SCHED_DEBUG */ 197 198 #ifdef CONFIG_COMPACTION 199 static int min_extfrag_threshold; 200 static int max_extfrag_threshold = 1000; 201 #endif 202 203 #endif /* CONFIG_SYSCTL */ 204 205 #if defined(CONFIG_BPF_SYSCALL) && defined(CONFIG_SYSCTL) 206 static int bpf_stats_handler(struct ctl_table *table, int write, 207 void __user *buffer, size_t *lenp, 208 loff_t *ppos) 209 { 210 struct static_key *key = (struct static_key *)table->data; 211 static int saved_val; 212 int val, ret; 213 struct ctl_table tmp = { 214 .data = &val, 215 .maxlen = sizeof(val), 216 .mode = table->mode, 217 .extra1 = SYSCTL_ZERO, 218 .extra2 = SYSCTL_ONE, 219 }; 220 221 if (write && !capable(CAP_SYS_ADMIN)) 222 return -EPERM; 223 224 mutex_lock(&bpf_stats_enabled_mutex); 225 val = saved_val; 226 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos); 227 if (write && !ret && val != saved_val) { 228 if (val) 229 static_key_slow_inc(key); 230 else 231 static_key_slow_dec(key); 232 saved_val = val; 233 } 234 mutex_unlock(&bpf_stats_enabled_mutex); 235 return ret; 236 } 237 #endif 238 239 /* 240 * /proc/sys support 241 */ 242 243 #ifdef CONFIG_PROC_SYSCTL 244 245 static int _proc_do_string(char *data, int maxlen, int write, 246 char *buffer, size_t *lenp, loff_t *ppos) 247 { 248 size_t len; 249 char c, *p; 250 251 if (!data || !maxlen || !*lenp) { 252 *lenp = 0; 253 return 0; 254 } 255 256 if (write) { 257 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) { 258 /* Only continue writes not past the end of buffer. */ 259 len = strlen(data); 260 if (len > maxlen - 1) 261 len = maxlen - 1; 262 263 if (*ppos > len) 264 return 0; 265 len = *ppos; 266 } else { 267 /* Start writing from beginning of buffer. */ 268 len = 0; 269 } 270 271 *ppos += *lenp; 272 p = buffer; 273 while ((p - buffer) < *lenp && len < maxlen - 1) { 274 c = *(p++); 275 if (c == 0 || c == '\n') 276 break; 277 data[len++] = c; 278 } 279 data[len] = 0; 280 } else { 281 len = strlen(data); 282 if (len > maxlen) 283 len = maxlen; 284 285 if (*ppos > len) { 286 *lenp = 0; 287 return 0; 288 } 289 290 data += *ppos; 291 len -= *ppos; 292 293 if (len > *lenp) 294 len = *lenp; 295 if (len) 296 memcpy(buffer, data, len); 297 if (len < *lenp) { 298 buffer[len] = '\n'; 299 len++; 300 } 301 *lenp = len; 302 *ppos += len; 303 } 304 return 0; 305 } 306 307 static void warn_sysctl_write(struct ctl_table *table) 308 { 309 pr_warn_once("%s wrote to %s when file position was not 0!\n" 310 "This will not be supported in the future. To silence this\n" 311 "warning, set kernel.sysctl_writes_strict = -1\n", 312 current->comm, table->procname); 313 } 314 315 /** 316 * proc_first_pos_non_zero_ignore - check if first position is allowed 317 * @ppos: file position 318 * @table: the sysctl table 319 * 320 * Returns true if the first position is non-zero and the sysctl_writes_strict 321 * mode indicates this is not allowed for numeric input types. String proc 322 * handlers can ignore the return value. 323 */ 324 static bool proc_first_pos_non_zero_ignore(loff_t *ppos, 325 struct ctl_table *table) 326 { 327 if (!*ppos) 328 return false; 329 330 switch (sysctl_writes_strict) { 331 case SYSCTL_WRITES_STRICT: 332 return true; 333 case SYSCTL_WRITES_WARN: 334 warn_sysctl_write(table); 335 return false; 336 default: 337 return false; 338 } 339 } 340 341 /** 342 * proc_dostring - read a string sysctl 343 * @table: the sysctl table 344 * @write: %TRUE if this is a write to the sysctl file 345 * @buffer: the user buffer 346 * @lenp: the size of the user buffer 347 * @ppos: file position 348 * 349 * Reads/writes a string from/to the user buffer. If the kernel 350 * buffer provided is not large enough to hold the string, the 351 * string is truncated. The copied string is %NULL-terminated. 352 * If the string is being read by the user process, it is copied 353 * and a newline '\n' is added. It is truncated if the buffer is 354 * not large enough. 355 * 356 * Returns 0 on success. 357 */ 358 int proc_dostring(struct ctl_table *table, int write, 359 void *buffer, size_t *lenp, loff_t *ppos) 360 { 361 if (write) 362 proc_first_pos_non_zero_ignore(ppos, table); 363 364 return _proc_do_string(table->data, table->maxlen, write, buffer, lenp, 365 ppos); 366 } 367 368 static size_t proc_skip_spaces(char **buf) 369 { 370 size_t ret; 371 char *tmp = skip_spaces(*buf); 372 ret = tmp - *buf; 373 *buf = tmp; 374 return ret; 375 } 376 377 static void proc_skip_char(char **buf, size_t *size, const char v) 378 { 379 while (*size) { 380 if (**buf != v) 381 break; 382 (*size)--; 383 (*buf)++; 384 } 385 } 386 387 /** 388 * strtoul_lenient - parse an ASCII formatted integer from a buffer and only 389 * fail on overflow 390 * 391 * @cp: kernel buffer containing the string to parse 392 * @endp: pointer to store the trailing characters 393 * @base: the base to use 394 * @res: where the parsed integer will be stored 395 * 396 * In case of success 0 is returned and @res will contain the parsed integer, 397 * @endp will hold any trailing characters. 398 * This function will fail the parse on overflow. If there wasn't an overflow 399 * the function will defer the decision what characters count as invalid to the 400 * caller. 401 */ 402 static int strtoul_lenient(const char *cp, char **endp, unsigned int base, 403 unsigned long *res) 404 { 405 unsigned long long result; 406 unsigned int rv; 407 408 cp = _parse_integer_fixup_radix(cp, &base); 409 rv = _parse_integer(cp, base, &result); 410 if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result)) 411 return -ERANGE; 412 413 cp += rv; 414 415 if (endp) 416 *endp = (char *)cp; 417 418 *res = (unsigned long)result; 419 return 0; 420 } 421 422 #define TMPBUFLEN 22 423 /** 424 * proc_get_long - reads an ASCII formatted integer from a user buffer 425 * 426 * @buf: a kernel buffer 427 * @size: size of the kernel buffer 428 * @val: this is where the number will be stored 429 * @neg: set to %TRUE if number is negative 430 * @perm_tr: a vector which contains the allowed trailers 431 * @perm_tr_len: size of the perm_tr vector 432 * @tr: pointer to store the trailer character 433 * 434 * In case of success %0 is returned and @buf and @size are updated with 435 * the amount of bytes read. If @tr is non-NULL and a trailing 436 * character exists (size is non-zero after returning from this 437 * function), @tr is updated with the trailing character. 438 */ 439 static int proc_get_long(char **buf, size_t *size, 440 unsigned long *val, bool *neg, 441 const char *perm_tr, unsigned perm_tr_len, char *tr) 442 { 443 int len; 444 char *p, tmp[TMPBUFLEN]; 445 446 if (!*size) 447 return -EINVAL; 448 449 len = *size; 450 if (len > TMPBUFLEN - 1) 451 len = TMPBUFLEN - 1; 452 453 memcpy(tmp, *buf, len); 454 455 tmp[len] = 0; 456 p = tmp; 457 if (*p == '-' && *size > 1) { 458 *neg = true; 459 p++; 460 } else 461 *neg = false; 462 if (!isdigit(*p)) 463 return -EINVAL; 464 465 if (strtoul_lenient(p, &p, 0, val)) 466 return -EINVAL; 467 468 len = p - tmp; 469 470 /* We don't know if the next char is whitespace thus we may accept 471 * invalid integers (e.g. 1234...a) or two integers instead of one 472 * (e.g. 123...1). So lets not allow such large numbers. */ 473 if (len == TMPBUFLEN - 1) 474 return -EINVAL; 475 476 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len)) 477 return -EINVAL; 478 479 if (tr && (len < *size)) 480 *tr = *p; 481 482 *buf += len; 483 *size -= len; 484 485 return 0; 486 } 487 488 /** 489 * proc_put_long - converts an integer to a decimal ASCII formatted string 490 * 491 * @buf: the user buffer 492 * @size: the size of the user buffer 493 * @val: the integer to be converted 494 * @neg: sign of the number, %TRUE for negative 495 * 496 * In case of success @buf and @size are updated with the amount of bytes 497 * written. 498 */ 499 static void proc_put_long(void **buf, size_t *size, unsigned long val, bool neg) 500 { 501 int len; 502 char tmp[TMPBUFLEN], *p = tmp; 503 504 sprintf(p, "%s%lu", neg ? "-" : "", val); 505 len = strlen(tmp); 506 if (len > *size) 507 len = *size; 508 memcpy(*buf, tmp, len); 509 *size -= len; 510 *buf += len; 511 } 512 #undef TMPBUFLEN 513 514 static void proc_put_char(void **buf, size_t *size, char c) 515 { 516 if (*size) { 517 char **buffer = (char **)buf; 518 **buffer = c; 519 520 (*size)--; 521 (*buffer)++; 522 *buf = *buffer; 523 } 524 } 525 526 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp, 527 int *valp, 528 int write, void *data) 529 { 530 if (write) { 531 if (*negp) { 532 if (*lvalp > (unsigned long) INT_MAX + 1) 533 return -EINVAL; 534 *valp = -*lvalp; 535 } else { 536 if (*lvalp > (unsigned long) INT_MAX) 537 return -EINVAL; 538 *valp = *lvalp; 539 } 540 } else { 541 int val = *valp; 542 if (val < 0) { 543 *negp = true; 544 *lvalp = -(unsigned long)val; 545 } else { 546 *negp = false; 547 *lvalp = (unsigned long)val; 548 } 549 } 550 return 0; 551 } 552 553 static int do_proc_douintvec_conv(unsigned long *lvalp, 554 unsigned int *valp, 555 int write, void *data) 556 { 557 if (write) { 558 if (*lvalp > UINT_MAX) 559 return -EINVAL; 560 *valp = *lvalp; 561 } else { 562 unsigned int val = *valp; 563 *lvalp = (unsigned long)val; 564 } 565 return 0; 566 } 567 568 static const char proc_wspace_sep[] = { ' ', '\t', '\n' }; 569 570 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table, 571 int write, void *buffer, 572 size_t *lenp, loff_t *ppos, 573 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 574 int write, void *data), 575 void *data) 576 { 577 int *i, vleft, first = 1, err = 0; 578 size_t left; 579 char *p; 580 581 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) { 582 *lenp = 0; 583 return 0; 584 } 585 586 i = (int *) tbl_data; 587 vleft = table->maxlen / sizeof(*i); 588 left = *lenp; 589 590 if (!conv) 591 conv = do_proc_dointvec_conv; 592 593 if (write) { 594 if (proc_first_pos_non_zero_ignore(ppos, table)) 595 goto out; 596 597 if (left > PAGE_SIZE - 1) 598 left = PAGE_SIZE - 1; 599 p = buffer; 600 } 601 602 for (; left && vleft--; i++, first=0) { 603 unsigned long lval; 604 bool neg; 605 606 if (write) { 607 left -= proc_skip_spaces(&p); 608 609 if (!left) 610 break; 611 err = proc_get_long(&p, &left, &lval, &neg, 612 proc_wspace_sep, 613 sizeof(proc_wspace_sep), NULL); 614 if (err) 615 break; 616 if (conv(&neg, &lval, i, 1, data)) { 617 err = -EINVAL; 618 break; 619 } 620 } else { 621 if (conv(&neg, &lval, i, 0, data)) { 622 err = -EINVAL; 623 break; 624 } 625 if (!first) 626 proc_put_char(&buffer, &left, '\t'); 627 proc_put_long(&buffer, &left, lval, neg); 628 } 629 } 630 631 if (!write && !first && left && !err) 632 proc_put_char(&buffer, &left, '\n'); 633 if (write && !err && left) 634 left -= proc_skip_spaces(&p); 635 if (write && first) 636 return err ? : -EINVAL; 637 *lenp -= left; 638 out: 639 *ppos += *lenp; 640 return err; 641 } 642 643 static int do_proc_dointvec(struct ctl_table *table, int write, 644 void *buffer, size_t *lenp, loff_t *ppos, 645 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 646 int write, void *data), 647 void *data) 648 { 649 return __do_proc_dointvec(table->data, table, write, 650 buffer, lenp, ppos, conv, data); 651 } 652 653 static int do_proc_douintvec_w(unsigned int *tbl_data, 654 struct ctl_table *table, 655 void *buffer, 656 size_t *lenp, loff_t *ppos, 657 int (*conv)(unsigned long *lvalp, 658 unsigned int *valp, 659 int write, void *data), 660 void *data) 661 { 662 unsigned long lval; 663 int err = 0; 664 size_t left; 665 bool neg; 666 char *p = buffer; 667 668 left = *lenp; 669 670 if (proc_first_pos_non_zero_ignore(ppos, table)) 671 goto bail_early; 672 673 if (left > PAGE_SIZE - 1) 674 left = PAGE_SIZE - 1; 675 676 left -= proc_skip_spaces(&p); 677 if (!left) { 678 err = -EINVAL; 679 goto out_free; 680 } 681 682 err = proc_get_long(&p, &left, &lval, &neg, 683 proc_wspace_sep, 684 sizeof(proc_wspace_sep), NULL); 685 if (err || neg) { 686 err = -EINVAL; 687 goto out_free; 688 } 689 690 if (conv(&lval, tbl_data, 1, data)) { 691 err = -EINVAL; 692 goto out_free; 693 } 694 695 if (!err && left) 696 left -= proc_skip_spaces(&p); 697 698 out_free: 699 if (err) 700 return -EINVAL; 701 702 return 0; 703 704 /* This is in keeping with old __do_proc_dointvec() */ 705 bail_early: 706 *ppos += *lenp; 707 return err; 708 } 709 710 static int do_proc_douintvec_r(unsigned int *tbl_data, void *buffer, 711 size_t *lenp, loff_t *ppos, 712 int (*conv)(unsigned long *lvalp, 713 unsigned int *valp, 714 int write, void *data), 715 void *data) 716 { 717 unsigned long lval; 718 int err = 0; 719 size_t left; 720 721 left = *lenp; 722 723 if (conv(&lval, tbl_data, 0, data)) { 724 err = -EINVAL; 725 goto out; 726 } 727 728 proc_put_long(&buffer, &left, lval, false); 729 if (!left) 730 goto out; 731 732 proc_put_char(&buffer, &left, '\n'); 733 734 out: 735 *lenp -= left; 736 *ppos += *lenp; 737 738 return err; 739 } 740 741 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table, 742 int write, void *buffer, 743 size_t *lenp, loff_t *ppos, 744 int (*conv)(unsigned long *lvalp, 745 unsigned int *valp, 746 int write, void *data), 747 void *data) 748 { 749 unsigned int *i, vleft; 750 751 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) { 752 *lenp = 0; 753 return 0; 754 } 755 756 i = (unsigned int *) tbl_data; 757 vleft = table->maxlen / sizeof(*i); 758 759 /* 760 * Arrays are not supported, keep this simple. *Do not* add 761 * support for them. 762 */ 763 if (vleft != 1) { 764 *lenp = 0; 765 return -EINVAL; 766 } 767 768 if (!conv) 769 conv = do_proc_douintvec_conv; 770 771 if (write) 772 return do_proc_douintvec_w(i, table, buffer, lenp, ppos, 773 conv, data); 774 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data); 775 } 776 777 static int do_proc_douintvec(struct ctl_table *table, int write, 778 void *buffer, size_t *lenp, loff_t *ppos, 779 int (*conv)(unsigned long *lvalp, 780 unsigned int *valp, 781 int write, void *data), 782 void *data) 783 { 784 return __do_proc_douintvec(table->data, table, write, 785 buffer, lenp, ppos, conv, data); 786 } 787 788 /** 789 * proc_dointvec - read a vector of integers 790 * @table: the sysctl table 791 * @write: %TRUE if this is a write to the sysctl file 792 * @buffer: the user buffer 793 * @lenp: the size of the user buffer 794 * @ppos: file position 795 * 796 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 797 * values from/to the user buffer, treated as an ASCII string. 798 * 799 * Returns 0 on success. 800 */ 801 int proc_dointvec(struct ctl_table *table, int write, void *buffer, 802 size_t *lenp, loff_t *ppos) 803 { 804 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL); 805 } 806 807 #ifdef CONFIG_COMPACTION 808 static int proc_dointvec_minmax_warn_RT_change(struct ctl_table *table, 809 int write, void *buffer, size_t *lenp, loff_t *ppos) 810 { 811 int ret, old; 812 813 if (!IS_ENABLED(CONFIG_PREEMPT_RT) || !write) 814 return proc_dointvec_minmax(table, write, buffer, lenp, ppos); 815 816 old = *(int *)table->data; 817 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos); 818 if (ret) 819 return ret; 820 if (old != *(int *)table->data) 821 pr_warn_once("sysctl attribute %s changed by %s[%d]\n", 822 table->procname, current->comm, 823 task_pid_nr(current)); 824 return ret; 825 } 826 #endif 827 828 /** 829 * proc_douintvec - read a vector of unsigned integers 830 * @table: the sysctl table 831 * @write: %TRUE if this is a write to the sysctl file 832 * @buffer: the user buffer 833 * @lenp: the size of the user buffer 834 * @ppos: file position 835 * 836 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer 837 * values from/to the user buffer, treated as an ASCII string. 838 * 839 * Returns 0 on success. 840 */ 841 int proc_douintvec(struct ctl_table *table, int write, void *buffer, 842 size_t *lenp, loff_t *ppos) 843 { 844 return do_proc_douintvec(table, write, buffer, lenp, ppos, 845 do_proc_douintvec_conv, NULL); 846 } 847 848 /* 849 * Taint values can only be increased 850 * This means we can safely use a temporary. 851 */ 852 static int proc_taint(struct ctl_table *table, int write, 853 void *buffer, size_t *lenp, loff_t *ppos) 854 { 855 struct ctl_table t; 856 unsigned long tmptaint = get_taint(); 857 int err; 858 859 if (write && !capable(CAP_SYS_ADMIN)) 860 return -EPERM; 861 862 t = *table; 863 t.data = &tmptaint; 864 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos); 865 if (err < 0) 866 return err; 867 868 if (write) { 869 int i; 870 871 /* 872 * If we are relying on panic_on_taint not producing 873 * false positives due to userspace input, bail out 874 * before setting the requested taint flags. 875 */ 876 if (panic_on_taint_nousertaint && (tmptaint & panic_on_taint)) 877 return -EINVAL; 878 879 /* 880 * Poor man's atomic or. Not worth adding a primitive 881 * to everyone's atomic.h for this 882 */ 883 for (i = 0; i < TAINT_FLAGS_COUNT; i++) 884 if ((1UL << i) & tmptaint) 885 add_taint(i, LOCKDEP_STILL_OK); 886 } 887 888 return err; 889 } 890 891 #ifdef CONFIG_PRINTK 892 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write, 893 void *buffer, size_t *lenp, loff_t *ppos) 894 { 895 if (write && !capable(CAP_SYS_ADMIN)) 896 return -EPERM; 897 898 return proc_dointvec_minmax(table, write, buffer, lenp, ppos); 899 } 900 #endif 901 902 /** 903 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure 904 * @min: pointer to minimum allowable value 905 * @max: pointer to maximum allowable value 906 * 907 * The do_proc_dointvec_minmax_conv_param structure provides the 908 * minimum and maximum values for doing range checking for those sysctl 909 * parameters that use the proc_dointvec_minmax() handler. 910 */ 911 struct do_proc_dointvec_minmax_conv_param { 912 int *min; 913 int *max; 914 }; 915 916 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp, 917 int *valp, 918 int write, void *data) 919 { 920 int tmp, ret; 921 struct do_proc_dointvec_minmax_conv_param *param = data; 922 /* 923 * If writing, first do so via a temporary local int so we can 924 * bounds-check it before touching *valp. 925 */ 926 int *ip = write ? &tmp : valp; 927 928 ret = do_proc_dointvec_conv(negp, lvalp, ip, write, data); 929 if (ret) 930 return ret; 931 932 if (write) { 933 if ((param->min && *param->min > tmp) || 934 (param->max && *param->max < tmp)) 935 return -EINVAL; 936 *valp = tmp; 937 } 938 939 return 0; 940 } 941 942 /** 943 * proc_dointvec_minmax - read a vector of integers with min/max values 944 * @table: the sysctl table 945 * @write: %TRUE if this is a write to the sysctl file 946 * @buffer: the user buffer 947 * @lenp: the size of the user buffer 948 * @ppos: file position 949 * 950 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 951 * values from/to the user buffer, treated as an ASCII string. 952 * 953 * This routine will ensure the values are within the range specified by 954 * table->extra1 (min) and table->extra2 (max). 955 * 956 * Returns 0 on success or -EINVAL on write when the range check fails. 957 */ 958 int proc_dointvec_minmax(struct ctl_table *table, int write, 959 void *buffer, size_t *lenp, loff_t *ppos) 960 { 961 struct do_proc_dointvec_minmax_conv_param param = { 962 .min = (int *) table->extra1, 963 .max = (int *) table->extra2, 964 }; 965 return do_proc_dointvec(table, write, buffer, lenp, ppos, 966 do_proc_dointvec_minmax_conv, ¶m); 967 } 968 969 /** 970 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure 971 * @min: pointer to minimum allowable value 972 * @max: pointer to maximum allowable value 973 * 974 * The do_proc_douintvec_minmax_conv_param structure provides the 975 * minimum and maximum values for doing range checking for those sysctl 976 * parameters that use the proc_douintvec_minmax() handler. 977 */ 978 struct do_proc_douintvec_minmax_conv_param { 979 unsigned int *min; 980 unsigned int *max; 981 }; 982 983 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp, 984 unsigned int *valp, 985 int write, void *data) 986 { 987 int ret; 988 unsigned int tmp; 989 struct do_proc_douintvec_minmax_conv_param *param = data; 990 /* write via temporary local uint for bounds-checking */ 991 unsigned int *up = write ? &tmp : valp; 992 993 ret = do_proc_douintvec_conv(lvalp, up, write, data); 994 if (ret) 995 return ret; 996 997 if (write) { 998 if ((param->min && *param->min > tmp) || 999 (param->max && *param->max < tmp)) 1000 return -ERANGE; 1001 1002 *valp = tmp; 1003 } 1004 1005 return 0; 1006 } 1007 1008 /** 1009 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values 1010 * @table: the sysctl table 1011 * @write: %TRUE if this is a write to the sysctl file 1012 * @buffer: the user buffer 1013 * @lenp: the size of the user buffer 1014 * @ppos: file position 1015 * 1016 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer 1017 * values from/to the user buffer, treated as an ASCII string. Negative 1018 * strings are not allowed. 1019 * 1020 * This routine will ensure the values are within the range specified by 1021 * table->extra1 (min) and table->extra2 (max). There is a final sanity 1022 * check for UINT_MAX to avoid having to support wrap around uses from 1023 * userspace. 1024 * 1025 * Returns 0 on success or -ERANGE on write when the range check fails. 1026 */ 1027 int proc_douintvec_minmax(struct ctl_table *table, int write, 1028 void *buffer, size_t *lenp, loff_t *ppos) 1029 { 1030 struct do_proc_douintvec_minmax_conv_param param = { 1031 .min = (unsigned int *) table->extra1, 1032 .max = (unsigned int *) table->extra2, 1033 }; 1034 return do_proc_douintvec(table, write, buffer, lenp, ppos, 1035 do_proc_douintvec_minmax_conv, ¶m); 1036 } 1037 1038 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp, 1039 unsigned int *valp, 1040 int write, void *data) 1041 { 1042 if (write) { 1043 unsigned int val; 1044 1045 val = round_pipe_size(*lvalp); 1046 if (val == 0) 1047 return -EINVAL; 1048 1049 *valp = val; 1050 } else { 1051 unsigned int val = *valp; 1052 *lvalp = (unsigned long) val; 1053 } 1054 1055 return 0; 1056 } 1057 1058 static int proc_dopipe_max_size(struct ctl_table *table, int write, 1059 void *buffer, size_t *lenp, loff_t *ppos) 1060 { 1061 return do_proc_douintvec(table, write, buffer, lenp, ppos, 1062 do_proc_dopipe_max_size_conv, NULL); 1063 } 1064 1065 static void validate_coredump_safety(void) 1066 { 1067 #ifdef CONFIG_COREDUMP 1068 if (suid_dumpable == SUID_DUMP_ROOT && 1069 core_pattern[0] != '/' && core_pattern[0] != '|') { 1070 printk(KERN_WARNING 1071 "Unsafe core_pattern used with fs.suid_dumpable=2.\n" 1072 "Pipe handler or fully qualified core dump path required.\n" 1073 "Set kernel.core_pattern before fs.suid_dumpable.\n" 1074 ); 1075 } 1076 #endif 1077 } 1078 1079 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write, 1080 void *buffer, size_t *lenp, loff_t *ppos) 1081 { 1082 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos); 1083 if (!error) 1084 validate_coredump_safety(); 1085 return error; 1086 } 1087 1088 #ifdef CONFIG_COREDUMP 1089 static int proc_dostring_coredump(struct ctl_table *table, int write, 1090 void *buffer, size_t *lenp, loff_t *ppos) 1091 { 1092 int error = proc_dostring(table, write, buffer, lenp, ppos); 1093 if (!error) 1094 validate_coredump_safety(); 1095 return error; 1096 } 1097 #endif 1098 1099 #ifdef CONFIG_MAGIC_SYSRQ 1100 static int sysrq_sysctl_handler(struct ctl_table *table, int write, 1101 void *buffer, size_t *lenp, loff_t *ppos) 1102 { 1103 int tmp, ret; 1104 1105 tmp = sysrq_mask(); 1106 1107 ret = __do_proc_dointvec(&tmp, table, write, buffer, 1108 lenp, ppos, NULL, NULL); 1109 if (ret || !write) 1110 return ret; 1111 1112 if (write) 1113 sysrq_toggle_support(tmp); 1114 1115 return 0; 1116 } 1117 #endif 1118 1119 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, 1120 int write, void *buffer, size_t *lenp, loff_t *ppos, 1121 unsigned long convmul, unsigned long convdiv) 1122 { 1123 unsigned long *i, *min, *max; 1124 int vleft, first = 1, err = 0; 1125 size_t left; 1126 char *p; 1127 1128 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) { 1129 *lenp = 0; 1130 return 0; 1131 } 1132 1133 i = (unsigned long *) data; 1134 min = (unsigned long *) table->extra1; 1135 max = (unsigned long *) table->extra2; 1136 vleft = table->maxlen / sizeof(unsigned long); 1137 left = *lenp; 1138 1139 if (write) { 1140 if (proc_first_pos_non_zero_ignore(ppos, table)) 1141 goto out; 1142 1143 if (left > PAGE_SIZE - 1) 1144 left = PAGE_SIZE - 1; 1145 p = buffer; 1146 } 1147 1148 for (; left && vleft--; i++, first = 0) { 1149 unsigned long val; 1150 1151 if (write) { 1152 bool neg; 1153 1154 left -= proc_skip_spaces(&p); 1155 if (!left) 1156 break; 1157 1158 err = proc_get_long(&p, &left, &val, &neg, 1159 proc_wspace_sep, 1160 sizeof(proc_wspace_sep), NULL); 1161 if (err) 1162 break; 1163 if (neg) 1164 continue; 1165 val = convmul * val / convdiv; 1166 if ((min && val < *min) || (max && val > *max)) { 1167 err = -EINVAL; 1168 break; 1169 } 1170 *i = val; 1171 } else { 1172 val = convdiv * (*i) / convmul; 1173 if (!first) 1174 proc_put_char(&buffer, &left, '\t'); 1175 proc_put_long(&buffer, &left, val, false); 1176 } 1177 } 1178 1179 if (!write && !first && left && !err) 1180 proc_put_char(&buffer, &left, '\n'); 1181 if (write && !err) 1182 left -= proc_skip_spaces(&p); 1183 if (write && first) 1184 return err ? : -EINVAL; 1185 *lenp -= left; 1186 out: 1187 *ppos += *lenp; 1188 return err; 1189 } 1190 1191 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write, 1192 void *buffer, size_t *lenp, loff_t *ppos, unsigned long convmul, 1193 unsigned long convdiv) 1194 { 1195 return __do_proc_doulongvec_minmax(table->data, table, write, 1196 buffer, lenp, ppos, convmul, convdiv); 1197 } 1198 1199 /** 1200 * proc_doulongvec_minmax - read a vector of long integers with min/max values 1201 * @table: the sysctl table 1202 * @write: %TRUE if this is a write to the sysctl file 1203 * @buffer: the user buffer 1204 * @lenp: the size of the user buffer 1205 * @ppos: file position 1206 * 1207 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 1208 * values from/to the user buffer, treated as an ASCII string. 1209 * 1210 * This routine will ensure the values are within the range specified by 1211 * table->extra1 (min) and table->extra2 (max). 1212 * 1213 * Returns 0 on success. 1214 */ 1215 int proc_doulongvec_minmax(struct ctl_table *table, int write, 1216 void *buffer, size_t *lenp, loff_t *ppos) 1217 { 1218 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l); 1219 } 1220 1221 /** 1222 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values 1223 * @table: the sysctl table 1224 * @write: %TRUE if this is a write to the sysctl file 1225 * @buffer: the user buffer 1226 * @lenp: the size of the user buffer 1227 * @ppos: file position 1228 * 1229 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 1230 * values from/to the user buffer, treated as an ASCII string. The values 1231 * are treated as milliseconds, and converted to jiffies when they are stored. 1232 * 1233 * This routine will ensure the values are within the range specified by 1234 * table->extra1 (min) and table->extra2 (max). 1235 * 1236 * Returns 0 on success. 1237 */ 1238 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 1239 void *buffer, size_t *lenp, loff_t *ppos) 1240 { 1241 return do_proc_doulongvec_minmax(table, write, buffer, 1242 lenp, ppos, HZ, 1000l); 1243 } 1244 1245 1246 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp, 1247 int *valp, 1248 int write, void *data) 1249 { 1250 if (write) { 1251 if (*lvalp > INT_MAX / HZ) 1252 return 1; 1253 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ); 1254 } else { 1255 int val = *valp; 1256 unsigned long lval; 1257 if (val < 0) { 1258 *negp = true; 1259 lval = -(unsigned long)val; 1260 } else { 1261 *negp = false; 1262 lval = (unsigned long)val; 1263 } 1264 *lvalp = lval / HZ; 1265 } 1266 return 0; 1267 } 1268 1269 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp, 1270 int *valp, 1271 int write, void *data) 1272 { 1273 if (write) { 1274 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ) 1275 return 1; 1276 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp); 1277 } else { 1278 int val = *valp; 1279 unsigned long lval; 1280 if (val < 0) { 1281 *negp = true; 1282 lval = -(unsigned long)val; 1283 } else { 1284 *negp = false; 1285 lval = (unsigned long)val; 1286 } 1287 *lvalp = jiffies_to_clock_t(lval); 1288 } 1289 return 0; 1290 } 1291 1292 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp, 1293 int *valp, 1294 int write, void *data) 1295 { 1296 if (write) { 1297 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp); 1298 1299 if (jif > INT_MAX) 1300 return 1; 1301 *valp = (int)jif; 1302 } else { 1303 int val = *valp; 1304 unsigned long lval; 1305 if (val < 0) { 1306 *negp = true; 1307 lval = -(unsigned long)val; 1308 } else { 1309 *negp = false; 1310 lval = (unsigned long)val; 1311 } 1312 *lvalp = jiffies_to_msecs(lval); 1313 } 1314 return 0; 1315 } 1316 1317 /** 1318 * proc_dointvec_jiffies - read a vector of integers as seconds 1319 * @table: the sysctl table 1320 * @write: %TRUE if this is a write to the sysctl file 1321 * @buffer: the user buffer 1322 * @lenp: the size of the user buffer 1323 * @ppos: file position 1324 * 1325 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 1326 * values from/to the user buffer, treated as an ASCII string. 1327 * The values read are assumed to be in seconds, and are converted into 1328 * jiffies. 1329 * 1330 * Returns 0 on success. 1331 */ 1332 int proc_dointvec_jiffies(struct ctl_table *table, int write, 1333 void *buffer, size_t *lenp, loff_t *ppos) 1334 { 1335 return do_proc_dointvec(table,write,buffer,lenp,ppos, 1336 do_proc_dointvec_jiffies_conv,NULL); 1337 } 1338 1339 /** 1340 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds 1341 * @table: the sysctl table 1342 * @write: %TRUE if this is a write to the sysctl file 1343 * @buffer: the user buffer 1344 * @lenp: the size of the user buffer 1345 * @ppos: pointer to the file position 1346 * 1347 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 1348 * values from/to the user buffer, treated as an ASCII string. 1349 * The values read are assumed to be in 1/USER_HZ seconds, and 1350 * are converted into jiffies. 1351 * 1352 * Returns 0 on success. 1353 */ 1354 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 1355 void *buffer, size_t *lenp, loff_t *ppos) 1356 { 1357 return do_proc_dointvec(table,write,buffer,lenp,ppos, 1358 do_proc_dointvec_userhz_jiffies_conv,NULL); 1359 } 1360 1361 /** 1362 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds 1363 * @table: the sysctl table 1364 * @write: %TRUE if this is a write to the sysctl file 1365 * @buffer: the user buffer 1366 * @lenp: the size of the user buffer 1367 * @ppos: file position 1368 * @ppos: the current position in the file 1369 * 1370 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 1371 * values from/to the user buffer, treated as an ASCII string. 1372 * The values read are assumed to be in 1/1000 seconds, and 1373 * are converted into jiffies. 1374 * 1375 * Returns 0 on success. 1376 */ 1377 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, void *buffer, 1378 size_t *lenp, loff_t *ppos) 1379 { 1380 return do_proc_dointvec(table, write, buffer, lenp, ppos, 1381 do_proc_dointvec_ms_jiffies_conv, NULL); 1382 } 1383 1384 static int proc_do_cad_pid(struct ctl_table *table, int write, void *buffer, 1385 size_t *lenp, loff_t *ppos) 1386 { 1387 struct pid *new_pid; 1388 pid_t tmp; 1389 int r; 1390 1391 tmp = pid_vnr(cad_pid); 1392 1393 r = __do_proc_dointvec(&tmp, table, write, buffer, 1394 lenp, ppos, NULL, NULL); 1395 if (r || !write) 1396 return r; 1397 1398 new_pid = find_get_pid(tmp); 1399 if (!new_pid) 1400 return -ESRCH; 1401 1402 put_pid(xchg(&cad_pid, new_pid)); 1403 return 0; 1404 } 1405 1406 /** 1407 * proc_do_large_bitmap - read/write from/to a large bitmap 1408 * @table: the sysctl table 1409 * @write: %TRUE if this is a write to the sysctl file 1410 * @buffer: the user buffer 1411 * @lenp: the size of the user buffer 1412 * @ppos: file position 1413 * 1414 * The bitmap is stored at table->data and the bitmap length (in bits) 1415 * in table->maxlen. 1416 * 1417 * We use a range comma separated format (e.g. 1,3-4,10-10) so that 1418 * large bitmaps may be represented in a compact manner. Writing into 1419 * the file will clear the bitmap then update it with the given input. 1420 * 1421 * Returns 0 on success. 1422 */ 1423 int proc_do_large_bitmap(struct ctl_table *table, int write, 1424 void *buffer, size_t *lenp, loff_t *ppos) 1425 { 1426 int err = 0; 1427 bool first = 1; 1428 size_t left = *lenp; 1429 unsigned long bitmap_len = table->maxlen; 1430 unsigned long *bitmap = *(unsigned long **) table->data; 1431 unsigned long *tmp_bitmap = NULL; 1432 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c; 1433 1434 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) { 1435 *lenp = 0; 1436 return 0; 1437 } 1438 1439 if (write) { 1440 char *p = buffer; 1441 size_t skipped = 0; 1442 1443 if (left > PAGE_SIZE - 1) { 1444 left = PAGE_SIZE - 1; 1445 /* How much of the buffer we'll skip this pass */ 1446 skipped = *lenp - left; 1447 } 1448 1449 tmp_bitmap = bitmap_zalloc(bitmap_len, GFP_KERNEL); 1450 if (!tmp_bitmap) 1451 return -ENOMEM; 1452 proc_skip_char(&p, &left, '\n'); 1453 while (!err && left) { 1454 unsigned long val_a, val_b; 1455 bool neg; 1456 size_t saved_left; 1457 1458 /* In case we stop parsing mid-number, we can reset */ 1459 saved_left = left; 1460 err = proc_get_long(&p, &left, &val_a, &neg, tr_a, 1461 sizeof(tr_a), &c); 1462 /* 1463 * If we consumed the entirety of a truncated buffer or 1464 * only one char is left (may be a "-"), then stop here, 1465 * reset, & come back for more. 1466 */ 1467 if ((left <= 1) && skipped) { 1468 left = saved_left; 1469 break; 1470 } 1471 1472 if (err) 1473 break; 1474 if (val_a >= bitmap_len || neg) { 1475 err = -EINVAL; 1476 break; 1477 } 1478 1479 val_b = val_a; 1480 if (left) { 1481 p++; 1482 left--; 1483 } 1484 1485 if (c == '-') { 1486 err = proc_get_long(&p, &left, &val_b, 1487 &neg, tr_b, sizeof(tr_b), 1488 &c); 1489 /* 1490 * If we consumed all of a truncated buffer or 1491 * then stop here, reset, & come back for more. 1492 */ 1493 if (!left && skipped) { 1494 left = saved_left; 1495 break; 1496 } 1497 1498 if (err) 1499 break; 1500 if (val_b >= bitmap_len || neg || 1501 val_a > val_b) { 1502 err = -EINVAL; 1503 break; 1504 } 1505 if (left) { 1506 p++; 1507 left--; 1508 } 1509 } 1510 1511 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1); 1512 first = 0; 1513 proc_skip_char(&p, &left, '\n'); 1514 } 1515 left += skipped; 1516 } else { 1517 unsigned long bit_a, bit_b = 0; 1518 1519 while (left) { 1520 bit_a = find_next_bit(bitmap, bitmap_len, bit_b); 1521 if (bit_a >= bitmap_len) 1522 break; 1523 bit_b = find_next_zero_bit(bitmap, bitmap_len, 1524 bit_a + 1) - 1; 1525 1526 if (!first) 1527 proc_put_char(&buffer, &left, ','); 1528 proc_put_long(&buffer, &left, bit_a, false); 1529 if (bit_a != bit_b) { 1530 proc_put_char(&buffer, &left, '-'); 1531 proc_put_long(&buffer, &left, bit_b, false); 1532 } 1533 1534 first = 0; bit_b++; 1535 } 1536 proc_put_char(&buffer, &left, '\n'); 1537 } 1538 1539 if (!err) { 1540 if (write) { 1541 if (*ppos) 1542 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len); 1543 else 1544 bitmap_copy(bitmap, tmp_bitmap, bitmap_len); 1545 } 1546 *lenp -= left; 1547 *ppos += *lenp; 1548 } 1549 1550 bitmap_free(tmp_bitmap); 1551 return err; 1552 } 1553 1554 #else /* CONFIG_PROC_SYSCTL */ 1555 1556 int proc_dostring(struct ctl_table *table, int write, 1557 void *buffer, size_t *lenp, loff_t *ppos) 1558 { 1559 return -ENOSYS; 1560 } 1561 1562 int proc_dointvec(struct ctl_table *table, int write, 1563 void *buffer, size_t *lenp, loff_t *ppos) 1564 { 1565 return -ENOSYS; 1566 } 1567 1568 int proc_douintvec(struct ctl_table *table, int write, 1569 void *buffer, size_t *lenp, loff_t *ppos) 1570 { 1571 return -ENOSYS; 1572 } 1573 1574 int proc_dointvec_minmax(struct ctl_table *table, int write, 1575 void *buffer, size_t *lenp, loff_t *ppos) 1576 { 1577 return -ENOSYS; 1578 } 1579 1580 int proc_douintvec_minmax(struct ctl_table *table, int write, 1581 void *buffer, size_t *lenp, loff_t *ppos) 1582 { 1583 return -ENOSYS; 1584 } 1585 1586 int proc_dointvec_jiffies(struct ctl_table *table, int write, 1587 void *buffer, size_t *lenp, loff_t *ppos) 1588 { 1589 return -ENOSYS; 1590 } 1591 1592 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 1593 void *buffer, size_t *lenp, loff_t *ppos) 1594 { 1595 return -ENOSYS; 1596 } 1597 1598 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, 1599 void *buffer, size_t *lenp, loff_t *ppos) 1600 { 1601 return -ENOSYS; 1602 } 1603 1604 int proc_doulongvec_minmax(struct ctl_table *table, int write, 1605 void *buffer, size_t *lenp, loff_t *ppos) 1606 { 1607 return -ENOSYS; 1608 } 1609 1610 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 1611 void *buffer, size_t *lenp, loff_t *ppos) 1612 { 1613 return -ENOSYS; 1614 } 1615 1616 int proc_do_large_bitmap(struct ctl_table *table, int write, 1617 void *buffer, size_t *lenp, loff_t *ppos) 1618 { 1619 return -ENOSYS; 1620 } 1621 1622 #endif /* CONFIG_PROC_SYSCTL */ 1623 1624 #if defined(CONFIG_SYSCTL) 1625 int proc_do_static_key(struct ctl_table *table, int write, 1626 void *buffer, size_t *lenp, loff_t *ppos) 1627 { 1628 struct static_key *key = (struct static_key *)table->data; 1629 static DEFINE_MUTEX(static_key_mutex); 1630 int val, ret; 1631 struct ctl_table tmp = { 1632 .data = &val, 1633 .maxlen = sizeof(val), 1634 .mode = table->mode, 1635 .extra1 = SYSCTL_ZERO, 1636 .extra2 = SYSCTL_ONE, 1637 }; 1638 1639 if (write && !capable(CAP_SYS_ADMIN)) 1640 return -EPERM; 1641 1642 mutex_lock(&static_key_mutex); 1643 val = static_key_enabled(key); 1644 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos); 1645 if (write && !ret) { 1646 if (val) 1647 static_key_enable(key); 1648 else 1649 static_key_disable(key); 1650 } 1651 mutex_unlock(&static_key_mutex); 1652 return ret; 1653 } 1654 1655 static struct ctl_table kern_table[] = { 1656 { 1657 .procname = "sched_child_runs_first", 1658 .data = &sysctl_sched_child_runs_first, 1659 .maxlen = sizeof(unsigned int), 1660 .mode = 0644, 1661 .proc_handler = proc_dointvec, 1662 }, 1663 #ifdef CONFIG_SCHED_DEBUG 1664 { 1665 .procname = "sched_min_granularity_ns", 1666 .data = &sysctl_sched_min_granularity, 1667 .maxlen = sizeof(unsigned int), 1668 .mode = 0644, 1669 .proc_handler = sched_proc_update_handler, 1670 .extra1 = &min_sched_granularity_ns, 1671 .extra2 = &max_sched_granularity_ns, 1672 }, 1673 { 1674 .procname = "sched_latency_ns", 1675 .data = &sysctl_sched_latency, 1676 .maxlen = sizeof(unsigned int), 1677 .mode = 0644, 1678 .proc_handler = sched_proc_update_handler, 1679 .extra1 = &min_sched_granularity_ns, 1680 .extra2 = &max_sched_granularity_ns, 1681 }, 1682 { 1683 .procname = "sched_wakeup_granularity_ns", 1684 .data = &sysctl_sched_wakeup_granularity, 1685 .maxlen = sizeof(unsigned int), 1686 .mode = 0644, 1687 .proc_handler = sched_proc_update_handler, 1688 .extra1 = &min_wakeup_granularity_ns, 1689 .extra2 = &max_wakeup_granularity_ns, 1690 }, 1691 #ifdef CONFIG_SMP 1692 { 1693 .procname = "sched_tunable_scaling", 1694 .data = &sysctl_sched_tunable_scaling, 1695 .maxlen = sizeof(enum sched_tunable_scaling), 1696 .mode = 0644, 1697 .proc_handler = sched_proc_update_handler, 1698 .extra1 = &min_sched_tunable_scaling, 1699 .extra2 = &max_sched_tunable_scaling, 1700 }, 1701 { 1702 .procname = "sched_migration_cost_ns", 1703 .data = &sysctl_sched_migration_cost, 1704 .maxlen = sizeof(unsigned int), 1705 .mode = 0644, 1706 .proc_handler = proc_dointvec, 1707 }, 1708 { 1709 .procname = "sched_nr_migrate", 1710 .data = &sysctl_sched_nr_migrate, 1711 .maxlen = sizeof(unsigned int), 1712 .mode = 0644, 1713 .proc_handler = proc_dointvec, 1714 }, 1715 #ifdef CONFIG_SCHEDSTATS 1716 { 1717 .procname = "sched_schedstats", 1718 .data = NULL, 1719 .maxlen = sizeof(unsigned int), 1720 .mode = 0644, 1721 .proc_handler = sysctl_schedstats, 1722 .extra1 = SYSCTL_ZERO, 1723 .extra2 = SYSCTL_ONE, 1724 }, 1725 #endif /* CONFIG_SCHEDSTATS */ 1726 #endif /* CONFIG_SMP */ 1727 #ifdef CONFIG_NUMA_BALANCING 1728 { 1729 .procname = "numa_balancing_scan_delay_ms", 1730 .data = &sysctl_numa_balancing_scan_delay, 1731 .maxlen = sizeof(unsigned int), 1732 .mode = 0644, 1733 .proc_handler = proc_dointvec, 1734 }, 1735 { 1736 .procname = "numa_balancing_scan_period_min_ms", 1737 .data = &sysctl_numa_balancing_scan_period_min, 1738 .maxlen = sizeof(unsigned int), 1739 .mode = 0644, 1740 .proc_handler = proc_dointvec, 1741 }, 1742 { 1743 .procname = "numa_balancing_scan_period_max_ms", 1744 .data = &sysctl_numa_balancing_scan_period_max, 1745 .maxlen = sizeof(unsigned int), 1746 .mode = 0644, 1747 .proc_handler = proc_dointvec, 1748 }, 1749 { 1750 .procname = "numa_balancing_scan_size_mb", 1751 .data = &sysctl_numa_balancing_scan_size, 1752 .maxlen = sizeof(unsigned int), 1753 .mode = 0644, 1754 .proc_handler = proc_dointvec_minmax, 1755 .extra1 = SYSCTL_ONE, 1756 }, 1757 { 1758 .procname = "numa_balancing", 1759 .data = NULL, /* filled in by handler */ 1760 .maxlen = sizeof(unsigned int), 1761 .mode = 0644, 1762 .proc_handler = sysctl_numa_balancing, 1763 .extra1 = SYSCTL_ZERO, 1764 .extra2 = SYSCTL_ONE, 1765 }, 1766 #endif /* CONFIG_NUMA_BALANCING */ 1767 #endif /* CONFIG_SCHED_DEBUG */ 1768 { 1769 .procname = "sched_rt_period_us", 1770 .data = &sysctl_sched_rt_period, 1771 .maxlen = sizeof(unsigned int), 1772 .mode = 0644, 1773 .proc_handler = sched_rt_handler, 1774 }, 1775 { 1776 .procname = "sched_rt_runtime_us", 1777 .data = &sysctl_sched_rt_runtime, 1778 .maxlen = sizeof(int), 1779 .mode = 0644, 1780 .proc_handler = sched_rt_handler, 1781 }, 1782 { 1783 .procname = "sched_rr_timeslice_ms", 1784 .data = &sysctl_sched_rr_timeslice, 1785 .maxlen = sizeof(int), 1786 .mode = 0644, 1787 .proc_handler = sched_rr_handler, 1788 }, 1789 #ifdef CONFIG_UCLAMP_TASK 1790 { 1791 .procname = "sched_util_clamp_min", 1792 .data = &sysctl_sched_uclamp_util_min, 1793 .maxlen = sizeof(unsigned int), 1794 .mode = 0644, 1795 .proc_handler = sysctl_sched_uclamp_handler, 1796 }, 1797 { 1798 .procname = "sched_util_clamp_max", 1799 .data = &sysctl_sched_uclamp_util_max, 1800 .maxlen = sizeof(unsigned int), 1801 .mode = 0644, 1802 .proc_handler = sysctl_sched_uclamp_handler, 1803 }, 1804 #endif 1805 #ifdef CONFIG_SCHED_AUTOGROUP 1806 { 1807 .procname = "sched_autogroup_enabled", 1808 .data = &sysctl_sched_autogroup_enabled, 1809 .maxlen = sizeof(unsigned int), 1810 .mode = 0644, 1811 .proc_handler = proc_dointvec_minmax, 1812 .extra1 = SYSCTL_ZERO, 1813 .extra2 = SYSCTL_ONE, 1814 }, 1815 #endif 1816 #ifdef CONFIG_CFS_BANDWIDTH 1817 { 1818 .procname = "sched_cfs_bandwidth_slice_us", 1819 .data = &sysctl_sched_cfs_bandwidth_slice, 1820 .maxlen = sizeof(unsigned int), 1821 .mode = 0644, 1822 .proc_handler = proc_dointvec_minmax, 1823 .extra1 = SYSCTL_ONE, 1824 }, 1825 #endif 1826 #if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_CPU_FREQ_GOV_SCHEDUTIL) 1827 { 1828 .procname = "sched_energy_aware", 1829 .data = &sysctl_sched_energy_aware, 1830 .maxlen = sizeof(unsigned int), 1831 .mode = 0644, 1832 .proc_handler = sched_energy_aware_handler, 1833 .extra1 = SYSCTL_ZERO, 1834 .extra2 = SYSCTL_ONE, 1835 }, 1836 #endif 1837 #ifdef CONFIG_PROVE_LOCKING 1838 { 1839 .procname = "prove_locking", 1840 .data = &prove_locking, 1841 .maxlen = sizeof(int), 1842 .mode = 0644, 1843 .proc_handler = proc_dointvec, 1844 }, 1845 #endif 1846 #ifdef CONFIG_LOCK_STAT 1847 { 1848 .procname = "lock_stat", 1849 .data = &lock_stat, 1850 .maxlen = sizeof(int), 1851 .mode = 0644, 1852 .proc_handler = proc_dointvec, 1853 }, 1854 #endif 1855 { 1856 .procname = "panic", 1857 .data = &panic_timeout, 1858 .maxlen = sizeof(int), 1859 .mode = 0644, 1860 .proc_handler = proc_dointvec, 1861 }, 1862 #ifdef CONFIG_COREDUMP 1863 { 1864 .procname = "core_uses_pid", 1865 .data = &core_uses_pid, 1866 .maxlen = sizeof(int), 1867 .mode = 0644, 1868 .proc_handler = proc_dointvec, 1869 }, 1870 { 1871 .procname = "core_pattern", 1872 .data = core_pattern, 1873 .maxlen = CORENAME_MAX_SIZE, 1874 .mode = 0644, 1875 .proc_handler = proc_dostring_coredump, 1876 }, 1877 { 1878 .procname = "core_pipe_limit", 1879 .data = &core_pipe_limit, 1880 .maxlen = sizeof(unsigned int), 1881 .mode = 0644, 1882 .proc_handler = proc_dointvec, 1883 }, 1884 #endif 1885 #ifdef CONFIG_PROC_SYSCTL 1886 { 1887 .procname = "tainted", 1888 .maxlen = sizeof(long), 1889 .mode = 0644, 1890 .proc_handler = proc_taint, 1891 }, 1892 { 1893 .procname = "sysctl_writes_strict", 1894 .data = &sysctl_writes_strict, 1895 .maxlen = sizeof(int), 1896 .mode = 0644, 1897 .proc_handler = proc_dointvec_minmax, 1898 .extra1 = &neg_one, 1899 .extra2 = SYSCTL_ONE, 1900 }, 1901 #endif 1902 #ifdef CONFIG_LATENCYTOP 1903 { 1904 .procname = "latencytop", 1905 .data = &latencytop_enabled, 1906 .maxlen = sizeof(int), 1907 .mode = 0644, 1908 .proc_handler = sysctl_latencytop, 1909 }, 1910 #endif 1911 #ifdef CONFIG_BLK_DEV_INITRD 1912 { 1913 .procname = "real-root-dev", 1914 .data = &real_root_dev, 1915 .maxlen = sizeof(int), 1916 .mode = 0644, 1917 .proc_handler = proc_dointvec, 1918 }, 1919 #endif 1920 { 1921 .procname = "print-fatal-signals", 1922 .data = &print_fatal_signals, 1923 .maxlen = sizeof(int), 1924 .mode = 0644, 1925 .proc_handler = proc_dointvec, 1926 }, 1927 #ifdef CONFIG_SPARC 1928 { 1929 .procname = "reboot-cmd", 1930 .data = reboot_command, 1931 .maxlen = 256, 1932 .mode = 0644, 1933 .proc_handler = proc_dostring, 1934 }, 1935 { 1936 .procname = "stop-a", 1937 .data = &stop_a_enabled, 1938 .maxlen = sizeof (int), 1939 .mode = 0644, 1940 .proc_handler = proc_dointvec, 1941 }, 1942 { 1943 .procname = "scons-poweroff", 1944 .data = &scons_pwroff, 1945 .maxlen = sizeof (int), 1946 .mode = 0644, 1947 .proc_handler = proc_dointvec, 1948 }, 1949 #endif 1950 #ifdef CONFIG_SPARC64 1951 { 1952 .procname = "tsb-ratio", 1953 .data = &sysctl_tsb_ratio, 1954 .maxlen = sizeof (int), 1955 .mode = 0644, 1956 .proc_handler = proc_dointvec, 1957 }, 1958 #endif 1959 #ifdef CONFIG_PARISC 1960 { 1961 .procname = "soft-power", 1962 .data = &pwrsw_enabled, 1963 .maxlen = sizeof (int), 1964 .mode = 0644, 1965 .proc_handler = proc_dointvec, 1966 }, 1967 #endif 1968 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW 1969 { 1970 .procname = "unaligned-trap", 1971 .data = &unaligned_enabled, 1972 .maxlen = sizeof (int), 1973 .mode = 0644, 1974 .proc_handler = proc_dointvec, 1975 }, 1976 #endif 1977 { 1978 .procname = "ctrl-alt-del", 1979 .data = &C_A_D, 1980 .maxlen = sizeof(int), 1981 .mode = 0644, 1982 .proc_handler = proc_dointvec, 1983 }, 1984 #ifdef CONFIG_FUNCTION_TRACER 1985 { 1986 .procname = "ftrace_enabled", 1987 .data = &ftrace_enabled, 1988 .maxlen = sizeof(int), 1989 .mode = 0644, 1990 .proc_handler = ftrace_enable_sysctl, 1991 }, 1992 #endif 1993 #ifdef CONFIG_STACK_TRACER 1994 { 1995 .procname = "stack_tracer_enabled", 1996 .data = &stack_tracer_enabled, 1997 .maxlen = sizeof(int), 1998 .mode = 0644, 1999 .proc_handler = stack_trace_sysctl, 2000 }, 2001 #endif 2002 #ifdef CONFIG_TRACING 2003 { 2004 .procname = "ftrace_dump_on_oops", 2005 .data = &ftrace_dump_on_oops, 2006 .maxlen = sizeof(int), 2007 .mode = 0644, 2008 .proc_handler = proc_dointvec, 2009 }, 2010 { 2011 .procname = "traceoff_on_warning", 2012 .data = &__disable_trace_on_warning, 2013 .maxlen = sizeof(__disable_trace_on_warning), 2014 .mode = 0644, 2015 .proc_handler = proc_dointvec, 2016 }, 2017 { 2018 .procname = "tracepoint_printk", 2019 .data = &tracepoint_printk, 2020 .maxlen = sizeof(tracepoint_printk), 2021 .mode = 0644, 2022 .proc_handler = tracepoint_printk_sysctl, 2023 }, 2024 #endif 2025 #ifdef CONFIG_KEXEC_CORE 2026 { 2027 .procname = "kexec_load_disabled", 2028 .data = &kexec_load_disabled, 2029 .maxlen = sizeof(int), 2030 .mode = 0644, 2031 /* only handle a transition from default "0" to "1" */ 2032 .proc_handler = proc_dointvec_minmax, 2033 .extra1 = SYSCTL_ONE, 2034 .extra2 = SYSCTL_ONE, 2035 }, 2036 #endif 2037 #ifdef CONFIG_MODULES 2038 { 2039 .procname = "modprobe", 2040 .data = &modprobe_path, 2041 .maxlen = KMOD_PATH_LEN, 2042 .mode = 0644, 2043 .proc_handler = proc_dostring, 2044 }, 2045 { 2046 .procname = "modules_disabled", 2047 .data = &modules_disabled, 2048 .maxlen = sizeof(int), 2049 .mode = 0644, 2050 /* only handle a transition from default "0" to "1" */ 2051 .proc_handler = proc_dointvec_minmax, 2052 .extra1 = SYSCTL_ONE, 2053 .extra2 = SYSCTL_ONE, 2054 }, 2055 #endif 2056 #ifdef CONFIG_UEVENT_HELPER 2057 { 2058 .procname = "hotplug", 2059 .data = &uevent_helper, 2060 .maxlen = UEVENT_HELPER_PATH_LEN, 2061 .mode = 0644, 2062 .proc_handler = proc_dostring, 2063 }, 2064 #endif 2065 #ifdef CONFIG_CHR_DEV_SG 2066 { 2067 .procname = "sg-big-buff", 2068 .data = &sg_big_buff, 2069 .maxlen = sizeof (int), 2070 .mode = 0444, 2071 .proc_handler = proc_dointvec, 2072 }, 2073 #endif 2074 #ifdef CONFIG_BSD_PROCESS_ACCT 2075 { 2076 .procname = "acct", 2077 .data = &acct_parm, 2078 .maxlen = 3*sizeof(int), 2079 .mode = 0644, 2080 .proc_handler = proc_dointvec, 2081 }, 2082 #endif 2083 #ifdef CONFIG_MAGIC_SYSRQ 2084 { 2085 .procname = "sysrq", 2086 .data = NULL, 2087 .maxlen = sizeof (int), 2088 .mode = 0644, 2089 .proc_handler = sysrq_sysctl_handler, 2090 }, 2091 #endif 2092 #ifdef CONFIG_PROC_SYSCTL 2093 { 2094 .procname = "cad_pid", 2095 .data = NULL, 2096 .maxlen = sizeof (int), 2097 .mode = 0600, 2098 .proc_handler = proc_do_cad_pid, 2099 }, 2100 #endif 2101 { 2102 .procname = "threads-max", 2103 .data = NULL, 2104 .maxlen = sizeof(int), 2105 .mode = 0644, 2106 .proc_handler = sysctl_max_threads, 2107 }, 2108 { 2109 .procname = "random", 2110 .mode = 0555, 2111 .child = random_table, 2112 }, 2113 { 2114 .procname = "usermodehelper", 2115 .mode = 0555, 2116 .child = usermodehelper_table, 2117 }, 2118 #ifdef CONFIG_FW_LOADER_USER_HELPER 2119 { 2120 .procname = "firmware_config", 2121 .mode = 0555, 2122 .child = firmware_config_table, 2123 }, 2124 #endif 2125 { 2126 .procname = "overflowuid", 2127 .data = &overflowuid, 2128 .maxlen = sizeof(int), 2129 .mode = 0644, 2130 .proc_handler = proc_dointvec_minmax, 2131 .extra1 = &minolduid, 2132 .extra2 = &maxolduid, 2133 }, 2134 { 2135 .procname = "overflowgid", 2136 .data = &overflowgid, 2137 .maxlen = sizeof(int), 2138 .mode = 0644, 2139 .proc_handler = proc_dointvec_minmax, 2140 .extra1 = &minolduid, 2141 .extra2 = &maxolduid, 2142 }, 2143 #ifdef CONFIG_S390 2144 { 2145 .procname = "userprocess_debug", 2146 .data = &show_unhandled_signals, 2147 .maxlen = sizeof(int), 2148 .mode = 0644, 2149 .proc_handler = proc_dointvec, 2150 }, 2151 #endif 2152 #ifdef CONFIG_SMP 2153 { 2154 .procname = "oops_all_cpu_backtrace", 2155 .data = &sysctl_oops_all_cpu_backtrace, 2156 .maxlen = sizeof(int), 2157 .mode = 0644, 2158 .proc_handler = proc_dointvec_minmax, 2159 .extra1 = SYSCTL_ZERO, 2160 .extra2 = SYSCTL_ONE, 2161 }, 2162 #endif /* CONFIG_SMP */ 2163 { 2164 .procname = "pid_max", 2165 .data = &pid_max, 2166 .maxlen = sizeof (int), 2167 .mode = 0644, 2168 .proc_handler = proc_dointvec_minmax, 2169 .extra1 = &pid_max_min, 2170 .extra2 = &pid_max_max, 2171 }, 2172 { 2173 .procname = "panic_on_oops", 2174 .data = &panic_on_oops, 2175 .maxlen = sizeof(int), 2176 .mode = 0644, 2177 .proc_handler = proc_dointvec, 2178 }, 2179 { 2180 .procname = "panic_print", 2181 .data = &panic_print, 2182 .maxlen = sizeof(unsigned long), 2183 .mode = 0644, 2184 .proc_handler = proc_doulongvec_minmax, 2185 }, 2186 #if defined CONFIG_PRINTK 2187 { 2188 .procname = "printk", 2189 .data = &console_loglevel, 2190 .maxlen = 4*sizeof(int), 2191 .mode = 0644, 2192 .proc_handler = proc_dointvec, 2193 }, 2194 { 2195 .procname = "printk_ratelimit", 2196 .data = &printk_ratelimit_state.interval, 2197 .maxlen = sizeof(int), 2198 .mode = 0644, 2199 .proc_handler = proc_dointvec_jiffies, 2200 }, 2201 { 2202 .procname = "printk_ratelimit_burst", 2203 .data = &printk_ratelimit_state.burst, 2204 .maxlen = sizeof(int), 2205 .mode = 0644, 2206 .proc_handler = proc_dointvec, 2207 }, 2208 { 2209 .procname = "printk_delay", 2210 .data = &printk_delay_msec, 2211 .maxlen = sizeof(int), 2212 .mode = 0644, 2213 .proc_handler = proc_dointvec_minmax, 2214 .extra1 = SYSCTL_ZERO, 2215 .extra2 = &ten_thousand, 2216 }, 2217 { 2218 .procname = "printk_devkmsg", 2219 .data = devkmsg_log_str, 2220 .maxlen = DEVKMSG_STR_MAX_SIZE, 2221 .mode = 0644, 2222 .proc_handler = devkmsg_sysctl_set_loglvl, 2223 }, 2224 { 2225 .procname = "dmesg_restrict", 2226 .data = &dmesg_restrict, 2227 .maxlen = sizeof(int), 2228 .mode = 0644, 2229 .proc_handler = proc_dointvec_minmax_sysadmin, 2230 .extra1 = SYSCTL_ZERO, 2231 .extra2 = SYSCTL_ONE, 2232 }, 2233 { 2234 .procname = "kptr_restrict", 2235 .data = &kptr_restrict, 2236 .maxlen = sizeof(int), 2237 .mode = 0644, 2238 .proc_handler = proc_dointvec_minmax_sysadmin, 2239 .extra1 = SYSCTL_ZERO, 2240 .extra2 = &two, 2241 }, 2242 #endif 2243 { 2244 .procname = "ngroups_max", 2245 .data = &ngroups_max, 2246 .maxlen = sizeof (int), 2247 .mode = 0444, 2248 .proc_handler = proc_dointvec, 2249 }, 2250 { 2251 .procname = "cap_last_cap", 2252 .data = (void *)&cap_last_cap, 2253 .maxlen = sizeof(int), 2254 .mode = 0444, 2255 .proc_handler = proc_dointvec, 2256 }, 2257 #if defined(CONFIG_LOCKUP_DETECTOR) 2258 { 2259 .procname = "watchdog", 2260 .data = &watchdog_user_enabled, 2261 .maxlen = sizeof(int), 2262 .mode = 0644, 2263 .proc_handler = proc_watchdog, 2264 .extra1 = SYSCTL_ZERO, 2265 .extra2 = SYSCTL_ONE, 2266 }, 2267 { 2268 .procname = "watchdog_thresh", 2269 .data = &watchdog_thresh, 2270 .maxlen = sizeof(int), 2271 .mode = 0644, 2272 .proc_handler = proc_watchdog_thresh, 2273 .extra1 = SYSCTL_ZERO, 2274 .extra2 = &sixty, 2275 }, 2276 { 2277 .procname = "nmi_watchdog", 2278 .data = &nmi_watchdog_user_enabled, 2279 .maxlen = sizeof(int), 2280 .mode = NMI_WATCHDOG_SYSCTL_PERM, 2281 .proc_handler = proc_nmi_watchdog, 2282 .extra1 = SYSCTL_ZERO, 2283 .extra2 = SYSCTL_ONE, 2284 }, 2285 { 2286 .procname = "watchdog_cpumask", 2287 .data = &watchdog_cpumask_bits, 2288 .maxlen = NR_CPUS, 2289 .mode = 0644, 2290 .proc_handler = proc_watchdog_cpumask, 2291 }, 2292 #ifdef CONFIG_SOFTLOCKUP_DETECTOR 2293 { 2294 .procname = "soft_watchdog", 2295 .data = &soft_watchdog_user_enabled, 2296 .maxlen = sizeof(int), 2297 .mode = 0644, 2298 .proc_handler = proc_soft_watchdog, 2299 .extra1 = SYSCTL_ZERO, 2300 .extra2 = SYSCTL_ONE, 2301 }, 2302 { 2303 .procname = "softlockup_panic", 2304 .data = &softlockup_panic, 2305 .maxlen = sizeof(int), 2306 .mode = 0644, 2307 .proc_handler = proc_dointvec_minmax, 2308 .extra1 = SYSCTL_ZERO, 2309 .extra2 = SYSCTL_ONE, 2310 }, 2311 #ifdef CONFIG_SMP 2312 { 2313 .procname = "softlockup_all_cpu_backtrace", 2314 .data = &sysctl_softlockup_all_cpu_backtrace, 2315 .maxlen = sizeof(int), 2316 .mode = 0644, 2317 .proc_handler = proc_dointvec_minmax, 2318 .extra1 = SYSCTL_ZERO, 2319 .extra2 = SYSCTL_ONE, 2320 }, 2321 #endif /* CONFIG_SMP */ 2322 #endif 2323 #ifdef CONFIG_HARDLOCKUP_DETECTOR 2324 { 2325 .procname = "hardlockup_panic", 2326 .data = &hardlockup_panic, 2327 .maxlen = sizeof(int), 2328 .mode = 0644, 2329 .proc_handler = proc_dointvec_minmax, 2330 .extra1 = SYSCTL_ZERO, 2331 .extra2 = SYSCTL_ONE, 2332 }, 2333 #ifdef CONFIG_SMP 2334 { 2335 .procname = "hardlockup_all_cpu_backtrace", 2336 .data = &sysctl_hardlockup_all_cpu_backtrace, 2337 .maxlen = sizeof(int), 2338 .mode = 0644, 2339 .proc_handler = proc_dointvec_minmax, 2340 .extra1 = SYSCTL_ZERO, 2341 .extra2 = SYSCTL_ONE, 2342 }, 2343 #endif /* CONFIG_SMP */ 2344 #endif 2345 #endif 2346 2347 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86) 2348 { 2349 .procname = "unknown_nmi_panic", 2350 .data = &unknown_nmi_panic, 2351 .maxlen = sizeof (int), 2352 .mode = 0644, 2353 .proc_handler = proc_dointvec, 2354 }, 2355 #endif 2356 2357 #if (defined(CONFIG_X86_32) || defined(CONFIG_PARISC)) && \ 2358 defined(CONFIG_DEBUG_STACKOVERFLOW) 2359 { 2360 .procname = "panic_on_stackoverflow", 2361 .data = &sysctl_panic_on_stackoverflow, 2362 .maxlen = sizeof(int), 2363 .mode = 0644, 2364 .proc_handler = proc_dointvec, 2365 }, 2366 #endif 2367 #if defined(CONFIG_X86) 2368 { 2369 .procname = "panic_on_unrecovered_nmi", 2370 .data = &panic_on_unrecovered_nmi, 2371 .maxlen = sizeof(int), 2372 .mode = 0644, 2373 .proc_handler = proc_dointvec, 2374 }, 2375 { 2376 .procname = "panic_on_io_nmi", 2377 .data = &panic_on_io_nmi, 2378 .maxlen = sizeof(int), 2379 .mode = 0644, 2380 .proc_handler = proc_dointvec, 2381 }, 2382 { 2383 .procname = "bootloader_type", 2384 .data = &bootloader_type, 2385 .maxlen = sizeof (int), 2386 .mode = 0444, 2387 .proc_handler = proc_dointvec, 2388 }, 2389 { 2390 .procname = "bootloader_version", 2391 .data = &bootloader_version, 2392 .maxlen = sizeof (int), 2393 .mode = 0444, 2394 .proc_handler = proc_dointvec, 2395 }, 2396 { 2397 .procname = "io_delay_type", 2398 .data = &io_delay_type, 2399 .maxlen = sizeof(int), 2400 .mode = 0644, 2401 .proc_handler = proc_dointvec, 2402 }, 2403 #endif 2404 #if defined(CONFIG_MMU) 2405 { 2406 .procname = "randomize_va_space", 2407 .data = &randomize_va_space, 2408 .maxlen = sizeof(int), 2409 .mode = 0644, 2410 .proc_handler = proc_dointvec, 2411 }, 2412 #endif 2413 #if defined(CONFIG_S390) && defined(CONFIG_SMP) 2414 { 2415 .procname = "spin_retry", 2416 .data = &spin_retry, 2417 .maxlen = sizeof (int), 2418 .mode = 0644, 2419 .proc_handler = proc_dointvec, 2420 }, 2421 #endif 2422 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86) 2423 { 2424 .procname = "acpi_video_flags", 2425 .data = &acpi_realmode_flags, 2426 .maxlen = sizeof (unsigned long), 2427 .mode = 0644, 2428 .proc_handler = proc_doulongvec_minmax, 2429 }, 2430 #endif 2431 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN 2432 { 2433 .procname = "ignore-unaligned-usertrap", 2434 .data = &no_unaligned_warning, 2435 .maxlen = sizeof (int), 2436 .mode = 0644, 2437 .proc_handler = proc_dointvec, 2438 }, 2439 #endif 2440 #ifdef CONFIG_IA64 2441 { 2442 .procname = "unaligned-dump-stack", 2443 .data = &unaligned_dump_stack, 2444 .maxlen = sizeof (int), 2445 .mode = 0644, 2446 .proc_handler = proc_dointvec, 2447 }, 2448 #endif 2449 #ifdef CONFIG_DETECT_HUNG_TASK 2450 #ifdef CONFIG_SMP 2451 { 2452 .procname = "hung_task_all_cpu_backtrace", 2453 .data = &sysctl_hung_task_all_cpu_backtrace, 2454 .maxlen = sizeof(int), 2455 .mode = 0644, 2456 .proc_handler = proc_dointvec_minmax, 2457 .extra1 = SYSCTL_ZERO, 2458 .extra2 = SYSCTL_ONE, 2459 }, 2460 #endif /* CONFIG_SMP */ 2461 { 2462 .procname = "hung_task_panic", 2463 .data = &sysctl_hung_task_panic, 2464 .maxlen = sizeof(int), 2465 .mode = 0644, 2466 .proc_handler = proc_dointvec_minmax, 2467 .extra1 = SYSCTL_ZERO, 2468 .extra2 = SYSCTL_ONE, 2469 }, 2470 { 2471 .procname = "hung_task_check_count", 2472 .data = &sysctl_hung_task_check_count, 2473 .maxlen = sizeof(int), 2474 .mode = 0644, 2475 .proc_handler = proc_dointvec_minmax, 2476 .extra1 = SYSCTL_ZERO, 2477 }, 2478 { 2479 .procname = "hung_task_timeout_secs", 2480 .data = &sysctl_hung_task_timeout_secs, 2481 .maxlen = sizeof(unsigned long), 2482 .mode = 0644, 2483 .proc_handler = proc_dohung_task_timeout_secs, 2484 .extra2 = &hung_task_timeout_max, 2485 }, 2486 { 2487 .procname = "hung_task_check_interval_secs", 2488 .data = &sysctl_hung_task_check_interval_secs, 2489 .maxlen = sizeof(unsigned long), 2490 .mode = 0644, 2491 .proc_handler = proc_dohung_task_timeout_secs, 2492 .extra2 = &hung_task_timeout_max, 2493 }, 2494 { 2495 .procname = "hung_task_warnings", 2496 .data = &sysctl_hung_task_warnings, 2497 .maxlen = sizeof(int), 2498 .mode = 0644, 2499 .proc_handler = proc_dointvec_minmax, 2500 .extra1 = &neg_one, 2501 }, 2502 #endif 2503 #ifdef CONFIG_RT_MUTEXES 2504 { 2505 .procname = "max_lock_depth", 2506 .data = &max_lock_depth, 2507 .maxlen = sizeof(int), 2508 .mode = 0644, 2509 .proc_handler = proc_dointvec, 2510 }, 2511 #endif 2512 { 2513 .procname = "poweroff_cmd", 2514 .data = &poweroff_cmd, 2515 .maxlen = POWEROFF_CMD_PATH_LEN, 2516 .mode = 0644, 2517 .proc_handler = proc_dostring, 2518 }, 2519 #ifdef CONFIG_KEYS 2520 { 2521 .procname = "keys", 2522 .mode = 0555, 2523 .child = key_sysctls, 2524 }, 2525 #endif 2526 #ifdef CONFIG_PERF_EVENTS 2527 /* 2528 * User-space scripts rely on the existence of this file 2529 * as a feature check for perf_events being enabled. 2530 * 2531 * So it's an ABI, do not remove! 2532 */ 2533 { 2534 .procname = "perf_event_paranoid", 2535 .data = &sysctl_perf_event_paranoid, 2536 .maxlen = sizeof(sysctl_perf_event_paranoid), 2537 .mode = 0644, 2538 .proc_handler = proc_dointvec, 2539 }, 2540 { 2541 .procname = "perf_event_mlock_kb", 2542 .data = &sysctl_perf_event_mlock, 2543 .maxlen = sizeof(sysctl_perf_event_mlock), 2544 .mode = 0644, 2545 .proc_handler = proc_dointvec, 2546 }, 2547 { 2548 .procname = "perf_event_max_sample_rate", 2549 .data = &sysctl_perf_event_sample_rate, 2550 .maxlen = sizeof(sysctl_perf_event_sample_rate), 2551 .mode = 0644, 2552 .proc_handler = perf_proc_update_handler, 2553 .extra1 = SYSCTL_ONE, 2554 }, 2555 { 2556 .procname = "perf_cpu_time_max_percent", 2557 .data = &sysctl_perf_cpu_time_max_percent, 2558 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent), 2559 .mode = 0644, 2560 .proc_handler = perf_cpu_time_max_percent_handler, 2561 .extra1 = SYSCTL_ZERO, 2562 .extra2 = &one_hundred, 2563 }, 2564 { 2565 .procname = "perf_event_max_stack", 2566 .data = &sysctl_perf_event_max_stack, 2567 .maxlen = sizeof(sysctl_perf_event_max_stack), 2568 .mode = 0644, 2569 .proc_handler = perf_event_max_stack_handler, 2570 .extra1 = SYSCTL_ZERO, 2571 .extra2 = &six_hundred_forty_kb, 2572 }, 2573 { 2574 .procname = "perf_event_max_contexts_per_stack", 2575 .data = &sysctl_perf_event_max_contexts_per_stack, 2576 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack), 2577 .mode = 0644, 2578 .proc_handler = perf_event_max_stack_handler, 2579 .extra1 = SYSCTL_ZERO, 2580 .extra2 = &one_thousand, 2581 }, 2582 #endif 2583 { 2584 .procname = "panic_on_warn", 2585 .data = &panic_on_warn, 2586 .maxlen = sizeof(int), 2587 .mode = 0644, 2588 .proc_handler = proc_dointvec_minmax, 2589 .extra1 = SYSCTL_ZERO, 2590 .extra2 = SYSCTL_ONE, 2591 }, 2592 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) 2593 { 2594 .procname = "timer_migration", 2595 .data = &sysctl_timer_migration, 2596 .maxlen = sizeof(unsigned int), 2597 .mode = 0644, 2598 .proc_handler = timer_migration_handler, 2599 .extra1 = SYSCTL_ZERO, 2600 .extra2 = SYSCTL_ONE, 2601 }, 2602 #endif 2603 #ifdef CONFIG_BPF_SYSCALL 2604 { 2605 .procname = "unprivileged_bpf_disabled", 2606 .data = &sysctl_unprivileged_bpf_disabled, 2607 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled), 2608 .mode = 0644, 2609 /* only handle a transition from default "0" to "1" */ 2610 .proc_handler = proc_dointvec_minmax, 2611 .extra1 = SYSCTL_ONE, 2612 .extra2 = SYSCTL_ONE, 2613 }, 2614 { 2615 .procname = "bpf_stats_enabled", 2616 .data = &bpf_stats_enabled_key.key, 2617 .maxlen = sizeof(bpf_stats_enabled_key), 2618 .mode = 0644, 2619 .proc_handler = bpf_stats_handler, 2620 }, 2621 #endif 2622 #if defined(CONFIG_TREE_RCU) 2623 { 2624 .procname = "panic_on_rcu_stall", 2625 .data = &sysctl_panic_on_rcu_stall, 2626 .maxlen = sizeof(sysctl_panic_on_rcu_stall), 2627 .mode = 0644, 2628 .proc_handler = proc_dointvec_minmax, 2629 .extra1 = SYSCTL_ZERO, 2630 .extra2 = SYSCTL_ONE, 2631 }, 2632 #endif 2633 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE 2634 { 2635 .procname = "stack_erasing", 2636 .data = NULL, 2637 .maxlen = sizeof(int), 2638 .mode = 0600, 2639 .proc_handler = stack_erasing_sysctl, 2640 .extra1 = SYSCTL_ZERO, 2641 .extra2 = SYSCTL_ONE, 2642 }, 2643 #endif 2644 { } 2645 }; 2646 2647 static struct ctl_table vm_table[] = { 2648 { 2649 .procname = "overcommit_memory", 2650 .data = &sysctl_overcommit_memory, 2651 .maxlen = sizeof(sysctl_overcommit_memory), 2652 .mode = 0644, 2653 .proc_handler = proc_dointvec_minmax, 2654 .extra1 = SYSCTL_ZERO, 2655 .extra2 = &two, 2656 }, 2657 { 2658 .procname = "panic_on_oom", 2659 .data = &sysctl_panic_on_oom, 2660 .maxlen = sizeof(sysctl_panic_on_oom), 2661 .mode = 0644, 2662 .proc_handler = proc_dointvec_minmax, 2663 .extra1 = SYSCTL_ZERO, 2664 .extra2 = &two, 2665 }, 2666 { 2667 .procname = "oom_kill_allocating_task", 2668 .data = &sysctl_oom_kill_allocating_task, 2669 .maxlen = sizeof(sysctl_oom_kill_allocating_task), 2670 .mode = 0644, 2671 .proc_handler = proc_dointvec, 2672 }, 2673 { 2674 .procname = "oom_dump_tasks", 2675 .data = &sysctl_oom_dump_tasks, 2676 .maxlen = sizeof(sysctl_oom_dump_tasks), 2677 .mode = 0644, 2678 .proc_handler = proc_dointvec, 2679 }, 2680 { 2681 .procname = "overcommit_ratio", 2682 .data = &sysctl_overcommit_ratio, 2683 .maxlen = sizeof(sysctl_overcommit_ratio), 2684 .mode = 0644, 2685 .proc_handler = overcommit_ratio_handler, 2686 }, 2687 { 2688 .procname = "overcommit_kbytes", 2689 .data = &sysctl_overcommit_kbytes, 2690 .maxlen = sizeof(sysctl_overcommit_kbytes), 2691 .mode = 0644, 2692 .proc_handler = overcommit_kbytes_handler, 2693 }, 2694 { 2695 .procname = "page-cluster", 2696 .data = &page_cluster, 2697 .maxlen = sizeof(int), 2698 .mode = 0644, 2699 .proc_handler = proc_dointvec_minmax, 2700 .extra1 = SYSCTL_ZERO, 2701 }, 2702 { 2703 .procname = "dirty_background_ratio", 2704 .data = &dirty_background_ratio, 2705 .maxlen = sizeof(dirty_background_ratio), 2706 .mode = 0644, 2707 .proc_handler = dirty_background_ratio_handler, 2708 .extra1 = SYSCTL_ZERO, 2709 .extra2 = &one_hundred, 2710 }, 2711 { 2712 .procname = "dirty_background_bytes", 2713 .data = &dirty_background_bytes, 2714 .maxlen = sizeof(dirty_background_bytes), 2715 .mode = 0644, 2716 .proc_handler = dirty_background_bytes_handler, 2717 .extra1 = &one_ul, 2718 }, 2719 { 2720 .procname = "dirty_ratio", 2721 .data = &vm_dirty_ratio, 2722 .maxlen = sizeof(vm_dirty_ratio), 2723 .mode = 0644, 2724 .proc_handler = dirty_ratio_handler, 2725 .extra1 = SYSCTL_ZERO, 2726 .extra2 = &one_hundred, 2727 }, 2728 { 2729 .procname = "dirty_bytes", 2730 .data = &vm_dirty_bytes, 2731 .maxlen = sizeof(vm_dirty_bytes), 2732 .mode = 0644, 2733 .proc_handler = dirty_bytes_handler, 2734 .extra1 = &dirty_bytes_min, 2735 }, 2736 { 2737 .procname = "dirty_writeback_centisecs", 2738 .data = &dirty_writeback_interval, 2739 .maxlen = sizeof(dirty_writeback_interval), 2740 .mode = 0644, 2741 .proc_handler = dirty_writeback_centisecs_handler, 2742 }, 2743 { 2744 .procname = "dirty_expire_centisecs", 2745 .data = &dirty_expire_interval, 2746 .maxlen = sizeof(dirty_expire_interval), 2747 .mode = 0644, 2748 .proc_handler = proc_dointvec_minmax, 2749 .extra1 = SYSCTL_ZERO, 2750 }, 2751 { 2752 .procname = "dirtytime_expire_seconds", 2753 .data = &dirtytime_expire_interval, 2754 .maxlen = sizeof(dirtytime_expire_interval), 2755 .mode = 0644, 2756 .proc_handler = dirtytime_interval_handler, 2757 .extra1 = SYSCTL_ZERO, 2758 }, 2759 { 2760 .procname = "swappiness", 2761 .data = &vm_swappiness, 2762 .maxlen = sizeof(vm_swappiness), 2763 .mode = 0644, 2764 .proc_handler = proc_dointvec_minmax, 2765 .extra1 = SYSCTL_ZERO, 2766 .extra2 = &two_hundred, 2767 }, 2768 #ifdef CONFIG_HUGETLB_PAGE 2769 { 2770 .procname = "nr_hugepages", 2771 .data = NULL, 2772 .maxlen = sizeof(unsigned long), 2773 .mode = 0644, 2774 .proc_handler = hugetlb_sysctl_handler, 2775 }, 2776 #ifdef CONFIG_NUMA 2777 { 2778 .procname = "nr_hugepages_mempolicy", 2779 .data = NULL, 2780 .maxlen = sizeof(unsigned long), 2781 .mode = 0644, 2782 .proc_handler = &hugetlb_mempolicy_sysctl_handler, 2783 }, 2784 { 2785 .procname = "numa_stat", 2786 .data = &sysctl_vm_numa_stat, 2787 .maxlen = sizeof(int), 2788 .mode = 0644, 2789 .proc_handler = sysctl_vm_numa_stat_handler, 2790 .extra1 = SYSCTL_ZERO, 2791 .extra2 = SYSCTL_ONE, 2792 }, 2793 #endif 2794 { 2795 .procname = "hugetlb_shm_group", 2796 .data = &sysctl_hugetlb_shm_group, 2797 .maxlen = sizeof(gid_t), 2798 .mode = 0644, 2799 .proc_handler = proc_dointvec, 2800 }, 2801 { 2802 .procname = "nr_overcommit_hugepages", 2803 .data = NULL, 2804 .maxlen = sizeof(unsigned long), 2805 .mode = 0644, 2806 .proc_handler = hugetlb_overcommit_handler, 2807 }, 2808 #endif 2809 { 2810 .procname = "lowmem_reserve_ratio", 2811 .data = &sysctl_lowmem_reserve_ratio, 2812 .maxlen = sizeof(sysctl_lowmem_reserve_ratio), 2813 .mode = 0644, 2814 .proc_handler = lowmem_reserve_ratio_sysctl_handler, 2815 }, 2816 { 2817 .procname = "drop_caches", 2818 .data = &sysctl_drop_caches, 2819 .maxlen = sizeof(int), 2820 .mode = 0200, 2821 .proc_handler = drop_caches_sysctl_handler, 2822 .extra1 = SYSCTL_ONE, 2823 .extra2 = &four, 2824 }, 2825 #ifdef CONFIG_COMPACTION 2826 { 2827 .procname = "compact_memory", 2828 .data = &sysctl_compact_memory, 2829 .maxlen = sizeof(int), 2830 .mode = 0200, 2831 .proc_handler = sysctl_compaction_handler, 2832 }, 2833 { 2834 .procname = "extfrag_threshold", 2835 .data = &sysctl_extfrag_threshold, 2836 .maxlen = sizeof(int), 2837 .mode = 0644, 2838 .proc_handler = proc_dointvec_minmax, 2839 .extra1 = &min_extfrag_threshold, 2840 .extra2 = &max_extfrag_threshold, 2841 }, 2842 { 2843 .procname = "compact_unevictable_allowed", 2844 .data = &sysctl_compact_unevictable_allowed, 2845 .maxlen = sizeof(int), 2846 .mode = 0644, 2847 .proc_handler = proc_dointvec_minmax_warn_RT_change, 2848 .extra1 = SYSCTL_ZERO, 2849 .extra2 = SYSCTL_ONE, 2850 }, 2851 2852 #endif /* CONFIG_COMPACTION */ 2853 { 2854 .procname = "min_free_kbytes", 2855 .data = &min_free_kbytes, 2856 .maxlen = sizeof(min_free_kbytes), 2857 .mode = 0644, 2858 .proc_handler = min_free_kbytes_sysctl_handler, 2859 .extra1 = SYSCTL_ZERO, 2860 }, 2861 { 2862 .procname = "watermark_boost_factor", 2863 .data = &watermark_boost_factor, 2864 .maxlen = sizeof(watermark_boost_factor), 2865 .mode = 0644, 2866 .proc_handler = proc_dointvec_minmax, 2867 .extra1 = SYSCTL_ZERO, 2868 }, 2869 { 2870 .procname = "watermark_scale_factor", 2871 .data = &watermark_scale_factor, 2872 .maxlen = sizeof(watermark_scale_factor), 2873 .mode = 0644, 2874 .proc_handler = watermark_scale_factor_sysctl_handler, 2875 .extra1 = SYSCTL_ONE, 2876 .extra2 = &one_thousand, 2877 }, 2878 { 2879 .procname = "percpu_pagelist_fraction", 2880 .data = &percpu_pagelist_fraction, 2881 .maxlen = sizeof(percpu_pagelist_fraction), 2882 .mode = 0644, 2883 .proc_handler = percpu_pagelist_fraction_sysctl_handler, 2884 .extra1 = SYSCTL_ZERO, 2885 }, 2886 #ifdef CONFIG_MMU 2887 { 2888 .procname = "max_map_count", 2889 .data = &sysctl_max_map_count, 2890 .maxlen = sizeof(sysctl_max_map_count), 2891 .mode = 0644, 2892 .proc_handler = proc_dointvec_minmax, 2893 .extra1 = SYSCTL_ZERO, 2894 }, 2895 #else 2896 { 2897 .procname = "nr_trim_pages", 2898 .data = &sysctl_nr_trim_pages, 2899 .maxlen = sizeof(sysctl_nr_trim_pages), 2900 .mode = 0644, 2901 .proc_handler = proc_dointvec_minmax, 2902 .extra1 = SYSCTL_ZERO, 2903 }, 2904 #endif 2905 { 2906 .procname = "laptop_mode", 2907 .data = &laptop_mode, 2908 .maxlen = sizeof(laptop_mode), 2909 .mode = 0644, 2910 .proc_handler = proc_dointvec_jiffies, 2911 }, 2912 { 2913 .procname = "block_dump", 2914 .data = &block_dump, 2915 .maxlen = sizeof(block_dump), 2916 .mode = 0644, 2917 .proc_handler = proc_dointvec, 2918 .extra1 = SYSCTL_ZERO, 2919 }, 2920 { 2921 .procname = "vfs_cache_pressure", 2922 .data = &sysctl_vfs_cache_pressure, 2923 .maxlen = sizeof(sysctl_vfs_cache_pressure), 2924 .mode = 0644, 2925 .proc_handler = proc_dointvec, 2926 .extra1 = SYSCTL_ZERO, 2927 }, 2928 #if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \ 2929 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT) 2930 { 2931 .procname = "legacy_va_layout", 2932 .data = &sysctl_legacy_va_layout, 2933 .maxlen = sizeof(sysctl_legacy_va_layout), 2934 .mode = 0644, 2935 .proc_handler = proc_dointvec, 2936 .extra1 = SYSCTL_ZERO, 2937 }, 2938 #endif 2939 #ifdef CONFIG_NUMA 2940 { 2941 .procname = "zone_reclaim_mode", 2942 .data = &node_reclaim_mode, 2943 .maxlen = sizeof(node_reclaim_mode), 2944 .mode = 0644, 2945 .proc_handler = proc_dointvec, 2946 .extra1 = SYSCTL_ZERO, 2947 }, 2948 { 2949 .procname = "min_unmapped_ratio", 2950 .data = &sysctl_min_unmapped_ratio, 2951 .maxlen = sizeof(sysctl_min_unmapped_ratio), 2952 .mode = 0644, 2953 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler, 2954 .extra1 = SYSCTL_ZERO, 2955 .extra2 = &one_hundred, 2956 }, 2957 { 2958 .procname = "min_slab_ratio", 2959 .data = &sysctl_min_slab_ratio, 2960 .maxlen = sizeof(sysctl_min_slab_ratio), 2961 .mode = 0644, 2962 .proc_handler = sysctl_min_slab_ratio_sysctl_handler, 2963 .extra1 = SYSCTL_ZERO, 2964 .extra2 = &one_hundred, 2965 }, 2966 #endif 2967 #ifdef CONFIG_SMP 2968 { 2969 .procname = "stat_interval", 2970 .data = &sysctl_stat_interval, 2971 .maxlen = sizeof(sysctl_stat_interval), 2972 .mode = 0644, 2973 .proc_handler = proc_dointvec_jiffies, 2974 }, 2975 { 2976 .procname = "stat_refresh", 2977 .data = NULL, 2978 .maxlen = 0, 2979 .mode = 0600, 2980 .proc_handler = vmstat_refresh, 2981 }, 2982 #endif 2983 #ifdef CONFIG_MMU 2984 { 2985 .procname = "mmap_min_addr", 2986 .data = &dac_mmap_min_addr, 2987 .maxlen = sizeof(unsigned long), 2988 .mode = 0644, 2989 .proc_handler = mmap_min_addr_handler, 2990 }, 2991 #endif 2992 #ifdef CONFIG_NUMA 2993 { 2994 .procname = "numa_zonelist_order", 2995 .data = &numa_zonelist_order, 2996 .maxlen = NUMA_ZONELIST_ORDER_LEN, 2997 .mode = 0644, 2998 .proc_handler = numa_zonelist_order_handler, 2999 }, 3000 #endif 3001 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \ 3002 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL)) 3003 { 3004 .procname = "vdso_enabled", 3005 #ifdef CONFIG_X86_32 3006 .data = &vdso32_enabled, 3007 .maxlen = sizeof(vdso32_enabled), 3008 #else 3009 .data = &vdso_enabled, 3010 .maxlen = sizeof(vdso_enabled), 3011 #endif 3012 .mode = 0644, 3013 .proc_handler = proc_dointvec, 3014 .extra1 = SYSCTL_ZERO, 3015 }, 3016 #endif 3017 #ifdef CONFIG_HIGHMEM 3018 { 3019 .procname = "highmem_is_dirtyable", 3020 .data = &vm_highmem_is_dirtyable, 3021 .maxlen = sizeof(vm_highmem_is_dirtyable), 3022 .mode = 0644, 3023 .proc_handler = proc_dointvec_minmax, 3024 .extra1 = SYSCTL_ZERO, 3025 .extra2 = SYSCTL_ONE, 3026 }, 3027 #endif 3028 #ifdef CONFIG_MEMORY_FAILURE 3029 { 3030 .procname = "memory_failure_early_kill", 3031 .data = &sysctl_memory_failure_early_kill, 3032 .maxlen = sizeof(sysctl_memory_failure_early_kill), 3033 .mode = 0644, 3034 .proc_handler = proc_dointvec_minmax, 3035 .extra1 = SYSCTL_ZERO, 3036 .extra2 = SYSCTL_ONE, 3037 }, 3038 { 3039 .procname = "memory_failure_recovery", 3040 .data = &sysctl_memory_failure_recovery, 3041 .maxlen = sizeof(sysctl_memory_failure_recovery), 3042 .mode = 0644, 3043 .proc_handler = proc_dointvec_minmax, 3044 .extra1 = SYSCTL_ZERO, 3045 .extra2 = SYSCTL_ONE, 3046 }, 3047 #endif 3048 { 3049 .procname = "user_reserve_kbytes", 3050 .data = &sysctl_user_reserve_kbytes, 3051 .maxlen = sizeof(sysctl_user_reserve_kbytes), 3052 .mode = 0644, 3053 .proc_handler = proc_doulongvec_minmax, 3054 }, 3055 { 3056 .procname = "admin_reserve_kbytes", 3057 .data = &sysctl_admin_reserve_kbytes, 3058 .maxlen = sizeof(sysctl_admin_reserve_kbytes), 3059 .mode = 0644, 3060 .proc_handler = proc_doulongvec_minmax, 3061 }, 3062 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS 3063 { 3064 .procname = "mmap_rnd_bits", 3065 .data = &mmap_rnd_bits, 3066 .maxlen = sizeof(mmap_rnd_bits), 3067 .mode = 0600, 3068 .proc_handler = proc_dointvec_minmax, 3069 .extra1 = (void *)&mmap_rnd_bits_min, 3070 .extra2 = (void *)&mmap_rnd_bits_max, 3071 }, 3072 #endif 3073 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS 3074 { 3075 .procname = "mmap_rnd_compat_bits", 3076 .data = &mmap_rnd_compat_bits, 3077 .maxlen = sizeof(mmap_rnd_compat_bits), 3078 .mode = 0600, 3079 .proc_handler = proc_dointvec_minmax, 3080 .extra1 = (void *)&mmap_rnd_compat_bits_min, 3081 .extra2 = (void *)&mmap_rnd_compat_bits_max, 3082 }, 3083 #endif 3084 #ifdef CONFIG_USERFAULTFD 3085 { 3086 .procname = "unprivileged_userfaultfd", 3087 .data = &sysctl_unprivileged_userfaultfd, 3088 .maxlen = sizeof(sysctl_unprivileged_userfaultfd), 3089 .mode = 0644, 3090 .proc_handler = proc_dointvec_minmax, 3091 .extra1 = SYSCTL_ZERO, 3092 .extra2 = SYSCTL_ONE, 3093 }, 3094 #endif 3095 { } 3096 }; 3097 3098 static struct ctl_table fs_table[] = { 3099 { 3100 .procname = "inode-nr", 3101 .data = &inodes_stat, 3102 .maxlen = 2*sizeof(long), 3103 .mode = 0444, 3104 .proc_handler = proc_nr_inodes, 3105 }, 3106 { 3107 .procname = "inode-state", 3108 .data = &inodes_stat, 3109 .maxlen = 7*sizeof(long), 3110 .mode = 0444, 3111 .proc_handler = proc_nr_inodes, 3112 }, 3113 { 3114 .procname = "file-nr", 3115 .data = &files_stat, 3116 .maxlen = sizeof(files_stat), 3117 .mode = 0444, 3118 .proc_handler = proc_nr_files, 3119 }, 3120 { 3121 .procname = "file-max", 3122 .data = &files_stat.max_files, 3123 .maxlen = sizeof(files_stat.max_files), 3124 .mode = 0644, 3125 .proc_handler = proc_doulongvec_minmax, 3126 .extra1 = &zero_ul, 3127 .extra2 = &long_max, 3128 }, 3129 { 3130 .procname = "nr_open", 3131 .data = &sysctl_nr_open, 3132 .maxlen = sizeof(unsigned int), 3133 .mode = 0644, 3134 .proc_handler = proc_dointvec_minmax, 3135 .extra1 = &sysctl_nr_open_min, 3136 .extra2 = &sysctl_nr_open_max, 3137 }, 3138 { 3139 .procname = "dentry-state", 3140 .data = &dentry_stat, 3141 .maxlen = 6*sizeof(long), 3142 .mode = 0444, 3143 .proc_handler = proc_nr_dentry, 3144 }, 3145 { 3146 .procname = "overflowuid", 3147 .data = &fs_overflowuid, 3148 .maxlen = sizeof(int), 3149 .mode = 0644, 3150 .proc_handler = proc_dointvec_minmax, 3151 .extra1 = &minolduid, 3152 .extra2 = &maxolduid, 3153 }, 3154 { 3155 .procname = "overflowgid", 3156 .data = &fs_overflowgid, 3157 .maxlen = sizeof(int), 3158 .mode = 0644, 3159 .proc_handler = proc_dointvec_minmax, 3160 .extra1 = &minolduid, 3161 .extra2 = &maxolduid, 3162 }, 3163 #ifdef CONFIG_FILE_LOCKING 3164 { 3165 .procname = "leases-enable", 3166 .data = &leases_enable, 3167 .maxlen = sizeof(int), 3168 .mode = 0644, 3169 .proc_handler = proc_dointvec, 3170 }, 3171 #endif 3172 #ifdef CONFIG_DNOTIFY 3173 { 3174 .procname = "dir-notify-enable", 3175 .data = &dir_notify_enable, 3176 .maxlen = sizeof(int), 3177 .mode = 0644, 3178 .proc_handler = proc_dointvec, 3179 }, 3180 #endif 3181 #ifdef CONFIG_MMU 3182 #ifdef CONFIG_FILE_LOCKING 3183 { 3184 .procname = "lease-break-time", 3185 .data = &lease_break_time, 3186 .maxlen = sizeof(int), 3187 .mode = 0644, 3188 .proc_handler = proc_dointvec, 3189 }, 3190 #endif 3191 #ifdef CONFIG_AIO 3192 { 3193 .procname = "aio-nr", 3194 .data = &aio_nr, 3195 .maxlen = sizeof(aio_nr), 3196 .mode = 0444, 3197 .proc_handler = proc_doulongvec_minmax, 3198 }, 3199 { 3200 .procname = "aio-max-nr", 3201 .data = &aio_max_nr, 3202 .maxlen = sizeof(aio_max_nr), 3203 .mode = 0644, 3204 .proc_handler = proc_doulongvec_minmax, 3205 }, 3206 #endif /* CONFIG_AIO */ 3207 #ifdef CONFIG_INOTIFY_USER 3208 { 3209 .procname = "inotify", 3210 .mode = 0555, 3211 .child = inotify_table, 3212 }, 3213 #endif 3214 #ifdef CONFIG_EPOLL 3215 { 3216 .procname = "epoll", 3217 .mode = 0555, 3218 .child = epoll_table, 3219 }, 3220 #endif 3221 #endif 3222 { 3223 .procname = "protected_symlinks", 3224 .data = &sysctl_protected_symlinks, 3225 .maxlen = sizeof(int), 3226 .mode = 0600, 3227 .proc_handler = proc_dointvec_minmax, 3228 .extra1 = SYSCTL_ZERO, 3229 .extra2 = SYSCTL_ONE, 3230 }, 3231 { 3232 .procname = "protected_hardlinks", 3233 .data = &sysctl_protected_hardlinks, 3234 .maxlen = sizeof(int), 3235 .mode = 0600, 3236 .proc_handler = proc_dointvec_minmax, 3237 .extra1 = SYSCTL_ZERO, 3238 .extra2 = SYSCTL_ONE, 3239 }, 3240 { 3241 .procname = "protected_fifos", 3242 .data = &sysctl_protected_fifos, 3243 .maxlen = sizeof(int), 3244 .mode = 0600, 3245 .proc_handler = proc_dointvec_minmax, 3246 .extra1 = SYSCTL_ZERO, 3247 .extra2 = &two, 3248 }, 3249 { 3250 .procname = "protected_regular", 3251 .data = &sysctl_protected_regular, 3252 .maxlen = sizeof(int), 3253 .mode = 0600, 3254 .proc_handler = proc_dointvec_minmax, 3255 .extra1 = SYSCTL_ZERO, 3256 .extra2 = &two, 3257 }, 3258 { 3259 .procname = "suid_dumpable", 3260 .data = &suid_dumpable, 3261 .maxlen = sizeof(int), 3262 .mode = 0644, 3263 .proc_handler = proc_dointvec_minmax_coredump, 3264 .extra1 = SYSCTL_ZERO, 3265 .extra2 = &two, 3266 }, 3267 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE) 3268 { 3269 .procname = "binfmt_misc", 3270 .mode = 0555, 3271 .child = sysctl_mount_point, 3272 }, 3273 #endif 3274 { 3275 .procname = "pipe-max-size", 3276 .data = &pipe_max_size, 3277 .maxlen = sizeof(pipe_max_size), 3278 .mode = 0644, 3279 .proc_handler = proc_dopipe_max_size, 3280 }, 3281 { 3282 .procname = "pipe-user-pages-hard", 3283 .data = &pipe_user_pages_hard, 3284 .maxlen = sizeof(pipe_user_pages_hard), 3285 .mode = 0644, 3286 .proc_handler = proc_doulongvec_minmax, 3287 }, 3288 { 3289 .procname = "pipe-user-pages-soft", 3290 .data = &pipe_user_pages_soft, 3291 .maxlen = sizeof(pipe_user_pages_soft), 3292 .mode = 0644, 3293 .proc_handler = proc_doulongvec_minmax, 3294 }, 3295 { 3296 .procname = "mount-max", 3297 .data = &sysctl_mount_max, 3298 .maxlen = sizeof(unsigned int), 3299 .mode = 0644, 3300 .proc_handler = proc_dointvec_minmax, 3301 .extra1 = SYSCTL_ONE, 3302 }, 3303 { } 3304 }; 3305 3306 static struct ctl_table debug_table[] = { 3307 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE 3308 { 3309 .procname = "exception-trace", 3310 .data = &show_unhandled_signals, 3311 .maxlen = sizeof(int), 3312 .mode = 0644, 3313 .proc_handler = proc_dointvec 3314 }, 3315 #endif 3316 #if defined(CONFIG_OPTPROBES) 3317 { 3318 .procname = "kprobes-optimization", 3319 .data = &sysctl_kprobes_optimization, 3320 .maxlen = sizeof(int), 3321 .mode = 0644, 3322 .proc_handler = proc_kprobes_optimization_handler, 3323 .extra1 = SYSCTL_ZERO, 3324 .extra2 = SYSCTL_ONE, 3325 }, 3326 #endif 3327 { } 3328 }; 3329 3330 static struct ctl_table dev_table[] = { 3331 { } 3332 }; 3333 3334 static struct ctl_table sysctl_base_table[] = { 3335 { 3336 .procname = "kernel", 3337 .mode = 0555, 3338 .child = kern_table, 3339 }, 3340 { 3341 .procname = "vm", 3342 .mode = 0555, 3343 .child = vm_table, 3344 }, 3345 { 3346 .procname = "fs", 3347 .mode = 0555, 3348 .child = fs_table, 3349 }, 3350 { 3351 .procname = "debug", 3352 .mode = 0555, 3353 .child = debug_table, 3354 }, 3355 { 3356 .procname = "dev", 3357 .mode = 0555, 3358 .child = dev_table, 3359 }, 3360 { } 3361 }; 3362 3363 int __init sysctl_init(void) 3364 { 3365 struct ctl_table_header *hdr; 3366 3367 hdr = register_sysctl_table(sysctl_base_table); 3368 kmemleak_not_leak(hdr); 3369 return 0; 3370 } 3371 #endif /* CONFIG_SYSCTL */ 3372 /* 3373 * No sense putting this after each symbol definition, twice, 3374 * exception granted :-) 3375 */ 3376 EXPORT_SYMBOL(proc_dointvec); 3377 EXPORT_SYMBOL(proc_douintvec); 3378 EXPORT_SYMBOL(proc_dointvec_jiffies); 3379 EXPORT_SYMBOL(proc_dointvec_minmax); 3380 EXPORT_SYMBOL_GPL(proc_douintvec_minmax); 3381 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies); 3382 EXPORT_SYMBOL(proc_dointvec_ms_jiffies); 3383 EXPORT_SYMBOL(proc_dostring); 3384 EXPORT_SYMBOL(proc_doulongvec_minmax); 3385 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax); 3386 EXPORT_SYMBOL(proc_do_large_bitmap); 3387