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