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