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. */ 99 extern int suid_dumpable; 100 #ifdef CONFIG_COREDUMP 101 extern int core_uses_pid; 102 extern char core_pattern[]; 103 extern unsigned int core_pipe_limit; 104 #endif 105 extern int pid_max; 106 extern int pid_max_min, pid_max_max; 107 extern int percpu_pagelist_fraction; 108 extern int compat_log; 109 extern int latencytop_enabled; 110 extern int sysctl_nr_open_min, sysctl_nr_open_max; 111 #ifndef CONFIG_MMU 112 extern int sysctl_nr_trim_pages; 113 #endif 114 115 /* Constants used for minimum and maximum */ 116 #ifdef CONFIG_LOCKUP_DETECTOR 117 static int sixty = 60; 118 #endif 119 120 static int __maybe_unused neg_one = -1; 121 122 static int zero; 123 static int __maybe_unused one = 1; 124 static int __maybe_unused two = 2; 125 static int __maybe_unused four = 4; 126 static unsigned long one_ul = 1; 127 static int one_hundred = 100; 128 #ifdef CONFIG_PRINTK 129 static int ten_thousand = 10000; 130 #endif 131 132 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */ 133 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE; 134 135 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */ 136 static int maxolduid = 65535; 137 static int minolduid; 138 139 static int ngroups_max = NGROUPS_MAX; 140 static 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 144 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ); 145 #endif 146 147 #ifdef CONFIG_INOTIFY_USER 148 #include <linux/inotify.h> 149 #endif 150 #ifdef CONFIG_SPARC 151 #endif 152 153 #ifdef __hppa__ 154 extern int pwrsw_enabled; 155 #endif 156 157 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW 158 extern int unaligned_enabled; 159 #endif 160 161 #ifdef CONFIG_IA64 162 extern int unaligned_dump_stack; 163 #endif 164 165 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN 166 extern 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 1 174 175 static int sysctl_writes_strict = SYSCTL_WRITES_WARN; 176 177 static int proc_do_cad_pid(struct ctl_table *table, int write, 178 void __user *buffer, size_t *lenp, loff_t *ppos); 179 static 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 184 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write, 185 void __user *buffer, size_t *lenp, loff_t *ppos); 186 #endif 187 188 static 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 191 static 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 */ 197 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE; 198 199 static 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 216 217 static struct ctl_table kern_table[]; 218 static struct ctl_table vm_table[]; 219 static struct ctl_table fs_table[]; 220 static struct ctl_table debug_table[]; 221 static struct ctl_table dev_table[]; 222 extern struct ctl_table random_table[]; 223 #ifdef CONFIG_EPOLL 224 extern struct ctl_table epoll_table[]; 225 #endif 226 227 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT 228 int sysctl_legacy_va_layout; 229 #endif 230 231 /* The default sysctl tables: */ 232 233 static 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 263 static int min_sched_granularity_ns = 100000; /* 100 usecs */ 264 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */ 265 static int min_wakeup_granularity_ns; /* 0 usecs */ 266 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */ 267 #ifdef CONFIG_SMP 268 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE; 269 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1; 270 #endif /* CONFIG_SMP */ 271 #endif /* CONFIG_SCHED_DEBUG */ 272 273 #ifdef CONFIG_COMPACTION 274 static int min_extfrag_threshold; 275 static int max_extfrag_threshold = 1000; 276 #endif 277 278 static 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 1138 static 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 1534 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE) 1535 static struct ctl_table binfmt_misc_table[] = { 1536 { } 1537 }; 1538 #endif 1539 1540 static struct ctl_table fs_table[] = { 1541 { 1542 .procname = "inode-nr", 1543 .data = &inodes_stat, 1544 .maxlen = 2*sizeof(long), 1545 .mode = 0444, 1546 .proc_handler = proc_nr_inodes, 1547 }, 1548 { 1549 .procname = "inode-state", 1550 .data = &inodes_stat, 1551 .maxlen = 7*sizeof(long), 1552 .mode = 0444, 1553 .proc_handler = proc_nr_inodes, 1554 }, 1555 { 1556 .procname = "file-nr", 1557 .data = &files_stat, 1558 .maxlen = sizeof(files_stat), 1559 .mode = 0444, 1560 .proc_handler = proc_nr_files, 1561 }, 1562 { 1563 .procname = "file-max", 1564 .data = &files_stat.max_files, 1565 .maxlen = sizeof(files_stat.max_files), 1566 .mode = 0644, 1567 .proc_handler = proc_doulongvec_minmax, 1568 }, 1569 { 1570 .procname = "nr_open", 1571 .data = &sysctl_nr_open, 1572 .maxlen = sizeof(int), 1573 .mode = 0644, 1574 .proc_handler = proc_dointvec_minmax, 1575 .extra1 = &sysctl_nr_open_min, 1576 .extra2 = &sysctl_nr_open_max, 1577 }, 1578 { 1579 .procname = "dentry-state", 1580 .data = &dentry_stat, 1581 .maxlen = 6*sizeof(long), 1582 .mode = 0444, 1583 .proc_handler = proc_nr_dentry, 1584 }, 1585 { 1586 .procname = "overflowuid", 1587 .data = &fs_overflowuid, 1588 .maxlen = sizeof(int), 1589 .mode = 0644, 1590 .proc_handler = proc_dointvec_minmax, 1591 .extra1 = &minolduid, 1592 .extra2 = &maxolduid, 1593 }, 1594 { 1595 .procname = "overflowgid", 1596 .data = &fs_overflowgid, 1597 .maxlen = sizeof(int), 1598 .mode = 0644, 1599 .proc_handler = proc_dointvec_minmax, 1600 .extra1 = &minolduid, 1601 .extra2 = &maxolduid, 1602 }, 1603 #ifdef CONFIG_FILE_LOCKING 1604 { 1605 .procname = "leases-enable", 1606 .data = &leases_enable, 1607 .maxlen = sizeof(int), 1608 .mode = 0644, 1609 .proc_handler = proc_dointvec, 1610 }, 1611 #endif 1612 #ifdef CONFIG_DNOTIFY 1613 { 1614 .procname = "dir-notify-enable", 1615 .data = &dir_notify_enable, 1616 .maxlen = sizeof(int), 1617 .mode = 0644, 1618 .proc_handler = proc_dointvec, 1619 }, 1620 #endif 1621 #ifdef CONFIG_MMU 1622 #ifdef CONFIG_FILE_LOCKING 1623 { 1624 .procname = "lease-break-time", 1625 .data = &lease_break_time, 1626 .maxlen = sizeof(int), 1627 .mode = 0644, 1628 .proc_handler = proc_dointvec, 1629 }, 1630 #endif 1631 #ifdef CONFIG_AIO 1632 { 1633 .procname = "aio-nr", 1634 .data = &aio_nr, 1635 .maxlen = sizeof(aio_nr), 1636 .mode = 0444, 1637 .proc_handler = proc_doulongvec_minmax, 1638 }, 1639 { 1640 .procname = "aio-max-nr", 1641 .data = &aio_max_nr, 1642 .maxlen = sizeof(aio_max_nr), 1643 .mode = 0644, 1644 .proc_handler = proc_doulongvec_minmax, 1645 }, 1646 #endif /* CONFIG_AIO */ 1647 #ifdef CONFIG_INOTIFY_USER 1648 { 1649 .procname = "inotify", 1650 .mode = 0555, 1651 .child = inotify_table, 1652 }, 1653 #endif 1654 #ifdef CONFIG_EPOLL 1655 { 1656 .procname = "epoll", 1657 .mode = 0555, 1658 .child = epoll_table, 1659 }, 1660 #endif 1661 #endif 1662 { 1663 .procname = "protected_symlinks", 1664 .data = &sysctl_protected_symlinks, 1665 .maxlen = sizeof(int), 1666 .mode = 0600, 1667 .proc_handler = proc_dointvec_minmax, 1668 .extra1 = &zero, 1669 .extra2 = &one, 1670 }, 1671 { 1672 .procname = "protected_hardlinks", 1673 .data = &sysctl_protected_hardlinks, 1674 .maxlen = sizeof(int), 1675 .mode = 0600, 1676 .proc_handler = proc_dointvec_minmax, 1677 .extra1 = &zero, 1678 .extra2 = &one, 1679 }, 1680 { 1681 .procname = "suid_dumpable", 1682 .data = &suid_dumpable, 1683 .maxlen = sizeof(int), 1684 .mode = 0644, 1685 .proc_handler = proc_dointvec_minmax_coredump, 1686 .extra1 = &zero, 1687 .extra2 = &two, 1688 }, 1689 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE) 1690 { 1691 .procname = "binfmt_misc", 1692 .mode = 0555, 1693 .child = binfmt_misc_table, 1694 }, 1695 #endif 1696 { 1697 .procname = "pipe-max-size", 1698 .data = &pipe_max_size, 1699 .maxlen = sizeof(int), 1700 .mode = 0644, 1701 .proc_handler = &pipe_proc_fn, 1702 .extra1 = &pipe_min_size, 1703 }, 1704 { } 1705 }; 1706 1707 static struct ctl_table debug_table[] = { 1708 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE 1709 { 1710 .procname = "exception-trace", 1711 .data = &show_unhandled_signals, 1712 .maxlen = sizeof(int), 1713 .mode = 0644, 1714 .proc_handler = proc_dointvec 1715 }, 1716 #endif 1717 #if defined(CONFIG_OPTPROBES) 1718 { 1719 .procname = "kprobes-optimization", 1720 .data = &sysctl_kprobes_optimization, 1721 .maxlen = sizeof(int), 1722 .mode = 0644, 1723 .proc_handler = proc_kprobes_optimization_handler, 1724 .extra1 = &zero, 1725 .extra2 = &one, 1726 }, 1727 #endif 1728 { } 1729 }; 1730 1731 static struct ctl_table dev_table[] = { 1732 { } 1733 }; 1734 1735 int __init sysctl_init(void) 1736 { 1737 struct ctl_table_header *hdr; 1738 1739 hdr = register_sysctl_table(sysctl_base_table); 1740 kmemleak_not_leak(hdr); 1741 return 0; 1742 } 1743 1744 #endif /* CONFIG_SYSCTL */ 1745 1746 /* 1747 * /proc/sys support 1748 */ 1749 1750 #ifdef CONFIG_PROC_SYSCTL 1751 1752 static int _proc_do_string(char *data, int maxlen, int write, 1753 char __user *buffer, 1754 size_t *lenp, loff_t *ppos) 1755 { 1756 size_t len; 1757 char __user *p; 1758 char c; 1759 1760 if (!data || !maxlen || !*lenp) { 1761 *lenp = 0; 1762 return 0; 1763 } 1764 1765 if (write) { 1766 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) { 1767 /* Only continue writes not past the end of buffer. */ 1768 len = strlen(data); 1769 if (len > maxlen - 1) 1770 len = maxlen - 1; 1771 1772 if (*ppos > len) 1773 return 0; 1774 len = *ppos; 1775 } else { 1776 /* Start writing from beginning of buffer. */ 1777 len = 0; 1778 } 1779 1780 *ppos += *lenp; 1781 p = buffer; 1782 while ((p - buffer) < *lenp && len < maxlen - 1) { 1783 if (get_user(c, p++)) 1784 return -EFAULT; 1785 if (c == 0 || c == '\n') 1786 break; 1787 data[len++] = c; 1788 } 1789 data[len] = 0; 1790 } else { 1791 len = strlen(data); 1792 if (len > maxlen) 1793 len = maxlen; 1794 1795 if (*ppos > len) { 1796 *lenp = 0; 1797 return 0; 1798 } 1799 1800 data += *ppos; 1801 len -= *ppos; 1802 1803 if (len > *lenp) 1804 len = *lenp; 1805 if (len) 1806 if (copy_to_user(buffer, data, len)) 1807 return -EFAULT; 1808 if (len < *lenp) { 1809 if (put_user('\n', buffer + len)) 1810 return -EFAULT; 1811 len++; 1812 } 1813 *lenp = len; 1814 *ppos += len; 1815 } 1816 return 0; 1817 } 1818 1819 static void warn_sysctl_write(struct ctl_table *table) 1820 { 1821 pr_warn_once("%s wrote to %s when file position was not 0!\n" 1822 "This will not be supported in the future. To silence this\n" 1823 "warning, set kernel.sysctl_writes_strict = -1\n", 1824 current->comm, table->procname); 1825 } 1826 1827 /** 1828 * proc_dostring - read a string sysctl 1829 * @table: the sysctl table 1830 * @write: %TRUE if this is a write to the sysctl file 1831 * @buffer: the user buffer 1832 * @lenp: the size of the user buffer 1833 * @ppos: file position 1834 * 1835 * Reads/writes a string from/to the user buffer. If the kernel 1836 * buffer provided is not large enough to hold the string, the 1837 * string is truncated. The copied string is %NULL-terminated. 1838 * If the string is being read by the user process, it is copied 1839 * and a newline '\n' is added. It is truncated if the buffer is 1840 * not large enough. 1841 * 1842 * Returns 0 on success. 1843 */ 1844 int proc_dostring(struct ctl_table *table, int write, 1845 void __user *buffer, size_t *lenp, loff_t *ppos) 1846 { 1847 if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN) 1848 warn_sysctl_write(table); 1849 1850 return _proc_do_string((char *)(table->data), table->maxlen, write, 1851 (char __user *)buffer, lenp, ppos); 1852 } 1853 1854 static size_t proc_skip_spaces(char **buf) 1855 { 1856 size_t ret; 1857 char *tmp = skip_spaces(*buf); 1858 ret = tmp - *buf; 1859 *buf = tmp; 1860 return ret; 1861 } 1862 1863 static void proc_skip_char(char **buf, size_t *size, const char v) 1864 { 1865 while (*size) { 1866 if (**buf != v) 1867 break; 1868 (*size)--; 1869 (*buf)++; 1870 } 1871 } 1872 1873 #define TMPBUFLEN 22 1874 /** 1875 * proc_get_long - reads an ASCII formatted integer from a user buffer 1876 * 1877 * @buf: a kernel buffer 1878 * @size: size of the kernel buffer 1879 * @val: this is where the number will be stored 1880 * @neg: set to %TRUE if number is negative 1881 * @perm_tr: a vector which contains the allowed trailers 1882 * @perm_tr_len: size of the perm_tr vector 1883 * @tr: pointer to store the trailer character 1884 * 1885 * In case of success %0 is returned and @buf and @size are updated with 1886 * the amount of bytes read. If @tr is non-NULL and a trailing 1887 * character exists (size is non-zero after returning from this 1888 * function), @tr is updated with the trailing character. 1889 */ 1890 static int proc_get_long(char **buf, size_t *size, 1891 unsigned long *val, bool *neg, 1892 const char *perm_tr, unsigned perm_tr_len, char *tr) 1893 { 1894 int len; 1895 char *p, tmp[TMPBUFLEN]; 1896 1897 if (!*size) 1898 return -EINVAL; 1899 1900 len = *size; 1901 if (len > TMPBUFLEN - 1) 1902 len = TMPBUFLEN - 1; 1903 1904 memcpy(tmp, *buf, len); 1905 1906 tmp[len] = 0; 1907 p = tmp; 1908 if (*p == '-' && *size > 1) { 1909 *neg = true; 1910 p++; 1911 } else 1912 *neg = false; 1913 if (!isdigit(*p)) 1914 return -EINVAL; 1915 1916 *val = simple_strtoul(p, &p, 0); 1917 1918 len = p - tmp; 1919 1920 /* We don't know if the next char is whitespace thus we may accept 1921 * invalid integers (e.g. 1234...a) or two integers instead of one 1922 * (e.g. 123...1). So lets not allow such large numbers. */ 1923 if (len == TMPBUFLEN - 1) 1924 return -EINVAL; 1925 1926 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len)) 1927 return -EINVAL; 1928 1929 if (tr && (len < *size)) 1930 *tr = *p; 1931 1932 *buf += len; 1933 *size -= len; 1934 1935 return 0; 1936 } 1937 1938 /** 1939 * proc_put_long - converts an integer to a decimal ASCII formatted string 1940 * 1941 * @buf: the user buffer 1942 * @size: the size of the user buffer 1943 * @val: the integer to be converted 1944 * @neg: sign of the number, %TRUE for negative 1945 * 1946 * In case of success %0 is returned and @buf and @size are updated with 1947 * the amount of bytes written. 1948 */ 1949 static int proc_put_long(void __user **buf, size_t *size, unsigned long val, 1950 bool neg) 1951 { 1952 int len; 1953 char tmp[TMPBUFLEN], *p = tmp; 1954 1955 sprintf(p, "%s%lu", neg ? "-" : "", val); 1956 len = strlen(tmp); 1957 if (len > *size) 1958 len = *size; 1959 if (copy_to_user(*buf, tmp, len)) 1960 return -EFAULT; 1961 *size -= len; 1962 *buf += len; 1963 return 0; 1964 } 1965 #undef TMPBUFLEN 1966 1967 static int proc_put_char(void __user **buf, size_t *size, char c) 1968 { 1969 if (*size) { 1970 char __user **buffer = (char __user **)buf; 1971 if (put_user(c, *buffer)) 1972 return -EFAULT; 1973 (*size)--, (*buffer)++; 1974 *buf = *buffer; 1975 } 1976 return 0; 1977 } 1978 1979 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp, 1980 int *valp, 1981 int write, void *data) 1982 { 1983 if (write) { 1984 if (*negp) { 1985 if (*lvalp > (unsigned long) INT_MAX + 1) 1986 return -EINVAL; 1987 *valp = -*lvalp; 1988 } else { 1989 if (*lvalp > (unsigned long) INT_MAX) 1990 return -EINVAL; 1991 *valp = *lvalp; 1992 } 1993 } else { 1994 int val = *valp; 1995 if (val < 0) { 1996 *negp = true; 1997 *lvalp = (unsigned long)-val; 1998 } else { 1999 *negp = false; 2000 *lvalp = (unsigned long)val; 2001 } 2002 } 2003 return 0; 2004 } 2005 2006 static const char proc_wspace_sep[] = { ' ', '\t', '\n' }; 2007 2008 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table, 2009 int write, void __user *buffer, 2010 size_t *lenp, loff_t *ppos, 2011 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 2012 int write, void *data), 2013 void *data) 2014 { 2015 int *i, vleft, first = 1, err = 0; 2016 unsigned long page = 0; 2017 size_t left; 2018 char *kbuf; 2019 2020 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) { 2021 *lenp = 0; 2022 return 0; 2023 } 2024 2025 i = (int *) tbl_data; 2026 vleft = table->maxlen / sizeof(*i); 2027 left = *lenp; 2028 2029 if (!conv) 2030 conv = do_proc_dointvec_conv; 2031 2032 if (write) { 2033 if (*ppos) { 2034 switch (sysctl_writes_strict) { 2035 case SYSCTL_WRITES_STRICT: 2036 goto out; 2037 case SYSCTL_WRITES_WARN: 2038 warn_sysctl_write(table); 2039 break; 2040 default: 2041 break; 2042 } 2043 } 2044 2045 if (left > PAGE_SIZE - 1) 2046 left = PAGE_SIZE - 1; 2047 page = __get_free_page(GFP_TEMPORARY); 2048 kbuf = (char *) page; 2049 if (!kbuf) 2050 return -ENOMEM; 2051 if (copy_from_user(kbuf, buffer, left)) { 2052 err = -EFAULT; 2053 goto free; 2054 } 2055 kbuf[left] = 0; 2056 } 2057 2058 for (; left && vleft--; i++, first=0) { 2059 unsigned long lval; 2060 bool neg; 2061 2062 if (write) { 2063 left -= proc_skip_spaces(&kbuf); 2064 2065 if (!left) 2066 break; 2067 err = proc_get_long(&kbuf, &left, &lval, &neg, 2068 proc_wspace_sep, 2069 sizeof(proc_wspace_sep), NULL); 2070 if (err) 2071 break; 2072 if (conv(&neg, &lval, i, 1, data)) { 2073 err = -EINVAL; 2074 break; 2075 } 2076 } else { 2077 if (conv(&neg, &lval, i, 0, data)) { 2078 err = -EINVAL; 2079 break; 2080 } 2081 if (!first) 2082 err = proc_put_char(&buffer, &left, '\t'); 2083 if (err) 2084 break; 2085 err = proc_put_long(&buffer, &left, lval, neg); 2086 if (err) 2087 break; 2088 } 2089 } 2090 2091 if (!write && !first && left && !err) 2092 err = proc_put_char(&buffer, &left, '\n'); 2093 if (write && !err && left) 2094 left -= proc_skip_spaces(&kbuf); 2095 free: 2096 if (write) { 2097 free_page(page); 2098 if (first) 2099 return err ? : -EINVAL; 2100 } 2101 *lenp -= left; 2102 out: 2103 *ppos += *lenp; 2104 return err; 2105 } 2106 2107 static int do_proc_dointvec(struct ctl_table *table, int write, 2108 void __user *buffer, size_t *lenp, loff_t *ppos, 2109 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 2110 int write, void *data), 2111 void *data) 2112 { 2113 return __do_proc_dointvec(table->data, table, write, 2114 buffer, lenp, ppos, conv, data); 2115 } 2116 2117 /** 2118 * proc_dointvec - read a vector of integers 2119 * @table: the sysctl table 2120 * @write: %TRUE if this is a write to the sysctl file 2121 * @buffer: the user buffer 2122 * @lenp: the size of the user buffer 2123 * @ppos: file position 2124 * 2125 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2126 * values from/to the user buffer, treated as an ASCII string. 2127 * 2128 * Returns 0 on success. 2129 */ 2130 int proc_dointvec(struct ctl_table *table, int write, 2131 void __user *buffer, size_t *lenp, loff_t *ppos) 2132 { 2133 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2134 NULL,NULL); 2135 } 2136 2137 /* 2138 * Taint values can only be increased 2139 * This means we can safely use a temporary. 2140 */ 2141 static int proc_taint(struct ctl_table *table, int write, 2142 void __user *buffer, size_t *lenp, loff_t *ppos) 2143 { 2144 struct ctl_table t; 2145 unsigned long tmptaint = get_taint(); 2146 int err; 2147 2148 if (write && !capable(CAP_SYS_ADMIN)) 2149 return -EPERM; 2150 2151 t = *table; 2152 t.data = &tmptaint; 2153 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos); 2154 if (err < 0) 2155 return err; 2156 2157 if (write) { 2158 /* 2159 * Poor man's atomic or. Not worth adding a primitive 2160 * to everyone's atomic.h for this 2161 */ 2162 int i; 2163 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) { 2164 if ((tmptaint >> i) & 1) 2165 add_taint(i, LOCKDEP_STILL_OK); 2166 } 2167 } 2168 2169 return err; 2170 } 2171 2172 #ifdef CONFIG_PRINTK 2173 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write, 2174 void __user *buffer, size_t *lenp, loff_t *ppos) 2175 { 2176 if (write && !capable(CAP_SYS_ADMIN)) 2177 return -EPERM; 2178 2179 return proc_dointvec_minmax(table, write, buffer, lenp, ppos); 2180 } 2181 #endif 2182 2183 struct do_proc_dointvec_minmax_conv_param { 2184 int *min; 2185 int *max; 2186 }; 2187 2188 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp, 2189 int *valp, 2190 int write, void *data) 2191 { 2192 struct do_proc_dointvec_minmax_conv_param *param = data; 2193 if (write) { 2194 int val = *negp ? -*lvalp : *lvalp; 2195 if ((param->min && *param->min > val) || 2196 (param->max && *param->max < val)) 2197 return -EINVAL; 2198 *valp = val; 2199 } else { 2200 int val = *valp; 2201 if (val < 0) { 2202 *negp = true; 2203 *lvalp = (unsigned long)-val; 2204 } else { 2205 *negp = false; 2206 *lvalp = (unsigned long)val; 2207 } 2208 } 2209 return 0; 2210 } 2211 2212 /** 2213 * proc_dointvec_minmax - read a vector of integers with min/max values 2214 * @table: the sysctl table 2215 * @write: %TRUE if this is a write to the sysctl file 2216 * @buffer: the user buffer 2217 * @lenp: the size of the user buffer 2218 * @ppos: file position 2219 * 2220 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2221 * values from/to the user buffer, treated as an ASCII string. 2222 * 2223 * This routine will ensure the values are within the range specified by 2224 * table->extra1 (min) and table->extra2 (max). 2225 * 2226 * Returns 0 on success. 2227 */ 2228 int proc_dointvec_minmax(struct ctl_table *table, int write, 2229 void __user *buffer, size_t *lenp, loff_t *ppos) 2230 { 2231 struct do_proc_dointvec_minmax_conv_param param = { 2232 .min = (int *) table->extra1, 2233 .max = (int *) table->extra2, 2234 }; 2235 return do_proc_dointvec(table, write, buffer, lenp, ppos, 2236 do_proc_dointvec_minmax_conv, ¶m); 2237 } 2238 2239 static void validate_coredump_safety(void) 2240 { 2241 #ifdef CONFIG_COREDUMP 2242 if (suid_dumpable == SUID_DUMP_ROOT && 2243 core_pattern[0] != '/' && core_pattern[0] != '|') { 2244 printk(KERN_WARNING "Unsafe core_pattern used with "\ 2245 "suid_dumpable=2. Pipe handler or fully qualified "\ 2246 "core dump path required.\n"); 2247 } 2248 #endif 2249 } 2250 2251 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write, 2252 void __user *buffer, size_t *lenp, loff_t *ppos) 2253 { 2254 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos); 2255 if (!error) 2256 validate_coredump_safety(); 2257 return error; 2258 } 2259 2260 #ifdef CONFIG_COREDUMP 2261 static int proc_dostring_coredump(struct ctl_table *table, int write, 2262 void __user *buffer, size_t *lenp, loff_t *ppos) 2263 { 2264 int error = proc_dostring(table, write, buffer, lenp, ppos); 2265 if (!error) 2266 validate_coredump_safety(); 2267 return error; 2268 } 2269 #endif 2270 2271 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write, 2272 void __user *buffer, 2273 size_t *lenp, loff_t *ppos, 2274 unsigned long convmul, 2275 unsigned long convdiv) 2276 { 2277 unsigned long *i, *min, *max; 2278 int vleft, first = 1, err = 0; 2279 unsigned long page = 0; 2280 size_t left; 2281 char *kbuf; 2282 2283 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) { 2284 *lenp = 0; 2285 return 0; 2286 } 2287 2288 i = (unsigned long *) data; 2289 min = (unsigned long *) table->extra1; 2290 max = (unsigned long *) table->extra2; 2291 vleft = table->maxlen / sizeof(unsigned long); 2292 left = *lenp; 2293 2294 if (write) { 2295 if (*ppos) { 2296 switch (sysctl_writes_strict) { 2297 case SYSCTL_WRITES_STRICT: 2298 goto out; 2299 case SYSCTL_WRITES_WARN: 2300 warn_sysctl_write(table); 2301 break; 2302 default: 2303 break; 2304 } 2305 } 2306 2307 if (left > PAGE_SIZE - 1) 2308 left = PAGE_SIZE - 1; 2309 page = __get_free_page(GFP_TEMPORARY); 2310 kbuf = (char *) page; 2311 if (!kbuf) 2312 return -ENOMEM; 2313 if (copy_from_user(kbuf, buffer, left)) { 2314 err = -EFAULT; 2315 goto free; 2316 } 2317 kbuf[left] = 0; 2318 } 2319 2320 for (; left && vleft--; i++, first = 0) { 2321 unsigned long val; 2322 2323 if (write) { 2324 bool neg; 2325 2326 left -= proc_skip_spaces(&kbuf); 2327 2328 err = proc_get_long(&kbuf, &left, &val, &neg, 2329 proc_wspace_sep, 2330 sizeof(proc_wspace_sep), NULL); 2331 if (err) 2332 break; 2333 if (neg) 2334 continue; 2335 if ((min && val < *min) || (max && val > *max)) 2336 continue; 2337 *i = val; 2338 } else { 2339 val = convdiv * (*i) / convmul; 2340 if (!first) { 2341 err = proc_put_char(&buffer, &left, '\t'); 2342 if (err) 2343 break; 2344 } 2345 err = proc_put_long(&buffer, &left, val, false); 2346 if (err) 2347 break; 2348 } 2349 } 2350 2351 if (!write && !first && left && !err) 2352 err = proc_put_char(&buffer, &left, '\n'); 2353 if (write && !err) 2354 left -= proc_skip_spaces(&kbuf); 2355 free: 2356 if (write) { 2357 free_page(page); 2358 if (first) 2359 return err ? : -EINVAL; 2360 } 2361 *lenp -= left; 2362 out: 2363 *ppos += *lenp; 2364 return err; 2365 } 2366 2367 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write, 2368 void __user *buffer, 2369 size_t *lenp, loff_t *ppos, 2370 unsigned long convmul, 2371 unsigned long convdiv) 2372 { 2373 return __do_proc_doulongvec_minmax(table->data, table, write, 2374 buffer, lenp, ppos, convmul, convdiv); 2375 } 2376 2377 /** 2378 * proc_doulongvec_minmax - read a vector of long integers with min/max values 2379 * @table: the sysctl table 2380 * @write: %TRUE if this is a write to the sysctl file 2381 * @buffer: the user buffer 2382 * @lenp: the size of the user buffer 2383 * @ppos: file position 2384 * 2385 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 2386 * values from/to the user buffer, treated as an ASCII string. 2387 * 2388 * This routine will ensure the values are within the range specified by 2389 * table->extra1 (min) and table->extra2 (max). 2390 * 2391 * Returns 0 on success. 2392 */ 2393 int proc_doulongvec_minmax(struct ctl_table *table, int write, 2394 void __user *buffer, size_t *lenp, loff_t *ppos) 2395 { 2396 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l); 2397 } 2398 2399 /** 2400 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values 2401 * @table: the sysctl table 2402 * @write: %TRUE if this is a write to the sysctl file 2403 * @buffer: the user buffer 2404 * @lenp: the size of the user buffer 2405 * @ppos: file position 2406 * 2407 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 2408 * values from/to the user buffer, treated as an ASCII string. The values 2409 * are treated as milliseconds, and converted to jiffies when they are stored. 2410 * 2411 * This routine will ensure the values are within the range specified by 2412 * table->extra1 (min) and table->extra2 (max). 2413 * 2414 * Returns 0 on success. 2415 */ 2416 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 2417 void __user *buffer, 2418 size_t *lenp, loff_t *ppos) 2419 { 2420 return do_proc_doulongvec_minmax(table, write, buffer, 2421 lenp, ppos, HZ, 1000l); 2422 } 2423 2424 2425 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp, 2426 int *valp, 2427 int write, void *data) 2428 { 2429 if (write) { 2430 if (*lvalp > LONG_MAX / HZ) 2431 return 1; 2432 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ); 2433 } else { 2434 int val = *valp; 2435 unsigned long lval; 2436 if (val < 0) { 2437 *negp = true; 2438 lval = (unsigned long)-val; 2439 } else { 2440 *negp = false; 2441 lval = (unsigned long)val; 2442 } 2443 *lvalp = lval / HZ; 2444 } 2445 return 0; 2446 } 2447 2448 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp, 2449 int *valp, 2450 int write, void *data) 2451 { 2452 if (write) { 2453 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ) 2454 return 1; 2455 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp); 2456 } else { 2457 int val = *valp; 2458 unsigned long lval; 2459 if (val < 0) { 2460 *negp = true; 2461 lval = (unsigned long)-val; 2462 } else { 2463 *negp = false; 2464 lval = (unsigned long)val; 2465 } 2466 *lvalp = jiffies_to_clock_t(lval); 2467 } 2468 return 0; 2469 } 2470 2471 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp, 2472 int *valp, 2473 int write, void *data) 2474 { 2475 if (write) { 2476 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp); 2477 2478 if (jif > INT_MAX) 2479 return 1; 2480 *valp = (int)jif; 2481 } else { 2482 int val = *valp; 2483 unsigned long lval; 2484 if (val < 0) { 2485 *negp = true; 2486 lval = (unsigned long)-val; 2487 } else { 2488 *negp = false; 2489 lval = (unsigned long)val; 2490 } 2491 *lvalp = jiffies_to_msecs(lval); 2492 } 2493 return 0; 2494 } 2495 2496 /** 2497 * proc_dointvec_jiffies - read a vector of integers as seconds 2498 * @table: the sysctl table 2499 * @write: %TRUE if this is a write to the sysctl file 2500 * @buffer: the user buffer 2501 * @lenp: the size of the user buffer 2502 * @ppos: file position 2503 * 2504 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2505 * values from/to the user buffer, treated as an ASCII string. 2506 * The values read are assumed to be in seconds, and are converted into 2507 * jiffies. 2508 * 2509 * Returns 0 on success. 2510 */ 2511 int proc_dointvec_jiffies(struct ctl_table *table, int write, 2512 void __user *buffer, size_t *lenp, loff_t *ppos) 2513 { 2514 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2515 do_proc_dointvec_jiffies_conv,NULL); 2516 } 2517 2518 /** 2519 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds 2520 * @table: the sysctl table 2521 * @write: %TRUE if this is a write to the sysctl file 2522 * @buffer: the user buffer 2523 * @lenp: the size of the user buffer 2524 * @ppos: pointer to the file position 2525 * 2526 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2527 * values from/to the user buffer, treated as an ASCII string. 2528 * The values read are assumed to be in 1/USER_HZ seconds, and 2529 * are converted into jiffies. 2530 * 2531 * Returns 0 on success. 2532 */ 2533 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 2534 void __user *buffer, size_t *lenp, loff_t *ppos) 2535 { 2536 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2537 do_proc_dointvec_userhz_jiffies_conv,NULL); 2538 } 2539 2540 /** 2541 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds 2542 * @table: the sysctl table 2543 * @write: %TRUE if this is a write to the sysctl file 2544 * @buffer: the user buffer 2545 * @lenp: the size of the user buffer 2546 * @ppos: file position 2547 * @ppos: the current position in the file 2548 * 2549 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2550 * values from/to the user buffer, treated as an ASCII string. 2551 * The values read are assumed to be in 1/1000 seconds, and 2552 * are converted into jiffies. 2553 * 2554 * Returns 0 on success. 2555 */ 2556 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, 2557 void __user *buffer, size_t *lenp, loff_t *ppos) 2558 { 2559 return do_proc_dointvec(table, write, buffer, lenp, ppos, 2560 do_proc_dointvec_ms_jiffies_conv, NULL); 2561 } 2562 2563 static int proc_do_cad_pid(struct ctl_table *table, int write, 2564 void __user *buffer, size_t *lenp, loff_t *ppos) 2565 { 2566 struct pid *new_pid; 2567 pid_t tmp; 2568 int r; 2569 2570 tmp = pid_vnr(cad_pid); 2571 2572 r = __do_proc_dointvec(&tmp, table, write, buffer, 2573 lenp, ppos, NULL, NULL); 2574 if (r || !write) 2575 return r; 2576 2577 new_pid = find_get_pid(tmp); 2578 if (!new_pid) 2579 return -ESRCH; 2580 2581 put_pid(xchg(&cad_pid, new_pid)); 2582 return 0; 2583 } 2584 2585 /** 2586 * proc_do_large_bitmap - read/write from/to a large bitmap 2587 * @table: the sysctl table 2588 * @write: %TRUE if this is a write to the sysctl file 2589 * @buffer: the user buffer 2590 * @lenp: the size of the user buffer 2591 * @ppos: file position 2592 * 2593 * The bitmap is stored at table->data and the bitmap length (in bits) 2594 * in table->maxlen. 2595 * 2596 * We use a range comma separated format (e.g. 1,3-4,10-10) so that 2597 * large bitmaps may be represented in a compact manner. Writing into 2598 * the file will clear the bitmap then update it with the given input. 2599 * 2600 * Returns 0 on success. 2601 */ 2602 int proc_do_large_bitmap(struct ctl_table *table, int write, 2603 void __user *buffer, size_t *lenp, loff_t *ppos) 2604 { 2605 int err = 0; 2606 bool first = 1; 2607 size_t left = *lenp; 2608 unsigned long bitmap_len = table->maxlen; 2609 unsigned long *bitmap = *(unsigned long **) table->data; 2610 unsigned long *tmp_bitmap = NULL; 2611 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c; 2612 2613 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) { 2614 *lenp = 0; 2615 return 0; 2616 } 2617 2618 if (write) { 2619 unsigned long page = 0; 2620 char *kbuf; 2621 2622 if (left > PAGE_SIZE - 1) 2623 left = PAGE_SIZE - 1; 2624 2625 page = __get_free_page(GFP_TEMPORARY); 2626 kbuf = (char *) page; 2627 if (!kbuf) 2628 return -ENOMEM; 2629 if (copy_from_user(kbuf, buffer, left)) { 2630 free_page(page); 2631 return -EFAULT; 2632 } 2633 kbuf[left] = 0; 2634 2635 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long), 2636 GFP_KERNEL); 2637 if (!tmp_bitmap) { 2638 free_page(page); 2639 return -ENOMEM; 2640 } 2641 proc_skip_char(&kbuf, &left, '\n'); 2642 while (!err && left) { 2643 unsigned long val_a, val_b; 2644 bool neg; 2645 2646 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a, 2647 sizeof(tr_a), &c); 2648 if (err) 2649 break; 2650 if (val_a >= bitmap_len || neg) { 2651 err = -EINVAL; 2652 break; 2653 } 2654 2655 val_b = val_a; 2656 if (left) { 2657 kbuf++; 2658 left--; 2659 } 2660 2661 if (c == '-') { 2662 err = proc_get_long(&kbuf, &left, &val_b, 2663 &neg, tr_b, sizeof(tr_b), 2664 &c); 2665 if (err) 2666 break; 2667 if (val_b >= bitmap_len || neg || 2668 val_a > val_b) { 2669 err = -EINVAL; 2670 break; 2671 } 2672 if (left) { 2673 kbuf++; 2674 left--; 2675 } 2676 } 2677 2678 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1); 2679 first = 0; 2680 proc_skip_char(&kbuf, &left, '\n'); 2681 } 2682 free_page(page); 2683 } else { 2684 unsigned long bit_a, bit_b = 0; 2685 2686 while (left) { 2687 bit_a = find_next_bit(bitmap, bitmap_len, bit_b); 2688 if (bit_a >= bitmap_len) 2689 break; 2690 bit_b = find_next_zero_bit(bitmap, bitmap_len, 2691 bit_a + 1) - 1; 2692 2693 if (!first) { 2694 err = proc_put_char(&buffer, &left, ','); 2695 if (err) 2696 break; 2697 } 2698 err = proc_put_long(&buffer, &left, bit_a, false); 2699 if (err) 2700 break; 2701 if (bit_a != bit_b) { 2702 err = proc_put_char(&buffer, &left, '-'); 2703 if (err) 2704 break; 2705 err = proc_put_long(&buffer, &left, bit_b, false); 2706 if (err) 2707 break; 2708 } 2709 2710 first = 0; bit_b++; 2711 } 2712 if (!err) 2713 err = proc_put_char(&buffer, &left, '\n'); 2714 } 2715 2716 if (!err) { 2717 if (write) { 2718 if (*ppos) 2719 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len); 2720 else 2721 bitmap_copy(bitmap, tmp_bitmap, bitmap_len); 2722 } 2723 kfree(tmp_bitmap); 2724 *lenp -= left; 2725 *ppos += *lenp; 2726 return 0; 2727 } else { 2728 kfree(tmp_bitmap); 2729 return err; 2730 } 2731 } 2732 2733 #else /* CONFIG_PROC_SYSCTL */ 2734 2735 int proc_dostring(struct ctl_table *table, int write, 2736 void __user *buffer, size_t *lenp, loff_t *ppos) 2737 { 2738 return -ENOSYS; 2739 } 2740 2741 int proc_dointvec(struct ctl_table *table, int write, 2742 void __user *buffer, size_t *lenp, loff_t *ppos) 2743 { 2744 return -ENOSYS; 2745 } 2746 2747 int proc_dointvec_minmax(struct ctl_table *table, int write, 2748 void __user *buffer, size_t *lenp, loff_t *ppos) 2749 { 2750 return -ENOSYS; 2751 } 2752 2753 int proc_dointvec_jiffies(struct ctl_table *table, int write, 2754 void __user *buffer, size_t *lenp, loff_t *ppos) 2755 { 2756 return -ENOSYS; 2757 } 2758 2759 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 2760 void __user *buffer, size_t *lenp, loff_t *ppos) 2761 { 2762 return -ENOSYS; 2763 } 2764 2765 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, 2766 void __user *buffer, size_t *lenp, loff_t *ppos) 2767 { 2768 return -ENOSYS; 2769 } 2770 2771 int proc_doulongvec_minmax(struct ctl_table *table, int write, 2772 void __user *buffer, size_t *lenp, loff_t *ppos) 2773 { 2774 return -ENOSYS; 2775 } 2776 2777 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 2778 void __user *buffer, 2779 size_t *lenp, loff_t *ppos) 2780 { 2781 return -ENOSYS; 2782 } 2783 2784 2785 #endif /* CONFIG_PROC_SYSCTL */ 2786 2787 /* 2788 * No sense putting this after each symbol definition, twice, 2789 * exception granted :-) 2790 */ 2791 EXPORT_SYMBOL(proc_dointvec); 2792 EXPORT_SYMBOL(proc_dointvec_jiffies); 2793 EXPORT_SYMBOL(proc_dointvec_minmax); 2794 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies); 2795 EXPORT_SYMBOL(proc_dointvec_ms_jiffies); 2796 EXPORT_SYMBOL(proc_dostring); 2797 EXPORT_SYMBOL(proc_doulongvec_minmax); 2798 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax); 2799