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 #endif /* CONFIG_SMP */ 353 #ifdef CONFIG_NUMA_BALANCING 354 { 355 .procname = "numa_balancing_scan_delay_ms", 356 .data = &sysctl_numa_balancing_scan_delay, 357 .maxlen = sizeof(unsigned int), 358 .mode = 0644, 359 .proc_handler = proc_dointvec, 360 }, 361 { 362 .procname = "numa_balancing_scan_period_min_ms", 363 .data = &sysctl_numa_balancing_scan_period_min, 364 .maxlen = sizeof(unsigned int), 365 .mode = 0644, 366 .proc_handler = proc_dointvec, 367 }, 368 { 369 .procname = "numa_balancing_scan_period_max_ms", 370 .data = &sysctl_numa_balancing_scan_period_max, 371 .maxlen = sizeof(unsigned int), 372 .mode = 0644, 373 .proc_handler = proc_dointvec, 374 }, 375 { 376 .procname = "numa_balancing_scan_size_mb", 377 .data = &sysctl_numa_balancing_scan_size, 378 .maxlen = sizeof(unsigned int), 379 .mode = 0644, 380 .proc_handler = proc_dointvec_minmax, 381 .extra1 = &one, 382 }, 383 { 384 .procname = "numa_balancing", 385 .data = NULL, /* filled in by handler */ 386 .maxlen = sizeof(unsigned int), 387 .mode = 0644, 388 .proc_handler = sysctl_numa_balancing, 389 .extra1 = &zero, 390 .extra2 = &one, 391 }, 392 #endif /* CONFIG_NUMA_BALANCING */ 393 #endif /* CONFIG_SCHED_DEBUG */ 394 { 395 .procname = "sched_rt_period_us", 396 .data = &sysctl_sched_rt_period, 397 .maxlen = sizeof(unsigned int), 398 .mode = 0644, 399 .proc_handler = sched_rt_handler, 400 }, 401 { 402 .procname = "sched_rt_runtime_us", 403 .data = &sysctl_sched_rt_runtime, 404 .maxlen = sizeof(int), 405 .mode = 0644, 406 .proc_handler = sched_rt_handler, 407 }, 408 { 409 .procname = "sched_rr_timeslice_ms", 410 .data = &sched_rr_timeslice, 411 .maxlen = sizeof(int), 412 .mode = 0644, 413 .proc_handler = sched_rr_handler, 414 }, 415 #ifdef CONFIG_SCHED_AUTOGROUP 416 { 417 .procname = "sched_autogroup_enabled", 418 .data = &sysctl_sched_autogroup_enabled, 419 .maxlen = sizeof(unsigned int), 420 .mode = 0644, 421 .proc_handler = proc_dointvec_minmax, 422 .extra1 = &zero, 423 .extra2 = &one, 424 }, 425 #endif 426 #ifdef CONFIG_CFS_BANDWIDTH 427 { 428 .procname = "sched_cfs_bandwidth_slice_us", 429 .data = &sysctl_sched_cfs_bandwidth_slice, 430 .maxlen = sizeof(unsigned int), 431 .mode = 0644, 432 .proc_handler = proc_dointvec_minmax, 433 .extra1 = &one, 434 }, 435 #endif 436 #ifdef CONFIG_PROVE_LOCKING 437 { 438 .procname = "prove_locking", 439 .data = &prove_locking, 440 .maxlen = sizeof(int), 441 .mode = 0644, 442 .proc_handler = proc_dointvec, 443 }, 444 #endif 445 #ifdef CONFIG_LOCK_STAT 446 { 447 .procname = "lock_stat", 448 .data = &lock_stat, 449 .maxlen = sizeof(int), 450 .mode = 0644, 451 .proc_handler = proc_dointvec, 452 }, 453 #endif 454 { 455 .procname = "panic", 456 .data = &panic_timeout, 457 .maxlen = sizeof(int), 458 .mode = 0644, 459 .proc_handler = proc_dointvec, 460 }, 461 #ifdef CONFIG_COREDUMP 462 { 463 .procname = "core_uses_pid", 464 .data = &core_uses_pid, 465 .maxlen = sizeof(int), 466 .mode = 0644, 467 .proc_handler = proc_dointvec, 468 }, 469 { 470 .procname = "core_pattern", 471 .data = core_pattern, 472 .maxlen = CORENAME_MAX_SIZE, 473 .mode = 0644, 474 .proc_handler = proc_dostring_coredump, 475 }, 476 { 477 .procname = "core_pipe_limit", 478 .data = &core_pipe_limit, 479 .maxlen = sizeof(unsigned int), 480 .mode = 0644, 481 .proc_handler = proc_dointvec, 482 }, 483 #endif 484 #ifdef CONFIG_PROC_SYSCTL 485 { 486 .procname = "tainted", 487 .maxlen = sizeof(long), 488 .mode = 0644, 489 .proc_handler = proc_taint, 490 }, 491 { 492 .procname = "sysctl_writes_strict", 493 .data = &sysctl_writes_strict, 494 .maxlen = sizeof(int), 495 .mode = 0644, 496 .proc_handler = proc_dointvec_minmax, 497 .extra1 = &neg_one, 498 .extra2 = &one, 499 }, 500 #endif 501 #ifdef CONFIG_LATENCYTOP 502 { 503 .procname = "latencytop", 504 .data = &latencytop_enabled, 505 .maxlen = sizeof(int), 506 .mode = 0644, 507 .proc_handler = proc_dointvec, 508 }, 509 #endif 510 #ifdef CONFIG_BLK_DEV_INITRD 511 { 512 .procname = "real-root-dev", 513 .data = &real_root_dev, 514 .maxlen = sizeof(int), 515 .mode = 0644, 516 .proc_handler = proc_dointvec, 517 }, 518 #endif 519 { 520 .procname = "print-fatal-signals", 521 .data = &print_fatal_signals, 522 .maxlen = sizeof(int), 523 .mode = 0644, 524 .proc_handler = proc_dointvec, 525 }, 526 #ifdef CONFIG_SPARC 527 { 528 .procname = "reboot-cmd", 529 .data = reboot_command, 530 .maxlen = 256, 531 .mode = 0644, 532 .proc_handler = proc_dostring, 533 }, 534 { 535 .procname = "stop-a", 536 .data = &stop_a_enabled, 537 .maxlen = sizeof (int), 538 .mode = 0644, 539 .proc_handler = proc_dointvec, 540 }, 541 { 542 .procname = "scons-poweroff", 543 .data = &scons_pwroff, 544 .maxlen = sizeof (int), 545 .mode = 0644, 546 .proc_handler = proc_dointvec, 547 }, 548 #endif 549 #ifdef CONFIG_SPARC64 550 { 551 .procname = "tsb-ratio", 552 .data = &sysctl_tsb_ratio, 553 .maxlen = sizeof (int), 554 .mode = 0644, 555 .proc_handler = proc_dointvec, 556 }, 557 #endif 558 #ifdef __hppa__ 559 { 560 .procname = "soft-power", 561 .data = &pwrsw_enabled, 562 .maxlen = sizeof (int), 563 .mode = 0644, 564 .proc_handler = proc_dointvec, 565 }, 566 #endif 567 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW 568 { 569 .procname = "unaligned-trap", 570 .data = &unaligned_enabled, 571 .maxlen = sizeof (int), 572 .mode = 0644, 573 .proc_handler = proc_dointvec, 574 }, 575 #endif 576 { 577 .procname = "ctrl-alt-del", 578 .data = &C_A_D, 579 .maxlen = sizeof(int), 580 .mode = 0644, 581 .proc_handler = proc_dointvec, 582 }, 583 #ifdef CONFIG_FUNCTION_TRACER 584 { 585 .procname = "ftrace_enabled", 586 .data = &ftrace_enabled, 587 .maxlen = sizeof(int), 588 .mode = 0644, 589 .proc_handler = ftrace_enable_sysctl, 590 }, 591 #endif 592 #ifdef CONFIG_STACK_TRACER 593 { 594 .procname = "stack_tracer_enabled", 595 .data = &stack_tracer_enabled, 596 .maxlen = sizeof(int), 597 .mode = 0644, 598 .proc_handler = stack_trace_sysctl, 599 }, 600 #endif 601 #ifdef CONFIG_TRACING 602 { 603 .procname = "ftrace_dump_on_oops", 604 .data = &ftrace_dump_on_oops, 605 .maxlen = sizeof(int), 606 .mode = 0644, 607 .proc_handler = proc_dointvec, 608 }, 609 { 610 .procname = "traceoff_on_warning", 611 .data = &__disable_trace_on_warning, 612 .maxlen = sizeof(__disable_trace_on_warning), 613 .mode = 0644, 614 .proc_handler = proc_dointvec, 615 }, 616 { 617 .procname = "tracepoint_printk", 618 .data = &tracepoint_printk, 619 .maxlen = sizeof(tracepoint_printk), 620 .mode = 0644, 621 .proc_handler = proc_dointvec, 622 }, 623 #endif 624 #ifdef CONFIG_KEXEC 625 { 626 .procname = "kexec_load_disabled", 627 .data = &kexec_load_disabled, 628 .maxlen = sizeof(int), 629 .mode = 0644, 630 /* only handle a transition from default "0" to "1" */ 631 .proc_handler = proc_dointvec_minmax, 632 .extra1 = &one, 633 .extra2 = &one, 634 }, 635 #endif 636 #ifdef CONFIG_MODULES 637 { 638 .procname = "modprobe", 639 .data = &modprobe_path, 640 .maxlen = KMOD_PATH_LEN, 641 .mode = 0644, 642 .proc_handler = proc_dostring, 643 }, 644 { 645 .procname = "modules_disabled", 646 .data = &modules_disabled, 647 .maxlen = sizeof(int), 648 .mode = 0644, 649 /* only handle a transition from default "0" to "1" */ 650 .proc_handler = proc_dointvec_minmax, 651 .extra1 = &one, 652 .extra2 = &one, 653 }, 654 #endif 655 #ifdef CONFIG_UEVENT_HELPER 656 { 657 .procname = "hotplug", 658 .data = &uevent_helper, 659 .maxlen = UEVENT_HELPER_PATH_LEN, 660 .mode = 0644, 661 .proc_handler = proc_dostring, 662 }, 663 #endif 664 #ifdef CONFIG_CHR_DEV_SG 665 { 666 .procname = "sg-big-buff", 667 .data = &sg_big_buff, 668 .maxlen = sizeof (int), 669 .mode = 0444, 670 .proc_handler = proc_dointvec, 671 }, 672 #endif 673 #ifdef CONFIG_BSD_PROCESS_ACCT 674 { 675 .procname = "acct", 676 .data = &acct_parm, 677 .maxlen = 3*sizeof(int), 678 .mode = 0644, 679 .proc_handler = proc_dointvec, 680 }, 681 #endif 682 #ifdef CONFIG_MAGIC_SYSRQ 683 { 684 .procname = "sysrq", 685 .data = &__sysrq_enabled, 686 .maxlen = sizeof (int), 687 .mode = 0644, 688 .proc_handler = sysrq_sysctl_handler, 689 }, 690 #endif 691 #ifdef CONFIG_PROC_SYSCTL 692 { 693 .procname = "cad_pid", 694 .data = NULL, 695 .maxlen = sizeof (int), 696 .mode = 0600, 697 .proc_handler = proc_do_cad_pid, 698 }, 699 #endif 700 { 701 .procname = "threads-max", 702 .data = NULL, 703 .maxlen = sizeof(int), 704 .mode = 0644, 705 .proc_handler = sysctl_max_threads, 706 }, 707 { 708 .procname = "random", 709 .mode = 0555, 710 .child = random_table, 711 }, 712 { 713 .procname = "usermodehelper", 714 .mode = 0555, 715 .child = usermodehelper_table, 716 }, 717 { 718 .procname = "overflowuid", 719 .data = &overflowuid, 720 .maxlen = sizeof(int), 721 .mode = 0644, 722 .proc_handler = proc_dointvec_minmax, 723 .extra1 = &minolduid, 724 .extra2 = &maxolduid, 725 }, 726 { 727 .procname = "overflowgid", 728 .data = &overflowgid, 729 .maxlen = sizeof(int), 730 .mode = 0644, 731 .proc_handler = proc_dointvec_minmax, 732 .extra1 = &minolduid, 733 .extra2 = &maxolduid, 734 }, 735 #ifdef CONFIG_S390 736 #ifdef CONFIG_MATHEMU 737 { 738 .procname = "ieee_emulation_warnings", 739 .data = &sysctl_ieee_emulation_warnings, 740 .maxlen = sizeof(int), 741 .mode = 0644, 742 .proc_handler = proc_dointvec, 743 }, 744 #endif 745 { 746 .procname = "userprocess_debug", 747 .data = &show_unhandled_signals, 748 .maxlen = sizeof(int), 749 .mode = 0644, 750 .proc_handler = proc_dointvec, 751 }, 752 #endif 753 { 754 .procname = "pid_max", 755 .data = &pid_max, 756 .maxlen = sizeof (int), 757 .mode = 0644, 758 .proc_handler = proc_dointvec_minmax, 759 .extra1 = &pid_max_min, 760 .extra2 = &pid_max_max, 761 }, 762 { 763 .procname = "panic_on_oops", 764 .data = &panic_on_oops, 765 .maxlen = sizeof(int), 766 .mode = 0644, 767 .proc_handler = proc_dointvec, 768 }, 769 #if defined CONFIG_PRINTK 770 { 771 .procname = "printk", 772 .data = &console_loglevel, 773 .maxlen = 4*sizeof(int), 774 .mode = 0644, 775 .proc_handler = proc_dointvec, 776 }, 777 { 778 .procname = "printk_ratelimit", 779 .data = &printk_ratelimit_state.interval, 780 .maxlen = sizeof(int), 781 .mode = 0644, 782 .proc_handler = proc_dointvec_jiffies, 783 }, 784 { 785 .procname = "printk_ratelimit_burst", 786 .data = &printk_ratelimit_state.burst, 787 .maxlen = sizeof(int), 788 .mode = 0644, 789 .proc_handler = proc_dointvec, 790 }, 791 { 792 .procname = "printk_delay", 793 .data = &printk_delay_msec, 794 .maxlen = sizeof(int), 795 .mode = 0644, 796 .proc_handler = proc_dointvec_minmax, 797 .extra1 = &zero, 798 .extra2 = &ten_thousand, 799 }, 800 { 801 .procname = "dmesg_restrict", 802 .data = &dmesg_restrict, 803 .maxlen = sizeof(int), 804 .mode = 0644, 805 .proc_handler = proc_dointvec_minmax_sysadmin, 806 .extra1 = &zero, 807 .extra2 = &one, 808 }, 809 { 810 .procname = "kptr_restrict", 811 .data = &kptr_restrict, 812 .maxlen = sizeof(int), 813 .mode = 0644, 814 .proc_handler = proc_dointvec_minmax_sysadmin, 815 .extra1 = &zero, 816 .extra2 = &two, 817 }, 818 #endif 819 { 820 .procname = "ngroups_max", 821 .data = &ngroups_max, 822 .maxlen = sizeof (int), 823 .mode = 0444, 824 .proc_handler = proc_dointvec, 825 }, 826 { 827 .procname = "cap_last_cap", 828 .data = (void *)&cap_last_cap, 829 .maxlen = sizeof(int), 830 .mode = 0444, 831 .proc_handler = proc_dointvec, 832 }, 833 #if defined(CONFIG_LOCKUP_DETECTOR) 834 { 835 .procname = "watchdog", 836 .data = &watchdog_user_enabled, 837 .maxlen = sizeof (int), 838 .mode = 0644, 839 .proc_handler = proc_watchdog, 840 .extra1 = &zero, 841 .extra2 = &one, 842 }, 843 { 844 .procname = "watchdog_thresh", 845 .data = &watchdog_thresh, 846 .maxlen = sizeof(int), 847 .mode = 0644, 848 .proc_handler = proc_watchdog_thresh, 849 .extra1 = &zero, 850 .extra2 = &sixty, 851 }, 852 { 853 .procname = "nmi_watchdog", 854 .data = &nmi_watchdog_enabled, 855 .maxlen = sizeof (int), 856 .mode = 0644, 857 .proc_handler = proc_nmi_watchdog, 858 .extra1 = &zero, 859 #if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR) 860 .extra2 = &one, 861 #else 862 .extra2 = &zero, 863 #endif 864 }, 865 { 866 .procname = "soft_watchdog", 867 .data = &soft_watchdog_enabled, 868 .maxlen = sizeof (int), 869 .mode = 0644, 870 .proc_handler = proc_soft_watchdog, 871 .extra1 = &zero, 872 .extra2 = &one, 873 }, 874 { 875 .procname = "watchdog_cpumask", 876 .data = &watchdog_cpumask_bits, 877 .maxlen = NR_CPUS, 878 .mode = 0644, 879 .proc_handler = proc_watchdog_cpumask, 880 }, 881 { 882 .procname = "softlockup_panic", 883 .data = &softlockup_panic, 884 .maxlen = sizeof(int), 885 .mode = 0644, 886 .proc_handler = proc_dointvec_minmax, 887 .extra1 = &zero, 888 .extra2 = &one, 889 }, 890 #ifdef CONFIG_SMP 891 { 892 .procname = "softlockup_all_cpu_backtrace", 893 .data = &sysctl_softlockup_all_cpu_backtrace, 894 .maxlen = sizeof(int), 895 .mode = 0644, 896 .proc_handler = proc_dointvec_minmax, 897 .extra1 = &zero, 898 .extra2 = &one, 899 }, 900 #endif /* CONFIG_SMP */ 901 #endif 902 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86) 903 { 904 .procname = "unknown_nmi_panic", 905 .data = &unknown_nmi_panic, 906 .maxlen = sizeof (int), 907 .mode = 0644, 908 .proc_handler = proc_dointvec, 909 }, 910 #endif 911 #if defined(CONFIG_X86) 912 { 913 .procname = "panic_on_unrecovered_nmi", 914 .data = &panic_on_unrecovered_nmi, 915 .maxlen = sizeof(int), 916 .mode = 0644, 917 .proc_handler = proc_dointvec, 918 }, 919 { 920 .procname = "panic_on_io_nmi", 921 .data = &panic_on_io_nmi, 922 .maxlen = sizeof(int), 923 .mode = 0644, 924 .proc_handler = proc_dointvec, 925 }, 926 #ifdef CONFIG_DEBUG_STACKOVERFLOW 927 { 928 .procname = "panic_on_stackoverflow", 929 .data = &sysctl_panic_on_stackoverflow, 930 .maxlen = sizeof(int), 931 .mode = 0644, 932 .proc_handler = proc_dointvec, 933 }, 934 #endif 935 { 936 .procname = "bootloader_type", 937 .data = &bootloader_type, 938 .maxlen = sizeof (int), 939 .mode = 0444, 940 .proc_handler = proc_dointvec, 941 }, 942 { 943 .procname = "bootloader_version", 944 .data = &bootloader_version, 945 .maxlen = sizeof (int), 946 .mode = 0444, 947 .proc_handler = proc_dointvec, 948 }, 949 { 950 .procname = "kstack_depth_to_print", 951 .data = &kstack_depth_to_print, 952 .maxlen = sizeof(int), 953 .mode = 0644, 954 .proc_handler = proc_dointvec, 955 }, 956 { 957 .procname = "io_delay_type", 958 .data = &io_delay_type, 959 .maxlen = sizeof(int), 960 .mode = 0644, 961 .proc_handler = proc_dointvec, 962 }, 963 #endif 964 #if defined(CONFIG_MMU) 965 { 966 .procname = "randomize_va_space", 967 .data = &randomize_va_space, 968 .maxlen = sizeof(int), 969 .mode = 0644, 970 .proc_handler = proc_dointvec, 971 }, 972 #endif 973 #if defined(CONFIG_S390) && defined(CONFIG_SMP) 974 { 975 .procname = "spin_retry", 976 .data = &spin_retry, 977 .maxlen = sizeof (int), 978 .mode = 0644, 979 .proc_handler = proc_dointvec, 980 }, 981 #endif 982 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86) 983 { 984 .procname = "acpi_video_flags", 985 .data = &acpi_realmode_flags, 986 .maxlen = sizeof (unsigned long), 987 .mode = 0644, 988 .proc_handler = proc_doulongvec_minmax, 989 }, 990 #endif 991 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN 992 { 993 .procname = "ignore-unaligned-usertrap", 994 .data = &no_unaligned_warning, 995 .maxlen = sizeof (int), 996 .mode = 0644, 997 .proc_handler = proc_dointvec, 998 }, 999 #endif 1000 #ifdef CONFIG_IA64 1001 { 1002 .procname = "unaligned-dump-stack", 1003 .data = &unaligned_dump_stack, 1004 .maxlen = sizeof (int), 1005 .mode = 0644, 1006 .proc_handler = proc_dointvec, 1007 }, 1008 #endif 1009 #ifdef CONFIG_DETECT_HUNG_TASK 1010 { 1011 .procname = "hung_task_panic", 1012 .data = &sysctl_hung_task_panic, 1013 .maxlen = sizeof(int), 1014 .mode = 0644, 1015 .proc_handler = proc_dointvec_minmax, 1016 .extra1 = &zero, 1017 .extra2 = &one, 1018 }, 1019 { 1020 .procname = "hung_task_check_count", 1021 .data = &sysctl_hung_task_check_count, 1022 .maxlen = sizeof(int), 1023 .mode = 0644, 1024 .proc_handler = proc_dointvec_minmax, 1025 .extra1 = &zero, 1026 }, 1027 { 1028 .procname = "hung_task_timeout_secs", 1029 .data = &sysctl_hung_task_timeout_secs, 1030 .maxlen = sizeof(unsigned long), 1031 .mode = 0644, 1032 .proc_handler = proc_dohung_task_timeout_secs, 1033 .extra2 = &hung_task_timeout_max, 1034 }, 1035 { 1036 .procname = "hung_task_warnings", 1037 .data = &sysctl_hung_task_warnings, 1038 .maxlen = sizeof(int), 1039 .mode = 0644, 1040 .proc_handler = proc_dointvec_minmax, 1041 .extra1 = &neg_one, 1042 }, 1043 #endif 1044 #ifdef CONFIG_COMPAT 1045 { 1046 .procname = "compat-log", 1047 .data = &compat_log, 1048 .maxlen = sizeof (int), 1049 .mode = 0644, 1050 .proc_handler = proc_dointvec, 1051 }, 1052 #endif 1053 #ifdef CONFIG_RT_MUTEXES 1054 { 1055 .procname = "max_lock_depth", 1056 .data = &max_lock_depth, 1057 .maxlen = sizeof(int), 1058 .mode = 0644, 1059 .proc_handler = proc_dointvec, 1060 }, 1061 #endif 1062 { 1063 .procname = "poweroff_cmd", 1064 .data = &poweroff_cmd, 1065 .maxlen = POWEROFF_CMD_PATH_LEN, 1066 .mode = 0644, 1067 .proc_handler = proc_dostring, 1068 }, 1069 #ifdef CONFIG_KEYS 1070 { 1071 .procname = "keys", 1072 .mode = 0555, 1073 .child = key_sysctls, 1074 }, 1075 #endif 1076 #ifdef CONFIG_PERF_EVENTS 1077 /* 1078 * User-space scripts rely on the existence of this file 1079 * as a feature check for perf_events being enabled. 1080 * 1081 * So it's an ABI, do not remove! 1082 */ 1083 { 1084 .procname = "perf_event_paranoid", 1085 .data = &sysctl_perf_event_paranoid, 1086 .maxlen = sizeof(sysctl_perf_event_paranoid), 1087 .mode = 0644, 1088 .proc_handler = proc_dointvec, 1089 }, 1090 { 1091 .procname = "perf_event_mlock_kb", 1092 .data = &sysctl_perf_event_mlock, 1093 .maxlen = sizeof(sysctl_perf_event_mlock), 1094 .mode = 0644, 1095 .proc_handler = proc_dointvec, 1096 }, 1097 { 1098 .procname = "perf_event_max_sample_rate", 1099 .data = &sysctl_perf_event_sample_rate, 1100 .maxlen = sizeof(sysctl_perf_event_sample_rate), 1101 .mode = 0644, 1102 .proc_handler = perf_proc_update_handler, 1103 .extra1 = &one, 1104 }, 1105 { 1106 .procname = "perf_cpu_time_max_percent", 1107 .data = &sysctl_perf_cpu_time_max_percent, 1108 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent), 1109 .mode = 0644, 1110 .proc_handler = perf_cpu_time_max_percent_handler, 1111 .extra1 = &zero, 1112 .extra2 = &one_hundred, 1113 }, 1114 #endif 1115 #ifdef CONFIG_KMEMCHECK 1116 { 1117 .procname = "kmemcheck", 1118 .data = &kmemcheck_enabled, 1119 .maxlen = sizeof(int), 1120 .mode = 0644, 1121 .proc_handler = proc_dointvec, 1122 }, 1123 #endif 1124 { 1125 .procname = "panic_on_warn", 1126 .data = &panic_on_warn, 1127 .maxlen = sizeof(int), 1128 .mode = 0644, 1129 .proc_handler = proc_dointvec_minmax, 1130 .extra1 = &zero, 1131 .extra2 = &one, 1132 }, 1133 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) 1134 { 1135 .procname = "timer_migration", 1136 .data = &sysctl_timer_migration, 1137 .maxlen = sizeof(unsigned int), 1138 .mode = 0644, 1139 .proc_handler = timer_migration_handler, 1140 }, 1141 #endif 1142 { } 1143 }; 1144 1145 static struct ctl_table vm_table[] = { 1146 { 1147 .procname = "overcommit_memory", 1148 .data = &sysctl_overcommit_memory, 1149 .maxlen = sizeof(sysctl_overcommit_memory), 1150 .mode = 0644, 1151 .proc_handler = proc_dointvec_minmax, 1152 .extra1 = &zero, 1153 .extra2 = &two, 1154 }, 1155 { 1156 .procname = "panic_on_oom", 1157 .data = &sysctl_panic_on_oom, 1158 .maxlen = sizeof(sysctl_panic_on_oom), 1159 .mode = 0644, 1160 .proc_handler = proc_dointvec_minmax, 1161 .extra1 = &zero, 1162 .extra2 = &two, 1163 }, 1164 { 1165 .procname = "oom_kill_allocating_task", 1166 .data = &sysctl_oom_kill_allocating_task, 1167 .maxlen = sizeof(sysctl_oom_kill_allocating_task), 1168 .mode = 0644, 1169 .proc_handler = proc_dointvec, 1170 }, 1171 { 1172 .procname = "oom_dump_tasks", 1173 .data = &sysctl_oom_dump_tasks, 1174 .maxlen = sizeof(sysctl_oom_dump_tasks), 1175 .mode = 0644, 1176 .proc_handler = proc_dointvec, 1177 }, 1178 { 1179 .procname = "overcommit_ratio", 1180 .data = &sysctl_overcommit_ratio, 1181 .maxlen = sizeof(sysctl_overcommit_ratio), 1182 .mode = 0644, 1183 .proc_handler = overcommit_ratio_handler, 1184 }, 1185 { 1186 .procname = "overcommit_kbytes", 1187 .data = &sysctl_overcommit_kbytes, 1188 .maxlen = sizeof(sysctl_overcommit_kbytes), 1189 .mode = 0644, 1190 .proc_handler = overcommit_kbytes_handler, 1191 }, 1192 { 1193 .procname = "page-cluster", 1194 .data = &page_cluster, 1195 .maxlen = sizeof(int), 1196 .mode = 0644, 1197 .proc_handler = proc_dointvec_minmax, 1198 .extra1 = &zero, 1199 }, 1200 { 1201 .procname = "dirty_background_ratio", 1202 .data = &dirty_background_ratio, 1203 .maxlen = sizeof(dirty_background_ratio), 1204 .mode = 0644, 1205 .proc_handler = dirty_background_ratio_handler, 1206 .extra1 = &zero, 1207 .extra2 = &one_hundred, 1208 }, 1209 { 1210 .procname = "dirty_background_bytes", 1211 .data = &dirty_background_bytes, 1212 .maxlen = sizeof(dirty_background_bytes), 1213 .mode = 0644, 1214 .proc_handler = dirty_background_bytes_handler, 1215 .extra1 = &one_ul, 1216 }, 1217 { 1218 .procname = "dirty_ratio", 1219 .data = &vm_dirty_ratio, 1220 .maxlen = sizeof(vm_dirty_ratio), 1221 .mode = 0644, 1222 .proc_handler = dirty_ratio_handler, 1223 .extra1 = &zero, 1224 .extra2 = &one_hundred, 1225 }, 1226 { 1227 .procname = "dirty_bytes", 1228 .data = &vm_dirty_bytes, 1229 .maxlen = sizeof(vm_dirty_bytes), 1230 .mode = 0644, 1231 .proc_handler = dirty_bytes_handler, 1232 .extra1 = &dirty_bytes_min, 1233 }, 1234 { 1235 .procname = "dirty_writeback_centisecs", 1236 .data = &dirty_writeback_interval, 1237 .maxlen = sizeof(dirty_writeback_interval), 1238 .mode = 0644, 1239 .proc_handler = dirty_writeback_centisecs_handler, 1240 }, 1241 { 1242 .procname = "dirty_expire_centisecs", 1243 .data = &dirty_expire_interval, 1244 .maxlen = sizeof(dirty_expire_interval), 1245 .mode = 0644, 1246 .proc_handler = proc_dointvec_minmax, 1247 .extra1 = &zero, 1248 }, 1249 { 1250 .procname = "dirtytime_expire_seconds", 1251 .data = &dirtytime_expire_interval, 1252 .maxlen = sizeof(dirty_expire_interval), 1253 .mode = 0644, 1254 .proc_handler = dirtytime_interval_handler, 1255 .extra1 = &zero, 1256 }, 1257 { 1258 .procname = "nr_pdflush_threads", 1259 .mode = 0444 /* read-only */, 1260 .proc_handler = pdflush_proc_obsolete, 1261 }, 1262 { 1263 .procname = "swappiness", 1264 .data = &vm_swappiness, 1265 .maxlen = sizeof(vm_swappiness), 1266 .mode = 0644, 1267 .proc_handler = proc_dointvec_minmax, 1268 .extra1 = &zero, 1269 .extra2 = &one_hundred, 1270 }, 1271 #ifdef CONFIG_HUGETLB_PAGE 1272 { 1273 .procname = "nr_hugepages", 1274 .data = NULL, 1275 .maxlen = sizeof(unsigned long), 1276 .mode = 0644, 1277 .proc_handler = hugetlb_sysctl_handler, 1278 }, 1279 #ifdef CONFIG_NUMA 1280 { 1281 .procname = "nr_hugepages_mempolicy", 1282 .data = NULL, 1283 .maxlen = sizeof(unsigned long), 1284 .mode = 0644, 1285 .proc_handler = &hugetlb_mempolicy_sysctl_handler, 1286 }, 1287 #endif 1288 { 1289 .procname = "hugetlb_shm_group", 1290 .data = &sysctl_hugetlb_shm_group, 1291 .maxlen = sizeof(gid_t), 1292 .mode = 0644, 1293 .proc_handler = proc_dointvec, 1294 }, 1295 { 1296 .procname = "hugepages_treat_as_movable", 1297 .data = &hugepages_treat_as_movable, 1298 .maxlen = sizeof(int), 1299 .mode = 0644, 1300 .proc_handler = proc_dointvec, 1301 }, 1302 { 1303 .procname = "nr_overcommit_hugepages", 1304 .data = NULL, 1305 .maxlen = sizeof(unsigned long), 1306 .mode = 0644, 1307 .proc_handler = hugetlb_overcommit_handler, 1308 }, 1309 #endif 1310 { 1311 .procname = "lowmem_reserve_ratio", 1312 .data = &sysctl_lowmem_reserve_ratio, 1313 .maxlen = sizeof(sysctl_lowmem_reserve_ratio), 1314 .mode = 0644, 1315 .proc_handler = lowmem_reserve_ratio_sysctl_handler, 1316 }, 1317 { 1318 .procname = "drop_caches", 1319 .data = &sysctl_drop_caches, 1320 .maxlen = sizeof(int), 1321 .mode = 0644, 1322 .proc_handler = drop_caches_sysctl_handler, 1323 .extra1 = &one, 1324 .extra2 = &four, 1325 }, 1326 #ifdef CONFIG_COMPACTION 1327 { 1328 .procname = "compact_memory", 1329 .data = &sysctl_compact_memory, 1330 .maxlen = sizeof(int), 1331 .mode = 0200, 1332 .proc_handler = sysctl_compaction_handler, 1333 }, 1334 { 1335 .procname = "extfrag_threshold", 1336 .data = &sysctl_extfrag_threshold, 1337 .maxlen = sizeof(int), 1338 .mode = 0644, 1339 .proc_handler = sysctl_extfrag_handler, 1340 .extra1 = &min_extfrag_threshold, 1341 .extra2 = &max_extfrag_threshold, 1342 }, 1343 { 1344 .procname = "compact_unevictable_allowed", 1345 .data = &sysctl_compact_unevictable_allowed, 1346 .maxlen = sizeof(int), 1347 .mode = 0644, 1348 .proc_handler = proc_dointvec, 1349 .extra1 = &zero, 1350 .extra2 = &one, 1351 }, 1352 1353 #endif /* CONFIG_COMPACTION */ 1354 { 1355 .procname = "min_free_kbytes", 1356 .data = &min_free_kbytes, 1357 .maxlen = sizeof(min_free_kbytes), 1358 .mode = 0644, 1359 .proc_handler = min_free_kbytes_sysctl_handler, 1360 .extra1 = &zero, 1361 }, 1362 { 1363 .procname = "percpu_pagelist_fraction", 1364 .data = &percpu_pagelist_fraction, 1365 .maxlen = sizeof(percpu_pagelist_fraction), 1366 .mode = 0644, 1367 .proc_handler = percpu_pagelist_fraction_sysctl_handler, 1368 .extra1 = &zero, 1369 }, 1370 #ifdef CONFIG_MMU 1371 { 1372 .procname = "max_map_count", 1373 .data = &sysctl_max_map_count, 1374 .maxlen = sizeof(sysctl_max_map_count), 1375 .mode = 0644, 1376 .proc_handler = proc_dointvec_minmax, 1377 .extra1 = &zero, 1378 }, 1379 #else 1380 { 1381 .procname = "nr_trim_pages", 1382 .data = &sysctl_nr_trim_pages, 1383 .maxlen = sizeof(sysctl_nr_trim_pages), 1384 .mode = 0644, 1385 .proc_handler = proc_dointvec_minmax, 1386 .extra1 = &zero, 1387 }, 1388 #endif 1389 { 1390 .procname = "laptop_mode", 1391 .data = &laptop_mode, 1392 .maxlen = sizeof(laptop_mode), 1393 .mode = 0644, 1394 .proc_handler = proc_dointvec_jiffies, 1395 }, 1396 { 1397 .procname = "block_dump", 1398 .data = &block_dump, 1399 .maxlen = sizeof(block_dump), 1400 .mode = 0644, 1401 .proc_handler = proc_dointvec, 1402 .extra1 = &zero, 1403 }, 1404 { 1405 .procname = "vfs_cache_pressure", 1406 .data = &sysctl_vfs_cache_pressure, 1407 .maxlen = sizeof(sysctl_vfs_cache_pressure), 1408 .mode = 0644, 1409 .proc_handler = proc_dointvec, 1410 .extra1 = &zero, 1411 }, 1412 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT 1413 { 1414 .procname = "legacy_va_layout", 1415 .data = &sysctl_legacy_va_layout, 1416 .maxlen = sizeof(sysctl_legacy_va_layout), 1417 .mode = 0644, 1418 .proc_handler = proc_dointvec, 1419 .extra1 = &zero, 1420 }, 1421 #endif 1422 #ifdef CONFIG_NUMA 1423 { 1424 .procname = "zone_reclaim_mode", 1425 .data = &zone_reclaim_mode, 1426 .maxlen = sizeof(zone_reclaim_mode), 1427 .mode = 0644, 1428 .proc_handler = proc_dointvec, 1429 .extra1 = &zero, 1430 }, 1431 { 1432 .procname = "min_unmapped_ratio", 1433 .data = &sysctl_min_unmapped_ratio, 1434 .maxlen = sizeof(sysctl_min_unmapped_ratio), 1435 .mode = 0644, 1436 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler, 1437 .extra1 = &zero, 1438 .extra2 = &one_hundred, 1439 }, 1440 { 1441 .procname = "min_slab_ratio", 1442 .data = &sysctl_min_slab_ratio, 1443 .maxlen = sizeof(sysctl_min_slab_ratio), 1444 .mode = 0644, 1445 .proc_handler = sysctl_min_slab_ratio_sysctl_handler, 1446 .extra1 = &zero, 1447 .extra2 = &one_hundred, 1448 }, 1449 #endif 1450 #ifdef CONFIG_SMP 1451 { 1452 .procname = "stat_interval", 1453 .data = &sysctl_stat_interval, 1454 .maxlen = sizeof(sysctl_stat_interval), 1455 .mode = 0644, 1456 .proc_handler = proc_dointvec_jiffies, 1457 }, 1458 #endif 1459 #ifdef CONFIG_MMU 1460 { 1461 .procname = "mmap_min_addr", 1462 .data = &dac_mmap_min_addr, 1463 .maxlen = sizeof(unsigned long), 1464 .mode = 0644, 1465 .proc_handler = mmap_min_addr_handler, 1466 }, 1467 #endif 1468 #ifdef CONFIG_NUMA 1469 { 1470 .procname = "numa_zonelist_order", 1471 .data = &numa_zonelist_order, 1472 .maxlen = NUMA_ZONELIST_ORDER_LEN, 1473 .mode = 0644, 1474 .proc_handler = numa_zonelist_order_handler, 1475 }, 1476 #endif 1477 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \ 1478 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL)) 1479 { 1480 .procname = "vdso_enabled", 1481 #ifdef CONFIG_X86_32 1482 .data = &vdso32_enabled, 1483 .maxlen = sizeof(vdso32_enabled), 1484 #else 1485 .data = &vdso_enabled, 1486 .maxlen = sizeof(vdso_enabled), 1487 #endif 1488 .mode = 0644, 1489 .proc_handler = proc_dointvec, 1490 .extra1 = &zero, 1491 }, 1492 #endif 1493 #ifdef CONFIG_HIGHMEM 1494 { 1495 .procname = "highmem_is_dirtyable", 1496 .data = &vm_highmem_is_dirtyable, 1497 .maxlen = sizeof(vm_highmem_is_dirtyable), 1498 .mode = 0644, 1499 .proc_handler = proc_dointvec_minmax, 1500 .extra1 = &zero, 1501 .extra2 = &one, 1502 }, 1503 #endif 1504 #ifdef CONFIG_MEMORY_FAILURE 1505 { 1506 .procname = "memory_failure_early_kill", 1507 .data = &sysctl_memory_failure_early_kill, 1508 .maxlen = sizeof(sysctl_memory_failure_early_kill), 1509 .mode = 0644, 1510 .proc_handler = proc_dointvec_minmax, 1511 .extra1 = &zero, 1512 .extra2 = &one, 1513 }, 1514 { 1515 .procname = "memory_failure_recovery", 1516 .data = &sysctl_memory_failure_recovery, 1517 .maxlen = sizeof(sysctl_memory_failure_recovery), 1518 .mode = 0644, 1519 .proc_handler = proc_dointvec_minmax, 1520 .extra1 = &zero, 1521 .extra2 = &one, 1522 }, 1523 #endif 1524 { 1525 .procname = "user_reserve_kbytes", 1526 .data = &sysctl_user_reserve_kbytes, 1527 .maxlen = sizeof(sysctl_user_reserve_kbytes), 1528 .mode = 0644, 1529 .proc_handler = proc_doulongvec_minmax, 1530 }, 1531 { 1532 .procname = "admin_reserve_kbytes", 1533 .data = &sysctl_admin_reserve_kbytes, 1534 .maxlen = sizeof(sysctl_admin_reserve_kbytes), 1535 .mode = 0644, 1536 .proc_handler = proc_doulongvec_minmax, 1537 }, 1538 { } 1539 }; 1540 1541 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE) 1542 static struct ctl_table binfmt_misc_table[] = { 1543 { } 1544 }; 1545 #endif 1546 1547 static struct ctl_table fs_table[] = { 1548 { 1549 .procname = "inode-nr", 1550 .data = &inodes_stat, 1551 .maxlen = 2*sizeof(long), 1552 .mode = 0444, 1553 .proc_handler = proc_nr_inodes, 1554 }, 1555 { 1556 .procname = "inode-state", 1557 .data = &inodes_stat, 1558 .maxlen = 7*sizeof(long), 1559 .mode = 0444, 1560 .proc_handler = proc_nr_inodes, 1561 }, 1562 { 1563 .procname = "file-nr", 1564 .data = &files_stat, 1565 .maxlen = sizeof(files_stat), 1566 .mode = 0444, 1567 .proc_handler = proc_nr_files, 1568 }, 1569 { 1570 .procname = "file-max", 1571 .data = &files_stat.max_files, 1572 .maxlen = sizeof(files_stat.max_files), 1573 .mode = 0644, 1574 .proc_handler = proc_doulongvec_minmax, 1575 }, 1576 { 1577 .procname = "nr_open", 1578 .data = &sysctl_nr_open, 1579 .maxlen = sizeof(int), 1580 .mode = 0644, 1581 .proc_handler = proc_dointvec_minmax, 1582 .extra1 = &sysctl_nr_open_min, 1583 .extra2 = &sysctl_nr_open_max, 1584 }, 1585 { 1586 .procname = "dentry-state", 1587 .data = &dentry_stat, 1588 .maxlen = 6*sizeof(long), 1589 .mode = 0444, 1590 .proc_handler = proc_nr_dentry, 1591 }, 1592 { 1593 .procname = "overflowuid", 1594 .data = &fs_overflowuid, 1595 .maxlen = sizeof(int), 1596 .mode = 0644, 1597 .proc_handler = proc_dointvec_minmax, 1598 .extra1 = &minolduid, 1599 .extra2 = &maxolduid, 1600 }, 1601 { 1602 .procname = "overflowgid", 1603 .data = &fs_overflowgid, 1604 .maxlen = sizeof(int), 1605 .mode = 0644, 1606 .proc_handler = proc_dointvec_minmax, 1607 .extra1 = &minolduid, 1608 .extra2 = &maxolduid, 1609 }, 1610 #ifdef CONFIG_FILE_LOCKING 1611 { 1612 .procname = "leases-enable", 1613 .data = &leases_enable, 1614 .maxlen = sizeof(int), 1615 .mode = 0644, 1616 .proc_handler = proc_dointvec, 1617 }, 1618 #endif 1619 #ifdef CONFIG_DNOTIFY 1620 { 1621 .procname = "dir-notify-enable", 1622 .data = &dir_notify_enable, 1623 .maxlen = sizeof(int), 1624 .mode = 0644, 1625 .proc_handler = proc_dointvec, 1626 }, 1627 #endif 1628 #ifdef CONFIG_MMU 1629 #ifdef CONFIG_FILE_LOCKING 1630 { 1631 .procname = "lease-break-time", 1632 .data = &lease_break_time, 1633 .maxlen = sizeof(int), 1634 .mode = 0644, 1635 .proc_handler = proc_dointvec, 1636 }, 1637 #endif 1638 #ifdef CONFIG_AIO 1639 { 1640 .procname = "aio-nr", 1641 .data = &aio_nr, 1642 .maxlen = sizeof(aio_nr), 1643 .mode = 0444, 1644 .proc_handler = proc_doulongvec_minmax, 1645 }, 1646 { 1647 .procname = "aio-max-nr", 1648 .data = &aio_max_nr, 1649 .maxlen = sizeof(aio_max_nr), 1650 .mode = 0644, 1651 .proc_handler = proc_doulongvec_minmax, 1652 }, 1653 #endif /* CONFIG_AIO */ 1654 #ifdef CONFIG_INOTIFY_USER 1655 { 1656 .procname = "inotify", 1657 .mode = 0555, 1658 .child = inotify_table, 1659 }, 1660 #endif 1661 #ifdef CONFIG_EPOLL 1662 { 1663 .procname = "epoll", 1664 .mode = 0555, 1665 .child = epoll_table, 1666 }, 1667 #endif 1668 #endif 1669 { 1670 .procname = "protected_symlinks", 1671 .data = &sysctl_protected_symlinks, 1672 .maxlen = sizeof(int), 1673 .mode = 0600, 1674 .proc_handler = proc_dointvec_minmax, 1675 .extra1 = &zero, 1676 .extra2 = &one, 1677 }, 1678 { 1679 .procname = "protected_hardlinks", 1680 .data = &sysctl_protected_hardlinks, 1681 .maxlen = sizeof(int), 1682 .mode = 0600, 1683 .proc_handler = proc_dointvec_minmax, 1684 .extra1 = &zero, 1685 .extra2 = &one, 1686 }, 1687 { 1688 .procname = "suid_dumpable", 1689 .data = &suid_dumpable, 1690 .maxlen = sizeof(int), 1691 .mode = 0644, 1692 .proc_handler = proc_dointvec_minmax_coredump, 1693 .extra1 = &zero, 1694 .extra2 = &two, 1695 }, 1696 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE) 1697 { 1698 .procname = "binfmt_misc", 1699 .mode = 0555, 1700 .child = binfmt_misc_table, 1701 }, 1702 #endif 1703 { 1704 .procname = "pipe-max-size", 1705 .data = &pipe_max_size, 1706 .maxlen = sizeof(int), 1707 .mode = 0644, 1708 .proc_handler = &pipe_proc_fn, 1709 .extra1 = &pipe_min_size, 1710 }, 1711 { } 1712 }; 1713 1714 static struct ctl_table debug_table[] = { 1715 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE 1716 { 1717 .procname = "exception-trace", 1718 .data = &show_unhandled_signals, 1719 .maxlen = sizeof(int), 1720 .mode = 0644, 1721 .proc_handler = proc_dointvec 1722 }, 1723 #endif 1724 #if defined(CONFIG_OPTPROBES) 1725 { 1726 .procname = "kprobes-optimization", 1727 .data = &sysctl_kprobes_optimization, 1728 .maxlen = sizeof(int), 1729 .mode = 0644, 1730 .proc_handler = proc_kprobes_optimization_handler, 1731 .extra1 = &zero, 1732 .extra2 = &one, 1733 }, 1734 #endif 1735 { } 1736 }; 1737 1738 static struct ctl_table dev_table[] = { 1739 { } 1740 }; 1741 1742 int __init sysctl_init(void) 1743 { 1744 struct ctl_table_header *hdr; 1745 1746 hdr = register_sysctl_table(sysctl_base_table); 1747 kmemleak_not_leak(hdr); 1748 return 0; 1749 } 1750 1751 #endif /* CONFIG_SYSCTL */ 1752 1753 /* 1754 * /proc/sys support 1755 */ 1756 1757 #ifdef CONFIG_PROC_SYSCTL 1758 1759 static int _proc_do_string(char *data, int maxlen, int write, 1760 char __user *buffer, 1761 size_t *lenp, loff_t *ppos) 1762 { 1763 size_t len; 1764 char __user *p; 1765 char c; 1766 1767 if (!data || !maxlen || !*lenp) { 1768 *lenp = 0; 1769 return 0; 1770 } 1771 1772 if (write) { 1773 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) { 1774 /* Only continue writes not past the end of buffer. */ 1775 len = strlen(data); 1776 if (len > maxlen - 1) 1777 len = maxlen - 1; 1778 1779 if (*ppos > len) 1780 return 0; 1781 len = *ppos; 1782 } else { 1783 /* Start writing from beginning of buffer. */ 1784 len = 0; 1785 } 1786 1787 *ppos += *lenp; 1788 p = buffer; 1789 while ((p - buffer) < *lenp && len < maxlen - 1) { 1790 if (get_user(c, p++)) 1791 return -EFAULT; 1792 if (c == 0 || c == '\n') 1793 break; 1794 data[len++] = c; 1795 } 1796 data[len] = 0; 1797 } else { 1798 len = strlen(data); 1799 if (len > maxlen) 1800 len = maxlen; 1801 1802 if (*ppos > len) { 1803 *lenp = 0; 1804 return 0; 1805 } 1806 1807 data += *ppos; 1808 len -= *ppos; 1809 1810 if (len > *lenp) 1811 len = *lenp; 1812 if (len) 1813 if (copy_to_user(buffer, data, len)) 1814 return -EFAULT; 1815 if (len < *lenp) { 1816 if (put_user('\n', buffer + len)) 1817 return -EFAULT; 1818 len++; 1819 } 1820 *lenp = len; 1821 *ppos += len; 1822 } 1823 return 0; 1824 } 1825 1826 static void warn_sysctl_write(struct ctl_table *table) 1827 { 1828 pr_warn_once("%s wrote to %s when file position was not 0!\n" 1829 "This will not be supported in the future. To silence this\n" 1830 "warning, set kernel.sysctl_writes_strict = -1\n", 1831 current->comm, table->procname); 1832 } 1833 1834 /** 1835 * proc_dostring - read a string sysctl 1836 * @table: the sysctl table 1837 * @write: %TRUE if this is a write to the sysctl file 1838 * @buffer: the user buffer 1839 * @lenp: the size of the user buffer 1840 * @ppos: file position 1841 * 1842 * Reads/writes a string from/to the user buffer. If the kernel 1843 * buffer provided is not large enough to hold the string, the 1844 * string is truncated. The copied string is %NULL-terminated. 1845 * If the string is being read by the user process, it is copied 1846 * and a newline '\n' is added. It is truncated if the buffer is 1847 * not large enough. 1848 * 1849 * Returns 0 on success. 1850 */ 1851 int proc_dostring(struct ctl_table *table, int write, 1852 void __user *buffer, size_t *lenp, loff_t *ppos) 1853 { 1854 if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN) 1855 warn_sysctl_write(table); 1856 1857 return _proc_do_string((char *)(table->data), table->maxlen, write, 1858 (char __user *)buffer, lenp, ppos); 1859 } 1860 1861 static size_t proc_skip_spaces(char **buf) 1862 { 1863 size_t ret; 1864 char *tmp = skip_spaces(*buf); 1865 ret = tmp - *buf; 1866 *buf = tmp; 1867 return ret; 1868 } 1869 1870 static void proc_skip_char(char **buf, size_t *size, const char v) 1871 { 1872 while (*size) { 1873 if (**buf != v) 1874 break; 1875 (*size)--; 1876 (*buf)++; 1877 } 1878 } 1879 1880 #define TMPBUFLEN 22 1881 /** 1882 * proc_get_long - reads an ASCII formatted integer from a user buffer 1883 * 1884 * @buf: a kernel buffer 1885 * @size: size of the kernel buffer 1886 * @val: this is where the number will be stored 1887 * @neg: set to %TRUE if number is negative 1888 * @perm_tr: a vector which contains the allowed trailers 1889 * @perm_tr_len: size of the perm_tr vector 1890 * @tr: pointer to store the trailer character 1891 * 1892 * In case of success %0 is returned and @buf and @size are updated with 1893 * the amount of bytes read. If @tr is non-NULL and a trailing 1894 * character exists (size is non-zero after returning from this 1895 * function), @tr is updated with the trailing character. 1896 */ 1897 static int proc_get_long(char **buf, size_t *size, 1898 unsigned long *val, bool *neg, 1899 const char *perm_tr, unsigned perm_tr_len, char *tr) 1900 { 1901 int len; 1902 char *p, tmp[TMPBUFLEN]; 1903 1904 if (!*size) 1905 return -EINVAL; 1906 1907 len = *size; 1908 if (len > TMPBUFLEN - 1) 1909 len = TMPBUFLEN - 1; 1910 1911 memcpy(tmp, *buf, len); 1912 1913 tmp[len] = 0; 1914 p = tmp; 1915 if (*p == '-' && *size > 1) { 1916 *neg = true; 1917 p++; 1918 } else 1919 *neg = false; 1920 if (!isdigit(*p)) 1921 return -EINVAL; 1922 1923 *val = simple_strtoul(p, &p, 0); 1924 1925 len = p - tmp; 1926 1927 /* We don't know if the next char is whitespace thus we may accept 1928 * invalid integers (e.g. 1234...a) or two integers instead of one 1929 * (e.g. 123...1). So lets not allow such large numbers. */ 1930 if (len == TMPBUFLEN - 1) 1931 return -EINVAL; 1932 1933 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len)) 1934 return -EINVAL; 1935 1936 if (tr && (len < *size)) 1937 *tr = *p; 1938 1939 *buf += len; 1940 *size -= len; 1941 1942 return 0; 1943 } 1944 1945 /** 1946 * proc_put_long - converts an integer to a decimal ASCII formatted string 1947 * 1948 * @buf: the user buffer 1949 * @size: the size of the user buffer 1950 * @val: the integer to be converted 1951 * @neg: sign of the number, %TRUE for negative 1952 * 1953 * In case of success %0 is returned and @buf and @size are updated with 1954 * the amount of bytes written. 1955 */ 1956 static int proc_put_long(void __user **buf, size_t *size, unsigned long val, 1957 bool neg) 1958 { 1959 int len; 1960 char tmp[TMPBUFLEN], *p = tmp; 1961 1962 sprintf(p, "%s%lu", neg ? "-" : "", val); 1963 len = strlen(tmp); 1964 if (len > *size) 1965 len = *size; 1966 if (copy_to_user(*buf, tmp, len)) 1967 return -EFAULT; 1968 *size -= len; 1969 *buf += len; 1970 return 0; 1971 } 1972 #undef TMPBUFLEN 1973 1974 static int proc_put_char(void __user **buf, size_t *size, char c) 1975 { 1976 if (*size) { 1977 char __user **buffer = (char __user **)buf; 1978 if (put_user(c, *buffer)) 1979 return -EFAULT; 1980 (*size)--, (*buffer)++; 1981 *buf = *buffer; 1982 } 1983 return 0; 1984 } 1985 1986 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp, 1987 int *valp, 1988 int write, void *data) 1989 { 1990 if (write) { 1991 if (*negp) { 1992 if (*lvalp > (unsigned long) INT_MAX + 1) 1993 return -EINVAL; 1994 *valp = -*lvalp; 1995 } else { 1996 if (*lvalp > (unsigned long) INT_MAX) 1997 return -EINVAL; 1998 *valp = *lvalp; 1999 } 2000 } else { 2001 int val = *valp; 2002 if (val < 0) { 2003 *negp = true; 2004 *lvalp = (unsigned long)-val; 2005 } else { 2006 *negp = false; 2007 *lvalp = (unsigned long)val; 2008 } 2009 } 2010 return 0; 2011 } 2012 2013 static const char proc_wspace_sep[] = { ' ', '\t', '\n' }; 2014 2015 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table, 2016 int write, void __user *buffer, 2017 size_t *lenp, loff_t *ppos, 2018 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 2019 int write, void *data), 2020 void *data) 2021 { 2022 int *i, vleft, first = 1, err = 0; 2023 unsigned long page = 0; 2024 size_t left; 2025 char *kbuf; 2026 2027 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) { 2028 *lenp = 0; 2029 return 0; 2030 } 2031 2032 i = (int *) tbl_data; 2033 vleft = table->maxlen / sizeof(*i); 2034 left = *lenp; 2035 2036 if (!conv) 2037 conv = do_proc_dointvec_conv; 2038 2039 if (write) { 2040 if (*ppos) { 2041 switch (sysctl_writes_strict) { 2042 case SYSCTL_WRITES_STRICT: 2043 goto out; 2044 case SYSCTL_WRITES_WARN: 2045 warn_sysctl_write(table); 2046 break; 2047 default: 2048 break; 2049 } 2050 } 2051 2052 if (left > PAGE_SIZE - 1) 2053 left = PAGE_SIZE - 1; 2054 page = __get_free_page(GFP_TEMPORARY); 2055 kbuf = (char *) page; 2056 if (!kbuf) 2057 return -ENOMEM; 2058 if (copy_from_user(kbuf, buffer, left)) { 2059 err = -EFAULT; 2060 goto free; 2061 } 2062 kbuf[left] = 0; 2063 } 2064 2065 for (; left && vleft--; i++, first=0) { 2066 unsigned long lval; 2067 bool neg; 2068 2069 if (write) { 2070 left -= proc_skip_spaces(&kbuf); 2071 2072 if (!left) 2073 break; 2074 err = proc_get_long(&kbuf, &left, &lval, &neg, 2075 proc_wspace_sep, 2076 sizeof(proc_wspace_sep), NULL); 2077 if (err) 2078 break; 2079 if (conv(&neg, &lval, i, 1, data)) { 2080 err = -EINVAL; 2081 break; 2082 } 2083 } else { 2084 if (conv(&neg, &lval, i, 0, data)) { 2085 err = -EINVAL; 2086 break; 2087 } 2088 if (!first) 2089 err = proc_put_char(&buffer, &left, '\t'); 2090 if (err) 2091 break; 2092 err = proc_put_long(&buffer, &left, lval, neg); 2093 if (err) 2094 break; 2095 } 2096 } 2097 2098 if (!write && !first && left && !err) 2099 err = proc_put_char(&buffer, &left, '\n'); 2100 if (write && !err && left) 2101 left -= proc_skip_spaces(&kbuf); 2102 free: 2103 if (write) { 2104 free_page(page); 2105 if (first) 2106 return err ? : -EINVAL; 2107 } 2108 *lenp -= left; 2109 out: 2110 *ppos += *lenp; 2111 return err; 2112 } 2113 2114 static int do_proc_dointvec(struct ctl_table *table, int write, 2115 void __user *buffer, size_t *lenp, loff_t *ppos, 2116 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 2117 int write, void *data), 2118 void *data) 2119 { 2120 return __do_proc_dointvec(table->data, table, write, 2121 buffer, lenp, ppos, conv, data); 2122 } 2123 2124 /** 2125 * proc_dointvec - read a vector of integers 2126 * @table: the sysctl table 2127 * @write: %TRUE if this is a write to the sysctl file 2128 * @buffer: the user buffer 2129 * @lenp: the size of the user buffer 2130 * @ppos: file position 2131 * 2132 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2133 * values from/to the user buffer, treated as an ASCII string. 2134 * 2135 * Returns 0 on success. 2136 */ 2137 int proc_dointvec(struct ctl_table *table, int write, 2138 void __user *buffer, size_t *lenp, loff_t *ppos) 2139 { 2140 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2141 NULL,NULL); 2142 } 2143 2144 /* 2145 * Taint values can only be increased 2146 * This means we can safely use a temporary. 2147 */ 2148 static int proc_taint(struct ctl_table *table, int write, 2149 void __user *buffer, size_t *lenp, loff_t *ppos) 2150 { 2151 struct ctl_table t; 2152 unsigned long tmptaint = get_taint(); 2153 int err; 2154 2155 if (write && !capable(CAP_SYS_ADMIN)) 2156 return -EPERM; 2157 2158 t = *table; 2159 t.data = &tmptaint; 2160 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos); 2161 if (err < 0) 2162 return err; 2163 2164 if (write) { 2165 /* 2166 * Poor man's atomic or. Not worth adding a primitive 2167 * to everyone's atomic.h for this 2168 */ 2169 int i; 2170 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) { 2171 if ((tmptaint >> i) & 1) 2172 add_taint(i, LOCKDEP_STILL_OK); 2173 } 2174 } 2175 2176 return err; 2177 } 2178 2179 #ifdef CONFIG_PRINTK 2180 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write, 2181 void __user *buffer, size_t *lenp, loff_t *ppos) 2182 { 2183 if (write && !capable(CAP_SYS_ADMIN)) 2184 return -EPERM; 2185 2186 return proc_dointvec_minmax(table, write, buffer, lenp, ppos); 2187 } 2188 #endif 2189 2190 struct do_proc_dointvec_minmax_conv_param { 2191 int *min; 2192 int *max; 2193 }; 2194 2195 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp, 2196 int *valp, 2197 int write, void *data) 2198 { 2199 struct do_proc_dointvec_minmax_conv_param *param = data; 2200 if (write) { 2201 int val = *negp ? -*lvalp : *lvalp; 2202 if ((param->min && *param->min > val) || 2203 (param->max && *param->max < val)) 2204 return -EINVAL; 2205 *valp = val; 2206 } else { 2207 int val = *valp; 2208 if (val < 0) { 2209 *negp = true; 2210 *lvalp = (unsigned long)-val; 2211 } else { 2212 *negp = false; 2213 *lvalp = (unsigned long)val; 2214 } 2215 } 2216 return 0; 2217 } 2218 2219 /** 2220 * proc_dointvec_minmax - read a vector of integers with min/max values 2221 * @table: the sysctl table 2222 * @write: %TRUE if this is a write to the sysctl file 2223 * @buffer: the user buffer 2224 * @lenp: the size of the user buffer 2225 * @ppos: file position 2226 * 2227 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2228 * values from/to the user buffer, treated as an ASCII string. 2229 * 2230 * This routine will ensure the values are within the range specified by 2231 * table->extra1 (min) and table->extra2 (max). 2232 * 2233 * Returns 0 on success. 2234 */ 2235 int proc_dointvec_minmax(struct ctl_table *table, int write, 2236 void __user *buffer, size_t *lenp, loff_t *ppos) 2237 { 2238 struct do_proc_dointvec_minmax_conv_param param = { 2239 .min = (int *) table->extra1, 2240 .max = (int *) table->extra2, 2241 }; 2242 return do_proc_dointvec(table, write, buffer, lenp, ppos, 2243 do_proc_dointvec_minmax_conv, ¶m); 2244 } 2245 2246 static void validate_coredump_safety(void) 2247 { 2248 #ifdef CONFIG_COREDUMP 2249 if (suid_dumpable == SUID_DUMP_ROOT && 2250 core_pattern[0] != '/' && core_pattern[0] != '|') { 2251 printk(KERN_WARNING "Unsafe core_pattern used with "\ 2252 "suid_dumpable=2. Pipe handler or fully qualified "\ 2253 "core dump path required.\n"); 2254 } 2255 #endif 2256 } 2257 2258 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write, 2259 void __user *buffer, size_t *lenp, loff_t *ppos) 2260 { 2261 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos); 2262 if (!error) 2263 validate_coredump_safety(); 2264 return error; 2265 } 2266 2267 #ifdef CONFIG_COREDUMP 2268 static int proc_dostring_coredump(struct ctl_table *table, int write, 2269 void __user *buffer, size_t *lenp, loff_t *ppos) 2270 { 2271 int error = proc_dostring(table, write, buffer, lenp, ppos); 2272 if (!error) 2273 validate_coredump_safety(); 2274 return error; 2275 } 2276 #endif 2277 2278 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write, 2279 void __user *buffer, 2280 size_t *lenp, loff_t *ppos, 2281 unsigned long convmul, 2282 unsigned long convdiv) 2283 { 2284 unsigned long *i, *min, *max; 2285 int vleft, first = 1, err = 0; 2286 unsigned long page = 0; 2287 size_t left; 2288 char *kbuf; 2289 2290 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) { 2291 *lenp = 0; 2292 return 0; 2293 } 2294 2295 i = (unsigned long *) data; 2296 min = (unsigned long *) table->extra1; 2297 max = (unsigned long *) table->extra2; 2298 vleft = table->maxlen / sizeof(unsigned long); 2299 left = *lenp; 2300 2301 if (write) { 2302 if (*ppos) { 2303 switch (sysctl_writes_strict) { 2304 case SYSCTL_WRITES_STRICT: 2305 goto out; 2306 case SYSCTL_WRITES_WARN: 2307 warn_sysctl_write(table); 2308 break; 2309 default: 2310 break; 2311 } 2312 } 2313 2314 if (left > PAGE_SIZE - 1) 2315 left = PAGE_SIZE - 1; 2316 page = __get_free_page(GFP_TEMPORARY); 2317 kbuf = (char *) page; 2318 if (!kbuf) 2319 return -ENOMEM; 2320 if (copy_from_user(kbuf, buffer, left)) { 2321 err = -EFAULT; 2322 goto free; 2323 } 2324 kbuf[left] = 0; 2325 } 2326 2327 for (; left && vleft--; i++, first = 0) { 2328 unsigned long val; 2329 2330 if (write) { 2331 bool neg; 2332 2333 left -= proc_skip_spaces(&kbuf); 2334 2335 err = proc_get_long(&kbuf, &left, &val, &neg, 2336 proc_wspace_sep, 2337 sizeof(proc_wspace_sep), NULL); 2338 if (err) 2339 break; 2340 if (neg) 2341 continue; 2342 if ((min && val < *min) || (max && val > *max)) 2343 continue; 2344 *i = val; 2345 } else { 2346 val = convdiv * (*i) / convmul; 2347 if (!first) { 2348 err = proc_put_char(&buffer, &left, '\t'); 2349 if (err) 2350 break; 2351 } 2352 err = proc_put_long(&buffer, &left, val, false); 2353 if (err) 2354 break; 2355 } 2356 } 2357 2358 if (!write && !first && left && !err) 2359 err = proc_put_char(&buffer, &left, '\n'); 2360 if (write && !err) 2361 left -= proc_skip_spaces(&kbuf); 2362 free: 2363 if (write) { 2364 free_page(page); 2365 if (first) 2366 return err ? : -EINVAL; 2367 } 2368 *lenp -= left; 2369 out: 2370 *ppos += *lenp; 2371 return err; 2372 } 2373 2374 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write, 2375 void __user *buffer, 2376 size_t *lenp, loff_t *ppos, 2377 unsigned long convmul, 2378 unsigned long convdiv) 2379 { 2380 return __do_proc_doulongvec_minmax(table->data, table, write, 2381 buffer, lenp, ppos, convmul, convdiv); 2382 } 2383 2384 /** 2385 * proc_doulongvec_minmax - read a vector of long integers with min/max values 2386 * @table: the sysctl table 2387 * @write: %TRUE if this is a write to the sysctl file 2388 * @buffer: the user buffer 2389 * @lenp: the size of the user buffer 2390 * @ppos: file position 2391 * 2392 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 2393 * values from/to the user buffer, treated as an ASCII string. 2394 * 2395 * This routine will ensure the values are within the range specified by 2396 * table->extra1 (min) and table->extra2 (max). 2397 * 2398 * Returns 0 on success. 2399 */ 2400 int proc_doulongvec_minmax(struct ctl_table *table, int write, 2401 void __user *buffer, size_t *lenp, loff_t *ppos) 2402 { 2403 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l); 2404 } 2405 2406 /** 2407 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values 2408 * @table: the sysctl table 2409 * @write: %TRUE if this is a write to the sysctl file 2410 * @buffer: the user buffer 2411 * @lenp: the size of the user buffer 2412 * @ppos: file position 2413 * 2414 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 2415 * values from/to the user buffer, treated as an ASCII string. The values 2416 * are treated as milliseconds, and converted to jiffies when they are stored. 2417 * 2418 * This routine will ensure the values are within the range specified by 2419 * table->extra1 (min) and table->extra2 (max). 2420 * 2421 * Returns 0 on success. 2422 */ 2423 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 2424 void __user *buffer, 2425 size_t *lenp, loff_t *ppos) 2426 { 2427 return do_proc_doulongvec_minmax(table, write, buffer, 2428 lenp, ppos, HZ, 1000l); 2429 } 2430 2431 2432 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp, 2433 int *valp, 2434 int write, void *data) 2435 { 2436 if (write) { 2437 if (*lvalp > LONG_MAX / HZ) 2438 return 1; 2439 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ); 2440 } else { 2441 int val = *valp; 2442 unsigned long lval; 2443 if (val < 0) { 2444 *negp = true; 2445 lval = (unsigned long)-val; 2446 } else { 2447 *negp = false; 2448 lval = (unsigned long)val; 2449 } 2450 *lvalp = lval / HZ; 2451 } 2452 return 0; 2453 } 2454 2455 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp, 2456 int *valp, 2457 int write, void *data) 2458 { 2459 if (write) { 2460 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ) 2461 return 1; 2462 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp); 2463 } else { 2464 int val = *valp; 2465 unsigned long lval; 2466 if (val < 0) { 2467 *negp = true; 2468 lval = (unsigned long)-val; 2469 } else { 2470 *negp = false; 2471 lval = (unsigned long)val; 2472 } 2473 *lvalp = jiffies_to_clock_t(lval); 2474 } 2475 return 0; 2476 } 2477 2478 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp, 2479 int *valp, 2480 int write, void *data) 2481 { 2482 if (write) { 2483 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp); 2484 2485 if (jif > INT_MAX) 2486 return 1; 2487 *valp = (int)jif; 2488 } else { 2489 int val = *valp; 2490 unsigned long lval; 2491 if (val < 0) { 2492 *negp = true; 2493 lval = (unsigned long)-val; 2494 } else { 2495 *negp = false; 2496 lval = (unsigned long)val; 2497 } 2498 *lvalp = jiffies_to_msecs(lval); 2499 } 2500 return 0; 2501 } 2502 2503 /** 2504 * proc_dointvec_jiffies - read a vector of integers as seconds 2505 * @table: the sysctl table 2506 * @write: %TRUE if this is a write to the sysctl file 2507 * @buffer: the user buffer 2508 * @lenp: the size of the user buffer 2509 * @ppos: file position 2510 * 2511 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2512 * values from/to the user buffer, treated as an ASCII string. 2513 * The values read are assumed to be in seconds, and are converted into 2514 * jiffies. 2515 * 2516 * Returns 0 on success. 2517 */ 2518 int proc_dointvec_jiffies(struct ctl_table *table, int write, 2519 void __user *buffer, size_t *lenp, loff_t *ppos) 2520 { 2521 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2522 do_proc_dointvec_jiffies_conv,NULL); 2523 } 2524 2525 /** 2526 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds 2527 * @table: the sysctl table 2528 * @write: %TRUE if this is a write to the sysctl file 2529 * @buffer: the user buffer 2530 * @lenp: the size of the user buffer 2531 * @ppos: pointer to the file position 2532 * 2533 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2534 * values from/to the user buffer, treated as an ASCII string. 2535 * The values read are assumed to be in 1/USER_HZ seconds, and 2536 * are converted into jiffies. 2537 * 2538 * Returns 0 on success. 2539 */ 2540 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 2541 void __user *buffer, size_t *lenp, loff_t *ppos) 2542 { 2543 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2544 do_proc_dointvec_userhz_jiffies_conv,NULL); 2545 } 2546 2547 /** 2548 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds 2549 * @table: the sysctl table 2550 * @write: %TRUE if this is a write to the sysctl file 2551 * @buffer: the user buffer 2552 * @lenp: the size of the user buffer 2553 * @ppos: file position 2554 * @ppos: the current position in the file 2555 * 2556 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2557 * values from/to the user buffer, treated as an ASCII string. 2558 * The values read are assumed to be in 1/1000 seconds, and 2559 * are converted into jiffies. 2560 * 2561 * Returns 0 on success. 2562 */ 2563 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, 2564 void __user *buffer, size_t *lenp, loff_t *ppos) 2565 { 2566 return do_proc_dointvec(table, write, buffer, lenp, ppos, 2567 do_proc_dointvec_ms_jiffies_conv, NULL); 2568 } 2569 2570 static int proc_do_cad_pid(struct ctl_table *table, int write, 2571 void __user *buffer, size_t *lenp, loff_t *ppos) 2572 { 2573 struct pid *new_pid; 2574 pid_t tmp; 2575 int r; 2576 2577 tmp = pid_vnr(cad_pid); 2578 2579 r = __do_proc_dointvec(&tmp, table, write, buffer, 2580 lenp, ppos, NULL, NULL); 2581 if (r || !write) 2582 return r; 2583 2584 new_pid = find_get_pid(tmp); 2585 if (!new_pid) 2586 return -ESRCH; 2587 2588 put_pid(xchg(&cad_pid, new_pid)); 2589 return 0; 2590 } 2591 2592 /** 2593 * proc_do_large_bitmap - read/write from/to a large bitmap 2594 * @table: the sysctl table 2595 * @write: %TRUE if this is a write to the sysctl file 2596 * @buffer: the user buffer 2597 * @lenp: the size of the user buffer 2598 * @ppos: file position 2599 * 2600 * The bitmap is stored at table->data and the bitmap length (in bits) 2601 * in table->maxlen. 2602 * 2603 * We use a range comma separated format (e.g. 1,3-4,10-10) so that 2604 * large bitmaps may be represented in a compact manner. Writing into 2605 * the file will clear the bitmap then update it with the given input. 2606 * 2607 * Returns 0 on success. 2608 */ 2609 int proc_do_large_bitmap(struct ctl_table *table, int write, 2610 void __user *buffer, size_t *lenp, loff_t *ppos) 2611 { 2612 int err = 0; 2613 bool first = 1; 2614 size_t left = *lenp; 2615 unsigned long bitmap_len = table->maxlen; 2616 unsigned long *bitmap = *(unsigned long **) table->data; 2617 unsigned long *tmp_bitmap = NULL; 2618 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c; 2619 2620 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) { 2621 *lenp = 0; 2622 return 0; 2623 } 2624 2625 if (write) { 2626 unsigned long page = 0; 2627 char *kbuf; 2628 2629 if (left > PAGE_SIZE - 1) 2630 left = PAGE_SIZE - 1; 2631 2632 page = __get_free_page(GFP_TEMPORARY); 2633 kbuf = (char *) page; 2634 if (!kbuf) 2635 return -ENOMEM; 2636 if (copy_from_user(kbuf, buffer, left)) { 2637 free_page(page); 2638 return -EFAULT; 2639 } 2640 kbuf[left] = 0; 2641 2642 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long), 2643 GFP_KERNEL); 2644 if (!tmp_bitmap) { 2645 free_page(page); 2646 return -ENOMEM; 2647 } 2648 proc_skip_char(&kbuf, &left, '\n'); 2649 while (!err && left) { 2650 unsigned long val_a, val_b; 2651 bool neg; 2652 2653 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a, 2654 sizeof(tr_a), &c); 2655 if (err) 2656 break; 2657 if (val_a >= bitmap_len || neg) { 2658 err = -EINVAL; 2659 break; 2660 } 2661 2662 val_b = val_a; 2663 if (left) { 2664 kbuf++; 2665 left--; 2666 } 2667 2668 if (c == '-') { 2669 err = proc_get_long(&kbuf, &left, &val_b, 2670 &neg, tr_b, sizeof(tr_b), 2671 &c); 2672 if (err) 2673 break; 2674 if (val_b >= bitmap_len || neg || 2675 val_a > val_b) { 2676 err = -EINVAL; 2677 break; 2678 } 2679 if (left) { 2680 kbuf++; 2681 left--; 2682 } 2683 } 2684 2685 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1); 2686 first = 0; 2687 proc_skip_char(&kbuf, &left, '\n'); 2688 } 2689 free_page(page); 2690 } else { 2691 unsigned long bit_a, bit_b = 0; 2692 2693 while (left) { 2694 bit_a = find_next_bit(bitmap, bitmap_len, bit_b); 2695 if (bit_a >= bitmap_len) 2696 break; 2697 bit_b = find_next_zero_bit(bitmap, bitmap_len, 2698 bit_a + 1) - 1; 2699 2700 if (!first) { 2701 err = proc_put_char(&buffer, &left, ','); 2702 if (err) 2703 break; 2704 } 2705 err = proc_put_long(&buffer, &left, bit_a, false); 2706 if (err) 2707 break; 2708 if (bit_a != bit_b) { 2709 err = proc_put_char(&buffer, &left, '-'); 2710 if (err) 2711 break; 2712 err = proc_put_long(&buffer, &left, bit_b, false); 2713 if (err) 2714 break; 2715 } 2716 2717 first = 0; bit_b++; 2718 } 2719 if (!err) 2720 err = proc_put_char(&buffer, &left, '\n'); 2721 } 2722 2723 if (!err) { 2724 if (write) { 2725 if (*ppos) 2726 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len); 2727 else 2728 bitmap_copy(bitmap, tmp_bitmap, bitmap_len); 2729 } 2730 kfree(tmp_bitmap); 2731 *lenp -= left; 2732 *ppos += *lenp; 2733 return 0; 2734 } else { 2735 kfree(tmp_bitmap); 2736 return err; 2737 } 2738 } 2739 2740 #else /* CONFIG_PROC_SYSCTL */ 2741 2742 int proc_dostring(struct ctl_table *table, int write, 2743 void __user *buffer, size_t *lenp, loff_t *ppos) 2744 { 2745 return -ENOSYS; 2746 } 2747 2748 int proc_dointvec(struct ctl_table *table, int write, 2749 void __user *buffer, size_t *lenp, loff_t *ppos) 2750 { 2751 return -ENOSYS; 2752 } 2753 2754 int proc_dointvec_minmax(struct ctl_table *table, int write, 2755 void __user *buffer, size_t *lenp, loff_t *ppos) 2756 { 2757 return -ENOSYS; 2758 } 2759 2760 int proc_dointvec_jiffies(struct ctl_table *table, int write, 2761 void __user *buffer, size_t *lenp, loff_t *ppos) 2762 { 2763 return -ENOSYS; 2764 } 2765 2766 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 2767 void __user *buffer, size_t *lenp, loff_t *ppos) 2768 { 2769 return -ENOSYS; 2770 } 2771 2772 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, 2773 void __user *buffer, size_t *lenp, loff_t *ppos) 2774 { 2775 return -ENOSYS; 2776 } 2777 2778 int proc_doulongvec_minmax(struct ctl_table *table, int write, 2779 void __user *buffer, size_t *lenp, loff_t *ppos) 2780 { 2781 return -ENOSYS; 2782 } 2783 2784 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 2785 void __user *buffer, 2786 size_t *lenp, loff_t *ppos) 2787 { 2788 return -ENOSYS; 2789 } 2790 2791 2792 #endif /* CONFIG_PROC_SYSCTL */ 2793 2794 /* 2795 * No sense putting this after each symbol definition, twice, 2796 * exception granted :-) 2797 */ 2798 EXPORT_SYMBOL(proc_dointvec); 2799 EXPORT_SYMBOL(proc_dointvec_jiffies); 2800 EXPORT_SYMBOL(proc_dointvec_minmax); 2801 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies); 2802 EXPORT_SYMBOL(proc_dointvec_ms_jiffies); 2803 EXPORT_SYMBOL(proc_dostring); 2804 EXPORT_SYMBOL(proc_doulongvec_minmax); 2805 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax); 2806