1 /* 2 * sysctl.c: General linux system control interface 3 * 4 * Begun 24 March 1995, Stephen Tweedie 5 * Added /proc support, Dec 1995 6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas. 7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver. 8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver. 9 * Dynamic registration fixes, Stephen Tweedie. 10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn. 11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris 12 * Horn. 13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer. 14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer. 15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill 16 * Wendling. 17 * The list_for_each() macro wasn't appropriate for the sysctl loop. 18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling 19 */ 20 21 #include <linux/module.h> 22 #include <linux/mm.h> 23 #include <linux/swap.h> 24 #include <linux/slab.h> 25 #include <linux/sysctl.h> 26 #include <linux/bitmap.h> 27 #include <linux/signal.h> 28 #include <linux/printk.h> 29 #include <linux/proc_fs.h> 30 #include <linux/security.h> 31 #include <linux/ctype.h> 32 #include <linux/kmemcheck.h> 33 #include <linux/kmemleak.h> 34 #include <linux/fs.h> 35 #include <linux/init.h> 36 #include <linux/kernel.h> 37 #include <linux/kobject.h> 38 #include <linux/net.h> 39 #include <linux/sysrq.h> 40 #include <linux/highuid.h> 41 #include <linux/writeback.h> 42 #include <linux/ratelimit.h> 43 #include <linux/compaction.h> 44 #include <linux/hugetlb.h> 45 #include <linux/initrd.h> 46 #include <linux/key.h> 47 #include <linux/times.h> 48 #include <linux/limits.h> 49 #include <linux/dcache.h> 50 #include <linux/dnotify.h> 51 #include <linux/syscalls.h> 52 #include <linux/vmstat.h> 53 #include <linux/nfs_fs.h> 54 #include <linux/acpi.h> 55 #include <linux/reboot.h> 56 #include <linux/ftrace.h> 57 #include <linux/perf_event.h> 58 #include <linux/kprobes.h> 59 #include <linux/pipe_fs_i.h> 60 #include <linux/oom.h> 61 #include <linux/kmod.h> 62 #include <linux/capability.h> 63 #include <linux/binfmts.h> 64 #include <linux/sched/sysctl.h> 65 66 #include <asm/uaccess.h> 67 #include <asm/processor.h> 68 69 #ifdef CONFIG_X86 70 #include <asm/nmi.h> 71 #include <asm/stacktrace.h> 72 #include <asm/io.h> 73 #endif 74 #ifdef CONFIG_SPARC 75 #include <asm/setup.h> 76 #endif 77 #ifdef CONFIG_BSD_PROCESS_ACCT 78 #include <linux/acct.h> 79 #endif 80 #ifdef CONFIG_RT_MUTEXES 81 #include <linux/rtmutex.h> 82 #endif 83 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT) 84 #include <linux/lockdep.h> 85 #endif 86 #ifdef CONFIG_CHR_DEV_SG 87 #include <scsi/sg.h> 88 #endif 89 90 #ifdef CONFIG_LOCKUP_DETECTOR 91 #include <linux/nmi.h> 92 #endif 93 94 95 #if defined(CONFIG_SYSCTL) 96 97 /* External variables not in a header file. */ 98 extern int sysctl_overcommit_memory; 99 extern int sysctl_overcommit_ratio; 100 extern int max_threads; 101 extern int suid_dumpable; 102 #ifdef CONFIG_COREDUMP 103 extern int core_uses_pid; 104 extern char core_pattern[]; 105 extern unsigned int core_pipe_limit; 106 #endif 107 extern int pid_max; 108 extern int pid_max_min, pid_max_max; 109 extern int percpu_pagelist_fraction; 110 extern int compat_log; 111 extern int latencytop_enabled; 112 extern int sysctl_nr_open_min, sysctl_nr_open_max; 113 #ifndef CONFIG_MMU 114 extern int sysctl_nr_trim_pages; 115 #endif 116 #ifdef CONFIG_BLOCK 117 extern int blk_iopoll_enabled; 118 #endif 119 120 /* Constants used for minimum and maximum */ 121 #ifdef CONFIG_LOCKUP_DETECTOR 122 static int sixty = 60; 123 #endif 124 125 static int zero; 126 static int __maybe_unused one = 1; 127 static int __maybe_unused two = 2; 128 static int __maybe_unused three = 3; 129 static unsigned long one_ul = 1; 130 static int one_hundred = 100; 131 #ifdef CONFIG_PRINTK 132 static int ten_thousand = 10000; 133 #endif 134 135 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */ 136 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE; 137 138 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */ 139 static int maxolduid = 65535; 140 static int minolduid; 141 static int min_percpu_pagelist_fract = 8; 142 143 static int ngroups_max = NGROUPS_MAX; 144 static const int cap_last_cap = CAP_LAST_CAP; 145 146 #ifdef CONFIG_INOTIFY_USER 147 #include <linux/inotify.h> 148 #endif 149 #ifdef CONFIG_SPARC 150 #endif 151 152 #ifdef CONFIG_SPARC64 153 extern int sysctl_tsb_ratio; 154 #endif 155 156 #ifdef __hppa__ 157 extern int pwrsw_enabled; 158 #endif 159 160 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW 161 extern int unaligned_enabled; 162 #endif 163 164 #ifdef CONFIG_IA64 165 extern int unaligned_dump_stack; 166 #endif 167 168 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN 169 extern int no_unaligned_warning; 170 #endif 171 172 #ifdef CONFIG_PROC_SYSCTL 173 static int proc_do_cad_pid(struct ctl_table *table, int write, 174 void __user *buffer, size_t *lenp, loff_t *ppos); 175 static int proc_taint(struct ctl_table *table, int write, 176 void __user *buffer, size_t *lenp, loff_t *ppos); 177 #endif 178 179 #ifdef CONFIG_PRINTK 180 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write, 181 void __user *buffer, size_t *lenp, loff_t *ppos); 182 #endif 183 184 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write, 185 void __user *buffer, size_t *lenp, loff_t *ppos); 186 #ifdef CONFIG_COREDUMP 187 static int proc_dostring_coredump(struct ctl_table *table, int write, 188 void __user *buffer, size_t *lenp, loff_t *ppos); 189 #endif 190 191 #ifdef CONFIG_MAGIC_SYSRQ 192 /* Note: sysrq code uses it's own private copy */ 193 static int __sysrq_enabled = SYSRQ_DEFAULT_ENABLE; 194 195 static int sysrq_sysctl_handler(ctl_table *table, int write, 196 void __user *buffer, size_t *lenp, 197 loff_t *ppos) 198 { 199 int error; 200 201 error = proc_dointvec(table, write, buffer, lenp, ppos); 202 if (error) 203 return error; 204 205 if (write) 206 sysrq_toggle_support(__sysrq_enabled); 207 208 return 0; 209 } 210 211 #endif 212 213 static struct ctl_table kern_table[]; 214 static struct ctl_table vm_table[]; 215 static struct ctl_table fs_table[]; 216 static struct ctl_table debug_table[]; 217 static struct ctl_table dev_table[]; 218 extern struct ctl_table random_table[]; 219 #ifdef CONFIG_EPOLL 220 extern struct ctl_table epoll_table[]; 221 #endif 222 223 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT 224 int sysctl_legacy_va_layout; 225 #endif 226 227 /* The default sysctl tables: */ 228 229 static struct ctl_table sysctl_base_table[] = { 230 { 231 .procname = "kernel", 232 .mode = 0555, 233 .child = kern_table, 234 }, 235 { 236 .procname = "vm", 237 .mode = 0555, 238 .child = vm_table, 239 }, 240 { 241 .procname = "fs", 242 .mode = 0555, 243 .child = fs_table, 244 }, 245 { 246 .procname = "debug", 247 .mode = 0555, 248 .child = debug_table, 249 }, 250 { 251 .procname = "dev", 252 .mode = 0555, 253 .child = dev_table, 254 }, 255 { } 256 }; 257 258 #ifdef CONFIG_SCHED_DEBUG 259 static int min_sched_granularity_ns = 100000; /* 100 usecs */ 260 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */ 261 static int min_wakeup_granularity_ns; /* 0 usecs */ 262 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */ 263 #ifdef CONFIG_SMP 264 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE; 265 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1; 266 #endif /* CONFIG_SMP */ 267 #endif /* CONFIG_SCHED_DEBUG */ 268 269 #ifdef CONFIG_COMPACTION 270 static int min_extfrag_threshold; 271 static int max_extfrag_threshold = 1000; 272 #endif 273 274 static struct ctl_table kern_table[] = { 275 { 276 .procname = "sched_child_runs_first", 277 .data = &sysctl_sched_child_runs_first, 278 .maxlen = sizeof(unsigned int), 279 .mode = 0644, 280 .proc_handler = proc_dointvec, 281 }, 282 #ifdef CONFIG_SCHED_DEBUG 283 { 284 .procname = "sched_min_granularity_ns", 285 .data = &sysctl_sched_min_granularity, 286 .maxlen = sizeof(unsigned int), 287 .mode = 0644, 288 .proc_handler = sched_proc_update_handler, 289 .extra1 = &min_sched_granularity_ns, 290 .extra2 = &max_sched_granularity_ns, 291 }, 292 { 293 .procname = "sched_latency_ns", 294 .data = &sysctl_sched_latency, 295 .maxlen = sizeof(unsigned int), 296 .mode = 0644, 297 .proc_handler = sched_proc_update_handler, 298 .extra1 = &min_sched_granularity_ns, 299 .extra2 = &max_sched_granularity_ns, 300 }, 301 { 302 .procname = "sched_wakeup_granularity_ns", 303 .data = &sysctl_sched_wakeup_granularity, 304 .maxlen = sizeof(unsigned int), 305 .mode = 0644, 306 .proc_handler = sched_proc_update_handler, 307 .extra1 = &min_wakeup_granularity_ns, 308 .extra2 = &max_wakeup_granularity_ns, 309 }, 310 #ifdef CONFIG_SMP 311 { 312 .procname = "sched_tunable_scaling", 313 .data = &sysctl_sched_tunable_scaling, 314 .maxlen = sizeof(enum sched_tunable_scaling), 315 .mode = 0644, 316 .proc_handler = sched_proc_update_handler, 317 .extra1 = &min_sched_tunable_scaling, 318 .extra2 = &max_sched_tunable_scaling, 319 }, 320 { 321 .procname = "sched_migration_cost_ns", 322 .data = &sysctl_sched_migration_cost, 323 .maxlen = sizeof(unsigned int), 324 .mode = 0644, 325 .proc_handler = proc_dointvec, 326 }, 327 { 328 .procname = "sched_nr_migrate", 329 .data = &sysctl_sched_nr_migrate, 330 .maxlen = sizeof(unsigned int), 331 .mode = 0644, 332 .proc_handler = proc_dointvec, 333 }, 334 { 335 .procname = "sched_time_avg_ms", 336 .data = &sysctl_sched_time_avg, 337 .maxlen = sizeof(unsigned int), 338 .mode = 0644, 339 .proc_handler = proc_dointvec, 340 }, 341 { 342 .procname = "sched_shares_window_ns", 343 .data = &sysctl_sched_shares_window, 344 .maxlen = sizeof(unsigned int), 345 .mode = 0644, 346 .proc_handler = proc_dointvec, 347 }, 348 { 349 .procname = "timer_migration", 350 .data = &sysctl_timer_migration, 351 .maxlen = sizeof(unsigned int), 352 .mode = 0644, 353 .proc_handler = proc_dointvec_minmax, 354 .extra1 = &zero, 355 .extra2 = &one, 356 }, 357 #endif /* CONFIG_SMP */ 358 #ifdef CONFIG_NUMA_BALANCING 359 { 360 .procname = "numa_balancing_scan_delay_ms", 361 .data = &sysctl_numa_balancing_scan_delay, 362 .maxlen = sizeof(unsigned int), 363 .mode = 0644, 364 .proc_handler = proc_dointvec, 365 }, 366 { 367 .procname = "numa_balancing_scan_period_min_ms", 368 .data = &sysctl_numa_balancing_scan_period_min, 369 .maxlen = sizeof(unsigned int), 370 .mode = 0644, 371 .proc_handler = proc_dointvec, 372 }, 373 { 374 .procname = "numa_balancing_scan_period_reset", 375 .data = &sysctl_numa_balancing_scan_period_reset, 376 .maxlen = sizeof(unsigned int), 377 .mode = 0644, 378 .proc_handler = proc_dointvec, 379 }, 380 { 381 .procname = "numa_balancing_scan_period_max_ms", 382 .data = &sysctl_numa_balancing_scan_period_max, 383 .maxlen = sizeof(unsigned int), 384 .mode = 0644, 385 .proc_handler = proc_dointvec, 386 }, 387 { 388 .procname = "numa_balancing_scan_size_mb", 389 .data = &sysctl_numa_balancing_scan_size, 390 .maxlen = sizeof(unsigned int), 391 .mode = 0644, 392 .proc_handler = proc_dointvec, 393 }, 394 #endif /* CONFIG_NUMA_BALANCING */ 395 #endif /* CONFIG_SCHED_DEBUG */ 396 { 397 .procname = "sched_rt_period_us", 398 .data = &sysctl_sched_rt_period, 399 .maxlen = sizeof(unsigned int), 400 .mode = 0644, 401 .proc_handler = sched_rt_handler, 402 }, 403 { 404 .procname = "sched_rt_runtime_us", 405 .data = &sysctl_sched_rt_runtime, 406 .maxlen = sizeof(int), 407 .mode = 0644, 408 .proc_handler = sched_rt_handler, 409 }, 410 { 411 .procname = "sched_rr_timeslice_ms", 412 .data = &sched_rr_timeslice, 413 .maxlen = sizeof(int), 414 .mode = 0644, 415 .proc_handler = sched_rr_handler, 416 }, 417 #ifdef CONFIG_SCHED_AUTOGROUP 418 { 419 .procname = "sched_autogroup_enabled", 420 .data = &sysctl_sched_autogroup_enabled, 421 .maxlen = sizeof(unsigned int), 422 .mode = 0644, 423 .proc_handler = proc_dointvec_minmax, 424 .extra1 = &zero, 425 .extra2 = &one, 426 }, 427 #endif 428 #ifdef CONFIG_CFS_BANDWIDTH 429 { 430 .procname = "sched_cfs_bandwidth_slice_us", 431 .data = &sysctl_sched_cfs_bandwidth_slice, 432 .maxlen = sizeof(unsigned int), 433 .mode = 0644, 434 .proc_handler = proc_dointvec_minmax, 435 .extra1 = &one, 436 }, 437 #endif 438 #ifdef CONFIG_PROVE_LOCKING 439 { 440 .procname = "prove_locking", 441 .data = &prove_locking, 442 .maxlen = sizeof(int), 443 .mode = 0644, 444 .proc_handler = proc_dointvec, 445 }, 446 #endif 447 #ifdef CONFIG_LOCK_STAT 448 { 449 .procname = "lock_stat", 450 .data = &lock_stat, 451 .maxlen = sizeof(int), 452 .mode = 0644, 453 .proc_handler = proc_dointvec, 454 }, 455 #endif 456 { 457 .procname = "panic", 458 .data = &panic_timeout, 459 .maxlen = sizeof(int), 460 .mode = 0644, 461 .proc_handler = proc_dointvec, 462 }, 463 #ifdef CONFIG_COREDUMP 464 { 465 .procname = "core_uses_pid", 466 .data = &core_uses_pid, 467 .maxlen = sizeof(int), 468 .mode = 0644, 469 .proc_handler = proc_dointvec, 470 }, 471 { 472 .procname = "core_pattern", 473 .data = core_pattern, 474 .maxlen = CORENAME_MAX_SIZE, 475 .mode = 0644, 476 .proc_handler = proc_dostring_coredump, 477 }, 478 { 479 .procname = "core_pipe_limit", 480 .data = &core_pipe_limit, 481 .maxlen = sizeof(unsigned int), 482 .mode = 0644, 483 .proc_handler = proc_dointvec, 484 }, 485 #endif 486 #ifdef CONFIG_PROC_SYSCTL 487 { 488 .procname = "tainted", 489 .maxlen = sizeof(long), 490 .mode = 0644, 491 .proc_handler = proc_taint, 492 }, 493 #endif 494 #ifdef CONFIG_LATENCYTOP 495 { 496 .procname = "latencytop", 497 .data = &latencytop_enabled, 498 .maxlen = sizeof(int), 499 .mode = 0644, 500 .proc_handler = proc_dointvec, 501 }, 502 #endif 503 #ifdef CONFIG_BLK_DEV_INITRD 504 { 505 .procname = "real-root-dev", 506 .data = &real_root_dev, 507 .maxlen = sizeof(int), 508 .mode = 0644, 509 .proc_handler = proc_dointvec, 510 }, 511 #endif 512 { 513 .procname = "print-fatal-signals", 514 .data = &print_fatal_signals, 515 .maxlen = sizeof(int), 516 .mode = 0644, 517 .proc_handler = proc_dointvec, 518 }, 519 #ifdef CONFIG_SPARC 520 { 521 .procname = "reboot-cmd", 522 .data = reboot_command, 523 .maxlen = 256, 524 .mode = 0644, 525 .proc_handler = proc_dostring, 526 }, 527 { 528 .procname = "stop-a", 529 .data = &stop_a_enabled, 530 .maxlen = sizeof (int), 531 .mode = 0644, 532 .proc_handler = proc_dointvec, 533 }, 534 { 535 .procname = "scons-poweroff", 536 .data = &scons_pwroff, 537 .maxlen = sizeof (int), 538 .mode = 0644, 539 .proc_handler = proc_dointvec, 540 }, 541 #endif 542 #ifdef CONFIG_SPARC64 543 { 544 .procname = "tsb-ratio", 545 .data = &sysctl_tsb_ratio, 546 .maxlen = sizeof (int), 547 .mode = 0644, 548 .proc_handler = proc_dointvec, 549 }, 550 #endif 551 #ifdef __hppa__ 552 { 553 .procname = "soft-power", 554 .data = &pwrsw_enabled, 555 .maxlen = sizeof (int), 556 .mode = 0644, 557 .proc_handler = proc_dointvec, 558 }, 559 #endif 560 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW 561 { 562 .procname = "unaligned-trap", 563 .data = &unaligned_enabled, 564 .maxlen = sizeof (int), 565 .mode = 0644, 566 .proc_handler = proc_dointvec, 567 }, 568 #endif 569 { 570 .procname = "ctrl-alt-del", 571 .data = &C_A_D, 572 .maxlen = sizeof(int), 573 .mode = 0644, 574 .proc_handler = proc_dointvec, 575 }, 576 #ifdef CONFIG_FUNCTION_TRACER 577 { 578 .procname = "ftrace_enabled", 579 .data = &ftrace_enabled, 580 .maxlen = sizeof(int), 581 .mode = 0644, 582 .proc_handler = ftrace_enable_sysctl, 583 }, 584 #endif 585 #ifdef CONFIG_STACK_TRACER 586 { 587 .procname = "stack_tracer_enabled", 588 .data = &stack_tracer_enabled, 589 .maxlen = sizeof(int), 590 .mode = 0644, 591 .proc_handler = stack_trace_sysctl, 592 }, 593 #endif 594 #ifdef CONFIG_TRACING 595 { 596 .procname = "ftrace_dump_on_oops", 597 .data = &ftrace_dump_on_oops, 598 .maxlen = sizeof(int), 599 .mode = 0644, 600 .proc_handler = proc_dointvec, 601 }, 602 #endif 603 #ifdef CONFIG_MODULES 604 { 605 .procname = "modprobe", 606 .data = &modprobe_path, 607 .maxlen = KMOD_PATH_LEN, 608 .mode = 0644, 609 .proc_handler = proc_dostring, 610 }, 611 { 612 .procname = "modules_disabled", 613 .data = &modules_disabled, 614 .maxlen = sizeof(int), 615 .mode = 0644, 616 /* only handle a transition from default "0" to "1" */ 617 .proc_handler = proc_dointvec_minmax, 618 .extra1 = &one, 619 .extra2 = &one, 620 }, 621 #endif 622 623 { 624 .procname = "hotplug", 625 .data = &uevent_helper, 626 .maxlen = UEVENT_HELPER_PATH_LEN, 627 .mode = 0644, 628 .proc_handler = proc_dostring, 629 }, 630 631 #ifdef CONFIG_CHR_DEV_SG 632 { 633 .procname = "sg-big-buff", 634 .data = &sg_big_buff, 635 .maxlen = sizeof (int), 636 .mode = 0444, 637 .proc_handler = proc_dointvec, 638 }, 639 #endif 640 #ifdef CONFIG_BSD_PROCESS_ACCT 641 { 642 .procname = "acct", 643 .data = &acct_parm, 644 .maxlen = 3*sizeof(int), 645 .mode = 0644, 646 .proc_handler = proc_dointvec, 647 }, 648 #endif 649 #ifdef CONFIG_MAGIC_SYSRQ 650 { 651 .procname = "sysrq", 652 .data = &__sysrq_enabled, 653 .maxlen = sizeof (int), 654 .mode = 0644, 655 .proc_handler = sysrq_sysctl_handler, 656 }, 657 #endif 658 #ifdef CONFIG_PROC_SYSCTL 659 { 660 .procname = "cad_pid", 661 .data = NULL, 662 .maxlen = sizeof (int), 663 .mode = 0600, 664 .proc_handler = proc_do_cad_pid, 665 }, 666 #endif 667 { 668 .procname = "threads-max", 669 .data = &max_threads, 670 .maxlen = sizeof(int), 671 .mode = 0644, 672 .proc_handler = proc_dointvec, 673 }, 674 { 675 .procname = "random", 676 .mode = 0555, 677 .child = random_table, 678 }, 679 { 680 .procname = "usermodehelper", 681 .mode = 0555, 682 .child = usermodehelper_table, 683 }, 684 { 685 .procname = "overflowuid", 686 .data = &overflowuid, 687 .maxlen = sizeof(int), 688 .mode = 0644, 689 .proc_handler = proc_dointvec_minmax, 690 .extra1 = &minolduid, 691 .extra2 = &maxolduid, 692 }, 693 { 694 .procname = "overflowgid", 695 .data = &overflowgid, 696 .maxlen = sizeof(int), 697 .mode = 0644, 698 .proc_handler = proc_dointvec_minmax, 699 .extra1 = &minolduid, 700 .extra2 = &maxolduid, 701 }, 702 #ifdef CONFIG_S390 703 #ifdef CONFIG_MATHEMU 704 { 705 .procname = "ieee_emulation_warnings", 706 .data = &sysctl_ieee_emulation_warnings, 707 .maxlen = sizeof(int), 708 .mode = 0644, 709 .proc_handler = proc_dointvec, 710 }, 711 #endif 712 { 713 .procname = "userprocess_debug", 714 .data = &show_unhandled_signals, 715 .maxlen = sizeof(int), 716 .mode = 0644, 717 .proc_handler = proc_dointvec, 718 }, 719 #endif 720 { 721 .procname = "pid_max", 722 .data = &pid_max, 723 .maxlen = sizeof (int), 724 .mode = 0644, 725 .proc_handler = proc_dointvec_minmax, 726 .extra1 = &pid_max_min, 727 .extra2 = &pid_max_max, 728 }, 729 { 730 .procname = "panic_on_oops", 731 .data = &panic_on_oops, 732 .maxlen = sizeof(int), 733 .mode = 0644, 734 .proc_handler = proc_dointvec, 735 }, 736 #if defined CONFIG_PRINTK 737 { 738 .procname = "printk", 739 .data = &console_loglevel, 740 .maxlen = 4*sizeof(int), 741 .mode = 0644, 742 .proc_handler = proc_dointvec, 743 }, 744 { 745 .procname = "printk_ratelimit", 746 .data = &printk_ratelimit_state.interval, 747 .maxlen = sizeof(int), 748 .mode = 0644, 749 .proc_handler = proc_dointvec_jiffies, 750 }, 751 { 752 .procname = "printk_ratelimit_burst", 753 .data = &printk_ratelimit_state.burst, 754 .maxlen = sizeof(int), 755 .mode = 0644, 756 .proc_handler = proc_dointvec, 757 }, 758 { 759 .procname = "printk_delay", 760 .data = &printk_delay_msec, 761 .maxlen = sizeof(int), 762 .mode = 0644, 763 .proc_handler = proc_dointvec_minmax, 764 .extra1 = &zero, 765 .extra2 = &ten_thousand, 766 }, 767 { 768 .procname = "dmesg_restrict", 769 .data = &dmesg_restrict, 770 .maxlen = sizeof(int), 771 .mode = 0644, 772 .proc_handler = proc_dointvec_minmax_sysadmin, 773 .extra1 = &zero, 774 .extra2 = &one, 775 }, 776 { 777 .procname = "kptr_restrict", 778 .data = &kptr_restrict, 779 .maxlen = sizeof(int), 780 .mode = 0644, 781 .proc_handler = proc_dointvec_minmax_sysadmin, 782 .extra1 = &zero, 783 .extra2 = &two, 784 }, 785 #endif 786 { 787 .procname = "ngroups_max", 788 .data = &ngroups_max, 789 .maxlen = sizeof (int), 790 .mode = 0444, 791 .proc_handler = proc_dointvec, 792 }, 793 { 794 .procname = "cap_last_cap", 795 .data = (void *)&cap_last_cap, 796 .maxlen = sizeof(int), 797 .mode = 0444, 798 .proc_handler = proc_dointvec, 799 }, 800 #if defined(CONFIG_LOCKUP_DETECTOR) 801 { 802 .procname = "watchdog", 803 .data = &watchdog_enabled, 804 .maxlen = sizeof (int), 805 .mode = 0644, 806 .proc_handler = proc_dowatchdog, 807 .extra1 = &zero, 808 .extra2 = &one, 809 }, 810 { 811 .procname = "watchdog_thresh", 812 .data = &watchdog_thresh, 813 .maxlen = sizeof(int), 814 .mode = 0644, 815 .proc_handler = proc_dowatchdog, 816 .extra1 = &zero, 817 .extra2 = &sixty, 818 }, 819 { 820 .procname = "softlockup_panic", 821 .data = &softlockup_panic, 822 .maxlen = sizeof(int), 823 .mode = 0644, 824 .proc_handler = proc_dointvec_minmax, 825 .extra1 = &zero, 826 .extra2 = &one, 827 }, 828 { 829 .procname = "nmi_watchdog", 830 .data = &watchdog_enabled, 831 .maxlen = sizeof (int), 832 .mode = 0644, 833 .proc_handler = proc_dowatchdog, 834 .extra1 = &zero, 835 .extra2 = &one, 836 }, 837 #endif 838 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86) 839 { 840 .procname = "unknown_nmi_panic", 841 .data = &unknown_nmi_panic, 842 .maxlen = sizeof (int), 843 .mode = 0644, 844 .proc_handler = proc_dointvec, 845 }, 846 #endif 847 #if defined(CONFIG_X86) 848 { 849 .procname = "panic_on_unrecovered_nmi", 850 .data = &panic_on_unrecovered_nmi, 851 .maxlen = sizeof(int), 852 .mode = 0644, 853 .proc_handler = proc_dointvec, 854 }, 855 { 856 .procname = "panic_on_io_nmi", 857 .data = &panic_on_io_nmi, 858 .maxlen = sizeof(int), 859 .mode = 0644, 860 .proc_handler = proc_dointvec, 861 }, 862 #ifdef CONFIG_DEBUG_STACKOVERFLOW 863 { 864 .procname = "panic_on_stackoverflow", 865 .data = &sysctl_panic_on_stackoverflow, 866 .maxlen = sizeof(int), 867 .mode = 0644, 868 .proc_handler = proc_dointvec, 869 }, 870 #endif 871 { 872 .procname = "bootloader_type", 873 .data = &bootloader_type, 874 .maxlen = sizeof (int), 875 .mode = 0444, 876 .proc_handler = proc_dointvec, 877 }, 878 { 879 .procname = "bootloader_version", 880 .data = &bootloader_version, 881 .maxlen = sizeof (int), 882 .mode = 0444, 883 .proc_handler = proc_dointvec, 884 }, 885 { 886 .procname = "kstack_depth_to_print", 887 .data = &kstack_depth_to_print, 888 .maxlen = sizeof(int), 889 .mode = 0644, 890 .proc_handler = proc_dointvec, 891 }, 892 { 893 .procname = "io_delay_type", 894 .data = &io_delay_type, 895 .maxlen = sizeof(int), 896 .mode = 0644, 897 .proc_handler = proc_dointvec, 898 }, 899 #endif 900 #if defined(CONFIG_MMU) 901 { 902 .procname = "randomize_va_space", 903 .data = &randomize_va_space, 904 .maxlen = sizeof(int), 905 .mode = 0644, 906 .proc_handler = proc_dointvec, 907 }, 908 #endif 909 #if defined(CONFIG_S390) && defined(CONFIG_SMP) 910 { 911 .procname = "spin_retry", 912 .data = &spin_retry, 913 .maxlen = sizeof (int), 914 .mode = 0644, 915 .proc_handler = proc_dointvec, 916 }, 917 #endif 918 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86) 919 { 920 .procname = "acpi_video_flags", 921 .data = &acpi_realmode_flags, 922 .maxlen = sizeof (unsigned long), 923 .mode = 0644, 924 .proc_handler = proc_doulongvec_minmax, 925 }, 926 #endif 927 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN 928 { 929 .procname = "ignore-unaligned-usertrap", 930 .data = &no_unaligned_warning, 931 .maxlen = sizeof (int), 932 .mode = 0644, 933 .proc_handler = proc_dointvec, 934 }, 935 #endif 936 #ifdef CONFIG_IA64 937 { 938 .procname = "unaligned-dump-stack", 939 .data = &unaligned_dump_stack, 940 .maxlen = sizeof (int), 941 .mode = 0644, 942 .proc_handler = proc_dointvec, 943 }, 944 #endif 945 #ifdef CONFIG_DETECT_HUNG_TASK 946 { 947 .procname = "hung_task_panic", 948 .data = &sysctl_hung_task_panic, 949 .maxlen = sizeof(int), 950 .mode = 0644, 951 .proc_handler = proc_dointvec_minmax, 952 .extra1 = &zero, 953 .extra2 = &one, 954 }, 955 { 956 .procname = "hung_task_check_count", 957 .data = &sysctl_hung_task_check_count, 958 .maxlen = sizeof(unsigned long), 959 .mode = 0644, 960 .proc_handler = proc_doulongvec_minmax, 961 }, 962 { 963 .procname = "hung_task_timeout_secs", 964 .data = &sysctl_hung_task_timeout_secs, 965 .maxlen = sizeof(unsigned long), 966 .mode = 0644, 967 .proc_handler = proc_dohung_task_timeout_secs, 968 }, 969 { 970 .procname = "hung_task_warnings", 971 .data = &sysctl_hung_task_warnings, 972 .maxlen = sizeof(unsigned long), 973 .mode = 0644, 974 .proc_handler = proc_doulongvec_minmax, 975 }, 976 #endif 977 #ifdef CONFIG_COMPAT 978 { 979 .procname = "compat-log", 980 .data = &compat_log, 981 .maxlen = sizeof (int), 982 .mode = 0644, 983 .proc_handler = proc_dointvec, 984 }, 985 #endif 986 #ifdef CONFIG_RT_MUTEXES 987 { 988 .procname = "max_lock_depth", 989 .data = &max_lock_depth, 990 .maxlen = sizeof(int), 991 .mode = 0644, 992 .proc_handler = proc_dointvec, 993 }, 994 #endif 995 { 996 .procname = "poweroff_cmd", 997 .data = &poweroff_cmd, 998 .maxlen = POWEROFF_CMD_PATH_LEN, 999 .mode = 0644, 1000 .proc_handler = proc_dostring, 1001 }, 1002 #ifdef CONFIG_KEYS 1003 { 1004 .procname = "keys", 1005 .mode = 0555, 1006 .child = key_sysctls, 1007 }, 1008 #endif 1009 #ifdef CONFIG_RCU_TORTURE_TEST 1010 { 1011 .procname = "rcutorture_runnable", 1012 .data = &rcutorture_runnable, 1013 .maxlen = sizeof(int), 1014 .mode = 0644, 1015 .proc_handler = proc_dointvec, 1016 }, 1017 #endif 1018 #ifdef CONFIG_PERF_EVENTS 1019 /* 1020 * User-space scripts rely on the existence of this file 1021 * as a feature check for perf_events being enabled. 1022 * 1023 * So it's an ABI, do not remove! 1024 */ 1025 { 1026 .procname = "perf_event_paranoid", 1027 .data = &sysctl_perf_event_paranoid, 1028 .maxlen = sizeof(sysctl_perf_event_paranoid), 1029 .mode = 0644, 1030 .proc_handler = proc_dointvec, 1031 }, 1032 { 1033 .procname = "perf_event_mlock_kb", 1034 .data = &sysctl_perf_event_mlock, 1035 .maxlen = sizeof(sysctl_perf_event_mlock), 1036 .mode = 0644, 1037 .proc_handler = proc_dointvec, 1038 }, 1039 { 1040 .procname = "perf_event_max_sample_rate", 1041 .data = &sysctl_perf_event_sample_rate, 1042 .maxlen = sizeof(sysctl_perf_event_sample_rate), 1043 .mode = 0644, 1044 .proc_handler = perf_proc_update_handler, 1045 }, 1046 { 1047 .procname = "perf_cpu_time_max_percent", 1048 .data = &sysctl_perf_cpu_time_max_percent, 1049 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent), 1050 .mode = 0644, 1051 .proc_handler = perf_cpu_time_max_percent_handler, 1052 .extra1 = &zero, 1053 .extra2 = &one_hundred, 1054 }, 1055 #endif 1056 #ifdef CONFIG_KMEMCHECK 1057 { 1058 .procname = "kmemcheck", 1059 .data = &kmemcheck_enabled, 1060 .maxlen = sizeof(int), 1061 .mode = 0644, 1062 .proc_handler = proc_dointvec, 1063 }, 1064 #endif 1065 #ifdef CONFIG_BLOCK 1066 { 1067 .procname = "blk_iopoll", 1068 .data = &blk_iopoll_enabled, 1069 .maxlen = sizeof(int), 1070 .mode = 0644, 1071 .proc_handler = proc_dointvec, 1072 }, 1073 #endif 1074 { } 1075 }; 1076 1077 static struct ctl_table vm_table[] = { 1078 { 1079 .procname = "overcommit_memory", 1080 .data = &sysctl_overcommit_memory, 1081 .maxlen = sizeof(sysctl_overcommit_memory), 1082 .mode = 0644, 1083 .proc_handler = proc_dointvec_minmax, 1084 .extra1 = &zero, 1085 .extra2 = &two, 1086 }, 1087 { 1088 .procname = "panic_on_oom", 1089 .data = &sysctl_panic_on_oom, 1090 .maxlen = sizeof(sysctl_panic_on_oom), 1091 .mode = 0644, 1092 .proc_handler = proc_dointvec_minmax, 1093 .extra1 = &zero, 1094 .extra2 = &two, 1095 }, 1096 { 1097 .procname = "oom_kill_allocating_task", 1098 .data = &sysctl_oom_kill_allocating_task, 1099 .maxlen = sizeof(sysctl_oom_kill_allocating_task), 1100 .mode = 0644, 1101 .proc_handler = proc_dointvec, 1102 }, 1103 { 1104 .procname = "oom_dump_tasks", 1105 .data = &sysctl_oom_dump_tasks, 1106 .maxlen = sizeof(sysctl_oom_dump_tasks), 1107 .mode = 0644, 1108 .proc_handler = proc_dointvec, 1109 }, 1110 { 1111 .procname = "overcommit_ratio", 1112 .data = &sysctl_overcommit_ratio, 1113 .maxlen = sizeof(sysctl_overcommit_ratio), 1114 .mode = 0644, 1115 .proc_handler = proc_dointvec, 1116 }, 1117 { 1118 .procname = "page-cluster", 1119 .data = &page_cluster, 1120 .maxlen = sizeof(int), 1121 .mode = 0644, 1122 .proc_handler = proc_dointvec_minmax, 1123 .extra1 = &zero, 1124 }, 1125 { 1126 .procname = "dirty_background_ratio", 1127 .data = &dirty_background_ratio, 1128 .maxlen = sizeof(dirty_background_ratio), 1129 .mode = 0644, 1130 .proc_handler = dirty_background_ratio_handler, 1131 .extra1 = &zero, 1132 .extra2 = &one_hundred, 1133 }, 1134 { 1135 .procname = "dirty_background_bytes", 1136 .data = &dirty_background_bytes, 1137 .maxlen = sizeof(dirty_background_bytes), 1138 .mode = 0644, 1139 .proc_handler = dirty_background_bytes_handler, 1140 .extra1 = &one_ul, 1141 }, 1142 { 1143 .procname = "dirty_ratio", 1144 .data = &vm_dirty_ratio, 1145 .maxlen = sizeof(vm_dirty_ratio), 1146 .mode = 0644, 1147 .proc_handler = dirty_ratio_handler, 1148 .extra1 = &zero, 1149 .extra2 = &one_hundred, 1150 }, 1151 { 1152 .procname = "dirty_bytes", 1153 .data = &vm_dirty_bytes, 1154 .maxlen = sizeof(vm_dirty_bytes), 1155 .mode = 0644, 1156 .proc_handler = dirty_bytes_handler, 1157 .extra1 = &dirty_bytes_min, 1158 }, 1159 { 1160 .procname = "dirty_writeback_centisecs", 1161 .data = &dirty_writeback_interval, 1162 .maxlen = sizeof(dirty_writeback_interval), 1163 .mode = 0644, 1164 .proc_handler = dirty_writeback_centisecs_handler, 1165 }, 1166 { 1167 .procname = "dirty_expire_centisecs", 1168 .data = &dirty_expire_interval, 1169 .maxlen = sizeof(dirty_expire_interval), 1170 .mode = 0644, 1171 .proc_handler = proc_dointvec_minmax, 1172 .extra1 = &zero, 1173 }, 1174 { 1175 .procname = "nr_pdflush_threads", 1176 .mode = 0444 /* read-only */, 1177 .proc_handler = pdflush_proc_obsolete, 1178 }, 1179 { 1180 .procname = "swappiness", 1181 .data = &vm_swappiness, 1182 .maxlen = sizeof(vm_swappiness), 1183 .mode = 0644, 1184 .proc_handler = proc_dointvec_minmax, 1185 .extra1 = &zero, 1186 .extra2 = &one_hundred, 1187 }, 1188 #ifdef CONFIG_HUGETLB_PAGE 1189 { 1190 .procname = "nr_hugepages", 1191 .data = NULL, 1192 .maxlen = sizeof(unsigned long), 1193 .mode = 0644, 1194 .proc_handler = hugetlb_sysctl_handler, 1195 .extra1 = (void *)&hugetlb_zero, 1196 .extra2 = (void *)&hugetlb_infinity, 1197 }, 1198 #ifdef CONFIG_NUMA 1199 { 1200 .procname = "nr_hugepages_mempolicy", 1201 .data = NULL, 1202 .maxlen = sizeof(unsigned long), 1203 .mode = 0644, 1204 .proc_handler = &hugetlb_mempolicy_sysctl_handler, 1205 .extra1 = (void *)&hugetlb_zero, 1206 .extra2 = (void *)&hugetlb_infinity, 1207 }, 1208 #endif 1209 { 1210 .procname = "hugetlb_shm_group", 1211 .data = &sysctl_hugetlb_shm_group, 1212 .maxlen = sizeof(gid_t), 1213 .mode = 0644, 1214 .proc_handler = proc_dointvec, 1215 }, 1216 { 1217 .procname = "hugepages_treat_as_movable", 1218 .data = &hugepages_treat_as_movable, 1219 .maxlen = sizeof(int), 1220 .mode = 0644, 1221 .proc_handler = hugetlb_treat_movable_handler, 1222 }, 1223 { 1224 .procname = "nr_overcommit_hugepages", 1225 .data = NULL, 1226 .maxlen = sizeof(unsigned long), 1227 .mode = 0644, 1228 .proc_handler = hugetlb_overcommit_handler, 1229 .extra1 = (void *)&hugetlb_zero, 1230 .extra2 = (void *)&hugetlb_infinity, 1231 }, 1232 #endif 1233 { 1234 .procname = "lowmem_reserve_ratio", 1235 .data = &sysctl_lowmem_reserve_ratio, 1236 .maxlen = sizeof(sysctl_lowmem_reserve_ratio), 1237 .mode = 0644, 1238 .proc_handler = lowmem_reserve_ratio_sysctl_handler, 1239 }, 1240 { 1241 .procname = "drop_caches", 1242 .data = &sysctl_drop_caches, 1243 .maxlen = sizeof(int), 1244 .mode = 0644, 1245 .proc_handler = drop_caches_sysctl_handler, 1246 .extra1 = &one, 1247 .extra2 = &three, 1248 }, 1249 #ifdef CONFIG_COMPACTION 1250 { 1251 .procname = "compact_memory", 1252 .data = &sysctl_compact_memory, 1253 .maxlen = sizeof(int), 1254 .mode = 0200, 1255 .proc_handler = sysctl_compaction_handler, 1256 }, 1257 { 1258 .procname = "extfrag_threshold", 1259 .data = &sysctl_extfrag_threshold, 1260 .maxlen = sizeof(int), 1261 .mode = 0644, 1262 .proc_handler = sysctl_extfrag_handler, 1263 .extra1 = &min_extfrag_threshold, 1264 .extra2 = &max_extfrag_threshold, 1265 }, 1266 1267 #endif /* CONFIG_COMPACTION */ 1268 { 1269 .procname = "min_free_kbytes", 1270 .data = &min_free_kbytes, 1271 .maxlen = sizeof(min_free_kbytes), 1272 .mode = 0644, 1273 .proc_handler = min_free_kbytes_sysctl_handler, 1274 .extra1 = &zero, 1275 }, 1276 { 1277 .procname = "percpu_pagelist_fraction", 1278 .data = &percpu_pagelist_fraction, 1279 .maxlen = sizeof(percpu_pagelist_fraction), 1280 .mode = 0644, 1281 .proc_handler = percpu_pagelist_fraction_sysctl_handler, 1282 .extra1 = &min_percpu_pagelist_fract, 1283 }, 1284 #ifdef CONFIG_MMU 1285 { 1286 .procname = "max_map_count", 1287 .data = &sysctl_max_map_count, 1288 .maxlen = sizeof(sysctl_max_map_count), 1289 .mode = 0644, 1290 .proc_handler = proc_dointvec_minmax, 1291 .extra1 = &zero, 1292 }, 1293 #else 1294 { 1295 .procname = "nr_trim_pages", 1296 .data = &sysctl_nr_trim_pages, 1297 .maxlen = sizeof(sysctl_nr_trim_pages), 1298 .mode = 0644, 1299 .proc_handler = proc_dointvec_minmax, 1300 .extra1 = &zero, 1301 }, 1302 #endif 1303 { 1304 .procname = "laptop_mode", 1305 .data = &laptop_mode, 1306 .maxlen = sizeof(laptop_mode), 1307 .mode = 0644, 1308 .proc_handler = proc_dointvec_jiffies, 1309 }, 1310 { 1311 .procname = "block_dump", 1312 .data = &block_dump, 1313 .maxlen = sizeof(block_dump), 1314 .mode = 0644, 1315 .proc_handler = proc_dointvec, 1316 .extra1 = &zero, 1317 }, 1318 { 1319 .procname = "vfs_cache_pressure", 1320 .data = &sysctl_vfs_cache_pressure, 1321 .maxlen = sizeof(sysctl_vfs_cache_pressure), 1322 .mode = 0644, 1323 .proc_handler = proc_dointvec, 1324 .extra1 = &zero, 1325 }, 1326 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT 1327 { 1328 .procname = "legacy_va_layout", 1329 .data = &sysctl_legacy_va_layout, 1330 .maxlen = sizeof(sysctl_legacy_va_layout), 1331 .mode = 0644, 1332 .proc_handler = proc_dointvec, 1333 .extra1 = &zero, 1334 }, 1335 #endif 1336 #ifdef CONFIG_NUMA 1337 { 1338 .procname = "zone_reclaim_mode", 1339 .data = &zone_reclaim_mode, 1340 .maxlen = sizeof(zone_reclaim_mode), 1341 .mode = 0644, 1342 .proc_handler = proc_dointvec, 1343 .extra1 = &zero, 1344 }, 1345 { 1346 .procname = "min_unmapped_ratio", 1347 .data = &sysctl_min_unmapped_ratio, 1348 .maxlen = sizeof(sysctl_min_unmapped_ratio), 1349 .mode = 0644, 1350 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler, 1351 .extra1 = &zero, 1352 .extra2 = &one_hundred, 1353 }, 1354 { 1355 .procname = "min_slab_ratio", 1356 .data = &sysctl_min_slab_ratio, 1357 .maxlen = sizeof(sysctl_min_slab_ratio), 1358 .mode = 0644, 1359 .proc_handler = sysctl_min_slab_ratio_sysctl_handler, 1360 .extra1 = &zero, 1361 .extra2 = &one_hundred, 1362 }, 1363 #endif 1364 #ifdef CONFIG_SMP 1365 { 1366 .procname = "stat_interval", 1367 .data = &sysctl_stat_interval, 1368 .maxlen = sizeof(sysctl_stat_interval), 1369 .mode = 0644, 1370 .proc_handler = proc_dointvec_jiffies, 1371 }, 1372 #endif 1373 #ifdef CONFIG_MMU 1374 { 1375 .procname = "mmap_min_addr", 1376 .data = &dac_mmap_min_addr, 1377 .maxlen = sizeof(unsigned long), 1378 .mode = 0644, 1379 .proc_handler = mmap_min_addr_handler, 1380 }, 1381 #endif 1382 #ifdef CONFIG_NUMA 1383 { 1384 .procname = "numa_zonelist_order", 1385 .data = &numa_zonelist_order, 1386 .maxlen = NUMA_ZONELIST_ORDER_LEN, 1387 .mode = 0644, 1388 .proc_handler = numa_zonelist_order_handler, 1389 }, 1390 #endif 1391 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \ 1392 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL)) 1393 { 1394 .procname = "vdso_enabled", 1395 .data = &vdso_enabled, 1396 .maxlen = sizeof(vdso_enabled), 1397 .mode = 0644, 1398 .proc_handler = proc_dointvec, 1399 .extra1 = &zero, 1400 }, 1401 #endif 1402 #ifdef CONFIG_HIGHMEM 1403 { 1404 .procname = "highmem_is_dirtyable", 1405 .data = &vm_highmem_is_dirtyable, 1406 .maxlen = sizeof(vm_highmem_is_dirtyable), 1407 .mode = 0644, 1408 .proc_handler = proc_dointvec_minmax, 1409 .extra1 = &zero, 1410 .extra2 = &one, 1411 }, 1412 #endif 1413 { 1414 .procname = "scan_unevictable_pages", 1415 .data = &scan_unevictable_pages, 1416 .maxlen = sizeof(scan_unevictable_pages), 1417 .mode = 0644, 1418 .proc_handler = scan_unevictable_handler, 1419 }, 1420 #ifdef CONFIG_MEMORY_FAILURE 1421 { 1422 .procname = "memory_failure_early_kill", 1423 .data = &sysctl_memory_failure_early_kill, 1424 .maxlen = sizeof(sysctl_memory_failure_early_kill), 1425 .mode = 0644, 1426 .proc_handler = proc_dointvec_minmax, 1427 .extra1 = &zero, 1428 .extra2 = &one, 1429 }, 1430 { 1431 .procname = "memory_failure_recovery", 1432 .data = &sysctl_memory_failure_recovery, 1433 .maxlen = sizeof(sysctl_memory_failure_recovery), 1434 .mode = 0644, 1435 .proc_handler = proc_dointvec_minmax, 1436 .extra1 = &zero, 1437 .extra2 = &one, 1438 }, 1439 #endif 1440 { 1441 .procname = "user_reserve_kbytes", 1442 .data = &sysctl_user_reserve_kbytes, 1443 .maxlen = sizeof(sysctl_user_reserve_kbytes), 1444 .mode = 0644, 1445 .proc_handler = proc_doulongvec_minmax, 1446 }, 1447 { 1448 .procname = "admin_reserve_kbytes", 1449 .data = &sysctl_admin_reserve_kbytes, 1450 .maxlen = sizeof(sysctl_admin_reserve_kbytes), 1451 .mode = 0644, 1452 .proc_handler = proc_doulongvec_minmax, 1453 }, 1454 { } 1455 }; 1456 1457 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE) 1458 static struct ctl_table binfmt_misc_table[] = { 1459 { } 1460 }; 1461 #endif 1462 1463 static struct ctl_table fs_table[] = { 1464 { 1465 .procname = "inode-nr", 1466 .data = &inodes_stat, 1467 .maxlen = 2*sizeof(int), 1468 .mode = 0444, 1469 .proc_handler = proc_nr_inodes, 1470 }, 1471 { 1472 .procname = "inode-state", 1473 .data = &inodes_stat, 1474 .maxlen = 7*sizeof(int), 1475 .mode = 0444, 1476 .proc_handler = proc_nr_inodes, 1477 }, 1478 { 1479 .procname = "file-nr", 1480 .data = &files_stat, 1481 .maxlen = sizeof(files_stat), 1482 .mode = 0444, 1483 .proc_handler = proc_nr_files, 1484 }, 1485 { 1486 .procname = "file-max", 1487 .data = &files_stat.max_files, 1488 .maxlen = sizeof(files_stat.max_files), 1489 .mode = 0644, 1490 .proc_handler = proc_doulongvec_minmax, 1491 }, 1492 { 1493 .procname = "nr_open", 1494 .data = &sysctl_nr_open, 1495 .maxlen = sizeof(int), 1496 .mode = 0644, 1497 .proc_handler = proc_dointvec_minmax, 1498 .extra1 = &sysctl_nr_open_min, 1499 .extra2 = &sysctl_nr_open_max, 1500 }, 1501 { 1502 .procname = "dentry-state", 1503 .data = &dentry_stat, 1504 .maxlen = 6*sizeof(int), 1505 .mode = 0444, 1506 .proc_handler = proc_nr_dentry, 1507 }, 1508 { 1509 .procname = "overflowuid", 1510 .data = &fs_overflowuid, 1511 .maxlen = sizeof(int), 1512 .mode = 0644, 1513 .proc_handler = proc_dointvec_minmax, 1514 .extra1 = &minolduid, 1515 .extra2 = &maxolduid, 1516 }, 1517 { 1518 .procname = "overflowgid", 1519 .data = &fs_overflowgid, 1520 .maxlen = sizeof(int), 1521 .mode = 0644, 1522 .proc_handler = proc_dointvec_minmax, 1523 .extra1 = &minolduid, 1524 .extra2 = &maxolduid, 1525 }, 1526 #ifdef CONFIG_FILE_LOCKING 1527 { 1528 .procname = "leases-enable", 1529 .data = &leases_enable, 1530 .maxlen = sizeof(int), 1531 .mode = 0644, 1532 .proc_handler = proc_dointvec, 1533 }, 1534 #endif 1535 #ifdef CONFIG_DNOTIFY 1536 { 1537 .procname = "dir-notify-enable", 1538 .data = &dir_notify_enable, 1539 .maxlen = sizeof(int), 1540 .mode = 0644, 1541 .proc_handler = proc_dointvec, 1542 }, 1543 #endif 1544 #ifdef CONFIG_MMU 1545 #ifdef CONFIG_FILE_LOCKING 1546 { 1547 .procname = "lease-break-time", 1548 .data = &lease_break_time, 1549 .maxlen = sizeof(int), 1550 .mode = 0644, 1551 .proc_handler = proc_dointvec, 1552 }, 1553 #endif 1554 #ifdef CONFIG_AIO 1555 { 1556 .procname = "aio-nr", 1557 .data = &aio_nr, 1558 .maxlen = sizeof(aio_nr), 1559 .mode = 0444, 1560 .proc_handler = proc_doulongvec_minmax, 1561 }, 1562 { 1563 .procname = "aio-max-nr", 1564 .data = &aio_max_nr, 1565 .maxlen = sizeof(aio_max_nr), 1566 .mode = 0644, 1567 .proc_handler = proc_doulongvec_minmax, 1568 }, 1569 #endif /* CONFIG_AIO */ 1570 #ifdef CONFIG_INOTIFY_USER 1571 { 1572 .procname = "inotify", 1573 .mode = 0555, 1574 .child = inotify_table, 1575 }, 1576 #endif 1577 #ifdef CONFIG_EPOLL 1578 { 1579 .procname = "epoll", 1580 .mode = 0555, 1581 .child = epoll_table, 1582 }, 1583 #endif 1584 #endif 1585 { 1586 .procname = "protected_symlinks", 1587 .data = &sysctl_protected_symlinks, 1588 .maxlen = sizeof(int), 1589 .mode = 0600, 1590 .proc_handler = proc_dointvec_minmax, 1591 .extra1 = &zero, 1592 .extra2 = &one, 1593 }, 1594 { 1595 .procname = "protected_hardlinks", 1596 .data = &sysctl_protected_hardlinks, 1597 .maxlen = sizeof(int), 1598 .mode = 0600, 1599 .proc_handler = proc_dointvec_minmax, 1600 .extra1 = &zero, 1601 .extra2 = &one, 1602 }, 1603 { 1604 .procname = "suid_dumpable", 1605 .data = &suid_dumpable, 1606 .maxlen = sizeof(int), 1607 .mode = 0644, 1608 .proc_handler = proc_dointvec_minmax_coredump, 1609 .extra1 = &zero, 1610 .extra2 = &two, 1611 }, 1612 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE) 1613 { 1614 .procname = "binfmt_misc", 1615 .mode = 0555, 1616 .child = binfmt_misc_table, 1617 }, 1618 #endif 1619 { 1620 .procname = "pipe-max-size", 1621 .data = &pipe_max_size, 1622 .maxlen = sizeof(int), 1623 .mode = 0644, 1624 .proc_handler = &pipe_proc_fn, 1625 .extra1 = &pipe_min_size, 1626 }, 1627 { } 1628 }; 1629 1630 static struct ctl_table debug_table[] = { 1631 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE 1632 { 1633 .procname = "exception-trace", 1634 .data = &show_unhandled_signals, 1635 .maxlen = sizeof(int), 1636 .mode = 0644, 1637 .proc_handler = proc_dointvec 1638 }, 1639 #endif 1640 #if defined(CONFIG_OPTPROBES) 1641 { 1642 .procname = "kprobes-optimization", 1643 .data = &sysctl_kprobes_optimization, 1644 .maxlen = sizeof(int), 1645 .mode = 0644, 1646 .proc_handler = proc_kprobes_optimization_handler, 1647 .extra1 = &zero, 1648 .extra2 = &one, 1649 }, 1650 #endif 1651 { } 1652 }; 1653 1654 static struct ctl_table dev_table[] = { 1655 { } 1656 }; 1657 1658 int __init sysctl_init(void) 1659 { 1660 struct ctl_table_header *hdr; 1661 1662 hdr = register_sysctl_table(sysctl_base_table); 1663 kmemleak_not_leak(hdr); 1664 return 0; 1665 } 1666 1667 #endif /* CONFIG_SYSCTL */ 1668 1669 /* 1670 * /proc/sys support 1671 */ 1672 1673 #ifdef CONFIG_PROC_SYSCTL 1674 1675 static int _proc_do_string(void* data, int maxlen, int write, 1676 void __user *buffer, 1677 size_t *lenp, loff_t *ppos) 1678 { 1679 size_t len; 1680 char __user *p; 1681 char c; 1682 1683 if (!data || !maxlen || !*lenp) { 1684 *lenp = 0; 1685 return 0; 1686 } 1687 1688 if (write) { 1689 len = 0; 1690 p = buffer; 1691 while (len < *lenp) { 1692 if (get_user(c, p++)) 1693 return -EFAULT; 1694 if (c == 0 || c == '\n') 1695 break; 1696 len++; 1697 } 1698 if (len >= maxlen) 1699 len = maxlen-1; 1700 if(copy_from_user(data, buffer, len)) 1701 return -EFAULT; 1702 ((char *) data)[len] = 0; 1703 *ppos += *lenp; 1704 } else { 1705 len = strlen(data); 1706 if (len > maxlen) 1707 len = maxlen; 1708 1709 if (*ppos > len) { 1710 *lenp = 0; 1711 return 0; 1712 } 1713 1714 data += *ppos; 1715 len -= *ppos; 1716 1717 if (len > *lenp) 1718 len = *lenp; 1719 if (len) 1720 if(copy_to_user(buffer, data, len)) 1721 return -EFAULT; 1722 if (len < *lenp) { 1723 if(put_user('\n', ((char __user *) buffer) + len)) 1724 return -EFAULT; 1725 len++; 1726 } 1727 *lenp = len; 1728 *ppos += len; 1729 } 1730 return 0; 1731 } 1732 1733 /** 1734 * proc_dostring - read a string sysctl 1735 * @table: the sysctl table 1736 * @write: %TRUE if this is a write to the sysctl file 1737 * @buffer: the user buffer 1738 * @lenp: the size of the user buffer 1739 * @ppos: file position 1740 * 1741 * Reads/writes a string from/to the user buffer. If the kernel 1742 * buffer provided is not large enough to hold the string, the 1743 * string is truncated. The copied string is %NULL-terminated. 1744 * If the string is being read by the user process, it is copied 1745 * and a newline '\n' is added. It is truncated if the buffer is 1746 * not large enough. 1747 * 1748 * Returns 0 on success. 1749 */ 1750 int proc_dostring(struct ctl_table *table, int write, 1751 void __user *buffer, size_t *lenp, loff_t *ppos) 1752 { 1753 return _proc_do_string(table->data, table->maxlen, write, 1754 buffer, lenp, ppos); 1755 } 1756 1757 static size_t proc_skip_spaces(char **buf) 1758 { 1759 size_t ret; 1760 char *tmp = skip_spaces(*buf); 1761 ret = tmp - *buf; 1762 *buf = tmp; 1763 return ret; 1764 } 1765 1766 static void proc_skip_char(char **buf, size_t *size, const char v) 1767 { 1768 while (*size) { 1769 if (**buf != v) 1770 break; 1771 (*size)--; 1772 (*buf)++; 1773 } 1774 } 1775 1776 #define TMPBUFLEN 22 1777 /** 1778 * proc_get_long - reads an ASCII formatted integer from a user buffer 1779 * 1780 * @buf: a kernel buffer 1781 * @size: size of the kernel buffer 1782 * @val: this is where the number will be stored 1783 * @neg: set to %TRUE if number is negative 1784 * @perm_tr: a vector which contains the allowed trailers 1785 * @perm_tr_len: size of the perm_tr vector 1786 * @tr: pointer to store the trailer character 1787 * 1788 * In case of success %0 is returned and @buf and @size are updated with 1789 * the amount of bytes read. If @tr is non-NULL and a trailing 1790 * character exists (size is non-zero after returning from this 1791 * function), @tr is updated with the trailing character. 1792 */ 1793 static int proc_get_long(char **buf, size_t *size, 1794 unsigned long *val, bool *neg, 1795 const char *perm_tr, unsigned perm_tr_len, char *tr) 1796 { 1797 int len; 1798 char *p, tmp[TMPBUFLEN]; 1799 1800 if (!*size) 1801 return -EINVAL; 1802 1803 len = *size; 1804 if (len > TMPBUFLEN - 1) 1805 len = TMPBUFLEN - 1; 1806 1807 memcpy(tmp, *buf, len); 1808 1809 tmp[len] = 0; 1810 p = tmp; 1811 if (*p == '-' && *size > 1) { 1812 *neg = true; 1813 p++; 1814 } else 1815 *neg = false; 1816 if (!isdigit(*p)) 1817 return -EINVAL; 1818 1819 *val = simple_strtoul(p, &p, 0); 1820 1821 len = p - tmp; 1822 1823 /* We don't know if the next char is whitespace thus we may accept 1824 * invalid integers (e.g. 1234...a) or two integers instead of one 1825 * (e.g. 123...1). So lets not allow such large numbers. */ 1826 if (len == TMPBUFLEN - 1) 1827 return -EINVAL; 1828 1829 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len)) 1830 return -EINVAL; 1831 1832 if (tr && (len < *size)) 1833 *tr = *p; 1834 1835 *buf += len; 1836 *size -= len; 1837 1838 return 0; 1839 } 1840 1841 /** 1842 * proc_put_long - converts an integer to a decimal ASCII formatted string 1843 * 1844 * @buf: the user buffer 1845 * @size: the size of the user buffer 1846 * @val: the integer to be converted 1847 * @neg: sign of the number, %TRUE for negative 1848 * 1849 * In case of success %0 is returned and @buf and @size are updated with 1850 * the amount of bytes written. 1851 */ 1852 static int proc_put_long(void __user **buf, size_t *size, unsigned long val, 1853 bool neg) 1854 { 1855 int len; 1856 char tmp[TMPBUFLEN], *p = tmp; 1857 1858 sprintf(p, "%s%lu", neg ? "-" : "", val); 1859 len = strlen(tmp); 1860 if (len > *size) 1861 len = *size; 1862 if (copy_to_user(*buf, tmp, len)) 1863 return -EFAULT; 1864 *size -= len; 1865 *buf += len; 1866 return 0; 1867 } 1868 #undef TMPBUFLEN 1869 1870 static int proc_put_char(void __user **buf, size_t *size, char c) 1871 { 1872 if (*size) { 1873 char __user **buffer = (char __user **)buf; 1874 if (put_user(c, *buffer)) 1875 return -EFAULT; 1876 (*size)--, (*buffer)++; 1877 *buf = *buffer; 1878 } 1879 return 0; 1880 } 1881 1882 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp, 1883 int *valp, 1884 int write, void *data) 1885 { 1886 if (write) { 1887 *valp = *negp ? -*lvalp : *lvalp; 1888 } else { 1889 int val = *valp; 1890 if (val < 0) { 1891 *negp = true; 1892 *lvalp = (unsigned long)-val; 1893 } else { 1894 *negp = false; 1895 *lvalp = (unsigned long)val; 1896 } 1897 } 1898 return 0; 1899 } 1900 1901 static const char proc_wspace_sep[] = { ' ', '\t', '\n' }; 1902 1903 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table, 1904 int write, void __user *buffer, 1905 size_t *lenp, loff_t *ppos, 1906 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 1907 int write, void *data), 1908 void *data) 1909 { 1910 int *i, vleft, first = 1, err = 0; 1911 unsigned long page = 0; 1912 size_t left; 1913 char *kbuf; 1914 1915 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) { 1916 *lenp = 0; 1917 return 0; 1918 } 1919 1920 i = (int *) tbl_data; 1921 vleft = table->maxlen / sizeof(*i); 1922 left = *lenp; 1923 1924 if (!conv) 1925 conv = do_proc_dointvec_conv; 1926 1927 if (write) { 1928 if (left > PAGE_SIZE - 1) 1929 left = PAGE_SIZE - 1; 1930 page = __get_free_page(GFP_TEMPORARY); 1931 kbuf = (char *) page; 1932 if (!kbuf) 1933 return -ENOMEM; 1934 if (copy_from_user(kbuf, buffer, left)) { 1935 err = -EFAULT; 1936 goto free; 1937 } 1938 kbuf[left] = 0; 1939 } 1940 1941 for (; left && vleft--; i++, first=0) { 1942 unsigned long lval; 1943 bool neg; 1944 1945 if (write) { 1946 left -= proc_skip_spaces(&kbuf); 1947 1948 if (!left) 1949 break; 1950 err = proc_get_long(&kbuf, &left, &lval, &neg, 1951 proc_wspace_sep, 1952 sizeof(proc_wspace_sep), NULL); 1953 if (err) 1954 break; 1955 if (conv(&neg, &lval, i, 1, data)) { 1956 err = -EINVAL; 1957 break; 1958 } 1959 } else { 1960 if (conv(&neg, &lval, i, 0, data)) { 1961 err = -EINVAL; 1962 break; 1963 } 1964 if (!first) 1965 err = proc_put_char(&buffer, &left, '\t'); 1966 if (err) 1967 break; 1968 err = proc_put_long(&buffer, &left, lval, neg); 1969 if (err) 1970 break; 1971 } 1972 } 1973 1974 if (!write && !first && left && !err) 1975 err = proc_put_char(&buffer, &left, '\n'); 1976 if (write && !err && left) 1977 left -= proc_skip_spaces(&kbuf); 1978 free: 1979 if (write) { 1980 free_page(page); 1981 if (first) 1982 return err ? : -EINVAL; 1983 } 1984 *lenp -= left; 1985 *ppos += *lenp; 1986 return err; 1987 } 1988 1989 static int do_proc_dointvec(struct ctl_table *table, int write, 1990 void __user *buffer, size_t *lenp, loff_t *ppos, 1991 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 1992 int write, void *data), 1993 void *data) 1994 { 1995 return __do_proc_dointvec(table->data, table, write, 1996 buffer, lenp, ppos, conv, data); 1997 } 1998 1999 /** 2000 * proc_dointvec - read a vector of integers 2001 * @table: the sysctl table 2002 * @write: %TRUE if this is a write to the sysctl file 2003 * @buffer: the user buffer 2004 * @lenp: the size of the user buffer 2005 * @ppos: file position 2006 * 2007 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2008 * values from/to the user buffer, treated as an ASCII string. 2009 * 2010 * Returns 0 on success. 2011 */ 2012 int proc_dointvec(struct ctl_table *table, int write, 2013 void __user *buffer, size_t *lenp, loff_t *ppos) 2014 { 2015 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2016 NULL,NULL); 2017 } 2018 2019 /* 2020 * Taint values can only be increased 2021 * This means we can safely use a temporary. 2022 */ 2023 static int proc_taint(struct ctl_table *table, int write, 2024 void __user *buffer, size_t *lenp, loff_t *ppos) 2025 { 2026 struct ctl_table t; 2027 unsigned long tmptaint = get_taint(); 2028 int err; 2029 2030 if (write && !capable(CAP_SYS_ADMIN)) 2031 return -EPERM; 2032 2033 t = *table; 2034 t.data = &tmptaint; 2035 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos); 2036 if (err < 0) 2037 return err; 2038 2039 if (write) { 2040 /* 2041 * Poor man's atomic or. Not worth adding a primitive 2042 * to everyone's atomic.h for this 2043 */ 2044 int i; 2045 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) { 2046 if ((tmptaint >> i) & 1) 2047 add_taint(i, LOCKDEP_STILL_OK); 2048 } 2049 } 2050 2051 return err; 2052 } 2053 2054 #ifdef CONFIG_PRINTK 2055 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write, 2056 void __user *buffer, size_t *lenp, loff_t *ppos) 2057 { 2058 if (write && !capable(CAP_SYS_ADMIN)) 2059 return -EPERM; 2060 2061 return proc_dointvec_minmax(table, write, buffer, lenp, ppos); 2062 } 2063 #endif 2064 2065 struct do_proc_dointvec_minmax_conv_param { 2066 int *min; 2067 int *max; 2068 }; 2069 2070 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp, 2071 int *valp, 2072 int write, void *data) 2073 { 2074 struct do_proc_dointvec_minmax_conv_param *param = data; 2075 if (write) { 2076 int val = *negp ? -*lvalp : *lvalp; 2077 if ((param->min && *param->min > val) || 2078 (param->max && *param->max < val)) 2079 return -EINVAL; 2080 *valp = val; 2081 } else { 2082 int val = *valp; 2083 if (val < 0) { 2084 *negp = true; 2085 *lvalp = (unsigned long)-val; 2086 } else { 2087 *negp = false; 2088 *lvalp = (unsigned long)val; 2089 } 2090 } 2091 return 0; 2092 } 2093 2094 /** 2095 * proc_dointvec_minmax - read a vector of integers with min/max values 2096 * @table: the sysctl table 2097 * @write: %TRUE if this is a write to the sysctl file 2098 * @buffer: the user buffer 2099 * @lenp: the size of the user buffer 2100 * @ppos: file position 2101 * 2102 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2103 * values from/to the user buffer, treated as an ASCII string. 2104 * 2105 * This routine will ensure the values are within the range specified by 2106 * table->extra1 (min) and table->extra2 (max). 2107 * 2108 * Returns 0 on success. 2109 */ 2110 int proc_dointvec_minmax(struct ctl_table *table, int write, 2111 void __user *buffer, size_t *lenp, loff_t *ppos) 2112 { 2113 struct do_proc_dointvec_minmax_conv_param param = { 2114 .min = (int *) table->extra1, 2115 .max = (int *) table->extra2, 2116 }; 2117 return do_proc_dointvec(table, write, buffer, lenp, ppos, 2118 do_proc_dointvec_minmax_conv, ¶m); 2119 } 2120 2121 static void validate_coredump_safety(void) 2122 { 2123 #ifdef CONFIG_COREDUMP 2124 if (suid_dumpable == SUID_DUMP_ROOT && 2125 core_pattern[0] != '/' && core_pattern[0] != '|') { 2126 printk(KERN_WARNING "Unsafe core_pattern used with "\ 2127 "suid_dumpable=2. Pipe handler or fully qualified "\ 2128 "core dump path required.\n"); 2129 } 2130 #endif 2131 } 2132 2133 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write, 2134 void __user *buffer, size_t *lenp, loff_t *ppos) 2135 { 2136 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos); 2137 if (!error) 2138 validate_coredump_safety(); 2139 return error; 2140 } 2141 2142 #ifdef CONFIG_COREDUMP 2143 static int proc_dostring_coredump(struct ctl_table *table, int write, 2144 void __user *buffer, size_t *lenp, loff_t *ppos) 2145 { 2146 int error = proc_dostring(table, write, buffer, lenp, ppos); 2147 if (!error) 2148 validate_coredump_safety(); 2149 return error; 2150 } 2151 #endif 2152 2153 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write, 2154 void __user *buffer, 2155 size_t *lenp, loff_t *ppos, 2156 unsigned long convmul, 2157 unsigned long convdiv) 2158 { 2159 unsigned long *i, *min, *max; 2160 int vleft, first = 1, err = 0; 2161 unsigned long page = 0; 2162 size_t left; 2163 char *kbuf; 2164 2165 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) { 2166 *lenp = 0; 2167 return 0; 2168 } 2169 2170 i = (unsigned long *) data; 2171 min = (unsigned long *) table->extra1; 2172 max = (unsigned long *) table->extra2; 2173 vleft = table->maxlen / sizeof(unsigned long); 2174 left = *lenp; 2175 2176 if (write) { 2177 if (left > PAGE_SIZE - 1) 2178 left = PAGE_SIZE - 1; 2179 page = __get_free_page(GFP_TEMPORARY); 2180 kbuf = (char *) page; 2181 if (!kbuf) 2182 return -ENOMEM; 2183 if (copy_from_user(kbuf, buffer, left)) { 2184 err = -EFAULT; 2185 goto free; 2186 } 2187 kbuf[left] = 0; 2188 } 2189 2190 for (; left && vleft--; i++, first = 0) { 2191 unsigned long val; 2192 2193 if (write) { 2194 bool neg; 2195 2196 left -= proc_skip_spaces(&kbuf); 2197 2198 err = proc_get_long(&kbuf, &left, &val, &neg, 2199 proc_wspace_sep, 2200 sizeof(proc_wspace_sep), NULL); 2201 if (err) 2202 break; 2203 if (neg) 2204 continue; 2205 if ((min && val < *min) || (max && val > *max)) 2206 continue; 2207 *i = val; 2208 } else { 2209 val = convdiv * (*i) / convmul; 2210 if (!first) 2211 err = proc_put_char(&buffer, &left, '\t'); 2212 err = proc_put_long(&buffer, &left, val, false); 2213 if (err) 2214 break; 2215 } 2216 } 2217 2218 if (!write && !first && left && !err) 2219 err = proc_put_char(&buffer, &left, '\n'); 2220 if (write && !err) 2221 left -= proc_skip_spaces(&kbuf); 2222 free: 2223 if (write) { 2224 free_page(page); 2225 if (first) 2226 return err ? : -EINVAL; 2227 } 2228 *lenp -= left; 2229 *ppos += *lenp; 2230 return err; 2231 } 2232 2233 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write, 2234 void __user *buffer, 2235 size_t *lenp, loff_t *ppos, 2236 unsigned long convmul, 2237 unsigned long convdiv) 2238 { 2239 return __do_proc_doulongvec_minmax(table->data, table, write, 2240 buffer, lenp, ppos, convmul, convdiv); 2241 } 2242 2243 /** 2244 * proc_doulongvec_minmax - read a vector of long integers with min/max values 2245 * @table: the sysctl table 2246 * @write: %TRUE if this is a write to the sysctl file 2247 * @buffer: the user buffer 2248 * @lenp: the size of the user buffer 2249 * @ppos: file position 2250 * 2251 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 2252 * values from/to the user buffer, treated as an ASCII string. 2253 * 2254 * This routine will ensure the values are within the range specified by 2255 * table->extra1 (min) and table->extra2 (max). 2256 * 2257 * Returns 0 on success. 2258 */ 2259 int proc_doulongvec_minmax(struct ctl_table *table, int write, 2260 void __user *buffer, size_t *lenp, loff_t *ppos) 2261 { 2262 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l); 2263 } 2264 2265 /** 2266 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values 2267 * @table: the sysctl table 2268 * @write: %TRUE if this is a write to the sysctl file 2269 * @buffer: the user buffer 2270 * @lenp: the size of the user buffer 2271 * @ppos: file position 2272 * 2273 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 2274 * values from/to the user buffer, treated as an ASCII string. The values 2275 * are treated as milliseconds, and converted to jiffies when they are stored. 2276 * 2277 * This routine will ensure the values are within the range specified by 2278 * table->extra1 (min) and table->extra2 (max). 2279 * 2280 * Returns 0 on success. 2281 */ 2282 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 2283 void __user *buffer, 2284 size_t *lenp, loff_t *ppos) 2285 { 2286 return do_proc_doulongvec_minmax(table, write, buffer, 2287 lenp, ppos, HZ, 1000l); 2288 } 2289 2290 2291 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp, 2292 int *valp, 2293 int write, void *data) 2294 { 2295 if (write) { 2296 if (*lvalp > LONG_MAX / HZ) 2297 return 1; 2298 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ); 2299 } else { 2300 int val = *valp; 2301 unsigned long lval; 2302 if (val < 0) { 2303 *negp = true; 2304 lval = (unsigned long)-val; 2305 } else { 2306 *negp = false; 2307 lval = (unsigned long)val; 2308 } 2309 *lvalp = lval / HZ; 2310 } 2311 return 0; 2312 } 2313 2314 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp, 2315 int *valp, 2316 int write, void *data) 2317 { 2318 if (write) { 2319 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ) 2320 return 1; 2321 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp); 2322 } else { 2323 int val = *valp; 2324 unsigned long lval; 2325 if (val < 0) { 2326 *negp = true; 2327 lval = (unsigned long)-val; 2328 } else { 2329 *negp = false; 2330 lval = (unsigned long)val; 2331 } 2332 *lvalp = jiffies_to_clock_t(lval); 2333 } 2334 return 0; 2335 } 2336 2337 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp, 2338 int *valp, 2339 int write, void *data) 2340 { 2341 if (write) { 2342 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp); 2343 } else { 2344 int val = *valp; 2345 unsigned long lval; 2346 if (val < 0) { 2347 *negp = true; 2348 lval = (unsigned long)-val; 2349 } else { 2350 *negp = false; 2351 lval = (unsigned long)val; 2352 } 2353 *lvalp = jiffies_to_msecs(lval); 2354 } 2355 return 0; 2356 } 2357 2358 /** 2359 * proc_dointvec_jiffies - read a vector of integers as seconds 2360 * @table: the sysctl table 2361 * @write: %TRUE if this is a write to the sysctl file 2362 * @buffer: the user buffer 2363 * @lenp: the size of the user buffer 2364 * @ppos: file position 2365 * 2366 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2367 * values from/to the user buffer, treated as an ASCII string. 2368 * The values read are assumed to be in seconds, and are converted into 2369 * jiffies. 2370 * 2371 * Returns 0 on success. 2372 */ 2373 int proc_dointvec_jiffies(struct ctl_table *table, int write, 2374 void __user *buffer, size_t *lenp, loff_t *ppos) 2375 { 2376 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2377 do_proc_dointvec_jiffies_conv,NULL); 2378 } 2379 2380 /** 2381 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds 2382 * @table: the sysctl table 2383 * @write: %TRUE if this is a write to the sysctl file 2384 * @buffer: the user buffer 2385 * @lenp: the size of the user buffer 2386 * @ppos: pointer to the file position 2387 * 2388 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2389 * values from/to the user buffer, treated as an ASCII string. 2390 * The values read are assumed to be in 1/USER_HZ seconds, and 2391 * are converted into jiffies. 2392 * 2393 * Returns 0 on success. 2394 */ 2395 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 2396 void __user *buffer, size_t *lenp, loff_t *ppos) 2397 { 2398 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2399 do_proc_dointvec_userhz_jiffies_conv,NULL); 2400 } 2401 2402 /** 2403 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds 2404 * @table: the sysctl table 2405 * @write: %TRUE if this is a write to the sysctl file 2406 * @buffer: the user buffer 2407 * @lenp: the size of the user buffer 2408 * @ppos: file position 2409 * @ppos: the current position in the file 2410 * 2411 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2412 * values from/to the user buffer, treated as an ASCII string. 2413 * The values read are assumed to be in 1/1000 seconds, and 2414 * are converted into jiffies. 2415 * 2416 * Returns 0 on success. 2417 */ 2418 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, 2419 void __user *buffer, size_t *lenp, loff_t *ppos) 2420 { 2421 return do_proc_dointvec(table, write, buffer, lenp, ppos, 2422 do_proc_dointvec_ms_jiffies_conv, NULL); 2423 } 2424 2425 static int proc_do_cad_pid(struct ctl_table *table, int write, 2426 void __user *buffer, size_t *lenp, loff_t *ppos) 2427 { 2428 struct pid *new_pid; 2429 pid_t tmp; 2430 int r; 2431 2432 tmp = pid_vnr(cad_pid); 2433 2434 r = __do_proc_dointvec(&tmp, table, write, buffer, 2435 lenp, ppos, NULL, NULL); 2436 if (r || !write) 2437 return r; 2438 2439 new_pid = find_get_pid(tmp); 2440 if (!new_pid) 2441 return -ESRCH; 2442 2443 put_pid(xchg(&cad_pid, new_pid)); 2444 return 0; 2445 } 2446 2447 /** 2448 * proc_do_large_bitmap - read/write from/to a large bitmap 2449 * @table: the sysctl table 2450 * @write: %TRUE if this is a write to the sysctl file 2451 * @buffer: the user buffer 2452 * @lenp: the size of the user buffer 2453 * @ppos: file position 2454 * 2455 * The bitmap is stored at table->data and the bitmap length (in bits) 2456 * in table->maxlen. 2457 * 2458 * We use a range comma separated format (e.g. 1,3-4,10-10) so that 2459 * large bitmaps may be represented in a compact manner. Writing into 2460 * the file will clear the bitmap then update it with the given input. 2461 * 2462 * Returns 0 on success. 2463 */ 2464 int proc_do_large_bitmap(struct ctl_table *table, int write, 2465 void __user *buffer, size_t *lenp, loff_t *ppos) 2466 { 2467 int err = 0; 2468 bool first = 1; 2469 size_t left = *lenp; 2470 unsigned long bitmap_len = table->maxlen; 2471 unsigned long *bitmap = (unsigned long *) table->data; 2472 unsigned long *tmp_bitmap = NULL; 2473 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c; 2474 2475 if (!bitmap_len || !left || (*ppos && !write)) { 2476 *lenp = 0; 2477 return 0; 2478 } 2479 2480 if (write) { 2481 unsigned long page = 0; 2482 char *kbuf; 2483 2484 if (left > PAGE_SIZE - 1) 2485 left = PAGE_SIZE - 1; 2486 2487 page = __get_free_page(GFP_TEMPORARY); 2488 kbuf = (char *) page; 2489 if (!kbuf) 2490 return -ENOMEM; 2491 if (copy_from_user(kbuf, buffer, left)) { 2492 free_page(page); 2493 return -EFAULT; 2494 } 2495 kbuf[left] = 0; 2496 2497 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long), 2498 GFP_KERNEL); 2499 if (!tmp_bitmap) { 2500 free_page(page); 2501 return -ENOMEM; 2502 } 2503 proc_skip_char(&kbuf, &left, '\n'); 2504 while (!err && left) { 2505 unsigned long val_a, val_b; 2506 bool neg; 2507 2508 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a, 2509 sizeof(tr_a), &c); 2510 if (err) 2511 break; 2512 if (val_a >= bitmap_len || neg) { 2513 err = -EINVAL; 2514 break; 2515 } 2516 2517 val_b = val_a; 2518 if (left) { 2519 kbuf++; 2520 left--; 2521 } 2522 2523 if (c == '-') { 2524 err = proc_get_long(&kbuf, &left, &val_b, 2525 &neg, tr_b, sizeof(tr_b), 2526 &c); 2527 if (err) 2528 break; 2529 if (val_b >= bitmap_len || neg || 2530 val_a > val_b) { 2531 err = -EINVAL; 2532 break; 2533 } 2534 if (left) { 2535 kbuf++; 2536 left--; 2537 } 2538 } 2539 2540 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1); 2541 first = 0; 2542 proc_skip_char(&kbuf, &left, '\n'); 2543 } 2544 free_page(page); 2545 } else { 2546 unsigned long bit_a, bit_b = 0; 2547 2548 while (left) { 2549 bit_a = find_next_bit(bitmap, bitmap_len, bit_b); 2550 if (bit_a >= bitmap_len) 2551 break; 2552 bit_b = find_next_zero_bit(bitmap, bitmap_len, 2553 bit_a + 1) - 1; 2554 2555 if (!first) { 2556 err = proc_put_char(&buffer, &left, ','); 2557 if (err) 2558 break; 2559 } 2560 err = proc_put_long(&buffer, &left, bit_a, false); 2561 if (err) 2562 break; 2563 if (bit_a != bit_b) { 2564 err = proc_put_char(&buffer, &left, '-'); 2565 if (err) 2566 break; 2567 err = proc_put_long(&buffer, &left, bit_b, false); 2568 if (err) 2569 break; 2570 } 2571 2572 first = 0; bit_b++; 2573 } 2574 if (!err) 2575 err = proc_put_char(&buffer, &left, '\n'); 2576 } 2577 2578 if (!err) { 2579 if (write) { 2580 if (*ppos) 2581 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len); 2582 else 2583 bitmap_copy(bitmap, tmp_bitmap, bitmap_len); 2584 } 2585 kfree(tmp_bitmap); 2586 *lenp -= left; 2587 *ppos += *lenp; 2588 return 0; 2589 } else { 2590 kfree(tmp_bitmap); 2591 return err; 2592 } 2593 } 2594 2595 #else /* CONFIG_PROC_SYSCTL */ 2596 2597 int proc_dostring(struct ctl_table *table, int write, 2598 void __user *buffer, size_t *lenp, loff_t *ppos) 2599 { 2600 return -ENOSYS; 2601 } 2602 2603 int proc_dointvec(struct ctl_table *table, int write, 2604 void __user *buffer, size_t *lenp, loff_t *ppos) 2605 { 2606 return -ENOSYS; 2607 } 2608 2609 int proc_dointvec_minmax(struct ctl_table *table, int write, 2610 void __user *buffer, size_t *lenp, loff_t *ppos) 2611 { 2612 return -ENOSYS; 2613 } 2614 2615 int proc_dointvec_jiffies(struct ctl_table *table, int write, 2616 void __user *buffer, size_t *lenp, loff_t *ppos) 2617 { 2618 return -ENOSYS; 2619 } 2620 2621 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 2622 void __user *buffer, size_t *lenp, loff_t *ppos) 2623 { 2624 return -ENOSYS; 2625 } 2626 2627 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, 2628 void __user *buffer, size_t *lenp, loff_t *ppos) 2629 { 2630 return -ENOSYS; 2631 } 2632 2633 int proc_doulongvec_minmax(struct ctl_table *table, int write, 2634 void __user *buffer, size_t *lenp, loff_t *ppos) 2635 { 2636 return -ENOSYS; 2637 } 2638 2639 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 2640 void __user *buffer, 2641 size_t *lenp, loff_t *ppos) 2642 { 2643 return -ENOSYS; 2644 } 2645 2646 2647 #endif /* CONFIG_PROC_SYSCTL */ 2648 2649 /* 2650 * No sense putting this after each symbol definition, twice, 2651 * exception granted :-) 2652 */ 2653 EXPORT_SYMBOL(proc_dointvec); 2654 EXPORT_SYMBOL(proc_dointvec_jiffies); 2655 EXPORT_SYMBOL(proc_dointvec_minmax); 2656 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies); 2657 EXPORT_SYMBOL(proc_dointvec_ms_jiffies); 2658 EXPORT_SYMBOL(proc_dostring); 2659 EXPORT_SYMBOL(proc_doulongvec_minmax); 2660 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax); 2661