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_CORE 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 static struct ctl_table fs_table[] = { 1542 { 1543 .procname = "inode-nr", 1544 .data = &inodes_stat, 1545 .maxlen = 2*sizeof(long), 1546 .mode = 0444, 1547 .proc_handler = proc_nr_inodes, 1548 }, 1549 { 1550 .procname = "inode-state", 1551 .data = &inodes_stat, 1552 .maxlen = 7*sizeof(long), 1553 .mode = 0444, 1554 .proc_handler = proc_nr_inodes, 1555 }, 1556 { 1557 .procname = "file-nr", 1558 .data = &files_stat, 1559 .maxlen = sizeof(files_stat), 1560 .mode = 0444, 1561 .proc_handler = proc_nr_files, 1562 }, 1563 { 1564 .procname = "file-max", 1565 .data = &files_stat.max_files, 1566 .maxlen = sizeof(files_stat.max_files), 1567 .mode = 0644, 1568 .proc_handler = proc_doulongvec_minmax, 1569 }, 1570 { 1571 .procname = "nr_open", 1572 .data = &sysctl_nr_open, 1573 .maxlen = sizeof(int), 1574 .mode = 0644, 1575 .proc_handler = proc_dointvec_minmax, 1576 .extra1 = &sysctl_nr_open_min, 1577 .extra2 = &sysctl_nr_open_max, 1578 }, 1579 { 1580 .procname = "dentry-state", 1581 .data = &dentry_stat, 1582 .maxlen = 6*sizeof(long), 1583 .mode = 0444, 1584 .proc_handler = proc_nr_dentry, 1585 }, 1586 { 1587 .procname = "overflowuid", 1588 .data = &fs_overflowuid, 1589 .maxlen = sizeof(int), 1590 .mode = 0644, 1591 .proc_handler = proc_dointvec_minmax, 1592 .extra1 = &minolduid, 1593 .extra2 = &maxolduid, 1594 }, 1595 { 1596 .procname = "overflowgid", 1597 .data = &fs_overflowgid, 1598 .maxlen = sizeof(int), 1599 .mode = 0644, 1600 .proc_handler = proc_dointvec_minmax, 1601 .extra1 = &minolduid, 1602 .extra2 = &maxolduid, 1603 }, 1604 #ifdef CONFIG_FILE_LOCKING 1605 { 1606 .procname = "leases-enable", 1607 .data = &leases_enable, 1608 .maxlen = sizeof(int), 1609 .mode = 0644, 1610 .proc_handler = proc_dointvec, 1611 }, 1612 #endif 1613 #ifdef CONFIG_DNOTIFY 1614 { 1615 .procname = "dir-notify-enable", 1616 .data = &dir_notify_enable, 1617 .maxlen = sizeof(int), 1618 .mode = 0644, 1619 .proc_handler = proc_dointvec, 1620 }, 1621 #endif 1622 #ifdef CONFIG_MMU 1623 #ifdef CONFIG_FILE_LOCKING 1624 { 1625 .procname = "lease-break-time", 1626 .data = &lease_break_time, 1627 .maxlen = sizeof(int), 1628 .mode = 0644, 1629 .proc_handler = proc_dointvec, 1630 }, 1631 #endif 1632 #ifdef CONFIG_AIO 1633 { 1634 .procname = "aio-nr", 1635 .data = &aio_nr, 1636 .maxlen = sizeof(aio_nr), 1637 .mode = 0444, 1638 .proc_handler = proc_doulongvec_minmax, 1639 }, 1640 { 1641 .procname = "aio-max-nr", 1642 .data = &aio_max_nr, 1643 .maxlen = sizeof(aio_max_nr), 1644 .mode = 0644, 1645 .proc_handler = proc_doulongvec_minmax, 1646 }, 1647 #endif /* CONFIG_AIO */ 1648 #ifdef CONFIG_INOTIFY_USER 1649 { 1650 .procname = "inotify", 1651 .mode = 0555, 1652 .child = inotify_table, 1653 }, 1654 #endif 1655 #ifdef CONFIG_EPOLL 1656 { 1657 .procname = "epoll", 1658 .mode = 0555, 1659 .child = epoll_table, 1660 }, 1661 #endif 1662 #endif 1663 { 1664 .procname = "protected_symlinks", 1665 .data = &sysctl_protected_symlinks, 1666 .maxlen = sizeof(int), 1667 .mode = 0600, 1668 .proc_handler = proc_dointvec_minmax, 1669 .extra1 = &zero, 1670 .extra2 = &one, 1671 }, 1672 { 1673 .procname = "protected_hardlinks", 1674 .data = &sysctl_protected_hardlinks, 1675 .maxlen = sizeof(int), 1676 .mode = 0600, 1677 .proc_handler = proc_dointvec_minmax, 1678 .extra1 = &zero, 1679 .extra2 = &one, 1680 }, 1681 { 1682 .procname = "suid_dumpable", 1683 .data = &suid_dumpable, 1684 .maxlen = sizeof(int), 1685 .mode = 0644, 1686 .proc_handler = proc_dointvec_minmax_coredump, 1687 .extra1 = &zero, 1688 .extra2 = &two, 1689 }, 1690 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE) 1691 { 1692 .procname = "binfmt_misc", 1693 .mode = 0555, 1694 .child = sysctl_mount_point, 1695 }, 1696 #endif 1697 { 1698 .procname = "pipe-max-size", 1699 .data = &pipe_max_size, 1700 .maxlen = sizeof(int), 1701 .mode = 0644, 1702 .proc_handler = &pipe_proc_fn, 1703 .extra1 = &pipe_min_size, 1704 }, 1705 { } 1706 }; 1707 1708 static struct ctl_table debug_table[] = { 1709 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE 1710 { 1711 .procname = "exception-trace", 1712 .data = &show_unhandled_signals, 1713 .maxlen = sizeof(int), 1714 .mode = 0644, 1715 .proc_handler = proc_dointvec 1716 }, 1717 #endif 1718 #if defined(CONFIG_OPTPROBES) 1719 { 1720 .procname = "kprobes-optimization", 1721 .data = &sysctl_kprobes_optimization, 1722 .maxlen = sizeof(int), 1723 .mode = 0644, 1724 .proc_handler = proc_kprobes_optimization_handler, 1725 .extra1 = &zero, 1726 .extra2 = &one, 1727 }, 1728 #endif 1729 { } 1730 }; 1731 1732 static struct ctl_table dev_table[] = { 1733 { } 1734 }; 1735 1736 int __init sysctl_init(void) 1737 { 1738 struct ctl_table_header *hdr; 1739 1740 hdr = register_sysctl_table(sysctl_base_table); 1741 kmemleak_not_leak(hdr); 1742 return 0; 1743 } 1744 1745 #endif /* CONFIG_SYSCTL */ 1746 1747 /* 1748 * /proc/sys support 1749 */ 1750 1751 #ifdef CONFIG_PROC_SYSCTL 1752 1753 static int _proc_do_string(char *data, int maxlen, int write, 1754 char __user *buffer, 1755 size_t *lenp, loff_t *ppos) 1756 { 1757 size_t len; 1758 char __user *p; 1759 char c; 1760 1761 if (!data || !maxlen || !*lenp) { 1762 *lenp = 0; 1763 return 0; 1764 } 1765 1766 if (write) { 1767 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) { 1768 /* Only continue writes not past the end of buffer. */ 1769 len = strlen(data); 1770 if (len > maxlen - 1) 1771 len = maxlen - 1; 1772 1773 if (*ppos > len) 1774 return 0; 1775 len = *ppos; 1776 } else { 1777 /* Start writing from beginning of buffer. */ 1778 len = 0; 1779 } 1780 1781 *ppos += *lenp; 1782 p = buffer; 1783 while ((p - buffer) < *lenp && len < maxlen - 1) { 1784 if (get_user(c, p++)) 1785 return -EFAULT; 1786 if (c == 0 || c == '\n') 1787 break; 1788 data[len++] = c; 1789 } 1790 data[len] = 0; 1791 } else { 1792 len = strlen(data); 1793 if (len > maxlen) 1794 len = maxlen; 1795 1796 if (*ppos > len) { 1797 *lenp = 0; 1798 return 0; 1799 } 1800 1801 data += *ppos; 1802 len -= *ppos; 1803 1804 if (len > *lenp) 1805 len = *lenp; 1806 if (len) 1807 if (copy_to_user(buffer, data, len)) 1808 return -EFAULT; 1809 if (len < *lenp) { 1810 if (put_user('\n', buffer + len)) 1811 return -EFAULT; 1812 len++; 1813 } 1814 *lenp = len; 1815 *ppos += len; 1816 } 1817 return 0; 1818 } 1819 1820 static void warn_sysctl_write(struct ctl_table *table) 1821 { 1822 pr_warn_once("%s wrote to %s when file position was not 0!\n" 1823 "This will not be supported in the future. To silence this\n" 1824 "warning, set kernel.sysctl_writes_strict = -1\n", 1825 current->comm, table->procname); 1826 } 1827 1828 /** 1829 * proc_dostring - read a string sysctl 1830 * @table: the sysctl table 1831 * @write: %TRUE if this is a write to the sysctl file 1832 * @buffer: the user buffer 1833 * @lenp: the size of the user buffer 1834 * @ppos: file position 1835 * 1836 * Reads/writes a string from/to the user buffer. If the kernel 1837 * buffer provided is not large enough to hold the string, the 1838 * string is truncated. The copied string is %NULL-terminated. 1839 * If the string is being read by the user process, it is copied 1840 * and a newline '\n' is added. It is truncated if the buffer is 1841 * not large enough. 1842 * 1843 * Returns 0 on success. 1844 */ 1845 int proc_dostring(struct ctl_table *table, int write, 1846 void __user *buffer, size_t *lenp, loff_t *ppos) 1847 { 1848 if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN) 1849 warn_sysctl_write(table); 1850 1851 return _proc_do_string((char *)(table->data), table->maxlen, write, 1852 (char __user *)buffer, lenp, ppos); 1853 } 1854 1855 static size_t proc_skip_spaces(char **buf) 1856 { 1857 size_t ret; 1858 char *tmp = skip_spaces(*buf); 1859 ret = tmp - *buf; 1860 *buf = tmp; 1861 return ret; 1862 } 1863 1864 static void proc_skip_char(char **buf, size_t *size, const char v) 1865 { 1866 while (*size) { 1867 if (**buf != v) 1868 break; 1869 (*size)--; 1870 (*buf)++; 1871 } 1872 } 1873 1874 #define TMPBUFLEN 22 1875 /** 1876 * proc_get_long - reads an ASCII formatted integer from a user buffer 1877 * 1878 * @buf: a kernel buffer 1879 * @size: size of the kernel buffer 1880 * @val: this is where the number will be stored 1881 * @neg: set to %TRUE if number is negative 1882 * @perm_tr: a vector which contains the allowed trailers 1883 * @perm_tr_len: size of the perm_tr vector 1884 * @tr: pointer to store the trailer character 1885 * 1886 * In case of success %0 is returned and @buf and @size are updated with 1887 * the amount of bytes read. If @tr is non-NULL and a trailing 1888 * character exists (size is non-zero after returning from this 1889 * function), @tr is updated with the trailing character. 1890 */ 1891 static int proc_get_long(char **buf, size_t *size, 1892 unsigned long *val, bool *neg, 1893 const char *perm_tr, unsigned perm_tr_len, char *tr) 1894 { 1895 int len; 1896 char *p, tmp[TMPBUFLEN]; 1897 1898 if (!*size) 1899 return -EINVAL; 1900 1901 len = *size; 1902 if (len > TMPBUFLEN - 1) 1903 len = TMPBUFLEN - 1; 1904 1905 memcpy(tmp, *buf, len); 1906 1907 tmp[len] = 0; 1908 p = tmp; 1909 if (*p == '-' && *size > 1) { 1910 *neg = true; 1911 p++; 1912 } else 1913 *neg = false; 1914 if (!isdigit(*p)) 1915 return -EINVAL; 1916 1917 *val = simple_strtoul(p, &p, 0); 1918 1919 len = p - tmp; 1920 1921 /* We don't know if the next char is whitespace thus we may accept 1922 * invalid integers (e.g. 1234...a) or two integers instead of one 1923 * (e.g. 123...1). So lets not allow such large numbers. */ 1924 if (len == TMPBUFLEN - 1) 1925 return -EINVAL; 1926 1927 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len)) 1928 return -EINVAL; 1929 1930 if (tr && (len < *size)) 1931 *tr = *p; 1932 1933 *buf += len; 1934 *size -= len; 1935 1936 return 0; 1937 } 1938 1939 /** 1940 * proc_put_long - converts an integer to a decimal ASCII formatted string 1941 * 1942 * @buf: the user buffer 1943 * @size: the size of the user buffer 1944 * @val: the integer to be converted 1945 * @neg: sign of the number, %TRUE for negative 1946 * 1947 * In case of success %0 is returned and @buf and @size are updated with 1948 * the amount of bytes written. 1949 */ 1950 static int proc_put_long(void __user **buf, size_t *size, unsigned long val, 1951 bool neg) 1952 { 1953 int len; 1954 char tmp[TMPBUFLEN], *p = tmp; 1955 1956 sprintf(p, "%s%lu", neg ? "-" : "", val); 1957 len = strlen(tmp); 1958 if (len > *size) 1959 len = *size; 1960 if (copy_to_user(*buf, tmp, len)) 1961 return -EFAULT; 1962 *size -= len; 1963 *buf += len; 1964 return 0; 1965 } 1966 #undef TMPBUFLEN 1967 1968 static int proc_put_char(void __user **buf, size_t *size, char c) 1969 { 1970 if (*size) { 1971 char __user **buffer = (char __user **)buf; 1972 if (put_user(c, *buffer)) 1973 return -EFAULT; 1974 (*size)--, (*buffer)++; 1975 *buf = *buffer; 1976 } 1977 return 0; 1978 } 1979 1980 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp, 1981 int *valp, 1982 int write, void *data) 1983 { 1984 if (write) { 1985 if (*negp) { 1986 if (*lvalp > (unsigned long) INT_MAX + 1) 1987 return -EINVAL; 1988 *valp = -*lvalp; 1989 } else { 1990 if (*lvalp > (unsigned long) INT_MAX) 1991 return -EINVAL; 1992 *valp = *lvalp; 1993 } 1994 } else { 1995 int val = *valp; 1996 if (val < 0) { 1997 *negp = true; 1998 *lvalp = -(unsigned long)val; 1999 } else { 2000 *negp = false; 2001 *lvalp = (unsigned long)val; 2002 } 2003 } 2004 return 0; 2005 } 2006 2007 static const char proc_wspace_sep[] = { ' ', '\t', '\n' }; 2008 2009 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table, 2010 int write, void __user *buffer, 2011 size_t *lenp, loff_t *ppos, 2012 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 2013 int write, void *data), 2014 void *data) 2015 { 2016 int *i, vleft, first = 1, err = 0; 2017 unsigned long page = 0; 2018 size_t left; 2019 char *kbuf; 2020 2021 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) { 2022 *lenp = 0; 2023 return 0; 2024 } 2025 2026 i = (int *) tbl_data; 2027 vleft = table->maxlen / sizeof(*i); 2028 left = *lenp; 2029 2030 if (!conv) 2031 conv = do_proc_dointvec_conv; 2032 2033 if (write) { 2034 if (*ppos) { 2035 switch (sysctl_writes_strict) { 2036 case SYSCTL_WRITES_STRICT: 2037 goto out; 2038 case SYSCTL_WRITES_WARN: 2039 warn_sysctl_write(table); 2040 break; 2041 default: 2042 break; 2043 } 2044 } 2045 2046 if (left > PAGE_SIZE - 1) 2047 left = PAGE_SIZE - 1; 2048 page = __get_free_page(GFP_TEMPORARY); 2049 kbuf = (char *) page; 2050 if (!kbuf) 2051 return -ENOMEM; 2052 if (copy_from_user(kbuf, buffer, left)) { 2053 err = -EFAULT; 2054 goto free; 2055 } 2056 kbuf[left] = 0; 2057 } 2058 2059 for (; left && vleft--; i++, first=0) { 2060 unsigned long lval; 2061 bool neg; 2062 2063 if (write) { 2064 left -= proc_skip_spaces(&kbuf); 2065 2066 if (!left) 2067 break; 2068 err = proc_get_long(&kbuf, &left, &lval, &neg, 2069 proc_wspace_sep, 2070 sizeof(proc_wspace_sep), NULL); 2071 if (err) 2072 break; 2073 if (conv(&neg, &lval, i, 1, data)) { 2074 err = -EINVAL; 2075 break; 2076 } 2077 } else { 2078 if (conv(&neg, &lval, i, 0, data)) { 2079 err = -EINVAL; 2080 break; 2081 } 2082 if (!first) 2083 err = proc_put_char(&buffer, &left, '\t'); 2084 if (err) 2085 break; 2086 err = proc_put_long(&buffer, &left, lval, neg); 2087 if (err) 2088 break; 2089 } 2090 } 2091 2092 if (!write && !first && left && !err) 2093 err = proc_put_char(&buffer, &left, '\n'); 2094 if (write && !err && left) 2095 left -= proc_skip_spaces(&kbuf); 2096 free: 2097 if (write) { 2098 free_page(page); 2099 if (first) 2100 return err ? : -EINVAL; 2101 } 2102 *lenp -= left; 2103 out: 2104 *ppos += *lenp; 2105 return err; 2106 } 2107 2108 static int do_proc_dointvec(struct ctl_table *table, int write, 2109 void __user *buffer, size_t *lenp, loff_t *ppos, 2110 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 2111 int write, void *data), 2112 void *data) 2113 { 2114 return __do_proc_dointvec(table->data, table, write, 2115 buffer, lenp, ppos, conv, data); 2116 } 2117 2118 /** 2119 * proc_dointvec - read a vector of integers 2120 * @table: the sysctl table 2121 * @write: %TRUE if this is a write to the sysctl file 2122 * @buffer: the user buffer 2123 * @lenp: the size of the user buffer 2124 * @ppos: file position 2125 * 2126 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2127 * values from/to the user buffer, treated as an ASCII string. 2128 * 2129 * Returns 0 on success. 2130 */ 2131 int proc_dointvec(struct ctl_table *table, int write, 2132 void __user *buffer, size_t *lenp, loff_t *ppos) 2133 { 2134 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2135 NULL,NULL); 2136 } 2137 2138 /* 2139 * Taint values can only be increased 2140 * This means we can safely use a temporary. 2141 */ 2142 static int proc_taint(struct ctl_table *table, int write, 2143 void __user *buffer, size_t *lenp, loff_t *ppos) 2144 { 2145 struct ctl_table t; 2146 unsigned long tmptaint = get_taint(); 2147 int err; 2148 2149 if (write && !capable(CAP_SYS_ADMIN)) 2150 return -EPERM; 2151 2152 t = *table; 2153 t.data = &tmptaint; 2154 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos); 2155 if (err < 0) 2156 return err; 2157 2158 if (write) { 2159 /* 2160 * Poor man's atomic or. Not worth adding a primitive 2161 * to everyone's atomic.h for this 2162 */ 2163 int i; 2164 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) { 2165 if ((tmptaint >> i) & 1) 2166 add_taint(i, LOCKDEP_STILL_OK); 2167 } 2168 } 2169 2170 return err; 2171 } 2172 2173 #ifdef CONFIG_PRINTK 2174 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write, 2175 void __user *buffer, size_t *lenp, loff_t *ppos) 2176 { 2177 if (write && !capable(CAP_SYS_ADMIN)) 2178 return -EPERM; 2179 2180 return proc_dointvec_minmax(table, write, buffer, lenp, ppos); 2181 } 2182 #endif 2183 2184 struct do_proc_dointvec_minmax_conv_param { 2185 int *min; 2186 int *max; 2187 }; 2188 2189 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp, 2190 int *valp, 2191 int write, void *data) 2192 { 2193 struct do_proc_dointvec_minmax_conv_param *param = data; 2194 if (write) { 2195 int val = *negp ? -*lvalp : *lvalp; 2196 if ((param->min && *param->min > val) || 2197 (param->max && *param->max < val)) 2198 return -EINVAL; 2199 *valp = val; 2200 } else { 2201 int val = *valp; 2202 if (val < 0) { 2203 *negp = true; 2204 *lvalp = -(unsigned long)val; 2205 } else { 2206 *negp = false; 2207 *lvalp = (unsigned long)val; 2208 } 2209 } 2210 return 0; 2211 } 2212 2213 /** 2214 * proc_dointvec_minmax - read a vector of integers with min/max values 2215 * @table: the sysctl table 2216 * @write: %TRUE if this is a write to the sysctl file 2217 * @buffer: the user buffer 2218 * @lenp: the size of the user buffer 2219 * @ppos: file position 2220 * 2221 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2222 * values from/to the user buffer, treated as an ASCII string. 2223 * 2224 * This routine will ensure the values are within the range specified by 2225 * table->extra1 (min) and table->extra2 (max). 2226 * 2227 * Returns 0 on success. 2228 */ 2229 int proc_dointvec_minmax(struct ctl_table *table, int write, 2230 void __user *buffer, size_t *lenp, loff_t *ppos) 2231 { 2232 struct do_proc_dointvec_minmax_conv_param param = { 2233 .min = (int *) table->extra1, 2234 .max = (int *) table->extra2, 2235 }; 2236 return do_proc_dointvec(table, write, buffer, lenp, ppos, 2237 do_proc_dointvec_minmax_conv, ¶m); 2238 } 2239 2240 static void validate_coredump_safety(void) 2241 { 2242 #ifdef CONFIG_COREDUMP 2243 if (suid_dumpable == SUID_DUMP_ROOT && 2244 core_pattern[0] != '/' && core_pattern[0] != '|') { 2245 printk(KERN_WARNING "Unsafe core_pattern used with "\ 2246 "suid_dumpable=2. Pipe handler or fully qualified "\ 2247 "core dump path required.\n"); 2248 } 2249 #endif 2250 } 2251 2252 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write, 2253 void __user *buffer, size_t *lenp, loff_t *ppos) 2254 { 2255 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos); 2256 if (!error) 2257 validate_coredump_safety(); 2258 return error; 2259 } 2260 2261 #ifdef CONFIG_COREDUMP 2262 static int proc_dostring_coredump(struct ctl_table *table, int write, 2263 void __user *buffer, size_t *lenp, loff_t *ppos) 2264 { 2265 int error = proc_dostring(table, write, buffer, lenp, ppos); 2266 if (!error) 2267 validate_coredump_safety(); 2268 return error; 2269 } 2270 #endif 2271 2272 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write, 2273 void __user *buffer, 2274 size_t *lenp, loff_t *ppos, 2275 unsigned long convmul, 2276 unsigned long convdiv) 2277 { 2278 unsigned long *i, *min, *max; 2279 int vleft, first = 1, err = 0; 2280 unsigned long page = 0; 2281 size_t left; 2282 char *kbuf; 2283 2284 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) { 2285 *lenp = 0; 2286 return 0; 2287 } 2288 2289 i = (unsigned long *) data; 2290 min = (unsigned long *) table->extra1; 2291 max = (unsigned long *) table->extra2; 2292 vleft = table->maxlen / sizeof(unsigned long); 2293 left = *lenp; 2294 2295 if (write) { 2296 if (*ppos) { 2297 switch (sysctl_writes_strict) { 2298 case SYSCTL_WRITES_STRICT: 2299 goto out; 2300 case SYSCTL_WRITES_WARN: 2301 warn_sysctl_write(table); 2302 break; 2303 default: 2304 break; 2305 } 2306 } 2307 2308 if (left > PAGE_SIZE - 1) 2309 left = PAGE_SIZE - 1; 2310 page = __get_free_page(GFP_TEMPORARY); 2311 kbuf = (char *) page; 2312 if (!kbuf) 2313 return -ENOMEM; 2314 if (copy_from_user(kbuf, buffer, left)) { 2315 err = -EFAULT; 2316 goto free; 2317 } 2318 kbuf[left] = 0; 2319 } 2320 2321 for (; left && vleft--; i++, first = 0) { 2322 unsigned long val; 2323 2324 if (write) { 2325 bool neg; 2326 2327 left -= proc_skip_spaces(&kbuf); 2328 2329 err = proc_get_long(&kbuf, &left, &val, &neg, 2330 proc_wspace_sep, 2331 sizeof(proc_wspace_sep), NULL); 2332 if (err) 2333 break; 2334 if (neg) 2335 continue; 2336 if ((min && val < *min) || (max && val > *max)) 2337 continue; 2338 *i = val; 2339 } else { 2340 val = convdiv * (*i) / convmul; 2341 if (!first) { 2342 err = proc_put_char(&buffer, &left, '\t'); 2343 if (err) 2344 break; 2345 } 2346 err = proc_put_long(&buffer, &left, val, false); 2347 if (err) 2348 break; 2349 } 2350 } 2351 2352 if (!write && !first && left && !err) 2353 err = proc_put_char(&buffer, &left, '\n'); 2354 if (write && !err) 2355 left -= proc_skip_spaces(&kbuf); 2356 free: 2357 if (write) { 2358 free_page(page); 2359 if (first) 2360 return err ? : -EINVAL; 2361 } 2362 *lenp -= left; 2363 out: 2364 *ppos += *lenp; 2365 return err; 2366 } 2367 2368 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write, 2369 void __user *buffer, 2370 size_t *lenp, loff_t *ppos, 2371 unsigned long convmul, 2372 unsigned long convdiv) 2373 { 2374 return __do_proc_doulongvec_minmax(table->data, table, write, 2375 buffer, lenp, ppos, convmul, convdiv); 2376 } 2377 2378 /** 2379 * proc_doulongvec_minmax - read a vector of long integers with min/max values 2380 * @table: the sysctl table 2381 * @write: %TRUE if this is a write to the sysctl file 2382 * @buffer: the user buffer 2383 * @lenp: the size of the user buffer 2384 * @ppos: file position 2385 * 2386 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 2387 * values from/to the user buffer, treated as an ASCII string. 2388 * 2389 * This routine will ensure the values are within the range specified by 2390 * table->extra1 (min) and table->extra2 (max). 2391 * 2392 * Returns 0 on success. 2393 */ 2394 int proc_doulongvec_minmax(struct ctl_table *table, int write, 2395 void __user *buffer, size_t *lenp, loff_t *ppos) 2396 { 2397 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l); 2398 } 2399 2400 /** 2401 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values 2402 * @table: the sysctl table 2403 * @write: %TRUE if this is a write to the sysctl file 2404 * @buffer: the user buffer 2405 * @lenp: the size of the user buffer 2406 * @ppos: file position 2407 * 2408 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 2409 * values from/to the user buffer, treated as an ASCII string. The values 2410 * are treated as milliseconds, and converted to jiffies when they are stored. 2411 * 2412 * This routine will ensure the values are within the range specified by 2413 * table->extra1 (min) and table->extra2 (max). 2414 * 2415 * Returns 0 on success. 2416 */ 2417 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 2418 void __user *buffer, 2419 size_t *lenp, loff_t *ppos) 2420 { 2421 return do_proc_doulongvec_minmax(table, write, buffer, 2422 lenp, ppos, HZ, 1000l); 2423 } 2424 2425 2426 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp, 2427 int *valp, 2428 int write, void *data) 2429 { 2430 if (write) { 2431 if (*lvalp > LONG_MAX / HZ) 2432 return 1; 2433 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ); 2434 } else { 2435 int val = *valp; 2436 unsigned long lval; 2437 if (val < 0) { 2438 *negp = true; 2439 lval = -(unsigned long)val; 2440 } else { 2441 *negp = false; 2442 lval = (unsigned long)val; 2443 } 2444 *lvalp = lval / HZ; 2445 } 2446 return 0; 2447 } 2448 2449 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp, 2450 int *valp, 2451 int write, void *data) 2452 { 2453 if (write) { 2454 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ) 2455 return 1; 2456 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp); 2457 } else { 2458 int val = *valp; 2459 unsigned long lval; 2460 if (val < 0) { 2461 *negp = true; 2462 lval = -(unsigned long)val; 2463 } else { 2464 *negp = false; 2465 lval = (unsigned long)val; 2466 } 2467 *lvalp = jiffies_to_clock_t(lval); 2468 } 2469 return 0; 2470 } 2471 2472 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp, 2473 int *valp, 2474 int write, void *data) 2475 { 2476 if (write) { 2477 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp); 2478 2479 if (jif > INT_MAX) 2480 return 1; 2481 *valp = (int)jif; 2482 } else { 2483 int val = *valp; 2484 unsigned long lval; 2485 if (val < 0) { 2486 *negp = true; 2487 lval = -(unsigned long)val; 2488 } else { 2489 *negp = false; 2490 lval = (unsigned long)val; 2491 } 2492 *lvalp = jiffies_to_msecs(lval); 2493 } 2494 return 0; 2495 } 2496 2497 /** 2498 * proc_dointvec_jiffies - read a vector of integers as seconds 2499 * @table: the sysctl table 2500 * @write: %TRUE if this is a write to the sysctl file 2501 * @buffer: the user buffer 2502 * @lenp: the size of the user buffer 2503 * @ppos: file position 2504 * 2505 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2506 * values from/to the user buffer, treated as an ASCII string. 2507 * The values read are assumed to be in seconds, and are converted into 2508 * jiffies. 2509 * 2510 * Returns 0 on success. 2511 */ 2512 int proc_dointvec_jiffies(struct ctl_table *table, int write, 2513 void __user *buffer, size_t *lenp, loff_t *ppos) 2514 { 2515 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2516 do_proc_dointvec_jiffies_conv,NULL); 2517 } 2518 2519 /** 2520 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds 2521 * @table: the sysctl table 2522 * @write: %TRUE if this is a write to the sysctl file 2523 * @buffer: the user buffer 2524 * @lenp: the size of the user buffer 2525 * @ppos: pointer to the file position 2526 * 2527 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2528 * values from/to the user buffer, treated as an ASCII string. 2529 * The values read are assumed to be in 1/USER_HZ seconds, and 2530 * are converted into jiffies. 2531 * 2532 * Returns 0 on success. 2533 */ 2534 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 2535 void __user *buffer, size_t *lenp, loff_t *ppos) 2536 { 2537 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2538 do_proc_dointvec_userhz_jiffies_conv,NULL); 2539 } 2540 2541 /** 2542 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds 2543 * @table: the sysctl table 2544 * @write: %TRUE if this is a write to the sysctl file 2545 * @buffer: the user buffer 2546 * @lenp: the size of the user buffer 2547 * @ppos: file position 2548 * @ppos: the current position in the file 2549 * 2550 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2551 * values from/to the user buffer, treated as an ASCII string. 2552 * The values read are assumed to be in 1/1000 seconds, and 2553 * are converted into jiffies. 2554 * 2555 * Returns 0 on success. 2556 */ 2557 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, 2558 void __user *buffer, size_t *lenp, loff_t *ppos) 2559 { 2560 return do_proc_dointvec(table, write, buffer, lenp, ppos, 2561 do_proc_dointvec_ms_jiffies_conv, NULL); 2562 } 2563 2564 static int proc_do_cad_pid(struct ctl_table *table, int write, 2565 void __user *buffer, size_t *lenp, loff_t *ppos) 2566 { 2567 struct pid *new_pid; 2568 pid_t tmp; 2569 int r; 2570 2571 tmp = pid_vnr(cad_pid); 2572 2573 r = __do_proc_dointvec(&tmp, table, write, buffer, 2574 lenp, ppos, NULL, NULL); 2575 if (r || !write) 2576 return r; 2577 2578 new_pid = find_get_pid(tmp); 2579 if (!new_pid) 2580 return -ESRCH; 2581 2582 put_pid(xchg(&cad_pid, new_pid)); 2583 return 0; 2584 } 2585 2586 /** 2587 * proc_do_large_bitmap - read/write from/to a large bitmap 2588 * @table: the sysctl table 2589 * @write: %TRUE if this is a write to the sysctl file 2590 * @buffer: the user buffer 2591 * @lenp: the size of the user buffer 2592 * @ppos: file position 2593 * 2594 * The bitmap is stored at table->data and the bitmap length (in bits) 2595 * in table->maxlen. 2596 * 2597 * We use a range comma separated format (e.g. 1,3-4,10-10) so that 2598 * large bitmaps may be represented in a compact manner. Writing into 2599 * the file will clear the bitmap then update it with the given input. 2600 * 2601 * Returns 0 on success. 2602 */ 2603 int proc_do_large_bitmap(struct ctl_table *table, int write, 2604 void __user *buffer, size_t *lenp, loff_t *ppos) 2605 { 2606 int err = 0; 2607 bool first = 1; 2608 size_t left = *lenp; 2609 unsigned long bitmap_len = table->maxlen; 2610 unsigned long *bitmap = *(unsigned long **) table->data; 2611 unsigned long *tmp_bitmap = NULL; 2612 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c; 2613 2614 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) { 2615 *lenp = 0; 2616 return 0; 2617 } 2618 2619 if (write) { 2620 unsigned long page = 0; 2621 char *kbuf; 2622 2623 if (left > PAGE_SIZE - 1) 2624 left = PAGE_SIZE - 1; 2625 2626 page = __get_free_page(GFP_TEMPORARY); 2627 kbuf = (char *) page; 2628 if (!kbuf) 2629 return -ENOMEM; 2630 if (copy_from_user(kbuf, buffer, left)) { 2631 free_page(page); 2632 return -EFAULT; 2633 } 2634 kbuf[left] = 0; 2635 2636 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long), 2637 GFP_KERNEL); 2638 if (!tmp_bitmap) { 2639 free_page(page); 2640 return -ENOMEM; 2641 } 2642 proc_skip_char(&kbuf, &left, '\n'); 2643 while (!err && left) { 2644 unsigned long val_a, val_b; 2645 bool neg; 2646 2647 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a, 2648 sizeof(tr_a), &c); 2649 if (err) 2650 break; 2651 if (val_a >= bitmap_len || neg) { 2652 err = -EINVAL; 2653 break; 2654 } 2655 2656 val_b = val_a; 2657 if (left) { 2658 kbuf++; 2659 left--; 2660 } 2661 2662 if (c == '-') { 2663 err = proc_get_long(&kbuf, &left, &val_b, 2664 &neg, tr_b, sizeof(tr_b), 2665 &c); 2666 if (err) 2667 break; 2668 if (val_b >= bitmap_len || neg || 2669 val_a > val_b) { 2670 err = -EINVAL; 2671 break; 2672 } 2673 if (left) { 2674 kbuf++; 2675 left--; 2676 } 2677 } 2678 2679 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1); 2680 first = 0; 2681 proc_skip_char(&kbuf, &left, '\n'); 2682 } 2683 free_page(page); 2684 } else { 2685 unsigned long bit_a, bit_b = 0; 2686 2687 while (left) { 2688 bit_a = find_next_bit(bitmap, bitmap_len, bit_b); 2689 if (bit_a >= bitmap_len) 2690 break; 2691 bit_b = find_next_zero_bit(bitmap, bitmap_len, 2692 bit_a + 1) - 1; 2693 2694 if (!first) { 2695 err = proc_put_char(&buffer, &left, ','); 2696 if (err) 2697 break; 2698 } 2699 err = proc_put_long(&buffer, &left, bit_a, false); 2700 if (err) 2701 break; 2702 if (bit_a != bit_b) { 2703 err = proc_put_char(&buffer, &left, '-'); 2704 if (err) 2705 break; 2706 err = proc_put_long(&buffer, &left, bit_b, false); 2707 if (err) 2708 break; 2709 } 2710 2711 first = 0; bit_b++; 2712 } 2713 if (!err) 2714 err = proc_put_char(&buffer, &left, '\n'); 2715 } 2716 2717 if (!err) { 2718 if (write) { 2719 if (*ppos) 2720 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len); 2721 else 2722 bitmap_copy(bitmap, tmp_bitmap, bitmap_len); 2723 } 2724 kfree(tmp_bitmap); 2725 *lenp -= left; 2726 *ppos += *lenp; 2727 return 0; 2728 } else { 2729 kfree(tmp_bitmap); 2730 return err; 2731 } 2732 } 2733 2734 #else /* CONFIG_PROC_SYSCTL */ 2735 2736 int proc_dostring(struct ctl_table *table, int write, 2737 void __user *buffer, size_t *lenp, loff_t *ppos) 2738 { 2739 return -ENOSYS; 2740 } 2741 2742 int proc_dointvec(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_minmax(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_jiffies(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_userhz_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_ms_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_doulongvec_minmax(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_ms_jiffies_minmax(struct ctl_table *table, int write, 2779 void __user *buffer, 2780 size_t *lenp, loff_t *ppos) 2781 { 2782 return -ENOSYS; 2783 } 2784 2785 2786 #endif /* CONFIG_PROC_SYSCTL */ 2787 2788 /* 2789 * No sense putting this after each symbol definition, twice, 2790 * exception granted :-) 2791 */ 2792 EXPORT_SYMBOL(proc_dointvec); 2793 EXPORT_SYMBOL(proc_dointvec_jiffies); 2794 EXPORT_SYMBOL(proc_dointvec_minmax); 2795 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies); 2796 EXPORT_SYMBOL(proc_dointvec_ms_jiffies); 2797 EXPORT_SYMBOL(proc_dostring); 2798 EXPORT_SYMBOL(proc_doulongvec_minmax); 2799 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax); 2800