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 #include <linux/bpf.h> 68 69 #include <asm/uaccess.h> 70 #include <asm/processor.h> 71 72 #ifdef CONFIG_X86 73 #include <asm/nmi.h> 74 #include <asm/stacktrace.h> 75 #include <asm/io.h> 76 #endif 77 #ifdef CONFIG_SPARC 78 #include <asm/setup.h> 79 #endif 80 #ifdef CONFIG_BSD_PROCESS_ACCT 81 #include <linux/acct.h> 82 #endif 83 #ifdef CONFIG_RT_MUTEXES 84 #include <linux/rtmutex.h> 85 #endif 86 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT) 87 #include <linux/lockdep.h> 88 #endif 89 #ifdef CONFIG_CHR_DEV_SG 90 #include <scsi/sg.h> 91 #endif 92 93 #ifdef CONFIG_LOCKUP_DETECTOR 94 #include <linux/nmi.h> 95 #endif 96 97 #if defined(CONFIG_SYSCTL) 98 99 /* External variables not in a header file. */ 100 extern int suid_dumpable; 101 #ifdef CONFIG_COREDUMP 102 extern int core_uses_pid; 103 extern char core_pattern[]; 104 extern unsigned int core_pipe_limit; 105 #endif 106 extern int pid_max; 107 extern int pid_max_min, pid_max_max; 108 extern int percpu_pagelist_fraction; 109 extern int compat_log; 110 extern int latencytop_enabled; 111 extern int sysctl_nr_open_min, sysctl_nr_open_max; 112 #ifndef CONFIG_MMU 113 extern int sysctl_nr_trim_pages; 114 #endif 115 116 /* Constants used for minimum and maximum */ 117 #ifdef CONFIG_LOCKUP_DETECTOR 118 static int sixty = 60; 119 #endif 120 121 static int __maybe_unused neg_one = -1; 122 123 static int zero; 124 static int __maybe_unused one = 1; 125 static int __maybe_unused two = 2; 126 static int __maybe_unused four = 4; 127 static unsigned long one_ul = 1; 128 static int one_hundred = 100; 129 #ifdef CONFIG_PRINTK 130 static int ten_thousand = 10000; 131 #endif 132 133 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */ 134 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE; 135 136 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */ 137 static int maxolduid = 65535; 138 static int minolduid; 139 140 static int ngroups_max = NGROUPS_MAX; 141 static const int cap_last_cap = CAP_LAST_CAP; 142 143 /*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */ 144 #ifdef CONFIG_DETECT_HUNG_TASK 145 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ); 146 #endif 147 148 #ifdef CONFIG_INOTIFY_USER 149 #include <linux/inotify.h> 150 #endif 151 #ifdef CONFIG_SPARC 152 #endif 153 154 #ifdef __hppa__ 155 extern int pwrsw_enabled; 156 #endif 157 158 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW 159 extern int unaligned_enabled; 160 #endif 161 162 #ifdef CONFIG_IA64 163 extern int unaligned_dump_stack; 164 #endif 165 166 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN 167 extern int no_unaligned_warning; 168 #endif 169 170 #ifdef CONFIG_PROC_SYSCTL 171 172 #define SYSCTL_WRITES_LEGACY -1 173 #define SYSCTL_WRITES_WARN 0 174 #define SYSCTL_WRITES_STRICT 1 175 176 static int sysctl_writes_strict = SYSCTL_WRITES_WARN; 177 178 static int proc_do_cad_pid(struct ctl_table *table, int write, 179 void __user *buffer, size_t *lenp, loff_t *ppos); 180 static int proc_taint(struct ctl_table *table, int write, 181 void __user *buffer, size_t *lenp, loff_t *ppos); 182 #endif 183 184 #ifdef CONFIG_PRINTK 185 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write, 186 void __user *buffer, size_t *lenp, loff_t *ppos); 187 #endif 188 189 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write, 190 void __user *buffer, size_t *lenp, loff_t *ppos); 191 #ifdef CONFIG_COREDUMP 192 static int proc_dostring_coredump(struct ctl_table *table, int write, 193 void __user *buffer, size_t *lenp, loff_t *ppos); 194 #endif 195 196 #ifdef CONFIG_MAGIC_SYSRQ 197 /* Note: sysrq code uses it's own private copy */ 198 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE; 199 200 static int sysrq_sysctl_handler(struct ctl_table *table, int write, 201 void __user *buffer, size_t *lenp, 202 loff_t *ppos) 203 { 204 int error; 205 206 error = proc_dointvec(table, write, buffer, lenp, ppos); 207 if (error) 208 return error; 209 210 if (write) 211 sysrq_toggle_support(__sysrq_enabled); 212 213 return 0; 214 } 215 216 #endif 217 218 static struct ctl_table kern_table[]; 219 static struct ctl_table vm_table[]; 220 static struct ctl_table fs_table[]; 221 static struct ctl_table debug_table[]; 222 static struct ctl_table dev_table[]; 223 extern struct ctl_table random_table[]; 224 #ifdef CONFIG_EPOLL 225 extern struct ctl_table epoll_table[]; 226 #endif 227 228 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT 229 int sysctl_legacy_va_layout; 230 #endif 231 232 /* The default sysctl tables: */ 233 234 static struct ctl_table sysctl_base_table[] = { 235 { 236 .procname = "kernel", 237 .mode = 0555, 238 .child = kern_table, 239 }, 240 { 241 .procname = "vm", 242 .mode = 0555, 243 .child = vm_table, 244 }, 245 { 246 .procname = "fs", 247 .mode = 0555, 248 .child = fs_table, 249 }, 250 { 251 .procname = "debug", 252 .mode = 0555, 253 .child = debug_table, 254 }, 255 { 256 .procname = "dev", 257 .mode = 0555, 258 .child = dev_table, 259 }, 260 { } 261 }; 262 263 #ifdef CONFIG_SCHED_DEBUG 264 static int min_sched_granularity_ns = 100000; /* 100 usecs */ 265 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */ 266 static int min_wakeup_granularity_ns; /* 0 usecs */ 267 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */ 268 #ifdef CONFIG_SMP 269 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE; 270 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1; 271 #endif /* CONFIG_SMP */ 272 #endif /* CONFIG_SCHED_DEBUG */ 273 274 #ifdef CONFIG_COMPACTION 275 static int min_extfrag_threshold; 276 static int max_extfrag_threshold = 1000; 277 #endif 278 279 static struct ctl_table kern_table[] = { 280 { 281 .procname = "sched_child_runs_first", 282 .data = &sysctl_sched_child_runs_first, 283 .maxlen = sizeof(unsigned int), 284 .mode = 0644, 285 .proc_handler = proc_dointvec, 286 }, 287 #ifdef CONFIG_SCHED_DEBUG 288 { 289 .procname = "sched_min_granularity_ns", 290 .data = &sysctl_sched_min_granularity, 291 .maxlen = sizeof(unsigned int), 292 .mode = 0644, 293 .proc_handler = sched_proc_update_handler, 294 .extra1 = &min_sched_granularity_ns, 295 .extra2 = &max_sched_granularity_ns, 296 }, 297 { 298 .procname = "sched_latency_ns", 299 .data = &sysctl_sched_latency, 300 .maxlen = sizeof(unsigned int), 301 .mode = 0644, 302 .proc_handler = sched_proc_update_handler, 303 .extra1 = &min_sched_granularity_ns, 304 .extra2 = &max_sched_granularity_ns, 305 }, 306 { 307 .procname = "sched_wakeup_granularity_ns", 308 .data = &sysctl_sched_wakeup_granularity, 309 .maxlen = sizeof(unsigned int), 310 .mode = 0644, 311 .proc_handler = sched_proc_update_handler, 312 .extra1 = &min_wakeup_granularity_ns, 313 .extra2 = &max_wakeup_granularity_ns, 314 }, 315 #ifdef CONFIG_SMP 316 { 317 .procname = "sched_tunable_scaling", 318 .data = &sysctl_sched_tunable_scaling, 319 .maxlen = sizeof(enum sched_tunable_scaling), 320 .mode = 0644, 321 .proc_handler = sched_proc_update_handler, 322 .extra1 = &min_sched_tunable_scaling, 323 .extra2 = &max_sched_tunable_scaling, 324 }, 325 { 326 .procname = "sched_migration_cost_ns", 327 .data = &sysctl_sched_migration_cost, 328 .maxlen = sizeof(unsigned int), 329 .mode = 0644, 330 .proc_handler = proc_dointvec, 331 }, 332 { 333 .procname = "sched_nr_migrate", 334 .data = &sysctl_sched_nr_migrate, 335 .maxlen = sizeof(unsigned int), 336 .mode = 0644, 337 .proc_handler = proc_dointvec, 338 }, 339 { 340 .procname = "sched_time_avg_ms", 341 .data = &sysctl_sched_time_avg, 342 .maxlen = sizeof(unsigned int), 343 .mode = 0644, 344 .proc_handler = proc_dointvec, 345 }, 346 { 347 .procname = "sched_shares_window_ns", 348 .data = &sysctl_sched_shares_window, 349 .maxlen = sizeof(unsigned int), 350 .mode = 0644, 351 .proc_handler = proc_dointvec, 352 }, 353 #endif /* CONFIG_SMP */ 354 #ifdef CONFIG_NUMA_BALANCING 355 { 356 .procname = "numa_balancing_scan_delay_ms", 357 .data = &sysctl_numa_balancing_scan_delay, 358 .maxlen = sizeof(unsigned int), 359 .mode = 0644, 360 .proc_handler = proc_dointvec, 361 }, 362 { 363 .procname = "numa_balancing_scan_period_min_ms", 364 .data = &sysctl_numa_balancing_scan_period_min, 365 .maxlen = sizeof(unsigned int), 366 .mode = 0644, 367 .proc_handler = proc_dointvec, 368 }, 369 { 370 .procname = "numa_balancing_scan_period_max_ms", 371 .data = &sysctl_numa_balancing_scan_period_max, 372 .maxlen = sizeof(unsigned int), 373 .mode = 0644, 374 .proc_handler = proc_dointvec, 375 }, 376 { 377 .procname = "numa_balancing_scan_size_mb", 378 .data = &sysctl_numa_balancing_scan_size, 379 .maxlen = sizeof(unsigned int), 380 .mode = 0644, 381 .proc_handler = proc_dointvec_minmax, 382 .extra1 = &one, 383 }, 384 { 385 .procname = "numa_balancing", 386 .data = NULL, /* filled in by handler */ 387 .maxlen = sizeof(unsigned int), 388 .mode = 0644, 389 .proc_handler = sysctl_numa_balancing, 390 .extra1 = &zero, 391 .extra2 = &one, 392 }, 393 #endif /* CONFIG_NUMA_BALANCING */ 394 #endif /* CONFIG_SCHED_DEBUG */ 395 { 396 .procname = "sched_rt_period_us", 397 .data = &sysctl_sched_rt_period, 398 .maxlen = sizeof(unsigned int), 399 .mode = 0644, 400 .proc_handler = sched_rt_handler, 401 }, 402 { 403 .procname = "sched_rt_runtime_us", 404 .data = &sysctl_sched_rt_runtime, 405 .maxlen = sizeof(int), 406 .mode = 0644, 407 .proc_handler = sched_rt_handler, 408 }, 409 { 410 .procname = "sched_rr_timeslice_ms", 411 .data = &sched_rr_timeslice, 412 .maxlen = sizeof(int), 413 .mode = 0644, 414 .proc_handler = sched_rr_handler, 415 }, 416 #ifdef CONFIG_SCHED_AUTOGROUP 417 { 418 .procname = "sched_autogroup_enabled", 419 .data = &sysctl_sched_autogroup_enabled, 420 .maxlen = sizeof(unsigned int), 421 .mode = 0644, 422 .proc_handler = proc_dointvec_minmax, 423 .extra1 = &zero, 424 .extra2 = &one, 425 }, 426 #endif 427 #ifdef CONFIG_CFS_BANDWIDTH 428 { 429 .procname = "sched_cfs_bandwidth_slice_us", 430 .data = &sysctl_sched_cfs_bandwidth_slice, 431 .maxlen = sizeof(unsigned int), 432 .mode = 0644, 433 .proc_handler = proc_dointvec_minmax, 434 .extra1 = &one, 435 }, 436 #endif 437 #ifdef CONFIG_PROVE_LOCKING 438 { 439 .procname = "prove_locking", 440 .data = &prove_locking, 441 .maxlen = sizeof(int), 442 .mode = 0644, 443 .proc_handler = proc_dointvec, 444 }, 445 #endif 446 #ifdef CONFIG_LOCK_STAT 447 { 448 .procname = "lock_stat", 449 .data = &lock_stat, 450 .maxlen = sizeof(int), 451 .mode = 0644, 452 .proc_handler = proc_dointvec, 453 }, 454 #endif 455 { 456 .procname = "panic", 457 .data = &panic_timeout, 458 .maxlen = sizeof(int), 459 .mode = 0644, 460 .proc_handler = proc_dointvec, 461 }, 462 #ifdef CONFIG_COREDUMP 463 { 464 .procname = "core_uses_pid", 465 .data = &core_uses_pid, 466 .maxlen = sizeof(int), 467 .mode = 0644, 468 .proc_handler = proc_dointvec, 469 }, 470 { 471 .procname = "core_pattern", 472 .data = core_pattern, 473 .maxlen = CORENAME_MAX_SIZE, 474 .mode = 0644, 475 .proc_handler = proc_dostring_coredump, 476 }, 477 { 478 .procname = "core_pipe_limit", 479 .data = &core_pipe_limit, 480 .maxlen = sizeof(unsigned int), 481 .mode = 0644, 482 .proc_handler = proc_dointvec, 483 }, 484 #endif 485 #ifdef CONFIG_PROC_SYSCTL 486 { 487 .procname = "tainted", 488 .maxlen = sizeof(long), 489 .mode = 0644, 490 .proc_handler = proc_taint, 491 }, 492 { 493 .procname = "sysctl_writes_strict", 494 .data = &sysctl_writes_strict, 495 .maxlen = sizeof(int), 496 .mode = 0644, 497 .proc_handler = proc_dointvec_minmax, 498 .extra1 = &neg_one, 499 .extra2 = &one, 500 }, 501 #endif 502 #ifdef CONFIG_LATENCYTOP 503 { 504 .procname = "latencytop", 505 .data = &latencytop_enabled, 506 .maxlen = sizeof(int), 507 .mode = 0644, 508 .proc_handler = proc_dointvec, 509 }, 510 #endif 511 #ifdef CONFIG_BLK_DEV_INITRD 512 { 513 .procname = "real-root-dev", 514 .data = &real_root_dev, 515 .maxlen = sizeof(int), 516 .mode = 0644, 517 .proc_handler = proc_dointvec, 518 }, 519 #endif 520 { 521 .procname = "print-fatal-signals", 522 .data = &print_fatal_signals, 523 .maxlen = sizeof(int), 524 .mode = 0644, 525 .proc_handler = proc_dointvec, 526 }, 527 #ifdef CONFIG_SPARC 528 { 529 .procname = "reboot-cmd", 530 .data = reboot_command, 531 .maxlen = 256, 532 .mode = 0644, 533 .proc_handler = proc_dostring, 534 }, 535 { 536 .procname = "stop-a", 537 .data = &stop_a_enabled, 538 .maxlen = sizeof (int), 539 .mode = 0644, 540 .proc_handler = proc_dointvec, 541 }, 542 { 543 .procname = "scons-poweroff", 544 .data = &scons_pwroff, 545 .maxlen = sizeof (int), 546 .mode = 0644, 547 .proc_handler = proc_dointvec, 548 }, 549 #endif 550 #ifdef CONFIG_SPARC64 551 { 552 .procname = "tsb-ratio", 553 .data = &sysctl_tsb_ratio, 554 .maxlen = sizeof (int), 555 .mode = 0644, 556 .proc_handler = proc_dointvec, 557 }, 558 #endif 559 #ifdef __hppa__ 560 { 561 .procname = "soft-power", 562 .data = &pwrsw_enabled, 563 .maxlen = sizeof (int), 564 .mode = 0644, 565 .proc_handler = proc_dointvec, 566 }, 567 #endif 568 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW 569 { 570 .procname = "unaligned-trap", 571 .data = &unaligned_enabled, 572 .maxlen = sizeof (int), 573 .mode = 0644, 574 .proc_handler = proc_dointvec, 575 }, 576 #endif 577 { 578 .procname = "ctrl-alt-del", 579 .data = &C_A_D, 580 .maxlen = sizeof(int), 581 .mode = 0644, 582 .proc_handler = proc_dointvec, 583 }, 584 #ifdef CONFIG_FUNCTION_TRACER 585 { 586 .procname = "ftrace_enabled", 587 .data = &ftrace_enabled, 588 .maxlen = sizeof(int), 589 .mode = 0644, 590 .proc_handler = ftrace_enable_sysctl, 591 }, 592 #endif 593 #ifdef CONFIG_STACK_TRACER 594 { 595 .procname = "stack_tracer_enabled", 596 .data = &stack_tracer_enabled, 597 .maxlen = sizeof(int), 598 .mode = 0644, 599 .proc_handler = stack_trace_sysctl, 600 }, 601 #endif 602 #ifdef CONFIG_TRACING 603 { 604 .procname = "ftrace_dump_on_oops", 605 .data = &ftrace_dump_on_oops, 606 .maxlen = sizeof(int), 607 .mode = 0644, 608 .proc_handler = proc_dointvec, 609 }, 610 { 611 .procname = "traceoff_on_warning", 612 .data = &__disable_trace_on_warning, 613 .maxlen = sizeof(__disable_trace_on_warning), 614 .mode = 0644, 615 .proc_handler = proc_dointvec, 616 }, 617 { 618 .procname = "tracepoint_printk", 619 .data = &tracepoint_printk, 620 .maxlen = sizeof(tracepoint_printk), 621 .mode = 0644, 622 .proc_handler = proc_dointvec, 623 }, 624 #endif 625 #ifdef CONFIG_KEXEC_CORE 626 { 627 .procname = "kexec_load_disabled", 628 .data = &kexec_load_disabled, 629 .maxlen = sizeof(int), 630 .mode = 0644, 631 /* only handle a transition from default "0" to "1" */ 632 .proc_handler = proc_dointvec_minmax, 633 .extra1 = &one, 634 .extra2 = &one, 635 }, 636 #endif 637 #ifdef CONFIG_MODULES 638 { 639 .procname = "modprobe", 640 .data = &modprobe_path, 641 .maxlen = KMOD_PATH_LEN, 642 .mode = 0644, 643 .proc_handler = proc_dostring, 644 }, 645 { 646 .procname = "modules_disabled", 647 .data = &modules_disabled, 648 .maxlen = sizeof(int), 649 .mode = 0644, 650 /* only handle a transition from default "0" to "1" */ 651 .proc_handler = proc_dointvec_minmax, 652 .extra1 = &one, 653 .extra2 = &one, 654 }, 655 #endif 656 #ifdef CONFIG_UEVENT_HELPER 657 { 658 .procname = "hotplug", 659 .data = &uevent_helper, 660 .maxlen = UEVENT_HELPER_PATH_LEN, 661 .mode = 0644, 662 .proc_handler = proc_dostring, 663 }, 664 #endif 665 #ifdef CONFIG_CHR_DEV_SG 666 { 667 .procname = "sg-big-buff", 668 .data = &sg_big_buff, 669 .maxlen = sizeof (int), 670 .mode = 0444, 671 .proc_handler = proc_dointvec, 672 }, 673 #endif 674 #ifdef CONFIG_BSD_PROCESS_ACCT 675 { 676 .procname = "acct", 677 .data = &acct_parm, 678 .maxlen = 3*sizeof(int), 679 .mode = 0644, 680 .proc_handler = proc_dointvec, 681 }, 682 #endif 683 #ifdef CONFIG_MAGIC_SYSRQ 684 { 685 .procname = "sysrq", 686 .data = &__sysrq_enabled, 687 .maxlen = sizeof (int), 688 .mode = 0644, 689 .proc_handler = sysrq_sysctl_handler, 690 }, 691 #endif 692 #ifdef CONFIG_PROC_SYSCTL 693 { 694 .procname = "cad_pid", 695 .data = NULL, 696 .maxlen = sizeof (int), 697 .mode = 0600, 698 .proc_handler = proc_do_cad_pid, 699 }, 700 #endif 701 { 702 .procname = "threads-max", 703 .data = NULL, 704 .maxlen = sizeof(int), 705 .mode = 0644, 706 .proc_handler = sysctl_max_threads, 707 }, 708 { 709 .procname = "random", 710 .mode = 0555, 711 .child = random_table, 712 }, 713 { 714 .procname = "usermodehelper", 715 .mode = 0555, 716 .child = usermodehelper_table, 717 }, 718 { 719 .procname = "overflowuid", 720 .data = &overflowuid, 721 .maxlen = sizeof(int), 722 .mode = 0644, 723 .proc_handler = proc_dointvec_minmax, 724 .extra1 = &minolduid, 725 .extra2 = &maxolduid, 726 }, 727 { 728 .procname = "overflowgid", 729 .data = &overflowgid, 730 .maxlen = sizeof(int), 731 .mode = 0644, 732 .proc_handler = proc_dointvec_minmax, 733 .extra1 = &minolduid, 734 .extra2 = &maxolduid, 735 }, 736 #ifdef CONFIG_S390 737 #ifdef CONFIG_MATHEMU 738 { 739 .procname = "ieee_emulation_warnings", 740 .data = &sysctl_ieee_emulation_warnings, 741 .maxlen = sizeof(int), 742 .mode = 0644, 743 .proc_handler = proc_dointvec, 744 }, 745 #endif 746 { 747 .procname = "userprocess_debug", 748 .data = &show_unhandled_signals, 749 .maxlen = sizeof(int), 750 .mode = 0644, 751 .proc_handler = proc_dointvec, 752 }, 753 #endif 754 { 755 .procname = "pid_max", 756 .data = &pid_max, 757 .maxlen = sizeof (int), 758 .mode = 0644, 759 .proc_handler = proc_dointvec_minmax, 760 .extra1 = &pid_max_min, 761 .extra2 = &pid_max_max, 762 }, 763 { 764 .procname = "panic_on_oops", 765 .data = &panic_on_oops, 766 .maxlen = sizeof(int), 767 .mode = 0644, 768 .proc_handler = proc_dointvec, 769 }, 770 #if defined CONFIG_PRINTK 771 { 772 .procname = "printk", 773 .data = &console_loglevel, 774 .maxlen = 4*sizeof(int), 775 .mode = 0644, 776 .proc_handler = proc_dointvec, 777 }, 778 { 779 .procname = "printk_ratelimit", 780 .data = &printk_ratelimit_state.interval, 781 .maxlen = sizeof(int), 782 .mode = 0644, 783 .proc_handler = proc_dointvec_jiffies, 784 }, 785 { 786 .procname = "printk_ratelimit_burst", 787 .data = &printk_ratelimit_state.burst, 788 .maxlen = sizeof(int), 789 .mode = 0644, 790 .proc_handler = proc_dointvec, 791 }, 792 { 793 .procname = "printk_delay", 794 .data = &printk_delay_msec, 795 .maxlen = sizeof(int), 796 .mode = 0644, 797 .proc_handler = proc_dointvec_minmax, 798 .extra1 = &zero, 799 .extra2 = &ten_thousand, 800 }, 801 { 802 .procname = "dmesg_restrict", 803 .data = &dmesg_restrict, 804 .maxlen = sizeof(int), 805 .mode = 0644, 806 .proc_handler = proc_dointvec_minmax_sysadmin, 807 .extra1 = &zero, 808 .extra2 = &one, 809 }, 810 { 811 .procname = "kptr_restrict", 812 .data = &kptr_restrict, 813 .maxlen = sizeof(int), 814 .mode = 0644, 815 .proc_handler = proc_dointvec_minmax_sysadmin, 816 .extra1 = &zero, 817 .extra2 = &two, 818 }, 819 #endif 820 { 821 .procname = "ngroups_max", 822 .data = &ngroups_max, 823 .maxlen = sizeof (int), 824 .mode = 0444, 825 .proc_handler = proc_dointvec, 826 }, 827 { 828 .procname = "cap_last_cap", 829 .data = (void *)&cap_last_cap, 830 .maxlen = sizeof(int), 831 .mode = 0444, 832 .proc_handler = proc_dointvec, 833 }, 834 #if defined(CONFIG_LOCKUP_DETECTOR) 835 { 836 .procname = "watchdog", 837 .data = &watchdog_user_enabled, 838 .maxlen = sizeof (int), 839 .mode = 0644, 840 .proc_handler = proc_watchdog, 841 .extra1 = &zero, 842 .extra2 = &one, 843 }, 844 { 845 .procname = "watchdog_thresh", 846 .data = &watchdog_thresh, 847 .maxlen = sizeof(int), 848 .mode = 0644, 849 .proc_handler = proc_watchdog_thresh, 850 .extra1 = &zero, 851 .extra2 = &sixty, 852 }, 853 { 854 .procname = "nmi_watchdog", 855 .data = &nmi_watchdog_enabled, 856 .maxlen = sizeof (int), 857 .mode = 0644, 858 .proc_handler = proc_nmi_watchdog, 859 .extra1 = &zero, 860 #if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR) 861 .extra2 = &one, 862 #else 863 .extra2 = &zero, 864 #endif 865 }, 866 { 867 .procname = "soft_watchdog", 868 .data = &soft_watchdog_enabled, 869 .maxlen = sizeof (int), 870 .mode = 0644, 871 .proc_handler = proc_soft_watchdog, 872 .extra1 = &zero, 873 .extra2 = &one, 874 }, 875 { 876 .procname = "watchdog_cpumask", 877 .data = &watchdog_cpumask_bits, 878 .maxlen = NR_CPUS, 879 .mode = 0644, 880 .proc_handler = proc_watchdog_cpumask, 881 }, 882 { 883 .procname = "softlockup_panic", 884 .data = &softlockup_panic, 885 .maxlen = sizeof(int), 886 .mode = 0644, 887 .proc_handler = proc_dointvec_minmax, 888 .extra1 = &zero, 889 .extra2 = &one, 890 }, 891 #ifdef CONFIG_SMP 892 { 893 .procname = "softlockup_all_cpu_backtrace", 894 .data = &sysctl_softlockup_all_cpu_backtrace, 895 .maxlen = sizeof(int), 896 .mode = 0644, 897 .proc_handler = proc_dointvec_minmax, 898 .extra1 = &zero, 899 .extra2 = &one, 900 }, 901 #endif /* CONFIG_SMP */ 902 #endif 903 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86) 904 { 905 .procname = "unknown_nmi_panic", 906 .data = &unknown_nmi_panic, 907 .maxlen = sizeof (int), 908 .mode = 0644, 909 .proc_handler = proc_dointvec, 910 }, 911 #endif 912 #if defined(CONFIG_X86) 913 { 914 .procname = "panic_on_unrecovered_nmi", 915 .data = &panic_on_unrecovered_nmi, 916 .maxlen = sizeof(int), 917 .mode = 0644, 918 .proc_handler = proc_dointvec, 919 }, 920 { 921 .procname = "panic_on_io_nmi", 922 .data = &panic_on_io_nmi, 923 .maxlen = sizeof(int), 924 .mode = 0644, 925 .proc_handler = proc_dointvec, 926 }, 927 #ifdef CONFIG_DEBUG_STACKOVERFLOW 928 { 929 .procname = "panic_on_stackoverflow", 930 .data = &sysctl_panic_on_stackoverflow, 931 .maxlen = sizeof(int), 932 .mode = 0644, 933 .proc_handler = proc_dointvec, 934 }, 935 #endif 936 { 937 .procname = "bootloader_type", 938 .data = &bootloader_type, 939 .maxlen = sizeof (int), 940 .mode = 0444, 941 .proc_handler = proc_dointvec, 942 }, 943 { 944 .procname = "bootloader_version", 945 .data = &bootloader_version, 946 .maxlen = sizeof (int), 947 .mode = 0444, 948 .proc_handler = proc_dointvec, 949 }, 950 { 951 .procname = "kstack_depth_to_print", 952 .data = &kstack_depth_to_print, 953 .maxlen = sizeof(int), 954 .mode = 0644, 955 .proc_handler = proc_dointvec, 956 }, 957 { 958 .procname = "io_delay_type", 959 .data = &io_delay_type, 960 .maxlen = sizeof(int), 961 .mode = 0644, 962 .proc_handler = proc_dointvec, 963 }, 964 #endif 965 #if defined(CONFIG_MMU) 966 { 967 .procname = "randomize_va_space", 968 .data = &randomize_va_space, 969 .maxlen = sizeof(int), 970 .mode = 0644, 971 .proc_handler = proc_dointvec, 972 }, 973 #endif 974 #if defined(CONFIG_S390) && defined(CONFIG_SMP) 975 { 976 .procname = "spin_retry", 977 .data = &spin_retry, 978 .maxlen = sizeof (int), 979 .mode = 0644, 980 .proc_handler = proc_dointvec, 981 }, 982 #endif 983 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86) 984 { 985 .procname = "acpi_video_flags", 986 .data = &acpi_realmode_flags, 987 .maxlen = sizeof (unsigned long), 988 .mode = 0644, 989 .proc_handler = proc_doulongvec_minmax, 990 }, 991 #endif 992 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN 993 { 994 .procname = "ignore-unaligned-usertrap", 995 .data = &no_unaligned_warning, 996 .maxlen = sizeof (int), 997 .mode = 0644, 998 .proc_handler = proc_dointvec, 999 }, 1000 #endif 1001 #ifdef CONFIG_IA64 1002 { 1003 .procname = "unaligned-dump-stack", 1004 .data = &unaligned_dump_stack, 1005 .maxlen = sizeof (int), 1006 .mode = 0644, 1007 .proc_handler = proc_dointvec, 1008 }, 1009 #endif 1010 #ifdef CONFIG_DETECT_HUNG_TASK 1011 { 1012 .procname = "hung_task_panic", 1013 .data = &sysctl_hung_task_panic, 1014 .maxlen = sizeof(int), 1015 .mode = 0644, 1016 .proc_handler = proc_dointvec_minmax, 1017 .extra1 = &zero, 1018 .extra2 = &one, 1019 }, 1020 { 1021 .procname = "hung_task_check_count", 1022 .data = &sysctl_hung_task_check_count, 1023 .maxlen = sizeof(int), 1024 .mode = 0644, 1025 .proc_handler = proc_dointvec_minmax, 1026 .extra1 = &zero, 1027 }, 1028 { 1029 .procname = "hung_task_timeout_secs", 1030 .data = &sysctl_hung_task_timeout_secs, 1031 .maxlen = sizeof(unsigned long), 1032 .mode = 0644, 1033 .proc_handler = proc_dohung_task_timeout_secs, 1034 .extra2 = &hung_task_timeout_max, 1035 }, 1036 { 1037 .procname = "hung_task_warnings", 1038 .data = &sysctl_hung_task_warnings, 1039 .maxlen = sizeof(int), 1040 .mode = 0644, 1041 .proc_handler = proc_dointvec_minmax, 1042 .extra1 = &neg_one, 1043 }, 1044 #endif 1045 #ifdef CONFIG_COMPAT 1046 { 1047 .procname = "compat-log", 1048 .data = &compat_log, 1049 .maxlen = sizeof (int), 1050 .mode = 0644, 1051 .proc_handler = proc_dointvec, 1052 }, 1053 #endif 1054 #ifdef CONFIG_RT_MUTEXES 1055 { 1056 .procname = "max_lock_depth", 1057 .data = &max_lock_depth, 1058 .maxlen = sizeof(int), 1059 .mode = 0644, 1060 .proc_handler = proc_dointvec, 1061 }, 1062 #endif 1063 { 1064 .procname = "poweroff_cmd", 1065 .data = &poweroff_cmd, 1066 .maxlen = POWEROFF_CMD_PATH_LEN, 1067 .mode = 0644, 1068 .proc_handler = proc_dostring, 1069 }, 1070 #ifdef CONFIG_KEYS 1071 { 1072 .procname = "keys", 1073 .mode = 0555, 1074 .child = key_sysctls, 1075 }, 1076 #endif 1077 #ifdef CONFIG_PERF_EVENTS 1078 /* 1079 * User-space scripts rely on the existence of this file 1080 * as a feature check for perf_events being enabled. 1081 * 1082 * So it's an ABI, do not remove! 1083 */ 1084 { 1085 .procname = "perf_event_paranoid", 1086 .data = &sysctl_perf_event_paranoid, 1087 .maxlen = sizeof(sysctl_perf_event_paranoid), 1088 .mode = 0644, 1089 .proc_handler = proc_dointvec, 1090 }, 1091 { 1092 .procname = "perf_event_mlock_kb", 1093 .data = &sysctl_perf_event_mlock, 1094 .maxlen = sizeof(sysctl_perf_event_mlock), 1095 .mode = 0644, 1096 .proc_handler = proc_dointvec, 1097 }, 1098 { 1099 .procname = "perf_event_max_sample_rate", 1100 .data = &sysctl_perf_event_sample_rate, 1101 .maxlen = sizeof(sysctl_perf_event_sample_rate), 1102 .mode = 0644, 1103 .proc_handler = perf_proc_update_handler, 1104 .extra1 = &one, 1105 }, 1106 { 1107 .procname = "perf_cpu_time_max_percent", 1108 .data = &sysctl_perf_cpu_time_max_percent, 1109 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent), 1110 .mode = 0644, 1111 .proc_handler = perf_cpu_time_max_percent_handler, 1112 .extra1 = &zero, 1113 .extra2 = &one_hundred, 1114 }, 1115 #endif 1116 #ifdef CONFIG_KMEMCHECK 1117 { 1118 .procname = "kmemcheck", 1119 .data = &kmemcheck_enabled, 1120 .maxlen = sizeof(int), 1121 .mode = 0644, 1122 .proc_handler = proc_dointvec, 1123 }, 1124 #endif 1125 { 1126 .procname = "panic_on_warn", 1127 .data = &panic_on_warn, 1128 .maxlen = sizeof(int), 1129 .mode = 0644, 1130 .proc_handler = proc_dointvec_minmax, 1131 .extra1 = &zero, 1132 .extra2 = &one, 1133 }, 1134 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) 1135 { 1136 .procname = "timer_migration", 1137 .data = &sysctl_timer_migration, 1138 .maxlen = sizeof(unsigned int), 1139 .mode = 0644, 1140 .proc_handler = timer_migration_handler, 1141 }, 1142 #endif 1143 #ifdef CONFIG_BPF_SYSCALL 1144 { 1145 .procname = "unprivileged_bpf_disabled", 1146 .data = &sysctl_unprivileged_bpf_disabled, 1147 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled), 1148 .mode = 0644, 1149 /* only handle a transition from default "0" to "1" */ 1150 .proc_handler = proc_dointvec_minmax, 1151 .extra1 = &one, 1152 .extra2 = &one, 1153 }, 1154 #endif 1155 { } 1156 }; 1157 1158 static struct ctl_table vm_table[] = { 1159 { 1160 .procname = "overcommit_memory", 1161 .data = &sysctl_overcommit_memory, 1162 .maxlen = sizeof(sysctl_overcommit_memory), 1163 .mode = 0644, 1164 .proc_handler = proc_dointvec_minmax, 1165 .extra1 = &zero, 1166 .extra2 = &two, 1167 }, 1168 { 1169 .procname = "panic_on_oom", 1170 .data = &sysctl_panic_on_oom, 1171 .maxlen = sizeof(sysctl_panic_on_oom), 1172 .mode = 0644, 1173 .proc_handler = proc_dointvec_minmax, 1174 .extra1 = &zero, 1175 .extra2 = &two, 1176 }, 1177 { 1178 .procname = "oom_kill_allocating_task", 1179 .data = &sysctl_oom_kill_allocating_task, 1180 .maxlen = sizeof(sysctl_oom_kill_allocating_task), 1181 .mode = 0644, 1182 .proc_handler = proc_dointvec, 1183 }, 1184 { 1185 .procname = "oom_dump_tasks", 1186 .data = &sysctl_oom_dump_tasks, 1187 .maxlen = sizeof(sysctl_oom_dump_tasks), 1188 .mode = 0644, 1189 .proc_handler = proc_dointvec, 1190 }, 1191 { 1192 .procname = "overcommit_ratio", 1193 .data = &sysctl_overcommit_ratio, 1194 .maxlen = sizeof(sysctl_overcommit_ratio), 1195 .mode = 0644, 1196 .proc_handler = overcommit_ratio_handler, 1197 }, 1198 { 1199 .procname = "overcommit_kbytes", 1200 .data = &sysctl_overcommit_kbytes, 1201 .maxlen = sizeof(sysctl_overcommit_kbytes), 1202 .mode = 0644, 1203 .proc_handler = overcommit_kbytes_handler, 1204 }, 1205 { 1206 .procname = "page-cluster", 1207 .data = &page_cluster, 1208 .maxlen = sizeof(int), 1209 .mode = 0644, 1210 .proc_handler = proc_dointvec_minmax, 1211 .extra1 = &zero, 1212 }, 1213 { 1214 .procname = "dirty_background_ratio", 1215 .data = &dirty_background_ratio, 1216 .maxlen = sizeof(dirty_background_ratio), 1217 .mode = 0644, 1218 .proc_handler = dirty_background_ratio_handler, 1219 .extra1 = &zero, 1220 .extra2 = &one_hundred, 1221 }, 1222 { 1223 .procname = "dirty_background_bytes", 1224 .data = &dirty_background_bytes, 1225 .maxlen = sizeof(dirty_background_bytes), 1226 .mode = 0644, 1227 .proc_handler = dirty_background_bytes_handler, 1228 .extra1 = &one_ul, 1229 }, 1230 { 1231 .procname = "dirty_ratio", 1232 .data = &vm_dirty_ratio, 1233 .maxlen = sizeof(vm_dirty_ratio), 1234 .mode = 0644, 1235 .proc_handler = dirty_ratio_handler, 1236 .extra1 = &zero, 1237 .extra2 = &one_hundred, 1238 }, 1239 { 1240 .procname = "dirty_bytes", 1241 .data = &vm_dirty_bytes, 1242 .maxlen = sizeof(vm_dirty_bytes), 1243 .mode = 0644, 1244 .proc_handler = dirty_bytes_handler, 1245 .extra1 = &dirty_bytes_min, 1246 }, 1247 { 1248 .procname = "dirty_writeback_centisecs", 1249 .data = &dirty_writeback_interval, 1250 .maxlen = sizeof(dirty_writeback_interval), 1251 .mode = 0644, 1252 .proc_handler = dirty_writeback_centisecs_handler, 1253 }, 1254 { 1255 .procname = "dirty_expire_centisecs", 1256 .data = &dirty_expire_interval, 1257 .maxlen = sizeof(dirty_expire_interval), 1258 .mode = 0644, 1259 .proc_handler = proc_dointvec_minmax, 1260 .extra1 = &zero, 1261 }, 1262 { 1263 .procname = "dirtytime_expire_seconds", 1264 .data = &dirtytime_expire_interval, 1265 .maxlen = sizeof(dirty_expire_interval), 1266 .mode = 0644, 1267 .proc_handler = dirtytime_interval_handler, 1268 .extra1 = &zero, 1269 }, 1270 { 1271 .procname = "nr_pdflush_threads", 1272 .mode = 0444 /* read-only */, 1273 .proc_handler = pdflush_proc_obsolete, 1274 }, 1275 { 1276 .procname = "swappiness", 1277 .data = &vm_swappiness, 1278 .maxlen = sizeof(vm_swappiness), 1279 .mode = 0644, 1280 .proc_handler = proc_dointvec_minmax, 1281 .extra1 = &zero, 1282 .extra2 = &one_hundred, 1283 }, 1284 #ifdef CONFIG_HUGETLB_PAGE 1285 { 1286 .procname = "nr_hugepages", 1287 .data = NULL, 1288 .maxlen = sizeof(unsigned long), 1289 .mode = 0644, 1290 .proc_handler = hugetlb_sysctl_handler, 1291 }, 1292 #ifdef CONFIG_NUMA 1293 { 1294 .procname = "nr_hugepages_mempolicy", 1295 .data = NULL, 1296 .maxlen = sizeof(unsigned long), 1297 .mode = 0644, 1298 .proc_handler = &hugetlb_mempolicy_sysctl_handler, 1299 }, 1300 #endif 1301 { 1302 .procname = "hugetlb_shm_group", 1303 .data = &sysctl_hugetlb_shm_group, 1304 .maxlen = sizeof(gid_t), 1305 .mode = 0644, 1306 .proc_handler = proc_dointvec, 1307 }, 1308 { 1309 .procname = "hugepages_treat_as_movable", 1310 .data = &hugepages_treat_as_movable, 1311 .maxlen = sizeof(int), 1312 .mode = 0644, 1313 .proc_handler = proc_dointvec, 1314 }, 1315 { 1316 .procname = "nr_overcommit_hugepages", 1317 .data = NULL, 1318 .maxlen = sizeof(unsigned long), 1319 .mode = 0644, 1320 .proc_handler = hugetlb_overcommit_handler, 1321 }, 1322 #endif 1323 { 1324 .procname = "lowmem_reserve_ratio", 1325 .data = &sysctl_lowmem_reserve_ratio, 1326 .maxlen = sizeof(sysctl_lowmem_reserve_ratio), 1327 .mode = 0644, 1328 .proc_handler = lowmem_reserve_ratio_sysctl_handler, 1329 }, 1330 { 1331 .procname = "drop_caches", 1332 .data = &sysctl_drop_caches, 1333 .maxlen = sizeof(int), 1334 .mode = 0644, 1335 .proc_handler = drop_caches_sysctl_handler, 1336 .extra1 = &one, 1337 .extra2 = &four, 1338 }, 1339 #ifdef CONFIG_COMPACTION 1340 { 1341 .procname = "compact_memory", 1342 .data = &sysctl_compact_memory, 1343 .maxlen = sizeof(int), 1344 .mode = 0200, 1345 .proc_handler = sysctl_compaction_handler, 1346 }, 1347 { 1348 .procname = "extfrag_threshold", 1349 .data = &sysctl_extfrag_threshold, 1350 .maxlen = sizeof(int), 1351 .mode = 0644, 1352 .proc_handler = sysctl_extfrag_handler, 1353 .extra1 = &min_extfrag_threshold, 1354 .extra2 = &max_extfrag_threshold, 1355 }, 1356 { 1357 .procname = "compact_unevictable_allowed", 1358 .data = &sysctl_compact_unevictable_allowed, 1359 .maxlen = sizeof(int), 1360 .mode = 0644, 1361 .proc_handler = proc_dointvec, 1362 .extra1 = &zero, 1363 .extra2 = &one, 1364 }, 1365 1366 #endif /* CONFIG_COMPACTION */ 1367 { 1368 .procname = "min_free_kbytes", 1369 .data = &min_free_kbytes, 1370 .maxlen = sizeof(min_free_kbytes), 1371 .mode = 0644, 1372 .proc_handler = min_free_kbytes_sysctl_handler, 1373 .extra1 = &zero, 1374 }, 1375 { 1376 .procname = "percpu_pagelist_fraction", 1377 .data = &percpu_pagelist_fraction, 1378 .maxlen = sizeof(percpu_pagelist_fraction), 1379 .mode = 0644, 1380 .proc_handler = percpu_pagelist_fraction_sysctl_handler, 1381 .extra1 = &zero, 1382 }, 1383 #ifdef CONFIG_MMU 1384 { 1385 .procname = "max_map_count", 1386 .data = &sysctl_max_map_count, 1387 .maxlen = sizeof(sysctl_max_map_count), 1388 .mode = 0644, 1389 .proc_handler = proc_dointvec_minmax, 1390 .extra1 = &zero, 1391 }, 1392 #else 1393 { 1394 .procname = "nr_trim_pages", 1395 .data = &sysctl_nr_trim_pages, 1396 .maxlen = sizeof(sysctl_nr_trim_pages), 1397 .mode = 0644, 1398 .proc_handler = proc_dointvec_minmax, 1399 .extra1 = &zero, 1400 }, 1401 #endif 1402 { 1403 .procname = "laptop_mode", 1404 .data = &laptop_mode, 1405 .maxlen = sizeof(laptop_mode), 1406 .mode = 0644, 1407 .proc_handler = proc_dointvec_jiffies, 1408 }, 1409 { 1410 .procname = "block_dump", 1411 .data = &block_dump, 1412 .maxlen = sizeof(block_dump), 1413 .mode = 0644, 1414 .proc_handler = proc_dointvec, 1415 .extra1 = &zero, 1416 }, 1417 { 1418 .procname = "vfs_cache_pressure", 1419 .data = &sysctl_vfs_cache_pressure, 1420 .maxlen = sizeof(sysctl_vfs_cache_pressure), 1421 .mode = 0644, 1422 .proc_handler = proc_dointvec, 1423 .extra1 = &zero, 1424 }, 1425 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT 1426 { 1427 .procname = "legacy_va_layout", 1428 .data = &sysctl_legacy_va_layout, 1429 .maxlen = sizeof(sysctl_legacy_va_layout), 1430 .mode = 0644, 1431 .proc_handler = proc_dointvec, 1432 .extra1 = &zero, 1433 }, 1434 #endif 1435 #ifdef CONFIG_NUMA 1436 { 1437 .procname = "zone_reclaim_mode", 1438 .data = &zone_reclaim_mode, 1439 .maxlen = sizeof(zone_reclaim_mode), 1440 .mode = 0644, 1441 .proc_handler = proc_dointvec, 1442 .extra1 = &zero, 1443 }, 1444 { 1445 .procname = "min_unmapped_ratio", 1446 .data = &sysctl_min_unmapped_ratio, 1447 .maxlen = sizeof(sysctl_min_unmapped_ratio), 1448 .mode = 0644, 1449 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler, 1450 .extra1 = &zero, 1451 .extra2 = &one_hundred, 1452 }, 1453 { 1454 .procname = "min_slab_ratio", 1455 .data = &sysctl_min_slab_ratio, 1456 .maxlen = sizeof(sysctl_min_slab_ratio), 1457 .mode = 0644, 1458 .proc_handler = sysctl_min_slab_ratio_sysctl_handler, 1459 .extra1 = &zero, 1460 .extra2 = &one_hundred, 1461 }, 1462 #endif 1463 #ifdef CONFIG_SMP 1464 { 1465 .procname = "stat_interval", 1466 .data = &sysctl_stat_interval, 1467 .maxlen = sizeof(sysctl_stat_interval), 1468 .mode = 0644, 1469 .proc_handler = proc_dointvec_jiffies, 1470 }, 1471 #endif 1472 #ifdef CONFIG_MMU 1473 { 1474 .procname = "mmap_min_addr", 1475 .data = &dac_mmap_min_addr, 1476 .maxlen = sizeof(unsigned long), 1477 .mode = 0644, 1478 .proc_handler = mmap_min_addr_handler, 1479 }, 1480 #endif 1481 #ifdef CONFIG_NUMA 1482 { 1483 .procname = "numa_zonelist_order", 1484 .data = &numa_zonelist_order, 1485 .maxlen = NUMA_ZONELIST_ORDER_LEN, 1486 .mode = 0644, 1487 .proc_handler = numa_zonelist_order_handler, 1488 }, 1489 #endif 1490 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \ 1491 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL)) 1492 { 1493 .procname = "vdso_enabled", 1494 #ifdef CONFIG_X86_32 1495 .data = &vdso32_enabled, 1496 .maxlen = sizeof(vdso32_enabled), 1497 #else 1498 .data = &vdso_enabled, 1499 .maxlen = sizeof(vdso_enabled), 1500 #endif 1501 .mode = 0644, 1502 .proc_handler = proc_dointvec, 1503 .extra1 = &zero, 1504 }, 1505 #endif 1506 #ifdef CONFIG_HIGHMEM 1507 { 1508 .procname = "highmem_is_dirtyable", 1509 .data = &vm_highmem_is_dirtyable, 1510 .maxlen = sizeof(vm_highmem_is_dirtyable), 1511 .mode = 0644, 1512 .proc_handler = proc_dointvec_minmax, 1513 .extra1 = &zero, 1514 .extra2 = &one, 1515 }, 1516 #endif 1517 #ifdef CONFIG_MEMORY_FAILURE 1518 { 1519 .procname = "memory_failure_early_kill", 1520 .data = &sysctl_memory_failure_early_kill, 1521 .maxlen = sizeof(sysctl_memory_failure_early_kill), 1522 .mode = 0644, 1523 .proc_handler = proc_dointvec_minmax, 1524 .extra1 = &zero, 1525 .extra2 = &one, 1526 }, 1527 { 1528 .procname = "memory_failure_recovery", 1529 .data = &sysctl_memory_failure_recovery, 1530 .maxlen = sizeof(sysctl_memory_failure_recovery), 1531 .mode = 0644, 1532 .proc_handler = proc_dointvec_minmax, 1533 .extra1 = &zero, 1534 .extra2 = &one, 1535 }, 1536 #endif 1537 { 1538 .procname = "user_reserve_kbytes", 1539 .data = &sysctl_user_reserve_kbytes, 1540 .maxlen = sizeof(sysctl_user_reserve_kbytes), 1541 .mode = 0644, 1542 .proc_handler = proc_doulongvec_minmax, 1543 }, 1544 { 1545 .procname = "admin_reserve_kbytes", 1546 .data = &sysctl_admin_reserve_kbytes, 1547 .maxlen = sizeof(sysctl_admin_reserve_kbytes), 1548 .mode = 0644, 1549 .proc_handler = proc_doulongvec_minmax, 1550 }, 1551 { } 1552 }; 1553 1554 static struct ctl_table fs_table[] = { 1555 { 1556 .procname = "inode-nr", 1557 .data = &inodes_stat, 1558 .maxlen = 2*sizeof(long), 1559 .mode = 0444, 1560 .proc_handler = proc_nr_inodes, 1561 }, 1562 { 1563 .procname = "inode-state", 1564 .data = &inodes_stat, 1565 .maxlen = 7*sizeof(long), 1566 .mode = 0444, 1567 .proc_handler = proc_nr_inodes, 1568 }, 1569 { 1570 .procname = "file-nr", 1571 .data = &files_stat, 1572 .maxlen = sizeof(files_stat), 1573 .mode = 0444, 1574 .proc_handler = proc_nr_files, 1575 }, 1576 { 1577 .procname = "file-max", 1578 .data = &files_stat.max_files, 1579 .maxlen = sizeof(files_stat.max_files), 1580 .mode = 0644, 1581 .proc_handler = proc_doulongvec_minmax, 1582 }, 1583 { 1584 .procname = "nr_open", 1585 .data = &sysctl_nr_open, 1586 .maxlen = sizeof(int), 1587 .mode = 0644, 1588 .proc_handler = proc_dointvec_minmax, 1589 .extra1 = &sysctl_nr_open_min, 1590 .extra2 = &sysctl_nr_open_max, 1591 }, 1592 { 1593 .procname = "dentry-state", 1594 .data = &dentry_stat, 1595 .maxlen = 6*sizeof(long), 1596 .mode = 0444, 1597 .proc_handler = proc_nr_dentry, 1598 }, 1599 { 1600 .procname = "overflowuid", 1601 .data = &fs_overflowuid, 1602 .maxlen = sizeof(int), 1603 .mode = 0644, 1604 .proc_handler = proc_dointvec_minmax, 1605 .extra1 = &minolduid, 1606 .extra2 = &maxolduid, 1607 }, 1608 { 1609 .procname = "overflowgid", 1610 .data = &fs_overflowgid, 1611 .maxlen = sizeof(int), 1612 .mode = 0644, 1613 .proc_handler = proc_dointvec_minmax, 1614 .extra1 = &minolduid, 1615 .extra2 = &maxolduid, 1616 }, 1617 #ifdef CONFIG_FILE_LOCKING 1618 { 1619 .procname = "leases-enable", 1620 .data = &leases_enable, 1621 .maxlen = sizeof(int), 1622 .mode = 0644, 1623 .proc_handler = proc_dointvec, 1624 }, 1625 #endif 1626 #ifdef CONFIG_DNOTIFY 1627 { 1628 .procname = "dir-notify-enable", 1629 .data = &dir_notify_enable, 1630 .maxlen = sizeof(int), 1631 .mode = 0644, 1632 .proc_handler = proc_dointvec, 1633 }, 1634 #endif 1635 #ifdef CONFIG_MMU 1636 #ifdef CONFIG_FILE_LOCKING 1637 { 1638 .procname = "lease-break-time", 1639 .data = &lease_break_time, 1640 .maxlen = sizeof(int), 1641 .mode = 0644, 1642 .proc_handler = proc_dointvec, 1643 }, 1644 #endif 1645 #ifdef CONFIG_AIO 1646 { 1647 .procname = "aio-nr", 1648 .data = &aio_nr, 1649 .maxlen = sizeof(aio_nr), 1650 .mode = 0444, 1651 .proc_handler = proc_doulongvec_minmax, 1652 }, 1653 { 1654 .procname = "aio-max-nr", 1655 .data = &aio_max_nr, 1656 .maxlen = sizeof(aio_max_nr), 1657 .mode = 0644, 1658 .proc_handler = proc_doulongvec_minmax, 1659 }, 1660 #endif /* CONFIG_AIO */ 1661 #ifdef CONFIG_INOTIFY_USER 1662 { 1663 .procname = "inotify", 1664 .mode = 0555, 1665 .child = inotify_table, 1666 }, 1667 #endif 1668 #ifdef CONFIG_EPOLL 1669 { 1670 .procname = "epoll", 1671 .mode = 0555, 1672 .child = epoll_table, 1673 }, 1674 #endif 1675 #endif 1676 { 1677 .procname = "protected_symlinks", 1678 .data = &sysctl_protected_symlinks, 1679 .maxlen = sizeof(int), 1680 .mode = 0600, 1681 .proc_handler = proc_dointvec_minmax, 1682 .extra1 = &zero, 1683 .extra2 = &one, 1684 }, 1685 { 1686 .procname = "protected_hardlinks", 1687 .data = &sysctl_protected_hardlinks, 1688 .maxlen = sizeof(int), 1689 .mode = 0600, 1690 .proc_handler = proc_dointvec_minmax, 1691 .extra1 = &zero, 1692 .extra2 = &one, 1693 }, 1694 { 1695 .procname = "suid_dumpable", 1696 .data = &suid_dumpable, 1697 .maxlen = sizeof(int), 1698 .mode = 0644, 1699 .proc_handler = proc_dointvec_minmax_coredump, 1700 .extra1 = &zero, 1701 .extra2 = &two, 1702 }, 1703 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE) 1704 { 1705 .procname = "binfmt_misc", 1706 .mode = 0555, 1707 .child = sysctl_mount_point, 1708 }, 1709 #endif 1710 { 1711 .procname = "pipe-max-size", 1712 .data = &pipe_max_size, 1713 .maxlen = sizeof(int), 1714 .mode = 0644, 1715 .proc_handler = &pipe_proc_fn, 1716 .extra1 = &pipe_min_size, 1717 }, 1718 { } 1719 }; 1720 1721 static struct ctl_table debug_table[] = { 1722 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE 1723 { 1724 .procname = "exception-trace", 1725 .data = &show_unhandled_signals, 1726 .maxlen = sizeof(int), 1727 .mode = 0644, 1728 .proc_handler = proc_dointvec 1729 }, 1730 #endif 1731 #if defined(CONFIG_OPTPROBES) 1732 { 1733 .procname = "kprobes-optimization", 1734 .data = &sysctl_kprobes_optimization, 1735 .maxlen = sizeof(int), 1736 .mode = 0644, 1737 .proc_handler = proc_kprobes_optimization_handler, 1738 .extra1 = &zero, 1739 .extra2 = &one, 1740 }, 1741 #endif 1742 { } 1743 }; 1744 1745 static struct ctl_table dev_table[] = { 1746 { } 1747 }; 1748 1749 int __init sysctl_init(void) 1750 { 1751 struct ctl_table_header *hdr; 1752 1753 hdr = register_sysctl_table(sysctl_base_table); 1754 kmemleak_not_leak(hdr); 1755 return 0; 1756 } 1757 1758 #endif /* CONFIG_SYSCTL */ 1759 1760 /* 1761 * /proc/sys support 1762 */ 1763 1764 #ifdef CONFIG_PROC_SYSCTL 1765 1766 static int _proc_do_string(char *data, int maxlen, int write, 1767 char __user *buffer, 1768 size_t *lenp, loff_t *ppos) 1769 { 1770 size_t len; 1771 char __user *p; 1772 char c; 1773 1774 if (!data || !maxlen || !*lenp) { 1775 *lenp = 0; 1776 return 0; 1777 } 1778 1779 if (write) { 1780 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) { 1781 /* Only continue writes not past the end of buffer. */ 1782 len = strlen(data); 1783 if (len > maxlen - 1) 1784 len = maxlen - 1; 1785 1786 if (*ppos > len) 1787 return 0; 1788 len = *ppos; 1789 } else { 1790 /* Start writing from beginning of buffer. */ 1791 len = 0; 1792 } 1793 1794 *ppos += *lenp; 1795 p = buffer; 1796 while ((p - buffer) < *lenp && len < maxlen - 1) { 1797 if (get_user(c, p++)) 1798 return -EFAULT; 1799 if (c == 0 || c == '\n') 1800 break; 1801 data[len++] = c; 1802 } 1803 data[len] = 0; 1804 } else { 1805 len = strlen(data); 1806 if (len > maxlen) 1807 len = maxlen; 1808 1809 if (*ppos > len) { 1810 *lenp = 0; 1811 return 0; 1812 } 1813 1814 data += *ppos; 1815 len -= *ppos; 1816 1817 if (len > *lenp) 1818 len = *lenp; 1819 if (len) 1820 if (copy_to_user(buffer, data, len)) 1821 return -EFAULT; 1822 if (len < *lenp) { 1823 if (put_user('\n', buffer + len)) 1824 return -EFAULT; 1825 len++; 1826 } 1827 *lenp = len; 1828 *ppos += len; 1829 } 1830 return 0; 1831 } 1832 1833 static void warn_sysctl_write(struct ctl_table *table) 1834 { 1835 pr_warn_once("%s wrote to %s when file position was not 0!\n" 1836 "This will not be supported in the future. To silence this\n" 1837 "warning, set kernel.sysctl_writes_strict = -1\n", 1838 current->comm, table->procname); 1839 } 1840 1841 /** 1842 * proc_dostring - read a string sysctl 1843 * @table: the sysctl table 1844 * @write: %TRUE if this is a write to the sysctl file 1845 * @buffer: the user buffer 1846 * @lenp: the size of the user buffer 1847 * @ppos: file position 1848 * 1849 * Reads/writes a string from/to the user buffer. If the kernel 1850 * buffer provided is not large enough to hold the string, the 1851 * string is truncated. The copied string is %NULL-terminated. 1852 * If the string is being read by the user process, it is copied 1853 * and a newline '\n' is added. It is truncated if the buffer is 1854 * not large enough. 1855 * 1856 * Returns 0 on success. 1857 */ 1858 int proc_dostring(struct ctl_table *table, int write, 1859 void __user *buffer, size_t *lenp, loff_t *ppos) 1860 { 1861 if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN) 1862 warn_sysctl_write(table); 1863 1864 return _proc_do_string((char *)(table->data), table->maxlen, write, 1865 (char __user *)buffer, lenp, ppos); 1866 } 1867 1868 static size_t proc_skip_spaces(char **buf) 1869 { 1870 size_t ret; 1871 char *tmp = skip_spaces(*buf); 1872 ret = tmp - *buf; 1873 *buf = tmp; 1874 return ret; 1875 } 1876 1877 static void proc_skip_char(char **buf, size_t *size, const char v) 1878 { 1879 while (*size) { 1880 if (**buf != v) 1881 break; 1882 (*size)--; 1883 (*buf)++; 1884 } 1885 } 1886 1887 #define TMPBUFLEN 22 1888 /** 1889 * proc_get_long - reads an ASCII formatted integer from a user buffer 1890 * 1891 * @buf: a kernel buffer 1892 * @size: size of the kernel buffer 1893 * @val: this is where the number will be stored 1894 * @neg: set to %TRUE if number is negative 1895 * @perm_tr: a vector which contains the allowed trailers 1896 * @perm_tr_len: size of the perm_tr vector 1897 * @tr: pointer to store the trailer character 1898 * 1899 * In case of success %0 is returned and @buf and @size are updated with 1900 * the amount of bytes read. If @tr is non-NULL and a trailing 1901 * character exists (size is non-zero after returning from this 1902 * function), @tr is updated with the trailing character. 1903 */ 1904 static int proc_get_long(char **buf, size_t *size, 1905 unsigned long *val, bool *neg, 1906 const char *perm_tr, unsigned perm_tr_len, char *tr) 1907 { 1908 int len; 1909 char *p, tmp[TMPBUFLEN]; 1910 1911 if (!*size) 1912 return -EINVAL; 1913 1914 len = *size; 1915 if (len > TMPBUFLEN - 1) 1916 len = TMPBUFLEN - 1; 1917 1918 memcpy(tmp, *buf, len); 1919 1920 tmp[len] = 0; 1921 p = tmp; 1922 if (*p == '-' && *size > 1) { 1923 *neg = true; 1924 p++; 1925 } else 1926 *neg = false; 1927 if (!isdigit(*p)) 1928 return -EINVAL; 1929 1930 *val = simple_strtoul(p, &p, 0); 1931 1932 len = p - tmp; 1933 1934 /* We don't know if the next char is whitespace thus we may accept 1935 * invalid integers (e.g. 1234...a) or two integers instead of one 1936 * (e.g. 123...1). So lets not allow such large numbers. */ 1937 if (len == TMPBUFLEN - 1) 1938 return -EINVAL; 1939 1940 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len)) 1941 return -EINVAL; 1942 1943 if (tr && (len < *size)) 1944 *tr = *p; 1945 1946 *buf += len; 1947 *size -= len; 1948 1949 return 0; 1950 } 1951 1952 /** 1953 * proc_put_long - converts an integer to a decimal ASCII formatted string 1954 * 1955 * @buf: the user buffer 1956 * @size: the size of the user buffer 1957 * @val: the integer to be converted 1958 * @neg: sign of the number, %TRUE for negative 1959 * 1960 * In case of success %0 is returned and @buf and @size are updated with 1961 * the amount of bytes written. 1962 */ 1963 static int proc_put_long(void __user **buf, size_t *size, unsigned long val, 1964 bool neg) 1965 { 1966 int len; 1967 char tmp[TMPBUFLEN], *p = tmp; 1968 1969 sprintf(p, "%s%lu", neg ? "-" : "", val); 1970 len = strlen(tmp); 1971 if (len > *size) 1972 len = *size; 1973 if (copy_to_user(*buf, tmp, len)) 1974 return -EFAULT; 1975 *size -= len; 1976 *buf += len; 1977 return 0; 1978 } 1979 #undef TMPBUFLEN 1980 1981 static int proc_put_char(void __user **buf, size_t *size, char c) 1982 { 1983 if (*size) { 1984 char __user **buffer = (char __user **)buf; 1985 if (put_user(c, *buffer)) 1986 return -EFAULT; 1987 (*size)--, (*buffer)++; 1988 *buf = *buffer; 1989 } 1990 return 0; 1991 } 1992 1993 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp, 1994 int *valp, 1995 int write, void *data) 1996 { 1997 if (write) { 1998 if (*negp) { 1999 if (*lvalp > (unsigned long) INT_MAX + 1) 2000 return -EINVAL; 2001 *valp = -*lvalp; 2002 } else { 2003 if (*lvalp > (unsigned long) INT_MAX) 2004 return -EINVAL; 2005 *valp = *lvalp; 2006 } 2007 } else { 2008 int val = *valp; 2009 if (val < 0) { 2010 *negp = true; 2011 *lvalp = -(unsigned long)val; 2012 } else { 2013 *negp = false; 2014 *lvalp = (unsigned long)val; 2015 } 2016 } 2017 return 0; 2018 } 2019 2020 static const char proc_wspace_sep[] = { ' ', '\t', '\n' }; 2021 2022 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table, 2023 int write, void __user *buffer, 2024 size_t *lenp, loff_t *ppos, 2025 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 2026 int write, void *data), 2027 void *data) 2028 { 2029 int *i, vleft, first = 1, err = 0; 2030 unsigned long page = 0; 2031 size_t left; 2032 char *kbuf; 2033 2034 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) { 2035 *lenp = 0; 2036 return 0; 2037 } 2038 2039 i = (int *) tbl_data; 2040 vleft = table->maxlen / sizeof(*i); 2041 left = *lenp; 2042 2043 if (!conv) 2044 conv = do_proc_dointvec_conv; 2045 2046 if (write) { 2047 if (*ppos) { 2048 switch (sysctl_writes_strict) { 2049 case SYSCTL_WRITES_STRICT: 2050 goto out; 2051 case SYSCTL_WRITES_WARN: 2052 warn_sysctl_write(table); 2053 break; 2054 default: 2055 break; 2056 } 2057 } 2058 2059 if (left > PAGE_SIZE - 1) 2060 left = PAGE_SIZE - 1; 2061 page = __get_free_page(GFP_TEMPORARY); 2062 kbuf = (char *) page; 2063 if (!kbuf) 2064 return -ENOMEM; 2065 if (copy_from_user(kbuf, buffer, left)) { 2066 err = -EFAULT; 2067 goto free; 2068 } 2069 kbuf[left] = 0; 2070 } 2071 2072 for (; left && vleft--; i++, first=0) { 2073 unsigned long lval; 2074 bool neg; 2075 2076 if (write) { 2077 left -= proc_skip_spaces(&kbuf); 2078 2079 if (!left) 2080 break; 2081 err = proc_get_long(&kbuf, &left, &lval, &neg, 2082 proc_wspace_sep, 2083 sizeof(proc_wspace_sep), NULL); 2084 if (err) 2085 break; 2086 if (conv(&neg, &lval, i, 1, data)) { 2087 err = -EINVAL; 2088 break; 2089 } 2090 } else { 2091 if (conv(&neg, &lval, i, 0, data)) { 2092 err = -EINVAL; 2093 break; 2094 } 2095 if (!first) 2096 err = proc_put_char(&buffer, &left, '\t'); 2097 if (err) 2098 break; 2099 err = proc_put_long(&buffer, &left, lval, neg); 2100 if (err) 2101 break; 2102 } 2103 } 2104 2105 if (!write && !first && left && !err) 2106 err = proc_put_char(&buffer, &left, '\n'); 2107 if (write && !err && left) 2108 left -= proc_skip_spaces(&kbuf); 2109 free: 2110 if (write) { 2111 free_page(page); 2112 if (first) 2113 return err ? : -EINVAL; 2114 } 2115 *lenp -= left; 2116 out: 2117 *ppos += *lenp; 2118 return err; 2119 } 2120 2121 static int do_proc_dointvec(struct ctl_table *table, int write, 2122 void __user *buffer, size_t *lenp, loff_t *ppos, 2123 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 2124 int write, void *data), 2125 void *data) 2126 { 2127 return __do_proc_dointvec(table->data, table, write, 2128 buffer, lenp, ppos, conv, data); 2129 } 2130 2131 /** 2132 * proc_dointvec - read a vector of integers 2133 * @table: the sysctl table 2134 * @write: %TRUE if this is a write to the sysctl file 2135 * @buffer: the user buffer 2136 * @lenp: the size of the user buffer 2137 * @ppos: file position 2138 * 2139 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2140 * values from/to the user buffer, treated as an ASCII string. 2141 * 2142 * Returns 0 on success. 2143 */ 2144 int proc_dointvec(struct ctl_table *table, int write, 2145 void __user *buffer, size_t *lenp, loff_t *ppos) 2146 { 2147 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2148 NULL,NULL); 2149 } 2150 2151 /* 2152 * Taint values can only be increased 2153 * This means we can safely use a temporary. 2154 */ 2155 static int proc_taint(struct ctl_table *table, int write, 2156 void __user *buffer, size_t *lenp, loff_t *ppos) 2157 { 2158 struct ctl_table t; 2159 unsigned long tmptaint = get_taint(); 2160 int err; 2161 2162 if (write && !capable(CAP_SYS_ADMIN)) 2163 return -EPERM; 2164 2165 t = *table; 2166 t.data = &tmptaint; 2167 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos); 2168 if (err < 0) 2169 return err; 2170 2171 if (write) { 2172 /* 2173 * Poor man's atomic or. Not worth adding a primitive 2174 * to everyone's atomic.h for this 2175 */ 2176 int i; 2177 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) { 2178 if ((tmptaint >> i) & 1) 2179 add_taint(i, LOCKDEP_STILL_OK); 2180 } 2181 } 2182 2183 return err; 2184 } 2185 2186 #ifdef CONFIG_PRINTK 2187 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write, 2188 void __user *buffer, size_t *lenp, loff_t *ppos) 2189 { 2190 if (write && !capable(CAP_SYS_ADMIN)) 2191 return -EPERM; 2192 2193 return proc_dointvec_minmax(table, write, buffer, lenp, ppos); 2194 } 2195 #endif 2196 2197 struct do_proc_dointvec_minmax_conv_param { 2198 int *min; 2199 int *max; 2200 }; 2201 2202 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp, 2203 int *valp, 2204 int write, void *data) 2205 { 2206 struct do_proc_dointvec_minmax_conv_param *param = data; 2207 if (write) { 2208 int val = *negp ? -*lvalp : *lvalp; 2209 if ((param->min && *param->min > val) || 2210 (param->max && *param->max < val)) 2211 return -EINVAL; 2212 *valp = val; 2213 } else { 2214 int val = *valp; 2215 if (val < 0) { 2216 *negp = true; 2217 *lvalp = -(unsigned long)val; 2218 } else { 2219 *negp = false; 2220 *lvalp = (unsigned long)val; 2221 } 2222 } 2223 return 0; 2224 } 2225 2226 /** 2227 * proc_dointvec_minmax - read a vector of integers with min/max values 2228 * @table: the sysctl table 2229 * @write: %TRUE if this is a write to the sysctl file 2230 * @buffer: the user buffer 2231 * @lenp: the size of the user buffer 2232 * @ppos: file position 2233 * 2234 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2235 * values from/to the user buffer, treated as an ASCII string. 2236 * 2237 * This routine will ensure the values are within the range specified by 2238 * table->extra1 (min) and table->extra2 (max). 2239 * 2240 * Returns 0 on success. 2241 */ 2242 int proc_dointvec_minmax(struct ctl_table *table, int write, 2243 void __user *buffer, size_t *lenp, loff_t *ppos) 2244 { 2245 struct do_proc_dointvec_minmax_conv_param param = { 2246 .min = (int *) table->extra1, 2247 .max = (int *) table->extra2, 2248 }; 2249 return do_proc_dointvec(table, write, buffer, lenp, ppos, 2250 do_proc_dointvec_minmax_conv, ¶m); 2251 } 2252 2253 static void validate_coredump_safety(void) 2254 { 2255 #ifdef CONFIG_COREDUMP 2256 if (suid_dumpable == SUID_DUMP_ROOT && 2257 core_pattern[0] != '/' && core_pattern[0] != '|') { 2258 printk(KERN_WARNING "Unsafe core_pattern used with "\ 2259 "suid_dumpable=2. Pipe handler or fully qualified "\ 2260 "core dump path required.\n"); 2261 } 2262 #endif 2263 } 2264 2265 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write, 2266 void __user *buffer, size_t *lenp, loff_t *ppos) 2267 { 2268 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos); 2269 if (!error) 2270 validate_coredump_safety(); 2271 return error; 2272 } 2273 2274 #ifdef CONFIG_COREDUMP 2275 static int proc_dostring_coredump(struct ctl_table *table, int write, 2276 void __user *buffer, size_t *lenp, loff_t *ppos) 2277 { 2278 int error = proc_dostring(table, write, buffer, lenp, ppos); 2279 if (!error) 2280 validate_coredump_safety(); 2281 return error; 2282 } 2283 #endif 2284 2285 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write, 2286 void __user *buffer, 2287 size_t *lenp, loff_t *ppos, 2288 unsigned long convmul, 2289 unsigned long convdiv) 2290 { 2291 unsigned long *i, *min, *max; 2292 int vleft, first = 1, err = 0; 2293 unsigned long page = 0; 2294 size_t left; 2295 char *kbuf; 2296 2297 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) { 2298 *lenp = 0; 2299 return 0; 2300 } 2301 2302 i = (unsigned long *) data; 2303 min = (unsigned long *) table->extra1; 2304 max = (unsigned long *) table->extra2; 2305 vleft = table->maxlen / sizeof(unsigned long); 2306 left = *lenp; 2307 2308 if (write) { 2309 if (*ppos) { 2310 switch (sysctl_writes_strict) { 2311 case SYSCTL_WRITES_STRICT: 2312 goto out; 2313 case SYSCTL_WRITES_WARN: 2314 warn_sysctl_write(table); 2315 break; 2316 default: 2317 break; 2318 } 2319 } 2320 2321 if (left > PAGE_SIZE - 1) 2322 left = PAGE_SIZE - 1; 2323 page = __get_free_page(GFP_TEMPORARY); 2324 kbuf = (char *) page; 2325 if (!kbuf) 2326 return -ENOMEM; 2327 if (copy_from_user(kbuf, buffer, left)) { 2328 err = -EFAULT; 2329 goto free; 2330 } 2331 kbuf[left] = 0; 2332 } 2333 2334 for (; left && vleft--; i++, first = 0) { 2335 unsigned long val; 2336 2337 if (write) { 2338 bool neg; 2339 2340 left -= proc_skip_spaces(&kbuf); 2341 2342 err = proc_get_long(&kbuf, &left, &val, &neg, 2343 proc_wspace_sep, 2344 sizeof(proc_wspace_sep), NULL); 2345 if (err) 2346 break; 2347 if (neg) 2348 continue; 2349 if ((min && val < *min) || (max && val > *max)) 2350 continue; 2351 *i = val; 2352 } else { 2353 val = convdiv * (*i) / convmul; 2354 if (!first) { 2355 err = proc_put_char(&buffer, &left, '\t'); 2356 if (err) 2357 break; 2358 } 2359 err = proc_put_long(&buffer, &left, val, false); 2360 if (err) 2361 break; 2362 } 2363 } 2364 2365 if (!write && !first && left && !err) 2366 err = proc_put_char(&buffer, &left, '\n'); 2367 if (write && !err) 2368 left -= proc_skip_spaces(&kbuf); 2369 free: 2370 if (write) { 2371 free_page(page); 2372 if (first) 2373 return err ? : -EINVAL; 2374 } 2375 *lenp -= left; 2376 out: 2377 *ppos += *lenp; 2378 return err; 2379 } 2380 2381 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write, 2382 void __user *buffer, 2383 size_t *lenp, loff_t *ppos, 2384 unsigned long convmul, 2385 unsigned long convdiv) 2386 { 2387 return __do_proc_doulongvec_minmax(table->data, table, write, 2388 buffer, lenp, ppos, convmul, convdiv); 2389 } 2390 2391 /** 2392 * proc_doulongvec_minmax - read a vector of long integers with min/max values 2393 * @table: the sysctl table 2394 * @write: %TRUE if this is a write to the sysctl file 2395 * @buffer: the user buffer 2396 * @lenp: the size of the user buffer 2397 * @ppos: file position 2398 * 2399 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 2400 * values from/to the user buffer, treated as an ASCII string. 2401 * 2402 * This routine will ensure the values are within the range specified by 2403 * table->extra1 (min) and table->extra2 (max). 2404 * 2405 * Returns 0 on success. 2406 */ 2407 int proc_doulongvec_minmax(struct ctl_table *table, int write, 2408 void __user *buffer, size_t *lenp, loff_t *ppos) 2409 { 2410 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l); 2411 } 2412 2413 /** 2414 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values 2415 * @table: the sysctl table 2416 * @write: %TRUE if this is a write to the sysctl file 2417 * @buffer: the user buffer 2418 * @lenp: the size of the user buffer 2419 * @ppos: file position 2420 * 2421 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 2422 * values from/to the user buffer, treated as an ASCII string. The values 2423 * are treated as milliseconds, and converted to jiffies when they are stored. 2424 * 2425 * This routine will ensure the values are within the range specified by 2426 * table->extra1 (min) and table->extra2 (max). 2427 * 2428 * Returns 0 on success. 2429 */ 2430 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 2431 void __user *buffer, 2432 size_t *lenp, loff_t *ppos) 2433 { 2434 return do_proc_doulongvec_minmax(table, write, buffer, 2435 lenp, ppos, HZ, 1000l); 2436 } 2437 2438 2439 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp, 2440 int *valp, 2441 int write, void *data) 2442 { 2443 if (write) { 2444 if (*lvalp > LONG_MAX / HZ) 2445 return 1; 2446 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ); 2447 } else { 2448 int val = *valp; 2449 unsigned long lval; 2450 if (val < 0) { 2451 *negp = true; 2452 lval = -(unsigned long)val; 2453 } else { 2454 *negp = false; 2455 lval = (unsigned long)val; 2456 } 2457 *lvalp = lval / HZ; 2458 } 2459 return 0; 2460 } 2461 2462 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp, 2463 int *valp, 2464 int write, void *data) 2465 { 2466 if (write) { 2467 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ) 2468 return 1; 2469 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp); 2470 } else { 2471 int val = *valp; 2472 unsigned long lval; 2473 if (val < 0) { 2474 *negp = true; 2475 lval = -(unsigned long)val; 2476 } else { 2477 *negp = false; 2478 lval = (unsigned long)val; 2479 } 2480 *lvalp = jiffies_to_clock_t(lval); 2481 } 2482 return 0; 2483 } 2484 2485 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp, 2486 int *valp, 2487 int write, void *data) 2488 { 2489 if (write) { 2490 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp); 2491 2492 if (jif > INT_MAX) 2493 return 1; 2494 *valp = (int)jif; 2495 } else { 2496 int val = *valp; 2497 unsigned long lval; 2498 if (val < 0) { 2499 *negp = true; 2500 lval = -(unsigned long)val; 2501 } else { 2502 *negp = false; 2503 lval = (unsigned long)val; 2504 } 2505 *lvalp = jiffies_to_msecs(lval); 2506 } 2507 return 0; 2508 } 2509 2510 /** 2511 * proc_dointvec_jiffies - read a vector of integers as seconds 2512 * @table: the sysctl table 2513 * @write: %TRUE if this is a write to the sysctl file 2514 * @buffer: the user buffer 2515 * @lenp: the size of the user buffer 2516 * @ppos: file position 2517 * 2518 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2519 * values from/to the user buffer, treated as an ASCII string. 2520 * The values read are assumed to be in seconds, and are converted into 2521 * jiffies. 2522 * 2523 * Returns 0 on success. 2524 */ 2525 int proc_dointvec_jiffies(struct ctl_table *table, int write, 2526 void __user *buffer, size_t *lenp, loff_t *ppos) 2527 { 2528 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2529 do_proc_dointvec_jiffies_conv,NULL); 2530 } 2531 2532 /** 2533 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds 2534 * @table: the sysctl table 2535 * @write: %TRUE if this is a write to the sysctl file 2536 * @buffer: the user buffer 2537 * @lenp: the size of the user buffer 2538 * @ppos: pointer to the file position 2539 * 2540 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2541 * values from/to the user buffer, treated as an ASCII string. 2542 * The values read are assumed to be in 1/USER_HZ seconds, and 2543 * are converted into jiffies. 2544 * 2545 * Returns 0 on success. 2546 */ 2547 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 2548 void __user *buffer, size_t *lenp, loff_t *ppos) 2549 { 2550 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2551 do_proc_dointvec_userhz_jiffies_conv,NULL); 2552 } 2553 2554 /** 2555 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds 2556 * @table: the sysctl table 2557 * @write: %TRUE if this is a write to the sysctl file 2558 * @buffer: the user buffer 2559 * @lenp: the size of the user buffer 2560 * @ppos: file position 2561 * @ppos: the current position in the file 2562 * 2563 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2564 * values from/to the user buffer, treated as an ASCII string. 2565 * The values read are assumed to be in 1/1000 seconds, and 2566 * are converted into jiffies. 2567 * 2568 * Returns 0 on success. 2569 */ 2570 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, 2571 void __user *buffer, size_t *lenp, loff_t *ppos) 2572 { 2573 return do_proc_dointvec(table, write, buffer, lenp, ppos, 2574 do_proc_dointvec_ms_jiffies_conv, NULL); 2575 } 2576 2577 static int proc_do_cad_pid(struct ctl_table *table, int write, 2578 void __user *buffer, size_t *lenp, loff_t *ppos) 2579 { 2580 struct pid *new_pid; 2581 pid_t tmp; 2582 int r; 2583 2584 tmp = pid_vnr(cad_pid); 2585 2586 r = __do_proc_dointvec(&tmp, table, write, buffer, 2587 lenp, ppos, NULL, NULL); 2588 if (r || !write) 2589 return r; 2590 2591 new_pid = find_get_pid(tmp); 2592 if (!new_pid) 2593 return -ESRCH; 2594 2595 put_pid(xchg(&cad_pid, new_pid)); 2596 return 0; 2597 } 2598 2599 /** 2600 * proc_do_large_bitmap - read/write from/to a large bitmap 2601 * @table: the sysctl table 2602 * @write: %TRUE if this is a write to the sysctl file 2603 * @buffer: the user buffer 2604 * @lenp: the size of the user buffer 2605 * @ppos: file position 2606 * 2607 * The bitmap is stored at table->data and the bitmap length (in bits) 2608 * in table->maxlen. 2609 * 2610 * We use a range comma separated format (e.g. 1,3-4,10-10) so that 2611 * large bitmaps may be represented in a compact manner. Writing into 2612 * the file will clear the bitmap then update it with the given input. 2613 * 2614 * Returns 0 on success. 2615 */ 2616 int proc_do_large_bitmap(struct ctl_table *table, int write, 2617 void __user *buffer, size_t *lenp, loff_t *ppos) 2618 { 2619 int err = 0; 2620 bool first = 1; 2621 size_t left = *lenp; 2622 unsigned long bitmap_len = table->maxlen; 2623 unsigned long *bitmap = *(unsigned long **) table->data; 2624 unsigned long *tmp_bitmap = NULL; 2625 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c; 2626 2627 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) { 2628 *lenp = 0; 2629 return 0; 2630 } 2631 2632 if (write) { 2633 unsigned long page = 0; 2634 char *kbuf; 2635 2636 if (left > PAGE_SIZE - 1) 2637 left = PAGE_SIZE - 1; 2638 2639 page = __get_free_page(GFP_TEMPORARY); 2640 kbuf = (char *) page; 2641 if (!kbuf) 2642 return -ENOMEM; 2643 if (copy_from_user(kbuf, buffer, left)) { 2644 free_page(page); 2645 return -EFAULT; 2646 } 2647 kbuf[left] = 0; 2648 2649 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long), 2650 GFP_KERNEL); 2651 if (!tmp_bitmap) { 2652 free_page(page); 2653 return -ENOMEM; 2654 } 2655 proc_skip_char(&kbuf, &left, '\n'); 2656 while (!err && left) { 2657 unsigned long val_a, val_b; 2658 bool neg; 2659 2660 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a, 2661 sizeof(tr_a), &c); 2662 if (err) 2663 break; 2664 if (val_a >= bitmap_len || neg) { 2665 err = -EINVAL; 2666 break; 2667 } 2668 2669 val_b = val_a; 2670 if (left) { 2671 kbuf++; 2672 left--; 2673 } 2674 2675 if (c == '-') { 2676 err = proc_get_long(&kbuf, &left, &val_b, 2677 &neg, tr_b, sizeof(tr_b), 2678 &c); 2679 if (err) 2680 break; 2681 if (val_b >= bitmap_len || neg || 2682 val_a > val_b) { 2683 err = -EINVAL; 2684 break; 2685 } 2686 if (left) { 2687 kbuf++; 2688 left--; 2689 } 2690 } 2691 2692 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1); 2693 first = 0; 2694 proc_skip_char(&kbuf, &left, '\n'); 2695 } 2696 free_page(page); 2697 } else { 2698 unsigned long bit_a, bit_b = 0; 2699 2700 while (left) { 2701 bit_a = find_next_bit(bitmap, bitmap_len, bit_b); 2702 if (bit_a >= bitmap_len) 2703 break; 2704 bit_b = find_next_zero_bit(bitmap, bitmap_len, 2705 bit_a + 1) - 1; 2706 2707 if (!first) { 2708 err = proc_put_char(&buffer, &left, ','); 2709 if (err) 2710 break; 2711 } 2712 err = proc_put_long(&buffer, &left, bit_a, false); 2713 if (err) 2714 break; 2715 if (bit_a != bit_b) { 2716 err = proc_put_char(&buffer, &left, '-'); 2717 if (err) 2718 break; 2719 err = proc_put_long(&buffer, &left, bit_b, false); 2720 if (err) 2721 break; 2722 } 2723 2724 first = 0; bit_b++; 2725 } 2726 if (!err) 2727 err = proc_put_char(&buffer, &left, '\n'); 2728 } 2729 2730 if (!err) { 2731 if (write) { 2732 if (*ppos) 2733 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len); 2734 else 2735 bitmap_copy(bitmap, tmp_bitmap, bitmap_len); 2736 } 2737 kfree(tmp_bitmap); 2738 *lenp -= left; 2739 *ppos += *lenp; 2740 return 0; 2741 } else { 2742 kfree(tmp_bitmap); 2743 return err; 2744 } 2745 } 2746 2747 #else /* CONFIG_PROC_SYSCTL */ 2748 2749 int proc_dostring(struct ctl_table *table, int write, 2750 void __user *buffer, size_t *lenp, loff_t *ppos) 2751 { 2752 return -ENOSYS; 2753 } 2754 2755 int proc_dointvec(struct ctl_table *table, int write, 2756 void __user *buffer, size_t *lenp, loff_t *ppos) 2757 { 2758 return -ENOSYS; 2759 } 2760 2761 int proc_dointvec_minmax(struct ctl_table *table, int write, 2762 void __user *buffer, size_t *lenp, loff_t *ppos) 2763 { 2764 return -ENOSYS; 2765 } 2766 2767 int proc_dointvec_jiffies(struct ctl_table *table, int write, 2768 void __user *buffer, size_t *lenp, loff_t *ppos) 2769 { 2770 return -ENOSYS; 2771 } 2772 2773 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 2774 void __user *buffer, size_t *lenp, loff_t *ppos) 2775 { 2776 return -ENOSYS; 2777 } 2778 2779 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, 2780 void __user *buffer, size_t *lenp, loff_t *ppos) 2781 { 2782 return -ENOSYS; 2783 } 2784 2785 int proc_doulongvec_minmax(struct ctl_table *table, int write, 2786 void __user *buffer, size_t *lenp, loff_t *ppos) 2787 { 2788 return -ENOSYS; 2789 } 2790 2791 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 2792 void __user *buffer, 2793 size_t *lenp, loff_t *ppos) 2794 { 2795 return -ENOSYS; 2796 } 2797 2798 2799 #endif /* CONFIG_PROC_SYSCTL */ 2800 2801 /* 2802 * No sense putting this after each symbol definition, twice, 2803 * exception granted :-) 2804 */ 2805 EXPORT_SYMBOL(proc_dointvec); 2806 EXPORT_SYMBOL(proc_dointvec_jiffies); 2807 EXPORT_SYMBOL(proc_dointvec_minmax); 2808 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies); 2809 EXPORT_SYMBOL(proc_dointvec_ms_jiffies); 2810 EXPORT_SYMBOL(proc_dostring); 2811 EXPORT_SYMBOL(proc_doulongvec_minmax); 2812 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax); 2813