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