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/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 #include <linux/sched/coredump.h> 66 #include <linux/kexec.h> 67 #include <linux/bpf.h> 68 #include <linux/mount.h> 69 70 #include <linux/uaccess.h> 71 #include <asm/processor.h> 72 73 #ifdef CONFIG_X86 74 #include <asm/nmi.h> 75 #include <asm/stacktrace.h> 76 #include <asm/io.h> 77 #endif 78 #ifdef CONFIG_SPARC 79 #include <asm/setup.h> 80 #endif 81 #ifdef CONFIG_BSD_PROCESS_ACCT 82 #include <linux/acct.h> 83 #endif 84 #ifdef CONFIG_RT_MUTEXES 85 #include <linux/rtmutex.h> 86 #endif 87 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT) 88 #include <linux/lockdep.h> 89 #endif 90 #ifdef CONFIG_CHR_DEV_SG 91 #include <scsi/sg.h> 92 #endif 93 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE 94 #include <linux/stackleak.h> 95 #endif 96 #ifdef CONFIG_LOCKUP_DETECTOR 97 #include <linux/nmi.h> 98 #endif 99 100 #if defined(CONFIG_SYSCTL) 101 102 /* External variables not in a header file. */ 103 extern int suid_dumpable; 104 #ifdef CONFIG_COREDUMP 105 extern int core_uses_pid; 106 extern char core_pattern[]; 107 extern unsigned int core_pipe_limit; 108 #endif 109 extern int pid_max; 110 extern int pid_max_min, pid_max_max; 111 extern int percpu_pagelist_fraction; 112 extern int latencytop_enabled; 113 extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max; 114 #ifndef CONFIG_MMU 115 extern int sysctl_nr_trim_pages; 116 #endif 117 118 /* Constants used for minimum and maximum */ 119 #ifdef CONFIG_LOCKUP_DETECTOR 120 static int sixty = 60; 121 #endif 122 123 static int __maybe_unused neg_one = -1; 124 125 static int zero; 126 static int __maybe_unused one = 1; 127 static int __maybe_unused two = 2; 128 static int __maybe_unused four = 4; 129 static unsigned long one_ul = 1; 130 static int one_hundred = 100; 131 static int one_thousand = 1000; 132 #ifdef CONFIG_PRINTK 133 static int ten_thousand = 10000; 134 #endif 135 #ifdef CONFIG_PERF_EVENTS 136 static int six_hundred_forty_kb = 640 * 1024; 137 #endif 138 139 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */ 140 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE; 141 142 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */ 143 static int maxolduid = 65535; 144 static int minolduid; 145 146 static int ngroups_max = NGROUPS_MAX; 147 static const int cap_last_cap = CAP_LAST_CAP; 148 149 /* 150 * This is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs 151 * and hung_task_check_interval_secs 152 */ 153 #ifdef CONFIG_DETECT_HUNG_TASK 154 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ); 155 #endif 156 157 #ifdef CONFIG_INOTIFY_USER 158 #include <linux/inotify.h> 159 #endif 160 #ifdef CONFIG_SPARC 161 #endif 162 163 #ifdef __hppa__ 164 extern int pwrsw_enabled; 165 #endif 166 167 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW 168 extern int unaligned_enabled; 169 #endif 170 171 #ifdef CONFIG_IA64 172 extern int unaligned_dump_stack; 173 #endif 174 175 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN 176 extern int no_unaligned_warning; 177 #endif 178 179 #ifdef CONFIG_PROC_SYSCTL 180 181 /** 182 * enum sysctl_writes_mode - supported sysctl write modes 183 * 184 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value 185 * to be written, and multiple writes on the same sysctl file descriptor 186 * will rewrite the sysctl value, regardless of file position. No warning 187 * is issued when the initial position is not 0. 188 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is 189 * not 0. 190 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at 191 * file position 0 and the value must be fully contained in the buffer 192 * sent to the write syscall. If dealing with strings respect the file 193 * position, but restrict this to the max length of the buffer, anything 194 * passed the max lenght will be ignored. Multiple writes will append 195 * to the buffer. 196 * 197 * These write modes control how current file position affects the behavior of 198 * updating sysctl values through the proc interface on each write. 199 */ 200 enum sysctl_writes_mode { 201 SYSCTL_WRITES_LEGACY = -1, 202 SYSCTL_WRITES_WARN = 0, 203 SYSCTL_WRITES_STRICT = 1, 204 }; 205 206 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT; 207 208 static int proc_do_cad_pid(struct ctl_table *table, int write, 209 void __user *buffer, size_t *lenp, loff_t *ppos); 210 static int proc_taint(struct ctl_table *table, int write, 211 void __user *buffer, size_t *lenp, loff_t *ppos); 212 #endif 213 214 #ifdef CONFIG_PRINTK 215 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write, 216 void __user *buffer, size_t *lenp, loff_t *ppos); 217 #endif 218 219 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write, 220 void __user *buffer, size_t *lenp, loff_t *ppos); 221 #ifdef CONFIG_COREDUMP 222 static int proc_dostring_coredump(struct ctl_table *table, int write, 223 void __user *buffer, size_t *lenp, loff_t *ppos); 224 #endif 225 static int proc_dopipe_max_size(struct ctl_table *table, int write, 226 void __user *buffer, size_t *lenp, loff_t *ppos); 227 228 #ifdef CONFIG_MAGIC_SYSRQ 229 /* Note: sysrq code uses its own private copy */ 230 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE; 231 232 static int sysrq_sysctl_handler(struct ctl_table *table, int write, 233 void __user *buffer, size_t *lenp, 234 loff_t *ppos) 235 { 236 int error; 237 238 error = proc_dointvec(table, write, buffer, lenp, ppos); 239 if (error) 240 return error; 241 242 if (write) 243 sysrq_toggle_support(__sysrq_enabled); 244 245 return 0; 246 } 247 248 #endif 249 250 static struct ctl_table kern_table[]; 251 static struct ctl_table vm_table[]; 252 static struct ctl_table fs_table[]; 253 static struct ctl_table debug_table[]; 254 static struct ctl_table dev_table[]; 255 extern struct ctl_table random_table[]; 256 #ifdef CONFIG_EPOLL 257 extern struct ctl_table epoll_table[]; 258 #endif 259 260 #ifdef CONFIG_FW_LOADER_USER_HELPER 261 extern struct ctl_table firmware_config_table[]; 262 #endif 263 264 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT 265 int sysctl_legacy_va_layout; 266 #endif 267 268 /* The default sysctl tables: */ 269 270 static struct ctl_table sysctl_base_table[] = { 271 { 272 .procname = "kernel", 273 .mode = 0555, 274 .child = kern_table, 275 }, 276 { 277 .procname = "vm", 278 .mode = 0555, 279 .child = vm_table, 280 }, 281 { 282 .procname = "fs", 283 .mode = 0555, 284 .child = fs_table, 285 }, 286 { 287 .procname = "debug", 288 .mode = 0555, 289 .child = debug_table, 290 }, 291 { 292 .procname = "dev", 293 .mode = 0555, 294 .child = dev_table, 295 }, 296 { } 297 }; 298 299 #ifdef CONFIG_SCHED_DEBUG 300 static int min_sched_granularity_ns = 100000; /* 100 usecs */ 301 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */ 302 static int min_wakeup_granularity_ns; /* 0 usecs */ 303 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */ 304 #ifdef CONFIG_SMP 305 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE; 306 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1; 307 #endif /* CONFIG_SMP */ 308 #endif /* CONFIG_SCHED_DEBUG */ 309 310 #ifdef CONFIG_COMPACTION 311 static int min_extfrag_threshold; 312 static int max_extfrag_threshold = 1000; 313 #endif 314 315 static struct ctl_table kern_table[] = { 316 { 317 .procname = "sched_child_runs_first", 318 .data = &sysctl_sched_child_runs_first, 319 .maxlen = sizeof(unsigned int), 320 .mode = 0644, 321 .proc_handler = proc_dointvec, 322 }, 323 #ifdef CONFIG_SCHED_DEBUG 324 { 325 .procname = "sched_min_granularity_ns", 326 .data = &sysctl_sched_min_granularity, 327 .maxlen = sizeof(unsigned int), 328 .mode = 0644, 329 .proc_handler = sched_proc_update_handler, 330 .extra1 = &min_sched_granularity_ns, 331 .extra2 = &max_sched_granularity_ns, 332 }, 333 { 334 .procname = "sched_latency_ns", 335 .data = &sysctl_sched_latency, 336 .maxlen = sizeof(unsigned int), 337 .mode = 0644, 338 .proc_handler = sched_proc_update_handler, 339 .extra1 = &min_sched_granularity_ns, 340 .extra2 = &max_sched_granularity_ns, 341 }, 342 { 343 .procname = "sched_wakeup_granularity_ns", 344 .data = &sysctl_sched_wakeup_granularity, 345 .maxlen = sizeof(unsigned int), 346 .mode = 0644, 347 .proc_handler = sched_proc_update_handler, 348 .extra1 = &min_wakeup_granularity_ns, 349 .extra2 = &max_wakeup_granularity_ns, 350 }, 351 #ifdef CONFIG_SMP 352 { 353 .procname = "sched_tunable_scaling", 354 .data = &sysctl_sched_tunable_scaling, 355 .maxlen = sizeof(enum sched_tunable_scaling), 356 .mode = 0644, 357 .proc_handler = sched_proc_update_handler, 358 .extra1 = &min_sched_tunable_scaling, 359 .extra2 = &max_sched_tunable_scaling, 360 }, 361 { 362 .procname = "sched_migration_cost_ns", 363 .data = &sysctl_sched_migration_cost, 364 .maxlen = sizeof(unsigned int), 365 .mode = 0644, 366 .proc_handler = proc_dointvec, 367 }, 368 { 369 .procname = "sched_nr_migrate", 370 .data = &sysctl_sched_nr_migrate, 371 .maxlen = sizeof(unsigned int), 372 .mode = 0644, 373 .proc_handler = proc_dointvec, 374 }, 375 #ifdef CONFIG_SCHEDSTATS 376 { 377 .procname = "sched_schedstats", 378 .data = NULL, 379 .maxlen = sizeof(unsigned int), 380 .mode = 0644, 381 .proc_handler = sysctl_schedstats, 382 .extra1 = &zero, 383 .extra2 = &one, 384 }, 385 #endif /* CONFIG_SCHEDSTATS */ 386 #endif /* CONFIG_SMP */ 387 #ifdef CONFIG_NUMA_BALANCING 388 { 389 .procname = "numa_balancing_scan_delay_ms", 390 .data = &sysctl_numa_balancing_scan_delay, 391 .maxlen = sizeof(unsigned int), 392 .mode = 0644, 393 .proc_handler = proc_dointvec, 394 }, 395 { 396 .procname = "numa_balancing_scan_period_min_ms", 397 .data = &sysctl_numa_balancing_scan_period_min, 398 .maxlen = sizeof(unsigned int), 399 .mode = 0644, 400 .proc_handler = proc_dointvec, 401 }, 402 { 403 .procname = "numa_balancing_scan_period_max_ms", 404 .data = &sysctl_numa_balancing_scan_period_max, 405 .maxlen = sizeof(unsigned int), 406 .mode = 0644, 407 .proc_handler = proc_dointvec, 408 }, 409 { 410 .procname = "numa_balancing_scan_size_mb", 411 .data = &sysctl_numa_balancing_scan_size, 412 .maxlen = sizeof(unsigned int), 413 .mode = 0644, 414 .proc_handler = proc_dointvec_minmax, 415 .extra1 = &one, 416 }, 417 { 418 .procname = "numa_balancing", 419 .data = NULL, /* filled in by handler */ 420 .maxlen = sizeof(unsigned int), 421 .mode = 0644, 422 .proc_handler = sysctl_numa_balancing, 423 .extra1 = &zero, 424 .extra2 = &one, 425 }, 426 #endif /* CONFIG_NUMA_BALANCING */ 427 #endif /* CONFIG_SCHED_DEBUG */ 428 { 429 .procname = "sched_rt_period_us", 430 .data = &sysctl_sched_rt_period, 431 .maxlen = sizeof(unsigned int), 432 .mode = 0644, 433 .proc_handler = sched_rt_handler, 434 }, 435 { 436 .procname = "sched_rt_runtime_us", 437 .data = &sysctl_sched_rt_runtime, 438 .maxlen = sizeof(int), 439 .mode = 0644, 440 .proc_handler = sched_rt_handler, 441 }, 442 { 443 .procname = "sched_rr_timeslice_ms", 444 .data = &sysctl_sched_rr_timeslice, 445 .maxlen = sizeof(int), 446 .mode = 0644, 447 .proc_handler = sched_rr_handler, 448 }, 449 #ifdef CONFIG_SCHED_AUTOGROUP 450 { 451 .procname = "sched_autogroup_enabled", 452 .data = &sysctl_sched_autogroup_enabled, 453 .maxlen = sizeof(unsigned int), 454 .mode = 0644, 455 .proc_handler = proc_dointvec_minmax, 456 .extra1 = &zero, 457 .extra2 = &one, 458 }, 459 #endif 460 #ifdef CONFIG_CFS_BANDWIDTH 461 { 462 .procname = "sched_cfs_bandwidth_slice_us", 463 .data = &sysctl_sched_cfs_bandwidth_slice, 464 .maxlen = sizeof(unsigned int), 465 .mode = 0644, 466 .proc_handler = proc_dointvec_minmax, 467 .extra1 = &one, 468 }, 469 #endif 470 #ifdef CONFIG_PROVE_LOCKING 471 { 472 .procname = "prove_locking", 473 .data = &prove_locking, 474 .maxlen = sizeof(int), 475 .mode = 0644, 476 .proc_handler = proc_dointvec, 477 }, 478 #endif 479 #ifdef CONFIG_LOCK_STAT 480 { 481 .procname = "lock_stat", 482 .data = &lock_stat, 483 .maxlen = sizeof(int), 484 .mode = 0644, 485 .proc_handler = proc_dointvec, 486 }, 487 #endif 488 { 489 .procname = "panic", 490 .data = &panic_timeout, 491 .maxlen = sizeof(int), 492 .mode = 0644, 493 .proc_handler = proc_dointvec, 494 }, 495 #ifdef CONFIG_COREDUMP 496 { 497 .procname = "core_uses_pid", 498 .data = &core_uses_pid, 499 .maxlen = sizeof(int), 500 .mode = 0644, 501 .proc_handler = proc_dointvec, 502 }, 503 { 504 .procname = "core_pattern", 505 .data = core_pattern, 506 .maxlen = CORENAME_MAX_SIZE, 507 .mode = 0644, 508 .proc_handler = proc_dostring_coredump, 509 }, 510 { 511 .procname = "core_pipe_limit", 512 .data = &core_pipe_limit, 513 .maxlen = sizeof(unsigned int), 514 .mode = 0644, 515 .proc_handler = proc_dointvec, 516 }, 517 #endif 518 #ifdef CONFIG_PROC_SYSCTL 519 { 520 .procname = "tainted", 521 .maxlen = sizeof(long), 522 .mode = 0644, 523 .proc_handler = proc_taint, 524 }, 525 { 526 .procname = "sysctl_writes_strict", 527 .data = &sysctl_writes_strict, 528 .maxlen = sizeof(int), 529 .mode = 0644, 530 .proc_handler = proc_dointvec_minmax, 531 .extra1 = &neg_one, 532 .extra2 = &one, 533 }, 534 #endif 535 #ifdef CONFIG_LATENCYTOP 536 { 537 .procname = "latencytop", 538 .data = &latencytop_enabled, 539 .maxlen = sizeof(int), 540 .mode = 0644, 541 .proc_handler = sysctl_latencytop, 542 }, 543 #endif 544 #ifdef CONFIG_BLK_DEV_INITRD 545 { 546 .procname = "real-root-dev", 547 .data = &real_root_dev, 548 .maxlen = sizeof(int), 549 .mode = 0644, 550 .proc_handler = proc_dointvec, 551 }, 552 #endif 553 { 554 .procname = "print-fatal-signals", 555 .data = &print_fatal_signals, 556 .maxlen = sizeof(int), 557 .mode = 0644, 558 .proc_handler = proc_dointvec, 559 }, 560 #ifdef CONFIG_SPARC 561 { 562 .procname = "reboot-cmd", 563 .data = reboot_command, 564 .maxlen = 256, 565 .mode = 0644, 566 .proc_handler = proc_dostring, 567 }, 568 { 569 .procname = "stop-a", 570 .data = &stop_a_enabled, 571 .maxlen = sizeof (int), 572 .mode = 0644, 573 .proc_handler = proc_dointvec, 574 }, 575 { 576 .procname = "scons-poweroff", 577 .data = &scons_pwroff, 578 .maxlen = sizeof (int), 579 .mode = 0644, 580 .proc_handler = proc_dointvec, 581 }, 582 #endif 583 #ifdef CONFIG_SPARC64 584 { 585 .procname = "tsb-ratio", 586 .data = &sysctl_tsb_ratio, 587 .maxlen = sizeof (int), 588 .mode = 0644, 589 .proc_handler = proc_dointvec, 590 }, 591 #endif 592 #ifdef __hppa__ 593 { 594 .procname = "soft-power", 595 .data = &pwrsw_enabled, 596 .maxlen = sizeof (int), 597 .mode = 0644, 598 .proc_handler = proc_dointvec, 599 }, 600 #endif 601 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW 602 { 603 .procname = "unaligned-trap", 604 .data = &unaligned_enabled, 605 .maxlen = sizeof (int), 606 .mode = 0644, 607 .proc_handler = proc_dointvec, 608 }, 609 #endif 610 { 611 .procname = "ctrl-alt-del", 612 .data = &C_A_D, 613 .maxlen = sizeof(int), 614 .mode = 0644, 615 .proc_handler = proc_dointvec, 616 }, 617 #ifdef CONFIG_FUNCTION_TRACER 618 { 619 .procname = "ftrace_enabled", 620 .data = &ftrace_enabled, 621 .maxlen = sizeof(int), 622 .mode = 0644, 623 .proc_handler = ftrace_enable_sysctl, 624 }, 625 #endif 626 #ifdef CONFIG_STACK_TRACER 627 { 628 .procname = "stack_tracer_enabled", 629 .data = &stack_tracer_enabled, 630 .maxlen = sizeof(int), 631 .mode = 0644, 632 .proc_handler = stack_trace_sysctl, 633 }, 634 #endif 635 #ifdef CONFIG_TRACING 636 { 637 .procname = "ftrace_dump_on_oops", 638 .data = &ftrace_dump_on_oops, 639 .maxlen = sizeof(int), 640 .mode = 0644, 641 .proc_handler = proc_dointvec, 642 }, 643 { 644 .procname = "traceoff_on_warning", 645 .data = &__disable_trace_on_warning, 646 .maxlen = sizeof(__disable_trace_on_warning), 647 .mode = 0644, 648 .proc_handler = proc_dointvec, 649 }, 650 { 651 .procname = "tracepoint_printk", 652 .data = &tracepoint_printk, 653 .maxlen = sizeof(tracepoint_printk), 654 .mode = 0644, 655 .proc_handler = tracepoint_printk_sysctl, 656 }, 657 #endif 658 #ifdef CONFIG_KEXEC_CORE 659 { 660 .procname = "kexec_load_disabled", 661 .data = &kexec_load_disabled, 662 .maxlen = sizeof(int), 663 .mode = 0644, 664 /* only handle a transition from default "0" to "1" */ 665 .proc_handler = proc_dointvec_minmax, 666 .extra1 = &one, 667 .extra2 = &one, 668 }, 669 #endif 670 #ifdef CONFIG_MODULES 671 { 672 .procname = "modprobe", 673 .data = &modprobe_path, 674 .maxlen = KMOD_PATH_LEN, 675 .mode = 0644, 676 .proc_handler = proc_dostring, 677 }, 678 { 679 .procname = "modules_disabled", 680 .data = &modules_disabled, 681 .maxlen = sizeof(int), 682 .mode = 0644, 683 /* only handle a transition from default "0" to "1" */ 684 .proc_handler = proc_dointvec_minmax, 685 .extra1 = &one, 686 .extra2 = &one, 687 }, 688 #endif 689 #ifdef CONFIG_UEVENT_HELPER 690 { 691 .procname = "hotplug", 692 .data = &uevent_helper, 693 .maxlen = UEVENT_HELPER_PATH_LEN, 694 .mode = 0644, 695 .proc_handler = proc_dostring, 696 }, 697 #endif 698 #ifdef CONFIG_CHR_DEV_SG 699 { 700 .procname = "sg-big-buff", 701 .data = &sg_big_buff, 702 .maxlen = sizeof (int), 703 .mode = 0444, 704 .proc_handler = proc_dointvec, 705 }, 706 #endif 707 #ifdef CONFIG_BSD_PROCESS_ACCT 708 { 709 .procname = "acct", 710 .data = &acct_parm, 711 .maxlen = 3*sizeof(int), 712 .mode = 0644, 713 .proc_handler = proc_dointvec, 714 }, 715 #endif 716 #ifdef CONFIG_MAGIC_SYSRQ 717 { 718 .procname = "sysrq", 719 .data = &__sysrq_enabled, 720 .maxlen = sizeof (int), 721 .mode = 0644, 722 .proc_handler = sysrq_sysctl_handler, 723 }, 724 #endif 725 #ifdef CONFIG_PROC_SYSCTL 726 { 727 .procname = "cad_pid", 728 .data = NULL, 729 .maxlen = sizeof (int), 730 .mode = 0600, 731 .proc_handler = proc_do_cad_pid, 732 }, 733 #endif 734 { 735 .procname = "threads-max", 736 .data = NULL, 737 .maxlen = sizeof(int), 738 .mode = 0644, 739 .proc_handler = sysctl_max_threads, 740 }, 741 { 742 .procname = "random", 743 .mode = 0555, 744 .child = random_table, 745 }, 746 { 747 .procname = "usermodehelper", 748 .mode = 0555, 749 .child = usermodehelper_table, 750 }, 751 #ifdef CONFIG_FW_LOADER_USER_HELPER 752 { 753 .procname = "firmware_config", 754 .mode = 0555, 755 .child = firmware_config_table, 756 }, 757 #endif 758 { 759 .procname = "overflowuid", 760 .data = &overflowuid, 761 .maxlen = sizeof(int), 762 .mode = 0644, 763 .proc_handler = proc_dointvec_minmax, 764 .extra1 = &minolduid, 765 .extra2 = &maxolduid, 766 }, 767 { 768 .procname = "overflowgid", 769 .data = &overflowgid, 770 .maxlen = sizeof(int), 771 .mode = 0644, 772 .proc_handler = proc_dointvec_minmax, 773 .extra1 = &minolduid, 774 .extra2 = &maxolduid, 775 }, 776 #ifdef CONFIG_S390 777 #ifdef CONFIG_MATHEMU 778 { 779 .procname = "ieee_emulation_warnings", 780 .data = &sysctl_ieee_emulation_warnings, 781 .maxlen = sizeof(int), 782 .mode = 0644, 783 .proc_handler = proc_dointvec, 784 }, 785 #endif 786 { 787 .procname = "userprocess_debug", 788 .data = &show_unhandled_signals, 789 .maxlen = sizeof(int), 790 .mode = 0644, 791 .proc_handler = proc_dointvec, 792 }, 793 #endif 794 { 795 .procname = "pid_max", 796 .data = &pid_max, 797 .maxlen = sizeof (int), 798 .mode = 0644, 799 .proc_handler = proc_dointvec_minmax, 800 .extra1 = &pid_max_min, 801 .extra2 = &pid_max_max, 802 }, 803 { 804 .procname = "panic_on_oops", 805 .data = &panic_on_oops, 806 .maxlen = sizeof(int), 807 .mode = 0644, 808 .proc_handler = proc_dointvec, 809 }, 810 #if defined CONFIG_PRINTK 811 { 812 .procname = "printk", 813 .data = &console_loglevel, 814 .maxlen = 4*sizeof(int), 815 .mode = 0644, 816 .proc_handler = proc_dointvec, 817 }, 818 { 819 .procname = "printk_ratelimit", 820 .data = &printk_ratelimit_state.interval, 821 .maxlen = sizeof(int), 822 .mode = 0644, 823 .proc_handler = proc_dointvec_jiffies, 824 }, 825 { 826 .procname = "printk_ratelimit_burst", 827 .data = &printk_ratelimit_state.burst, 828 .maxlen = sizeof(int), 829 .mode = 0644, 830 .proc_handler = proc_dointvec, 831 }, 832 { 833 .procname = "printk_delay", 834 .data = &printk_delay_msec, 835 .maxlen = sizeof(int), 836 .mode = 0644, 837 .proc_handler = proc_dointvec_minmax, 838 .extra1 = &zero, 839 .extra2 = &ten_thousand, 840 }, 841 { 842 .procname = "printk_devkmsg", 843 .data = devkmsg_log_str, 844 .maxlen = DEVKMSG_STR_MAX_SIZE, 845 .mode = 0644, 846 .proc_handler = devkmsg_sysctl_set_loglvl, 847 }, 848 { 849 .procname = "dmesg_restrict", 850 .data = &dmesg_restrict, 851 .maxlen = sizeof(int), 852 .mode = 0644, 853 .proc_handler = proc_dointvec_minmax_sysadmin, 854 .extra1 = &zero, 855 .extra2 = &one, 856 }, 857 { 858 .procname = "kptr_restrict", 859 .data = &kptr_restrict, 860 .maxlen = sizeof(int), 861 .mode = 0644, 862 .proc_handler = proc_dointvec_minmax_sysadmin, 863 .extra1 = &zero, 864 .extra2 = &two, 865 }, 866 #endif 867 { 868 .procname = "ngroups_max", 869 .data = &ngroups_max, 870 .maxlen = sizeof (int), 871 .mode = 0444, 872 .proc_handler = proc_dointvec, 873 }, 874 { 875 .procname = "cap_last_cap", 876 .data = (void *)&cap_last_cap, 877 .maxlen = sizeof(int), 878 .mode = 0444, 879 .proc_handler = proc_dointvec, 880 }, 881 #if defined(CONFIG_LOCKUP_DETECTOR) 882 { 883 .procname = "watchdog", 884 .data = &watchdog_user_enabled, 885 .maxlen = sizeof(int), 886 .mode = 0644, 887 .proc_handler = proc_watchdog, 888 .extra1 = &zero, 889 .extra2 = &one, 890 }, 891 { 892 .procname = "watchdog_thresh", 893 .data = &watchdog_thresh, 894 .maxlen = sizeof(int), 895 .mode = 0644, 896 .proc_handler = proc_watchdog_thresh, 897 .extra1 = &zero, 898 .extra2 = &sixty, 899 }, 900 { 901 .procname = "nmi_watchdog", 902 .data = &nmi_watchdog_user_enabled, 903 .maxlen = sizeof(int), 904 .mode = NMI_WATCHDOG_SYSCTL_PERM, 905 .proc_handler = proc_nmi_watchdog, 906 .extra1 = &zero, 907 .extra2 = &one, 908 }, 909 { 910 .procname = "watchdog_cpumask", 911 .data = &watchdog_cpumask_bits, 912 .maxlen = NR_CPUS, 913 .mode = 0644, 914 .proc_handler = proc_watchdog_cpumask, 915 }, 916 #ifdef CONFIG_SOFTLOCKUP_DETECTOR 917 { 918 .procname = "soft_watchdog", 919 .data = &soft_watchdog_user_enabled, 920 .maxlen = sizeof(int), 921 .mode = 0644, 922 .proc_handler = proc_soft_watchdog, 923 .extra1 = &zero, 924 .extra2 = &one, 925 }, 926 { 927 .procname = "softlockup_panic", 928 .data = &softlockup_panic, 929 .maxlen = sizeof(int), 930 .mode = 0644, 931 .proc_handler = proc_dointvec_minmax, 932 .extra1 = &zero, 933 .extra2 = &one, 934 }, 935 #ifdef CONFIG_SMP 936 { 937 .procname = "softlockup_all_cpu_backtrace", 938 .data = &sysctl_softlockup_all_cpu_backtrace, 939 .maxlen = sizeof(int), 940 .mode = 0644, 941 .proc_handler = proc_dointvec_minmax, 942 .extra1 = &zero, 943 .extra2 = &one, 944 }, 945 #endif /* CONFIG_SMP */ 946 #endif 947 #ifdef CONFIG_HARDLOCKUP_DETECTOR 948 { 949 .procname = "hardlockup_panic", 950 .data = &hardlockup_panic, 951 .maxlen = sizeof(int), 952 .mode = 0644, 953 .proc_handler = proc_dointvec_minmax, 954 .extra1 = &zero, 955 .extra2 = &one, 956 }, 957 #ifdef CONFIG_SMP 958 { 959 .procname = "hardlockup_all_cpu_backtrace", 960 .data = &sysctl_hardlockup_all_cpu_backtrace, 961 .maxlen = sizeof(int), 962 .mode = 0644, 963 .proc_handler = proc_dointvec_minmax, 964 .extra1 = &zero, 965 .extra2 = &one, 966 }, 967 #endif /* CONFIG_SMP */ 968 #endif 969 #endif 970 971 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86) 972 { 973 .procname = "unknown_nmi_panic", 974 .data = &unknown_nmi_panic, 975 .maxlen = sizeof (int), 976 .mode = 0644, 977 .proc_handler = proc_dointvec, 978 }, 979 #endif 980 #if defined(CONFIG_X86) 981 { 982 .procname = "panic_on_unrecovered_nmi", 983 .data = &panic_on_unrecovered_nmi, 984 .maxlen = sizeof(int), 985 .mode = 0644, 986 .proc_handler = proc_dointvec, 987 }, 988 { 989 .procname = "panic_on_io_nmi", 990 .data = &panic_on_io_nmi, 991 .maxlen = sizeof(int), 992 .mode = 0644, 993 .proc_handler = proc_dointvec, 994 }, 995 #ifdef CONFIG_DEBUG_STACKOVERFLOW 996 { 997 .procname = "panic_on_stackoverflow", 998 .data = &sysctl_panic_on_stackoverflow, 999 .maxlen = sizeof(int), 1000 .mode = 0644, 1001 .proc_handler = proc_dointvec, 1002 }, 1003 #endif 1004 { 1005 .procname = "bootloader_type", 1006 .data = &bootloader_type, 1007 .maxlen = sizeof (int), 1008 .mode = 0444, 1009 .proc_handler = proc_dointvec, 1010 }, 1011 { 1012 .procname = "bootloader_version", 1013 .data = &bootloader_version, 1014 .maxlen = sizeof (int), 1015 .mode = 0444, 1016 .proc_handler = proc_dointvec, 1017 }, 1018 { 1019 .procname = "io_delay_type", 1020 .data = &io_delay_type, 1021 .maxlen = sizeof(int), 1022 .mode = 0644, 1023 .proc_handler = proc_dointvec, 1024 }, 1025 #endif 1026 #if defined(CONFIG_MMU) 1027 { 1028 .procname = "randomize_va_space", 1029 .data = &randomize_va_space, 1030 .maxlen = sizeof(int), 1031 .mode = 0644, 1032 .proc_handler = proc_dointvec, 1033 }, 1034 #endif 1035 #if defined(CONFIG_S390) && defined(CONFIG_SMP) 1036 { 1037 .procname = "spin_retry", 1038 .data = &spin_retry, 1039 .maxlen = sizeof (int), 1040 .mode = 0644, 1041 .proc_handler = proc_dointvec, 1042 }, 1043 #endif 1044 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86) 1045 { 1046 .procname = "acpi_video_flags", 1047 .data = &acpi_realmode_flags, 1048 .maxlen = sizeof (unsigned long), 1049 .mode = 0644, 1050 .proc_handler = proc_doulongvec_minmax, 1051 }, 1052 #endif 1053 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN 1054 { 1055 .procname = "ignore-unaligned-usertrap", 1056 .data = &no_unaligned_warning, 1057 .maxlen = sizeof (int), 1058 .mode = 0644, 1059 .proc_handler = proc_dointvec, 1060 }, 1061 #endif 1062 #ifdef CONFIG_IA64 1063 { 1064 .procname = "unaligned-dump-stack", 1065 .data = &unaligned_dump_stack, 1066 .maxlen = sizeof (int), 1067 .mode = 0644, 1068 .proc_handler = proc_dointvec, 1069 }, 1070 #endif 1071 #ifdef CONFIG_DETECT_HUNG_TASK 1072 { 1073 .procname = "hung_task_panic", 1074 .data = &sysctl_hung_task_panic, 1075 .maxlen = sizeof(int), 1076 .mode = 0644, 1077 .proc_handler = proc_dointvec_minmax, 1078 .extra1 = &zero, 1079 .extra2 = &one, 1080 }, 1081 { 1082 .procname = "hung_task_check_count", 1083 .data = &sysctl_hung_task_check_count, 1084 .maxlen = sizeof(int), 1085 .mode = 0644, 1086 .proc_handler = proc_dointvec_minmax, 1087 .extra1 = &zero, 1088 }, 1089 { 1090 .procname = "hung_task_timeout_secs", 1091 .data = &sysctl_hung_task_timeout_secs, 1092 .maxlen = sizeof(unsigned long), 1093 .mode = 0644, 1094 .proc_handler = proc_dohung_task_timeout_secs, 1095 .extra2 = &hung_task_timeout_max, 1096 }, 1097 { 1098 .procname = "hung_task_check_interval_secs", 1099 .data = &sysctl_hung_task_check_interval_secs, 1100 .maxlen = sizeof(unsigned long), 1101 .mode = 0644, 1102 .proc_handler = proc_dohung_task_timeout_secs, 1103 .extra2 = &hung_task_timeout_max, 1104 }, 1105 { 1106 .procname = "hung_task_warnings", 1107 .data = &sysctl_hung_task_warnings, 1108 .maxlen = sizeof(int), 1109 .mode = 0644, 1110 .proc_handler = proc_dointvec_minmax, 1111 .extra1 = &neg_one, 1112 }, 1113 #endif 1114 #ifdef CONFIG_RT_MUTEXES 1115 { 1116 .procname = "max_lock_depth", 1117 .data = &max_lock_depth, 1118 .maxlen = sizeof(int), 1119 .mode = 0644, 1120 .proc_handler = proc_dointvec, 1121 }, 1122 #endif 1123 { 1124 .procname = "poweroff_cmd", 1125 .data = &poweroff_cmd, 1126 .maxlen = POWEROFF_CMD_PATH_LEN, 1127 .mode = 0644, 1128 .proc_handler = proc_dostring, 1129 }, 1130 #ifdef CONFIG_KEYS 1131 { 1132 .procname = "keys", 1133 .mode = 0555, 1134 .child = key_sysctls, 1135 }, 1136 #endif 1137 #ifdef CONFIG_PERF_EVENTS 1138 /* 1139 * User-space scripts rely on the existence of this file 1140 * as a feature check for perf_events being enabled. 1141 * 1142 * So it's an ABI, do not remove! 1143 */ 1144 { 1145 .procname = "perf_event_paranoid", 1146 .data = &sysctl_perf_event_paranoid, 1147 .maxlen = sizeof(sysctl_perf_event_paranoid), 1148 .mode = 0644, 1149 .proc_handler = proc_dointvec, 1150 }, 1151 { 1152 .procname = "perf_event_mlock_kb", 1153 .data = &sysctl_perf_event_mlock, 1154 .maxlen = sizeof(sysctl_perf_event_mlock), 1155 .mode = 0644, 1156 .proc_handler = proc_dointvec, 1157 }, 1158 { 1159 .procname = "perf_event_max_sample_rate", 1160 .data = &sysctl_perf_event_sample_rate, 1161 .maxlen = sizeof(sysctl_perf_event_sample_rate), 1162 .mode = 0644, 1163 .proc_handler = perf_proc_update_handler, 1164 .extra1 = &one, 1165 }, 1166 { 1167 .procname = "perf_cpu_time_max_percent", 1168 .data = &sysctl_perf_cpu_time_max_percent, 1169 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent), 1170 .mode = 0644, 1171 .proc_handler = perf_cpu_time_max_percent_handler, 1172 .extra1 = &zero, 1173 .extra2 = &one_hundred, 1174 }, 1175 { 1176 .procname = "perf_event_max_stack", 1177 .data = &sysctl_perf_event_max_stack, 1178 .maxlen = sizeof(sysctl_perf_event_max_stack), 1179 .mode = 0644, 1180 .proc_handler = perf_event_max_stack_handler, 1181 .extra1 = &zero, 1182 .extra2 = &six_hundred_forty_kb, 1183 }, 1184 { 1185 .procname = "perf_event_max_contexts_per_stack", 1186 .data = &sysctl_perf_event_max_contexts_per_stack, 1187 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack), 1188 .mode = 0644, 1189 .proc_handler = perf_event_max_stack_handler, 1190 .extra1 = &zero, 1191 .extra2 = &one_thousand, 1192 }, 1193 #endif 1194 { 1195 .procname = "panic_on_warn", 1196 .data = &panic_on_warn, 1197 .maxlen = sizeof(int), 1198 .mode = 0644, 1199 .proc_handler = proc_dointvec_minmax, 1200 .extra1 = &zero, 1201 .extra2 = &one, 1202 }, 1203 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) 1204 { 1205 .procname = "timer_migration", 1206 .data = &sysctl_timer_migration, 1207 .maxlen = sizeof(unsigned int), 1208 .mode = 0644, 1209 .proc_handler = timer_migration_handler, 1210 .extra1 = &zero, 1211 .extra2 = &one, 1212 }, 1213 #endif 1214 #ifdef CONFIG_BPF_SYSCALL 1215 { 1216 .procname = "unprivileged_bpf_disabled", 1217 .data = &sysctl_unprivileged_bpf_disabled, 1218 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled), 1219 .mode = 0644, 1220 /* only handle a transition from default "0" to "1" */ 1221 .proc_handler = proc_dointvec_minmax, 1222 .extra1 = &one, 1223 .extra2 = &one, 1224 }, 1225 #endif 1226 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU) 1227 { 1228 .procname = "panic_on_rcu_stall", 1229 .data = &sysctl_panic_on_rcu_stall, 1230 .maxlen = sizeof(sysctl_panic_on_rcu_stall), 1231 .mode = 0644, 1232 .proc_handler = proc_dointvec_minmax, 1233 .extra1 = &zero, 1234 .extra2 = &one, 1235 }, 1236 #endif 1237 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE 1238 { 1239 .procname = "stack_erasing", 1240 .data = NULL, 1241 .maxlen = sizeof(int), 1242 .mode = 0600, 1243 .proc_handler = stack_erasing_sysctl, 1244 .extra1 = &zero, 1245 .extra2 = &one, 1246 }, 1247 #endif 1248 { } 1249 }; 1250 1251 static struct ctl_table vm_table[] = { 1252 { 1253 .procname = "overcommit_memory", 1254 .data = &sysctl_overcommit_memory, 1255 .maxlen = sizeof(sysctl_overcommit_memory), 1256 .mode = 0644, 1257 .proc_handler = proc_dointvec_minmax, 1258 .extra1 = &zero, 1259 .extra2 = &two, 1260 }, 1261 { 1262 .procname = "panic_on_oom", 1263 .data = &sysctl_panic_on_oom, 1264 .maxlen = sizeof(sysctl_panic_on_oom), 1265 .mode = 0644, 1266 .proc_handler = proc_dointvec_minmax, 1267 .extra1 = &zero, 1268 .extra2 = &two, 1269 }, 1270 { 1271 .procname = "oom_kill_allocating_task", 1272 .data = &sysctl_oom_kill_allocating_task, 1273 .maxlen = sizeof(sysctl_oom_kill_allocating_task), 1274 .mode = 0644, 1275 .proc_handler = proc_dointvec, 1276 }, 1277 { 1278 .procname = "oom_dump_tasks", 1279 .data = &sysctl_oom_dump_tasks, 1280 .maxlen = sizeof(sysctl_oom_dump_tasks), 1281 .mode = 0644, 1282 .proc_handler = proc_dointvec, 1283 }, 1284 { 1285 .procname = "overcommit_ratio", 1286 .data = &sysctl_overcommit_ratio, 1287 .maxlen = sizeof(sysctl_overcommit_ratio), 1288 .mode = 0644, 1289 .proc_handler = overcommit_ratio_handler, 1290 }, 1291 { 1292 .procname = "overcommit_kbytes", 1293 .data = &sysctl_overcommit_kbytes, 1294 .maxlen = sizeof(sysctl_overcommit_kbytes), 1295 .mode = 0644, 1296 .proc_handler = overcommit_kbytes_handler, 1297 }, 1298 { 1299 .procname = "page-cluster", 1300 .data = &page_cluster, 1301 .maxlen = sizeof(int), 1302 .mode = 0644, 1303 .proc_handler = proc_dointvec_minmax, 1304 .extra1 = &zero, 1305 }, 1306 { 1307 .procname = "dirty_background_ratio", 1308 .data = &dirty_background_ratio, 1309 .maxlen = sizeof(dirty_background_ratio), 1310 .mode = 0644, 1311 .proc_handler = dirty_background_ratio_handler, 1312 .extra1 = &zero, 1313 .extra2 = &one_hundred, 1314 }, 1315 { 1316 .procname = "dirty_background_bytes", 1317 .data = &dirty_background_bytes, 1318 .maxlen = sizeof(dirty_background_bytes), 1319 .mode = 0644, 1320 .proc_handler = dirty_background_bytes_handler, 1321 .extra1 = &one_ul, 1322 }, 1323 { 1324 .procname = "dirty_ratio", 1325 .data = &vm_dirty_ratio, 1326 .maxlen = sizeof(vm_dirty_ratio), 1327 .mode = 0644, 1328 .proc_handler = dirty_ratio_handler, 1329 .extra1 = &zero, 1330 .extra2 = &one_hundred, 1331 }, 1332 { 1333 .procname = "dirty_bytes", 1334 .data = &vm_dirty_bytes, 1335 .maxlen = sizeof(vm_dirty_bytes), 1336 .mode = 0644, 1337 .proc_handler = dirty_bytes_handler, 1338 .extra1 = &dirty_bytes_min, 1339 }, 1340 { 1341 .procname = "dirty_writeback_centisecs", 1342 .data = &dirty_writeback_interval, 1343 .maxlen = sizeof(dirty_writeback_interval), 1344 .mode = 0644, 1345 .proc_handler = dirty_writeback_centisecs_handler, 1346 }, 1347 { 1348 .procname = "dirty_expire_centisecs", 1349 .data = &dirty_expire_interval, 1350 .maxlen = sizeof(dirty_expire_interval), 1351 .mode = 0644, 1352 .proc_handler = proc_dointvec_minmax, 1353 .extra1 = &zero, 1354 }, 1355 { 1356 .procname = "dirtytime_expire_seconds", 1357 .data = &dirtytime_expire_interval, 1358 .maxlen = sizeof(dirtytime_expire_interval), 1359 .mode = 0644, 1360 .proc_handler = dirtytime_interval_handler, 1361 .extra1 = &zero, 1362 }, 1363 { 1364 .procname = "swappiness", 1365 .data = &vm_swappiness, 1366 .maxlen = sizeof(vm_swappiness), 1367 .mode = 0644, 1368 .proc_handler = proc_dointvec_minmax, 1369 .extra1 = &zero, 1370 .extra2 = &one_hundred, 1371 }, 1372 #ifdef CONFIG_HUGETLB_PAGE 1373 { 1374 .procname = "nr_hugepages", 1375 .data = NULL, 1376 .maxlen = sizeof(unsigned long), 1377 .mode = 0644, 1378 .proc_handler = hugetlb_sysctl_handler, 1379 }, 1380 #ifdef CONFIG_NUMA 1381 { 1382 .procname = "nr_hugepages_mempolicy", 1383 .data = NULL, 1384 .maxlen = sizeof(unsigned long), 1385 .mode = 0644, 1386 .proc_handler = &hugetlb_mempolicy_sysctl_handler, 1387 }, 1388 { 1389 .procname = "numa_stat", 1390 .data = &sysctl_vm_numa_stat, 1391 .maxlen = sizeof(int), 1392 .mode = 0644, 1393 .proc_handler = sysctl_vm_numa_stat_handler, 1394 .extra1 = &zero, 1395 .extra2 = &one, 1396 }, 1397 #endif 1398 { 1399 .procname = "hugetlb_shm_group", 1400 .data = &sysctl_hugetlb_shm_group, 1401 .maxlen = sizeof(gid_t), 1402 .mode = 0644, 1403 .proc_handler = proc_dointvec, 1404 }, 1405 { 1406 .procname = "nr_overcommit_hugepages", 1407 .data = NULL, 1408 .maxlen = sizeof(unsigned long), 1409 .mode = 0644, 1410 .proc_handler = hugetlb_overcommit_handler, 1411 }, 1412 #endif 1413 { 1414 .procname = "lowmem_reserve_ratio", 1415 .data = &sysctl_lowmem_reserve_ratio, 1416 .maxlen = sizeof(sysctl_lowmem_reserve_ratio), 1417 .mode = 0644, 1418 .proc_handler = lowmem_reserve_ratio_sysctl_handler, 1419 }, 1420 { 1421 .procname = "drop_caches", 1422 .data = &sysctl_drop_caches, 1423 .maxlen = sizeof(int), 1424 .mode = 0644, 1425 .proc_handler = drop_caches_sysctl_handler, 1426 .extra1 = &one, 1427 .extra2 = &four, 1428 }, 1429 #ifdef CONFIG_COMPACTION 1430 { 1431 .procname = "compact_memory", 1432 .data = &sysctl_compact_memory, 1433 .maxlen = sizeof(int), 1434 .mode = 0200, 1435 .proc_handler = sysctl_compaction_handler, 1436 }, 1437 { 1438 .procname = "extfrag_threshold", 1439 .data = &sysctl_extfrag_threshold, 1440 .maxlen = sizeof(int), 1441 .mode = 0644, 1442 .proc_handler = sysctl_extfrag_handler, 1443 .extra1 = &min_extfrag_threshold, 1444 .extra2 = &max_extfrag_threshold, 1445 }, 1446 { 1447 .procname = "compact_unevictable_allowed", 1448 .data = &sysctl_compact_unevictable_allowed, 1449 .maxlen = sizeof(int), 1450 .mode = 0644, 1451 .proc_handler = proc_dointvec, 1452 .extra1 = &zero, 1453 .extra2 = &one, 1454 }, 1455 1456 #endif /* CONFIG_COMPACTION */ 1457 { 1458 .procname = "min_free_kbytes", 1459 .data = &min_free_kbytes, 1460 .maxlen = sizeof(min_free_kbytes), 1461 .mode = 0644, 1462 .proc_handler = min_free_kbytes_sysctl_handler, 1463 .extra1 = &zero, 1464 }, 1465 { 1466 .procname = "watermark_scale_factor", 1467 .data = &watermark_scale_factor, 1468 .maxlen = sizeof(watermark_scale_factor), 1469 .mode = 0644, 1470 .proc_handler = watermark_scale_factor_sysctl_handler, 1471 .extra1 = &one, 1472 .extra2 = &one_thousand, 1473 }, 1474 { 1475 .procname = "percpu_pagelist_fraction", 1476 .data = &percpu_pagelist_fraction, 1477 .maxlen = sizeof(percpu_pagelist_fraction), 1478 .mode = 0644, 1479 .proc_handler = percpu_pagelist_fraction_sysctl_handler, 1480 .extra1 = &zero, 1481 }, 1482 #ifdef CONFIG_MMU 1483 { 1484 .procname = "max_map_count", 1485 .data = &sysctl_max_map_count, 1486 .maxlen = sizeof(sysctl_max_map_count), 1487 .mode = 0644, 1488 .proc_handler = proc_dointvec_minmax, 1489 .extra1 = &zero, 1490 }, 1491 #else 1492 { 1493 .procname = "nr_trim_pages", 1494 .data = &sysctl_nr_trim_pages, 1495 .maxlen = sizeof(sysctl_nr_trim_pages), 1496 .mode = 0644, 1497 .proc_handler = proc_dointvec_minmax, 1498 .extra1 = &zero, 1499 }, 1500 #endif 1501 { 1502 .procname = "laptop_mode", 1503 .data = &laptop_mode, 1504 .maxlen = sizeof(laptop_mode), 1505 .mode = 0644, 1506 .proc_handler = proc_dointvec_jiffies, 1507 }, 1508 { 1509 .procname = "block_dump", 1510 .data = &block_dump, 1511 .maxlen = sizeof(block_dump), 1512 .mode = 0644, 1513 .proc_handler = proc_dointvec, 1514 .extra1 = &zero, 1515 }, 1516 { 1517 .procname = "vfs_cache_pressure", 1518 .data = &sysctl_vfs_cache_pressure, 1519 .maxlen = sizeof(sysctl_vfs_cache_pressure), 1520 .mode = 0644, 1521 .proc_handler = proc_dointvec, 1522 .extra1 = &zero, 1523 }, 1524 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT 1525 { 1526 .procname = "legacy_va_layout", 1527 .data = &sysctl_legacy_va_layout, 1528 .maxlen = sizeof(sysctl_legacy_va_layout), 1529 .mode = 0644, 1530 .proc_handler = proc_dointvec, 1531 .extra1 = &zero, 1532 }, 1533 #endif 1534 #ifdef CONFIG_NUMA 1535 { 1536 .procname = "zone_reclaim_mode", 1537 .data = &node_reclaim_mode, 1538 .maxlen = sizeof(node_reclaim_mode), 1539 .mode = 0644, 1540 .proc_handler = proc_dointvec, 1541 .extra1 = &zero, 1542 }, 1543 { 1544 .procname = "min_unmapped_ratio", 1545 .data = &sysctl_min_unmapped_ratio, 1546 .maxlen = sizeof(sysctl_min_unmapped_ratio), 1547 .mode = 0644, 1548 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler, 1549 .extra1 = &zero, 1550 .extra2 = &one_hundred, 1551 }, 1552 { 1553 .procname = "min_slab_ratio", 1554 .data = &sysctl_min_slab_ratio, 1555 .maxlen = sizeof(sysctl_min_slab_ratio), 1556 .mode = 0644, 1557 .proc_handler = sysctl_min_slab_ratio_sysctl_handler, 1558 .extra1 = &zero, 1559 .extra2 = &one_hundred, 1560 }, 1561 #endif 1562 #ifdef CONFIG_SMP 1563 { 1564 .procname = "stat_interval", 1565 .data = &sysctl_stat_interval, 1566 .maxlen = sizeof(sysctl_stat_interval), 1567 .mode = 0644, 1568 .proc_handler = proc_dointvec_jiffies, 1569 }, 1570 { 1571 .procname = "stat_refresh", 1572 .data = NULL, 1573 .maxlen = 0, 1574 .mode = 0600, 1575 .proc_handler = vmstat_refresh, 1576 }, 1577 #endif 1578 #ifdef CONFIG_MMU 1579 { 1580 .procname = "mmap_min_addr", 1581 .data = &dac_mmap_min_addr, 1582 .maxlen = sizeof(unsigned long), 1583 .mode = 0644, 1584 .proc_handler = mmap_min_addr_handler, 1585 }, 1586 #endif 1587 #ifdef CONFIG_NUMA 1588 { 1589 .procname = "numa_zonelist_order", 1590 .data = &numa_zonelist_order, 1591 .maxlen = NUMA_ZONELIST_ORDER_LEN, 1592 .mode = 0644, 1593 .proc_handler = numa_zonelist_order_handler, 1594 }, 1595 #endif 1596 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \ 1597 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL)) 1598 { 1599 .procname = "vdso_enabled", 1600 #ifdef CONFIG_X86_32 1601 .data = &vdso32_enabled, 1602 .maxlen = sizeof(vdso32_enabled), 1603 #else 1604 .data = &vdso_enabled, 1605 .maxlen = sizeof(vdso_enabled), 1606 #endif 1607 .mode = 0644, 1608 .proc_handler = proc_dointvec, 1609 .extra1 = &zero, 1610 }, 1611 #endif 1612 #ifdef CONFIG_HIGHMEM 1613 { 1614 .procname = "highmem_is_dirtyable", 1615 .data = &vm_highmem_is_dirtyable, 1616 .maxlen = sizeof(vm_highmem_is_dirtyable), 1617 .mode = 0644, 1618 .proc_handler = proc_dointvec_minmax, 1619 .extra1 = &zero, 1620 .extra2 = &one, 1621 }, 1622 #endif 1623 #ifdef CONFIG_MEMORY_FAILURE 1624 { 1625 .procname = "memory_failure_early_kill", 1626 .data = &sysctl_memory_failure_early_kill, 1627 .maxlen = sizeof(sysctl_memory_failure_early_kill), 1628 .mode = 0644, 1629 .proc_handler = proc_dointvec_minmax, 1630 .extra1 = &zero, 1631 .extra2 = &one, 1632 }, 1633 { 1634 .procname = "memory_failure_recovery", 1635 .data = &sysctl_memory_failure_recovery, 1636 .maxlen = sizeof(sysctl_memory_failure_recovery), 1637 .mode = 0644, 1638 .proc_handler = proc_dointvec_minmax, 1639 .extra1 = &zero, 1640 .extra2 = &one, 1641 }, 1642 #endif 1643 { 1644 .procname = "user_reserve_kbytes", 1645 .data = &sysctl_user_reserve_kbytes, 1646 .maxlen = sizeof(sysctl_user_reserve_kbytes), 1647 .mode = 0644, 1648 .proc_handler = proc_doulongvec_minmax, 1649 }, 1650 { 1651 .procname = "admin_reserve_kbytes", 1652 .data = &sysctl_admin_reserve_kbytes, 1653 .maxlen = sizeof(sysctl_admin_reserve_kbytes), 1654 .mode = 0644, 1655 .proc_handler = proc_doulongvec_minmax, 1656 }, 1657 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS 1658 { 1659 .procname = "mmap_rnd_bits", 1660 .data = &mmap_rnd_bits, 1661 .maxlen = sizeof(mmap_rnd_bits), 1662 .mode = 0600, 1663 .proc_handler = proc_dointvec_minmax, 1664 .extra1 = (void *)&mmap_rnd_bits_min, 1665 .extra2 = (void *)&mmap_rnd_bits_max, 1666 }, 1667 #endif 1668 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS 1669 { 1670 .procname = "mmap_rnd_compat_bits", 1671 .data = &mmap_rnd_compat_bits, 1672 .maxlen = sizeof(mmap_rnd_compat_bits), 1673 .mode = 0600, 1674 .proc_handler = proc_dointvec_minmax, 1675 .extra1 = (void *)&mmap_rnd_compat_bits_min, 1676 .extra2 = (void *)&mmap_rnd_compat_bits_max, 1677 }, 1678 #endif 1679 { } 1680 }; 1681 1682 static struct ctl_table fs_table[] = { 1683 { 1684 .procname = "inode-nr", 1685 .data = &inodes_stat, 1686 .maxlen = 2*sizeof(long), 1687 .mode = 0444, 1688 .proc_handler = proc_nr_inodes, 1689 }, 1690 { 1691 .procname = "inode-state", 1692 .data = &inodes_stat, 1693 .maxlen = 7*sizeof(long), 1694 .mode = 0444, 1695 .proc_handler = proc_nr_inodes, 1696 }, 1697 { 1698 .procname = "file-nr", 1699 .data = &files_stat, 1700 .maxlen = sizeof(files_stat), 1701 .mode = 0444, 1702 .proc_handler = proc_nr_files, 1703 }, 1704 { 1705 .procname = "file-max", 1706 .data = &files_stat.max_files, 1707 .maxlen = sizeof(files_stat.max_files), 1708 .mode = 0644, 1709 .proc_handler = proc_doulongvec_minmax, 1710 }, 1711 { 1712 .procname = "nr_open", 1713 .data = &sysctl_nr_open, 1714 .maxlen = sizeof(unsigned int), 1715 .mode = 0644, 1716 .proc_handler = proc_dointvec_minmax, 1717 .extra1 = &sysctl_nr_open_min, 1718 .extra2 = &sysctl_nr_open_max, 1719 }, 1720 { 1721 .procname = "dentry-state", 1722 .data = &dentry_stat, 1723 .maxlen = 6*sizeof(long), 1724 .mode = 0444, 1725 .proc_handler = proc_nr_dentry, 1726 }, 1727 { 1728 .procname = "overflowuid", 1729 .data = &fs_overflowuid, 1730 .maxlen = sizeof(int), 1731 .mode = 0644, 1732 .proc_handler = proc_dointvec_minmax, 1733 .extra1 = &minolduid, 1734 .extra2 = &maxolduid, 1735 }, 1736 { 1737 .procname = "overflowgid", 1738 .data = &fs_overflowgid, 1739 .maxlen = sizeof(int), 1740 .mode = 0644, 1741 .proc_handler = proc_dointvec_minmax, 1742 .extra1 = &minolduid, 1743 .extra2 = &maxolduid, 1744 }, 1745 #ifdef CONFIG_FILE_LOCKING 1746 { 1747 .procname = "leases-enable", 1748 .data = &leases_enable, 1749 .maxlen = sizeof(int), 1750 .mode = 0644, 1751 .proc_handler = proc_dointvec, 1752 }, 1753 #endif 1754 #ifdef CONFIG_DNOTIFY 1755 { 1756 .procname = "dir-notify-enable", 1757 .data = &dir_notify_enable, 1758 .maxlen = sizeof(int), 1759 .mode = 0644, 1760 .proc_handler = proc_dointvec, 1761 }, 1762 #endif 1763 #ifdef CONFIG_MMU 1764 #ifdef CONFIG_FILE_LOCKING 1765 { 1766 .procname = "lease-break-time", 1767 .data = &lease_break_time, 1768 .maxlen = sizeof(int), 1769 .mode = 0644, 1770 .proc_handler = proc_dointvec, 1771 }, 1772 #endif 1773 #ifdef CONFIG_AIO 1774 { 1775 .procname = "aio-nr", 1776 .data = &aio_nr, 1777 .maxlen = sizeof(aio_nr), 1778 .mode = 0444, 1779 .proc_handler = proc_doulongvec_minmax, 1780 }, 1781 { 1782 .procname = "aio-max-nr", 1783 .data = &aio_max_nr, 1784 .maxlen = sizeof(aio_max_nr), 1785 .mode = 0644, 1786 .proc_handler = proc_doulongvec_minmax, 1787 }, 1788 #endif /* CONFIG_AIO */ 1789 #ifdef CONFIG_INOTIFY_USER 1790 { 1791 .procname = "inotify", 1792 .mode = 0555, 1793 .child = inotify_table, 1794 }, 1795 #endif 1796 #ifdef CONFIG_EPOLL 1797 { 1798 .procname = "epoll", 1799 .mode = 0555, 1800 .child = epoll_table, 1801 }, 1802 #endif 1803 #endif 1804 { 1805 .procname = "protected_symlinks", 1806 .data = &sysctl_protected_symlinks, 1807 .maxlen = sizeof(int), 1808 .mode = 0600, 1809 .proc_handler = proc_dointvec_minmax, 1810 .extra1 = &zero, 1811 .extra2 = &one, 1812 }, 1813 { 1814 .procname = "protected_hardlinks", 1815 .data = &sysctl_protected_hardlinks, 1816 .maxlen = sizeof(int), 1817 .mode = 0600, 1818 .proc_handler = proc_dointvec_minmax, 1819 .extra1 = &zero, 1820 .extra2 = &one, 1821 }, 1822 { 1823 .procname = "protected_fifos", 1824 .data = &sysctl_protected_fifos, 1825 .maxlen = sizeof(int), 1826 .mode = 0600, 1827 .proc_handler = proc_dointvec_minmax, 1828 .extra1 = &zero, 1829 .extra2 = &two, 1830 }, 1831 { 1832 .procname = "protected_regular", 1833 .data = &sysctl_protected_regular, 1834 .maxlen = sizeof(int), 1835 .mode = 0600, 1836 .proc_handler = proc_dointvec_minmax, 1837 .extra1 = &zero, 1838 .extra2 = &two, 1839 }, 1840 { 1841 .procname = "suid_dumpable", 1842 .data = &suid_dumpable, 1843 .maxlen = sizeof(int), 1844 .mode = 0644, 1845 .proc_handler = proc_dointvec_minmax_coredump, 1846 .extra1 = &zero, 1847 .extra2 = &two, 1848 }, 1849 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE) 1850 { 1851 .procname = "binfmt_misc", 1852 .mode = 0555, 1853 .child = sysctl_mount_point, 1854 }, 1855 #endif 1856 { 1857 .procname = "pipe-max-size", 1858 .data = &pipe_max_size, 1859 .maxlen = sizeof(pipe_max_size), 1860 .mode = 0644, 1861 .proc_handler = proc_dopipe_max_size, 1862 }, 1863 { 1864 .procname = "pipe-user-pages-hard", 1865 .data = &pipe_user_pages_hard, 1866 .maxlen = sizeof(pipe_user_pages_hard), 1867 .mode = 0644, 1868 .proc_handler = proc_doulongvec_minmax, 1869 }, 1870 { 1871 .procname = "pipe-user-pages-soft", 1872 .data = &pipe_user_pages_soft, 1873 .maxlen = sizeof(pipe_user_pages_soft), 1874 .mode = 0644, 1875 .proc_handler = proc_doulongvec_minmax, 1876 }, 1877 { 1878 .procname = "mount-max", 1879 .data = &sysctl_mount_max, 1880 .maxlen = sizeof(unsigned int), 1881 .mode = 0644, 1882 .proc_handler = proc_dointvec_minmax, 1883 .extra1 = &one, 1884 }, 1885 { } 1886 }; 1887 1888 static struct ctl_table debug_table[] = { 1889 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE 1890 { 1891 .procname = "exception-trace", 1892 .data = &show_unhandled_signals, 1893 .maxlen = sizeof(int), 1894 .mode = 0644, 1895 .proc_handler = proc_dointvec 1896 }, 1897 #endif 1898 #if defined(CONFIG_OPTPROBES) 1899 { 1900 .procname = "kprobes-optimization", 1901 .data = &sysctl_kprobes_optimization, 1902 .maxlen = sizeof(int), 1903 .mode = 0644, 1904 .proc_handler = proc_kprobes_optimization_handler, 1905 .extra1 = &zero, 1906 .extra2 = &one, 1907 }, 1908 #endif 1909 { } 1910 }; 1911 1912 static struct ctl_table dev_table[] = { 1913 { } 1914 }; 1915 1916 int __init sysctl_init(void) 1917 { 1918 struct ctl_table_header *hdr; 1919 1920 hdr = register_sysctl_table(sysctl_base_table); 1921 kmemleak_not_leak(hdr); 1922 return 0; 1923 } 1924 1925 #endif /* CONFIG_SYSCTL */ 1926 1927 /* 1928 * /proc/sys support 1929 */ 1930 1931 #ifdef CONFIG_PROC_SYSCTL 1932 1933 static int _proc_do_string(char *data, int maxlen, int write, 1934 char __user *buffer, 1935 size_t *lenp, loff_t *ppos) 1936 { 1937 size_t len; 1938 char __user *p; 1939 char c; 1940 1941 if (!data || !maxlen || !*lenp) { 1942 *lenp = 0; 1943 return 0; 1944 } 1945 1946 if (write) { 1947 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) { 1948 /* Only continue writes not past the end of buffer. */ 1949 len = strlen(data); 1950 if (len > maxlen - 1) 1951 len = maxlen - 1; 1952 1953 if (*ppos > len) 1954 return 0; 1955 len = *ppos; 1956 } else { 1957 /* Start writing from beginning of buffer. */ 1958 len = 0; 1959 } 1960 1961 *ppos += *lenp; 1962 p = buffer; 1963 while ((p - buffer) < *lenp && len < maxlen - 1) { 1964 if (get_user(c, p++)) 1965 return -EFAULT; 1966 if (c == 0 || c == '\n') 1967 break; 1968 data[len++] = c; 1969 } 1970 data[len] = 0; 1971 } else { 1972 len = strlen(data); 1973 if (len > maxlen) 1974 len = maxlen; 1975 1976 if (*ppos > len) { 1977 *lenp = 0; 1978 return 0; 1979 } 1980 1981 data += *ppos; 1982 len -= *ppos; 1983 1984 if (len > *lenp) 1985 len = *lenp; 1986 if (len) 1987 if (copy_to_user(buffer, data, len)) 1988 return -EFAULT; 1989 if (len < *lenp) { 1990 if (put_user('\n', buffer + len)) 1991 return -EFAULT; 1992 len++; 1993 } 1994 *lenp = len; 1995 *ppos += len; 1996 } 1997 return 0; 1998 } 1999 2000 static void warn_sysctl_write(struct ctl_table *table) 2001 { 2002 pr_warn_once("%s wrote to %s when file position was not 0!\n" 2003 "This will not be supported in the future. To silence this\n" 2004 "warning, set kernel.sysctl_writes_strict = -1\n", 2005 current->comm, table->procname); 2006 } 2007 2008 /** 2009 * proc_first_pos_non_zero_ignore - check if first position is allowed 2010 * @ppos: file position 2011 * @table: the sysctl table 2012 * 2013 * Returns true if the first position is non-zero and the sysctl_writes_strict 2014 * mode indicates this is not allowed for numeric input types. String proc 2015 * handlers can ignore the return value. 2016 */ 2017 static bool proc_first_pos_non_zero_ignore(loff_t *ppos, 2018 struct ctl_table *table) 2019 { 2020 if (!*ppos) 2021 return false; 2022 2023 switch (sysctl_writes_strict) { 2024 case SYSCTL_WRITES_STRICT: 2025 return true; 2026 case SYSCTL_WRITES_WARN: 2027 warn_sysctl_write(table); 2028 return false; 2029 default: 2030 return false; 2031 } 2032 } 2033 2034 /** 2035 * proc_dostring - read a string sysctl 2036 * @table: the sysctl table 2037 * @write: %TRUE if this is a write to the sysctl file 2038 * @buffer: the user buffer 2039 * @lenp: the size of the user buffer 2040 * @ppos: file position 2041 * 2042 * Reads/writes a string from/to the user buffer. If the kernel 2043 * buffer provided is not large enough to hold the string, the 2044 * string is truncated. The copied string is %NULL-terminated. 2045 * If the string is being read by the user process, it is copied 2046 * and a newline '\n' is added. It is truncated if the buffer is 2047 * not large enough. 2048 * 2049 * Returns 0 on success. 2050 */ 2051 int proc_dostring(struct ctl_table *table, int write, 2052 void __user *buffer, size_t *lenp, loff_t *ppos) 2053 { 2054 if (write) 2055 proc_first_pos_non_zero_ignore(ppos, table); 2056 2057 return _proc_do_string((char *)(table->data), table->maxlen, write, 2058 (char __user *)buffer, lenp, ppos); 2059 } 2060 2061 static size_t proc_skip_spaces(char **buf) 2062 { 2063 size_t ret; 2064 char *tmp = skip_spaces(*buf); 2065 ret = tmp - *buf; 2066 *buf = tmp; 2067 return ret; 2068 } 2069 2070 static void proc_skip_char(char **buf, size_t *size, const char v) 2071 { 2072 while (*size) { 2073 if (**buf != v) 2074 break; 2075 (*size)--; 2076 (*buf)++; 2077 } 2078 } 2079 2080 #define TMPBUFLEN 22 2081 /** 2082 * proc_get_long - reads an ASCII formatted integer from a user buffer 2083 * 2084 * @buf: a kernel buffer 2085 * @size: size of the kernel buffer 2086 * @val: this is where the number will be stored 2087 * @neg: set to %TRUE if number is negative 2088 * @perm_tr: a vector which contains the allowed trailers 2089 * @perm_tr_len: size of the perm_tr vector 2090 * @tr: pointer to store the trailer character 2091 * 2092 * In case of success %0 is returned and @buf and @size are updated with 2093 * the amount of bytes read. If @tr is non-NULL and a trailing 2094 * character exists (size is non-zero after returning from this 2095 * function), @tr is updated with the trailing character. 2096 */ 2097 static int proc_get_long(char **buf, size_t *size, 2098 unsigned long *val, bool *neg, 2099 const char *perm_tr, unsigned perm_tr_len, char *tr) 2100 { 2101 int len; 2102 char *p, tmp[TMPBUFLEN]; 2103 2104 if (!*size) 2105 return -EINVAL; 2106 2107 len = *size; 2108 if (len > TMPBUFLEN - 1) 2109 len = TMPBUFLEN - 1; 2110 2111 memcpy(tmp, *buf, len); 2112 2113 tmp[len] = 0; 2114 p = tmp; 2115 if (*p == '-' && *size > 1) { 2116 *neg = true; 2117 p++; 2118 } else 2119 *neg = false; 2120 if (!isdigit(*p)) 2121 return -EINVAL; 2122 2123 *val = simple_strtoul(p, &p, 0); 2124 2125 len = p - tmp; 2126 2127 /* We don't know if the next char is whitespace thus we may accept 2128 * invalid integers (e.g. 1234...a) or two integers instead of one 2129 * (e.g. 123...1). So lets not allow such large numbers. */ 2130 if (len == TMPBUFLEN - 1) 2131 return -EINVAL; 2132 2133 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len)) 2134 return -EINVAL; 2135 2136 if (tr && (len < *size)) 2137 *tr = *p; 2138 2139 *buf += len; 2140 *size -= len; 2141 2142 return 0; 2143 } 2144 2145 /** 2146 * proc_put_long - converts an integer to a decimal ASCII formatted string 2147 * 2148 * @buf: the user buffer 2149 * @size: the size of the user buffer 2150 * @val: the integer to be converted 2151 * @neg: sign of the number, %TRUE for negative 2152 * 2153 * In case of success %0 is returned and @buf and @size are updated with 2154 * the amount of bytes written. 2155 */ 2156 static int proc_put_long(void __user **buf, size_t *size, unsigned long val, 2157 bool neg) 2158 { 2159 int len; 2160 char tmp[TMPBUFLEN], *p = tmp; 2161 2162 sprintf(p, "%s%lu", neg ? "-" : "", val); 2163 len = strlen(tmp); 2164 if (len > *size) 2165 len = *size; 2166 if (copy_to_user(*buf, tmp, len)) 2167 return -EFAULT; 2168 *size -= len; 2169 *buf += len; 2170 return 0; 2171 } 2172 #undef TMPBUFLEN 2173 2174 static int proc_put_char(void __user **buf, size_t *size, char c) 2175 { 2176 if (*size) { 2177 char __user **buffer = (char __user **)buf; 2178 if (put_user(c, *buffer)) 2179 return -EFAULT; 2180 (*size)--, (*buffer)++; 2181 *buf = *buffer; 2182 } 2183 return 0; 2184 } 2185 2186 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp, 2187 int *valp, 2188 int write, void *data) 2189 { 2190 if (write) { 2191 if (*negp) { 2192 if (*lvalp > (unsigned long) INT_MAX + 1) 2193 return -EINVAL; 2194 *valp = -*lvalp; 2195 } else { 2196 if (*lvalp > (unsigned long) INT_MAX) 2197 return -EINVAL; 2198 *valp = *lvalp; 2199 } 2200 } else { 2201 int val = *valp; 2202 if (val < 0) { 2203 *negp = true; 2204 *lvalp = -(unsigned long)val; 2205 } else { 2206 *negp = false; 2207 *lvalp = (unsigned long)val; 2208 } 2209 } 2210 return 0; 2211 } 2212 2213 static int do_proc_douintvec_conv(unsigned long *lvalp, 2214 unsigned int *valp, 2215 int write, void *data) 2216 { 2217 if (write) { 2218 if (*lvalp > UINT_MAX) 2219 return -EINVAL; 2220 *valp = *lvalp; 2221 } else { 2222 unsigned int val = *valp; 2223 *lvalp = (unsigned long)val; 2224 } 2225 return 0; 2226 } 2227 2228 static const char proc_wspace_sep[] = { ' ', '\t', '\n' }; 2229 2230 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table, 2231 int write, void __user *buffer, 2232 size_t *lenp, loff_t *ppos, 2233 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 2234 int write, void *data), 2235 void *data) 2236 { 2237 int *i, vleft, first = 1, err = 0; 2238 size_t left; 2239 char *kbuf = NULL, *p; 2240 2241 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) { 2242 *lenp = 0; 2243 return 0; 2244 } 2245 2246 i = (int *) tbl_data; 2247 vleft = table->maxlen / sizeof(*i); 2248 left = *lenp; 2249 2250 if (!conv) 2251 conv = do_proc_dointvec_conv; 2252 2253 if (write) { 2254 if (proc_first_pos_non_zero_ignore(ppos, table)) 2255 goto out; 2256 2257 if (left > PAGE_SIZE - 1) 2258 left = PAGE_SIZE - 1; 2259 p = kbuf = memdup_user_nul(buffer, left); 2260 if (IS_ERR(kbuf)) 2261 return PTR_ERR(kbuf); 2262 } 2263 2264 for (; left && vleft--; i++, first=0) { 2265 unsigned long lval; 2266 bool neg; 2267 2268 if (write) { 2269 left -= proc_skip_spaces(&p); 2270 2271 if (!left) 2272 break; 2273 err = proc_get_long(&p, &left, &lval, &neg, 2274 proc_wspace_sep, 2275 sizeof(proc_wspace_sep), NULL); 2276 if (err) 2277 break; 2278 if (conv(&neg, &lval, i, 1, data)) { 2279 err = -EINVAL; 2280 break; 2281 } 2282 } else { 2283 if (conv(&neg, &lval, i, 0, data)) { 2284 err = -EINVAL; 2285 break; 2286 } 2287 if (!first) 2288 err = proc_put_char(&buffer, &left, '\t'); 2289 if (err) 2290 break; 2291 err = proc_put_long(&buffer, &left, lval, neg); 2292 if (err) 2293 break; 2294 } 2295 } 2296 2297 if (!write && !first && left && !err) 2298 err = proc_put_char(&buffer, &left, '\n'); 2299 if (write && !err && left) 2300 left -= proc_skip_spaces(&p); 2301 if (write) { 2302 kfree(kbuf); 2303 if (first) 2304 return err ? : -EINVAL; 2305 } 2306 *lenp -= left; 2307 out: 2308 *ppos += *lenp; 2309 return err; 2310 } 2311 2312 static int do_proc_dointvec(struct ctl_table *table, int write, 2313 void __user *buffer, size_t *lenp, loff_t *ppos, 2314 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 2315 int write, void *data), 2316 void *data) 2317 { 2318 return __do_proc_dointvec(table->data, table, write, 2319 buffer, lenp, ppos, conv, data); 2320 } 2321 2322 static int do_proc_douintvec_w(unsigned int *tbl_data, 2323 struct ctl_table *table, 2324 void __user *buffer, 2325 size_t *lenp, loff_t *ppos, 2326 int (*conv)(unsigned long *lvalp, 2327 unsigned int *valp, 2328 int write, void *data), 2329 void *data) 2330 { 2331 unsigned long lval; 2332 int err = 0; 2333 size_t left; 2334 bool neg; 2335 char *kbuf = NULL, *p; 2336 2337 left = *lenp; 2338 2339 if (proc_first_pos_non_zero_ignore(ppos, table)) 2340 goto bail_early; 2341 2342 if (left > PAGE_SIZE - 1) 2343 left = PAGE_SIZE - 1; 2344 2345 p = kbuf = memdup_user_nul(buffer, left); 2346 if (IS_ERR(kbuf)) 2347 return -EINVAL; 2348 2349 left -= proc_skip_spaces(&p); 2350 if (!left) { 2351 err = -EINVAL; 2352 goto out_free; 2353 } 2354 2355 err = proc_get_long(&p, &left, &lval, &neg, 2356 proc_wspace_sep, 2357 sizeof(proc_wspace_sep), NULL); 2358 if (err || neg) { 2359 err = -EINVAL; 2360 goto out_free; 2361 } 2362 2363 if (conv(&lval, tbl_data, 1, data)) { 2364 err = -EINVAL; 2365 goto out_free; 2366 } 2367 2368 if (!err && left) 2369 left -= proc_skip_spaces(&p); 2370 2371 out_free: 2372 kfree(kbuf); 2373 if (err) 2374 return -EINVAL; 2375 2376 return 0; 2377 2378 /* This is in keeping with old __do_proc_dointvec() */ 2379 bail_early: 2380 *ppos += *lenp; 2381 return err; 2382 } 2383 2384 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer, 2385 size_t *lenp, loff_t *ppos, 2386 int (*conv)(unsigned long *lvalp, 2387 unsigned int *valp, 2388 int write, void *data), 2389 void *data) 2390 { 2391 unsigned long lval; 2392 int err = 0; 2393 size_t left; 2394 2395 left = *lenp; 2396 2397 if (conv(&lval, tbl_data, 0, data)) { 2398 err = -EINVAL; 2399 goto out; 2400 } 2401 2402 err = proc_put_long(&buffer, &left, lval, false); 2403 if (err || !left) 2404 goto out; 2405 2406 err = proc_put_char(&buffer, &left, '\n'); 2407 2408 out: 2409 *lenp -= left; 2410 *ppos += *lenp; 2411 2412 return err; 2413 } 2414 2415 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table, 2416 int write, void __user *buffer, 2417 size_t *lenp, loff_t *ppos, 2418 int (*conv)(unsigned long *lvalp, 2419 unsigned int *valp, 2420 int write, void *data), 2421 void *data) 2422 { 2423 unsigned int *i, vleft; 2424 2425 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) { 2426 *lenp = 0; 2427 return 0; 2428 } 2429 2430 i = (unsigned int *) tbl_data; 2431 vleft = table->maxlen / sizeof(*i); 2432 2433 /* 2434 * Arrays are not supported, keep this simple. *Do not* add 2435 * support for them. 2436 */ 2437 if (vleft != 1) { 2438 *lenp = 0; 2439 return -EINVAL; 2440 } 2441 2442 if (!conv) 2443 conv = do_proc_douintvec_conv; 2444 2445 if (write) 2446 return do_proc_douintvec_w(i, table, buffer, lenp, ppos, 2447 conv, data); 2448 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data); 2449 } 2450 2451 static int do_proc_douintvec(struct ctl_table *table, int write, 2452 void __user *buffer, size_t *lenp, loff_t *ppos, 2453 int (*conv)(unsigned long *lvalp, 2454 unsigned int *valp, 2455 int write, void *data), 2456 void *data) 2457 { 2458 return __do_proc_douintvec(table->data, table, write, 2459 buffer, lenp, ppos, conv, data); 2460 } 2461 2462 /** 2463 * proc_dointvec - read a vector of integers 2464 * @table: the sysctl table 2465 * @write: %TRUE if this is a write to the sysctl file 2466 * @buffer: the user buffer 2467 * @lenp: the size of the user buffer 2468 * @ppos: file position 2469 * 2470 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2471 * values from/to the user buffer, treated as an ASCII string. 2472 * 2473 * Returns 0 on success. 2474 */ 2475 int proc_dointvec(struct ctl_table *table, int write, 2476 void __user *buffer, size_t *lenp, loff_t *ppos) 2477 { 2478 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL); 2479 } 2480 2481 /** 2482 * proc_douintvec - read a vector of unsigned integers 2483 * @table: the sysctl table 2484 * @write: %TRUE if this is a write to the sysctl file 2485 * @buffer: the user buffer 2486 * @lenp: the size of the user buffer 2487 * @ppos: file position 2488 * 2489 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer 2490 * values from/to the user buffer, treated as an ASCII string. 2491 * 2492 * Returns 0 on success. 2493 */ 2494 int proc_douintvec(struct ctl_table *table, int write, 2495 void __user *buffer, size_t *lenp, loff_t *ppos) 2496 { 2497 return do_proc_douintvec(table, write, buffer, lenp, ppos, 2498 do_proc_douintvec_conv, NULL); 2499 } 2500 2501 /* 2502 * Taint values can only be increased 2503 * This means we can safely use a temporary. 2504 */ 2505 static int proc_taint(struct ctl_table *table, int write, 2506 void __user *buffer, size_t *lenp, loff_t *ppos) 2507 { 2508 struct ctl_table t; 2509 unsigned long tmptaint = get_taint(); 2510 int err; 2511 2512 if (write && !capable(CAP_SYS_ADMIN)) 2513 return -EPERM; 2514 2515 t = *table; 2516 t.data = &tmptaint; 2517 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos); 2518 if (err < 0) 2519 return err; 2520 2521 if (write) { 2522 /* 2523 * Poor man's atomic or. Not worth adding a primitive 2524 * to everyone's atomic.h for this 2525 */ 2526 int i; 2527 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) { 2528 if ((tmptaint >> i) & 1) 2529 add_taint(i, LOCKDEP_STILL_OK); 2530 } 2531 } 2532 2533 return err; 2534 } 2535 2536 #ifdef CONFIG_PRINTK 2537 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write, 2538 void __user *buffer, size_t *lenp, loff_t *ppos) 2539 { 2540 if (write && !capable(CAP_SYS_ADMIN)) 2541 return -EPERM; 2542 2543 return proc_dointvec_minmax(table, write, buffer, lenp, ppos); 2544 } 2545 #endif 2546 2547 /** 2548 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure 2549 * @min: pointer to minimum allowable value 2550 * @max: pointer to maximum allowable value 2551 * 2552 * The do_proc_dointvec_minmax_conv_param structure provides the 2553 * minimum and maximum values for doing range checking for those sysctl 2554 * parameters that use the proc_dointvec_minmax() handler. 2555 */ 2556 struct do_proc_dointvec_minmax_conv_param { 2557 int *min; 2558 int *max; 2559 }; 2560 2561 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp, 2562 int *valp, 2563 int write, void *data) 2564 { 2565 struct do_proc_dointvec_minmax_conv_param *param = data; 2566 if (write) { 2567 int val = *negp ? -*lvalp : *lvalp; 2568 if ((param->min && *param->min > val) || 2569 (param->max && *param->max < val)) 2570 return -EINVAL; 2571 *valp = val; 2572 } else { 2573 int val = *valp; 2574 if (val < 0) { 2575 *negp = true; 2576 *lvalp = -(unsigned long)val; 2577 } else { 2578 *negp = false; 2579 *lvalp = (unsigned long)val; 2580 } 2581 } 2582 return 0; 2583 } 2584 2585 /** 2586 * proc_dointvec_minmax - read a vector of integers with min/max values 2587 * @table: the sysctl table 2588 * @write: %TRUE if this is a write to the sysctl file 2589 * @buffer: the user buffer 2590 * @lenp: the size of the user buffer 2591 * @ppos: file position 2592 * 2593 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2594 * values from/to the user buffer, treated as an ASCII string. 2595 * 2596 * This routine will ensure the values are within the range specified by 2597 * table->extra1 (min) and table->extra2 (max). 2598 * 2599 * Returns 0 on success or -EINVAL on write when the range check fails. 2600 */ 2601 int proc_dointvec_minmax(struct ctl_table *table, int write, 2602 void __user *buffer, size_t *lenp, loff_t *ppos) 2603 { 2604 struct do_proc_dointvec_minmax_conv_param param = { 2605 .min = (int *) table->extra1, 2606 .max = (int *) table->extra2, 2607 }; 2608 return do_proc_dointvec(table, write, buffer, lenp, ppos, 2609 do_proc_dointvec_minmax_conv, ¶m); 2610 } 2611 2612 /** 2613 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure 2614 * @min: pointer to minimum allowable value 2615 * @max: pointer to maximum allowable value 2616 * 2617 * The do_proc_douintvec_minmax_conv_param structure provides the 2618 * minimum and maximum values for doing range checking for those sysctl 2619 * parameters that use the proc_douintvec_minmax() handler. 2620 */ 2621 struct do_proc_douintvec_minmax_conv_param { 2622 unsigned int *min; 2623 unsigned int *max; 2624 }; 2625 2626 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp, 2627 unsigned int *valp, 2628 int write, void *data) 2629 { 2630 struct do_proc_douintvec_minmax_conv_param *param = data; 2631 2632 if (write) { 2633 unsigned int val = *lvalp; 2634 2635 if (*lvalp > UINT_MAX) 2636 return -EINVAL; 2637 2638 if ((param->min && *param->min > val) || 2639 (param->max && *param->max < val)) 2640 return -ERANGE; 2641 2642 *valp = val; 2643 } else { 2644 unsigned int val = *valp; 2645 *lvalp = (unsigned long) val; 2646 } 2647 2648 return 0; 2649 } 2650 2651 /** 2652 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values 2653 * @table: the sysctl table 2654 * @write: %TRUE if this is a write to the sysctl file 2655 * @buffer: the user buffer 2656 * @lenp: the size of the user buffer 2657 * @ppos: file position 2658 * 2659 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer 2660 * values from/to the user buffer, treated as an ASCII string. Negative 2661 * strings are not allowed. 2662 * 2663 * This routine will ensure the values are within the range specified by 2664 * table->extra1 (min) and table->extra2 (max). There is a final sanity 2665 * check for UINT_MAX to avoid having to support wrap around uses from 2666 * userspace. 2667 * 2668 * Returns 0 on success or -ERANGE on write when the range check fails. 2669 */ 2670 int proc_douintvec_minmax(struct ctl_table *table, int write, 2671 void __user *buffer, size_t *lenp, loff_t *ppos) 2672 { 2673 struct do_proc_douintvec_minmax_conv_param param = { 2674 .min = (unsigned int *) table->extra1, 2675 .max = (unsigned int *) table->extra2, 2676 }; 2677 return do_proc_douintvec(table, write, buffer, lenp, ppos, 2678 do_proc_douintvec_minmax_conv, ¶m); 2679 } 2680 2681 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp, 2682 unsigned int *valp, 2683 int write, void *data) 2684 { 2685 if (write) { 2686 unsigned int val; 2687 2688 val = round_pipe_size(*lvalp); 2689 if (val == 0) 2690 return -EINVAL; 2691 2692 *valp = val; 2693 } else { 2694 unsigned int val = *valp; 2695 *lvalp = (unsigned long) val; 2696 } 2697 2698 return 0; 2699 } 2700 2701 static int proc_dopipe_max_size(struct ctl_table *table, int write, 2702 void __user *buffer, size_t *lenp, loff_t *ppos) 2703 { 2704 return do_proc_douintvec(table, write, buffer, lenp, ppos, 2705 do_proc_dopipe_max_size_conv, NULL); 2706 } 2707 2708 static void validate_coredump_safety(void) 2709 { 2710 #ifdef CONFIG_COREDUMP 2711 if (suid_dumpable == SUID_DUMP_ROOT && 2712 core_pattern[0] != '/' && core_pattern[0] != '|') { 2713 printk(KERN_WARNING 2714 "Unsafe core_pattern used with fs.suid_dumpable=2.\n" 2715 "Pipe handler or fully qualified core dump path required.\n" 2716 "Set kernel.core_pattern before fs.suid_dumpable.\n" 2717 ); 2718 } 2719 #endif 2720 } 2721 2722 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write, 2723 void __user *buffer, size_t *lenp, loff_t *ppos) 2724 { 2725 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos); 2726 if (!error) 2727 validate_coredump_safety(); 2728 return error; 2729 } 2730 2731 #ifdef CONFIG_COREDUMP 2732 static int proc_dostring_coredump(struct ctl_table *table, int write, 2733 void __user *buffer, size_t *lenp, loff_t *ppos) 2734 { 2735 int error = proc_dostring(table, write, buffer, lenp, ppos); 2736 if (!error) 2737 validate_coredump_safety(); 2738 return error; 2739 } 2740 #endif 2741 2742 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write, 2743 void __user *buffer, 2744 size_t *lenp, loff_t *ppos, 2745 unsigned long convmul, 2746 unsigned long convdiv) 2747 { 2748 unsigned long *i, *min, *max; 2749 int vleft, first = 1, err = 0; 2750 size_t left; 2751 char *kbuf = NULL, *p; 2752 2753 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) { 2754 *lenp = 0; 2755 return 0; 2756 } 2757 2758 i = (unsigned long *) data; 2759 min = (unsigned long *) table->extra1; 2760 max = (unsigned long *) table->extra2; 2761 vleft = table->maxlen / sizeof(unsigned long); 2762 left = *lenp; 2763 2764 if (write) { 2765 if (proc_first_pos_non_zero_ignore(ppos, table)) 2766 goto out; 2767 2768 if (left > PAGE_SIZE - 1) 2769 left = PAGE_SIZE - 1; 2770 p = kbuf = memdup_user_nul(buffer, left); 2771 if (IS_ERR(kbuf)) 2772 return PTR_ERR(kbuf); 2773 } 2774 2775 for (; left && vleft--; i++, first = 0) { 2776 unsigned long val; 2777 2778 if (write) { 2779 bool neg; 2780 2781 left -= proc_skip_spaces(&p); 2782 2783 err = proc_get_long(&p, &left, &val, &neg, 2784 proc_wspace_sep, 2785 sizeof(proc_wspace_sep), NULL); 2786 if (err) 2787 break; 2788 if (neg) 2789 continue; 2790 val = convmul * val / convdiv; 2791 if ((min && val < *min) || (max && val > *max)) 2792 continue; 2793 *i = val; 2794 } else { 2795 val = convdiv * (*i) / convmul; 2796 if (!first) { 2797 err = proc_put_char(&buffer, &left, '\t'); 2798 if (err) 2799 break; 2800 } 2801 err = proc_put_long(&buffer, &left, val, false); 2802 if (err) 2803 break; 2804 } 2805 } 2806 2807 if (!write && !first && left && !err) 2808 err = proc_put_char(&buffer, &left, '\n'); 2809 if (write && !err) 2810 left -= proc_skip_spaces(&p); 2811 if (write) { 2812 kfree(kbuf); 2813 if (first) 2814 return err ? : -EINVAL; 2815 } 2816 *lenp -= left; 2817 out: 2818 *ppos += *lenp; 2819 return err; 2820 } 2821 2822 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write, 2823 void __user *buffer, 2824 size_t *lenp, loff_t *ppos, 2825 unsigned long convmul, 2826 unsigned long convdiv) 2827 { 2828 return __do_proc_doulongvec_minmax(table->data, table, write, 2829 buffer, lenp, ppos, convmul, convdiv); 2830 } 2831 2832 /** 2833 * proc_doulongvec_minmax - read a vector of long integers with min/max values 2834 * @table: the sysctl table 2835 * @write: %TRUE if this is a write to the sysctl file 2836 * @buffer: the user buffer 2837 * @lenp: the size of the user buffer 2838 * @ppos: file position 2839 * 2840 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 2841 * values from/to the user buffer, treated as an ASCII string. 2842 * 2843 * This routine will ensure the values are within the range specified by 2844 * table->extra1 (min) and table->extra2 (max). 2845 * 2846 * Returns 0 on success. 2847 */ 2848 int proc_doulongvec_minmax(struct ctl_table *table, int write, 2849 void __user *buffer, size_t *lenp, loff_t *ppos) 2850 { 2851 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l); 2852 } 2853 2854 /** 2855 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values 2856 * @table: the sysctl table 2857 * @write: %TRUE if this is a write to the sysctl file 2858 * @buffer: the user buffer 2859 * @lenp: the size of the user buffer 2860 * @ppos: file position 2861 * 2862 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 2863 * values from/to the user buffer, treated as an ASCII string. The values 2864 * are treated as milliseconds, and converted to jiffies when they are stored. 2865 * 2866 * This routine will ensure the values are within the range specified by 2867 * table->extra1 (min) and table->extra2 (max). 2868 * 2869 * Returns 0 on success. 2870 */ 2871 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 2872 void __user *buffer, 2873 size_t *lenp, loff_t *ppos) 2874 { 2875 return do_proc_doulongvec_minmax(table, write, buffer, 2876 lenp, ppos, HZ, 1000l); 2877 } 2878 2879 2880 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp, 2881 int *valp, 2882 int write, void *data) 2883 { 2884 if (write) { 2885 if (*lvalp > INT_MAX / HZ) 2886 return 1; 2887 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ); 2888 } else { 2889 int val = *valp; 2890 unsigned long lval; 2891 if (val < 0) { 2892 *negp = true; 2893 lval = -(unsigned long)val; 2894 } else { 2895 *negp = false; 2896 lval = (unsigned long)val; 2897 } 2898 *lvalp = lval / HZ; 2899 } 2900 return 0; 2901 } 2902 2903 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp, 2904 int *valp, 2905 int write, void *data) 2906 { 2907 if (write) { 2908 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ) 2909 return 1; 2910 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp); 2911 } else { 2912 int val = *valp; 2913 unsigned long lval; 2914 if (val < 0) { 2915 *negp = true; 2916 lval = -(unsigned long)val; 2917 } else { 2918 *negp = false; 2919 lval = (unsigned long)val; 2920 } 2921 *lvalp = jiffies_to_clock_t(lval); 2922 } 2923 return 0; 2924 } 2925 2926 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp, 2927 int *valp, 2928 int write, void *data) 2929 { 2930 if (write) { 2931 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp); 2932 2933 if (jif > INT_MAX) 2934 return 1; 2935 *valp = (int)jif; 2936 } else { 2937 int val = *valp; 2938 unsigned long lval; 2939 if (val < 0) { 2940 *negp = true; 2941 lval = -(unsigned long)val; 2942 } else { 2943 *negp = false; 2944 lval = (unsigned long)val; 2945 } 2946 *lvalp = jiffies_to_msecs(lval); 2947 } 2948 return 0; 2949 } 2950 2951 /** 2952 * proc_dointvec_jiffies - read a vector of integers as seconds 2953 * @table: the sysctl table 2954 * @write: %TRUE if this is a write to the sysctl file 2955 * @buffer: the user buffer 2956 * @lenp: the size of the user buffer 2957 * @ppos: file position 2958 * 2959 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2960 * values from/to the user buffer, treated as an ASCII string. 2961 * The values read are assumed to be in seconds, and are converted into 2962 * jiffies. 2963 * 2964 * Returns 0 on success. 2965 */ 2966 int proc_dointvec_jiffies(struct ctl_table *table, int write, 2967 void __user *buffer, size_t *lenp, loff_t *ppos) 2968 { 2969 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2970 do_proc_dointvec_jiffies_conv,NULL); 2971 } 2972 2973 /** 2974 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds 2975 * @table: the sysctl table 2976 * @write: %TRUE if this is a write to the sysctl file 2977 * @buffer: the user buffer 2978 * @lenp: the size of the user buffer 2979 * @ppos: pointer to the file position 2980 * 2981 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2982 * values from/to the user buffer, treated as an ASCII string. 2983 * The values read are assumed to be in 1/USER_HZ seconds, and 2984 * are converted into jiffies. 2985 * 2986 * Returns 0 on success. 2987 */ 2988 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 2989 void __user *buffer, size_t *lenp, loff_t *ppos) 2990 { 2991 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2992 do_proc_dointvec_userhz_jiffies_conv,NULL); 2993 } 2994 2995 /** 2996 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds 2997 * @table: the sysctl table 2998 * @write: %TRUE if this is a write to the sysctl file 2999 * @buffer: the user buffer 3000 * @lenp: the size of the user buffer 3001 * @ppos: file position 3002 * @ppos: the current position in the file 3003 * 3004 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 3005 * values from/to the user buffer, treated as an ASCII string. 3006 * The values read are assumed to be in 1/1000 seconds, and 3007 * are converted into jiffies. 3008 * 3009 * Returns 0 on success. 3010 */ 3011 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, 3012 void __user *buffer, size_t *lenp, loff_t *ppos) 3013 { 3014 return do_proc_dointvec(table, write, buffer, lenp, ppos, 3015 do_proc_dointvec_ms_jiffies_conv, NULL); 3016 } 3017 3018 static int proc_do_cad_pid(struct ctl_table *table, int write, 3019 void __user *buffer, size_t *lenp, loff_t *ppos) 3020 { 3021 struct pid *new_pid; 3022 pid_t tmp; 3023 int r; 3024 3025 tmp = pid_vnr(cad_pid); 3026 3027 r = __do_proc_dointvec(&tmp, table, write, buffer, 3028 lenp, ppos, NULL, NULL); 3029 if (r || !write) 3030 return r; 3031 3032 new_pid = find_get_pid(tmp); 3033 if (!new_pid) 3034 return -ESRCH; 3035 3036 put_pid(xchg(&cad_pid, new_pid)); 3037 return 0; 3038 } 3039 3040 /** 3041 * proc_do_large_bitmap - read/write from/to a large bitmap 3042 * @table: the sysctl table 3043 * @write: %TRUE if this is a write to the sysctl file 3044 * @buffer: the user buffer 3045 * @lenp: the size of the user buffer 3046 * @ppos: file position 3047 * 3048 * The bitmap is stored at table->data and the bitmap length (in bits) 3049 * in table->maxlen. 3050 * 3051 * We use a range comma separated format (e.g. 1,3-4,10-10) so that 3052 * large bitmaps may be represented in a compact manner. Writing into 3053 * the file will clear the bitmap then update it with the given input. 3054 * 3055 * Returns 0 on success. 3056 */ 3057 int proc_do_large_bitmap(struct ctl_table *table, int write, 3058 void __user *buffer, size_t *lenp, loff_t *ppos) 3059 { 3060 int err = 0; 3061 bool first = 1; 3062 size_t left = *lenp; 3063 unsigned long bitmap_len = table->maxlen; 3064 unsigned long *bitmap = *(unsigned long **) table->data; 3065 unsigned long *tmp_bitmap = NULL; 3066 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c; 3067 3068 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) { 3069 *lenp = 0; 3070 return 0; 3071 } 3072 3073 if (write) { 3074 char *kbuf, *p; 3075 3076 if (left > PAGE_SIZE - 1) 3077 left = PAGE_SIZE - 1; 3078 3079 p = kbuf = memdup_user_nul(buffer, left); 3080 if (IS_ERR(kbuf)) 3081 return PTR_ERR(kbuf); 3082 3083 tmp_bitmap = kcalloc(BITS_TO_LONGS(bitmap_len), 3084 sizeof(unsigned long), 3085 GFP_KERNEL); 3086 if (!tmp_bitmap) { 3087 kfree(kbuf); 3088 return -ENOMEM; 3089 } 3090 proc_skip_char(&p, &left, '\n'); 3091 while (!err && left) { 3092 unsigned long val_a, val_b; 3093 bool neg; 3094 3095 err = proc_get_long(&p, &left, &val_a, &neg, tr_a, 3096 sizeof(tr_a), &c); 3097 if (err) 3098 break; 3099 if (val_a >= bitmap_len || neg) { 3100 err = -EINVAL; 3101 break; 3102 } 3103 3104 val_b = val_a; 3105 if (left) { 3106 p++; 3107 left--; 3108 } 3109 3110 if (c == '-') { 3111 err = proc_get_long(&p, &left, &val_b, 3112 &neg, tr_b, sizeof(tr_b), 3113 &c); 3114 if (err) 3115 break; 3116 if (val_b >= bitmap_len || neg || 3117 val_a > val_b) { 3118 err = -EINVAL; 3119 break; 3120 } 3121 if (left) { 3122 p++; 3123 left--; 3124 } 3125 } 3126 3127 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1); 3128 first = 0; 3129 proc_skip_char(&p, &left, '\n'); 3130 } 3131 kfree(kbuf); 3132 } else { 3133 unsigned long bit_a, bit_b = 0; 3134 3135 while (left) { 3136 bit_a = find_next_bit(bitmap, bitmap_len, bit_b); 3137 if (bit_a >= bitmap_len) 3138 break; 3139 bit_b = find_next_zero_bit(bitmap, bitmap_len, 3140 bit_a + 1) - 1; 3141 3142 if (!first) { 3143 err = proc_put_char(&buffer, &left, ','); 3144 if (err) 3145 break; 3146 } 3147 err = proc_put_long(&buffer, &left, bit_a, false); 3148 if (err) 3149 break; 3150 if (bit_a != bit_b) { 3151 err = proc_put_char(&buffer, &left, '-'); 3152 if (err) 3153 break; 3154 err = proc_put_long(&buffer, &left, bit_b, false); 3155 if (err) 3156 break; 3157 } 3158 3159 first = 0; bit_b++; 3160 } 3161 if (!err) 3162 err = proc_put_char(&buffer, &left, '\n'); 3163 } 3164 3165 if (!err) { 3166 if (write) { 3167 if (*ppos) 3168 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len); 3169 else 3170 bitmap_copy(bitmap, tmp_bitmap, bitmap_len); 3171 } 3172 *lenp -= left; 3173 *ppos += *lenp; 3174 } 3175 3176 kfree(tmp_bitmap); 3177 return err; 3178 } 3179 3180 #else /* CONFIG_PROC_SYSCTL */ 3181 3182 int proc_dostring(struct ctl_table *table, int write, 3183 void __user *buffer, size_t *lenp, loff_t *ppos) 3184 { 3185 return -ENOSYS; 3186 } 3187 3188 int proc_dointvec(struct ctl_table *table, int write, 3189 void __user *buffer, size_t *lenp, loff_t *ppos) 3190 { 3191 return -ENOSYS; 3192 } 3193 3194 int proc_douintvec(struct ctl_table *table, int write, 3195 void __user *buffer, size_t *lenp, loff_t *ppos) 3196 { 3197 return -ENOSYS; 3198 } 3199 3200 int proc_dointvec_minmax(struct ctl_table *table, int write, 3201 void __user *buffer, size_t *lenp, loff_t *ppos) 3202 { 3203 return -ENOSYS; 3204 } 3205 3206 int proc_douintvec_minmax(struct ctl_table *table, int write, 3207 void __user *buffer, size_t *lenp, loff_t *ppos) 3208 { 3209 return -ENOSYS; 3210 } 3211 3212 int proc_dointvec_jiffies(struct ctl_table *table, int write, 3213 void __user *buffer, size_t *lenp, loff_t *ppos) 3214 { 3215 return -ENOSYS; 3216 } 3217 3218 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 3219 void __user *buffer, size_t *lenp, loff_t *ppos) 3220 { 3221 return -ENOSYS; 3222 } 3223 3224 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, 3225 void __user *buffer, size_t *lenp, loff_t *ppos) 3226 { 3227 return -ENOSYS; 3228 } 3229 3230 int proc_doulongvec_minmax(struct ctl_table *table, int write, 3231 void __user *buffer, size_t *lenp, loff_t *ppos) 3232 { 3233 return -ENOSYS; 3234 } 3235 3236 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 3237 void __user *buffer, 3238 size_t *lenp, loff_t *ppos) 3239 { 3240 return -ENOSYS; 3241 } 3242 3243 3244 #endif /* CONFIG_PROC_SYSCTL */ 3245 3246 /* 3247 * No sense putting this after each symbol definition, twice, 3248 * exception granted :-) 3249 */ 3250 EXPORT_SYMBOL(proc_dointvec); 3251 EXPORT_SYMBOL(proc_douintvec); 3252 EXPORT_SYMBOL(proc_dointvec_jiffies); 3253 EXPORT_SYMBOL(proc_dointvec_minmax); 3254 EXPORT_SYMBOL_GPL(proc_douintvec_minmax); 3255 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies); 3256 EXPORT_SYMBOL(proc_dointvec_ms_jiffies); 3257 EXPORT_SYMBOL(proc_dostring); 3258 EXPORT_SYMBOL(proc_doulongvec_minmax); 3259 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax); 3260