1 /* 2 * sysctl.c: General linux system control interface 3 * 4 * Begun 24 March 1995, Stephen Tweedie 5 * Added /proc support, Dec 1995 6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas. 7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver. 8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver. 9 * Dynamic registration fixes, Stephen Tweedie. 10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn. 11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris 12 * Horn. 13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer. 14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer. 15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill 16 * Wendling. 17 * The list_for_each() macro wasn't appropriate for the sysctl loop. 18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling 19 */ 20 21 #include <linux/module.h> 22 #include <linux/aio.h> 23 #include <linux/mm.h> 24 #include <linux/swap.h> 25 #include <linux/slab.h> 26 #include <linux/sysctl.h> 27 #include <linux/bitmap.h> 28 #include <linux/signal.h> 29 #include <linux/printk.h> 30 #include <linux/proc_fs.h> 31 #include <linux/security.h> 32 #include <linux/ctype.h> 33 #include <linux/kmemcheck.h> 34 #include <linux/kmemleak.h> 35 #include <linux/fs.h> 36 #include <linux/init.h> 37 #include <linux/kernel.h> 38 #include <linux/kobject.h> 39 #include <linux/net.h> 40 #include <linux/sysrq.h> 41 #include <linux/highuid.h> 42 #include <linux/writeback.h> 43 #include <linux/ratelimit.h> 44 #include <linux/compaction.h> 45 #include <linux/hugetlb.h> 46 #include <linux/initrd.h> 47 #include <linux/key.h> 48 #include <linux/times.h> 49 #include <linux/limits.h> 50 #include <linux/dcache.h> 51 #include <linux/dnotify.h> 52 #include <linux/syscalls.h> 53 #include <linux/vmstat.h> 54 #include <linux/nfs_fs.h> 55 #include <linux/acpi.h> 56 #include <linux/reboot.h> 57 #include <linux/ftrace.h> 58 #include <linux/perf_event.h> 59 #include <linux/kprobes.h> 60 #include <linux/pipe_fs_i.h> 61 #include <linux/oom.h> 62 #include <linux/kmod.h> 63 #include <linux/capability.h> 64 #include <linux/binfmts.h> 65 #include <linux/sched/sysctl.h> 66 #include <linux/sched/coredump.h> 67 #include <linux/kexec.h> 68 #include <linux/bpf.h> 69 #include <linux/mount.h> 70 71 #include <linux/uaccess.h> 72 #include <asm/processor.h> 73 74 #ifdef CONFIG_X86 75 #include <asm/nmi.h> 76 #include <asm/stacktrace.h> 77 #include <asm/io.h> 78 #endif 79 #ifdef CONFIG_SPARC 80 #include <asm/setup.h> 81 #endif 82 #ifdef CONFIG_BSD_PROCESS_ACCT 83 #include <linux/acct.h> 84 #endif 85 #ifdef CONFIG_RT_MUTEXES 86 #include <linux/rtmutex.h> 87 #endif 88 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT) 89 #include <linux/lockdep.h> 90 #endif 91 #ifdef CONFIG_CHR_DEV_SG 92 #include <scsi/sg.h> 93 #endif 94 95 #ifdef CONFIG_LOCKUP_DETECTOR 96 #include <linux/nmi.h> 97 #endif 98 99 #if defined(CONFIG_SYSCTL) 100 101 /* External variables not in a header file. */ 102 extern int suid_dumpable; 103 #ifdef CONFIG_COREDUMP 104 extern int core_uses_pid; 105 extern char core_pattern[]; 106 extern unsigned int core_pipe_limit; 107 #endif 108 extern int pid_max; 109 extern int pid_max_min, pid_max_max; 110 extern int percpu_pagelist_fraction; 111 extern int latencytop_enabled; 112 extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max; 113 #ifndef CONFIG_MMU 114 extern int sysctl_nr_trim_pages; 115 #endif 116 117 /* Constants used for minimum and maximum */ 118 #ifdef CONFIG_LOCKUP_DETECTOR 119 static int sixty = 60; 120 #endif 121 122 static int __maybe_unused neg_one = -1; 123 124 static int zero; 125 static int __maybe_unused one = 1; 126 static int __maybe_unused two = 2; 127 static int __maybe_unused four = 4; 128 static unsigned long one_ul = 1; 129 static int one_hundred = 100; 130 static int one_thousand = 1000; 131 #ifdef CONFIG_PRINTK 132 static int ten_thousand = 10000; 133 #endif 134 #ifdef CONFIG_PERF_EVENTS 135 static int six_hundred_forty_kb = 640 * 1024; 136 #endif 137 138 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */ 139 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE; 140 141 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */ 142 static int maxolduid = 65535; 143 static int minolduid; 144 145 static int ngroups_max = NGROUPS_MAX; 146 static const int cap_last_cap = CAP_LAST_CAP; 147 148 /*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */ 149 #ifdef CONFIG_DETECT_HUNG_TASK 150 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ); 151 #endif 152 153 #ifdef CONFIG_INOTIFY_USER 154 #include <linux/inotify.h> 155 #endif 156 #ifdef CONFIG_SPARC 157 #endif 158 159 #ifdef __hppa__ 160 extern int pwrsw_enabled; 161 #endif 162 163 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW 164 extern int unaligned_enabled; 165 #endif 166 167 #ifdef CONFIG_IA64 168 extern int unaligned_dump_stack; 169 #endif 170 171 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN 172 extern int no_unaligned_warning; 173 #endif 174 175 #ifdef CONFIG_PROC_SYSCTL 176 177 /** 178 * enum sysctl_writes_mode - supported sysctl write modes 179 * 180 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value 181 * to be written, and multiple writes on the same sysctl file descriptor 182 * will rewrite the sysctl value, regardless of file position. No warning 183 * is issued when the initial position is not 0. 184 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is 185 * not 0. 186 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at 187 * file position 0 and the value must be fully contained in the buffer 188 * sent to the write syscall. If dealing with strings respect the file 189 * position, but restrict this to the max length of the buffer, anything 190 * passed the max lenght will be ignored. Multiple writes will append 191 * to the buffer. 192 * 193 * These write modes control how current file position affects the behavior of 194 * updating sysctl values through the proc interface on each write. 195 */ 196 enum sysctl_writes_mode { 197 SYSCTL_WRITES_LEGACY = -1, 198 SYSCTL_WRITES_WARN = 0, 199 SYSCTL_WRITES_STRICT = 1, 200 }; 201 202 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT; 203 204 static int proc_do_cad_pid(struct ctl_table *table, int write, 205 void __user *buffer, size_t *lenp, loff_t *ppos); 206 static int proc_taint(struct ctl_table *table, int write, 207 void __user *buffer, size_t *lenp, loff_t *ppos); 208 #endif 209 210 #ifdef CONFIG_PRINTK 211 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write, 212 void __user *buffer, size_t *lenp, loff_t *ppos); 213 #endif 214 215 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write, 216 void __user *buffer, size_t *lenp, loff_t *ppos); 217 #ifdef CONFIG_COREDUMP 218 static int proc_dostring_coredump(struct ctl_table *table, int write, 219 void __user *buffer, size_t *lenp, loff_t *ppos); 220 #endif 221 222 #ifdef CONFIG_MAGIC_SYSRQ 223 /* Note: sysrq code uses it's own private copy */ 224 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE; 225 226 static int sysrq_sysctl_handler(struct ctl_table *table, int write, 227 void __user *buffer, size_t *lenp, 228 loff_t *ppos) 229 { 230 int error; 231 232 error = proc_dointvec(table, write, buffer, lenp, ppos); 233 if (error) 234 return error; 235 236 if (write) 237 sysrq_toggle_support(__sysrq_enabled); 238 239 return 0; 240 } 241 242 #endif 243 244 static struct ctl_table kern_table[]; 245 static struct ctl_table vm_table[]; 246 static struct ctl_table fs_table[]; 247 static struct ctl_table debug_table[]; 248 static struct ctl_table dev_table[]; 249 extern struct ctl_table random_table[]; 250 #ifdef CONFIG_EPOLL 251 extern struct ctl_table epoll_table[]; 252 #endif 253 254 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT 255 int sysctl_legacy_va_layout; 256 #endif 257 258 /* The default sysctl tables: */ 259 260 static struct ctl_table sysctl_base_table[] = { 261 { 262 .procname = "kernel", 263 .mode = 0555, 264 .child = kern_table, 265 }, 266 { 267 .procname = "vm", 268 .mode = 0555, 269 .child = vm_table, 270 }, 271 { 272 .procname = "fs", 273 .mode = 0555, 274 .child = fs_table, 275 }, 276 { 277 .procname = "debug", 278 .mode = 0555, 279 .child = debug_table, 280 }, 281 { 282 .procname = "dev", 283 .mode = 0555, 284 .child = dev_table, 285 }, 286 { } 287 }; 288 289 #ifdef CONFIG_SCHED_DEBUG 290 static int min_sched_granularity_ns = 100000; /* 100 usecs */ 291 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */ 292 static int min_wakeup_granularity_ns; /* 0 usecs */ 293 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */ 294 #ifdef CONFIG_SMP 295 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE; 296 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1; 297 #endif /* CONFIG_SMP */ 298 #endif /* CONFIG_SCHED_DEBUG */ 299 300 #ifdef CONFIG_COMPACTION 301 static int min_extfrag_threshold; 302 static int max_extfrag_threshold = 1000; 303 #endif 304 305 static struct ctl_table kern_table[] = { 306 { 307 .procname = "sched_child_runs_first", 308 .data = &sysctl_sched_child_runs_first, 309 .maxlen = sizeof(unsigned int), 310 .mode = 0644, 311 .proc_handler = proc_dointvec, 312 }, 313 #ifdef CONFIG_SCHED_DEBUG 314 { 315 .procname = "sched_min_granularity_ns", 316 .data = &sysctl_sched_min_granularity, 317 .maxlen = sizeof(unsigned int), 318 .mode = 0644, 319 .proc_handler = sched_proc_update_handler, 320 .extra1 = &min_sched_granularity_ns, 321 .extra2 = &max_sched_granularity_ns, 322 }, 323 { 324 .procname = "sched_latency_ns", 325 .data = &sysctl_sched_latency, 326 .maxlen = sizeof(unsigned int), 327 .mode = 0644, 328 .proc_handler = sched_proc_update_handler, 329 .extra1 = &min_sched_granularity_ns, 330 .extra2 = &max_sched_granularity_ns, 331 }, 332 { 333 .procname = "sched_wakeup_granularity_ns", 334 .data = &sysctl_sched_wakeup_granularity, 335 .maxlen = sizeof(unsigned int), 336 .mode = 0644, 337 .proc_handler = sched_proc_update_handler, 338 .extra1 = &min_wakeup_granularity_ns, 339 .extra2 = &max_wakeup_granularity_ns, 340 }, 341 #ifdef CONFIG_SMP 342 { 343 .procname = "sched_tunable_scaling", 344 .data = &sysctl_sched_tunable_scaling, 345 .maxlen = sizeof(enum sched_tunable_scaling), 346 .mode = 0644, 347 .proc_handler = sched_proc_update_handler, 348 .extra1 = &min_sched_tunable_scaling, 349 .extra2 = &max_sched_tunable_scaling, 350 }, 351 { 352 .procname = "sched_migration_cost_ns", 353 .data = &sysctl_sched_migration_cost, 354 .maxlen = sizeof(unsigned int), 355 .mode = 0644, 356 .proc_handler = proc_dointvec, 357 }, 358 { 359 .procname = "sched_nr_migrate", 360 .data = &sysctl_sched_nr_migrate, 361 .maxlen = sizeof(unsigned int), 362 .mode = 0644, 363 .proc_handler = proc_dointvec, 364 }, 365 { 366 .procname = "sched_time_avg_ms", 367 .data = &sysctl_sched_time_avg, 368 .maxlen = sizeof(unsigned int), 369 .mode = 0644, 370 .proc_handler = proc_dointvec_minmax, 371 .extra1 = &one, 372 }, 373 #ifdef CONFIG_SCHEDSTATS 374 { 375 .procname = "sched_schedstats", 376 .data = NULL, 377 .maxlen = sizeof(unsigned int), 378 .mode = 0644, 379 .proc_handler = sysctl_schedstats, 380 .extra1 = &zero, 381 .extra2 = &one, 382 }, 383 #endif /* CONFIG_SCHEDSTATS */ 384 #endif /* CONFIG_SMP */ 385 #ifdef CONFIG_NUMA_BALANCING 386 { 387 .procname = "numa_balancing_scan_delay_ms", 388 .data = &sysctl_numa_balancing_scan_delay, 389 .maxlen = sizeof(unsigned int), 390 .mode = 0644, 391 .proc_handler = proc_dointvec, 392 }, 393 { 394 .procname = "numa_balancing_scan_period_min_ms", 395 .data = &sysctl_numa_balancing_scan_period_min, 396 .maxlen = sizeof(unsigned int), 397 .mode = 0644, 398 .proc_handler = proc_dointvec, 399 }, 400 { 401 .procname = "numa_balancing_scan_period_max_ms", 402 .data = &sysctl_numa_balancing_scan_period_max, 403 .maxlen = sizeof(unsigned int), 404 .mode = 0644, 405 .proc_handler = proc_dointvec, 406 }, 407 { 408 .procname = "numa_balancing_scan_size_mb", 409 .data = &sysctl_numa_balancing_scan_size, 410 .maxlen = sizeof(unsigned int), 411 .mode = 0644, 412 .proc_handler = proc_dointvec_minmax, 413 .extra1 = &one, 414 }, 415 { 416 .procname = "numa_balancing", 417 .data = NULL, /* filled in by handler */ 418 .maxlen = sizeof(unsigned int), 419 .mode = 0644, 420 .proc_handler = sysctl_numa_balancing, 421 .extra1 = &zero, 422 .extra2 = &one, 423 }, 424 #endif /* CONFIG_NUMA_BALANCING */ 425 #endif /* CONFIG_SCHED_DEBUG */ 426 { 427 .procname = "sched_rt_period_us", 428 .data = &sysctl_sched_rt_period, 429 .maxlen = sizeof(unsigned int), 430 .mode = 0644, 431 .proc_handler = sched_rt_handler, 432 }, 433 { 434 .procname = "sched_rt_runtime_us", 435 .data = &sysctl_sched_rt_runtime, 436 .maxlen = sizeof(int), 437 .mode = 0644, 438 .proc_handler = sched_rt_handler, 439 }, 440 { 441 .procname = "sched_rr_timeslice_ms", 442 .data = &sysctl_sched_rr_timeslice, 443 .maxlen = sizeof(int), 444 .mode = 0644, 445 .proc_handler = sched_rr_handler, 446 }, 447 #ifdef CONFIG_SCHED_AUTOGROUP 448 { 449 .procname = "sched_autogroup_enabled", 450 .data = &sysctl_sched_autogroup_enabled, 451 .maxlen = sizeof(unsigned int), 452 .mode = 0644, 453 .proc_handler = proc_dointvec_minmax, 454 .extra1 = &zero, 455 .extra2 = &one, 456 }, 457 #endif 458 #ifdef CONFIG_CFS_BANDWIDTH 459 { 460 .procname = "sched_cfs_bandwidth_slice_us", 461 .data = &sysctl_sched_cfs_bandwidth_slice, 462 .maxlen = sizeof(unsigned int), 463 .mode = 0644, 464 .proc_handler = proc_dointvec_minmax, 465 .extra1 = &one, 466 }, 467 #endif 468 #ifdef CONFIG_PROVE_LOCKING 469 { 470 .procname = "prove_locking", 471 .data = &prove_locking, 472 .maxlen = sizeof(int), 473 .mode = 0644, 474 .proc_handler = proc_dointvec, 475 }, 476 #endif 477 #ifdef CONFIG_LOCK_STAT 478 { 479 .procname = "lock_stat", 480 .data = &lock_stat, 481 .maxlen = sizeof(int), 482 .mode = 0644, 483 .proc_handler = proc_dointvec, 484 }, 485 #endif 486 { 487 .procname = "panic", 488 .data = &panic_timeout, 489 .maxlen = sizeof(int), 490 .mode = 0644, 491 .proc_handler = proc_dointvec, 492 }, 493 #ifdef CONFIG_COREDUMP 494 { 495 .procname = "core_uses_pid", 496 .data = &core_uses_pid, 497 .maxlen = sizeof(int), 498 .mode = 0644, 499 .proc_handler = proc_dointvec, 500 }, 501 { 502 .procname = "core_pattern", 503 .data = core_pattern, 504 .maxlen = CORENAME_MAX_SIZE, 505 .mode = 0644, 506 .proc_handler = proc_dostring_coredump, 507 }, 508 { 509 .procname = "core_pipe_limit", 510 .data = &core_pipe_limit, 511 .maxlen = sizeof(unsigned int), 512 .mode = 0644, 513 .proc_handler = proc_dointvec, 514 }, 515 #endif 516 #ifdef CONFIG_PROC_SYSCTL 517 { 518 .procname = "tainted", 519 .maxlen = sizeof(long), 520 .mode = 0644, 521 .proc_handler = proc_taint, 522 }, 523 { 524 .procname = "sysctl_writes_strict", 525 .data = &sysctl_writes_strict, 526 .maxlen = sizeof(int), 527 .mode = 0644, 528 .proc_handler = proc_dointvec_minmax, 529 .extra1 = &neg_one, 530 .extra2 = &one, 531 }, 532 #endif 533 #ifdef CONFIG_LATENCYTOP 534 { 535 .procname = "latencytop", 536 .data = &latencytop_enabled, 537 .maxlen = sizeof(int), 538 .mode = 0644, 539 .proc_handler = sysctl_latencytop, 540 }, 541 #endif 542 #ifdef CONFIG_BLK_DEV_INITRD 543 { 544 .procname = "real-root-dev", 545 .data = &real_root_dev, 546 .maxlen = sizeof(int), 547 .mode = 0644, 548 .proc_handler = proc_dointvec, 549 }, 550 #endif 551 { 552 .procname = "print-fatal-signals", 553 .data = &print_fatal_signals, 554 .maxlen = sizeof(int), 555 .mode = 0644, 556 .proc_handler = proc_dointvec, 557 }, 558 #ifdef CONFIG_SPARC 559 { 560 .procname = "reboot-cmd", 561 .data = reboot_command, 562 .maxlen = 256, 563 .mode = 0644, 564 .proc_handler = proc_dostring, 565 }, 566 { 567 .procname = "stop-a", 568 .data = &stop_a_enabled, 569 .maxlen = sizeof (int), 570 .mode = 0644, 571 .proc_handler = proc_dointvec, 572 }, 573 { 574 .procname = "scons-poweroff", 575 .data = &scons_pwroff, 576 .maxlen = sizeof (int), 577 .mode = 0644, 578 .proc_handler = proc_dointvec, 579 }, 580 #endif 581 #ifdef CONFIG_SPARC64 582 { 583 .procname = "tsb-ratio", 584 .data = &sysctl_tsb_ratio, 585 .maxlen = sizeof (int), 586 .mode = 0644, 587 .proc_handler = proc_dointvec, 588 }, 589 #endif 590 #ifdef __hppa__ 591 { 592 .procname = "soft-power", 593 .data = &pwrsw_enabled, 594 .maxlen = sizeof (int), 595 .mode = 0644, 596 .proc_handler = proc_dointvec, 597 }, 598 #endif 599 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW 600 { 601 .procname = "unaligned-trap", 602 .data = &unaligned_enabled, 603 .maxlen = sizeof (int), 604 .mode = 0644, 605 .proc_handler = proc_dointvec, 606 }, 607 #endif 608 { 609 .procname = "ctrl-alt-del", 610 .data = &C_A_D, 611 .maxlen = sizeof(int), 612 .mode = 0644, 613 .proc_handler = proc_dointvec, 614 }, 615 #ifdef CONFIG_FUNCTION_TRACER 616 { 617 .procname = "ftrace_enabled", 618 .data = &ftrace_enabled, 619 .maxlen = sizeof(int), 620 .mode = 0644, 621 .proc_handler = ftrace_enable_sysctl, 622 }, 623 #endif 624 #ifdef CONFIG_STACK_TRACER 625 { 626 .procname = "stack_tracer_enabled", 627 .data = &stack_tracer_enabled, 628 .maxlen = sizeof(int), 629 .mode = 0644, 630 .proc_handler = stack_trace_sysctl, 631 }, 632 #endif 633 #ifdef CONFIG_TRACING 634 { 635 .procname = "ftrace_dump_on_oops", 636 .data = &ftrace_dump_on_oops, 637 .maxlen = sizeof(int), 638 .mode = 0644, 639 .proc_handler = proc_dointvec, 640 }, 641 { 642 .procname = "traceoff_on_warning", 643 .data = &__disable_trace_on_warning, 644 .maxlen = sizeof(__disable_trace_on_warning), 645 .mode = 0644, 646 .proc_handler = proc_dointvec, 647 }, 648 { 649 .procname = "tracepoint_printk", 650 .data = &tracepoint_printk, 651 .maxlen = sizeof(tracepoint_printk), 652 .mode = 0644, 653 .proc_handler = tracepoint_printk_sysctl, 654 }, 655 #endif 656 #ifdef CONFIG_KEXEC_CORE 657 { 658 .procname = "kexec_load_disabled", 659 .data = &kexec_load_disabled, 660 .maxlen = sizeof(int), 661 .mode = 0644, 662 /* only handle a transition from default "0" to "1" */ 663 .proc_handler = proc_dointvec_minmax, 664 .extra1 = &one, 665 .extra2 = &one, 666 }, 667 #endif 668 #ifdef CONFIG_MODULES 669 { 670 .procname = "modprobe", 671 .data = &modprobe_path, 672 .maxlen = KMOD_PATH_LEN, 673 .mode = 0644, 674 .proc_handler = proc_dostring, 675 }, 676 { 677 .procname = "modules_disabled", 678 .data = &modules_disabled, 679 .maxlen = sizeof(int), 680 .mode = 0644, 681 /* only handle a transition from default "0" to "1" */ 682 .proc_handler = proc_dointvec_minmax, 683 .extra1 = &one, 684 .extra2 = &one, 685 }, 686 #endif 687 #ifdef CONFIG_UEVENT_HELPER 688 { 689 .procname = "hotplug", 690 .data = &uevent_helper, 691 .maxlen = UEVENT_HELPER_PATH_LEN, 692 .mode = 0644, 693 .proc_handler = proc_dostring, 694 }, 695 #endif 696 #ifdef CONFIG_CHR_DEV_SG 697 { 698 .procname = "sg-big-buff", 699 .data = &sg_big_buff, 700 .maxlen = sizeof (int), 701 .mode = 0444, 702 .proc_handler = proc_dointvec, 703 }, 704 #endif 705 #ifdef CONFIG_BSD_PROCESS_ACCT 706 { 707 .procname = "acct", 708 .data = &acct_parm, 709 .maxlen = 3*sizeof(int), 710 .mode = 0644, 711 .proc_handler = proc_dointvec, 712 }, 713 #endif 714 #ifdef CONFIG_MAGIC_SYSRQ 715 { 716 .procname = "sysrq", 717 .data = &__sysrq_enabled, 718 .maxlen = sizeof (int), 719 .mode = 0644, 720 .proc_handler = sysrq_sysctl_handler, 721 }, 722 #endif 723 #ifdef CONFIG_PROC_SYSCTL 724 { 725 .procname = "cad_pid", 726 .data = NULL, 727 .maxlen = sizeof (int), 728 .mode = 0600, 729 .proc_handler = proc_do_cad_pid, 730 }, 731 #endif 732 { 733 .procname = "threads-max", 734 .data = NULL, 735 .maxlen = sizeof(int), 736 .mode = 0644, 737 .proc_handler = sysctl_max_threads, 738 }, 739 { 740 .procname = "random", 741 .mode = 0555, 742 .child = random_table, 743 }, 744 { 745 .procname = "usermodehelper", 746 .mode = 0555, 747 .child = usermodehelper_table, 748 }, 749 { 750 .procname = "overflowuid", 751 .data = &overflowuid, 752 .maxlen = sizeof(int), 753 .mode = 0644, 754 .proc_handler = proc_dointvec_minmax, 755 .extra1 = &minolduid, 756 .extra2 = &maxolduid, 757 }, 758 { 759 .procname = "overflowgid", 760 .data = &overflowgid, 761 .maxlen = sizeof(int), 762 .mode = 0644, 763 .proc_handler = proc_dointvec_minmax, 764 .extra1 = &minolduid, 765 .extra2 = &maxolduid, 766 }, 767 #ifdef CONFIG_S390 768 #ifdef CONFIG_MATHEMU 769 { 770 .procname = "ieee_emulation_warnings", 771 .data = &sysctl_ieee_emulation_warnings, 772 .maxlen = sizeof(int), 773 .mode = 0644, 774 .proc_handler = proc_dointvec, 775 }, 776 #endif 777 { 778 .procname = "userprocess_debug", 779 .data = &show_unhandled_signals, 780 .maxlen = sizeof(int), 781 .mode = 0644, 782 .proc_handler = proc_dointvec, 783 }, 784 #endif 785 { 786 .procname = "pid_max", 787 .data = &pid_max, 788 .maxlen = sizeof (int), 789 .mode = 0644, 790 .proc_handler = proc_dointvec_minmax, 791 .extra1 = &pid_max_min, 792 .extra2 = &pid_max_max, 793 }, 794 { 795 .procname = "panic_on_oops", 796 .data = &panic_on_oops, 797 .maxlen = sizeof(int), 798 .mode = 0644, 799 .proc_handler = proc_dointvec, 800 }, 801 #if defined CONFIG_PRINTK 802 { 803 .procname = "printk", 804 .data = &console_loglevel, 805 .maxlen = 4*sizeof(int), 806 .mode = 0644, 807 .proc_handler = proc_dointvec, 808 }, 809 { 810 .procname = "printk_ratelimit", 811 .data = &printk_ratelimit_state.interval, 812 .maxlen = sizeof(int), 813 .mode = 0644, 814 .proc_handler = proc_dointvec_jiffies, 815 }, 816 { 817 .procname = "printk_ratelimit_burst", 818 .data = &printk_ratelimit_state.burst, 819 .maxlen = sizeof(int), 820 .mode = 0644, 821 .proc_handler = proc_dointvec, 822 }, 823 { 824 .procname = "printk_delay", 825 .data = &printk_delay_msec, 826 .maxlen = sizeof(int), 827 .mode = 0644, 828 .proc_handler = proc_dointvec_minmax, 829 .extra1 = &zero, 830 .extra2 = &ten_thousand, 831 }, 832 { 833 .procname = "printk_devkmsg", 834 .data = devkmsg_log_str, 835 .maxlen = DEVKMSG_STR_MAX_SIZE, 836 .mode = 0644, 837 .proc_handler = devkmsg_sysctl_set_loglvl, 838 }, 839 { 840 .procname = "dmesg_restrict", 841 .data = &dmesg_restrict, 842 .maxlen = sizeof(int), 843 .mode = 0644, 844 .proc_handler = proc_dointvec_minmax_sysadmin, 845 .extra1 = &zero, 846 .extra2 = &one, 847 }, 848 { 849 .procname = "kptr_restrict", 850 .data = &kptr_restrict, 851 .maxlen = sizeof(int), 852 .mode = 0644, 853 .proc_handler = proc_dointvec_minmax_sysadmin, 854 .extra1 = &zero, 855 .extra2 = &two, 856 }, 857 #endif 858 { 859 .procname = "ngroups_max", 860 .data = &ngroups_max, 861 .maxlen = sizeof (int), 862 .mode = 0444, 863 .proc_handler = proc_dointvec, 864 }, 865 { 866 .procname = "cap_last_cap", 867 .data = (void *)&cap_last_cap, 868 .maxlen = sizeof(int), 869 .mode = 0444, 870 .proc_handler = proc_dointvec, 871 }, 872 #if defined(CONFIG_LOCKUP_DETECTOR) 873 { 874 .procname = "watchdog", 875 .data = &watchdog_user_enabled, 876 .maxlen = sizeof(int), 877 .mode = 0644, 878 .proc_handler = proc_watchdog, 879 .extra1 = &zero, 880 .extra2 = &one, 881 }, 882 { 883 .procname = "watchdog_thresh", 884 .data = &watchdog_thresh, 885 .maxlen = sizeof(int), 886 .mode = 0644, 887 .proc_handler = proc_watchdog_thresh, 888 .extra1 = &zero, 889 .extra2 = &sixty, 890 }, 891 { 892 .procname = "nmi_watchdog", 893 .data = &nmi_watchdog_user_enabled, 894 .maxlen = sizeof(int), 895 .mode = NMI_WATCHDOG_SYSCTL_PERM, 896 .proc_handler = proc_nmi_watchdog, 897 .extra1 = &zero, 898 .extra2 = &one, 899 }, 900 { 901 .procname = "watchdog_cpumask", 902 .data = &watchdog_cpumask_bits, 903 .maxlen = NR_CPUS, 904 .mode = 0644, 905 .proc_handler = proc_watchdog_cpumask, 906 }, 907 #ifdef CONFIG_SOFTLOCKUP_DETECTOR 908 { 909 .procname = "soft_watchdog", 910 .data = &soft_watchdog_user_enabled, 911 .maxlen = sizeof(int), 912 .mode = 0644, 913 .proc_handler = proc_soft_watchdog, 914 .extra1 = &zero, 915 .extra2 = &one, 916 }, 917 { 918 .procname = "softlockup_panic", 919 .data = &softlockup_panic, 920 .maxlen = sizeof(int), 921 .mode = 0644, 922 .proc_handler = proc_dointvec_minmax, 923 .extra1 = &zero, 924 .extra2 = &one, 925 }, 926 #ifdef CONFIG_SMP 927 { 928 .procname = "softlockup_all_cpu_backtrace", 929 .data = &sysctl_softlockup_all_cpu_backtrace, 930 .maxlen = sizeof(int), 931 .mode = 0644, 932 .proc_handler = proc_dointvec_minmax, 933 .extra1 = &zero, 934 .extra2 = &one, 935 }, 936 #endif /* CONFIG_SMP */ 937 #endif 938 #ifdef CONFIG_HARDLOCKUP_DETECTOR 939 { 940 .procname = "hardlockup_panic", 941 .data = &hardlockup_panic, 942 .maxlen = sizeof(int), 943 .mode = 0644, 944 .proc_handler = proc_dointvec_minmax, 945 .extra1 = &zero, 946 .extra2 = &one, 947 }, 948 #ifdef CONFIG_SMP 949 { 950 .procname = "hardlockup_all_cpu_backtrace", 951 .data = &sysctl_hardlockup_all_cpu_backtrace, 952 .maxlen = sizeof(int), 953 .mode = 0644, 954 .proc_handler = proc_dointvec_minmax, 955 .extra1 = &zero, 956 .extra2 = &one, 957 }, 958 #endif /* CONFIG_SMP */ 959 #endif 960 #endif 961 962 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86) 963 { 964 .procname = "unknown_nmi_panic", 965 .data = &unknown_nmi_panic, 966 .maxlen = sizeof (int), 967 .mode = 0644, 968 .proc_handler = proc_dointvec, 969 }, 970 #endif 971 #if defined(CONFIG_X86) 972 { 973 .procname = "panic_on_unrecovered_nmi", 974 .data = &panic_on_unrecovered_nmi, 975 .maxlen = sizeof(int), 976 .mode = 0644, 977 .proc_handler = proc_dointvec, 978 }, 979 { 980 .procname = "panic_on_io_nmi", 981 .data = &panic_on_io_nmi, 982 .maxlen = sizeof(int), 983 .mode = 0644, 984 .proc_handler = proc_dointvec, 985 }, 986 #ifdef CONFIG_DEBUG_STACKOVERFLOW 987 { 988 .procname = "panic_on_stackoverflow", 989 .data = &sysctl_panic_on_stackoverflow, 990 .maxlen = sizeof(int), 991 .mode = 0644, 992 .proc_handler = proc_dointvec, 993 }, 994 #endif 995 { 996 .procname = "bootloader_type", 997 .data = &bootloader_type, 998 .maxlen = sizeof (int), 999 .mode = 0444, 1000 .proc_handler = proc_dointvec, 1001 }, 1002 { 1003 .procname = "bootloader_version", 1004 .data = &bootloader_version, 1005 .maxlen = sizeof (int), 1006 .mode = 0444, 1007 .proc_handler = proc_dointvec, 1008 }, 1009 { 1010 .procname = "io_delay_type", 1011 .data = &io_delay_type, 1012 .maxlen = sizeof(int), 1013 .mode = 0644, 1014 .proc_handler = proc_dointvec, 1015 }, 1016 #endif 1017 #if defined(CONFIG_MMU) 1018 { 1019 .procname = "randomize_va_space", 1020 .data = &randomize_va_space, 1021 .maxlen = sizeof(int), 1022 .mode = 0644, 1023 .proc_handler = proc_dointvec, 1024 }, 1025 #endif 1026 #if defined(CONFIG_S390) && defined(CONFIG_SMP) 1027 { 1028 .procname = "spin_retry", 1029 .data = &spin_retry, 1030 .maxlen = sizeof (int), 1031 .mode = 0644, 1032 .proc_handler = proc_dointvec, 1033 }, 1034 #endif 1035 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86) 1036 { 1037 .procname = "acpi_video_flags", 1038 .data = &acpi_realmode_flags, 1039 .maxlen = sizeof (unsigned long), 1040 .mode = 0644, 1041 .proc_handler = proc_doulongvec_minmax, 1042 }, 1043 #endif 1044 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN 1045 { 1046 .procname = "ignore-unaligned-usertrap", 1047 .data = &no_unaligned_warning, 1048 .maxlen = sizeof (int), 1049 .mode = 0644, 1050 .proc_handler = proc_dointvec, 1051 }, 1052 #endif 1053 #ifdef CONFIG_IA64 1054 { 1055 .procname = "unaligned-dump-stack", 1056 .data = &unaligned_dump_stack, 1057 .maxlen = sizeof (int), 1058 .mode = 0644, 1059 .proc_handler = proc_dointvec, 1060 }, 1061 #endif 1062 #ifdef CONFIG_DETECT_HUNG_TASK 1063 { 1064 .procname = "hung_task_panic", 1065 .data = &sysctl_hung_task_panic, 1066 .maxlen = sizeof(int), 1067 .mode = 0644, 1068 .proc_handler = proc_dointvec_minmax, 1069 .extra1 = &zero, 1070 .extra2 = &one, 1071 }, 1072 { 1073 .procname = "hung_task_check_count", 1074 .data = &sysctl_hung_task_check_count, 1075 .maxlen = sizeof(int), 1076 .mode = 0644, 1077 .proc_handler = proc_dointvec_minmax, 1078 .extra1 = &zero, 1079 }, 1080 { 1081 .procname = "hung_task_timeout_secs", 1082 .data = &sysctl_hung_task_timeout_secs, 1083 .maxlen = sizeof(unsigned long), 1084 .mode = 0644, 1085 .proc_handler = proc_dohung_task_timeout_secs, 1086 .extra2 = &hung_task_timeout_max, 1087 }, 1088 { 1089 .procname = "hung_task_warnings", 1090 .data = &sysctl_hung_task_warnings, 1091 .maxlen = sizeof(int), 1092 .mode = 0644, 1093 .proc_handler = proc_dointvec_minmax, 1094 .extra1 = &neg_one, 1095 }, 1096 #endif 1097 #ifdef CONFIG_RT_MUTEXES 1098 { 1099 .procname = "max_lock_depth", 1100 .data = &max_lock_depth, 1101 .maxlen = sizeof(int), 1102 .mode = 0644, 1103 .proc_handler = proc_dointvec, 1104 }, 1105 #endif 1106 { 1107 .procname = "poweroff_cmd", 1108 .data = &poweroff_cmd, 1109 .maxlen = POWEROFF_CMD_PATH_LEN, 1110 .mode = 0644, 1111 .proc_handler = proc_dostring, 1112 }, 1113 #ifdef CONFIG_KEYS 1114 { 1115 .procname = "keys", 1116 .mode = 0555, 1117 .child = key_sysctls, 1118 }, 1119 #endif 1120 #ifdef CONFIG_PERF_EVENTS 1121 /* 1122 * User-space scripts rely on the existence of this file 1123 * as a feature check for perf_events being enabled. 1124 * 1125 * So it's an ABI, do not remove! 1126 */ 1127 { 1128 .procname = "perf_event_paranoid", 1129 .data = &sysctl_perf_event_paranoid, 1130 .maxlen = sizeof(sysctl_perf_event_paranoid), 1131 .mode = 0644, 1132 .proc_handler = proc_dointvec, 1133 }, 1134 { 1135 .procname = "perf_event_mlock_kb", 1136 .data = &sysctl_perf_event_mlock, 1137 .maxlen = sizeof(sysctl_perf_event_mlock), 1138 .mode = 0644, 1139 .proc_handler = proc_dointvec, 1140 }, 1141 { 1142 .procname = "perf_event_max_sample_rate", 1143 .data = &sysctl_perf_event_sample_rate, 1144 .maxlen = sizeof(sysctl_perf_event_sample_rate), 1145 .mode = 0644, 1146 .proc_handler = perf_proc_update_handler, 1147 .extra1 = &one, 1148 }, 1149 { 1150 .procname = "perf_cpu_time_max_percent", 1151 .data = &sysctl_perf_cpu_time_max_percent, 1152 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent), 1153 .mode = 0644, 1154 .proc_handler = perf_cpu_time_max_percent_handler, 1155 .extra1 = &zero, 1156 .extra2 = &one_hundred, 1157 }, 1158 { 1159 .procname = "perf_event_max_stack", 1160 .data = &sysctl_perf_event_max_stack, 1161 .maxlen = sizeof(sysctl_perf_event_max_stack), 1162 .mode = 0644, 1163 .proc_handler = perf_event_max_stack_handler, 1164 .extra1 = &zero, 1165 .extra2 = &six_hundred_forty_kb, 1166 }, 1167 { 1168 .procname = "perf_event_max_contexts_per_stack", 1169 .data = &sysctl_perf_event_max_contexts_per_stack, 1170 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack), 1171 .mode = 0644, 1172 .proc_handler = perf_event_max_stack_handler, 1173 .extra1 = &zero, 1174 .extra2 = &one_thousand, 1175 }, 1176 #endif 1177 #ifdef CONFIG_KMEMCHECK 1178 { 1179 .procname = "kmemcheck", 1180 .data = &kmemcheck_enabled, 1181 .maxlen = sizeof(int), 1182 .mode = 0644, 1183 .proc_handler = proc_dointvec, 1184 }, 1185 #endif 1186 { 1187 .procname = "panic_on_warn", 1188 .data = &panic_on_warn, 1189 .maxlen = sizeof(int), 1190 .mode = 0644, 1191 .proc_handler = proc_dointvec_minmax, 1192 .extra1 = &zero, 1193 .extra2 = &one, 1194 }, 1195 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) 1196 { 1197 .procname = "timer_migration", 1198 .data = &sysctl_timer_migration, 1199 .maxlen = sizeof(unsigned int), 1200 .mode = 0644, 1201 .proc_handler = timer_migration_handler, 1202 .extra1 = &zero, 1203 .extra2 = &one, 1204 }, 1205 #endif 1206 #ifdef CONFIG_BPF_SYSCALL 1207 { 1208 .procname = "unprivileged_bpf_disabled", 1209 .data = &sysctl_unprivileged_bpf_disabled, 1210 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled), 1211 .mode = 0644, 1212 /* only handle a transition from default "0" to "1" */ 1213 .proc_handler = proc_dointvec_minmax, 1214 .extra1 = &one, 1215 .extra2 = &one, 1216 }, 1217 #endif 1218 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU) 1219 { 1220 .procname = "panic_on_rcu_stall", 1221 .data = &sysctl_panic_on_rcu_stall, 1222 .maxlen = sizeof(sysctl_panic_on_rcu_stall), 1223 .mode = 0644, 1224 .proc_handler = proc_dointvec_minmax, 1225 .extra1 = &zero, 1226 .extra2 = &one, 1227 }, 1228 #endif 1229 { } 1230 }; 1231 1232 static struct ctl_table vm_table[] = { 1233 { 1234 .procname = "overcommit_memory", 1235 .data = &sysctl_overcommit_memory, 1236 .maxlen = sizeof(sysctl_overcommit_memory), 1237 .mode = 0644, 1238 .proc_handler = proc_dointvec_minmax, 1239 .extra1 = &zero, 1240 .extra2 = &two, 1241 }, 1242 { 1243 .procname = "panic_on_oom", 1244 .data = &sysctl_panic_on_oom, 1245 .maxlen = sizeof(sysctl_panic_on_oom), 1246 .mode = 0644, 1247 .proc_handler = proc_dointvec_minmax, 1248 .extra1 = &zero, 1249 .extra2 = &two, 1250 }, 1251 { 1252 .procname = "oom_kill_allocating_task", 1253 .data = &sysctl_oom_kill_allocating_task, 1254 .maxlen = sizeof(sysctl_oom_kill_allocating_task), 1255 .mode = 0644, 1256 .proc_handler = proc_dointvec, 1257 }, 1258 { 1259 .procname = "oom_dump_tasks", 1260 .data = &sysctl_oom_dump_tasks, 1261 .maxlen = sizeof(sysctl_oom_dump_tasks), 1262 .mode = 0644, 1263 .proc_handler = proc_dointvec, 1264 }, 1265 { 1266 .procname = "overcommit_ratio", 1267 .data = &sysctl_overcommit_ratio, 1268 .maxlen = sizeof(sysctl_overcommit_ratio), 1269 .mode = 0644, 1270 .proc_handler = overcommit_ratio_handler, 1271 }, 1272 { 1273 .procname = "overcommit_kbytes", 1274 .data = &sysctl_overcommit_kbytes, 1275 .maxlen = sizeof(sysctl_overcommit_kbytes), 1276 .mode = 0644, 1277 .proc_handler = overcommit_kbytes_handler, 1278 }, 1279 { 1280 .procname = "page-cluster", 1281 .data = &page_cluster, 1282 .maxlen = sizeof(int), 1283 .mode = 0644, 1284 .proc_handler = proc_dointvec_minmax, 1285 .extra1 = &zero, 1286 }, 1287 { 1288 .procname = "dirty_background_ratio", 1289 .data = &dirty_background_ratio, 1290 .maxlen = sizeof(dirty_background_ratio), 1291 .mode = 0644, 1292 .proc_handler = dirty_background_ratio_handler, 1293 .extra1 = &zero, 1294 .extra2 = &one_hundred, 1295 }, 1296 { 1297 .procname = "dirty_background_bytes", 1298 .data = &dirty_background_bytes, 1299 .maxlen = sizeof(dirty_background_bytes), 1300 .mode = 0644, 1301 .proc_handler = dirty_background_bytes_handler, 1302 .extra1 = &one_ul, 1303 }, 1304 { 1305 .procname = "dirty_ratio", 1306 .data = &vm_dirty_ratio, 1307 .maxlen = sizeof(vm_dirty_ratio), 1308 .mode = 0644, 1309 .proc_handler = dirty_ratio_handler, 1310 .extra1 = &zero, 1311 .extra2 = &one_hundred, 1312 }, 1313 { 1314 .procname = "dirty_bytes", 1315 .data = &vm_dirty_bytes, 1316 .maxlen = sizeof(vm_dirty_bytes), 1317 .mode = 0644, 1318 .proc_handler = dirty_bytes_handler, 1319 .extra1 = &dirty_bytes_min, 1320 }, 1321 { 1322 .procname = "dirty_writeback_centisecs", 1323 .data = &dirty_writeback_interval, 1324 .maxlen = sizeof(dirty_writeback_interval), 1325 .mode = 0644, 1326 .proc_handler = dirty_writeback_centisecs_handler, 1327 }, 1328 { 1329 .procname = "dirty_expire_centisecs", 1330 .data = &dirty_expire_interval, 1331 .maxlen = sizeof(dirty_expire_interval), 1332 .mode = 0644, 1333 .proc_handler = proc_dointvec_minmax, 1334 .extra1 = &zero, 1335 }, 1336 { 1337 .procname = "dirtytime_expire_seconds", 1338 .data = &dirtytime_expire_interval, 1339 .maxlen = sizeof(dirty_expire_interval), 1340 .mode = 0644, 1341 .proc_handler = dirtytime_interval_handler, 1342 .extra1 = &zero, 1343 }, 1344 { 1345 .procname = "nr_pdflush_threads", 1346 .mode = 0444 /* read-only */, 1347 .proc_handler = pdflush_proc_obsolete, 1348 }, 1349 { 1350 .procname = "swappiness", 1351 .data = &vm_swappiness, 1352 .maxlen = sizeof(vm_swappiness), 1353 .mode = 0644, 1354 .proc_handler = proc_dointvec_minmax, 1355 .extra1 = &zero, 1356 .extra2 = &one_hundred, 1357 }, 1358 #ifdef CONFIG_HUGETLB_PAGE 1359 { 1360 .procname = "nr_hugepages", 1361 .data = NULL, 1362 .maxlen = sizeof(unsigned long), 1363 .mode = 0644, 1364 .proc_handler = hugetlb_sysctl_handler, 1365 }, 1366 #ifdef CONFIG_NUMA 1367 { 1368 .procname = "nr_hugepages_mempolicy", 1369 .data = NULL, 1370 .maxlen = sizeof(unsigned long), 1371 .mode = 0644, 1372 .proc_handler = &hugetlb_mempolicy_sysctl_handler, 1373 }, 1374 #endif 1375 { 1376 .procname = "hugetlb_shm_group", 1377 .data = &sysctl_hugetlb_shm_group, 1378 .maxlen = sizeof(gid_t), 1379 .mode = 0644, 1380 .proc_handler = proc_dointvec, 1381 }, 1382 { 1383 .procname = "hugepages_treat_as_movable", 1384 .data = &hugepages_treat_as_movable, 1385 .maxlen = sizeof(int), 1386 .mode = 0644, 1387 .proc_handler = proc_dointvec, 1388 }, 1389 { 1390 .procname = "nr_overcommit_hugepages", 1391 .data = NULL, 1392 .maxlen = sizeof(unsigned long), 1393 .mode = 0644, 1394 .proc_handler = hugetlb_overcommit_handler, 1395 }, 1396 #endif 1397 { 1398 .procname = "lowmem_reserve_ratio", 1399 .data = &sysctl_lowmem_reserve_ratio, 1400 .maxlen = sizeof(sysctl_lowmem_reserve_ratio), 1401 .mode = 0644, 1402 .proc_handler = lowmem_reserve_ratio_sysctl_handler, 1403 }, 1404 { 1405 .procname = "drop_caches", 1406 .data = &sysctl_drop_caches, 1407 .maxlen = sizeof(int), 1408 .mode = 0644, 1409 .proc_handler = drop_caches_sysctl_handler, 1410 .extra1 = &one, 1411 .extra2 = &four, 1412 }, 1413 #ifdef CONFIG_COMPACTION 1414 { 1415 .procname = "compact_memory", 1416 .data = &sysctl_compact_memory, 1417 .maxlen = sizeof(int), 1418 .mode = 0200, 1419 .proc_handler = sysctl_compaction_handler, 1420 }, 1421 { 1422 .procname = "extfrag_threshold", 1423 .data = &sysctl_extfrag_threshold, 1424 .maxlen = sizeof(int), 1425 .mode = 0644, 1426 .proc_handler = sysctl_extfrag_handler, 1427 .extra1 = &min_extfrag_threshold, 1428 .extra2 = &max_extfrag_threshold, 1429 }, 1430 { 1431 .procname = "compact_unevictable_allowed", 1432 .data = &sysctl_compact_unevictable_allowed, 1433 .maxlen = sizeof(int), 1434 .mode = 0644, 1435 .proc_handler = proc_dointvec, 1436 .extra1 = &zero, 1437 .extra2 = &one, 1438 }, 1439 1440 #endif /* CONFIG_COMPACTION */ 1441 { 1442 .procname = "min_free_kbytes", 1443 .data = &min_free_kbytes, 1444 .maxlen = sizeof(min_free_kbytes), 1445 .mode = 0644, 1446 .proc_handler = min_free_kbytes_sysctl_handler, 1447 .extra1 = &zero, 1448 }, 1449 { 1450 .procname = "watermark_scale_factor", 1451 .data = &watermark_scale_factor, 1452 .maxlen = sizeof(watermark_scale_factor), 1453 .mode = 0644, 1454 .proc_handler = watermark_scale_factor_sysctl_handler, 1455 .extra1 = &one, 1456 .extra2 = &one_thousand, 1457 }, 1458 { 1459 .procname = "percpu_pagelist_fraction", 1460 .data = &percpu_pagelist_fraction, 1461 .maxlen = sizeof(percpu_pagelist_fraction), 1462 .mode = 0644, 1463 .proc_handler = percpu_pagelist_fraction_sysctl_handler, 1464 .extra1 = &zero, 1465 }, 1466 #ifdef CONFIG_MMU 1467 { 1468 .procname = "max_map_count", 1469 .data = &sysctl_max_map_count, 1470 .maxlen = sizeof(sysctl_max_map_count), 1471 .mode = 0644, 1472 .proc_handler = proc_dointvec_minmax, 1473 .extra1 = &zero, 1474 }, 1475 #else 1476 { 1477 .procname = "nr_trim_pages", 1478 .data = &sysctl_nr_trim_pages, 1479 .maxlen = sizeof(sysctl_nr_trim_pages), 1480 .mode = 0644, 1481 .proc_handler = proc_dointvec_minmax, 1482 .extra1 = &zero, 1483 }, 1484 #endif 1485 { 1486 .procname = "laptop_mode", 1487 .data = &laptop_mode, 1488 .maxlen = sizeof(laptop_mode), 1489 .mode = 0644, 1490 .proc_handler = proc_dointvec_jiffies, 1491 }, 1492 { 1493 .procname = "block_dump", 1494 .data = &block_dump, 1495 .maxlen = sizeof(block_dump), 1496 .mode = 0644, 1497 .proc_handler = proc_dointvec, 1498 .extra1 = &zero, 1499 }, 1500 { 1501 .procname = "vfs_cache_pressure", 1502 .data = &sysctl_vfs_cache_pressure, 1503 .maxlen = sizeof(sysctl_vfs_cache_pressure), 1504 .mode = 0644, 1505 .proc_handler = proc_dointvec, 1506 .extra1 = &zero, 1507 }, 1508 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT 1509 { 1510 .procname = "legacy_va_layout", 1511 .data = &sysctl_legacy_va_layout, 1512 .maxlen = sizeof(sysctl_legacy_va_layout), 1513 .mode = 0644, 1514 .proc_handler = proc_dointvec, 1515 .extra1 = &zero, 1516 }, 1517 #endif 1518 #ifdef CONFIG_NUMA 1519 { 1520 .procname = "zone_reclaim_mode", 1521 .data = &node_reclaim_mode, 1522 .maxlen = sizeof(node_reclaim_mode), 1523 .mode = 0644, 1524 .proc_handler = proc_dointvec, 1525 .extra1 = &zero, 1526 }, 1527 { 1528 .procname = "min_unmapped_ratio", 1529 .data = &sysctl_min_unmapped_ratio, 1530 .maxlen = sizeof(sysctl_min_unmapped_ratio), 1531 .mode = 0644, 1532 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler, 1533 .extra1 = &zero, 1534 .extra2 = &one_hundred, 1535 }, 1536 { 1537 .procname = "min_slab_ratio", 1538 .data = &sysctl_min_slab_ratio, 1539 .maxlen = sizeof(sysctl_min_slab_ratio), 1540 .mode = 0644, 1541 .proc_handler = sysctl_min_slab_ratio_sysctl_handler, 1542 .extra1 = &zero, 1543 .extra2 = &one_hundred, 1544 }, 1545 #endif 1546 #ifdef CONFIG_SMP 1547 { 1548 .procname = "stat_interval", 1549 .data = &sysctl_stat_interval, 1550 .maxlen = sizeof(sysctl_stat_interval), 1551 .mode = 0644, 1552 .proc_handler = proc_dointvec_jiffies, 1553 }, 1554 { 1555 .procname = "stat_refresh", 1556 .data = NULL, 1557 .maxlen = 0, 1558 .mode = 0600, 1559 .proc_handler = vmstat_refresh, 1560 }, 1561 #endif 1562 #ifdef CONFIG_MMU 1563 { 1564 .procname = "mmap_min_addr", 1565 .data = &dac_mmap_min_addr, 1566 .maxlen = sizeof(unsigned long), 1567 .mode = 0644, 1568 .proc_handler = mmap_min_addr_handler, 1569 }, 1570 #endif 1571 #ifdef CONFIG_NUMA 1572 { 1573 .procname = "numa_zonelist_order", 1574 .data = &numa_zonelist_order, 1575 .maxlen = NUMA_ZONELIST_ORDER_LEN, 1576 .mode = 0644, 1577 .proc_handler = numa_zonelist_order_handler, 1578 }, 1579 #endif 1580 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \ 1581 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL)) 1582 { 1583 .procname = "vdso_enabled", 1584 #ifdef CONFIG_X86_32 1585 .data = &vdso32_enabled, 1586 .maxlen = sizeof(vdso32_enabled), 1587 #else 1588 .data = &vdso_enabled, 1589 .maxlen = sizeof(vdso_enabled), 1590 #endif 1591 .mode = 0644, 1592 .proc_handler = proc_dointvec, 1593 .extra1 = &zero, 1594 }, 1595 #endif 1596 #ifdef CONFIG_HIGHMEM 1597 { 1598 .procname = "highmem_is_dirtyable", 1599 .data = &vm_highmem_is_dirtyable, 1600 .maxlen = sizeof(vm_highmem_is_dirtyable), 1601 .mode = 0644, 1602 .proc_handler = proc_dointvec_minmax, 1603 .extra1 = &zero, 1604 .extra2 = &one, 1605 }, 1606 #endif 1607 #ifdef CONFIG_MEMORY_FAILURE 1608 { 1609 .procname = "memory_failure_early_kill", 1610 .data = &sysctl_memory_failure_early_kill, 1611 .maxlen = sizeof(sysctl_memory_failure_early_kill), 1612 .mode = 0644, 1613 .proc_handler = proc_dointvec_minmax, 1614 .extra1 = &zero, 1615 .extra2 = &one, 1616 }, 1617 { 1618 .procname = "memory_failure_recovery", 1619 .data = &sysctl_memory_failure_recovery, 1620 .maxlen = sizeof(sysctl_memory_failure_recovery), 1621 .mode = 0644, 1622 .proc_handler = proc_dointvec_minmax, 1623 .extra1 = &zero, 1624 .extra2 = &one, 1625 }, 1626 #endif 1627 { 1628 .procname = "user_reserve_kbytes", 1629 .data = &sysctl_user_reserve_kbytes, 1630 .maxlen = sizeof(sysctl_user_reserve_kbytes), 1631 .mode = 0644, 1632 .proc_handler = proc_doulongvec_minmax, 1633 }, 1634 { 1635 .procname = "admin_reserve_kbytes", 1636 .data = &sysctl_admin_reserve_kbytes, 1637 .maxlen = sizeof(sysctl_admin_reserve_kbytes), 1638 .mode = 0644, 1639 .proc_handler = proc_doulongvec_minmax, 1640 }, 1641 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS 1642 { 1643 .procname = "mmap_rnd_bits", 1644 .data = &mmap_rnd_bits, 1645 .maxlen = sizeof(mmap_rnd_bits), 1646 .mode = 0600, 1647 .proc_handler = proc_dointvec_minmax, 1648 .extra1 = (void *)&mmap_rnd_bits_min, 1649 .extra2 = (void *)&mmap_rnd_bits_max, 1650 }, 1651 #endif 1652 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS 1653 { 1654 .procname = "mmap_rnd_compat_bits", 1655 .data = &mmap_rnd_compat_bits, 1656 .maxlen = sizeof(mmap_rnd_compat_bits), 1657 .mode = 0600, 1658 .proc_handler = proc_dointvec_minmax, 1659 .extra1 = (void *)&mmap_rnd_compat_bits_min, 1660 .extra2 = (void *)&mmap_rnd_compat_bits_max, 1661 }, 1662 #endif 1663 { } 1664 }; 1665 1666 static struct ctl_table fs_table[] = { 1667 { 1668 .procname = "inode-nr", 1669 .data = &inodes_stat, 1670 .maxlen = 2*sizeof(long), 1671 .mode = 0444, 1672 .proc_handler = proc_nr_inodes, 1673 }, 1674 { 1675 .procname = "inode-state", 1676 .data = &inodes_stat, 1677 .maxlen = 7*sizeof(long), 1678 .mode = 0444, 1679 .proc_handler = proc_nr_inodes, 1680 }, 1681 { 1682 .procname = "file-nr", 1683 .data = &files_stat, 1684 .maxlen = sizeof(files_stat), 1685 .mode = 0444, 1686 .proc_handler = proc_nr_files, 1687 }, 1688 { 1689 .procname = "file-max", 1690 .data = &files_stat.max_files, 1691 .maxlen = sizeof(files_stat.max_files), 1692 .mode = 0644, 1693 .proc_handler = proc_doulongvec_minmax, 1694 }, 1695 { 1696 .procname = "nr_open", 1697 .data = &sysctl_nr_open, 1698 .maxlen = sizeof(unsigned int), 1699 .mode = 0644, 1700 .proc_handler = proc_dointvec_minmax, 1701 .extra1 = &sysctl_nr_open_min, 1702 .extra2 = &sysctl_nr_open_max, 1703 }, 1704 { 1705 .procname = "dentry-state", 1706 .data = &dentry_stat, 1707 .maxlen = 6*sizeof(long), 1708 .mode = 0444, 1709 .proc_handler = proc_nr_dentry, 1710 }, 1711 { 1712 .procname = "overflowuid", 1713 .data = &fs_overflowuid, 1714 .maxlen = sizeof(int), 1715 .mode = 0644, 1716 .proc_handler = proc_dointvec_minmax, 1717 .extra1 = &minolduid, 1718 .extra2 = &maxolduid, 1719 }, 1720 { 1721 .procname = "overflowgid", 1722 .data = &fs_overflowgid, 1723 .maxlen = sizeof(int), 1724 .mode = 0644, 1725 .proc_handler = proc_dointvec_minmax, 1726 .extra1 = &minolduid, 1727 .extra2 = &maxolduid, 1728 }, 1729 #ifdef CONFIG_FILE_LOCKING 1730 { 1731 .procname = "leases-enable", 1732 .data = &leases_enable, 1733 .maxlen = sizeof(int), 1734 .mode = 0644, 1735 .proc_handler = proc_dointvec, 1736 }, 1737 #endif 1738 #ifdef CONFIG_DNOTIFY 1739 { 1740 .procname = "dir-notify-enable", 1741 .data = &dir_notify_enable, 1742 .maxlen = sizeof(int), 1743 .mode = 0644, 1744 .proc_handler = proc_dointvec, 1745 }, 1746 #endif 1747 #ifdef CONFIG_MMU 1748 #ifdef CONFIG_FILE_LOCKING 1749 { 1750 .procname = "lease-break-time", 1751 .data = &lease_break_time, 1752 .maxlen = sizeof(int), 1753 .mode = 0644, 1754 .proc_handler = proc_dointvec, 1755 }, 1756 #endif 1757 #ifdef CONFIG_AIO 1758 { 1759 .procname = "aio-nr", 1760 .data = &aio_nr, 1761 .maxlen = sizeof(aio_nr), 1762 .mode = 0444, 1763 .proc_handler = proc_doulongvec_minmax, 1764 }, 1765 { 1766 .procname = "aio-max-nr", 1767 .data = &aio_max_nr, 1768 .maxlen = sizeof(aio_max_nr), 1769 .mode = 0644, 1770 .proc_handler = proc_doulongvec_minmax, 1771 }, 1772 #endif /* CONFIG_AIO */ 1773 #ifdef CONFIG_INOTIFY_USER 1774 { 1775 .procname = "inotify", 1776 .mode = 0555, 1777 .child = inotify_table, 1778 }, 1779 #endif 1780 #ifdef CONFIG_EPOLL 1781 { 1782 .procname = "epoll", 1783 .mode = 0555, 1784 .child = epoll_table, 1785 }, 1786 #endif 1787 #endif 1788 { 1789 .procname = "protected_symlinks", 1790 .data = &sysctl_protected_symlinks, 1791 .maxlen = sizeof(int), 1792 .mode = 0600, 1793 .proc_handler = proc_dointvec_minmax, 1794 .extra1 = &zero, 1795 .extra2 = &one, 1796 }, 1797 { 1798 .procname = "protected_hardlinks", 1799 .data = &sysctl_protected_hardlinks, 1800 .maxlen = sizeof(int), 1801 .mode = 0600, 1802 .proc_handler = proc_dointvec_minmax, 1803 .extra1 = &zero, 1804 .extra2 = &one, 1805 }, 1806 { 1807 .procname = "suid_dumpable", 1808 .data = &suid_dumpable, 1809 .maxlen = sizeof(int), 1810 .mode = 0644, 1811 .proc_handler = proc_dointvec_minmax_coredump, 1812 .extra1 = &zero, 1813 .extra2 = &two, 1814 }, 1815 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE) 1816 { 1817 .procname = "binfmt_misc", 1818 .mode = 0555, 1819 .child = sysctl_mount_point, 1820 }, 1821 #endif 1822 { 1823 .procname = "pipe-max-size", 1824 .data = &pipe_max_size, 1825 .maxlen = sizeof(int), 1826 .mode = 0644, 1827 .proc_handler = &pipe_proc_fn, 1828 .extra1 = &pipe_min_size, 1829 }, 1830 { 1831 .procname = "pipe-user-pages-hard", 1832 .data = &pipe_user_pages_hard, 1833 .maxlen = sizeof(pipe_user_pages_hard), 1834 .mode = 0644, 1835 .proc_handler = proc_doulongvec_minmax, 1836 }, 1837 { 1838 .procname = "pipe-user-pages-soft", 1839 .data = &pipe_user_pages_soft, 1840 .maxlen = sizeof(pipe_user_pages_soft), 1841 .mode = 0644, 1842 .proc_handler = proc_doulongvec_minmax, 1843 }, 1844 { 1845 .procname = "mount-max", 1846 .data = &sysctl_mount_max, 1847 .maxlen = sizeof(unsigned int), 1848 .mode = 0644, 1849 .proc_handler = proc_dointvec_minmax, 1850 .extra1 = &one, 1851 }, 1852 { } 1853 }; 1854 1855 static struct ctl_table debug_table[] = { 1856 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE 1857 { 1858 .procname = "exception-trace", 1859 .data = &show_unhandled_signals, 1860 .maxlen = sizeof(int), 1861 .mode = 0644, 1862 .proc_handler = proc_dointvec 1863 }, 1864 #endif 1865 #if defined(CONFIG_OPTPROBES) 1866 { 1867 .procname = "kprobes-optimization", 1868 .data = &sysctl_kprobes_optimization, 1869 .maxlen = sizeof(int), 1870 .mode = 0644, 1871 .proc_handler = proc_kprobes_optimization_handler, 1872 .extra1 = &zero, 1873 .extra2 = &one, 1874 }, 1875 #endif 1876 { } 1877 }; 1878 1879 static struct ctl_table dev_table[] = { 1880 { } 1881 }; 1882 1883 int __init sysctl_init(void) 1884 { 1885 struct ctl_table_header *hdr; 1886 1887 hdr = register_sysctl_table(sysctl_base_table); 1888 kmemleak_not_leak(hdr); 1889 return 0; 1890 } 1891 1892 #endif /* CONFIG_SYSCTL */ 1893 1894 /* 1895 * /proc/sys support 1896 */ 1897 1898 #ifdef CONFIG_PROC_SYSCTL 1899 1900 static int _proc_do_string(char *data, int maxlen, int write, 1901 char __user *buffer, 1902 size_t *lenp, loff_t *ppos) 1903 { 1904 size_t len; 1905 char __user *p; 1906 char c; 1907 1908 if (!data || !maxlen || !*lenp) { 1909 *lenp = 0; 1910 return 0; 1911 } 1912 1913 if (write) { 1914 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) { 1915 /* Only continue writes not past the end of buffer. */ 1916 len = strlen(data); 1917 if (len > maxlen - 1) 1918 len = maxlen - 1; 1919 1920 if (*ppos > len) 1921 return 0; 1922 len = *ppos; 1923 } else { 1924 /* Start writing from beginning of buffer. */ 1925 len = 0; 1926 } 1927 1928 *ppos += *lenp; 1929 p = buffer; 1930 while ((p - buffer) < *lenp && len < maxlen - 1) { 1931 if (get_user(c, p++)) 1932 return -EFAULT; 1933 if (c == 0 || c == '\n') 1934 break; 1935 data[len++] = c; 1936 } 1937 data[len] = 0; 1938 } else { 1939 len = strlen(data); 1940 if (len > maxlen) 1941 len = maxlen; 1942 1943 if (*ppos > len) { 1944 *lenp = 0; 1945 return 0; 1946 } 1947 1948 data += *ppos; 1949 len -= *ppos; 1950 1951 if (len > *lenp) 1952 len = *lenp; 1953 if (len) 1954 if (copy_to_user(buffer, data, len)) 1955 return -EFAULT; 1956 if (len < *lenp) { 1957 if (put_user('\n', buffer + len)) 1958 return -EFAULT; 1959 len++; 1960 } 1961 *lenp = len; 1962 *ppos += len; 1963 } 1964 return 0; 1965 } 1966 1967 static void warn_sysctl_write(struct ctl_table *table) 1968 { 1969 pr_warn_once("%s wrote to %s when file position was not 0!\n" 1970 "This will not be supported in the future. To silence this\n" 1971 "warning, set kernel.sysctl_writes_strict = -1\n", 1972 current->comm, table->procname); 1973 } 1974 1975 /** 1976 * proc_first_pos_non_zero_ignore - check if firs position is allowed 1977 * @ppos: file position 1978 * @table: the sysctl table 1979 * 1980 * Returns true if the first position is non-zero and the sysctl_writes_strict 1981 * mode indicates this is not allowed for numeric input types. String proc 1982 * hadlers can ignore the return value. 1983 */ 1984 static bool proc_first_pos_non_zero_ignore(loff_t *ppos, 1985 struct ctl_table *table) 1986 { 1987 if (!*ppos) 1988 return false; 1989 1990 switch (sysctl_writes_strict) { 1991 case SYSCTL_WRITES_STRICT: 1992 return true; 1993 case SYSCTL_WRITES_WARN: 1994 warn_sysctl_write(table); 1995 return false; 1996 default: 1997 return false; 1998 } 1999 } 2000 2001 /** 2002 * proc_dostring - read a string sysctl 2003 * @table: the sysctl table 2004 * @write: %TRUE if this is a write to the sysctl file 2005 * @buffer: the user buffer 2006 * @lenp: the size of the user buffer 2007 * @ppos: file position 2008 * 2009 * Reads/writes a string from/to the user buffer. If the kernel 2010 * buffer provided is not large enough to hold the string, the 2011 * string is truncated. The copied string is %NULL-terminated. 2012 * If the string is being read by the user process, it is copied 2013 * and a newline '\n' is added. It is truncated if the buffer is 2014 * not large enough. 2015 * 2016 * Returns 0 on success. 2017 */ 2018 int proc_dostring(struct ctl_table *table, int write, 2019 void __user *buffer, size_t *lenp, loff_t *ppos) 2020 { 2021 if (write) 2022 proc_first_pos_non_zero_ignore(ppos, table); 2023 2024 return _proc_do_string((char *)(table->data), table->maxlen, write, 2025 (char __user *)buffer, lenp, ppos); 2026 } 2027 2028 static size_t proc_skip_spaces(char **buf) 2029 { 2030 size_t ret; 2031 char *tmp = skip_spaces(*buf); 2032 ret = tmp - *buf; 2033 *buf = tmp; 2034 return ret; 2035 } 2036 2037 static void proc_skip_char(char **buf, size_t *size, const char v) 2038 { 2039 while (*size) { 2040 if (**buf != v) 2041 break; 2042 (*size)--; 2043 (*buf)++; 2044 } 2045 } 2046 2047 #define TMPBUFLEN 22 2048 /** 2049 * proc_get_long - reads an ASCII formatted integer from a user buffer 2050 * 2051 * @buf: a kernel buffer 2052 * @size: size of the kernel buffer 2053 * @val: this is where the number will be stored 2054 * @neg: set to %TRUE if number is negative 2055 * @perm_tr: a vector which contains the allowed trailers 2056 * @perm_tr_len: size of the perm_tr vector 2057 * @tr: pointer to store the trailer character 2058 * 2059 * In case of success %0 is returned and @buf and @size are updated with 2060 * the amount of bytes read. If @tr is non-NULL and a trailing 2061 * character exists (size is non-zero after returning from this 2062 * function), @tr is updated with the trailing character. 2063 */ 2064 static int proc_get_long(char **buf, size_t *size, 2065 unsigned long *val, bool *neg, 2066 const char *perm_tr, unsigned perm_tr_len, char *tr) 2067 { 2068 int len; 2069 char *p, tmp[TMPBUFLEN]; 2070 2071 if (!*size) 2072 return -EINVAL; 2073 2074 len = *size; 2075 if (len > TMPBUFLEN - 1) 2076 len = TMPBUFLEN - 1; 2077 2078 memcpy(tmp, *buf, len); 2079 2080 tmp[len] = 0; 2081 p = tmp; 2082 if (*p == '-' && *size > 1) { 2083 *neg = true; 2084 p++; 2085 } else 2086 *neg = false; 2087 if (!isdigit(*p)) 2088 return -EINVAL; 2089 2090 *val = simple_strtoul(p, &p, 0); 2091 2092 len = p - tmp; 2093 2094 /* We don't know if the next char is whitespace thus we may accept 2095 * invalid integers (e.g. 1234...a) or two integers instead of one 2096 * (e.g. 123...1). So lets not allow such large numbers. */ 2097 if (len == TMPBUFLEN - 1) 2098 return -EINVAL; 2099 2100 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len)) 2101 return -EINVAL; 2102 2103 if (tr && (len < *size)) 2104 *tr = *p; 2105 2106 *buf += len; 2107 *size -= len; 2108 2109 return 0; 2110 } 2111 2112 /** 2113 * proc_put_long - converts an integer to a decimal ASCII formatted string 2114 * 2115 * @buf: the user buffer 2116 * @size: the size of the user buffer 2117 * @val: the integer to be converted 2118 * @neg: sign of the number, %TRUE for negative 2119 * 2120 * In case of success %0 is returned and @buf and @size are updated with 2121 * the amount of bytes written. 2122 */ 2123 static int proc_put_long(void __user **buf, size_t *size, unsigned long val, 2124 bool neg) 2125 { 2126 int len; 2127 char tmp[TMPBUFLEN], *p = tmp; 2128 2129 sprintf(p, "%s%lu", neg ? "-" : "", val); 2130 len = strlen(tmp); 2131 if (len > *size) 2132 len = *size; 2133 if (copy_to_user(*buf, tmp, len)) 2134 return -EFAULT; 2135 *size -= len; 2136 *buf += len; 2137 return 0; 2138 } 2139 #undef TMPBUFLEN 2140 2141 static int proc_put_char(void __user **buf, size_t *size, char c) 2142 { 2143 if (*size) { 2144 char __user **buffer = (char __user **)buf; 2145 if (put_user(c, *buffer)) 2146 return -EFAULT; 2147 (*size)--, (*buffer)++; 2148 *buf = *buffer; 2149 } 2150 return 0; 2151 } 2152 2153 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp, 2154 int *valp, 2155 int write, void *data) 2156 { 2157 if (write) { 2158 if (*negp) { 2159 if (*lvalp > (unsigned long) INT_MAX + 1) 2160 return -EINVAL; 2161 *valp = -*lvalp; 2162 } else { 2163 if (*lvalp > (unsigned long) INT_MAX) 2164 return -EINVAL; 2165 *valp = *lvalp; 2166 } 2167 } else { 2168 int val = *valp; 2169 if (val < 0) { 2170 *negp = true; 2171 *lvalp = -(unsigned long)val; 2172 } else { 2173 *negp = false; 2174 *lvalp = (unsigned long)val; 2175 } 2176 } 2177 return 0; 2178 } 2179 2180 static int do_proc_douintvec_conv(unsigned long *lvalp, 2181 unsigned int *valp, 2182 int write, void *data) 2183 { 2184 if (write) { 2185 if (*lvalp > UINT_MAX) 2186 return -EINVAL; 2187 *valp = *lvalp; 2188 } else { 2189 unsigned int val = *valp; 2190 *lvalp = (unsigned long)val; 2191 } 2192 return 0; 2193 } 2194 2195 static const char proc_wspace_sep[] = { ' ', '\t', '\n' }; 2196 2197 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table, 2198 int write, void __user *buffer, 2199 size_t *lenp, loff_t *ppos, 2200 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 2201 int write, void *data), 2202 void *data) 2203 { 2204 int *i, vleft, first = 1, err = 0; 2205 size_t left; 2206 char *kbuf = NULL, *p; 2207 2208 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) { 2209 *lenp = 0; 2210 return 0; 2211 } 2212 2213 i = (int *) tbl_data; 2214 vleft = table->maxlen / sizeof(*i); 2215 left = *lenp; 2216 2217 if (!conv) 2218 conv = do_proc_dointvec_conv; 2219 2220 if (write) { 2221 if (proc_first_pos_non_zero_ignore(ppos, table)) 2222 goto out; 2223 2224 if (left > PAGE_SIZE - 1) 2225 left = PAGE_SIZE - 1; 2226 p = kbuf = memdup_user_nul(buffer, left); 2227 if (IS_ERR(kbuf)) 2228 return PTR_ERR(kbuf); 2229 } 2230 2231 for (; left && vleft--; i++, first=0) { 2232 unsigned long lval; 2233 bool neg; 2234 2235 if (write) { 2236 left -= proc_skip_spaces(&p); 2237 2238 if (!left) 2239 break; 2240 err = proc_get_long(&p, &left, &lval, &neg, 2241 proc_wspace_sep, 2242 sizeof(proc_wspace_sep), NULL); 2243 if (err) 2244 break; 2245 if (conv(&neg, &lval, i, 1, data)) { 2246 err = -EINVAL; 2247 break; 2248 } 2249 } else { 2250 if (conv(&neg, &lval, i, 0, data)) { 2251 err = -EINVAL; 2252 break; 2253 } 2254 if (!first) 2255 err = proc_put_char(&buffer, &left, '\t'); 2256 if (err) 2257 break; 2258 err = proc_put_long(&buffer, &left, lval, neg); 2259 if (err) 2260 break; 2261 } 2262 } 2263 2264 if (!write && !first && left && !err) 2265 err = proc_put_char(&buffer, &left, '\n'); 2266 if (write && !err && left) 2267 left -= proc_skip_spaces(&p); 2268 if (write) { 2269 kfree(kbuf); 2270 if (first) 2271 return err ? : -EINVAL; 2272 } 2273 *lenp -= left; 2274 out: 2275 *ppos += *lenp; 2276 return err; 2277 } 2278 2279 static int do_proc_dointvec(struct ctl_table *table, int write, 2280 void __user *buffer, size_t *lenp, loff_t *ppos, 2281 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 2282 int write, void *data), 2283 void *data) 2284 { 2285 return __do_proc_dointvec(table->data, table, write, 2286 buffer, lenp, ppos, conv, data); 2287 } 2288 2289 static int do_proc_douintvec_w(unsigned int *tbl_data, 2290 struct ctl_table *table, 2291 void __user *buffer, 2292 size_t *lenp, loff_t *ppos, 2293 int (*conv)(unsigned long *lvalp, 2294 unsigned int *valp, 2295 int write, void *data), 2296 void *data) 2297 { 2298 unsigned long lval; 2299 int err = 0; 2300 size_t left; 2301 bool neg; 2302 char *kbuf = NULL, *p; 2303 2304 left = *lenp; 2305 2306 if (proc_first_pos_non_zero_ignore(ppos, table)) 2307 goto bail_early; 2308 2309 if (left > PAGE_SIZE - 1) 2310 left = PAGE_SIZE - 1; 2311 2312 p = kbuf = memdup_user_nul(buffer, left); 2313 if (IS_ERR(kbuf)) 2314 return -EINVAL; 2315 2316 left -= proc_skip_spaces(&p); 2317 if (!left) { 2318 err = -EINVAL; 2319 goto out_free; 2320 } 2321 2322 err = proc_get_long(&p, &left, &lval, &neg, 2323 proc_wspace_sep, 2324 sizeof(proc_wspace_sep), NULL); 2325 if (err || neg) { 2326 err = -EINVAL; 2327 goto out_free; 2328 } 2329 2330 if (conv(&lval, tbl_data, 1, data)) { 2331 err = -EINVAL; 2332 goto out_free; 2333 } 2334 2335 if (!err && left) 2336 left -= proc_skip_spaces(&p); 2337 2338 out_free: 2339 kfree(kbuf); 2340 if (err) 2341 return -EINVAL; 2342 2343 return 0; 2344 2345 /* This is in keeping with old __do_proc_dointvec() */ 2346 bail_early: 2347 *ppos += *lenp; 2348 return err; 2349 } 2350 2351 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer, 2352 size_t *lenp, loff_t *ppos, 2353 int (*conv)(unsigned long *lvalp, 2354 unsigned int *valp, 2355 int write, void *data), 2356 void *data) 2357 { 2358 unsigned long lval; 2359 int err = 0; 2360 size_t left; 2361 2362 left = *lenp; 2363 2364 if (conv(&lval, tbl_data, 0, data)) { 2365 err = -EINVAL; 2366 goto out; 2367 } 2368 2369 err = proc_put_long(&buffer, &left, lval, false); 2370 if (err || !left) 2371 goto out; 2372 2373 err = proc_put_char(&buffer, &left, '\n'); 2374 2375 out: 2376 *lenp -= left; 2377 *ppos += *lenp; 2378 2379 return err; 2380 } 2381 2382 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table, 2383 int write, void __user *buffer, 2384 size_t *lenp, loff_t *ppos, 2385 int (*conv)(unsigned long *lvalp, 2386 unsigned int *valp, 2387 int write, void *data), 2388 void *data) 2389 { 2390 unsigned int *i, vleft; 2391 2392 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) { 2393 *lenp = 0; 2394 return 0; 2395 } 2396 2397 i = (unsigned int *) tbl_data; 2398 vleft = table->maxlen / sizeof(*i); 2399 2400 /* 2401 * Arrays are not supported, keep this simple. *Do not* add 2402 * support for them. 2403 */ 2404 if (vleft != 1) { 2405 *lenp = 0; 2406 return -EINVAL; 2407 } 2408 2409 if (!conv) 2410 conv = do_proc_douintvec_conv; 2411 2412 if (write) 2413 return do_proc_douintvec_w(i, table, buffer, lenp, ppos, 2414 conv, data); 2415 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data); 2416 } 2417 2418 static int do_proc_douintvec(struct ctl_table *table, int write, 2419 void __user *buffer, size_t *lenp, loff_t *ppos, 2420 int (*conv)(unsigned long *lvalp, 2421 unsigned int *valp, 2422 int write, void *data), 2423 void *data) 2424 { 2425 return __do_proc_douintvec(table->data, table, write, 2426 buffer, lenp, ppos, conv, data); 2427 } 2428 2429 /** 2430 * proc_dointvec - read a vector of integers 2431 * @table: the sysctl table 2432 * @write: %TRUE if this is a write to the sysctl file 2433 * @buffer: the user buffer 2434 * @lenp: the size of the user buffer 2435 * @ppos: file position 2436 * 2437 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2438 * values from/to the user buffer, treated as an ASCII string. 2439 * 2440 * Returns 0 on success. 2441 */ 2442 int proc_dointvec(struct ctl_table *table, int write, 2443 void __user *buffer, size_t *lenp, loff_t *ppos) 2444 { 2445 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL); 2446 } 2447 2448 /** 2449 * proc_douintvec - read a vector of unsigned integers 2450 * @table: the sysctl table 2451 * @write: %TRUE if this is a write to the sysctl file 2452 * @buffer: the user buffer 2453 * @lenp: the size of the user buffer 2454 * @ppos: file position 2455 * 2456 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer 2457 * values from/to the user buffer, treated as an ASCII string. 2458 * 2459 * Returns 0 on success. 2460 */ 2461 int proc_douintvec(struct ctl_table *table, int write, 2462 void __user *buffer, size_t *lenp, loff_t *ppos) 2463 { 2464 return do_proc_douintvec(table, write, buffer, lenp, ppos, 2465 do_proc_douintvec_conv, NULL); 2466 } 2467 2468 /* 2469 * Taint values can only be increased 2470 * This means we can safely use a temporary. 2471 */ 2472 static int proc_taint(struct ctl_table *table, int write, 2473 void __user *buffer, size_t *lenp, loff_t *ppos) 2474 { 2475 struct ctl_table t; 2476 unsigned long tmptaint = get_taint(); 2477 int err; 2478 2479 if (write && !capable(CAP_SYS_ADMIN)) 2480 return -EPERM; 2481 2482 t = *table; 2483 t.data = &tmptaint; 2484 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos); 2485 if (err < 0) 2486 return err; 2487 2488 if (write) { 2489 /* 2490 * Poor man's atomic or. Not worth adding a primitive 2491 * to everyone's atomic.h for this 2492 */ 2493 int i; 2494 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) { 2495 if ((tmptaint >> i) & 1) 2496 add_taint(i, LOCKDEP_STILL_OK); 2497 } 2498 } 2499 2500 return err; 2501 } 2502 2503 #ifdef CONFIG_PRINTK 2504 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write, 2505 void __user *buffer, size_t *lenp, loff_t *ppos) 2506 { 2507 if (write && !capable(CAP_SYS_ADMIN)) 2508 return -EPERM; 2509 2510 return proc_dointvec_minmax(table, write, buffer, lenp, ppos); 2511 } 2512 #endif 2513 2514 struct do_proc_dointvec_minmax_conv_param { 2515 int *min; 2516 int *max; 2517 }; 2518 2519 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp, 2520 int *valp, 2521 int write, void *data) 2522 { 2523 struct do_proc_dointvec_minmax_conv_param *param = data; 2524 if (write) { 2525 int val = *negp ? -*lvalp : *lvalp; 2526 if ((param->min && *param->min > val) || 2527 (param->max && *param->max < val)) 2528 return -EINVAL; 2529 *valp = val; 2530 } else { 2531 int val = *valp; 2532 if (val < 0) { 2533 *negp = true; 2534 *lvalp = -(unsigned long)val; 2535 } else { 2536 *negp = false; 2537 *lvalp = (unsigned long)val; 2538 } 2539 } 2540 return 0; 2541 } 2542 2543 /** 2544 * proc_dointvec_minmax - read a vector of integers with min/max values 2545 * @table: the sysctl table 2546 * @write: %TRUE if this is a write to the sysctl file 2547 * @buffer: the user buffer 2548 * @lenp: the size of the user buffer 2549 * @ppos: file position 2550 * 2551 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2552 * values from/to the user buffer, treated as an ASCII string. 2553 * 2554 * This routine will ensure the values are within the range specified by 2555 * table->extra1 (min) and table->extra2 (max). 2556 * 2557 * Returns 0 on success. 2558 */ 2559 int proc_dointvec_minmax(struct ctl_table *table, int write, 2560 void __user *buffer, size_t *lenp, loff_t *ppos) 2561 { 2562 struct do_proc_dointvec_minmax_conv_param param = { 2563 .min = (int *) table->extra1, 2564 .max = (int *) table->extra2, 2565 }; 2566 return do_proc_dointvec(table, write, buffer, lenp, ppos, 2567 do_proc_dointvec_minmax_conv, ¶m); 2568 } 2569 2570 struct do_proc_douintvec_minmax_conv_param { 2571 unsigned int *min; 2572 unsigned int *max; 2573 }; 2574 2575 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp, 2576 unsigned int *valp, 2577 int write, void *data) 2578 { 2579 struct do_proc_douintvec_minmax_conv_param *param = data; 2580 2581 if (write) { 2582 unsigned int val = *lvalp; 2583 2584 if ((param->min && *param->min > val) || 2585 (param->max && *param->max < val)) 2586 return -ERANGE; 2587 2588 if (*lvalp > UINT_MAX) 2589 return -EINVAL; 2590 *valp = val; 2591 } else { 2592 unsigned int val = *valp; 2593 *lvalp = (unsigned long) val; 2594 } 2595 2596 return 0; 2597 } 2598 2599 /** 2600 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values 2601 * @table: the sysctl table 2602 * @write: %TRUE if this is a write to the sysctl file 2603 * @buffer: the user buffer 2604 * @lenp: the size of the user buffer 2605 * @ppos: file position 2606 * 2607 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer 2608 * values from/to the user buffer, treated as an ASCII string. Negative 2609 * strings are not allowed. 2610 * 2611 * This routine will ensure the values are within the range specified by 2612 * table->extra1 (min) and table->extra2 (max). There is a final sanity 2613 * check for UINT_MAX to avoid having to support wrap around uses from 2614 * userspace. 2615 * 2616 * Returns 0 on success. 2617 */ 2618 int proc_douintvec_minmax(struct ctl_table *table, int write, 2619 void __user *buffer, size_t *lenp, loff_t *ppos) 2620 { 2621 struct do_proc_douintvec_minmax_conv_param param = { 2622 .min = (unsigned int *) table->extra1, 2623 .max = (unsigned int *) table->extra2, 2624 }; 2625 return do_proc_douintvec(table, write, buffer, lenp, ppos, 2626 do_proc_douintvec_minmax_conv, ¶m); 2627 } 2628 2629 static void validate_coredump_safety(void) 2630 { 2631 #ifdef CONFIG_COREDUMP 2632 if (suid_dumpable == SUID_DUMP_ROOT && 2633 core_pattern[0] != '/' && core_pattern[0] != '|') { 2634 printk(KERN_WARNING 2635 "Unsafe core_pattern used with fs.suid_dumpable=2.\n" 2636 "Pipe handler or fully qualified core dump path required.\n" 2637 "Set kernel.core_pattern before fs.suid_dumpable.\n" 2638 ); 2639 } 2640 #endif 2641 } 2642 2643 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write, 2644 void __user *buffer, size_t *lenp, loff_t *ppos) 2645 { 2646 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos); 2647 if (!error) 2648 validate_coredump_safety(); 2649 return error; 2650 } 2651 2652 #ifdef CONFIG_COREDUMP 2653 static int proc_dostring_coredump(struct ctl_table *table, int write, 2654 void __user *buffer, size_t *lenp, loff_t *ppos) 2655 { 2656 int error = proc_dostring(table, write, buffer, lenp, ppos); 2657 if (!error) 2658 validate_coredump_safety(); 2659 return error; 2660 } 2661 #endif 2662 2663 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write, 2664 void __user *buffer, 2665 size_t *lenp, loff_t *ppos, 2666 unsigned long convmul, 2667 unsigned long convdiv) 2668 { 2669 unsigned long *i, *min, *max; 2670 int vleft, first = 1, err = 0; 2671 size_t left; 2672 char *kbuf = NULL, *p; 2673 2674 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) { 2675 *lenp = 0; 2676 return 0; 2677 } 2678 2679 i = (unsigned long *) data; 2680 min = (unsigned long *) table->extra1; 2681 max = (unsigned long *) table->extra2; 2682 vleft = table->maxlen / sizeof(unsigned long); 2683 left = *lenp; 2684 2685 if (write) { 2686 if (proc_first_pos_non_zero_ignore(ppos, table)) 2687 goto out; 2688 2689 if (left > PAGE_SIZE - 1) 2690 left = PAGE_SIZE - 1; 2691 p = kbuf = memdup_user_nul(buffer, left); 2692 if (IS_ERR(kbuf)) 2693 return PTR_ERR(kbuf); 2694 } 2695 2696 for (; left && vleft--; i++, first = 0) { 2697 unsigned long val; 2698 2699 if (write) { 2700 bool neg; 2701 2702 left -= proc_skip_spaces(&p); 2703 2704 err = proc_get_long(&p, &left, &val, &neg, 2705 proc_wspace_sep, 2706 sizeof(proc_wspace_sep), NULL); 2707 if (err) 2708 break; 2709 if (neg) 2710 continue; 2711 val = convmul * val / convdiv; 2712 if ((min && val < *min) || (max && val > *max)) 2713 continue; 2714 *i = val; 2715 } else { 2716 val = convdiv * (*i) / convmul; 2717 if (!first) { 2718 err = proc_put_char(&buffer, &left, '\t'); 2719 if (err) 2720 break; 2721 } 2722 err = proc_put_long(&buffer, &left, val, false); 2723 if (err) 2724 break; 2725 } 2726 } 2727 2728 if (!write && !first && left && !err) 2729 err = proc_put_char(&buffer, &left, '\n'); 2730 if (write && !err) 2731 left -= proc_skip_spaces(&p); 2732 if (write) { 2733 kfree(kbuf); 2734 if (first) 2735 return err ? : -EINVAL; 2736 } 2737 *lenp -= left; 2738 out: 2739 *ppos += *lenp; 2740 return err; 2741 } 2742 2743 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write, 2744 void __user *buffer, 2745 size_t *lenp, loff_t *ppos, 2746 unsigned long convmul, 2747 unsigned long convdiv) 2748 { 2749 return __do_proc_doulongvec_minmax(table->data, table, write, 2750 buffer, lenp, ppos, convmul, convdiv); 2751 } 2752 2753 /** 2754 * proc_doulongvec_minmax - read a vector of long integers with min/max values 2755 * @table: the sysctl table 2756 * @write: %TRUE if this is a write to the sysctl file 2757 * @buffer: the user buffer 2758 * @lenp: the size of the user buffer 2759 * @ppos: file position 2760 * 2761 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 2762 * values from/to the user buffer, treated as an ASCII string. 2763 * 2764 * This routine will ensure the values are within the range specified by 2765 * table->extra1 (min) and table->extra2 (max). 2766 * 2767 * Returns 0 on success. 2768 */ 2769 int proc_doulongvec_minmax(struct ctl_table *table, int write, 2770 void __user *buffer, size_t *lenp, loff_t *ppos) 2771 { 2772 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l); 2773 } 2774 2775 /** 2776 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values 2777 * @table: the sysctl table 2778 * @write: %TRUE if this is a write to the sysctl file 2779 * @buffer: the user buffer 2780 * @lenp: the size of the user buffer 2781 * @ppos: file position 2782 * 2783 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 2784 * values from/to the user buffer, treated as an ASCII string. The values 2785 * are treated as milliseconds, and converted to jiffies when they are stored. 2786 * 2787 * This routine will ensure the values are within the range specified by 2788 * table->extra1 (min) and table->extra2 (max). 2789 * 2790 * Returns 0 on success. 2791 */ 2792 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 2793 void __user *buffer, 2794 size_t *lenp, loff_t *ppos) 2795 { 2796 return do_proc_doulongvec_minmax(table, write, buffer, 2797 lenp, ppos, HZ, 1000l); 2798 } 2799 2800 2801 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp, 2802 int *valp, 2803 int write, void *data) 2804 { 2805 if (write) { 2806 if (*lvalp > INT_MAX / HZ) 2807 return 1; 2808 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ); 2809 } else { 2810 int val = *valp; 2811 unsigned long lval; 2812 if (val < 0) { 2813 *negp = true; 2814 lval = -(unsigned long)val; 2815 } else { 2816 *negp = false; 2817 lval = (unsigned long)val; 2818 } 2819 *lvalp = lval / HZ; 2820 } 2821 return 0; 2822 } 2823 2824 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp, 2825 int *valp, 2826 int write, void *data) 2827 { 2828 if (write) { 2829 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ) 2830 return 1; 2831 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp); 2832 } else { 2833 int val = *valp; 2834 unsigned long lval; 2835 if (val < 0) { 2836 *negp = true; 2837 lval = -(unsigned long)val; 2838 } else { 2839 *negp = false; 2840 lval = (unsigned long)val; 2841 } 2842 *lvalp = jiffies_to_clock_t(lval); 2843 } 2844 return 0; 2845 } 2846 2847 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp, 2848 int *valp, 2849 int write, void *data) 2850 { 2851 if (write) { 2852 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp); 2853 2854 if (jif > INT_MAX) 2855 return 1; 2856 *valp = (int)jif; 2857 } else { 2858 int val = *valp; 2859 unsigned long lval; 2860 if (val < 0) { 2861 *negp = true; 2862 lval = -(unsigned long)val; 2863 } else { 2864 *negp = false; 2865 lval = (unsigned long)val; 2866 } 2867 *lvalp = jiffies_to_msecs(lval); 2868 } 2869 return 0; 2870 } 2871 2872 /** 2873 * proc_dointvec_jiffies - read a vector of integers as seconds 2874 * @table: the sysctl table 2875 * @write: %TRUE if this is a write to the sysctl file 2876 * @buffer: the user buffer 2877 * @lenp: the size of the user buffer 2878 * @ppos: file position 2879 * 2880 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2881 * values from/to the user buffer, treated as an ASCII string. 2882 * The values read are assumed to be in seconds, and are converted into 2883 * jiffies. 2884 * 2885 * Returns 0 on success. 2886 */ 2887 int proc_dointvec_jiffies(struct ctl_table *table, int write, 2888 void __user *buffer, size_t *lenp, loff_t *ppos) 2889 { 2890 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2891 do_proc_dointvec_jiffies_conv,NULL); 2892 } 2893 2894 /** 2895 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds 2896 * @table: the sysctl table 2897 * @write: %TRUE if this is a write to the sysctl file 2898 * @buffer: the user buffer 2899 * @lenp: the size of the user buffer 2900 * @ppos: pointer to the file position 2901 * 2902 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2903 * values from/to the user buffer, treated as an ASCII string. 2904 * The values read are assumed to be in 1/USER_HZ seconds, and 2905 * are converted into jiffies. 2906 * 2907 * Returns 0 on success. 2908 */ 2909 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 2910 void __user *buffer, size_t *lenp, loff_t *ppos) 2911 { 2912 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2913 do_proc_dointvec_userhz_jiffies_conv,NULL); 2914 } 2915 2916 /** 2917 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds 2918 * @table: the sysctl table 2919 * @write: %TRUE if this is a write to the sysctl file 2920 * @buffer: the user buffer 2921 * @lenp: the size of the user buffer 2922 * @ppos: file position 2923 * @ppos: the current position in the file 2924 * 2925 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2926 * values from/to the user buffer, treated as an ASCII string. 2927 * The values read are assumed to be in 1/1000 seconds, and 2928 * are converted into jiffies. 2929 * 2930 * Returns 0 on success. 2931 */ 2932 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, 2933 void __user *buffer, size_t *lenp, loff_t *ppos) 2934 { 2935 return do_proc_dointvec(table, write, buffer, lenp, ppos, 2936 do_proc_dointvec_ms_jiffies_conv, NULL); 2937 } 2938 2939 static int proc_do_cad_pid(struct ctl_table *table, int write, 2940 void __user *buffer, size_t *lenp, loff_t *ppos) 2941 { 2942 struct pid *new_pid; 2943 pid_t tmp; 2944 int r; 2945 2946 tmp = pid_vnr(cad_pid); 2947 2948 r = __do_proc_dointvec(&tmp, table, write, buffer, 2949 lenp, ppos, NULL, NULL); 2950 if (r || !write) 2951 return r; 2952 2953 new_pid = find_get_pid(tmp); 2954 if (!new_pid) 2955 return -ESRCH; 2956 2957 put_pid(xchg(&cad_pid, new_pid)); 2958 return 0; 2959 } 2960 2961 /** 2962 * proc_do_large_bitmap - read/write from/to a large bitmap 2963 * @table: the sysctl table 2964 * @write: %TRUE if this is a write to the sysctl file 2965 * @buffer: the user buffer 2966 * @lenp: the size of the user buffer 2967 * @ppos: file position 2968 * 2969 * The bitmap is stored at table->data and the bitmap length (in bits) 2970 * in table->maxlen. 2971 * 2972 * We use a range comma separated format (e.g. 1,3-4,10-10) so that 2973 * large bitmaps may be represented in a compact manner. Writing into 2974 * the file will clear the bitmap then update it with the given input. 2975 * 2976 * Returns 0 on success. 2977 */ 2978 int proc_do_large_bitmap(struct ctl_table *table, int write, 2979 void __user *buffer, size_t *lenp, loff_t *ppos) 2980 { 2981 int err = 0; 2982 bool first = 1; 2983 size_t left = *lenp; 2984 unsigned long bitmap_len = table->maxlen; 2985 unsigned long *bitmap = *(unsigned long **) table->data; 2986 unsigned long *tmp_bitmap = NULL; 2987 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c; 2988 2989 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) { 2990 *lenp = 0; 2991 return 0; 2992 } 2993 2994 if (write) { 2995 char *kbuf, *p; 2996 2997 if (left > PAGE_SIZE - 1) 2998 left = PAGE_SIZE - 1; 2999 3000 p = kbuf = memdup_user_nul(buffer, left); 3001 if (IS_ERR(kbuf)) 3002 return PTR_ERR(kbuf); 3003 3004 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long), 3005 GFP_KERNEL); 3006 if (!tmp_bitmap) { 3007 kfree(kbuf); 3008 return -ENOMEM; 3009 } 3010 proc_skip_char(&p, &left, '\n'); 3011 while (!err && left) { 3012 unsigned long val_a, val_b; 3013 bool neg; 3014 3015 err = proc_get_long(&p, &left, &val_a, &neg, tr_a, 3016 sizeof(tr_a), &c); 3017 if (err) 3018 break; 3019 if (val_a >= bitmap_len || neg) { 3020 err = -EINVAL; 3021 break; 3022 } 3023 3024 val_b = val_a; 3025 if (left) { 3026 p++; 3027 left--; 3028 } 3029 3030 if (c == '-') { 3031 err = proc_get_long(&p, &left, &val_b, 3032 &neg, tr_b, sizeof(tr_b), 3033 &c); 3034 if (err) 3035 break; 3036 if (val_b >= bitmap_len || neg || 3037 val_a > val_b) { 3038 err = -EINVAL; 3039 break; 3040 } 3041 if (left) { 3042 p++; 3043 left--; 3044 } 3045 } 3046 3047 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1); 3048 first = 0; 3049 proc_skip_char(&p, &left, '\n'); 3050 } 3051 kfree(kbuf); 3052 } else { 3053 unsigned long bit_a, bit_b = 0; 3054 3055 while (left) { 3056 bit_a = find_next_bit(bitmap, bitmap_len, bit_b); 3057 if (bit_a >= bitmap_len) 3058 break; 3059 bit_b = find_next_zero_bit(bitmap, bitmap_len, 3060 bit_a + 1) - 1; 3061 3062 if (!first) { 3063 err = proc_put_char(&buffer, &left, ','); 3064 if (err) 3065 break; 3066 } 3067 err = proc_put_long(&buffer, &left, bit_a, false); 3068 if (err) 3069 break; 3070 if (bit_a != bit_b) { 3071 err = proc_put_char(&buffer, &left, '-'); 3072 if (err) 3073 break; 3074 err = proc_put_long(&buffer, &left, bit_b, false); 3075 if (err) 3076 break; 3077 } 3078 3079 first = 0; bit_b++; 3080 } 3081 if (!err) 3082 err = proc_put_char(&buffer, &left, '\n'); 3083 } 3084 3085 if (!err) { 3086 if (write) { 3087 if (*ppos) 3088 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len); 3089 else 3090 bitmap_copy(bitmap, tmp_bitmap, bitmap_len); 3091 } 3092 kfree(tmp_bitmap); 3093 *lenp -= left; 3094 *ppos += *lenp; 3095 return 0; 3096 } else { 3097 kfree(tmp_bitmap); 3098 return err; 3099 } 3100 } 3101 3102 #else /* CONFIG_PROC_SYSCTL */ 3103 3104 int proc_dostring(struct ctl_table *table, int write, 3105 void __user *buffer, size_t *lenp, loff_t *ppos) 3106 { 3107 return -ENOSYS; 3108 } 3109 3110 int proc_dointvec(struct ctl_table *table, int write, 3111 void __user *buffer, size_t *lenp, loff_t *ppos) 3112 { 3113 return -ENOSYS; 3114 } 3115 3116 int proc_douintvec(struct ctl_table *table, int write, 3117 void __user *buffer, size_t *lenp, loff_t *ppos) 3118 { 3119 return -ENOSYS; 3120 } 3121 3122 int proc_dointvec_minmax(struct ctl_table *table, int write, 3123 void __user *buffer, size_t *lenp, loff_t *ppos) 3124 { 3125 return -ENOSYS; 3126 } 3127 3128 int proc_douintvec_minmax(struct ctl_table *table, int write, 3129 void __user *buffer, size_t *lenp, loff_t *ppos) 3130 { 3131 return -ENOSYS; 3132 } 3133 3134 int proc_dointvec_jiffies(struct ctl_table *table, int write, 3135 void __user *buffer, size_t *lenp, loff_t *ppos) 3136 { 3137 return -ENOSYS; 3138 } 3139 3140 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 3141 void __user *buffer, size_t *lenp, loff_t *ppos) 3142 { 3143 return -ENOSYS; 3144 } 3145 3146 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, 3147 void __user *buffer, size_t *lenp, loff_t *ppos) 3148 { 3149 return -ENOSYS; 3150 } 3151 3152 int proc_doulongvec_minmax(struct ctl_table *table, int write, 3153 void __user *buffer, size_t *lenp, loff_t *ppos) 3154 { 3155 return -ENOSYS; 3156 } 3157 3158 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 3159 void __user *buffer, 3160 size_t *lenp, loff_t *ppos) 3161 { 3162 return -ENOSYS; 3163 } 3164 3165 3166 #endif /* CONFIG_PROC_SYSCTL */ 3167 3168 /* 3169 * No sense putting this after each symbol definition, twice, 3170 * exception granted :-) 3171 */ 3172 EXPORT_SYMBOL(proc_dointvec); 3173 EXPORT_SYMBOL(proc_douintvec); 3174 EXPORT_SYMBOL(proc_dointvec_jiffies); 3175 EXPORT_SYMBOL(proc_dointvec_minmax); 3176 EXPORT_SYMBOL_GPL(proc_douintvec_minmax); 3177 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies); 3178 EXPORT_SYMBOL(proc_dointvec_ms_jiffies); 3179 EXPORT_SYMBOL(proc_dostring); 3180 EXPORT_SYMBOL(proc_doulongvec_minmax); 3181 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax); 3182