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 #include <linux/pipe_fs_i.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 { 1178 .procname = "panic_on_warn", 1179 .data = &panic_on_warn, 1180 .maxlen = sizeof(int), 1181 .mode = 0644, 1182 .proc_handler = proc_dointvec_minmax, 1183 .extra1 = &zero, 1184 .extra2 = &one, 1185 }, 1186 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) 1187 { 1188 .procname = "timer_migration", 1189 .data = &sysctl_timer_migration, 1190 .maxlen = sizeof(unsigned int), 1191 .mode = 0644, 1192 .proc_handler = timer_migration_handler, 1193 .extra1 = &zero, 1194 .extra2 = &one, 1195 }, 1196 #endif 1197 #ifdef CONFIG_BPF_SYSCALL 1198 { 1199 .procname = "unprivileged_bpf_disabled", 1200 .data = &sysctl_unprivileged_bpf_disabled, 1201 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled), 1202 .mode = 0644, 1203 /* only handle a transition from default "0" to "1" */ 1204 .proc_handler = proc_dointvec_minmax, 1205 .extra1 = &one, 1206 .extra2 = &one, 1207 }, 1208 #endif 1209 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU) 1210 { 1211 .procname = "panic_on_rcu_stall", 1212 .data = &sysctl_panic_on_rcu_stall, 1213 .maxlen = sizeof(sysctl_panic_on_rcu_stall), 1214 .mode = 0644, 1215 .proc_handler = proc_dointvec_minmax, 1216 .extra1 = &zero, 1217 .extra2 = &one, 1218 }, 1219 #endif 1220 { } 1221 }; 1222 1223 static struct ctl_table vm_table[] = { 1224 { 1225 .procname = "overcommit_memory", 1226 .data = &sysctl_overcommit_memory, 1227 .maxlen = sizeof(sysctl_overcommit_memory), 1228 .mode = 0644, 1229 .proc_handler = proc_dointvec_minmax, 1230 .extra1 = &zero, 1231 .extra2 = &two, 1232 }, 1233 { 1234 .procname = "panic_on_oom", 1235 .data = &sysctl_panic_on_oom, 1236 .maxlen = sizeof(sysctl_panic_on_oom), 1237 .mode = 0644, 1238 .proc_handler = proc_dointvec_minmax, 1239 .extra1 = &zero, 1240 .extra2 = &two, 1241 }, 1242 { 1243 .procname = "oom_kill_allocating_task", 1244 .data = &sysctl_oom_kill_allocating_task, 1245 .maxlen = sizeof(sysctl_oom_kill_allocating_task), 1246 .mode = 0644, 1247 .proc_handler = proc_dointvec, 1248 }, 1249 { 1250 .procname = "oom_dump_tasks", 1251 .data = &sysctl_oom_dump_tasks, 1252 .maxlen = sizeof(sysctl_oom_dump_tasks), 1253 .mode = 0644, 1254 .proc_handler = proc_dointvec, 1255 }, 1256 { 1257 .procname = "overcommit_ratio", 1258 .data = &sysctl_overcommit_ratio, 1259 .maxlen = sizeof(sysctl_overcommit_ratio), 1260 .mode = 0644, 1261 .proc_handler = overcommit_ratio_handler, 1262 }, 1263 { 1264 .procname = "overcommit_kbytes", 1265 .data = &sysctl_overcommit_kbytes, 1266 .maxlen = sizeof(sysctl_overcommit_kbytes), 1267 .mode = 0644, 1268 .proc_handler = overcommit_kbytes_handler, 1269 }, 1270 { 1271 .procname = "page-cluster", 1272 .data = &page_cluster, 1273 .maxlen = sizeof(int), 1274 .mode = 0644, 1275 .proc_handler = proc_dointvec_minmax, 1276 .extra1 = &zero, 1277 }, 1278 { 1279 .procname = "dirty_background_ratio", 1280 .data = &dirty_background_ratio, 1281 .maxlen = sizeof(dirty_background_ratio), 1282 .mode = 0644, 1283 .proc_handler = dirty_background_ratio_handler, 1284 .extra1 = &zero, 1285 .extra2 = &one_hundred, 1286 }, 1287 { 1288 .procname = "dirty_background_bytes", 1289 .data = &dirty_background_bytes, 1290 .maxlen = sizeof(dirty_background_bytes), 1291 .mode = 0644, 1292 .proc_handler = dirty_background_bytes_handler, 1293 .extra1 = &one_ul, 1294 }, 1295 { 1296 .procname = "dirty_ratio", 1297 .data = &vm_dirty_ratio, 1298 .maxlen = sizeof(vm_dirty_ratio), 1299 .mode = 0644, 1300 .proc_handler = dirty_ratio_handler, 1301 .extra1 = &zero, 1302 .extra2 = &one_hundred, 1303 }, 1304 { 1305 .procname = "dirty_bytes", 1306 .data = &vm_dirty_bytes, 1307 .maxlen = sizeof(vm_dirty_bytes), 1308 .mode = 0644, 1309 .proc_handler = dirty_bytes_handler, 1310 .extra1 = &dirty_bytes_min, 1311 }, 1312 { 1313 .procname = "dirty_writeback_centisecs", 1314 .data = &dirty_writeback_interval, 1315 .maxlen = sizeof(dirty_writeback_interval), 1316 .mode = 0644, 1317 .proc_handler = dirty_writeback_centisecs_handler, 1318 }, 1319 { 1320 .procname = "dirty_expire_centisecs", 1321 .data = &dirty_expire_interval, 1322 .maxlen = sizeof(dirty_expire_interval), 1323 .mode = 0644, 1324 .proc_handler = proc_dointvec_minmax, 1325 .extra1 = &zero, 1326 }, 1327 { 1328 .procname = "dirtytime_expire_seconds", 1329 .data = &dirtytime_expire_interval, 1330 .maxlen = sizeof(dirty_expire_interval), 1331 .mode = 0644, 1332 .proc_handler = dirtytime_interval_handler, 1333 .extra1 = &zero, 1334 }, 1335 { 1336 .procname = "swappiness", 1337 .data = &vm_swappiness, 1338 .maxlen = sizeof(vm_swappiness), 1339 .mode = 0644, 1340 .proc_handler = proc_dointvec_minmax, 1341 .extra1 = &zero, 1342 .extra2 = &one_hundred, 1343 }, 1344 #ifdef CONFIG_HUGETLB_PAGE 1345 { 1346 .procname = "nr_hugepages", 1347 .data = NULL, 1348 .maxlen = sizeof(unsigned long), 1349 .mode = 0644, 1350 .proc_handler = hugetlb_sysctl_handler, 1351 }, 1352 #ifdef CONFIG_NUMA 1353 { 1354 .procname = "nr_hugepages_mempolicy", 1355 .data = NULL, 1356 .maxlen = sizeof(unsigned long), 1357 .mode = 0644, 1358 .proc_handler = &hugetlb_mempolicy_sysctl_handler, 1359 }, 1360 { 1361 .procname = "numa_stat", 1362 .data = &sysctl_vm_numa_stat, 1363 .maxlen = sizeof(int), 1364 .mode = 0644, 1365 .proc_handler = sysctl_vm_numa_stat_handler, 1366 .extra1 = &zero, 1367 .extra2 = &one, 1368 }, 1369 #endif 1370 { 1371 .procname = "hugetlb_shm_group", 1372 .data = &sysctl_hugetlb_shm_group, 1373 .maxlen = sizeof(gid_t), 1374 .mode = 0644, 1375 .proc_handler = proc_dointvec, 1376 }, 1377 { 1378 .procname = "nr_overcommit_hugepages", 1379 .data = NULL, 1380 .maxlen = sizeof(unsigned long), 1381 .mode = 0644, 1382 .proc_handler = hugetlb_overcommit_handler, 1383 }, 1384 #endif 1385 { 1386 .procname = "lowmem_reserve_ratio", 1387 .data = &sysctl_lowmem_reserve_ratio, 1388 .maxlen = sizeof(sysctl_lowmem_reserve_ratio), 1389 .mode = 0644, 1390 .proc_handler = lowmem_reserve_ratio_sysctl_handler, 1391 }, 1392 { 1393 .procname = "drop_caches", 1394 .data = &sysctl_drop_caches, 1395 .maxlen = sizeof(int), 1396 .mode = 0644, 1397 .proc_handler = drop_caches_sysctl_handler, 1398 .extra1 = &one, 1399 .extra2 = &four, 1400 }, 1401 #ifdef CONFIG_COMPACTION 1402 { 1403 .procname = "compact_memory", 1404 .data = &sysctl_compact_memory, 1405 .maxlen = sizeof(int), 1406 .mode = 0200, 1407 .proc_handler = sysctl_compaction_handler, 1408 }, 1409 { 1410 .procname = "extfrag_threshold", 1411 .data = &sysctl_extfrag_threshold, 1412 .maxlen = sizeof(int), 1413 .mode = 0644, 1414 .proc_handler = sysctl_extfrag_handler, 1415 .extra1 = &min_extfrag_threshold, 1416 .extra2 = &max_extfrag_threshold, 1417 }, 1418 { 1419 .procname = "compact_unevictable_allowed", 1420 .data = &sysctl_compact_unevictable_allowed, 1421 .maxlen = sizeof(int), 1422 .mode = 0644, 1423 .proc_handler = proc_dointvec, 1424 .extra1 = &zero, 1425 .extra2 = &one, 1426 }, 1427 1428 #endif /* CONFIG_COMPACTION */ 1429 { 1430 .procname = "min_free_kbytes", 1431 .data = &min_free_kbytes, 1432 .maxlen = sizeof(min_free_kbytes), 1433 .mode = 0644, 1434 .proc_handler = min_free_kbytes_sysctl_handler, 1435 .extra1 = &zero, 1436 }, 1437 { 1438 .procname = "watermark_scale_factor", 1439 .data = &watermark_scale_factor, 1440 .maxlen = sizeof(watermark_scale_factor), 1441 .mode = 0644, 1442 .proc_handler = watermark_scale_factor_sysctl_handler, 1443 .extra1 = &one, 1444 .extra2 = &one_thousand, 1445 }, 1446 { 1447 .procname = "percpu_pagelist_fraction", 1448 .data = &percpu_pagelist_fraction, 1449 .maxlen = sizeof(percpu_pagelist_fraction), 1450 .mode = 0644, 1451 .proc_handler = percpu_pagelist_fraction_sysctl_handler, 1452 .extra1 = &zero, 1453 }, 1454 #ifdef CONFIG_MMU 1455 { 1456 .procname = "max_map_count", 1457 .data = &sysctl_max_map_count, 1458 .maxlen = sizeof(sysctl_max_map_count), 1459 .mode = 0644, 1460 .proc_handler = proc_dointvec_minmax, 1461 .extra1 = &zero, 1462 }, 1463 #else 1464 { 1465 .procname = "nr_trim_pages", 1466 .data = &sysctl_nr_trim_pages, 1467 .maxlen = sizeof(sysctl_nr_trim_pages), 1468 .mode = 0644, 1469 .proc_handler = proc_dointvec_minmax, 1470 .extra1 = &zero, 1471 }, 1472 #endif 1473 { 1474 .procname = "laptop_mode", 1475 .data = &laptop_mode, 1476 .maxlen = sizeof(laptop_mode), 1477 .mode = 0644, 1478 .proc_handler = proc_dointvec_jiffies, 1479 }, 1480 { 1481 .procname = "block_dump", 1482 .data = &block_dump, 1483 .maxlen = sizeof(block_dump), 1484 .mode = 0644, 1485 .proc_handler = proc_dointvec, 1486 .extra1 = &zero, 1487 }, 1488 { 1489 .procname = "vfs_cache_pressure", 1490 .data = &sysctl_vfs_cache_pressure, 1491 .maxlen = sizeof(sysctl_vfs_cache_pressure), 1492 .mode = 0644, 1493 .proc_handler = proc_dointvec, 1494 .extra1 = &zero, 1495 }, 1496 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT 1497 { 1498 .procname = "legacy_va_layout", 1499 .data = &sysctl_legacy_va_layout, 1500 .maxlen = sizeof(sysctl_legacy_va_layout), 1501 .mode = 0644, 1502 .proc_handler = proc_dointvec, 1503 .extra1 = &zero, 1504 }, 1505 #endif 1506 #ifdef CONFIG_NUMA 1507 { 1508 .procname = "zone_reclaim_mode", 1509 .data = &node_reclaim_mode, 1510 .maxlen = sizeof(node_reclaim_mode), 1511 .mode = 0644, 1512 .proc_handler = proc_dointvec, 1513 .extra1 = &zero, 1514 }, 1515 { 1516 .procname = "min_unmapped_ratio", 1517 .data = &sysctl_min_unmapped_ratio, 1518 .maxlen = sizeof(sysctl_min_unmapped_ratio), 1519 .mode = 0644, 1520 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler, 1521 .extra1 = &zero, 1522 .extra2 = &one_hundred, 1523 }, 1524 { 1525 .procname = "min_slab_ratio", 1526 .data = &sysctl_min_slab_ratio, 1527 .maxlen = sizeof(sysctl_min_slab_ratio), 1528 .mode = 0644, 1529 .proc_handler = sysctl_min_slab_ratio_sysctl_handler, 1530 .extra1 = &zero, 1531 .extra2 = &one_hundred, 1532 }, 1533 #endif 1534 #ifdef CONFIG_SMP 1535 { 1536 .procname = "stat_interval", 1537 .data = &sysctl_stat_interval, 1538 .maxlen = sizeof(sysctl_stat_interval), 1539 .mode = 0644, 1540 .proc_handler = proc_dointvec_jiffies, 1541 }, 1542 { 1543 .procname = "stat_refresh", 1544 .data = NULL, 1545 .maxlen = 0, 1546 .mode = 0600, 1547 .proc_handler = vmstat_refresh, 1548 }, 1549 #endif 1550 #ifdef CONFIG_MMU 1551 { 1552 .procname = "mmap_min_addr", 1553 .data = &dac_mmap_min_addr, 1554 .maxlen = sizeof(unsigned long), 1555 .mode = 0644, 1556 .proc_handler = mmap_min_addr_handler, 1557 }, 1558 #endif 1559 #ifdef CONFIG_NUMA 1560 { 1561 .procname = "numa_zonelist_order", 1562 .data = &numa_zonelist_order, 1563 .maxlen = NUMA_ZONELIST_ORDER_LEN, 1564 .mode = 0644, 1565 .proc_handler = numa_zonelist_order_handler, 1566 }, 1567 #endif 1568 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \ 1569 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL)) 1570 { 1571 .procname = "vdso_enabled", 1572 #ifdef CONFIG_X86_32 1573 .data = &vdso32_enabled, 1574 .maxlen = sizeof(vdso32_enabled), 1575 #else 1576 .data = &vdso_enabled, 1577 .maxlen = sizeof(vdso_enabled), 1578 #endif 1579 .mode = 0644, 1580 .proc_handler = proc_dointvec, 1581 .extra1 = &zero, 1582 }, 1583 #endif 1584 #ifdef CONFIG_HIGHMEM 1585 { 1586 .procname = "highmem_is_dirtyable", 1587 .data = &vm_highmem_is_dirtyable, 1588 .maxlen = sizeof(vm_highmem_is_dirtyable), 1589 .mode = 0644, 1590 .proc_handler = proc_dointvec_minmax, 1591 .extra1 = &zero, 1592 .extra2 = &one, 1593 }, 1594 #endif 1595 #ifdef CONFIG_MEMORY_FAILURE 1596 { 1597 .procname = "memory_failure_early_kill", 1598 .data = &sysctl_memory_failure_early_kill, 1599 .maxlen = sizeof(sysctl_memory_failure_early_kill), 1600 .mode = 0644, 1601 .proc_handler = proc_dointvec_minmax, 1602 .extra1 = &zero, 1603 .extra2 = &one, 1604 }, 1605 { 1606 .procname = "memory_failure_recovery", 1607 .data = &sysctl_memory_failure_recovery, 1608 .maxlen = sizeof(sysctl_memory_failure_recovery), 1609 .mode = 0644, 1610 .proc_handler = proc_dointvec_minmax, 1611 .extra1 = &zero, 1612 .extra2 = &one, 1613 }, 1614 #endif 1615 { 1616 .procname = "user_reserve_kbytes", 1617 .data = &sysctl_user_reserve_kbytes, 1618 .maxlen = sizeof(sysctl_user_reserve_kbytes), 1619 .mode = 0644, 1620 .proc_handler = proc_doulongvec_minmax, 1621 }, 1622 { 1623 .procname = "admin_reserve_kbytes", 1624 .data = &sysctl_admin_reserve_kbytes, 1625 .maxlen = sizeof(sysctl_admin_reserve_kbytes), 1626 .mode = 0644, 1627 .proc_handler = proc_doulongvec_minmax, 1628 }, 1629 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS 1630 { 1631 .procname = "mmap_rnd_bits", 1632 .data = &mmap_rnd_bits, 1633 .maxlen = sizeof(mmap_rnd_bits), 1634 .mode = 0600, 1635 .proc_handler = proc_dointvec_minmax, 1636 .extra1 = (void *)&mmap_rnd_bits_min, 1637 .extra2 = (void *)&mmap_rnd_bits_max, 1638 }, 1639 #endif 1640 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS 1641 { 1642 .procname = "mmap_rnd_compat_bits", 1643 .data = &mmap_rnd_compat_bits, 1644 .maxlen = sizeof(mmap_rnd_compat_bits), 1645 .mode = 0600, 1646 .proc_handler = proc_dointvec_minmax, 1647 .extra1 = (void *)&mmap_rnd_compat_bits_min, 1648 .extra2 = (void *)&mmap_rnd_compat_bits_max, 1649 }, 1650 #endif 1651 { } 1652 }; 1653 1654 static struct ctl_table fs_table[] = { 1655 { 1656 .procname = "inode-nr", 1657 .data = &inodes_stat, 1658 .maxlen = 2*sizeof(long), 1659 .mode = 0444, 1660 .proc_handler = proc_nr_inodes, 1661 }, 1662 { 1663 .procname = "inode-state", 1664 .data = &inodes_stat, 1665 .maxlen = 7*sizeof(long), 1666 .mode = 0444, 1667 .proc_handler = proc_nr_inodes, 1668 }, 1669 { 1670 .procname = "file-nr", 1671 .data = &files_stat, 1672 .maxlen = sizeof(files_stat), 1673 .mode = 0444, 1674 .proc_handler = proc_nr_files, 1675 }, 1676 { 1677 .procname = "file-max", 1678 .data = &files_stat.max_files, 1679 .maxlen = sizeof(files_stat.max_files), 1680 .mode = 0644, 1681 .proc_handler = proc_doulongvec_minmax, 1682 }, 1683 { 1684 .procname = "nr_open", 1685 .data = &sysctl_nr_open, 1686 .maxlen = sizeof(unsigned int), 1687 .mode = 0644, 1688 .proc_handler = proc_dointvec_minmax, 1689 .extra1 = &sysctl_nr_open_min, 1690 .extra2 = &sysctl_nr_open_max, 1691 }, 1692 { 1693 .procname = "dentry-state", 1694 .data = &dentry_stat, 1695 .maxlen = 6*sizeof(long), 1696 .mode = 0444, 1697 .proc_handler = proc_nr_dentry, 1698 }, 1699 { 1700 .procname = "overflowuid", 1701 .data = &fs_overflowuid, 1702 .maxlen = sizeof(int), 1703 .mode = 0644, 1704 .proc_handler = proc_dointvec_minmax, 1705 .extra1 = &minolduid, 1706 .extra2 = &maxolduid, 1707 }, 1708 { 1709 .procname = "overflowgid", 1710 .data = &fs_overflowgid, 1711 .maxlen = sizeof(int), 1712 .mode = 0644, 1713 .proc_handler = proc_dointvec_minmax, 1714 .extra1 = &minolduid, 1715 .extra2 = &maxolduid, 1716 }, 1717 #ifdef CONFIG_FILE_LOCKING 1718 { 1719 .procname = "leases-enable", 1720 .data = &leases_enable, 1721 .maxlen = sizeof(int), 1722 .mode = 0644, 1723 .proc_handler = proc_dointvec, 1724 }, 1725 #endif 1726 #ifdef CONFIG_DNOTIFY 1727 { 1728 .procname = "dir-notify-enable", 1729 .data = &dir_notify_enable, 1730 .maxlen = sizeof(int), 1731 .mode = 0644, 1732 .proc_handler = proc_dointvec, 1733 }, 1734 #endif 1735 #ifdef CONFIG_MMU 1736 #ifdef CONFIG_FILE_LOCKING 1737 { 1738 .procname = "lease-break-time", 1739 .data = &lease_break_time, 1740 .maxlen = sizeof(int), 1741 .mode = 0644, 1742 .proc_handler = proc_dointvec, 1743 }, 1744 #endif 1745 #ifdef CONFIG_AIO 1746 { 1747 .procname = "aio-nr", 1748 .data = &aio_nr, 1749 .maxlen = sizeof(aio_nr), 1750 .mode = 0444, 1751 .proc_handler = proc_doulongvec_minmax, 1752 }, 1753 { 1754 .procname = "aio-max-nr", 1755 .data = &aio_max_nr, 1756 .maxlen = sizeof(aio_max_nr), 1757 .mode = 0644, 1758 .proc_handler = proc_doulongvec_minmax, 1759 }, 1760 #endif /* CONFIG_AIO */ 1761 #ifdef CONFIG_INOTIFY_USER 1762 { 1763 .procname = "inotify", 1764 .mode = 0555, 1765 .child = inotify_table, 1766 }, 1767 #endif 1768 #ifdef CONFIG_EPOLL 1769 { 1770 .procname = "epoll", 1771 .mode = 0555, 1772 .child = epoll_table, 1773 }, 1774 #endif 1775 #endif 1776 { 1777 .procname = "protected_symlinks", 1778 .data = &sysctl_protected_symlinks, 1779 .maxlen = sizeof(int), 1780 .mode = 0600, 1781 .proc_handler = proc_dointvec_minmax, 1782 .extra1 = &zero, 1783 .extra2 = &one, 1784 }, 1785 { 1786 .procname = "protected_hardlinks", 1787 .data = &sysctl_protected_hardlinks, 1788 .maxlen = sizeof(int), 1789 .mode = 0600, 1790 .proc_handler = proc_dointvec_minmax, 1791 .extra1 = &zero, 1792 .extra2 = &one, 1793 }, 1794 { 1795 .procname = "suid_dumpable", 1796 .data = &suid_dumpable, 1797 .maxlen = sizeof(int), 1798 .mode = 0644, 1799 .proc_handler = proc_dointvec_minmax_coredump, 1800 .extra1 = &zero, 1801 .extra2 = &two, 1802 }, 1803 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE) 1804 { 1805 .procname = "binfmt_misc", 1806 .mode = 0555, 1807 .child = sysctl_mount_point, 1808 }, 1809 #endif 1810 { 1811 .procname = "pipe-max-size", 1812 .data = &pipe_max_size, 1813 .maxlen = sizeof(pipe_max_size), 1814 .mode = 0644, 1815 .proc_handler = &pipe_proc_fn, 1816 .extra1 = &pipe_min_size, 1817 }, 1818 { 1819 .procname = "pipe-user-pages-hard", 1820 .data = &pipe_user_pages_hard, 1821 .maxlen = sizeof(pipe_user_pages_hard), 1822 .mode = 0644, 1823 .proc_handler = proc_doulongvec_minmax, 1824 }, 1825 { 1826 .procname = "pipe-user-pages-soft", 1827 .data = &pipe_user_pages_soft, 1828 .maxlen = sizeof(pipe_user_pages_soft), 1829 .mode = 0644, 1830 .proc_handler = proc_doulongvec_minmax, 1831 }, 1832 { 1833 .procname = "mount-max", 1834 .data = &sysctl_mount_max, 1835 .maxlen = sizeof(unsigned int), 1836 .mode = 0644, 1837 .proc_handler = proc_dointvec_minmax, 1838 .extra1 = &one, 1839 }, 1840 { } 1841 }; 1842 1843 static struct ctl_table debug_table[] = { 1844 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE 1845 { 1846 .procname = "exception-trace", 1847 .data = &show_unhandled_signals, 1848 .maxlen = sizeof(int), 1849 .mode = 0644, 1850 .proc_handler = proc_dointvec 1851 }, 1852 #endif 1853 #if defined(CONFIG_OPTPROBES) 1854 { 1855 .procname = "kprobes-optimization", 1856 .data = &sysctl_kprobes_optimization, 1857 .maxlen = sizeof(int), 1858 .mode = 0644, 1859 .proc_handler = proc_kprobes_optimization_handler, 1860 .extra1 = &zero, 1861 .extra2 = &one, 1862 }, 1863 #endif 1864 { } 1865 }; 1866 1867 static struct ctl_table dev_table[] = { 1868 { } 1869 }; 1870 1871 int __init sysctl_init(void) 1872 { 1873 struct ctl_table_header *hdr; 1874 1875 hdr = register_sysctl_table(sysctl_base_table); 1876 kmemleak_not_leak(hdr); 1877 return 0; 1878 } 1879 1880 #endif /* CONFIG_SYSCTL */ 1881 1882 /* 1883 * /proc/sys support 1884 */ 1885 1886 #ifdef CONFIG_PROC_SYSCTL 1887 1888 static int _proc_do_string(char *data, int maxlen, int write, 1889 char __user *buffer, 1890 size_t *lenp, loff_t *ppos) 1891 { 1892 size_t len; 1893 char __user *p; 1894 char c; 1895 1896 if (!data || !maxlen || !*lenp) { 1897 *lenp = 0; 1898 return 0; 1899 } 1900 1901 if (write) { 1902 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) { 1903 /* Only continue writes not past the end of buffer. */ 1904 len = strlen(data); 1905 if (len > maxlen - 1) 1906 len = maxlen - 1; 1907 1908 if (*ppos > len) 1909 return 0; 1910 len = *ppos; 1911 } else { 1912 /* Start writing from beginning of buffer. */ 1913 len = 0; 1914 } 1915 1916 *ppos += *lenp; 1917 p = buffer; 1918 while ((p - buffer) < *lenp && len < maxlen - 1) { 1919 if (get_user(c, p++)) 1920 return -EFAULT; 1921 if (c == 0 || c == '\n') 1922 break; 1923 data[len++] = c; 1924 } 1925 data[len] = 0; 1926 } else { 1927 len = strlen(data); 1928 if (len > maxlen) 1929 len = maxlen; 1930 1931 if (*ppos > len) { 1932 *lenp = 0; 1933 return 0; 1934 } 1935 1936 data += *ppos; 1937 len -= *ppos; 1938 1939 if (len > *lenp) 1940 len = *lenp; 1941 if (len) 1942 if (copy_to_user(buffer, data, len)) 1943 return -EFAULT; 1944 if (len < *lenp) { 1945 if (put_user('\n', buffer + len)) 1946 return -EFAULT; 1947 len++; 1948 } 1949 *lenp = len; 1950 *ppos += len; 1951 } 1952 return 0; 1953 } 1954 1955 static void warn_sysctl_write(struct ctl_table *table) 1956 { 1957 pr_warn_once("%s wrote to %s when file position was not 0!\n" 1958 "This will not be supported in the future. To silence this\n" 1959 "warning, set kernel.sysctl_writes_strict = -1\n", 1960 current->comm, table->procname); 1961 } 1962 1963 /** 1964 * proc_first_pos_non_zero_ignore - check if firs position is allowed 1965 * @ppos: file position 1966 * @table: the sysctl table 1967 * 1968 * Returns true if the first position is non-zero and the sysctl_writes_strict 1969 * mode indicates this is not allowed for numeric input types. String proc 1970 * hadlers can ignore the return value. 1971 */ 1972 static bool proc_first_pos_non_zero_ignore(loff_t *ppos, 1973 struct ctl_table *table) 1974 { 1975 if (!*ppos) 1976 return false; 1977 1978 switch (sysctl_writes_strict) { 1979 case SYSCTL_WRITES_STRICT: 1980 return true; 1981 case SYSCTL_WRITES_WARN: 1982 warn_sysctl_write(table); 1983 return false; 1984 default: 1985 return false; 1986 } 1987 } 1988 1989 /** 1990 * proc_dostring - read a string sysctl 1991 * @table: the sysctl table 1992 * @write: %TRUE if this is a write to the sysctl file 1993 * @buffer: the user buffer 1994 * @lenp: the size of the user buffer 1995 * @ppos: file position 1996 * 1997 * Reads/writes a string from/to the user buffer. If the kernel 1998 * buffer provided is not large enough to hold the string, the 1999 * string is truncated. The copied string is %NULL-terminated. 2000 * If the string is being read by the user process, it is copied 2001 * and a newline '\n' is added. It is truncated if the buffer is 2002 * not large enough. 2003 * 2004 * Returns 0 on success. 2005 */ 2006 int proc_dostring(struct ctl_table *table, int write, 2007 void __user *buffer, size_t *lenp, loff_t *ppos) 2008 { 2009 if (write) 2010 proc_first_pos_non_zero_ignore(ppos, table); 2011 2012 return _proc_do_string((char *)(table->data), table->maxlen, write, 2013 (char __user *)buffer, lenp, ppos); 2014 } 2015 2016 static size_t proc_skip_spaces(char **buf) 2017 { 2018 size_t ret; 2019 char *tmp = skip_spaces(*buf); 2020 ret = tmp - *buf; 2021 *buf = tmp; 2022 return ret; 2023 } 2024 2025 static void proc_skip_char(char **buf, size_t *size, const char v) 2026 { 2027 while (*size) { 2028 if (**buf != v) 2029 break; 2030 (*size)--; 2031 (*buf)++; 2032 } 2033 } 2034 2035 #define TMPBUFLEN 22 2036 /** 2037 * proc_get_long - reads an ASCII formatted integer from a user buffer 2038 * 2039 * @buf: a kernel buffer 2040 * @size: size of the kernel buffer 2041 * @val: this is where the number will be stored 2042 * @neg: set to %TRUE if number is negative 2043 * @perm_tr: a vector which contains the allowed trailers 2044 * @perm_tr_len: size of the perm_tr vector 2045 * @tr: pointer to store the trailer character 2046 * 2047 * In case of success %0 is returned and @buf and @size are updated with 2048 * the amount of bytes read. If @tr is non-NULL and a trailing 2049 * character exists (size is non-zero after returning from this 2050 * function), @tr is updated with the trailing character. 2051 */ 2052 static int proc_get_long(char **buf, size_t *size, 2053 unsigned long *val, bool *neg, 2054 const char *perm_tr, unsigned perm_tr_len, char *tr) 2055 { 2056 int len; 2057 char *p, tmp[TMPBUFLEN]; 2058 2059 if (!*size) 2060 return -EINVAL; 2061 2062 len = *size; 2063 if (len > TMPBUFLEN - 1) 2064 len = TMPBUFLEN - 1; 2065 2066 memcpy(tmp, *buf, len); 2067 2068 tmp[len] = 0; 2069 p = tmp; 2070 if (*p == '-' && *size > 1) { 2071 *neg = true; 2072 p++; 2073 } else 2074 *neg = false; 2075 if (!isdigit(*p)) 2076 return -EINVAL; 2077 2078 *val = simple_strtoul(p, &p, 0); 2079 2080 len = p - tmp; 2081 2082 /* We don't know if the next char is whitespace thus we may accept 2083 * invalid integers (e.g. 1234...a) or two integers instead of one 2084 * (e.g. 123...1). So lets not allow such large numbers. */ 2085 if (len == TMPBUFLEN - 1) 2086 return -EINVAL; 2087 2088 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len)) 2089 return -EINVAL; 2090 2091 if (tr && (len < *size)) 2092 *tr = *p; 2093 2094 *buf += len; 2095 *size -= len; 2096 2097 return 0; 2098 } 2099 2100 /** 2101 * proc_put_long - converts an integer to a decimal ASCII formatted string 2102 * 2103 * @buf: the user buffer 2104 * @size: the size of the user buffer 2105 * @val: the integer to be converted 2106 * @neg: sign of the number, %TRUE for negative 2107 * 2108 * In case of success %0 is returned and @buf and @size are updated with 2109 * the amount of bytes written. 2110 */ 2111 static int proc_put_long(void __user **buf, size_t *size, unsigned long val, 2112 bool neg) 2113 { 2114 int len; 2115 char tmp[TMPBUFLEN], *p = tmp; 2116 2117 sprintf(p, "%s%lu", neg ? "-" : "", val); 2118 len = strlen(tmp); 2119 if (len > *size) 2120 len = *size; 2121 if (copy_to_user(*buf, tmp, len)) 2122 return -EFAULT; 2123 *size -= len; 2124 *buf += len; 2125 return 0; 2126 } 2127 #undef TMPBUFLEN 2128 2129 static int proc_put_char(void __user **buf, size_t *size, char c) 2130 { 2131 if (*size) { 2132 char __user **buffer = (char __user **)buf; 2133 if (put_user(c, *buffer)) 2134 return -EFAULT; 2135 (*size)--, (*buffer)++; 2136 *buf = *buffer; 2137 } 2138 return 0; 2139 } 2140 2141 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp, 2142 int *valp, 2143 int write, void *data) 2144 { 2145 if (write) { 2146 if (*negp) { 2147 if (*lvalp > (unsigned long) INT_MAX + 1) 2148 return -EINVAL; 2149 *valp = -*lvalp; 2150 } else { 2151 if (*lvalp > (unsigned long) INT_MAX) 2152 return -EINVAL; 2153 *valp = *lvalp; 2154 } 2155 } else { 2156 int val = *valp; 2157 if (val < 0) { 2158 *negp = true; 2159 *lvalp = -(unsigned long)val; 2160 } else { 2161 *negp = false; 2162 *lvalp = (unsigned long)val; 2163 } 2164 } 2165 return 0; 2166 } 2167 2168 static int do_proc_douintvec_conv(unsigned long *lvalp, 2169 unsigned int *valp, 2170 int write, void *data) 2171 { 2172 if (write) { 2173 if (*lvalp > UINT_MAX) 2174 return -EINVAL; 2175 *valp = *lvalp; 2176 } else { 2177 unsigned int val = *valp; 2178 *lvalp = (unsigned long)val; 2179 } 2180 return 0; 2181 } 2182 2183 static const char proc_wspace_sep[] = { ' ', '\t', '\n' }; 2184 2185 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table, 2186 int write, void __user *buffer, 2187 size_t *lenp, loff_t *ppos, 2188 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 2189 int write, void *data), 2190 void *data) 2191 { 2192 int *i, vleft, first = 1, err = 0; 2193 size_t left; 2194 char *kbuf = NULL, *p; 2195 2196 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) { 2197 *lenp = 0; 2198 return 0; 2199 } 2200 2201 i = (int *) tbl_data; 2202 vleft = table->maxlen / sizeof(*i); 2203 left = *lenp; 2204 2205 if (!conv) 2206 conv = do_proc_dointvec_conv; 2207 2208 if (write) { 2209 if (proc_first_pos_non_zero_ignore(ppos, table)) 2210 goto out; 2211 2212 if (left > PAGE_SIZE - 1) 2213 left = PAGE_SIZE - 1; 2214 p = kbuf = memdup_user_nul(buffer, left); 2215 if (IS_ERR(kbuf)) 2216 return PTR_ERR(kbuf); 2217 } 2218 2219 for (; left && vleft--; i++, first=0) { 2220 unsigned long lval; 2221 bool neg; 2222 2223 if (write) { 2224 left -= proc_skip_spaces(&p); 2225 2226 if (!left) 2227 break; 2228 err = proc_get_long(&p, &left, &lval, &neg, 2229 proc_wspace_sep, 2230 sizeof(proc_wspace_sep), NULL); 2231 if (err) 2232 break; 2233 if (conv(&neg, &lval, i, 1, data)) { 2234 err = -EINVAL; 2235 break; 2236 } 2237 } else { 2238 if (conv(&neg, &lval, i, 0, data)) { 2239 err = -EINVAL; 2240 break; 2241 } 2242 if (!first) 2243 err = proc_put_char(&buffer, &left, '\t'); 2244 if (err) 2245 break; 2246 err = proc_put_long(&buffer, &left, lval, neg); 2247 if (err) 2248 break; 2249 } 2250 } 2251 2252 if (!write && !first && left && !err) 2253 err = proc_put_char(&buffer, &left, '\n'); 2254 if (write && !err && left) 2255 left -= proc_skip_spaces(&p); 2256 if (write) { 2257 kfree(kbuf); 2258 if (first) 2259 return err ? : -EINVAL; 2260 } 2261 *lenp -= left; 2262 out: 2263 *ppos += *lenp; 2264 return err; 2265 } 2266 2267 static int do_proc_dointvec(struct ctl_table *table, int write, 2268 void __user *buffer, size_t *lenp, loff_t *ppos, 2269 int (*conv)(bool *negp, unsigned long *lvalp, int *valp, 2270 int write, void *data), 2271 void *data) 2272 { 2273 return __do_proc_dointvec(table->data, table, write, 2274 buffer, lenp, ppos, conv, data); 2275 } 2276 2277 static int do_proc_douintvec_w(unsigned int *tbl_data, 2278 struct ctl_table *table, 2279 void __user *buffer, 2280 size_t *lenp, loff_t *ppos, 2281 int (*conv)(unsigned long *lvalp, 2282 unsigned int *valp, 2283 int write, void *data), 2284 void *data) 2285 { 2286 unsigned long lval; 2287 int err = 0; 2288 size_t left; 2289 bool neg; 2290 char *kbuf = NULL, *p; 2291 2292 left = *lenp; 2293 2294 if (proc_first_pos_non_zero_ignore(ppos, table)) 2295 goto bail_early; 2296 2297 if (left > PAGE_SIZE - 1) 2298 left = PAGE_SIZE - 1; 2299 2300 p = kbuf = memdup_user_nul(buffer, left); 2301 if (IS_ERR(kbuf)) 2302 return -EINVAL; 2303 2304 left -= proc_skip_spaces(&p); 2305 if (!left) { 2306 err = -EINVAL; 2307 goto out_free; 2308 } 2309 2310 err = proc_get_long(&p, &left, &lval, &neg, 2311 proc_wspace_sep, 2312 sizeof(proc_wspace_sep), NULL); 2313 if (err || neg) { 2314 err = -EINVAL; 2315 goto out_free; 2316 } 2317 2318 if (conv(&lval, tbl_data, 1, data)) { 2319 err = -EINVAL; 2320 goto out_free; 2321 } 2322 2323 if (!err && left) 2324 left -= proc_skip_spaces(&p); 2325 2326 out_free: 2327 kfree(kbuf); 2328 if (err) 2329 return -EINVAL; 2330 2331 return 0; 2332 2333 /* This is in keeping with old __do_proc_dointvec() */ 2334 bail_early: 2335 *ppos += *lenp; 2336 return err; 2337 } 2338 2339 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer, 2340 size_t *lenp, loff_t *ppos, 2341 int (*conv)(unsigned long *lvalp, 2342 unsigned int *valp, 2343 int write, void *data), 2344 void *data) 2345 { 2346 unsigned long lval; 2347 int err = 0; 2348 size_t left; 2349 2350 left = *lenp; 2351 2352 if (conv(&lval, tbl_data, 0, data)) { 2353 err = -EINVAL; 2354 goto out; 2355 } 2356 2357 err = proc_put_long(&buffer, &left, lval, false); 2358 if (err || !left) 2359 goto out; 2360 2361 err = proc_put_char(&buffer, &left, '\n'); 2362 2363 out: 2364 *lenp -= left; 2365 *ppos += *lenp; 2366 2367 return err; 2368 } 2369 2370 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table, 2371 int write, void __user *buffer, 2372 size_t *lenp, loff_t *ppos, 2373 int (*conv)(unsigned long *lvalp, 2374 unsigned int *valp, 2375 int write, void *data), 2376 void *data) 2377 { 2378 unsigned int *i, vleft; 2379 2380 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) { 2381 *lenp = 0; 2382 return 0; 2383 } 2384 2385 i = (unsigned int *) tbl_data; 2386 vleft = table->maxlen / sizeof(*i); 2387 2388 /* 2389 * Arrays are not supported, keep this simple. *Do not* add 2390 * support for them. 2391 */ 2392 if (vleft != 1) { 2393 *lenp = 0; 2394 return -EINVAL; 2395 } 2396 2397 if (!conv) 2398 conv = do_proc_douintvec_conv; 2399 2400 if (write) 2401 return do_proc_douintvec_w(i, table, buffer, lenp, ppos, 2402 conv, data); 2403 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data); 2404 } 2405 2406 static int do_proc_douintvec(struct ctl_table *table, int write, 2407 void __user *buffer, size_t *lenp, loff_t *ppos, 2408 int (*conv)(unsigned long *lvalp, 2409 unsigned int *valp, 2410 int write, void *data), 2411 void *data) 2412 { 2413 return __do_proc_douintvec(table->data, table, write, 2414 buffer, lenp, ppos, conv, data); 2415 } 2416 2417 /** 2418 * proc_dointvec - read a vector of integers 2419 * @table: the sysctl table 2420 * @write: %TRUE if this is a write to the sysctl file 2421 * @buffer: the user buffer 2422 * @lenp: the size of the user buffer 2423 * @ppos: file position 2424 * 2425 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2426 * values from/to the user buffer, treated as an ASCII string. 2427 * 2428 * Returns 0 on success. 2429 */ 2430 int proc_dointvec(struct ctl_table *table, int write, 2431 void __user *buffer, size_t *lenp, loff_t *ppos) 2432 { 2433 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL); 2434 } 2435 2436 /** 2437 * proc_douintvec - read a vector of unsigned integers 2438 * @table: the sysctl table 2439 * @write: %TRUE if this is a write to the sysctl file 2440 * @buffer: the user buffer 2441 * @lenp: the size of the user buffer 2442 * @ppos: file position 2443 * 2444 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer 2445 * values from/to the user buffer, treated as an ASCII string. 2446 * 2447 * Returns 0 on success. 2448 */ 2449 int proc_douintvec(struct ctl_table *table, int write, 2450 void __user *buffer, size_t *lenp, loff_t *ppos) 2451 { 2452 return do_proc_douintvec(table, write, buffer, lenp, ppos, 2453 do_proc_douintvec_conv, NULL); 2454 } 2455 2456 /* 2457 * Taint values can only be increased 2458 * This means we can safely use a temporary. 2459 */ 2460 static int proc_taint(struct ctl_table *table, int write, 2461 void __user *buffer, size_t *lenp, loff_t *ppos) 2462 { 2463 struct ctl_table t; 2464 unsigned long tmptaint = get_taint(); 2465 int err; 2466 2467 if (write && !capable(CAP_SYS_ADMIN)) 2468 return -EPERM; 2469 2470 t = *table; 2471 t.data = &tmptaint; 2472 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos); 2473 if (err < 0) 2474 return err; 2475 2476 if (write) { 2477 /* 2478 * Poor man's atomic or. Not worth adding a primitive 2479 * to everyone's atomic.h for this 2480 */ 2481 int i; 2482 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) { 2483 if ((tmptaint >> i) & 1) 2484 add_taint(i, LOCKDEP_STILL_OK); 2485 } 2486 } 2487 2488 return err; 2489 } 2490 2491 #ifdef CONFIG_PRINTK 2492 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write, 2493 void __user *buffer, size_t *lenp, loff_t *ppos) 2494 { 2495 if (write && !capable(CAP_SYS_ADMIN)) 2496 return -EPERM; 2497 2498 return proc_dointvec_minmax(table, write, buffer, lenp, ppos); 2499 } 2500 #endif 2501 2502 struct do_proc_dointvec_minmax_conv_param { 2503 int *min; 2504 int *max; 2505 }; 2506 2507 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp, 2508 int *valp, 2509 int write, void *data) 2510 { 2511 struct do_proc_dointvec_minmax_conv_param *param = data; 2512 if (write) { 2513 int val = *negp ? -*lvalp : *lvalp; 2514 if ((param->min && *param->min > val) || 2515 (param->max && *param->max < val)) 2516 return -EINVAL; 2517 *valp = val; 2518 } else { 2519 int val = *valp; 2520 if (val < 0) { 2521 *negp = true; 2522 *lvalp = -(unsigned long)val; 2523 } else { 2524 *negp = false; 2525 *lvalp = (unsigned long)val; 2526 } 2527 } 2528 return 0; 2529 } 2530 2531 /** 2532 * proc_dointvec_minmax - read a vector of integers with min/max values 2533 * @table: the sysctl table 2534 * @write: %TRUE if this is a write to the sysctl file 2535 * @buffer: the user buffer 2536 * @lenp: the size of the user buffer 2537 * @ppos: file position 2538 * 2539 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2540 * values from/to the user buffer, treated as an ASCII string. 2541 * 2542 * This routine will ensure the values are within the range specified by 2543 * table->extra1 (min) and table->extra2 (max). 2544 * 2545 * Returns 0 on success. 2546 */ 2547 int proc_dointvec_minmax(struct ctl_table *table, int write, 2548 void __user *buffer, size_t *lenp, loff_t *ppos) 2549 { 2550 struct do_proc_dointvec_minmax_conv_param param = { 2551 .min = (int *) table->extra1, 2552 .max = (int *) table->extra2, 2553 }; 2554 return do_proc_dointvec(table, write, buffer, lenp, ppos, 2555 do_proc_dointvec_minmax_conv, ¶m); 2556 } 2557 2558 struct do_proc_douintvec_minmax_conv_param { 2559 unsigned int *min; 2560 unsigned int *max; 2561 }; 2562 2563 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp, 2564 unsigned int *valp, 2565 int write, void *data) 2566 { 2567 struct do_proc_douintvec_minmax_conv_param *param = data; 2568 2569 if (write) { 2570 unsigned int val = *lvalp; 2571 2572 if (*lvalp > UINT_MAX) 2573 return -EINVAL; 2574 2575 if ((param->min && *param->min > val) || 2576 (param->max && *param->max < val)) 2577 return -ERANGE; 2578 2579 *valp = val; 2580 } else { 2581 unsigned int val = *valp; 2582 *lvalp = (unsigned long) val; 2583 } 2584 2585 return 0; 2586 } 2587 2588 /** 2589 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values 2590 * @table: the sysctl table 2591 * @write: %TRUE if this is a write to the sysctl file 2592 * @buffer: the user buffer 2593 * @lenp: the size of the user buffer 2594 * @ppos: file position 2595 * 2596 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer 2597 * values from/to the user buffer, treated as an ASCII string. Negative 2598 * strings are not allowed. 2599 * 2600 * This routine will ensure the values are within the range specified by 2601 * table->extra1 (min) and table->extra2 (max). There is a final sanity 2602 * check for UINT_MAX to avoid having to support wrap around uses from 2603 * userspace. 2604 * 2605 * Returns 0 on success. 2606 */ 2607 int proc_douintvec_minmax(struct ctl_table *table, int write, 2608 void __user *buffer, size_t *lenp, loff_t *ppos) 2609 { 2610 struct do_proc_douintvec_minmax_conv_param param = { 2611 .min = (unsigned int *) table->extra1, 2612 .max = (unsigned int *) table->extra2, 2613 }; 2614 return do_proc_douintvec(table, write, buffer, lenp, ppos, 2615 do_proc_douintvec_minmax_conv, ¶m); 2616 } 2617 2618 struct do_proc_dopipe_max_size_conv_param { 2619 unsigned int *min; 2620 }; 2621 2622 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp, 2623 unsigned int *valp, 2624 int write, void *data) 2625 { 2626 struct do_proc_dopipe_max_size_conv_param *param = data; 2627 2628 if (write) { 2629 unsigned int val; 2630 2631 if (*lvalp > UINT_MAX) 2632 return -EINVAL; 2633 2634 val = round_pipe_size(*lvalp); 2635 if (val == 0) 2636 return -EINVAL; 2637 2638 if (param->min && *param->min > val) 2639 return -ERANGE; 2640 2641 *valp = val; 2642 } else { 2643 unsigned int val = *valp; 2644 *lvalp = (unsigned long) val; 2645 } 2646 2647 return 0; 2648 } 2649 2650 int proc_dopipe_max_size(struct ctl_table *table, int write, 2651 void __user *buffer, size_t *lenp, loff_t *ppos) 2652 { 2653 struct do_proc_dopipe_max_size_conv_param param = { 2654 .min = (unsigned int *) table->extra1, 2655 }; 2656 return do_proc_douintvec(table, write, buffer, lenp, ppos, 2657 do_proc_dopipe_max_size_conv, ¶m); 2658 } 2659 2660 static void validate_coredump_safety(void) 2661 { 2662 #ifdef CONFIG_COREDUMP 2663 if (suid_dumpable == SUID_DUMP_ROOT && 2664 core_pattern[0] != '/' && core_pattern[0] != '|') { 2665 printk(KERN_WARNING 2666 "Unsafe core_pattern used with fs.suid_dumpable=2.\n" 2667 "Pipe handler or fully qualified core dump path required.\n" 2668 "Set kernel.core_pattern before fs.suid_dumpable.\n" 2669 ); 2670 } 2671 #endif 2672 } 2673 2674 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write, 2675 void __user *buffer, size_t *lenp, loff_t *ppos) 2676 { 2677 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos); 2678 if (!error) 2679 validate_coredump_safety(); 2680 return error; 2681 } 2682 2683 #ifdef CONFIG_COREDUMP 2684 static int proc_dostring_coredump(struct ctl_table *table, int write, 2685 void __user *buffer, size_t *lenp, loff_t *ppos) 2686 { 2687 int error = proc_dostring(table, write, buffer, lenp, ppos); 2688 if (!error) 2689 validate_coredump_safety(); 2690 return error; 2691 } 2692 #endif 2693 2694 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write, 2695 void __user *buffer, 2696 size_t *lenp, loff_t *ppos, 2697 unsigned long convmul, 2698 unsigned long convdiv) 2699 { 2700 unsigned long *i, *min, *max; 2701 int vleft, first = 1, err = 0; 2702 size_t left; 2703 char *kbuf = NULL, *p; 2704 2705 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) { 2706 *lenp = 0; 2707 return 0; 2708 } 2709 2710 i = (unsigned long *) data; 2711 min = (unsigned long *) table->extra1; 2712 max = (unsigned long *) table->extra2; 2713 vleft = table->maxlen / sizeof(unsigned long); 2714 left = *lenp; 2715 2716 if (write) { 2717 if (proc_first_pos_non_zero_ignore(ppos, table)) 2718 goto out; 2719 2720 if (left > PAGE_SIZE - 1) 2721 left = PAGE_SIZE - 1; 2722 p = kbuf = memdup_user_nul(buffer, left); 2723 if (IS_ERR(kbuf)) 2724 return PTR_ERR(kbuf); 2725 } 2726 2727 for (; left && vleft--; i++, first = 0) { 2728 unsigned long val; 2729 2730 if (write) { 2731 bool neg; 2732 2733 left -= proc_skip_spaces(&p); 2734 2735 err = proc_get_long(&p, &left, &val, &neg, 2736 proc_wspace_sep, 2737 sizeof(proc_wspace_sep), NULL); 2738 if (err) 2739 break; 2740 if (neg) 2741 continue; 2742 val = convmul * val / convdiv; 2743 if ((min && val < *min) || (max && val > *max)) 2744 continue; 2745 *i = val; 2746 } else { 2747 val = convdiv * (*i) / convmul; 2748 if (!first) { 2749 err = proc_put_char(&buffer, &left, '\t'); 2750 if (err) 2751 break; 2752 } 2753 err = proc_put_long(&buffer, &left, val, false); 2754 if (err) 2755 break; 2756 } 2757 } 2758 2759 if (!write && !first && left && !err) 2760 err = proc_put_char(&buffer, &left, '\n'); 2761 if (write && !err) 2762 left -= proc_skip_spaces(&p); 2763 if (write) { 2764 kfree(kbuf); 2765 if (first) 2766 return err ? : -EINVAL; 2767 } 2768 *lenp -= left; 2769 out: 2770 *ppos += *lenp; 2771 return err; 2772 } 2773 2774 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write, 2775 void __user *buffer, 2776 size_t *lenp, loff_t *ppos, 2777 unsigned long convmul, 2778 unsigned long convdiv) 2779 { 2780 return __do_proc_doulongvec_minmax(table->data, table, write, 2781 buffer, lenp, ppos, convmul, convdiv); 2782 } 2783 2784 /** 2785 * proc_doulongvec_minmax - read a vector of long integers with min/max values 2786 * @table: the sysctl table 2787 * @write: %TRUE if this is a write to the sysctl file 2788 * @buffer: the user buffer 2789 * @lenp: the size of the user buffer 2790 * @ppos: file position 2791 * 2792 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 2793 * values from/to the user buffer, treated as an ASCII string. 2794 * 2795 * This routine will ensure the values are within the range specified by 2796 * table->extra1 (min) and table->extra2 (max). 2797 * 2798 * Returns 0 on success. 2799 */ 2800 int proc_doulongvec_minmax(struct ctl_table *table, int write, 2801 void __user *buffer, size_t *lenp, loff_t *ppos) 2802 { 2803 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l); 2804 } 2805 2806 /** 2807 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values 2808 * @table: the sysctl table 2809 * @write: %TRUE if this is a write to the sysctl file 2810 * @buffer: the user buffer 2811 * @lenp: the size of the user buffer 2812 * @ppos: file position 2813 * 2814 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long 2815 * values from/to the user buffer, treated as an ASCII string. The values 2816 * are treated as milliseconds, and converted to jiffies when they are stored. 2817 * 2818 * This routine will ensure the values are within the range specified by 2819 * table->extra1 (min) and table->extra2 (max). 2820 * 2821 * Returns 0 on success. 2822 */ 2823 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 2824 void __user *buffer, 2825 size_t *lenp, loff_t *ppos) 2826 { 2827 return do_proc_doulongvec_minmax(table, write, buffer, 2828 lenp, ppos, HZ, 1000l); 2829 } 2830 2831 2832 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp, 2833 int *valp, 2834 int write, void *data) 2835 { 2836 if (write) { 2837 if (*lvalp > INT_MAX / HZ) 2838 return 1; 2839 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ); 2840 } else { 2841 int val = *valp; 2842 unsigned long lval; 2843 if (val < 0) { 2844 *negp = true; 2845 lval = -(unsigned long)val; 2846 } else { 2847 *negp = false; 2848 lval = (unsigned long)val; 2849 } 2850 *lvalp = lval / HZ; 2851 } 2852 return 0; 2853 } 2854 2855 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp, 2856 int *valp, 2857 int write, void *data) 2858 { 2859 if (write) { 2860 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ) 2861 return 1; 2862 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp); 2863 } else { 2864 int val = *valp; 2865 unsigned long lval; 2866 if (val < 0) { 2867 *negp = true; 2868 lval = -(unsigned long)val; 2869 } else { 2870 *negp = false; 2871 lval = (unsigned long)val; 2872 } 2873 *lvalp = jiffies_to_clock_t(lval); 2874 } 2875 return 0; 2876 } 2877 2878 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp, 2879 int *valp, 2880 int write, void *data) 2881 { 2882 if (write) { 2883 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp); 2884 2885 if (jif > INT_MAX) 2886 return 1; 2887 *valp = (int)jif; 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 = jiffies_to_msecs(lval); 2899 } 2900 return 0; 2901 } 2902 2903 /** 2904 * proc_dointvec_jiffies - read a vector of integers as seconds 2905 * @table: the sysctl table 2906 * @write: %TRUE if this is a write to the sysctl file 2907 * @buffer: the user buffer 2908 * @lenp: the size of the user buffer 2909 * @ppos: file position 2910 * 2911 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2912 * values from/to the user buffer, treated as an ASCII string. 2913 * The values read are assumed to be in seconds, and are converted into 2914 * jiffies. 2915 * 2916 * Returns 0 on success. 2917 */ 2918 int proc_dointvec_jiffies(struct ctl_table *table, int write, 2919 void __user *buffer, size_t *lenp, loff_t *ppos) 2920 { 2921 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2922 do_proc_dointvec_jiffies_conv,NULL); 2923 } 2924 2925 /** 2926 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds 2927 * @table: the sysctl table 2928 * @write: %TRUE if this is a write to the sysctl file 2929 * @buffer: the user buffer 2930 * @lenp: the size of the user buffer 2931 * @ppos: pointer to the file position 2932 * 2933 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2934 * values from/to the user buffer, treated as an ASCII string. 2935 * The values read are assumed to be in 1/USER_HZ seconds, and 2936 * are converted into jiffies. 2937 * 2938 * Returns 0 on success. 2939 */ 2940 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 2941 void __user *buffer, size_t *lenp, loff_t *ppos) 2942 { 2943 return do_proc_dointvec(table,write,buffer,lenp,ppos, 2944 do_proc_dointvec_userhz_jiffies_conv,NULL); 2945 } 2946 2947 /** 2948 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds 2949 * @table: the sysctl table 2950 * @write: %TRUE if this is a write to the sysctl file 2951 * @buffer: the user buffer 2952 * @lenp: the size of the user buffer 2953 * @ppos: file position 2954 * @ppos: the current position in the file 2955 * 2956 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer 2957 * values from/to the user buffer, treated as an ASCII string. 2958 * The values read are assumed to be in 1/1000 seconds, and 2959 * are converted into jiffies. 2960 * 2961 * Returns 0 on success. 2962 */ 2963 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, 2964 void __user *buffer, size_t *lenp, loff_t *ppos) 2965 { 2966 return do_proc_dointvec(table, write, buffer, lenp, ppos, 2967 do_proc_dointvec_ms_jiffies_conv, NULL); 2968 } 2969 2970 static int proc_do_cad_pid(struct ctl_table *table, int write, 2971 void __user *buffer, size_t *lenp, loff_t *ppos) 2972 { 2973 struct pid *new_pid; 2974 pid_t tmp; 2975 int r; 2976 2977 tmp = pid_vnr(cad_pid); 2978 2979 r = __do_proc_dointvec(&tmp, table, write, buffer, 2980 lenp, ppos, NULL, NULL); 2981 if (r || !write) 2982 return r; 2983 2984 new_pid = find_get_pid(tmp); 2985 if (!new_pid) 2986 return -ESRCH; 2987 2988 put_pid(xchg(&cad_pid, new_pid)); 2989 return 0; 2990 } 2991 2992 /** 2993 * proc_do_large_bitmap - read/write from/to a large bitmap 2994 * @table: the sysctl table 2995 * @write: %TRUE if this is a write to the sysctl file 2996 * @buffer: the user buffer 2997 * @lenp: the size of the user buffer 2998 * @ppos: file position 2999 * 3000 * The bitmap is stored at table->data and the bitmap length (in bits) 3001 * in table->maxlen. 3002 * 3003 * We use a range comma separated format (e.g. 1,3-4,10-10) so that 3004 * large bitmaps may be represented in a compact manner. Writing into 3005 * the file will clear the bitmap then update it with the given input. 3006 * 3007 * Returns 0 on success. 3008 */ 3009 int proc_do_large_bitmap(struct ctl_table *table, int write, 3010 void __user *buffer, size_t *lenp, loff_t *ppos) 3011 { 3012 int err = 0; 3013 bool first = 1; 3014 size_t left = *lenp; 3015 unsigned long bitmap_len = table->maxlen; 3016 unsigned long *bitmap = *(unsigned long **) table->data; 3017 unsigned long *tmp_bitmap = NULL; 3018 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c; 3019 3020 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) { 3021 *lenp = 0; 3022 return 0; 3023 } 3024 3025 if (write) { 3026 char *kbuf, *p; 3027 3028 if (left > PAGE_SIZE - 1) 3029 left = PAGE_SIZE - 1; 3030 3031 p = kbuf = memdup_user_nul(buffer, left); 3032 if (IS_ERR(kbuf)) 3033 return PTR_ERR(kbuf); 3034 3035 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long), 3036 GFP_KERNEL); 3037 if (!tmp_bitmap) { 3038 kfree(kbuf); 3039 return -ENOMEM; 3040 } 3041 proc_skip_char(&p, &left, '\n'); 3042 while (!err && left) { 3043 unsigned long val_a, val_b; 3044 bool neg; 3045 3046 err = proc_get_long(&p, &left, &val_a, &neg, tr_a, 3047 sizeof(tr_a), &c); 3048 if (err) 3049 break; 3050 if (val_a >= bitmap_len || neg) { 3051 err = -EINVAL; 3052 break; 3053 } 3054 3055 val_b = val_a; 3056 if (left) { 3057 p++; 3058 left--; 3059 } 3060 3061 if (c == '-') { 3062 err = proc_get_long(&p, &left, &val_b, 3063 &neg, tr_b, sizeof(tr_b), 3064 &c); 3065 if (err) 3066 break; 3067 if (val_b >= bitmap_len || neg || 3068 val_a > val_b) { 3069 err = -EINVAL; 3070 break; 3071 } 3072 if (left) { 3073 p++; 3074 left--; 3075 } 3076 } 3077 3078 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1); 3079 first = 0; 3080 proc_skip_char(&p, &left, '\n'); 3081 } 3082 kfree(kbuf); 3083 } else { 3084 unsigned long bit_a, bit_b = 0; 3085 3086 while (left) { 3087 bit_a = find_next_bit(bitmap, bitmap_len, bit_b); 3088 if (bit_a >= bitmap_len) 3089 break; 3090 bit_b = find_next_zero_bit(bitmap, bitmap_len, 3091 bit_a + 1) - 1; 3092 3093 if (!first) { 3094 err = proc_put_char(&buffer, &left, ','); 3095 if (err) 3096 break; 3097 } 3098 err = proc_put_long(&buffer, &left, bit_a, false); 3099 if (err) 3100 break; 3101 if (bit_a != bit_b) { 3102 err = proc_put_char(&buffer, &left, '-'); 3103 if (err) 3104 break; 3105 err = proc_put_long(&buffer, &left, bit_b, false); 3106 if (err) 3107 break; 3108 } 3109 3110 first = 0; bit_b++; 3111 } 3112 if (!err) 3113 err = proc_put_char(&buffer, &left, '\n'); 3114 } 3115 3116 if (!err) { 3117 if (write) { 3118 if (*ppos) 3119 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len); 3120 else 3121 bitmap_copy(bitmap, tmp_bitmap, bitmap_len); 3122 } 3123 *lenp -= left; 3124 *ppos += *lenp; 3125 } 3126 3127 kfree(tmp_bitmap); 3128 return err; 3129 } 3130 3131 #else /* CONFIG_PROC_SYSCTL */ 3132 3133 int proc_dostring(struct ctl_table *table, int write, 3134 void __user *buffer, size_t *lenp, loff_t *ppos) 3135 { 3136 return -ENOSYS; 3137 } 3138 3139 int proc_dointvec(struct ctl_table *table, int write, 3140 void __user *buffer, size_t *lenp, loff_t *ppos) 3141 { 3142 return -ENOSYS; 3143 } 3144 3145 int proc_douintvec(struct ctl_table *table, int write, 3146 void __user *buffer, size_t *lenp, loff_t *ppos) 3147 { 3148 return -ENOSYS; 3149 } 3150 3151 int proc_dointvec_minmax(struct ctl_table *table, int write, 3152 void __user *buffer, size_t *lenp, loff_t *ppos) 3153 { 3154 return -ENOSYS; 3155 } 3156 3157 int proc_douintvec_minmax(struct ctl_table *table, int write, 3158 void __user *buffer, size_t *lenp, loff_t *ppos) 3159 { 3160 return -ENOSYS; 3161 } 3162 3163 int proc_dopipe_max_size(struct ctl_table *table, int write, 3164 void __user *buffer, size_t *lenp, loff_t *ppos) 3165 { 3166 return -ENOSYS; 3167 } 3168 3169 int proc_dointvec_jiffies(struct ctl_table *table, int write, 3170 void __user *buffer, size_t *lenp, loff_t *ppos) 3171 { 3172 return -ENOSYS; 3173 } 3174 3175 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, 3176 void __user *buffer, size_t *lenp, loff_t *ppos) 3177 { 3178 return -ENOSYS; 3179 } 3180 3181 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, 3182 void __user *buffer, size_t *lenp, loff_t *ppos) 3183 { 3184 return -ENOSYS; 3185 } 3186 3187 int proc_doulongvec_minmax(struct ctl_table *table, int write, 3188 void __user *buffer, size_t *lenp, loff_t *ppos) 3189 { 3190 return -ENOSYS; 3191 } 3192 3193 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write, 3194 void __user *buffer, 3195 size_t *lenp, loff_t *ppos) 3196 { 3197 return -ENOSYS; 3198 } 3199 3200 3201 #endif /* CONFIG_PROC_SYSCTL */ 3202 3203 /* 3204 * No sense putting this after each symbol definition, twice, 3205 * exception granted :-) 3206 */ 3207 EXPORT_SYMBOL(proc_dointvec); 3208 EXPORT_SYMBOL(proc_douintvec); 3209 EXPORT_SYMBOL(proc_dointvec_jiffies); 3210 EXPORT_SYMBOL(proc_dointvec_minmax); 3211 EXPORT_SYMBOL_GPL(proc_douintvec_minmax); 3212 EXPORT_SYMBOL_GPL(proc_dopipe_max_size); 3213 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies); 3214 EXPORT_SYMBOL(proc_dointvec_ms_jiffies); 3215 EXPORT_SYMBOL(proc_dostring); 3216 EXPORT_SYMBOL(proc_doulongvec_minmax); 3217 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax); 3218